A randomized clinical trial on the effects of exercise on muscle remodelling following bariatric surgery

Abstract Background Muscle atrophy and strength loss are common adverse outcomes following bariatric surgery. This randomized, controlled trial investigated the effects of exercise training on bariatric surgery‐induced loss of muscle mass and function. Additionally, we investigated the effects of the intervention on molecular and histological mediators of muscle remodelling. Methods Eighty women with obesity were randomly assigned to a Roux‐en‐Y gastric bypass (RYGB: n = 40, age = 42 ± 8 years) or RYGB plus exercise training group (RYGB + ET: n = 40, age = 38 ± 7 years). Clinical and laboratory parameters were assessed at baseline, and 3 (POST3) and 9 months (POST9) after surgery. The 6 month, three‐times‐a‐week, exercise intervention (resistance plus aerobic exercise) was initiated 3 months post‐surgery (for RYGB + ET). A healthy, lean, age‐matched control group was recruited to provide reference values for selected variables. Results Surgery resulted in a similar (P = 0.66) reduction in lower‐limb muscle strength in RYGB and RYGB+ET (−26% vs. −31%), which was rescued to baseline values in RYGB + ET (P = 0.21 vs. baseline) but not in RYGB (P < 0.01 vs. baseline). Patients in RYGB+ET had greater absolute (214 vs. 120 kg, P < 0.01) and relative (2.4 vs. 1.4 kg/body mass, P < 0.01) muscle strength compared with RYGB alone at POST9. Exercise resulted in better performance in timed‐up‐and‐go (6.3 vs. 7.1 s, P = 0.05) and timed‐stand‐test (18 vs. 14 repetitions, P < 0.01) compared with RYGB. Fat‐free mass was lower (POST9‐PRE) after RYBG than RYGB + ET (total: −7.9 vs. −4.9 kg, P < 0.01; lower‐limb: −3.8 vs. −2.7 kg, P = 0.02). Surgery reduced Types I (~ − 21%; P = 0.99 between‐group comparison) and II fibre cross‐sectional areas (~ − 27%; P = 0.88 between‐group comparison), which were rescued to baseline values in RYGB+ET (P > 0.05 vs. baseline) but not RYGB (P > 0.01 vs. baseline). RYGB + ET showed greater Type I (5187 vs. 3898 μm2, P < 0.01) and Type II (5165 vs. 3565 μm2, P < 0.01) fCSA than RYGB at POST9. RYGB + ET also resulted in increased capillarization (P < 0.01) and satellite cell content (P < 0.01) than RYGB at POST9. Gene‐set normalized enrichment scores for the muscle transcriptome revealed that the ubiquitin‐mediated proteolysis pathway was suppressed in RYGB + ET at POST9 vs. PRE (NES: −1.7; P < 0.01), but not in RYGB. Atrogin‐1 gene expression was lower in RYGB + ET vs. RYGB at POST9 (0.18 vs. 0.71‐fold change, P < 0.01). From both genotypic and phenotypic perspectives, the muscle of exercised patients resembled that of healthy lean individuals. Conclusions This study provides compelling evidence—from gene to function—that strongly supports the incorporation of exercise into the recovery algorithm for bariatric patients so as to counteract the post‐surgical loss of muscle mass and function.


