Skeletal muscle transcriptome response to a bout of endurance exercise in physically active and sedentary older adults

Abstract

Age-related declines in cardiorespiratory fitness and physical function are mitigated by regular endurance exercise in older adults. This may be due, in part, to changes in the transcriptional program of skeletal muscle following repeated bouts of exercise. However, the impact of chronic exercise training on the transcriptional response to an acute bout of endurance exercise has not been clearly determined. Here, we characterized baseline differences in muscle transcriptome and exercise-induced response in older adults who were active/endurance trained or sedentary. RNA-sequencing was performed on vastus lateralis biopsy specimens obtained before, immediately after, and 3 h following a bout of endurance exercise (40 min of cycling at 60%–70% of heart rate reserve). Using a recently developed bioinformatics approach, we found that transcript signatures related to type I myofibers, mitochondria, and endothelial cells were higher in active/endurance-trained adults and were associated with key phenotypic features including V̇o_2peak, ATP_max, and muscle fiber proportion. Immune cell signatures were elevated in the sedentary group and linked to visceral and intermuscular adipose tissue mass. Following acute exercise, we observed distinct temporal transcriptional signatures that were largely similar among groups. Enrichment analysis revealed catabolic processes were uniquely enriched in the sedentary group at the 3-h postexercise timepoint. In summary, this study revealed key transcriptional signatures that distinguished active and sedentary adults, which were associated with difference in oxidative capacity and depot-specific adiposity. The acute response signatures were consistent with beneficial effects of endurance exercise to improve muscle health in older adults irrespective of exercise history and adiposity.

NEW & NOTEWORTHY Muscle transcript signatures associated with oxidative capacity and immune cells underlie important phenotypic and clinical characteristics of older adults who are endurance trained or sedentary. Despite divergent phenotypes, the temporal transcriptional signatures in response to an acute bout of endurance exercise were largely similar among groups. These data provide new insight into the transcriptional programs of aging muscle and the beneficial effects of endurance exercise to promote healthy aging in older adults.

INTRODUCTION

Age-related decline in cardiorespiratory fitness and physical function is associated with reduced quality of life and increased morbidity in older adults (1, 2). Skeletal muscle is an important mediator of these impairments, due to an age-related loss of mitochondrial content, function, and atrophy of muscle fibers (3, 4). The aging muscle phenotype is underpinned in part by an altered muscle gene expression program in older individuals as compared with their younger counterparts (5–7). In addition, there is a growing awareness of the impact that physical activity and adiposity has on age-associated decline in cardiovascular fitness, muscle mass, and mitochondrial function. For example, apparent age-related impairment in skeletal muscle mitochondrial energetics in sedentary individuals was more strongly related to higher BMI and lower cardiorespiratory fitness (8). In contrast, life-long endurance exercise attenuates the age-related loss in cardiorespiratory fitness and muscle mass and function compared with age-matched sedentary counterparts (9, 10). We and others also reported that age-related impairment in skeletal muscle mitochondrial energetics was attenuated in physically active older adults engaged in chronic endurance exercise training (11, 12). Despite the clear impact that physical activity level and adiposity have on age-associated decline of muscle, the molecular mechanisms that distinguish active/endurance trained from sedentary older adults have not been clearly defined. Thus, studies that comprehensively phenotype older adults with respect to training status, objectively measured physical activity and body composition in conjunction with molecular profiling are needed to better disentangle the role of physical activity and adiposity on loss of cardiorespiratory fitness and decline in physical function in aging.

Despite the beneficial effects of exercise for older adults, some loss of plasticity and adaptive response of muscle to exercise has been reported (13–16). This may reflect age-related differences in exercise-induced transcriptional programs. Specifically, the transcriptional response to an acute resistance exercise bout is attenuated in skeletal muscle of older adults compared with young adults (17), which may lead to an attenuation in resistance training adaptations such as single fiber size and contractile function (18, 19). Interestingly, despite the positive adaptations to endurance exercise training, the transcriptome response to an acute bout of endurance exercise has not been clearly defined in older adults (20, 21). As the repeated transient alterations in gene expression play a key role in mediating phenotypic adaptations (22), it is important to define the transcriptional response to acute endurance exercise. The majority of studies to date have focused on comparing young and older subjects (20, 21), and without considering the impact of decreased physical activity or adiposity that occurs with aging. As such, the muscle acute responses to endurance exercise have not been examined between older adults who are active/endurance trained versus sedentary.

To address this paucity in the literature, we generated muscle RNA-seq datasets from groups of well-phenotyped older adults that were either active/endurance trained or sedentary before and at two timepoints following an acute endurance exercise session. To infer relevant biological insight for our transcriptome datasets, we used a recently developed data-driven approach for analysis of cell-type proportion variation and pathway activity in global gene expression data [Pathway-Level Information ExtractoR (PLIER)] (23) in conjunction with deep human phenotyping. PLIER analysis enables insight into complex muscle tissue containing multiple cell types and identifies transcriptomic signatures (i.e., latent variables) that define specific cell types and the skeletal muscle phenotype (mitochondrial energetics, fiber type characteristics, etc.). We reasoned that differences in physical activity status would be reflected in diverse skeletal muscle transcriptome signatures and would result in altered response to exercise. Our study highlights several cell specific transcriptome signatures in muscle that underlie important phenotypic and clinical characteristics of older adults who are endurance trained. Surprisingly, the transcriptome programs activated by an acute exercise bout were largely similar among active/endurance trained and sedentary older adults. These findings are consistent with the beneficial effects of endurance exercise to improve muscle health in older adults.

METHODS

Participant Recruitment

Older men and women (65–90 yr of age) were recruited from the Orlando, FL area. Participants were in good general health, defined as not having chronic medical conditions, no contraindications for exercise, weight stable for the last 6 mo, and normal resting blood pressure (<150 mmHg systolic, <90 mmHg diastolic). Recruitment of participants occurred between January 2015 and June 2017, whereas experimental procedures occurred between February 2015 and August 2017. The participants who were consented but did not complete the entire experimental protocol were withdrawn from the study, primarily due to 1) not meeting inclusion criteria in the initial screening period, 2) contraindications to exercise (high blood pressure, abnormal EKG, etc.), or 3) inability to contact following initial recruitment. Participants who completed the study were assigned to either the older active/endurance trained (n = 10; 8 males, 2 females) or older sedentary group (n = 9; 7 males, 2 females). All experimental procedures were approved by the AdventHealth Orlando (Orlando, FL) institutional review board (IRB ID: 554559) and were performed in accordance with the standards set forth by the Declaration of Helsinki. Participants provided written informed consent before completing any data collection procedures.

