Induction of ATF4-Regulated Atrogenes Is Uncoupled from Muscle Atrophy during Disuse in Halofuginone-Treated Mice and in Hibernating Brown Bears

Activating transcription factor 4 (ATF4) is involved in muscle atrophy through the overexpression of some atrogenes. However, it also controls the transcription of genes involved in muscle homeostasis maintenance. Here, we explored the effect of ATF4 activation by the pharmacological molecule halofuginone during hindlimb suspension (HS)-induced muscle atrophy. Firstly, we reported that periodic activation of ATF4-regulated atrogenes (Gadd45a, Cdkn1a, and Eif4ebp1) by halofuginone was not associated with muscle atrophy in healthy mice. Secondly, halofuginone-treated mice even showed reduced atrophy during HS, although the induction of the ATF4 pathway was identical to that in untreated HS mice. We further showed that halofuginone inhibited transforming growth factor-β (TGF-β) signalling, while promoting bone morphogenetic protein (BMP) signalling in healthy mice and slightly preserved protein synthesis during HS. Finally, ATF4-regulated atrogenes were also induced in the atrophy-resistant muscles of hibernating brown bears, in which we previously also reported concurrent TGF-β inhibition and BMP activation. Overall, we show that ATF4-induced atrogenes can be uncoupled from muscle atrophy. In addition, our data also indicate that halofuginone can control the TGF-β/BMP balance towards muscle mass maintenance. Whether halofuginone-induced BMP signalling can counteract the effect of ATF4-induced atrogenes needs to be further investigated and may open a new avenue to fight muscle atrophy. Finally, our study opens the way for further studies to identify well-tolerated chemical compounds in humans that are able to fine-tune the TGF-β/BMP balance and could be used to preserve muscle mass during catabolic situations.


Introduction
Many unloading conditions (e.g., microgravity, bed rest, or physical inactivity) lead to a loss of muscle mass and strength. This muscle atrophy is associated with adverse health effects such as autonomy decline and increased morbidity and mortality [1][2][3]. Considering the lack of proven, easy-to-use therapeutic or preventive treatment, muscle atrophy remains a major public health issue (World Health Organisation data) [4].
The underlying molecular mechanisms of muscle atrophy involve the dysregulation of a complex network of intracellular pathways leading to an imbalance in protein turnover [5][6][7][8][9]. Activating transcription factor 4 (ATF4) is overexpressed in many conditions of muscle atrophy [10][11][12][13][14] and is, therefore, considered as an atrogene, i.e., one of the genes with expression at the mRNA level that is commonly altered during atrophy [14].
istration at the end of each week, we measured the mRNA levels for some ATF4 target genes, involved in muscle atrophy, i.e., Trib3, Cdkn1a, Gadd45a, and Eif4ebp1 ( Figure 1A-F). Except for Atf4, for which mRNA levels were elevated during the first 2 weeks of HF treatment, Trib3, Cdkn1a, and Gadd45a were all overexpressed in muscles after 2 weeks of HF treatment compared to H2O-treated mice. Of note, Eif4ebp1 mRNA levels increased in mouse muscles after 4 weeks of HF-treatment compared to H2O, with a noticeable trend after 3 weeks of treatment ( Figure 1F). We also investigated the regulation of other ATF4 target genes and showed the overexpression of Asns over the 4 weeks of HF treatment as well as a trend for Ddit3 and Ppp1r15a (Supplementary Figure S1). week and presented as individual values with mean bars ± SEM. Statistics are described in Section 4. * p adj < 0.05; ** p adj < 0.01; *** p adj < 0.001; **** p adj < 0.0001. These data underline an activation of the ATF4 transcriptional activity. However, despite the overexpression of the ATF4-regulated atrogenes, the mass of the gastrocnemius, soleus, tibialis anterior, and extensor digitorum longus (EDL) muscles remained unchanged during the 4 weeks of HF treatment ( Figure 1G, Supplementary Figure S1). Altogether, these data suggest that a long-term halofuginone administration induced ATF4-regulated atrogenes without leading to muscle atrophy.

