You are on page 1of 32


L i f e s c i e n c e s at W h i t e head I nstitute for Biomedica l Research | spring 2007

Cancer pa g e 1 6
Why do discoveries take
so long to reach the clinic?
When your genes aren’t
to blame, what is?
Is tumor sequencing
ready to ramp up?

Tuning into
protein networks
big tent
Window on Whitehead On the cover
Researchers study
Science for me and YouTube an enormous range
of possible causes
My daughter is studying biology in high school, and her experi- and cures for can-
ence is both amazingly like and amazingly unlike mine at her age. cers—and the early
The amazingly unlike part isn’t hard to figure out. My old biology text- results of more powerful under-
book doesn’t mention recombinant DNA, which had barely been invented. standings are just now filtering into
My daughter lives in a world in which the human genome has always been clinics. Here, a cultured cell with
sequenced, sheep have always been cloned, and a certain number of your structure typical of melanoma or
friends have entered this world via in vitro fertilization. breast cancer. Image by Stone
What’s amazingly like is how she’s learning: teacher, textbook and a lit-
tle time in the lab. Oh, her biology teacher is fond of educational websites,
but those aren’t terribly important in class (yet). Among our contributors
However dramatically medicine has changed
The popularity of in my lifetime, that’s nothing compared to what DAVID CAMERON
she’ll see. As both a medical consumer and citi- manages media rela-
Web videos gives zen, she’ll need to understand the strengths and tions at Whitehead
us new ways limitations of the major advances now lurking Institute. He tells us
just over the horizon. he is secretly biding
to learn about But I don’t think she’ll spend much time his time until he lands a position
today’s biomedical reading about them in print. as staff writer for Jon Stewart’s
We still get print newspapers, news maga- Daily Show.
research zines and science publications delivered at home.
My daughter rarely reads any of them. ALYSSA KNELLER is
What she does, like her friends (and her parents), is spend time on the Whitehead’s Web
Web. Lots of time. The Web will be her main channel for tracking the editor. She has lived
future of biomedicine, as it will be for so many other topics. on three islands and
And when she can, she’ll be watching videos on the Web. worked as an envi-
For years, Whitehead has been filming our principal investigators as ronmental consultant and commu-
they give lectures to our non-scientific staff, and posting those films on our nity journalist.
website (
But, as everyone knows, the popularity of Web videos is soaring now JAMES robert
with the combination of powerful PCs, fast Internet connections, inexpen- O’BRIEN is a nation-
sive digital video hardware and software, and Web video aggregators. ally recognized
That’s excellent news for public understanding of biology. Along with illustrator whose
the extraordinary power and promise of today’s research comes extraordi- work is created using
nary complexity. The popularity of Web videos gives us new ways to dive Adobe Photoshop and Illustrator
through all those details to learn about today’s biomedical research. along with found and hand-made
Enjoy an animation of proteins doing their dances together, and sud- elements. In his rare free time, he
denly you understand the basic concept. Watch a researcher explain what enjoys entertaining his daughter,
her lab studies, and it becomes clear. Show students a postdoc describing running marathons, reading impor-
how he got excited about his field, and you can inspire them too. tant novels and collecting and cre-
What’s really new is that people don’t have to wander across your web- ating music.
site to find this great stuff. Today, for instance, YouTube’s most famous
science video shows the startling results of dropping a Mentos mint into a JAMES YANG has
bottle of Diet Coke. But the video aggregation website also is becoming a received over 200
major resource for high school teachers swapping classroom videos. awards for design
So we at Whitehead and our colleagues at other research institutions and illustration excel-
will be expanding our use of video, along with podcasts and other Web lence. He also has
goodies. It’s another way for our scientists to tell their stories—and some- exhibited sculpture at the Smith­
times, my daughter and her peers will tune in. —Eric Bender sonian Institute’s National Museum
of American History and created
the children’s book Joey & Jet.
paradigm Life sciences at Whitehead Institute for Biomedical Research spring 2007

Cover stories: Questions on cancer
12 A slow saga of success
Ever wondered why the journey from lab discovery to the clinic
takes so long?

16 The unusual suspect
Cancer researchers look beyond the genome to the epigenome

21 Out of sequence?
Scientists debate whether it’s time to tackle tumor genomes
and epigenomes

Features 16
5 The RNA connection
Joint projects by David Bartel’s lab highlight the crucial role
of collaborations

6 Biofuels and the gene pool
The power of yeast genetics might make ethanol fuel much
more cost-effective

8 Mouth to mouth
What can a frog mouth tell us about human birth defects?

10 Network news 6
Hui Ge sifts through oceans of data to explore how genes collaborate

24 Biology’s big tent
Meet some of the young researchers pouring into life sciences
from other fields

26 Targeting the agents of disease
In the war on infectious disease, are we spending smart?

2 Science digest 26
Cracking open the black box of autoimmune disease, and a gene that
shuts itself off

22 Whiteboard
The link between epigenetics and cancer

28 Fast FAQs
For all the controversies, it’s still early days for stem cell science

29 On the Web
Twenty-five years ago, Jack Whitehead signed the agreement creating
the Institute that bears his name
29 spring 2007 paradigm []
science digest This schematic rep-
resents the strategy
to identify where the
leads to autoimmune disease.
Scientists previously discov-
ered that regulatory T cells are
master gene regulator
Foxp3 physically inter-
themselves controlled by a mas-
acts with the genome in ter gene regulator called Foxp3.
T cells. The background Master gene regulators bind to
is a microarray where specific genes and control their
the red probes reveal level of activity, which in turn
regions of DNA where
Foxp3 is bound.
affects the behavior of cells. In
fact, when Foxp3 stops func-
tioning, the body can no longer
produce working regulatory T cells. When this hap-
pens, the frontline T cells damage multiple organs and
cause symptoms of type 1 diabetes and Crohn’s disease.
However, until now, scientists have barely under-
stood how Foxp3 controls regulatory T cells because
they knew almost nothing about the actual genes
within Foxp3’s purview.
Researchers in Richard Young’s Whitehead lab,
working closely with immunologist Harald von
Boehmer of the Dana-Farber Cancer Institute, used a
DNA microarray technology developed by Young to
scan the entire genome of T cells and locate the genes
controlled by Foxp3. There were roughly 30 genes
found to be directly controlled by Foxp3. One, called
Cracking open the Ptpn22, showed a particularly strong affinity.
“This relation was striking because Ptpn22 is
black box of strongly associated with type 1 diabetes, rheumatoid
arthritis, lupus and Graves’ disease, but the gene had
By David
Cameron autoimmune disease not been previously linked to regulatory-T-cell func-
tion,” says Alexander Marson, an MD/PhD student in
Autoimmune diseases such as type 1 diabetes, the Young lab and lead author on the Nature paper.
lupus and rheumatoid arthritis occur when the immune “Discovering this correlation was a big moment for us.
system fails to regulate itself. But researchers have not It verified that we were on the right track for identify-
known precisely where the molecular breakdowns ing autoimmune-related genes.”
responsible for such failures occur. Now, a team of
scientists from Whitehead Institute and Dana-Farber
Cancer Institute have identified a key set of genes that “ With this new list of
lie at the core of autoimmune disease, findings that may
help scientists develop new methods for manipulating
genes, we can now look
immune system activity. for possible therapies
“This may shorten the path to new therapies for
autoimmune disease,” says Whitehead Member and
with far greater
MIT professor of biology Richard Young, senior precision.” —Richard Young
author on the paper that appeared online in January in
Nature. “With this new list of genes, we can now look The researchers still don’t know exactly how Foxp3
for possible therapies with far greater precision.” enables regulatory T cells to prevent autoimmunity. But
The immune system is often described as a kind of the list of the genes that Foxp3 targets provides an ini-
military unit, a defense network that guards the body tial map of the circuitry of these cells, which is impor-
from invaders. Seen in this way, a group of white blood tant for understanding how they control a healthy
cells called T cells are the frontline soldiers of immune immune response.
defense, engaging invading pathogens head on. “Autoimmune diseases take a tremendous toll on
These T cells are commanded by a second group human health, but on a strictly molecular level, auto-
of cells called regulatory T cells. Regulatory T cells immunity is a black box,” says Young. “When we
tom dicesare

prevent biological “friendly fire” by ensuring that the discover the molecular mechanisms that drive these
T cells do not attack the body’s own tissues. Failure of conditions, we can migrate from treating symptoms to
the regulatory T cells to control the frontline fighters developing treatments for the disease itself.”

[] paradigm spring 2007
antisense madness. When conditions
Dueling RNAs protect cells around yeast cells are good and
rich in nutrients, the cells divide by
Researchers have found that mitosis—that is, the DNA duplicates
a class of RNA molecules, previ- so that each daughter cell receives
ously thought to have no function, exactly the same number of chromo-
may protect sex cells from self- somes as the original cell. But when
destructing. These findings were the yeast cells are starving, IME4
published in November in Cell. switches on and activates a process
Central to this discovery is the called meiosis. Here, the cells divide
process of gene expression. When a into germ-cell spores that, like mam-
gene is ready to produce a protein, malian egg and sperm cells, have half
the two strands of DNA that con- the normal number of chromosomes.
stitute the gene unravel. The first Yeast spores withstand this harsh
strand produces a molecule called environment far better than the
messenger RNA, which acts as the larger cells from which they spring.
protein’s template. Biologists call But in some cases, flipping the
this first strand of DNA the “sense” meiotic switch can be catastrophic.
or “coding” transcript. Even though If a cell with only one copy of
the other strand doesn’t contain a each chromosome (a haploid cell)
protein recipe, it may also, on occa- Haploid cells, which contain one copy of is forced into meiosis, its progeny
sion, produce an “antisense” RNA each chromosome, should not undergo won’t survive. Fortunately, such
further reduction of their chromosome
molecule, one whose sequence is number through meiosis. To prevent such
destructive meiotic division is
complementary to that of the mes- a catastrophe, haploids block the produc- avoided in haploid cells because
senger (sense) RNA. Antisense RNA tion of a protein called IME4 by expressing they continually produce IME4 anti-
has been detected for a number of antisense IME4. Failure to block sense sense RNA, blocking the production
genes but is largely considered a IME4 results in the aberrant meiotic events of sense RNA. Antisense IME4,
shown here in purple.
genetic oddity. then, safeguards against meiosis in
Using baker’s yeast, Cintia Hon- cells that can’t handle it.
gay, a postdoctoral researcher in the specific function in a higher cell for “This is really the first time we’ve
lab of Whitehead Member Gerald antisense RNA has been found,” seen a gene regulate itself in this
Fink, discovered that in the case of says Fink, senior author on the way,” says Hongay. “Considering
a gene called IME4, the antisense paper. “This points to an entirely how widespread these antisense tran-
RNA blocks the sense RNA. In new process of gene regulation that scripts are,” adds Fink, “I wouldn’t
other words, the gene disables its we’ve never seen before in eukary- be surprised if these findings eventu-
own ability to make protein. otic cells.” ally lead us to discover an entirely
“This is the first case where a There is a method to this sense/ new level of gene regulation.”

