P. 1
sc-8-07_downmagaz

sc-8-07_downmagaz

|Views: 151|Likes:
Published by Luky Barabas

More info:

Published by: Luky Barabas on Jul 18, 2011
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

08/30/2013

pdf

text

original

CONTENTS Volume 333 Issue 6039

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 127
COVER
Composite image of the Sombrero galaxy captured by
the Hubble Space Telescope in 2003. Like the Milky Way,
the Sombrero is a spiral galaxy with a central stellar bulge
and a thin disk containing spiral arms of newly formed stars.
The special section beginning on page 169 examines the ways
in which galaxies change over time.
Image: NASA and The Hubble Heritage Team (Space Telescope Science
Institute/Association of Universities for Research in Astronomy)
DEPARTMENTS
133 This Week in Science
137 Editors’ Choice
139 Science Staff
243 New Products
244 Science Careers
page 148
EDITORIAL
136 A Threat to Medical Innovation
Michael Rosbash
NEWS OF THE WEEK
140 A roundup of the week’s top stories
NEWS & ANALYSIS
143 Once on ‘Fast Track,’
Avastin Now Derailed
144 Foreign Researchers Begin
to Make Their Mark
145 Unseen World of Clinical Trials
Emerges From U.S. Database
146 Cash Advance, New Approach
Aim to Relaunch Biosphere 2
147 DARPA Offers $30 Million to
Jump-Start Cellular Factories
NEWS FOCUS
148 NUCLEAR WASTE
Waste Panel Expected to
Back Interim Storage
>> Science Podcast
150 Light at the End of the Radwaste Disposal
Tunnel Could Be Real
153 Picking Up the Pieces at
Ravaged Tohoku University
Facilities Plot Research Revival
LETTERS
156 Seeds of Change for Restoration Ecology
R. Aerts and O. Honnay
Manufacturing Decline Yields
Drug Shortages
P. Connelly and B. P. Quinn
NSF’s Struggle to Articulate Relevance
R. Frodeman and J. B. Holbrook
158 CORRECTIONS AND CLARIFICATIONS
BOOKS ET AL.
159 Science as Psychology
L. M. Osbeck et al.,
reviewed by R. N. Giere
160 Loving and Hating Mathematics
R. Hersh and V. John-Steiner,
reviewed by L. A. Steen
POLICY FORUM
161 A Dark Age for Space Astronomy?
R.-M. Bonnet and J. A. M. Bleeker
>> Galaxy Evolution section p. 169
PERSPECTIVES
163 Lipases in Cachexia
P. Arner
>> Report p. 233
164 Precision, Not Power
E. E. Eyler
>> Report p. 196
165 A Critical Point for Turbulence
B. Eckhardt
>> Research Article p. 192
166 Sex, Death, and the Red Queen
M. A. Brockhurst
>> Report p. 216
CONTENTS continued >>
SPECIAL SECTION
Galaxy Evolution
INTRODUCTION
169 A Universe of Galaxies
NEWS
170 Milky Way Researchers’
Home Away From Home
173 Galaxy Zoo Volunteers Share
Pain and Glory of Research
PERSPECTIVE
176 Galactic Paleontology
E. Tolstoy
REVIEWS
178 The Cosmic History of Star Formation
J. S. Dunlop
182 The Coevolution of Galaxies
and Supermassive Black Holes:
A Local Perspective
T. M. Heckman and G. Kauffmann
>> Policy Forum p. 161
Published by AAAS
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 129
CONTENTS
page 159
pages 164 & 196
BREVIA
186 Adult Neural Function Requires MeCP2
C. M. McGraw et al.
An epigenetic program regulated by MeCP2
needs to be maintained throughout life for
normal neurological function.
RESEARCH ARTICLES
187 Glycolytic Oscillations and Limits
on Robust Efficiency
F. A. Chandra et al.
Gylcolytic oscillations in yeast are a
by-product of a trade-off between robustness
and efficiency.
192 The Onset of Turbulence in Pipe Flow
K. Avila et al.
The lifetimes of injected jet puffs are used to
determine the critical point at which turbulent
pipe flow is sustained.
>> Perspective p. 165
REPORTS
196 Frequency Metrology in Quantum
Degenerate Helium: Direct Measurement
of the 2
3
S
1
→ 2
1
S
0
Transition
R. van Rooij et al.
Measurement of an extremely weak
spectroscopic transition in helium hones
fundamental atomic theories.
>> Perspective p. 164
199 An Extremely Luminous Panchromatic
Outburst from the Nucleus of a
Distant Galaxy
A. J. Levan et al.
203 A Possible Relativistic Jetted Outburst
from a Massive Black Hole Fed by a
Tidally Disrupted Star
J. S. Bloom et al.
A recent bright emission observed by the Swift
satellite is due to the sudden accretion of a
star onto a massive black hole.
206 Observation of Transient
Structural-Transformation Dynamics
in a Cu
2
S Nanorod
H. Zheng et al.
Structural fluctuations between two
equilibrium phases are observed in
copper sulfide nanoparticles.
209 Palladium-Catalyzed Aerobic
Dehydrogenation of Substituted
Cyclohexanones to Phenols
Y. Izawa et al.
Phenol derivatives are prepared from
nonaromatic ring compounds that can
bear a wide variety of substitutents.
213 High Pre-Eruptive Water Contents
Preserved in Lunar Melt Inclusions
E. H. Hauri et al.
Primitive magmatic melt inclusions from
the Moon contain as much water as some
terrestrial mid-ocean ridge magmas.
216 Running with the Red Queen:
Host-Parasite Coevolution Selects
for Biparental Sex
L. T. Morran et al.
Outcrossing provides better survival than
self-fertilization during coevolution
between a host and its parasite.
>> Perspective p. 166
218 Isolation of Single Human Hematopoietic
Stem Cells Capable of Long-Term
Multilineage Engraftment
F. Notta et al.
Proteins are identified that underlie the early
commitment steps of human hematopoietic
stem cell differentiation.
222 Coupled, Circumferential Motions of the
Cell Wall Synthesis Machinery and MreB
Filaments in B. subtilis
E. C. Garner et al.
225 Processive Movement of MreB-Associated
Cell Wall Biosynthetic Complexes in
Bacteria
J. Domínguez-Escobar et al.
Bacteria elongation involves moving
synthetic complexes around the cell wall
228 Phosphorylation of the Autophagy
Receptor Optineurin Restricts Salmonella
Growth
P. Wild et al.
Phosphorylation of an autophagy receptor
restricts pathogenic cytosolic bacterial growth.
233 Adipose Triglyceride Lipase Contributes to
Cancer-Associated Cachexia
S. K. Das et al.
Ablation of a gene controlling fat breakdown
can protect mice from cancer-associated
uncontrolled loss of fat and muscle.
>> Perspective p. 163
238 A Pericyte Origin of Spinal Cord Scar
Tissue
C. Göritz et al.
Scars formed in response to damage to the
central nervous system show unexpected
complexity.
>> Science Podcast
CONTENTS continued >>
page 238
Published by AAAS
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 131
CONTENTS
SCIENCEXPRESS
www.sciencexpress.org
Herschel Detects a Massive Dust Reservoir
in Supernova 1987A
M. Matsuura et al.
The large amount of dust produced by this
supernova may help explain the dust observed
in young galaxies.
10.1126/science.1205983
CRASY: Mass- or Electron-Correlated
Rotational Alignment Spectroscopy
C. Schröter et al.
A technique merging rotational spectroscopy with
mass spectrometry facilitates analysis of a complex
isotopic mixture.
10.1126/science.1204352
A Highly Conserved Neutralizing Epitope
on Group 2 Influenza A Viruses
D. C. Ekiert et al.
An antibody against a conserved epitope broadly
neutralizes group 2 influenza viruses.
10.1126/science.1204839
>> Science Podcast
Schema-Dependent Gene Activation
and Memory Encoding in Neocortex
D. Tse et al.
New hippocampal-dependent learning is in
parallel consolidated with existing memories
in the neocortex.
10.1126/science.1205274
Drosophila Sex lethal Gene Initiates Female
Development in Germline Progenitors
K. Hashiyama et al.
Primordial germ cells are directed toward
oogenesis even before they migrate to the
gonads of the fruit fly.
10.1126/science.1208146
High Value and Long Life—Double Jeopardy
for Tunas and Billfishes
B. B. Collette et al.
10.1126/science.1208730
SCIENCENOW
www.sciencenow.org
Highlights From Our Daily News Coverage
Magnetic Nanoparticles Fry Tumors
“Magnetic hyperthermia” proves more effective
in mice than a traditional cancer drug.
http://scim.ag/_hyperthermia
Long-Dead Cane Toads Continue
to Haunt Australian Wildlife
Road-kill “toad jerky” is highly toxic,
even after months in the Sun.
http://scim.ag/cane-toads
Chimps Are Good Listeners, Too
Like humans, chimps can pick out meaning
even when the sounds of words are distorted,
suggesting yet again that human language
has roots in other species.
http://scim.ag/chimp-listen
SCIENCEONLINE
SCIENCE (ISSN 0036-8075) is published weekly on Friday, except the last
week in December, by the American Association for the Advancement of
Science, 1200 New York Avenue, NW, Washington, DC 20005. Periodicals Mail
postage (publication No. 484460) paid at Washington, DC, and additional mailing
offices. Copyright © 2011 by the American Association for the Advancement of
Science. The title SCIENCE is a registered trademark of the AAAS. Domestic individual
membership and subscription (51 issues): $149 ($74 allocated to subscription).
Domestic institutional subscription (51 issues): $990; Foreign postage extra: Mexico,
Caribbean (surface mail) $55; other countries (air assist delivery) $85. First class,
airmail, student, and emeritus rates on request. Canadian rates with GST available
upon request, GST #1254 88122. Publications Mail Agreement Number 1069624.
Printed in the U.S.A.
Change of address: Allow 4 weeks, giving old and new addresses and 8-digit account
number. Postmaster: Send change of address to AAAS, P.O. Box 96178, Washington,
DC 20090–6178. Single-copy sales: $10.00 current issue, $15.00 back issue prepaid
includes surface postage; bulk rates on request. Authorization to photocopy
material for internal or personal use under circumstances not falling within the fair
use provisions of the Copyright Act is granted by AAAS to libraries and other users
registered with the Copyright Clearance Center (CCC) Transactional Reporting Service,
provided that $25.00 per article is paid directly to CCC, 222 Rosewood Drive, Danvers,
MA 01923. The identification code for Science is 0036-8075. Science is indexed in the
Reader’s Guide to Periodical Literature and in several specialized indexes.
SCIENCESIGNALING
www.sciencesignaling.org
The Signal Transduction Knowledge Environment
5 July issue: http://scim.ag/ss070511
RESEARCH ARTICLE: Sequential Phosphorylation
of Smoothened Transduces Graded Hedgehog
Signaling
Y. Su et al.
It takes two kinases and two phosphatases acting
on a single membrane protein Smoothened to
interpret the Hedgehog morphogen gradient.
RESEARCH ARTICLE: Identification of a Lysosomal
Pathway That Modulates Glucocorticoid Signaling
and the Inflammatory Response
Y. He et al.
PODCAST
Y. He and A. M. VanHook
Inhibition of lysosome function promotes the
anti-inflammatory effects of glucocorticoid signaling.
PERSPECTIVE: Bacterial Scaffolds Assemble
Novel Higher-Order Complexes to Reengineer
Eukaryotic Cell Processes
C. F. Lesser and J. M. Leong
Bacterial effector proteins can act as novel scaffolds
for signaling proteins in infected mammalian cells.
SCIENCETRANSLATIONAL MEDICINE
www.sciencetranslationalmedicine.org
Integrating Medicine and Science
6 July issue: http://scim.ag/stm070611
COMMENTARY: Reengineering Translational
Science—The Time Is Right
F. S. Collins
The new National Center for Advancing Translational
Sciences aims to revamp the process of biomedical
research translation.
PERSPECTIVE: Development of Newborn
and Infant Vaccines
G. Sanchez-Schmitz and O. Levy
Insights into immune ontogeny will inform
translation of new vaccines that are safe
and effective for newborns and infants.
RESEARCH ARTICLE: Optimization of Dosing
for EGFR-Mutant Non–Small Cell Lung Cancer
with Evolutionary Cancer Modeling
J. Chmielecki et al.
Predictive models of EGFR-mutant tumor behavior
point to alternative drug dosing strategies to prevent
and treat acquired resistance.
RESEARCH ARTICLE: CXCL5 Mediates UVB
Irradiation–Induced Pain
J. M. Dawes et al.
The cytokine CXCL5 is a peripheral mediator of pain
induced by UVB irradiation to the skin.
RESEARCH ARTICLE: Deep Sequencing
of the Human TCRγ and TCRβ Repertoires
Suggests That TCRβ Rearranges After αβ
and γδ T Cell Commitment
A. M. Sherwood et al.
Deep sequencing provides new insights about T cell
receptor rearrangement in humans.
SCIENCECAREERS
www.sciencecareers.org/career_magazine
Free Career Resources for Scientists
A Nobel Prize–Winning Culture
V. Raper
Scientists at the MRC Laboratory of Molecular
Biology attribute the lab’s success to bold thinking,
secure funding, and a collaborative culture.
http://scim.ag/mrc_lmb_nobel
A Farmer’s Son Cultivates a Career
in Translational Research
C. Wald
Molecular geneticist Robert Kralovics traces
his character as a scientist to growing up
as a farmer’s son in a Communist regime.
http://scim.ag/kravlovics
SCIENCEPODCAST
www.sciencemag.org/multimedia/podcast
Free Weekly Show
On the 8 July Science Podcast: the cellular origin
of spinal cord scar tissue, neutralizing influenza,
the future of nuclear power in the U.S., and more.
SCIENCEINSIDER
news.sciencemag.org/scienceinsider
Science Policy News and Analysis
SCIENCESIGNALING
Lysosomes regulate glucocorticoid signaling.
C
R
E
D
I
T
:

E
R
I
C

H
U
D
S
O
N
/
L
A
B
O
R
A
T
O
R
Y

O
F

A
N
A
L
Y
T
I
C
A
L
,

C
E
L
L
U
L
A
R
,

A
N
D

M
O
L
E
C
U
L
A
R

M
I
C
R
O
S
C
O
P
Y
,

V
A
N

A
N
D
E
L

R
E
S
E
A
R
C
H

I
N
S
T
I
T
U
T
E
Published by AAAS
133
EDITED BY STELLA HURTLEY
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

M
A
R
K

G
A
R
L
I
C
K
/
U
N
I
V
E
R
S
I
T
Y

O
F

W
A
R
W
I
C
K
;

H
A
U
R
I

E
T

A
L
.
Continued on page 135
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011
Operating at the Limits
Control theory provides a mathematical basis for
engineering the dynamic behavior of a system by
using feedback. Design is constrained by trade-
offs between making the system efficient and ro-
bust while minimizing output error. Glycolysis is
a central metabolic pathway that consumes glu-
cose to generate adenosine triphosphate (ATP)
with two key enzymes that are feedback-inhibited
by ATP. Under certain conditions, glycolytic
intermediates in yeast oscillate, but the basis for
these oscillations has been unclear. Chandra et
al. (p. 187) have now applied control theory to
analysis of a minimal model of glycolysis and
show that the oscillations are a consequence
of operating at the hard limits of maximizing
robustness, while minimizing metabolic overhead
and enzyme complexity.
Volatile Moon Rocks
The presence of water in the Moon is a highly
debated question. Hauri et al. (p. 213, pub-
lished online 26 May) present measurements
of dissolved volatiles, including water, in melt
inclusions from a sample of ancient lunar
magma brought back by Apollo 17. Melt inclu-
sions are small pieces of molten rock that get
trapped in crystals that
grow in magma. Melt
inclusions thus preserve
their volatile content,
which would otherwise
evaporate during volca-
nic eruption. The results
suggest that parts of the
lunar interior are more
volatile-rich than prior
studies have indicated, and may be more akin to
Earth’s modern upper mantle. This is surpris-
ing because the Moon was thought to have lost
most of its volatile content immediately after
its formation, which was caused by a Mars-sized
object hitting Earth.
Turbulent Times
In 1883, Osborne Reynolds described the key
factors that influence the transition of a flowing
fluid from a smooth, laminar flow to a choppy,
chaotic, turbulent flow. Known as the Reynolds
number (Re), the ratio of inertial forces to
viscous forces is used to predict the change in
flow behavior at a critical value for a specific
flow geometry. In simple pipe flow, the onset of
turbulence has been estimated to occur at Re
values between 1900 and 2100, but it has not
been possible to pin down a critical transition
point. Avila et al. (p. 192; see the Perspective
by Eckhardt) examined the fate of water jet
puffs injected into a stream of flowing water.
At low Re, the puff eventually decayed, while
at high Re, a puff split into two by absorbing
energy from the flowing liquid. Finding the
point where the lifetime of a single puff reached
a maximum allowed the minimum Re required
to sustain turbulent flow to be determined.
Forbidden Territory
In the parlance of quantum mechanics, certain
transitions between specific energy states
are termed “forbidden.” These transitions,
however, do occur, but they are extremely
improbable. Using precise laser spectroscopy in
tandem with laser cooling, van Rooij et al.
(p. 196; see the Perspective by Eyler) mea-
sured the frequency of one such transition in
helium (from the lowest triplet state to the
second-lowest singlet state), relying on extend-
ed interaction times between the atoms and
light field, as well as highly sensitive detection,
to characterize the rare excitation events. This
frequency, in turn, afforded a stringent test of
the theoretical framework describing atomic
structure and light-matter interactions.
Caught in the Act
Under conditions where a structural phase
transition can occur, multiple phases of a mate-
rial coexist as it transforms from one equilib-
rium phase to the other. Zheng et al. (p. 206)
devised a method to study the transformation
between two solid phases in copper sulfide
nanoparticles. Extremely high-quality transmis-
sion electron microscopy and image analysis
were able to distinguish between the two phases
within the nanoparticle as the transition temper-
ature was approached. Regions were observed
to flip back and forth between the two phases,
which could be described in terms of simplified
thermodynamic fluctuation arguments.
A Different Route
to Phenols
Phenol derivatives are essential intermedi-
ates in the preparation of many commercial
organic compounds, such as drugs, pesticides,
pigments, and plastics. Traditionally, these in-
termediates are prepared through modification
of intact aromatic rings. Izawa et al. (p. 209,
published online 9 June) present an alterna-
tive strategy: palladium-catalyzed oxidation of
cyclohexanone derivatives (essentially hydroge-
nated precursors of the desired phenols), with
oxygen accepting the liberated hydrogen to
form water. The advantage of this approach is
that the cyclohexanone precursors are available
with substitution patterns that are difficult to
achieve selectively when using standard proto-
cols for direct aromatic ring elaboration.
Off With Her Head!
The Red Queen hypothesis suggests that co-
evolution results from the evolutionary race
between interacting species resulting in a
On 28 March 2011, the NASA
Swift satellite detected a high-
energy outburst that behaved
completely differently from the
classical gamma-ray bursts that
the telescope was designed to
study and detect. Levan et al. (p.
199, published online 16 June)
present comprehensive obser-
vations of this unusual event,
concluding that it occurred in the
nucleus of a galaxy with a redshift
of 0.35. Bloom et al. (p. 203,
published online 16 June) pre-
sent a physical model to explain
the origin of the burst: The tidal
disruption of a star as it passed
close to the galaxy’s central black
hole produced a relativistic jet
pointed toward Earth. Swift Encounter
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

This Week in Science
C
R
E
D
I
T
:

D
O
M
Í
N
G
U
E
Z
-
E
S
C
O
B
A
R

E
T

A
L
.
seemingly stationary situation. Morran et al. (p. 216; see the Perspective by Brockhurst) explore
the role of the Red Queen in the case of a host-pathogen system with the presence and absence of
outcrossing and coevolution. Caenorhabditis elegans nematodes were infected with the bacteria
Serratia marcescens. The populations of C. elegans were either wild-type (can cross with other in-
dividuals or self-fertilize), obligate outcrossers, or obligately self-fertilizing lines. Outcrossers had
lower mortality rates when infected than obligate selfers, even if infected by coevolving bacterial
strains. Wild-type populations initially showed higher levels of outcrossing in the presence of the
virulent bacteria, but reduced levels of outcrossing in later generations. Thus, selection imposed
by coevolving pathogens may account for the widespread prevalence of outcrossing in nature.
Cord Blood Delivers
Continuous blood-cell production during an organism’s life depends on rare, self-renewing hemato-
poietic stem cells (HSCs). The ability to characterize human HSCs has been limited because isolation
methods cannot distinguish HSCs from progenitors that only participate in transient hematopoietic
regeneration. Notta et al. (p. 218) now report isolation of a near-homogeneous population of HSCs,
from human umbilical cord blood and successful transplantation of a single HSC into immunodeficient
mice. These findings provide a method to assist in systematic studies of human HSC biology and the
development of approaches to harness the regenerative potential of HSCs for clinical applications, such
as transplantation.
Bacterial Elongation Mechanism
One of the defining concepts of recent research on bacterial cytoskeleton
organization and morphogenesis is the helical organization of actin-like MreB
proteins and the role of these helices in the spatial organization of cell wall
biosynthesis. Two papers now challenge
this picture. Domínguez-Escobar et al.
(p. 225, published online 2 June) and
Garner et al. (p. 222, published online
2 June) studied the dynamic interactions
between bacterial actins and the cell
wall elongation machinery in Bacillus
subtilis cells and found that MreB pro-
teins in actively growing cells did not form helical filamentous structures. Instead, the proteins formed
discrete patches that moved processively along peripheral tracks perpendicular to the cell axis.
Optineurin in Autophagic Bacterial Clearance
Autophagy receptors bind both ubiquitin and autophagy markers, including microtubule-associated
protein light chain 3 (LC3), and promote the specific clearance of protein aggregates, defective
organelles, and intracellular pathogens. Wild et al. (p. 228, published online 26 May) describe
optineurin (OPTN) as an autophagy receptor whose function is regulated by phosphorylation of its
LC3-interacting motif. Phosphorylation by the protein kinase Tank binding kinase 1 (TBK1) increased
the affinity of OPTN for autophagy modifiers by 13-fold. OPTN is also a ubiquitin-binding protein
and was recruited to cytosolic Salmonella to promote bacterial clearance via the autophagy pathway.
Thus, TBK1 and OPTN represent critical components of the cell defense system for restricting the
growth of bacteria in the cell.
Waste Not
Cancer-associated cachexia is a wasting syndrome characterized by uncontrolled loss of fat and
muscle mass, which kills about 15% of cancer patients. Studying two models of tumor-bearing
mice, Das et al. (p. 223, published online 16 June; see the Perspective by Arner) show that
adipose triglyceride lipase (ATGL), an enzyme that breaks down stored fat, is essential to the
pathogenesis of cancer-associated cachexia. Mutant mice that were genetically deficient in ATGL
were protected from cancer–associated cachexia; they retained a normal fat mass and showed
reduced loss of skeletal muscle. Pharmacological inhibition of ATGL may thus merit investigation
as a potential treatment for cachexia.
Continued from page 133
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011
17050 Montebello Road
Cupertino, California 95014
Email: AAASInfo@betchartexpeditions.com
www.betchartexpeditions.com
For a detailed brochure,
please call (800) 252-4910
All prices are per person twin share + air
Venice:
Roundthe
Lagoon
October18-30, 2011
Explore Trieste and Aquileia (the
second Rome), Venice, the Po Delta
andCalera Botanical Garden. Alsosee
the miniature Venice (Comacchio)
withsuperb leadership. $3,695 + air
Turkey:
Archaeology &Draconid
Meteor Shower
October 3-14, 2011
Discover the
cultural wonders
of Turkey during
the rain of falling
stars known as the
Draconid Meteor
Shower. See Istanbul, Cappadocia,
Ephesus, Pergamum, Assos, Troy
&more. $3,995 + air
BETCHART EXPEDITIONS Inc.
To meet the challenge of the compe-
titive economy in the newmillennium,
AAAS started Entry Point!, a program
that offers students with disabilities
competitive internship opportunities
in science, engineering, mathematics,
computer science, and some fields of
business. Join us. Together we can
make a difference.
To learn more, visit:
aaas.org/plusyou/entrypoint
Summer Internships
Students with Disabilities
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

A Threat to Medical Innovation
THESE ARE DIFFICULT TIMES FOR THE U.S. NATIONAL INSTITUTES OF HEALTH (NIH). ITS 2011 BUDGET
is 1% less than in 2010, and it is increasingly common for NIH to administratively decrease
the size of awards across the board, before money goes out the door. The federal stimulus pack-
age buffered the impact of NIH budget issues. For example, approximately 36% of NIH grant
awards derived from the stimulus in FY 2009 were individual research awards (R01s).* But
the prospects moving forward are bleak, given the absence of stimulus funds and the build-
ing momentum for cutting federal discretionary spending. There have been 5 years of no real
growth in the NIH budget, explaining in part why the number of grants awarded for research
ideas originated by individual scientists (untargeted R01s) has been stagnant. It also explains
why paylines (the percent of applications funded) are at or near record lows. NIH Director
Francis Collins recently testified before a Senate subcommittee that in FY 2011 only 17
to 18% of grant applications would be funded, the lowest level on
record. Because collective funding efforts, such as program project
and center grants, as well as targeted R01s (responses to specific pro-
gram announcements from individual NIH institutes), also compete
with the bottom-up R01 program for funding, innovative investiga-
tor-initiated research is in dire straits. This critical part of the NIH
research effort forms the basis for future medical progress and must
be returned to good health.
Roy Vagelos, the former CEO of Merck, is highly critical of new
NIH efforts to put increased resources into translational and applied
research efforts, saying that NIH should stick to supporting new knowl-
edge and discoveries.‡ Note that the most important breakthroughs
often come from unexpected areas of inquiry. For example, recom-
binant DNA and monoclonal antibodies emerged from fundamental
research in bacteriology and immunology, respectively. These tech-
nologies gave birth to the biotechnology industry and underlie many therapeutics approved by
the U.S. Food and Drug Administration. A more recent example is RNA interference, which
came from research on plants and worms. The key concept is that we do not know from which
life science discipline or even organism the next great medical advance will emerge.
A “fund people, not projects” strategy is already practiced by the Howard Hughes Medical
Institute. The Wellcome Trust and individual European Research Council grants are moving
in this direction. NIH programs, including the Pioneer Awards and New Innovator Awards,
recognize innovation by outstanding scientists, but these programs make up a small frac-
tion of total NIH funding. For example, only 17 Pioneer Awards were awarded in FY 2010.
Dramatic movement in the “fund people, not projects” direction throughout the NIH will be
essential to promote true innovation.
With 10 to 15% paylines at some institutes (or even less), the current situation makes grant
evaluation nearly impossible and is putting truly excellent laboratories out of business. In the
spirit of “never waste a good crisis,” a serious evaluation of many NIH extramural policies and
programs is warranted. They include centers and other large collective funding efforts as well as
expensive clinical and epidemiological research. Although long-standing constituencies make
it hard to consider ending or even reducing these programs, their cost/benefit ratios should be
honestly examined. In addition, the NIH intramural research program receives approximately
10% of NIH dollars without being subject to the same level of competitive merit review as
the rest of NIH-supported research. Moving forward in an era of decreasing real budgets will
require hard work as well as the courage to see and tell the truth, the same qualities required for
innovative research contributions.
C
R
E
D
I
T
S
:
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 136
EDITORIAL
C
R
E
D
I
T
S
:

(
T
O
P
)

B
R
A
N
D
E
I
S

U
N
I
V
E
R
S
I
T
Y
;

(
L
E
F
T
)

T
O
D
D

D
A
V
I
D
S
O
N
/
A
L
A
M
Y
10.1126/science.1210374
– Michael Rosbash
Michael Rosbash is
an Investigator of
the Howard Hughes
Medical Institute
and a professor of
Biology at Brandeis
University. E-mail:
rosbash@brandeis.edu.
*NIH Recovery Act Grant Awards: Fiscal Year 2009 spreadsheet (http://report.nih.gov/recovery).
http://report.nih.gov/NIHDatabook/Charts/Default.aspx?showm=Y&chartId=34&catId=2.
‡http://blogs.wsj.com/health/2011/04/15/former-merck-head-vagelos-says-nih-should-stick-to-basic-research.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

137
EDITORS’CHOICE
EDITED BY KRISTEN MUELLER AND JAKE YESTON
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

C
O
U
R
T
E
S
Y

O
F

T
H
E

C
A
R
N
E
G
I
E

I
N
S
T
I
T
U
T
I
O
N

F
O
R

S
C
I
E
N
C
E
;

N
A
S
A
Continued on page 138
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011
E C O L O G Y
Carnegie Blows Up Biodiversity
To assess the effects of anthropogenic changes on biodiversity, we need to know not
only what exists but what existed, Dolan contends in an analysis of historic plankton
samples collected on the final cruise of the ship Carnegie during 1928–1929. Built
for oceanography, on its last global voyage it was equipped to systematically collect
plankton. Plankton were obtained from 160 sampling stations. Three groups were
chosen for identification and counting to represent distinct trophic levels: small
copepod crustaceans, Ceratium dinoflagellate alga, and ciliate zooplankton called
tintinnids. Reanalysis of these collections revealed that changes in species richness
were correlated for the three groups, and more species were collected at the tropics
than at high latitudes. Interestingly, of the several hundred species, most were rare
and few were common. Although potentially a rich source of now-unfunded taxo-
nomic expertise, historical data do have gaps and study design issues that cannot
now be resolved. Sadly, the Carnegie’s adventure came to an end when it, and its
scientist, Captain Ault, were blown up while refueling in Samoa. — CA
J. Plankton Res. 33, 10.1093/plankt/fbr060 (2011).
D E V E L O P ME N T
Controlling Sex
Chromosomes are packaged into transcriptionally silent hetero-
chromatin and transcriptionally active euchromatin. The highly
conserved protein HP1a marks heterochromatin in Drosophila,
and heterochromatin-rich telomeres are capped by a protein
complex composed of HP1a and HP1/ORCAssociated Protein
(HOAP). Because knockdown of HP1a is associated with loss of
male viability, and the HP1a-HOAP complex shows similarity to
the mammalian sex-determining region of the Y chromosome
(SRY), Li et al. sought to investigate the role of this complex
in regulating sex determination in Drosophila. The authors
carried out gene expression analysis in HOAP-deficient flies
and found that the majority of down-regulated genes were
those associated with the testis. Further analyses showed that
this was due to repressive activity by HOAP and both repressive
and activating functions of HP1a that affected the function of
the establishment promoter of Sex-lethal, the master regulator
of sex determination. Flies mutant in HP1a or HOAP exhibited
defects in sex determination. Thus, proteins typically associated
with heterochromatin are critical for regulating the changes in
gene expression required for sex determination in flies. — LMZ
PLoS Genet. 7, e1002122 (2011).
C L I MAT E S C I E N C E
In Synch with the Weather
Cooling climates in Africa over the past 10 million years have
led to substantial environmental changes that might have
influenced human evolution. Demonstrating a specific relation,
however, has been difficult because the human fossil record
is sparse, and uncertainties in dating of both these fossils and
climate records make correlation at the necessary resolution
(within 100,000 years) problematic. To help tackle some of
these issues, Joordens et al. examined strontium isotopes from
fish fossils in sediments from Lake
Turkana that also contain human
fossils, focusing on an interval
around 2 million years ago. Rivers
feeding the lake drain rocks with
different Sr isotope compositions,
and thus the lake Sr chemistry
varied during monsoon wet and
dry periods. The record, which
extends over about 150,000 years,
implies that this variation primarily
reflects the precession of Earth’s
orbit, which varies over about
21,000 years. The sediments were
deposited during a well-known
magnetic field reversal, allowing an accurate time-climate
sequence to be constructed. The authors were able to place 12
hominid fossils into the finely calibrated sequence. Although
the age span is too brief to establish a larger relation between
climate change and human evolution, the approach could lead
to longer records here and elsewhere. — BH
Earth Planet. Sci. Lett. 307, 1 (2011).
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org
S T R U C T U R A L B I O L O G Y
Chromatin Regulation Unraveled
Histones are proteins that act to package DNA
into chromatin. The methylation of lysine 4 on
histone H3 is correlated with active gene expres-
sion in eukaryotes. In mammals, H3K4 is methyl-
ated by the MLL family of histone methyl transfer-
ases (HMTs). The catalytic subunits all share a SET
domain, but full activity requires the presence of
other complex components. One core component
is Ash2L, which comprises a zinc-finger motif and
a carboxy-terminal SP1a and ryanodine receptor
(SPRY) domain. Chen et al. and Sarvan et al.
have determined the structure of the N-terminal
domain of human Ash2L. Both found that besides
a zinc finger, the domain includes a winged-helix
motif, a DNA binding motif that is often involved
in transcription regulation. The zinc-finger
domain was previously proposed to bind histone
tails; however, in the context of the structure, it
lacks features required for such binding. Sarvan
et al. show that the Ash2L winged-helix domain
is required for binding to the β-globin locus
control region, H3K4 methylation, and maximal
expression of the β-globin gene, whereas Chen et
al. show that it binds preferentially to an active
chromatin domain in a Homeobox gene locus and
suggest that it may play a role in chromosome
demarcation. — V V
EMBO Rep. 12, 10.1038/embor.2011.101 (2011);
Nat. Struct. Mol. Biol. 18, 10.1038/nsmb.2093 (2011).
D E V E L O P ME N T
Feedback Forms Frogs
Signaling by gradients of the morphogen BMP
(bone morphogenetic protein) controls the
contributions of individual cells to the develop-
ing embryo. Genes encoding components of the
BMP signaling pathway are expressed together in
groups that include both components that favor
BMP signaling and also feedback inhibitors of
the pathway. Paulsen et al. used a combination
of experiments and mathematical modeling to
explore what advantage this organization of the
BMP signaling module might confer. Human
embryonic kidney cells responded to a range of
BMP concentrations that varied almost 100-fold.
But when a feedback inhibitor was removed, the
dynamic range of response was reduced by a fac-
tor of 10. Another important role of the feedback
inhibition was to reduce cell-to-cell variation
in the response to BMP. Indeed, in Xenopus
tadpoles, loss of feedback inhibitors led to a
several-fold increase in variability in tail length
during development. Thus, this developmental
regulatory system is optimized to promote a
robust, relatively constant phenotype in the face
of genetic, environmental, and stochastic varia-
tion or noise. — LBR
Proc. Natl. Acad. Sci. U.S.A. 108, 10202 (2011).
P H Y S I C S
The Pulling Power of Pairs
The functionality and characteristics of many
semiconductor devices depend crucially on the
movement of electron-hole pairs in response
to an applied electric field. In such ambipolar
devices, it is typically the heavier holes that
dominate the transport properties. As such, the
packets of charged pairs move in the direction of
the holes. Some experiments, however, exhibit
behavior opposite to that expected motion and
have been interpreted in terms of a correla-
tion field, or friction, between the electrons
and holes. Yang et al. use a transient grating
spectroscopy technique to probe the electron and
hole pair packets as they drift and diffuse in a
single quantum-well structure. An electron-hole
density grating is imprinted in the quantum well
by two interfering laser beams. Monitoring the
diffraction of a probe beam through the grating
as it evolves and decays provides a detailed
picture of the electrons’ and holes’ dynamics.
With a model, the authors are then able to
determine the degree of friction between them.
The technique should provide a route to better
understanding the operation of such ambipolar
devices and, from the polarization dependence,
the operation of spintronic devices based on the
spin properties of the carriers. — ISO
Phys. Rev. Lett. 106, 247401 (2011).
C H E MI S T R Y
Solvated Past the Finish Line
As concerns mount about the adverse impact of
atmospheric CO
2
on climate, there is increasing
interest in diverting some of the greenhouse gas
toward use as a feedstock for the industrial prepa-
ration of commodity chemicals. One promising
reaction in this vein is hydrogenation to formic
acid (HCOOH). This process is enthalpically favor-
able, but the entropic penalty for turning two
gases into one liquid molecule pushes the overall
equilibrium back toward the reactant side. Amines
can deliver an enthalpic kick by deprotonating
the acid. Schaub and Paciello found, however,
that when trihexylamine is used as a base for ease
of product isolation, the kick isn’t quite vigorous
enough. Adding a diol solvent inches the reaction
over the line to thermodynamic favorability, pre-
sumably by stabilizing the ionic products through
hydrogen bonding. — JSY
Angew. Chem. Int. Ed. 50, 10.1002/
anie.201101292 (2011).
EDITORS’CHOICE
Continued from page 137
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

139 8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org
www.sciencemag.org
SENIOR EDITORIAL BOARD
Cori Bargmann, The Rockefeller Univ.
John I. Brauman, Chair, Stanford Univ.
Richard Losick, Harvard Univ.
Michael S. Turner, University of Chicago
BOARD OF REVIEWING EDITORS
Adriano Aguzzi, Univ. Hospital Zürich
Takuzo Aida, Univ. of Tokyo
Sonia Altizer, Univ. of Georgia
Sebastian Amigorena, Institut Curie
Angelika Amon, MIT
Kathryn Anderson, Memorial Sloan-Kettering Cancer Center
Siv G. E. Andersson, Uppsala Univ.
Peter Andolfatto, Princeton Univ.
Meinrat O. Andreae, Max Planck Inst., Mainz
John A. Bargh, Yale Univ.
Ben Barres, Stanford Medical School
Marisa Bartolomei, Univ. of Penn. School of Med.
Jordi Bascompte, Estación Biológica de Doñana, CSIC
Facundo Batista, London Research Inst.
Ray H. Baughman, Univ. of Texas, Dallas
David Baum, Univ. of Wisconsin
Yasmine Belkaid, NIAID, NIH
Philip Benfey, Duke Univ.
Stephen J. Benkovic, Penn State Univ.
Gregory C. Beroza, Stanford Univ.
Peer Bork, EMBL
Bernard Bourdon, Ecole Normale Superieure de Lyon
Ian Boyd, Univ. of St. Andrews
Paul M. Brakefield, Univ. of Cambridge
Christian Büchel, Universitätsklinikum Hamburg-Eppendorf
Joseph A. Burns, Cornell Univ.
William P. Butz, Population Reference Bureau
Gyorgy Buzsaki, Rutgers Univ.
Mats Carlsson, Univ. of Oslo
Mildred Cho, Stanford Univ.
David Clapham, Children’s Hospital, Boston
David Clary, Univ. of Oxford
J. M. Claverie, CNRS, Marseille
Jonathan D. Cohen, Princeton Univ.
Alan Cowman, Walter & Eliza Hall Inst.
Robert H. Crabtree, Yale Univ.
Wolfgang Cramer, Potsdam Inst. for Climate Impact Research
F. Fleming Crim, Univ. of Wisconsin
Jeff L. Dangl, Univ. of North Carolina
Tom Daniel, Univ. of Washington
Stanislas Dehaene, Collège de France
Emmanouil T. Dermitzakis, Univ. of Geneva Medical School
Robert Desimone, MIT
Claude Desplan, New York Univ.
Ap Dijksterhuis, Radboud Univ. of Nijmegen
Dennis Discher, Univ. of Pennsylvania
Scott C. Doney, Woods Hole Oceanographic Inst.
Jennifer A. Doudna, Univ. of California, Berkeley
Julian Downward, Cancer Research UK
Bruce Dunn, Univ. of California, Los Angeles
Christopher Dye, WHO
David Ehrhardt, Carnegie Inst. of Washington
Michael B. Elowitz, Calif. Inst. of Technology
Tim Elston, Univ. of North Carolina at Chapel Hill
Gerhard Ertl, Fritz-Haber-Institut, Berlin
Barry Everitt, Univ. of Cambridge
Paul G. Falkowski, Rutgers Univ.
Ernst Fehr, Univ. of Zurich
Tom Fenchel, Univ. of Copenhagen
Alain Fischer, INSERM
Wulfram Gerstner, EPFL Lausanne
Karl-Heinz Glassmeier, For Geophysics & Extraterrestrial
Physics, TU Braunchweig
Diane Griffin, Johns Hopkins Bloomberg School of
Public Health
Elizabeth Grove, Univ. of Chicago
Taekjip Ha, Univ. of Illinois at Urbana-Champaign
Christian Haass, Ludwig Maximilians Univ.
Steven Hahn, Fred Hutchinson Cancer Research Center
Gregory J. Hannon, Cold Spring Harbor Lab.
Martin Heimann, Max Planck Inst., Jena
Isaac Held, NOAA
James A. Hendler, Rensselaer Polytechnic Inst.
Janet G. Hering, Swiss Fed. Inst. of Aquatic
Science & Technology
Ray Hilborn, Univ. of Washington
Michael E. Himmel, National Renewable Energy Lab.
Kai-Uwe Hinrichs, Univ. of Bremen
Kei Hirose, Tokyo Inst. of Technology
David Hodell, Univ. of Cambridge
Ove Hoegh-Guldberg, Univ. of Queensland
David Holden, Imperial College
Lora Hooper, UT Southwestern Medical Ctr at Dallas
Jeffrey A. Hubbell, EPFL Lausanne
Steven Jacobsen, Univ. of California, Los Angeles
Kai Johnsson, EPFL Lausanne
Peter Jonas, Universität Freiburg
William Kaelin, Dana-Farber Cancer Inst.
Barbara B. Kahn, Harvard Medical School
Daniel Kahne, Harvard Univ.
Bernhard Keimer, Max Planck Inst., Stuttgart
Joel Kingsolver, Univ. of North Carolina at Chapel Hill
Robert Kingston, Harvard Medical School
Alberto R. Kornblihtt, Univ. of Buenos Aires
Leonid Kruglyak, Princeton Univ.
Mitchell A. Lazar, Univ. of Pennsylvania
David Lazer, Harvard Univ.
Virginia Lee, Univ. of Pennsylvania
Ottoline Leyser, Cambridge Univ.
Olle Lindvall, Univ. Hospital, Lund
Marcia C. Linn, Univ. of California, Berkeley
John Lis, Cornell Univ.
Jianguo Liu, Michigan State Univ.
Richard Losick, Harvard Univ.
Jonathan Losos, Harvard Univ.
Ke Lu, Chinese Acad. of Sciences
Laura Machesky, CRUK Beatson Inst. for Cancer Research
Andrew P. MacKenzie, Univ. of St Andrews
Anne Magurran, Univ. of St Andrews
Oscar Marin, CSIC & Univ. Miguel Hernández
Charles Marshall, Univ. of California, Berkeley
Martin M. Matzuk, Baylor College of Medicine
Grahma Medley, Univ. of Warwick
Yasushi Miyashita, Univ. of Tokyo
Richard Morris, Univ. of Edinburgh
Edvard Moser, Norwegian Univ. of Science and Technology
Sean Munro, MRC Lab. of Molecular Biology
Naoto Nagaosa, Univ. of Tokyo
James Nelson, Stanford Univ. School of Med.
Timothy W. Nilsen, Case Western Reserve Univ.
Pär Nordlund, Karolinska Inst.
Helga Nowotny, European Research Advisory Board
Luke O'Neill, Trinity College, Dublin
Stuart H. Orkin, Dana-Farber Cancer Inst.
Christine Ortiz, MIT
Elinor Ostrom, Indiana Univ.
Andrew Oswald, Univ. of Warwick
Jane Parker, Max-Planck Inst. of Plant Breeding Research
Donald R. Paul, Univ. of Texas at Austin
P. David Pearson, Univ. of California, Berkeley
Reginald M. Penner, Univ. of California, Irvine
John H. J. Petrini, Memorial Sloan-Kettering Cancer Center
Simon Phillpot, Univ. of Florida
Philippe Poulin, CNRS
Colin Renfrew, Univ. of Cambridge
Trevor Robbins, Univ. of Cambridge
Barbara A. Romanowicz, Univ. of California, Berkeley
Jens Rostrup-Nielsen, Haldor Topsoe
Edward M. Rubin, Lawrence Berkeley National Lab
Mike Ryan, Univ. of Texas, Austin
Shimon Sakaguchi, Kyoto Univ.
Miquel Salmeron, Lawrence Berkeley National Lab
Jürgen Sandkühler, Medical Univ. of Vienna
Randy Seeley, Univ. of Cincinnati
Christine Seidman, Harvard Medical School
Vladimir Shalaev, Purdue Univ.
Joseph Silk, Univ. of Oxford
Denis Simon, Univ. of Oregon
Alison Smith, John Innes Centre
Davor Solter, Inst. of Medical Biology, Singapore
John Speakman, Univ. of Aberdeen
Allan C. Spradling, Carnegie Institution of Washington
Jonathan Sprent, Garvan Inst. of Medical Research
Elsbeth Stern, ETH Zürich
Ira Tabas, Columbia Univ.
Yoshiko Takahashi, Nara Inst. of Science and Technology
John Thomas, Duke Univ.
Herbert Virgin, Washington Univ.
Bert Vogelstein, Johns Hopkins Univ.
Cynthia Volkert, Univ. of Gottingen
Bruce D. Walker, Harvard Medical School
Douglas Wallace, Leibniz Inst. of Marine Sciences
Ian Walmsley, Univ. of Oxford
David A. Wardle, Swedish Univ. of Agric Sciences
Detlef Weigel, Max Planck Inst., Tübingen
Jonathan Weissman, Univ. of California, San Francisco
Sue Wessler, Univ. of California, Riverside
Ian A. Wilson, The Scripps Res. Inst.
Timothy D. Wilson, Univ. of Virginia
Jan Zaanen, Leiden Univ.
Mayana Zatz, University of Sao Paolo
Jonathan Zehr, Ocean Sciences
Huda Zoghbi, Baylor College of Medicine
Maria Zuber, MIT
BOOK REVIEW BOARD
John Aldrich, Duke Univ.
David Bloom, Harvard Univ.
Angela Creager, Princeton Univ.
Richard Shweder, Univ. of Chicago
Ed Wasserman, DuPont
Lewis Wolpert, Univ. College London
1200 New York Avenue, NW
Washington, DC 20005
Editorial: 202-326-6550, FAX 202-289-7562
News: 202-326-6581, FAX 202-371-9227
Bateman House, 82-88 Hills Road
Cambridge, UK CB2 1LQ
+44 (0) 1223 326500, FAX +44 (0) 1223 326501
SUBSCRIPTION SERVICES For change of address, missing issues, new
orders and renewals, and payment questions: 866-434-AAAS (2227)
or 202-326-6417, FAX 202-842-1065. Mailing addresses: AAAS, P.O.
Box 96178, Washington, DC 20090-6178 or AAAS Member Services,
1200 New York Avenue, NW, Washington, DC 20005
INSTITUTIONAL SITE LICENSES please call 202-326-6755 for any
questions or information
REPRINTS: Author Inquiries 800-635-7181
Commercial Inquiries 803-359-4578
PERMISSIONS 202-326-7074, FAX 202-682-0816
MEMBER BENEFITS AAAS/Barnes&Noble.com bookstore www.aaas.org/bn;
AAAS Online Store www.apisource.com/aaas/ code MKB6; AAAS
Travels: Betchart Expeditions 800-252-4910; Apple Store www.
apple.com/eppstore/aaas; Bank of America MasterCard 1-800-833-6262
priority code FAA3YU; Cold Spring Harbor Laboratory Press
Publications www.cshlpress.com/affiliates/aaas.htm; GEICO Auto
Insurance www.geico.com/landingpage/go51.htm?logo=17624;
Hertz 800-654-2200 CDP#343457; Office Depot https://bsd.
officedepot.com/portalLogin.do; Seabury & Smith Life Insurance 800-
424-9883; Subaru VIP Program 202-326-6417; VIP Moving Services
www.vipmayflower.com/domestic/index.html; Other Benefits: AAAS
Member Services 202-326-6417 or www.aaasmember.org.
science_editors@aaas.org (for general editorial queries)
science_letters@aaas.org (for queries about letters)
science_reviews@aaas.org (for returning manuscript reviews)
science_bookrevs@aaas.org (for book review queries)
Published by the American Association for the Advancement of Science
(AAAS), Science serves its readers as a forum for the presentation and
discussion of important issues related to the advancement of science,
including the presentation of minority or conflicting points of view,
rather than by publishing only material on which a consensus has been
reached. Accordingly, all articles published in Science—including
editorials, news and comment, and book reviews—are signed and reflect
the individual views of the authors and not official points of view adopted
by AAAS or the institutions with which the authors are affiliated.
AAAS was founded in 1848 and incorporated in 1874. Its mission is to
advance science, engineering, and innovation throughout the world for
the benefit of all people. The goals of the association are to: enhance
communication among scientists, engineers, and the public; promote and
defend the integrity of science and its use; strengthen support for the
science and technology enterprise; provide a voice for science on societal
issues; promote the responsible use of science in public policy; strengthen
and diversify the science and technology workforce; foster education in
science and technology for everyone; increase public engagement with
science and technology; and advance international cooperation in science.
INFORMATION FOR AUTHORS
See pages 784 and 785 of the 11 February 2011 issue or
access www.sciencemag.org/about/authors
EDITOR-IN-CHIEF Bruce Alberts
EXECUTIVE EDITOR NEWS EDITOR
Monica M. Bradford Colin Norman
MANAGING EDITOR, RESEARCH JOURNALS Katrina L. Kelner
DEPUTY EDITORS R. Brooks Hanson, Barbara R. Jasny, Andrew
M. Sugden
EDITORIAL SENIOR EDITORS/COMMENTARY Lisa D. Chong, Brad Wible; SENIOR
EDITORS Gilbert J. Chin, Pamela J. Hines, Paula A. Kiberstis (Boston), Marc
S. Lavine (Toronto), Beverly A. Purnell, L. Bryan Ray, Guy Riddihough,
H. Jesse Smith, Phillip D. Szuromi (Tennessee), Valda Vinson, Jake
S. Yeston, Laura M. Zahn (San Diego); ASSOCIATE EDITORS Kristen L.
Mueller, Jelena Stajic, Sacha Vignieri, Nicholas S. Wigginton; BOOK REVIEW
EDITOR Sherman J. Suter; ASSOCIATE LETTERS EDITOR Jennifer Sills; EDITORIAL
MANAGER Cara Tate; SENIOR COPY EDITORS Jeffrey E. Cook, Cynthia Howe,
Harry Jach, Lauren Kmec, Barbara P. Ordway, Trista Wagoner; COPY EDITOR
Chris Filiatreau; SENIOR EDITORIAL COORDINATORS Carolyn Kyle, Beverly
Shields; EDITORIAL COORDINATORS Joi S. Granger, Anita Wynn; PUBLICATIONS
ASSISTANTS Ramatoulaye Diop, Emily Guise, Jeffrey Hearn, Michael Hicks,
Lisa Johnson, Scott Miller, Jerry Richardson, Brian White; EDITORIAL
ASSISTANT Patricia M. Moore; EXECUTIVE EDITORIAL ASSISTANT Yolanda O'Bannon
(San Francisco); EXECUTIVE ASSISTANT Alison Crawford; ADMINISTRATIVE SUPPORT
Maryrose Madrid; EDITORIAL FELLOW Melissa R. McCartney
EDITORIAL DIRECTOR, WEB AND NEW MEDIA Stewart Wills; SENIOR WEB EDITOR
Tara S. Marathe; RESEARCH ASSOCIATE Corinna Cohn; WEB DEVELOPMENT
MANAGER Martyn Green; WEB DEVELOPER Andrew Whitesell; INTERNS
Andrew Green, Kerry Klein
NEWS DEPUTY NEWS EDITORS Robert Coontz, David Grimm (Online), Eliot
Marshall, Jeffrey Mervis, Leslie Roberts, John Travis; CONTRIBUTING EDITORS
Elizabeth Culotta, Polly Shulman; NEWS WRITERS Yudhijit Bhattacharjee,
Adrian Cho, Jennifer Couzin-Frankel, Carolyn Gramling, Jocelyn Kaiser,
Richard A. Kerr, Eli Kintisch, Greg Miller, Elizabeth Pennisi,
Sara Reardon, Robert F. Service (Pacific NW), Erik Stokstad; WEB DEVELOPER
Daniel Berger; INTERNS Daniel Strain, Yasmin Ogale, Natalie Villacorta;
CONTRIBUTING CORRESPONDENTS Jon Cohen (San Diego, CA), Daniel
Ferber, Ann Gibbons, Sam Kean, Andrew Lawler, Mitch Leslie,
Charles C. Mann, Virginia Morell, Gary Taubes; COPY EDITORS Linda
B. Felaco, Melvin Gatling, Melissa Raimondi; ADMINISTRATIVE SUPPORT
Scherraine Mack; BUREAUS San Diego, CA: 760-942-3252, FAX 760-942-
4979; Pacific Northwest: 503-963-1940
PRODUCTION DIRECTOR Wendy K. Shank; ASSISTANT MANAGER Rebecca
Doshi; SENIOR SPECIALISTS Steve Forrester, Chris Redwood, Anthony
Rosen; PREFLIGHT DIRECTOR David M. Tompkins; MANAGER Marcus Spiegler;
SPECIALISTS Jason Hillman, Tara Kelly
ART DIRECTOR Yael Fitzpatrick; ASSOCIATE ART DIRECTOR Laura Creveling;
SENIOR ILLUSTRATORS Chris Bickel, Katharine Sutliff; ILLUSTRATOR Yana
Hammond; SENIOR ART ASSOCIATES Holly Bishop, Preston Huey,
Nayomi Kevitiyagala, Matthew Twombly; ART ASSOCIATE Kay Engman;
PHOTO EDITOR Leslie Blizard
SCIENCE INTERNATIONAL
EUROPE (science@science-int.co.uk) EDITORIAL: INTERNATIONAL MANAGING
EDITOR Andrew M. Sugden; SENIOR EDITOR/COMMENTARY Julia Fahren-
kamp-Uppenbrink; SENIOR EDITORS Caroline Ash, Stella M. Hurtley,
Ian S. Osborne, Peter Stern; ASSOCIATE EDITOR Maria Cruz; LOCUM
EDITOR Helen Pickersgill; EDITORIAL SUPPORT Samantha Hogg, Alice Whaley;
ADMINISTRATIVE SUPPORT John Cannell, Janet Clements, Louise
Hartwell; NEWS: DEPUTY NEWS EDITOR, U.K. Daniel Clery; CONTRIBUTING
EDITOR, EUROPE Martin Enserink; CONTRIBUTING CORRESPONDENTS Michael
Balter (Paris), John Bohannon (Vienna), Gretchen Vogel (Berlin)
LATIN AMERICA CONTRIBUTING CORRESPONDENT Antonio Regalado
ASIA Japan Office: Asca Corporation, Tomoko Furusawa, Rustic Bldg. 7F,
77 Tenjin-cho, Shinjuku-ku, Tokyo 162-0808, Japan; +81 3
6802 4616, FAX +81 3 6802 4615, inquiry@sciencemag.jp;
ASIA NEWS EDITOR Richard Stone (Beijing: rstone@aaas.org); CONTRIBUTING
CORRESPONDENTS Dennis Normile [Japan: +81 (0) 3 3391 0630, FAX
+81 (0) 3 5936 3531; dnormile@gol.com]; Hao Xin [China:
cindyhao@gmail.com]; Mara Hvistendahl [China: mara@
marahvistendahl.com]; Pallava Bagla [South Asia: +91 (0) 11 2271
2896; pbagla@vsnl.com]
EXECUTIVE PUBLISHER Alan I. Leshner
PUBLISHER Beth Rosner
FULFILLMENT SYSTEMS AND OPERATIONS (membership@aaas.org); CUSTOMER
SERVICE SUPERVISOR Pat Butler; SPECIALISTS Latoya Casteel, LaVonda
Crawford, Vicki Linton, April Marshall; DATA ENTRY SUPERVISOR Cynthia
Johnson; SPECIALISTS Shirlene Hall, Tarrika Hill, William Jones
BUSINESS OPERATIONS AND ADMINISTRATION DIRECTOR Deborah Rivera-
Wienhold; BUSINESS SYSTEMS AND FINANCIAL ANALYSIS DIRECTOR Randy Yi;
MANAGER, FULFILLMENT SYSTEMS Frits Buningh; MANAGER, BUSINESS ANALYSIS Eric
Knott; MANAGER, BUSINESS OPERATIONS Jessica Tierney; BUSINESS ANALYSTS
Priti Pamnani, Celeste Troxler; Christine Wehrli; RIGHTS AND
PERMISSIONS: ADMINISTRATOR Emilie David; ASSOCIATE Elizabeth Sandler;
MARKETING DIRECTOR Ian King; MARKETING MANAGERS Allison Pritchard, Alison
Chandler, Julianne Wielga, Samantha Smith; MARKETING ASSOCIATES
Aimee Aponte, Mary Ellen Crowley; SENIOR MARKETING EXECUTIVE Jennifer
Reeves; DIRECTOR, SITE LICENSING Tom Ryan; DIRECTOR, CORPORATE RELATIONS
Eileen Bernadette Moran; SENIOR PUBLISHER RELATIONS SPECIALIST
Kiki Forsythe; PUBLISHER RELATIONS MANAGER Catherine Holland; PUBLISHER
RELATIONS, EASTERN REGION Phillip Smith; PUBLISHER RELATIONS, WESTERN REGION
Ryan Rexroth; CUSTOMER RELATIONS MANAGER Iquo Edim; CUSTOMER RELATIONS
COORDINATOR David Lee; MARKETING MANAGER Christina Schlecht; MARKETING
ASSOCIATES Laura Tutino, Chad Johnson; ELECTRONIC MEDIA: DIRECTOR Lizabeth
Harman; ASSISTANT MANAGER Lisa Stanford; SENIOR PRODUCTION SPECIALIST Ryan
Atkins; PRODUCTION SPECIALISTS Antoinette Hodal, Nichele Johnston, Kimberly
Oster; DIRECTOR, WEB AND NEW MEDIA Will Collins; PROJECT MANAGER Trista Snyder;
SENIOR PRODUCTION SPECIALIST Christopher Coleman; COMPUTER SPECIALISTS
Walter Jones, Kai Zhang
ADVERTISING DIRECTOR, WORLDWIDE AD SALES Bill Moran
COMMERCIAL EDITOR Sean Sanders: 202-326-6430
ASSISTANT COMMERCIAL EDITOR Tianna Hicklin 202-326-6463
PRODUCT ( sci ence_advert i si ng@aaas. org) ; MI DWEST Ri ck
Bongiovanni: 330-405-7080, FAX 330-405-7081; EAST COAST/
E. CANADA Laurie Faraday: 508-747-9395, FAX 617-507-8189;
WEST COAST/W. CANADA Lynne Stickrod: 415-931-9782, FAX 415-520-
6940; UK/EUROPE/ASIA Roger Goncalves: TEL/FAX +41 43 243 1358;
JAPAN ASCA Corporation, Makiko Hara: +81 (0) 3 6802 4616, FAX
+81 (0) 3 6802 4615; ads@sciencemag.jp; CHINA/TAIWAN Ruolei Wu:
+86 1367 1015 294 rwu@aaas.org
WORLDWIDE ASSOCIATE DIRECTOR OF SCIENCE CAREERS Tracy Holmes: +44 (0)
1223 326525, FAX +44 (0) 1223 326532
CLASSIFIED (advertise@sciencecareers.org); U.S.: MIDWEST/WEST COAST/
SOUTH CENTRAL/CANADA Tina Burks: 202-326-6577; EAST COAST/INDUSTRY
Elizabeth Early: 202-326-6578; SALES ADMINISTRATOR: Marci Gallun;
EUROPE/ROW SALES: Susanne Kharraz, Dan Pennington, Alex Palmer; SALES
ASSISTANT Lisa Patterson; JAPAN ASCA Corporation, Makiko Hara +81 (0)
3 6802 4616, FAX +81 (0) 3 6802 4615; careerads@sciencemag.jp;
CHINA/TAIWAN Ruolei Wu: +86 1367 1015 294 rwu@aaas.org;
ADVERTISING SUPPORT MANAGER Karen Foote: 202-326-6740; ADVERTISING
PRODUCTION OPERATIONS MANAGER Deborah Tompkins; SENIOR PRODUCTION
SPECIALIST/GRAPHIC DESIGNER Amy Hardcastle; PRODUCTION SPECIALIST
Yuse Lajiminmuhip; SENIOR TRAFFIC ASSOCIATE Christine Hall; SALES
COORDINATOR Shirley Young
AAAS BOARD OF DIRECTORS RETIRING PRESIDENT, CHAIR Alice Huang;
PRESIDENT Nina Fedoroff; PRESIDENT-ELECT William Press; TREASURER
David E. Shaw; CHIEF EXECUTIVE OFFICER Alan I. Leshner; BOARD Nancy
Knowlton, Stephen Mayo, Raymond Orbach, Julia M. Phillips, Sue V.
Rosser, David D. Sabatini, Inder Verma, Thomas A. Woolsey
Published by AAAS
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 140
NEWS OF THE WEEK
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

E
S
A
/
J
-
L
.
A
T
T
E
L
E
Y
N
;

R
I
E
D
I
N
G
E
R

P
H
I
L
I
P
P
E
/
M
A
X
P
P
P
/
L
A
N
D
O
V

South Korea 2
Korean Students Are Top
Digital Readers
South Korean teenagers beat out their peers
from 18 other countries in the first assess-
ment of digital literacy. The Paris-based Pro-
gramme for International Student Assess-
ment (PISA) added a computer component
to its 2009 reading test for 15-year-olds, and
Korea chalked up a score of 568, 31 points
above New Zealand and Australia and 49
points ahead of Japan. The average was
500. (The United Kingdom, Germany, and
the United States did not participate.) Girls
scored higher than boys in every country—
an average of 24 points overall—but the dif-
ference was less marked than in assessments
of print reading. One surprising finding:
Greater use of the computer at home didn’t
necessarily translate into a better score.
The digital assessment measured a
student’s ability to follow hyperlinks to
obtain information on, for example, how
to find a summer job. The test is part of a
plan to expand PISA beyond pencil-and-
paper assessments; future assessments will
include computer-based tests of problem-
solving skills.
Sacramento, California 3
State Universities Reeling
After More Cuts
California Governor Jerry Brown signed a
budget last week that will inflict yet another
round of steep cuts on the state’s already
cash-strapped university systems. The bud-
get will reduce payments for the next aca-
demic year by $650 million each to the
10-school University of California (UC) sys-
tem and to the 23 campuses that make up the
California State University. The UC system
has seen state funding drop from $15,020
per student in 2000–01 to $8220 in 2010–11,
revenue that the institutions have tried to
recoup by raising tuition substantially. Some
administrators say the cuts have already hurt
the university’s ability to recruit and retain
top researchers.
California is not alone in its fiscal pain.
This year neighboring Arizona cut its contri-
butions to its three state universities by
$170 million, or about 20%. Pennsylva-
nia State University will take a 19% cut. In
Michigan, state universities will see a 15%
drop for the 2011–12 academic year. To
make ends meet, university administrators
across the country will have to choose from
an unappetizing menu of layoffs, salary caps,
enrollment limits, and tuition hikes.
Paris 1
Shuffle Results in New
Research Minister
The arrest in New York City of former Inter-
national Monetary Fund (IMF) Manag-
ing Director Dominique Strauss-Kahn on
charges of sexual assault has had a ripple
effect in the French cabinet. As a result,
France has a new minister for higher educa-
tion and research.
After finance minister Christine Lagarde
replaced Strauss-Kahn as IMF chief on
29 June, Valérie Pécresse, who had held the
research post for 4 years, was appointed
junior budget
minister and
government
spokesperson.
Her place was
in turn filled
by Laurent
Wauquiez,
formerly in
charge of Euro-
pean affairs.
Wauquiez, who
at 36 is the
youngest mem-
ber of the cabi-
net, studied history, English, German, and
Arabic, and graduated from two of France’s
most elite higher education institutions, the
École Normale Supérieure and the École
Nationale d’Administration.
Bertrand Monthubert, national secre-
tary for higher education and research of
the opposition French Socialist Party, calls
Wauquiez a surprising choice. He will
have “a tough job quelling the research and
academic communities’ deep fears for the
future,” Monthubert says.
http://scim.ag/Wauquiez_minister
>The National Institutes of Health’s high-
profile initiative to diagnose mystery
illnesses has stopped taking new appli-
cations, likely for just a few months. The
program, started in 2008, has been
overwhelmed by some 5400 inquiries
and nearly 2000 full applications. It
has investigated around 350 cases and
has built up a backlog of about 100
enrolled patients.
NOTED
1
2
3
5
AROUND THE WORLD
Darmstadt, Germany 4
Scientists Reconnect With
Cluster Mission
The European Space Agency’s Cluster mis-
sion is back in the fold after controllers fixed
a serious glitch.
Cluster is comprised of four identical sat-
ellites that have been studying Earth’s space
environment and the solar wind since 2000.
They pirouette around each other as they
orbit, enabling scientists to map space in 3D.
Each craft carries 11 sensors, of which five
make up the Wave Experiment Consortium.
4
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 141
NEWS
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

P
H
O
T
O

B
Y

K
A
T
H
E
R
I
N
E

M
E
R
T
E
N

S
T
U
D
I
O
;

M
I
C
H
A
Ë
L

C
A
T
A
N
Z
A
R
I
T
I
/
W
I
K
I
M
E
D
I
A

C
O
M
M
O
N
S
;

U
.
S
.

F
I
S
H

A
N
D

W
I
L
D
L
I
F
E

S
E
R
V
I
C
E
THEY SAID IT
“ It’s like piranha. They sense
blood in the water.”
—Gene Lucas, University of
California, Santa Barbara’s executive
vice chancellor, said of other universities
recruiting UC scientists, in an interview
with the Los Angeles Times.
FINDINGS
for their social
frolicking and
impressive acro-
batic displays,
were hunted to
extinction in
these same waters
during the 19th
and 20th centu-
ries’ era of indus-
trial whaling. A
small popula-
tion managed to survive near remote, sub-
Antarctic islands south of New Zealand.
In recent years, a few dozen females found
their way back to the same bays their ances-
tors used for bearing their young. Normally,
such cultural knowledge is passed from
mother to daughter, the researchers say. But
the tradition had been lost, until these pio-
neering females began making the >>
NEWSMAKERS
A Final Plan for the Spotted Owl
The decline of the northern spotted owl in the
1980s led to dramatic restrictions on logging
in Oregon and Washington. But the popula-
tion has continued to shrink by about 3% a
year, and there may be as few as 7000 owls
left. On 30 June, the U.S. Fish and Wildlife
Service (FWS) released a final version of a
recovery plan for the owl. Topping the list is
protection of forest habitat, preventing for-
est fires, and dealing with the barred owl,
which has invaded the spotted owls’ terri-
tory. Federal officials will have to consult the
recovery plan whenever they consider major
actions that might impact the owl, such as
revising how they manage forests.
Controversy is far from over, however:
FWS is considering an experimental killing
of barred owls to see if that will benefit the
spotted owls. And, as a result of a lawsuit,
FWS will reexamine the amount and location
of critical habitat before 15 November.
Lonsdale, a 79-year-old
retired chemist in Bend,
Oregon, made a fortune
as the founder of a drug
development and research
company. He ran for the
U.S. Senate as a Democratic
nominee in the 1990s.
His award would provide $50,000 for
the best proposal to study the origin of life
and up to $2 million in potential research
funding. Applicants are invited to submit “a
cogent hypothesis for how life first arose,
including its plausible chemistry, and for
how primitive life could have evolved to
modern biological cells, including the pres-
ent genetic material and metabolism.” A
panel of prominent origin researchers—such
as Harvard Medical School Nobelist Jack
Szostak and NASA astrobiologist Chris
McKay—will review proposals starting
this month. An award, or multiple, will be
announced early next year.
http://scim.ag/Lonsdale_prize
Scientist-Entrepreneur Offers
Origin-of-Life Prize
Harry Lonsdale, a millionaire scientist
who is an avowed atheist, has announced
a $50,000 prize to promote research on
the origin of life. He hopes that research-
ers working on the question will eventu-
ally prove that life’s origins can be fully
explained by physical and chemical pro-
cesses without invoking a creator.
Right Whales Finally
Coming Home
After more than 100 years, right whales
have returned to their calving grounds in
New Zealand, an international team of sci-
entists reports. The 100-ton whales, known
But in March the five on satellite 3 failed to
respond to ground commands.
Several weeks of intense work traced the
problem to the simultaneous locking of all
five power switches to the instruments—
an eventuality that designers thought too
unlikely to consider. But Cluster’s control
team came up with a scheme to use a power
cable and its backup cable simultaneously—
normally forbidden in satellite operations.
During a tense day last month, controllers
sent up a series of commands that success-
fully flipped each of the power switches.
“We were lucky to get it back on,” says
deputy project scientists Arnaud Masson.
Team members are working to ensure that
something similar doesn’t happen again.
http://scim.ag/esa_cluster
Zagreb, Croatia 5
Medical Journal Editors
Resign Amid Acrimony
The two editors-in-chief of the Croatian
Medical Journal (CMJ) resigned on 28 June,
along with the majority of the journal’s edi-
torial board, in the latest episode of a decade-
long, bitter fight between the journal’s edi-
tors and its management board.
Founded in 1991, CMJ has the high-
est impact factor of any Croatian journal,
and it has won international plaudits. But
its editors have long battled the University
of Zagreb’s medical faculty, which, along
with three other medical schools, owns the
journal (Science, 18 April 2008, p. 304). The
current editors-in-chief, Ana Marušić of the
University of Split School of Medicine and
Ivan Damjanov of the University of Kansas
School of Medicine, say they threw in the
towel because “obstruction” by the manage-
ment board has made running the journal
virtually impossible. A proposed change
in the journal’s management structure may
have cost them their jobs anyway.
http://scim.ag/CMJ_resign
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

142 8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

I
N
S
T
I
T
U
T
E

O
F

H
E
R
I
T
A
G
E

R
E
S
T
O
R
A
T
I
O
N
,

P
O
L
Y
T
E
C
H
N
I
C

U
N
I
V
E
R
S
I
T
Y

O
F

V
A
L
E
N
C
I
A

(
2
)
;

C
O
U
R
T
E
S
Y

O
F

T
H
E

P
R
O
G
E
R
I
A

R
E
S
E
A
R
C
H

F
O
U
N
D
A
T
I
O
N
;

J
E
R
O
M
E

S
U
E
U
R
BY THE NUMBERS
12,000–24,000 Tons of plas-
tic that deep-dwelling North Pacific
fish ingest each year, based on find-
ings from the Scripps Environmental
Accumulation of Plastic Expedition,
known as SEAPLEX.
€80.2 billion Initial proposal
by the European Commission for
research and innovation funding in
2014–20 under the Horizon 2020
program (formerly called Frame-
work). That would be a 46% increase
over current spending.
50% Loss in the amount of land
suitable for cultivating premium
wine grapes in high-value areas
of northern California by 2040
because of global warming, accord-
ing to a projection in Environmental
Research Letters.
Random Sample
Miniature
Art Masters
Microbiologist Rosa María
Montes Estellés once infected
a church mural with bacteria. But it
was for a good cause: The bacteria ate
their way through 4 centuries of grime
encrusted on a mural at Santos Juanes
Church in Valencia, Spain, exposing the
underlying colors.
Bacteria are only the latest tool in the art restorer’s arsenal. Restorers use microabrasion,
burly bristles, and chemical washes to strip layers of pollution from buildings, statues, and
paintings. But each method has shortcomings: They can put the underlying artwork at risk or
poison workers, and they often require slow and painstaking manual labor. So in 2005, a group
of Italian art restorers tried a new tack: They bred bacteria to remove an obstinate layer of col-
lagen from the murals of Campo Santo di Pisa.
At Santos Juanes, the offending material was a crusty white mixture of salt, sulfates, nitrates,
and carbon, originating from centuries of rainwater mixing with deposits from nesting birds
and insects in the roof above the murals. Over time, the material slid downward, encrusting the
paintings, where it fermented together with atmospheric pollutants. Montes and colleagues at
the Polytechnic University of Valencia selected and “trained” a nitrogen-loving type of bacte-
ria, Pseudomonas stutzeri, to eat the noxious blend.
How to apply the bacteria was a challenge. After testing different materials, Montes and
biologist Pilar Bosch chose a gel that keeps the bacteria wet and alive but doesn’t sink into the
underlying paint. The team’s preliminary results will appear in a forthcoming issue of Arché.
Meanwhile, Montes hopes to develop more and better treatments customized for different
surfaces and pollutants. >>FINDINGS
journey once again. Reporting in Marine
Ecology Progress Series, the scientists—
from Canada, the United States, Australia,
and New Zealand—confirmed that some of
the females had migrated from the south-
ern islands to New Zealand by comparing
the DNA in tissue samples collected from
seven whales at both sites. Now that the tra-
dition has been restored, scientists expect
more whales to follow the pioneers.
Tiny Bug Makes a Riot
With Its Privates
The world’s loudest
animal relative to its
size has been revealed
to be a tiny bug with a
big organ. Specifically,
the water boatman,
Micronecta scholtzi,
rattles its penis along
grooves in its abdomen
to produce a chattering
song—that registers at 99.2 decibels—about
the volume of a loud orchestra when heard
from the front row. Scientists presenting
at the Society for Experimental Biology’s
annual conference in Glasgow recorded the
bug and analyzed its volume compared to
various other loud animals. Even though the
water boatman does its “singing” from the
bottom of rivers to attract mates, humans
walking along the riverbank can clearly hear
it. The area along its abdomen that the bug
uses to make the noise is about the width of
only a human hair, and researchers aren’t
sure exactly how it produces so loud a song.
http://scim.ag/Water_boatman
New Drug Hope for ‘Aging‘ Kids
A drug already approved to treat cancer and
prevent the rejection of transplanted organs
may be the next hope for children with pro-
geria, a rare disease that resembles acceler-
ated aging and typically kills those afflicted
by their teen years. In studies on cells from
those with the disorder, the drug, rapamy-
cin, promoted clearance of the mutant pro-
tein dubbed progerin and extended the cells’
survival, reports a group that includes
Francis Collins, director of the U.S.
National Institutes of
Health, in the 29 June
issue of Science Trans-
lational Medicine.
The buildup of
progerin inside the
nuclei of cells dis-
torts their shape and
somehow initiates a
diverse set of normally
age-related symptoms
such as as loss of hair,
brittle bones, stiff skin, and cardiovascu-
lar disease. Three drugs that seek to impair
the synthesis of progerin are already being
tested in about 50 afflicted children, but
rapamycin may offer an alternative, com-
plementary strategy, says Collins. Given
the short life span of someone with pro-
geria, researchers and phyicians are now
debating whether mouse studies are needed
before proceeding with a clinical trial of
an oral form of rapamycin that has shown
tolerable side effects in children.
http://scim.ag/agingkids
NEWS OF THE WEEK
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

NEWS & ANALYSIS
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 143
NEWS & ANALYSIS
C
R
E
D
I
T
S
:

G
E
T
T
Y
/
B
L
O
O
M
B
E
R
G
;

J
B

R
E
E
D
,

G
E
T
T
Y
/
B
L
O
O
M
B
E
R
G

(
I
N
S
E
T
)
SILVER SPRING, MARYLAND—A unanimous
vote last week by advisers to the U.S. Food
and Drug Administration (FDA) is likely to
sharply curtail access in America to the breast
cancer drug Avastin, exposing deep rifts
among patient advocates, researchers, and
regulators over what kind of scientific back-
ing a drug needs to remain in use.
Avastin was approved for breast can-
cer in early 2008 under FDA’s accelerated
approval process. The process was designed
to get promising drugs to very sick patients
quickly, but FDA acknowledges that those
drugs also run the risk of later failing to live
up to their promise. That’s what happened
with Avastin. Two large follow-up studies
with about 2000 women showed only a sliver
of benefit compared with earlier, more com-
pelling results. And all the studies recorded
serious health risks, including heart prob-
lems and fatal gastrointestinal perforations.
In December, FDA announced that it was
revoking Avastin’s breast cancer approval, a
rare but not unprecedented move for a fast-
tracked drug. What was unprecedented was
what came next: protests by Avastin’s maker,
the California company Genentech, which
sought a reversal.
Last week’s meeting of FDA’s oncologic
drug advisory committee was the culmination
of the appeal, and it didn’t go well for Genen-
tech. The company argued that few treatments
are available to women with metastatic breast
cancer and that Avastin’s performance could
vary depending on the chemo therapy drug
with which it’s paired. FDA officials were
skeptical. Avastin’s potency against breast
cancer, they said, simply hasn’t held up.
Abigail Brandel, an FDA attorney, explained
at the hearing that when it comes to acceler-
ated approvals, “it is entirely possible that in
some cases clinical benefit will not be veri-
fied.” FDA then has the legal right to withdraw
the drug, she said.
FDA has yet to make a final decision on
revoking Avastin’s approval for metastatic
breast cancer, although it seems likely to;
either way, Avastin will stay on the market in
the United States for several other forms of
cancer. But if it loses its breast cancer label,
insurance companies may decline to cover
it for that use. Few patients can afford the
approximately $90,000 a year Avastin costs.
At the hearing, patients pleaded to have
Avastin remain available. “It is morally and
ethically wrong” to take this drug away, said
Crystal Hanna, a 36-year-old breast cancer
patient. She was joined by dozens of others,
many describing how Avastin had helped
them. Although some oncologists back them,
others think it’s all too easy to attribute relative
good health to a drug when it’s not clear the
two are related.
“The FDA’s role is not to look at those
individual stories, it’s to look at science,”
says Fran Visco, president of the National
Breast Cancer Coalition in Washing-
ton, D.C., and a breast cancer sur-
vivor herself. She’s grateful for
the 6-0 anti-Avastin vote by the
committee and dismayed that so
many breast cancer patients are
fighting for this drug.
Breast cancer studies of Avas-
tin, including the one that led to its
approval, have shown no advantage
in survival and no improved qual-
ity of life. There were indications that
tumors were less likely to grow for up
to 3 months. The study on which Avastin
was approved showed a 5½-month pause
in tumor growth, which many hoped would
translate to better survival. It did not. When
a colleague of Visco’s spoke about Avastin’s
flaws at the hearing, she was booed and had
to be escorted to her car by security. “We’ve
lowered the bar so much” in terms of what’s
expected from cancer drugs, Visco says.
Like several drugs on the market, Avas-
tin is designed to choke off a tumor’s blood
supply. This strategy, says Robert Kerbel,
a cancer biologist at Sunnybrook Research
Institute at the University of Toronto in Can-
ada, was based on the theory that all solid
tumors—like those in the breast, lung, and
colon—need to stimulate new blood ves-
sels to grow, a process called angiogenesis.
Now, “that whole theory is undergoing a lot
of reexamination.”
One issue, Kerbel says, is whether a nar-
rowly aimed drug can be effective in many
cancers when there are dozens of potential
angiogenesis targets. Avastin goes after only
one, called vascular endothelial growth fac-
tor; it’s possible VEGF doesn’t play a big
role in breast cancer, Kerbel says. Further-
more, a 2009 paper in the British Journal
of Cancer by a Swiss group reported that
breast tumors had unexpectedly low lev-
els of VEGF and other angiogenic factors,
suggesting that blocking them may not be a
promising strategy.
But Kerbel’s view is nuanced. He agrees
that clinical trials of Avastin showed little
benefit but notes that other evidence sug-
gests that women with an aggressive subtype
of breast cancer, called “triple negative,” can
benefit from it.
The elephant in the room is cost. Avastin
is one of the most expensive cancer drugs on
Once on ‘Fast Track,’
Avastin Now Derailed
FDA
Pro-Avastin. A breast cancer patient advocate testi-
fies in favor of the drug during last week’s hearing.
ti
ap
in
ity o
tumo
to 3 m
was ap
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 144
NEWS&ANALYSIS
C
R
E
D
I
T
:

C
O
U
R
T
E
S
Y

O
F

E
D

G
R
U
M
B
I
N
E
BEIJING—As one of the country’s poorer
provinces, Yunnan in southwestern China
has struggled to balance economic develop-
ment and conservation. “Few researchers
in China are equipped with the necessary
expertise or perspective” to help Yunnan
officials craft sound environmental policies,
says Xu Jianchu, a conservation biologist
at the Kunming Institute of Botany of the
Chinese Academy of Sciences (CAS). That’s
where Edward Grumbine comes in.
With more than 2 decades of expe-
rience as an environmental policy
expert, Grumbine came to Kunming
from Prescott College in Arizona on
a 1-year fellowship and just reupped
for a second year. He and Xu are try-
ing to persuade local officials to take
a holistic approach toward land-use
management and to enlighten them
on the perils of unrestrained hydro-
power development.
China has been going all out to
persuade ethnic Chinese researchers
stationed abroad to return and bolster
science in their motherland (Science,
31 July 2009, p. 534). Conditions in China
are improving so quickly, research chiefs say,
that labs are now wooing top overseas scien-
tists, no matter their ethnicity. Last month,
CAS held a workshop here to assess a pair
of fellowship programs launched in 2009 to
bring non-Chinese scientists to CAS labs.
Thus far the programs have supported 179
postdocs and 477 professors, paying annual
stipends as high as 500,000 yuan ($77,000).
At the meeting, Grumbine and 69 other
foreign researchers on long-term stints
hailed the fellowships. China, they say,
offers a number of attractions. Some sci-
entists came here because there is ample
funding for research and new instruments.
Others followed partners to China or say
they are curious about the rising power.
Whatever their reason for flocking to
China, foreign fellows have had an impact.
For starters, they have helped catalyze inter-
disciplinary studies that are sorely lacking
in China, says Zhou Zhonghe, director of
CAS’s Institute of Vertebrate Paleontology
and Paleoanthropology here. One example
is Romain Amiot, a foreign postdoc who
came here in 2006 to reconstruct ancient cli-
mates based on evidence gleaned from the
teeth and bones of dinosaurs. “Nobody had
used our materials to study past climate,”
Zhou says. Fellows also spark connections.
“Behind each fellow is a network of over-
seas expertise that is valuable for academic
exchange,” Xu says. Some visitors super-
vise graduate students, write papers, and
edit English-language Chinese journals.
“This will help to boost language skills
and educate young generations of Chinese
scientists,” Zhou says.
Not all is rosy. Researchers say CAS
should offer more fellowships and that these
should permit longer stays and include
funds for research materials. CAS should
also host more top-notch conferences so
foreigners can become better acquainted
with Chinese science, says David Yuen, a
geologist at the University of Minnesota,
Twin Cities, and a CAS fellow here at the
academy’s Graduate University.
Fellows have also vented frus-
tration over the difficulty of get-
ting materials, such as cell lines and
transgenic animals, in and out of
China. Other concerns are poor data
sharing and weak enforcement of
intellectual-property rights. To lure
more top researchers to China, says
Zheng Chunmiao, a hydrologist here
at Peking University, “these obstacles
must be overcome.”
Most fellows say they have had
a gratifying experience. Last year,
Thorjørn Larssen, a geochemist at the
University of Oslo, won a 4-month
fellowship to study mercury pollution and
mitigation policies at CAS’s Institute of Geo-
chemistry in Guiyang. China is responsible
for more than a third of all mercury released
into the environment. “What we do in China
matters to the rest of the world,” Larssen
says. He helped initiate a provincewide sur-
vey of mercury pollution and human expo-
sure that he and his Chinese colleagues hope
to scale up nationwide. For scientists who
are game to take on such challenges, CAS
has rolled out the red carpet.
–JANE QIU
Jane Qiu is a writer in Beijing.
Eco-guru. CAS fellow Edward Grumbine is advising Yunnan Province
on environmental policy.
Foreign Researchers Begin to Make Their Mark
CHI NA
the market. FDA and its advisers are not per-
mitted to consider cost in their calculus, but “I
don’t believe … for a second” that they ignore
it, Kerbel says.
The fundamental question remains: Just
how good does a drug have to be to merit
approval? Richard Pazdur, who heads
FDA’s office of oncology drug products,
told Genentech that “all we wanted was one
single trial to show a clinical benefit” and
reassure FDA that the early results held up.
But Paul Schmidt, an attorney for the com-
pany, countered that Genentech thought it
had done that in a trial in which women get-
ting Avastin had 0.8 months more time—just
over 3 weeks—during which their tumors
didn’t grow. Did FDA ever warn Genen-
tech that this “will not be enough to support
approval?” Schmidt asked. The answer, Paz-
dur admitted, was no.
Advocates for breast cancer patients are
divided, demanding both choice in therapy
and effective treatments. Other groups have
experienced the same tension. The acceler-
ated approval program began in 1992 after
AIDS patients insisted on faster access to
drugs. But that community is now “advo-
cating for rigorous investigation rather than
open access to whatever is out there,” says
Alastair Wood, a pharmacologist and part-
ner at Symphony Capital in New York City.
“There’s been a real evolution, … really
a swing away from ‘Give me everything
you’ve got and I’ll take it’ to … ‘We want
better therapies, but we want evidence that
the therapies are working.’ ”
Genentech’s options, meanwhile, have
narrowed. It plans to appeal to FDA Com-
missioner Margaret Hamburg for “a middle
ground” resolution, says Genentech spokes-
person Krysta Pellegrino, perhaps asking
FDA to restrict Avastin use to the most aggres-
sive breast cancers.
–JENNIFER COUZIN-FRANKEL AND
YASMIN OGALE
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 145
NEWS&ANALYSIS
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

E
.

