NEW YORK (GenomeWeb) – Researchers at the H. Lee Moffitt Cancer Center and Research Institute have used a chemical proteomic and phosphoproteomics-based approach to explore the molecular mechanisms underpinning the function of the anaplastic lymphoma kinase inhibitor ceritinib in lung cancer.
The study, which they published last week in Nature Chemical Biology, identified several off-target effects key to ceritinib's activity. More generally, the research demonstrates the potential of such work to enable rational selection of combination therapies as well as identification of candidate biomarkers for predicting patient responses to therapy, said Uwe Rix, an assistant professor at Moffitt and senior author on the paper.
Combining chemical proteomics to identify protein interactors of ceritinib (marketed by Novartis as Zykadia) with phosphoproteomics to assess signaling activity across various protein pathways allowed the researchers to develop an understanding of both the drug's activity and the behavior of the cancer cells more broadly, Rix noted.
"What you can achieve with this [information] is that you might understand the mechanism of action of that particular drug, but then you can also use that understanding to, for instance, design a new drug combination that actually has enhanced efficacy, and at the same time maybe identify a new biomarker candidate that has a mechanistic rationale behind it," he said.
Rix and his colleagues began the study with a phenotypic screen of 240 compounds against 20 non-small cell lung cancer cell lines harboring a variety of driver mutations. Using unsupervised hierarchical clustering they found that drugs to the same target clustered together in terms of their effects on these cell lines. Ceritinib, however, was a notable exception. Like the other ALK inhibitors tested, ceritinib was effective against the ALK-rearranged cell line used in the study, but it was also effective in several additional cell lines where other ALK inhibitors were not.
The researchers then tried to identify the mechanisms responsible for this unanticipated activity. Using a chemical proteomics platform they probed the NSCLC lines to identify ceritinib's protein targets. In addition to its intended target ALK and the known off-target IGF1R, the drug bound to kinases including FAK1, RSK1/2, ERK1/2, CAMKK2, and FER, with FAK1 and RSK1/2 being the most highly enriched in the cell lines where ceritinib was effective.
They then used mass spec-based phosphoproteomics to identify protein signaling changes due to ceritinib treatment. Combining immunoprecipitation of tyrosine phosphopeptides with mass spec they identified 435 unique phosphopeptides, while enriching with immobilized metal ion affinity chromatography (IMAC) and using SILAC mass spec for quantification they identified 4,433 phosphopeptides. Of these, they identified 121 phosphopeptides that were significantly upregulated and 165 that were significantly downregulated in all replicates, which they used to build a ceritinib effector network consisting of 139 protein nodes and 312 interactions, with RSK1/2, IGF1R, and FAK1 among the most central nodes of this network.
Using both siRNA knockdown and small molecule-based targeting, Rix and his colleagues looked at how inhibition of these kinases affected cell viability, finding that targeting them singly had little effect. Simultaneous inhibition of these targets, however, significantly decreases cell viability, indicating that ceritinib's polypharmacological activity is key to its effectiveness in non-ALK-rearranged cell lines.
Further exploration of downstream molecules affected by ceritinib treatment indicated that the drug inhibited phosphorylation of the transcription factor YB1, which has been found to facilitate resistance to microtubule-targeting agents like paclitaxel. This led the researchers to test a combined ceritinib-paclitaxel treatment, which they found to be particularly effective, including in NSCLC lines where single-drug treatment was not effective.
Rix said the study provides an example of what is likely a common phenomenon among targeted drugs generally.
"I think it's fairly common that targeted drugs actually have a much broader target spectrum than anticipated," he said. "We have, for instance, recently seen a compound that has a dual mechanism of action wherein it hits two different kinases. One [target] was already known, but the other one was actually almost equally important for the overall mechanism of action. So sometimes you do see these things, and I believe there are going to be more cases like that and probably some hidden opportunities that can be looked into."
The challenge, though, is in teasing out these complex mechanisms of action — distinguishing between helpful and harmful off-target effects as well as identifying weak but nonetheless important off-target binding.
"It's relatively difficult to identify [such effects] and to link them up with mechanistic studies to really look at how they contribute to the overall mechanism of action," Rix said. "But it's only when you actually get to the point that you understand the mechanism of action and can utilize that for designing better drug combinations or finding mechanistic biomarker candidates that you start to reap the benefits of this."
Rix said he and his team are now hoping to apply their findings to clinical studies and that they are currently in discussion with their Moffitt colleague and study co-author Eric Haura about collaborating on such work.
He said they are also following up with studies of other drugs that, like ceritinib, showed activity not anticipated based on their stated targets.
He noted that among the benefits of such an approach is the possibility of identifying treatments for cancers not harboring specific mutations or alterations that make them conventional candidates for targeted therapies.
For instance, in the NCB study, he and his colleagues were "specifically looking [at cell lines] that don't really have any targeted therapies yet," Rix said. "So we focused more on [lines] that did not have an EGFR mutation or an ALK translocation. The panel of cell lines that we were looking at were more enriched for, for instance, KRAS mutations, which are largely un-druggable these days, and some others where there's really no known driver mutation one could target."
"I think this is an area where there are many new findings to be made," he said. "We're excited about exploring this further."