Cocoa flavanols, Nrf2 activation, and oxidative stress in peripheral artery disease: mechanistic findings in muscle based on outcomes from a randomized trial

Abstract

The pathophysiology of muscle damage in peripheral artery disease (PAD) includes increased oxidant production and impaired antioxidant defenses. Epicatechin (EPI), a naturally occurring flavanol, has antioxidant properties that may mediate the beneficial effects of natural products such as cocoa. In a phase II randomized trial, a cocoa-flavanol-rich beverage significantly improved walking performance compared with a placebo in people with PAD. In the present work, the molecular mechanisms underlying the therapeutic effect of cocoa flavanols were investigated by analyzing baseline and follow-up muscle biopsies from participants. Increases in nuclear factor erythroid 2-related factor 2 (Nrf2) target antioxidants heme oxygenase-1 (HO-1) and NAD(P)H dehydrogenase [quinone] 1 (NQO1) in the cocoa group were significantly associated with reduced accumulation of central nuclei, a myopathy indicator, in type II muscle fibers (P = 0.017 and P = 0.023, respectively). Protein levels of the mitochondrial respiratory complex III subunit, cytochrome b-c1 complex subunit 2 (UQCRC2), were significantly higher in the cocoa group than in the placebo group (P = 0.032), and increases in UQCRC2 were significantly associated with increased levels of Nrf2 target antioxidants HO-1 and NQO1 (P = 0.001 and P = 0.035, respectively). Exposure of non-PAD human myotubes to ex vivo serum from patients with PAD reduced Nrf2 phosphorylation, an indicator of activation, increased hydrogen peroxide production and oxidative stress, and reduced mitochondrial respiration. Treatment of myotubes with EPI in the presence of serum from patients with PAD increased Nrf2 phosphorylation and protected against PAD serum-induced oxidative stress and mitochondrial dysfunction. Overall, these findings suggest that cocoa flavanols may enhance antioxidant capacity in PAD via Nrf2 activation.

NEW & NOTEWORTHY The current study supports the hypothesis that in people with PAD, cocoa flavanols activate Nrf2, thereby increasing antioxidant protein levels, protecting against skeletal muscle damage, and increasing mitochondrial protein abundance. These results suggest that Nrf2 activation may be an important therapeutic target for improving walking performance in people with PAD.

INTRODUCTION

Peripheral artery disease (PAD) is defined by the narrowing of the peripheral arteries to the lower extremities due to the build-up of atherosclerotic plaque (1). PAD affects over 230 million individuals globally and is a major cause of morbidity and mortality (2). PAD is also associated with exercise limitations and accelerated functional decline over time (3). Despite the major health burden of PAD, the disease is understudied relative to other vascular diseases, and only a limited number of therapies have been found to enhance walking impairment in individuals with PAD (2).

Oxidative stress from repeated episodes of lower extremity ischemia/reperfusion during walking activity has been studied as a potential causal pathway for the pathophysiology of PAD (4). Preclinical studies suggest that the mitochondrial electron transport chain (ETC) may contribute to superoxide and hydrogen peroxide during reoxygenation of ischemic tissue in PAD (4–8). In patients with PAD, circulating levels of oxidative stress markers are higher than in non-PAD controls at rest and following exercise (9, 10). Increased reactive oxygen species (ROS) and oxidative stress markers have been demonstrated in gastrocnemius muscle of patients with PAD compared with non-PAD control muscle (11, 12).

Impaired antioxidant defenses may also increase oxidative stress in PAD. Lower concentrations of specific, scavenging antioxidants such as vitamin C and vitamin E that are consumed during oxidative stress have been reported in PAD (13–17). Furthermore, the activity of several major antioxidant enzymes, including manganese superoxide dismutase (SOD), the first line of defense against mitochondrial ROS, and glutathione peroxidase (GPX), is reduced in circulation and skeletal muscle of patients with PAD compared with controls (18–23). SOD, GPX, and other factors associated with cellular antioxidant defense mechanisms contain antioxidant response elements (AREs) in their promoter regions. One of the principal regulators of AREs is the transcription factor NF-E2-related factor 2 (NFE2L2, also known as Nrf2), a redox-sensitive “master regulator” of the antioxidant response (24). Evidence suggests that Nrf2 activation protects cerebral, myocardial, hepatic, and kidney tissue from adverse effects of ischemia (25–31). However, while impaired Nrf2 activity has been reported in PAD (32), little is known about the potential therapeutic effects of Nrf2 activation.

Several natural products, such as flavanols, activate Nrf2 (33, 34). For example, epicatechin (EPI), the main flavanol present in cocoa (35), activated the Nrf2 pathway in preclinical models (36–42). In a phase II randomized clinical trial of 44 people with PAD randomized to cocoa flavanol (15 g of cocoa and 75 mg EPI) or placebo beverage (the COCOA-PAD Trial), 6 mo of daily cocoa flavanols improved 6-min walk distance by 42.6 m, increased gastrocnemius cytochrome c oxidase (COX) enzyme activity, and lowered abundance of central nuclei in calf muscle biopsies, compared with placebo (43). In the current study, we aimed to investigate the mechanism behind the cocoa flavanol-mediated therapeutic effect. We hypothesized that cocoa flavanols increase skeletal muscle Nrf2 activity to improve skeletal muscle morphology and mitochondrial protein abundance. Baseline and follow-up gastrocnemius muscle samples from participants in the COCOA-PAD trial were analyzed to measure fiber type-specific central nuclei and cross-sectional area (CSA), as well as the protein levels of the Nrf2 target antioxidants, heme oxygenase-1 (HO1), and NAD(P)H dehydrogenase [quinone] 1 (NQO1), and oxidative phosphorylation (OXPHOS) protein subunits. Myotubes differentiated from primary myogenic progenitor cells (MPCs) isolated from muscle from healthy donors were treated with ex vivo serum obtained from patients with PAD in the COCOA PAD trial. Using this in vitro model of PAD, we tested the effect of EPI treatment on Nrf2 activation, mitochondrial function, and oxidative stress in myotubes.

MATERIALS AND METHODS

Human Tissues and Ethics Statement

Whole gastrocnemius tissue homogenates, fresh-frozen gastrocnemius tissues, and serum samples were obtained from 16 participants with PAD who had muscle biopsies at baseline and 6-mo follow-up in the COCOA-PAD randomized clinical trial (NCT02876887). Recruitment methods and details can be found in the primary outcomes report from the trial (43).

