Biologyarticle2026-09-18

Integrated multi-omics decodes the AKI-to-CKD transition: from ensemble discovery to structure-guided translational targeting

Open access0 citations

Abstract

The transition from acute kidney injury (AKI) to chronic kidney disease (CKD) constitutes a major global health burden. Identifying causal drivers remains challenging due to proteomic complexity and confounding biases. We aimed to construct a discovery framework to identify actionable targets driving this pathological conversion. Among 53,014 UK Biobank participants with Olink proteomic data, 2,532 AKI patients (410 progression events) were identified and included in the subsequent predictive modeling. We developed a ‘triple-orthogonal screening strategy’, integrating an ensemble machine learning framework (six algorithms including LASSO and RF-RFE), Cox proportional hazards models, and two-sample Mendelian randomization (MR) to filter 2,923 proteins. Findings were validated via single-cell transcriptomics and a mouse model of ischemia-reperfusion injury characterized by paired bulk RNA-sequencing and TMT-based quantitative proteomics. We identified 25 key proteins possessing both statistical prognostic value and genetic support. An ensemble prediction model based on these features achieved an ROC-AUC of 0.939. Using LGALS3 as a paradigm, multi-omics validation revealed a unique ‘transcriptional-translational discordance’—characterized by sustained protein accumulation despite transcriptional silence—during the maladaptive repair phase (day 14). Furthermore, structural profiling identified a hotspot on LGALS3 (residues Glu185 and Gln187), where the binding pocket for the inhibitor olitigaltin and lactose (anhydrous) spatially converges with high-scoring B-cell epitopes. We decoded the prioritized proteomic landscape of the AKI-to-CKD transition. This study establishes a framework integrating multi-layer prioritization with structure-guided validation, identifying pathogenic drivers characterized by transcriptional-translational discordance. This workflow provides atomic-level coordinates for precision therapeutic intervention against renal fibrosis.

// Source

View paper (DOI)Open access versionOpenAlexJournal of Translational MedicinePublished 2026-09-18

Authors: Xin Cao, Yuxi Xiao, Yiwei Wang, Hanwei Huang, Weimiao Kong, Zefan Chen, Yidong Zheng, Jialin Li, Enhui Li, Yizi Gong, Qiongjing Yuan, Huipeng Ge, Xiangcheng Xiao

Institutions: Central South University, Third Xiangya Hospital, Nanfang Hospital, Southern Medical University, Shenzhen University, Guangdong Academy of Medical Sciences, Hunan University, Guangdong Provincial People's Hospital, Xiangya Hospital Central South University