Transcriptomic and Proteomic of Gastrocnemius Muscle in Peripheral Artery Disease

Abstract

Background: Few effective therapies exist to improve lower extremity muscle pathology and mobility loss due to peripheral artery disease (PAD), in part because mechanisms associated with functional impairment remain unclear.

Methods: To better understand mechanisms of muscle impairment in PAD, we performed in-depth transcriptomic and proteomic analyses on gastrocnemius muscle biopsies from 31 PAD participants (mean age, 69.9 years) and 29 age- and sex-matched non-PAD controls (mean age, 70.0 years) free of diabetes or limb-threatening ischemia.

Results: Transcriptomic and proteomic analyses suggested activation of hypoxia-compensatory mechanisms in PAD muscle, including inflammation, fibrosis, apoptosis, angiogenesis, unfolded protein response, and nerve and muscle repair. Stoichiometric proportions of mitochondrial respiratory proteins were aberrant in PAD compared to non-PAD, suggesting that respiratory proteins not in complete functional units are not removed by mitophagy, likely contributing to abnormal mitochondrial activity. Supporting this hypothesis, greater mitochondrial respiratory protein abundance was significantly associated with greater complex II and complex IV respiratory activity in non-PAD but not in PAD. Rate-limiting glycolytic enzymes, such as hexokinase and pyruvate kinase, were less abundant in muscle of people with PAD compared with non-PAD participants, suggesting diminished glucose metabolism.

Conclusions: In PAD muscle, hypoxia induces accumulation of mitochondria respiratory proteins, reduced activity of rate-limiting glycolytic enzymes, and an enhanced integrated stress response that modulates protein translation. These mechanisms may serve as targets for disease modification.

Keywords: biomarkers; mitochondria; mitophagy; proteomics; unfolded protein response.

Figure 1.. Analysis of transcripts.

A. Volcano plot of differentially expressed protein-coding genes according to effect size and level of significance. B. Volcano plot of differentially expressed non-protein-coding genes according to effect size and level of significance. C. Boxplots of MIR210HG in PAD and non-PAD (Wilcoxon test p<0.05). D. Select pathways enriched in PAD or non-PAD (q-value <0.05) (full list in Supplementary Data 2). Gene set collections used were: Hallmark (HM), Reactome (RE), KEGG (KG), WikiPathways (WP), Pathway Interaction Database (PID), and BioCarta (BC).

Figure 2.. Analysis of proteins.

A. Volcano plot showing proteins over- and under-represented in PAD compared to non-PAD according to effect size and level of significance. B. Heatmap of 96 mitochondria respiratory proteins in PAD and non-PAD. Participants are ordered according to average protein level in ascending (non-PAD) and descending (PAD) order. C. Barchart showing ribosomal-related proteins over- and under-represented in PAD compared to non-PAD. D. Selected pathways enriched in PAD or non-PAD (q-value <0.05) (full list in Supplementary Data 4). Gene set collections used were: Hallmark (HM), Reactome (RE), KEGG (KG), WikiPathways (WP), Pathway Interaction Database (PID), and BioCarta (BC).

Figure 3.

Co-expression analysis of 96 mitochondrial respiratory proteins in PAD and non-PAD subjects. A-B. Undirected, weighted networks for each group, where nodes represent proteins (colored by protein complex) and links represent protein-protein correlations (whose widths increase monotonically with correlation). Different node shapes indicate whether proteins are encoded in nuclear or mitochondrial DNA. C-D. Hierarchically clustered heatmap of Pearson’s correlation matrix for electron transport chain (ETC) proteins in each cohort. E. Network closeness centrality measurements comparing PAD and non-PAD for each ETC complex. F. Pearson’s correlation measurements comparing PAD and non-PAD for each ETC complex.

Figure 4.

A. Scatterplots of correlations between the activity of Complex II and Complex IV in non-PAD and PAD participants. Shaded areas around the regression line indicate 95% confidence intervals. B. Scatterplot of the Mitochondria Protein Score, Complex II Protein Score and Complex IV Protein Score with Complex II and Complex IV activity assessed by respirometry. Rho values are Spearman’s correlation coefficients and p-values are from robust linear regression.

Figure 5.

A. Scatterplot of mRNA-protein correlations in PAD and non-PAD limited to the 5851 genes for which both mRNA and protein expression was available. Colored symbols indicate: 1) significant correlations (p-value<0.05) between mRNAs and proteins either in PAD patients or non-PAD controls, and 2) correlation coefficients in PAD outside the 95% CI defined for correlations in non-PAD samples. Other mRNA-protein correlation pairs are indicated by grey empty circles. Because of missing data, some pairs have high but still non-significant correlation values in both PAD patients and non-PAD controls. The filled circles on the upper-left quadrant represent transcript-protein pairs that have higher correlation in PAD than in controls. Enriched pathways buy GSEA for off-diagonal genes are shown on the right, namely “gamma-tubulin complex” (green circles), “extracellular vesicles” (blue circles and panel B for individual genes in the pathway), mitochondrion (red circles and panel C), and tRNA amino-acylation (cyan symbols and panel D).