Intracellular inactivation of thyroid hormone is a survival mechanism for muscle stem cell proliferation and lineage progression.

Precise control of the thyroid hormone (T3)-dependent transcriptional program is required by multiple cell systems, including muscle stem cells. Deciphering how this is achieved and how the T3 signal is controlled in stem cell niches is essentially unknown. We report that in response to proliferative stimuli such as acute skeletal muscle injury, type 3 deiodinase (D3), the thyroid hormone-inactivating enzyme, is induced in satellite cells where it reduces intracellular thyroid signaling. Satellite cell-specific genetic ablation of dio3 severely impairs skeletal muscle regeneration. This impairment is due to massive satellite cell apoptosis caused by exposure of activated satellite cells to the circulating TH. The execution of this proapoptotic program requires an intact FoxO3/MyoD axis, both genes positively regulated by intracellular TH. Thus, D3 is dynamically exploited in vivo to chronically attenuate TH signaling under basal conditions while also being available to acutely increase gene programs required for satellite cell lineage progression.

Introduction

Muscle regeneration is a multistep process that includes myofiber degradation, regeneration, and remodeling (Ten Broek et al., 2010). The repair process is characterized by the activation of a primary myogenic stem cell population referred to as “satellite cells,” which give rise to activated proliferating myoblasts or myoblast precursor cells (mpcs), followed by cell differentiation and fusion into regenerated myofibers. Satellite cells, that are normally quiescent, can be activated to proliferate and generate committed progeny in response to a variety of stimuli, including degenerative muscle diseases (Brack and Rando, 2012, Dhawan and Rando, 2005, Rudnicki et al., 2008).

The active thyroid hormone (TH), T3, derives in large part from the monodeiodination of the prohormone thyroxine (T4) by one of two iodothyronine selenodeiodinases (D1 or D2). Conversely, TH signaling terminates consequent to inactivation of T3 and T4 induced by removal of a tyrosyl ring iodine by type 3 deiodinase (D3). D3 converts the active hormone T3 to inactive metabolites thereby terminating TH action within cells. This provides a mechanism by which TH action can be terminated in a tissue-specific chronologically programmed fashion (Bianco et al., 2002). The high expression of D3 in fetal compartments and the growth retardation and partial neonatal mortality of D3-null mice (Hernandez et al., 2006) confirm that D3 exerts a critical function during development.

Normal TH levels are required for efficient muscle homeostasis, function, and regeneration (McIntosh et al., 1994, Simonides and van Hardeveld, 2008). Muscle is a major target of TH action. Indeed, a broad set of genes are positively or negatively regulated at the transcriptional level by TH (Salvatore et al., 2014, Simonides and van Hardeveld, 2008). One of the genes transcriptionally stimulated by T3 is MyoD (Muscat et al., 1995), which is a master regulator of the myogenic developmental and regeneration program. While it is well known that muscle function is altered in patients with thyrotoxicosis or hypothyroidism, it has also been shown that TH excess impairs the regeneration process in the mdx mouse (Anderson et al., 1994). The pathophysiological mechanism underlying this effect is unknown.

There are two sources of T3 in muscle tissue; one is the fraction that enters the cells directly from the plasma, the second is locally produced from T4-to-T3 conversion via D2 action (Dentice et al., 2010, Marsili et al., 2011). The factors involved in the modulation of TH availability at cell level are unknown. Similarly, little is known about how the balance between the T3-activating and -inactivating deiodinases in muscle and in muscle progenitor cells is determined. Clarification of these issues would be a significant advance in the understanding of the cellular pathways governing the progression of muscle stem cell lineage.

The aim of our study was to dissect the role of the intracellular TH metabolism and signaling in muscle progenitor cells. We identified D3 in satellite cells and mpcs, and found that it is induced upon stem cell activation early after muscle injury. This event was associated with the expansion of the satellite cell population that occurs after muscle injury. Despite normal plasma T3 concentrations, selective depletion of D3 in the satellite cell compartment resulted in severe cell apoptosis thereby disrupting the normal pattern of tissue response to acute injury and causing a marked delay in muscle regeneration. Thus, we demonstrate that D3 and modulation of local TH metabolism represent a survival mechanism during the progression of the muscle stem cell lineage.

Results

Upregulation of D3 in Proliferating Satellite Cells during Muscle Regeneration

To assess whether D3 is expressed in satellite cells, we measured its expression in FACS-sorted cells from Tg:Pax7-nGFP mice (Rocheteau et al., 2012) and found elevated D3 levels, which declined as differentiation proceeded (Figures 1A and 1B), a pattern observed also in C2C12 as well as in primary cultures of skeletal muscle enriched in satellite cells (pp6 cells; Qu et al., 1998) (Figure S1A available online). D3 levels increased rapidly during muscle development, and peaked in the second postnatal week, after which they progressively declined to reach very low levels in the adult (Figure S1B). D3 expression, detectable in isolated Pax7-positive cells, increased in dividing MyoD1-positive cells of muscle fibers (Figures 1C and S1C). Interestingly, D3 activity in limb muscles was significantly higher in the dystrophic mdx mouse than in controls (Figure 1D), which is consistent with the increased proliferation of satellite cells in the dystrophic context.

