Scientists who study how human chemistry can
permanently turn off genes have typically
focused on small islands of DNA believed to contain most of
the chemical alterations involved in those
switches. But after an epic tour of so-called DNA
methylation sites across the human genome in
normal and cancer cells, Johns Hopkins scientists have
found that the vast majority of the sites aren't
grouped in those islands at all but on nearby regions that
they've named "shores."
"Our study suggests that the real jackpot for
methylation isn't where we have all been looking
but in these shores located just nearby," said Andrew
Feinberg, professor and King Fahd Chair in
Molecular Medicine at the Johns Hopkins University School
The discovery is more than academic, Feinberg said,
since it may shift the current focus away
from the islands to thousands of new sites scattered
throughout the genome, each with the potential
to serve as novel targets for studying the development of
tissues, organs and animals, and for treating
diseases such as cancer that are already known to involve
Methylation is one of several so-called epigenetic
modifications that affect which genes are
expressed without changing the DNA sequence itself.
Previous studies have suggested that DNA
methylation plays an important role in guiding stem cells
to mature into a variety of cell types such as
hair, muscle and nerve cells. Methylation has also been
implicated in the abnormal gene expression
that cancer cells show.
The longtime focus in methylation has been CpG
islands, regions of the genome rich in the DNA
building block molecules cytosine and guanine. The reason
is that these islands tend to occur near the
"start" signal of a protein-coding gene, a place with the
potential to affect whether or not that gene is
expressed and to what extent.
However, Feinberg and co-investigator Rafael I.
Irizarri, of the Johns Hopkins University
Bloomberg School of Public Health, and their colleagues
wondered whether undiscovered methylated
sites were hiding unnoticed elsewhere in the genome. The
researchers performed their comprehensive
survey in human brain, liver and spleen tissues obtained
from five autopsies, identifying 16,379
methylated regions using a new method that searches all
DNA, not just CpG islands.
To their surprise, the researchers discovered that
about 76 percent of the genome's
methylated sites occur a short distance (between 200 and
2,000 kilobases) away from the islands. In
contrast, only 6 percent of methylated sites were situated
inside CpG islands. Because of the newly
discovered sites' proximity to the islands, the researchers
named them CpG shores.
"This finding is so unexpected because CpG islands are
the areas where scientists have really
concentrated their research," Feinberg said.
Taking their research a step further, Feinberg,
Irizarri and their colleagues searched for new
methylated areas in colon tumors as well as normal colon
tissue removed from 13 different-cancer
patients. The researchers identified 2,707 regions in both
CpG islands and shores that were either
more or less methylated between the two tissue types.
Previous studies by other research groups,
looking only at CpG islands, have suggested that the DNA in
cancer cells tends to be more methylated
compared to normal, healthy cells. However, comparing the
tumor and healthy colon cells' entire
genomes, including CpG islands as well as shores,
Feinberg's group found that cancer cells had a
roughly equal number of more- and less-methylated sites
than the normal cells.
When the researchers looked at where these differently
methylated sites were located in the
cancer cells, they found that these regions matched up with
many of the methylated areas they'd
located in the normal brain, liver and spleen tissues they
"This suggests that the epigenetic changes we see in
cancer are leading cells to look and behave
like other cells they're not at all supposed to be like,"
Feinberg said. "For example, colon cells that
become cancerous might start to act like brain, liver or
spleen cells. They're losing the characteristics
they should have and taking on those of other tissues."
He and his colleagues suggest in the study, published
online Jan. 18 in Nature Genetics, that
their newly developed map of where these methylation sites
differ in colon cancer and healthy tissue
may eventually help researchers develop treatments for
cancer by restoring normal methylation
patterns or by targeting the gene pathways they have
identified in this study.
Others who participated in this research are Christine
Ladd-Acosta, Bo Wen, Carolina Montano,
Patrick Onyango, Hengmi Cui, Kevin Gabo, Michael Rongione,
Hong Ji, James Potash and Sarven
Sabunciyan, all of the Johns Hopkins University School of
Medicine; Zhijin Wu, of Brown University;
and Maree Webster, of the Uniformed Services University of