NEW YORK (GenomeWeb) – A proteogenomic analysis of breast cancer xenografts suggests that protein patterns in tumors may complement and expand on information available from tumor genomic profiles, sometimes leading to potential treatment targets.
As they reported in Nature Communications yesterday, researchers at Washington University, the McDonnell Genome Institute, and the Broad Institute used a combination of DNA sequencing, RNA sequencing, and mass spectrometry to profile two dozen patient-derived xenografts developed from primary breast cancer tumors, metastatic breast tumors, or Epstein Barr virus-positive lymphoproliferations.
In general, proteomic patterns in the tumors clustered in ways that lined up with mutations in the tumor and with known, expression-based breast cancer subtypes. Even so, the team detected additional, protein-based clues to the pathways that are altered in the tumors, including enhanced expression of AKT proteins or proteins produced from genes such as ARAF, BRAF, and HSP90AB1 in tumors without telltale genetic changes in the gene.
Several breast cancer xenografts had protein patterns that appeared to make them susceptible to treatments that target the HER2 or PI3K pathways, for example. And the combined proteomic and genomic data "revealed that the dual activation of PIK3CA at the genomic level and AKTs at the protein level may be a common signature of breast tumors, affecting more than [20 percent of patient-derived xenografts] in this cohort," the authors noted. "Importantly, our results demonstrate the potential utility of combinatorial inhibitor treatments to treat breast tumors showing these proteogenomic signatures."
Despite the promise of personalized cancer treatments, the team explained, it is often tricky to find and successfully target seemingly druggable mutations, even in tumors that are subjected to sequencing.
To learn more about the actual consequences of somatic alterations and search for additional clues that could impact targeted treatment outcomes, the group established 24 patient-derived xenografts in immunodeficient mice — representing 10 basal breast cancers, nine luminal B tumors, four breast tumors with elevated HER2, and one tumor with low levels of the claudin marker.
The researchers did Sanger targeted DNA sequencing aimed at hotspot mutations on one of the xenografts, while the remaining xenografts were subjected to genome, transcriptome, and/or exome sequencing to characterize their mutation, copy number, and expression patterns.
To that, the team also added proteomic and phosphoproteomic data generated with mass spec, with or without isobaric mass-tag labels. From the tens of thousands of proteins and phosphorylation sites detected in the samples, it narrowed in on 10,069 proteins and more than 36,600 phosphorylation sites that were more completely characterized in the subsequent analyses.
Most recurrent mutations and variant allele patterns in the xenografts lined up well with those described in a subset of the original tumors, the researchers reported. Likewise, gene and protein expression profiles for housekeeping genes coincided with one another quite well in the xenografts, though those expression patterns were less closely linked for some other gene types.
The team's data pointed to somatic mutations, copy number variants, gene expression profiles, and protein patterns that might be druggable in the xenografts. And preliminary experiments on a subset of the xenografts offered a glimpse at some of the alterations that may be susceptible to targeted treatments used alone or in combination.
"[T]his initial work using proteogenomic integration coupled with patient-derived xenograft validation, has demonstrated a strategy that, in principle, may enable more accurate prediction of the efficacy of mechanism-based cancer therapeutics," the authors concluded.