Multi-dimensional immunoproteomics coupled with in vitro recapitulation of oncogenic NRAS^Q61R identifies diagnostically relevant autoantibody biomarkers in thyroid neoplasia

Tumor-associated antigen (TAA)-specific autoantibodies have been widely implicated in cancer diagnosis. However, cancer cell lines that are typically exploited as candidate TAA sources in immunoproteomic studies may fail to accurately represent the autoantigen-ome of lower-grade neoplasms. Here, we established an integrated strategy for the identification of disease-relevant TAAs in thyroid neoplasia, which combined NRAS^Q61R oncogene expression in non-tumorous thyroid Nthy-ori 3-1 cells with a multi-dimensional proteomic technique DISER that consisted of profiling NRAS^Q61R-induced proteins using 2-dimensional difference gel electrophoresis (2D-DIGE) coupled with serological proteome analysis (SERPA) of the TAA repertoire of patients with thyroid encapsulated follicular-patterned/RAS-like phenotype (EFP/RLP) tumors. We identified several candidate cell-based (nicotinamide phosphoribosyltransferase NAMPT, glutamate dehydrogenase GLUD1, and glutathione S-transferase omega-1 GSTO1) and autoantibody (fumarate hydratase FH, calponin-3 CNN3, and pyruvate kinase PKM autoantibodies) biomarkers, including NRAS^Q61R-induced TAA phosphoglycerate kinase 1 PGK1. Meta-profiling of the reactivity of the identified autoantibodies across an independent SERPA series implicated the PKM autoantibody as a histological phenotype-independent biomarker of thyroid malignancy (11/38 (29%) patients with overtly malignant and uncertain malignant potential (UMP) tumors vs 0/22 (p = 0.0046) and 0/20 (p = 0.011) patients with non-invasive EFP/RLP tumors and healthy controls, respectively). PGK1 and CNN3 autoantibodies were identified as EFP/RLP-specific biomarkers, potentially suitable for further discriminating tumors with different malignant potential (PGK1: 7/22 (32%) patients with non-invasive EFP/RLP tumors vs 0/38 (p = 0.00044) and 0/20 (p = 0.0092) patients with other tumors and healthy controls, respectively; СNN3: 9/29 (31%) patients with malignant and borderline EFP/RLP tumors vs 0/31 (p = 0.00068) and 0/20 (p = 0.0067) patients with other tumors and healthy controls, respectively). The combined use of PKM, CNN3, and PGK1 autoantibodies allowed the reclassification of malignant/UMP tumor risk in 19/41 (46%) of EFP/RLP tumor patients. Taken together, we established an experimental pipeline DISER for the concurrent identification of cell-based and TAA biomarkers. The combination of DISER with in vitro oncogene expression allows further targeted identification of oncogene-induced TAAs. Using this integrated approach, we identified candidate autoantibody biomarkers that might be of value for differential diagnostic purposes in thyroid neoplasia.