enobosarm (gtx-024) modulates adult skeletal muscle mass independently of the androgen receptor in...

Enobosarm (GTx-024) Modulates Adult SkeletalMuscle Mass Independently of the AndrogenReceptor in the Satellite Cell Lineage

Vanessa Dubois, Ioannis Simitsidellis, Michaël R. Laurent, Ferran Jardi,Philippa T. K. Saunders, Dirk Vanderschueren, and Frank Claessens

Molecular Endocrinology Laboratory (V.D., M.R.L., F.C.), Department of Cellular and MolecularMedicine, Department of Gerontology and Geriatrics (M.R.L.), and Clinical and ExperimentalEndocrinology (F.J., D.V.), Department of Clinical and Experimental Medicine, KU Leuven, 3000 Leuven,Belgium; and Medical Research Council Centre for Inflammation Research (I.S., P.T.K.S.), University ofEdinburgh, Edinburgh EH16 4SB, United Kingdom

Androgens increase skeletal muscle mass, but their clinical use is hampered by a lack of tissueselectivity and subsequent side effects. Selective androgen receptor modulators elicit muscle-anabolic effects while only sparingly affecting reproductive tissues. The selective androgen re-ceptor modulator, GTx-024 (enobosarm), is being investigated for cancer cachexia, sarcopenia, andmuscle wasting diseases. Here we investigate the role of muscle androgen receptor (AR) in theanabolic effect of GTx-024. In mice lacking AR in the satellite cell lineage (satARKO), the weight ofthe androgen-sensitive levator ani muscle was lower but was decreased further upon orchidec-tomy. GTx-024 was as effective as DHT in restoring levator ani weights to sham levels. Expressionof the muscle-specific, androgen-responsive genes S-adenosylmethionine decarboxylase and myo-statin was decreased by orchidectomy and restored by GTx-024 and DHT in control mice, whereasthe expression was low and unaffected by androgen status in satARKO. In contrast, insulin-likegrowth factor 1Ea expression was not different between satARKO and control muscle, decreasedupon castration, and was restored by DHT and GTx-024 in both genotypes. These data indicate thatGTx-024 does not selectively modulate AR in the satellite cell lineage and that cells outside this lineageremain androgen responsive in satARKO muscle. Indeed, residual AR-positive cells were present insatARKO muscle, coexpressing the fibroblast-lineage marker vimentin. AR positive, muscle-residentfibroblasts could therefore be involved in the indirect effects of androgens on muscle. In conclusion,bothDHTandGTx-024targetARpathways in thesatellitecell lineage,butcellsoutsidethis lineagealsocontribute to the anabolic effects of androgens. (Endocrinology 156: 4522–4533, 2015)

Androgens play important roles in diverse biologicalprocesses such as development and maintenance of

the male phenotype. In addition, they have effects on sev-eral nonreproductive tissues including bone, smooth mus-cle, and skeletal muscle (1–3). Their biological effects areexerted by activating the androgen receptor (AR). Thisligand-inducible transcription factor binds to specificDNA sequences called androgen response elements(AREs) to regulate transcription of target genes (4).

Despite their well-documented effect of increasingmuscle mass and strength (5, 6), the clinical use of andro-gens is limited due to a lack of tissue selectivity and po-tential subsequent side effects. Together with their poororal bioavailability and pharmacokinetic profile, this hasstimulated the development of selective AR modulators(SARMs). A well-established body of evidence supportstheir in vivo tissue selectivity in animal models, as assessedby their ability to increase levator ani muscle weight and

ISSN Print 0013-7227 ISSN Online 1945-7170Printed in USACopyright © 2015 by the Endocrine SocietyReceived June 1, 2015. Accepted September 16, 2015.First Published Online September 22, 2015

Abbreviations: Amd1, S-adenosylmethionine decarboxylase; AR, androgen receptor; ARE,androgen response element; BC, bulbocavernosus; DAPI, 4�,6�-diamino-2-phenylindole;EDL, extensor digitorum longus; FGF-2, fibroblast growth factor 2; GASTR, gastrocnemius;IGF-1Ea, insulin-like growth factor 1Ea; LA, levator ani; Mstn, myostatin; orx, orchidecto-mized; SARM, selective AR modulator; satARKO, satellite cell-specific knockout of the AR;SV, seminal vesicles; TBS, Tris-buffered saline; vim, vimentin.

O R I G I N A L R E S E A R C H

4522 press.endocrine.org/journal/endo Endocrinology, December 2015, 156(12):4522–4533 doi: 10.1210/en.2015-1479

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 23 March 2016. at 08:45 For personal use only. No other uses without permission. . All rights reserved.

bone mineral density while only sparingly affecting theweight of androgenic tissues (7).

GTx-024 (enobosarm) is a nonsteroidal SARM thatincreases muscle mass with only limited effects on seminalvesicles in preclinical studies (8). Trials in healthy elderlymen (9) and in cancer patients with muscle wasting (10)have reported increased lean body mass and improvedphysical function in the GTx-024-treated groups com-pared with placebo, making this compound a strong can-didate for further clinical development.

Skeletal muscle is a complex tissue, and the impact ofandrogens can be mediated via one or all of the AR-pos-itive cell types. Indeed, androgens can have direct effectson cells of the satellite cell lineage including satellite cells,myoblasts, and myocytes, which all express the AR (1, 11,12). In addition, indirect effects may contribute to theeventual muscle hypertrophy, with endothelial cells andfibroblasts within the muscle being candidate target cellsbecause AR expression has been reported in these celltypes (13, 14).

Whether the anabolic effect of GTx-024 on muscle isdue to selective action on muscle cell AR is presently un-known. To clarify this issue, we performed a castrationand drug replacement study in mice selectively lacking theAR in satellite cells and hence in myoblasts and myocytes(satellite cell-specific AR knockout [satARKO]). The non-aromatizable androgen DHT was used as a positivecontrol.

Materials and Methods

AnimalsMice with a satellite cell-specific knockout of the AR

(satARKO) (ARfl/Y;MyoD-iCre�/�) and control mice (ARfl/Y;MyoD-iCre�/�) (C57BL/6J genetic background) were generatedas described elsewhere (15). The animals were housed in stan-dard cages at 20°C with a 12-hour dark, 12-hour light cycleaccording to our institutional guidelines. They had ad libitumaccess to tap water and standard chow. Offspring were weanedat 3 weeks of age and genotyped by PCR-based analysis of DNAsamples obtained via tail biopsy. The experimental protocol wasconducted with approval of the KU Leuven Ethical Committee(P078–2010).

