The science of epigenetics blossomed in the mid-1990s and has remained in
full flower ever since. Epigenetic regulation is now known to play
fundamental roles in everything from cancer to cloning to
stem cells. Leading this epigenetic revolution is
Rudolf Jaenisch of the Massachusetts Institute of Technology. In the
last two years, Jaenisch has contributed to a baker's dozen Hot Papers
on reprogramming fibroblasts into a pluripotent state comparable to
embryonic stem cells—a distinction that earned him a spot in this
issue's annual roundup of
authors. And his prior citation record is no less impressive: a 1992
Cell paper on mutation of the methyltransferase gene, for
example, has amassed over 1,600 citations, while his Nature
Genetics review on the epigenetic regulation of gene expression has
been cited more than 700 times in just six years (see the second below,
paper #1). More recently, Jaenisch’s 2007 Nature article
on "In vitro reprogramming of fibroblasts into a pluripotent
ES-cell-like state" has repeatedly appeared in the
Biology Top Ten and now ranks at #9.
Jaenisch, 66, received his M.D. from the University of Munich in 1967. Over
the next five years, he did postdoctoral research at the Max Planck
Institute for Biochemistry in Munich, at Princeton University, and at the
Fox Chase Cancer Research Center in Philadelphia. In 1972, he became an
assistant research professor at the Salk Institute in La Jolla, and then in
1977 moved back to Germany to become head of the department of tumor
virology at the Heinrich Pette Institute for Experimental Virology and
Immunology in Hamburg. Since 1984, Jaenisch has been a professor of biology
at MIT and a member of the Whitehead Institute for Biomedical Research.
You’ve been a driving force in modern
epigenetic research since its earliest days. Can you tell us what
prompted your move into the science of epigenetics?
It came from an odd observation in the 1970s, when I was interested in
making what would turn out to be the first transgenic mice, although
that term wasn’t even invented yet. The idea was to use
retroviruses to put some foreign genetic information into the germ line
of mice. A group from the Wistar Institute had reported that when you
do this, these viruses get activated. But when I infected embryos with
retroviruses, I couldn’t reproduce what they saw. In my lab,
viruses could infect the embryos and integrate into the genome but they
were silenced. Retroviruses—in this case, leukemia
viruses—have very strong promoters. They are strongly expressed
in somatic cells, in blood cells, in fibroblasts, but when they infect
embryos they are not expressed at all—they seemed to get
silenced. I was very puzzled by this. I also observed that if I put
these viruses into early embryonic cells such as ES cells, they were
not expressed. They did efficiently integrate into the genome, but
became silenced. If I infected cells from later stages in development,
the viruses expressed very highly. So what was the reason for this?
Cited Papers by Rudolf Jaenisch and
Colleagues, Published Since
by total citations)
R. Jaenisch, A. Bird,
"Epigenetic regulation of
gene expression: how the genome
integrates intrinsic and
Nature Genetics, 33:
L.A. Boyer, et al.,
regulatory circuitry in human
embryonic stem cells,"
Cell, 122(6): 947-56,
F. Gaudet, et al.,
"Induction of tumors in
mice by genomic
B.E. Bernstein, et
al., "A bivalent
chromatic structure marks key
developmental genes in embryonic
Cell, 125(2): 315-26, 21
T.I. Lee, et al.,
"Control of developmental
regulators by polycomb in human
embryonic stem cells,"
Cell, 125(2): 301-13, 21
I then realized that as soon as these viruses integrate into the genome
they become methylated in cytosine residues. This is what’s
called CpG methylation, a phenomenon that had just been discovered and
was very new at the time. Methylation seemed to correlate with gene
inactivity. If a gene was methylated, it was inactive. If it
wasn’t methylated, it was active. What we found was that as soon
as a virus integrates into the genome it becomes methylated, but this
happens only in the embryonic cell. In a somatic cell or a later cell,
the viral DNA stays unmethylated and so it's expressed. We were the
first to demonstrate this, and many people confirmed it.
Were you able to show that methylation was the
causal factor in silencing these genes? And what promoted the
For this we needed more genetic tools, so we made one of the first
knockouts, knocking out the gene for methyltransferase, which is the
main enzyme that accomplishes this methylation. Knocking out that gene
is lethal; the embryos die in mid-gestation. So that made it clear that
DNA methylation is absolutely essential for the development and health
of the organism, and that led me into this concept of epigenetic
control of gene expression. When we infected embryonic stem cells with
these viruses, and the methyltransferase gene had been knocked out, the
virus would be expressed; it would not be silenced. This told us that
there was a causal relationship between methylation and silencing. It
was the methylation that silenced the gene in the embryonic stem cells.
