Exploration of new biomarkers for rheumatoid arthritis based on plasma proteomics and transcriptome integrated with genome-wide Mendelian randomization.

Objective

Early diagnosis of seronegative rheumatoid arthritis (SNRA) is often challenging due to lack of reliable serological markers. The aim of this study was to explore protein biomarkers of SNRA and identify their potential as therapeutic targets.

Methods

Third-eight seropositive (SPRA) patients and 20 SNRA patients were enrolled in our omics study. Differentially expressed proteins (DEPs) between the two groups were identified via the Astral mass Analyzer for quantitative proteomics using data-independent acquisition (DIA) mass spectrometry, and the dataset of GSE93272 was used for the validation test. A variety of bioinformatics algorithms such as random forest, support vector machine, nomogram construction, consensus clustering were applied to identify key regulators, expression patterns, and molecular subtypes. We also performed functional enrichment analysis of differentially expressed genes (DEGs) across clusters and evaluated immune cell infiltration using single-sample gene set enrichment analysis (ssGSEA). Mendelian randomization (MR) analyses were conducted to identify druggable targets, and colocalization analyses were performed to determine whether RA risk and the expression of these druggable genes were influenced by shared SNPs. Additionally, we validated these findings through both in vitro and in vivo experiments, using clinical peripheral blood samples, RAW264.7 cells, and the collagen-induced arthritis (CIA) rat model to assess the functions of key genes in RA.

Results

Using quantitative proteomics, we identified 22 DEPs between SPRA and SNRA, including 12 upregulated and 10 downregulated proteins. Notably, CAMK2G protein was positively correlated with inflammation indicators in RA patients. Differential expression analysis between RA samples and controls identified nine key DEPs as potential biomarkers. A random forest classifier and a nomogram incorporating these nine biomarkers were constructed, and their clinical utility was supported by decision curve analysis. In addition, consensus clustering based on these hub genes stratified RA patients into distinct subgroups characterized by differential immune cell infiltration patterns and polygenic risk scores. Significant association between NR1H3 and disease by colocalization analysis in MR suggests its potential as druggable target. Notably, our experiments have confirmed that CAMK2G expression is associated with enhanced osteoclastogenic capacity, and enhanced CAMK2G expression accompanies joint damage in CIA rats.

Conclusion

Our research on polygenic risk model indicates CAMK2G protein could serve as a promising biomarker and NR1H3 might be a potential candidate for targeting therapy for RA.