New class of RNAs is revealed
An entirely new type of RNA molecules has been dis- additional 21U-RNA genes might reside,” says Bartel.
covered by scientists in the lab of Whitehead Member and “Combining these predictions with the 5,000 that we
Howard Hughes Medical Institute Investigator David Bartel. experimentally identified, we suspect that there are more
Reporting in January in Cell, the team describes iden- than 12,000 different 21U-RNA genes in the genome.”
tifying more than 5,000 of these new molecules, termed Because each gene typically produces a unique 21U-RNA,
21U-RNAs, in the C. elegans worm. Each molecule con- a very large diversity of molecules is made.
tains 21 chemical building blocks (nucleotides), and each While the genes are “spread out over a wide swathe
begins with the chemical uridine, represented by the letter of the genome, they all share common requirements for
U. In addition, each of the 5,000 different 21U-RNA mol- expression and common structural features,” says Bartel-
ecules comes from one of two chromosomal regions. And lab PhD student J. Graham Ruby, lead author on the paper.
while the 21Us themselves have diverse sequence pat- “The fact that 21U-RNAs share this common structure
terns, the DNA sequences residing just outside those that and origin suggests an important function,” comments
cintia hongay

give rise to each 21U are identical. MIT professor and Nobel laureate Phillip Sharp, who was
“Using the sequence pattern, we can predict where not part of the research team. spring 2007 paradigm []
This mouse embryo has been environmental nutrients
labeled with a reporter that and protein growth fac-
detects the high level of tors, which in turn influ-
mTOR protein activity. mTOR
is stained red here, while DNAence the size of a cell.
is blue. When rapamycin blocks
mTOR, it tricks the cells
responsible for organ
rejection into believing that they are starving.
But scientists now realize that mTOR’s significance
reaches beyond its relation to rapamycin. mTOR also
plays an integral role in many cancers, including pros-
tate and brain cancers.
Reporting in December in Developmental Cell,
postdoctoral scientist David Guertin and others in
Sabatini’s lab describe using genetic tools to show that
mTOR is a critical regulator of a prominent cancer
protein called AKT.
In a previous paper, then-postdoctoral researcher
Dos Sarbassov, Sabatini, Guertin and colleagues
showed that if proteins critical for one aspect of mTOR
activity were inhibited, AKT could not activate. This
implied that blocking mTOR might prevent AKT
Cancer pathway exposed from driving tumor growth. “But that paper relied on
biochemical techniques, like RNAi, to interfere with
In the mid-1990s when Whitehead Member David mTOR, and so not everyone in the scientific commu-
Sabatini, then a graduate student at Johns Hopkins nity accepted it,” says Sabatini. “To get the conclusive
University, discovered a protein called mTOR, he had evidence that mTOR is a major player, you need to
no idea that within a decade this finding would catch knock it out altogether. And that’s what we did here.”
the attention of drug companies worldwide. In the recent study, Sabatini’s lab developed mouse
Sabatini and others had been investigating the models in which genes necessary for this aspect of
mechanisms behind the success of rapamycin, a mTOR activity were deleted. Again, AKT was signifi-
drug that helps prevent organ rejection in transplant cantly inhibited in these animal models. “Discovering
patients. They found that the drug works by blocking this new branch of mTOR signaling has changed how
a previously unknown protein, which was eventually we think about mTOR’s role in cancer,” says Guertin.
dubbed mTOR (for mammalian target of rapamycin). “This work has opened the door to new therapeutic
Scientists soon found that mTOR helps cells detect strategies that could have a broad impact in the clinic.”

A new drug target for herpes?
For a family of viruses as wide- to remove ubiquitin (in
spread as herpes, relatively few magenta), a small mol-
drug targets exist. Now, scientists ecule that flags broken david guertin (top); christian schlieker (bottom)
in the labs of Whitehead Member proteins for disposal.
Hidde Ploegh and Harvard’s Rachelle “We don’t know why
Gaudet have solved the complex the virus needs this partic-
structure of a recently discovered ular function, but because
protein that is found in a wide range all herpes viruses contain
of herpes viruses. This protein may an activity very much like
prove to be a potential drug target. M48, it must be impor-
Reporting in the March 9 issue of tant,” says Schlieker. “The
Molecular Cell, Whitehead postdoc fact that the enzyme’s
Christian Schlieker described the use called M48 (pictured here in yellow). architecture is distinct from host pro-
of X-ray crystallography to delineate It turns out that M48 belongs to a teins makes it an attractive target for
the intricate structure of this protein, class of proteins whose function is therapeutic intervention.”

[] paradigm spring 2007
Manolis Kellis, MIT,
and Eric Lai,
Sloan Kettering
(not shown)
bartel group, sive: kim furnald; burge: tim gray; lodish, ge, camargo: sam ogden; ambros: rosalind lee; kellis, horvitz: donna coveney/mit; bartel: thomas lavergne/rice university; nusbaum: maria nemchuk

MicroRNA genomics
and targets in flies
Lee Lim, Rosetta
MicroRNA target
recognition in
human cells
Harvey Lodish,
Michael Hemann, Whitehead Craig Mello,
MIT MicroRNAs in University of
MicroRNAs blood cell Massachusetts/
involved in development Worcester
cancer Small RNAs in worms

Bonnie Bartel, Cliff Tabin,
Rice University Harvard
Roles and MicroRNAs in
targets of plant mammalian
microRNAs development

The Bartel lab

Chris Burge, Hazel Sive,
MIT Whitehead
Exploring genomics MicroRNAs in
Victor Ambros, Dartmouth, and functions zebrafish and frog Chad Nusbaum, Broad,
and Robert Horvitz, MIT of microRNAs development and Hui Ge, Whitehead
MicroRNAs in computationally High-throughput
worm development sequencing of microRNAs,
analysis of small interfer-
Fernando Camargo, ing RNAs in worms
target recognition
in blood cells

More than a third of the human genome is partially
regulated by microRNAs—tiny snippets of RNA that can
disable a gene’s ability to create proteins.
So it’s no surprise that the lab of Whitehead Member

David Bartel, the first to report this surprisingly widespread
role for microRNAs, has found many colleagues happy to
collaborate. At the same time, “as our lab looks at the par-
ticular targets of particular microRNAs, then we become
A snapshot of joint projects by interested in what’s going on in other labs that specialize in
those targets,” Bartel says.
David Bartel’s lab highlights the Here’s a glimpse at some current connections for the 20-
person lab—and it is just a glimpse. It shows only the prin-
crucial role of collaborations cipal investigators, not the postdocs and students who do all
the bench work, let alone the ongoing streams of informal
discussions. spring 2007 paradigm []
Last year, four billion gallons of
ethanol were produced in the
United States, while we con-
sumed about 140 billion gallons
of gasoline.

The power of yeast
genetics might make
ethanol fuel much
more cost-effective

By David Cameron

Biofuels and the gene pool
Your kitchen is stocked with Now, the benefits of so intimately On top of that, you need energy to
knowing this microscopic life form are ship the ethanol to regions of the coun-
one of the mightiest tools of reaching beyond biomedicine into the try where corn isn’t plentiful.
modern biology: yeast. realm of global warming. It’s pretty easy for critics to start
poking holes in this schematic. Because
Common, everyday baker’s Farming fuel when it comes to ethanol as an alter-
yeast, living in little packets As politicians finally get serious about native to oil, the energy return on the
the need for the U.S. to decrease energy investment is much slimmer
in your fridge, and diffused dependency on fossil fuels, there is than desired.
throughout your bread, beer one partial solution that they all like: This is precisely where yeast genet-
ethanol. ics can help.
and wine. But ethanol isn’t like crude oil. You Whitehead Member and yeast
This mundane single-cell can’t just drill down and then catch expert Gerald Fink has teamed up with
it as it gushes out. Instead, it takes a chemical engineer Gregory Stepha-
organism not only allows lot of energy to produce this colorless nopoulos of Massachusetts Institute
researchers to beta-test countless grain alcohol. The trick is to use the of Technology to create a genetically
least energy possible to produce the altered strain of yeast that promises to
genetic tools (many of which are most ethanol allowable. make ethanol production far more effi-
eventually scaled up for human So far, that’s been an elusive goal. cient—50 percent more efficient.
In the United States, ethanol is Ethanol is produced through fer-
cells) but is employed to screen produced chiefly from corn, and work- mentation. After the corn has been
drugs and even to study certain ing with corn demands a lot of energy. made into sugar, baker’s yeast metabo-
Everything from the growing process lizes the sugar, producing ethanol.
diseases such as Parkinson’s. For
stockbyte platinum

to producing fertilizers to harvesting But there’s an unfortunate irony
any molecular biologist working the crop requires oil. Then the corn to this procedure. Yeast doesn’t tol-
needs to be made into sugar, which is erate ethanol very well. In fact, at
today, it’s hard to overstate the turned into ethanol—which still needs certain levels, ethanol is toxic to it.
contributions of yeast genetics. to be distilled prior to commercial use. And because yeast is indispensable to

