A Global Loss of Dio2 Leads to Unexpected Changes in Function and Fiber Types of Slow Skeletal Muscle in Male Mice

Abstract

The type 2 iodothyronine-deiodinase (D2) enzyme converts T₄ to T₃, and mice deficient in this enzyme [D2 knockout (D2KO) mice] have decreased T₃ derived from T₄ in skeletal muscle despite normal circulating T₃ levels. Because slow skeletal muscle is particularly susceptible to changes in T₃ levels, we expected D2 inactivation to result in more pronounced slow-muscle characteristics in the soleus muscle, mirroring hypothyroidism. However, ex vivo studies of D2KO soleus revealed higher rates of twitch contraction and relaxation and reduced resistance to fatigue. Immunostaining of D2KO soleus showed that these properties were associated with changes in muscle fiber type composition, including a marked increase in the number of fast, glycolytic type IIB fibers. D2KO soleus muscle fibers had a larger cross-sectional area, and this correlated with increased myonuclear accretion in myotubes formed from D2KO skeletal muscle precursor cells differentiated in vitro. Consistent with our functional findings, D2KO soleus gene expression was markedly different from that in hypothyroid wild-type (WT) mice. Comparison of gene expression between euthyroid WT and D2KO mice indicated that PGC-1α, a T₃-dependent regulator of slow muscle fiber type, was decreased by ∼50% in D2KO soleus. Disruption of Dio2 in the C2C12 myoblast cell line led to a significant decrease in PGC-1α expression and a faster muscle phenotype upon differentiation. These results indicate that D2 loss leads to significant changes in soleus contractile function and fiber type composition that are inconsistent with local hypothyroidism and suggest that reduced levels of PCG-1α may contribute to the observed phenotypical changes.

Thyroid hormone (TH) plays important roles in development, growth, and metabolism, acting on many different target tissues (1). TH mediates its effects on gene expression at both the genomic and nongenomic level (1–3). Most of the well-described effects of TH are attributed to its genomic action, where the biologically active form of TH, T₃, binds to nuclear TH receptors (TRs) to either positively or negatively regulate gene transcription (1). The action of this potent transcriptional regulator is controlled at many levels, including production, secretion, transport into the cell, and by tissue- and cell-specific local deiodination either activating or inactivating TH (1). The iodothyronine deiodinase family of selenoenzymes mediates the activation and inactivation of TH and includes three members, types 1, 2, and 3 deiodinases (D1, D2, and D3). D1 and D2 activate the prohormone T₄ to T₃ by removal of an outer-ring iodine, whereas D1 and D3 can inactivate T₄ and T₃ by inner-ring deiodination, generating reverse T₃ and 3,3′ T₂, respectively (4). D1 and D2 contribute to the circulating pool of T₃, whereas increases in D2 expression can augment T₃ content within a specific cell type or tissue, thereby providing local control of intracellular T₃ (5). Conversely, increases in D3 lead to a specific reduction of T₃ levels within cells and tissues, thus attenuating the TH signal (5). Changes in the expression and activity of both D2 and D3 provide a powerful mechanism for dynamic local changes in intracellular T₃ leading to a vast array of specific downstream effects (6).

Skeletal muscle is a TH target tissue, and patients with either hypothyroidism or thyrotoxicosis often develop muscle dysfunction (7–11). Hypothyroidism induces a shift toward slow, oxidative fiber characteristics with increased fatigue resistance, whereas thyrotoxicosis increases the content of fast, predominantly glycolytic fibers (7, 8, 12–14). Functionally, this leads to muscle contraction and relaxation rates being decreased during hypothyroidism and increased during thyrotoxicosis, with concomitant changes in endurance (7, 15, 16). TH administration induces a sequential transition from the slowest to the fastest myosin heavy chain (MHC) isoform in muscle fibers, that is, MHC I to IIa, to IIx, to IIb, constituting type I, IIA, IIX, and IIB fibers, which is in part mediated indirectly through changes in miR-133a1 and the transcription factor TEA domain family member 1 (17–20). Additionally, TH regulates the expression of the sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase (SERCA) type 1 and 2a proteins that mediate Ca²⁺ reuptake into the sarcoplasmic reticulum and determine muscle relaxation properties (8, 21). TH also acts directly on the muscle stem cell equivalent, the satellite cell, and is necessary for muscle regeneration after injury (22). Recent studies have shown that in satellite cells a programmed local inactivation of TH by D3 followed by a properly timed activation of T₄ by D2 is required for normal muscle repair (23, 24). T₃ also stimulates autophagy in skeletal muscle, and T₃-dependent increases in mitophagy are coupled with increased mitochondrial biogenesis and activity (25).

