The largest catalog of the best antibodies in the world

./ Primary antibodies

./ Secondary antibodies ./ Proteinslpeptides

./ Kits

./ Lysates ./ Slides ./ Custom antibody arrays

www.abcam.com

Other resources I Downloadable pathways and posters

We work hard to gather new pathways that complement our large range of antibodies. We have pathways and posters for each of the following research areas:

./ Cancer

./ Microbiology

./ Immunology ./ Stem Cells

www.abcam.com/poster

./ Neuroscience ./ Cardiovascular

./ Chromatin and Nuclear Signaling

Other resources I Protocol library

We have over 100 protocols for antibody-related applications as well as detailed troubleshooting tips to assist with your research .

./ Western blot

./ ELISA

./ Immunoprecipitation ./ ChiP www.abcam.com/protocol

./ Immunostaining ./ Dot blot

./ Flow CytometrylFACS ./ ELiSPOT

We have offices in the US, UK, Japan and China. abeam'

See website for local contact info: www.abcam.com/contactus ----...

R&D SYSTEMS 2010 CATALOG AVAILABLE NOW!

R&D Systems is a leading supplier of cell biology reagents with over 20 years of experience serving the research community. More than 97% of the products supplied by R&D Systems are developed and manufactured at our own faci lities. Th is enables us to mal ntain strict manufacturi ng and qual ity control standards, ensuring that our reagents meet the high levels of performance a nd consistency expected by 0 ur customers.

OFFERING MORE THAN 15,000 QUALITY PRODUCTS including:

Proteins Antibodies

ELiSAs Multiplex Immunoassays Stem Cell Products

& More!

Request a catalog online: www.RnDSystems.com/go/Catal og

R&D Systems Tools for Cell Biology Research™

USA & Canada R&D Syslems, In~. leI: (800) 343-7475 infu@RnDSy,lems.(Om Europe R&D Systems Europe, Ltd. Ie), +44 (0)1235529449 info@RnDSystems,co.uk

China R&D System, China CI>., lid. Tel: +86 (21 J 52380373 info@RnDSySlems(hina,com_cn

Selection expanding weekly-visit www.RnOS-ystems.wmigo/request to sign up for weekl:y new product updates.

GE Healthcare Life Sciences

Drop. Measure. Done.

NanoVue™ Plus: intelligent performance across all spectrophotometer applications

It's all the convenience you wont in a spectrophotometer, packaged in a portable, ergonomic device. NanoVue Plus features a new hydrophobic, gold-colored sample plate coati ng that del ivers outstandi ng results for sub-microliter amounts of proteins and nucleic acids. I t operates without a PC and does not requi re time-consum in 9 thirdparty path length recalibration

NanoVue Plus enables easy protocol selection using advanced software that includes intuitive drop-down lists for the full range of CyOye'" fluorescent dyes, as well as lists for common fluors. Results can exported using a USB cable or Bluetooth™ connections for print via computer (PVC) or stored using the new SO card option. An integrated printer is also available.

imagination at work

\~

• Swift. accurate analysis of 0.5 i-JI samples of nucleic acids and proteins

• Practical drop and measure mechanism

• Outstanding sample recovery

• Reliable and reproducible measurements

• Automatic self-calibration on start up

• Path length recalibration kit available as accessory

Experience the NanoVue Plus first hand. Register for a trial at:

www.gelifesciences.com/trY.Na noVuePI us

I AKTA I Amersham I Biacore I IN Cell Analysis I Whotmon I GE Service I

EDITORIAL

504 Science Ed ucation Web Sites Bruce Albert;

0» Science Prize Essay p. 538

NEWS OF THE WEEK

508 Bird-Dinosaur Link Firmed Up, and in B rilliant Iechnicclor »Report p. 571

509 Haiti's Quake Shifts Clinic's Focus From AIDS to Aid

510 Cli mate Sc1ence lea der

Rajend fa Pachauri Confronts t he Critics

511 NRC Urges U.s. to Rethink Sale of Helium Reserve

512 In Central California, Coho Sa lmon Are on the Brin k

513 From Science's Online Dally News Site

514 Test Sh ots Show Laser-Fusion Experiment Is on Ia rget

» Science Exprl'55 Report by

S. H. Glel1Zer et ilL

515 Did a Battering Rain of Comets Bring Ganymede to Geologic Life?

515 From the Science Policy Blog

NEWS FOCUS

516 Faintest Thrum Hera Ids Quantum Machines

>0> Science Podcast

519 !n the Deep Blue Sea

520 Little Castle on the Prairie

lETIERS

522 Tracking the Source of' Glacier Misinformation l. G. Cogley et at

A Role for Postdocs in

U nderqrad !lale Ed ecation 5 D. Bw;/} et a!

Taki ng a (lie from the Silver Screen Z. Fautkes

Give the "Fair Sex" a Fair Shake M. Nestle

BOOKS ET AL.

524 Reading in the Brain

S Dehael1l?, reviewed by c. 6 Gross

525 Hausch recken [Locusts]

5. Kaegi, reviewed by M. D. Laubichler andG: Honegger

52S Browsings

POUCYFORUM

page 516

527 The Politics of Geoengineering ]. I Blackstock: and]. C. S. L0l19 »s-Perspecttve p. 530

PERSPECTIVES

528 Arsonists in Rheumatoi d Arth ritis G. A. Zimmer mall and A. S. Weyrirh »Report p. 580

529 Bllndling with X·rays C. R. Safinya and r: Li »Report p. 555

530 A Test for Geoengin'eering?

A. Robock et at. »Policy Forum p. 527

532 Apes Among the TangLed Branches of Human Origins

I. Harrison

534 Measuring Subjective We!l·Being R. Loyard

»Report p. 576

535 Turninq Away from High Symmetry ]. C (rocker

»s-Repott p. 560

537 Retrospective:

Edwil1 G. Krebs (1918-2009) W. A. Catterall and]. D. Scott

page 525

CONTENTS continued»

COVER

Spore- bearing funqa I pa rasites emerge from the digested

co rpses of three bdelloid rotifers. The,e freshwateri nvertebrates (length <0.5 mi llimetersl prese nt an evolutionary puzzle beta use they have reprod uced without sex fo r millions of yea rs but have not been d riven extinct by relentlessly coevolvi fig parasites. Bdelloids can escape funga I parasites in spate and time

through complete desiccation and dispersal by wind to

u ni nleeted habitats. See pag e 574.

Image: Kent Loeffler, Kathie T. Hodge, Chris Wilsan

DEPARTMENTS

501 This Week in Science:

50S Editors' Choice

506 Science Staff

507 Random Samples

540 AAAS News & Notes

597 New Products

59B Science Careers

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

497

I CONTENTS

page 525

pages 534 & 576

pages 508 & 571

498

SCIENCE PRIZE ESSAY

538 Making Genetics Easy to Understand L. A. Stork and K. Pompei

»Editorial p. 504

REVIEW

542 Coexistence of Quiescenta nd Active Adult Stem Cells in Mammals

L li and H_ Clevets

BREVIA

546 Role of ABA and ABH in Desiccation Tolerance

A. Khandelwal et al.

Til e horrn 0 ne pati;way Illa t s tabili zes seed s may have se rve d rn ore primitive seed Less plants in su ppo rting de siccatie n 11111' r <In (e.

RESEARCH ARTICLE

547 local a nd Long· Range Red proca L ReguLation of cAMP and cGMP

in Axon/Dendrite Formation

M. Shelly et al.

Cyclic ade no 5i ne rno nopho sph ate a nd eye! ic gltanGsine monophusphate regulate axon formation and exert opposite actions on den dri Ie forma Ii 0 n,

REPORTS

552 Phase Transitions of Adsorbed Atoms on the Surface of a Ca rbon Na notube Z. Wong et al,

T h e a dso rption be havior of rare g<l5es is

10 ltowe d through eha ng es in reso nance frequency of a singLe-waiLed carbon nanctube.

555 Spontaneous and X-ray-Triggered CrystaLlization at Long Range in SeLf-AssembLing Filament Networks H. Cui et al,

Dilute solutions of alkyl'lerminated

pep Ii de fila m e nts ca n un de r!lo ord en ng upon X-I <ly expo su re,

»P~rsp~rJiv:e p. 529

560 The Free-Energy Landscape of Clusters of Attractive Hard Spheres

G. M~ng et al,

Entropic effects favor the formation 01 small dusters of coLloidal particles that have Lower sy mm e try.

»s-Perspecttve p. 535

564 A Tricyclic Aromatic Isomer of Hexasilabenzene

K. Abers_felder et al.

A structural isomer 01 benzene in which carbon is replaced by silicon exhibits

u nexp ected electro ni c stabilizatio n.

566 Combined Effects on SeLectivity

in Fe-Catalyzed Methylene Oxidation M. S. Chell and M. C. White

A" iron ca lalyst sh ows s ele etiv ity tor the oxidatiun 01 51' co ndary C-H bonds ill org<lrric molecules,

571 A Basal Alvarezsauroid Theropod from the Early Late Jurassic

of Xinjiang, China

j. N_ ChoilJiere et al,

The Alvarezsallroidea qroup extends the dade conl<li"il1g birds and their theropod relatives back to 1.60 million ye<lf5 ag().

»News story p. 508

574 Anciently Asexuql BdeUoid Rotiiers Escape Letha 1 Fungal Parasites

by Drying Up and BLowing Away

C. G. Wilson and P. W. Sherman

Ase~u<ll bd~lIoid rotifers avoid the evo lutiona ry parasil ~ trap thOlfg h

des icc a tion resi sta lie e_ and dis persal. »Science Podcas:

576 Objective Confirmation of Subjective Measures of Human Well-Being:

Evidence from the U.S.A. A.J_ O>wo/d and S. WII

5 U bj ective Ii Ie- satist a clio II S cores a gl e e with obi ective mea su re s ot well -bei ng across 50 American states. »Perspective p. 534

580 Platelets Amplify Inflammation

in Arthritis via Collagen-Dependent MicroparticLe Production

E. Boilard N al,

Micropartides released by platelets contribute I() ioflammeticn underlying rheumatoid arthritis.

»c-Persoective p: 528

584 DecorreLated NeuronaL Firing i n Cortical Microcircuits

A. S. Ecker et al.

Des pite de n se (0 n 11 ecti vi ty and sha red i np lit, the firing rail'S of nearby neurons are la.rgely u 11'(0 r relate d.

587 The Asynchronous State in Cortical Circuits A. Renart et al,

A gel1eral theoretical description of c orrela,li 011 sin hig h I y con" eele d

re (U rre n t Ile uron a I drcui Is.

590 Direct Restart 01 a Replication Fork Stalled by a Head-On RNA Polymerase R. t. Pomerantz and M. O'Donnell

Head-on ccllisions between DNA and RNA polymerases <Ire resolved in vitre.

593 p53 Controls Radiation-Induced Gastrointestinal Syndrome in Mice Independent of Apoptosis D. G. Kirsch et al,

Ionizing radtation destroys gastrointe~lina\ epithelial cells by a mechanism that appears to be independent of apoptosts,

29 JANUARY 2010 VO L 327 SCI ENCE www.sciencernaq.orq

SCIENCEONLINE

CONTENTS

SC IE NCEXPRESS

www.sciencexpress.org

N~Terminill Acetylation of Cellula r Proteins Creates Specific Degradation Signals

C~S Hwong et al.

111 vivll protein stllichiometries are f;egulaled through rHermil1a\Ly aeetylated d~gmI15.

10 .1126/sciel1(e.118314 7

Generating a Priori wit h Bacterially Exp ressed R ecom bin ant Pri 0 n Protei n f Wong et al.

Purified recombinant prion protein recapitulates the characteristics of tll~ i"llf~clious i1gent in

p rio n dis ease,

10.1126/scien(e.1183748

Contributions of Stratcspheric

Waler Vapor to Decadal Changes

in the Rale of Global. Warming

S. Solomon et at,

Deueases in stratospheric waler vapor after the year 2000 slowed full rille of increase in !jlobill5urface

Ie mp era lure.

10 .1126/science.1182488

Symmetric Inertial Confinement Fusion lm ploslens at Ult ra "High laser Energies 5. H. Glenzer et al,

10 .1126/scie()(e.1185634 »NeW5 !tl1ly p. 514

(harged·;f>a rtide f>robing of X·ray-Driven I ne rtia l-Fusi on Implosions

C. K: u et.al.

Laser-driven temperatures and implosion symmetry are dose to the requirements for in e mal- fllsi ()n ig ni tion,

10 .1l26/scien(e.118574 7

Food Security: The Cha llenge of Feeding 9 Billion People H. C. ]. Godfray et al.

10 .1126/scien(e.118S 383

SCiENCENOW

www.sciencenow.org

Highlights From Our Daily News Coverage

Coll.iding Particles Can Make Black Holes Simulations do not prove that the Lilrge Hildron Collider ~ wilt produce them, however.

~

~ Birdlike Dinosallr Was Adept Glider

~- Foam mode L mily dadfy orig i n of bi rd fLig ht.

=> '"

~ H ear That? Bats and Whales Share Sonar Protein

~ S ;mita r ge nef c chang es helped @:Gh!llnealion @voLve ~ in disparate species.

z

z

"'"

3

;;:

'" "'" i

~

:.

'"'

~

'" u

SCIENCESIGNAI..ING www.sciencesignaling .. org

The Signal Transduction Knowledge Environmem

RESEARCH ARTtCLE: Molecular Basis

of the Death-Associated Protei 11 Kinase--Calcillm/(a lmod uli n Reg ulator Comp lex 1_ de Diego et ~1.

The structure ot the: dea,ll1-assodated protein kinase and calmodulin comptex reveals how calmodulin binding leads to kinase i1divation.

RESEARCH ARTICLE: A Crucial Role for RACK! in the Regulation of Glucnse-Stimulated !REla Activation in Pancreatic l3'cells

l\ Qiu et al,

RACKl diclat .. s the- response of ttl .. i'nlrilc .. llular stress s en so r IRE 10. til di fferellt ext r ilc.elluta r 5ti mo li,

PERSpEalVE: Proteomics Modifies Our Understanding of Cell Cycle Complexity M. C. Hall

Tho LIS ands of m 0 din ell protei n 5 w .. re revealed in global cell cycle studies ot phospllorylation and Q·GlycNAcylatioll.

POOCAST

M. Lafon and A. M. VanHoQk

A si IIg le ami 11 0 acid 5 ubrti!uti en ina vi r al protei II is sufficient to i1ttenlla;te rabies virulence,

SCIENC ECAREERS

www .. sciencecareers.org/career ~magaz.ine Free Career Resources for Scientists

Perspective: From 'Pet' to peer S. Pfifl!!OIl et aL

Diverse prcbationary faculty members may be de-nietl a fair chance til become peers.

Mind Matters: Gossip inthe Lab I. S. Levine

Wllrkptace gossip may not be all bad, but i I nee d s to be handled with cal e.

Funding News D.Adams

Gel th e latest i11l1l0Un ceme lib oj' res earch fun di 11 g, Ie lIows hips, and interns hi ps.

SCIENCETRANSLATIONAL MEDICINE www.sciencetrilnslationillmedicine.org Integrating Medicine and Science

PIERSPECTIVE: Pathogen Microevolulion

in High Resolution

R_ 1(. kil and V. Niret

Ne-J[t-generatilln high-tl1f'llughput ~equendllg provi de-s insight i lit 0 Ih e 0 ri gi nand spread

of mkrcbial pilthllgen~.

SCIENCES!GNALING Structural insights into kinase activation by calmodul in,

COMMENtARY: Transforming Medicine via Digitallnnova!ion

E.]. Topol

Oi gital devi< es ha,ve ex~epti on a I p romi se for ei;anging the future I)f medicine.

RESEARCH ARTICLE: Resiliency and Vulnerability in the H ER2- H ER3 Tu morige nit Drive r

D. N. Amin et al,

The abi! i II' to over co me drug -resistant breast ca nc ers may lie in do~age frequency.

RESEARCH ARTICLE: Metabolomic Imaging for Human Prostate Cancer Detection

CoL. W!I et al,

M .. lab 0 I it jm a gi ng in til e pro stale reve.a ts can cercus reglons, guides biopsies, and helps proqnesis,

SCI EN CEPODCAST www.sdencemag.org/multimediil/podcast Free Weekly Show

Download the 2: 9 Jan uary S cience Podcilst to h .. ar about dealing with parasites in tne absence of sex, Ih e advent of q uant um rna chi nes, y!l II r (etten

to Science,and more.

SCI EN C EI NSIDE.R blogs.5ciencemag.org/sciel1ceinsider Science Policy News and Analysis

S(I ENa: {!SS"ttI; 'lJ1l~~u j 5-) 1:5: puMi !i~d weekly () n Fr[lh)lj ~(~t OU~ Lint wttk in, Dt{.errLb~r! ~y t:b~ Am~rfGiiJ" .A.!iS:Dr~ati.ofi fer the Al1va:lu~ment.of :Sdllllt'tj, nan N1!:w '!fork I!~ en ue, NW. W~~hitl gH~l!r [) c 2:(1 (100 5,. Perlcdlcels ~\ai I pcsteqe (Pil bli~.atil;i II Na. 4.a .... 1I6f1) ~id .at ~;5.hil1 !;Ita II, IX, and add itio IH! I IDoii I i r!g offices. tQpytight O~"Q 10 Eiv the Americall ~iMioii for the Ad\lali~~rn of Science, "fh~ rille SClt NeE is ill re:g i~:E:red lri!ldeinqJ~ 0 f the A_MS. IMm~rti'" i!1divid lIaIl membershi p end SIJb$triptiol'l {Sl i~$IIe:~): :$1.116 U1t1 i;lllo~·I«I to :5Ilb~i;.rlpfiOIl)_ i)rQme~ic il1stitlilioll.;.1 slIkcrip.lior! j,S.1 ~1Irt!s): -';'910: Foi'"0!)11 postagl2 ~rJ: Mc->:KO, C<1ri!:J.bc:.:J,1I (sllrf.x~ m.:Jil) :s.S,S; OIl1N eeu ntncs -(.ai r csstst d~liV(!ry) s:3S, F-·irst dess • .1irm.1i I, student, end emeritus rates 0 n request, G1lli:ld lan rates 'wil h G'Sf .1Vt.IilJbl12 upcn ANjU5f." GSf Lft2.SlI 8:812:~. PiJ bli(.}tioilS MJiI Agrffi"n~llt Numoor IM'9'62i1. l'1ilJ.tt~ in the U.S.A.

Qumg e 01 addrflH: Allow q ~ks. givillg old and Ilew ;lddr~s and .a-digit .1{(OUllt nu mlJ.N. P'Mtrnmer: ScJrd d!.al'grt! 0 f cdd ress to fJAAS. P'.O~ Box '9'-617.8, Wilstlil'glOI\ DC 200'9"(J~6173. "S ingl ~opy 1atul. :5010,00 aJ rrcnt issue, :S 15. ,00 bJc:k: iss lie prelMid iB:cllldes SlJrf,KI2 pcrSfJg~; bulk: rates 011 roqlJ~t. Au:t1:Ioir1rntioo to phot(j(opy fUlI!Irt'fit:lL ror irnel'll.!!i91 ·~~I1ii!IIJs:e under Ch1)l~a1~TIO"I falijr!g y,I"nhir! the f.!!iir lise prrlrjj~kior!s of 1f1~ COP'frig lit Ao; is 9 ranted ~ MAS to Iibri3rie:s end ether IlY'l"S ~~~ wilh the OJ p!}tig In Cl029fi,!1~ (e!1I:E:r (C(O f.a!l:5ai;:lioll.ai Repciortir!!;I Servict, PlD\'ide:d that 5-2.Q.OO ~t ~rride ls pCii.d d ir«tlV 1)] (.c(, ~l~ ~sewoiJ.d Dri~, ~mr~~ N!A Ql9U- 'the ideruiflmio"!"! ~ (or Sda:r.re is 00..36:·8.075, St".Ifr:!(1." is iildG.'OO ill the ReaM! Gurtro 10 PuirJd. irol UtefOWro and ill several spccialilfd illdct{!S.

"'AAAS

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

499

STUNNING Q_UALITY

Reagents for Sample Preparation from New England Biolabs

Introducing NEBNext"', a series of highly pure reagents that f.1cilitate sample preparation for downstream applications such as next generation sequencing and expression library construction. Available in sets, master mixes and modules, these robust reagents undergo stringent quality controls and functional validation, ensuring maximum yield, convenience and value.

Sequencing coverage map of the E. coli genome after using NEBNext''' DNA Sample Prep Reagent Set 1 for Sample Preparation

txlO'

2xl0'

3xlO'

4xl0'

Genome Pos il i on

E. coli SlraJIl MG 1655 gDNA was prepared with NEBNext DNA Sample Prep Reagent Sel I and sef]lJIJIICf!d on an lIIumina Gellome Amllyzer II.

For more information about NEBNext, including customized solutions, please OOlHaCl NEBNcxt@neb.com.

CLONING & MAPPING

DNA AMPlifICATION & PCR

RNA ANALYSIS

Now available:

NEBNext""' dsDNA Fragmentase'" an enzyme-based solution for the fragmentation of DNA

~\~G~

3;-'?> .• NEW ENGLAND

/80_ iolabs:

u YEARS _ /ffl<1b1ing r~lmQIQgiliS in rb. Itftscirnm

PROTEIN EXPRESSION & ANALYSIS

GENE EXPRESSION

, & CELLULAR ANALYSIS

In the (Stem Cell) Zone

Efforts to Isolate and characterize various types of stem tells have revealed much about the molecular mechanisms of stem cell self-renewa'[ and differentiation. as well as the function of the rnicroenvironrnent, or stem (ell niche. li and (Levers (p. 542) review what is known about stem cells in the hair follicle, bone marrow, and intestinal epithelium and suggest that two stem cell populations, one that is quiescent and one that is actively in the cell cycle, may coexist in these tissues in a so-called "zoned" stem cell model.

Phase Transitions on Carbon Nanotubes

The natu re of ph ase transition s cha nges with system dimensionality. Many aspects oftwodimensional systems have been explored by adsorbi ng ra re gases on graphite su rfaces, Wallget al. (p, 552) reduce the dimensionality f urth er by exami ning phase tran sitions of argo n and krypton on slngle·walled carbon nanotubes,

..l following the extent of surface coverage and

..

!;; mappin gout ph ase tra nsitions by usin g the

~ nanotube as a resonator. Changing the conduc!!: ta nee and th us th e d ensi!y of su fface electro ns i also allowed exploration of the effect of adsor.;;. bate-su rtace interactions.

t;;

:5

o

t Packinq Puzzle

1= .

g The packing 01 a large number of spheres is a

~ welL·studied problem with maximal packing

g_ based on the arran gement of nearest n eig h bors, ~ With much smaller numbers of particles, 1t is the 5 free en ergy that gave rns which perkin g arranqe-

rnents dominate. Meng. et 01. (p, 560; see the Perspective by Crocker) looked at th e assern bly of colloidal clusters where the number of parndes was limited from 2 to 10. For five particles or fewer, onLy one packing arrangement was found. For six or more particles, while a number of similar energ y structures (auld Iorrn, the proba bility of formation was biased toward th ose structures with the greater n umbe r of n ea rest-

n eig h bar conn ections.

Bi,g Finger

Alvarezsauroidea are an enigmatic group of theropods presumed to be closely related to birds, though most specimens are younger than Archaeopteryx. Choiniere et at. (p. 571; see news story by Stone) now describe a more complete early specimen, dating to about 160 mil" lion years ago, which supports the conclusion that Alva reZSi!uroidea a re a baSilL 9 roup of th e dad e containi ng both birds and their close theropod relatives. The fossil also he Ips to reveal the evolution of this group's peculiar forelimb, whkh includes one enla rged functional ling er.

Sex Avoidance Strategy

In m etaz 0 am, sex, as a means of reprod uction, is almost universal, with popular theories suggesting that the high rate of genome mixing that sex promotes helps protect us from deleterious mutations an dlor rapid ty evolving parasites. Asexual

a rga nisms a re both rare and often evolutionarily short-lived. If this rarity is d lie to failure to deal with pa rasites, then bdelloid rotifers, which have been a bstain'ing from sex for millions of years, must be able to dodge the parasite bullet in some

EDITED BY STEllA HURTLEY

other way. Wi lson and Sherman (p. 574; see the cover) suggest that bdeUoid rotifers do so by being both so resistant to d esictation and so

a rnena ble to dispersal on the air that their parasites can neither su rvive such punish ing conditio ns nor spread asia r nor as fast, a llow-

-] ing the asexual bdelloids to start afresh in

(pa rasite-free) pastures anew.

Aromatic Silicon

f Benzene has long intrigued chemists on

a ceo unt of th e energy stabi lizetio n, termed a roo

matielty,. w. hith arises fr.om wele(tr.o.n.d.eLoc. aliza-

non around Its rmg framework. A persistent

q uesti on has been how su ch stabilization would be impacted were the carbons to be

replaced by heavier atoms such as silicon. Abers felder et ai. (p, 564) have preps red a

benzene analog with 51 atoms in place of all six-rinq carbons, but a slightLy altered

bonding framework in whkh substituents outside the ring are no longer evenly distributed. ! nstead,

I he su bstituents

pair up at two 5i

sites, leaving two oth er ri ng sites with no external appendages. The resulting compound no longer has a continuo

ous network of n-electrons, but retains a degree of aromatic stabilization involving sigma and non bonding electrons.

.--

The Best Things in Life Are Free?

Does money buy happiness? Answers to this question differ, depen ding, in pa rt, on whether one asks a n economist or a psychologist. The former would point to correlations between higher incomes and greater self-reported well-beinq, whereas the latter wou ld arg lie that happiness shows little correlation with absolute material goods and is instead dictated largely by an indio vidual's so-called set-point, Another strand of research invokes a hedonic treadmill, whereby income matters until subsistence requirements are met, at which point comparisons with one's neigh, hers are what influence one's sense 01 life sarisfaction. Oswald and Wu (p. 576, published

Continued on page 503

www.sclencemag.org SCIENCE VOL 327 29 JANUARY 2010

501

1 2
25010Da
Is.oc..Oa
10M..,
T~
5l)1o(Ja
37ma
261D.J
20k0a
-- --
15l.[)')
1(1£1.11 We do proteins too!

In addition to our comprehensive range of over 45,000 primary antibodies, did you also know we are the place to shop for your protein and peptide needs?

Gamma H2AX phospho 8139 detection in C€!II hi stone p repa rat ions in a bsenC€! (1, 2) or presence of the phospho 8139 H2AX peptide ab15645 (3, 4), or in presenC€! of the non-phospho S 139 H2AX peptide (5, 6).

• Over 3500 full length proteins and enzymes

• 4300 immunogen peptides and g.rowing

• Both recombinant and purified native formats

• Western blotting positive controls, ELISA and functional assays

Abcam Inc. 1 Kendall Square, Ste 341 Cambridge, MA 02139-1517 USA

Tel: 1-617-225-2272 Toll free: 1-888-77~ABCAM Toll free Fax: 1-866-739-9884

abeam' www.abcam.com

----.

Continued from page 501

This Week in Sdence

online 17 December; see the Perspective by layard) establish that the subjective responses from 1 million adults, collected within health surveys conducted by the u.s. Centers for Disease Control and Prevention, do indeed correlate with objective measures of quality of life.

Platelet Microparticles Drive Inflammatory Arthritis

Platelets a re best known for th eir critical ro Ie in blood

dot formation during wound repair, but an appreciation for their role in inflammatory processes is growing. Platelet-derived cellular rnicropartides (MPs) are small membrane vesicles released from platelets in response to cell activation that can transport biomolecules throughout the body that have also been implicated in inflammalory processes. Boilard et ol: (p. 580; see the Perspective by Zimmerman and Weyrich) have now found that platelet-derived MPs probably contribute to the inflarnmato ry processes un derlym g rh eu ma tuid art h ritis, an autoimmune disease. The majority of MPs in synovial fluid fro m patients with va rio us types of infla rnrnatory a rth ritis

were platelet-derived and, importantly, platelet-derived MPs were lacking in synovial fluid from osteua rth ritrs patie nts. Fu rth errnore, platelet d eple tion a b rog ated disease developrne nt ina mouse model of inflammatory arthritis.

Columns, Connections, and Correlations

What is the nature of interactions between neurons in neural circuits? The prevalent hypothesis suggests th at dense local con nectivity ca uses nearby cortica L neu rons to receive su bstantial amounts of common input, which in turn leads to strong correlations between them. Now two studies challenge this view, which impacts our fundamental understanding of coding in the cortex. Ecker et at. (p, 584) investigated the statisti ts of correlated firing in pairs of neuro ns from area Vl of awake ma caque

men keys, In contrast to previous studies, (0 rrelation s turned out to be very low, irrespective of the stimu Ius bein g shown to the a nirnals, the distances of the recording sites, a nd the similarity of the neuron's receptive fields or response properties. In an accompanying modeling and recording paper, Renart et al. (p. 587) demon strate how it is possible to have zero noise correLation, even arno ng cells wlth common input.

When Polymerases Collide

Genomes must be rephcated by DNA polymerase and copied by RNA polymerase. The two processes occu r toncu rren tly in barte ria and, alth ough the majority of bacteria I genes are co-oriented with replication, head-on collisions still ottur and can result in double-strand breaks in the genome. Pomerantz and O'Dnnnell (p. 590) show, in vitro, that both Escherichia coli DNA and RNA polyrnerases g rind to a ha It when they collide bu mper to bum per but that th e replication fork remains intact, with the RNA polymerase being shunted off the DNA and out of the way. Mfd, a transcriptionrepair couplingfactor, promotes DNA replication restart probably by rewinding separated DNA stands ahead of the halted transcription complex, promoting its displacement from the DNA.

Gut Check

The gastrointestinal (G I) tract is particu larly sensitive to dama ge by ionizin g radiation. Despite

decad es of study, fu nda mental questio ns such as which cells and which rnolecu lar mechanisms medi· ate this G1 damage remain a source of great controversy. Studying a series of genetically manipulated mice, Kirsch et al. (p, 593, pub lish ed online 17 December) co nclude that GI epithelia I cells, rather

; t han end othelial cells, are the critira I cellu la r ta rgets of radiation damage an d that apoptosis (a wellg studied mechanism of cell death) is not a major contributor to the damage. Rather, an alternative

ffi cell-death pathway whose actTvity is in hi hited by the tu rnor suppressor protein p53 a pp ears to medi- 6 ate GI damage. Further insights into this pathway may assist the development of medical counter-

II>

5 measures for preventing an d treatin 9 radiatiu n-tnd uced tissu e damage.

www.sciencemag.org SCIENCE VOL 321 29 JANUARY 2010

Science Signaling

Science Signaling, from the publisher of Science, AAAS, features top-notch, peerreviewed, original research weekly. Su brnit your manuscripts in the following areas of cellu lar regulation:

• Biochemistry

• Bioinformatics

• Cell Biology

• Development

• Immunology

• Microbiology

• Molecular Biology

• Neuroscience

• Pharmacology

• Physiology and

Medicine

• Systems Biology

Subscribing to Sdence Signaling ensu res that you and your lab

h ave the latest cell signaling resources. For more information visit www.Scienc.eSignaling.org

Chief Scientific Editor Michael B, Yaffe, M.D., Ph.D.

Associate Professor, Department of Biology Massachusetts I nSI ltute of lech no logy

Editor

Nancy R. Gough, Ph.D.

AMS

Submit your research at: www.sciencesignaling.org/ about/ help / research .dtl

EDITORIAL

I

Science Education Web Sites

IN THIS ISSUE, WE ANNOUNCE THE FIRST OF 12 WINNERS OF A COMPETITION FOR WEB SITES,THAT BmeeAlbertsis Editor-

best promote SCl eace education. Each month this year, Science will pu blisb an essay by the in-thief of 5ciellce.,

creators of a winning Web site that describes their online resource This month '8 featured site

focuses on teaching and learning genetics, and it originates from the U uiversity ofUtab (see

p. 538). The Science Prize for Online Resources in Education (SPO RE) recognizes outstand-

ing fie ely available online materials that enrich science education .. There were nearly 100

entries fOT 2009 from many nations. They spanned diverse subjects, ranging from astronomy,

chemistry, and physics to geology and biology, Most sites targeted students, ranging from

elementary through graduate school, whereas others focused on the general public, Many

included videos, animations, real-world data sets, or teaching materials.

A panel of 16 scientists and nine teachers performed the challenging task of selecting the winners from the excellent entri es, In the end, two that were judged to be. of the very highest quality were nevertheless not chosen, The Physics Education Technology (PhET) Web site, created at.the University of'Colorado, Boulder, was considered ineligible because Science bad recently published an, Education Forum that describes how to use PbET's physics simulations," In fact, this article provided file inspiration for the SPORE contest .. An entry from the Howard Hughes Medical Instrtute (HHMI) WIIS also H,Ot selected.although it produces. an outstanding eduoation Web site (www; bhrni.org/biointeractive], HHMI is Scfence's partner in, producing the Education Forum, and we felt uncomfortable awarding them ODe of only 12 slots.

Why did Science create such a competition? There are many prizes for those who produce excellent scientific research, bnt only a few awards for educators. Yet being an outstanding science educator is as demanding and valuable to society as being an exceptional research scientist. And, as it does for research, highlighting education excellence sets a standard for others to aim at, while simultaneously emphasizing the enOlIDOIlS value of the endeavor. There is another important reason for the recognition that this competition brings. The World Wide Web is a fantastic information resource, but it can be overwhelming. Many had hoped, for example, that the US, National Science Digital Library Project might go OJ long way toward solving this problem. * >I< But the collection of science. education Web sites that resulted, although a valuable resource, contains so many entries that additional guidance seems warranted. With a limit of 12 Web sites a year, Science aims to make it easier to find valuable materials, both for one's intellectual growtli and for teaching.

This last point raises <1 broader issue .. When I began my academic career as all assistant profesSOl" at Princeton University in J 966, I so light to learn everything about what Otl1CI"S had discovered previously, before beginning my research all chromosome replication. Yet when I taught, I rarely sought to build on what otherteachers had developed before me. This difference. between how scientists approach theirresearch and their teaching goes a long \V'dY, T believe, to explain why the quality ofuniversity science education lags so fat behind the quality of science itself

Through the Web, a rapidly expanding OpenCollfseWare Consortium, with more than 150 universities from 36 nations, makes different approaches to teaching readily observable globally .. Based on tills wi de 'Vi sibility, many more contests can be developed to reward innovation in science education, Scientific societies might; for example, annually recognize the best I-mouth teaching modules for-au introductory science course in college, or provide an award for the best set of laboratory modules for a science class that ate lnquiry-based and require ouly modest resources (thereby being readily exportablej.The nomination process for Science's 201O SPORE contesthas just begun (WW'iIl.aa as. org/golspore}. According to Wikipedia, a "spore is a reproductive structure that is adapted for dispersal and surviving for extended periods of time in wJ..favOl1:lbJe conditions." Analogously, we hope 111M SP.ORE seeds the proliferation of many other education awards, adapted for dispersal and survival m the world ofeducation, - Bruce Atbens

lO.112619Cie!itl',1187267

504

'C E. Wiemafl rot CI/., Science 32:2, 6821200$). ~'J, Mervis, Science 323, 54 (2009).

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemaq.orq

CEll BIOLOGY

Meet U at the Terminal

Parkinson's disease is characterized by progressive neurodegeneration and has been linked to mutations in the gene parkin, which encodes a protein that recruits enzymes that catalyze the conjugation of ubiquitin to target su bstrates, Parkin itself contains a ubiquitin-like (U bl) domain, which is structurally similar to ubiquitin but is not rich in prolin e residues. The Src homology 3 (SH3) domain is found in hundreds of copies in the human protenrne, and it was originally discovered as a mediator of protetn-p rotein interactions, particu larly via its affinity for proline-rich regions. Irernpe et af. have found that the Ubl domain in parkin binds to the SH3 domain in endophihn A, which is involved in the retrieval of synaptic vesicles du ring neu ronal

~

W a ctivi ty. Ph os p h oryla tio n inc reased th e 'inter il cti 0 n

~ of parkin with endophilin A, and also stimu lated ,i the ubiquttination of a9 roup of protei ns in synap~ J- tic nerve terminals in wi'ld'-type mice, but not in

~ parkin-deficient animals. Thus, a regulated interI action between ubiquitination and endocytosis

~ may provide a due to the pathophysiology in

~ patients with Parkinson's disease. - H P

9 Mol. Celf 36, 1034 (2009).

E

D

§ MICROBIOLOGY o

~ Lhling Off Anesthetic

E.

g Toxicity concerns have largely ellminared ehtcro-

"'

5 form's erstwhile lise as an anesthetic" though the

MI(ROfll'IJlLGY

Working Backwards

EDITED BY GILBERT CHIN AND JAKE YESTON

What do you get when you throw anti-idiotypic antibodies into the Sargasso Sea? Well, probably not even a tiny splashbut conceptually what can be harvested is a bounty of previously undescribed a nd unsuspected virus genomes. When viruses infed plants, the plants tight back by generating an immune response that makes use of and is tailored to the specitic virus; small RNAs, roug hly 21 to 25 n udeotides in size, a re synthesized by the host a nd used to til [gel and destroy complementary viral RNAs. In developing a diagnostic method for two sweet potato viruses that (a use corky lesi ens in the roots, Kreuze et 0/. discovered that it was possible through deep sequencing technology to recover the vi ral genom 1"5 by assern bling the ove rla ppi n9 .sequen ces of the small RNA, produced by infected' pia nts, In addition to the expe cted culprits, they identified new members of the hadnavirus and mastrevirus genera, which eontain species that infed banana and sugarcane plants, respectively. WU et al: have analyzed libraries of small RNA sequences derived from the fruit fly and the mosquito, and they were also able, via deep sequencing and metagenomic analysis, 10 identify the presence of a number of new viruses, whose complete genomes could Ihen be reconstructed by PCR. - GjC

Vir%gf388, 1 (20,09); Prof. Norl. Acad. Sci. U.S.A 107, 10,1073fpoas.0911353107 (2010).

APPLIED PHYSICS

Plasmons on Separate Paths

In the quest tortaster signal processing speeds, the re are essentia Ily two approach es: sh rin k the size of the circu itry, 0 ruse ligh t r ath er

tha n ele ctro nie cu rrent as th e info rrnati on carrier. In the former approach, the opsratinq frequency faces a fundamental upper limit

of several tens of GHz due to the properties of the materials required in the desiq n of cu r re nt- based devices. Ligh t, in (on trast. offers the ultima Ie 5 peed, bu t req ui re scire uit size to remain 0 n the 0 rder of its ro Llg h ly m icro meter· scale wavelength. Surface plasrnons-elight-induced elertronic excitations

n ear I he su rface of a meta l/die lertric interface-offer the possibility of

ex p loitin g the speed and stze vi rtu es of the two a pp roach es, H o \iIIe v er, excitin g the plasmon s is u su ally a resona nt ph e-

nomencn, meaning that only one kind of

plasmon can be excited at 11 particular wavelength. Williams et al. present a technique based on a desig ned array of an n u lar holes on a textu red copper su rf ace tha tallows two tightly bound Ihz-induced plasmon modes to propagate independently, thus providing potential for a number of applications in chemical and biological sensing, security screening, and communications. - ISO

Appl. Phys. Lett. 96,011101 (2010},

com po und still plays a role in in dustrial chernistry, Low levels of ch loroform

(trichloromethane) inhibit the metabolisms of bacteria responsible for detoxifying many other ch 10 ri nated hyd rocarbons: th eretore, its persistence in the environment not only presents a possible human health risk but also slows down natural remediation processes. Now, however, Grostern et at. have discovered a popu lation of Dehalobacter that can not only tolerate a high level of ch 10 rotc rrn but actually

grow 011 it by coupling

derh lorination to respiration. Kin etics studies suqgest th at the ba rteria use a red u mve dehaloqenase

en zyme specific to chloro-

fo rrn an d the structurally similar trirh lorneth ane.

Dired reduction of chloroform ccu p led to bacterial growth 15 advantageous for remediation because

it 15 more efficient and does not requi re the addition of separate growth su bstrates. Conta minated aquifers-such as the one from which these bacteria were originally isolated-may be lerti Ie g rou nds for oth er bacteria with specialized adaptations to tolerate toxins, - NW

Environ. Microbiol. 12, 10.11111j.1462- 2920.2009 .021S0.x. (2010).

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

505

•

-e

1IiIWW. sci en (e!l1a 9.or9

lZQO New York Avenue, NW Wa.hingmn,DC zoom; Editon"l; 202-326.-6550, FA):; 202-28_907562 N~w" 202-326.6581, fAX l02-, 11-9Z27 Batem~n Hau""" 82-BB Hilh Road Cambridge, UK (B2 llQ

-1-<14 (0) un 3.26500, fAX +44 {OJ 12,23 326501

S'\JlI~rn Ifllll~1 SEA V'j' f~ For dlang ~ Q i add re.s.Si'~ m i Si5'ing 1 ~SlJ IZ SI n iliIW erde 1'0 'ond re 11 ewill;l:, and pilyment. que,tio fl,: 8~6-43~ -MAS (nn) or <,02-326-64 1;. rJl).:.z02·842·106S, Moi!ing addresses: MM., P.O, Box 96178" Wl,hingtlln, DC JOO_90-6178 er MAS M.flibd S."","" 1200 N~wYork M~nuQ, NW, Wil,hlngton, DC 2QOOS

I"'nl""o~" 5", LI«"'" please ~JU 102'· 326·6755 rcr ally ~ U.5~O ns or infolma I ion

R ... ,~'" AUlhor Inquirie$ AOO-63 ,-71B1

Citmnlerdallnqllilic, BOLI59·4S7{1

P"'''I'~''N' 20Z-S'26-707Q. FAX lDZ·682·pil16

M" .... B"",,,,, AAAS/!!am es &Noblo,colii bookstore ~'''';,"i1'''.<ligJbn; AAI'S Online )to'·~ ..-ww.aphOUrIQ.com/ailJI/ <ode MKB6; AM:'> Iravels: S.I.harl EApeditioo. 1l0()-'252-~910; ApfM Store " ... ~. iJppl e. r,om'cpp<lor""""ils;· funk of ArneriCd M,jst",C,rd 1-800·813·6262 priorily code fAl13¥U; (ola Spring Harbor labor~t,;ry Pre" Publ i<.IiPIlI www.t'hlpro.l',(OmWf1lf~t~ll'WI~.i1lm: GfilW Alito lnsurance w\'1V.geimcom~l"ridingp"9.lgo51.htm!logo-176.2~; Hem OOO·6S~ ·2200 CDP~3d3~5 7.: Olfice D~pot htlpl;l/bs.:1. oJfi«(>d .polrornlpor'l1lllogin .do; Smliury '& S rtrith We Ins Uta n(e 800- ~Z~-9883; Sub.',u VIP PiogrJm202·326.M17; VIP Moving 5 .. \'10(>, Ivww.vipmayIlO\'l<!r.com~dom.stidilide'Jitml; Dlil,r IlOn.~l5: AAAS MrI11~r Si'IVICes 202-326·6il17 01 \wNi.<lil"'!Mmbcr.or~.

~det1;:vdlmrl@"""""ol!J !lQr gene",1 edit~rial querill5)

,ci"n(2_I.lwrs@ail",.0I9 (lor qurui., "ooot li>ltcrs)

,ci"n(2_r""' ..... @aJJ'.org (lor r~tuming rTIiln!l5rnp' .w;"y,;)

,den(e_bookreVl@"".~",r9 (lor bOOK r<!'lie", que"",,)

Publiihed 11Y the Ameri(lln i\$OC;;'lioI1 forlhe iUt-fol\Celil~1 01 Sctonce (MAS), Stie{lte '.f'I~' it$ I.,ad." ii' " fOJlffli for th e p""'htation ;'md di,tJJsslon 'of imporl~nt i55U"~, .. Mod.to th. adv.n<omon( of -.;100<". indl1'liog the ~ro<enj;ltlon 01 ml"orlly 'Q' ,mnmnlng PQhlt< of vifW, ri1thef thall by publishing only m.Wi,,1 0!1 wl1ith a '0!1S<1!15U1 has b,"'n r~achNI. A({ordin.91y, all ,1rtill., publ[lhed In SdenC2-intlmling ~dilorm~, new; ~nd tollln1en~ "nd booR I'evi(!\'f.;------fl!~ si~ne<1 and reftw the il)tlividu~1 '<Iem of 11\e;l!J111Qf< ~!1d !101 ollie;;,1 PQinlS ~fViI?\'l odoptlld by AAM or Ill. fnsHIUlionl' \\ill" IIhi'" _tll~ allthor, ar.~ Jfliljated,

MAS ."", foun~"d in U~8 an~ iJICorpOliltod iii 1a7~. It> fliis5ion ,,10 "dv"n{e stion,e, "ngin •• rirl9, and inoo .... fian thrGughout·!h. \\'{)rid fm Ihe berlefit Qf .U ~opt. _ The gQDi:; QI lf1e .,<oei.tion • r~ ro~ enh~",<" (ommllllfcatio!1 Qn:JO!1(j 5(~n~ru, fllgln!'ffi,. ano !he publ):; promote ;wj del.n d th. rntcgnty 01 science and lIS Il>ll; SlI1lI1g lhen support tor th\\ !ciell(e and lecllt1OlO9Y ente!'prlle; prQV\de .~ voice for <Cic:!1ce 0!1 ",del'll 1"LJ05; p~olilote tile r""",nsible uso.ol 5d""e in pUlilic p<llic;y; ,nen,9lh"ll -. nd divor.lify the .... ne. ~nd IQrnnology ~"'rkfilra;; 1051e' cd UGltion in ,denr.:. and ,.dlnolog}' Jor every<llle: in<;rea>e. pub lic engagemenl with Klon", J!ld todlOOI"'9\'; Jlld adv"na, intematllnaL COOf>'1r.ltion fn 90011(0.

INFORMATION FOR AIFI~IlRS

See pages 352 and 3'!>:5 of the. 15" Jaouary 20:10 Issue or .access www.\Cien,emag.orgla.bQuvauthQfS

to!IQR!~ wltr lI.u(e Arb"efts

D:fW1M wnOR ~[m [Dim'

Mo n iea M. g.-aclfo rd Co I ill No rm.in

M~'~G'~" (Oro,. ~~,tll~CI)lGIJ~~~\~ Kat.;n,. L. Kelner

DEPUTY fD'TO~> R. B"ooh Han.on, Ba,bO'fa R. J~.ny, Andrl'.w M. Sugden

Eono","'_ 'I"'OR "-""0""<"",,",,,,",, U'" D_ (l1ong. 8md Wibl.: ""!O' '""''''' Gili>e~·l. Cmn. Porn!!l" I, Hih",. Paula A. ~ille!'Sli" (e",!onJ, Mar( >,_ ~vio~ (TOf(l!ltO), 8ev~rl f A. Pu rncll, L Bryan R;!y, GlIY Ridcljllough. H. lesw_ Sm I! h. ph il lip D_ S-,ur{lm i (T enn6~~eJ, \I~lda Vin~on. J" I<~ S, Y",[on; ..,"""" "'''''''' K~sten L Mu.ll{>~, )elen" SI"Iic. Nichola" S. WlgglOIO'1i.aura M'. lahO; m ... .,. """"-'1£ Alexi> WijnllC' Mo9U~~ uN< !UlnO'" gewflrt Wills; "'~.:5{)g.Al~ ClN:L:JP.\IE nmtJPB Rol.:J~rt Fr~den-c;~~ T'I'a S. MilT"! 11 e; "" ,0 '''~I o .... c ".. MM'!vn {i reen, And rew WhjlGell; '<><II< """ .... -..;n~ 5~.I1l1·a!1 J. 5" ter: A.,,,,,,,,, 'm,,,, ,onu_. leMit!)f 5.il!>; ,ono'""-\ .... ~Wl C"", Iule: ... , .. <o!')''''~'''·lef[r·l:\'' E. Uiok, Cymhl~ How~, Hnny 1(1(11. Sarb"'", P. Ondwil)', Tri,!il W~90Ii~r; coev mn tI es (I, n"$ filial rfJJ u, lJ lJ rcn Krn(lc~· EDllC!'RJAl l:(nH~rnft"lIlJi C,.mlyn Kyle, Se~"rly SmoliK; """u,.",, ...... "" .... ttR"rnaloul,,y" Diop, J.oi s. r;rHl1ger, l~ffroy He"r~, US" )QI1O>on, Scoli MiIL~r, )erry Rici1ilrdson, j e nnltcr A. 5eibert. Br'i,m WhitQ, Ai1"i1a Wynn; 'on,,",," .,,,,,,.,, Eflill\, Guise; Micha.l Hicks, Patricia M. MOOre, Miti~m Weinberg; U<E~lIH'II'rE "»I,5..UNT Sylvtli 5. Kih.r.r.a; ,.tl-M~HI::!il"A.lon'll'!E '""0"' M~ryrose Madrid; <rnlO."lt ,,,no,, Mdil5il ~. M ,CaI't 11ey N.~' "" ",. ~"" '""0"' Robert Coo n I;' ~Hot Mnr,ha II, j~1I ro t Me l'Ii" Le,lle Rnbem; <"'''" .. ''''"~ '" "0" Wu,i>e!h Cillo iM, PQlly .shulman; ~<V!. ""'!"' Yudhljlt Bh.n.(ha~~e, I\dnan ClIO. lellnifer (aUlin, DoVid Grimm, Cons!il nto Holden, )o(d\,n !Ill isec, Sam Kmo, J\idl~rd A. Km, EILKio!trll. Gl'1l» MUler, Eli<ab~!h_ F\l!1 !li5i, RQIkrt F. S"rWc, (p"eitlc NWJ. Erik Stakstad"jue Wany: ,,,,,,,,,,liluren "11~nkm.n: <P>Il.""" •• c __ ,Jon (OnOll (';;m DI~90' CA), Don",IF~~bor, Ann Gibooos. Robert KP<1I11g, IIndrelV lawler; M.itd, lA'slie, Cfl<lrt~~ C. Ma 11 n, Virgillio Mor~lt GofY Iauhe s: <0" '"'10"' lindi\ B, f~I.(O, Mflvln G:ilHtngj M!2'Li.s..~8 R.ijlmondi; ADj.ull'il~sl.R.At~VIt '~.ILH:lI'Olf1: 5c11Qrrai n c Mack; )lu".",So n DT~o. CA.: 760· g ~.1'3 2 S2, I'AA 760·942-<19r?; f';leitk North\'l6l: 'OH63-19~0

P'D.oman 01",,10' )ifme, liIndry: ••• " DO ... """ •• Wendy K,. S 11 '10k; Mo5t~lA[f1' IMNAr.a. Rc b(!t(() Do~bl; 'Satmlil' j,:r-E.ct~i ... ur~ s.tcv~ F orr~l-C:t, en ri~ R~dwood; .. ,! .. ~" Mttoony Rosen; ~"fJ-1G1<' "'"'o;IOR D.vi~ /II. Tn!"~~in~ ~ ...... MM,U,S:piegler, "'0"-,,,, ).WI1 HiUm~n

,A'Rl DIREi;;lo~ Yael KaB; "~'D~~"lE".AU LlIIiIEctg~ lallrl1'l tl'c'l;.l@lll'\l; lU~ID~ l,,-<m •• 1O", (oris Ii i<:~el. KotMarine Sut6ff: 1"""",,,0' Ya n~ (ire- h Inan: "'"'"' .orr ""'''''" '" Holly flilholl, Pr~,tOIl H l! e1, N~yomi Kev;; iy~g"La; NO ",CO)J"'" )l>SIica N ewli<lW, Maltllf\1t l'wombty: _'0 ","".lBio Bl!!:Jm

Sa,;;YlE 'l'l-l~LO~ONr'd

E.uFt~ff (Si(,jeo(~@.s(j~ntG-int.c.o.ukj Hlll1Jft.lAl! i'NlDnt1l!l~Cl1fIj,,, ~rulF«) 'EDh"'" A ndlt'w M. S "9d~n; ",,,"" , .. ,O!II",,,M_. Mia Fail rcn~a",p· Oppeoonnk; "H'o, <OllO'~ c",olloe /Il;h. Stella M_ Hurtle'j, '1M '>. Osoornel [letoRISt£rlt AS:5.Q[J.I!,1E Blll0R., Murid Quz; tOm"", Enll0R Hr,:l~n Pldlcrlgill: <D "0"", >~"." DoborJh Denn isiln. Rach.1 RobertI" Alit.(' What.y: ~~M'~"I'''I'''' ,"""on John <~on.U, Jonel q~ment', lQ~i'i" Moore; IN,EWS; UI~OPE 1'i~~.1 W,"OFt ].01"11"1 Tra'Ll'jsi DgJ'tlIn' ft EW~ I!DnClll Ofl;l"lid CI.~ry; '0."",1"'" '00.", 'O"n<N\' 'M iell ,1 C I Sa I t~, (p iHil 1, )oh 11 B"oho" non (Vi "1100), Mil !11 n Ensenok (Am<lerdom "nd P,jri:;l, G r~ldl1'n Vo<jel !Be~in)

LP..n~ A ri'I ERrClj -CONlf;!1 iI LmW,1j [Olitfl,£if.'(ji1(D~1 AntG t'l i;o Rcg(,llado

_A"A )opon otf1c,", A>ca Corpora~O!1, Tomol:io fUr1ll<M\l, RUltic BI(Ig, 7 r. 77 T enji n-oh 0, Sh inju ku.ku, To kyo 162-0808, Jil pan; + 81 , 6802 ~.616, FAX -h31 , 6802 ~ 615, jnqui!)'@ld~n'em~g._jI1; "'t~ ~'1'" ''''0' R !chald S tone (B~lji n9: r< lan~ @ila"'i_Qrg); ",,,,",, \IT'NO {o •• "'o.o ... ,, OMnl; Normile Oap"n: +111 (01 3 B91 06-30, FA)( +(H (tI) 1 59% 3S31, dhorlnile@gof_{0Il1J; Hao Xin [ChIM: +86 (01 to 6307 4~ 39 or ,,;,0'7 3671>, FAA +86 (0) 10 ~"o1 ~ 3'i~; dndyh"o@~"""lwmJ; Palkwo Bagl;r [Soul;' /Il;i": +91 (0) 11 2271 28%: P~I.@V5I1Lcoml

cxw III'! f ,~~\ I1I1 tR A~ n I. l,,~h,",r FIJ BU)II ,iI B~lh R!I;<n~.

!'u",,,eM';" 5""M' .. 0 0 .... ,,""; (m~mbNship@"a".I,org); m .. <10. WlIylon Butler, '5'f"to!l' !!i'l'!iH·&I;5 A.N'AlV!l1 NOIHun~ NYllil1.::1l1; W,5lQ.!,Ij!:P1' ".,," ,u_ .. "'" Po t Bun m; s ..... Au'" W toy" (.5t"<'I, L.Vond:. Crawfo!·d. Vicki lillIan, April MJr~hall; 0." ""'"" >11",.,,0. Cynl~ia Jol~ ~50n; '"UM,"' •. S!1iiI. n ~ H.!~ Iarri ~ Ifill, will i~~1 JO[l~1

SU5IN'E5:50 -01<'i~MiiJ"s il!HJ Ab-,ijlliN.tS.tUTfOB Dj~l[{lO'R Deboml, Rf"'l2'rtj~ Wlenhold;" oQiS!i.!ili11:rn DloR Ea(I:n;~ !I VSiIlJilCS. O-patA'T! j)~,~ Rli ndy Vi; !&AN,'r4! Il. ~!OiOUS: AGM VSoj5- E 1'1 c I< no It; M.I!.D!lAfiER.... !l!!1,lS.Of;~:!i OJi:QI.ill n ON s. J ess TC..a firm.y; " ... " .. " •• (,,'" Prill r"!)l~'an-j, (~Ifste Troxler; ~'"""" ",,0 RfRMf_5St(l~,~ ,ll.UHlLNI,5:tItA.'i c It Em iLk DaVid; M~a n IIIiE' [.ttlb h (]t h 5~ nt1[{! r; ~AAIfFfI"'G Q;1'RJ,;CI'Q.R tan i( ing; M,II!IIIKE'tHIIIG MAtlili!.-(l!::RS Allisrm P.ritwn I'd, A l ison

Chan cllor"juliJ!l ne Wi elg~; ""'",""IH 0<1",.,; Almer Ap OOle, 'M.!1 EH~n

Crowloy. W~hdy Wi,.; " •• , en ,'""""' )"nJiif.r R".v,,;

DnH.(lOR, SUE U~IEI'I!!~Hfllii Tom Rv~n; DHt~ClO~,t {O~P{):II!MIi' RiElAl'l-O"S Eil(l~n Ilc:rntldQu~ MOfilll: i'UJlU:iIiEIiI: ih.EL\lU;lN'~, ~~'[JUit(E~ '5!i',ff!~us't Klki Fo~the; ~'ENlalt rUBuSflElt Qfi.A1'1'IlN5 :5'PE';;;I~~ Clltlierine t-iollnnd~ PIHll_dSHl'a ~.a4'JP!J r.:!i!1 !;A5Y tDIB,;!>T P.11illlp Smi th; PH,ra.u !iiHi;J!, ~E~Al1QI't~; WOl ((jUl Pt1illp TSQlakld.l!i; Fm!FJI:.l~!'" S-LU!:tiI:·vu,ml: '.quo Ediin~ F'UlIf'H.W!;Nl ",'''''''.''00· C. rrie M"cDonilld; MAPJ!"'~" ~~ •• _ (~ri'~n" SdM{I1l; ~.AlUI;P1NI!I .A!j:~O(r"H Mary lfl90iloui; EU,Cl,~I'J!'IIIC M-ED'A~ ,,",ANliitiEf;I ~iiabelh ~~rn1"n.; ""OJ!'" " ... AG·'" TMj. Sn~d~r, "l"<fAJl"r ""'~'rn lllil Sl<lo'lto~d;m<'o" """""cr,,,~ s .. """,,. Ry:.n Atkin" €hli't<J~hor CoI..","n, waLI~r JGOC's: pooo"" 'ION ,e",...,m- Nldl~IQ )011 nstcn, Ki mberl'y D,ter; [)iI~H1Q:Ilr W.ElI ~D NEW i!,IiEIlIA Will Collin5

Ai}lIfi"tl511!1.1i D!1iIf ClO~ WOliH.DWU'f SlID tAla; Bill Mr:lQn ,IlMM""", "JllO~Sc.~ '>lnde,,: 20Z·32o·6'!30 PROj i:d nJ~ECit-o~. oiftitEAtii Brianna Bld.s~r

l>.o.V<1 l>ci~",:f~~dveni~ing@""ill.org); ~''''' • .--rl'', "">D" Rick, Bongipvannl: 330"~ 05·7080, FAX 330"~ 05·70&1: EMl <0.;,1 •• , ... "" laurIUil'.O"Y: ~OS·7q7·9J95. FAX 617·507"8189; _, I~.o,s' l~nno St i(k roo: 415-931"9182, fAX ~ lS·S 20"6940; "",OJ .0 ... 1 .... Roge r Goncillve,: lH/fAX .... 11 13 2~ 3 BSa: L ..... A.5CA Corporation, Nan~ko Ide +81 (0) 3 630, Q6).6, FA~ +'31 (Dl3 680.2 ~ 6 ,~;ads@ sctenCllm"9.jR; "'.,"" ".m< .,,"""" Pei~ndl'i1 S4l11nl,

rtrolRlO,WlI;l:E MSOa,Al:E !:]j~EtlI:IR"{If ,5"aENa" 'U~iSllS rl'~cy HOIIll~$': ~i1 (0,) in 3 ;l26".2~, F/lIt +~A (01 nZl ~26 532

CW51frED (.adv-c-rli:S-e@5d@n{;I2{~1~8-er:s..org).,.-1J"s.: :5.MB IIot.IUf!.GElt Ddryl A!1{ler;o!1: 202-326"oS4O: M'"_' TiM Bllrk;~20l-326-6577; ,." to,,, Alc~js FlQ!nin,g: 202-326-6'>78; <I ... ,/,a"'" , .. .-r .. , Nt(I~oliJ_I Hiobbi;j,e: 202- 326,6503; M'" <O","'~.'o." Rohan ~dmon.on, S h i r Ley Young; ,,, .. Susanne Klwtilt, DM Pennington, AI~x Palmer; ">"'> A>;LS1I\N' [jsa P~I!er~n: .. "'~ MCA Corporation, Jie Chin +81 (0) 3 6802 ~616, JAX +81(0) 3 6802 ~615; (ArQ,""dl@;clen(onia~.jp; ."' .... I~.' .. "".-O'f .• ,,,,.,, .. " K.ren F<lotc: 202-326-6740; .0 ... "~nG P1!!OOUtT[ClN; j):PE~~l~PilS. MII,JU,1>ER O~bGr2lh romplrln:i~ !iEN~OR Pf!OOULHOM 5PEW!'U:51/G~ffil-[ DBI6N<l:fI ArJlY H~lrdGtl~tle-; 5~10lit ffiODUrr!Ofi sf'ruAIJ'n RDbert B\!dt "_'"' ""m< ''''0(''''' ChM!~nc HiLll

AAA5 .kAFlD' af" DillE{l'OR,i If~lfnN!ii f'RE.'iiOEffi, tH.PtHIl ]otllne5" J~ Mt(lIrtlW; ~.~'(]~1 PflJr C. ·A.gre: p.p:tEsfD£HT""ElEn Alke HUCUlg; T'p;fA:.51.j)!:£R D~1IkI E. Shdw; '"'" ",,"'IV. ",,"". AI.)" r. l.!'5hner; 00"'0 AU(~ Gfi,t, Lind" P. B_ Kate hi, Nall(y Knowllo n, (Il~ny A, M l! rray, J uti" M. Phillip" ThOlll"5 D. Pallar{J, David ~ So.bil!lni. ThOOIO, A Wools~V

I.AAAS

SENIOR EDITORIAL SOARD 10M I. 8r'l!Jl1.'n. Oirrir, ~r1m1Q,J Ul,,_ Ridlatd I;osidif. f/Qilltm1 Until.

L1rK1ii P_d~, UtIiv. Co/IrgI!lM""n Midlael S-. n:&nll!!rj' Vl1r~firi td (hir:rJ!Ji'

506

29 JANUARY 20m VOl327 SCIENCE www.sciencernag.org

Springs are a major source of water in Himal.3yan villages, But spring water has become starter, even drying up in summers because of deforestation and chang,es in rainfall, Harve~ting rainwater can help, but convoluted mountain topography often hides the recharge areas Where preopitation goes u nderg tnu nd to feed the spnngs below,

To find those areas, scientists ate turning to radioisotopes, A team headed by physicist K, Shivanna of til€. Mumbaj~based Bhabha Atomic Rese.;lfch Centre (BARe) took: samples of rain~ water a nd spring water from va ri ous lorati ons, Theisotopic com position of rainwater va ries by season, latitude, altitude, and distance from the coast; the differenceS leave their mark on water flaWing fra m spri nqs,

In a 2-year pilot project, the researchers and a local group, the Himalayan Environmental Studies and Conservatien Organization (HESCO), compared water samples in the Alaknanda River Basin in the Himalayan state of Uttarakhand in India, From isotope signatures of the rainwaterin the upper reaches and the water in the various sprinqs, the BARe teamidentifred possible reeha 'gf! zones for til e sprinqs in the th ree valleys, "This made it pessi ble. to accu rately locate darns, trenches, and underg round dikes to give rainwater time to seep into the sprin g5 below," says HESCO fa under Ani I leshi. As a result, the discharge rate of existing springs almost dou bled and two new springs appeared. Shivanna says there are now plans for similar projects in seven other Himalayan regions.

g Hormesis Who?

z o

~ Many biologists have noticed that when they

5: feed a plant or a njrnal a tiny a rnou nt of a poi-

z

o son, the resu It 15 the op poslte of what they see

S_ .

at higher doses-the plant grows faster, for

~ example, or the mouse develops fewer tumors * than it normally would, That "biphasic" phsg nomenon is known, am.ong other terms, as

9

~ > o

~ C 2000

~ ,g

~ 6 1600

z x

~

o i:o

"' o ,...

g

t:o ~

ce <.!

The co-founder of the company that Introduced the popu tar Baby Ei nstem videos has sued to

o blain the records of researche rs whose work cast dou bt 0 n th e value of the series,

h ormesis, Mainstrea m scientists have mostly ignored such observations (Sdence, 17 October 2003, p, 376), But lately, hormesis has been

gai ning grou nd, especially in Asia, A search of the Web of Science database fi nds that citation s of papers usi,ng the term "horrnesis" or "horrnetic" have soared from about 200 in 1999 to more than 2400 in 2009. Toxicologist Edward Cala brese of th e University of Massach usstts,

Am herst, who did th e search, predicts that citations "will profoundly grow" as people realize the i mplicatio ns of horrnesis in areas SUGh as finding new anticancer dru g5,

2400

Baby Einstein Goes to (OU rt

'0 1200
~
-0 800
E
'"
z 400 EDITED sv CONSTANCE HOLDEN

William Clark, who founded The

Ba by Einstein Co, with his wife in 1996, su ed the University of Washington this month for access to materials held by UW pediatric researchers Dimitri Christakts and Frederick Zimmerman, The pair wrote two widely publicized studies in 2004 and 2007: One suqqested that TV'immersed toddlers may be at risk for later atte ntion problems; th e oth er lin ked watchi.ng baby videos with delays in language development (Science, 24August 2007, p. 1015),

"Given that oth er research studies have not shown the same outcomes, we would like the raw data and analytical methods from the Washi ngto n studies so we can aud it th eir methodology," Clark said in an 11 January statement. The Clarks sold their business to Disney in 2001,

Both papers were influential in deflating educationa I claims used to market baby videos, says Josh Golin of the Cam paig n for a Com rnercialFree Chlldhood, which in 2006 complained to the Federal Trade Cornrrussion about Ba,by Einstein, Disney has since toned down its claims, And last September, facing the threat of a class-action suit, the company announced that it would give refunds to dIssatisfied parents, "It's a brand trying 1.0 save itself," Galin says,

OLd Man of the Sea

Oceanographer Walter Munk, 92, has won the 2010 Crafoord Prize for planet-spanninq work explaining everything from why Earth wobbles to how the wind moves ocean waters, The prize, with a $555,000 award, comes 10 years after he won the Kyoto Prlze, the other major honor billed as a No bel equivalent. "I'm so delighted about both, " he says, "I thought I was beyond that"

Not rna ny scientists have authored papers "still cited 70 years later," says fellow physical oceanograph er Carl Wu nsch of the Massachusetts Institute of Technology in Cambridge, "He has a knack of getting to the hea rt of en e subject after another." Mu nk, who still d rives to his offfce al the Scri pps In sntution of Oceanography in San Diego, California, almost daily, is now working on a neglected sort of wave that he thinks is vital 10 d riVing ocean currents, Then there's a sort of 3D x-ray of the ocean he wa nts 10 appty to addressing sea-level rise, But thai, he says, "is a hope for the future,"

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

507

PALEONTOLOGY

Bird-Dinosaur Link Firmed Up, And in Brilliant Technicolor

B EI)IN G- Which came first, the chicken or the dina sam egg? That one's a cinch. Less obvious is the riddle of kinship. Most scientists think birds evolved from dinosaurs about. 150 million years ago. But a sparse fossil record has providedammunition to those who insist that birds arose independently. A stunning new fossilmakes that idea virtually untenable. And a second paper this week brings dinosaur feathers vividly to life, offering new clues to why this instrument of flight first evolved in. flightI ess creatures.

On page 571, fossil-hunter extraordinaire Xu Xing of the Institute of Vertebrate Paleontologyand Paleoanthropology (IVPP) here and colleagues unveil Haplocheirus sellers, a new genus of alvarezsauroid-a group of dinosaurs once thought to be flightless birds. The nearly complete skeleton, unearthed from I60-million-yeaJ-old mudstone deposits in northwestern China's Junggar Basin, extends the fossil record of alvarezsauroids back in time by a whopping 63 million yearsmaking it about 15 million years older than the earl jest known bi rd, Archaeopteryx.

In 2006, specirnens from another theropod g:roup-tyrannosauroids-challenged the SO~ called temporal paradox: the fact that

508

irrefutable birdlike dinosaurs appear millions of yea:rs after Archaeopteryx in the fossil record. (Among other birdlike features, tyrannesauroids SPOlt three-toed teet, hollow bones, and even wishbones.) Along with those and other recent (rods, Haplocheirus "establishes once and for all that there is no temporal paradox," asserts Hans-Dieter Sues, curator of vertebrate paleontology at the National Museum ofNatural History in Washington, D.c.

The newest, oldest alvarezsauroid dispels some of the group's mystique. "Alvarezsauroids were: collected for the first time in the 1920$ but were so enigmatic that they were never recognized and described until the 1990s;' says Philip Currie,

a paleontologist at the University of Alberta in Edmonton, Canada. As specimens turned lip in Asia, Europe, and N01th America, debate raged over whether they represented dinosaurs or flightless birds. Haplocheirus "definitely settles this dispute," Sues says. The exquisitely preserved specimen "is so primitive," says JVPP's ZhOH Zhonghe, that it lacks some birdlike features seen in later alvarezsauroids such as fused wrist bones, a backward-facing pubis, or a crest near tile top ofthe shinbone.

Birds of a feather. A new avia nlike alva rezsau roid (below) predates the fi rst bi rd, Arc{Jaeopteryxpowerfu I evidence that bi rds arose fro m

di nosa u rs. Fossi uzed

pig me nt organelles reveal the true colors of Sinosatuooteryx (letO.

The controversy highlights how hard it can be to distinguish birds from dinosaurs, "Deep in evolutionary history, it's extremely difficult to draw the line," XlI says. The discovery of feathered dinosaurs in the late 1990s turned classification upside down.

Why feathers evolved in dinosaurs is a puzzle. One idea is that feathers provided insulation .. Another is that they evolved as ! camouflage or to attract a mate. But evidence -0: such as distinctive. colors 01" patterns had ~ eluded researchers=-untilnow This week in ~

:l:

Nature, paleontologist Mike Benton of the i§.

University of Bristol in the United Kingdom ~f

.. =

and co lleagues offer tile first report 0 f ~

organelles bearing the pigment melanin in. ~ dinosaurs, They found melanosomes in the ~ theropods Sinornithosaurus and Sino- ~

z

sauropteryx and in a bird, Confuciusomis, that ~

lived roughly 125 million years ago. ~ Although many scientists expected ~. dinosaur feathers to contain melanosornes, ~ finding them wasn't easy. In Benton's lab in ~

z

Bristol, rvPP's Zhang Fucheng says be "spent :'!i

day and night for more than a year" studying ~

"-

scanning electron microscope images of scores ~

of fossil specimens. Zhang's diligence paid ~ off-he uncovered melanosomes containing ~ the two most common kinds of the pigment: 9

z enmelanin, which gives a. gray or black tinge, 2

and pheomelanin, which gives a chestnut to ~

oJ

reddish-brown color, Tile find could settle §

z

a debate over whether bristles studding g

Sinosauropteryx were primitive feathers ~ 01" collagen fibers. "Melanosomes prove ~ that those structures are indeed feathers," ~ ~

~

Benton asserts.

The discovery also reveals the true :;(

colors of dinosaurs .. Smosauropteryx's !oj

protofeathers ran in alternating orange ~ and white rings down its tail like a ~ barbershop pole. To Benton, the ~. vibrant pattern suggests that "feath- ~

D

ers first arose as agents for color dis- ~

play and only later in their evolution- :;; e

ary history did they become useful ~

for flight and insulation." Others '" :;;

are not so SllIC. ZhOH, a co-author ~

u

of the Nature paper, still thinks i'.i

feathers evolved fOT insulation. ~ To Xu, the mystery has deep- ~ ened, "Now we are realizing that f feather evolution is more complex cl li"i

than we thought," he savs, t:;

.... 'J~ ~

-RICHARD STONE u

29 JA NUA RY 2010 va L 327 SCI EN CE www.sciencemag.org

GLOBAL HEALTH

Haiti's Quake Shifts Clinic's Focus From AIDS to Aid

On one and a quarter hectares in downtown Port-au-Prince, the AIDS clin:ic called GHESKlO bas developed a reputation over the past 25 years as a place that, improbably, providesthousands of Haitians with free care and conducts world-class clinical research, Led by Jean "BiJJ" Pape, GHESKIO (the Haitian Group for the Study of Kaposi's Sarcoma and Opportunistic Infections) has continued to function I

since the 12 January earthquake. The clinic has now become one of me few places where locals can receive emergency care in that part of the devastated city and has converted itself into a 110s- Helping hands. GHESKIO head Pape conpita] with a surgica 1 unit .and a verted his downtown Pnrt-au-Prince AI DS makeshift outdoor home for thousands eli ni c into a rlffug ee ca m p and emerq ency

of refugees, hospital co mplete with a su rgical unit.

GHESKIO collaborates with the

Center for Global Health at Weill Cornell Medical College in New York City, and together they receive about $3 million a year from the U.S. National Institutes of Health to conduct HN / AIDS clinical research and training. All of that has now been put on indefinite hold, as Pape and his sta ff of more than 3 00 people WOrk to save the lives of those who have nowhere else to tum-while they still keep delivering HTV/AIDS and tuberculosis testing, couuseling , and treatment to more than I. 00,000 others who each year regularly visit the clinic and its newer satellite near the airport. In many ways, GHESl(JO is Port-au-Prince's equivalent of Partners in Health, the nonprofit program run by Harvard Medica I School's Paul

Q Farmer that serves rural Haiti.

~ Pape was in a meeting with the prime minsister, minister 0 fhealth, and officials from the ~ World Health Organization and other inter~ national groups when the magnitude-7.0 ~ earthquake bit~ All escaped from the room, ~ which collapsed, relatively unscathed, On ~ 2 I J allll~y, he respon~ed to questions from ~ Science m an e-mail; his responses have been ~ edited and condensed.

OJ

~

-c

"

'"'

~ Q: What is the status of your staff?

§ J.P.: The earthquake. affected us a lot We. have ~ documented the loss of 4 staff members

Reliet] personnel were delivering ARV drugs even during the aftershocks,

-]ONCOHEN

including 1 MD and the head of the microbiology lab,and 12 injured staff Twenty-one people have not reported to work.

Q: What damage did your dimes suffer?

J.P.; Three important buildings are severely damaged, two at one site and one at another. The other buildings can be repaired. [In an earI ier e-mail to colleagues at Cornell, Pape noted that GHESKIO's tuberculosis lab had to be

sealed until it is decontaminated, as tubes containing multidrug-resistant TB ba cteria and other strains had broken.]

Q: What stockpile do you have of antiretroviral drugs [ARVs] and are you in danger of running out?

1. P.: Our AIDS care program is going well. We activated am contingency plan the morning after the earthquake, This was easy as we have been used to uns table political. situations and hurricanes. Basically, all patients have an extra supply of 2 weeks of important drugs (ARVsandTB).

Q: Are you able to distribute ARVs tathe people most in need?

J.P.: We are now seeing about 80% of all patients we were seeing before. PEPF.A.R [the U.S, President's Emergency Plan for AIDS

Q: What assistance are you receiving? J.P.: We have the full back-up of both [J,S, Ambassador Eric Goosby, who even volunteered to come and help [IS, as well as Michel Kazatchkiue executive director of the Global Pund and his entire staff. They have found another way to help us through COPRESIDA. theagency in charge of the AIDS program in the Dominican Republic, This catastrophe has revealed that the DR is a true friend of Haiti. We are grateful for the world response and particularly that of ORand the persona! involvement of'President Leonel Iemandez,

Q.: 'How do you restart the clinical trials and training you had under way?

J.P.: We have stopped enrolling new volunteers in study projects, and we have stopped training activities. We have two new missions:

Provi ding h umau ita ria n aid re lief to 5000 refugees who are 011 OUl' premises after the earthquake-c-and need everything-and taking care offlie injured.

Q: Who else is working with you?

J.P.: This new mission has strengthened om collaboration with Quisqueya University to educate the population and avoid a major epidemic 0 f waterborne in fectiou s di seases We're also collaborating with other new partners, such as World Water [Relief] to provide potable water and Action centre la Faim to setup latrines. The U S, Department of Health and Human Services has supported us in setting IIp a complete hospital able to handle all medical and surgical emergencies.

Q: How are you holding up?

J.P.: We did not choose this situation and it's tough, but we arc up to this new challenge,

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

509

I

NEWS OF THE WEEK

NEWSMAKER INTERVIEW

Climate Science Leader Rajendra Pachauri Confronts the Critics

NEW DELHI-Ithas been a long, hot winter for the Intergovernmental Panel on Climate Change OPCC) and its chair. Rajendra Kumar Pachauri, E-mails leaked last November cast doubt on the integrity of a few of the 4000 scientists who produce consensus reports for the UN. body on climate change science (Science, 4 December 2009, p. 1329). Then IPCC earlier this month offered regret for having included an unsupported prediction in its fourth assessment in 2007 that Himalayan glaciers would melt away by 2035 (Science, 13 November 2009, p. 924).