M
A
R
S
H
A
L
L
/
S
C
I
E
N
C
E
;

S
O
U
R
C
E
:

D
E
B
O
R
A
H

Z
A
R
I
N
No one knew what to expect when Congress
created a registry of clinical trials in 1997—
then expanded it in 2007, asking researchers
to submit a summary of results to a public
Web site. Even the number of studies likely
to be registered was a complete unknown,
and no one had any idea if the results would
be informative, says Deborah Zarin, who
directs the database at the National Insti-
tutes of Health (NIH), called ClinicalTrials.
gov. But to date more than 108,000 trials
have been registered, and results have been
posted for more than 3600 (see
graph). Most intriguing, Zarin
says, the data are revealing a
whole “world of clinical trials”
never seen before.
The database includes clini-
cal research funded by the U.S.
government and clinical stud-
ies conducted by private entities
to meet U.S. standards. (Early
studies such as phase I trials are
exempt from registering.) Since
2008, trials must report outcome
data within 1 year of comple-
tion. Delays aren’t granted for
publication.
At a conference in June spon-
sored by NIH’s National Library
of Medicine and AAAS (Science’s pub-
lisher), Zarin shared observations based on
results submitted so far. At first, only 25%
of the trials with results in the database had
been published; over time, the published
fraction rose to about 52% in that cohort.
Some studies are distressingly complex,
with tens of “primary endpoints.” But so
far, despite worries that the reporting for-
mat would be rigid, no trial has been too
exotic to summarize, Zarin says.
The system wasn’t designed to do qual-
ity control, but database managers do
check data. This has led to surprises, too,
Zarin said, including more confusion than
expected in summary reports—revealed for
example by the use of the phrase “time to
survival,” or by the study with results, Zarin
said, on “more eyeballs than two times
the number of subjects.” Zarin, a Harvard
Medical School graduate, psychiatrist, and
expert in evidence-based medicine, recently
spoke with Science; the interview was
edited for brevity and clarity.
–ELIOT MARSHALL
Q: Did you initially meet resistance trying
to get the summary data?
A: Total resistance. Certain high-volume
drug companies were clearly under pres-
sure [to release trial data] and put a lot of
effort into it and are doing an excellent job.
Other companies don’t seem to be jump-
ing in with two feet. … Academic research-
ers have been slow to comply. … It really
requires a mindset change. The first reac-
tion of academic researchers was, “This
will interfere with publication. No one can
tell me when to present the results. I’m
working on a journal article.” It is very
clear in the law that this has to be done on a
timeline that’s legally determined and does
not depend on when the publication occurs.
… The journal editors … have said that
posting of summary results will not inter-
fere with publication. But people need to
get the message.
Q: How many trials are you missing?
A: There simply is no way to know. … You
hear of sporadic cases. Whether there are
whole islands of unregistered trials, whether
the sponsors have stashes of unregistered
trials, we just don’t know. If there are any
islands, they are getting smaller and smaller.
Q: Were you surprised by what you have
been learning from the data?
A: I call it my introduction to the sausage
factory. It appears that there are a number of
practices in the world of clinical trials that
I hadn’t been aware of; it surprised a lot of
people. For example, researchers might say,
this is a trial of 400 subjects, 200 in each
arm, and when they came to report results,
they would be talking about 600 people. We
would ask them to explain. They
would say, “We are including
200 people from this other study
because we had always intended
to do that.” … There were a lot
of—what would I call it?—
nonrigorous practices.
Q: Were the lapses more than
clerical errors?
A: We are finding that in some
cases, investigators cannot
explain their trial, cannot explain
their data. Many of them rely on
the biostatistician, but some bio-
statisticians can’t explain the
trial design.
So there is a disturbing sense
of some trials being done with no clear
intellectual leader. That may be too strong
a statement, but that’s the feeling we are
left with.
Q: Are reporting requirements changing
how research is done?
A: I have not seen the impact on new trials,
but I’ve heard anecdotes. Groups say they
are prespecifying their outcome measures
much more carefully. I know that journal
editors are using the ClinicalTrials.gov site
to scrutinize the prespecified outcome mea-
sures to make sure that they are in sync with
what is being reported in a manuscript—
and I know that editors have found prob-
lems that way.
Q: Who is the database for?
A: For right now, we view the basic results
database as helping experts—whether it
be journal editors or people we count on
to write good systematic reviews. … It
improves clinical practice, and in that way
it benefits the public.
120,000
100,000
80,000
60,000
40,000
20,000
0
Number of New Records (since 1 May 2005)
2005 2006 2007 2008 2009 2010 2011
~25/wk
~250/wk
~330/wk
N
u
m
b
e
r

o
f

r
e
c
o
r
d
s
Clinical cornucopia. Deborah Zarin oversees an
expanding registry of clinical trials and their results.
Unseen World of Clinical Trials
Emerges From U.S. Database
NEWSMAKER I NTERVI EW: DEBORAH ZARI N
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 146
C
R
E
D
I
T
:

U
N
I
V
E
R
S
I
T
Y

O
F

A
R
I
Z
O
N
A
,

B
I
O
S
P
H
E
R
E

2
Looking like a futuristic spaceship that has
landed in the desert, Biosphere 2 has always
been an oddity. The early days of this mas-
sive, enclosed facility were marked by high-
profile attempts to have people live inside a
sealed ecosystem for up to 2 years. After that
flopped, Columbia University took over and
conducted ambitious climate change experi-
ments. But officials abruptly pulled the plug
in 2003 after costs became unsustainable
(Science, 19 December 2003, p. 2053).
Since 2007, the University of Arizona
(UA) has managed a modest amount of
research at the facility, primarily focusing on
ecohydrology. Now the university, aided by
two major gifts last week, is pushing ahead
with plans to revive the scientific fortunes of
Biosphere 2. On 1 July, the real estate devel-
opment company that owned the 16-hectare
campus donated it and the iconic buildings
to UA. At the same time, financier—and
longtime Biosphere patron—Edward Bass
announced he would chip in $20 million by
October to support operations and research.
“It’s on a good trajectory,” says plant biologist
Joseph Berry of the Carnegie Institution for
Science in Stanford, California, who led the
scientific advisory committee for Columbia.
The challenge will be for UA to attract
enough research grants to make Biosphere
2 viable in the long term. University offi-
cials are betting big on a set of experimen-
tal hillslopes, which will eventually take up
more than 25% of the facility. Much more
space could ultimately be developed for other
research into the impact of climate change
on temperature and rainfall. “I am excited
about the options for experiments in large,
well-controlled, replicated habitats,” says
Chris Field, head of the Carnegie Institution’s
Department of Global Ecology.
Located 50 kilometers from Tucson,
Biosphere 2 was built in the late 1980s for
$200 million by Space Biospheres Ven-
tures. Funded primarily by Bass, the com-
pany aimed to develop technology for space
colonization. The campus boasts 13,000
square meters of airtight greenhouses, a
3-megawatt power plant, dormitories, and
a conference center. Biomes include an
“ocean” with coral reef, mangroves, tropi-
cal rainforest, and savanna. Two high-profile
experiments in the early 1990s, which sealed
crews of “biospherians” inside the glass pyra-
mids, encountered many problems, including
falling oxygen levels.
In 1994, Bass’s investment company
took over and brought in Columbia Univer-
sity to manage the site. Researchers spent
7 years revamping Biosphere 2 and investi-
gating the effects of elevated carbon dioxide
levels on ocean and terrestrial ecosystems.
These experiments didn’t need the facility to
remain airtight, but they required complex
ventilation and air conditioning to control the
atmosphere. After Columbia ended its con-
tract, because of expenses and an inability
to win enough government grants, Bass sold
the property and surrounding ranch to a real
estate development company. The research
equipment was auctioned off, and the site
became just a tourist attraction. (Biosphere 2
is still a major draw; more than 100,000 peo-
ple a year pay up to $20 to visit.)
UA started up research again in 2007, and
Bass has donated about $4 million per year
for salaries, repairs, and operations. The uni-
versity hired nine scientists, led by plant ecol-
ogist Travis Huxman, to work at Biosphere 2.
In a big change, they essentially opened the
windows, shifting Biosphere 2 from a “par-
tially closed” system, in which CO
2
con-
centration was manipulated, to “open flow”
with external air. This saves roughly 75% in
energy costs, and Huxman says it also allows
for better measurement of water vapor. “To
be honest, there are many facilities and sites
in natural settings tackling the CO
2
ques-
tion,” Huxman says. Since 2007, a number of
small-scale experiments have taken place at
Biosphere 2, including a study of the impact
of temperature on pinyon pine mortality.
A more ambitious flagship effort is now
gaining momentum. Called the Landscape
Evolution Observatory (LEO), three large
artificial hillslopes will be built in what was
the 2000-square-meter farm biome. The first
goal is to study how water flows through soil
and alters its chemistry. Then researchers will
investigate how vegetation impacts the inter-
actions of carbon and water cycles under vari-
ous climatic conditions. The first phase of the
$6 million construction, all of which has been
funded by Bass, should be completed in Janu-
ary. Although this kind of research could be
done in a traditional greenhouse, the scale of
Biosphere 2 allows for large plants and “semi-
realistic” ecosystem settings, says William
Schlesinger, who directs the Cary Institute of
Ecosystem Studies in Millbrook, New York.
The overall approach makes sense to
plant biologist Barry Osmond of the Austra-
lian National University in Canberra, who
led Biosphere 2 from 2001 to 2003. “It is a
good move, securing long-term focus for a
large number of really talented researchers to
a unique experiment,” he says. “This seems
a much more robust, more sustainable and
achievable operating model” than Columbia’s
attempt to quickly build a major research
facility, like a national laboratory, and fund it
with a large government contract.
Whether UA can truly revive Biosphere
2 as a scientific destination in the desert
depends in part on whether its researchers
can win enough grants. (So far they’ve been
awarded more than $624,000 from the U.S.
National Science Foundation.) Joaquin Ruiz,
dean of the UA College of Science, says the
goal is to use the funding from Bass to get
LEO running, then invest the remainder as
a quasi-endowment to help cover operations.
Salaries would be covered by grants and
income from tourism.
The handover of Biosphere 2’s owner-
ship, from CDO Ranching & Development
to the university, should also help science.
Researchers will have more flexibility to
modify the buildings, if necessary. “It does
offer some stability to a facility with an oth-
erwise confusing past,” Schlesinger says.
“Biosphere 2 is a big commitment on the
part of UA.” –ERIK STOKSTAD
Cash Advance, New Approach Aim
To Relaunch Biosphere 2
ECOLOGY
Gaining ground. Biosphere 2’s
massive greenhouse will host a unique
experiment in soil processes.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 147
NEWS&ANALYSIS
C
R
E
D
I
T
:

A
M
Y
R
I
S
For Alicia Jackson and her bosses at the
Defense Advanced Research Projects
Agency (DARPA), synthetic biology is
still too much of an art form. That’s why
Jackson, who was trained as a materials
scientist and is now a DARPA program
manager, is hoping her agency can help
scientists get much better at modifying the
metabolic and genetic machinery of cells
to produce useful products such as fuels.
“We want to create a new manufacturing
capability for the United States” using liv-
ing systems, she says. And toward that end,
DARPA is about to sponsor a $30 million
program called Living Foundries.
Although an official announcement is still
weeks away, researchers who learned
about the program last week from
Jackson at a DARPA industry work-
shop at its Arlington, Virginia, head-
quarters, are thrilled. “It really is the
first time that our funding agencies
have embraced the idea of supporting
not individual projects but rather the
vision of creating a technology which
can be used for many, many different
applications,” says Thomas Knight,
a synthetic biologist at Ginkgo Bio-
Works in Boston.
Turning bacteria, fungi, or other
kinds of cells into molecular facto-
ries, with DNA as the instruction set,
has been difficult and expensive. It
took DuPont from 1992 to 2007 and
many million dollars before it suc-
ceeded at efficiently making a poly-
mer building block in genetically
engineered bacteria fed on corn syrup, for
example. Biosynthesis by engineered yeast
of artemisinin, an antimalarial medication
typically derived from plants, cost $25 mil-
lion to develop. “And we are not getting any
better; we’re not getting any faster,” Jackson
notes. It now takes hundreds of thousands of
experiments to work out the kinks for each
new cell factory, she says.
To foster a general capability in synthetic
biology, DARPA plans to push for the con-
tinued standardization of “parts”: the genes,
regulatory DNA, and other genetic instruc-
tions researchers add to cells. It will also
promote the development of compatible,
easier-to-use tools for assembling DNA into
the right instructions, quicker and cheaper
methods for synthesizing DNA, and more
streamlined methods for assessing these cel-
lular factories. Jackson pledged that Living
Foundries will, unlike most grants programs,
fund all meritorious proposals. The new
DARPA program will also support informat-
ics and the development of new ways to char-
acterize cells, areas not always lumped into
synthetic biology.
Such funding is more than welcomed by
synthetic biologists, who acknowledge that
their field is still in its infancy. “There’s a lot
of foundation work that needs to be done,”
says Julius Lucks of Cornell University. The
fundamental tools that would make synthetic
biology more affordable and commercially
viable are lacking, and their development is
not really in the purview of typical academic
research, Lucks explains. He sees the new
DARPA program as “a funding source to
bridge the gap between what can be done in
the lab and what industry will do.”
The National Science Foundation (NSF)
took a first cut at laying this foundation
5 years ago, when it established a Synthetic
Biology Engineering Research Center, pri-
marily based at the University of Califor-
nia, Berkeley. The $40 million, 10-year NSF
award supports about 10 senior investigators
who “are really going from basic science to
[the] translation to industry,” says NSF pro-
gram manager Theresa Good.
But the funding for that center is spread
too thinly to push the technology forward
the way it needs to be, says Drew Endy, a
synthetic biologist at Stanford University in
Palo Alto, California. The Living Foundries
program can build on what that NSF effort
has achieved to take synthetic biology to the
next level, he predicts.
Endy is particularly eager to see gains in
commercial DNA synthesis and in making
tools for genetically engineering cells com-
patible with each other. Too often, he says,
researchers who design a new genetic pro-
gram, such as one to make a useful chemical,
have to synthesize the needed DNA in frag-
ments and assemble it, a time-consuming task
requiring at-the-bench expertise. The alterna-
tive of buying that DNA commercially can
also be challenging and expensive: It costs
at least 45 cents a base and can take up to
2 months to procure. Endy hopes the DARPA
program will make buying DNA routine,
with a quick turnaround time. Jackson agrees
that DNA should not be an impedi-
ment. “DNA should cost next to noth-
ing,” she insists.
Ultimately, DARPA’s goal is to
“break open the field to new players
so [they] will not have to be experi-
enced in genetics to design new bio-
logical systems,” Jackson says. At the
meeting, she drew parallels between
synthetic biology and the semicon-
ductor industry, pointing out that
applications for integrated circuits
really took off once transistors and
other components were standardized,
enabling many more people to come
up with new circuits. She would like
to see that happen with synthetic
biology, as newcomers to the field are
likely to devise products and appli-
cations not considered by the typical
synthetic biologist.
DARPA “has a very bold vision,” says
Daniel Drell, a biology program manager at
the U.S. Department of Energy who attended
the Arlington meeting. “But I don’t think
this is going to be easy.” He predicts it will
be challenging to make living systems con-
form to engineering principles. And Steven
Benner, a synthetic biologist at the Foun-
dation for Applied Molecular Evolution in
Gainesville, Florida, says Jackson doesn’t
appreciate that commercially produced syn-
thetic DNA is quite reasonably priced, given
the cost of the chemicals used to make it.
Finally, the time it takes to develop a new liv-
ing foundry can only be sped up so far, says
Dieter Söll of Yale University, as research-
ers are constrained by how fast the cells used
reproduce. Concludes Drell, “Biology is
going to fight them.”
–ELIZABETH PENNISI
DARPA Offers $30 Million to Jump-Start Cellular Factories
SYNTHETI C BI OLOGY
Living foundry. Synthetic biologists have made yeast (blue) that pro-
duce and release an oily precursor (brown) for making diesel and other
useful chemicals.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 148
C
R
E
D
I
T
:

T
H
E

M
O
N
R
O
E

E
V
E
N
I
N
G

N
E
W
S
/
A
P
AFTER KILLING THE PROJECT TO ESTABLISH A
nuclear waste repository at Yucca Mountain,
Nevada, last year, President Barack Obama
set up a commission to chart a new course
for U.S. nuclear waste policy. The group,
which is set to deliver its interim findings
at the end of this month, is expected to say
that the answers to America’s nuclear waste
conundrum are technically feasible. The
problem, however, is that those solutions are
likely to be, in political terms, radioactive.
The Blue Ribbon Commission on Amer-
ica’s Nuclear Future confronts a challenge
that has stymied Washington for 40 years:
the nation’s relentless production of nuclear
waste. In recent decades, U.S. reactors have
created more than 2000 metric tons of highly
radioactive spent fuel each year. Codified in
law in 1987, the Yucca plan meant the spent
fuel, held in bundles of 4-meter-long zirco-
nium alloy tubes, would be cooled for up to
a decade in storage pools at U.S. reactors.
Then it would be shipped to Yucca Moun-
tain, transferred into steel cylinders, and fur-
ther cooled by fans for 50 years. When the
facility contained 70,000 tons of waste, it
would be closed up.
In the past 24 years, the Department of
Energy (DOE) has built an 8-kilometer-long
tunnel at Yucca and has conducted experi-
ments to ensure that the reposi-
tory could hold the waste for up to
1 million years without releasing
dangerous amounts of radiation.
But although the government has
spent roughly $10 billion on the
project, lawsuits, red tape, and
political opposition have prevented DOE
from disposing of a single ton of commercial
fuel. About 65,000 tons of spent nuclear fuel
are piled up at U.S. reactors in cooling pools
and in steel-and-concrete casks stored out-
doors. Some experts believe the pools rep-
resent unacceptable safety or environmental
risks in the case of natural calam-
ity or terrorist attack. But moving
cooled-down fuel from U.S. pools
into casks, which are considered
safer, would cost utilities billions
of dollars.
To alleviate this pressure on
reactor sites and buy the govern-
ment time to establish a perma-
nent repository, the commission
will likely call for an impor-
tant new step: interim storage of
the fuel in one or several central
locations. After cooling in pools
for a decade, fuel would be trans-
ferred to such a facility to be stored
in outdoor steel-and-concrete
casks for “multiple decades up to
100 years or possibly more.” Over
that period, the commission envi-
sions, expanded federal research
into fuel recycling or other tech-
nologies might reduce the amount
of fuel requiring disposal. Meanwhile, fed-
eral officials would have a second chance to
establish a permanent U.S. geologic reposi-
tory—this time, perhaps, in a fashion less
acrimonious than the Yucca effort.
Buying time
“Consolidated interim storage preserves
options while other aspects of an integrated
waste management strategy can be devel-
oped,” says one of three commission sub-
committee reports released in June. Later
this month, an interim report from the full
commission is expected; given the hundreds
of hours of public testimony and published
documents cited in the lengthy subcommit-
tee reports, experts expect the final version,
scheduled to be released in January 2012, to
offer substantially the same conclusions.
In addition to paving the political and
logistical route to disposal, cen-
tral, interim storage sites could
make the repository easier to
design and build. Building the
Yucca repository required a num-
ber of engineering tradeoffs,
explains physicist Charles Fors-
berg of the Massachusetts Institute of Tech-
nology (MIT) in Cambridge, and the site’s
“awkward” design was a stumbling block to
getting it licensed.
One reason was that Yucca Mountain had
to cool waste before permanently storing
it. Spent fuel straight from the reactor can
Online
sciencemag.org
Podcast interview
with author
Eli Kintisch.
Hot stuff. Spent fuel being lowered
into a storage cask at a commercial
nuclear reactor in Virginia.
NEWSFOCUS
Waste Panel Expected
To Back Interim Storage
A blue-ribbon commission signals that spent nuclear fuel should cool
above ground while the United States figures out long-term disposal
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 149
WA
OR
CA
NV
I D
MT
WY
UT
CO
KS
OK
MO
AR
L A
MS AL
GA
F L
T N
KY
VA
MD
DE
WV
NC
SC
T X
HI
NM
AZ
AK
NE
I A
SD
ND
MN
WI
MI
I L
I N
OH
PA
NJ
NY
ME
MA
CT
VT
NH
RI
100
Spent fuel in
metric tons
1000
3000
5000
States with
<1 ton of
spent fuel
States with
no spent fuel
Spent nuclear fuel is piling up at
77 sites (small dots) in 35 states.
Where the Waste Is
M
A
P

S
O
U
R
C
E
:

D
E
P
A
R
T
M
E
N
T

O
F

E
N
E
R
G
Y
,

N
U
C
L
E
A
R

E
N
E
R
G
Y

I
N
S
T
I
T
U
T
E
;

G
R
A
P
H

S
O
U
R
C
E
:

E
L
E
C
T
R
I
C

P
O
W
E
R

R
E
S
E
A
R
C
H

I
N
S
T
I
T
U
T
E
quickly reach 1500ºC, hot enough
to destroy the tubes that hold it.
Cooling for about a decade in stor-
age pools dissipates most of the
heat from the shortest-lived iso-
topes. But after being bundled
together and entombed in the
mountain for centuries, it might
still gradually create enough
heat to aid corrosion of the
tubes, create dangerous steam
within the tunnels, or even, over
time, alter the geology of the
site. First cooling the waste for
at least 5 decades at interim stor-
age sites could eliminate the need
for fans at a permanent reposi-
tory, says Forsberg, co-author of
several influential MIT reports on
nuclear waste. The interior of the
long-term repository—compared
with Yucca Mountain—would also
require less ventilation and less
access by remote devices to han-
dle the fuel after emplacement,
and it could be more easily sealed with an
appropriate fill.
In any case, experts agree, some new plan
for waste storage is essential. Waste currently
stored in pools and casks at U.S. sites does
not pose “unmanageable … safety or security
risks,” says a subcommittee report. But every
ton that stays at reactor sites makes those
risks slightly greater. Fuel in U.S. spent fuel
pools is packed four times as densely as it was
25 years ago, raising concerns about the risk
of explosions or meltdown if the pools were
to empty in an accident. The tsunami that dev-
astated the Fukushima nuclear plant in Japan
in March may have resulted in a loss of water
in one of its ponds (Science, 1 April, p. 24).
A draft commission report says the issue of
the safety of keeping fuel densely packed in
pools should be “reexamined,” although “it is
still too early to draw definitive conclusions”
from the Fukushima accident. It calls for an
expert panel at the National Academies to
tackle the subject.
If an interim storage site could
get licensed—a big if, given polit-
ical sensibilities—it might save
money for utilities and the govern-
ment, which is currently paying
hundreds of millions of dollars in
legal claims to utilities for the waste.
Nine decommissioned reactor sites
in the United States currently house
nuclear fuel in aboveground casks.
Centralized storage could save utili-
ties billions in security costs and by
freeing the land for other uses.
As for long-term disposal, commis-
sioners say the government should “expe-
ditiously” move to set up a geologic
repository—they were told not to specify
where. “There is no ethical basis for abroga-
tion of responsibility” for securing nuclear
waste “to future generations,” a subcom-
mittee report says. To avoid repeating the
Yucca Mountain experience, which was
plagued by opposition from the state of
Nevada (see p. 150), the process of choos-
ing a site should include “consultation,
transparency, accountability, and scientific
and technical credibility,” a draft report
says. Commissioners are also likely to call
for the project to be managed by a new,
independent entity.
The commission is also likely to recom-
mend expanding research into technolo-
gies such as reprocessing, in which nuclear
waste is converted back into nuclear fuel,
and into advanced or more efficient reac-
tors that might produce less waste. Nuclear
power will never be completely clean, how-
ever. “No currently available or reasonably
foreseeable reactor and fuel-cycle technolo-
gies … have the potential to fundamentally
alter the waste management challenge,” the
research subcommittee draft says.
The road ahead
Forecasting what the Blue Ribbon Com-
mission will recommend is one thing; pre-
dicting what the Obama Administration and
its successors will actually do with them is
much harder. Several environmental and
antinuclear groups have already spoken out
against creating new storage sites for waste,
and a commission subcommittee admits
in a report that it’s a “contentious issue.”
Commissioners hope the track record of the
57 licensed fuel storage facilities—most at
U.S. reactor sites—will alleviate some fears,
and that an open site-selection process cou-
pled with “incentives” like training and jobs
for local communities and utilities will ulti-
mately carry the day.
But even simply getting money for more
federal research into nuclear power could be a
challenge. As the budget process in Washing-
ton grows ever more contentious, lobbyists
and activists alike are increasingly skeptical
that substantial increases can happen soon. “I
don’t think the budgets are going to expand
beyond what they are now,” says physicist
Thomas Cochran of the Natural Resources
Defense Council in Washington, D.C.
–ELI KINTISCH
Fuel Stored
in Dry Casks
Fuel Stored in
Cooling Pools
2010
20,000
40,000
60,000
80,000
100,000
M
e
t
r
i
c

t
o
n
s

o
f

u
r
a
n
i
u
m
120,000
140,000
2015 2020 2025 2030 2035 2040 2045 2050 2055 2060
Plain spent. Assuming no new plants, used fuel stores will
more than double by 2050.
NUCLEAR WASTE NEWSFOCUS
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 150
NEWSFOCUS
C
R
E
D
I
T
:

D
E
P
A
R
T
M
E
N
T

O
F

E
N
E
R
G
Y
The Obama Administration’s shutdown of
the quarter-century-long, $15 billion effort
to dispose of 65,000 tons of U.S. spent
nuclear reactor fuel in Nevada’s Yucca
Mountain could be the latest of many les-
sons learned around the world. Unforeseen
technical problems have abounded there and
at proposed disposal sites around the world,
but no certain deal breakers have turned up.
Yet, despite the absence of insurmount-
able geologic or engineering obstacles,
no permanent repository for spent reactor
fuel has been built anywhere. Every coun-
try looking for a place to dispose of its
wastes has stumbled in its early tries to site
repositories. Almost invariably, a govern-
ment decides which site would be suitable,
it announces its decision, an uproar ensues
from the locals, the government defends
its chosen site, but eventually it is forced to
abandon its choice as untenable.
In the wake of the 2010 abandonment of
the Yucca Mountain program, the U.S. advi-
sory Nuclear Waste Technical Review Board
(NWTRB) puts the United States among
those nations whose waste-disposal pro-
grams “either have lost public trust and con-
fidence or seem never to have merited it at
all,” as the board stated in an April report to
Congress. (Yet, ironically, the United States
is the only country in the world to open and
operate a nuclear waste repository: a facility
for storing waste from the nuclear weapons
program that doesn’t include spent fuel.)
So as the Administration’s Blue Ribbon
Commission on America’s Nuclear Future
prepares to deliver its draft report (see
p. 148), many authoritative groups have
been driving home the lessons learned from
Yucca Mountain and around the world.
NWTRB put it most succinctly: “The
interdependencies, both subtle and overt,
between the technical, social, and political
forces are inescapable.”
Uniformly dismal failure
In the 1950s, when coun-
tries first started ponder-
ing how to dispose of
spent nuclear fuel from
power plants and radio-
active waste from nuclear
weapons production, the
solution seemed straight-
forward enough. Nuclear
waste contains isotopes of
elements that will remain
radioactive for thousands
to many hundreds of
thousands of years. Rock
formations hundreds of
meters beneath the sur-
face have been there, lit-
tle disturbed, for millions
if not billions of years.
So put the wastes in tun-
nels in the rock, seal the
tunnels, and the problem
would be solved. Layers of
deeply buried salt were an early favorite; if
there’s salt still there after millions of years,
water—which can corrode stored waste and
carry it back into the environment—won’t
be a factor. But whatever medium was at
hand—salt, granite, clay, or volcanic ash
turned to stone called tuff—looked promis-
ing to the government’s experts charged with
finding a suitable site.
Despite the promising geology, the top-
down approach just didn’t pan out. Whether
it was salt in Germany or old bedrock in the
United Kingdom, Canada, or Scandinavia,
“almost all countries that have tried to site
repositories have had one or more failures,”
notes a June draft report, Spent Fuel From
Nuclear Power Reactors, from the Inter-
national Panel on Fissile Materials (IPFM),
an independent group of nuclear experts.
In the United States, 4 decades of govern-
ment efforts were marked “by heavy hand-
edness on the part of the federal government
and political uprisings in a succession of
states where it proposed to site repositories,”
notes the IPFM report. An early setback
came near Lyons, Kansas, where in 1970
the Atomic Energy Commission (AEC)—
the forerunner of today’s Department of
Energy (DOE)—decided to entomb highly
radioactive wastes from nuclear weapons
production in an abandoned salt mine. In
1957, a U.S. National Academies report had
Light at the End of the Radwaste
Disposal Tunnel Could Be Real
A long run of failures could finally drive the United States to follow other countries’
lead and accept radioactive waste disposal as the sociotechnological problem that it is
RADI OACTI VE WASTE DI SPOSAL
Yucca Mountain. The now-
abandoned repository site’s
rock is perhaps the planet’s
most thoroughly studied.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 151
NEWSFOCUS
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

C
O
U
R
T
E
S
Y

O
F

S
V
E
N
S
K

K
Ä
R
N
B
R
Ä
N
S
L
E
H
A
N
T
E
R
I
N
G

A
B
;

N
U
C
L
E
A
R

R
E
G
U
L
A
T
O
R
Y

C
O
M
M
I
S
S
I
O
N

(
3
)
;

W
I
K
I
M
E
D
I
A

C
O
M
M
O
N
S
recommended layered salt formations for
such wastes because, over time, salt would
flow to seal in the wastes.
But the head of the Kansas Geological
Survey urged more study of the integrity of
the proposed Lyons repository. AEC agreed
but continued its preparatory work anyway.
Fearing a fait accompli, the IPFM report
says, Kansans and their politicians rose to
oppose the plan. Technical revelations then
lit the fuse on a by-now-politically-unstable
situation. It turned out that the site had long
ago been peppered with oil and gas wells
with no assurance they had all been securely
plugged. And several years earlier, a min-
ing company pumping water into the for-
mation nearby to dissolve and extract salt
had 640 cubic meters of water go missing,
suggesting that the salt geology was more
complex and less well understood than the
academies had assumed. AEC abandoned
the site in 1971.
Yucca Mountain’s cycle from decision
to abandonment was far more protracted.
Acting under the 1982 Nuclear Waste
Policy Act, DOE had selected three can-
didate sites: salt in Texas, basalt in Wash-
ington state, and volcanic tuff at Nevada’s
Yucca Mountain.
Technically, Yucca Mountain looked
promising. The spent fuel would be well
above the water table and therefore exposed
to the vanishingly small amount of water
seeping from the desert above. The decisive
factor, however, was political. At the time, the
Democratic Party controlled both houses of
the U.S. Congress, and the powerful speaker
of the House represented Texas while the
House majority leader represented Washing-
ton state. Nevada’s delegation, however, was
split between Democrats and Republicans
and was new to Congress. In 1987, Congress
struck the Texas and Washington sites from
the list, leaving Yucca Mountain the only
candidate in the running. Nevadans still call
the act the “screw Nevada bill.”
If Yucca Mountain had proved to be the
perfect repository site, it might not have
been abandoned, but investigations soon
started to tarnish its luster. That’s normal
for site evaluations; as geophysicist Wendell
Weart told Science in 1999, “You never feel
quite as comfortable about a site as the day
you start to study it.”
Yucca Mountain’s unwelcome surprises
included the possibility of earthquakes and
volcanic eruptions (Science, 8 November
1996, p. 913) and fears, later allayed, that the
repository itself might explode like
a nuclear bomb (Science, 30 June
1995, p. 1836). But the overarching
concern has been the discovery that
water seeps down through the moun-
tain many times faster than had been
thought. So, in the mountain’s oxygen-
rich interior, water laden with salt dis-
solved from the rock would drip onto
spent-fuel assemblies still hot from
their lingering radioactivity. That’s a
great recipe for corrosion. The seep-
ing brine would release radionuclides
from the spent fuel and carry them on
through the rock as far as the water
is going.
To make matters worse, the plan-
ning horizon for Yucca Mountain got
extended by a factor of 100. An acad-
emies study committee requested by
Congress concluded that the risk
of human exposure to radioactivity
should be estimated out to the time
of maximum exposure, when con-
tainment has failed and wastes have
spread. That upped the time scientists had to
predict the behavior of the repository and its
wastes from 10,000 years to 1 million years.
In response to such surprises, DOE hun-
kered down. “DOE lacked transparency in
developing its plans for the Yucca Mountain
repository,” an April report, Commercial
Nuclear Waste, from the U.S. Government
Accountability Office (GAO) concluded.
For example, instead of polling the broad
community for ideas, DOE designed tita-
nium drip shields on its own to protect the
waste. DOE did not “establish independent
scientific panels or any form of state over-
sight that might have given affected parties
more confidence in the solutions,” the report
says; nor did it “promote state involvement
in key decisions and oversight.”
Could more transparency and coopera-
tion have saved the day? Mineralogist Rod-
ney Ewing of the University of Michigan,
Ann Arbor, a longtime critic of the Yucca
Mountain program, thinks so. “I really think
if there’s a strong scientific basis combined
with public empowerment, you can make
progress,” he says.
A more successful path
As an example of how to do things right,
Ewing and other critics often cite another
The Swedish way.
The KBS design:
copper-clad wastes (yellow)
encased in clay (pink) beneath
500 meters of granite.
Nuclear Waste Repositories —Past and Future?
WIPP Carlsbad, NM
Operating
Salt
Operating
Yucca Mountain, NV
Abandoned 2010
Mostly dry volcanic
tuff, oxidizing
$15 billion spent;
well studied
Hanford, WA
Dropped
Wet basalt reducing
TBD
Status:
Geologic medium:
Advantages:
Deaf Smith County, TX
Dropped
Salt
TBD
NUCLEAR WASTE NEWSFOCUS
Forsmark, Sweden
Selected
Wet basalt reducing
Local support
engineered barriers
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 152
NEWSFOCUS
C
R
E
D
I
T
:

C
O
U
R
T
E
S
Y

O
F

C
O
N
N
E
C
T
I
C
U
T

Y
A
N
K
E
E

A
T
O
M
I
C

P
O
W
E
R

C
O
.
DOE radwaste repository, the Waste Isola-
tion Pilot Plant (WIPP) near Carlsbad, New
Mexico (Science, 12 March 1999, p. 1626).
“It wasn’t an easy sell,” Ewing says. “There
were substantial objections to WIPP. But
there was a process of public and scientific
engagement, so at the end you could say, this
makes sense. It worked, but it took time”—
30 years of time.
WIPP did start with several advantages
over Yucca Mountain. First, the people of
Carlsbad wanted it. When community lead-
ers heard about the abandonment of the
Lyons site, they offered their own layered
salt as a replacement. They had just lost a
major employer, a potash mining company,
and were looking for an economic boost.
And they were already familiar with the
risks of mining, not to mention those from
nearby nuclear testing. Their interest would
never waiver.
The state of New Mexico, having con-
stituencies other than Carlsbad to consider
and a long and sometimes strained history of
state-federal relations, was not so receptive.
Nongovernmental organizations (NGOs),
including environmental groups, objected
to the Carlsbad site as well. But unlike the
way Yucca Mountain turned out, a flurry of
lawsuits brought by the state and NGOs led
to concessions from DOE and constructive
interventions by Congress. A quasi-inde-
pendent Environmental Evaluation Group
with both state and federal funding provided
credible scientific information to the state
and the public when inevitable technical
issues arose. And a signed agreement made
the state “equal partners with DOE in the
development of WIPP,” says Mark Gaffigan,
lead author of the GAO report.
Congressional legislation helped, too.
It limited WIPP to defense-related waste
such as rags, protective clothes, and tools
contaminated with toxic, long-lived radio-
nuclides including plutonium. The absence
of high-level wastes such as thermally hot
spent fuel eased relations between DOE
and the state and simplified the repository
design. Legislation also gave oversight of
the repository to the U.S. Environmental
Protection Agency, which was able to retain
the 10,000-year standard for maintaining
repository integrity without going to a mil-
lion-year standard. And legislation provided
New Mexico with $280 million in compen-
sation over 14 years. WIPP received its first
wastes in 1999; today, 9000 shipments of
wastes totaling 71,000 cubic meters have
been stored there.
No one else has managed to open a
repository for anything but low-level wastes,
but two countries—Sweden and Finland—
have gotten as far as selecting sites, although
they have not yet given them final approval.
As laid out in an IPFM report chapter by
physicist Johan Swahn of the NGO Office
for Nuclear Waste Review (MKG), Sweden
started with some advantages. For one, Swe-
den began its site search with a relatively
robust repository design in hand. Unlike the
approach at Yucca Mountain, the Swedish
KBS method developed by SKB, the nuclear
waste company responsible for ultimate dis-
posal, does not depend solely on geology for
containment. Spent fuel would be encased
in 5 centimeters of copper surrounded by
extremely low-permeability clay.
Sweden was also able to make changes in
midstream. After provoking public outcries
with uninvited exploratory drilling for a site,
SKB backtracked and asked communities to
volunteer as repository sites with the right to
back out at any point. As in the case of WIPP,
volunteers were looking for economic ben-
efits. And they also were familiar with things
nuclear; each of the two finalists already
had a nuclear plant and one had a low-level
waste site, the other a centralized facility for
temporary spent-fuel storage. The Swedish
government instituted a relatively open and
consultative site-approval process, going so
far as to fund NGOs such as MKG to moni-
tor the process. And, in contrast to Yucca
Mountain, that process sets a more practi-
cal standard to shoot for. It does not depend
solely on quantitative calculations of the risk
of repository failure out to a million years.
Beyond a few hundreds of thousands of
years, more qualitative arguments for repos-
itory safety can be made.
A U.S. way ahead?
After seeing these and other reports, hearing
testimony, and making site visits, the Obama
Administration’s Blue Ribbon Commission
looks set to recommend later this month a
consent-based, transparent, and flexible
approach to nuclear waste disposal. How-
ever, even strictly applying lessons learned
from Sweden and WIPP won’t guarantee
smooth sailing. For one thing, the technical
challenges of storing nuclear waste safely
for many millennia have not gone away.
In Sweden, for example, after 30 years of
development, the KBS disposal system has
developed what could be a major problem.
Laboratory studies have lately raised the
possibility that the copper cladding meant to
shield the waste from groundwater may be
much more prone to corrosion than anyone
had suspected. The site-approval process,
now under way, will consider just how sig-
nificant a problem that is.
On the social side, the United States is
not Sweden, Swahn points out. “Your polit-
ical system has more difficulties than we
have dealing with these sorts of issues,”
he says. “Trust is important in this country,
and people trust the system. That is very,
very different.”
–RICHARD A. KERR
Awaiting disposal. These casks contain the spent fuel from the production of 110 billion kilowatt-hours
of electricity during 28 years by the Connecticut Yankee nuclear power plant.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 153
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

Y
O
S
H
I
Y
U
K
I

S
U
Z
U
K
I
/
F
I
E
L
D

S
C
I
E
N
C
E

C
E
N
T
E
R

(
T
H
U
M
B
N
A
I
L
S
)
;

M
I
N
O
R
U

I
K
E
D
A
/
F
I
E
L
D

S
C
I
E
N
C
E

C
E
N
T
E
R

(
1
)
;

D
.

N
O
R
M
I
L
E
/
S
C
I
E
N
C
E

(
1
)
SENDAI, JAPAN—The first tremors to arrive
at the chemistry building on Tohoku Uni-
versity’s Aobayama campus on 11 March
were no stronger than others that routinely
rattle this earthquake-prone region. But it
was a matter of course for chemist Hiromi
Tobita to follow the department’s standard
procedure: He left his eighth-floor office
and called for those in the labs to join him
in the hallway to wait out the quake. The
department made safe zones of the
hallways by strictly limiting storage
there to reprints and light, nonvola-
tile supplies kept in lockers bolted to
the walls. The wisdom of that policy
became evident moments later as the
main shocks from the magnitude-9
earthquake sent fume hoods and ven-
tilators, glassware, and chemicals
raining down on the lab benches,
and tossed desks and chairs about
as if they were toys. “If I had stayed
in my office, I would have been seri-
ously injured if not killed,” Tobita
says, as would his colleagues in the
labs. The shaking went on for longer
than any earthquake Tobita had ever felt.
“It was impossible to stand. I was holding
onto a locker; many students lay down on
the floor,” he says. When it subsided, every-
one dashed for the stairs, hearing the hiss of
escaping gas and sniffing toxic chemicals
and smoke from a fire on the seventh floor.
They wondered if they would ever return to
the labs.
With communications cut, no one knew
what was happening off
campus. But Tohoku
geologist Koji Minoura
had an inkling. After
years of matching field
evidence and histori-
cal records, he had pub-
lished a claim in 2001
that in the year 869 C.E.,
during Japan’s Jogan era,
a once-in-1000-years
earthquake triggered a
stupendous tsunami that
ran up to 3 kilometers
inland along hundreds
of kilometers of Japan’s eastern coast. As
he huddled under a table in his sixth-floor
office to avoid the books and papers cascad-
ing from shelves, the ferocity of the shak-
ing made him think: “This is it! This is the
return of Jogan!”
It would be several days before Mino-
ura appreciated how awfully prescient his
2001 paper had been. Some Tohoku col-
leagues found out within minutes. As those
at the Aobayama campus were fleeing build-
ings, a dozen students and staff members
at a marine Field Science Center 54 kilo-
meters east in Onagawa were “in a leisurely
way thinking maybe we should evacuate,”
says Manami Kanno, an assistant professor
who works on marine population genetics.
The center’s two-story buildings rode out
the earthquake with minimal damage. But
town loudspeakers were warning of a “big
tsunami.” “After an earthquake, they often
issue tsunami warnings,” Kanno says. So the
researchers coolly collected their belong-
ings and drove to a designated spot deemed
safe for 10-meter-high waves. They milled
around, not knowing quite what to expect.
Then it came. “It wasn’t water. It was a
mountain of houses, cars, and debris coming
toward us,” Kanno says. They fled on foot
for higher ground. One center technician
had stayed behind and climbed a hill behind
the center snapping pictures as the waters
swamped the first floors, the second floors,
and finally topped the roofs of the build-
ings. Then the tsunami receded as quickly
as it had come, sweeping away their cars,
Picking Up the Pieces
At Ravaged Tohoku University
Four months after the earthquake and tsunami disaster, Tohoku researchers are
finding new meaning in their work
JAPAN DI SASTER
Going, going … The 11
March tsunami swamped a
Tohoku University marine
science center (right), wash-
ing away equipment and
leaving debris (left).
NEWSFOCUS
Zeroed out. The marine population genetics group lost all
of its equipment and most of its data and samples.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 154
research equipment, computers, and almost
every bit of data and sample collected over
the past 8 years.
Tohoku University lost three students. But
through good fortune, neither it nor any other
major university or institute in Japan reported
deaths or casualties among staff members.
The devastation inflicted upon labs and
experiments is a different story. The educa-
tion ministry last May estimated that the
11 March earthquake and tsunami caused
$664 million in damage to buildings and
major facilities at 72 national universities
and institutes. That figure does not include
personal computers, bench-top microscopes,
and other instruments, or research materials.
Factoring such items in, Tohoku University,
the region’s higher education flagship, says
that it alone suffered $990 million in damage.
Money for repairs and
new equipment is flowing,
thanks to an emergency
budget announced in April.
For some scientists, resum-
ing research simply meant
waiting for utilities to come
back online and cleaning
up. “A lot of groups are
already back to work,” says
materials scientist Akihisa
Inoue, president of Tohoku
University.
But scores more face months of idleness.
Some are waiting for major equipment to be
restored (see sidebar, below). Others lost
exceptional or irreplaceable samples and
lab animals. For a few, the tsunami wiped
out a lifetime of work. “We are starting
from zero,” says Minoru
Ikeda, a marine popula-
tion geneticist at Onagawa
field station. He lost all of
the data and samples for a
completed but unpublished
study on genetic diversity of
sweetfish, or Plecoglossus
altivelis, in East Asia that
he had spent 8 years collecting. “There is
nothing else to do but to resume research
activities as quickly as possible,” he says.
As scientists, “we have to produce results.”
For many other researchers, the disaster has
sent them in new directions.
Fear of heights
In Tohoku’s chemistry department, Tobita is
among the unlucky few unable to really get
back to work. Labs on the sixth floor and
below are mostly back to normal. Groups
requiring fume hoods had been located on the
seventh and eighth floors to minimize duct-
work needed to reach rooftop vents. Those
floors are still off-limits because of extensive
damage to walls, ceilings, and equipment,
and the continuing danger from aftershocks.
Plus, Tobita says, the thought of returning to
work on the upper floors “terrifies everyone.”
Tobita lost more than his lab. He works on
inorganometallic compounds, such as silicon
and germanium or silicon and ruthenium.
These little-understood exotic combina-
tions may eventually prove valuable as cata-
lysts. But they are sensitive to air, moisture,
and temperature and were stored in glove
boxes. After the building lost electricity, 100
or more painstakingly created compounds
decomposed.
Tohoku plans to erect a prefabricated
building to house the displaced chemistry
groups, but it won’t be ready until fall. In the
meantime, Tobita’s group is scattered in labs
on two campuses. With no fume hoods and
one glove box, they can only prepare pre-
cursor compounds to use when the new lab
is ready. Tobita says his research plans have
been set back at least 6 months. He is work-
ing on niche compounds, so he’s not worried
about being scooped. But he frets about the
effect of the delay on his graduate students’
thesis work, especially those hoping to get
their degrees next March. The faculty is dis-
cussing how to factor the disaster into stu-
dent evaluations.
The effect on students is a major consid-
eration for Tohoku developmental biologist
Onagawa
Sendai
Tokai
Tsukuba
TOKYO
Naka
Hachinohe
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

D
.

N
O
R
M
I
L
E
/
S
C
I
E
N
C
E
;

I
S
T
O
C
K
P
H
O
T
O
.
C
O
M
NEWSFOCUS
ence buckled roads and damaged site piping. Meanwhile, ground water seeped into the linear
accelerator tunnel. But J-PARC buildings sit on deep piles that protect them from soil liquefaction,
and they withstood the shaking with minimal damage. The 2-year-old, $1.5 billion lab features
a 50-gigaelectronvolt synchrotron supporting physics, materials science, and biomedical studies.
Repairs will take until fall; experiments could resume by the end of the year.
G The Naka Fusion Institute will test superconductive coils for the ITER fusion reactor now under
construction in Cadarache, France. Engineers have declared the test building unsafe, and repairs
will take another 6 months. The status of the test equipment and the impact on ITER’s construc-
tion are unclear. The institute is also home to the JT-60 experimental fusion reactor, now being
upgraded. Mostly unscathed, the revamped reactor expects to see first plasma in 2016.
G Damage to buildings was light at the High Energy Accelerator Research Organization (KEK)
in Tsukuba. But powerful magnets and detectors were shifted out of alignment or knocked to the
floor. Scientists are now realigning beamlines and hope to restart experiments this fall.
G Up the road from KEK, the University of Tsukuba sustained about $87 million in damage to
buildings and research equipment. Most research groups have returned to work after cleaning up
and buying new instruments. But the earthquake may have delivered a knockout blow to Tsuku-
ba’s 12UD Pelletron tandem accelerator, used for nuclear physics and accelerator mass spectrom-
etry. The facility is 35 years old, “and it would be very difficult to restore it,” says physicist Eiji Kita,
who oversees the facility. He says it may be time to explore other research themes. –D.N.
Unscathed. Hallway refuges
prevented injuries to chemist
Hiromi Tobita’s group.
TOKYO—Among major research institutions, Tohoku University in
Sendai was dealt the fiercest blow by the 11 March earthquake and
tsunami (see main text). But scientists across northeastern Japan are
striving to get their research back on track.
G Buffeted by the tsunami while docked in
Hachinohe, the deep-sea drilling vessel Chikyu
lost one of six thrusters that position the ship
during drilling. After temporary repairs, Chikyu
is now undergoing sea trials before embarking
on a commercial oil exploration cruise to run
until the end of the year. Plans for resuming
scientific drilling are uncertain.
G Liquefaction of silty soils took a heavy toll
at the Japan Proton Accelerator Research
Complex (J-PARC) in Tokai. Ground subsid-
Facilities Plot Research Revival
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 155
NEWSFOCUS
C
R
E
D
I
T
S
:

D
E
N
N
I
S

N
O
R
M
I
L
E
/
S
C
I
E
N
C
E

(
3
)
Junken Aoki. He was at a seminar in Tokyo
when the earthquake struck. He quickly
confirmed that his family was safe. He wor-
ried about his team, until a colleague got
through by phone and let him know every-
one was okay. Then Aoki thought about the
4000 zebrafish his group uses to study bone
and blood vessel formation. Temperatures
in Sendai were hovering around 0˚C, and
with the power out, labs were growing cold.
“Below 16˚[C] the fish die,” he says.
The next day, Aoki loaded his car with
water, instant noodles, flashlights, and bat-
teries and drove to Sendai, taking a circu-
itous route over back roads because the
expressway was closed. The normally
5-hour trip took 12 hours. He stayed the
night and then packed a couple of specimens
from each of his 50 lines in special contain-
ers and returned to Tokyo, delivering the fish
to the lab of a friend at Tokyo Medical and
Dental University.
Early last month, Aoki’s group brought
the zebrafish back to Sendai and resumed
their research. Before the quake, he had
also experimented on mice to probe the
metabolic pathways and functions of lyso-
phospholipids. These studies are on hold.
The department had to destroy its 500 mice
because the earthquake cracked walls,
breaching the pathogen-free environment.
It will take a year to breed a sufficient
number of animals to restart research, Aoki
says. That means a change of plan for three
Ph.D. students who were using knockout
mice for thesis research. One is switching
to zebrafish. Another plans in vitro studies.
The third will try blocking a target gene in
wild-type mice. With offers of assistance
pouring in, several group members moved
temporarily to labs at other Japanese univer-
sities, sparking collaborative projects. “That
was one of the good things that came out of
this,” Aoki says.
Hardest hit at Tohoku is the marine pop-
ulation genetics group. “Our equipment is
all gone, almost all of our samples are gone,
and almost all of our data is gone,” Ikeda
says. In its place, the tsunami filled the cen-
ter with fishing gear and other flotsam from
elsewhere in town. And it left, like a pair of
exclamation points, houses perched atop
each of the center’s two buildings.
The center’s small research vessel sur-
vived because two staffers, following
standard procedure, put to sea when the
shaking stopped. Standard procedures
didn’t extend to backing up data to remote
servers before the tsunami carried away
their computers. “We never thought of this
risk,” Ikeda says. One grad student was
away for the day and had his laptop—and
thus his data—with him. Some tissue sam-
ples were recovered, and the researchers
will try to extract DNA.
The group is now working out of a lab
at one of Tohoku’s Sendai campuses, about
2 hours from Onagawa. They are just start-
ing to accumulate basic lab equipment and
arranging to borrow time on sequencers.
They are still mulling over how to restart
their research. One thing that will keep them
busy is a proposal to study the impact of
the tsunami on marine ecosystems and the
recovery of fisheries.
The university is still studying whether
to rebuild the center in its present location,
build anew on higher ground, or move to
another city. Ph.D. student Azuma Kami-
yama isn’t waiting for a decision: She’s going
back to Onagawa. She had lived in Onagawa
for 6 years and counts friends and acquain-
tances among the town’s 872 dead and miss-
ing. Some 80% of the town’s buildings were
washed away; the tsunami left Kamiyama,
who had been living in a seaside dormitory,
with only the clothes on her back and her
iPhone. She has taken leave from the uni-
versity to work for the local government. “I
want to help in the reconstruction of the town
rather than work on research,” she says.
Other scientists also say they were pro-
foundly affected by the disaster. “My life
and work have been completely changed,”
Minoura says. He had never thought about
how his findings could influence public pol-
icy, nor had he courted the media. If Minoura
confirms his suspicions that a similarly mas-
sive tsunami occurred in 869 C.E., farther to
the north on an adjacent segment of the fault
that ruptured in March, he says he will speak
up more forcefully about the risk, particu-
larly to a cluster of nuclear facilities on the
coast smack in the middle of the region. “I
can’t be silent,” he says.
The 11 March disaster has many in the
Tohoku University community thinking
beyond the campus. Dozens of students and
faculty members volunteered in evacuation
centers. And now, Inoue says, “researchers
are thinking of what they can contribute to
the restoration.” One group wants to study
how to strengthen phone systems such that
they function through a disaster. Others have
considered how to make coastal communi-
ties less vulnerable to tsunamis. There are
proposals to investigate new approaches to
energy conservation and alternative elec-
tricity generation. And economists and engi-
neers have ideas for rebuilding the region’s
shattered economy.
To implement this research agenda,
Tohoku plans to open the International
Research Institute of Disaster Science, which
will bring together 300 engineers and scien-
tists from various disciplines. As the region’s
only comprehensive university, Tohoku feels
a responsibility “to take this experience and
use our creativity to contribute to society,”
Inoue says. He hopes the effort will promote
disaster preparedness and postdisaster recov-
ery not only in Tohoku but worldwide.
–DENNIS NORMILE
Back to work. Geologist Koji Minoura (top) searches
for ancient tsunamis, and biologist Junken Aoki
(above) breeds zebrafish. Tohoku President Akihisa
Inoue (below) is planning new programs.
“Researchers
are thinking
of what
they can
contribute
to the
restoration.”
— AKIHISA INOUE,
TOHOKU
UNIVERSITY
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Astronomy in peril?
COMMENTARY
161
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 156
LETTERS
edited by Jennifer Sills
LETTERS I BOOKS I POLICY FORUM I EDUCATION FORUM I PERSPECTIVES
The psychology
of science
159
C
R
E
D
I
T
:

D
A
R
W
I
N
E
K
/
W
I
K
I
M
E
D
I
A

C
O
M
M
O
N
S
Manufacturing Decline
Yields Drug Shortages
FROM OUR VANTAGE POINT WITHIN THE WORLD
of pharmaceutical development and manu-
facturing, we welcome your effort to draw
attention to shortages of the irreplaceable
cancer drug cytarabine (“Shortages of can-
cer drugs put patients, trials at risk,” J. Kaiser,
News & Analysis, 29 April, p. 523). The scar-
city of not only cytarabine, but also numer-
ous other drugs, is an urgent public health
problem that both the drug industry and the
government have an obligation to address. To
do this successfully, both the causes and the
solutions of this issue need to be understood
in their broader context: the neglect of man-
ufacturing, especially high-tech manufactur-
ing, in the United States.
The proximate cause of the most recent
cytarabine shortage was a failure of produc-
tion. However, what turned this error into a
tragedy was a failure of the market. As the
News story explains, the profit margin on
generic drugs such as cytarabine is often very
small. At the same time, the technical and
Seeds of Change for Restoration Ecology
FORESTS PROVIDE A WIDE VARIETY OF ECOSYSTEM SERVICES, INCLUDING PROVISIONS SUCH AS
food and fuel and services that affect climate and water quality (1). In light of the increasing
global population pressure, we must not only conserve, but also restore forests to meet the
increasing demands for ecosystem services and goods
that they provide (2). Ecological restoration has recently
adopted insights from the biodiversity-ecosystem func-
tion (BEF) perspective (3). This emphasis on functional
rather than taxonomic diversity (3, 4), combined with
increasing acceptance of perennial, global-scale effects
on the environment (5, 6) and the associated species
gains and losses (“Terrestrial ecosystem responses to
species gains and losses,” D. A. Wardle et al., Review,
10 June, p. 1273), may be the beginning of a paradigm
shift in forest conservation and restoration ecology. As
a result, we may see increased tolerance toward and the
use of nonnative tree species in forests worldwide.
RAF AERTS
1
* AND OLIVIER HONNAY
2
1
Division Forest, Nature and Landscape, University of Leuven, Celestijnenlaan 200E-2411, BE-3001 Leuven, Belgium.
2
Laboratory of Plant Ecology, University of Leuven, Kasteelpark Arenberg 31-2435, BE-3001 Leuven, Belgium.
*To whom correspondence should be addressed. E-mail: raf.aerts@biw.kuleuven.be
References and Notes
1. Millennium Ecosystem Assessment, Ecosystems and Human Well-Being (Island Press, Washington, DC, 2005).
2. D. Lamb, P. D. Erskine, J. A. Parrotta, Science 310, 1628 (2005).
3. S. Naeem et al., Eds., Biodiversity, Ecosystem Functioning, and Human Wellbeing (Oxford Univ. Press, Oxford, 2009).
4. M. A. Davis et al., Nature 474, 153 (2011).
5. S. H. M. Butchart et al., Science 328, 1164 (2010).
6. M. Hoffmann et al., Science 330, 1503 (2010).
7. R.A. was supported by a postdoctoral fellowship of the Research Foundation–Flanders (FWO).
regulatory burden—especially for cytotoxic
sterile injectibles (again, like cytarabine)—is
as substantial as it is for extremely profitable,
brand-name drugs. Few businesses could be
expected to enter such a market, and indeed,
few have.
In this respect, the pharmaceutical indus-
try is far from exceptional. Business leaders
like Andrew Liveris (CEO, Dow Chemical),
Michael Porter (founder, Monitor Group),
and Andy Grove (CEO, Intel) have sounded
the alarm at the decline of America’s manu-
facturing capacity. They note that manufac-
turing has a tremendous multiplier effect on
economic growth and job creation (1–3).
Nonetheless, a perfect storm of laissez-faire
public policy and bottom-line business deci-
sions has allowed the sector to atrophy. Con-
stituents of both the public and the private
sectors work under the mistaken belief that
American firms can continue to innovate
domestically while leaving manufacturing to
someone else. The flaws in that logic are evi-
dent in our industry. We have seen that R&D
and manufacturing are most effective when
communication is reciprocal, the lessons
learned in the plant feeding back to the draw-
ing board (or the lab bench) (4). As a result, in
many industries, where manufacturing goes,
innovation tends to follow.
The cytarabine shortage shows us that the
wasting away of the U.S. manufacturing sec-
tor may not just shut us out of future pros-
perity; it can actively do harm in the present
day. Today, across the United States, patients
with leukemia are being told that they can’t
have the one drug that could extend or save
their lives.
Letters to the Editor
Letters (~300 words) discuss material published in
Science in the past 3 months or matters of gen-
eral interest. Letters are not acknowledged upon
receipt. Whether published in full or in part, Let-
ters are subject to editing for clarity and space.
Letters submitted, published, or posted elsewhere,
in print or online, will be disqualified. To submit a
Letter, go to www.submit2science.org.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Fat loss
and cancer
Sex & the Red Queen
163 166
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 157
We join with Liveris, Porter, and the like-
minded thinkers mentioned in the News story
in their call for more meaningful govern-
ment incentives to make manufacturing in
the United States a logical choice. Targeted
incentives can be an investment not only in
the future of the economy, but in the future of
individual Americans as well.
PATRICK CONNELLY* AND BRIAN PATRICK QUINN
Vertex Pharmaceuticals, Cambridge, MA 02139, USA.
*To whom correspondence should be addressed. E-mail:
patrick_connelly@vrtx.com
References
1. A. Grove, “How America can create jobs,” Bloom-
berg Businessweek, published online 1 July 2010
(www.businessweek.com/magazine/content/10_28/
b4186048358596.htm).
2. A. N. Liveris, Make It in America: The Case for Re-
Inventing the Economy (Wiley, Hoboken, NJ, 2011).

3. M. E. Porter, M. R. Kramer, Harv. Business Rev. 89, 62
(2011).
4. P. R. Connelly, B. P. Quinn, P. Hurter, J. Condon,
P. Mueller, Pharm. Outsourcing 11, 16 (2010).
NSF’s Struggle
to Articulate Relevance
PUBLIC SCIENCE TODAY FINDS ITSELF CAUGHT
between competing demands: Researchers
need autonomy to pursue questions wher-
ever they lead, whereas funders demand that
research meet societal needs. The National
Science Foundation (NSF) offers a case
study of the balancing of scientific auton-
omy and societal accountability. NSF is
charged with funding basic (i.e., nonmis-
sion) research. Yet Congress funds basic
science in the hope that societal benefits
will result. In 1997, NSF added a “broader
impacts” review criterion to address con-
cerns about relevance: Justify research in
terms of societal outcomes.
Over the past decade our nation’s concern
with accountability has increased, and in
response, the NSF recently issued new draft
criteria for the review of submitted propos-
als (1). The new plan will require researchers
to identify the broader good of their research
by selecting from a list of national priorities.
No doubt, scientists who complained about
the vagueness of the “benefits to society”
clause in the former criterion will welcome
the proposed changes as providing much-
needed clarity and direction to the idea of
“broader impacts.” But specifying impacts
raises three potential problems.
First, the list focuses on economics and
national security, but excludes protecting the
environment and addressing other social prob-
lems. Aside from the consequences of neglect-
ing these areas, this new focus may under-
mine the attractiveness of STEM disciplines
to more idealistic students who are interested
in meeting human needs rather than foster-
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 158
LETTERS
Learn howcurrent events
are impacting your work.
ScienceInsider, the newpolicy blog fromthe journal
Science, is your source for breaking news and instant
analysis fromthe nexus of politics and science.
Produced by an international teamof science journalists,
ScienceInsider offers hard-hitting coverage on a range of
issues including climate change, bioterrorism, research
funding, and more.
Before research happens at the bench, science policy is
formulated in the halls of government. Make sure you
understand howcurrent events are impacting your work.
Read ScienceInsider today.
www.ScienceInsider.org
Breaking news and analysis from
the world of science policy
ing economic competitiveness. Second, under
the proposed new criteria, applicants and
reviewers are restricted to the provided list of
national needs, which will complicate efforts
to respond to new challenges as they develop.
Third, addressing these national needs is
now supposed to happen “collectively.” This
reopens the question of whether each individ-
ual proposal must address broader impacts.
The new criterion thus replaces vagueness
regarding what counts as a broader impact
with vagueness regarding who is responsible
for addressing broader impacts.
The new criteria are not without merit.
For example, the guidelines on how to
implement proposed broader impacts are
improved. Once one identifies the national
goal to be pursued, the new broader impacts
criterion focuses on logistical questions that
should be asked in peer review.
The proposed changes in the merit review
criteria move too far in the direction of
accountability, at the cost of scientific cre-
ativity and autonomy. The set of principles
(in terms of national goals) also suffers from
excessive detail at the cost of flexibility. Of
course, revising the criteria is a perennial
process of renegotiation as cultural values
change, and these are only draft criteria and
principles. NSF invites comments until 14
July. Both the scientific community and the
science policy community need to make their
voices heard. ROBERT FRODEMAN*
AND J. BRITT HOLBROOK
Center for the Study of Interdisciplinarity, University of
North Texas, Denton, TX 76203–5017, USA.
*To whom correspondence should be addressed. E-mail:
frodeman@unt.edu
Reference
1. National Science Board, “NSB/NSF seeks input on
proposed merit review criteria revision and principles”
(http://nsf.gov/nsb/publications/2011/06_mrtf.jsp).
CORRECTIONS AND CLARIFICATIONS
Reports: “Realizing all-spin–based logic operations atom by
atom” by A. A. Khajetoorians et al. (27 May, p. 1062). The fol-
lowing errors were introduced during proofs. On page 1063,
the third complete sentence should read, “In the next step,
the spin lead is built atom-by-atom by subsequently adding
Fe atoms with an interatomic distance d = 0.923 nm, where
the interatomic exchange coupling is antiferromagnetic
(|J
l
| » 0.1 meV) (Fig. 2, A to E) for spin leads with lengths of
up to six atoms.” On the same page, the third sentence of
the second complete paragraph should begin, “Given that the
exchange interaction between each spin lead and its island
J
isl
dominates, and that the mutual interaction between the
end atoms in both leads is smaller than J
l
, which is….” The
caption above Table 1 should read, “Possible logical opera-
tions as a function of the relative orientation of biasing field
and tip magnetization, and of the parity of each spin lead.”
On page 1063, third column, second paragraph, the equation
in the third sentence was incorrect. The correct equation is
B
crit
= –|J
α
+ J
β
|/m ≈ –0.05 meV/3.5 µB = –0.25 T.
News Focus: “New work reinforces megaquake’s harsh
lessons in geoscience” by R. A. Kerr (20 May, p. 911).
The story gave incorrect URLs for three Science papers
it cited. The correct URLS are www.sciencemag.org/
content/early/2011/05/18/science.1206731 (Simons),
www.sciencemag.org/content/early/2011/05/18/sci-
ence.1207020 (Ide), and www.sciencemag.org/content/
early/2011/05/18/science.1207401 (Sato). The online HTML
version has been corrected.
Education Forum: “Inquiry-based writing in the laboratory
course” by C. Moskovitz and D. Kellogg (20 May, p. 919).
Two similar sentences appear in the first column on page
920. One was mistakenly included when a correction was
made. The correct sentence is: “Now, imagine that students
are given, at random, either contaminated or uncontami-
nated reagents, but they do not know who received which.”
Letters: “Counting India’s wild tigers reliably” by K. U.
Karanth et al. (13 May, p. 791). The photo credit should
have been “Pallava Bagla.”
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

BOOKS ET AL.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011
159
C
R
E
D
I
T
:

M
A
R
K

M
I
L
L
E
N
E
/
W
W
W
.
M
I
L
L
E
N
E
.
C
O
M
A
book titled “Psychology as Science”
would be unremarkable. But Science
as Psychology? What can that be?
The subtitle gives hints but no clear answer.
For that, one must turn to the “Introduction,”
where the authors introduce two quite differ-
ent but related large problems: “integration”
and “grab-bag general psychology.” The first
concerns how one can integrate the “cogni-
tive” and the “social” within science studies.
The second underlies the authors’ desire to
reform “general psychology.” The connec-
tion between the two problems is that Lisa
Osbeck, Nancy Nersessian, Kareen Malone,
and Wendy Newstetter believe their approach
to general psychology promotes the integra-
tion they desire in science studies. In keeping
with my own interests and expertise, here I
focus on the first problem, the one of greater
interest for general scientific readers.
The work’s major contributions lie not in
the conclusions reached but in the richness of
its starting points. One is its “unit of analysis,”
which they characterize as “the acting person
in normatively structured contexts of practice
within which sense-making and identity are
two central tasks.” Or, less tendentiously, “the
acting person as scientist.” By way of contrast,
a favorite unit of analysis in the philosophy of
science is something abstract, such as
a scientific theory, and in the sociol-
ogy of science something larger, such
as a social network. Many historians
of science could claim an acting per-
son as their unit of analysis, although
they would not describe their work in
this manner. Adopting an acting person
as basic could lead philosophers of sci-
ence to think of theories as representa-
tional devices created and modified by
scientists and lead sociologists of sci-
ence to focus more on the particularity
of the actors in their networks.
The potential for integration in sci-
ence studies is real because the sug-
gested concept of the acting person
as scientist is quite rich. Acting per-
sons are embodied, so they necessar-
ily exhibit qualities such as race and
gender. Acting persons also differ in
their experience, education, and cog-
nitive abilities. Most important, act-
ing persons are enculturated; in fact they are
members of various cultures. In this book, the
culture of interest is primarily that of individ-
ual laboratories, although the
larger cultures of universities,
scientific fields, and nations
also play a role. The cognitive
and the social come together in
the individual acting scientist.
The authors devote five of
their eight chapters to a sin-
gle large empirical case study.
The first and last chapters are
historical and theoretical. The
second explains the authors’
methodology, which is thor-
oughly qualitative and broadly characterized
as “ethnographic observation.” The subjects
were members of two biomedical engineer-
ing laboratories at a large research university.
One was a well-established tissue-engineer-
ing laboratory directed by a senior scientist;
the second, a new neural engineering labora-
tory directed by a much younger researcher.
Both labs comprised roughly 15 members,
including the director, a lab manager, a post-
doctoral researcher, several doctoral stu-
dents, a couple of master’s students, and a
few undergraduates. The authors thoroughly
familiarized themselves with the laborato-
ries: reading grant proposals, attending labo-
ratory meetings, and generally “hanging out”
in the labs. Their principal data, however,
consist of a series of unstructured interviews
(about 75 for each lab) extending over a two-
year period. “Our process,” they write, “was
broadly informed by grounded theory … in
the sense that we endeavored to use constant
comparison and analytic induction to develop
coding schemes and concep-
tual categories to character-
ize the laboratory practices
we observed and the discur-
sive productions obtained
through interviews.” But,
they immediately continue,
“it is most accurate to say
that we used a variety of
interpretative strategies.”
Their ground-up strategy
(my term) is exemplified
by the fact that “the role of
emotion in research practices” (the topic of
the chapter “The Feeling Person”) was not
originally part of the project but emerged
only from their analysis of the interviews.
The authors’ second major innovation is to
consider a laboratory as a distributed cogni-
tive system:
We propose extending the traditional
notions of “mental model” and “men-
tal modeling” by means of the notion of
“distributed model-based cognition”….
On our interpretation, the device model-
system and the researcher mental models
constitute a distributed inferential sys-
tem through which problem solutions are
achieved by means of simulation pro-
cesses. We understand this to mean that
part of the simulation process occurs in
the mind or imagination and part in the
real-world manipulation of the model.
The real-world models they have in mind
are artificial blood vessels and dishes of neu-
ronal cells. Distributed cognition is not an
idea that emerges from their research, but
their study well illustrates its usefulness in
understanding scientific practice. Scientific
cognition, especially problem solving, is not
something that goes on just in the heads of sci-
entists. It is a process of interaction among a
group of scientists together with the artifacts
and instruments that populate modern labora-
tories. Science as Psychology contains many
insights but few memorable general conclu-
sions. The best I could find is: “Glaringly evi-
dent in the interview material is that cognitive,
social, material, cultural/historical, and affec-
Distributed Cognition in the Lab
SCIENCE STUDIES
Ronald N. Giere
*
*E-mail: giere@umn.edu
Science as Psychology
Sense-Making and Identity
in Science Practice

by Lisa M. Osbeck,
Nancy J. Nersessian,
Kareen R. Malone, and
Wendy C. Newstetter
Cambridge University Press,
Cambridge, 2011. 287 pp. $90,
£55. ISBN 9780521882071.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

BOOKS ET AL.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 160
M
athematics used to be an invisible
culture. Today, however, headlines
announce solutions of major prob-
lems (Fermat, Poincaré), journalists explain
extreme weather using esoteric mathematical
theories (“butterfly effect”), and consultants
promote the economic potential of powerful
algorithms (data mining, three-dimensional
modeling). Popular plays and movies (e.g.,
Arcadia, Proof, and A Beautiful Mind) and
the television show Numb3rs have brought
mathematicians into the theater and people’s
living rooms. At the same time, educators
worry about sagging mathematics test scores,
parents fret about the rigor of their children’s
math education, and students agonize about
preparing for SAT and Advanced Placement
exams. One consequence has been a spate of
passionate “math wars” that have riven par-
ents, teachers, and school boards across the
United States.
As public interest in mathematics has
grown, so has a popular literature of books
Distilling Truth
from Emotion
MATHEMATICS
Lynn Arthur Steen
The reviewer is at St. Olaf College, Northfield, MN 55057,
USA. E-mail: steen@stolaf.edu
and articles written to explain mathematics
to the general public. In Loving and Hating
Mathematics, retired mathematician Reuben
Hersh and linguist Vera John-Steiner (both
at the University of New Mexico) seek to
explain mathematicians, rather than math-
ematics. They do this primarily through an
anthology of anecdotes, many oft-told, drawn
largely from biographies of 20th-century
mathematicians. Chapter notes document
sources for these vignettes (some legend-
ary, a few likely apocryphal), and appendices
offer annotated bibliographies of published
biographies along with
very brief biographies of
over 250 mathematicians
named in the book.
A major goal of the
volume is to overcome a
variety of “myths” about
mathematicians and, by
extension, mathematics.
To refute the notion that
mathematicians lack emo-
tional complexity, the authors offer examples
of mathematicians’ early traits (intense curi-
osity and self-confidence) and of tensions
between competition and community that
form the social glue of mathematical culture.
Further examples show how mathematics has
served some as an escape from unpleasant
reality, notably war and prison. A discussion
of the risks that mathematics poses for minds
susceptible to addiction and madness ends
with Kurt Gödel (who, fearing food was poi-
son, starved himself to death) and the “Una-
bomber,” Ted Kaczynski (whose mathemat-
ics papers were “paragons of precision”).
To rebut the widespread idea that math-
ematicians are recluses engaged in solitary
pursuits, Hersh and John-Steiner offer sev-
eral extended stories of intense, productive
friendships (e.g., Grace Chisholm and Wil-
liam Young) and effective communities (e.g.,
Bourbaki and Göttingen) that have both nur-
tured mathematicians and advanced math-
ematics. Several personal accounts illustrate
the struggles of women and minorities to
be taken seriously in a mathematical culture
long dominated by white males. To refute
the myth that mathematics is for the young
(probably of concern only to some mathema-
ticians), the authors offer examples of excep-
tional work by older mathematicians as well
as comments gleaned from an informal sur-
vey of retired mathematicians.
In concluding, the authors try—less suc-
cessfully—to counter the widespread belief
that mathematics is an effective filter for
higher education. Because mathematicians
are “thinking creatures of flesh and blood”
who, “like all people, think socially and emo-
tionally,” it follows, Hersh and John-Steiner
argue, that mathematicians’ actions in teach-
ing and research are freighted with ethical
implications. Morever, they assert, math-
ematical knowledge, although “relevant for
engineering,” is much less appropriate for
other professions. “[D]o you know a doctor,
lawyer, or business person who uses calculus,
or even an algebraic equation or a theorem
from geometry?”
It follows in their analysis that the use of
mathematics “to decide who may get into
graduate or professional
programs” is ethically
questionable. The authors
extend this reasoning
down into the undergradu-
ate and high school years,
arguing that mathematics
education would be signifi-
cantly improved if students
were free to decide which
mathematics to take and
when to take it. In particular, they contend that
using calculus as a filter is counterproductive
to sound education—a belief shared by many
college and university mathematicians.
Notwithstanding the idealism of the
authors’ case, it would take substantial empir-
ical evidence to convincingly refute the
widely held view that mathematics is a useful
filter for higher education. Hersh and John-
Steiner offer examples and opinions that
support their views but make little effort to
counter the deep emotional, social, and cur-
ricular pressures that have created and main-
tain mathematics as a filter. Without an analy-
sis of the forces for stasis, their proposals for
reform are unpersuasive.
In the end, one is left uncertain about
the intended audience for the book, which
seems in many ways a rich 20th-century
sequel to Robert Moritz’s classic Memora-
bilia Mathematica ( 1). Its central appeal will
be to prospective mathematicians and others
working closely with (or married to) math-
ematicians, who will find “a native’s guide to
the culture of mathematics.” (And that might
be a more illuminating title than Loving and
Hating Mathematics, which is both mysteri-
ous and misleading.) Wisely used, the book
could infuse the important issue of students’
emotions into debates about mathematics
education and in that way support some of
the authors’ desired educational goals.
References
1. R. E. Moritz, Memorabilia Mathematica (Macmillan, New
York, 1914).
Loving and Hating Mathematics
Challenging the Myths of
Mathematical Life
by Reuben Hersh and
Vera John-Steiner
Princeton University Press,
Princeton, NJ, 2011. 428 pp. $29.95,
£20.95. ISBN 9780691142470.
10.1126/science.1208719
tive dimensions of human practice intertwine
in ways that are vital to problem solving and
innovation in the laboratory.” However, the
authors do demonstrate the unifying power of
their principal ideas—the unit of analysis as
the acting person and the laboratory as a site
of distributed cognition. These alone make
their study a singular contribution to the psy-
chology of science. Its only recent competi-
tor of comparable scope is Gregory Feist’s The
Psychology of Science and the Origins of the
Scientific Mind ( 1). The authors do not say so,
but Feist’s book seems an example of what
they reject as “grab-bag general psychology”
applied to science. Nevertheless, that there
can be so strong a clash at such a fundamen-
tal methodological level is itself a sign that
maybe the psychology of science is finally
ready to take its rightful place among the other
major components of science studies.
References
1. G. Feist, The Psychology of Science and the Origins of the
Scientific Mind (Yale Univ. Press, New Haven, CT, 2006).
10.1126/science.1207754
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 161
POLICYFORUM
A
rtificial satellites
have given astrono-
mers access to hid-
den portions of the elec-
tromagnetic spectrum and
dramatically changed our
perception of the universe.
Space-based astronomy
remains an integral part
of contemporary research,
resting on powerful tele-
scopes developed by major
space agencies—the U.S.
National Aeronautics and
Space Admi ni st rat i on
(NASA), the European
Space Agency (ESA), and
the Japan Aerospace Explo-
ration Agency (JAXA).
Together, these telescopes
encompass the whole elec-
tromagnetic spectrum: e.g., Fermi observa-
tory and International Gamma-Ray Astro-
physics Laboratory (INTEGRAL) at gamma-
ray energies, Chandra, and X-ray Multi-Mir-
ror Mission (XMM)–Newton at x-rays, Hub-
ble at near-ultraviolet and visual wavelengths,
and Herschel and Planck in the infrared and
submillimeter domain. Although all of these
are now in operation, some will reach the
end of their lifetimes during the next 5 years.
In the high-energy domain, no major x-ray
and gamma-ray observatories will be avail-
able to replace XMM-Newton, Chandra,
INTEGRAL, and Fermi ( 1).
Future demands for larger apertures and
more complex detectors will impose heavier
demands on the limited resources of space
agencies. For example, NASA’s development
of the James Webb Space Telescope (JWST)
is blocking the start of other new large ven-
tures that had been anticipated ( 2). The cost
of JWST has increased almost 10-fold since
its start in the mid-1990s, and its launch is
delayed until 2018 from the original goal
of 2007. Similarly, within the ESA Cosmic
Vision program ( 3), no new large mission has
been approved in the last 10 years. We believe
these problems have to do with lack of global
strategic planning and cost capping and that
lessons learned during the ESA Horizon
2000 program could help.
Astronet and the Decadal Surveys
Several disconnected survey efforts in
Europe and the United States reflect the
lack of global strategic planning. In 2007,
a joint European initiative (Astronet) devel-
oped a comprehensive “Science Vision for
European Astronomy.” This included both
ground- and space-based segments aimed
at answering by 2025 the highest-priority
scientific questions. Drawing on the ESA
Cosmic Vision program, the Astronet infra-
structure roadmap for European astronomy
was set up in 2008 ( 4). First priority for large
space facilities was given to a gravitational
wave observatory [the Laser Interferometer
Space Antenna (LISA)] and to the Interna-
tional X-ray Observatory (IXO) (see the fig-
ure), both missions envisaged as coopera-
tively shared by ESA and NASA .
A similar, but disconnected, decadal sur-
vey exercise was completed in 2010 by the
U.S. National Academies of Science ( 5). In
astronomy, the science objectives for 2013–
2022 are a Wide-Field Infrared Survey
Telescope (WFIRST) to address questions
in dark energy and exoplanet research, an
Explorer program to rapidly respond to new
scientific and technological breakthroughs,
LISA, and IXO. But JWST
cost overruns cast doubts on
the feasibility of developing
WFIRST, LISA, and IXO
within the decade ( 6, 7).
In March 2011, ESA
announced that IXO coopera-
tion with NASA was no longer
feasible because of incompat-
ible schedules and budgets ( 8).
All partners involved in merg-
ing the NASA Constellation-X
and ESA X-ray Evolving Uni-
verse Spectroscopy (XEUS)
projects into IXO, successful
after a 10-year process, were
put back where they started.
A similar situation affects the
infrared research community
because of the failure to nego-
tiate a compromise for merg-
ing WFIRST with the ESA Euclid mission,
both aiming to map the geometry of the uni-
verse as shaped by its dark matter and dark
energy. Planetary missions do not fare bet-
ter. At a NASA-ESA review of their joint
Mars exploration program, NASA informed
ESA that anticipated scenarios for interna-
tional cooperation had to be revisited because
of lack of financial means ( 9). International
cooperation, in particular between ESA and
NASA, faces a major impasse, with “sur-
veys” doomed to obsolescence in terms of
timeliness and affordability.
Programmatic “Design-to-Cost” Approach
Although the global economic downturn
certainly affected space science funding, we
think deeper management issues are largely
to blame for stagnation in the development of
global space astronomy. Although priorities
were set in the Astronet and decadal surveys,
they basically entail a “wish list” of proj-
ects, not embedded in a coherent program
that could be assessed in an integral way
regarding scientific balance, technical feasi-
bility, cost, and international collaboration.
This severely limits timely implementation
and cost control. When the cost of individual
missions exceeds annual national space sci-
ence budgets, a balanced long-term (10- to
20-year) program should be established that
fits into an overall multiyear budget envelope,
coupled to a strict “design-to-cost” imple-
A Dark Age for Space Astronomy?
ASTROPHYSICS
Roger-Maurice Bonnet
1
* and Johan A. M. Bleeker
2 3

Horizon 2000 offers lessons to improve global
planning and priority setting and to implement
on-time, on-budget programs.
C
R
E
D
I
T
:

E
S
A
;

B
A
C
K
G
R
O
U
N
D

I
M
A
G
E
:

E
S
A
/
N
A
S
A
,

T
H
E

A
V
O

P
R
O
J
E
C
T
,

A
N
D

P
A
O
L
O

P
A
D
O
V
A
N
I
*Author for correspondence. E-mail: rmbonnet@issibern.ch
1
ISSI: International Space Science Institute, CH-3012 Bern,
Switzerland.
2
SRON: Netherlands Institute for Space Research,
NL-3584 Utrecht, Netherlands.
3
Astronomical Institute, Uni-
versity of Utrecht, NL-3508 Utrecht, Netherlands.
Artist conception of the IXO.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 162
POLICYFORUM
mentation of all program elements, that is,
constrain design goals of individual missions
according to allocated funds. This approach
was taken in ESA’s space-science Horizon
2000 program that we helped formulate in
1984 ( 10), and it proved to be a strong asset in
maintaining both the scientific scope and the
cost-to-completion of that program.
Under this approach, a survey committee,
in consultation with the space science com-
munity, selected science priorities by iden-
tifying a few large $1 billion missions (cor-
nerstones) and associated programs for real-
ization of cutting-edge technologies. This
ensured a good combination of state-of-the-
art tools and reliable cost control. Corner-
stone missions were to be implemented only
when the technology was proven and the mis-
sion science requirements frozen, possibly
after a descope, or scaling back. This avoided
frequent and costly redesigns, which helped
minimize the time needed for industrial
development. Although the user communities
did not know a priori the launch date of their
cornerstone mission, they were assured from
the outset that it would not be reopened for
competition or even cancellation, and once
development of the mission started, it would
be developed in time.
To balance the program, to address spe-
cific science objectives, and to respond to
emerging discoveries, a set of medium and
small missions ($300 million to $500 mil-
lion) were introduced for competitive selec-
tion during Horizon 2000. If their estimated
cost was above the limit, a descope was man-
datory before development. The full program
had to be implemented within a predefined
budget envelope that required augmentation
of the ESA scientific budget by more than
50% over a period of 7 years. Approval was
obtained in 1985 for initiating the program,
including the 50% increase, to be reached
over a period of 10 years. This would allow
completion of the full program in 22 years,
instead of 20, that is, in 2007. In reality, it
was completed in 2009 with only 2 years’
delay, relatively minor compared with the
10 years’ delay currently facing JWST. Hori-
zon 2000 saw the launches of all corner-
stones [Solar and Heliospheric Observatory
“SOHO,” Cluster, Rosetta, XMM-Newton,
and Herschel (the largest space telescope
ever launched)] all presently operating, and
12 competitively selected medium-size mis-
sions embodying the required flexibility in
the program. Relaunch of the four Cluster
satellites, victims of a 1996 launcher failure,
was included, with no extra funds allocated.
The ESA design-to-cost approach allowed
the promised balance, as all missions stayed
within original fixed financial limits. Corner-
stones were open to international cooperation
but could also be implemented in a purely
European context, although with less ambi-
tious science objectives as a consequence of
cost capping. Although Horizon 2000 became
an international reference ( 11), NASA never
attempted such an approach, admittedly more
difficult to implement in the U.S. system of
yearly budget allocations as opposed to the
long-term financial commitment necessary
in the case of Horizon 2000.
We believe the Horizon 2000 approach
offers a solution to the present budgetary and
schedule difficulties facing the space agen-
cies. Key ingredients are strategic long-term
planning of science priorities embodied in
large space observatories, flexibility through
smaller-scale missions to address science
niches and respond to emerging discover-
ies, and cost capping through a design-to-
cost philosophy for the individual missions.
Long-term strategic planning on a global
scale is emphatically missing now in space
agencies’ planning. The present piecemeal
individual project approach leads to open-
ended design, budgeting, and scheduling.
The JWST problem would not have arisen
in its present magnitude if JWST had been
part of a coherent program with political and
financial approval at the outset, incorporat-
ing other missions to be implemented within
fixed schedules and budget. In the case of
Cosmic Vision, ESA abandoned the Hori-
zon 2000 approach and has since not been
able to set priorities and decide on a coherent
sequence of new missions.
Toward a New Global Planning Process
We believe large-scale space facilities required
for ground-breaking astronomy in the coming
decades can only be accommodated by global
pooling of financial resources and technologi-
cal expertise. The disconnected approach with
separate surveys in Europe, the United States,
and elsewhere is no longer viable. The lack of
a common strategy is becoming a major stum-
bling block. Strategic long-term planning is
needed to establish a science-driven roadmap
based on global cooperation with clearly iden-
tified science priorities.
Recently, the Committee on Space
Research (COSPAR) and the International
Astronomical Union (IAU) determined
that a planning exercise on a world scale is
needed. COSPAR set up a working group
aimed at establishing a global program in
space astronomy ( 1). That program will be
integrated within a similar program formu-
lated by the IAU for ground-based astron-
omy and should result in a “global decadal
survey” for contemporary astronomy.
However, implementation of this global
survey has to reside in the space agencies by
forging an interagency coordination group
on large observatory–class space astronomy
facilities. Agencies have already done this
successfully for other efforts, e.g., the Com-
mittee on Earth Observation Satellites, the
International Space Exploration Coordina-
tion Group, and the International Committee
on Global Navigation Satellite Systems, with-
out the necessity for intergovernmental agree-
ments on budgets. As was the case of Hori-
zon 2000 for the cornerstone projects, such
an interagency group would have to agree
on major flagship missions to be developed,
contingent on an agreed mission concept, on
the development of necessary technologies,
and on a cost-to-completion budget. Such an
interagency roadmap should extend over at
least two decades, given the long lead times
for development and qualification of tech-
nologies. This interagency group, involving
the leading scientists, should also distribute,
up front, the leadership role for each selected
mission. Medium- and smaller-scale missions
will safeguard the flexible part in the program
and need to be selected in competition.
New resources may also come from new
partners that are demonstrating increas-
ing capability in space science, like Russia,
China, India, and Brazil. They have impres-
sive plans for the future—including col-
laborative ventures with NASA, ESA, and
JAXA—and need to be involved in discuss-
ing priorities for interagency roadmaps. A
dark age for space astronomy can be avoided
through vision and political will, with clear
appreciation that leadership of large space
missions should be globally shared.
References
1. P. Ubertini et al., Future of Space Astronomy: A Global
Road Map for the Next Decades (Report V2.4, COSPAR-
Working Group on the Future of Space Astronomy,
COSPAR, Paris, 2011); http://cosparhq.cnes.fr/
AstronomyWG_CSAC_v22-mar-2011.pdf.
2. L. Billings, Nature 467, 1028 (2010).
3. G. F. Bignami, P. Cargill, B. Schutz, C. Turon, Cosmic
Vision: Space Science for Europe 2015-2025 (ESA
BR-247, ESA, Noordwijk, Netherlands, 2005).
4. M. F. Bode, M. J. Cruz, F. J. Molster, Eds., The Astronet
Infrastructure Roadmap, A Strategic Plan for European
Astronomy (Astronet, Paris, 2008).
5. National Research Council, New Worlds, New Horizons in
Astronomy and Astrophysics (National Academies of Sci-
ence, Washington, DC, 2010).
6. A. Lawler, Y. Bhattacharjee, Science 330, 1028 (2010).
7. Y. Bhattacharjee, Science 331, 1121 (2011).
8. E. Samuel Reich, Nature 471, 421 (2011).
9. P. de Selding, Space News, 25 April 2011.
10. R. M. Bonnet, J. A. M. Bleeker, Space Science Horizon
2000 (ESA Special publ. 1070, ESA, Paris, 1985).
11. D. Clery, Science 261, 540 (1993).
10.1126/science.1206409
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011
163
PERSPECTIVES
Lipases in Cachexia
MEDICINE
Peter Arner
Enzymes expressed in adipose tissue are
involved in the loss of fat mass that occurs
in a cancer-associated wasting disorder.
C
achexia is a wasting dis-
order that is frequently
observed in cancer—
particularly in gastrointestinal
malignancy—and is associated
with poor response to chemother-
apy and decreased survival ( 1).
The most prominent symptom is
uncontrolled loss of body weight
due to depletion of adipose tissue
and skeletal muscle. There is no
effective cure for cancer cachexia,
and the underlying mechanisms
are poorly understood. Longitudi-
nal studies of humans suggest that
loss of fat mass is an early event
in the pathogenesis of the condi-
tion ( 2). In humans, this deple-
tion is not due to loss of fat cells
(adipocytes), but is attributed to a
decrease in lipids stored in these
cells, causing them to be smaller
( 3). On page 233 of this issue, Das
et al. ( 4) present strong evidence
that lipases in adipose tissue break
down stored fat, contributing to
cancer-associated cachexia.
The lipid droplet makes up
most of the adipocyte volume,
and the fat consists predominantly
of triglycerides. The breakdown of these tri-
glycerides in fat cells (lipolysis) is regulated
by two enzymes. Adipose triglyceride lipase
(ATGL) catalyzes the first step (formation of
diglycerides), and hormone-sensitive lipase
(HSL) catalyzes further hydrolysis to gener-
ate the end products fatty acids and glycerol.
The involvement of increased adipocyte
lipolysis in cancer cachexia–associated fat
loss was suggested by the increased circulat-
ing concentrations of fatty acids and glycerol
observed in patients ( 1). In vivo studies in
humans also demonstrated increased whole-
body lipolysis rates in cancer cachexia ( 5).
In fat cells from patients with the syndrome,
the lipolytic effect of catecholamines—
the major lipolysis-stimulating hormones
in humans—is enhanced in vitro, and the
amounts of HSL, and to some extent ATGL,
are increased in the adipocytes ( 6). These
results implicated adipocyte lipases in the
loss of fat mass in cancer cachexia.
Das et al. investigated the cachexia
effect of lung carcinoma or melanoma in
mice. The growth of these malignant tumors
caused increased rates of lipolysis, loss of
fat mass, and reduction of skeletal muscle
volume in wild-type mice, which was inde-
pendent of food intake. However, mice with
ATGL or HSL deficiency were resistant to
the cachexia effects of the tumors, and the
protective effect was strongest in ATGL
deficiency. In the wild-type mice and also in
parallel human studies, the authors observed
that the enzymatic activities of the two
lipases were increased and were correlated
with the severity of cachexia.
Although one might expect lipases to be
involved in loss of fat mass in cancer cachexia,
their effect on muscle wasting observed by
Das et al. is surprising. A decrease of muscle
mass appears after fat mass loss, at later and
often final stages of cancer cachexia ( 2). This
muscle loss involves decreased synthesis and
increased degradation of muscle
protein as well as programmed
cell death (apoptosis) of muscle
cells ( 1). The authors speculate
that the lipase effect on muscle
is indirect and induced by fatty
acids that are produced through
accelerated lipolysis in fat cells
of cachexia mice. The oxidation
of fatty acids is increased in skel-
etal muscle in cancer cachexia
( 1), which was confirmed by Das
et al. In muscle, glucose and fatty
acids compete as energy sub-
strates for oxidation. An overrid-
ing role of fatty acid oxidation
could have secondary effects on
protein metabolism and turnover
of muscle cells. However, these
assumptions need to be sup-
ported by direct experiments.
Cancer cachexia might be
caused by circulating factors that
are directly or indirectly gener-
ated by the tumor (see the figure).
Cytokines, in particular tumor
necrosis factor–α, have been pro-
posed as such factors. However,
their effects on lipolysis do not
resemble the lipolytic phenotype
of cancer cachexia ( 6). By contrast, cytokines
may be involved in the depletion of muscle
mass ( 1). For adipose tissue lipolysis, zinc-α2
glycoprotein-1 (AZGP1) is of particular
interest ( 1). AZGP1 is an abundant circulat-
ing protein that enhances the lipolytic effect
of catecholamines on fat cells. It is produced
by the host and certain tumors. Adipose tis-
sue secretes large amounts of AZGP1, and
this is accelerated in cancer cachexia patients
and correlates with the reported weight loss
( 7). Das et al. observed an increase in AZGP1
concentration in cachexia mice. Other circu-
lating lipolytic factors derived from malig-
nant tumors have also been described but
have not yet been defined ( 8).
It is tempting to speculate that drugs tar-
geted to adipose lipases could prevent cancer
cachexia and its devastating effects. Selective
inhibitors of HSL have already been devel-
oped ( 9), one of which blocks HSL-induced
lipolysis in fat cells of cancer cachexia
patients ( 6). Mice lacking HSL have a rather
benign phenotype. Although Das et al. report
Adipose:
•Lipase activation
(ATGL, HSL)
•Fatty acid release
Tumor
Secreted factors Other factors:
•Catecholamines
•Natriuretic peptides
Brain
Liver
Muscle
Cachexia:
Fat loss
Muscle loss
Cancer cachexia. A possible model of action for circulating lipolytic factors is
shown. Factors released from a tumor act on adipocytes, increasing the expres-
sion and function of ATGL and HSL. This increases the lipolytic action of regula-
tory hormones (catecholamines and natriuretic peptides). Tumor-derived factors
may also act on nonadipose organs such as the liver and brain, causing them to
produce secondary circulatory lipolytic factors.
C
R
E
D
I
T
:

B
.

S
T
R
A
U
C
H
/
S
C
I
E
N
C
E
Department of Medicine, Karolinska Institutet at Karolin-
ska University Hospital, 141 86 Stockholm, Sweden. E-mail:
peter.arner@ki.se
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 164
PERSPECTIVES
that mice lacking ATGL have a stronger pro-
tective effect against cachexia compared to
those lacking HSL, ATGL deficiency in mice
and humans is associated with major impair-
ment of cardiac function ( 10). Further char-
acterization of the functions of both lipases
may inform the development of pharmaco-
therapeutic strategies to treat cachexia asso-
ciated with cancer and other chronic diseases.
The findings by Das et al. raise the ques-
tion of the role of lipases in muscle loss that
occurs in cachexia. ATGL and HSL also
regulate lipolysis in muscle cells, so myo-
cyte lipolysis could be involved. Further-
more, whether lipases are important for
other cachexia conditions, such as severe
infections, pulmonary or cardiac insuf-
ficiency, and kidney failure, is unclear.
Heart insufficiency is of particular interest
because cachexia in patients with this con-
dition is thought to involve increased circu-
lating concentrations of natriuretic peptides
( 11). These hormones have strong lipolytic
effects, but only in fat cells of humans and
other primates. Notably, the lipolytic effect
of natriuretic peptides is increased in fat cells
of cancer cachexia patients ( 6). The animal
models of cachexia developed by Das et al.
should further increase our understanding of
the pathophysiology behind this condition.
References
1. M. J. Tisdale, Physiol. Rev. 89, 381 (2009).
2. M. Fouladiun et al., Cancer 103, 2189 (2005).
3. M. Rydén et al., Cancer 113, 1695 (2008).
4. S. K. Das et al., Science 333, 233 (2011);
10.1126/science.1198973.
5. S. Klein, R. R. Wolfe, J. Clin. Invest. 86, 1403 (1990).
6. T. Agustsson et al., Cancer Res. 67, 5531 (2007).
7. T. Mracek et al., Br. J. Cancer 104, 441 (2011).
8. L. O. Byerley et al., Nutr. Cancer 62, 484 (2010).
9. T. H. Claus et al., J. Pharmacol. Exp. Ther. 315, 1396
(2005).
10. A. Lass, R. Zimmermann, M. Oberer, R. Zechner, Prog.
Lipid Res. 50, 14 (2011).
11. M. Lainscak, S. von Haehling, S. D. Anker, Int. J. Cardiol.
132, 303 (2009).
10.1126/science.1209418
Precision, Not Power
PHYSICS
Edward E. Eyler
A combination of optical trapping and
frequency combs is used to precisely probe
an extraordinarily weak transition in
atomic helium.
T
he essence of precision measure-
ment is to investigate elusive physi-
cal phenomena by achieving high
accuracy, or observing very weak events,
or both. This normally requires a system
sufficiently simple that quantitative com-
parisons with theory are possible. On page
196 of this issue van Rooij et al. ( 1) pres-
ent results of a study that encompasses all
of these elements. They report on a precise
measurement of laser-driven transi-
tions between the singlet and triplet
metastable excited states of atomic
helium, 2
1
S and 2
3
S, a transition
so strongly forbidden by spin and
parity selection rules that even 10
years ago, its excitation would have
been regarded as impossible. The
transition is weaker by 14 orders
of magnitude than a typical fully
allowed electric dipole transition
(see the figure) ( 2) and is one of
the weakest optical transitions to be
measured with accuracy. This feat
is accomplished not by resorting
to high laser powers but by using
modest powers in conjunction with
long interaction times and very nar-
row linewidths.
The approach taken by van Rooij
et al. combines optical trapping of
laser-cooled helium atoms and fre-
quency metrology using phase-sta-
bilized femtosecond frequency combs, with
a little borrowing from telecommunications
technology, in the form of a fiber laser oper-
ating near 1.557 µm (µm). The laser is sta-
bilized to within 75 kHz by direct referenc-
ing to a “tooth” of the frequency comb, which
simultaneously provides an absolute calibra-
tion. A total power of 0.5 W is sufficient for
both optical trapping and spectroscopy.
The measurement is unusual in using a
sample of metastable 2
3
S helium atoms so
cold and dense that it is at or near quantum
degeneracy. The atoms are sufficiently slow-
moving that Doppler shifts and interaction-
time broadening are negligible. Surprisingly,
the authors find that shifts due to atomic inter-
actions are extremely small even in Bose-
condensed
4
He. Apart from the laser line-
width, some additional line broadening, as
well as part of the uncertainty, is contributed
by inhomogeneous frequency shifts
due to the optical trap. The inherent
natural linewidth of the transition is
just 8 Hz, determined by the sponta-
neous two-photon decay rate of the 2
1
S state to the ground 1
1
S state.
The results for the 2
1
S
0
→ 2
3
S
1

transition frequencies in both
3
He and
4
He are accurate to about 1.5 kHz, or
8 parts in 10
12
, improving on previous
indirect experimental determinations
by a factor of 100. From the perspec-
tive of atomic theory, the experiment
tests calculations of quantum electro-
dynamics (QED) contributions to the
ionization energies of the two meta-
stable states. The experimental accu-
racy exceeds that of current QED
calculations by a factor of 1000, a
major challenge for future theoreti-
cal work. The results also agree with
theory within its estimated accuracy,
indicating that the calculations are
well understood within their present
limits. This is an encouraging sign
that advances in theory and experi-
ment for the nearby 2
3
P state may
Department of Physics, University of Connect-
icut, Storrs, CT 06269–3046, USA. E-mail:
edward.eyler@uconn.edu
Decay rate (s
–1
) Temperature (K)
Conventional atomic beams
Supersonic beams
Quantum degeneracy
Helium in optical trap
Optical transition energy
3
He nuclear radius
via electron scattering
10
10
10
12
10
10
10
8
10
6
10
4
10
2
10
0
10
–2
10
–4
10
–6
10
8
10
6
10
4
10
2
10
0
10
–2
10
–4
10
–6
10
–8
2
1
S
0
1
1
S
0
(2 ϫ E1)
2
1
P
0
1
1
S
0
(E1)
2
1
P
0
2
1
S
0
(E1)
3
1
D
2
2
1
S
0
(E2)
2
3
P
2
1
1
S
0
(M2)
2
3
S
1
1
1
S
0
(M1)
2
1
S
0
2
3
S
1
(M1)
Probing the forbidden. (Left) Radiative decay rates for representative
transitions in helium ( 7– 9), comparing the doubly-forbidden 2
1
S
0
→ 2
3
S
1
transition with allowed electric dipole transitions (E1), “forbidden” elec-
tric quadrupole, magnetic dipole, and magnetic quadrupole transitions (E2,
M1, and M2), and spontaneous two-photon transitions (2 × E1). (Right) A
comparison of equivalent temperature scales. The large ratio of the transi-
tion energy to the thermal energy, about 10
10
, helps make possible the pre-
cision needed to probe physics on the nuclear scale, normally seen only in
electron scattering above about 100 MeV, or 10
12
K. The helium atom sym-
bol labels parameters directly relevant to the experiment.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 165
PERSPECTIVES
soon be sufficiently accurate to provide a sta-
tistically significant value of the fine-structure
constant ( 3), which would be the first time in
many decades that it could be based on actual
atomic fine structure.
Another remarkable attainment of pre-
cise atomic spectroscopy is the measure-
ment of nuclear properties by high-resolu-
tion measurements of electronic transitions.
The effects of finite nuclear size show up as
tiny perturbations in the energy levels, which
can be analyzed quantitatively if sufficiently
accurate atomic structure calculations can be
completed. Van Rooij et al. have compared
the 2
1
S → 2
3
S frequencies in
3
He and
4
He
to accurately determine a
3
He nuclear charge
radius of 1.961(4) fm. The result is more accu-
rate than direct measurements by scattering of
energetic electrons ( 4), although it disagrees
with a previous atomic physics determination.
Looking ahead, the authors have also set
the stage for a new generation of deep-ultra-
violet (UV) measurements that would probe
ultracold helium atoms in the ground 1
1
S
state, which lies below the metastable states
by a remarkably large 20 electron volts (eV).
If newly developed UV frequency combs
can be used directly for such measurements,
improvements in accuracy by many orders
of magnitude would be possible ( 5, 6). The
first major step toward this goal is to pro-
duce an optically trapped sample of ultra-
cold atoms in the singlet metastable state.
This could be achieved by a variation of the
arrangement used by van Rooij et al., which
already can transfer the entire trapped atom
sample to the 2
1
S state. Unfortunately, their
present 1.55-µm trap is repulsive for 2
1
S
atoms, so a different trapping wavelength
would be necessary. It can be expected that
cold singlet-state helium atoms will soon be
available for spectroscopy and cold colli-
sion experiments, and ultimately for the pro-
duction, trapping, and spectroscopy of the
ground state.
References
1. R. van Rooij et al., Science 333, 196 (2011).
2. G. Łach, K. Pachucki, Phys. Rev. A 64, 042510 (2001).
3. K. Pachucki, V. A. Yerokhin, Phys. Rev. A 79, 062516
(2009).
4. C. R. Ottermann et al., Nucl. Phys. A. 436, 688 (1985).
5. E. E. Eyler et al., Eur. Phys. J. D 48, 43 (2008).
6. D. Z. Kandula, C. Gohle, T. J. Pinkert, W. Ubachs, K. S. E.
Eikema, Phys. Rev. Lett. 105, 063001 (2010).
7. National Institute of Standards and Technology (NIST)
Atomic Spectra Database Version 4, NIST Standard
Reference Database 78; www.nist.gov/pml/data/asd.cfm
(2010).
8. G. Łach, K. Pachucki, Phys. Rev. A 64, 042510 (2001).
9. S. A. Alexander, R. L. Coldwell, Int. J. Quantum Chem.
108, 2813 (2008).
10.1126/science.1208276
A Critical Point for Turbulence
APPLIED PHYSICS
Bruno Eckhardt
The conditions that mark the transition
from laminar to turbulent flow in a pipe
have been determined by experiments and
numerical simulations.
W
hen a wind goes from a gentle
breeze to a gusty storm, it changes
from a state of smooth laminar
flow to one that is complex and turbulent.
One of the great triumphs of early 20th-cen-
tury science was determining the exact con-
ditions for the occurrence of the transition
between these dynamical states for many
types of flows ( 1). For most cases, a well-
defined critical flow speed could be deter-
mined where laminar flow becomes suscep-
tible to small perturbations and gives way to
turbulence. One of nature’s whims is that the
technologically important case of pressure-
driven flow through a cylindrical pipe does
not fit into this classification ( 2). On page 192
of this issue, Avila et al. ( 3) show how a criti-
cal point for turbulent pipe flow may finally
be identified.
Avila et al. studied the flow of water at
speeds of about 0.5 m/s in a pipe 4 mm in
diameter and 15 m in length. All of these
dimensional quantities can be combined into
a single relevant dimensionless parameter,
the Reynolds number Re = UD/ν, formed by
the mean velocity U, the diameter D of the
pipe, and the kinematic viscosity ν of the
fluid. A transition from laminar to turbulent
flow is often observed near Re of about 2000,
but the quoted values vary considerably, not
only between publications or experimental
facilities but also between runs at the same
facility ( 2). So should this be construed as a
transition without a critical point?
Most of the progress made in characteriz-
ing transitions was achieved by studying the
response of the laminar flow state to small
perturbations, which are expected to decay
below the critical Reynolds numbers and to be
amplified above it. Fluid flows are described
by the Navier-Stokes equations, which are
nonlinear partial differential equations that
can be solved exactly for a limited set of con-
ditions only. For small perturbations in the
vicinity of the laminar flow, linearized flow
equations that are more readily solved provide
an approximate description ( 1). However, cal-
culations for pipe flow show that there is no
change in behavior and that no critical point
can be found this way. The sufficiently small
perturbations required for a linear analysis
will always decay at arbitrarily high Reynolds
numbers (see the figure, panel A) ( 4).
Turbulence can only arise through large
perturbations for which the full nonlinear
equations of motion have to be studied.
Direct numerical simulations of the
Navier-Stokes equations provided an
initial suggestion for a critical Reyn-
Fachbereich Physik, Philipps-Universität Marburg, 35032
Marburg, Germany, and J. M. Burgerscentrum, TU Delft,
2628 CD Delft, Netherlands. E-mail: bruno.eckhardt@
physik.uni-marburg.de
A
B
Reynolds number
0 81.5 773 2040
Re
PT
Re
EXP
Re
ECS
Re
E
Laminar
Transient
~1600
Turning on turbulence. (A) The different flow states in pipe flow can be distinguished by their Reynolds num-
bers (Re). Below Re
E
= 81.5, all perturbations decay monotonically in energy (E). Spatially extended coherent
states (ECS) appear above Re
ECS
= 773. Experimentally, it is possible to induce transient localized turbulence
above Re
EXP
≈ 1600. Avila et al. show that a spatially intermittent but temporally persistent turbulence (PT)
forms above Re
PT
= 2040. At some higher but as yet undetermined Reynolds number, one may expect the flow
to become spatially homogeneous. (B) Avila et al. could observe splitting of localized “puffs” of turbulence in
numerical simulations when flow was suddenly increased so that Re jumped from 2200 to 2300. Fluid flow is
from left to right. A puff stretches in the downstream direction (bottom frame), a new puff is created (middle
frame), and once the puffs are sufficiently separated, they can stretch (top frame) and eventually break up again.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 166
PERSPECTIVES
olds number with the discovery of fully three-
dimensional, spatially extended and persis-
tent flow structures ( 5, 6)—coherent struc-
tures—that were subsequently also identified
in experiments ( 7). These structures appear
at specific flow speeds that can be computed
numerically with high precision and can pro-
vide a critical Reynolds number. However,
we do not have any a priori information con-
cerning where these critical points are and
what the associated flows look like. At pres-
ent, the lowest Re where some structures have
been found is 773 ( 8).
The presence of many coherent structures
of different shapes suggested that they pro-
vide a scaffold that could support turbulent
dynamics by creating a multitude of connec-
tions between these states ( 2). For low Reyn-
olds numbers, it was accepted that the tangle
of connections was not woven with sufficient
tightness to capture the turbulent dynamics
forever. It was expected that at higher Reyn-
olds numbers exceeding a critical value, the
turbulence would become persistent ( 9), but
more extensive experimental and numerical
studies contradicted the initial agreement:
The lifetimes increased rapidly, but there was
no finite number at which they would diverge
( 10). Accordingly, the critical Reynolds num-
ber would be infinity, and all turbulence
in pipe flow would be transient, albeit with
excessively long lifetimes.
Avila et al. resolved this puzzling behav-
ior and identified the missing feature that had
not received sufficient attention: Turbulence
in pipe flow has the unusual property that
for Reynolds numbers below about 2300, it
remains localized in short “puffs” that move
downstream without much change in form.
Because of their finite lifetime, the puffs
should disappear one by one, and only the
laminar profile would remain at long times.
However, Nishi et al. ( 11) showed that puffs
can split. In one process, fluctuations in the
middle of the puff may become strong enough
to introduce a laminar region that then pushes
the two elements apart (see the figure, panel
B, for an example from a numerical simula-
tion). In another case, patches of turbulence
swept downstream in the center of the fluid
may attach to the walls and start new turbu-
lent puffs. Such processes introduce connec-
tions between the puffs so that they can no
longer be considered in isolation. In particu-
lar, if a puff manages to split before it decays,
the sibling may carry on the turbulence, spa-
tial and temporal couplings become impor-
tant ( 12), and there may always be some tur-
bulence somewhere along the pipe.
Avila et al. compared the lifetime of puffs
with the time it takes for them to split. They
S
ex is hard to explain. Since males can’t
reproduce by themselves and often
contribute nothing except genes to off-
spring, a population of asexual females can
grow at double the rate of a population that
reproduces sexually ( 1). Why then, given this
“cost of males,” do most plants and animals
indulge in biparental sex? One possible solu-
tion is that sex accelerates adaptation; the Red
Queen hypothesis, for example, proposes that
sex gives plants and animals an edge in the
never-ending battle against their coevolv-
ing parasites ( 2– 4). Although researchers
have collected empirical field data consis-
tent with the Red Queen hypothesis from a
range of natural host-parasite systems, direct
experimental evidence that coevolving para-
sites select for sex in their hosts has proven
elusive. On page 216 of this issue, Morran et
al. ( 5) pin down some of that direct evidence.
In laboratory experiments, they grew several
populations of nematode worms, some with
and some without a bacterial parasite, to pro-
vide the most definitive support yet for the
Red Queen’s answer to why sex evolved.
As first conceived in 1973 by evolution-
ary biologist Leigh Van Valen, the Red Queen
hypothesis had little to do with sex. Van Valen
used the Red Queen’s race, from Lewis Car-
roll’s Through the Looking-Glass, as an anal-
ogy for nature ( 6). In Carroll’s story, Alice
and the Red Queen run as fast as they can but
never get anywhere ( 7). In Van Valen’s view
of nature, species continually evolve but their
fitness never increases because each adapta-
tion is countered by adaptations by their com-
petitors and enemies ( 6). He suggested that
this coevolutionary mechanism could explain
why rates of extinction within animal groups
remain near constant through geological
time. Biologists later co-opted the Red Queen
analogy into a new coevolutionary hypoth-
esis for the evolution of sex ( 4). Mathemati-
Sex, Death, and the Red Queen
EVOLUTION
Michael A. Brockhurst
Experiments involving host-parasite interactions demonstrate the evolutionary benefits of sexual
reproduction.
Institute of Integrative Biology, University of Liverpool, Liver-
pool L69 7ZB, UK. E-mail: michael.brockhurst@liv.ac.uk
overcame the difficulty of inducing turbulence
at these low Reynolds numbers by creating a
stepwise perturbation—they injected a water
jet into the flow to create puffs of turbulence.
With increasing Reynolds number, the life-
times of puffs increased rapidly and the time
to split decreased. In the critical region where
these two times were similar, only one split-
ting or decay event occurred for every 10,000
injections of the jet. Such rare events are inac-
cessible in numerical simulations. Avila et al.
provide convincing evidence for a crossing of
the two curves at Re = 2040. On the basis of
previous studies ( 12, 13), a higher value might
be expected, but the difference presumably
comes from a poorer statistical method that
missed the important rare events.
The findings of Avila et al., and even more
so their method of analysis, bring into focus
the spatiotemporal aspects of the transition
problem ( 14). They pave the way for a better
understanding of the transition in pipe flows
and related shear flows, such as plane Couette
flows and perhaps even boundary-layer flows,
and connect the transition to the spatial inter-
mittency and phase transitions in directed
percolation ( 15). They provide not only the
long-sought critical Reynolds number for
pipe flow, but also define a critical change in
our approach to studying turbulence transi-
tions in spatially extended systems.
References
1. P. Drazin, W. Reid, Hydrodynamic Stability (Cambridge
Univ. Press, Cambridge, 2004).
2. B. Eckhardt, Philos. Trans. R. Soc. London Ser. A 367,
449 (2009).
3. K. Avila et al., Science 333, 192 (2011).
4. A. Meseguer, L. Trefethen, J. Comput. Phys. 186, 178
(2003).
5. H. Faisst, B. Eckhardt, Phys. Rev. Lett. 91, 224502 (2003).
6. H. Wedin, R. R. Kerswell, J. Fluid Mech. 508, 333 (2004).
7. B. Hof et al., Science 305, 1594 (2004).
8. C. C. T. Pringle, R. R. Kerswell, Phys. Rev. Lett. 99,
074502 (2007).
9. H. Faisst, B. Eckhardt, J. Fluid Mech. 504, 343 (2004).
10. B. Hof, J. Westerweel, T. M. Schneider, B. Eckhardt,
Nature 443, 59 (2006).
11. M. Nishi, B. Ünsal, F. Durst, G. Biswas, J. Fluid Mech.
614, 425 (2008).
12. D. Moxey, D. Barkley, Proc. Natl. Acad. Sci. U.S.A. 107,
8091 (2010).
13. J. Rotta, Ing. Archiv 24, 258 (1956).
14. P. Manneville, Phys. Rev. E. 79, 025301 (2009).
15. H. Hinrichsen, Adv. Phys. 49, 815 (2000).
10.1126/science.1208261
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 167
PERSPECTIVES
cal models showed that coevolving parasites
could, over time, select against common gene
variants (alleles) in the host, thereby favor-
ing rarer host alleles. These once-rare alleles
then increase in frequency and become com-
mon, thus establishing sustained oscillating
changes in host and parasite allele frequencies
( 3). This continual selection for rarity favors
sexual reproduction over asexual reproduc-
tion; sexual recombination allows hosts to
reshuffle their pack of alleles and generate
new, rare combinations in their offspring.
Empirical field data, most notably from
studies of freshwater snails that can repro-
duce sexually or asexually (facultative
reproduction) and their trematode parasites
(flukes), broadly support the Red Queen
hypothesis. Trematodes are best adapted to
infect locally common snail genotypes ( 8),
and the frequency of male snails (a proxy for
the frequency of sexual reproduction) is high-
est in the shallows where the risk of infection
is greatest ( 9). This suggests that infection
promotes sex. However, as in any field study,
it is difficult to definitively ascribe causation,
because researchers can never rule out selec-
tion by other environmental variables that
also correlate with the frequency of males.
Another issue with field data is that coevolu-
tion itself must necessarily be inferred, since
hosts from the past and future are not avail-
able to directly test whether today’s parasites
actually are best adapted to contemporary
hosts. Testing the causality of the Red Queen
hypothesis requires controlled, real-time evo-
lution experiments and the ability to keep a
“living fossil record” of past populations in
suspended animation.
Experimental evolution has tradition-
ally involved microbes ( 10). However, larger
short-lived organisms, such as fruit flies and
nematodes, are amenable to experimental
evolution. Nematodes, like microbes, can also
be frozen in suspended animation, and revived
at a later date, allowing direct comparison of
descendants with their evolutionary ancestors
(see the figure). In their experiments, Mor-
ran et al. used the nematode, Caenorhabdi-
tis elegans, and its natural bacterial parasite,
Serratia marcescens. C. elegans is faculta-
tively sexual; males typically constitute 20 to
30% of a wild-type population. In experimen-
tal populations raised without parasites, the
authors report that the proportion of the popu-
lation reproducing sexually remained at 20%.
However, in the presence of coevolving para-
sites, the frequency of sex rapidly increased
and stabilized at 80 to 90%. These results sug-
gested that the coevolving parasites selected
for sex. This conclusion was reinforced by
results from a third set of experimental nem-
atode populations, in which the researchers
exposed the worms to a fixed, nonevolving
strain of S. marcescens while allowing C. ele-
gans to adapt. Here, after an initial increase
in the frequency of males, sexual reproduction
subsequently declined to 20%. Morran et al.
concluded that coevolution with parasites, not
parasites per se, provides sustained selection
for the long-term maintenance of sex.
Morran et al. were also able to measure
the benefits of sex by enforcing or preventing
sex in certain nematode populations, using
mutants that were either obligate-sexuals or
obligate self-fertilizers. When coevolving
with parasites, all selfing C. elegans popula-
tions became extinct within 20 generations;
in contrast, sexual C. elegans populations
never became extinct. Similarly, the advan-
tages of sex were revealed in experiments that
involved reviving earlier, ancestral nematodes
and infecting them with newer, coevolved
parasites. The parasites had become more
deadly over time, but coevolved sexual C.
elegans populations showed resistance; in
contrast, coevolved selfing C. elegans did
not. These observations support Van Valen’s
original macroevolutionary version of the
Red Queen hypothesis, and demonstrate that
species that lag behind in the coevolutionary
race are prone to extinction.
The Red Queen hypothesis places host-
parasite coevolution, with its demand for
rapid and continual adaptation, at the heart
of evolution. Van Valen recognized, however,
that such pairwise associations are only a
subset of the rich and varied coevolutionary
interactions inherent to natural communities.
The challenge for theorists and empiricists
alike is to understand how pairwise coevolu-
tionary processes scale up when embedded in
a broader and more complex network of spe-
cies interactions. As more runners join the
race, do the benefits of sex multiply?
References
1. J. Maynard Smith, The Evolution of Sex (Cambridge Univ.
Press, Cambridge, 1978).
2. J. Jaenike, Evol. Theory 3, 191 (1978).
3. W. D. Hamilton, Oikos 35, 282 (1980).
4. G. Bell, The Masterpiece of Nature (Croom Helm, London,
1982).
5. L. T. Morran et al., Science 333, 216 (2011).
6. L. Van Valen, Evol. Theory 1, 1 (1973).
7. L. Carroll, Through the Looking-Glass and What Alice
Found There (Macmillan, London, 1871).
8. C. M. Lively, M. F. Dybdahl, Nature 405, 679 (2000).
9. K. C. King et al., Curr. Biol. 19, 1438 (2009).
10. A. Buckling et al., Nature 457, 824 (2009).
11. S. Paterson et al., Nature 464, 275 (2010).
H
0
P
0
H
1
P
1
H
2
P
2
H
3
P
3
Coevolution
H
0
P
0
Frozen
stocks
etc.…
etc.…
R
a
t
e

o
f

s
e
x
u
a
l

r
e
p
r
o
d
u
c
t
i
o
n
Time
Control Evolution Coevolution
H
1
P
0
H
2
P
0
H
3
P
0
Evolution
Hobbling the Red Queen. Researchers can study
the impact of parasite-host interactions on the evo-
lution of sexual reproduction by conducting experi-
ments that create different host-parasite populations,
and allowing them to evolve over many generations.
In this example, if researchers allow a nematode
worm host (H) and a parasite (p) to coevolve (top
series of boxes), then high rates of sexual reproduc-
tion are sustained (graph, right). If they use frozen
parasite stocks to reinfect each new generation of the
host with a fixed, nonevolving ancestral strain of the
parasite (p
0
bottom series of boxes), rates of sexual
reproduction can decline. Such experiments can also
replace the host, rather than the parasite ( 11).
C
R
E
D
I
T
:

C
.