Overexpression of ATF4-Regulated Atrogenes during Hindlimb Suspension Is Uncoupled from Muscle Atrophy in HF-Treated Mice
We then investigated the effect of the induction of the ATF4 pathway by HF treatment during the muscle atrophy induced by hindlimb suspension (HS). Briefly, mice received HF three times a week for 3 weeks and were then, 3 days after the last HF administration, hindlimb-suspended or not for 3 or 7 days (Figure 2A). Measurements were thus performed at least 6 days after the last HF administration. HF induces ATF4 activation through the phosphorylation of eIF2α [47]. We observed that HF treatment resulted in overall higher phosphorylated and total eIF2α protein levels compared to H 2 O-treated mice ( Figure 2B-D). In addition, HS led to an overall decrease in phosphorylated eIF2α protein levels compared to the controls (Ctrls) ( Figure 2B,C). In addition, levels of the mRNA encoding the phosphatase GADD34 (Ppp1r15a) were higher at 3 days of hindlimb suspension (HS3) compared to the Ctrls in both H 2 O-and HF-treated mice (Supplementary Figure S2). The expression of ATF4-regulated genes was not different between Ctrl-HF and Ctrl-H 2 O groups ( Figure 2 and Supplementary Figure S2). The mRNA levels of Atf4 and its target genes involved in muscle atrophy, i.e., Trib3, Cdkn1a, and Eif4ebp1, were higher during HS in both H 2 O-and HF-treated mice ( Figure 2E-H), while the mRNA levels of the ATF4-regulated atrogene Gadd45a remained unchanged ( Figure 2I). Of note, mRNA expression of other ATF4 target genes remained unchanged for Asns or slightly decreased upon HS for Ddit3. (Supplementary Figure S2). Altogether, Figures 1 and 2 shows that (i) HF administration induced ATF4-regulated atrogenes after 6 h, but no more after 6 days, indicating that this effect was rapid and transient, and (ii) HS induced overexpression of these atrogenes.
We next investigated the outcomes on skeletal muscle. Gastrocnemius muscle had atrophied only in H 2 O-treated mice after 3 days of HS (H 2 O-HS3) and in both H 2 O-and HF-treated mice after 7 days of HS. Surprisingly, the average of the fibre cross-sectional area (CSA) did not change in H 2 O-HS3 mice compared to the Ctrls but was lower after HS7 regardless of the treatment ( Figure 3B). We further analysed the distribution of gastrocnemius fibre CSA (Supplementary Figure S3A,B). We reported (i) a lower proportion of small fibres and (ii) a higher proportion of large fibres in HF-treated mice compared to the H 2 O group (Supplementary Figure S3A). This observation was restricted to fast-twitch fibres (i.e., 2X/2B) (Supplementary Figure S3B). Taken together, our data suggest that induction of ATF4-regulated atrogenes is not associated with muscle atrophy after 3 days of hindlimb suspension in HF-treated mice and, thus, that HF slightly preserves muscle mass during HS.  3 times a week for 3 weeks (WK) (black arrows) and were then subjected to hindlimb suspension for 3 (HS3, light grey bars) or 7 (HS7, white bars) days or kept unsuspended (Ctrl, dark grey). The dotted arrows represent the time when the muscles were collected. (B-D) Relative protein levels in gastrocnemius for phosphorylated and total eIF2α were measured by Western blotting, quantified, and normalised to the total protein content. Representative Western blots are shown. (E-I) Relative mRNA levels in gastrocnemius for Atf4, Trib3, Cdkn1a, Eif4ebp1, and Gadd45a were measured by RT-qPCR and were normalised using Tbp. Data are expressed as fold change vs. H 2 O-Ctrl and are presented as individual values normalised mean bars ± SEM. Statistics are described in Section 