Dicarbonyl stress and glyoxalase enzyme system regulation in human skeletal muscle

Abstract

Skeletal muscle insulin resistance is a hallmark of Type 2 diabetes (T2DM) and may be exacerbated by protein modifications by methylglyoxal (MG), known as dicarbonyl stress. The glyoxalase enzyme system composed of glyoxalase 1/2 (GLO1/GLO2) is the natural defense against dicarbonyl stress, yet its protein expression, activity, and regulation remain largely unexplored in skeletal muscle. Therefore, this study investigated dicarbonyl stress and the glyoxalase enzyme system in the skeletal muscle of subjects with T2DM (age: 56 ± 5 yr.; BMI: 32 ± 2 kg/m²) compared with lean healthy control subjects (LHC; age: 27 ± 1 yr.; BMI: 22 ± 1 kg/m²). Skeletal muscle biopsies obtained from the vastus lateralis at basal and insulin-stimulated states of the hyperinsulinemic (40 mU·m⁻²·min⁻¹)–euglycemic (5 mM) clamp were analyzed for proteins related to dicarbonyl stress and glyoxalase biology. At baseline, T2DM had increased carbonyl stress and lower GLO1 protein expression (−78.8%), which inversely correlated with BMI, percent body fat, and HOMA-IR, while positively correlating with clamp-derived glucose disposal rates. T2DM also had lower NRF2 protein expression (−31.6%), which is a positive regulator of GLO1, while Keap1 protein expression, a negative regulator of GLO1, was elevated (207%). Additionally, insulin stimulation during the clamp had a differential effect on NRF2, Keap1, and MG-modified protein expression. These data suggest that dicarbonyl stress and the glyoxalase enzyme system are dysregulated in T2DM skeletal muscle and may underlie skeletal muscle insulin resistance. Whether these phenotypic differences contribute to the development of T2DM warrants further investigation.

INTRODUCTION

The prevalence of Type 2 diabetes (T2DM) is a well-defined public health concern (27) and places a tremendous burden on health care systems (40). Chronic hyperglycemia is the hallmark of T2DM and damages cells and tissues from a multitude of established and yet-to-be defined pathways. One proposed mechanism of hyperglycemia-associated tissue damage is through accumulation of methylglyoxal (MG), a highly reactive α-dicarbonyl (36). MG modifies proteins through covalent interactions, largely targeted at arginine residues, forming a hydroimidazolone adduct (MG-H1), altering the protein’s structure and charge. MG is primarily formed by spontaneous oxidation of the glycolytic intermediates glyceraldehyde-3-phosphate (G3P) and dihydroxyacetone phosphate (DHAP) (9, 17, 52). Thus, cellular glycolytic flux plays a major role in MG formation (31, 34, 59). Under physiological conditions, MG generation is balanced by detoxification through the glyoxalase enzyme system that consists of glyoxalase-1 (GLO1) and glyoxalase-2 (GLO2) (18, 50). GLO1 is under transcriptional regulation by nuclear factor erythroid 2-related factor 2 (NRF2), which is constitutively produced in the cytosol (56). However, NRF2 activity is repressed via Kelch-like ECH-associating protein 1 (Keap1), which binds NRF2 and signals its degradation via the ubiquitin-proteasomal system. When GLO1 protein expression is reduced, the rate of MG formation exceeds the rate of detoxification causing MG-modified proteins to accumulate, a state termed “dicarbonyl stress” (18).

Indeed, plasma and tissue levels of MG-modified proteins are elevated in the diabetic condition, concomitant with reductions in GLO1 protein expression (23, 35) and dysregulation of NRF2 and Keap1 protein expression (1, 24). Additionally, MG-modified proteins are well-documented contributors to diabetic complications, including nephropathy, neuropathy and retinopathy. Further, in mouse L6 skeletal muscle cell culture, dicarbonyl stress induced by exogenous MG treatment, has been shown to inhibit insulin signaling and disrupt mitochondrial function (15, 29, 38). Certainly, the importance of skeletal muscle health in maintaining insulin sensitivity and preventing the onset of T2DM is well known, as skeletal muscle is responsible for >80% of whole body insulin-stimulated glucose disposal (10). Because of this high glycolytic flux, skeletal muscle may be particularly susceptible to reductions in GLO1 protein and subsequent increases in dicarbonyl stress. Notably, a recent clinical trial examining the effects of a pharmacologic GLO1 inducer, in obese, insulin-resistant adults, improved glycemic control comparable to that of metformin (57). In light of these findings, skeletal muscle GLO1 regulation may be critical for whole body glucose metabolism. However, MG-GLO1 biology in human skeletal muscle remains largely undefined, and the effects of T2DM are unknown.