Assessment of Exercise History and Objectively Assessed Physical Activity Behavior

Active/endurance-trained participants were required to be engaged in endurance exercise (running, cycling, and/or swimming) ≥3 days/wk without extensive lay off over the previous 6 mo, whereas sedentary individuals exercised ≤1 day/wk. Exercise history was determined by self-report questionnaire and a thorough interview. Additional assessments were undertaken to evaluate physical activity of the participants. Specifically, we used the Physical Activity Survey for the Elderly (PASE) questionnaire, which assigns a score (0–793) based on self-reported assessments of activity (walking, recreational activities, exercise, housework, yard work, caring for others), frequency, duration, and intensity level of activity over a week span (24, 25). Along with self-reported exercise habits, physical activity was objectively assessed using a triaxial accelerometer (SenseWear Pro Armband, BodyMedia Inc., Pittsburgh, PA) between study visits 3 and 4. The participant wore the monitor consistently for at least 7 days on their upper left arm, except while showering or bathing. Days with wear time of at least 85% were used for analysis.

Aerobic Capacity

Cardiorespiratory fitness (V̇o_2peak) was examined by a graded exercise protocol on a cycle ergometer as previously described (26). Heart rate, blood pressure, and ECG were constantly monitored throughout the test. Tests were terminated when volitional exhaustion was reached or when criteria outlined by the American College of Sports Medicine guidelines was observed in participants (27). These criteria include 1) plateau or decline in V̇o₂, 2) respiratory exchange ratio (RER) increased to 1.10 or higher, and/or 3) participant’s heart rate increased to within 10 beats of the age-predicted maximum [208 – (0.7 × age)].

Body and Muscle Composition

Body composition was assessed by dual energy X-ray absorptiometry (DXA) using a GE lunar iDXA whole body scanner. Lean and fat mass was analyzed with encore software, and skeletal muscle index (SMI) was calculated as appendicular lean mass per height (ALM/m²). In addition, thigh muscle and adipose composition was determined by magnetic resonance imaging (MRI) using a 3 T Philips Acheiva magnet as previously described (11). Thigh muscle, subcutaneous fat, and intermuscular fat volume were determined by Analyze 11.0 software (Biomedical Imaging Resource, Mayo Clinic, Rochester, MN).

Magnetic Resonance Spectroscopy

MRS was used to measure in vivo mitochondrial energetics via a 3 T Philips Acheiva magnet as previously described (11, 28). Briefly, a ³¹P surface coil was used to measure PCr, ATP, and Pi using standard one pulse acquisition experiment. The experimental spectra (6 s each) consisted of four (NSA), 1.5 s (TR) partially saturated free induction decays. Following a baseline measure, participants were instructed to perform isometric contractions of the quadriceps for ∼45 s to deplete PCr. The participants remained still to allow PCr peaks to return to normal (∼8 min). Established formulas were used to determine the intracellular pH based on the PCr and Pi chemical shifts (28). The changes in areas/heights of each peak in the spectrum during the experiment were determined using the AMARES (Advances Method for Accurate, Robust and Efficient Spectral fitting) algorithm within the jMRUI software; the changes in areas/heights were used to calculate the ATP_max.

Acute Exercise Testing

Participants performed an acute bout of endurance exercise as previously described (11). Briefly, the participants performed a 6-min warm-up consisting of light exercise on an electronically braked cycle ergometer (Lode Excalibur, Lode B.V., Groningen, The Netherlands). After the warm-up period, participants cycled for 40 min at a target of 70% heart rate reserve (HRR) that was calculated from the maximal heart rate attained during the V̇o₂ peak test. Participants were encouraged to finish the entire 40-min cycling bout, and pedaling resistance was adjusted to aid in completing the exercise protocol. Heart rate, perceived exertion, power output, and blood pressure were recorded every 5 min. Indirect calorimetry was performed using an open-circuit spirometry metabolic monitoring system (Parvo Medics, Sandy, UT). Power, heart rate, and V̇o₂ were taken as an average throughout the exercise bout once steady-state was achieved (∼5 min into the 40-min bout).

Percutaneous Muscle Biopsies

Following an overnight fast, participants consumed a standardized low-glycemic meal (200 kcal, 15% protein, 35% fat, and 50% carbohydrate). Fifteen minutes later, a percutaneous muscle biopsy was collected, and again immediately post and 3 h following the acute exercise bout. Participants were instructed to refrain from exercise for 48 h before the muscle biopsy to minimize the effects of acute exercise on baseline transcript profiles. Biopsies were obtained with suction from the middle region of the vastus lateralis under local anesthesia (2% buffered lidocaine). A separate portion of muscle (∼30 mg) was mounted on a small piece of cork with mounting medium (Shandon Cryochrome, Thermo Fisher Scientific, Pittsburgh, PA), frozen in liquid nitrogen-cooled isopentane, and stored at −80°C until histological analysis. The remaining tissue was flash frozen in liquid nitrogen and stored at −80°C until RNA extraction.

Histological Analysis for Fiber Type and Cross-Sectional Area

Muscle biospecimens were used for histological analysis of fiber type and cross-sectional area as previously described (29). Although single muscle fiber SDS-PAGE is considered the gold-standard, recent work has shown the reliability of immunostaining procedures to accurately assess fiber type proportion in human skeletal muscle biospecimens (30). A subset of participants [active/trained = 9 (7 male/2 female); sedentary = 8 (6 male/2 female] were used for this analysis due to insufficient tissue obtained from biopsy procedure from some participants. Mounted muscle samples were sectioned (10 µm) on a cryostat (Cryotome E, Thermo Shando, Pittsburgh, PA) at −20°C and placed on glass slides. For muscle fiber type, sections were blocked in 10% goat serum and incubated overnight with primary antibodies directed toward type I [BA-F8 (IgG2b); Developmental Studies Hybridoma Bank (DSHB), Iowa City, IA], type IIa [SC-71 (IgG1); DSHB], and type IIx (6H1 IgM; DSHB). The following day, sections were washed, incubated with fluorescent-labeled secondary antibodies [DyLight 405 (IgG2b), Alexa Fluor 488 (IgG1) and Alexa 555 (IgM); Thermo Fisher Scientific], and coverslipped with fluorescent mounting media (Prolong Gold; Thermo Fisher Scientific). Sections were imaged using an inverted fluorescent microscope (Nikon), and image analysis was performed using NIS elements software 4.20.01.