[] paradigm spring 2007
the process, there’s no way of getting “What we have provided is an controlled by many genes,” says
around using it. The end result is inef- enabling technology,” says Stepha- Alper. “Now others can apply this
ficient production. nopoulos. “A key component of this approach for making ethanol produc-
Many scientists have tried to engi- is that when we think of a cell that tion or other phenotypes of interest
neer ethanol-tolerant strains of yeast, makes a biofuel, the production of far more efficient.”
usually by tinkering with one or two that biofuel is not a property of a “This is a major contribution,”
key genes at a time. Hal Alper, a post- single gene or a single enzyme. The comments Michael Ladisch, a profes-
doctoral researcher in both the Fink production of ethanol is a property of sor at Purdue University’s Laboratory
and Stephanopoulos labs, decided to a whole network of reactions, all of of Renewable Resources Engineering.
take a different approach. which need to work together so that “This research demonstrates that etha-
Rather than homing in on a single the cell can make the molecule at effi- nol tolerance is not a simple phenome-
gene, he thought, why not target a reg- cient rates.” non. The fact that they’ve identified the
ulatory molecule that can affect many The greatest significance of this genes involved and can efficiently track
genes at once? research is that it opens up a new them is a major step forward.”
avenue for thinking about engineering “Yeast has been key to advances
Power yeast other desirable properties in a cell, the in basic biology and medicine,” notes
Transcription factors are nature’s researchers say. Fink. “I am very optimistic that this
equivalent to circuit breakers. Much as “Before this, we had very few yeast will also contribute to improving
one circuit breaker activates the elec- tools for improving a process that is our ability to make alternative fuels.”
tricity in many rooms in your house,
one transcription factor can control the
activity of a whole network of genes rather than hoMing in on a single gene,
in a cell. If the one-gene-at-a-time
approach couldn’t make yeast more
why not target a regulatory molecule
tolerant of ethanol, perhaps transcrip- that can affect many genes at once?
tion factors could.
Alper decided to focus on two tran-
scription factors. One of them, called
the TATA-binding protein, yielded
startling results in ethanol. When Alper
altered this transcription factor, it over-
expressed many genes, of which at
least a dozen proved sufficient to elicit
an improved ethanol tolerance. As a
result, this altered strain of yeast could
survive high ethanol concentrations.
Over a 21-hour period, it produced
50 percent more ethanol than normal

Gregory Stephanopoulos,
Hal Alper and Gerald Fink
celebrate their success in
illustration: tom dicesare; photo: donna coveney/mit

creating yeast that shows
higher tolerance for ethanol.

Here’s how
ethanol fuel is Enzyme Enzyme Yeast
created from
corn. Yeast
does all the
heavy lifting in
the fermenta-
tion process. Corn Grind Cook Turn Convert Ferment Distill Ethanol
to liquid to sugar spring 2007 paradigm []
to mouth
What can a frog mouth
tell us about human birth defects
and evolution?
By Alyssa Kneller

How to make a mouth
As early embryos, humans bear a striking resemblance to frogs.
Postdoctoral researcher Amanda
Both species comprise three basic cell types, arranged in the same Dickinson managed to map the steps
general pattern. And that isn’t surprising, considering we evolved required to make the primary mouth,
the first opening to form in the embryo,
from a common ancestor. and published her results last July in
But where does the likeness end? The jury is still out on which the journal Developmental Biology.
The primary mouth connects the
developmental processes we share. gut to the outside and allows feed-
“Our wild and crazy idea is that animals as different as sea ing. At a later stage, the secondary
mouth—which includes the jaws, teeth
urchins and humans use the same biological mechanisms to orga- and tongue—grows around this hole,
nize their heads,” says Whitehead Member Hazel Sive. allowing organisms to chew.
“We think the primary mouth might
Her lab is beginning to test this idea primitive animals. play a major role in positioning other
in the frog, Xenopus, which is easy to Sive hopes her research will yield parts of the face, which is one of the
study, and whose mouth is very similar clues about evolution. “This idea isn’t reasons it’s critical to understand how
to the human mouth. in any developmental biology text- this initial opening forms,” says Sive.
Once researchers understand this books yet, but the position of the cells Although other labs had studied
animal, they will look at other organ- at the front of all animal embryos is
isms, including more primitive animals remarkably similar,” she says. Amanda Dickinson, shown here with Whitehead
whose heads comprise just a hollow The similarity of mouth formation Member Hazel Sive, was inspired to probe the con-
cylinder of cells, to compare processes. in frogs and humans also will shed nections between species after playing with ani-
mals in Nova Scotia tide pools as a child.
Slice a frog embryo approximately light on the mechanisms behind human
12 hours after it is fertilized and you’ll craniofacial birth defects.
see three layers—endoderm, mesoderm Such defects, which range from
and ectoderm—which eventually give cleft palates to underdeveloped jaws,
rise to all of the animal’s tissues. account for three-fourths of all struc-
But the extreme front end (anterior) tural birth defects, according to the
of the embryo lacks mesoderm, (the National Institute of Dental and Cra-
middle layer), just comprising the ecto- niofacial Research. Cleft palates alone
derm and endoderm. Human embryos afflict roughly 8,000 newborns in the
exhibit the same pattern. United States each year.
In the most primitive animals, If frogs and humans use the same
there is no mesoderm anywhere in genetic circuitry to control formation
the animal. Sive thinks that the lack of the mouth and head, then scientists
of mesoderm in the extreme front of might be able to apply findings from
higher animals is a relic of ancient evo- one species to the other. That’s one rea-
lutionary processes, and a persistence son Sive’s lab is studying mouth devel-
of a process that occurred in the most opment in the frog.

[] paradigm spring 2007
particular aspects of primary mouth Whitehead postdoctoral researcher Amanda Dickinson mapped how a hole pokes through in a frog
development in a variety of model to form the primary mouth, or the first opening in the organism. First, a membrane between two
organisms, Dickinson was the first to layers of cells dissolves where the hole is destined to be. The cells begin to mix, and some of them
die as part of a thinning process. Finally, a hole forms, connecting the gut to the outside world.
tie this work together in a single critter.
Further, she is the first to approach this
process using molecular tools, with the stick to solid surfaces so they won’t
ability to identify the genes involved. “ the position be washed away. The Sive lab has pio-
Dickinson traced the movement of the cells at neered use of the cement gland as a
of individual cells as a frog embryo’s
gut tube poked through to the outside
the front of all “marker” for the extreme anterior of
the embryo.
world (see the image series above). Ini- animal embryos Humans lack cement glands, but
tially, the endoderm cells at the end of
the gut tube are encased by ectoderm.
is remarkably many aquatic vertebrates depend on
them for survival. A cement gland con-
The endoderm and ectoderm are kept similar.” —Hazel Sive sists of just one layer of cells, so it is
separate by a basement membrane easy to study. Li and others have iden-
between them. This basement mem- tified many of the proteins that turn on
brane begins to disappear where the thin like the membrane of a balloon, the genes that make the cement gland.
hole is destined to form, an oddity in pops open in the middle, producing the Li is building a detailed genetic circuit
biological terms. primary mouth. Dickinson and postdoc diagram that describes how this organ
“Normally, the endoderm and Colin DeBakker are now working to is positioned at the extreme anterior.
ectoderm never mix,” says Sive. “As pinpoint the networks of genes that “My job is to isolate the factors
the basement membrane breaks down, control each step of this process. that control which genes are expressed
these layers have to overcome their “The animal would have real prob- in the region, and more and more
hatred of one another.” lems if holes started appearing all over evidence shows that some of these
The sworn enemies grow flexible as the body, so we’re interested in the regulatory mechanisms play a role in
the materials between them disappear. genetic circuitry that coordinates for- multiple organs and multiple species,
In fact, the endoderm and ectoderm mation of just the right size hole in just perhaps even humans,” he says.
lose so much of their stiffness that they the right place,” says Sive. Thus Li’s analysis might help
cave toward the center of the embryo, Dickinson and DeBakker unravel the
forming a dimple on its surface. Cementing evolution factors that control primary mouth
opposite: kim furnald for furnald/gray

“At around the same time, the two New genetic analyses tie into other formation in frogs and shed light on
layers begin to mix, which is truly work in the Sive lab on the extreme the interactions between genes and
remarkable,” says Dickinson. “It’s as anterior region of embryos. proteins in humans. And the Sive lab’s
if you shuffled two decks of playing Postdoc Shuhong Li concentrates studies highlight how basic research on
cards.” on the genes and proteins that control seemingly esoteric organs or systems
As endoderm and ectoderm cells a much simpler organ—the cement can yield information that aids human
mix, some of them begin to die. The gland, which sits just below the pri- biomedical research.
mixed mass grows thinner and thin- mary mouth. Sive likens this organ
ner until it’s just a single layer of cells. to an underwater Post-it note, which David Cameron contributed to this
Finally, this layer, which is stretched secrets mucus to help frog embryos story. spring 2007 paradigm []
Today’s advances in systems biology began with
genome sequencing. Hui Ge is among those cre-
ating more powerful next-generation platforms
that integrate protein interactions and other
kinds of high-volume analyses as well.

as opposed to the more traditional
hypothesis-driven approaches,” she
adds. “We put together all this high-
throughput information, and then we
can find predictions for uncharacter-
ized genes and connections between
them, which we can test. And this can
be an iterative process in the lab. You
make predictions and validate them,
and then that validation can help your
prediction techniques.”

Reading the “interactome”
Ge graduated with a bachelor’s degree
in biochemistry and molecular biology
from the Beijing University in 1999,

Network news

Hui Ge sifts through oceans of data to explore
how genes collaborate By Eric Bender

Consider the worm. This is the realm of systems biol-
For a multi-cellular organism, ogy, and of Whitehead Fellow Hui
Ge, who studies embryonic develop-
Caenorhabditis elegans keeps ment in the worm.
things simple. A typical adult “I’ve always been interested in
how a fertilized egg develops into Ge and her co-workers in Marc Vidal’s Harvard
roundworm has 959 cells, no a whole organism like us,” says Ge. lab mapped out interactions between many
proteins in the C. elegans worm, in this 2004
more and no less, and scientists She gets the big picture on this process Science paper.
by combining data from several high-
have traced the exact lineage throughput analysis techniques that and won a scholarship to Harvard
of each cell. The animal goes cut wide swaths through the worm Medical School.
genome, and using advanced statistical She arrived just as systems biology
through life without a brain or methods to sort through the results. began to soar.
much of a sex life (almost all are This systems biology approach “I fell in love with the idea that you
photo: sam ogden; graphic courtesy of science

starts with the components of a system, can study how the organism works,
hermaphrodites). and studies how those components not just by studying individual genes
But C. elegans also has about work together to achieve a certain but understanding what a lot of genes
function, such as protein synthesis do at a time,” she says.
19,000 genes—almost as many or protein degradation. “It’s impor- In a homework assignment for
genes as humans. And just as in tant to know not just the individual a class taught by Harvard’s George
components of a cell but how they Church, Ge came up with a computa-
humans, no gene does its work are mapped together,” Ge notes. “It’s tional strategy that eventually turned
alone. Instead, tasks are accom- like a subway system; if you remove into a paper in Nature Genetics.
a station, the effect on the system will The strategy was about correlating
plished through highly complex depend on its position.” data from large-scale studies of genes
networks of protein interactions. “These are data-driven approaches with large-scale studies of protein

