Three Months of Bed Rest Induce a Residual Transcriptomic Signature Resilient to Resistance Exercise Countermeasures

Received: 27 November 2019 | Revised: 19 February 2020 | Accepted: 29 March 2020 DOI: 10.1096/fj.201902976R

RESEARCH ARTICLE

Three months of bed rest induce a residual transcriptomic signature resilient to resistance exercise countermeasures

1Department of Laboratory Medicine, Division of Clinical Physiology, Karolinska Abstract Institutet, and Unit of Clinical Physiology, This study explored the muscle genome-wide response to long-term unloading (84- Karolinska University Hospital, Stockholm, day bed rest) in 21 men. We hypothesized that a part of the bed rest-induced gene Sweden expression signature would be resilient to a concurrent flywheel resistance exercise 2Department of Physiology & Pharmacology, Karolinska Institutet, (RE) countermeasure. Using DNA microarray technology analyzing 35 345 gene- Stockholm, Sweden level probe-sets, we identified 335 annotated probe-sets that were downregulated, 3 Department of Orthopaedics, Region and 315 that were upregulated after bed rest (P < .01). Besides a predictable differen- Jönköping County, Eksjö, Sweden tial expression of genes and pathways related to mitochondria (downregulation; false- 4Department of Biomedical and Clinical Sciences, Linköping University, Linköping, discovery rates (FDR) <1E-04), ubiquitin system (upregulation; FDR = 3E-02), and Sweden skeletal muscle energy metabolism and structure (downregulation; FDR ≤ 3E-03), 84- day bed rest also altered circadian rhythm regulation (upregulation; FDR = 3E-02). Correspondence Rodrigo Fernandez-Gonzalo, Department While most of the bed rest-induced changes were counteracted by RE, 209 transcripts of Laboratory Medicine, Division of were resilient to the exercise countermeasure. Genes upregulated after bed rest were Clinical Physiology, Karolinska Institute, particularly resistant to training (P < .001 vs downregulated, non-reversed genes). and Unit of Clinical Physiology, Karolinska University Hospital, ANA Futura, Specifically, “Translation Factors,” “Proteasome Degradation,” “Cell Cycle,” and Alfred Nobels Allé 8, 141 52 Huddinge, “Nucleotide Metabolism” pathways were not normalized by RE. This study provides Stockholm, Sweden. Email: [email protected] an unbiased high-throughput transcriptomic signature of one of the longest unloading periods in humans to date. Classical disuse-related changes in structural and meta- Funding information bolic genes/pathways were identified, together with a novel upregulation of circadian Swedish National Space Agency; European Space Agency (ESA); Swedish National rhythm transcripts. In the context of previous bed rest campaigns, the latter seemed Center for Research in Sports to be related to the duration of unloading, suggesting the transcriptomic machinery continues to adapt throughout extended disuse periods. Despite that the RE training offset most of the bed rest-induced muscle-phenotypic and transcriptomic alterations,

Abbreviations: BR, bed rest; BRE, bed rest and exercise; DAVID, The Database for Annotation, Visualization and Integrated Discovery; FC, fold change; FDR, false discovery rate; Go-term, gene ontology-term; Kcal, kilocalorie; LIMMA, linear models for microarray data; MEDES, Space Medicine and Physiology clinic, Toulouse, France; NUSE, normalized unscaled standard error; PCA, principal component analysis; RE, resistance exercise; t-SNE, t-distributed stochastic neighbor embedding Eric Rullman and Thomas Gustafsson Joint senior authors.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2020 The Authors. The FASEB Journal published by Wiley Periodicals LLC on behalf of Federation of American Societies for Experimental Biology

wileyonlinelibrary.com/journal/fsb2 | 1 The FASEB Journal. 2020;00:1–12. 2 | FERNANDEZ-GONZALO et al.

we contend that the human skeletal muscle also displays a residual transcriptomic signature of unloading that is resistant to an established exercise countermeasure.

KEYWORDS atrophy, microarray, muscle wasting, unloading

1 | INTRODUCTION paradigm was able to preserve mm. quadriceps mass, and halved the atrophy in m. triceps surae after 90 days of bed rest In response to reduced mechanical stimuli, the skeletal in men.1 We also showed that this protocol was effective in muscle experiences loss of mass and function, increased fa- counteracting some, but not all, of the metabolic alterations tigability and insulin resistance, and a shift toward a faster in muscle triggered by 84-day bed rest.21 However, whether muscle fiber phenotype.1-6 The loss of muscle mass during those positive adaptations induced by RE were due to a re- short periods of disuse or unloading, such as during recovery versal effect on the specific molecular pathways controlling from injury or illness, is characterized by a rapid increase muscle integrity during unloading, or to an activation of dif- in protein breakdown,7,8 along with reduced muscle protein ferent pathways, or a combination, remains to be elucidated. synthesis.9 In contrast, knowledge about the physiological It follows that the potential existence of groups of genes that and molecular basis for muscle loss during prolonged disuse are not reversed by the countermeasure employed might be periods warrants further research, since this has mainly been of particular interest, since these could provide mechanistic characterized in cross-sectional studies. insight into disuse-related changes that are resilient to current Hitherto attempts to provide mechanistic information on interventions employed to combat muscle loss. the molecular programs governing unloading-induced mus- The main purpose of the current study was to analyze the cle deconditioning have included the exploration of tran- genome-wide muscle transcriptome response to long-term scriptome changes in muscle after short-term bed rest or bed rest with or without RE. We hypothesized that pathways unilateral lower limb suspension/cast immobilization, rang- related to metabolism, cell structure, and protein synthesis ing from 24 hours to 21 days,10-17 or longer periods of bed would be downregulated, while molecular routes direct- rest (60 days).18,19 A common disuse signature, regardless of ing protein breakdown would be upregulated by 84 days of the differences in methods and/or study duration, is a marked bed rest. In addition, while we hypothesized that RE would downregulation of gene networks related to both energy me- restore the majority of the transcriptional signature of bed tabolism and structural components, together with an upreg- rest, we specifically sought to explore the potential existence ulation of the ubiquitin-proteasome pathway.10-12,14,15,17-19 of genes that would be resilient to the countermeasure em- In addition, we recently identified the MEF2 family of tran- ployed. Finally, we used an exon usage analysis to determine scription factors, in particular MEF2C, as potential master splice events induced by long-term bed rest. regulators of skeletal muscle alterations to short-term unloading (21-day bed rest),15 while microRNAs seemed to play a negligible role.16 In contrast, the molecular basis of 2 | MATERIALS AND METHODS long-term disuse atrophy have remained scantly research. Thus, although unloading models, in particular bed rest, in- 2.1 | General design duce robust muscle atrophy solely due to lack of mechanical stimuli, it is plausible that the transcriptomic information Twenty-one healthy men (age range 26-41 years) were ran- gathered from long-term bed rest could further our under- domly assigned to performed 90-day bed rest with (bed rest standing of more complex and multifactorial muscle wasting and exercise, BRE; n = 9) or without (BR; n = 12) concur- conditions.20 Indeed, additional mechanistic information on rent iso-inertial RE every third day targeting the quadriceps the driving forces behind disuse-induced muscle loss would muscle group (ie, supine squat; four sets of seven maximal allow for better characterization of nutritional/pharmacolog- concentric-eccentric repetitions) and the calf muscles (ie, ical and/or exercise countermeasures to prevent, or at least calf press; 4 sets of 14 concentric-eccentric repetitions) attenuate, muscle loss during long-term lack of mechanical employing flywheel technology. Muscle biopsies from m. stimuli such as during extended disease episodes. vastus lateralis were obtained from all subjects before and One of the most efficient strategies to preserve muscle in- after 84 days of bed rest. The muscle specimens were used tegrity during unloaded conditions is high-intensity resistance to conduct a DNA microarray analysis. We and others have exercise (RE). The use of an iso-inertial flywheel exercise previously reported on functional, structural, and metabolic FERNANDEZ-GONZALO et al. | 3 muscle adaptations during the same bed rest project,1,3,6,21-23 force, power, work, joint angle, and angular velocity were as well as tendon24 and bone properties,25,26 and vascular recorded.1 alterations.27