Healthy participant serum samples (free of diabetes and cardiovascular disease) were obtained from the previously completed MASTERS trial (44) (NCT02308228). These samples were used as control serum samples in cell culture experiments.

Human samples were collected with written informed consent from each participant and the approval of the respective Institutional Review Boards. Participants’ written consent forms and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki.

Immunohistochemistry

Immunohistochemistry (IHC) measures of central nuclei, satellite cells, embryonic myosin heavy chain expression, and capillary density were analyzed; methods and results were previously published (43). Briefly, 7-μm sections were taken from frozen muscle at −22 to −25°C and placed onto SuperFrost Plus slides (Thermo Fisher Scientific, Waltham, MA #12–550-15). All slides were subjected to downstream assays after drying for 1 h.

Hematoxylin and eosin (H&E) histochemistry was performed and used to assess the quality of each biopsy, and only samples lacking freeze artifacts were used for downstream analyses. Standard protocols were used for H&E (45). Distribution of muscle fiber types was assessed using primary antibodies against myosin heavy chain (MyHC) isoforms according to our established protocol (46). Briefly, following sectioning and drying, slides were incubated overnight in a primary antibody cocktail (all MyHC antibodies obtained from Developmental Studies Hybridoma Bank, DSHB, Iowa City, IA): Mouse IgG2b BA.D5 concentrate (Type I, #BA.D5-c, RRID: AB_2235587, 1:100), Mouse IgG1 SC.71 supernatant (Type IIa, #SC.$11-s, RRID: AB_2147165, 1:50), Mouse IgM 6H.1 supernatant (Type IIx, #6H.1-s, RRID: AB_1157897, neat), and rabbit anti-laminin (fiber borders) (MilliporeSigma, Burlington, MA, # L9393, RRID: AB_477163, 1:200). Following washing with phosphate-buffered saline (PBS, 10 mM, pH 7.4), slides were incubated with secondary antibodies: anti-mouse [IgG2b AlexaFluor 647 (Thermo Fisher Scientific, #A21242, RRID: AB_2535811, 1:250), IgG1 AlexaFluor 488 (Thermo Fisher Scientific, #A21121, RRID: AB_2535764, 1:250), and IgM AlexaFluor 555 (Thermo Fisher Scientific, #A21426, RRID: AB_2535847, 1:250)] and anti-rabbit biotin (Jackson ImmunoResearch Labs, West Grove, PA, #111-065-144, RRID: AB_2337965, 1:500). Streptavidin-AMCA (Vector Laboratories, Newark, CA, # CI-1000, RRID: AB_2336195, 1:100) was used to visualize laminin, and coverslips were applied with a 1:1 mixture of PBS and glycerol (VWR #BT142040).

For analysis of central nuclei by fiber type, slides were incubated with primary antibodies targeting dystrophin (Abcam, Cambridge, UK, #15277, RRID: AB_301813, 1:200) to detect fiber borders and Type I MyHC (same as above). DAPI (4′6′-diamidino-2-phenylindole) (Thermo Fisher Scientific, #D1306, 1:10,000) was used to visualize nuclei, and central nuclei were defined as located within, but not touching, dystrophin borders. Fiber type-specific central nuclei were quantified manually by experienced researchers. Sections were imaged using an Olympus BX61 VS microscope (Olympus Corporation, Shinjuku City, Tokyo, Japan), and stitched images of entire muscle biopsies were acquired with a ×20 objective. On average, ∼500 fibers were counted per sample (range: 74–1,902 fibers). Fiber type-specific data for two participants were not obtained due to no remaining sample. Fiber-type distribution and cross-sectional area (CSA) were processed and analyzed with MyoVision analytical software (47).

Western Blot Analysis

For Western blotting, proteins prepared in Laemmli sample buffer (Bio-Rad Laboratories, Hercules, CA, # 1610737), were boiled, and proteins (20 μg) were separated by SDS-PAGE using 4–20% Criterion TGX Stain-Free Protein Gels (Bio-Rad #5678094). Total protein stains were activated by brief ultraviolet (UV) excitation using a ChemiDoc MP system (Bio-Rad), then proteins were transferred for 1 h 30 min at 200 mA onto a PVDF membrane (Bio-Rad #1620263) in 20% methanol Tris-glycine buffer at 4°C. Membranes were subsequently blocked with SuperBlock Blocking Buffer (Thermo Fisher Scientific, #37515) for 1 h at room temperature, followed by overnight incubation in primary antibody at 4°C.

The antibodies (all obtained from Thermo Fisher Scientific) and their respective concentrations used were as follows: anti-Heme Oxygenase-1 (HO-1) (#MA1-112, RRID: AB_2536823, 1:1,000), anti-NAD(P)H dehydrogenase [quinone] 1 (NQO1) (#39–3700, RRID: AB_2533410, 1:1,000), anti-OxPhos Human Antibody Cocktail (#45–8199, RRID: AB_2533836, 1:200), anti-Phospho-Nrf2 (Ser40) (#PA5-67520, RRID: AB_2691678, 1:3,000), anti-Nrf2 (#PA5-27882, RRID: AB_2545358, 1:500), anti-Kelch-like ECH-associated protein 1 (Keap1) (#10503-2-AP, RRID: AB_2132625, 1:2,000), and anti-beta actin (#AM4302, RRID: AB_2536382, 1:1,000). All antibodies met the Thermo Fisher Scientific advanced verification, verified by cell treatment to ensure that the antibody binds to the antigen stated. Membranes were washed in Tris-buffered saline-Tween (TBS-T, 0.1% Tween-20) before incubation in goat anti-mouse IgG, horseradish peroxidase (HRP)-conjugated secondary antibody (#31430, 1:10,000) or goat anti-rabbit IgG, HRP-conjugated secondary antibody (#31460, 1:10,000) for 1 h at room temperature. Blots were developed with enhanced chemiluminescence (Clarity Western ECL Substrate, Bio-Rad #1705061), imaged using a ChemiDoc MP system, and quantified using Image Lab Software (Bio-Rad).