Transcript Profiling

Baseline and exercise-induced transcript profiles were examined in muscle biopsy specimens by RNA-sequencing and informatic analysis. Approximately 10 mg of skeletal muscle tissues were homogenized in a Bead Ruptor Elite bead mill homogenizer (Omni International, Kennesaw, GA) cooled by liquid nitrogen and homogenized for 2 × 20 s at 4.2 m/s. Muscle homogenates were processed using the Agencourt RNAdvance Tissue kit (Beckman Coulter, Indianapolis, IN) on a BioMek FXP Laboratory Automation Workstation (Beckman Coulter, Indianapolis, IN) according to the manufacturer’s instructions. The purity and concentration of the RNA samples were evaluated on a NanoDrop One/One spectrophotometer (Thermo Fisher Scientific) and quantified using Quant-iT RiboGreen RNA Assay kit (Thermo Fisher Scientific). The RNA integrity was assessed with an RNA Nano 6000 Assay kit of the Agilent Bioanalyzer 2100 system (Agilent Technologies, Palo Alto, CA). RNA integrity numbers (RINs) for all samples were above 8.0 with a median RIN of 8.9. mRNA libraries were prepared using the Universal Plus mRNA-Seq Library preparation kit (NuGEN). Briefly, mRNA was isolated from purified 400 ng total RNA using oligo-dT beads and used to synthesize cDNA following the NuGEN’s instructions. The transcripts for ribosomal RNA (rRNA) and globin were further depleted using the AnyDeplete kit (NuGEN) before the amplification of libraries except for a subset of 15 sedentary muscle samples. The purified libraries were quantified using the Qubit dsDNA HS kit (Thermo Fisher Scientific). Library size and distribution were determined by Bioanalyzer using the High Sensitivity DNA kit (Agilent) or Fragment Analyzer using High Sensitivity NGS Analysis kit (Agilent).

RNA-Sequencing and Preprocessing

Following library preparation, samples were pooled, and the library integrity was confirmed by sequencing on an Illumina MiSeq. Deep sequencing was subsequently performed using an S2 flow cell in a NovaSeq 6000 Sequencing System (Illumina, Inc., San Diego, CA), with an average depth of 30 million 50 bp paired-end reads per sample. Raw data were processed using bcl2fastq Conversion Software (Illumina) to obtain FASTQ files and reads were subjected to quality control using FastQC (31) and RNASeqMetrics (Broad Institute, Picard Toolkit). The reads were aligned to the Human GENCODE hg38 genome using STAR (32). Transcripts were quantified with rsem (33). The raw FASTQ data and the final transcript expression matrix were deposited with GEO (GSE151066).

Differentially Expressed Transcripts

We used DESeq2 to identify transcripts that differed between cohorts at baseline and transcripts that differed between timepoints for each cohort with a |log₂-fold change| > 1 and a false discovery rate (FDR) of < 0.05 (34).

PLIER Analysis

We used PLIER for the functional interpretation of RNA-seq datasets (23). The PLIER default gene sets were augmented with skeletal muscle fiber-type and mononuclear cell-type specific gene signatures generated in a previous study (35). For each fiber-type, 50 unique marker transcripts were chosen from the fiber-type specific transcriptomes. Only transcripts that were present in our data set were used for the selection of marker transcripts. Mononuclear cell-type (e.g., endothelial cells, FAPs) gene signatures were taken directly from Rubenstein et al. (35). The parameter k, which controls the number of LVs generated, was set to 34, which is twice the number of statistically significant principal components, as has been recommended for PLIER. Otherwise, default parameters were used for PLIER.

The PLIER framework identifies groups of correlated transcripts, or latent variables (LV), which may change in a coordinated manner due to a specific transcriptional program or cell-type composition variation. PLIER provides a metric for the prevalence of each LV within each sample by which a statistical measure of variation of the LV across groups or timepoints can be computed. LVs are further associated with prior knowledge in the form of skeletal muscle fiber-type and mononuclear cell-type gene signatures (35), as well as biological pathways from publicly available databases, where the association has statistical support. When LVs are associated with cell-type marker gene signatures, they are interpretable as a proxy for cell-type proportion.

Statistical Analyses

Baseline characteristics were compared using a Student’s t test. χ² test was used to examine differences in the sex ratio (males/females) between the groups. Pearson partial correlation analyses were used to evaluate the relationship between transcript profiles and biological or clinical measures at baseline. For time course data generated during the acute exercise bout, a repeated-measures ANOVA with the group, timepoint, and their interaction as factors in the model was utilized, followed by Tukey’s post hoc test. Data are presented as means ± standard deviation (SD). All analyses were performed in SAS (9.4) and statistical significance was set at P < 0.05.

RESULTS

Participant Characteristics and Physiological Response to Exercise

The overarching goal of this study was to identify transcriptomic signatures that underlie differences between physically active/endurance trained and sedentary older adults at baseline and in response to an acute bout of exercise. We used a deep phenotyping approach that combined clinical assessments with molecular profiling of skeletal muscle biospecimens obtained before, immediately after, and 3 h following an acute bout of endurance exercise. All testing procedures were completed over four study visits to the AdventHealth Translational Research Institute (Orlando, FL) and the study design is depicted in Fig. 1. Overall, 19 older participants who were either physically active/endurance trained or sedentary were recruited and completed the study protocol. Participant characteristics can be found in Table 1. Age was similar among the groups. Consistent with a sedentary lifestyle, the sedentary older adults had a higher body weight and BMI, as well as whole body, visceral, and thigh fat mass when compared with active/endurance-trained participants. Per study design, older active/endurance-trained adults had a higher PASE score, absolute (L/min) and relative [mL/kg fat-free mass (FFM)/min] V̇o_2peak, activity levels (steps per day), and in vivo mitochondrial capacity (ATP_max) when compared with sedentary counterparts.

Figure 1.

Study design outline. Baseline and exercise-induced clinical phenotyping was performed across four separate visits. Following written informed consent, medical history, assessment of self-reported physical activity [physical activity survey for the elderly (PASE)], and assessment of body composition [dual X-ray absorptiometry (DXA)] were assessed on the screening visit for each participant. Following the screening visit, assessments of cardiorespiratory fitness (V̇o₂ peak test), muscle volume, and mitochondrial function [magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS)] on visits 2 and 3. Objective physical activity was assessed for at least 7 days between visits 3 and 4 via the SenseWear Pro Armband. During the exercise visit (visit 4), muscle biopsy specimens were obtained before (Pre), immediately (0 h) following an acute bout of endurance (cycling) exercise [40 min at 60%–70% heart rate reserve (HRR)] and 3 h into recovery postexercise (3-h post). Biopsies were used for immunohistology and transcriptome profiling by RNA-seq followed by informatics. Exercise intensity was evaluated throughout the exercise bout by examining heart rate, power output, ventilatory gas exchange volume of oxygen consumed (V̇o₂) and volume of carbon dioxide produced (V̇co₂) by indirect calorimetry and rating of perceived exertion (RPE).

Table 1.