4. * p adj < 0.05; ** p adj < 0.01; *** p adj < 0.001; **** p adj < 0.0001. ured by RT-qPCR and were normalised using Tbp. Data are expressed as fold change vs. H2O-Ctrl and are presented as individual values normalised mean bars ± SEM. Statistics are described in Methods section. * padj < 0.05; ** padj < 0.01; *** padj < 0.001; **** padj < 0.0001. We next investigated the outcomes on skeletal muscle. Gastrocnemius muscle had atrophied only in H2O-treated mice after 3 days of HS (H2O-HS3) and in both H2O-and HF-treated mice after 7 days of HS. Surprisingly, the average of the fibre cross-sectional area (CSA) did not change in H2O-HS3 mice compared to the Ctrls but was lower after HS7 regardless of the treatment ( Figure 3B). We further analysed the distribution of gastrocnemius fibre CSA (Supplementary Figure S3A,B). We reported (i) a lower proportion of small fibres and (ii) a higher proportion of large fibres in HF-treated mice compared to the H2O group (Supplementary Figure S3A). This observation was restricted to fast-twitch fibres (i.e., 2X/2B) (Supplementary Figure S3B). Taken together, our data suggest that induction of ATF4-regulated atrogenes is not associated with muscle atrophy after 3 days of hindlimb suspension in HF-treated mice and, thus, that HF slightly preserves muscle mass during HS.

Halofuginone Treatment Inhibits TGF-β while Promoting BMP Signalling in Gastrocnemius Muscle
HF inhibits the TGF-β pathway [36,48]. We and others have recently reported that inhibition of the TGF- signalling is associated with the concomitant activation of BMP signalling [45,46]. Therefore, we investigated how HF treatment and subsequent HS treatment affected these pathways in skeletal muscle. The nuclear localisation of SMADs mirrors the upstream activation of the TGF-β or BMP pathway [49]. We, thus, measured protein levels for the transcription factors SMAD2/3 (TGF- signalling), SMAD1/5 (BMP signalling), and SMAD4 (TGF- and BMP signalling) in nuclear and cytosolic fractions (  Figure S4A-F). The ratio of nuclear SMAD2/3 to total SMAD2/3 was very low in HF-treated mice compared to H2O-treated mice and decreased

Halofuginone Treatment Inhibits TGF-β While Promoting BMP Signalling in Gastrocnemius Muscle
HF inhibits the TGF-β pathway [36,48]. We and others have recently reported that inhibition of the TGF-β signalling is associated with the concomitant activation of BMP signalling [45,46]. Therefore, we investigated how HF treatment and subsequent HS treatment affected these pathways in skeletal muscle. The nuclear localisation of SMADs mirrors the upstream activation of the TGF-β or BMP pathway [49]. We, thus, measured protein levels for the transcription factors SMAD2/3 (TGF-β signalling), SMAD1/5 (BMP signalling), and SMAD4 (TGF-β and BMP signalling) in nuclear and cytosolic fractions ( Figure 4A-D and Supplementary Figure S4A-F). The ratio of nuclear SMAD2/3 to total SMAD2/3 was very low in HF-treated mice compared to H 2 O-treated mice and decreased upon HS only in H 2 O-treated mice ( Figure 4A,B). Consistently, the overall mRNA levels of several collagens, which are well-known target genes of TGF-β signalling, decreased upon HS (Supplementary Figure S5). Moreover, the ratio of nuclear SMAD1/5 to total SMAD1/5 was higher in HF-Ctrl mice than in H 2 O-Ctrl mice. This ratio was reduced at HS7 compared to the Ctrl in HF-treated mice, while it was increased in H 2 O-treated mice ( Figure 4A,C). Finally, the ratio of nuclear SMAD4 to total SMAD4 was overall lower in HFvs. H 2 O-treated mice ( Figure 4A-D), with a decrease in HF-treated mice at HS7 compared to the Ctrl.