Pachauri, a 69-year-old industria] engineer and director general of The Energy and Resources Institute (TERJ) here, has headed IPCC since 2002, He r outine ly puts in 18-hour days and is not known to have taken a.vacation in 3 years. The workaholic has recently come under attack in the U.K. press for his lucrative stints as an adviser to companies including the Toyota Motor Corp. and Deutsche Bank-eam:ings that he insists go to TERI. On a cool, smoggy morn:ing here earlier this week, Pachauri defended. rPCC's work and shot back at.critics who want to see him ousted as panel chair. -PAlLAVABA6LA

Q: lhe big issue dogging IPce this winter is the inclusion of a prediction in the fourth assessment that Himalayan glaciers would melt by 2035. IPec has offered regret-but not an apology.

510

R.K.P.: We have made a mistake and we have admitted that. Our job is essentially to bring the science into our assessments from the best sources that exist. Look at the extent of the glaciology work thai has been done in this country It is pathetic. I mean, that is really where we need to come up with an apology ..

0: In a 20 January statement, IPCC still says that India's glaciers are melting away. Isn't that a tall claim?

R.K.P.: OUI" glaci ers are under the same i.nfl tlences, the same temperature changes as other glaciers in the world. So you know we cannot possibly assume if all the other glaciers are melting, that for some reason we are insulated from those influences, The lay pub! ic . , . call see with their eyes what is happening to our glaciers,

Q: What is your stance on linking global warming with extreme events? Has IPCC made a blunder by suggesting the link? R.K.P.: No, we have not made a blunder, and we are going to issue a statement on that. We decided weU over a year ago to do a special report on climate change and extreme events. We would like to assess all the new information and research now available,

Q: Some critics contend that while !pce was projecting that it was doing great science, it is turning out to have done some sloppy work.

R.K.P.: While I am sure there are some peopie who believe that, I a lso can tell you that there is a large body of people wholook at the entirety of what lPCC h~LS done. We have placed before the world ... a defining piece of work, which clearly tells you about the scientific reasons for climate change.

The veracity, the honesty, the scrupulousness with which we carry out our assessment bas been the hallmark of the IPee, and we are never going to compromise on that.

Q: What have YOll learned from these episodes?

R.K.P.: We have got to ensure that aU our procedures are followed in letter and spirit. and with a huge amount of due diligence. J will personally make sure that all the lead author teams that are going to work on the fifth assessment report and om special reports observe this scrupulously, go the extra mile in making sure that we don't use any information that is questionable. What has happened only highlights the importance of the procedures that we have establi shed. J f they had been followed, we wouldn't have got into this unfortunate error.

0·: The other issue that dogged IPCC is the leaked e-mails from the [Climatic Research Unit of the University of East Ang\ia in Norwich. U.K.].

R.K.P.: Those e-mails represent nothing more than private communications, privateairing of anguish or anger OI emotion. It was indiscreet.

Q: Has all that has happened this winter dented the credibility of IPCe?

R.K.P.: J don 't think the credibrl ity of the IPee can 'be dented. If the lPCC wasn't there, why would anyone be worried about climate change?

There are those who would wish to demolish the science of climate change. Our vindication will Iie in our performance.

Q:Are you being made a faU guy?

R.K.P.: I am not a fall guy, but you know the buck stops here, I am the chairman; I am not going to shirk responsibility.

Q: Is there a conflict of interest between your role as IPce chair and your work advising 1)

'"

companies? ~

R,.K.P.: I don't see any conflict at all, Science has :;,

::J:

to be used for decision-makinz, lPec's work is z

.... s_

supposed to be very clearly policy relevant, ~

How can I establish policy relevance JII shut ~ mysel fin au ivory tower and say J will not say ~ anything about climate change? I feel totally g comfortable i11 the role of adviser to anybody. ~

29 JANUA RY 2010 va L 327 SCI EN CE www.sciencemag.org

Q: A statement from fERI lists the number of companies you are associated with, the money which has Hawed back to you and the a rga nizatio n: no 0,000 from Deutsche Bank, S8D,OOO from Toyota, and so forth. Vou don't think this is conflict of interest? R.K.P.:'Where is the conflict ef'interest? I am a paid employee of my institute, not of the IPCC. 1 don't see why 1 shouldn't advise anybody anywhere in the world ... as long as T am not making money out of it. [TIle money] is going to my institute.

0: Some people disagree; they believe that

PHYSICS

you have to be cleaner than Caesar's wife. R.K.P.: Yeah, but Caesar was also murdered by Brutus, wasn't he? Caesar was murdered by a group of people for their own interest, an right? So I cannot possibly be held acco untabl e for all the I ies that the media are writing about in a certain section of'the U.K. press. 1 mean, if they arc going to influence public opinion, I can assure you it is hot going to last forever. 1 am absolutely convinoed the truth will prevail in the end.

0: You put up a brave face, but some in the scientific community feel let down. They say

NEWS OF THE WEEK I

that you are carrying too much baggage, that it's time for you to move on.

R.K.P.: 1 certainly have 110 intenrion to quit. T will continue as the chairman of'the lPCe till 1 have completed the fifth assessment report.

Q: Are you becoming a thorn 10 the side of vested interests-a th orn they wish to eliminate?

R.K.P.: No question about that. But I have no intentions of backing eff I am not going to tailor the truth to suit the vested interests of those wbo would like to continue with business as usual.

NRC Urges U.S. to Rethink Sale of Helium Rese.rve

1[1 1996, the U S. Congress decided to sell the l billion cub i cm eters of gaseous heli UITJspecifically the heavi.er- isotope, helium-s-cthllt the couney bad stockpiled. But conditions it imposed 0[1 the sales are keeping the price of he Hum artificially low ~U1d encouraging waste of a substance indispensable for numerous scientific and technological applications, says a National Research Council report released last week.

"Helium is being sold at fire-sale prices, ",and low prices are not going to encourage the ~ recycling, conservation, and substitution that ~ might prolong the existing supply," says ~ Charles Groat. a geologist at the University of ~ texas, Austin, and co-chair of the committee i that wrote the report.

g Produced in radioactive decay, helium col~ le.cts in the same rock formations that trap other :5 gases and is primarily a byproduct of the natu~ ral gas. industry. It i.s tliC only element that ~ remains 11 liquid at absolute zero, making it an ~ unparalle led cooling agent 01' '·c1yogel1."Witl1- ; 0 ut heli urn, the superconducting magnets in ~' MRI machines won't work and myriad I incs of ~ physics research would grind to a halt. Helium § is also essential to purge the tanks and lines in ~ rockets that burn liquid hydrogen,

~ In 1960, Congress told tile now-defunct § Bureau of Mines to stockpile helium piped ~ from gas fields in Kansas, Oklahoma, and ~. Texas in a rock formation called the Bush ~ Dome Reservoir near Amarillo, Texas. By ~ 1973. the dome held 1 billion cubic meters of ~ gas. But the bureau's helium sa les were weaker ~ than expected, and the reserve was losing :€ money. So t3 ycars ago, Congress told the E Bureau of Land Management (ELM), whicJl ~ had taken control of the helium, to sell almost g all of'itby 2015.

is Congressrequired BLM to sell the gas fat

enough money to pay. off the reserve's debt~ $1.66 per cubic meter with increases for inflation, At the time, B.LM's price for crt!de helium was above the market price for refilled helium, Since 1995, however, global demand forhelium bas increased by nearly 70%, and BLM's current price of$2.29pC1' cubic meter is below the price from private sources,

TIle 60 million cubic meters pumped from the reserve each year make up half the crude helium brought to market in the United States

(hrom~to9rallhy, Leak detection

liftifl9 gas, heat transfer 4% Breathing

7~ I- mixtures {diving) 2"/0

Controlled atmosphere {mil n ul a (Iuri n g)

13%----

for big COJ1SlLm.ers such as NASA and the Department ofDefense that would ensure a supply in times a f shortage, the report says.

The report even suggests tlia.:t Congress rethink the sale of the reserve, as the world's resources could be depleted within 40 years and demand could exceed supply within a decade. "Probably 10 or 15 years ago it was heresyto say we need a reserve," Groat says. "Nowfnatthe situation has changed, I thinkthat may be revisited." At the least, he says, COI1~

Pfe5Sl1re and purge (rn eke try) :

26%

Wise use? Helium is indispensable for rhnting the superconductinq maqnets in the large Hadron Collider (right) a lid rnanutactu ring optical ii bers, but not. for weldi ng a lid f lling balloons,

and a third of the total worldwide. So, the report says, the low price, Which BLM sticks to as a matter Q f policy, drives the market and spms needless consumption, such as the J 5 million cubic meters. used annually by welders in the United States. (Europeans. use argon.)

BLM should establish a higher marketbased price, the report says, although thatmay be trie_ky; as only four refiners have access to tile pipeline to the dmne. To soften the blow to scientists, those with grants from agencies such as the. National Science FOltl1ciatioJi, the NationalInstitutes of Health. and tIle Department of Energy should be allowed to boy BLM helium under terms currently reserved

gress will have to tell BLM what to do after 1015, as the bureau will miss the deadline for selling the remaining 650 million cubic meters of gas by years.

Will Congress heed the report? Maybe, says one congressional staffer. An acute shortage of the lighter isotope of helium, helium-J, has already grabbed legislators' attention, he says, because: it may derail the Department of Homeland Security's plan to deplOY thousa[lds of hehum- 3-fi Iled radiat ion det~ctors (Science, 6 November 2009, p. 778"). "At least )IOU call say to members, 'You were working on this. and here's this other part of the probIQm you Should be aware of'" -ADRIAN CHO

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

511

I

FISHERIES

In Central California, Coho Salmon Are on the Brink

Lagunitas Creek starts high on the north flank of Mount Tamalpais, just north of San Francisco, Cal ifornia, and makes a short run to the Paci fie Ocean, passing through a rural valley and a coastal redwood forest. It was once a thriving breeding ground for coho salmon. Local legends tell of streams so thick with fish returning from the sea to spawn that a person could walk from one side to the other on the fishes' backs. The state record coho, a 1 O-kilogram whopper, was caught on a tributary in 1959.

But those days are long gone. The subspecies of coho that lives along the central California coast is the most endangered of the many troubled salmon populations on the West Coast of North America, says Charlotte Ambrose, a recovery coordinator with the National Marine Fisheries Service (NMFS) in Santa Rosa, Cali forni a, Listed as an endangered species in 2006, the cen-

Central Califor_nia Coast Coho Salmon Historical Range

Survival. plan. A federal plan to save the central California. coast coho salmon will focus on 28 waters h £Ids wh e re CD n serva tio n maasu res (a n have the mostimpact.

... Cily

..",_ Recent cce Coho Presence (2003-2009)

00 ccc Coho Priority Watershed o CCC Coho Historical Range

En c ific Ocean

512

tral coast coho's numbers have recently taken an even sharper tum for the worse. As this year's winter spawning season draws to an end, biologists who've been surveying streams and rivers throughout the fish's range are reporting dismal numbers. A federal species recovery plan to be released next month has morphed into a species survival plan, Ambrose says: "We truly are at the brink of extinction."

The recovery plan will focus on28 watersheds where NMFS thinks habitat restora-

tion efforts-such as restoring floodpJ ains, preserv ing forested areas along creek sides, and placing woody debris in streams to provide shelter for

fish=can have an immediate impact on the coho's survival Lagunitas Creek, which has one of the strongest remaining runs of wild central coast coho, is one. A tour of'the watershed last week illustrated why it may be one of the coho's last best hopes-and why success is far from guaranteed.

In a soaking min, Greg Andrew, a fishery program manager wi til the Marin Municipal Water District, unlocked a gate and piloted his hybrid SUV up a steep, unpaved road that parallels the creek. After a kilometer, what look like two giant concrete slides come into view: spillways for Peters Dam, the largest of seven dams built in the area between 1872 and 1979 to create drinking water reservoirs. The. dams blocked off about half of the. former coho habitat, Andrew says: "We're trying to make what's left as good as possible."

These efforts include periodic dips in the road and other drainage features added to reduce the amount of sediment that washes into the creek, where it can suffocate salmon eggs and clog nooks and crannies in the ~ streambed that young fish use for shelter, 2li Further downstream, Andrew points out a ~ woody debris structure in the creek, one of ~ about 60 built by the water district. These ~ strategically placed piles oflogs create slow i eddies where fish can escape the raging ~ flows created by winter StOlIDS. ~

The final stretch of Lagunitas Creek ~ passes through Point Reyes National :2 o

Seashore before emptying into Tomales Bay. ~

ill a $6.2 million project completed ill 2008, ~-

o

the National Park Service knocked down ~

levees at the mouth of the creek and restored ~

~more th an 100 hectares from cattle pasture -

into a tidal wetland. The project provides ~ crucial floodplain habitat for coho and other 11

29 JANUA RY 2010 VO L 327 SCI EN CE www.sciencernaq . org

wildlife, says park service hydrologist Brannon Ketcham. For coho, the wetlands provide another refuge from the rushing water in winter and a place for smelts to bulk lip before heading out to sea in spring.

If only there were more fish to take advantage of it. As of last week. b iolegists surveying tbe creek for the M arin water district bad counted only 64 coho .in Lagunitas Creek and its tributaries. In the already lean years prior to the coho's addition to the endangered species list, nearly 600 fish returned on average. Because the run usually peaks in early January, Andrew doesn't expect the count to rise substantially, if at all. Across the entire range, Ambrose estimates that only 500 fish have returned this year.

In Lagunitas Creek and elsewlrere, this marks the third 'straight year of abysmal returns for the coho, aL1 especially ominous milestone, biologists say, because of the fish's 3·yea_r life cycle. Salmon born ill a given year migrate out to sea and return to their natal stream 3 years later to spawn and die, This creates three distinct "year classes," each of which returns every 3 years. This year's feeble return suggests that all three classes are faring poorly across the coho's range.

The precipitous decline is probably dne to a combination of factors, beginning with a ISO-year history ofland and water useincluding dams. mining. agriculture, and nrbenization=-that have degraded freshwater habitat throughout the central coast coho's range. More recently, the seasonal upwel Ii ng of nutrient-rich deep water off the California coast has been delayed, which may have reduced the availability offood for h ungry smelts beginni ng the ir III igration out to sea, says Brian Spence, 11 research fishery biologist with NMFS ill Santa Cruz. Moreover, California's 3-year drought has dried up streams and delayed access to spawning grounds, he says; "When yon stack all that on top of the already diminished freshwa tel' capacity, 1 think that's why some of these populations are getting precariously close to blinking out altogether."

Conserving and improving what's left of the coho's freshwater habitat is the best hope for the fish's survival, says Ambrose. Lagunitas Creek illustrates the best case scenario, she adds. Everyone from the National Park Service to the county water district to local cOl1Serv~l.tion groups bas done work On the coho's behal L "There is nothing like tllj s kind of collaboration anywhere else in the range of the central coast coho," Am brose says.

'But even here there is some resistance.

At all animated public meeting last week,

some residents voiced opposition to a salmon conservation plan commissioned by the county government on the grounds that it could violate their property rights. Mere than 100 people braved the rain ~U1d packed into a 1'00111 at a local se]1001. As a woman from the env iroum ental consulting firm that. prepared the pian explained one of its main recommendations-preserving an undeveloped btl ffer zone of up to :3 0 meters along the sides of the creek-many in the audience groaned. lo a narrow valley where many Jots hew closely to the creek, the idea did not.seem to sit weU .. One man yelled out:

"Are you being realistic? Are you on planet Earth?" In the public comment session, residents' concerns seemed roughly split between preserving property rights and aiding the salmon,

Aside from habitat 'restoration, the only remaining way for humans to help coho is to raise them m captivity and 'release them into the Wild. Unfortunately, couservati on hatcheries have at best a mixed success rate, says John Carlos Garza. an NNr_F8 geneticist in Santa Cruz. "Historically, our best guess is that hatcheries have overall had a detrimental effect on salmon populations," he says. One rca son is inbreeding: When populations dwindle, the chances of breeding related individuals goes up, and the inbred offspring have poor survival rates in the wild. Garza explains. To mitigate this problem. Garza has helped the two coho hatcheries in centra 1 Cal i forni a irnpl ernent a genetic matchmaking service. They now use DNA tests to se lect which pairs of fish to breed in order to max imi ze geneti c diversity.

Garza also oversees an experimental project ill which captive-bred fish are raised in fresh water and released into streams as adults. This offers advantages over the standard approac h 0 f releasing young fish born in the hatchery; POT one, the offspring of the adult-released fish imprint on the stream where they hatched rather than on the hatchery Garza thinks this approach may be valuable for repopulating streams where salmon have been completely wiped out.

These innovative methods may help make conservation hatcheries more successful, but they won't be enough to save the coho on their own, Garza and others say. "They're very expensive and intens ive strategies that should be adopted only when til ere: aren't a lot of other good strategies available," Garza says. Unfortunately .. the C entral coast co ho aTC. ru 11 III ng out of options. "Let's be. clear," Garza says, "these are last-ditch efforts,"

-(iRE6 MILLER

NEWS OF THE WEEK

I

ScienceNOW.org

JIt

From Science's

Online Daily News Site

Colliding PartiCles Can Make Black Holes You've heard the controversy. Particle physi" osts predict that the world's new highestenergy atom smasher, the Large Hadron Collider near Geneva, Switzerland, might create tiny black holes. Curiously, though, nobody had ever shown that Einst"in 's theory of gen.' eral relativity predicts that a black hole can be made this way. Now a computer model shows conclusively for the first time that a particle celtision really (an makea black hole. http://bit.!Y/blackholes

Bats and Whales Share Sonar Protein

Bats and dolphins areabout as different as mammals get, Yet, beth hOiTIe inon their prey by emitbn;Q sound waves and sensing the reflections, a process ca lied echolocation. A new study shows that in both qroups the Si;lme protein evolved in the same way to make that possible, Researchers say it's sure prising to discover a moleclJ tar convergence in these very distantly related groups of anim<J!s. http://bit.ly/batwhale

Cold War Split Birds, T09

The Cold War Ir---------= .... divided the people of Europe for nearly half a century; and it turns out humans I

weren't theonly ones sturk behind the Iron Curtain. Trade blockades Led to vastly dirferent n umbers and types of invasive birds in We.stem and Eastern Europe, new research reveals, Thefindinqs, say experts, hig hlight the dr arnatic impact hums n activity can have on the success of alien species. http://bit.ly/coldwarbirds

Birdlike Dinos<jur Was Adept Glider

The fight over bird fligh tevolution is one of the longest-run ningand most heated debates in paleontology, Were the first flyers arboreal creatures that initially glided from treetops to the ground? Or were they bipedal ground runners With evolving wings that allowed them to take progressively longer jumps? The first flight tests of a foam model of a four-winged, feathered dinosaur lend credence to the former hypothesis .. http://bit.ly/dinoflight

Read the full postinqs, comments, and more on seiencenow.sciencemaq .. org.

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

513

I NEWS OF THE WEEK

PHYSICS

Test Shots Show Laser-Fusion Experiment Is on Target

As the managers of the Large Hadron Collider ~t CERN near Geneva, Switzerland, call testify, building your gjant research facility i.s one thing, but getting it to work properly when you switch it 0]] is definitely another. Perhaps w j th CERN'5 setbacks in mind, those in charge of the National Ignition Facility (NJF), a huge laser for nuclear fusion experiments at Lawrence Livermore National Laboratory in California, have been gingerly putting thei [huge machine through Its paces since it WaS completed nearly a year ago.

ILl a study published online this week by Science (www. sci encemag, org/cgil content/ abstraot/l 1.&5634), NlF researchers describe their first experiments using all 192 of the facility's beams on test targets empty cf'fuel, They were able to couple the laser's energy into the target efficiently and implode the target symmetrically=so far, so good, The scientists say they are on target to attempt ignition-s-a self-sustaining fusion reaction that produces excess encrgy=-befbre the end of this year. "It's come up better than anyone thought. They're ahead of the curve predicted," says Mike Dunne, director of the Central Laser Facility of'Rutherford Appleton Laboratory Mar Oxford, U'K,

NIT is the sledgehammer to crack a nut, writ huge (Science, 17 April 2009, p, 326). The sledgehammer is a.laser that occupies 11 build ing the size of three football fields and 10 stories high. Inside it, hundreds of optica I amplifiers, beam splitters, and other devices ta ke a norma! laser beam, spl it it 192 ways, and boost the combined energy of the beams to 1.8 megajoules.

The nut is a tiny spherical capsule the size of a pepp erc om made of'beryllium, which in later experiments will encase a dash of deuterium and tritium (D - T)~isotopes of hydrogen, The aim is to use the power of the laser to heat the capsule so fast that it explodes, propelling the D-1 fuel inward toward the center and crushing it to atemperature arid pressure greaterthen those in the core of the sun, As nuclei in the Ver'} center begin to fuse, the energy produced will cause all the D- T nUCll to burn 'it1 a flash of energywith luck, mort" energy th~n was pumped into tIle capsule in the first place.

Hut the peril of this "indirect' approach is. that gold. atoms kicked off the in si d e wall of the hohlraum create a plasma that can interfere with the incoming beams in unpredictable ways, In the experiments, the N1F team managed to tune the many beams to keep these laser-plasma interactions to an acceptable level. "They were sufficiently benign at this energy, wbich,is a. huge success," says Dunne,

In fact, the team turned some of these in.ter<1cti ons to their advantage. In the past few years, theorists had suggested that with so many beams converging into the-narrow ends of the lrohlraum,

the beams could nudge the plasma into a regular repeating pattern, producing a sort of diffraction grating that might scatter the beams, It was potentially a "terminal pro b~ lem," says Dunne. But early last year, Livermore researchers suggested .that they could use the gratings to steer the beams toward hard-to-reach parts of the hohlraum interior-> in particular halfway down the inside wall, farthest from the entrance holes. The recent experiments have proved their theory right. "You can deposit energy where you need It,." Glenzer says. LLE's David Meyerhofcr is impressed, "It's the first time laser-plasma interactions have been used in a beneficial manner; Usually, yon try to avoid them," he says. "There's no suggestion that this won't work at full energy."

The nex t 1Tl aj or hurd I e is ens uri 11 g th <It th e implosion of the fuel is smooth and symmetLical.This remains "an open question," Dunne says, because experiments so far have used empty capsules and the explosions "were not uniform on a microscopic scale, and that can cause problems later." Even so, he says, he's "not too worried yet," because the capsules used were not machined to the same precision used for ignition shots,

Experiments ,~Lt NJF arc currently stopped arid the 'ignition carnpaign will be-gin '11 &1IT)C$t. in May Glenzer says. If all goes well, Dunne says, adecision wi II be made 1]'1 July on whether to push ahead with full D-T fusion oJ

c

experiments and an attempt at ignition ill ~

October. Successful ignition th is year is "not ~ out of the question," McO·01Y says. "But I'd ~ be surprised iftt happens," :5

-DANIEL ClE:RV ~

Hot €eU. To implode the central fuel capsule, NI~ fires. 192 beams into an erasersized gold cyli nder, he~ting it to x-ray temperatu res.

514

Livermore. has spent more than 10 years and $3,5 billioll bui.ldiug NIl?, and researchers hope successful ignition experiments wi U pay back that investment in future fusion power stations. But a m a in part of NIF's role is to test computer simulations of nuclear explosions to ensure that tbe u.s.. nuclear weapons stockpile is reliable.

In the paper, Siegfried Glenzer of Livermore and his colleagues there, at Los Alamos National La bora.tory, and at General Atomics in S<i11 Diego, Cali fornia, deseri be shots using a beam energy of 0.7 rncgajcules, about 40% ofNJF's max irnum. "We're doing the local th ing, and it's going better than expected' Glenzer says.

The team addressed the problem of getting the most laser power onto the capsule find do ing so symmctri cally so that it implodes evenly. The lasers output is in the ultraviolet, but lor a better implosion you need. x-rays. So instead of shining the beams directly onto the capsule, they put it in the center of a gold cylinder about the size of a penci 1 eraser, called a hohlraum, By shining the beams through holes in the ends of the hohlraum, 1heycan make its inner surface hot enough to emit x-rays, which cause the capsule to implode. In the experiments described, the team produced a radiation temperature inside (he hohlraum of3.3 million kelvin, exactly in I inc with models. Robert McCrOJY, director of the Laboratory for Laser Energetics (LLE) at the University of Rochester in New York state, ca lis the feat a major achievement, "If you don 't get the hohfranm temperature you're not goi.ng to get the implosion you need,"

29 JANUARY 2010 va L 327 SCI ENCE www.science mag. org

PLANETARY SCIENCE

Did a Battering Rain of Comets Bring Ganymede to Geologic Life?

Scientists have come up with 11 promising explanation of a planetary odd couple: Jupiter's major moons Ganymede and Callisto, Much like the disparate pair of Earth and Venus, the two moons have similar sizes and similar compositions but have followed vastly different paths of development, Once formed, Ganymede separated into layers of ice, rock, and molten metal, and then something reshaped parts of its surface, Meanwhile, Callisto went almost nowhere. It has not moved much beyond the bland, newborn ball ofmixed .lee and rock it started outas,

A study published this week in NCH1lre Geoscience offers a possible explanation for why Ganymede: evolved so much more than Callisto did: Ganymede suffered a [ar worse beating 3.9 billion yearsago when a min of cornets and ast.er,oids battered much of the solar system. "An interesting idea, well argued," says p 1 auetary physicist Dav id

of the LHB's impacts 011 Ganymede and Callisto, tbey found plenty of discrimination, Jupiter's powerful gravity would have accelerated incoming comets and drawn more of them near the planet, the researchers note, Ganymede, being closer to Jupiter, would have suffered twice as many impacts as Callisto and at higher velocities. So Ganymede would have received 3,5 time-s more energy than Ca Wsto.

In their modeling, that was more than enough heat to begin me ltin g the ice in Ganymede's natal mixture of ice and rock. That thaw would have allowed the moon's rock to start Sinking through the increasingly slushy intcFior. Then the sinking rock would have givenup its gravitational energy as heat, accelmati ng the 111 0 on's separation in to layers. E ventually, enough heat from r,adioac;tive: decay would have built IIp to separate the rocks iron into a molten core, drive a magnetic field, and perhaps form Ganymede's grooved surface geology. Shortchanged on impact energy, Callisto 'would not have melted enough to achieve "runaway" heating during separation, leaving it cold and without a core, Wifh 11 negligible heat source below it, Callisto's surface would be geologically dead for cons.

"It's 11 classic goodscience paper," says planetary geologist James Head of Brown

No Longer two peas in a pod. Some sou rcc of heat-possi bly a storm 01 impacts eOM ago-led to the resurtacino of Ganymede (left) but lett Callisto (n"ght), la rgely unchanged.

Stevenson of the California Institute of Technology in Pasadena "I think it's promising .. "

For 30 years, researchers have wondered what process could have got enough heat into Ganymede to drive its geological evolution without setting off Callisto as welL In search of a Ilea! source that could discriminate between the two moons, planetary scientists Amy Barr and Robin Canup of the Southwest Research Institute in Boulder, Colorado, con-

'" sidered the late heavy bombardment (tHB). ~ Thatsa hypothesized storm of comets' and (in ~ the inner sO.lal" syste~') asteroids that many 1 planetary scrennsts think pummeled the solar g system 3.9 billion years ago.

!3 When Barrand Cal\Up simu luted the effects

University. "It's very exciting because it combines a couple of seemingly disparate problems.TI'he scenario offers a promising, testable explanation for the Ganymede-Callisto dichotomy. be says.And to the extent that it proves to be an attractive explanation of the dichotomy, it also lends support to the reality of the LHB. In particular, Barr says, urn beating "fits nicely with the Nice model," Thar.model (named after the French city where jt was developed) shows how Jupiter and Saturn could have stirred up ali LHB while migrating outward through the solar systern (Sc fence , 17 July 2009, p. 262). But researchers agree tbat much analysisremains to be done, and another mission to the Jupiter system Would be nice. -RICHARD A.KERR

NEWS OF THE WEEK

I

From the Science Policy Blog

Writing in f)f!r Spiegel" three prominent climate scien tists have criticized th e pol \c\es of the I ntergovemmental Panel on Climate Change and its chair, Rajendra KlJmar Pachallri. The Wall Street Journal reprinted the column, which says the pa nel shou ld adopt conflictof-interest policies, a mechanism for dealing with errors, and more transparent policies for selecting its leadershi p and authors. http://bit.lyI1WXOph

Director Peter Sawicki has been Let go by the German Institute for QuaLity and Efficiency in HeaLth Care, and Foli~ Kalatas· is voluntarily steppi ng down from his position ~s president of the European Research Council. http://blt.ly/SaNC]F

Budyet woes have imperiled the dream of Alain Beaudet, president of the: Canadian Institutes of Health Researdi, to launch a massive: initiative to combat ALzheimer's disease and dementia. A govern men t d eficit has sea led back th e proposal from hundreds 01 millions of dollars to a far more modest S 10 million. http :lfbit.lyJ7 d RBWi

A new report says thattar fewer people than believed have died in the Democratic RepubLic of the Congo from a decadelong civil war. The lnternerional Rescue Committee has cited a widely quoted figure of 5.4 million, but the new analysis, part of a publication called the Human Security Report 2009, faults the comrnittee's methods and suggests that the number is closer to 900,000_ http://biUy/4TJZAc

A Ca lifornia physiCist suggests that d~liv" ering feed to Haiti by small, widely dispersed air drops would be- more effective than deliveri ng it toa few sitoe~ in large packages, which can be hard to distribute because of poor tnfrastructu re 0 r thieving thugs. Sill Wattenburg says the u.S. milita ry has been wa ry of the technique, however, despite its successful use in Bosnia and Afg hanistan. http://bit.ly/6Xfg20

for the full postings and more, go to blogs.scieJlcemag .org/scienceinsider.

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

51.5

.... Meow. A gizlllo occupying two places at ..,. once wo u ld be an a flalog to Schrodinger's dead-and-a live cat.

exploded in terms of the number of people working in it," says Robert Knobel of Queen's University in Kingston, Canada. "They're bringing their toolboxes from all these different fields,"

Quantum machines could lead to devices that blur the lines between el ectrcnics, optics, and: mechanics-humming widgets th.at coax light into odd states or translate information encoded in quantum states of photons into electron ic signals. They might even probe a

fundamental mystery: Why don't human-scale objects behave quantum. mechanically? Reaching the ground

state of mechanical motion is "tile kind of result that will create a new field in itself," says Jack Harris, an experimenter at Yale U n.i versity. "This will be a door opening-s-a big one."

aintest hrum Heralds OuantumMach·nes

Hig,h frequencies and low temperatures But first physicists must make machines that make the slightest movement. Twentieth century physicist Werner Heisenberg's famous uncertainty principle implies that, like a small child, no object can stand perfectly still. Even in its ground state, it must possess a last, inextricable half-quantum of energy andjiggle maybe a half-dozen groups observing this," with zero-point motion. Physicists would like he says. to see that minimal tremor in a mechanical

If so, then the age of quantum machines widget. To spot it, they are tinkering with will finally have arrived. Molecules, atoms, nanometer-sized beams of semiconductor, and subatomic particles all obey the mind- micrometer-sized bits of glass, and even bending dictates of quantum theory. But mirrors weighing kilograms.

physicists have never observed such odd Those objects all share a key property: behavior in the movement of a humanrnade When each is nudged away from its equilibobject. So their first goa I has been to put the ri urn pos ition or shape. it oscill ates at a well-

simplest machine-a vibrating defined frequency like a tuning

beam or some other "oscillator'v= 0 n lin e fork. Quantum mechanically,

into its ground state, which would such a "harmonic oscillator" can

be a crucial first step toward sClencemag.org absorb energy only in quanta

Podrast interview

machines that oscillate around. with author whose size is proportional to its

two different positions at once and Adrian Cho, frequency. So to reach the ground

other weird. states of motion. state, physicists need to extract ali

Within the past 6 months, four different. but the last, irretrievable half-quantum. That's groups have come within a few dozen quanta easier said than done. Each quantum is so oftha!. goal, and researcherssay one may have small that to remove them all, researchers

reached it. must. lower the oscillator's temperature-Land hence its energy ----{l ea fly toa bs 01 ute zero.

Nevertheless, physicists tried the direct route to Ole ground state. To Tnake the energy quanta as large as possible. they etched beams of semiconductor that vibrated at high frequencies-s-up to I billion cycles per second, or 1 gigahertz, To make. the beams

Aft.er years of trying, physicists are on the verge of making tiny vibrating devices that make the slightest possible movement. Far weirde.r mechanical devices could follow

IF YOU WANTIO BE ,REMEMBERED, MAKE A bad prediction. In 2003, Science reported that within 6 months physicists might create tiny machines whose movement obeyed the weird rules of quantum mechanics, which state that an object can absorb energy on Iy in discrete "quanta" and can be in two places at once (Science, 3 January 2003, p. 36). That didn't happen, but Tobias Kippen berg, all experimenter at the Swiss Federal Institute of Technology, Lausanne. still shows a slide of the article in his talks. "It's entertaining, and it also reminds people exactly how challenging the problems are in this field," Kippenberg says.

Those obstacles are not insurmountable, however. Kippenberg's team and several others have nearly entered the realm ofquantum motion. Ironically, to get there, they're racing to make gizmos that make literally the slightest movement. vibrating widgets drained of every possible bit of energy and quivering witll only an unquenchable "zeropoint motion." Recent progress toward that "ground state" of motion has come so fast that even Kippenberg is willing to prognosticate: "1 expect in the next year there will be

516

Curiously, as physicists have homed in on their objective, the field has grownmore diverse. Seven years ago, a few groups of condensed matter physicists were pursuing a single strategy for reaching the ground state. Now, teams from optics and astrophysics have jo ined the cha se. "The field has

29 JA NUA RY 2010 va L 327 SCI EN CE www.sciencemag.org

as cold as possible, they relied 01.1 so-called passive cooling: sticking them in the best 1 i q uid-heli urn refrigerators, w-hich can reach a few rnillikelvin, or thousandths of a degree above absolute zero. To sell se a beam's motion, they employed a scheme called "capacitive coupling," applying a voltage between the beam and an electrode (see diagram) .. The beam's motion causes the voltage to vary.

Researchers tan into a few roadblocks, however. To push the frequencies of their vibrating beams higher, experimenters made them ever stiffer: But that meant that the size of the beam's already minuscule oscillations decreased even further, making them diffiell [t to detect

The scheme for measuring the beam's motion actually jostled the beam as we1J. Physicists tracked the varying voltage by observing how it affected the current througha device called a single-electron transistor, But as th~ elec.trotlS hopped through 0 ue by one, they tugged on the beam, creating "back ac.tion.""This back action is quite a bit bigger than you need for th 6S e quantum me as nrerne nts ," Kno be J says. "And 1 don't think we understood that theoretically or experimentally at the time."

Coot new ideas

Even as the straight path to the ground state proved difficult, new avenues toward that goal began to open. In particular, experts in quantum optics began experimenting with tcchniqu es to control the m oti on 0 f micrometer-sized objects with laser light. '"It turns out that YOl! can use the whole quantum-optics toolbox to prepare, manipulate, and read out a mechanical system," says Markus Aspelmeyer, a physicist at the Uuiversity of Vienna,

Ironically, physicists call cool an osci llator by shining light on it. Conceptually, such active cooling works as follows: Researchers put a tiny rnir r or on a11 oscillating beam (see diagram). It and a larger, fixed mirror forman "optical cavity" that resonates with light of a frequency set by the mirrors' spacing, just as an organ pipe whistles at a pitch set by its length. If the little mirror could [Jot move, then only light ofthis frequency could shine

:;: through the large mirror and into the cavity,

e But if the mirror oscillates at a definite ~ frequency, then laser light whose frequency ~ has beer; lowered by that amount can also

o· .

8': enter tile cavity, To do so, however, each

o .

t photon must absorb a quantum of energy

~ from the mirror to make up for tile energy it ~ is lacking, So that "detuned" light saps 5 energy from the osc.illator. The light wave

reemerging from the cavity also reveals the minor's motion through shifts in the alignmen t of its peaks a 11d troughs, or "phase," allowing researchers to detect the motion with greater sensitivity.

Using such "resolved-sideband cooling," three groups have reduced the energy of a micrometer-sized oscillator to between 32 and 67 quanta, as they reported in July 2009 in Nature Physics. The experiments varied ill details. Aspelmeyer's team used a mirror on a beam; Kippenberg s shined light into a glass ung that served as both optical cavity and oscillator. Hailin Wang's team at the University of Oregon, Eugene, used a glass bead .in a similar way. All three worked in fridges that reacheda few kel vin; they think they can reach the ground state by starting at mill ikelvin temperatures, "It doesn't loo]; like we arc bumping into ~U1y fundamental issues" that prevent it, Aspelrneyer says.

Physicists working with nanometer-sized oscillators have found better ways to chill and measure their devices, too. In fact, they've borrowed the concept of resolved-sideband cooling. For example, Keith Schwab of the California Institute of Technology (Caltech) in Pasadena and colleagues have applied it to a silicon-nitride beam 30 micrometers long and about 15 0 nanometers thick and wide that thrums at 6.3 megahertz.

The beam couples to a nearby microwave resonator-simply <1 long strip' of superconducting niobium that can ring with microwaves of a particular frequency, As in the optical experiments, detuned

Chillin'. 111 the old·school apPI'O~Qh (teft) , physicists sim ply mot a v:i bra tiog beam i[l a heUu 111 refrigl'f3tor. J n a flew tack, Laser light

sa ps.a beam's energy as it enters an optiGilI Gay" tty with the beam at ens end.

BRUTE FORCE

NEWSFOCUS

microwaves e<U1 enter the niobium strip only by absorbing energy from the beam. Starting <It milli ke lvin temperatures, the researchers r educed the be<l111'S energy to a handfu I of quanta .. 11 s they reported this month in Nature. "We pushed the experiment as hard as we could, and in the end we came dawn to four," Schwab says.

Brute. force carries the day

In the race to the ground state, however, the straightforward approach seems to have-won out. Andrew Cleland, John Martinis, and colleagues at the University of California. Santa Barbara, have succeeded by using a "brute force" combination of passive cooling and a very high oscillator frequency, say several physicists who have seen preprints describing the work. The key to the experiment lies rn a clever scheme to detect the oscillator's motion, they say.

Cleland and Martinis's gizmo is a beam that vibrate-s at a whopping 6 gigahertz, researchers say. B ut rather than swinging up and down, it gets thinner and thicker. It also consists of so-called piezoelectric materia! that creates all oscillating electric field (IS it expands and contracts, making that motion easy to detect.

To do that, Cleland and Martinis rely on a widget called a "phase qubit," a strip of superconductor with a nonsuperconducting patch in it that acts a bit like a sandbar in a stream offree-flowing electrons. The details aside, the phase qubit is itself a highly controllable quantum-mechanical system with a

lIGHTTQUCH

L:)se.r !ig~!. ";,que.m;v _f'."-I.- /"1

Mirror

www.screncernaq.orq SCI EN CE VOL 327 29 JANUARY 2010

517

I NEWSFOCUS

ground state and oue higher-energy state. Researchers can case the qubit from one state to the other-or even put it into both states at once-by applying microwaves of a specific frequency, Moreover, they can change that frequency by adjusting the current flowing through the qubit,

So Cleland and Martinis can feed energy quanta into the oscillating beam one by one. They first put the qubit into its energetic state and then adjust the qnbit's frequency to matcb that of the oscillator to shuffle the quantum of energy over, They can also run the process backward 1,0 pull quanta out of the oscillator. And the team has pulled out every last one to reach the ground state, others say. "1 would say that Andrew and John have achieved it," Schwab says. "Wc'vo/ gotten damn dose, but these guys are deep ill to 'it."

Cleland and Martinis had previously used a phase qubit to fiddle with, a microwave cav! ty. They showed they could put the cavity into any delicate quantum state, including tbe ground state- or one In which it contained t\1>'O di fferent n urn bers of rni crowave pho to I.}$. simultaneously, as they reported in May in Nature. Now, they have simply replaced the gigahertz microwave cavity with a gigahertz mechanical oscillator, others say.

Not everyone is convinced that the Santa Barbara team has reached the desired goal. The researchers detect not the motion of the beam but the electric field the material produces, Kippenberg argues: "It's not a purely mechanical oscillator." The experiment could be done with a different material that would produce such a field through internal stresses, without mechanical motion, he says. Others say that's 11 quibble.as the Santa Barbara device does move.

Testing the limits of reality

Just what quantum machines will be good for remains to be seen. Most immediately, they might have technical applications in basic research. A gigahertz nanomechauical oscillator in its ground state would make an exquisitely sensitive force detector, says physicist Konrad Lehnert of .fILA, a laboratory run j cintly by the National Institute of Standards and Technology and the University of Colorado, Boulder. "That gives you a way to interrogate the nanoscale world in a very gentle, noninvasive way," he says ..

Micrometer-sized widgets 1:l1 quantum motionmight prove particularly useful in quantum optics experi ments, says Yale's Harris. They might serve as "nonlinear" optical elements that perform tasks such as splitting a single nigher-energy photon into

two lower-energy ones, he says. Currently, researchers do such things bypassing light through certain crystals, which typically absorb al arge fracti em of the photons. Quantum machines might do the job with(JU( such high losses, Harris says,

The tiny machines might also bridge the gaps between various technologies. Physicists can already control the quantum behavlor of electrons and of electromagnetic waves such as light or microwaves, Quantum

Many sizes .. Physicist are I;xperim,enting with nanometer-sized beams (top), nucrorneter-slzed rings ("middle), and macroscopic mirrors,

machines would enable them to control quantized vibrations, or "phonons," and forge connections among all three.says Oskar Painter, an applied physicis t at Caltech, "You 've got phonons, photons, and electrons" working together, he says. "That's where the revolution is going to come from:"

PainteJ: is already pushing in that direction, His team recently fashioned a beam of silicon 30 micrometers long and 1 A micrometers wide arid patterned it with a' ladderlike arrangement of IIOleS, The pattern sirnultaneously traps light and vibrations. In fact, the pressure from trapped light can set the beam

vibrati ng, and that motion reveals itself' in microwaves, emanating from the. beam, the team reported in October 2009 in Nature. Such <1 structure could convert optical signals to microwaves and vice versa, Painter says.

Quantum machines might ultimately probe the or ig ins of everyday reality. Although the rules of quantum mechanics allow an object to be in two places at once, human-sized "classical" obj ects do uot behave that way. "Systems either behave quantum mechanically or classically," says Nergis Mavalvala, all astrophysicist at tile Massachusetts Institute of Techno logy in li Cambridge. "Is there something rumley in ~ between? I dO[l't think anyone h as an iii answer for that," 6"

o

Many physicists think that in principle a ;;;

large object could be put mto such. a two- ~ places-at-once state-if it were shielded 'g from vibrations, radiation, and other environ- ~

J'

mental i lIn nen CeS, whi ch cause such deli- ,8

cate states to "collapse." Others argue that ~ as-yet-unknown factors may prevent large 0

:;:

objects from behaving quantum rnechani- i

cally; Earned British theorist Roger Penrose "

(1

argues that if a large object were put into a l'5

here-and-there state, its own gravity would ~ pull it to one place or the other, taking the 6i

g

quantum weirdness-and perhaps, some of ti

theftm-out of'the everyday world. ~ To test such ideas, many researchers -:; would. like to try to put a human-sized ~ object in two places at OI'ICC. 'There's noth- g

to

ing better than all experiment that proves B

that it works 01" it doesn't," Mavalvala says. fi d Such experiments are a ways off. hut physi- ~

cists arc surprisingly close to reaching the ~ "ffi

ground state of a j umbo osci llator. Using '"

~,-. optical techniques, Mava lvala, and col- -

lea gues cooled a 1- gram mirror to 1 00,000 ~ quanta, as they reported in 2007 inPhysicat ~, Review Letters. . .•

More recently, the group has cooled mir- ~ tors weighing 10,.8 kilograms even furth er. ~ The four mirrors form two crossed 4-kilome- ~ ter optical cavities in the Laser Interferometer ~ Gravitational Wave Observatory (LlGO) in ~

l!:

Hanford, Washington, which is designed to ~

detect rippl es in spacetime. Using Ll GO's t;,; electronic stabilization system in Iieu of ~ lasers. the team cooled the mirrors' relative 3

'0 '" z:

~ §

motion to just 234 quanta', as they reported in July 2009 in the New Joumal of Physics.

"Exactly where, quantum machil'icS will

~ lead may be impossible to say, But once. 0

physicists can put gadgets into quantum g motion, the. possibilities may be limited ~ on ly by researchers' ingen ui ty Sam ething ~ new and wild seems sure to shake loose. ~

-ADRIAN CHO 5

518

29 JANUARY 2010 va L 327 SCI EN CE www.sciencemag . org

EVOLUTION

In the Deep Blue Sea

A survey of corals and their kin shows the ocean's evolutionary potential

The deep sea may noise em like a crucible 0 f evolution. But, to the surprise of biologists, a new construction of the coral family tree suggests that evolution proceeds at full bore :iJi waters well below where sunlight peuetrates, Moreover, some coral diversity may have bloomed there first, before spreading coastward-c-the reverse of'what has long been thought. "As people look in the deep sea, they are finding much more diversity than they expected," says Clifford Cunningham, an evolutionary biologist at Duke Uui ~

Z versityin Durham, North Carolina. "We're ''5 just at the very beginning of understanding fu deep-sea evolution."

~ Ata meeting" in Seattle earlier this month, ~ a dozen experts celebrated progress in assem~ bling an improved family tree showing the ~ relatedness of jellyfish, corals, anemones, ~ and hydra, wh.ich are collectively known as a cnidarians, Relatively simple creatures char~ acterized by their production of specialized i sti ngjllg. eel Is, en idari ails are i ~po:t.allt to ~ the marine ecosystem. Some live 111 deep ~ water; others in the shallows. Some, like

0(

§ reef-forming corals, live in large colonies

~ and partner with algae; others go it alone.

t By comparing a variety of genes among ~ cnidarians, Cunninrrham and the others

§ -

:."w~ gathered in Seattle have pieced together

'" large swaths of their family tree. This ~ Cnidarian Tree of Life project and other g studies have unveiled where some of the

'"'

il'" species arose and where traits such as colo-

~ niality came from. The most paradigm~ shaking resu lt so far concerns certain 1 corals. Researchers have Jong thought that ~

1::; deep-sea. corals were derived from a few

~ species of shallow-water reef dwellers that ~ migrated mit to sea over time. Those new~ comers didn't diversify very much in their ~ new environs, it was thought.

a

~

6l *The Society for Integ rative and Compa rativ.e Bio logy

5 m~e!illg look place 3-7 Jan uary in Seattle, Washington.

But 40 years ago, IDaline biologists discovered quite a diversity of worms and other softbodied creatures in deep-sea sediments, with up to 100 species per shovelful of mud. Marveling over this surprising variety, Scott France of the University of Louisiana, Lafayette, wondered whether bard, rocky sea floors were also rich in coral species. Since 1992, he and his colleagues have been collectiug specimens and data from submersible expeditions and from museum collections, focusing on black corals-s-nanred for their dark, flexible, treelike skeletons-s-and octocorals, all of which have eight tentacles as opposed to the black coral's six.

To figure out who lived where, Eric Pante, one of his graduate students, charted the data on depths at which 3100 museum specimens were collected and incorporated specimens that he and France collected a.t sea. They found, for example, that most of 420 metallic" colored chrysogorgiid corals documented lived ill the deep sea, but a few were confined to sha Ilow water,

France and his colleagues also sequenced parts of several genes from representatives of three families of o ctocorals, those commonly known as sea fans and sea Whips. Comparisons of the DNA sequences allowed them to make a family tree, upon which they could overlay the depths of each species' home, The deep-sea species in the three octocoral families "all seem to be coming from a single common ancestor" that. lived in deep water, France concluded at the meeting. Mercer Bru gler, another 0 f Fran ce 's gradu ate students, looked at three families of black corals, which are valued for jewelry, and concluded that they too diversified in the deep ocean

Thi s emerging story is 11 't limi ted to cora Is.

Marymegan Daly, a systematist at OJljO State University, Columbus, and co-coordinator of the Cnidarian Tree of Life project, has found a similar pattern among the sea anemones she's analyzed. "We. see Jots of radiations of deep-

Diversity in the dark. (left to right) A colony of bam boo coral. a solitary deep-sea (0 ral, a nd a chry,ogorgiid cctccoral speak to the ri th ness of the ocean depths.

sea forms," she reported. In short, concludes France, "we can 't say that the deep sea is a boring environment. ill terms of evolution"

And although the Cnidarian Tree of Life project has confirmed that other types of corals have shallow origins, those lineages may find refuge in the deep sea during tough times, says Marcos Barbeitos, an evolutionary biologist at the University of Kansas (KU), Lawrence, Barbeitos studies solitary corals, which are individual polyps, and some of their reef-forming counterparts, which are large colonies of polyps that tend to live in shallow water. Taxonomists have tended to group reef corals with each other, putting deep-water solitary corals in a separate group. They have also assumed that the colonial corals arose from an ancestor that was a solitary coral. But Barbeitos's study, based on DNA samples from 97 coral species, suggests that some soliI:<IlY corals are closer" kin to shallow-reef colonial species than to other deep-water solitary corals and that the evolutionary path to colonia1.ity may equally well go the other way.

A statistical analysis revealed that at least. two of the solitary lineages from the deep sea evolved from colonial forms rather than the other way around. These colonial forms apparently "diversify in shallow water and give rise to [solitary] species that live in deeper water," says Barbeitos. Furthermore, his data indicate that colonia I forms are about as likely to arise from solitary forms as the other way around, suggesting that colouiality can be lost and gained over evolutionary time.

Barbeitos suggests that as reefs waxed and waned over the planet's history, corals have managed to survive in part because deepwater lineages persist. They then expand and diversify into sha llow-water, reef-forming corals when conditions are again favorable.As With the black and octocorals, deep water is critical to Cnidarian evolution, says KU's P aulyn Carrwrjgh t, co-coordinator of the Cnidarian Tree of Life project. Corals "may be more robust because of the connection to deep-water relatives." -su ZABETH PEN N IS I

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

519

The flagship building for Sanford Health ill Sio-ux Falls, South Dakota, really does fly flags, colored pennants over ramparts and turrets. Shields with pseudoheraldic symbols hang inside,and wooden thrones overlook the lobby. This children's hospital runs on what it calls the "Disney principle": Just as children never see the Mickey Mouse mascot at Disneyland without his head on, Sanford never lets the fantasy lapse with a bare surface. CT scanners have twisting otters painted on them, and doors of tiny, Wonderland size appear in the hallway, open ing onto dioramas of secret worlds. Had Alice fallen ill, she would have convalesced at something like the Sanford Castle.

With so much invested in decoration, the castle is decidedly a patients' hospital. Indeed, the focus of San ford Health (a cluster of clinics formerly known as Sioux Valley Hospital and Health) has always been patient care, not clinical science. But recently, the institution's stepsister of a research arm was transformed. T. Denny Sanford, a banking mogul, bad already donated $16 million to build the castle in 2004, an as toundi ng amount for South Dakota. To help push beyond caring for patients, in February 2007 Sanford donated $400 million more, largely for medical research--one of the largest gifts ever bestowed on any U.S. hospital. Rich Adcock, art executive vice president at Sanford, said the hosp ital had a vision for growth, "but we knew we needed an angel. Denny Sanford was that angel."

With Sanford's blessing, officials announced what's known as the Sanford Project: a plan to cure one widespread disease within Sanford's lifetime. The deadline is a challenge; Sanford turns 75 this year. After interviewing experts, project directors selected type 1 diabetes (juvenile diabetes), a disorder in which immune cells destroy the body's beta cells, the insulin-secreting cells in the pancreas. Newt Gingrich, the form er U. S, House 0 f Representatives speaker and a health care-reform advocate, showed up to help announce what Sanford Health is calling a "Manhattan Project."

Conquering diabetes is ambitious enough, but then the project declared it had already settled all the approach it would take: harnessing the body's ability to regenerate beta. cells. The choice startled some diabetes scientists who saw equal or more p romis e in oth er appro aches-i-s tern - ce U therapy or beta-cell transplants And the ultimate goal ofthe Sanford Project may be even more audacious: enticing enough scientific talent to create a permanent biomedical hub, a Boston or San Diego, in the 46th most populous state in the union,

'let's st.a.rt tomorrow'

Paul Bum, a biochemist and he ad of research for the Sanford Proj eet, is exasperated that most clinical diabetes research focuses on preventing overactive immune cells from destroying beta cells. Doctors normally don't diagnose chil dren with dia-

Night tights. Sanford Health plans to build about 20 children's clinics worldwide, all shaped like castles.

betes until they've lost 90% of their beta cells, so corralling their immune systems alone won't reverse the damage, he argues. He thinks it's essential to regenerate healthy beta cells to cure the disease,

But if Burn merely wanted to realign research priorities, he could have stayed at his old job as vice president of research for the Juvenile Diabetes Research Foundation (JDRF) in New York City. He came to Sioux Falls to run a differen t kind of cl inic, on e that, he says, "is making big bets," His team won't be doing basic research, though: "Our focus will really be on proof-of-concept studies" including clinical trials.

Burn sees a gap between research and drug development for type 1 diabetes and wants Sanford to bridge it. Based 011 the years he worked for drug companies, B(U:n argues that they have little financial incentive to '11m trials for type [ diabetes when there's a larger and far more lucrative market for the distinct problem of adult-onset, or type 2, diabetes. (Burn estimates that 90% of research dollars spent on diabetes fund type 2 work.) "Even beyond 2015, there's no cure in the pip eline" for type 1, Burn says. "It's the same old story of trying to deliver insulin in a new way."

Because Sanford does little basic research, Bum hopes that places like the Diabetes Center of Ex cellence at the University of Florida, Gainesville, or the Burnham Institute based in San Diego (and the recipient of its own $30 million gift from Sanford), will funnel promising lab results to him for clinical trials. Burn drew 011 his JDRF contacts to set up these partnerships, but the ties are loose, and no one w111 coordinate the direction of basic research among the dillies,

Bum is now interviewing for six positions to run clinical trials at Sanford, and he already has his core research team in place. The project impressed many doubters in June 2009 by unveiling a surprisingly international group, including scientists from China and Russia. (Burn, who is Swiss, has a Ph.D. from the University of Basel.) One early recruit was Alexander Rabinovitch, a gaunt Canadian forced into retirement because of his age by the University of Alberta ill Edmonton. His lab in Canada had significantly boosted the.

r

numbers of beta cells in mice by using 1:\1{0 ~

drugs already approved by the Food and Drug ~ Administration for other purposes, and he i2_ was eager to expand into hmnan trials, No one ~ seemed likely to sponsor the work, however. ~ When BUrri heard this, be immediately (3

520

29 JANUARY 2010 va L 327 SCI EN CE www.sciencernaq.orq

dipped into the $400 million and offered to fund it. It will start this March 01' April. the first trial initi ated at Sanford.

It's a 11 example of the fl ex fb.lI ity that nine figures can b 1.1 y. Sanford Health-which partners with the University of South Dakota School of Mcdicine=-ma intai IlS that its annual research budget for diabetes will approach $100 million per year. Burn hopes his scientists can win grants to cover much of that. but he can also cut through bureaucratic tape himself. "If they have a good project, I tell tbem, 'Here's the money, let's start tomorrow.' "

Bum sees advantages even in the remoteness of Sioux Falls. With a dearth of competition from other research clinics, Sanford will have its pick of diabetes patients from the nearby population of 155-,000. Sioux Falls does not support a deep scientific cornmunity like larger cities do, he admits, "but it's a paradise from the-point of view oftrying to set up a clinical trial." What's more, Sanford Castle isjust the first of20~some castles planned, Another one financed by Sanford opened in Augustin Oklahoma, and adIl1IDistratorsare scanting sites ill Belize. Ireland, and elsewhere. Burn hopes to recruit diabetes patients through each castle.

Creeping doubts

NEWSFOCUS

Big bets. Paul Burn wa nts to cu re type 1 diabetes a nd build ~ rases rch hub in Si OU)( Fa lis.

thai pig cells are not a long-term solntion-> the donor pigs must be raj sed in superstenle bays at costs of$50 per day pel' pig-but feels they are the best short-term hope (Science, 20 November 2009, p. 1049).

Weir touts regenerating beta cells from embryonic stem cells or induced pluripotent stem cells (iPS cells): adult cells that have been reprogrammed to a stemlike state. Adult cells present challenges, such as the need to remove epigenetic imprinting, Weir says, "and we're still figuring out whether there are [other] differences that would not allow them to succeed." For

Not everyone imagines a research utopia in Sioux Falls. "I think some people really share my concern that it's not an established scientific community," says Gordon Weir. a Harvard University molecular biologist who works on regenending beta cells.

Hendrik-Jan Schuurman, a diabetes researcher at th~ Ulliversiry of Minnesota, Twin Cities, is more supportive but echoes Weir's concerns: "Everybody k].10WS the university ill

ce South Dakota is not all institute i with a reputation for biomedical ~ research." That's why he thought Dnuhts, Harvard's G ord on i Sanford. was prudent to estab- Weir and others ask if the

~ Iish ties with Burnham and tile project can succeed.

,~ Florida diabetes center, places

i that can provide support and advice for San~ ford while it builds up the long-term research 6in:frastructme it will need

~ Scientists have also qt;estioned putting so ~ much money into just one option-beta-cell ,~ regeneration-when other ideas seem worth g trying. Schuurman and colleagues in New @ Zealand have developed ways to transplant pig g_ pancreatic cells into patients, by coating the ~ cells to enable them to escape detection by the 5 immune system. Schuurrnan acknowledges

these reasons, he feels that generating beta cells from embryonic stem cells (ES) is the most attractive path.

Burn argues that stem-cell therapies. adult or embryonic, are 20 01' 30 years distant, and he's impatient. Yet there's another reason Sanford will not sponsor embryonic stem cell work: South Dakota has banned it .. A local advocacy group, South Dakotans for Lifesaving Cures, calls the ban the most restrictive in the country, as it might prohibit citizens from receiving ES treatments developed elsewhere. The grotlp plans to introduce a bill into the state legislature to liftthe ban, partly to aid research, says spokesperson Nathan Peterson: "We have top-notch medical facilities, and there isn't any reason the scientisrsshouldn't have the option of conducting [RS] work in an ethical manner,' Sanford, however, refuses to takea public stand 011 lifting the ES ban.

Room to grow

While tile Sanford Project still publicly proclaims it will oversee $100 million of research a year and cure diabetes through beta-cell regeneration alone, in private Burn backs offa bit The project will spend about $30 million in 2010, and he diplomatically calls $100 million an "ambitious" number. He also says Sanford will explore areas beyond regeneration, perhaps dabbling in iPS or even immunological work.

At the same time, Burn isn't shying away from the Sanford Project's vow to cure type 1 diabetes, soon. He's overseeing a. move into a new headquarters, a 28,OOO-m:! building 011 the outskirts of Sioux Palls. Burn hopes to move 150 people into one of its two _5,600-mi research bays by this summer and fill a. second hay later. "A t universities, you're often fi ghting over every square foot." B urn notes from a snowy arid truly capacious parking lot. «Here space should be 110 issue."

As for the future, Sanford Health plans to open a 75-hecta"e research campus south of town and keep the momentum ofihe Sanford Project going after it cures type 1 diabetes. Burn believes beta-cell regeneration holds great promise for curing type 2 as well. Indeed, Burn sees no reason Sanford can't build a green oasis of research in an otherwise barren scientific state. And even skeptics of the proj ect admit that $400 millieu is a lot of green. "YOl! can create a scie 11 ti fie en v iro nmen t. in even Timb uktu,' says Harvard's Weir. "If you make it big enough and get enough people there, you cau succeed,"

-SAM K.EAN

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

521

LETTERS I BOOK.S I POLICY FORUM I EDUCATION FORUM I PERSPECTIVES

LE.TTERS

edited by Jennifer Sif/s

Tracking the Source of Glacier Misinformation

A RECENT NEWS OF THE WEEK STORY ON HIMALAYAN GLACIERS ("NO SIGN YET OF HIMALAYAN meltdown. Indian report finds," P Bagla, 13 November 2009, p. 924) highlights how inadequately reviewed material' makes its way into the. public consciousness. One source, Working GI'OlJP II (W 0- m of the Intergovernmenta I Panel on Climate Change (IPCe) [pp. 493 and 494 in (1)] reproduces several errors. The Working Group writes that "[gjlaciers in the Himalaya are r-eceding faster than in any other part of the-world" and that "the likelihood of them disappearing by the year 2035 and perhaps sooner is very high if the Earth keeps warm-

ing at tile current rate, Its total area will likely shrink from the present 500,000 to 100,000 kro2 by the year 2035," Another SOUI1::e (2) advances a no-less mistaken conjecture, not discussed in Baglas News ofthe Week story; that Himalayan glaciers are responding to the climate ofas long as 15,000 years ago.

The IPCC Fourth Assessment, particularly of the physical science basis forthe changes, is mostly accurate, but the first WG-JI sentence above derives from a World Wildlife Ptmd report (3), which cites It news story (4) about an unpublished study (5) that neither compares Himalayan glaciers with other rates 0 f recession 110r estimates a date for disappearance of Himalayan glaciers. H ima layan rates of'recession in (he WG-IT report (1) are not exceptional (6).111 the second WG-IJ sentence, "its" cannot refer to Himalayan glaciers [area. about 33,000 kll12 (7)], and may refer to the world total area of glaciers and ice caps. A bibliographic search suggests that the second WG-ll sentence is copied inaccurately from (8), in which the predicted date for shrinkage of the world total from 500,000 to 100,000 km2 is 2350, not 2035.

The claim that Himalayan glaciers may disappear by 2035 requires a 25-fold. greater loss rate from 1999 to 2035 than that estimated. for 1960 to 1999 (7). It conflicts with knowledge of glacier-climate relationships and is wrong. Nevertheless, it has captured the global imagination and has been repeated in good faith often, including recently by the ]»CC's chairman (9).

These errors could have been avoided had the norms of scientific publication, including peer review and concentration lIpan peer-reviewed work, been respected.

J. GRAHAM (OGlEY,l* JEFfREY S. KARGEL," G. KASER,3 C. J. VAN DER VEEN4 -Department of Geography, Trent University, Peterborouoh, ON K9J 7 88, Csnada, lDep~rtmeot of Hydrology ~n~ Water Resources, University of Arizona, Tucson, AZ 85721, USA. 'Institllt fiir Geoqraphie, University of lnnsbruck, A·6020 lnnsbruck, Amtria. 'D~p"rtment of G~ogr~phy, University of Kansas, Lawrence, KS 66045, USA.

~1(l whom correspondence should b~ addressed, f·mail: gcog\cy@trentu .C~

5.2.2

Relefell(e~"nd Notes

1. R. V. Cruz et cl., in Climate Change 2007: Impacts, Maplalion and Vulnerability. Contribution of Working Group II III the Fourth Assessfli en! Report of the Intergovernmental Panel lin (Iimate Change, M. l. Parry el ol., EM. (Cambridge Univ. Prass, ("mb~dg~, (007), pp. 469~506.

2 . V. K. Rai na, H in/o/ayon G/a(ler5 (0; leu ,sion Paper, Ministry of E nvi ro nment il nd for~,t" G ove rn rnen t 01 lndis. New Delhi, (009).

3. WW F, An 0 ~ef'ijew of G/a, tets, Glader lIetteal, ond Sllbsequenllmpat:/s in Nepal, India and China (WW F Nepal P,o gr~ rn, kath rna nd u, 2005).

4. F. Pearce, New Scientist, 162 (2189). 18 (1999).

5. S .. I. Hasnain, "Report on Himalayan gl.Joology: Appendix 6, unpublished. minute) 01 the July 1999 rneetinq" (Bureau of the intematicnat Commissfun for Snow imd le<;, 1999); W\'IW.cryil,pnei;csc1ences.orgldocs.tJtml.

6. WGMS, Ffll(/uations of Glaciers 2000-2005, Vol. IX (World Gl~(ier Moniloring <;~rvi(e, Zurich, 2008).

7. M, B. OYU;9 e rov, M .. F M ~ier, Occasiono! Paper 58 (Institute of Arctic and Alpine Relearch, Univ. of Colorado, Boulder, CO,2005).

8. V. M. KotlY.Jkov, Tech. Doc. Hydwl. 1, 61 (1996).

9. C. Morris. ''Why >ire the Him~[;Jyan glaoErs m~lting?" (h ltp:/Im!w,. bb c.ce.u k1Uhilsou th_" si al8 3 37 604.1 tml, 10 .. Ihls tetter was published online 20 lanu~ry 2010 (WWWCiC; ~ nc e mag. or giSt; extfele tters/ 1f129 4 9).

A Role for Postdocs in Undergraduate Education

IN HIS EDITORIAL, "GALVANIZING SCIENCE departments" (4 September 2009, p, ll8J),

C. Wieman described ongoing programs

at University of Colorado, Boulder and University of British Columbia in Vancouver that are successfully implementing new effective, research-based teaching methods in several science, technology, engineering, and mathematics (STEM) departments. As Wieman points out, transformations in the STEM teaching culture at large research universities are sorely needed, but such institutional change is notoriously difficult to bring "'i about. It is therefore worth a closer look at z;

~ how these two programs work. Their success ,.;

has been primarily due to the science ednca- ~ ti.on specialists Wieman mentions, who are ~ called Science Teaching Fellows (STFs) in. 5

~ Boulder. It may not. have been clear from the :r

Editorial that these are postdocs. Most earned ~ Ph.D.s iii their respective science disciplines ~ (110t education}, but developed strong interests 0. in pedagogy and educational research during ~

~

their training. ~

29 JA NUA RY 2010 va L 327 SCI EN CE www.sciencemag.org

One of their missions is to assist departmental faculty with the kinds of course transformation that Wieman describes" As postdocs in the discipline, they have the content knowledge required for effective development of educational materials, and. they are not threatening to faculty, as outsiders with educational degrees might be. Their ability to effect faculty change derives from their familiarity with the educational research evidence, their enthusiasm and people skills, and the assistance they can offer inimplemeuting new teaching approaches, which c~] be labor-intensive.

The- other mission for STFs is to gain science" education training, which is. uncommon within science departments and is. valuable in light of the growing number of coUege and university science departments desiring permanent Science Faculty with Education Specialties (SPES) (1, 2). These individuals are discipline-based science faculty who make scholarly work in science education part of the fabric of the science disciplines themselves. SPES are undertaking efforts in the three science education arenas of'undergradaate science education, k-12 science education, and discipline-based science education research, as well as in basic: science research (.1). furthering the current push to improve STEM education at.all levels.

The Boulder and Vancouver programs, unique. to our knowledge, should be transferable to any institution that can provide stron g, pedagogi ca Ily ill form ed leadership (preferablyfrom within STEM departments) and financial support for STFs. Funding agencies and foundations could have a major impact on improving STEM education by supporting su ch postdoctoral positions, thereby enabling the replication of these programs at other universi ties and promoting the training of more SFES.

SETH D. BUSH,' NAN.CY}'. PELAEZ.' JAMES A. RUDD,3 MICHAEL T. STEVENS,q KIMBERLY D. TANNER/ KATHY S. WllUAMS,6 WILLJAM B. woon-

1 Depa rtment of Chemistry and Biochemistry, California Polytechnic State University, San luis Obispo, CA 93407, USA. lOepartment 01 Biotogical Sciences, Purdue Universily, West lafayette, IN 47907, USA. ,J Department of Chemistry and Biochemistry, Calilorni~ Slate UniVCf'lity, Los

Angele" Los Angele" CA 90032, USA. ~ Department of Biotogical Scien(es,(,;li~ornia State Univer~ty, Stanislaus, Tu rtock, CA 95332, USA. 'Department of Biotogy and SEPAL, SM Fr<l nci $(0 State UnivNsi II', San Francisco, CA 94132, USA. "Department 01 Biology, San Diego State University, San Diego, CA 92182, USA. I Department 01 MCD Biology and Science fduc~tf(li1 I "illative, Universily of Colm ado, Boulder, CO 80309, USA,

'fo whom correspondence should be addressed. f·mal!: wood@(olorndo,edu

References

1. S. D. Bu,h et.o}, [BELlieS(i. £due. s, 297 (2006). 2.. S. D. Bu,h et ot., Science 322,1795 (2008).

Taking a Cue from the Silver Screen

IN HIS EDITORIAL "PROMOTING SCIENTIFIC standards" (1 January; p. 12), R Alberts notes that many scientific projects are carried out by large teams, which makes attributing author contributions a problem. The concept of authorship is derived from a literary tradition, but novels and poems are written by no more than one ill' two people. Accordingly, authorship preStmles that everyone makes an equal contribution to the piece. The International Committee of Medical Journal Editors guidelines all authorship, also known as the Vancouver guidelines (www.icmje.org) explicitly state that eVClY author has equal responsibility for all material ill the paper, That the new Science policies described by Alberts do not follow the Vancouver guidelines suggests that we llCed. a new model for assigning credit to scientific projects.

Films might provide a better model for assigning credit than literature. Movie productions, like large scientific projects, represent the collaborative efforts oflarge teams, often working semi-independently of each other. The credits spell out who did what-> director; cinematographer, screenwriter, and so Oil. There is no pretense that everyone who contributed to the film is an author oftbe film.

Honorary authorships are often given to principal vinvestigators who provide. resources, but minimal scientific input. StIch investigators.amanalogous to film producers, who often set up financ ing a nd handle adrni 11- istration, It is appropriate fhat this important

work receives due credit, but that credit should not imply involvement in the creative process. Stich contributions would probably not be recognized if the film industry were using, ~IS science still does, the blunt instrumont of authorship, tEN fAU LKES Departrnent 01 B,iology, The Universi ty of Texas- Pan American, Edinbur\l. TX 73S39, USA. E·mail: zl~ulhs@ utpa.edu

Give the nFair Sex" a Fair Shake

AS A LONGTIME READER OF SCIENCE, ANDTHE invited food speaker to the New York Academy of Science's series "Girls Night Out," 1 take exception to the idea that the choice of topics noudeseends to women ("Scicnce for the fair sex," Random Samples, 18 December 2009, p. 1597). When I see the statement, "Guess girls ar-e interested jJ1 science only if you can find a. link to food, love, or makeup," I see tile attitude-s-all too familiar to those of tLS whose Work crosses into social science-that nothing but cell biology and genetics constitute real science, Tbe statement suggests that work dealing with quotidian matters such as food, love. or even makeup cannot possibly be scientifically rigorous or interesting. 1 would argue instead that rigorous scientific thinking thoroughly informs my research on the influence of politics on agricultural production and consumption, particularly with respect to obesity and food safety. My lecture to the "girls" on 16 February will be much the same as the talks 1 give to mixed-gender audiences of researchers, university professors. health professionals. goverrunent officials, arid business leaders. 1 am curious to know whether social scientists arc as tired as r am of colleagues characterizing our work as insufficiently scientifi c to be taken seri onsly

MARION NESTLE

Department of Nutrition, ~ood 5tudi~, Public Health, New York universtty, N~w York, NY 10012, USA. E-mail: marion.ne,tle@nyu.~du

letters to the Editor

letter, (- 300 words) discuss material pu blished

in Science in the previa us 3 months or issues of general interest. They ca n be su bmitted through the Web (www.submit2scfence.org) or by regula r mai I (1200 New York Ave., NW, Was hing ton, DC 20005, USA). letters are not acknowledged upon receipt, nor are authors generally consu lied before pu blicstion. Whether publtshed in r ull or in part, letters are subject to edi ling for d~rity a nd space.

www.sciencemag ... org SCIENCE VOL 327 29 JANUARY 2010

523

NEUROSCIENCE

Making Sense of Printed Symbols

Charles G. Gross

Reading ill tile Brain

The Science Bod Evolution of a Human Invention

by Stanis/as Dehlfflne

Vil(i 119, New York, 2009,

400 pp, $27.95, C$35.

ISBN 9780670021109.

T hcr~ is, a, swa,tb O,f,~, orrex III the human TIght

. , tempo ral 10 he t hat is activated by the sight of words and. letters, Damage to this "letter box area" results in a specific inability to read. The central "paradox" that Stanislas Dehaeue's Reading in the Brain tries to resolve is bow

the brain, the result 0 f millions of years of evolution, can have mechanisms for reading (specifically, a parti cular area for processing words) when writing evolved only about 5400 years ago,

Dehaene, a leading cognitive scientist at tile College de France, has made substantial contributi om; to our understanding of numbersand of'reading, largely through imaging

sel ectivity over changes i 11 sti 111 ul us si ze, contrast, and exact location within the central visual field; and

often have large receptive fields (tbe area of tire ViS1Jal field that acti vates them). Pnrfhermore, their selectivity for shape can be modi-

fied (indeed programmed) by experience, and their destruction results .in a severe deficit in visual pattern recognition. Although studied in most detail in macaque monkeys, neurons with similar properties exist in the human temporal lobe as well. Thus these neurons are generally believed to underlie all types of visual pattern recognition even though they presumably evolved to recognize such things as the facial expms$.i011$ of other monkeys, food, predators, and other visual patterns in the natural world.

Dehaene seems unaware of how plastic Ute properties of inferior temporal neurons are. For example, in monkeys, they can be taught to distingui.sh pet dogs from domestic cats (3), which hardly seems the result of natural selection. Similarly in imaging studies, there are regions in the temporal lobe of human car experts that respond differentially to different models of cars (4), again hardly the result of evolution. There is no paradox (or recycling) here: Inferior tempera I neurons are generalpurpose analyzers of visual fbrms-eincluding cats, cars, words, and letters-e-that arc relevant to the individual primate because of previous experience.

Although the author's treatment of behavioral and imaging studies of reading is quite sophisticated, his presentation oftbesupposed underlying physiology is often oversimplified and sometimes misleading, For example, he claims that many inferior temporal neurons code for shapes ( 'such as t" F, Y, or 0"). In fact, neurons spec i:ficaUy responsive for these shapes over changes in size, contrast, and location have neve!" been reported. Dehaerte notes that "[mjore complex curves, shapes, fragments of objects, or even entire. objects 0.£ faces are, however, needed to trigger neurous at the higher levels" of the brain such as in fenor temporal cortex. But the only objects tJ]<tt

studies in humans. In the book, he discusses how we read, how we learn to read, and how reading can be. disordered. His account is critical, cautious, reasonably comprehensive, and accessible to laymen.

The author identifies the answer to his word-area paradox as the properties of neurons in the inferior temporal cortex, which he describes as baving been recycled for reading letters and words. These visual neurons have been studied in monkeys for over 40 years (1, 2),. 111ey are selectively responsive to complex shapes; frequently [Iiainta.in their

The reviewer is at the Department of Psy,ho\ogy and lnsti(tile 01 Netl roscienre, Printeton University, Green Hall, Pri O( eton, NJ 08 S 44, USA, [- mail: cg g ross@plincelan. ~du

524

have ever been reported to be coded by single neurons (i.e., with responses invariant across changes in size, contrast, and location) are faces and hands (2). The hundreds of thousands of in feri or temporal neurons studied by many groups of investigators provide no good evidence for.aneuron coding any other object or any letter with such invariance, Even specific faces are not coded by a. specific dedicated neuron or "grandmother cell"-a term, incidentally, introduced by Jerry Lettvin, not Horace Barlow (5). Dehaene seems to misnnderstaud the nature of'visual receptive fields. Although the average size of receptive fields increases as one moves down the ventral visual system, nus in no way imp! ies "that tIle part of the retina to which the pre ferred object must be presented for the neuron to fire doubles or triples in diameter at each step,"

These errors suggest that Dehaene does not realize that individual faces and other obj (Jets are coded by the pattern 0 ffiring over a small population of neurons 'rather than by. individual grandmother cells (5'), For example, a given inferior temporal neuron may respond to four different faces, another nellron to two of these and to a fifth face, a third neuron to a different set offaces, and so on. Thus, each specific face is coded by the pattern of activity over a set of neurons rather than by the firing of a single dedicated neuron or a set of dedicated neurons with the same properties (4). Such population coding is presumably also the way objects and letters are coded.

The book ends with a (Faustian") incantation to "the blooming of, .. a 'culture of neurollS'" that will extend neuronal recycling to aspects of human cu lture 'beyond reading:

r would like to defend the idea that

readi ng is only one of many examples of how cultural invention is constrained by ou r neuronal arch itecture. I f WI:! exten d the neuronal recycling model to other

hu man activities, we should be able to

:> o w o 5

:J:

5

Q

I;; ~

o

~

8

"

Q

Something sim ilar to Dehaeues reading ~

area paradox led Alfred Russel Wallace to a :!: major break with Charles Darwin over the ~

]}

evolution of human cognitive abilities. On 8

the basis of tile yearS Wallace spent among ~ Pacific Isl anders, he be lieved them in '110 w-ay ~ intellectually inferior to Englishmen. Yet he ~ thought their cognitive abilities in ~l1ch areas a

con nect them to their corresponding brain mechanisms .... (W]e may OM day be able to list the essential ingredients of all

human culture (including family; society,

religion, music, rut, and so on) and understand how each of them relates to

the vast array of om brain capabilities.

