Exercise reduces the protein abundance of TXNIP and its interacting partner REDD1 in skeletal muscle: potential role for a PKA-mediated mechanism

Abstract

Thioredoxin-interacting protein (TXNIP) negatively effects the redox state and growth signaling via its interactions with thioredoxin (TRX) and regulated in development and DNA damage response 1 (REDD1), respectively. TXNIP expression is downregulated by pathways activated during aerobic exercise (AE), via posttranslational modifications (PTMs; serine phosphorylation and ubiquitination). The purpose of this investigation was to determine the effects of acute AE on TXNIP expression, posttranslational modifications, and its interacting partners, REDD1 and TRX. Fifteen healthy adults performed 30 min of aerobic exercise (80% V̇o_2max) with muscle biopsies taken before, immediately following, and 3 h following the exercise bout. To explore potential mechanisms underlying our in vivo findings, primary human myotubes were exposed to two models of exercise, electrical pulse stimulation (EPS) and palmitate-forskolin-ionomycin (PFI). Immediately following exercise, TXNIP protein decreased, but returned to preexercise levels 3 h after exercise. These results were replicated in our PFI exercise model only. Although not statistically significant, there was a trending main effect in serine-phosphorylation status of TXNIP (P = 0.07) immediately following exercise. REDD1 protein decreased 3 h after exercise. AE had no effect on TRX protein expression, gene expression, or the activity of its reducing enzyme, thioredoxin reductase. Consequently, AE had no effect on the TRX: TXNIP interaction. Our results indicate that AE leads to acute reductions in TXNIP and REDD1 protein expression. However, these changes did not result in alterations in the TRX: TXNIP interaction and could not be entirely explained by alterations in TXNIP PTMs or changes in TRX expression or activity.

NEW & NOTEWORTHY Aerobic exercise is an effective tool in the prevention and treatment of several chronic metabolic diseases. However, the mechanisms through which these benefits are conferred have yet to be fully elucidated. Our data reveal a novel effect of aerobic exercise on reducing the protein expression of molecular targets that negatively impact redox and insulin/growth signaling in skeletal muscle. These findings contribute to the expanding repository of molecular signatures provoked by aerobic exercise.

INTRODUCTION

Skeletal muscle is a metabolically active organ, accounting for 40%–50% of total body mass and 80%–90% of insulin-stimulated glucose uptake following a meal (1). As such, impaired skeletal muscle metabolism has been implicated in the pathogenesis of diabetes (1), metabolic syndrome (2), nonalcoholic fatty liver disease (3), and sarcopenia (4). Regular aerobic exercise offers protection from these diseases, in part by enhancing skeletal muscle insulin sensitivity (5), muscle protein synthesis (6), antioxidant capacity (7), and enhanced mitochondrial biogenesis (8, 9). A protein target with known implications in several of these metabolic processes is thioredoxin-interacting protein (TXNIP).

TXNIP, a member of the ∝-arrestin protein family, elicits its cellular effects by forming complexes with other protein targets including thioredoxin (TRX) and regulated in development and DNA damage responses 1 (REDD1) (10). By binding to redox-sensitive cysteine residues of TRX, TXNIP can reduce its activity to promote oxidative stress (10). Interestingly, TXNIP is also vital for REDD1’s stability, as TXNIP knockdown increases the rate of REDD1 degradation, thus relieving the suppression of mTOR signaling (11). Considering the role of redox balance in regulating insulin sensitivity and mTOR signaling in regulating muscle mass, an abundance of TXNIP protein may promote an insulin-resistant, sarcopenic environment, through its protein-protein interactions (3, 11–18). Recent evidence from cell culture models have indicated an increase in AMPK activity, cyclic AMP (cAMP), and/or activation of protein kinase A (PKA) can downregulate TXNIP expression by augmenting its rate of proteasome degradation via coupled serine phosphorylation-ubiquitination (19–21). AMPK and PKA are also readily activated during aerobic exercise, and therefore may be involved with mediating the reductions in TXNIP expression in rodent skeletal muscle seen with swimming and treadmill exercise (22). However, it is not known whether acute aerobic exercise is an adequate stimulus to reduce TXNIP expression in human skeletal muscle.