Subject Characteristics

	Active/Trained	Sedentary	P Value
Clinical Characteristics
Age, yr	68.4 ± 3.8	70.7 ± 4.0	0.2245
Sex, Males/Females	8/2	7/2	1.0000^
Weight, kg	72.1 ± 14.9	85.4 ± 13.7	0.0593
BMI, kg/m2	24.2 ± 4.2	28.7 ± 3.3*	0.0200
PASE score	173.5 ± 76.5	96.4 ± 40.2*	0.0151
SMI, ALM/m2	7.99 ± 1.17	8.07 ± 1.08	0.8853
Total fat mass, kg	17.1 ± 8.8	31.0 ± 1.2	0.0033
Visceral adipose tissue mass, kg	0.70 ± 0.65	1.78 ± 1.06	0.0151
Thigh muscle volume, cm3	2,216.1 ± 373.7	2,064.2 ± 407.7	0.4085
Thigh subcutaneous fat volume, cm3	704.7 ± 403.0	1,257.8 ± 663.5*	0.0400
Thigh intermuscular fat volume, cm3	201.5 ± 135.4	330.3 ± 114.1*	0.0397
Absolute V̇o2peak, L/min	2.50 ± 0.63	1.78 ± 0.38*	0.0085
Relative V̇o2peak, mL/kg FFM/min	44.4 ± 6.5	32.1 ± 3.4*	0.0001
Daily physical activity, steps/24 h	8,366.6 ± 2,913.3	4,626.6 ± 2,396.6*	0.0075
ATPmax, mM/s	1.14 ± 0.43	0.60 ± 0.13*	0.0024
Histological Characteristics
Type I muscle fiber, %	59.3 ± 21.5	42.6 ± 9.8	0.0618
Type II muscle fiber, %	40.7 ± 21.5	57.5 ± 9.8	0.0617
Type I CSA, µm2	5,921.2 ± 1,305.2	4,704.3 ± 707.7*	0.0333
Type IIa CSA, µm2	5,499.4 ± 1,195.2	4,447.7 ± 1,230.9	0.0943
Physiological Variables during Exercise
AVG HR, bpm	120.8 ± 17.6	114.4 ± 16.7	0.4679
%HRR, %	67 ± 8	61 ± 14	0.2346
AVG V̇o2, L/min	1.6 ± 0.4	1.1 ± 0.3*	0.0000
%V̇o2peak, %	64 ± 7	64 ± 8	0.9735
AVG power output, W	100.2 ± 26.2	49.7 ± 18.8*	0.0014
RQ, V̇co2/ V̇o2	0.89 ± 0.03	0.84 ± 0.02	0.4133

Data are presented as means ± standard deviation (SD). absolute V̇o₂, oxygen consumption during acute exercise bout; ALM, appendicular lean mass; AVG, average; BMI, body mass index; cm, centimeters; CSA, cross-sectional area of myofibers; hr, hour; kg, kilograms; L/min, liters per minute; m, meters; PASE, physical activity score for the elderly; RQ, respiratory quotient [carbon dioxide produced (V̇co₂)/oxygen consumed (V̇o₂)] during acute exercise bout; SMI, skeletal muscle index; V̇o_2peak, maximal oxygen consumption during graded exercise test; %, proportion of type I or type II myofibers; %HRR, exercise intensity presented as percentage (%) of heart rate reserve (HRR); %V̇o_2max, exercise intensity presented as percentage (%) of maximal oxygen consumption obtained during a graded exercise test (V̇o_2peak).

*Significant differences between groups (P < 0.05). ^χ² analysis.

After the initial baseline biopsy, participants completed a 40-min bout of cycling at 60%–70% of heart rate reserve (HRR), after which biopsies of the vastus lateralis were taken immediately and 3-h postexercise for RNA-seq analysis. Physiological responses recorded during the acute exercise bout can be found in Table 1. Active/endurance-trained adults exercised at a higher absolute V̇o₂ and power output during the acute bout of exercise. However, all participants performed the exercise bout at a similar relative intensity (% HRR and % V̇o_2peak) and respiratory quotient (RQ). The small low glycemic meal consumed before exercise may have influenced the RQ response to exercise. However, the change in RQ is similar to previous described work using a fixed exercise intensity (22).

Global Transcriptome Profiling of Skeletal Muscle

Molecular profiling of skeletal muscle was performed by RNA-seq analysis. The final data set had an average read depth of 33,733,605 reads and 93.8% of base pairs were mapped to the transcriptome. A subset of samples was subjected to targeted ribo-depletion which produced a batch effect (Supplemental Fig. S1 and Supplemental Table S1; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.16888711). Computational correction of this batch effect is described in the supplemental methods section. We initially performed a principal components analysis (PCA), which revealed a clear group-specific separation indicative of global differences in baseline transcriptomes (Principal component 2, Supplemental Fig. S2A). Hierarchical clustering of the differentially expressed transcripts at baseline (Pre) resulted in a clear separation of all samples by group (Supplemental Fig. S2B). In response to exercise, the PCA showed that the data separated by timepoint within the general exercise response trajectory along PC1 (Supplemental Fig. S2A). The heatmaps presented in Supplemental Fig. S2C–E show differentially expressed genes across timepoints in both study groups and show the transcripts separated by timepoint. Next, we explored the impact of interparticipant differences in exercise intensity on the transcriptional response to exercise. To do this, we used ANCOVA models and found that variance in exercise intensity [i.e., %V̇o_2peak, %HRR, and % ventilatory threshold (VT)] did not impact time, group, or interaction effects on differentially expressed genes (data not shown). These initial analyses indicated clear global differences in muscle transcriptomes between active/endurance trained and sedentary groups at baseline, whereas the transcriptional directional and magnitude response to exercise was similar among the groups.

To further investigate group differences, we utilized PLIER, a recently developed bioinformatics framework that finds and interprets transcript expression variation in one data-driven step by aligning prior knowledge pathways to sources of variation in a transcriptome data set (23). PLIER serves as an effective alternative to the standard pipeline of performing functional or pathway enrichment analyses on sets of differentially expressed genes that is enabled by tools such as DAVID (36) or ingenuity pathway analysis (IPA) (37). Using PLIER, we identified 42 LVs, of which 30 mapped to prior knowledge pathways or cell-type signatures (complete list of LVs can be found in Supplemental Table S2). The heatmap in Fig. 2 includes LVs that showed differences between active/endurance trained and sedentary groups, distinct temporal responses to acute exercise, or both group and time differences, as well as cell signatures that have been implicated in skeletal muscle adaptations with aging. Specifically, these include LV3 and 41 that includes transcripts for fibro-adipogenic progenitor (FAP) cells (38), LV18 that includes transcript involved in the ribosome, and LV22 that includes transcripts for the lysosome.

Figure 2.

Cell type and pathways associated with PLIER latent variables (LVs). Heatmap of selected latent variable (LV) pathway associations (colored entries) that showed differences between active/endurance trained and sedentary groups, distinct temporal responses to acute exercise, or both group and time differences, as well as cell signatures that have been implicated in skeletal muscle adaptations with aging (i.e., FAP cells, ribosome, lysosome). Prior knowledge inputs provided to PLIER included canonical pathways (KEGG, Reactome), muscle fiber type-specific pathways, and immune cell-specific pathways. Darker colors indicate greater (LV) pathway associations. LVs were selected based on the presence of significant differences between groups, timepoints, or both. Star indicates gene signatures that were derived from single cell (gold star) or individual muscle fiber (blue star) RNA-seq analysis described by Rubenstein et al. (35). PLIER, pathway-level information ExtractoR.