TGF-β catabolic action involves the inhibition of protein synthesis [40,50,51], whereas the anabolic action of BMP involves its promotion [44]. The overall protein-synthesis rates measured by puromycin incorporation were reduced during hindlimb suspension. However, this decrease was only significant in H 2 O-HS7 compared to H 2 O-Ctrl mice ( Figure 5A,B). TGF-β signalling acts also as a negative regulator of muscle mass through the induction of the atrogenes TRIM63/MurF1 and FBXO32/Atrogin-1 [38,39], while BMP signalling acts as a positive regulator with the transcriptional repression of the atrogene FBXO30/Musa1 [44,45]. Trim63 and Fbxo32 mRNA levels were upregulated only at HS3 in both H 2 O and HF-treated mice, while mRNA levels for the atrogene Fbxo30 remained unchanged ( Figure 5C-E). Our data suggest that while HF inhibits TGF-β signalling, it also promotes BMP signalling in the control gastrocnemius muscles. We also showed that HF partially attenuates the drop in protein synthesis during hindlimb suspension.

ATF4-Regulated Atrogenes Are Overexpressed in Atrophy-Resistant Hibernating Brown Bear Muscle
Our data strongly suggest that the induction of ATF4 signalling is not always associated with muscle atrophy, either in basal conditions or in HS-induced muscle atrophy. We took advantage of a natural model, i.e., the hibernating brown bear, which experiences only a moderate loss of muscle protein content while remaining completely inactive for up to 6 months [52][53][54]. As with HF treatment, we recently reported a concomitant TGF-β pathway inhibition and BMP pathway activation in hibernating brown bear muscle [46]. We, thus, explored whether ATF4-regulated atrogenes may also be induced in this model. Interestingly, as shown in Figure 6, Atf4 was upregulated in hibernating brown bear muscle compared to the active counterpart. In addition, the two main ATF4-regulated atrogenes (Gadd45a and Cdkn1a) and Trib3 were also induced in hibernating brown bear muscle. Other ATF4 target genes were either down-(Eif4ebp1, Ppp1r15a, and Asns) or upregulated (Ddit3). These data show that ATF4-regulated atrogenes are induced in hibernating brown bear muscle, even if they are resistant to atrophy. Our data strongly suggest that the induction of ATF4 signalling is not always associated with muscle atrophy, either in basal conditions or in HS-induced muscle atrophy. We took advantage of a natural model, i.e., the hibernating brown bear, which experiences only a moderate loss of muscle protein content while remaining completely inactive for up to 6 months [52][53][54]. As with HF treatment, we recently reported a concomitant TGF- pathway inhibition and BMP pathway activation in hibernating brown bear muscle [46].
We, thus, explored whether ATF4-regulated atrogenes may also be induced in this model. Interestingly, as shown in Figure 6, Atf4 was upregulated in hibernating brown bear muscle compared to the active counterpart. In addition, the two main ATF4-regulated atrogenes (Gadd45a and Cdkn1a) and Trib3 were also induced in hibernating brown bear muscle. Other ATF4 target genes were either down-(Eif4ebp1, Ppp1r15a, and Asns) or upregulated (Ddit3). These data show that ATF4-regulated atrogenes are induced in hibernating brown bear muscle, even if they are resistant to atrophy. Figure 6. ATF4-regulated atrogenes are induced in atrophy-resistant hibernating brown bear muscles. Gene expression levels for ATF4, GADD45A, CDKN1A, TRIB3, EIF4EBP1, PPP1R15A, ASNS, and DDIT3 in vastus lateralis muscle of active and hibernating brown bears (n = 6 bears/season, the same individuals were sampled and analysed in summer and winter, log2FC winter/summer). Data are presented as individual values as log2FC with mean bars ± lfcSE (log2 fold change standard error). Statistics are described in [46]. * padj < 0.05; ** padj < 0.01; **** padj < 0.0001. FC: fold change; W: winter (hibernating season); S: summer (active season)