And this methyltransferase knockout led us to many, many more
interesting insights into epigenetics. It told us not only that
methylation is essential for life—but also that it has a causal
role in cancer development and in genomic imprinting. Many issues came
into focus when we used this particular mutant to understand epigenetic
What determines whether the viral DNA is
methylated or non-methylated at different stages of
You have to realize that the genome is full of viruses, and they are
generally silent, which means they’re methylated. And organisms
have a major interest in silencing viral genes that get into the
genome. Otherwise they’d screw up the genome. So this is an
evolutionarily conserved mechanism to silence these transposable
elements, these viruses. And you can argue that this is most important
in the early embryonic lineage, which gives rise to germ cells. We
later found out that these early embryonic cells express several
methyltransferases. The one we knocked out is the main one, but it only
serves to maintain methylation once methylation has already been
established by a different enzyme. It propagates the methylation signal
from one cell division to the next. It doesn’t establish new
methylation on its own.
The job of establishing methylation is done by enzymes called de
novo methyltransferases. So if you put a virus into an early cell
and it’s not methylated, then the maintenance methyltransferase
that we used would not do anything for this virus, but the de
novo methyltransferase will—or it will if the viral DNA is a
target for it. That’s how these two enzymes collaborate. As it
turns out, only embryonic cells express these de novo
methyltransferases. Later they’re not expressed. And this
explains why if a virus gets into a later cell, it’s not
methylated. If it gets into an earlier cell, it is. This resolved the
issue that you rightly raised.
Did you ever knock out de novo
That was done by a former student of mine, En Li. He actually did the
first methyltransferase knockout in my lab as a student. He started his
own lab and knocked out the de novo methyltransferases, and
that really showed us that they are very important in cancer
So what is the role of DNA methylation in
cancer development, and how do you use that knowledge, if you can,
to treat or prevent cancer?
We know now, for example, that cancers need to silence tumor suppressor
genes. This can be done either by a mutation, which disrupts the gene
and is irreversible, or by silencing the gene with methylation. In
colon cancers, for instance, certain tumor suppressor genes are always
silenced by methylation. We wondered what the mechanism for that is,
and we learned that in these colon cancers, the de novo
methyltransferases become inappropriately re-expressed. Then they
target certain genes—specific sequences—and they become
causally involved in cancer by silencing key tumor suppressor genes.
And that, of course, immediately provokes the question of whether or
not these insights are useful for therapy. I always thought it would
be. In looking at this maintenance methyltransferase in our original
experiments, we saw that when it was inhibited, either genetically by a
mutation, or by a drug, it really prevented cancer development in these
mice. The drug is actually a weird drug. It is very similar to
cytosine. So DNA incorporates this drug, and when it replicates it does
so with this altered cytosine, and now it cannot be methylated. It
binds the enzyme covalently and leads to demethylation of DNA. And when
we did this in mice, it protected the mice against cancer. This
approach is now being tested in some big clinical trials for certain
leukemias and head and neck cancers—at Johns Hopkins and at M.D.
What about side effects, since it seems
you’re dealing with some very fundamental mechanisms that
could have effects system-wide?
If you use it for a cancer patient, it seems to be okay. But if you
want to use it for children, for instance, who have these mutations
that predispose them to colon cancer, you can’t use any drug that
has these kinds of potential side effects. But I think this is where
the de novo methyltransferase will come in very handy. If you
inhibit this with a drug there are virtually no side effects, based on
what we understand from all this work. I think that will be very
interesting to do, and we’re hoping that some drug company finds
a drug that inhibits the de novo methyltransferase.
Why virtually no side effects?
Well, as I told you, the enzyme, de novo methyltransferase, is
never expressed at high level in somatic cells—it has no role in
somatic cells. It gets re-expressed in cancer, but that’s an
error. The maintenance methyltransferase is expressed in all cells that
replicate DNA. So when that’s inhibited, it’s lethal to the
cell, which dies. This enzyme is needed in somatic cells. De
novo methyltransferases are only needed in embryonic cells. When
you knock them out in later cells, there’s really no phenotype.
