A single bout of cycling exercise induces nucleosome repositioning in skeletal muscle of lean and obese individuals

Abstract

Acute aerobic exercise improves, whereas being overweight impairs, skeletal muscle mitochondrial function in correlation with reduced insulin sensitivity. Here, the molecular and metabolic effects of a single exercise bout in the skeletal muscle was compared in men, aged 19-30 years, that were either lean (BMI<25, 18.5-24.1 kg/m², n=15) or overweight/obese (Ov/Ob; BMI≥25, 25.5-36.9 kg/m², n=15). Four hours after a high-carbohydrate breakfast (7 kcal/kg; 60% carbohydrate, 25% fat, 15% protein), participants performed cycling exercise (50% VO₂max, expending ~650 kcal). Muscle biopsies and peripheral blood samples were collected 30 minutes before the meal and immediately after exercise. Muscle long-chain acylcarnitines were increased in Ov/Ob compared to lean, with or without exercise. A single exercise bout increased mRNA abundance of genes related to mitochondria and insulin signaling in both lean and Ov/Ob. Nucleosome mapping by micrococcal nuclease digestion with deep sequencing (MNase-seq) revealed that exercise repositioned the −1N nucleosome away from the transcription start site of the PGC1a promoter and of other mitochondrial genes, but did not affect genes related to insulin signaling, in both lean and Ov/Ob. In conclusion, these data suggest that a single exercise bout induced epigenetic alterations in skeletal muscle in a BMI-independent manner.

Introduction:

The prevalence of obesity is ~36% among adults and 17% among young individuals in the United States [1]. Approximately 30 million people in the United States suffer from diabetes, the majority diagnosed with type 2 diabetes, and 84 million people are pre-diabetic [2]. Obesity is associated with reduced skeletal muscle mitochondrial abundance, size, and efficiency [3, 4]. Correspondingly, altered fatty acid oxidation in skeletal muscle is indicated with obesogenic diets, increased BMI, and type 2 diabetes and has been implicated in the development of insulin resistance [3, 5–9]. Altered fatty acid oxidation is often termed incomplete beta-oxidation and is revealed by abnormal acylcarnitine accumulation which signifies either a blockage in the fatty acid oxidation pathway or byproduct accumulation due to increased fatty acid oxidative flux. However, our recent work has cast doubt on a causal link between muscle long-chain acylcarnitine accumulation and insulin resistance [10]. Additional alternative mechanisms that may link obesogenic diets and diabetes to mitochondrial function include the obesity-related downregulation of transcription factors that drive nuclear encoded mitochondrial genes (NEMG), such as nuclear respiratory factor-1 (NRF-1) and peroxisome proliferator-activated receptor gamma coactivator 1 alpha (PGC-1α) [11, 12]. Conversely, exercise is an effector of both oxidative metabolic flux and an activator of transcription factors such as PGC-1α [13–16]. Exercise-induced effects on gene control are predicted to play an essential role in adaptations that enhance mitochondrial number and function as well as insulin sensitivity Therefore, we questioned if the transcriptional repression of genes, such as PGC-1α, that occurs in obesity would interfere with exercise-induced transcriptional activity.

Transcriptional activity is controlled, in part, by nucleosome modifications that influence nucleosome positioning. Nucleosomes provide a scaffold for DNA condensation through the formation of heterochromatin, from which transcription factors are unable to bind to promoter regions for transcription initiation. During transcription, DNA is post-translationally modified, via epigenetic modifications in various manners, to alter the binding affinity of the DNA to the histone proteins that compromise the nucleosome. Thus, heterochromatin is remodeled into euchromatin which allows transcription factors and pol II to bind for transcription initiation and elongation. The nucleosome directly upstream of the transcriptional start site (TSS) is termed the −1 nucleosome (−1N). Thus, the presence or absence of this nucleosome at the TSS is important in determining whether transcription of the downstream gene is permissible. Analyzing nucleosome positioning allows the determination of epigenetically regulated genes unbiasedly as the combined effect of all epigenetic modifications [17].

Exercise is known to epigenetically induce gene expression of NEMGs including PGC-1α through DNA hypomethylation in the promoter region [18]. Conversely, type 2 diabetic subjects have hypermethylated PGC-1α promoter leading to lower PGC-1α expression, as well as lower mitochondrial content [19, 20]. Therefore, we investigated whether a single exercise bout could differentially affect nucleosome repositioning in the promoter region of PGC-1α, and of insulin resistance-related genes, in lean compared to obese subjects. Our data suggest that exercise in both overweight/obese and lean individuals results in similar epigenetic regulation of genes related to mitochondrial function, thus exercise has beneficial effects on muscle gene control in a BMI-independent manner.