2.5 | Skeletal muscle biopsies 2.2 | Subjects Before (PRE) and after (POST) 84-day bed rest, yet prior to Specialized personnel from the Institute for Space Medicine re-ambulation, muscle biopsies were obtained from m. vas- and Physiology clinic (MEDES) interviewed potential can- tus lateralis of the dominant leg from all subjects. Despite didates. Finally, 21 healthy men were recruited, and after the bed rest period lasting 90 days, biopsies were obtained at a thorough medical examination, nine men (33 ± 5 years, day 84 to avoid any interference with the demanding testing 176 ± 5 cm, and 71 ± 6 kg) were randomly assigned to 90-day schedule following the intervention. All pre- and post-bed bed rest with concurrent flywheel resistance exercise (BRE), rest biopsies were taken in the morning within a 2-3-hour and 12 men (32 ± 4 years, 173 ± 3 cm, and 72 ± 5 kg) to window. Briefly, after local anesthesia had been applied, 90-day bed rest only (BR). One of the subjects in BRE had a tissue samples were collected using 5- or 6-mm Bergström previous unreported knee injury that made him failed to com- needles. Samples were cleansed from fat, connective tissue, plete some training sessions and post-bed rest physical tests. and excess blood, before being frozen in liquid nitrogen and Three subjects from BR received pamidronate supplementa- stored at −80°C. tion. The purposes, premises, and risks associated with the experiments were clearly described before subjects signed a written consent form to participate. The study was designed 2.6 | RNA extraction and and performed according to the Declaration of Helsinki. All microarray analysis the experiments included in the study were approved by the local Ethics Committee. Using a bead beater device (BioSpec Products, Inc, Bartlesville, OK) and TRIzol (Invitrogen Life Technologies, Carlsbad, CA), total RNA was extracted from ~20-mg frozen 2.3 | Head-down tilt bed rest muscle samples. Then, the RNA was purified using PureYield RNA Midiprep System (Promega Corporation, Madison, Subjects from both BR and BRE were subjected to 6º head- WI) and RNA quality was checked using an Agilent 2100 down tilt position at all times (ie, rest, exercise training, toi- bioanalyzer (Agilent Technologies, Amsterdam, NL). HTA let procedures, shower, transportation, etc). The participants 2.0 gene-chips were processed according to the manufac- were only allowed to rest on their elbows during meals, and turer's protocol. Fragmented (5 μg) cDNA was hybridized to to move in the horizontal plane. Video surveillance and pres- each array and scanned using a Gene Chip Scanner 30007G. sure-sensitive mattresses were used to assess compliance. On Probesets were scaled, normalized and summarized on a daily basis, physiotherapists performed massages and ankle Ensbl-gene- and exon-level, respectively using the Aroma- circumduction movements in all participants. Diet and time pipeline29 and chip definition files from Brainarray V19.30 of the day for each meal were controlled during the bed rest Quality control was performed using normalized unscaled period. The caloric intake was based on subjects' body weight, standard error (NUSE) and principal components analysis and it corresponded to 34.8 kcal kg−1 day−1 (50%-55% carbo- (PCA). The complete data set is publicly available at Gene hydrates, 30%-35% lipids, and 12%-15% proteins).28 Expression Omnibus with accession ID (GSE148152). Unsupervised cluster-analysis was carried out using t-distributed stochastic neighbor embedding (t-SNE)31,32 2.4 | Flywheel RE utilizing Rt-SNE.33 Gene-level differential expression was assessed using Limma, analyzing PRE vs POST block wise The RE performed by BRE consisted of 6° head-down tilt su- per subject as well as differences between BR and BRE pine squat and calf press exercises using iso-inertial flywheel over time: ENSG~(BR_POST – BR_PRE) − (BRE_POST technology (Yo-Yo Technology Inc, Stockholm, Sweden1) – BRE_PRE). False discovery rates (FDR) were calculated every third day (2-3 days per week; ~3 minutes of actual based on Bayesian moderated t-statistics where an FDR of muscle activation per week). The exercise training began on 1% was considered significant. day 5 of bed rest. Four sets of 7 or 14 maximal, coupled con- Gene-ontology analysis and pathway enrichment analysis centric-eccentric repetitions were performed in each session was performed using the Database for Annotation, Visualization for the supine squat and calf press, respectively. Two min- and Integrated Discovery (DAVID) (biological function and cel- utes of rest were allowed between sets. During all repetitions lular compartment) for overrepresentation among differentially 4 | FERNANDEZ-GONZALO et al. expressed genes, with all genes tested for differential expression among the downregulated genes, and eight biological func- as background. Pathway analysis was performed using gene- tions and seven cellular compartments among the upregu- set enrichment analysis where pathway annotation of genes was lated genes (Figure 1A, Table S2). retrieved from wiki-pathways (current as of December 2018).34 Of the downregulated ontologies, the most notable were For both over-representation and gene-set enrichment, an FDR related to mitochondrial compartment and function. In fact, of 5% was considered significant. 