The relevance of D2 activity in skeletal muscle is suggested by its active regulation during hypothyroidism, cold exposure, critical illness, chronic inflammation, and with exercise (26–32). Although a global targeted disruption of the Dio2 gene in mice [D2 knockout (D2KO)] leads to increased levels of T₄ and TSH, circulating levels of T₃ remain normal (33, 34). Despite this, isotopic tracer studies have shown that these mice have a reduced amount of T₃ derived from local T₄ to T₃ conversion in their skeletal muscle and, in line with this, tissue T₃ content was found to be decreased by about one-third in skeletal muscle of D2KO mice (35, 36). Consistent with these results, D2KO mice have lower expression of T₃-responsive genes such as MyoD, SERCA2a, and Troponin 1 and 2 in their tibialis anterior muscle (23).

Notably, D2 activity in the soleus is approximately sixfold higher than in the gastrocnemius, indicating that D2 activity is greater in slow-twitch vs fast-twitch muscle (28). Tonic slow-twitch muscles such as the soleus are considerably more responsive to TH than is fast-twitch skeletal muscle (8, 10, 37). Accordingly, complete disruption of T₃ signaling in TRα and TRβ knockout mice showed marked alterations in gene expression and contractile properties in soleus muscle, a typical slow muscle, with little or no changes in the pure fast extensor digitorum longus muscle (38, 39).

Given these findings, we anticipated that D2KO mice would have functional alterations in their soleus muscle that would be similar to those of hypothyroidism. Surprisingly, contractile properties of D2KO soleus muscle showed less, rather than more, slow characteristics, and gene expression patterns were not consistent with reduced TH activity. These findings were supported by immunohistochemical staining showing an increased percentage of fast type IIB muscle fibers in D2KO soleus. Additional analyses indicated reduced expression of PGC-1α, a major determinant of muscle fiber type, in the absence of D2. Our results indicate that the soleus phenotype observed in D2KO mice is complex and cannot be simply attributed to lower intramuscular levels of T₃.

Materials and Methods

Animals

All animal studies were approved by the Institutional Animal Care and Use Committees of Brigham and Women’s Hospital, Harvard University, and the VU University Medical Center, and all mice used in these studies were male. D2KO mice were described previously and were backcrossed into a C57BL/6J background (33, 40). Hypothyroidism was induced by feeding 8-week-old mice chow supplemented with 0.15% propylthiouracil (PTU) (Harlan Teklad) for 42 days, whereas controls were fed a regular diet without PTU.

Ex vivo muscle analysis

Soleus muscles from 9- to 11-week-old wild-type (WT) and D2KO mice were dissected and mounted vertically between a dual-mode lever arm and a fixed hook in a tissue bath containing continuously gassed Tyrode solution (95% O₂/5% CO₂ at 24°C) (intact muscle test system; Aurora Scientific). The muscle was stimulated directly with platinum electrodes, and following a 5-minute equilibration at slack length, preload force was adjusted to give maximal twitch force at optimal muscle length (pulse duration of 0.4 ms). Contraction characteristics were determined essentially as described previously (41). A series of tetani (pulse duration of 1 second, 45-second intervals) with increasing frequency (1, 5, 10, 20, 40, 60, 80, and 100 Hz) was used to determine the force–frequency relationship and maximal tetanic force. Muscle fatigue was induced using a total of 20 tetani (20 Hz, pulse duration of 600 ms, 3-second intervals), followed by 15 minutes of rest with a single tetanus at 1, 5, 10, and 15 minutes to monitor recovery. The contralateral muscle was dissected during the 5-minute equilibration period, mounted, and frozen in liquid nitrogen–cooled isopentane for further immunohistochemical analyses.

Cell isolation and fluorescence-activated cell sorting

Myofiber-associated cells were isolated from intact limb muscles (extensor digitorum longus, gastrocnemius, quadriceps, soleus, tibialis anterior, and triceps brachii) from 8- to 12-week-old mice by collagenase and dispase digestion as described previously (42, 43). After isolation, myofiber-associated cells were incubated in Hanks balanced salt solution containing 2% calf serum (Invitrogen) on ice for 20 minutes with fluorescence-activated cell sorting (FACS) antibodies as detailed in an online repository [(44); also, see Refs. (45–50)]. Cells were sorted using FACSAria II, MoFlo, or SORP Aria flow cytometers (BD Biosciences). Live cells were identified as calcein blue positive (1:1000, Invitrogen) and propidium iodide negative (1 mg/mL, Invitrogen). CD45⁻Sca1⁻CD11b⁻Ter119⁻CXCR4⁺β₁-integrin⁺ cells were identified as satellite cells and were double sorted to maximize purity (42, 43).