Materials and Methods

Participants:

Fifteen participants with BMI<25 were recruited into the lean group and fifteen with BMI≥25 in the Ov/Ob group. The eligibility criteria were: healthy males, ages 18-30 years, completion of a physical examination from a State Licensed M.D. or D.O. within the past 2 years and completion of a physical activity readiness questionnaire (PARQ) [21], the International Physical Activity Questionnaires (IPAQ) [22] and a medical history questionnaire. The study procedures, consent forms and other study-related documents were approved by the Purdue University Institutional Review Board, IRB 1407015074. Participants were recruited by word of mouth and flyers. Exclusion criteria included: not meeting the BMI requirements, unwillingness to have a physical examination completed within the timeframe, and scheduling conflicts. To address possible complications during the muscle biopsy procedure, participants completed a questionnaire regarding allergies to glue, anesthetics or iodine and current medications.

Visit 1:

Participants were instructed to refrain from exercising on the day of testing and reported to the Purdue University Nutrition Science Bionutrition Core (PUNSBC) between 0700-1000 h. Height, weight, waist circumference, hip circumference, resting heart rate and blood pressure were measured. Body composition was assessed by dual-energy X-ray absorptiometry (DXA; GE Lunar iDXA). A VO₂ max test was performed on a stationary cycle ergometer using a protocol with the TrueOne 2400 Metabolic Measurement System (Parvo Medics), as previously described [23].

Visit 2:

Participants were instructed to refrain from vigorous physical activity for 24 h prior to visit 2. Following an overnight fast, participants reported to the PUNSBC between 0700 and 1000 h where they sat quietly for 15 min prior to antecubital blood and quadriceps muscle biopsy sampling (Pre-samples). Blood samples were collected into EDTA vacutainers (BD-Pharmigen) by a phlebotomist from the antecubital vein and plasma was obtained, flash frozen in liquid nitrogen and stored at −80°C until further analysis. Skeletal muscle biopsies were obtained from the vastus lateralis muscle using the percutaneous muscle biopsy technique, as previously described [24]. Briefly, 2.5 mL of a 1% lidocaine solution was injected into the thigh and a 6 mm Bergstrom biopsy needle (Pelomi Medical, Denmark) under suction was used collect the muscle sample, which was quickly rinsed in PBS, blotted, flash frozen in liquid nitrogen and stored at −80°C until further processing. Participants then consumed a standardized, high carbohydrate meal, calories consumed based on body weight (7 kcal/kg; 60% carbohydrate, 25% fat, 15% protein) under the supervision of a registered dietician to ensure adherence. Following the meal, participants were asked to refrain from eating or drinking, except for water, and to refrain from physical activity for the next 4 h, at which time they reported back to the laboratory for the exercise testing trial. The exercise testing trial was completed with each participant expending a total of 650 kcal. Immediately following the exercise testing trial, a second antecubital blood and quadriceps muscle biopsy (Post-samples) were obtained from each participant.

Plasma assays:

Commercially available assay kits were used to measure plasma glucose (10009582, Cayman Chemical Company), triglycerides (10010303, Cayman Chemical Company), insulin (KAQ1251, Invitrogen), cholesterol (STA-384, Cell Biolabs), HDL-C, LDL-C and VLDL-C (STA-391, Cell Biolabs), and free fatty acids (STA-618, Cell Biolabs) per the manufacturer’s protocols.

Metabolomics of acylcarnitines:

Acylcarnitine profiling was performed by the Metabolite Profiling Facility at Purdue University as previously described [25]. Briefly, approximately 45 mg of skeletal muscle were spiked with internal standards and homogenized in 1.5 mL cold acetonitrile to extract the acylcarnitines. Samples were centrifuged at 12,000 x g for 5 min at 4°C, and the resulting supernatant was analyzed using tandem MS via the 6400 Series Triple Quadrupole LC/MS System (Agilent) to measure levels of 38 acylcarnitines species ranging from 0 to 22 carbons in length.

RNA-seq:

Raw and analyzed RNA-seq data can be found in NIH GEO, accession GSE108643. Skeletal muscles were homogenized in Trizol reagent (Thermo Fisher Scientific) per the manufacturer’s protocol, and column purified (Qiagen). RNA-seq was performed by the Purdue University Genomics Core Facility. cDNA libraries were constructed using the TruSeq stranded mRNA library prep kit (Illumina), and sequenced on the Illumina HiSeq 2500 Ultra-High-Throughput Sequencing System. Reads were aligned to the human (hg38) genome using Tophat 2 [26], allowing up to 2 nucleotide mismatches, ~97% of sequenced reads aligned to genome, median FPKM was 0.2 for 56,891 transcripts, which represented 26,376 unique genes.