Transcriptomic Signatures That Distinguish Active/Endurance Trained and Sedentary Phenotypes

We used previously reported gene expression patterns from RNA-seq of subtype-pooled single human muscle fibers and single cell RNA-seq of mononuclear cells from human vastus lateralis to examine cell-type proportion in our data (35). Two of the PLIER-identified LVs, LV10, and LV17, were significantly associated with fiber type transcript signatures. LV10 contained transcripts that characterize slow fibers and were elevated in active/endurance-trained older adults (Fig. 3, A and C), whereas LV17 contained transcripts associated with fast fibers and were higher in sedentary adults (Fig. 3, B and D). Along with the known fiber type specific transcripts derived from our original work, LV10 contained several novel transcripts that have not been described. These include: 1) MYLK3, a serine-threonine kinase that has previously been shown to be specific to cardiac tissue that acts on myosin light chain isoform to influence contractile performance of myocytes (39), 2) INPP4B, a lipid phosphatase expressed in skeletal muscle that has been shown to suppress PI3K-Akt signaling through binding to phosphatidic acid and PI(3,4,5)P3 (40), 3) CFL2, a skeletal muscle-specific cofilin where human genetic defects were associated with myopathy (41), and 4) GBP2, a GTPase induced by interferon signaling that is associated with apoptosis and has recently been shown to be upregulated in skeletal muscle of Parkinson’s disease patients (42). Compared with our previous work that examined transcript profiles in individual muscle fibers (35), our results were derived from bulk RNA-seq analysis of whole muscle, a heterogenous tissue that contains various cell types along with contractile proteins (i.e., satellite cells, endothelial cells, neuronal cells, etc.). Thus, presence of these transcripts may be attributed to a greater abundance of these cell types surrounding type I muscle fibers. The differences in LV10 and LV17 between groups mirrored myofiber proportion determined by MHC staining and imaging by immunohistology (Table 1). To examine this relationship, we computed Pearson correlation coefficients between the per sample LV estimations and their respective immunohistochemical measurements. We found strong positive correlations between transcriptome-derived fiber type-associated LVs and myofiber proportion (Fig. 3, E and F).

Figure 3.

Muscle fiber transcriptome signatures are associated with histological analysis of muscle fiber type proportion in older active/endurance trained and sedentary adults. Heatmaps represent transcripts for the top thirty transcripts that are associated with LV10 (slow fiber type) (A) and LV17 (fast fiber type) (B). Each column represents one sample; columns are ordered by timepoint (Pre, light purple; Post, orange; 3-h post, maroon) and group [active/trained, green (n = 10); sedentary, blue (n = 9)]. Black squares indicate transcripts that mapped to specific prior knowledge pathway. Data are presented as z-scored regularized log-transformed gene expression data. Transcript signatures for slow fiber type (LV10) (C) and fast fiber type (LV17) (D) were evaluated in older active/endurance trained (green circles and lines, n = 10) and sedentary adults (blue circles and lines, n = 9) adults at before (Pre), immediately following (Post), and 3 h after (3-h Post) an acute bout of endurance exercise. Data are presented as means ± SD. ^Significant difference (P < 0.05) between groups; †P < 0.05 vs. Pre; §P < 0.05 vs. Post using a repeated-measures ANOVA with group, timepoint, and their interactions as factors in the model, followed by Tukey’s post hoc test. Pearson partial correlations were then used to examine the relationship between LV signatures and type I myofiber proportion (E) and type II proportion (F) in older active/endurance trained [green circles, n = 9 (7 male/2 female)] and older sedentary [blue circles, n = 8 (6 male/2 female)].

Cardiorespiratory fitness is a marker of overall health and progressively decreases with age (1). This age-related decrease in cardiorespiratory fitness and physical function is paralleled by age-related impairment in muscle mitochondrial energetics. We found that LV11 contained transcripts associated with mitochondrial energy metabolism and was significantly higher in the active/endurance-trained group than in the sedentary group (Fig. 4, A and B) and were positively associated with cardiorespiratory fitness and ATP_max, a noninvasive measure of mitochondrial capacity to produce ATP (Fig. 4, C and D).

Figure 4.

Mitochondrial transcriptome signatures are associated with whole body and muscle-specific oxidative capacity in older active/endurance trained and sedentary adults. A: heatmaps represent transcripts for the top 30 transcripts that are associated with LV11. Each column represents one sample; columns are ordered by timepoint (Pre, light purple; Post, orange; 3-h Post, maroon) and group [active/trained, green (n = 10); sedentary, blue (n = 9)]. Black squares indicate transcripts that mapped to specific prior knowledge pathway. Data are presented as z-scored regularized log-transformed gene expression data. B: transcript signatures for mitochondria (LV11) were evaluated in older active/endurance trained (green circles and lines, n = 10) and sedentary adults (blue circles and lines, n = 9) adults at before (Pre), immediately following (Post), and 3 h after (3-h Post) an acute bout of endurance exercise. Data are presented as means ± SD. ^Significant difference (P < 0.05) between groups. Pearson partial correlations were then used to examine the relationship between LV signatures and cardiorespiratory fitness (Relative V̇o_2peak) (C) and in vivo mitochondrial energetics (ATP_max) (D) in older active/endurance trained (green circles, n = 10) and older sedentary (blue circles, n = 9).

Muscle capillaries are comprised primarily of endothelial cells and contribute to muscle oxidative capacity by supplying nutrients and oxygen to contracting muscle and are reduced in older adults (43). We identified that LV33 was enriched with endothelial cell-related genes and was higher in the active/endurance-trained group than in the sedentary group (Fig. 5, A and B). In addition, the endothelial cell-associated LV33 was positively correlated with V̇o_2peak (Fig. 5C) and ATP_max (Fig. 5D). Along with oxidative capacity, the endothelial cell signature was related to markers of whole muscle mass including SMI (Pearson r = 0.5567, P value = 0.0164), thigh muscle volume (Pearson r = 0.4813, P value = 0.0369), and type I myofiber cross-sectional area (Pearson r = 0.5101, P value = 0.0361). These findings are in line with observations by Prior et al. (44) who showed that the capillary-to-muscle fiber ratio was positively associated with V̇o_2peak, appendicular lean mass, and thigh muscle cross-sectional area in older adults with sarcopenia. Along with endothelial cells, LV23, which mapped to smooth muscle cell transcripts, was higher in the older active/endurance-trained group than in the sedentary group (Supplemental Fig. S3A and B). This, in combination with a higher endothelial cell signature, would suggest higher vascularization in skeletal muscle of older active/endurance-trained adults.

Figure 5.