Differentiation and myogenic colony formation analysis

Differentiation was assessed by seeding double-sorted satellite cells isolated from 8-week-old male WT and D2KO mice at 5000 cells per well in a type I collagen and laminin [1 μg/mL (Sigma-Aldrich) and 10 μg/mL (Invitrogen)]–coated flat-bottom 96-well plate. Satellite cells were cultured in growth media for 2 days (F10 plus 20% horse serum plus 1% penicillin/streptomycin plus 1% GlutaMax with daily 5 ng/mL basic fibroblast growth factor addition; Invitrogen) (51). Cells were switched to differentiation media (DMEM, 1% GlutaMax, 1% penicillin/streptomycin, and 2% fetal bovine serum) and maintained for an additional 3 days. Differentiated cells were fixed in 4% paraformaldehyde and stained as described with mouse anti-fast MHC and anti-slow MHC (Sigma-Aldrich) and goat anti-mouse Alexa Fluor 594 and 4′,6-diamidino-2-phenylindole (Invitrogen) (52–54). Whole-well images were taken with a Celigo microwell image cytometer (Nexcelom Bioscience) and nuclei per myotube were quantitated using ImageJ (43).

Myogenic colony formation was assessed as described by plating cells individually in a 96-well dish and then culturing cells in growth media for 5 days (43, 55). Wells were scored positive when they contained two or more cells, and the number of cells in positive wells was also determined.

Immunohistochemistry staining and quantitation

Staining of the mouse soleus to determine fiber type was performed as described previously (56). In brief, 10-μm frozen serial sections of soleus taken from the midpoint of the muscle from the contralateral leg of WT or D2KO mice used for ex vivo muscle analysis were stained for MHC I and IIa, MHC IIb, or MHC IIx (57–61). Information regarding the primary and secondary antibodies and staining conditions is summarized in an online repository (44).

Stained sections were imaged on a Nikon i90 microscope using NIS-Elements AR software (Nikon) at ×4. The percentages of positively stained fibers, fiber cross-sectional area (CSA), and fiber minimum feret diameter were determined using a publicly available MATLAB-based application for semiautomatic muscle analysis using segmentation histology and were scored by a blinded observer (62). Ferret diameter was determined from the smallest rectangle that bounds the fiber (62).

Quantitative real-time PCR

Sequences of primers used are listed in an online repository (44). For the time course of gene expression in neonatal WT and D2KO soleus, total RNA was extracted from tissues using TRIzol (Invitrogen), and 2.5 μg of RNA was reverse transcribed into cDNA using SuperScript VILO (Invitrogen). Postnatal day (P)1 and P5 required pooling of RNA from multiple mice to obtain sufficient amounts for cDNA synthesis in some cases, and when this was necessary equal amounts of RNA were used per mouse. cDNAs were amplified by PCR in a LightCycler 480 II (Roche) using LightCycler 480 SYBR Green (Roche). Quantitative real-time PCR (qPCR) was performed as described previously using the ΔΔCT method, with the geometric mean of cyclophilin A, hypoxanthine-guanine phosphoribosyltransferase, and TATA-binding protein used for normalization (28, 63). For measurement of gene expression of adult WT vs D2KO, and experiments with C2C12 cells, a similar procedure was used except that only cyclophilin A was used for normalization. Measurement of gene expression levels in adult WT, D2KO, and WT hypothyroid mice was performed as described previously (64). In brief, DNase I–treated total RNA was transcribed to cDNA using cloned AMV first-strand cDNA synthesis (Invitrogen). RNA concentration was adjusted to 12.5 ng/µL. qPCR was performed using MESA GREEN qPCR MasterMix Plus for SYBR assay (Eurogentec) on an Applied Biosystems qPCR machine model 7700. Expression of hypoxanthine-guanine phosphoribosyltransferase was used for normalization.

Hormone measurements

Serum T₃ and T₄ were measured as described previously using an in-house RIA (65–67).

Generation of a C2C12 cell line with a disruption of the Dio2 gene and measurement of D2 activity

C2C12 cells were obtained from the American Type Culture Collection. The Dio2 gene was disrupted in C2C12 cells by using the CRISPR-Cas9 system, as described previously (68). In brief, the guide RNA was designed to target the selenocysteine-containing active site of Dio2 and cloned into px459 v2.0 (Addgene; donated by Feng Zhang) using the following primers: 5′-CACCGCCTAGTGAAAGGTGGTCAGG-3′ (forward) and 5′-AAACCCTGACCACCTTTCACTAGGC-3′ (reverse).

Cells were transfected using Lipofectamine 3000 (Invitrogen) and subjected to antibiotic selection (puromycin, 2 µg/mL) 48 hours after transfection. After 3 days, viable cells were trypsinized, and single cells were plated onto 96-well plates and subsequently expanded. Clones were sent for sequencing to screen for gene disruption, which was confirmed by measurement of D2 activity in cell sonicates with 1 nM T₄ to assess specific activity, and 500 nM T₄ to assess nonspecific background deiodination as described previously (69).

C2C12 culture

WT and D2KO C2C12 were cultured in DMEM supplemented with 20% fetal bovine serum and 1% penicillin/streptomycin/glutamine. To differentiate, cells were grown to confluence and switched to 2% horse serum supplemented with 1% insulin-transferrin-selenium (Invitrogen). To assess T₃ responsiveness, cells were differentiated for 4 days in resin-stripped horse serum and treated with the indicated doses of T₃ (70).