E
L
E
G
A
N
S

I
M
A
G
E

B
Y

B
O
B

G
O
L
D
S
T
E
I
N
,

U
N
I
V
E
R
S
I
T
Y

O
F

N
O
R
T
H

C
A
R
O
L
I
N
A
,

C
H
A
P
E
L

H
I
L
L
,

C
R
E
A
T
I
V
E

C
O
M
M
O
N
S

L
I
C
E
N
S
E
10.1126/science.1209420
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

SPECIALSECTION
C
R
E
D
I
T
:

N
A
S
A
/
J
P
L
-
C
A
L
T
E
C
H
CONTENTS
News
170 Milky Way Researchers’ Home Away
From Home
173 Galaxy Zoo Volunteers Share Pain
and Glory of Research
Perspective
176 Galactic Paleontology
E. Tolstoy
Reviews
178 The Cosmic History of
Star Formation
J. S. Dunlop
182 The Coevolution of Galaxies and
Supermassive Black Holes:
A Local Perspective
T. M. Heckman and G. Kauffmann
See also Policy Forum p. 161.
Galaxy
Evolution
I NT RODUCT I ON
A Universe of Galaxies
IT WASN’T UNTIL THE 1920S THAT ASTRONOMERS REALIZED THAT THERE WERE OTHER
galaxies in the universe besides our own. Using the 100-inch telescope at Mount
Wilson in California, Edwin Hubble determined the distance to Andromeda
(M31) and to the Triangulum (M33) and concluded that each was an “isolated
system of stars and nebulae, lying far outside the limits of the galactic system.”
Before that, these and other galaxies were classified as nebulae, extended objects
other than planets or comets; although their location was a matter of great debate,
they were generally thought to be within our galaxy.
Nowadays there is no doubt that the universe extends well beyond the con-
fines of the Milky Way and that our galaxy is just one among many. Telescopes
much more powerful than those used by Hubble have produced ever-larger and
more comprehensive surveys of galaxies. The detailed understanding of our gal-
axy has also evolved dramatically. As explained by Tolstoy (p. 176), the study
of individual stars in the Milky Way can help us understand the history of our
galaxy and the physical processes that happened when the first stars and galaxies
formed. More generally, the history of star formation in galaxies tells us how the
structure and average chemical composition of the universe have changed over
its 14-billion-year history. On page 178, Dunlop discusses the different methods
used to quantify this history back to about 500 million years after the big bang.
The biggest surprise in recent years may be the discovery that the life cycles
of galaxies and those of the black holes that reside in their centers are intimately
linked. To understand how galaxies formed and evolved, it is necessary to
understand the role black holes play in the evolution of galaxies. Heckman and
Kauffmann (p. 182) summarize the progress made in the past decade by studying
the interplay between galaxies and their central black holes.
Researchers studying galactic evolution work on a wide range of scales.
Some scrutinize the structures of individual galaxies; others analyze statistical
patterns from huge data sets. In a pair of News stories, Bhattacharjee (p. 170)
describes insights being gleaned from Andromeda, a nearby spiral galaxy simi-
lar in many ways to the Milky Way. Clery (p. 173) tells the story of the Galaxy
Zoo, an online project that enlisted more than 100,000 volunteers to classify
galaxies recorded by automated sky surveys.
Much of the progress described in this special section rests on the availability
of state-of-the-art space telescopes, such as the Hubble Space Telescope. In a
Policy Forum, Bonnet and Bleeker (p. 161) propose that the far-future, large-
scale space facilities required to make groundbreaking discoveries in astronomy
should be globally shared, and they suggest ways to improve international plan-
ning and cooperation. Although we need to start planning for those facilities
now, we do not need to wait for them to make further progress. Some of the
telescopes nearing completion, or in planning, both on Earth and in space, can
potentially address some of the most pressing questions. Hopefully, they will
surprise us again.
– MARIA CRUZ AND ROBERT COONTZ
169 www.sciencemag.org SCIENCE VOL 333 8 JULY 2011
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 170
GALAXY EVOLUTI ON
Milky Way Researchers’
Home Away From Home
Right next door and easy to study, Andromeda provides an
excellent model of how our own galaxy probably evolved
If you were stuck forever inside your house, you could still learn a lot
about houses by studying the one across the street. For astronomers
trapped inside the Milky Way, the Andromeda galaxy is that neighboring
house. And unlike the Milky Way, it can be imaged and studied in its
entirety. At a distance of 2.5 million light-years, it is our closest galactic
neighbor comparable to the Milky Way: far enough to offer a global view
and close enough for telescopes to take a good look at individual stars
inside it.
As such, Andromeda has come to be regarded as one of the best models
available for understanding the evolution of galaxies, including our own.
That’s why it has acquired a dedicated fan club of astronomers who have
been studying the galaxy in ever-increasing detail over the past 15 years.
These studies have helped astronomers define Andromeda’s structure with
unprecedented clarity, showing that the galaxy’s disk and halo extend far
beyond the boundaries that had been assumed before. The details of the
picture are providing clues about how Andromeda came to be the galaxy
it is today.
GGA GA GALA LAXY XY EEVO VOLU LUTI TI ON ON
Right next door and easy to study, Andromeda provides an
excellent model of how our own galaxy probably evolved
If you were stuck forever inside your house, you could still learn a lot
about houses by studying the one across the street. For astronomers
trapped inside the Milky Way, the Andromeda galaxy is that neighboring
house. And unlike the Milky Way, it can be imaged and studied in its
entirety. At a distance of 2.5 million light-years, it is our closest galactic
neighbor comparable to the Milky Way: far enough to offer a global view
and close enough for telescopes to take a good look at individual stars
inside it.
As such, Andromeda has come to be regarded as one of the best models
available for understanding the evolution of galaxies, including our own.
That’s why it has acquired a dedicated fan club of astronomers who have
been studying the galaxy in ever-increasing detail over the past 15 years.
These studies have helped astronomers define Andromeda’s structure with
unprecedented clarity, showing that the galaxy’s disk and halo extend far
beyond the boundaries that had been assumed before. The details of the
picture are providing clues about how Andromeda came to be the galaxy
it is today.
C
R
E
D
I
T
:

R
O
B
E
R
T

G
E
N
D
L
E
R
/

W
W
W
.
R
O
B
G
E
N
D
L
E
R
A
S
T
R
O
P
I
C
S
.
C
O
M
GALAXY EVOLUTION
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 171
SPECIALSECTION
C
R
E
D
I
T

(
T
O
P

T
O

B
O
T
T
O
M
)
:

J
O
H
N

D
U
B
I
N
S
K
I
/
U
N
I
V
E
R
S
I
T
Y

O
F

T
O
R
O
N
T
O

(
4
)
;

C
O
U
R
T
E
S
Y

P
U
R
A
G
R
A

G
U
H
A
T
H
A
K
U
R
T
A
Over the years, astro-
nomers have discovered a
dozen stellar streams falling
into the outer regions of
the galaxy that have been
shown to be the remnants of
smaller, satellite galaxies. The
presence of these so-called
tidal streams indicates that
Andromeda has swallowed
up numerous smaller galaxies
in the past, incorporating
their stars, gas, and other
matter into its structure over
billions of years. By studying
the velocities and chemical
com positions of stars within
these streams and elsewhere
in the galaxy, astronomers are
beginning to piece together a
detailed evolutionary history
of Andromeda.
The results confirm the
long-held idea that large
galaxies like our own have
formed through the hier-
archical assembly of smaller
galaxies. The picture is what
astronomers expected to see based on current
theories of dark matter, the invisible stuff
that makes up 80% of the universe. “The
fact that Andromeda seems to be surrounded
by all of these streams is a smoking gun
that hierarchical structure formation is
happening,” says James Bullock, a theoretical
cosmologist at the University of California
(UC), Irvine. “So indirectly, these numerous
streams provide compelling evidence that we
are not so far off base in our understanding
of dark matter.”
Mistaken identity
Andromeda can be seen in the night sky
with the naked eye. Like the Milky Way and
other spiral galaxies, it consists of a flat,
rotating disk of stars with a central bulge,
nested inside a more-tenuous spherical
halo of stars. Understanding the precise
characteristics of its structure has gone
hand in hand with efforts to understand
its evolution.
Until the early 2000s, astronomers
thought Andromeda’s halo was metal-rich:
that is, made up of stars containing heavier
elements in addition to hydrogen and
helium. Because heavy elements formed
relatively late in the universe’s history, their
presence is a marker of youth. Researchers
studying the halo in 2002 estimated that a
third of the stars were between 6 billion and
8 billion years old—nearly 4 billion years
younger than most of their counterparts in
the halo of the Milky Way. One possible
explanation for the difference was that
Andromeda began to form much later than
the Milky Way did.
But that age estimate turned out to be a case
of mistaken identity: The stars that researchers
had identified as belonging to Andromeda’s
halo were in fact from the outer reaches of the
galaxy’s bulge, where stars had been recently
forming. Using the Keck telescopes atop
Mauna Kea in Hawaii, a collaboration led
by Puragra Guhathakurta of the University
of California, Santa Cruz (UCSC), spotted
faint stars far out from the
bulge—up to five times
farther out than where stars
had been sighted before.
In 2005, Guhathakurta’s
collaboration—which goes
by the name SPLASH (Spec-
tro scopic and Photomet ric
Landscape of Andromeda’s
Stellar Halo)—reported that
these newly discovered stars
were part of Andromeda’s
true halo and were as ancient
as stars in the halo of the
Milky Way.
The finding meant that
Andromeda and the Milky
Way were likely to have
evolved through similar pro-
cesses: interactions with
other, ancient galaxies. Be-
cause the newly dis covered
stars could not possibly have
been born so far from the
central star-forming regions
in Andromeda’s disk, astro-
nomers concluded that they
probably came from other
gal axies that were sucked into
Andromeda.
Death plunge
Astronomers had already
glimpsed one clear incident of
cannibalism by Andromeda.
In 2001, peering through the
Isaac Newton Telescope on
the Canary Islands, Rodrigo
Ibata of the Strasbourg Ob-
ser vatory in France and
his colleagues discovered a
dense line of stars poking
through Andromeda’s south-
ern halo and pointed toward its disk. They
realized that it was a giant stream of stars
moving in unison. “It was plunging radially
toward the center of the galaxy,” says
Scott Chapman, an astronomer at the
University of Cambridge in the U.K. and
a co-author of a Nature paper reporting
the discovery.
In 1994, Ibata had discovered a similar
stream flowing into the Milky Way, which
researchers determined to be a flow of stars
from the Sagittarius dwarf elliptical galaxy,
a satellite galaxy being shredded by the
Milky Way. In Andromeda’s case, the Giant
Southern Stream—as the structure came to
be called—was speculatively linked to two
of Andromeda’s satellites,
M32 and M110, although
there was no firm evidence
for the connection.
The stream quickly
became the subject of
de tailed studies aimed at
nailing its progenitor and
figuring out exactly how the
cataclysmic shredding had
proceeded over time. Mark
Fardal, now a post doc at the
University of Massachusetts,
Amherst, was at the forefront
of these efforts. Using data
Steady loss. A simulation showing how the Triangulum galaxy is beginning to
lose stars and gas to Andromeda as it orbits the larger galaxy.
Star man. Guhathakurta helped
locate Andromeda’s true halo.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 172
GALAXY EVOLUTION
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

C
O
U
R
T
E
S
Y

O
F

A
L
A
N

M
C
C
O
N
N
A
C
H
I
E
;

N
I
C
O
L
A
S

M
A
R
T
I
N
/
M
A
X

P
L
A
N
C
K

I
N
S
T
I
T
U
T
E

F
O
R

A
S
T
R
O
N
O
M
Y
on the velocities and chemical abundances
of stars within the stream, which revealed
bifurcations and narrower currents inside
it, Fardal came up with simulations of
the event. He concluded that the stream
had not come from M32 or M110 after
all. It was likely some other local satellite
that had been completely cannibalized
by Andromeda, leaving no trace outside
Andromeda’s halo.
Fardal suspects that the vanished
satellite may have begun its plunge into
Andromeda 700 million years ago. By that
time, Andromeda’s gravitational pull had
gradually stripped the satellite of dark matter
and outer stars over hundreds of millions of
years. “It’s the same phenomenon as tides
on Earth, only much more extreme,” Fardal
says. “You get a stretching of the satellite
that results in stars being unbound from it.”
In Fardal’s estimate, the satellite behind the
Giant Southern Stream may have been as
massive as 3 billion suns, less than 1% of
the mass of the Milky Way. Its shredding not
only brought new material into Andromeda
but also likely sent shock waves rippling
through the galaxy, which might have
induced the formation of new stars in other
regions and shaped the overall structure of
the galaxy in unknown ways.
After the discovery of the Giant Southern
Stream, astronomers stepped up efforts to
look for other streams within Andromeda.
Looking through the 8-meter Subaru
telescope in Hawaii, SPLASH collaborators
Mikito Tanaka and Masashi Chiba of Tohoku
University in Japan dis covered two dense lines
of stars in the northwest region of the galaxy.
An analysis of the stars’ spectra
yielded their velocities, from
which researchers at UCSC
confirmed that the lines were
indeed streams. The stars were
moving coherently “as a group
that ‘remembers’ the orbital
speed of the progenitor dwarf
galaxy to which the stars once
belonged,” Guhathakurta says.
Mergers everywhere
More recently, several more
tidal streams have been dis-
covered by the Pan-Andro-
meda Archaeological Survey
(PAndAS), a collaboration
led by Alan McConnachie
at the Herzberg Institute
of Astrophysics in Victoria,
Can ada. Observing with the
Canada- Fr ance- Hawai i
Telescope, the researchers
sur veyed the entire outskirts
of Andromeda, identifying
17 dwarf satellites that had
not been known before.
“Some of the satellites
are in a stage of significant
disruption; others don’t ap-
pear to be getting disrupted
at all,” McConnachie says.
“We also see a bunch of
streams that don’t seem to
have a satellite connected
to them, which means
the satellite has exerted
completely con sumed.”
The survey also showed
just how powerful an influ-
ence Andromeda has exerted
on its neighborhood. Mc-
Connachie says he and his
colleagues were surprised to
discover that Andromeda was
starting to nibble stars away
from the Triangulum galaxy,
a neighbor that is half the size
of the Milky Way and several times larger
than any dwarf. The researchers expected
Triangulum’s significant size and heft to
protect it from Andromeda’s gravitational tug.
But on the edge of Triangulum, they
saw “a vast, extended, distorted distribution
of stars, which is exactly what you would
expect of stars pulled off of Triangulum,”
McConnachie says. His group has done
simulations showing that “about 3 billion years
ago, the Triangulum galaxy passed relatively
close to Andromeda, and
Andromeda’s gravity caused
several of Triangulum’s stars
to unravel,” McConnachie
says. At some point in the
future, Triangulum could
end up as a line streaking
through Andromeda’s halo
into its center.
The discovery of the
various tidal streams has
convinced astronomers that
the chewing up of small
galaxies by bigger ones—
and more generally galaxy
mergers—are a frequent oc-
currence in the universe.
“We’ve gone from not
really knowing that galaxy
cannibalism occurs to hav-
ing a few examples of the
process to knowing that this
is an extremely common pro-
cess,” McConnachie says.
Guhathakurta says the
number of streams sug gests
Andromeda may have had
“a more troubled past than the Milky Way.”
Researchers are now digging deeper into the
streams to get a better handle on questions
such as the durations of encounters with
different satellites, the orbits of the progenitor
galaxies, and the structural and chemical
characteristics of the satellites before they
were swallowed.
The study of Andromeda has provided
theorists with a rich trove of data to test
ideas of galaxy evolution, says Bullock.
“We can actually predict
fairly accurately how many
little dwarf galaxies should
have been accreted and
destroyed,” he says. “The first
of these predictions were in
place before PAndAS and
SPLASH made their dis cover-
ies, and things look pretty
good in comparison.” He says
continued investigations of
Andromeda will be critical for
gaining key insights about the
universe at large. “For example,
how small do the dark matter
clumps have to be before they
stop making galaxies?” he
says. “Andromeda provides a
perfect place to explore these
questions.”
–YUDHIJIT BHATTACHARJEE
Assimilated. Reddish streak of stars high in heavy elements shows remains
of a dwarf galaxy on Andromeda’s outskirts.
“We’ve gone from
not really knowing
that galaxy cannibal-
ism occurs to having
a few examples of
the process to
knowing that this
is an extremely
common process.”
—ALAN MCCONNACHIE,
HERZBERG INSTITUTE
OF ASTROPHYSICS
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 173
SPECIALSECTION
C
R
E
D
I
T
:

N
A
S
A
,

E
S
A
,

W
.

K
E
E
L

(
U
N
I
V
E
R
S
I
T
Y

O
F

A
L
A
B
A
M
A
)
,

A
N
D

T
H
E

G
A
L
A
X
Y

Z
O
O

T
E
A
M
The automated surveys that are becoming
increasingly common in astronomy are
producing an embarrassment of riches for
researchers. Projects such as the Sloan
Digital Sky Survey (SDSS) are generating so
much data that, in some cases, astronomers
don’t know what to do with them all. SDSS
has compiled a list of more than 1 million
galaxies. To glean information about galaxy
evolution, however, astronomers need to
know what type of galaxy each one is: spiral,
barred spiral, elliptical, or something else.
At present, the only reliable way to classify
galaxies is to look at each one. But the SDSS
list is so long that all the world’s astronomers
working together couldn’t muster enough
eyeballs for the task.
Enter the “wisdom of crowds.” An
online effort called Galaxy Zoo, launched
in 2007, set a standard for citizen-scientist
participation projects. Zealous volunteers
astonished the project’s organizers by
classifying the entire catalog years ahead
of schedule. The results have brought real
statistical rigor to a field used to samples too
small to support firm conclusions.
But that’s not all. Buoyed by the curiosity
and dedication of the volunteers, the Galaxy
Zoo team went on to ask more-complicated
classification questions that led to studies
they hadn’t thought possible. And in an
online discussion forum on the Galaxy
Zoo Web site, volunteers have pointed to
anomalies that on closer inspection have
turned out to be genuinely new astronomical
objects. “I’m incredibly impressed by what
they’ve managed to achieve,” says University
of Oxford astronomer Roger Davies,
president of the Royal Astronomical Society.
“They’ve made it possible to do things with a
huge survey.”
Uncaging astronomy
Galaxy Zoo was born out of necessity—and
tedium. Several groups have attempted to
develop image-analysis programs to classify
galaxies automatically. But galaxies are hard
for a computer program to get a handle on.
In general, they are starry hazes, some with
spiral arms, some without; some tightly
wound and others looser; sporting central
bars of various lengths, or no bar; some
viewed face-on and others edge-on; not to
mention the universe’s many misshapen
oddball galaxies and messy galactic mer-
gers in progress. As a result, software class-
ification isn’t yet a reliable method.
A quick and easy way to sort galaxies into
spirals (flat disks with arms) and ellipticals
(featureless blobs with bright cores) is to
look at their spectra. Spirals tend to have
active star-forming regions that emit a lot of
blue light, whereas ellipticals are generally
composed of older, cooler stars emitting in
the red. Astronomers have spotted exceptions
to the rule—red spirals and blue ellipticals—
but they haven’t been sure how large these
subsets were. In 2006, Kevin Schawinski,
then a postgrad student at Oxford, set out
to answer that question by eyeballing some
50,000 SDSS galaxies to pick out all the
ellipticals and find out what proportion of
them were blue. Schawinski succeeded, but
the work was almost unbearably tedious. His
main conclusion, he says, was that “there had
to be a better way to do this.”
Chatting in a pub, Schawinski and fellow
astronomer Chris Lintott came up with the
concept of Galaxy Zoo: an online appeal
to thousands of citizen-scientists for help
with visual galaxy classifications. Projects
such as SETI@home had already shown
that members of the public were willing
to hand over their spare computer time to
scientists, and some had enlisted real citizen
participation. Stardust@Home, for example,
asked volunteers to spot tracks of interstellar
dust in samples brought back by NASA’s
Stardust comet-chasing mission. Stardust@
Home recruited 20,000 people. Schawinski
and Lintott figured that if they could get a
quarter of that number to do one classification
per day, they could work through the whole
SDSS galaxy list in 3 years with each galaxy
inspected by five different people.
The Galaxy Zoo Web site was launched
on 11 July 2007. A simple interface gave
volunteers a brief tutorial and then set them
off, asking simply whether each galaxy
image was spiral or elliptical and whether it
spun clockwise or anticlockwise. Participants
could classify as many or as few as they liked
and could break off and come back whenever
they wanted. A news story on a BBC Web
site set the ball rolling; after just 3 hours,
Schawinski recalls, traffic was so heavy that
Galaxy Zoo’s site, hosted by Johns Hopkins
University, crashed. The university quickly
Galaxy Zoo Volunteers Share
Pain and Glory of Research
A project to “crowdsource” galactic classifications has paid off in ways the
astronomers who started it never expected
GALAXY EVOLUTI ON
Space oddity. Greenish “voorwerp” spotted
by a Dutch volunteer still intrigues scientists.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 174
GALAXY EVOLUTION
C
R
E
D
I
T
S

(
T
O
P

T
O

B
O
T
T
O
M
)
:

S
L
O
A
N

D
I
G
I
T
A
L

S
K
Y

S
U
R
V
E
Y
;

C
O
U
R
T
E
S
Y

O
F

L
A
S

C
U
M
B
R
E
S

O
B
S
E
R
V
A
T
O
R
Y

G
L
O
B
A
L

T
E
L
E
S
C
O
P
E

N
E
T
W
O
R
K
/
G
A
L
A
X
I
E
S

O
B
S
E
R
V
E
D

W
I
T
H

F
A
U
L
K
E
S

T
E
L
E
S
C
O
P
E

N
O
R
T
H
,

H
A
W
A
I
I
cloned the site onto other servers and got it
working again in a few hours.
Twelve hours after launch, Galaxy Zoo
was receiving 20,000 classifications per
hour. Two days later, the figure was 60,000
per hour, and after 10 days the public had
submitted 8 million classifications. “There
were so many classifying for a while, it
was a bit scary,” says Karen Masters of
the University of Portsmouth in the United
Kingdom. By the time the Galaxy Zoo team
submitted its first paper 9 months later,
more than 100,000 volunteers had classified
the whole SDSS catalog, with each galaxy
viewed 38 times on average.
This ability to do multiple classifications
of the same object is one of the key strengths
of Galaxy Zoo data. Fuzzy images of distant
galaxies are difficult to classify, and even
galaxy experts often disagree. But with
Galaxy Zoo, if many volunteers click on the
same type for a galaxy, you can be pretty
sure of the result; if there is a spread of
different classifications, you know you have
to treat the result with care. “With a large
number of evaluations, you can get a level of
uncertainty,” Davies says. “This is a unique
thing that Galaxy Zoo is able to do.”
Seeing red
One early result of the first phase of Galaxy
Zoo was testing the traditional link between
galactic color and shape: blue spirals and red
ellipticals. Galaxy Zoo found that 80% of
galaxies follow this pattern,
but sizable minorities show
different correlations. “It’s
reinvigorated the idea that
morphology is important,”
Masters says. “There was a
trend to look only at color,
but color is not the only
way to look at a galaxy.”
Whereas color tells you
about a galaxy’s stars, its
shape gives you dynamical
information, she says.
Astronomers currently
think that galaxies form
when dark matter clusters
together under its own
gravity and then pulls in
gas, dust, stars, and smaller
galaxies to form the flat,
extended disk of a spiral
galaxy. The dark matter
stays in a spherical “halo”
that surrounds the whole
galaxy. In spiral galaxies,
stars keep forming in the disk as neutral
hydrogen gas rains down from the halo into
the spiral arms. Elliptical galaxies, in contrast,
result when two galaxies of comparable
masses collide. If one or both are spirals, their
spiral structure is destroyed, leaving stars in
random orbits around the galactic center.
The collision also slams together gas clouds,
causing a burst of intense star formation that
flares and burns out quickly, using up all the
hydrogen gas. The end result is a structureless
blob of a galaxy with no gas clouds and few
young stars.
Masters set out to study the red spiral
galaxies identified in the first phase of Galaxy
Zoo. First she weeded out all the galaxies
that weren’t viewed face-
on. (Spiral galaxies viewed
edge-on look redder than
they really are, she explains,
because the dust in their
galactic plane tends to redden
starlight passing through it.)
“There were so many galaxies
that you can be selective,” she
says. Masters noticed that
barred spirals, galaxies with
a linear structure through the
center that links into the spiral
arms, were twice as common
in red spirals as in blue ones.
Data from the second phase of
Galaxy Zoo (GZ2) confirmed
the pattern. In general, bluish
light shows that a galaxy is still producing
many young stars; reddish light indicates
that it is not. Masters wondered whether
the bars might be halting star formation,
perhaps by channeling hydrogen gas from the
arms into the core. “Do bars kill galaxies?”
she wondered.
If gas were pouring into a galaxy’s center,
Masters reasoned, the supermassive black hole
there would heat it and spew out material as a
huge, hot beacon of energy: an active galactic
nucleus (AGN). When she checked galaxies’
spectra for signs of such beacons, however,
Masters failed to find a strong correlation
between AGNs and red barred spirals. Now
she suspects that something external, such
as small neighboring galaxies, is stripping
neutral hydrogen from the galactic halo of the
red spirals and so depriving their arms of fuel
for star formation. “A bar is a side effect of
the process,” she says.
As for the blue ellipticals, “we’ve got
a better picture,” Schawinski says. The
blue ones that SDSS is seeing tend to be
small, recently formed ones that are still
undergoing vigorous star formation before
moving to a calmer phase. “Ellipticals
quench star formation rapidly,” Schawinski
says. Astronomers can see the process more
clearly because of the huge sample provided
by SDSS and the clear classification from
Galaxy Zoo. “This very good consensus,
on a million objects, you can’t get any other
way,” he says.
Little green galaxies
The success of Galaxy Zoo goes beyond the
big number of eyes involved, researchers say.
Volunteers have played pivotal roles in some
new discoveries.
In the days after Galaxy Zoo was launched,
Pick one. Humans still outperform software in placing galaxies
within Hubble’s classification scheme.
Rarity. Statistics from Galaxy Zoo led one researcher to probe
unusual red spiral galaxies.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 175
SPECIALSECTION
C
R
E
D
I
T
:

H
A
N
N
Y

&

T
H
E

M
Y
S
T
E
R
Y

O
F

T
H
E

V
O
O
R
W
E
R
P

,

T
H
E

Z
O
O
N
I
V
E
R
S
E
,

H
T
T
P
:
/
/
H
A
N
N
Y
S
V
O
O
R
W
E
R
P
.
Z
O
O
N
I
V
E
R
S
E
.
O
R
G
the research team was inundated with
queries, both technical and scientific,
from volunteers. Un able to field all of
the questions, the researchers set up
an online forum so that the volunteers,
or “zooites” as they soon called them-
selves, could interact, veterans helping
newcomers. The researchers could
add comments when necessary and
contribute occasional blog posts. “Set-
ting up the forum was very important.
It acted as an escape valve for things
that didn’t fit,” says William Keel of the
University of Alabama, Tuscaloosa.
One zooite, Dutch schoolteacher
Hanny Van Arkel, spotted several
oddities. They didn’t appear to be
galaxies at all but were more like
green stars. She started a thread on the
forum called “Give peas a chance,”
and other volunteers soon contributed
similar objects they’d found. Before
long they had collected more than 100
“green peas.”
Carolin Cardamone, a colleague
of Schawinski’s at the Massachusetts
Institute of Technology in Cambridge,
started to investigate. “They looked
like stars,” she says, but when she
checked their spectra, “the colors
were very unlike stars.” Instead, the objects
seemed to be very compact galaxies
undergoing extreme star formation and
emitting light strongly at a wavelength
characteristic of highly ionized oxygen,
which appears green in SDSS images.
Using the peas identified by the zooites as
a model, Cardamone developed a program
to find others in the SDSS catalog. Zooites
helped her analyze the new sample, too.
“It’s like having research assistants around
the world working 24 hours a day,” she says.
When Cardamone and others published
a paper on the green pea sample in 2009,
10 zooites were acknowledged for their
contributions. There’s still much debate
about what these objects actually are, but
Schawinski says they most resemble the
protogalaxies thought to have existed very
early in the history of the universe. “They’re
living fossils that we can study in detail
because they’re close by,” he says.
Van Arkel is much better known for
another unusual object she found in Galaxy
Zoo: a glowing green blob now known as
Hanny’s Voorwerp (Dutch for “object”). The
blob happened to be in the image of a galaxy,
IC 2497, but astronomers had no idea whether
it was connected with the galaxy or was much
more distant. Unusually, it was glowing
very brightly at one particular wavelength.
Astronomers got so curious that numerous
telescopes were trained on the voorwerp,
including Kitt Peak in Arizona, the orbiting
Swift, Suzaku, and Hubble telescopes, the
Westerbork radio telescope array in the
Netherlands, and the Merlin array in the
United Kingdom. It was eventually found
to be a gas cloud in the vicinity of IC 2497,
possibly a tidal tail drawn out when another
galaxy passed nearby. Its emissions suggest
it is being illuminated by intense light from
a highly energetic AGN called a quasar, but
there is no quasar in sight.
The voorwerp is still puzzling astro-
nomers. One current theory is that the
center of IC 2497 was a quasar that has now
switched off. But because the voorwerp is
tens of thousands of light-years away, it’s still
being illuminated by light from the quasar. So
in the voorwerp we’re seeing the afterglow
of the quasar although the quasar itself is
gone. There are also suggestions that a jet of
gas streaming out of the center of IC 2497 is
impacting on the voorwerp, causing areas of
star formation.
Hoping to learn more about the life
cycles of quasars, Keel asked the zooites to
find other clouds similarly illuminated
by now-dormant quasars. They sifted
through 18,000 objects. Keel inspected
the best candidates using telescopes
at Kitt Peak and the Lick Observatory
and found 19 promising clouds more
than 30,000 light-years from a potential
quasar. Keel has won time on the
Hubble telescope to examine seven
of them in detail. Working on the
voorwerp, he says, has been “the most
rewarding little chunk of science I’ve
done in the past decade.”
The Galaxy Zoo team learned an
important lesson from the project’s first
phase: Don’t underestimate the abilities
of the zooites. In GZ2 they took the
250,000 brightest galaxies from the
SDSS catalog and scrutinized them for
more-detailed information, such as the
number of spiral arms, how tightly they
are wound, and the length of the bars.
The response was again astonishing:
60 million classifications in 14 months.
The first results using GZ2 data are just
starting to come out.
Meanwhile, zooites are hard at
work on the third phase of the project,
known as Hubble Zoo. It asks almost
the same questions as GZ2 but applies
them to a much deeper data set: Hubble
Space Telescope images of more than
2 million galaxies covering 75% of the age
of the universe. With this temporally deep
sample, astronomers “can compare with
the local universe to quantify things to do
with galaxy evolution that they couldn’t do
before,” Keel says.
In the future, zooites may grapple with
data from even-more-powerful survey tele-
scopes now being planned, such as the Square
Kilometre Array and the Large Synoptic
Survey Telescope (LSST). By the time
LSST starts work later this decade, Keel
says, astronomers expect that computers will
be able to do much of the heavy lifting of
classification. Human classifiers will be used
to improve the software and look at anything
odd the computer has trouble with. “I can
see a tighter synergy between observers and
software,” Keel says.
Even as the use of software expands,
however, Davies says it’s “likely that [Galaxy
Zoo–like] techniques will be widely used in
the future.” That will be a relief to thousands
of galaxy spotters around the globe. “The
community of Galaxy Zoo gives them
the opportunity to participate that they’re
looking for.” –DANIEL CLERY
Artist’s conception. The saga of the Galaxy Zoo has been
dramatized in an online graphic novel.
Published by AAAS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

PERSPECTIVE
Galactic Paleontology
Eline Tolstoy
Individual low-mass stars have very long lives, comparable to the age of the universe, and can thus
be used to probe ancient star formation. At present, such stars can be identified and studied only
in the Milky Way and in the very closest of our neighboring galaxies, which are predominantly small
dwarf galaxies. These nearby ancient stars are a fossil record that can provide detailed information
about the physical processes that dominated the epoch of galaxy formation and subsequent evolution.
O
ne of the major successes inastronomyover
the past century must be our very detailed
and fundamental understanding of stars,
and specifically the way they live out their lives,
burninghydrogenintoheliumandproducingheavier
elements such as carbon, nitrogen, and oxygen. We
also understand howstars end their lives—either as
supernova explosions, producing much heavier and
more exotic elements in the process (such as iron,
calcium, magnesium, and titanium), or slowly fad-
ing away as white dwarfs that contribute almost
nothing to the universe beyond a few photons.
The work of Enjar Hertzsprung and Henry
Norris Russell, at the beginning of the last century,
showed that the colors and luminosities of indi-
vidual stars form distinct sequences for different
groups of stars. The physical reason could not be
understood until the work of Hans Bethe in the
1930s, which explained nuclear burning processes
inside stars. He showed that these color and lumi-
nosity trends were evolutionary sequences, which
depend on the age and mass of a star. This physical
understanding means that we can use these se-
quences to determine the ages of individual stars
by measuring their luminosities. From these se-
quences we knowthat low-mass stars, like our Sun,
have lifetimes comparable to the age of the uni-
verse. If such stars formed in the early universe,
then they will still be around today, in more or less
the same state as they were then, representing
living fossils of past star formation. This provides a
powerful tool to measure when and how many
stars were formed at different ages of the universe,
back to the earliest times (1, 2) (Fig. 1). Thus, for
all galaxies that are near enough to be resolved into
individual stars (e.g., the Sculptor dwarf spheroidal
galaxy, Fig. 2) down to the faint magnitudes of the
lowest-mass stars [about 10
6
parsec distance (3), or
the boundaries of our Local Group (4)], we can
determine detailed star formation histories star by
star, going back to the first low-mass stars (5, 6). It
is only by resolving individual stars that we can
pick out the fewfaint fossil remnants of the most
ancient stellar population, and use their proper-
ties to understand the link between the formation
of the first stars and the first galaxies.
During the 1950s, the work of M. Burbidge,
G. Burbidge, W. Fowler, and F. Hoyle provided a
further critical advance in our understanding of
nucleosynthesis. They explained how(and which)
chemical elements can be produced in a variety of
different conditions, including the interiors of stars
andsupernovae, throughout the universe. This means
that it is only as part of the life cycle of stars that all
chemical elements in the universe heavier than he-
lium(andsome lithium) are formed. Over time, with
each successive generation of stars and their super-
nova deaths, the gas ingalaxies, fromwhichnewstars
form, becomes more enriched in heavy elements.
The outer region of stars, the photosphere,
contains the remnants of the gas from which the
star originally formed. Because the properties of
the photosphere are largely unaffected by the
internal processes of nuclear burning going on in
the core of the star, the gas trapped there can be
considered a fossil record of the gas properties of
the galaxy at the time the star formed and can be
used to trace its chemical evolution.
In the 1920s, thanks to recent advances in quan-
tummechanics, Cecilia Payne was able todetermine
how to interpret the absorption lines seen in the
spectra of stars in terms of the abundances of dif-
ferent chemical elements in the photosphere. She
realized that the multitude of iron absorption lines in
the spectrumof our Sun (andall other stars) does not
imply high iron content, but merely the fact that iron
has numerous possible quantum mechanical tran-
sitions (which produce spectral lines). Proper treat-
ment of the probability of line formation shows that
these lines do not correspond to a high fraction of
iron. In fact, she found that all stars contain mainly
hydrogen and helium, with only a tiny fraction (less
than 2%) of all the other elements. The abundance
of the chemical elements found in a star’s photo-
sphere is referred to as the star’s metallicity (astrono-
mers call all elements heavier thanhelium“metals”).
Low-metallicity stars formed from gas that had not
beenvery much enriched by previous generations of
stars, and thus they are presumed to be older than
higher-metallicity stars.
Kapteyn Astronomical Institute, University of Groningen,
Landleven 12, 9747AD Groningen, Netherlands. E-mail:
etolstoy@astro.rug.nl
-0.5 0 0.5 1 1.5 2
V- I
Main
sequence
Horizontal branch
Red giant branch
Tip
AGB
Blue loop
4
2
0
-2
-4
-6
-8
Fig. 1. The theoretical color-magnitude diagram (or Hertzsprung-Russell
diagram) made from stellar evolutions models, in the filters V (visual)
and I (red), for a galaxy that has formed stars constantly for the last 13
billion years. Each data point represents an individual star. The stars are
color-coded according to their age: blue, stars <300 million years old;
cyan, 300 million to 1.1 billion years old; red, 1.1 billion to 3 billion
years old; green, 3 billion to 8 billion years old; black, >8 billion years
old. Also labeled are a few key stellar evolution phases that clearly mark
the presence of stellar populations of different ages. The Red Giant
Branch contains stars >1 billion years old, which makes it a poor age
discriminator, whereas on the Main Sequence (the longest phase in a
stellar lifetime when hydrogen is fusing to form helium in the core) the
different ages of stellar populations are nicely spread out. The Horizontal
Branch (an almost horizontal sequence of stars at I ~ 0), only contains
stars >10 billion years old and is thus an unequivocal indicator of the
presence of an ancient stellar population. The upper Main Sequence and
Blue Loop stars are uniquely identified with very young stellar pop-
ulations (<1 billion years old), and the extended Asymptotic Giant Branch
(AGB) above the tip of the Red Giant Branch is a mix of intermediate-age
stellar populations. This color-magnitude diagram was made using the
IACstar synthetic computation code (2).
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 176

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Our detailed understanding of stellar properties
means that today we can use photometric and spec-
troscopic studies of individual stars in our Milky
Way and other nearby galaxies to trace the history
of star formation throughout the universe, back to
the first low-mass stars that may have formed even
before the first galaxies (7–9). Extensive surveys
(10) have found small samples of extremely metal-
poor stars, which are presumably the oldest and
formed at very early times. The lowest-metallicity
star ever found anywhere in the universe is an
otherwise unremarkable star in the halo of the
Milky Way, with an iron abundance ~3 × 10
–9
that
of the Sun (11). So far no zero-metallicity star has
been found. This implies that the very first stars
must have been more massive, and thus much
shorter lived, than our Sun. Extremely metal-poor
stars have also been found in nearby dwarf galaxies
(12–17), although their properties are generally
distinct fromthose of the metal-poor stars found in
the Milky Way (Fig. 3).
One of the fundamental tenets of the currently
favored theory of galaxy formation and evolution,
called hierarchical structure formation, is that a
series of mergers combined smaller structures
(dwarf galaxies) to form the large galaxies we see
today, such as the Milky Way (18). The chemical
properties of the individual stars found in different
galaxies can show whether they have had a com-
mon history within the Local Group. By compar-
ing the abundance properties of the individual
stars in the Milky Way to those in the surrounding
swarmof dwarf galaxies, we can see whether such
small galaxies could have combined to form the
Milky Way, and can gain some information about
the time scale over which this could have occurred.
Relative to the Milky Way, dwarf galaxies ap-
pear to show more scatter in abundance ratios, as
well as a much slower enrichment of alpha elements
(Fig. 3). It remains an open question exactly how
well the distribution of extremely low-metallicity
stars in dwarf galaxies compares to that in the Milky
Way, as at present the samples of these stars (rep-
resented on the left side of Fig. 3) are still rather
small (13, 14, 19). The abundance patterns of the
individual stars that make up the different compo-
nents of the Milky Way (halo, thin/thick disk, bulge)
(Fig. 3) show evidence of coherent large-scale star
formation that allows all the components of the
Milky Way to efficiently enrich with metals, often
very rapidly. This suggests that large galaxies can
retain the enriching products of supernovae explo-
sions and can rapidly build up metals in the disk,
bulge, and halo. In contrast, the small dwarf galaxies
we observe today in the Local Group have had a
wide range of star formation histories and do not
contain very metal-rich stars, such as those typical
of the Milky Way disk or bulge (5). This strongly
suggests that in low-mass dwarf galaxies, star for-
mation is never a continuous process; it is instead
liable to progress in fits and starts, probably because
the galaxies are so small that the energetic effects of
star formation tend to disrupt the gas (from which
stars form) so severely that only a very low level
of star formation can allow the galaxy to remain
bound (20, 21). This difficulty for small galaxies
to retain gas, and the products of star formation,
results in a slow buildup in metallicity (22). This
very irregular and inefficient mode of star forma-
tion also leaves its mark in the stellar abundance pat-
tern of a range of chemical elements (5, 7) (Fig. 3).
A comparison between the abundance patterns
of stars in the Milky Way and dwarf galaxies indi-
cates that large and small systems evolve differ-
ently. This means that the bulk of the Milky Way
stellar mass cannot have come fromthe merging of
other galaxies, but that most stars must have formed
fromgas within the Milky Way. This restricts merg-
ing as a dominant process to redshifts, z > 5, which
corresponds to a time when the universe had been
forming stars for 1 billion years or less. It is not yet
Fig. 2. A true color image (35 arc min by 35 arc min) of the Sculptor dwarf spheroidal galaxy taken with the
NOAOCTIOBlanco 4-mtelescope and MOSAIC camera. All the small blue and red dots are individual stars. Some
of the brighter stars (with a red fuzzy halo) are foreground stars in the halo of the Milky Way in front of Sculptor.
The Sculptor dwarf spheroidal galaxy is 9 × 10
4
parsec fromour Sun and well within the Local Group; it contains
~10
5
stars, compared to ~10
11
stars in the Milky Way. This galaxy is a prototype ancient dwarf galaxy and has
been the subject of extensive imaging and spectroscopic surveys of the individual stars [e.g., (5, 13, 19, 23–26)].
[Image: T. J. L. de Boer, Kapteyn Astronomical Institute, University of Groningen, Netherlands]
-4 -3 -2 -1 0
-0.4
-0.2
0
0.2
0.4
0.6
Fe/H

/
F
e
Fig. 3. Abundance measurements of iron ([Fe/H]), with increasing metallicity going from left to right, and
alpha-elements ([a/Fe], an average of calcium, magnesium, and titanium) for individual stars in the Milky
Way (triangles) and a diverse sample of dwarf galaxies within 2.5 × 10
5
parsec of the Milky Way (star
symbols). About half of the dwarf galaxy measurements come from the Sculptor dwarf spheroidal galaxy
shown in Fig. 2. The Milky Way symbols are color-coded according to whether their kinematics suggests
membership in the halo (red), the thick disk (green), or the thin disk (blue) [(5, 8, 9, 12–14, 24, 27, 28)
and references therein]. The average error bars on the measurements are given in the top left hand
corner, in red for the Milky Way stars, and in black for the dwarf galaxy stars. The errors for the Milky Way
stars are on average smaller than those for the dwarf galaxy measurements.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 177
SPECIALSECTION

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

possible to know whether our Milky Way and its
environment are typical, and therefore how general
this conclusion is. The next logical step is to ex-
tend the comparison to individual stars in M31
and its surroundings, but this must wait for a high-
resolution spectrograph on the next generation of
extremely large telescopes. We also need to obtain
larger samples of stellar abundances in our Milky
Way and surrounding dwarf galaxies. The future
of these kinds of study is promising, as large sur-
veys on ever larger telescopes can train their mir-
rors on ever more distant resolved stellar systems.
References and Notes
1. B. Tinsley, Fundam. Cosm. Phys. 5, 287 (1980).
2. A. Aparicio, C. Gallart, Astrophys. J. 128, 1465
(2004).
3. The parsec is a commonly used measure of distance in
astronomy corresponding to the distance from the Sun to
an object that has a parallax of 1 arc sec. This is
equivalent to ~3.26 light years or 3.1 × 10
13
km.
4. The group of galaxies that includes the Milky Way
and where all galaxies are bound together by gravity.
It comprises around 50 small- to medium-sized dwarf
galaxies, two large spiral galaxies (the Milky Way and
M31), and two smaller spirals (the Large Magellanic
Cloud and M33). It has a very irregular shape, but it
is 2 × 10
6
parsec at its broadest extent.
5. E. Tolstoy, V. Hill, M. Tosi, Annu. Rev. Astron. Astrophys.
47, 371 (2009).
6. S. L. Hidalgo et al., Astrophys. J. 730, 14 (2011).
7. A. McWilliam, Annu. Rev. Astron. Astrophys. 35, 503 (1997).
8. B. Edvardsson et al., Astron. Astrophys. 275, 101 (1993).
9. M. D. Shetrone, P. Côté, W. L. W. Sargent, Astrophys. J.
548, 592 (2001).
10. T. C. Beers, N. Christlieb, Annu. Rev. Astron. Astrophys.
43, 531 (2005).
11. N. Christlieb et al., Nature 419, 904 (2002).
12. W. Aoki et al., Astron. Astrophys. 502, 569 (2009).
13. M. Tafelmeyer et al., Astron. Astrophys. 524, A58 (2010).
14. A. Frebel, J. D. Simon, M. Geha, B. Willman, Astrophys. J.
708, 560 (2010).
15. E. N. Kirby, J. D. Simon, M. Geha, P. Guhathakurta,
A. Frebel, Astrophys. J. 685, L43 (2008).
16. J. G. Cohen, W. Huang, Astrophys. J. 701, 1053 (2009).
17. J. E. Norris et al., Astrophys. J. 689, L113 (2008).
18. S. D. M. White, C. S. Frenk, Astrophys. J. 379, 52
(1991).
19. E. Starkenburg et al., Astron. Astrophys. 513, A34
(2010).
20. M.-M. Mac Low, A. Ferrara, Astrophys. J. 513, 142
(1999).
21. Y. Revaz et al., Astron. Astrophys. 501, 189 (2009).
22. G. A. Lanfranchi, F. Matteucci, Astron. Astrophys. 468,
927 (2007).
23. T. J. L. de Boer et al., Astron. Astrophys. 528, A119 (2011).
24. M. D. Shetrone et al., Astrophys. J. 125, 684 (2003).
25. E. Tolstoy et al., Astrophys. J. 617, L119 (2004).
26. G. Battaglia et al., Astrophys. J. 681, L13 (2008).
27. B. Letarte et al., Astron. Astrophys. 523, A17 (2010).
28. K. A. Venn et al., Astrophys. J. 128, 1177 (2004).
Acknowledgments: This was written while I was a visitor at
l’Observatoire de la Cote d’Azur, and I am grateful for the
support of the visitor program there. Supported by a
VICI grant from the Netherlands Organization for
Scientific Research. I thank F. Fraternali, V. Hill, and
E. Starkenburg for comments and discussion.
10.1126/science.1207392
REVI EW
The Cosmic History of Star Formation
James S. Dunlop
Major advances in observational astronomy over the past 20 years have revolutionized our view of cosmic
history, transforming our understanding of how the hot, smooth, early universe evolved into the complex
and beautiful universe of stars and galaxies in which we now live. I describe how astronomers have
used a range of complementary techniques to map out the rise and fall of star formation over 95% of
cosmic time, back to the current observational frontier only ~500 million years after the Big Bang.
T
he cosmic history of star formation is our
own history, or at least our prehistory. It is
only through the lives and deaths of suc-
cessive generations of stars that the atomic compo-
sition of the universe has been enriched (albeit only
slightly) to contain atoms such as carbon, oxygen,
and nitrogen, atoms that are essential for organic-
basedlife. The past historyof star-formationactivity
even affects today’s financial markets, with the seem-
ingly ever-rising price of rare commodities such as
gold being due, in large part, to the rarity and brevity
of the violent supernova explosions in which all
gold was originally forged.
The formation of one particular star, our Sun,
has of course been especially important to us, as
it provides all the energy to power life on Earth.
Fortunately, or perhaps inevitably (otherwise we
probably wouldn’t be here), this energy source is
remarkably stable and long-lived. The geological
evidence fromwithin the solar systemindicates
that the Sun has been burning for ~5 billion years,
and astronomers now understand enough about
stellar evolution to be confident that the Sun will
burn for a similar amount of time again, before
expanding into a red giant en route to eventual
death as a white dwarf stellar remnant.
We thus live in a fairly stable and peaceful
corner of the universe, and indeed our entire
Milky Way Galaxy of ~100 billion stars appears
to be evolving gently and steadily, forming stars
at the relatively modest rate of ~3 solar masses per
year. However, there are many pieces of evidence
to suggest that the universe was once a much more
violent place, with stars being formed at a much
higher rate than is seen around us today. At the
same time, we also knowthat, at very early times,
in the so called “dark ages,” there can have been
no stars at all until the inferno following the Big
Bang cooled to a temperature that allowed the first
clouds of primordial gas to collapse. Charting the
cosmic history of star-formation activity, from
the first stars to the present day, has thus long
been a fundamental goal of astronomy.
How to Measure Star-Formation History
Astronomers are fortunate in having four inde-
pendent lines of evidence through which to map
out the past history of star formation in galaxies.
Institute for Astronomy, University of Edinburgh, Royal Ob-
servatory, Edinburgh, EH9 3HJ, UK. E-mail: jsd@roe.ac.uk
Fig. 1. As shown by the Hubble Space Telescope image of the elliptical galaxy ESO 325-G004 (left),
massive galaxies in the nearby universe are overwhelmingly dominated by old red/yellow stars formed
more than 10 billion years ago and show little evidence of recent star-formation activity. By contrast,
lower-mass galaxies generally continue to form stars in the present day, as shown by the young blue star
clusters, red clumps of ionized hydrogen gas, and dark regions of cool dust and gas shown in the Hubble
Space Telescope image of the spiral galaxy NGC 3982 (right).
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 178
C
R
E
D
I
T
:
N
A
S
A
/
E
S
A
/
S
T
S
C
I
/
A
U
R
A
/
T
H
E
H
U
B
B
L
E
H
E
R
I
T
A
G
E
T
E
A
M

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

First, like archaeologists, astronomers can ex-
amine the “fossil evidence” of past star-formation
trends in the stellar populations of present-day gal-
axies. Detailed spectroscopic observations of the
light from these relatively nearby galaxies are in
principle straightforward, and the challenge is to
dissect their integrated starlight into stellar pop-
ulations of measurable mass, age, and chemical
composition (so-called “metallicity”). This ap-
proach is especially powerful for probing recent
star-formation activity but, like all archaeology,
provides decreasing clarity with increasing look-
back time. It also provides little direct information
on where the various stellar populations found in
present-day galaxies actually formed (i.e., in situ or
in separate subunits that subsequently merged).
Second, unlike any other area of science,
astronomers have the luxury of being able to look
directly back in time, exploiting the finite speed
of light to study the star-formation activity in
galaxies at ever earlier times simply by looking at
galaxies at ever greater distances. The relevance
of this approach relies on an extension of the
Copernican principle that we live in an ordinary
region of the universe and that, averaged over
sufficiently large scales, the universe is essential-
ly the same everywhere. Only then can the ob-
served behavior of “galaxies back then over
there” be meaningfully related to “galaxies back
then over here,” allowing us to infer what our
own region of the universe may have looked
like at comparably early times. Provided one ac-
cepts this (observationally supported) assump-
tion, the challenges are primarily technical due
to the extreme faintness of the most distant gal-
axies, the progressive redshifting of their light
with increasing distance [due to the expansion
of the universe, observed wavelength = (1 + z) ×
emitted wavelength, where z = redshift], the po-
tentially confusing effects of interstellar dust on the
visibility of young stars, and issues over the best
observational tracers of star-formation activity.
Third, again using observations of high-
redshift galaxies to look back to earlier epochs,
astronomers can use infrared observations to mea-
sure the stellar masses of galaxies and hence chart
the buildup of stellar mass with cosmic time. Be-
cause most stars (like our Sun) are relatively long-
lived, the global stellar mass density at a particular
time should reflect the time integral of all preceding
star-formation activity. To some extent, this mea-
surement can therefore be viewed as simply a con-
sistency check on the second method described
above. However, it is in fact of more value than
this, because the stellar masses of galaxies are
dominated by large numbers of relatively low-mass
long-lived stars, whereas the direct measurement of
star-formation activity is, in practice, confined to
observations of the most massive, luminous short-
lived stars. Thus, comparison of the results fromthis
and the previous method has the potential to provide
information on the ratio of the numbers of low-mass
to high-mass stars formed in star-forming regions
(the so-called “Initial Mass Function”) and whether
this has changed substantially over cosmic time.
Finally, it is possible to set constraints on the
history of cosmic star formation by attempting to
measure how the average chemical composition
of the universe has changed over cosmic time.
All elements heavier than hydrogen have been
produced by nuclear fusion, and the very early hot
universe certainly provided the necessary high
temperatures and densities for fusion to take place.
However, the rapid expansion and consequent cool-
ing of the universe in the immediate aftermath of
the Big Bang meant that nuclear fusion could only
be sustained for ~15 min, with the result that only
the first stage of fusion, fromhydrogen to helium,
was properly completed before the universe en-
tered the “dark ages.” The first stars, referred to as
Population III stars, must therefore have formed
from material with the so-called “primordial com-
position” of 75% hydrogen and 25% helium (by
mass; 92%and 8%byatomic number density), with
only minute trace amounts of light metals such as
lithium. These first Population III
stars did not last long (none have
ever been discovered surviving to
the present day). However, through
internal nuclear fusion followed
by supernova explosions, they
must have commenced the pro-
cess of chemical enrichment that,
through recycling in the inter-
stellar medium, produced suc-
cessive generations of increasingly
metal-rich stars (so called Popu-
lation II, and then Population I,
such as our own Sun). Fromsen-
sitive spectroscopic observations
of known emission and absorp-
tion lines fromdifferent elements
in galaxies and more diffuse re-
gions of the universe at various
redshifts, it is possible to con-
struct the cosmic history of chem-
ical enrichment, which then sets
constraints on the history of star
formation.
The Fossil Evidence
Over the past decade there have
been several major studies of the stellar popula-
tions in the present-day nearby galaxy population.
Akey advance has been the creation of very large
spectroscopic surveys, in particular the Sloan
Digital Sky Survey (SDSS) (1), which has now
provided a public release of fully calibrated op-
tical spectra for ~1 million galaxies in the local
universe. The SDSS covers a large enough volume
of the local universe for the results to be regarded as
representative of the universe of galaxies in the
present day. With such a large, complete, and re-
presentative galaxy sample, it is possible to explore
how, for example, galaxy age depends on mass,
size, or morphological type. In addition, because
the consistent and well-calibrated SDSS data have
nowbeen made public, different groups of research-
ers have been able to undertake different indepen-
dent analyses and openly explore the robustness
and accuracy of the derived results (2–5).
Given such a vast and rich data set, the chal-
lenge is to maximize the information that can be
reliably extracted from the spectral database. This
is done by comparing the data with the predictions
of computer-generated models of what the inte-
grated spectra of stellar populations of different
ages, initial mass functions, and metallicity should
look like as a function of age. A lot of effort has
been invested in the development of such models
over the last ~30 years (6–8), and they rely on an
accurate theoretical description of how a star of
given mass and chemical composition will evolve
over time (9) combined with an accurate predic-
tion of precisely what spectrum of light will be
produced by the stellar atmosphere of a given star
at a given stage in its life. To tackle this latter issue,
theoretical predictions can, at least for a subset of
stars, be cross-checked with detailed spectroscopy
of nearby stars (10).
The models are certainly not yet perfect, and
even with clever and efficient data analysis tech-
niques (4) there are limitations to how robustly the
true star-formation history of a galaxy can be deter-
mined from the final integrated spectrum of all its
constituent stellar populations. Nevertheless, some
clear and unambiguous trends have been estab-
lished by this work.
First, it is clear that the most massive (gen-
erally elliptical) galaxies are the oldest, and the
simplest, with generally very little recent star for-
mation (4, 11) and spectra that can be described
10
10
-3
0.01
0.1
5 0
Time (Gyr)

S
F
R
(
M
s
u
n

y
r
-
1

M
p
c
-
3
)
Fig. 2. A simple representation of our current knowledge of the rise
and fall of globally averaged star-formation activity over the 13.7
billion years of cosmic history. The black line indicates our best es-
timate of how the density of star formation (in solar masses formed
per year per unit of co-moving volume) grew rapidly in the first 2
billion years after the Big Bang, stayed roughly constant for a further
~2.5 billion years, then has declined almost linearly with time since
the universe was ~5 billion years old. The red lines indicate the typical
current uncertainty in the measurement, which rises to approximately
an order of magnitude at the earliest times.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 179
SPECIALSECTION

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

by between one and three distinct stellar popula-
tions (12). By contrast, galaxies with more typical
masses have younger average stellar ages and
more complicated star-formation histories, with
as many as five different stellar populations being
required to fit the SDSS spectra (Fig. 1). Second,
integrating over the entire galaxy population, over-
all star-formation density (i.e., star formation per
unit volume in the universe) is inferred to have
declined monotonically since redshift z ~ 2, when
the universe was ~3 billion years old, at which
point this approach to deducing cosmic star-
formation history basically runs out of steam.
These are not all new results [for example,
there has long been evidence that star formation
has moved fromthe most massive to the least mas-
sive systems over cosmic time (13)], but this work
has clarified and robustly quantified the key trends
that must be explained by any successful model of
galaxy formation and evolution. For example, the
lack of substantial present-day star formation in
massive galaxies has led theorists to invoke feed-
back from the central super-massive black hole to
explain how residual gas in and around the most
massive galaxies can be prevented from cooling
and forming yet more stars (14).
Looking Back in Time
As already mentioned, astronomers are not just
stuck with the fossil evidence, and the past 20 years
have seen a veritable explosion in the detection
and study of galaxies at ever greater distances,
which are hence viewed at ever earlier times. This
work has been technologically driven, both by
the advent of giant (8- to 10-mdiameter) ground-
based telescopes and by a series of spectacularly
successful new space observatories (especially
the Hubble Space Telescope, the Spitzer Space
Telescope, and the Herschel Space Observatory)
that, free from the effects of Earth’s atmosphere,
have facilitated sensitive observations over the in-
frared region of the electromagnetic spectrum.
Amultifrequency, multifacility approach is im-
portant in this work because there are a number
of different probes of the level of star-formation
activity in distant galaxies, and ideally one would
like to be able to exploit these at all distances and
epochs, despite the progressive redshifting of the
light as one looks back to the earliest galaxies.
It has also become clear that, because young stars
are born in clouds of gas and dust, much of the
blue or ultraviolet light produced by hot young
stars is in fact often absorbed by interstellar dust
and then re-emitted at much longer (infrared) wave-
lengths. Thus a panchromatic view is required to
obtain a complete census of the “action.”
Observationally and theoretically, it is clear
(and unsurprising) that stars formover a wide range
of masses. However, although the precise shape
of the initial mass function is a subject of contin-
ual study and debate (15–17), it is the relatively rare,
most-massive stars that, in practice, provide the
most useful tracers of star-formation activity, both
because they shine only briefly (i.e., for a few mil-
lion years) and because they are so bright and hot.
Young massive stars provide a number of use-
ful indicators (at different wavelengths) of the level
of star-formation activity (18, 19). These include
bright optical-ultraviolet continuum light [from the
stars themselves (20, 21)], bright hydrogen and
oxygen emission lines [from the surrounding gas
ionized by the hot stars (22–24)], enhanced emis-
sion at mid- and far-infrared wavelengths [from
dust warmed by the ultraviolet light from the stars
(25–27)], and radio emission [fromrelativistic elec-
trons accelerated by the shock waves produced by
the supernovae explosions that mark the deaths
of these short-lived massive objects (28)].
Various authors have discussed the pros and
cons of each of these star-formation indicators
and assessed the prospects for combining different
measures to produce a complete, unbiased, and
consistent measurement of star-formation activity
over the widest possible range of redshifts (29–33).
This is work in progress but, nevertheless, in recent
years a consistent picture has emerged, at least out
to redshift z ~ 2. As illustrated in Fig. 2, essentially
all tracers of star-formation activity indicate that the
star-formation rate (per unit volume) in the universe
was an order of magnitude greater at z ~ 1 than in
the present day (34, 20, 35) and that star-formation
density stays at comparable or even higher levels
out to at least redshift z ~ 2 (24, 36–38), with a
plausible peak at z ~ 3 (39). Notably, this basic
form of evolution over the past 12 billion years is
in excellent agreement with that derived from the
latest analyses of the fossil evidence.
At higher redshifts the picture becomes less
clear, in part because the galaxies are more dis-
tant and hence fainter, but also because the most
easily accessible star-formation indicators are
ultraviolet continuum and emission lines (i.e.,
Lyman-alpha), which, although helpfully red-
shifted into the optical wavebands, are also the
most susceptible to the highly uncertain effects
of dust extinction (40, 41). Nevertheless, in the
past ~10 years, enormous strides have been
made in the discovery and study of large sam-
ples of galaxies at z = 3 to 6 (42–44), and the
ultraviolet colors of the galaxies themselves can
be used to make plausible corrections for dust
extinction. At present, the available evidence sug-
gests that beyond z ~ 3, as we look back into
the first 2 billion years of cosmic history, star-
formation density declines gradually but steadily
out to at least z ~ 6, when the universe was ~1
billion years old (21, 45). At extreme redshifts,
z > 7, even the ultraviolet light from young stars
is redshifted out of the optical and into the in-
frared. As a result, the effective study of the first
billion years of star-formation activity has had to
await the advent of sensitive near-infrared imaging
on the Hubble Space Telescope.
The Buildup of Mass and Metals
Sustained star formation over the majority of
cosmic time must inevitably produce a gradual
Fig. 3. The deepest ever near-infrared image of the sky, taken with the new WFC3 on board the
refurbished Hubble Space Telescope (lower right). This image, shown at left, is in a small region of sky
called the Hubble Ultra Deep Field (HUDF) and has led to the discovery of the first galaxies at redshift
z = 7 to 8 (indicated by yellow circles). The upper-right panel shows a zoom in to the 1.6-mm image of
one of these galaxies, at redshift z = 7.2, seen when the universe was only 750 million years old.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 180
C
R
E
D
I
T
:
J
.
S
.
D
U
N
L
O
P
A
N
D
N
A
S
A
/
E
S
A
/
S
T
S
C
I

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

buildup in the total mass of stars in the universe.
In recent years, improvements in near- and mid-
infrared observational facilities have made it pos-
sible to estimate the stellar masses of galaxies out
to the highest redshifts, hence enabling a direct
check on whether the whole picture hangs to-
gether (46, 47). The answer seems to be that there
is excellent agreement out to z ~ 1, but at higher
redshifts some researchers have argued that there
do not seemto be as many stars in place as would
be expected, given the integrated total of all the
observed preceding star-formation activity [after
accounting for mass loss caused by stellar evo-
lution processes such as supernovae and stellar
winds (48)]. It would be a mistake to exaggerate
the severity of this discrepancy [the numbers are
only in disagreement by a factor ~2 to 3, or ar-
guably less (49)], but the direction of disagreement
suggests that either the level of star-formation
activity in the young universe has been over-
estimated (e.g., by excessive upward corrections
for dust extinction) or the integrated stellar masses
of all galaxies at each epoch have been system-
atically underestimated. Both of these options
are still possible (49), as is the more speculative
possibility of time evolution of the stellar initial
mass function (50, 51). However, the direction of
this tension certainly makes it hard to argue that
the star-formation density at early times has been
seriously underestimated, thus reinforcing the
argument for a gradual decline in universal star-
formation activity beyond z ~ 3 (Fig. 2). Broadly
speaking, the data all suggest that about half the
stars in the present-day universe were in place by
z ~ 2, when the universe was ~3 billion years old.
A second long-lived legacy of preceding star-
formation activity is the abundance of the heavier
chemical elements. Because they are only formed
in massive stars that live for less than 10 million
years, the growth in the cosmic abundance of
elements such as carbon or iron should provide
a fairly prompt (in cosmological terms) indicator
of recent star-formation activity, delayed only
by the time scale for expulsion of the newly
formed atomic nuclei into the wider interstellar
medium by supernovae explosions. However,
conducting a full observational census of the
cosmic budget of chemical elements at all epochs
is extremely challenging because the “metals”
(i.e., any element heavier than helium) can hide
in different places at different redshifts (52). Spe-
cifically, in the present day most of the heavier
elements are now either locked up in stars and
planets (like our Earth) or are found in the hot-gas
phase between galaxies. By contrast, when the
universe was young, those heavy elements that
had already been produced almost certainly re-
sided primarily in cool gas and dust grains within
galaxies (53). In addition, heavy element abun-
dance is also a strong function of environment;
it appears that massive galaxies are, and always
have been, more metal rich than their lower-mass
counterparts (54). Still, the search has now been
conducted with sufficient thoroughness that as-
tronomers are now reassured that the measured
abundances of the elements at z ~ 2 to 3 are in
agreement with that expected from the integrated
star-formation activity at earlier epochs (55). In-
terestingly, there now appears to be evidence of
a relatively rapid downturn in carbon abundance
as we look back toward z = 6, suggestive of an
extremely rapid buildup of stars in the immedi-
ately preceding ~500 million years (55).
Epilogue: Searching for the First Galaxies
and Stars
The recent refurbishment of the Hubble Space
Telescope with the sensitive near-infrared Wide
Field Camera 3 (WFC3) has enabled the first
detections of star-forming galaxies at redshifts
z > 7 (Fig. 3 (56–59)), providing a first glimpse
into the “epoch of reionization,” the ~600-million-
year era in which a range of circumstantial evi-
dence indicates the first stars and galaxies switched
on and reionized the previously cold, neutral, and
dark universe (60).
We now know that stars and galaxies existed
at z ~ 8.5, extending our study of cosmic star-
formation history back to within 500 million
years of the Big Bang (56, 61, 62), and a lot of
work is currently being expended in trying to
determine whether there are enough young gal-
axies at these early times to explain cosmic re-
ionization (63). The very first galaxies are of
course, by definition, expected to contain very
young stellar populations of very low metallicity,
and claims that some of the galaxies at z > 7
discovered with WFC3 are exceptionally blue
have sparked an ongoing and lively debate as to
whether such primitive stellar populations have
indeed now been observed (64–66). It remains
to be seen whether the next generation of even
more powerful astronomical facilities, in partic-
ular the James Webb Space Telescope, can dis-
cover the elusive first generation of Population III
stars, thus completing our journey through the
cosmic history of star formation.
References and Notes
1. D. G. York et al., Astron. J. 120, 1579 (2000).
2. C. A. Tremonti et al., Astrophys. J. 613, 898 (2004).
3. G. Kauffmann et al., Mon. Not. R. Astron. Soc. 353, 713
(2004).
4. A. F. Heavens, B. Panter, R. Jimenez, J. Dunlop, Nature
428, 625 (2004).
5. H. Mathis, S. Charlot, J. Brinchmann, Mon. Not. R.
Astron. Soc. 365, 385 (2006).
6. A. G. Bruzual, S. Charlot, Astrophys. J. 405, 538 (1993).
7. R. Jimenez, J. MacDonald, J. S. Dunlop, P. Padoan,
J. A. Peacock, Mon. Not. R. Astron. Soc. 349, 240 (2004).
8. C. C. Conroy, J. E. Gunn, Astrophys. J. 712, 833 (2010).
9. A. Bressan et al., Astron. Astrophys. Suppl. Ser. 100, 647
(1993).
10. P. Coelho, B. Barbuy, J. Meléndez, R. P. Schiavon,
B. V. Castilho, Astron. Astrophys. 443, 735 (2005).
11. B. Panter, R. Jimenez, A. F. Heavens, S. Charlot,
Mon. Not. R. Astron. Soc. 378, 1550 (2007).
12. R. Tojeiro et al., Astrophys. J. Suppl. Ser. 185, 1 (2009).
13. L. L. Cowie, A. Songaila, E. M. Hu, J. G. Cohen, Astron. J.
112, 839 (1996).
14. T. M. Heckman, G. Kauffmann, Science 333, 182 (2011).
15. E. E. Salpeter, Astrophys. J. 121, 161 (1955).
16. G. Chabrier, Publ. Astron. Soc. Pac. 115, 763 (2003).
17. C. Weidner et al., Mon. Not. R. Astron. Soc. 412, 979 (2011).
18. R. C. Kennicutt Jr., Annu. Rev. Astron. Astrophys. 36, 189
(1998).
19. A. M. Hopkins et al., Astrophys. J. 599, 971 (2003).
20. P. Madau et al., Mon. Not. R. Astron. Soc. 283, 1388
(1996).
21. R. J. Bouwens et al., Astrophys. J. 705, 936 (2009).
22. L. J. Kewley, M. J. Geller, R. A. Jansen, Astron. J. 127,
2002 (2004).
23. M. Doherty et al., Mon. Not. R. Astron. Soc. 370, 331
(2006).
24. D. Sobral et al., Mon. Not. R. Astron. Soc. 398, 75
(2009).
25. D. H. Hughes et al., Nature 394, 241 (1998).
26. K. I. Caputi et al., Astrophys. J. 660, 97 (2007).
27. D. Elbaz et al., Astron. Astrophys. 518, L29 (2010).
28. J. J. Condon, Annu. Rev. Astron. Astrophys. 30, 575 (1992).
29. N. A. Reddy et al., Astrophys. J. 633, 748 (2005).
30. E. Daddi et al., Astrophys. J. 670, 156 (2007).
31. R. C. Kennicutt Jr. et al., Astrophys. J. 703, 1672 (2009).
32. D. Burgarella et al., Publ. Astron. Soc. Jpn. 61, 177 (2009).
33. G. E. Magdis et al., Astrophys. J. 714, 1740 (2010).
34. S. J. Lilly, O. Le Fèvre, F. Hammer, D. Crampton,
Astrophys. J. 460, L1 (1996).
35. K. G. Noeske et al., Astrophys. J. 660, L47 (2007).
36. E. Le Floc’h et al., Astrophys. J. 632, 169 (2005).
37. N. A. Reddy et al., Astrophys. J. Suppl. Ser. 175, 48 (2008).
38. R. R. Chary, A. Pope, Astrophys. J.,
http://arxiv.org/abs/1003.1731 (2011).
39. A. M. Hopkins, J. F. Beacom, Astrophys. J. 651, 142 (2006).
40. D. Calzetti, Astron. J. 113, 162 (1997).
41. S. Charlot, S. M. Fall, Astrophys. J. 539, 718 (2000).
42. C. C. Steidel et al., Astrophys. J. 592, 728 (2003).
43. R. J. Bouwens, G. D. Illingworth, M. Franx, H. Ford,
Astrophys. J. 686, 230 (2008).
44. R. J. McLure, M. Cirasuolo, J. S. Dunlop, S. Foucaud,
O. Almaini, Mon. Not. R. Astron. Soc. 395, 2196 (2009).
45. M. Ouchi et al., Astrophys. J. 723, 869 (2010).
46. K. I. Caputi, R. J. McLure, J. S. Dunlop, M. Cirasuolo,
A. M. Schael, Mon. Not. R. Astron. Soc. 366, 609 (2006).
47. A. Fontana et al., Astron. Astrophys. 459, 745 (2006).
48. S. M. Wilkins, N. Trentham, A. M. Hopkins, Mon. Not. R.
Astron. Soc. 385, 687 (2008).
49. N. A. Reddy, C. C. Steidel, Astrophys. J. 692, 778 (2009).
50. R. Dave, http://arxiv.org/abs/1008.5283 (2010).
51. P. G. van Dokkum, C. Conroy, Nature 468, 940 (2010).
52. M. Pettini, http://arxiv.org/abs/astro-ph/0303272 (2004).
53. L. Dunne, S. A. Eales, M. G. Edmunds, Mon. Not. R.
Astron. Soc. 341, 589 (2003).
54. B. Panter, R. Jimenez, A. F. Heavens, S. Charlot,
Mon. Not. R. Astron. Soc. 391, 1117 (2008).
55. E. V. Ryan-Weber, M. Pettini, P. Madau, B. J. Zych,
Mon. Not. R. Astron. Soc. 395, 1476 (2009).
56. R. J. McLure et al., Mon. Not. R. Astron. Soc. 403, 960
(2010).
57. P. A. Oesch et al., Astrophys. J. 709, L16 (2010).
58. S. L. Finkelstein et al., Astrophys. J. 719, 1250 (2010).
59. A. J. Bunker et al., Mon. Not. R. Astron. Soc. 409, 855
(2010).
60. J. Dunkley et al., Astrophys. J. 701, 1804 (2009).
61. R. J. Bouwens et al., Astrophys. J. 709, L133 (2010).
62. N. R. Tanvir et al., Nature 461, 1254 (2009).
63. B. E. Robertson, R. S. Ellis, J. S. Dunlop, R. J. McLure,
D. P. Stark, Nature 468, 49 (2010).
64. R. J. Bouwens et al., Astrophys. J. 708, L69 (2010).
65. D. Schaerer, S. de Barros, Astron. Astrophys. 515, A73
(2010).
66. J. S. Dunlop et al., Mon. Not. R. Astron. Soc., http://arxiv.
org/abs/1102.5005 (2011).
Acknowledgments: The author acknowledges the support of
the Royal Society through a Wolfson Research Merit
award and the support of the European Research Council
through the award of an Advanced Grant.
10.1126/science.1200644
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 181
SPECIALSECTION

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

REVI EW
The Coevolution of Galaxies
and Supermassive Black Holes:
A Local Perspective
Timothy M. Heckman
1
* and Guinevere Kauffmann
2
One of the most fascinating discoveries in the past decade was that galaxies typically contain a centrally
located black hole with a mass that is millions or even billions of times that of the Sun. There is
now compelling evidence that we cannot understand how galaxies formed and evolved without
understanding the life cycles of these supermassive black holes (and vice versa). We summarize the
current understanding of this coevolution of galaxies and supermassive black holes (based largely
on observations of the local, present-day universe) and describe prospects for the future.
O
f all the denizens of the cosmos studied
by astronomers, none are as fascinating
as black holes: objects whose gravita-
tional field is so strong that light itself cannot
escape from inside their aptly named “event
horizon.” Predicted long ago, the existence of
these exotic objects has now been established
observationally. Black holes come in two vari-
eties: The first have masses roughly 10 times
that of our Sun and are formed during the cata-
clysmic events at the death of a very massive star.
The second are found in the centers (“nuclei”)
of galaxies and have masses that range roughly
between one million and one billion times the
Sun’s mass. For this reason, they are dubbed su-
permassive black holes (1).
One of the most unexpected and important
discoveries of the past decade has been that the
lives of galaxies and their supermassive black
holes are inextricably intertwined. This is note-
worthy, because the scales of the two phenome-
na are so different: The galaxy is more than one
billion times larger than the black hole and con-
tains more than one thousand times as much
mass. Many astronomers now believe that we
cannot understand how galaxies formed and
evolved without understanding their black holes
in detail. Understanding the link between gal-
axies and their black holes has become one of
the most important problems in astrophysics.
Background
The mass of an object can be determined by its
gravitational effect on satellite systems orbiting
about the object: Measure the speed of the satellite
and the radius of its orbit to determine the mass of
the central object. Such measurements have re-
vealed the presence of a very large but invisible
mass concentration in the centers of 50 nearby
galaxies (1). In our own Milky Way (Fig. 1), it is
established that the mass is concentrated within
such a small radius that ordinary matter would
have no other option but to collapse into the event
horizon of a supermassive black hole (2, 3).
Today, most supermassive black holes are in a
relatively dormant state, producing little energy (4).
However, these black holes do experience occa-
sional episodes of intense growth (“feeding fren-
zies”) when they swallownearby gas clouds. As this
gas swirls around the black hole onto the event
horizon, it is heated to high temperatures and emits
an immense amount of radiation. Once beyond the
event horizon, this gas can no longer be seen. How-
ever, it does add to the mass of the black hole. This
is a major way by which black holes grow, but as
we will discuss, in some feeding frenzies black holes
may also gain mass when they swallow each other!
In addition to the radiation emitted by in-falling
hot gas, supermassive black holes can produce
powerful “jets.” These jets are two very narrow
streams of material traveling in opposite directions
outward from the black hole at velocities close to
the speed of light. The particles in the jet travel out
to distances far larger than the galaxy, where they
finally decelerate as they interact with surrounding
gas and light up in two radio-emitting lobes.
Einstein’s theory of relativity tells us that the
maximumamount of energy Ethat can be extracted
froma mass mis E = mc
2
(the complete conversion
of mass into energy) (c, the speed of light in a vac-
uum). Accretion of matter by a
black hole releases roughly 10%
of mc
2
, whereas nuclear reactions
in stars extract only 0.7%. This
high efficiency means that super-
massive black holes in a feeding
frenzy are the most powerful
known sources of energy: They
can literally outshine the entire
galaxy in which they live. They
are cosmic beacons visible clear
across the universe. By measuring
their numbers and their properties,
we can trace the growth of super-
massive black holes across the
whole sweepof cosmic time (5, 6).
A galaxy such as our Milky
Way contains ordinary matter in
the form of stars and gas. The
stars are arrayed in two primary
components: (i) a roughly spheri-
cal and slowly rotating “bulge,”
which dominates the inner part
of the galaxy, and (ii) a very flat
and more rapidly rotating “disk”
that dominates over the bulge at
larger radii (Fig. 2). These two
components make up the clas-
sical visible region of a galaxy,
which lies at the center of a much
larger halo. The halo consists pri-
marily of dark matter, but also contains a major
reservoir of gas that will accrete onto the galaxy
and provide the rawmaterial to formnewstars and
grow the black hole.
Galaxies can be divided into two main pop-
ulations (7, 8). One population consists of galaxies
that have young, hot stars that are currently forming
out of cold, dense gas. These are the “living” gal-
axies. The other class consists of galaxies with low
star-formation rates and gas fractions. These are the
“passive” galaxies. Observations showthat the frac-
tion of passive galaxies has increased substantially
over the past 8 billion years (9, 10), presumably
R.A. (")
D
e
c

(
"
)
0.4
0.4
0.2
-0.2
-0.4
0.
0.2 -0.2 -0.4 0.
Fig. 1. Plot of the orbits of stars around the supermassive black hole in
the nucleus of our own Milky Way. The supermassive black hole lies at
the origin of the coordinate system, and the region shown is ~0.15 light
years across. Applying Newton’s law of gravity to these orbits yields a
black hole mass 4 million times that of the Sun. Dec (declination) and
R.A. (right ascension) are the sky coordinates relative to the supermas-
sive black hole measured in seconds of arc ("). [Source: S. Gillesen et al.,
Max Planck Institute for Extraterrestrial Physics]
1
Center for Astrophysical Sciences, Department of Physics
and Astronomy, Johns Hopkins University, Baltimore, MD
21218, USA.
2
Max Planck Institute for Astrophysics, Garching,
Germany.
*To whom correspondence should be addressed. E-mail:
heckman@pha.jhu.edu
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 182

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

because some process has disrupted the cold gas
supply in many members of the living population.
Coevolution: The Evidence
There are two main lines of evidence that the life
cycles of galaxies and supermassive black holes
are strongly linked. With the help of the Hubble
Space Telescope as well as large ground-based
observatories, astronomers were able to assem-
ble a sample of supermassive black holes with
accurate mass determinations (1, 11). To their
surprise, the astronomers then found that the
mass of each supermassive black hole was rough-
ly a fixed fraction (about one part in one thou-
sand) of the mass of its bulge (1, 11–13). This
discovery implied that processes acting over bil-
lions of years of cosmic time in galaxies must have
caused 0.1 percent of the mass of stars forming
the bulge to be fed into the black hole. Because
bulge formation is believed to be a highly com-
plex process, this is a rather surprising conclusion.
A second, complementary line of evidence
comes from using telescopes as time machines to
probe the past histories of galaxies. Because of
the finite velocity of light, the deeper out we probe
in distance, the farther we look back in time. We
have now managed to detect galaxies and super-
massive black holes at a time when the universe
was less than one billion years old. When we track
the growth rates of galaxies and supermassive black
holes as a function of cosmic time, we see very
similar trends for both populations (6). In particular,
the cosmic histories of the growth rates of both
galaxies and supermassive black holes depend very
strongly on mass (5, 14, 15). The most massive
galaxies and black holes grew rapidly early in the
history of the universe, and their masses then
remained nearly constant. In contrast, lower-mass
galaxies and black holes have grown at a more
uniform rate over the entire history of the universe.
Put another way, the sites of the most vigorous star-
formation and the most active black hole growth
have moved from the most massive galaxies to
progressively smaller and smaller systems as the
universe evolved (16–18).
Given the marked similarity in the histories
of galaxy and black hole growth, the natural con-
clusion is that the formation and evolution of
these two kinds of objects are tightly linked. The
important question is how we can explain this
in physical rather than phenomenological terms.
The Current Theoretical Paradigm
Theoretical models for the coevolution of galaxies
and supermassive black holes are based on com-
bining analytic models or numerical simulations
of structure formation in the dark matter com-
ponent of the universe, with educated guesses
about how small-scale processes such as star for-
mation and black hole accretion operate in prac-
tice (19–22).
Over the course of cosmic time, galaxies grow
through two main mechanisms: accretion of gas
from the external environment and mergers be-
tween two or more galaxies. In galaxy mergers, the
thin and highly ordered disk component of each
galaxy is scrambled. During the merger, tidal forces
between the two galaxies drain away angular mo-
mentumfromthe cold gas in the disk of the galaxy,
allowing it to flow into the inner region, where it
fuels an intense burst of star for-
mation and delivers a fresh supply
of gas to the supermassive black
hole. The scrambled disk material
settles into a newly created bulge.
If the merging galaxies contained
their own supermassive black
holes, these too would merge to
form a single larger one.
The release of energy from the
merger-induced starburst is so in-
tense that it may blow away most
or all of the remaining gas in a
powerful outflow. In addition, jets
from the black hole may pump
energy into the surrounding gas,
preventing it from cooling, falling
into the galaxy, and forming new
stars. The end result of this process
is a single galaxy with a larger bulge
than was present in either progeni-
tor galaxy, as well as a substantially
more massive black hole. In some
cases, star formationwill cease, caus-
ing a living galaxy to be converted
into a passive one. Because jets re-
lease their energy most efficiently in
the hot gaseous material that sur-
rounds galaxies located in massive dark-matter
halos, death comes earliest to galaxies in the
crowded environments within such halos.
Testing the Paradigm with
Present-Day Galaxies
This paradigm does a reasonably good job of re-
producing the global statistical properties of the
present-day galaxy population, such as the relative
numbers of galaxies with different stellar masses
and star-formation rates. However, this does not
prove that the assumptions incorporated in the
model are correct. To test this, we need to demon-
strate that the same model can reproduce a wide
variety of observational data, not only for present-
day galaxies and supermassive black holes, but
also in the early universe.
Much more detailed information is currently
available for present-day galaxies, because the
objects are nearer and easier to study. We will there-
fore emphasize observational tests in the present-
day universe in this article. However, we will also
comment on future prospects for investigating the
early (distant) universe when the most massive
galaxies and black holes were rapidly growing.
The most massive galaxies and their black
holes. The most massive black holes today (masses
of about 100 million to a few billion times the
mass of the Sun) are no longer accreting a sub-
stantial amount of gas; thus, their masses are
growing very slowly (14, 15). These black holes
are found in the most massive galaxies with the
most massive bulges (12, 13). Such galaxies are
currently forming stars at a much smaller rate
than in the distant past, and are usually lacking a
Fig. 2. Hubble Space Telescope image of the Sombrero Galaxy. There are two components to spiral galaxies like this one: (i) a
roughly spherical bulge whose light dominates the inner region of the galaxy, and (ii) a thin disk component that dominates the
light in the outer region. This disk component is seen nearly edge-on in the Sombrero Galaxy. The disk contains opaque clouds of
gas and dust that appear in this image as a black band of obscuration. [Source: NASA/European Space Agency (ESA)/Space
Telescope Science Institute (STScI)/Association of Universities for Research in Astronomy (AURA)/The Hubble Heritage Team]
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 183
SPECIALSECTION