Cell Culture

Primary human MPCs were obtained from the University of Kentucky Center for Muscle Biology Healthy Muscle Bank. These cells were isolated from human vastus lateralis muscle biopsies by magnetic antibody cell sorting (MACS) using anti-CD56 microbeads and an Automacs Pro cell sorter (Miltenyi Biotec, Auburn, CA), as previously described (48). MPCs used in this study were from n = 3 biological replicates (2 males, 1 female) with a mean age of 59 ± 3 yr. MPCs were expanded on 10% Matrigel-coated (Corning Inc., Corning, NY, #354234) Primaria culture plates (Corning #353846) in growth media consisting of Hams F-10 (Corning #10–070-CV), 20% fetal bovine serum (FBS) (Corning #35–070-CV), 1% penicillin-streptomycin (VWR #97063-708), and 10 ng/mL basic fibroblast growth factor (bFGF) (Corning #354060). MPCs were split by mild trypsinization (VWR #L0154-0100) at around ∼40% confluence until the time of differentiation.

For differentiation to myotubes, MPCs were allowed to reach ∼85–95% confluency before switching from growth medium to differentiation medium, which consisted of DMEM (ATCC, Manassas, VA #30–2006) supplemented with 2% horse serum (HS) (Corning #35-030-CV). MPCs were allowed to differentiate for 5 days before switching the 2% HS to either 10% pooled PAD serum (n = 10) or 10% pooled healthy participant serum (n = 10) for 48 h, as previously described (49). For EPI-treated myotubes, differentiation media or serum-treated media were supplemented with (-)-Epicatechin (Indofine Chemical Company, Hillsborough Township, NJ, #020977S (1–10 μM final concentration). Sulforaphane (SFN) at a 10 μM concentration (Enzo Life Sciences, Farmingdale, NY, # 89149-298) was used as a positive control for Nrf2 activation. A subset of myotubes exposed to 10% pooled PAD serum was also treated with the Nrf2 antagonist ML385 (50, 51) (MilliporeSigma, #SML1833) along with EPI to selectively inhibit Nrf2 at a final concentration of 10 μM.

Measurement of Oxygen Consumption Rate

For oxygen consumption rate (OCR) measurement, MPCs were cultured and differentiated into myotubes in Matrigel-coated Seahorse XF24 cell culture microplates (Agilent Technologies, Santa Clara, CA, #100777-004). Before OCR analysis using a Seahorse XF24 Analyzer (Agilent), media was replaced with Seahorse XF DMEM assay media supplemented with glucose (10 mM), pyruvate (1 mM), and glutamine solution (2 mM) (Agilent #103680-100). A Seahorse XF Cell Mito Stress Test Kit (Agilent #103015-100) was used to load the following compounds for injection into the wells: oligomycin (Oligo, for ATP synthase inhibition and measurement of ATP-linked respiration, 1.5 μM), carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP, for uncoupling to measure maximal OCR, 1 μM), and antimycin A/rotenone (AmA/ROT, inhibitors of Complex III and I, to determine nonmitochondrial OCR, 0.5 μM). Immediately after OCR measurement, the protein concentration was determined using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific #23252). All data were generated from four technical replicates (averaged) for each biological replicate and condition.

Measurement of Hydrogen Peroxide Production

Cellular hydrogen peroxide production was measured using an Amplex UltraRed (AmR) Assay Kit (Thermo Fisher #A36006). Myotubes were incubated for 1 h in reaction buffer consisting of 50 μM AmR and 0.1 U/mL horseradish peroxidase (HRP) in Krebs-Ringer phosphate (KRPG), followed by fluorescence measurement (excitation 560 nm, emission 590 nm) with a microplate reader (BioTek Cytation 5 Multimode Reader, BioTek, Winooski, VT). Each assay included a positive control (10 μM hydrogen peroxide in reaction buffer) and a negative control (reaction buffer without the addition of HRP). AmR fluorescence was corrected for background fluorescence and normalized to corresponding protein concentrations, determined using a bicinchoninic acid (BCA) assay. Four technical replicates were averaged for each condition.

Citrate Synthase Activity Assay

Mitochondrial content was estimated by measuring the activity of citrate synthase (CS). The assay involves the CS-catalyzed reaction between the thiol group of acetyl coenzyme A (CoA) and 5,5′-Dithiobis-(2-nitrobenzoic acid) (DTNB), resulting in the production of 5-thio-2-nitrobenzoic acid (TNB) in the presence of oxaloacetic acid (OAA). To perform the assay, myotubes lysed in RIPA lysis buffer (VWR #786-489) were mixed in a buffer containing cell lysate, 100 mM Tris (pH 8.0), 10 mM DTNB, 30 mM acetyl CoA, and 10 mM OAA. CS activity was measured by monitoring the change in absorbance at 412 nm, indicating the formation of TNB, using a Cytation 5 Multimode reader. All samples were analyzed in quadruplicate, and CS activity was normalized to protein concentrations, which were determined using a BCA protein assay.

Glutathione Peroxidase Activity Assay

A GPX Assay Kit (Abcam #ab102530) was used as a marker of antioxidant activity. Specifically, this assay quantifies the activity of glutathione-dependent peroxidases. The assay involved the colorimetric detection of nicotinamide adenine dinucleotide phosphate (NADPH) consumption during the reduction of oxidized glutathione by glutathione reductase. GPX activity was assessed by measurement of the optical density of wells at 340 nm using a Cytation 5 Multimode reader. Four technical replicates were averaged for each condition, and GPX activity was normalized to protein concentrations.

Immunocytochemistry

For immunostaining of myotubes, cells were fixed in 4% paraformaldehyde (PFA) (VWR #200058-220) before gentle washing in PBS and incubation in 100 mM glycine for 10 min at room temperature. Cells were subsequently blocked for 1 h at room temperature in SuperBlock, followed by overnight incubation in anti-4-hydroxynonal (4-HNE) primary antibody (Abcam #ab46545, 1:200) at 4°C. Cells were gently washed in PBS-0.1% Tween-20 (PBST) and then incubated in goat anti-rabbit IgG AlexaFluor 488 (Thermo Fisher #A11034, 1:250) and DAPI (1:1,000) for 1 h at room temperature. Images were obtained with an inverted fluorescent microscope (Axio Observer 7, Carl Zeiss AG, Oberkochen, Baden-Württemberg, Germany) with a ×10 objective. Four technical replicates were averaged for each condition, and fluorescence intensity was quantified using ImageJ (US National Institutes of Health, Bethesda, Maryland) and expressed as arbitrary units (a.u.).