Western blotting

C2C12 cell extracts were prepared by lysing in T-PER tissue protein extraction reagent (Thermo Scientific) with protease inhibitor cocktail III (Calbiochem). Protein concentration was measured using Bradford reagent (Bio-Rad Laboratories). Twenty micrograms of total protein was separated on a 4% to 20% polyacrylamide gel and transferred using the Trans-Blot Turbo system (Bio-Rad Laboratories) onto a polyvinylidene difluoride membrane. Blocking was performed in 5% BSA with 0.1% Tween 20. Antibody conditions and concentrations are listed in an online repository [(44); also see Refs. (53, 71)]. An Immun-Star horseradish peroxidase chemiluminescent kit (Bio-Rad Laboratories) was used for detection. Blots were imaged on a ChemiDoc XRS+ system (Bio-Rad Laboratories), and densitometry analysis was performed using Image Laboratory v5.1 (Bio-Rad Laboratories).

Adenoviral transduction

D2KO C2C12 cells were induced to differentiate for 24 hours and then infected with GFP or GFP and PGC-1α–expressing adenovirus. Cells were differentiated for 72 hours postinfection before collecting for downstream analysis. Adenoviruses were a gift from Dr. Alexander Banks and have been described previously (72).

Statistical analysis

Data were analyzed used GraphPad Prism 7.05 software. Details of analyses are described in the figure legends, with P < 0.05 being considered significant.

Results

Functional assessment of D2KO soleus ex vivo

We found no difference in body weight and tibia length between WT and D2KO mice. However, absolute and normalized weights of the soleus were significantly increased by 8% and 9%, respectively, in D2KO mice, with a small but nonsignificant increase in CSA (Table 1). When muscle was evaluated ex vivo, we found that both twitch force and maximal tetanic force were slightly increased in D2KO soleus; however, when these were normalized to muscle CSA, the differences were no longer significant [Fig. 1(a)–1(d)]. Because hypothyroidism leads to a fiber type transition toward type I slow fibers, we predicted that D2KO mice would have lower rates of contraction and relaxation. Surprisingly, the rates of contraction and relaxation of D2KO soleus were significantly faster, being increased by 28% and 52%, respectively, when compared with WT controls, resulting in a significant 14% reduction of the time to 50% relaxation [Fig. 1(e)–1(g)].

Table 1.

Comparison of General Parameters and Soleus of WT and D2KO Mice

	WT (n = 15)		D2KO (n = 13)		P Value
	Mean	± SEM	Mean	± SEM
General
Body weight	25.0	0.4	24.7	0.3	NS
Tibia length, cm	1.81	0.01	1.79	0.01	NS
Soleus
Muscle CSA, mm2	0.94	0.03	0.96	0.02	NS
Muscle length, cm	1.0	0.0	1.1	0.1	NS
Muscle weight, mg	8.8	0.2	9.6	0.2	<0.01
Muscle weight/tibia length, mg/cm	4.9	0.1	5.3	0.1	<0.01

Abbreviation: NS, not significant.

Figure 1.

Assessment of twitch force and rates of contraction and relaxation of WT and D2KO soleus ex vivo. (a) Twitch force, (b) twitch force normalized to muscle CSA, (c) maximal tetanic force, (d) maximal tetanic force normalized to muscle CSA, (e) half-time of relaxation, (f) rate of contraction, and (g) rate of relaxation of WT and D2KO soleus were determined ex vivo. The mean ± SEM are shown (n = 9 to 15). *P < 0.05, **P < 0.01, ***P < 0.001 by an unpaired Student t test.

WT and D2KO soleus have different fiber type composition

Soleus is a typical slow, fatigue-resistant muscle enriched in oxidative type I and IIA fibers, with a small proportion of type IIX fibers. Type IIA and IIX fibers have faster contractile properties, and the latter relies more on glycolysis for energy production. Fast, glycolytic type IIB fibers are virtually absent in soleus muscle. Thus, the increased rates of contraction and relaxation observed in D2KO soleus suggested a greater proportion of IIA and/or IIX fibers. To further understand this phenotype, we stained soleus sections from the contralateral leg of mice used for the ex vivo functional measurements for MHC I, IIa, IIx, and IIb and quantitated fiber type composition. Relatively small but significant changes were found in the number of type I and IIA fibers, with an increase in MHC I–positive and a decrease in MHC IIa–positive fibers in D2KO mice [Fig. 2(a)–2(c)]. The small number of hybrid fibers positive for both MHC I and IIa was also significantly decreased [Fig. 2(j)]. Taken together, these changes resulted in a slight yet not statistically significant net reduction of the percentage of primarily oxidative fibers in D2KO soleus. In contrast to the MHC I– and MHC IIa–expressing fibers, there were no significant changes in the percentage of fibers staining for MHC IIx, either alone or in combination with MHC IIa or IIb [Fig. 2(d)–2(f), 2(k), and 2(l)]. Additionally, there was a marked increase in the percentage of fibers staining exclusively for MHC IIb, the fastest isoform [Fig. 2(g)–2(i)]. These type IIB fibers are normally enriched in glycolytic fast muscle. Representative whole-section staining is shown in an online repository (44).