Experimental designAt 12 weeks of age, male mice were randomly divided into

four groups, which each contained eight mice per genotype. An-imals were weighed and body composition was assessed bywhole-body dual-energy X-ray absorptiometry. In group 1, micewere sham operated (sham) under sodium pentobarbital anes-thesia. In groups 2, 3, and 4, mice were orchidectomized (orx).Mice from groups 3 and 4 were treated with 7 mg/kg�d DHT(number 10300 from Sigma-Aldrich) or 3 mg/kg�d GTx-024(number 43974 from Sigma-Aldrich) for 2 weeks. The drugs

were dissolved in dimethylsulfoxide-polyethylene glycol 300(1:9 [vol/vol]) and administered via daily sc injections in thecervical region. The sham and orx animals were treated withvehicle during the treatment period. At the end of the treatment,animals were weighed and body composition was assessed bywhole-body dual-energy X-ray absorptiometry. Seminal vesiclesand levator ani muscle as well as several limb muscles (gastroc-nemius, extensor digitorum longus, soleus) were collected andweighed after euthanasia by cardiac puncture.

Dual-energy X-ray absorptiometryWhole-body lean and fat mass were analyzed in vivo using the

PIXImus mouse densitometer (Lunar Corp) with ultrahigh res-olution (0.18 � 0.18 pixels, 1.6 line pairs/mm) and softwareversion 1.45.

Serum IGF-I measurementAfter acid-ethanol extraction, serum IGF-I concentrations

were measured by an in-house RIA in the presence of an excessof IGF-2 (25 ng/tube) (16).

Quantitative real-time PCRTotal RNA was extracted from murine muscle using TRIzol

reagent (Invitrogen) according to the manufacturer’s protocols.After digestion with deoxyribonuclease I (Fermentas), cDNAwas synthesized from 300 ng (levator ani muscle) or 1 �g (gas-trocnemius muscle) RNA using the RevertAid Moloney murineleukemia virus reverse transcriptase kit (Fermentas) and randomhexamer primers (Fermentas). The PCR mixtures (10 �L) con-tained 1� platinum SYBR Green quantitative PCR SuperMix-uracil-DNA glycosylase (Invitrogen), 0.15 �M of each primer,and 50 nM ROX reference dye (Invitrogen). The 7500 Fast real-time PCR system (Applied Biosystems) was used with the FastRT-PCR two-step protocol (2 min at 50°C, 20 sec at 95°C, and40 cycles of 3 sec at 95°C and 30 sec at 60°C). The primersequences are listed in Supplemental Table 1. All primers weredesigned to hybridize to separate exons, and the generation ofsingle correct amplicons was checked by DNA sequencing and inmelting curve assays. Gene expression values are expressed rel-ative to the levels of 18S rRNA.

Luciferase reporter assayMouse C2C12 myoblasts were purchased from the American

Type Culture Collection and cultured at 37°C and 5% CO2 inDMEM containing 4,5 g/L glucose, 4 mM L-glutamine, and pen-icillin (100 IU/mL)-streptomycin (100 �g/mL) (Sigma-Aldrich)and supplemented with 10% fetal calf serum (Invitrogen). Fortransfection experiments, cells were seeded in 96-well plates at adensity of 104 cells/well in DMEM supplemented with 5% char-coal-stripped serum (Sigma-Aldrich) and transiently transfectedusing the X-tremeGENE transfection reagent (Roche) with 10 ngof expression plasmid for full-size human AR (17), 100 ng ofluciferase reporter plasmid containing four copies of the ARE ofinterest, and 10 ng of �-galactosidase expression plasmid (Strat-agene), which served as an internal control for transfection ef-ficiency. The luciferase reporter plasmids were generated as de-scribed before (18), and the ARE sequences are listed elsewhere(15). Cells were lysed after stimulation for 24 hours with 100 nMDHT or GTx-024 in the presence or absence of 100 �MMDV3100 (number SRP016825M from Sequoia Research

doi: 10.1210/en.2015-1479 press.endocrine.org/journal/endo 4523


http://press.endocrine.org/doi/suppl/10.1210/en.2015-1479/suppl_file/en-15-1479.pdf

Products), and luciferase and �-galactosidase activities weremeasured and calculated as previously described (19).

Double-antibody fluorescent immunohistochemistryDouble-antibody fluorescent immunohistochemistry was

performed as described previously (20). Briefly, levator ani andgastrocnemius muscles were fixed in 10% neutral-buffered for-malin (Sigma-Aldrich) overnight at 4°C, rinsed with PBS, andthen stored in 70% ethanol at 4°C. After embedding into par-affin wax, 5-�m sections were cut onto microslides, deparaf-finized, rehydrated, and subjected to heat-induced antigen re-trieval in a decloaking chamber containing 0.01 M citrate buffer.The tissue sections were then incubated with 3% hydrogen per-oxide in methanol for 30 minutes at room temperature to blockendogenous peroxidase activity, rinsed with water, and blockedfor 30 minutes in goat serum (Biosera) diluted 1:4 in Tris-buff-ered saline (TBS; 50 mM Tris, pH 7.4; 0.85% saline) containing5% BSA. Primary antibodies were diluted in goat serum/TBS/BSA and incubated at 4°C overnight. Details of the primary an-tibodies are provided in Table 1. Secondary antibody (goat an-tirabbit peroxidase [Abcam; number ab7171] or goat antimouseperoxidase [Abcam; number ab6823]) was diluted 1:500 in goatserum/TBS/BSA and applied to sections for 45 minutes at roomtemperature. Finally, slides were incubated with the TyramideSignal Amplification kit (PerkinElmer; fluorescein for AR andcyanine 3 for the other markers) for 10 minutes and then with4�,6�-diamino-2-phenylindole (DAPI; Sigma-Aldrich) diluted1:500 in TBS to enable nuclear counterstaining. Washes betweenthe several incubations were performed three times for 3 minuteseach using TBS. Immunostaining of negative controls, which didnot show any antiserum immunolabeling, included sequentialelimination of either the primary or secondary antibody from thestaining procedure. Sections were examined using a Zeiss LSM510 Meta-confocal microscope equipped with a digital camera,and images were analyzed by ImageJ software (National Insti-tutes of Health, Bethesda, Maryland). The percentage of the AR-positive cells corresponding to fibroblasts (identified as vimentinpositive) in the levator ani was calculated of 286 � 73 AR-positive cells per animal. All procedures were carried out usingcoded slides to avoid bias.

