Multiomic Analysis of Calf Muscle in Peripheral Artery Disease and Chronic Kidney Disease

Abstract

Background: Chronic kidney disease (CKD) has emerged as a significant risk factor that accelerates atherosclerosis, decreases muscle function, and increases the risk of amputation or death in patients with peripheral artery disease (PAD). However, the modulators underlying this exacerbated pathobiology are ill-defined. Recent work has demonstrated that uremic toxins are associated with limb amputation in PAD and have pathological effects in both the limb muscle and vasculature. Herein, we use multiomics to identify novel modulators of disease pathobiology in patients with PAD and CKD.

Methods: A cross-sectional study enrolled 4 groups of participants: controls without PAD or CKD (n=28), patients with PAD only (n=46), patients with CKD only (n=31), and patients with both PAD and CKD (n=18). Both targeted (uremic toxins) and nontargeted metabolomics in plasma were performed using mass spectrometry. Calf muscle biopsies were used to measure histopathology, perform bulk and single-nucleus RNA sequencing, and assess mitochondrial function. Differential gene and metabolite analyses, as well as pathway and gene set enrichment analyses, were performed.

Results: Patients with both PAD and CKD exhibited significantly lower calf muscle strength and smaller muscle fiber areas compared with controls and those with only PAD. Compared with controls, mitochondrial function was impaired in patients with CKD, with or without PAD, but not in PAD patients without CKD. Plasma metabolomics revealed substantial alterations in the metabolome of patients with CKD, with significant correlations observed between uremic toxins (eg, kynurenine and indoxyl sulfate) and both muscle strength and mitochondrial function. RNA sequencing analyses identified downregulation of mitochondrial genes and pathways associated with protein translation in patients with both PAD and CKD. Single-nucleus RNA sequencing further highlighted a mitochondrial deficiency in muscle fibers along with unique remodeling of fibro-adipogenic progenitor cells in patients with both PAD and CKD, with an increase in adipogenic cell populations.

Conclusions: CKD significantly exacerbates ischemic muscle pathology in PAD, as evidenced by diminished muscle strength, reduced mitochondrial function, and altered transcriptome profiles. The correlation between uremic toxins and muscle dysfunction suggests that targeting these metabolites may offer therapeutic potential for improving muscle health in PAD patients with CKD.

Keywords: mitochondria; muscles; peripheral arterial disease; renal insufficiency, chronic; uremia.

Figure 1.

Calf muscle strength and muscle fiber area is lowest in patients with peripheral artery disease (PAD) and chronic kidney disease (CKD). A, Calf muscle strength in absolute and normalized to body weight (n=102). B, Representative immunofluorescence images of gastrocnemius muscle biopsies with fiber types and sizes labeled. C, Violin plots showing the cross-sectional areas of type I (n=8816 myofibers) and type II (n=8395) myofibers in each patient group (n=63 patients). D, Quantification of the percentage of type I fibers (n=63 patients). E, Pearson correlation between calf muscle strength and mean myofiber area (both type I and II myofibers; n=43 patients). A was analyzed using a Kruskal-Wallis test with Dunn test for pairwise comparisons. C was analyzed using a linear mixed effects model. D was analyzed using a 1-way ANOVA with Tukey post hoc testing. Error bars represent the SD. MVIC indicates maximal voluntary isometric contraction.

Figure 2.

Skeletal muscle mitochondrial function is significantly impaired in patients with chronic kidney disease (CKD) and peripheral artery disease (PAD)+CKD. A, Graphical depiction of mitochondrial analyses using permeabilized myofibers. B, Citrate synthase activity in gastrocnemius muscle (n=81 patients). C, Relationship between oxygen consumption (JO₂) or JO₂ normalized to citrate synthase activity and energy demand (ΔG_ATP) when mitochondria were fueled with pyruvate, malate, and octanoylcarnitine. D, Quantification of the conductance (the slope of JO₂ and ΔG_ATP relationship; n=86 patients) using both unnormalized and normalized to citrate synthase activity JO₂ data. E, Relationship between mitochondrial hydrogen peroxide production (JH₂O₂) and energy demand (ΔG_ATP) when mitochondria were fueled with pyruvate, malate, and octanoylcarnitine. F, Quantification of JH₂O₂ under state 2 (no energy demand) conditions (n=82 patients). G, Quantification of JH₂O₂ with mitochondrial fueled with succinate followed by inhibition of the matrix antioxidant systems (with auranofin/N,N'-Bis(2-chloroethyl)-N-nitrosourea, bis-chloroethylnitrosourea; n=85 patients). Analyzed via 2-way ANOVA with Tukey post hoc. H, Representative images and quantification of succinate dehydrogenase (SDH) activity in muscle sections (n=53 patients). B, D, F, G, and H were analyzed with a 1-way ANOVA with Tukey post hoc. Error bars represent the SD. OXPHOS indicates oxidative phosphorylation.