53 of the 335 downregulated genes were associated with mi- The continuous relationship between BR and BRE was in- tochondrial function (Figure 1A, Table S1). Mitochondrion vestigated using FC-FC plot using the logFC of BR (x-axis) was overwhelmingly dominating in cellular compartment, and logFC of BRE (y-axis) for each gene, thus creating a polar that is, 109 of the 335 genes were annotated as localized coordinate system illustrating similarities/dissimilarities of within the mitochondrion rendering a >20-fold enrich- gene-expression between the two conditions. The resulting cir- ment for mitochondrial related genes. Consistent with the cle is divided into eight sectors each covering 45° of the cir- substantial loss of muscle mass, ontologies related to muscle. Each sector will contain genes with a similar regulation in cle structural components were highly enriched among the both conditions, significant effect after BR but not BRE, or vice downregulated genes (Figure 1A, Table S2). While most of versa. Pathway-level comparison of BR vs BRE differential the myosin-related genes were downregulated, for example, expression was calculated using two-sided Wilcoxon tests on MYH7 and MYL2, the fast-twitch myosin MYH1 and peri- empirical cumulative distributions where an FDR of 5% was natal MYH8 (P = .0107) were both upregulated (Table S1). considered significantly different between BR and BRE. Of the upregulated ontologies, the most highly enriched were Transcript isoform expression was analyzed by re-anno- related to genes involved in mRNA-processing and regulation tation of the probe-data from the microarray to the transcript of transcription, followed by members of the ubiquitin system level Ensbl-transcript probeset version 19 from Brainarray. (Figure 1A, Table S2). The expression of several genes pre- Differential expression after bed rest was analyzed in the same viously reported to be involved in the regulation of circadian manner as gene-level data. To identify isoforms within the same rhythm was also greater after bed rest (9.7-fold enrichment) gene with different response to bed rest, the transcript-isoforms (Figure 1A, Table S2), including the well-known Bmal1 tran- for each differentially expressed gene were analyzed for dif- scription factor regulators RORA and NR1D2. While RORA ferences in response to the bed rest intervention by compar- acts as an activator of Bmal1 transcription, NR1D2, a mem- ing the pairwise transcripts using Wilcoxon-test. Alternative ber of the nuclear hormone receptor subgroup Rev-erb, inhib- exon usage was analyzed using the diffSplice function within its Bmal1 expression.35,36 Limma, where t-statistics (P < .05) is used to compare each A substantial number (ie, 85) of transcription factors exon to the average of all other exons for the same gene. and co-factors were differentially expressed after the bed rest period. There were significantly (P < .001) more transcription factors among the upregulated (n = 64) than the 3 | RESULTS downregulated genes (n = 21) (Figure 1B, Table S3). A majority of these transcription factors changed in a pattern Changes in muscle size and function from the 90-day bed coherent with the observed gene expression profile, re- rest project have been described previously. In brief, the un- flecting their documented regulatory functions. Several of loading intervention induced an 18% decreased of quadriceps these factors have previously been linked to metabolic al- muscle volume, which was counteracted by the RE protocol teration and atrophy processes in skeletal muscle following employed.1 Similarly, decrements in muscle function of knee unloading, such as the downregulation of PGC1-alpha and extensor muscles after the bed rest period were partially off- PPARA, and the upregulation of Foxo3 and NF-kappa-beta. set by the flywheel RE training program,1 as well as some Yet, several transcription factors not previously associated of the metabolic alterations induced by long-term bed rest.21 with lack of mechanical stimuli were also identified. For example, TEF-1 (TEAD1), a transcription factor implicated in the fast-to-slow fiber-type transition,37 and PROX1, 3.1 | Bed rest which is specifically expressed in slow muscle fibers,38 were downregulated in BR. This was in accordance with Following 84 days of bed rest, 35 345 gene-level probe sets the upregulation of genes related to fast-twitch myosin (see were analyzed for differential expression, of these 17 207 above). Other downregulated transcription factors with a were un-annotated. Gene-level differential expression iden- known impact on molecular regulation of skeletal muscle tified 335 annotated probe sets that were downregulated, were the glucose-sensing MLXIPL, the circadian rhythm and 315 that were upregulated after BR at an FDR of 0.01 factor NPAS2, and ESRRA, an important component of the (Figure 1A, Table S1). Fifteen biological functions and eight regeneration-to-injury process. Upregulated transcription cellular compartments were enriched at an FDR of <1% factors following BR, with a role in muscle remodeling and FERNANDEZ-GONZALO et al. | 5 (A)