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

major supply of the cold, dense gas needed to
form new stars (8, 23).
Studies of massive galaxies with x-ray observ-
atories in space indicate that the majority are sur-
rounded by a halo of hot and tenuous gas (24).
Black holes in these galaxies are observed to
produce large-scale jets. The jets emit electro-
magnetic radiation that is most easily observed
using radio telescopes. The combination of radio
and x-ray observations of nearby groups and clus-
ters led to the discovery that jets are able to dump
considerable energy into the hot
gas atmospheres surrounding
nearby massive galaxies (Fig. 3).
If cold, dense gas is present it too
can be heated by the jet. The heat-
ing of gas may be sufficient to
prevent the gas from cooling and
forming stars (24–26). These ob-
servations provided the main mo-
tivation for including feedback
fromjets in the theoretical models.
Intermediate mass galaxies
and black holes. Less massive
black holes in the present uni-
verse (masses of roughly 1 million
to 100 million times the mass
of the Sun) are still accreting gas
at a substantial rate, on average
(14, 15). However, only a small
fraction of these black holes are
accreting at very high rates at
any given time. This means that
the growth occurs sporadically,
presumably when gas is able
to refuel the accretion disk sur-
rounding the black hole. When
activated, these black holes pro-
duce a copious supply of light,
but they do not produce large
and powerful jets (24).
There is a strong connection
observed between the growth of
these black holes and the for-
mation of stars in the bulge-
dominated inner region of the
galaxy (27–30). If we average
over a suitably large sample of
such black holes, we find that
the ratio between the rate of star-
formation in the inner regions of the galaxy and
the rate of black hole growth is about 1000 (14).
One thousand-to-one is the same as the ratio
of the mass of stars to that in the black hole for
the most massive galaxies. As we have discussed,
stars and black holes in such systems were formed
at very early epochs. Thus, the process that deter-
mined the 1000-to-1 ratio in the early universe is
still in action today. What exactly is this process?
If we knew, it would be very good news for ob-
servers, because they would then be able to
study its inner workings in a tremendous amount
of detail, using observational facilities on the
ground and in space. For the moment, all we
have are some educated guesses.
Testing Model Assumptions
Fueling. The hypothesis made in many theoretical
models is that supermassive black holes experience
major growth episodes during a merger between
two galaxies, because this process delivers a fresh
supply of fuel to the black hole. Observations show
that there is a very high concentration of gas and a
high star-formation rate in the central regions of
merging galaxies (31). During a merger, galaxies
approach each other very closely, and tidal forces
cause distortions in their shapes. In recent years,
it has become possible to test the hypothesis of fuel-
ing via mergers by comparing the environments
and shapes of galaxies with actively growing black
holes to carefully chosen control samples of gal-
axies with dormant black holes. These studies
have demonstrated that there is no clear con-
nection between mergers and the fueling of black
holes in the present-day universe (32, 33).
There is a strong connection observed between
star formation in the central region of a galaxy
and the growth of a supermassive black hole, sug-
gesting a common source of fuel. Star formation
requires a supply of cold dense gas, but a merger
is only one way to deliver this gas into the central
region of the galaxy. Irrespective of howcold dense
gas is transported to the central region of the gal-
axy where it forms stars, some of this gas is also
transported all the way inward to a region that is
millions of times smaller, where it can be ingested
by the supermassive black hole. Howthis happens
is still a mystery. In particular, we do not knowwhy
the ratio of the amount of star
formation and black hole growth
is 1000 in bulges.
Feedback. If large-scale dy-
namical processes such as merg-
ers do not regulate the growth
of supermassive black holes,
some other mechanism must be
found to explain the fixed 1000-
to-1 ratio between bulge mass
and black hole mass. The co-
evolution of galaxies and su-
permassive black holes clearly
requires more than a common
source of fuel. It also requires
some physical communication
between the black hole and gal-
axy that limits the growth of each
component—in other words,
some type of feedback.
Two modes of feedback are
well established observationally.
In the case of the most massive
black holes, feedback is observed
to be provided by jets (Fig. 3).
Upon reaching the galaxy’s halo,
the jets deposit energy that dis-
places and heats the halo gas. This
process may limit the growth of
the most massive galaxies and su-
permassive black holes by in-
hibiting the ability of the hot gas
to cool and form stars (24–26).
In lower-mass galaxies, the
rapid growth of black holes is
associated with high rates of star
formation that produce many
young stars that emit copious
amounts of ultraviolet radiation.
This radiation from these stars exerts pressure
on the surrounding dusty medium, blowing apart
the dense, cold cocoons of gas in which the stars
were born. The stars rapidly burn through their
supply of nuclear fuel and die in spectacular
fashion in explosive events called supernovae,
which violently heat the surrounding gas. When
the rate at which massive stars are formed and
die becomes large enough, the combined effect
of all the supernova explosions powers an out-
flowing galactic wind (Fig. 4) that can sweep
away the galaxy’s gas and carry it out into the
halo (34, 35).
Fig. 3. Twin jets produced by the supermassive black hole in the galaxy Messier 84 are
strongly interacting with the hot tenuous gas in the galaxy’s halo. The black hole is
located at the center of the brightest region in this image. The jets are traced by their
radio emission (shown in red) as mapped with the Very Large Array (VLA), and the hot
gas is traced by its x-ray emission (shown in blue), as mapped with the Chandra X-ray
Observatory. The jets decellerate as they travel out into the halo and light up as two
lobes of radio emission. These lobes form two cavities in the hot gas as they push it out
of their way. [Source: x-ray, NASA/Chandra X-ray Center (CXC)/Max Planck Institute for
Extraterrestrial Physics/A. Finoguenov; radio, NSF/National Radio Astronomy Observa-
tory/VLA/European Southern Observatory/R. A. Laing; optical, Sloan Digital Sky Survey]
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 184

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

It is not yet clear whether this kind of feed-
back can really remove so much gas from the
galaxy that the galaxy is transformed from liv-
ing to passive, as is assumed in some theoret-
ical models. It may be that gas is removed only
from the central region of the galaxy, limiting
the growth of the inner bulge and the black hole
(36, 37). Could this process lead to the 1000-
to-1 mass ratio?
Although star formation rates in these re-
gions may decrease for some time, fresh gas that
is brought in from the outer disk or ejected from
dying stars may rejuvenate the galaxy once more,
leading to a cycle of birth, death, and rebirth (38).
Summary and Future Prospects
Over the past decade, we have learned that not
only do supermassive black holes exist, but that
they are a crucial component of a complex cos-
mic ecosystem that links them together with
dark matter, stars, and gas. This ecosystem has
evolved dramatically over the history of the uni-
verse, and supermassive black holes have played
a surprisingly major role in this evolution. For-
tunately, we now have a working conceptual
framework for understanding this ecosystem
and the role played by supermassive black holes.
However, there are serious gaps in our knowledge.
On the theoretical side, we can make a lot of
progress as the ever-increasing power of com-
puters enables us to create more complete and
realistic numerical simulations. Eventually these
simulations may enable us to calculate the im-
portant physical processes directly, without having
to make the kinds of reasonable (but possibly
wrong) assumptions that we are forced to make
in current simulations.
On the observational side, progress is likely
to come on several fronts. First, with the Atacama
Large Millimeter Array (ALMA), we will be
able to map out the detailed properties of the gas
in the regions very close to supermassive black
holes in present-day galaxies. This is the “food
source” for the supermassive black hole, and we
will learn precisely how this food is delivered to
the black hole. The James Webb Space Telescope
and the next generation of giant ground-based
optical-infrared telescopes and x-ray observato-
ries will work in concert with ALMA to provide
detailed information about how star-formation,
gas, and feedback processes are related to the
formation and growth of supermassive black
holes in the early universe. NASA’s NuStar mis-
sion will be able to peer into the dust-enshrouded
regions where many black holes are growing.
Finally, there is an urgent need for large, complete
surveys of the early universe that combine imag-
ing with spectroscopy. This would give us a ro-
bust picture of the overall demographics of galaxies
and supermassive black holes at early times. This
could be compared to the demographics of the
present-day universe, as has been determined
by the Sloan Digital Sky Survey. Several such
projects are under development.
It has been an exciting and surprising dec-
ade. The future looks even better.
References and Notes
1. L. Ferrarese, H. Ford, Space Sci. Rev. 116, 523 (2005).
2. R. Genzel, F. Eisenhauer, S. Gillessen, Rev. Mod. Phys.
82, 3121 (2010).
3. A. Ghez et al., Astrophys. J. 689, 1044 (2008).
4. L. Ho, Annu. Rev. Astron. Astrophys. 46, 475 (2008).
5. W. N. Brandt, G. Hasinger, Annu. Rev. Astron. Astrophys.
43, 827 (2005).
6. A. Marconi et al., Mon. Not. R. Astron. Soc. 351, 169 (2004).
7. I. Baldry et al., Astrophys. J. 600, 681 (2004).
8. D. Schiminovich et al., Astrophys. J. Suppl. Ser. 173, 315
(2007).
9. M. Brown et al., Astrophys. J. 654, 858 (2007).
10. S. Faber et al., Astrophys. J. 665, 265 (2007).
11. K. Gültekin et al., Astrophys. J. 698, 198 (2009).
12. A. Marconi, L. Hunt, Astrophys. J. 589, L21 (2003).
13. N. Häring, H.-W. Rix, Astrophys. J. 604, L89 (2004).
14. T. M. Heckman et al., Astrophys. J. 613, 109 (2004).
15. J. Greene, L. Ho, Astrophys. J. 667, 131 (2007).
16. L. Cowie et al., Astron. J. 112, 839 (1996).
17. K. Noeske et al., Astrophys. J. 660, L47 (2007).
18. B. Panter, R. Jimenez, A. F. Heavens, S. Charlot,
Mon. Not. R. Astron. Soc. 378, 1550 (2007).
19. D. Croton et al., Mon. Not. R. Astron. Soc. 365, 11 (2006).
20. P. Hopkins et al., Astrophys. J. Suppl. Ser. 163, 1 (2006).
21. G. Kauffmann, M. Haehnelt, Mon. Not. R. Astron. Soc.
311, 576 (2000).
22. V. Springel, T. Di Matteo, L. Hernquist, Mon. Not. R.
Astron. Soc. 361, 776 (2005).
23. A. Saintonge et al., Mon. Not. R. Astron. Soc.,
http://arxiv.org/abs/1103.1642 (2011).
24. B. R. McNamara, P. E. J. Nulsen, Annu. Rev. Astron. Astrophys.
45, 117 (2007).
25. P. Best, C. R. Kaiser, T. M. Heckman, G. Kauffmann,
Mon. Not. R. Astron. Soc. 368, L67 (2006).
26. E. Churazov, M. Bruggen, C. R. Kaiser, H. Bohringer,
W. Forman, Astrophys. J. 554, 261 (2001).
27. G. Kauffmann et al., Mon. Not. R. Astron. Soc. 346, 1055
(2003).
28. G. Kauffmann et al., Astrophys. J. Suppl. Ser. 173, 357 (2007).
29. G. Kauffmann, T. M. Heckman, Mon. Not. R. Astron. Soc.
397, 135 (2009).
30. H. Netzer, Mon. Not. R. Astron. Soc. 399, 1907 (2009).
31. D. B. Sanders, I. F. Mirabel, Annu. Rev. Astron. Astrophys.
34, 749 (1996).
32. C. Li, G. Kauffmann, T. M. Heckman, S. D. M. White,
Y. P. Jing, Mon. Not. R. Astron. Soc. 385, 1915 (2008).
33. T. Reichard et al., Astrophys. J. 691, 1005 (2009).
34. T. M. Heckman, L. Armus, G. Miley, Astrophys. J.
Suppl. Ser. 74, 833 (1990).
35. S. Veilleux, G. Cecil, J. Bland-Hawthorn, Annu. Rev.
Astron. Astrophys. 43, 769 (2005).
36. R. Davies et al., Astrophys. J. 671, 1388 (2007).
37. V. Wild, T. M. Heckman, S. Charlot, Mon. Not. R.
Astron. Soc. 405, 933 (2010).
38. L. Ciotti, J. Ostriker, Astrophys. J. 665, 1038 (2007).
10.1126/science.1200504
Fig. 4. Feedback driven by a burst of formation of massive stars in the galaxy Messier 82. This picture
combines a true-color optical image taken by the Hubble Space Telescope of the galaxy itself (the aqua and
pink band of light running from left to right across the image) with an image taken by the Chandra X-ray
Observatory of a hot wind of gas traveling out into the halo (shown in blue) and with an infrared image taken
by the Spitzer Space Telescope of much cooler gas and dust being carried in the outflow (shown in red).
[Source: x-ray, NASA/CXC/Johns Hopkins University/D. Strickland; optical, NASA/ESA/STScI/AURA/Hubble
Heritage Team; infrared, NASA/Jet Propulsion Laboratory–Caltech/University of Arizona/C. Engelbracht]
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 185
SPECIALSECTION

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Adult Neural Function Requires MeCP2
Christopher M. McGraw,
1,2
Rodney C. Samaco,
3
Huda Y. Zoghbi
1,3,4
*
R
ett syndrome (RTT) is a postnatal neuro-
logical disorder characterized by autistic
symptoms, cognitive and motor abnor-
malities, as well as decreased brain growth during
childhood (1). RTTis due to mutations in MECP2,
which encodes the epigenetic regulator methyl-
CpG-binding protein 2 (MeCP2). The onset of
RTT symptoms during a critical period of brain
development suggests that the function of MeCP2
in the maturing nervous system is critical for
establishing normal adult neurological function.
Although recent evidence (2) has shown that re-
expression of MeCP2 in symptomatic mice that
lack Mecp2 rescues several features of disease, it
remains unknown whether providing MeCP2 func-
tion exclusively during early postnatal life might
be sufficient to prevent or mitigate disease in adult
animals. In other words, if the nervous system es-
tablishes a normal epigenetic program during early
life, would neurological function be protected
after later loss of MeCP2?
To address this question, we developed an
adult onset model of RTT by crossing mice har-
boring a floxed Mecp2 allele [Mecp2
f lox
(3)] and a
tamoxifen-inducible CreERallele [CAGGS-CreER
(4)] to delete Mecp2 when animals are fully mature
(postnatal day 60 or older). Thus, MeCP2 expres-
sion is eliminated only during adult life. Tamoxifen
given daily intraperitoneally at 100 mg/kg for
20 days effectively reduces whole-brain MeCP2
levels in Mecp2
f lox/y
; CreER
+/−
mice (Fig. 1, A
andB). Vehicle-treatedMecp2
f lox/y
; CreER
+/−
mice
did not experience substantial recombination
(fig. S1, A and B).
Mice lacking Mecp2 as adults (AKO) develop
symptoms of disease and behavioral deficits sim-
ilar to germline null (KO) mice. By 10 weeks after
dosing, AKO mice are less active, have abnormal
gait, and develop hind-limb clasping, similar to
10- to 11-week-old KOmice (Fig. 1C). AKOmice
also develop motor abnormalities and impaired
nesting ability, as observed in KO mice (Fig. 1D).
In addition, both AKO and KO mice show im-
paired learning and memory (Fig. 1D).
Adult deletion of Mecp2 also demonstrates
that some genes whose expression levels are
sensitive to MeCP2 abundance are altered in its
absence (5). In total, we tested 10 genes whose
expression levels are known to be altered in KO
mice (Sst, Grin2a, Htr1a, Oprk1, Tac1, Nxph4,
Bdnf, Gal, Lphn2, and Odz3), and 60% are sig-
nificantly altered in AKOanimals compared with
that in wild-type controls (P < 0.05) (Fig. 1E and
fig. S2). However, four of these altered genes
(Htr1a, Oprk1, Tac1, and Nxph4) are also signif-
icantly altered in control Mecp2
f lox
mice (P<0.05),
suggesting increased sensitivity of these loci to
MeCP2 function (fig. S2).
Lastly, both AKO and KO mice died pre-
maturely with similar median time to death [13
weeks after dosing period (n = 20) versus 13.3
weeks of life (n = 13), respectively] (Fig. 1F).
Taken together, multiple features of disease in
a mouse model of RTTcan be recapitulated after
adult deletion of Mecp2, including disease symp-
toms, behavioral deficits, gene expression changes,
and premature death, indicating that expression
of MeCP2 during early life provides little if any
protection against the disease. Therefore, unlike
the effects of some long-lasting epigenetic instruc-
tions that are programmed during early life (6), the
effects of MeCP2 on gene expression and neu-
rological function appear to be lost soon after
deletion. Moreover, this result argues that the
temporal association of disease with the postnatal
period of neurodevelopment may be unrelated to
any “developmental” or stage-restricted function
of MeCP2, at least in mouse models. The in-
terpretation of MeCP2 function presented here is
consistent with the findings of Guy et al. (2).
However, the previous study does not exclude
the possibility that rescue following reexpression
of MeCP2 could have been achieved in part by
reinvigorating stalled neurodevelopmental pro-
cesses, which would have predicted that AKO
mice should be partly protected against disease.
Our results rule out this possibility and decisively
show the dependence of the mature brain on
MeCP2 function. Lastly, these findings suggest
that therapies for RTT, like MeCP2 function,
must be continuously maintained.
References and Notes
1. M. Chahrour, H. Y. Zoghbi, Neuron 56, 422 (2007).
2. J. Guy, J. Gan, J. Selfridge, S. Cobb, A. Bird, Science 315,
1143 (2007).
3. J. Guy, B. Hendrich, M. Holmes, J. E. Martin, A. Bird,
Nat. Genet. 27, 322 (2001).
4. S. Hayashi, A. P. McMahon, Dev. Biol. 244, 305 (2002).
5. M. Chahrour et al., Science 320, 1224 (2008).
6. I. C. G. Weaver et al., Nat. Neurosci. 7, 847 (2004).
Acknowledgments: We thank S. Baker, Y. Lee, A. Flora,
and L. Chen for discussions and NIH (grants NS057819,
HD024064 to H.Z.; F31-NS073317 to C.M.; T32-NS043124
to R.S.), Baylor Research Advocates for Student Scientists
(C.M.), International Rett Syndrome Foundation, Simons
Foundation, and Rett Syndrome Research Trust (H.Z.) for support.
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1206593/DC1
Materials and Methods
Figs S1 and S2
References
5 April 2011; accepted 19 May 2011
Published online 2 June 2011;
10.1126/science.1206593
BREVIA
0 5 10 15 20 25
0
20
40
60
80
100
Weeks post-treatment
P
e
r
c
e
n
t

s
u
r
v
i
v
i
n
g
WT
CreER
Flox
AKO
0 5 10 15 20 25
0
20
40
60
80
100
Weeks of life
P
e
r
c
e
n
t

s
u
r
v
i
v
i
n
g

WT
KO
0
0.5
1.0
1.5
2.0
2.5
R
e
l
a
t
i
v
e

n
o
r
m
a
l
i
z
e
d

l
e
v
e
l
s

o
f

m
R
N
A
*
**
0
3
6
9
12
S
c
o
r
e

(
0

-

1
2
)
* ***
F
o
o
t
s
l
i
p
s

p
e
r

u
n
i
t

d
i
s
t
a
n
c
e

(
x
1
0
-
2
)
0
2.5
5
*** ***
0
1
2
3 * ***
S
c
o
r
e

(
0

-

3
)
0
50
100
** **
%

t
i
m
e

f
r
e
e
z
i
n
g
0
50
150
250
L
a
t
e
n
c
y

t
o

f
a
l
l

(
s
e
c
)
* ***
Symptoms
DAPI MeCP2 merge
F
l
o
x
A
K
O
A
D
F
B
E
C
R
e
l
a
t
i
v
e

n
o
r
m
a
l
i
z
e
d

M
e
C
P
2

l
e
v
e
l
s
***
0
0.2
0.4
0.6
0.8
1.0
1.2
KO Wild-type
Germline Mecp2 KO
Wild-type CreER Flox AKO
Adult Mecp2 KO
Grid walk
Sst Grin2a
Rotarod Nest-building Fear conditioning
Fig. 1. Adult deletion of Mecp2 recapitulates germline knock-out. (A and B) MeCP2 is depleted in AKO
mice by Western blot of brain lysates [(A), N = 3 to 4 mice per genotype] and by immunofluorescence
in cerebellum (B). Scale bar indicates 50 µm. (C) AKO mice display symptoms of disease. N = 6 to 12
per genotype. (D) AKO mice develop motor and learning impairments similar to those of germline
Mecp2
null/y
(KO) mice. N = 10 to 26 per genotype. (E) Sst and Grin2a mRNA levels are altered in AKO
mice. N = 4 to 12 per genotype. (F) AKO mice die prematurely (left) similar to KO mice (right). N = 10
to 26 per genotype. Data presented as mean T SEM. *P < 0.05; **P < 0.01; ***P < 0.001. WT, wild
type; DAPI, 4´,6´-diamidino-2-phenylindole.
1
Program in Developmental Biology, Baylor College of Medi-
cine, and Jan and Dan Duncan Neurological Research Institute,
Houston, TX 77030, USA.
2
Medical Scientist Training Program,
Baylor College of Medicine, Houston, TX 77030, USA.
3
Depart-
ment of Molecular and Human Genetics, Baylor College of
Medicine, and Jan and Dan Duncan Neurological Research
Institute, Houston, TX77030, USA.
4
Department of Neuroscience
and Howard Hughes Medical Institute, Baylor College of Medi-
cine, and Jan and Dan Duncan Neurological Research Institute,
Houston, TX 77030, USA.
*To whom correspondence should be addressed. E-mail:
hzoghbi@bcm.edu
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 186

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Glycolytic Oscillations and Limits on
Robust Efficiency
Fiona A. Chandra,
1
* Gentian Buzi,
2
John C. Doyle
2
Both engineering and evolution are constrained by trade-offs between efficiency and robustness,
but theory that formalizes this fact is limited. For a simple two-state model of glycolysis, we
explicitly derive analytic equations for hard trade-offs between robustness and efficiency with
oscillations as an inevitable side effect. The model describes how the trade-offs arise from
individual parameters, including the interplay of feedback control with autocatalysis of network
products necessary to power and catalyze intermediate reactions. We then use control theory to
prove that the essential features of these hard trade-off “laws” are universal and fundamental, in
that they depend minimally on the details of this system and generalize to the robust efficiency
of any autocatalytic network. The theory also suggests worst-case conditions that are consistent
with initial experiments.
M
inimizing waste, resource use, and fra-
gility to perturbations in system com-
ponents, operation, and environment
(1) is crucial to the sustainability of systems ranging
fromcells to engineering infrastructure. Hard lim-
its on computation, prediction, energy conversion,
communication, control, and even measurement
are at the heart of modern theories of systems in
engineering and science (2). Unfortunately, lack
of coherence among these subjects makes it dif-
ficult to explore the trade-offs between these
limits, and a more unified theory is needed to
understand and design complex systems. Using
the well-studied problem of glycolytic oscillation
as a case study, we integrate concepts from bio-
chemistry and control theory (3, 4) to explore
the hard limits of robust efficiency.
Glycolytic oscillation, in which the concen-
trations of metabolites fluctuate, has been a clas-
sic case for both theoretical and experimental
study in control and dynamical systems since the
1960s (5–8). Numerous mathematical models have
been developed, from minimal models (9, 10) to
those with extensive mechanistic detail (11). Be-
sides being the most studied control system and
the most common, glycolysis is also conserved
from bacteria to humans and, presumably, has
been under intense evolutionary pressure for ro-
bust efficiency. Thus, new insights are less likely
to be confounded by either gaps in the literature
or evolutionary accidents compared with less well
studied biological circuitry. Nevertheless, the func-
tion of the oscillations, if any, remains a mystery
and one we aim to resolve.
The first step is development of the simplest
possible model of glycolysis that illustrates the
trade-offs caused by autocatalysis. Biologically
motivated minimal models of glycolytic oscilla-
tions exist, but analysis of robustness and effi-
ciency trade-offs has not received much attention.
Such analysis can provide a much deeper under-
standing of the underlying basis of glycolytic os-
cillations as well as illustrate universal laws that
are broadly applicable.
Minimal model of glycolysis. Glycolysis is a
central energy producer in a living cell, consum-
ing glucose to generate adenosine triphosphate
(ATP), which is used throughout the cell. The
first steps of the reaction require ATP, making it
autocatalytic. In early experiments in Saccharo-
myces cerevisiae, investigators observed two syn-
chronized pools of oscillating metabolites (12),
which suggested that a two-state model incorpo-
rating phosphofructokinase (PFK) might capture
some aspects of system dynamics, and indeed,
such simplified models (9, 10) qualitatively repro-
duce the experimental behavior. We propose a
minimal systemwith three reactions (Fig. 1A), for
which we can identify specific mechanisms both
necessary and sufficient for oscillations (Table 1).
x˙ ¼
2y
a
1 þ y
2h

2kx
1 þ y
2g
y˙ ¼ −q
2y
a
1 þ y
2h
þ ðq þ 1Þ
2kx
1 þ y
2g
− ð1 þ dÞ
"


#
¼
"
1
−q
#
2y
a
1 þ y
2h

PFK
þ
"
−1
ðq þ 1Þ
#
2kx
1 þ y
2g

PK
þ
"
0
−ð1 þ dÞ
#

Consumption
ð1:1Þ
In the first reaction in Eq. 1.1, PFK consumes
q molecules of y (ATP) with allosteric inhibition
by ATP. We lump the intermediate metabolites
into one variable, x. In the second reaction, pyr-
uvate kinase (PK) produces q + 1 molecules of y
for a net (normalized) production of one unit,
which is consumed in a final reaction modeling
the cell’s use of ATP. In glycolysis, two ATP
molecules are consumed upstream and four are
produced downstream, which normalizes to q = 1
(each y molecule produces two downstream) with
kinetic exponent a = 1. To highlight essential
trade-offs with the simplest possible analysis, we
normalize the concentration such that the un-
perturbed ( d ¼ 0) steady states are y ¼ 1 and
x ¼ 1=k [the system can have one additional
steady state, which is unstable when (1, 1/k) is sta-
ble]. [See the supporting online material (SOM)
part I]. The basal rate of the PFK reaction and
the consumption rate have been normalized to
1 (the 2 in the numerator and feedback coeffi-
cients of the reactions come fromthese normaliza-
tions). Our results hold for more general systems
as discussed below and in SOM, but the analysis
is less transparent.
As most research does, we focus on allosteric
activation of the enzyme PFK by adenosine
monophosphate (AMP) as the main control point
of glycolysis. We assume that the total concentra-
tion of adenosine phosphates, including adenosine
diphosphate (ADP), in the cell [A
tot
] = [ATP] +
[ADP] + [AMP] remains constant, and the acti-
vating effects of AMP can be modeled as ATP
inhibition. ATP also inhibits PKactivity, although
this has been largely ignored in most models [ex-
cept (13, 14)]. We emphasize its importance and
model both inhibitions through exponents h
and g. We use linearization to focus initially on
RESEARCHARTICLES
1
Department of Bioengineering, California Institute of Tech-
nology, Pasadena, CA 91125, USA.
2
Department of Control
and Dynamical Systems, California Institute of Technology,
Pasadena, CA 91125, USA.
*To whom correspondence should be addressed. E-mail:
fiona@caltech.edu
A B
Fig. 1. (A) Diagram of two-state glycolysis model.
ATP, along with constant glucose input, produce a
pool of intermediate metabolites (phosphorylated
six-carbon sugars), which then produces two ATPs.
ATP inhibits both the first (PFK or PFK-like) and
second (PK or PK-like) reactions. (B) Control theo-
retical diagramof the same system(arrows represent
logical connections, not fluxes). The system without
inhibition or feedback is labeled the “Plant” (P)
[solid box, solid and dotted loop in (A)], whereas the
inhibitory mechanism is considered the “Controller”
(here labeled by its inhibitory strength, H) [dashed
loop in (A) and (B)]. The effect of disturbance d in
ATP demand is modeled as the systemW(see text for
definition).
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 187

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

steady-state error and instability, while highlight-
ing disturbance and control:
"
Dx˙
Dy˙
#
¼
"
−k a þ g
ðq þ 1Þk −qa − gðq þ 1Þ
#"
Dx
Dy
#
þ
"
0
−1
#
d

Disturbance
þ
"
−1
q
#
hDy

Control
ð1:2Þ
The first term on the right-hand side (RHS) gives
the dynamics of the “open loop” plant [P, defined
as Eq. 1.2 when there is no control, i.e., h = 0]
(solid and dotted loop in Fig. 1A or solid box in
Fig. 1B) in response to the second term (dis-
turbance in demand); the third term is the control
on PFK (dashed loop in Fig. 1A).
Elementary analysis. The simplest robust
performance requirement (motivated by the need
to maintain high energy charge) is that the
concentration of y remains nearly constant de-
spite fluctuating demand d. In our model, this
requires that the steady-state error ratio be
small. This ratio is computed by solving for
jDy=dj when
"
Dx˙
Dy˙
#
¼
"
0
0
#
to be:

Dy
d

¼

1
h − a

ð2:1Þ
This ratio is small when |h − a| is large, and
jDy=dj → 0 if and only if h→∞. One trade-off
is that large h requires either high cooperativity
or very tight ATP-enzyme binding, and the re-
sulting complex enzymes are more costly for
the cell to produce. A more interesting trade-off
arises because Eq. 1.2 is stable (see SOM, part II)
if and only if
0 < h − a <
k þ gð1 þ qÞ
q
ð2:2Þ
The left-hand side (LHS) bounds the minimum
feedback strength h required to stabilize the
system, so autocatalysis requires some minimal
enzyme complexity for stability, which is com-
patible with making Eq. 2.1 small. More impor-
tant, combining Eq. 2.1 and Eq. 2.2 constrains
the minimum stable steady-state error to

Dy
d

¼

1
h − a

>
q
k þ gð1 þ qÞ
ð2:3Þ
Equation 2.3 and Fig. 2A [showing the error
bound Eq. 2.3 versus k] illustrate a simple and
elegant trade-off between robustness and effi-
ciency (as measured by complexity and metabol-
ic overhead). Low error requires large h, but to
allow this to be stable, k and/or g must also be
large enough. Large k requires either a more
efficient or a higher level of enzymes, and large g
requires a more complex allosterically controlled
PK enzyme; both would increase the cell’s meta-
bolic load. Thus, fragility directly trades off against
complexity and high metabolic overhead (low
efficiency).
The steady-state error is minimized when h
is chosen so that Eq. 2.3 is an equality, but Eq. 1.1
enters sustained oscillations at this hard limit
(this boundary is called a supercritical Hopf bi-
furcation). Thus, at least in this model, oscil-
lations have no direct purpose but are side effects
of hard trade-offs crucial to the functioning of
the cell and can be avoided at some expense. Note
that robustness means making fragility (steady-state
error and oscillations) small, and efficiency means
making metabolic overhead (enzyme amount and
complexity) small.
Hard limits on robust efficiency. Thus far,
we have described simple trade-offs based on
basic biochemical features of a minimal model.
Our elementary analysis of Eq. 1.2 is consistent
with existing literature yet clarifies in Eq. 2.3
how oscillations are the inevitable side effect of
robust efficiency and trade-offs between steady-
state error and stability. An important next step is
to expand to a more detailed and comprehensive
model and also to extend the analysis to study
global nonlinear stability, stochastics, and worst-
case disturbances. We have explored such dimen-
sions, and the results are consistent, although often
less accessible (most additional modeling details
make the trade-offs worse).
A more fundamental direction, however, is to
rigorously prove that the trade-offs suggested by
Eq. 2.3 are unavoidable regardless of these ne-
glected details, depend only on the basic proper-
ties of autocatalytic and control feedbacks, and
are unlikely to be either artifacts of model sim-
plifications or “frozen accidents” of evolution (of
course, in principle, anything is possible because
there is always some gap between models and
reality.) Fortunately, control theory has been de-
veloped precisely to address such questions in
engineering. Unfortunately, although well known
to engineers and mathematicians, control theory
has not been integrated into other fields. A good
background is given in (4).
Control theory focuses our attention on a more
complete picture of the transient response to dis-
Table 1. Description of model variables, parameters, and control theoretic terms.
Model parameters Definition of terms
x Lumped variable of intermediate metabolites P(s) Open loop response (h = 0) in frequency (s) domain
y output, ATP level
k Intermediate reaction rate WS(s) Weighted response to a disturbance d
d Perturbation in ATP consumption WS(s) = W(s)S(s) where W(s) is the weight
q Autocatalytic stoichiometry S(s) Impulse response to a disturbance d
a Cooperativity of ATP binding to PFK z Zero, the solution to P(z) = 0
h Feedback strength of ATP on PFK p Pole, the solution to W( p) = P( p) = ∞, or D( p) = 0
g Feedback strength of ATP on PK
A B
Fig. 2. Trade-offs between waste, fragility, and complexity due to enzyme complexity and amount.
Enzyme amount affects the intermediate reaction rate k (x axis), plotted for g = 0 (solid line) and g = 1
(dashed line). Large k requires high metabolic overhead, and large g requires high enzyme complexity.
Even small g > 0 enhances the trade-offs, particularly at low k. (A) The y axis shows the system’s steady-
state error, and the curves denote the boundary between stable (above) and oscillatory (below) regions.
(B) The y axis shows the lower bound of the hard limits in Eqs. 3.4 and 3.6.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 188
RESEARCH ARTICLES

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

turbances. Because even temporary ATP depletion
can induce cell death, large amplitude oscillation
can be detrimental (15). Therefore, static steady-
state response alone provides insufficient informa-
tion, and the dynamics must be analyzed carefully.
To this end, we reconsider the linearized model
Eq. 1.2 and allow d = d(t) to be an arbitrary func-
tion of time, although the figures only show re-
sponses of the nonlinear system Eq. 1.1 to step
changes in d(t). The theory is most conveniently
written using frequency-domain transforms % yðsÞ ≜


−∞
yðtÞe
−st
dt, where s ∈ ℂis the (complex) Laplace
transform variable, and frequency w with s = jw
is the Fourier transform variable. We consider
three cases of control: (i) “wild type” with con-
stant h (the case studied above); (ii) a general case
where h is replaced by a controller H with arbi-
trarily complex internal dynamics, constrained
only to stabilize Eq. 1.2; and (iii) no control (h =
H = 0). H is assumed linear and time invariant,
and we write H = H(s).
The weighted sensitivity transfer function de-
fined as WSðsÞ ≜ % yðsÞ=
%
dðsÞ is the response from
d to y. Given Eq. 1.2 and controller H, we can
factor WSðsÞ ¼ WðsÞSðsÞ, where S is called the
sensitivity function and W is the weight, equal to
the uncontrolled (H= h = 0) response fromd to y.
For disturbance d, W(s), S(s), and the open-loop
response P(s) (see SOM, part III) are given by
WðsÞ ¼
s þ k
DðsÞ
SðsÞ ≜
1
1 þ PðsÞHðsÞ
¼
DðsÞ
DðsÞ þ HðsÞð−qs þ kÞ
PðsÞ ¼
−qs þ k
DðsÞ
ð3:1Þ
where D(s) = s
2
+ [k + g + q(a + g)]s − ka. With
constant, stabilizing H(s) = h > a, it follows from
Eq. 3.1 and Eq. 2.3 that the response at frequency
w = 0 is equal to the steady-state error ratio:

Dy
d

¼ jWSð0Þj ¼ jWð0ÞSð0Þj
¼

1
a

a
h − a

¼

1
h − a

>
q
k þ gð1 þ qÞ
ð3:2Þ
S is the primary robustness measure for feed-
back control (3), and |S(s = jw)| measures how
much a disturbance is attenuated (|S( jw)| < 1) or
amplified (|S( jw)| > 1) at frequency w. SðsÞ ≡ 1
when H(s) = 0. The response of y to any other
disturbance can be treated with the appropriate
weight W.
When there is autocatalysis, we can derive
stricter bounds on the response WS and S using
the maximum modulus theorem from complex
analysis (16). In Eq. 3.1, when q > 0, P(s) has a
zero at z = k/q defined as P(z) = 0, which is
positive real [Re(z) > 0]. When a > 0, both W(s)
and P(s) have an unstable pole ( p > 0), defined as
where W( p) = P( p) = ∞, and can be computed by
solving D( p) = 0. So for any stabilizing H: S(z) =
1, S( p) = 0, and neither S(s) nor WS(s) have poles
Fig. 4. Log sensitivity
log|S(jw)| (left) and step
response of the nonlinear
system in Eq. 1.1 to step
changeindemandd (right).
(A) The two-state glycol-
ysis model allows higher
feedback gain h and bet-
ter performance when
thereis anadditional feed-
back loop on PK (g = 1).
h = 4 does not drive the
system into sustained os-
cillation as in the g = 0
case in Fig. 3B. Com-
pared with Fig. 3B, both
the peaks and total area
in log|S(jw)| are lower.
(B) The effects of varying
intermediate reactionrate
k given particular inhibi-
tionstrengths (inthis case,
h = 3 and g = 1). Lower k
results inbothhigher peak
and area under the curve
(left), which translate to
more oscillatory transients
(right).
Fig. 3. Log sensitivity
log|S( jw)| (left) without
ATPfeedback onPK(g =0)
and step response of the
nonlinear system Eq. 1.1
to step change in demand
d (right). The integral of
log|S(jw)| is constrained
by Eq. 3.5 in (A) (left) and
Eq. 3.6 in (B) (left) and is
the same for all h. Only
the shape changes with in-
creasing h. Higher h gives
better steady-state error
with more oscillatory tran-
sient. (A) With no auto-
catalysis (q =0) the system
is stable for all h > 0. (B)
When q = 1, log|S(jw)| is
more severely constrained
by Eq. 3.6 and the system
has sustained oscillations
for large h.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 189
RESEARCH ARTICLES

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

in Re(s) ≥ 0. Hence, the maximum modulus
theorem holds for WS(s) in the positive real
domain Re(s) ≥ 0 (SOM, part III) and
‖WS‖

≜ max
jw
jWSð jwÞj
¼ max
ReðsÞ≥0
jWSðsÞj ≥ jWðzÞSðzÞj
¼
q
k þ qg ð3:3Þ
‖S‖

≜ max
jw
jSð jwÞj ¼

s þ p
s − p
S


z þ p
z − p

ð3:4Þ
The norm ‖WS‖

has a variety of interpretations
(3), the simplest of which is the maximum
sinusoidal steady-state response for any frequen-
cy w. Ideally, both WS and S should be low at all
frequencies, but this contradicts Eq. 3.3 and Eq.
3.4, which hold regardless of the controller used.
The peak ‖WS‖

is always larger than the bound
in Eq. 3.3 for any h, and minimizing steady-state
error |WS(0)| leads to‖WS‖

→∞and oscillations
(fig. S2). How the RHS of Eq. 3.4 varies with k
and g is shown in Fig. 2B; both Eq. 3.3 and Eq.
3.4 are aggravated by small k and g. These are
hard constraints on any stabilizing controller from
y to the first enzyme, no matter how complex the
implementation and, thus, are much deeper than
Eq. 2.3, which applies only for constant H = h.
Conditions such as those in Eqs. 3.3 and 3.4
can be applied to other transfer functions and
weights to provide a rich theoretical framework
for exploring additional trade-offs and details,
including the realistic frequency content of d(t),
appropriate error penalties in y(t) and other sig-
nals, and other sources of noise and uncertainty
(3, 4). A complementary focus is on constraints
that are independent of these details, such as
Bode’s integral formula (3)
1
p


0
lnjSð jwÞjdw ≥ 0 ð3:5Þ
which holds for any linear, stabilizing H that is
causal (i.e., H cannot depend on future values of
y(t). H = h depends only on current values). This
“water bed” effect implies that the net disturbance
attenuation (ln|S( jw)| < 0) is at least equaled by
the net amplification (ln|S( jw)| > 0). It is a general
constraint on WS(s) for any W, which transparent-
ly factors out [lnjWSð jwÞj ¼ lnjWð jwÞSð jwÞj ¼
lnjWð jwÞj þlnjSð jwÞj]. For q = 0, constant con-
trollers H = h achieve Eq. 3.5 with equality, as
illustrated in Fig. 3A. More controller complex-
ity can thus fine-tune the shape of lnjSð jwÞj but
cannot uniformly improve it. Autocatalysis q > 0,
however, makes things worse, because z = k/q is
finite, and Eq. 3.5 can be strengthened to
1
p


0
lnjSð jwÞj
z
z
2
þ w
2

dw ≥
max 0, lnj
z þ p
z − p
j
( )
ð3:6Þ
with z and p as defined above (for proof, see
SOM, part V). It is easily shown that p > 0 when
a > 0 and, otherwise, Eq. 3.6 is just bounded by
0. Hence, autocatalysis always causes positive z
and p, and the integral in Eq. 3.6 is bounded
similarly to that in Eq. 3.4. The low-pass filter
z/(z
2
+w
2
) constrains the water bed effect to below-
frequency w = z. Small z = k/q produces a more
severe limitation, because any disturbance at-
tenuation must be repaid with amplification with-
in a more limited frequency range. The trade-off
in three criteria is shown in Fig. 2B: High k both
stabilizes the system and reduces the bound but
implies high metabolic overhead. How auto-
catalysis and Eq. 3.6 affect dynamics is illustrated
in Fig. 3B. S(0) gives the steady-state error, where-
as the peak in S( jw) corresponds to how “ringy”
the transient y(t) dynamics are at frequency w.
At h = 2, S(0) is large, the peak ‖S‖

is low, and
y(t) has a large steady-state error, which h = 3
lowers but with more transient fluctuations. At
h =4, the systemoscillates at the frequency where
S( jw)→∞. Larger q makes z smaller and perform-
ance worse (more ringy), as shown in fig. S3. The
trade-off in Eq. 2.3 and the difference between
Eq. 3.5 and Eq. 3.6 disappears with no auto-
catalysis (q→0), because the RHS bound in Eq.
2.3→∞and in Eq. 3.6→0. Zero steady-state error
with stability is then possible by taking h→∞.
Complexity and robustness. We have taken
PFK feedback as the main controller, but the
often neglected PK feedback increases enzyme
complexity and plays an important, but subtle,
role in robustness. Put most simply, increasing g
uniformly improves the stability bound in Eq.
2.3. From Eq. 2.2, if q = a = 1, then the system is
stable for all k > 0 if and only if 0 < h − 1 < 2g.
Thus g > 0 is necessary to simultaneously main-
tain acceptable steady-state error S(0) = 1/(h − 1)
and stability for all k > 0. Replacing g = 0 (Fig.
3B) with g = 1 (Fig. 4A) does not change S(0),
but yðtÞ is more damped (and the peaks and in-
tegral in Eq. 3.6 are lower). The h = 4 case is
unstable in Fig. 3B but stable in Fig. 4A. The
effect of g > 0 on the robustness versus efficiency
trade-off involving k gives us insight into how the
system is designed. Although a and q are essen-
tially fixed by the network’s autocatalytic struc-
ture, h and g can be tuned on evolutionary time
scales. Thus, 0 < h − 1 < 2g is biologically plau-
sible and, in fact, is consistent with most es-
timates, which ensures stability for all k > 0 (13).
This allows individual cells to further fine-tune
k > 0 through the many mechanisms that control
enzyme levels, but stability for all k > 0 also
provides robustness to unavoidable noise in gene
expression and enzyme levels (17). Quantifying
this effect would require more detailed modeling
and integration of our hard limits on robustness
to external disturbances with those in (17) on
robustness to internal noise in transcription.
From an engineering perspective, this is a
remarkably clever control architecture, and the
presence of g > 0 suggests that, at least in this
case, evolution favors higher complexity in ex-
change for flexibility in k and robustness. Further
insights come fromthe bound in Eq. 3.6. Because
z = k/q, increasing k improves both sides of Eq.
3.6 and uniformly improves robustness (Fig. 4B),
at the expense of higher enzyme levels. Increas-
ing g decreases p, while leaving z unchanged (the
dependency of p on g is given in equation S3.9),
decreasing ln|(z + p)/(z − p)| (Fig. 2B). This
improves the constraint in Eq. 3.6 and enables
more aggressive controller gains h on PFK. By
itself (when h < a), however, g > 0 cannot stabilize,
and a stabilizing G(s) needs very high complexity
(see SOM, part VI).
Our simple model thus far restricts the con-
troller implementation to ATP inhibition, but other
intermediate metabolites can also have inhibitory
effects. We show in SOM, part VIII, that control
Table 2. Summary of the performance, metabolic overhead, and stability trade-offs in glycolysis. Each parameter in the two-state model presents its
own set of trade-offs.
Parameter Pros Cons
Low q Improves performance limit
Can stabilize the system
Reduces metabolic efficiency
High k Improves performance limit
Can stabilize the system
Increases enzyme complexity
Increases metabolic load
High h Stabilizes the system
Improves steady-state error
Increases enzyme complexity
High h can lead into a limit cycle
Worsens transient oscillations
Additional feedback loop (g > 0) Improves performance limit
Improves stability bounds
Increases pathway complexity
Increases enzyme complexity
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 190
RESEARCH ARTICLES

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

by intermediate metabolites can relax stability
and performance constraints at the cost of lower
efficiency. Intermediate inhibition on PFK can
change both the steady-state error and stability
bounds, whereas intermediate activation of PK
can lift performance constraint (ultimately, the
effects of both are limited by enzyme saturation).
Fructose 1,6-bisphosphate (the product of PFK)
has been thought to both inhibit PFKand activate
PK, which also suggests that nature accepts greater
complexity in return for robustness.
Experiments revisited. Our theory shows
both how autocatalysis makes glycolysis more
prone to sustained oscillations and howsufficient-
ly complex feedback control ameliorates this
potential fragility. The trade-offs summarized in
Table 2 suggest that ringy transient dynamics
would be more likely under specific worst-case
conditions that we have attempted to create ex-
perimentally. Small z = k/q has the most obvious
impact on overall fragility, and this occurs at
high autocatalytic stoichiometry q and/or low
k. Thus, to get a worst-case high-q and low-k
condition, wild-type S. cerevisiae cells (strain
W303) were first grown in ethanol and briefly
starved in phosphate-buffered saline, then rapidly
shifted into anaerobic glucose metabolism (18).
Transcription levels of some glycolytic genes are
decreased when S. cerevisiae is grown in ethanol
(19), which could decrease k. Flow cytometry of
fluorescence-tagged proteins indeed shows lower
concentrations of glycolytic enzymes involved in
the intermediate reactions in cells grown in eth-
anol compared with glucose (18).
Our single-cell autofluorescence measure-
ments of the reduced form of nicotinamide ade-
nine dinucleotide (NADH) showed that a portion
of the cells indeed exhibited fluctuating transients
before settling into a higher NADHlevel (fig. S6),
as expected from a robust controller and roughly
corresponds to 1 ≤ k ≤ 3 in Fig. 4B (right). The
period is in good agreement with the 36-s period
in cell suspensions (20), and this transient does
not occur in cells grown in glucose (fig. S7), also
as expected for high k [e.g., k = 5 in Fig. 4B
(right)]. We observe no sustained oscillation re-
gardless of the experimental perturbations ap-
plied, which suggests that the intact single cell is
indeed rather robust.
In fact, despite intense experimental study,
spontaneous sustained oscillations in yeast have
only been observed in cell-free extracts or in in-
tact cells in dense suspensions but not when iso-
lated (20). Our single-cell model is too simplistic
to be as predictive as the detailed models in the
literature, but because the analysis highlights
fundamental trade-offs, it can give insights into
these different behaviors. For example, in cell-
free extracts, parameters can be pushed into re-
gimes exposing extreme fragilities that wild-type
cells have evolved to avoid. In SOM, part X, we
show that our model and theory are consistent
with observed patterns of oscillations in well-
known extract experiments (5). Of course, the
possibility of single-cell oscillation cannot be
ruled out, and there is much more to be done
theoretically and experimentally to fully resolve
this. The tools and analysis presented here can be
applied to more complete models and, it is hoped,
can clarify future directions. In SOM, part XI, we
further discuss what is needed to address both
dense cell suspensions and isolated cells.
Discussion. Our analysis illustrates the power
of control theory to clarify biological phenome-
na and biology so as to motivate new theoretical
directions (21). In this simple model of glycoly-
sis, oscillation is neither directly purposeful nor
an evolutionary accident but a necessary conse-
quence of autocatalysis and hard trade-offs be-
tween robustness and efficiency (or fragility and
overhead). Nature has evolved a control structure
finely tuned to effectively manage these trade-
offs with flexibility to adapt to changes in supply
and demand, at the cost of higher enzyme com-
plexity. Consistent with engineering, purposeful
complexity in biology is primarily driven by ro-
bustness, not minimal functionality (1), and there
are hard trade-offs that this complexity mediates.
The theory presented here is consistent
throughout in highlighting hard trade-offs, but
there are important differences in the details. Al-
though Eq. 2.3 is phenomenological and specific
to the model in Eq. 1.2, the theory in Eqs. 3.3 to
3.6 is more complete, holding for all frequencies
and arbitrarily complex causal controllers, and
also applying to other systems. However, Eq. 3.6
still requires substantial phenomenology, because
the formulas for z and p depend on assumptions
about autocatalysis (q and a) and enzyme effi-
ciencies and levels (k). It is hoped that this will
encourage efforts in further unification of con-
trol theory with thermodynamics and statistical
mechanics, and recent progress is encouraging
(22). It also leads to rethinking how biology
overcomes the “causality” limit with various
mechanisms that exploit predictable environ-
mental fluctuations (e.g., circadian rhythms) or
provide remote sensing (e.g., vision and hearing),
both of which can greatly mitigate hard limits such
as Eq. 3.6 (23). In the case of circadian rhythms,
oscillation is not just a side effect but has the
purpose of exploiting predictable periodicity in
the environment.
Although our minimal model has limited
quantitative predictive power, it can still provide
qualitative insights about experiments, such as
which parameters to perturb and why extracts
oscillate more easily than isolated cells (SOM,
parts X and XI). To maximize accessibility, we
used the simplest possible model that captures
the real system’s essential features, yet facilitates
theoretical analysis connecting network structure
with functional trade-offs and allows the results
to be carried out analytically [a model’s scope
and fidelity versus ease of theoretical analysis is
itself an inherent trade-off (24)]. The SOM cov-
ers various extensions to our model, including
a nonlinear model of arbitrary length (SOM, part
XII) (25) and reversible reactions (SOM, part
XIII). The effect of reversibility in the inter-
mediate (PK) reaction depends on PK inhibition
strength g and can either ameliorate performance
limit at the cost of efficiency, or make it worse. The
analysis readily scales to more complex models
with appropriate computer-aided design soft-
ware, but the results are far less accessible.
This research article ultimately raises more
questions than it answers, and there is much more
to be done experimentally and theoretically. Tuning
the autocatalytic and control feedbacks via enzyme
mutations to affect robustness is an interesting
direction for future experiments. A relatively easy
theoretical direction that is largely unexplored is to
generalize the bounds in Eqs. 3.3 and 3.4 to
complex multivariable feedback systems involv-
ing more enzymes and metabolites. Control of
additional complex autocatalytic processes, such
as redox balance and biosynthesis of building
blocks and enzymes, is crucial for a more complete
understanding. For example, without aerobic me-
tabolism, NADH is no longer an energy source
but a waste product that must be reduced to nico-
tinamide adenine dinucleotide via other cellular
mechanisms, and then recycled, a potentially
destabilizing autocatalytic loop. The hard limits
can also be generalized to nonlinear systems and
controllers with more complex definitions and
proofs, but many questions remain open (see SOM,
part XV, and references therein). Finally, auto-
catalytic recycling and control feedbacks must
increase and work together effectively in all hu-
man systems as we seek to be more sustainable,
efficient, and robust.
References and Notes
1. M. E. Csete, J. C. Doyle, Science 295, 1664
(2002).
2. These limits are associated with names such as Turing,
Gödel, Poincaré, Carnot, Shannon, Bode, Wiener, and
Heisenberg and are the foundation for subjects
with a vast literature, including textbooks and popular
expositions. Unfortunately, integrated treatments are
surprisingly lacking. For a recent survey of some
dimensions of this problem, see (26).
3. J. C. Doyle, B. A. Francis, A. Tannenbaum, Feedback
Control Theory (Macmillan, New York, 1992).
4. K. Astrom, R. M. Murray, Feedback Systems: An
Introduction for Scientists and Engineers (Princeton Univ.
Press, Princeton, NJ, 2008).
5. A. Ghosh, B. Chance, Biochem. Biophys. Res. Commun.
16, 174 (1964).
6. P. Richard, FEMS Microbiol. Rev. 27, 547 (2003).
7. B. Teusink, B. M. Bakker, H. V. Westerhoff, Biochim.
Biophys. Acta 1275, 204 (1996).
8. M. Bier, B. Teusink, B. N. Kholodenko, H. V. Westerhoff,
Biophys. Chem. 62, 15 (1996).
9. A. Goldbeter, Biochemical Oscillations and Cellular
Rhythms (Cambridge Univ. Press, Cambridge, 1996).
10. E. E. Sel’kov, Eur. J. Biochem. 59, 151 (1975).
11. F. Hynne, S. Danø, P. G. Sørensen, Biophys. Chem. 94,
121 (2001).
12. A. Betz, B. Chance, Arch. Biochem. Biophys. 109, 585
(1965).
13. Y. Termonia, J. Ross, Proc. Natl. Acad. Sci. U.S.A. 78,
2952 (1981).
14. P. H. Richter, J. Ross, Science 211, 715 (1981).
15. B. V. Chernyak, O. Y. Pletjushkina, D. S. Izyumov,
K. G. Lyamzaev, A. V. Avetisyan, Biochemistry (Moscow)
70, 240 (2005).
16. The maximum modulus theorem applied to rational
functions, such as WS(s), says that on any domain in the
complex plane that contains no poles of WS(s), the
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 191
RESEARCH ARTICLES

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

maximum modulus |WS(s)| is always achieved on the
boundary. See (3).
17. I. Lestas, G. Vinnicombe, J. Paulsson, Nature 467, 174 (2010).
18. Materials and methods are available as supporting
material on Science Online.
19. A. Ståhlberg et al., BMC Genomics 9, 170 (2008).
20. A. K. Poulsen, M. Ø. Petersen, L. F. Olsen, Biophys. Chem.
125, 275 (2007).
21. P. A. Iglesias, B. P. Ingalls, Control Theory and Systems
Biology (MIT Press, Cambridge, MA. 2010).
22. H. Sandberg, J. C. Delvenne, J. C. Doyle, IEEE Trans
Auto Control, 56, 293 (2011).
23. N. C. Martins, M. A. Dahleh, J. C. Doyle, IEEE Trans.
Automat. Contr. 52, 56 (2007).
24. A. D. Lander, PLoS Biol. 2, e164 (2004).
25. G. Buzi, U. Topcu, J. Doyle, Automatica, 47, 1123 (2011).
26. D. L. Alderson, J. C. Doyle, IEEE Trans Syst. Man Cybernet.
Part A Syst. Hum. 40, 839 (2010).
Acknowledgments: The authors thank H. El-Samad and
J. Stewart-Ornstein at the University of California,
San Francisco, for their laboratory space and assistance;
N. Pierce (Ray Deshaies’ lab) for the green fluorescent
protein library; O. Venturelli (Richard Murray’s lab) for
her help; and M. Csete for helpful feedback. Microscopy
was performed at the Nikon Imaging Center at UCSF.
Experimental data are available in SOM. This work is
supported by the NIH (award R01GM078992A) and
Institute of Collaborative Biotechnologies from the
U.S. Army Research Office (subaward KK4102, prime
award DAAD19-03-D-0004).
Supporting Online Material
www.sciencemag.org/cgi/content/full/333/6039/187/DC1
Materials and Methods
SOM Text
Figs. S1 to S10
Tables S1 and S23
References (27–31)
22 November 2010; accepted 2 May 2011
10.1126/science.1200705
The Onset of Turbulence in Pipe Flow
Kerstin Avila,
1
* David Moxey,
2
Alberto de Lozar,
1
Marc Avila,
1
Dwight Barkley,
2,3
Björn Hof
1
*
Shear flows undergo a sudden transition from laminar to turbulent motion as the velocity
increases, and the onset of turbulence radically changes transport efficiency and mixing
properties. Even for the well-studied case of pipe flow, it has not been possible to determine at
what Reynolds number the motion will be either persistently turbulent or ultimately laminar.
We show that in pipes, turbulence that is transient at low Reynolds numbers becomes sustained
at a distinct critical point. Through extensive experiments and computer simulations, we were able
to identify and characterize the processes ultimately responsible for sustaining turbulence. In
contrast to the classical Landau-Ruelle-Takens view that turbulence arises from an increase in
the temporal complexity of fluid motion, here, spatial proliferation of chaotic domains is the
decisive process and intrinsic to the nature of fluid turbulence.
T
he seemingly simple question as to when
the flow down an ordinary pipe turns tur-
bulent dates back to the pioneering study
of Osborne Reynolds in the late 19th century (1).
Reynolds proposed that below a critical velocity,
pipe flows are always laminar, whereas above
that critical velocity turbulence prevails, given
the right initial conditions. The observation that
this critical point can be expressed in a dimen-
sionless form was the basis of one of the central
concepts in fluid dynamics: the Reynolds number
(Re = UD/n, where U is the mean velocity, D is
the pipe diameter, and n is the kinematic viscos-
ity). Curiously, although Reynolds similarity has
proved to be valid throughout fluid mechanics
the value of the critical point in pipe flowhas been
debated ever since. In an early attempt to deter-
mine its value (2), Reynolds rewrote the equations
of motion, separating quantities into average and
fluctuating parts—a method that is nowcalled the
Reynolds decomposition. This contribution is gen-
erally regarded as the foundation of modern tur-
bulence research, but it has failed to clarify the
value of the critical point in pipe flow. Values
reported in contemporary textbooks and journal
papers vary widely, typically ranging from 1700 to
2300 (3–5), and occasionally even values in
excess of 3000 (6) are quoted.
One circumstance that complicates this prob-
lem is that laminar pipe flow is stable to infin-
itesimal perturbations (7, 8), and therefore in
order to trigger turbulence, a disturbance of finite
amplitude is required (1, 3, 9). What makes mat-
ters even more difficult is that at low Re, tur-
bulence is transient. Here, turbulence occurs in
the form of localized patches called puffs (10)
that are embedded in the surrounding laminar
flow and decay according to a memoryless pro-
cess (that is, independent of their previous history)
(11). The rapid increase in lifetime with Re has led
to various proposed values for a critical point at
which the lifetime would diverge and turbulence
would become sustained (4, 12, 13). However,
more detailed studies (14–18) have shown that
the lifetime of individual puffs remains finite and
only approaches infinity asymptotically with Re.
Qualitatively, this behavior is reminiscent of the
dynamics of a class of model systems called
coupled map lattices (19). Here, individual lattice
points can exhibit transient chaotic dynamics but
eventually settle to a stable laminar fixed point.
Because of the spatial coupling, these systems
exhibit a statistical phase transition as the control
parameter is increased. Below the critical point,
eventually all sites will end up in the laminar phase,
whereas above there is always a nonzero fraction
of chaotic sites, and with increasing control pa-
rameter the fraction of laminar (nonchaotic) sites
quickly diminishes. Analogies to fluid flows have
been pointed out in a number of studies (20–23)
that indicate the potential relevance of the spatial
dynamics for the long-term behavior in fluid sys-
tems. In a numerical study of pipe flow, Moxey
and Barkley (24) observed that at Re ≈ 2300 tur-
bulent puffs delocalize, and the turbulent fraction
increases, which is in qualitative agreement with
this picture. However, the stochastic nature of the
spatial coupling was not taken into account, and
the extremely long time-scales intrinsic to the flow
could not be resolved in the simulations. In this
work, we determined the critical point in pipe flow
and quantified the relevant process sustaining tur-
bulence in linearly stable shear flows.
Long-pipe experiments. Determining the point
at which the proliferation of turbulence outweighs
its decay and turbulence eventually becomes sus-
tained requires that the time scales of both decay
and spreading processes be captured. Because tur-
bulent puffs move downstreamat approximately
the mean flow velocity, a long pipe is required to
observe long time-scales. Using a precision glass
tube with a relatively small diameter (D = 4 T
0.01 mm) and overall length of 15 m, a total di-
mensionless length of 3750D is achieved. The
pipe is composed of 14 sections joined by ma-
chined perspex connectors that provide an ac-
curate fit. A smooth inlet together with careful
alignment of the individual pipe sections allows
the flow to remain laminar up to Re = 4400. De-
viations in Re were kept below T5 throughout
each set of measurements, which extended over
periods of up to 45 hours. This precision was
achieved with stringent control of both the pres-
sure difference driving the flowand the fluid (wa-
ter) temperature (T0.05 K). A detailed description
of the experimental setup can be found in (16).
Starting from a fully developed laminar flow
allows us to induce turbulence in a controlled
manner and quantify the spreading rate at some
downstream position. The experimental proce-
dure is to create a single turbulent puff close to
the pipe inlet and to monitor any changes in the
turbulent fraction at downstream positions. It is
important that a perturbation is chosen that effi-
ciently triggers turbulence. In many earlier studies,
such as (10, 25), turbulence was induced by in-
sertion of a static obstacle close to the pipe inlet.
Such obstacles provide a continuous perturbation,
and at high Reynolds numbers the flow down-
streamis fully turbulent, whereas in the transitional
1
Max Planck Institute for Dynamics and Self-Organization,
Bunsenstrasse 10, 37073 Göttingen, Germany.
2
Department of
Mathematics, University of Warwick, Coventry CV4 7AL, UK.
3
Physique et Mécanique des Milieux Hétérogènes, UMR 7636
CNRS–École Supérieure de Physique et de Chimie Industrielles
de la Ville de Paris–Univ Paris 06–Univ Paris 07, 10 rue
Vauquelin, 75005 Paris, France.
*To whom correspondence should be addressed. E-mail:
kavila@ds.mpg.de (K.A.); bhof@ds.mpg.de (B.H.)
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 192
RESEARCH ARTICLES

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

regime (Re ≲ 3000) irregular sequences of tur-
bulent and laminar phases are created (25). How-
ever, at lower Reynolds numbers (Re ≈ 2000)
sufficiently strong continuous perturbations cause
plug-shaped velocity profiles that are unable to
sustain turbulence (26) and lead to relaminariza-
tion. Hence, for large disturbance levels contin-
uous perturbations may fail to trigger turbulence
at low Reynolds numbers. To avoid such prob-
lems, in the present study an impulsive perturba-
tion was chosen, consisting of a water jet injected
through a circular hole of 0.2D in the pipe wall
250D downstream of the inlet. The duration of
the perturbation was adjusted for each set of mea-
surements (8 to 20 ms, corresponding to advec-
tion of approximately 1 to 2.5D at the mean flow
velocity), ensuring that only one single puff was
generated from each perturbation. Different am-
plitudes were tested, and the results were found to
be independent of the perturbation strength. In
relation to the mass flow in the pipe, typical in-
jection rates were about 2.5%. To establish that
results were independent of the perturbation, ad-
ditional measurements were carried out by using
an obstacle to impulsively disturb the flow. A
thin wire (0.8 mm in diameter and 10 mm in
length) was inserted into the pipe 150D from the
inlet. The wire was held against the pipe wall and
aligned with the pipe axis by using a small mag-
net attached to the outside of the pipe. When
against the wall, the disturbance created by the
wire is too small to trigger turbulence for the Re
investigated. When the wire was moved azimuth-
ally along the pipe wall by approximately 0.5 to
1 mmin a time interval of about 0.1 s (correspond-
ing to advection of about 10D, based on the mean
flowspeed), a single turbulent puff was triggered.
After triggering disturbances, the flow was
monitored by two downstream pressure sensors
(fig. S1). The first one, located 300D from the
inlet, confirmed that each perturbation results in
the creation of a single puff. The second one,
which can be positioned at various distances L
from the perturbation, was used to distinguish
cases in which multiple puffs arrive (Fig. 1A)
from those in which only the single puff arrives
(Fig. 1B) or no puff arrives. This is a direct mea-
surement of whether the turbulent fraction in the
flow has increased, remained constant, or de-
creased during downstream propagation.
Simulations. To complement experiments
and gain insights into the underlying spread of
turbulence, we have carried out extensive nu-
merical simulations. Two independent numerical
codes have been used; one is a spectral-element
Fourier code (27) that solves the Navier-Stokes
equations in Cartesian coordinates (DNS1), and
the other is a hybrid spectral finite-difference
0
0.5
1
p
r
e
s
s
u
r
e
(
a
.
u
.
)
time (D/U)
0
0.5
1
p
r
e
s
s
u
r
e
(
a
.
u
.
)
A
B
100 100 200
0 0
space (D)
t
i
m
e

(
D
/
U
)
1
5
0
0
0
C
D
Fig. 1. Puff splitting in experiment and numerical simulation. (A and B) Pressure signals from the ex-
periment are used to distinguish the case of (A) a split puff from(B) a single puff. A splitting is registered if
the signal has peaks separated by 20Dor more and if between peaks the signal drops by at least 30%. The
flow between the two puffs does not recover to the fully developed laminar profile. (C) Space-time
diagram from numerical simulation using the hybrid spectral finite-difference code (DNS2) at Re = 2300
showing the splitting process. The square of the cross-sectional average of streamwise vorticity is plotted
on a logarithmic scale in a co-moving reference frame (speed U
p
= 0.929U) and 100D of the 150D
simulation domain are shown. The upstream edge of a puff is relatively well defined, whereas the
downstreamedge is fuzzy and fluctuates. After a splitting, the two puffs propagate downstream, separated
by an approximately constant distance, and generate a twin-peaked pressure signal (A). (D) Visualization
of puff splitting in a cross-sectional (x, y) plane, with red as positive and blue as negative streamwise
vorticity on a linear scale, from the same run as (C) and showing 75D. At t = 0, Re is impulsively changed
from 2200 to 2300. Snapshots (from bottom to top) were taken at t = 500, 990, 1010, 1110, and 1600.
Once the puff extends far enough and the vorticity decays in its central section, a new puff emerges.
0
0
3
0
0
0
1
0
0
0
0 150
space (D)
t
i
m
e

(
D
/
U
)
t
i
m
e

(
D
/
U
)
B
A
Fig. 2. Spreading of turbulence in numerical sim-
ulation. Space-time diagramat (A) Re = 2300 and
(B) Re = 2450 from numerical simulation (DNS2),
showing how turbulence proliferates starting from
a localized puff at Re = 2200 as initial condition.
The cross-sectional average of streamwise vorticity
squared is plotted on a logarithmic scale in a co-
moving reference frame at speed 0.947U and 0.94U,
respectively. At Re = 2300, the expansion process is
dominated by discrete steps, corresponding to puff
splits, whereas at Re = 2450, expansion is more
smooth, more rapid, and individual puffs are no
longer easily identified.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 193
RESEARCH ARTICLES

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

code (28) that solves the equations in cylindrical
coordinates (DNS2). Both methods use periodic
boundary conditions in the streamwise direction
and impose constant unit mass flux, ensuring no
variation in Re during any run. The codes have
been tested both against each other, producing
statistically identical results, and by increasing
the resolution of the discretizations. Details have
been reported elsewhere (27–29).
Puff splitting. For all of the lower Re under
investigation, an increase in the turbulent fraction
manifests itself in the form of “puff splitting”
(10, 30), in which new puffs are seeded down-
stream of existing ones (Fig. 1, C and D). Di-
rectly downstream of a turbulent puff, the flow
has not recovered a parabolic profile and cannot
sustain turbulence (26). To successfully seed a
new puff, a patch of vorticity has to escape far
enough downstream from the initial puff to an
area where the velocity profile is sufficiently par-
abolic. This process can be observed in Fig. 1C,
where regions of large vorticity fluctuations re-
peatedly propagate downstream and decay. Only
during a sufficiently large excursion does a new
puff arise downstream of the old one. Starting
froma single puff, turbulence proliferates through
a sequence of splittings at Re = 2300, as illus-
trated in Fig. 2A. For Re ≤ 2300 for all of our mea-
surements (numerical as well as experimental),
spreading of turbulence and hence any increase
in turbulence fraction (fig. S5) was exclusively
observed in the form of puff splitting. Only at
Reynolds numbers somewhat larger than con-
sidered here (Re > 2400) do individual puffs start
to noticeably expand (Fig. 2B) so that the overall
spreading of turbulence becomes a complex mix-
ture of splitting, as well as growth and merging of
individual turbulent domains.
Characteristic time of memoryless process.
Because of the stochastic nature of the splitting
process, a statistical approach must be used. To
collect splitting statistics, we performed ensem-
bles of numerical simulations starting from inde-
pendent puffs (table S3), fromwhich we determined
the time for each to split. We let P(Re,t) denote
the probability that a puff at Reynolds number Re
splits before time t. Then 1 – P(Re,t) is the prob-
ability for a puff to remain a single turbulent
patch—not split, after time t at Reynolds number
Re. As shown in Fig. 3, 1 – P(Re, t) = exp[–(t –
t
0
)/t(Re)], where t is the observation time, t
0
is a
formation time, and t(Re) is the Re-dependent
characteristic time for the process. The formation
time t
0
includes any equilibration time for the
initial condition to evolve to the turbulence state
at the particular Re and the intrinsic time needed
for splitting. An offset time t
0
arises in puff decay
distributions for analogous reasons (14, 17, 31).
In principle, t
0
depends on Re, as well as on the
initial conditions used in ensemble runs, but from
the two detailed cases in Fig. 3, as well as from
the time of first splitting at other Re, we find
consistently that 100 ≤ t
0
≤ 200.
Experimental measurements (Fig. 3, circles
and squares) also reveal exponential distribu-
tions. Observations were carried out at different
distances L, which were translated into observa-
tion time by t = L/U
p
, where U
p
= 1.482 × 10
−4

2.416 × 10
−4
Re is an approximation to the mean
puff speed in this Re range, as obtained through
numerical simulation (fig. S2). Beyond the initial
formation time, distributions are exponential and
hence memoryless, reflecting that the probability
of splitting does not depend on the age of the puff
under investigation. The splitting probability is
constant in time and characterized solely by t,
which after the initial formation time gives the
mean time for a puff to split and the turbulence
fraction to increase.
Experimentally, it is far easier to keep the ob-
servation point fixed during one series of measure-
ment and vary Re in order to determine P(Re, t),
fromwhich t(Re) can be obtained. The results for
five distances L are shown in Fig. 4. For each data
point, typically 2000 measurements were per-
formed, but for the lowest Re, up to 60,000 mea-
surements were used. As expected, at large Re
the splitting probability is high and decreases
as Re is reduced. Curves for fixed L are S-shaped,
indicating that P →0 only asymptotically as Re
decreases. All measurements were well approxi-
mated by a single superexponential fit with only
two parameters, t = exp[exp(aRe + b)], where a =
–0.003115 and b = 9.161 (Fig. 4, solid lines). In
contrast to earlier studies (10, 24, 30), our data
indicate that there is no critical point at which the
spatial proliferation of turbulence abruptly sets
in. The tendency to split appears to be intrinsic to
turbulence even at low Reynolds numbers, and
turbulent patches are not in an equilibrium state
(32). It is hence the stochastic details of this pro-
cess that decide whether turbulence will either
spread or recede and eventually decay.
Fig. 4. Probability of puff
splitting after traveling a
fixed distance. The five data
sets correspondtodistances
L in the experiments, as in-
dicated inthe legend. Here,
the splitting probability P is
computed as P =r/n, where
r is the number of events
that split and n is the total
number of realizations. The
error bars in the vertical
direction are 95% confi-
dence intervals for the pa-
rameter P of a binomial
distribution as a function
of (n, r) obtained with the
Wilsonmethod(40), whereas
in the horizontal direction
they show the uncertainty
in Re during a set of measurements. The solid lines correspond to the superexponential fit from Fig. 5
without additional fitting parameters.
Fig. 3. Probability dis-
tributions for a puff to
remain in equilibrium. P
is the probability that a
puff will split before time
t. Hence, the plotted quan-
tity 1 − P is the proba-
bility that a puff remains
a single localized puff up
to time t. The numerical
distributions at Re = 2300
and Re = 2350 are ob-
tainedfromall first-splitting
times in ensembles of sim-
ulations by using both the
spectral-element Fourier
code(DNS1) andthehybrid
spectral finite-difference
code (DNS2). Experimental distributions at Re = 2195 and Re = 2255 are obtained from statistics
collected from fixed downstream locations L converted to time by the Re-dependent mean puff
propagation speed U
p
. All distributions are of the form exp[−(t − t
0
)/t], as illustrated by the dashed lines,
where t
0
is a development time for splitting to take place (t
0
≈150 for DNS, whereas for the experiment, t
0
is nearer to 100 and has greater uncertainty). The exponential form of the distributions indicates that
splitting is a memoryless process with characteristic time t.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 194
RESEARCH ARTICLES

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Critical point. To determine the critical point
for the onset of sustained turbulence, we com-
pared the time scale for turbulence to spread with
the time scale for turbulence to decay. The de-
pendence of the mean splitting time on Re is
plotted in Fig. 5. From each experimental data
point in Fig. 4, the mean time t was obtained and
plotted (Fig. 5, right branch, colored symbols),
together with the single superexponential fit
(Fig. 5, solid line). In order to obtain t from the
experimental probabilities, a formation time of t
0
=
100 was considered [uncertainties in t
0
= (50,150)
are included in the error bars]. Because in exper-
iments observation times are generally large, the
uncertainties in t
0
are negligible compared with
statistical errors.
Additionally, t values from simulations are
included in Fig. 5 (black solid triangles), showing
that results from both numerical codes are in
excellent agreement with the experimental data.
From ensemble simulations, we can obtain di-
rectly the times for each split and a maximum
likelihood estimate of t from the memoryless
character of the splitting process (17, 33). The
rapid increase in splitting times makes it in-
feasible to obtain t numerically at low Re.
The left branch in Fig. 5 summarizes pre-
viously measured mean lifetimes for turbulent
decay (15, 17, 18), together with a single super-
exponential fit for mean lifetimes. The intersec-
tion at Re ≈ 2040 marks where the mean lifetime
is equal to the mean splitting time, and to the right
of the intersection, splittings outweigh the decay
of puffs. Analyzing the data in terms of the tur-
bulent fraction results in the same critical point
(fig. S5), confirming the procedure applied here.
Typically in statistical phase transitions, crit-
ical points are not identical to the exact balance
point of two competing processes because of
correlations. For example, in the standard contact
process (34, 35) the spreading (contamination)
rate of an active phase has to outweigh its decay
(recovery) rate by a ratio of about 3 before the
active phase becomes sustained. Although in the
present case the long time scales make it im-
possible to measure other signatures of criticality
such as scale invariance, the superexponential
scaling of the two processes ensures that the crit-
ical point will be almost indistinguishable from
the intersection point. For an increase in Re of 10
(or 0.5%) above the critical point, the splitting
rate already outweighs the decay rate by a factor
of 4. Therefore, the difference between the inter-
section point and the critical point is of the same
order as the experimental uncertainty in Re, and
2040 T 10 provides a close estimate of the critical
point for the onset of sustained turbulence.
Conclusion. The complexity of the transition
process encountered in pipe flow is common to
many shear flows, including Couette, channel,
duct, and boundary layer flows. In all of these
flows, turbulence is found despite the stability of
the base flow and first takes the form of localized
patches, which are transient. The key to the ap-
proach here to determine the onset and sustain-
ment of turbulence has been to separate the
analysis of decay and proliferation mechanisms,
and this approach should be equally applicable
even though details of these mechanisms may dif-
fer from case to case. In all of these flows (analo-
gous to our findings for pipe flow), the spatial
coupling of transiently chaotic domains may give
rise to the sustainment of turbulence (23), breaking
with the classical viewthat turbulence arises through
an increase in temporal complexity (36, 37). The
intermittently alternating laminar and turbulent re-
gions encountered in pipe flow just above criti-
cality are intrinsic to the problem and place pipe
flow in the larger theoretical framework of spatio-
temporal intermittency (19, 38) and nonequilibrium
phase transitions in which universal scaling prop-
erties may be expected (20, 35, 39). Although in
the present study the spatial interaction is relatively
simple because of the clear separation of adjacent
puffs, further above the critical point (Re ≳ 2400)
the dynamics quickly become increasingly com-
plicated, with domains merging and annihilating.
To comprehend this increasing spatial complexity
is a challenge for future studies and is key to our
understanding the onset and nature of turbulence.
References and Notes
1. O. Reynolds, Philos. Trans. R. Soc. London Ser. A 174,
935 (1883).
2. O. Reynolds, Philos. Trans. R. Soc. London Ser. A 186,
123 (1895).
3. A. Darbyshire, T. Mullin, J. Fluid Mech. 289, 83
(1995).
4. H. Faisst, B. Eckhardt, J. Fluid Mech. 504, 343 (2004).
5. B. Eckhardt, Philos. Trans. R. Soc. London Ser. A 367,
449 (2009).
6. B. Eckhardt, Nonlinearity 21, T1 (2008).
7. P. Drazin, W. Reid, Hydrodynamic Stability (Cambridge
Univ. Press, Cambridge, 2004).
8. A. Meseguer, L. Trefethen, J. Comput. Phys. 186, 178
(2003).
9. B. Hof, A. Juel, T. Mullin, Phys. Rev. Lett. 91, 244502
(2003).
10. I. J. Wygnanski, F. H. Champagne, J. Fluid Mech. 59, 281
(1973).
11. B. Eckhardt, T. M. Schneider, B. Hof, J. Westerweel,
Annu. Rev. Fluid Mech. 39, 447 (2007).
12. A. P. Willis, R. R. Kerswell, Phys. Rev. Lett. 98, 014501
(2007).
13. J. Peixinho, T. Mullin, Phys. Rev. Lett. 96, 094501 (2006).
14. B. Hof, J. Westerweel, T. M. Schneider, B. Eckhardt,
Nature 443, 59 (2006).
15. B. Hof, A. de Lozar, D. J. Kuik, J. Westerweel, Phys. Rev.
Lett. 101, 214501 (2008).
16. A. de Lozar, B. Hof, Philos. Trans. R. Soc. London Ser. A
367, 589 (2009).
17. M. Avila, A. Willis, B. Hof, J. Fluid Mech. 646, 127
(2010).
18. D. Kuik, C. Poelma, J. Westerweel, J. Fluid Mech. 645,
529 (2010).
19. K. Kaneko, Prog. Theor. Phys. 74, 1033 (1985).
20. Y. Pomeau, Physica D 23, 3 (1986).
21. S. Bottin, H. Chaté, Eur. Phys. J. B 6, 143 (1998).
22. S. Bottin, F. Daviaud, P. Manneville, O. Dauchot,
Europhys. Lett. 43, 171 (1998).
23. P. Manneville, Phys. Rev. E Stat. Nonlin. Soft Matter Phys.
79, 25301 (2009).
24. D. Moxey, D. Barkley, Proc. Natl. Acad. Sci. U.S.A. 107,
8091 (2010).
25. J. Rotta, Arch. Appl. Mech. 24, 258 (1956).
26. B. Hof, A. de Lozar, M. Avila, X. Tu, T. M. Schneider,
Science 327, 1491 (2010).
27. H. M. Blackburn, S. J. Sherwin, J. Comput. Phys. 197, 759
(2004).
28. A. P. Willis, R. R. Kerswell, J. Fluid Mech. 619, 213 (2009).
29. Materials and methods are available as supporting
material on Science Online.
30. M. Nishi, B. Ünsal, F. Durst, G. Biswas, J. Fluid Mech.
614, 425 (2008).
31. T. M. Schneider, B. Eckhardt, Phys. Rev. E Stat. Nonlin.
Soft Matter Phys. 78, 046310 (2008).
32. I. Wygnanski, M. Sokolov, D. Friedman, J. Fluid Mech. 69,
283 (1975).
33. J. Lawless, Statistical Models and Methods for Lifetime
Data (Wiley, New Jersey, ed. 2, 2003).
34. T. Harris, Ann. Probab. 2, 969 (1974).
35. H. Hinrichsen, Adv. Phys. 49, 815 (2000).
Fig. 5. Mean lifetime of
a puff before decaying or
splitting. Solid colored sym-
bols correspond to experimen-
tal splitting measurements.
Measurements downstream
of jet injection are at dif-
ferent distances L as indi-
cated in Fig. 4. Measurements
downstream of an obsta-
cle perturbation are at L =
1700. The solid black tri-
angles show the character-
istic splitting time obtained
from direct numerical sim-
ulations using the spectral-
element Fourier code (DNS1)
and the hybrid spectral finite-
difference code (DNS2). The
solid curve is given by t = exp[exp(−0.003115 Re + 9.161)] and approximates the Re-dependence of
mean time until a second puff is nucleated and the turbulent fraction increases. This curve was obtained
by fitting the data sets with distances L = 1664, 2100, and 3380, whereas results from shorter pipes
were not fitted because of the stronger influence of the uncertainty in t
0
. The dashed curve is given by t =
exp[exp(0.005556 Re − 8.499)] and approximates the Re-dependence of the mean time until turbulence
decays and the flow relaminarizes. Both curves correspond to superexponential scaling with Re and have a
crossover at Re = 2040 T 10, determining the transition between transient and sustained turbulence in
pipe flow in the thermodynamic limit.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 195
RESEARCH ARTICLES

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

36. L. Landau, E. Lifshitz, Fluid Mechanics, vol. 225
(Pergamom, New York, 1959).
37. D. Ruelle, F. Takens, Commun. Math. Phys. 20, 167 (1971).
38. H. Chaté, P. Manneville, Physica D 32, 409 (1988).
39. J. Rolf, T. Bohr, M. Jensen, Phys. Rev. E Stat. Phys.
Plasmas Fluids Relat. Interdiscip. Topics 57, R2503 (1998).
40. L. D. Brown, T. T. Cai, A. DasGupta, Stat. Sci. 16, 101 (2001).
Acknowledgments: We thank A. P. Willis for sharing his hybrid
spectral finite-difference code. We acknowledge the
Deutsche Forschungsgemeinschaft (project FOR 1182), the
Max Planck Society, and the Engineering and Physical
Sciences Research Council (grant EP/F017413/2) for
financial support. D.B. thanks the Leverhulme Trust and the
Royal Society for their support. M.A. and B.H. acknowledge
computing resources from GWDG (Gesellschaft für
wissenschaftliche Datenverarbeitung Göttingen) and the
Jülich Supercomputing Centre (grant HGU16), where
DNS2 were performed. D.M. and D.B. acknowledge
computing resources from the Centre for Scientific
Computing, University of Warwick, and Grand Equipement
National de Calcul Intensif-Institut du Développement et
des Ressources en Informatique Scientifique (grants
2010-1119 and 2011-1119), where DNS1 were performed.
K.A. acknowledges support from the International Max
Planck Research School for the Physics of Biological and
Complex Systems and the Göttinger Graduate School for
Neurosciences and Molecular Biosciences.
Supporting Online Material
www.sciencemag.org/cgi/content/full/333/6039/192/DC1
Materials and Methods
Figs. S1 to S5
Tables S1 to S3
References
24 January 2011; accepted 19 May 2011
10.1126/science.1203223
REPORTS
Frequency Metrology in Quantum
Degenerate Helium: Direct Measurement
of the 2
3
S
1
—> 2
1
S
0
Transition
R. van Rooij,
1
J. S. Borbely,
1
J. Simonet,
2
M. D. Hoogerland,
3
K. S. E. Eikema,
1
R. A. Rozendaal,
1
W. Vassen
1
*
Precision spectroscopy of simple atomic systems has refined our understanding of the fundamental
laws of quantum physics. In particular, helium spectroscopy has played a crucial role in describing
two-electron interactions, determining the fine-structure constant and extracting the size of the helium
nucleus. Here we present a measurement of the doubly forbidden 1557-nanometer transition connecting
the two metastable states of helium (the lowest energy triplet state 2
3
S
1
and first excited singlet state
2
1
S
0
), for which quantum electrodynamic and nuclear size effects are very strong. This transition is
weaker by 14 orders of magnitude than the most predominantly measured transition in helium. Ultracold,
submicrokelvin, fermionic
3
He and bosonic
4
He atoms are used to obtain a precision of 8 × 10
−12
,
providing a stringent test of two-electron quantumelectrodynamic theory and of nuclear few-body theory.
T
he first observations of helium emission
spectra at the end of the 19th century
revealed two separate series of lines, asso-
ciated with orthohelium and parahelium, respec-
tively. In 1926, Heisenberg explained the distinction
between these two spectra on the basis of wave
mechanics, electron spin, and the Pauli exclusion
principle (1). The spectrum of orthohelium arises
fromtriplet states for which the electron spins are
parallel, whereas in paraheliumthe electron spins
are antiparallel, forming singlet states (Fig. 1).
From the lowest state of orthohelium, the 1s2s
3
S
1
state (denoted 2
3
S
1
), only excitations to triplet
states have been observed. Orthohelium transi-
tions from the 2
3
S
1
state and associated studies
of the n
3
P
0,1,2
(n = 2,3) fine-structure splittings
(2–7) have enabled tests of quantum electrody-
namics (QED) (8, 9), as well as a determination
of the fine-structure constant (5, 10). In the singlet
spectrum of helium (parahelium), electric-dipole
transitions from the 1
1
S
0
ground state (11) and
from the metastable 2
1
S
0
state (12, 13) have also
provided tests of high-precision QED calcula-
tions. All these frequency metrology studies have
been performed using either atomic beams or gas
discharges. However, helium in the metastable
2
3
S
1
state (He*, lifetime 8 × 10
3
s) can be laser-
cooled and trapped, which allows much longer
interaction times for excitation of weak transitions.
He* atoms have been cooled to mKtemperatures,
which revealed quantum statistical effects of
bunching and antibunching (14) and allowed
quantum degeneracy to be achieved for both the
bosonic isotope
4
He (15, 16) and the fermionic
isotope
3
He (17).
Here we observe an orthohelium-parahelium
transition, specifically, the 1557-nm transition
between the metastable 2
3
S
1
and 2
1
S
0
states
(Fig. 1), both in
4
He and
3
He. This transition is an
excellent testing ground for fundamental theory
of atomic structure. Because of a large electron
density at the nucleus, the energy of S states is the
most sensitive to QED and to nuclear size effects
(8). For the 2
3
S
1
and 2
1
S
0
metastable states, QED
terms contribute 4 and 3 GHz respectively, to a
total binding energy of 10
6
GHz (8, 9). The pres-
ent accuracy in the QED calculations is 2 MHz,
based on an estimate of non-evaluated higher-
order terms. Many of these terms are common be-
tween the isotopes. Therefore, in the calculation of
the isotope shift (i.e., the difference between the
transition frequencies for
4
He and
3
He), mass-
independent terms cancel, and the uncertainty is
reduced to the sub-kHz level (18). As the finite
nuclear charge radius shifts the 2
3
S
1
state by
2.6 MHz and the 2
1
S
0
state by 2.0 MHz, an ac-
curate isotope-shift measurement allows a sensi-
tive determination of the difference in the mean
charge radius of the a particle and of the
3
He
nucleus, which provides a stringent test of nuclear
charge radius calculations and experiments (19).
1
LaserLaB Vrije Universiteit, De Boelelaan 1081, 1081 HV Am-
sterdam, Netherlands.
2
École Normale Supérieure, Laboratoire
Kastler-Brossel, 24 rue Lhomond, 75005 Paris, France.
3
De-
partment of Physics, University of Auckland, Private Bag 92019,
Auckland, New Zealand.
*To whom correspondence should be addressed. E-mail:
w.vassen@vu.nl
Fig. 1. Relevant energy lev-
els, transition wavelengths,
and state lifetimes of helium.
The magnetic-dipole transi-
tion connecting the 2
3
S
1
state and the 2
1
S
0
state has
a wavelength of 1557 nm
and an Einstein A coefficient
of 9.1 × 10
−8
s
−1
. A focused
1557-nm laser also consti-
tutes a trap for ultracold at-
2
1
S
0
2
1
P
1
1
1
S
0
parahelium
2x~120 nm
1557 nm
2059 nm
A=9.1 x 10
-8
s
-1 2
3
S
1
2
3
P
0,1,2
1083 nm
orthohelium
=8000 s
=98 ns
=20 ms
oms in the 2
3
S
1
state because it is red detuned fromthe 2
3
S
1
→2
3
P
J
transitions. As the
1557-nmlaser light is blue detuned fromthe 2
1
S
0
→2
1
P
1
transition, atoms in the 2
1
S
0
state are antitrapped.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 196

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

The natural linewidth of the 2
3
S
1
→ 2
1
S
0
transition is 8 Hz, determined by the 20-ms lifetime
of the 2
1
S
0
state, which relaxes via two-photon
decay to the ground state (Fig. 1). This transition is
200,000 times as narrow as the natural linewidth
of the 2
3
P state, which is most prominently used
for spectroscopy in helium. The Einstein A
coefficient for the 2
3
S
1
→2
1
S
0
magnetic-dipole
transition is ~10
−7
s
−1
(20, 21), smaller by 14
orders of magnitude than for the electric-dipole
transitions from 2
3
S
1
to 2
3
P
0,1,2
states, which
indicates that excitation requires high power and/or
long interaction times.
The experiment described here was performed
using an apparatus designed for the production
of quantum degenerate gases of helium (17, 22).
Briefly, the metastable 2
3
S
1
state is populated
by electron impact in an electric discharge. The
atomic beam is collimated, slowed, and trapped
by using standard laser cooling and trapping
techniques on the 2
3
S
1
→ 2
3
P
2
transition at
1083 nm. The atoms, optically pumped to m
J
=
+1, are then transferred to an Ioffe-Pritchard–
type magnetic trap.
4
He* atoms are evaporative-
ly cooled toward Bose-Einstein condensation by
stimulating radio-frequency (RF) transitions to un-
trapped states. For
3
He* (in the F = 3/2 hyperfine
state), quantum degeneracy is reached by sym-
pathetic cooling with
4
He*. Either one or both
of the two isotopes are transferred into a crossed-
beam optical dipole trap. This trap consists of
two focused 1557-nm laser beams, intersecting
at their foci, as shown in Fig. 2. We transfer up
to 10
6
atoms to this optical trap.
After loading the optical trap, the atoms are
illuminated by a separate beam for spectroscopy,
which is derived from the same laser as the op-
tical trap beam, but is switched and frequency-
shifted by a 40-MHz acousto-optic modulator. A
heterodyne signal is set up between the 1557-nm
laser and a mode of a femtosecond frequency–
comb laser to deduce the absolute frequency of
the spectroscopy laser. The frequency comb is
based on a mode-locked erbium-doped fiber laser,
for which both the repetition rate and the carrier-
envelope offset frequency are referenced to a
global positioning system–controlled Rubidium
clock (23).
After a certain interaction time (typically 1 to
6 s), both the spectroscopy beam and the trap
beamare switched off, which allows the atoms to
fall because of gravity. The high internal energy
of He* (20 eVabove the 1
1
S
0
ground state) allows
for efficient detection on a microchannel plate
(MCP) detector (Fig. 2). The MCP signal reflects
both the number of atoms and their temperature.
In the case of
4
He, the signal has a bimodal
character that results from the combination of
Bose-condensed atoms and thermal atoms (Fig.
3A); a fit to this signal provides the number of
condensed atoms (23). Because the excited state
is antitrapped, the trap is depleted when the spec-
troscopy beam is resonant with the atomic tran-
sition. By deducing the remaining number of 2
3
S
1
atoms for various laser frequencies, the atomic res-
onance frequency is determined from a Gaussian
fit to the data (Fig. 3B). The observed linewidth
is largely due to the 75-kHz laser linewidth.
Several systematic shifts in the transition
frequency are taken into account (23). The largest
shift is due to the Zeeman effect. The measured
transition, 2
3
S
1
(m
J
= +1) →2
1
S
0
(m
J
= 0) for
4
He, and 2
3
S
1
, F = 3/2 (m
F
= +3/2) →2
1
S
0
, F =
1/2 (m
F
= +1/2) for
3
He, is shifted from res-
onance predominantly by Earth’s magnetic field.
The size of the shift is deduced by measuring the
resonance frequency of RF spin-flip transitions
between the 2
3
S
1
magnetic substates. An ad-
ditional shift is caused by the momentum transfer
from a 1557-nm photon to an atom. In the case
of
4
He, the high density of the condensate could
potentially cause a mean-field shift (24). How-
ever, by performing the experiment with reduced
atomic density, no shift is observed.
The second-largest systematic frequency per-
turbation is due to the AC Stark shift associated
with the intense 1557-nm light that induces the
dipole trap: The specific energy state of the trap-
ping potential for an atom determines the AC
Stark shift for that atom. For
4
He, only excita-
tions of atoms condensed in the ground state of
the dipole trap are taken into account in deter-
mining the transition frequency. As the trap depth
depends linearly on laser intensity, measuring the
resonance frequency for a range of applied laser
powers allows an extrapolation to zero laser in-
tensity. In contrast,
3
He atoms, because of their
fermionic nature, are distributed throughout the
energy states of the dipole trap, and as a result,
Fig. 2. Experimental setup. A small fraction of the 1557-nmlaser light is split off and coupled via a fiber-optic
link to be referenced to a fiber-based frequency comb. A heterodyne signal is monitored on a fast photodiode
(PD) to determine the absolute frequency of the 1557-nm laser. The remaining light is divided into the trap
beam and the spectroscopy beam. A crossed-beam dipole trap configuration is realized by focusing both the
incident and returning trap beam (with orthogonal linear polarizations) to a waist of ~85 mm at the center of
the magnetic trap (represented by the green coils) under a relative angle of 19 degrees, trapping atoms at the
intersection. The spectroscopy beam is frequency shifted by a 40-MHz acousto-optical modulator (AOM),
overlapped with the returning trap beam and absorbed by a thermopile power meter (PM). A microchannel
plate detector is positioned underneath the trap for temperature and atom number determination.
170 175 180 185 190 195 200
Time (ms)
0
20
40
60
80
100
H
e
*

u
x
(
a
r
b
.
u
n
i
t
s
)
Thermal atoms
Condensed atoms
A B
-150 -100 -50 0 50 100 150
f - f0 (kHz)
40
60
80
100
R
e
m
a
i
n
i
n
g
a
t
o
m
s
(
%
)
90 kHz
Fig. 3. (A) Bimodal time-of-flight distribution observed when He* atoms are detected on the MCP detector
~186 ms after the trapping laser light is turned off. The MCP signal is fit to determine the number of Bose-
condensed atoms. (B) The percentage of Bose-condensed
4
He atoms remaining in the 2
3
S
1
state as a function
of applied laser frequency (relative to the fitted center frequency f
0
). The line is a fit of a Gaussian to the data.
We measure linewidths varying from 75 to 130 kHz depending on the trap depth and on the isotope.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 197
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

the measured AC Stark shift does not equal the
trap depth (as is the case with
4
He) but is reduced
due to the density of states within the dipole trap.
Anonlinear shift can then potentially arise at high
laser intensities, where the larger trap depths al-
low for higher temperatures. To minimize this ef-
fect,
3
He atoms are sympathetically cooled to the
quantum degenerate regime to populate predom-
inantly the lowest energy states of the trapping
potential. Over the course of several months, 20
independent extrapolations were obtained (as
shown in Fig. 4) to deduce an absolute frequency
of the 2
3
S
1
→ 2
1
S
0
transition for
4
He of f
4
=
192,510,702,145.6(1.8) kHz and for
3
He (F =
3/2 → F = 1/2) of f
3
= 192,504,914,426.4(1.5)
kHz, where the one–standard deviation error in
parentheses includes all statistical and systematic
uncertainties.
For both isotopes, our result agrees with QED
calculations of the ionization energies of the two
metastable states (9, 25). The present experimen-
tal error in the transition frequency is smaller by
three orders of magnitude than estimates of non-
evaluated higher-order terms in state-of-the-art QED
calculations and presents a significant challenge
for groups involved in atomic structure theory.
An indirect value of the energy difference
between the 2
3
S
1
and the 2
1
S
0
states can be
obtained from the literature (only for
4
He) by
combining experimental transition frequencies from
both metastable states to high-lying S, P, and D
states with theoretical values for the ionization
energies of these states. This procedure yields
ionization energies for the 2
1
S
0
state (8, 12, 13)
and the 2
3
S
1
state (2, 8), and the difference be-
tween these values gives a transition frequency of
192,510,701.96(16) MHz, in agreement with our
result, although with much lower precision.
Isotope shift measurements, combined with
high-precision QEDtheory, provide a method for
isolating contributions due to finite nuclear size
effects. The difference in nuclear charge radii be-
tween
3
He and
4
He is determined by comparing
experiment and theory. The
4
He nuclear charge
radius is one of the most precisely known of all
nuclei (26), 1.681(4) fm. Avalue of the
3
He nu-
clear charge radius can then be deduced with
similar precision. In calculating the isotope shift,
QED theory is more precise than our measure-
ment, as mass-independent terms cancel. The
theoretical value for the isotope shift (if one as-
sumes pointlike nuclei) is 8,034,148.6(7) kHz (23).
Subtracting the measured transition frequencies
and correcting for the accurately known hyper-
fine structure ( f
hfs
) (25, 27), we find an isotope
shift of f
4
– f
3
+ f
hfs
= 8,034,367.2(2.3) kHz. The
218.6-kHz difference may be attributed to the
finite size of both nuclei. This nuclear shift is pro-
portional to the difference in the nuclear charge
radii squared, ∆r
c
2
≡ r
c
2
(
3
He) − r
c
2
(
4
He). Using
the theoretical proportionality constant of 4.6642
fm
2
/ MHz (18) for the measured transition, we
deduce ∆r
c
2
= 1.019(11) fm
2
. ∆r
c
2
represents a
more universal parameter than the value of the
isotope shift, as it is obtained fromvarious branches
of physics. Besides getting it through spectroscop-
ic means, it can be determined fromnuclear theory
and from electron-scattering experiments. Nuclear
few-body theory provides ∆r
c
2
= 1.16 (12) fm
2
(18, 23, 28), whereas from electron-scattering
experiments ∆r
c
2
= 1.01 (13) fm
2
(26, 29). Com-
paring the values of ∆r
c
2
, we find our result to
be in good agreement but more precise by an or-
der of magnitude. An independent spectroscopic
measurement in helium on the 2
3
S
1
→ 2
3
P
0
transition (30) gives ∆r
c
2
= 1.059(3) fm
2
, ob-
tained by using the most recent QEDcalculations
(19). Although the measurement precision of
the isotope shift for this transition is comparable
to our precision, the smaller uncertainty in ∆r
c
2
is
due to a larger sensitivity to differential nuclear
charge effects. At present, the accuracy to which
the
4
He charge radius is known sets a lower limit
on the uncertainty of the
3
He charge radius deter-
mined from helium spectroscopy. Our measure-
ment presents a value for the
3
He nuclear charge
radius of 1.961(4) fm.
We have also demonstrated that all of the
trapped atoms can be transferred to the 2
1
S
0
state
to produce a source of ultracold singlet helium.
Optically trapping these atoms simultaneously
with cold 1
1
S
0
ground-state atoms (produced
after two-photon decay) opens up the possibility
of performing two-photon spectroscopy on the
2
1
S
0
↔1
1
S
0
transition (11, 31), where QEDand
nuclear size effects are strongest.
References and Notes
1. W. Heisenberg, Z. Phys. 39, 499 (1926).
2. C. Dorrer, F. Nez, B. de Beauvoir, L. Julien, F. Biraben,
Phys. Rev. Lett. 78, 3658 (1997).
3. P. C. Pastor et al., Phys. Rev. Lett. 92, 023001
(2004).
4. P. Mueller et al., Phys. Rev. Lett. 94, 133001 (2005).
5. M. Smiciklas, D. Shiner, Phys. Rev. Lett. 105, 123001
(2010).
6. J. S. Borbely et al., Phys. Rev. A 79, 060503 (2009).
7. T. Zelevinsky, D. Farkas, G. Gabrielse, Phys. Rev. Lett. 95,
203001 (2005).
8. G. W. F. Drake, Z.-C. Yan, Can. J. Phys. 86, 45 (2008).
9. V. A. Yerokhin, K. Pachucki, Phys. Rev. A 81, 022507
(2010).
10. K. Pachucki, V. A. Yerokhin, Phys. Rev. Lett. 104, 070403
(2010).
11. D. Z. Kandula, C. Gohle, T. J. Pinkert, W. Ubachs,
K. S. Eikema, Phys. Rev. Lett. 105, 063001 (2010).
12. C. J. Sansonetti, J. D. Gillaspy, Phys. Rev. A 45, R1
(1992).
13. W. Lichten, D. Shiner, Z.-X. Zhou, Phys. Rev. A 43, 1663
(1991).
14. T. Jeltes et al., Nature 445, 402 (2007).
15. A. Robert et al., Science 292, 461 (2001).
16. F. Pereira Dos Santos et al., Phys. Rev. Lett. 86, 3459
(2001).
17. J. M. McNamara, T. Jeltes, A. S. Tychkov, W. Hogervorst,
W. Vassen, Phys. Rev. Lett. 97, 080404 (2006).
18. G. W. F. Drake, W. N. Nörtershäuser, Z.-C. Yan,
Can. J. Phys. 83, 311 (2005).
19. D. C. Morton, Q. Wu, G. W. F. Drake, Phys. Rev. A 73,
034502 (2006).
20. G. Łach, K. Pachucki, Phys. Rev. A 64, 042510 (2001).
21. K. A. H. van Leeuwen, W. Vassen, Europhys. Lett. 76, 409
(2006).
22. A. S. Tychkov et al., Phys. Rev. A 73, 031603 (2006).
23. Methods and calculations are further detailed in
supporting material at Science Online.
24. T. C. Killian et al., Phys. Rev. Lett. 81, 3807 (1998).
25. D. C. Morton, Q. Wu, G. W. F. Drake, Can. J. Phys. 84, 83
(2006).
26. I. Sick, Phys. Rev. C Nucl. Phys. 77, 041302 (2008).
27. S. D. Rosner, F. M. Pipkin, Phys. Rev. A 1, 571 (1970).
28. A. Kievsky, S. Rosati, M. Viviani, L. E. Marcucci,
L. Girlanda, J. Phys. G 35, 063101 (2008).
29. I. Sick, Lect. Notes Phys. 745, 57 (2008).
30. D. Shiner, R. Dixson, V. Vedantham V, Phys. Rev. Lett.
74, 3553 (1995).
31. E. Eyler et al., Eur. Phys. J. D 48, 43 (2008).
Acknowledgments: This work, as part of the European
Science Foundation EuroQUAM Programme, was
financially supported by the Dutch Foundation for
Fundamental Research on Matter (FOM). J.S.
acknowledges financial support from the EC’s Seventh
Framework Programme (LASERLAB-EUROPE). M.D.H. and
K.S.E.E acknowledge financial support from the
Netherlands Organization for Scientific Research (NWO).
We would like to thank J. Bouma for technical support,
and J. C. J. Koelemeij and K. A. H. van Leeuwen for
fruitful discussions. We also thank G. W. F. Drake,
K. Pachucki, and V. A. Yerokhin for sharing the
calculated theoretical parameters detailed in SOM.
Supporting Online Material
www.sciencemag.org/cgi/content/full/333/6039/196/DC1
Materials and Methods
SOM Text
Figs. S1 and S2
Table S1
Reference (32)
4 March 2011; accepted 16 May 2011
10.1126/science.1205163
1 2 3 4 5 6 7 8 9 10
Measurement number
-15
-10
-5
0
5
10
15
f
4
-
1
9
2
5
1
0
7
0
2
1
4
5
.
6
(
k
H
z
)
A
1 2 3 4 5 6 7 8 9 10
Measurement number
-15
-10
-5
0
5
10
15
f
3
-
1
9
2
5
0
4
9
1
4
4
2
6
.
4
(
k
H
z
)
B
Fig. 4. Measured transition frequencies for
4
He (A) and for
3
He (B). The error bar on each data point
includes contributions from various systematic sources, e.g., AC Stark shift and Zeeman shift (23). The
frequencies quoted on the y axes are the weighted averages of the data points. Their associated
uncertainties (with the uncertainties due to the frequency comb calibration and the mean-field shift added
in quadrature) are T1.8 kHz for
4
He and T1.5 kHz for
3
He, represented by the green bands.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 198
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

