Epigenetic patterns found in tumors
Scientists are quickly learning that the hardwired code within the arrangement of DNA bases does not fully determine the fate of an individual organism. This current understanding is built on the idea of epigenetics — that DNA expression can vary between individuals and is not caused by specific changes to the DNA sequence itself.
Héctor Corrada Bravo, an assistant professor at the University of Maryland Institute for Advanced Computer Studies and Center for Bioinformatics and Computational Biology, gave a seminar at Carnegie Mellon last Friday, sharing his team’s research exploring the epigenome — a record of the chemical changes to DNA and histone proteins — of tumor cells for detectable patterns.
Every cell in an organism is a descendant of the same original stem cell and, thus, all contain exactly the same code of DNA. The cells use the DNA to construct the necessary proteins for function. However, in each type of cell, a unique group of all the possible proteins will be made in that cell that determines the cell’s specific function. Bravo explained that his team’s goal is to look at the epigenome of tumor cells and identify patterns or specific regions of the DNA that are uniquely active in the cancer cells.
The challenge of studying the epigenome is that there are no specific changes in the DNA sequence that identify that a gene is activated at any given time. Bravo’s team became the first group to utilize whole genome bi-sulfite sequencing to help identify these activated genes in the genome of cancer cells.
The technique is based on the idea of DNA methylation, the binding of methyl groups to cytosine bases that commonly follow guanine bases within the DNA code. Simply put, high methylation has been hypothesized to contribute to a higher silencing of genes so fewer genes are expressed as proteins.
When the DNA is treated with bi-sulfite, cytosine bases that are not methylated are converted to uracil, another sugar base that is only found in RNA molecules. Cytosine that is bound to a methyl group remains cytosine. This DNA can then be sequenced and the cytosine bases in the DNA that remain are the ones that have been methylated.
For each cytosine site in the DNA, Bravo’s team calculated the probability that the site was methylated in cancer cells. This analysis helped his team observe patterns in the genome to identify more active genes in the cancer cells.
First, in their analysis with colon cancer, the team initially found that the genes that were less methylated were hyper-variably expressed, meaning their expression exceeded that of a normal cell.
Bravo claimed, “Instability at the DNA methylation level leads to instability at the gene expression level.”
The researchers further explored the areas of the genome that were hyper-variably expressed and saw a large number of potential genes that have been linked to cancer in other research. Cancer cells are cells that are experiencing an unregulated system of growth. There are numerous proteins that, if over-expressed in the cell, could cause the cell to grow at much higher rates, causing cancer.
After seeing this correlation with their data, Bravo and his team expanded their research and explored other cancers, such as lung, breast, and thyroid cancer. Their results showed that all the tumor types lose methylation in many cancer-specific genes.
Furthermore, Bravo explained their curiosity in whether methylation levels varied in different stages of tumors. Once again, their results proved that cancers which were further along and those capable of metastasis — the ability to spread in the body — had lower levels of methylation in specific regions.
Bravo explained that the repeated correlation that he and his team saw led them to believe that “there was something deeper at play.” They felt that these patterns should be explored further in depth.
Currently, the team is looking at cancer cell populations and comparing the epigenetic expression among the different cells in the population. They are actively working on a complex algorithm that generates a directed graph of all the possible methylated sites in the genome.
The algorithm Bravo and his team developed uses the graph to find the minimum cost path from the start to the end. In doing so, it identifies patterns in the genomic data, which can be extremely useful for more efficient cancer detection.
More importantly, it will lead to a deeper understanding of the biological roots behind the causes of cancer. Identifying these hyper-variable genes will contribute to better gene therapy and drug development.
Collaborators of Bravo’s work include Winston Timp, assistant professor of biomedical engineering at Johns Hopkins University (JHU); Andrew Feinberg, director of the Center for Epigenetics at JHU; and Rafael Irizarry of the Dana-Farber Cancer Institute at Harvard University.