Figure 1

Type 3 D3 Is Highly Expressed in Proliferating Muscle Precursor Cells and in Pax7-Positive Cells in Skeletal Muscle

(A) Type 3 deiodinase (D3) protein (inset top) and mRNA (inset bottom) levels in mouse GFP-sorted (FACS) satellite cells isolated from Tg:Pax7nGFP+ mice grown in culture under proliferative (PROL) or differentiating (DIFF) conditions.

(B) Immunofluorescence staining (IF) for D3, EdU, MHC, DAPI, and merged images in PROL and DIFF FACS-sorted Pax7-GFP-positive cells. Scale bar, 50 μm.

(C) D3 (red)- and Pax7 (green)-positive cells detected by IF in fibers cultured for 0, 24, and 72 hr (scale bar, 50 μm).

(D) D3 enzymatic activity in different muscles from control (WT) and mdx mice. Data are expressed as mean ± SEM of at least three independent experiments.

(E) D3 protein (top) and mRNA (bottom) measured by western blot and RT-PCR in uninjured muscles (0) and injured muscles (4–15) at different stages of muscle regeneration induced by cardiotoxin (CTX) injection into the tibialis anterior (TA) muscle.

(F) D3 enzymatic activity in TA muscles during muscle regeneration induced by CTX injection.

For (A) and (D)–(F) data are expressed as mean ± SEM of at least three independent experiments.

We examined D3 expression during regeneration after cardiotoxin (CTX) injection (Chargé and Rudnicki, 2004). D3 mRNA, protein, and enzymatic activity were significantly increased at early stages after CTX injection (4–8 days) in the tibialis anterior (TA) muscle. Then D3 declined, reaching nearly normal adult levels by day 15 (Figures 1E and 1F). Coimmunofluorescence analysis showed D3 staining in the Pax7-positive cells as well as in the newly regenerating fibers and in F4/80+ macrophages infiltrating the injury site (Figures 2A and 2B). D3 showed a dynamic cell-type specific expression, i.e., a robust expression at day 7 in the satellite cell compartment and declined thereafter (Figures 2A and 2B).

Figure 2

D3 Is Induced in the Early Phases of Muscle Regeneration

(A) Left: immunofluorescence of D3 (red), MHC, F4/80, and Pax7 (green), and merged images in tibialis anterior (TA) muscles at days 0, 7, and 15 of muscle regeneration. Right: immunofluorescence of D3 (red), Pax7, F4/80, and Desmin (green), DAPI (blue), and merged images of TA muscles at day 7 of muscle regeneration.

(B) Quantification of percentage of D3-positive cells in muscle section at 7 days of regeneration (top) and in specific cell populations (bottom) after damage.

(C) D3 mRNA levels in subpopulations of quiescent (adult) and activated Pax7-nGFPHi, Pax7-nGFPmid, and Pax7-nGFPLo cells isolated as described under Experimental Procedures. Activated satellite cells were obtained 5 days after TA injury. Normalized copies of D3 in quiescent, Pax7Hi cells were arbitrarily set as 1. Data are expressed as mean ± SEM of at least three independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

(D) TH-responsive gene expression measured by RT-PCR in the same samples as in (C). Data are expressed as mean ± SEM of at least three independent experiments.

TH Signaling in Satellite Cells and D3 Expression

To determine whether D3 might serve as a novel marker of a specific subset of Pax7-positive cells, we isolated satellite cells and fractionated them according to nGFP intensity by flow cytometry. We analyzed FACS-isolated Pax7-nGFPHi (top 20%), Pax7-nGFPMid (intermediate 60%), and Pax7-nGFPLo (bottom 20%) from CTX-injured or control-uninjured muscles. Interestingly, within the Pax7-nGFP-expressing cells, D3 expression was higher in Pax7-nGFPLo cells, which have the highest commitment to progress toward a myogenic lineage (Figure 2C) (Mourikis et al., 2012). Within the D3-positive Pax7-nGFPLo fraction, D3 expression inversely correlated with TH signaling. This is illustrated by the expression profile of well-characterized positively (SerCa2, FoxO3, PGC-1α, Troponin 2) and negatively (PGC-1β and Troponin 1) (Figures 2D and S2) regulated TH target genes, which points to a relatively “hypothyroid state” in the Pax7-nGFPLo/D3high-expressing cells (Figure 2D). The pattern of D3 expression and the expression profile of the TH-responsive gene in GFP-sorted cells provide functional evidence that intracellular TH concentrations are tightly regulated in satellite cells during lineage progression.