An Extremely Luminous Panchromatic
Outburst from the Nucleus of a
Distant Galaxy
A. J. Levan,
1
* N. R. Tanvir,
2
S. B. Cenko,
3
D. A. Perley,
3
K. Wiersema,
2
J. S. Bloom,
3
A. S. Fruchter,
4
A. de Ugarte Postigo,
5
P. T. O’Brien,
2
N. Butler,
3
A. J. van der Horst,
6
G. Leloudas,
5
A. N. Morgan,
3
K. Misra,
4
G. C. Bower,
3
J. Farihi,
2
R. L. Tunnicliffe,
1
M. Modjaz,
7
J. M. Silverman,
3
J. Hjorth,
5
C. Thöne,
8
A. Cucchiara,
3
J. M. Castro Cerón,
9
A. J. Castro-Tirado,
8
J. A. Arnold,
10
M. Bremer,
11
J. P. Brodie,
10
T. Carroll,
12
M. C. Cooper,
13
P. A. Curran,
14
R. M. Cutri,
15
J. Ehle,
12
D. Forbes,
16
J. Fynbo,
5
J. Gorosabel,
8
J. Graham,
4,17
D. I. Hoffman,
15
S. Guziy,
8
P. Jakobsson,
19
A. Kamble,
20
T. Kerr,
12
M. M. Kasliwal,
18
C. Kouveliotou,
21
D. Kocevski,
10
N. M. Law,
22
P. E. Nugent,
3,23
E. O. Ofek,
18
D. Poznanski,
3,23
R. M. Quimby,
18
E. Rol,
24
A. J. Romanowsky,
10
R. Sánchez-Ramírez,
8
S. Schulze,
19
N. Singh,
10,25
L. van Spaandonk,
1,26
R. L. C. Starling,
2
R. G. Strom,
24,27
J. C. Tello,
8
O. Vaduvescu,
28
P. J. Wheatley,
1
R. A. M. J. Wijers,
24
J. M. Winters,
11
D. Xu
29
Variable x-ray and g-ray emission is characteristic of the most extreme physical processes in
the universe. We present multiwavelength observations of a unique g-ray–selected transient
detected by the Swift satellite, accompanied by bright emission across the electromagnetic
spectrum, and whose properties are unlike any previously observed source. We pinpoint the
event to the center of a small, star-forming galaxy at redshift z = 0.3534. Its high-energy
emission has lasted much longer than any g-ray burst, whereas its peak luminosity was
∼100 times higher than bright active galactic nuclei. The association of the outburst with the
center of its host galaxy suggests that this phenomenon has its origin in a rare mechanism
involving the massive black hole in the nucleus of that galaxy.
S
urveys of the sky at short wavelengths
(x-ray and g-ray) reveal a much more dy-
namic universe than is seen in the optical
wavelengths. Many sources vary substantially;
the most extreme can go from invisibility to
being the brightest objects in the sky, some-
times on time scales of seconds. The sources of
such bursts of high-energy radiation have prov-
en difficult to trace, but dedicated observation-
al programs have shown that some fraction
originate in the Milky Way, either fromisolated
neutron stars with intense magnetic fields (1) or
from binary systems containing neutron stars
and black holes (2). Some long-lived but variable
x-ray and g-ray emissions originate in active gal-
axies (3), whereas the brightest are the long-
duration g-ray bursts (long-GRBs), which are
detected at a rate of approximately two per week
by current missions such as the Swift satellite (4)
and are now thought to originate from the col-
lapse of massive stars in the distant universe (5, 6).
GRB110328A/Swift J164449.3+573451 (here-
after Sw 1644+57) was detected with the Swift
Burst Alert Telescope (BAT) at 12:57:45 UT on
28 March 2011 (7). It required an unusually long
integration, in excess of 1000 s, to trigger the in-
strument because of its slowvariability time scale.
Follow-up observations with the Ultraviolet and
Optical Telescope (UVOT) and X-Ray Telescope
(XRT) onboard the Swift satellite began 1475 s
after the initial outburst. No source was seen in
the UVOTobservations, but a bright point source
was found with the XRT (7). Unlike any previ-
ously observed long-GRBs (which typically de-
cline substantially on a time scale of minutes),
Sw 1644+57 remained bright and highly varia-
ble for a prolonged period and went on to re-
trigger the BATon three further occasions over
the next 48 hours (8). Reexamination of previ-
ous g-ray observations of this region showed that
the source appears to have been present a few
days before the initial trigger but not at earlier
times (9). Equally unlike any normal long-GRB,
the source remained bright in the x-rays for more
than 2 weeks (Fig. 1). The early x-ray behavior
showed the same dramatic flaring seen with BAT,
with flares having time scales of hours and with
broadly similar shapes. After the first 48 hours,
the x-rays maintained a more constant level, albeit
with episodic brightening and fading spanning
more than an order of magnitude in flux.
Our first ground-based observations of Sw
1644+57 began approximately 2 hours after the
burst trigger, with the Gemini-North Telescope in
Hawaii. Unfortunately, poor weather conditions
meant that only shallow observations were pos-
sible, and these did not yield any candidate op-
tical counterpart to a limit of r ∼ 22.1 magnitude.
At 13 hours after the trigger, we obtained im-
aging with the Nordic Optical Telescope (NOT)
on La Palma, which revealed a R ≈ 22.5 mag-
nitude source that was consistent with the x-ray
position (10). Examination of archival images
obtained with the Palomar Transient Factory
(PTF) revealed this source to be present at ap-
proximately the same brightness more than a
year before the outburst. Our subsequent optical
monitoring (below) confirms that the optical
flux is dominated by the host galaxy. Early anal-
ysis of the x-ray/g-ray data was used to argue
that the transient was most likely a source within
the Milky Way (11). However, our spectroscopy
of the optical counterpart with Gemini-North
(12), the Gran Telescopio Canarias (GTC) in
La Palma (13), and the Keck Telescope in Hawaii
(14) [supplementary online material (SOM) text]
showed strong emission lines of hydrogen and
oxygen (as well as absorption lines from a mod-
erate age stellar population), which is consistent
with a star-forming galaxy at a systemic redshift
of z = 0.3534 T 0.0002 (Fig. 2). Thus, Sw 1644+57
is a source at cosmological distance with extreme-
ly unusual properties.
We continued to monitor the field from the
ground in the optical and near-infrared (near-IR)
with Gemini-North, the UK Infrared Telescope
(UKIRT), the NOT, the William Herschel Tele-
scope (WHT), PAIRITEL, the Telescopio Nazionale
Galileo (TNG), and GTC, obtaining observations
1
Department of Physics, University of Warwick, Coventry CV4
7AL, UK.
2
Department of Physics and Astronomy, University
of Leicester, Leicester LE1 7RH, UK.
3
Department of Astron-
omy, University of California, Berkeley, CA 94720–3411, USA.
4
Space Telescope Science Institute, 3700 San Martin Drive,
Baltimore, MD 21218, USA.
5
Dark Cosmology Centre, Niels
Bohr Institute, University of Copenhagen, 2100 Copenhagen,
Denmark.
6
Universities Space Research Association, National
Space Science and Technology Center, 320 Sparkman Drive,
Huntsville, AL 35805, USA.
7
Columbia Astrophysics Laboratory,
Columbia University, New York, NY 10024, USA.
8
Instituto de
Astrofísica de Andalucía–Consejo Superior de Investigaciones
Científicas (IAA-CSIC), Glorieta de la Astronomía s/n, E-18008
Granada, Spain.
9
Herschel Science Operations Centre, European
Space Astronomy Centre, European Space Agency (ESA), Post
Office Box 78, 28691 Villanueva de la Caada, Madrid, Spain.
10
University of California Observatories/Lick Observatory, Uni-
versity of California, Santa Cruz, 1156 High Street, Santa Cruz,
CA 95064, USA.
11
Institut de RadioAstronomie Millimétrique,
300 rue de la Piscine, Domaine Universitaire, 38406 Saint
Martin d’Hères, France.
12
Joint Astronomy center, 660 North
A’ohoku Place, University Park, Hilo, HI 96720, USA.
13
Center for
Galaxy Evolution, University of California, Irvine, 4129 Frederick
Reines Hall, Irvine, CA 92697, USA.
14
Astrophysique Interactions
Multi-échelles, Commissariat à l’Énergie Atomique/Direction des
Sciences de la Matière–CNRS, Irfu/Service d’Astrophysique,
Centre de Saclay, Bâtiment 709, FR-91191 Gif-sur-Yvette Cedex,
France.
15
Infrared Processing and Analysis Center, California
Institute of Technology, Pasadena, CA 91125, USA.
16
Centre for
Astrophysics andSupercomputing, SwinburneUniversity, Hawthorn
VIC 3122 Australia.
17
Department of Physics and Astronomy, Johns
Hopkins University, Baltimore, MD21218, USA.
18
Cahill Center for
Astrophysics, California Institute of Technology, Pasadena, CA
91125, USA.
19
Centre for Astrophysics and Cosmology, Science
Institute, University of Iceland, Dunhaga 5 IS-107 Reykjavik,
Iceland.
20
Center for Gravitation and Cosmology, University of
Wisconsin-Milwaukee, 1900 East Kenwood Boulevard, Milwaukee,
WI 53211, USA.
21
Space Science Office, VP62, NASA/Marshall
Space Flight Center Huntsville, AL 35812, USA.
22
Dunlap Institute
for Astronomy and Astrophysics, University of Toronto, Toronto,
M5S 3H4 Ontario, Canada.
23
Computational Cosmology Center,
Lawrence Berkeley National Laboratory, 1 Cyclotron Road,
Berkeley, CA 94720, USA.
24
Astronomical Institute, University of
Amsterdam, Science Park 904, 1098 XH Amsterdam, Nether-
lands.
25
Centre for Astronomy, National University of Ireland,
Galway, Ireland.
26
Centre for Astrophysics Research, Science and
Technology Research Institute, University of Hertfordshire,
Hatfield AL10 9AB, UK.
27
Netherlands Institute for Radio As-
tronomy (ASTRON), Postbus 2, 7990 AA Dwingeloo, Netherlands.
28
Isaac Newton Group of Telescopes, Apartado de correos 321
E-38700, Santa Cruz de la Palma, Canary Islands, Spain.
29
Benoziyo
Center for Astrophysics, Faculty of Physics, Weizmann Institute
of Science, Rehovot, 76100, Israel.
*To whom correspondence should be addressed. E-mail:
a.j.levan@warwick.ac.uk
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 199
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

from the B band (435 nm) to the L band (3780
nm). In contrast to the non-varying behavior in
the optical, these data showed that at near-IR
wavelengths the source fluctuated by more than
a factor of 3 in flux over several days, indicating
that the g-ray transient was also producing con-
siderable longer-wavelength emission. Our detec-
tion in the L band [270 T 50 microjansky (mJy)],
compared with quiescent limits from the WISE
satellite, implies that the transient is at least an
order of magnitude brighter than its host galaxy
at these wavelengths. The IR variations roughly
track those of the x-ray (Fig. 1) but are certainly
not perfectly correlated, suggesting multiple emis-
sion components.
We obtained an observation with the Chandra
X-ray Observatory, which took place about 6.5
days after the initial outburst (Fig. 1). This showed
that the x-ray emission continued to exhibit a
factor of ∼2 changes in flux on time scales as
short as ∼100 s even at this comparatively late
time after the early flaring. However, our pho-
tometry of individual optical and near-IR im-
ages (with a time resolution of 20 to 60 s) does
not reveal rapid variability in the near-IR light.
In the optical r band, little variability was seen
(<10%) on all time scales, indicating that the host
galaxy dominates the optical emission. The tran-
Fig. 1. The x-ray, IR, and radio lightcurves of Sw
1644+57. (Top) The XRT (0.3 to 10 keV; red) and
BAT (15 to 50 keV; black) flux against observed
time since the initial outburst trigger time; the
right hand axis indicates the luminosity of the
event. (Inset) The dense sampling of our Chandra
observation. The dashed blue vertical lines indi-
cate the times of subsequent triggers of the BAT.
(Middle) Our near-IR lightcurve of this event (host
flux not subtracted). (Bottom) Our 4.8- and 1.4-GHz
lightcurve’s obtainedfromthe WSRT showarisingradio
flux. The left-hand panels represent pre-outburst
observations of the location of Sw 1644+57 and the
limits on transient emission at this time (24), in the
x-ray (ROSAT, 1991), IR (WISE, Jan 2010), and radio
(VLA FIRST, 1998). They clearly demonstrate the
large amplitude of this outburst in the x-ray and IR.
Fig. 2. Spectrum of the host galaxy of Sw 1644+57, obtained at the GTC. (Left inset) the Hb line as seen
in the first Gemini Multi Object Spectograph spectrum. Prominent stellar atmosphere absorption is visible.
(Right inset) The first Gemini spectrum (red), the second Gemini spectrum (blue), the Keck spectrum
(purple), and the GTC spectrum (black) covering the Hb and [O III] doublet, all rebinned to the lower
resolution of the GTC spectra. No emission line variability is apparent over the 3-day span of these
observations.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 200
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

sient has a very red optical near-IR color, prob-
ably because of a high dust column along the line
of sight. The dust hypothesis would be consistent
with the high hydrogen column density inferred
from the x-ray spectrum (10
22
cm
−2
), which im-
plies host extinction of A
V
∼ 6 (15), reducing the
optical luminosity by a factor of ≳100. Together,
these findings suggest that the source is situated
in a dense and dusty region, such as a galactic
nucleus (SOM text).
Observations at still longer wavelengths
showed a bright radio (16) and millimeter source
(17) at the same location. Our millimeter obser-
vations from the Institut de Radioastronomie
Millimétrique (IRAM) confirm this, and radio
(1.4. and 4.8 GHz) observations fromWesterbork
Synthesis Radio Telescope (WSRT) show a
bright source, which in contrast to the optical
and IR light brightened over the first week after
the outburst (Fig. 1, bottom). These observations
demonstrate that Sw 1644+57 emitted strong
radiation across the electromagnetic spectrum,
whereas the differing behavior in each waveband
may be due to either strong spectral evolution or
distinct emission components.
The character of the host galaxy and the po-
sition of the transient within it are potentially im-
portant clues to the nature of Sw 1644+57, and
to this end, we obtained observations with the
Hubble Space Telescope (HST) on 4 April and
20 April 2011. In the near-IR, the image remains
unresolved, fading between the two epochs of
observation, and consistent with emission from
the transient still dominating. In the optical wave-
bands, we clearly detected the light of the host
galaxy. The Wide Field Camera 3 (WFC3) IR
position of the transient falls within 0.03 arc sec
(1 s, < 150 pc at z = 0.3534) of the center of
the host galaxy (Fig. 3). We also obtained Very
Long Baseline Array (VLBA) observations of
Sw 1644+57 on 1 April 2011. These provided
another precise astrometric position, with an
offset from the center of the host of 0.04 T 0.07
arc sec, further strengthening the association with
the nucleus of the host (18).
The host galaxy itself appears compact and
noninteracting, with a half-light radius in the op-
tical of r
h
= 1.04 kpc and an absolute magnitude
of M
V
= −18.19 (comparable in luminosity with
the Large Magellanic Cloud). Subtraction of a
point source from the HST F606W image sug-
gests an upper limit to the transient magnitude in
that band of 30%of the host light, or a magnitude
of 24.1 (AB). The measured ratios of emission
lines are consistent with an origin in a normal star-
forming galaxy that has not, at least until now, con-
tained anactive nucleus. The inferred star formation
rate of the host is 0.5 solar mass (M

) yr
−1
.
Our observations clearly show that the tran-
sient originates from the center of a galaxy at cos-
mological distances. At this redshift, the brightest
x-ray flare reached a luminosity of L
X
∼ 3 ×
10
48
erg s
−1
(isotropic equivalent) for ∼1000 s.
The total energy output in the first ∼10
6
s after
the outburst of ∼10
53
erg is equivalent to ∼10%of
Fig. 3. Discovery images of Sw 1644+57-and its host galaxy. (Top) Our ground-based imaging in the
optical r band (top left) and IR K-band (top right). The images are oriented north, up; east, left and are
approximately 1 arc min in height. The location of Sw 1644+57 is indicated with arrows. (Bottom)
Zoomed-in regions of our later time observations with HST in the IR (bottom right) and optical (bottom
left). The crosshair indicates the optically derived centroid of the host galaxy. The red circle shows the
location of the IR source inferred from our first epoch WFC IR observations, whereas the larger green
circle shows the offset (due to the systematic uncertainty in tying coordinate frames) from our very long
baseline interferometry position.
Fig. 4. The x-ray lumi-
nosity and optical/near-
IR absolute magnitude
of Sw 1644+57, at peak
(bold stars) and at 10
6
s
after the trigger. For com-
parison, we show the
properties of the most
luminous quasars and
blazars (3C 279 and Mrk
421); a sample of all ob-
jects within the 2XMM
survey with high confi-
dence (>2 s) association
with objects in the Sloan
Digital Sky Survey of
knownredshift z <3(25);
and a sample of more
local galaxies [from(26);
optical magnitudes in-
clude contribution from
the host galaxy]. We
also plot the late-time
luminosities of a sam-
ple of bright GRB af-
terglows [extrapolated
from (27, 28)], which
are relevant because
Sw 1644+57 stays within an order of magnitude of its brightest peak, even 10
6
s after the outburst
began. We also plot the location of the candidate tidal disruption event in RXJ 1242-119 (29).
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 201
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

the rest energy of the Sun. Although these num-
bers are not abnormal for long-GRBs, the proper-
ties of this outburst are clearly distinct from the
long-GRB population. First, the repetition of the
g-ray trigger four times in 48 hours is unheard
of for long-GRBs, which are destructive and
non-repeating events. Further, the duration of
bright x-ray emission is much longer than has
ever been seen for any long-GRB (19, 20), per-
sisting at L
X
∼ 10
47
erg s
−1
2 weeks after the
initial event. This, together with the origin in the
core of its host galaxy, implies that Sw 1644+57
most likely originates from the central massive
black hole. However, the x-ray luminosity of
Sw 1644+57 is well beyond the bright end of the
quasar luminosity function (21) and is more lu-
minous (by a factor of ∼100) than flares from the
brightest blazars (3). However, its optical lu-
minosity is a factor of ∼10
4
fainter than a bright
quasar (22), implying either different emission
processes or (as seems to be the case, owing to
the red color) a particularly high dust column
within the host. The overall energetics and long
duration, together with the order-of-magnitude
variations in flux over 100 s time scales, make it
clear that we are observing an unprecedented
astrophysical object (Fig. 4).
The peak luminosity corresponds to the
Eddington luminosity of a ∼10
10
M

black hole.
It is highly unlikely that a moderate-sized galaxy
such as the host of Sw 1644+57 could contain
such a massive black hole; our spectral energy
distribution fitting of the host galaxy implies
that its total stellar mass is less than this value
(SOM text and fig. S6), and for a typical stellar
mass–black hole mass relation (23), its black
hole mass is unlikely to be greater than ∼10
7
M

.
Hence, Sw1644+57 is either accreting at a super-
Eddington rate or has its total energy modified by
relativistic beaming (or both). Bloom et al. (24)
consider the possibility that the source of this
event is the tidal disruption of a star around the
central black hole.
The detection of a different class of extremely
energetic g-ray transient after many years of in-
tensive monitoring of the g-ray sky highlights the
rarity of this phenomenon. Although the bright
flares in Sw 1644+57 are of longer duration
than typical GRBs, it is likely that Swift would
have detected a similar event to at least z ∼ 0.7. If
we assume that galaxies within 1 magnitude of the
Sw 1644+57 host will typically contain similar-
mass black holes at their centers, then we can
estimate a space density of ∼5 × 10
7
Gpc
−3
po-
tential hosts. Thus, a single example in 6.5 years
of Swift operations would correspond to a rate
per galaxy of 1 in 3 × f
beam
gigayears, where we
allow for the possibility that the radiation is
beamed into a fraction f
beam
of a sphere.
Although Sw 1644+57 was detected through
its g-ray emission, it is its behavior at x-ray and
IR wavelengths, lasting at a bright flux level for
days to weeks, that most strikingly demonstrates
its difference from known classes of high-energy
transient. This raises the possibility that similar
events that are rather less variable or less lumi-
nous could be occurring but have so far evaded
detection with existing satellites.
References and Notes
1. C. Kouveliotou et al., Nature 393, 235 (1998).
2. F. Aharonian et al., Science 309, 746 (2005).
3. I. Donnarumma et al., Astrophys. J. 691, L13 (2009).
4. N. Gehrels, E. Ramirez-Ruiz, D. B. Fox, Annu. Rev. Astron.
Astrophys. 47, 567 (2009).
5. J. Hjorth et al., Nature 423, 847 (2003).
6. S. E. Woosley, J. S. Bloom, Annu. Rev. Astron. Astrophys.
44, 507 (2006).
7. J. C. Cummings et al., GRB Coord. Network 11823,
1 (2011).
8. T. Sakamoto et al., GRB Coord. Network 11842,
1 (2011).
9. H. Krimm, S. Barthelmy, GRB Coord. Network 11891,
1 (2011).
10. G. Leloudas et al., GRB Coord. Network 11830, 1 (2011).
11. S. D. Barthelmy et al., GRB Coord. Network 11824,
1 (2011).
12. A. Levan, N. Tanvir, K. Wiersema, D. Perley,
GRB Coord. Network 11833, 1 (2011).
13. C. Thoene et al., GRB Coord. Network 11834, 1 (2011).
14. S. B. Cenko et al., GRB Coord. Network 11874, 1 (2011).
15. P. Predehl, J. H. M. M. Schmitt, Astron. Astrophys. 293,
889 (1995).
16. A. Zauderer, E. Berger, D. A. Frail, A. Soderberg,
GRB Coord. Network 11836, 1 (2011).
17. A. Zauderer et al., GRB Coord. Network 11841,
1 (2011).
18. A. Brunthaler et al., GRB Coord. Network 11911,
1 (2011).
19. C. Kouveliotou et al., Astrophys. J. 608, 872 (2004).
20. J. A. Nousek et al., Astrophys. J. 642, 389 (2006).
21. Y. Ueda, M. Akiyama, K. Ohta, T. Miyaji, Astrophys. J.
598, 886 (2003).
22. D. W. Just et al., Astrophys. J. 665, 1004 (2007).
23. V. N. Bennert, M. W. Auger, T. Treu, J.-H. Woo,
M. A. Malkan, Astrophys. J. 726, 59 (2011).
24. J. Bloom et al., Science 333, XXXX (2011).
25. F. Pineau et al., Astron. Astrophys. 527, A126 (2011).
26. L. C. Ho, Astrophys. J. 699, 626 (2009).
27. D. A. Kann et al., Astrophys. J. 720, 1513 (2010).
28. P. A. Evans et al., Mon. Not. R. Astron. Soc. 397, 1177
(2009).
29. S. Komossa et al., Astrophys. J. 603, L17 (2004).
Acknowledgments: We gratefully acknowledge the efforts of
the many observatories whose data are presented here.
We particularly thank D. Malesani for assistance in the
calibration of the optical photometry, M. Irwin for
assistance with processing the UKIRT data, D. Fox for help
with the PTF data, and K. Hurley and J. Prochaska for
assistance in obtaining the Keck data. A.J.L. and N.R.T.
acknowledge support from the Science and Technology
Facilities Council. Swift, launched in November 2004, is a
NASA mission in partnership with the Italian Space
Agency and the UK Space Agency. Swift is managed by
NASA Goddard. Pennsylvania State University controls
science and flight operations from the Mission Operations
Center in University Park, Pennsylvania. Los Alamos
National Laboratory provides g-ray imaging analysis.
S.B.C. acknowledges generous support from G. Bengier
and C. Bengier, the Richard and Rhoda Goldman Fund,
NASA/Swift grant NNX10AI21G, NASA/Fermi grant
NNX1OA057G, and NSF grant AST-0908886. A.J.vdH.
was supported by NASA grant NNH07ZDA001-GLAST. G.L.
is supported by a grant from the Carlsberg foundation.
M.M. is supported by the Hubble Fellowship grant
HST-HF-51277.01-A, awarded by the The Space Telescope
Science Institute (STScI), which is operated by AURA
under NASA contract NAS5-26555. The Dark Cosmology
Centre is funded by the Danish National Research
Foundation. This work makes use of data obtained by the
Chandra X-ray Observatory (OBSID = 12920). This work
is based on observations made with the NASA/ESA Hubble
Space Telescope (program ID 12447), obtained from
the data archive at the Space Telescope Institute. STScI is
operated by the association of Universities for Research
in Astronomy under the NASA contract NAS 5-26555.
This work is also based on observations obtained at the
Gemini Observatory, which is operated by the Association
of Universities for Research in Astronomy, under a
cooperative agreement with the NSF on behalf of the
Gemini partnership: the National Science Foundation
(United States), the Science and Technology Facilities
Council (UK), the National Research Council (Canada),
Consejo Nacional de Investigaciones Científicas y
Técnicas (Chile), the Australian Research Council (Australia),
Ministério da Ciência e Tecnologia (Brazil), and Ministerio
de Ciencia, Tecnología e Innovación Productiva (Argentina).
The UK Infrared Telescope is operated by the Joint
Astronomy Centre on behalf of the Science and
Technology Facilities Council of the UK. UKIRT data were
processed by the Cambridge Astronomical Survey Unit.
This work is also based on observations made with the
Nordic Optical Telescope, operated on the island of
La Palma jointly by Denmark, Finland, Iceland, Norway,
and Sweden, in the Spanish Observatorio del Roque de
los Muchachos of the Instituto de Astrofisica de Canarias.
This work is also based on observations made with the
GTC and on observations obtained with the Samuel
Oschin Telescope at the Palomar Observatory as part of
the Palomar Transient Factory project. The National
Energy Research Scientific Computing Center, which is
supported by the Office of Science of the U.S. Department
of Energy under contract DE-AC02-05CH11231, provided
staff, computational resources, and data storage for this
project. The WHT is operated on the island of La Palma by
the Isaac Newton Group in the Spanish Observatorio del
Roque de los Muchachos of the Instituto de Astrofísica de
Canarias. This work is also based on observations made with
the Italian Telescopio Nazionale Galileo operated on the
island of La Palma by the Fundacion Galileo Galilei of the
INAF (Istituto Nazionale di Astrofisica) at the Spanish
Observatorio del Roque de los Muchachos of the Instituto de
Astrofisica de Canarias. The WSRT is operated by ASTRON,
with support from the Netherlands Foundation for Scientific
Research. This work is also based on observations carried
out with the IRAM Plateau de Bure Interferometer. IRAM
is supported by Institut National des Sciences de
l’Univers/CNRS (France), Max-Planck-Gesellschaft (Germany),
and Instituto Geográfico Nacional (Spain). Some of the
data presented here were obtained at the W.M. Keck
Observatory, which is operated as a scientific partnership
among the California Institute of Technology, the
University of California, and NASA. The Observatory was
made possible by the generous financial support of the
W.M. Keck Foundation. J.G., A.J.C.T., R.S.R., and S.G.
are partially supported by the Spanish programs
AYA-2007-63677, AYA-2008-03467/ESP, and
AYA-2009-14000-C03-01. R.L.C.S. is supported by a Royal
Society Fellowship. We acknowledge support by the Spanish
Ministry of Science and Innovation under project grants
AYA2008-03467/ESP and AYA2009-14000-C03-01 (including
Feder funds). J.P.B. and A.J.R. were supported by the NSF
through grants AST-0808099 and AST-0909237. A.N.M.
gratefully acknowledges support from a NSF Graduate
Research Fellowship. A.P.K. is partially supported by NSF
award AST-1008353. R.A.M.J.W. acknowledges support from
the European Research Council via Advanced Investigator
grant 247295. We acknowledge access to the major
research facilities program supported by the Commonwealth
of Australia under the International Science Linkages
program. N.B. and C.P. are NASA Einstein Fellows. M.C.C. is
a Hubble Fellow. Data used in this paper are presented in
the SOM or taken from the literature as cited.
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1207143/DC1
SOM Text
Figs. S1 to S12
Tables S1 to S9
References (30–75)
18 April 2011; accepted 8 June 2011
Published online 16 June 2011;
10.1126/science.1207143
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 202
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

A Possible Relativistic Jetted Outburst
from a Massive Black Hole Fed by a
Tidally Disrupted Star
Joshua S. Bloom,
1
* Dimitrios Giannios,
2
Brian D. Metzger,
2
S. Bradley Cenko,
1
Daniel A. Perley,
1
Nathaniel R. Butler,
1
Nial R. Tanvir,
3
Andrew J. Levan,
4
Paul T. O' Brien,
3
Linda E. Strubbe,
1,5
Fabio De Colle,
6
Enrico Ramirez-Ruiz,
6
William H. Lee,
7
Sergei Nayakshin,
3
Eliot Quataert,
1,5
Andrew R. King,
3
Antonino Cucchiara,
1,8
James Guillochon,
6
Geoffrey C. Bower,
9,1
Andrew S. Fruchter,
10
Adam N. Morgan,
1
Alexander J. van der Horst
11
Gas accretion onto some massive black holes (MBHs) at the centers of galaxies actively powers
luminous emission, but most MBHs are considered dormant. Occasionally, a star passing too
near an MBH is torn apart by gravitational forces, leading to a bright tidal disruption flare (TDF).
Although the high-energy transient Sw 1644+57 initially displayed none of the theoretically
anticipated (nor previously observed) TDF characteristics, we show that observations suggest a
sudden accretion event onto a central MBH of mass about 10
6
to 10
7
solar masses. There is
evidence for a mildly relativistic outflow, jet collimation, and a spectrum characterized by
synchrotron and inverse Compton processes; this leads to a natural analogy of Sw 1644+57 to a
temporary smaller-scale blazar.
A
lthough variability is common to all ac-
tive galactic nuclei (AGN), fundamen-
tally tied to the unsteady accretion flow
of gas toward the central massive black hole
(MBH), the time scale for active MBHs to dra-
matically change accretion rates (leading the source
to, for example, “turn off ”), is much longer than a
human lifetime. The most variable AGN are a
subclass called blazars, with typical masses M
BH

10
8
to 10
9
M

(M

is the mass of the Sun), orig-
inally found to be radio and optically bright but
with luminosities dominated by x-rays and gamma
rays. Substantial changes in the apparent lumi-
nosity over minutes- to hour-long time scales are
thought to be predominately caused by Doppler-
beamed emitting regions within a jetted outflow
moving relativistically [G
j
≈ 10 (1)] toward the
observer (2, 3). The high-energy emission is
thought to be caused by inverse Compton upscat-
tering of the accretion disk photons, photons
from within the jet itself, and/or photons from
structures external to the accretion disk (4, 5).
Inactive MBHs can suddenly “turn on” while
being fed by temporary mass accretion estab-
lished after the tidal disruption of a passing star
(6–10). If a star of mass M
*
and radius R
*
passes
within the disruption radius r
d
≈R
*
(M
BH
/M
*
)
1/3
≈5
M
7
−2/3
r
S
[withM
BH
=10
7
M

(M
7
), r
S
=2 GM
BH
/c
2
the Schwarzschild radius of the BH, M
*
= M

,
and R
*
= R

], then a mass of up to ∼M
*
/2 will
accrete onto the MBH with a peak accretion rate
on a time scale of weeks (7). The accretion rate
for typical scenarios with an M
7
BHcan be super-
Eddington (11) for months (10, 12). Candidate
tidal disruption flares (TDFs) have been observed at
x-ray, ultraviolet, and optical wavebands (13–16),
with inferred rates ∼10
−5
year
−1
gal
−1
(7), although
the observed light curves and spectra did not al-
ways match the simplest theoretical expectations.
Recently, it has been suggested (17) that a
long-lasting radio event (with a time scale to peak
of ∼1 year) could follow a TDF arising from a
jetted relativistic outflow as it interacted with
(and was slowed down by) the external ambient
medium, akin to the afterglowfromexternal shocks
after gamma-ray bursts (18). The supposition was
that the observer viewed the event off-axis from
the relativistic jet. Just what would be seen if in-
stead the jet were pointed nearly toward the ob-
server, as with the geometry inferred of blazars,
was not considered.
Sw 1644+57 was initially detected as a long-
duration gamma-ray burst (GRB110328A) by
the Swift satellite (19) at a time t
0
= 28 March
2011 12:57:45 UT. However, given the longevity
and flaring of the x-ray afterglow, it was quickly
realized that the high-energy emission was unlike
that associated with any previous GRB (20).
Based on the data available in the first 2 days
after the event, it was suggested (21–23) that Sw
1644+57, at a redshift of z = 0.3543, could be
analogous to a scaled-down version of a blazar
impulsively fed by ∼1 M

of stellar mass.
There are several lines of evidence to suggest
an accreting MBH origin for this unusual event.
First, the astrometric coincidence of the associ-
ated x-ray, optical, infrared (IR), and radio tran-
sient with the light centroid of the putative host
galaxy is strongly indicative of a positional con-
nection to an MBH(21–25). Second, the observed
x-ray variability time scales are consistent with
those of an accreting MBH [see below and sup-
porting online material (SOM)]. Last, the observed
correlation between the x-ray flux and spectral
hardness (SOM) is similar to that observed in
blazars (26, 27). Arguments against alternative
interpretations are considered in the SOM.
Accepting the accreting MBHhypothesis, we
nowexamine constraints on the BHmass and the
accretion characteristics. The x-ray light curve
implies a minimum host-frame variability time
scale of t
var,min
≈ 78 s (SOM and fig. S1). By
requiring (28, 29) that t
var,min
exceed the light-
crossing time of r
S
, we derived an upper limit on
the MBHmass M
BH
≲8 × 10
6
M

. Irrespective of
the timing argument, we can place approximate
upper limits on the mass of the central BH if we
assume that the whole mass of the galaxy [few ×
10
9
M

(20)] and its light [few× 10
9
L

(20); L

,
solar luminosity] arise from the host bulge (i.e.,
not in the disk) and apply the bulge mass–BH
mass and the bulge luminosity–BH mass corre-
lations (30–32). All such analyses suggest that
M
BH
≲ 10
7
M

, which is securely under the limit
(few × 10
8
M

) required for the tidal disruption
of a solar-mass star to occur outside the event
horizon of the MBH.
If the emission is isotropic, the average x-ray
luminosity of the outburst (SOM), L
x
≈ 10
47
erg
s
−1
, corresponds to the Eddington luminosity of
an ∼10
9
M

BH, which is incompatible with the
upper limit derived from variability. If the source
is relativistically beamed (SOM), with beaming
factor f
b
= (1 − cos qj) ≤ 1, the beaming-corrected
luminosity f
b
L
x
∼ 10
45
erg s
−1
becomes consistent
with the Eddington luminosity of an ∼10
7
M

MBH if q
j
= 1/G
j
≈ 0.1, as inferred in blazars (we
show below that this value of G
j
is also con-
sistent with the inferred rate of Sw 1644+57–like
events). We can also infer the presence of rela-
tivistic outflow (SOM) by requiring that the true
brightness temperature of the radio transient be
less than the inverse Compton catastrophe tem-
perature 10
12
K. Those constraints require a mean
G
j
≳ 1.9 from t
0
to the time of the very long base-
line interferometry (VLBI) observations reported
in (20). Separately, we can use the observed var-
iability of the radio counterpart to place con-
straints on the source size, finding G
j
≳ 10.
If the source had been active in the distant
past, we would expect to observe extended radio
emission (such as jets or other emission knots) in
VLBI imaging. Because this was not seen (20)
and archival searches spanning two decades
have yielded no evidence for prior AGN activity
from radio to gamma-ray wavebands (SOM),
the evidence thus suggests that an M
BH
= 10
6
to
10
7
M

BH underwent a dramatic turn on to
1
Department of Astronomy, University of California, Berkeley,
CA 94720-3411, USA.
2
Department of Astrophysical Sciences,
Peyton Hall, Princeton University, Princeton, NJ 08544, USA.
3
Department of Physics and Astronomy, University of Leices-
ter, University Road, Leicester LE1 7RH, UK.
4
Department of
Physics, University of Warwick, Coventry CV4 7AL, UK.
5
The-
oretical Astrophysics Center, University of California, Berkeley,
CA 94720, USA.
6
Astronomy and Astrophysics Department,
University of California, Santa Cruz, CA 95064, USA.
7
Instituto
de Astronomía, Universidad Nacional Autonoma de México,
Apartado Postal 70-264, Ciudad Universitaria, México DF
04510.
8
Computational Cosmology Center, Lawrence Berkeley
National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720,
USA.
9
Radio Astronomy Laboratory, University of California,
Berkeley, 601 Campbell Hall, Berkeley, CA 94720-3411, USA.
10
Space Telescope Science Institute, 3700 San Martin Drive,
Baltimore, MD 21218, USA.
11
Universities Space Research
Association, National Space Science and Technology Center,
320 Sparkman Drive, Huntsville, AL 35805, USA.
*To whom correspondence should be addressed. E-mail:
jbloom@astro.berkeley.edu
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 203
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

near-Eddington accretion rates, launching an
energetic relativistic outflow in the process. This
rapid increase in the accretion rate cannot result
from gas entering the sphere of influence (soi) of
the MBH, because this would require a time scale
∼R
soi
/s ≳ 10
4
year to appreciably alter the accre-
tion rate near the horizon, where R
soi
∼ 1 pc is the
radius of the sphere of influence and s ∼ 100 km
s
−1
is a typical bulge velocity dispersion. We sug-
gest instead that a TDF provides a natural ex-
planation for Sw 1644+57.
The observed x-ray fluence S
x
suggests an
energy release of E
x
= 1.6 × 10
53
f
b
erg for the
first ∼50 days. Assuming that the energy released
in the XRT band is about one-third of the bolo-
metric energy (Fig. 1) and adopting f
b
= 5 × 10
−3
,
the total energy release from the jet amounts to
0.3% of the maximum available mass energy to
be accreted if M
*
= M

. Given a typical accretion
efficiency of D
BH
≡ E
av
/m
acc
c
2
= 0.1 (m
acc
, ac-
creted mass), the jet need radiate only about 1/30th
of the available energy E
av
; if mass is lost during
the circularization phase or to subsequent disk
winds, then the required jet efficiency must be
higher. The duration of the x-ray light curve and
the requisite accretion rate are also broadly
compatible with the several-day fallback time
scale (SOM).
The broadband spectral energy distribution
(SED) of Sw1644+57 (Fig. 1) displays two peaks,
at far-IRand at x-ray/gamma-ray wavebands. Ther-
mal emission from the disk or accretion-powered
outflows (9, 10) do not naturally account for ei-
ther component. Instead, the overall spectral shape
is reminiscent of blazars, for which the peaks at
low and high energies are typically modeled as
synchrotron and inverse Compton (IC) emission,
respectively. The x-ray emission shows both a
bright/flaring and a dim/slower-varying (“qui-
escent”) state. Under the TDF hypothesis, what
could account for the observed spectrum and
temporal behavior? Possibilities are as follows:
1) Single-component synchrotron with dust
extinction. In low-luminosity BLLacertae (BLLac)
objects, the synchrotron spectral peak (nF
n
; with
specific flux density F
n
at frequency n) may oc-
cur at energies as high as hard x-rays. Thus, one
possibility is that the entire emission, from radio
to x-rays, is part of a single nonthermal syn-
chrotron spectrum originating from shocked rel-
ativistic electrons. In this scenario, the suppressed
optical emission and red IRcolors of the transient
could result from dust extinction in the visible-
light V band (A
V
) > 10 mag. Thus, although a
single extinguished synchrotron spectrum can-
not be ruled out, the large required extinction
may disfavor this interpretation (fig. S3). Fur-
thermore, although a synchrotron origin is still
likely for at least the radio emission, there is
evidence that the radio- and x-ray–emitting re-
gions may not be spatially coincident (SOM).
2) Two-component blazar emission. The
far-IRand hard x-ray peaks may, instead, represent
distinct spectral components, corresponding
to synchrotron and IC emission, respectively, as
in blazars (SOM and fig. S3). The nF
n
luminos-
ity of the low-energy peak is ∼1 to 2 orders of
magnitude weaker than that of the high-energy
peak (Fig. 1). This extreme ratio and the rela-
tively low frequency of the synchrotron peak are
both compatible with Eddington-accreting blazar
emission (5).
3) Forward shock emission from jet–interstellar
medium interaction. Although the above models
generally assume that the low- and high-energy
spectral components are directly related, evidence
suggests that they may originate from distinct radii,
at least during the x-ray flaring state. Although
the rapid variability of the x-ray emission strongly
indicates an “internal” origin (33), the radio-IR
emission varies more smoothly and could in-
stead result at larger radii from the interaction of
the jet with the surrounding interstellar medium
(SOM). If no AGN activity occurred before the
recent onset of emission, the jet must burrow its
way through the gas in the nuclear region (34).
Because of its fast motion, the newly formed jet
drives a shock into the external gas (forward
shock), while simultaneously a reverse shock
slows it down. Particles accelerated at these
shocks may power synchrotron afterglow emis-
sion beginning simultaneously when the jet forms,
yet lasting long after the internal emission has
faded. This model, the geometry of which is de-
picted in Fig. 2, appears to best accommodate the
data and predicts the long-term evolution of the
radio and IR transient (SOM).
No rising UV-optical transient nor slowly
evolving thermal x-ray component has been seen
to date from Sw 1644+57, in contrast with the
nominal expectations of TDFs. However, if Sw
1644+57 was obscured by dust, then UV-optical
suppression of the transient would be expected.
And if we understand the thermal x-ray emission
as being outshone by the jetted emission in the
first weeks, the thermal component may emerge
on a time scale of months. For this to occur, the
jet emission must be quenched due, for example,
Fig. 1. Multiwavelength spectral en-
ergy distribution of Sw 1644+57 at t
0
+
2.9 days. Our radio-through-UV mea-
surements are represented by solid
circles, with data from the published
circulars (20) represented by open cir-
cles (41). X-ray and soft gamma-ray
points from the Swift X-ray Telescope
(XRT) and Burst Alert Telescope (un-
corrected for host-galaxy absorption)
are shown as black crosses, and the
Fermi Large Area Telescope gamma-
ray upper limit (42) is shown at the
far right. The 90% uncertainty region
of a power-law fit to the XRT data (with
N
H
absorption removed) is represented
by the blue bowtie. (Inset) The same
data zoomed in on the optical–near-IR
window. Overplotted are two different
multicomponent models for the SED
(43) (Fig. 2). The orange curve shows
a model with synchrotron, synchrotron
self-Compton, and external Compton
(EC) contributions. The purple curve
shows a model in which the IR emis-
sion originates from a compact source
of synchrotron emission (∼4 × 10
14
cm).
Both models require moderate extinction (A
V
∼ 3 to 5 mag). Additional synchrotron models are shown in fig. S3. The model SEDs here and in fig. S3 were
generated using the computer code from (44, 45).
10
10
10
12
10
14
10
16
10
18
10
20
10
22
10
24
Observed frequency ν (Hz)
10
40
10
41
10
42
10
43
10
44
10
45
10
46
S
p
e
c
i
f
i
c

l
u
m
i
n
o
s
i
t
y

ν
L
ν

(
e
r
g

s

1
)
10
−17
10
−16
10
−15
10
−14
10
−13
10
−12
10
−11
10
−10
S
p
e
c
i
f
i
c

f
l
u
x

ν
F
ν

(
e
r
g

s

1

c
m

2
)
Synch + EC + self−Compton
Synch + EC + afterglow
Obs. wavelength (µm)
0.5 1.0 2.0 4.0
10
14
10
15
10
−15
10
−14
10
−13
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 204
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

to a transition of the accretion flowto a soft/thermal
state once the accretion rate becomes sub-Eddington,
in analogy to the behavior of stellar-mass x-ray
binaries. Even in this case, whether and when ther-
mal emission will be observable hinges on the de-
gree of dust extinction and its brightness relative
to the host bulge.
If the TDF hypothesis is correct, Sw1644+57
will fade over the coming year and will not re-
peat. If our interpretation about the relativistic
flowand spectral origin is correct, then we would
expect the transient emission to be polarized at a
(low) level similar to that seen in gamma-ray burst
afterglows [as opposed to blazars (35, 36)]. More-
over, we expect to see evidence for superluminal
motion of the radio source as seen in VLBI moni-
toring over the next few months; the source itself
may become resolved on time scales of a few
months if it remains bright enough to detect at
radio wavebands.
Adopting a beaming fraction f
b
≲ 10
−2
con-
sistent with that inferred fromSw1644+57 (SOM),
we conclude that for every on-axis event, there
will be 1/f
b
≳ 10
2
events pointed away from our
line of sight. Because Swift has detected only one
such event in ∼6 years of monitoring, the total
inferred limit on the rate of TDFs accompanied
by relativistic ejecta is ≳10 year
−1
out to a sim-
ilar distance. Although the majority of such events
will not produce prompt high-energy emission,
bright radio emission is predicted once the ejecta
decelerates to nonrelativistic speeds on a time
scale of ∼1 year (17). The predicted peak flux is
sufficientlyhigh(∼0.1to1 mJyat several-gigahertz
frequencies and redshifts similar to that of Sw
1644+57) that ∼10 to 100 may be detected per
year by upcoming radio transient surveys.
The emerging jet from the tidal disruption
event appears to be powerful enough to ac-
celerate cosmic rays up to the highest observed
energy (∼10
20
eV) (37). The observed rate of
jets associated with the tidal disruption of a
star,
˙
R
e
10
−11
Mpc
−3
year
−1
, and the energy
released per event of 3 × E
x
∼ 5 × 10
53
erg,
however, imply an energy injection rate of
˙
E
TDF
e
5 Â10
42
erg Mpc
−3
year
−1
. Despite the
large uncertainty, this rate is substantially smaller
than the injection rate
˙
E
inj
e
10
44
erg Mpc
−3
year
−1
required to explain the observed flux of cosmic
rays of energy >10
19
eV (38). This conclusion
is, however, subject to uncertainties associated
with the radiative efficiency of the jet.
There is much evidence that AGN jets are
accelerated by magnetohydrodynamic, rather
than hydrodynamic, forces (39). A key unsolved
question is whether the large-scale magnetic field
necessary to power the jet is advected in with the
flow (40), or whether it is generated locally in
the disk by instabilities or dynamo action. If the
jet is launched from a radius R
in
, the magnetic
field strength at its base (B) is related to the jet
luminosity by L
j
∼ pR
in
2
c × (B
2
/4p). If we as-
sume L
j
∼ 10
45
erg s
−1
, similar to the Eddington
limit for a ∼10
7
M

MBH (as appears necessary
to explain the bright nonthermal emission), the
required field strength is B∼10
5
Gfor R
in
∼1.5 r
S
.
This field is much higher than the average field
strengths of typical main-sequence stars (<10
3
G).
The stellar field is further diluted because of
flux freezing by a factor ∼(R
*
/R
in
)
2
as matter falls
into the BH, where R
*
∼ R

is the stellar radius
before disruption. Hence, the large-scale field
responsible for launching the jet associated with
Sw 1644+57 must have been generated in situ.
Placing similar constraints has not previously
been possible in the context of normal AGN or
x-ray binary disks, because of the much larger
ratio between the outer and inner disk radii in
such systems.
References and Notes
1. The inferred typical Lorentz factor of blazar jets
G
j
= (1 − b
2
)
−1/2
≈ 10, with velocity v = bc of the
jetted outflow (c is the speed of light in vacuum).
2. M. Ulrich, L. Maraschi, C. M. Urry, Annu. Rev. Astron.
Astrophys. 35, 445 (1997).
3. L. Maraschi, F. Tavecchio, Astrophys. J. 593, 667 (2003).
4. M. Böttcher, Astrophys. Space Sci. 309, 95 (2007).
5. G. Fossati, L. Maraschi, A. Celotti, A. Comastri,
G. Ghisellini, Mon. Not. R. Astron. Soc. 299, 433 (1998).
6. M. J. Rees, Nature 333, 523 (1988).
7. C. R. Evans, C. S. Kochanek, Astrophys. J. Lett. 346, L13
(1989).
8. J. Wang, D. Merritt, Astrophys. J. 600, 149 (2004).
9. J. E. Grindlay, AIP Conf. Proc. 714, 413 (2004).
10. L. E. Strubbe, E. Quataert, Mon. Not. R. Astron. Soc. 400,
2070 (2009).
11. The Eddington accretion rate is a theoretical upper
limit to spherical mass accretion, whereby mass inflow
due to gravitation attraction to a central source is
balanced by radiation pressure from that source.
Super-Eddington luminosities are possible when mass
accretion is not homogeneous, spherical, and/or the main
source of emission near the radiation source is not in
the form of photons (for example, neutrinos). For a
10
7
M
ʘ
(M
7
) BH, the Eddington luminosity is
L
Edd
= 1.3 × 10
45
erg s
−1
.
12. E. Ramirez-Ruiz, S. Rosswog, Astrophys. J. Lett. 697, L77
(2009).
13. S. Komossa, J. Greiner, Astron. Astrophys. 349, L45
(1999).
14. S. Gezari et al., Astrophys. J. 676, 944 (2008).
15. S. van Velzen et al., Optical discovery of stellar tidal
disruption flares (2010); available at http://arxiv.org/abs/
1009.1627.
16. S. B. Cenko et al., PTF10iya: A short-lived, luminous flare
from the nuclear region of a star-forming galaxy (2011);
available at http://arxiv.org/abs/1103.0779.
17. D. Giannios, B. D. Metzger, Radio transients from stellar
tidal disruption by massive black holes (2011); available at
http://arxiv.org/abs/1102.1429.
18. M. J. Rees, P. Meszaros, Mon. Not. R. Astron. Soc. 258,
41P (1992).
19. J. R. Cummings et al., “GRB 110328A: Swift detection of
a burst” (GCN Circular 11823, NASA Goddard Space
Flight Center, Greenbelt, MD, 2011).
20. A. J. Levan et al., Science 333, 199 (2011).
21. J. S. Bloom, N. R. Butler, S. B. Cenko, D. A. Perley, “GRB
110328A / Swift J164449.3+573451: X-ray analysis and a
mini-blazar analogy” (GCN Circular 11847, NASA Goddard
Space Flight Center, Greenbelt, MD, 2011).
22. Interpretations similar to that in (21) were put forward
later (23).
23. U. Barres de Almeida, A. De Angelis, Enhanced emission
from GRB 110328A could be evidence for tidal disruption
of a star (2011); available at http://arxiv.org/abs/
1104.2528.
24. E. Berger et al., “GRB 110328A / Swift J164449.3+573451:
Radio-optical/NIR Astrometry” (GCN Circular 11854, NASA
Goddard Space Flight Center, Greenbelt, MD, 2011).
25. Within the uncertainties from Hubble imaging [∼300 pc;
(20)], the central stellar and gas density could be high
enough to allow other progenitors, such as supernovae.
26. T. Takahashi et al., Astrophys. J. Lett., 470, L89 (1996).
27. G. Fossati et al., Astrophys. J. 541, 166 (2000).
28. As in gamma-ray burst light curves, even in the presence
of relativistic motion, the observed variability should
track that of the energy injection time scales from the
central engine. Shorter time scale variability could,
however, result from compact emitting regions moving
fast within the jet (29).
29. D. Giannios, D. A. Uzdensky, M. C. Begelman, Mon. Not.
R. Astron. Soc. 395, L29 (2009).
30. J. Magorrian et al., Astron. J. 115, 2285 (1998).
31. T. R. Lauer et al., Astrophys. J. 662, 808 (2007).
32. J. Kormendy, R. Bender, M. E. Cornell, Nature 469, 374
(2011).
33. M. Spada, G. Ghisellini, D. Lazzati, A. Celotti, Mon. Not.
R. Astron. Soc. 325, 1559 (2001).
34. This situation is not encountered in normal (long-lived)
blazars because a large ≳kiloparsec-scale cavity has been
carved by the preceding outflow.
35. Here the departure from the blazar analogy is worth
noting, in that the physics of the radio emission is likely
to be different in this case: We have suggested that the
emission is originating from the shocked surrounding
material (forward shock), not the shocked jet as in normal
blazars, which could contain large-scale fields. Even so, only
10% of flat-spectrum radio quasars and BL Lac objects have
polarization larger than our VLBI limits (36).
36. M. F. Aller, H. D. Aller, P. A. Hughes, Astrophys. J. 586,
33 (2003).
37. G. R. Farrar, A. Gruzinov, Astrophys. J. 693, 329 (2009).
Fig. 2. Schematic representation of the geometry
and emission regions for Sw 1644+57. A star is
disrupted at distance r
d
from an MBH of mass M
BH
with Schwarzschild radius r
S
. Half of the mass of
the star escapes on unbound orbits while the other
half remains bound. Shocked, circularized fallback
mass sets up a temporary accretion disk with inner
radius 3 r
S
(for a nonspinning BH). A two-sided jet
is powered starting at the time of accretion and
plows through the interstellar region surrounding
the BH at a Lorentz factor G
j
. At some later time,
the jet has reached a distance R
j
, where the forward
shock radiates the observed radio and IR light.
Emission from the accretion disk is Compton-
upscattered, giving rise to the observed x-rays.
Different viewing angles (whether the observer is
inside q
j
≈ 1/G
j
or not) determine what sort of
phenomena is observed. An analogy with blazars
and AGN for more massive BHs is given.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 205
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

38. E. Waxman, Astrophys. J. Lett. 452, L1 (1995).
39. N. Vlahakis, A. Königl, Astrophys. J. 605, 656 (2004).
40. H. C. Spruit, D. A. Uzdensky, Astrophys. J. 629, 960
(2005).
41. The uncertain relative contributions of the host galaxy
and the optical/IR transient result in very large
uncertainties for the J and z photometric data points. In
this model, the radio and IR emission are produced by
synchrotron radiation from an extended source, whereas
the x-ray emission is dominated by the Compton
scattering of external photons from the accretion disk (for
illustrative purposes, we assume a 10
6
M
ʘ
MBH). As in
the orange model, the x-ray emission is dominated by
external Compton scattering, whereas the peak at high
energies results from synchrotron self-Compton emission.
An additional synchrotron component from a mildly
relativistic blast-wave afterglow at larger radius is
invoked to explain the bright radio and millimeter fluxes.
42. S. Campana, L. Foschini, G. Tagliaferri, G. Ghisellini,
S. Covino, “GRB 110328/Swift J164449.3+573451:
Fermi observations” (GCN Circular 11851, NASA Goddard
Space Flight Center, Greenbelt, MD, 2011).
43. Although we have modeled the “quiescent” SED, a similar
external Compton model can be made to fit the “flaring”
x-ray state, provided that the jet luminosity is accordingly
increased.
44. H. Krawczynski et al., Astrophys. J. 601, 151 (2004).
45. For the models, G
j
= 10 and the magnetic field strength
B = 10 (0.001) Gauss for the synchrotron + external
Compton + afterglow (synchrotron + external Compton +
self-Compton) model. In the synchrotron + external
Compton + afterglow model, we find a disk luminosity of
L
disk
= 0.4 L
Edd
, with L
Edd
= 1.3 × 10
44
erg s
−1
,
corresponding to a BH with 10
6
M
ʘ
. In the synchrotron +
external Compton + self-Compton model, we find a disk
luminosity of L
disk
= 0.04 L
Edd
, with L
Edd
= 1.3 × 10
45
erg s
−1
,
corresponding to a BH with 10
7
M
ʘ
.
Acknowledgments: We thank R. Romani, C. McKee, L. Blitz,
and J. Hjorth for close reads of drafts of this work and
for helpful interactions. We are grateful to the entire
Swift team for work on their remarkable observatory
that enabled discovery of this event. Swift, launched in
November 2004, is a NASA mission in partnership with
the Italian Space Agency and the UK Space Agency.
Swift is managed by NASA Goddard. Pennsylvania State
University controls science and flight operations from
the Mission Operations Center in University Park, PA. Los
Alamos National Laboratory provides gamma-ray imaging
analysis. J.S.B. and his group were partially supported by
grants NASA/NNX10AF93G, NASA/NNX10AI28G, and
NSG/AST-100991. D.G. acknowledges support from the
Lyman Spitzer Jr. Fellowship awarded by the Department
of Astrophysical Sciences at Princeton University. B.D.M. is
a NASA Einstein Fellow and is supported by NASA through
Einstein Postdoctoral Fellowship grant number PF9-00065
awarded by the Chandra X-ray Center, which is operated
by the Smithsonian Astrophysical Observatory for NASA
under contract NAS8-03060. S.B.C. acknowledges
generous support from Gary and Cynthia Bengier, the
Richard & Rhoda Goldman fund, NASA/Swift grant
NNX10AI21G, NASA Fermi grant NNX10A057G, and
NSF grant AST-0908886. N.R.B. is a NASA Einstein Fellow.
W.H.L. is supported in part by Consejo Nacional de Ciencia
y Tecnología grant 83254. A.J.vdH. was supported by
NASA grant NNH07ZDA001-GLAST. The primary references
for the data presented here were given in the text and
may be found in (20) or in the NASA/Swift archive
(http://heasarc.nasa.gov/docs/swift/archive/).
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1207150/DC1
Materials and Methods
Figs. S1 to S3
References
18 April 2011; accepted 6 June 2011
Published online 16 June 2011;
10.1126/science.1207150
Observation of Transient
Structural-Transformation
Dynamics in a Cu
2
S Nanorod
Haimei Zheng,
1
* Jessy B. Rivest,
2
Timothy A. Miller,
3
Bryce Sadtler,
1,4
† Aaron Lindenberg,
3
Michael F. Toney,
5
Lin-Wang Wang,
1
Christian Kisielowski,
1,6
A. Paul Alivisatos
1,4
*
The study of first-order structural transformations has been of great interest to scientists in
many disciplines. Expectations from phase-transition theory are that the system fluctuates
between two equilibrium structures near the transition point and that the region of transition
broadens in small crystals. We report the direct observation of structural fluctuations within a
single nanocrystal using transmission electron microscopy. We observed trajectories of structural
transformations in individual nanocrystals with atomic resolution, which reveal details of the
fluctuation dynamics, including nucleation, phase propagation, and pinning of structural
domains by defects. Such observations provide crucial insight for the understanding of
microscopic pathways of phase transitions.
F
irst-order structural transformations in
solids play an important role in a variety
of processes ranging from information stor-
age (1, 2) to materials processing (3). An un-
derstanding of the microscopic mechanisms of
structural transformations is critical for under-
standing and controlling these processes. In nano-
scale systems, the energetic barrier to a structural
transformation scales with crystal size. When
the size of a nanocrystal is in a regime where
thermal energy is comparable to the energy bar-
rier for phase transformation, fluctuations be-
tween two stable structures occur at the transition
point (4, 5). This is relevant to many molecular
and solid-state phenomena near equilibrium,
and there have been numerous studies of the
ensemble fluctuations that accompany these
phase transformations (6–10). However, in en-
semble studies, only the average characteristics
of the fluctuations can be observed, and many
important features may be completely obscured
by parallel, unsynchronized transition processes.
In this study, we focus on observing the indi-
vidual structural fluctuations in a single nano-
crystal by taking advantage of recent advances in
transmission electron microscopy (TEM). Ad-
vances in electron optics and recording systems
for TEMhave enabled rapid imaging with single-
atom sensitivity across the periodic table and
with greater collection efficiency (11–13). This
provides extraordinary opportunities to study
the structural-transformation dynamics in situ
with atomic resolution.
A model system chosen for this study is the
structural transformation between the low- and
high-chalcocite phases of copper sulfide (Cu
2
S)
nanorods (i.e., with dimensions of 5 by 28 nm).
We synthesize Cu
2
S nanorods through a colloidal
solution process, and the nanorods showthe low-
chalcocite phase at room temperature (14). The
low-chalcocite is monoclinic with a space group
of P2
1
/c or Pc, in which copper atoms partially
occupy the lattice sites within a distorted hexag-
onal sulfur lattice frame (15). When Cu
2
S trans-
forms into the high-chalcocite structure (a space
group of P6
3
/mmc), the hexagonal sulfur sub-
lattice remains rigid, but copper atoms occupy
different lattice sites (16, 17) (see the Cu-S phase
diagram in fig. S1). In Cu
2
S nanorods, the struc-
tural transformation is poised relatively close to
roomtemperature [376 Kin bulk and ~337 T 4 K
in the nanorods (18)]; therefore, thermal energy
sufficient to induce the structural transformation
is low enough to avoid melting, ripening, defect
rearrangement, or other unwanted perturbations
of the nanocrystals. The nanorod geometry allows
for the atomic structure of the material to be re-
solved through the thickness of a fewnanometers,
whereas the ability to independently control the
length enables us to adjust the number of fluc-
tuating domains within the particle.
Another convenient feature of the Cu
2
S nano-
rod system is that because the transformation
temperature is relatively low, heating from the
electron beamcan be used to induce the structural
transformations. Samples for this investigation are
prepared by drop-casting a dilute solution of Cu
2
S
nanorods on a conductive carbon grid. While im-
aging by TEM, part of the energy dissipated from
the interaction between the nanocrystal and the
1
Materials Sciences Division, Lawrence Berkeley National Lab-
oratory, Berkeley, CA 94720, USA.
2
Department of Mechanical
Engineering, University of California, Berkeley, CA 94720, USA.
3
Department of Materials Science and Engineering, Photon
Science, Stanford University, Stanford, CA 94305, USA.
4
De-
partment of Chemistry, University of California, Berkeley, CA
94720, USA.
5
Stanford Synchrotron Radiation Laboratory, Palo
Alto, California, 94025, USA.
6
National Center for Electron
Microcopy, Lawrence Berkeley National Laboratory, Berkeley,
CA 94720, USA.
*To whom correspondence should be addressed. E-mail:
hmzheng@lbl.gov (H.M.); alivis@berkeley.edu (A.P.A.)
†Present address: Division of Chemistry and Chemical
Engineering, California Institute of Technology, Pasadena,
CA 91125, USA.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 206
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

electron beam (e.g., inelastic scattering of the
incident electrons) is converted into heat. The
equilibrium temperature of the nanocrystal is
determined by the electron current density (en-
ergy input) and thermal dissipation. There have
been many reports on the heating of a sample by
electron-beam irradiation, where a wide range of
temperature increases have been reported (19–21).
Here, we found that under controlled electron-
beamirradiation a Cu
2
S nanorod on a carbon thin
film can be heated above its phase-transition tem-
perature of 337 T 4 K [i.e., it is estimated that at
steady-state conditions, a Cu
2
S nanorod is heated
to ~347.8 Kunder electron-beamirradiation with
a current density of 5000 electrons per Å
2
per
second; see details in the supporting online
material (18)]. The structural-transformation
temperature of an ensemble of Cu
2
S nanorods
without electron-beam irradiation was verified
by x-ray diffraction studies of a powder of Cu
2
S
nanorods (18).
We have estimated that fluctuations between
the two structures in the Cu
2
S nanorods should
be observable under the above imaging condi-
tions. As the system approaches the critical
temperature, the relative probability of observing
the two phases is given by (22)
P
1
/P
2
¼ exp
−(E
1
− E
2
)DT
T
C
k
B
T

ð1Þ
where P
1
is the probability of observing phase 1,
P
2
is the probability of observing phase 2, T is
temperature, T
c
is the phase transition temperature,
k
B
is the Boltzmann constant, E
1
– E
2
= (e
1
– e
2
)N
is the latent heat (enthalpy of transition), (e
1
– e
2
)
is the transition enthalpy per Cu
2
S unit cell, and
N is the number of Cu
2
S unit cells. For a Cu
2
S
nanocrystal of a given size (for instance, N= 1000),
if we take ∆e = 3849 J/mol (40 meV per unit cell)
(23), T
c
= 337 T 4 K, and P
1
/P
2
= e
−1
(at the critical
phase-transition temperature, the thermal fluctua-
tion of kT is equal to the transition energy), we find
that ∆T ~ 0.2 K. This suggests that the structural
transformation in a Cu
2
S nanocrystal with N =
1000 (2- to 4-nm domains) occurs within a tem-
perature range of T
c
T 0.2 K. Our experiments
show that this broadening in the transition tem-
perature provides ample opportunity to collect
structural-transformation trajectories.
Figure 1 illustrates the image-processing tech-
nique that we have developed to visualize the
different structural domains within a Cu
2
S nano-
rod in a high-resolution TEM (HRTEM) image.
We apply digital masks in Fourier space, which
are characteristic of each of the two phases.
The corresponding filtered real-space images
allow for the identification of the low- or high-
chalcocite domains in a single nanocrystal. For
noise reduction, our mask design captures only
Fourier components above a given threshold
[e.g., signal/noise > 10; see II and III in Fig. 1A,
in which the MacTempas image simulation pro-
gram (24) was used for image processing and
Adobe Photoshop software was used for false
Fig. 1. Images and image process to visualize the low- and high-chalcocite structures in a Cu
2
S nanorod.
(A) Image process to highlight the low-chalcocite (green) and high-chalcocite (red) domains in HRTEM
images of a Cu
2
S nanorod. (I) Obtain the fast Fourier transform (FFT) pattern (left) of the original HRTEM
image (right). (II) Place a mask in the FFT pattern in (I) allowing the diffraction spots of the hexagonal
lattice to pass (left). The corresponding filtered image shows the high-chalcocite domain or the hexagonal
lattice frame of low-chalcocite (right). (III) Place a mask in the FFT pattern (I) only allowing the low-
chalcocite diffraction spots to pass (left). The corresponding filtered image shows the low-chalcocite domain
(right). (IV) Highlight the two structures in the Cu
2
S nanorod by overlapping the two filtered images (right)
and the two masks in II and III (left). (B) The low-chalcocite structure in the [110] zone axis. Images fromtop
to bottom: the atomic structure from the selected section in (A) (top), simulated image (second row),
structural model (third row), and simulated electron diffraction (bottom). In the atomic structure in the top
image, the low-chalcocite (in the core) is superimposed with the high-chalcocite structure (outer layer of the
nanorod). (C) The high-chalcocite structure in the [010] zone axis displayed in the same order as in (B).
Fig. 2. Trajectory of the struc-
tural transformations in a Cu
2
S
nanorod. (A) Sequential im-
ages showing the low-chalcocite
structure (green), mixed structure
(mixed green and red domains),
and the pure high-chalcocite
structure (red). (B) Trajectory of
structural transformations from
a low-chalcocite structure (L) to
the transitionperiodwithfluctua-
tions betweenthe twophases, and
the final stable high-chalcocite
structure (H).
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 207
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

coloring of the images (see fig. S2)]. As a result,
the filtered images do not represent all atom po-
sitions in the two phases, but rather reflect the size
and shape of the structural domains. By applying
this procedure to a time series of HRTEMimages,
the spatial distribution of the two phases within a
single Cu
2
S nanorod can be tracked in time.
We recorded a series of HRTEM images con-
tinuously at a rate of 0.5 s per frame using an
aberration-corrected transmission electron mi-
croscope operated at 80 kV. The microscope
is tuned to a spherical aberration (Cs) value of
–0.015 mm, and the defocus is set to ~8 nm.
Under these conditions, atomic columns appear
bright and the intensities reveal the positions of
copper and sulfur as long as the surface rough-
ness of the sample does not substantially exceed
the focal spread in the beam direction (~1 nm)
(13). Thus, the atomic structures of the two
phases are resolved in a single image (Fig. 1, B
and C; a comparison between the single image
and the phase image by exit-wave reconstruction
for each phase is shown in fig. S3). By applying
the above image processing to the HRTEM im-
age series collected under these conditions, the
dynamic spatial distribution of the structural do-
mains, as well as changes in the atomic structure
of a Cu
2
S nanorod during structural transforma-
tion, can be observed.
Figure 2 shows the trajectory of the structur-
al transformation from low- to high-chalcocite
phase in a single Cu
2
S nanorod (similar trajec-
tories of phase transitions have been observed in
many other Cu
2
S nanorods). The nanorod ini-
tially has the expected low-chalcocite structure.
Under electron-beam irradiation, either the com-
plete Cu
2
S nanorod or a portion of it temporarily
transforms into high-chalcocite phase, and the
two structures fluctuate for an extended period
of time. Ultimately, the nanorod transforms into
the stable high-chalcocite phase (Fig. 2; see de-
tails in movie S1). In Fig. 2B, “H” represents the
states with a pure high-chalcocite structure in the
Cu
2
S nanorod, and “L” represents the presence
of low-chalcocite phase in the nanorod, including
both the pure low-chalcocite structure and a mixed
structure. By decreasing (or increasing) the elec-
tron current density, the duration of the period
during which fluctuations are observed increases
(or decreases, respectively). We have further no-
ticed that once the Cu
2
S nanorod has transformed
into the stable high-chalcocite structure, the struc-
ture remains, even when the electron beam is
shut off for a period of time. This suggests that
the high-chalcocite structure tends to be trapped
in the nanorod (25).
The high level of detail of our observations
provides insight into the nature of structural trans-
formation in a Cu
2
S nanorod. For instance, we
can see differences in the nucleation processes
between the forward and reverse transformations.
At the onset of the transition from low- to high-
chalcocite, the high-chalcocite structure is ob-
served at the outer surface of the low-chalcocite
nanorod (Fig. 3A). The high-chalcocite propa-
gates inward concentrically until the whole nano-
rod is transformed into pure high-chalcocite phase.
This transition behavior is similar to a solid-liquid
phase transition, such as the high-temperature
melting of a metal nanoparticle (26). When the
low-chalcocite phase reappears (nucleates), it is
located at the core of the high-chalcocite Cu
2
S
nanorod (Figs. 2 and 3, also see movies S1, S3,
and S5). The low-chalcocite domain (i.e., 2 to 4 nm
in diameter) can propagate along the long-axis of
the nanorod or grow into a larger domain, which
suggests that the transition from high- to low-
chalcocite is a nucleation and growth process. It
is likely that there are rapid structural fluctuations
of even smaller domains during the nucleation
process.
The dynamics of the Cu
2
S structural trans-
formation are strongly affected by the presence
of defects. Defects, such as a stacking fault across
a Cu
2
S nanorod, separate the nanorod into dif-
ferent structural domains (Fig. 4 A and B). Tra-
jectories of structural fluctuations are different in
adjacent domains (Fig. 4C; also see movie S3).
The smaller domain (zone II) fluctuates more fre-
quently than a larger domain (zone I). We have
observed the prevalence of high-chalcocite struc-
ture at the domain boundary during the transition,
which is probably due to the higher energy of
the defect sites. One can imagine that complex
phase-transition kinetics occurs in bulk materials
and nanocrystal ensembles due to parallel phase-
transition processes in different parts of the sam-
ple. The nanorods present a simplified case where
fluctuations in just a fewdomains along the length
of the nanorod can be monitored.
We have estimated the fluctuation kinetics by
using a thermodynamic fluctuation model (27).
For a small systemembedded in a larger reservoir
(e.g., a low-chalcocite domain at the core of a high-
chalcocite nanorod), the probability for the small
system to have an internal energy of E is
P(E) ¼ exp
−(E − E
0
)
2
2k
B
T
2
C

ð2Þ
where E
0
is the average internal energy of the
small system and C is the specific heat for the
small system (27). We assume that the structural
transformation occurs when the energy change
(E – E
0
) is equal to the interface energy (E
s
) of
the small system. The fluctuation time t can be
expressed as
t ¼ t
0
exp
E
2
s
2k
B
T
2
C

ð3Þ
Here, t
0
is an attempt time for the atoms to
execute the transition, usually taken as a vibra-
Fig. 4. The effect of defects on the structural trans-
formations of a Cu
2
S nanorod. (A) HRTEM image
after image processinghighlightingthe low-chalcocite
domains (green). (B) Filtered image showing (110)
planes, where the regions marked with dashed lines
highlight two stacking faults. (C) Trajectories of struc-
tural fluctuations in zones I and II during the transi-
tion period.
Fig. 3. Sequential images showing the pathways of structural transformations in two Cu
2
S nanorods. (A)
The high-chalcocite structure (red) is formed at the outer surface of the nanorod. The low-chalcocite
domain is shown in green. (B) The low-chalcocite structure nucleates at the core of the high chalcocite.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 208
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

tional period, which is on the order of picoseconds.
E
s
between high- and low-chalcocite phases re-
sults from the different Cu arrangements in these
two phases and the related extra Ewald energy at
the interface. This is similar to the case of the inter-
face between wurtzite (WZ) and zincblende (ZB),
where the different second nearest-neighbor atomic
positions cause different Ewald energies in these
two phases. It has been found that the interface
energy between WZ and ZB per surface unit cell
is similar to the energy difference per unit cell
(28). Thus, if we take this approximation that the
interface energy between low- and high-chalcocite
is the same as the internal energy difference be-
tween these two phases, (e
1
– e
2
) = 40 meV per
unit cell (23), and assume that there are N Cu
2
S
unit formulae inside a spherical core, we have E
s
=
(36p)
1/3
N
2/3
(e
1
– e
2
). Note that C = C
unit
N, and
C
unit
= 52 J/mol·K (23). Thus, fromEq. 3, we get
t ~ 2 s when N = 1000 and t
0
= 1 ps (29). This
fluctuation time is of the same order as our ob-
served experimental value.
In summary, we have observed dynamic struc-
tural transformations of a single Cu
2
S nanorod
from a low- to a high-chalcocite structure. The
influence of the surface and interface energies
on nucleation and pinning phenomena of a par-
ticular phase by defects suggests strategies for
stabilizing metastable structures. The ability to
directly visualize these processes will aid in the
future design of materials with new and con-
trolled phases.
References and Notes
1. M. H. R. Lankhorst, B. W. S. M. M. Ketelaars,
R. A. M. Wolters, Nat. Mater. 4, 347 (2005).
2. S. H. Lee, Y. Jung, R. Agarwal, Nat. Nanotechnol. 2, 626
(2007).
3. J. Christian, The Theory of Transformations in Metals and
Alloys (Pergamon, London, 1965).
4. H. E. Stanley, Introduction to Phase Transitions and
Critical Phenomena (Oxford Univ. Press, New York, 1987).
5. M. S. S. Challa, D. P. Landau, K. Binder, Phase Transit.
24, 343 (1990).
6. K. Jacobs, D. Zaziski, E. C. Scher, A. B. Herhold, A. Paul
Alivisatos, Science 293, 1803 (2001).
7. Q. X. Guo et al., Nano Lett. 8, 972 (2008).
8. R. F. Smith et al., Phys. Rev. Lett. 101, 065701 (2008).
9. J. S. Wittenberg, M. G. Merkle, A. P. Alivisatos, Phys. Rev.
Lett. 103, 125701 (2009).
10. Z. W. Wang et al., Proc. Natl. Acad. Sci. U.S.A. 107,
17119 (2010).
11. M. Lentzen, Microsc. Microanal. 14, 16 (2008).
12. C. Ö. Girit et al., Science 323, 1705 (2009).
13. R. Erni et al., Phys. Rev. B 82, 165443 (2010).
14. B. Sadtler et al., J. Am. Chem. Soc. 131, 5285 (2009).
15. H. T. Evans, Nat. Phys. Sci. 232, 69 (1971).
16. M. J. Buerger, B. J. Wuensch, Science 141, 276 (1963).
17. A. Putnis, Am. Mineral. 62, 107 (1977).
18. Supporting material is available on Science Online.
19. T. Yokota, M. Murayama, J. M. Howe, Phys. Rev. Lett. 91,
265504 (2003).
20. R. F. Egerton, P. Li, M. Malac, Micron 35, 399 (2004).
21. A. Reguer et al., Ultramicroscopy 110, 61 (2009).
22. M. S. S. Challa, D. P. Landau, K. Binder, Phys. Rev. B 34,
1841 (1986).
23. D. J. Chakrabarti, D. E. Laughlin, J. Phase Equil. 4, 254
(1983).
24. www.totalresolution.com
25. R. W. Potter, Econ. Geol. 72, 1524 (1977).
26. S. L. Lai, J. Y. Guo, V. Petrova V, G. Ramanath,
L. H. Allen, Phys. Rev. Lett. 77, 99 (1996).
27. L. D. Landau, E. M. Lifshitz, Statistical Physics (Pergamon,
New York, 1980).
28. F. Glas, J. Appl. Phys. 104, 093520 (2008).
29. A. F. Voter, Phys. Rev. B 34, 6819 (1986).
Acknowledgments: We thank U. Dahmen at National Center
for Electron Microscopy (NCEM) and E. Borrero and
C. Dellago at University of Vienna for helpful discussions.
The authors are grateful to the support of Helios Solar
Energy Research Center (SERC) and NCEM, which are
funded by the director of the Office of Science, Office of
Basic Energy Sciences (BES), Materials Science and
Engineering Division of the U.S. Department of Energy
(DOE) under contract no. DE-AC02-05CH11231. TEM
experiments were performed using TEAM0.5 microscope
at NCEM. X-ray experiments were performed at the
Stanford Synchrotron Radiation Laboratory, a DOE-BES
user facility. J.B.R. was funded by a fellowship from Intel
Corporation, and T.A.M. and A.L. were supported by the
DOE BES Materials Sciences and Engineering Division.
H.Z. conceived the work and performed the experiments;
H.Z. and A.P.A. analyzed the data and wrote the paper;
J.B.R. prepared nanorod samples, took ensemble x-ray
data, and edited the paper; B.S. contributed to nanorod
sample preparation and edited the paper; M.F.T., A.L.,
and T.A.M. provided x-ray analysis infrastructure; L.-W.W.
provided theoretical calculation and edited the paper; and
C.K. contributed to image analysis and edited the paper.
Supporting Online Material
www.sciencemag.org/cgi/content/full/333/6039/206/DC1
Materials and Methods
SOM Text
Figs. S1 to S4
Table S1
References (30–35)
Movies S1 to S6
23 February 2011; accepted 25 May 2011
10.1126/science.1204713
Palladium-Catalyzed Aerobic
Dehydrogenation of Substituted
Cyclohexanones to Phenols
Yusuke Izawa, Doris Pun, Shannon S. Stahl*
Aromatic molecules are key constituents of many pharmaceuticals, electronic materials, and
commodity plastics. The utility of these molecules directly reflects the identity and pattern of
substituents on the aromatic ring. Here, we report a palladium(II) catalyst system, incorporating an
unconventional ortho-dimethylaminopyridine ligand, for the conversion of substituted
cyclohexanones to the corresponding phenols. The reaction proceeds via successive
dehydrogenation of two saturated carbon-carbon bonds of the six-membered ring and uses
molecular oxygen as the hydrogen acceptor. This reactivity demonstrates a versatile and
efficient strategy for the synthesis of substituted aromatic molecules with fundamentally
different selectivity constraints from the numerous known synthetic methods that rely on
substitution of a preexisting aromatic ring.
P
henols are common precursors and core
structures of industrial chemicals rang-
ing from pharmaceuticals to polymers. The
introduction of chemical functional groups with
specific patterns around the aromatic ring rep-
resents a key challenge in the preparation of these
molecules (1). Electrophilic aromatic substitutions
are classical chemical reactions that remain among
the most versatile methods for the synthesis of
substituted phenols; however, strong electronic
directing effects associated with these reactions
limit their utility to the preparation of ortho- and
para-substituted derivatives. This limitation has
inspired extensive efforts to identify complemen-
tary routes to substituted phenols, such as a recent
two-step arene C–H borylation/oxidation pro-
cedure for the introduction of a hydroxyl group
into an aromatic ring, guided by steric rather
than electronic effects (2). Recent advances in
palladium-catalyzed aerobic oxidation reactions
(3–5) suggested to us that diverse phenol deriv-
atives, including those with meta substitution,
could be accessed by dehydrogenation of cyclo-
hexanones via sequential Pd-mediated C–H
activation/b-hydride elimination steps, followed by
tautomerization of the resulting dienone product
(Fig. 1A). This strategy is appealing because
Pd
II
–hydride intermediates formed in this mech-
anism could be oxidized by molecular oxygen
(6, 7), thereby enabling the overall process to be
catalytic in Pd with water as the sole by-product
(Fig. 1B). Successful catalysts for this class of
reactions could find broad utility owing to the nu-
merous straightforward chemical reactions that
provide access to substituted cyclohexanones, in-
cluding enolate arylation and alkylation meth-
ods, conjugate addition to cyclohexenones, and
Robinson annulation and Diels-Alder reactions
(Fig. 1C).
The preparation of phenols from ketone pre-
cursors have been explored previously (8–16).
Condensation reactions of acyclic ketones, for
example, with b-ketoaldehydes or b-diketones,
enable direct access to substituted phenols (8),
but low product yields, limited access to starting
materials, and/or formation of isomeric products
Department of Chemistry, University of Wisconsin–Madison,
1101 University Avenue, Madison, WI 53706, USA.
*To whom correspondence should be addressed. E-mail:
stahl@chem.wisc.edu
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 209
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

have restricted the utility of these procedures.
Methods for formation of phenols via dehydro-
genation of cyclohexenones have been pursued
(8–13), but reactions of this type typically use
undesirable stoichiometric reagents, such as DDQ
(2,3-dichloro-5,6-dicyano-1,4-benzoquinone) (17);
use stepwise procedures, such as bromination/
dehydrobromination; or require harsh reaction
conditions (≥200°C) that limit functional group
compatibility. In contrast, no effective methods
for dehydrogenation of substituted cyclohexanones
exist, with relevant precedents almost exclusively
limited to reactions of unsubstituted cyclohexa-
none and yields of ≤30% (mostly <5%) (12–16).
Iridium complexes bearing tridentate “pincer”
ligands are among the most effective catalysts
for the dehydrogenation of saturated hydrocar-
bons (18, 19). These reactions are typically car-
ried out in the presence of a sacrificial hydrogen
acceptor, such as norbornene or tert-butylethylene,
and a recent investigation of the reaction of an
Ir-pincer complex with cyclohexanone resulted in
stoichiometric dehydrogenation. Catalytic turn-
over was inhibited by the formation of an Ir-
phenoxide product, ( pincer)Ir(H)(OPh) (20).
In order to explore prospects for the proposed
Pd-catalyzed dehydrogenation methods, we inves-
tigated the reactivity of 3-methylcyclohexenone
(1a) and 3-methylcyclohexanone (1b) under 1 atm
of O
2
with various Pd
II
catalysts. Preliminary anal-
ysis of Pd
II
sources, solvents, and reaction condi-
tions revealed that 3-methylphenol (meta-cresol)
could be obtained in modest yields from 1a
and 1b (28 to 52%) with 3 mol % Pd(OAc)
2
or
Pd(TFA)
2
(OAc is an acetate group and TFA is
trifluoroacetate) in dimethylsulfoxide (DMSO)
as the solvent under 1 atm of O
2
at 80°C (Fig. 2,
entries 1 and 2) (21, 22). Brønsted bases, such as
sodium acetate, and traditional monodentate and
bidentate nitrogen ligands (pyridine, bipyridine,
and phenanthroline) failed to increase the yield of
the phenol product (entries 3 to 6; for expanded
screening results, see table S2). The use of 4,5-
diazafluorenone (23) led to a substantial increase
in the yield of m-cresol from 1a (78%, entry 7);
however, conversion to phenol from the satu-
rated cyclohexanone 1b was still unsatisfactory.
Both key steps in the substrate oxidation sequence,
C–H activation and b-hydride elimination (Fig.
1A), should benefit from a more electrophilic
catalyst, and recent success with 2-fluoropyridine
(
2F
py) as an electron-deficient ligand in aerobic
oxidative coupling reactions (24) prompted us to
evaluate ligands of this type in the dehydrogenation
reactions (entries 8 to 15). The use of
2F
py as a
ligand led to a modest improvement in the yield
of 2 from 1b (44%, entry 10); however, the best
results were obtainedwhen2-(N,N-dimethylamino)
pyridine (
2NMe2
py) was used in combination with
p-toluenesulfonic acid (TsOH) (entry 18). We
speculate that the improved results obtained
with the
2NMe2
py/ TsOH combination relative to
those with
2NMe2
py alone (compare entries 15
and 18) reflect the ability of TsOH to protonate
the tertiary amine of the coordinated pyridine
and thereby afford a more electron-deficient pyr-
idine ligand. The position of the dimethylamino
group on the pyridine ligand is important, as
revealed by the inferior results obtained with
4-(N,N- dimethylamino)pyridine in place of the
2-substituted ligand (entries 19 and 20). Com-
mon heterogeneous palladium catalysts failed
to afford the desired m-cresol (entries 21 to 24).
The optimized catalytic conditions (Fig. 2, en-
try 18) proved to be effective in the preparation of
a number of substituted phenol derivatives (Fig. 3).
Varying the position of the methyl substituent on
the cyclohexanone had little effect on the out-
come of the reaction; the corresponding ortho-,
meta-, and para-cresols were each obtained in
good yield (Fig. 3 entries 1 to 3). Aryl-substituted
phenols, including a number of meta-substituted
derivatives, were accessed via the dehydrogena-
tion of the corresponding arylcyclohexanone de-
rivatives (entries 4 to 11). The 3-arylcyclohexanones
were readily obtained via conjugate addition of
aryl boronic acids to cyclohexenone, and both
electron-donating and electron-withdrawing groups
were tolerated in the dehydrogenation reaction.
Substrates with aryl groups bearing a para bro-
mide or iodide afforded only low yields of the
desired phenol (28% and 16%, respectively;
not shown in Fig. 3). Diels-Alder cycloaddition
and Robinson annulation represent classical
and widely used synthetic organic reactions that
provide efficient routes to cyclohexenones bear-
ing multiple substituents on the six-membered
ring. The aerobic oxidation method described
here provides an attractive alternative to the use
of stoichiometric reagents, such as DDQ, which
have been used previously in the dehydrogena-
tion of cyclohexenones (17), and such substrates
underwent very effective dehydrogenation under
the optimized catalytic conditions, including those
bearing alkyl, aryl, and/or ester substituents (en-
tries 12 to 17).
The 3,5-disubstituted cyclohexenones were pur-
sued further, because the corresponding phenol
derivatives exhibit important biological activity
and products of this type cannot be readily prepared
by classical aromatic-substitution, metal-catalyzed
cross-coupling, or related synthetic methods. As a
representative example, O-terphenylcarbamate 5
was recently identified as a potent in vitro alloste-
ric inhibitor of the human luteinizing hormone
Fig. 1. Strategy for the synthesis of phenols via aerobic oxidative dehydrogenation of cyclo-
hexanone derivatives. (A) Stepwise sequence for Pd-mediated dehydrogenation of cyclohexanone.
(B) Catalytic mechanism whereby cyclohexanone dehydrogenation can be achieved with O
2
as the
terminal oxidant. (C) Representative synthetic methods that afford facile access to substituted
cyclohexanone derivatives. L, a ligand; M, metal; R, organic substituent; X, halide or pseudohalide.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 210
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

receptor, which has been implicated in fertility
and ovarian cancer (25). This compound was pre-
pared by traditional Suzuki coupling methods
using 3,5-dibromophenol as the starting mate-
rial; however, introduction of the unsymmet-
rical aryl substitution pattern results in low yields
of the biaryl intermediate 6 and the desired 3,5-
diarylphenol 4 (30%and 62%yields, respectively;
Fig. 4). In contrast, the 3,5-diarylcyclohexenone
Fig. 2. Catalyst optimization for
aerobic oxidative dehydrogenation
of cyclohexanone derivatives 1a
and 1b. Reaction conditions are
as follows: cyclic ketone(1.0mmol),
PdX
2
(0.03 mmol), ligand/ TsOH
(mol %indicated), DMSO(0.4 ml),
80°C, 1atmO
2
, 24hours. Et, ethyl
group; Me, methyl group; TsOH,
p-toluenesulfonic acid. Yields de-
termined by gas chromatography.
Fig. 3. Palladium-catalyzed aerobic dehydrogenation of cyclic ketones. Reaction conditions are as follows: cyclic ketone (1.0 mmol), Pd(TFA)
2
(0.03 or 0.05 mmol),
2-(N,N-dimethylamino)pyridine (0.06 or 0.10 mmol), TsOH (0.12 or 0.20 mmol), DMSO (0.4 ml), 1 atm O
2
, 80°C, 24 hours. Isolated product yields are reported.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 211
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

derivative 3 is obtained readily from very inex-
pensive starting materials (4-methylacetophenone,
benzaldehyde, and acetone) via sequential aldol
condensation and Robinson annulation. Subse-
quent Pd-catalyzed dehydrogenation of 3 af-
forded phenol 4 in excellent yield.
Preliminary mechanistic analysis of these de-
hydrogenation reactions was performed by mon-
itoring the conversion of cyclohexanone to phenol
by gas chromatography. The kinetic time course
revealed the formation and disappearance of the
partially dehydrogenated intermediate, cyclohex-
enone (Fig. 5A). This result is consistent with the
overall catalytic mechanism in Fig. 1B, in which
the substrate dissociates from the catalyst after
each dehydrogenation step. A fit of the kinetic
data based on a simple sequential reaction model
reveals that the two dehydrogenation steps have
similar rate constants, k
1
= 0.12(T0.02) hour
−1
and
k
2
= 0.33(T0.04) hour
−1
(Fig. 5B). The dehydro-
genation of cyclohexenone, monitored indepen-
dently, exhibits a rate constant somewhat lower
than that obtained from the fit of the sequential
reaction, k
2
' = 0.19(T0.02) hour
−1
. Accurate in-
terpretation of these results will require further
investigation; however, the higher concentration of
cyclohexenone in the latter reaction may slowthe
catalytic turnover via alkene coordination to Pd.
The joint application of catalyst screening
and mechanistic studies will play an important
role in extending these reactions to different pro-
duct classes. For example, catalysts that can
effect the first step substantially more rapidly
than the second step (k
1
:k
2
> 10:1) would enable
selective formation of the enone products rather
than phenols. Moreover, it should be possible to
develop efficient catalysts for dehydrogenative
aromatization of other substrate classes, such as
substituted cyclohexenes (26–28). Substrates of
this type are readily accessed via Diels-Alder
cycloaddition reactions, and their dehydrogen-
ation could proceed via sequential allylic C–H
activation/b-hydride elimination steps (Fig. 5C).
Toward this end, preliminary studies reveal that
cyclohexene derivative 7 undergoes efficient
dehydrogenation to the trisubstituted arene 8 in
near-quantitative yield.
Methods for selective dehydrogenation of sat-
urated carbon-carbon bonds represent an im-
portant class of C–Hfunctionalization (18, 29), and
the reactions presented above highlight the pro-
spective utility of such methods in the synthesis of
substituted aromatic molecules. These reactions
achieve high conversions and product yields, they
are capable of using O
2
as the hydrogen acceptor,
Fig. 4. Application of palladium-catalyzed aerobic oxidative dehydrogenation in the preparation of a terphenyl-derived allosteric inhibitor of the luteinizing
hormone receptor. Ph, phenyl group.
Fig. 5. (A) Kinetic profile of Pd
II
-catalyzed aerobic dehydrogenation of cyclohexanone and cyclohexenone,
showing the formation and decay of cyclohexenone as an intermediate in the reaction. Reaction condi-
tions are as follows: cyclohexanone (0.5 mmol), Pd(TFA)
2
(0.025 mmol), 2-(N,N-dimethylamino)pyridine
(0.05 mmol), TsOH (0.10 mmol), DMSO (0.5 ml), 1 atm O
2
, 80°C. Error bars represent standard devia-
tions from five independent measurements. (B) Comparison of the rate constants obtained for the dehy-
drogenation steps in the sequential conversion of cyclohexanone to phenol and in the direct dehydrogenation
of cyclohexenone. (C) General strategy and a specific example of Pd-catalyzed aerobic dehydrogenation
of cyclohexenes.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 212
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