Statistical Analyses

For the IHC measures included in this manuscript, muscle characteristics (CSA, fiber-type frequency, and central nuclei by fiber type) at baseline and follow-up were compared with placebo and cocoa intervention groups by independent t tests. For Western blotting of skeletal muscle homogenate proteins, baseline and follow-up samples were consistently run on the same gel. The relative difference in signal intensity (normalized to total protein) in follow-up versus baseline samples was compared between placebo and cocoa intervention groups by independent t tests. For correlation analyses, a Pearson correlation was calculated using the change in Nrf2 targets as independent variables and the change in other muscle measures as dependent variables.

For in vitro assays, the difference between groups (3 or more) were compared by a one-way ANOVA using Tukey’s multiple comparisons test for post hoc analyses. All statistical analyses were performed in GraphPad Prism (Ver. 9.4), and significance for all statistical tests was set at P < 0.05.

A list of all abbreviations used throughout the manuscript is available in Supplemental Table S1.

RESULTS

Skeletal muscle Nrf2 activation by cocoa is associated with blunted accumulation of central nuclei within type II fibers.

Sixteen participants with PAD had muscle biopsy specimens at baseline and 6-mo follow-up, including 10 randomized to cocoa flavanols and six randomized to placebo. Overall, among participants with a muscle biopsy, the mean age was 73.70 ± 8.39 yr, mean ankle brachial index was 0.65 ± 0.13, 5 (31%) were females and 12 (75%) were Black. Participant demographics are provided in Table 1.

Table 1.

Patient characteristics of COCOA-PAD trial participants

	Group
	Cocoa Intervention (n = 10)	Placebo (n = 6)
Age, yr; means (SD)	70.13 (7.73)	77.31 (6.70)
Male, n (%)	7 (70)	4 (66.67)
Black, n (%)	8 (80)	4 (66.67)
ABI, means (SD)	0.67 (0.14)	0.61 (0.12)
BMI, kg/m2; means (SD)	25.82 (4.09)	28.50 (3.30)

Table 1 shows the demographics and characteristics of the participants from the COCOA-PAD trial for which muscle biopsies and serum were analyzed in the current study. Corresponding serum samples from the cocoa intervention group were used for in vitro treatment experiments. ABI, ankle-brachial index; BMI, body mass index; PAD, peripheral artery disease.

Previous studies in mice showed that epicatechin (EPI) increases skeletal muscle fiber size during development and regeneration (52, 53). EPI modulates fiber type and attenuates the loss in fiber size in different preclinical models of muscle atrophy (54–56). Thus, we sought to analyze fiber type-specific CSA and fiber-type distribution in calf muscle biopsies from the COCOA-PAD trial. Compared with placebo, cocoa did not significantly change the CSA of type I, type IIa, or type IIa/x fibers (Supplemental Fig. S1, A–C). Similarly, cocoa flavonols did not significantly change the frequency of type I, type IIa, or type IIa/x fibers compared with placebo (Supplemental Fig. S1, D–F). These data suggest that 6-mo treatment with EPI-rich cocoa did not affect fiber size or fiber-type distribution in PAD skeletal muscles. In our previous study, we found that central nuclei accumulated within muscle fibers over the 6-mo trial period only within the placebo group, suggesting cocoa prevented this accumulation (43). Central nuclei have been identified as a sign of myopathy in human skeletal muscles, and their presence has been linked to muscular disorders and muscle degeneration (57, 58). To further investigate whether accumulation of central nuclei in PAD was fiber type specific, we assessed the abundance of central nuclei in type I and type II fibers separately. Cocoa did not significantly change the percentage of type I fibers with central nuclei compared with placebo (Supplemental Fig. S1G). However, compared with placebo, cocoa significantly reduced the proportion of type II fibers with central nuclei (P = 0.0096) (Fig. 1, A and B).

Figure 1.

Six-month change in type II fiber central nuclei and Nrf2 targets in muscle biopsies from COCOA-PAD trial. A: representative images of dystrophin-demarcated (white) muscle fibers in gastrocnemius biopsies. Type I fibers shown in green. White arrows point to nuclei (DAPI, blue) that are centrally located. Scale bar = 100 μm. B: change in the proportion of type II fibers (%) with central nuclei in placebo (PLA) and cocoa intervention (COCOA) groups from baseline to follow-up. Difference between groups tested using an independent t test, n = 6 participants in PLA, n = 9 participants in COCOA, bar graphs indicate means ± SD. C: the data in B represented as the proportion of type II fibers (%) with central nuclei at baseline (Pre) and 6-mo follow-up (Post) for each participant at each time point. D: representative Western blot showing Pre- and Post-Nrf2 and corresponding total protein levels in PLA and COCOA groups, molecular weight marker shown on left. E: quantification of the change in Nrf2 after densitometric analysis of the levels of each sample normalized to corresponding total protein [(Nrf2 post/total protein)/(Nrf2 pre/total protein)]. F: representative Western blot showing Pre- and Post-HO-1 and NQO1 and corresponding total protein levels in PLA and COCOA groups, molecular weight markers shown on left. G: quantification of the change in HO-1 after densitometric analysis of the levels of each sample normalized to corresponding total protein [(HO-1 post/total protein)/(HO-1 pre/total protein)]. Pre and Post from each participant were run on the same blot. H: quantification of the change in NQO1. Data are means ± SD, difference in change between COCOA (n = 10 participants) and PLA (n = 6 participants) groups tested using an independent t test, P value shown, ns: not significant, **P < 0.01. Nrf2, nuclear factor erythroid 2-related factor 2; PAD, peripheral artery disease.