Statistical analysisStatistical analyses were performed using GraphPad Prism

software (version 6.00 for Windows). A Fisher’s exact test was

used for categorical variables, whereas for ordinal variables, aStudent’s t test and a one-way ANOVA were used to analyze thedifferences between two or more groups, respectively. A two-way ANOVA was used in experiments with more than two in-dependent variables. If an overall ANOVA revealed a significantdifference, Bonferroni’s post hoc test was used to analyze thedifferences between groups. All statistical tests were performedtwo tailed. Data are presented as means � SEM, and P � .05 wasconsidered statistically significant.

Results

GTx-024 is a tissue-selective compound withanabolic effect on murine androgen-sensitivemuscle

To validate the doses used and to confirm the tissueselectivity of GTx-024 in our experimental setting, a pilotexperiment was performed in which the weight of the LAand seminal vesicles (SV) of the control mice were com-pared among the different treatment groups. orx at 12weeks of age led to markedly reduced levator ani (LA) andSV weights after 2 weeks, and both DHT and GTx-024increased the weight of these organs when drug treatmentwas initiated immediately after orx (Figure 1). However,whereas DHT and GTx-024 restored LA weight to shamlevel (Figure 1A), they had a differential effect on SVweight. Indeed, DHT supplementation fully normalizedSV weight, whereas only partial restoration was observedin the GTx-024-treated group (Figure 1B). Gastrocnemius(GASTR) muscle mass was not affected by orx with orwithout drug replacement (Supplemental Figure 1A),which correlates with the lower levels of AR protein (Sup-plemental Figure 1, B and C). Thus, we conclude thatGTx-024 acts as a SARM with an anabolic effect on LAmuscle equivalent to physiological androgen concentra-tions in this experimental setting.

Table 1. Antibody Table: Details of the Primary Antibodies Used for Fluorescent Immunohistochemistry

Peptide/ProteinTarget Antigen Sequence (if Known) Name of Antibody

Manufacturer, Catalog Number,and/or Name of IndividualProviding the Antibody

Species Raised(Monoclonal orPolyclonal)

DilutionUsed

AR Synthetic peptide derived from near N terminusof human AR

Rabbit antihuman AR monoclonalantibody (clone SP107)

Spring Bioscience; M4070 Rabbit monoclonal 1:300

Myogenin Recombinant protein containing rat myogeninamino acids 30–224

Antimyogenin antibody (F5D) Abcam; ab1835 Mouse monoclonal 1:50

CD31 Synthetic peptide corresponding to C terminusof mouse CD31

Anti-CD31 antibody Abcam; ab28364 Rabbit polyclonal 1:500

Vim Synthetic peptide corresponding to residuessurrounding Arg45 of human vim

Vim (D21H3) XP rabbit monoclonalantibody

Cell Signaling Technology; 5741 Rabbit monoclonal 1:600

Laminin Protein purified from the basement membraneof Englebreth Holm-Swarm (EHS) sarcoma(mouse)

Antilaminin antibody Abcam; ab11575 Rabbit polyclonal 1:300

Pax7 Purified internal fragment of human recombinantPAX7 expressed in Escherichia coli

PAX7 antibody Thermo Scientific; PA1–117 Rabbit polyclonal 1:500

4524 Dubois et al Enobosarm Action on Skeletal Muscle Endocrinology, December 2015, 156(12):4522–4533


Muscle AR is not essential for modulation ofmuscle mass by GTx-024

To investigate whether the anabolic effect of GTx-024is mediated via AR in muscle cells, we performed the samecastration experiment with drug replacement in the sa-tARKO model. In these mice, the AR was selectively ab-lated in the muscle progenitor cells called satellite cells,leading to a more than 2-fold reduction in LA weight (15)(sham conditions in Figure 2A).

It is known that lean body mass or appendicular musclemass in mice is considerably less sensitive than LA muscle(11), especially in short-term experiments. Indeed, bodyweight and body composition were not different betweenthe groups, neither at baseline nor after the treatment pe-riod (Supplemental Figure 2, A and B). In addition, limbmuscle mass was not different between the satARKO andcontrol mice and was unaffected by orx without or withDHT or GTx-024, as evidenced by the similar weight ofGASTR, extensor digitorum longus (EDL), and soleusmuscles in the different treatment groups (SupplementalFigure 2C). Therefore, we focused in further experimentson the perineal LA muscle, which is widely used as a read-out for androgens in rodents including in SARM devel-opment (21).

orx decreased LA mass in control mice, an effect thatwas also observed in satARKO mice (Figure 2A). Inter-estingly, GTx-024 reversed the orx effect on LA muscle ofsatARKO mice, to the same extent as did DHT (Figure2A). These observations were confirmed in another peri-neal muscle, the bulbocavernosus (BC) (Figure 2B). As acontrol for the in vivo effectiveness of the treatments, theSV weights for each condition are depicted in Figure 2C.

The rescue of LA mass to sham level observed in cas-trated DHT- as well as GTx-024-treated satARKO micemay reflect an indirect nonmuscle AR contribution. Al-

ternatively, residual muscle AR ex-pression could be responsible for thisanabolic response. To elucidate thisquestion, we measured the expres-sion of S-adenosylmethionine decar-boxylase 1 (Amd1) and myostatin(Mstn), two genes shown to bestrongly androgen regulated in skel-etal muscle (15, 22). Indeed, a morethan 10-fold reduction in Amd1 andMstn mRNA was observed in LAmuscle of satARKO compared withcontrol mice (sham conditions inFigure 3, A and B). orx decreasedAmd1 and Mstn transcript levels incontrol mice, but no further decreasewas observed in the satARKO mice

(Figure 3, A and B), demonstrating that muscle AR is suf-ficient for androgen regulation of these muscle-specificgenes. In addition, these findings confirm effective disrup-tion of muscle AR in satARKO, at least at the functionallevel. An orx-mediated decrease in Amd1 and MstnmRNA was reversed by DHT and also by GTx-024 incontrol mice (Figure 3, A and B), indicating that GTx-024action on muscle is, at least in part, direct via muscle ARactivation. Importantly, however, neither DHT nor GTx-024 was able to induce Amd1 or Mstn expression in thesatARKO LA (Figure 3, A and B), although they had aclear effect on muscle mass, supporting the hypothesis ofan indirect nonmuscle AR contribution to muscle mass byandrogens and GTx-024. Similar Amd1 expression pat-terns were observed in GASTR muscle (Supplemental Fig-ure 3A). Mstn transcript levels were, however, not alteredby orx with or without drug replacement in GASTR mus-cle, albeit lower in the satARKO compared with the con-trol samples (two way ANOVA, P � .02 for genotype)(Supplemental Figure 3B).