So if you inactivate them by a drug, we would predict that they
wouldn’t have any phenotype and so would have no
effect—except in the cancer cells.
A couple years ago you got into the stem cell
field in a big way, reprogramming fibroblasts into pluripotent
stem cells. How did that come about?
"We and two other groups
independently showed that we could make
mice out of these reprogrammed stem
cells, and so these cells were really
indistinguishable from embryonic stem
cells," says Rudolf Jaenisch. "This
really electrified the
When Dolly the sheep was born back in 1996, I immediately got
interested in the technique that was used, called nuclear
transplantation. That’s nothing more than an epigenetic
phenomenon. So I immediately went to Hawaii, where they cloned the
first mice, and began a collaboration learning how to do this. We tried
to understand the mechanism of the reprogramming that goes on in this
procedure. But nuclear transplantation is a very complicated procedure
and not many can do it well.
Why is this nothing other than an epigenetic
Well, in Dolly they took the nucleus from a mammary cell and put it
into an egg cell that had its nucleus removed. Normally, development
goes in one direction. It goes from pluripotent stem cells to more
restricted stem cells to differentiated cells. In nuclear
transplantation, the egg reverses that. It makes out of a
differentiated cell—a mammary cell, in Dolly’s
case—an embryonic cell. So somehow the clock is turned back, and
that means you have to turn the clock of epigenetic changes back to the
embryonic ground state. It’s all about epigenetics. Nothing else.
So how does the egg accomplish that?
What’s the precise mechanism?
I was thinking about that a lot for many years, while I made all the
tools to address this question. It’s a very interesting question.
Then Shinya Yamanaka’s paper came out in 2006 (K. Takahashi, S.
Yamanaka, Nature, 126(4): 663-76, 1006).That was a really
important paper, showing that you could reprogram adult fibroblast
cells into pluripotent stem cells by using pretty much exactly what we
were thinking about. He used the same genes, but he also added two
cancer genes, two oncogenes, to do this. That was really brilliant and
a very important finding. It made the process efficient enough that he
could detect it. You don’t need the two cancer genes but they
make it much more efficient. That was a very important insight. When
Yamanaka's paper came out nobody really believed it—it looked
just too fantastic. We had all the tools to do this, and so we used the
same genes he used, including the two oncogenes, and it worked very
robustly. We showed—along with Yamanaka and Konrad Hochedlinger,
a former student of mine now at Massachusetts General Hospital, all
working independently—that we could make mice out of these
reprogrammed stem cells, and so these cells were really
indistinguishable from embryonic stem cells. This really electrified
the field; it clearly showed that these cells were pluripotent, and now
three independent groups had shown it. Nobody could doubt anymore that
this was correct.
In the couple of years since these papers were
published, how has this reprogramming technique been
We’ve all done a lot. We can do this now in human cells.
We’ve also done it with different cell types.
And this avoids the ethical and moral issues
about the use of human embryos for stem cells?
Is it reasonable to assume, then, that the
ethical debates about the use of embryos are now over?
I think the sharp edges are over. I think nuclear transplantation using
human embryos won’t be done often. I don’t think it has any
place in therapy. But some people may still do it for research. So
embryonic stem cells are still needed. When you make iPS cells, as
these are called—"induced pluripotent stem cells"—you have
to compare them to embryonic stem cells. So you still need additional
embryonic stem cells for the comparison, to standardize the iPS cells.
How do you see this research progressing over
the next five years?
I think we’ll see rapid and enormous progress. It’s a very,
very active field. What we really have to do, though, and it’s
very expensive proposition, is get into human cells more and make
patient-specific cells, really analyze them, and try to develop
potential therapies for some of these major diseases. This research,
however, is very expensive, but if we could do it, progress could be
rapid. In vitro reprogramming allows us to take skin cells
from a patient with Parkinson’s disease, for instance, and
make in vitro pluripotent stem cells; these could then be
differentiated into the neurons which are the problems in these
patients. Then we can study the disease, a major human complex
disease, in a Petri dish. The potential is enormous.
KEYWORDS: RUDOLF JAENISCH, MIT,
WHITEHEAD INSTITUTE, EPIGENETICS, METHYLATION, METHYLTRANSFERASE,
NUCLEAR TRANSPLANTATION, REPROGRAMMING CELLS, INDUCED PLURIPOTENT
STEM CELLS, IPS.