Introduction
Bariatric surgery is the treatment of choice for severe obesity as it yields substantial weight loss and reversal of cardiometabolic risk factors 1 ; however, this procedure is not free of adverse outcomes. Some post-operative patients exhibit significant loss of fat-free mass, 2 strength, and functionality, 3,4 which is important because these parameters are related to distinct physiological processes (e.g. immune response, regulation of glucose levels, protein synthesis, and basal metabolic rate), 5-7 frailty syndrome, 8 and most critically, mortality. 9,10 The underlying mechanisms related to muscle atrophy following bariatric surgery remain largely underexplored. Individuals with obesity are thought to exhibit a blunted anabolic response to stimuli like exercise bouts, a condition known as anabolic resistance. 11 This resistance has been ascribed to (i) reduced activation of proteins related to the mTOR pathway (a master regulator of muscle protein synthesis) 12 ; (ii) altered proteolytic activity (e.g. the ubiquitin-proteasome system and autophagic/lysosomal system) 13 ; (iii) microvascular rarefaction 14 ; and (iv) reduced myogenic response. 15 Whether and how these mechanisms contribute to surgery-induced muscle atrophy remains to be determined.
Exercise improves muscle mass, strength, and functionality in a wide spectrum of sub-populations; however, whether exercise preserves muscle mass in post-bariatric patients is controversial. [16][17][18][19] Indeed, Stegen et al. 18 observed higher muscle strength levels in post-bariatric surgery patients undergoing 12 weeks of exercise training than non-exercised patients; however, exercise did not affect fat-free mass sparing or functionality. Conversely, Herring et al. 17 demonstrated that 12 weeks of exercise training attenuated the loss in fat-free mass and resulted in greater strength and functionality gains in patients undergoing bariatric surgery vs. a non-exercised group. It is noteworthy that both studies evaluated fat-free mass using electrical bioimpedance, a method that is neither an accurate measurement of fat-free mass in individuals with obesity 20 nor of muscle mass. To the best of our knowledge, the only study that directly assessed muscle mass (i.e. magnetic resonance imaging) in post-bariatric surgery patients undergoing exercise training did not detect any effect of exercise on muscle mass sparing. 19 The dissonant findings are hard to reconcile, and in fact, literature is scarce on this subject, warranting further investigation. Moreover, the studies investigating the effects of exercise on muscle mass and function of post-bariatric surgery patients have failed to identify mechanisms underlying exercise-mediated muscle remodelling in this population. We 21,22 and others 23,24 have shown that exercise promotes beneficial physiological and metabolic effects that are additive to the weight loss effects of surgery, including effects on bone mass, inflammatory markers, endothelial function, and insulin sensitivity.
Herein, we performed a randomized, controlled, mechanistic trial to comprehensively examine the effects of exercise training on body composition (fat-free mass as primary outcome), muscle function and related cellular and molecular mechanisms (secondary outcomes) in women undergoing bariatric surgery. We hypothesized that exercise training following bariatric surgery would attenuate loss of fat-free mass and conserve muscle function when compared with standard of care. We further proposed that histological and molecular analysis would generate new insights on the muscle remodelling mechanisms resulting from the surgical procedure and exercise.

Study design and human subjects
This was a randomized controlled trial (registered at Clinicaltrials.gov as NCT02441361) that a priori aimed to investigate the effects of exercise training following bariatric surgery on multiple clinical outcomes. Specifically, this manuscript focuses on the effects of exercise training on muscle mass and function after bariatric surgery, as well as in the molecular and histological mediators of muscle remodelling. Additional reports from this trial contain data on endothelial function, bone mass, and insulin resistance were published elsewhere. 21,22,25 Data were collected between March 2015 and September 2019 in São Paulo (Brazil). The study was approved by the Ethics Committee of the Clinical Hospital of the School of Medicine of the University of São Paulo, and all procedures were in accordance with the recommendations of the Helsinki Declaration. All patients provided written informed consent before being enrolled in the study. This manuscript is reported according to the guidelines of the Consolidated Standards of Reporting Trials.
Patients were recruited from the Metabolic and Bariatric Surgery Unit of the Clinical Hospital of the School of Medicine of the University of São Paulo. Inclusion criteria were as follows: women who were eligible for bariatric surgery [body mass index (BMI) > 40 kg/m 2 or ≥35 kg/m 2 with associated comorbidities], 18-60 years, and not engaged in an exercise training programme for at least 1 year prior to the study.
Exclusion criteria involved cancer in the past 5 years, and any cardiovascular diseases, neurological disorders, or skeletal muscle impairment that would contraindicate exercise practice. None of the patients were receiving hormone replacement therapy.
Before surgery, patients were randomly assigned (1:1) into either a Roux-en-Y gastric bypass (RYGB) or Roux-en-Y gastric bypass plus exercise training group (RYGB + ET), using a computer-generated randomization code. Clinical and laboratory parameters were assessed at baseline (PRE), and 3 (POST3) and 9 (POST9) months after the surgery. The 6 month exercise intervention (for the exercise group) started at POST3 upon post-surgery medical clearance for high-intensity exercise participation. A healthy control group comprising lean, age-matched women was also recruited and assessed to provide reference values for selected variables.

Exercise intervention
The RYGB + ET group participated in a 6 month, three-timesa-week, supervised exercise training programme at the hospital, whereas the RYGB group received standard post-surgery care during follow-up. Training sessions included a 5 min light warm-up followed by strengthening exercises for the major muscle groups (leg-press 45°, leg extension, half-squat, bench press, lat pulldown, seated row, and calf raise) and aerobic exercise on a treadmill. The resistance exercise protocol was comprised of three sets of 8-12 repetition maximum with a 60 s rest interval between sets and exercises. Load progression (5%) was employed as soon as patients were able to perform two or more repetitions than previously determined. Aerobic training consisted of 30-60 min (10 min progression every 4 weeks) of treadmill walking at an intensity corresponding to 50% of the delta difference between the ventilatory anaerobic threshold and respiratory compensation point. Heart rate was monitored throughout every session to ensure proper exercise intensity (Polar®). Any adverse events or signs and symptoms were documented including feelings of general fatigue, soft tissue soreness, and injury or illness; and a record of attendance was kept ensuring adherence to the protocol.