We previously identified two conserved AREs in thepromotor and in exon 2 of the Mstn gene, referred to asARE1 and ARE2, both of them binding the DNA-bindingdomainof theARandconferringandrogen responsivenessto a heterologous promotor (15). ARE1 is part of an ARbinding site found in the chromatin of primary humanmyoblasts by chromatin immunoprecipitation-on-chipanalysis (23). In a transient transfection assay in theC2C12 muscle cell line, ARE1- and ARE2-based reportergenes displayed responsiveness to DHT and GTx-024,which was strongly reduced in the presence of the ARantagonist MDV3100 (Supplemental Figure 4). Thus, inline with the above-mentioned in vivo findings, these invitro data indicate that Mstn gene transcription is a goodreadout of muscle-specific androgen responsiveness and

Figure 1. GTx-024 is a tissue-selective compound with anabolic effect on muscle. A and B, LAmuscle weight (A) and SV weight (B) of 14-week-old control mice that were sham operated(sham), orchidectomized (orx), or orchidectomized and treated with DHT (orx�DHT) or GTx-024(orx�GTx-024) at 12 weeks of age (n � 3). Organ weights were corrected for body weight (BW).Error bars indicate SEM. *, P � .05.



that muscle AR is involved in the anabolic effect ofGTx-024.

Insulin-like growth factor 1Ea (IGF-1Ea), an IGF-1 iso-form locally expressed within muscle, is an important pos-itive regulator of muscle mass (24), and its expression isregulated by androgens (25, 26). Indeed, orx decreasedIGF-1Ea transcript levels in the LA of control mice, aneffect that was reversed by DHT and also by GTx-024

(Figure 3C). Importantly, however, the similar IGF-1Eatranscript levels in control and satARKO sham-operatedanimals as well as the decrease in IGF-1Ea mRNA in cas-trated satARKO mice (Figure 3C) indicate that androgenregulation of IGF-1Ea is indirect via nonmuscle AR. Anorx-mediated decrease in IGF-1Ea mRNA was reversedby DHT and also by GTx-024 in both genotypes (Figure3C), opening up the possibility that IGF-1 signaling may

Figure 2. GTx-024 effects on muscle mass in the satARKO model. LA weight (A), BC weight (B), and SV weight (C) of 14-week-old control andsatARKO mice that were sham operated (sham), orchidectomized (orx), or orchidectomized and treated with DHT (orx�DHT) or GTx-024(orx�GTx-024) at 12 weeks of age (n � 7–8). Organ weights were corrected for body weight (BW). Error bars indicate SEM. *, P � .05.



be involved in the indirect nonmuscle AR contribution tomuscle mass by androgens and GTx-024. Serum IGF-1levels were similar in satARKO and control mice and un-affected by orx without or with DHT or GTx-024 (Sup-plemental Figure 5).

Altogether we conclude that both muscle cell and non-muscle cell AR play a role in DHT and GTx-024 action onmuscle. The tissue selectivity of GTx-024 is, however, notexplained by the selective targeting of muscle cell AR, asevidenced by the retained efficacy in the satARKO model.

Fibroblasts may be involved in indirect androgenaction on muscle

We next examined which cell types within the muscleother than the muscle cells themselves may mediate theindirect anabolic effect of androgens. Although AR pro-tein is expressed predominantly in satellite cells and myo-nuclei, endothelial cells and fibroblasts within the musclehave also both been reported to express the AR (13, 14).Therefore, LA muscle sections from control and satARKOmice were double stained for the AR on the one hand and

Figure 3. Amd1, Mstn, and IGF-1Ea mRNA expression in LA muscle. Quantitative real-time PCR analysis of Amd1 (A), Mstn (B), and IGF-1Ea (C)mRNA in LA muscle of 14-week-old control and satARKO mice that were sham operated (sham), orchidectomized (orx), or orchidectomized andtreated with DHT (orx�DHT) or GTx-024 (orx�GTx-024) at 12 weeks of age (n � 7–8). Error bars indicate SEM. *, P � .05.



for myogenin, CD31, or vimentin (vim) on the other hand(markers for muscle fibers, endothelial cells, and fibro-blasts, respectively). In control mice, we observed AR�

nuclei coexpressing not only the muscle marker myogenin(Figure 4A) but also the fibroblast marker vim (Figure 4B).The stainings revealed no coexpression of AR and CD31

(Figure 4C). In satARKO LA muscle,the overall number of AR� cells wasconsiderably reduced (see quantifi-cation in Figure 6A), in line with thestrongly reduced mRNA levels re-ported previously (15). Still, we ob-served many remaining AR� nuclei,which appeared to be located almostentirely outside muscle fibers (whitearrows in Figure 4A). The nonsatel-lite cell nature of these remainingAR� nuclei was confirmed by theirlocalization outside the basal lamina(Figure 5A) as well as the absence ofPax7 coexpression (Figure 5B). Mostof the AR� nuclei in the satARKOLA were identified as fibroblasts, asdetermined by positive staining tovim (Figure 4B).

A more detailed examination ofthe latter double staining revealedthat, although control LA muscleconsists mostly of AR� vim� cells(white arrows in upper panel of Fig-ure 6B) with some cells being AR�

vim� (red arrows in upper panel ofFigure 6B), the vast majority of AR�

cells in satARKO LA muscle also ex-press vim (red arrows in lower panelof Figure 6B). Quantification of AR�

and vim� cells is shown in Figure 6C.These findings were confirmed inGASTR muscle (Supplemental Fig-ure 6). Altogether the intranuclearpresence of AR suggests that fibro-blasts could be involved in the indi-rect anabolic effect of androgens onskeletal muscle.