Type II-selective oxidative damage has been previously reported in PAD (59). To examine whether the protective effect of cocoa on type II fiber type-specific central nuclei was associated with an antioxidant effect of Nrf2 activation, we examined the change in the protein levels of Nrf2 as well as the Nrf2 targets, HO-1 and NQO1, in muscle biopsies from baseline to 6-mo follow-up by Western blot (Fig. 1, D–H). There was no significant difference in total Nrf2 abundance in muscle homogenates from the cocoa group relative to placebo (P = 0.317) (Fig. 1E). However, HO-1 and NQO1 are the major ARE-regulated phase II detoxifying enzymes regulated by Nrf2, and expression levels of these proteins are sensitive markers of Nrf2 activity (60, 61). HO-1 (Fig. 1G) and NQO1 (Fig. 1H) levels tended to increase (P = 0.097 and P = 0.106, respectively) within muscle homogenates from the cocoa group relative to placebo, but findings did not reach statistical significance. Furthermore, in all muscle biopsies, the increase in HO-1 and NQO1 levels from baseline to follow-up were significantly associated with decreased abundance of type II fibers with central nuclei (R² = 0.366, P = 0.017 and R² = 0.339, P = 0.023, respectively) (Supplemental Fig. S2, A and B). Taken together, these data support the hypothesis that cocoa may stimulate the Nrf2 pathway in PAD skeletal muscle, protecting against the preferential oxidative damage of type II myofibers. However, statistical power may have been insufficient to demonstrate a significant increase in HO-1 and NQO1, due to limited sample size.

Cocoa Increases Abundance of Complex III Subunit in Skeletal Muscle

In the COCOA-PAD trial, cocoa increased COX activity compared with placebo (43). To further interrogate the mechanism of preserved mitochondrial activity, we assessed the relative levels of the OXPHOS complexes NADH:ubiquinone oxidoreductase subunit B8 (NDUFB8, Complex I), succinate dehydrogenase complex iron sulfur subunit B (SDHB, Complex II), ubiquinol-cytochrome C reductase core protein 2 (UQCRC2, Complex III), and ATP synthase F1 subunit alpha (ATP5A, Complex V) at follow-up compared with baseline by Western blot (Fig. 2A). There was no significant difference between cocoa and placebo groups in the fold-change of NDUFB8 (Fig. 2B) or SDHB (Fig. 2C) from baseline to follow-up. However, UQCRC2 levels were significantly increased in the cocoa group relative to placebo (P = 0.032) (Fig. 2D), and ATP5A levels increased in the cocoa group relative to placebo, though the difference did not reach statistical significance (P = 0.096) (Fig. 2E) at 6-mo follow up. Moreover, in all muscle biopsies, there was a significant positive association between the change in HO-1 and NQO1 levels from baseline to follow-up and the change in the levels of UQCRC2 (R² = 0.529, P = 0.001 and R² = 0.279, P = 0.035, respectively) (Supplemental Fig. S2, C and D). Increases in UQCRC2 in all muscle biopsies were also significantly associated with improvements in 6-min walk distance (R2 = 0.379, P = 0.015) (Supplemental Fig. S2E). Therefore, cocoa treatment may enhance the antioxidant response via Nrf2 activation, leading to an increased abundance of OXHPOS proteins and preserving mitochondrial activity, thus improving walking performance.

Figure 2.

Six-month change in OXPHOS protein levels in muscle biopsies from COCOA-PAD trial. A: representative Western blot showing Pre and Post levels of OXPHOS proteins and corresponding total protein levels in PLA and COCOA groups, molecular weight markers shown on left. B–E: quantification of Western blots showing change in OXPHOS proteins [NDUFB8 (B), SDHB (C), UQCRC2 (D), and ATP5A (E)] after densitometric analysis of the levels of each sample normalized to corresponding total protein [(OXHOS protein post/total protein)/(OXPHOS protein pre/total protein)]. Pre and Post samples from each participant were run on the same blot. Difference in change between COCOA (n = 10 participants) and PLA (n = 6 participants) groups tested using an independent t test. P values shown, ns: not significant, *P < 0.05. ATP5A, ATP synthase F1 subunit alpha; NDUFB8, NADH:Ubiquinone oxidoreductase subunit B8; SDHB, succinate dehydrogenase complex iron sulfur subunit B; PAD, peripheral artery disease; PLA, placebo; Post, 6-mo follow-up; Pre, baseline; UQCRC2, ubiquinol-cytochrome C reductase core protein 2.

Treatment of Myotubes with PAD Serum Reduces Mitochondrial Respiration and Increases Hydrogen Peroxide Production

Recent studies have used ex vivo human serum to investigate mechanistic pathways and study the cellular response to factors such as aging (49, 62–68). To determine whether exposure to serum from patients with PAD can reproduce the disease-associated pathological hallmarks of mitochondrial dysfunction and oxidative stress, myotubes differentiated from primary MPCs isolated from muscle from healthy donors (n = 3 biological replicates) were treated with 10% serum pooled from the 10 patients in the cocoa intervention group (at baseline) or 10% pooled serum obtained from 10 age- and sex-matched healthy participants for 48 h. Characteristics for the healthy participants are provided in Table 2. Control myotubes were maintained under normal growth conditions (2% HS, Supplemental Fig. S3). We found that myotubes treated with PAD serum had a significantly lower maximal OCR than aged-matched, healthy serum-treated myotubes or myotubes maintained under normal growth conditions (P = 0.031 and P = 0.0002, respectively) (Supplemental Fig. S3, A and B). In addition, myotubes treated with both healthy serum (from age-matched, older participants) and PAD serum had higher hydrogen peroxide production relative to control myotubes, as indicated by elevated AmR fluorescence (P = 0.004 and P = 0.0006, respectively) (Supplemental Fig. S3C). These findings indicate that treatment of myotubes with serum from older participants leads to increased ROS production, regardless of health status; however, PAD serum treatment leads to greater increases in ROS production and greater impairments in mitochondrial respiration compared with serum from healthy older individuals.

Table 2.

Characteristics for MASTERS trial healthy participants

	Healthy Participant Serum (n = 10)
Age, yr; means (SD)	74.03 (5.41)
Male, n (%)	5 (50)
White, n (%)	10 (100)
BMI, kg/m2; means (SD)	24.63 (2.87)

Table 2 lists the demographics and characteristics of the participants from the MASTERS trial for which serum was used in the current study for in vitro treatments. BMI, body mass index.

EPI Treatment Activates Nrf2 in Myotubes

To examine the dose-dependent effects of epicatechin (EPI) on Nrf2 activation in vitro, myotubes were treated with increasing doses of EPI (1, 2.5, 5, and 10 μM) for 48 h (Supplemental Fig. S3, D and E). Sulforaphane (SFN, 10 μM), a well-studied Nrf2 inducer, was tested as a positive control. We found that 10 μM EPI and 10 μM SFN significantly increased Nrf2 phosphorylation, an indicator of activation (P = 0.016 and P = 0.005, respectively), relative to untreated myotubes, as indicated by increased phospho-Nrf2/total Nrf2 levels (69) (Supplemental Fig. S3, D and E). These data are consistent with a previous study showing a dose-dependent effect of EPI on Nrf2 signaling in human vascular endothelial cells (70). Based on these data, subsequent experiments were performed with doses of 10 μM EPI.