Endothelial cell transcriptome signatures are associated with oxidative capacity and skeletal muscle mass in older active/endurance trained and sedentary adults. A: heatmaps represent transcripts for the top 30 transcripts that are associated with LV33. Each column represents one sample; columns are ordered by timepoint (Pre, light purple; Post, orange; 3-h Post, maroon) and group [active/trained, green (n = 10); sedentary, blue (n = 9)]. Black squares indicate transcripts that mapped to specific prior knowledge pathway. Data are presented as z-scored regularized log-transformed gene expression data. B: transcript signatures for endothelial cells (LV33) were evaluated in older active/endurance trained (green circles and lines, n = 10) and sedentary adults (blue circles and lines, n = 9) adults at before (Pre), immediately following (Post), and 3 h after (3-h Post) an acute bout of endurance exercise. Data are presented as means ± SD. ^Significant difference (P < 0.05) between groups; †P < 0.05 vs. Pre; §P < 0.05 vs. Post using a repeated-measures ANOVA with group, timepoint, and their interactions as factors in the model, followed by Tukey’s post hoc test. Pearson partial correlations were then used to examine the relationship between LV signatures and cardiorespiratory fitness (Relative V̇o_2peak) (C) and in vivo mitochondrial energetics (ATPmax) (D) in older active/endurance trained (green circles, n = 10) and older sedentary (blue circles, n = 9).

LV42, which mapped to immune cell transcript profiles (myeloid cells, T cells, and NK cells), was higher in the older sedentary group than in the active/endurance-trained group (Fig. 6, A and B). We found that immune cell profile was positively associated with visceral adipose tissue mass (Fig. 6C) and thigh intermuscular adipose tissue (IMAT) volume (Pearson r = 0.437, P = 0.062), as well as IMAT expressed as percentage of total thigh adipose tissue volume (Fig. 6D). Interestingly, subcutaneous adipose tissue volume was not associated with LV42 (Pearson r=−0.058, P = 0.814). Along with adiposity, we also examined the relationship between immune cell signatures with markers of physical activity. Although LV42 was not associated with V̇o_2peak (Pearson r=−0.365, P = 0.125), we did observe a trend for immune cell signatures to correlate with daily physical activity (steps per 24 h; Pearson r= −0.420, P = 0.074). These data suggest an elevation in visceral and muscle-specific adiposity may be associated with higher immune cell infiltration in skeletal muscle from older sedentary adults.

Figure 6.

Immune cell signatures are elevated in older sedentary adults and are related to visceral and intermuscular adipose tissue accumulation. A: heatmaps represent transcripts for the top 30 transcripts that are associated with LV42 (immune cells). Each column represents one sample; columns are ordered by timepoint (Pre, light purple; Post, orange; 3-h Post, maroon) and group [active/trained, green (n = 10); sedentary, blue (n = 9)]. Black squares indicate transcripts that mapped to specific prior knowledge pathway. Data are presented as z-scored regularized log-transformed gene expression data. B: transcript signatures for immune cells (LV42) were evaluated in older active/endurance trained (green circles and lines, n = 10) and sedentary adults (blue circles and lines, n = 9) adults at before (Pre), immediately following (Post), and 3 h after (3-h Post) an acute bout of endurance exercise. Data are presented as means ± SD. ^Significant difference (P < 0.05) between groups using a repeated-measures ANOVA with group, timepoint, and their interactions as factors in the model, followed by Tukey’s post hoc test. Pearson partial correlations were then used to examine the relationship between LV signatures and visceral adipose tissue mass (C) and intermuscular adipose tissue (IMAT) volume (D) in older active/endurance trained (green circles, n = 10) and older sedentary (blue circles, n = 9).

Skeletal Muscle Transcriptional Response following a Bout of Endurance Exercise in Active/Endurance Trained and Sedentary Older Adults

Alterations in gene expression may occur across distinct trajectories during the minutes and hours following exercise. Despite the beneficial effects of endurance exercise training on whole body and skeletal muscle health, few studies have compared the skeletal muscle transcriptomic responses to an acute bout of endurance exercise in older active/endurance trained and sedentary adults (21). PLIER analysis identified several LVs that varied by timepoint and thus characterized the exercise response, including LVs that exhibited an early (LV2) (Fig. 7A), late (LV5 and LV28) (Fig. 7, B and C), and a gradual response to exercise (LV39) (Fig. 7D). Specifically, LV2 contained classic immediate early transcripts including JUN, EGR1, and NR4A1 (Fig. 7E), which likely indicates of activation of the mitogen-activated protein kinase (MAPK) family, a common stress response to acute exercise that regulates transcription of numerous cellular processes (22). In addition, several chemokine gene transcripts (CCL2, CXCL1, CXCL2) were elevated in skeletal muscle to similar extent in both groups immediately postexercise. Furthermore, LV5 included PPARGC1A and PPARGC1B (PGC-1α and β), the master regulators of mitochondrial biogenesis known to be induced by acute exercise (45–48) (Fig. 7F). The LVs that characterized acute responses to exercise exhibited no identifiable differences between the groups (Fig. 7, E–H). These data are consistent with the initial analysis of global transcriptomic response described in Supplemental Fig. S2(A, C–E), providing further evidence that the transcriptional responses to exercise were largely similar between the active/endurance trained and sedentary groups.

Figure 7.

Latent variables (LVs) that define time-dependent transcriptional responses to exercise are similar between active/endurance trained and sedentary older adults. LVs associated with the early (LV2; A), late (LV5; B and LV28; C), and gradual (LV39; D) transcriptional response to an acute bout of endurance exercise were evaluated in older active/endurance trained (green circles and lines, n = 10) and sedentary adults (blue circles and lines, n = 9). Data are presented as means ± SD. †P < 0.05 vs. Pre; §P < 0.05 vs. Post; ‡P < 0.05 vs. 3-h Post using a repeated-measures ANOVA with group, timepoint, and their interactions as factors in the model, followed by Tukey’s post hoc test. Heatmaps were generated using transcripts that are differentially expressed in LV2 (E), LV5 (F), LV28 (G), and LV39 (H). Data are presented as z-scored regularized log-transformed gene expression data. Each column represents an individual participant and are separated by timepoint (Pre, purple; Post, orange; 3-h Post, red) and group (active/trained, green; sedentary, blue).

To further evaluate the transcriptional response to acute exercise, differential gene expression analysis was performed separately for each group. The Venn diagrams presented in Fig. 8, A–C show differentially expressed genes (DEGs, both positive and negative; complete list of genes can be found in Supplemental Tables S3–S5) across timepoints in both study groups. Similar to the PLIER analysis, there was significant overlap in the response of individual transcripts to exercise between the two groups. Specifically, ∼71% (164) of DEGs were similar between the groups when comparing baseline (Pre) to immediately postexercise (Post) (Fig. 8A), 55% (176) of the DEGs were similar when comparing baseline to 3-h postexercise (3 h Post) (Fig. 8B), and 58% (96) of the DEGs were similar when comparing immediately postexercise to 3-h postexercise (Fig. 8C). We next performed a KEGG Enrichment analysis of DEGs that were similar between the groups (Fig. 8, D–G). In line with the upregulation of immediate-early transcripts found in LV2 (Fig. 7E), the MAPK signaling pathway was enriched at the immediately postexercise timepoint compared with baseline, as well as the 3-h postexercise timepoint compared with immediately postexercise. In addition, tumor necrosis factor (TNF) signaling pathway was enriched at the immediately postexercise timepoint compared with baseline, as well as the 3-h postexercise timepoint compared with to baseline. Along with TNF signaling pathway, the PI3K-Akt signaling pathway was enriched at the 3-h postexercise timepoint compared with baseline. Collectively, KEGG pathway analyses revealed that the transcriptional response for cellular stress (MAPK and TNF signaling) as well the metabolic (PI3K-Akt signaling) pathways following an acute bout of endurance exercise was largely similar among older active/endurance-trained adults and sedentary individuals.