TP63 MYH7B ARHGAP28 Cellular compartment BCKDK Cellular compartment SLC41A1 6 Description FC FDR DescriptionFCFDR SPATA25 ANKRD12 mitochondrial inner membrane 8.7 3.E-41 nucleus1.8 3.E-15 mitochondrion4.4 2.E-38 PPIF MYH1 nucleoplasm2.0 1.E-09 FOXO3 mitochondrial matrix 5.9 1.E-14 NMRK2 MYL3 PLCE1 chromatin 6.01.E-02 NINJ2 CIR1 PDE11A HSPB6 Biological process MYH7 NEDD4LUSP6ADH1C Biological process 4 DHCR24 RRAD DescriptionFCFDR Description FC FDR TNNT1 ABCA5 ADH1B mRNA processing 5.31.E-03 mitochondrial electron transport 17.4 7.E-10 CASQ2 MSTN transcription, DNA-templated1.7 6.E-02 tricarboxylic acid cycle 16.2 2.E-05 −logFDR ART3 protein ubiquitination 2.93.E-02 muscle filament sliding12.4 1.E-04 FABP3 regulation of circadian rhythm 8.33.E-02 MYOZ2 LMOD1 2

0 −2 −1 012 logFC

(B) Transcription factors

TP63

PPARGC1B FOXO3 MLXIPL HMBOX1 PPARA CIR1

SMARCC1 4 NFYC PREB NFKBIA NPAS2 ELP2MDM2 ARHGAP35 EPC2 NR1D2 −logFD R THAP4 DEK SFPQ PROX1

−1 01 logFC

FIGURE 1 V-plots of long-term bed rest. A, V-plot showing overall gene expression alterations by 84 days of bed rest (FDR < 0.01), with the significantly upregulated (red) and downregulated (blue) genes, together with representative cellular compartments and biological processes. B, V-plot highlighting differentially expressed genes after 84 days of bed rest belonging to transcription factors molecular regulation, were EPC1, KLF10, RBPJ, NFE2L2 was the only Gene ontology-term (GO-term) significantly (alt. Nfr2), and SP3 (skeletal muscle growth regulation); enriched among the genes with differentially expressed tran- MDM2 and TRIM32 (E3 ubiquitin-protein system); and script isoforms. RBFOX2 (exon splicing) (Figure 1B, Table S3). Using the diffSplice method in linear models for microarray data (LIMMA), transcript isoform and alternative exon 3.2 | Effect of exercise countermeasure usage was analyzed. We discovered 553 genes that had ≥1 exon with alternative exon usage at an FDR < 1%. Of these, To evaluate the gene expression profile following prolonged 219 could be attributed to genes that had ≥2 transcript iso- bed rest with and without RE countermeasure, data were ex- forms differentially expressed at an FDR < 1%, and 83 genes plored in an unsupervised data-driven fashion using t-SNE. had transcripts with isoforms where the response to bed rest This analysis revealed distinct clustering into three groups was significantly different among the transcripts (Table S4, (PRE, BR, and BRE), where the Euclidian distance between Figure S1). These transcripts belonged to genes primarily the PRE and BRE observations was significantly smaller associated with muscle structural components and contrac- and with a certain degree of overlap, compared with the tile machinery; in fact, “muscle filament sliding” (FDR 5%) distance between BR and BRE (Figure 2A). This was also 6 | FERNANDEZ-GONZALO et al.

(A)

(B)

Time Not Normalized by Exercise Normalized by Exercise BR BR + E Overcompensated by Exercise

Downregulated Upregulated by (C) Normalized by Exercise: by Bedrest Bedrest

20% 74% 6% 45% 54% 1% Yes ***

Yes plus overcompensated

n = 335 n = 315

FIGURE 2 Effects of high-intensity, low-volume resistance exercise to counteract the transcriptomic alterations induced by 84 days of bed rest. A, tSNE-plot with hierarchical clustering of global gene expression before (PRE) and after 84 days of bed rest with (BRE) and without (BR) resistance exercise countermeasure. B, Representation of the continuous relationship between BR and BRE using FC-FC plot using the logFC of BR (x-axis) and logFC of BRE (y-axis) for each gene. This polar coordinate system illustrates the similarities/dissimilarities of gene-expression between the two conditions. The resulting circle is divided into 8 sectors each covering 45° of the circle. Genes that were not normalized by exercise are shown in purple, those where exercise normalized the expression (related to PRE) are shown in soft green, and those where exercise actually overcompensated the expression of the particular transcript (related to PRE) are displayed in dark green. C, Pie charts showing the power of the exercise countermeasure employed to normalize transcriptome alterations induced by 84 days of bed rest with specific focus on both downregulated and upregulated factors. Note the greater proportion of genes not normalized by exercise among the upregulated genes (45%) when compared with the downregulated ones (20%) (***; P < .001) FERNANDEZ-GONZALO et al. | 7 evident from the differential expression analysis, where 52 data sets from young men in untrained-basal vs trained-basal of the downregulated and 22 of the upregulated genes had an conditions.40,41 The overlap between the BRE residual signa- interaction between BR and BRE over time (P < .05). The ture and the genes differentially expressed by RE (ie, Pre vs propensity of the genes that were reversed by the exercise Post differences P < .05) had little similarities. Thus, only 6 countermeasure were further analyzed based on co-variance and 28 genes out of the 209 genes included in the residual between BR and BRE.39 This approach allowed us to visual- signature were differentially regulated by RE in Raue et al41 ize the effects of the two interventions on the gene-level and and Damas et al,40 respectively. Overall, these results reflect to stratify genes based on their propensity to become reversed specific disuse transcriptomic alterations that are independ- (or amplified) in the BRE condition. The majority (268 of ent from RE-increased levels of muscle activation and/or 335) of the genes downregulated by BR were reversed by tension. exercise, and 20 genes were even over-compensated, that is, The difference regarding the effect of the exercise negative FC in BR and a positive FC in BRE (Figure 2B,C). countermeasure between the downregulated and upregulated Only 20% of the genes (67 of 335) that were downregulated genes persisted when we summarized the data on the level in BR remained downregulated to a similar extent follow- of pathways and biological functions (Figure 3). Thus, for ing BRE, which was in contrast to the 45% (142 out of 315) all downregulated pathways after BR, the effect was reduced of genes that were upregulated following both BR and BRE (ie, more similar to the pre-bed rest condition) in the BRE (Figure 2B,C; Table S5). To test whether the group of genes group, which was not the case for the upregulated pathways not reversed in the BRE group was due to a lack of exer- (see below). The highest degree of reversibility by exercise cise effects on these genes, or rather induced by both stimuli was found in those genes associated to “Electron Transport (unloading and exercise), the genes included in the residual Chain” (Wilcoxon-test 7e-21). This pathway was even signature were compared with RE training transcriptome slightly elevated compared with the pre-level in the BRE

Pathway FC FDR WX-P-value

Translation Factors 1.4 0.03 n.s BR

BRE Proteasome Degradation 1.4 0.02 n.s

Nucleotide Metabolism 1.5 0.03 n.s

Cell Cycle 1.3 0.02 n.s

TP53 Network 1.6 0.02 2.E-02

PPAR Alpha Pathway 0.7 0.02 7.E-03

TGF−beta Signaling Pathway 1.4 0.02 9.E-05

TCA Cycle 0.6 0.02 8.E-05

Circadian rythm related genes 1.3 0.02 3.E-05

Fatty Acid Beta Oxidation 0.7 0.02 1.E-06

Oxidative phosphorylation 0.6 0.02 3.E-18

Electron Transport Chain 0.6 0.02 7.E-21

−0.4 0.0 0.4 logFC

FIGURE 3 Log-change (related to PRE) of molecular pathways after 84 days of bed rest with (BRE) or without (BR) concurrent resistance exercise training 8 | FERNANDEZ-GONZALO et al.