Therefore, we sought to characterize the glyoxalase enzyme system and MG-modified protein profiles, in skeletal muscle biopsies obtained from lean, healthy control (LHC) and T2DM subjects, under basal and insulin-stimulated conditions of a hyperinsulinemic-euglycemic clamp. We hypothesized that basal GLO1 protein expression and activity would be reduced, whereas MG-modified proteins would be increased in T2DM skeletal muscle. Further, we hypothesized that insulin-stimulated glycolytic flux would exacerbate the formation of MG-modified proteins in T2DM. To further elucidate MG-GLO1 biology in human skeletal muscle, additional assays were performed to characterize the protein expression of key regulatory proteins (NRF2 and Keap1), potential sites for MG generation [G3P-dehydrogenase (GAPDH) and triose phosphate isomerase (TPI)], as well as an alternate control point of MG detoxification [aldoketoreductase-1B1 (AKR1B1)].

MATERIALS AND METHODS

Study design.

Our intent was to characterize MG-GLO1 biology in a population of young, LHCs–representing an optimal state of health–and directly contrast these observations to T2DM subjects, thus highlighting the importance of the disease phenotype. Hyperinsulinemic-euglycemic clamps were used to induce experimental hyperinsulinemia and characterize whole body insulin sensitivity. Skeletal muscle needle biopsies were obtained during basal and insulin-stimulated conditions of the clamp procedure for each subject. Skeletal muscle tissue samples were probed for protein expression, activity, and post-translational modifications, as it pertained to MG-GLO1 biology. Subject characteristics from the LHC group, and a larger subset of the T2DM group, have been published previously (25, 55).

Subjects.

Participants were recruited from the Chicago, Illinois, metropolitan area. All subjects were screened via health history, medical exam, resting EKG, and fasting blood chemistry in the Clinical Research Center of the University of Illinois at Chicago. Glucose tolerance was characterized by a 75-g oral glucose tolerance test following a standard fasting period (10–12 h) for all subjects. Individuals were excluded if they used nicotine, had undergone greater than 2-kg weight change in the last 6 mo, or had evidence of hematological, renal, hepatic, or cardiovascular disease. LHCs were excluded when results of OGTT indicated impaired fasting glucose or impaired glucose tolerance. Height and body weight were measured by standard procedures, and body composition was assessed by dual-energy X-ray absorptiometry (iDXA; Lunar, Madison, WI). All studies were approved by the Institutional Review Board of the University of Illinois at Chicago.

Metabolic control.

All subjects were instructed to maintain their regular dietary eating habits and completed 3-day diet records (two week days and one weekend). Before the metabolic testing day, subjects were instructed to abstain from alcohol consumption for 48 h and ingestion of caffeine for 24 h and further refrain from exercise for at least 24 h. On the day preceding metabolic testing, subjects were counseled to consume ~55% of calories as carbohydrate to meet a goal of 250 g (55). At ~6:00 PM on the evening before metabolic testing, participants were provided a balanced metabolic meal (55% carbohydrate, 35% fat, 10% protein: based on their estimated individual nutrient requirements and habitual physical activity levels) (19) to stabilize muscle and liver glycogen stores. After meal consumption, participants fasted overnight for a period of 10–12 h. All participants were asked to withhold medications and supplements on the morning of metabolic testing that were known to influence primary outcome variables. Female participants, who were premenopausal with regular menstrual cycles, were studied in the follicular phase.

Insulin sensitivity.

Whole body insulin sensitivity was assessed using the gold standard hyperinsulinemic (40 mU·m⁻²·min⁻¹)-euglycemic (5 mM) clamp procedure, performed as previously detailed (20, 21, 44, 45, 55). Blood glucose was measured every 5 min via Yellow Springs Instruments (YSI) glucose-lactate analyzer (YSI 2300; STAT Plus, Yellow Springs, OH) and was clamped at 90 mg/dl by use of a variable glucose infusion (20% dextrose) calculated according to the method of DeFronzo et al. (11). Blood samples were collected (5 ml) every 15 min for analysis of plasma glucose and insulin concentrations. Glucose disposal rates (GDR; mg·kg⁻¹·min⁻¹) were calculated as the mean rate obtained during insulin-stimulated conditions using Steele’s single-pool model of glucose kinetics, as previously described (46). Insulin sensitivity is reported as “M/I” and calculated from GDR normalized to steady-state insulin concentrations during the final 30 min of the clamp procedure.

Skeletal muscle biopsy.

Skeletal muscle biopsies (~200 mg) of the vastus lateralis were obtained under local anesthetic (Lidocaine HCl, 1%) (51), using a 5-mm Bergstrom cannula with suction (5, 13) during the baseline (0 min) and insulin-stimulated (120 min) periods of the hyperinsulinemic-euglycemic clamp procedure (25, 55). Infusion of steady-state insulin and glucose was not terminated until completion of the muscle biopsy procedure. Muscle tissue was blotted and trimmed of adipose and connective tissue, if necessary, and immediately (less than 90 s from extraction) flash frozen in liquid nitrogen and subsequently stored at −80°C until future processing and analysis.