Figure 8.

Pathway analysis of transcriptome pathways that are similarly provoked in skeletal muscle by endurance exercise in older active/endurance trained and sedentary adults. Venn diagrams represent DE transcripts for each cohort (green, active/endurance trained-specific transcripts; blue, sedentary (Sed)-specific transcripts; gray, transcripts with similar response between active/endurance trained and sedentary adults) in every comparison [Pre vs. Post (A), Pre vs. 3-h Post (B), Post vs. 3 h Post (C)]. Pathway analysis of transcripts that showed similar response between groups in each comparison [Pre vs. Post (D), Pre vs. 3-h Post (E and F), Post vs. 3 h Post (G)]. Color of dot indicates level of significance (P-adjusted value) and size of dot indicates number of transcripts found in each pathway.

Despite the similarities in transcript profiles following acute exercise, there were also transcriptional responses that distinguished the two groups. The greatest group-specific difference was observed at the 3-h postexercise timepoint compared with baseline, with 135 transcripts specific to the sedentary group, whereas only 10 were specific to the active/endurance-trained group (Fig. 8B). The top biological processes gene ontology (GO) terms that were enriched and specific to the sedentary group included cellular response to peptide, positive regulation of cellular catabolic process, catabolic processes, as well as ATP and ADP metabolic processes (Fig. 8F). These data are in line with previous research suggesting an elevated catabolic response in skeletal muscle of older adults following acute endurance exercise (49). In summary, although the transcriptional responses are largely similar between groups, subtle differences observed in muscle of older sedentary would suggest unique adaptations to a novel bout of endurance exercise.

DISCUSSION

In this study, we sought to explore novel skeletal muscle transcriptomic signatures at baseline and following an acute bout of endurance exercise in active/endurance trained and sedentary older adults. We reasoned that differences in physical activity status would be reflected in diverse skeletal muscle transcriptome signatures and would result in altered response to exercise. We demonstrate that 1) mitochondrial, type I muscle fiber type, and endothelial cell transcript profiles were elevated in muscle of older active/endurance-trained adults and were associated with key phenotypic characteristics (i.e., V̇o_2peak, ATP_max, fiber type proportion, and muscle mass), 2) immune cell signatures were elevated in muscle of older sedentary adults and were associated with visceral and intermuscular adipose tissue accumulation, and 3), despite baseline phenotypic differences, the transcriptional response to an acute bout of endurance exercise was largely similar between groups. These data suggest that differences in physical activity status and adiposity do not impact the muscle transcriptional response to endurance exercise in older adults in general, when performed at the same relative intensity.

Recent large-scale meta-analyses have identified novel transcriptional responses to exercise in humans (50, 51). However, although important to identify specific molecular targets, a limitation with these types of analyses is the inability to understand biological relevance to the transcript signatures. To infer biological relevance from the transcriptional profiles, we utilized a recently developed bioinformatics approach called PLIER in combination with human clinical and skeletal muscle phenotyping. PLIER performs an unsupervised data structure decomposition to identify sets of correlated transcripts, or latent variables (LVs), while associating the latent variables with specific pathways or cell types (23). PLIER has been used to measure muscle fiber type and mononuclear cell-type proportions in human skeletal muscle at rest and to identify biological processes that are induced by acute resistance exercise in young and older adults (35), as well as chronic resistance exercise training in older adults (52) and Parkinson’s disease patients (42). Our findings provide further evidence that PLIER can be used to assess cell-type composition in human skeletal muscle. Specifically, we observed a high correspondence between LV signatures associated with muscle fiber-type composition, mitochondrial content, and endothelial cells with histological analysis of fiber-type composition of type I slow and type II fast muscle fibers, cardiorespiratory fitness, and in vivo assessment of mitochondrial oxidative capacity. In addition, we were able to analyze baseline and exercise-induced responses of cell types that may be more difficult to assess in skeletal muscle biospecimens (i.e., immune cells). We believe our data provide evidence for the use of PLIER in large-scale studies, including RNA-seq results from previous meta-analyses, to aid in understanding potential changes in cell-type signatures in heterogenous tissues like skeletal muscle between phenotypically diverse human subjects, as well as in response to various acute and chronic exercise interventions.

Chronic inflammation is a hallmark of aging and is related to impaired muscle size and function in older adults (53). Furthermore, it has been shown that a sustained elevation in skeletal muscle inflammatory markers may be associated with an impaired adaptation to exercise training. Specifically, Merritt et al. (54) have shown that elevated inflammatory signal in skeletal muscle was associated with a reduced regenerative capacity in skeletal muscle of older adults following resistance exercise. In the current study, we identified a transcriptional profile (LV42) that was associated with immune cells in skeletal muscle. A possible explanation for the differences in immune cell composition may be due to increase whole body and muscle-specific adiposity in older sedentary adults (Table 1). Aging is associated with an expansion and redistribution of adipose tissue, particularly in visceral adipose tissue depots and skeletal muscle, which are associated with metabolic and function impairments in older adults (55–59). Furthermore, an increase of adipose tissue is associated with elevated inflammatory profile in older adults (56). In line with these findings, we observed that immune cell signatures in skeletal muscle were positively associated with both visceral and intermuscular adipose tissue accumulation (Fig. 6, C and D). In addition to increased adiposity, previous work has shown that a physically active lifestyle may minimize baseline inflammation in circulation (60–63), as well as in muscle (64). Although not statistically significant, we observed that LV42 was negatively associated with daily physical activity (steps per 24 h; Pearson r=−0.420, P = 0.074). Together, these data suggest lower adiposity and chronic exercise training may result in a reduction in circulating and muscle-specific inflammation in older adults.