and the catalyst tolerates useful substrate func-
tional groups, including aromatic and heteroatom
substituents. With the development of improved
methods for safe and scalable aerobic oxidation
reactions (30), dehydrogenation methods of this
type could have an important impact on laboratory-
and industrial-scale chemical synthesis.
References and Notes
1. J. H. P. Tyman, Synthetic and Natural Phenols (Elsevier,
New York, 1996).
2. R. E. Maleczka Jr., F. Shi, D. Holmes, M. R. Smith 3rd,
J. Am. Chem. Soc. 125, 7792 (2003).
3. S. S. Stahl, Angew. Chem. Int. Ed. 43, 3400 (2004).
4. K. M. Gligorich, M. S. Sigman, Chem. Commun. 2009,
3854 (2009).
5. X. Chen, K. M. Engle, D.-H. Wang, J.-Q. Yu, Angew.
Chem. Int. Ed. 48, 5094 (2009).
6. M. M. Konnick, S. S. Stahl, J. Am. Chem. Soc. 130, 5753
(2008).
7. M. C. Denney, N. A. Smythe, K. L. Cetto, R. A. Kemp,
K. I. Goldberg, J. Am. Chem. Soc. 128, 2508 (2006).
8. P. Bamfield, P. F. Gordon, Chem. Soc. Rev. 13, 441
(1984).
9. E. C. Horning, M. G. Horning, J. Am. Chem. Soc. 69,
1359 (1947).
10. P. P. Fu, R. G. Harvey, Chem. Rev. 78, 317 (1978).
11. T. Moriuchi, K. Kikushima, T. Kajikawa, T. Hirao,
Tetrahedron Lett. 50, 7385 (2009).
12. C. S. Yi, D. W. Lee, Organometallics 28, 947
(2009).
13. P. F. Schuda, W. A. Price, J. Org. Chem. 52, 1972
(1987).
14. J. Muzart, J. P. Pete, J. Mol. Catal. 15, 373 (1982).
15. T. T. Wenzel, J. Chem. Soc. Chem. Commun. 1989, 932
(1989).
16. J. Muzart, Eur. J. Org. Chem. 2010, 3779
(2010).
17. D. R. Buckle, in Encyclopedia of Reagents for Organic
Synthesis, D. Crich, Ed. (Wiley, New York, 2010).
18. J. Choi, A. H. R. MacArthur, M. Brookhart, A. S. Goldman,
Chem. Rev. 111, 1761 (2011).
19. R. Ahuja et al., Nat. Chem. 3, 167 (2011).
20. X. W. Zhang, D. Y. Wang, T. J. Emge, A. S. Goldman,
Inorg. Chim. Acta 369, 253 (2011).
21. R. C. Larock, T. R. Hightower, J. Org. Chem. 58, 5298
(1993).
22. R. A. T. M. van Benthem, H. Hiemstra, J. J. Michels,
W. N. Speckamp, J. Chem. Soc. Chem. Commun. 1994,
357 (1994).
23. A. N. Campbell, P. B. White, I. A. Guzei, S. S. Stahl, J. Am.
Chem. Soc. 132, 15116 (2010).
24. Y. Izawa, S. S. Stahl, Adv. Synth. Catal. 352, 3223
(2010).
25. L. H. Heitman et al., J. Med. Chem. 52, 2036
(2009).
26. R. A. Sheldon, J. M. Sobczak, J. Mol. Catal. 68,
1 (1991).
27. J. E. Bercaw, N. Hazari, J. A. Labinger, J. Org. Chem. 73,
8654 (2008).
28. J. E. Bercaw, N. Hazari, J. A. Labinger, P. F. Oblad,
Angew. Chem. Int. Ed. 47, 9941 (2008).
29. G. E. Dobereiner, R. H. Crabtree, Chem. Rev. 110, 681
(2010).
30. X. Ye, M. D. Johnson, T. Diao, M. H. Yates, S. S. Stahl,
Green Chem. 12, 1180 (2010).
Acknowledgments: We are grateful to the NIH
(RC1-GM091161), Mitsubishi Chemical Corporation,
and NSF (CHE-1041934 to D.P.) for financial
support of this work.
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1204183/DC1
Materials and Methods
Figs. S1 to S20
Tables S1 and S2
Characterization Data of New Compounds
References (31–43)
11 February 2011; accepted 1 June 2011
Published online 9 June 2011;
10.1126/science.1204183
High Pre-Eruptive Water Contents
Preserved in Lunar Melt Inclusions
Erik H. Hauri,
1
* Thomas Weinreich,
2
Alberto E. Saal,
2
Malcolm C. Rutherford,
2
James A. Van Orman
3
The Moon has long been thought to be highly depleted in volatiles such as water, and indeed
published direct measurements of water in lunar volcanic glasses have never exceeded 50 parts per
million (ppm). Here, we report in situ measurements of water in lunar melt inclusions; these
samples of primitive lunar magma, by virtue of being trapped within olivine crystals before
volcanic eruption, did not experience posteruptive degassing. The lunar melt inclusions contain
615 to 1410 ppm water and high correlated amounts of fluorine (50 to 78 ppm), sulfur (612 to
877 ppm), and chlorine (1.5 to 3.0 ppm). These volatile contents are very similar to primitive
terrestrial mid-ocean ridge basalts and indicate that some parts of the lunar interior contain as
much water as Earth’s upper mantle.
T
he Moon is thought to have formed in a
giant impact collision between a Mars-
sized object and an early-formed proto-
Earth (1). Though all of the inner planets,
including Earth, are depleted in water and other
volatiles when compared with primitive meteor-
ites, the more extreme depletion of volatiles in
lunar volcanic rocks has long been taken as key
evidence for a giant impact that resulted in high-
temperature catastrophic degassing of the mate-
rial that formed the Moon (2, 3). However, recent
work on rapidly quenched lunar volcanic glasses
has detected the presence of water dissolved in
lunar magmas at concentrations up to 46 parts per
million (ppm) (4), and water contents of lunar
apatite grains from mare basalts are consistent
with similarly minor amounts of water in prim-
itive lunar magmas (5–7). These results indicate
that the Moon is not a perfectly anhydrous plan-
etary body and suggest that some fraction of the
Moon’s observed depletion in highly volatile el-
ements may be the result of magmatic degassing
during the eruption of lunar magmas into the near-
vacuum of the Moon’s surface.
To bypass the process of volcanic degassing,
we conducted a search for lunar melt inclusions
in Apollo 17 sample 74220, a lunar soil contain-
ing ~99% high-Ti volcanic glass beads, the so-
called orange glass with 9 to 12 weight % (wt %)
TiO
2
(8). Melt inclusions are small samples of
magma trapped within crystals that grow in the
magma before eruption. By virtue of their en-
closure within their host crystals, melt inclusions
are protected from loss of volatiles by degassing
during magma eruption. Melt inclusions have
been used for decades to determine pre-eruptive
volatile contents of terrestrial magmas from sub-
duction zones (9, 10), hotspots (11, 12), and mid-
ocean ridges (13, 14), as well as volatile contents
of martian magmas (15, 16). Using standard pet-
rographic methods, we identified nine inclusion-
bearing olivine crystals (Fig. 1) and analyzed melt
inclusions hosted within (17). The measured wa-
ter contents of the melt inclusions range from
615 ppm to a maximum of 1410 ppm (Fig. 2);
these water contents are up to 100 times as high
as the water content of the matrix glass surround-
ing the olivine crystals (6 to 30 ppm H
2
O) and
the centers of individual volcanic glass beads
from the same sample (4). The melt inclusions
also contain high concentrations of fluorine (50
to 78 ppm), sulfur (612 to 877 ppm), and chlorine
(1.5 to 3.0 ppm) that are 2 to 100 times as high as
those of the matrix glasses and individual glass
beads from this sample (Fig. 3). Volatile contents
corrected for postentrapment crystallization are
on average 21% lower than the measured con-
centrations (17) and represent the best estimate
of the pre-eruptive concentrations of volatiles
in the 74220 magma.
There are few descriptions in the literature of
melt inclusions contained within olivine from
lunar samples, but these existing observations
provide important context for the volatile abun-
dances we have observed. Roedder and Weiblen
(18–20) noted the presence of silicate melt in-
clusions in the first samples returned from the
Apollo 11, Apollo 12, and Luna 16 missions and
reported that many primary melt inclusions con-
tained a vapor bubble, requiring dissolved vola-
tiles to have been present in the melt at the time
it was trapped within the host crystal. Klein and
Rutherford (21) and Weitz et al. (22) found sul-
fur contents of 600 to 800 ppm, similar to our
1
Department of Terrestrial Magnetism, Carnegie Institution
of Washington, Washington, DC 20015, USA.
2
Department of
Geological Sciences, Brown University, Providence, RI 02912,
USA.
3
Department of Geological Sciences, Case Western Re-
serve University, Cleveland, OH 44106, USA.
*To whom correspondence should be addressed. E-mail:
ehauri@ciw.edu
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 213
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

measurements, and Cl contents of <50 ppm that
were limited by electron microprobe detection
limits. Reheated Apollo 12 melt inclusions, con-
taining medium-Ti magmas (5 to 6 wt % TiO
2
),
show sulfur contents that are 20% higher than
our data on average (23). In our data set, we
observe a correlation of all the volatiles with
each other (Fig. 3), pointing toward the degassed
compositions of the matrix glass rinds and vol-
canic glass beads (4).
The most important aspect of our volatile data
on lunar melt inclusions is their similarity to melt
inclusions from primitive samples of terrestrial
mid-ocean ridge basalts (MORBs), like those
recovered from spreading centers located within
transform faults (24); the melt inclusions from
74220 are markedly similar to melt inclusions
from the Siqueiros Fracture Zone on the East
Pacific Rise, some of the most primitive mid-
ocean ridge magmas that have been measured
(Fig. 3). These similarities suggest that the vol-
atile signature of the lunar mantle source of the
high-Ti melt inclusions is very similar to that of
the upper mantle source of MORB.
It is important that we have made these
measurements on inclusions from olivine crys-
tals contained within primitive lunar volcanic
glasses. These inclusions were quenched within
minutes after their eruption (4), providing min-
imal opportunity for posteruptive hydrogen dif-
fusion out of the inclusions and affording a
direct H
2
O measurement on primary lunar mag-
ma samples that have not experienced poster-
uptive degassing and associated loss of volatiles.
The water concentrations that we measured are
20 to 100 times as high as previous direct mea-
surements of the lunar glass beads from this
same sample, which was estimated to have suf-
fered 95 to 98% loss of H
2
O via degassing (4),
and they are higher than estimates derived from
lunar apatite measurements, which require a 95
to 99% correction for fractional crystallization
to estimate primary magma volatile contents
(5, 6). Our results are direct measurements on
primary lunar magma compositions that require
no such extrapolations.
Our melt inclusion data allow us to place
some constraints on the volatile content of the
lunar mantle source that generated the high-Ti
picritic magmas. Using the most water-rich melt
inclusion composition after correction for post-
entrapment crystallization, and an estimation that
the high-Ti magmas originated from 5 to 30%
batch partial melting with partitioning similar to
that of terrestrial mantle-derived melts (17), we
estimate lunar mantle volatile concentrations of
79 to 409 ppm H
2
O, 7 to 26 ppm F, 193 to 352
ppm S, and 0.14 to 0.83 ppmCl. These estimates
overlap most estimates for the volatile content of
the terrestrial MORB mantle (24–27) and are
much higher than previous estimates for the lunar
mantle based on the volatile content of lunar
apatite (5, 6) and the variation of Cl isotopes in
lunar rocks (28), including the sample 74220 that
we have studied here. The melt inclusions indi-
cate definitively that some reservoirs within the
interiors of Earth and the Moon not only have
similar water contents, but also similar contents
of fluorine, sulfur, and chlorine associated with
this water, a volatile abundance signature shared
by both bodies.
These results show that the Moon is the only
planetary object in our solar system currently
identified to have an internal reservoir with a
volatile content similar to that of Earth’s upper
mantle, and that previous estimates of the lunar
inventory for highly volatile elements are biased
to low concentrations owing to the degassed na-
ture of lunar samples thus far studied. The Moon
has erupted a wide variety of magmas during its
history, and it remains to be seen whether other
A
B
F E D
C
16
OH/
30
Si
Fig. 2. (A to F) NanoSIMS scanning isotope images of olivines A1, A2, N3, N6, N8, and N9 fromApollo 17
sample 74220, showing the distribution of water within melt inclusions from the olivine grains shown in
Fig. 1. The images showthe distribution of the isotope ratio
16
OH/
30
Si indicated by the color scale shown
in (A), which ranges from dark regions corresponding to low
16
OH/
30
Si ratios (e.g., olivine surrounding
melt inclusions), to red regions within melt inclusions with
16
OH/
30
Si ratios approaching 0.25
(corresponding to ~1400 ppm H
2
O). The color scale is the same in all images, and all images showa scale
bar of 1 mm. Rectangular areas are regions of interest within which each isotope ratio is calculated and
converted to a concentration.
A C
E
B
D F
Fig. 1. (A to F) Optical photographs of olivines A1, A2, N3, N6, N8, and N9 from Apollo 17 sample
74220. Inclusions within circles indicate the inclusions that were imaged in Fig. 2. Scale bars are 10 mm
in all photos.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 214
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

lunar mantle sources are as volatile rich as the
source of Apollo 17 high-Ti magmas. Nevertheless,
the hydrated nature of at least part of the Moon’s
interior is a result that is not consistent with the
notion that the Moon lost its entire volatile in-
ventory to the vacuum of space during degassing
after a high-energy giant impact, which would
be expected to leave a highly desiccated lunar
interior.
If the bulk of the lunar interior has a volatile
content similar to our estimate for the high-Ti
mantle source, then our results present difficulties
for late-accretion models that require volatile
delivery to Earth and the Moon after their for-
mation, because these two bodies have very dif-
ferent accretion cross sections that would predict
different internal volatile contents. An Earth-Moon
similarity in volatiles could indicate that chemical
exchange of even the most volatile elements be-
tween the molten Earth and the proto-lunar disc
might have been pervasive and extensive, result-
ing in homogenization at the very high temper-
atures expected after a giant impact; this could
have been aided by the presence of a high-
temperature convective atmospheric envelope sur-
rounding Earth and the proto-lunar disc as the
Moon solidified (29). Alternatively, it is con-
ceivable that a portion of the lunar interior
escaped the widespread melting expected in the
aftermath of a giant impact and simply inherited
the inventory of water and other volatiles that is
characteristic of Earth’s upper mantle. Any model
for the formation of Earth-Moon system must
meet the constraints imposed by the presence of
H
2
O in the lunar interior, with an abundance sim-
ilar to that of Earth’s upper mantle and with a
complement of fluorine, sulfur, and chlorine also
present at terrestrial levels. To the extent that lu-
nar formation models predict very different vol-
atile contents of Earth and the Moon, our results
on the volatile content of lunar melt inclusions
suggest that we lack understanding on some crit-
ical aspects of the physics of planetary moon for-
mation by collisional impact.
Our findings also have implications for the
origin of water ice in shadowed lunar craters,
which has been attributed to cometary and
meteoritic impacts (30). It is conceivable that
some of this water could have originated from
magmatic degassing during emplacement and
eruption of lunar magmas (31). These results also
underscore the importance of pyroclastic volcan-
ic samples in unraveling the history and compo-
sition of the Moon’s interior; indeed, such deposits
have been identified and mapped on the surfaces
of all the terrestrial planets and many satellites.
References and Notes
1. R. M. Canup, Annu. Rev. Astron. Astrophys. 42, 441
(2004).
2. S. R. Taylor, in Origin of the Moon, W. K. Hartmann,
R. J. Phillips, G. J. Taylor, Eds. (Lunar Planetary Institute,
Houston, TX, 1986), pp. 125–143.
3. F. Albarède, Nature 461, 1227 (2009).
4. A. E. Saal et al., Nature 454, 192 (2008).
5. F. M. McCubbin et al., Proc. Natl. Acad. Sci. U.S.A. 107,
11223 (2010).
6. J. W. Boyce et al., Nature 466, 466 (2010).
7. J. P. Greenwood et al., Nat. Geosci. 4, 79 (2011).
8. J. W. Delano, J. Geophys. Res. 91, D201 (1986).
9. K. A. Kelley et al., J. Geophys. Res. 111, B09208
(2006).
10. A. M. Shaw, E. H. Hauri, T. P. Fischer, D. R. Hilton,
K. A. Kelley, Earth Planet. Sci. Lett. 275, 138 (2008).
11. E. H. Hauri, Chem. Geol. 183, 115 (2002).
12. J. C. Lassiter, E. H. Hauri, I. K. Nikogosian,
H. G. Barsczus, Earth Planet. Sci. Lett. 202, 525
(2002).
13. N. Shimizu, Phys. Earth Planet. Inter. 107, 183
(1998).
14. A. M. Shaw, M. D. Behn, S. E. Humphris, R. A. Sohn,
P. M. Gregg, Earth Planet. Sci. Lett. 289, 311
(2010).
15. H. Y. McSween Jr., R. P. Harvey, Science 259, 1890
(1993).
16. L. L. Watson, I. D. Hutcheon, S. Epstein, E. M. Stolper,
Science 265, 86 (1994).
17. Detailed information on melt inclusion identification
and analytical methods used in this study, as well as
all data, are contained in supporting online materials
available at Science Online.
18. E. Roedder, P. W. Weiblen, Proc. Lunar Sci. Conf. 1, 801
(1970).
19. E. Roedder, P. W. Weiblen, Proc. Lunar Sci. Conf. 2, 507
(1971).
20. E. Roedder, P. W. Weiblen, Earth Planet. Sci. Lett. 13,
272 (1972).
21. N. Klein, M. J. Rutherford, Lunar Planet. Sci. 29, 1448
(1998).
22. C. M. Weitz, M. J. Rutherford, J. W. Head III, D. S. McKay,
Meteorit. Planet. Sci. 34, 527 (1999).
23. D. J. Bombardieri, M. D. Norman, V. S. Kamenetsky,
L. V. Danyushevsky, Meteorit. Planet. Sci. 40, 679 (2005).
24. A. E. Saal, E. H. Hauri, C. H. Langmuir, M. R. Perfit,
Nature 419, 451 (2002).
25. A. Jambon, J. L. Zimmermann, Earth Planet. Sci. Lett.
101, 323 (1990).
26. P. Michael, Earth Planet. Sci. Lett. 131, 301 (1995).
27. J. E. Dixon, L. Leist, C. Langmuir, J. G. Schilling,
Nature 420, 385 (2002).
28. Z. D. Sharp, C. K. Shearer, K. D. McKeegan, J. D. Barnes,
Y. Q. Wang, Science 329, 1050 (2010).
29. K. Pahlevan, D. J. Stevenson, Earth Planet. Sci. Lett. 262,
438 (2007).
30. W. C. Feldman et al., J. Geophys. Res. 106, 23231
(2001).
31. A. P. S. Crotts, Astrophys. J. 687, 692 (2008).
Acknowledgments: This work was supported by the
Carnegie Institution of Washington, the NASA LASER
and Cosmochemistry programs, the NASA Lunar
Science Institute, and the NASA Astrobiology Institute.
We thank J. Wang for NanoSIMS assistance, and
L. Nittler and Z. Peeters for help with image
processing software.
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1204626/DC1
Materials and Methods
Figs. S1 and S2
Tables S1 and S2
References (32–37)
21 February 2011; accepted 10 May 2011
Published online 26 May 2011;
10.1126/science.1204626
Fig. 3. (A to C) Volatile abun-
dances for lunar melt inclusions
(orange circle with black rims) and
matrix glasses (orange circles) from
Apollo 17 sample 74220. Melt in-
clusions show the highest concen-
trations (>600 ppm H
2
O) whereas
matrix glasses show the lowest con-
centrations due to degassing (≤30
ppm H
2
O). The black curves show
lunar magma degassing trends,
scaled from the volatile-volatile
correlations observed in core-rim
NanoSIMS data on a lunar glass
bead reported by Saal et al. (4);
the core-rim data were scaled by
multiplying the originally reported
data for each element, by the ratio
of the highest melt inclusion com-
position to that of the core com-
position reported in table 2 of (4).
The gray field surrounds data for
melt inclusions from the Siqueiros
Fracture Zone on the East Pacific
Rise, as an example of depleted
MORB (24).
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0 500 1000 1500
H
2
O ppm
C
l

p
p
m
0
200
400
600
800
1000
S

p
p
m
0
20
40
60
80
100
120
140
F

p
p
m
A
B
C
Siqueiros MORB
Siqueiros MORB
Siqueiros MORB
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 215
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Running with the Red Queen:
Host-Parasite Coevolution Selects
for Biparental Sex
Levi T. Morran,* Olivia G. Schmidt, Ian A. Gelarden, Raymond C. Parrish II, Curtis M. Lively
Most organisms reproduce through outcrossing, even though it comes with substantial costs. The
Red Queen hypothesis proposes that selection from coevolving pathogens facilitates the persistence
of outcrossing despite these costs. We used experimental coevolution to test the Red Queen
hypothesis and found that coevolution with a bacterial pathogen (Serratia marcescens) resulted in
significantly more outcrossing in mixed mating experimental populations of the nematode
Caenorhabditis elegans. Furthermore, we found that coevolution with the pathogen rapidly drove
obligately selfing populations to extinction, whereas outcrossing populations persisted through
reciprocal coevolution. Thus, consistent with the Red Queen hypothesis, coevolving pathogens can
select for biparental sex.
O
utcrossing (mating between different in-
dividuals) is the most prevalent mode of
reproduction among plants and animals.
The maintenance of outcrossing on such a large
scale strongly suggests that there is a selective ad-
vantage for outcrossing relative to self-fertilization
or asexual reproduction. Nonetheless, the preva-
lence of outcrossing is puzzling, because it often
incurs costs that are not associated with uni-
parental modes of reproduction (1–3). For exam-
ple, many outcrossing species produce males
that facilitate outcrossing but are incapable of
bearing offspring themselves, resulting in the
“cost of males.” Every male takes the place of an
offspring-bearing progeny (female or hermaph-
rodite) that could have been produced (2). The
systematic loss of offspring-bearing progeny can
reduce the numerical contribution of a lineage
by as much 50% (2). Therefore, the selective ben-
efits of outcrossing must more than compensate
for this fitness deficit to achieve a high frequency
in nature.
One selective benefit of outcrossing, relative
to self-fertilization, is the capability to produce
offspring with greater fitness under novel envi-
ronmental conditions (4, 5). Outcrossing can in-
crease fitness and accelerate a population’s rate
of adaptation to novel conditions by permitting
genetic exchange between diverse lineages, pro-
moting genetic variation among offspring, and
allowing beneficial alleles to be quickly assembled
into the same genome (6, 7). In contrast, obligate
selfing can impede adaptation by preventing ge-
netic exchange, which results in the loss of within-
lineage genetic variation and ultimately confines
beneficial alleles to a single lineage (8, 9). Under
novel environmental conditions, the benefits of
outcrossing can compensate for the cost of male
production, but these benefits may be short-lived
(5). Outcrossing is less likely to be favored after
populations adapt to a novel environment, as ge-
netic exchange becomes less imperative or per-
haps even deleterious (8, 9). Hence, the long-term
maintenance of outcrossing would seemto require
that populations are constantly exposed to novel
environmental conditions.
The Red Queen hypothesis provides a pos-
sible explanation for the long-term maintenance
of outcrossing. Specifically, under the Red Queen
hypothesis, coevolutionary interactions between
hosts and pathogens might generate ever-changing
environmental conditions and thus favor the long-
term maintenance of outcrossing relative to self-
fertilization (10) or asexual reproduction (11, 12).
The reason is that hosts are under selection to
evade infection by the pathogen, whereas the
pathogen is selected to infect the hosts. Assuming
that some formof genetic matching between host
and pathogen determines the outcome of inter-
actions, pathogen genotypes that infect the most
common host genotypes will be favored by natu-
ral selection (11, 13). This may produce substan-
tial and frequent change in pathogen populations,
thus rapidly changing the environment for the
host population. Under these conditions, outcross-
ing can facilitate rapid adaptation by generating
offspring with rare or novel genotypes, which are
more likely to escape infection by coevolving path-
ogens (10–13). Conversely, selfing and asexual
reproduction generate offspring with little or no
genetic diversity, thus impeding the adaptive pro-
cess and leaving themhighly susceptible to infec-
tion by coevolving pathogens (10–13).
The Red Queen hypothesis has been empir-
ically supported in studies of natural snail popu-
lations, which show that sexual reproduction is
more common where parasites are common and
adapted to infect the local host population (14, 15).
Outcrossing also seems to reduce the degree of
infection relative to biparental inbreeding and
asexual reproduction in fish (16). Finally, the
capability of antagonistic interactions to drive rap-
id evolutionary change has also been determined
for several different systems (17–20). Nonetheless,
direct controlled tests for the effect of coevolution
on the maintenance of sex have proven difficult,
because they require biological systems in which
host and pathogen populations can coevolve for
multiple generations in a manner that selects for
increased infectivity by a pathogen as well as in-
creased resistance (or enhanced avoidance) by
the host. Further, the host species should exhibit
genetic variation in its degree of outcrossing. Thus,
we chose to examine the nematode Caenorhabditis
elegans and its pathogenic bacteria Serratia
marcescens, which exhibit these desired properties.
Populations of the host species, C. elegans,
are composed of males and hermaphrodites. The
hermaphrodites can reproduce through either
self-fertilization or by outcrossing with males (21).
Although usually low (<1% to 30%) (22), out-
crossing rates can be genetically manipulated to
produce either obligately selfing (5, 23) or ob-
ligately outcrossing (5, 24) populations. The path-
ogen, S. marcescens 2170, is highly virulent and
capable of exerting strong selection on C. elegans.
When consumed, live S. marcescens can produce
a systemic infection that kills the nematode with-
in 24 hours (25). This interaction has a heritable
genetic basis (26), which allows for a potential
response to selection. Moreover, C. elegans pop-
ulations are capable of evolving greater fitness
Department of Biology, Indiana University, 1001 East Third
Street, Bloomington, IN 47405, USA.
*To whom correspondence should be addressed. E-mail:
lmorran@indiana.edu
Fig. 1. Wild-type outcross-
ing rates over time. Out-
crossing rates in wild-type
populations were not ma-
nipulatedandfreetoevolve
during the experiment.
The wild-type populations
were exposed to three dif-
ferent treatments: control
(no S. marcescens; dotted
line), evolution(fixedstrain
of S. marcescens; dashed
line), and coevolution (co-
evolving S. marcescens;
solid line) for 30 gener-
ations. Error bars, 2 SEM.
0.2
0.4
0.6
0.8
1
0 4 8 12 16 20 24 28 32
Generation
O
u
t
c
r
o
s
s
i
n
g

R
a
t
e

(
±

2

s
.
e
.
)
0
Control
Evolution
Coevolution
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 216
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

in response to S. marcescens exposure (5), and
S. marcescens can evolve greater infectivity when
successful infection of C. elegans is its only means
of proliferation. Selection for increased infectiv-
ity can be imposed by propagating only those
bacterial cells that have been harvested from the
carcasses of hosts, which were killed by the bacte-
ria within 24 hours of exposure. Therefore, the
C. elegans/S. marcescens system can be used to
generate antagonistic coevolution when a host pop-
ulation and a pathogen population are repeatedly
passaged under selection together, thus permitting
a direct test of the Red Queen hypothesis.
We used experimental coevolution in the
C. elegans/S. marcescens system to test the pre-
diction that antagonistic coevolution between
host and pathogen populations can maintain high
levels of outcrossing despite the inherent cost of
males. We used obligately selfing, wild-type, and
obligately outcrossing populations of C. elegans
with a CB4856 genetic background (5). Where-
as the reproductive modes of the obligately self-
ing and obligately outcrossing populations are
genetically fixed, the wild-type populations can
reproduce by either selfing or outcrossing [the
baseline outcrossing rate is ~20 to 30%(5)], and
the rate of outcrossing can respond to selection
(5). Before the experiment, we mutagenized five
independent replicate populations of each mating
type (obligate selfing, wild-type, and obligate out-
crossing) by exposing them to ethyl methane-
sulfonate (EMS) to infuse novel genetic variation
in each population. The five replicate populations
were then passaged under three different para-
site treatments (table S1): (i) control (no exposure
to S. marcescens), (ii) evolution (repeated expo-
sure toa fixed, nonevolvingstrainof S. marcescens),
and (iii) coevolution. The coevolution treatment in-
volved repeated exposure (30 host generations) to
a potentially coevolving population of S. marcescens,
which was under selection for increased infectiv-
ity. S. marcescens Sm2170 served as the ancestral
strain in the coevolution treatment, as well as the
fixed strain in the evolution treatment.
The results were consistent with the Red
Queen hypothesis. In the coevolution treatment,
all of the obligately selfing populations became
extinct within 20 generations (fig. S1). However,
none of the obligately selfing populations went
extinct in either the evolution treatment or in the
control treatment. In addition, all of the obligately
outcrossing and wild-type populations persisted
throughout the experiment in all three treatment
types (fig. S1). Thus, extinction was only ob-
served in obligately selfing hosts when confronted
with coevolving pathogens.
We also found that the presence of coevolving
S. marcescens selected for and maintained high
levels of outcrossing in wild-type C. elegans pop-
ulations (Fig. 1). Over the first eight generations
of the experiment, outcrossing rates increased
from 20% to more than 70% in both the evo-
lution and coevolution treatments (Fig. 1) (F
2,11
=
8.26; P = 0.006). However, the wild-type popu-
lations consistently exposed to a fixed population
of S. marcescens (evolution treatment) exhibited
a steady decline in outcrossing rates after this ini-
tial increase, eventually returning to control levels
of outcrossing (Fig. 1), as previously observed (5).
In contrast, populations in the coevolution treat-
ment consistently maintained high levels of out-
crossing throughout the experiment, relative to
the control treatment (Fig. 1) (F
1,12
= 209.5; P <
0.0001). Coevolution with S. marcescens, there-
fore, favored the evolution and long-term main-
tenance of higher rates of outcrossing.
As also predicted by the Red Queen hypoth-
esis, outcrossing hosts adapted to changes in the
pathogen population, whereas selfing apparently
prevented an adaptive counter-response. The an-
cestral strain of the obligately selfing hosts suffered
higher mortality rates when exposed to bacteria
from the coevolution treatment than when ex-
posed to either the ancestral bacteria (Fig. 2A)
(c > a: F
1,71
= 21.2; P < 0.0001) or to the nonco-
evolving control bacteria (Fig. 2A) (c > b: F
1,71
=
31.9; P < 0.0001). Therefore, the bacteria in the
coevolution treatment evolved greater infectivity
in response to selection. Further, the obligately
selfing hosts did not adapt to the evolution of
increased infectivity in the bacteria, because
the bacteria from the coevolution treatment in-
duced greater levels of mortality against the worms
after 10 generations of coevolution than against
the ancestral hosts (Fig. 2A) ( f > c: F
1,71
= 69.2;
P < 0.0001). Finally, an increase in mortality
by more than a factor of 3 was observed in the
obligately selfing hosts in only 10 generations
(Fig. 2A) ( f > a: F
1,71
= 173.7; P < 0.0001),
which could explain why these hosts were driven
to extinction.
The pathogens that coevolved with the wild-
type and obligate outcrossing populations also
evolved greater infectivity (Fig. 2, Band C) (i >h:
F
1,104
= 69.5; P< 0.0001; i > g: F
1,104
= 32.9; P <
0.0001; o > n: F
1,60
= 141.1; P < 0.0001; o > m:
F
1,60
=50.9; P< 0.0001). However, the wild-type
and obligately outcrossing populations adapted
to the changes in their respective coevolving path-
ogen populations. Specifically, both the wild-type
and obligately outcrossing populations exhibited
lower mortality rates against the pathogens with
which they were currently evolving than did their
Fig. 2. Coevolutionary dynamics of
hosts and pathogens. We exposed
hosts evolved under the coevolution
treatment and their ancestral popu-
lations (before coevolution) to three
pathogen populations: (i) an ancestor
strain (ancestral to all S. marcescens
used in this study), (ii) a noncoevolv-
ing strain (evolved without selection),
and (iii) their respective coevolving
strain (coevolving with the host pop-
ulation). We evaluated host mortal-
ity after 24 hours of exposure to the
pathogens and present the means
across the replicate host populations.
(A) Three obligately selfing C. elegans
populations persisted beyond 10 host
generations in the coevolution treat-
ment. These populations were assayed
before extinction. (B) All five wild-
type C. elegans populations in the
coevolution treatment and their an-
cestors were assayed at the endpoint
of the experiment (30 host gener-
ations). (C) All five obligately out-
crossing C. elegans populations in the
coevolution treatment and their an-
cestors were assayed at the endpoint
of the experiment. Error bars, 2 SEM.
0
0.2
0.4
0.6
0.8
Ancestral Populations
0
0.2
0.4
0.6
0.8
Ancestor
Non-coevolving
Coevolving
S. marcescens
Obligately Selfing C. elegans
Ancestral Populations Generation 10
“Coevolution” Populations
0
0.2
0.4
0.6
0.8
Obligately Outcrossing C. elegans
Ancestral Populations
H
o
s
t

M
o
r
t
a
l
i
t
y

R
a
t
e

a
t

2
4

H
o
u
r
s

o
f

E
x
p
o
s
u
r
e

(
±

2

s
.
e
.
)
Wildtype (Mixed Mating) C. elegans
A
B
C
Generation 30
“Coevolution” Populations
Generation 30
“Coevolution” Populations
p q r
a b c
h i g
d e f
m n o
j k l
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 217
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

ancestors (Fig. 2, Band C) (i > l: F
1,104
= 27.9; P<
0.0001; o > r: F
1,60
= 166.2; P < 0.0001), thus
indicating reciprocal coevolution in the outcross-
ing host populations. Whereas the obligate selfing
populations in the coevolution treatment became
more infected over time (Fig. 2A), the wild-type
populations maintained the same level of infec-
tivity over the course of the experiment (Fig. 2B)
(g = l: F
1,104
= 0.35; P= 0.554), while the obligate
outcrossing populations were significantly less
infected at the end of the experiment relative to
the beginning (Fig. 2C) (m > r: F
1,60
= 33.1; P <
0.0001). Coupled with the maintenance of high
outcrossing rates in the coevolving wild-type
populations (Fig. 1), these results demonstrate the
ability of antagonistic coevolution to continually
generate novel environmental conditions under
which outcrossing is favored and populations per-
sist when interacting with a virulent pathogen.
A recent host/pathogen coevolution study in
C. elegans further supports the conclusion that
low levels of outcrossing impede the rate of
adaptive evolution. The C. elegans hosts in this
previous study appear to have primarily repro-
duced via self-fertilization and did not evolve
significantly greater resistance to a coevolving
pathogen over 48 generations of selection (27).
Contrary to our study, however, greater out-
crossing rates did not evolve in these mixed-
mating populations in response to the pathogen.
It may be that higher levels of genetic variation
and/or a greater level of pathogen virulence in
our study account for the difference in outcomes.
In summary, we found that obligately selfing
lineages were driven to extinction when con-
fronted with a coevolving parasite. These results
are consistent with the macroevolutionary aspects
of the Red Queen hypothesis, as originally formu-
lated by Van Valen (28). We also found that the
presence of a coevolving pathogen selected for and
maintained high levels of outcrossing in mixed-
mating populations, whereas elevated levels of
outcrossing were not maintained in populations
where the pathogen was not coevolving. These
results are consistent with the microevolutionary
predictions of the Red Queen. Taken together, the
results demonstrate that sex can facilitate adap-
tation to novel environments, but the long-term
maintenance of sex requires that the novelty does
not wear off.
References and Notes
1. G. C. Williams, Sex and Evolution (Princeton University
Press, Princeton, NJ, 1975).
2. J. Maynard Smith, The Evolution of Sex (Cambridge
University Press, Cambridge, UK, 1978).
3. G. Bell, The Masterpiece of Nature: The Evolution and
Genetics of Sexuality (University of California Press,
Berkeley, CA, 1982).
4. G. L. Stebbins, Am. Nat. 91, 337 (1957).
5. L. T. Morran, M. D. Parmenter, P. C. Phillips, Nature 462,
350 (2009).
6. H. J. Muller, Am. Nat. 66, 118 (1932).
7. R. A. Fisher, The Genetical Theory of Natural Selection
(Clarendon Press, Oxford, 1930).
8. R. Lande, D. W. Schemske, Evolution 39, 24 (1985).
9. D. Charlesworth, B. Charlesworth, Annu. Rev. Ecol. Syst.
18, 237 (1987).
10. A. F. Agrawal, C. M. Lively, Evolution 55, 869 (2001).
11. J. Jaenike, Evol. Theory 3, 191 (1978).
12. W. D. Hamilton, Oikos 35, 282 (1980).
13. W. D. Hamilton, R. Axelrod, R. Tanese, Proc. Natl. Acad.
Sci. U.S.A. 87, 3566 (1990).
14. C. M. Lively, Nature 328, 519 (1987).
15. K. C. King, L. F. Delph, J. Jokela, C. M. Lively, Curr. Biol.
19, 1438 (2009).
16. C. M. Lively, C. Craddock, R. C. Vrijenhoek, Nature 344,
864 (1990).
17. E. Decaestecker et al., Nature 450, 870 (2007).
18. B. Koskella, C. M. Lively, Evolution 63, 2213 (2009).
19. J. Jokela, M. F. Dybdahl, C. M. Lively, Am. Nat. 174
(suppl. 1), S43 (2009).
20. S. Paterson et al., Nature 464, 275 (2010).
21. S. Brenner, Genetics 77, 71 (1974).
22. H. Teotónio, D. Manoel, P. C. Phillips, Evolution 60, 1300
(2006).
23. L. M. Miller, J. D. Plenefisch, L. P. Casson, B. J. Meyer,
Cell 55, 167 (1988).
24. T. Schedl, J. Kimble, Genetics 119, 43 (1988).
25. C. L. Kurz et al., EMBO J. 22, 1451 (2003).
26. G. V. Mallo et al., Curr. Biol. 12, 1209 (2002).
27. R. D. Schulte, C. Makus, B. Hasert, N. K. Michiels,
H. Schulenburg, Proc. Natl. Acad. Sci. U.S.A. 107, 7359
(2010).
28. L. Van Valen, Evol. Theory 1, 1 (1973).
Acknowledgments: We thank H. Hundley and R. Matteson
for logistical assistance. We also thank F. Bashey,
L. Delph, P. Phillips, M. Parmenter, the Lively and Hall
laboratories, and two reviewers for helpful comments
and discussion, as well as the Wissenschaftskolleg zu
Berlin for a fellowship to C.M.L. during the preparation
of the manuscript. Funding was provided by the NSF
(DEB-0640639 to C.M.L) and the NIH (1F32GM096482-01
to L.T.M). Nematode strains were provided by the
Caenorhabditis Genetics Center, which is funded by the
NIH National Center for Research Resources (NCRR). Data
deposited at Dryad, 10.5061/dryad.c0q0h.
Supporting Online Material
www.sciencemag.org/cgi/content/full/333/6039/216/DC1
Materials and Methods
Fig. S1
Table S1
References 29 to 31
31 March 2011; accepted 24 May 2011
10.1126/science.1206360
Isolation of Single Human Hematopoietic
Stem Cells Capable of Long-Term
Multilineage Engraftment
Faiyaz Notta,
1,2
* Sergei Doulatov,
1,2
* Elisa Laurenti,
1,2
Armando Poeppl,
1
Igor Jurisica,
3,4
John E. Dick
1,2

Lifelong blood cell production is dependent on rare hematopoietic stem cells (HSCs) to
perpetually replenish mature cells via a series of lineage-restricted intermediates. Investigating
the molecular state of HSCs is contingent on the ability to purify HSCs away from transiently
engrafting cells. We demonstrated that human HSCs remain infrequent, using current purification
strategies based on Thy1 (CD90) expression. By tracking the expression of several adhesion
molecules in HSC-enriched subsets, we revealed CD49f as a specific HSC marker. Single CD49f
+
cells were highly efficient in generating long-term multilineage grafts, and the loss of CD49f
expression identified transiently engrafting multipotent progenitors (MPPs). The demarcation of
human HSCs and MPPs will enable the investigation of the molecular determinants of HSCs,
with a goal of developing stem cell–based therapeutics.
M
ature blood cell lineages are generated
from a network of hierarchically dis-
tinct progenitors that arise from self-
renewing hematopoietic stem cells (HSCs). The
extensive regenerative potential of HSCs makes
them attractive targets for cellular and genetic
therapies. The molecular regulation of specific
HSC properties such as long-term self-renewal is
beginning to be elucidated for murine HSCs (1).
However the biology of human HSCs remains
poorly understood because of their rarity and the
lack of methods to segregate HSCs from multip-
otent progenitors (MPPs) to obtain pure popula-
tions for biological and molecular analysis.
The bulk of HSCs are CD34
+
, as evidenced
by human transplantation and xenograft re-
population assays; however, most CD34
+
cells
are lineage-restricted progenitors and HSCs re-
main rare. HSCs can be enriched further on the
basis of CD45RA (2), Thy1 (3–5), and CD38
(6, 7) expression. Loss of Thy1 expression in the
CD34
+
CD38

CD45RA

compartment of lineage-
depleted cord blood (CB) was recently proposed
to be sufficient to separate HSCs from MPPs
(5). However, more than a third of Thy1

primary
recipients gave rise to engraftment in secondary
animals, raising uncertainty about whether Thy1
can absolutely segregate HSCs from MPPs. To
1
Division of Stem Cell and Developmental Biology, Campbell
Family Institute for Cancer Research/Ontario Cancer Institute,
Toronto, Ontario, Canada.
2
Department of Molecular Genetics,
University of Toronto, Toronto, Ontario, Canada.
3
Ontario
Cancer Institute and Campbell Family Institute for Cancer
Research, Toronto, Ontario, Canada.
4
Departments of Com-
puter Science and Medical Biophysics, University of Toronto,
Toronto, Ontario, Canada.
*These authors contributed equally to this work.
†To whom correspondence should be addressed. Toronto
Medical Discovery Tower, Room 8-301, 101 College Street,
Toronto, Canada M5G 1L7. E-mail: jdick@uhnres.utoronto.ca
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 218
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

resolve the relationship between these two sub-
sets, the number of cells in each subset that are
capable of short-term and long-term engraftment
must be quantified at clonal resolution. We re-
cently optimized the HSC xenograft assay by using
intrafemoral injection into female NOD-scid-
IL2Rgc
−/−
(NSG) mice (8–10). Flow-sorted CB
HSCs (CD34
+
CD38

CD45RA

Thy1
+
; Thy1
+
) (Fig.
1, P5) and MPPs (CD34
+
CD38

CD45RA

Thy1

;
Thy1

) (Fig. 1, P2) fractions were functionally
characterized with our HSC assay. A priori, HSCs
were operationally defined by lymphomyeloid
engraftment that persisted for at least 20 weeks
after transplant. This duration represents a strin-
gent test of long-term repopulation and encom-
passes the total engraftment time of primary
and secondary transplants historically used to
assess the self-renewal capacity of human HSCs
in xenograft models. At nonlimiting cell doses,
recipients of Thy1
+
and Thy1

cells had similar
levels of human chimerism and lineage distri-
bution (injected femur: P = 0.17; Fig. 2, A and B;
fig. S1; and table S1). To assess whether Thy1

cells would persist beyond the allotted 20-week
primary transplant period, we performed sec-
ondary transplants for an additional 12 to 14
weeks. This revealed that Thy1

cells could be
serially transplanted, albeit with lower effici-
ency than Thy1
+
cells (table S2), which is consist-
ent with previous work (5). These data suggest
that cells with extensive self-renewal potential
exist in both Thy1
+
and Thy1

subsets, although
the basis for the disparity in secondary trans-
fer efficiency between these subsets remained
unknown.
We next performed limiting dilution analysis
(LDA) to measure the frequency of HSCs within
Thy1
+
and Thy1

fractions. One in 20 Thy1
+
cells (5%) clonally initiated long-term hemato-
poiesis in NSG mice ascompared to 1 in 100
(1%) Thy1

cells (P = 0.0003, Fig. 2C). Double
sorting and high-stringency sort modes used in
our experimental design ruled out the possibility
that HSC activity from Thy1

cells was due to
residual contamination from Thy1
lo/+
cells (figs.
S2 and S3). The inability of prior studies to detect
engraftment from Thy1

cells was probably due
to the less sensitive xenograft models employed
(3). Thus, although Thy1
+
enriches for HSCs, long-
term repopulating activity persists in the Thy1

fraction previously believed to represent MPPs (5).
To examine the hierarchical relationship be-
tween the Thy1 subsets, we cultured sorted Thy1
+
and Thy1

cells with stroma cells known to ex-
press HSC-supportive ligands (11). Both Thy1
+
and Thy1

cells (>70%) remained CD34
+
CD38

onstromal cultures (fig. S4, column 3). Unexpected-
ly, Thy1

cells consistently generated Thy1
+
cells
on stroma (Fig. 2D, right panel) and also in vivo
within the bone marrow microenvironment of
NSG mice that received transplants (Fig. 2E).
Thy1
+
cells arising fromThy1

cells after culture,
as well as Thy1
+
cells that retained their Thy1
expression, had robust repopulating activity in
NSG mice 20 weeks after transplantation. En-
graftment and lineage potentials were identical
for Thy1
+
cells derived from either Thy1 subfrac-
tion (Fig. 2F and fig. S5). Cells that remained
Thy1

after being cultured on OP9 stroma did
not sustain a long-term graft (Fig. 2F, right panel);
however, they transiently repopulated (fig. S6).
These results demonstrate that the Thy1

com-
partment is heterogeneous and contains a small
fraction with repopulating activity and a larger
fraction with MPP-like activity.
To further purify HSCs in both Thy1
+
and
Thy1

subsets, we searched for a different cell
Fig. 1. Cell sorting scheme
used to isolate human HSCs
and MPPs. Lin

CB cells
were stained with mono-
clonal antibodies against
CD34, CD38, CD45RA, Thy1
(CD90), CD49f antigens,
andthe mitochondrial mem-
branedyeRho. Thefrequen-
cy of each subpopulation
(P1 to P10) is based on the
parent gate shown on the
top right of each plot.
Fig. 2. Functional characterization of the Thy1
+
and Thy1

subsets. (A and B) Engraftment and lineage
potential of Thy1
+
(P5) and Thy1

(P2) cells assessed in NSG mice 20 weeks after transplant (Thy1
+
: n =
65 mice; Thy1

: n = 30 mice, 6 experiments). IF, injected femur; BM, left femur, tibiae; SP, spleen; TH,
thymus. (C) Frequency of long-term repopulating cells within Thy1
+
and Thy1

populations measured by
LDA. (D) Phenotype of Thy1
+
(left) and Thy1

(right) cells cultured on OP9 stroma. Plots are gated on
CD34
+
CD38

cells (figs. S4 and S14). (E) Phenotype of Thy1
+
and Thy1

cells engrafted in NSG mice. (F)
Populations labeled + and – from (D) were isolated from stroma and transplanted into NSG mice (200 to
400 cells, two experiments, n = 10 mice per group). Their engraftment potential (as percent of CD45
+
cells) is shown in the bottom panel. All data are presented as means T SEM. PB, peripheral blood.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 219
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

surface marker. We hypothesized that integrins
would mark human HSCs, because they me-
diate niche interactions and have been used to
isolate murine HSCs and other somatic stem
cells (12, 13). We compared the surface expres-
sion of several adhesion molecules between
HSC-enriched (Thy1
+
) and -depleted (Thy1

)
fractions. Among our candidates, only ITGA6
(integrin a6, termed CD49f ) was differentially
expressed (Fig. 3A and fig. S7), with 50 to 70%
of Thy1
+
cells expressing CD49f versus 10 to
20% of Thy1

cells.
To determine whether human HSCs could
be delineated using CD49f expression, we parti-
tioned Thy1
+
cells into CD49f
+/hi
(here called
Thy1
+
CD49f
+
; Fig. 1, P9) and CD49f
lo/−
(here
called Thy1
+
CD49f

; Fig. 1, P8) subfractions
and evaluated their capacity for long-term multi-
lineage chimerism in NSG recipients. Mean chi-
merism in the injected femur was 6.7-fold higher
for Thy1
+
CD49f
+
than for Thy1
+
CD49f

cells
(22.7% versus 3.4%, P < 0.0001, Fig. 3B, left
panel), and only Thy1
+
CD49f
+
cells could be
serially transplanted (table S3). LDA revealed
that 9.5% (1 in 10.5) of Thy1
+
CD49f
+
cells had
long-termrepopulating activity as compared with
0.9% (1 in 111.3) Thy1
+
CD49f

cells (P = 9.9 ×
10
−9
, Fig. 3C, and tables S1 and S4). Because we
had found that the Thy1

fraction was heteroge-
neous, we tested whether CD49f expression also
marked Thy1

HSCs. Indeed, only Thy1

CD49f
+
cells reconstituted NSG mice 20 weeks after
transplant (Fig. 3B, right panel). LDA indicated
that approximately 4.5% (1 in 22.1) of cells in
this fraction had long-term multilineage engraft-
ment potential as compared to 0.13%(1 in 735.2)
of Thy1

CD49f

cells (Fig. 3Cand table S4). No
difference in lineage potential was observed be-
tween Thy1
+
CD49f
+
and Thy1

CD49f
+
cells
(fig. S8), although recipients of Thy1
+
CD49f
+
cells trended toward higher levels of chimerism
at similar cell doses (table S1). We estimate that
although most human HSCs are Thy1
+
, con-
sistent with prior work, 1 in 5.5 CB HSCs lack
Thy1 expression. These data indicate that hu-
man HSCs are marked by CD49f, and they es-
tablish the existence of Thy1

HSCs.
The absence of long-termgrafts inThy1

CD49f

recipients raised the possibility that the loss of
CD49f demarcated human MPPs in the Thy1

subset (5). To test this idea, we temporally mon-
itored the peripheral blood and marrow of NSG
recipients transplanted with all four Thy1 and
CD49f subsets for 30 weeks. Although levels of
chimerism gradually increased in the peripheral
blood of mice transplanted with CD49f
+
HSC
subsets, engraftment of Thy1

CD49f

cells peaked
between 2 and 4 weeks and then declined (Fig.
3D). The bone marrow of Thy1

CD49f

recip-
ients displayed significantly higher levels of chi-
merism at 2 weeks than did CD49f
+
HSCs in
both the injected femur and noninjected bones,
indicating that Thy1

CD49f

cells have a higher
engraftment and differentiation potential than
HSCs immediately after transplant (Fig. 3E). These
results also rule out the idea that Thy1

CD49f

cells have impaired capacity to home and prolif-
erate in the marrow. B cells, monocytes, gran-
ulocytes, and erythrocytes were detected in the
bone marrow of Thy1

CD49f

mice (Fig. 3E
and fig. S9). HSC-enriched fractions displayed
a delay in engraftment until 4 weeks (Fig. 3F).
The engraftment kinetics of Thy1
+
CD49f

cells
were intermediate to HSC and Thy1

CD49f

subsets (Fig. 3, Dto F). These data demonstrate
that Thy1

CD49f

cells can give rise to all major
hematopoietic lineages but fail to engraft long-
term, indicating that these are bona fide MPPs.
To remain consistent with previous work (5), we
defined transiently engrafting Thy1

CD49f

cells
as MPPs. However, considering that Thy1

CD49f

cells differ from CD49f
+
HSCs solely in the abil-
ity to engraft durably, an alternate interpretation
is that Thy1

CD49f

cells are short-term HSCs.
To provide an independent line of evidence
for distinguishing our functionally defined HSC
and MPP populations, we carried out global
gene expression analysis of sorted CD49f
+
and
Thy1

CD49f

subsets. Unsupervised hierarchi-
cal clustering revealed that the CD49f
+
HSCs
clustered together irrespective of Thy1 status
(fig. S10A). No significant differences in gene
expression were detected between these subsets,
Fig. 3. Human HSCs and MPPs are demarcated by CD49f expression. (A) Cell surface expression of
adhesion molecules (CD49b, CD49d, CD49e, CD49f, CD44, and CD184) measured as the ratio of mean
fluorescence intensity (MFI) between HSC-enriched (Thy1
+
) and -depleted (Thy1

) populations. (B) Mean
engraftment levels of cells fractionated by Thy1 and CD49f expression in NSG mice (n = 4 experiments).
Cell doses and number of mice are shown in table S1. (C) Long-term repopulating cell frequency in
Thy1 and CD49f subsets measured by LDA. (D) Kinetic analysis of peripheral blood engraftment
(measured as percent of CD45
+
cells) by cells fractionated on Thy1 and CD49f after transplantation into
NSG mice (100 to 200 cells per recipient). The legend is consistent with (C). (E and F) Short-term
engraftment by cells fractionated on Thy1 and CD49f at 2 (E) and 4 (F) weeks after transplant. The
human graft consisted of GlyA
+
erythroid (white bars) and CD45
+
myelolymphoid cells (black bars). All
data are presented as means T SEM. *P < 0.05; **P < 0.001. BM, non-injected bones.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 220
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

although Thy1

CD49f
+
HSCs displayed an inter-
mediate pattern to Thy1
+
CD49f
+
HSCs and MPPs
(fig. S10B), consistent with the lower frequency
of HSCs within this population. In contrast, the
Thy1

CD49f

MPPs clustered independently
fromboth CD49f
+
HSCsubsets (fig. S10A). Sev-
enty differentially expressed genes segregated
Thy1

CD49f

MPP versus HSC subsets (fig.
S10B and table S5), consistent with previous
murine studies (14). Stem and committed pro-
genitor cells (15) differed more dramatically (500
to 3000 genes), underscoring the close rela-
tionship between immature cell types. These
findings support our functional delineation of
Thy1
+
CD49f
+
and Thy1

CD49f
+
HSCs as dis-
tinct fromThy1

CD49f

MPPs and identify gene
expression changes associated with the earliest
steps of human HSC differentiation.
Although sorting based on CD49f enables the
highest reported purity of human HSCs, this test
still falls short of the most definitive assessment
of HSC potential: single-cell transplantation. A
single long-term mouse HSC provides lifelong
blood production (16). Because only 9.5% of
Thy1
+
CD49f
+
cells were HSCs by LDA, addi-
tional strategies were needed to efficiently assess
single human HSCs. High efflux of the mitochon-
drial dye rhodamine-123 (Rho) could enrich for
HSCs within the Lin

CD34
+
CD38

fraction (17).
To test whether Rho efflux marked HSCs in
conjunction with Thy1, we transplanted limiting
numbers of Thy1
+
cells sorted based on high
(Thy1
+
Rho
lo
; Fig. 1, P6) and low (Thy1
+
Rho
hi
;
Fig. 1, P7) Rho efflux. Recipients of Thy1
+
Rho
lo
cells exhibited 40-fold higher chimerism in the
injected femur (fig. S11) and displayed twofold
enrichment for HSCs as compared to Thy1
+
alone
(table S1).
We next questioned whether the addition of
Rho to Thy1
+
CD49f
+
would permit robust en-
graftment of single human HSCs. We flow-sorted
single Thy1
+
Rho
lo
CD49f
+
cells and transplanted
them into NSG recipients (Fig. 4A). In our first
experiment, 5 of 18 recipients (28%, Fig. 4B)
transplanted with single cells displayed multi-
lineage chimerism 20 weeks after transplant (Fig.
4C and fig. S12). Serial transfer was successful
in two of four secondary recipients despite the
fact that only 20% of total marrow was used
for transplantation (fig. S13), indicating that
individual Thy1
+
Rho
lo
CD49f
+
cells extensively
self-renew. In a second experiment, we observed
a lower frequency (14%) of single-cell transfer,
perhaps reflecting the genetic heterogeneity of
CB donors (Fig. 4B). Human cell engraftment at
marrow sites distant from the injected femur in-
dicated that single Thy1
+
Rho
lo
CD49f
+
cells could
give rise to systemic grafts (Fig. 4, D to E), ful-
filling a key criterion of HSCs. Based on historic-
al data showing xenotransplantation inefficiency
(18), we are probably underestimating HSC fre-
quency. Engraftment of single Lin

CD34
+
CD38

CD45RA

Thy1
+
Rho
lo
CD49f
+
cells provides
evidence that human HSCs express CD49f.
In this study, we purified human HSCs at
single-cell resolution, separated HSCs fromMPPs,
and identified CD34
+
Thy1

HSCs. The ability to
investigate two highly enriched and related multi-
potent cell populations that differ in their capacity
for self-renewal opens the way for studies to elu-
cidate developmental programs specific to human
HSCs. Because the cell number required for chro-
matin immunoprecipitation coupled with high-
throughput sequencing (ChIP-seq) and DNA
methylation profiling is constantly decreasing
(19–21), it will be critical to subject HSCs and
MPPs to these technologies. Such analyses will
aid in identifying gene regulatory networks that
govern human HSCfunction and in turn facilitate
manipulating and expanding human HSCs ex vivo
to overcome barriers to successful transplantation.
References and Notes
1. M. Sauvageau, G. Sauvageau, Cell Stem Cell 7, 299 (2010).
2. H. Mayani, W. Dragowska, P. M. Lansdorp, Blood 82,
2664 (1993).
3. C. M. Baum, I. L. Weissman, A. S. Tsukamoto,
A. M. Buckle, B. Peault, Proc. Natl. Acad. Sci. U.S.A. 89,
2804 (1992).
4. W. Craig, R. Kay, R. L. Cutler, P. M. Lansdorp, J. Exp. Med.
177, 1331 (1993).
5. R. Majeti, C. Y. Park, I. L. Weissman, Cell Stem Cell 1, 635
(2007).
6. Q. L. Hao, A. J. Shah, F. T. Thiemann,
E. M. Smogorzewska, G. M. Crooks, Blood 86, 3745 (1995).
7. M. Bhatia, J. C. Wang, U. Kapp, D. Bonnet, J. E. Dick,
Proc. Natl. Acad. Sci. U.S.A. 94, 5320 (1997).
8. L. D. Shultz et al., J. Immunol. 174, 6477 (2005).
9. J. L. McKenzie, O. I. Gan, M. Doedens, J. E. Dick, Blood
106, 1259 (2005).
10. F. Notta, S. Doulatov, J. E. Dick, Blood 115, 3704 (2010).
11. H. Ueno et al., Nat. Immunol. 4, 457 (2003).
12. J. Stingl et al., Nature 439, 993 (2006).
13. P. Benveniste et al., Cell Stem Cell 6, 48 (2010).
14. M. J. Kiel et al., Cell 121, 1109 (2005).
15. S. Doulatov et al., Nat. Immunol. 11, 585 (2010).
16. M. Osawa, K. Hanada, H. Hamada, H. Nakauchi, Science
273, 242 (1996).
17. J. L. McKenzie, K. Takenaka, O. I. Gan, M. Doedens,
J. E. Dick, Blood 109, 543 (2007).
18. L. D. Shultz, F. Ishikawa, D. L. Greiner, Nat. Rev.
Immunol. 7, 118 (2007).
19. M. Adli, J. Zhu, B. E. Bernstein, Nat. Methods 7, 615 (2010).
20. A. Goren et al., Nat. Methods 7, 47 (2010).
21. H. Ji et al., Nature 467, 338 (2010).
Acknowledgments: We thank K. Moore and the obstetrics
unit of Trillium Hospital (Mississauga, Ontario) for
providing CB, the Dick Lab and M. Cooper for critical
manuscript review, C. Nostro from the Keller lab for
assistance with immunofluorescence, and P. A. Penttilä
and the University Health Network/Sickkids flow
cytometry facility. This work was supported by Canadian
Institutes for Health Research (CIHR) studentships
(F.N. and S.D.) and Swiss National Science Foundation
and Roche fellowships (E.L.). Grants were received from
the CIHR, Canadian Cancer Society, Terry Fox Foundation,
Genome Canada through the Ontario Genomics
Institute, Ontario Institute for Cancer Research with funds
from the province of Ontario, and a Canada Research
Chair. This research was funded in part by the Ontario
Ministry of Health and Long Term Care (OMOHLTC).
The views expressed do not necessarily reflect those of
the OMOHLTC. Microarray data are available online at
the Gene Expression Omnibus Web page (accession
code GSE29105). F.N., S.D., and J.E.D. designed the
study; F.N., S.D., and A.P. performed experiments;
E.L., S.D., and I.J. analyzed gene expression data; F.N.
prepared the manuscript; S.D. and J.E.D. edited the
manuscript; and J.E.D supervised the study.
Supporting Online Material
www.sciencemag.org/cgi/content/full/333/6039/218/DC1
Materials and Methods
Figs. S1 to S14
Tables S1 to S5
References
3 December 2010; accepted 2 June 2011
10.1126/science.1201219
Fig. 4. Engraftment of single hu-
man HSCs. (A) Experimental design
used to sort and transplant single
P10 (Thy1
+
Rho
lo
CD49f
+
) cells into
NSGmice. (1) Single-cell fluorescence-
activated cell sorting deposition and
microscopic visualization to verify
the presence of a single cell. (2) Up-
take into syringe. (3) Verification of
cell aspiration. (4) Transplant into
NSG mice. (B) Cloning efficiency, (C)
B-lymphoid (CD19
+
) and myeloid
(CD33
+
) lineage distribution, and
(D) mean levels of human chime-
rism in the injected femur and non-
injected bones of single human
Thy1
+
Rho
lo
CD49f
+
HSCs from two
independent CB samples engrafting
NSGmice. (E) Flowcytometric analysis
of three representative mice engrafted
with single Thy1
+
Rho
lo
CD49f
+
cells.
Auto, auto-fluorescence.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 221
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Coupled, Circumferential Motions of
the Cell Wall Synthesis Machinery
and MreB Filaments in B. subtilis
Ethan C. Garner,
1
* Remi Bernard,
2
Wenqin Wang,
3
Xiaowei Zhuang,
4,3,5
*†
David Z. Rudner,
2
*† Tim Mitchison
1
*†
Rod-shaped bacteria elongate by the action of cell wall synthesis complexes linked to
underlying dynamic MreB filaments. To understand how the movements of these filaments relate
to cell wall synthesis, we characterized the dynamics of MreB and the cell wall elongation
machinery using high-precision particle tracking in Bacillus subtilis. We found that MreB and the
elongation machinery moved circumferentially around the cell, perpendicular to its length, with
nearby synthesis complexes and MreB filaments moving independently in both directions.
Inhibition of cell wall synthesis by various methods blocked the movement of MreB. Thus, bacteria
elongate by the uncoordinated, circumferential movements of synthetic complexes that insert
radial hoops of new peptidoglycan during their transit, possibly driving the motion of the
underlying MreB filaments.
T
he shape of most bacteria is maintained by
the cell wall peptidoglycan (PG), a three-
dimensional meshwork composed of gly-
can strands linked by peptide cross-bridges, but
how collections of genes confer defined shapes
and widths to this structure is unclear. The PGis a
single macromolecule, and the mechanisms that
localize the synthesis of this structure dictate the
shape of the organism.
The peptidoglycan elongation machinery
(PGEM) responsible for rod-shaped growth is
composed of synthetic enzymes [called penicillin
binding proteins (PBPs)] and conserved mem-
brane proteins (MreC, MreD, RodA, RodZ). These
proteins interact with MreB (1–4), a distant actin
homolog that assembles into cytoplasmic fila-
ments (5, 6). Depletions of PGEM proteins or
disruption of MreB filaments produces round
cells (7–9). MreB appears to form helical struc-
tures by light microscopy, which have been pro-
posed to organize the PGEM to facilitate the
construction of a rod-shaped cell (10–12).
MreB is dynamic, and moves directionally in
B. subtilis (13, 14) and Caulobacter crescentus
(15). However, it is unclear how this motion
relates to cell wall synthesis. To explore the or-
igin and function of MreB movement, we char-
acterized and compared the relative dynamics of
MreB and the PGEM in B. subtilis.
B. subtilis expresses three MreB paralogs
(MreB, Mbl, and MreBH). We began by imaging
their dynamics with confocal microscopy. Green
fluorescent protein (GFP)-Mbl expressed as the
sole source of Mbl in the cell displayed well-
separated foci that moved linearly across the
cell width (Fig. 1A and movies S1, A and B).
Maximal intensity projections of these movies
revealed closely spaced horizontal bands perpen-
dicular to the cell length, suggesting that the pre-
dominant movement occurred as a rotation around
the cell circumference. Kymographs drawn across
the cell width generated diagonal lines, indicative
of circumferential movement occurring at approx-
imately constant velocity (22 T 4 nm/s, n = 40).
Similar circumferential motion was observed with
GFP fusion proteins to all three MreB paralogs
regardless of expression context (Fig. 1B and
movies S2, A and B). The motions of MreB did
not arise from rotation of the cell itself: Imaging
of MreB and EpsE (a protein associated with the
flagellar bodies traversing the cell wall) demon-
strated that MreB filaments rotate within a static
cell (Fig. 1C and movies S3, A and B).
Previous studies have attributed the motion of
MreB to polymerization dynamics or treadmil-
ling (15, 16), models at odds with the observa-
tion that purified B. subtilis MreB displays no
difference in polymerization in adenosine 5′-
diphosphate (ADP)- and adenosine 5′-triphosphate
(ATP)-bound states (5). To investigate the role of
polymer dynamics in MreBmovement, we imaged
GFP-MreB containing two different mutations
thought to perturb ATP hydrolysis (7, 16, 17).
Consistent with an inhibition of MreB function,
expression of these mutants resulted in perturbed
cell morphologies (fig. S1). However, these mu-
tants displayed circumferential movements at
speeds similar to those observed above (24 T 4,
26 T 3 nm/s; n = 25) (Fig. 1B, fig. S1C, and
1
Department of Systems Biology, Harvard Medical School,
Boston, MA 02115, USA.
2
Department of Microbiology and
Molecular Genetics, Harvard Medical School, Boston, MA 02115,
USA.
3
Department of Physics, Harvard University, Cambridge,
MA 02138, USA.
4
Department of Chemistry and Chemical
Biology, Harvard University, Cambridge, MA 02138, USA.
5
Howard Hughes Medical Institute, Harvard University, Cam-
bridge, MA 02138, USA.
*To whom correspondence should be addressed. E-mail:
ethan.garner@hms.harvard.edu (E.C.G.); zhuang@chemistry.
harvard.edu (X.Z.); rudner@hms.harvard.edu (D.Z.R.);
timothy_mitchison@hms.harvard.edu (T.M.)
†These authors contributed equally to this work.
Fig. 1. MreB paralogs display circumferential motion independent of the cell body. (A) (Left) Montage
of GFP-Mbl motion (BDR2061) from movie S1B. (Middle) Maximal intensity projection (MIP) of movie
S1B. (Right) Kymographs drawn between lines in montage. Under our growth conditions, Bacillus grows
in long septate chains. (B) (Top) Kymographs of GFP-Mbl, GFP-MreB, and GFP-MreBH in merodiploid
strains. (Far right) Kymograph showing axial motion of GFP-MreB(D158A), a mutation believed to inhibit
ATP hydrolysis. (Bottom) MIP of movies of GFP-Mbl, GFP-MreB, and GFP-MreBH. (C) (Top) Kymograph of
EpsE-GFP. (Bottom) MIP of an EpsE-GFP movie.
REPORTS
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 222

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

movies S2B and S4), suggesting that a mecha-
nismother than polymerization dynamics drives
MreB motion.
Because MreBinteracts with the PGEM(1–4),
we hypothesized that MreB movement could be
driven by cell wall synthesis. To test this, we
monitored GFP-Mbl dynamics while depleting
three components of the PGEM: RodA, RodZ,
and Pbp2A(Fig. 2A, fig. S2, and movies S5, A to
F). As these proteins depleted over time, we ob-
served a gradual cessation of movement. At late
stages of depletion, most of the Mbl was mo-
tionless. Notably, these experiments revealed a dis-
connected structure: At intermediate depletion states
(~2 hours), cells displayed immobile filaments,
while adjacent particles still underwent rotary
movement.
Similar to genetic depletions, the addition of
antibiotics targeting different steps in PG synthe-
sis caused a cessation of filament motion (Fig. 2B
and movies S6, A to D). This effect was rapid:
Antibiotic addition to cells under thin agar pads
completely stopped filament motion within 10 to
30 s. High concentrations of antibiotics that target
other essential processes had no effect on filament
movement (Fig. 2C and movies S7, A and B),
suggesting that this effect is specific. Further-
more, the minimal inhibitoryconcentrations (MICs)
of antibiotics that stopped motion mirrored the
minimal concentrations that inhibited cell growth
(fig. S3, table S1, and movies S8, A to D), with
short treatments near the MICresulting in partially
frozen filaments.
Thus, PG synthesis appears to drive the mo-
tion of MreB. For this hypothesis to be correct,
both the PGEMand MreBparalogs should move
around the cell body in a similar manner. To test
this, we characterized the dynamics of the MreB
paralogs and three of the PGEM components
(MreC, MreD, Pbp2A) using high-precision par-
ticle tracking. We titrated the expression of GFP
fusion proteins to lowlevels to obtain diffraction-
limited foci, which we imaged using total inter-
nal reflection fluorescence (TIRF) microscopy
to examine their dynamics on the bottom half
of the bacterium. Low-level expression was ob-
tained in two ways: (i) with inducible expres-
sion in the background of endogenous protein
(merodiploids); and (ii) with inducible expres-
sion as the only source of protein (replacements).
The centroids of foci were fit to obtain posi-
tional information with subpixel precision (18),
and their positions were tracked over time (movie
S9A). These studies demonstrated that all six
proteins move in linear paths across the cell width
(Fig. 3).
We determined the velocity for all traces over
20 frames in length by two methods. These analy-
ses revealed that all six proteins move at similar
speeds under each expression condition (Fig. 4, A,
B, and G; figs. S6 to S8; and table S2). When the
proteins were expressed at low levels as replace-
ments, their mean velocity increased (Fig. 4, A
and B), with the exception of the minor paralog
MreBH (fig. S9A). This increase in speed may
arise from a cellular response to the reduction
in the overall levels of the cell wall synthesis ma-
chinery, and we could reproduce this effect in
trans by tracking merodiploid GFP-Mbl foci in
an MreB deletion strain (fig. S9B).
Next, we determined the angles at which these
proteins crossed the cell. The mean angle was
90° for all proteins, with most trajectories fall-
ing within 15° of this mean (Fig. 4, C and G, and
fig. S11). Analysis of the scaling exponents from
the mean squared displacement (MSD) versus t
plots indicated that all six proteins move in a
directed manner (fig. S12 and table S2). Thus, all
three MreB paralogs and three PGEM compo-
nents exhibit directed motions occurring at the
same speed and angle to the cell body. Therefore,
these proteins move in concert, functioning as a
macromolecular unit (1–4). Consistent with this
idea, vancomycin stopped the directed movements
of all proteins: Foci of MreB paralogs became
immobile, as did chromosomal replacements of
MreCand MreD. By contrast, directionally moving
Fig. 2. Filament motion requires cell wall synthesis. (Kymographs are drawn between lines.) (A) Kymographs
of GFP-Mbl during depletions of isopropyl-b-D-thiogalactopyranoside (IPTG)–inducible genes: (i) RodA, a
membrane-spanning component of the PGEM; (ii) RodZ, a protein that links MreB to the PGEM; and (iii)
Pbp2A, an elongation-specific transpeptidase, which was depleted in a strain lacking the redundant
transpeptidase PbpH. Strains were grown in 2 mM IPTG, shifted to media without IPTG, then imaged at
the indicated times. (B) Kymographs showing antibiotics targeting cell wall synthesis freeze GFP-Mbl
motion. BDR2061 was imaged after addition of 2 ml of antibiotics to a 600-ml agar pad. Initial con-
centrations: 10 mg/ml ampicillin (blocks transpeptidation), 5 mg/ml mecillinam (blocks transpeptidation),
80 mg/ml vancomycin (blocks transglycosylation and transpeptidation), 50 mg/ml phosphomycin (blocks
PG precursor synthesis; 6 ml added). (C) Kymographs showing that off-target antibiotics do not affect GFP-
Mbl motion. BDR2061 was incubated with the indicated antibiotics for 2 min and immediately imaged.
Final concentrations: 500 mg/ml rifampicin (inhibits transcription), 500 mg/ml kanamycin (inhibits
translation), 340 mg/ml chloramphenicol (inhibits translation).
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 223
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Pbp2A foci transitioned into a rapidly diffusive
state (fig. S13 and movie S12A).
Similar to the fragmented structure suggested
during PGEMdepletions, our single-particle track-
ing data were not consistent with the existence
of a coherent, long-range MreBcytoskeleton. Rath-
er, the directional motions were often discontin-
uous and independent: Foci of both the MreB
paralogs and PGEM displayed pauses and re-
versals in motion, while adjacent foci continued
to move unidirectionally (fig. S14 and movies
S12, B to D). These reversals could not have been
due to movement of the proteins around the cell
body, as our TIRF imaging only reports move-
ments of foci on the bottom half of the cell.
Having observed discontinuous motions of
single foci, we next examined the relative direc-
tionality between foci to test if their motions were
coordinated along the cell length. We calculated
the fraction of traces moving in the same direc-
tion over the imaging period as a function of their
separation. We found no correlation even at the
shortest separations: The relative orientations of
all proteins were randomly distributed at all dis-
tances (Fig. 4D). This uncoordinated movement
was evident in our tracking movies (movies S9,
A to C, and S10, A to C), and we have high-
Fig. 3. Particle tracking of MreB paralogs and the PGEM shows linear
movements across the cell. Representative traces of (A) MreB paralogs and
(B) PGEM components from each expression condition. Trace color encodes
time (blue to red) in 300-ms steps. Cell outline is blue, midline green. Low-
level expression of MreB, Mbl, MreC, and MreD in replacements resulted in
wider cells, which we stabilized with magnesium (figs. S4 and S5). For the
MreB paralogs, both expression methods yielded large numbers of foci that
moved in linear paths across the cell width (movies S9, B and C). Ex-
pression of PGEM proteins by both methods revealed that PGEM foci par-
tition into two populations, one moving slowly and directionally and one
moving rapidly and nondirectionally, which we interpret to be diffusion
within the membrane. When expressed at high levels as replacements, a
dense mix of both populations was observed. As PGEM expression levels
were reduced, the diffusing population effectively disappeared, leaving
predominantly directionally moving foci that traversed the cell width
(movies S10, A to C). When expressed at low levels in merodiploids, both
populations of PGEM foci were observed, with the directionally moving
population comprising the minority (movie S11). Because we could only
accurately track the slow directionally moving particles, all our data refer
to this population.
Fig. 4. Relative dynamics of the cell wall synthesis machinery and MreB. His-
tograms of velocity of GFP fusion proteins expressed as (A) merodiploids and
(B) replacements. Velocity (V) was calculated by fitting MSDversus t (fig. S6) to
MSD = (Vt)
2
+ 4Dt, yielding two distinct populations, high (>5 × 10
−4
nm/s) and
low (≤5 × 10
−4
nm/s) (table S2 and fig. S7). Displayed are high-velocity traces
that moved in a consistent manner during their lifetime [>0.95 r
2
fit to log(MSD)
versus log(t)]. Plots of all data without r
2
screening are in fig. S7. (C) Distributions
of the angles that traces cross the cell, determined by combining trajectories with
segmented brightfield images (fig. S10) (17). Shown are traces combined from
both expression conditions over 20 frames in length with linear r
2
fits > 0.5.
Separate plots of each expression condition and different r
2
screening criteria
are in fig. S11. (D) Linear traces [as in (C)] extracted from 100 s of imaging
were evaluated pairwise to determine their relative directionality. The fraction
of traces moving in the same direction is plotted as a function of distance in
bins of 160 nm. (E and F) Kymographs of proximal foci in merodiploid GFP-
Mbl (E) and replacement GFP-Pbp2A (F) strains moving in opposing directions
across one surface. Distance between kymograph bars is indicated in italics
(See also fig. S15 and movies S13, A to D.) (G) Summary table of tracking data.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 224
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

lighted examples of proximal opposing move-
ments (Fig. 4, E and F; fig. S15; and movies
S13, A to D). Thus, PGEM and MreB filaments
move in both directions around the cell, and we
could not resolve any coordination along the
cell length.
It thus appears that the coupled motions of the
PGEM and MreB reflect the active process of
cell wall synthesis: a circumferential motion of
disconnected MreB-PGEM complexes moving
around the cell in both directions, synthesizing
discrete radial bands of PG oriented perpendic-
ular to the cell length (fig. S16). This model is
consistent with the arrangement of B. subtilis PG
observed by atomic force microscopy (19).
MreB filaments are required for elongation-
specific PGsynthesis (2, 20), suggesting that they
are integral components of these translocating
machines. These filaments may serve as coordi-
nating scaffolds to link the PGEMto the enzymes
that synthesize PGprecursors (1–4, 10) (fig. S17).
We cannot completely rule out the contribution
of polymer dynamics to these motions, because
there are no methods to inhibit MreBpolymeriza-
tion without disrupting existing filaments. How-
ever, we did not observe any directed motion in the
absence of PGsynthesis, even with high-precision
measurements (fig. S13), suggesting that PG syn-
thesis is the predominant process driving these
motions. If these motions are driven by MreB po-
lymerization or another process, this would re-
quire induction of equivalent rigor states during
depletions of all PGEM components and the anti-
biotic inhibition of PG cross-linking, polymeriza-
tion, and precursor synthesis.
Rather than a contiguous helical structure,
these observations reveal the mobile, fragmented
nature of MreB. Thus, although MreBis required
for rod-shape maintenance, it cannot function
as a cell-spanning structure, much less a coher-
ent “cytoskeleton” in B. subtilis. It remains to
be determined how the short-range activities of
these independent biosynthetic complexes im-
part a long-range order to the cell wall.
References and Notes
1. T. Kruse, J. Bork-Jensen, K. Gerdes, Mol. Microbiol. 55,
78 (2005).
2. Y. Kawai, K. Asai, J. Errington, Mol. Microbiol. 73, 719 (2009).
3. F. O. Bendezú, C. A. Hale, T. G. Bernhardt, P. A. de Boer,
EMBO J. 28, 193 (2009).
4. C. L. White, A. Kitich, J. W. Gober, Mol. Microbiol. 76,
616 (2010).
5. J. A. Mayer, K. J. Amann, Cell Motil. Cytoskeleton 66, 109
(2009).
6. F. van den Ent, L. A. Amos, J. Löwe, Nature 413, 39
(2001).
7. T. Kruse, J. Møller-Jensen, A. Løbner-Olesen, K. Gerdes,
EMBO J. 22, 5283 (2003).
8. L. J. Jones, R. Carballido-López, J. Errington, Cell 104,
913 (2001).
9. Z. Gitai, N. A. Dye, A. Reisenauer, M. Wachi, L. Shapiro,
Cell 120, 329 (2005).
10. R. M. Figge, A. V. Divakaruni, J. W. Gober,
Mol. Microbiol. 51, 1321 (2004).
11. R. A. Daniel, J. Errington, Cell 113, 767 (2003).
12. R. Carballido-López, Microbiol. Mol. Biol. Rev. 70, 888
(2006).
13. R. Carballido-López, J. Errington, Dev. Cell 4, 19
(2003).
14. H. J. Soufo, P. L. Graumann, Curr. Biol. 13, 1916
(2003).
15. S. Y. Kim, Z. Gitai, A. Kinkhabwala, L. Shapiro,
W. E. Moerner, Proc. Natl. Acad. Sci. U.S.A. 103, 10929
(2006).
16. H. J. Defeu Soufo, P. L. Graumann, Mol. Microbiol. 62,
1340 (2006).
17. See supporting material on Science Online.
18. R. E. Thompson, D. R. Larson, W. W. Webb, Biophys. J.
82, 2775 (2002).
19. E. J. Hayhurst, L. Kailas, J. K. Hobbs, S. J. Foster, Proc.
Natl. Acad. Sci. U.S.A. 105, 14603 (2008).
20. T. Uehara, J. T. Park, J. Bacteriol. 190, 3914 (2008).
Acknowledgments: We thank P. Graumman, J. Errington,
R. Carballido-Lopez, D. Kearns, D. Popham, and
D. Scheffers for strains; D. Kahne, S. Walker,
T. Bernhardt, and T. Uehara for discussions; Q. Justman
for editing; S. Layer for inspiration; T. Emonet for
MicrobeTracker; and the HMS Nikon Imaging Center
for the use of their facilites. Funding was provided by
NIH grants R01-GM073831 (D.Z.R.), R01-GM096450
(X.Z.), and R01-GM39565 (T.M.). X.Z. is a Howard
Hughes Medical Investigator. E.C.G. was supported
by the Helen Hay Whitney Foundation.
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1203285/DC1
Materials and Methods
SOM Text
Figs. S1 to S17
Tables S1 to S5
Movies S1 to S13
References
25 January 2011; accepted 20 May 2011
Published online 2 June 2011;
10.1126/science.1203285
Processive Movement of
MreB-Associated Cell Wall
Biosynthetic Complexes in Bacteria
Julia Domínguez-Escobar,
1
Arnaud Chastanet,
2,3
* Alvaro H. Crevenna,
1
* Vincent Fromion,
4
Roland Wedlich-Söldner,
1
† Rut Carballido-López
2,3

The peptidoglycan cell wall and the actin-like MreB cytoskeleton are major determinants of cell
shape in rod-shaped bacteria. The prevailing model postulates that helical, membrane-associated
MreB filaments organize elongation-specific peptidoglycan-synthesizing complexes along sidewalls.
We used total internal reflection fluorescence microscopy to visualize the dynamic relation between
MreB isoforms and cell wall synthesis in live Bacillus subtilis cells. During exponential growth,
MreB proteins did not form helical structures. Instead, together with other morphogenetic factors,
they assembled into discrete patches that moved processively along peripheral tracks perpendicular
to the cell axis. Patch motility was largely powered by cell wall synthesis, and MreB polymers
restricted diffusion of patch components in the membrane and oriented patch motion.
T
he peptidoglycan (PG) layer, or sacculus,
which is composed of glycan strands cross-
linked by peptide bridges, forms a load-
bearing network that maintains bacterial cell
shape. Synthesis and chemical composition of
PG are well understood (1), but the structure of
the sacculus and the mechanisms controlling its
growth remain elusive. It is currently assumed
that the actin-like MreB proteins form filamentous
helical structures along the membrane, which di-
rect cell wall growth by positioning multienzyme
complexes that mediate sidewall elongation (2).
These elongation complexes are thought to con-
tain the essential transmembrane proteins MreC
and MreD, RodA and RodZ, PG hydrolases, and
penicillin-binding proteins (PBPs), the enzymes
that catalyze PG elongation and cross-linking
(3, 4). Studies using fluorescently labeled vanco-
mycin (Van-FL) have revealed helical incorpora-
tion patterns of PG precursors into the sidewall
in B. subtilis (5) and Caulobacter crescentus (6),
but could not resolve the dynamics of cell wall
synthesis. We used total internal reflection fluo-
rescence microscopy (TIRFM), a sensitive method
for studying events at cell surfaces (7, 8), to ob-
serve the dynamics of the MreB cytoskeleton and
its relationship with cell wall growth.
Functional green fluorescent protein (GFP)
fusions to the three MreB isoforms in B. subtilis
(MreB, Mbl, and MreBH) expressed at wild-
type levels (fig. S1) formed discrete patches in
exponentially growing cells (Fig. 1A). Patches
were restricted to the cell periphery (fig. S2A)
and exhibited continuous movement along lin-
ear tracks roughly perpendicular to the long axis
of the cell (Fig. 1B and movie S1). To reconcile
our findings with the helical structures described
for MreB proteins (2, 9), we simultaneously im-
aged cells by TIRFM and conventional epiflu-
orescence. Owing to the increased depth of field,
MreB patterns visualized by epifluorescence could
be misinterpreted as “helical” (Fig. 1C). In addi-
tion, MreB localized to transverse bands as cells
entered stationary phase (fig. S2B). We found
1
Max Planck Institute of Biochemistry, Am Klopferspitz 18,
D-82152 Martinsried, Germany.
2
INRA, UMR1319 Micalis,
F-78352 Jouy-en-Josas, France.
3
AgroParisTech, UMR Micalis,
F-78350 Jouy-en-Josas, France.
4
INRA, UR1077 Mathématique,
Informatique et Génomes, F-78352 Jouy-en-Josas, France.
*These authors contributed equally to this work.
†To whom correspondence should be addressed. E-mail:
wedlich@biochem.mpg.de (R.W.-S.); rut.carballido-lopez@jouy.
inra.fr (R.C.-L.)
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 225
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

similar patch-like localization and dynamics of
MreB in the Gram-negative bacteria Escherichia
coli and C. crescentus (fig. S2C), suggesting that
these are widely conserved features in bacteria.
Detailed analysis of patch dynamics using
kymographs (10) showed that patches moved
bidirectionally across the cell at constant velocity
(fig. S2D). We often observed reversal and cross-
over of patches (fig. S2D), but did not see abrupt
patch appearances or disappearances at mid-
trajectory, indicating tight association of MreBs
with the cell periphery. To determine patch veloc-
ities, we measured the slopes of kymograph traces.
Patches formed by MreB, Mbl, and MreBH moved
at significantly different speeds (Fig. 1D, fig. S3,
and table S1; P < 0.001 for all pairs). Speeds were
mildly dependent on growth temperature (fig.
S4A) but were not affected by high concentrations
of magnesium (fig. S4B) or untagged endogenous
copies of the respective MreB isoforms (fig. S4C
and table S1). Patch trajectories in maximum
projections were oriented at angles close to 90°
relative to the long axis of the cell (Fig. 1E and
table S1). Although tracks were evenly spaced
along the length of the cell, with major distance
peaks between 0.5 and 1 mm (Fig. 1F), high var-
iability of autocorrelation between cells (fig. S5,
A to C) and over time (fig. S5D) argued against
a constrained periodic structure. However, we did
find a correlation between numbers of tracks and
cell length (Fig. 1G), suggesting an average dis-
tance between tracks of ~0.5 mm, as previously
reported (11).
MreB, Mbl, and MreBH have been reported
to colocalize and interact extensively with each
other (12, 13). To determine whether differences
in patch speeds (Fig. 1D) reflected distinct isoform
composition or simply different strain backgrounds,
we performed two-color TIRFM on pairs of GFP-
and mRFPruby- (monomeric RFP, red fluorescent
protein) tagged proteins. All tested pairs displayed
extensive overlap (Fig. 1H) with colocalization in
more than 75% of kymograph traces (Fig. 1I
and table S3) and moved at similar speeds (fig.
S6 and table S1), indicating that the three
MreB isoforms coexist in motile patches.
MreB and Mbl have been implicated in the
spatial organization of lateral cell wall synthetic
A
E
B
D
C
F
0 12 8 4 max
I
n
t
e
n
s
i
t
y
200
0 6 5 4 3 2 1
600
I
C
F
-0.5
0
0.5
1
Length (µm)
G

MreB
Mbl
MreBH
0
0
8 6 4 2
4
8
16
12
Length (µm)
n

(
t
r
a
j
e
c
t
o
r
i
e
s
)
BF EPI TIRFM
H
3
0

s
I
0 100
0
100
I
n
t
e
n
s
i
t
y



(
a
.
u
.
)
S
p
e
e
d

(
n
m
/
s
)
MreB Mbl MreBH
120
0
60
A
n
g
l
e

(
d
e
g
r
e
e
s
)
130
50
MreB Mbl MreBH
90
% traces 100 0
G-MreB Mbl-R
G-MreBH R-MreB
G-MreBH Mbl-R
G-Mbl Mbl-R
Mbl-G R-Mbl
% traces 100 0
G-PbpH R-MreB
R-Mbl
RodA-G R-MreB
RodA-G R-Mbl
G-PbpH
M
S
p
e
e
d

(
n
m
/
s
)
120
0
60
P
b
p
H
M
r
e
D
M
r
e
C
P
B
P
2
a
R
o
d
A
J K
A
n
g
l
e

(
d
e
g
r
e
e
s
)
130
50
P
b
p
H
M
r
e
D
M
r
e
C
V
a
n
-
F
L
P
B
P
2
a
R
o
d
A
L
0 90
0
100
Position (pix)
3
0

s
i
n
t
e
n
s
i
t
y



(
a
.
u
.
)
Position (pix)
Fig. 1. Motile cell wall elongation complexes. (A) GFP-MreB, GFP-Mbl, and
GFP-MreBH (left to right) imaged by TIRFM. Data in (A) to (G) are from
strains 3723, 2523, and 2566J (table S4). (B) Movement of a GFP-MreB
patch (arrow) and trajectory in maximum projection. Time is in seconds. (C)
Simultaneous visualization of GFP-Mbl by TIRFM and epifluorescence (EPI).
(Left) Corresponding bright-field (BF) image. (D) Typical kymograph and
patch speed boxplots of MreB isoforms. (E) Typical maximum projection
and angle boxplots for MreB isoform trajectories. (F) Distribution of GFP-
MreB trajectories in maximum projection, linescan (along dotted line)
and intensity correlation function (ICF). (G) Number of patch trajectories
increases with cell length (linear fit: R
2
= 0.61). (H) Colocalization of
GFP-MreB and Mbl-RFP in maximum projections, linescan (dotted line)
and kymographs. (I) Quantification of MreBs colocalization from kymo-
graph traces. G: GFP; R: RFP; gray bars: colocalization; green and red bars:
single GFP and RFP color traces. (J) Boxplots of patch speed of morpho-
genetic proteins. (K) Orientation of patch trajectories of morphogenetic
proteins and Van-FL bands. (L) Colocalization of GFP-PbpH and RFP-MreB
in maximum projections, linescan, and kymograph traces. (M) Quantification
of MreB and Mbl colocalization with PbpH and RodA from kymograph traces.
Labels as in (I). Scale bars: 1 mm.
Fig. 2. Patch motility is not driven by
treadmilling. (A) Time series and kymo-
graph showing fission and fusion (arrow-
head) and reversal of a GFP-Mbl patch
(asterisk, RWSB10). Red: TIRFM; green:
epifluorescence; blue (cell outline): bright
field. (B) Time series and kymograph
(dotted line) showing splitting of a GFP-
MreB patch (RWSB1). (C) TIRF-FRAP with
partial photobleaching of a moving GFP-
MreB patch (RWSB1). (D) Kymographs
along patch traces of GFP-Mbl in the
DmrebDmbl background (RWSB10), with
corresponding intensity profiles showing
lack of fluorescence recovery upon par-
tial bleaching. (E) Inverse FRAP (iFRAP).
A cell (RWSB10) was bleached within the
region outlined, omitting only a GFP-Mbl
patch. Kymograph (dotted line) shows
movement of the patch with no loss of fluorescence. Scale bars: 1 mm; time bars: 30 s in (A) and (B); 10 s
in (D) and (E). Time is in seconds. Asterisks and arrow at initial position in (C) and (E).
A B
E
D C
0 20 16 12 8 4
24 44 40 36 32 28
I
n
t
e
n
s
i
t
y

(
a
.
u
.
)
Time (s)
0
0 10 20
1
0
-6
-2
-4
0
4
2
6
8
0 20 16 12 8 4 -4 -8
* * * * *
*
*
*
*
* *
*
* *
*
*
*
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 226
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

complexes (4, 14). We analyzed the localization
of proteins associated with sidewall elongation
by TIRFM. MreC and MreD, transmembrane
proteins reported to localize in helical patterns
and to couple the cytosolic MreBs to the extra-
cellular PG synthetic machinery (4, 15), formed
discrete patches that behaved like MreB patches
in speed and orientation (Fig. 1, J and K, table
S1, and movie S2). In B. subtilis, several PBPs
(16) and the autolysin LytE (12) localize to the
sidewall in distinct foci and bands. Of the 11
vegetatively expressed PBPs of B. subtilis, only
the transpeptidase PbpH formed patches that
moved around the cell periphery (Fig. 1, J and
K, table S1, and movie S3). All other PBPs and
LytE localized to patches that randomly moved
along the cell surface (fig. S7A). In cells lacking
pbpH, PBP2a patches displayed directional
movement along circumferential tracks (Fig. 1,
J and K, table S1, and movie S3), consistent with
the proposed reciprocal redundancy of these two
transpeptidases in sidewall synthesis (17). Fur-
thermore, the integral membrane protein RodA,
which has been linked to cell wall elongation and
to PbpH and PBP2a (17, 18), also formed circum-
ferentially moving patches (Fig. 1, J and K,
table S1, and movie S3). Finally, the orientation
of Van-FL–labeled tracks was also centered around
90°C (Fig. 1K) instead of helical (5). Whereas
MreBs and MreC patches exclusively displayed
circumferential motion, MreD, PBP2a, PbpH,
and RodA patches also frequently exhibited rapid
diffusion along the membrane (fig. S7B). Cir-
cumferentially motile PbpH and RodA strongly
colocalized with MreB and Mbl (Fig. 1, L and
M, fig. S6, and table S1). Thus, MreBs, MreC/D,
RodA, and PbpH/2a form part of cell wall elon-
gation complexes that display circumferential
processive motility in B. subtilis cells.
What is the molecular basis for the proces-
sive movement? One possibility would be tread-
milling of MreB, as previously suggested in
B. subtilis (11) and C. crescentus (19). However,
several dynamic behaviors of MreB patches such
as fusion and fission (Fig. 2A and movie S4),
reversal of direction (Fig. 2A), or splitting (Fig.
2B) argue against treadmilling-driven motion. To
directly test for turnover of MreBs within patches,
we performed FRAP (fluorescence recovery after
photobleaching) experiments. When individu-
al GFP-MreB or GFP-Mbl patches were partial-
ly bleached, their motility was unaffected and
no fluorescence recovery occurred during their
movement across the cell (Fig. 2, C and D). Sim-
ilarly, when a whole cell was bleached, with the
exception of a single GFP-Mbl patch (Fig. 2E),
the signal from this patch did not decrease dur-
ing its movement across the cell, again ruling out
treadmilling-driven motion. Absence of treadmill-
ing does not, however, preclude polymerization
and slow global turnover of MreB (20), which is
predicted from the correlation between cell length
and patch numbers (Fig. 1F).
Alternatively, a motive force for MreB patches,
and thus for the elongation complexes, could
be provided by PG synthesis itself. To test this
possibility, we treated cells with vancomycin
(100 mg/ml) to inhibit incorporation of PG pre-
cursors into the sacculus (21), or phosphomycin
(700 mg/ml), which blocks the first cytosolic
step in PG precursor synthesis (22). Addition of
either antibiotic completely stopped movement
of PbpH, MreB, and Mbl patches (Fig. 3, A and
B). After removal of the drugs, motility grad-
ually resumed (Fig. 3, C and D, and movie S5).
Finally, upon disruption of the PG backbone with
lysozyme, patch movement also ceased (Fig. 3E
and movie S6), and many patches rapidly dif-
fused in the cytosol (Fig. 3E and movie S6). If
PG assembly directly drives motility of MreB
patches, the concentration of PG precursors avail-
able should influence patch speed. Reducing
the precursor pool with low concentrations (1 to
5 mg/ml) of phosphomycin did not affect the di-
rection of MreB and Mbl patches (fig. S9A),
but their velocities were significantly lower (Fig.
3F and table S1). FRAP experiments showed
no turnover of MreB or Mbl patches in the pres-
ence of either high or low concentrations of phos-
phomycin (fig. S8), confirming the absence of
MreB turnover in patches. To target PG synthe-
sis genetically without inducing drastic changes
in cell morphology, we analyzed MreB, Mbl,
and RodA patches in mutants null for pbpH and
pbpA (encodes PBP2a), which display a wild-type
growth rate and morphology during vegetative
growth (17, 23). Both deletions significantly
slowed down MreB and Mbl patches (Fig. 3G),
suggesting a role for transpeptidases in setting
Fig. 3. Contribution of PG syn-
thesis to patch motility. (A and B)
Immobile MreB, Mbl, and PbpH
patches after treatment with van-
comycin (100 mg/ml) for 8 min (A)
and with phosphomycin (700 mg/ml)
for 30 min (B). Time is in seconds.
(C and D) Effects of vancomycin
(C) and phosphomycin (D) are revers-
ible. Kymographs were taken before
(−) and immediately after addition
of the drug (0) and at the indicated
times (in minutes) after washout. Red
arrows: partial recovery of motility. (E)
Kymographs of GFP-Mbl patches
(2523) showing partial arrest (a),
complete arrest (b), or diffusive mo-
tility (c) after treatment with lysozyme
(1 mg/ml) for 5 min. (F and G) Patch
speed boxplots upon low phosphomy-
cin (1 to 5 mg/ml) treatment relative
to untreated cells (F) and in the pres-
ence and absence of pbpH or pbpA
(G). Scale bars: 1 mm; time bars: 30 s.
Fig. 4. Roles of MreBs in patch mo-
tility. (A and B) Boxplots of patch
speeds (A) and trajectory angles
(B) in the presence and absence of
mreB. (C) Movement of a GFP-Mbl
patch (asterisk) along the cell axis
in the DmrebDmbl strain (RWSB10).
Red: TIRFM; green: epifluorescence;
blue (cell outline): bright field. Time
is in seconds. (D) Normalized PbpH
and RodA patch motility (patch tra-
jectories per kymograph trace per
minute) after deletion of mreB or
mbl. (E) Typical kymographs for
PbpH and RodA patches in wild-
type and mreB backgrounds. Scale
bars: 1 mm; time bar: 30 s.
A
C
0 4 8 12 16 20 24
D E
B
A
n
g
l
e

(
d
e
g
r
e
e
s
)
180
0
Mbl MreBH PbpH
∆mreB ∆mreB
90
w
t

m
b
l

m
r
e
B
w
t

m
b
l

m
r
e
B
0
1
PbpH RodA
M
o
t
i
l
e

p
a
t
c
h
e
s

(
a
.
u
.
)
S
p
e
e
d

(
n
m
/
s
)
120
0
60
PbpH Mbl MreBH RodA
wt
∆mreB
PbpH RodA
F
A
MreB
0 20 0 20 0 20
Mbl PbpH
B
0 20 0 20 0 20
MreB Mbl PbpH
S
p
e
e
d