MNase-seq:

Raw and annotated MNase-seq data can be found in NIH GEO, accession GSE108642. Mononucleosomal DNA was isolated from ~10 mg of skeletal muscle (N=15 per group) as described [20]. Briefly, samples were dounce homogenized in sucrose buffer (0.25 M sucrose, 10 mM Tris-acetate pH 8.1, 1 mM EDTA, 1 mM DTT, 1 mM sodium orthovanadate, and 1X complete protease inhibitor tablet (Roche)). Centrifuged pellets were resuspended in sucrose buffer containing 0.1 mM CaCl2 and 4 mM MgCl2 and incubated with micrococcal nuclease (Roche) for 15 min at 37°C, 5 mM EDTA was added. After centrifugation, pellets were suspended in 0.25 M lysis buffer (50 mM Tris, pH 8.1, 10 mM EDTA, 1% SDS, and 1X complete protease inhibitor). Samples were treated with 0.1 mg/mL of proteinase K (Qiagen) overnight at 37°C. Mononucleosomal DNA sequencing was performed by the Genomics Core Facility at Pennington Biomedical Research Center. The sequenced reads were aligned to the human (hg38) genome using Bowtie 2 [27], while allowing 0 mismatches in a seed alignment during multiseed alignment. On average, ~98% of sequenced reads aligned to genome.

Statistical analysis:

Normality and equal variance between groups were assessed prior to each statistical test. All results are reported as mean ± SEM unless stated otherwise.

Anthropometrics and plasma assays:

GraphPad Prism 5.0 software was used to perform Student’s t-test or paired t-test where applicable. P < 0.05 was considered statistically significant. N=15 per treatment group and per time point.

Metabolomics:

Pairwise correlations were determined using JMP 7.0 software and correlation matrices were generated. Statistical significance between groups was determined by paired t-test with P < 0.05. N=15 per treatment group and per time point.

RNA-seq:

Differential gene expression was determined using Cuffdiff with the P value and false discovery rate (FDR) cutoffs set at 0.05 [28]. Gene Ontology (GO) and KEGG pathway analyses were performed using DAVID, with an EASE cutoff of 0.1 to determine significantly enriched functional annotations, as previously described [29]. N=14-15 per treatment group and per time point for RNA-seq experiments and data analysis.

MNase-seq:

Analysis was performed using HOMER and downstream custom scripts as previously described [20]. N=14-15 per treatment group per time point for experimentation. Within treatment groups and time points, for statistical analysis only, three individual samples were randomly pooled to give an overall N=5 per group, such that every individual enrolled in the study was represented in the MNase-seq experimentation and the number of reads per pooled sample were sufficient for robust statistical analysis between treatment groups and time, as consistent with analysis protocols previously utilized and published by our group and others for MNase-seq analysis [17, 30, 31].

Results:

Descriptive Characteristics of subjects

Out of the 55 subjects screened, 37 were eligible and consented to the study. A total of 30 subjects completed the totality of the study (Figure 1). All eligible subjects were considered healthy and were assigned to either the lean (n=15) or the Ov/Ob (n=15) group by BMI. One lean sample was excluded from RNA-seq and MNase-seq due to tissue mass limitations, and one Ov/Ob sample was excluded from plasma analyses, due to lack of Post blood sample procurement (Figure 1). The lean group (BMI, 22.03 ± 0.46 kg/m², 69.38 ± 1.81 kg) had significantly lower body weight (BMI, 28.95 ± 0.80 kg/m², 93.75 ± 3.49 kg), lower percent fat mass assessed via DXA (19.11 ± 1.25 vs. 27.52 ± 2.22), lower waist circumference (80.01 ± 1.54 vs. 93.88 ± 2.49 cm), lower hip circumference (91.41 ± 0.98 vs. 105.10 ± 1.54 cm), and lower percent lean mass assessed via DXA (76.49 ± 1.22 vs. 68.70 ± 2.12 ) than the Ov/Ob group (Table 1). There were no significant differences in age, waist-to-hip ratio, or physical activity levels according to the self-report IPAQ-short form questionnaire between groups (Table 1).

Figure 1. Study Overview.

A) Flowchart summarizing the subject screening and recruitment and B) schematic of experimental time course. Illustrations were created with BioRender.com

Table 1. Subject anthropometric and plasma measurements.

Results are expressed as means ± SEM.