Body weight and composition
Body weight was assessed on a calibrated digital scale and height was evaluated with the aid of a stadiometer, from which BMI was calculated. All patients underwent a wholebody dual-energy x-ray absorptiometry scan (GE Healthcare®) to quantify fat mass and lean mass using CoreScan ™ software. All dual-energy x-ray absorptiometry measurements were carried out by the same trained technician.

Muscular strength rest (1-RM) and functionality
Before testing, all patients performed two familiarization sessions separated by at least 72 h. Lower-and upper-limb maximal dynamic strength was assessed in the leg-press (45°) and bench-press, respectively (Nakagym, São Paulo, Brazil). Coefficient of variation for 1RM in the leg-press and bench-press were 2.8% and 1.1%, respectively. Functionality was measured by the timed-stands test and the timed-up-and-go test. The timed-stands test evaluates the maximum number of stands that an individual can perform from a standard-height armless-chair (i.e. 45 cm), whereas the timed-up-and-go test registers the minimal time (in seconds) that each individual requires to rise from a standard chair, walk to a line on the floor 3 m away, turn around, and sit down again. Two attempts were allowed for each test. All tests were conducted by the same investigator to avoid bias. Coefficient of variation for timed-stands test and timed-up-and-go test were 2.1% and 3.5%, respectively.

Physical activity monitoring
Physical activity levels and sedentary behaviour were objectively measured using ActiGraph GT3X® accelerometers (ActiGraph, Pensacola, Florida). Patients were instructed to wear the accelerometer for seven consecutive days during waking hours, except when bathing. Physical activity was assessed during non-training periods [i.e. at PRE, before the commencement of exercise intervention (POST3), and after the completion of the exercise intervention (POST9)]. All patients were required to accumulate at least 10 h of valid activity recordings per day for at least five days. The accelerometer was worn on an elastic belt at the waistline on the right side of the hip. To accurately register when the device was worn and removed, patients kept a daily log. Data were exported from the device in 60 s epochs, using ActiLife 6 software (ActiGraph, Pensacola, Florida). Non-wear periods were defined as intervals of at least 60 min of zero activity counts, assuming a tolerance of no more than 2 min of counts between 0 and 100. Freedson et al. 26 cut-points were used to define epochs as follows: sedentary time (<100 counts per minute), light-intensity physical activity (≥100 to <1952 counts per minute), and moderate-to-vigorous physical activity (≥1952 counts per minute). Sedentary behaviour and moderate-to-vigorous physical activity data are presented as minute per day.

Muscle biopsies
All percutaneous muscle biopsy procedures were performed by a trained physician. The patients assumed a comfortable reclining position with both legs outstretched. The vastus lateralis biopsy site, (at a point 10-15 cm proximal from the tibial tuberosity and 5 cm lateral from the midline of the femoral course), was prepared with chlorhexidine gluconate solution. Skin and subcutaneous tissue were infiltrated with a 3 mL lidocaine hydrochloride 2% w/v solution, ensuring that the fascia lata was well infiltrated. After ensuring adequate local anaesthesia, an incision was made in the lateral thigh at the biopsy site directly through the overlying skin, subcutaneous fat, and fascia lata. A 5 mm modified Bergström needle was then inserted through the fascia, and an assistant immediately applied suction using a 60 mL syringe connected to a canister and attached to the top of the needle. A muscle sample was removed, and the skin was closed with skin closure tape. To minimize risk of infection and bruising, a pressure dressing was immediately applied and maintained for 72 h. All post-intervention biopsies were performed 72 h following the last exercise session at an adjacent distal incision to the baseline site, approximately 0.5 cm apart. There were no adverse events related to the procedure.

Immunohistochemistry
Frozen muscles were embedded in optimal cutting temperature compound and were sectioned (7 μm thick) on a cryostat. For fibre immunostaining assay, muscle samples were brought to room temperature and fixed in methanol for 10 min, washed [three 10 min washes with phosphate-buffered saline (PBS)], then blocked for 60 min in a blocking buffer solution (containing 1% PBS, 5% bovine serum albumin, and 0.3% Triton X-100). The slides were then incubated with primary antibody overnight at 4°C. The next day, slides were again washed (three 10 min washes with PBS) and incubated in appropriate secondary antibody in the dark for 1 h at room temperature. Nuclei were labelled using fluoroshield mounting medium with DAPI (Ab104139, Abcam), prior to cover slipping slides. Images were captured with an Olympus BX51 Fluorescence microscope with a magnification of ×20. For Type I and Type II fibre cross-sectional area (fCSA), myonuclei and satellite cell content (SC), a minimum of 120 fibres per subject at each time point was analysed. 27 The quantification of muscle fibre capillaries [capillary contacts (CC, the number of capillaries around a fibre), capillary-to-fibre ratio on an individual fibre basis (C/Fi), and capillary-to-fibre perimeter exchange index (CFPE, an estimate of the capillary-to-fibre surface area)] were performed on 50 muscle fibres/subject/time point, as previously described. 28 All immunohistochemical images were analysed by a single researcher in a blinded manner. Coefficient of variation between two blinded measurements were 3.7%, 3.5%, 3.7%, 3.3%, 3.3%, 4.4%, 2.6%, and 3.4% for Types I and II muscle fCSA, Types I and II muscle fibre myonuclei content, Types I and II muscle fibre myonuclear domain, CC, C/Fi, CFPE, and SC, respectively.

