(A-4) Transcriptomic analysis suggests biological processes are shared among highly migratory cancer cells but accomplished through unique transcriptional programs

Introduction:: Intratumor heterogeneity presents a significant challenge by compromising the accuracy of cancer progression prediction and affecting the response to therapy. Individual cell behaviors can be obscured by population-based averages, which clouds experimental results. Intratumor heterogeneity is caused by several factors, including genetic and epigenetic differences between cells in the primary tumor. As tumors progress, cancer cells accumulate mutations, resulting in variances within the population and forming distinct phenotypes, including migratory ability. An improved understanding of the impact of intratumor heterogeneity on cancer cell migration has the potential to facilitate the development of novel strategies to prevent or stop the spread of cancer, thereby improving patient outcomes. Additionally, studying the mechanisms of cancer cell migration can provide insights into the basic biology of cell migration, which has implications for other diseases and areas of biology. Several studies have used siRNA libraries or other genetic knock-out systems in high throughput cell migration assays to identify key genes involved in cell migration. This study utilizes a novel approach that may enable us to identify cancer-specific biomarkers of highly motile cells. Instead of knocking out numerous genes in one or two cell lines, we have sorted five cell lines into highly and weakly migratory subpopulations and subjected them to RNA sequencing to identify common specific molecular determinants of migration phenotype or other common biological processes.


Materials and Methods:: Five highly metastatic cancer cell lines (A375, MCF10a-CA1a, MDA-MB-231, SUM159, SW480) were sorted using consecutive transwell assays resulting in subpopulations of highly and weakly migratory cells (Fig. 1). RNA was isolated using the RNeasy mini kit with on-column DNA digestion. RNA samples were submitted to the VANTAGE core at Vanderbilt University for sequencing on the NOVAseq 6000. Raw sequencing files were aligned via STAR and differential expression analysis was conducted using DESeq2 in R. Gene Ontology (GO) term enrichment analysis was carried out via the over-representation analysis on WebGestalt.org using the ‘Biological Process noRedundant’ functional database, the ‘genome’ as the select reference set, and retaining all terms with an FDR lower than 0.05.


Results, Conclusions, and Discussions:: To determine molecular similarities between highly migratory subpopulations of cancer cells compared to their weakly migratory counterparts, 5 cancer cell lines (A375, CA1a, MDA-MB-231, SUM159, SW480) were sorted using a migration-based transwell assay to isolate a ‘highly migratory’ and a ‘weakly migratory’ fraction. Cells migrated through a collagen matrix and a membrane with 8µm sized pores in a serum gradient (Fig 1). After sorting, bulk RNA sequencing was performed for each cell subpopulation. Interestingly, we did not find many transcripts that were shared between highly migratory cells when compared to their weakly migratory counterparts. We identified only 3 genes (GGT5, EVA1A, and TM4SF18) that were significantly upregulated in all 5 highly migratory subpopulations and 0 genes that were significantly upregulated in all the weakly migratory subpopulations (Fig 2). However, each of the cell lines had many transcripts that were only significantly differentially regulated in that cell line (Fig 2), indicating that there are robust transcriptional differences between highly migratory and weakly migratory cells but that the particular changes are context-specific.

To investigate possible biological processes that are common to highly migratory cells compared to weakly migratory cells, we performed Gene Ontology (GO) term enrichment analysis which revealed 37 GO terms significantly enriched and shared amongst all 5 cell lines (Fig 3A). Further investigation of the significant GO terms revealed transcriptional signatures that are unique to each of the cell lines (Fig 3B,C). For example, terms such as ‘positive regulation of cell motility’ and ‘ERK1 and ERK2 cascade’ were significantly enriched in all 5 cell lines but importantly each cell line contained a partially unique set of genes associated with the particular term (Fig 3B,C). Thus our data suggests that highly migratory cancer cells may use unique transcriptional programs to achieve a highly motile phenotype via similar biological processes. All together our data indicates transcriptional networks underlying migratory phenotypes are highly heterogeneous and highlights the importance of studying numerous cell lines to reveal novel molecular mechanisms by which cancer cells migrate.


Acknowledgements (Optional): :

References (Optional): :