NEDD4

UBE2D1

COX15 UCHL3 PSMD1

ATP5F1C NDUFB8 PSMD7 UBE2B PSMD11 NDUFB6 ATP5MC2 PSMD5 ATP5MC3 PSMD10 SDHC PSMD3 COX5A ATP5PB PSMC3 SURF1 PSMD4 ATP5MC1 NDUFV3 COX6B1 HLA−E ATP5IF1

Time Time

BR + E BR + E

EIF2AK3

RXRG CD36 EIF5B EIF5 EIF3E EIF4A2 EIF2AK2 EIF4H EIF2S3 EIF4B FABP1 EIF2B1 CYP8B1 EIF3B PPARA UCP1 FABP3 FABP6 CYP27A1

AQP7

Time Time

BR + E BR + E

FIGURE 4 Representation of the continuous relationship between BR and BRE using FC-FC plot using the logFC of BR (x-axis) and logFC of BRE (y-axis) for genes related to (a) electron transport and oxidative phosphorylation, (b) proteasome, (c) PPAR-signaling, and (d) translation factors. Purple and green arrows indicate the direction of the transcript expression alterations in a specific sector induced by bed rest and bed rest plus exercise, respectively group (FC = 1.05; Figures 3 and 4). Other downregulated 4 | DISCUSSION pathways with high degree of normalization in BRE were “Fatty acid beta-oxidation,” “TCA cycle,” and “PPAR Alpha This study provides unique information about the transcrip- pathway” (Figures 3 and 4). Among the upregulated path- tomic changes occurring in skeletal muscle during one of the ways that were normalized in BRE we found “TGF − beta longest bed rest campaigns ever performed. In comparison to Signaling Pathway,”42 “Circadian rhythm related genes,” shorter bed rest studies, the results indicate that the transcrip- and “TP53 Network.” However, despite a general counter- tomic machinery continues to adapt throughout extended dis- acting effect of RE at the transcriptomic level, a number of use periods, and that high-intensity, low-volume RE, capable bed rest-induced upregulated pathways were not normal- of offsetting unloading-induced muscle atrophy, neutralizes ized by the exercise countermeasure; “Translation Factors,” most but not all of the molecular changes induced by 84 days “Proteasome Degradation,” “Cell Cycle,” and “Nucleotide of bed rest. Thus, we reveal a skeletal muscle residual tran- Metabolism” (Figures 3 and 4). scriptomic signature of unloading that is resilient to RE. FERNANDEZ-GONZALO et al. | 9 The current data show that the biological perturbations to be essential for proper insulin handling, lipic homeostasis, in skeletal muscle after 84 days of bed rest are at least as and myokine secretion.56 large, if not greater, than what has been described after 2115 Past studies have shown that increased levels of physical or 60 days of bed rest.18,19 For example, pathways related to activity can induce differential isoform-specific expression mitochondria, energy metabolism, and structural components in skeletal muscle.57-58 Our exon usage analysis is one of were downregulated, and those involved in the ubiquitin sys- the first reports showing that genes associated with skele- tem upregulated, which is consistent with findings in skeletal muscle structural components and contractile machinery tal muscle after short-10-15,17 and moderately long periods of presented isoforms that responded to long-term disuse in an unloading.18,19 Similarly, we found a transcriptomic profile individual fashion. In the vast majority of the cases, however, indicating slow-to-fast fiber transition, a common event in the isoforms followed an expression pattern that was similar unloading studies.6,43-46 Thus, MYH1 and MYH8, myosin to that of the common gene, ie, if the gene was downregu- isoforms much more abundant in fast-twitch fibers,47 were lated by bed rest, the different isoforms were downregulated highly upregulated after bed rest. Among the factors regulat- as well, albeit the magnitude of change differed among iso- ing fiber type composition, PPAR48 and MEF2,49 were down- forms. With the current knowledge, it is difficult to conclude regulated with bed rest, which is in line with shorter bed rest if and how these splice changes are related to the observed campaigns.15 Our data add Prox1 as a factor involved in the skeletal muscle phenotype alterations. However, the finding unloading-induced fiber shifting. Prox1, which was downreg- that isoforms responded to long-term unloading in an in- ulated after 84-day of bed rest, is specifically expressed in dividual fashion may be relevant in future attempts to find slow muscle fibers.38 This factor is an important component putative targets and biomarkers for predicting the response of the molecular program protecting the slow fiber type and of a particular countermeasure to muscle deconditioning, as activates the NFAT pathway (a key regulator of slow muscle suggested for other diseases.59 fiber phenotype).38 RE-countermeasures are effective in preserving most of Despite the similarities with shorter unloading studies, the muscle mass and function with prolonged bed rest, but 84 days of bed rest also resulted in unique transcriptomic whether this is associated with a complete or partial nor- alterations not noted after shorter campaigns. Thus, in con- malization of the transcriptomic signature, or due to an acti- trast to the somewhat generalized idea that the majority of vation of a different gene expression profile, was unknown. molecular alterations occur at the beginning of the unload- Our unsupervised analysis (ie, t-SNE) revealed that the over- ing period, our data suggest that the muscle transcriptomic all transcriptomic signature of prolonged bed rest was, to a responses to unloading are ongoing and continue to adapt relatively high degree, reversed by flywheel RE. Thus, the (and change) throughout extended periods of disuse. An in- 3-min-per-week of contractile activity (4 sets × 7 reps em- triguing observation was that 84 days of bed rest induced al- ploying flywheel RE) offset the transcriptomic alterations re- terations in several genes and transcription factors involved lated to aerobic energy metabolism (eg, “Electron Transport in circadian rhythm regulation. Muscle activity and nutrient Chain,” “Fatty Acid Beta-Oxidation,” “TCA Cycle,” “PPAR- intake are known cues for the circadian rhythm regulation in signaling pathway”), which is further supported by the met- skeletal muscle.50,51 Given that diet and time of