Immunoblotting.

Approximately 10–15 mg (wet weight) of frozen muscle tissue was homogenized by ceramic beads (lysing matrix D; FastPrep-24 homogenizer; MP Biomedicals, Santa Ana, CA) in 20 volumes of ice-cold buffer consisting of 20 mM Tris·HCl (pH 7.5), 150 mM NaCl, 1 mM Na₂ EDTA, 1 mM EGTA, 2.5 mM Na pyrophosphate, 1 mM β-glycerophosphate, 1 mM Na₃VO₄, 1% Triton, and 1 μg/ml leupeptin (Cell Signaling Technology, Beverly, MA) with an added protease and phosphatase inhibitor cocktail (MS-SAFE; Sigma Aldrich, St. Louis, MO). Total protein concentration was determined via BCA protein assay (Pierce Biotechnology, Rockford, IL). Proteins (10–20 µg) were separated via 10% SDS-PAGE and transferred to either PVDF or nitrocellulose membranes followed by blocking with 3–5% nonfat dry milk or BSA in TBST. Conditions were optimized individually for each protein of interest. Immunoblotting proceeded with the antibodies as follows: GLO1 (sc133144; Santa Cruz Biotechnologies, Santa Cruz, CA), anti-methylglyoxal (STA-011; Cell Biolabs, San Diego, CA), NRF2 (ab62352; Abcam, Cambridge, MA), Keap1 (A1820; NeoScientific, Cambridge, MA), TPI (sc30145; Santa Cruz Biotechnologies), GLO2 (sc365233; Santa Cruz Biotechnologies), AKR1B1 (GTX113381; GeneTex, Irvine, CA), and/or GAPDH (2118S; Cell Signaling Technology)/β-actin (612657; BD Biosciences, San Jose, CA). Secondary antibodies (LI-COR Biosciences, Lincoln, NE) were selected in a species-specific manner, according to manufacturer’s instructions. Protein expression was visualized via near-infrared fluorescence imaging (Odyssey Clx, LI-COR Biosciences), quantified via Image Studio, and normalized to GAPDH or β-actin protein expression as appropriate. MG-modified proteins were quantified using the full range of detected bands between 15 and 260 kDa. This approach is similar to the methods reported by others to assess MG modification of cellular proteins with minor modifications for human skeletal muscle tissue (39, 58). As additional validation that our immunoblotting approach was accurate and sensitive in detecting MG-modified proteins, we performed a titration curve with known quantities of MG-modified BSA (MG-modified BSA, STA-306; Cell Biolabs). The antimethylglyoxal antibody, which recognizes MG-HI residues, was able to detect MG-modified BSA as low as 0.625 µg with a linear response within the range of protein quantities examined, which is expected to encompass physiologically relevant quantities present in human skeletal muscle tissue (Fig. 1).

Fig. 1.

Validation of quantification technique for methylglyoxal (MG)-modified proteins. A: representative Western blot image. B: MG-modified proteins quantified by our methods using an anti-MG monoclonal antibody and LI-COR immunofluorescent secondary antibody (fully described in materials and methods) are increased in a linear relationship with quantities of MG-modified BSA standard.

GLO1 enzymatic activity.

GLO1 enzymatic activity (QuantiChrom glyoxalase I assay kit, BioAssay Systems, Hayward, CA) was determined via conversion of S-lactoylglutathione (Δ in molar extinction coefficient: Δε240 = 3.37 mM/cm) after incubating skeletal muscle homogenates (10 µg) with MG and reduced glutathione, where 1 unit of GLO1 forms 1 μmol of S-lactoylglutathione per minute at pH 6.6 and 25°C (26). Optical density of the samples was quantified via photospectrometry at 240 nm, and data are expressed as U/l of S-lactoylglutathione.

Protein carbonyl assay.

Protein carbonyls are the most common form of protein modification by glycation, oxidative stress, or other oxidative by-products and consist of carbonyl derivatives of arginine, but also lysine, threonine, and proline residues. Protein carbonyl content of skeletal muscle was analyzed by a commercially available kit (Oxiselect, Cell Biolabs), according to the manufacturer’s instructions. Briefly, skeletal muscle homogenate was diluted to 10 µg/ml and loaded into a precoated 96-well plate for 2 h at 37°C. Following the manufacturer’s instructions, we quantified carbonyl content by derivatization (of all protein carbonyls present within the skeletal muscle homogenate) with DNP hydrazone, followed by an anti-DNP primary antibody, a horseradish peroxidase-conjugated secondary antibody, and microplate quantification of absorbance at 450 nm. Comparison to a standard curve of oxidized BSA allows for estimation of carbonyl content reported as nanomoles of protein carbonyls/mg total protein.

Statistics.