29 JAN UARY 2010 va L 327 SCIEN CE www.sciencem a 9 .orq

as mathematics and music could not have arisen by natural selection because these abilities were not adaptive for "savages," He resolved this paradox by attributing the evolution of hum all cognition to some spiritual force rather than natural selection. (6), aview that, caused Darwin to groan (7).

Some neuroscientists may groan over the presentation offamiliar ideas in Readingi n I he Brain as if they have been newly discovered by the author-for instance, tile ideas that brain plasticity, while considerable, is constrained by genetics; that neurons that. evolved for one function can take on another through experience (Dehaene's "neuron recycling"); andthat humans can teach themselves far beyond the capacity of other animals, Alas, even the best popular science writing often leads experts to complain that what is well known to ther .. is misleadingly presentedas fresh and original Nonetheless, for- rnany readers those details may really be. new,

References

1. C. G. Gro", D. 6. ~ende" C. E. RQ(~a·Mjrand.l, S(f~lI[e 166, 1303 (I969).

2. c. G. GrO\I, Neuropsychologia 46, 841 (Zi)OS).

3. D.]. Freedman, M Rie,enhub"" T. Pog~io, E. K. Miller.}.

NeurGsci. 23, 5235 (2003).

4. I. Go U lhi N, P. S k udlarski, J. C. GD(~. A W. A,liderson, Nat, Neurosd. 3. 191 (2000).

5. c. G. Gro,5, Neuroscientist 8., 512 ('200Z).

6,. A. R. Wall.ce, Contnbufions (0 the Theory of Natural Selection (Mil rmi Iii! n, l nod 0 n. ed. ,2, 1871).

7. lettel', C. D~rwin 10 A R. Wililace, 26 [anuary 1810, in IJI/red susse! Wallace: Leiter! ond Remfni!cence~ (Helper, NEW Yor~, ~n6), p. 200.

lO.1l26Is(;ience.118S49S

THEATER

Of Locusts and Scientists

Manfred D. Laubichler and Gitta Honegger

Throughout history, the arrival of swarms of migratory locusts has brought famine, strife, and the dissolution of order. Even today, despite intensive worldwide monitoring under the auspices of

.:5" the United Nations' Food and Agriculture.

o

ui Organization, locusts continue 10 wreak havoc :1:

.~ by destroying thousands of acres of farmland

9 in such places as Somalia, which are at the ~ brink of starvati on even without the devastat@

\i:

"" The revi ewers ~re in th e S ch 001 of Llf e Sci en ces {M. D. U. ·tll e

~ School of Theater and Film (G.H)., and the (enter for Biol~ ogy and Soriety, Arizona State University. P05t Oifire Box ~ 87285-4801, lemp~, AZ 87285" USA. Ecm~jt M~nfr~ct. 5 iaubichter@astl.edll

BROWSINGS

BOOKS ETAL

Strange Maps: An Atlas of Cartographic Curiosities. Frank Jacobs. Vjking Studio, New York. 2009. 256 pp, Paper. $30, ($37.50. ISBN 9780142005255.

This "anti-atlas" grew out of the blog (http://strangemaps. wordpress.corn) that Jacobs started "to collect cartographic cU)iO$ja" from the Intern et, It opens with a few examples of geographers' misccnceptions.such as Mercator's late-16thcentury Sepfentrionaiium terrarum, which placed a black and very high cliff at the center of a whirlpool around the North Pole. SeveraL chapters present out-of-the-ordinary geographies: unrealized poLitical proposa ls, ephemeral state" odd borders. and "bltsof national territory ad rift from the motherland." Jacobs chose some examples for their u nusual approaches, such as distinctive perspective" transformations into [iviag creatures, or distortions 01 propaganda and parody. HI.' offers cartograms (statistics masquerading as maps") and other depictions of Linguistic. cu~nary, (lIttural and economic differences. There are literary creations (such as Captain Nemo '5 Mysterious Island and Dorothy's Land of Oz), as well as images formed by jam (the Amelita,) or douds (Great Britain) or from beef (Braz.iU. Jacobs even includes a few extraterrestrial examples" such as the U.S. Geological Survey's striking representation of the effects of meteorite impacts in The Central Far Side oj the Moon (above). The author provides background details and often whimsical commentary for each of the more than 100 maps.

Heuscluecken [locusts] by Stefiln Kilegi:

scenography by Dominic Huber: music by Bo Wiget

Rimini Proto 1(011. Seh auspielha us ZOri ch, Octo her 2009. HAU Berlin, 28 Ma reh to 10 April 201 O. www. rimini-prolokoll.de/wel:rsite/en/ proie ct 4200.1 Itm I

ing eff~cts of locust swarms. It is thus highly surprising, to say the least, to find roughly 10,000 African migratory locusts tLocusta migratoria) ina 60-m2 terrarium at the center of a theater production entitled simply Heuschrecken [Locus!s].

The production could be seen at the Zurich Schauspielhaus's new performance space Schiffban this past fall, and this coming spring it will appear at the Hebbel am Ufer Theater in Berlin. Responsible for this unusual event is S te-

fan Kaegi, cofounder ofthe innovative and award-winning German-Swiss theater collective Rimini Protokoll, Kaegi and his partners HeJgard Haugand DaJ1ieI Wetzel have developed a unique theatrical approach to both highlighting and interweaving diverse interests and objectives in exploring the

"big questions" and a variety of contemporary issues and concerns. Casting "real" people-whom they call the "experts of everyday life"-rather than professional actors, the team draws from their protagonists' diverse fields of specialization, expertise, and life experiences to illuminate a given topic from a variety cf perspectives. Tbey have based previous projects ana reexamination of Das KapUa lit] a globaJ economy (Karl Marx, Das Kapital, Erster Balld); the problems of identity and adoption (Welcome to You), physics at CERN and th~ quest for the Higgs patti-

cle (Physik), and the impending change in Cairo from live muezzins to recorded calls forpmyer (Radio Muezzin.).

In Heuschrecken, the plot unfolds following the standardized routine of a laboratory notebook that details observations and measurements related to the events in the terrarinm (referred to as "the system"). An actress (Lara Korte) reads the entries from a detached position above the system and the audience. She first accounts for time and place

and Some basic parameten, (such as temperature, humidity, numberoflocusts, and number of audience members) before introducing the experts. These are Zakaria Farah, Somali-bern and Swiss-trained professor emeritus of food chemistry at the ETH Zurich; Jorg

Samietz, an expert in agriCl1J rural entomology and the ecology of IO(1)Sts; Barbara Burtscher, a physics teacher and financial adviser; the musician Bo Wiget; and the video artist Andi

A. Mldler.

For much of the performance, the terrarium is a closed system monitored by th.e observing experts from the outside, Occasiena Ily, someone ventures inside (usually the entomologist and, toward the end of the two- 11O(lT prcducti OD, the physicist in a space. sui t). The. audience, seated on risers along two sides of'the terrarium, watches the activities inside the system both directly and through video

www.science.r:n.ag.org SCIENCE VOL 327 29 JANUARY2010

525

BOOKS ETAL

projections fed by several cameras directed by the experts placed at the Sh01t sides.

As the performance unfolds, the audience, guided by the experts, observes life and death inside the system-including mating, instances of cannibalism, and the striking speed with which freshly supplied green plants are devoured. In addition, the performance contains experiments that test various hypotheses, such as how changing the temperature in different parts of the system affects the location of locusts. There is also the unexpected, in the form of "locust music," when induced movements lead insects to land on different sections of strings coupled with an amplifier. For all of these experiments and observations, the experts provide explanations of locust biology and behavior.

But Heuschrecken is much more than a docudrama Illustrating locust biology. Like all Rimini Protokoll productions, it creates its own laboratory setting by concurrently staging multiple interwoven narratives based on the life stories of the experts and their relations to the events inside the system. The main character ill this production is Farah, who connects his childhood experiences with swarms of migratory locusts in Somalia and Kenya with his subsequent career as food chemist and agricultural expert in different parts of the developing world and the problems he encountered along the way. Burtscher-whose perspective as a physicist and financial adviser brings together concerns about resource management and commodity prices, complex dynamics, and, finally, the fragility ofthe global ecosystem and humanity's precarious place in it-provides another

Making "locust music." The performance includes sounds produced by insects" landj!lg on different sections of strings connected to an amplifier,

thread. Entering the system in a space snit, she suggests simultaneously what the future of om planet might look like (the audience had previously been told bow much more robust locusts are than humans) and what other worlds we might encounter at some future time may resemble (with free associations to both science and science fiction).

So what has all of this to do with theater?

The theater is, by definition, a place for observation. Its stories are about sex, conflict, and death. It is driven by curiosity leading to discovery. Its very setting is thus experimental, suggesting a common root of science and theater, which was recognized by Aristotle, elaborated by Renaissance artists and scholars, and largely forgotten in more recent times.

The scientist as explorer of new worlds.

526

The production of Heuschrecken brings back this deep connection between science and theater. Its main [OCLlS is on observation: The audience observes (along with the experts) the life of the locusts in the system. It also observes the experts observing nature and connecting their observations to their life stories. And lastly, the audience observes itself observing the multiple dimensions of the production. These multiple layers of observation and reflection are at the core of both the scientific and theatrical experience.

This experience is simultaneously entertaining and insightful. The unfolding stories have all the elements of good drama. However, this theater production is not simply about science. It stages the drama of science, as the spectator is kept in suspense about events in nature and in the lrfe of the experts and becomes a participant in a scientific and theatrical experiment. The science and theater each illuminate the other without sacrificing scientific accuracy or theatrical imagination,

Heuschrecken thus differs remarkably g from many recent attempts to bridge the gap ~ between the "two cultures" of the sciences ~ and the humanities and arts. Countless sym- I posia have explored "the arts and science," §. and the Sloan Foundation has spent consid- 8 erable resources to bring "science" to the ~ main stage. However, many of these efforts ~ have been less about understanding science ~

~ and more a sequence of cliches about the sci- »

z

eutist as tragically mad genius (Froo!), mis- 2

chievous genius (QED), or misunderstood ~ and neglected genius (Arcadia). By risking ~ a different, experimental approach, Rimini ~ Protokoll has succeeded in bringing science g to life on stage.

'"

~

lO.1l2615(ien(e.1l87039 0

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

CLiMATE CHANGE

The Politics of Geoengineering

Jason J. ;Blackstocku* and Jane C. S. long;

Des~. ite rnOtl.llt.i.l1g.· (JV .. idel~. CC. that seve. re CIU11Atc change could emerge rap'

idly, the global reduction of carbon emissions remains alarmingly elusive (1, 2). As aresult, concerned scientists are now asking whether geoengureering=-tbe intentional, large-scale alteration oftbe climate systemmight be able to limit climate change impacts. Recent prominent reviews have emphasized that such schemes are fraught with uncertainties and potential negative effects and, thus, cannot be a substitute for comprehensive mitigation (3, 4). Butaa unabated climate change could jtself prove extremely risky, these reviews also recommend expanding geoengineering research. As streh research is considered (5-7), a process for ensuring global transparency and cooperati on iSJ1eeded.

Geoengineering schem.cs can be divided into two very different categories: carbon dioxide removal (CDR,) and solar radjatlon management (SRM) (4), CDR schemes such as direct air capture (8) or ocean fertilization (9) would remove the cause, of climate change. However, technical challenges and large uncertainties surrounding large-scale CDR deployment, along with long delays in the-elimalic response to carbon forcing, mean that it would take decades to have notable effect (4).

Conversely, SR..L'vl could substantially influence the climate in months, but with much greater uncertainty about the net effects, SRiVl schemes such as stratospheric aerosols and cloud brightening aim to cool the planet by reflecting a fraction of the incoming sunlight aWdY from Earth. "Natural experiments" ca used by volcanoes have demonstrated the rapid impact potential of SRM, and such schemes should be technically simple to deploy at low cost, relative to mitigation (3, 4). But SR.i\1 would at best unevenly ameliorate regional c! imarie change and may ha:ve serious unintended consequences. For example, SRM could PJOduce droughts and delay the recovery of the ozone layer by decades, while doing almost nothing to address ocean acidification. SR.M is thus not a substinnefor mitigation (3, 4).

Despite the limitations and risks, banning responsible SR..M research would be a mistake. The ability to rapidly influence the climate

ilnlQroation,Ji lnstitute for Applied System, Analysis, LliK~nburg, A2 361, Awslrf~. '~otre for I ntornationat Gov~rnance IMoViilioll. Waterloo; N2L 6(2. Caoada. 'LaWrence livermore National laboratoiY, Livermore, CA 94550, USA (D E-AC52"07NA27344).

'Author lor correspondence; jjb@iiasa.ac.at

means SRM might be tile only recourse should a climate crisis materialize, Moreover.the rapid impact, simple deployment, and low cost of SRM schemes make unilateral deployment a very real concern (10). More knowledge will help tIS craft good international governance and avoid rash.unilateral actions.

Until recently, SR.\\1 research was largely politically benign, as it consisted only of model studies published in the open literature (3). In contrast, emerging laboratory-based development of SRM technologies raises the prospect that. national or corporate interests might try (or appC<U" to try) to cal] 1:1'0 I Of prof t from these schemes, Such a perception would be particularly I ike Iy if SRM roscarch were framed in terms of national security, especially if the resCu1s were classified (J 1).

l?ield tests of snell technologies can exacerbate these issues. Subscale fieldexpernnents designed to have demonstrably negligible environmental and transboundary impacts, such as those' recently conducted in Russia (12), can be valuable for testing technologies and identifying the environmental risks of scaling up. But the controversy surrounding a 2009 IndoGerman ocean fertilization experiment highlights both the difficulty of proving that the risks 0 f any gecengineering field test are in fatt demonstrably negligible, and the political sensitivities such tests Cali evoke (13).

As such, nations l11USt careft.11y consider the signals that any unilateral field test sends to the iuternationa] cemrnunity If conducted without interuational approval, a test perceived (even just politically) to present transboundary risks could spark intemational tensions, creating OJ global "crisis of'Iegitimacy" (I(J). For subscale testing, we need politically-acceptable scientific standards and oversight mechanisms for ensuring demonstrably negligible impacts.

Eventually, confirming the effectiveness ofan SRM scheme would require large-scale tests with demonstrable climatic impacts-aessentially low-level deployment. But because of the complexity and variability of the climate, it. will beextremely difficult to attribute impacts and unintended consequences to any test (3, 14). 'This means liability for damages, rea] Or perceived, would become a political challenge .. For example, if the Asian or African monsoon were to weaken in a year following an SRM tcsf~at the edge of natural variability, but still causing droughts and food shodages---=tU1cm1ainTy about causatjon could 11.101 a.ccnsatlons ofrespcnsi bility.

I

Nations commencing geoengin'r;e:ring research must commit to lulL inlernationall collaboration and transparency.

As a first step to addressing some of these issues, scientists should propose intemational 110lmS and best practices for research (fO). The upcoming Asilomar conference on Climate Intervention Technologies in March 20 10 will bring together ~ 150 scientists to begin. this process, However, although necessary, such n011nS are [lot sufficient

Issues of acceptable risk for snbscale testing; if, when, and where climatic impacts testing should begin; or how SRM teclmologies Sh01:11d be managed are not just scientific. Suchquestions require a broadly acces"Sible, transparent, atld international political process. Vulnerable developing countries so far absent from SRM discussions must be engagedand all stak;eholders need to G011Sider whether existing frameworks can faciJit:ate this process, or whether new forums, treaties, and organizations are required.

Emerging national research prograrns=and even individual scientists=-urust forswear climatic impacts testing find carefully restrict subscale field-testing until approved by (1 broad legitimate international process. All SRM research should be in the public domain and should be integrated into any subsequent international research framework, Programs should include international collaboration, communicate with developing nations, and prioritize research that has global versus national benefits. These stepswill lirnit the new problems .geoengineering research heaps on an already strained global climate agenda. preserving options for future intemationa I cooperation.

R e fe rences and No.tes

1. S. Rahm510rfet ol.. Science 316. 709 (2G01).

2. c. P. McMullen. J. Jabbour. Eds., climate Chanf}e Science (ompendhlnl 2009 (UN~P, Nairobi, 2009).

3. J.]. Blackstock ft al., Climate £ngine~n·ng Responses to dimate Emergencies (Novim. Sont.; Barbora, CA, 2009); ilvdlL<lble at htlp,'larOv.orlllpdfI0907.514Q.

·4. J. Shepherd el at., Geoengineen·n{j the Climate: 5tfente, Goverrwl)(e and Uncertainly (The. Royol Society, london, 2009),

5. US (on!) ressional a n d u. K. p. rlia nlfll lary hea ri ng 5 on geo"nginee~i1g have recently been tonvened: see

sd e II ce.h ouse ,Qovlp rf5slP !lArn de. o,pxl N ew~ I 0=2 61' 6. 6·. In Z009, the Euwpean Uni()n began the IMPure g~oeng1neering 1"'E'of(n pmje~t (nttp;lliI)ipljcun'rQw.del).

7. The UX R",eoTCr.CounciLs &1efgy Pmglo:rhme

""nou" {eo sup port 101 ;,~ g~oengi nee ri n 9 re,.;a rrh: see www.;,p" c.~ ("uk/C:on !~ntllllew'/g eoe n gi ne Elin g. him. a D. w. hit~. S[j~nce 325. 1654 (2()09).

9. K. O. Bue'iSeler et ol .• :iri~n[e 319; ~62 (2008).

10. D. G. Vi(l()r e/ 01 .• Foreign A/f. 88. 6:4 (2009).

H. E. Kinti~{h. Scisnmlnlider (2009), blog5.lcien(~mag.org!

sc1 B nceins i derl200910 31eKdusiv e-rnl Ilt.h tm!.

12. Y. A. lzrael tl 01 .• Russ. Me/eorol. Hydtol. 34. 265 (2009).

13. Editori.I .• Nature Geosd. 2. 153 (20'09).

14. A. Roback,Science 321,530' (2010):

10 .. 11261>(ieM~.11S302 9

www.sciencernaq.orq SCIENCE VOL 327 29 JANUARY2010

527

IMMUNOLOGY

Arsonists in Rheumatoid Arthritis

Guy A. Zimmerman and Andrew S. Weyrich

PlI:.t.ele.ts a:e., SJII.al .. I.an.,l1Cleate .. c. e.'lls.' that CIrculate m the blood of mammals and

are critical effectors of hemostasis, blood clotting, and wound repair. Rheumatoid arthritis is a chronic inflammatory disease of humans that involves the innate and adaptive limbs of the host immune system, and is thought to be autoimmune in nature, Traditionally, platelets and rheumatoid arthritis don't go together. On page 5&0 of this issue, Boilard et al. (1) report. that they domicroparticles released by platelets may be incendiary devices in the conflagration of a bot, swollen, and painful rhenrnatoidjoint.

Although rheumatoid arthritis affects severa] tissues, destructive joint inflammatiou is a central feature (see tile figure). The normal knee is a synovial joint that encloses a space containing .1;1 clear, viscous, largely acellular fluid filtrate of plasma and is bordered by synovium, a tissue consisting of lining cells, stromal matrix molecules, and blood vessels. Boilard et al. have discovered that platelet microparticles-c-vesicles shed by activated platelets (2}--are present in knee joint fluid in

Department of Medicine, University of Utah; Sat! Lake City,. UT 64112, USA. E-mili!: guy.rimmerman@hmbg.utah.edu,

rheumatoid arthritis and other inflammatory arthritic conditions. The authors show, using amouse model , tbatplatelets are critical in the development of inflammatory arthritis. Activation of glycoprotein VI, a platelet-specific receptor for collagen, induces microparticle shedding. In addition, fibroblast-like cells that line the synovial Cavity of the joint can also trigger microparticle release. Because these fibroblast-like synoviocytes and collagen are present in the inflamed synovium, platelet interactions in this milieu could lead to local release of microparticles and their translocation into the joint space.

Boilard et al, also determined that rnicroparticles from the joint fluid of patients with rheumatoid arthritis can reciprocally activate fibroblast-like synoviocytes, and this interaction induces synoviocytes to secrete inflammatory chsmokines and cytokines, Interieukin-l-c-a pleiotropic cytokine that is rapidly synthesized by activated human platelets (3) and is packaged into microparticles (1)~accOlmted for much of this stimu!atory activity .. Thus, a vicious cycle ensues:

Fibroblast-like synoviocytes induce formation of platelet-derived microparticles, The microparticles then deliver interleukin-L,

Inflammation associated with rheumatoid

a rth ritis is stoked by plate lets, expandi ng thei r roles into the immune system.

which triggers synoviocytes to synthesize other cytokines and chemokines, some of which attract polymorphonuclear leukocytes. and thereby fan the fire of inflammation,

A role for platelets has previously been described in inflammatory joint diseases, yet this is not commonly mentioned in summaries of their pathogenesis (4, 5). It's worth noting that platelets are previous suspects in rheumatoid arthritis: Theyaccumulate in the joints of affected patients; platelet thrombi were observed lTi synovial vessels of patients with. rheumatoid arthritis; and increased numbers of platelets in synovial fluid and ofmieropartides in blood are associated with the condition (6, 7-9). Why the lack of mechanistic study? Platelets have generally been viewed as cells with limited short-term hemostatic activities that exclusively take place in the vascular compartment. Recently discovered "new biology" of platelets is changing these perceptions, however. For example, they have an intricate transcriptome (entire repertoire of RNAs produced), activation-dependent posttranscriptional pathways, influences on extravascular events, and, perhaps, longer life spans than previously appreciated (10). Furthermore, there is a wealth of information

'" u

i

;j

~

and synoviotytes may trigger the release of rnirropa rtides fro m platelets. These '"

microparticles then enter the jOlllt space a nd further amplify inflam mation by PfO~ ~

j.!-; ducing interleukin-L This then activates synoviocytes, thus supporting a tyde of !;}

inflammation. :J

Perpetrators in a painful [oint, In rheurnatoi d a rthri tis, the joi nt 5ynovi urn becomes inflamed with dilated blood vessel, and immune cells. Polymorphonuclear leu korytes (PMN), cytoki nes, che mokines, and other mediators of inflarnmarion, accurn ulate in the synovial flu id. Local activation of platelets by collagen

528

29 JAN UARY 2010 VO L 327 SCI EN CE www.scie ncerna 9 .or9

indicating that plate I ets have inflammatory functions and a substantial arsenal of factors that confer the ability to signal to monccytes, dendritic cells, and. other immune effector celts. There is also evidence that platelet rnicroparticles activate adaptive immune cells in specific tissue compartments in response to cues thai nigger antibody synthesis and alter lymphocyte activities (11).

A perplexing issue is that Boilard et al. did not detect iniac t platelets in the synovial fluid of patients with rheumatoid arthritis, although other investigators have found them (7, 8), How, then, did the incendiary micropartic les gain access to the j oint space? One possibility is that microparticles associate with transmigrating leukocytes and are then carried into the synovial fluid, a mechanism supported by microscopic and dow cytornetric observations of Boilard et al.

This could happen in synovial blood vessels, where the numbers of microparticles may be much higher than reported in peripheral blood (9). It could also occur in extravascular synovial tissue if release ofrnicroparti des occurs as a res nit of platelet interactions with collagen and/or fibroblast-like synoviocytes in this compartment (1). Platelets adhere to activated polymorphonuclear leukocytes and monocytes in tbe circulation in many inflammatory conditions (12), including rheumatoid arthritis (13). Thus, leukocytes may deliver platelets and/or microparticles to extravascular sites by this, 01' other, mechanisms (14). Polymorphonuclear leukocytes may well be accomplices that deliver microparticles to the rheumatoid arthritis joint space and, conceivably, facilitate their local formation ill transit through the synoviurn,

PERSPECTIVES

References and Notes

1. [. Boilard I!I at, Science 3Z1, 5BO (2010).

2. S. Pere,· Pujol et ol., (y/omelry A 71, 38 (2007).

3. M_ M. Deni, et 01., (el( 122, 379 (2005).

4_ A. E. Ro,enberg. ·i n Robbjn, and CWan Pillholagi, Ba~j, of Di5ea,e, V. Kumar, A- K. Abba', N. F"v,~o,). C. A,ter, Eds (S~u nders, Phi la delphi~, ed. 8, 2010) i pp. 1205-:1,256

5. E. H. (hoy, G. s. Pan~yi, N. fngl.]. M~d 344. 907 (2001),

6. F. Di~z"Gof!1>1lez, M H. Gi",b~rg, in Kelly's Textbook qf Rheumolol09Y, E. Q. IWrl.1 Jr. 1[1 01 .. Ed,. (El>evier Sauo" ders, Philadelphia, 2005), vol. 1, pp. 252-259.

7 C. M. Herd. C. P. Page, in ImmullDpharmorology ot Platr!" leis, /III. [oseph, Ed. (Academic Pr5', London, 1995).

8. M. Yawn, M. Ojaldetti, ArthritIs Rheum. 21, 607 (1978).

9. E. A.]. Kl1ijfi·Dutmer et 01., Arthrifis Rheum. 46, 1498 (2002).

10. G. A, Zimmerman, A. 5. Weyrlch, AI1:eriosc/er, rhromb, Vasco Bioi. 28, ,17 (Z008).

11. D, L Sprague et oi., BioarJ :tU, 5028 (ZOOS).

12 A. 5. WeyMch, G. A. Zimmermon, rrendslmmuno/. 25,

489 (2Q04).

13 J E. Joseph et 0/., Br. j. Ha~matol. 115,.451 (2001)

14. T. Wej,smuller et 01.,). riil). Invert. UB, 3682 (2008).

15. We thank D. Lim for ,mi51i1me with the figure.

lO.11261Itien~e.1185a69

MATERIALS SCIENCE

Bundling with X-rays

Cyrus R Safinya1 and Youli lP

More than a century after its discovely, x-ray irradiation continues to profoundly impact a wide range of fields, from screening at airports to wholebody imaging diagnostics in health care (1). In the natural sciences, x-ray crystallography has clarified how the shapes of proteins and related complexes relate to their cellular function, and x-ray scattering has elucidated the structure and dynamics, mechanical properties, and intermolecular interactions of countless materials (2-5). OIl page 555 of this issue, Cui et (II. (6) report a new twist ill the application of x-ray scattering, where synchrotron x-ray irradiation, in addition to its usual role in probing structure, acts as a reversible switch for self-assembly from 11. disordered to all ordered state of bundled filaments (see the figure).

Bundling offilamentous proteins is widely observed in the eukaryotic cell cytoskeleton (7). In. cells, bundles are assembled and disassembled to enable a wide range of functions. For example, filamentous actin can assembl e with associ ating proteins into stress fibers ending in focal adhesion spots associated with cell adhesion, or form dynamically

lMaterial.5, Physics, and Molecular, Cellular, and Developmental Biology Departm~nts, University 01 Calflornj~, Santa Barbara, CA 93106, USA. 'Mat~rials Researrh taboratory, Univmity 01 California, Sonta Barba ra, CA 93106, USA. E-mail: s~liny~@mrI.U(5b.edu; youli@mrLucsb

active bundles within membrane-protruding filopodia in cell crawling. Another example is microtubule bundles, which are important in the development and extension of axons in neurons (8). The bundles are stabilized by microtubule-associated proteins and possibly other factors, whichnoncovalently cross-link neighboring microtubules.

Oosawa was the first theorist to develop a model explaining how nanofibers carrying the same charge may form bundles (9).

High-brilliance synchrotron rediationis used to trigger a reversible transition from randomly oriented filaments to hexagonal bundles.

The model, which applies generally to both biological and nonbiological fibers, is connterintuitive, because similarly charged fibers normally repel each other due to electrostatic forces. Oosawa showed how fluctuations in the bound counterions of neighboring filaments become correlated and prod.uce an attractive force (si 111i lar to an effective van derWaals attraction), which leads to bundles, More recent theories also show the possibility ofbundle formation through salt-bridge-

)'.-rayon

X-ray off

.x-ray irradiation switch. I n (Ill et 0('5 experi ment, transmission 01 a hig h-bri IUa nee synchrotron 'beam through the sample charges up peptide-based nanorods .. Asa result, the randomly oriented filaments transition to two"dimen5lonal ionit crystals of hexagonal bu nd les, The filament bu ndles revert to the disordered state after the bes m is tu rned off.

www.sciencernaq.orq SCIENCE VOL 327 29 JAN UARY 2010

529

PERSPECTIVES

like, exponentially decaying attractive forces, between neighboring filaments (9).

Most experimental data on nonspecific bundling interactions appear to be consistent with theoretical predictions of densely packed bundles resulting frOn1 counterion-induced attractive forces, but substantia'! discrepancies remain. Microtubules call form various bundling architectures, from tight hexagonal bundles to loose two-dimensional necklacelike morphologies with linear, branched, and loop morphologies (10) that are not predicted by theory

In contrast to the filament bundling described so far, which typically arises from attractive forces, long-range electrostatic repulsion aptJears to play the dominant role in inducing tb.e formation of widely spaced, stable hexagonal filament bundles reported by Cujel{!l. The authors hypothesize that x·ray irra diati 0[1 induces a reversible chemiCOIL reaction, with deprotonation of carboxyl groups on. glutamic acid residues leading to highly charged film.ncuts (see the figure). Their observation that the indllced ordered phase occurs only above a certain x-ray dose I'<tte rather than accumulated dose is consis-

tent with a reversible switching process (see the figure). The largeequilibrium spacing observed by the authors seems to result from repulsive filaments in a confined geometry, reminiscent of a two-dimensional Wignet crysta 1 [wh ere 1111n i rnization 0 f p otenti a 1 energy at low concentration leads to a twodimensional crystal (3)].

Radiation-induced structural changes are usually detrimental .. The term "radiation damage" is widely used to describe the resulting structural degradation, What is unusual and interesting in Cui et tll.'S study is that x-ray irradiation induces rather than destroys ordering. The system seems to be highly susceptible to hexagonal order due to the built-in e leetrostatie repulsive force; above a eri tic ,,1 concentration, bundles form spontaneously without x-ray radiation, By increasing the charge of'the filaments. irradiation tips the transition point to much lower concentrations, Where spontaneous bundl ing does not occur,

Purther studies are needed to clarify the detailed x-ray-inducedionization process responsible for the ordering observed by Cui et al. Nevertheless, the x-ray ·switcli Introduced by this study opens up entirely new

directions in nanoscale assembly. VVc expect that future work will extend the discovcry to other systems, such as other peptidebased geometric shapes. including sheets and SlJ heres , Other new directions may involve using grazing-incidence x-ray irradiation for the controlled growth of ultrathin ordered phases at interfaces.

.Rl'ferl'nces

1. S. G. Benko, G. B. lubkin, pnys. Today 48, 23 (1995). 2_ J. Als·l\leilsen, G. M~teriik, P/iys" today 4.8, M (1995).

3. 1'. M. Cha iki n, r.r, luiJemky, Printiples oj Condensed Matter Ph}"iirr (Camb lid ge un iv. Pr ~S5, (a m bri dg e, 1995).

4, P. G. De Geooel,j. Prost, fhe PhySiCS o!UQufdCrysto/s to~fQ rd Unfv. Pre», Oxford, ed, 2,. 1993 i.

5. C. R. 5afjj1ya, in rh~ N~w Physics/or the Twenty nrst Ce~tury, G. F'a.let, Ed. (Colmbrldge Ulliv. Prell,

Col m bridg E, 2006).

f" H. Cui et al., Science 327, 555. (2010); published online 17 D"(~mher 2009 uo 1126Isden(~,:I..l~2' 40),

7, [) Bf~Y, CeJrMov~m~"I$' From MolEcUle!;. to Motility (Gorllmd, N ew Yo(~, ed.2, 2M 1).

ll, E. R. Kilndeil{t al., Principles of Neural Scien[e (McGill"" Hill, 1\1 eW Yo rk, S i nga pore, e d. 4, ;!DO 0),

9. W. M. Gelb.rt, R. F. Bruinsma, P. A. Pincus, V. A. p<lrse· gian, pnys. Today 53, 38 (2i)OO).

10 D, J. Nl:edleman el 0/0, Proc. Natl.Acad, sd U.S.A 101, 16099 (20M).

ATMOSPHERIC SCIENCE

1O.11261Icience.1185868

A Test for Geoengineering?

Alan Robock,l Martin Bunzl} Ben Kravitz,l Georgiv L Ste nell ikov'l

Scientific and political interest 111 the possibility of geoengineering the climate is rising (l). There. arc currently no means of i rnplementi ng geoenginecring,

but if 11 viable technology is produced in the next decade. how could it be tested? We argue that geoengineering cannotbe tested without furl-scale implementation. The initial production of aerosol droplets can be tested on a small scare, but 110W they will grow in size (which determines the injection rate needed to produce a particular cooling) can Ol1Jy be tested by injection into an existing aerosol cloud, which cannot. be confined to one Iocation. Furthermore, weather and. climate variability preclude observation of the climate response without a' large, decade-long. forcing. Such full-scale implementation could

'Dep~rtm~llt 01 Environmental Science" Rutgers University, 14 College Farm Road, N~w BrUll5wick, NJ 089.01, USA. 'D ep <1 rtm ent of Philosophy, Ru 192 rs U n jv~rsi!y, 191 Ryd ers La ne, (II ~W B r u IlswitK N J .089.0 1, USA. 'Di\li ~ on 01 PhYsical 5 dences ~ n d Engi n~ering, 4700 Ki ng Abd ullil h University of Science ¥ld T~,~no\Qgy, Thuwi)1 2395.5,6900, Kingdom or Sa udi Arabia. E- maih robod(@ehvsci .ru t9 er I J:'d u

530

disrupt food production on a large scale,

We usc the term "geoenginecring'tto refer to solar radiationmanagemerufSk M), partieularly the injection of aeroso Is Into the stratosphere to emulate volcanic emissions. We consider the best case for conducting experiments in the atmosphere. putting aside. some of the worst-ease reservations tim! have been raised about the atmospheric risks of geoengineering (1., 3).

If ongoing climate modeling and limited experiments to test insertion methodology were to indicate that SR.M would reverse many negative aspects of global warming, could these results be validated. with in situ experiments to test the creation of a stratospheric aerosol cloud and the resulting climate response? Some authors have argued that the effects of polar testing could be confined to the Aldie (4). However, we have shown (5), on the basis of analogs from past volcanio eruptions and climate mode! experiments. that Arctic injection would cool the atmosphere down to latitude 30cN" weakening the summer mOTiSOOn over A frica and Asi a

Stratospheri c geoengi naeri ng can not be testen i n th~ atm espnere without fu tt-sea le implementation.

and reducing precipitation, just like tropical injections of stratospheric aerosols, Indeed, any high-latitude sulfate aerosol production would affect large parts of the planet.

Even if insertion does indeed have to end IIp as planetwide, it rn ight be tho light that one could at least proceed at row rates of insertion and look for any untoward side effects before increasing the dose, But two major issues prevent lise ful testing of stratospheric aerosol injection with small amounts.

First, to produce an aerosol cloud of sufficient thickness that lasts long enough to detectably cool Earth's surface, regular injectious would be needed into air that already contains. an aerosol cloud. One can fly aircraft or balloons into the stratosphere andtest nozzles and injection of material into the wake of' the planes (see the figure) (6) and thereby measure the creation of aerosols in the fir$i minutes or hourS into it pristine stratosphere. However, current theory tells us that continued rnniSsioIl of sulfur gases or sulfate particles would cause existing particles to grow to larger sizes, largerthau vo lcanic eruptions typ~

29 JANUARY 2010 VOL 327 SCIENCE www.sciencernaq.orq

PERSPECTIVES

ically produce. These larger particles would be less effective at cooling Earth, requiring even more injections (7). Such effects could no] be tested except at full scale.

Second, the signal of sma 1J injecti ons would be indistinguishable from the noise of weather and climate variations. The only way to separate the signal from noise is to get a large signal from a large forcing, maintained for a substantial period. Different model simulations [e.g .• (5)] have SJ10WIl that injection of 5 Tg (5 x l012 &) of SOl into the tropical lower stratosphere every year-s-the equivalent of one 1991 Mount Pinatubo emption every 4 years=-could lower global average surface air temperature. but African and Asian summer precipitation would also be reduced, potentially affecting the water and. food supplies of more than 2 billion people. If much less S02 were injected, any potential effect on the monsoon would be indistinguishable from climate noise.

Volcanic eruptions serve a~ an excellent natural example of this. In 1991, the Mount Pinatubo vo lcano injected 2() Tg of 502 into the stratosphere (8). The. planet cooled by -O.5"C in 1992 and then warmed back IIp gradually as the volcanic. cloud fell out of the atmosphere over the next year or so, There was a farge reduction of the Asian 111011S0011 iii the summer of 1992 (9) and a measurable ozone depletion in the stratosphere (10). The eruptions of the Kasatochi volcano in 2008 (l.5 T g of SO) and the Sarychev volcano in 2009 (estimated 2 Tg of S02) did not produce a climate response fhal. could be measured against the noise 0 f chaotic weather variability.

Climate model simulations suggest that the equivalent of one Pinatubo every 4 years would be required to" counteract global warming for dl~ next few decades. The cloud would have to be maintained in file stratosphere to allow (he cHmate system to cool in response, unlike for the Pinatubo case, when the cloud

tell out oftbe atmosphere before the climate system could react f\JI1y, Such an experirnentwould essentially be: implementation of geoeugineering. No ma.tter what the results, it would be difficult to stop such an experiment q 1I i c lily. Firs t, alJ mod el sim ulatio us co 11- ducted so far, starting .vith Wigley (J 1), show tbat upon cessation of geoengineering, the climate: would wa rm much more. rapidly than if no geoengineering bad been conducted. This rapid warming would be much more disruptive than the gradual change we are experiencing now. Second, the geoengineering infrastructure, includ ing different industrial interests involving many jobs, would lobby to keep the program going.

Furthermore, no stratospheric aerosol observing system exists to monitor the effects of any in situ testing. After the 1991 PiORtubo eruption, data from the Stratospheric Aerosol and Gas Experiment n (SAGE IT) instrurnent on the Earth Radiation Budget Satellite showed how the aerosols spread. A limb-scanning design such as that of SAGE n is optimal fOJ" measuring the vertical distribution of aerosols. SAGE IJI Hew from 2()02 to 2006, but there are no plans for a follow-on mission. A spare SAGE lIT sits all a shelf at a NASA Jab and could be used now. The only limb-scanner currently in orbit, the Optical Spectrograph and Infrared Imaging System (OSIRIS) on the Odin satellite, is not used to regularly monitor stratospheric aerosols. These current and past successes could serve asa model for a robust stratospheric observing system; which could also be used to measure the effects of episodic volcanic eruptions.

Finally, loca] impacts are: particularly difficult to predict. Modeling to date has raised concerns that large-scale sul fUT insertion might produce lUlfoward local elimate responses affecting both ternperature and moisture. G I oba 1 clirnate n1 odel S SUTl-

Potential techniques for pwducing aerosols in the- stratosphere. The effects of creati n9 aerosol, in a pristine stratosphere can be tested, but lonqerterm effetts as the particles groW to larger sizes are difficult to predict and attribute.

ulate monsoon circulation at a fairly large scale. The more local the interest, the Iongeran experiment would have to run to rule out adverse side effects. In a 1 O-year experiment to test for a climate signal over noise, the chance of a local adverse response could not be ruled out prior to the experiment. As such, a prudently designed experiment would have to make provision for such outcomes. Although even a major disruption of agricultural output would be difficult to attribute to geoengincering, wert:' such outcomes to occur, necessitating an end to the experiment, the sulfate aerosol density would need. to be decreased slowly to avoid ecological shocks. All these issues will need to be considered in policy and governance deliberations (1).

References and Notes

1. J.]. Blad'SlIl[~. J C S Lo~g, SCience 327, 527 (2010)

2. G. C. Heg~rl, S Solomon, 5qef),e 325, 955 (2009); pub!bh~d online 6 Augu;t 2009 (lQ.1126Mien,e. 1178530).

3. A. RobQck, Bull. AI. Sci. 64, 14 (2008).

4. K. (aldeir~, L. Wood, philos. Tran$. R. SOt. London Set. A 366, 4039 «ODS).

s. A. Robock, L. Orna n, G L. Stench ikov, J Geophy5 Re5. 113, D16101 (2008).

6. A. Robo(k, A B. Milrquardt, B. Kr;lVjll:, G. Sten(hfkov, GeophYI. Res. Lett. 36, L19703 (~009k

7. f'. Heckendorn Bt ai., Elll.Ifron. Res. Left. 4, 045108 (2009).

S_ G. J. S_ Bluth, S. D. Doiron, A]. Krveger, L. s. Wal\~r, C ( Srh net' I.,. Geaph Y5_ Res, Lett. :L 9., 151 (1992)

9 1(. E_ Tren~~rth, A Dot Geophys. Res. Lett: 34, U570.l W)07}.

10. S. Solomon, Rev. Geophys. 37, 275 U 999).

11. I. M.l. Wigley, Science 3iQ, 'lS2 (2006); publish~d

onl i Of 14 Se ptem ber .2 006 (10.11 Z 6/sti en ce, 1 Ul71B}.

12. Supporled by NSF grant ATM-on0452.

10.1126Isd~n(~. 11-S623 7

www.sciencemag.org SCIENCE VOL 327 29 JAN UARY 201 0

531

PERSPECTIVES

ANTHROPOLOGY

Apes Among the Tangled Branches of Human Origins

T!!rry Harrison

Tbe detailed description of ArdipitheCliS ramidus (1) more than lived up to the buzz of anticipation that preceded it in the paleoanthropological community. A. ramidus is a purported hominio (the group comprising humans and their extinct relatives after they diverged from our closest living relatives, the chimpanzees) from the Middle Awash region of Ethiopia. The focus of attention has been on bow A. ramidus may r~lllte to later fossil horrunins and to living apesand humans (see the first figure), but to appreciate the place: of A_ ramidus. in human origins, we must also view it from the perspective of the hominoids (apes) that lived in the Miocene, 23 to 5 million years ago (see the second figure).

A. ramidus ts known from more than tOO specimens, including a remarkably preserved partial skeleton, that date back to 4.4 million years ago (2). Several other hominin contenders are known from the late Miocene (1 to 5 million years ago), including Ardipithe('11S kadabba, Orrorin tugenensis, and Sahelanthropus tchadensis, but our knowledge of

Center for the Stu dy 01 Hum an Origins, Departm ent of Anthropology, New York Unjv~rsiW, New York, NY 100Q3, USA. E-mail: lerry.harrison@nY~I.edu

Cercepjt~eco_idea ((ercopithe"oid~ or Old WorLd mo~keys)

Catarrhini (rafarrhines)

Hornlncidea (homiooi05 or apes)

their anatomy is much less complete. Not till paleoanthropolcgists (including this author) accept. thatA. ramidus is a hominin or agree with the evolutionary and. paleobiological interpretations that have be~n proposed (2), but there is no doubt about its critical importance for understanding human origins (3, 4). The unveiling of A. ramidus has required. a major rethinking of what the last common ancestor of humans and ch impanzees looked like and which initial evolutionary steps may have characterized the earliest hominies. A. /"(Indd!ls also helps to close the gap between the last common ancestor of humans and chimpanzees (estimated at 7 to 5 million years- ago) and the earliest undoubted hominin, Austmlopithecns anamensis (4.2milliou years ago) (5).

Duringfhe early Miocene (23 to 16 million years ago), the precursors ofhominoids-s-the proconsuloids=-were a remarkably diverse group of catarrhine primates (the gi"D"I.IP comprising Old World monkeys and apes., see the first figure) restricted to the tropical forests and woodlands of Afiica and the Arabian Peninsula (6). Between 17 and 14 million years ago, environments in Africa became drier and increasingly more seasonal. Proconsuloid diversity declined, and cercopith-

The evolution of apes between 23 and '5 million years ago set the scene for the emergence ofthe first hominins in Africa,

ecoids (Old World monkeys) and early homiuoids, such as Kenyapithecus, Equatortus, and Nacholapuheous, bee-arne the dominant taxa (see the second figure). These hominoids and other catarrhines responded. to increased seasonality by developing dietary.adaptations fOT eating leaves or fOT processing hard food items, and by developing a range, of specialized locomotor behaviors (~8).

Abo-ut l 6 to 15 mi Ilion years ago, apes expanded their geographic range out of Africa to colonize much of Eurasia, This influx of hominoids into Eurasia coincided withthe middle Miocene climatic optimum, fa phase of global warming that allowed tropical and subtropical mammals to extend theiJ' ranges northward. The earliest Eurasian apes, Griphopithecus and Kenyapithecus, are known from sites in Turkey and central Europe, Like their African contemporaries, they had thickenameled molars and robust jaws, adaptations for exploiting a broad spectrum of seasonally available foods.

Between 13 and 9 million years ago, hominoid diversity in western and central Europe increased to include Pierolapithecus, Anoiapithecus, and at least four species ofDIJ'opitheC1JS (9, 10). Pierolapithecus and Anoiapuhecus from Spain are probably stem hominids

Gorilla (gorillas)

.

• Homininae (h()minill<!s)

Pan

(chirn pallzee5')

Hominidae (horninids)

• fjominini (hcmlmns)

Ponginae (pon\ljne~)

Pongo (oranqutans)

Hylobatidae ---------------(hylobatids)

A classification of Ihe extant catarthines (Old Wo~ld monkeys and apes).

532.

Hylabates

(giboolls and ~iam~ng)

29 JAN UARY 2010 va L 327 SCIEN CE www.sciencemag.org

(great a pes and h tnn ans, see the first figure), Dryopithecus has been inferred to be <1 stern hominid. an early member of the orangutan lineage, or a stem hominine (African great. apes and humans, see the first figure), but. the first of these options is the most plausible,

A diversity ofhominoids also occurred ill Asia during the middle and late Miocene, extending from Indo -Pakistan to Thailand. Of these, Ankurapithecus, Sivopithecus, Lt!fengpithecus, Kharotpithecus, and Gigal'ltopilhecus are all likely to be closely related to the extant ora ngutan {ill.

Gradual cooling during the middle Miocene led. to greater seasonali ty in western and central Europe and a shift. from subtropical evergreen forests to predominantly deciduous broadleaved woodlands, This shift was accompanied by a dramatic turnover of the mammalian fauna at 9.6 million years ago, termed the Mid-Vallesian Crisis, when most horninoids became extinct (1]). The highly specialized stem hominid Oreopithecus survived on European island refugia until 6 to 7 million years ago. In southeast Europe and southwest Asia, hominoids specialized for dry open woodlands, including Ouranopithecus and Udabnopithecus, survived well into the late Miocene (10 to 7 million years ago). Oura nopuhecus probably offers the best evidence of an early hominine in Eurasia, which implies thatAfrican great apes extended their range from Africa into southeast Europe and southwest Asia about 10 million years ago.

About 7 to 8 million years ago, upfift of the Tibetan Plateau and increased intensity of the Asian rnenscon, together with the global expansion of C4 grasses, led to a further dec] ine in the diversity of'Eurasian horrrinoids, By 5 mi Ilion years ago, hominoids had "become extinct throughout Eurasia, except for those surviving in the present-day range of Asian hcminoids (orangutans and hylobands, see the first figure), extending from southern China to Southeast Asia.

In Africa, the fossil record for hominoids between 13 and 7 million years ago is r clatively sparse. This has led some authors to postulate that the homin.ines initially diverged in Eurasia before migrating back into Africa (13, 14). However, recent discoveries and a growing appreciation of later Miocene hominoid diversity in Afri ca make this an untenable scenario. The recently described l O-millionyear-old Nakalipithecus from Kenya is closely related to Ouranopithecus but is older and has more primitive teeth, implying that these taxa shared .3 last common ancestor in Africa V 5). It has been suggested fhat Samburupithecus (9.S million years old) from Kenya and Chororapuhecus (10 to 10.5mjUjonYe'dJsold)

PERSPECTIVES

18

Millien years age

1.2 H) 8

4

'I

16

6

[(en yapilhecus Griphopi1hecUs

OrftQpithetus

i

i Homo

UdabnopltheCl15" I

.,.. !: AU5(mlopfthew~

OuronppifhetlJs 'i

~":-·"h.~_u • i

'. '··7 :,

'I' ".~., 7 ..... 3"7.... 5aneianthrOPU5

· ..... Nakalfpilhrcu5 ',h

...... .. r·~'··1 afro/in

l(Mrorapilhecus \,

•.•• v, Ardipithews

Saf))bl)rupnh!!cu~

Proconsliloids

Khoralpnhrc:u~

? ----'

Gigantopithec:us

Pongo

Hylobatiol

Ste 01 homi ~ld s

E~t~nl Africiln .great ~p~>

H 0011 n i ns POYlgin ~5

Slemhominrne> or5tem homlnins _ Hylobiltidl

Hominoid relationships. Schematic representation of the interred evolutionary relationships between Miotene apes, early h omini ns, ·a nd exta nt homi noids, Solid 9 ray bars represent the known ti me fa nge of each genus, thin dark lines are inferred reletinnships between the genera, and thin Bashed lines wHh "7:" denote uncertai I) relationships.

from Ethiopia are related to gorillas, but the evidence is slim .. and they are probably stem hominines (16). Further recent fossil finds confirm that in the middle and late Miocene. Africa WdS populated by a multitude of hominoids, but the current material is too scanty to designate additional species (J 5,17, 18).

As paleontological exploration intensifies across Africa, our knowledge of hominoids in this critical time period will steadily grow. Rather than just a few relictnal evolutionary strands surviving to the end of the Miocene and giving rise to modem hominine lineages. as was previously thought ape: diversity in Africa during the late Miocene looks very bnshlike, The relationships betweeIiArdlpithecus and earlier hominids will remain enigmatic until the quality oftJ1e fossil evidence

from the late Miocene of Africa improves, but. this wi II eventua Ily prove critical in resolvi ng its affinities to Inter horninins, The important questions then become: Where did ArdipitheClIS and the other early hominin contenders come from? Are. they truly members of the hominin lineage, or simply apes among the tangled branches that constitute the basal hominine bush?

Referenc.es

L A. Gibbon" Science 3Z6, 36 (2009/.

2 1. Whi1e et 0/., Science 316 .• 64 (2009).

3. A. Gibbon,. Science 326; 1598 (2009).

4. P. Andr~ws, 1. Harrison. rn Inlerpret(ng·the POlil: EW.lYS on Human, Primate, and Mammal Evolution, D. E. lie· berman, R. l. Sltlirll, J .. l<ell~Yi Ed,. (Bnll, B~St{)n, 2005), pp.103-121.

5. 5. Kum~ r el al .• Pi"oc. NatL Acad. S( i. tl.S A. 1 G 2, 18842 (.2005).

www.sciencemag.org SCIENCE VOL 327 29 JAN UARY 2010

533

PERS PECTIVES

6. 1 Hamson, in rile Primate fo."il Record, W. C. Hartwig, Ed. (Cambfidg~ Vniv. Pre:ss, Cambridge, 2002),111- 338.

7. T. Harrisen.j. Hum. Evol. 18, 537 (1989).

S_ M .. L M(CfD'iSin el 01., iii Ptimote Locomotion: R~en( ildvatlw', ~. Strasser.J. Heaqle, A. Ro,~nberger, H. McHenry, Ed,. (Plenum, New York, 1998), pp. 353-396.

9. 5_ Mo\,~·SoliJ et ol., S,ience 306, 1339 (2004)_

10. S. Moy.1·Sola et ol., Ptoc. Nat/. ncad: Sci. U.s.A. '106, 9601 (2009).

11. D. Beg un, in Handbook of Poleoonthropoiogy Vol. II, W.

Henk~, L Tatter,all Ed) (Spri~9~r, Berlin, 2007). pp.

921-977-

12;, J. AgU51i e/ ol.,J. Hum. EVa/. ~.5. 1115 (Z003). 13 D. B~gun el cl., Deitneo 10, 23 (2003).

14. c-a. Stewart, T. R. Disotell CUlT. Bioi. s, RS8.2 (1998). 15, Y Kunim~tsu et ai, Prtu; Natl-lkod $". USA. 104,

1'9220 (2001).

16. G, SUWil et 0/., NO/1Jre ~~8, 921 (2007).

17.. M. Pi ckf ord, B. Sen ut, Anthropo/. Sd 113, 11 S (2.005) 18. M. Pickford ~I 0/., C. R. Pillevolll, 413 (2009).

10. 112611(ien(E.l1S~703

ECONOMICS

Measuring Subjective Well-Being

Richard Layard

How should human happinessand life-satisfaction be assessed?

What is progress, and how should we measure tile well-being of a population'? The Organization for Economic Cooperation and Development has held two major conferences on the subject, and last year, President Sarkozy of France established a distinguished commissian to report on the samequestions (J). This major debate reflects the fact that higher national income has not brought the better quality of life that many expected, and surveys in the United States show 110 increase in happiness over the past 60 years. These surveys rely on questions about subjective well-being, find it is reasonable to ask how reliable survey answers are as measures of the quality oflife as people experience it. On page. 576 of this issue (2), Oswald and Wu carry out an interesting test of this. First they measure subjective wen-being in each U.S. state, and then compare it with the average objectively measured wage in the same U.S. state (both variables being controlled ferpersonal factors). The negative correlation of the two variables is remarkably high=-as it should be ifhigher wages are compensating for a lower experienced quality of !ife (and vice versa), The study will likely stimulate some lively debate across many disciplines, including scientists, economists, sociologists, psychologists, and policy-makers.

But should we really adopt SUbjective well-being as om measure of the quality of Ii fe? Philosophically, many would say "yes," as they have ever since the l Sth-century Enlightenment. But, practically, can subjective well-being really be measured well enough to be used III policy analys is? Is' what

London School 01 Economics, London WC2A 2AE, UK. E-mail: f.l~yard@ls~.~(_uk

534

people say about their subjective state well enough correlated with the inner reality'?

The science is, of course, very young, but it is well enough developed fOJ us to say "yes," In the typical question, an individual is asked, "Taking a.1I things together, how happy are you?" The possible answers range from 0' (extremely unhappy) to I 0 (extremely happy). To evaluate the information con-

Moreover, questions on happiness and life satisfaction have-now been asked ill hundreds of routine population surveys, and ill multiple regressions within COUntries, the following causal factors are always irnportautrphysical health, family status, employment, income, and age. This is true both in cross-section studies and in panel studies that include an individual fixed effect. Moreover, the sizes of the effects are remarkably similar in widely different studies done within different countries (3) ..

Similarly, responses on life satisfaction can be used to explain behavior such as quitting one'sjob and exiting from marriage. They can also, as Oswald and \Vll show, be used to measure quality-o f-life differences across the United States in a way that is consistent with the pattern of wage differences.

Answers about happiness are also well correlated with measurements of bodily function, such as amounts of salivary cortisol, fibrinogen stress 'responses, blood pressure, heart rate, and (in some cases) immune system responses to <I flu vaccination, These correlations hold across individuals, as in the famous cross-sectional study 0 f B ritish Whitehall civil servants (4), and also in some cases within the same individual over time.

tent in the answers to such questions, we can examine whether these answers are well enough correlated. with other relevant factors. They are in fact well-correlated with at least five relevant sets of variables: the reports of friends; the plausible causes of well-being; some plausible effects of well-being; physical functioning, such as levels of cortisol; and measures of brain activity.

When a subject's friends are- asked about the subject's happiness, the answers correlate well with the subject's own report. (Were it not so, human society would find it hard to fUllction.)

Finally, there are reported correlations with brain activity across individuals, and within individuals over time. The best

:i'

known of these is the correlation of'positive $_

affect with activity in the left dorsolateral g prefrontal cortex (PFC) and negative affect ~ with activity-in the right dorsolateral PFC ~ (5) .. this area of work is in itsinfancy but, ~

o

if s~lcc~ssfnl, it -:iU rej~force t~e_ v.iew that ~

subjective experrence IS an objective real- ~ ity. Bec;:lJ!se this is so often questioned, it ! is worth repeating the findings of Coghill g (6) who applied the same very hot pad to 5

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

the legs of many subjects; in that case, the reported pain varied widely and was well correlated with activity in the cortex.

There are, of co urse, many different ways to measure happiness and life satisfaction. Such measures can be based on a single question or (to reduce measurement error) on many questions, whioh can be combined into a single index using weights that reflect their average impact. ou answers to tile single question. For most purposes, we would like the measurements to cover a substantial period of time, but this can also be achieved by repeated questioning, And as the Enlightenment tradition suggests, the questions should mainly aim at capturing pure affect (feeling), without too much involvement of judgm cut (7), Fifty years ago, there was considerable debate on how to measure depressian, but by now this has bocome much Jess controversial.Jn all likelihood, the measure'In ent of happ iness w il] become simi lady less controversial.

Now is the time for every govemrrrent to collect data on a uniform basis on. the happiness of its population. That was the most rrnportant of the recoml11end~ll.ions of Presi-

dent Sarkozy's commission (although it also recommended better measurement of GDP). Once there is good information on levels of happiness, three things will be possible: the monitoring of trends, the identification of problem groups in the population, and the ana lysis of WIlY some people are happy and others are not, To fully understand intercountry differences in happiness, we shall probably have to rely on biomarkers. in addition to answers to ve:rb~ questions asked in different languages.

As for policy analysis, this has to be based on solid science, using either naturalistic data or controlled experiments. But the metric has to change. Today's cost-benefit analysis measures benefit in terms of the money that the beneficiary has shown he or she would be willing to pay for a change of state. But formany key areas of public policy, such measurements make no sense because little individual choice- is mvclved=-think, for example, of physical heaJth, mental health, responsible parents, family stability, (unjernploymeni.; and community life. In these areas, we can got much better measures of the benefits of a policy change through direct measures of subjective well-being. It is therefore time

PERSPECTIVES

to begin developing all alternative system of cost-benefit analysis in which the units me units of subjective well-being.

But sound measnrernent will only become possible if social science (indnding psycho Logy) takes as a prime objective the quantitative study of the determinants of well-being, Every survey of iudividua Is shoul d. automatically measure their well-being, so that in time we can really say wha t. matters to people and by bow much. Wilen we do, it will produce very different priorities for our-society

References

1_ J Stiglil:'( et ol., lIep~(( by (he Commissio« on the Meo5vr'em~"t of Economic performance and Sociol Pri>gr",s, '~pt"rI1bef ~OO9 (WWW_sliglilz-,en-fitDui,i./r)' 2_ A,l. Oswald .. S_ Wu. "{i~n,e l27. 576 (2010); publj)h~d Dnline.1] D"c~rnbef 2'009 (10,} )26/

5d~~[ '" 118060 Ii)-

1. R. diTalla etai.,.Kev. fron. Sfar. 85, .809 (2(03).

4. A. Sleptl}~.l W~rdle, Ne!lmbfol. Aging 26 (suppl, 1), 5108 (ZOOS).

S. H. L. Urry et 01., Psycho!. Sd.1S. 367 (2004)-

6. R. C. CoghlU et oi., Prot. Nl3tl. Awd. Set. U.S.A.. lO(}. 8538 (2003).

7. O. l(ahnemon, m Well-Being: n,e Foundarions of HedO/lic psychology, D _ Kiln nem;ln el. aI., E ds. (Russell Sag e

Fo und ati on. New Vork, 1999). chW 1_

10 . liZ6/! ,ience.1186 31,

MATERIALS SCIENCE

Turning Away from High Symmetry

John C.Crocker

AcommOll sight in cold climates is the growth of crystalline ice "feathers" 01' "ferns" on (1, cold window pMC. These structures arc initiated by heterogeneous nucleation-c-the i ce crystals form on the surface of a different subst.ance, in this case, specks of dust. It is much harder to start crystal formation in pure fluids because the seed crystal must form spontaneously in solution (hence pure water can supercool and remain as a liquid for long times below its freezing point), Homogeneous nucleation is largely a mystery; observing small clusters of molecules directly is difficult, and their formation is a rare event, On page 560 of this issue, Meng et al. (1) report an exhaustive experimental study of the equilibrium cluster configurations in a model system consisting of microscopic plastic spheres that were designed to have a short-ranged, reversible attraction that drives them together (2). This

U n;ver.~ity of Pennsylva ni~_. (hem leal and BiornnlecuIilf Enginc~ring, ~2Q South 33rd Strn9t, Ph;\a~dpmCJ.. PA 19104, USA. E-mail: i(rocker@5ea5.up~nn.edll

work underscores the subtle geometrical difficulties associated with crystal nucleation, .111d helps us to understand the "niles" by which nature self-assembles small structures and ultimately crystals,

III earlier wor]; assembl ing clusters, Mancharan and co-workers (3) found that highly symmetric polyhedra! clusters were the most stable in terms of bond energies. However, in their current work on clusters in thermodynamic equilibrium, the most common structures are typically the least symmetric. The essential difference lies .in eutropic effects-in thermal equilibrium, the most common clusters are not those with the lowest internal binding energy (4), but rather those with the lowest free energy, which is favored by higher entropy.

Entropy has been equated in popular culture with unavoidable d isorder and. decay, but its true nature can be appreciated in the context or self-assembly. hi equilibrium, every possible way to build a duster (having the same' binding energy) will occur with the same probability=cthere are simply fewer

Ught microscopy studies of cluster formation by colloidal particles show that [eSS"symmetri cal stru etures a re favored under eq ui! i bri u rncerrditicns.

ways ofbuilding symmetric clusters. (see the figure, panel A). Specifically, Meng et al. hypothesized that the symmetric polyhedra we uld encounter two entropy penalties. Theil' rotational symmetry leads them to have very 10\'1 rotationa I enttol"IY, and their hi gh degree of Internal coordination makes them very rigid, which causes them to have very low vibratronal clumpy (sec the figure, panel B).

To experimentally test their hypothesis, Meng et al. first prepared a suspension of rnicrospheres that ate large enough to image with a light microscope, but small enough t.o undergo random thermal fluctuations and obey equilibrium statistical mechanics. This sample was then loaded into an array of'microfabricated chambers, each of which was small enough tbal.it would contain only a small number of'microspheres. An additional entropic cffect-c-called a depletion attraclion-was induced between tll~ rnicrospheres by the addition Of1l111Ch smaller, nanoscale particles (2) .. This force causes the microspheres to join np to (min clusters, one per chamber. Each duster was then SCl'utiliized

www.sciencemaq.orq SCIENCE VOL 327 29 JAN UARY 2010

535

PERS PECTIVES

A.

No rol~!jon syrl1 metry

B

or

or

Of

Twofold rotation symmetry

c

Octahedron

Rigid

to pure polytetrahedra through a combination of high-entropy floppy modes (as illustrated in the figure, panel B) associated with the square pyramids and the higher degree of bonding possible for such structures, These stru ctures a I so resem b I e familiar crysta I symmetries, such as

hexagonal close-packed (hcp),

The sphere-sphere attractions in this system are quite short-ranged and sti If compared to the microsphere diameter-the so-called sticky-sphere limit. The authors note that for the effectively longerranged potentials o f'ato mic systems, tbeir free-energy model predicts that the dramatic eutropic effects

become somewhat more subdued and the preferr-ed cluster structures become temperature dependent, Generalizing their model to nonspherical molecules will be a nontrivial extension. Moreover, recent experiments with similar microsphere mode 1 material shave SJlOWD

very slow (5, 6) or nonexistent (7) crystal nucleation. which seems inconsistent with the hop-like cluster structures observed here.

Unlike the conditions here, where clusters can equilibrate at leisure in complete isolation, the clusters in an unbounded fluid are continuously bombarded by ami grow by absorbing smaller clusters and can form transient gel-like chai 11$ of clusters, all 0 fwhich may frustrate the equilibration of internal modes. It may well be that to

understand nucleation, we need to understand its absence-the breakdown of equilibration that prevails during the. glass or gelation transition, Still, the field guide of equilibrium clusters provided by Meng et al. will provide a crucial baseline against which uouequilibrium clusters may be compared.

let me count the ways. A pa rtiru la r duste r is m ore Ii kely to occur at eq ui Ii briurn, beta use of entropic effects, when there are more ways to assemble the structure and when the structure has lower symmetry. (A) To understand rotational entropy, consider the simplified problem of adding a third "square" 10 a dirner of two squares (middle). There are four sites for adding the squa re that create a bent conformation (top), but on Iy two sites for creati ng the linea r co nfo rmanon (bottom). The bent site is favored by having twice as many possible configurations, which is formally equivalent 10 its having half as many rotational symmetries. (B) To understand vibrational entropy, consider a duster of disks that can assemble into rigi d tria n9 les or floppy sq uares, which flex to form a rhombus (middle). If ~ disk is a dded to the center structure adjacent to the squa re (top) when the square is fully flexed, it (an be locked into and fa rm two rigid tria ng les, The disk [an also bea dd led?) elsewhere and leave the flo ppy mode intact (bottom). The rig id stru cture with lower vi bratic nal entropy is l~ss likely 10 form, as the ti me wi ndow for its Iorrnatio n is limited, whereas the: bi nding sites that leave the flo ppy struttu re j ntart a ra always available. In eq uili briurn, the top duste r is far less likely to form than the bottom one. (e) When assern bli n9 six spheres, Meng et at. fou nd that the most syrn metric cluster, the ottahedron (top), only occurs 4% of the time compared to 96% for the pnlytetrahedron (bottom). Rotational entropy leads to a lactor of 12 difference in proba bility; the remai ning factor of -.2 is f romditterences in vi brationa I entropy.

under a microscope, and followed as it randornly tumbled end over end and diffused around its chamber by Brownian motion.

The d.i ffi cult part 'I i es i 11 the determination of three-dimensional structure, Meng et al, recorded a short video for each cluster with enough frames to show the tumbling cluster from different angles. Then, like bird watchers identifying species, the authors matched each cluster's distinctive appearance in the video, by eye, to a field guide of computer-generated images of cluster structures in different orientations, To collect a statistically complete sample of all clusters of up to eight spheres, they manua lly i denti:fi ed more than 500 clusters in this way. A. partial survey of the structures With lip to 12 spheres was also completed.

The hypothesis of Meng et al. was validated by this analysis: The most symmetric

536

24-fold rotationa l symmetry

floppy mode

Six spheres

Refere!lfe.s

1. G. Meng, N. Arkus, M. P. Brenner, v. N. Manon.Jran, 5(ien~e 327, 560 (2010).

2. J. r Crocker, J. A. MaHeD, A. D. Dinsmore, 1\. G. Yildh, Phys. sev. Lett .. 82, 4352 (1999),

3. v. N. Manohara~, M. T. Elsesser, D.]. Pine" Science 301, 4S'i (2003).

4. N Arku" V. N. Ma~Dnar"n, M P. Brenner, PhY$. Rev. Lett. 101, 11330.3 (2009).

S. P. L Biam;,mie~o, A. J. Kim,], C. Crutker, PhY5. Rev. Left. 94, 058302 (Z005)

6. A.]. Kim, r. L. Biancanietlo, J. C (rm;ker; Langmuir 22:, 1991 (2006).

7. f>. J. LU e/ 01., Nature 453, 499 (2008)..

10 .112 611de~ (e .1184457

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

96% Poly tetra hadron

Floppy mode

Twofold rotatlen symmetry

and rigid structures we r e dramatically unde r - represented in their sample, ill quantitative agreement with their free-energy model. In one example, clusters of six spheres can form two different structures with the same number of sphere-sphere bonds and binding energy: a regular octahedron, and a "polytetrahedron" consisting of three tetrahedra joined by adjacent: faces (see the figure, panel C). Remarkably, the latter low-symmetry structure was observed 96% ofthe time, compared to just 4% for the octahedron. According to the authors' model, this difference is almost entirely accounted for by the octahedron's high rotations I symmetry,

Similarly, the most symmetric and lowestenergy case for 12 spheres, the icosahedron, was never observed. For clusters ofrnore than 10 spheres, structures containing both square pyramids and tetrahedra were favored relative

RETROSPECTIVE

Edwin G~ Krebs (1918-2009)

Wi II iam A. CaHera II and John D. Scott

Edwin G. Krebs, a giant ofbiomedical science in the 20th centtlry,. died on 21 December from congestrve heart failure in Seattle at the age of 91. His discovery (with Edmond H. Fischer) ofprotein phosphorylation as a regulatory mechanism touched all aspects ofbiomedical science and pro foundly infiuencedtherapemic approaches now usedin clinical care. Ed's life epitomizes commitment to family, excel-

lence m research, and academic service.

Edwin O. Kr-ebs was born in. La using, Iowa, in 1918, the SOl1 of a Presbyterian minister and a schoolteacher. His father died suddenly when Ed was I s: and at fhe height of the Depression, the family moved to Urbana, Illinois, where :Ed earned a degree in chemistry from the University of Illinois in 1940. As an undergraduate, he became enamored witb organic chemistry but eventually chose to study medicine at the Washington University School of Medicine in St. Louis, Missouri. Although the principal responsibility of a medical school during World War II was to train physicians for the armed forces. Ed also participated in medica! research. After medical school and residency training at Barnes Hospital in St. LOllis, he went on active duty as a medica! officer in the Navy. Following his d i scharge ill 1946, he returned to St. Louis and was accepted as a postdoctoral fellow in thel aboratory of Nobel Laureates Carl and Gerty Cori in the Department of Biochemistty. After 2 years of postdoctoral research on the interaction of protamine (a small sperm protein) with rabbit muscle phosphorylase, Ed became so captivated with biochemistry that he never returned to' clinical medicine,

During his naval service, Ed enjoyed a hi; ef ,6 visit to Puget Sound, so he happily accepted ~ a position as assistant professor of biochem~ istry in the fledgling University of Washing~ ton School of Medicine in 194 K Under the g Visionary leadership of Hans Neurath, the

v.. "" "' ,.

z

;; D efJa rIm en t or Ph,mfla w(ogy, School of Medici n e, Urii ver-

Ii! sity of Washinglon, Se<ltlle, WI!. 98195-7280, USA. E·mail: 5 wCi1tt@u,washington.edu

Department of Biochemistry expanded in protein chemistry and enzymology, including recru ttment in 1953 of Edmond Fischer, a talented and charismatic Swiss biochemist studying potato phosphorylase. Thus, a litelong friendship and. a formidable research partnership were forged,

Together Ed (Krebs) and Eddy (Fischer) determined the mechanism by which adeuo-

sine 5'-mollophosphate (AMP) servedas an activator of ph osphorylase bin.skeletal muscle. They found that adenosine 5' -tnphospbate (ATP) was required for phosphorylase activation, and in an unusual experiment discovered that calcium, leaching from filter paper used to clarity the muscle extract, was an important cofactor. They demonstrated that phosphate was incorporated into a specific serine residue of phosphorylase b, thereby yieldIng an activated form called phosphorylase (I, Their landmark paper was published in the Journai of Biological Chemisrry ill 1955. Subseqnently, Krebs, Fischer, and colleagues confirmed that this phosphorylation is mediated by another enzyme (Phosphorylase b kinase), which itself is conhoned by yet another enzyrrte, a cyclic AMP (cALv£P)-respOllsive kinase. This finding led directly to the concept of kinase cascades In 1968,. Ed purified the cAMP-dependent protein kinase (protein kinase A). The recognition that protein kinase A is' the primary cellular effector for cAl'vIP signaling placed protein phosphorylation in a central position in hormone action and broadened the importance of this protein modification in cell biology, physiology, and pharmacology,

At the same time as his discovery of protein kinase A, Ed's interests in teaching and academic administration led him to seek new opportunities and c 11Cl llenges as founding cll!)ir of'the Department of Biological Chernistry at the University of California, Davis. He also embarked on a long association with: the AJUerican Society for Biochemistry and Molecular Biology as-associate editor of the Journal oj Biologfc.al Chemistry fbr 20 years and as presiden; of the American Society

PERSPECTIVES

Denfli I1g how protein phosphorylation is regulated led to an explosion of knowledge a bout most cellular processes.

fOT Biochemistry and Molecular Biology in 1985,. In 1977, Ed returned to tile Universi ty of Washington as investigator of the Howard. Hughes Medical Institute and chair of the Department of Pharm aco logy, where he inaugurated a msjor new research direction in molecular pharmacology What be liked most about both onus chair positions, he said, was tlre responsibility of selecting outstanding faculty members for the departments.

After achieving his goals as department chair in 198.3, Ed refocused his efforts OD research, training j unior' scientists, and solving new problems in cell signaling. His laboratory studied signaling events that involved the phosphorylation of protems on tyrosine residues. His lab WaS also instrumental in the discovery of the mitogen-activated protein kinase pathway, a sequence of'protein kinases that respond to extracellular stimuli and regulate a wide range of cellular processes including gene expression and cell growth, differentiation, and survival,

Krebs received many major scientific awards for his insights, into the principles goyewing eel! u 1 ar regulation ill health and disease, i ncllldil1g election. to the Nationa I Academy of Sciences (1973), the Passano Foundation Award ( l 988), the Horwitz Prize (1989), toe Lasker Medica! Research Award (1989), the 3M Life Sciences Award (L989). and the WclchAward in Chemistry (199 I). At age 74, he and Edmond Fischer were honored with the 1992 Nobel Prize ill Physiology or Medicine for their discovery of protein phosphorylation. In 1997, Ed finally closed his lab but was frequently spotted wandering 111e halls of the University of Washington Medical Center on his way to bear the latest research seminars. Ed is survived by his wife of 64 years Virginia (Deedy) Krebs, three children, and several grandchildren.

Edwin G. Krebs will be remembered for his keen intellect, astonishing research productivity, and iconic status within the research community Those who were privileged to work closely with him will rernember him fondly as a k-ind and gentle mentor who passed 011 extraoedinary insights in a quiet and dignified manner The legacy ofthis self-proclaimed reluctant biochemist should be a wonderful inspiration to the next generation of our profession.

10, 1126/'(i~nc~.1l869 13

www.sciencemaq.orq SCIENCE VOL 327 29 JAN UARY 2010

537

WINNER OF SCIENCE PRIZE FOR ONLINE RESOURCES IN EDUCATION

Making Genetics Easy to Understand

Louisa A. Stark' and Kevin Pompei

!The Human Genome Project and the subsequent explosion of genomic information are transforming our knowledge of how organisms function and how genes and the environment interact. These insights have led to advances in personal ized medicine, stem cell treatments, and genetic testing. Students, teachers, and the public must be prepared to make informed decisions about participation in genornics research, genome-related health care, use of genetically modified agricultural products, and public funding for stern cell research, Education has been identified as a crosscutting element that is critical to achieving the potential of genomics research (1).,

To address this need for genomic literacy, we have developed two related Web sites. Learn.Genetics (see figure, right, from http:// 1 ea rn, gen etics .uta h. edn/) provides educational materials ttmt currently cover 15 topic areas ranging from DNA to epigenetics, Classroom activities designed to support and extend these materials, as well as other

resources for educators. are

available on Teach.Genet-

ics (http://teach.genetics. utah.edu/).

Our goal in developing these Web sites has been to make genetics and geuomics easy for everyone to understand, The grants and contracts funding the sites havesupported development of curriculum-supplement materials for grades 5 through 12 that address the National Science Education Standards (2) and gaps in standards-related, free, online. multimedia educational materials.

Web-site and ill-person feedback indicate that the materials are used by a much broader audience, Students from middle school through

An integrated pair 01 Web sites lor students

a nd teachers supports genetics and genomics education worldwide ..

to "viral" dissemination through link-sharing Web sites and blogs, Although this type of dissemination is unpredictable, both am "Mouse party" and "Cell size and scale" (see figure below) interactive animations have spread this way, engendering discussions about science in over 30 languages around the world. "The new science of addiction: Genetics and the brain" module has received the most unanticipated LIse; i.t has been incorporated into police officer training and addiction treatment in several countries,

We use a participatory design approach to developing our materials, involving teachers and scientists along with the science educators, instructional designers, science writers, teacher professional developers, scientists, multimedia designers, Web developers, and evaluators that comprise our team. Om method emerged from extensive. work with teachers in professional development programs and capitalized OIl teachers' real-world expertise ill successful teaching approaches, knowledge of engaging topies and materials, knowledge of the gaps in available online materials, and familiarity with

the state science education standards guiding curricula. It also

J:

d.raws on scientists' depth of 5

expertise in their fields. lnvolv- (;

,..

iug the center's entire team ~

builds understanding of the CQn- ~ tent aud learning objectives and ~ enables each to contribute his or ~

lj

her perspective and expertise to ':;f

the process=from writing, visu- ! alization, production, Classroom w evaluation, and teacher profes- ~

sional develop ment, iF: <,i

Om development process for III

1 ffi

a modu e begins, with a summer '"

workshop, advertised to teach- ~ ers through our e-mail 'list All £ online application enables us to E! select an outstanding group of 12 g to 18 grade-appropriate teach- ~ ers who represent a diversity of u

graduate school use the site to better understand content their instructors present, to assist in completing assignments, and to explore science independently. Higher-education faculty usc the materials for courses ranging from introductory biology to professional preparation in education and nursing,

Animations presenting science concepts ill all accessibleand engaging way attract mom hers of the genera I publ ic, which leads

--

--

---

--

--

--

learn.Genetics. The site provides educational materiels on 15 topic areas, ranging from DNA to epigenetics.

Genetic Science l.earniuq Center, Univ~r5ity of Utah. Salt. Lake City, UT 84112, USA.