RNA-sequencing and bioinformatics analysis
For RNA sequencing (RNA-seq), muscle RNA was isolated (six patients per group) using the RNeasy Fibrous Tissue Mini Kit (Qiagen®). Total RNA concentration was assessed spectrophotometrically at 260 nm (GE Healthcare®) and RNA integrity number (RIN ≥ 7) was checked using a Bioanalyzer 2100 (Agilent®). RNA library construction and sequencing were performed in the Genomics Core at Pennington Biomedical Research Center. Sample concentration was normalized and cDNA pools were created for each sample, and subsequently tagged with a barcoded oligo adapter to allow for sample specific resolution using the Quant-seq® 30 mRNA-Seq Library Prep Kit (Lexogen®). Sequencing was carried out using an Illumina Seq 2500 platform (Illumina®) with 75 BP single end reads. Reads were aligned to the human-reference genome (GRCh38.77). Annotation and preparation of the sequencing data for downstream analysis was performed by custom R scripts 29 with annotations downloaded from BioMart (R. V3.4.3, Rstudio V1.1.423 and biomaRt V2.24.1). Quant-Seq raw data are available in the NCBI GEO database (accession GSE 137631).
Raw count data were normalized via TMM normalization in edgeR 30 and adjusted for mean-variance effects via the voom function in limma. 31 Differential gene expression analysis was performed via the limma empirical Bayes analysis pipeline by adjusting the model design matrix using paired or unpaired analysis, depending upon the specific contrasts. Genes with a false discovery rate (FDR) ≤ 0.05 were considered as differentially expressed. Pathway enrichment analysis was carried out via the gene set enrichment analysis (GSEA) pre-ranked method by querying a 'custom' database consisting of genesets from Wikipathways (www.wikipathways.org) plus user-defined gene sets corresponding to specific biological functions. Pathway enrichments with an FDR ≤ 10% and the absolute normalized enrichment score (NES) ≥ 1.5 were considered significant.

Quantitative real-time PCR
Total muscle RNA was lysed in 1 mL of TRIzol reagent (Invitrogen®) and RNA was isolated using standard techniques. The Superscript Platinum One-Step kit (Invitrogen®), with incorporated Maxima SYBR Green/ROX qPCR Master Mix (Thermo Fisher Scientific®) on a Step One ™ Real-Time PCR System (Applied Biosystems®) was used to perform the qRT-PCR experiments. All primers were purchased from Invitrogen®. The fold changes from PRE were calculated using Exercise, bariatric surgery, and skeletal muscle the 2 ÀΔΔCt method, 32 and all mRNA levels were normalized to beta-2-microglobulin (β2M).

Western blotting
Protein extraction followed by quantification was performed via Western blotting as described previously. 33 In brief, cell lysates (30-40 μg) were resolved by sodium dodecyl polyacrylamide gel electrophoresis, electrophoresis to PVDF membrane, and incubated overnight with appropriate primary antibodies followed by incubation in appropriate HRP tagged secondary antibody. Signals were revealed by enhanced chemiluminescence ECL reagent (SuperSignal ™ West Femto Substrate, Thermo Fisher Scientific). Gel-to-gel variation and equal protein loading were controlled using a standardized sample on each gel. Results were corrected to Ponceau red staining (0.5%, w:v) of the membrane for all immunoblot assays. 34 Blots were quantified using the Image-J software®. Details of all reagents and resources are provided in supporting information, Table S1.