The goal of this study was to determine the effects of aerobic exercise on skeletal muscle TXNIP protein expression, posttranslational modifications, and intermolecular interactions with binding partners TRX and REDD1. We hypothesized that exercise would activate enzymes (i.e., PKA, AMPK) capable of phosphorylating TXNIP, increasing its susceptibility to ubiquitination and proteasome degradation, and subsequently result in a reduction in TXNIP protein expression. The lower expression of TXNIP would decrease the REDD1’s stability; we therefore hypothesized that there would be a delayed reduction in REDD1 protein expression. The dynamics of the TRX: TXNIP interaction, however, is slightly more complicated, as acute exercise has been shown to increase components of the antioxidant defense system, including TRX (23, 24). Furthermore, the TRX: TXNIP interaction is negatively regulated by thioredoxin reductase 1 (TRXR1) by reducing intramolecular and intermolecular disulfide bridges on thioredoxin (25). Therefore, we also examined TRX expression and TRXR1 activity to gain a comprehensive understanding of exercise’s effect on the TRX:TXNIP protein interaction. To further dissect the possible effects of exercise, we utilized two in vitro models of exercise (electrical pulse stimulation and a palmitate/forskolin/ionomycin cocktail) to examine responses in primary human myotubes.

METHODS

Study Design

During the first visit, all procedures and potential risks involved with the study were explained to the participants. Upon providing verbal and written informed consent, the subjects were enrolled and height, weight, body fat percentage via dual X-ray absorptiometry (DXA), and maximal aerobic capacity (V̇o_2max) were collected. Visits 2, 3, and 4 consisted of familiarization trials consisting of aerobic treadmill exercise at 40%, 65%, and 80% of V̇o_2max, for 60, 30, and 30 min, respectively. Each exercise bout was separated by at least 4 days. Three days before each visit, subjects completed diet and physical activity logs, which they were asked to replicate in the days leading up to the subsequent visits. Subjects were also instructed to abstain from vigorous exercise and alcohol consumption 48 h and caffeine consumption 24 h before each visit. In addition, subjects fasted for at least 12 h. All experimental protocols were approved by the Institutional Review Board at the University of Illinois of Chicago (IRB Approval No.: 2015-0127).

Participants

Fifteen healthy adults, (8 M, 7 F, age: 25 ± 4 yr) participated in the study. An eligibility checklist was used to screen potential participants over the phone to determine eligibility and health history. Exclusion criteria included smoking within the past year, a diagnosis of diabetes, cardiovascular disease, kidney disease, major depression, high blood pressure, or high blood cholesterol. Inclusion included 18–35 yr of age and body mass index (BMI) between 18 and 26 kg/m². Baseline characteristics for all participants are presented in Table 1.

Table 1.

Subject characteristics

n	15
Age, yr	25.7 ± 4.0
Sex, %F	47%
Weight, kg	68.3 ± 8.3
BMI, kg/m2	22.4 ± 2.6
BF (%)	23.1 ± 5.7
V̇o2max, mL/kg/min	47.7 ± 7.4

Subject baseline characteristics and treadmill parameters for 80% V̇o_2max trial. Data are depicted as means ± SD. BMI, body mass index.

V̇o_2max and Aerobic Exercise

V̇o_2max was determined using a treadmill protocol during which the subjects ran at a uniform speed while grade increased 2% every 2 min, until volitional fatigue. Expired air was analyzed via the PARVO Medics metabolic cart (Salt Lake City, UT). Heart rate was continuously monitored and subjects provided a rating of perceived exertion (RPE) (26) at the end of each 2-min interval. V̇o_2max was achieved if the subjects met three of the four criteria: a plateau in V̇o₂ despite an increasing workload, an RPE > 17, RER > 1.1, and an HR > 85% age-predicted maximal heart rate. Oxygen consumption at 40%, 65%, and 80% of their V̇o_2max was calculated and inserted into the appropriate ACSM metabolic equations to derive the proper treadmill speed and grade to achieve the appropriate V̇o₂ for the submaximal exercise bouts (27). The average speed and grade performed during the final visit were 6.5 ± 1.1 and 1.0 ± 1.0, respectively. All exercise testing sessions and biopsies were completed in the morning following an overnight fast.