We first observed that the overexpression of the ATF4-regulated atrogenes Trib3, Cdkn1a Gadd45a, and Eif4ebp1, as well as Atf4 itself, induced by halofuginone treatment for up to 4 weeks, did not coincide with atrophy in all hindlimb muscles, including gastrocnemius. Subsequently, we then reported that pre-treatment with halofuginone mitigated the atrophy of the gastrocnemius muscle during hindlimb suspension. These positive effects of halofuginone treatment are consistent with reports showing that this dose (i.e., 0.25 μg/g) and frequency of administration (i) were very well-tolerated in mice for up to 3 months and (ii) improved muscle-cell survival, promoted membrane repair, and improved muscle performances in models of muscular dystrophies [31,33,[58][59][60]. However, none of these studies explored whether the potential effect of HF would involve the ATF4 pathway. Here, we showed that induction of ATF4-regulated atrogenes was uncoupled from muscle atrophy during hindlimb suspension. Indeed, although ATF4-regulated atrogenes were overexpressed during hindlimb suspension, halofuginone-treated mice displayed a partial preservation of gastrocnemius muscle mass and CSA. In addition, we took advantage of a natural model of resistance to muscle atrophy to examine the expression of ATF4-regulated atrogenes. The brown bear remains completely inactive during hibernation for up to 6 months but, surprisingly, is not sensitive to muscle atrophy [52][53][54], which provides an interesting model for finding new molecular mechanisms to fight muscle atrophy in humans. We showed that the atrogenes CDKN1A, GADD45A, ATF4 itself, and TRIB3 were upregulated in atrophy-resistant hibernating brown bear muscle compared to active bear muscle. Of note, ATF4 is mainly regulated at the translational level [15]. However, in most of the previous studies on the topic, the authors only measured Atf4 mRNA expression and expression of its target genes as evidence of its activity Figure 7. Graphical abstract. The red and green lines represent catabolic and anabolic effects, respectively. Dotted lines represent hypothetical connections. The arrows/T bars above the ATF4 atrogenes, SMAD2/3, and SMAD1/5 boxes represent the induction/inhibition by halofuginone or by an as-yet-unknown mechanism in mouse or bear muscle, respectively. Created with BioRender.com.
We first observed that the overexpression of the ATF4-regulated atrogenes Trib3, Cdkn1a Gadd45a, and Eif4ebp1, as well as Atf4 itself, induced by halofuginone treatment for up to 4 weeks, did not coincide with atrophy in all hindlimb muscles, including gastrocnemius. Subsequently, we then reported that pre-treatment with halofuginone mitigated the atrophy of the gastrocnemius muscle during hindlimb suspension. These positive effects of halofuginone treatment are consistent with reports showing that this dose (i.e., 0.25 µg/g) and frequency of administration (i) were very well-tolerated in mice for up to 3 months and (ii) improved muscle-cell survival, promoted membrane repair, and improved muscle performances in models of muscular dystrophies [31,33,[58][59][60]. However, none of these studies explored whether the potential effect of HF would involve the ATF4 pathway. Here, we showed that induction of ATF4-regulated atrogenes was uncoupled from muscle atrophy during hindlimb suspension. Indeed, although ATF4-regulated atrogenes were overexpressed during hindlimb suspension, halofuginone-treated mice displayed a partial preservation of gastrocnemius muscle mass and CSA. In addition, we took advantage of a natural model of resistance to muscle atrophy to examine the expression of ATF4regulated atrogenes. The brown bear remains completely inactive during hibernation for up to 6 months but, surprisingly, is not sensitive to muscle atrophy [52][53][54], which provides an interesting model for finding new molecular mechanisms to fight muscle atrophy in humans. We showed that the atrogenes CDKN1A, GADD45A, ATF4 itself, and TRIB3 were upregulated in atrophy-resistant hibernating brown bear muscle compared to active bear muscle. Of note, ATF4 is mainly regulated at the translational level [15]. However, in most of the previous studies on the topic, the authors only measured Atf4 mRNA expression and expression of its target genes as evidence of its activity in skeletal muscle. Indeed, endogenous ATF4 protein cannot be reliably detected in skeletal muscle, presumably due to its low abundance, very short half-life, and lack of a high quality antibody [11][12][13]17,18,61]. Altogether, these data strongly suggest that the induction of the ATF4 pathway can be dissociated from muscle atrophy. ATF4 target genes are highly dependent on the type and duration of stress stimuli [21,62], and the ability to restore homeostasis may be overwhelmed when the stress is too severe or sustained, resulting in cell death through the transcription of pro-apoptotic genes [63][64][65][66]. Therefore, to avoid chronic and acute activation, halofuginone was administrated periodically to activate the eIF2α-ATF4 pathway in mice. In our conditions, the ATF4 transcriptional program may, thus, (i) differ from the transcriptional program induced by a severe and sustained activation and (ii) include genes that might counteract the effect of ATF4-induced atrogenes.
