Role of Hypoxia in breast cancer - Alternative Splicing and Methylation

Missing out on a great opportunity can be suffocating. Suffocation or asphyxiation is the deficiency of oxygen supply in the body. Such deprivation of oxygen is known as hypoxia. Hence, it is probably not surprising that looking for opportunities in hypoxia research can result in such outcomes. The recent Nobel Prize in Physiology was awarded to Semenza and colleagues for their pioneering work on hypoxia. No doubt these scientists have avoided suffocation for so long by incorporating some of the defense mechanisms that the body has developed to counteract the lack of oxygen. The role of hypoxia in tumour microenvironment makes it even more pertinent to understand hypoxia and how it is regulated.

One has to realize that while the body as a whole can experience hypoxia, it is the individual cells that respond to this condition. While some cells might die due to the acute lack of oxygen, certain adjoining cells might manage to survive as the oxygen deficiency was not as pronounced. Understanding this cell to cell heterogeneity in dealing with the lack of oxygen would be next step in unraveling the tumor micro environment. However, this needs sophisticated equipment like the 10x Chromium System that can generate single cell RNA-seq libraries from a cell suspension. Only when you have the instrument you can generate pilot data required by the grant agency. Not surprisingly, you need the said grant money to buy the instrument in the first place. This is probably what smart people call a catch 22. 

We decided to overcome this by categorizing entire tumor samples as hypoxic or normoxic as single cell resolution continues to evade our simple minds. In order to classify the tumor into hypoxic or normoxic we needed a signature that could act as a set of features. The molecular signature database is a great source for obtaining such lists of genes involved in a specific molecular function. These signatures have been meticulously assembled by combing through numerous other published datasets. The hypoxia signature on MSigDb is actually based on approximately 80 other lists taken from various studies (including those published by the great Semenza). In order to evaluate the biological meaningfulness of this signature, we used the ShinyGO tool (as it is very Shiny) to visualize the molecular functions that are prominent in this signature set consisting of 200 genes. 

Hypoxia hallmark signature from MSIGDB enrichment for molecular function (200 genes)

While functions involving sugar metabolism are prominent and make sense, they might not be the ideal set of genes to find hypoxic (single cell like) tumor patients. Hence, we decided to make our own new signature. Details of how this was done is described in the manuscript of Pant et. al., 2020. the new custom signature identified by Pant and colleagues is a leaner list that might help achieve the goal of studying hypoxia heterogeneity by looking at inter-individual variation until intra-individual variation becomes accessible. 

Hypoxia custom signature identified by Pant et.al., 2020

Using this super fantastic new signature that is identified by cleverly combining ChIP-seq and micro-array data, we stratified the public TCGA breast cancer data into hypoxia and normoxic patients. By comparing these two groups of tumors, we hoped to understand the differences that might exists between hypoxic and normoxic regions within the tumor. Since, we are specifically interested in Alternative Splicing (due to its ability to increase complexity without any increase in gene number) and its role in hypoxia, we identified differential spliced isoforms between these two groups. Exonic regions that were isoform specific among these isoforms were identified. Only for these exons, the role of DNA methylation was assessed by looking at correlations between the expression level of these exons and the methylation level of proximal DNA measured using arrays. Code required to replicate the results is available on the github repo here: Hypoxia splicing methylation correlation.