Discussion

Muscle wasting is a hallmark of ag-ing (27) and also occurs in variouschronic disorders such as cancer (27,28). Androgens could potentially beexploited in these patient groups be-

cause they increase both muscle mass and strength (5, 6).Their anabolic action is thought to be mediated by a dualmechanism, ie, direct activation of muscle AR as well asindirect action through nonmuscle AR pathways (1).However, because androgen treatment is associated with

Figure 4. Microscopy of cell types identified by double staining for AR and cell type-specificmarkers. A, LA muscle sections from 8-week-old control and satARKO mice were stained for AR(green), myogenin (red), and DAPI (blue). Myogenin is a marker for muscle cells. The remainingAR� nuclei in satARKO LA muscle appear to be located outside the muscle fibers (white arrows).B, LA muscle sections from 8-week-old control and satARKO mice were stained for AR (green),vim (red), and DAPI (blue). vim is a marker for fibroblasts. Positive staining for vim was detectedin AR� cells (red arrows). C, LA muscle sections from 8-week-old control and satARKO mice werestained for AR (green), CD31 (red), and DAPI (blue). CD31 is a marker for endothelial cells. CD31staining was detected around blood vessels (white arrowheads) and did not colocalize with AR�

cells (white arrows). Scale bar, 20 �m.



potential cardiovascular and pros-tate cancer risks, SARMs were devel-oped as an alternative strategy tocounteract muscle atrophy (7). GTx-024 or enobosarm (also known asostarine and S-22) is clinically wellcharacterized and has consistentlydemonstrated increases in lean bodymass across various populations(29). The molecular mechanisms be-hind its tissue selectivity and ana-bolic action on muscle itself, how-ever, remain elusive. The aim of thepresent study was therefore to inves-tigate whether the anabolic effect ofGTx-024 is mediated via AR in mus-cle cells. To this end, we performed acastration and drug replacement ex-periment in mice selectively lackingthe AR in satellite cells and hence inmyoblasts and myocytes (satARKO)(15) and compared the effects ofGTx-024 with those of DHT.

Skeletal muscles differ markedlyin their responsiveness to androgens.For example, the perineal skeletalmuscles LA and BC are highly an-drogen responsive and depend onandrogens for their normal mainte-nance and function, whereas thelimb skeletal muscle EDL is rela-tively unresponsive to androgensand does not depend on androgens tomaintain fiber size (30). In accor-dance, selective AR ablation in mus-cle reduces BC/LA but not limbmuscle mass (11, 15). Immunohisto-chemical staining of muscle sectionsrevealed that the BC/LA complexcontains much more AR proteinthan do less responsive muscles likeGASTR or EDL (13, 31), a findingthat was confirmed in this study.Thus, differences in AR protein con-tent of skeletal muscles seem to un-derlie differences in androgen re-sponsiveness. This difference inandrogen sensitivity is illustrated bythe fact that in this study 2 weeks oftreatment with DHT or GTx-024were sufficient to increase LA massin orx animals but not to alter lean

Figure 5. The remaining AR� nuclei in satARKO muscle are not satellite cells. A, LA musclesections from 8-week-old control and satARKO mice were stained for AR (green), laminin (red),and DAPI (blue). Laminin is a structural component of the basal lamina. B, LA muscle sectionsfrom 8-week-old control and satARKO mice were stained for AR (green), Pax7 (red), and DAPI(blue). Pax7 is a marker for satellite cells. The nonsatellite cell nature of the remaining AR� nucleiin satARKO muscle is confirmed by their localization outside the basal lamina as well as theabsence of Pax7 coexpression. Scale bar, 50 �m.



body mass, in contrast with the increase in lean mass ob-served upon 8 weeks of treatment with DHT or the struc-turally related GTx-007 compound (32). Due to its highandrogen responsiveness, the LA muscle is widely ac-cepted as a readout for androgen anabolic action in pre-clinical and pharmacological studies including in SARMdevelopment (21). For this reason, we decided to focus onLA muscle to study the mechanisms of GTx-024 action.

A direct role for muscle AR in mediating androgen an-abolic action was demonstrated by various mouse modelsin which the AR was specifically ablated in progenitor (15)or mature (11, 12) muscle cells. These three models allshowed a muscle phenotype, albeit generally mild. To de-termine the implication of muscle AR in mediating GTx-024 action, we assessed the expression of two musclegenes, Amd1 and Mstn, in response to GTx-024 treat-ment. Amd1 is a key enzyme in the synthesis pathway ofpolyamines, which are increased in rodent models of mus-cle hypertrophy (33). Mstn is a strong negative regulatorof muscle growth because disruption of the Mstn gene

induces a dramatic increase in mus-cle mass (34). These genes are andro-gen regulated in skeletal muscle be-cause Amd1 (22) and Mstn (35)mRNA levels were decreased in orxmice and restored by the administra-tion of T. Recently we identifiedmuscle AR as a mediator of this an-drogen regulation. Indeed, Mstn aswell as Amd1 transcript levels werereduced in satARKO muscle (15).GTx-024 up-regulated the mRNAlevels of both genes in the LA muscleof castrated control mice, indicatinga role for the muscle AR in GTx-024action.

None of the muscle-specificARKO models fully reproduces themuscle phenotype of the globalARKO (22, 36), suggesting thatmuscle cells are not the sole target forandrogen action in muscle. In accor-dance, cell-specific overexpressionof wild-type AR in skeletal muscle oftesticular feminized rats is not suffi-cient to rescue BC/LA mass (37). Ad-ditional evidence for indirect andro-gen action on muscle is provided bythe observation that orx further de-creases BC/LA mass in satARKOmice, an effect that is fully reversedby DHT supplementation (15).

Thus, the satARKO model could be further exploited todemonstrate a tissue-selective mechanism via muscle AR.However, whereas regulation of the muscle-specific an-drogen target genes Amd1 and Mstn was completely ab-lated in satARKO, GTx-024 still reversed the orx effect onBC/LA mass as efficiently as DHT in vivo. Hence, our datasuggest that GTx-024 action is both direct via muscle ARand indirect via nonmuscle AR pathways, just as for otherandrogens.