Halofuginone is well-described to also target the TGF-β pathway [35,36]. The nuclear translocation of the TGF-β transcription factors SMAD2/3 requires the formation of a complex with SMAD4 [43]. Halofuginone-induced eIF2α phosphorylation has been reported to inhibit the nuclear translocation of this complex in intestinal porcine enterocyte cells in vitro [67]. Consistently, we reported here a concomitant overall (i) increase in phosphorylated eIF2α protein levels and (ii) reduction in SMAD2/3 and SMAD4 nuclear protein levels in HF-treated mice. Thus, this highlights in skeletal muscle, for the first time, the possible role of HF-induced eIF2α phosphorylation in TGF-β inhibition. Although, the concomitant collagen downregulation and decrease in SMAD2/3 nuclear protein levels during hindlimb suspension in H 2 O-treated mice suggest a decrease in TGF-β signalling, we cannot exclude that these events are disconnected. Indeed, the SMAD2/3 nuclear protein levels are consistently low in HF-treated mice and, thus, cannot explain the decreased collagen expression during hindlimb suspension. Much remains to be clarified about the mechanisms of action of HF. Indeed, HF is used for its antifibrotic properties mediated by TGF-β inhibition in situations already characterised by fibrosis [48]. This is, however, not the case in our study. In addition, the inhibition of TGF-β signalling in muscles of HF-treated mice could have led to transcriptional changes that remain to be explored. It is possible that TGF-β signalling is induced later during hindlimb suspension. In fact, the TGF-β signalling pathway has previously been reported to be either unchanged in skeletal muscle after 1-3 days or induced after 7-10 days of unloading [46,68,69]. We also reported an upregulation of Trim63 and Fbxo32 during hindlimb suspension. Although these atrogenes are targets of the TGF-β signalling activation, they are also regulated by other signalling pathways [5].
We and others reported that the balance between TGF-β and BMP signalling seems crucial for muscle-mass maintenance during catabolic situations [39,[44][45][46]70]. Indeed, using the hibernating bear model, we recently reported that TGF-β signalling, i.e., a negative regulator of muscle mass, was downregulated at the transcriptomic level in muscles that are resistant to atrophy, while BMP signalling, i.e., a positive regulator of muscle mass, was maintained [46]. Previous data suggested that TGF-β inhibition would release SMAD4, i.e., the common actor in TGF-β and BMP signalling, which could, thus, be recruited to BMP signalling and promote hypertrophy and/or counteract atrophy [45]. Here, we reported an increase in the SMAD1/5 nuclear protein levels in the halofuginonetreated control mice, suggesting there was concomitant BMP signalling activation and TGF-β inhibition. In addition, BMP activation was reported to increase during denervation, intensive care disuse, and amyotrophic lateral sclerosis and was described as essential to counteract excessive muscle wasting [44,45]. In agreement, we reported here that BMP transcription factors SMAD1/5 accumulated in the nucleus in H 2 O-treated mice but, surprisingly, declined in HF-treated mice after 7 days of hindlimb suspension. Whether the higher basal pools of nuclear SMAD1/5 and their maintenance after 3 days of hindlimb suspension in HF-treated mice contributed to attenuating skeletal muscle atrophy during hindlimb suspension remains to be explored. Of note, BMP signalling has been reported to promote protein synthesis in muscle [44]. Maintenance of the BMP pathway after 3 days of hindlimb suspension may have contributed to the partial preservation of protein synthesis and muscle mass in HF-treated mice. Nuclear translocation of SMAD1/5 represses the transcription of FBXO30/Musa1 [44,45]. However, we did not observe any change in Fbxo30 expression. Mechanisms by which BMP signalling controls muscle mass are still very poorly understood and will require further studies, particularly with a comprehensive characterisation of the BMP target genes in skeletal muscle. We can also speculate that HF-induced BMP activation has helped to limit muscle atrophy induced by ATF4-regulated atrogenes (Figure 7).