Despite the elevated immune cell markers in the sedentary group, the acute response to exercise was similar between the two groups. Furthermore, our pathway analysis of transcripts altered by exercise revealed TNF signaling pathway was similarly enriched immediately following and 3-h postexercise timepoint in both groups. This contrasts with previous work which showed an increase in proinflammatory factors (transcripts for TNF-α, TGF-β, and IL-8) in skeletal muscle of healthy older nonexercisers following resistance exercise, that was not apparent in life-long aerobic exercisers (65). A possible explanation for the discrepancy between our studies and those by Lavin and colleagues (65) may be due to different exercise modalities (resistance vs. endurance). An unaccustomed resistance exercise bout, which consists of concentric and eccentric muscle contractions, is associated with muscle damage leading to a potential increased inflammatory response to stimulate repair (66). Interestingly, consequent bouts of resistance exercise are associated with lower muscle damage, leading to a reduced inflammatory response and increased anabolic response to exercise (66, 67). Similar to unaccustomed resistance exercise, endurance exercise, including cycling, increases inflammatory signaling in skeletal muscle (68–70). These changes may also be mediated by muscle damage, a novel bout of endurance exercise can increase serum markers of muscle damage in sedentary younger adults (71, 72), with elevated responses relating to exercising at a higher metabolic intensity (72). However, unlike trained resistance exercise models, muscle from athletes presents an increase in inflammatory signaling following cycling exercise (70), suggesting this acute inflammatory response may provide an important stimulus for both active/endurance trained and sedentary adults. Collectively, these data suggest that chronic endurance exercise training may minimize the effect of aging on muscle inflammation in older adults, yet acute changes in the inflammatory profile of muscle following endurance exercise are similar between active/endurance trained and sedentary older adults.

Previous work has revealed that acute bouts of exercise induce transient transcriptional changes in skeletal muscle that, following cumulative bouts of exercise, result in accrual of key proteins associated with improved muscle health (46). Unfortunately, age appears to be associated with an impaired adaptive response to exercise compared with younger counterparts (14–16), resulting in an inability to increase muscle mass and function following an exercise training intervention. Specifically, studies that have examined the effects of resistance exercise on whole muscle and individual myofiber size and function have shown a limited adaptive response in octogenarian men and women (18, 19). Iversen et al. (21) have shown that skeletal muscle of older sedentary adults can respond to an acute bout of endurance exercise to a similar or greater effect than physically active older counterparts; however, whole transcriptome profiling has not been previously performed following exercise in older active/endurance trained and sedentary adults. Despite differences in physical activity status and adiposity, we observed that both groups responded similarly to an acute bout of endurance exercise. Amar et al. (51) recently performed a linear mixed effect meta-regression of 43 studies examining muscle and blood samples derived from 739 individuals before and after exercise and identified distinct transcriptional patterns in response to acute exercise. When comparing our work to this recent study (51), many of the transcriptional response pathways were similar in regard to time-trajectory between studies, including similar patterns in pathways related to energy metabolism (PPARGC1A and LDLR), blood vessel regulation (VEGFA) that were upregulated at the early mid timepoints and early response in gene regulation mediators (EGR).

In addition, we also observed novel signatures associated with chemokine signaling. Specifically, muscle from both active and sedentary older adults showed an elevation in several chemoattractants, including CCL2, CXCL1, CXCL2, immediately postexercise. A primary role of chemoattractants is to recruit inflammatory cells, such as macrophages and monocytes, to damaged tissue to aid in repair and remodeling processes (73, 74). Investigations on chemokine signaling postexercise have been mostly focused on resistance exercise due to the damaging nature of both concentric and eccentric muscle contractions (75, 76). Although not as well studied, nondamaging stimuli (i.e., endurance exercise) appear to increase macrophage and monocyte infiltration in muscle of young adults (77) potentially due to an increase in chemoattractants such as CXCL2 (78, 79). Interestingly, compared with studies in young adults, we observed an elevation in the expression of chemoattractants immediately postexercise in skeletal muscle of older adults. This is similar to findings from Mathers et al. (75), who observed an early increase in chemoattractants in skeletal muscle of older males and females following a bout of resistance exercise. Whether this early chemokine response aids in adaptations important to aged-muscle (i.e., repair and remodeling) is unclear and requires future examination.

A novel finding from our study was the enrichment in transcripts associated with catabolic processes that was specific to sedentary older adults in muscle samples obtained immediately postexercise. Protein turnover, which is defined by changes in protein synthesis and breakdown rates, is an important mediator of the adaptative responses to exercise (80). Though a primary focus has centered on exercise-induced changes in skeletal muscle protein synthesis rates, acute exercise is also associated with an increase in protein breakdown in skeletal muscle (49, 81). Sheffield-Moore et al. showed that older sedentary adults have an elevation in protein breakdown in skeletal muscle following an acute bout of endurance exercise in comparison with younger counterparts (49). These findings are generally in line with our data showing transcript profiles associated with catabolic processes are enriched in skeletal muscle older sedentary adults following acute endurance exercise. Interestingly, these changes appeared to be unique to older sedentary adults and were not altered in muscle of older active/endurance-trained adults. A possible explanation for this may stem from differences in resting protein turnover in trained versus untrained muscle. Both resistance and endurance exercise training increase protein turnover at rest (82, 83). Interestingly, this increase in resting protein turnover leads to a reduction in protein turnover following an acute bout of resistance exercise (82). Thus, compared with the untrained state in which a novel bout of exercise alters the molecular machinery to stimulate protein turnover, trained muscle may minimize this molecular response due to an elevated protein turnover in the resting state. Although changes in protein turnover have not been examined in the trained and untrained state with endurance exercise in older adults, data from Pikosky et al. (82) suggest that similar to resistance exercise training, endurance exercise training may attenuate the protein turnover response following an acute bout of endurance exercise (83). Future work is needed to test this hypothesis.

Collectively, our data suggest that aged-muscle is able to respond to external exercise stimuli regardless of physical activity or adiposity levels. A caveat to our data is the potential influence of the small meal on the transcriptional response to an acute bout of exercise. Due to the lack of a nonexercise control group, we were not able to delineate postexercise transcriptional signatures that may have been attributed to feeding; thus, future work examining how acute exercise, feeding, and a combination of the two alter the transcriptional program in skeletal muscle of phenotypically diverse older adults is warranted.

In summary, by combining deep human phenotyping and RNA-sequencing, we identified unique transcript signatures that define baseline and exercise-induced cell/pathway specific profiles in skeletal muscle of older adults who are either active/endurance trained or sedentary. These data suggest that elevated mitochondrial and endothelial cell transcript profiles are associated with higher cardiorespiratory fitness in older active/endurance-trained adults, whereas immune cells signatures in skeletal muscle are associated with depot-specific adiposity. Furthermore, we identified unique transcriptional profiles in skeletal muscle defining the immediate and late responses to acute endurance exercise that were similar between phenotypically different groups of older adults. The protective effects of endurance exercise in older active/endurance trained adults, along with the ability to transcriptionally respond to a single exercise bout, suggest that chronic physical activity may be a valuable intervention to improve muscle mass and function in older adults.

SUPPLEMENTAL DATA

Supplemental Figs. S1 and S2 and Tables S1–S5: https://doi.org/10.6084/m9.figshare.16888711.

GRANTS

This work was supported by National Institute on Aging at the National Institutes of Health Grants K01AG044437 (to P.M.C.) and R01AG038576 (to S.W.T.).

DISCLOSURES

P.M.C. is a consultant for Astellas/Mitobridge, Incorporated. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

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