A first potential pathway of indirect androgen actionon muscle is by affecting the nonmuscular fraction of thetissue. Indeed, AR expression within the muscle is notrestricted to satellite cells and myonuclei but also has beendescribed in endothelial cells and fibroblasts (13, 14).Therefore, in this study, LA muscle sections were doublestained for the AR on the one hand and for CD31 or vimon the other hand (markers for endothelial cells and fi-broblasts, respectively). In our conditions, CD31� cellswere scarce throughout the LA and did not show AR ex-pression. However, resident fibroblasts within the LA ex-

Figure 6. Fibroblasts may be involved in indirect androgen action on muscle. A, Number of AR�

cells per field in LA muscle sections from 8-week-old control and satARKO mice (n � 3). B, LAmuscle sections from 8-week-old control and satARKO mice were stained for AR (green), vim(red), and DAPI (blue). Whereas control LA muscle consists mostly of AR� vim� cells (whitearrows in upper panel) with some cells being AR� vim� (red arrows in upper panel), the vastmajority of AR� cells in satARKO LA muscle also expresses vim (red arrows in lower panel). Scalebar, 20 �m. C, Percentage of AR� cells that stained negatively (white) or positively (black) for vimwas calculated from LA muscle sections from 8-week-old control and satARKO mice (n � 3).Error bars indicate SEM. *, P � .05.



press nuclear AR, thus suggesting that these cells could beinvolved in the androgen action on muscle. This hypoth-esis is further supported by the fact that in satARKO LAmuscle, which shows androgen responsiveness in the ab-sence of muscle AR, the vast majority of remaining AR�

cells also expresses vim. Also in GASTR muscle, vim isexpressed in some AR� cells from the control mice as wellas in most AR� cells from the satARKO animals. Thus,although our study focused on LA muscle, our findingscan be extended to limb skeletal muscles.

Several studies support the involvement of muscle fi-broblasts in the development, maintenance, and regener-ation of muscle structure and function. Mice devoid ofTranscription factor 7-like 2, which is abundantly ex-pressed in muscle fibroblasts but not in myogenic cells,display an abnormal muscular morphology, with severalmuscles being aberrantly split along with an altered mus-cle fiber type (38). Similarly, the ablation of Transcriptionfactor 7-like 2 during muscle regeneration leads to pre-mature satellite cell differentiation and smaller myofibers(39). The exact mechanisms underlying the paracrine con-trol of myoblast proliferation and differentiation by fi-broblasts are unclear. Early studies have shown that incultures composed of both myogenic and fibroblast-likecells, myoblasts exhibit a prolonged proliferative phase,resulting in delayed but increased production of fusedmyotubes. Subsequent experiments with conditioned me-dia indicated that the fibroblast-like cells produce solublemyogenic growth factors but failed to determine theiridentity (40). Recently fibroblast growth factor 2 (FGF-2)was proposed as a possible mediator because neutralizingantibodies to FGF-2 were able to block the fibroblast ef-fects on myotubes. However, inhibition due to FGF-2 wasnoticed only in coculture experiments allowing direct cell-cell contact between both cell types and not in conditionedmedia experiments (41). Thus, the exact nature and iden-tity of the fibroblast-derived paracrine factors stimulatingmyogenesis remain presently unknown.

Although androgen effects on fibroblasts have been in-tensively studied in prostate cancer (42), androgen actionon muscle fibroblasts remains to be investigated. Recentlya mouse model was generated in which the AR was ablatedin mesenchymal cells (43). In these mutant mice, theBC/LA muscle complex failed to develop. Moreover, be-cause the number of proliferating undifferentiated myo-blasts was reduced, the authors suggested that the mesen-chymal AR may regulate the proliferation of musclemyoblasts in a paracrine way (43). Because muscle fibro-blasts are from mesenchymal origin, this study supportsthe involvement of fibroblasts in indirect androgen actionon muscle. Additional evidence is provided by in vitroexperiments, in which androgen treatment of the C3H

10T1/2 pluripotent mesenchymal cell line up-regulatedmyogenic differentiation markers (44). To confirm, how-ever, that the indirect pathway, potentially via muscle-resident fibroblasts, affects adult muscle homeostasis,studies using an inducible Cre-LoxP model targeting thesefibroblasts (with minimal changes in other organs) wouldbe required.

Transcript levels of IGF-1Ea were not different be-tween the satARKO and control muscle but were de-creased upon castration in both genotypes, indicating in-direct androgen regulation of IGF-1Ea via nonmuscle AR.DHT as well as GTx-024 rescued the orx-mediated de-crease in IGF-1Ea mRNA, suggesting that local IGF-1signaling may be involved in the indirect nonmuscle ARcontribution to muscle mass by androgens and GTx-024.There were no differences in serum IGF-1 levels. There isincreasing evidence that, in contrast to circulating IGF-1(45), locally produced IGF-1 is an important mediator ofandrogen action in muscle. Indeed, androgen treatmentincreases and orchidectomy decreases IGF-1 mRNA inmuscle (12, 25). However, because both fibroblasts (46)and muscle cells (47) produce IGF-1, further studies areneeded to determine whether, in addition to a direct effecton IGF-1 production by muscle cells (26), androgens stim-ulate IGF-1 production by fibroblasts with subsequentproliferative effects on neighboring muscle cells or, alter-natively, enhance the secretion of myogenic factors by fi-broblasts, which will lead to increased IGF-1 productionby the muscle cells themselves.

Although in this study we were interested in the localeffects of androgens on muscle, it must be taken into ac-count that androgens might also promote muscle functionby acting on other organs or systems (1, 48). A limitednumber of studies have assessed this possibility, focusingmainly on the effects of androgens on muscle innervationand, in particular, on the highly androgen-sensitive spinalmotor neurons supplying the LA (49, 50). Although thesestudies suggest that the neuronal AR is not crucial formaintaining LA mass, this has not been confirmed yet witha neuron-specific AR knockout model.

In summary, the mechanism of action of GTx-024 onskeletal muscle is partly direct via muscle AR activation. Inaddition, part of the anabolic effect seems to be indirect vianonmuscle AR pathways. Muscle fibroblasts may play arole in these indirect pathways, although additional stud-ies are needed to clarify the exact molecular mechanismsof this indirect activity. Moreover, further investigation isrequired to confirm whether other SARMs exist or can bedesigned that have a more muscle-specific action.