[10] paradigm spring 2007
interactions. More specifically, it was four differentiated cells, in the first says. “These genes are not function-
to correlate transcriptome data (reflect- hour after fertilization. ally equivalent, but they complement
ing which genes a cell expresses under The scientists combined data each other’s function. Knowing these
certain conditions) with interactome from three sources: protein-protein kinds of genetically buffering pairs
data (mapping interactions between interaction, gene expression, and is very important for understanding
the proteins). Where the two data sets loss-of-function profiling based on development but also for understand-
agreed, clearer pictures of protein roles RNA interference. They then made ing disease.” She gives the example of
would emerge. predictions about how the embryonic the mammalian p53 tumor suppressor
“She was one of the first research- “molecular machines” work. Testing gene: “Even if you knock down p53,
ers to suggest putting data together 10 uncharacterized proteins by seeing a mouse will not get cancer immedi-
from very different data sets to get where they popped up in live animals, ately. It increases the chance that when
something that was better than the the researchers found that the locations something else is damaged, the mouse
sum of the parts,” says Harvard’s generally were consistent with the pro- will get cancer.”
Marc Vidal, in whose lab Ge ended teins’ predicted roles, findings reported The second major effort is to take a
up. in a 2005 Nature paper. dynamic view of gene expression that
Doing research on yeast, Ge and Completing her PhD in genetics, integrates time and space information.
her co-workers demonstrated “the first Ge was picked as a Whitehead Fellow. “In multi-cellular organisms, at differ-
global evidence that genes with similar Before starting at the Institute, though, ent locations, the molecular networks
expression profiles are more likely to she spent six months at the lab of Har- are actually different, because not all
encode interacting proteins,” as their vard’s Craig Hunter, learning the craft of the genes are expressed at the same
paper put it. And they showed that the of worm wet-lab work. time,” Ge says. One student in her

In a 2005 Nature paper, Ge and colleagues Next, the researchers filtered out a network When the researchers zeroed in on “sub-
combined maps of protein interactions (blue), “backbone” that grouped proteins shown to networks,” they found that proteins whose
gene expression (red) and loss-of-function be interacting by at least two sets of data. functions were already known helped to char-
profiling from RNA interference studies (red). acterize unknown proteins nearby.

integrated data could help to improve Healthy pairs of genes
hypotheses generated from either At Whitehead, Ge and colleagues have lab, for instance, is adding data about
approach alone. embarked on two main projects with location and trying to predict the genes
the worm, the first being further explo- that are expressed in certain tissues
Getting worm rations of genetic interactions during such as muscles or skin.
Also at the Vidal lab, Ge worked on embryonic development. “Quantitative science is providing
a big project to map out much of the RNA interference studies have us with a process that helps us under-
interactome of C. elegans. Eventually highlighted about 2,500 genes whose stand biology more quickly and in a
published in Science, this paper had loss kills the worm embryo. That num- systematic way,” says Ge. “Little by
no fewer than eight other co-first- ber seems pretty small compared to the little, we are learning how to achieve
graphics courtesy of nature

authors. (The worm was the likely total of around 19,000 genes, she says, these projects.”
target because it was the first multi-cel- and she suggests that it’s because genes Today’s attempts at detailed molec-
lular organism to be sequenced com- can back up each other’s functions. ular modeling are “still first draft,
pletely, in 1998.) “We are combining the protein still relatively fuzzy—just like genome
Next, Ge and colleagues tackled interaction map with the genetic inter- sequencing once was,” acknowledges
very early embryogenesis—the process action map to predict these pairs that Ge’s mentor Vidal. “But they are really
by which the worm divides twice, into give you a synthetic phenotype,” Ge shaping up.” spring 2007 paradigm [11]
“Find something that interests you, and go for it.”
That’s what David Baltimore told postdoctoral researcher
Naomi Rosenberg when she entered his MIT lab back in
1973. And one topic that interested Rosenberg was leukemia.
In graduate school Rosenberg had used cell culture
techniques to study different diseases, but those techniques
didn’t yet exist for this deadly cancer of the blood.
After perusing the scientific literature, she came across
the Abelson virus, a pathogen that
caused rapid tumor growth when
injected in mice. “The speed with
1960 which it caused leukemia intrigued
Peter Nowell and me,” recalls Rosenberg, now a
David Hungerford professor at Tufts University Medi-
discover that patients cal School.
Rosenberg began trying vari-

A slow
with chronic myelog-
enous leukemia (CML) ous methods for infecting healthy
have a unique chro- blood cells in culture. Then, in
mosome, soon 1975, she published her success,

saga of
named the Philadel- using the Abelson virus to induce
phia chromosome.

success leukemia in mouse blood cells. “For the first time we had
If you’ve ever won- a way to study in a controlled environment how this virus
dered why the journey interacts with its target,” she says.
What Rosenberg didn’t know at the time was that this
from lab discovery to postdoctoral success was one link in a profound—and
the clinic takes so long, unsuspected—chain of events that three decades later would
culminate in a spectacular cancer drug: Gleevec.
follow the decades- Gleevec treats chronic myelogenous leukemia (CML),
which strikes about one-fifth of all leukemia patients—
long story of Gleevec roughly 1.5 cases per 100,000 people. Before Gleevec, the
fatality rate of this disease was 100 percent.
By David Cameron
Brought to market in 2002, Gleevec represents a revo-
lution in cancer treatment. Rather than carpet-bombing
the body with toxins that wipe out the cancer but incite a
range of devastating side-effects, and very often fail any-
way, Gleevec targets a specific molecular abnormality of
CML. Although the drug doesn’t eliminate the cancer, for
the majority of patients it knocks it into a kind of perma-
nent remission. As long as patients take Gleevec daily, CML
becomes a chronic, and manageable, disease.
Gleevec points to the future of cancer treatment, but it
also typifies the serendipitous nature of basic research—and
the agonizingly long road that even the most dramatic suc-
cess stories must follow.
“The journey from the lab bench to the clinic is a slow
process,” says Whitehead Member Robert Weinberg. “Most
people fail to appreciate the time it takes for a discovery to
result in a drug. It would be wonderful if these things turned
around quickly. But this process always has—and probably
always will—take time.”

[12] paradigm spring 2007
Fusing research The human connection
One can argue that the first major Gleevec-related Gleevec is so effective because CML has a clear target.
discovery occurred in 1914 when Theodor Boveri, a As horrible as the disease is, all of its symptoms completely
German cytologist, proposed that cancer results from depends on a single protein, one that the drug disrupts.
defects in a cell’s chromosomes. However, most people In 1984, researchers from both the United States and the
mark 1960 as the year the story began. That’s when Peter Netherlands discovered the strange genetic signature that
Nowell of the University of Pennsylvania and David produced this key protein.
Hungerford of Fox Chase Cancer Center noticed that cell Again, while human can-
samples from CML patients often contained an abnor-
mally small chromosome, soon dubbed the Philadelphia
1978- cers generally are not caused by
viruses, understanding how the
chromosome. 1980 virus worked, and which genes
In 1973, Janet Rowley of the University of Owen Witte finds the it interacted with in animals,
Chicago found that this chro- protein that the Abel- provided scientists with clues for
mosome was a kind of hybrid, son virus produces where to look in human cells. A
the result of chromosomes and discovers its func- team led by Gerard Grosveld, then
9 and 21 swapping genetic 1973 tion. In the same way at Genetics Erasmus University in
material. Janet Rowley finds that that the Philadelphia the Netherlands, announced that it
Between these two the Philadelphia chromo- chromosome is a had located the human version of
findings, another piece of the some is a hybrid of two hybrid, the Abelson the Abelson virus gene. While nor-
Gleevec puzzle was discov- normal chromosomes viral protein appears mal human cells contain a healthy
that have fused together. to be a hybrid.

1969 ered—one that then lacked any
1975 1984
Herbert Abelson Naomi Rosenberg Gerard Grosveld discovers the gene that
apparent connection with the
discovers a virus that uses Abelson virus to causes CML. The BCR/Abl gene, located
Philadelphia chromosome.
causes virulent leuke- transform normal cells on the Philadelphia chromosome, cre-
Herbert Abelson, then an MD
mia in mice. The virus into leukemic cells in ates a protein that, just like the Abelson
at Children’s Hospital in Bos-
is named after him. a Petri dish, develop- viral protein, is a fusion product.
ton, discovered in 1969 a virus
ing a platform with
that caused leukemia in mice.
which researchers can
(Named after its discoverer, this
more effectively study
was the virus that Rosenberg would use to develop her version of the Abelson gene, in
the disease.
cell culture platform six years later.) CML patients, this gene turned out
Since the Abelson virus was one of many able to pro- to be located right on the Philadel-
duce tumors in animals, scientists reasoned that viruses phia chromosome.
also caused tumors in humans. But this view was gradually Now that scientists had the gene’s address, Witte imme-
overturned in the late 1970s and early 1980s, when Wein- diately started to look for the protein that the human Abel-
berg and many other researchers demonstrated that gener- son gene expressed. He discovered that cells from CML
ally the real culprits were genetic mutations. patients expressed an over-sized variant of the normal Abel-
Still, years of research with cancer viruses were son protein. “It was very strange,” he says.
hardly in vain. In 1976, shortly after Rosenberg’s success Like the viral protein, the human CML protein was a
in culturing leukemia, a new postdoc in the Baltimore lab, kind of hybrid, the byproduct of two unrelated genes (the
Owen Witte, collaborated with Rosenberg to identify the BCR and Abl genes) fusing together. Called BCR/Abl, this
protein expressed by the Abelson virus. Like many other mutant gene on the Philadelphia chromosome created a pro-
cancer viruses, the Abelson virus is so tiny that it produces tein that, like the gene, consisted of two components that
a single major protein. Witte and Rosenberg found that this in normal cells existed apart from each other. This was the
protein was a hybrid, most likely a result of the viral gene over-sized protein that Witte had discovered in CML cells.
fusing with a normal cellular gene. But observing the protein was one thing. Proving it
“This was a mystery,” recalls Witte. “We had this pecu- caused cancer was another.
liar protein, but we still didn’t understand what it did.”
At this time, there was no obvious relation between this A silver bullet
fusion viral protein in mice and the fused human chromo- In 1984, David Baltimore, who had become Whitehead Insti-
some that Rowley had found. tute’s first Director, moved his lab across the street from MIT spring 2007 paradigm [13]
into the new Whitehead building. Shortly after this, MIT ture, this fusion protein evaded cellular regulation. The
graduate student George Daley joined Baltimore’s group. BCR/Abl protein was a kinase gone wild.
Daley (who would eventually become a Whitehead Fellow) By the mid-80s a number of researchers were inves-
arrived at the lab already interested in CML in general, and tigating whether kinases could be potential drug targets.
in the newly discovered BCR/Abl gene in particular. One of these was Brian Druker, an oncologist working
“What wasn’t clear,” says Daley, now a professor at with Thomas Roberts at Dana-Farber Cancer Institute.
Harvard Medical School, “was whether or not BCR/Abl Druker had chosen Roberts’s lab over the clinic because
was the gene that initiated the disease. Weinberg’s research, he wanted to understand cancer on the molecular level.
for example, was suggesting that most cancers needed muta- “Even though we were getting better at using chemother-
tions in a number of key genes to develop, not just one.” apy to treat cancers like childhood leukemia, the drugs
Experiments with cell cultures from both the Baltimore had terrible side effects,” he says. “What’s more,
lab and Witte’s UCLA lab suggested that the gene could we didn’t even understand what they did.”
transform normal human cells During his nine years at
into cancer cells that resembled Dana-Farber, Druker began
CML. to focus more on CML, col-
“But an animal model was 1990 1996 laborating on and off with
still missing,” says Daley. “In George Daley creates Brian Druker and Nicholas two Swiss pharma compa-
order to provide the conclusive animal models of CML, Lydon discover a chemical nies, Ciba-Geigy and Sandoz.
evidence that BCR/Abl was proving that the BCR/Abl compound that blocks Although few companies
indeed the culprit, we needed gene is sufficient to the BCR/Abl protein in were investing heavily in
cause the disease. human cells.