meals were abolic adaptations at the enzymatic level reported from controlled for in the current study, and the observed differ- the same bed rest intervention.21 Similarly, other pathways ences between bed rest with vs without exercise (see below), important for skeletal muscle integrity were normalized by our results support the idea of a disuse-sensitive circadian the exercise intervention, including “TP53 Network” and molecular clock in the skeletal muscle.52-54 These findings “TGF-beta Signaling Pathway.” While TGF-beta superfam- add information from in vivo human muscle to past reports ily is associated with an inhibition of muscle regeneration describing atypical expression of core clock genes/transcrip- following injury,60 P53 can exacerbate the stress load upon tion factors after muscle denervation in mice.55 The Bmal1- the muscle in situations of inactivity-induced atrophy.61 Our CLOCK complex is the central modulator of the circadian data support the connection of these two pathways with a rhythm system. In the current study, the antagonistic factors molecular environment favoring blunted muscle protein syn- RORA and NR1D2 were both upregulated after long-term thesis and an increase in atrophy-related factors in unloaded, bed rest, which may be related to a myocellular effort to sta- non-injured human skeletal muscle. Interestingly, the bed bilize the regulatory feedback loop controlling Bmal1 expres- rest-induced upregulation of “Circadian Rhythm Related sion, and thus, decrease the impact of the lack of mechanical Genes” was counteracted by the RE countermeasure em- stimuli on circadian rhythms regulation. The implications of ployed. This result supports the existence of muscle tension/ the unloading-induced circadian rhythm alterations reported activation-specific circadian rhythm regulatory mechanisms here should not be overlooked, since the cell-autonomous (discussed above). skeletal muscle clocks have been shown to impact important Despite the counteracting effects of RE, there was a group pathways involved in human skeletal muscle remodeling, and of genes that was resilient to RE (ie, they were upregulated 10 | FERNANDEZ-GONZALO et al. or downregulated to a similar extent in both conditions). ACKNOWLEDGMENTS Other studies analyzing different muscle groups and/or The experimental part of this study (bed rest intervention) women during shorter bed rest periods have also suggested was supported by grants from the Swedish National Space the existence of genes resilient to diverse exercise counter- Agency (SNSA), the European Space Agency (ESA), and measures.18,19 Interestingly, the current analyses showed the Swedish National Centee for Research in Sports (CIF) to that those genes that were upregulated by bed rest in the PAT. The authors would like to thank BEA—the core facility current study were less likely to be reversed by the exer- for Bioinformatics and Expression Analysis service facility cise countermeasure than the downregulated transcripts. In at Novum, Karolinska Institutet in Huddinge, for their help- addition, this group of genes had very little overlap with ful contribution. transcriptomic changes after RE only,40,41 suggesting that the residual signature was of an “unloading-specific” na- CONFLICT OF INTEREST ture and independent of the particular exercise intervention Rodrigo Fernandez-Gonzalo, Per A. Tesch, Tommy R. conducted. It could be that the different muscle activity/ Lundberg, Björn A. Alkner, Eric Rullman, and Thomas tension pattern of the RE protocol used (ie, high-intensity, Gustafsson declare they have no competing interests. low-volume), when compared with that required for the maintenance of an ambulatory position (ie, low-intensity, AUTHOR CONTRIBUTIONS high-volume), could explain the inefficiency of the counter- P. A. Tesch and B. A. Alkner designed and conducted the measure employed to normalize the residual signature bed rest with/without exercise countermeasure intervention genes. In addition, other bed rest-induced changes, such and collected the muscle biopsies. R. Fernandez-Gonzalo, as microenvironment and/or hemodynamics alterations T. Gustafsson, and E. Rullman designed the gene array ex- could have contributed to the differential expression of the periments. E. Rullman performed the statistical analysis. E. genes identified in the residual signature. When analyzed Rullman, T. R. Lundberg, T. Gustafsson, and R. Fernandez- at a pathway level, several molecular pathways were upreg- Gonzalo interpreted the results. R. Fernandez-Gonzalo, T. ulated to a similar extent after bed rest with and without R. Lundberg, and T. Gustafsson drafted the manuscript. B. RE, constituting the pathway-level residual signature of A. Alkner, P. A. Tesch, and E. Rullman read and provided bed rest; “Translation Factors,” “Proteasome Degradation,” feedback on the drafted manuscript. All authors read and ap- “Cell Cycle,” and “Nucleotide Metabolism.” Altogether, proved the submitted version. these results suggest that, despite a rather successful reversal of the phenotypic alterations induced by long-term bed REFERENCES 1 rest (ie, muscle mass), the particular RE employed could 1. Alkner BA, Tesch PA. Knee extensor and plantar flexor muscle not fully counteract all of the transcriptomic changes. The size and function following 90 days of bed rest with or without clinical and/or functional impact of the residual signature resistance exercise. Eur J Appl Physiol. 2004;93:294-305. needs to be examined in future studies. 2. Berg HE, Larsson L, Tesch PA. Lower limb skeletal muscle function after 6 wk of bed rest. J Appl Physiol. 1997;82:182-188. In conclusion, this study provides the most comprehen- 3. Haus JM, Carrithers JA, Carroll CC, Tesch PA, Trappe TA. sive human skeletal muscle transcriptome profile to unload- Contractile and connective tissue protein content of human skeletal ing to date, as well as delivers novel information about the muscle: effects of 35 and 90 days of simulated microgravity and ex- reversal of transcriptomic changes through RE. We report ercise countermeasures. Am J Physiol Regul Integr Comp Physiol. that (a) besides a predictable differential expression of genes 2007;293:R1722-R1727. and pathways related to mitochondria, ubiquitin system, and 4. Trappe S. Effects of spaceflight, simulated spaceflight and counter- J Gravit Physiol skeletal muscle metabolism and structure, 84 days of bed measures on single muscle fiber physiology. . rest altered other less-expected transcripts involved in, for 2002;9:P323-P326. 