Background: Cancer is a great medical concern and can be very difficult to treat due to its wide variety of forms and origins in different people. Unlike some common illnesses, where we know the virus or bacteria that causes it, know the symptoms that most people exhibit when infected, and know what medications will help alleviate symptoms of rid the body of the infection, cancer is very different. It originates from a person’s own cells (from almost any tissue type), can be caused by a variety of different combinations of mutations in the genome (changes in the DNA sequence of the cell), many of which we are still discovering, and the symptoms can vary drastically. Because of these features, each tumor has to be treated almost as if it were its own unique illness, and thus it is extremely hard to try and find treatments that work across different cancers without really hammering the whole body. Advances in cancer care have done a lot to improve prevention and treatment of cancer, but there is still much to be learned about how tumors from different tissues or with different sets of mutations will act and the best ways to treat them.
Today’s article is an open access paper (meaning that the full article is freely available to everyone) that is part of a special issue of Nature focusing on the human epigenome (characteristics of the genome that affect which genes are on or off beyond the sequence itself). Polak et al. examine the epigenome of different cancer cells to uncover more about how they work and where they originated. Knowing this could help to better treat each patient and destroy tumors without damaging the rest of the body’s healthy tissues.
Problem/Question: Cancerous tumors are unique in different patients. How do we figure out the best ways to treat each one?
Results: The cells that make up different tissues in our bodies (muscle, liver, skin, etc.) have different arrangements of their genomes. While the underlying genome sequence is the same, different regions are open or compact, and different genes are turned on or off. This is important for proper cell function and discerning what each cell looks and acts like. The authors of today’s paper also suggest that this knowledge can be used to better understand cancer cells.
Because cancer is caused by mutations (changes) in the DNA sequence, the locations of these changes are likely going to be affected by the organization of the genome in a cell. To verify that this is the case and determine what else this line of thinking can tell us about cancer cells, the authors examined 173 cancer cell genomes from eight different cancer types, spanning a variety of tissues and mutational differences. They compared the locations of mutations in each cancer genome to epigenome data from many different healthy human tissues. In melanoma cells, and many other types that they tested, mutations were much more likely to occur in regions that were more compact and contained genes that are turned off than open regions with active genes. However, in liver cancer cells, mutations were more common in regions enriched for a specific histone mark (H3K4me1) that denotes genes that are on. Histones are complexes of proteins that have DNA wrapped around them. Many histone proteins have “tails” that can be modified in different ways and these modifications are like flags that direct the actions of other proteins in the cell. These seemingly conflicting findings suggest that mutations can occur in different regions of the genome dependent upon the tissue and not all tissues have the same mutation distribution or the same guidelines for mutation distributions.
One important thing to remember as we go through this paper is that while the authors are finding correlations that are statistically significant, correlation does not equal causation. Meaning that just because mutations occur preferentially in specific regions of one tissue type and not others does not mean that that in itself is causing the specific type of cancer from that tissue. It also does not necessarily mean the characteristics like specific histone marks are causing different regions to be mutated more frequently than others, simply that we can observe a pattern.
When compared overall, the authors found that epigenome features and cell replication timing (time between cell divisions) can together account for 74-86% of the differences in mutation locations between seven of their eight cancer types examined. This is much higher than previous studies and the high percentage of coverage means that this information could be used to predict the tissue of origin for a cancer genome.
One difficulty when extrapolating these findings to individual cancers is that there are less mutations from a single source than from a pool of cancer genomes from multiple tumors. The authors tested a variety of designed and real-life examples and found that their chromatin feature comparison still performed better than any previous analyses. They also tested to see if adding data such as gene expression and nucleotide content would improve their accuracy, but it did not. This indicates that these chromatin features (such as histone marks) are strongly correlated with differences in mutation distribution between tissues.
Lastly, the authors compared the locations of mutations to chromatin features of cancer cell lines (cells from a tumor that are grown and maintained in a lab). They found that the locations of mutations more strongly correlates with chromatin features from the tissue of origin than a cell line of the same type of cancer. There are a number of possible reasons for this, two of the main ones the authors addressed are that many mutations that are found in cancer cells may occur before there are changes to epigenetic features associated with the cell being cancerous. Second, the cell lines used are considered an advanced stage of their respective cancers, and there may be differences between epigenetic features of different stages of cancer.
Big Picture: Tumors are highly variable, making cancer very difficult to treat on a broad scale and necessitating personalized characterization and treatment. The authors of this study showed that epigenetic features strongly correlate with the locations of mutations in the genome of a cancerous cell. This can enable scientists and doctors to predict the tissue of origin for cancer cells and may help detect the location of tumors and narrow down the best treatment options. Personalized medicine is becoming more of a reality as whole genome sequencing comes down in price and can enable more optimized treatment of not only cancer, but other diseases as well. This paper is one more step in the direction of better understanding cancer and finding ways to detect and treat this difficult to manage occurrence.
The Ribosome 