	Lean		Ov/Ob
Anthropometric Characteristics
Number of Subjects	15		15
Age (years)	23.33 ± 0.96		23.73 ± 0.79
Body weight (kg)	69.38 ± 1.81		93.75 ± 3.49****
BMI (kg/m2)	22.03 ± 0.46		28.95 ± 0.80****
% lean mass	76.49 ± 1.22		68.70 ± 2.12**
% fat mass	19.11 ± 1.25		27.52 ± 2.22**
Waist circumference (cm)	80.01 ± 1.54		93.88 ± 2.49****
Hip circumference (cm)	91.41 ± 0.98		105.10 ± 1.54****
Waist-to-hip ratio	0.88 ± 0.01		0.89 ± 0.02
Kcal intake	7kcal/kg BW		7kcal/kg BW
IPAQ physical activity level	11 of 15 subjects in high category		9 of 15 subjects in high category
VO2 max (mL/kg/min)	39.73 ± 1.73		33.66 ± 2.31*
RER at 50% VO2max	0.86 ± 0.02		0.86 ± 0.01
Energy expenditure (kcal/min at 50% VO2max)	6.80 ± 0.37		7.69 ± 0.54
Exercise duration (mins to expend 650kcal)	99.94 ± 5.82		91.20 ± 7.26
Cycle ergometer resistance (for 50% VO2max)	7.93 ± 0.69		8.33 ± 0.71
Plasma measurements
	Pre	Post	Pre	Post
Number of subjects	15	15	15	14
Glucose (mg/dL)	76.9 ± 2.5	63.8 ± 1.7****	74.0 ± 1.9	68.0 ± 1.7*
Insulin (μU/mL)	14.6 ± 0.6	14.3 ± 0.4	18.2 ± 1.4#	16.7 ± 1.1#*
HOMA-IR	2.8 ± 0.17		3.3 ± 0.28
Triglycerides (mg/dL)	58.9 ± 6.1	65.5 ± 4.7	78.1 ± 10.6#	106.2 ± 15.4#***
Free fatty acids (mM)	3.7 ± 0.5	4.5 ± 0.17****	3.8 ± 0.07	4.4 ± 0.1***
Total cholesterol (mg/dL)	235.2 ± 32.3	291.5 ± 23.7	284.2 ± 30.3	374.5 ± 39.6*
LDL/VLDL- cholesterol (mg/dL)	105.5 ± 13.2	108.5 ± 14.1	187.7 ± 17.0##	183.5 ± 19.6##
HDL cholesterol (mg/dL)	49.7 ± 4.1	67.5 ± 5.9**	51.7 ± 5.7	56.0 ± 5.6

^*denotes significant difference between Pre and Post within the same group with p ≤ 0.05,

^**p ≤ 0.01,

^***p ≤ 0.001,

^****p ≤ 0.0001,

^#denotes significant difference between different BMIs either pre or post exercise p ≤ 0.05,

^##p ≤ 0.01,

^###p ≤ 0.001

Energy expenditure during stationary cycling exercise was determined during a VO₂ max test during the first visit. The VO₂ max test revealed a lower cardiorespiratory fitness in Ov/Ob (33.66 ± 2.31 mL/kg/min) compared to lean (39.73 ± 1.73 mL/kg/min). Interestingly, at 50% VO₂ max, there was no significant difference in substrate utilization ratio or energy expenditure between groups (Table 1). Based on the participants’ energy expenditure measured at 50% VO₂ max, the duration of and resistance during a single exercise bout for 650 kcal expenditure in visit 2 was calculated and measured and were not different between groups (Table 1).

Exercise improves plasma metabolite profile in both lean and overweight/obese

Prior to exercise, fasting plasma glucose levels did not differ significantly between lean and Ov/Ob, but plasma insulin was significantly increased in Ov/Ob (18.2 ± 1.4) compared to lean (14.6 ± 0.6), potentially suggesting pre-diabetes in subjects of the Ov/Ob group (Table 1). Prior to exercise, fasting HOMA-IR was not statistically different but trending higher in Ov/Ob (3.34 ± 0.28) compared to lean (2.79 ± 0.17) (Table 1). In response to exercise, plasma glucose levels decreased in both lean and Ov/Ob; however, plasma insulin levels decreased only in Ov/Ob, not in lean (Table 1). Despite the exercise-induced decrease in plasma insulin levels in Ov/Ob, insulin remained significantly higher in exercised Ov/Ob (16.7 ± 1.1) relative to exercised lean (14.3 ± 0.4) (Table 1).

The profile of free fatty acids, total cholesterol and HDL cholesterol were similar between lean and Ov/Ob prior to exercise; however, plasma triglycerides (78.1 ± 10.6 vs. 58.9 ± 6.1) and LDL/VLDL cholesterol (187.7 ± 17.0 vs. 105.5 ± 13.2) were significantly elevated in Ov/Ob compared to lean (Table 1). After exercise, plasma free fatty acids increased in both lean and Ov/Ob indicative of activated lipolysis, whereas plasma triglycerides and total cholesterol increased in Ov/Ob, but not in lean participants (Table 1). Conversely, exercise increased HDL cholesterol in lean but not Ov/Ob (Table 1), consistent with previous literature [32]. Together, these data suggest that lean individuals have better insulinemia and plasma lipid profiles compared to Ov/Ob, but that both lean and Ov/Ob experience beneficial exercise-induced improvements in plasma metabolite profiles.

