New technique widens the lens on cancer, could lead to new therapies

By Jennifer O'Brien

With the help of data from the Human Genome Project, UCSF scientists have developed a new technique for screening tumors that highlights the role of a non-genetic mechanism in cancer. The researchers say the approach, which they applied to the study of human brain tumors, “widens the lens” on the cancer genome, and could lead to new therapeutic targets.

The technique allows scientists for the first time to detect whether individual tumor suppressor genes, often inactivated in cancer, have been corrupted by damage to the genes themselves or by—or in conjunction with—another mechanism, called DNA methylation.

The implications of the finding are significant, says senior author Joe Costello, PhD, UCSF assistant professor of neurological surgery, and a member of the UCSF Comprehensive Cancer Center. “We show that DNA methylation is most often affecting a totally different set of genes. This means that our current understanding of the genes that contribute to cancer is far from complete, and we have a source of the therapeutic targets for fighting cancer that is not fully realized.  Until now, it’s as though we’ve been looking at the cancer genome with one eye closed. With this approach, we now have stereo-vision. We can see the cancer genome in depth.”

Traditionally, cancer researchers have focused on trying to identify tumor suppressor genes that have been damaged, either by deletion or mutation. They have known for 15 years that some tumor suppressor genes are methylated - an enzymatic process that can inactivate genes - but they have not been able to detect the process with classical genetic screenings because the genes are not damaged, only silenced, and therefore appear normal.

As a result, scientists have not been able to discern the relative role or extent of interaction of the two different mechanisms in undermining the two copies, or alleles, of tumor suppressor genes; consequently, genes undermined primarily by methylation have remained hidden.

In the current study of human brain tumors, however, researchers determined not only that DNA methylation had occurred in numerous tumor suppressor genes, but that methylation primarily affected different genes than those damaged by deletion or mutation.

“The study raises the distinct possibility,” says Costello, “that there are additional undetected genes that are disabled by DNA methylation. If this proves to be the case, he says, it’s possible that new targets for therapy and diagnosis will be discovered through this integrated analysis.” Costello is a member of the UCSF Brain Tumor Research Center.

The study is published on-line Sept. 30, and appears in the November issue of Nature Genetics.

In their study, the researchers analyzed DNA from 26 human brain tumors, as well as from 11 non-tumor brain samples from the UCSF Neurosurgery Tumor Bank. In most cases, DNA methylation in the tumors did not occur on genes that also had deletions or mutations. In many cases, methylation affected both copies of a gene, essential for actually “turning off” the gene’s activity. Notably, the all-methylated genes would have been invisible in traditional genetic screenings. In the rare cases where convergence of methylation with deletions did occur, the convergence identified inactivated genes that had been missed by traditional approaches.

Two observations were particularly notable. In one case, the scientists discovered that both copies of a gene known as CISH1 were aberrantly methylated in 18 of 26 tumors. In three other tumors, the gene was both methylated and deleted. The researchers say that these observations, coupled with a study by another lab showing that CISH1 is methylated in liver cancer, and cell culture studies indicating CISH1 is involved in growth control, implicate the gene as a candidate tumor suppressor gene.

In another case, the scientists discovered that a gene known as COE3 was aberrantly methylated and deleted in high grade tumors and methylated on both alleles in others, indicating that it is a target of inactivation in brain tumors. COE3 had not been detected in traditional genetic studies.

“Our integrated approach addresses a large gap in the understanding of tumor genomes,” says lead author, Giusepee Zardo, PhD, UCSF visiting postdoctoral fellow in the UCSF Department of Neurological Surgery. The other lead author is Maarit Tiirikainen, Ph.D., associate specialist in the UCSF Department of Neurological Surgery. Both are members of the UCSF Brain Tumor Research Center and the UCSF Comprehensive Cancer Center.

The next step in investigating the CISH1 and COE3 genes, says Tiirikainen, will be to determine if they are “functionally” important - ie, capable of causing or driving cancer development in cells in culture and in mice.