Sample size calculation and statistical analysis
Sample size was determined with the aid of a G-Power software (version 3.1.2-Universitat Kiel, Germany) package. The analysis was conducted by inputting α error (0.05), power (1 À β error = 0.90), and effect size (Hedges' g = 0.20), considering previous data indicating moderate between-group effect size (RYGB vs. RYGB + ET) for fat-free mass in individuals undergoing bariatric surgery. 18 Calculations were based on an ANOVA with repeated measures (withinbetween interactions) and the total sample size was determined to be 56 patients. To account for midtrial withdrawals, the sample was increased by~40% (i.e. 80 patients).
Data are presented as mean ± standard deviation (SD), estimated mean difference between groups (EMD) at POST9 (only in the presence of group x time interaction), 95% confidence interval (95%CI), unless otherwise indicated. Baseline characteristics were compared between the two groups using either Student's t test or Fisher exact test. The data were analysed using an intention-to-treat (ITT) approach to preserve the integrity of randomization. Dependent variables were analysed by a mixed model analysis for repeated measures, using group (RYGB and RYGB + ET group) and time (PRE, POST3, and POST9) as fixed factors, and subjects as a random factor using SAS software version 9.1. Post-hoc tests with Tukey's adjustment were performed for multiple comparisons. Additionally, in order to verify efficacy of the exercise training in body composition parameters, independent t tests were performed to test possible between-group differences in absolute changes (Δ) between POST9 to PRE. To compare experimental groups (RYGB vs. RYGB + ET) at POST9 with reference values from healthy women lean control group, values of muscle strength, functionality, capillarization, and SC content were transformed into standardized scores (i.e. Z-score) having mean ± SD values of the CTRL as reference for each assessment. In this analysis it is assumed that the closer the Z-scores are to zero the more comparable they are to the healthy women in the lean control group. In addition, in order to statistically compare the Z-score of each dependent variable of the non-exercise and exercise group with women in the healthy lean control group, a one-way ANOVA was performed, and in case of F-value significant, post-hoc tests with Dunnett's adjustment were performed for multiple comparisons. Significance level was set at P ≤ 0.05. To facilitate the reader, main effects of time are described as effects of surgery (POST3-PRE), whereas group × time interactions at POST9 are described as effects of exercise, unless otherwise stated.

Flow of participants
The flow of participants through the study is depicted in Figure S1. Three-hundred and thirty-eight patients were screened for participation. Eighty patients met inclusion criteria and were randomly (computer-generated randomization code) assigned (1:1) to either RYGB (n = 40) or RYGB + ET (n = 40). Thirteen patients dropped out of the RYGB group (did not perform surgery: n = 9; personal reasons: n = 4) and 12 dropped out from the RYGB + ET group before the intervention (did not perform surgery: n = 7; personal reasons: n = 5). Patients from the RYGB + ET group attended 83 ± 11% of the exercise training sessions. Both groups were comparable regarding baseline characteristics (Table 1).
Muscle fibre cross-sectional area, myonuclei content, myonuclear domain, muscle fibre capillarization, and satellite cell content To further examine the effect of exercise on muscle remodelling, we performed immunostaining assays from vastus lateralis samples to assess adaptations at the muscle fibre level, specifically Type I and Type II muscle fCSA, satellite cell content, and capillarization ( Figure 3, Panels A, H, and L, respectively). Type I and Type II fCSA decreased 3 months following surgery in both groups (main effect of time: both P < 0.0001). The RYGB + ET group showed greater Type I fCSA (EMD: 1289.2 μm 2 , 95% CI: 355.7 to 2222.6, P = 0.0025) and Type II fCSA (EMD: 1600.1 μm 2 , 95% CI: 468.4 to 2731.8, P = 0.0019) than the RYGB group (Figure 3, Panels B and C) at POST9. Of relevance, Types I and II fCSA in the RYGB + ET reached pre-surgery levels (P > 0.05). Data are expressed as mean ± SD or number of patients (%). 1-RM: maximal dynamic strength; ACE, angiotensin-converting enzyme; MVPA, moderate-to-vigorous physical activity; RCP, respiratory compensation point; RYGB + ET: Roux-en-Y gastric bypass plus exercise training group; RYGB: Roux-en-Y gastric bypass plus non-exercise group; VAT, ventilatory anaerobic threshold; VO 2 peak , peak oxygen uptake.
Exercise, bariatric surgery, and skeletal muscle The satellite cell pool was increased after the intervention in the RYGB + ET group, whereas no changes were noted in the RYGB group across time (both P > 0.05). Of relevance, the RYGB + ET group had a larger satellite cell pool than the RYGB group after the intervention (EMD: 0.04 satellite cell/fibre; 95% CI: 0.01 to 0.06; P < 0.0001) (Figure 3, Panel I). Both groups demonstrated reduced Type I muscle fibre myonuclei content at POST3 and POST9 vs. PRE (P < 0.0001 and P = 0.0308, respectively).
Our data indicated a decreased expression of the autophagy system-related genes BNIP3 and CTSL1 at 3 and 9 months after the surgery, independent of exercise (both P < 0.0001) ( Figure S3, Panels A and C, respectively). mTOR pathway-related proteins, p-4E-BP1/4E-BP1 was reduced at POST9 and POST3 vs. PRE, independent of exercise (main effect of time: P = 0.0125 and P = 0.0412, respectively) ( Figure S4, Panel I). However, no significant changes were observed in mTOR and P70S6K because of surgery and exercise ( Figure S4, Panels A-H).