5. Trappe S, Creer A, Slivka D, Minchev K, Trappe T. Single mus- example, circadian rhythms. This suggests that muscle tran- cle fiber function with concurrent exercise or nutrition counter- scriptomic modifications are ongoing throughout extended measures during 60 days of bed rest in women. J Appl Physiol. periods of disuse. (b) As little as 3 minutes per week of mus- 2007;103:1242-1250. cle activation through RE is enough to offset the majority 6. Gallagher P, Trappe S, Harber M, et al. Effects of 84-days of be- of the bed rest-induced transcriptomic alterations. However, drest and resistance training on single muscle fibre myosin heavy (c) there were transcripts that were resilient to this counter- chain distribution in human vastus lateralis and soleus muscles. Acta Physiol Scand measure, suggesting the presence of a residual signature of . 2005;185:61-69. 7. Reid MB, Judge AR, Bodine SC. CrossTalk opposing view: the skeletal muscle disuse. Collectively, the current study paves dominant mechanism causing disuse muscle atrophy is proteolysis. the way for further exploration of skeletal muscle alterations J. Physiol. 2014;592:5345-5347. to extreme disuse, and the potential, and limitations, of both 8. Tesch PA, von Walden F, Gustafsson T, Linnehan RM, Trappe TA. current and new countermeasures employed to fight muscle Skeletal muscle proteolysis in response to short-term unloading in deconditioning. humans. J Appl Physiol. 2008;105:902-906. FERNANDEZ-GONZALO et al. | 11 9. Phillips SM, Mcglory C. CrossTalk proposal: the dominant mecha- 25. Watanabe Y, Ohshima H, Mizuno K, et al. Intravenous pamidro- nism causing disuse muscle atrophy is decreased protein synthesis. nate prevents femoral bone loss and renal stone formation during J Physiol. 2014;592:5341-5343. 90-day bed rest. J Bone Min. Res. 2004;19:1771-1778. 10. Alibegovic AC, Sonne MP, Hojbjerre L, et al. Insulin resistance 26. Rittweger J, Frost HM, Schiessl H, et al. Muscle atrophy and bone induced by physical inactivity is associated with multiple tran- loss after 90 days' bed rest and the effects of flywheel resistive scriptional changes in skeletal muscle in young men. Am J Physiol exercise and pamidronate: results from the LTBR study. Bone. Endocrinol Metab. 2010;299:E752-E763. 2005;36:1019-1029. 11. Chen YW, Gregory CM, Scarborough MT, Shi R, Walter GA, 27. Capri A, Kerbeci P, Pascaud L, Arbeille P. Assessment of the calf Vandenborne K. Transcriptional pathways associated with vein cross section change during stand-test after a 90 day bed rest skeletal muscle disuse atrophy in humans. Physiol Genomics. by echography. J Gravit Physiol. 2004;11:P83-P84. 2007;31:510-520. 28. ESA-CNES-NASDA. Final report. Simulation of a mission aboard 12. Jones SW, Hill RJ, Krasney PA, O’Conner B, Peirce N, Greenhaff the international space station by a long-term (90 days) anti-ortho- PL. Disuse atrophy and exercise rehabilitation in humans pro- static bed rest at -6° on healthy subjects; European Space Agency, foundly affects the expression of genes associated with the regula- Noordwjik, The Netherlands, 2003. tion of skeletal muscle mass. FASEB J. 2004;18:1025-1027. 29. Bengtsson H, Simpson K, Bullard J, Hansen K. aroma.affy- 13. Mahmassani ZS, Reidy PT, McKenzie AI, Stubben C, Howard MT, metrix: A Generic Framework in R for Analyzing Small to Very Drummond MJ. Age-dependent skeletal muscle transcriptome re- Large Affymetrix Data Sets in Bounded Memory. Berkeley, CA: sponse to bed rest-induced atrophy. J Appl Physiol. 2019;126:894-902. Department of Statistics, University of California, Berkely; 2008. 14. Ogawa T, Furochi H, Mameoka M, et al. Ubiquitin ligase gene ex- Tech. Report 745. https://statistics.berkeley.edu/sites/default/files/ pression in healthy volunteers with 20-day bedrest. Muscle Nerve. tech-reports/745.pdf. Accessed February 18, 2020. 2006;34:463-469. 30. Dai M, Wang P, Boyd AD, et al. Evolving gene/transcript defi- 15. Rullman E, Fernandez-Gonzalo R, Mekjavic IB, Gustafsson T, nitions significantly alter the interpretation of GeneChip data. Eiken O. MEF2 as upstream regulator of the transcriptome sig- Nucleic Acids Res. 2005;33:e175. nature in human skeletal muscle during unloading. Am J Physiol 31. Van Der Maaten L, Hinton G. Visualizing data using t-SNE. J Regul Integr Comp Physiol. 2018;315:799-809. Mach Learn Res. 2008;9:2579-2605. 16. Rullman E, Mekjavic IB, Fischer H, Eiken O. PlanHab (Planetary 32. Van Der Maaten L. Accelerating t-sne using tree-based algorithms. Habitat Simulation): the combined and separate effects of 21 J Mach Learn Res. 2014;15:3221-3245. days bed rest and hypoxic confinement on human skeletal muscle 33. Krijthe JH. T-Distributed Stochastic Neighbor Embedding Using miRNA expression. Physiol Rep. 2016;4:e12753. Barnes-Hut Implementation; 2015. https://github.com/jkrijthe/ 17. Urso ML, Scrimgeour AG, Chen YW, Thompson PD, Clarkson Rtsne. Accessed January 20, 2020. PM. Analysis of human skeletal muscle after 48 h immobilization 34. Slenter DN, Kutmon M, Hanspers K, et al. WikiPathways: a mul- reveals alterations in mRNA and protein for extracellular matrix tifaceted pathway database bridging metabolomics to other omics components. J Appl Physiol. 2006;101:1136-1148. research. Nucleic Acids Res. 2018;46:D661-D667. 18. Chopard A, Lecunff M, Danger R, et al. Large-scale mRNA anal- 35. Ramakrishnan SN, Muscat GEO. The orphan Rev-erb nuclear re- ysis of female skeletal muscles during 60 days of bed rest with and ceptors: a link between metabolism, circadian rhythm and inflam- without exercise or dietary protein supplementation as counter- mation? Nucl Recept Signal. 2006;4:e009. measures. Physiol Genomics. 2009;38:291-302. 