In conclusion, halofuginone treatment reproduced the muscle features of hibernating bears in gastrocnemius mice muscles with (i) the activation of ATF4-regulated atrogenes and (ii) the concurrent inhibition of TGF-β signalling and promotion of BMP signalling, without resulting in muscle atrophy (Figure 7). These characteristics were associated with mitigated muscle atrophy during physical inactivity. To date, clinical trials have all attempted to inhibit the TGF-β pathway, mostly with side effects or minimal efficiency [71].
Our study suggests halofuginone, as a well-tolerated chemical compound, already used in human clinical trials [36], was able to tune the TGF-β/BMP balance in vivo and likely sustained muscle mass. Moreover, our data open new ways to further decipher by which precise mechanisms ATF4 induces atrophy and how BMP activation can interfere.

Materials and Methods
Ethics, animals housing, and experimental design. All experiments were conducted with the approval of the regional ethics committee (agreement no. D6334515) following the European Directive 2010/63/EU on the protection of vertebrate animals used for experimental and scientific purposes. This study was performed with 12-week-old C57BL6/J male mice (25-30 g), purchased from Janvier Labs (Le Genest-Saint-Isle, France). They were housed individually upon arrival for 10 days of acclimatisation in a controlled room (22 ± 2 • C, 60 ± 5% humidity, 12 h light/dark cycle, and light period starting at 8 h), fed ad libitum a standard rodent diet (pellets A03 from Safe, Augy, France), and given free access to water. Two distinct animal experiments were performed. To evaluate the effects of a periodic halofuginone (HF) (#32481, Sigma, Saint-Quentin-Fallavier, France) administration, we performed a first protocol where mice received either HF (0.25 µg/g) or water (H 2 O) by gavage 3 times a week for 1 to 4 weeks (n = 6 animals per group). This dose was reported as well tolerated over longer periods [31,33,60]. Gastrocnemius muscle was sampled 6 h after the last HF/H 2 O administration at the end of each week. Subsequently, we performed a second protocol to test whether HF administration before hindlimb unloading had a positive effect on muscle mass and function. For that purpose, we performed two separate animal experiments. In each experiment, mice received either HF (0.25 µg/g) or H 2 O by gavage 3 times a week for 3 weeks and were afterwards subjected either to hindlimb unloading through tail suspension (HS) or kept unsuspended (Ctrl) for 3 or 7 days, as previously described [46] (n = 8-19 animals per group). We did not record any difference between Ctrl mice at 3 or 7 days for all the measurements reported in this manuscript. We, therefore, pooled the two groups of Ctrl mice for further analysis and data representation. Food intake and body weight were recorded throughout the different protocols. Unloading in control mice resulted in only a small body weight loss (<10%) that occurred within the first 3 days concomitantly with a decrease in food intake, whereas HF treatment did not modify food intake or body weight ( see Supplementary Figures S1 and S2).
Tissue collection. At the end of the experiments, mice were euthanised by cervical dislocation. The soleus, gastrocnemius, tibialis anterior, and extensor digitorum longus (EDL) muscles were carefully collected and weighed prior to immediate freezing in liquid nitrogen and storage at −80 • C until analyses.