Statistical analysis was performed using PRISM (GraphPad Software, La Jolla, CA). Baseline differences in variables of interest were compared using an independent-samples t-test. A two-way [(Group (LHC vs. T2DM) × Condition (basal vs. insulin)] ANOVA with repeated measures for “condition” was used to investigate protein expression, carbonyl stress, and enzymatic activity and Bonferroni post hoc adjustments were performed as necessary. Pearson’s correlation coefficient was used to investigate relationships among GLO1 protein expression and markers of metabolic health. Linear regression models were developed using SPSS (SPSS, Chicago, IL). Significance was set at P < 0.05. Data are presented as means ± SE.

RESULTS

Baseline characteristics.

Baseline characteristics of the study population are presented in Table 1. By design, T2DM subjects were older and more obese than LHC. Additionally, T2DM subjects were insulin resistant and glucose intolerant, according to HOMA-IR, clamp-derived insulin sensitivity (M/I) and 2 h OGTT glucose values.

Table 1.

Subject characteristics

	LHC	T2DM
n (M, F)	10 (4,6)	5 (2,3)
Age, yr	27 ± 1	56 ± 5*
BMI, kg/m2	22.3 ± 1.1	32.3 ± 1.6*
Body fat, %	24.1 ± 1.8	40.7 ± 3.0*
Fat-free mass, kg	48.8 ± 3.1	59.4 ± 4.0*
Total cholesterol, mg/dl	154 ± 7	151 ± 6
HDL cholesterol, mg/dl	60 ± 5	52 ± 4
LDL cholesterol, mg/dl	80 ± 11	81 ± 7
Triglycerides, mg/dl	75 ± 7	86 ± 14
Fasting glucose, mg/dl	90 ± 4	151 ± 24*
HbA1c, %	5.3 ± 0.1	7.7 ± 1.0*
Fasting insulin, µU/ml	5.6 ± 0.6	7.8 ± 1.3
HOMA-IR, AU	1.0 ± 0.1	2.2 ± 0.4*
2-h OGTT glucose, mg/dl	101 ± 4	262 ± 60*
OGTT glucose iAUC, mg·dl−1·2 h−1	2,884 ± 306	11,682 ± 1,059
OGTT insulin AUC, µU·ml−1·2 h−1	3,748 ± 307	2,676 ± 948
Matsuda Index, AU	8.3 ± 1.0	4.5 ± 0.7
GDR, mg kg−1 min−1	7.1 ± 0.5	2.7 ± 0.5*
Clamp insulin, µU/ml	5.3 ± 1.0	7.3 ± 1.4
M/I, mg kg−1 min−1·µU−1·ml	0.11 ± 0.01	0.03 ± 0.01*

Data are presented means ± SE. LHC, Lean healthy control subjects; T2DM, individuals with Type 2 diabetes mellitus; BMI, body mass index; HOMA-IR, homeostatic model assessment-insulin resistance; OGTT, 75-g oral glucose tolerance test; GDR, glucose disposal rate as calculated from the hyperinsulinemic-euglycemic clamp; M/I, clamp-derived glucose disposal rate normalized to prevailing insulin concentration.

^*Significantly different from LHC, P < 0.05.

GLO1/2 protein expression and enzymatic activity.

Skeletal muscle basal GLO1 protein expression was markedly reduced by −78.8% in T2DM compared with LHC (Fig. 2, A and B) (LHC: basal, 6327 ± 665 AU; insulin, 6681 ± 886 AU; T2DM: basal, 1340 ± 308 AU; insulin, 1714 ± 353 AU; Group, P < 0.001; Condition, P = 0.515; Interaction, P = 0.986), where GLO2 remained similar (Fig. 2, C and D) (LHC: basal, 0.441 ± 0.049 AU; insulin, 0.428 ± 0.049 AU; T2DM: basal, 0.517 ± 0.76 AU; insulin, 0.461 ± 0.064 AU; Group, P = 0.523; Condition, P = 0.088; Interaction, P = 0.264). There was no effect of insulin stimulation on GLO1 or GLO2 protein expression (Fig. 2, B and D). Because the T2DM group was significantly older and more obese than the LHC group and because chronologic age and obesity have previously been associated with reductions in GLO1 protein expression, we generated linear regression models to investigate the effects of T2DM status on GLO1 protein expression controlling for age and BMI. Models are presented in Table 2. Model 1 used basal GLO1 protein expression (AU) as the dependent variable and age and BMI as independent variables and expectedly revealed a strong R value of 0.697. However, the addition of group (T2DM) to the model (model 2) markedly increased the R value to 0.845 with a significant effect of group only, indicating T2DM as a strong predictor of reduced GLO1 protein expression in our participants. Correlational analysis was used to investigate potential relationships between skeletal muscle basal GLO1 protein expression and markers of obesity and metabolic health. Basal skeletal muscle GLO1 protein expression was inversely correlated with BMI (Fig. 3A), body fat percentage (Fig. 3B), and HOMA-IR (Fig. 3C), while GLO1 positively correlated with skeletal muscle insulin sensitivity (M/I) measured by the hyperinsulinemic-euglycemic clamp (Fig. 3D). To further characterize the glyoxalase enzyme system, a commercially available GLO1 enzymatic activity kit was modified for application in human skeletal muscle. GLO1 enzymatic activity was not different between groups or conditions (LHC: basal, 150.3 ± 6.0 U/l; insulin, 162.5 ± 3.4 U/l; T2DM: basal, 159.6 ± 5.7 U/l; insulin, 146.0 ± 4.9 U/l; Group, P = 0.172; Condition, P = 0.907; Interaction P = 0.150)

Fig. 2.

