Phosphoproteomic portraits of triple-negative breast cancer (TNBC).

1071 Background: Next-gen sequencing studies demonstrate that each TNBC case harbors a unique set of mutations. We hypothesized that all the different genetic aberrations would converge in discrete patterns of activation of signaling nodes. Pinpointing these nodes would be a parsimonious solution for functional classification. Such nodes (kinases) would also constitute targets. Methods: Training set (n = 34): two groups of early TNBC paired by T, N, G: (A) relapsed in < 4 years; (B) relapse-free > 12 years. Cell lines: 3 TNBC cell lines that develop metastases and kill recipient mice in < 4 months; 7 TNBC indolent cell lines (no metastases in 60 weeks). 250 ug of protein from frozen tumors or cell pellets were purified by Ti+4-IMAC and run in triplicate in an Orbitrap Elite mass spectrometer. Software: Maxquant for peptide identification, PHOSIDA for kinase prediction, Gene-E for clustering. Validation set: phosphohits and hyperactive kinases identified were validated in 113 consecutive early TNBC cases. Mass spectra were translated into H-Scores with an in-house built algorithm and an Ariol scanner. Top hits were pharmacologically targeted on metastatic xenograft models. Results: > 15000 unique phosphopeptides (pPs)were identified and quantified in the training set. 161 and 541 pPs were up-regulated in the A vs B and B vs A groups, respectively. The hyperactivity of 13 kinases (enrichment FDR < 0.1 all) accounted for the pPs upregulated in A. Three of them (PAK2, PIM1, PRKCE) were shared with those driving pPs of aggressive cell lines. The top 40 hits (13 kinases and 27 pPs) enriched in A were validated. A signature characterized by positivity of either STAT3, CYCB, CDK6, P70S6K and/or PRKCE identified 93% of the relapsing cases. RFS of patients positive vs. negative for the signature: 6.7 vs. 11.9 years, LogRank P < 0.001; Cox´s hazard > 550% increase in relapse risk, adjusted by T, N, G, age; P < 0.001. Compared to vehicle, pharmacological blockade of CDK6, STAT3, and P70S6K induced > 2-fold increase in tumor growth control and animal survival in graft models of metastatic TNBC (P < 0.01 all. Conclusions: an unbiased interrogation of the phosphoproteome identifies kinases driving relapsing vs. non-relapsing TNBC. Such nodes constitute targets.