to demonstrate that it, and it alone, could initiate CML in drugs that blocked kinases, Druker
Clinical trials begin for
an animal.” was convinced that CML could
this compound, which
Many research labs were trying to do this, mostly respond to such a novel approach.
is named Gleevec.
through incorporating BCR/Abl into the animals’ germ “It was a well-defined disorder that
lines. But the BCR/Abl gene was toxic to germ cells, and we knew resulted from an acti-
most of these mice died in utero. vated kinase,” he says. “Block the
Daley tried a different route. kinase, and you’d topple the disease. Plain and simple.”
Using the methods that then-Whitehead Member Rich- In 1993 Druker left Dana-Farber and took a position
ard Mulligan had developed for transferring genes into at Oregon Health & Science University. “I had one goal at
blood stem cells, Daley transferred the BCR/Abl gene into the time: to find a company that had an inhibitor for BCR/
the bone marrow of mice. Abl and to bring it into the clinic.”
He then took that bone marrow and transplanted it Druker contacted Nick Lydon, a scientist at Ciba-
into a second group of mice whose own marrow had been Geigy. Lydon had developed a number of small kinase-
destroyed by radiation. blocking compounds that Druker wanted to test. Since
And the second group of mice developed CML. kinases pass their phosphate messages by physically inter-
“If you think about it,” jokes Daley, “this was really acting with other proteins, almost like two Lego pieces
gene anti-therapy.” This experiment proved that the BCR/ snapping together, the hope was to find a tiny molecule
Abl gene was sufficient to cause CML. “We now knew that that could wedge itself inside the exact spot where the two
for this disease, BCR/Abl was the fundamental drug target,” proteins fit, thus obstructing the message. The caveat was
he says. that such a molecule must be so specific that it could only
disrupt BCR/Abl. And while kinases are not as uniform as a
Killing with kinase box of Legos, they are similar enough to make this a daunt-
kim furnald for furnald/gray

Back in 1980, when Owen Witte first discovered the Abel- ing challenge.
son virus fusion protein, he also found that it belonged to a While Druker was screening Lydon’s compounds in
family of proteins called kinases. A kinase sends messages human bone marrow cells, one named STI571 stood out. By
through the cell by adding a phosphate to other proteins, targeting a section of the BCR/Able protein called the “cata-
which in turn affects those proteins’ activity. lytic cleft,” this compound immobilized its ability to transfer
What distinguished the over-sized BCR/Abl protein that phosphates to other proteins. Healthy cells were unaffected.
Witte had identified from ordinary kinases—and from the “At that point I knew we had a potential drug,” says
normal Abl protein—was that because of its mutant struc- Druker.

[14] paradigm spring 2007
and Australia). After the treatment, over
90 percent of people diagnosed with
a fairly early stage of the disease were
free of symptoms. About 60 percent of
patients with advanced CML experienced
brief remission, with relapse often occur-
ring after a few months.
“But understand,” says Druker, “for
years I’d been treating patients with the
disease, telling every one of them that
they’d be lucky if they lived five years.
And then this! This is one of the best
examples I’ve ever seen of science tri-
umphing over disease.”
“It’s unlikely that we’ll find a drug
for breast or prostate cancers that works
exactly like Gleevec,” cautions Daley.
“These more common cancers typically
aren’t caused by a single mutated protein.
There are usually a few BCR/Abl-like pro-
teins at work. But what we can do is try to
develop cocktail
2002 drugs, therapies
that have two or
After stunning success
three compounds
in CML patients, Gleevec
that knock out a
is approved by the Food
handful of path-
and Drug Administration.
ways at once.”
While Gleevec
has also proven
effective for certain rare forms of gastro-
intestinal cancer and the blood condi-
tion hypereosinophilic syndrome, it isn’t
the only such show in town. Iressa and
Tarveca, lung cancer drugs, and Her-
ceptin, a breast cancer drug based partly
on Weinberg’s research, also successfully
target specific molecular signatures. Many
similar drugs are in clinical trials.
Many people taking Gleevec today
were not even alive when the Philadel-
phia chromosome was first discovered,
nearly 50 years ago. During this period,
In 1996, Druker and Lydon published these findings, the our understanding of the very nature of cancer dramatically
same year that Ciba-Geigy and Sandoz merged and formed changed, and thanks to the persistence of researchers such
the pharmaceutical giant Novartis. While Druker became as Druker, the genetic revolution that began in the 1970s
the academic advocate for STI571, Lydon and Alex Matter, has finally arrived at the clinic.
director of the Novartis Oncology Research unit, continued Still, the need for fundamental science has never been
to push STI571 into clinical trials, despite lingering skepti- greater.
cism that simply blocking a kinase could hamper such a “Every once in a while I’ll hear someone suggest that
deadly disease. we’ve done enough basic research, and now all our energy
should be focused on applying it,” says Weinberg. “Nothing
On trial can be farther from the truth. There are still many signaling
The first human trials of the drug occurred in 1998. All 31 pathways operating within cancer cells that we just don’t
patients experienced complete remission. know about yet. Only by meticulously researching how all
Over the next four years, 6,000 people entered into clini- these other cancers work will we be able to build an arsenal
cal trials with STI571, renamed Gleevec (Glivec in Europe of drugs that disable this disease at the root.” spring 2007 paradigm [15]
Cancer researchers look beyond the genome
to the epigenome—and the role of methyl marks

By Alyssa Kneller

More than 10,000 people in the U. S. will die
of kidney cancer in 2007, the American Cancer Society predicts.
And often, it won’t be genes gone bad that get them.
Scientists have known for decades that cancer can be caused by genetic mutations. But they’ve recently
discovered another culprit—the tiny methyl group, which consists of a single carbon atom surrounded by
hydrogen atoms. When this chemical group shows up in the wrong places on an otherwise normal strand
of DNA, it can cause cancer.
In 1994, two groups showed that about 57 His lab uncovered an interesting pattern. In
percent of patients with the most common form roughly 20 percent of the tumors, the DNA bases
of kidney cancer harbor a mutation on the von forming the VHL promoter (the region where
Hippel-Lindau (VHL) tumor-suppressor gene. proteins bind to activate the gene) have acquired
This finding led some doctors to wonder about extra methyl groups. However, the sequence of
the remaining 43 percent—how do they arise? the DNA bases in the whole VHL gene is usually
Stephen Baylin, professor of oncology at the normal, indicating that the gene has not suffered a
Johns Hopkins University, and his colleague mutation. Baylin and his colleagues hypothesized
James Herman, now an associate professor at that the extra methyl miscreants were guilty of
the same institution, decided to delve deeper into shutting down the otherwise normal gene. Scien-
yasuhiro yamada

this medical mystery by taking a closer look at tists had already shown that methyl groups block
the VHL gene in patients with the non-hereditary access to DNA, preventing it from being read out,
form of this cancer. so this was a logical conclusion.

