Transcriptomic identification of mitophagy-related plaque markers HAND2, NRP1 and RNF213, with external replication and in vivo validation of MLXIP

Abstract

Atherosclerosis is a leading cause of cardiovascular mortality, and mitophagy dysfunction contributes to its pathogenesis. However, the mitophagy-related molecular signatures that could serve as plaque-associated markers remain incompletely characterized. Paired differential expression analysis of GSE43292 (32 plaque vs. 32 paired macroscopically intact carotid samples), combined with weighted gene co-expression network analysis, defined a plaque-associated module that was intersected with 4,922 mitophagy-associated genes (GeneCards) to identify mitophagy-related DEGs. LASSO, Random Forest and XGBoost were applied as an auxiliary, patient-grouped, cross-validated feature-selection step; a logistic-regression model fitted on the shared genes was evaluated by receiver operating characteristic (ROC) analysis with 95% confidence intervals (CIs). External validation used GSE100927 (69 plaque vs. 35 control peripheral artery samples). Immune infiltration was assessed by CIBERSORT, and MLXIP protein abundance was explored by Western blot in a small ApoE ⁻/⁻ murine model. Comparator specimens were macroscopically intact tissue (Stary types I–II). Paired analysis of GSE43292 identified 1,249 differentially expressed genes, of which 107 were mitophagy-related. Patient-grouped machine-learning feature selection nominated three consensus markers— HAND2 , NRP1 and RNF213 —and a three-gene logistic-regression model reached an out-of-fold area under the curve (AUC) of 0.892 (95% CI 0.810–0.973); because selection and evaluation used the same cohort, these results are exploratory. All three were directionally reproduced in the independent external cohort GSE100927 (direction-locked AUC 0.814, 0.942 and 0.848 respectively). Of three genes examined a priori , MLXIP and EPB41L4A-AS1 replicated in the external cohort (AUC 0.834 for both), whereas RNU4ATAC did not (AUC 0.321). CIBERSORT showed higher macrophage and memory B-cell fractions and lower regulatory T-cell fractions in plaques. In an exploratory in vivo experiment ( n  = 3 per group), aortic MLXIP protein was 37.7-fold higher in atherosclerotic mice than in controls ( P  = 0.11) with unchanged MLXIP mRNA; these data were directional only. In this exploratory study, bioinformatics and machine learning nominated HAND2 , NRP1 and RNF213 as candidate mitophagy-related plaque markers, with MLXIP and EPB41L4A-AS1 replicated in an independent cohort, and pointed to a mitophagy–immune axis in plaque pathogenesis. These findings are hypothesis-generating and require confirmation in larger cohorts.