Skeletal muscle GLO1 protein expression. A: representative Western blot image for GLO1. GAPDH used as a loading control. B: GLO1 protein expression is reduced in skeletal muscle of T2DM during basal and insulin-stimulated states of the hyperinsulinemic-euglycemic clamp. *Significant effect of group, P < 0.001. C: representative Western blot image for GLO2. GAPDH used as loading control. D: GLO2 protein expression was not different with Condition or Group (fully detailed in results). LHC, n = 10; T2DM, n = 5. B, Basal, t = 0 min; I, Insulin, t = 120 min.

Table 2.

Summary of regression analysis for variables predicting basal GLO1 protein expression

	Model 1				Model 2
Variable	B	SE B	β	Significance	B	SE B	β	Significance
Age	−45.5	66.2	−0.239	0.505	93.8	69.8	0.492	0.206
BMI	−258.4	182.8	−0.491	0.183	9.6	168.8	0.018	0.956
Group					−7,780	2,629	−1.273	0.013*
R2		0.486				0.636

Linear regression models to determine the effect of T2DM (group) on the primary outcome measure (skeletal muscle basal GLO1 protein expression). B, unstandardized coefficients; SE B, standard error of B; β, standardized coefficients; R², coefficient of determination.

^*Significant predictor of basal GLO1 protein expression.

Fig. 3.

Baseline GLO1 protein correlations. Correlations observed between skeletal muscle GLO1 protein and BMI (A), body fat percentage (B), and fasting insulin resistance (HOMA-IR; C). A positive correlation was observed between basal GLO1 protein expression and peripheral insulin sensitivity determined by hyperinsulinemic-euglycemic clamp (M/I; D).

MG-modified proteins and carbonyl stress.

During basal conditions, total skeletal muscle MG-modified proteins were similar between LHC and T2DM. However, there was a significant effect of condition (basal vs. insulin; P < 0.001) and an interaction effect [Group (LHC vs. T2DM) × Condition (basal vs. insulin); P < 0.001], whereby insulin stimulation increased MG-modified proteins only in T2DM skeletal muscle (Fig. 4; Bonferroni’s post hoc comparison, P < 0.001), implicating an acute increase in dicarbonyl stress. A global carbonyl stress assay was also performed to assess oxidative and carbonyl modifications of skeletal muscle proteins. This analysis includes MG modification of proteins, but also quantifies oxidative damage, which provides an indicator of overall cellular oxidative stress. Under basal conditions, skeletal muscle of individuals with T2DM presented with higher levels of carbonyl stress (Fig. 5A) (LHC: basal, 3.6 ± 0.2 nmol/mg protein; insulin, 3.7 ± 0.1 nmol/mg protein; T2DM: basal, 4.2 ± 0.2 nmol/mg protein; insulin, 4.1 ± 0.1 nmol/mg protein; Group, P = 0.026; Condition, P = 0.987; Interaction, P = 0.627). Unlike MG-modified proteins, carbonyl stress was not affected by insulin stimulation. Additional correlational analysis revealed an inverse correlation between basal carbonyl stress and basal skeletal muscle GLO1 protein expression (Fig. 5B; r = 0.517, P = 0.049).

Fig. 4.

Skeletal muscle MG-modified proteins. A: MG-modified proteins are increased in T2DM with insulin stimulation during a hyperinsulinemic-euglycemic clamp. LHC, n = 10; T2DM, n = 5. *Significant effect of Condition, P < 0.001 and Interaction (Group × Condition), P < 0.001. Bonferroni post hoc comparison revealed a significant effect of insulin in T2DM. #P < 0.001. B: representative Western blot image. B, basal, t = 0 min; I, insulin, t = 120 min.

Fig. 5.

Skeletal muscle carbonyl stress. A: carbonyl stress was elevated in the skeletal muscle of T2DM; *Significant effect of Group, P = 0.026. B: an inverse correlation was observed between basal carbonyl stress and basal GLO1 protein expression (r = −0.517, P = 0.049).

NRF2 and Keap1 protein expression.