Skeletal Muscle Biopsy

On the final visit day, subjects performed 30 min of high-intensity aerobic exercise (80% V̇o_2max). Skeletal muscle biopsies were taken from the vastus lateralis before, immediately after, and 3 h after the exercise bout. Pre- and postexercise biopsies were alternated between right and left legs. On the final biopsy (3-h post), we sampled the muscle from a new distal incision sight on the leg that was sampled before the exercise bout. Before the biopsy, local anesthetic (lidocaine HCl 1%) was administered followed by a small incision (∼0.5 cm) at the biopsy site. A Bergstrom needle was inserted with suction extracting ∼200 mg of muscle tissue. Muscle tissue was cleared of all visible connective tissue and fat, blotted with gauze to remove blood, immediately flash frozen in liquid nitrogen and stored at −80°C until further analysis was performed.

Tissue Homogenization and Protein Quantification

Muscle tissue (∼10 mg) was weighed and homogenized with ceramic beads (Lysing Matrix D; FastPrep-24; MP Bio, Santa Ana, CA) in 20 volumes of 1X cell lysis buffer [20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM-glycerophosphate, 1 mM Na₃PO₄, 1 g/mL leupeptin; Cell Signaling Technology, Danvers, MA] supplemented with protease/phosphatase inhibitor cocktail (1: 10,000: MSSAFE; Sigma, St. Louis, MO). Protein concentration for each sample homogenate was determined by a commercially available bicinchoninic acid (BCA) protein assay kit (Pierce, Rockford, IL).

Western Blotting

Aliquots containing 30 µg of total protein were diluted in equal volumes of 2X Laemmli Buffer (BioRad, Hercules, CA), with 5% β-mercaptoethanol (βME) before heating at 70°C for 10 min. Denatured samples were brought to room temperature, loaded onto a 12.5% polyacrylamide gel, separated by SDS-PAGE, and transferred to nitrocellulose membranes. Membranes were blocked with Odyssey Blocking Buffer (OBB; Li-Cor, Lincoln, NE) for 1 h and were incubated with primary antibodies TXNIP (1: 1,000; Abcam, ab188865, Cambridge, UK), TRX (1:1,000; Abcam, ab16965), REDD1 (1:1,000; Proteintech, 10638-1-AP), and GAPDH (1:5,000; Cell Signaling, 2118) overnight at 4°C in OBB with 0.1% Tween-20. Membranes were washed with TBST and incubated with an anti-rabbit or anti-mouse fluorophore-conjugated secondary antibody (1:20,000, Li-Cor) in OBB supplemented with 0.1% Tween-20 for 1 h. The membranes were washed with TBST followed by TBS before being scanned on the Odyssey CLx Imaging System (Li-Cor) and quantified on Image Studio software (V4.0.21; Li-Cor). GAPDH was used as a loading control. TXNIP and REDD1 antibodies were validated in-house (see Supplemental material; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.16992013.v4). Antibodies against TRX, phosphoserine, and ubiquitin have been validated elsewhere (28–30). Uncropped images of all immunoblots can be found in the supplementary data files.

