Background/Purpose: Rheumatoid arthritis (RA) remains a significant unmet need despite improved therapy. Defining the interlaced nature of gene regulation and disease pathogenesis using integrative systems biology analysis is a powerful approach that can lead to experimentally verifiable predictions of novel therapeutic targets. We have developed a network that can potentially predict gene interactions that contribute to the pathogenic behavior of RA FLS and identify drug synergy to restore homeostasis.

Methods: Rheumatoid arthritis (RA) remains a significant unmet need despite improved therapy. Defining the interlaced nature of gene regulation and disease pathogenesis using integrative systems biology analysis is a powerful approach that can lead to experimentally verifiable predictions of novel therapeutic targets. We have developed a network that can potentially predict gene interactions that contribute to the pathogenic behavior of RA FLS and identify drug synergy to restore homeostasis. Methods: A novel computational systems biology method, known as RAnet, was designed and builds a probabilistic network model capturing the key regulations pertaining to the disease state of imprinted RA FLS. By comparing RA and osteoarthritis (OA) FLS, we curated genes that are critical in RA based on those that satisfied two of the three conditions: 1) differentially methylated genes (DMGs) in RA cells from 28 FLS lines, 2) differentially expressed genes (DEGs) in RA from 37 public microarray FLS lines, and 3) risk genes identified from GWAS studies. A set of 140 genes was selected to create a RA regulatory network. The network was determined by maximizing and adding or removing edges between randomly selected pairs of nodes. Using two and three gene combinations, predicted effects on gene expression were determined by calculating the Pearson and Spearman rank correlation coefficients.

Results: 403 gene-gene interactions were identified in RA Net. Because of the large number of potential combinations, we ranked and prioritized them in terms of the frequency in 1574 predicted models. The top gene/gene interactions included NFkB/GCLC, TP53/MDM2, CACL10/LINC0043, CYSLTR1/BHLHE22, and CARD16/C2orf88 because all were identified in at least 75% of the predicted models. We were also able to identify genes that regulate pathogenic pathways and cytokines in RA. For TNF regulation, top hits included TP53, PSMA4, IL1RN, CCR2, AIRE, JUN, and BMX. For IL-1 regulation, top hits included IL2RA, BMX, TNF, AIRE, and TP53. These core gene sets represent possible therapeutic targets that regulate these cytokines in RA. To determine combinations of genes that could be targeted to convert the RA FLS phenotype to a non-RA phenotype, we predicted the effect of knocking down combinations of genes. The maximum efficacy achievable with three-gene recipes (r_avg = 0.576) is significantly improved compared with two-gene recipes (r_avg = 0.517). The best three gene combination was the knockdown of genes the ring finger protein RNF144B (E3 ubiquitin-protein ligase), UGT2A3 and PLCH2. Interestingly, RNF144B expression is decreased by NSAIDs and RNF144B inhibitors are being developed as therapeutic agents.

Conclusion: This approach identifies synergistic therapeutic targets in RA using systems biology. The genes that regulate pathogenic pathways repeatedly appear in top gene “recipes” and could be promising candidates for combination therapy. This novel in silico method could offer a new way to determine disease mechanisms in RA.

« Back to 2016 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/ra-net-a-systems-biology-approach-to-identify-genes-regulating-pathogenic-pathways-in-rheumatoid-arthritis-ra-fibroblast-like-synoviocytes-fls/