Effective Satellite Cell D3 Deletion in a Conditional dio3-Null Mouse

We next asked what role D3 plays in proliferating myoblast precursors. To this aim, we generated a satellite cell-specific conditional dio3 knockout mouse (see Experimental Procedures). The homozygous dio3fl/fl mice were viable and fertile, with no alterations in D3 protein, body mass, fertility, or serum TH concentrations (Figures S1 and S3; data not shown). Adeno-Cre infection of primary myoblasts resulted in effective deletion of the selenocysteine insertion sequence (SECIS) element and inactivation of D3 (see Experimental Procedures), without affecting D3 mRNA levels (Figure S1D) in agreement with the genetic strategy used.

We crossed the dio3 floxed mouse with a Tg:Pax7-CreERT2 mouse (Mourikis et al., 2012) in order to specifically delete D3 in the satellite cells. We subjected 2-month-old dio3fl/fl; Tg:Pax7-CreERT2 mice (cD3KO) to tamoxifen (TAM) treatment followed by CTX injury (Figure 3A). After TAM treatment, D3 in satellite cells was markedly and specifically reduced as demonstrated by immunofluorescence staining (Figures 3B and S3). As expected, D3 protein levels were significantly reduced, but not abolished, in muscle by TAM treatment, in agreement with the expression of D3 in Pax7-negative cells, e.g., in infiltrating macrophages (Figures 2B and 3B). Thus, the cD3KO mice sustained specific postnatal D3 depletion specifically in Pax7-positive cells.

Figure 3

Conditional Pax7-Specific D3 Knockout Mice

(A) Diagram of the experimental design used for CTX and tamoxifen (TAM) injection.

(B) IF of D3 and Pax7 in the TA of control (WT) and cD3KO mice treated with TAM as indicated 7 days after injury. Scale bar, 25 μm.

(C) Percentage of Pax7- and TUNEL-positive cells evaluated by IF in TA 7 days after injury. Data are expressed as mean ± SEM of at least three independent experiments. ∗p < 0.05.

(D) IF of Pax7- (red) and TUNEL-positive (green) cells in CTX-injured TA of euthyroid control (CTR EU), euthyroid cD3KO (cD3KO EU), and hypothyroid cD3KO (cD3KO HYPO) mice treated with TAM as indicated. Scale bar, 50 μm.

(E) Quantification of double-Pax7/TUNEL-positive cells from the experiment in (D). Data are expressed as mean ± SEM of at least three independent experiments. ∗p < 0.05.

(F) Percentage of Pax7-positive cells in TA muscles as in (E).

D3 Depletion Causes Massive Apoptosis of Pax7-Expressing Cells, Impairment of the Initial Wave of Muscle Regeneration, and Failure of the Repair Process

After CTX injection (4 days), the pattern of regeneration in the TA was similar in the control and cD3KO mice. However, following D3 depletion at day 7 post-CTX, we observed a significant reduction in the number of Pax7-positive cells (Figures 3A–3C). This reduction was due to massive apoptosis, as measured by TUNEL assay and PARP cleavage in regenerating muscles (Figures 3C–3F, S3D, and S3E). Examination of muscle sections at later times showed a striking difference between the cD3KO and wild-type (WT) animals, a difference particularly evident at day 14 postinjury and thereafter (Figure 4A). Indeed, while regeneration was active at this time in control muscle, with most of the myonuclei in the central position of myofibers and a sharp increase in the myogenic markers MyoD, Myogenin, and MHC, these markers were greatly decreased in cD3KO muscles (Figures 4B and 4C). There was also a dramatic reduction in the number of regenerating fibers in the mutant animals (Figure 4D), which is consistent with a block in the initial wave of muscle regeneration. At day 40, when regeneration was almost complete and the number of centrally located nuclei was similar in the cD3KO and WT mice (Figure 4D), regeneration was persistently aberrant and incomplete in the mutant tissues (Figure 4A). In addition, the cross-sectional area was significantly reduced in cD3KO mice versus control mice (Figure 4E). Systemic hypothyroidism significantly rescued apoptosis in cD3KO mice (Figures 3D–3F), indicating that apoptosis was due to an excessive intracellular TH concentration. Overall, these results indicate that D3 depletion in satellite cells and the consequent cell apoptosis significantly impair the regeneration process.

Figure 4

D3 Depletion Impairs Muscle Regeneration

(A) Histological analysis (H&E staining) of the TA of control and cD3KO mice 4, 14, and 40 days after CTX injection.

(B) Expression of myogenic markers MyoD, Myogenin, and MHC measured by IF of the injured TA muscles at day 14 as in (A).