Figure 2.

Percentage of myofibers staining positive for MHC I, IIa, IIb, and IIx in WT and D2KO soleus. (a–i) Immunohistochemistry was performed on WT and D2KO soleus to determine the percentage of fibers that stained with (a–c) MHC I and IIa, (d–f) MHC IIx, and (g–i) MHC IIb. (j–l) The percentage of mixed fibers staining positive with more than one MHC is shown and was determined by either staining in the same section (I/IIa) or in serial sections (IIax and IIbx). Sections from each mouse were stained two to three times for each MHC isoform, and the fraction of positive fibers was averaged. Scale bars, 50 μm. The mean ± SEM from the average section staining of five to six mice for each genotype is shown. *P < 0.05, **P < 0.01 from WT by an unpaired Student t test. Representative whole-section stains are shown in an online repository (44).

D2KO soleus is less fatigue resistant

A fatigue and recovery protocol was used to test the functional relevance of the slightly higher proportion of glycolytic fibers in D2KO soleus. Muscles were stimulated with 20 consecutive tetani 3 seconds apart, and force was recorded at 0, 15, 30, 45, and 60 seconds (corresponding to 0, 5, 10, 15, and 20 tetani). The decay of maximal tetanic force indicates that D2KO soleus is significantly less fatigue resistant than WT muscle [Fig. 3]. Recovery of force was monitored by recording a single tetanus at 1, 5, 10, and 15 minutes after the fatigue protocol. Although force returned to >90% of the initial values in both phenotypes, recovery was slower in D2KO soleus with significantly less force being generated 5 minutes into the recovery period [Fig. 3]. These data indicate that the decreased ratio of oxidative-to-glycolytic fibers in D2KO mice has significant consequences in terms of an increased susceptibility to fatigue, as well as delayed recovery time after muscle stimulation.

Figure 3.

Assessment of fatigue resistance and recovery in WT and D2KO soleus. Muscle fatigue was induced using a total of 20 tetani (stimulation frequency 20 Hz, pulse duration 600 ms, 3-s intervals). Force was recorded at 0, 15, 30, 45 and 60 s (corresponding to 0, 5, 10, 15, and 20 tetani) and normalized to the force at t = 0 for each muscle. Values shown are the mean ± SEM of 11 (WT) and 7 (D2KO) muscles. Average absolute force at t = 0 was not significantly different between both genotypes: WT, 15.1 ± 0.9 (g); D2KO, 14.8 ± 0.6 (g). Following the last tetanus, single tetani were recorded at 1, 5, 10, and 15 min to monitor recovery. Fatigue and recovery were considered as separate groups, and significant differences were determined by a two-way ANOVA for repeated measures followed by a Sidak multiple comparison test. *P < 0.05; **P < 0.01.

D2KO soleus has increased myofiber CSA consistent with a greater in vitro fusion index of differentiated satellite cells

Fiber CSA for WT and D2KO soleus was also quantitated. The average CSA of D2KO soleus myofibers was significantly greater [1197 ± 31 μm² for WT vs 1300 ± 22 μm² for D2KO (mean ± SEM), P < 0.05], with the overall fiber size distribution being shifted toward the right [Fig. 4(a)]. This shift is solely the result of the greater average CSA of type IIA and type IIX fibers in D2KO soleus because average CSA of type I fibers was not altered [Fig. 4(b)–4(d)]. CSA was not quantitated in MHC IIB fibers owing to the small number of positive fibers in each muscle section (ranging from 0 to 71 per sample). Similar results were obtained when the minimum feret diameter was used for quantitation (44, 62).

Figure 4.

Myofiber CSA distribution of WT and D2KO soleus and evaluation of WT and D2KO satellite cell fusion index. WT and D2KO soleus were stained with laminin to outline myofibers, and CSA was determined using semiautomatic muscle analysis using segmentation of histology (62). (a–d) Images quantitated were the same as in Fig. 2, and the CSA distributions of (a) all fibers or (b) type I–positive, (c) type IIa–positive, and (d) type IIx–positive fibers are shown. Values were binned into 200 μM groups with the upper limit shown on the x-axis. Data shown are the mean ± SEM from four to six mice per genotype. Two-way ANOVA determined a significant interaction between genotype and CSA (P < 0.001) for (a), (c), and (d). *P < 0.05, **P < 0.01, ***P < 0.001 when compared with WT by a Sidak multiple comparison test. (e and f) FACS-purified satellite cells from WT and D2KO mice were differentiated and the (e) fusion index (number of nuclei in MHC⁺ myotubes with three or more nuclei per total nuclei in MHC⁺ cells) and the (f) number of nuclei per myotube were quantitated. Values shown are mean ± SEM from three biological replicates in two to three separate experiments. (e) No difference was found by an unpaired Student t test, whereas genotype and nuclei number per myotube were found to have a significant interaction by two-way ANOVA (P < 0.001). (f) ***P < 0.001 for WT and D2KO values by a Sidak multiple comparison test. (g) Representative images of differentiated WT and D2KO satellite cells. Fast and slow MHC protein are stained red, and nuclei are stained blue with 4′,6-diamidino-2-phenylindole. Images were obtained at 10× magnification. Cells with more nuclei characteristic of differentiated D2KO satellite cells are indicated by white arrows.