Muscle genotype and phenotype of post-bariatric patients in comparison with healthy lean controls
To investigate the extent to which exercise may rescue muscle genotype and phenotype in patients who underwent bariatric surgery, we compared the RNA-seq data from RYGB + ET and RYGB groups (at POST9) vs. healthy lean control group. In comparison with controls, the RYGB + ET group showed a much lower number of differentially expressed genes (276 genes) than the RYGB group (544 To further examine whether skeletal muscle phenotype also follows the same pattern, muscle strength, functionality, fCSA, capillarization, and SC, data were transformed into standardized scores (i.e. Z-score), and then the RYGB + ET Figure 5 Normalized skeletal muscle cell signalling pathways and physiological outcomes. RYGB + ET: Roux-en-Y gastric bypass plus exercise training group; RYGB: Roux-en-Y gastric bypass plus non-exercise. Venn diagram of the total number of differentially expressed genes (FDR ≤ 0.05), and gene set enrichment analysis (GSEA) related to muscle protein synthesis and breakdown in the RYGB + ET and RYGB groups at POST9 in comparison with healthy lean control group (Panels A and B, respectively). Comparison of the RYGB + ET and RYGB groups at POST9 with reference values from healthy lean control group: Z-score for absolute and relative lower-limb (Panels C and D, respectively) and upper-limb (Panels E and F, respectively) strength performance in the 1-RM test, functionality in the timed-up-and-go test and timed-stand test (Panels G and H, respectively), Types I and II muscle fibre cross-sectional area (CSA) (Panels I and J, respectively), muscle fibre capillarization parameters (Panels K, L, and M, respectively) and satellite cell content (Panel N). Data are expressed as mean ± SD. Red arrow indicates an enrichment with a false discovery rate (FDR) ≤ 10% and the absolute log2 foldchange >1.5; PRE: before surgery (baseline); POST3: 3 months following surgery; POST9: 9 months following surgery. * indicates P < 0.05 in comparison with healthy lean control group. and RYGB groups were compared with controls. The RYGB + ET group and controls showed similar values in absolute and relative lower-limb muscle strength and absolute upper-limb muscle strength (P = 0.0531, P = 0.0620, P = 0.9999, respectively). In contrast, the RYGB group exhibited reduced values of absolute and relative lower-limb muscle strength and absolute upper-limb muscle strength vs. controls (P = 0.0002, P < 0.0001, P = 0.0171, respectively) ( Figure 5, Panels C-F). The RYGB + ET group demonstrated even faster scores in the timed-stand test than controls (P < 0.0025), while the RYGB group showed slower scores than controls (P < 0.0001). Both groups exhibited lower performance in the relative upper-limb muscle strength test and timed-up-and-go test as compared with controls (P < 0.05 for all) (Figure 5, Panels F and G, respectively).

