Skip to main content
Premium Trial:

Request an Annual Quote

Bypassing Phosphorylation, Vandy Team Gauges Kinase Inhibitor Response via Protein Expression Levels


By Adam Bonislawski

Researchers led by Vanderbilt University Director of Proteomics Daniel Liebler have published a proof-of-principle study suggesting that it may be possible to assess patient response to drugs like kinase inhibitors using protein expression profiles.

While the majority of proteomics work in this area has revolved around direct measurement of protein phosphorylation, a protein expression-based approach to assessing drug response could prove more straightforward and stable, since measuring protein phosphorylation levels often requires affinity enrichment of phosphorylated proteins and is highly susceptible to preanalytical variation.

Given phosphorylation’s key role in protein signaling, proteomic analyses of cell signaling networks have typically relied on measurements of protein phosphorylation levels. Proteomic analyses of patient response to protein kinase inhibitors have likewise focused on measuring phosphorylation levels of proteins in the targeted signaling networks.

Questions about the lability of protein phosphorylation – particularly with regard to clinical samples – have complicated efforts to measure these post-translational modifications. Tumor biopsies and other samples taken from patients are still live tissue, meaning that biological processes affecting protein phosphorylation are still ongoing even after the cells have been collected. Such post-excision changes could potentially result in misleading measurements of phosphorylation levels.

This issue has recently become a significant area of concern for proteomics researchers, impacting high-profile projects like the National Cancer Institute's Clinical Proteomic Technologies for Cancer initiative, of which Liebler’s group is a participant. As Amanda Paulovich, a scientist at the Fred Hutchinson Cancer Research Center and also a CPTC participant, told ProteoMonitor in an September interview, questions of sample collection with regard to phosphoproteomics work were "given a lot of airtime" during the August kick-off meeting for the initiative’s second phase (PM 9/2/2011).

A major concern, Paulovich noted at the time, was that tissue samples the group has obtained from the NCI-funded Cancer Genome Atlas might not be appropriate for phosphoprotein analysis given that they were originally collected for genomic, not phosphoproteomic, research.

"Those samples were collected very rigorously with great quality control with genomic analysis in mind, but of course they had no idea that downstream [the samples] would be co-opted for use in proteomics," she said. "So they weren't really collected under ideal conditions for [proteomic] analysis of things like post-translational modifications."

"There are different considerations when you're talking about molecules as labile as the phosphoproteome that respond within seconds to cells being stimulated or perturbed," she added.

“Much of the signaling work that’s been done in proteomics has been done in cell culture models where you have complete control of the system, and it’s very easy to stop an experiment and harvest the cells immediately and control all the preanalytical variables,” Liebler told ProteoMonitor this week. “But if you’re working with human tissues, whether it’s surgical specimens or biopsies, you just don’t have that luxury.”

“So we though that maybe since signaling networks drive genes, which drive transcripts, which drive protein expression, maybe protein expression could provide a readout for the state of a network,” he said.

To test this hypothesis, the Vanderbilt researchers used shotgun analysis on a Thermo Scientific LTQ-XL mass spectrometer to compare protein expression in proliferating A431 cells, EGF-stimulated A431 cells, and A431 cells co-treated with the EGFR inhibitors cetuximab – marketed by Merck, ImClone, and Bristol-Myers Squibb as Erbitux – and gefitinib – marketed by AstraZeneca and Teva as Iressa.

This analysis identified thirteen proteins whose EGF-induced expression changes were reversed by the two inhibitors. These changes were verified in 12 of these proteins using multiple reaction-monitoring mass spec on a Thermo Scientific TSQ Vantage machine.

The researchers then used MRM-MS to measure these twelve proteins in three additional models: a comparison of EGRF inhibitor-sensitive DiFi and EGFR inhibitor-insensitive cell lines; formalin-fixed, paraffin-embedded mouse xenograft DiFi and HCT116 tumor samples; and in tissue from a sufferer of Menetrier’s disease who had been treated with cetuximab. According to the study results, which are detailed in a paper currently in press at Molecular & Cellular Proteomics, a core group of proteins, including c-Jun, jagged-1, and claudin 4, showed decreased expression upon EGFR inhibition in all three models.

These proteins, Liebler noted, should not be considered an actual clinically useful EGFR inhibition signature, but, he said, “I think we established proof of concept for the idea that a protein expression signature demonstrated in a well-controlled model of inhibition – a drug acting on a signaling target – can then be applied in other systems to detect the same event.”

“We’ve established the proof of concept that protein expression changes reflect [changes in] the signaling network,” he said. “If you think logically, that’s not that surprising, but I think no one has showed that before. So we think this kind of approach is now worth exploring.”

George Mason University researcher Emanuel Petricoin – whose research focuses in large part on using phosphoproteomics to analyze cell signaling and drug response and who, with his GMU colleague Lance Liotta, recently developed a new fixative specifically for phosphoproteomic sample collection – called the study “a very interesting paper,” but added that a number of questions still remain.

In particular, he noted, the study doesn’t establish that the observed protein changes are necessarily specific to EGFR inhibition.

“They didn’t compare this and look at other drug effects,” he said. “They didn’t use an mTOR inhibitor, for instance. They didn’t really check for the specificity.”

The MCP study also didn’t establish that protein expression is, in fact, less labile than protein phosphorylation, Petricoin said.

“The underlying premise was that measuring phosphorylation is difficult because it’s so labile,” he said. “But it’s possible that these protein expression changes are maybe just as labile as the underpinning phosphoproteins. In fact, you could argue, that because they are drug responsive, their expression level is obviously changing in response to the signaling that the drug is hitting. So if the signaling proteins are changing because they are so labile, it could be that these expression changes are just as labile.”

Liebler said that the study’s underlying assumption that expression levels are more stable than protein phosphorylation is “part conventional wisdom and part evidence-based on comparisons of [protein expression levels] in fresh and formalin-fixed paraffin-embedded tissues.”

He noted that work was ongoing under the CPTC program to better establish just how stable protein expression is.

“One of the questions [CPTC is] really interested in is the stability of different proteome components to sample handling and processing – particularly in the case of surgical specimens from biopsies,” Liebler said. So CPTC is actually in the process of doing some studies to specifically address that question: What is the effect of preanalytical variables – processing ischemic time, time between excision and freezing, for example – on [post-translational modification] and protein stability?”

“We’ll have data on that by spring,” he noted. “It should provide a much more definitive answer to the questions of what’s stable and what’s unstable under conditions on a time course.”

Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.