To investigate whether proteins regulating GLO1 expression were altered in T2DM skeletal muscle in the basal and metabolic state of the hyperinsulinemic-euglycemic clamp, we performed Western blot analysis as previously described for NRF2 and Keap1. In agreement with reduced GLO1 protein expression in T2DM, NRF2 was 31.2% lower in T2DM compared with LHC (Fig. 6, A and B) (LHC: basal, 0.0037 ± 0.0005 AU; insulin, 0.0031 ± 0.0005 AU; T2DM: basal, 0.0018 ± 0.0002 AU; insulin, 0.0022 ± 0.0022 AU; Group, P = 0.095; Condition, P = 0.601; Interaction, P = 0.009). Of note, there was an interaction effect of Group (LHC vs. T2DM) × Condition (basal vs. insulin), where skeletal muscle NRF2 decreased in LHC, but increased in T2DM with insulin stimulation. Additionally, Bonferroni’s post hoc comparison revealed a significant effect of insulin in LHC (P < 0.05), but not in T2DM. In agreement with reduced GLO1 and NRF2 protein expression in T2DM skeletal muscle, Keap1 protein was more than twofold higher in T2DM compared with LHC (Fig. 6, C and D) (LHC: basal, 0.054 ± 0.010 AU; insulin, 0.074 ± 0.015 AU; T2DM: basal, 0.113 ± 0.015 AU; insulin, 0.059 ± 0.014 AU; Group, P = 0.238; Condition, P = 0.156; Interaction, P = 0.007). In agreement with the NRF2 results, there was also a significant interaction effect of Group (LHC vs. T2DM) × Condition (basal vs. insulin) on Keap1 protein expression where LHC increased, but T2DM decreased with hyperinsulinemia. Additionally, Bonferroni’s post hoc comparison revealed a significant effect of insulin only in T2DM (#P < 0.05).

Fig. 6.

Skeletal muscle NRF2 and Keap1 protein expression. A: representative Western blot image for NRF2. B: interaction effect of Condition × Group on NRF2 protein expression. LHC, n = 10; T2DM, n = 5; *P = 0.009. Bonferroni’s post hoc comparison revealed a significant effect of insulin in LHC. #P < 0.05. C: representative Western blot image for Keap1. D: interaction effect of Condition × Group on Keap1 protein expression. LHC, n = 10; T2DM, n = 5; *P = 0.007. Bonferroni post hoc comparison revealed a significant effect of insulin in T2DM. #P < 0.05. B, basal, t = 0 min; I, insulin, t = 120 min.

GAPDH, TPI, and AKR1BI protein expression.

To investigate potential mechanisms of MG generation, we quantified protein expression of the glycolytic enzymes, GAPDH and TPI, in skeletal muscle at basal and insulin-stimulated states of the hyperinsulinemic-euglycemic clamp. Dysregulation of these proteins causes triose phosphate intermediates to accumulate and increases MG generation in the context of in vitro T2DM models. There was no difference in GAPDH protein expression (Fig. 7, A and B) when normalized to β-actin (LHC: basal, 1.00 ± 0.12 AU; insulin, 0.96 ± 0.10 AU; T2DM: basal, 0.84 ± 0.09 AU; insulin, 0.87 ± 0.11 AU; Group, P = 0.346; Condition, P = 0.980; Interaction, P = 0.767). This provides additional validation of GAPDH as an adequate loading control in our T2DM samples. Similarly, TP1 protein expression (Fig. 7, C and D) remained unchanged (LHC: basal 0.202 ± 0.020 AU, insulin 0.226 ± 0.022 AU; T2DM: basal 0.249 ± 0.016 AU, insulin 0.266 ± 0.029 AU; Group, P = 0.207; Condition, P = 0.065; Interaction, P = 0.766).

Fig. 7.

Skeletal muscle GAPDH, TPI, and AKR1B1 protein expression. A: representative Western blot image for GAPDH. B: no effect of Condition or Group on GAPDH protein expression (fully detailed in results). C: representative Western blot image for TPI. D: no effect of Condition or Group on TPI protein expression (fully detailed in results). A trend for increased TPI was observed with insulin stimulation (Condition), P = 0.067. E: representative Western blot image for AKR1B1. F: no effect of Condition or Group on AKR1B1 protein expression (fully detailed in results); LHC, n = 10; T2DM, n = 5. B, basal, t = 0 min; I, insulin, t = 120 min.

To investigate potential mechanisms regulating MG detoxification, we quantified AKR1B1 protein expression in skeletal muscle at basal and insulin-stimulated states of the hyperinsulinemic-euglycemic clamp. AKR1B1 is the predominant aldoketoreductase isoform in human skeletal muscle (14) and plays only a minor role in MG detoxification under normal conditions (42, 48, 53). However, preclinical models show that MG detoxification by AKR1B1 is increased when GLO1 protein expression is reduced (3). There was no effect of Group or Condition on AKR1B1 protein expression (Fig. 7, E and F) (LHC: basal, 1.32 ± 0.20 AU; insulin, 0.99 ± 0.1 AU; T2DM: basal, 0.91 ± 0.14 AU; insulin, 0.85 ± 0.09 AU; Group, P = 0.157; Condition, P = 0.261; Interaction, P = 0.444).