[16] paradigm spring 2007
In one breed of mice,
intestinal tumors
form in two distinct
stages, which are
partially regulated
by epigenetic events,
including misplaced
methyl groups. First,
microscopic tumors,
such as the one in
the center of the
top image, develop.
Given the right cir-
cumstances, these
growths progress
into macroscopic
tumors (bottom). spring 2007 paradigm [17]
Although Baylin was hardly the first scientist to observe
odd methylation patterns in the DNA of tumors, he was
the epigenome lets each type of
among the first to produce evidence that this might play a cell access parts of the genome
major role in cancer formation. A deluge of papers came out
around that time, including a key one by Whitehead Mem-
for its own particular needs
ber Rudolf Jaenisch, providing irrefutable proof that mis-
placed methyl marks can contribute to cancer formation. of a skin cell. At least 200 different types of cells comprise a
“I think the VHL gene was precociously trying to tell us human being, and each one contains a different epigenome.
something,” Baylin says. “If you find a gene that has lost its Given their essential functions, epigenetic marks hardly
function via a mutation, then you can probably find cases serve as DNA accessories. But they can be changed like
where that gene has lost its function via a modification to a pair of earrings or a necklace. For example, an enzyme
the epigenome.” called Dnmt3a places methyl marks on previously unmeth-
ylated DNA. Typically active in developing embryos, this
Marking up DNA—and passing it on enzyme helps to establish tissue-specific DNA methylation
So what’s the epigenome? patterns.
You can think of it as the system that lets each type of Importantly, such marks are replicated during cell divi-
cell access parts of the genome for its own particular needs. sion and passed to daughter cells. Thus epigenetic marks are
The epigenome serves as a firewall, hiding certain genes transient in one sense, yet heritable in another.
while exposing others. For example, a few methyl groups on “This dichotomy is one of the reasons we’re study-
the promoter of a gene can keep it concealed and silent in a ing epigenetics as it relates to cancer,” says Heinz Linhart,
particular tissue. Though methyl marks are the best under- an MD/PhD in Jaenisch’s lab. “Epigenetic marks provide
stood epigenetic marks, there’s another major group—pack- potent therapeutic targets because they can be added or
aging proteins. For example, some proteins block access to stripped, but we wouldn’t be talking about them if they
genes by coiling bits of the sequence into neat “spools.” weren’t heritable. Neither mutations nor misplaced methyl
Epigenetic mechanisms usually help cells express genes marks would induce tumors if they were diluted out when
at the right time and place. While all of an organism’s cells cells divided.”
share the same genes, epigenetics ensures that a brain cell Linhart manipulates the epigenomes of mice to explore
produces dopamine, serotonin and other “brain” chemicals methylation patterns previously linked to tumor formation.
rather than keratin, fats and oils, which are characteristic He tinkers with methyl marks and watches the results—an
approach that allows him to establish cause and effect.
Jaenisch used a similar approach more than 10 years ago
Cancer stem cells to silence critics of the first studies that provided evidence
that epigenetic changes can produce tumors.
and epigenetics Tale of a tumor
A growing number of Nature Genetics, includ- In 1994, the same year Baylin completed his kidney cancer
scientists accept that not ing one by Stephen Baylin, study, Jaenisch began to study methylation in tumor-prone
all cells in a tumor are cre- professor of oncology at the mice. Though healthy in most respects, these animals
ated equal. They believe Johns Hopkins University, develop large numbers of tumors in their intestines.
that a small population of link methylation patterns Ironically, Jaenisch suspected that missing methyl groups
“stem cells” gives rise to in cancer cells to patterns might be to blame. Around 1980, scientists had noticed that
the slightly differentiated of DNA-packaging proteins the DNA of many tumor cells was missing methyl marks,
cells that form the bulk of a in embryonic stem cells. but they didn’t have the tools to probe the relationship.
tumor. The cancer stem cells The DNA-packaging pro- Furthermore, renowned cancer researcher Bert Vogelstein
divide less frequently than teins could leave particular of Johns Hopkins had observed this “hypomethylation”
their specialized daughter genes, those involved in pattern in the tumor-prone mice and proposed that it was a
cells, but live forever. keeping a cell specialized prerequisite for polyp formation.
“It’s still a matter of faith rather than immature, vul- In collaboration with Whitehead Member Robert Wein-
that the stem cell model nerable to methylation in berg, Jaenisch and postdoctoral researcher Peter Laird (now
applies to all cases of can- adult cells. a professor at the University of Southern California) stripped
cer, though the evidence is “It’s certainly possible methyl groups from the DNA of their pint-sized subjects and
compelling for a small num- that these patterns are waited for the animals to develop tons of tumors.
ber of solid tumors,” says fundamentally linked to The results, published in Cell in 1995, were startling.
Whitehead Member Robert formation of cancer stem Rather than mimicking tumor formation, these mice
Weinberg. cells, but this needs to be produced fewer tumors. “Though we were puzzled by the
Three recent studies in proven,” says Baylin. outcome, we were pleased to establish a causal relationship
between methylation and cancer,” says Jaenisch, who also is

[18] paradigm spring 2007
Eva Moran and Heinz a biology professor at Massachusetts methylated these hot spots, the mice developed more macro-
Linhart manipulate mice Institute of Technology. scopic intestinal tumors than usual.
epigenomes by tinker- A decade later, Japanese patholo- The pair dug deeper and identified a key growth-control
ing with methyl marks,
and then see how that gist Yasuhiro Yamada joined the gene affected by the misplaced methyl groups. Their find-
affects tumor formation. lab. Yamada was particularly ings, which should be published this spring, provide an
knowledgeable about the mice used interesting twist to the intestinal-tumor tale.
in the experiment. He knew that “In these mice, intestinal tumors arise through a complex
their intestinal tumors developed in two distinct stages. First interplay between genetic events, global hypomethylation
come microscopic tumors that resemble flowers. Given the and local hypermethylation,” says Linhart, who is still teas-
right conditions, these grow into massive irregular tumors ing apart the details of this relationship.
that can be seen with the naked eye (see photos on page 17). The story of the intestinal tumor demonstrates once
Yamada repeated the 1995 study and discovered that again that cancer is rarely simple. The term encompasses a
hypomethylation increases the number of tiny tumors but multitude of diseases characterized by the abnormal pro-
decreases the number of large tumors. The earlier research- liferation of cells. Each of these diseases has its own story
ers missed the microscopic effect. filled with its own characters, ranging from genes to viruses
“Our lab had just shown that global hypomethylation to methyl groups.
destabilizes DNA big time, so we reasoned that the small “Epigenetics will not provide a universal cure for cancer
tumors result from chromosomal instability rather than epi- because it does not cause every instance of the ‘disease,’”
genetic silencing,” Jaenisch explains. says Linhart. In fact, it might offer more promise as a diag-
Linhart and MIT diploma student Eva Moran took nostic tool. A growing body of evidence suggests that most
kim furnald for furnald/gray

the study one step further by setting new methyl marks tumors exhibit epigenetic changes regardless of their origin.
randomly on the DNA of the tumor-prone mice—a gain- So epigenetic patterns could be used to diagnose particular
of-function study as opposed to the many loss-of-function types of cancer, even those caused by genetic mutations. But
studies done previously. scientists caution against losing sight of the big picture.
In most tumor cells, DNA is unusually short on methyl “Although methylation changes can be just as important
groups. Yet the same cells often contain short sequences as mutations in particular cases, epigenetics is just one very
replete with methyl groups, hot spots that typically fall on narrow part of the broad cancer research field,” Weinberg
the regulatory regions of genes. After Linhart and Moran explains. spring 2007 paradigm [19]
Strategically placed methyl groups (shown in red)
block access to key regions of DNA, keeping specific 1980s. He was also the first to change
genes silent. There are two known types of epigenetic those patterns by treating cells with
marks—methyl groups and DNA-packaging proteins. chemicals.
To see how methyl groups work on one gene, see
Whiteboard on page 22.
One of the chemicals he used was
azacytidine—which became Vidaza,
25 years later. Jones believes other
success stories will follow.
“I think these drugs will find much
more use in the future because they’re
very good at resetting the epigenetic
program, which has gone awry in a
cancer cell,” he says.
But Jaenisch worries that compa-
nies will rush to create drugs before
methyl group fully understanding the consequences
methyl group of taking them. He cautions scientists
to search for side effects before apply-
ing epigenetic therapies. This warn-
ing comes from experience. When
the Jaenisch lab reduced the number
of methyl marks on the DNA of
tumor-prone mice, the animals developed fewer macroscopic
“ The good news is tumors in their intestines. But in another study, the lab found
that epigenetic marks are that loss of methyl marks can cause aggressive lymphomas.
reversible.” —Rudolf Jaenisch “If you want to use methylation changes as a therapeutic
tool, you have to know what you’re doing,” says Jaenisch.
Stephen Baylin is more optimistic. He points out that
Ready for drugs? Jaenisch tinkered with methylation patterns in mouse
Epigenetic marks have attracted attention from pharma- embryos, when enzymes were still busy setting and stripping
ceutical companies hoping to reverse them. In 2004, the methyl marks. He wonders if the lymphomas can be blamed
Food and Drug Administration approved Vidaza, a DNA- on timing, rather than on the treatment itself. Would mice
demethylating drug manufactured by Pharmion Corpo- develop these lymphomas if they were exposed to a demeth-
ration, for use in certain blood diseases such as chronic ylating drug as adults?
myelomonocytic leukemia. Vidaza is believed to work Despite this debate over side effects, Jaenisch agrees that
indirectly by reducing DNA methylation and directly by epigenetic therapies will eventually become a reality. “These
killing cells. therapies should materialize after we develop a robust
This approval was the realization of a dream for Peter understanding of the mechanisms involved,” he says. “The
Jones, director of the University of Southern California/Nor- good news is epigenetic marks are reversible, which gives us
ris Comprehensive Cancer Center. hope to treat thousands of cancer patients someday.”
Jones was one of the first researchers to observe methyla-
tion patterns in cancer cells during the late 1970s and early David Cameron contributed to this story.

Epigenetics and the environment her lab at the Queensland Institute of
Medical Research. She points out that
“The epigenome allows the genome to lower their risk of colon cancer by 20 some epigenetic marks in plants fluctu-
talk to the environment,” says White- percent. Folate happens to be a methyl ate throughout the day as light levels
head Member Rudolf Jaenisch. In fact, group donor, so perhaps it protects the change. It’s not unreasonable to hypoth-
the epigenome might explain the link women by acting on the epigenome. esize that epigenetic marks fluctuate in
between particular diets and increased Or perhaps not. “We don’t know humans over a period of days or years in
or reduced risk of cancer. if folate modifies the transcriptional response to diet.
The long-term Harvard Nurses’ state of certain genes, but I do suspect “I think we’re going to discover a
christina ullman

Health Study, for example, showed that that people have underestimated the lot of layers of epigenetic modification
women who take a multivitamin pill plasticity of epigenomes,” says Emma (beyond methylation), and some will be
containing folate (a form of vitamin B9) Whitelaw, who studies epigenetics in more stable than others,” she says.