Immunoprecipitation

Aliquots of skeletal muscle tissue homogenate, prepared as described in the Tissue Homogenization and Protein Quantification section, were incubated overnight while rotating at 4°C with an anti-TRX antibody (2 µg antibody: 100 µg protein; Abcam) or anti-TXNIP antibody (4 µL antibody: 200 μL total protein; Cell Signaling), which were bound to magnetic beads (Dynabeads Protein G; Thermo Fisher Scientific, Waltham, MA). The beads containing the antigen-antibody complex were washed three times with ice-cold PBS. Following the last wash, the beads were resuspended in Laemmli Buffer (BioRad) with 5% βME and heated at 70°C for 10 min to denature and elute the proteins from the beads. The eluate was then loaded onto a 12.5% polyacrylamide gel, separated by SDS-PAGE, and transferred to a nitrocellulose membrane. Membranes were blocked in OBB (Li-Cor) for 1 h followed by incubation with primary antibodies overnight at 4°C. Membranes containing TXNIP immunoprecipitate were probed with anti-phosphoserine (1:500; EMD Millipore, AB1603), anti-ubiquitin (1:500; Santa Cruz, sc-8017), REDD1 (1:1,000), and TXNIP antibodies (1:1,000; Cell Signaling, 14715). Multiple proteins were able to be detected on a single membrane by a combined approach of membrane striping and utilization of differential secondary antibodies labeled with spectrally distinct, near-infrared fluorescent dyes at 700 nm and 800 nm emission wavelengths (Li-Cor). Merged images confirmed the absence of spectral overlap. Membranes containing TRX immunoprecipitate were probed for TXNIP (1:1,000; Cell Signaling, 14715) and TRX (1:1,000; Abcam, ab16965). Membranes were washed with TBST and incubated with appropriate fluorophore-conjugated secondary antibody in OBB supplemented with 0.1% Tween-20 for 1 h. To avoid detection of the IgG heavy chain (50 kD) from the antibody coeluted with the protein of interest, a light chain-specific anti-rabbit secondary (1:100,000, Jackson ImmunoResearch, 211–622-171) was utilized. The membranes were washed with TBST followed by TBS before being scanned and quantified on the Odyssey CLx Imaging System (Li-Cor). All protein signals were normalized to the protein target that was immunoprecipitated from the sample.

RNA Extraction and Reverse Transcription

At the time of collection, a portion of the biopsy sample (∼20 mg) was fixed in RNAlater (Invitrogen) and immediately frozen in liquid nitrogen. Samples were then stored at −80°C until RNA extraction. RNA extraction was performed using Qiagen’s RNeasy kit as per kit protocol with modifications to increase yield for fibrous tissue. Briefly, RNAlater-fixed samples were homogenized in RLT buffer with βME via bead homogenization, as described above. Samples were treated with 10 µL proteinase K (Qiagen), incubated at 55°C for 10 min and then centrifuged at 10,000 g for 3 min at room temperature. Supernatants were transferred to a sterile microfuge tube and 450 µL of ethanol was added to each sample before transfer to the RNeasy spin columns. Ethanol extracts were collected via centrifugation at 9,000 g for 30 s at room temperature and the Qiagen protocol provided was followed thereafter. Two microliter of each extraction was analyzed for RNA concentration and purity via Nanodrop (Thermo Fisher Scientific). Average RNA concentration was 128 ± 66.7 ng/µL (means ± SD) for all samples. Absorbance values (A260, A280) and ratios (A260/A280, A260/A230) are provided in the Supplemental material. Reverse transcription was performed with iScript advanced reverse transcriptase kit (BioRad) via manufacturer protocols to generate 150 ng of cDNA, which was then diluted 1:4 with nuclease-free water.

Droplet Digital PCR

Droplet digital PCR (ddPCR) (BioRad) was used to quantify transcripts of TRX, thioredoxin reductase 1 (TRXR1), as previously described (31). Primers for TRX, thioredoxin, and TRXR1 were designed using Roche’s universal probe library (UPL) assay design center (lifescience.roche.com). Forward and reverse primer sequences used for TXNIP were designed, as described previously (32, 33). Primers and probes used for each of the targets are provided in the Supplementary material. In brief, ddPCR assay was performed by combining 2 µL of cDNA (3.75 ng) with ddPCR mix for probes (BioRad), along with appropriate probes, primers, and nuclease-free water yielding 20 μL for each reaction. A triplicate of no-template negative controls in which the cDNA was substituted for nuclease-free water was performed alongside each of the assays and subtracted from the final signal. The 20 µL reaction was then combined with 70 μL of droplet oil for probes (BioRad) and droplets were generated using a droplet generator (BioRad). 40 µL of the resultant droplet suspension was then carefully pipetted onto a 96 well plate, sealed and placed in a thermocycler where 40 cycles of PCR were performed. The droplets were analyzed using a QX200 Droplet Digital PCR system (BioRad) by counting the droplets positive for FAM fluorescent probes. Transcript copy number was corrected for nonspecific signal by subtracting the number of copies detected in the NTC samples.