Muscle long-chain acylcarnitines are increased in Ov/Ob compared to lean with and without exercise.

Build-up of long- and medium-chain acylcarnitines in skeletal muscle can indicate increased flux of fatty acids through mitochondrial beta-oxidation and/or incomplete beta-oxidation. Elevated acylcarnitine levels have been implicated in impacting mitochondrial function as observed in obesity and insulin resistance [5, 6]. Before exercise, lean and Ov/Ob muscles presented with similar levels of free-, acetyl-, short-, and medium-chain acylcarnitines (Figures 2A,B). However, the vast majority of long-chain acylcarnitines were increased in Ov/Ob subjects compared to lean (Figure 2C). These findings agree with reported accumulation of muscle long-chain acylcarnitines in obesity [5, 6]. In response to a single bout of exercise, both lean and Ov/Ob muscle showed increased acetyl-, short-, and medium-chain acylcarnitines likely reflecting increased macronutrient oxidative flux (Figures 2A,B). Exercise increased total short- and medium-chain acylcarnitines in lean subjects by 7-fold, but only by 3-fold in Ov/Ob (Figure 2D).

Figure 2. Muscle long-chain acylcarnitines are increased in Ov/Ob compared to lean with and without exercise.

Muscle free and acetyl (A), short and medium-chain (B) and long-chain (C) acylcarnitine levels in subjects before (Pre) or after (Post) acute exercise bout, respectively. (D) Relative fold-change from Pre to Post exercise in Lean and Ov/Ob subjects for free, acetyl, short- and medium-, or long-chain acylcarnitines. Data are expressed as mean ± SEM, # denotes significant difference between Lean and Ov/Ob,* denotes significant difference between Pre and Post by Student’s t-test, p ≤ 0.05, n=15.

Interestingly, despite elevated baseline levels in Ov/Ob, acute exercise did not further increase long-chain acylcarnitines in these subjects, nor in lean participants, with the exception of minor species C20:1 and C20:2 (Figure 2C). Together, these data demonstrate that at basal conditions, muscle long-chain acylcarnitines are elevated in Ov/Ob compared to lean subjects and that exercise does not impact long-chain acylcarnitine levels but rather increases acetyl-, short-, and medium-chain acylcarnitines in both lean and Ov/Ob subject likely reflecting exercise-induced increased rates of oxidative metabolism.

Exercise induces transcriptional reprogramming independent of BMI

To determine the effects of acute exercise on metabolism-related gene expression in skeletal muscle, RNA-seq was performed. Prior to exercise, 297 genes were differentially expressed between lean and Ov/Ob, while after acute exercise only 161 genes were distinct between lean and Ov/Ob (Figure 3A,B). Of all the genes that were differentially expressed in lean versus Ov/Ob, only 14% overlapped before and after exercise, suggesting that, in general, the gene expression profile is highly dependent upon BMI status before and after exercise (Figure 3C). A single exercise bout induced differential expression of a large number of genes in both lean and Ov/Ob, 1,230 and 1,123, respectively (Figure 3D). Nearly half of the exercise-induced genes overlapped between lean and Ov/Ob, suggesting that exercise-induced genes are largely independent of BMI status (Figure 3E). Pathway analysis of exercise-induced genes that overlapped between Ov/Ob and lean showed an enrichment for positive regulation of metabolism; however, the extent of this enrichment was larger in Ov/Ob compared to lean participants (Figure 3F), with over double the number of DEGs in each pathway. This suggests that changes in gene expression related to the positive regulation of metabolism in Ov/Ob subjects is more susceptible to exercise compared to lean. Gene ontology (GO) analysis revealed that genes upregulated in both lean and Ov/Ob following exercise were similarly enriched for terms related to lipid metabolism (GO Term ID 0033993), mitochondrial biogenesis (GO Term ID 0007006), and insulin signaling (GO Term ID 0032868) (Figure 3G). These data are consistent with previous studies showing that exercise improves insulin sensitivity [33, 34] and lipid metabolism [35] probably through enhancing mitochondrial biogenesis and function [36].

Figure 3. Lean and Ov/Ob subjects are similarly transcriptionally responsive to acute exercise.