(C) Confirmation of decreased expression of MyoD, Myogenin, and nMyHC confirmed by RT-PCR (top) and western blot analysis (bottom) from the same muscles as in (B).

(D) Percentage of centrally located nuclei evaluated in H&E-stained TA sections from WT and cD3KO mice 4, 14, and 40 days after CTX injection.

(E) Cross-sectional area of TA sections from WT and cD3KO mice 14 days after CTX injection. Results are representative of at least three independent experiments. Scale bar, 100 μm.

For (C) and (D) data are expressed as mean ± SEM of at least three independent experiments.

D3 Depletion, as well as Excessive TH, Causes Caspase-3-Dependent Apoptosis in Activated Satellite Cells

In dio3fl/fl myoblasts, apoptosis followed D3 depletion or TH treatment (Figures 5A, 5B, S4A, and S4B). This did not occur in Adeno-Cre-infected cells from WT mice or in the fibroblast-like primary cell population from dio3fl/fl mice (Figures S4C–S4F). Since D3 serves only to inactivate TH, we speculated that apoptosis would not occur in the absence of TH in the serum, which was in fact the case (Figure 5A). Furthermore, apoptosis was dependent on, and related to, cell proliferation. When differentiated dio3fl/fl cells were infected with the Adeno-Cre virus or treated with supraphysiological TH concentrations, no apoptosis was detected, nor did apoptosis occur in cells arrested in S phase before TH treatment (Figures S4G and S4H). These results indicate that proliferating, but not cell cycle-arrested or differentiated, mpcs, undergo apoptosis following exposure to elevated TH concentrations. Accordingly, in cultured fibers, TH treatment significantly induced apoptosis in proliferating myoblasts (Figures S4I and S4L).

Figure 5

D3 Depletion Causes Massive Apoptosis of Activated Myogenic Precursor Cells

(A) Proliferating primary myogenic precursor cells from cD3KO mice infected with CRE-expressing adenovirus (Ad-CRE) or control GFP (Ad-GFP). Apoptosis measured by western blot analysis of PARP cleavage (right panels).

(B) FACS-sorted Pax7-GFP-positive cells exposed to TH for the indicated times and apoptosis assessed by measuring Annexin-V FACS profile.

(C) Immunostaining of Pax7- and TUNEL-positive cells in TA of untreated control mice and TH-treated mice (n = 3 CTR, n = 3 TH; experiments were run three times).

(D) Pax7-positive and TUNEL-positive cells from the same experiment as in (C). Data are expressed as mean ± SEM of at least three independent experiments.

Finally, we tested whether apoptosis occurred also in vivo after TH treatment. In CTX-injured TH-treated mice, impairment of the regeneration process (Figures S5A–S5C) was associated with a significant reduction in Pax7-positive cells and a corresponding increase in apoptosis (Figures 5C and 5D). Interestingly, after TH treatment, the expression levels of TH-regulated myogenic markers were normally increased in uninjured muscle, but not in CTX muscle (Figure S5D). This finding is in agreement with the concept that systemic thyrotoxicosis causes massive apoptosis of activated myogenic precursor cells. These results demonstrate that D3 expression protects activated satellite cells from a physiological, but spatiotemporally excessive, intracellular TH concentration.

FoxO3 Is a Novel TH-Responsive Gene in Myoblasts

To identify the molecular mechanisms underlying D3-dependent apoptosis, we evaluated the effects of D3 depletion on the expression of various transcription and myogenic factors. D3 depletion was associated with a spike in TH-dependent signaling (Figure S5E) and a transient induced transcription of differentiation markers (Figures 6A and 6B), which is in agreement with the marked similarity between the apoptotic cascade and muscle differentiation (Fernando et al., 2002). Interestingly, FoxO3 was significantly induced upon D3 depletion in vitro and confirmed in vivo (Figures 6A, S5F, and S5G). The upregulation of FoxO3 preceded the peak of MyoD, a well-established TH-responsive gene, thereby confirming data showing that FoxO3 induces MyoD in muscle cells (Hu et al., 2008) (Figures 6A and 6B).

Figure 6

Increased TH Signal Activates a Proapoptotic Cascade Involving FoxO3 and MyoD

(A) Time-dependent induction of TH target genes after Ad-CRE-induced D3 depletion in mpc cells from cD3KO mice or WT mice as indicated.

(B) FoxO3 and MyoD mRNA measured by RT-PCR in the same samples as in (A).

(C) FoxO3 mRNA (bottom) and protein (top) in TH-treated mpc from WT mice.

(D) FoxO3 mRNA measured by RT-PCR in the Mpc from WT mice transfected with empty vector (CMV-FLAG) or TRβ dominant-negative-expressing plasmid (TRPV) as indicated. Top: FoxO3 protein levels were measured by western blot analysis in the same samples.