Discussion
This trial demonstrated that a 6 month, three-times-a-week, supervised exercise training programme attenuated the loss of fat-free mass and reversed the loss of muscle strength among women who had bariatric surgery. The exercise training programme completely reversed surgery-induced muscle fibre atrophy, and this was accompanied by increases in capillarization and satellite cell content. Interestingly, RNA-seq and RT-PCR analyses support the hypothesis that exercise suppressed the ubiquitin-proteasome system, which may partially explain the beneficial effect of exercise on muscle remodelling. Of relevance, genotype and phenotype data revealed that skeletal muscle of patients who exercised resembled that of healthy lean individuals.
Bariatric surgery is the treatment of choice for severe obesity, but this approach is not free of adverse effects. In consonance with previous studies, 3,4 we observed a drastic decrease in the absolute and relative lower-limb and upper-limb muscle strength 3 months following surgery, whereas no significant change was observed in functionality and relative strength. In contrast, the addition of a post-surgery exercise training programme demonstrated to be an effective strategy to reverse the loss of absolute and relative lower-limb muscle strength and relative upper-limb muscle strength following surgery. It is noteworthy that exercised patients also increased absolute upper-limb muscle strength following intervention, but it did not return to baseline values. Moreover, the combination of surgery plus exercise training provided improvements in functionality to a greater extent than that seen with surgery/standard of care 9 months following intervention. These data corroborate the hypothesis that exercise can offset the decline in muscle strength and function that is associated with bariatric surgery. 16,17 Conversely, physical activity levels were unchanged in both groups throughout the study, supporting the idea that neither bariatric surgery nor structured exercise training are effective to promote profound increases in unstructured physical activity levels in this population. 35 Muscle atrophy and dysfunction following bariatric surgery is partially attributed to a substantial decrease in fat-free mass. [16][17][18] In agreement with previous studies, BMI, fat mass, total fat-free mass, and lower-limb and upper-limb fat-free mass were reduced at 3 and 9 months after surgery. While lifestyle interventions (e.g. exercise training and nutrition) are recommended to attenuate the loss of fat-free mass among patients undergoing bariatric surgery, 36,37 efficacy is still debated. [16][17][18] In this study, the losses in total and lower-limb fat-free mass were attenuated in the RYGB + ET in comparison with the RYGB group, supporting previous findings. 16,17 Of note, the benefit of exercise on total and lower-limb fat-free mass was independent of the relatively low protein consumption reported by the patients (i.e. <1 g·kg À1 ·day À1 ) ( Table S2), suggesting that exercise per se is an important anabolic stimulus for this population, although the between-group differences for total and lowerlimb fat-free mass were relatively small. Contrary to total and lower-limb fat-free mass data, exercise training was not able to mitigate upper-limb fat-free mass loss, despite the observed increase in upper-limb muscle strength. A possible explanation may be related to the effects of aerobic exercise on muscle mass in untrained individuals. 38 In this context, it is possible to speculate that the aerobic exercise (treadmill) may have provided an additional anabolic stimulus to the lower-limb but not to the upper-limb muscles, especially considering that patients still presented with high body mass values, even after surgery (BMI~36 kg/m 2 ). Gains in upper-limb muscle strength may be linked to neural adaptations such as increase in the firing rate, inter-and intra-muscular coordination, motor unit synchronic improvement. 39 To further examine the effect of exercise on muscle remodelling at the muscle fibre level, we performed immunostaining assays from vastus lateralis samples. Our findings demonstrate a decrease in Type I and Type II fCSA 3 months following surgery in both groups. In contrast, the addition of a post-surgery exercise training routine increased Type I and Type II fCSA when compared with the non-exercised group. Furthermore, exercise was able to fully reverse the muscle fibre atrophy induced by surgery, as Types I and II fCSA were increased back to pre-surgery levels in RYGB + ET group.
Muscle fibre remodelling depends on the pool of satellite cells, which are localized between the basal lamina and sarcolemma. When activated, satellite cells proliferate and differentiate to supply additional myonuclei to the fibre or replenish the resident pool of satellite cells. 40 We observed that exercise training increased the pool of satellite cells after the intervention, but muscle fibre myonuclei content remained unchanging, challenging the myonuclear domain hypothesis, according to which, during the hypertrophic process, myonuclei accretion is required to maintain a given surrounding territory of cytoplasm under the control of a given number of myonuclei. In another perspective, satellite cells are also thought to regulate fibrogenic cell collagen expression and, hence, muscle fibre extracellular matrix remodelling, another key factor related to muscle fibre remodelling. 41 In this regard, we recently demonstrated that bariatric surgery decreased muscle extracellular matrix deposition, an effect that was potentiated by exercise. 25 Taken together, these findings suggest that satellite cells may have been predominantly driven towards skeletal muscle extracellular matrix remodelling, rather than myonuclear accretion.
The delivery of molecular signals [e.g. insulin-like growth factor-1 (IGF-1), hepatocyte growth factor (HGF), interleukin-6 (IL-6) and myostatin] to skeletal muscle cells is highly dependent on muscle fibre capillarization, a key aspect in muscle remodelling. 42 The formation of new capillaries in myofibres depends on a complex process termed angiogenesis, which is regulated by anti-angiogenic and pro-angiogenic factors. Our data indicate that exercise increased the expression of different angiogenesis markers (i.e. ANGP1, ANGP2, TEK, and VEGF), which was accompanied by increased muscle fibre capillarization. Collectively, these findings suggest that exercise increased muscle fibre perfusion via increased capillarization (mediated by alteration in the expression of numerous angiogenesis markers), possibly resulting in a more efficient delivery of oxygen, nutrients, and growth factors necessary for muscle fibre growth.
To gather mechanistic insights on the effects of surgery combined with exercise on muscle remodelling-related pathways, we performed RNA-seq and RT-PCR analysis of muscle samples that revealed that the ubiquitin-mediated proteolysis pathway and Atrogin-1 gene expression was suppressed in the RYGB + ET group at POST9 vs. PRE, but not in the RYGB group. This suggests that exercise mitigated muscle protein breakdown mediated by the ubiquitin-proteasome system and may represent an important part of the effect of exercise on muscle size.
The autophagic system is responsible for removing and eliminating unfolded and toxic proteins as well as abnormal and dysfunctional organelles, through the formation of a double membrane vesicle that engulfs portion of the cytoplasm, organelles, glycogen, and protein aggregates, such as those that accumulate within the muscle cells of individuals with obesity. 43 Of interest, we verified a decrease in the expression of autophagy system-related genes (i.e. BNIP3 and CTSL1) following surgery, which was not affected by exercise. These findings suggest that bariatric surgery induces a lower autophagic flux, possibly because of the substantial reduction in adiposity, which, in turn, may reduce the number of toxic proteins and dysfunctional organelles. This may represent a new mechanism associated with the beneficial effect of bariatric surgery on decreased systemic inflammation and endoplasmic reticulum stress. mTOR pathway is recognized as a key regulator of muscle anabolism. Previous studies have shown that calorie restriction suppresses the expression of mTOR pathway-related proteins, 44 while mechanical stimuli are known to activate them. 45 We were unable to identify significant changes in mTOR pathway-related proteins (i.e. mTOR and P70S6K) as a result of surgery and exercise, with the exception of p-4E-BP1/4E-BP1 that was reduced at POST9 and POST3 in comparison with baseline values. It is possible that the absence of change in these proteins was influenced by the timing of the muscle biopsy, because activation of these proteins appears to be dependent on an acute anabolic stimulus (e.g. protein ingestion or resistance exercise). 44,46 Therefore, caution should be exercised in concluding that exercise and surgery did not influence the mTOR pathway in this study.
Muscle mass is regulated by the coordinated balance between the rates of muscle protein synthesis and muscle protein breakdown. 47 Our findings suggest a complex network of mechanisms related to muscle fibre growth in patients with obesity undergoing bariatric surgery. Despite cellular adaptations occurring in the absence of chronic adaptations in proteins related to mTOR signalling pathway and in myonuclei content, exercise was shown to mitigate muscle protein breakdown, an observation that is supported by reduced gene expression in ubiquitin-proteasome system related proteins. The increased muscle fibre capillarization and satellite cell content may also help explain our results, as these adaptations may have positively affected the spatial proximity between satellite cell and capillaries, facilitating the delivery of growth factors, thereby improving the myogenic response to exercise. 42 Although it is reported that exercise increases myonuclei content during hypertrophy-oriented conditions, 48 our data suggest that the existing myonuclei in the exercised patients are capable of sustaining new transcriptional requirements of the muscle fibre in order to promote protein accretion under conditions that are unfavourable to hypertrophy (e.g. negative energy balance). It is difficult to put these novel data into perspective with the literature, as these are the first data to have assessed changes in fat-free mass and histological and molecular parameters related to muscle fibre adaptations in obese patients undergoing bariatric surgery and exercise training.