A) Heat map showing raw z-scores for differentially expressed genes (DEGs). The complete-linkage method was used to compute hierarchal clustering and the Pearson method was used to measure distance between rows and columns. Green indicates that the expression level of gene higher, and red indicates lower, than the mean of the row. B) Total number of DEGs, either up or down regulated, and the overlap (C) comparing Lean to Ov/Ob either before (Pre) or immediately after (Post) exercise. D) Total number of DEGs, either up or down regulated, and the overlap (E) comparing before (Pre) to immediately after (Post) exercise in either Lean or Ov/Ob. F) Functional and G) GO Term enrichment of DEGs for either Lean or Ov/Ob comparing Pre to Post exercise. N=14-15/group.

We next examined several genes known to be important for metabolism and insulin signaling that were regulated by exercise in our study. Specifically, we found that both lean and Ov/Ob subjects upregulated phosphofructokinase 2 (PFK2), PGC-1α, protein phosphatase 2 (PP2A), insulin receptor 2 (IRS-2), phosphatidylinositol 3-kinase (PI3K), sirtuin-1 (SIRT-1), cyclic AMP-responsive element-binding protein 5 (CREB-5), and cyclin D1 in response to exercise (Figure 4A). In lean, but not Ov/Ob, exercise upregulated forkhead box protein O1 (FOXO-1), AMP-activated protein kinase (AMPK), and phosphofructokinase-1 (PFK-1) (Figure 4B). Conversely, in Ov/Ob, but not lean, exercise increased an insulin signaling-related gene, ras homolog enriched in brain (RHEB) (Figure 4C). Genes related to fatty acid oxidation, such as carnitine palmitoyltransferase 1 (CPT1) and 2 (CPT2), carnitine-acylcarnitine translocase (SLC25A20/CACT) or long-chain acyl-CoA dehydrogenase (ACADL) were not regulated by exercise, agreeing with the lack of long-chain acylcarnitine reduction in response to exercise (Figure 2). Importantly, these data support the idea that exercise can revert the obese transcriptome to resemble more closely that of a lean individual. These data also suggest that the transcriptional response to exercise-induced changes in lipid, mitochondria, and insulin-related genes occurs independent of BMI.

Figure 4. Exercise increases metabolism-related genes in both Lean and Ov/Ob.

Effect of acute exercise on skeletal muscle gene expression by RNAseq, Fragments Per Kilobase of transcript per Million (FKPM) of (A) PFK-2, PGC-1α, PP2A, IRS-2, PI3K, SIRT-1, CREB-5, Cyclin D1; and B) FOXO-1, AMPK, PFK-1; and C) RHEB in lean and Ov/Ob subjects before exercise and in lean and Ov/Ob immediately after exercise. **denotes significant difference between groups with p ≤ 0.01, ***p ≤ 0.001, ****- ≤ 0.0001, N=14-15/treatment group.

Skeletal muscle nucleosome mapping

A major determinant of transcriptional activity is the nucleosome positioning at and upstream (−1N) of the TSS regulating the accessibility of transcription factor binding. MNase-seq was performed in skeletal muscle to determine: a) whether obese and lean individuals had differential nucleosome positioning surrounding TSS; b) whether acute exercise would induce nucleosome remodeling; and c) whether this nucleosome remodeling would differ in lean compared to Ov/Ob. Similar to others [20], we observed nucleosome depletion in the region directly upstream of the (TSS) in the whole genome in all groups (Figure 5A). Interestingly, genome-wide −1N occupancy was increased in Ov/Ob Pre-exercise compared to lean Pre-exercise, and this difference was completely abolished post-exercise (Figure 5A). In genes downregulated by exercise in either lean or Ov/Ob, there was no difference in −1N occupancy and position pre or post. For exercise-induced upregulated genes, despite the fact that the −1N occupancy was greater before exercise in both lean and Ov/Ob (Figure 5B), the magnitude of −1N occupancy was higher in Ov/Ob compared to lean before and after exercise, suggesting that these genes may be epigenetically repressed in Ov/Ob compared to lean (Figure 5B). Nucleosome positioning in the genes included in the GO Terms for mitochondrial biogenesis (GO Term ID 0007006) and insulin signaling (GO Term ID 0032868) showed a similar effect wherein the Ov/Ob −1N occupancy was increased before exercise but reverted to levels similar to lean following exercise (Figure 5C). Thus, exercise may epigenetically regulate genes to alleviate pre-exercise differences between Ov/Ob and lean individuals.

Figure 5. Nucleosome occupancy in lean and Ov/Ob pre and post exercise.

