PHILADELPHIA (GenomeWeb) – Researchers from the National Cancer Institute's Clinical Proteomic Tumor Analysis Consortium (CPTAC) presented findings from the group's work at the American Association for Cancer Research annual meeting here this week, making the case for proteomics as a necessary complement to more widespread cancer genomics techniques.
A five-year, $100 million-plus project launched in 2011, the CPTAC initiative has done protein biomarker discovery and verification studies in tumor tissue samples previously characterized at the genomic and transcriptomic level by the NCI's Cancer Genome Atlas (TCGA) team. Specifically, the consortium, which comprises researchers from institutions around the country, including eight primary centers, undertook analysis of three tumor types – breast, colorectal, and ovarian – with the aim of profiling around 100 samples of each.
CPTAC researchers working on each of the three tumor types presented at this week's meeting, highlighting significant findings from their work and making the case for the utility of a proteogenomic approach to cancer research.
For example, Vanderbilt University researcher Daniel Lieber presented findings from his team's analysis of 95 colon and rectal tumors – results of which were published last year in Nature – in which they determined that while gene copy number alterations correlated well with changes in mRNA abundance, few CNAs showed effects at the protein level.
Among the CNAs that did lead to strong changes at both the mRNA and protein level was the chromosome 20q amplicon, and by combining the genomic and proteomic data the researchers were able to identify proteins including HNF4A, TOMM34, and SRC as potentially key players in the disease.
Liebler and his colleagues also looked at correlation between mRNA and protein levels more broadly, finding that while abundances of the two correlated well in individual tumors, they correlated less well across the full collection of tumors. Looking at the specific functions of these gene products, they found that the level of correlation between mRNA and protein was significantly dependent on their biological function. For instance, genes involved in metabolic processes had strong correlation between mRNA and protein levels, while those involved in ribosome and spliceosome processes as well as oxidative phosphorylation correlated poorly, or, in some cases showed negative correlation.
The other CPTAC analyses had similar findings, with Broad Institute researcher Philipp Mertins, who presented on the group's breast cancer analysis, and Pacific Northwest National Laboratory researcher Karin Rodland, who presented on ovarian cancer, noting that mRNA-to-protein correlation depended significantly on function, and highlighting largely the same relationships between biological pathways and mRNA-to-protein correlation as did Liebler. The findings provide additional insight into the question of the relationship between mRNA and protein levels, an issue of considerable interest within omics research, particularly as many have sought to use mRNA measurements as a proxy for protein measurement.
Analyzing 174 ovarian cancer tumors, Rodland and her colleagues identified a number of CNAs linked to changes in protein abundance, many of which were linked to processes like proliferation, cell motility and invasion, and immune regulation that are involved in cancer. They also identified five proteomic subtypes in the TCGA cohort, three of which correlated to mRNA subtypes and two of which were not seen at the RNA level.
Looking for explanations of the high chromosomal instability characteristic of ovarian cancer, the CPTAC researchers also identified a number of proteins linked to chromatin remodeling upregulated in patient tumors and also established a link between high levels of acetylation on histone H4 and poor patient survival.
They also identified through phosphoproteomic analysis a set of signaling pathways upregulated in patients with poor survival, information that could be used to develop markers for selecting patients likely to benefit from specific therapies or at particular risk of developing resistance to therapy.
Mertins and his colleagues likewise looked for the effects of genomic aberrations in the breast cancer proteome and phosphoproteome, and similarly found that while evidence of gene variants was rare at the protein level, protein changes linked to known key mutations could be detected in some cases. For instance, tumors with the common breast cancer mutations PIK3CA and TP53 exhibited specific phosphorylation changes in signaling pathways downstream of these mutations.
The CPTAC researchers were able to detect protein changes in the case of only 1 to 2 percent of the roughly 9,600 single amino acid variants and 36,000 alternative splice junctions identified through the TCGA's genomic characterization of the breast cancer samples.
Washington University researcher Matthew Ellis, who is also working on the CPTAC breast cancer analysis discussed work he and his colleagues are doing using the multi-inhibitor kinase bead affinity chromatography technology developed by University of North Carolina, Chapel Hillresearcher Gary Johnson to identify possible drug targets for the disease.
Running samples through chromatography columns functionalized with kinase inhibitors, the researchers are able to pull out active kinases from tumor samples, which they can then identify and quantify via mass spec, allowing them to look for outliers that are especially upregulated in disease cases.
And Fred Hutchinson Cancer Research Center researcher Amanda Paulovich presented on the CPTAC group's efforts to develop multiple-reaction monitoring assays for targeted protein analysis, with the aim of ending researchers' reliance on antibody-based techniques like Western blots for protein quantitation.
Echoing the concerns of many proteomics researchers, she noted the poor reliability and limited availability of antibodies and highlighted targeted mass spec – and MRM and immuno-MRM specifically – as a promising alternative.
Demonstrating the potential of the technique, Paulovich presented data on a pair of assays she and her colleagues have developed for measuring proteins in the DNA damage response network – an 82-plex immuno-MRM assay for measuring signaling activity in this pathway and a 135-plex IMAC-MRM assay targeting the same pathway but using IMAC enrichment of phosphopepetides instead of antibody enrichment.
Paulovich also highlighted the CPTAC assay portal, which currently contains 554 MRM assays to 306 proteins as well as 314 monoclonal antibodies for use in immuno-MRM assays that researchers can obtain at cost. She noted that by the end of summer 2016, the portal will contain around 2,000 MRM assays provided by the various CPTAC groups.
Generating MRM content will be key to widespread adoption of the technology, she said, noting that ultimately MRM will ideally be provided as a service where researchers can easily obtain kits with all the reagents and standards required to measure their proteins of interest. Currently, she said, the technology suffers from something of a "chicken or the egg" problem, with vendors reluctant to invest in building MRM offerings due to low researcher demand, and research demand being stifled by a lack of easily accessible MRM assays.
To that end, Paulovich and other researchers are exploring public-private partnerships that could foster MRM development with the ultimate aim of making developing and offering assays a viable commercial business.