• Auth or fa r C crrespo n dence, E-m ai Houisa, sta rk@utilh.edu

antibody

tRNA

711ff!

hemoqlobtn

6·5nrr

Cell size and scale. This interactive animation allows users to

zcorn from a coffee bean down to a carbon atom. cornpa ri ng the relative sizes of representative (ells, rnirrorqanisms, orga nelles and molecules along the way.

538

29 JANUARY 2010 VOL 327 SCIENCE www.screncernaq.orq

teaching experience, student populations, and locales; about 5 to 10% of applicants are accepted. Participants receive travel expenses and a stipend,

A typicaJ 4Yz~day summer workshop begins with talks by scientists and discussions 0 f scientific alticles, from WJ11ch participants distill important concepts for their students. The teachers and om staff work together to define the "big ideas" that emerge from these concepts, around which the module will be organized. Small groups of teachers then develop each big idea, drafting online and classroom learning experiences designed to assist students in learning, The workshops offer a rare opportunity for teachers to develop creative ideas. for curriculum materials that will be used worldwide. to interact with scientists, to update their content knowledge, and to work with other teachers from across the country. A glimpse into one Summer workshop can be Seen at http://lcClm.genctics.utah.cdu/ content/epigenetics/credits/, IIi it, teachers describe ideas that became the Insights From Identical Twins movie and "Gene control" interactive animation 0.11 Learn.Genetics and the '·DNA and histone model" activity on Teach. Genetics.

After the summer worksho-p, om team works with the materia Is the teachers drafted, Ideas may be combined, modified, expanded or contracted, as we plan a: module that addresses the big ideas while fitting the anticipated cost within the available budget.

Our materials evolve each year as we learn from past modules and feedback obtained via classroom testing, teacher workshops, and Web-site feedback. At present, key module conceptsare addressed il'i animated and interactive activities, because these appeal to the broadest range of learners. Narration allows users to concentrate on the animation and leads to deeper learning (3); text is available for those who are hearing-impaired or in computer labs lacking headphones. "Learn More" pages pro-

~ vide additional information for those inter~ ested in exploring beyond the basics, All ¥ pages are designed following standard Web g usability guidelines (4), including meaning~ fbi visuals and text that is clear, concise, and ~ easily scanned.

~ Good instrnction addresses multiple is learning styles, such as v isual, auditory, ~ and kinesthetic. (5. 6), Therefore, our modg ules include non-computer-based classg room materials designed to support, extend, g and assess online learning. For example, g the somatic cell nuclear transfer (SeNT) ~ technique uti I bed in the "eli ck and clone" ('j interactive animation is minored in the

ESSAY

Louisa A. Stark, Ph.D., is director of the Genetic Science Learning Center (G5i.C) at th e University of Utah, Dtlring _gradtla.te school, S"tark had the op porttlnity to take hands-on science into Denver i nner-ci ty schools as a member· of the "Science Squad," spa rki ng a passion for working with kindergar!,en Ihrough gra.de.'12 students and teachers. After completing her doctorate. in evolutionary genetics, she baqan

a career with scienre education partnership p rog rams, joining the GSlC 1 n 1999. Her rscenrawards indude.the.. 2008 Frie nd of $cien ce Education Award~ the 2008 Award for Excellence in 1:1 uman Genetics Education, anda 2009 Distinguished Alumni Award.

Kevin Pompei is the GS LC associate director and leads the center's educational materia I development team. He has served as CEO of two software com pa niesa nd has extensive experience in software plld Web design, Web architecture, and database development, as well as managing teams of diverse technology professionals. He is worki ng on a 9 ra duate ~egree in Educationall'sychology with a focus on lnstructlonel Design and EducationaL Te(hn.ology.

Once a module. is produced, it is tested in the classroom with students and teachers who were not involved if] the. development process. Students complete knowledge assessments before, after, and 2 weeks after studying a module. Teachers receive a stipend for thei I" participation and feedback.

We have begun expanding beyond genetics with our "Amazing cells" and "Great Salt Lake ecology" modules, We hope to develop additional materials in the areas of I:i fe science and health and other scientific fields.

The Web has become the primary infermation source for individuals with access to the Internet. Online educational materials thus have the ability to affect science literacy, preparing individuals to participate in the workforce, as well as to become jnformed hea lth -care con sumers and citizens of the 21 st century.

References and Note.s

1. F 5. Collin, er al., Not~re 4n, 835 (2003).

2_ Nabonal Re:;Eorch Coundl, Notional Science Edt1qJuofl Standards (Nil tiona L !leil demy Pres I, W ash i ng ten, 0 C, 1996),

3. R. E. Moyer, N. D1r. Teach. Learn. 2002(89), 55 (2002).

4. 1- Nielsen, u~eitmm'l~kob NieL>en's_ Wi>b,ile (~99S,2009); \IIIWW.useiuoml.

5. R. Dunn, K. Dunn, Teaching Studenrr Through their Iml.ivid.ual teamilig $tyley,·.A. Prm;tkqi Approach (R~,ton. ReslOn, VA, 1978).

6_ H. Gardner, Frames of Mind: The Theory oj Multiple WeL· ligences (Plume, Ne\VYor~, 1993).

7. A. E. Ni~~elle, !Ii esis, U niv, 01 M~lbou("" (21)09).

g. Sup ported by NIH I Natio nil I (en terfo r Res earc h

Res ources, Nil tion ill I nstitute Oil Drug A buse, ~ nd N"bon"~ Human G~~anl'" Reseilrdi lnstitute), Howald Hug hes Mroi G'I L I nsf tute, Mil rrh of Di me>, H ealtn

R~,ou I( es Servi ce Ad min i strati O~, It Ha rold BurlDo Found.tion, Soren50n Molec~l.Jr Geneillq.gy Foundation, Utah Deportment of Health, Great Sillt lake lnstitute ill WeSlmin,ler College, C~ildleo'5 Tumor Foundation, ;md University of Utah. With thilnb to Ihe entire GeMtir S,ielite L~"[Jii~g ~.nt~r team.

www.sclencemaq.orq SCIENCE VOL 327 29 JANUARY 2010

paper-based "Let's clone a mouse, mouse, mouse" activity. We suggest using the latter as all assessment for the on 1 Inc activi ty:, <l sking students to use the mouse and petri dish paper cut-outs to create a poster describing the SeNT technique, without giving them the i nsrructions, A dditionally, worksh eets are provided for many other animations to guide students' learning.

An entire module comprises 2 to 10 hours of instruction. However, in designing our materials, we recognize that many teachers do not use a- curriculum supplement module in its entirety (7). They select materials that address the science standards they are required to teach, ale appro· priate fOI their students' level, and £t their instructional designs (7). For example, in the" "Arnaz i I1g cell s" rn odule, a seven I hgrade teacher might only use "Inside a cell" and HCl;"l1 size" and scale," A high-school biology teacher might. use these activities plus "Build-a-rnembrane," "Coffee to carbon,') and "The fight-or-flight response." To address!l1ultiple grade levels and teachers' d iffering use of the e urricula, each animation an d classroom activity focuses On as i ugle main learning objective, making it easier for teachers to incorporate the materials into their lessons.

As our team produces a module, they consult the scientists who participated in the summer workshop and others for additional information and to verify scientific accuracy. Currently; module production also includes development of valid and reliable assessment instruments, used in classroom field tests. and later added to the resource materials. on Teach.Genetics A mediumsized module, such as "Epigenetics," takes 3 to 4 months 10 produce.

539

AAASN EWS&NOTES

EDITED BY EDWARD W_ LEMPINEN

ASSOCIATION AFFAIRS

Science and Technology. She has consulted on science policy for governmentagencies in China, Taiwan, and Singapore; at the Singapore Botanic Gardens, there's a species of orchid named for her.

Born in the city of Nanchang in southeastern China, Huang came to the United States with her family as a young girl. Her father had been bishop of the Anglican Episcopal mmistry in Southwest China, but if not for his calling to the clergy, he told her, he might have been a doctor. HeT mother was a [Jll1·SC".

From. age 7, Huang set out to become a physician. Attending parochial girls' schools, "1 had some wonderful teachers," she recalled. "They understood why 1 was curiousabout things and gave me the tools to realize the answers to roy curiosity."

It was at Johns Hopkins that she was first exposed to scientific research, and that diverted her from medicine. But she was involved in medical research for much of her career, working on viruses, cancer, HIV, and other diseases.

She served on the AAAS Board of'Directors from 1997 to 2001 and was elected a AAAS Fellow in 2000.,. She will succeed Nobel laureate Peter Agre as president when the AAAS Annual Meeting closes on 22 February; Agre will begin a l-year term as chairman oftheAAAS Board.

Huang sees AAAS and Science as strongly positioned to support the scientific enterprise on issues ranging from political manipulation of data to reducing laboratolies' paperwork requirements. She cited two priority areas for continuing MAS efforts: supporting women and minorities, and international science engagement.

International science collaboration can propagate a scientific culture based on. free inquiry and meritocracy, Huang said. "The unfettering of curiosity and freedom to ask questions in pursuit of scientific discoveries can become a subtle force for a freer society and less authoritarian governing system."

In the United States, science must help Women and minorities break through tile glass ceiling to leadership positions. "They're generally hired into the profession and 'not promoted in proportion to (heir numbers," she said, "It's a tragic loss of individuals who have the talentand capability and who really have the ability to con. tribute to our society,"

Alice Huang: Passion, Freedo,m Are Crucial to Global S&T Progress

The old year was ending and. a new one beginning, and Alice Huang was scanning tbe popular news media lists of the most influen tial p eop le 0 f 2009 an d the decade past, hoping to see a scientist or a11 engineer. This year, she didn't find one.

Though there are many worthy candidates, the lack of public recognition defines a crucial' challenge for American science, the incoming AAAS president said in a recent interview. Scientists and science teachers m ust do jnore to convey to the public-s-end to science strldeJ1ts~the ideal.iS1J1, creativity, and passion tha:t drive many breakthroughs.

Huang, a distinguished virologist now at the California Institute of Technology, described how the thrill of her own first disco-veries inspired a career of research. But too often these days, she said, even excellent students lack that passion.

"We don't focus on that when we first talk to people about science," Huang said. "When young people come to me and ask for ad-vice. I find that they're- very focused all 'how do I get ahead, and how do I get the position I want?' Somehow they've forgotten ... the joy of living a passionate life. Every scientist I've met who is successful is indeed passionate about what they do."

In Huang's view, talking about passion and idealism is crucial for addressing chalI enges ill h ea lth, energy, th e envi ronm en t, and other fields. It can help build public understanding of science and support for investment in research and development. It can help recruit more students Into science and engineering, especially from the uuderutilized pool of women and minorities. Overseas, it can help drive horne the point that. a culture of freedom is critical to scientific advancement.

Huang knows her subject well-s-since earning her Ph.D. at Johns Hopkins in 1966, she has come to QCClIPY ali influential position at the juncture of research, education, and diplomacy.

As a graduate student, Huang discovered and characterized defective interfer-

lncorninq AAAS president, Ali'ce 5. Huang, a senior faclllty associate in biology at Caltech, will become the president of MAS on 22 Febrl)ary, when the associa~oll's annual meeting doses in San Diego.

ing viruses=-viruses that have the pcrential to help control viral diseases in plants, animats .. and humans. Her work raised the possibility that defective interfering viral particles could be used for disease prevention. Her postdoctora I work wi til David Ba I timon: at the Salk Institute and Mrr on vesicular stomatitis virus led. the way to Baltimore's Nobel Prize-winning discovery of reverse transcrrptase.

Huang spent 20 years on tile faculty at Harvard Medical School; from 1979 to 1990, she also directed the Laboratories 0 f Infectious Diseases at Children's Hospital in Boston. She was appointed dean of science at. New Yark University in 1991, and in 1997 moved west to serve as senior councilor for external relations at Caltech, Today, sJU!;'S a senior faculty associate in b io logy there.

Huang is a past president of tIm American Society for Microbiology, and served from 2004 to 2009 on the California Council 011

540

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag_org

PUBLIC ENGAGEMENT

Science Books & Film Gets Digital Makeover

After 44 years In print, AAAS's Science Boob &. Film has gone online with a host of new features, ensuring the venerable journal's position as one ofthe world's leading authorities on science media resources.

Podcast interviews with authors, a weekly editor's blog, and a searchable database of thousands of reviews are among the valuable new tools for librarians, parents, and readers. Although best known for its reviews of children's books, SB&F also covers science television and film, software, and Web sites.

"SB&F has been a one-of-a-kind journal for decades," said. Heather Malcomson, the journal's editor for the past 8 years, "But I thi n k that the online pre SCI1ce, using a mun ber of plarforms, will really give usan opportunity to reach more people, and. more young people. Weknow the enU1tlSiasm for SCIence reading is out tllere,and We want to COI1I]ett those- readers with. the best the 'SCIenCe world has to offer."

Each monthly online issue, available at www.shfonline.eom, contains about. 7S reviews written by scientists, educators. and media specialists. But the new blog for subscribers expands these offerings with II weekly look at new books and software, science on television, and ongoing SB&F projects. Malcomson hopes that the blog will also "become an interactive place for me to communicate with our readers."

Librarians, who rely 011 the reviews to guide critical purchasing decisions for their science sections, make til) the bul k of more tha 11 1500 journal subscribers, according to Malcomson. Patrons of'rnany public libraries, as well as individual subscribers" have full access to all of SB&F's features, including lists ofthe past years Best Books for Chi ldren, Junior High, and High School Students, and Best Video and Software,

Subscribers call also plan their rn onthly viewing schedule with the. help of SB& F s "Science on TV" column, which offers shortdescriptions of select programs with a scientific bent.

Even without a subscription, visitors to SB&F online can listen to the journal's podcast series, ''AAAS Book Talks." Tne series, supported by the William T. Golden Bndowment Fund for Program Innovation, interviews award-winning science authors to learn more about their inspiration and future projects.

"The authors have really interesting stories to share about where the idea for the book came from or stories behind the bOOK, like research expeditions," said Malcomson WilO does many of the interviews along with SB&FEditorci1'i·ChiefMal'la Sosa "All offhe

authors have a desire to communicate science in a way that is exciting and encourages listeners to take all interest in science."

Along with auto manufacturer Subaru, SB&F sponsors an annual award honoring individual science books. The AAASJSuharll SB& F Prizes for Excellence in Science Books celebrateoutstanding science writing and illustration for childrenand young adults,

-Mally McElroy and Becky Ham

AAAS

Call for Nomination of 201.0 Fellows

AAAS Fellows who are current members of the association are invited to nominate members for election as Fellows, A Fellow is defined as 11 member "whose efforts 01) behalf of the advancement of science or itsapplications are scientifically or socially

ANNUAL MEE:TING

Se;eking Global Solutions, Scientists Meet in San Diego

Ihe MAS Annual. Meetin_g has Long been known asthe premier mu ltidisci plina ry scierrce gather.ing ill the United States, hu t tbis year it wi ll (;on lin ue its evolution to a prime international affair;: When the meeting convenes in 'Jan Diego from 18 to 22 February, scientists, journslists, and educators Irom more than 50 nations will be there,

In his lnl11 tactio n to the rneetln g, MAs President and Nob.elLaureatePe-ter G. 'Agre urged his colleagues to embrace a global lOi5;0 n of sden ceo The meeting "(arts on every setenrist and engineer to make their work both benefical and understandable," said Agre. '1ft 'is a call to action that resonates around theworld."

(onvening under theba nnar ~.BJidg'ing Science and Society.," tDIlI researchers will discuss.

their findings 1 nthe context of global challenges In the environment, economy. health. and education. Attendees can explore (Lltting-edge research In the neurssciences, energYt'astrobiolog1, public hl'lalth, and enVffonmentat nhanqe, amI learn liow these advances directly affed eenrtrnom tria ls, ra re for I heeld erly, sustaina b Ie cjtles, bl] rd er secu rityj and other pll bHt ~Oncern5, f~p~rts wiLL also diS{u5s :efforts tu

di stiilguish ed" A nomination must be sponsored by three AAAS Fellows, two of whom must have no affiliatie]1 with the nominee's iusritntion.

Nominations undergo review by the steering groups of the associ ation 's, sections (the chair, chair-elect, retiring chair, secretary, and four members-at-large of each section). Each steeling group reviews only those nominations dcsjgnated for its section. Names of Fellow nominees who are approved by the steering groups are presented to the MAS Council for election.

Nominations with complete documentation must be reoeivedby 10 May 201 O. Nominations received alter that date will be held for the following year. The nomination form and a list of Clul'entAAA.S Fellews can be found at www. aaas.org/aboufaaas/fellows, To request a hard copy of the nomination fO!ID, please contact the AAAS EXeqltive Office, 1200 New York Avenue NW, Washington, DC, 20005, USA; at202-326-6635; or at gseiler@aaas.org.

open up laborate ries a nd factories and brin g researeh to the classroom in an unprecedented effort te s how students th e day-to-day ,world of discovery and a pplication.

Agre will open the ~76th Ann'±lal Meeting wfth a presidential address; other preminent speakers include Us. Ilresjclenti~l adviser Eric Lander, who will discuss science and technology in the administration ot U,5, Presid!;Rt Ba~ack Obama: Nobel I<ltlleate Carol Grelde~ Who, will talk an the genet- 1(5 of deg.eneratf1Je disease; U.S. GE'ologital Survey Dlrector Marcia McNut'1;" who Will .,pea k on ocea n '50- ellce;, and If]ternationalUnear Coll~der D'irertar Barry C. Barish, ~ho$e address will f oru s on neW fro ntiers in pa rticle pJ1Jsics.

The .staff of ScJence, SpenceNOW, and MAS's Stiens« Update radie program will provide coverage [rom Sao Diego in news reports, pndrasts, and bloqs, AAAS.org's Annual Meeting NeW5 Blog will p rovi de exte'051ve rep a rt-

ing on sympasta and briefj ngs, a Ion !:I with lin ks to U.s. and internarienal news (OVenl.ge. Find the full spectrum of Annual Meeting coverage at www,aa<is.org/golt1ews.

For registration and other info~matjon about the 2010 Annual Meeting, visit www,aaas.org/ meetings. Information from the meeting wlll also be posted at the 2010 AAAS Annual Meet· ing and SfienceNOW p~e5 on Facebeok, and via Twitter at #AA~StO. ~BeGky Ham

wwwsciencemaq.org SCIENCE VOL 327 29 JANUARY2010

541

Coexistence of Quiescent and Active Adult Stem Cells in Mammals

lin!leng Lil ~nd Hans CLeversz

Adu Lt stem cells are crucia L. for physi oLogical tissue renewal. a nd regeneration after inj ury, PrevaiLing models assume the existence of a singLe quiescent population of stem cells residing in a specialized niche ofa given tissue. Emerging evidence indicates that both quiescent (out of cell cyde and in a lower metabolic state) and active (in cell cycle and not abLe to retain DNA labels) stem cell subpopulations may coexist in several tissues, in separate yet adjoining locations. Here, we summarize these findings and propose that quiescent and active stem

cell populations have separate but cooperative functional roles.

Stem cells constimtea long-lived populau .. on of c.ellS tha . .t p. assess.' the ability 10 self. renew (a process of duplication without losing developmental potential) and give rise to multiple cell types (differentiation) (1). In DnJsophilu and C{)fw;whabdifis elegans, a single population of genu stem cells (GSCs) resides in a single niche (2, 3). The asymmetric architecture of the stem cell niche dictates that stem cells normally divide asymmetrically into a new stem cell (self-renewal) and a committed progenitor (differentiation), GSCs only divide symmetrically (self"renewaJ only) when expansion of the stem cell pool is required (3). The invertebrate GSCs normally undergo constant cycling (either asymmetric or symmetric division) (4, 5) and become quiescent when animals are challenged outritionally (6). 'ill contrast, mammalian adult stem cells are generally detected as in .a predominantly quiescent state (7-9). Bow can long-term quiescent stem cells support rapidly regenerating tissues (e.g., producing billions of blood and intestinal cells daily) during normal physiology'! The mechanism wbeJ"Cby quiescent stem cells give rise to transit amplifying CTA) cells, which in tum differentiate into mature cells, may not provide a satisfactory answer because TA cells are short-lived and cannot self-renew. Recently, populations of stew cells that are long-lived yet constantly cycling have been identified (10). Here, 'We integrate insights from bone marrow, intesfinal epithelium, and bail: fallicle to. fhrrnulate an alternative and complementary model in which subpopulations of quiescent and active adult stem cells coexist in the same tissue.

Hair Follicle

The hair follicle provides an excellent system for studying stem cell biology. Each hair follicle is composed of a permanent portion, which in-

1 Stow~rs Institute for M~di(ill Resililfch (51 MR), Kafl5dS City, MQ 64110, U SA and Depertrn ent of Pathology, Kansas University Mt>dicaL Center, kansas City, KS 66160, USA. E"mail:

L i l@slo\V€rs.org 1 H~ bre(h t I nstitute, Utrecht 3584 cr, Nether· lands and Uriivefsity Medical Center, Utrecht 35840(, NeUlerlands, E-mili!: h.devefs@lltIbre~ht.e\1

542

eludes the sebaceous gland and the underlying bulge region, and a temporary portion (cycling segment) that cycles through anagen (active

A

t

PermaneITl segmen1

I

",,,,o,:'Z- - IJ

__ I ;. q Half maid,

Dermal papilla (DP)

s·

T

Transit amp lify in.g (TA) cells

u

CBCs

growth), catagen (apoptosis-driveo retraction), and telcgeu (a resting period) (Figs, IA and 2A) (11, j 2). SIgnals cmam'J.tlog from the 'underlying dermal papilla (DP) induce activation of the stem cells in the. hair genn (a cluster of cells located at the bottom of bulge) and 'the bulge (13). During anagen, the distance between the bulge and DP increases, which modifies the activation state of the stem cells (Fig. 2A).

Cotsarelis and colleagues were the first to identify label-retaining cells (LRCs) in the bulge region (7). However, because of the inability to isolate live 5-bromo-2' -deoxyuridine positive (Brdu) LRCs. they could not determine whether bulge LRCs were stem cells. Fuchs and colleagues developed an elegpnt method to isolate live LRCs by generntin,g a doxycycline-controlled mstooe2Bgreen fluorescent protein (H2I3-GFP) mouse line (J4). EXpression of H2B"GFP Was seen. in almost all of the epidermal (including stem) cells without adding doxycycline. When doxycycline

o Quiescent (LRO) stem cells Activt,l stem/progeni1or Ct,liJs

-- 0steoblasts

OS! eobla.s1fc lining qeJls

t

LROs

fiSC

Endosteum

Fig. 1. Stem cellloGltions in hair folLicLe, gut, and bone ma.rrow. (A) Hair follicle structure with quie5H'nt (bulge) ilnd active (hair germ) stem/progenitor cells. BuLge area typically maintains quiescent stem cells, whereas DP provides stimulatory signals. Only during development and under injury condition, bulge stem cells give rise to stem cells in epi de rm is. (8) I n test! na I crypt structure with qui es cent (+4 ) and acti ve CBC (lgrST) stem cells, .as well as TA and mesenchymal cells. (e) Quiescent H5Cs located in the endosteal region where osteoblastic lining, endothelial, CAR, and other cells form the endosteal region and active HSCs located in the central. marrow region, which Lacks osteoblastic cells.

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

was administered to young mice fur 1 month, most cells diluted out the H2B~GFP label by division during the chasing process, Some cells in the bulge, however, retained H2B-QFP and were thus identified as CFF-expressing LRCs. 11.e majority of these H2B-CFP LRCs expressed the stem cell marker CD34 (14). Sorted and cultured CD34+ bulge cells from a keratinl4-GFP transgenic mouse were transplanted and shown to generate the entire hair follicle (J 5). Taken together, these studies demonstrate that LRCs (here identified as CD34+ and K14') in the hair follicle represent functional stem cells (J 5) .

.AJthough many bulge cells can be activated from the quiescent state to enter the cell cycle during anagen (15), some LRCs retain labels over many months suggesting that they exist in a quiescent slate and are not actively involved in hair regeneration (J 6) (Fig. 28). These observations suggest that bulge cells are heterogeneous in terms of cycling state (I 7).

Bulge stem cells "Were recently found not to directly generate T:A. cells but rather give rise to an intermediate stem cell population located at the hair germ. These hair germ cells in turn produced the TA cells, which further differentiated into the hair shaft in the hair matrix (13, 18).. Leucine-rich repeat-containing heterotrimeric guanine nucleotide- binding protein (0 proteinj-coupled receptor (Lgr) 5 is expressed in the bulge region and the hair germ during late telogen/early anagen as well as in the bair matrix during late anagen (Fig. 2C) (J 9). In the bulge, tile CD34' compartment consists of both label-retaining Lgr5" cells and active Lgr5' cells located SIde by side (Fig. lC). This is consistent with the detection of Akt-phosphorylated (activated) .LRCs in the lower bulge region and hair germ during anagen (Fig. 2BJ (20). Indeed, the Lgr -5' cells seem to be til e f rst to proli fern te upon induction of anagen, and. some LgrS" cells persist through sever-ill hair cycles (19) as revealed by ill vivo lineage tracing, Thus, the location of Lgr5+ cells ill the lower bulge and extending into the hair gell11 is consistent with the recent finding that activation of bulge and hair genn stem cells occurs in two steps (13). Activation of bulge cells follows activation of hair genn cells, indicating that the bulge cells are replenishing hair germ cells that have entered the hair matrix to support hair growth (Fig. 2A) (J S?- On tbe basis of these observations, the hair follicle appears to contain both quiescent and active stem cell populations in separate yet adjacent locations.

Gut

In the intestinal epithelium, stem cells and their short-lived 'fA cells reside in crypts (Fig. ill). Cells exiting the crypts and. entering the villi terminally differentiate into enterocyres, goblet cells, or enteroendocrine cells. Paneth cells escape this course by migrating to crypt bottoms, The intestinal epithelium is renewed about every 5 days (21). Potten and others have proposed

REVIEW I

Earlyanagen

A

Telogen

{

:'

Hair cycle

)i( ,... + Olffitru ctlon

OP ...... and

• retraction

Cat(lgen

o Quieseent stem cells

• Active stem/progenitor ceils __ Stimulatory signals

Late .anagen

B

G

Telogen

D

Hair gelffi

,i3ui.ge

I ",ulge L.gr5+ I [Ja.iT germ

Fi g. 2. DP position determines state of stem/progenitor ceUs in hair follicle. (A) Hai r cycle. At the start of anag en, DP is proximal to hair germ and bulge. Stem/pro genitor ceUs in hai r germ (a structure below th e bu Lg e) a.re a ctivated by DP first, whereas stem celts rema; n quie5C en t in bulge. When hai r germ cell5 enter the hair matrb, then stem ceLis in bulge are actiir.:rted to repLenish Lost hair germ cells. During anagen, active hair genn stem cells give rise to TA cells, which in rum support hair growth. DP is pushed down from bulge because of fast expansio n of p rogenito r cells, In catag en, DP retracts toward bulge. (B) (0 -stafning of BrdU with phosphorylated -Akt shows relatio nship between quiescent stem cells (yellow arrow) and a cti ve s te m cells (red arrows) in bulge and hair germ. [Reprinted by permission from John Wiley and Sons, Incorporated, AlphaMed Press, p. 2834 of (20).] (C and 0 ) CD34+LgrS- and CD34"'19r5+ stem rells, respectively, toea te din si de and n ext 10 b u 19 e a rea (C), which is con 5i stent with Lg r5 + stern cells located a dj acsnt to LRCs (D). [Reprinted by permission from Macmillan Publishers, Limited, pp, 129.2 and 1293 of (.19).]

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

543

REVIEW

that LRCs (on average 2. to 4 out of 16 cells in a ring) at the +4 position located above the Panetb cells (Fig. IB) represent stem cells (8). A recent study using in vivo lineage tracing bas shown that cells expressing Bmi I predominantly mark +4 position and ate able to give rise to all four epithelial lineages ()1). Although Bmil + cells were suggested to be slowly cycling, it is CUl1"COtly unknown whether they are LRCs (i.e., quiescent). Some cells at the +4 cell position that express the potential stern cells marker Mesashi - I ate Dot sensitive to inactivation of the cell cycle protein CDC25, suggesting they have quiescent features (23),

In an alternative model. crypt-based columnar cells (CBes) specifically expressing Lgr5 located at crypt bottoms among Paneth cells represent intestinal stem cells (fSCs) (Fig. 1 B) (14), Lineage tracing has shown that Lg1."5 , cells represent a long-lived and cycling multipotent stem cell population (15). A single ~r5+ stem cell can fOnTI a long-lived, se If-reo ewing "minigut" to culture (26). Lgr5" SIe1D cells are distinct from +4 LRCs in that LgrS stem cells do not retain DNA labels. and are sensitive to CDC25 inactivation, supporting their proliferating feature (23). Thus, the intestine contains a quiescent stem cell population III the +4 position and a cycling .Lgr5 stern cell population among the Panerh cells, Future ~rudie.s are needed to determine the relationship between these cell populations.

Bone Marrow

'Studies of hcmatcpoietic stem cells (HSCs) have shaped our thinking mi. most basic features of mammalian stem cells (J} H5Gs give lise to a hierarchically organized set of progenitors for erythroid; myeloid, lymphoid, and megakaryocyte lineages (1). The mainstay of HSC analysis bas been marker-based cell sorting followed by transplantation in lethally irradiated recipients, This assay tests self-renewal as well as multilineage potentia L

Most primitive HSCs are quiescent (27) Using DNA label BrdU and H28-GFP incorporation respectively (14) .. LRCs were found to be predominantly located in tile endosteum (Fig, Ie) (28) find subsequently confirmed to be a primitive HSC population (29. 3(1). The quiescent HSCs have superior long-term reconstitution poteutial; however, fISCs in this population cycle only once evcty 145 days on average and thus may not provide ongoing support for production of billions of blood cells, although they can be activated fur this function under injury conditions (;19)_ Other studies based on the DNA label BrdU and H2B.,.GFP incorporation have suggested mal the majority of murine HSCs undergo more ti:equent cycling (29,31 33). These seemingly disparate findings may be reconciled by posrulating the existence of two subpopulations of HSCs, one being long-term quiescent ("reserved") and the other being more actively cycling ("primlXl") (Fig, IC) (29,33,34). Primed

544

HSCs may be the workhorse that supports the daily produ, .. 1l0n of billions of blood cells, whereas reserved HSCs function as. a "backup" (to replenish lost active stem cells under homeostasis and particularly in response to .i.ili ury or pathological challenge) (29, 33).

Stem Cell Zones .and Associated Microenvi ronmental Signals

The observations described above do not easily confbrm to a model in which asingle stem cell population mairnains tissue self-renewal (Fig.

A Single stem cell populaUon

Stimlliatory

Proliferation and differentiation

migrating upward from the crypt bottom and passing +4 may potentially provide a negative feedback. to the quiescent stem cells. Thus, loss of mes or their progeny may trigger tbe activation of quiescent stem cells. Different assays may be biased in revealing one or the other population of stem cells: transplantation assays document "stem cell potential" and may favorably detect thereserved population, particularly ill the serial transplantation assays, whereas in vivo lineage tracing assays visualize the primed population.

How can these two very diffen .. nt states of stern cell subpopulations be maintained at 'Separate but adjoining locations? The nature of the locations of

stern cell su bpopulati OI'Is in hair fcllicle and intestine suggests that there are. segregated zones that

del ermin e the active or qui escent state of the stem cells. Individual SIe:lD cell Riches are maintained by the secretion of specific proreins. for example, the hair tbllic1e bulge constitutes a canonical Wnt-

oIDbol'le morphogenetic protein (BMP) on microenvironment rhrough the expression of secreted Dkk-I, sFRP, Wif, and HMP (14). to contrast, the OF secreresWnt signals and noggin"m antagonist of Bfvl.P signaling (J 1) (Fig. 4A). Similarly, cycling LgrS'; stem cells at the crypt bottom in intestine arc in a micmeovironmeot "With high Wot activity (JO), whereas BMI' signaling is inhibited by noggin

and gremlin produced by submucosal tissue below the crypts. (37, 38). Stern cells at +4 positions encounter BMP4 and the Wnt inhibitor sJ1RP5 (37, 39) (Fig. 48). In bone marrow, endosteal region (composed of osteoblastic, vascular (40);and C){CLl2 abundant reticular (CAR) cells (41), OS well as osteoclast'> (41)J and central marrow region (lacking osteoblastic signals) have been shown to form inhibitory

and stimulatory zones respectively (43,44) (Fig. 4C).lnm.guingly, HSCs in aged mice localized more distantly to the endosteum than in young mice, supporting the existence of two zones in bone man-ow, with the central IllaTTOW zone favoringproliferarion ofFJSCs (45). This correlates with the increasing number but decreasing function of aged HSCs (46). Under homcostarie conditions, BJI..1Ps (28), Osteopontin (47), Wnt signaling inhibitors sFRPI (48), as well as. potentially nnncanonical Wms are expressed in the endosteal zone, providing an inhibitory micrcenvironmeer. Fibroblast growth fadors (FGFs) and canonical Wnts expressed from endo-

Active state

Fig. 3. Two models fOf stem cell-baseq tissue se[f-ten.eWaI <Jnd regeneration. (A) Prellaili n 9 mod 1"1 of a si n 9 [I" stem ce II po pu lati 0 n located in the niche. Asymmetric division is the key mechanism to maintain bal<ince betwe.en self-renewal and differentiation. (B) PrO" posed a[ternative and mmplementary model: co-existing quiescent and active stem cell popalations lOCilte!:l in adjacent zones with co rres p ondi ng in h i bito ry. and sti mula tory si gila ls, Q uiesce nt stem cells replace damaged active stem c.ells (heavy arro'(i). Conve.rsely, active stem cells may 'replace lost quiescent stem <:ells ([jght arrow). A negative feedback from either active stem cells (dashed line) or their prQgeny (solid line) may contribute to preve n ti on of q u ie seen t ste m ce lls. fro m activati on.

29 JANUARY 2010 VOL 327 SCIENCE www.sciencernag.0rg

B

Two siem cell population.s

fnhlbilGry zone

Q "~"'Q ~~;d'~;k"

Q'Uiesl:ent state

Ptellfe r.d i0n and differentia\lon

]A) . An alternative "zoned" model in which quiescent and active Stem cells coexist within the same tissue, may better explain these observations (3j). In the zoned stem cell model, active stem cells are the primed subpopulation that account. for rncst vef the replenishment of corresponding tissues, whereas quiescent stem cells function as a backup or reserved subpopulation. This reserved population can be activated either by a stochastic mechanism (36) or by feedback upon Joss of active stem cells or extensive tissue damage {Fig. 3B), For example, the cycling eBC stem cells and/or their progeny

A

Inl11blrory

.,

stem. / (OS) 1"'11/

\ Bulge

,

\ ,

, "

,

_,

\ ,

\, ,~

'\~~f (I'lP)

'. sFliIP

, ,

" Sti'mulatory··,

NoggfnlWnl (DP)

B

In"lbitorv

r I

I ,

I

r

"

I

•

• ( ,

Me

"

, .; "" GBC's

,~

, "

,,'" # Stimulatory ~ .... , ....

Noggin I GremlinIWnt (su~ m UC<lsa)

REVIEW

c

fnhl,bllory

Osteoclast

Qulescent

@

T

'. ••

(Osteoblastic

lI~ing cells)

I r

" ~'

.-

.,,'" • Adivil ,

,.1 ,j H6G '"

, "

'CAR

/

, ,

, ,

Endosteal zone

Central marrow zone

Fig. 4. Zoned stem cell population and theassodated microenvironmental signals,. (A) Bulge area typically provides Wnt~offand BMP-on signals thus maintaining quiescent stem cells, whereqs OP provides stimulatory (Wnt~on and BMP-off) signals. (B) Similarly, in intestinal crypt (+4) ISCs are rnai ntained by Wnt-off and BMP-on signals. In contrast, CBC (lg6''') stem cells

are exposed to Wnt-on but BMP-off signals. (C) Quiescent HSCs located in the endosteal lone whine osteobla~c lining cells provide dominant inhibitory signals including BMP, OPN, and sFRP.t. In contrast, hlSCs located in the central marrow lOne are stimulated by endothelial, megakaryocyte, and CAR cells secreting Wnt, FGF, and SO Flo

thelial megakaryocytes, and CAR cells may have a dominant and stimulatory influence on the central marrow zone under homeostatic conditions (33, 49,50) (Fig, 4C). The locations of these cells and the associated signals might change" however, under stressed conditions (43)_

Functional Significanc.e of Zoned Stem Cell subpopulations

What could be the advantage of maintaining zoned subpopulations of stem cells? We speeulate that increasing life span and body size during evolution exerted selective pressure 10 increase the longevity and output of adult stem cell pools, particularly in the rapidly regenerating tissues, without increasing the risk for (tumorigenic) mutations. The adaptation of reservedquiescent stem cell pools may provide a mechanism to increase the longevity of ad.u1t stem cells through population replacement (Fig. 38). Here, populalion replacement describes a mechanism by which one populatiou of cells is replaced by another and should be distinguished from the classic mechanism of self-renewal at the individual cell level. This is exemplified by lost stem cells in hair genn being replenished by stem cells located in the bulge during each bah cycle (1 3). Nonnall y, qui escen t stem cells wo ul d be expected to replace damaged active stem cells (Fig. 38). TIlls would prevent the active stem cell pool from becoming exhausted and protect against accumulating potentially rumor(genic mutations during DNA replication. Conversely; active stem cells would possibly have the capacity to replace lost or damaged quiescent

stem cells under very spec ial circumstances (Fig. 3 B). Toe eombination of these reciprocal backup systems would provide a robusr mechanism to ensure a higb rate of physiological self-renewal as well as flexible damage repair, after which the original hierarchy could be reestablished,

Reference's and Notes

L L L W"i.5s.rna~, Cell 100, 157 (20oo)-

2. J. E. Kimble, ]. G. While, Dev. Bioi, 81, 208 (19Bl).

3". Iii. Spr.~dliog, D. Drummond-Barbosa, T. Kai, Nature 414, 93 (2001).

4. S. L Eritti'nden, K_ A. l"onhaf~, D, T, Byrd, J. Kimbl,E, Mol Bioi. Cell .17, 3 OS 1 (zOO 6).

5. X. SQng, G .. B_ Call D. Kirilly, r, xte, Developmenl 134, 1071 (2007).

6, P. Narbonne, R Rgy, Development 133, 6 U (2006).

7. G. Col>ar€li., T. f. Sun, R, M. Lavlrnr, Cell 61, 1]29 (19.90)"

8. C. s Potlsn, C. Boolo. D. M.. PrillhOld. lilt ).Exp. PO/bDl. 78, 219, (199'1).

9. F. IIrai er al., CellllS, 149 (2004),

1 p. N, Sa rker et 01., Na lure 449, 1003 (2007).

11. E. Fuchs,. I. A. 5~gte, Cell 100, 143 (2000).

12. C. Nleni"nl1, F. M Wall, Trend s Cdl BiOI. 12, laS (~002) D. V. Greco et ai. Cell Stem Celi 4, iss (2009).

14. T Iumbar et 11/., Science 101, 359 (2004).

15. C. Blanpain, W. E. lowry, /I. GeoghEgan, L. P.ol~k,

E. Fuchs, Ceil 1l B, 635 (2004).

16. 5. K. Waghmare ef al; EMBO J zr, 13 09 (200B).

17. E. Fuch" (ell 137, 611 (2009).

HI. Y. V. Zhang, ]. Cheong, N. Ciapurin, D.]. McDermitt,

T. TUiil~or, Ce/I Slem Cell 5. 267 (2009).

19. V. jaks et ol, Hat. Genet. 40, 1291 (2008).

20. J. Zhang e/ 0/, Stem (ells 24, 2526 (2006).

21. 1\1, Barker, M. van de Welering, H. devers, Gillll1S Dev. 22, 1856 (2008),

22. E. Sangiargi, M, it Capecchi, Nat. Gel1et. 40, 915 (2008),

23. G. Lee, L S. White. K. E. Hurov, T. S. Stappenbstk, H. PiWIiica- Wornis, Prot; NaIL Aca,I So. U.S.A 11M. 4701 (.20U9).

24. H, Cheng. C. P. leblnnd, Am, J. Anal. 141, 5)7 (1974).

25. N, B.r~l', H. Oevers, GastroentemlOgy 133, 175. (2001).

26. T. S.lo et al., Nature 459, 262 (2009). 27 F_ Ara.i et 111., Cell 118. :tA9 (2004).

28. J. Zhang et al., Nature 425,836 (2003).

29. A_ Wib;Dn f'f 01., Ce//135, 1118 (200S).

3 O. A. fou di er 01., Nat. Biotechnol. "1.1, 34 (2 0(9).

31. s, H. Cheshier, s. J_ MOrrison, x,. liao, L L. W~i»manl

Ptoc. Noli. Acad. Sci. U. S. A 'J 6, 31:!O (1999). 32 M.]. Kiel el ol, NatUre 449, 238 (~007).

33. J. 5. Hauq et ol., Cell Stem Cell z, 367 (2008).

"34_ I- L. AbkDWilz ~t ~I., pm,. Nat/. Awrj, sc'. U,S.A- 87. 9062 (1990),

35. 0, H. Scoville, T. 5alo, X. C He, L Li, Gastroenlerology 13 4, .849 (2008).

36, I, L McKenzie, O_ LGan, M, Doedens, ]. C Wong, J. [ Dick., Nat. Im!JIllnol. 7, 122S (20'06).

37. )(. C H~ el al., Not. Genet 36, 1117 (2004).

3 a. c. Kosin; ki ~t ot; Ptoe. NaiL Acad. 5ci, U.S.A. 104"

15418 (20'07)

3 9. A. Gr"gOliHf, H. Clever" Genes Dev. 19,377 (2M 5). 40'. M.]. Kiel et 0/" Cell ~ 2'1, 1109 (,2O'D5)

41. T. Sugiyama, H. K6h,m, M. Noda, T. Na.g~sa\'la, Immunify

25, 977 (2006).

42. O. KoU€I et 111., Nat. Mer:I. 12, 657 (2006).

43. Y. Xie et 01, Nature 457, 97 (20'09).

44. c. lc Celsc a/ al .. Nowre 457, 92 (2009).

45. A_ Kiihler et 111., Blood 1:L4, nO (2009).

46. s, M. Cha mbefl et 01., !'Lo's Bioi. 5, e.201 (Z 00'7).

47. S. K. Nii';,on et 0/., BloQd~06, 1232 (2005),

48. T. Yokota eI 0/., j. Imlllul10L sas, 60'61 (2DOS).

49. M. Alillline et al" Growth Factors 23, 87 (200S), SO. A. R. WecheZ<lk, 0', E. Coan, £xp. Cell se: 288, ~ ~5

(2003).

51_ We 'Ihiln~ J, ~Imb~ and E .. Fuchs fOr \!lejr crih",! raadinq an d very va luab Ie co m m ents; colleague s, i ndud: n 9

I. Grindley, D SCOville, x, He, I. Hauq, v. 'Greco, etc., lor I ugg estions and discussion>; end J. Pe rry end K. To n nen for pro DIre;, d io g. WE [~gret Ih;; t we are vn~llle to d Ie other relevant studies be(aUlf 01 spate constraints.

L L. wa~ supported by SIMR and by gr~ol UOIDKOBS 507 from Naliona I In5 ~ tu te of Dia betel a;n d Dig~ stive an d

Ki dr1~y Di,ea:>e.

10. 112 61sci en ce. 1180 194·

Www:sclencemag.org SCIENCE VOL 327 29 JANUARY 2010

545

Role of ABA andABI3 Desiccation Tolerance

..

In

A,. Kl1andl!lwal, 1 S. H. eho,2 H, M~reUi.i,J Y. Sakata} P.-F. Perroud,5 A. Pan,1 R,. 5, Q\.latrano5.-

To survive on hod, the earliest land plants bad, to ,develop mechanisms to tolerate desiccation. Modern vascular plants possess an array of morphological features to retain water (such as conductive tissues, cuticle. and stomata) a nd have retained desi ccation to leran ce in only a few specialized structures (e.g., seeds) Present-day bryop bytes ( mosses), in contrast, lack; water transport and retention tissues. presumably like early land plants. As, a result, their vegetative state is at equilibrium with tbe surrounding air, creating a water-deficit condition that most angiesperms could not tolerate (l) Phylogenetic analyses.suggest that desiccation tolerance in vegetative tissue cf'bryophytes WaS lost in the first vascular plants (2). Here, we evaluate whether desiccation tolerance in angiosperm seedsand in 'vegetative tissues of tlre moss Physcomitrella parens use similar regulatory pathways.

The phytohormone abscisic acid (ABA) proteets seeds during water stress by activating genes through transcription factors such as ABSClSlC ACID lNS£NSITlYE 3 (AEl3) (3),

ABA IS also found in non seed plants such as algae and P patens (4) and uses similar signaling pathways. Far example, a wheat ABAresponsive promotercan be activated by ABA in cells of P patens (5), and one of three homoLogs of ABl3 found in P patens partially complements the ArabidO{isis abi3-6 mutant (6).

Untreated wild-type (WT) filaments of P patens can survive up to 92% water Loss (7) butcannot recover from complete desiccation (Fig. IA), We generated two independent lines (!;,.abi3-J and 6abi3-2) in Which all throe P.

patens AB13 genes. (A, B, and C) We113 deleted by using sequential gene targeting (fig. S I) (8,9). WT lines survived if incubated with AHA (10 10 100 f1M) fer 24 hours before desiccation, whereas two t:.abi31ineS" did not survive, even at 100 ).1.M ABA (Fig, lA), The 8abi3 hoes were also nat responsive to an ABA-responsive promoter from moss (PpLEA 1 tl-GUS'j, whereas WT exhibited an increase (fig. £2). Expression of 22 ABA up-regulated genes from WT P. patens (that are presumably required fur tolerance) were compared with those of Aabi3 at 24 bOUIS after AB A treatment, 24 hours after drying, and 5 min and J 5 min after rehydration (Fig. 1 B). Without PPAB13. only a few transcripts had reduced expression after ABA treatment and drying, whereas the others maintained their expression. The loss ofPpABJ3 bad little effect on this subset of ABA up-regulated genes before re hydrati on. However, all 22 genes assayed at 5 and 15 min after rehydration showed drastically reduced transcripts or none at all in the illIbi3-J line when compared with \VT (Fig. 1 B). For successful recovery from desiccation, PrABIJ appears to be essential for the maintenanee, either by synthesis OT stabilization, of those transcripts in" dueed during the ABA pretreatment that are critical for tolerance.

We conclude that both ABA and ABJ3 are required for P. pO/1m,S', vegetative. tissue to surviv.e desiccation. Because the P patens genome lacks the transcription factors r1US3 and LEO (10) that are required for seed maturation like AB13 (3), the role of ABU in this nonseed plant appears to be directly in desiccation tolerance,

Fig. 1. (A) ABA and PpABI3 are re- A

q uired for desiccation toleran ce. Ti5- AS"'

sues from 6-day~old WT, 6abj3~1, and (jIM)

6obi3-2 were treated with ABA (0, 1,0,

50, and 100 )lM) for 24 hours. Tissueswr were dried for .24 hours, rehydrated with $terile distilled water, and in{lJimted for

2 weeks. (8) Reverse transcription poly- _ Jof

merase chain reaction (RT-PCR) analysis

of ABA-induced transcripts in WT and

6abi3-.1 during ABA treatment. dr,ying,

and rehydration. RNA wqs extracted from ~

6-day-old ti~ues 24 hOlJrs a.fter ABA treat-

ment, 24 hours after drying, and 5 and

15 min after rehydration in basal medi-

um. cDNA was synthesized with use of 2 )1g of RNA, and peR was performed with- Use of genespecific primers (table 51).

primarily in the recovery stage. OLIT working hypothesis is tliat gene regulatory pathways tllat include both ABA and ABB originally evolved for ce U ular protection fiom water def cits but independently have been used to provide desiccation tolerance in vegetative tissues ofbryophytes and in angiosperm seeds.

References and Notes

~ D, fO G~lf, in Mechani5ln5 of "nlliranrom/ai SIre,s R e5i510 nee in PI an I!;, A. S. Basra, R. 1(', Sa sra, Ed,. (H~rwood Academic, londorr, 199/), pp, 43-58.

2. M. J. Oliver, ]. Vetten, B .. D. Mis:hier, illfegr. Compo Bioi, 45, 788 (2005).

3. R. it Finkelstein, 'II, R~~V€5, r. Anizumi, c. Steber, At/flU.

Rev; Plan! Bioi. 59, 387 (20PS).

4. M. M. John. P/!y:>M MoL Bial. p/an~ 14. ~3. (2008).

5. c. D. Knight et ot, pll1nt Cell 7, 499 (1995).

6. H, H. M<lr"€lla,. Y. S;jkata, R S. Quatrano, Plant J.46, 1032 (20Q6).

7. W. Fho~, D ~ Raloa de,.,;, R. Rej~i, PI liMa 22 O, 384 (2005).

'It P.· F. Perroud, R. $. Qu~lrilno, Plant Cell 20,411 (2008).

9. Mal~ri~ll; and method, are ~v:;i\llbL~ a, 5uppmting material 00 science Online.

10. 5. A. Ren~ing. et iii., Science 319, 64 (ZOOS); pubLished onliae 13 ueeember 2007 {1O.11261>d~nce.ll SM46).

11. We thank l , GUnther il.nd l. Main'" for technic",l support and D. Cove and S. McDanieL for many ·hetpfuL

d ls cuss ions. Su pported by fun ds f rom ~ 5 F IE F· 04 25749) ;3 n d Wash in glon Univ~r5i ty.

Supporting Online Material

'VWW.l riencernaq.or g/rgi/co nten tlflllL/3.2 715 9 65154 6/DC1 Mil I;l:rials a nd M~tli 0 ds

Fig,. 51 and 52

Tabl€ 51

RefetenC5

22 O,lober 2:009; ol crepted 18 D~ [em ber 2:009 to 1126Ijden~e.~1836 72

'Monsanto (QmP1lny, 700 Che51l!rfi~ld PoOOI'<lY We-;t, 51. Lo~!is, MO 63"017, USA. 'D~ilrtment of Biology, Pennsylvani~ S(ate- UniverSity, g~te College, PA 16802,· USA. lDonal£j Danforth Pla nt Sden ce Center, 5 t, louis, MO 63132, USA. 'Department of BioSdffice, TGkyo Ul1iYg{s1ry of Agri(U!lure, Tokyo 15&.8502, Japan. 'Department of Biology, Washington University, 51. La uis MO 6313 0, USA~

*To wh om corre spo n den (1" ,hould be ad d,e ssed. E -rna I (: rsq@wustl.edu

B

546

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

Local and Lang,-Range Reciprocal Regulation of cAMP and cGMP

in Axon/Dendrite Formation

Maya SheUy,l~ Byung Kook lim,l~ laura Cancedda,u Sarah C. Heilshorn,it Hongfeng Gao,l Mu-ming POOl;

cytosglit cyclic adenosine monophosphate (tAMP) <lnd cydic guanosine rnonophosphate

(cGMP) often mediate antagonistic ce llula r actions of ex;tracel(ul;:lf factors, from the requ lation of ion channels 10 (ell volume control and axon guidance. We found that localized cAMP

and cGMP activities in undifferentiated neurite~ (If cultured hippocampal neurons promote

and supp re 5S axon formation, r<2spectively, a nd exert opposite effects on dendrite formation .. Fluorescence resona nee enerl;1Y transfer imaging showed that atteratio ns of Ih e amou nf of cAMP resulted in opposite changes in the amount of cGMP, and vice versa, through the activation of specific phosphodiesterases and protein kinases. Local elevation of cAMP in one neurite resulted in cAMP reduction in all other Ml.!rite~. of the same neuron. Thus, 10Gai and long·range reciprocal regulation of cAMP and cGMP together e/1s\.lfes coordinated development of oneaxon and multiple dendrites.

The polarization of posunitotic neurons involves the di.:rerentiati~n of'a single axon and multiple dendrites, The process of neuronal polarization begins with the specification of the axon/dendrite identity of undifferentiated neurites, followed by selective localization of molecules that are responsible for differentia] growth and specific functieus of tho axon and dendrites (I., 2). The initial axon/ den drite sp ecification may result fro III cytoplasmic asymmetry caused by the last mitotic division of the neural progenitor cell (3), the stochastic fluctuation of cytoplasmic deterrninants (1, 2, 4, 5), or the action of extracellular polarizing factors (6-12), In cultured hippocampal neurons, neuronal polarization appears to begin with selective accumutaricn/activation of key components into one undifferentiated neurite U, 11, 9, 12 j 9) that triggers its rapid growth audaxon differentiation. To ensure the formation of one axon. and multiple dendrites, axon i.nitiatiQ1J at one neurite must be accompanied by long-range signals fhat suppress axon fonnation or promote dendrite formation in other neurites, We investigated the role of two second messengers, cyclic. adenosine monophosphate (cAMP) and cyclic gU8110sme monophosphate (cOlVlP), in mediating the co-

J Division of N €U ro bio I 09 y, Departrnen t (If Mol eeular and (en Biology, Helen Will, Neur05cience Institute, University crf C~liforl1id. Berkeley, CA 94720, USA. ?Oep<lft!11f'nt of Neuroscience and Brain Te[hnologie.l. Italian Instiltlte of 1 eci1no\·ogy. Via Morego 30, Genoa 16163, Iwly.

'The~~ authors (onll'jbuted equilily 10 this l'Iol~,

t,Pr€sellt address: Department .of Mat€ria\s SOl!nre and fngineering, Geball@ Laboralo!)l for AIlvan[ed Maleriills, Stanford UniVersity. St~l1iprd, CA 943 OS, USA.

:]:To whom correspondence should be addressed, E-mail: mpoo@udink.berkeley.edu

ordinated differentiation of one axon and multiple dendrites.

Antagonistic axon/dendrite i nitlation by local cAMP/cGMP a ctivi ty_ Dissociated embryorne hippocampal neW·0I1S undergo spontaneous polarization 01'1 a uniform culture substratum, from: a morphologicatly symmetric cell after plati.l"I.g to a. polarized neuron exhibiting a single axon and multiple dendrites within a few days (20, 21). First, we examined the effect of localized cA11P and cGMP activities in axon/dendrite initiation by plating these neurons on substrates coated with snipes of rnembrane-permeant flucrescent analogs of cAMP or cGIVlP [F -cAMP or F-cOMP; see Sl1PPOltillg online material (SO:M)]. The neurons were imaged at 12 bours after plating and were immunostained after further incu bation of 48 hours with the axon marker Smi- 312 and the somatodendritic marker MAP2 .. At 12 hours, cells exhibited several short neurites of similar lengths (Fig. L A and B). By 60 hours, nons were found on the F-c..A.MP stripe or off the F-cGMP stripe, whereas dendrites were mostly found off the F-cAMP stripe or on the F-cGfltiP stripe (Fig. I, A and B). We analyzed. the distribution of axon/dendrite initiation sites on the srnna at 48" tD 60 hoursafter plating, when neurons had completed the polarization process, for all polarized rolls with the soma located at the stripe boundary (Fig. I Ca). This retrospective analysts ofaxon/dendrire initiation was possible because neurite initiation sites on the SOlID did Dot move during neuronal polarization. (Fig. I, A and B). The distribution of the axon initio, tionsite showed preference for initiation on the F-GAMP stripe and off tile F-cGMP stripe, but no preference tor stripes coated with bovine serum albumin (Fig. I Cb). The distribution of dendrite initiation sites showed opposite pref-

erence (Fig. ICb). Analysis of the preference index {PI = [(% on snipe) (% offstripe)]!IOO%} showed that localized cAMP and cOM}> aotivities were sufficient to induce preferential initiatinn ofaxons and dendrites, respectrvely (Fig. ID). These effects are distinct from Ole known effects of cAMP and cGM P on The guidaoce of grcflNtl, cones (21, 13) that were also revealed by tile axnu/dendrite pathfinding at the stripe boundary (Fig, I, :A and B).

Next, we plated these neurons on u substrate striped with either the adenyhne cyclase (AG) inhibitor 5Q-22536 or the protein kinase A (PKA) inhibitor KT5720 or Rp-R-Br-cAMPS, to reduce locally basal cAMP/PKA activity. For neurons with the soma located at the stripe boundary, axons were mostly initiated offthe stripes (Fig. 1, Cb and D), similar to that found on the F-cGMP- striped substrate, Conversely, when the neurons were plated on a substrate striped with the soluble guanylate cyclase (soq inhibitor ODQ or the protein kinase G (PKO) inhibitor KT5R23 or Rp-8-pCPT-cGMPS, aX01JS were initiated mostly on the stripe (Fig. 1. Cb and D), similar to that found for F -cAW stripes. There was also a preference for the dendrite. to be initiated off the ODQiKT5823IRp-8-pCPTcGMPS stripes, but on the SQ-22536IKT5720! Rp-8-Br-!:AMPS stripes (Fig,. J. ce nnd. DJ. Thus, asymmetry in the basal cAIvIJ' or cOMP acti vity is suffiei en t to tri gger axo 0/ dendrite polarization, and cAMP and tGMP activities e:XCI:t antagonistic actions on this process,

Reciprocal regulation between tAMP and tGMP_ Antagonistic cellular actions mediated by oAMP and cGMP (J3- 28) C01)ld result from opposing actions of cAMP and cGMP pathways on common cellular targets, However, crosstalk between these two pathways could also occur by reciprocal regulation between cAl'vIT' and cGMP, as suggested by findings from 1100" nenrcnal systems (28; 19), We used genetically encoded fluorescence resonance energy transfer (FRET) reporters to examine directly the reciprocal regulation between cAMP and cGMP in cultured hippocampal neurons. The neurons were ti:ansfected 2 hours after cell plating with a. construct encoding one of the three FRET reporters: indicator of cAMP using Gpac. {lCUR) (30), cGMP energy transfer sensor derived from PDE5A (cOEScDE5) (3/), or A-kinase activity reporter (AKAR) (32) (SOM text), which were designed to monitor rhe amount of cA1vTP, cGMP. and PKA activity, respectively. All FRET measuremcnts were made 10 to 16 hours after cell plating, before axon/dendrite differentiation (:lO, 21). Bath application of the membranepermeantcANlP analog Sp-8-Br-cAMPS (LO !-1M) or the ACactivlItor ferskolin (20 pM) resulted in a global increase of cAMP and PKA signals in JCUE- and AKAR-AMpressing cells, respectively, as measured by the increase in the ratio of yellow fluorescent protein (Y FP) to cyan fluorescent protein (CFP) fluorescence at the neurite (Fig. 2,

www'.scieneemag.org SCI EN CE va L 327 29 JAN UARY 2010

547

RESEARCH ARTICLE

A, e, 0, E, and G). In contrast, the same treat- 1116nt induced a reduction of cGMP in C6US expressing cGBS-DE5 (Fig. 2, C, F and G), Conversely, tbe membrane-pertneaut cOMP analog 8-pCPT-cGMP (20 11M) or the nitric oxide donor (DBA NONOate,. 200 )lM), which is known to activate sOC, increased cGMP (Fig. 2. C. F and G) and reduced cAMP and PKA activity (F.ig, 2, 8,0, E and 0).

The reciprocal regulation was also observed when the basal level of either cAMP or cOW

was modulated by altering its synthesis OJ" degradation. Bath application of the AC inhibitor SQ~22536 (10 ).1M) resulted in not only the expected reduction of cAMP and PKA activity, but also in cGMP elevation (Fig. 20). Treatment wiih the sGC inhibitor ODQ (111M) yielded opposite effects: reduction of cGMP and elevation of cAl\1P and PKA activity (pig. 20), Furthermore, the nonspecific. phosphodiesterase (PDE) inhibitor 3-isobutyl-l-mcthylxantbine (lBMX, 50 ~Lfvl) elevated both the basal cAMPfPKA

A

B

Ca

• •

Ii! .. : '\ ... ... 4. .'

.. e :

- ,., ~ .

... .

.. -~

....

• It, ~ ... \ ... . ..

"IJ,;.~ ,.,.

,., ,.,- 'iII':,!!_ .

,.~p

01)0

D

0..1

ii OA

on· '"

gi Oi---'__-"r-_"-_""-

11

"D t -0,.

s!-

.o,e

Fig. 1. Antagonistic ,effects of cAMP and cGMP on axon/dendrite initiation. (A ,!tid H) Images of cultured hippocampal neurons on substrates (Dated with stripes (blue) of F-cAMP or F~(GMP at 12 and 60 hours after cell plating and immunostained (at 60 'hours) for axons and dendrites Mth smi-312 and MAP2 antibodies, respectively. White arrows and arrowheads: neurites that later became axons and dendrites. respectively. Yellow symbols: axon/dendrite turning at stripe boundary. Scale bars, 10 urn, (C) Preferential axon/dendrite initiation. (a) Diagram depicting the distribljtir;ln of axon/dendrite initiation site (circles) relative to the center of the stripe boundary intersecting the soma. (b) Distribution of initiation sites on-stripe (blue) or off-stripe (black), for axons (30 cells) and all dendrites (10 (ells) initiated on various striped substrates. Scale bar, 5 urn, (0) Preferential index for axon/dendrite initiation, shown as average ± so [n = 3 to 5 cultures, 150 to 200 cells each, P < 0.001, two-tailed t test or KolrnoqorovSmirnov test, compared to bovine serum albumin (S5A)].

548.

and cGMP levels, whereas the cAMP-spec.ific PDE4 inhibitor rolipram (I )lM) caused an expeeted increased cAMPfPKA level and a slight cGl\1P reduction (pig. 20), A peptide-based as, say also confirmed that I13MX could elevate PKA activity in hippocampal cultures (fig. S3Cb). and that rolipram treatment or transfectioo with PDE4D small interfering RNAs (sim'lAs) caused similar PKA elevation iJ;I. cultured HEK-293T cells (fig. S I A). Thu s, reciprocal regulation between cA11P and cGrvJP is constitutively present in these developing neurons, allowing adjustment of the basal level of these cyclic nucleotides bi-directionally. The antagonisticactions of cAMP/cOMP on axon/dendrite differentianon described above. CFtg. 1) could be attributed directly to the reciprocal regulation between these two cyclic nucleotides.

Mechanisms underlying reciprocaL regulation. hi variousuon-neurooal systems, cOMP-induced activation of PDEs inhibits cAMP signaling by increasing cAMP hydrolysis (29 .. 33), Measurements with FRET reporters ib cultured neurons showed that. after pre-incubation with JBrvIX, subsequent addition of NO-donor (or 8-pGPTcGMP) and forskoim failed to induce reciprocal down-regulation of cAMPIPKA activity (Fig. 3, Au and Ab) and cOMP (Fig. 3Ac), respectively, demonstrating the involvement of PDEs III this reciprocal regulation, Pre-incubation with roliprarn, which by itself elevated the basal level of cAMPIPKA activity (Fig, 2G), abolished the g'pCPT ;{:GtviP-induced reduction of the PKA activity (Fig. 3Ad). The roliprarn effect was specific, because the PDE2-S1Jecific inhibitor BAY 60"7550 bad "DO effuct (Fig. 3Ae), and it was not attributable to saturated cAMP elevation caused by rolipram, because further cAMP elevation could be achieved by adding forskolin to rolipram-treated cells (Fig. 3Af), Similarly, for the crosstalk from cAMP to cOMP, the forskolin-induced cOMP reduction could be abolished by the PDE3 inhibitor cilostamide (300 nM) (Fig. 3Ag) and was significantly diminished by the inhibitor of cGMP-specific POE5 MY-5445 (30 ~tM) (Fig. 3Ab). Furthermore, although an NO donor greatly reduced the forskolin-induoed PKA activity (Fig. 3Ba), it had no significant effect on the PKA activity induced by the nnnhydrolyzable cAM:P analog Sp-8~BHAMPS (Fig .. ]Bb). Similarly, fcrskolin suppressed NO donor-induced cOMP elevation (Fig. 3Bc) 'but had no significant effect on the cOMP signal induced by the nonhydrolyzable cGMP analog 8-pCPT-cGMP (Fig. 3Bd).

In non-neuronal systems (29, 33), POE .activities can be modulated by PKNPKG,dependent phosphorylation. We found that forskolin-induced cOW reduction was prevented by KT5720 (200 nM, Fig. JAi), and the reduction of PKA activity that was caused by 8-pCPT-cGMP was prevented by the PKG inhibitor K.T582] (200 11M, Pig. 3Aj). TIlUS, whereas reciprocal regulation between cAMP and cOMP is modulated by specific PDEs, PKAlPKG activation might

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

Fig. 2. Reciprocal regulation between cAMP and cGMP revealed by FRET imaging. (A) Forskolin-induced FRET signal for PKA activity. (a) YFP image of a hippocampal neuron expressing AKAR at 16 hours. Scale bar, 10 ~Lin. (b) Forskolin-induced cha nges in YFP and (FP fluorescence at a neurite tip (red square in a), normalized by the average value before forskolin application. (c) Ratio of normalized YFP fluorescence [F(YFP)] to CFP fluorescence [F(CFP)], representing the FRET signal (B and () FRET" signals observed at the neurite tip of 16-hollrs neuronsexpressi ng ICUE or cGE5-DE5. The black and white images show YFP fluorescence and the color show FRET signals at different times (in min) after forskoli n or 8-pePT -cGMP ap-

A

Fig. 3. The role of PDEs and P·iWPKG in reciprocal cAMPIcGMP regulation and phosphorylation of axon determinants LKBl and GSK-311 (A and 8) Average FRET signals for cAMP, PKA, and cGMP at 10 to 20 min a fter bath applicati a n of the. compound indicated above, in the presence of the drug inditated below (±5EM, S to 8 cells each, 1 or .2 per cell). In (A) and (B), data significantly different from control are marked by an asterisk (P < 0.05, Mann-Whitney Utes!). (e) Immunoblot "nalysis of cAMP-induced phosphorylation of lKBl, GSK-3~, and Akt on lotal. cell lysate.s of S-day cottu res of corbcill neur on s, using a nti bodies specific to the phosphoryla tio n site or the protein. Forsk., forskolin, Histo~ grams show the average phosphorylation level (±SD, n = 5), normalized to ~-acti n an d shown as fold of control (D and E) Antagonistic action of cGMP activity on tAMP-induced lKBl phosphorylaton is mediated bycAMP-specific PDEs. Immunoblot analysis as in (0. CeLls were treated with forskolin (D) or Sp""8-Br~PS (El, either alone or

eAMIi'

PKA

111M)(:

RESEARCH ARTICLE

I

plication, coded in pseudocolors (scale ba r on the right). Inset.: Higher magnification images of FRET signals at the neurite tip. Scale ba r, 10 urn, (D to F) fRET signal (at neurite tips) for all neurons expressi ng the ICU E (D), cAMP], AKAR [(E), PKA] or cG£S-OE5 [(F), cGMPj, induced by Ihe compound as indicated, averaged over 40-5 bins and normalized by the mean control value before drug application (±S.EM, n = 5 to 8 cells each, 1 or 2 per cem. (G) Reciprocal regulation between cAMP and cGMP and the role of PDEs and PKA/PKG. Oata represent average FRET signals similar to that in (D) to (F) for cAMP, PKA, or cGMP at 10 to 20 min after drug application. Error bars indicate ±SEM, n = 5 to 8 cells each, 1 Of 2 per cell.

8

I

cGMP

PKA.

c

E

.F

101111101111:

• • ... '*'

~.n; Ilk

pI:..KB' CS«431 I

• 10'80'

....

0.5 2 10 5(1

...........

GSK4J! pJIokt IS~.4131

Roll.".",: 11I:,I1.1C81 ,....,~_"...,..,~P!I. (S.r4311

pGSK~lP (S.F9~

GSK4p I

,GSK411 (s.t9)

LIC81

LKBI

~~ .. -, ~ r ;. j . I \

.... L. ~

'" ..... - . .

I a .

· •. ptjilB1 ·C~ 1:1"",""

ICorI.iCoI; .... Sj

lM"'ln ... -_ ... 1,

in combination with increasing concentrations of 8-pCPT-cGMP, and in the absence or preSence of IBMX or roliprarn. (f) Inhibition of PKG by KT5823 resulted in LKB1 and GSK-3~ phosphorylation. lmrnunoblot analysis was as in (0.

www.sciencernaq.orq SCIENCE VOL 327 29 JANUARY 2010

549

RESEARCH ARTICLE

also regulate AC/sGC or selective PDEs (29, 33). This hypothesis is supported by the finding that KT5720 alone not only reduced the PKA activity, but also increased cOMPand reduced cA11P, whereas KT5823 alone elevated cAN[P/PKA and reduced cGMP (Fig. 20).

cGMP antagonizes the phosphorylation of axon determinants. To further explore the rnechanisrns underlying the antagonistic actions of cAMP and cGfvlP 00 axon initiation, we inquired whether this antagonism is reflected in the phosphorylation of LKBI (12), GSK-3~ (8, 16), and

B

a;

E

...

o

.s

«I c Ql

...... ~. 1'°1'·

IX' 0.8 I&.

b ...... ,-_.....--_".--.......- ......... -.'

a 10 20

e 14

...

o

c 1 .. 2 1'1:1

c

!Ie 1.0

': "°1

. ~ 0.8.

c__ 4 ...,...-0---'.-0-......-""'2 ... 0--.

E -,

! '.2

c 10.

~

III 01..°8'"

t

It:

II-

~-4-- __ ~ __ -- __ ~

Ca [

1 41.

1.2

10

cGMP~AM~bead 1

1.2

o

1.2

10 20

Time (min)

Fig. 4. local cAMP elevation induced long-range cAMP reduction, but local cGMP elevation had no long-range effect. (A) YFP fluorescence and FRET signals for cAMP in an ICUE"expressing hippocampal neuron at 16 hours, at different times (in minutes) after the contact by a forskclin-ccated glass bead (dashed circle) ill the bead-contacted (upper panels) and noncontacted (lower panels) neurite. Scale bar, 10 urn, (B) Average FRET signals for cAMP, PKA, or cGMP in cells expressing ICUE (a), AKAR (b), Of cGES'DES (c), respectively, detected at different locations along the Icrskolin bead--(ontacted neurite (1 to 3) and a II other neurites (4 to 6). For bead-contacted neurites, the signals we re measured at the bead contact site at the tip (1) a nd at sites 113 (2) and 2/3 (3) neurite lengths from the tip [see illustration in (All. For all other neurites, the signals were measured at sites 113 (4) or .213 (5) of neurite length from the soma, and at the tip of the neurite (6). Data represent average ± SEM (n ;=; 5 to 10 cells, including all neurites). (C) Same as in (B), except that the bead was coated with cGMp·AM.

550

o

Akt (8, ]5), which are proteins that promote axon formation after their phosphoryl ati on. Immunobloning of lysates of cultured cortical neurons using phosphory lation sire-specific antibodies showed that elevating cAMP synthesis with forskolin caused a PKA-dependent increase of LKB 1 phosphorylation at serine 43J (S43l)and of GSK-3~ at S9 (Fig. 3C), and this effect correlated with the increased level of total LKB J (l2} Furthermore, when cultures that were plated on F "cAf\.1P stripe were transfected with siRNA to LKB I (12) and examined at 60 hours, the polarized population of these trsnsfected neurons showed reduction in preferential axon initiation at the stripe boundary (fig. S2A). Forskolin caused 110 detectable Akt phosphorylation at S473 (Fig. 3C). Conversely, elevating cGMP by bath application of 8-pCPT.cOrvfP did [lot cause detectable reduction of LKBI and GSK.3~ phosphorylation (Fig. 3C), probably because of'the low resolution of the immunoblotting method. However, in REK-293T cells expressing an LKB I construct, 8-pCPTcG.MI' markedly reduced LKBI phosphorylation and the total amount of LKB I (fig. 82B). Activation ofPKA activity byfcrskoun was confirmed in hippocampal cultures by a peptide-based assay (fig. S3, A and Ca), which also showed a dose, dependent reduction of the basal PKA activity in these cultured neurons by 8-pCPT-cGMP treatment (fig. 83B). The 8-pCPT-cGNIP effects on the phosphorylation and stability of LKB I depended on the PKA site 431, because they were absent in HE K-293 T cells expressing LKSI with a serineto-alanine mutation (12) at this site (fig. 828) .

We further examined the mechanism underlying the effect of cGMP on PKA -dependeot LKBI phosphorylation, Bath-applied 8-pCPT-cGtvIP dose-dependently reduced the phosphorylatton (at S431) and stabilization of LKBJ induced by forskolin in cultured cortical neurons (Fig. 3D) . This antagonistic cGtvIPeffect was abolished by either IBMX or rolipram (Fig .. 3D), indicating that cGIV[P reduced the cAMP level primarily by PDE4 activation .. The antagonistic effect of 8-pCPT"CGtvrP on LKB 1 phosphorylation was largely absent when PDE-resistant Sp-8-Br-c.AMPS was used instead of forskolin (Fig, 3E)_ The peptide-based PKA assay also confirmed fhese findings in cuhured hippocampal neurons (fig. 83), Treatment -with KT5823 resulted ina slight elevation of PKA activity in hippocampal neurons (fig. 83Cb) and LKBI phosphorylation/stabilization in cortical cultures (Fig. 3 F), consistent with the fRET imaging result showing that PKNPKO inhibition could modulate the cAMPI cGrvrP level (Fig. 2).

Perturbation of cAMP/cGMP affects neuronal pelarizaticn in vivo. We also investigated the role of cAMP/cGtvrP in the development of cortical neurons in vivo by elevating cAMP and reducing cGMP ill newly generated cortical neurons ill rat embryos. This was achieved by in utero electro" poration (34) of constructs expressing enhanced green fluorescent protein (EGFP) and specific slRNAs against either cAMP-selective phosphodiesterase 40 (PDE4D) (12) or soluble guanylate

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

cyda~e-~l subunit (,sGC-~l) (fig. SIB) into a subpopulation of neural progenitor cells at embryooic day 18 ([18). We found that cortical neurons expressing fhe si:Ri'lA against either PDI:!4:D or sGGPI exhibited impaired radial migretion to the cortical plate (CP) in E21 embryos (fig. 54, A to C), Whereas most neurons bad a single-neurite (unipolar) or two-neurite (bipolar) morphology in control E21 rat embryQs,a. large percent<1ge of PDE4DCir sGC.~1 siRNA-expressiQg neurons Were multipolar (fig. S4, D and E). The polarization defects correlated positively with the level of s,iRNA expression (fig. S4, F to H). Similar polarization and migration defects were also observed fur PDE4D or sGG~ 1 siRNA-ex.:pressing neurons in postnatal day 0 (PO) rateortices (fig. S5). Thus, elevating cAMP and reducing cGIv1P produced simi lar defeers in developing cortical nCUIOns in vivo. However; whether there is acausal ret<itionship between the tightly linked processes of neuronal po larization and radial migration during this stage. of development and whether cAMPI cGMJ' signaling is involved in both processes remain to be further elucidated.

Local cAMP elevation Causes long"rallge cAMp suppression, The reciprocal regulation between cANlP and cdMP could facilitate neurite differentiation along the axonal and dendritic route when cAMP and cGMP are locally elevated, respectively, To examine how rhis local cAMPI cGIvIP elevation affects the global cAMP/cGM!' signaling of the entire neuron, we manipulated a glass bead coated with furskolm (SOM text) into contact with an undifferentiated neurite of hippocampal neurons expressing the FRET reporter fur cAMP or PKA activity. We found a marked and persistent increase in. eAMP or PKA signals at the bead contact site (Fig. 4, A. Ba, and Bb, and fig. S6) and transient cA.MPfPKA signals alongthe

A a

shaft of bead-conacted neurites, with decreasing arnpliltide and longer delay of onset at more proximal locations toward the soma (Fig. 4, A, Ba, and Bb, and fig. S6). The apparent spread of cAMP!PKi),_ activity along the neurite was not caused by the diffusion of forskolin exrracellulady, because no signal was observed when the forskolin-coated bead was plaoed in close pIOXim ity (<2 ),J.tll) but not in direct contact with the neurite (fig, S6). After the forskolin-bead contact, we observed a striking reduction of cAMP/PKA signals in all other neurites (Fig. 4, A, Ba, and Bb, and fig, S6), with increasing extent of reduction Bod a longer delay of onset at more distal locations. In a reciprocal manner, in cells expressing the cGMP rep alter cOBS-DES, the r;:GMP signal was reduced at the forskolin bead-contacted neurite but increased in all other neurites, with a pattern exactly mirroring tllat of cANIP!PKA (Fig. 4Bc).