Thioredoxin Reductase Activity Assay

TRXR activity was assayed by a commercially available kit (Cayman Chemical) in a subset of the muscle samples previously used to measure the TRX: TXNIP interaction. The assay was performed according to the manufacturer protocol apart from homogenization buffer, which was the same buffer described for Western blotting analysis (cell lysis buffer and cell signaling protease phosphatase inhibitor). In brief, the principal of the assay is to monitor the reduction of DTNB with NADPH to TNB. Muscle homogenate (20 µL) was assayed in duplicate with and without sodium aurothiomalate (ATM), which is a specific TRXR inhibitor. Reduction of DTNB results in a yellow color change that when corrected for the color change that occurs in ATM-treated samples is proportional to TRXR activity. Color development was monitored at 405 nm over 5 min and TRXR activity was calculated by the following equation:

$$\frac{\mathrm{C}\mathrm{o}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{e}\mathrm{d}\mathrm{ }\mathrm{\Delta }\mathrm{A}/\mathrm{t}\mathrm{i}\mathrm{m}\mathrm{e}}{7.92\mathrm{m}{\mathrm{M}}^{-1}}\times \frac{0.2\mathrm{m}\mathrm{L}}{0.02\mathrm{ }\mathrm{m}\mathrm{L}}\times \mathrm{S}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{ }\mathrm{D}\mathrm{i}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}.$$

Values were corrected for protein loaded and expressed as µmol/min/µg protein.

In Vitro Exercise Stimulation

As described previously (34), satellite cells were isolated from a muscle biopsy obtained from the vastus lateralis of a lean, healthy, and recreationally active male subject (age = 22 yr, BMI = 25.7 kg/m²) and amplified on type-I collagen-coated plates in growth media (DMEM low glucose medium supplemented with 10% FBS, 0.5 mg/mL BSA, 0.05% fetuin, 20 ng/mL human epidermal growth factor, 0.39 µg/mL dexamethasone, and 100 µg/mL penicillin/streptomycin) in a 5% CO₂ and 37°C humidified atmosphere. Upon reaching 80%–90% confluence, myoblasts were switched to differentiation media for 5 days (DMEM low glucose medium supplemented with 2% horse serum, 0.3% BSA, 0.05% fetuin, and 100 µg/mL penicillin/streptomycin) to induce the formation of multinucleated, mature myotubes. On day 4, mature myotubes were stimulated for 24 h with repeated bipolar electrical stimuli (11.5 V, 1 Hz and 2 ms), provided by a pulse generator (C-Dish, IonOptix LLC, Milton, MA). During the final 2 h of EPS, another set of myotubes were treated with 30 µM palmitate and 0.5 µM Ionomycin, with or without the addition of 4 µM forskolin to isolate the effect of cAMP/PKA signaling. This pharmacological cocktail has been used previously (35) to mimic certain molecular events that occur during exercise in vivo. For example, palmitate is used to mimic the lipolytic environment resulting from epinephrine/norepinephrine stimulation. Forskolin is used to increase the bioavailability of cAMP and increase PKA activation to mimic β-adrenergic stimulation. Finally, the calcium ionophore, ionomycin, increases intracellular calcium concentrations to activate calcium-sensitive pathways normally activated during exercise. Immediately following stimulation, cells were rinsed with DPBS, harvested in ice-cold lysis buffer [50 mM HEPES, 100 mM sodium fluoride, 50 mM sodium pyrophosphate, 10 mM sodium orthovanadate, 10 mM EDTA, 1% Triton X-100, and protease and phosphatase (1 and 2) inhibitor cocktails (Sigma, St. Louis, MO)] per well, and sonicated for 5 s, followed by a 2 h rotation at 4°C. Cell lysates were subjected to immunoblot analysis as described in Western Blotting and probed for TXNIP (1:1,000; Abcam), REDD1 (1:1,000; Proteintech), and Phospho-PKA substrate (1:1,000; Cell Signaling, 9624) to verify forskolin-induced PKA activity.

Statistical Analysis

All data are presented as the means ± standard deviation. Statistical analysis was performed using Prism 4.0 software (GraphPad Software, Inc., La Jolla, CA). Differences in protein signals across the three time points were analyzed via one-way repeated-measures ANOVA. Pearson’s r was performed to analyze the relationship between dependent variables. Significance was set at P ≤ 0.05.