The increased myofiber CSA found in D2KO soleus in vivo is consistent with our in vitro findings. When FACS-purified WT satellite cells, the muscle stem cell equivalent, were cultured in vitro we found no difference in the number of myogenic colonies formed from a single cell, and no difference in the number of cells per colony, suggesting that there are no intrinsic differences between WT and D2KO satellite cells in their ability to survive and expand under conditions that are supportive of myogenic cell proliferation (44). Differentiation of WT satellite cells in vitro by culture under low serum conditions led to a robust 15-fold induction of Dio2 gene expression along with a >200-fold induction of the differentiation marker myogenin (44). When WT- and D2KO-derived satellite cells were differentiated in vitro there was no difference in the percentage of cells staining positive for MHC between the genotypes, suggesting that both cell types differentiated to a similar extent (44). Furthermore, there was no difference in the fusion index between WT and D2KO-derived satellite cells [Fig. 4(e)]. Despite this, differentiated D2KO satellite cells had a different pattern of myonuclear accretion, forming significantly fewer myotubes with 3 to 9 nuclei per myotube, and a greater number of myotubes with 10 to 19 nuclei per myotube [Fig. 4(f) and 4(g)]. This suggests that D2KO satellite cells have abnormalities in the fusion process that lead to more cells fusing per myotube formed, and it is consistent with the increased myofiber CSA found in D2KO soleus in vivo.

Gene expression patterns in D2KO muscle do not mimic hypothyroidism

Our functional data combined with our immunohistochemistry results suggest that D2KO mice do not exhibit a hypothyroid phenotype in soleus. However, at the time of these studies, previous work showing that low circulating TH levels lead to phenotypic changes in slow muscle was only done in rats or rabbits, not mice (21, 37, 73–75). To test whether the effects of low TH levels could be species-specific, we made WT control mice hypothyroid by feeding them chow with 0.15% PTU, and then compared the pattern of gene expression of their soleus to euthyroid WT control and D2KO mice. As reported earlier, T₄ levels were higher in euthyroid D2KO mice compared with euthyroid WT controls, whereas serum T₃ was unchanged (33, 34) (Table 2). Serum T₄ and T₃ levels were decreased in PTU chow–fed mice to 11% and 69% of euthyroid WT control values, respectively, confirming hypothyroidism (Table 2). Overall, patterns of gene expression in hypothyroid WT mice were similar to what has been previously reported in other species. Hypothyroidism led to significant decreases in expression of MHC IIx and SERCA1, whereas expression of MHC I and SERCA2a was increased when compared with WT euthyroid controls [Fig. 5]. MHC IIb expression also decreased in hypothyroid WT soleus; however, these changes were not significant due to the large variation in expression levels found in euthyroid animals [Fig. 5(d)]. Of note, none of the changes in gene expression found during hypothyroidism in WT mice was apparent in D2KO soleus. Furthermore, MHC IIx was the only gene for which the expression was significantly different between WT and D2KO soleus, with expression levels being increased compared with WT [Fig. 5(c)].

Table 2.

Serum T₃ and T₄ Values of WT, D2KO, and WT Mice Fed Chow With 0.15% PTU

		T3			T4
	n	Mean	± SEM	P Value vs WT	Mean	± SEM	P Value vs WT
WT	9	0.81	0.05	—	4.6	0.0	—
D2KO	12	0.83	0.04	NS	5.7	0.0	< 0.001
WT + PTU	6	0.56	0.03	< 0.01	0.5	0.0	< 0.001

Abbreviation: NS, not significant.

Figure 5.

Comparison of gene expression in WT, D2KO, and hypothyroid WT soleus. Relative gene expression levels in WT, D2KO, and WT hypothyroid soleus were measured using qPCR. (a–f) The mean expressions ± SEM normalized to WT expression of (a) MHC IIx (Myh1), (b) MHC IIa (Myh2), (c) MHC IIb (Myh4), (d) MHC I (Myh7), (e) SERCA1 (Atp2a1), and (f) SERCA2a (Atp2a2) are shown. n = 5 to 12 animals per group. Samples were compared by one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001 by a Tukey multiple comparison test.