(E) FoxO3 transcriptional activity measured in proliferating (PROL) and differentiating (DIFF) mpc treated with TH or rT3 and transfected with the FoxO3-responsive promoter (DBE-LUC) and CMV-Renilla plasmids as internal standard. The LUC/Renilla value from untreated, proliferating cells was arbitrarily set as 1. Data are expressed as mean ± SEM of at least three independent experiments. ∗p < 0.05.

(F) FoxO3 mRNA from WT mice untreated (EU), treated with TH (HYPER), or treated with MMI and KClO4 (HYPO) as indicated (see Experimental Procedures).

(G) FoxO3 mRNA levels in mpc from control (WT) and D2KO mice measured by RT-PCR.

(H) FoxO3 mRNA from WT and D2KO muscles at different postnatal days as indicated.

For (B)–(H) data are expressed as mean ± SEM of at least three independent experiments.

To verify that TH induces FoxO3 levels, we measured the response of FoxO3 to TH treatment in myogenic cells. FoxO3 was induced by TH at mRNA and protein levels (Figure 6C). This effect was abrogated when a dominant-negative TH receptor (TRPV) was transfected (Figures 6D and S4B), which indicates that it requires a functional TR. TH also significantly increased FoxO3 transcriptional activity, as demonstrated by the significant increase in the activity of a FoxO3-responsive artificial promoter upon treatment with TH, as well as its reduction by intracellular T3 depletion obtained by reverse T3-mediated (Dentice et al., 2010) blockade of D2 (Figure 6E). Importantly, the dependence of FoxO3 on TH levels was confirmed in vivo by the appropriate FoxO3 levels in both hypo- and hyperthyroid muscles (Figure 6F). Interestingly, FoxO3 was reduced in D2KO cells and muscles (Figures 6G and 6H), in which intracellular T3 concentrations are reduced, despite normal circulating T3 (Marsili et al., 2011), indicating a direct correlation between intracellular TH and FoxO3 levels.

In summary, this set of experiments provides compelling evidence that TH is a master regulator of FoxO3 and demonstrates that FoxO3 responds to the modulation of TH signaling in muscle.

D3 Depletion Induced Apoptosis via Upregulation of the FoxO3-MyoD Axis

To assess the functional relevance of the FoxO3-MyoD axis in TH-dependent apoptosis, we analyzed separately the roles of FoxO3 and MyoD in TH-treated cells. Interestingly, TH failed to induce apoptosis in cells expressing a dominant-negative FoxO3 protein (Figures 7A and S4B) and in myoblasts genetically depleted of FoxO3 (Figures 7B and 7C), which indicates that FoxO3 plays a crucial role in TH-induced apoptosis. We also assessed the effects of TH in myoblasts in which MyoD expression was efficiently blocked by RNAi (Figures S6A and S6B). Again, TH failed to induce apoptosis (Figures 7D and 7E). This result indicates that an active MyoD is also required for TH-induced apoptosis, and is consistent with a linear cascade in which TH induces FoxO3, which lies upstream of MyoD. Consistent with this model, both FoxO3 and MyoD deficiency were rescued by transfecting a constitutively active pAKT-resistant FoxO3 (Figure 7C) as well as by MyoD overexpression (Figure S6B), thereby restoring in both cases the capacity of TH to induce apoptosis. These gain- and loss-of-function analyses clearly indicate that FoxO3 is a critical mediator of the apoptosis driven by TH.

Figure 7

The FoxO3-MyoD Axis Is Required for TH-Induced Apoptosis

(A) Mpc from WT mice transiently transfected with empty vector (CMV-FLAG) or FoxO dominant-negative (FoxOd.n.) expression plasmid and treated or not with TH. Apoptosis was measured based on western blot of PARP cleavage.

(B) Western blot analysis of PARP cleavage in mpc from WT and FoxO-KO mice treated or not with TH.

(C) PARP cleavage in mpc from FoxO-KO mice. Cells were transfected with empty vector or FoxO3 expression vector and treated or not with TH as indicated.

(D) Mpc from WT mice transiently transfected with control iRNA oligonucleotides (iCTR) or oligonucleotide for specific MyoD silencing (iMyoD) and treated with TH as indicated. PARP cleavage was evaluated by western blot from total lysates as marker of apoptosis.

(E) Representative images of the same cells as in (D).

(F) TH target gene expression in mpc from WT and FoxO-KO mice treated or not with TH evaluated by RT-PCR. Cyclophilin A served as internal control. For each gene, normalized copies of the target gene in untreated (CTR) cells from WT mice were arbitrarily set as 1. Data are expressed as mean ± SEM of at least three independent experiments.

(G) The TH-FoxO3-MyoD axis promotes apoptosis in activated myogenic precursor cells.

