NEW YORK (GenomeWeb) – A team led by researchers from the Technical University of Munich has used chemical and phosphoproteomics to characterize the targets of 243 kinase inhibitors that have either been approved as drugs or tested in humans.
Detailed in a study published last week in Science, the work is part of a larger effort by the researchers to build a database containing target information for as many kinase inhibitors as possible, said Bernhard Küster, chair of proteomics and bioanalytics at TUM and senior author on the paper. He and his colleagues are currently targeting an additional 1,000 compounds provided by Roche and GlaxoSmithKline and plan to add data on new inhibitors to the database as they become available.
This sort of target data has a number of potential uses, Küster said, noting that it will provide researchers and clinicians with a better understanding of the molecules' off-target effects as well as help with drug repurposing efforts.
"I think that it's just a fundamentally useful piece of information, knowing for any one molecule what the target spectrum is," he said. "Generally people think that if something is labeled as a MEK inhibitor, or as an EGFR inhibitor, this is what it will do. Whereas many, most of these molecules, actually do many more things."
In some cases, these off-target effects are linked to drug side-effects. In other cases, they may be linked to the actual therapeutic benefit of these agents. For instance, in an October study in Nature Chemical Biology, researchers at the H. Lee Moffitt Cancer Center and Research Institute used an approach similar to that employed by Küster and his colleagues to explore the molecular mechanisms underpinning the function of the anaplastic lymphoma kinase inhibitor ceritinib in lung cancer and found that several off-target interactions are key to that drug's activity.
Both Küster and Moffit researcher Uwe Rix, senior author on the Nature Chemical Bio study, previously worked at Cellzome, a German chemical proteomics company where Küster was formerly an executive. That firm, which GSK acquired for $99 million in 2012, developed a version of the chemical proteomics technology used by both researchers in their recent work.
A better understanding of the full range of an agent's molecular targets could also help researchers and clinicians optimize the use of existing drugs as well as repurpose them for new indications. This is becoming particularly relevant in areas like cancer where tumor sequencing is increasingly used to guide therapy, Küster suggested.
"If that patient has a certain CDK genetic phenotype, you would at least be able to consider which of the CDK inhibitors you might want to try," he said. "There are maybe twenty CDK inhibitors out there, but they all at different specificities. And maybe, in my patient I have the CDK activating mutation and I also have an AKT activating mutation. Maybe there is a compound that would actually address both AKT and CTK at the same time. That might be a molecule one would consider using for management of these patients."
In the Science study the researchers identified a potential repurposing opportunity for the MET/VEGFR inhibitor cabozantinib, which is marketed by Exelixis as Cabometyx for advanced kidney cancer and as Cometriq for medullary thyroid cancer. Their chemical proteomics work found that this compound was also an effective inhibitor of the mutated tyrosine kinase FLT3-ITD, which plays a role in forms of acute myeloid leukemia. Based on these findings, they tested the drug in AML cell lines and in AML mouse xenografts, finding that the drug reduced tumor cell growth and improved mouse survival by a statistically significant amount.
Küster and his colleagues highlighted the case of MELK inhibitors in non-small cell lung cancer as another example of the approach's potential in drug development. In an analysis of lung cancer tissue microarrays, the researchers found that high MELK expression in squamous cell carcinoma correlated with poor survival, bolstering previous research that has identified that protein as a potential prognostic marker and therapeutic target.
However, as they noted, the role of MELK in SCC-NSCLC remains cloudy. Cell line work has shown it is not required for cancer growth. At the same time, it could still play a role in tumor maintenance or other processes linked to disease. The MELK inhibitor OTS-167 is currently in phase I clinical trials, but the TUM researchers found that this drug has broad action across many different kinases, and so, they noted, its effectiveness may not be due to MELK inhibition at all, leaving the question of MELK's role in lung cancer uncertain.
To explore this question further, they co-crystalized MELK with several other inhibitors, identifying a pair of cysteine residues present in only 1 percent of human kinases that, they noted, could be targeted for development of more select MELK inhibitors.
Kuster noted that, particularly in the case of more recently developed kinase inhibitors, pharma companies often have generated broad target data themselves, but, he said, that data is not necessarily published.
"Pharma companies often have this information, but it doesn't necessarily filter out to the public domain," he said. He cited a literature search he and his co-authors did as part of the study in which they found that of the 200,000 publications listed in PubMed or SciFinder involving kinase inhibitors, more than half focused on one of just five drugs: rapamycin, imatinib, sorafenib, gefitinib, and erlotinib. Of the 243 compounds they looked at, 17 had no PubMed entry and 70 were included in fewer than 10 publications in PubMed.
"In the public domain a lot of times these compounds have only been tested in a limited set of assays," Küster said. "And clearly what our work argues is that these compounds should be tested against a wider panel [of potential targets]."
The 243 compounds explored in the Science study have all been tested in humans, making them potential therapeutic agents. As the TUM team expands its database, it plans to begin looking at agents not tested in humans but which are being used or could be used as research probes, Küster said.
"We are quite sure that people don't know that the tools that they are using [to study a specific kinase] have many more effects, and that therefore they need to control much better their experiments so that they are able to attribute the biological effect that they are measuring to the action of the compound and the target that they think they are inhibiting," he said.
He said he and his colleagues are also using the database to inform their own internal drug discovery programs.
"We are an academic lab, so we can't compete with pharma, but we see this as an opportunistic way of making more use for compounds starting from what are already very good molecules," he said. "Anything that has made it into [testing in] people has passed all the pre-clinical check boxes, so that makes them very valuable molecules."
Additionally, Küster said his lab hopes to take the target data and combine it with more extensive phosphoproteomic analyses to better determine the mechanisms of action of these compounds.
"We would like to know what other molecular consequences in the cancer cell follow from inhibition of one or two or 10 different targets," he said. "That's a question that is very relevant to understanding how those compounds work, because it's often not necessarily that the inhibition of the primary target is responsible for the actual effect that the compound has on the cell."
"This is something my pharma collaborators say is often something that they don't really know," he added. "Is the compound engaging the target in the cell, or is it actually engaging the whole pathway in the cell? And how do I measure that?"