Metastatic castration-resistant prostate cancer reveals intrapatient similarity and interpatient heterogeneity of therapeutic kinase targets

Abstract

In prostate cancer, multiple metastases from the same patient share similar copy number, mutational status, erythroblast transformation specific (ETS) rearrangements, and methylation patterns supporting their clonal origins. Whether actionable targets such as tyrosine kinases are also similarly expressed and activated in anatomically distinct metastatic lesions of the same patient is not known. We evaluated active kinases using phosphotyrosine peptide enrichment and quantitative mass spectrometry to identify druggable targets in metastatic castration-resistant prostate cancer obtained at rapid autopsy. We identified distinct phosphopeptide patterns in metastatic tissues compared with treatment-naive primary prostate tissue and prostate cancer cell line-derived xenografts. Evaluation of metastatic castration-resistant prostate cancer samples for tyrosine phosphorylation and upstream kinase targets revealed SRC, epidermal growth factor receptor (EGFR), rearranged during transfection (RET), anaplastic lymphoma kinase (ALK), and MAPK1/3 and other activities while exhibiting intrapatient similarity and interpatient heterogeneity. Phosphoproteomic analyses and identification of kinase activation states in metastatic castration-resistant prostate cancer patients have allowed for the prioritization of kinases for further clinical evaluation.

Keywords: combination therapy; metastasis; personalized medicine; phosphotyrosine; resistance.

Fig. 1.

Anatomical location and histological characterization of metastatic CRPC samples used for phosphoproteomics. Metastatic CRPC tissues were obtained from the Rapid Autopsy Program at the University of Michigan. Sixteen samples from 12 different patients are represented and prepared as previously described for phosphoproteomics (8). Red dots indicate the approximate location of the metastatic lesions analyzed. Same-colored lines represent tissues from the same patient. Patient RA53 left lung and left femur were combined due to limiting material (yellow lines). Only tissues with greater than 350 mg and 50% tumor content were evaluated. (Scale bar, 50 μm.)

Fig. 2.

Phosphoproteomic analyses of cell line-derived xenografts, treatment-naïve prostate cancer, and metastatic CRPC reveal distinct phosphopatterns. (A) Unsupervised hierarchical clustering of phosphotyrosine-enriched peptides separates cell line-derived xenograft tumors from primary prostate or metastatic tissue. (B) Further evaluation of a separate run of 10 metastatic CRPC lesions reveals patient-specific and metastatic site similarity of phosphotyrosine peptide patterns. (C) Western blot validation of four different activated kinases identified from both phosphoproteomics and inferred kinase activities confirms the heterogeneity observed across five different patients, as each patient exhibited a unique phosphopattern. Western blot data were separated to highlight each individual patient but were performed on the same western blot. Yellow, hyperphosphorylation; blue, hypophosphorylation. Intensity bar in Fig. 2B is applicable to Fig. 2A.

Fig. 3.

Related phosphokinase and substrate expression patterns are observed within distinct anatomical metastatic lesions of the same patient. Western blot analyses from seven different sets of patients with three or four distinct metastatic lesions were evaluated for kinase activation patterns that were identified in the phosphoproteomic datasets and kinase–substrate relationships or RTKs that have been previously targeted clinically. Each patient expressed similar activated kinase patterns independent of the anatomical location of the metastatic lesions. The unique phosphopatterns are also depicted schematically below the Western blot data.

Fig. 4.

Large-scale analyses of kinase activation patterns confirm intrapatient similarity across multiple, anatomically distinct metastases. (A) Phosphokinase and phospho-RTK arrays were used to analyze metastatic lesions from five different patients from anatomically distinct metastatic lesions. (B) Unique phosphopatterns were observed for each patient, and similar patterns were observed within the same patient, as shown with like-colored circles. Each observable phospho- or total protein spot from the phosphokinase and RTK arrays were used for PCA. LN, lymph node. (C) PCA analysis of all five patients confirms intrapatient kinase expression similarity and interpatient dissimilarity. (D) Grouping metastatic lesions by similar anatomical site shows no significant grouping of samples. Each phosphokinase and phospho-RTK array are spotted in duplicate, and positive control spots are located in the top left, right, and bottom left of each array. The first three principal components represent 77% of the total variance. Adrenal, adrenal gland lesions; LN, distant lymph node lesions; marrow, bone marrow lesion.