RESULTS

Effect of Acute Exercise on TXNIP Protein Expression and Posttranslational Modifications

To determine the effect of acute aerobic exercise on TXNIP protein expression, we conducted Western blot analysis on the tissue collected before, immediately following, and 3 h following exercise (Fig. 1). Immediately following exercise, TXNIP protein expression was significantly reduced (P < 0.05) but returned to baseline values at 3 h (Fig. 1A). To detect changes in serine phosphorylation and ubiquitination status, we immunoprecipitated TXNIP from the tissue lysate and conducted Western blot analysis. There was a trending main effect for serine phosphorylation (P = 0.07; Fig. 1B, but not for ubiquitination P = 0.39; Supplemental material). Furthermore, we assessed changes in TXNIP mRNA levels and found no differences in across the time points (Fig. 1C).

Figure 1.

Skeletal muscle TXNIP expression and posttranslational modifications. Representative Western blot images and quantification for TXNIP expression (A) and phosphorylated serine (P-Ser) residues (B). Exercise reduced TXNIP protein expression immediately following exercise but returned to baseline 3 h following. There was trending significance for the main effect of exercise for serine phosphorylation (P = 0.07). TXNIP mRNA expression was assessed via droplet digital PCR (C). TXNIP mRNA transcript levels were similar across the timepoints. n = 9–15, *significantly different from Pre values, P < 0.05. TXNIP protein expression was normalized to GAPDH. P-Ser signals were normalized to TXNIP. TXNIP mRNA expression was normalized to total cDNA. TXNIP, thioredoxin-interacting protein.

Potential Mechanisms Regulating TXNIP Abundance during Acute Exercise

To explore these potential mechanisms, we stimulated human primary myotubes with two in vitro models of exercise, electrical pulse stimulation (EPS), a combination of 30 µM palmitate, and 0.5 µM Ionomycin (PI), or a combination of 30 µM palmitate, 0.5 µM Ionomycin, and 4 µM forskolin (PFI) for 2 h. PFI treatment significantly reduced TXNIP expression whereas EPS or PI alone had no effect (Fig. 2, A and B). Furthermore, only PFI treatment resulted in a significant elevation in phosphorylation status of PKA substrates (P < 0.05) (Fig. 2D).

Figure 2.

Mechanisms of altered TXNIP expression by exercise signals in human primary myotubes. TXNIP and PKA-phosphorylated substrates (p-PKA) protein levels were analyzed in human primary myotubes after treatment with exercise signals including 24 h of electrical stimulation (EPS) (A and C), or a combination of palmitate with ionomycin (PI) or palmitate, forskolin, and ionomycin (PFI) (B and D). Two hours of PFI treatment attenuated TXNIP protein expression and increased p-PKA. n = 6 replicates per condition. *Significantly different from vehicle control (VCTRL), P < 0.05. †Significantly different from PI P < 0.05. Both TXNIP and p-PKA signals were normalized to β-actin. TXNIP, thioredoxin-interacting protein.

Effect of Acute Aerobic Exercise on REDD1 Protein Expression

To understand the effects of acute aerobic exercise on the expression of REDD1, we conducted Western blot analysis on the tissue lysate collected before, immediately following, and 3 h following exercise (Fig. 3). Before exercise, REDD1 protein expression significantly correlated with TXNIP expression (r = 0.61, P < 0.05) (Fig. 3A). There was a significant main effect of exercise on REDD1 protein expression, and post hoc analysis revealed a significant decrease 3 h following exercise (P < 0.05) (Fig. 3B).

Figure 3.

Skeletal muscle REDD1 expression. REDD1 expression was analyzed via Western blot analysis. Preexercise REDD1 expression was correlated with preexercise TXNIP expression (r = 0.610, P = 0.016) (A). Three hours following exercise, there was a significant decrease in REDD1 protein expression (B). n = 15, P < 0.05. Due to sample limitations, we were unable to measure one subject’s pre- and one subject’s post-TRX protein expression values; therefore these values were mean substituted before analysis via one-way repeated measures ANOVA. TXNIP and REDD1 signals were normalized to GAPDH. REDD1, regulated in development and DNA damage response 1; TRX, thioredoxin; TXNIP, thioredoxin-interacting protein.