(
n
m
/
s
)
MreB Mbl
120
0
60
phosphomycin
E
0
-2
10
5
C
D
0 -30 30
a
b
c
G
S
p
e
e
d

(
n
m
/
s
)
MreB Mbl
120
0
60
∆pbpH ∆pbpA
RodA
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 227
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

the pace of new PG strand assembly. Indeed, the
essential transpeptidase PBP2 of E. coli plays a
direct role in glycan strand synthesis apart from
its role in cross-linking (24). Again, track orienta-
tion was unaffected (fig. S9B). Mutants in pbpH
and pbpA were slightly impaired in growth (fig.
S9C), consistent with a link between patch mo-
tility and cell growth.
Our results indicate that cell wall synthesis
itself provides the driving force for the proces-
sive motility of sidewall elongation complexes.
This is also supported by recent analysis show-
ing that PG glycosyl transferases are processive
enzymes (25). What, then, is the biological func-
tion of the essential actin-like MreB? MreB posi-
tions cytosolic enzymes involved in PG precursor
synthesis in C. crescentus (6, 14). Thus, MreB
patches might simply act as passive scaffolds for
synthesis and export of PG precursors and/or
regulate their delivery to sites of active PG syn-
thesis (24). Our observation that RodA and PBPs
localize not only to MreB-associated motile elon-
gation complexes but also to rapidly diffusing
patches (fig. S7A) suggested an alternative sce-
nario, in which MreB polymers might actively
restrict and/or control the mobility of cell wall
elongation complexes. To test this hypothesis,
we monitored patch motility in the absence of
individual MreB isoforms. Patches moved much
faster in the mreB mutant, whereas deletion of
mbl or mreBH had no significant effect (Fig. 4A,
fig. S10, and table S1). In addition, GFP-Mbl
patches in a DmblDmreB background exhibited
less uniform directionality (Fig. 4B), sometimes
even following trajectories along the cell axis
(Fig. 4C and movie S7). An even stronger effect
was observed on RodA and PbpH: Processive-
ly motile patches were almost completely elimi-
nated, and the few remaining covered very short
distances, albeit faster (Fig. 4, D and E). Thus,
MreB may function directly in restricting and/or
organizing motility of cell wall elongation com-
plexes, although an additional role in the recruit-
ment of PG precursors cannot be excluded.
We propose a model for sidewall elongation
in B. subtilis, where motile membrane-associated
elongation complexes insert new PG along bands
largely perpendicular to the long cell axis. Old PG
strands are used as guiding scaffolds, and motil-
ity of the complexes is powered by PG polymer-
ization. MreB-like polymers restrict the diffusion
of the complexes within the membrane to achieve
processive and correctly oriented movement. A
similar scenario has been proposed for plant cells,
where processive motility of cellulose synthase
complexes is suggested to be constrained by cor-
tical microtubules (26).
References and Notes
1. J. V. Höltje, Microbiol. Mol. Biol. Rev. 62, 181
(1998).
2. R. Carballido-López, Microbiol. Mol. Biol. Rev. 70, 888
(2006).
3. R. Carballido-López, A. Formstone, Curr. Opin. Microbiol.
10, 611 (2007).
4. W. Margolin, Curr. Biol. 19, R812 (2009).
5. R. A. Daniel, J. Errington, Cell 113, 767 (2003).
6. A. V. Divakaruni, C. Baida, C. L. White, J. W. Gober,
Mol. Microbiol. 66, 174 (2007).
7. J. H. Yu, A. H. Crevenna, M. Bettenbühl, T. Freisinger,
R. Wedlich-Söldner, J. Cell Sci. 124, 1533 (2011).
8. T. C. Fleming et al., Genes Dev. 24, 1160 (2010).
9. M. T. Cabeen, C. Jacobs-Wagner, Annu. Rev. Genet. 44,
365 (2010).
10. Materials and methods are available as supporting
material on Science Online.
11. H. J. Defeu Soufo, P. L. Graumann, EMBO Rep. 5, 789
(2004).
12. R. Carballido-López et al., Dev. Cell 11, 399 (2006).
13. H. J. Defeu Soufo, P. L. Graumann, Mol. Microbiol. 62,
1340 (2006).
14. C. L. White, A. Kitich, J. W. Gober, Mol. Microbiol. 76,
616 (2010).
15. D. Claessen et al., Mol. Microbiol. 68, 1029 (2008).
16. D. J. Scheffers, L. J. Jones, J. Errington, Mol. Microbiol.
51, 749 (2004).
17. Y. Wei, T. Havasy, D. C. McPherson, D. L. Popham,
J. Bacteriol. 185, 4717 (2003).
18. A. O. Henriques, P. Glaser, P. J. Piggot, C. P. Moran Jr.,
Mol. Microbiol. 28, 235 (1998).
19. S. Y. Kim, Z. Gitai, A. Kinkhabwala, L. Shapiro,
W. E. Moerner, Proc. Natl. Acad. Sci. U.S.A. 103, 10929
(2006).
20. R. Carballido-López, J. Errington, Dev. Cell 4, 19
(2003).
21. D. Kahne, C. Leimkuhler, W. Lu, C. Walsh, Chem. Rev.
105, 425 (2005).
22. E. D. Brown, E. I. Vivas, C. T. Walsh, R. Kolter, J. Bacteriol.
177, 4194 (1995).
23. T. Murray, D. L. Popham, P. Setlow, J. Bacteriol. 179,
3021 (1997).
24. T. Uehara, J. T. Park, J. Bacteriol. 190, 3914 (2008).
25. D. Barrett et al., J. Biol. Chem. 282, 31964 (2007).
26. A. R. Paredez, C. R. Somerville, D. W. Ehrhardt, Science
312, 1491 (2006).
Acknowledgments: We thank R. Daniel and L. J. Wu for
sharing unpublished strains and for helpful suggestions,
D.-J. Scheffers for strains, N. Campo for the antibody
against GFP, and P. Hardy for manuscript editing. This
work was supported by a Human Frontier Science
Program (HFSP) Young Investigator grant (R.C.-L. and
R.W.-S., HFSP-RGY0067/2009-C); a European Molecular
Biology Organization short-term fellowship (R.C.-L),
The French National Research Agency (ANR)
(R.C.-L., ANR-08-JCJC-0024–01), and the Max Planck
society (R.W.-S.).
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1203466/DC1
Materials and Methods
Figs. S1 to S10
Tables S1 to S5
References (27–37)
Movies S1 to S7
27 January 2011; accepted 20 May 2011
Published online 2 June 2011;
10.1126/science.1203466
Phosphorylation of the Autophagy
Receptor Optineurin Restricts
Salmonella Growth
Philipp Wild,
1
Hesso Farhan,
2
David G. McEwan,
1
Sebastian Wagner,
3
Vladimir V. Rogov,
4,5
Nathan R. Brady,
6
Benjamin Richter,
1
Jelena Korac,
7
Oliver Waidmann,
1
Chunaram Choudhary,
3
Volker Dötsch,
4
Dirk Bumann,
2
Ivan Dikic
1,7
*
Selective autophagy can be mediated via receptor molecules that link specific cargoes to the
autophagosomal membranes decorated by ubiquitin-like microtubule-associated protein light chain 3
(LC3) modifiers. Although several autophagy receptors have been identified, little is known about
mechanisms controlling their functions in vivo. In this work, we found that phosphorylation of an
autophagy receptor, optineurin, promoted selective autophagy of ubiquitin-coated cytosolic Salmonella
enterica. The protein kinase TANK binding kinase 1 (TBK1) phosphorylated optineurin on serine-177,
enhancing LC3 binding affinity and autophagic clearance of cytosolic Salmonella. Conversely,
ubiquitin- or LC3-binding optineurin mutants and silencing of optineurin or TBK1 impaired Salmonella
autophagy, resulting in increased intracellular bacterial proliferation. We propose that phosphorylation
of autophagy receptors might be a general mechanism for regulation of cargo-selective autophagy.
M
acroautophagy (hereafter referred to as
autophagy) is an evolutionarily con-
served catabolic process by which cells
deliver bulk cytosolic components for degradation
to the lysosome (1–4). Selectivity in cargo target-
ing is mediated via autophagy receptors that simul-
taneously bind cargoes and autophagy modifiers,
autophagy-related protein 8 (ATG8)/ microtubule-
associated protein light chain 3 (LC3)/g-aminobutyric
acid receptor-associated protein (GABARAP) pro-
teins, which are conjugated to the autophagosomal
membranes (5, 6). The regulatory mechanisms
controlling the spatiotemporal dynamics of the
autophagy receptor-target interaction in cells re-
main unclear (7).
1
Frankfurt Institute for Molecular Life Sciences and Institute of
Biochemistry II, Goethe University School of Medicine, Theodor-
Stern-Kai 7, D-60590 Frankfurt (Main), Germany.
2
Infection
Biology, Biozentrum, University Basel, Klingelbergstr. 50/70,
CH-4056 Basel, Switzerland.
3
The Novo Nordisk Foundation
Center for Protein Research, Faculty of Health Sciences, Uni-
versity of Copenhagen, Blegdamsvej 3, 2200 Copenhagen,
Denmark.
4
Institute of Biophysical Chemistry and Center for
Biomolecular Magnetic Resonance, Goethe University, Marie
Curie Strasse 9, D-60439 Frankfurt (Main), Germany.
5
Institute
of Protein Research, 142290, Pushchino, Russia.
6
Systems
Biology of Cell Death Mechanisms, German Cancer Research
Center, Bioquant, Im Neuenheimer Feld 267, 69120 Heidel-
berg, Germany.
7
University of Split, School of Medicine, De-
partment of Immunology and Medical Genetics, Soltanska 2,
21 000 Split, Croatia.
*To whom correspondence should be addressed. E-mail:
ivan.dikic@biochem2.de
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 228
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Multiple autophagy receptors have been iden-
tified with the yeast two-hybrid system(8, 9), which
included an N-terminal fragment of optineurin
(OPTN), a ubiquitin-binding protein also known
as NF-kB essential modulator–related protein
(Fig. 1, Aand B). The specific interactions between
OPTN and LC3/GABARAP proteins were veri-
fied by pull-down assays in mammalian cells, di-
rected yeast two-hybrid transformations, and in
vitro using purified proteins (Fig. 1Cand fig. S1, A
and B) (10). OPTNbound to ubiquitin chains and
autophagy modifiers ATG8/LC3/GABARAP pro-
teins but not to mono-ubiquitin or other ubiquitin-
like proteins (Fig. 1C and fig. S1C). Deletion
mapping of the N-terminal region of OPTNiden-
tified an LC3 interacting motif (LIR), a linear
tetrapeptide sequence present in known autophagy
receptors that binds directly to LC3/GABARAP
modifiers (9, 11, 12). The LIR was located be-
tween the coiled-coil domains of OPTN encom-
passing amino acids 169 to 209 (Fig. 1A) and
was essential for in vitro and in vivo binding be-
tween OPTN and LC3/GABARAP (Fig. 1, B
and C, and figs. S1A and S2A). Single point mu-
tations at either OPTN Phe
178
→Ala
178
(F178A)
or I181A (13), corresponding to the WxxL of
p62, were sufficient to abrogate the interaction
with LC3/GABARAP proteins, whereas these
mutants were still able to bind to linear ubiquitin
chains fused to glutathione S-transferase (GST-
4xUb) (Fig. 1A and fig. S2B).
OPTN localized in LC3-positive vesicles upon
induction of autophagy (Fig. 1D and figs. S1D
and S3, A to F). An LC3-binding–deficient mu-
tant of OPTN (OPTN F178A) did not cluster
into cytoplasmic LC3 structures under autophagy-
inducing conditions (fig. S3B). Surprisingly, this
was also the case for ubiquitin-binding–deficient
OPTN variants (fig. S3C). The localization of
mCherry-LC3B with enhanced green fluorescent
protein (EGFP)–OPTN wild-type (WT), DLIR,
and F178A mutants was quantified under nutri-
ent deprivation–induced autophagy conditions in
human MCF-7 cells with the use of high-sampling
(>1000 cells per condition), high-resolution imag-
ing (fig. S3, D and E). Quantification of the pop-
ulation responses revealed that mCherry-LC3B:
EGFP-OPTN WT colocalization was enhanced
in cells treated with bafilomycin A1 (Fig. 1E). Co-
localization of EGFP-OPTN F178A and EGFP-
OPTNDLIRwithmCherry-LC3Bwas significantly
suppressed compared with WTEGFP-OPTN(Fig.
1E). Thus, OPTN is an autophagy receptor that
binds and localizes with LC3/GABARAP via a
phenylalanine-containing LIR motif and ubiquitin
via its ubiquitin binding in ABIN and NEMO
(UBAN) domains (14).
The hydrophobic core sequence of the LIR
motif in OPTN is preceded by multiple serine
residues, which are evolutionarily conserved (fig.
S4A). One prime kinase candidate was the Tank-
binding kinase (TBK1), due to the presence of a
conserved TBK1 consensus sequence (SxxxpS).
Indeed, TBK1 binds directly to OPTN (15) and
coexpression of OPTNtogether with TBK1 in vivo
increased OPTN phosphorylation, which was
reversed by either l-phosphatase treatment or a
TBK1 inhibitor (BX795) (fig. S4, B to D) (16).
Moreover, recombinant TBK1 phosphorylated
purified GST-OPTN in an in vitro kinase assay
(fig. S4, E and F), indicating that TBK1 is a bona
fide kinase that directly phosphorylates OPTN.
To identify the TBK1 phosphorylation sites
on OPTN, we performed stable isotope labeling
with amino acids in cell culture (SILAC)–based
mass spectrometry (fig. S5A). Phosphorylated
Ser
177
, adjacent to the OPTN LIR motif (Fig. 1A
and fig. S4A), showed substantially higher SILAC
Fig. 1. OPTN is an autophagy receptor. (A) Schematic representation of
OPTN’s domain architecture. An alignment of known LIR motifs is shown
underneath, with the tetra-peptide LIR highlighted in bold. N-terminal end of
the LIR, acidic residues (green), and potential phosphorylation sites (red) are
shown and are considered part of an extended LIR. CC, coiled-coil; aa, amino
acids; ZnF, zinc finger. (B) Directed yeast two-hybrid of bait proteins (pYTH9:
scAtg8, GABARAPL-1, LC3A, and SUMO1) and prey OPTN variants [pACT2:
full-length OPTN WT, LIR mutant (F178A) or ubiquitin binding mutant
(E478G)]. p62 LIR (aa 311 to 444) was also included. Interaction was
assessed using a b-galactosidase assay. (C) GST pull-down assay of EGFP-
OPTN from stable MCF-7 cells using GST or GST-ubiquitin like proteins. IB,
immunoblot. (D) Representative confocal images of HeLa cells overexpressing
EGFP-OPTN and treated with nutrient deprivation plus lysosomal bafilomycin
A1 (ND/BafA1) localized to endogenous LC3B and ubiquitin (inset). Scale bars,
10 mm. (E) Colocalization quantification of ≥1000 cells expressing mCherry-
LC3B and EGFP-OPTN [WT and deletion; DLIR (D178-181) or point mutation
F178A] within single cells by ImageStreamanalysis. Error bars indicate mean T
SD; n = 3 independent experiments. *P < 0.05; one-tailed, paired t test.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 229
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

intensity in cells coexpressing TBK1 in compari-
son with control (fig. S5B). Multiple phosphory-
lated LIR-peptides with up to three phosphorylated
groups were identified in the presence of over-
expressed TBK1. Endogenous OPTN was phos-
phorylatedat Ser
177
(measuredbyusing anti-pSer
177
OPTN antibody) (fig. S6A) after stimulation
of mouse embryonic fibroblasts (MEFs) with
microbe-derived lipopolysaccharide (LPS), which
Fig. 2. TBK1 phosphorylates OPTNwithin the extended LIR
motif. (A) MEFs were stimulated with 1 mg/ml LPS in the
presence or absence of 1 mM BX795 for the indicated time
points. Cell lysates were analyzed with antibodies against
OPTN, phospho-OPTN (Ser
177
), total IRF3, and phospho-
IRF3 (pSer
396
). pSer
177
OPTN was lost after inhibition of
TBK1 with BX795. (B) MEFs were left untreated or stim-
ulated with LPS for 30 min, lysed, and endogenous OPTN
precipitated with GST-LCB. (C) SILAC-labeled untreated or 30-min LPS-stimulated MEF cells and OPTN precipitated with GST-GABARAPL1 and analyzed by mass
spectrometry (MS). MS spectrum showed enriched phosphor-Ser
177
OPTN in LPS-stimulated cells compared with control. (D) ITC titration of LC3B with OPTN
peptides—namely LIR_P0 (no phosphorylation), LIR_P1 (pSer
177
) (middle), and LIR_Ptot (all serines phosphorylated, left)—corresponding to human OPTN
amino acids 169 to 184 (right). Raw data (upper boxes) and the integrated heat per titration step (points) and best-fit curves (lower panels) are shown.
Calculations assume a one-site binding model. The dissociation constant (K
d
), in the presence of phosphorylated OPTN serine residues showed a 5- to 13-fold
decrease indicating enhanced affinity of LC3B to OPTN LIR. Ka, acid constant; DH, change in enthalpy; DS, change in entropy.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 230
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

activates TBK1 via Toll-like receptor 4 (Fig. 2A).
OPTN and phospho-TBK1 (active) formed an
inducible complex with GST-LC3 (Fig. 2B). Next,
phosphorylated OPTNcomplexed with the autoph-
agy modifier GST-GABARAPL1 was analyzed
using SILAC-labeled MEFs from untreated (light)
or LPS-stimulated MEFs (heavy). The most abun-
dant phosphopeptide of OPTN identified was a
LIR spanning fragment, which carried a phosphate
group at mouse OPTN Ser
187
corresponding to hu-
man Ser
177
(Fig. 2C and fig. S4A). SILAC-based
relative quantification showed an increased intensity
of the peptide from LPS-stimulated cells in com-
parisontocontrol cells (no LPStreatment) (Fig. 2C).
Next, we investigated the potential impact of
phosphorylation of the LIRmotif on the interaction
of OPTN with LC3 proteins. Isothermal titration
calorimetry (ITC) experiments with LC3B and
three different peptides, spanning OPTN’s LIRmo-
tif and the preceding (N-terminal) serine-rich region,
representing different states of OPTNLIRphospho-
rylation (P0, no phosphorylation; P1, pSer
177
; and
Ptot, all five serines being phosphorylated) re-
vealed that the binding affinity of OPTN LIR to
LC3Bwas considerably enhanced by the presence
of phosphate groups (Fig. 2D). The dissociation
constant (K
d
) decreased from67 mMfor the non–
phosphorylated state (P0) to 13 mM for the mono-
phosphorylated form (P1) and to 5 mM for the
penta-phosphorylated form(Ptot) (Fig. 2D). Nuclear
magnetic resonance titration experiments revealed
that phospho-OPTN peptide does not substantially
change the binding mode, but that the presence of
phospho-serines preceding the LIR alters hydro-
gen bond formation and increases binding affin-
ity (fig. S7A).
Using phospho-mimicking OPTN mutants
with serine to aspartic acid (5xS->D) or glutamic
acid (5xS->E) and nonphosphorylatable alanine
(5xS->A), we tested the interaction of OPTNand
LC3B. Both phospho-mimicking versions of OPTN
bound to LC3B with a higher affinity than OPTN
wild type (fig. S7B). In contrast, the OPTN5xS->A
mutant was strongly impaired in its ability to bind
LC3B, demonstrating that additional negative
charges contribute to binding of OPTN to LC3B
in cells (fig. S7B). Accordingly, LC3Bcoimmuno-
precipitated more efficiently phosphorylated OPTN
(induced by TBK1 coexpression) than unmodi-
fied OPTN (fig. S7C).
TBK1 is activated by the cell wall compo-
nents of Gram-negative bacteria (such as LPS),
and TBK1 kinase activity is required to maintain
the integrity of Salmonella-containing vacuoles
(SCVs) and restrict Salmonella enterica serovar
Typhimurium (S. Typhimurium) growth in the
cytosol (17, 18). Upon infection, most Salmonella
reside in SCVs. However, a fraction of Salmonella
escape from the SCVs to the host cell cytosol
where they hyperproliferate, which has been ob-
served in vivo in the gall bladder epithelium of
infected mice and is likely to play an important
role in Salmonella dissemination to new hosts
(19). As a cellular defense mechanism, cytosolic
Salmonella is rapidly coated with ubiquitin (20)
and delivered to the autophagy clearance path-
way (17, 18). Therefore, we tested a potential role
for OPTN in regulating autophagy of cytosolic
Salmonella. In Salmonella-infected HeLa cells,
OPTNwas recruited to a subpopulation of cytosolic
bacteria stained with antibodies against LC3 and
ubiquitin (anti-LC3 and anti-ubiquitin) (Fig. 3A),
but not with LAMP1 (a marker for the SCV) (fig.
S8A). Aubiquitin-binding–deficient OPTNmutant
(DF474, 475NA) failed to be recruited to Salmo-
nella (fig. S8, Band D), whereas the LC3-binding–
deficient mutant OPTNF178Awas readilydetected
around Salmonella (fig. S8, C and D). To test
OPTN recruitment to cytosolic Salmonella, we
used an EGFP-expressing Salmonella sifA

mutant
that resides predominantly in the cytosol (21).
OPTN localized with TBK1, LC3, and ubiquitin
to cytosolic Salmonella (Fig. 3B), but not LAMP1,
a marker for Salmonella SCV (fig. S8E). OPTN
that was recruited to cytosolic Salmonella (both
WTand sifA) was phosphorylated on Ser
177
, as mon-
itored by an anti-pSer
177
OPTN antibody (Fig. 3,
A and B), and inhibition of TBK1 using BX795,
a specific inhibitor (19), led to loss of phospho-
rylated OPTN (pS177), but not total OPTN to
ubiquitin positive, cytosolic Salmonella (Fig. 3C
and fig. S8E). Thus, OPTN can be recruited to
ubiquitinated cytosolic Salmonella, with TBK1,
which subsequently phosphorylates OPTN, re-
sulting in enhanced LC3 binding. Consistent with
these observations, a nonphosphorylatable OPTN
mutant (5xS->A), when transfected in cells, failed
to direct LC3 to Salmonella as efficiently as a phos-
phomimicking OPTN variant (5xS->D) (fig. S8F).
Fig. 3. Phosphorylated OPTN colocalizes with ubiquitin- and LC3B-positive cytosolic
Salmonella. (A) Confocal images of HeLa cells expressing EGFP-OPTN and mCherry-LC3B
4 hours post infection (hpi) with S. Typhimurium (SL1344). EGFP-OPTN colocalizes with
LC3B (upper panel) on a fraction of Salmonella. Phospho-Ser
177
OPTN localizes to EGFP-
expressing Salmonella (MW57) with Ub at 1 hpi. Scale bar, 10 mm. (B) EGFP-OPTN
(green) and pS177 OPTN (purple) colocalized with TBK1 (upper panel), mCherry-LC3B
(middle panel), and ubiquitin (lower panel) at 4 hpi with cytosolic Salmonella sifA (blue). (C) Quantification of colocalization of cytosolic Salmonella sifA

with
EGFP-OPTN, ubiquitin, and pSer
177
OPTN from cells represented in (B), as compared with cells treated with 1 mM BX795. Error bars indicate mean T SEM.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 231
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Depletion of OPTN in Hela cells, either tran-
siently by small interfering RNA(siRNA) or stably
by short hairpin RNA (shRNA), resulted in en-
hanced proliferation of Salmonella as compared
with control siRNA or shRNA (Fig. 4, A to C).
Introduction of an shRNA-resistant OPTN gene
resulted in substantial suppression of intracellu-
lar Salmonella proliferation (Fig. 4C). However, a
ubiquitin-binding–deficient OPTN mutant (E478G)
or LIR mutant (F178A) was not able to rescue Sal-
monella suppression to the level of WT OPTN
(Fig. 4C). Thus, OPTNrequires both Ub and LC3-
binding domains to restrict bacterial growth in cells
and can therefore be classified as a bona fide au-
tophagy receptor for ubiquitinated bacteria.
Salmonella growth restriction mediated by
both autophagy receptors OPTN and NDP52 ap-
pears comparable (Fig. 4, A and B), and both pro-
teins colocalized to cytosolic bacteria (Fig. 3A)
(18, 22). Therefore, we tested whether they are
directed to the same or different populations of
cytosolic bacteria. OPTN targeted the same bacte-
ria as NDP52 and p62 (Fig. 4D) However, OPTN
and NDP52 colocalized to the same patches on
Salmonella, whereas OPTN and p62 were present
on different subdomains (Fig. 4D). Similarly,
NDP52 and p62 also formed nonoverlapping sub-
domains around the ubiquitinated bacteria, as ob-
served previously (23). Thus, NDP52 and OPTN
localize to common microdomains on the surface
of ubiquitinated bacteria separate from p62. Silenc-
ing either OPTN/NDP52 (Fig. 4A) or NDP52/p62
(23) had no additive effect in increasing bacterial
proliferation, indicating that autophagy receptors
(OPTN, NDP52, and p62) act along the same path-
way and are mutually dependent in promoting
autophagy of cytosolic bacteria.
Thus, OPTN appears to function in innate im-
munity against cytosolic bacteria by linking the
TBK1 signaling pathway to autophagic elimina-
tion of cytosolic pathogens. TBK1 appears to be
recruited to Salmonella via OPTNUBANdomain
binding to ubiquitin-decorated bacteria, where
TBK1 becomes activated and mediates phospho-
rylation of Ser
177
adjacent to the OPTN’s LIRmotif.
In such a model, phosphorylation of an autophagy
receptor acts as a molecular trigger to promote au-
tophagic clearance of cytosolic bacteria. Multi-
ple Salmonella-sensing receptors, including p62,
NDP52, and OPTN, bind to ubiquitin-coated Sal-
monella. The ability of their respective UBDs to
bind to different ubiquitin chains might account
for partitioning of autophagy receptors to different
subdomains on the bacterium (fig. S9). The iden-
tity of the E3 ubiquitin ligase and, consequently,
which type of ubiquitin chains are conjugated to
the bacterial surface components, is still unknown.
Thus, whether OPTN, p62, and NDP52 are re-
cruited to subdomains by interaction with differ-
ent ubiquitin chains or via other complexes on
Salmonella surface remains an open question.
This study also indicates more general and
diverse roles for phosphorylation in the control
of signaling networks mediated by ubiquitin or its
related modifiers. For example, small ubiquitin-like
modifier (SUMO) interaction with its respective
Fig. 4. Salmonella prolif-
eration is enhanced in the
absence of OPTN in vivo.
(A) HeLa cells transfected
with indicated siRNAs were
infected with Salmonel-
la (SL1344) and lysed at
the indicated time points
post-infection, and bacte-
rial colonies were counted
on selective agar plates.
(B) Numbers of bacteria
recovered from HeLa cells
transfected with the indi-
cated siRNAs and infected
with Salmonella for 8 hpi.
Intracellular Salmonella
replication was calculated
as fold increase at the 2-
hour time point. Depletion
of OPTN, NDP52, or TBK1
resulted in increased Sal-
monella intracellular rep-
lication. Error bars indicate mean T SD of n = 3 independent experiments. *P <
0.002, two-tailed t test. CFU, colony-forming units. (C) ShRNA OPTN-depleted
HeLa cells were transiently reconstituted with shRNA-resistant OPTN WT, shR-
OPTNF178A, and shR-OPTNE478G. Both LIR mutants and Ub-binding mutants
of OPTN failed to rescue OPTN-depleted HeLa cells 8 hpi compared with OPTN
WT. Error bars indicate mean T SD of n = 3 independent experiments. (D)
Three-dimensional reconstitution of confocal image z-stacks of Salmonella-
infected HeLa cells at 4 hpi. EGFP-OPTN WT (green), NDP52 (red, left panel),
and endogenous p62 (red right panel) form distinct “patches” on the surface
of cytosolic Salmonella.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 232
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

SUMO-interacting motif also enhances binding af-
finity (24), similar to LIRmodification. Notably, a
number of known autophagy receptors contain
conserved serine residues adjacent to their LIRs,
including NIXand NBR1, indicating a potentially
broader impact of phosphorylation of autophagy
processes. Interestingly, LC3A and LC3B have
phosphorylated serine/threonine residues in their
N-terminal extensions that are crucial for the inter-
action with LIR motifs (25, 26). Taken together,
phosphorylation of ubiquitin-like modifiers and
their binding domains brings another layer of com-
plexity in controlling ubiquitin and autophagy sig-
naling networks (27).
References and Notes
1. H. Nakatogawa, K. Suzuki, Y. Kamada, Y. Ohsumi,
Nat. Rev. Mol. Cell Biol. 10, 458 (2009).
2. Z. Yang, D. J. Klionsky, Curr. Opin. Cell Biol. 22, 124 (2010).
3. B. Levine, N. Mizushima, H. W. Virgin, Nature 469, 323
(2011).
4. V. Deretic, Curr. Opin. Cell Biol. 22, 252 (2010).
5. V. Kirkin, D. G. McEwan, I. Novak, I. Dikic, Mol. Cell 34,
259 (2009).
6. C. Kraft, M. Peter, K. Hofmann, Nat. Cell Biol. 12, 836
(2010).
7. D. G. McEwan, I. Dikic, Trends Cell Biol. 21, 195 (2011).
8. I. Novak et al., EMBO Rep. 11, 45 (2010).
9. V. Kirkin et al., Mol. Cell 33, 505 (2009).
10. Materials and methods are available as supporting
material on Science Online.
11. S. Pankiv et al., J. Biol. Chem. 282, 24131 (2007).
12. C. Behrends, M. E. Sowa, S. P. Gygi, J. W. Harper, Nature
466, 68 (2010).
13. Single-letter abbreviations for the amino acid residues
are as follows: A, Ala; D, Asp; E, Glu; F, Phe; G, Gly; I, Ile;
and S, Ser.
14. S. Wagner et al., Oncogene 27, 3739 (2008).
15. S. Morton, L. Hesson, M. Peggie, P. Cohen, FEBS Lett.
582, 997 (2008).
16. K. Clark, L. Plater, M. Peggie, P. Cohen, J. Biol. Chem.
284, 14136 (2009).
17. A. L. Radtke, L. M. Delbridge, S. Balachandran,
G. N. Barber, M. X. O’Riordan, PLoS Pathog. 3, e29
(2007).
18. T. L. Thurston, G. Ryzhakov, S. Bloor, N. von Muhlinen,
F. Randow, Nat. Immunol. 10, 1215 (2009).
19. L. A. Knodler et al., Proc. Natl. Acad. Sci. U.S.A. 107,
17733 (2010).
20. A. J. Perrin, X. Jiang, C. L. Birmingham, N. S. So,
J. H. Brumell, Curr. Biol. 14, 806 (2004).
21. C. R. Beuzón et al., EMBO J. 19, 3235 (2000).
22. Y. T. Zheng et al., J. Immunol. 183, 5909 (2009).
23. M. Cemma, P. K. Kim, J. H. Brumell, Autophagy 7, 341
(2011).
24. P. Stehmeier, S. Muller, Mol. Cell 33, 400 (2009).
25. H. Jiang, D. Cheng, W. Liu, J. Peng, J. Feng, Biochem.
Biophys. Res. Commun. 395, 471 (2010).
26. S. J. Cherra 3rd et al., J. Cell Biol. 190, 533 (2010).
27. F. Ikeda, N. Crosetto, I. Dikic, Cell 143, 677 (2010).
Acknowledgments: We thank P. Cohen, S. Mueller,
K. Rajalingam, C. Behrends, and the members of
Dikic laboratory for constructive comments and critical
reading of the manuscript; P. Cohen for OPTN and
TBK1 reagents; and V. Kirkin for the initial yeast
two-hybrid screens. This work was supported by grants
from Deutsche Forschungsgemeinschaft, the Cluster
of Excellence “Macromolecular Complexes” of the
Goethe University Frankfurt (EXC115), the Landes-Offensive
zur Entwicklung Wissentschaftlich-ökonomischer
Exzellenz–funded Onkogene Signaltransduktion Frankfurt
network, and the European Research Council (ERC)
under the European Union’s Seventh Framework
Programme (FP7/2007-2013)/ERC grant agreement
no. (250241-LineUb) to I.D. and partly by Swiss
National Fonds (3100A0-121834/1) to D.B. The
Center for Protein Research is funded by a generous
grant from the Novo Nordisk Foundation. J.K. is supported
by a scholarship from the Split, Croatia, government, S.W.
by a postdoctoral fellowship from the Danish Council for
Independent Research (FSS: 10-085134), N.R.B. by the
Initiative and Networking Fund of the Helmholtz
Association, and H.F. by Swiss National Science Foundation
(31003A-121834).
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1205405/DC1
Materials and Methods
Figs. S1 to S9
Table 1
References
10 March 2011; accepted 17 May 2011
Published online 26 May 2011;
10.1126/science.1205405
Adipose Triglyceride Lipase Contributes
to Cancer-Associated Cachexia
Suman K. Das,
1
Sandra Eder,
2
Silvia Schauer,
1
Clemens Diwoky,
3
Hannes Temmel,
1
Barbara Guertl,
1
Gregor Gorkiewicz,
1
Kuppusamy P. Tamilarasan,
1
Pooja Kumari,
1,4
Michael Trauner,
4
Robert Zimmermann,
2
Paul Vesely,
1
Guenter Haemmerle,
2
Rudolf Zechner,
2
* Gerald Hoefler
1
*
Cachexia is a multifactorial wasting syndrome most common in patients with cancer that is
characterized by the uncontrolled loss of adipose and muscle mass. We show that the inhibition
of lipolysis through genetic ablation of adipose triglyceride lipase (Atgl) or hormone-sensitive
lipase (Hsl) ameliorates certain features of cancer-associated cachexia (CAC). In wild-type
C57BL/6 mice, the injection of Lewis lung carcinoma or B16 melanoma cells causes tumor
growth, loss of white adipose tissue (WAT), and a marked reduction of gastrocnemius muscle.
In contrast, Atgl-deficient mice with tumors resisted increased WAT lipolysis, myocyte apoptosis,
and proteasomal muscle degradation and maintained normal adipose and gastrocnemius
muscle mass. Hsl-deficient mice with tumors were also protected although to a lesser degree.
Thus, functional lipolysis is essential in the pathogenesis of CAC. Pharmacological inhibition
of metabolic lipases may help prevent cachexia.
C
achexia (kakos hexis, Greek for “bad con-
dition”) is a devastating syndrome that
frequently occurs in patients suffering from
chronic infection, such as tuberculosis or AIDS,
and other diseases, including chronic obstructive
pulmonary disease, chronic kidney disease, and
chronic heart failure. Most commonly, however,
cachexia is observed in cancer. The highest
frequency of cancer-associated cachexia (CAC)
occurs in pancreatic and gastric cancer (1–3).
CAC is an important adverse prognostic factor
and the immediate cause of death in an estimated
15% of all cancer patients (3–5). Wasting results
fromdepletion of both adipose tissue and skeletal
muscle (6, 7). In contrast to starvation, the non-
muscle protein compartment of the body is rela-
tively unaffected in CAC patients (7), implying a
tumor-associated metabolic condition that spe-
cifically targets adipose tissue and muscle. Thus,
anorexia is unlikely to be solely responsible for
the loss of skeletal muscle in patients with CAC.
Indeed, nutritional supplementation has largely
failed to reverse the wasting process (8).
The pathogenesis of CAC is multifactorial
(9). Central mechanisms regulate appetite, food
intake, and energy consumption. Contributing
peripheral mechanisms control lipid and carbo-
hydrate metabolism in various tissues. Severe
lipid loss in CAC is driven by changes in lipid
catabolism (10–15) and, possibly, lipogenesis
(16). The concept that increased triacylglycerol
(TG) degradation may contribute decisively to
CAC is strongly underscored by increased plasma
levels of fatty acids (FAs) and glycerol; increased
lipolytic rates upon epinephrine stimulation; and
increased expression of lipid-mobilizing factors,
such as zinc-a2 glycoprotein-1 (AZGP1), tumor
necrosis factor a (TNF-a), and interleukins
(IL)-1 and -6 (17, 18). The breakdown of fat re-
quires lipolysis of TG stored in cellular lipid
droplets and is mediated by adipose triglyceride
lipase (ATGL) and hormone-sensitive lipase
(HSL) (19). This led to our hypothesis that dis-
ruption of fat catabolism may prevent the initia-
tion and/or progression of CAC. In vertebrates,
lipolysis is most active in adipose tissue, with
ATGL predominantly responsible for the initial
step of TGhydrolysis (formation of diacylglycerol)
and HSL for the hydrolysis of diacylglycerol. We
investigated whether one or both of these lipases
are essential for CAC.
To assess the role of lipases in CAC, we used
two different cachexia models in mice lacking
1
Institute of Pathology, Medical University of Graz, 8036 Graz,
Austria.
2
Institute of Molecular Biosciences, University of Graz,
8010 Graz, Austria.
3
Institute of Medical Engineering, Graz
University of Technology, 8010 Graz, Austria.
4
Division of
Gastroenterology and Hepatology, Department of Internal
Medicine, Medical University of Graz, 8036 Graz, Austria.
*To whom correspondence should be addressed. E-mail:
rudolf.zechner@uni-graz.at (R.Z.); gerald.hoefler@medunigraz.
at (G.H.)
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 233
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

either Atgl or Hsl. Cachexia was induced in wild-
type C57BL/6 (WT) mice, Atgl-deficient (Atgl
−/−
)
mice (20), and Hsl-deficient (Hsl
−/−
) mice (21) by
subcutaneous injection of Lewis lung carcinoma
(LLC) cells (22) or B16 melanoma cells (23).
Tumor growth was observed in 100% of treated
animals. Tumor weights tended to be lower in
nonfasted (Fig. 1, A to C) and overnight (o/n)-
fasted (fig. S1, A and B) (24) Atgl
−/−
and Hsl
−/−
mice than in WT mice. However, none of the
differences reached statistical significance.
Next we analyzed body weight, plasma me-
tabolite concentrations, and fat mass in mice of
all genotypes with or without tumors. Because
all of these parameters strongly depend on the
feeding status of the mice and to account for any
nutritional bias, they were determined in non-
fasted and o/n-fasted animals. Total body weight
(after subtraction of tumor weight) differed dras-
tically in lipase-deficient mouse models com-
pared with WT mice in response to LLCand B16
(Fig. 1, Dto F). Whereas non–tumor-bearing WT
mice on normal chow diet gained weight over
a period of 3 weeks, WT mice with LLC started
to lose weight 14 days after tumor injection, re-
sulting in an average weight loss of 1.8 g after
21 days. In contrast, body weight in Atgl
−/−
mice
with LLCwas identical to Atgl
−/−
without tumors
at all times. Hsl
−/−
mice exhibited an intermediate
phenotype. Compared to non–tumor-bearing Hsl
−/−
mice, body weight was reduced. However, the
loss was less extreme than in WT mice carrying
the tumor. Similar results were obtained in B16-
treated mice (Fig. 1 F). Compared with WT mice
without tumors, B16-treated mice weighed 3.3 g
less 16 days after tumor injection. In contrast,
B16-treated Atgl
−/−
mice maintained weights sim-
ilar to those of untreated Atgl
−/−
mice. Hsl
−/−
ani-
mals were less protected than Atgl
−/−
mice and
lost on average 2.7 g of body weight. The differ-
ences were even more pronounced in o/n-fasted
mice, a condition when lipolysis is physiologi-
cally induced (fig. S2). Whereas WT mice with
LLC weighed 2.1 g (after 14 days) and 5.5 g
(after 21 days) less than WT mice without tu-
mors, Atgl
−/−
mice were totally resistant to weight
loss and Hsl
−/−
animals reached intermediate
values. Differences in weight loss in response
to the tumors were not explained by variable food
intake (fig. S3, A and B), because it was similar
in all animals during the initial phase of the
experiment and decreased uniformly in all
tumor-carrying mice during the final 2 to 4 days.
Thus, in the mouse, protection from CAC-
associated weight loss can be entirely conferred
by the lack of ATGL and partially by the absence
of HSL.
Fig. 1. Ablation of Atgl
protects mice from cancer-
associated weight loss and
cancer-associated loss of
adipose tissue. (A to C)
Tumor weights 14 days
(d) and 21 days after in-
jecting LLC and 16 days
after injecting B16 mel-
anoma (B16) cells were
slightly lower in lipase-
deficient mice compared
with WT. (D to F) WT
mice significantly lost
weight with tumor pro-
gression after injection
of LLC and B16 tumor
cells, whereas Atgl
−/−
ani-
mals did not develop ca-
chexia. Hsl
−/−
mice also
lost weight but less than
WT mice did. (G) Nor-
malized gonadal and epi-
didymal WAT was reduced
by about 55% in non-
fasted tumor-bearing
WT mice 21 days after
LLC tumor implantation,
whereas lipase-deficient
mice were protected from
WAT loss. (H to J) No or
minimal gonadal and
epididymal WAT was de-
tected upon dissection of
nonfasted WT B16 tumor-
bearing mice after 16 days and fasted WT LLC tumor-bearing mice after
21 days compared to saline injected control mice. Atgl
−/−
tumor-bearing
mice retained gonadal and epididymal WAT, whereas partial loss of gonadal
and epididymal WAT was observed in Hsl
−/−
tumor-bearing mice. WAT was
dissected, and its mass normalized to tibia length as described in (24). Black
bars indicate control (saline-injected) animals; white bars, tumor-bearing
animals. To allow direct comparison, values were determined after removal
of the respective tumor. ***P < 0.001, **P < 0.01, *P < 0.05, n = 5 to 7
except for (A) to (E) and (G) LLC, Hsl
−/−
, n = 3 and (D) LLC, 14 days, normal,
n = 2. ( J) Scale bars represent a length of 1 mm.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 234
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Consistent with our previous findings (20, 21),
plasma glucose, FA, and TG levels were reduced
in o/n-fasted Atgl
−/−
versus WT mice (figs. S4 to
S6). In o/n-fasted Hsl
−/−
mice, FA and TG levels
were also reduced (although less than in Atgl
−/−
mice), whereas plasma glucose concentrations
were unchanged compared with those of WT
mice (figs. S4 to S6). The presence of LLC for
14 days did not affect plasma glucose levels in o/n-
fasted mice of any genotype (fig. S4A). After
21 days, plasma glucose levels were decreased in
tumor-bearing, o/n-fasted WT(–43.2%) and Hsl
−/−
mice (–27.3%) but remained unchanged in Atgl
−/−
mice when compared to mice of the same geno-
type without tumors (fig. S4B). These differences
in glucose levels in LLC-treated WT and Hsl
−/−
mice were not observed in nonfasted animals
(fig. S4C). Similarly, plasma glucose and FA lev-
els of nonfasted B16-treated mice matched those
in untreated animals (fig. S4D). Presence of LLC
caused an increase in FA levels in WT (fasted:
+24.0% at 14 days and +9.9% at 21 days; non-
fasted: +54.5%at 21 days) and Hsl
−/−
mice (fasted:
+23.4%at 14 days and +25.0%at 21days) (fig. S5).
In contrast, FA levels in tumor-bearing Atgl
−/−
mice remained unchanged compared with those
of Atgl
−/−
mice without tumors independent of
the feeding status. Increased FA levels were also
observed in nonfasted B16-tumor-bearing WT
mice (+27%) (fig. S5D). Serum TG levels were
not significantly different in o/n-fasted animals
with or without tumors (fig. S6).
To assess the contribution of adipose tissue
loss to the tumor-induced weight loss, we deter-
mined white adipose tissue (WAT) mass by vi-
sual inspection, weighing of surgically removed
adipose depots (gonadal and epididymal adi-
pose tissue), and in vivo nuclear magnetic res-
onance (NMR) WAT quantitation. In nonfasted
Fig. 2. Loss of WAT in tumor-bearing animals is mainly attributable to
lipolysis. Both LLC and B16 tumors significantly increased FAs (A and C) and
glycerol (B and D) release from WAT explants in nonfasted tumor-bearing
WT mice. LLC-bearing lipase-deficient mice did not exhibit increased FA or
glycerol release from WAT explants, whereas increased FA and glycerol re-
lease was observed from WAT explants of B16-tumor-bearing Hsl
−/−
mice.
WAT explants from B16-tumor-bearing Atgl
−/−
mice did not show increased
FA or glycerol release. (E to J) Lipolytic agonists (TNF-a, IL-6, and AZGP1)
were increased in both LLC- and B16-tumor–bearing mice from all geno-
types. Black bars indicate control (saline-injected) animals; white bars, tumor-
bearing animals. ***P < 0.001, **P < 0.01, *P < 0.05, n = 7 except for (A)
and (B), n = 3 to 5.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 235
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Fig. 3. Atgl
−/−
micewere
protected from tumor-
associated skeletal mus-
cle loss. (Aand D) Muscle
mass of gastrocnemius
(normalized to tibia
length) of surgically pre-
pared gastrocnemius il-
lustrate skeletal muscle
loss in tumor-bearing
WT and Hsl
−/−
mice but
not in Atgl
−/−
mice. (B
and E) Proteasome ac-
tivity in gastrocnemius
muscle homogenates is
significantly increased
in tumor-bearing WT and
Hsl
−/−
mice but not in
Atgl
−/−
mice. (C and F) In-
creased caspase 3 and 7
activity in gastrocnemius
muscle homogenates dem-
onstrates activation of
apoptosisintumor-bearing
WT and Hsl
−/−
mice. Caspase 3 and 7 activity in gastrocnemius muscle of Atgl
−/−
mice was not affected by tumor growth. Black bars indicate control (saline-injected)
animals, white bars, tumor-bearing animals. ***P < 0.001, **P < 0.01, *P < 0.05, n = 7.
Fig. 4. CAC patients show increased TG hydrolase activity compared with
noncachectic patients. Total lipase activity (A), specifically inhibited (HSL,
76-0079) lipase activity (mainly ATGL activity) (B), and HSL activity
(determined by subtraction of HSL-inhibited lipase activity from total lipase
activity) (C) in visceral WAT of cancer patients compared with those of
noncancer patients. Lipase activities in WAT of cachectic cancer patients are
significantly higher than in noncachectic cancer patients. Ranges indicate
mean T standard deviation. (D to F) Total, HSL-inhibited, and HSL lipase
activities show negative correlation with BMI of cancer patients. FFA, free
fatty acid.
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 236
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

LLC-bearing WT mice, WATweight decreased
by 55% after 21 days of tumor growth (Fig. 1G),
which corresponded to a loss of 1.7 g of WAT
in NMR analysis (fig. S7A). Nonfasted B16-
bearing WT mice lost 85% of WAT weight after
16 days of tumor growth (Fig. 1H) (–2.0 g of
adipose tissue mass in NMR analysis, fig. S7B).
In contrast, none of the tumors affected WAT
mass of Atgl
−/−
mice. In fact, weight of body fat
depots and total body fat was increased in Atgl
−/−
mice independent of the presence or absence of
tumors when compared to non–tumor-bearing
WT mice (Fig. 1, G and H, and fig. S7). This
confirms our earlier observation that Atgl defi-
ciency causes obesity in mice kept on a normal
chow diet (20) and shows that WAT mass is in-
dependent of the presence of the tumor. In
nonfasted Hsl
−/−
mice, LLC did not affect the
weight of WAT depots. However, NMR analyses
revealed a total WAT reduction by 0.7 g in (Fig.
1G and fig. S7A). B16 in Hsl
−/−
mice caused a
32% reduction of adipose tissue weight and a
0.9 g WAT loss in NMR analysis (Fig. 1H and
fig. S7B). In o/n-fasted animals, the differences
were even more striking (Fig. 1, I and J, and fig.
S7C). After 21 days, LLC tumors caused the loss
of more than 95% of WAT weight (2 g in NMR
analysis) in WT mice, whereas adipose mass was
again completely retained in LLC-treated Atgl
−/−
mice. Hsl
−/−
mice exhibited an intermediate loss
of 37% of WATweight. Results for LLC-bearing
animals were substantiated by magnetic reso-
nance imaging analysis. Taken together, these
results show that, independently of feeding status
and tumor type, ATGLdeficiency completely and
HSL deficiency partially protects mice from the
loss of WAT.
Tumor-associated loss of adipose tissue in
animal models has been mostly attributed to an
increase in WAT lipolysis (10–16). Consistent
with these reports, the release of FAs and glycerol
from WAT explants was increased in WT mice
with LLC (38% and 31%, respectively) and B16
(39% and 21%, respectively) compared with
mice without tumors (Fig. 2, A to D). This in-
crease in lipolysis did not occur in LLC- or B16-
tumor–bearing Atgl
−/−
mice. WAT lipolysis was
also attenuated in tumor-bearing Hsl
−/−
mice.
Whereas FA and glycerol release from WATwas
similar in Hsl
−/−
mice with or without LLC, B16
melanoma formation caused 28% and 19%
increases, respectively. To investigate whether
changes in TG hydrolase activities underlie the
observed differences in lipolytic rates, we mea-
sured ATGL and HSL enzyme activities in WAT
in response to tumor growth (fig. S8). LLC in
o/n-fasted WT mice caused a twofold increase in
total lipase activity because of increased ATGL
and HSL activity. No tumor-induced increase in
WATlipase activity was observed in Atgl
−/−
mice,
whereas in Hsl
−/−
mice total TG lipase activity
increased by 2.1-fold because of increased ATGL
activity. Thus, LLC and B16 cause an induction
of WAT TG hydrolase activity, leading to an
increased release of FAs and glycerol from WAT.
This induction of lipolysis is not observed in
the absence of ATGL and significantly reduced
in the absence of HSL. Thus, lipase deficiency
blocks tumor-induced WAT loss.
Several reports have shown that the induc-
tion of lipolysis during CACis mediated by inflam-
matory cytokines (17), AZGP1 (also designated
lipid mobilizing factor, LMF) (18), or cytotoxin-
induced death executor proteins (CIDE), such
as CIDE-A (25). To investigate whether these
lipolytic agonists are increased in cachexia mod-
els lacking specific lipases, we assessed their
plasma levels. In response to LLC, TNF-a and
IL-6 levels were increased (between 2- to 3.2-
fold) in all genotypes (Fig. 2, E and F). Similarly,
quantitative Western blot analysis revealed that
plasma AZGP1 levels were higher in LLC-
bearing animals of all genotypes (Fig. 2G). In
response to B16, the induction of cytokine re-
lease was even more pronounced (Fig. 2, H and
I). Plasma levels of TNF-a and IL-6 increased
about five- to ninefold and 18- to 21-fold, respec-
tively. Plasma AZGP1 levels were also consist-
ently higher in B16-treated mice of all genotypes
compared with untreated mice (Fig. 2J). This
suggests that in WT mice the increased concen-
tration of inflammatory and lipolytic agonists
induce lipolysis via ATGL and HSL leading to
the uncontrolled loss of WAT and cachexia. In
the absence of lipases, particularly in the absence
of ATGL, this process is disrupted and WAT is
retained.
In cancer patients, CAC not only emaciates
adipose tissue but also consumes skeletal muscle
and cardiac muscle (9, 22, 26, 27). Similarly, we
observed that LLC and B16 melanoma in mice
lowers skeletal muscle mass and heart weight.
Surgically removed gastrocnemius muscle of
WT mice injected with LLC (21 days) and B-16
(16 days) weighed 36% and 25% less than that
of WT mice, respectively (Fig. 3, A and D), sug-
gesting that muscle loss was less pronounced
than the loss of adipose mass. Remarkably, Atgl
−/−
mice suffered no significant loss of gastrocnemius
muscle weight in response to the tumors (Fig. 3, A
and D). Similarly as in WT mice, LLCand B16 in
Hsl
−/−
mice diminished gastrocnemius muscle
weight (Fig. 3, A and D), albeit to a lesser degree
(–27% and –18%, respectively). Wasting of
gastrocnemius muscle was also reflected by a
reduction of total muscle protein in WT (–36%)
and Hsl
−/−
mice (–22%) in response to B16, where-
as Atgl
−/−
mice with B16 melanoma were re-
sistant to the loss of muscle protein (fig. S9). The
weight of soleus muscle also decreased in LLC-
injected WT mice after 21 days (33%, not sta-
tistically significant) and B16-injected WT mice
(–27%) (fig. S10). Both LLC and B16 caused a
moderate decrease in heart weight (–7.4% and
–9.9%, respectively) and total cardiac protein
content (–20.1% and –24.9%, respectively) in
WT mice but not in Atgl
−/−
or Hsl
−/−
mice (fig.
S11). Consistent with our previous observations
(20), Atgl
−/−
mice exhibited a twofold increased
TG content in gastrocnemius muscle and a 12- to
15-fold increased TG content in cardiac muscle
(fig. S12) compared with WT animals.
Muscle atrophy may originate from a de-
crease in protein synthesis (28) or an increase
in protein degradation (9, 22). Studies in a num-
ber of experimental models of cachexia suggest
that both processes occur simultaneously (9).
Animal models of cancer cachexia as well as
studies in cancer patients provided evidence
that the ubiquitin-proteasome pathway mainly
degrades myofibrillar proteins, particularly at
later stages of cachexia when patients lost more
than 10% of their body weight (29). Addition-
ally, apoptotic cell death characterized by in-
creased activity of caspases contributes to the
loss of gastrocnemius muscle in mice bearing
the cachexia-inducing MAC16 tumor (30). Con-
sistent with these observations, we detected a
marked increase in proteasome activity (Fig. 3,
B and E) and an increase in caspase 3 and 7 ac-
tivity (Fig. 3, Cand F) in gastrocnemius muscle
of WT mice and Hsl
−/−
mice 21 days after LLC
injection and 16 days after B16 injection. In
contrast, no significant change in proteasome
or caspase 3 and 7 activity was observed in gas-
trocnemius muscle of Atgl
−/−
mice in response
to both tumors (Fig. 3 B, C, E, and F). Changes
in the weight of gastrocnemius muscle, caspase
3 and 7 activity, and proteasome activity were not
observed in WT and Hsl
−/−
mice, 2 weeks after
LLC injection (fig. S13), suggesting that loss of
WAT precedes the loss of muscle mass, which
is consistent with earlier observations (10).
Previous work showed that cachexia is as-
sociated with increased FA oxidation in gas-
trocnemius muscle (9). Our data confirm these
findings, showing 1.8- to 2.5-fold increased
mRNA expression levels of genes involved in
the regulation of cellular FA uptake (CD36 and
fatty acid transport protein 1, Fatp-1), FA transport
into mitochondria (carnitin palmitoyltransferase-1b,
Cpt-1b), and mitochondrial function (peroxisome
proliferator-activated receptor-g coactivator-1a,
Pgc-1a) in gastrocnemius muscle samples of
WT mice with LLC (fig. S14). mRNA levels in
muscle samples of Atgl
−/−
mice were not affected
in response to LLC. In Hsl
−/−
mice, mRNAlevels
increased in the presence of the tumor, although
to a lesser extent than in WT mice. This sug-
gests that the catabolic state in CACmobilizes FA
from adipose tissue, leading to an energy sub-
strate switch from glucose to FA use in skeletal
muscle.
To test whether the activity of metabolic
lipases in WAT also associates with CAC in hu-
mans, we determined ATGL- and HSL-mediated
TG hydrolase activities of visceral WAT from
autopsy samples of 27 patients. Twelve of these
individuals had been diagnosed with various
forms of malignancies (two adenocarcinomas of
the lung, two adenocarcinomas of the colon, two
ductal adenocarcinomas of the breast, two adeno-
carcinomas of the prostate, one hepatocellular
carcinoma, one clear cell carcinoma of the kidney,
one squamous cell carcinoma of the esophagus,
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 237
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

and one malignant germ cell tumor). Six out of
the 12 patients were designated as cachectic ac-
cording to the definition of Evans et al. (31).
Total lipase, ATGL, and HSL activities were
significantly higher in visceral WAT of cancer
patients compared with individuals without can-
cer and significantly higher in cancer patients
with cachexia compared with cancer patients
without cachexia (Fig. 4, A to C). Lipase ac-
tivities in cancer patients without cachexia were
similar to those of noncancer patients. A signif-
icant inverse correlation was found between
total lipase, ATGL, and HSL activities in WAT
of cancer patients and their body mass index
(BMI) (Fig. 4, D to F). In contrast, lipolytic ac-
tivities in WAT of noncancer patients showed
no correlation with their BMI (fig. S15). Thus,
our study provides compelling evidence that
the previously observed increase in FA and glyc-
erol release fromWATof patients with CAC(9, 10)
is due to up-regulation of ATGL and HSL acti-
vities and that increased lipase activities strong-
ly correlate with cachexia.
In summary, our data are consistent with the
view that lipolysis plays an instrumental role in
the pathogenesis of CAC. The increased catabo-
lism of adipose lipid stores leads to the complete
loss of WAT followed by a reduction in muscle
mass. The absence of ATGL and, to a lesser de-
gree, HSL reduces FA mobilization, retains WAT
and muscle mass, and prevents CAC. Whether
the protection of adipose and muscle loss in
lipase-deficient mice is a consequence of defec-
tive tissue autonomous lipolysis or due to endo-
crine signaling from the tumor or WAT remains
to be elucidated. However, pharmacological
inhibition of lipases may represent a powerful
strategy to avoid the devastating condition of
cachexia in response to cancer or other chronic
diseases.
References and Notes
1. M. J. Tisdale, Nat. Rev. Cancer 2, 862 (2002).
2. J. E. Morley, D. R. Thomas, M. M. Wilson, Am. J. Clin.
Nutr. 83, 735 (2006).
3. W. D. Dewys et al., Am. J. Med. 69, 491 (1980).
4. C. Deans, S. J. Wigmore, Curr. Opin. Clin. Nutr.
Metab. Care 8, 265 (2005).
5. K. C. Fearon, A. G. Moses, Int. J. Cardiol. 85, 73
(2002).
6. M. Fouladiun et al., Cancer 103, 2189 (2005).
7. K. C. Fearon, Proc. Nutr. Soc. 51, 251 (1992).
8. M. Lainscak, G. S. Filippatos, M. Gheorghiade,
G. C. Fonarow, S. D. Anker, Am. J. Cardiol. 101, 8E
(2008).
9. M. J. Tisdale, Physiol. Rev. 89, 381 (2009).
10. T. Agustsson et al., Cancer Res. 67, 5531 (2007).
11. A. Hyltander, P. Daneryd, R. Sandström, U. Körner,
K. Lundholm, Eur. J. Cancer 36, 330 (2000).
12. S. Klein, R. R. Wolfe, J. Clin. Invest. 86, 1403
(1990).
13. A. Legaspi, M. Jeevanandam, H. F. Starnes Jr.,
M. F. Brennan, Metabolism 36, 958 (1987).
14. M. Rydén et al., Cancer 113, 1695 (2008).
15. J. H. Shaw, R. R. Wolfe, Ann. Surg. 205, 368
(1987).
16. M. Jeevanandam, G. D. Horowitz, S. F. Lowry,
M. F. Brennan, Metabolism 35, 304 (1986).
17. J. M. Argilés, S. Busquets, M. Toledo, F. J. López-Soriano,
Curr. Opin. Support. Palliat. Care 3, 263 (2009).
18. C. Bing et al., Proc. Natl. Acad. Sci. U.S.A. 101,
2500 (2004).
19. R. Zechner, P. C. Kienesberger, G. Haemmerle,
R. Zimmermann, A. Lass, J. Lipid Res. 50, 3 (2009).
20. G. Haemmerle et al., Science 312, 734 (2006).
21. G. Haemmerle et al., J. Biol. Chem. 277, 4806
(2002).
22. M. van Royen et al., Biochem. Biophys. Res. Commun.
270, 533 (2000).
23. I. Kawamura et al., Anticancer Res. 19, 341
(1999).
24. Materials and methods are available as supporting
material on Science Online.
25. J. Laurencikiene et al., Cancer Res. 68, 9247
(2008).
26. X. Zhou et al., Cell 142, 531 (2010).
27. S. Busquets et al., Clin. Nutr. 26, 239 (2007).
28. M. J. Rennie et al., Clin. Physiol. 3, 387 (1983).
29. J. Khal, A. V. Hine, K. C. Fearon, C. H. Dejong,
M. J. Tisdale, Int. J. Biochem. Cell Biol. 37, 2196 (2005).
30. J. E. Belizário, M. J. Lorite, M. J. Tisdale, Br. J. Cancer 84,
1135 (2001).
31. W. J. Evans et al., Clin. Nutr. 27, 793 (2008).
Acknowledgments: We thank E. Zechner and C. Schober-
Trummler for reviewing the manuscript. The research
was supported by the doctoral program Molecular
Medicine of the Medical University of Graz (S.D.);
GOLD, Genomics of Lipid-Associated Disorders as
part of the Austrian Genome Project GEN-AU funded by
Forschungsförderungsgesellschaft and Bundesministerium
für Wissenschaft und Forschung (Ru.Ze.); SFB LIPOTOX
grant no. F30 (Ru.Ze., G.H.), the Wittgenstein Award 2007
grant no. Z136 funded by the Austrian Fonds zur Förderung
der Wissenschaftlichen Forschung (Ru.Ze.). S.K.D., Ro.Zi.,
G.H., and Ru.Ze. hold a patent related to the modulation
of ATGL for prevention and treatment of cachexia.
Supporting Online Material
www.sciencemag.org/cgi/content/full/science.1198973/DC1
Materials and Methods
Figs. S1 to S15
References
12 October 2010; accepted 27 May 2011
Published online 16 June 2011;
10.1126/science.1198973
A Pericyte Origin of Spinal
Cord Scar Tissue
Christian Göritz,
1
David O. Dias,
1
Nikolay Tomilin,
2
Mariano Barbacid,
3
Oleg Shupliakov,
2
Jonas Frisén
1
*
There is limited regeneration of lost tissue after central nervous system injury, and the lesion is
sealed with a scar. The role of the scar, which often is referred to as the glial scar because of
its abundance of astrocytes, is complex and has been discussed for more than a century. Here we
show that a specific pericyte subtype gives rise to scar-forming stromal cells, which outnumber
astrocytes, in the injured spinal cord. Blocking the generation of progeny by this pericyte subtype
results in failure to seal the injured tissue. The formation of connective tissue is common to
many injuries and pathologies, and here we demonstrate a cellular origin of fibrosis.
M
ost studies on the scar tissue that forms
at injuries in the central nervous sys-
tem (CNS) have focused on astrocytes,
and it is often referred to as the glial scar (1–5).
There is also a connective tissue or stromal, non-
glial, component of the scar (6–10), but it has
received much less attention. The generation of
connective tissue, with large numbers of fibroblasts
depositing extracellular matrix (ECM) proteins, is
a general feature of scarring and fibrosis in all
organs and in diverse types of pathology (11). In
spite of being a major clinical problem that has
been extensively studied, the origin of scar-forming
fibroblasts has been difficult to establish. Most
studies have suggested that they may derive from
circulating cells, proliferating resident fibroblasts,
endothelial cells, or epithelial cells (12–14). There
are also data indicating that pericytes, perivascu-
lar cells enwrapping the endothelial cells of cap-
illaries, may differentiate into collagen-producing
cells in models of dermal scarring and in kidney
fibrosis (15–17).
We have explored the role of pericytes in
scar formation after spinal cord injury. We found
that Glast-CreER transgenic mice (18) enabled
recombination of the R26R-yellow f luorescent
protein (R26R-YFP) reporter allele (19) in a sub-
set of pericytes lining blood vessels in the spinal
cord parenchyma, which allowed us to stably and
heritably label these cells (20) (Fig. 1 and figs. S1
to S5). The recombined cells had the typical ul-
trastructural features of pericytes (21), including
being encased in the vascular basal lamina, which
separates them from endothelial cells and astro-
cytes (Fig. 1, A to D). The recombined cells
represent a distinct pericyte subpopulation that
constitutes ~10% of all pericytes in the adult
spinal cord [assessed by electron microscopy
(EM)]. At positions where processes intersect,
the Glast-CreER–expressing pericytes were in-
variably located abluminal to the other pericyte
subtype (Fig. 1A and fig. S6). We refer to the
pericyte subclass that is recombined in Glast-
CreER mice as type A pericytes and the other
1
Department of Cell and Molecular Biology, Karolinska Insti-
tute, SE-171 77 Stockholm, Sweden.
2
Department of Neuro-
science, Karolinska Institute, SE-171 77 Stockholm, Sweden.
3
Molecular Oncology Programme, Centro Nacional de Inves-
tigaciones Oncológicas, 28029 Madrid, Spain.
*To whom correspondence should be addressed. E-mail:
jonas.frisen@ki.se
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 238
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

subclass as type B pericytes. Most or all CNS
pericytes express platelet-derived growth factor
receptor (PDGFR) a and b; and CD13 (Fig. 1, E
and F, and fig. S3) (22–24), but the expression of
some other markers is heterogeneous (25). Type
B pericytes can be distinguished by the expres-
sion of desmin and/or alpha smooth muscle actin
(Fig. 1G and fig. S4).
To address whether pericytes participate in
scar formation after spinal cord injury, we gen-
etically labeled type A pericytes before a dorsal
funiculus incision or a dorsal hemisection and
assayed the fate of recombined cells for up to
7 months after the lesion (fig. S1B). The injury
was made following a 7-day clearing period
without tamoxifen, which ensures that all re-
combination occurs before the insult, so that
even if cells other than type A pericytes would
start to express the Glast-CreER transgene in re-
sponse to the injury, it would not result in re-
combination (26). The injury induced an increase
in the number of recombined cells (Fig. 2, Ato G).
This reaction was restricted to the injured seg-
ment. After 9 days, the number of recombined
cells had increased more than 25-fold in the in-
jured segment. The number of pericyte-derived
cells peaked at 2 weeks and then, as the scar
condensed (9), decreased after 4 months to a lev-
el at which it remained for at least 7 months after
the injury (Fig. 2G). This can be compared with
the dynamics of the astrocyte population, which
undergoes an approximate doubling in the first
2 weeks in the same injury paradigm and then
decreases thereafter with similar kinetics (27).
There are about 10 times as many astrocytes as
type A pericytes in an uninjured spinal cord seg-
ment (480 T 22 and 49 T 3 cells in a 20-mm
coronal section, respectively) (27), but 2 weeks
after a lesion, there are about two times as many
pericyte-derived cells as newly generated
astrocytes in an injured spinal cord segment
(Fig. 2H). The scar is compartmentalized, with
pericyte-derived cells located in the center
Fig. 1. Genetic labeling of type A pericytes. (A to C) Electron micrographs showing a recombined type A
pericyte (arrow) on a blood vessel. (A) Pseudocolors indicate a recombined pericyte (green), nonre-
combined type B pericytes (blue), endothelial cells (red), and astrocytes (cyan). The inset shows the light
microscopic image of the section with the same recombined pericyte revealed by DAB reaction (arrow)
before cutting ultrathin sections. Higher magnifications of the recombined pericyte from(A) show (B) that
it is surrounded by basal lamina (bl, arrowheads) and (C) the plasma membrane (pm, arrowheads) of the
astrocyte. (D) Two recombined pericytes encapsulating two endothelial tubes (stained for von Willebrand
factor, vWF), surrounded by glial fibrillary acidic protein (GFAP)
+
astrocyte processes. Type A pericytes
express CD13 (E) and PDGFRb (F) but not desmin (G). Cell nuclei are visualized with 4′,6′-diamidino-2-
phenylindole (DAPI) in (D) to (G) and appear blue. Scale bars: 2 mmin (A), 0.5 mmin (C), and 10 mmin (G).
Fig. 2. Pericytes form the core of the scar in the injured spinal cord. (A)
Distribution of recombined type A pericytes (YFP) and astrocytes (GFAP) in an
uninjured thoracic spinal cord segment. (B to F) Type A pericyte progeny
occupy the core of the scar and are surrounded by astrocytes after a dorsal
funiculus incision. (G) Number of type A pericyte–derived cells at the lesion
site. (H) Net addition of type A pericyte–derived cells compared with astro-
cytes [data from (27)] 14 days and 4 months after injury. Cell nuclei are
visualized with DAPI in (A) to (F). (A) to (E) show coronal sections and (F) a
sagittal section. The quantifications show the average number of recombined
cells per 20-mm coronal section. Error bars represent SD. Scale bar, 200 mm.
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 239
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

surrounded first by a layer of astrocytes originat-
ing from ependymal cells and then by a layer of
astrocytes originating by self-duplication of resi-
dent astrocytes (fig. S7) (27).
To gain insight into the dynamics of the
pericyte injury response, we analyzed the early
events of pericyte recruitment. The lesion center
was nearly devoid of blood vessels on days
1 and 2 after the injury, but on days 3 to 5 after
the injury, blood vessel sprouts, with an increased
density of associated pericytes, appeared at the
lesion (Fig. 3A). All recombined pericytes were
tightly associated with endothelial cells outside
the injury site, but many recombined cells had
lost contact with blood vessels at the lesion (Fig.
3, B and C). Ultrastructural analysis showed an
increase in the number of type A pericytes and
a change in their morphology. Five days after
injury, they had detached from the basal lamina
encasement and developed thin processes, some
of which penetrated through the basal lamina to
invade the surrounding tissue (Fig. 3, D to F, and
fig. S8). Type A, but not type B, pericytes de-
posited abundant ECM within their basal lamina
encasement. Type A pericyte progeny that had
invaded the tissue were also surrounded by ECM
(Fig. 3, D, F, and I, and fig. S8 and S9). Five days
after injury, the number of pericytes associated
with blood vessels had significantly increased
in number (0.016 T 0.001 pericyte nuclear planes
per 10-mm capillary surface by EM in the un-
injured situation; n = 94 versus 0.151 T 0.045
after injury; n = 82 (mean T SEM); P < 0.001
Student’s t test). Type A pericytes increased
three times more in number as compared with
type B pericytes, and most important, only type
A pericytes showed signs of leaving the blood
vessel wall (Fig. 3, C to G, and fig. S6). The
recombined cells that no longer had contact with
Fig. 3. Pericytes give
rise to stromal cells and
deposit ECM in the in-
jured spinal cord. (A) Ac-
cumulation of type A
pericyte–derived cells
(YFP) and their detach-
ment from the vascular
wall (arrowheads) in the
lesion area, 5 days after
injury. (B) A blood ves-
sel crossing the border
(dashed line) between
intact and injured (upper
right) tissue, 5 days after
injury. Type A pericytes
(YFP) densely cover en-
dothelial cells (vWF) with-
in the intact tissue and
their progenydetachfrom
the blood vessel wall and
invade the surrounding
damaged tissue (arrow-
heads). (C) A blood ves-
sel within the lesion with
type B pericytes (visual-
ized with antibodies
against desmin) and type
A pericytes (YFP), 5 days
after injury, showing the
expansion and detach-
ment of the YFP-labeled
cells. (D) Pseudocolored
electron micrograph
showing a blood vessel
with three recombined
type A pericytes (green)
5 days after injury. Type
A pericytes detach from
thesurroundingbasal lam-
ina (bl), form thin pro-
cesses, and deposit ECM.
A type B pericyte (blue)
remains tightly attached
to the basal lamina (see
also fig. S8). Its ultrastructure is retained, similar to that in uninjured tissues
(Fig. 1A). An endothelial cell is colored red and astrocytes cyan. Boxed
area shows the fibrous ECM deposited around type A pericytes. (E) Three-
dimensional reconstruction of a series of electron micrographs showing a
leading process (lp) of a recombined type A pericyte breaking through its
basal lamina (bl) encasement (gray). (F) Electron micrograph of the lesion
area 14 days after injury. Several type A pericyte–derived cells (green)
have left the vascular wall and show abundant fibrous ECM (arrows) in
their immediate surrounding. Boxed area shows abundant fibrous ECM.
(G) Large numbers of type A pericyte–derived cells (YFP) distant to blood
vessels (platelet endothelial cell adhesion molecule, PECAM) 14 days after
injury. Dashed line outlines the ependymal layer. (H) Type A pericyte–derived
cells express smooth muscle actin (SMA) 5 days after injury. (I) The
distribution of recombined cells overlaps with that of fibronectin 14 days
after injury. Scale bars: 20 mm in (A) to (C) and (G) to (I), 2 mm in (D) and
(E), and 5 mm in (F).
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 240
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

blood vessels lost their expression of CD13 and
PDGFRa, but remained positive for PDGFRb
and became positive for the fibroblast marker
fibronectin and transiently expressed the myofi-
broblast marker smooth muscle actin (28) up to
9 days after the injury (Fig. 3, H and I). They
were never positive for astrocyte or oligodendro-
cyte lineage markers (figs. S10 and S11). The
vast majority of cells expressing markers of
Fig. 4. Pericyte-derived cells are essential for regaining tissue integrity. (A
and B) Many type A pericytes are Ki67
+
5 days after the injury and incorporate
BrdUduring the first 5 days after injury. (C) Schematic depiction of the strategy
to block the generation of progeny by type A pericytes. (D and E) The gen-
eration of type A pericyte progeny is abrogated in Glast-Rasless mice 5 days
after spinal cord injury. (F) Comparison of the scar core volume within the glial
borders that is occupied by PDGFRb
+
stromal cells in vehicle- and tamoxifen-
treated animals (Student’s two-tailed t test). (G) The percentage of the scar core
volume occupied by PDGFRb
+
stromal cells correlates with the recombination
efficacy in Glast-Rasless mice (Pearson’s correlation coefficient). (H) Correla-
tion of the tissue defect volume to the recombination efficacy in Glast-Rasless
mice. Individual animals are indicated with the same color in (G) and (H). (I
and J) The injury site (indicated by dashed line) of dissected spinal cords froma
vehicle-treated (I) and a tamoxifen-treated animal (J) 18 weeks after injury.
Arrows point to the tissue defect in (J). (Kto P) Sections of the spinal cords from
(I) and (J) showing a scar with PDGFRb
+
stromal cells and fibronectin in the
vehicle animal and the absence of a corresponding stroma in the tamoxifen
animal, which has an open tissue defect lined by GFAP
+
astrocytes. The animal
in (J) is represented by a green dot in (G) and (H). Scale bars: 20 mm in (B),
50 mm in (D), 0.5 mm in (I), and 0.1 mm in (O).
www.sciencemag.org SCIENCE VOL 333 8 JULY 2011 241
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

stromal cells at the lesion, such as fibronectin,
PDGFRb, and smooth muscle actin, were re-
combined, which established type A pericytes
as the main source of the scar connective tissue.
Thus, pericytes enter the lesion area with blood
vessel sprouts, and type A pericytes give rise
to cells that leave the blood vessel wall and
form the stromal component of the scar tissue
(fig. S12).
Analysis of bromodeoxyuridine (BrdU) in-
corporation and the mitotic marker Ki67 revealed
abundant proliferation of type A pericytes during
the first days after the spinal cord injury. By day 5
after the injury, 36.4 T 3.9% of YFP
+
cells were
Ki67
+
, and 95.5 T 1.9% had incorporated BrdU,
which indicated that the large increase in the
number of recombined cells is the result of pro-
liferation of type A pericytes (Fig. 4, A and B).
The absence of recombined cells in bone mar-
row or blood excluded a circulating source of re-
combined cells to the injured spinal cord (fig.
S13). Furthermore, the Glast-CreER line did not
recombine microglia or macrophages, and type A
pericyte–derived cells were distinct from these
cell types (fig. S14).
To assess the role of type A pericyte–derived
cells in the injured spinal cord, we devised a
genetic strategy to inhibit their generation. We
established mice that, in addition to carrying the
Glast-CreER and R26R-YFP alleles, were homo-
zygous for H-ras and N-ras null alleles and for
floxed K-ras alleles, in which type A pericytes
would lack all ras genes after induction of re-
combination (Fig. 4C, we refer to these mice as
Glast-Rasless). ras genes are necessary for cell
cycle progression and mitosis (29), and inducing
recombination before spinal cord injury drasti-
cally reduced the appearance of recombined cells
at the lesion site (Fig. 4, D and E). Deleting all
ras genes did not result in any apparent alteration
of the morphology or number of type Apericytes
outside the lesion nor did it alter the number or
distribution of blood vessels (fig. S15).
Adult Glast-Rasless mice in which recom-
bination had been induced by five daily injec-
tions of tamoxifen, followed by a 7-day clearing
period, were subjected to a dorsal spinal cord
hemisection and analyzed 18 weeks after the
injury. Mice of the identical genotype that re-
ceived vehicle without tamoxifen served as con-
trols and were indistinguishable from wild-type
mice with regard to spinal cord scar formation.
The scar of vehicle control animals was com-
posed of a core of PDGFRb-expressing cells en-
cased in fibronectin and surrounded by astrocytes,
similar to mice wild type for ras genes (Fig. 4).
The tamoxifen group had significantly less
PDGFRb-positive stromal cells in the scar core
compared with the vehicle group (P = 0.001,
Student’s t test) (Fig. 4F). Tamoxifen-induced
genetic recombination with CreER is seldom
complete, and we asked if the variation in the
generation of stromal cells within the tamoxi-
fen group was related to variation in recombina-
tion efficacy. The size of the stromal component
in individual animals did indeed negatively cor-
relate to the recombination efficacy (P = 0.0015,
r = –0.8857, Pearson’s correlation coefficient)
(Fig. 4G).
It became obvious when analyzing the injured
spinal cords that the generation of progeny by
pericytes is important for sealing the injury, as
33% of the tamoxifen-treated Glast-Rasless ani-
mals had failed to close the lesion and had an
open tissue defect at the site of the injury (com-
pared with none of the vehicle controls) (Fig. 4).
We found a correlation between the recombina-
tion efficacy and the failure to regain tissue in-
tegrity, with the animals showing the highest
recombination efficacy having open tissue defects
at the lesion site (Fig. 4H). The tamoxifen-treated
Glast-Rasless mice with the highest recombina-
tion efficacy were largely devoid of a stromal cell
and fibronectin scar core, which demonstrated that
type A pericyte–derived cells are required to seal
spinal cord lesions (Fig. 4, I to P).
We have identified pericytes as a source of
scar-forming cells in the adult spinal cord. Pre-
vious studies have demonstrated altered pericyte
morphology in response to traumatic brain injury
and suggested that they may leave the vessel wall
(30, 31), which in the light of our data indicates
that scar formation by pericytes may be a general
response to injuries in the CNS and potentially in
other organs. It is well known that pericytes are
heterogeneous on the basis of the expression
of markers and morphology (25, 32). Here we
demonstrate functional heterogeneity of pericyte
populations, with scar formation restricted to a
distinct subclass. Although the presence of stromal
cells in CNS scar tissue has been long recognized
(6–9), their role has been difficult to establish in
the absence of knowledge on their origin. We con-
clude that the generation of progeny by pericytes
is essential to regain tissue integrity after spinal
cord injury.
References and Notes
1. J. W. Fawcett, J. Neurotrauma 23, 371 (2006).
2. J. Silver, J. H. Miller, Nat. Rev. Neurosci. 5, 146 (2004).
3. A. Rolls, R. Shechter, M. Schwartz, Nat. Rev. Neurosci.
10, 235 (2009).
4. S. Okada et al., Nat. Med. 12, 829 (2006).
5. M. V. Sofroniew, Trends Neurosci. 32, 638 (2009).
6. M. A. Matthews, M. F. St Onge, C. L. Faciane,
J. B. Gelderd, Neuropathol. Appl. Neurobiol. 5, 161
(1979).
7. W. F. Windle, W. W. Chambers, J. Comp. Neurol. 93, 241
(1950).
8. J. G. Krikorian, L. Guth, E. J. Donati, Exp. Neurol. 72, 698
(1981).
9. E. Camand, M. P. Morel, A. Faissner, C. Sotelo, I. Dusart,
Eur. J. Neurosci. 20, 1161 (2004).
10. M. C. Shearer, J. W. Fawcett, Cell Tissue Res. 305, 267
(2001).
11. G. C. Gurtner, S. Werner, Y. Barrandon, M. T. Longaker,
Nature 453, 314 (2008).
12. G. Krenning, E. M. Zeisberg, R. Kalluri, J. Cell. Physiol.
225, 631 (2010).
13. I. A. Darby, T. D. Hewitson, Int. Rev. Cytol. 257, 143
(2007).
14. M. Zeisberg, R. Kalluri, J. Mol. Med. 82, 175 (2004).
15. B. D. Humphreys et al., Am. J. Pathol. 176, 85
(2010).
16. S. L. Lin, T. Kisseleva, D. A. Brenner, J. S. Duffield,
Am. J. Pathol. 173, 1617 (2008).
17. C. Sundberg, M. Ivarsson, B. Gerdin, K. Rubin,
Lab. Invest. 74, 452 (1996).
18. M. Slezak et al., Glia 55, 1565 (2007).
19. S. Srinivas et al., BMC Dev. Biol. 1, 4 (2001).
20. Materials and methods are available as supporting
material on Science Online.
21. M. Krueger, I. Bechmann, Glia 58, 1 (2010).
22. A. Armulik et al., Nature 468, 557 (2010).
23. R. Daneman, L. Zhou, A. A. Kebede, B. A. Barres, Nature
468, 562 (2010).
24. R. D. Bell et al., Neuron 68, 409 (2010).
25. C. Bondjers et al., FASEB J. 20, 1703 (2006).
26. K. Meletis et al., PLoS Biol. 6, e182 (2008).
27. F. Barnabé-Heider et al., Cell Stem Cell 7, 470 (2010).
28. R. J. McAnulty, Int. J. Biochem. Cell Biol. 39, 666
(2007).
29. M. Drosten et al., EMBO J. 29, 1091 (2010).
30. P. Dore-Duffy et al., Microvasc. Res. 60, 55 (2000).
31. P. Dore-Duffy, Curr. Pharm. Des. 14, 1581 (2008).
Acknowledgments: We thank A. Simon and the Frisén and
Shupliakov groups for valuable discussions. This study
was supported by the Swedish Research Council, the
Swedish Cancer Society, Tobias Stiftelsen, Hjärnfonden,
Knut och Alice Wallenbergs Stiftelse, the Swedish Agency
for Innovation Systems, and the European Research
Council (ERC-AG/250297-RAS AHEAD). D.D. was
supported by the Foundation for Science and Technology
from the Portuguese government (SFRH/BD/63164/2009).
Supporting Online Material
www.sciencemag.org/cgi/content/full/333/6039/238/DC1
Materials and Methods
Figs. S1 to S15
References
21 January 2011; accepted 13 May 2011
10.1126/science.1203165
8 JULY 2011 VOL 333 SCIENCE www.sciencemag.org 242
REPORTS

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

243
Newly offered instrumentation, apparatus, and laboratory materials of interest to researchers in all disciplines in academic, industrial, and governmental organizations are
featured in this space. Emphasis is given to purpose, chief characteristics, and availability of products and materials. Endorsement by Science or AAAS of any products or
materials mentioned is not implied. Additional information may be obtained from the manufacturer or supplier.
Electronically submit your new product description or product literature information! Go to www.sciencemag.org/products/newproducts.dtl for more information.
LIFE SCIENCE TECHNOLOGIES
NEW PRODUCTS
AAAS/Science Business Office Feature
DNA REpAiR kNockDowN cEll liNEs
A range of highly effcient knockdown cell lines covering the base
excision DNA repair pathway enable scientists to study the molecu-
lar etiology of tumor genomic instability and to exploit it for oncol-
ogy research. The new range includes 20 DNA repair knockdown
cell lines, which are based upon target-specifc lN428 glioblastoma
shRNA lentivirus-transduced cells that are rigorously qualifed and
mycoplasma free. DNA repair pathways maintain the integrity of the
genome, reducing the onset of cancer, disease, and aging pheno-
types. The DNA repair knockdown cell lines provide essential tools
to address these needs. All DNA repair knockdown cell line products
are evaluated by reverse transcription polymerase chain reactions
(RT-pcR) and western blot analysis, and functional assays are per-
formed when feasible. The knockdown effciency of the new DNA
repair knockdown cell lines on target genes, as measured by RT-
pcR, is 63 to 98%.
AMS Biotechnology
For info: +44-(0)-1235-828200 www.amsbio.com
HiGH THRouGHpuT DNA lAbEliNG
New high throughput genomic DNA labeling kits and sample track-
ing spike-ins can further streamline workfow and minimize sample-
tracking errors for researchers performing array comparative genom-
ic hybridization (acGH). The new cytosure HT Genomic DNA label-
ing kit makes it possible to simultaneously label up to 96 samples
using a 96-well plate format, boosting cost-effectiveness, increasing
processing speed, and facilitating automated sample handling. The
unique formulation and mastermix-based approach maximizes sig-
nal-to-noise ratio and reduces sample-to-sample variation, facilitating
the confdent detection of even the smallest aberrations. cytosure
iscA sample Tracking spike-ins ensure that any potential sample
mix-up during array processing can be easily identifed, safeguard-
ing against mistakes and maximizing data confdence. when used in
combination with cytosure iscA arrays and the cytosure interpret
software, these new products make it easy to correlate individual
data sets with sample origin.
oxford Gene Technology
For info: +44-(0)-1865-856826 www.ogt.co.uk
DNA/RNA TRANsfoRmATioN
The newly launched Eppendorf Eporator offers a fast, simple, and safe way to trans-
form bacteria, yeast, and other microorganisms with DNA/RNA. Results are highly
reproducible and, compared with chemical methods, electroporation yields to signif-
cantly higher transformation effciency. Designed to deliver ideal conditions for elec-
troporation of bacteria and yeast, the Eppendorf Eporator has been shown to give
transformation effciencies 10 times higher than with chemical transformation (heat
shock method). The new Eppendorf Eporator not only saves valuable time and deliv-
ers higher transformation rates, but, crucially, the instrument is also extremely user
friendly. Two new program buttons allow for storage and recall of the most commonly
used parameters, and a simple one-button operation ensures intuitive use for faster
sample handling. Eppendorf Eporator has a compact, space-saving design for easy
storage and transport and comes with a usb port facilitating export of data for analysis
and Glp-compliant documentation.
Eppendorf
For info: 800-645-3050 www.eppendorf.com
HiGH THRouGHpuT sAmplE pREpARATioN
The samplicity filtration system is an innovative new technology
that provides a convenient, high throughput alternative to syringe-tip
flters when preparing samples for chromatography. The samplicity
system allows up to eight samples—even those with high viscosity
or particulates—to be simultaneously vacuum-fltered in seconds.
samples are quickly and easily loaded using a pipettor and are fl-
tered directly into lc vials. The fltered samples are immediately
ready for subsequent analyses. The samplicity system provides re-
lief from the repetition of manual fltration and offers a throughput
capacity well-aligned with the needs of most labs. The samplicity
system is designed for use with millex samplicity flter units with
a hydrophilic Tefon membrane flter in either a 0.45 or 0.2 μm pore
size. millex samplicity flter units have low extractables, low analyte
binding properties, and a low hold-up volume, which allows process-
ing of samples as small as 200 μl.
EmD millipore
For info: 800-645-5476 www.millipore.com
plATE wAsHER
The Hydrospeed plate washer is an advanced system optimized
for washing of cells, beads, and enzyme-linked immunosorbent as-
says (ElisA) in 96- and 384-well formats. it offers full control over
critical wash parameters via an intuitive touchscreen interface, with
extra gentle drop-wise dispensing, and tunable aspiration settings
to help avoid loss of material and maximize assay effciency. The Hy-
drospeed features advanced cell protection settings for washing of
adherent and loosely adherent cells, allowing the user to dial-in extra
gentle wash parameters to suit their cells, microplates, and applica-
tion. The system’s innovative Anti clogging function takes the hassle
out of ElisA washing by automatically rinsing and soaking the wash
head when the system is idle between runs. The Hydrospeed uses
two magnets per well for high performance magnetic bead wash-
ing, offering fast bead settling and excellent recovery rates, and can
also be equipped with a vacuum fltration module for processing of
nonmagnetic beads.
Tecan
For info: +41-(0)-44-922-81-11 www.tecan.com
www.sciencemag.org/products SCIENCE VOL 333 8 JULY 2011

o
n

J
u
l
y

7
,

2
0
1
1
w
w
w
.
s
c
i
e
n
c
e
m
a
g
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

S
u
b
m
i
s
s
i
o
n
d
e
a
d
l
i
n
e
A
u
g
u
s
t
1
imagination at work
* For the purpose of this prize, molecular biology is
defined as “that part of biology which attempts to
interpret biological events in terms of the physico-
chemical properties of molecules in a cell”.
(McGraw-Hill Dictionary of Scientific and
Technical Terms, 4th Edition).
GE Healthcare Bio-Sciences AB,
Björkgatan 30, 751 84 Uppsala, Sweden.
©2011 General Electric Company
– All rights reserved.
28-9402-06AB
Imagine standing on the podium at the Grand Hotel in Stockholm, making your
acceptance speech for the GE & Science Prize for Young Life Scientists. Imagine
having your essay read by your peers around the world. Imagine discussing your
work in a seminar with other prize winners and Nobel Laureates. Imagine what you
could do with the $25,000 prize money. Nowstop imagining. If you were awarded your
Ph.D. in molecular biology in 2010, then submit your 1000-word essay by August 1,
and you can make it a reality.
Want to build a better reality? Go to www.gescienceprize.org
The GE & Science Prize for Young Life Scientists.
Because brilliant ideas build better realities.
Your
name
here.
©2011 Life Technologies Corporation. All rights reserved. The trademarks mentioned herein are the property of Life Technologies Corporation
or their respective owners. For research use only. Not intended for any animal or human therapeutic or diagnostic use. CO19167 0511
headline essence
headline brow
Go to www.invitrogen.com/nupagesci
Get the free mobile app for your phone at http://gettag.mobi
You can depend on NuPAGE
®
gels to deliver straight, sharp bands
Life Technologies offers you a reason to smile with Novex
®
NuPAGE
®
precast gels for protein
separation. Accurate and easy separation is at your fingertips. Enjoy the convenience of
consistent, superior performance.
here's a good reason
to smile
To find your local sales office, visit www.bio-rad.com/contact/
In the U.S., call toll free at 1-800-4BIORAD (1-800-424-6723)
Visit us at www.bio-rad.com
That’s qqPPCCCRReeevvooluutiiooonaarry.
The Bio-Rad CFX96 Touch

and CFX384 Touch

real-time
PCR systems let you start your run with the easy-to-use
touch screen, view your data on the system in real-time,
and have your results emailed to you at the end of the run.
The CFX96 Touch and CFX384 Touch systems’ revolutionary
features give you a better way to increase your qPCRproductivity.

