Agents believed to selectively "restart" genes that
limit cancer's growth — a potential treatment option
already in early clinical studies — instead turn off
as many genes as they turn on, a team of researchers from
the National Cancer Institute and Johns Hopkins has
"We don't know what effect all these changes might
have, but it's clear that when scientists are looking only
at the agents' effects on a particular gene or a few
particular genes, they aren't seeing the whole picture,"
says Andrew Feinberg, King Fahd Professor of Medicine at
Johns Hopkins. Their report appears in the October issue of
The research team probed the global effects of each of
three approaches to unhooking methyl groups from genes'
DNA. Cells normally use methyl groups to "mark" certain
genes, indicating whether their instructions should or
shouldn't be used for making proteins, but the marks are
frequently disrupted in cancer cells.
For example, in cancer cells genes that normally
stifle cell growth — so-called tumor suppressor genes
— are shut down because extra methyl groups are
hanging on to them. If these extra methyl groups could be
removed, the thinking has gone, the gene could be restarted
and the cancer slowed or stopped.
But the new work shows that while the agents tested do
restart cancer-suppressing genes, they also knock methyl
groups off other genes. Moreover, some of the unexpectedly
affected genes are turned on, but an equal number —
hundreds — of other genes are turned off.
The findings don't mean automatic failure for clinical
trials of so-called demethylation agents, but they do
indicate that careful attention should be paid to results
of laboratory experiments and clinical trials that use the
agents, since so many genes are affected, Feinberg says.
"It was kind of assumed that removing methyl groups
would turn some genes on and others off, but the
deactivation side of the coin had been largely ignored as
being a minor effect," adds David Gius, chief of molecular
radiation oncology at the NCI. "Now we know for sure that
removing methyl groups has both consequences and to equal
In their experiments, the researchers examined the
expression of nearly 8,000 genes simultaneously in a colon
cancer cell line (called HCT116). By studying the genetic
"fingerprint" of a sample before and after demethylation,
they could measure how the treatments affected the extent
to which the genes' instructions were being used to make
One chemical agent they tested,
5-aza-2'-deoxycytidine, blocks addition of methyl groups to
DNA and is currently in early clinical trials for leukemia.
The researchers also tested the effects of knocking out of
the cancer cell line either of two genes that encode
proteins (DNA methyltransferases) that hook methyl groups
onto the DNA, as well as knocking out both genes in the
same cell. They compared these methyl-based mechanisms of
gene regulation to a chromosome-based one, also in early
clinical trials, using a chemical called trichostatin A, or
TSA, that gently unravels the chromosomes, exposing genes
and allowing their instructions to be read.
Much to the team's surprise, both chemical agents
— one methyl-based and one chromosome-based —
created similar patterns of changes in gene expression in
the cell lines, says Hengmi Cui, assistant professor of
medicine. However, the genetic knockouts' patterns of
genetic changes were not similar to those of chemical
While the number of affected genes differed, all the
methods turned off as many genes as they turned on,
including entire gene families that might play a role in
cancer's development or contribute to its aggressive
The researchers are now investigating further some of
the individual genes and gene families affected by the
various treatments, and are studying the mechanisms by
which the chemical agents and the gene knockouts affect
methylation and gene expression.
For example, the demethylation chemical was thought to
act indirectly — preventing re-methylation, so to
speak — as proteins were turned over and recycled in
the cell. Because the effects of this agent were similar at
both one day and five days after treatment, the researchers
suggest the agent might have a more direct effect than
Methylation of genes is an example of epigenetics,
which are inheritable modifications to chromosomes and
genes other than changes in the DNA sequence itself.
The research was funded by the National Cancer
Institute. Authors on the study are Cui, Feinberg, Sheri
Brandenburg and Yali Hu, of Johns Hopkins; Gius, Matthew
Bradbury, John Cook, DeeDee Smart, Shuping Zhao, Kheem
Bisht, Allen Ho, David Mattson, Lunching Sun, Eric Chuang
and James Mitchell, of the National Cancer Institute; and
Lynn Young and Peter Munson, of the Center for Information
Technology at the National Institutes of Health.