36. Sato TK, Panda S, Miraglia LJ, et al. A functional genomics strat- 19. Salanova M, Gambara G, Moriggi M, et al. Vibration mechanosig- egy reveals rora as a component of the mammalian circadian clock. nals superimposed to resistive exercise result in baseline skeletal Neuron. 2004;43:527-537. muscle transcriptome profiles following chronic disuse in bed rest. 37. Tsika RW, Schramm C, Simmer G, Fitzsimons DP, Moss RL, Ji J. Sci Rep. 2015;5:17027. Overexpression of TEAD-1 in transgenic mouse striated muscles 20. Vernikos J, Schneider VS. Space, gravity and the physiology produces a slower skeletal muscle contractile phenotype. J Biol of aging: parallel or convergent disciplines? A mini-review. Chem. 2008;283:36154-36167. Gerontology. 2010;56:157-166. 38. Kivela R, Salmela I, Nguyen YH, et al. The transcription factor 21. Irimia JM, Guerrero M, Rodriguez-Miguelez P, et al. Metabolic Prox1 is essential for satellite cell differentiation and muscle fi- adaptations in skeletal muscle after 84 days of bed rest with and bre-type regulation. Nat Commun. 2016;7:13124. without concurrent flywheel resistance exercise. J Appl Physiol. 39. Boström J, Sramkova Z, Salašová A, et al. Comparative cell cycle 2017;122:96-103. transcriptomics reveals synchronization of developmental transcrip- 22. Nielsen RO, Schjerling P, Tesch P, Stal P, Langberg H. Collagen tion factor networks in cancer cells. PLoS ONE. 2017;12:e0188772. content in the vastus lateralis and the soleus muscle following a 90- 40. Damas F, Ugrinowitsch C, Libardi CA, et al. Resistance training in day bed rest period with or without resistance exercises. Muscles young men induces muscle transcriptome-wide changes associated Ligaments Tendons J. 2015;5:305-309. with muscle structure and metabolism refining the response to ex- 23. Rittweger J, Felsenberg D, Maganaris C, Ferretti JL. Vertical jump ercise-induced stress. Eur J Appl Physiol. 2018;118:2607-2616. performance after 90 days bed rest with and without flywheel resis- 41. Raue U, Trappe TA, Estrem ST, et al. Transcriptome signa- tive exercise, including a 180 days follow-up. Eur J Appl Physiol. ture of resistance exercise adaptations: mixed muscle and fiber 2007;100:427-436. type specific profiles in young and old adults. J Appl Physiol. 24. Reeves ND, Maganaris CN, Ferretti G, Narici MV. Influence of 2012;112:1625-1636. 90-day simulated microgravity on human tendon mechanical prop- 42. Kandasamy K, Sujatha Mohan S, Raju R, et al. NetPath: a public erties and the effect of resistive countermeasures. J Appl Physiol. resource of curated signal transduction pathways. Genome Biol. 2005;98:2278-2286. 2010;11:R3. 12 | FERNANDEZ-GONZALO et al. 43. Hather BM, Adams GR, Tesch PA, Dudley GA. Skeletal muscle 55. Nakao R, Yamamoto S, Horikawa K, et al. Atypical expression responses to lower limb suspension in humans. J Appl Physiol. of circadian clock genes in denervated mouse skeletal muscle. 1992;72:1493-1498. Chronobiol Int. 2015;32:486-496. 44. Adams GR, Hather BM, Dudley GA. Effect of short-term un- 56. Perrin L, Loizides-Mangold U, Chanon S, et al. Transcriptomic weighting on human skeletal muscle strength and size. Aviat Sp analyses reveal rhythmic and CLOCK-driven pathways in human Env Med. 1994;65:1116-1121. skeletal muscle. eLife. 2018;7:e34114. 45. Cotter JA, Yu A, Haddad F, et al. Concurrent exercise on a grav- 57. Lundberg TR, Fernandez-Gonzalo R, Norrbom J, Fischer H, Tesch ity-independent device during simulated microgravity. Med Sci PA, Gustafsson T. Truncated splice variant PGC-1alpha4 is not as- Sport. Exerc. 2015;47:990-1000. sociated with exercise-induced human muscle hypertrophy. Acta 46. Trappe S, Costill D, Gallagher P, et al. Exercise in space: human Physiol. 2014;212:142-151. skeletal muscle after 6 months aboard the International Space 58. Gustafsson T, Ameln H, Fischer H, Sundberg CJ, Timmons JA, Station. J Appl Physiol. 2009;106:1159-1168. Jansson E. VEGF-A splice variants and related receptor expression 47. Stuart CA, Stone WL, Howell ME, et al. Myosin content of indi- in human skeletal muscle following submaximal exercise. J Appl vidual human muscle fibers isolated by laser capture microdissec- Physiol. 2005;98:2137-2146. tion. Am J Physiol Cell Physiol. 2016;310:C381-C389. 59. Safikhani Z, Smirnov P, Thu KL, et al. Gene isoforms as expres- 48. Lin J, Wu H, Tarr PT, et al. Transcriptional co-activator PGC-1 sion-based biomarkers predictive of drug response in vitro. Nat alpha drives the formation of slow-twitch muscle fibres. Nature. Commun. 2017;8:1126. 2002;418:797-801. 60. Kim J, Lee J. Role of transforming growth factor-β in muscle dam- 49. Wu H, Naya FJ, McKinsey TA, et al. MEF2 responds to multiple age and regeneration: focused on eccentric muscle contraction. calcium-regulated signals in the control of skeletal muscle fiber J Exerc Rehabil. 2017;13:621-626. type. EMBO J. 2000;19:1963-1973. 61. Beyfuss K, Hood DA. A systematic review of p53 regulation of 50. Wolff G, Esser KA. Scheduled exercise phase shifts the cir- oxidative stress in skeletal muscle. Redox Rep. 2018;23:100-117. cadian clock in skeletal Muscle. Med Sci Sports Exerc. 2012;44:1663-1670. 51. Shavlakadze T, Anwari T, Soffe Z, et al. Impact of fasting on SUPPORTING INFORMATION the rhythmic expression of myogenic and metabolic factors Additional supporting information may be found online in in skeletal muscle of adult mice. Am J Physiol Cell Physiol. the Supporting Information section. 2013;305:C26-35. 52. Gabriel BM, Zierath JR. Circadian rhythms and exercise - Nat Rev Endocrinol re-setting the clock in metabolic disease. . How to cite this article: Fernandez-Gonzalo R, Tesch 2019;15:197-206. PA, Lundberg TR, Alkner BA, Rullman E, Gustafsson 53. Schroder EA, Esser KA. Circadian rhythms, skeletal muscle molecular clocks, and exercise. Exerc Sport Sci Rev. T. Three months of bed rest induce a residual 2013;41:224-229. transcriptomic signature resilient to resistance exercise 54. Zambon AC, McDearmon EL, Salomonis N, et al. Time- and exer- countermeasures. The FASEB Journal. 2020;00:1–12. cise-dependent gene regulation in human skeletal muscle. Genome https://doi.org/10.1096/fj.201902976R Biol. 2003;4:R61.