DISCUSSION

Here, we report that GLO1 protein expression is markedly reduced in the skeletal muscle of individuals with T2DM compared with healthy control subjects. While previous research has demonstrated a reduction in GLO1 protein, expression in insulin-independent tissues in the context of T2DM, to our knowledge, this is the first report of reduced GLO1 protein expression in the highly metabolic skeletal muscle tissue. This is particularly exciting given the recent findings by Xue et al. (57) that demonstrated GLO1 inducer therapy improved whole body glycemic control in insulin-resistant individuals comparable to the first line T2DM medication, metformin. In this lens, it is likely that GLO1 inducer therapy acted upon metabolic tissue (skeletal muscle, adipose, or liver), which is known to play a role in whole body glucose homeostasis. Furthermore, the inverse correlations between skeletal muscle GLO1 protein expression and multiple indices of obesity and insulin resistance provide strong incentive to investigate potential therapeutic mechanisms of normalizing skeletal muscle GLO1 protein expression.

The primary consequence of reduced GLO1 protein expression is an increase in dicarbonyl stress; thus, we sought to characterize MG-specific protein modifications in the skeletal muscle of LHC and T2DM individuals. We performed immunoblotting analysis of MG-modified proteins using an antibody specific to MG-H1 modifications in skeletal muscle of LHC and T2DM in the basal and insulin-stimulated conditions of the hyperinsulinemic-euglycemic clamp procedure. The observed increase in MG-modified proteins with insulin stimulation in the T2DM group may signify an acute state of dicarbonyl stress. It is plausible that acute dicarbonyl stress resulted from a combination of an increased rate of MG generation (through insulin-stimulated glycolytic flux) and a diminished rate of MG-detoxification (due to reduced GLO1 protein expression) in the skeletal muscle of individuals with T2DM. To better characterize dicarbonyl stress in T2DM skeletal muscle, we used a global carbonyl assay, which quantifies oxidative protein modification, as well as MG-specific modifications, as a heterogeneous marker of cellular stress. Basal levels of carbonyl stress were higher in T2DM skeletal muscle compared with the LHC group and produced an inverse correlation when compared with basal GLO1 protein expression. Taken together, these data describe a distinctive phenotype of MG-GLO1 biology in the skeletal muscle of T2DM.

Although further research is required to understand the implications of a dysregulated MG-GLO1 axis in relation to insulin resistance and T2DM, reductions in GLO1 protein are well documented to be the primary cause of exacerbated dicarbonyl stress in insulin-independent tissues, and evidence from in vitro and animal models (12, 38) indicates a potential causative role of dicarbonyl stress in the development of skeletal muscle insulin resistance. Within the perspective of available literature and recent advances in GLO1 inducer therapy, these results highlight the need for additional research to determine whether therapeutic enhancement of GLO1 protein expression is protective to dicarbonyl stress in skeletal muscle in vivo. The potential benefit that this may confer to the prevention and treatment of the T2DM condition has yet to be determined.

To identify potential mechanisms underlying the observed reduction in skeletal muscle GLO1 protein expression in T2DM, we analyzed two primary GLO1 regulatory proteins, NRF2 and Keap1, at basal and insulin-stimulated states of the hyperinsulinemic-euglycemic clamp. NRF2 and Keap1 have a well-defined inverse relationship, whereby NRF2 is constitutively produced, but degraded in the cytosol by a Keap1-dependent mechanism. Keap1, itself, is transiently degraded with increased oxidative stress, allowing NRF2 protein to accumulate and activate the cell’s antioxidant response, which includes GLO1 protein transcription. Expectedly, we show an inverse relationship between NRF2 and Keap1 protein expression at each sample collection. Both NRF2 and Keap1 protein expression were different between LHC and T2DM in the basal and insulin-stimulated states of the clamp, implicating dynamic differential regulation of these proteins in skeletal muscle of individuals with T2DM compared with LHC. In agreement with basal reductions in GLO1 protein expression, T2DM skeletal muscle presented with reduced NRF2 protein expression and increased Keap1 protein expression. Although the role of these regulatory proteins have been studied extensively (56), to our knowledge, we are the first to demonstrate basal reductions of NRF2 in the skeletal muscle of T2DM. The literature supports a potential mechanistic explanation as insulin-stimulated activation of the PI3K/Akt pathway has been previously shown to regulate NRF2-dependent pathways (28). Given that T2DM skeletal muscle is characterized by insulin resistance and a dysregulation of PI3K/Akt signaling, our data are in agreement with other preclinical models (54). Future research should continue to investigate the role NRF2/Keap1 regulation on skeletal muscle GLO1 protein regulation, but also on its independent potential role in insulin resistance pathology (7). Although, not classically considered cross-validation, the consistency of our immunoblotting results across the entire MG-GLO1-NRF2-Keap1 axis, agreement with known regulatory molecular mechanisms, and alignment with available literature, provides cogent evidence to the dysregulation of this entire regulatory system in human skeletal muscle.