Measurement of protein synthesis in gastrocnemius. At the end of protocol 2, mice received an intraperitoneal injection of 0.040 µmol/g puromycin (#P8833, Sigma, Saint-Quentin-Fallavier, France) dissolved in 100 µL of a saline solution before euthanasia, as described previously [72]. At exactly 30 min post-puromycin injection, gastrocnemius muscle was dissected and frozen in liquid nitrogen for Western blot analysis, as follows.
Histology and morphometric measurements. A part of the gastrocnemius muscle was collected at the end of protocol 2 and frozen in isopentane chilled with liquid nitrogen and stored at −80 • C until use. Serial muscle cross-sections (10 µm thick) were obtained using a cryostat (HM500M Microm International, Fisher Scientific, Illkirch, France) at −20 • C. Cross-sections were labelled with anti-laminin-α1 (L9393 Sigma, Saint-Quentin-Fallavier, France) to outline the fibre cross-sectional area (CSA) and BFF3 antibody (#AB_2266724, DSHB, Iowa City, IA, USA) to determine myosin heavy chain type 2B fibre. Both were subsequently hybridised with a corresponding secondary antibody conjugated to Alexa-Fluor (Invitrogen, Cergy-Pontoise, France). Image acquisitions were performed with a high-resolution ORCA-Flash4.0 LT+ Digital CMOS camera coupled to a IX-73 microscope (Olympus, Münster, Germany) and Cell-Sens dimension software (Olympus Soft Imaging Solutions, Münster, Germany). The CSA was determined for 1000-1500 fibres per animal, using ImageJ software 1.53f51 (http://rsb.info.nih.gov/ij/, accessed on 3 April 2018).
Protein isolation. Gastrocnemius muscles were pulverised in liquid nitrogen.
RT-qPCR. Total RNA from gastrocnemius muscle samples was extracted with Macherey-Nagel™ NucleoSpin™ 96 RNA Kit and KingFisher™ Duo Prime Purification System, in accordance with the instructions of the manufacturer (Macherey-Nagel, Hoerdt Cedex, France). RNA was quantified by measuring the absorbance at 260 nm on a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Then, 500 ng of RNA were treated with DNase I (Invitrogen, Cergy-Pontoise, France) prior to reverse transcription using random primers and SuperScript II (Invitrogen, Cergy-Pontoise, France), in accordance with the instructions of the manufacturer. Real-time PCR was carried out using the CFX96 Real-Time PCR detection system (Biorad, Marnes-la-Coquette, France). Primer sequences are provided in Supplementary Table S1. PCR reactions were performed using the IQ SYBR Green Supermix (Biorad, Marnes-la-Coquette, France), in accordance with the instructions of the manufacturer. The comparative threshold cycle (2∆∆CT) method was used to compare the relative mRNA expression between each group, using TBP (TATA binding protein) as a reference gene for muscle. The relative mRNA abundance was arbitrarily set to 1 for the H 2 O-Ctrl group.
Statistics. All data are means ± SEM and were analysed for normality of residuals using the Shapiro-Wilk test. No set of data was transformed for non-normality distribution. For protocol 1 (n = 6/group), we performed a multiple Welch t-test within each week. For protocol 2 (n = 8-19/group), we performed a two-way ANOVA with the factors "Hindlimb suspension" and "Halofuginone" and corrected the data for multiple comparisons using Tukey's test. These analyses were performed using Prism 9 (GraphPad Prism 9, San Diego, CA, USA).
Transcriptomic Data. We used transcriptomic data from already published studies [46]. The transcriptomic bear data supporting Figure 6 of this study are openly available in the GEO repository database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi, reference no. GSE144856, accessed on 1 September 2021). To identify the differentially expressed genes (DEGs) from this list, we selected a winter/summer (FC) > 1.0 with an adjusted p-value < 0.05 as cut-off for the up-regulated genes.