When, similar experjmems were done using a glass bead coated witb1he urembrane-permeehle IflMP-AM (aclllDxymethyl ester), we found a.local elevation of'the cGMP signal and a reduction of the cAMPIPKA signal at the bead-contacted neurite, but a complete absence of any long-range effect an either cGMP or cAMP/PKA in all other neurires (Fig, 4C and fig. S6} Because the cGMP-AM bead also induced a local cAMf!/PKA reduction" the lack of effect in other neurites indicates that the longrange cAMP sign?1ing is unidirectional, rriggered only by cA.M:P elevation. 'The existence of longrange self-down-regulation of cAMP but not L'GMP immediately suggests a mechanism fo!' the Kumatioo of one axon arid multiple dendrites: Axon induction by local eAW elevation leads to suppression of axon f011lmtion and promotion of dendrite ti;!lJl1Ilti.onat a 1I other neurites, whereas dendrite induction by localcG!v[P' elevation has no longrange eftect

'8

RESEARCH ARTICLE

I

Differential growth reglilation by cAMP and cGMP. Polarization of cuhured hippocampal neurons begins with growth acceleration of a single undifferentiared neurite (1. 2, 4, 21), suggesting that axon initiation roay be achieved by signals acting directly on specific eytoskeletal components (35-37) that accelerate neurite growth. Differential gr~1h regulation may ool'lllibule"to the antagonistic at:tioo by (:AMP and cGMP on alton/dendrite differentiation. We thus examined the cried of global manipulation ofcAMP/cGMP activities 00 the growth of undifferentiated neurites and axons/dendntes. Bath application of'various pharmacological agertts at 2 hours after cell plating and neurite lengtb measurements at 9 hours. showed that global cAMP elevation (with forskolin) or cGtvlP reduction (with ODQ) enhanced the growth primarily of one or two neurites (Fig .. SA), whereas global. cOMP elevation (with 8-pCPT-cGMP) or cAMP reduction (with SQ;22S36) res uhed in relatively uniform growth promotion of all neuritcs (Fig. SA). The asymmetry in neurite growth promotion by cAMP elevation lIlay reflect the dominance of a single neurite that first acquired the highest cAMP elevation, together wah lo~g-Tllrrge self-down-regulation of cA11P thar SllPP1"eSSed the growth of most other neurites, By 48 hours after plating, global elevation of cAMP (or reduction of cGMP) increased the axon length but reduced the dendrite length> whereas gJQb~1. elevation of cGNIP (or reduction of cAMP) had opposite effects (Fig. 58), consistent with (lAMP/clJMP antagonism. These results also suggest that axon and dendrite growth involve cAMP-:and eGMP-dependent activation of specific cellular components; respectively.

DisclIssion. Many extracellular factors known to regulate neuron pol arizati on, inel udin g brainderived neurotrophic factor (BDNF) (8, 12), nave growth fgctor (NGF) (9), Sema3A (38), netrin-]

fig. 5. Differential effects of cAMP and cGMP. on the growth of undifferentiated neurites and axonsldendrites. (A) Hippocampal neurons were incubated with various compounds at 2 noursafter plating, and neurite lengths were measured at'9 hours. (a) Composite tr'd;cing5 of aline urites from 10 ra ndo mly sa m pled he (Ire ns in control a nd treated c u Itl) res. Sca Ie ba r, 50 urn, (b) Percentage ,difference in neurite length (average ± SD; n = 3 to 5 cultures, 7.0 to 100 cells each) relative to the average neurite length for 100 neurons in parallel control cultures, for up to the first seven longest neurites (blue) and for all neurites (yellow). (B) Axon and dendrite lengths at 60 hours (compound added at 2 hours), determined by immunQstaining with axon- and dendrite"ipecific markers, smi-312 and MAP2, respectively. (a) Composites of tracings of randomly sampled cells taxons, 25 cells, green; dendrites, 1S cells, red} in control and treated cultures. Scale bar, .100 (axon5) and SO' (dendrites)WIl. (b) Percentage difference of 'axon and dendrite lengths (average L SD; n = 3 to 5 cultures, 70 to 100 cells eath) relative to those in 100 control neuron~ in paraLlel cultures.

www.sclencemag.org SCIENCE VOL 327 29 JANUARY 2010

551

(1O). Wnt (ll), and larrrmin (6, 7), could modulate the cAM.P or cOMP level ill. 1l6UlUilS (12, 38 43). Localized expos we to (me or more of such factors may create R cytoplasmic asymmetry fur axon/dendrite initiation. Perturbation of the cAMP/cGMP level resulted in neuronal polarization defects in the developing cortex (figs. 34 and S5). but the cAMP/cGMP-modulating extracellular factors I,har are responsible for pe lariz.ing the neurons in vivo remain to be identified. In a model for distinct actions of local versus global cAMP/cGW signaling (fig, S7), we propose that axons and dendrites are induced via localized cAMP and cG1V1P signals, and local antagoni~11c interactions between cAMP and cOMP pathways ensure that axon initiation is accompanied by the inhibition of dendrite fOl1nation, and vice versa. The cAMP s@ml acts through phosphorylated LKB I (J 2) and OSK-3~ (8, Hi) and their downstream effectors, which may converge witb the PI3K pathway at several levcis to promote -a,"\(DU initiation US. 19.,. 44-46), whereas the cGMT' s.igllal suppresses 8){OO formation via reciprocal down-regulation of cA1\.fPfPKA-deptmdent phosphorylation ofl.KBl and GSK~]~ (Fig. 3), as well as specifically promotes dcndrirc gro~1h (Fig. 5) vID cellular processes yet to be identified (fig. S7).

Long-range inhibitory signaling in neurons has been reported, Local contact of a neurite with a larger oell (47) Of laminin-coated surface (6, 7), or local perfusion of's neurite wnh forskolm (4?\), all led to growth inhibition of distant neurites, TIllS mhibitory effect may be caused by the long-range self-suppression of cAMP, although the mechanism underlying this long-range cAMP selfanragonisrn remains to he elucidated, The. absence oflong-rnngesignaling resulting fiom local cGtvIP elevation suggests. the framework of axon dominance sjgnaRng In the coordinated axon/dendrite

differentiation; whereas local cAMP/cGMP reciprocal regulation helps tocbaanel the 'differentiation process along either all axonal or dendritic mute, tile long-range cAMp self-suppression ensures the formaticn of only one axon,

Referen res iI n d N IJ teo;

1. N. Arimura, K. Koibu(hi, Nat. R~v. NeuroID, 8', 194 {lOOn

z. A. p,_ B~mes, D. Solecki, F. POlleui, ClI". Opf/). NeurobioL

U, 44 (2008).

,_ F_ [_ d. And, et 01, Na/Ure 436, 704 (2005)_

4. K. Goslin, G, Bankef,). Cell BioI. 108, 1501 (198-9).

5. Cl<lcobson. B, Schnapp, G. A. Banker, Neuron 49, 797 (2006).

6. T. Esd1. V. lemm on, G. II ~nker, j. Neuro,ci. 19. 6417 ( 1999). 7- C. M"n~ger, N. Mmura, Y_ Fuk~T;!, I\. I(aibuchl,

j. Neurochem. iJ'l, 109 (2004),

a. 1. YosMmu rilet at, Cell a2 e. 137 (,;: 0 OS)

9. ]. $. Do 54 Iva , T. Has@g.wa, T. Miyagi, C. G. DOlli, 1- Ab~d- Rodrigu~, Nat. NeU(9Ki. 8. 606 (200~).

10. C. E, Adler:, R. O. Fetter, C. I. Bargm~nD, Nal. Neuros(i. 9, 511 12.0G6)_

11. M. A Hilliard, C, I. B ~fgma nil. Dev. Cell 1 c. 379 (2006). 12 _ M, S~elly, L (. ~(ed dol, S, He i ls ~'Qm, G_ Su m bre,

M. M. Poo, Cell 129, S65 .(2007).

13. /III. rival, 5 _ Ban da ,a" T. I neue, T. Mey" f, Curf 8 lol. 1.11,

44,(2008)-

14, M,. Tmiy3ma et at" 1- C~I{ Bio/. l7S, 147 (2006).

15. S. H. Sh;, L, Y. Ian, Y. N. [an, (elflU, 63 C;!003).

16. H. Ji"ng, W: Guo., X. Liang, V. Rao, Cel/1Z0, 123 120051.

17. J. C. Schwa mborl"l, A. W. Piis'hel. Not NMOSO. 7. 923 (2004). is. M, Ki,hi, Y_ fl._ pa~, I, G. Crump, 1- R. S.me" 5t/ellce 307,

929 (2005).

19. A. P. Barnes el ai., r~11129, 549 (2007).

20. c. G. ncnt, C. A. Su~ivan, G. A. eanker,l. NeUrDsd. 8, 1454 (1965),

.21. A. M, Craig, G" lianlwr, Annu. Rev. Neufl)!Ki. 11, l~ 7 (1994).

22. A_ M. lohof; /III. Quillon, Y _ D"n, /III. 111_ Poo, J. NellrG_,ci_

uz, 1253 (1991).

21. M. i'lb~iYartoa flt Qt. NaMe 423, 990 (2003)_

24. III. ButtrMOr, 5, A. SiegeibaunU. NeiJropl!ysio/. 90, S86 (2003), "25. H. (_ H~r1Zell. R. Pischmeist~r) NaMe 323., 273 (1986).

26. A. Hempel. T. Noll, I\. Muhs, H M. PipN, 11m. 1. PhysioL no., H~264 (1996)_

27. C. Peilil'Rasgado, V. A Kimler, K. D. M[Gruder, 1. TIe,

H. Rasgado-Flor-es, Am. J. physiol. 7.67. C1319 (1994)_ zs. N" D_ Goldberg ~toLAdv, (yfli,Nudeatid/! Res,S, 307 (1975),

2'9. M_ 2~(C<ilD, M_ A. MdVSf' io n, fire Res. Hm, 156~ (2007L 3 O. l. M. Di Pilo 10, )(. Chen g. ] .. loa n 9, noc. N att Amd: Sd U_S.A 101, 16513 (2004).

3-1. V. 0, N ikolaev, 5. G. mba iyil~, M. J. tohse. Nat. Metbods 3. 23 (2006)

32. j. lha~g. V . .Mo. S. S. Taylor, R. Y. T~~l"\r rroc: Notl. Aead.

SCi. I},S.A 91r, 14997 (21)01).

3-3. F. S. Menni!i, W S_ Farocl, C. J. Schmidt Nat: Rev. Drug

D;;coV, 5, 660 (201)6)_

34. T. Saito, N.ll'a"ka15uji, ow srot. 240, 237 (2001). 35_ F. 6,"dk~, (_ G. Dotti, SCieJ1ce283, 19-31 (1999).

36. H, Witte. D. lIIeukir'h~n, F. Bradke,j. C~II Bioi. 180,619

(200S)-

37. H. Witte. F. Bradke.,Curr. Opin. Ne.urobio/. 18,479 (Z008). ~1l_ F. Polleux, T. Morro'll, A_ Ghosh, Nalure 404, 567 (2(100)

39. H. 1. Song, G. 6. Ming, M. M. Poo, Nature 388, 275 11997).

40. G. L Ming ~( ol., Neuron 19,1225 (1997)

41, Y. GilO, E. Nikulina, W. Mellado, M. T. Filbio, I. Neurom: l3, 11770 (2003)_

~2. K. Toga5hi et ot.. Neuran 58, -694 (2008).

43_ V_ H. Hopket, D, Shl!Wlm, M_ T"""iBt-lo'"gne, M M- Poo, e. Holt, Na/we qOl, 59 (1999).

44. O. O,si POVOl, N. Bo rdl'~Y, It A_ De P-; nho, ]. B. A "feen, Nat. (ell BfoL 5, 889 (2003).

45 Y. I. ~hbi I!I al., &;n~ Dev_ 22, 2485 (2008)_

46. 1. Wildon ger, L Y. I~ n, t. i'I. Ian, Genes oe« 22., 24 4 7 (2008).

~7. O. j, GDldberg. s Schacher, Dev. BfoL 124. 35 (1987). 48_ 1- Q_ Zh~ng, t. ~h~ng, M, M. PQo,), Cell BiaL 1,27, ~693 Wl94).

49_ We tnan k R_ Tha\;ar, S_ Li, M, lIIo,lr, a nd D. Liepm;, nn (VniVl'lSlty of California ill Berkeley) for help WiI!1 pol.y{dilllethyl5flm':;nel O"litroflui!llc mold" J, Zhong

(] ahn > HoPI::i 11> U nil/!'r>lty) for the !CU E iln dAKAR FRET p wtles: M. 1- Lohse (U nivef$lty of .WIl ilburgl for the (GESrOE~ FRET plob~ M f.elll'r and S_ l'aU\<l1 Mfiiver5ity 01 Califrn'ni" ~t a~rl<el~) for "dvic~' on

FRET i moging and bead (Aatlrrg; and M. Hung lor help wi th wlillre preparanons, Thl) wOJ"l w<I_I.luppotttd in part by il gtilnt tram the lII.tional IIllilute5 of Health (N5-227611)_

Supporti fig Onlifl!, Moi!t!'ri<il

\'/WIY. sriencern ag . Drg/,giltDn I ell tlfu Ul3 2 7/r; 96S/~4 7/0Cl Ma tfrial, ;3 n d Me tll 0 ds

SOM T~~t

Fig,. S 1 to 57

27 JlJly 2009; octepted 10 December 2009 10_112U,dim,e_U 79735

Phase Transitions of Adsorbed Atoms on the Surface of a Carbon Nanotube

Zenghui Wang, Jiang Wei, Peter Morse, }. Gregory Dash, Oscar E. Vitd'les, David H. Cobd:en*

Phase transitions of adsorbed atoms' and molecules (In twoLdimensional substrate~ are 'well e-xploreq, but similar transitions in the one-dimensional limit have been more difficult to study experimentally. Suspended carbon nanotubes Gill act as nanoscale resonators with remarkable electromechanical properti ss a nd the ability to detect adsorption at the level of single atoms. We used single-walled carbon nanotube resonators to study the phase behavior of adsorbed argon and krypton atoms as well as their coupling to the substrate electrons. By monitoring the resonance- frequency in the p~esence of gaS£!5, we- observed the formation of rnonolayers on. the cylindrical surface, phase: transitions within them, and simulttineOl!S modifiCiltioll of the e[ettric.a I conductance.

F ilms Of.' atoms or 1TI.ole.oUI.es adsorbed on. surfaces exhibit many kinds of ordering that reflect interactions between the adsorbates as well as with the surface, One of

552

the simplest model systems, rate gases on bulk exfoliated graphite, exhibits a buge range of twodimensional (20) phenomena within the first adsorbed layer, including 2.0 melting, transitions

between solids that are either commensurate or incommensurate with (he graphene lattice, and critical behavior (I, J). Here, 'we explore the phase behavior of argon snd krypton on carbon nsnotubes, wh~~ the dimensionality of the subsuste approaches the 1 D limit, aud report effeers on the electronic conductance that reflect adsorbate-substrate interactions. We map out this phase behavior using nanomeohanical resonators based on individua! suspended single-walled nanotubes (SWNTs) (3-6)- This platfcrm cornbines remarkable electrical (7-9) and electromechanical (10-1 2) properties with extreme mass sensitivity U 3-15) and allows simultaneous measurement of both the. precise amount of adsorbed substance and its cffee: ou tbc electrical

D€partment 01 Ph!{Sii:5, University 01 Wa,hington, 5~altl~, WA 98195-1560, USA

"To Whom (orre,pondem:e ,hallid be addressed, E'mail: ccbden@u.wJ,hington.edu

29 JANUARY 2010 VOL 327 SCIENCE www.sciencernag.org

properties. Our approach avoids problems of heterogeneity that have complicated previous explorations of this regime based on eonventiona] techniques that require the use of bulk naoourbe samples U 6 . J 8),

Suspended nanotube devices were made by prefabricating the electrodes and trench, and then growing the SWNTs 111 the last step (19) to avoid any chemica] exposure that might contaminate the pristine nanotube surface (Fig. lA). Mechanical resonances. appear as sharp features in the, current signal during frequency sweeps (20)_ The frequency f"", of each, resonance decreases with increasing pressure P as gas adsorbs (20), as 'illustrated in Fig. IB for Kr at 77 K. Equilibrium is established in seconds after a small pressure change (Fig. IC), and 1"= is a reproducible function of P (Fig, I DJ.

We deduce the adsorbed mass by assuming thai at a given temperature jiili varies as p -ll2, where tbe total mass per unit length p is the, sum of that of the bare nanotube, p(). and ofthe adsorbates, "p. (This assumption will be justified

B

C 85

350 400

(h1 .z.)

below.) The fractional mass increase of the nanobalance is then 6.p/po = (jiJr ra,i I, where jij = lim» • Id_. Because j~ is determined separately at each temperature 1, any variation of the resonance frequency with T, such as might result from thermal expansion. is factored out. If mc and m,,{], are the atomic masses of carbon -and the adsorbed speciesrespectively, then the quantity

\p = (Aphn'(IJ/CPoIwc) = InC1mo& [Urlf;"$)2 -IJ

(I)

is the number of adsorbed atoms per carbon atom; Fig. IE shows an example of'an isotherm of qJ versus P dedved in this way.

111.e assumed p -Ir:l scaling off_ requires that the change in elastic properties is negligible compared with the fractional change In mass" This assumption can be made. because the covalent C-C bond is two orders of magnitude stronger than the V!:IO der Waals attraction between adsorbates. We also must assume that the mass i5 distributed uniformly over the nano-

o

E

1

P (Tarr)

if ,

-a.:!5V

-7.55V

o -6.J15v

Fig. 1. Detecting adsorption on a single vibrating nanotuba (A) Electrorl micrograph of a nanombe spanning a 2iLmwide trench between Pt contacts. (S) line trace~ of the mixing current signal showing resonances shifting with Krgas pressure for device YB3 (O.5-~lm gap) at 77.4 Kand Vg = 958 V. (c) Grayscale of the mixing current signal for a series of frequency sweeps taken during a sequence of Kr pressure increments, one roughly every 3 min, demonstrating rapid eql.litihration, for device YBl (2ilm gap) at 77.4 K. (OJ Variation of resonance frequency with Ar pressurl) for VBl, showing high reprodudbllity. (E) Variation of coverage parameter <p obtained using Eq. 1 with Kr pressure for VBl, demonstrating little sensitivity to gate voltage.

Fig. 2. Isotherms of 'COverage parameter <p for fAr on device YB3. The temperatures are 66.1, 67.7, 68.8, 71.Q, 73.9, and' 77.4 K. The large ste p 0 t~urs withi n the SLj percriticalituid (F), with a first-order L-V transition expe cted below a (rili ca I point 'at -0.01 Torr. The smaller step, a sepqrate measurement of which is shown in the inset, 0(( UfS.at the Ira nsitio n toa n I S. Dotted and dashed I in es indicate boundari es of coexistence r6giof15.

0.2

0.0 0-01

.- .. ---.-----.--~~---- .. -- .-.----~.~~

FtW __ • __ • __ ._~~--

_. -._- _ .. __ .... . ~~~

F 6G, ~;f /-,?

'Iltr f ,/ I .r:

If! ) ..... / ...... , TOfl

$ ··.//n<tK

I

0.1

1 P(Torf)

REPORTS

tUbe surface, which would not be the case if part of the surface W6Ill contaminated. or if a denser phase appeared preferentially at the ends or in the middle in respocse to long-range forces. Nonunifunnity wouldcause different vibrational modes to shift in different ways. However. we have found tl:!,atfV(tts does not depend on which mode is' used a 11 d -a 1 SOl hat j t is insensitive to substrate gate voltage II g> up to about 8 V (see, for example, Fig. IE). Taken together with the results described belnw, these observations indicate that many of our devices consist of single-walled nanotubes in which qiis a good measure of the coverage (number of adsorbates per surface atom] and that the coverageis uniform.

Isotherms of (p versus P for AI are dominated by .a large, smooth step, as shown in Fig. 2 for device YB3. A similar step is well known in COI1- vennonal vo tumetric isotherms on bulk exfoliated graphite and is characteristic of the densificatiou of a supercritieal 2D fluid (P). which occurs above 56 K" the 1D liquid (L)-vapor M critical point of Ar 00 graphite (2. 2}, 22). The colder Ar isotherms also show a second, smallerstep at 'P 0::. 0.24 (see the inset of Fig. 2). A similar second step is seen for Aron graphite (22) when the fluid freezes to a 20 incommensurate solid (IS). Accordingly, we antioipate an L+V coexistence region et lower teroperaturss as indicmed by the dotted line. and we expect the F+IS coexistence region to have the form indicated by "the dashed lines. We note that corresponding features occur at higher pressures on nanotubes than on graphite, reflecting the expected weaker binding of atoms to a nanotube surface than to bulk graphite (23).

We now discuss how this phase behavior changes for Kr, which is larger and more polarizable than Ar. Isotherms ofKr on device VB3 exhibit a dramatic vertical step followed by two smaller steps (Fig, 3). Again, these resemble cooventional volumetric isotherms of the same substance on exfoliated graphite (24) but shifted to higher pressures. The size and sharpness of the large step implies not only a first-order phase transltiQn but also excellent substrate homogeneity; consistent with the absence of grain boundaries or imperfections on the surface of this SWNT. The first plateau, between the first two steps, is narrow and easily missed; it is resolved in the 77.4 K data here but not the 73.7 K. data,

Whereas AI does not form any commensurate phase on graprute, Kr condenses 6:0111 a lowdensity :m vapor (V) to a commensurate. solid (CS) with one Kr atom per six C atoms (25). as indicated in the left inset to Fig. 3. At higher pressure, it con verts to 1In IS (26). The first plateau in OUl· Kr iSQthel1TIS, whenever resolved, is centered at 'P =1/6, corresponding to the COy, erage of the commensurate solid and implying that the first step is-a V-CS transition. The highest plateau reached is. likely to be the IS, whereas the intermediate plateau is ofunknown nature but may be related to a proposed reentrant fluid phase (17).

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

553

I

REPORTS

The fact that (p takes the value of 116 fur the commensurate phase strongly indicates that tbe assumption behind Eq, I that!."" oc p 112 is justilled fur this device. SOme other devices, including VB I (Fig. I E), showed similar responses but exhibited relatively small values of 4'. This difference can be ex pI ained if these deviceshave a reduced ratio of availilble surface .area to nanorube mass, which would be the case for a multiwalled naaotubc or if part of the nanotube's surface were contaminated.

The identification of commensurate and incommensurate 20 solids on the cylindricalnanorube surface raises many interesting questions. The soHd is subject both to the curvature, which breaks the isotropy of the graphene lattice, and 10 the cylindrical boundary condition. It may be rolled seamlessly like tbe underlying graphene, O!" alternatively it may contain a. domain. wall running along the nanotube, Forthe CS, which is in registry with the carbon surface, the seamless case occurs only when {N M)/3 is <In integer, where (/'f,MI'J is the nanotube's roll-up vector, lnte.r<:stingly, this is precisely the same condition as for the nanotube to be metallic (28). From the

Vg dependence ofits conductance G (right inset to Fig. J), we ideutify YB3 asa small-gap metallie nanotube that obeys this condition

Because W6 can measure-the electrical properties of the SWNT simultaneously with the mass absorption, we can quantitatively investigate the coupling between adsorbates and electrons, This is nor possible in conventional adsorption experiments, Figure 4 shows the {P~p isotherm of Kr at 77 K for another device, YB8.1t exhibits a large, sharp step at.sbout 161llTorr similar to that into YB3, (The smaller step pressure in YB8 can be explained by a larger binding energy dire to a larger nanotube diameter} The upper left Inset shows G·Vg characteristics measured at pressures on either side of the phase transition. The conductance is suppressed at the higher pressure for positive Vg• The other trace in the main panel shows the resistance measured at a fixedpositive gate voltage. It increases gradually at low eoverages and jumps suddenly at the transition. This behavior is reproducible and reversible.

One immediate consequence of'rhis observed coupling is that we can investigate the dynamics of'such a phase transition. 111e right inset ID Fig. 4

Fig. 3. Adsorption of Kr on device VB3, .showing firstorder p ha se tra n!litions. (The 73.7 K isotherm has incomplete data at ttl e dashed Ii ne.) The dotted horizontal line at <p = 116 corresponds to the expected coverage of 1 iJdsorbate p er6 carb on atoms in a commensurate solid On a deem SWNT. The left inset shows the commensurate a rra ngement of adsorbates (shaded d rdes) sitting on ev~ ery third (arb on hexago n. The ri ght inset shows con d ucta n ce versus gate voltage (solid: vamLlm; dashed: in air, where the re is h yslere5is in dkated by the arrows) at room temper· ature; whose form is. that of a srna[l-bandga,p nanotube.

~ (l) C>

ro 0 _ <v .1

> o

o

v

0,0 ~

-

.'1,

......

o 10 V (V)

II

0.01

Q.1 P (Torr)

1

Fig. 4. Combined mass and transport mgjlsu rem ents on d evi (9 YB 8 (l·f.1m gap) exposed to Kr at 77 K. Measurements are shown of both coverage parameter q> (l!if!: axi~) and resista nee R (fig ht axis) as a function of pressure. The left inset shows conducta nee versus gate voltage at pressures just below (15 mTorr) and a Dove (17 m T grr) th e tran&itiof\, wi til an arrow indimting the gate voltage of +3.1 V at which R was measured. The ri 9 ht inset shows r esi sta nee versus time during a rapid upward pressure sweep across the tra n smon.

, ...... ;'r ...

o co - ,;:-=-----..----~

o 10

P (rnTclfl

554

shows the resistance morritored with a 10·ms instrumental response time as the Kr pressure is increased rapidly across the transition. Tile step OCC\J1'S in about 001 s, indicating that this is the iotrinsie time scale of the phase nansitio» on the SWNT.

Analysis of'the C t- Vg_ characteristics indicates that this SV\lNT has a small gap (-60 meV) and lhat the couductance is limited by tunneling across the gap at positive Vg but not at negative Vg (20). The larger decrease in conductance at the phase transition fbr positive Vg could thus be related to an increase in the gap. We note that the CS has a reciprccal lattice vector that connects the Dime paints in the graphene Brtllouin zone, and hence coherent scattering from commensurate Kr offers a possible mechanism for modifying the gap Also, the K.r atoms will be statically polarized by the gate-inducedelectric field perpcndicular to the nanotubc stufaee (2U), thereby modifying their interactlcns with each other aria with the substrate at high gate voltages. Further experiments of this' type could yield many more important it)sigJns into the interaction of adsorbed substances with ihe electrons in graphitic carbon,

R.eferences and Notes

t- [, G" p~)h, 1- RLN.ld5, Phas~ Trafl>itiQf1~ if! SWjar;e Film,'Nato Ad\l<mcf d Sludy lnsti lutes Series B (Plenum, New Yort 1930), vel, 51"

2. L~W. Brudl, M. W. (01.." E. Zilremb.J, Phy,iml Msmptirm: Forces a[li'i Pnellam<!f1IJ_ (Qttfbld Umv.rsity PI!"', D):forn, 1991L

J. V. SillOflOva ~t ol., N~Iut€ 411, 284 (2004).

4. B. Wilk<1mp. M. Poo~ H. S" J" van de, Zao~ ~ral1G Lett. 6, 2904 (20(6).

s. H. B_ Peng, C. w_ (~ang, S Alo~i, L 0. Vu;Ninsky,

A. Z enl, "hys. Re\l. 1 en. 97, 081.20 J {20061. 6_ A. K, Hultel et at., NallG Lett. '.1,2547 (Z()09L

7 1. Ca~, O. Wong, H. Dai, Naf_ Mofllr. 4, 745 (20(5)-

11_ b. A. 5te~e, G. Gou, L_ P. KlluwennOllen, Nat. Nanll/J;(hnoL 4, 36~ (2;009).

9, V. 'V, lJeshp;mde et ai" 5cfM(e ~2'3, 106 (2009J_ 10. G_ A. SfeeLe rr al., 5rJel1ce 3l5" 1103 (2009).

1 L S- l,;j" 0 gn", Y. T ~ !'aka nov, [, Kin~ ret, D _ G)ncia -Sanchez, A. Bodltold, Science 325, nDI (2009).

12_ R. leturcq st at; Na: Phy". S, 327 (2009)_

13 K. jense n, K. K1 m, A_ ZeU!, teat: NanGtrc hnoL 1, 533 (2006!.

14. H. Y. Chlu, P. Hung, H, W. C. P'o>rma, M. Bo(kralh, Nono Lett. 8, 4342 (200S)

15. B. lJ!>.'la»M, D. G.!(i~"Samhel, 1\. Aguesca, A. Biichlold, NerIO (et/, 8. 3735 (200a!

16. W. Shi, J. K'. Johoson, phJ15. liev. !eft 91, 015504 (;!()O3l"

17 M. R. jo~nsan rt 0/., aiem. Pbys. 293, 217 ,(2003).

18. 1- (_ l.a sja unias, K_ B ajitkovi( 1 l. Sa uva j ol, P. Man caau, phys. Rev. lett. 91, ~2 5901 (2003).

19, 1.010, Q. w.~ng, D. w. Wang, H, 1. Dai, 5maU.l', 138 (2005).

20. Mateliill'; and methodt'3re avaaablf as 5uppOtting material an SdfflKE Online.

21. A. D. Migone, i. R_ Li, M. H. W. Chan, p-hys. Rev. Lett. 53,

ali) (1984).

2?- F_ ,MiIloq Phy.;. /.I!tt. 40, 9 (i979)-

23. G, Stan, M, W. (Ole, 5111. St:i. 395, 2BO (199B).

24. A. ThornY, x" Duval. J Chim. Plly;. 66, 1,966 (:t.969).

;:S, M. D. Chinn,S. C. Fein, Phys. ReV_ Lett: 39, 146 (1977)_ 26, s, c. Pai~, M D. (hinn, R. D_ Di~hl, Phys. ReY, B 21,

4170 (19BO).

27 E" D. SpechUt al., Z ph~ B Condens, Marter69, 347 (1987)-

28. ]. C. Charlier, x, Bl~!~, s, Iio(he, 8e~. Mod. Phys" 79, 677 (2~07)_

29. W e a~~n owledge us ef ul di>~u!.,wn. wi Il1 M. Bm:krOl th, H,,·Y, Chfu, M, W. CDI~, M, den ~ij" M. 5(hic~, and A. YMI d er z. ~ de. Thl.l work wa I supported by g r a 015 from N51'

20

29 JANUARY 2010 VOL 327 SCIENCE www.scieneemag.0rg

REPORTS

\OM R Q6Q./iO 7 8 <md 090769 Q)l tli" Am~riCil n C henil<:" I Society Pet!Oleum Research fund, the University· of Wash in gton (UW) Roya It)' Rese~ rrh fu nd, and ·a UW University Initiatives Fund fellowship. Pmlions ot this work were tlDn~ In the Uni\'~<>ity 01 (~lil\)rnia at S~nL'l Barbara r.lanofabritation Facility and in the UW

Nanotechnology' Center, whi~h M" parts of Ihe iIlSF- Figs. 5t !II 53

funded Natio~al Nanote(h~ology In frastructure Network. References

Supporting Online Material

WlVw,5c1e~(ernag,org/cgiICGn\"ntlfli\ll32 ;.1596 'I15S 2/00 ~a SeplPmber 2009; mepled 1 Oecemb~; 2009

MOl.terial> and Methods 10. l1l6/5dem:e.1 18.2507

factor peak. around 0.1 A-I (fig. s::n (:20, 21). Successive irtadiatien 00 the same spot yielded a series of Bra gg peaks, The relative positions of the peaks full ow the qlg"" ratios of I : ,.j3 : J4 : .,f7 : ...;g. : v'i2 (where q* is !he principal peak position), obar1\.Cter'istlc of a hi.ghly ordered :m hexagonal lstnce (space group p6mm). TIle. development of scattering profiles it; characterized by the emergence and continued increase in intensity of the principal Bragg peak, as well as the appearance of additional peaks at higher q values. Figure 1 C shows the 2D scattering patterns of the first and the-last exposures, revealing the disorder-to-order transition, The low volume fra~tion of tll41Dents (0.5 wt 'Yo) indicates that these bexagenally packed 10 ObJt!L'fS must exist as bundles. We measured the final values of full width at half maximum arid used the Scherrer equation to estimate a bundle size. The calculalion yields a value ofuhout 1 urn: 'however, given the size regime the absolu~ value obtained from the Scherrer eqvatiO.L1 is of questionable accuracy, The Debye-Sherrer ring-like pattern after x-ray exposure is typical of scattering from H powder sample, implying that these hexagonal crystalline bundles are randomly distributed in solution {Fig. ID},

We foand tbat this x-ray triggered structural rearraegement only occurred at relatively low concentrstions, Figure '2A shows a similar disorder-toorder transition observed at 1 WI: % upon continued x-ray irradiation, However, at higher concentrations (2 wt % and higher), the Braggpeaks could be observed without the need for continuous x-ray irradiation (Fig. 2, B and C). The hexagonally stacked filaments at 2. wt % and 5 wt % were stable during x-ray exposure, and their corresponding Bragg peak positions did not shift by more than 3% with accumulated exposure time.

These observations suggest '!hat hexagonal stacking of filaments at higher coneentration is a spontaneous process. arid not associated with x-ray irradiaticn. We pll)t the )hray profiles corresponding to the last exposure for various concenrrations in Fig. 20. Five to seven Bragg peaks ru-e registered 111 each scattering. profile and have the expected relative ratios of hexagonal structures, The Bragg peak positions shift smoothly to higher values of q with an increase in peptide concentration U1 both spontaneous and x-ray triggered hexagonal structures. These changes COT" respond to a decrease. in irnerfilament separation as the concentration rises (Fig, 2E) and suggests the spontaneous and x-ray-triggered hexagonal structures emerge through similar mechanisms"

To determine whether x-ray beam heating could contribute to this structural disorder-to-order

Spontaneous and X-ray-Triggered Crystallization at Long Range In Self-Assembling Filament Networks

H ooggang Clii,l E. Thomas Pas h lick, 1 Yun S. Velithko,1 Steven 1- Weigand,'Z An drew G. Cheetha ill, 3 Cl'Iristina J. Neww rnb, 1 Sa rnuel I. Stuppl.3.4,S ..

We report here crystallization at lo ng ranqe in networks of li ke-charqe supreme lecu lar peptide filaments mediated by repulsive forces. The crystallization is spontaneous beyond a g,iven concentration of the molecules that form the filaments but can be triggered by x-rays at [ower concentrations. The crystalline domai n~ formed by x~ray irr~djation, with interf La ment separatio ns of up to 320 angstrom S, ra n be stable for he urs afte r the beam is turned off, and ions that. screen '( harges on the fila rnents suppress ordering. We hypothesize that the stability of crystalline dom"i ns emerges Irom a balance of roepulsivf! tensions linked tt;) n<1tiv~ or x-ray-induced ~harge$ and

the mechanical compressive entrapment of 'filaments withina. network. Similar phenomena may occur naturally in the cytoskeleton of cells and, if induced externally in bi ological or artifidal systems, lead to possible biomedical and lithographic functions.

O ne-di1TI6I1 .. SiO. nal (JU). objects in S.OIUtiOl1, such as carbon nanorubes (1), filamen-

tous viruses (2), and rigid molecules (1), can spontaneously form orientaticnally ordered domains or networks as a result of their Shape, This excluded volume effect is useful in the design of devices, liquid crystals, high-strength materials, bioactive hydrogels, and other fimctional structures, In biological systems; the bundling, orientation, and mechanical networking of ID cytoskeleton components such as filamentous actin and microtubules mediate cellular events such as mitosis, protein transport, and signal transduction (4 9). Small-angle x-ray scattering (SAXS) can provide infbrmatioo on the size, shape, and symmetry of the internal domains of materials (l0,. J J) and is extremely useful to study struttures with the lengih .scales <;In D networks fonnecl by nanoseale filaments (4). However, x-ray beams can cause irreversible cbemical changes that lead to detectable products at high fluxes and .affect the formation of 3D structures (12 14) The. typical

mechanisms involve ejection of energetic electrons from molecules via the well-known photoelectric, Compton, and Auger effects and form fiee radicals and charged molecules, as well as chemical reactions (I5 19).

We report the spontaneous.aud x-ray triggered crystallization of supramolecular filaments. within 3D networks at unexpectedly large distances (up to 320 A). The filaments described here are funned by self-assembly in water of a symhetic molecule containing the short peptide sequence AiIl6Glli] (A6E:;) covalently grafted to an alkyl chain of 16 carbons (fig. S1) (20). A representative cryogenic transmission electron microscopy (crye- TEM) image is shown in Fig. I A, revealing cylindrical nanofibers measuring -10.2 A in diameter corresponding to twice the fully extended length of the molecules, The length of the filaments cannot be directly obtained from eryoTEM, but it is estimated to be on the scale of tens of micrometers. The combined effect of intermolecular hydrogen bonding among the peptide segrnerus and hydrophobic collapse of hydrocarbon tails in these molecules leads 10 formation of the 10 nanosnuetures in dilute solution (21,21).

We observed thai the SAXS profiles of 0.5 weight percent (WI: %) solutions dramatically changed upon continuous x-ray exposure. Figure 1 B displays 50 sequential SA XS prof les of the same spot with an exposure time of 4 s each, plotted on a double-logarithmic scale, The first spectrum (red) shows a typical scattermg profile of cylindrical objec1s ill solution, as' suggested by the l slope in the low q l-egion and a diffuse funn

] Deparbn ent 01 Materia Is 5d~~ ce a nd Engl·neeri nq, N onhw~~t~m UnIVersity, 222 0 Campus Drive, Eval1Sto~, I L b020S, uSA. 1DuPont-Northw5t~rn'Dow ColtilbO!iltiw At:C55 Team (DND·CAT) Synchrotron Research Center, Northwestern \..Ioiversi \y, Advanced photon Sou[(·e (AP5)IArgonne Natlonal L~bor~tQI")I 4~2·A004, 9700 50U~1 C~S~ Awnu~J Argonne, I L 604 39, U SA ~I nstitut€ tor Bi 0 N 0 notedm 010 gy' in Medi d n~, NorthwestErn Univer,ity, (hie~go, Il 60611, USA. ~Departrn ent of en em ishy, Nonhwertern Un iVe!sity, 22 20 ca mpus Dn"ye, Eva~5ton, I L 602 OS, USA. ~Depdl'im!i'nt of M ~di d n e, Northwestern UI;iversity, 22~0 campus DriVl!, Evaoston, IL

60203, USA. .

'To whom correspondence should be- addressed, E·mail: s·slupp@!1orthw€stero,edu

555

Www',soeneemag.org SCI EN CE va L 327 29 JAN UARY 2010

REPORTS

transition, We studied the 1 wt % A6EJ solution at25~ and 40"C. Ws found that Bragg peaks did not appear after thefirst 4~s exposure at 40°C and that the overall scattering profiles underwent a similar transition as those 0 brained at 25 °C (fig. S3) (20). Figure 2F shows the relative change in the g'" peak intensity plotted with respect 10 the exposure time; we observed no obvious difference between 25° and 40QC. In addition, Caffrey and co-workers reported only a O. T GOC temperature rise as a resultof continuous x "YJY exposure of Miili-Q (Millipore) water (16). Beam heatfig resuhs typ icall y in on! y a small temperature change because ofthe rapid dissipation of hear j11tO the local environment; therefore, one does not expect this thermal effect alone to contribute to the observed structural transition (fig. S3).

Another important observation was the reversal of the x -ray -triggered crystalline structures to a disordered Slate after removal of me beam. MaGroscopically, the 1 wt % solution in the capillary was clear before x-ray irradiation. After x-ray exposure (accumulated 200 S), the spot of interest became opaque (marked with a white arrow in Fig. 3A). This appearance of turbidity is indicative of a transition between two. states with different Structutal order, possibly because of scattering by crystalline filament domains. After removal of the x-ray beam, the solution gradually became clear and completely lost its opacity within about 40 min (Fig. 3, A to D .. and fig . .s4). TI1is reversible opacity was further investigated with scattering experiments. Figure 3 E shows the transition from disordered filaments (black curve, the first 4~s exposure) to the hexagonally ordered bundles (redcurve the last 4-s exposure), Two hours after removing the x-ray beam, we collected scattering profiles on the same location. We fonnd that all peaks corresponding 10 hexagonal filament stacking vanished (green curve) .. thus suggesting the Joss of positional order Restarting Hay irradiation causes the reappearance of Bragg scattering peaks (blue curve). This reversibility is also observed when adjacent regions to the originally irradiated spot are exposed" to x-rays, This experiment suggests that reversibility is. not caused by sim ple di ffusion of c h emi cally unm edified materia! in to th e origin ally inadi ared spot. I n fact, analytical high-performance liquid chromatography (liPLC) and mass spectrometry of samples after a large x-ray dose conflrrned thar chemical dam. age is Tiot involved in the observed reversible phenomenon (figs. SlO to S36) (20).

In order to gain more insight into the observed x-ray-induced phenomenon, we slUdied the effects of beam flux and photon energy. A longer exposure time was required to induce this disorder-order transition whcn the x-ray beam was gradually attenuated (Fig. 31"). However, in order to induce filament orderiug, a larger sceumulated dose [supporting online mated,,11 (SOM) text 52 3!'I.d S6 and figs. S5· and S6] (20) was needed when samples were exposed to x-rays at a lower dose rate (a more attenuated beam) (Fig, 30). Eventually. at a beam attenuation of 0.012,

556

the Bragg peaks corresponding to the hexagonally ordered filaments did not appear even witb an accumulated 1120-s exposure and a total dose about twice as large as that needed to induce the filament ordering with .:1 nouanenuated beam [4.39 x J 04 grays (Gy)] (fig. 56) (20). Radiation damage by x-rays bas been reported as more severe when. samples are "irradiated with a fixed accumulated dose but at a lower dose rate [the inverse dose-rate effect (18)]. Thus, in this context, if chemical damage did occur and the resuiting produets were responsible for filament ordering, the same accmnulared dose at a lower dose rate should have enhanced the effect, These dose rate-dependent experiments again support that a reversible process is involved. F.ililluent ordering can be induced by using x-ray photons with a lower energy (9 keY versus 15 keV) (fig. S7) (20); which suggests that much weaker beams can be used to promote ordering given that lower energy photons have a higher massenergy absorption coefficiem,

An unexpected feature of me observed hexagonal crystalline structures tanned spontaneously or triggered by x-rays is the large II spacing among filaments (Fig. 2]]). The center-to-center spacing is as large as 320 A. fOJ" 0.5 wt % solutions (x-raytriggered) (Fig. IF). Given tbc iOz-A diameter of

the filaments, the wall-to-wall distances measured range between 218 A and 90 A for concentrations of 0.5 wt % and 5 wt %, respectively, These distanees are much larger than those reported fur cytoskeleton filaments and DNA strands (4,24, .35). In these systems, the distances measured are in the mnge of 30 to 55 A and always in the presence of multivalent counterions th?t should introduce attractive forces among like-charge objects. The rnuhivalency of added counterions, not added in our systems, is critical to induce positional ordering thro ugh attractive fore as in t11 ese b i 0- molecular systems. In one example, Safinya and co-workers have observed both hexagonal and necklace bundles in microtubule filaments after add1qg multivalent counterions (4). 111e structures are viewed as resulting from the balance among electrostatic repulsion and attraction as well as hydration forces (repulsive) and van. del." Waals interactions. Because multivalent counterioos are not added in am: systems, attractive electrostatic inler"il.ctlonS. cannot be considered as offsetting repulsive forces among the filaments, At [he same lime, van der Waals forces are not expected to stabilize repulsions at the interfilament distances of hundreds of angstroms observed here, We therefore hypothesize that crystalline. stacking of filaments emerges from tlre minimization of

0.5 WI %

fig. 1. X-ray irradiation triggered crystallization of self-assembling filaments at 0.5 wt % aqueous solutions of a peptide amphiphile. (A) Representative (ryo-TEM image of filaments formed in O.S wt % solwtiot1~ of peptide amphiphile. (8) A series of 50 conseet<l1ve scattering profile~ of O.S wt % solution~ obtained from the same sample area with4~s exposure each (dose rate: 4806 Gy 5-1). The measurements were performed at room temperature with a monochromic x-ray at 15 keV, and the typical incident /i-fay flux on the sample was -1 x 101c photons per s wilh a 0.2 mm-by-O.3 mm collimator (samples were placed inside quartz capillaries of a diameter of 2 rnm). The first a nd· the last spectra are colored red and green, respectively. (C) The 2D scattering patterns corresJ:lOnding to" the first and last exposures in (S). (0) Schematic representation of Hay--triggered organization of peptide filaments from it diso rdered state to a hexagonally ordered state,

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.erg

repulsive electrostatic interactions balanced by other attractive forces associated with network S tructures .

To understand the observed disorder-to-order transition, we eonsidered first the origin of bundle formation, Our insight is that bundle formation fromfilaments with very high aspect ratios must occur during their supramolecular growth, because these extremely long and networked 61- aments are unlike ly to reorganize into bundles on short time scales once they are fanned. TEM reveals the coexistence of very short and very long fll aments at the v cry early stages of self-ass ern bly (Fig. 4EQ. Based on this observation, we hypothesize that long filaments funned at early stages create a Stable network that templates the formation of bundles as growth of short filaments continues" As depicted to Fig. 4, A to D, this "templating" effect Can be envisioned as further parallel growth of'fllaments around the originally formed network, At the moment the precursor network is formed, topological constraints miffing from interlocking of filaments would 1[111i1 their mobility within a tubelike channel ar0U11d each of them (Fig, 4A), Thus-the dimension of the channel would actually limit the size of the bundles formed. Long fllamsms can only move freely within their own channels, whereas shorter filaments can move io and Out without any restrictions, Newly foemed long filaments will be preferentially confined within one of these channels because of their local rigidity and high aspect

ratio, as well as the long-range electrostatic repulsions among tbe filaments (Fig. 4B). We assume that the growth of spatially correlated filaments in the limited space provides the foundation for the observed filament stacking, At I wt % or lower, subsequent filament growth within the. spaceconfined channels eventua lly results in formation of loosely packed filament bundles that lack positional order (Fig, 4C). The lncrea~iflg Dumber of til aments within tIl e b undles as concen tration is increased can be readily observed by cryoTE:!vI: (Fig. 4, E to G). We suggest that, beyond a critical conceotrauon, electrostatic repulsions within bundles combined with our hypothesized network forces lead to hexagonal ordering of tilamynts (Fig, 4D)_ The interactions, between these Like-charge filaments depend not only on the interfilament distance and their mutual orientation but also on the 3D structure of the electric double layers arising fromcondensed.and mobile ions, The theoretical analysis ofelectrostatic interactionswithin the bundles would require solving Poisson's equation with boundary conditions for a system of parallel filaments.

ii'.(EE(r) + P(F)) = p(t)

where £: is the dielectric pennimvity tensor, [(1') is the electric field that determines the strength of el ectrostatic interactions, P (f) deseri bes -the polarizafien of dielectric mterfaces, and p{y) is the free charge .densiry. For systems with high Sl.U"-

REPORTS

I

moe charge in the presence of d.ielectric interfaces .• charge correlations and interface polarization lead to intractable tnatbernatical problems (26), clearly beyond the scope of'this work,

The crystallization at long range driven by 200-8 exposure to x-rays supports strongly the presence ofbundles evert in very dilute networks. The gradual rise in intensity of the principal Bmgg peak. suggests th at more crystal]; ne bU11 dles are being formed upon COntinuous x-ray exposure. Also, 00 the basis of the Scherrer equation, the observed decrease with further exposure in full width at half maximum of this peak suggests that crystalline bundles. are increasing in size. During this period of time, we do not expect reorganization ofthe extremely long filaments across a huge volume, and therefore crystallization must be occurring locally wilhOJthe bundles, At the same time, the observed reversibility between disordered and hexagonally ordered states at low conceutratious suggests tbat mterfilamenr forces must be changing as a result of x-ray exposure. X-ray irradiation is likely to increase the charge density on f lamen t surfs ces ( /8, 19" 2 7), thus altering th e balance ot'forces Within the bundles oftbe dilute network,

This new balance of forces induced by x-rays must resemble that prevailing in the sponraneous crystallization ofbundles at high ccoceotrariou. If electrostatir; repulsive forces among the filaments are mediating crystal lizati on at long range, they must be acting in concert with a balancing at-

Fig. 2. Concentration de- A -

pendente and temperature effect of x-ray-triggered f la rnent rrystalli za ti on. (A) Set of SO consecutive sea tteri n9 prof les from

1 wt %.aqueous solutions ;g:

\eJ!'posure time: 4 5). (8)

Set of 30 consecutive scat-

tering profiles from 2 wt % 111

solutions (exp osure t\ me: •

2 s), and 5 wt % solutions ~

(exposure time: 1 s). (() '--~_'-~~~--~"'"

(The dose rate for the experiments in (Al to (C) was 0 4806 Gy 5-1.J The first

a nd the last p rofile~it] (A)

to (C) ar:e colored red .and green, r:espectively. (0) A replot of x-ray sea ns com:sponding to the last expo- 5Uf~ at co ncentrations of 0.5 wt%, lwt%, 2 wt%, and 5 wt %. (E) A plot of (enter-to-center spadngs among filaments asa function of peptide amphiphile {Once nrra tipn (th e inset shows a schematic represe ntation of th e. La rg.est

spacing observed among filaments in 0.5 wt % solutions), (F) Eff·ect of temperature on Hay-triggered ordering of the filaments in a 1 wt % solution (the relative intensity is given by (I f1 - 14)J!~, where n denotes the

o

2S0C - 40°C

'0 5 20

Exposure TIme (sec)

aceum ulatedexposure ti me and f4 is the intensity value at the location of the pri ndpal Bragg pea kin the first scattering profile obtained after x-ray irradiatiQn) .

Www:sciencemag.org SCI EN CE va L 327 29 JAN UARY 2010

557

REPORTS

tractive force. We hy.pothesize that this attractive. force could he rooted in the structural imegrity of tile entire network, The spatial confinement imposed byueighbcring bundles in the network acts effectively as an "attractive" force balancing repulsion among the like-charge filaments. The stability of these networks containing crystalline bundles of like-charge filaments may originate in tension linked to repulsive forces within each bundle balanced by compression forces associated with the network. In this context, it is interesting that crystallization within the bundles provides a mechanism to minimize the local tension from repulsive forces, This balance is reminiscent of that existing in self-stabilizing tensegrity networks (8, 9),

We carried out experiments to demonstrate the involvement of repulsive electrostatic forces incrystallization at long range within the filament bundles. These experiments involved adding NaCl and eaCb to I wt % solutions in order to screen electrostatic. forces among like-charge filaments. As [he amount of NaCl was increased, the primm)' Bragg peak developed more slowly during continuous x-ray irradiation (Fig. 41 and fig. S8) (20). At a concentration of 50 111M NaG or higher, the x-ray-induced crysrallizatiou was not observed at all even after an accumulated exposure time of 120 s, l<ignte 4J shows 50 sequential scattering profiles of J WI % solution ]11 the presence of 50 mM NaC!. When the divalent salt CaCb WaS used even in concen-

trations as low as 5111M. the crystalline structures were never observed. Addition of 5 rnM NaOt-L can promote filament ordering at 1 wt % without exposure of samples to x,-ray irradiation (fig. S9) (20). NaOH can increase charge density on the acidic nanotiber surfaces by deprotonating COOH groups in glutamic residues_ These observations support our hypothesis that eleetrostatio repulsions mediate crystallization of the filaments at long range, Ow' experiments also suggest that the observed reversible x-ray triggered crysta Uization results from increasing charge density on filament surfaces.

To investigate funherwherher this charge density increase is associated with permanent chemical modifications by the x-ray photons, we carried out chemical analysis on a sample uradiatedwnh a large dose of x-rays. Even when about 601% of the total volume of a sample. was irradiated with a dose 44 times greater than is needed "to generste filament ordering, chemical changes were still not observed by using an ultraviolet-visible light detector, Only mass spectrometry detection could reveal a very small amount of chemically modified by-products (SOM text S I 0 and figs. S 10 to 536) 12Q). in acidic, air-saturated aqueous solutions, the dominant species of water radiolysis are hydroxyl radicals (·OJ!) and hydrated electrons (e~); both could potentially react with peptide backbones and side chains lJ.t various ways and yield products such as ammonia, keto acids, aldehydes, and other peptide derivatives

Fig. 3_ Reversible crystallization of filaments triggered by x-ray irradiation and the effects of the x-ray beam attenuation. (A to 0) A series of seq\lentitil photographs of the quartzrapiUary loaded with sample reveals the optical appea ranee changes before and after x-ray irradiation. The parallel red lines in (A) demarcate the edges of the capfllary, and the white

arrows point to the spot irradiated with Nays for 200 s q(A -1) 0.1

(the cylindrical sba pe vtsible within the space of the

capillary is an artifact 01 the photograph). Photograph (Al corresponds to a sample-loaded capillary after 200 5 of x-ray exposure (indicated as time 00:00 in the lower left-hand corner of the photograph). Sub~equent photographs [(a) to (D)J reveal changes in the appearance of the spot after the elapsed times indicated in the left-hand corner of the photographs (minutes:seconds). The opacity in the irradiated spots dimmed eventually in the absence of the x-ray beam. {El Scattering profiles at the position marked with a white arrow in (Al after the first x-ray exposure of 4 s (black curve) and after the last of 49 additional expos u re s, (red cu.rve). Afte r a waiting p e nod of 2 ho U 1':5 with the b eil rn off, (he e xp erimen t wa s repea te d of') the same spot, and similar S(iltt~rin~ profilesw~re obtai~ed (ween and blue (urves). All exposures lasted. 4 s and the dose rate was 4806 Gy s- . A.V., arbll1:ary units. (F) A plot of accumulated x-ray exposure time required for observation of the principal Bragg peak as a function of relative beam intensity. (6) A plot of th:e reciprocal accumulated x-ray dose (SOM text 52 and 5"6 and figs. S5and 5"6) r;o) as a function of relative beam intensity.

ssg,

(19). Among these products .. a-hydroxyl carboxylic acids and I;l·keto acids could potentially contribute to an increase in charge density because of their slightly lower pKa (where K~ is the acid dissociation constant) values. Interestingly, de. tailed chemical analysis using mass spectrometry indicates that hydroxylation products are only associated with the alkyl tails-and the first alanine residue adjacent to the tail Our data also show that the very small amount of a-hydroxylation occurring in the glutamic acid residues results in main-chain eleavage and formation of amideterminated products (80M text S 10 and figs. S to to S36) (20). We conclude that none of these species contributes substantially, jf at all, to the increase III charge density 00 the surfaces of filaments.

Because oxygen plays a very Important role in mediating this irreversible chemical reaction to generate tbeu-bydroxylation products, we perfunned experiments in an oxygen-free solution and found that x-ray irradiatio» still induced ordering of filaments (ng_ 837) (10). This experiment provides .additional evidence that permanent chemical damage mediated by soluble oxygen in water is not necessary to facilitate filament ordering, In addition, we have observed filament ordering in ether peptide amphiphile molecules with differen; sequences, for example, VVAABEOGREDK:ETV, VVJ\AEEGGTKREEVD, aad AAEEGGREDKETV (figs. 538 to 840) (2n, 28), further suggestin.II thar the mechanism of charge den,~i1Y increase

F '0200

Q)

..!.

IP 150 E

i=

! 100

_:,

8. 50 .n

1.0 0.8 0.60.4 0.2 0.0 Relative Beam Jntellslty

29 JANUARY 2010 VOl327 SCIENCE www.sciencemag.org

on filament surfaces is not specifically linked to glutamic acid residues but is instead a more general phenomenon.

On the basis of evidence described above We propose that the main mechanism responsible for an increase in charge density on peptide nanofiber surfaces is linked to the reversible ionization of the COOH groups by x-ray irradiation, Because we obtained direct experimental evidence for filament ordering as a result of carboxyl ionization with strong base, We speculate that tbe enhanced electrostatic repulsions among filaments leading to crystallization arise from deprotonation ofCOOH directly caused by x-ray i.rradiation (SOM text S 11) (20). Hydroxide ions produced by water radielysis may also contribute to the

enhanced ionization ofcarboxyl groups. Because of the bigb packing density of eOOH on the surfaces of these extremely long filaments, a "VelY small percentage of ionized COOH groups could potentially lead to a considerable increase in electrostatic repulsion among filaments and thus produce the observed ordering. Hydrated electrons produced by water radio lysis could also contribute to electrostatic repulsion through their addition to the C=O bonds V 9). The charged species can return to theiroriginal structures through a previously proposed reconstitution mechanism (SOM text S II) (J 9, 20).

We have reported crystallization mediated by repulsive forces of high aspect ratio like-charge filaments within a network environment. We be-

Fi.g. 4. (A to D) Schematic illustration of the proposed templating model for filament bundle formation during self-assembly, (A) At ea fiy stage5 of self-assembly, interlotking of long filaments leads to the formation of a channel around each one of them. The tong filament has miJ!ny degrees of freedom within its channel but cannot leave it or allow others to enter its space. The volume of this imaginary channel should be determined mostly by the bulk density of long filaments. (B) Newly formed long filaments from ~maller ones and from monomers are confined to grow within these channels. (e) Further growth causes. the formation of bundles of aligned long filaments. (D) At high concentrations., hexagonally ordered filaments are sponte neQusly formed _as a result of repulsive electrostatic forces a nd network confinement. Cryo-TEM image of filament bundles 'in 0.5 wt % solutions (E), 1 wt % solutions (F), and 2 wt % solutions (6). (H) Negatively stained TEM image of very earlystaqes of self-assembly revealing the presence of small aggregates (marked with black arrow) coexisting With loosely packed long filaments at 0.1 wt % [this image offers support for the ternplatinq merhanism for bundle formation explained in (A) to (D)]. (I) The. effect of added salts on x·ray-triggered crystallization of filaments: plot of the intensity of the primary 8ragg peak as a function of exposure time to x·rays in 1 wt % solutions; when the intensity plotted 1.S below the dotted line, the Bragg peak was not observed. (J) Set of 50 x"fay scattering profiles corresponding to a 1 wt % solution containing 50 mM concentration of NaCl obtained sequentially after 50 exposures of 4 s each to the x-ray beam (dose rate: 4806 Gy.,-l),

REPORTS

li eve that arrangement of the filaments into aligned bundles trapped within a network is necessary for the observed erystallizarion and can be templated during growth of the very long filaments by selfassembly. The. supramolecular nature and role of electrostatics in' artificial systems studied here create a connection to cytoskeleton filaments, where similar crystalline bundles embedded in networks could occur naturally or be created by externally regulated charge density.

Reference, and Not!,1

1. W. H. Song, I. A, Kin toe h, A. H. Win die, Sc ience 3 OZ. 1363 120031.

2. S.-W. Lee, C. Mao. C. E .. Flynn, A_ M. aelcher. Scienee zss, 8~2 (2002)

3. M. Nakata et at., science 318. 1276 (wo7).

4-. D,). Ne~dleman et al., Pr(K sa« !lead. SCi. U.s.A 101,

16099 (2004).

5. 0, Pe\letier et al .• Phys. Rev~ tett, 91, l48102 (200~). 6, M. L GardeL ef at., Science 304. UOl (2004).

7 M. B~the, C. H~lI',ing~r, M. M. ClaeI5en!;, A. R. Baul(h, E. Frey. 8iGphys.). 94, 295S (2008).

B. D. E, I~gb"r, An"'} Rev Phy,iol. 59, S75 (1997)

9. N. W~ng·.l. P. Butler,.D. E. Ingbet,Sael1(e 1.61J; U24 (1993).

10. B_ Chu, B_ S. Hsiao, Chem Rev. 101, 1127 (2001).

11. J. tlpfert, 5. Doniach, AnIlU. Rev. fJiaphys. BiomoL strua. 3'6, 307 (Z007).

12. V. (herelOv. K. M. Ri€dL, M. Caffrey, J. Synchrotron

Radial. 9, 313 lZ0011.

13. R. B. G. Ravelli. 5. M. M(Sweeney, StrucMt 8. 315 (2000).

14. T. Y. T~ng. K. Moffet J, Synchrotron Radjal. 7. 3B (2000)_

15. C. NaV€, Radiat: I'h)'5. Chern. 45, 4i!3 (1995.).

16. A_ (_ Cheng, M. C"ffrey, Biophys. J 70, 2212' (l996).

17. M. Weik e/ 01., rroc Nail. scad. Sci. U.S.A 97, 823 (2000).

18. G. Stark. Bio(him. Biophys.. /1(10 1071, 103 (1991).

19. w. M. Gaflison, (hem. R.ev. 87. 381 (1937).

20. M~tEri~ls, methods, and additional fiqures are ~volilable ~) wppgrting material an Science Ohlin",.

21. 1- D. Hartge(ink. E. B~niash. s. L Stupp, S,i~lice 294, 1684 (2001).

22. Y. S. Veli,hko, 5. I. StllpP, M. O. de la Cru1.,). PhYJ.

(hem, B 112, 2326 (200B).

23. c. l. Pizzey et al .• J (hem. I'hY5, lZ9. 095103 (200a).

24. V. A. BloomfieLd. Biopolymer:; 44, 269 11997l.

15. ]. Pelto ) r., D. Du ra n d, J. Doucet, F. Livola n I, BiophYJ. J. 71, 4B (1996)

26. V. A. Parlegian, Von det WlIal.l (Cambridge univ, Prell, Carnbridqe, 2005).

27. E. Collet et al.. Science 300, 612 (2003).

28. S i I'Ig le- L"tt~r abh re viations fo r the ami no ,Kid re;idue5 ore as follows: A, Ala; D. Asp; E, GLu; G. GLy; I\. ly,;

R, A19; T, Thr; and v, V~l

29. This work was I upport ed by tll e u..S. 0 epa rlm en t of E~~f9Y !gl a nt ne. OHG 02- OOER4 5 a 1 0)_ The SAl! S experiment. were performed at the OND·CAT lorated at Sector S of the APS. DND-CAT is supported by

E. I. DuPont de NemDurs and Company, Dow Chemical (om pa ny. a nd !Me sta te ol I Hi nois U,e of the Il,PS wa 5 supported by the u·,s .. Department of Energy. Office of S(i~n ce, Offi ce of Bali c E ner!IY Stien(l~'. un der con tra ct no. DE-AC02-06CHIB57 _ Some additionaL supportfer this wqrk w~ obtained from NIH National 1~5titutf of DenIal and Craniofadal Rese,mh grant 5ROl Df01592il and N5Fgr.aot DMR·060S~27, We thank the 6iologi(;11 Imaging Facility (BIF) at Nortnw!Csl!lm ior the use oi TEM equipmen!.

Supporting Online Material

www.,([el1cemag ,orglcg ileo ntent/fu IV" ience .11 S"Z3 401 DC 1 M~ If'riall and Meth 0 ds

SOM Text

Fig,. 5l to 51,4 Ref'~ren(€'

23 September 2 009 ;a((~p ted 3 0 N ovemb er 2009 Pu bLi III ed onl i O@ 17 D~(emb er 2009; 10_1U<\lIdeo<;e.ll82340

Include this information when dung tbj~ paper.

WWW.sCleneemag.org SCI EN CE va L 327 29 JAN UA RY 2010

559

REPORTS

The Free-E'nergy Landscape of Clusters of Attractive Hard Spheres

Guangnan Meng,1 Natalie Arkus,z Mkhael P. Brenner,z Vinothan N. Manoharanl,Z~

The study of clusters has provided a tangible. link between local geometry and bulk condensed matter, but experiments have not yet systematically explored the thermodynamics of the smallest dusters. Here we present experimental measurements of the structures and free energies of cclloida I clusters in which the particles a ct as ha rd spheres with short-range attractions. We found that high ly sym metric clusters ate strongly suppressed by rotationa I entropy, whereil> the most stable dusters have anharmonic vibrational modes or extra bonds. Many of these clusters are subsets of close-packed lattices. As the number of particles increases from 6 to 10, we 'observe the emergence of a complex free"energy landscape with a small number of grount! states and many local minima.

All isolated system of 10 interacting atoms ormolecules willgenerally adopt a struoturc that differs in symmetry and average energy from tbat of a bulk liquid, solid" Or even a. systetn contain lng ! 00 particles. Yet tbe srudy of such small clusters bas shed light on a wide variety of phenomena that are observed in the ne'lds of'condensed-matrer physics and physical chemistry. Since- Frank first predicted (I) that icosahedral short-range order would be a hallmark of liquid structure, th e study of small- cluster geomctry 11M provided key jhsight$ into. U1C frustration underlying nonequilibrium phenomena such as nucleation and the glass transition (2-4)~ Experimental studies (5,. 6) have confirmed this approach tbrough the discovery of local cluster-like order in bulk liquids and glasses, with recent resuits (7) suggesting th;ll structural arrest in condensed phases may be related to geometrical con straints at the- scale- of a few parti cI es.

There remain many unresolved questions about cluster geometry and itsconnection to bulk behavior, Altboug,h experiments and simulations have determined the rnmurmm potential-energy clusters for various interactions (8 10), the likelihood of observing a particular cluster structure depends on its free energy U1, 12). \Vbat cluster structuresare favored by entropy? How does ihe competition between potential energy and entropy evolve as the number of particles N approaches the bulk limit? Experiments on atomic clusters b a va DOt systematica Uy ex plored tb ese questions: they are limited by short duster lifetimes, nonequilibrium conditions, and the difficulties of 01'1· raining real-space structures of'individual clusters in free space (13).

Here we report experimental results for the structures and me energies of small equilibrium clusters as a function of N, with N'5 10. TIle experimental system is described in Fig. 1" We use colloidal particles rather than atoms, because

lO~pilrtm~l1! cf Phjl'iicl. Hafll~rd Univer,ity. Glmbndg~, 'MA 0213 8, USA, I H~rvard School 0.1 En gi neeri ng an d Appli ed 5riel1(~~, HElrv~rd Univer,jty, Cambridge, MA ,0213e, USA.

'To. whom correspondence should be addressed. E·mail:' vn m @~ea, .ha rva rd. ed u

560

we can precisely control the interactions arid directly observe the three-dimensional (3D) snuctures of the clusters using eptical microscopy, To a good approximation, our particlesact as "sticky" hard spheres, .arguably the Simplest nontrivial interaction that leads to clustering. The attraction arises from a depletion interaction with a range of about 1.05 times rhe partiele diameter and a depth of about 41q, T, where ks is Boltzmann's constant and Tis temperature. Because the pair potential is short-ranged, the total. potential energy U of a given structure is well approximated by U = CUm, where C is the number of contacts or depletion hands and Um is the depth of the pair potential (14). Although these particles form a gel

in bulk. the range and depth of'the interaction are consistent with an equilibrium phase diagram showing a fluid-to-crystal transition U 5).

We, created clusters by isolating small numbers of polystyrene lPS) microspheres in cylindrical microwells filled with water and poly(Nisopropylacrylarmde) (polyNIPAM) nanoparticles, which cause the depletion interactiQn. Wechemically funcuonalized the micrewells so that particles could not stick to the surfaces, This allows :m dusters 10 fbnn in the middle of the wells. unaffected by the boundaries. After the clusters reached equilibrium, we used optical microscopy to observe the cluster structures, and we collected statistics by scanning througb the microwell plate, which contains thousands of isolated clusters, Although tbe number of particles per wen is not controlled, we generated enough clusters at each value of N cS 10 to measure their occurrencefrequencies. We then determined the free energies from the ensemble statistics through the Boltzmann distnbution, M' = -k1l11nP. where P is the probabiliry of observing a given cluster,

We classified ow' clusters by comparing them to finire sphere paekings. A previous theoretical study (16) enumerated the mechanically stable clusters nf idealized bard spheres with infiniresimally short-ranged interactions, revealing the minima of the potential-ecergy landscape as a function of N_ All oftbe minima at each value of N S 9 have the same potentia I energy, which is a result not observed with longer-ranged potentials (17). We explored the structures and proba-

E

fig. 1. (A) Diagram of experimental systern (14). We used lithography to. make rnicrowells with depth and diameter of 30 urn (see '1,150 fig. 51). Tnll5e are filled with a suspension of 1.0e)J.m-diaffieter PS spheres and aO-om polyN1PAM mkrogel particles, which induce a depletion attraction <l~ illustmted in (B). The number of PS particles per wllll varies, but the 1!vel'<lge is about 10. {() Pair potential as estimated from the Vrij approximation to. the Asakura-Oosawa potential (28, 29). Because the range of the depletion attraction is less than 1110 of the PS sphere diameter, the interaction is strictly pairwise additive. (D) Optical micrograph of microwetls with assembled colloidal dusters suspended inside. The drdes highlig ht individual dusters in differeot mkrowells. There are about 104 microwells per 5lide. Scale bar, 20 urn. (E) High-magnification optical microgr.aphs of conoidal dusters in rnicrowells with N = 2, 3. 4, and 5 parti des. These are the only structures that form for N :-; 5. Scale bar. 1 urn,

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

bilities of these packmgs at finite temperature, wroth allows us to map the free-energy landscape (18, 19, Il). All ef the observed duster structures agree with the theoretical predictions. For example, for N < 6, we observed one unique structure for each N: a dimer for N = 2, trimer tor N = 3, tetrahedron for N == 4, and triangular dipyramid for N = 5. The optlcal11J1ierogrllphs in Fig, 1 show the structures of the smallest clusters.

The 'first interesting case is N = 6. We observed two structures (Fig. '2 arid fig. S2), botb with C = 12 contacts and therefore equivalent potential energy. The first is the octahedron, a Platonic solid. The second, we call B "polytetrahedron.' It consists ofa triangular dipyramid wiih a third tetrahedron added to one of the faces. We observed transitions between the two states on time scales ofminmes.lndicating that the system is at equilibn um (mov ie 5 I).

Although these two structures have the same potential energies, the poly tetrahedron ocelli'S about 20 times more often than the octahedron, implying a free-energy difference of about 3kBT. This difference can beartrfbuted only to entropy. As shown ill Fig. 2, the measured probabilities for the two structures agree well with theoretical calculations based on standard approximations for the rotational and vibrational entropies ill the classical limit (14).

Tile rotartonal entropy makes the largest contribution to the free-energy difference between the two structures (fig, 52). The rotational partition funcri on is rela ted to two geometrical q uan titi.e~: the number of orientations, which is. proportional to the moment of inertia, and the rotational symmetry of fue cl uster, or, altemati vely, tb e num ber of ways om} can assemble the same cluster by rermutil1g particle I abels (3D). Formally, the ratio of the permutational degeneracies of two clusters is inversely proportional to the ratio of their symmetry numbers (2J). This permutational degeneracy accounts fora factor of 12 in the polytetrahedrorroctabedron probability ratio .. The remaining factor of 2 comes from the differences in the moments of inertia. and the vibrational entropies.

This result ill ustrstes a genera I rul e for clusters 'with short-range attractions: ~m01Jg clusters with the same potential energy, highly symmetric structures are extremely unfavorable at equilibrium .. By contrast, fur the longer-ranged LennardJones 6-12 potential, the ocrahedron bas lower potential energy than the polytetrahcdron does (17), so that the dominant structure depends on temperamre. The dominance of the polytetrahedroll in out system may have consequences for nucleation; the equilibrium phase of attractive hard spheres is a face-centered cubic (FCC) L11'8< tal (15), Which contains octahedral, Dot polytetrahcdral, subunits,

At N = 7, the first chiral structures mise. We observed .six duster structures, two of which are chiral enannomers. The experimental measurements agree well with the theoretical values for the probabilities of , each structure (Fig. 2). For

REPORTS

these small clusters, the most pronounced infWence on the probabilities comes. from symmetry, At N = 8, 3 of the 16 different possible sphere packings never Occur in the experiments. These three snuetures have the highest symmetry numbers, IJ = 4, 6, and 12_

A few structures differ by such small changes in particle spacing th1'!t we cannot differentiate between them U'~jng our microscope, All of these are variants of pentagonal dipyramids, In a pentagonal dipyramid of seven spheres, the top ana bottom spheres of the pyramid are separated by a small gap of 0:;; O.05d,-where d is the sphere. diameter. If these two spheres are brought together, a gap cf = O.09d opens between two of the

spheres on the pentagon. Because we cannot resolve this. gap in our experiment'>, we have binned fhese structures together at both N = 7 and N = 8. Tho one statisticallysignificant discrepancy between experiment and theory OCCUl'S at N = 8; it arises because the experimental potential has a range that is comparable to the g?P distance, Although we account fur this extra potential energy In the probability calculations, the probabU~ties are sensitive: to the magnitude of the potential at the gap distance, If the interaccion energy differs from our estimated value by only O.lk1)Tin the gap, the theoretical calculation fa U s within eITOr of the experirnental value. Thls difference could be due to polydispersity in either the depletsnt

N=6 poly- N=7
100 tetr:,h ad ron 35
30
80
;? 25
.'.'!...-
~ 60 20
:b
<11
.D 15
E 40
!L
10
20
5
0 0
Czv· Oh ~~ 05r, c,,_. C2 C2 C3V N",,'}

80

70

eo

~

.'.'!...- 50 >-

=:

~ 40

.D

e

!L 30

20

10

Fig, .2. Comparison of experimental and theoretTfill (14) duster probabilities P at N = 6, 7., and 8. Structures that are difficult to differentiate experimeo@lly have been binned together af N = 7 and 8 to (Qmpate to theory, The t<Jkulated probabilities for the individual states are shown a'l light gray bars, and binned probabilities are dark gray. Orange dots indicate the experimental measurements,with 95% confidence intervals given by the error bars (14) (table 51). Renderings and point groups in Schonfiies hQtation are shown for €i!ch structure. The number in the subscript of each symbol indicates the order of the highest. rotational symmetry 'axis, and the letter indicates the symmetry group. The highest symmetry structures' ilre those in D, T, and 0 groups, Structures in (I and C, groups occur in chiral pairs.

www:sClencemag.org SCIENCE VOL 327 29 JANUARY 2010

561

REPORTS

or large spheres or to another interaction such as van der Waals forces,

The probability distributions in Fig. 2 are "field guides" to the free-energy landscapes at N = 6, 7, and 8. Each structure represents a local free-energy minimum. the depth of which is proportional to the probability. We note two topographical features besides the trend toward structures with low symmetry: first, the number of local minima increases dramatically with N;. and second, the landscape is relatively flat for N = 7 and 8. In other words, there are many shallow minima, but no one minimum has a lice energy muohlarger than any other_

The landscape undergoes a qualitative change for N ~ 9. Theoretically we expect some n = 77 structures at N = 9 and n = 393 at N = 10, too many to caralog experimentally, We tberefore measured ouly a subset of structures identified by our theoretical study (16). The subset we choose CO[l4 sists of clusters tbar fall into either of two categories: nonrigid structures, in which one of the vibrational modes is a large-amplitude, anharmonic shear mode; and structures with more than 3N ~ 6 bonds Nonrigidity arises when a duster contains half-octahedra that share at least one vertex, allowing the cluster to twist over a finite distance without breaking or forming another bond. We expect these packings to have high vibrational <entropy. Structures with more tban 3N - 6 bonds can occur for N::: 10. These are the expected ground states.

562

Indeed, these special packings do occur "villi high frequency (Fig. 3alJd table 82). Because most clusters at N = 9 and 10 have equal potential eoergy, low symmetry, and tberefbre comparable rotational entropy, we expect the average probability of anyone structure in a set orQ possible clusters to be of order lin. At N = 9, we expect an average' probability of about J %. and at N = 10, about 0.25%,

By contrast, the one nonrigid structure at N= 9 occurs with p;::;_ 10%. By 'Using the theoretical ,Q and the experimental P, we estimate that the free energy of tbe nonrigid snucture is about '2kBT lower than that of an avera,ge structure at N = 9 (14). Thus-the structure is highly stable by nearly balfthe free energy of an extra bond. The stabilization comes from the vibrational entropy associated with the nonrigid mode (movie 83). Our theoretical calculations (14) predict P ::::c 3%, whieb is lower than the observed probability but higher than ali other clusters at N = 9 .. The discrepancy is due to the sensitive dependence of thevibrational partition function on the curvature of the pair potential near the minimum, a consequence of the nonrigid mode. A more precise calculation requires. an accurate measurement of electrostatic effects in the experimental pair potential ncar the depletion well.

At N ,.., 10.; only 3 of the 393 theoretically possible clusters have 3N - 5 ,.., 25 contacts, ~et these occur about 10% of the time. Although we have only limited statistics forhlgiler N. we con-

Fig. 3;. (A) OpticaL micrographs and renderir)gs of nonrigid structures at N = 9 (see also movies 52 and 53) and (8) N = 10 (movies SA to 56). (e) 5ll'tlctufe~s of 3N ~ 5 = 25 bond packinqs <1t N = 10 (movies 57 to 59). The anharmonic vibrational modes of the nonrigid structures are shown by red arrows. Experimentally rneasu red probabiliti es _are listed at.to p. Annotations in micrographs indicate clusters correspcndinq to subscts of HCP or FCC lattices. Scale bars, 1 urn.

tinue to observe the prevalence ofa few packmgs with 3N - 5 or more bonds. The structures with extra bonds have combined probabilities of 20 to 30% at N = 11 and 12 (table S2). Again these probabilities are large compared 10 !In, even though, in several cases, the clusters have high symmetry. The potential-energy gain is therefore large enough to overcome the deficiency in rotational entropy.

