Phenotype scanning results.

BackgroundOsteoarthritis (OA) is a prevalent chronic joint disease for which there is a lack of effective treatments. In this study, we used Mendelian randomization analysis to identify circulating proteins that are causally associated with OA-related traits, providing important insights into potential drug targets for OA.MethodCausal associations between 1553 circulating proteins and five OA-related traits were assessed in large-scale two-sample MR analyses using Wald ratio or inverse variance weighting, and the results were corrected for Bonferroni. In addition, sensitivity analyses were performed to validate the reliability of the MR results, including reverse MR analysis and Steiger filtering to ensure the causal direction between circulating proteins and OA; Bayesian co-localization and phenotypic scanning were used to eliminate confounding effects and horizontal pleiotropy. External validation was performed to exclude incidental findings using novel plasma protein quantitative trait loci. Finally, the online analysis tool Enrichr was utilized to screen drugs and molecular docking was performed to predict binding modes and energies between proteins and drugs to identify the most stable and likely binding modes and drugs.ResultFour proteins were ultimately found to be reliably and causally associated with three OA-related features: DNAJB12 and USP8 were associated with knee OA, IL12B with spinal OA, and RGMB with thumb OA. The ORs for the above proteins were 1.51 (95% CI, 1.26–1.81), 1.72 (95% CI, 1.42–2.08), 0.87 (95% CI, 0.81–0.92), and 0.59 (95% CI, 0.47–0.75), respectively. Drug-predicting small molecules (doxazosin, XEN 103, and montelukast) that simultaneously target three proteins, DNAJB12, USP8, and IL12B, docked well.ConclusionBased on our comprehensive analysis, we can draw the conclusion that there is a causal relationship between the genetic levels of DNAJB12, USP8, IL12B, and RGMB and the risk of respective OA.They may be potential options for OA screening and prevention in clinical practice. They can also serve as candidate molecules for future mechanism exploration and drug target selection.

**背景**：骨关节炎（Osteoarthritis, OA）是一种高发的慢性关节疾病，目前尚无有效的治疗方案。本研究采用孟德尔随机化分析（Mendelian randomization analysis）筛选与骨关节炎相关表型存在因果关联的循环蛋白，为骨关节炎潜在药物靶点的发掘提供重要理论支撑。
**方法**：本研究基于大规模两样本孟德尔随机化分析，采用瓦尔德比率（Wald ratio）或逆方差加权（inverse variance weighting）法，评估1553种循环蛋白与5种骨关节炎相关表型间的因果关联，并通过邦费罗尼校正（Bonferroni correction）对结果进行多重检验校正。此外，本研究开展敏感性分析以验证孟德尔随机化结果的可靠性：包括反向孟德尔随机化分析（reverse MR analysis）与施泰格过滤（Steiger filtering），以明确循环蛋白与骨关节炎之间的因果方向；采用贝叶斯共定位（Bayesian co-localization）与表型扫描（phenotypic scanning）排除混杂效应与水平多效性（horizontal pleiotropy）。本研究还利用新型血浆蛋白数量性状位点（plasma protein quantitative trait loci）开展外部验证，以排除偶然发现的假阳性结果。最后，本研究借助在线分析工具Enrichr筛选潜在治疗药物，并通过分子对接（molecular docking）预测蛋白与药物的结合模式及结合能，以筛选出稳定性与结合可能性最优的组合。
**结果**：最终共鉴定出4种蛋白与3种骨关节炎相关表型存在可靠的因果关联：DNAJB12和USP8与膝骨关节炎相关，IL12B与脊柱骨关节炎相关，RGMB与拇指骨关节炎相关。上述蛋白的比值比（Odds Ratio, OR）分别为1.51（95%置信区间，95% CI：1.26~1.81）、1.72（95% CI：1.42~2.08）、0.87（95% CI：0.81~0.92）与0.59（95% CI：0.47~0.75）。同时靶向DNAJB12、USP8与IL12B三种蛋白的预测小分子药物（多沙唑嗪（doxazosin）、XEN 103与孟鲁司特（montelukast））均展现出良好的对接效果。
**结论**：基于本研究的综合分析，可得出以下结论：DNAJB12、USP8、IL12B与RGMB的遗传水平与对应骨关节炎的发病风险存在因果关联。上述蛋白可作为临床骨关节炎筛查与预防的潜在靶点，同时也可作为未来机制探索与药物靶点筛选的候选分子。