Easier — factory calibration, quick setup, and intuitive
software let you get better results right away

Faster — save time and costs optimizing assays
in a single run using the thermal gradient

Smarter — innovative technology with long-life
LEDs provides optimal results and maximal reliability
without time-consuming, costly maintenance
Better qPCR is at your fingertips.
Visit www.bio-rad.com/ad/qPCRevolutionary
or contact your Bio-Rad sales representative to learn more.
RRe esse eaarccch h. TToog geeethheeer.
qPCR that stands alone.
AMPLI FI CATI ON // CFX96 TOUCH AND CFX384 TOUCH
Perform an Apples-to-Apples
Comparison to any
384-Well PCR System
LightCycler
®
480 Real-Time PCR System
Does Life Technologies’ launch of the new ViiA 7 Real-Time PCR System
have you seeking a replacement for your 384-well system? Before rushing into
any purchase, check out the time-tested, peer-published LightCycler® 480
Real-Time PCR System from Roche Applied Science — the 384-well solution
of choice for hundreds of researchers during the past five years.
í Generate consistent, reliable data—without Rox dye. Why rely on Rox dye
to normalize data between wells? Achieve reproducible results well-to-well
and between runs through the LightCycler® 480 System’s greater temperature
homogeneity (Figure 1).
í Readily integrate the LightCycler
®
480 System into your lab’s robotics.
Why limit your automation choices to only the one robot supplied by your
qPCR instrument manufacturer?
í Rely on >900 publications and your colleagues’ experience with our
service. Ask any LightCycler® 480 System owner about the ongoing
customer support you can expect.
Compare the LightCycler
®
480 System to any 384-well system — old or new.
Call 800 262 4911 or visit www.roche-applied-science.com/usa/compare
to learn more.
For life science research only.
Not for use in diagnostic procedures.
LIGHTCYCLER is a trademark of Roche.
Other brands or product names are trademarks of their respective holders.
© 2011 Roche Diagnostics. All rights reserved.
Figure 1: Influence of the Therma-Base layer on
temperature homogeneity across a 384-well
plate. (a): LightCycler
®
480 Real-Time PCR System
including Therma-Base; (b): 384-well plate on
Applied Biosystems 7900HT without Therma-Base.
Data generated at Roche Applied Science.
T
m
T
m
1
0.5
0
-0.5
-1
1
5
9
13
17
21
P
K
F
A
1
0.5
0
-0.5
-1
1
5
9
13
17
21
P
K
F
A
Columns Rows
Columns Rows a
b
Roche Diagnostics Corporation
Roche Applied Science
Indianapolis, Indiana
Carl R. Alving, M.D.
Chief of the Department of Adjuvant &
Antigen Research, Division of
Retrovirology at the Walter Reed Army
Institute of Research
AAAS member
“A dream
toldmeto
do it.”
MemberCentral is the new website
that looksat sciencethroughtheeyes
of AAAS members. It celebrates their
achievements—likeDr. Alving’svaccine
patch—and their shared belief in the
transformative power of science. Use
MemberCentral to connect with other
members, learn about work being
done in other fields, and get fresh
perspectives on issues ranging from
speciationtoSTEMeducation.
Visit MemberCentral today and get to
know the AAAS member community
inawholenewway.
Blogs I Videos I Webinars
Discounts I Downloads | Community
MemberCentral.aaas.org
Dr. Carl Alving
on his inspiration
for inventing
the vaccine patch.
“A dream
toldmeto
do it.”
Michelin strives for innovation in sustainable mobility
and initiates ambitious research partnerships
with world class academic institutions
The ever-growing worldwide demand for efficient road transportation translates
into the need to reduce urban pollution, green house gas emissions, traffic
congestion, energy consumption, and hazards. Michelin, a world leader in
tyre manufacturing, works to invent new materials and processes that help
the automotive industry tackle these enormous challenges. Convinced that
innovation will not come only from the optimization of existing materials but
from technological and conceptual leaps, Michelin has initiated an ambitious
research programthat promotes close and in-depth partnerships with world class
academic institutions.
“The Materials Science Chair at ESPCI ParisTech” was launched in 2008 as part
of a partnership between Michelin and ESPCI ParisTech (École supérieure de
physique et de chimie industrielles de la Ville de Paris). Located in the heart of
Paris, within the famous Quartier Latin, ESPCI ParisTech is a leading French
Grande École training scientists and engineers at the graduate level, as well as
a world renowned research institution and an engine of innovation for industry.
It counts a number of prominent scientists and Nobel laureates, including
Pierre-Gilles de Gennes and Georges Charpak.The Chair fosters collaborations
and exchanges between scientists at ESPCI ParisTech and research engineers
at Michelin, and supports the research-driven education provided by ESPCI
ParisTech.
To mark two years of successful collaboration and to strengthen its field of
expertise, the Chair recently organized a two-day international workshop at
ESPCI ParisTech. The event brought together chemists, materials scientists,
physicists, and engineers committed to the understanding, processing, and
applications of polymer nanocomposites, rubbers, filled elastomers, and multi-
component polymer nanostructures. The workshop featured best-in-their-field
speakers who reviewed the latest achievements in the dynamics of polymeric
materials at different scales. The lectures, which combined simulation, theory
and advanced experimentation, emphasized the need for multi-scale approaches
connecting the microscopic properties (macromolecular architecture, local
deformation, glass dynamics) to the macroscopic behaviour (aging, mechanics
at large deformation, fracture, rheology) of polymeric materials.
Information on “The Materials Science Chair at ESPCI ParisTech” and on the
“Multi-scale Dynamics of Structured Polymeric Materials” workshop is available
at: www.chairemichelin.espci.fr/home/
Contact with the Scientific Direction of Michelin is through:
pierrick.travert@fr.michelin.com
Metal and textile reinforcements
Tyres: multi-component products …..
More than 30 different
rubbers
Mesoscale
Dynamics
Molecular
Dynamics
…with multi-scale
physics
About 200 chemical
compounds
A Journal with Impact fromAAAS, the publisher of Science
Science Translational Medicine
Integrating Medicine and Science
“The 2010 selection for the Nobel Prize in Physiology or Medicine
as well as the three Lasker Awards brought welcome opportunities
to celebrate truly groundbreaking translational research.”*
This quote illuminates the importance
of translational medicine discoveries.
A recent journal article features the
sequencing of fetal DNA from plasma
of a pregnant woman to permit prenatal,
noninvasive genome-wide screening
to diagnose fetal genetic disorders.
* Sci Transl Med 22 December 2010:
Vol. 2, Issue 63, p. 63ed9
DOI: 10.1126/scitranslmed.3001816
Recommend
an institutional
subscription
to your library
today!
ScienceOnline.org/recommend
ScienceTranslationalMedicine.org
Indexed in
MEDLINE/PubMed
243
Newly offered instrumentation, apparatus, and laboratory materials of interest to researchers in all disciplines in academic, industrial, and governmental organizations are
featured in this space. Emphasis is given to purpose, chief characteristics, and availability of products and materials. Endorsement by Science or AAAS of any products or
materials mentioned is not implied. Additional information may be obtained from the manufacturer or supplier.
Electronically submit your new product description or product literature information! Go to www.sciencemag.org/products/newproducts.dtl for more information.
LIFE SCIENCE TECHNOLOGIES
NEW PRODUCTS
AAAS/Science Business Office Feature
DNA REPAIR KNOCKDOWN CELL LINES
A range of highly efficient knockdown cell lines covering the base
excision DNA repair pathway enable scientists to study the molecu-
lar etiology of tumor genomic instability and to exploit it for oncol-
ogy research. The new range includes 20 DNA repair knockdown
cell lines, which are based upon target-specific LN428 glioblastoma
shRNA lentivirus-transduced cells that are rigorously qualified and
mycoplasma free. DNA repair pathways maintain the integrity of the
genome, reducing the onset of cancer, disease, and aging pheno-
types. The DNA repair knockdown cell lines provide essential tools
to address these needs. All DNA repair knockdown cell line products
are evaluated by reverse transcription polymerase chain reactions
(RT-PCR) and Western blot analysis, and functional assays are per-
formed when feasible. The knockdown efficiency of the new DNA
repair knockdown cell lines on target genes, as measured by RT-
PCR, is 63 to 98%.
AMS Biotechnology
For info: +44-(0)-1235-828200 www.amsbio.com
HIGH THROUGHPUT DNA LABELING
New high throughput genomic DNA labeling kits and sample track-
ing spike-ins can further streamline workflow and minimize sample-
tracking errors for researchers performing array comparative genom-
ic hybridization (aCGH). The new CytoSure HT Genomic DNA Label-
ing Kit makes it possible to simultaneously label up to 96 samples
using a 96-well plate format, boosting cost-effectiveness, increasing
processing speed, and facilitating automated sample handling. The
unique formulation and mastermix-based approach maximizes sig-
nal-to-noise ratio and reduces sample-to-sample variation, facilitating
the confident detection of even the smallest aberrations. CytoSure
ISCA Sample Tracking Spike-ins ensure that any potential sample
mix-up during array processing can be easily identified, safeguard-
ing against mistakes and maximizing data confidence. When used in
combination with CytoSure ISCA arrays and the CytoSure Interpret
Software, these new products make it easy to correlate individual
data sets with sample origin.
Oxford Gene Technology
For info: +44-(0)-1865-856826 www.ogt.co.uk
DNA/RNA TRANSFORMATION
The newly launched Eppendorf Eporator offers a fast, simple, and safe way to trans-
form bacteria, yeast, and other microorganisms with DNA/RNA. Results are highly
reproducible and, compared with chemical methods, electroporation yields to signifi-
cantly higher transformation efficiency. Designed to deliver ideal conditions for elec-
troporation of bacteria and yeast, the Eppendorf Eporator has been shown to give
transformation efficiencies 10 times higher than with chemical transformation (heat
shock method). The new Eppendorf Eporator not only saves valuable time and deliv-
ers higher transformation rates, but, crucially, the instrument is also extremely user
friendly. Two new program buttons allow for storage and recall of the most commonly
used parameters, and a simple one-button operation ensures intuitive use for faster
sample handling. Eppendorf Eporator has a compact, space-saving design for easy
storage and transport and comes with a USB port facilitating export of data for analysis
and GLP-compliant documentation.
Eppendorf
For info: 800-645-3050 www.eppendorf.com
HIGH THROUGHPUT SAMPLE PREPARATION
The Samplicity Filtration System is an innovative new technology
that provides a convenient, high throughput alternative to syringe-tip
filters when preparing samples for chromatography. The Samplicity
system allows up to eight samples—even those with high viscosity
or particulates—to be simultaneously vacuum-filtered in seconds.
Samples are quickly and easily loaded using a pipettor and are fil-
tered directly into LC vials. The filtered samples are immediately
ready for subsequent analyses. The Samplicity system provides re-
lief from the repetition of manual filtration and offers a throughput
capacity well-aligned with the needs of most labs. The Samplicity
system is designed for use with Millex Samplicity filter units with
a hydrophilic Teflon membrane filter in either a 0.45 or 0.2 µm pore
size. Millex Samplicity filter units have low extractables, low analyte
binding properties, and a low hold-up volume, which allows process-
ing of samples as small as 200 µL.
EMD Millipore
For info: 800-645-5476 www.millipore.com
PLATE WASHER
The HydroSpeed plate washer is an advanced system optimized
for washing of cells, beads, and enzyme-linked immunosorbent as-
says (ELISA) in 96- and 384-well formats. It offers full control over
critical wash parameters via an intuitive touchscreen interface, with
extra gentle drop-wise dispensing, and tunable aspiration settings
to help avoid loss of material and maximize assay efficiency. The Hy-
droSpeed features advanced Cell Protection settings for washing of
adherent and loosely adherent cells, allowing the user to dial-in extra
gentle wash parameters to suit their cells, microplates, and applica-
tion. The system’s innovative Anti Clogging function takes the hassle
out of ELISA washing by automatically rinsing and soaking the wash
head when the system is idle between runs. The HydroSpeed uses
two magnets per well for high performance magnetic bead wash-
ing, offering fast bead settling and excellent recovery rates, and can
also be equipped with a vacuum filtration module for processing of
nonmagnetic beads.
Tecan
For info: +41-(0)-44-922-81-11 www.tecan.com
www.sciencemag.org/products SCIENCE VOL 333 8 JULY 2011
Jumpstart your
biomarker research
with a Biomarker
Discovery Pilot Grant
+1-888-494-8555 +1-770-729-2992 http://www.RayBiotech.com/Grant_sci grant@raybiotech.com
Discover more. Publish faster.
RayBiotech oers the widest selection of protein proling
tools in today's market. In a single experiment, our protein
arrays can detect over 500 proteins or measure the concen-
tration of 280 proteins, including cytokines, growth factors,
soluble receptors, and other molecules related to:
• Alzheimer’s disease • Angiogenesis
• Obesity & diabetes • Inflammation
• Cardiovascular disease • Apoptosis
• Cancer • and many more
More data with less sample.
• High-content screening
• Adaptable to high-throughput automated testing
• Compatible with most sample types
• Quantitative and screening arrays available
• Biostatistic and bioinformatic services available
• Proven technology, hundreds of publications
APPLYTODAY to win
up to $20,000 worth
of Raybiotech products
and services.
Complete your application for Raybiotech’s Biomarker
Discovery Grant ProgramTODAY and get a 10% discount on
all Raybiotech products and services (good through the end
of 2011). DEADLINE FOR APPLICATIONS IS AUGUST 1, 2011.
For more details, visit our website:,
www.Raybiotech.com/Grant_sci
or email us grant@raybiotech.com.
Withastaff of 4700, ForschungszentrumJülich–amember of theHelmholtz Association–
is one of the largest interdisciplinary research centres in Europe. Work with us on
the grand challenges in the elds of health, energy & environment, and information
technology, as well as on the many and varied tasks of research management.
The Nuclear Waste Management and Reactor Safety part of the Institute for Energy and
Climate Research contributes fundamental and applied research for the safe management
of nuclear waste and the safe operation of nuclear reactors. Research topics are to a large
extent material science oriented and include the long-term safety of nuclear waste dispo-
sal, innovative waste management strategies (partitioning, conditioning, transmutation),
structure research (actinide solid state chemistry), non-destructive characterisation of
nuclear waste and nuclear safeguards.
This team currently has a vacancy for a
GROUP LEADER (m/f)
in Computational Science
Five year funding to establish a research group, tenure-option
Job description:
ForschungszentrumJülich plans to establish a Young Investigators Group focused on furthering
our understanding of materials relevant to the safe management of high level nuclear waste
on the basis of simulations on the atomistic, nano- and mesoscale. Typical research projects
would include the modeling of the incorporation of actinides into crystalline solids, or the
interaction of aqueous actinide species with solid surfaces.
Forschungszentrum Jülich operates one of the most powerful computer systems for scientic
and technical applications in Europe, including the petaop computer JUGENE and the
supercomputer JUROPA, and provides the adequate scientic infrastructure for the project.
The position offers you ve years of funding to establish your own research group. The grant
will include the group leader position, PhDand/or postdoctoral fellows. It is envisaged that the
group will consist of four teammembers with complementary computational skills. The tenure
option (permanent position) will depend on the positive outcome of a midterm evaluation
which will be conducted about four years after start. A joint appointment as junior professor
(W1) will be made wherever applicable with our partner RWTH Aachen University.
Qualications and skills required:
In order to contribute to the research programme of the Institute for Energy and Climate
Research/Nuclear Waste Management and Reactor Safety, it is expected that a wide
variety of computational approaches, ranging from highly accurate parameter-free
atomistic modeling approaches to semi-empirical and/or empirical methods capable of
describing interactions on the nano- or mesoscale will be adopted and developed within
the Young Investigators Group.
The successful candidate is expected to conduct visionary independent research and
will have extensive experience in code development, for example as a co-developer of
a DFT package or in the context of developing a QM/MM approach. He/she will closely
cooperate with the experimental research programme of the institute and is expected
to acquire third party funding. As this group will initially be established as a Young
Investigators Group, the candidate must have obtained a PhD within the last 6 years and
should have spent at least 6 months abroad.
Salary and social benets will conform to the provisions of the German civil service.
The implementation of equal opportunities is a cornerstone of our staff policy at Forschungs-
zentrumJülich, for which we have received the „TOTAL E-QUALITY“ Award. Applications from
women are therefore particularly welcome. We also welcome applications from disabled
persons.
Please send your application following the specic requirements (as indicated in the job
advertisement on our website www.fz-juelich.de/careers), quoting the reference code
138/2011SC, until August 19th, 2011 to:
Forschungszentrum Jülich GmbH
Geschäftsbereich Personal
- Personalentwicklung -
52425 Jülich
Germany
contact:
Barbara Küppers
phone: +49 2461 61-5358
www.fz-juelich.de
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g
Tracy Holmes
Worldwide Associate Director
Science Careers
Phone: +44 (0) 1223 326525
UNITED STATES & CANADA
E-mail: advertise@sciencecareers.org
Fax: 202-289-6742
Tina Burks
Midwest/West Coast/
South Central/Canada
Phone: 202-326-6577
Elizabeth Early
East Coast & Industry
Phone: 202-326-6578
Marci Gallun
Sales Administrator
Phone: 202-326-6582
Online Job Posting Questions
Phone: 202-326-6577
EUROPE & REST OF WORLD
E-mail: ads@science-int.co.uk
Fax: +44 (0) 1223 326532
Alex Palmer
Phone: +44 (0) 1223 326527
Susanne Kharraz
Phone: +44 (0) 1223 326529
Dan Pennington
Phone: +44 (0) 1223 326517
Lisa Patterson
Phone: +44 (0) 1223 326528
JAPAN
ASCA Corporation
Phone: +81-3-6802-4616
Fax: +81-3-6802-4615
E-mail: careerads@sciencemag.jp
CHINA &TAIWAN
Ruolei Wu
Phone: +86-1367-1015-294
E-mail: rwu@aaas.org
For full advertising details, go to
ScienceCareers.org and click For Employers,
or call one of our representatives.
All ads submitted for publication must comply
with applicable U.S. and non-U.S. laws. Science
reserves the right to refuse any advertisement
at its sole discretion for any reason, including
without limitation for offensive language or
inappropriate content, and all advertising is
subject to publisher approval. Science encour-
ages our readers to alert us to any ads that
they feel may be discriminatory or offensive.
S
c
ie
n
c
e
C
a
r
e
e
r
s
A
d
v
e
r
t
is
in
g
EcoHealth Alliance Announces 3 New Positions
in Emerging Disease Ecology
EcoHealth Alliance is expanding its research programs in emerging disease
ecology and seeks outstanding candidates for three positions. The positions
will play core roles in our Hotspots of Emerging Infectious Diseases proj-
ect. Full position descriptions are available at www.ecohealthalliance.org/
about/careers.
Macroecologist/GIS Researcher with a disease interest and experience
managing large, spatial databases. Project includes examining the roles of
biodiversity, land-use change, climate change and other socio-economic,
demographic, and environmental drivers on disease emergence. PhDin Ecol-
ogy, Biology, Mathematics, or related field and experience with GIS and
database software is required. Prior postdoctoral experience preferred.
Public Health Research Scientist/Epidemiologist with an interest in human
medicine, to investigate the origins and recent trends in emerging infectious
diseases, working to increase the rigor of the underlying datasets on disease
emergence and conducting analyses on past, present, and future trends for
disease emergence. PhD in Epidemiology or Disease Ecology and proven
experience with infectious disease modeling is required.
One Analyst/Technician Position: GIS Analyst/Technician to work with
infectious disease data sets, as well as environmental and socioeconomic data
sets, to support model development for the Hotspots of Emerging Infectious
Disease project. Agraduate degree in quantitative or computational sciences
is preferred, however we will consider candidates with undergraduate level
degrees in geography, epidemiology, or ecology as well.
Further details for all positions may be found at www.ecohealthalliance.org/
about/careers. All positions are based in NewYork and require some interna-
tional travel. Review of applications will begin August 1 and continue until
positions are filled. Candidates should submit a CV, 2-page cover letter stating
clearly the position of interest and career goals, and e-mail addresses for two
references to jobs@ecohealthalliance.org. Please include the position title
in the subject of the e-mail.
BE PART OF THE FOUNDING FACULTY OF SCIENTISTS AND ENGINEERS.
Successful candidates can look forward to internationally competitive remuneration, and assistance for relocation to Singapore.
If you want to be part of the founding faculty that will shape the world of tomorrow, please apply to SUTD at www.sutd.edu.sg
The Singapore University of Technology and Design (SUTD), established in collaboration with the Massachusetts Institute of Technology (MIT), is seeking
exceptional faculty members for this new university slated to matriculate its first intake of students in April 2012.
SUTD, the first university in the world with a focus on design accomplished through an integrated multi-disciplinary curriculum, has a mission to advance knowledge
and nurture technically grounded leaders and innovators to serve societal needs. SUTD is characterised by a breadth of intellectual perspectives (the“university”),
a focus on engineering foundations (“technology”) and an emphasis on innovation and creativity (“design”). The Universityʼs programmes are based on four pillars
leading to separate degree programmes in Architecture and Sustainable Design, Engineering Product Development, Engineering Systems and Design, and
Information Systems Technology and Design. Design, as an academic discipline, cuts across the curriculum and will be the framework for novel research and
educational programmes.
MITʼs multi-faceted collaboration with SUTD includes the development of new courses and curricula, assistance with the early deployment of courses in Singapore,
assistance with faculty and student recruiting, mentoring, and career development, and collaborating on a major joint research projects, through a major new
international design centre and student exchanges. Many of the newly hired SUTD faculty will spend up to year at MIT in a specially tailored programme for
collaboration and professional development.
FACULTY POSITIONS
The qualifications for the faculty position include: an earned doctorate in Biology, Biomedical Sciences, Chemistry, Chemical Engineering, a strong commitment to
teaching at the undergraduate and graduate levels, a demonstrated record of, or potential for scholarly research, and excellent communication skills.
We invite applications for interdisciplinary faculty appointments at all levels, with many opportunities available in particular at the Assistant and Associate Professor
levels. Duties include teaching of graduate and undergraduate students, research, supervision of student research, advising undergraduate student projects, and
service to SUTD and the community. Faculty will be expected to develop and sustain a strong research programme. Attractive research grant opportunities are
also available.
TENURE TRACK FACULTY POSITIONS
IN CANCER GENETICS
The Department of Genetics and Comprehensive Cancer Center at the Uni-
versity of Alabama at Birmingham (UAB) invites applications from highly
qualified individuals for tenure-track faculty positions at the level of Assistant
or Associate Professor. Applications are sought from candidates with exper-
tise in the broad area of cancer genetics. However candidates with a strong
background in mitochondrial genetics, genetics, epigenetics of cancer cell
metabolism, genomics, epigenomics, inter-genomics and functional genomics
based approaches to understanding the biology of cancer will be preferred.
Researchers studying the molecular genetics of cancer susceptibility involving
energy balance and obesity are also encouraged to apply. Candidates must be
committed to developing a robust, extramurally funded research program in
cancer genetics. Candidates at the senior level should have outstanding and
independent research programsupported by peer-reviewed grants. The success-
ful candidate will also contribute to departmental responsibilities associated
with graduate and medical student training.
UAB is a major research university and academic health center that offers a
highly interactive scientific environment with state-of-the-art research facilities
and multidisciplinary work environment. The Department of Genetics fosters
collaboration with other basic science as well as clinical departments and
the UAB Comprehensive Cancer Center. The Department provides excellent
laboratory facilities, a highly competitive salary and start-up funds, and access
to numerous core facilities.
Applicants should send a curriculum vitae, a description of research accom-
plishments and future research objectives (pdf format), and have three
references addressed to Keshav K. Singh, Ph.D., Joy and Bill Harbert
Endowed Chair in Cancer Genetics, Professor of Genetics, Pathology and
Environmental Health, Department of Genetics, UABSchool of Medicine,
Birmingham, Alabama. Please e-mail documents to mito@uab.edu.
The University of Alabama at Birmingham is an Equal Opportunity,
Affirmative Action Employer with a strong commitment to ethnic and
cultural diversity among its faculty, students and staff. Applications from
women and ethnic minorities are encouraged.
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g
The Novo Nordisk Foundation Center for Basic Metabolic Research
Group Leaders and Postdocs
We invite letters of interest for a
number of positions as group leaders
and postdocs at the Novo Nordisk
Foundation Center for Basic Metabolic
Research at the University of
Copenhagen (www.metabol.ku.dk).
We wish to recruit highly committed
group leaders and post docs to
establish new research groups
at the Center, carrying out basic
metabolic research of the highest
scientific impact and standard on
an independent basis but in close
collaboration with the other groups
at the Center. It is anticipated that
the group leaders should be able to
attract significant research funding in
addition to the Center contribution.
You should have an excellent track-
record, international reputation and
documented abilities. Our goal is to
build and maintain a highly integrated
research environment to harvest
synergies across the whole Center
and its International collaborators,
thus an enthusiastic, dedicated and
collaborative attitude is essential.
The positions in the Section of
Metabolic Genetics headed by
Professor Oluf B. Pedersen
www.metabol.ku.dk/scientific_
sections/metabolic_genetics/
E-mail: borbye@sund.ku.dk
• Two group leaders at the level of
full or associate professor - to build
and lead development of 1) a unit
for advanced statistical genetics
involving whole genome sequencing
studies of cardio-metabolic traits
and 2) a unit for metagenomics-
driven studies of the human gut
microbiome
• Two postdocs or assistant
professors - to work in the fields
of human genomics discovery,
genetic epidemiology and genetic
physiology in collaboration with
present investigators at the section.
The positions in the Section of
Integrative Physiology headed by
Professor Juleen Zierath
www.metabol.ku.dk/scientific_
sections/integrative_physiology/
E-mail: juleen.zierath@sund.ku.dk
• Group leader at the level of
assistant or associate professor -
to build and lead research aiming
at understanding the mechanism
by which lifestyle factors including
nutrition and physical exercise
influence whole body insulin
sensitivity and energy homeostasis
in Type 2 diabetes and obesity. The
potential candidate should have
documented experience in research
focused on the identification and
biological validation of metabolic
pathways and key regulatory genes
that cause insulin resistance in Type
2 diabetes.
• Two postdocs - to work in the area
of epigenetics and development
of insulin resistance together with
group leader, assistant professor
Romain Barres and on molecular
mechanisms controlling metabolism
in skeletal muscle together with
group leader, assistant professor
Jonas Thue Treebak.
The Novo Nordisk Foundation
Center for Basic Metabolic Research
The Novo Nordisk Foundation Center for
Basic Metabolic Research is an integral part
of Faculty of Health Sciences financed by
an 885 mill DKK (approx. 110 mio EURO)
unconditional, 10 year grant to University
of Copenhagen from the Novo Nordisk
Foundation (www.novonordiskfonden.dk/en).
The vision of the Center is to become
a leading center of excellence for
interdisciplinary basic metabolic research
evolving from the three thematic areas
to generate cutting edge and profound
knowledge about metabolic functions as the
basis for the development of novel means
of diagnosis, prevention and treatment of
common metabolic diseases.
uxi viis i r y oi coiixuacix
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g
www.metabol.ku.dk
The research activities at the Center include
interdisciplinary basic metabolic research in
the fields of
• Metabolic Genetics
• Integrative Physiology
• Metabolic Receptology &
Enteroendocrinology.
The Center has five Scientific Directors:
Professor Oluf B. Pedersen, Professor
Juleen Zierath, Professor Thue W. Schwartz,
Professor Jens Juul Holst, Professor Gerald
Shulman and a Director for Science
Communication, Professor Thomas
Söderqvist.
The positions in the Section of
Metabolic Receptology and
Enteroendocrinology headed by
Professor Thue W. Schwartz
www.metabol.ku.dk/scientific_
sections/metabolic_receptiology_and_
enteroendocrinology/
E-mail: tws@sund.ku.dk
• Two group leaders at the
assistant or associate professor
level - to build and lead research
in novel, complementing areas of
Enteroendocrinology. Candidates
with a background in, for example
in vitro and in vivo bioimaging,
sensory biology or endocrine
stem cell biology are particularly
encouraged to send letters of
interest.
• Two postdocs - to work in the
research area of molecular and
cellular Enteroendocrinology and/or
gut-brain axis in collaboration with
the research director or with group
leader, associate professor Birgitte
Holst.
If you are interested, please
send letter of interest asap or
preferably before August 15 to
metabol@sund.ku.dk.
Please note that this advertisement is
only a call for expression of interest.
The Faculty of Health Sciences comprises
app. 4500 students, app. 900 PhD
students and app. 2100 employees.
The Faculty creates new knowledge and
recognition through its core activities:
research, teaching, knowledge sharing and
communication. With basic research elds
ranging from molecular studies to studies of
society, the Faculty contributes to a healthy
future through its graduates, research
findings and inventions for the benefit of
patients and the community.
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g
P
o
s
t
d
o
c
S
u
r
v
e
y
S
pecial
C
ar
eer
F
eat
ur
e:
A
ugus
t
2
6,
2
0
1
1
R
eserve
your
ad
by
A
ugust
9
to
guarantee
space.
*
*Ads accepted until August 22 if space is still available.
Po ost stdo doct ctor orat atee
ggg
tdd
ttor orat atee
de de egr gr gree ee eesss
hhh
DDDDD
S
c
S
c
SSS
c
ie ie e
n
t
n
t t
n
t t
n
t
is is is is is
t
s
t
s
t
s
t
s
t
s
tt
s
P
h
P
h
P
hh
P
h
P
h
DDD
Why you should advertise in this issue of Science:
Content: Ph.D. scientists ready to make their
next career move rely on Science for sound
analysis and career advice. In this issue, we survey
postdoc supervisors who share insights on how to find the right fit
for a postdoc and career planning during the postdoctoral experience.
Reach: Science reaches over 130,000 subscribers every week and 66% of those
readers hold Ph.D. degrees, so you can be confident that you are reaching
your target market.
Results: When it comes to finding a postdoc, Science offers a simple formula:
relevant content that spotlights your ad + a large, qualified audience = your
hiring success.
To Book Your Ad, Contact: Midwest/West Coast/
South Central/Canada: Tina Burks Phone: 202-326-6577
E-mail: tburks@aaas.org
East Coast/Industry: Elizabeth Early Phone: 202-326-6578
E-mail: eearly@aaas.org
ScienceCareers.org
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g
Read inspiring stories
of women working in
“Green Science”
who are blending
a unique combination of
enthusiasm for science
and concern for others
to make the world
a better place.
This booklet is brought to you by the
AAAS/Science Business Office
in partnership with the
L’Oreal Foundation
Download this
free booklet
ScienceCareers.org/
LOrealWiS
WOMEN
IN SCIENCE
forging
new pathways in
green
science
California State Polytechnic University, Pomona (Cal Poly Pomona) invites nominations and appli-
cations for the position of Dean of the College of Science. The Dean is the chief academic officer of
the College of Science and provides academic leadership for faculty, staff, and students in the col-
lege. We seek a scholar with a vision for promoting excellence in undergraduate and graduate teach-
ing, research and other scholarly activities, as well as service. Candidates should have outstanding
leadership, management, and interpersonal skills, a record of excellence in teaching and externally-
funded scholarly activity, and a significant record of department, college, and university service. The
successful candidate will be expected to articulate and implement a vision for expanding faculty
research productivity in a way that enhances our educational programs. Detailed information about
this position may be found online at http://sci.csupomona.edu/dean_search/.
One of eight Colleges on the campus, the College of Science has approximately 2,800 undergradu-
ate majors in nineteen degree programs offered by seven academic departments: Biological
Sciences, Chemistry, Computer Science, Geological Sciences, Kinesiology & Health Promotion,
Mathematics and Statistics, and Physics. Master's degree programs enrolling approximately 230 stu-
dents are offered in Biological Sciences, Chemistry, Computer Science, Kinesiology, and
Mathematics. The College of Science is committed to excellence in teaching and scholarship and is
the largest single recipient of extramural funding on campus. Additional information about the
College may be found online at http://sci.csupomona.edu/.
Cal Poly Pomona, located about 30 miles east of downtown Los Angeles, is one of 23 campuses in
the California State University system. The University has an ethnically diverse student population
of 20,000. Students are enrolled in 62 undergraduate majors and 19 master's degree programs.
Additional information about the University may be found online at http://www.csupomona.edu/.
Review of applications will begin on September 16, 2011 and will continue until the position is
filled. The University seeks to fill this position by spring 2012, however, later dates are possible.
Applicants must forward (1) a letter of interest, which explains in detail the candidate’s administra-
tive philosophy and vision; (2) a curriculum vitae; and (3) the names and contact information of five
references (with a brief note describing the relationship of the references to the applicant) to: Sharon
Tanabe, Partner, stsearch@storbeckpimentel.com, RE: CPP DS2, Storbeck/Pimentel &
Associates, 1111 Corporate Center Drive, Suite 106, Monterey Park, CA 91754.
For a confidential inquiry or nomination, please contact Ms. Tanabe at (323) 260-5045.
California State Polytechnic University, Pomona is an Equal Opportunity, Affirmative Action Employer. Cal
Poly Pomona subscribes to all State and Federal regulations and prohibits discrimination based on gender,
race, sexual orientation, national origin, disability, marital status, age, religion, or covered veteran’s status.
DEAN
COLLEGE OF SCIENCE
(REOPENED)
POSTDOCTORAL
FELLOWSHIPS
CENTER FOR EPIGENETICS
Johns Hopkins University
School of Medicine
Postdoctoral fellowships are available
for the study of the epigenetic basis
of common human disease, including
genomic and mechanistic studies (see
epigenetics.jhu.edu; Nature Genetics
2009, 41:178-186; Nature 2010, 467:
285-290; Nature Biotech 2010, 10:
1949-1952; Nature Genetics 2011, doi:
10.1038/ng.865). The candidate should
be a prospective or recent recipient
of a Ph.D. or M.D./Ph.D. with good
publication(s) in genetics or genomics.
Please e-mail Dr. Andrew P. Feinberg
a curriculumvitae, statement of research
interests and names and email addresses
of three references, but no other attach-
ments, to afeinberg@jhu.edu or visit
our website at epigenetics.jhu.edu/
feinberglab.
Johns Hopkins University is an
Equal Opportunity/Afrmative Action
Employer.
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g
ScienceCareers.org
An exceptional career requires insightful planning and management.
That’s where Science Careers comes in. From job search to career
enhancement, Science Careers has the tools and resources to help
you achieve your goals. Get yourself on the right track today and get a
real career plan that works. Visit ScienceCareers.org.
Get a Career Plan that Works.
Download your free copy today at
ScienceCareers.org/booklets
Brought to you by the
AAAS/Science Business Office
FULBRIGHT/NORD-PAS DE CALAIS REGIONAL COUNCIL
awards will be granted to three to four Americans to conduct
research in Nord-Pas de Calais for periods of three to twelve
months. Bordering Belgium, Nord-Pas de Calais is France’s
northernmost region situated in the heart of the “Golden
Triangle” of Europe’s three great capital cies: London, Paris
and Brussels.
Eligible Candidates: Associate, assistant and full professors &
postdoctoral researchers
Eligible Fields: Hard, social and human sciences
Frenchlanguageabilityis commensuratewiththerequirements
of the project.
Starng Date: September 2012
Applicaon Deadline: August 1, 2011
For further informaon please visit:
www.cies.org, contact Zaneta Bertot, ZBertot@iie.org
www.fulbright-france.org, contact Ms. Patricia Janin, Franco-
American Commission for Educaonal Exchange at pjanin@
fulbright-france.org.
FULBRIGHT RESEARCH GRANTS
TO NORD PAS DE CALAIS, FRANCE
EUROPE’S “GOLDEN TRIANGLE’’
Neuroscience Position at
Georgia State University
Georgia State University (www.gsu.edu) invites applications for a senior
neuroscience faculty position at the associate or full professor level. In
exceptional cases, applicants at the assistant professor level will be con-
sidered. This position is part of a major initiative to enhance existing
strengths in neuroscience at GSU over the next three years coordinated
by the Center for Behavioral Neuroscience (www.cbn-atl.org) and the
GSU Brains and Behavior Program (http://brainsbehavior.gsu.edu)
working with GSU’s newly formed Neuroscience Institute (www.
neuroscience.gsu.edu). Preference will be given to applicants working
in behavioral or systems neuroscience. We are particularly interested in
research that complements the CBN’s focus on the mechanisms of social
behavior, neuroendocrinology, and the neuroscience of emotion. We are
open both to applicants using standard laboratory animal models and to
applicants with a neuroethological perspective using vertebrate or inver-
tebrate species. Successful applicants will be individuals who will take
advantage of the interdisciplinary collaborative research opportunities
available within the Center for Behavioral Neuroscience and the Brains
and Behavior Program.
Applicants must have a Ph.D. degree with relevant postdoctoral experience
and demonstrated ability to conduct an independent research program
using modern techniques. They must also have skills and interest in
teaching. Submit curriculum vitae, bibliography, statements of research
and teaching interests, and the names of three references, either electroni-
cally to Ms. Yolanda Barrow at ybarrow@gsu.edu, with the subject line
“Neuroscience Search”, or by mail to Neuroscience Institute, Attn. Ms.
TaraAlexander, Georgia State University, P.O. Box 5030, Atlanta, GA
30302-5030, USA. The review of applications will begin immediately,
and will continue until position is filled.
Georgia State University is an
Affirmative Action/Equal Opportunity Employer.
Assistant/Associate Professor in
Plant Sciences – Whole Plant Physiologist
Department of Plant Sciences, University of California, Davis
The successful candidate’s research will focus on integrative, whole plant physiology with emphasis on
environmental, developmental and/or stress physiology. Research areas may include but are not limited
to (1) physiological functioning to optimize crop management strategies under different environmental
conditions (2) identifying key phenotypic traits that could increase efficiencies in sustainable annual or
perennial plant cropping systems (3) understand and predict effects of environmental changes and manage-
ment practices on ecosystem dynamics. Candidate will be expected to operate well in multi-disciplinary
teams focused on the development of practical solutions to critical issues related to sustainable crop
production and ecosystem management practices in California and across the world.
The successful candidate is expected to teach core courses in the Plant Sciences undergraduate curriculum
and graduate courses within her/his area of research expertise in the Horticulture and Agronomy or Plant
Biology Graduate Group. Advising and mentoring of undergraduate and graduate students is required.
Participation in departmental, college, and campus committees and with state, regional and national
organizations is expected.
Expanded position description can be located at: http://www.plantsciences.ucdavis.edu/plantsciences/
databases/directory/news/jobs.asp
Qualifications: Ph.D. or equivalent level of experience in plant physiology, crop physiology, plant
biology, or related fields.
Salary: Commensurate with qualifications and experience.
TO APPLY: Candidates should begin the application process by registering online to apply: http:
//recruitments.plantsciences.ucdavis.edu. Please include statements of research and teaching interests,
curriculum vitae, publication list, copies of 3 of your most important research publications, copies of
undergraduate and graduate transcripts (if within 5 years of either degree), and the names, e-mail addresses,
and telephone numbers of at least 5 professional references. For technical or administrative questions
regarding the application process please e-mail plantsciences@ucdavis.edu. Reviewof the applications
for all positions will begin August 1, 2011. The position will remain open until filled.
UC Davis is an Affirmative Action/Equal Employment Opportunity Employer and is dedicated to
recruiting a diverse faculty community. We welcome all qualified applicants to apply, including
women, minorities, veterans, and individuals with disabilities.
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g
AAAS is here – helping scientists achieve career success.
Every month, over 400,000 students and scientists visit ScienceCareers.org in search of the information, advice, and opportuni-
ties they need to take the next step in their careers.
A complete career resource, free to the public, Science Careers offers a suite of tools and services developed specifically for
scientists. With hundreds of career development articles, a grants and scholarships database, webinars and downloadable
booklets filled with practical advice, a community forum providing real-time answers to career questions, and thousands of
job listings in academia, government, and industry, Science Careers has helped countless individuals prepare themselves for
successful careers.
As a AAAS member, your dues help AAAS make this service freely available to the scientific community. If you’re not a member,
join us. Together we can make a difference.
To learn more, visit aaas.org/plusyou/sciencecareers
Prince Albert II of Monaco/Instut
Pasteur Award
TheInstut Pasteur, theMonacoScienc Center
and the Prince Albert II of Monaco Foundaon
are pleased to invite applicaons for the Prince
Albert II of Monaco/Institut Pasteur Award
2012.
This biennial awardwill begiventoaninvesgator
in the eld of:
“Environmental changes and impacts
on Human Health”
The prize will be awarded on the occasion of the
SciencSymposiumdedicatedtoenvironmental
changes and their impacts on human health, to
be held on March 23rd, 2012.
The Award will honor an investigator under
fifty years old who has made outstanding
contribuonstotheeld. Thehonoreewill receive
a 40,000 Euros personal award. Candidates from
all relevant disciplines (ecological, biomedical,
and social sciences) are invited to submit their
applicaons, which will be evaluated by a Jury
of renowned scienc experts.
The dossier should include:
a one-page letter written by the nominee •
descri bi ng the i mportance of hi s/her
contribuon to the eld of “Environmental
changes and impacts on Health”.
a curriculum vitae, •
a list of scienc publicaons and/or books, •
a one-page outline of her/his plans for future •
research,
two leers of support. •
The deadline for submitting nominations is
October 30th, 2011.
All les should only be sent in electronic format,
as a single pdf le to: award@ec2h-monaco.
org
To learn more, please visit:
hp://www.ec2h-monaco.org
Assistant or Associate Research
Professor Position in Coastal Physical
Oceanography at the University of Maine
(Non-tenured)
The University of Maine School of Marine Sciences, in Orono, Maine, seeks
an innovative investigator with focus on ocean observing systems and general
observational physical oceanography. The successful candidate will join the
Physical Oceanography Group (PhOG), which designed and operates data buoy
arrays in Coastal Ocean Observing Systems (COOS) including the Gulf of
Maine Northeast Regional Association of Coastal Ocean Observing Systems
(NERACOOS) and the Caribbean Integrated Ocean Observing System (CarI-
COOS). The successful candidate is also expected to develop an active, funded
research program and assist in the teaching mission at the undergraduate and
graduate levels. This position has an anticipated start date on or before Janu-
ary 2, 2012. UMaine will provide up to nine months of salary in each of the
first two years. Subsequent support from UMaine will depend on performance
of the successful candidate and UMaine finances, but he or she is expected to
raise the bulk of his or her salary in subsequent years. Salary for this position
is competitive and will be commensurate with experience and potential.
Minimum Qualifications: Qualified applicants are required to have a PhD in
physical oceanography or a closely related physical science and have several
years of relevant professional experience. Experience with proposal writing
and observational analysis of data from Coastal Ocean Observing Systems
is preferred.
To Apply: Inquiries and applications should be directed to Neal R. Pettigrew,
Chair of Search Committee, at nealp@maine.edu and Susanne_Thibodeau
@UMIT.MAINE.EDU. Documents required include: curriculum vitae, a
letter of application, statement of research and teaching interests, and names
and contact information of three references. Review of applications will begin
immediately and will continue until the position is filled.
On January 1, 2011, UMaine joined a growing number - now nearly 400 - of
colleges and universities around the country that are tobacco-free. Imple-
mentation followed more than three years of study and planning managed
by a committee of faculty members, staff members and students. Agood deal
of information about UMaine’s plan and the new policy is online at http:
//umaine.edu/tobaccofree/.
The University of Maine is an
Equal Employment Opportunity/Affirmative Action Employer.
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g
AWARDS
POSITIONS OPEN
POSTDOCTORAL RESEARCH
POSITION
Virginia Commonwealth University (VCU)
School of Medicine invites applicants for post-
doctoral fellowships on their NIH Institutional
Training Grant entitled Training in Functional
Lipidomics in Cardiovascular and Respiratory Dis-
eases directed by Dr. Sarah Spiegel, chair of
Biochemistry and Molecular Biology. Fellowships
provide up to three years of funding for training
with outstanding faculty in laboratory and other
skills necessary for establishing an independent re-
search career. This interdepartmental training
program stems from VCU_s strength in signaling,
functional lipidomics, and cardiovascular and res-
piratory research. Candidates must have a Ph.D.,
M.D., or M.D.-Ph.D., and must be a U.S. citizen
or green card holder and should have experience
in biochemistry, molecular biology, or other rel-
evant fields. Submit a cover letter with curriculum
vitae and three reference letters to Dr. Michael
Maceyka, e-mail: mwmaceyka@vcu.edu.
RESEARCH FELLOW wanted to design, code,
and conduct psychophysical, brain imaging experi-
ments in area of psychology of perception. Must have
Master_s Degree in Environmental Studies or related
field and two years of experience in same type of job or
two years of experience in research laboratory setting,
including designing and conducting psychophysical,
fMRI, and MEG experiments; designing and compos-
ing study proposals for funding; conducting mathe-
matical modeling; and developing Matlab programs
capable of producing detailed moving stimuli. Send re-
sume to: Dr. Takeo Watanabe, Dir., Vision Sciences
Lab., Boston University Department of Psychology,
64 Cummington Street, Boston, MA 02215.
POSTDOCTORAL FELLOW
The Musculoskeletal Research Center at New York
University Langone Medical Center is recruiting for a
postdoctoral position investigating the immunological
mechanism of progranulin_s anti-inflammatory effect
(see: W. Tang et al., Science, 332(6028):478, 2011).
Qualified applicants must hold an M.D. or Ph.D. Pref-
erence will be given to those who have experience in
Treg, Th17, and autoimmune diseases. Interested appli-
cants should send a cover letter, curriculum vitae, and
three letters of recommendation to Kevin Rochford,
e-mail: kevin.rochford@nyumc.org, or Dr. Chuanju
Liu, e-mail: chuanju.liu@med.nyu.edu.
CAREER OPPORTUNITY—Doctor of Optom-
etry (O.D.) degree in 27 months for Ph.D.s in sci-
ence and M.D.s. Excellent career opportunities for
O.D.-Ph.D.s and O.D.-M.D.s in research, educa-
tion, industry, and clinical practice. This unique pro-
gram starts in March of each year, features small
classes, and 12 months devoted to clinical care.
Contact the Admissions Office, telephone: 800-
824-5526 at: The New England College of Op-
tometry, 424 Beacon Street, Boston, MA 02115.
Additional information at website: http://www.
neco.edu, or by e-mail: admissions@neco.edu.
We deliver
customized job alerts.
www.ScienceCareers.org
MARKETPLACE
8¢/u
Truncated
Taq DNA
Polymerase
Withstand 99
o
C
US Pat #5,436,149 e-mail: abpeps@msn.com
Call: Ab Peptides 1•800•383•3362
Fax: 314•968•8988 www.abpeps.com
Widely
Recognized
Original &
Guaranteed
KlenTaq1
C
A
R
E
E
R
T
R
E
N
D
S
Careers Away
from
the Bench
Advice and Options for Scientists
This booklet is brought to you by
the AAAS/Science Business Office
Fromtechnology specialists to patent
attorneys to policy advisers, learn more
about the types of careers that scientists
can pursue and the skills needed in order
to succeed in nonresearch careers.
Download
your free copy
today.
ScienceCareers.org/booklets
8 JULY 2011 VOL 333 SCIENCE www.sciencecareers.org 254
o
n
l
i
n
e

@
s
c
i
e
n
c
e
c
a
r
e
e
r
s
.
o
r
g

CONTENTS
SPECIAL SECTION

Volume 333 Issue 6039

Galaxy Evolution
INTRODUCTION 169 A Universe of Galaxies NEWS 170 Milky Way Researchers’ Home Away From Home
173

PERSPECTIVE 176 Galactic Paleontology
E. Tolstoy

REVIEWS 178 The Cosmic History of Star Formation
J. S. Dunlop
182

Galaxy Zoo Volunteers Share Pain and Glory of Research

The Coevolution of Galaxies and Supermassive Black Holes: A Local Perspective
T. M. Heckman and G. Kauffmann

>> Policy Forum p. 161

page 148

EDITORIAL 136 A Threat to Medical Innovation
Michael Rosbash

NEWS FOCUS
148 NUCLEAR WASTE

NEWS OF THE WEEK 140 A roundup of the week’s top stories NEWS & ANALYSIS 143 Once on ‘Fast Track,’ Avastin Now Derailed 144 Foreign Researchers Begin to Make Their Mark 145 Unseen World of Clinical Trials Emerges From U.S. Database 146 Cash Advance, New Approach Aim to Relaunch Biosphere 2 147 DARPA Offers $30 Million to Jump-Start Cellular Factories

Waste Panel Expected to Back Interim Storage
>> Science Podcast

BOOKS ET AL. 159 Science as Psychology
L. M. Osbeck et al., reviewed by R. N. Giere

150 153

Light at the End of the Radwaste Disposal Tunnel Could Be Real Picking Up the Pieces at Ravaged Tohoku University
Facilities Plot Research Revival

160

R. Hersh and V. John-Steiner, reviewed by L. A. Steen

Loving and Hating Mathematics

POLICY FORUM 161 A Dark Age for Space Astronomy?
R.-M. Bonnet and J. A. M. Bleeker >> Galaxy Evolution section p. 169

LETTERS 156 Seeds of Change for Restoration Ecology
R. Aerts and O. Honnay

Manufacturing Decline Yields Drug Shortages
P. Connelly and B. P. Quinn

PERSPECTIVES 163 Lipases in Cachexia
P. Arner >> Report p. 233 E. E. Eyler >> Report p. 196
165

R. Frodeman and J. B. Holbrook
158

NSF’s Struggle to Articulate Relevance CORRECTIONS AND CLARIFICATIONS

164

Precision, Not Power A Critical Point for Turbulence Sex, Death, and the Red Queen

B. Eckhardt >> Research Article p. 192
166

M. A. Brockhurst >> Report p. 216

CONTENTS continued >>

COVER Composite image of the Sombrero galaxy captured by the Hubble Space Telescope in 2003. Like the Milky Way, the Sombrero is a spiral galaxy with a central stellar bulge and a thin disk containing spiral arms of newly formed stars. The special section beginning on page 169 examines the ways in which galaxies change over time.
Image: NASA and The Hubble Heritage Team (Space Telescope Science Institute/Association of Universities for Research in Astronomy)

DEPARTMENTS
133 137 139 243 244

This Week in Science Editors’ Choice Science Staff New Products Science Careers

www.sciencemag.org

SCIENCE

VOL 333

8 JULY 2011

127

Published by AAAS

CONTENTS

BREVIA
186

213

Adult Neural Function Requires MeCP2

High Pre-Eruptive Water Contents Preserved in Lunar Melt Inclusions

C. M. McGraw et al. An epigenetic program regulated by MeCP2 needs to be maintained throughout life for normal neurological function.

E. H. Hauri et al. Primitive magmatic melt inclusions from the Moon contain as much water as some terrestrial mid-ocean ridge magmas.
216

RESEARCH ARTICLES
187

Glycolytic Oscillations and Limits on Robust Efficiency

Running with the Red Queen: Host-Parasite Coevolution Selects for Biparental Sex

F. A. Chandra et al. Gylcolytic oscillations in yeast are a by-product of a trade-off between robustness and efficiency.
192

L. T. Morran et al. Outcrossing provides better survival than self-fertilization during coevolution between a host and its parasite. >> Perspective p. 166
218

The Onset of Turbulence in Pipe Flow

K. Avila et al. The lifetimes of injected jet puffs are used to determine the critical point at which turbulent pipe flow is sustained. >> Perspective p. 165
222

Isolation of Single Human Hematopoietic Stem Cells Capable of Long-Term Multilineage Engraftment

F. Notta et al. Proteins are identified that underlie the early commitment steps of human hematopoietic stem cell differentiation.

page 159

REPORTS
196

Frequency Metrology in Quantum Degenerate Helium: Direct Measurement of the 2 3S1 → 2 1S0 Transition
R. van Rooij et al. Measurement of an extremely weak spectroscopic transition in helium hones fundamental atomic theories. >> Perspective p. 164

Coupled, Circumferential Motions of the Cell Wall Synthesis Machinery and MreB Filaments in B. subtilis
E. C. Garner et al.

225

199

An Extremely Luminous Panchromatic Outburst from the Nucleus of a Distant Galaxy
A. J. Levan et al.

J. Domínguez-Escobar et al. Bacteria elongation involves moving synthetic complexes around the cell wall
228

Processive Movement of MreB-Associated Cell Wall Biosynthetic Complexes in Bacteria

Phosphorylation of the Autophagy Receptor Optineurin Restricts Salmonella Growth

pages 164 & 196

203

A Possible Relativistic Jetted Outburst from a Massive Black Hole Fed by a Tidally Disrupted Star

P. Wild et al. Phosphorylation of an autophagy receptor restricts pathogenic cytosolic bacterial growth.
233

J. S. Bloom et al. A recent bright emission observed by the Swift satellite is due to the sudden accretion of a star onto a massive black hole.

Adipose Triglyceride Lipase Contributes to Cancer-Associated Cachexia
S. K. Das et al. Ablation of a gene controlling fat breakdown can protect mice from cancer-associated uncontrolled loss of fat and muscle. >> Perspective p. 163

206

Observation of Transient Structural-Transformation Dynamics in a Cu2S Nanorod
H. Zheng et al. Structural fluctuations between two equilibrium phases are observed in copper sulfide nanoparticles.

238

A Pericyte Origin of Spinal Cord Scar Tissue

page 238

209

Palladium-Catalyzed Aerobic Dehydrogenation of Substituted Cyclohexanones to Phenols

C. Göritz et al. Scars formed in response to damage to the central nervous system show unexpected complexity. >> Science Podcast

Y. Izawa et al. Phenol derivatives are prepared from nonaromatic ring compounds that can bear a wide variety of substitutents.

CONTENTS continued >>

www.sciencemag.org

SCIENCE

VOL 333

8 JULY 2011

129

Published by AAAS

484460) paid at Washington. Caribbean (surface mail) $55.1204352 RESEARCH ARTICLE: Identification of a Lysosomal Pathway That Modulates Glucocorticoid Signaling and the Inflammatory Response Y. giving old and new addresses and 8-digit account number. Long-Dead Cane Toads Continue to Haunt Australian Wildlife Road-kill “toad jerky” is highly toxic. Raper Scientists at the MRC Laboratory of Molecular Biology attribute the lab’s success to bold thinking.org SCIENCEXPRESS SCIENCESIGNALING Herschel Detects a Massive Dust Reservoir in Supernova 1987A www. chimps can pick out meaning even when the sounds of words are distorted. airmail. Copyright © 2011 by the American Association for the Advancement of Science.sciencetranslationalmedicine.sciencecareers.sciencenow. Authorization to photocopy material for internal or personal use under circumstances not falling within the fair use provisions of the Copyright Act is granted by AAAS to libraries and other users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service. First class.org The Signal Transduction Knowledge Environment 5 July issue: http://scim.00 per article is paid directly to CCC. MA 01923. CRASY: Mass.O. bulk rates on request. Dawes et al.sciencemag.or Electron-Correlated Rotational Alignment Spectroscopy C. Washington. Science is indexed in the Reader’s Guide to Periodical Literature and in several specialized indexes.ag/mrc_lmb_nobel Schema-Dependent Gene Activation and Memory Encoding in Neocortex D.ag/stm070611 C. other countries (air assist delivery) $85. M. Publications Mail Agreement Number 1069624. SCIENCECAREERS A Highly Conserved Neutralizing Epitope on Group 2 Influenza A Viruses PODCAST www. M. Box 96178. www.sciencemag. Primordial germ cells are directed toward oogenesis even before they migrate to the gonads of the fruit fly. B. High Value and Long Life—Double Jeopardy for Tunas and Billfishes B. VAN ANDEL RESEARCH INSTITUTE C.org SCIENCE VOL 333 8 JULY 2011 131 Published by AAAS .ag/kravlovics Drosophila Sex lethal Gene Initiates Female Development in Germline Progenitors K. DC 20005. Sherwood et al.sciencesignaling. 10. and a collaborative culture.ag/cane-toads RESEARCH ARTICLE: CXCL5 Mediates UVB Irradiation–Induced Pain J. M.00 current issue.1205983 RESEARCH ARTICLE: Sequential Phosphorylation of Smoothened Transduces Graded Hedgehog Signaling Y. except the last week in December. GST #1254 88122. P.ag/ss070511 M. Canadian rates with GST available upon request. Danvers.1208730 PERSPECTIVE: Development of Newborn and Infant Vaccines G. The large amount of dust produced by this supernova may help explain the dust observed in young galaxies. provided that $25. 1200 New York Avenue. and additional mailing offices.S. by the American Association for the Advancement of Science. 10. Foreign postage extra: Mexico.1126/science.org/career_magazine Free Career Resources for Scientists D.S. DC 20090–6178.1208146 COMMENTARY: Reengineering Translational Science—The Time Is Right SCIENCEPODCAST F.1126/science. The title SCIENCE is a registered trademark of the AAAS.sciencemag. 10. Periodicals Mail postage (publication No. http://scim. Domestic individual membership and subscription (51 issues): $149 ($74 allocated to subscription).org Integrating Medicine and Science 6 July issue: http://scim. http://scim. Postmaster: Send change of address to AAAS. Change of address: Allow 4 weeks. SCIENCESIGNALING Lysosomes regulate glucocorticoid signaling.00 back issue prepaid includes surface postage. AND MOLECULAR MICROSCOPY. A Nobel Prize–Winning Culture PERSPECTIVE: Bacterial Scaffolds Assemble Novel Higher-Order Complexes to Reengineer Eukaryotic Cell Processes V. 10. Chimps Are Good Listeners. $15. CELLULAR. C. An antibody against a conserved epitope broadly neutralizes group 2 influenza viruses. Collins The new National Center for Advancing Translational Sciences aims to revamp the process of biomedical research translation. Sanchez-Schmitz and O..ag/_hyperthermia J. http://scim. 10. www. F. and more. http://scim.1126/science.org Highlights From Our Daily News Coverage SCIENCENOW Magnetic Nanoparticles Fry Tumors RESEARCH ARTICLE: Optimization of Dosing for EGFR-Mutant Non–Small Cell Lung Cancer with Evolutionary Cancer Modeling “Magnetic hyperthermia” proves more effective in mice than a traditional cancer drug.ag/chimp-listen RESEARCH ARTICLE: Deep Sequencing of the Human TCRγ and TCRβ Repertoires Suggests That TCRβ Rearranges After αβ and γδ T Cell Commitment A. secure funding. Too Like humans. Predictive models of EGFR-mutant tumor behavior point to alternative drug dosing strategies to prevent and treat acquired resistance. The identification code for Science is 0036-8075. Chmielecki et al. Domestic institutional subscription (51 issues): $990. Leong Bacterial effector proteins can act as novel scaffolds for signaling proteins in infected mammalian cells. Levy Insights into immune ontogeny will inform translation of new vaccines that are safe and effective for newborns and infants.1126/science. The cytokine CXCL5 is a peripheral mediator of pain induced by UVB irradiation to the skin. 10. Schröter et al. Wald Molecular geneticist Robert Kralovics traces his character as a scientist to growing up as a farmer’s son in a Communist regime. student.A.org/multimedia/podcast Free Weekly Show On the 8 July Science Podcast: the cellular origin of spinal cord scar tissue. Washington. He and A.1204839 >> Science Podcast Y.org/scienceinsider Science Policy News and Analysis www. Single-copy sales: $10. He et al. DC. Printed in the U. Hashiyama et al. 222 Rosewood Drive. VanHook Inhibition of lysosome function promotes the anti-inflammatory effects of glucocorticoid signaling. http://scim. SCIENCE (ISSN 0036-8075) is published weekly on Friday. suggesting yet again that human language has roots in other species. the future of nuclear power in the U. It takes two kinases and two phosphatases acting on a single membrane protein Smoothened to interpret the Hedgehog morphogen gradient. Tse et al. Ekiert et al. even after months in the Sun.1126/science. S. A technique merging rotational spectroscopy with mass spectrometry facilitates analysis of a complex isotopic mixture. Matsuura et al. Su et al.CONTENTS SCIENCEONLINE www. Collette et al.1205274 CREDIT: ERIC HUDSON/LABORATORY OF ANALYTICAL. M. New hippocampal-dependent learning is in parallel consolidated with existing memories in the neocortex. and emeritus rates on request.1126/science. Deep sequencing provides new insights about T cell receptor rearrangement in humans. Lesser and J. A Farmer’s Son Cultivates a Career in Translational Research SCIENCETRANSLATIONAL MEDICINE www. NW. SCIENCEINSIDER news. neutralizing influenza.sciencexpress.

The advantage of this approach is that the cyclohexanone precursors are available with substitution patterns that are difficult to achieve selectively when using standard protocols for direct aromatic ring elaboration. Finding the point where the lifetime of a single puff reached a maximum allowed the minimum Re required to sustain turbulent flow to be determined. Regions were observed to flip back and forth between the two phases. and plastics. EDITED BY STELLA HURTLEY Volatile Moon Rocks The presence of water in the Moon is a highly debated question. Osborne Reynolds described the key factors that influence the transition of a flowing fluid from a smooth. HAURI ET AL. Extremely high-quality transmis- Off With Her Head! The Red Queen hypothesis suggests that coevolution results from the evolutionary race between interacting species resulting in a Continued on page 135 www. Swift Encounter At low Re. 209. which could be described in terms of simplified thermodynamic fluctuation arguments. On 28 March 2011. but they are extremely improbable. in turn. The results suggest that parts of the lunar interior are more volatile-rich than prior studies have indicated. Known as the Reynolds number (Re). Zheng et al. but it has not been possible to pin down a critical transition point. glycolytic intermediates in yeast oscillate. Turbulent Times In 1883. This is surprising because the Moon was thought to have lost most of its volatile content immediately after its formation. (p. Caught in the Act Under conditions where a structural phase transition can occur.” These transitions. CREDITS (TOP TO BOTTOM): MARK GARLICK/UNIVERSITY OF WARWICK. (p. (p. Design is constrained by tradeoffs between making the system efficient and robust while minimizing output error. multiple phases of a material coexist as it transforms from one equilibrium phase to the other. 203. Under certain conditions. pigments. Traditionally. Bloom et al. Izawa et al. 196. which would otherwise evaporate during volcanic eruption. the puff eventually decayed. relying on extended interaction times between the atoms and light field. pesticides.Operating at the Limits Control theory provides a mathematical basis for engineering the dynamic behavior of a system by using feedback. 2011 . Chandra et al. published online 26 May) present measurements of dissolved volatiles. see the Perspective by Eckhardt) examined the fate of water jet puffs injected into a stream of flowing water. certain transitions between specific energy states are termed “forbidden. which was caused by a Mars-sized object hitting Earth. Avila et al. published online 16 June) present a physical model to explain the origin of the burst: The tidal disruption of a star as it passed close to the galaxy’s central black hole produced a relativistic jet pointed toward Earth. (p.org SCIENCE VOL 333 Published by AAAS 8 JULY 2011 133 Downloaded from www. 199. the ratio of inertial forces to viscous forces is used to predict the change in flow behavior at a critical value for a specific flow geometry. In simple pipe flow. to characterize the rare excitation events. van Rooij et al.sciencemag. and may be more akin to Earth’s modern upper mantle. published online 9 June) present an alternative strategy: palladium-catalyzed oxidation of cyclohexanone derivatives (essentially hydrogenated precursors of the desired phenols). with oxygen accepting the liberated hydrogen to form water. Levan et al. see the Perspective by Eyler) measured the frequency of one such transition in helium (from the lowest triplet state to the second-lowest singlet state). the onset of turbulence has been estimated to occur at Re values between 1900 and 2100. a puff split into two by absorbing energy from the flowing liquid. 213. as well as highly sensitive detection. including water. 206) devised a method to study the transformation between two solid phases in copper sulfide nanoparticles. turbulent flow. afforded a stringent test of the theoretical framework describing atomic structure and light-matter interactions. (p. while at high Re. 192. however.35. 187) have now applied control theory to analysis of a minimal model of glycolysis and show that the oscillations are a consequence of operating at the hard limits of maximizing robustness. chaotic. these intermediates are prepared through modification of intact aromatic rings. (p. Forbidden Territory In the parlance of quantum mechanics.sciencemag. This frequency. such as drugs. A Different Route to Phenols Phenol derivatives are essential intermediates in the preparation of many commercial organic compounds. but the basis for these oscillations has been unclear. Hauri et al. in melt inclusions from a sample of ancient lunar magma brought back by Apollo 17. the NASA Swift satellite detected a highenergy outburst that behaved completely differently from the classical gamma-ray bursts that the telescope was designed to study and detect. Melt inclusions thus preserve their volatile content. concluding that it occurred in the nucleus of a galaxy with a redshift of 0. sion electron microscopy and image analysis were able to distinguish between the two phases within the nanoparticle as the transition temperature was approached. Glycolysis is a central metabolic pathway that consumes glucose to generate adenosine triphosphate (ATP) with two key enzymes that are feedback-inhibited by ATP. Melt inclusions are small pieces of molten rock that get trapped in crystals that grow in magma. while minimizing metabolic overhead and enzyme complexity. (p. laminar flow to a choppy. Using precise laser spectroscopy in tandem with laser cooling. published online 16 June) present comprehensive observations of this unusual event. do occur.org on July 7. (p.

Together we can make a difference. www. published online 2 June) and Garner et al. (p. which kills about 15% of cancer patients. Morran et al. please call (800) 252-4910 17050 Montebello Road Cupertino. computer science. All prices are per person twin share + air For a detailed brochure. Pergamum. AAAS started Entry Point!. Caenorhabditis elegans nematodes were infected with the bacteria Serratia marcescens. Also see the miniature Venice (Comacchio) with superb leadership. Instead. 2011 Explore Trieste and Aquileia (the second Rome). Outcrossers had lower mortality rates when infected than obligate selfers. and intracellular pathogens. Thus. Domínguez-Escobar et al. To learn more. (p. they retained a normal fat mass and showed reduced loss of skeletal muscle. including microtubule-associated protein light chain 3 (LC3). Assos. 2011 October 3-14. engineering. Wild et al. published online 26 May) describe optineurin (OPTN) as an autophagy receptor whose function is regulated by phosphorylation of its LC3-interacting motif. 223. 228. 222. Summer Internships Students with Disabilities To meet the challenge of the competitive economy in the new millennium.com Optineurin in Autophagic Bacterial Clearance Autophagy receptors bind both ubiquitin and autophagy markers. Studying two models of tumor-bearing mice. Thus. 2011 Discover the cultural wonders of Turkey during the rain of falling stars known as the Draconid Meteor Shower. see the Perspective by Arner) show that adipose triglyceride lipase (ATGL). Venice.org/plusyou/entrypoint CREDIT: DOMÍNGUEZ-ESCOBAR ET AL.com www. Das et al. visit: aaas. Two papers now challenge this picture.This Week in Science Continued from page 133 seemingly stationary situation. elegans were either wild-type (can cross with other individuals or self-fertilize). see the Perspective by Brockhurst) explore the role of the Red Queen in the case of a host-pathogen system with the presence and absence of outcrossing and coevolution. (p. 225. a program that offers students with disabilities competitive internship opportunities in science.995 + air Archaeology & Draconid Meteor Shower Turkey: .sciencemag. Waste Not Cancer-associated cachexia is a wasting syndrome characterized by uncontrolled loss of fat and muscle mass. Ephesus. These findings provide a method to assist in systematic studies of human HSC biology and the development of approaches to harness the regenerative potential of HSCs for clinical applications.695 + air Venice: Round the Lagoon Cord Blood Delivers Continuous blood-cell production during an organism’s life depends on rare. Email: AAASInfo@betchartexpeditions. the Po Delta and Calera Botanical Garden. and some fields of business. but reduced levels of outcrossing in later generations. obligate outcrossers.org on July 7. (p. October 18-30. mathematics. (p.org SCIENCE VOL 333 8 JULY 2011 Published by AAAS Downloaded from www. See Istanbul. even if infected by coevolving bacterial strains. self-renewing hematopoietic stem cells (HSCs). Mutant mice that were genetically deficient in ATGL were protected from cancer–associated cachexia. Troy & more. Phosphorylation by the protein kinase Tank binding kinase 1 (TBK1) increased the affinity of OPTN for autophagy modifiers by 13-fold. TBK1 and OPTN represent critical components of the cell defense system for restricting the growth of bacteria in the cell. Join us. OPTN is also a ubiquitin-binding protein and was recruited to cytosolic Salmonella to promote bacterial clearance via the autophagy pathway. The ability to characterize human HSCs has been limited because isolation methods cannot distinguish HSCs from progenitors that only participate in transient hematopoietic regeneration.betchartexpeditions. published online 2 June) studied the dynamic interactions between bacterial actins and the cell wall elongation machinery in Bacillus subtilis cells and found that MreB proteins in actively growing cells did not form helical filamentous structures. The populations of C. selection imposed by coevolving pathogens may account for the widespread prevalence of outcrossing in nature. is essential to the pathogenesis of cancer-associated cachexia. $3. or obligately self-fertilizing lines. from human umbilical cord blood and successful transplantation of a single HSC into immunodeficient mice. and promote the specific clearance of protein aggregates. such as transplantation.sciencemag. $3. the proteins formed discrete patches that moved processively along peripheral tracks perpendicular to the cell axis. 216. Bacterial Elongation Mechanism One of the defining concepts of recent research on bacterial cytoskeleton organization and morphogenesis is the helical organization of actin-like MreB proteins and the role of these helices in the spatial organization of cell wall biosynthesis. Cappadocia. 218) now report isolation of a near-homogeneous population of HSCs. (p. defective organelles. Pharmacological inhibition of ATGL may thus merit investigation as a potential treatment for cachexia. Wild-type populations initially showed higher levels of outcrossing in the presence of the virulent bacteria. California 95014 BETCHART EXPEDITIONS Inc. Notta et al. an enzyme that breaks down stored fat. published online 16 June.

(LEFT) TODD DAVIDSON/ALAMY is 1% less than in 2010. not projects” direction throughout the NIH will be essential to promote true innovation. their cost/benefit ratios should be honestly examined. recombinant DNA and monoclonal antibodies emerged from fundamental research in bacteriology and immunology. Downloaded from www. Although long-standing constituencies make it hard to consider ending or even reducing these programs. the current situation makes grant evaluation nearly impossible and is putting truly excellent laboratories out of business. innovative investigator-initiated research is in dire straits.gov/NIHDatabook/Charts/Default. before money goes out the door. not projects” strategy is already practiced by the Howard Hughes Medical Institute. Food and Drug Administration. the former CEO of Merck. Dramatic movement in the “fund people. In addition. In the spirit of “never waste a good crisis. – Michael Rosbash Michael Rosbash is an Investigator of the Howard Hughes Medical Institute and a professor of Biology at Brandeis University.aspx?showm=Y&chartId=34&catId=2.edu. NIH Director Francis Collins recently testified before a Senate subcommittee that in FY 2011 only 17 to 18% of grant applications would be funded. the same qualities required for innovative research contributions. Moving forward in an era of decreasing real budgets will require hard work as well as the courage to see and tell the truth.S. which came from research on plants and worms. They include centers and other large collective funding efforts as well as expensive clinical and epidemiological research. Because collective funding efforts. but these programs make up a small fraction of total NIH funding.S. 2011 . For example. There have been 5 years of no real growth in the NIH budget.* But the prospects moving forward are bleak. NIH programs. respectively. These technologies gave birth to the biotechnology industry and underlie many therapeutics approved by the U. also compete with the bottom-up R01 program for funding.EDITORIAL A Threat to Medical Innovation THESE ARE DIFFICULT TIMES FOR THE U. The Wellcome Trust and individual European Research Council grants are moving in this direction.sciencemag. saying that NIH should stick to supporting new knowledge and discoveries.sciencemag.org Published by AAAS CREDITS: (TOP) BRANDEIS UNIVERSITY. Roy Vagelos. and it is increasingly common for NIH to administratively decrease the size of awards across the board.1126/science.com/health/2011/04/15/former-merck-head-vagelos-says-nih-should-stick-to-basic-research.nih. as well as targeted R01s (responses to specific program announcements from individual NIH institutes). only 17 Pioneer Awards were awarded in FY 2010.nih. such as program project and center grants. 136 8 JULY 2011 VOL 333 SCIENCE www. given the absence of stimulus funds and the building momentum for cutting federal discretionary spending. For example. The federal stimulus package buffered the impact of NIH budget issues.1210374 *NIH Recovery Act Grant Awards: Fiscal Year 2009 spreadsheet (http://report. is highly critical of new NIH efforts to put increased resources into translational and applied research efforts.” a serious evaluation of many NIH extramural policies and programs is warranted.org on July 7. ITS 2011 BUDGET 10.gov/recovery). With 10 to 15% paylines at some institutes (or even less). the lowest level on record. A more recent example is RNA interference. approximately 36% of NIH grant awards derived from the stimulus in FY 2009 were individual research awards (R01s). This critical part of the NIH research effort forms the basis for future medical progress and must be returned to good health. explaining in part why the number of grants awarded for research ideas originated by individual scientists (untargeted R01s) has been stagnant.‡ Note that the most important breakthroughs often come from unexpected areas of inquiry. The key concept is that we do not know from which life science discipline or even organism the next great medical advance will emerge. It also explains why paylines (the percent of applications funded) are at or near record lows. the NIH intramural research program receives approximately 10% of NIH dollars without being subject to the same level of competitive merit review as the rest of NIH-supported research. recognize innovation by outstanding scientists. A “fund people. including the Pioneer Awards and New Innovator Awards.wsj. http://report. E-mail: rosbash@brandeis. ‡http://blogs. For example. NATIONAL INSTITUTES OF HEALTH (NIH).

Captain Ault.sciencemag. — BH Earth Planet. Sadly. 10. the master regulator of sex determination. the approach could lead to longer records here and elsewhere. Three groups were chosen for identification and counting to represent distinct trophic levels: small copepod crustaceans.1093/plankt/fbr060 (2011).000 years. Ceratium dinoflagellate alga. Lett. the Carnegie’s adventure came to an end when it. 1 (2011). which extends over about 150. The sediments were deposited during a well-known magnetic field reversal. we need to know not only what exists but what existed. Built for oceanography. — LMZ PLoS Genet. sought to investigate the role of this complex in regulating sex determination in Drosophila.EDITORS’CHOICE EDITED BY KRISTEN MUELLER AND JAKE YESTON DEVELOPMENT Controlling Sex Chromosomes are packaged into transcriptionally silent heterochromatin and transcriptionally active euchromatin. and heterochromatin-rich telomeres are capped by a protein complex composed of HP1a and HP1/ORCAssociated Protein (HOAP). NASA ECOLOGY Carnegie Blows Up Biodiversity To assess the effects of anthropogenic changes on biodiversity.org on July 7. on its last global voyage it was equipped to systematically collect plankton. and uncertainties in dating of both these fossils and climate records make correlation at the necessary resolution (within 100. and the HP1a-HOAP complex shows similarity to the mammalian sex-determining region of the Y chromosome (SRY). The authors were able to place 12 hominid fossils into the finely calibrated sequence. Further analyses showed that this was due to repressive activity by HOAP and both repressive and activating functions of HP1a that affected the function of the establishment promoter of Sex-lethal. Flies mutant in HP1a or HOAP exhibited defects in sex determination. — CA J. e1002122 (2011). proteins typically associated with heterochromatin are critical for regulating the changes in gene expression required for sex determination in flies.000 years) problematic. 7. www.sciencemag. Plankton Res. Joordens et al. Plankton were obtained from 160 sampling stations. Demonstrating a specific relation. were blown up while refueling in Samoa. Li et al. Although the age span is too brief to establish a larger relation between climate change and human evolution. most were rare and few were common. Dolan contends in an analysis of historic plankton samples collected on the final cruise of the ship Carnegie during 1928–1929. 307. however. Thus.000 years. and more species were collected at the tropics than at high latitudes. allowing an accurate time-climate sequence to be constructed. Reanalysis of these collections revealed that changes in species richness were correlated for the three groups. The authors carried out gene expression analysis in HOAP-deficient flies and found that the majority of down-regulated genes were those associated with the testis. Although potentially a rich source of now-unfunded taxonomic expertise. Interestingly. which varies over about 21. Continued on page 138 CREDITS (TOP TO BOTTOM): COURTESY OF THE CARNEGIE INSTITUTION FOR SCIENCE. and ciliate zooplankton called tintinnids. C L I M AT E S C I E N C E In Synch with the Weather Cooling climates in Africa over the past 10 million years have led to substantial environmental changes that might have influenced human evolution. and thus the lake Sr chemistry varied during monsoon wet and dry periods. has been difficult because the human fossil record is sparse. examined strontium isotopes from fish fossils in sediments from Lake Turkana that also contain human fossils. historical data do have gaps and study design issues that cannot now be resolved. Sci. To help tackle some of these issues. Rivers feeding the lake drain rocks with different Sr isotope compositions. and its scientist. The highly conserved protein HP1a marks heterochromatin in Drosophila. Because knockdown of HP1a is associated with loss of male viability. focusing on an interval around 2 million years ago.org SCIENCE VOL 333 8 JULY 2011 137 Published by AAAS Downloaded from www. implies that this variation primarily reflects the precession of Earth’s orbit. 33. The record. 2011 . of the several hundred species.

H3K4 methylation. Sarvan et al. DEVELOPMENT of genetic.S.org Downloaded from www. and Sarvan et al. however. — JSY Angew. Adding a diol solvent inches the reaction over the line to thermodynamic favorability. the kick isn’t quite vigorous enough.2011.EDITORS’CHOICE Continued from page 137 STRUCTURAL BIOLOGY Chromatin Regulation Unraveled Histones are proteins that act to package DNA into chromatin. 247401 (2011). The methylation of lysine 4 on histone H3 is correlated with active gene expression in eukaryotes. and maximal expression of the β-globin gene. This process is enthalpically favorable.sciencemag. but full activity requires the presence of other complex components. Schaub and Paciello found. With a model. 50. the authors are then able to determine the degree of friction between them. Natl.sciencemag. Indeed. In such ambipolar devices.201101292 (2011). relatively constant phenotype in the face 8 JULY 2011 Published by AAAS Solvated Past the Finish Line As concerns mount about the adverse impact of atmospheric CO2 on climate. whereas Chen et al. In mammals. 10. Both found that besides a zinc finger. The zinc-finger domain was previously proposed to bind histone tails. 10.A. the dynamic range of response was reduced by a factor of 10. a DNA binding motif that is often involved in transcription regulation. Chem. — V V EMBO Rep. As such. Acad. loss of feedback inhibitors led to a several-fold increase in variability in tail length during development. and stochastic variation or noise. have determined the structure of the N-terminal domain of human Ash2L. Ed. in the context of the structure. 2011 . Thus. Human embryonic kidney cells responded to a range of BMP concentrations that varied almost 100-fold. it lacks features required for such binding.1038/embor. Rev. 10202 (2011). Int. Nat. The technique should provide a route to better understanding the operation of such ambipolar devices and. Lett. Another important role of the feedback inhibition was to reduce cell-to-cell variation in the response to BMP. Struct. the domain includes a winged-helix motif. An electron-hole density grating is imprinted in the quantum well by two interfering laser beams. presumably by stabilizing the ionic products through hydrogen bonding. the operation of spintronic devices based on the spin properties of the carriers. or friction. there is increasing interest in diverting some of the greenhouse gas toward use as a feedstock for the industrial preparation of commodity chemicals.1038/nsmb. that when trihexylamine is used as a base for ease of product isolation. 10. 12. environmental. 106. between the electrons and holes. H3K4 is methylated by the MLL family of histone methyl transferases (HMTs). however. the packets of charged pairs move in the direction of the holes. VOL 333 SCIENCE www.101 (2011). The catalytic subunits all share a SET domain. exhibit behavior opposite to that expected motion and have been interpreted in terms of a correlation field. however. But when a feedback inhibitor was removed. U. One promising reaction in this vein is hydrogenation to formic acid (HCOOH). this developmental regulatory system is optimized to promote a robust. PHYSICS The Pulling Power of Pairs The functionality and characteristics of many semiconductor devices depend crucially on the movement of electron-hole pairs in response to an applied electric field. Some experiments. — LBR Proc. Genes encoding components of the BMP signaling pathway are expressed together in groups that include both components that favor BMP signaling and also feedback inhibitors of the pathway. which comprises a zinc-finger motif and a carboxy-terminal SP1a and ryanodine receptor (SPRY) domain.2093 (2011). Yang et al. Paulsen et al. Sci. show that it binds preferentially to an active chromatin domain in a Homeobox gene locus and suggest that it may play a role in chromosome demarcation. use a transient grating spectroscopy technique to probe the electron and hole pair packets as they drift and diffuse in a single quantum-well structure. show that the Ash2L winged-helix domain is required for binding to the β-globin locus control region. 18. from the polarization dependence. Chen et al. Monitoring the diffraction of a probe beam through the grating as it evolves and decays provides a detailed picture of the electrons’ and holes’ dynamics. used a combination of experiments and mathematical modeling to explore what advantage this organization of the BMP signaling module might confer. — ISO Phys. but the entropic penalty for turning two gases into one liquid molecule pushes the overall equilibrium back toward the reactant side. CHEMISTRY Feedback Forms Frogs Signaling by gradients of the morphogen BMP (bone morphogenetic protein) controls the contributions of individual cells to the developing embryo. in Xenopus tadpoles. One core component is Ash2L. 108. Mol. Biol. Amines can deliver an enthalpic kick by deprotonating the acid. it is typically the heavier holes that dominate the transport properties.org on July 7.1002/ anie.

Princeton Univ. Univ. UK CB2 1LQ orders and renewals. Univ. UT Southwestern Medical Ctr at Dallas Jeffrey A. promote and defend the integrity of science and its use. of Oslo Mildred Cho. strengthen and diversify the science and technology workforce. Andrews Paul M. engineers. MIT Claude Desplan. DIRECTOR. SENIOR EDITORS/COMMENTARY Lisa D. Stella M. Melissa Raimondi. Univ. Granger. Yoshiko Takahashi. ADVERTISING SUPPORT MANAGER Karen Foote: 202-326-6740. Marseille Jonathan D. ASSISTANT MANAGER Lisa Stanford. Ray H. New York Univ. promote the responsible use of science in public policy. of Warwick Jane Parker. SENIOR EDITORS Caroline Ash. Nilsen. San Francisco Sue Wessler. FAX 202-289-7562 News: 202-326-6581. EUROPE Martin Enserink. EDITORIAL COORDINATORS Joi S.jp.org WORLDWIDE ASSOCIATE DIRECTOR OF SCIENCE CAREERS Tracy Holmes: +44 (0) 1223 326525. Sugden.aaas. Scott Miller. of Tokyo James Nelson. Moore. DIRECTOR. Apple Store www. SALES COORDINATOR Shirley Young AAAS B OARD OF D IRECTORS RETIRING PRESIDENT .www. Bank of America MasterCard 1-800-833-6262 priority code FAA3YU. Makiko Hara +81 (0) 3 6802 4616. Baylor College of Medicine Grahma Medley. EXECUTIVE ASSISTANT Alison Crawford. of Washington Michael B.com]. of Vienna Randy Seeley. Tokyo 162-0808. Hurtley. Jena Isaac Held. Pär Nordlund. Felaco.sciencemag. LaVonda Crawford. Ann Gibbons.htm?logo=17624. Alex Palmer. Inst. Erik Stokstad. Robert F. CA: 760-942-3252. of Florida Philippe Poulin. Its mission is to advance science. Irvine John H. The goals of the association are to: enhance communication among scientists. Suter. Univ. of Science and Technology Sean Munro. MIT Kathryn Anderson. of Cambridge Paul G. Univ. Crabtree. Jasny. Univ. Estación Biológica de Doñana. SENIOR EDITOR/COMMENTARY Julia Fahren- INFORMATION FOR AUTHORS See pages 784 and 785 of the 11 February 2011 issue or access www. Alice Whaley. Nara Inst. Trista Wagoner. University of Chicago BOARD OF REVIEWING EDITORS Adriano Aguzzi. Preston Huey. of Chicago Ed Wasserman. ADMINISTRATIVE SUPPORT Maryrose Madrid. Univ. PREFLIGHT DIRECTOR David M. Univ. Harvard Univ. Riverside Ian A.aaasmember. Rensselaer Polytechnic Inst. Mueller.com/aaas/ code MKB6. of Aquatic Science & Technology Ray Hilborn. of Washington Michael E. WEB AND NEW MEDIA Stewart Wills. Andersson. Univ. CHAIR Alice Huang. foster education in science and technology for everyone. Ian S. of Texas at Austin P. Rosser. Universität Freiburg William Kaelin. Harvard Univ. SALES ADMINISTRATOR: Marci Gallun. of Texas. Bargh. Princeton Univ. Lund Marcia C. of Oxford Denis Simon. Yale Univ. Univ. SENIOR MARKETING EXECUTIVE Jennifer Reeves. CONTRIBUTING CORRESPONDENTS Michael Balter (Paris). John Innes Centre Davor Solter. ASSOCIATE Elizabeth Sandler. Ludwig Maximilians Univ. Michael Hicks. H. Woolsey SENIOR EDITORIAL BOARD Cori Bargmann. Mitch Leslie. of North Carolina at Chapel Hill Robert Kingston. Andrew Oswald. BOARD Nancy Knowlton. Mary Ellen Crowley. of Oxford J. J. Cohen. SENIOR WEB EDITOR Tara S. Univ. ads@sciencemag. Univ. Rutgers Univ. David Grimm (Online). pbagla@vsnl. School of Med. Richard A.com/portalLogin. ASSOCIATE LETTERS EDITOR Jennifer Sills. of St Andrews Oscar Marin. of Warwick Yasushi Miyashita. Bernhard Keimer. Hering. of Chicago Taekjip Ha. P. WEST COAST/W. CANADA Lynne Stickrod: 415-931-9782. of Buenos Aires Leonid Kruglyak. of California. Lauren Kmec. Stuttgart Joel Kingsolver. Univ. AAAS Online Store www. PRODUCTION SPECIALIST Yuse Lajiminmuhip. and the public. Peter Andolfatto. FAX 760-9424979. EPFL Lausanne Karl-Heinz Glassmeier. Max-Planck Inst. COPY EDITORS Linda B. McCartney EDITORIAL DIRECTOR. Univ. Univ. MRC Lab.S.htm. of Pennsylvania Ottoline Leyser. Gregory C. Orkin.org). MARKETING ASSOCIATES Aimee Aponte. The Rockefeller Univ. Brad Wible. Lawrence Berkeley National Lab Mike Ryan. EAST COAST/INDUSTRY Elizabeth Early: 202-326-6578. FAX +44 (0) 1223 326532 CLASSIFIED (advertise@sciencecareers.. EDITORIAL FELLOW Melissa R. Phillip D. For Geophysics & Extraterrestrial Physics. EUROPE/ROW SALES: Susanne Kharraz. Butz. Univ. Alan Cowman. Sacha Vignieri. Andrew Lawler. College London 139 8 JULY 2011 VOL 333 SCIENCE www. ADMINISTRATIVE SUPPORT Scherraine Mack. Nayomi Kevitiyagala. Michigan State Univ. Chin. Christine Wehrli. Jerry Richardson. Trinity College.sciencemag. WEB DEVELOPMENT MANAGER Martyn Green. Brian White. Children’s Hospital. all articles published in Science—including editorials. MANAGER Marcus Spiegler. National Renewable Energy Lab. U. Bryan Ray. Timothy D. PRESIDENT . Jeffrey Mervis. of Cambridge Ove Hoegh-Guldberg. Kiberstis (Boston). of Penn. Potsdam Inst. Doney. Univ. Christine Ortiz. Univ. Univ. Max Planck Inst. FAX 415-5206940. Doudna. Janet G. Hospital. Peter Stern. Tokyo Inst. Beroza. Chong. April Marshall. NOAA James A. of North Carolina Tom Daniel. Lavine (Toronto). Jesse Smith. Stanford Univ. Stanford Univ. Romanowicz. Stanford Univ. Eli Kintisch.com] M I D W E S T Rick Bongiovanni: 330-405-7080. DC 20005 I NSTITUTIONAL S ITE L ICENSES please call 202-326-6755 for any questions or information REPRINTS: Author Inquiries 800-635-7181 Commercial Inquiries 803-359-4578 PERMISSIONS 202-326-7074. Univ. apple.com]. Jonathan Losos. FAX +81 (0) 3 6802 4615. Penn State Univ. Columbia Univ. of California. Himmel. Ernst Fehr. Polly Shulman. Tara Kelly ART DIRECTOR Yael Fitzpatrick. Peer Bork. Harvard Univ. CSIC & Univ. of California. science_editors@aaas. WEB DEVELOPER Daniel Berger. Harry Jach. careerads@sciencemag. Barbara R.org MANAGING EDITOR. Lisa Johnson. Turner. Laura M. Jennifer Couzin-Frankel.com bookstore www.O. Ordway. Shank. Spradling. FAX +81 (0) 3 5936 3531. Jake S. PRODUCTION SPECIALISTS Antoinette Hodal. Berkeley John Lis. Michael S. Harvard Medical School Vladimir Shalaev. Tompkins. EDITORIAL SUPPORT Samantha Hogg. Chris Redwood. Purdue Univ. Pacific Northwest: 503-963-1940 PRODUCTION DIRECTOR Wendy K. officedepot. Brauman. Johns Hopkins Univ. Science serves its readers as a forum for the presentation and discussion of important issues related to the advancement of science. RIGHTS AND PERMISSIONS: ADMINISTRATOR Emilie David.. DATA ENTRY SUPERVISOR Cynthia Johnson. Univ. of Cincinnati Christine Seidman. ETH Zürich Ira Tabas.com/eppstore/aaas. SENIOR TRAFFIC ASSOCIATE Christine Hall. ART ASSOCIATE Kay Engman. SALES ASSISTANT Lisa Patterson.org. Garvan Inst. ILLUSTRATOR Yana Hammond. Accordingly. Berkeley Jens Rostrup-Nielsen. Hubbell. SENIOR ART ASSOCIATES Holly Bishop.co. DuPont Lewis Wolpert. Inst. Kimberly Oster. Box 96178. Harvard Medical School Daniel Kahne. provide a voice for science on societal issues. Univ. Angela Creager. MARKETING MANAGERS Allison Pritchard. Univ. Yasmin Ogale. of Edinburgh Edvard Moser. Benkovic. Kahn. Chinese Acad. Karolinska Inst. David D. JAPAN ASCA Corporation. 82-88 Hills Road Cambridge. of California. Baughman. PUBLISHER RELATIONS MANAGER Catherine Holland.org). FAX +44 (0) 1223 326501 SUBSCRIPTION SERVICES For change of address. Ecole Normale Superieure de Lyon Ian Boyd. of Plant Breeding Research Donald R. UK/EUROPE/ASIA Roger Goncalves: TEL/FAX +41 43 243 1358. CONTRIBUTING EDITOR. and advance international cooperation in science. Yale Univ. Univ. of Tokyo Sonia Altizer. Sue V. Univ. Univ. Population Reference Bureau Gyorgy Buzsaki.org. Leshner PUBLISHER Beth Rosner FULFILLMENT SYSTEMS AND OPERATIONS (membership@aaas. LOCUM EDITOR Helen Pickersgill. Cambridge Univ. FAX +81 3 6802 4615. AAAS was founded in 1848 and incorporated in 1874. Mainz John A. CONTRIBUTING CORRESPONDENTS Jon Cohen (San Diego. CHIEF EXECUTIVE OFFICER Alan I. strengthen support for the science and technology enterprise. WEB AND NEW MEDIA Will Collins. SENIOR PRODUCTION SPECIALIST Christopher Coleman. Wardle. Valda Vinson. NEWS WRITERS Yudhijit Bhattacharjee. Dermitzakis. SENIOR PRODUCTION SPECIALIST / GRAPHIC DESIGNER Amy Hardcastle. of Pennsylvania David Lazer. Barbara P. Mitchell A. John Bohannon (Vienna). Stephen J. of Aberdeen Allan C. Dana-Farber Cancer Inst. Lazar. BUSINESS OPERATIONS Jessica Tierney. Vicki Linton. Stanford Medical School Marisa Bartolomei. Tomoko Furusawa. Ke Lu. of Zurich Tom Fenchel. David Bloom. of Cambridge Christian Büchel. Nicholas S. Beverly A.do. E. of Nijmegen Dennis Discher. inquiry@sciencemag. INTERNS Daniel Strain. of Technology David Hodell. of Copenhagen Alain Fischer. INTERNS Andrew Green. of St Andrews Anne Magurran. Jordi Bascompte. Institut Curie Angelika Amon. Ben Barres. Case Western Reserve Univ. Baylor College of Medicine Maria Zuber. for Climate Impact Research F. Anita Wynn. Paul. of Marine Sciences Ian Walmsley. Sara Reardon. Richard Shweder.jp. Gretchen Vogel (Berlin) LATIN AMERICA CONTRIBUTING CORRESPONDENT Antonio Regalado ASIA Japan Office: Asca Corporation. Harvard Medical School Douglas Wallace. Hao Xin [China: cindyhao@gmail. Cold Spring Harbor Lab. Swiss Fed. Charles C. Collège de France Emmanouil T. Martin Heimann. Johns Hopkins Bloomberg School of Public Health Elizabeth Grove. DC 20090-6178 or AAAS Member Services. NW. CSIC Facundo Batista. Carnegie Inst. DIRECTOR. Other Benefits: AAAS Member Services 202-326-6417 or www. Makiko Hara: +81 (0) 3 6802 4616. Mayana Zatz.uk) EDITORIAL: INTERNATIONAL MANAGING EDITOR Andrew M. Princeton Univ. Univ. Service (Pacific NW). COPY EDITOR Chris Filiatreau. University of Sao Paolo Jonathan Zehr. 77 Tenjin-cho. of Sciences Laura Machesky.com/affiliates/aaas. Chad Johnson. ELECTRONIC MEDIA: DIRECTOR Lizabeth Harman. John Travis. Brakefield. COMPUTER SPECIALISTS Walter Jones. Emily Guise. Univ. Leibniz Inst. Mara Hvistendahl [China: mara@ marahvistendahl. The Scripps Res. of Medical Research Elsbeth Stern. Berkeley Martin M. Carnegie Institution of Washington Jonathan Sprent. MARKETING ASSOCIATES Laura Tutino. FAX 202-682-0816 MEMBER BENEFITS AAAS/Barnes&Noble. Celeste Troxler. WORLDWIDE AD SALES Bill Moran COMMERCIAL EDITOR Sean Sanders: 202-326-6430 ASSISTANT COMMERCIAL EDITOR Tianna Hicklin 202-326-6463 BUSINESS OPERATIONS Wienhold. Hines. Petrini. Berkeley Julian Downward. of Cambridge Barbara A. Claverie. Steven Hahn. Univ. Janet Clements. Swedish Univ. Falkowski. CA). Bradford EDITOR-IN-CHIEF Bruce Alberts EXECUTIVE EDITOR NEWS EDITOR Colin Norman EXECUTIVE PUBLISHER Alan I. MANAGER. Univ. Miquel Salmeron. Gary Taubes. Norwegian Univ. Kerry Klein NEWS DEPUTY NEWS EDITORS Robert Coontz. William Jones SERVICE SUPERVISOR ADMINISTRATION DIRECTOR Deborah RiveraRandy Yi. Rubin. CONTRIBUTING EDITORS Elizabeth Culotta. of Washington Stanislas Dehaene. Max Planck Inst.org (for returning manuscript reviews) science_bookrevs@aaas.org (for book review queries) Published by the American Association for the Advancement of Science (AAAS). Daniel Clery. Julia M.com/domestic/index. of California. Eliot Marshall. ADVERTISING PRODUCTION OPERATIONS MANAGER Deborah Tompkins. Meinrat O. CUSTOMER M. Univ. FAX +81 (0) 3 6802 4615. Ocean Sciences Huda Zoghbi. Thomas A. Mats Carlsson. AAAS Travels: Betchart Expeditions 800-252-4910. Samantha Smith. Beverly Shields. Rustic Bldg. Sugden Andrew Editorial: 202-326-6550.: MIDWEST/WEST COAST/ SOUTH CENTRAL/CANADA Tina Burks: 202-326-6577. Univ. INSERM Wulfram Gerstner.geico. RESEARCH ASSOCIATE Corinna Cohn. and innovation throughout the world for the benefit of all people. NIH Philip Benfey. CONTRIBUTING CORRESPONDENTS Dennis Normile [Japan: +81 (0) 3 3391 0630. Cynthia Volkert. Max Planck Inst. Jocelyn Kaiser. Andreae. Chair. Univ. SPECIALISTS Jason Hillman. NW Washington. GEICO Auto Insurance www. Max Planck Inst. and book reviews—are signed and reflect the individual views of the authors and not official points of view adopted by AAAS or the institutions with which the authors are affiliated. Univ. CUSTOMER RELATIONS COORDINATOR David Lee. Hannon. ASSISTANT MANAGER Rebecca Doshi. Szuromi (Tennessee). Cynthia Howe. of Pennsylvania Scott C. for Cancer Research Andrew P. Calif. SENIOR SPECIALISTS Steve Forrester. including the presentation of minority or conflicting points of view.. Julianne Wielga. Barbara B. of Medical Biology. Alison Chandler. Univ. PUBLISHER RELATIONS. Radboud Univ. FAX 202-842-1065. ASIA NEWS EDITOR Richard Stone (Beijing: rstone@aaas. EPFL Lausanne Steven Jacobsen. Berkeley Reginald M. ADMINISTRATIVE SUPPORT John Cannell. CHINA/TAIWAN Ruolei Wu: +86 1367 1015 294 rwu@aaas. Univ. DC 20005 Bateman House. Cancer Research UK Bruce Dunn. Memorial Sloan-Kettering Cancer Center Siv G. Univ. SENIOR EDITORIAL COORDINATORS Carolyn Kyle. PRESIDENT Nina Fedoroff. Uppsala Univ. Indiana Univ. Jianguo Liu. Shinjuku-ku. Univ. Harvard Univ. VIP Moving Services www.org (for queries about letters) science_reviews@aaas. Burns. engineering. Richard Losick. Duke Univ. Univ. Imperial College Lora Hooper.. BOOK REVIEW EDITOR Sherman J. Mailing addresses: AAAS. FAX 202-371-9227 +44 (0) 1223 326500. L. Marathe. 1200 New York Avenue. Jelena Stajic. Tarrika Hill. Univ. SENIOR PRODUCTION SPECIALIST Ryan Atkins. PUBLISHER RELATIONS. Wilson. Univ. Hertz 800-654-2200 CDP#343457.K. Univ. of California. Paula A. Washington Univ. Fritz-Haber-Institut. Univ. Hospital Zürich Takuzo Aida. of Technology Tim Elston.org/about/authors kamp-Uppenbrink. MANAGER. Cornell Univ. Miguel Hernández Charles Marshall. Univ. David Clapham. news and comment. Timothy W. Carolyn Gramling. Jennifer A. EDITORIAL ASSISTANT Patricia M. TREASURER David E. Leslie Roberts. Joseph Silk.ELECT William Press. School of Med. CNRS.com/landingpage/go51. Medical Univ. Sam Kean. ASSOCIATE ART DIRECTOR Laura Creveling. new 1200 New York Avenue.org). Matzuk. European Research Advisory Board Luke O'Neill. PHOTO EDITOR Leslie Blizard EDITORIAL EDITORS Pat Butler. Virginia Lee. of Bremen Kei Hirose. ASSOCIATE EDITORS Kristen L.apisource. and payment questions: 866-434-AAAS (2227) or 202-326-6417. of Texas. Elowitz. of Virginia Jan Zaanen. CANADA Laurie Faraday: 508-747-9395. MARKETING DIRECTOR Ian King. EDITORIAL MANAGER Cara Tate. Subaru VIP Program 202-326-6417. Washington. of California. Tübingen Jonathan Weissman. M. London Research Inst. Wilson. Inder Verma. Daniel Ferber. Univ.sciencemag.org Published by AAAS . Wigginton. Los Angeles Kai Johnsson. Cold Spring Harbor Laboratory Press Publications www. Fleming Crim. Universitätsklinikum Hamburg-Eppendorf Joseph A. SPECIALISTS Latoya Casteel. Univ. Stephen Mayo. of North Carolina at Chapel Hill Gerhard Ertl. Univ. FULFILLMENT SYSTEMS BUSINESS SYSTEMS AND FINANCIAL ANALYSIS DIRECTOR P R O D U C T (science_advertising@aaas. Univ.com]. Monica M. CHINA / TAIWAN Ruolei Wu: +86 1367 1015 294 rwu@aaas. Adrian Cho. EASTERN REGION Phillip Smith. +81 3 6802 4616. of Wisconsin Yasmine Belkaid. Japan. Univ. rather than by publishing only material on which a consensus has been reached. PUBLICATIONS ASSISTANTS Ramatoulaye Diop. CNRS Colin Renfrew. Guy Riddihough. SENIOR COPY EDITORS Jeffrey E. CRUK Beatson Inst. Haldor Topsoe Edward M. Univ. of Wisconsin Jeff L. of Queensland David Holden. William P. Duke Univ. MacKenzie. Univ.org). Elizabeth Pennisi. missing issues. EPFL Lausanne Peter Jonas. ASSOCIATE EDITOR Maria Cruz. NIAID. MIT Elinor Ostrom. Olle Lindvall. Univ. Kerr. Univ. Univ. Seabury & Smith Life Insurance 800424-9883. Office Depot https://bsd. Melvin Gatling. Louise Hartwell. of Oxford David A. MIT BOOK REVIEW BOARD John Aldrich.org/bn. David Pearson. Austin Shimon Sakaguchi. dnormile@gol. Dallas David Baum. Stanford Univ. Yeston. Linn. SENIOR PUBLISHER RELATIONS SPECIALIST Kiki Forsythe. Purnell. of Science and Technology John Thomas. U. Raymond Orbach. Kyoto Univ. increase public engagement with science and technology. Natalie Villacorta. Rutgers Univ. Hendler. BUREAUS San Diego. Leshner. Katharine Sutliff. Los Angeles Christopher Dye. NEWS: DEPUTY NEWS EDITOR. SCIENCE INTERNATIONAL EUROPE (science@science-int. of Agric Sciences Detlef Weigel. Osborne. Univ. Wolfgang Cramer. AND MANAGER. Robert H. RESEARCH JOURNALS Katrina L. Zahn (San Diego). Pallava Bagla [South Asia: +91 (0) 11 2271 2896. JAPAN ASCA Corporation. Univ. Harvard Univ. Ap Dijksterhuis.html. Univ. Sabatini. Fred Hutchinson Cancer Research Center Gregory J. of Molecular Biology Naoto Nagaosa. SPECIALISTS Shirlene Hall. Walter & Eliza Hall Inst. of Georgia Sebastian Amigorena. Singapore John Speakman.org (for general editorial queries) science_letters@aaas. PROJECT MANAGER Trista Snyder. Dana-Farber Cancer Inst. Kai Zhang ADVERTISING DIRECTOR. FAX 617-507-8189. John I. EMBL Bernard Bourdon. BUSINESS ANALYSTS Priti Pamnani. SENIOR ILLUSTRATORS Chris Bickel. 7F. SENIOR Gilbert J. BUSINESS ANALYSIS Eric Knott. Nichele Johnston. Dublin Stuart H. Marc S. Memorial Sloan-Kettering Cancer Center Simon Phillpot. Pamela J. Inst. of Tokyo Richard Morris. Shaw. CUSTOMER RELATIONS MANAGER Iquo Edim. EXECUTIVE EDITORIAL ASSISTANT Yolanda O'Bannon (San Francisco). Univ. CORPORATE RELATIONS Eileen Bernadette Moran. Harvard Univ. Leiden Univ. Dangl. WHO David Ehrhardt. of Cambridge Trevor Robbins. Princeton Univ. Washington. Coo