Finally, we asked whether TH-mediated muscle differentiation also requires an intact FoxO3. The expression of several well-known T3-dependent myogenic markers was not responsive to TH in FoxO3-depleted cells (Figure 7F), and this was not due to altered expression of TH receptors (Figure S6C). These results indicate that FoxO3 is a critical mediator of TH action in skeletal muscle and that, notwithstanding exposure to excessive TH, FoxO3-depleted myoblasts do not undergo apoptosis upon TH treatment and are phenotypically hypothyroid at nuclear level.

Discussion

Here we report that D3 is a survival factor for proliferating satellite cells. We show that a time-dependent and cell-autonomous degradation of TH signaling via D3 expression is required for correct satellite cell amplification and muscle repair. In the absence of D3, proliferating myoblasts undergo apoptosis, and this is functionally critical in regenerating muscle, but less evident under homeostatic conditions (Figure S7). To our knowledge, this study is the first to describe how a ubiquitous endocrine signal—TH—is integrated and adapted in the stem cell niche to enable cell lineage progression.

In muscle stem cells, only activated proliferating cells apoptose, not those that are quiescent or arrested in S phase (Figure 5). We previously demonstrated (Dentice et al., 2010) that “physiological” T3 is too low to allow proper cell differentiation. Therefore, differentiating cells express D2 in order to locally produce extra T3 to allow proper differentiation. Vice versa, proliferating myoblasts require lower than circulating TH levels, and to this aim activate D3-mediated T3 inactivation. This mechanism protects them from apoptosis induced by the “excessive” levels of T3 present in the plasma. Therefore, satellite cells are able to customize their intracellular T3 concentrations based on their functional state and dynamic needs. It is not yet clear why this is the case, but we speculate that temporally excessive TH creates a metabolic or mitotic catastrophe, which cannot be overcome by the proliferating cells.

In vitro, D3 is potently induced by growth factors (EGF, FGF, and PDGF) and cytokines (Dentice et al., 2009). We demonstrated in neoplastic contexts that the Wnt and Shh pathways are potent inducers of Dio3 (Dentice et al., 2009, Dentice et al., 2012). No upstream regulators of D3 in myoblasts are known. In the present study, we found that β-catenin, Sonic Hedgehog, c-Met, cMyc, and Pax7 induced D3 in myoblasts in vitro (Figure S6D). It is conceivable that D3 is induced by a set of integrated signals from growth factors, cytokines, morphogens, and other signaling pathways that are active at the injury site. Dissecting their specific role will be the object of future studies.

T3 is a circulating hormone whose plasma concentrations are relatively stable in the healthy state. In plasma, the hypothalamic-pituitary-thyroid axis prevents potentially adverse oscillations of T4 and T3 via a potent homeostatic mechanism that controls hormone secretion by the thyroid gland (Bianco and Kim, 2006). A major question in this context is how muscle stem cells customize their TH signature and adapt it to their physiology. The rapid increase in D3 in proliferating muscle stem cells and the consequent attenuation of T3 nuclear receptor saturation is an ideal solution to this problem.

The local metabolism of TH is a highly regulated process during embryonic and fetal life, as well as during muscle regeneration. In general, D3 is reciprocally regulated with D2 to maintain local control of the T3 concentration. These interlocking pathways facilitate specific chronotropic and tissue-specific changes in T3 concentrations during periods when developmental or regenerative programs call for a transient increase or decrease in specific cells (Gereben et al., 2008). The rapid increase of D3 in proliferating myoblasts clearly indicates that the ambient plasma and cellular concentrations of TH are too high to sustain their proliferation. D3 is also expressed in activated macrophages at the injury site (Boelen et al., 2011), although its role in this cell context is presently unknown. This mechanism also occurs in the mdx mouse model, where D3 is expressed at significantly higher levels compared with healthy muscle (Figure 1D).

Skeletal muscle stem cells are a heterogeneous population of cells that span from stem-like cells to cells more committed toward myogenic lineage progression and differentiation. There are at least two populations of Pax7-positive satellite cells resident in skeletal muscle. One population rapidly contributes to muscle repair; the second is more stem cell-like and remains longer in a quiescent state in the recipient muscle. These functional differences are reflected by the different levels of Pax7 (Rocheteau et al., 2012). In this context, Pax7Hi represents a reversible dormant stem cell state during homeostasis (Mourikis et al., 2012). Here we show that T3 availability is differentially regulated in these two subpopulations of myogenic cells via D3 action. During muscle homeostasis, D3 is expressed in the satellite cells that are poised for myogenic commitment (Pax7Lo), wherein D3 expression does not simply reflect a secondary homeostatic regulation of this enzyme, but causes an intracellular attenuation of the TH signal compared to the Pax7Hi cells. This is reflected in the divergent expression of oppositely regulated TH-responsive genes, which change depending on the reduced TH signaling mediated by D3 (Figures 2E and 2F). Such a regulation could correlate with the metabolic status of the cells, e.g., Pax7Lo quiescent cells are more metabolically active than the corresponding Pax7Hi (Rocheteau et al., 2012). This raises intriguing issues about whether and how T3 signaling might influence stemness. Here we show that the control of TH signaling in the muscle stem compartment occurs, and that it is necessary for muscle cell lineage progression. This is in agreement with the critical role of TH action in neurons, where TH signaling represses Sox2 levels and regulates neurogenesis (López-Juárez et al., 2012).