Perhaps the most striking feature of these clusters 1& that many aft Subsets of lattice packings, and in particular of the hexagonally close packed (RCP) lattice. The lattice packings are marked in Fig_ 3. The underlying reason appears to be that both nonrigidity and extra bonds require the clusters to have octahedral subunits, The propensity tor icosahedra (8, 22) in longer-range systems is absent in ours, We observed TIO icosahcdra at either N = 12 or 13, presumably because neither 12-sph6re or I3~sphere icosahedra are special clusters fOJ short-range interactions; neither are nonrigid, neither have more than 3N - 6 bonds, and both have very high symmetry numbers (0 = 60).

By using the same statistical mechanical approxirnationsthat were used to estimate probabilities for N~ S, we can calculate the free energies of ali mechanically stable sphere packings that have been enumerated (16) up 10 N = 10. This yieldsthefree-energy landscape shown in [;"ig,4, We see that, in general, the locus of states is correlated with the rotational entropy. which is proportional to k13]r(v'i/cr), where J is the moment of inert ia, The 0 nly slates that lie below tills locusoccur aIN = 9 and 10. These correspond to either nonrigid structures or structures with extra bonds, both of'wbich appear as deeper minima,

The diagram also reveals some new features.

First, low-symmetry polytenahedral states proliferate as N increases. At N = 1 O. where clusters with extra bonds first appear, the absolute probability of observing these ground slates is low because of the large [lumber of low-symmetry sta tes that heat slightly high er free energy,

Second, the highest free-energy structures for N = 6 to 10 are convex deltahedra (13), polyhedra with a 100lg history in the field of condensedmaner physics (24, 3). These are not always the most symmetric structures; ar N = 8 the highest free-energy state is the deltabedron, a snub disphenoid that. has lower symmetry than an eightsphere tetrahedral cluster The convex. deltahedra also happen to be the same "minimal-moment" structures that are funned in capillary-driven assembly of colloidal particles (25). The optimal paekmgs under these nonequihbrium conditions fherefore correspond to the least-optimal packings at equilibrium,

Our results suggest that nucleation barriers and StruLtUJ1.1i motifs in attractive hard-sphere systems such as colloidal suspensions will be different from those in systems witb longer -range potentials, which tend to favor symmetric structures at sufficiently low temperatures, For N <: 9 all of ow- clusters havenearly equivalent potential

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

-I

-r

-01

~111

j~l1.rjgjtll

I! I.. ~

-1

• of

.' II

-5

-r

tu

(;

12 1J

Fl mann ria I Ilan ~11111 illilr.lIQf1 = ,l/tl

Fig. 4. Calculated minima of the free,energy landscape for 6 5 N S; 10 (,14). The x axis is in units of the rotational partition function, where I is. the moment of ine rtia (t<r\culated for a pa rtide mass equal to 1) and c is' the rotational symmetry number. The bond stre ngth for the cakulatio n is Urn = 4k$T. Each black symbol represents the, free en €rgy of an individual duster. The numbe r of spo kes in each symbol indicates the symmetry number (got = 1, line ~egment = 2, and 50 onl. Orange symbols are nonrigid structures, which first appear at N = 9,and violet symbols have extra bonds, first appearing at N = 10. VerticaL gray lines indicate the contribution to the free energy due to. rotational and vibrational entropy. The reference states me chosen to be the highest free-energy states at each value of N. The general trend islor Low-symmetry states to be favored in proportion to their rotational entropy, k.s[nh!llcs). The potential-enerqy contribution accounts tor the vertical space between the violet symbols and their grily lines.

energy, and therefore the rotational entropy so> I eets again st S ymmetri c structures at all temperarures, Specifically, the symmetry number and the permutational degeneracy have the greatest effect on the free energy; differences in moment of inertia do not contribute as much. 111U$, even if cluster rotations are- hindered, as they may be in a bul]; supercooled liquid, the perrnutetional degeneracy might still influence the probability of formation, At higher n. the most probable structures we observe involve cemainations of octahedra and ietrahedra, M~IlY of these structures ate compatible with an He? lattice, but not with FCC. Our results also suggest that the curvature of the pair potential near the min. imum should affect nucleation, because the CIJIvature detennines the free energy of the nonrigid clusters.

Structures with fivefold symmetry, such as the pentagonal dipyramid and icosahedron, are 'high ly unravorable in our system. Therefore we clo not expect icosahedra or other clusters with

fivefold symmetry to be a structural motif U1 !I'j:.tractive bard-sphere gels or fluid cluster phases (16); where the attraction IS short-ranged.

We find that the most stable small clusters of hard spheres with short-ranged attractions ClIP be dererrrdned by geometrical rules: (1) rotational eotropy favors structures with fewer symmetry clements; (2),vibrational entropy favors nonrigid clusters, which have balf-octahedral substructures sharing at least one vertex: and (3) potential energy favors clusters with both octahedral and letrshedral substructures, allowing them ID have extra bonds,

Out picture of the- free-energy landscape is still incomplem, The qualitative features of the landscape ate independent of'remperarure for ow' experimemal system because thedepletion interaction is fundamentally entropic (14). This will not be (he case for other types of interactions, such as DNA-mediated attractions. (27), Also, under nonequilibrium conditions; we expect a different distribution of'snuotures than the ones

REPORTS

shown here. Finally, our model for the landscape does not account for energy barriers or thee-Oil' neetivity between minima, It will be interesting to see if further studies can explain the emergeoce o f bulk crystallizati On or structural arrest in terms of'these topographical features and their geometrical underpinnings.

References and Notes

L F. t Frank, pro" R. So,. Lqnd, A Mati) Pilys. Sd 21,5, 4J (1%2).

2. F, H, StiUing"r, I, A. Webel', Science 225,983 (19S4).

3. [1, ~elsan, F. Sp3~pe~, Solid State Phys. 42." 1

(19S9).

4, l. P. K, Doye, O. ]: Wa(es., snence 271, 484 (lTJ5i. 5, H. Reichert et 11/" Nafure 408,. 839 (2000).

6. H. W. Sh~~g, w. K_ LUQ, F. M. lllilmgir, J. M. BOli, E. Ma.

Nature 439, 419 \2006/,

7, c. PaJrkk !loyall, S. R. Wall.ml, T. Qht!uka. H. Tanaka, Nat. Ma~r. 7. S~~ (2D~81

8. M. Hoare, P. Pal, fldv. Phys. 24, 645 (l975).

'9. M. Hoare, J. McI~l1e!, Ild~; Phy" ~2, 791 (:1;983).

10. D. J. Willes at M, Cambridge Cluster Database, WWWiPJilleufH,Qi'Lil(.uI1CCD,htnil (2ooS).

11, D. J. Wol€l, T. V_ Gogdatl, J. Phys. (/iem, B110, 20165 (2006).

l~, j, p, K. Doye, F. C.lvo, J. [hem. Phys. 116, 8301 (2002).

13. R_ 1.1ohMtoo, Atomic and MolewlQr Clusters (Taylor Ii" franq., New York,. 2.002).

14, Materials and metncds are available a 1 lupporting materia i on Science Ohlin" ..

15. V_ J. Andel',on, H. N. L"kker~~rker; Nature 416, 811 (2002).

16, 1'1, Arkul., V. N. Manohoton. M. P_ Brel1ner. Phys. Rtv. Lett 103, 118303 (2009),

17. M. Hoare, P. Pal, Adv. Phys.. 2.0, 161 ·(1971).

18. C L. Brooks 3rd, j, 1'1, ()nuc~ic, D. l. Wales, Science 293, 612 (200l).

19. K. A. Dill, H, S. Chan, Na!, Struct, SioL 4, 10 (19<)7).

2.0, 0 ur pi! rt ides "re rnacfo,wpf c an d th e re fore dis~ 09~ish;3 ble in prindple. But we do n 01 di5(tng ul,h thl<! parttdes in "ilher tI1e oct~h"(lra~ Of polyt.!i'anedral ma((Dstlte. The rot" tional partition funaion i5 therefore exactly the same as if W~ were 10 assume the p"rtlcles to be

i ndis nng~ is h. ble.

2 t G. FJ;m~~, i: ~ R" liil!.. P [iDrrmoll~, I C/leJII Phy,. 28. 14% (1993).

22 ), Doye, D Wales, l. [hem. Soc., Faraday trons. 93, 4233

(1997).

;;3. N. 10 bnsnn, [ali. J, Mat/I. 18, 169 (196 6). 24, I, Bernat Nature 18S., ~8 (1960),

25, V. N. Manoharan, M. IE[<;e'l~t, 0.]. Pi~e. Scfeflce 301. 48, (2(1m).

26. A. 5tradner lOt al., Na/llre 432, 492 (2004).

27. P. L Biancaniello, A, I Kim, l. C. Cro'k'~r, P/)JIS, Rev. Lett_ 94, D58302 (2005).

2B. 5. Makura, F. O~5aw~, }, (hem. P/1YS, 2".2, 1~5 (1954).

29, II. V~j, Pure Appl. (hem. 48, 471 (1976).

3 (I, We 'j ha ok F. Sjlal'p.eo an 1'1 Z, eheng lor h el plul dj)OJ5siDns. W~ a(~nDwledge ilJPpor'l from N5F untl~f award number> DMR'OBio4B4, CBE'r'07 476.2 5, and

DM 5-0907985; from the o~tense A dva nc€ d Re!e;1 r c h Projects Agenql tinder contract BAA 07·21; and from [he Kavli In sti [U Ie 1m Bian ~ no Sci. no" and T "dmllill9Y ~ t

H arva rd U~jve rsi ly.

Supportiflg Online Mi1t(Hjal

WWW,,(iencemag ,Ofg/Cg IIw ntel1l1fu 1113 2 715 9 65156 OID( 1 Ma lerials ;1Il 1'1 Meth 0 ds

Ftg,. 51 and 52

Tables 51 and S2

MoviE'S 51 to 59

~ a Aug us I 2009; a [["pled 10 Dec ern her 2009 W.11261Ide~ C€ .11 B 12 63

www:soeneemag.or9 SCI ENCE va L 327 29 JAN UARY 2010

563

REPORTS

A Tricyclic Aromatic Isomer of Hexasilabenzene

Kai Abersfelder, Andrew J. P. White, Henry S. Rzepa," David S,heschkewitz*

Benzene represents the ~hQwcase of Hikkelaromaticity. The silicon analog, hexasilabenzene, has consequently been targeted for decades, We now report anintensely green isomer of Si6R6 (R being .2,4,6·trHsopropyLphenyl) with a tricyclic. structure in the solid state featuring silicon atoms with two, one, and no subsrituents outside the ring framework. The highly dispersed 29Si nuclear magnetic resonance: shifts in solution ranging from +125 to -90 parts per million indicate an inhomogeneous electron distribution due to the dismutation of formal oxidation numbers as compared with that of benzene. Theoretical analysis reveals. nQnetheless the cyclic deLocatilation of six mobile electron~ of the 11>, (J- and non-bonding type across the central four-membered ring. For this alternative form of aromatidty, in principle applicable to many Huckel aromatic species, We propose the term dismutational aromatidty,

B 6IJZSnS has stimulated synthetic and theoretic a] chem ists sine e its d iseovery by Faraday in 1825 (1). It provided the basis for the famous rule associated with Hackel that the presence of 4n + 2 cyclically delocalized n-electroos imp lies aremati city and related stabiIization (2). Controversies regarding the concept continue to this day to provoke much productive research (3)_ Not surprisingly. the potential energy surface (PES) of Ce;He; is one of the best Studied, More than 200 isomers have been identified, all being -at least 60 kcal mol ' hi_gheJ in energy than benzene (4).

The prospect ofarornaticity incompounds involving silicon was ignored for a long time because silicon was generally considered unable to f01111 stable n-bonds (5)< However, in 1981 West ei at, reported a stable disilene, a compound wuh a Si-Si double bond sterically protected by bulky mesityl (2,4,6·nimetbylpheoyl) groups (6)~ Since then, about lOO disilenes (7) ami evert a few compounds with Si-Si triple bonds (8, 9) have been isolated, taking advantage of steric and/or electronic stabilization.

Cyclically del ocalized 1\ -sy sterns involvi ng sj Iicon sbifted into focus (10), and numerous breekthroughs have been achieved, such as the isolation of stable sila- and 1,z"disilaben7.6nes withplanar, Hlickel·arornatic structures (11, 12). PotootIDJly aromatic com pound s based on srli con alone are so fat limited to persila-analogs of eyelopropenium cation (13) and {.'Yclobmudiene dianion (14).

Because of the special role of benzene, a particularly entidng gila-aromatic compound is the elusive hexasilabenzene, After early predictions. of a sixfeld symmetry analogous to that of benzene (15), a chair, like conflrmatien resembling that of eye lobexane Was established through theoretical calculations U 6, 17), Experimentally, only an isomeric, nonaromatic bexasileprismane

I mperiiil Co![eg~ London, Department of Chemistry, london swr 2A1, U i{.

'To whom rorrespol1denc~ should be addressed. E-mail; ds[heo;,hkewltz@imperiat.ac.uk (D.S.); nepa@impertilt.dc.tlk (H.5.R.)

564

has been prepared by the reduction of a 1,1,2,:2- tetrabalodisilane (18), suggesting that the choice of precursor may be ·important in any synthetic approach.

Awcl)' to prearrange half of the Si5-seafi'old became apparent to us it! light of out previous results conecming the reaction of disilenide I_ with au excess of dichlorosilanes affordiog cyclofrisi lanes (J 9).

Extrapo laring these results to silicon tetrach 10' ride (SiC4l, we anticipated that cyclotrisilane 2 lnay become accessible, although the reaction of 1 with a shortage of SiC", yields a Si;51'ipo-cluster (20)., Conversely, rapid addition of an excess of SiC11 to disilenide 1 (21) afforded 2 in 59% isolated yield (Scheme I) (22). Given the unsymmerric 1,1.2-sum,'1:itutionpattem of2 (confirmed by means of x-ray diffraction) (figs. 82 to S4) (22), we expeered it to provide an alternative entry to Si"R,; chemistry. Reduction of 2 with three equivalents of lithiluTIIllilphthalcne in tetl'llhydrofi.mllv dietbyletber (3:5 ratio) led to the product 3a (R being 2,4,6-triisopropylpbenyl, or Tip), which was isolated as dark green crystals in 52% yield (Scheme 1) (22)- The surprisingly stable 33 can be exposed tb airtor hours as a solid or for minutes in solutioo without detectable changes, Meltin,gassociated decomposition was observed at 216"C.

The highest mass in the nJ3ES spectrum of3a is observed at 13MB mass/charge ratio (m/z) with the correct isotopic dismbution for Si6C~{)HI3g indicating the formarion of a dirneric structure, On the basts of a two-dimensionaj29SV' H nuclear magnetic resonance (NlVIR) correlation, !be 29Si NMRsignaisat l24.6, -84.8, and -89.3 parts per million (ppm) were assigned to tWO silicon atoms eaeb with one, two, and no directly attached Tip

2

3a

Scheme 1. Synthesis of the dismutational hexasilabenzene isomer 3.11 (Tip, 2,4,6-triisopropylphenyO.

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

fig, 1. Structure of 3a·2C>loH~ (thermaL ellipsoids at 50%). Hydrogen atoms and naphthalene fnol~w\e~ are omitted fof cia rity. (I mer!) Photographic image of darkgreen crystals of 3.11 'C6H6• Se Lected bond Len gths in angstroms (estimated SD) are Sil-Si2: = 2.3581 (5), Sil-Si3 = 23275(5), Sil-Si3' "" 2.3034{S), snSi1' = 2.7287\J), and SilSi3 = 23375(5).

substituent, respectively. TIle most positive 29Si NMR oh ernie at shift it; observed in the region typical for terrasilyl-substituted Si=Si bonds (7) and fal' from the negative region where signals of tetracoordinare silicon atoms within small rings are found (18-20). This observation excluded any saturaredstructure as 'a possible product The groen color of 3o, due to the longest wavelength absorption in the ultraviolet (UV)/vis spectrum at " = 62] nrn, would also be highly atypical of saturated silicon compounds.

An x-ray diffiaetion study on single crystals of 3:. con finned the features deduced from the spectroscopic data (Fig. 1) (22). The hexasilabenzene isomer 3a crystallizes as a naphthalene solvate (3a·2 ClOHg). Despite Its tricyclic con· nectivity, the chair-like confbrmatiun of 3;1 resembles the predicted puckered structure of hexasilabcnzcue (16, J 7). 3a exhibits a rhomboid S4-ring in the center (Si L, Sil and symmetry equivalents Si I', Si3'), with two opposing SiTip! bridges (Si2, Si2') pointing up- and downwards with respect to the S4 plane [the angle to Si 1· Sa-sil is 66.80(2)"'; with estimatedSD in paremheses]. Sil and Si I' showhernispherical coordination; all bonds poin; into one-half of an imaginruy sphere, which is a common coordination in case oflcad (23), tin, and germanium (24) but rare for neutral silicon compounds (20, 15). Despite tbe SiTip2. bridges, Si I-Si3 [2.3275(5) A] is Duly slightly longer than Si 1-5i3' (2.3034(5) A], both being at the shorter end of'typioal Si-Si single bonds, TI1e rhomboidal disrornoo becomes apparent in one shorter diagonal distance lSi I-Si]' 2.7287(7) A], which nonetheless is almost 17% longer than typical Si-Si single bonds.

00 the basis of the analytical data, a resonanc-e hybrid between A and 8 seemed best

suited to describe the electronic structure of 3a (Scheme 2). Contrasting the usual six n-electrons of Hiickel aromatic systems, the mobile electrons in 3ll would nominally be ofthreediffereot types. Two n-, two 0-, and two nonbonding electrons could be cyclically delocalized over four SiliC011 centers with formal interruption of the a-framework by twosanaated SiTiP2 homobridges (26), Because of the topological similarities to singlet diradicals of the Nieeke type (27), however; we initially did not exclude acertain diradical character in 3a (resonance form C).

Density functional theory (DIT) calculations (22) on the Dip-substituted model compound 3b (R being 2.6·diisopropylphenyl, or Dip) at the B3LYP/6~31G(d)-Level of theory clarified ihe bonding situation (interactive table), The experimenral geometry of 33 is reasonably well reproduced despite substitution of the. paraisopropyl _groups with hydrogen atoms [for example, diagonal. distance 31:): Si l-Si l ' 2.779A (calculated), 2.7287{7) (observed) A]. The experimencal trends to the 29Si NN.IR shins. of 3u agree welJ with those calculated for the model compound 3b [I) = 170, -4o, and -83 pprn, B3LYP/6-311 +G(2df) basis on Si; 6·3 IG(d). on C,R]. Additionally, the UV/vis spectrum of 3b simulated with time-dependem OFT (TDOFT) calculations is very similar to the experimental spectrum of 3a. The" closed-shell model 3b appearedjustified by the conformity of experiment and theory, which we lent further SUpp01t by optimization of'thetriplet :3-A" state of3b, revealing a much longer diagonal distance (3,069 A) between the substituent- free silicon atoms. The adiabatic singlet-triplet gap ofEs_T= 24.l keal mol I and the dominant (>90%) contribution of'a single closed-shell CQn figuratio n to the ccmplete acti ve

R

I

R SI R

\ ." -

SI~St I

/..St SI,

R-Si"" R

R

A

Scheme 2. Resonance formulae (A) to. (t) for 3a-c [for 3a, R '" Tip; for 3b, R = Dip (2,.6- diisopropylpheny[); for 3(, R = Me).

HOMO

IHOMQ..1

REPORTS

I

space self-consistent field (CASSCF) (8.8) computed multireference wave function for the parent model (3c, using the Si geotnctry computed for 311) supports a low diradical character for 3b (18).

Inspection of the relevant molecular orbitals (MOs) of3b (Fig. 2 and interactive table) revea Is that the highest occupied molecular orbital (HOMO) (orbital 309), HOMQ--Z, and HOM0-3 (the disrnutative trio) are ali cyclically delocalized around the central S~ ring, The HOMD-3 is also conjugated with the aryl rings of four of the Dip groups (which is cbaraeteristic ofn-aromatic bonds). Tn contrast, the HOMO~ I IS dearly atnibutableto the larger c-framework of3b, which includes the two bridging Si atoms and displays the typical appearance of an oligosilane system (29). TIle lowest unoccupied MO (LUMO) oOb is anrihonding between the substiruem-fiee silicon atoms, corresponding more closely to the diradica1 resonance fonn C (Scheme 2), which is in agreement with the longer distance between those atoms calculated for tile :3 A,! state Of 3b,

The topology of the total electron density in 3c (R being Me) was initially probed by Bader's atoms-in-molecules (AlNt) method (3(J). TIns analysis revealed bond critical points (Reps) for both the annulated three-membered rings and for the central Si.rll10tifbutoot for the long diagonal.disranee; corroborating the absence ofdirect bonding between Si 1 and Si I ' (interactive table), The value of the eleerrcn density p(r) for BCPSil-Si3' [0.0'78 atomic unit (.,u)] was higher than for the homobond BCPSil~O;-'3 (D.D?]), which hints at a cyclically delocalized through-bond interaction. An electr-on localization fimctiou (ELF) analysis (31) computed for the total electrons shows two disynaptic basins fur Si {-SB/Si l' -Si3' [1.42 electrons (e) each], two further disynaptio basins for Sil-Si3'/sil'cSi3 (1.96 e) and two monosynaptic (ncnbonding) basins. located on sn and Si t' (1.60 e), for which resonance between fonns A ana B (Scheme 2) is it fair zero-order representation. TIle disynaptic basin centroids approximately correspond to the AIM HCPs. If the ELF is restricted to contributions from the top four MOs (disirmrative trio plus HOM 0-1), the same basin pattern is revealed (with further basins arising from ihe HOMO-J (interactive.tsblej]. TI.e totalelectron ELF isosurf a ce includes a. delocalized

Fig. 2. HOMO, HOMO-l. HOMO-2, and H.OMO-3 of mode! ~b contoured at 0.035 atomic units (interactive table),

HOM~?

WWW.sclencemag.org SCIENCE VOL 327 29 JANUARY'2010

565

REPORTS

region around the S~ ring with a bifurcation threshold (O.755) similar to that "reported for homoaramatie carbon rings (32), This feature is also present in the BLF isosurface, including j list HOMO to HOMO-]. When the HOMQ-I is excluded, the BLF surface now starts to resemble the topology of the torus. link computed USing the 1t-MOs of benzene (33):

To order to quamify the aromaticiry of 33,b, we calculated the nueleus-independeut chemical shift, NICS{O), at the center of'tbe Si.J ring of3b (-23.8 ppm), which indicates substantial aromaticity (benzene --IO ppm) butmay also include shielding effects from thea-framework (34), To estimate these latter effects, we computed the NIC.s(O) value for 4, the hypothetical saturated hydrogenation product of 3b. This in silico reduction has the eJfeot of seqlle~1ering the two Si 1001e pair electrons and hence suppressing tbe dismutatioual resonauee, The result (--6.4 ppm) suggests that the strongly diatropic NleS(O) value of 31> is truly due to aromaticity. Further confirmation is obtained from the NJCS(O) value of-3.3 ppm for !he :;A,I triplet (and presumed nonarornatic) st;Jte of3b (resonance C) (Scheme 2).

In order to place 3b in terms of relative en.ergy, we calculated two of lis isomers with Dip substitucuts; the experimentally known hexasilaprismane U 8) and the hypothetical bexasilabenzene, Both turned out [0 be lower in free energy .1.~s [B3L"':fP/6-JIG(d); 3b, 0.0; pnsmane, -1 L 7; benzene, -4J kcal mol 1] with tbe hexasilabenzene surprisingly situated midway. which "raises the intriguing possibility of II future synthesis of'a stable bexasilabenzene ..

The general formalism leading to the type of aromatioity exemplified by Jll-C is a twofold formal 1,2"'shifi of substituents in the classical

Huckel aromatic compounds: a double. intramolecular dismutarion. The. formal oxidation states of the silicon atoms in 3a-c arc +2 (SiR2), +1 (SiR), and 0 (Si) as opposed to the unifonn oxidation state of + I in hexasilabenzenes .. We propose the term dismurarional arematicity for a phenomenon that in principle should be applicable to any classical Huckel aromatic compound with at least six ring atoms,

References and Note~

1. M. ~afoday, P/Jilos. trans: ff. Soc lend. B Bioi. "5,;. U5, 440 (1825).

2. w. von E. Doering, F. L. De,er!; 1. 11m. Cliem. 50(". 13, 876 (19~1).

1. A r. BaiaMO, p. R. Schleyer, H. S. R~pa, (hem. Rev- 105, 3436 (20051'.

4. T. C. DI~adayaliln~, U. D. prjyakumar, G, ill Sa!lty. j. I'IiY5. Chern. A 108, 1Mfj' (2004).

5. F-. A Cotto~, G. Wili<.in"on. Advanced InorgQllir Chemistry (Wiley, New York. ed 4, 19M).

~. R,. We!l, M~ 1 Fjnk. J., Mio;~l, Sderr,e 214, 1343 (1 ~81)_

7. M, Kira, T. Iwamoto, Mv. Organomef. {ileme 54, 73 (2006).

S. A. 5ekiguchl, R. K.illjQ, M. khioQne, Science 305, IJ 5S (2.004).

9. T. Sagmo~ eI al,J AIJI. Chern. SOt. no. 13856 (1008).

10. V. y, lee, A. Seki'luchi .. Angew. Ch"m. Int. [d. 46. 6596 (2007).

11. K. Waki!a, N. Tbkiioh, R. Ok:r,;,ki. s. Nag~,e, An'leW_ (hem. Int'. Ed. 39, &:l4 (~aOo).

12. R. Kinjoet al., I. Am. Chem. SiJ(. 1~9, "1766 {lpon,

13. M. I chi no~ e, M. Igor <lIM, It. Sa nu k.i, A. Sek1g~chi, j. 11m.

Chern. Sot; in, 9978 (2005)

14. V. Y, lee, K: Ta~iln.J'hi, T .. Mat;uno. M. khinohe,

A. 5ekigul'hi, J. Am. (hem. 5Jl(. 126, 4758 (2004).

15. M; J. 5. Dewar. D. H. L{), C. H. Ramlden, 1. Am. Chem, Sb,. 97. 1~11 (1975")_

16. S. Nagas~, H, teramae, T. Kudo, J. {hem. p.hy,. 86, 45.13 (1987).

1 J. K. K .. Sa ldfid ge, D. U zan.], M. 6. ME rtln, Org(lIlomet aI/irs 19, 1477 (2(100)

111. A. 5ekigu{hi, T. varabe, C. l\abutQ, H. Sakul"lIi. j. Am, Chem. Soc, rts, 5853 (1993).

19, K. Ab"J:<; IeId~r eO. S, n""qh ke.l'/i tIe}. Am. Ch~",. So C. BO,

4114 (2008).

2.0. D. Scheschkewjtl:, Allgew: (hem. lilt. {d. 44, 2954 (200S). 21. D. 5chl".smkewitz, Allgew. Chem. IIlL Ed. ,U_ 2965 (2004). 22 _ Mil ~~ Ii;, ~ "nd m"th 0 d, are ava-iii! blio a, $Up ~o rti iIg

material on ScielKe Online and (ontain details of €xpertmenta I p rocedores, a nalyb cal d ata, <1 n d x· r. y 111 uctur~ determinatiens. Derail> of the corn putstlo nal

p rocedures a'e a vaila bl.e vi" the. i [lte,activl". ta ble and tne dlQ ital re 1'0 lilory lie ks therein, Regll(di ngge neflll

info r mation on the di gilal r.epo>i to!)', 5 ee (J 51'.

23. l. 5hlmoni·Uvny, J. P. Gllll"k€f. C. w. fiod:. l/lor9. Chem.

37. 18'>3 a 998)

24. A. Sehn1l;pf, Cliem. soc. Rev. 36,745 ('2007).

25. G\ Fil(h~, e/ et; AllgElv. 'hem, Int. Ed. 44, 78&4 (2005). 26 O. Z~ang tf ai., J. Am. Chem. Soc. 131, 9789 (2009).

27. o. roIiee!..'!', A, Fuc.hs, F. Baumfi,ter, M. N leger.

W. W. Schoeller, Angew. (/Iem. 1'11. Ed. Ell'll. 34, .555" (1995).

Z8. M, 5<'. i ersta d, c. R. ~ insi nger, c, J.. o "me" I1ngI;'W. CIi~If/.

In!: Ed. n, j 894 (20021.

29. T. Sthep~r5.l. Mid1~ j. I'nys. Org, {/leJ!1.c 15, 490 (:2002).

30. R. F. W. Bader, Atoms in Molecules: A Quantum Theory (OXfQld U 'Ii". Prej~, O#ord. 1'~ 0).

3:1,. A. S~vin ei. ol; An!!e>I', (hem. / 'It [d. Engl n, 187 1;1.992).

32. C. S. M. Allan. H. 5. Ri;ep;J, ],Chem, Theory Compol. 4, 1811-1 (2 008).

33. L 5, W~n~Eri' e( al.,!, Phr" them, A 1:1), l1619 (2009)

34. l. Chen, C. S. Wan n ere, C. Corm i nboeuf, R. pu chta, P. R. S(hle~" Chem. Rev. res, 3842 (,DOS).

35. ]. DOIMling et 01 .• J Chem. In]. Modd 49, 1571 (2008).

36. WI! tha nil ihe Di'utst:h ~ f or5~ho n9'ge [I1<'.i11"""1t (0 FG S(H E 901iJ3- Z) iI nd th6 KiI rl- Wi nna eker- ru nd 0 r I~ e Avent;, Foundation for ffn"oc;~1 5UPPQrt. Cryst~llo9r"phi( d eta-II> were Ifepoli ted wi th the Cii mh I'i dg e (.ys:tallo~tilphic o"r.. ~e~!(g as (('0(-7456<\0 (.2) and

7 ~5,681 (3a).

Supporting. Online Matefiill

IWJW .>Lien(e~'a9.{)rglrgil(o nte11l1fu 11/3 2715 9 55/S6 ~!DC! Mil.l!'l'iall and Methods

f1g,. S L to 56

Ref~rence

Inte'ilctive Table

10 September 2009; accepted 24 Nov~mb€1 2009 1O.ltl6hdence.l1S1771

Combined Effects on Se,lectivity

In Fe-Catalyzed Methylene Dxidation

Mark S. Chen and M. Christina Wh'ite~

Methylene (-11 bonds ate among the most difficult chemical bond~ to selectlvely functionaljze because of their abundance in organic structures and' inertness to most chemical reagenl5, Their s~lective oxidations 1tl biosynthetic pathways underscore the power of such reactions for streamlining the synthesis of molecules with complex oxygenation pattern s. We report that an iron catalyst can achieve methylene (-11 bond oxidations in diverse natural-product settings with predictable and high

cherno-, site-, and even diastereoselectivities. El.edronic, steric, and stereoelertronic factors, which individuaUy promote se[e~tivity with this catalyst, ate demonstrated to be PQwerful (antral elements when operating in wmbination in complex molecules. This small-molecule catJlyst di~p[i;ly5 site selectivities complementary to Ihose attained through enzymatic catalysis.

McthYleile (secondary) C H bonds are ubiquitous in organic structures and are often viewed by organic chemists as the mert scaffold upon which tbe traditional chemistry of "reactive" functionaL groups is performed, In contrast, the enzymatic oxidation of -methylenes (i.e, C-H to C-O) 1:; OJ fundamental tran sforma tio n in bi ologi cal system s an d is crit-

566

ieal for drug metabolism and the. biosynthesis of secondary metabolites (1, 2). .Selecti vity in enzymatic catalysis is dictated by the local chemical environment of the enzyme active site, a fea ture that inh e rent I y lim its su bstrate scope. Despite important advances in the discovery of catalysts for C -H oxidation; the abi lity to directly funetionalize isolated, unactivated second-

at)' C~H bonds with usefirl levels of selectivity U1 complex molecule settings under preparatively useful conditions (i.e., limiting amounts of substrate) has been restricted to- the realm of enzymatic catalysis, A small-molecule catalyst capable of performing methylene oxidations with broad scope, predictable selectivities, and iJ:I preparatively useful yields weuld have a transformative effect on streamlining the practice of organic synthesis.

The paucity of methods for the oxidation of isel ated, unactivated, and noneq Divalent secondary C H 'bonds underscores that they are, arguably, the most challenging chemical bonds to selectivyiy functionaliz~\ Reactivity for oxidizing such bonds had been observed with several catalysts, but generally in substrates where selectivity issues arc circumvented {e.g"llyciollCXanc --j. cyclohexaaone) (3-11} or else where reactive sites are either electronically activated (i.e., adjacent

Departm ~nt of Chemistry, Rog ~r A do ms labora lory, University of lI!inols, Urbana, Il 61801, USA.

"To Whom rorrespondente shollid be addressed, E'mail; whlte@S(s.t~u(.edu

29 JANUARY 2010 VOL 327 SCIENCE www.seieneemag.org

to 1t-systems or heteroaroms) (12-15) or oriented favorably toward a directing group U 6-19). The electron-rich nature of tertiary C H bonds and their relative scarcity makes them more facile targets fur selective oxidation (3-5, 19. 20). Bioinspired catalysts with elaborate binding rockets have been viewed as (he mO~1 promising candidates to effect site selectivity in unactivated, nondirected methylene oxidations (21).

We recently reported a bulky, electrophilic catalyst [Fc{S,S-PDP) 1] (Fig, IC) (3, 4) that uses H202 with acetic acid as an additive (22) to stereospecifically oxidize isolated, unactivated tertiary G-H bonds across a broad range of substrares, Surpnsmgly, predictably high levels of site selectivity were shown to be possible with 1 based on subtle steric and electronic differences among multiple tertiary C-H bonds. Encouraged by these results, we hypothesized that such a catalyst that is highly responsive to II combination of electronic and steric effects would be well suited for selective secondary C H bond oxidation. Chernoselectiviry advantages for secondary C --H bond oxidation may arise from increased steric accessibility compared to tertiary C-H bondsand greater electron-rich character compared to primary C H bonds (Fig .. IA). Because 'secondary C-H bonds have intermediate electronic properties, they may be particularly

~

Sl!Jer:ttvJ,,.

Fig. 1. tAl Chemical properties of aliphatic C-H bonds. (B) Synergistic effects on site selectivity. Steric, stereoelectronic (influence of 0 rientation of electron orbitals in spate on reactivity), and e lectronic influences on reactivity with c"talyst 1 have an additive effect in complex molecule settings, which can lead to highly predictable and selective outcomes .. (c) Electrophilic, bulky catalyst Fe($S)-PDP (1) for predictably selective aliphatic C-H oxidation,

amenable to tuning by both electronic activation and deactivation. Moreover, since secondary C H bonds m-e prevalent in ring systems, a variety of stereo electronic. effects (i.e., th e influence of orientation of electron orbitals in space on chemical reactivity) may be readily exploited to achieve high site selectivity in their functiona lizatien.

Here, we report that the same bulky, electrophilie Fe(S,S-PDP) I catalyst is capable of site-selective oxidation of isolated, unactivated secondary C- H bonds to afford mono-oxygenared products in preparatively useful yields=without the use of directing or activating groups. The site selectivities can be. predicted in complex rno leeular settings 00 the basis of steric, electronic, and stereoelecrronic rules derived from simple compounds, Whereas individuaJJy these effects ill some cases afford modest selectivity, when manifest in combination ill natural-product settings they result in useful levels of chemo-, site-, and even diastereoselective methylene oxidauons (Fig, l8). Moreover; because site selectivity in the case of small-molecule catalyst 1 is driver) by its sensitivity to the local chemical enviromnent of the substrate, we can achieve site

Fi g. 2. Ele rtronici nflue n ces (ind uctive effects) on site selectivityi n the oxidation of secondary {-+I bonds of aeyeli c and cydi c substra las. An ite ra tive a ddi Ii a n protoc 0 I was used in which catalyst 1, H20, o:<ida[lt, and AcOH additive were added in three portions avera 30-min perio d to decrea se the rate of bior rnultirnolecular catalyst de composition relative to substrate oxidation (3, 4, 23). Methylene C-+I oxidation with 1 orcu rs preferen tia Uy at the methylene site most remote from EWGS; !Sm, % recovB(ed unoxidized starting material. The average mass balance for entries 2 to 13 is 94%.

REPORTS

selectivities complementary to those obtained in biotransforroations.

TIle highly electrophilic oxidant generated with (S,Sj-l and A202 allows for substantial electronic control aver secondary C-H oxidation site selectivities (Fig. 2). n-Hexane, a hydrocarbon wifh no electronic biasing elements, is oxidized with no site selectivity (primary C-H bond oxidation bas not been observed with I), fonning ketones 2 and 3 in equal amounts (Fig. 2, entry I) (23). The oxidation of methylene C-H bonds fumishes ketone products through alcohol intermediates via two stepwise, rapid oxidations (alcohols have been detected under conditions of limiting oxidant). A single electron-withdrawing group (EWG), such as a terminal ester, deactivates proximal sites: via inductive effects, thereby introducing differential electronic environments that bias the site of oxidation. Specifically, high selectivity for ketone. formationis observed at the most electron-rich methylene site (~50% yield), corresponding to that farthest from the aWG (entries 2 to 5). Consistent with our expectations (see above), the electronic biasing effect 00 methylene oxidations with (s'S)-J appears to be very strcng; electronic deactivation by the ester

EWG. . __ '-'k. . ~,j( -....

H H

Entry

Millor Oxidation Pwduct Isolated % ViaJdt ('iorsm)

Sittl~o' MInor ,Oxidation %'II'lold

5

C5,18% CI.,14%

7,4iJ%,II22%)

6 R 8, A.O 0% (sr.,)
1 (] 9 A OPII 3O%l41%) C2.14%
G I~O. FI .. CO~Me 5'2%137%1 C2. 2Q"
9 rt, R '" CHzOPl1l ElO%'(2S-") C2. Ir~
10 b n,R ... O 38~133%) C3. 26%
11 13. R",OPIV ~,(22%) C3,24'lo
12 14, A =- CO-,!Me 45~131%) 03,28%
13 1:!ii. A= CH10P,i 43%110(,%) ca,31% • ~1I1!l1l11O ddlU:)tl pr~OOI)l Un OINt· nClr.-dnll ClGl"nPlllI1lCh; Wf'W ~Il:JIOO All !Jog. CMdlzod proc!IJcua l Be VlakI.' IMllniH III ~. ~ .. nscn l!JII'f!b.I1iCIQ!i1 IlnIfurG CII[C5;t;;:i5J 0lC,Wl1lOIl f!lOdUdt 'I«II~JOd 11'1 ~ ¥Ira as

nb'''IHII'b!7111 I.I1"ot'PC2!~ 0

www:screncemag.org SCIENCE VOL 327 29 JANUARY 2010

567

REPORTS

moiety in methylheptancate 5 results in high selectivity fur oxidation at the remote C-6 position relative to four other proximal secondary sites (entry 3, ;) J % yield). Wbeo tertiary sites arc positioned a or ~ to EWGs, the above trend still holds, highlighting the utility of this tactic for effecting the chemoselective oxidation of secondary C-H bonds in preference to tertiary C-H bonds (entries 4 and 5)"

Electronic deactivation of proximal mcthyl ene sites also leads to predictably selective outcomes for oxidations within five- and seven, membered ring systems, ill all cases examined, the major product is a result of keronization at the methylene site most remote from the EWG (Fig. 2, entries 6 to 13). Consistent with selectivity trends round withacyclic substrates, cyclepentanone shows no conversion due to strong electronic deactivation of secondary sites 11 or ~ to the carbonyl (entry 6). Increased product formation as high as 60% (11, entry 9) is observed as the EWG is rendered less inductively withdrawing and/or more remote from the ling (entries 7 to 9). Similar reactivity trends are observed witb the oxidation of seven-membered rings; however, the benefits of an EWGort site selectivities are less pronounced (entries 10 to 13). For example, as tbe EWG is rendered less inductively withdrawing, the increased reactj vlty is 0 ffset by th e erosion of site sel ectivi ty due to increased oxidation at the more pro xirnal C3 position (entries 12 and 13), TIlls trend reveals that a careful balance between substrate reactivity and selectivity must be struck wheo relying on electronic deactivation alone to achieve site selectivity with these simple substrates.

Examinatiop ofa series of six-membered cyclic substrates with BWGs demonstrated no pref erence fur the more electrorucally favored C4 methylene site over the more proximal C3 site (Fig. 3A, entries 1 and 2). However, an increase in [C3:C4] site selectivity was observed witb a sterically bulkier ring substituent that was no. longer electronically deactivating (i.e., tert-butyl, entry 3) These trends suggest that stereoelectronic parameters based on confbrmarioual ef~ fects are strong conthbuting factors to the observed product distribution in six-membered ring oxidations with 1 (entries J to 3). ln these cases, torsional strain between the bulky equatorial group and vicinal rnethylenes, generated by unfavorable 1,3.dia.xia! interactions at Claud C3, IS alleviated upon C3 C H oxidation (24). III addition, such sterically bulky substituents a lso contribute to selectivity by inhibiting oxidation at adiaeent sites .. Methylene sites adjacent to bulky z-butyl or gem.dimethyl groups are disfavored in oxidations. with (8,5)-1; a positions are consisteatly oxidized in the lowest statistical yields (2% C2-ketone, entry 3; Fig. 3B, emry 4).

The bulky nature of the (s'S)-t catalyst even allows for subtle steric influences to have notable effects on the chernoselecrivity of secondaJY versus tertiary C -H bond oxidations (Fig. 3B, entries 5 and 6). ds~Isomer 2:9, containing

568.·

equatorial tertiary sites, undergoes preferential tertiary bydroxylation to provide alcohol 30 in 55% yield (tertiary.secondary = 4:1, entry 5). In contrast, oxidation of lrans·isomer 33, containing only axial tertiary sites, affords predominantly secondary C-H bond oxidation products 35 and 36 io a combined 50% yield (tertiary: secondary = 1 :2.enlly 6). Although a preference

for equatorial versus axial tertiary C-H bond oxidation has been noted for dioxirane oxidants (5), a reversal in chemoseleetivity to favor secondary C. H bond oxidation bas not been previously reported. This is most likely because the strong electronic preference for tertiary hydroxylation with an electrophilic oxidant can be overridden only with a very bulky oxidant

?'7

1~1"'.

E ·1"" StIlrtil"llll.. Mat.hl'

11 • r (I'iIm)

O.lCid,Dtian Froduets "'f9 Yloldt

5

fI_£':::Y

17·;,43% 2O~,.%

33,j11%)0"

18f 21% it.t,2O'

.~-f-pd0

2',1.2%

o . ~O

~+~11+e

o . 0

~.+~+~

l1enl8ry seoond aryl <1 1 o

._:p::::1 + ..::1Jt::J + _::z::::::::r

OH $4, 29"tO as, 22% I '6.2M4

lleft1a'Y IiE!CXIIlti aryl .. , '2

• Il;)Ui1'1Il:! 4IIt.f1t",," Ploill«l L • (111 W U I .!KlI'lc!d 4r" !$01t!1Id • pl!llI prodlld:J. 1111_ Ill ...... wlli!;l flol."j I IliDluled 1'S.;m '~l"I'bI!" m .. llln;l III 1:3. lllld c.4d.u~.... I' C"-

kelOlIB was Ol:PSefll9Cll 2".. oc ,.19111. EId~mlne!l1ly cae al!a'vSls

Fig. 3 .. Stereoelectronk, sterie, and electronic influences (hyperconjugative activation) on chemoselectivity and site selectivity in the oxidation of secondary C-H bonds in six-membered rings. (Al Alleviation of torsional ring sltain in six· membered rings favors formation of C3- over C4cketone products.. Average mass balance for entries 1 to 3 is 89%. (B) Steric factors can lead to both site selectivity and chemoselectivity for secondary C-H bond oxidation in preference to tertiary (.-H bonds with bulky catalyst 1. The average mass balance for enmes 4 to 6 is 86%. (C) Electron activation (via hyperconjugation) of aliphatic secondary C-H bonds for site-~elective methylene oxidation; rsm, % recovered unoxidizsd starting material.

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.erg

11

like that formed with catalyst (S,5')-I/R 201. We hypothesized rhat the collective influence of these individual steric effects au site selectivity can have a. combined effect on the selective oxidation of substrates with increased conformational rigidity (see below),

Although not required, eleetrome activating groups (EAGs) can be used to achieve site selectivities witb catalyst 1 that are orthogonal to those based on inductive or steric effects (Fig. 3C). Groups capable of byperconjugative activatiou, i.e., donation of electron density from a fined orbital to tbe antibonding orbital of a C H bond, may be used 10 activate adjacent C-H bonds toward oxidation. The strained cyclopropane ring, having appreciable p-character in the C-C

A

3

second'S1)' C -H S'ltes : 1 Imlaty C-H sfles·:2

($,$).1 (25 moI%)

o AC:OH (0.5 '€!qUill )

HA (5.0 eqUIV.' 'I.., _~-..J

slow addition 1~rsm

" +49,(1 equlv.

bonding orbitals. can activate the cyclopropylearbinyl uietbylenes through hypercoojugation (25). For example, cyclopropanes can direct secondary C H bond oxidations at sites that are sterieally or inductively deactivated, Although quaternary alkyl substitution diverts oxidation away from o-bydrogens on II cyclohexane ring (Fig. 38. enny 4), selective oxidation at these sites occurs when a cyclopropyJ group is installed (Fig .. 3C, entry 7). The site selectivity in the oxidation of methylbeptanoate can also bc altered to favor the CS position (disfavored based 011 inductive effects: Fig, 2, eotry 3, C5:C6= 1:3) via the incorporation of a cyolopropyl moiety (Fig" 3C, entry 8, CS ;Cfi = 6; I). Following tlte previously established electronic treeds,

H {. ·50. 53~

B

(+51. (I equiv )

REPORTS

I

the more electron-rich of the two methylene sites flanking the cyclopropyl ring is still selectively oxidized. No oxidation of tertiary C H bonds on the. cyclopropyl ring was observed Another powerful form of hypercoojugative activation is the interaction of lone-pair electrons of an ethereal oxygen with adjacent C:':A bonds (26, 27). This electronic activating e'l'tect' enables the selective oxidation of five- and six-membered cyclic ethers into lactone products (entries 9 and 10). Lower product yields and poor mass balance in these cases can be attributed to overoxidation of the open-chain hydroxyaldehyde form of the intermediate lactol, Consistent with this explanation, when a seven-membered cyclic ether is oxidized. lactone is not observed but adipic acid (48) is

SIJCOIJdsry C-H sIles : S

tenliJtyC-H sItes: 3

(R.FI)-1i (26 mol%) AcOH (0 IS equrV')

HP, (5.0 eqUlv,)

sfo w 8IkJ1tiO(J

5"4 rsm

e

11

2~iCO

A B.

, ,

, 5

~ . .;;.

H'amb,oXlde 68 f1 equiv)

Enzymatic Oxidation Botryl'fS Cinerea

St1COndary ,C H ,ltcs : 7 tertiary C-H fllles:2

(R. R)ot (15 moI%) AcOH ('5 equi ,)

HA ,(3,'6 eqvlv ) jtOfal! II(J UdlJifl()iJ

1'S%rsm

o 0 ,.. 2%( ).$4

and IWO otner oxidation prodtlct6

" OH

f!O%.

set:Ondary C-H s-'tes : 11 retlJary C-H sties :2

(1'l,R)~1 t25mOI%) .~e Me J~O

AcOH (0 5 eqUI'I. ,'. 0

;t '32°

H"O" (~ OequlV ) Me HC3- alOne

$/oW dOtlron ... H (sse b6law}

~rsm t +).2-oxo-1iCla.J solldeSS, 4D%

( ),s~&l"iElOlicte 54 80% l' equrv )

Enzymatic Oxldatlon Cunfllnghamofla blakesleeana

D

1

(

7

:l!

(-)-6$ (1 equrv)

14 days

Of; . 'SftrcO$Ofccllvo Me.''''ytene H'ydrO.Kflallon

(8.5)- .. "I. (25 mcl%) AcOH (0 5, equhll

HAl5.0 'equllI' ) .s'Cw eddrltQl!

2A%rsm

Fig .• 4. Additive effects of steric, electronic, and stereoelectronic factors on the selectivity of secondary C-H bond oxidations of terpenoids. A slow addition protocol was used in which ca:talyst 1 and HZ02 oxidant were added via syringe pump OVer a i-hour period to detrea~e the rate of hi- or multimolecular catalyst decomposition relative to substrate oxidation (4, 23). (A) Electronic and steric deactivation, as well as stereoelectronic activati on via alleviation of 1,3-diaxial six-membered ring interactions, lead to predictable and highly selective oxidations by 1 on rnanooi-rterived diane {-)c49. (B) Cyclopropane-induced methylene activation (via hyperconjugative elec-

IOlal. C7 oxldalion = &2%

tron activation), as well as electronic deactivation, leads to selective oxidation of (-)-51. (C) Transformation of (-}-ambroxide (53) into (+)·55 via sequential, i ntermolecula r secondary C -H o.X:idation. Microbial biotransformation~ of {-)-53 and (+)-54 demonstrate site seLectivitie5 orthogonal to those of Fe(R, R-PDP) 1. For a complete list of all th e oxidation products formed in the microbiaL oxidations, see figs. 51 and 5.2 (23). (D) Energyminimized stn,lttur:e of (-)-dihydropteuromutilone (56). (E) Diastereoselective methylene hydroxylation of (-)·56; rsrn, % recovered unoxidized starting material.

www.sciencemag.org SCIENCE VOL 327 29 JANUARY 2010

569

I

REPORTS

isolated in 87% yield (entry 11). In more confermationally rigid substrates, ovcroxidatioo is not observed (see below). Benzylic sites arcalso activated toward selective oxidation, although this effuct is limited by the requiremcrrr for CWO substitution on the aromatic ring to prevent its: oxidation (compound til) (23).

Having established a framework of selectivity rules forca\1lIYSI i wilb relatively simple 11101- ccules, we sought to apply the system to complex natural products, In this respect, we sought to mimic the action of selective oxidative enzymes like the cytochrome P-450s. Terpecuids, the largest and most diverse class of natural products; are biosynthesized through' pathways that introduce oxygen functional lty independent of C-C bondforming events. Topologjeally diverse hydrocarbon skeletons are forged through a sequence of carbonyl condensations, reductions that remove postcondeusation oxygen functionality, and cyelizaiions (18). Oxidation patterns are subsequently installed via late-stage oxidative tailoring enzyme modifications (1), We hypotbesized that the individual electronic. steric, and stereoelectroruc effeots on selectivity in oxidations of simple compounds with catalyst I lsee above) would be heigbtened when. combined in complex molecule settings.

We first examined the oxidation of diane (-)49, a molecule derived IT('}lTI the diterpenoid (+)-manool. that comprises a substituted transdecalin C>QTC and contains a total of 28 C-H bonds, 14 of which are secondary and two tertiary O'ig. 4A). According to our selectivity rules, we hypotbesized that the two tcrtiruy sites should be electronically and stericslly deactivated toward oxidation: C9 is a to a ketone, and C5 ~s both in an axial orientation and adjacent to a g(1/11- dimethyl group. With regards to methylene OXldation. we recognized that rhe two ketones electronically deactivate the side chain lind the entire B nag, leaving the secondary C- H bonds of the A ring as the most likely sites for oxidation. The most distal site from the srerically bulky quaternary centers on the A ring is C2; moreover, repulsive l,3-di.axial interactions With two methyl subsritueats (C15 and C16j may be relieved by C2 C,H bond oxidation. Despite nine possible sites of oxidation, we observed oxidation at only theC2 and C3 sires. Consistent with om analysis, C2 is the major site of oxidation affording (-)-50 in 53')1" isolated yield (12% recovered starting material, Ism). This example highlights tile capacity for distinct selectivity factors to become mutually reinforcing ill a complex molecule setting, thereby producing predictable and highly selective oxidation of secondary C-H bonds,

Site selectivity through hypercoojugative activation iii demonstrated with the F6(PDP)catalyzed oxidation of the terpenoid derivative (-)-5'J, a compound with a sensitive carbogeoic framework composed of a cyclopropane- and cyclobutane-annulated tune-membered ring (Fig, 4H). We hypothesized that the cyclopropane, moiety would effectively override sterie effects

570

and activate the two <ldjaoont metbykne groups, with Cj being favored fut oxidation relative to C6 because of its remoteness from the ketone moiety. Upon exposure of (-)-51 to catalyst 1, il number of minor. unidentified oxidation products were observed; however, oxidation at C3 occurred preferentially to furnish ketone (-)-52 as the major product in "1\ preparatively usefu145% yield (5% rsm). This oxidation proceeds in higher yield and selectivity with CR. R}-I than {S.S)·I, whic h we rationalize as a better matching of catalyst geometry with substrate topology.

Selective, sequential oxidations (21) may be achieved with an electrophilic oxidant by engaging an electrou-activating group, in an otherwise electronically unbiased hydrocarbon skeleton, to direct the initial site of oxidation. Upon oxidation" the newly installed ketone, acting as an UWO, will inductively bias the molecule during further' oxidation. Th6 two-step oxidative traasfcrmation of (-)-ambroxide (53) into (+)-2-oxo-sdareolide (55) demonstrates the power of tbis strategy to effect higbly selective sequential oxidations with (R,R)-l (Fig. 4C). The first methylene oxidation exploits' theactivared C- H bonds on the fivc-mernberedeyclic ether (C ring) to achieve high site selectivity at one site among nine other possible sites of oxidlltiol1. Oxidation of the c-ethereal C-H bonds furnishes (+)-(3R)-sclarcolide (54) in 80% yield, even amid numerous other electronically and sterically accessible secondary and tertiary sites of oxidation. The newly installed lactone of (+)-54 now serves as an EWG to inductively deactivate the Band C lings, making the secondary C-H bonds on the A dog rnust favorable toward further oxidation, As noted above with an analogous Iruns-decalin structure [( - )-49J, C2 and C3 are t he only notable sites of oxidation with catalyst J!H102. Among these two sites, C2 is favored on the basis of destabilizing diaxial interactions and results in (+)r2coxo·sclar60lida (55) being the major oxidation product in 46% isolated yield (9% rsm), Oxidation of (-)-53 with catalyst I afforded very little yield of (+),55, even at higher catalyst and oxidant loadings. This observation sugg~S(s that rapid catalyst decomposition may playa role in preventing overoxidations with this mild oxidant (4). Through an interplay of electronic activation arid deactivation, opposite hemispheres of a complex molecule can be selectively oxidized through Into11:110- lecular axidarion steps with a single catalyst. Oxidations oiH-53 and (+}-54 with microbial enzymes, which are reported to furnish some of the oxidation products we observe witb catalyst 1, require substantially longer reaction times (3 to 14 days), provide {+)-54 from (~)-53 in only minimal yield «5%), ami demonstrate no observable funnati011 of (+)-55 from (+)-54 (Fig. 4C) (29, 30)_ These examples illustrare that site-selective oxidations can be pred.icmbfy achieved with catalyst 1. that are often complementary to those accessible via enzymatic oxidations.

PleumlTlutilin and its derivatives have attracted extensive anentioo from both the pbennaceutcal and academic communities because of then- poruot activities against drug-resistant Gram-positive bacteria and their unusual tricyclic ditcrpenoid strucnne. Despite potent antibaetenal activities. their hydrophobic nature limits their solubility and promotes rapid degradation by cytochrome P-450 at the C2 and C8 positions, resulting in inactive metabolites (31). On the basis of our findings with (-),53 and {+}-S4, We anticipated that catalyst (S.S)-l would selectively oxidize at an alternative site, providing a distinct ftl u cti onal group handle for further chemical modification. High levels of regio-, chemo-cand stereoseleetivity were observed in the oxidation of pleuromutilin derivative dihydrepleutomutilone [(-)"56)] with (.s:S)-1!H20l (Fig. 4,0 and E} 111e C!4 glycolic acid side chain, a prevalent site for chemical rnodification, remained intact during oxidation. This result illustrates the extraordinary chemoselectivity possible with this oxidant; even a primary alcohol OWl be electronically deactivated from oxidation "men a to a eatbonyL Blectron-whhdrawing carbonyl and ester moieties effectively deactivate the five- and eight-membered rings toward dectmphilic oxidation with (S~S)-1/H201.' rendering tbe six-membered ring as the mostly likely region for oxidation. Altbough the axial. tertiary C H bond ar C6 is electronically accessible, it is situated in the sterically congested cavity of the five- and six-membered fused ring system (Fig- 40). We hypothesized that the C7 methylene is the single most electron-rico and sterically accessible site fer oxidation by catalyst {~~0,-lfH202' Consistent with this analysis, we observed that 62% of the starling mateeial is oxidized specifically at C7 (24% rsm) among II possible sitescf oxidation, The major oxidation product is the result of diastereoselective hydroxylation at the more sterieally accessible equatorial seeondary C H bond, providing (-}57 as a single isomer in 42% isolated yield (l<'ig. 4E). In comparison, ketone at C7 [(-)-S8] is provided in only 20% yield. Although the C>H bond of a secondary alcohol is highly electronically activated toward oxidation, energy minimization of(-)-56 shows that this bond is sterieally deactivated by virtue of its: proximity to the eight-membered ring (Fig. 40).

A small-molecule reagent sensitiveto the local chemical en:vil'OD1lloot of the substrate has enabled selective methylene C H oxidations that are predictable using the fundamental concepts of electronics, sterios, and stereoeleorronies. We anticipate that these findings will contribute to redefining how C-H bonds 'are viewed in synthetic planning, future methods development, and trw exploration .of molecular diversity

ReferenCH andNotes

1. J. Clardy. C Wa15h. Narure 432, 829 (2004}.

2. P. R. Ortiz de Monlelldno, Ed .• Cjitlxhrl1me' PIl50:' Structure. Mf!I;fwnism and Biac~~mistry {Pliln!Jm, 'New York, e d. 3, 2005).

3. M. 5. (~en, M. C White. Science 31B. 78'~ (2m)]),

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

4 _ N A. V~r meu len, M. 5._ (hen. M ,c.. While, Tetrllhedron 65. 30m (2009).

5. R, Mello, 'M. Fioreotioo, C. Fusco, R. Curci, j. Am_ -Chem. Soc 111,6749 (1989).

D_ D- H. R- a~rtDn, E .C5uh~i, ill. Oi:b~lik, rf'/t~~edton 4(;, ~743 (1990).

7. (, Kim, K, I. Chen, j. Kim, L Que ]r,,). Am_ [hem. 5Qc. 119, 596/1 (1997)_

3_ L okuna. 5_ lto, S. Ohba, Y _ r.lj)htdil" 1- (hem, Soc.

Dalton Trans. 19'17, 3547 (19<r7).

9. E xo se I.ee ti~ e OX-i do ti ons of oorbomyl d ~ ,i va Uves hi! lie been obs erved under fen ton cOl1ditiOn5 (iron 5~ It> Wi til H~OI) (J2)

10. J. P. Collman, H. Tanaka, R. T. Hernbre, i. I. Broumao, ], Am. (h~m_ 5e,_ 11Z, )6B9 (1990)_

ll_ A rare example of regioseledlv~, unactfllatl'd m€thyleM O~ ida ti 0 n 1m be~n reported. U>in 9 a ~ iron pm phy~ n t~t.1ly5~ hexane o~id.tjoo using excess substrate gav~ 2·hexaMI and ,·hExano[ i" 13 ;md 7% yields, r,"!p"(tilj~ly, rn>ed. on Ol\id.nl (n)..

1 Z'. M. S. Ch en, M, C, Whitt;.]. Am. (hem. Soc. 126, 13 46 (2004).

13. K. J. Ftaunhotl~r, D. A, Bo(hovchin, M. C. Whitt, Ol'{j_ Lett 7, 2 2~ (1'0051

14. L M. Stal1g, M. C. Whit~, Nat. (h~m.l, 547 (2009).

15. M_ A. Umb'~it, K, B. Sharpless, 1- Am. them. Soc. 99, 5526 (19771

16_ s_ D~s, c.. D_ l\lcaJ\liID, R H. C(abt"'~, G_ W B'lIdvig, Sdel1re 312, 1941 (2006).

J 7 _ 1- Yil~g, B .. Gabriele, 5_ Bel~eder~} y, HIJang; R BI~sl!lw, ). Orf), (hem. 67, 50507 (2002).

ts. L, V. Oesai, K L, Hull. M_ 5_ Sanford, 1- Am_ (hllm_ Soc, U6, 9542 (2004).

19 _ K_ C~e!i, P_ S_ Bar:.n. Nafure 459., BM (~009.).

20. B-. H. Brod.ky, J. O~ Boil, ). Am. Chero. Soc Ill, 15391 (200S)_

21. R. Mo_,·ealleste, .L. Que Jr., Stieno: 31Z. 18.S5 (2.006). 22_ Fqr the ben~fida( "'I!,,(~ of AcOH on nonhems

iron- rata lyz€d o~jdatiom (blefi G epoxi do Ii on), see tN·).

23. M alert ~ Is ~"d meth a ds ore avo ila ble a s s upportm g material at 5den[~ Online.

24. Eo l, ~(iet, 5. H. S~~raell'r, T, J- 6'et1. r. J- Silos, 1. C. Rkh~r, j. Am. (h~rn. Soc. 1i8, j 327 (1966).

25. G. A Qlah, P. Reddy, G_ K. 5, Pra~ash, (/Iem. Rev. 92, 69 (1992).

26, P_ A_ W~nder, M. K_ Hilin)~i, A. V~ W_ Mo)IWeg, Oro- lett.

7, 79 (ZOOS).

27 _ j, S. Lee, R, L_ FudlS, Org. WI. 5., 2247 (2003).

211. E. M. Dav15, R. Croteau, fop. C[lrr. Chern. ~09, 53 (lilO!)). 29 _ A. Fa/ooq, S_ T~hat~, Z /'JaWr!or'Scll, C 55, 341 (2000)-

30. A. Ala, L, j. Conti, 1. ee tt e~ dg~, I. Qrhan" B. S ener, Chern_ Phattn. Bull_ (Tekyo) 5. S, 11S (2007).

31. R. L HM5im~t 01 .. , Org. Process Ref. De». 6, 482 (1'M2)_

12. j, T. Groves, T E_ Nerno, R, S. M~r.;, 1- Am. Chem_ Soc. 101,~2 90 (1979).

33. M, H. Lim, Y_ j, lee, Y_ M, GDh, W, Nam, C Kin;, Bull. Chem. 50(. lpn. 'n. 107. (199,9).

REPORTS

34. M _ (_ Wh ite, A. G. DoylE, Eo ill. [a cobsen, ], An], (_hem Soc ~B, 7lWl (2001).

35_ Dedk~ted to Professor Eric N_ lacobSEn on hi, SOth birthday ior his insplration"l work on lelectlve o;:id~"OOl, '(II" gr;'\ef\J~y a(~nDwI"dg" R. ], P~kula for perfo(fTlin9

9 a I chrorn ate gra pay experi men 15 tOWol rd eluddati n,g the sterie ~ffect5 on Qxidation ~lte s~lectMty; M. A. Bigi,

D_ J. Covell, and E. M, Slang lor helpful dildmions and (hecij~g our IpeC\fO'l;!lp ie d;lta; an d 5_ A_ Reed 10' tielpfu I di~ru~sionl an d dn ~dI:i~g our e~pe~ menIal procedure.

M5. C. "'0' a Harv~rd Unive,si II' 9 ra I'lu ~te stud ent who (ompleted hi, dlldmal work witn M .:C W. a t In e Unlven;i!y of Iilinai) a! Urb"f);I-Chamf"';!i~- Tm) \'/(lrk WoS suhrnitted til M.s.t:l rorrmittee as part of hi! thesis on 111 August 2009 " nd was presented by M. C. W. at Ih~ Welch 5ympo,iu m on 27 Octuher 2009. We are grateful 10 the A_ P. Sloan FOOnd~tilln, the Cmrille and Henry me,yNs H:rund:ltilm, Bo.lo! ·Myel' Squ ibb, Plilfr,lIb bon, and the Univer>ity of Illinois for fln~nciall(Jpport. 'MSC is a .2008, B,istol·Myer. Squibb Gradu~te fellow in Synthetic: organic: Chemislly.

A U 5. patent on ',e\i>ctive AI i phil~ [ (-H Ol<idatjon" j) pending (app lkation WZ 45, De 6).

Supporting Online Material

www.$dencemag .org/cg ilcMten t/fu IV3 2715 96 5156 610C 1 Ma tertals ;1 n d Meth 0 ds

Refe"'nee.s

20 October 2009; an:epted 2S Novemb~r 2009 1 O.l1l61l"d ence .11B36a2

A Basal Alvarezsauroid Theropod from the Early Late Jurassic of

Xinjiang, China

Jonah N, Choini'ere,l"'" Xing Xu/ James M. Clark,l Catherine A. Forster,1 Yu GUQ,l Fenglu Han;!.

Th~fo5sil record of Jura~~ic theropod dinosaurs clos~ly related to birds remain~ poor. A neW theropod fro m the earlie~t Late J UI asstc of western China represents the ",arliest divergi nq member of the enigmatic theropod group Alvi!relSaUroidea and confirms that this g,roup is a basal member of Manlraptora, the dade containing birds and their closest theropod relatives, It extends the fossil record of Alva rezsa uroidea by 63 million years a nd provides evidence for marriraptorans earlier in the t05511 record than ArchaeDpteryx, The new taxon confirms extreme morphological convergence between birds and derived alvarezsauroids and illuminates incipient stages Qf the highly modified ;:il\lijrezs<iurid forelimb.

T be presence of fhe basal evialan (1 J) Archagopttryx- itl ihe latest Late Jurass,ic (Tithonian) and the poor fOs,,,n representation of more basal msniraptoran taxa in centemporaneous or slightly older deposits indicate either a gnp in the stratigraphie record or, mote controversially, that birds are not relined to theropods (4). Recent discoveries of Middle-Late Jurassic manlraptorans (5~7) from China are startin g to filli n the temporal gap, but me a."oes ofthese new taxa are poorly resolved (8, 9), and they do not clarity basal maniraptoran diversification because very little character evidence separates them fiom birds (5, 7).

1 Department of Biolo!lirnl Sdences, The Grorge Washington Unive,sity, W.Jshinglon, DC 20052. l)SA. 'Key l,.Jboraloo::y of Evo\Ubonary Syst~mati(.l 01 VeMbrate~, InstilUte of Verte-b,aw P<lleontology' and P.alellilnlhropology, S'eiFng 100044, Chillo.

'To whom correspondence should be addressed. E-mail: jonah,(hoinier~@gm~il.com

Here, we describe a three-dimensionally PIer served, nearly complete skeleton of an alvarezsaureid Ihero_]Xld, [{aplocilli'lhlS sellers, gen, et spec. nov, (10), from orange mudstone beds in the upper part of the Shishugou Formation In Ww:aiwan area, Junggar Basin, Xinjiang, China. Radiometric dating constrains the age of this fossil to between 158.7 ± 0.3 and 161.2 ± Q.2 Ma (million years ago) (11), correepondtng to the Oxfordiao marine smge in the early late Jurassic. (i2} The dates. for Haplocheirus reduce the conflict between the fossil record and phylogenetic hypotheses thor early manirapreran diversification took pia ce in the Jurassic. Alvarezsaurords are known BUill South America (13- 15), Asia (16 19), Nmth Arnerica (20, 11), and Europe (22, 23). The basal phylogenetic positien and early temporal position of Haploclteirus imply that Alvarezsauroidea originated -in Asia rather th;m South America (14, 10), Derived 1ht,-'mhcrs of tbe Alvarctsaur-

oidea were originally thought to be flightless basal avialans U 8) because they share many morphological charactsristics wiih birds, including a loosely sutured skull. a keeled sternum, fused wrist elements, and a posteriorly directed pubis, Haplochsirus preserves plesiomorphie morphological characteristics that confirm a basal position for Alvarezsauroid ea Viii thin Maniraptura (24), demonstrating that tbese features of derived alvarezsaiaoids represent dramatic convergenees with birds.

rvpp (Institute of Vertebrate Paleontology and Paleoanthropnlogy) V 1598E (Figs> 1 and ':! and figs, S4 to 89) is -140 em in total body lengthas preserved but is missing the end of the tail (estimated total length 190 to 230 em; table S I}. The bones of the braincase are coosslfied and thencuroccntral sutures ate visible, suggesting that the animal is. Likely II young adult or latestage subadult,

The gracile.low skull (Fig, 2Aand fly. 54 to 57) is well preserved in three dimensions, The narrow. elongate.rostrum becomes taller-and wider just anterior to the large, ,mter-ohrtemlly facing orbits. The dorsoventrally thin jugal is triradiate, unlike the rodlike jUgll! of more derived alvarezsaurolds (17). Tn lateral view. the basisphenoid is oriented at45Q 10 hOlimntzrl. a condition presentm some aeodonnds (Z;i) and in. alvarezsauroids ( f 7). The basipterygoid ptocessesere long and project vemrolaterally, a morphology known only in primitive birds aile! alvarezseuroids among Thoropcda (17).

At least 30 small maxillary teeth are present in Haplocketrus. as in Shuvuuia (J'7). Pelecanimimu« (16), therizinosauroids (17), and treodontids (28), Unlike tile conical, unserrated teeth of Mononylcrls (18) and ShUliuliia (29), however, the maxillruy teeth of llaplndJeklls are

www'.scieneemag.org SCI EN CE va L 327 29 JAN UARY 2010

571

REPORTS

Fig. 1. Haplocheirus A

sollers (lvpp V1598B). (A) Photograph of holetype in dorsal View. (8) Une drawing of holotype in dorsal view. Dashed line indicates crocodyli-

form fossil underlying B

cervical vertebrae. (0 Reconstruction. G rilY fi II i neli ca tes portio ns of the 5 keleto n that are not preserved. Abbreviatio ns: cav, caudal vertebrae; ev, cervical vertebr ill"; gs, qastra lia; i I, left i lium; If, left femur; lh, left humerus; lrn, left manus; p, pubis; rf; nghtfemLlr; rh, right humerus; rm, right manus; rmt, right rnetata rsus; 'Pi ri ght pes; rs, right scapula; rtf, right tibia and fibula; sk sku l~ Sol, sacra l vertebrae; tr, tho ra DC ri bs; tv,

tho raci c vertebrae. Sca le ba r: 25 cm.

recurved and bear small serrations p osteriorly (fi g. S4). Haplocheims shows marked hererudonty in both the maxilla and the dentary, with the maxillary teeth diminishing in size posteriorly (fig- S5) and the lower jaw bearing large, su bconical, unsenated anterior teeth and S111a.ll er, recurved, serrated posterior teeth (f g. S4).

TIle cervical centra lack the strong opisdrocoelous condition of Sbuvuui« and MDflonYh1JS (29)The five sacral vertebrae all lack the ventral keel present in other alvarezsauroids (29). The eeotrum of sa cral five is am phiplatyan , whereas in M(JIJolrykus and SllIIvuuia (29) the last sacral centrum bears a convex posterior surface for articulation with the procoelous first caudal vertebra. Similar to other alvarezsauroids, the mid-caudal vertebrae have short and anterodorsally oriented prezygapopbyses, whereas in 1110St coelurosaurs the prezygapophyses of the midcaudal vertebrae are long and horizontally oriented" A sternum was not found with the specimen.

The ovate coracoid bas along posrerovemral precess and bears a well-developed biceps tuberde, which is lost in other alvarezsauroids, Similar 10 the basal alvarezsauroid H11Q.gotrylrlls (30), the hypertrophied rectangular internal tuberosity of the humerus is separated from the humeral head by a distinct notch, and the humeral ectepicondyle is large and bemispherical distally. The incomplete ulna bears a well-developed olecranon process, as do all alvarezsauroids. Proximally, the radius articul ares with a lateral concavity on the ulna.

Unlike the avianlike oarpornetacarpus (29) of more derived alvarezsauroids, the distal carpa Is are unfused, Metacarpal II (the medial metacarpal) is 1.9 tunes the width of metacarpal ill (Fig, 3), wider than in most maniraptorans but narrower

512

than in MonDnyku,~ and Shuvuliia {29j. Metacarpal fV is only half the length of metacarpal ill, a condition unknown in other theropod dino~, The middle digit is the longest in the manus, unlike in Shuvuuia, where the second and third digits are both much sh,Olter than the first digit (3J). Manual phalanx [[-J bears a deep axial furrow ventrally, as in all alvarezsauroids (14). The- manual unguals are curved and bear distally placed flexor tubercles, Milke the unusual unguals of derived alvarezsaurcids (3 f), which are flat and lack flexor tubercles.

The pubis projects anterovemrally (figs. S8 and S9), which is the plesiomnrpbic c~ndilion fur msniraptorans, in contrast with tbe vertical pubes of Paragop~\lkl.ls (30) and highly retroverted pubes of derived alvarezsauroids (29). TIle curved femur bears a plesiomorphic alariform lesser trochanter, consistent with the femoral morphology of Alvarezsaurus (14) and Po!agOl1ykllS, (30). The lareral distal condyle of me femur is conical (fig. 59) and projects further distally than the medial condyle, a morphology known oilly in some trcodonrids and alvarezsauroids (14). There is no developmeot of an avianlike medial cnemial crest, as .in MOl1onykus and Shuvuuia \26). The third metatarsal is fully exposed in anterior view and is a component of Ih~ ankle joint, as in the basal alvarezsauroids PtJlagollykus and Alvarezsaurus (13,3"0).

A phylogenetic analysis of 99 theropod taxa shows tbatJlaplocheirus is the basalmostmember of the Alvarezsauroidea [Fig. 3 and fig. 51; see Supporting Online Material (SOM) for full details], Haplocheirus possesses unambiguous synapomorphies that support the tnonophyly of the Alvarezsauroidea, including long basiprerygoid processes, 1:1 vertically inclined basisphenoid,

Fig. 2. HapiocheitU5 sollers (lVPP V15988) skull and forelimb elements. (A) Skull in left lateral view. (8) Right scapula in lateral view. (Cj Right coracoid in lateral view. (0) Right humerus in latera l view. (~) Left di sta I h u me rus in flexo r vi ew. (f) Left distal humerus in distal view. (G) Right ulna in medial view. (H) (eft manus in lateraL view, with digit II separated from manus. Abbreviations: ar, acromion process; 'If, axial furrow; aof, antetbital fenestra; <if, (oramid foramen; d, coracoid tuberosity; dpc, deuopectoral crest; eec, ectepicondyle of humerus) emf, external mandibular fenestrae; en, external naris; ft, 'flexor tubercle; g, glenoid; hh, humeral head; it, internal tubercsity; mell to mdV, metacarpals II ro IV; mf, maxillary fenestra; 0, orbitj op,olecranon process of the ulna: rap, retroarticular process; sic, semilunate carpal. ScaLe bar: 4 ern.

a b ypertropbiedintemal tuberosity of the hUlIlBl.1JS proximally level with the humeral bead, a large humeral ectepicondyle, a pronounced axial fwTOW on the flexor surface of manual phalanx IT-I, and a conical lateral distal condyle of the femur.

Although a basal position of alvarezsauroids within Maniraptora has been proposed. (24,32-34), the global. phylogeuetic placement of this group within Theropoda has been untested by the recent description of new alvarezsauroid material tram Canada (2l1), A-.;gentioa (15), and MongOlia (J 9). Early analyses placed alvarezsauroids as either sister to, or nested within, Avialae (J8, 29), Alvarezsauroidea bas also been hypothesized as

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

Jurassic

REPORTS

Fig,. a.. Phylogenelic relir liouships of Haplochein.JS so/lets and hand comparisons ,of se~ theropod:;. (A) Simp! ified, temporally calibrated dadogram of the strict conse-nsU5 of 69 most pa rsirno nio us trees prod u ced fr1 this' a fl<llysis (see a lso fig, 51), Thick lines with rounded end caps i ndi ra te straligraphic ran ge of known taxa, Thin black lines with square ends indicate phylcqenef c relationship. Dashed limos represt'nt ghost lineages implied by the stratigra phic dis< tribution of fossils with respect to the phyloqenetic I'l'[ationshi ps shown here (not.e the exceptionally long ghost lineage for AlVa relsau(oid~a). Arabic numerals indicate ages in millions of yea rs before present. Boxes represent named clades. (8 to E) Reconstructed theropod mani scaled to digit two length. (B) Allosaurus tragi/is after Madsen (40). (C) Hapiocileirus soliers (lVPP V1598S). (D) Shuvuui(1deserti after Suzuki et al. (31). (E) Deinonychus a ntirrhopus a fte r astra m (41). M etilta rpa l ele me nts a re shad e £1, dashed Ii n e5

Ornitholesle.s

r------

Cretaceous

Outgroups

c

AfvarezsaurU5 - Achi/lesBurl)5

_~agonykus Pervicutso:

Shuv.u.uia

Parvicursorinae . _1'v!.0~O~tk~5_

Alberlonykus

. . Th_ert:!n.5''!a.u .... ro~id .. e.a - •• -.- - - _

___ ~ . O\liraptorosauri~ . •

- -. - - __ Archaeopteryx

~-.!!~~ -. ~Qs2rloEt~rygld!~+ _O!nitl:!.o~h~~c!>!_

Avialae TroodontJdae

-,_----, --,-----------_,

Drornaeosau rldae

------------

Compsognalhfdae

r---'- - -- - - -. - - -- - -,-

. ... ~ - HapJocheirus sollers an I ra P.lo ra

Alvarezsau ridae

Paraves

me .s($[.el" taxon to Ornithomimosauria (33), Our phylogenetic analysis agrees with still other research t2, 14) in pla.cillg Alvarezsauroidea as a relatively basal marriraptoran group. OUf most parsimonious trees do nat support either avialan or ernitbernirnosaurianaffiuities for the Alvarezsanmidea, although the omitbomimosaur bypothesis is nearly as. well supported, 111e shortest tree found wllen constrainmg the Alvarezsauroidea to be sister to or nested Within Avialae is seven steps longer than the optimal tree. The miuimum length of trees produced when COIlstraining Alvarezsauroidea to be sister to or nested within Omithomimosauria is one step longer than the optima! tree (sec SOM for phylogenetic details).