Effect of Acute Exercise on TRX: TXNIP Protein Interactions

To explore the potential effect of TXNIP protein expression on its interactions with TRX, we coimmunoprecipitated TRX from the tissue lysates and conducted Western blot analysis to quantify TRX-associated TXNIP protein expression (Fig. 4). Following exercise, TRX protein expression did not change (P > 0.05) (Fig. 4A). Similarly, TRX (Fig. 4B) and TrxR1 (Fig. 4C) mRNA transcripts were not altered following exercise (P > 0.05), and neither was their ratio (P > 0.05) (Fig. 4D). Furthermore, the activity of TrxR1 was not altered by acute exercise (Fig. 4E). In line with these findings, we did not see any changes in the TRX-TXNIP interaction following exercise (P > 0.05) (Fig. 4F). Collectively, these data suggest that acute exercise mainly alters TRX/TXNIP biology in muscle through degradation of TXNIP, which may be triggered by its serine phosphorylation.

Figure 4.

Skeletal muscle TRX, TrxR1 expression, TrxR1 enzymatic activity, and TRX: TXNIP protein-protein interaction. Western blot analysis and ddPCR revealed exercise had no effect on TRX protein expression (A) or transcripts (B), TRXR1 transcripts (C), nor the ratio of TRX: TRXR1 transcripts (D). Furthermore, exercise had no effect on TRXR1 activity (E). As expected, coimmunoprecipitation analysis revealed that exercise had no effect on TRX: TXNIP protein interaction (F). n = 9–15, P > 0.05. Due to sample limitations, we were unable to measure one subject’s pre- and one subject’s post-TRX protein expression values; therefore these values were mean substituted before analysis via one-way repeated measures ANOVA. ddPCR, droplet digital PCR; TRX, thioredoxin; TXNIP, thioredoxin-interacting protein.

DISCUSSION

In the present study, we sought to determine the effects of acute aerobic exercise on TXNIP biology including protein expression, protein-protein interactions, PTMs, and expression of its interacting partners, TRX, and REDD1. Our hypotheses were that acute aerobic exercise would reduce TXNIP expression via a coupled phosphorylation-ubiquitination event, which has been shown to increase its susceptibility to degradation. Evidence suggests these phosphorylation and ubiquitination events are triggered by elevations in AMPK, [cAMP], and/or PKA activity (19–21). Therefore, we expected an increase in the serine-phosphorylation and ubiquitination signatures on TXNIP following exercise as well.

In agreement with our hypothesis, we observed decreased TXNIP expression immediately following exercise. These data are consistent with a previous report of a reduction in TXNIP protein expression immediately following exercise in rats, which returned to baseline 3 h after the bout (22). However, reductions in TXNIP could not be explained by PTMs, as there were no changes in serine phosphorylation or ubiquitination status following exercise. Furthermore, we found no differences in TXNIP mRNA levels across the timepoints, indicating changes in protein expression were not due to changes in the mRNA abundance. However, exercise failed to increase AMPK phosphorylation on its activation site (Thr¹⁷²) (Supplemental material), which agrees with recent observations in endurance-trained athletes (36). Interestingly, our in vitro models of exercise revealed that TXNIP protein expression was significantly reduced in the presence of forskolin. As expected, levels of phosphorylated substrates of PKA only increased in the presence of forskolin. These data suggest that exercise may elicit its effects on TXNIP protein expression via elevations in [cAMP] and/or PKA activation, not AMPK, even though we failed to verify a significant serine-phosphorylation signature following exercise. It is possible; however, the window of time during which PTM signatures were highest, occurred during the exercise bout and were therefore not captured with our immediate or 3-h post-tissue collection.