EPI Treatment Prevents Oxidative Stress in PAD Serum-Treated Myotubes by Preventing the Reduction in Nrf2 Activity

We aimed to investigate whether EPI treatment reduced ROS production and oxidative stress induced by PAD serum in muscle cells. EPI treatment significantly reduced hydrogen peroxide production after exposure to PAD serum (PAD serum vs. PAD serum + EPI, P = 0.0002) (Fig. 3A). Lipid peroxidation, assessed by 4-HNE staining, increased in myotubes treated with PAD serum compared with control untreated myotubes (P < 0.0001) (Fig. 3, B and C). EPI treatment significantly attenuated the increase in lipid peroxidation induced by PAD serum (PAD serum vs. PAD serum + EPI, P = 0.0004).

Figure 3.

Effects of epicatechin (EPI) on PAD serum-induced hydrogen peroxide production and oxidative stress in myotubes. A: hydrogen peroxide production, assessed by Amplex UltraRed (AmR) fluorescence in control myotubes (CON, standard 2% horse serum media), 10% pooled PAD serum-treated myotubes (PAD serum), and 10% pooled PAD serum-treated myotubes supplemented with 10 μM EPI (PAD serum + EPI). Data expressed as fold-regulation relative to CON myotubes. B: representative immunofluorescence images of 4-HNE staining in myotubes, image acquired at ×20 magnification. Blue: DAPI indicates nuclei and Green: 4-HNE used as a marker of lipid peroxidation and oxidative stress. Scale bar shown in bottom right. C: quantification of 4-HNE fluorescence intensity, expressed as arbitrary units (a.u.). Experiments are averages of four technical replicates. Difference between groups (n = 3 independent biological replicates/group) tested using a one-way ANOVA, results of pairwise post hoc analyses (Tukey’s) displayed as **P < 0.01, ***P < 0.001, ****P < 0.0001. PAD, peripheral artery disease.

We next evaluated whether the effect of EPI to reduce oxidative stress after treatment with PAD serum was mediated via activation of the Nrf2 pathway. First, we measured the activity of the Nrf2 target GPX as a marker of antioxidant activity. PAD serum treatment significantly reduced GPX activity in myotubes relative to control (P = 0.004), whereas there was no significant difference in GPX activity between control myotubes and myotubes treated with PAD serum + EPI (Fig. 4A). EPI treatment also increased GPX activity in PAD serum-treated compared with control myotubes, but results did not reach statistical significance (PAD serum vs. PAD serum + EPI, P = 0.0889). PAD serum reduced levels of Keap1, a protein that represses Nrf2 transcriptional activity, relative to control, and EPI had no additional effect on Keap1 levels (PAD serum vs. control: P = 0.022, PAD serum + EPI vs. control: P = 0.020) (Fig. 4, B and C). However, while PAD serum treatment also reduced Nrf2 phosphorylation, this was restored by addition of EPI (phospho-Nrf2: PAD serum vs. PAD serum + EPI, P = 0.009, Fig. 4, D and E). HO-1 protein levels were reduced by PAD serum relative to control, which was restored by EPI treatment (Fig. 4, F and G). Moreover, NQO1 protein levels were also reduced by PAD serum relative to control (P = 0.049), and although not totally restored, there were no significant differences in NQO1 levels between control myotubes and myotubes treated with PAD serum + EPI (Fig. 4, F and H). These data suggest that the protective effect of EPI against elevated ROS production and oxidative stress in muscle cells in response to PAD serum treatment may be mediated by EPI effects to prevent reduction in Nrf2 activity.

Figure 4.

Effects of EPI on antioxidant activity and Nrf2 activation in myotubes. A: glutathione peroxidase (GPX) enzymatic activity in lysates from CON, PAD serum, or PAD serum + EPI myotubes, normalized to protein concentrations. Experiments are averages of four technical replicates. B: representative Western blot of Keap1 and β-actin in myotube lysates, molecular weight markers shown on left. C: quantification of Keap1 levels after densitometric analysis of the levels of each sample normalized to corresponding loading control (β-actin), expressed as fold-regulation relative to CON. D: representative blot of phosphor-Nrf2 (p-Nrf2) and Nrf2 levels in myotube lysates. E: quantification of p-Nrf2 levels after densitometric analysis of the levels of each sample normalized to corresponding Nrf2 (p-Nrf2/Nrf2), expressed as fold-regulation relative to CON. F: representative Western blot of HO-1, NQO1, and β-actin in myotube lysates. G: quantification of HO-1 levels normalized to β-actin. H: quantification of NQO1 levels normalized to β-actin. Difference between groups (n = 3 independent biological replicates/group) tested using a one-way ANOVA, results of pairwise post hoc analyses (Tukey’s) displayed as *P < 0.05, **P < 0.01, ***P < 0.001. EPI, epicatechin; PAD, peripheral artery disease.

EPI Treatment Protects Myotubes against PAD Serum-Induced Mitochondrial Decrements

Next, we sought to test whether EPI-induced Nrf2 activation and enhanced antioxidant responses could mitigate mitochondrial respiratory decline in myotubes exposed to PAD serum. EPI treatment attenuated the PAD serum-induced reduction in the basal OCR (PAD serum vs. PAD serum + EPI: P = 0.036) (Fig. 5, A and B). There were no significant differences between PAD serum and PAD serum + EPI treatments in oligo-induced OCR or FCCP-induced OCR. Likewise, CS activity was significantly lower in both PAD serum- and PAD serum + EPI-treated myotubes compared with control myotubes (P = 0.007 and P = 0.039, respectively) (Fig. 5B). Although there was no significant difference in CS activity between PAD serum- and PAD serum + EPI-treated myotubes, treatment with EPI significantly reduced the decrease in NDUFB8, SDHB, and UQCRC2 levels after exposure to PAD serum (PAD serum vs. PAD serum + EPI, P = 0.0097, P = 0.012, and P = 0.013, respectively) (Fig. 5, C–F). Furthermore, ATP5A levels tended to increase in PAD serum + EPI-treated myotubes compared with myotubes treated with PAD serum alone (P = 0.062) (Fig. 5G).

