NEW YORK (GenomeWeb) – Researchers with the National Cancer Institute's Clinical Proteomic Tumor Analysis Consortium have identified a pair of proteins as possible keys to the development of drug resistance in triple-negative breast cancer (TNBC).
The findings, published last month in Cancer Research, demonstrate the utility of proteogenomic and phosphoproteomics analyses to uncover mechanisms of cancer drug resistance while also yielding a pair of specific targets for follow up in human clinical trials, said Cynthia Ma, associate professor of oncology at Washington University School of Medicine and senior author on the paper.
The study also provides a look at the third stage of the CPTAC project, which represents a more translation-focused effort than the previous two stages of the initiative. Studies of patient drug response and the development of resistance are at the center of this latest stage.
In the study, the researchers set out to explore the question of why PI3K inhibitors have shown little efficacy in treating TNBC despite the fact that such cancers often exhibit heightened PI3K signaling.
To do this they analyzed a set of six patient-derived xenograft (PDX) models of TNBC that showed varying responsiveness to the pan-PI3K inhibitor buparlisib. Using a combination of transcriptomics, reverse phase protein arrays, and mass spec-based proteomic and phosphoproteomics analysis, the researchers generated transcriptomic, proteomic, and phosphoproteomic profiles of the xenograft tumors with and without buparlisib treatment and at two hours and 50 hours after treatment. These TNBC tumor models had previously undergone exome sequencing, and that information was also included in the analysis.
For their mass spec analyses, the researchers used six-plex tandem mass tag labeling, which enabled them to run each PDX model across five conditions in a single experiment along with a common internal reference sample that allowed them to integrate and compare the data collected across the six total mass spec experiments (one each for each of the six PDX models). Running the samples on a Thermo Fisher Scientific Q Exactive instrument, they quantified an average of 10,027 proteins and 35,412 phosphosites per sample.
Repeating this experiment using an upfront enrichment step to focus only on protein kinases, they quantified an average of 257 kinases per sample and a total of 348 kinases across all 30 samples.
A number of observations came from these analyses, including the "recapitulation of well-known biology," said Filip Mundt, first author on the study and a postdoc in the lab of Broad Institute researcher Steve Carr (also an author on the paper).
For instance, Mundt said, the researchers saw expected cross-talk with the MEK signaling pathway as resistance to PI3K inhibition increased. "Also, we saw the loss of phosphorylation of the well-known AKT substrate AKT1S1/PRAS40," he said, noting that "this has been shown before, and loss of phosphorylation of threonine-residue 246 is a bona fide marker of PI3K inhibition."
Beyond recapitulating these established observations, the CPTAC researchers also identified several other AKT1S1 residues that showed changes in phosphorylation after PI3K inhibition as well as significant phosphorylation changes in the protein Ki-67, which is an established marker of cellular proliferation and a prognostic marker in breast cancer.
"Ki-67 was, in fact, the protein that showed the most significant number of phosphorylation changes after treatment across all [PDX] models," Mundt said, adding that while the researchers haven't determined the exact meaning of these changes, they occurred prior to a decrease in Ki-67 protein expression levels. High Ki-67 expression has been associated in breast cancer with a higher incidence of metastasis and recurrence.
The fact that changes in Ki-67 phosphorylation patterns preceded drops in protein expression suggest "that the phosphorylation status of this cell cycle marker might be a more sensitive readout than its protein level," Mundt said.
The study also identified the proteins NEK9 and MAP2K4 as key players in the development of buparlisib resistance. Mundt cited this as perhaps the most significant finding, noting that these proteins have been relatively little studied.
They identified these proteins by looking for kinase phosphorylation events that were highly enriched both in the tumor phosphoproteomes overall and then in the buparlisib-resistant tumors compared to the sensitive tumors. Using shRNAs to knock down these proteins in a cell line derived from one of the PDX models, they found that knockdown of either inhibited tumor cell growth while increasing sensitivity to buparlisib. Combining proteomic and phosphoproteomic analysis of NEK9 and MAP2K4 and the broader PI3K signaling network with genomic data, the researchers linked the role of these proteins in buparlisib resistance to a specific PIK3CA mutation, demonstrating the potential of proteogenomic analyses to piece together disease-linked cellular processes across multiple levels of omics data.
The move towards proteogenomics within CPTAC began with the initiative's second stage, during which researchers undertook proteomic analyses of three tumor types — breast, colorectal, and ovarian — previously characterized at the genomic and transcriptomic levels by the NCI's Cancer Genome Atlas project.
Henry Rodriguez, director of the Office of Cancer Clinical Proteomics Research, Center for Strategic Initiatives at NCI, noted last year upon announcing the sites participating in the third phase of the project that there is "great potential for new insights to come from the combined analysis of cancer proteomic and genomic data, as proteomic data provides additional critical information that is difficult or impossible to infer from genomic data alone."
In the case of the Cancer Research study, the phosphoproteomic analysis provided many of the most notable insights, Ma said. Because protein phosphorylation is key to protein signaling, it is an area of keen interest within proteomics and cancer research particularly. This is especially the case given the growing use of kinase inhibitors like buparlisib in cancer treatment.
"For the evaluation of kinase inhibitors, phosphoproteomic assessment is the most robust and dynamic method for evaluating on-target and off-target effects as well as adaptive resistance mechanisms upon drug exposure," Ma said. "In addition, levels of phosphoproteins reflect pathway signaling activity that total protein may not. One could imagine that phosphorylated protein levels are better predictors of response to a kinase inhibitor."
In the past, much clinical phosphoproteomics work has been done using RPPA, with researchers like MD Anderson Cancer Center's Gordon Mills and George Mason University's Emanuel Petricoin making extensive use of this antibody-based approach to measure protein phosphorylation in tumor samples and, in some cases, helping to guide treatment.
Mass spec has been less commonly used for this sort of analysis in human clinical samples due to its lower sensitivity, but as the technology's performance has improved, it has become better suited for phosphoproteomic analysis of tumor samples.
The Cancer Research study looked at PDX models and cell lines rather than human clinical samples, but, Ma noted, the findings demonstrated the potential advantages of mass spec.
"Mass spectrometry-based proteomics technology allows more unbiased biomarker research compared to RPPA, which is limited in the number of biomarkers [by antibody availability]," she said. "For example, NEK9 and MAP2K4, which are two markers not on the RPPA panel that we used, would have been missed if we had not done the mass spectrometry-based proteomic analysis."
Ma said she and her colleagues now planned to follow up their findings with additional investigations into the role of NEK9 and MAP2K4 in breast cancer.
"Both NEK9 and MAP2K4 are understudied kinases," she said. "Future studies will focus on validation of this finding in human clinical trials and [trying] to further elucidate their molecular functions in breast cancer and drug resistance."