The Middle to Late JUrasSLU (BarhonianKimmeridgian) age (8, 9) ofnewly described maniraptorans from China (6, 7) implies mat the Alvarezsaurotdea have a minimum ghost Lineage (3-'1 of -63 million years (Fig, 3)~ The pI\."SCIJCe of Naplochei1"l.ls in the early Late Jurassic of China confirms the prediction made from this ghost lineage that alvarezsauroids were present at trus time. To explore the effect of l-Jap/m:heirlls on congruence between srr3i:(grnphk. infonnarion and phylogenetic hypotheses for the Coelurosauria (36--39), we compared the gaps in Ole fussil record implied before and after inclusion of l1aplocheinl8 (37,39). The addition of me earliest Late Jurassic lfap/ochei.ms improves the stratigraphic fit of the mauiraptoran fossil record to the topologies recovered in our analysis. by au average of 140/"" from a mean fit of 0.41 to 0.48 (fig. S3).

Hapioci1eirlls is the largest alvarezsauroi.d kilOW!; fj'om l'Olnplcle rrlaterial (see SOM), arid its

~.

in (0) indicate inferred digital elements. Abbrevfations: I! to IV, manual digits I! to IV; Aa, Aalenian; Baj, Bajocian; Bat, Bathonian; Cal, Callovian; Ox, Oxfordian;.K;, Kimmeridgian; Ii, Tithonian; Be, Berriasian; Va, Valanginian; Ha, Hauterivian; Bar, Barremian; Ap, Aptiall; A~ Albian; Ce, Cenomilnian; Tu, Turonian; Co, Coniacian; Sa, Santonian; Cam, Campanian; Ma, Maastrichtian. Scale bar. 5 em (B), 4 cm (C and El, 05 em (OJ.

basal pbylogenetic position suggests. a patrern of miniaturization for the Alvarezsauroides, relatively rare in dinosaurs but collvergently evolved in Paraves (2). Derived alvarezsaurotds have a simplified, homogeneous dentition. 00 uverge IJt with that of some extant insectivorous mammals (20), but fIaplochllirus basrecurved, serrated tooth aad caninifonns that suggesr camivory W<lS the primitive condition fur the clade. The presence in HaplocheiJ"l;,I,~ of only slIghtly reduced second and third manual digits and curved unguals with flexor tubercles on these digits implies that the hand was fully functional and Haplocheirus retained some grnsping ability, unlike the presumably limited function of the greatly reduced lateral manual elements of Mommykus and Shuvuuia (3l). 111e mediolaterally nj1JTOW Mom (meescarpal rhree) and the greatly shortened and slender MeN suggest that the eXl;ensiv{} digital reduction and fusion seen in derived alvarezsauroids was already under way by the earliest Late Jurassic, proceeded fum lateral to media; on the manus and, sUl'prisingly, initially involved reduction in length of only M.clY.

References and Notes

1. We' use the definition of Mialae from (2), the least indU;lve d.de wntilioi"g, Archaeopteryx Ift!iQgraphica .nd trown group birds. (other than the d~nni'tion hom (}j. at( \il x~ ;ha ~ ng a more fe, en t mmm on ~"' 'O, tor wi to Arc;N(](,oplezyx than with rroodon jJrJf!agUI.

2. A. H. Turner, D. Po~ 1. fl.., Clarke, G M. Erick,o~, M, A, N"I~U, Sr:ience 317, 1378 (2007).

3. F. Z~ang. Z.Zho~. X: X~. X. Wang. C. 5ultiv~~, NrJltJre 455. 1105 (iOoe),

4. fl.. feduGcia, Auk 11'1, 1187 (20Q2),

s. F. Z~ang. Z. lho~, X. X~. x. Wang, Narunvi5ten!;t;Mffe~ 89, 394 (2002).

6. X. Xu, F. Zhang, Na/urwimnsr:haflw 91, 173 CWOS).

7. o, Hu, l. Hou, L Zhang, X. Xu. Nature 461, MO (20M), B.. Y. Liu, Y. Lw. s, 1;' z, y oMg, Chill, Sci. i!ulL 51. 2634 (2Q06).

9. K.)(U et al., jurassic System in Ihe Nort/) of China (VII): rJ'~ stratigraphic Regia!! Qf No rfhe~st Chira {Petroleum In dus try Pr ~S5, Beiji~ g, i OO} ).

to. Sy<;temati( pilleomolo9Y: rnerepoda Mal'5h. HllIl:

Coelu,os~urio von HUEne. 1914; Alv"~'ElSaUroldea Bon"p"ftot 199~; HaplQcMitus soli~rs, gen.l't spec, ~QV. Etyni ology, H~plorh~iw" latin lzsd from ~apjo(h~ r (Greek), "simple hand," and referring to the lack af 111e i IHl d a Ii led "' ~n us oi derived • Iva rem UlOid,; soltet«, I"~~ lor "skiIlM: ["ll'rr;09 to the presumed ability 01

I h i5 ta xon to perform di gila I a( n ons th at wauld be impos,ible for dEfived alvarezsiluridl. Holot)ljl~: IVPP V1S93!i is a neatly wmplele skeleton, misling only the tti) till <;a ud~l ve rte bra" a nd the d orsat asp e(I., of Ihe ilia (Fi 9'. 1 and 2 a!1 d fIg~. 54 to S9). Lorn li ty an d horizon; lWI1.GG;H B~si~, Xinjla>1jJ. China. lower orange mudstone b ed, in th~ upp er part of 111 e Sni!~ ugou Formalion (Oxfurdi a n). OJ,, g~GSk La rg e "tva r elSll uroid Wi lh tn e tollowing autspcmcrphtes; aWl'550ry mandibular tenesna "nl~ rod or," I to the rna od ib ular len es tra, f€ nestra e

divi ded by posterior p recess of the de Ill. ry; In etara rpa L III one-h~lf the t~~gth of metacarpal II. Dlflers from all other alvarEzsauroid, ill the fo[lowin,g deMved f~oturf" he terod on I den!" ry tooth row wi then la tged den tilfY tooth Il.; .alveolar .matgin ·af a~lerior end oi dent.ry is dorsa~1' mnv&; rliaxillazy "no dent:"y ~I'th with

peste nor serrations, Pie >i01]10 rp hie ; n the Ahra r~LlitUrQidea for the foUowiog characters: postorbit;tl contacts the jugal to form <t postorbital bar; recurved mil~illi!ry and den ta,!}, teeth Wi In serra t;o~); di~ 1;0 l co 'I" Is not fused to metacarpals; manual digit (II longer than digit II; manual dig,t> III and IV bear returvad unguilis; pubis prQie(ts anlerovenlrally; ventral surface ot sacrum without keel; pnslerinr su rfa te of I", t ,aci a 1 ,I' n trum flat

11.. I. M, Oarl< d Ill" in 9th tmernotiono! Symposium on Mesozoic rermslrial Ecosyslems and Biota, P. M. Barrett. S, E. Evans, Eds. (M;mf~€ster, 1)1(, 20(6), pp. n-28",

t z, F. M. Ciradlt€in. J. G. Ogg. A. G. Smith, Ed!" 11 Geologic lime Scale 20M ((ambridge Unl'''" Press, C~mbrfdge. 2()04)

1'1. I. F. BOfl~p" rte, R""i>ta' del Mus"" Ar!l~rhllO de (j'eri{i~!l Naturale!; Beriwrr;fina lIivadollio.' Pa/eonta/ogia 4. 17 (1991).

www.scieneemag.org SCI EN CE va L 327 29 JAN UARY 2010

573

I

REPORTS

14_ f_ E. Novas, Melli. QU~n!ir MU5 39, 675 (J996)_ 15'. A. G, Martinelli, E. I. Vera, ZooJoxo 1582, 1 (zoOl).

16. A_ A_ K;1rhu, A. 5_ Rautlan, PlJleO/1r~l, J- 30, 58] (1996)_

17. l. M. Chiappe, M. Nore(l, j. M. Clark. NaturE! 392, 275 (1~98)

IS. Pfrle A" M. A. Norell, L. M. Chiappe, I. M. Clark, Nature ~6l, 62J 11993").

19. A. H. Turner, 5. Nelbi!t~ M. A. Nor~ll, Am. M[I';. Novil- 364f1, 1 (2009)

:20. N. R. Longrkh, r. I. OJ rrie, C(erae. Re5. ~ 0, 239 (2009)

21. J. R. Hutshinsnn, l. M_ Chiappe,). venebr. i>o/roflfoL 18, 447 (1~8).

2 Z. D. Naish, G. I. Dyke, Ne/Je5 jalirb. Geal. Falaeom ol.

Montalihejle 7, 385 ('2004).

23. E. Kelll~r, D. tin gOTSS'(U, Z, (sUd, Arf a Poloeon/oL Rliliianlae 5', 349 (200§)

l4. f. E, Novos, D. Pol, in M~5-ozoi( Birds: Above /he Heads of Dillo5aurs, L. 'M, (hiappe, L Witmer, Eels. (Vnlv. of

Ca lilornia Pre", B~rkel~y, 200 2J, P p_ U 1-12 8,

l5. P. I. M;, k\lVi~ky, M, A _ Norell. in The D ino'iQlIrio, D. B. Weilhampel, P. Oodson, H. OSlT1olska. Ed!:. (Vniv. 01 Ca Ii lorn ta Press, B~rkf:ley, ed, 2, 20(4), pp, 184--195.

.20_ B. P. PeIW-MoreM t!t ol., NatUre 370, 363 (1994)_

27. J. M_ Clilrk, T_ Moryailska, R. Barsbuld, ill Tile Dil1osaurfo, D. B. Wtfsha mp el, P. Dodson, H _ 01 m otska, Ed,. (Vniv _ ot (alUmni;, Preis, Berk€ley, ed, "2, 2(04), pp, lS1-1M

28, A. p~rle, L M_ Chi~ppe, R, Barshold, l M" Clar~, M, Norell, Am. Mm; Novil_ 3105. 1 (1994).

29. L, M, Chiilppe, M. A. Norell, 1, M. C(;uk, in MesIJlOic Birds~ Above rhe Head. of Di1)D501lr5, L. M. (hl.pp~ l. M.

Wi ~mer, Ed". (U~i\I of Cllilml1 i~ P fIl,), a~rl:el~y, ~ 00 2), pp. 87-120.

31). F_ E, NO'ias.J veneo« PaleolltoL 17, 137 119971. 31. 5. 5u1lJ~i e/ 01., Coo/rib. Sci. 494, 112002).

3~. T, R. Holll Jr-,)c Veltebr. P(J./eontol 14, 480 (1994).

31. P. 5ertno, in New Persper:live, on me Origin and' Early [voilltion of Bird.l, 1- Gauthier, L F. G~Il, Eds. (Pe~bQdy Muwum of NoMal Hi>lory. Neew Haven. cr, 20(1),

.PP - 6'9-9a-

14. j, M. Clark, M. A. Norell, P. I. Mako\'i,ky. in M~sozoi[ Bird,_- AbGve tire Heads aj Dinosaurs, L M. Chiappe.

6. M. Witmer, Eds. (Vnw.· 01 Cililfornia Prm, Bl"r~l'ley, 2(02), pp, 31-64.

35. M, A. Nor"l!. Am. i SCi. 2;93A, 407 U 993)

36. We rneasu r~d (ong rue no~ of SIr atig rsp fly with phylog€ny usiog !hI! rnodified Milnhattan 5tratigrophk M~osW.re {M5 M') U 7) wi th the phylogene~ ~ 50ftwa re package TNt tJ 8) .md pub Ii sh ed IC~ p [5 that allow for un tertai n ty in first appearance dates tor fo,\siltaM (39).

37. D. PoL. M, Norell, r;fndisMi 17, 285 (200.1)_

3::1. INt version 1.1; P. A. GoloboH', S. rmil, K. C Ni~on, (onsejo N ~ci OM I d~ I nvestiq a ,1 one) Cientifl cas Y tik~ka.! Miguel LlLlo. Iucurnan, Argen~nil, 2003.

39_ D, Pol. M. A. iIIore~, Sy,r_ pilil. 5S, 512 (2006).

40. 1- H.l M~d'e~, Utah Gro~ Minn SurV. Bull. 109c I (1976).

41. I. I:L Ostrom, Peabody iVJ!IS. Nat. Hist_ Bu/f_ 30 (1969)~2. We thank W. Ding a~d' T. Yu ior findih9 the specimen. an;:! the ,few of the 2004 Sjno--A!)ien(an ~eld expedition fa r e scavati n_g the sperimen: H .. ' j. Wan g, X. Di ng, and

t. Xiong for prep;Jrfng the 10,;1(; A. Bumlioni,

c. MeMi "9, and M. iIIorell j or sp ~ci ~1 e naccess: il nd

P _ Barrett, H.. [ameton, J< Conrad, M, Hlison, D. Hone, C. sulliVan, and A. Turner for dlsrussions. Support for this research was prav i dad by the iIIationill 5 dena! Fa undati on Division of I1iIrih Sciences and Off[cli! of lotem.tianal

Sri enee <\nd Engine."~ 1\9 "I (he USA, tli~ tMimial No tor 0 I Sci ence~ ~oundation of Chi nil, and the Chi nese AColdemy of S,I enees; and th~ C;WI) F~dliUlting Fu nd, the MaSSI,

FOil ndation, the Hilm or S~llee beq ~51, a nd [he George Wdshington Univer;1ty.

Supporting Online Material

www_<cieneerMg . Drg/egllmnt €fl tffu U/3 2 7/5 9 6515 71/flC 1 Ma teriats ;3 Q d Me In a d,

SOM Te~t

Fig,. Slill ~

T~bla< Sl and S2

R~f~renCe5

18 September 2:00.9; accepted 1 December 2002 10 1126(stienee.1182143

Anc.iently Asexual Bdelloid Rotifers Escape Lethal Fungal Parasites

by Drying Up and Blowing Away

Christopher G. Wi!son* and Paul W. Sherman

Asexuality has major theoretical advantages over sexual reproduction. An important evolutionary puzzle, therefore. is Whyex:clusiv!!ly asexual metazoan lineages rarely endure. The Red Queen hY[:lothesis posits that asexuality is rapidly extinguished by relentlessly cotl\lolving parasites and pathogens. If 50 •. any long-lasting asexual line<!g.e must have unusual alternative rneehanisms to deal with these biotk enemies. Bdelloid rotifers ari:' freshwater invertebrates that abandoned sexuil\ reproduction millions of years ago. Here, we show that (ultur~d populations of bdelloids can rid themselves. of a deadly fungal parasilI' tl)mugn complete desiccation (anhydrobiosis) and disperse by wind to establish new populations in its absence. In Red Queen models, spatiotemporal escape can decouple and pralI'ct asexuals from (oevolving enemies. Thus. our results may help to explain the persistence of the anciently asexual. Bdelloidea.

S eXlJ1l1 reprod. uction reduces .the effiCien. C¥. of gene trensrnesion by up to 50%, dis- 11JplE favorable gene combinations, spreads disease, and is energetically expensive (1) Yet, paradoxically, sex is nearly ubiquitous: Obligate asexuality occurs in less than I% of ani" mal sp ec ies (1, 2), and IlE s ca fie-red distri bu tion at the tips of phylogenetic trees implies that abandoning sex condemns a clade to extinction be-fore it can radiate sufficiently to achieve high tax 0 norni c rank.

The class Bdelloidea (phylum Rotifera) is a famous exception, During three cennnies of observation, more than 450 species of these tiny freshwater invertebrates have been described, but

Departmwtof Neurobiology and Behavior, (omell Univ€l>ity, I thaea, NY 14850, USA.

"To. Whom (Orr espon den re ~h nul d be add ressed. E· mail; cgw8@wmell.edu

574

neither males nor mcietic eggs have ever been recorded (3). Molecular evidence supports the inference that bdelloid retifers have been obll_gately asexual for tens ofmilJions of years (2-5).

Hypotheses for sexual reproduction suggest that it is maintained because it IOC.mOVes deleterious mutations, facilitates coevolution witb parasites and pathogens, or bath (2, j. 8). However, mutational hypotheses have been challenged empirically U. 5, 6, 9), ~nd muranoeal load may be less problematic than predicted, because the bdelloids have persisted despite acournulating mutations faster than related sexual clades (4). ill contrast, the coevohmon (Red Queen) bypothesis has received considerable empirical lind theoretical support (1, 5 11).

Under Red Queen models, biotic interactions. favor sex by relentlessly imposing fluctnating, time-lagged, frequency-dependent selection (6 10.1. However, asexuality can be maintained

inone special case: when vulnerable hosts can temporarily shed locally coadapted parasites and pathogens and dispense without them to unmfecred habitats (In Migration lind clonal diversi ty at the POP]) lati on I eve I can th <en 51.) bstirute for recombination and genetic diversity at the individual level. allowing mobile hosts to avoid the costs of sex while continuing to "outrun" their enemies (1 I, 11). Dispersalfurther favors asexuality by reducing intergenerational transmission of infections (J 3).

Three unusual characteristics of bdelloid roo tifers suggest that this scenario may apply to them (5, H, 12). First; bdelloids can survive extended (up to 9 years) a nd repeated bouts of complete desiccation (anhydrobiosisi at any life stage (14, 15). Second, anhydrobiotie bdelloids have an extraordinary potential for wind dispersal as tiny (usually <300 urn) ovoid propagules {called "nms") (l4 11), resulting in circumglcbal distribution of some taxa (18). Third, bdelloids can thrive in almost MY moist habitat. rapidly coJor-tizing even the most ephemeral patches of moss or rainwater on every continent (14-18),

All identified parasites of bdelln:id rotifers are comyectes or hyphomycete fungi. Most belong to Rotijer()_phdlOra~ agenus of obligate, lethal :fun. gal endoparasites that arc exclusive to bdelloids (J9, 21T), Infections spread when rotifers ingest spores (conidia), which lodge in their pharynx and produce assimilative hyphae. As the rotifer is killed and digested, hyphae puncture its integumont and, at the. an'lWiltcr interface, produce eonidiophores carrying hundreds of new conidia.

We investigated whether populations of the bdelloid rorifer Habrotrocho elusa can. escape the fungal -parns\teR0!if€ropht}-u;wa angustispora (19) in space and time, through anhydrobiosis and subsequent wind dispersal, We transferred rorifers from a rncnoclonal population singly to

29 JANUARY 2010 VOL 327 SCIENCE www.scienesmag.org

Petri dishes (20). A control group was allowed toprolifsrate without parasites, whereas six experimeutal groups were inoculated after 9 days with -712 (± 210 SD) conidia ofR_ angustispora. The first experimental groupremained hydrated throughout the experiment, but the remaining five were desiccated 72 hours after inoculation, maintained at 39.8% relative 'humidity (- 1.8% SD) for 7, 14, 21, 28, and 35 days, respective ly, and then rehydrated.

Without anhydrobiosis, R. angustispara exterminated Ii. elusa populations in 13.8 ± 2.6 days (mean ± SD, n = 1,2 populations), during which time the uninfeeted control populations 'were peaking (Fig. 1 A). In the dishes that were rehydrated after 7 days of desiccation, (he parasne initially seemed absent Recovered routers were healthy, desiccation had fractured and soattered the fungal conidiophores, and water samples were not infectious to fiesh rotifer cultures, suggesting absence of viable conidia (2U). However, within 48 hours, hyphae began to re-emerge from the remains of dead rotifers, generating new COilidiophores and exterminating the populations in 18.6 .t 4.0 days (Fig. J B), Excluding a four-day lag while the fungus regenerated, the time coarse of the resurgent infection it) the rehydrated populations did 110t differ from that observed iu the hydrated, inoculated group (i.e., complete extermination in 14.1? ± 4_0 days versus 13.8 .!. 2.6 days; unpaired I test, It = 22, t = 0.544, P = 0"59). Identical runga.l regeneration was observed after 14 days of desiccation (with exterrninarion in 14.7 ± 2.3 days).

However, dramatically different results occurred in the other three experimental groups. After 21 days of anhydrobiosis, 60"/0 of rotifer populations remained free of fungal infections tor the duration ofthe experiment (20 weeks), Longer periods of desiccation were even more effective:

After 28 and 35 days, 85 and 90 .5%, respectively, of rehydrated populations remained fungus-free and thus grew significantly ITIOre (.Fig. I C).

In nature, desiccated bdelloids arc dispersed ~y wind, adding a second dimension in which they might evade parasites (j 6), To simulate wind dispersal, we established a heavily infected popularion of H. elusa in a eli sh with a natura I, friable substrate (sterilized moss and silt) and placed it in a wind chanber with a mrbulentflow equivalent to a light breeze (20). Desiccated substrate particles were blown toward empty target dishes 30 to 40 em away After" 7 days in the chamber, these "wind dispersal" dishes h<!d accumulated 5_01 mg (:12,52 SD) of material; they were then hydrated and monitored for (i weeks.

Enol;lgh roufers survived to establish populations in 17 of 24 target dishes. Fungal infectionaappeared in seven of the newly established populations and exterminated them within 16.4 ± 5.0 days (Fig. 2A), However, the \0 remaining wind-dispersed populations (58.8%) never exhibited infections. We replicated this experiment and obtained similar results (i.e., 63.6% of winddispersed populations remained fungus-free; see

REPORTS

I

fig, S2). For comparison. "wet dispersal" dishes (Ii = 24) were created by pipetting 5.0 mg of well-mixed, suspended substrate from an identical infected source dish that was not desiccated (20), Rotifers established populations in all dishes, but fungal conidia also were readily transmitted, and ail populations were exterminated in 22.3 .t 5,7 days (Fig. 2B).

Our first experiment indicates thai anhydrobiosis allows II elusa to shed R. angustispora, apparently because the fungus is less resistant to extended desiccation than its host (Fig. Ie and fig. 51). Our second experiment demonstrates that wind dispersal acts in concert with anhydrobiosis to facilitate escape from RotIjeroplllhol'lJ, which is primarily waterborne. Three weeks of in situ desiccation (Fig. lB) were required to

achieve the same rate of parasite elimination (-60%) that was seen after only 7 days in the wind chamber (Fig. 2 and fig .. S2). In nature, dispersal distances and maximal durations of :JOe hydrobiosis are certainly fur greater (U 18,2il). Together, our results demonstrate that anhydrobiosis, even for relatively short periods, enables this bdelloid rotifer to disperse without aCC0111- paniment by a lethally coadapted fungal parasite.

Although facultative or recently evolved asexuality is common among organisms resistant to physical extremes, as well .3$ those with passively dispersed dormant propagules tl), the extraordinary capabilities of the bdelloids in both regards sets them apart. Few animals or plants can withstand complete loss of cellular water, and even then the ability is usually restricted

.. '

---~==~~·~·I j

!!II

-...___~ ·~o f

(18 ~

[

'Fig. 1. (A) Growth curve for rotifer populations (H. eli15o) inoculated with R, onglJ5tisporo veffijS urn medal controls. Error bars indi(ilte 95% confidente intervals (as). (B) Growth rurves for inoculated po pu la ti ons su bjected to extended d~siceati 0 n, The pro portio n of infected rotifers in e<jeh dish W<j5 used to construct a mea n rno rtality Ulrv!;'. Because all parasitized populations were even tuaLLy extermi nated, thi s curve fina Uy sta bilizes at a vallie equ a l to the proportion of dishes in which infections returned after desiccation. (e) Statistica [ summary of treatment differences from (B). PopuLation growth was quantified for each dish as the natural leqarilhm of !he tota I a rea under its growth cu rYe; shaded bars show tne mean for each treatment group {:!;9S% CIl. Open bars shoW the proportion of populations that were exterminated by the fungus (±9S% CI). "'P < 0.05 by one-way analysis of variance, Tukey post hoc (for growth), and 'l' (for mortality), to rrected post hoc.

www.sciencemag.org SCIENCE VOL 327 29 JANUARY2010

575

A C
! !JD
rn 1500' f3
... 11.0
CD t
:::: AW 2 10
- 110,
0' iI4lIO .. c GO
I&.l,; Q
01 aoo '=' 5.0
ji.
'I: 1.
-;: 100
:::i 0 g_
0 5 to 15 10 2S 30
B
1.00- I c e ;:

J!!

:I CL o Q..

... GI CL III .... GI

-

;;

a

DI c

i!

211

~~~------------------~u

lCO

10 IiQ

a

I:!'.

ii' ;:;t

~ Rr Q.

0-

m !.

~~--~~~~~~~------~!~"' ~ !ill: -

_~.~~ : 1

+4~ __ ~~~ ~~~~~O

HIO

Fi!l' 2. (A) Growth A

cu rves (erro r ba rs indi-' .

call" :t.95% ell and mortality for 17 H. elusa popl,lliitions founded by artificial wind dispersal of substrate (5.01 mg ± 2.52 SD) from an infected source culture desiccated Ior 7 days. Parasites exterminated 7 populations, but 10 remained uninfected (S8.8%), compared with complete mortality of all 24 control populii-

tions founded concurrently by wet transfer

of 5 mg of the same infected source material. (8) Statistical summary of differences in (A) [see Fig. l( legend); "P < 0.0001, unpaired t test; ~p < 0.0001, Fisher's exact test

REPORTS

1011

o o

to specific life stages (e.g., seeds, larvae, eggs) (5). Only bdelloid rotifers and certain tardigrades and nematodes can tolerate repealed bouts of desiccation at any Ii fe stage (14, J j), and .. of these, only the rotifers occur frequently in samples of'rain and wind (l6). With their smaller, more aerodynamic tuns, bdelloids can colonize tiny, isolated microhabitats more rapidly than t:ardigrndesand nematodes (l7).

The Bdelloidea have been called an "evolutionary scandal" (1). because their ancient asexuality seemed to challenge 'all hypotheses for the long-term maintenance ofsex, However, if anhydroaiotic dispersal enables bdelloid species to escape temporally and spatially from some or many natural enemies .. their coevolutienary burden would be substantially reduced. Therefore, our results are consistent with a scenario in which bdelloids have evaded parasites and pathogens over evolunonary time, without incurring the costs of sexuality, by playing a never-ending game of "hide-and-seek" (11, 1Z'r Decoupliag from coevolving enemies could help to explain the ancieot asexuality of the Bdelloidea under the broad array of models that derive from or incorporate Red Queen dynamics.

Referen(es "lid Not~s

1. G. Bell, rhe Maslerpiece oj Nature (Univ. of California Press, Be ,keley. C.A, 198 2).

2. O. P. Judlon, B. B. i'lormatk, Trends fali. fvoL ~1. 41 (1996)

3. D. B. Mark WE lch .• J .. L. M ark w~ teh, M. Me >~lso~, Ptoc. Naif. Awd SCi. U.S.A 105, 5145 (200S).

4. T. G. Barr.dough. D. FOl1.t~neto, C. RicCI, •. A. Herniou, Mol. BioI, fvo~ 24. t95? (2007)_

5. B. B. No rmark. O. p.. Judson. N. A. I'll orono BioL}. Lil/n. Soc. 79. 69 (2003).

6. 5. A. W5t. C. M. Live(y. A. F. Read. I. [voL BioL 1l, 1003 (1999)

7. W. D. Hamilton, Oiko5 35. 282 a980).

8_ W. D. Hamilton, R. Axelrod, R. T~ne"" Pan::. Natf. Acari Sci. U.s,A. 87., 3566 (1990).

9 M_ '"liltne. R. o. KOL!jl05, S, Bonhoeffer, Ttrnrk TitoL f!;tJL 23. 419 (;moe).

10. I. Jekela, M. F. Dybda~l C. I'll Lively, Am. NQI, 1,74 (suppL 1), 543 (2009).

U. R, l Ladle, Il'. A. ]b~n5tone. o. 1'. [udson, Prot. R So<;, London Ser. B Bioi. Sci. 253. 155 (1993).

576

............... _---~.

.Y

5

10

15 2.0 2.5 30 TlmetdaysJ

35

B

0.0

12. o. P. )udlon,}. TheOf_ Bio/. 186,33 (19'97). 13, A f_ Agr~Wil~ ?loS Bioi. 4. "265 (2006).

14. LN. Ric.;i. HYdrtJbiG/9gi{J 446J447. ~ aOO).).

15. A. lunnadiffe, J. Lapinski, Phi/os. nails. R. Soc. Wildon Set. 11 Bioi. Sci. 3.5B, 1755 (2003).

16. D. G. ]er'll;in$', M 0 U~d"'\VGDd, Hydtobiologia )871388. 15 (1998).

17. C E CiCer€5, D. A. Soluk, Oew/ggia 1)1, 402 ('2002). H!. D. fonlil~Ho, T _ G. Barraclough, K. Chen, C. Ried,

E< A. H~mioul Mo/_ Ecol. ,7, 31~6 (2.008)

19. G. L. Barron. COl). 1- Bot. 69,494 (1991).

20. S~~ supporting matenal 00 SCience Online.

21. We (hank D. Slotlje. G. Maltet>. and 1 Hue tor rMe.arch os,;i5!<1n~e: IC Hodge, S. Glilr'kling. D I'm'r(;l!ie!o, "i\~

G. Barron tor he IpM ad\lk~; U;S, AgliflJ uurat R~lear(h

9.0

1.0

8.0

0.8

7.0 6.0

0.6

5.0

0.4

0.2

4.0 3.0

Wrtr. Wind DJspersal treatment

Service collEction of ,ntomopdtnogenic fungi (IIRSEf) fur III ngal ilO1il1es; and 1. Seelfy. J. Scarl. K. R@eve. R. lia,gulll, C Liw.ly. N. Hairston, "rid R. BoQ~er fo, u,eful

(0 mmentaries; Fin~ nri al support wa, provi d~d by Sigma Xi. V S. Department of AgM Cl! ltu fe, ;md Cornell Unive'lity (Oep artmenl of Neurobiology and Behavior ~nd ;3 n S. H. Wei!.> Prf5ide n~ al FellllVll>hip to P. w .S.).

Supporting Online Material

www .s dencemag.orglcg i/co n tent[fu 1V12715 95515 7 4/D(1 Ma teriels an d Meth 0 ds

Fig,. 51 .and 52

Releref\''''

16 July 2009; o((epted 10 [)e(~Oiber 2009 1 O.lll6/,{j ence .1179 2 5·2

Objective Confirmation of Subjective Measures of Human Well-Being:

Evidence from the U.S.A.

Andrew J. OSW<![ql" and Stephen Wu2

A huge research literatUre, across the be havio fal and social sciences, uses info rrnation on individuals' s'ubjettive well-being. These are responses to questions-asked by survey interviewers or medical personnel-such as. "How happy do you feel on.a scale from 1 to 4?" Yet there is little scientific evidence that such data are meaningful This study 'examines a 2005-.2008 Behaviorel Risk Factor Surveillance System random sample of 1.3 million U.S. citizens. Life satisfaction in I!i!ch U.S. state is meas!)red. AcrQSs America, people's answers trace out the same pattern of quality of life as previ Ol.!s[y estimated, from solely non~u bjectivI'! datil, inane branch of economics (:so-called "compensating differentials" neoclassical theory, originally from AdiJm Smith). There is a state-by-state match tr = 0.6, P < 0.001) between subjective and objective well-being. This result has some potential to help to unify disciplines.

The conce~t ofhuman wel.l-bei?~ is important but difficult to. study empirically, One approach is to listen to what buman beings say Research across the fields of psychology, decision science, medical science, economics. and other social sciences draws upon questionnaire dam 011 people's subjective well-being (1-13). These are numerical scores (e.g., from very satisfied to very dissatisfied) in response to survey

questions such as: How happy are yon with your lire? Sample sizes in these statistical analyses typically vary from a few dozen individuals ill a

1 IJEp.o[lIi1 en! of E'OfIOniiCI, UrJiversityof WarwiCK. (oV~n!Jy CV 4 7 A~ UK. 2Dep.lrtmen! of Eco norn irs. Harnilton (o\leg e, Oi nlon, NY 133:23, lJSA

"To Whom correspondence shOUld be addressed, E·mail; and rew ,oswa ld@warwi(k.ac.uk

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

laborsrory to many tens of ihousandsof people in !I household survey (2, 4).

If reported well-being numbers provide aceutate infbnnation about human experience- so are not merely random, deliberately Of accidentally untruthfu', or irredeemably affected by the possibility that they !TIP}' not be comparable from one person. 10 ·anothe.t'----SUch data offer important intellectual opportunities [0' research scientists and practical ones to policy-makers. However, at present, there is little empirical evidence, of'tbe sort able to convince a skeptic, that they do. Perhaps the closest to a validation (of the believability) ofsubjective well-being scores is the findiQg that ihere are correlations between reported happiness and blood pressure, and among emotions, relative reward. and the brain (14-18). This argument, although suggestive, faces a difficulty. The demonstration of a statistical link between pbys· iological.IIle<JSLn·cs!lI1d subjective weU-ootng answers usefully establishes that the latter are not random numbers, But skeptics can reasonably argue that it does little more than that. l3iok)gica.l indicators are not themselves unambiguous treasures of human happiness or unhappiness.

This study focuses not on people bur on places (J 9). P laces have characteristics that human beings find objectively pleasant (Hawaiian sunshine or Colorado scenery) arid unpleasant (CQnnecticl)t land prices or New York City traffic fumes); many are cardinally measurable, The study bJends new data from the U.S. Behavioral Rill'k Pactor Surveillance System (BRFSS) (20,21), elements of tbe economist's coropcnsanng-differennals theory CZ? 24), and recent research on so-called amenity effects in happiness regression equations (25, 26). TIle jargon tenus "compensating" and "amenity" here capture.the following idea In a country where people can live wherever they please. a place with a harsh environment- 'think of commuters in Anchorageat 5:30 a.m, in February- -has to offer i Is 10 h abitams some offsettin g feature, 0 fted an income advantage, to persuade them to stay there rather than leave to live in a different region. Si 11111arly. intrinsically pleasant environments have to offer less, other Thing» being equal.

The study examines life satigfumion among a. recent random sample of 1.3 milllon U.S. inhabitants. The size of the dataset, gafuered between 2005 and 2008, provides opportunities denied to previous investigators. The often-used General Social Survey, fOI" example, samples only -·3000 Americans biannually; it is too small mallow Sl:1lT1.l-by-StBTe analysis.

The BRFSS can be downloaded at www.cdc. gov!BRFSSa:ndis a state-based system of health surveys that gathers {daily) infonua tion on health risk behaviors and disease prevalence in America. for many states, it is the only source ofup-rodale data. The survey was started in 1984 by the Centers fOT Diseuse Control and Prevention (CDC). The data are collected in all 50 states, the. District of Columbia, Puerto Rico, the US. Virgin Islands, and Guam. TIle aim is to "identify

emerging health problems, establishand track bealth objectives, and develop and evaluate public health policies and programs." Although initially smaller in its sample, 350,000 adults are !lOW interviewed each year; the BRfSS has become the largest (random-digit dialing) telephone health :SlJf\ley in the world. Although the nature of the datil set means that it is' not possible to record clinical data on people, the advantage is that its samples provide representative snapshots of the self-described health of people in the United States. Thcsupplememary online material (SOM) gives further inforrnation (l9).

The exact wording of the BRFSS lifesatisfaction question is, "In general, bow satisfied are YQl.I with your life?" Here people are able to answer with one of the follOWing: very satisfied, sari sf ed, dissati sf! ed, or very dissatisfi ed, Th e questionnaire is publicly availableat the BRFSS Web site, Information on BRESS individuals' lite sansfaotiou was collected fur thefirst time in 2005. Published research on I ife sat isfaetion US! ng this datil. set is tbas only beginning.

Life satisfaction in the United. States can be treated in a cardinal way by assigning I to 4 to the four answers, where "very satisfied" is assigned a 4. The mean of life satisfaction iii modern u.s. data arc then 3.4:! 0,6 (SO). Well-being-answers are skewed, they lire more commonly in the upper end of the possible distribution presented to an interviewee.

Building on now-l;tru1dard methods used in the literature (4, 9), a life-satisfaction regression equation W<lS estimated, using 1.3 million data points, ill which a number of'iudependent variables were included, These were (seven) banded variables fur different levels of household jnceme; variables fur the survey reepondent's a."oe,age squarefl. and gender; five variables for different ethnic types; and variables fur the person's level of education and marital starus,as. well as for employment status: being self-employed, retired, .a student, unemployed, or a homemaker. The sample used in the study was all t110SC respondents between the ages of 18 and !l5. Also included as independent control variables were eleven "durnl11y"variabJes fur the month offnterview; and a set of U.S. stare dummy variables (because DE, is included, the regression sample number is 51 states). It can be Seen that a dtunmy variable that is coded zero or lInity acts as a y axis intercept shifter ill a. regrossi on equation. The study used a linear Ordinary Least Squares estimator in which fhe four possible satisfaction values of ihe dependent variable, were assigned the integers from a high of 4 to a low of 1, and standard errors were acljusted for cl ustering at th e state level. Sti bstan tive tin dings were not altered by switching to an ordered estimator. The simpler method allowscoefficieors (given U1 the first column of Table I, each of which measures a vertical intercept shifter in a lifesansfacrion equation) to be read off as cardin a 1 lire-seristactien points,

111e key contribution here-now feasible because of the availability of BRFSS data- is

REPORTS

that it is possible to measure the pattern of peopie's feelings of well-being (their life-satisfaction scores) across the geography of'the United States. It was helpfu] that the state-by-state durumyvariable panern of coefficients was not very sensitive to which exact demographic and personal variables were Included in the lifesetisfaction regression equatiol1,

Information about how much Americans enj oy their lives is shown iu Table. 1 . The numbers in the first col umn are regression-corrected life-satisfacti(m patterns across space in the United States; they are stsre-dunmy coefflcieats. Alae bama, because it comes first alphabetically, is the base category against which other states are measured; this flOnnaliz;atiotl makes no difference to the analytical conclusions.

A life-satisfaction regression equation L '" L~) fur individual Americans lies behind Table 1. The statistical structure of these American equations was empirically similar to those round for many industrialized nations within the existing "happiness" f terature (19). II ceutrols for (that is, includes as. independent z variables) the incomes and demographic eharacteristics of same pled individuals. This is necessary to the design of the study's test. The test is not primarily au attempt to assess the different kinds of people who are "happy." but rather the kinds of geegraphic areas.

However. there is another, and older, way to try to assess the quality of life 3.Q!uSS regions. In all elementary form, it continues to be published in magazines. ill the form of .quality- of-life rankings, More fonnally, it appears in the eeonomics regression-equatiou literature and uses obj 6Clive data rather than S1.I bjeotive information Its focus is people's overall utility (or well-being), The approach relies on the Idea that people enjoy having high income and spatial amenities such as sunshine hours, but that they dislike disamcnities such as traffic congestion. Regression equations are estimated ill which wagearenrs, and house prices arc the dependent variables, The independent variahles in these regression equations are measures of the amenities and disamenities of the areas of the United States, The result is then estimates of "compensating differentials" by geographical area. This allows economists to read off the implied number of dollars (positive or negative) in workers' pay packets in Los Angeles tilat would be required to offset the unpleasantness of. say, poor air quality. and the pleasantness of high numbers of sunshine hours. The theory of compensating diffunmtials is attributed 10 the Scottish economist Adam Smith (27).

Some ofthc most recent and thorough research in this vein, by Stuart Gabriel aadoolleagues (24), was published in 2003, It used, among other attributes. objective indicators fur each U.S, state. such as precipi1;<ltiol1, tempemture, wmd.speed, sunshine, coastal land, inland water, public laud, National Parks, hazardous Waste sites. en viromneota 1 "greenness, .,. commuting time, violent crime, ail"

www.sciencemag.org SCIENCE VOL 327 29 JANUARY2010

577

REPORTS

quality, studeot-teecher ratio. load taxes, local sp0lidiilg on education and highways, and cost of living. Gabriel's analysis used no subjective data. Instead, these objective data were corobined in 11 weighted average (24), where, as expiamed in the SOM (19). the weights were not chosen arbitrarily but rather according to co effic lent s izes rak en (r0111 regional wage and price equations.

In nontechnical terms, the compensatingdifferentials methodology of Gabriel and earlier economists argues indeed, assumes that higher house prices are a revealed signal of higher quality of Life, other things constant, because humans will move toward the areas they find attractive, which, in turn, drives up housing prices. A spatial equilibrium is reached when a representative citizen obtains the same total utility 10 each region. This total uriliry depends 011 a combination of the inherent quality of life (sunny days, clean air, and so on) plus the earned income from living in that area, This methodology a llows these two to be separated conceptually, It is heJpful to realize tbat the Gabriel et at. state-by-state quality-of-life ranking is not exceptional, The authors note its approximate similarity to those in earlier writings (23),

The ranking, according to the Gabriel 131 a.l. methodology, of the inherent quality ofIife across the U.S. is shown 10 column 3 of table. L The ranking is computed using data for the year 1 990,

which appears to be the most recent estimate in tb e pub lished economics Iiteranus, A.~ with column I, column 3 is 110t to be thought of as a representati on of tota I utility: instead, it c-aptures nomncorne elements of human well-being.

Informarion on both methods of life-evaluation is provided in Table 1> The first state-dummy coefficient in column I of Table 1 can be interpreted as Shl)wing mat con-cered satisfactionwith-life 00 average in Alaska is 0.013 cardinal life-satisfaction points. below that io the base stare of Alabama; Arizona is indistinguishable from Alabama, and so on down the listed states. Sansfaotion with life is lowest in New York. The particularly high-satisfaction states are Louisiana lind Hawaii. In Table I, the 95% confidence intervals correspond, in each case, to a test of the null hypothesis of zero 00 the coefficient; it should not be presumed that there is a statistically siguiflcam difference between each of the states within Icw-satisfaerion and high-satisfaction groupings of states.

Although it is natural to be guided by tormal survey data .. it might be thought unusual that Louisiana-c-a state affected by Hurricane Katrina comes so high in the state lifesatisfaction Jeague table. Various checks were done [discussed in SOM (19)]. It was f-ound that Louisiana showed up strongly before Katrina and in a menjal-health ranking done

Table 1. Two·ways or measuring the quality oHif!:! in America. The BRFSS lifesatistacti9n equation uses coefficient dumrny-vanable values for each state. The (omp<'nsating,differentiaLs methodology of objectiVe qua\ity--:of-life rank, by stnte,is from G;;Ibriel et ai, (23). Alabama is included in tnlt data {and in Fig. 1}. It

by Mental Health America and the Office of Applied Studies of the U.S. Substance Abuse and Mental Health Services Administration, based on data from the National Survey on Drug Usc and Health 1004 05 .. Nevenhelcss, it is likely that Katrina altered the composition of tbis state-u3meIy, that those who left were not 11. random sample of "the popularicn=-sc Some caution in inrerpreration is caDed for about this state's ranked position, and that position may repay finurestatistical investigation,

Differences in well-being across states are not minor. In cardinal teJTIlS, us demonstrated in the SOM (19), they correspond to up to 0 12 life-sstisfaction points across U.S. states, which IS 3i ttl i lari u size to the individual eross-secdona 1 effect on li fe satisfll.cti on of t:nari tal separation 01" uneroploymeet, other things being equal.

These two approaches ate brought together in Fig. J . II reveals a notable match between the ful.ly adjusted lifs-satisfactinn levels of Table I and the objectively calculated Gabriel ranking, The subjective. well-being and l,he objective well-being measures tally. The estimated linear equation is written out in full I n Fig. 1 and is apprexirnately of equation form y = (J,003 O.OOb:.

The gradient in Fig. 1 isnegative, This is because the Gabriel method, beinga ranking of quality of life across states. uses the convention that 1 is the higbes: raakieg Q'f quality of life

is the base category, against which the other states' coefficients are normalized. tn effect., A.labama has a lif~satisfaction coefficient of zero [and a raoking of 26 in (23)J. Number of observations, 1,213,,992. The standard errors were adjusted to r cLusterin gat th e sta te l.eveL CI, co nft dem;e interval; N A, not a pp liea ble.

BRFSS (ife-5atisfaction meHlod Camp BRFSS 'life-satisfaction method (omp
din. diU.
Sta.te Coefficient 195%(1] methodS .State Coefficient [95% en methods
Alaska -{I.OB [-OmS, -0.008] 23 MQntana 0.001 [-{J.OOl, 0.Q04] 4
ArizOJ1<l 0.001 [-0.002, o.oosj 20 Nebraska -{J.044 [-0.047, -0.041] 16
Arkansas -{J.D17 [-0.019, -0.D15J 3 Nevada -{I.065 [-0.068, -0.0621 29
Califol'fiia -{).076 [-0.080, -0.0721 42 New Hampshire -{).033 [-0.1136, -0.030] 43
Colorado -0.027 [-{).030, -0.024] 34 New Jersey -0;078 [-O.OBl, -0:075] 47
Connectitu! -{).OS1 [-0.OB4, -{).D781 32 New MexIco -{).029 1-0.034, -0.0241 14
Delaware -{).027 {-O.029, -{).025J 30 New York -O,08B [-0.090, -0,085] 50
District of Columbia -{I.048 [-O.OS!, -0.045] N.A. North Carolina -0.013 [-0.Q15, -0.012] 17
Florida 0.004 [0.002, 0.006J 10 North Di,lkoia -{).030 [-0.032, -0.027] 6
Georg.ia -{I.021 [-0.023. -0.020] 36 Ohio -{I.07D [-0.071, -0.06Bl 33
Hawaii 0.011 [0.004, 0.018] 38 Oklahoma -{).026 [-0.0il.9, -0.024J 21
Idaho -{J.014 [-0.017, -0.0111 5 Oregon -0.040 [-0.04.4-, -0.037] 22
Illinois -{).072 [-{J.ON, -{).069] 48 PennsylVania -0.067 [-0.069, -0.0651 35
Indiana -0.078 {-O.OBO, -0.0771 44 Rhode .isla nd -O.06B [-0;071, -0.066] 12
IoWa -{).041 ~-D .. 044, -a.U3B] IS South Carolina, 0.001 [0.000, 0.002] 18
Kamas -0,044 [-0.046, -O.D'n] 19 SOUl" Dakota -0.014 [-0.017, -{).010] 2
Kenlucky -{I.04S [-0.047, -{J.0.43j 24 Tennessee 0.003 [0.001, 0.0041 28
Louisiana 0 . .033 [0.032, 0.034] 8 Texas -0.014 [-0.018, -{).010] 25
Maine -{I.OQ6 [-0.009, -0.003] 9 Utah -0.026 [-0.030, -0.023] 39
lVIaryland -0.066 [-0.069, -0.064] 45 Vermont -0.017 [-0.020, -0.014] 13
Ma 5 sa ch [JS etts -{).070 {-O.OJ3, -0.068] 27 Virginia -0,033 r-M35, -0,031] 31
Mkhigan -{).079 [-D.OB1, -a.an] 49 Washington. -0.046 [-0.049, -0.0421 41
Minnesota -0,.031 [-0.034, -0.0271 46 West Virginia -0.0<14 [~,046, -0:042:] 11
Mississippi 0.001 [0.000, 0.001] 7 Wis(onsin -D.038 [-{I.040, -D.03S] 37
Missouri -0.064 [-0.066, -0 . .062] 40 Wyoming -0.013 [--0.016, -G.OI0] 1
578- 29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org Fig. 1. Fitte.d equation:

Adj Listed Life Sa tisfacti on = -o.ooss - 0.0012 Objective Rank; r = 0.598, Each dot is § a state. The wfrelQtion i~ -~

'>igninGlIIt at P < 0.001 ona ~

two-tailed test. This figure W

2 plots state dummy coeffi- :'i

dents from a life-satisfaction -g

1ii

equation against state rank ~

in quality of life from the 1

compensati nq-differentia Is ,~

"'

results, based on objective ~

a rnenitles Ii ~,e su nshine hours, £.

of Gabriel et al. (24). Life satisfacticlO is· coded for each .individual from a score of 4

(VI" ry sati sfi ed) to 1 (very dissatisfied). On the y axis,

the [egression controls for household income. as well as the sLirvey respondent's gender, age, age squared, education. marital status, employment status, and race,and also year dummies and month-af-interview dummies. Alabama is included, Washington. DC. is omitted from 1.24) and thus fror,n here .. The bottom right-hand observation is New York. Wording of the question in the BRFSS qu€stionntl!ire (questionnaire line 206):

(ahd hence the. number 50, for the state of New York, corresponds to the lowest quality-of-life area in the United States). The calculated Pearson r coefficient is 0.6. With 48 degrees of'freedorn, t11C nail hypothesis of'no correlation is therefore rejected at the P <: O.OOJ level (in fact, approximately P = 0,0001) on a two-tailed lest. The I). is 0.36: thar is. -36% ofthe variance in the y axis variable is explained by the"" axis variable. As shown more fully in the SOM (19) and in (20), the higb-satisfactiou states in Fig. 1 are not simply the high-income ones,

A correlation 1J0efficientc of 0.6 is unusual by the standards of behavioral science, It is high by the cut-offs suggested by Cohen's (28) rules-ofthumb (which argued that in human data an r value over 0.5 should be seen as a large association, and 0.3 B medium one). An r = 0"6 is the same degree of correlation, for exampit, as bas been found for people's own lifu-satisfdction readings taken 2 weeks apart (tbat is, using the same well-being question, asked of the same person) (19). An r = 0.) has been demonstrated for subjective well-being correlated with BEG 1"\symmetry in the brain (J 5). TIle variable on the x axis in Fig. I is an ordinal rank. As would be expected, an arguably preferable Spearman rank test confirms the same relation as the Pearson test An alternative estimation method would be muJtilevelmodeliog (30, 3n TIle data set is too large to allow this e.asi1y, but experiments em subsamples were done. These suggested that equivalent lire-satisfaction equations. emerge.

Although this study is designed; as all. analyrical contribution, measures of subjective well-

O.GII-

0.02

o

-0.02

-0.06

-0.08

-o.i

being are becoming more widely discussed outside the laboratory. As the. draft of this article was being reviseq, the Stiglitz-Sen-Firoussi Commission 00 the Measurement of Social and Economic Progress (-?2), set up by Nicholas Sarkozy to assess the appropriate goals of WestelTl gOV" ernments over (he remainder of the century, issued a report, Worldwide attention was garnered for its argument; "A ", 1)JJit)'ulg theme of the report ... is that the time- is ripe for dur measuremen! system to shlll: emphasis from messuring economic production to measuring people's well-being." "Measures of both objective and subjet:tiw well-being provide key Information about people's quality of l:ife. Statistical offices [worldwide] should incorporate questions to capture people's lire evaluations, bedonic experiences and priorities in their own survey." [Executive StnmMl:!yof Oonnnission Report (32), pp. 12 and 16],

III this vein, the CWTCnt study's principal conrribution, Fig. I, offers a possible bridge between different ways of tbinking--between, in particular, the fields of hedonic psychology and neoclassical economics (the latter has traditionally been hostile 10 the lise of data on subjectively reported feelings), It. offers a crosscheck on the spatial compensa ring-differentials theory of economies and regional science, It 111aY\lJso be relevant to the work of behavieml Scientists, geographers, applied psychologists, and mental-health specialists, The study's finding suggests rhatsubjcetive well-being data eontain genuine informauon about the quality of human lives.

•

•

.'.

.'

.,

'. . '.

•

•

•

• •

•

_.

..

' ..

• •

•

o

30

In general, how satisfied are you with your life? 1 Very satisfied

2S;gtisfi!1d

3 Di ssaflsfied

4 Very dissatisfied

10

4(l

REPORTS

"20

Obje:clive. Q-uality-ol-Ule Ranking (where 1 Is high and so Is low)

References and Notes

1.- R. Po, E~~erli~, ?roc Notl: ilc"d. sa U,;i,A 1M, 11176 (200}).

2. E. Dien~r, E, M, SUIT. ~_ E. luras, H. L. Sn1ith,

{'sjt/wi. ElIllI. 125, Z76 (199~t

3 I. H_ Fowler. N. A. Chrlstakis. 8M) B7,aB38 (200BJ. 4, A. J. O,wald, Icon. J. 107, isis (J. 997).

5 Y _ Y!mQ. A·m. Soo"l. Rev. 73, 204 (200B)

6. B. topes CardozQ, A. VHgara, F. Agani, C. A. GolWay, 1MIi! 284, 569 (2000),

7 B. R adcU II, Am. Polt: Sti. Rev. 95, 93 <J (2001). '!L Y. I, Hu, 5. Sle\VMI;·BrDWn. L T'wi99, 5. wekb, P5YC/IO/. Med. 37. IOO) (loon

9. E. F. P. Wilmer, Q. J. tcon. rzo, 963 (200S/.

10, D'. Gilllert. Stumbling I1/l Happinrss (Knopf:. New Y crk, 2006). n, a- 5Ieve!],Son.)_ Wqtferl, J, (.egal 'jIurj. ~7. (S~), SB (ZOOB). 12. A. t Clark, J. iaaot tcon. 21, :i 23 (2003).

13 E. W, Dunn. L B, Akoin. M. r. Norton. science 319. 16S7 (Z008).

14. E. Sandvik. E. Diener, L Seidlitz,J Per', 50~. P!;ycho~ 61 •. )11 (1,!9~).

15. H, L_ Urry et 0/', !';yehol. se: 15, 367 (2004),

16, A. Steptoe, J. Wa(dle, NeumbioL AfJing 2-6 L>uppl. 1), 10a (2005)_

17. K_ Fll""bach et 01" Stieoce 318, 1305 (2007)_

lB. D, G. Bt~nenIIQ\v~r. A. J. O,wa~d, J" Jiealth [con, 27, ?18 (20061.

19. Mate~als and melhods are availilble .1, lup~rting rna teria I on 5, ience Onti ne.

;1.0, S. D. BMg~r. (_ J. Donoho. H. A .. Wa\IInenl, Qual. ale Reo;. 18, 179 (2009)_

21. 11_ J o,wald, 5. Wu, unpublished manuscript, UniVerSity of Warwi,k lMgu!1 2009); presented at the lnstitute for ~he Study of Labor (IZA) Prize Conference, Wa!~lngtQn, DC, 22 October 2009; JZA discussion paper 4600

(N overn her 20(9) lrom Vli\vw. i "'. olg,

22. j. Roba{k, j. polit. fwl!. 90, 1251 (1982),

23. J. Gyourko, 1. Trac)", J. Pollt. f(on. 99; 7 H (1991).

24. S. A. Ga~~~I, J. P. Malley, i\i, L, W.1H~er, Reg, SCI: Urban [con, n. 619 (2003),

25, M. Moro. F. B~"retol1, Q. ferreira, J. p, ClInch. [(OL [(On. 65, 448 (2008),

26. 5, l(Jet~lng"r, £W<I.o J. 1}9, 482 (2009).

27 A. Smith. illl Inquiry into the Natuie' and Causes oj th€ Wealth 9/ Nations (UniV, of (hic090 pres" Chic~gD. 1716, reprinted 1977).

28. I. Co~~n, Stotis/ica/Power ill/airs;, tor tne BelwviQro/ 5ciellf€5 tE'rlba.um, Hillsdale. NJ. ed, 2. 19,88).

29 II. B. Krueg"r, D. A. S[hk:.tde. I P~/JII, [con, 92 • UiD (2003),

3D. M. G. Pilmu .• R. ZeUi, A. Gelman. Soc.lndic Re5 .• published 0111;0 e 28 Apti I 2009 (10.100 71s '1.12 05"009-9 481.2).

3:l 11_ Gelln~n, B. Shor, J. BaIUn'li, 0, P~r~. Q, J Po[iI_ S'i, Z, 1<15 (2Don

32. I. Stiglftl tt ol, Comml"ion on the Me~;urement of Eron ornic P'Erfmma nee and Sodol. ProgtBis ((m'nnji,';nn 01 t~e Goyemnl~nt of France, Pari), 5 eptemcer 2009); ;3 vaila ble at www,5tiglil2-I€W fitoussifr,

:n For tliwmi ens ~ nd creta; led poi n Is, we Ihoi nk four

ref" r ees i3 n d P J. H am rnond, M. Kimbil ll, D. Krupka"

E, luttrner, ). Mile" E. Prol0. and D. 5groi, \laluabl~ sugges"oos were mode by D. Beoj~mi". D. B(;;ochHowet, A. CI. r~. M Clemlifl ts, E. Di ene r, D, Oorli nil, D Gilbert, II. GDG(t;Ilt, C, Gr~h~m", o, Heifetz. S, Lu"(liing~r.

S. Mukhfrj~e, R. Put~am, and I. Smith. The first ilulhof5 work wa.l<u pported by a u. K. Eco~omi c an d Sod" I R~5Mfth reieilrw proiellcflhip, 'Nil {ol~fli,1> of interest.

Supporting Online Material

WW\V. <ciencerMg . Dr9/[9 ilmnl ell tlfu lVscienc <,.118 06 OlifD( 1 Ma I~iial, and Me In 0 d,'

Fig,. S 1 and 52:

T~blel Silo 54 R~IE'rence5

60

l~ Aug ust 2tJO 9; a c (epD~d 1 Decemb er 2009 Pu bli shed online 17 D~{emh er 2009;

10.11~ 61, ti ence _ll8Q61l6

Includ~ IMj, iolormation \Vhf" dting this pa per ..

www'.saeneemag.org SCI EN CE va L 327 29 JAN UARY 2010

579

REPORTS

Platelets Amplify Inflammation

in Arthritis via Collagen-Dep,endent Microparticle Production

Eric BoiL;;Jrd,l Pete t A. 'Nigrovk.1,l!: Katherine La rabee/ Gerald F. M. W",tt5,l JOflathan S. CobLyn,l Michael E. WeinbLatt/ Elena M. Massarotti,l,

Eileen Remold·O'OonneLl,3 Richard W. Farndale," Jerry Ware,!> David M'. Leel*

In addition to their pivotal role in thrombosis and wound repair, platelets' participate in inflammatory responses. We investigated the role of platelets in the autoimmune diseas~ rheumatoid arthritis. We identified plat~l!:~t miuop,Htides-submiuometer veskles elaborated by activated platelets-v-i n joint fluid from patients with rheumatoid a rthritis and oth er forms of

i nflam matory a rth ritis, but not in joint fluid from patien 15 with osteoa rth ritis, P latelet micropartides were proi nflammato ry, eliciting cytokine. respo nses from synoviaL fibroblasts via

i nterleekin-L Consistent with thesefi ndi nqs, dep letion of platelets attenuated murine inflammatory arthritis. Using both pharmacologic and genetic approaches, we identified the

colla gen receptor glycoprotein VI as a key trigger for plate[~t rnicropartide generatio n ina rth ritis 'pathophysioLogy. Thus, these findings demonstrate a previou~ly unappreciated role for platelets and their activation-induced micropartides in inflammatory joint diseases.

P latelets are highl~ abundant hem~lOpoietic cells, outnumbering leukocytes ill the peripheral circulation by almost two orders of magnitude (1). The role of platelets in hemostasis. ~md wound repair after vascular tojUl."Y is well !cn0\VO (2); however, there is a growing appreciation for their role in intlammatiorr. This role has been studied most carefully in srherosclerosis, a chronic inflammatory disease of the blood vessels in wbicll platelets release a broad range of inflaromatory mediators that support endothelial cell activation, leukocyte adhesion and transmigration, monocyte. maturation, and elaboration of cytokines and reactive oxygen species [reviewed in (3)].

The growing literature regarding the protoflarrunatory capacity of platelets prompted us to investigate whether they could participate in another common inflairnuaiory conditi on, inflammatory arthritis, Of the inflammatory arthritides, rheumatoid arthritis (RA) is tile most common (4), RA manifests as chronic mflarmnaticn of the synovial lining of the joint, resulting ill pain, swelling, and ultimately dcsuuction of cartilage and bone (5). Although evidence bas implicated lymphocytes, innate immune cells such as neutrophile and mast cells, and synovial tissue cells in t1'le evolution ofRA, to dare platelets have no known functional role. We first sought [0' determine whether platelets are present in

IOMsion 01 Ri1eumatotOlJY, lmmunoloqy and Allergy, Brigham aod Women's Ho~pit;t HJ]\/ard Medi(a.l Scheel, Boston, MA 02115, USA. IOMsion (I! Iminu~ology, Children's Hospital BO,tOI1, Bosion, MIA 02115, USA. 'Immune Di,eil,e Institute, H arva rd Medi ca l Schoo\. Bostuo MA 02 115, USA. ~ Univer5i ly of Cambridge, Departm~~t of BiochemiStry, Downing Site, Cam brid ge C B2 lQW, UK. ' U nivefsity of Arkilllsa:; for Medi ca I 5 ~ ences, Little Rock, AR 7 221) 5-7199, USA

'To whom correspondence should be addressed. E"mail:' d I e~@ Ii (5 .bwh .ha rva rd. ed u

580

RA synovial. fluids (SF) by use of the plateletspecific marker CD41 (GPllb/a2b from the platelet-specific integrin GPITbill.a/«2bj:Bj (6, 7). Plow cytomeuie analyses detected a substantial number ofCD41-positiv(,! events in RA SF (Fig.

1, A and B). Unexpectedly, most CD41-positive events Were smaller in size than intact leukocytes Of platelets (fig. S I), suggesting that these particles are platelet micrcparticles (MPs). Platelet MPs are intact vesicles (0.2 to I um in diameter) that {ann by budding from tbe membranes of activated platelets (8-1 0), We found on average just under 2 x 105 CD4r' MPs permicrohter of SP fi:mn patients with RA (Fig. IC). 10 conImst to RA SF, where MPs were present in all samples, CD41' tviPs were undetectable in 19 out of 20 osrcoarthriris (GA) SF samples CFtg. Ie). We also detected platelet MPs ill ethel: inflammatory arthritides (Fig. 10). We further investigated the presence of MPs o,nginating from other hematopoeitically derived cell types. Consistent with previous. observations (1 I). MPs expressing neutrophil, T cell, or macrophage markers were present albeit in substantially smaller amoinns than observed for platelet lVlPs iJ.1 RA SF (Fig. IE). Recognizing that platelets can adhere to migrating leukocytes (J 2, 13) .. we l;I)SQ assessed the presence of plarelets associated with SF leukocytes. Intriguingly, we found that a substli:ntiaJ Dumber of SF leukocytes costained with CD41 and the leukocyte marker CD45 (fig. S 1, 0 and B). Microscopic examination I·Cveeled that CD41 staining in these cells was 1"1:)· stricted to discrete particles of MP size adherent to the leukocyte surface; we could not detect

B

... ..

:I

'Ei ,

c 4I)CiOOO .' D -- e «IOCCO •
. ,:- .' ..
JCIIIIIOII' '. ~ *'* _. •
• ~
~ , •• a ;§ ..-
l' 1!!DXIi:I' '" I(IlQOQ -'._
.1, ~
•• ~ • "II!!!I
~ • • III: • • s '.,.'"
·toOOO . '
-'. 1- ~I.' ,,' ,- .-
• II I. ... ,
,.' '.'
.~ 0iIi 'COl' Fig ... 1_ Platelet MP5 are abundant in inftammatory SF. Cells in freshly is!Jlated RA SF were stained with lineage marker~: (D15 (neutrophiLs), C03 (T cell~), (014 (monccytesend macrophaqes] and (041 (platelets), or the appropriate isotype controls and analyzed by flow cytometry. {Al Forward· by sidescatter profiles of events in RA SF. Populations identified by furlher gating and lineage marker staining ,;:m, labeled. (8) Repre~enfatlve hiswgram of (D41+ (b~[k fill) pliltele!. MP~ resident in RA SF. [vents were gated ba~ed on the forward-scatter parameters indicflred in (A). (Gray fill" isotype controL) Data are (<'presentative of profiles from eight RA patients. (e) Flow cytometric qU<Jntification of CD41+ platelet MPs' (.:::1 urn as determined by size calibration beads) in RA and OA SF after removal of Leukocytes by centrifugation in = 20 donors per group). (D) Flow cytometric quantification of platelet ((1)41,+) MPs il'l SF from juvenile idiopathic arthriti~ OIA, fi = 6), psoriatic arthritis (PA, n = 19), and gout (n = 14). (E) Flow cytometric quantification of MPs ·in RA SF derived from the indicated cell types (n = 19 donors).

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.erg

Fig. 2, Platelets ilr!:' in- A volved' in arthritis development (A) Arthritis severity' after KlBxN seru m tra nsfer in mi ce administered a plateletde p leti nga n ti body (triangles) or l>alype control (squares); n = 10 mice per group. Data are the mean t SEM pooled from C two independent experi-' rnents, P < o.oo~ Arrows, parenteral administration of platelet-depleting antibody; arrowheads" K/BxN serum adrninistration, (8) ,Histomorphometric quantification of arthritis severity in ankle jQints, of platelet-E depleted and control mice at ellperiment 'termination; n = 10 mice per gwup. ~p -c 0-01- (c to ~) Arthritis severity was measured after administration of KlBxN

serum in mice (C) defi- .""r

,ient in thrornboxane synthase (Tbxos.J-I-), treated daily with (P) th~ thrombQxan~ A2 (TP) antagonist-5Q 29,548 or (E) ADP:P2Y12 inhibitor clopidogrel, or in mice deficient in (F) GPlb (Gplba -1-), Data are the, mean ± SEM, n = 10 mice per group. (e), (0), (F) P = not significant; {El Po::: 0.001.

intact platelets associated with Sf leukocytes (fig, SID).

We next explored the pathophysielogic importauce of platelets and platelet MP!; in inflammarory arthritis invivo. Here, we used the KJBltN serum transfer model of inflarnmatery arthritis, 111e progressive distal symmetric erosive polyarthritis observed in KlBxN T cell reeepter transgenie mice results from T cell recognition of a ubiquitous autnantigen, glucose-e-phosphate isomerase, presented by major histocompatability c0111pieJ1 class if l-A'P, driving high-titer arthritogenic autoantibody production [reviewed in (14)]. Arthritis can also be induced by passive transfer of immunoglobulin G (IgG) autoantibodies from KlBxN mice into wild-type mice. Numerous effector mechanisms have been onplica ted in the pathogenesis of the IgG-dr.iven effector phase of KlBxN serum-transfer arrbritis, including neutrophils, mast cells, Fe gamma receptors, and soluble mediators [interleukin-I (lL-I), tumor necrosis factor (TNF), complement C5af('5a receptor, eicosanoids, and the mast cell protease tryptsse] (14-J6). To study the role of platelets in this system; we initiated KiBxN serum tlllnsfer arthritis in animals treated with a platelet-depleting antibody regimen that rapidly (within 60 min) reduces platelet numbers by >95% for at least 6 days (17). We found that platelet-depleted mice ex hibited a marked reduclion in arthritis as assessed hy clinical scoring and by histological analysis (Fig. 2, A and B).

These. findings demonstrate that platelets are required for inflammatory arthritis development in vivo,

To gain further insight into the link between platelets and joint inflammation, we explored the mechanisms by which platelets are activated to release MP in the context of arthritis. Platelets can be triggered via several pathways, many of whicb have already been targeted tor the prevention of thrombosis. Among these pathways we considered were tbromboxaue A2 sthnulation of its receptor eTP) on platelets (blocked by TP antagonist SQ 29548), ligation of the P2Y12 receptor by adenosine 5c-diphosp}1,ate (inhibited by clopidogrel), and GPrb-IX. a platelet me1TIbrane glycoprotein complex. that binds to von Willebrand factor. By using genetically deficient mice and pharmaoclogic blo];;lqtde, we determined that inrerferencewith these. pathways did not impede development of joint inflammation, suggesting that these pathways do net regulate pli!'!tele;t MP generation in inflarmnatory arthritis (Fig. 2j C to .G).

Wbat other pathways to platele; activation could be operativeiu arthritis? Knowing that !he concentration of platelet-derived MPs within RA SF (2 x 105 per. microliter] greatly exceeds that ill RA peripheral blood (600 per microliter) (111), we hypothesized that platelet activation Likely occurs locally, The vasculature of the joint is in intimate contact with fibroblast-like synoviccyres (FLS) and the extracellular matrix

B

A '"

i

'._""'1fG c __ ...,_.

D

• •

REPORTS

(ECM) elaborated by These cells (/9), and thus we modeled the interaction of platelets with synoviocytes and their ECM in vitro. Mouse platelets coincubated with primm)' mouse FLS promptly released MPs (Fig, 3A). 1b determine the tligger fOj' MP release, we assessed MP formarion after treatment wIth specific. pharmacologie inhibitors or using platelets isolated from mice deficient 'in candidate genes. Consistent with our in vivo results, we found tIlat cycleoxygenase (Plgsrl-), tbromboxane, GP!Ibilla (ifgbr.l-),. Grlb (GpJba -1-), and ADP-P2YI7 pathways were dispensable (Fig. 38). We next explored whether the ECM generated by pri- 111m), cultured FLS could be the relevant stimulus, particularly collagen, Previous studies have demonstrated that collagen can activate platelets to form MPs (8) and that glycoprotein VT .(GPVI) is the predominant collagen receptor on platelets (2{J)_ GP\i1 is an immunoglobulin superfamily member expressed exclusively by rnegakaryocytes and platelets that signals via noncovalent interaction with the common '1- chain of the Fe receptor (2 J). Using platelets lacking either FcR-y-chain (Feel'} g I ) or GrVl (Gpol-), we found that generation of MPs by primary FLS was mediated predominantly via this pathway (Fig. 38). After confirming that human platelets also released MPs upon ccincubation with primary human FLS (Fig, 3, C and D), we validated the relevance of GPVT in MP release in humans using the GP"Vl.specitic: agonist eollagen-relared peptide (CRP) (22) (Fig. 3D). Further characterization of collagen-stimulated platelet MPs showed that their phenotype is congruent witb that of platelet MPs Dum SF and distinct from Ibm of intact platelets (table S 1). Finally. we assessed whether GPVT is relevant. for p~telet activation in vivo by administering KlBxN serum to Gp6-.J- and control mice and assessing the development of synovitis, Both clinical and bistomorphometric assessment CODe finned that arthritis in Gpo!- mice was signif icantly reduced (Fig, 3, E and F). These results confirm that activation of plate lets via the collagen receptor GPVI-a pathway resulting in MP generation-plays an imponant role in the pathogenesis of arthritis.

Having demonstrated platelet participation in inflammatory arthritis in vivo, we aimed to identify platelet MP effector activities that contribute to joint inflemmsrion, The most abundant cell in the pathologic rheumatoid pannus tissue IS we FLS (19), This lineage plays a substantial tole in the perpetuation of joint inflammation and in the destruction of cartilage (J9, 13), We surveyed the capacity of collagen-stimulated human platelet MPs to dicit a range of oytokines and chemokines frem FLS and observed prominent production of the broadly inflammatory cytokine 1L-6 and the neutrophil cbemoartractant lL-R (Fig, 4A and fig, S2). Consistent with this observation, incubation of tvlPs isolated from RA Sf induced PLS to release substantial quan-

WWW.SClencemag.org SCIENCE VOL 327 29 JANUARY 2010

581

I REPORTS

A

B

..

~ 1 : ~~_'~~~~~r-r-r-'_

o ~ 2 , , , , 7 ~ 9 ro

CIIy

D

Fig. 3. Platelets form MPs and partiei pa te ina rth riti 5 P atho p h ysio logy viii stimulation of the collagen receptor GPVI. (A) Representative flow cytometry forward- and side-scatter plots of C041 + mouse platelets incubated in the presence or absence of FtS. (B) Examinanon of candidate stimuli of murine platelet MP formation upon co-culture with FLS. Mouse platelets incubated in the presence of cyclooxygenase inhibitor salicylic acid (ASA), iso la.ted from mi ce treated with ADP:PlY12 inhibitor dopidogrel, or from the indicated gene-targeted mice were coincubated with mouse FtS. MP formation was quantified by flow cytometry. Data are the mean .t SEM, n ~ 3 independent experiments in duplicate. (C) Scanning electron micrograph of human platelets eXhibiting MP budding when incubated in the p~5ence 01 FlS. Arrows indicate -the edge Qf -the fibrobla5t~ like synoviocyte. Upper and lower panels are 9800x and 69,270x magnifications, respectively. (O) HUman platelets form MPs when incubated with FLS and when exposed to a GPVI-specific peptide ligand (eRP) but not in the presence of a related control peptide (GPP). (f) Arthritis severity after KlBxN serum transfer was quantified in GPVI-null (Gp6 _,I, (triangLes) or wild-type control (squares) mice. Data are ttle me"n 1. SEM. pooled from three ihdependent experimen~; n = 25 mice per group. P < 0.001. (F) Histomorphometric quantification of arthritis severity in ankLe tissues from GPVI-null (Gp6 -1-) and \NT mke a! experimentaL termination. Data are the mean 1. SEM; n = 25 mite per group. *P ~ 0.019, ""P < 0.05, ***P < 0.01.

titles of both eytokines (Fig. 48 and fig. S3). We focused our further studies on MP stimulation of ll..-8 by FLS because SF from the inflammatory arthritides is rich in n eutrop hils. To elucidate a mechanism by which platelet MPs stimulate FLS, we used a genetic approach with MPs generated from mice deficient in specific candidate genes. We found that platelet MPs generated from rni ce lacking both lL-II;( and lIA~ (llluib -.'-) were incapable of stimulating murine FLS to produce the murine IL-S ortholog KG (Fig, 4C). Furthermore, FLS generated from mice deficient in the !L-l receptor (lllrrt ) were unresponsive to platelet lVlPs, though release of KC remained intact after TNF stimulation (Fig. 4D). By contrast, IY1Ps from mouse platelets. deficient in prostaglandin synthesis capacity (Pigs rl-) retained their ability to stimulate KC production from FLS (Fig. 4C).

Platelets exhibit membrane-associated IL-l activity (24), and we confirmed that both forms of this cytokine were present in wild-type murine

582

t:Jr~LI'p.~,lf!1

",",,- j

.1'I'1~ II ml

.s,.r... .ll'

M1's., although !L-Ia was predominant OLe la, 87 1. 7 pglmg protein: JL.l~, 2 .1. 0.2 pg/mg 'protein), Blocking both forms of TL-I with neutralizing antibodies was necessary to fully bltmt FLS activation by Mrs (Fig,4E). We obtained similar results in the human system. Platelet MPs from RA SL~ expressed surface JL.la, which as in murine MPspredominatcd over IL-re (fig. S4). Similarly, human platelet MPs elicited by in vitro collagen sdmularion expressed both ll..-la and JL-W (19.1 versus OJ pg/mg protein) and triggered RA FLS to release 1l.-8 in a dosedependent manner, indeed, more robustly than either ll..-l~ or Wi< (Fig. 41'). Botb forms of IL-l participated in human FLS stimulation because neutralization of platelet NlP IL-l activity required blocking antibodies against both IL-I a and IL-l ~ (Fig. 4G). Together, these results show that platelet MPs likely contribute to joint inflammation via mechanisms including highly potent IL-l-mediated activation of resident synoviocytes,

These results provide an explanation for several disparate previous observations. Radiolabeled platelets localize to inflamed joints {25} and RA SF displays appreciable amounts of soluble platelet proteins (26), yet intact platelets are rare in arthritic SF, Whereas SF NIP eoncennations exceed those in the circulation of RJ\ patients by several orders of magnitude (18), platelet activation appears to be primarily all articular process wherein MEs dlssemlnaie platelet-derived cytokmes i-nto the arthritic joint. Subsyaovial capillaries exhibit fenestrations and are prone to enhanced permeabihty after stimulation (27, 28). We speculate that Circulating platelets contact ECM via these fenestrations when local conditions favor permeability, activating GPY1. Alternatively, platelet "sampling" ofECM via fenestrations may be routine. activating platelets as ECM constituents are modified. Thus, synovium is enriched for collagen type TV (29), and we have observed that FLS deficient in collagen type rv demonstrate

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

'10

A .

B

NP

c

o

I) lOD tOo .w 1OO1*UGol_IOOO IIC"'_Odr~

+l"\l·ll·,, _ .... __ -1.,' T1II1-lL-lo

... _"!I.ll.lnI~ ..... oIlMrnl

G 10 ,Coo 1_

bIod<Irro ,.!1lkl>ciodV (mal"'"

Fig.4. MPs activate FLS in an IVL--dependent manner. (A) MPs generated by tollagen stimulation of human platelets were coincubated with human as, and cytokine release was qua n tified by P roteome Profi le r, Da ta are re pre.sen til tille 0 f th ree independent expe ri men 15. (8) MPs isolated from RA Sf were coinwbated with FLS, and su p ern a til nts were assayed for IL·g release by EUSA (enzyme-linked imrnunosorbent assay). (() Mouse platelet MPs generated by coUagen stimulation of platelets from the indicated genotypes were coincubated with mouse Fl5, and supernatants were assayed for KC release by ELISA. (D) Mouse MPs generated by collagen stimulation (if WT platelets were coin~u bated with Il:-IR1-nuU (/11r1-1, FL5, and supernatants were assayed for KC release by EliSA. RecombInant TNF (10 ng/ml) wa~ added' to FLS to induce KC release as a positive control. (E) Mouse platelet MPs were coincubated with fLS in thepresence of I L-1-neutralizing a ntibodies, a nd supernatants were assayed fo r KC release by ELISA. (f) Potency of human MP stimulation of FLS. Fl5 were exposed to graded ccncentrations of IL -1~, TNJ, or platelet MPs, and IL-8 release '11<1'5 quantified in culMe supernatants by [LISA, (6) Human platelet MPs were coincubated with Fl5 in the, presence of IL-l-neutralizing antibodies, and supernatants. were assayed for I L -8 release by ELISA. Data for (B) to (G) .are the mean ± SEM of three independent experiments.

"'tL!III-IL-I .. (l1>1L-lll ..... , "I>IL 10111 ..,dl.,..m

I DlHk11l9 anflllQo(!t (!!WI!II)

reduced capacity to stimulate platelet MP release (fig. S5).

Ojhcr pathways linking platelets to arthritis await discovery. The effeet ofOPYl deficiency 00 gene-RIling NlPs and suppressing arthritis is incomplete, and IL-I is unlikely to be the only relevant platelet me-diator in the syoovial environment Further, we have observed direct binding ofMPs by SF leukocytes (fig. S I, 0 and E), an interaction whose functional consequences remain unknown,

The relevance of'these results for humsn disease is substantial. Because mice or humaus lacking GPVI remain healthy (30), antagonism of this receptor represents a novel therapeutic approach. The observation that membraneassociated MP IL-I is unusually difficult to antagonize (Fig, 4) may help to e.xplaiXl the limited effect of 1L-I blockade in RA (31), and the prominence of lL-lu within rvtPs poses potential constraints on the efficacy of lL-lf:} specific

REPORTS

agents. Ow- results provide compelling evidence that platelets play an -amplifying role ill. the pathophysiology of inflammatory arthritis, liberating proinflamrnatory MPs that represent tbe most abundant cellular element in SF.

Referenueo; and Notes

1. L p, Gartner, i. L Hi an, J. M. Stru m, BR5 Cell BioioBY and Histology (Up pi ncott WilLi" rns & Wilkr ns, Phllad~tphia, PA, ed, S, 20061,

.~. 1_ N_ Gl'Org~, Lan.el 355, lS31 (2000)_

3. G. Oavi, C. Patlono., /'l, Eagl. J. Med. 357, 2482

(200/).

4: C. G, Helm iLk et oi; Art hn:/js R/!eum. 58. 15 (ZOO S).

5. D. M. Lee. M. E. Wejnblatt, tnnce: 358, 903 (lOOt}

6. K. B. Pa,takio, N. E. Br0Wl1501'\, D. A. Terle, L. Ha!'\lalh, (;lil1. Mol. Pa tnot. q 9, M 17 (1996)_

7. P. /I, Valanl. W. )y, L. L HOTsm1~n, W. W. Mall, Y. S. A~n, Gr, ,. Haetrlato/, tOO, .24 (1993)_

flo S. Perel," Puj 01, P. H, IIMrker, tI. S. Key, Cylomelry il 71, 38 (200,)_

9.. P. Thiagarajan, J,F. T.ft,}. BioL them. 266, 2,4302 (1991)-

10, P. Wolf, Hr: J. ttaemato; 13, 269 (1967),

11. R_], Berckmol1s ft (1/_, Arthritis Rheum. 46, 2851 12002.).

W- R_ B_ Levene. E. M. Robellino/ Bload 67. 2m (1986)_

13. I. L joseph, p, Harrison, I. I. Ma{kie, D. A. Isenberg, S, 1- Machin. Sr; J Haemmol, 115, 461 (200 D.

14: D. KybIJrz, M. Con, Springer 5emin. Illlmullopalho[

ZS. 79 (2003)

15_ M_ (h~n el 01, Arthritis Rheum. 58, 1354 (2008)_ 16_ K. ShfF) el at, J. Imm!lno~ lsz, 647 (2.0(9)_

17. M._ taMamtl~ et at .. Nat MM. 11. 1167 (i100S)_ 13-. o. A. Knil,ff·(iutmer, I. Ko~rt>, R. Nieuwland,

E'. M. Ka Is beeH,alt> nbu rg. M. A, Vil n de lilaf. Arthritis RbeUm 46, 1498 (20Q2).

19_ D. M. Lee, H. p, Kiener, M. B_ Brenner. in K~U~y Textbook of Rheumatology (Sounders, St. LouiS, MO,

ed. 7, 20Q4), pp. US-laS.

20. B_ Nieswandtet oi., J. Cliol_ them. Z7 5, 23998 (2000).

21.. M._ MQr~j, S_ M. lung, IIrromb. Res_ 114, 221 (2004)_

22. P. A. Smethurst et al., J. Bioi. Chern. 282, 129'6

(~DD7).

23. D, WI. L~ e/ aI., S(ience 315, 1006 (2007).

24. S, Lindemilnn et al., j, (eIlIMI. 154, 485 (2001),

25. M_ Fan er al."Ann, Rheum. Dis' 42,545 (1983).

26_ M, H- Gin~berg. G. Breln., 1 L 5~osey, /lrf/iritis Ilhfll(li, za, 994 (1978).

27_ S- A Bfn,tadt et 01 .. , Nat. ImmMol_ 7, 284 (2006)_

28. H. R" SchumaCher Jr., Ann. (till. Lab. Sci. 5, 489

(;l,975)_

29_ P. Poduval sr (1/_, Arthritis Rheum. 56, 3959 (20071,

30. B" ~i~,w"ndt, 5_ P_ W,atson, Blood 102, 449 1;W03)_

31. M, Merlen), J. A. Singh. Cochrone Databose S,V$! Rev_ 1 •.

CD OOS 121 (2-009).

n. I'i na nd a I support lor thi! work wa;- provi ded by N) H grant POI AI06S8S8, en Arthritis Fou~dation Inv!l',iiga.tor Award, ~~d !he Cogan I'il!)'rily Foundiltion (D.M-U: an Arthritis Foundation Postdoctoral Fellow,bip Award (U.); NIH gran!., KPSAR051321 (PAN_) and HL091269 (E. R.~ 0 ,J: the B ri ti sh HeM! "oun dstlon il nd Medlc~ I R"5ea(n (oundl. UK (R.W.F'-1j and iIIationall{eilTl, lUng. and stood Institute grant l1l50545 (J.W.!. O.M, L., E.B., and J.W. have filEd" provi$ioo~1 pat~nl ;,ppliGitkm

for glyGojlroce1n VI., ;r t:argel for treatment for innommillmy il[1ori\i,

Supporting Online Material

'I/WIv _ ,ciencema.g . Drg/[gllmnt ell tlfu un 2 7/5 9 65/5 B DIDC 1 Mil teriats and Me tn 0 d5-

Fig,. 51 to S5

Table 51

R~feren(es

~4 5~pt~mber 2009; ~((~pted '8 DeCember 2009 1 O.l1l61,{j ence.11819 2 a

www:sClencemag,org SC1ENCE VOL 327 29 JANUARY 2010

58.3

REPORTS

Decorrelated Neuronal F'iring in Cortical Microcircuits

Alexander S .. Ecker,1.l·3 Philipp Berens,1.2·3 Georgios A. Keliris} Mafthias Bethge,1,2 Nikos K. logothetis/A Andreas S. Tolias3•5•6*

Correlates' trial-to"triaL variability in the activity of cortical neurons is thought 10 reflect the functio nal con neetivity of the d rcuit. Many cortica I areas are 'Qrga nized into functional columns. in which neurons are believed to be denseLy connected and to share common inpl!1. NUmerous studies report a high degree of correlated variability between nearby cells, We deveLoped chronically implanted multitetrode- arrays offering unprecedented recording quality to reexamine this question in the primary visual cortex of awake macaques. We found that even nearby neurons with similar orientation tu ning show virtue lly no correlated va riabi lity, 01,1 r findin g5 suggest a refinement of current models of cortical mi erocircuit architecture and function: Either adjacent neurons sha re only a few percent of their inputs or, alternatively, their activity is activ~ly deccrrelated.

Co.n. '.eI8led re.SP.611S. 'e flu.ctuations am .. ong simultaneously recorded neurons have

been observed in 'a number of cortical areas (1-14). TI,e prevailing hypothesis is that these correlations (referred to as "noise correlations") are caused by random fluctuations in the activity of neurons presynaptic to it pair of cells (5, 6, 15 f 7). Noise correlations nrc reported to be particularly strong in nearby cells with similar response properties (5 j 3, 18), which supports the idea that nearby cells within it funcrional column are densely connected and share a substantial amount of common input (17). Theoretical walk shows that noise correlations with such a '%nited-rnnge" structure are particularly detrimental fOJ" population ceding (5, jfJ.~21). Thus, knowledge of the precise nature of noise correlations can advance out understanding of cbs structure and function of cortical microcircuits in vivo,

Although the prevalent finding of spike count correlations in the range of 0.1 to 0,3 (2-JJ) seems to suggest that their magnitude and cause b a v e been fumly esta blish ed, the re are. se veral technical challenges in the measurement of noise correlations, Spike count correlations Call be gencrated in the absence of shared presynaptic noise by a number of'taetors: First, it is difficult to control for internal variables, which modulate firing rates, such as motor plans, or cognitive states, like attention. Second, recordings fh,)m electrodes that are not chronically implanted often suffer from instabilitiesin tbe electrodes' positions. Third, if multiple cells are recorded from the same electrode, suboptimal single-unit isolation is a concern (12). Fourth, In experiments conducted under anesthesia, con-elations may arise fiom spon-

1 Ma x Plilnt k I nst it ute for B i 0 logi C ill (yb ern sties, 72. 076, Hibi O1g en, Germ a ny, '-Cent re for 111 leg ra tive Nell rosd ence .and Institute to r theoret iea! Phy>i (5. U ni versity of 1 (J bingen. 72076 TiibfngenJ Germ.:my. sD€pilrtmeni oi Neuroscience, Baylor Co!leg e of Medici ne, HOIl sto Ii; TX 77 030, USA, 'DivisiQn of Imaging Sdenr€ and BiomediG<JI El1gin~~ring, U11ivsrsi ty 01 Mil n diester, Manchester M 1 7 H L. UK. 5 M i d1ileL E. DeB a ~ey Wl~( ans Alfai rs M~ di cal Cen ter, Houston, 1)( 7 7030, USA. 'Depe rtrnen I of Corn pll ta If onal and A ppli ed M <I Lh E'-rnatics, Rke Ullil'l'(Sity, Houston. 'IX n005, U5A

'T o whom corresp and ~nce sh au ld be add ressad: oliO I ia s@ tn,.bem.edu

584

taneous oscillations that rue absent in behaving animals (21). Given that all these factors will artiticially increase estimates of noise correlations, it is important to control for every single one,

We reexamined this issue and measured spike count con-elations by using- aJ1~yS of chronically implanted tetrodes (fig. Sl) to simultaneously record the activity of local groups of neurons in the primary visual cortex (area vn of awake monkeys (24,25), Tetrodes provide a superior qut1lity of single-unit isolation of nearby neurons (20) compared with conventional single electrodes 0;: rigid rrmltieleclrode arrays. An example of a ten-ode isolating multiple cells Is shown in Fig. 1. Because

A

c

o

:l.~_.,.. g40~

.$O~

~20~

:~ ol=:-·~

u,

10[

0-'1: .1\. .

o 90 180270

Direction of motion

the signal is recorded simultaneously by four adjacent mierowires, the Jocation of the neurons can be tri angulated, resul ling in distinct cl ustcrs, each representing the action potentials of a single neuron (Fig. 1, A and B). It: for example" only chanue14 had been recorded (as could be the case with a single electrode), eel Is L, 2, 4 and 5 would have been nearly impossible to distmguish. For our analyses, we only considered cells that were quantitatively detennined to be Vel), well isolated, in this case discarding the second neuron, which had ~8% falsely assigned spikes (25). Cells recorded on one tenode bad highly overlapping, receptive fields (Fig. I C, colored outlines). We presented sine wave gratings drifting in 16 directions of motion perpendicular to the grating orientation (Fig. lC). Consistent with the columnar organization of VI, three of the four neurons had very similar preferred orientations (Fig. ] D). When we examined the spike count correlations (1'"") of the six pail's, they were extremely low (Fig, I E), 'With an r..., average value of 0,D2.

We collected data m a total of 46 recording sessions from two monkeys (0; 2'7: H, 19). Gratings were presented at eight different orientations and were either static or drifting in the direction orthogonal to the orieotation, The gratings were large eoough to cover the receptive fields of all neurons recorded by the a[ray (Fig. Ie). Spatial frequency and speed were chosen such that a large number of neurons was driven,

B

I toO~JV

1 ~.f'-,,-~Jl,

~

contarn.: 1. 7"10

0.4%

2.2%

0.20/0

E

4

"i\' (,

~

• '"~'';ji.;r

• 0.21

..... _ ~ .. :'

.~;\

•

~ --41 0.04

8

":[ :~;:~

I •• -0.14 --4-----

-4 0 4-4

·rr~~