Figure 5.

Effects of EPI on PAD serum-induced mitochondrial decrements in myotubes. A: oxygen consumption rate (OCR) trace over time after injections of indicated mitochondrial respiration modulators in CON, PAD serum, and PAD serum + EPI myotubes (left) and quantification of basal OCR (before Oligo addition), ATP-linked OCR (after Oligo addition), and maximal OCR (FCCP-stimulated) (right), normalized to protein concentration. B: citrate synthase (CS) enzymatic activity in primary myotube lysates normalized to protein concentrations. Experiments are averages of four technical replicates. C: representative Western blot of OXPHOS protein levels and corresponding total protein in primary myotube lysates, molecular weight markers shown on left. D–G: quantification of OXPHOS protein levels after densitometric analysis of the levels of each sample normalized to corresponding total protein. Data are expressed as fold-regulation relative to CON myotubes. Difference between groups (n = 3 independent biological replicates/group) tested using a one-way ANOVA, results of pairwise post hoc analyses (Tukey’s) displayed as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. EPI, epicatechin; PAD, peripheral artery disease.

Nrf2 Inhibition Reverses the Protective Effect of EPI Treatment

To further confirm the role of Nrf2 activation in the EPI-induced protection against ROS production and mitochondrial dysfunction, myotubes exposed to PAD serum and treated with EPI were also treated with the small molecule antagonist of Nrf2, ML385. Nrf2 inhibition by ML385 was demonstrated by a significant reduction in Nrf2 phosphorylation (Fig. 6, A and B) (PAD serum + EPI vs. PAD serum + EPI + ML385, P < 0.0001). Moreover, ML385 prevented the reduction in hydrogen peroxide production by EPI in myotubes exposed to PAD serum (PAD serum + EPI vs. PAD serum + EPI + ML385, P = 0.0002) (Fig. 6C). Likewise, ML385 reversed the improvement by EPI in FCCP-stimulated OCR (PAD serum + EPI vs. PAD serum + EPI + ML385, P < 0.0001) (Fig. 6D). Taken together, these data provide further evidence that protective effects of EPI treatment are mediated via Nrf2 activation.

Figure 6.

Effects of Nrf2 inhibitor ML385 on EPI-induced effects during PAD serum treatment in myotubes. A: Western blot of phospho-Nrf2 (p-Nrf2) and Nrf2 levels in PAD serum, PAD serum + EPI, and PAD serum + EPI + ML385 myotube lysates. B: quantification of p-Nrf2 levels after densitometric analysis of the levels of each sample normalized to corresponding Nrf2 (p-Nrf2/Nrf2), expressed as fold-regulation relative to PAD serum group. C: hydrogen peroxide production, assessed by AmR fluorescence. Data expressed as fold-regulation relative to PAD serum myotubes. D: OCR trace over time after injections of indicated mitochondrial respiration modulators (left), and quantification of basal OCR (before Oligo addition), ATP-linked OCR (after Oligo addition), and maximal OCR (FCCP-stimulated) (right), normalized to protein concentration. Experiments are averages of four technical replicates. Difference between groups (n = 3 independent biological replicates/group) tested using a one-way ANOVA, results of pairwise post hoc analyses (Tukey’s) displayed as *P < 0.05, ***P < 0.001, ****P < 0.0001. EPI, epicatechin; OCR, oxygen consumption rate; PAD, peripheral artery disease.

DISCUSSION

In PAD, repeated cycles of ischemia/reperfusion are thought to increase oxidative stress that contributes to the myopathy of PAD. Associations between oxidative stress and muscle damage have been reported in both preclinical models of PAD (6, 7, 71) and in humans with PAD (11, 18, 59). In addition to increased ROS production, defective cellular antioxidant defense may contribute to PAD-associated oxidative stress, and antioxidant therapy is a potentially promising therapeutic approach in PAD (72, 73). Our findings of increased Nrf2 activation may explain the large meaningful effect of cocoa flavanols in the Phase II COCOA-PAD trial (43) and may also explain benefits observed in other clinical trials in PAD using compounds with antioxidant properties including resveratrol (RESTORE trial) (74) and nicotinamide riboside (NICE trial) (under review).

Flavanols, natural phytochemicals such as epicatechin, exert their antioxidant function via Nrf2 activation (33, 34). Specifically, chemical modifications by phytochemicals cause Nrf2 to translocate to the nucleus, where it binds to the ARE to induce antioxidant gene expression (75). In the current study, we mechanistically explored the potential role of Nrf2 activation in cocoa flavanol-mediated improvements of skeletal muscle. In muscle biopsies from the COCOA-PAD trial, there was a trend for the Nrf2 targets HO-1 and NQO1 to be higher in skeletal muscle biopsies of participants randomized to cocoa flavanols compared with placebo, although differences were not statistically significant. Although the mean values of HO-1 and NQO1 in the cocoa group did not show a significant change over placebo, this may be due to the limited sample size and heterogeneity in responses. Within the cocoa group, there were some participants with large changes in Nrf2 targets, whereas others had minimal changes. Still, the increases in HO-1 and NQO1 were significantly associated with reduced abundance of type II fibers with central nuclei (P = 0.017, R² = 0.366 and P = 0.023, R² = 0.339, respectively). Compared with placebo, cocoa flavanols significantly reduced the percentage of type II fibers with central nuclei within muscle. The increase in central nuclei, indicative of muscle myopathy and degeneration (57, 58), specifically in type II fibers observed in those on placebo during the course of the 6-mo trial is consistent with previous findings indicating type II-selective oxidative damage in PAD (59). One potential mechanism to explain this fiber type difference in oxidative damage is that type II fibers have been shown to produce greater amounts of hydrogen peroxide compared with type I fibers as a result of mitochondrial free radical leak (76). Moreover, type I fibers may be protected from oxidative damage, compared with type II fibers, due to a relatively higher activity of endogenous antioxidant enzymes, including SOD and GPX (77). The lower abundance in type II fibers with central nuclei in the cocoa flavanol intervention group compared with placebo, together with the negative association of centrally nucleated type II fibers with Nrf2 targets, suggest that cocoa flavanols may have induced an antioxidant effect to protect type II fibers against oxidative damage.