[20] paradigm spring 2007
Out of sequence?
of the budget twice as efficient,” he
says. “And the Human Genome Proj-
ect taught us that setting a goal often
causes costs to drop as the technology
Scientists debate whether it’s time to tackle grows more efficient. This makes more
science possible.”
tumor genomes and epigenomes By Alyssa Kneller
Sequencing the epigenomes
A group of scientists recently
On a cold January after- proposed another giant proj-
noon, Eric Lander sits in his ect—sequencing several human
office at the Broad Institute, epigenomes—to advance cancer
explaining a new effort to research. The epigenome lies
sequence the DNA of tumors, above the DNA sequence and
when his cell phone rings. He includes methyl marks as well
answers and listens. as proteins that package DNA
“Do you want me to come (for more information see “The
over to the hospital tonight?” he Unusual Suspect” on page 16).
asks, then covers the mouthpiece Aberrant methylation plays a
and whispers, “I just want to role in many types of cancer.
make arrangements to visit my As president of the Ameri-
cousin. Unfortunately, he has a can Association for Cancer
very serious cancer, so this is quite Research, Peter Jones helped to
relevant to our conversation.” develop a blueprint for an inter-
Lander’s cousin was diagnosed national project to sequence
with a type of bile duct cancer at several human epigenomes.
the end of December. This form of Eventually, he hopes to com-
cancer is relatively rare, and with- pare the epigenomes of cancer
out known therapies. patients with those of healthy
Getting off the phone, Lander individuals.
notes that “we could try to use “The epigenome is the miss-
therapies that were developed for other funding should emphasize small, inves- ing piece between genes and proteins,
cancers to treat my cousin, but we tigator-initiated projects rather than and we need the sequence to fully
don’t know which genes are involved, large, collaborative ones. understand cancer and make accurate
and that’s very frustrating.” “I’m not at all convinced that, diagnoses,” says Jones, director of the
The Broad is participating in a new given the current NIH grant funding USC/Norris Comprehensive Cancer
federally funded venture that could climate, we can afford to invest enor- Center. Like Lander, he believes that
help patients such as Lander’s cousin. mous amounts of money in sequencing larger-scale projects make smaller-scale
The National Cancer Institute recently cancer genomes,” says Weinberg. projects more efficient.
kicked off a pilot project to sequence “The larger-scale projects generate But the proposal faces critics. Some
the DNA of tumors from patients with large databases, which are useful to wonder if we know enough about the
particular types of cancer. If the pilot smaller-scale scientists, but on their epigenome to launch such a monumen-
succeeds, the government might fund own, larger-scale projects hardly yield tal project. Labs are still uncovering
a massive cancer-sequencing project to as much conceptual bang for the buck layer upon layer of epigenetic marks
determine the genetic components of as smaller-scale projects,” he declares. that current sequencing technologies
all types of cancer. Many involve muta- “I think there are still some major miss. And the project requires sequenc-
tions to the same genes, and the result- unsolved conceptual problems that ing numerous epigenomes, because
ing information would help researchers only smaller, more focused research each cell type contains a different set of
map relationships between cancers. efforts can address.” epigenetic marks.
But some scientists feel that NCI Lander, who is both a Whitehead “My feeling is an epigenome proj-
should wait to launch the project when Member and Director of the Broad, ect is a bit premature,” says Emma
federal funding for biomedical research counters that the pilot project will cost Whitelaw, of Queensland Institute of
expands. The National Institute of just 0.5 percent of the NCI budget. Medical Research. “We could save a
Health’s budget doubled between 1998 “I can’t imagine why you wouldn’t lot of money by waiting a few years,
james o’brien

and 2003 but is now dropping in real spend 0.5 percent of the budget on by which time we should know more
dollars each year. Whitehead Member cancer genome sequencing, because about what to look for and where to
Robert Weinberg suggests that grant it would make the other 99.5 percent look for it.” spring 2007 paradigm [21]

whiteboard the link
between Epigenetics and Cancer
A. Non-cancerous cell One copy of a single chromosome pair is methylated.


helps turn on
the H19 gene.

H19 gene

Insulator protein binds between
two genes, keeping the enhancer
away from the first one.

Interaction results in
H19 gene product
lgf2 gene

helps turn on
the lgf2 gene.

Interaction results in
lgf2 gene product H19 gene

paternal Methyl groups block access to the
chromosome DNA between the two genes, pre-
venting the insulator protein and
enhancer from binding.

[22] paradigm spring 2007
here are two known types of of a carbon atom surrounded by hydrogen. graphic
epigenetic marks—methyl groups and In the example on the right, misplaced methyl Christina Ullman
DNA-packaging proteins—which help groups on one copy of a single chromosome SCIENTIFIC ADVISOR
cells turn on specific genes at the right time contribute to cancer by disrupting the balance Heinz Linhart
and place. Strategically placed methyl groups between two gene products. Ordinarily, only
(shown here in red) can block access to key one copy of the chromosome pair is methylated
regions of DNA. Each methyl group consists at the location illustrated.

B. cancerous cell Both copies of a single chromosome pair are methylated.

chromosome Insulator protein can no
longer bind to the region
between the two genes.
helps turn on
the lgf2 gene.

Misplaced methyl groups block access
to the DNA between two genes.

H19 gene
lgf2 gene Interaction results in
lgf2 gene product

helps turn on
the lgf2 gene.

lgf2 gene

Interaction results in
lgf2 gene product H19 gene

Methyl groups block access to the
chromosome DNA between the two genes, pre-
venting the insulator protein and
enhancer from binding. spring 2007 paradigm [23]
Biology’s big tent
are mathematics and biology
separate universes? Oliver King, whose
doctoral thesis sorted out a problem in
a mind-bending 32 dimensions, says
Meet some of the young researchers pouring into that’s one way to visualize his transi-
tion from number theory to protein-
life sciences from other fields By Carol Cruzan Morton folding biology.

Christopher Love shawdee eshghi

Age: 30 his best chance to bridge the Age: 30 Engineers and biologists
gap. From the perspective of think differently. “In engineer-
Bachelor’s degree: Chemistry, Bachelor’s degree: Chemical
University of Virginia a surface chemist, immunol- engineering, MIT ing, you start with physical
ogy seemed the easiest entry laws you know are irrefutable,
When Christopher Love started point. “I knew there was a lot In a recent meeting of the and if the data don’t support
his postdoctoral fellowship of contact between cells and Harvey Lodish lab, discussion them, you know that the data
in the immunology research that it involved surface interac- turned to the evolution of red are wrong,” Eshghi says. “In
group of Whitehead Member tions—something familiar,” blood cells: Why does a red biology, you don’t have that
Hidde Ploegh, he had not Love says. “It took me a year blood cell lack a nucleus? starting point. It’s very empiri-
taken a biology class since to understand the terminology. “The first thing I thought cal. Classical biology papers
high school. It’s a different language from of was the mechanical proper- use a lot of inductive reason-
Love knew how to make even other areas of biology.” ties of the cell,” says Shawdee ing: ‘This is our hypothesis.
magnetic nanoparticles orga- He needed even more Eshghi, who is just finishing Here are some data to support
nize themselves into micro- time to understand how biolo- her doctoral work in biological it. Maybe this is what’s going
scale structures and how to gists think. “Biology has an engineering under the joint on. We did another experiment
kim furnald for furnald/gray

create nanometer-thin crystal- extra level of complexity from oversight of Lodish and Linda to show this isn’t it.’ They
line coatings of molecules on materials science, physics and Griffith in the MIT biological present all the hypotheses and
metals. He wanted to apply chemistry,” Love comments. engineering division. “The knock them down.”
such tools from the physical “I’m amazed at the types of nucleus is stiff and cannot She adds that engineers
sciences to advance medical insights biologists can draw bend. The hallmark of the have a versatile common
knowledge and public health. from experiments, where it red blood cell is its flexibility. language: mathematics. But
Jumping feet-first into a tends to be difficult to control That’s not what comes to the biological systems need more
biology lab, he figured, was the variables.” minds of most biologists.” levels of explanation.

[24] paradigm spring 2007
“The math department was sort Biology at Whitehead and else- in string theory, a cornerstone of mod-
of austere,” says King, a postdoctoral where is increasingly infiltrated by ern physics.
researcher in the lab of Whitehead computational scientists, engineers, Meet the future of biology, rep-
Member Susan Lindquist. “Here we mathematicians and others who didn’t resented by King and the scientists
get to play with robots. We get to read train in biology. Whitehead Fellow below, all part of the next wave of
about sea slugs and cannibals. Every Paul Wiggins, for example, switched to researchers drawn to Whitehead by the
gene has its own story.” biology after starting graduate studies challenge of today’s life sciences.

danielle cook france kyle farh

Age: 28 properties of the stretchy stalk Age: 28 inquiries. Recently, Farh and
that affixes the tiny pond crit- collaborator Andrew Grimson
Bachelor’s degree: Biomedical Bachelor’s degree: Computer
engineering, Washington Univ. ter to a rock or crustacean. science, Rice University found that mammalian genes
When she publishes papers, have evolved to avoid target-
A biologist can be hard to find she considers which commu- Kyle Farh’s computer science ing by microRNAs that would
in the Whitehead lab of Paul nity she wants to reach, either training brings much-needed otherwise reduce or compro-
Matsudaira. And the lab’s cell biologists for the subject expertise to the Whitehead lab mise the genes’ function.
expertise ranges from simulat- matter or biophysicists for the of David Bartel. But it did not For all the differences, Farh
ing colliding stars on super- underlying imaging. “People help during his first two years has found a lot of common
computers to building joints either peg you as a biologist or at Harvard Medical School. ground between computa-
for robotic arms. an engineer,” she says. “I probably went to medi- tional and experimental biol-
“I wish we could get a France, whose mom is a cal school knowing the least ogy. “The thing that biologists
biology graduate student to math teacher, set her sights on molecular biology of all my are really good at, compared
work on Vorticella,” a genus of engineering earlier than most classmates,” remarks Farh, to other scientists, is doing
protozoa, muses Danielle Cook girls. “I see a lot of future in who initially joined a controls, because there is so
France. “There are tons of using the biology we know to startup company after college. much you don’t know about
open questions.” In the mean- engineer new things, such as At Whitehead, Farh’s the system, which is so com-
time, lab technicians provide building materials from basic bench work remains limited plex,” Farh comments. “You
the biological expertise and biological components,” she to occasional and relatively really have to be as rigorous
tutoring in protein purification. notes. “MIT has given me more simple procedures. He has about controls in computa-
France, a biological engi- confidence about starting my surrounded himself with tion.” In the end, he says, test
neering graduate student, own company. That spirit is in experimentalists who inform results can be equally enlight-
studies the rubber-band-like the air.” and inspire his computational ening or enigmatic. spring 2007 paradigm [25]
“I would like to see funds focused on the
diseases that people are actually dying from,
rather than ones that someone might imagine
could be a problem in the event of warfare,”
says Harvard epidemiologist Megan Murray.

out. Some of her postdocs “are spend-
ing more time writing grants than writ-
ing papers,” she adds.
At the National Institute of Allergy
and Infectious Diseases, where the
interim budget for 2007 was $3.8 bil-
lion, the proportion of grant propos-
als that will be funded has fallen over
several years to just 10 percent. “We
are being crunched,” acknowledges
Anthony Fauci, longtime NIAID direc-
tor. “We’ve had flat funding for the last
three years, and with inflation, we’ve
had a 10 percent decrease in purchas-
ing power. It’s a bad signal to send for
the young people.”