Evidence has suggested REDD1 and TXNIP’s protein stability is mutually dependent. Therefore, we expected reductions in TXNIP protein would lead to a similar, albeit delayed, reduction in REDD1 protein. In line with our hypothesis, REDD1 protein expression decreased 3 h following exercise. Most reports utilizing aerobic exercise has been assessed in untrained rodents and have shown that aerobic exercise can increase REDD1 expression at both the mRNA and protein levels (37–39). In humans, however, Masschelein et al. (40) reported that REDD1 expression in human skeletal muscle failed to increase following a high-intensity bout of exercise performed in a hypoxic environment. However, hypoxia alone can upregulate REDD1 protein levels, rendering it difficult to assess the independent effect of exercise (40). Our study design utilized a relatively high-intensity (80% V̇o_2max) aerobic exercise bout, and we found a significant decrease in REDD1 protein expression 3 h following the exercise. Previous evidence indicates that REDD1 protein stability is dependent on its interaction with TXNIP, with Jin et al. (11) showing that TXNIP can dimerize with REDD1 to increase its cellular lifespan. However, when TXNIP gene expression was silenced, this led to concomitant reduction in REDD1 protein expression (11). Therefore, our results agree with previous data and our hypothesis regarding TXNIP-dependent REDD1 protein stability. Considering the relatively high aerobic fitness (recreationally trained) of our subjects, and how our data conflicts with what has been shown in untrained rodents, future investigations should seek to understand the impact that prior aerobic fitness has on the acute effects of exercise on REDD1 protein expression. Furthermore, recent data in rodents indicate that REDD1 gene expression is regulated by the circadian rhythm. Specifically, the authors observed expression in rodent skeletal muscle is highest during the dark cycle followed by a rapid decline during the light cycle (41). These data may have implications for the current findings considering the subjects were exercised in the morning, transitioning from the dark to light cycle. Therefore, we cannot completely disregard the role of circadian rhythm in the observed decline in REDD1 protein expression.

Last, we sought to determine the effect of acute aerobic exercise on TRX-TXNIP protein-protein interactions. As mentioned, this interaction is important for the redox state as TXNIP is an endogenous inhibitor of TRX and can reduce its antioxidant capabilities. We acknowledge that this interaction is not solely driven by changes in TXNIP protein expression, as TRX expression and enzymatic activity of TrxR1, also have independent effects on this interaction. Therefore, we measured TRX mRNA/protein expression as well as mRNA expression and enzymatic activity of TrxR1 in the tissue lysates. TrxR1 is the cytosolic isotype of the enzyme responsible for reducing cysteines of TRX thereby breaking any intermolecular disulfide bridges between other interacting proteins such as TXNIP. Contrary to our hypothesis, we did not find any changes in TRX mRNA or protein expression following acute aerobic exercise. Moreover, we did not see changes in TrxR1 gene expression nor its enzymatic activity. Other studies have shown TRX expression in the brain and peripheral blood mononuclear cells increases following acute and chronic aerobic exercise (23, 24, 42); however, the current study was the first in human skeletal muscle, which points to a possible tissue-specific regulation of this protein target. Furthermore, the exercise bout may not have been robust enough to elicit changes in TRX due to the relatively high fitness of the participants. In agreement, we were unable to detect changes in the TRX: TXNIP interaction following acute aerobic exercise. TXNIP has been shown to dissociate rapidly from TRX following treatment with an oxidative stress inducer, uric acid (13). Aerobic exercise has been shown to promote transient elevations in oxidative stress, which could elicit a similar dissociation. However, because we recruited aerobically fit individuals, it is possible they underwent minimal changes in the redox state, thereby preserving the disulfide bridges between TRX and TXNIP.

Conclusions

Our data indicate that acute aerobic exercise was able to significantly reduce TXNIP and REDD1 protein expression, which may be mediated by a PKA- or cAMP-related mechanism, as indicated by our in vitro experiments. However, we did not observe changes in TRX or TrxR1 expression and/or TrxR1 activity following acute aerobic exercise in human skeletal muscle and likewise we did not observe changes in its interaction with TXNIP. Overall, this is the first study to examine the effect of acute aerobic exercise on TXNIP protein expression and PTM modifications in the skeletal muscle of healthy individuals. Although other investigations have looked at the effect of acute aerobic exercise on TXNIP and TRX independently, this is first study to examine its effect on their interaction. Last, this is the first investigation to examine the effect of high-intensity acute aerobic exercise alone on REDD1 protein expression in humans.

SUPPLEMENTAL DATA

All Supplemental Material is available at https://doi.org/10.6084/m9.figshare.16992013.v4.

GRANTS

This work was supported by the American Diabetes Association 1-14-JF-32, National Institutes of Health (NIH) Grants R01DK109948 and UL1RR029879 (to J.M.H), and ACSM Grant 19-01128 (to A.B.C.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

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