Our in vitro model of PAD-associated oxidative stress and mitochondrial dysfunction adds to the growing number of studies using ex vivo human serum to condition skeletal muscle myotubes to study changes in metabolism in differing disease states (49). Results showed that treatment of myotubes with 10% serum pooled from patients with PAD increased hydrogen peroxide production, reduced Nrf2 activation, lowering the antioxidant response, and reduced mitochondrial respiration. Thus, in vitro treatment of myotubes with ex vivo PAD serum may provide a valuable experimental model to study the mechanisms of PAD-associated skeletal muscle myopathy. This model can also be a useful means for testing novel therapeutics for PAD treatment.

Importantly, our mechanistic studies in myotubes highlight a potential role for the cocoa flavanol epicatechin in protecting against skeletal muscle oxidative stress and mitochondrial respiratory deficits by preventing PAD-induced reduction of Nrf2 activation. Nrf2 regulation is mediated by Keap1-dependent and Keap1-independent mechanisms (78). Although the primary regulation of Nrf2 is thought to involve its interaction with Keap1, which keeps Nrf2 sequestered in an inactive form in the cytosol, an increasing body of literature has also revealed the importance of other mechanisms of Nrf2 regulation, including phosphorylation (78). Our data revealed that supplementation of PAD serum-treated muscle cells with epicatechin led to an increase in Nrf2 target antioxidants without altering Keap1 levels. However, Nrf2 phosphorylation was significantly enhanced with epicatechin. These data suggest a phosphorylation-mediated mechanism of Nrf2 activation, consistent with other previous reports demonstrating that epicatechin promotes Nrf2 phosphorylation (36, 41, 42).

Oxidative damage to mitochondrial proteins leads to their degradation, which can result in a failure of protein maintenance. Likewise, preventing oxidative damage by antioxidants can maintain mitochondrial protein homeostasis (79, 80). Nrf2 activation results in increased OXPHOS protein expression, and, conversely, the loss of Nrf2 downregulates OXPHOS proteins (81–86). We observed an increase in the levels of the Complex III protein, UQCRC2, in skeletal muscle biopsies of participants in the cocoa flavanol intervention group, which was associated with increased HO-1 and NQO1, as well as improved 6-min walk distance. Notably, Complex III is the major site for ROS generation in the ETC, especially during hypoxia, and Complex III is especially susceptible to ROS-induced oxidative damage (87–90). Furthermore, our in vitro experiments demonstrated that while exposure of myotubes to PAD serum reduced OXPHOS protein expression, this was attenuated by treatment with epicatechin. Taken together, these findings suggest that the antioxidant effects of cocoa flavanols may also maintain Complex III protein abundance in PAD, which may directly impact mitochondrial activity and walking performance. Positive effects of cocoa flavanols, including epicatechin, on mitochondrial function in several different cell types and disease models have been reported (91–98). Cocoa flavanols have been studied as prime candidates that may positively affect skeletal muscle metabolism specifically (99, 100). In patients with Becker muscular dystrophy, epicatechin therapy led to increased skeletal muscle mitochondrial biogenesis and functional exercise improvements (91). Similarly, in patients with heart failure and type 2 diabetes, cocoa flavanols improved muscle antioxidant expression, oxidative stress markers, and mitochondrial biogenesis (101–103). Our work suggests that cocoa flavanols may also directly target skeletal muscle in PAD to modulate the Nrf2 pathway and alleviate oxidative stress and mitochondrial dysfunction. Interestingly, our in vitro experiments also demonstrated that inhibition of Nrf2 prevents epicatechin-mediated protection, further implicating the Nrf2 pathway in improvements mediated by cocoa flavanols.

Our study has several limitations. First, the sample size for patients providing skeletal muscle biopsies was small. Second, the in vitro cell culture conditions do not entirely reflect in vivo conditions, where myofibers are in close contact with the extracellular matrix and a variety of different mononuclear cell populations. Although the focus of these studies was myotubes differentiated from MPCs, we cannot rule out that the effect of epicatechin on skeletal muscle involves modulation of other cell types. Third, the relatively small sample size was associated with some imbalances in characteristics between the cocoa and placebo groups. These imbalances may have confounded associations reported here. Finally, cocoa administration results in the appearance of many metabolites in plasma; however, our in vitro studies only assessed (−)-epicatechin. We cannot conclude that in vivo effects of cocoa were solely due to epicatechin rather than other metabolites or epicatechin conjugates. Still, epicatechin is the major flavanol found in cocoa and its antioxidant properties are mainly attributed to epicatechin (104–106). Moreover, beneficial effects of cocoa on vascular function are thought to be mediated by epicatechin (107). Lastly, the cocoa treatment in the COCOA-PAD trial was supplemented with 75 mg of epicatechin and structurally related epicatechin metabolites were shown to be elevated in plasma of participants randomized to cocoa (43).

DATA AVAILABILITY

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

SUPPLEMENTAL DATA

Supplemental Table S1 and Supplemenatl Figs. S1–S3: https://doi.org/10.6084/m9.figshare.24891714.v1.

GRANTS

This research was supported in part by the intramural research program of the US Department of Agriculture, National Institute of Food and Agriculture (Agriculture and Food Research Initiative, Grant No. 2022–67012-38533, Project Accession No. 1029340). This study was also funded by the National Institute on Aging (R21-AG050897 and R01-AG066724).

DISCLAIMERS

The findings and conclusions in this publication have not been formally disseminated by the US Department of Agriculture and should not be construed to represent any agency determination or policy.

DISCLOSURES

M. M. McDermott has received research funding from Hlixmith and has received other research support from Mars, Chromadex, ReserveAge, Helixmith, and ArtAssist. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

A.I., M.M.M., L.F., C.A.P., and K.K. conceived and designed research; A.I., J.K.J., J.C.S., K.J.H., R.S., and K.K. performed experiments; A.I., J.C.S., D.Z., and K.K. analyzed data; A.I., M.M.M., K.J.H., R.S., L.F., C.A.P., and K.K. interpreted results of experiments; A.I. and K.K. prepared figures; A.I. drafted manuscript; M.M.M., K.J.H., R.S., L.F., C.A.P., and K.K. edited and revised manuscript; A.I., M.M.M., J.K.J., J.C.S., D.Z., K.J.H., R.S., L.F., C.A.P., and K.K. approved final version of manuscript.