Nucleosome occupancy in the region adjacent to the transcription start site (TSS) from nucleotide −1000 to nucleotide +1000 in the (A) whole genome, (B) all upregulated genes, and (C) GO term genes related to mitochondrial biogenesis and insulin signaling in skeletal muscle following a single exercise bout. Nucleosome occupancy is shown as a smoothed graph, yellow arrow highlights shift in −1N position, N=14-15/group.

We next examined the effect of exercise on −1N position in individual genes. Exercise caused the −1N to reposition away from the TSS in PGC-1α, RHEB, and PFK-1 promoters in both lean and Ov/Ob, consistent with increased expression of these genes post-exercise, with the exception of PFK-1 which was not upregulated by exercise in Ov/Ob (Figure 6A–C and 3). In the FOXO-1 and AMPK promoters, −1N occupancy repositioned towards the TSS following exercise in both groups but to a greater extent in Ov/Ob (Figure 6D–E), suggesting that failure of acute exercise to increase expression of these genes in Ov/Ob is compounded by nucleosome repositioning towards the TSS.

Figure 6. Effect of a single exercise bout on −1N nucleosome gene positions.

(A) PGC-1α, (B) RHEB, (C) PFK-1, (D) FOXO-1, (E) AMPK, (F) PP2A, (G) IRS-1, and (H) PI3K. −1 nucleosome positions are depicted within −600 to 0 of the transcription start site (TSS), N=14-15. Black arrow represents a shift for both lean and Ov/Ob, red arrows represents a shift for Ov/Ob only, and green line represents no overall shift.

In the PP2A, IRS-2, and PI3K promoter, exercise induced nucleosome repositioning closer to the TSS in the Ov/Ob group only (PP2A and IRS-2) or no repositioning (PI3K). Despite this, all three genes were upregulated similarly by exercise in both lean and Ov/Ob (Figure 6F–H and 3). Together, these data suggest that −1N repositioning in response to acute exercise may be important for expression of metabolism and mitochondrial genes, but changes observed in genes related to insulin signaling cannot be accounted for by epigenetic modifications.

In summary, these data suggest that epigenetic regulation of nucleosome positioning in skeletal muscle is differential by BMI status, but that exercise is a strong effector of nucleosome repositioning and potentially the regulation of differential gene expression in a manner that is largely independent of BMI.

Discussion

Skeletal muscle is a major site for post-prandial (insulin-dependent) and post-exercise (insulin-independent) glucose uptake and is one of the first tissues to exhibit insulin resistance in overweight and obese individuals [37, 38]. Additionally, in subjects with impaired glucose tolerance and type 2 diabetes, exercise increases insulin sensitivity [39–41]. Although the subjects in the present study were considered healthy, the Ov/Ob subjects did exhibit elevated fasting plasma insulin and HOMA-IR compared to lean, which may be indicative of prediabetes. Exercise-induced insulin-independent uptake of glucose in skeletal muscle of Ov/Ob and diabetic individuals is one of the mechanisms through which exercise improves health-related outcomes in these individuals [42]. In Ov/Ob subjects herein, exercise led to reduced plasma insulin levels, which can be considered beneficial as persistent elevated levels of insulin leads to loss of insulin sensitivity in the skeletal muscle [42–45]. In addition to increased uptake and oxidation of glucose, exercise induces concomitant hydrolysis and release of fatty acids from adipose and muscle triglycerides, and oxidation of fatty acids in order to meet energy demands particularly during low and moderate intensity exercise bouts [46]. Accordingly, plasma free fatty acids increased in lean and Ov/Ob after exercise. Of note, Pre samples were taken in a fasted state and Post samples were taken ~5 hours postprandially, as such serum triglycerides are known to be elevated 3-5 h following a meal which may have differentially influenced serum triglycerides in the Pre compared to Post states [46, 47]. Increased triglycerides observed in Ov/Ob but not in lean may be indicative of decreased lipoprotein lipase activity in Ov/Ob, which resulted in elevated postprandial triglycerides levels independent of the exercise bout [48]. Another explanation for the exercise-induced changes in triglycerides and free fatty acids may be the energy balance obtained in the Ov/Ob vs lean group, as both treatment groups were given a standardized 7kcal/kg breakfast prior to their exercise bout and both groups expended 650kcal total during the exercise bout. Thus, due to the trending difference in energy expenditure rates between the groups, the Ov/Ob group’s energy intake matched energy expenditure, resulting in energy balance, whereas, the lean group theoretically had a ~100kcal excess following the exercise bout. Carbohydrate intake is known to blunt the exercise-induced upregulation of metabolic genes, whereas glycogen depletion stimulates the expression of genes involved in metabolic adaptations [49, 50]. Thus, differences observed in the present study may be in response to differences in energy excess vs balance at the end of the exercise bout.