Hyperthyroidism in humans may cause a severe muscle myopathy associated with a failure in muscle regeneration (Anderson et al., 1994, Simonides and van Hardeveld, 2008). While it was thought that the effects of hyperthyroidism occurred at the level of myofibers (Simonides and van Hardeveld, 2008), the reduced expression of Pax7 and increased apoptosis in hyperthyroid muscle in mice characterized in this study point to a failure at stem cell level, which may well occur also in humans.

Here we show that FoxO3 is a positively regulated downstream effector of TH signaling. Importantly, the identification of FoxO3 as a T3 target links two potent systems controlling multiple aspects of muscle physiology. In silico examination of the mouse FoxO3 locus showed multiple potential candidate sequences containing TR-binding elements, although none was evolutionally conserved, and ChIP assays failed to identify functional relevance under our experimental conditions. Nevertheless, the accumulating evidence indicating crosstalk between T3 and the AKT pathway (Figure 6) points to a complex network involving FoxO3 regulation by T3. Intriguingly, FoxO3 is a potent inducer of D2—the T3-producing enzyme—thus suggesting that a positive autoregulatory loop might be one mechanism by which FoxO3 is a target of T3, and also indirectly sustains T3 intracellular concentration by inducing D2 (Figure 7G).

Asakura and colleagues suggested that MyoD is a proapoptotic transcription factor in muscle stem cells (Asakura et al., 2007). Here we show that MyoD, a target of both T3 and FoxO3 (Dentice et al., 2010, Hu et al., 2008), is required for the proapoptotic cascade triggered by T3. Finally, we show that not only does FoxO3 act as a TH target, but also that its action is required to sensitize muscle stem cells to TH levels.

In summary, the absence of D3 in activated muscle stem cells causes a time-dependent and irreversible stem cell death due to excessive intracellular nuclear T3. D3 depletion renders these cells incapable of overcoming the T3 excess, thus promoting the apoptotic pathway. The finding that D3 plays a crucial role in controlling nuclear T3 in muscle stem cells and in the triggering of a death pathway via the FoxO3-MyoD axis has broad implications and opens a future research scenario for the development of translational therapeutic strategies based on the local control of potent endocrine signals by modulating their metabolic pathways.

Experimental Procedures

Cell Cultures, Transfections, and Reagents

Primary muscle cultures (pp6) were isolated as described (Qu et al., 1998) from the indicated mouse lines. FoxO3-depleted (FoxO-KO) pp6 cells were prepared from foxO3 mice as previously described (Dentice et al., 2010). C2C12 cells were obtained from ATCC. In some experiments, endogenous T3 and T4 were removed from the FBS by charcoal absorption (Larsen, 1972). Transient transfections were performed using Lipofectamine 2000 (Life Technologies) according to the manufacturer’s instructions. Single myofibers were prepared from the extensor digitorumlongus and gastrocnemius muscles of 6- to 12-week-old mice as previously described (Rosenblatt et al., 1995). Pax7Hi-Lo isolation by FACS has been described elsewhere (Rocheteau et al., 2012). Anti-MyoD (sc-304), myogenin (sc-12732), tubulin (sc-8035), and anti-FoxO3 antibodies were purchased from Santa Cruz Biotechnology. Polyclonal anti-MHC antibody (MF-20a) and anti-Pax7 antibody were from Developmental Studies Hybridoma Bank. Anti-D3 antibody is described elsewhere (Huang et al., 2003). Anti-Desmin antibody was from MP Biomedicals (#10519). Anti-total Akt was from Upstate, and anti-pAkt and anti-PARP were from Cell Signaling Technology. TUNEL assay was performed using the ApopTag kit from Millipore. Caspase-3/Caspase-7 activity was measured by using the CaspaseGlo kit from Promega.

FACS-Sorted Satellite Cells

Cells isolated by FACS were prepared from Tg:Pax7-nGFP mice as previously described (Rocheteau et al., 2012). In brief, muscle dissection was done with a scalpel by removing the tissue from the bone in DMEM. Muscles were then chopped and digested with collagenase 0.1% and trypsin 0.25% at 37°C. The collagenase/trypsin solution was added to continue the digestion until the process was completely terminated and muscle totally digested. Satellite cells have been cultured in 1:1 DMEM:MCDB containing 20% serum FBS and ITS (1 ×, Insulin-Transferrin-Selenium; Gibco). Cells were plated on Matrigel (BD Biosciences; catalog #354234) or collected in a lysis buffer (RNeasy Micro Kit, Quiagen) for RNA extraction. Cells were prepared for FACS analysis using a MoFlo (Beckman Coulter).