We further examined alternate pathways of MG production and detoxification to better characterize the MG-GLO1 axis in T2DM skeletal muscle. Reduced activity of GAPDH and TPI has been documented to cause a backlog of glycolysis in T2DM, generating an accumulation of the triose phosphate intermediates G3P and DHAP, resulting in greater spontaneous generation of MG. However, no differences in GAPDH or TPI protein expression were observed between LHC and T2DM. It is possible that our cohort was underpowered and unable to realize any true difference in GAPDH or TPI protein expression. However, it is more likely that performing enzymatic activity assays for GAPDH and TPI would have been more appropriate, as both oxidative stressors inhibit GAPDH and TPI enzymatic activity without affecting total protein expression (9, 16). In addition, to better characterize alternative regulators of MG detoxification, we investigated GLO2 and AKR1B1 protein expression. In line with previous research, skeletal muscle GLO2 protein expression remained unchanged with T2DM status or insulin stimulation. An alternate pathway of MG detoxification occurs via the aldoketoreductase AKR1B1. In healthy physiology, AKR1B1 plays a minor role of MG detoxification, as GLO1 has a much higher affinity for MG. However, in conditions when GLO1 is insufficient to prevent accumulation of MG, such as observed in our T2DM group, AKR1B1 compensates with an increase in MG detoxification capacity. We observed no compensatory increases in AKR1B1 protein expression in T2DM skeletal muscle; however, we did not examine enzyme activity.

Given the dramatic reduction in GLO1 protein expression in T2DM skeletal muscle, we performed GLO1 activity assays where we hypothesized a parallel finding to GLO1 protein. However, contrary to our hypothesis, GLO1 enzymatic activity was not different in LHC compared with T2DM. It may be that skeletal muscle GLO1 enzymatic activity is upregulated to account for the loss of total enzyme content, as this phenomenon has been similarly observed in red blood cells of T2DM (49) and, additionally, in immune cells of individuals with diabetes. There also remains the potential for confounding factors present in the skeletal muscle homogenate of LHC or T2DM to alter the GLO1 enzyme kinetics or assay output. For example, the GLO1 activity assay involves provision of highly reactive substrates (MG and reduced glutathione), which could reasonably react with other constituents within the skeletal muscle homogenate. Further, other components within the skeletal muscle homogenate may allosterically inhibit or enhance glyoxalase system enzyme activity. Therefore, future work should continue to investigate the importance of GLO1 protein expression and enzyme activity in the context of T2DM and probe for potential confounding factors within skeletal muscle. Alternatively, our experimental conditions may not have been appropriately optimized for human skeletal muscle.

These results should be interpreted with our study limitations in mind. The sample size of our T2DM group was modest (n = 5); however, the data were normally distributed and achieved statistical significance through the stringent two-way repeated-measures ANOVA. By design, our T2DM subjects were older and more obese than the LHC group, which prevents our ability to draw strict conclusions on the effect of T2DM per se on the MG-GLO1-NRF2-Keap1 axis. To address these issues, we generated linear regression models controlling for both age and BMI, evidencing the independent effect of T2DM in our cohort. Still, future studies should use larger, well-matched cohorts to elucidate independent effects of diabetes, obesity, and chronologic age on GLO1 biology. Our interpretation of MG-directed protein modifications may also be limited by our approach, which used immunoblotting as a semiquantitative measure of MG-directed protein modification. This approach is similar to previous reports (6, 8, 31); however, we supplemented our assay with a titration curve of MG-modified BSA to verify a working range of sensitivity. Further, the MG-modified protein immunoblotting results were cross-validated using a chemical assay for total carbonyl protein modifications. Certainly, targeted mass spectrometry is the most sensitive and accurate method for quantifying MG-directed protein modifications (32, 33, 41), and future studies will be needed to verify our results.

Perspectives and Significance

To our knowledge, this study represents the first characterization of the MG-GLO1 axis in human skeletal muscle. Importantly, these data provide a link between dysregulation of the MG-GLO1 axis and skeletal muscle metabolic health, which is in agreement with previous in vitro (2, 4, 22, 47) and human (30, 37, 43) studies. In the context of the emerging evidence surrounding GLO1-inducer therapy for metabolic disease, these results support a mechanistic role within skeletal muscle. Future investigations are warranted to determine the therapeutic potential of normalizing the MG-GLO1 axis.

GRANTS

This work was supported by the American Diabetes Association Grant 1-14-JF-32 (to J. M. Haus), National Institutes of Health Grants R01-DK-109948 and UL1-RR-029879 (to J. M. Haus), and Sigma Xi Research Society Grant-in-Aid (to J. T. Mey).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

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