Genetic prioritization of HSDL2 in suicide and intentional self-harm: multi-omics evidence linking lipid metabolism and neuronal vulnerability

作者：Z Zhang, Y H Wu, Fanxu Meng, Hanbing Ji, Hai Zhong, Shucheng Si, Yun WEI, Yilei Ge, Sijia Wu, Qingxin Luo, Le Wang, Tiemei Liu, Jiawei Xiu, Yi Guo, Yue Li, Lei Hou, Hao Chen, Xiaoru Sun, Y Yu, Qingzhen Hou, Shanshan Gao, Fuzhong Xue, Hongkai Li · 发表于：BMC Medicine · 年份：2026 · DOI：10.1186/s12916-026-05066-6 · 研究领域：Suicide and Self-Harm Studies、Tryptophan and brain disorders、Genetic Neurodegenerative Diseases

Suicide or other intentional self-harm (SSH) represents a critical global public health burden, yet biologically informed targets remain poorly defined due to its complex and incompletely understood etiology. This study aimed to prioritize genetically supported candidate targets for SSH and characterize their biological and translational relevance. We developed a multi-omics evidence framework integrating transcriptomic and proteomic Mendelian randomization, colocalization, and sensitivity diagnostics. The primary SSH outcome was from FinnGen R12 (11,538 cases; 488,810 controls). eQTLGen and GTEx were used for transcriptomic analyses, and UKB-PPP and ARIC for proteomic validation. Downstream analyses included functional enrichment, tissue expression profiling, exploratory metabolomic pathway analysis, brain cell-type-specific eQTL analyses, phenome-wide liability scanning, structural modeling with docking, virtual screening and molecular dynamics simulations, and baseline SSH classification in UKB-PPP individual-level proteomic data. Transcriptome-wide MR identified 1,164 FDR-significant genes, of which 55 were retained as Tier 1–4 prioritized candidates after proteomic validation, colocalization, and sensitivity assessment. HSDL2 was the sole Tier 1 candidate, supported by consistent eQTL and pQTL evidence, adequate instrument strength, and colocalization at the HSDL2 locus (PP.H4 = 0.797). Genetically predicted HSDL2 expression was associated with higher SSH risk in eQTLGen...