Interestingly, total cholesterol levels also increased in Ov/Ob but not in lean following the single exercise bout. Although others have shown that total cholesterol increases in amateur but not in trained, professional cyclists following exercise [51], it is difficult to determine the cause (postprandial or due to exercise) of the increases observed in the present study. In lean but not Ov/Ob, HDL-C level increased following the acute exercise bout, which may be an exercise-induced benefit leading to the reduction of cardiometabolic disease risk [52]. As others have noted and as we observed here, even a single bout of exercise may have beneficial effects on lipid metabolism and insulin sensitivity in healthy and diseased populations [34, 53].

The beneficial effects of exercise on cardiometabolic health may be due to its effect on specific mechanisms regulating insulin signaling in the skeletal muscle. In the skeletal muscle, mechanisms accused of the development of insulin resistance and subsequent progression to type 2 diabetes in Ov/Ob individuals include the interference in insulin signaling by excess lipid accumulation [54], mitochondrial dysfunction, and a reduction in mitochondrial biogenesis [55]. These can lead to incomplete beta-oxidation and subsequently to accumulation of acylcarnitines, that are purported to interfere with insulin signaling [56–59]. However, we recently reported that excessive accumulation of long-chain acylcarnitines in muscle with or without high-fat diet challenge did not significantly impair muscle insulin responsiveness [10]. In this study, Ov/Ob subjects had elevated long-chain acylcarnitines compared to lean subjects. Exercise did not impact the levels of long-chain acylcarnitines in either subject group, but did increase medium-, short-, and acetylcarnitine in both lean and Ov/Ob subjects. Thus, while exercise is known to improve metabolic parameters, insulin sensitivity, and glucose metabolism, we show that exercise failed to attenuate elevated long-chain acylcarnitines in Ov/Ob subjects suggesting that insulin-sensitizing mechanisms occur independent of long-chain acylcarnitine reduction.

Exercise induces gene expression of mitochondrial genes, including PGC-1α and other NEMGs, through alterations in nucleosome occupancy and −1N repositioning in the promoter region of these genes [18–20, 31]. In this study, we reported the first whole genome nucleosome maps in lean and Ov/Ob individuals in response to acute exercise in order to analyze nucleosome occupancy across the genome and to determine epigenetically regulated genes in a BMI-dependent manner. Surprisingly, the nucleosome occupancy in the MitoCarta database genes was not different between lean and Ov/Ob with or without exercise. We observed an overall shift in −1N positioning in Ov/Ob compared to lean at baseline. Importantly, exercise caused a shift in nucleosome occupancy for Ov/Ob towards that seen in lean, particularly for genes upregulated, but not for genes downregulated. In conjunction with our previous studies [20, 31], exercise led to −1N repositioning away from the TSS in the PGC-1α promoter in both lean and Ov/Ob. Other studies have demonstrated that acute exercise epigenetically induces gene expression of PGC-1α, and other genes related to mitochondrial function in skeletal muscle of human subjects [18]. On the other hand, PGC-1α is also epigenetically downregulated in type 2 diabetic subjects [19]. PGC-1α expression was significantly increased and nucleosome repositioning occurred in both lean and Ov/Ob in response to exercise. In the promoters of other genes upregulated by exercise, the −1N repositioning in response to exercise showed similar patterns in lean and Ov/Ob in conjunction with gene expression. These genes include those involved in AMPK and/or insulin signaling (CREB-5, RHEB, FOXO-1) and epigenetic modulators (SIRT-1, Cyclin D1). However, in the AMPK promoter or in those of components related to insulin signaling (IRS-2, PI3K, PP2A and PFK-1), the −1N repositioning show no change or changes inconsistent with upregulation following exercise in lean vs. Ov/Ob, despite increases in gene expression in either lean or Ov/Ob. These findings suggest that other epigenetic modulators that directly or indirectly impact nucleosome position, such as the histone acetyltransferases or deacetylases, do not have major effects on regulation of insulin responsiveness in skeletal muscle. Thus, although −1N repositioning may be important for determining expression of some genes known to regulate metabolism and mitochondrial adaptations in response to exercise training, it may not be an important regulatory mechanism for determining the expression of genes specific to the insulin signaling pathway.

In conclusion, these data suggest that differential expression of metabolism and mitochondrial adaptation-related genes are due in part to epigenetic alterations following exercise. The responsiveness of both Ov/Ob and lean individuals suggests that baseline BMI does not hamper the exercise-induced beneficial changes in transcriptome or epigenetic nucleosome repositioning mechanisms. In future studies, understanding specific epigenetic mechanisms through which genes related to metabolism and mitochondrial adaptations are regulated by exercise can help design therapeutic solutions to modulate expression levels in skeletal muscle which may be beneficial in obesity and type 2 diabetes.