A bioterror money pit?
Many areas of the world are awash
in infectious diseases. Some are long-
standing and endemic, such as malaria.
Some are of recent origin, such as
HIV/AIDS. Others represent new or
re-emerging threats such as Ebola or
drug-resistant TB, a scourge that’s

Targeting the
spreading through South Africa.
Scientists are working on many
fronts to unravel the basic biology of
dangerous microbes.

agents of disease
But funding for these research

efforts is a major bottleneck.
Fauci points out that existing
resources are being directed at emerg-
In the war on infectious disease, are we ing and re-emerging infections in the
context of global health and security.
spending enough—in the right places? He believes that the intense concern
about a possible bird-flu pandemic is
By Richard Saltus an opportunity to reduce the toll of
ordinary seasonal influenza with better
Megan Murray, an epidemiologist at the Harvard School of Pub- vaccines and therapies. Spending on
influenza has been ratcheted up 10-
lic Health, is very worried about where her next research dollars fold to $222 million, says Fauci. Other
will come from. At times, she thinks she may have to fall back on NIAID priorities include developing an
effective HIV/AIDS vaccine, prevent-
being a practicing physician to make ends meet. ing mother-to-child HIV infections and
The co-leader of a large study in Peru aimed at identifying attacking the worrisome emergence of
drug-resistant TB.
risk factors for drug-resistant tuberculosis, Murray has published Though controversial, the infusion
important papers in the field. But she’s finding that it has become of $1.2 billion in new federal funding
kent dayton

for research on potential bioterrorism
highly challenging to get money from the National Institutes of agents has been a “boon” to NIAID-
Health, especially for new projects and for young scientists starting supported investigations generally, says

[26] paradigm spring 2007
Fauci. While the money is earmarked Private progress outstanding basic science,
for research on a list of “select agents” Unexpectedly and dra- but it’s less well-suited to
including anthrax, plague, tularemia matically, the funding programmatic approaches
and smallpox, it will yield insights and arena has been trans- to solving problems.”
diagnostic tools for fighting civilian formed in the past several Enormous gifts from
infections as well. years as wealthy philan- donors such as Gates and
But Hidde Ploegh, Whitehead thropists have taken up his buddy Warren Buffett
Member and immune-system expert, the cause of global public make a big splash and
expresses some leeriness about this health. can make a difference. By
funding direction. “To my knowledge, Most notably, the contrast, the NIH—prin-
no realistic threat assessment exists for Gates Foundation pro- cipally NIAID—has a
most of the agents on this select list, vided a huge shot in record of major sustained
nor has there been much of a debate the arm by committing funding that adds up
over it,” he says. “But because there is $450 million in 2003 to “We are being crunched,” over the years. “Over
more money in biodefense-related proj- launch the Grand Chal- acknowledges Anthony the last 25 years,” Fauci
Fauci, longtime director
ects, you will see investigators move lenges in Global Health of the National Institute points out, “we’ve spent
toward these agents.” Initiative. A high priority of Allergy and Infectious over $30 billion on HIV/
“There’s a contrast between on the Gates agenda is Diseases. AIDS.” And in that same
what infectious disease people worry malaria, which causes an period, he adds, NIAID
about—emerging communicable dis- estimated 1.5 to 2.7 million deaths and has grown, budget-wise, from the
eases—and what the bioterrorism infects 300 to 500 million people each eighth- to the second-largest Institute
people worry about, which is someone year, mostly in sub-Saharan Africa. within the NIH.
appropriating or creating a weapon Patrick Duffy, a prominent malaria
out of biological material,” remarks expert formerly at Walter Reed Army Tightened belts
Nobel laureate David Baltimore, who Institute of Research, was lured to the Still, after Congress doubled the NIH
recently retired as president of the Seattle Biomedical Research Institute, budget to $26.7 billion between 1999
California Institute of Technology and which receives significant Gates fund- and 2003, the increase in scientists
continues his research on HIV/AIDs. ing. Duffy, whose research focuses on applying for grants combined with the
discovering and evaluating antigens abrupt end of the budget largesse has
“Our representatives in government have a for a malaria vaccine, says that the spread resources thin.
grave responsibility to choose whether they Gates model “is more like corporate Funders also need to balance basic
advocate a war on terror or a war on microbes
or a war on cancer,” says Whitehead Member funding, in the sense that there are tar- and applied research. In Nature Immu-
Hidde Ploegh. “And you can’t spend the same geted goals and milestones to monitor nology in 2005, Fauci wrote, “An
dollar twice.” progress. The NIH model has yielded important challenge for the NIAID
is to find a way to preserve a robust
commitment to the fundamental,
investigator-initiated research that
is the bedrock of the research enter-
prise while meeting expectations for
more applied research, including the
advanced development of vaccines,
therapeutics, and diagnostics.”
sam ogden (ploegh); shaun heasley/reuters/corbis (fauci)

Time will tell whether the federal
government will eke out enough
funding, in the right places, to encour-
age the next generation of young
“I personally would largely allocate
money on the basis of global disease
burden, bearing in mind that epidemic
disease can happen without warning,”
says Harvard’s Murray. “There is still
a need to study potentially epidemic
agents. But with malaria, TB and HIV
as the main contributors to infectious-
disease deaths, those would seem the
obvious priorities.” spring 2007 paradigm [27]
fast FAQs and the egg would be used to gener-
ate patient-specific ES cells.] All the
genetic mutations causing the dis-
ease would be present in the ES cell.
If we could derive ES cells from
a Parkinson’s patient, we would like
to coax these cells into forming neu-
rons, and as we study the process,
hopefully find the defects that cause
the disease. In other words, we can
study a very complex disease in the
Petri dish with the potential to look
for compounds that treat it.

But does SCNT work in humans?
It works in mice, cows, sheep and
rabbits. It will work in humans. We
need to resolve the technical issues.

Why do we need to keep deriving
additional lines of human ES cell?
The presidentially approved ES lines
are useful to an extent. But they
Promises and realities in were created under conditions that
we no longer use.
embryonic stem cell research The Harvard ES lines [not
presidentially approved] are also
For all the controversies, it’s still early days for the science very useful. But what’s become
clear in the last few months is that
Whitehead Member Rudolf Jaenisch weighs in on the way you isolate or culture the
cells determines how the cells react
what we can, can’t and (hopefully) will one day do with
later. For example, all the human
human embryonic stem cells. lines that we work with have been
isolated under high oxygen con-
What do we know for sure that We need to learn how to make centrations. But if you grow these
embryonic stem cells can do? the human ES cells as easy to work cells under lower oxygen, they do
Embryonic stem [ES] cells have in with as mouse ES cells. The issues much better. We need to grow these
principle an enormous potential are strictly technical. cells under different conditions and
for research and therapy. We can decide what the best conditions are.
extrapolate much from our knowl- How successful have we been in turn-
edge of mouse ES cells. From these ing ES cells into specific tissues? What about alternative means for
cells, we know we can generate any There are major efforts to derive deriving ES cells?
tissue type in the Petri dish. We also neurons, heart muscle and blood These other methods are driven by
know they’re useful for therapy. cells. There have been major suc- ethical objections. We did a proof-
We’ve used them to treat a mouse cesses, but we’re not yet able to of-principle experiment in mice to
variant of the human disease severe produce a tissue for patients. show that one can generate embry-
combined immunodeficiency. We onic stem cells without destroying a
showed that one can restore the What do ES cells offer for basic viable embryo. Generating ES cells
immune system using customized research? from amniotic fluid also has got-
embryonic stem cells. These cells have enormous value as ten a lot of attention recently, but I
But the human system is much research tools. The hope would be think that approach is overblown.
more complex. These cells don’t to use somatic cell nuclear transfer The real goal of the field is to
grow rapidly, they’re difficult to [SCNT] to generate human models generate ES cells without using
grow as single cells, and they suffer of complex diseases like Parkin- eggs. Take a skin cell and treat it in
james yang

chromosomal aberrations quite eas- son’s, Alzheimer’s and diabetes. some way that redirects it back to
ily. All these issues we can handle in [In SCNT, an egg’s DNA would be an ES-cell-like cell. Nuclear transfer
the mouse. replaced with DNA from a patient, shows that the egg can do it.

[28] paradigm spring 2007
on the Web
paradigm Our silver anniversary
Editor Whitehead celebrates its 25th birthday with a special section of
Eric Bender our website that features interviews, videos and slide shows
Scientists gone wild
Watch former Associate Director John Pratt recount
associate Editors a near-death experience in the Himalayas with
David Cameron Whitehead Members Paul Matsudaira and Rudolf Jaenisch. (The hikers narrowly avoided avalanches
617.324.0460 during their long trek.) This is just one of the videos
that highlight the Institute’s tight-knit community.
Alyssa Kneller Visit
Tapping your memory
An interactive memory board lets scientists and
friends of Whitehead contribute photos and stories
Eric Mongeon
about the Institute. Whitehead Members and former
Mongeon: Projects
postdocs divulge some of their proudest moments in
the labs and a few of their colleagues’ most striking
character traits. To browse, visit
Paradigm is published
twice a year by the Office
of Communications and
Public Affairs at Whitehead
Institute for Biomedical Who was Jack Whitehead?
Research. The magazine Meet the extraordinary character who made White-
reports on life sciences head Institute possible. After making millions
research and innovations on medical devices, Edwin C. (Jack) Whitehead
at Whitehead, and explores dreamed of creating the ideal research environment
public issues related to for smart young scientists. At the end of a decade-
the conduct of biological long quest, he inked the deal with Massachusetts
research. To subscribe, Institute of Technology that launched Whitehead
send your address to Institute. “People developed a real respect for Jack,” recalls David Baltimore, Founding Director.
Text, photographs and “There was sort of a love affair between him and
artwork may not be reused the Institute, which is very rare with founders.” For
without written permission more about Jack Whitehead, visit
from the editor. edu/about/25th/jack.

Office of Communications
and Public Affairs

In this seven-minute video
Whitehead Institute for overview, Whitehead scien-
Biomedical Research tists describe a research envi-
Five Cambridge Center ronment that inspires
Cambridge, MA creativity and collaboration.
02142-1479 Visit www.whitehead.
617.258.5183 spring 2007 paradigm [29]
Whitehead Fellow Hui Ge studies how
networks of proteins interact in the
embryonic C. elegans worm. Here is a
network backbone with proteins grouped
into predicted “molecular machines.”
This is from a 2005 Nature paper co-
authored by Ge, who was then in the lab
of Harvard’s Marc Vidal. For more infor-
mation, see page 10. Courtesy Nature.

Whitehead Institute US Postage
for Biomedical Research
Cambridge, MA
Nine Cambridge Center
Permit No. 56998
Cambridge, Massachusetts 02142-1479