Online supplementary material
Additional supporting information may be found online in the Supporting Information section at the end of the article. Table S1. List of reagents and resources. Table S2. Dietary intake. Figure S1. Study Recruitment. RYGB+ET: Roux-en-Y Gastric Bypass plus Exercise Training group; RYGB: Roux-en-Y Gastric Bypass plus non-exercise. ITT: intention-to-treat. Figure S2. Expression of Genes and Protein Related to Angiogenesis. RYGB+ET (n = 14): Roux-en-Y Gastric Bypass plus Exercise Training group; RYGB (n = 14): Roux-en-Y Gastric Bypass plus non-exercise. Gene expression of the HIF1-α, ANGPT1, ANGPT2, MDM2, TEK and THSP1 (Panel A, B, C, D, E and F, respectively). Protein expression of the VEGF (Panel G). Representative image of the Western blot bands for VEGF (Panel H). Data are expressed as mean ± SD. PRE: before surgery (baseline); POST3: 3 months following surgery; POST9: 9 months following surgery.^indicates P < 0.05 for main effect of time; * indicates P < 0.05 in comparison to PRE; # indicates P < 0.05 in comparison to POST3; $ indicates P < 0.05 for between-group comparison at POST9. as mean ± SD. PRE: before surgery (baseline); POST3: 3 months following surgery; POST9: 9 months following surgery.^indicates P < 0.05 for main effect of time. Figure S4. Expression of Proteins Related to mTOR Pathway Along the Intervention Period. RYGB+ET (n = 12): Roux-en-Y Gastric Bypass plus Exercise Training group; RYGB (n = 12): Roux-en-Y Gastric Bypass plus non-exercise. Protein expression of the p-mTOR, mTOR, p-mTOR/mTOR ratio (Panel A, B and C, respectively). Protein expression of the p-p70S6K, p70S6K, p-p70S6K/p70S6K ratio (Panel D, E and F, respectively). Protein expression of the p-4E-BP1, 4E-BP1, p-4E-BP1/4E-BP1 ratio (Panel G, H and I, respectively). Representative image of the Western blot of the proteins related to mTOR pathway (Panel J). Data are expressed as mean ± SD. PRE: before surgery (baseline); POST3: 3 months following surgery; POST9: 9 months following surgery.^indicates P < 0.05 for main effect of time.