As the cell biology lead at Perlstein Lab, I spend a lot of time thinking about ways to better understand the mechanism of action of the hits from our primary screens. There are many experiments one can do to try to narrow down what pathway their compound is affecting such as cell biology experiments in the presence of appropriate agonists or antagonists, metabolomics, immunohistochemistry and much much more. But one way to get a broad picture of how your compound of interest might be affecting cell biology is with RNAseq. RNAseq uses the power of next generation sequencing (NGS) to measure the transcriptome or RNA expression profile of the target material.
For prepping RNA for deep sequencing, one has to isolate the desired population of RNA and convert it to cDNA or complementary DNA. The samples are then fragmented for attachment of adaptor sequences (required for binding of the DNA to the chip). The cDNA can be read from one end or both ends (paired end read) and the resulting sequence information can be aligned to a known template or assembled for novel sequence information.
Alternative methods such as microarrays and cDNA sequencing have various limitations with regard to background noise, throughput, expression quantification, and amounts of RNA required. RNAseq resolves those issues by allowing for high-throughput, high-resolution expression information on low quantities of input information.
Let’s go through an example of how a group used RNAseq to characterize mechanisms of action of their compound of interest. Wacker et al used unbiased transcriptome analysis to identify the target of BI 2536, an anti-cancer drug. To do this they used a human colon cancer line, HCT-116, which is deficient in the multidrug resistance pump and mismatch repair so it effectively behaves like a mutagenized line. They then treated these cells with BI 2536 and isolated resistant clones. Those clones were sent for deep sequencing to identify the mutations that conferred resistance to the drug.
There were independent mutations in most of the clones however, PLK1, Polo-like kinase 1, was mutated in multiple isolated clones. To determine if these gene mutations confer resistance, the researchers expressed the mutant PLK1 genes in hTERT-RPE1 and HeLa cells and determined that cells expressing the PLK1 mutant proteins were less susceptible to BI 2536 induced toxicity.
Of course there were other genes in the clones identified as potential targets for resistance to BI 2536. For example, after going through the sequencing data it was noted that ABCB1 (P-gp, a drug efflux transporter) was the one of the most highly expressed mRNAs in the resistant clones.
They used the same approach to see if they could identify the target of Bortozemib. Researchers treated HCT-116 cells with various concentrations of Bortozemib and isolated nineteen resistant clones. A subset of these clones was deep sequenced, and the already known target of Bortozemib was mutated in multiple clones. As the authors point out, if the target was not known, the other five mutated genes would have to be characterized in independent studies.
Another example is using RNAseq to study the toxicity profile of compounds in rat liver, as described in Gong et al. There are many ways to use this technology to help uncover modes of action!