Acknowledgments

We thank Arantza Esnal-Zufiaurre, Rita Bollen, Hilde Debruyn,Dieter Schollaert, Erik Van Herck, and Ludo Deboel for theirexcellent technical assistance. We are grateful to James Daltonand Ramesh Narayanan for their comments on the experimentaldesign of the in vivo experiments.

Address all correspondence and requests for reprints to: FrankClaessens, PhD, Molecular Endocrinology Laboratory, Depart-ment of Cellular and Molecular Medicine, KU Leuven, CampusGasthuisberg O and N1 PO Box 901, Herestraat 49, 3000 Leuven,Belgium. E-mail: [email protected].

This work was supported by Grant GOA/15/017 from KULeuven and Grant G.0858.11 from the Research FoundationFlanders. V.D. and M.R.L. are holders of a doctoral fellowshipof the Research Foundation Flanders. I.S. and P.T.K.S. are sup-ported by Grant G1100356/1 from the Medical Research Coun-cil United Kingdom.

Disclosure Summary: The authors have nothing to disclose.

References

1. Dubois V, Laurent M, Boonen S, Vanderschueren D, Claessens F.Androgens and skeletal muscle: cellular and molecular action mech-anisms underlying the anabolic actions. Cell Mol Life Sci. 2012;69:1651–1667.

2. Welsh M, Moffat L, Jack L, et al. Deletion of androgen receptor inthe smooth muscle of the seminal vesicles impairs secretory functionand alters its responsiveness to exogenous testosterone and estra-diol. Endocrinology. 2010;151:3374–3385.

3. Vanderschueren D, Laurent MR, Claessens F, et al. Sex steroid ac-tions in male bone. Endocr Rev. 2014;35(6):906–960.

4. Claessens F, Denayer S, Van Tilborgh N, Kerkhofs S, Helsen C,Haelens A. Diverse roles of androgen receptor (AR) domains inAR-mediated signaling. Nucl Recept Signal. 2008;6:e008.

5. Bhasin S, Storer TW, Berman N, et al. The effects of supraphysi-ologic doses of testosterone on muscle size and strength in normalmen. N Engl J Med. 1996;335:1–7.

6. Finkelstein JS, Lee H, Burnett-Bowie SA, et al. Gonadal steroids andbody composition, strength, and sexual function in men. N EnglJ Med. 2013;369:1011–1022.

7. Mohler ML, Bohl CE, Jones A, et al. Nonsteroidal selective andro-gen receptor modulators (SARMs): dissociating the anabolic andandrogenic activities of the androgen receptor for therapeutic ben-efit. J Med Chem. 2009;52:3597–3617.

8. Kim J, Wu D, Hwang DJ, Miller DD, Dalton JT. The para substitu-ent of S-3-(phenoxy)-2-hydroxy-2-methyl-N-(4-nitro-3-trifluorom-ethyl-phenyl)-propionamides is a major structural determinant of invivo disposition and activity of selective androgen receptor modu-lators. J Pharmacol Exp Ther. 2005;315:230–239.

9. Dalton JT, Barnette KG, Bohl CE, et al. The selective androgenreceptor modulator GTx-024 (enobosarm) improves lean body massand physical function in healthy elderly men and postmenopausalwomen: results of a double-blind, placebo-controlled phase II trial.J Cachexia Sarcopenia Muscle. 2011;2:153–161.

10. Dobs AS, Boccia RV, Croot CC, et al. Effects of enobosarm onmuscle wasting and physical function in patients with cancer: a dou-ble-blind, randomised controlled phase 2 trial. Lancet Oncol. 2013;14:335–345.

11. Ophoff J, Van Proeyen K, Callewaert F, et al. Androgen signaling inmyocytes contributes to the maintenance of muscle mass and fiber

type regulation but not to muscle strength or fatigue. Endocrinol-ogy. 2009;150:3558–3566.

12. Chambon C, Duteil D, Vignaud A, et al. Myocytic androgen recep-tor controls the strength but not the mass of limb muscles. Proc NatlAcad Sci USA. 2010;107:14327–14332.

13. Johansen JA, Breedlove SM, Jordan CL. Androgen receptor expres-sion in the levator ani muscle of male mice. J Neuroendocrinol.2007;19:823–826.

14. Sinha-Hikim I, Taylor WE, Gonzalez-Cadavid NF, Zheng W, Bha-sin S. Androgen receptor in human skeletal muscle and culturedmuscle satellite cells: up-regulation by androgen treatment. J ClinEndocrinol Metab. 2004;89:5245–5255.

15. Dubois V, Laurent MR, Sinnesael M, et al. A satellite cell-specificknockout of the androgen receptor reveals myostatin as a directandrogen target in skeletal muscle. FASEB J. 2014;28:2979–2994.

16. Verhaeghe J, van Herck E, Visser WJ, et al. Bone and mineral me-tabolism in BB rats with long-term diabetes. Decreased bone turn-over and osteoporosis. Diabetes. 1990;39:477–482.

17. Helsen C, Dubois V, Verfaillie A, et al. Evidence for DNA-bindingdomain-ligand-binding domain communications in the androgenreceptor. Mol Cell Biol. 2012;32:3033–3043.

18. Denayer S, Helsen C, Thorrez L, Haelens A, Claessens F. The rulesof DNA recognition by the androgen receptor. Mol Endocrinol.2010;24:898–913.

19. Verrijdt G, Schauwaers K, Haelens A, Rombauts W, Claessens F.Functional interplay between two response elements with distinctbinding characteristics dictates androgen specificity of the mousesex-limited protein enhancer. J Biol Chem. 2002;277:35191–35201.

20. Cousins FL, Murray A, Esnal A, Gibson DA, Critchley HO, Saun-ders PT. Evidence from a mouse model that epithelial cell migrationand mesenchymal-epithelial transition contribute to rapid restora-tion of uterine tissue integrity during menstruation. PLoS One.2014;9:e86378.

21. MacLean HE, Handelsman DJ. Unraveling androgen action in mus-cle: genetic tools probing cellular mechanisms. Endocrinology.2009;150:3437–3439.

22. MacLean HE, Chiu WS, Notini AJ, et al. Impaired skeletal muscledevelopment and function in male, but not female, genomic andro-gen receptor knockout mice. FASEB J. 2008;22:2676–2689.