Figure 3.

Chronic kidney disease (CKD) profoundly impacts the plasma metabolome independent of peripheral artery disease (PAD). A, Targeted plasma metabolomic quantification of uremic toxins (n=97 patients). Analysis done using 1-way ANOVA with Tukey post hoc. B, Graphic of experimental design and overall metabolite detection for untargeted (global) metabolomics in patient plasma. C, Principal component analysis demonstrates separation between CKD and non-CKD patients, independent of PAD. D, A heat map with hierarchical clustering of metabolites and patients. E, Venn diagram showing significant metabolite differences across group comparisons. Error bars represent the SD.

Figure 4.

Uremic toxin levels have an inverse relationship with calf muscle mitochondrial function and strength. A Pearson correlations between oxidative phosphorylation (OXPHOS) conductance and L-kynurenine, L-Kyn/Trp ratio, and indoxyl sulfate levels (n=81 patients). B, Pearson correlations between calf muscle strength and L-kynurenine, L-Kyn/Trp ratio, and indoxyl sulfate levels (n=72 patients). Statistical analyses performed using 2-tailed Pearson correlation.

Figure 5.

Whole muscle and single-nucleus RNA sequencing identify mitochondrial deficiency and cytoplasmic translation defects in patients with peripheral artery disease and chronic kidney disease (PAD+CKD). A, RNA sequencing on total RNA from the gastrocnemius was performed (n=88 patients). Partial least squares discriminant analysis (PLS-DA) revealed clear separation between PAD and PAD+CKD patients. B, Volcano plot of gene expression shows differentially expressed genes in PAD and PAD+CKD patients. C, Gene set enrichment analysis in significantly upregulated and downregulated genes between PAD and PAD+CKD patients. D, Single-nucleus RNA sequencing was performed on gastrocnemius muscle specimens from 20 PAD and 12 PAD+CKD patients. Uniform Manifold Approximation and Projections (UMAP) presented by group and by cell types are shown. E, Percentage of nuclei within each cluster determined by group. F, Dot plots of the top 4 marker genes for each cell type. G, UMAPs of the subclustering of myofiber nuclei by group and type. H, Venn diagrams shown differentially expressed genes by myonuclei type. I, UMAPs showing the mitochondrial gene module score for each nuclei demonstrated the mitochondrial deficiency in PAD+CKD muscles. J, Gene set enrichment analysis results for myonuclei populations. The x axis is the normalized enrichment score and the values for each bar are the adjusted P values. DEG indicates differentially expressed gene; EC, endothelial cell; FAP, fibro-adipogenic progenitor cells; FDR, false discovery rate; MF, myofiber; NADH, nicotinamide adenine dinucleotide; NES, normalized enrichment score; and NMJ, neuromuscular junction.

Figure 6.

Fibro-adipogenic progenitor cells (FAPs) are uniquely remodeled in muscles from patients with peripheral artery disease and (PAD+CKD). A, Uniform Manifold Approximation and Projections (UMAPs) of the subclustering of FAPs presented by group and by subpopulation type are shown. B, Donut plots showing the abundance of subpopulations of FAPs in both groups. C, Predicted differential potentials using Palantir. D, Trajectory inference in FAPs showed 3 distinct fates in PAD+CKD. E, Feature plots showing normalized gene expression levels for inflammatory, profibrotic, and adipogenic genes. F, Gene expression changes across those 3 fates across pseudotime for both groups.

Figure 7.

CellChat analysis predicts changes in intercellular communication of peripheral artery disease and chronic kidney disease (PAD+CKD) patients. A, Circle plots showing the overall intercellular communication occurring in PAD and PAD+CKD. Circle sizes represent the number of cells and edge width represents communication probability. B, Comparison of outgoing and incoming interaction strengths for all cells type in PAD and PAD+CKD. C, Dot plot showing cell-specific ligand-receptor interaction strengths for incoming and outgoing signals. D, Ranked significant ligand-receptor communications for relative information flow between PAD and PAD+CKD muscles. EC indicates endothelial cell; FAP, fibro-adipogenic progenitor cells; FMuSC, muscle stem cell; MF, myofiber; NMJ, neuromuscular junction; and SMC, smooth muscle cell.