Cancer occurs when there is widespread disruption of the genes that normally regulate a cell’s growth. Disruption of tumor suppressor genes releases the normal brakes on cell growth.

Disruption of the genes that drive cell growth leads to abnormal acceleration of cell growth; these latter genes are known as “oncogenes.” Disturbance of enough of these two types of regulatory genes can push a cell into replication overdrive, the hallmark of cancer.

Treating tumors generally requires therapies that target several of the many cancer-causing genes within a tumor, so a drug aimed at one tumor suppressor gene would be unlikely to halt the disease. However, a drug aimed at aberrantly methylated tumor suppressor genes could help to undermine the process.

DNA methylation, while implicated in cancer, normally plays a critical role in the life of most cells, assisting in a process that maintains stability of chromosomes and regulates gene expression.

Methylation occurs when one of several enzymes latches onto a particular cytosine, one of the four building blocks of a gene. The event takes place all over the genome, including in regions of genes known as the “promoter”, which regulate gene activity, or “expression.”

Encompassing the promoter of many genes are regions rich in cytosine and an adjacent guanine, linked together by a phosphate molecule. In cancers, it is these regions, known as CpG islands, that some times are aberrantly methylated in tumor suppressor genes.

Scientists have long been able to detect aberrant DNA methylation within a cell’s genome using an enzyme known as Not1. The enzyme specifically targets CpG islands near the promoter, where abnormal methylation has the strongest effect on the gene. When the islands are not methylated, the enzyme cuts them. When the islands are methylated, the enzyme cannot cut them.  As a result, this process, known as Restricted Landmark Genomic Scanning (RLGS), illuminates methylated CpG islands against the backdrop of a molecular depiction of the genome, a kind of DNA fingerprint of methylation patterns. RLGS has enabled scientists to identify the methylation status of thousands of genes this way.

However, scientists have not been able to integrate methylation studies with genetic maps indicating the sites of deletion and mutation patterns in tumor suppressor genes, because the location of most RLGS fragments within given chromosomes, let alone specific genes, is unknown.

In the current study, the UCSF team developed a genome-wide approach for mapping genomic and methylation alterations and integrating them. In developing the approach, the team relied heavily on the DNA sequences from the Human Genome Project.  A computer program was written specifically to create an RLGS profile from the sequences, a kind of in silico RLGS profile.  Actual RLGS profiles of DNA were then matched to the in silico profiles.  This process identified the genes associated with each RLGS fragment and allowed the team to make methylation site maps of chromosomes.  Now, they are able to compare the RLGS-based methylation results directly to the high-resolution maps of deletions.  The approach, says Costello, should offer scientists a much more complete image of cancer’s genome landscape.
All of the authors of the Nature Genetics paper are members of the UCSF Comprehensive Cancer Center.

## Other co-authors include:

* Chibo Hong, PhD, visiting postdoctoral fellow in the department of neurological surgery and a member of the UCSF Brain Tumor Research Center;
* Anjan Misra, PhD, visiting postdoctoral fellow in the department of neurological surgery and a member of the UCSF Brain Tumor Research Center;
* Burt G. Feuerstein, PhD, UCSF professor of laboratory medicine and neurological surgery and a member of the UCSF Brain Tumor Research Center;
* Stanislav Volik, PhD, of the UCSF Cancer Research Institute;
* Colin C. Collins, PhD, UCSF assistant adjunct professor UCSF Cancer Research Institute;
* Kathleen R. Lamborn, PhD, UCSF professor of neurological surgery and a member of the UCSF Brain Tumor Research Center;
* Andrew Bollen, DVM, MD, UCSF professor of pathology; and
* Daniel Pinkel, PhD UCSF professor of laboratory medicine, and Donna G.. Albertson, PhD, UCSF associate professor in the UCSF Cancer Research Institute.

The study was funded by the James S. McDonnell Foundation and the National Institutes of Health.