~~~,.

. '

, -0.01

o

I • ~ "

'l~'ii: .

• 1~~"

• 0.02

o 4-4

-001

o

4

z-score

Fig., 1_ Example of five single units recorded from one tetrode. Colors a re, matched in all panels. (A) SCijtter plots shoWing amplittrde of first principal component nf spike waveform5 for all pairs of rhannels, Clusters are modeled as multivariate GalJssians, which allows quantification of their separation. (B) Example waveforms of multiunit (blac.k) and the five sinqle units {colored}. Each row corresponds to one tetrode thannel, Estimated fa15e-assignment rates (25) are shown below each column. Nellron 2 (orange) is.disca rd ~ d be ca use of ins uff den t isola ti on. (0 Grati ng sti m uius ave rlaid wi Ih re te pti VI" fie ld outli n 1"5 of 24 simu[taneou~ly recorded neurons. Red dot fixation spot. (0) Tuning wrves. Error bars are SEM. (E:) Scatter plots of z score-transformed responses for all pairs obtained from the four neurons. r" values are indicated. Pair identities are coded by colored dots.

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

but not optimized fbi" any specific cell. In total, we recorded 9 L 7 single units, After discarding cells that were not well isolated (>5% falsely assigned spikes), not visually responsive, Or not tuned to orientation, we obtained 407 (O, 262; H, 145) single units. This corresponds 10 1907 CD, J 3]5; H, 572) simultaneously recorded pairs, in 406 of which (D. 361; H, 45) both neurons were recorded by the same terrode;

Neurons recorded from OTIC tetrcde are Pl1YSically dose to each other, have highly overlapping receptive fields, and are believed to receive strong common input. N evertheleas, spike count correlations in pairs of neurons recorded by the same tetrode were exceedingly low (r...., = 0.005.!. 0004; mean z SEM) (Fig. n Even cells with Similar preferred orientations (I·"'gnal > 0.5) had very weak correlations (I'", = 0.028 i. 0.0 I 0). This also held if pairs were snnngly driven by gratings with orientations dose to the- cells' preferred orientations. Under such stimulation, spike count correlations were not larger than those under stim-

A

0.4
0.3
0.2
0.1
JiE 0
-0,1
-0.2 11=406

. .

-0.3

-0.4

L- ~ ~ _

-1

-0.5

o

0.5

ulation ",ii1h less optima! gratings. (I;"" = 0.021 1. 0.013 versus 0.016.!: 0.0 I l , two-sample I test: p= 0.80, n = 361 J. Only -14% of all pairs with cells recorded from the same tetrode bad correlations significantly different from zC!'O (a = 0.05; for r"':p"L"'" 0.5: 13.2% positive. 2.4% negative; for r.lgc1ll < 0,5: 5,9% positive, 7.4% negative). Theoretical censiderarions and numerical studies indl~te that much of'the scatter in the di'lmbution (Fig. 23) lUUy result from estimating correlation coefficients from finite d ala (figs, S 2 and $3). Even though there were cases where similarly tuned neurons Were correlated, these constituted only a small minority of pairs. Under our experi- 11lental conditions, spike count correlations fur local ensembles were smaller [han previously reponed by more than an order of magnitude (Fig, 2C)

Previous studies report 'high correlations also for similarly tuned cells recorded on different electrodes separated by up to several millimeters (7 11). Therefore, we analyzed all simultaneously recorded pairs, including those pairs where the

B

c

0.4 0.3

Pr'l;,vious stUdies

• V1

~ MT OIT

0.4

Ol.!:r data r~I!I""I" Q 5 r''l1''.1 -c 0.5

0.3
01
0 A <S>
:0 0.2
Y
-0.1
~
0.1
-Q .. 3 - Sampled
-QA -Withjiller 1:3
0 a
0 0.2 0,.4
Fraction Fig. 2. Spike count correlations of pairs of neurons. recor.ded by the same tetrode, (A) Relation between r"gnat and r" for all pairs of nearby neurons {both neurons reco rded by the same tetrode), (B) Distribution ofr,e (mean r SEM = 0.005 ± 0.004). Colo red lines are distributions obtained by generating a rtificial datil with the same number of trials as in the experiments (red. fixed f" = 0.01; blue; av,erilge fsc_ = 0.01. and S.D. 0.1), which indicate that most of the scatter in the empirical distribution is due to estimating correlations from finite data [see supporting online material (SCM) text sections 1 and 2 for discussion]. (e) Average ro;c compared with previously reported values [bladtsymbols,. V1 (3, 4,7-9), MT (5.6); IT (2)]. Celts with similar Hming (["gnat Qf >0.5) have slightly higher correlations (r" = 0.028 ~ 0..010) than dissimilar (ells (r" = -0.001 .L 0.004, two-sample t test, P = 0..002, n = 406), but flc is an order of magnitude smatter than in previm)5, reports.

A

B

C
0.4
0,2
_lll 0
-0"2 ,
-0.4:
0 . r- ~ .,._.._... ~ _._.. '*

o 250 ;;00 0 250 500

Tlme:(ms)

..MJ "wi ,. 7 t

REPORTS

I

two neurons were recorded by different tetrodes. Average spike count correlations were low (rsc = O.OlD ± 0.002, mean ± SEM) (Fig. 3, A and B). There Was only Ii. weak relation between tuning similarity mid spike count eorrelarion (two-sample t test, rMgnnt «; 0.5 versus ':'\i!J",l1al > 0.5: P = 0.003, n = J 90.7) (Fig, 3C), and even similarly tuned cells had an average correlation close to zero (r.i!!,.Q.J > 0.5: r", = 0_0;!3 .L 0,00.5, rnean z SEM) (Fig. 3C), Correlations did not depend on the distance between the two neurons (linear regression stope: P = 0.99; two-sample I test within versus across tetrcdes: P '" 0.16) (Fig. ]0). Although there was a weak relation between two neurons' average firing nne and their spike count

A

B 014

0.4 n" 1907

(1,2
-~ 0
-0.2
-0,4
-1
C
0.03
~
W 0,02
ui
4<1
~ 0001
.....,. -0.,2

'- -0,4,- _

0.5 0 0,2

Fraa~on

-o.s 0

D

n'------

'-0,5 :>0.5 0 0.2 0,4 0.8 HI 3.2

f'Illn.o' Distance (mrtl)

Fig. 3. (A) Relation between r slgn.1 and r sc for all pairs of>imu(ti!nePllsly recorded neUfOfl~. (:B) Distribution of ro< (mean ± SEM = 0.010 ± 0.002). 'C) Cells with similar tuning (r <igna\ of ::-0.5) have slightlY higher correlations (r st = 0.023 j_ ().005) than the remaining cell's (r« ~ 0.008 .i 0,002. two· sample t test, P = 0.0.25, fI = 1907), (D) f", values do not depend on the distance between neurons (linear regression: P = 0,.99, n = 1901), The bin at zero contains neurons recorded by the same tetrode (Fig .. 2). Error bars show SEM.

o

E

0.02

n.s,

::ii O.IJ!

w

ui

11' 0

_..,.....

.... ~

V" -0.01

-0.02 L...-_~_ <05 >0,5

Fig. 4. Spike count correlations for natural images. (A) Example of the stirn· ulus with receptive. field outlines of 13 simulta neously recorded neurons. Red dQt.· fixation spot. (B) Peristimu[u5 time histaqrarns of t~[ls in (A) for this stimulus. In each triai, the stimulus was flashed four times fOf 200 ms, with' 50'ms pauses, For consis~nC)1 with the grating stimulus, spike coun!!; were computed

RF dlsl~nce {oJ Fr.acllon RF dlsl. (~)

from the first 500 rns (two stimulus flashes). (() Relation between receptive field distance and r", is not significant (linear regression: P = 0.11, n = 329). (D) Distribution 01 f<e (mean .l S EM = 0.001.1 0.005), (E) Avera.ge f" for pairs with overlapping re(eptive fields versus pairs with nonoverlapping receptive fields. Two·sample t test P= 0.89, n = .3 29.

www.sclencemag.org SCIENCE VOL 327 29 JANUARY 2010

585

REPORTS

(r.,)= 0.12 <r~) =- 0.01

f"" =0

----

1000 1445 2000

3000

NUfl1bElr of neurons

Fig. 5. Impact of correlation strength on encoding accuracy of neural populations, The Cramer-Rae bound (minimum achievable decoding error) is shown for different corre[atio n structti res: a previous rep' ort [blac k, refs, (5, 6n, our data (red), and ani nde p endent population of neurons (black dashed). Dotted lines highlight the number of neurons that would be necessary to achi~ve a decoding error of 2". For details, see fig. 58 and SOM tex\- section 6.

correlations, this relation arose on time scales longer than one trial and therefore appears to be unrelated 10 shared presynaptic noise (fig. 84).

To investigate whether low correlations also o ceur under more naturali slit stimulu s condition s, we. conducted additional experiments io otle, of the monkeys, (H). We first mapped the neurons' receptive fields before presenting natural images (Fig. 4, A and B) (25). The average rsc was close ill zero (1"," = 0.001 l: 0.005, mean :t: SEM., onesample I test: P= 0.89, n = 329) (Fig.4, C and D); with 00 relation between receptive field overlap and spike count COITe1ati01]S (linear regression slope: P = O. 12, n = 319) (Fig. 4C). Neurons with receptive fields within 050 of visual angle bad spike count correlations similar- 10 neurons with more distant receptive fields (two-sample ttest:

P> 0.'13, 11 = 329) (l;'jg. 4E).

We recorded another 56 pairs ofneurons while a third monkey (8) was presented with moving bars .. As with the other sti111uli, spike count correlations. were dose to zero under these conditions (0.014 .L 0.011, mean 1: S,E.M, P= D.21, n = 56) (fig. S5).

Under II variety of stimulation conditions ranging :&om classic stlrl1uli (sueb as bars and gratings] to natural images, spike count L'OITelations in the primal)' visual cortex of awake monkeys were extremely low, These results stand in contrast to a number of previous studies. which report correlations of the order 0_1 ro OJ. Above we suggested four factors that could lead 1:0 spike count correlations unrelated to shared presynaptic noise {l7). We now demonstrate how artificially introducing anyone offhese into our data produces correlations similar to previo usty pub lished results. First, uncontrolled external or internal variables ean be studied by assuming the neurons.' firing rates are gain-modulated by a common underlying process. Shared modulations of only -15% can lead to correlations on the order of 0.2, as

586

shewn in fig, S6., It is extremely hard. if not impossible, to control precisely the effects of at, teetional state, reward expectancy, task -solving strategy, Or other cognitive factors (28). In con" tmst to extra striate arms like V 4 or MT, area VI is much less affected by such modulations. Se0- ond, slow drifts over time or abrupt movements of the electrode tipcan lead to changing waveforms and, thus, lost spikes or increased contamination by multiunit activity because of decreasing signal-to-noise ratio. Because movement is likely to affect 011 neurons recorded by aile electrode, it can. be modeled as a common gain modulation, and the aboveargumems apply, Our recordings were extraordinarily stable, as demonstrated by our ability to track neurons over several days (24). Third, contamination of waveform clusters identified as single units by spikes of other cells. can create artificiall yhi gil ccrrclations and L-.lTI even give ri se to or amplifY the Iirnited-ra u ge correlation smrcture {fig. S7 shows that -10% false assignments during spike sorting can produce correlations of order O.IJ Fourth, during anesthesia, till and down states or even subtle variations in the level of'anes(hesia will inevitably cause changes in firing rates common to many cells (analogous to point I but on a relatively rapid time scale; see also .(29)], potentially having a stronger impact on nearby cells. Thus, any meaningful characterizarion of the impact of noise correlations on population coding critically depends 011 the ability to obtain stable recordings from large populations of well-isolated adjacent neurons, ideally in an awake animal and in a cortical region like VI., which is not modulated strongly by variables that cannot be precisely controlled. Interpreting spike CQWlt cerrelarionsin tenus of their effects on encoding oapacities of cortical microeircuits or drawing conclusions about fuoctiona i connectivity only makes sense jf one can separate covariability because of unconrrclled variables from that reflecting intrinsic noise in toe circuit.

Our findings have implications for models of cortical circuit architeerure, The current view on the generation of correlations in cortical circuits rests on two major assumptions: (i) nearby cortical neurons receive a substantial amount of common input (6,17 .. 30,31); (ii) such common input leads to correlations (15 17,32). In light of'our data, al least one of these assumptions cannot be correct.

Based 00 measured spike count correlations, an influential modeling study inferredthat, on average. nearby cells share up to 30"/., of their inputs (l7)c Under the same model. our data suggest that at most, 5% of the inputs are shared. Note that anatomical studies report -10% common inputs fur excitatory nCW'OilS (30, 31), In addition. cortical excitatory connections may be very precisely structured (33) to form many independent subunits. In this ease, most recorded pairs consist of neurons belonging II') different subunits, and average correlations are very low.

Assumption (ii) has been challenged by recent network models in which a dynamic

balance. of exci ta tory al1 d j 11 hi bitory fl uctuati on s counteracts correlations induced by common in, puts (29, 34). This results in correlations tbat are positive On average but very 10'111 (--C.,OJ), a prediction in good agreement with OUl" data. To prevent small correlations u'OlTI accumulating and dominating network activity, such a decorrelation mechanism might be a crucial prerequisite of hierarch (O!i] cortical processing,

Whatever the mechanism behind the decorrelated state of the neocortex, it offers substantial advantages fur irrformation processing Consider a downstream 'neuron reading out the orientation of a grating from the actiVity of VI neurons. If con-elations were ..;Q.t2 on average (5, 6), the runnber of neurons necessary to achieve 20 precision (mot mean square error) would be five times larger than those In the scenario in which the average correlations m-e -(l.OT {Fig, 5 and fig, S8). Moreover, it is unclear whether neurons have access to the correlation structure of (heir synaptic inputs. If the network is in the decorrelated state, however. the effect of not taking any remaining correlations into account is small, and decoding is greatly simplified.

Referen<e5 and Notes

1. M. B~ch, I. Kruger, Exp. Brain Res. 61, 451 (19B6).

2. 'r.j, Gawne. B.]. Richmond,). Neuro5d. 11,2758 (1'193),

3. 1. 1 UaWll".1. W Kia"" I. A. Hertz. B. 1- Richmond.

Cer-eb, Colt ex 6, 48Z (19 9 6).

4. D. S. Rei.h, F_ Mechler, ). D Victor, Science 294, 2566 (2001).

S. E.. Zoh~ry. M.. N. Sh~dlen, W. T_ N~wsQn\e, NdWr'e 37'0, 140 (1')94l.

6. W. Bai" E, ZohaJjl, W. T. Newsome, J. Ne~rosci_ 21, 1676

120(1).

I. A_ Kohn. M. 11_ Smith. I Neuill~tr 2S. 3661 (2005) a. D, A. Gutni,ky, V. Dragoi. Natu(e 452.,220 (2008). 9. M. A. Smith, A. Kohn,1- NeurOSfi, 28, 125n tWOS).

10, M. R. Cllhen. W. T. New!mne, Neuroll 6{), 162 (21)08). 11, X. I1l1"ng, s. G. lisberger, J. N~ujo{)~y!iial, ~01, 30~2 (2009).

12. D. lee, N, L. POri. W. Kru,e, A p, G~orgopoulos. ). Neumsd. 18, 116'1 (199&).

1,~. L (b!i.lt;llitinidi~, p_ 5. GDldm~n-R:.~k, J- NeUrophf.;iat S8., 34a7 (l002).

14. B_ S, Averbeck, D, Lee. j. Neurosr:i. 23, 7630 (~003). 1,~. H. I,_ Boy" ntjr; A_ R. Mareos,]c P - S.,gundo, J- Neurophy;/tlL 36, 20S (19,73)

16. A. oM, Aemen, G. l., Gemfi~, M. K., Habib, G. Palm, ). NeurophY3101. 61, 900 (l989).

17. M. N. S~adlen, W_ 1 NewiQn\e,], NeVr'a$<'i. 18. 3Bl'~

(;l,99S).

18. B_ B. A.v~rb~ck, D. Lee,}. Neuropby.!io/. 9)}, 36~3 (2006). 19, L. F. Abhmt; P. Dilya n, Neutal (olniJut 11. 91 (1,9991. 20_ S D_ W;lk c. W. EU~(~, Neural (anlput. H, 1~5 (2002). 21, H. 50ll1poiinlky. H. ~oon, K. Kang, /III, 5hamir, Phys. Rei:

E SMI. Non/in. Soft Mdtter ph!". 64, 051904 (2001).

22. K. D. H a rris, D. A. Henze, J. Csi (%Iii, H, Hir.,., G. BUlS.lki.

1. N~U"9P/lysio[ 84, 401 (2000).

23. j. F. Poulet. C, C. Petersen. NatJJre 454, aar (2008). 2:4. A_ S. Iclias et al., J Neurophysio!, 9S. 3780 QO(7). 25 Materia15 ~~d methodi are ~vailable ill supporting

material on Science Online,

26. C. M. Gray. p., ~. Maldonado, 1\1. Wilson. B. McNaughton, j. NeurQsci- Methods 63, H (1995)_

27. I\hhoug.h many studies were i" VI U, 3, 4, 7-9), for sorn g s IlJdi.,. aT ea d illerene., could pl".~ a rnlB

(l, 5, 6. la-B). In addltion, dltfell'nt,re(ording method I m; gh't be biased towa ~d rg(Ordi n g certa 1 n types of cells. However, tetrodes have atso been U5~d before (4).

2B,. In addition, miniature BY"- movements are known

to rna duia te ~ fin g ra tes a problem that is a mpl iii ~d if

29 JANUARY 2010 VOL 327 SCIENCE www.sciencernag.org

'li'Ioli stimuli, Jiar~ly' ~oV~~1J9 the. neuron~' reo;p~Ve. 1i~ld.l, are used,

.29_ A, Ren;lrI et ot, Stience H7, 587 (W10).

30. V. e ra ile~betg, A. 5 d1 (il, Anatomy of the (000; StDti~ti,:; and' Geotri~tfY 15 (Iii ng~r, Bertili, 1991)

31. B. Hellwig, A. Schili, A.. A~m€n, Bioi, CYbem. 71, 1 (1994).

32_ J. de la Ro(~a, B. Doiron, E. Sheil- B !Own, K. )osie, A. Reyel, Na.ture 448. 1l~2 (2(IOJ)_

33. y, C Yu, It S. Bulfje, X. Wang, S. H. ,hi, NlffOre 458, SOl (2009)

34. l Hertz, Neural CompuL published online 20 October 2009; 101l62/~~lO}OO9,06-0S-S06,

3 5. We tha n R M. Su bra man ian, A. H uansel oil r, T. J Willi ford, and D. Murray for help with experiments: R. I: Cotton for h iscon trih u ti 00 in setn ng up record [0 g "qOi pmen t;il od the Siapas laboratery at CilltEl.ch for providing lhe

Ie wrdi;, 9 software, We thank, R. J, i;:i1\ton, P. !)lIy~p, S. Deneve, K. n Harris, f. v. tobenov, W. J. Mar

J. H. Macke, A, Reoilfr, I. de ta Ro(h~, A, G_ Siapos, and 5. M. Smim~h, for comments on fll€ manuscript and di~(u\'ion5_ This WIllI. wa. supported by ,he National Eye tnstltute, NI H (ROI EY018847), the M~X Pl;mlit Society, the V_S, Department. of D~t·.n;e (WB1XWH-OS·2-0i4V), " VA Merit Award from the Department of Veterans AtfaiT5, and an Atpald and Mabel B~ckman Foundatien Young

REPORTS

I nv"s tig~ tOT Awa rd tc A.S ,1, and by the. German F"d"f1I ~ Mini.try 0 t EdlJ catioo and Resea r(h (BM B~) t hroujjh t~e Bemst€l n awar d to M. B. (F KZ'O t.GQ06 0 I)_

Supporting Online Mi1terial.

www .s de neemaq, orgkg lim n ten tlfu IV) 27/59 55'/584/0Cl Mil lenals and Meto 0 ds

'10M Text

Figs, 51 til 59

Referenci',

29 July 2009; a (Cl'p~ed lODe (ember 2 OM 1~,1126/~oen(e.11nB67

The Asynchronous State in Cortical Circuits

ALfO!l~o Ren<!rt?*.t Jai.me de La Rocha,'l,2* Peter .Bartho,l,'3 Liad HoLLender,l Nestor Pargil:iI Alex Reyes,2 Kenneth D. Harri'S1.5t

Correlated spiking is often observed in corticaL dreuits, but its functionaL role is controversial, It is. believed that correlations are a consequence of shared inputs between nearby neurons and could severely constrain information decoding. Here we show theoretically that recurrent neural networks can generate an asynchronous state characterized by a rbitrariLy low mean spiking correlations despite substantial amounts of shared input I n this state, spontaneous fluctuations in the activity of excitatory a nd inhibitory populations accurately trtlck each other, generating negative correlations in synaptic {:urrents which (antel the effed of shared input Near-zero mean rorrelations were seen experimentally in recordin.gs from rodent neocortex in viVo. Our results suggest a .reex!lrflination of the sources underlying observed correlations and their functional consequences for information processing.

The spiking activity of neurons, is often correlated within local cortiCal, populations (1-4). Although correlations could be a signature cf.ecrive information process tug (5, 6), they can also impair the estimation of informatioa conveyed by the firing rates of neural populations (7, 1, 8) and might limit (he efficiency of an organism for performing sensory dis-eliminations (7, 2). Under special conditions, correlated spiking is an inevitable consequence of shared presynaptic input (9, W). In general, bowever, the overall contribution of shared input 10 correlation uiJJgnirudes tneasured ill vivo is unclear, as measured, eorrelations could reflect mostly COVID;, iations in activity due to cognitive or external variables outside the control of the experimenter {J} -13). To investigate the relation between correlations and shared input, we studied theoreti cally the eoirelation structures characteristic of densely connected recurrent networks.

We start by considering bow t11C eorrelation between a single neuronal pair depends on the

1 Center for M olew lar and Beh aviora! Neu ro science, Rl! tgers u 11 iversi ~y, N ew~rk c N] ono 2, USA. 'Center for 'Neur~ I Sdenre, N elY York U niversi ty, N ~W York, NY 10003, U.5A. 31 nstiMe of fxperifTIent~t Medicine, Hungarian A[a(jemy 01 5 nences, Bu dapest 1M3, H II og a ry. • Depa rt a mento de Ff sica Tt6ric;<l} Universidad Auto!Jorna de Madrid, Madrid 28049, 5 pai n. 'SmilQW Relea((h C1!nfef, New York Un iversiljl M~di cal 5 rhoo\, New York. NY' 10016, USA.

'The~e author'> contributed ~quaUy 10 this WQlk.

tPr~lent address: Department> iiI Bioenginee~ng and Et~triGll and Electronit Engin~ering, lroperi~l College, tond on SW7 2Al., UK.

1:,To whom correspondence should be addressed, E'mail: ar€ Ilart@il nd rome da. ru tg ers, ed \1

B '~ C £II1PIJI:s"OI"Ily'
to ~~"'
0
E 0.5 .. i.' ...... -/ V' -'"
Qi
tr
0 ~ --:S;_'" __J."'"
u l-J lllJ "mY
o ' ,.. ~~~_: _ ~-:ru.-......r"
0 0.2 0.4-
Shared in put iraGlioll P
E F E arc! J jr"lpl..llS fraction fJ of shared inputs and the degree rio to which the. inputs are themselves correlated. The effuct of shared input com be isolated by considering presynaptic neurons that fire independent-

A

~~ ~~ nO

i: > 0

o V~1Ii\..

II c> 0

V~J'v._ ,

4_':I'+t.N

'I .c> 0

~,/1 !r;r

[5 ~

Qi 0 .. 5

5

U

ly (ljn = 0). Both excitatory (E) and inhibitory (1) S11Jred inputs Gause positive correlations of a moderate magnitude in the synaptic input and spiking activity of the postsynaptic pair (Fig. I, A and B) (? 14). Spiking correlations rll;) between inputs, however, have a major Impact on the output correlation to;mt of the postsynaptic: pair. When all inputs are E, weak input correlations give rise to stro(Jgly correlated synaptic currents and output spikes (Fig. IC), This occurs because, when p and rill are small, the correlation c of the two input currents is approximately equal to

(1)

u '.f\II.._

~\rllc<o

0.2

In put spl kl ng corre I anon r"

(15), wbere N is the number of synaptic inc puts, resulting in a large gain in the relation between rin and r,;h! (rtg. 1 E, upper solid curve). The situation changes when both neurons receive I as well as E inputs. Correlations between E or between 'neurons lead to strongly cor-

0 ...

Fig. 1. Effect of shared inputs and correlated inputs on output correlation. (A) 5haredexcitatory I.E, green) or inhibitory (I, red) inputs induce positive correlations in the synaptic currents 01 two cells (c > 0). (8) Corr~liltior:i coefficient of synaptic currents c (dilshed line) and output spikes fOUl (circles, (OU nt window 50 ms) ofa postsynaptic pair of integrate"and·fire neurons as a function of the shared input fraction p (21), Earh postsynaptic cell received Nl; = 250 Poisson input spike trains. {el Input spike raster (top),. synaptic currents (middle), and membrane potentials (bottom) of a postsynaptic pair remiving weakly correlated £ inputs [black circle in (El, fin = 0.025J. (D) Whereas correlations between £inputs or between I inputs contribute po5itively to c, correlations between E and I inputs have a decorrelatinq effect. (E) Correlations .c {dashed Line} and rout (circle» as a function of the rnpl)t spike correlatkm (rri al fixed p = 0.2. E inputs only: Each cell receives N~ = 250 correlated Poisson spike trains (211; E and I input5: Nj =- 220 inhibitory input trains were added with identical statistics and correlations. (F) Same as (0 but for the Case with E and I inputs [bLue cirde in (EJ, tin = 0.025]. f and t currents are shown separateLy fr-om the total currents (b[ackand gray). A5terisks indicate Large fluctuations in the excitatory and inhibitory currents that occur simultaneously.

WWW.sclencemag.org SCIENCE VOL 327 29 JANUARY 2010

587

REPORTS

related excitatory and inhibitory synaptic currants (Fig. IF, red and greet! traces). However, when E and 1 inputs are fhemselves coirelsted, large fluctuations in the excitatory and inhibitory currents occur simultaneously (Fig. I F', asterisks) and cancel, which leads to a significant reduction in the correlation of the total synaptic currents o and output spikes (Fig. IF, black and gray traces). Correlations between E and T inputs thus decorrelate the synaptic currents to postsynaptic neurons (J 6).

To investigate whether such dccerrelation can arise spontaneously from the dynamics of a recurrent network, we characterized the behavlor of correlarions in a simple rCCUl1"Cot circuit of binary neurons U 7-/9). The network consists of two populations (ofsize N) of E and 1 neurons connected randomly, both receiving excitatory projections from an external (X) population of N cells (Fig. 2A). The network has two key prop .. CIties: First, the connectivity is "dense" so that the coonection probability p (and thus the mean fraction of shared input) is fixed independently of the network size [e.g., [.I = 0.2 (20) as in Fig, 2). Second, the synaptic couplings are "strong,' such that only a small fraction of a cell's excitatory inputs. is enough to evoke firing (Fig. 2B); in the model, although the average number of ioputs is proportional to N, the number of E inplJ~ needed toinduce firing is 0111y proportional to ..IN (19). Our analysis showed that, even in the presence of shared input, the network settles into a stationary state in which the population-averaged firing COt· relation r is weak, if inhibition is sufficiently strong and fast. It'i tact, in networks of different sizes, T" dcereases jn a way inversely propertiona I to N (Fig. 2C,. open squares) [(17) section 2.3], a Signature of asynchronous networks (18). In an asynchronous state, the variance of the population-averaged instantaneous activity scales ill the S<lU1e way as if the neurons were completely independeut=es u». Thus, correlalions in the asynchronous stale do not qualitatively constrain averaging of activity across neural populations (fig. S2) (18)"

Asynchronous activity persists in the presence of shared input because of a spontaneously generated tracking of fluctuations in the populatioo-avereged instantaneous activities m"d,t) and IIIIJ) of the E and incmons (Fig. 2D) [U 7) section 2.4]. Specifically, mAl) tracks md:l) with a smalllag (El-Lag), and" they both closely follow the external instantaneous activity 1n,1'(1). In larger networks, tracking becomes mOT<;: accurote and becomes perfect in the large N" limit;

where .Ile and A) are constants that depend on the network architecture. Tracking occurs because, when the connectivity is strong and dense, even small "random" fluctuations in instantaneous excitatory activity, of order 1/..JN, are large enough to recruit inhibitory feedback

588·

A4~

x

P /

B

c

·2e '----------

1000

1000

Tims (ms)

e

.H"lI

~~-6~~~~~?

1)-'

c;

,g

'"

1!! 0.01

8

100

1.000

N alwork size N

~

--

F Asyn c~ renous Ilrlng

r -l/N -

"Y '\

WsaldY,Bsynp1'IuJflol;JS- Syothronoi..js

lo'tal C'urrent-s current. compon-e I'IIls

c,. wfdlh -!fIN "ff.C, •. C·",.c",

~

N.,[)2' G

E
o..~
tr
.9
19
~
8 0
C'
~
U
-(l,4 -10 0 10

Lag (ms)

..Q.03

o

Firing correlation f

Fig. 2.. Asynchronous activity in a binary recurrent netv.rork. (A) Schematic of the network architecture. The shared input fraction is p. (B) Strong coupling produces irregular spiking activity because. of a dynamic balance between the large excitatory (£ and Xl and inhibitory (I) currents to each (eLL (19, 22). Dashed line represents threshold. (e) Population-averaged correlation rceffirients of the firing activity Vi open squares); total current (C, filled 5quares);and current components versus network size N. Dashed Lines show lJ.JN and 1JN scaljnq for cQmparison. (,D) Instmtaneous populatton-averaged activities (transformed to z scores) of the E, I, and X neurons showing that tracking becomes more accurate with increasing N. (Insets) Instances of the lag between E and I activities (£I-Lag). Color code as in (B). (~) PopUlation-averaged CCGs of the current components (N '" 8192). Color (Ope as in (C). (Insets) Magnification of the peak of the IE and EI ceGs (bottom) shows that the EI-Lag decreases with N, which leads to the decrease in the magnitude and width of the totaL current C::CG (top). (F) Description of the asynchronous self-consistent solution (see text), (G) The histogram of firing correlations in the netYl'orll (if pairs; N '" B192) is wide: Cfr» i,

Tracking of the instantaneous population 3Ctivities is equivalent to a precise cancellerion of the dtfferem components of the (zero·:rog) populationaveraged current correlation c (Fig. 2EJ [(11) section 2.5]. Because the synaptic CUl1"ent to each cell consists of an excitatory and an inhibitory component. the average current correlation across Dell pairs, c, can be decomposed into eEl';, elf, and Cf!J Cthe tenn CIE. = cm). Both CRE and en are positive and large, i.e., independent of N for large networks (Pig. 2C, colored squares) because of amplification of weak tiring correlations and of shared E or J inputs (Fig, 21', i; and Fig. IF, red and green traces). However, eEl is large and negative because of correlations between E and I cells generated by tracking, which leads to the cancellation:

c = CHI! + elr + 2cEl - 1/1N (3)

(Fig. 2C,. filled squares; 2 F, ii: find fig. S3). Even a:fter tbis cancellation, bowever, the instantaneous current correlation c is still larger than the-correlaticn to tiring r (fig. 2C, filled and open squares). This is- possible because neurons in-

legtate their inputs over time, so that the instantaneous correlation r is related to the area under the current eross-correlogram (Ceo). Because tracking becomes faster fur llIrge!' networks. both the width of the CU11'ent COO (effectively set by the El-.Lag) and Its magnitude decrease as II,jN (Fig. 28, insets). Its areais thus liN. as required for asynchronous firing (Fig. 2F, iii), Because the asynchronous state just de-scribed is a dynamical pbenomecon, it does not require finetuning ofnetwork parameters (fig. S4). Parameter changes lead to adjustments in rates and con-elations such that the cancellarion in TIq. 3 still holds.

Although the theory predi cIS that the populationaveraged correlation r should be close to zero, it does not predict that every pair of cells should be as weakly correlated. Rather, the distribution of r across pairs is "wide," witha standard deviarion 0,. much larger than its mean r for large networks (0"" decays only as INN), which resubs in similar numbers of positively and negatively correlated pairs, (Fig, 2G). This is because the hard-wired sources of correlation have a strong impact.on :individual r values (of order l/..JN) and

29 JANUARY 2010 VOL 327 SCIENCE www.sciencemag.org

A

B C
, (},2
A c
0
=af 0.1
~
(; 0
o
>"" -01
-0.05 0 0,,05
S piking co rre latia n r ,50 -25 0 25 50

REPORTS

0
J E'P\l1'. '::?SF
IP8P-IFSP to 0_2
v; .vm 0
""
'" 0,15
~
:;;
0 0.1
>10 0.05 Fig. 3. Cancellation of correlationsin a recurrent network of spikinq neurons. (A) Raster (top) of 500 E (green) and I (red) neurons sorted by rate in a conductance-based inteqrate-and-fire network receiving shared independent Poisson inputs (p = 0.2). Bottom CUIVeS shol'\l tracking of in~tantaneoUS pop Lila tlon-ave ra ged act] viti es (tra nsfor med to t scores, bi n si ze 3 ms). Average firing rates of E and I cells were 1 and 3.6 spike per s, respectively. (B) Histogr.am of spike count correlations (black; count window SO rnsl and

Lag (ms) Hol,dlng Potential {mV}

of jittered spike trains [gray, jitter ± 500 ms (21)]. (C) Population-averaged (CGs of the membrane, potential containing mostly EPSPs (green) or IPSPs (red) in both cells, or EPSPs for one cell and tPSPs for the other (gold). The black curve is from pairs at resting potential. (D) Peak height of the mernbrane potential C(G as a function of the mean holding potential of both neurons in '111e pair. 'Green and red circles correspond to the reversal of inhibition and excitation, a nd the black circle corresponds to rest.

A

B

Fig. 4. Distribution of correlations in the rat neocortex in vivo. (A) Raster (lap) and instan ta n eo us p OpU laf on a ctivily (bottom) for a popuLation of 100 ~imultarleousLy recorded neu rons (sorted by rate) during a period of cortical activation (ACT). (B) Histogram of spike .(Ol!f\t correlations of th~population in (A) is wide (crf» n. The white curve is the mean hi sto g ra m of th e ji tiered spike trains [jitter ± .200 .ms, gray 5hade 95% confidence intervaL; count window 50 ms (21)]. I n sets show average raw e ro S5- correlograms of all negativeLy (left) and positively (right) significantly correlated pairs (P < 0.Q1). (( and DJ Same as (A and B) for the same population of cells during a period of (0(tical inactivation OnAcn. Histogram of correlations during InACT is biased toward positive values (red). Restri ding the analysis t.o up-state activity by removing down-state periods [black brackets in (C), (21)J !argeLy elimin~tes the positive bias (Up-Sf, orange). (E) Box-whisker plots showing the distribution of mean correlations across experiments for different conditions. Crosses represent outliers. (f) Median fraction of significantly correLated pairs (P < 0.01, white bars) a nd of significantly and negativeLy correlated pairs (filled bars) across experiments. Error bars represent i nterqua rtile range.

'~r ~5LY-

-Il,' 001

:_'l\1_"

'" £ 0 1,;,' .:...L"":":'..;..&."":"":"';_:"";"L...J:......:.....a.:..

-0.1

o

0,'

c

Spiking correlation r

o

~

II> _,.,_

Q.

j : I ....... _,._.:...;....;:....:...._.:....>.. ......... '-'-""'"-_

InACT

E

-0.2 F

SpIking ccrretanon r

o

0,2,

J

therefore generate large heterogeneity across pairs (fig. S3), Asynchronous activity is. also possible under nonstationary conditions; NumericaJ simulations with time-varying inputs display a similar correlati On structure if r is computed with respect to. tl,e time-varying instantaneous average activity of each cell (fig" S5).

Active decorrelation of synaptic currents also occurs in a more biologically plausible network of spiking neurons, We simulated large networks of randomly connected conductance-based imegrete-and- fire neurons (21), with parameters chosen to produce a balanced state (19, 2])

where neurons tired irregularly (Fig. 3A). Iv; predicted, the distribution of spike count correlation cceffieients r is, wide (Fig" 3R), with an extremely low average (in .EE pairs i: < 0.001 for all count window sizes, fig. S6C). This happened fora. large range of average firing rates and connection probabilities (fig. S7, A to F), with synchrony developing only when inhibition was substantially slower than excitation (fig. S7, G to I). The distribution of r conditioned on the response to time-varying external inputs was also wide' for a largo range of modulation frequencies (fig, S8). To determine

0.4

whether a cancellation between the components of the current correlation (Eq, 3) underlies the small value of i: observed, we injected different levels of constant current into cell pairs in which we: bad disabled the spiking mechanism, The range of current levels was adjusted to isolate the excitatory postsynaptic potential (EPSP)and the inhibitory postsynaptic potential (IPSP) comp onents n ear their respective reversal. potentials (13), or combinations ofEPSPs and ll'SPs at inc termediate potentials. TIle correlation between isolated EPSPs (Fig. 3C green) and between isolated IPSPs (red) was much larger than the correlation measured with no injected current (black) because of a cancellation with the largenegative correlation between EPSPs and IPSPs (gold). The Vsbaped relation between membrane potential correlation and holding potential (Fig. 3D) is an experimentally testable prediction of our jheory,

The existence of a wide distribution of spike count correlations was confirmed in neuronal population recordings collected with silicon m:icrcelectrodes in somatosensory and auditory cortices of urethane-anesthetized rats (21). Under urethane anesthesia; cortical activity displays spontaneous changes in state (24) homologous to those seen during sleep (25). Network activity alternates between an "activated" (ACf) state of tonic firing, resembling thaI seen in rapid eye movement (REM) sleep (Fig. 4, A and B, blue). and sn r'inactivated" state (InACT), characterized by global fluctuations. in population activity {up-down transitions) resembling slow-wave sleep (Fig. 4, C and D~ red). During ACT periods, correlations were, OD average, remarkably small and the correlation histogram was wide, (Fig. 48 sbows one experiment: F = 0.0075; 47% of pairs negatively correlated) These values were typical of ACT state- correlations across different animals [(Fig. 4E) fl = II recording sessions in nine rats, r mediae was 0.00.53 (O.D024 to 0.0094 interquartile range); across all 3.0,772 pairs, I' = 0.0.052, arid 4.7% bad r < .0]. This behavior did not depend strongly on the time scale at which correlations were measured (fig. S9). Although r- in the ACT state was systematically Low, it was po siti ve and significant! y different froth zero

www:sClencemag.org SCIENCE VOL 327 29 JANUARY 2010

589

I

REPORTS

in all experiments (Fig, 4E; P <: 0.D05). A mit10rilY of both positive and Ti ega tive eorrel ations were statistjcally significant (Fig. 4, Band F, blue versus gray) .. Pairs with significant negative (positive) correlations showed clear troughs (peaks) in their cross-correlograms on average (Fig, 4Bj insets, and fig_ S10)_ Fipally, {he correlation histogram during the ACT state was' still wide, even if only neWUIJS recorded in the same shank were considered (fig. S lI). During lnACT periods, as expected from eomodulation by the slow oscillation, the distribution of r was consistently biased toward positive values [(Fig. 4, C to F, red) for seven sessions in five Tats, r median-was Q,0953 (0.088100.109 lnterquartile mage); across all .8,916 pairs, r = 0,096, and 9%, had r < 0]. Within up-State periods, however, average correlarions were again weak (26);, removal of down-states fioin the recorded spike trains PJ) resulted in correlation histograms similar to those during tbe ACT state [(Fig. 4, D and .8, orange] r median wes 0..0.163 (0.,00.66 to 0_023 Interquartile range); across all 18,916 pairs, r = 0..0.1311, and 45% had" r <: 0]. Thus, correcting for C01l'l1110n modulations in activity revealed a wide distribution of correlations even under nonstationary conditions ill vivo.

In sumluary, we demonstrate theoretically that recurrent network dyuamies C!.II] lead to an active decorrelarion of synaptic: currents, resultjug in a state of arbitrarily low mean cerrelatien. We therefore conclude that shared input does not inevitably cause correlated activity By preventing uncootrolled network-wide synchrony, this mechanism generates 1:) background of weakly eorrel;nedspj]_cing, as required fill" efficient information processing based on either firing rates or coordinated spike timing patterns (27, 28). Both simulations and ill vivo recordings showed a wide distribution of correlations under stationary conditions, In noastationary conditions,

global activity modulations can result in positively biased correlations, but con-elations around the mean activity imposed by these modulations tim still be extremely small (Fig. 4, 0 arid E; and figs. S5 and SS). Similarly weak eorrelations have been reported in visually driven neural populations in area VI of awake behaving monkeys (29). However, asthe constellation of inputs driving a cortical circuit is, in general, \.)nknown to the experimenter; positrve correlations may persist even after all experimentally controlled variables are accounted fur ut, 12). Whether "residual" correlati on s of this nature will ha vo a strong Impact 00 coding will depend on the extent 10 which downstream networks are able to disambiguate modulations in activity due. to different sources. In either case, we suggest that cortical ci reui try does n or itself consti tote an 1 rreducible source of '<noise. ,.

Ref~ren(e5 and Notes

1. T.): Gown~, B. J. Richmond. J. New"tJ5ci. 13, nS2 ,(1993).

2. ~ ZOhill¥, M. N S~~dlen. W. T. NeWIDl'ne. Nature 37tl, 140. (1994).

3'. E. Vaild'iaet al -r Noltlre 373, 515 (1995).

4. D, lee. N. L Port, W. Krul€, Ii. p. G~orgopoolo~. 1- Nelll'tJ~'J .1B. HoI (1998)"

5. C. M. Gray, W. Singer, Ptoc. Natl. Acad. Sri U . .£Il. 86, 1698 (1989).

6. A. 'R i enle, 5. G run. M. OJ esm a no. A. Aertsen, Science 27 8., 1950 U997)-

7. K. H. Britten, M. ill. 5rn.dlen, W. T. Newsome, J. A. M()\IShon, J Neumxi. 12, 4745 (rim).

8. H. 50mpolinsky. H, Yoen, K. Kang. M. Shamir, Phys. R'eY.

E 64, 051904 (2001).

~. M, N. Shad len. 'IV, T. Newsome.): NeilfO!id. 1B, jjJ·70 (1998), 1(1. B. Krfener. T. TeulMf, A. Aert>en, M. Die,m~nn. 5. R~tter, Neural,Colllput 20. 2185, ~008}.

1L P_ R. Ra~lf!iern~, V, A. F. L'l'nm~, H. Spekteij~.

NDI. Net1W5d. 1, 9S·2 (2Q04).

12. M. R. Coh~n, w. T. New,ome, NeUron 60, 162' ("2008-).

13. H, Nienbolg, B. G. Cumming, Nature 459. 89 Q.009), 14_ R, MDlemj-BD1~, N. Parg~, Phy> ReI<: lett. 96, 02&101

(2;006).

15. The co ,reta tioo between two qUil~'tltie5,eil ch given hy sum oi N V~ Ii il bl"", correlated by a n

amount riO' au! 01 wflkh Np are common, is ~g~il( to

ip I 110 (N - p)11 [1 + ~,(N - 1)1. which ts oppmKinlaMy equ~l to p + Nfl. when p - nnN < < 1-

16. When the !'iring st~tistfu ot the [ ilnd I populations are identi~"l, th .. leilding older effect \if po'i~lie Mng correlari ons on c ca n on ly be ~ o. If eXQ to Ii on an d inhibition are precisely b"lanced., the eqUillil\li5 t~aliled ;3 n d Ul e frMtioll 01 sMa red il1p u! s ets the popillil Ii on" il~er:.getl hrin g cat r~~ ti on (il).

17. Dl!laillol ibe lhe.,ory are availobl~ in the Sunportin,g o oli ne Materia I on 5 cience On I i~ e.

tao I. G1mburg I. H. 50mpolinsry. pnys. R~~. E 50, 3171 (1994).

19. (_ van Vreeswijk. H. SQilipoliMky. Science 274, 1724 (19%).

<'D. (_ Holm9ren, T. H<lrlrimy. S, Svenoento". Y. Zil.bertEf, ). Pl!jIIioL 551, 139 (2003).

2:t M~reli~ls and method, are av:.lWbl~ as supporting

rna II! ria L on SdeiKe O"1i ne,

22. D, 1- Amit, 'N. Brunei, cereb. Cortex 7, 237 (19,97).

23. M. Okun, I. t.o I'np I, Nat. NeurmCi_ 11, 535 (200S)_ 2:4. M. 5!~ri~ele. Elffiroenaph/JliJgr~p/iy (Williams &

Wilkin,. Billtimore, MD, ed, 4. 1999).

25, 6. A. (lfmen! et oi.. PloS ONe 3,e2004 (2008).

26. E. A_ Stem .. D. l~eg~r, (_ J. WlL.an. Nature 394, 475 f1.99S)-

27. T_ P. Vogej" L e. Abbot:t.. 1- Neurosci. 25, 10736 (2005). 23. A, Kumar. S. Rotter. A. Aertsen,1- Neumsci. 28, 526B (200al. 29. A. S. Ecker e/ 01., Stieno: 327 .• 584 (2009).

30_ We th~nk A. Art"ra;ingnilm, No' Brunet. G. Buz,~.k;-.

B. Doiron, A, E(ker, 5, FijiXlwa, A. Heirnet, A. ~obn. D Robbe, 5. Sd~at<i, E, Stark, and A. Tollos for comments on iln earlier ver~ion at 'this milnul{flp~

(_ 'V2 n V'''!;5Wjjk lv, d i, russ ian" ;10 d A- COll)P Ie. S. Ardiel, and l, Manrique for ,haling their codes. Thi! study

was supportEd by NIH granl, MH073l45 and DC009947; NSF gran! 5BE·0542013 to the TemporaL Dynamics of Learni'ng Certter, a n illS F ,tie nee 0 f leo rn in g" Ceo. II' ri a iIlationa lin, ti tu te on Dea fn~>-s and Other CD m rn un i cation Disorders, N1H,granl DC·01l5787-01Al; "nd a 5panl,h gral11 FI5 2006-09294. K,[XH., is an Alfred P. Slaa~

r ~U DW. We would li ~e. to d~di<;o tB- t~i' WllfK 10 the. m~!lio!,/ of D. j. Amit,

Supporting Online Material

WW\V. )(jencem ag ,orglc:g ileo ntent/fu IV32 715 9 6 515 8 710C1 Material~ and Method~

SOM Text

Fig,. 51 10 511 R€f~renCe5

n lui y 2:Q09; accecred 10 De eember 2 OD9 10.112 6I5c1 enee .1179 a so

Direct Restart of a Replication Fork Stalled by a Head-On RNA Polymerase

Richard'T. Pomerantz and Mike O'Donnell*

I n vivo studies suggest th at replication forks are arrested by encounters with head -on transcri'ptio n complexes. Yet, the fate of the replisorne and R.NA polymerase (RNAP) after a head-on collision is unknown. We found that the Escherichia coli replisome stalls upon collision with a head-on transcription complex, but instead of ~oUap5ing, th~ replication fork remains highly stable and eventually resumes elongation after displacing the RNAP from DNA. We also found that the transcri ption·repai r coupling factor Mfd promotes direct resta rt of the fork after the collision by fadlitating displacement of the RNAP. These. finding.s demonstrate. the intrinsic stability of the replication apparatus and a previously unknown role for the transcripticn-coupled repair pathway in promoting replication past a R.NAP block.

In vivo studies suggest that replication forks are arrested by head-on transeriprion complexes, but are unaffucted by ccdirectional transcriptron complexes (1) [supporting online" materia! (SOM) Text S I].. Mechanisms that

590

resolve bead-on collisions in favor of the replisome are therefore necessary fQl" chromosome duplication and may preserve genomic integrity by preventing fork collapse, In vivo data, indicate 1hat head-on replis"ome-RNA polymerase

(RNAP) collisions cause chromosomal deletions, which suggests dissociation of the replisome (2). GenetlC stod i es irn plicate recombi nation aJ repair in resolving conflicts between replication and transcription, which also suggests tile possibility of fork collapse (3, 4). Similarly, in vitro data imply that the replisome dissociates after eocountering a lac repressor, which arrests the fork (5). In contrast, several in vivo studies indicstethat althoug'h repllcation forks st;i:ll at protein barriers, the repiisorne remains stable and resumes elongatiouafler removal of the block (6), Thus, replisome stalling may not necessitate fork collapse (1). We investigated the stability ofthe Escherichia CQlt replisome after it encounters .11 head-on RJ'lAP in vitro.

1he Rorfreleller University, Heward Hughe, Medk~llll,bnll€. 1230 York Avenue, N~w York, NY l002i USA,

"To Whom correspondence ~hould be addressed, E'mail: c don nell@mait, fod<eleller .edtl

29 JANUARY 2010 VOL 327 SCIENCE www.sciencernag.org

The E. co Ii replisorne is a 111ul tiprote in complex that copies DNA with high speed [-..630 nucleotidcs (m) .'tl] and processivity (~50 kb) (8). A solid-phase assay was used to study a rcplisome-RNAP head-on collision (Fig. I A). An RNAP-halted elongation complex was assembled on linear DNA, immobilized to streptavidin beads. then washed with B high concentration of salt to remove unstable RNAP- DNA complexes (fig. Sl) (9), Next, the replisome was assembled in two steps: Fi:rst, the replicative Dnafl helicase that encircles the lagging strand was added: second, DNA polymerase ill (Pol. ilf}, the clamp loader, and the (3: clamp were added along 'with A TP (ad enosine 5' -triphosphate], ['::'P](l-dCTP (2· -deoxycytidine 5' -triphosp hare), and e2p]-a-dGTP (2'-deoxyguanosine 5'triphosphate). Fork movement was initiated by adding [~1p]-ct--dArp and [32Pla-dITP wjth single-strand DNA binding protein (SSB). We observed a :l.5-kb product equal to the distance from the fork to the promoter, indicating that the

A

replisome was impeded by the RNAP (Fig. 1 B, lane 2). Full-length DNA {3.6 kb) was also produced" suggesting incomplete promoter occupancy by RNAP or replisome read-through of the transcription complex. (Fig. I B. lane 2). Omitting the prom o ter sp eci fi city factor, cr 10, resulted in only full-length DNA (Fig, IB, lane 1). An average of 50.% (n = 8) of the repfisomes produced full-length DNA in the presence of a bead-on RNAP (Fig. IB, light), which exceeded the Dumber of templates that lacked RNAI' (24%; fig, S2). This suggested that -26% of the replisomes passed a head-on RNAP during the IOc min time course. We determined whether the replisome reads-through RNAI' directly by perfonning a pulse-chase experiment in which cold dNTPs (deoxynucleoside triphosphates) were added after 5 111in, and extension of the 2.5-kb productwas monitored (Fig. IC), A. steady in, crease In. tb e ratio of full-len gth to intermedi atelength product was observed, indicating that the replisome passes tbeR:t'lAP, albeit after pausing

8

Tra nsenpti o n com plex assembly

" 2.5kb i , 1,1 kb

//~~==r::J:#=~ :::::::::::6

., Wa,s h t -, RNAP

DnaB t

Raplisome I assembly t

\ \

Leading ~

strand

<

...J

II

Laggj~g strand

Replicalion I ~-clamp Pal III elongali[)n t

_.,. ...-

Repliootion T ranscrl phon

Transerlptlol'l complex .as~embly

,

Wa~h

I ~e~1icaIIOn

+ t

Xhol digesllt).O

;---~:

A

(RU) 12K101

10x107 <:C

5 8.x107

'Z 6x107 1ii

~ 4)[107 i5.

2)1;107

tb

4.0

block

0.5

0.611

~ 0.5

'0 C 0,.4.'. r

C.c

.g"& 0.3

c .. .s::

E.1: 0,2 :a 0.1 0-

REPORTS

for a considerable duration (Fig. I C). The relatively large. amount of,fuU-leogtb product (39%) observed after 5 min suggests fbat some replisomes might have passed the RNAP MID high cffic:icI1CY. Rcplisomc rea:d"~lfOl.lgh ofRNAP Oil a different template ruled out any sequence-specrfic etreo\S (Og. 53). The stalled repllsome remained active for 60 mm after the collision without the need for primcsomal proteins, su!;h as PriNe, that are necessary to reload DnnB ento SSB...coated single-strand DNA (Fig. IC) uo, 11). DOllE therefore stayed bound to tbe stalled fork. Consistent with previous. studies, we demonstrated that DnaB was required for replication and that SSB prevented DnaB loading in the absence of prinrosomal proteins (tig. S4)_ Thus, although the replication fork stalls upon encountering a bead-on RNAP~ the replisome remains intact and resumes elongation, presumably after displacing the trenscriptioacotnplex,

We detennined whether the rcplisoine displaces the RNAP from DNA by Xho 1 digestion

3.

2.

0.5

C

kb

• .-RNAP block

~.5

2.1

Fig_ 1 .. The replisorne slowly passes. a head-on RNAP. (A) A head-on RNAP was assembled on immobilized DNA. DllaB was added; the n th e replisome waS ;l5sembli1d and replication initiated. Replisome component fu n rtions (8): Pol III core (ora n ge), synthesi.zes DNA~ ~ damp (dark bLue)' confers processivity to Pol HI; damp-loader (light blue), assembles~· damps onto primed sites; Dna6 (yellow), unwinds DNA. (B) leading-stl'Clfld syn-

thesis was performed in the presence (lane 2) or absence (!anI" 1) of a head-on .RNAP. (RightHhe mean traction (:t. 58 of MHength DNA (0.5, n '" B) produced in the presence of a head-on RNAP is plotted. (C) Time course of leadin.g"stTanil

1 2 100 41 % F'LlIl-leng'th

1 2 3 4 0.9 5 15 25 35 45 55 65

39 44 5652 % Full.I .. ~gI~ Mill

5 10 30 60 Min

synthesis on DNA containing <I head-on RNAP. The ratio of full-length to i nte rmedia te-length DNA UfIJlI) is shown in the. rig hi plot (If!., i nt~nsity of !utilength DNA; ii, intensity of intermediate-length DNA).

FiNAPfootpri nt

B

C"' ·5'

.L3 L.! ~

20 4(l 60 Min

Fig. Z. The replisome displaces a head-on RN.AP from DNA. (A) Replisome displacement of RNA? was probed by Xho I digestion 01 the promoter-proximal seq!.lence in the presence or absence of repltcation, A halted RNAP was assembled on DNA., and the reaction was divided. Replication was ((anes 2, 4, and 6; blue bars) or was not (lanes 1, 3, and

Transcrlp! of challenge

... -I.emp.lale

5; red bars) initiated, and samples were treated with xho I at the indicated times. RU, relative units. (B) RNAP displacement in the presence (lane 3) or absertce (lane 2) Q·f r.eplication was determined by monitoring transcription of 11 challenge template. Transcription of challenge template control (lane 1).

www:sClencemag.org SCIENCE VOL 327 29 JANUARY2010

591

REPORTS

Fig. 3. Mfd promotes fork progression folLowing a head-on collision. Lea di n g-stra n d synth esi s was performed on DNA containing a head-on RNAP. Reactions were divided before (8)' or after (A and C) replication was performed for 5 min. (B) DnaB was removed from solution by washing after the collision wa s formed. B1Jffer (la nes 1) or either wl1.dtype [(A) and (B), lane 2] or K634N Mfd [(C), lane 2] was then added for a further 5 min.

A

Transcription c~mplex assembly

J

,

Replicali~n

I

jMld

kb + Mfd

__ Fulllength .... RNAP block

2.0-

1,5-

1.,0-

0,5-

1 2

52: ao %,Full"lang1h

of the promoter-proximal sequence, which is protected by RNAP (Fig, iA). RNAP occupancy of the promoter blocks digestion, whereas RNAP displacement allows digestion. Removing ribonucleotides (NTPs) by washing prevents reassembly of the halted RNAP. A significant increase in digestion was observed only when replication was performed (Fig. 2A., rigbt; compare blue and red bars) .. Little RNAP displacement was detected <It 2.0 min, which is' likely due to replication of only 8. $% of'tbe DNA, probably as a result of RNAP binding to the primertemplate (J 2) (fig. S5). Nevertheles», the large increase in digestion ~e to replication indicates that the replisome displaces the RNAP. Similar results were observed In the presence of OreH and limiting NTPs. which inhibit R,NAP backtracking (fig. S6). We further investigated RNAP displacement by monitoring transcription of a cballenge template in the presence of 010 and limiting NTPs. Transcription of the challenge template was only observedafierrephcetion, which indicates tha); the replisorne displaced tll_e· RNAP (Fig. 2B, compare lanes 2 and 3). These results are consistent with the ability of DnaB to displace a protein bloclc during DNA unwinding (lJ).

Genetic data implicate RNAP modulators such as Mfd in resolving conflicts between replication <!TId transcription (3}. Mfa displaces a halted RNA? from DNA and recruits the nucleotide excision repair machmery 10 the site, which results in preferential repair of the transcribed strand when RNAP stalls at lesion referred to as tranacription-coupled repair tTCR) (14 f 8). TCR bas been postulated to promote fork progression by displacing RNAP blocks nom DNA (16) .. We examined whether Mfd fucilitates restart of the fork after a head-on collision, After providing 5 min to allow the replisome 10 collide with RNAP, we divided the reaction and treated it with either MfU. or bufferfora further 5 min. Nearly full extension of the

592

B

c

current view of transcription, We find that Mfci promotes direct restart of the fork following the collision by facilitating displacement of the RNAP (fig, S9, right). Codirectional collisions are resolved without auxiliary factors; the replisome uses mRNA as: a primer to reinitiate leading-straad synthesis after disp lacing a codireetional RNAP from DNA {9}, Pol HI extension of the RNA was not observed iu our stuc!y; probably due to displacement of the uanseript, Genetic data suggest that recombinational repair and other RNAP modulators help resolve repl:isome-RNAP conflicts, thus explaining the normal growth rate of cellslacking Mid (3). mid cells, however, demonstrate a greater lapse in replication and cell growth fuIJowing ultraviolet irradiation, which supports a role for Mfd in facil itatlng repl icaticn through transcripn on complexes arrested by lesions in vivo (20). Ow' data demoustratea new role for TCR in promating replication past an RNAP block and may have implications fOT human disorders that result from defieiem TCR, such as Coekayne syndrome (15, 16).

References and Notes

1.. C. l, R udolp~, P. DhiUan, T. Moor€, R. G. Lloyd, DNIl Reporr (Amst.) 6, 9 a 1 {;WO 7).

~. D, Vi!ette. S D, Ehllich, B. M i [11 e ~ Mol, Gen. Genet. 2 5.2, }9!1 (1<)96).

3, B. W. Tr~utin!i~t, R. p, J,,~t"j i, f. RUSokava, R, G. lloyd, Mol. Cell 19, . 247 (2005).

4. P. McGly~~, R. G. 11 oyd, tel r 10 I, 35 (lOOO).

5. P. McGlynn, c. P. Guy, J. M~I. BiD/. 181, 249 (2008).

o. K. tabib, B. Hodgsorl, fM80 I/ep. 8,346 (Z007).

7. V. Bidnenko, S. D. ~hrh(h, B. Miche!, fMtlfJ j. 21,. 3898 (201:)2).

8. R. T. PomerimlZ, M. O'oonneil. ttends Mir:robiol. 15, 156 (2007).

9. R. T. Pomerantz, M, O'l)omieU, N"tUr~ 456, 762 (2008).

10. R. C. Heller, K. j, Mari~ns, Nature 4.39, 557 (2006) 1L A. Yu!hakov, J. Turner, M. O'Oonn~ll, Cdl 86, 877 (1996),

12 D. C. Hfnkl€, I. Rin~, M. I. Chamberlin.". Mol, Bfai 70, 1<),7 (.1972).

is. o, c. Kapliln, M. O'Donnell, MoL Cell IS, 453 (2004).

14. I. s, P.r~, M. T. Marr, J, W. RobertI, Cel/lO'J, 757 {2002).

15 P. C. Hanawall, G. Spivak, NCiI. Rev. Mol. eel/Bioi. 9, 9SS (2008).

16. P. C. lianawall, science 266, 1957 (1994').

17. c. p, Selby, A Sancar, J. 8iol. (hem. 270, 4882

(1995).

ra 1\. M. Oe"clmE:>cu et oi., Cell 1<'.4. 507 (2006).

19. B .. tiu, B. 1\11. Ail)~rI5, Sciew~ .267, ll~l (1995).

20. D, L. George, E. M. Wftki~, Mol. se« Genet. B3, ZS3 [1974).

24 We are g;;Mfullo S, D~r't 10f pn;widing RIIIAP and Mid protein> and e~prnsion vectors. We .il15Q thank

S, 6orukho~ 10 r p rovi ~i n 9 GreS a nd D. Zh a ng lor tschnital support This \York W~5 supported by a _grant from the NIH iM.O,O) Imd by a Marie-]o,ee and It.my Kravis fello\\"hip at the Ro{ke[eller UniV€rsity (R.T.P.).

Supporting Online M .. terial

\VW\Y .>rlence~1~g.orgfcgil co I~ tenllfu U/3 2 715 9 65159 olOn Mali'rials and M,lliod,

SOM T<'xt

Fig~. 51 (0 511

B July 1'009; ac(~pted 30 ~ovem.h~f Z009 W.1126/!{je!(c€.1179595

29 JANUARY 2010 VOL 327 SCIENCE www.sciencernag.org

'Transcription complex assembly

M

Rep!icBtio~

, ,

Wash .awa,y Dn'lB

I M~d

kIlf t

4.Q... - + MId

. I .... Full-

length

.. RNAP block

Transcription complex assembly

4

Replication

A34NMfd

kb + K634N Mfd

... Fulllength

1.5-

1.5-

plock

2.Q...

2.0-

1.0-

1 .. 0-

O.S-

e.s-

12

52 7S % Full.";mgl~

25-kb product Was observed uponaddinon of Mfd (Fig, 3A, compare lanes 1 and 2), which removed the RNAP block (fig. S7). We determined whether Mfd-mediated replication restart requires a supply of DnaB in solution (Fig. 3B). For this experiment, the beads were wasbed after the collision then resuspended in buffer containing dNTPs and all the replication proteins except DnaB either 10 the presence or absence of Mfd, The addition of Mfd again resulted in extension ofthe 2.5-kb product, providing further evidence that DoaE stays bound to the stalled fork. Next, the experiment of Fig, 3A was repeated with Mfd K634N, which is defective in ATPbydrolys:is and can no longer dislodge RNAP (17) Mfd K634N failed to promote extension of the 2.5-kb product (Pig. 3C, compare lanes f and 2). These results. indicate that Mffi promotes fork progression after the collision by using the energy of ATP, and hence trauslocase activity, to dissociate the transcription complex ahead of the stalled fork, Finally. we demonstrated that the ability of Mfdto promote replication pasta head-on RNAP was' unaffected by the addition of all four NTPs and Grel], which promote transcription elongation (fig, S 8).

ill conclusion, W6 find that although the replication furk stalls upon collision with a. head-on transcription complex, the replisome remains stable and resumes elongation 'alter displacing the RNAP (fig. 89, left). It is conceivable that the collision may induce RNAP backtrac.kiog. However, the lack of stiroulatiou of f{l\JAP endonuclease activity after the collision suggests. that this is not the ease (fig. SI0). Moreover, the addition of GreR and NTPs, which inhibit backtracking, have no effect on our assays (figs. S6, S8, and SJ 1). A previous study of the T4 rep' isomereported that a bead-on R.NAP remalns bound to the DNA, but the RNAP' and transcript switch strands as the replication fork passes {I 9), This result is difficult 10 reconcile with the