Researchers from the University of Michigan have developed a strategy for identifying potential cancer drug targets from tumor RNA-seq data by honing in on outlier expression of kinases.
Now, the researchers plan to test their proof-of-concept strategy, published recently in Cancer Discovery, in xenograft models and in vitro tumor models. Once optimized, they plan to extend the work to patients who enroll in the university's cancer sequencing protocol, Mi-OncoSeq. The goal is to eventually be able to culture tumor tissue from patients in real time to identify outlier expression of kinases that would point to a relevant therapy.
"This represents another dimension in which you could explore a therapeutic target, and expands the window of available approaches to therapy," Chandan Kumar-Sinha, a research assistant professor of pathology at the University of Michigan, told Clinical Sequencing News.
Kinase mutations are frequently good targets for drugs, with many targeted therapies on the market and in development focused on kinases. For instance, the compounds imatinib, trastuzumab, and lapatinib are approved to target kinase mutations in chronic myeloid leukemia, breast cancer, and lung cancer, respectively. And, there are a number of other compounds in development that target kinases like PIK3CA, BRAF, and others.
The idea behind the Michigan study is that if mutations to all known 468 kinases, or the kinome, are assessed, outlier expression could potentially be an indicator of therapies that could be used either alone or in combination with other targeted therapies.
To demonstrate this concept, the researchers performed RNA-seq of 482 cancer and benign samples from 25 different tissue types, focusing their analysis on the 468 kinases.
First, the team tested the protocol on breast cancer samples, including 43 cell lines and 67 tissue samples. They then looked for the kinases that displayed the highest level of absolute expression, defined as greater than 20 reads per kilobase transcript per million total reads in the sequencing run, as well as the highest levels of differential expression compared to the median level of expression for a given gene.
In cell lines known to be positive for ERBB2 amplifications, all samples demonstrated outlier expression of the gene, as expected. However, many of the samples also had outlier expression in additional kinase genes, including CDK12, FGFR4, and RET.
To test the therapeutic implications of these findings, the researchers first tested the cell lines' susceptibility to trastuzumab, marketed as Herceptin by Genentech, which targets cells with ERBB2 amplifications. As expected, they found that the cell lines responded to the therapy.
Next, they treated the cells with both Herceptin and an FGFR inhibitor, and found that the cell lines harboring outlier expression of both ERBB2 and FGFR4 showed increased sensitivity to the combination of drugs, while those expressing ERBB2 but without outlier expression of FGFR4 did not.
Finally, the team created Herceptin-resistant cells from two ERBB2-positive lines. The cells that also had FGFR4 outlier expression remained sensitive to the FGFR inhibitor, while cells without it continued to proliferate.
Building on this initial approach, Kumar-Sinha then sought to look for outlier expression of kinases in pancreatic cancer, for which there are currently no available targeted therapies.
Using the same RNA-seq approach and analysis of the kinome for outlier expression in 22 pancreatic cancer cell lines and 13 pancreatic tissue samples, which included cancer samples, benign tissue, and xenograft models, revealed outlier kinases that were overexpressed in the pancreatic cancer samples, including EPHA2, MET, PLK2, MST1R, and AKT2.
"We found multiple kinases with known inhibitors showing outlier expression," Kumar-Sinha said.
AXL and EGFR showed outlier expression in both pancreatic and breast cancer samples, but those were the only two genes that overlapped.
The researchers then tested the impact of knocking down genes with outlier expression on cell proliferation, and in all cases the knockdown of a gene demonstrating outlier expression inhibited cell growth, while knocking down a non-outlier gene had no impact on growth.
"These observations strongly support the notion that outlier kinases represent specific therapeutic targets in individual cancer samples," the authors wrote.
To test the impact of targeting outlier expression in vivo, the researchers created xenograft models of pancreatic tumors from two different cell lines that had demonstrated outlier expression of MET. Treating the mice with a MET inhibitor, cabozantinib, which is marketed by Exelixis as Cometriq to treat medullary thyroid cancer, reduced the size and weight of the tumors in the mice. One of the cell lines had also demonstrated outlier expression of AKT2, a nonspecific target of the inhibitor. Following treatment, the tumors created from this cell line also showed decreased levels of phosphorylated AKT.
Kumar-Sinha said the approach could help open up approved drugs that have been narrowly defined to additional patients that might benefit. "There are very few bona fide targets and those are very narrowly defined," he said.
The University of Michigan maintains a clinical sequencing pipeline run by Arul Chinnaiyan's lab, dubbed Mi-OncoSeq, in which cancer patients receive exome and transcriptome sequencing (CSN 12/19/2012).
Kumar-Sinha said that while he has not yet applied his technique to any of those samples, one next step would be to analyze consented samples that are being sequenced as part of that protocol.
First, though, the team needs to optimize techniques for culturing patient tumor tissue in vitro.
"The idea is to carry out the sequencing analysis and culture in parallel, so that we can test therapeutic candidates derived from clinical sequencing, including outlier kinases," he said.