23. Wyce A, Bai Y, Nagpal S, Thompson CC. Research resource: theandrogen receptor modulates expression of genes with critical rolesin muscle development and function. Mol Endocrinol. 2010;24:1665–1674.

24. Shavlakadze T, Winn N, Rosenthal N, Grounds MD. Reconcilingdata from transgenic mice that overexpress IGF-I specifically in skel-etal muscle. Growth Horm IGF Res. 2005;15:4–18.

25. Lewis MI, Horvitz GD, Clemmons DR, Fournier M. Role of IGF-Iand IGF-binding proteins within diaphragm muscle in modulatingthe effects of nandrolone. Am J Physiol Endocrinol Metab. 2002;282:E483–E490.

26. Kamanga-Sollo E, Pampusch MS, Xi G, White ME, Hathaway MR,Dayton WR. IGF-I mRNA levels in bovine satellite cell cultures:effects of fusion and anabolic steroid treatment. J Cell Physiol. 2004;201:181–189.

27. Gielen E, Verschueren S, O’Neill TW, et al. Musculoskeletal frailty:a geriatric syndrome at the core of fracture occurrence in older age.Calcif Tissue Int. 2012;91:161–177.

28. Ebner N, Springer J, Kalantar-Zadeh K, et al. Mechanism and noveltherapeutic approaches to wasting in chronic disease. Maturitas.2013;75:199–206.

29. Dalton JT, Taylor RP, Mohler ML, Steiner MS. Selective androgenreceptor modulators for the prevention and treatment of musclewasting associated with cancer. Curr Opin Support Palliat Care.2013;7:345–351.

30. Lubischer JL, Bebinger DM. Regulation of terminal Schwann cell



mailto:[email protected]

number at the adult neuromuscular junction. J Neurosci. 1999;19:RC46.

31. Monks DA, O’Bryant EL, Jordan CL. Androgen receptor immuno-reactivity in skeletal muscle: enrichment at the neuromuscular junc-tion. J Comp Neurol. 2004;473:59–72.

32. Gao W, Reiser PJ, Coss CC, et al. Selective androgen receptor mod-ulator treatment improves muscle strength and body compositionand prevents bone loss in orchidectomized rats. Endocrinology.2005;146:4887–4897.

33. Lee NK, MacLean HE. Polyamines, androgens, and skeletal musclehypertrophy. J Cell Physiol. 2011;226:1453–1460.

34. McPherron AC, Lawler AM, Lee SJ. Regulation of skeletal musclemass in mice by a new TGF-� superfamily member. Nature. 1997;387:83–90.

35. Ibebunjo C, Eash JK, Li C, Ma Q, Glass DJ. Voluntary running,skeletal muscle gene expression, and signaling inversely regulated byorchidectomy and testosterone replacement. Am J Physiol Endocri-nol Metab. 2011;300:E327–E340.

36. Ophoff J, Callewaert F, Venken K, et al. Physical activity in theandrogen receptor knockout mouse: evidence for reversal of andro-gen deficiency on cancellous bone. Biochem Biophys Res Commun.2009;378:139–144.

37. Niel L, Shah AH, Lewis GA, et al. Sexual differentiation of the spinalnucleus of the bulbocavernosus is not mediated solely by androgenreceptors in muscle fibers. Endocrinology. 2009;150:3207–3213.

38. Mathew SJ, Hansen JM, Merrell AJ, et al. Connective tissue fibro-blasts and Tcf4 regulate myogenesis. Development. 2011;138:371–384.

39. Murphy MM, Lawson JA, Mathew SJ, Hutcheson DA, Kardon G.Satellite cells, connective tissue fibroblasts and their interactions arecrucial for muscle regeneration. Development. 2011;138:3625–3637.

40. Quinn LS, Ong LD, Roeder RA. Paracrine control of myoblastproliferation and differentiation by fibroblasts. Dev Biol. 1990;140:8 –19.

41. Rao N, Evans S, Stewart D, et al. Fibroblasts influence muscle pro-genitor differentiation and alignment in contact independent anddependent manners in organized co-culture devices. Biomed Micro-devices. 2013;15:161–169.

42. Yu S, Xia S, Yang D, et al. Androgen receptor in human prostatecancer-associated fibroblasts promotes prostate cancer epithelialcell growth and invasion. Med Oncol. 2013;30:674.

43. Ipulan LA, Suzuki K, Sakamoto Y, et al. Nonmyocytic androgenreceptor regulates the sexually dimorphic development of the em-bryonic bulbocavernosus muscle. Endocrinology. 2014;155:2467–2479.

44. Singh R, Artaza JN, Taylor WE, Gonzalez-Cadavid NF, Bhasin S.Androgens stimulate myogenic differentiation and inhibit adipo-genesis in C3H 10T1/2 pluripotent cells through an androgen re-ceptor-mediated pathway. Endocrinology. 2003;144:5081–5088.

45. Serra C, Bhasin S, Tangherlini F, et al. The role of GH and IGF-I inmediating anabolic effects of testosterone on androgen-responsivemuscle. Endocrinology. 2011;152:193–206.

46. Clemmons DR. Multiple hormones stimulate the production of so-matomedin by cultured human fibroblasts. J Clin EndocrinolMetab. 1984;58:850–856.

47. Tollefsen SE, Lajara R, McCusker RH, Clemmons DR, Rotwein P.Insulin-like growth factors (IGF) in muscle development. Expressionof IGF-I, the IGF-I receptor, and an IGF binding protein duringmyoblast differentiation. J Biol Chem. 1989;264:13810–13817.

48. Bonewald LF, Kiel DP, Clemens TL, et al. Forum on bone andskeletal muscle interactions: summary of the proceedings of an AS-BMR workshop. J Bone Miner Res. 2013;28:1857–1865.

49. Fishman RB, Breedlove SM. Neonatal androgen maintains sexuallydimorphic muscles in the absence of innervation. Muscle Nerve.1988;11:553–560.

50. Raskin K, Marie-Luce C, Picot M, et al. Characterization of thespinal nucleus of the bulbocavernosus neuromuscular system inmale mice lacking androgen receptor in the nervous system. Endo-crinology. 2012;153:3376–3385.



enobosarm (gtx-024) modulates adult skeletal muscle mass independently of the androgen receptor in...

Documents