Differences in gene expression in D2KO muscle are evident in soleus by P15

The above results indicate that there are significant differences between WT and D2KO soleus that are not explained simply by T₃ deficiency in adult mice. We sought to determine whether these were due to changes in D2KO muscle that could be traced to the neonatal period when there are dramatic and dynamic changes in both circulating T₄ and T₃, as well as Dio2 and Dio3 expression, with serum T₃ and T₄ peaking at P15 (23, 76–80). We evaluated gene expression at postnatal days 0, 5, 10, and 15 in WT and D2KO soleus. We found that Dio2 expression increased steadily over time in WT mice, and Dio3 expression peaked on P5 in WT, but not in D2KO, soleus [Fig. 6(a) and 6(b)]. By P15, MHC I, MHC IIa, and MyoD expression were significantly decreased in D2KO soleus, whereas PGC-1α and Crym expression trended toward being lower [Fig. 6(c)–6(g)]. MHC IIx, embryonic MHC, MHC IIb, perinatal MHC, SERCA1, SERCA2a, cyclin D1, Pax7, Ki67, MCT8, MCT 10, and myogenin expression were unchanged (44). Notably, the decreases in MyoD, PGC-1α, and Crym persisted in D2KO adult mice (44), whereas expression of MHC I and MHC IIa were no longer different [Fig. 5(a) and 5(b)]. These results show that even by P15 differences in gene expression between WT and D2KO soleus exist that persist into adulthood.

Figure 6.

Genes with differential expression in neonatal WT and D2KO soleus. Relative gene expression levels in WT and D2KO soleus at P1, P5, P10, and P15 were measured using qPCR. (a–g) The mean expressions ± SEM of (a) D2 (Dio2), (b) D3 (Dio3), (c) MHC I (Myh7), (d) MHC IIa (Myh2), (e) PGC-1α (Ppargc1a), and (f) MyoD (Myod1), and (g) Crym, in soleus are shown. Expression levels are normalized to the average value of WT expression at P1. All tissues were obtained from male pups. P1 and P5 required pooling of tissue from one to three mice per sample to obtain sufficient mRNA (2.5 μg) for cDNA synthesis. P1, n = 4 to 6 samples per genotype; P5, n = 4 samples per genotype; P10, n = 6 to 7 samples per genotype; P15, n = 6 samples per genotype. Two-way ANOVA results for all genes are summarized in an online repository (44). For (b) and (g), there was significant interaction between age and genotype. For (b)–(e), age had a significant effect, and for (b) and (d) genotype had a significant effect. *P < 0.05, ***P < 0.001 when compared with WT by a Sidak multiple comparison test. Dio2 expression was analyzed by one-way ANOVA and was significant (P < 0.01). **P < 0.01 vs P1 by a Tukey multiple comparison test. ND, not determined.

A loss of Dio2 leads to less PCG-1α expression, which may contribute to increased fast characteristics in D2KO muscle

The T₃-dependent transcriptional coactivator PGC-1α plays an important role in muscle fiber type determination, with overexpression leading to a dramatic increase in slow-twitch fibers, whereas a muscle-specific deletion results in an increase in fast-twitch fibers (81, 82). This suggests a possible role for muscle D2 activity in providing sufficient T₃-dependent expression of PGC-1α in the development of the slow muscle phenotype. Such a role could then explain the partial impairment of the normal development of soleus muscle in D2KO mice observed in this study. Indeed, PGC-1α expression was significantly decreased by 46% in adult D2KO mice (P < 0.001) (44), although a decrease in neonatal soleus was not apparent until P15 [Fig. 6(e)].

To further address the suggested relationship between D2, PGC-1α, and fiber type, we created a C2C12 line where expression of Dio2 was disrupted using the CRISPR-Cas9 system. As expected, these cells had no D2 activity (44). Upon differentiation, expression of PGC-1α was decreased by 72% [Fig. 7(a)]. Moreover, MHC I expression was markedly decreased, whereas expression of MHC IIa was increased, indicating a transition toward a faster, more glycolytic phenotype [Fig. 7(a)]. Western blotting confirmed the reciprocal change in expression of the fast and slow MHC isoforms [Fig. 7(b)].

Figure 7.

Dio2 loss leads to decreased PGC-1α expression and a transition from slow-twitch to fast-twitch MHC expression in C2C12 cells. (a) WT and D2KO C2C12 cells were differentiated for 5 d by serum withdrawal, and gene expression levels were quantitated using real-time PCR. Results shown are the mean ± SEM of two separate experiments performed in triplicate, and were compared by an unpaired Student t test. *P < 0.05; ***P < 0.001. (b) Western blotting for fast (left panel) and slow (right panel) MHC expression in differentiated WT and D2KO C2C12 cells. Band density was quantitated using ImageJ and was normalized to α-tublin. (c) WT C2C12 cells were differentiated for 4 d in media containing T₃-free resin-stripped serum supplemented with the indicated amount of T₃, and then gene expression levels were determined qPCR as above. The data shown represent the mean ± SEM of data from three experiments run in triplicate. Significant differences were determined using one-way ANOVA followed by a Sidak multiple comparison test. *P < 0.05, **P < 0.01, ***P < 0.001 when compared with no T₃ condition.

