Diagnostic and Prognostic Biomarkers of Rheumatoid Arthritis (RA) in Pakistani Population

Abstract

Rheumatoid arthritis is an autoimmune disorder of complex disease etiology. The serological and genetic markers of rheumatoid arthritis have not been completely deciphered. Currently available serological diagnostic markers lack in terms of sensitivity and specificity and thus additional biomarkers are warranted for early disease diagnosis and management. Genome-wide association studies have enabled simultaneous identification of single nucleotide polymorphisms associated with complex disorders such as rheumatoid arthritis. However, polymorphisms that are unable to match the significance threshold can be missed. Data integration can help in identification of these variants. Current study aimed to screen and compare the serum proteome profiles of rheumatoid arthritis serotypes with healthy controls in the Pakistani population for identification of potential disease biomarkers. The present work then intended to identify novel candidate non-coding risk variants for rheumatoid arthritis using a data integration pipeline. Serum samples were collected from Pakistani rheumatoid arthritis patients and healthy controls. The samples were enriched for low abundance proteins using ProteoMinerTM columns. Patients were assigned to one of the four serotypes based on anti‐citrullinated peptide antibodies and rheumatoid factor. Serum protein profiles were analyzed via liquid chromatography-tandem mass spectrometry (LC-MS/MS). The changes in the protein abundance were determined using label-free quantification software ProgenesisQI™. Ingenuity pathway analysis was used to analyse the pathways associated with the differentially expressed proteins. Findings were validated in an independent cohort of patients and healthy controls using enzyme-linked immunosorbent xix assay. 340 significant single nucleotide polymorphisms for rheumatoid arthritis were chosen from published genome wide association studies. SNipA proxy search tool was used to identify single nucleotide polymorphisms linked to query polymorphisms that were then scored using RegulomeDB, hereby named as proxy single nucleotide polymorphisms. Single nucleotide polymorphisms with scores less than three were annotated. Expression quantitative trait loci linked to these single nucleotide polymorphisms were studied for protein-protein interactions using STRING database. Single nucleotide polymorphisms linked to key proteins were further annotated using the SNPfunc tool. A total of 213 proteins were identified. Comparative analysis of all groups (false discovery rate less than 0.05, greater than 2-fold change, and identified with more than 2 unique peptides) identified ten proteins that were differentially expressed between rheumatoid arthritis serotypes and healthy controls including pregnancy zone protein, selenoprotein P, C4b-binding protein beta chain, apolipoprotein M, N-acetylmuramoyl-L alanine amidase, catalytic chain, oncoprotein-induced transcript 3 protein, carboxypeptidase N subunit 2, apolipoprotein C-I and apolipoprotein C-III. Pathway analysis predicted inhibition of liver X receptor/ retinoid X receptor activation pathway and production of nitric oxide and reactive oxygen species pathway in macrophages in all serotypes. Protein interaction analysis identified 13 ‗hub proteins‘ expressed by the expression quantitative trait loci linked to 54 single nucleotide polymorphisms. Of these, nine were already reported for rheumatoid arthritis. Remaining 45 novel polymorphisms, mapped to 11 genomic loci, are novel candidate risk variants for rheumatoid arthritis. Of 9194 proxy single nucleotide polymorphisms, 492 single nucleotide polymorphisms returned significant RegulomeDB scores and mapped to 94 expression quantitative trait loci. xx Conclusively, the current study has explored the untapped proteomics of Pakistani rheumatoid arthritis patients and identified catalogue of serum biomarkers with diagnostic and prognostic potential of Pakistani rheumatoid arthritis patients. These serum biomarkers can be further tested in larger cohorts for evaluation of their diagnostic potential. The study also used a data integration pipeline to identify the putative risk variants for rheumatoid arthritis that might have been missed by genome wide association studies. These missed variants can help to fill the current gaps in the knowledge of rheumatoid arthritis‘ genetics. Further, the proposed data integration pipeline can be incorporated into a ready-to-use computational package to accelerate the identification of missed variants for other complex disorders.