A team from the Barts Cancer Institute at the Queen Mary University of London has completed a quantitative phosphoproteomic analysis of acute myeloid leukemia cells that suggests that resistance to kinase inhibitor therapies in cancer may be keyed by multiple pathways outside those targeted by the drug.
The study, published online in Molecular and Cellular Proteomics last month, measured global phosphorylation in AML cell lines with differing sensitivities to inhibitors targeting PI3K, MEK, and JAK pathways to determine whether resistance to the inhibitors was due only to activation of each drug's kinase target or instead due to activity of parallel pathways.
The group found that intrinsic resistance could not be explained by activity in only the target kinase in two out of the three cases studied. According to Pedro Cutillas, a senior lecturer at the Barts institute and lead author on the paper, this indicates that the widely held paradigm of "oncogene addiction" may not hold true for all cancers.
Cutillas told ProteoMonitor in an e-mail this week that the results suggest that "trying to predict sensitivity to a given inhibitor by examining the activity of the pathway to be targeted only will give lots of false positives and false negatives."
"I think that oncologists will need to start looking more broadly at the circuitry of the biological network in each individual patient before trying to predict the outcome of a given therapeutic strategy," he said.
While "addiction" to a single pathway may be the case in a small proportion of cancers, he added, "there is no reason why a given cancer cell population can only over-activate one of the pathways out of the many different ones present in cells."
According to Cutillas, it is becoming increasingly apparent through clinical examples that the paradigm does not always hold. "For example, recent clinical trials have shown that a very small proportion of patients with activating mutations on … PI3K respond to therapies that target this enzyme," he said. The group's phosphoproteomic study, he added, provided molecular evidence to show how this could be so.
In their analysis, Cutillas and his team used a label-free nanoflow-liquid chromatography tandem mass spectrometry workflow on a Thermo Scientific LTQ-Orbitrap XL instrument.
The researchers created their own informatics software, called Pescal, to support the label-free method, Cutillas said, because at the time they started the project they were not satisfied with existing software tools for non-labeled LC-MS/MS.
While isotope labeling approaches like SILAC and iTRAQ offer highly accurate quantification, the team needed an approach better suited for clinical studies. "We wanted to compare large sample numbers and biological and technical replicates. In the long run we would like to apply these techniques in primary samples as part of clinical trials," Cutillas said. "Therefore, labeling techniques were not appropriate for this work."
"What we do instead is to compare the intensities of the phosphopeptide ions across samples directly without having to derivatize proteins with isotope labels," he explained, saying the method is "more complex than quantitative proteomics based on labeling because we have to compare each identified peptide across different LC-MS runs.
In the MCP paper, the researchers noted they have found that more than 90 percent of phosphopeptides analyzed using the label-free strategy could be quantified with "good precision" and accuracy deviation below 50 percent in earlier experiments.
In the AML study, the group quantified global phosphorylation across several resistant and sensitive cell lines either treated or untreated with three kinase inhibitors — MEK1, JAK-1, and PI-103 — targeting a trio of pathways known to have a role in AML.
The researchers chose cell line pairs of opposing sensitivity and resistance to the inhibitors for further analysis and measured phosphorylation in several replicates of each line. They investigated whether any of the phosphorylation sites measured correlated with the cell line's sensitivity to inhibition.
Several phosphorylation sites that were inhibited by the three compounds had intensities that "did not correlate with growth inhibition sensitivity," the authors wrote. Meanwhile, hundreds of phosphorylation sites that did correlate with sensitivity and resistance were not inhibited by the compounds, they found.
Overall, the group found that markers of kinase activity targeted by MEK1 were more intense in the cell line that was sensitive to this inhibitor. In contrast, the markers of kinases inhibited by PI-103 and JAK-1 were more intense in resistant cells, the authors wrote.
Mining data, the group also found that signals of phosphorylated PKCs were overall more intense in resistant cells than in sensitive cells, suggesting PKCs may also be implicated in AML resistance to various kinase inhibitors.
As a whole, the results indicated that while in some cases the activation status of a targeted pathway can indicate sensitivity to its associated inhibitor, this may not be sufficient to predict sensitivity in all cases.
Cutillas said that he and his team are now expanding the project to study a larger number of cell lines as well as "primary AML tissue," building on their proteomic and bioinformatics strategy from this initial study.
In addition to his position at Barts, Cutillas is a founding director of the biotech firm Activiomics, which, he said, provides phosphoproteomic services for several companies, including Genentech and GlaxoSmithKline.
He suggested that if the group's findings hold, phosphoproteomic analysis could be a valuable tool to better predict the best combination therapy for patients.
"Currently, oncologists ask… 'Is drug X likely to work for patient Y?,' [while] phosphoproteomics [could] give oncologists the opportunity to ask open questions [like] 'which drug or drug combination is likely to work for patient Y?," he said.
A number of researchers and firms aside from Cutillas and Activiomics are also pursuing similar efforts using phosphoproteomics to guide drug treatment.
For instance, proteomics firm Theranostics Health has amassed a database of phosphorylation profiles in a number of cancers that it hopes to use in companion diagnostic development (PM 8/13/2010). Last month, a team led by University of North Carolina researcher Gary Johnson published a paper in Cell in which they used phosphoproteomics to predict drug combinations that would overcome kinase inhibitor resistance in triple-negative breast cancer patients (PM 4/20/2012).