Plasmids and Expression Constructs

FoxO3, FoxOdn, and DBE plasmids are described elsewhere (Dentice et al., 2010). TRPV plasmid was kindly provided by Dr. S. Cheng. To knock down MyoD, two different Stealth siRNA targeting two different regions of the coding sequence (MyoD-7, GACGACUUCUAUGAUGACCCGUGUU; and MyoD-8, CACUACAGUGGCGACUCAGAUGCAU) were designed with the Block-iT technology (Invitrogen) and purchased from Invitrogen Corporation. Effective MyoD knockdown was assessed by measuring endogenous MyoD levels in satellite cells following 48 hr Stealth siRNA transfection (Figure S6A).

Generation of the cD3KO Mouse Model

All three deiodinases contain a critical selenocysteine residue in the catalytic domain, which is encoded by a UGA codon and requires a specific secondary mRNA loop (SECIS element) in the 3′ UTR of the mRNA that overrides the stop function and inserts selenocysteine (Berry et al., 1991). Given the absolute requirement of a SECIS element for correct D3 protein translation, we selectively ablated the SECIS region in the dio3 gene, thus preserving the dio3as transcript encoded in the antisense orientation in the dio3 locus, the function of which is not known (Figure S2). We generated a plasmid harboring floxed sites in the dio3 locus, and specifically flanking the SECIS mRNA structure located at nt 1,001 and nt 1,706 from the ATG in the dio3 mRNA (Figure S2). In the absence of the SECIS, the TGA codon, within the D3 catalytic domain, is recognized as stop codon, and protein translation terminated. Once we obtained the homozygous dio3fl/fl mice, we crossed these mice with heterozygous Tg:Pax7-CreERT2 mice, expressing the Cre-ERT2 fusion protein. The Cre-ERT2 protein consists of Cre recombinase fused to a triple-mutant form of the human estrogen receptor and catalyzes the lox sites recombination TAM inducible. The expected recombination event occurred in the progeny, as documented in the PCR in Figure S1D. Dio3fl/fl not expressing recombinase, but TAM treated, were used as negative controls.

Immunofluorescence and Immunohistochemistry

For immunofluorescent staining, cells were fixed with 4% formaldehyde and permeabilized in 0.1% Triton X-100, then blocked with 0.5% goat serum and incubated with primary antibody. Dissected muscle was snap frozen in liquid nitrogen, sectioned (7 μm thick), and stained with hematoxylin and eosin stain (H&E) or immunofluorescence using standard protocols. Alexa 595-conjugated secondary antibody was used. Images were acquired with an IX51 Olympus microscope and the Cell∗F software. Images were assembled using Adobe Photoshop. Western blots were run in triplicate, and bands were quantitated in one representative gel.

Muscle Regeneration Studies

CTX was injected as described previously (Yan et al., 2003). Briefly, 10–20 μl of 10 μM CTX (Naja mossambica mossambica, Sigma-Aldrich) was injected into the right TA and gastrocnemius muscles of anesthetized 12-week-old C57BL/6 male mice (The Jackson Laboratory) (WT), and age- and gender-matched cD3KO mice. The uninjected left TA muscles served as controls. For CTX experiments in cD3KO mice, animals were treated daily with 75 μg/g TAM for 5 days, and CTX was injected on day 4. Mice were killed 4, 14, or 40 days after CTX injection. For CTX experiments in WT mice, animals were treated daily with TH (2.5 μg/mouse T3 + 10 μg/mouse T4) 4 days before CTX injection; mice were killed 4 and 14 days after CTX injection. Hypothyroidism was obtained by supplying the drinking water with 0.1% MMI and 1% KClO4 for 6 weeks. Animal experimental protocols were approved by the Animal Research Committee of the University of Naples Federico II.

Statistical Analysis

Errors quoted are SEM throughout. Differences between samples were assessed by the Student’s two-tailed t test for independent samples. In all experiments, differences were considered significant when p was less than 0.05. Asterisks indicate significance at ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001 throughout.

Author Contributions

V.D., A.S., C.L., O.G., S.Y., and P.Z. performed in vitro and in vivo experiments; R.A. generated mouse models; G.M., A.C., S.B., and L.D.V. provided observations and scientific interpretations; P.R.L. and S.T. provided essential reagents and scientific insight; M.D. performed most experiments and analysis, prepared figures, and wrote the paper; D.S. designed the overall study, supervised the experiments, analyzed the results, and wrote the paper; and all authors discussed the results and provided input on the manuscript.