Proteomics of Metastatic and Non-Metastatic Uveal Melanoma

Abstract

Purpose : Proteomic analysis of primary uveal melanoma (pUM) that were either metastatic and non-metastatic was pursued for insights into UM pathobiology, mechanisms of UM metastasis, and to facilitate the identification of protein biomarkers of UM metastasis.

Methods : pUM (33 that metastasized & 43 that were non-metastasizing) were collected from enucleated eyes at the University of Liverpool and the Cleveland Clinic and were subjected to quantitative proteomic analysis using LC MS/MS iTRAQ technology. Thirteen choroid specimens excised from UM eyes distant from the tumors (6 Mets & 7 NonMets) were pooled following tryptic digestion and utilized as a reference control. Tryptic digests of pUM were labeled with unique iTRAQ tags, mixed with the pooled control in batches of 8, and fractionated by RPHPLC at pH10. Chromatography fractions were subjected to LC MS/MS on an Orbitrap Fusion Lumos Tribrid mass spectrometer. Protein identification utilized the Mascot 2.6.2 search engine, the human UniProt database, required a minimum of 2 unique peptides, and a false discovery rate ≤ 1%. Protein quantitation utilized the weighted average method. Differentially expressed (DE) proteins exhibited group differences with p ≤ 0.05 (t-test) after correction for multiple comparisons and were used for bioinformatics analyses, principal component analyses (PCA) and for developing predictive models.

Results : Total proteins quantified per UM sample ranged 2168-2818 (average 2580) with 1541 proteins quantified in all 76 tumors. More decreased than increased proteins were detected between the two groups. Of 424 DE proteins identified, tryptophan-tRNA ligase and HLA class II histocompatibility antigen DR-α exhibited the greatest difference and were >2-fold more abundant in metastasizing UM than non-metastasizing UM. Ingenuity Pathway Analysis of 131 DE proteins with UM ratios differing by about 1.3-fold implicates immunological and protein synthesis molecular and cellular functions in UM pathobiology. PCA using 35 DE proteins with Met/NonMet differing by 1.5-fold or more allowed >80% correct prediction of metastatic status of 20 pUM.

Conclusions : We are pursuing a rigorous proteomic characterization of pUM and expect our final results to provide new insights into mechanisms of UM metastasis as well as reveal the capabilities of proteomics technology for discriminating between metastatic and non-metastatic pUM.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.