Apart from PGC-1α, the examined muscle genes are themselves transcriptionally regulated by T₃in vivo to different degrees. We therefore tested the T₃ responsiveness of these genes, as well as PGC-1α, in differentiating C2C12 cells to assess a possible independent contribution of the reduction of T₃ concentration to the observed effects in D2KO C2C12. However, no marked differences in T₃ responsiveness were seen that could account for these effects [Fig. 7(c)].

We also assessed whether restoring PGC-1α expression by infection with a PGC-1α–expressing adenovirus would normalize the decreased expression of MHC I and increased expression of MHC IIa upon differentiation in the D2KO C2C12 cells (44). Although transduction with a PGC-1α–expressing adenovirus did significantly increase PGC-1α expression, there was no change in the expression of either MHC I or IIa in differentiated D2KO C2C12 cells when compared with cells that had been transduced with GFP-expressing adenovirus, suggesting that an increase in PGC-1α alone is not sufficient to normalize the shift toward faster MHC expression in differentiated D2KO C2C12 cells.

Discussion

Our results further define and underscore the complex role of T₃ in muscle. Skeletal muscle is composed of both slow oxidative and fast glycolytic fibers that contribute to the overall contractile and metabolic properties of a muscle. Fiber type, in turn, is determined by a complex interplay between myogenic potential and innervation (83). T₃ is an additional factor that contributes to development of the adult phenotype of the individual muscle fibers. T₃ drives the expression of a faster, more glycolytic phenotype characterized by a progressive transition from type I, to IIa, to IIx, to IIb MHC expression, including shifts in associated faster isoforms of both regulatory and accessory contractile proteins. This responsiveness of the contractile and metabolic phenotype to changes in T₃ levels is particularly evident in slow muscle (8).

T₃ content can be increased locally in a particular cell or tissue by the outer-ring deiodination of T₄ by D2 (5). D2-mediated increases in T₃ are important in numerous processes, including brown adipose tissue thermogenesis, cochlea development, the negative regulation of TSH by T₄, maintenance of T₃ levels in the brain, bone strength and mineralization, normal motor phenotype, and muscle regeneration after injury (23, 36, 76, 80, 84–87). In muscle, D2 can be found in the differentiating muscle stem cell; however, overall D2 activity is low in adult muscle (23, 27, 28, 30, 88–90). Despite this, previous studies from our group using isotopic tracers have shown that a significant fraction of T₃ in muscle comes from the conversion of T₄ (35). Nevertheless, analysis of models of early developmental disruption of D2 activity in myocytes did not suggest a significant role for D2 in determining muscle phenotype, although the degree of myocyte-specific knockdown of D2 in these models could not be assessed with certainty (90, 91). In contrast, in the global D2KO model changes were found in the fast glycolytic tibialis anterior muscle consistent with a mildly hypothyroid condition (23). In the same model, Bárez-López et al. (36) reported that D2KO did not affect gross structural features of fast and slow muscle as assessed by hematoxylin and eosin staining, despite a 33% decrease in T₃ content in the quadriceps.

With this in mind, and considering that fast muscle is typically much less responsive to a reduction of T₃ levels than is slow muscle, we expected that the soleus muscle of D2KO mice would be functionally hypothyroid and display increased characteristics of oxidative muscle, that is, slower contractile properties and increased resistance to fatigue. Unexpectedly, D2KO soleus muscle exhibited increased characteristics of fast-twitch glycolytic muscle with higher rates of contraction and relaxation, and a greater susceptibility to fatigue. Although the combined fiber type data in Fig. 2 show no net increase in the percentage of type II fibers, there is a significant decrease in mixed oxidative/glycolytic type IIA fibers, whereas the fastest, pure glycolytic type IIB fibers are markedly increased [Fig. 2(a) and 2(g)]. Importantly, the average CSA of the type IIA and type IIX fibers is significantly increased in D2KO soleus, with no effect on this parameter in type I fibers. Taken together, these quantitative and qualitative differences in the fiber type composition in D2KO soleus can account for the observed faster twitch characteristics.

The relevance of the change in fiber type composition is underscored by the greater drop in force during repetitive stimulation and the slower rate of recovery. This reflects the exhaustion of the greater percentage of pure glycolytic type IIB fibers found in the D2KO soleus muscle, and it can provide an explanation for the reduced endurance of D2KO mice reported by Bárez-López et al. (36). Our data showing the effect of the altered fiber type composition on fatigability in D2KO soleus are quantitatively similar to those reported in a study by Feng et al. (92) in which hindlimb unloading induced a shift from type IIA to IIX and IIB fibers in soleus. This shift was reversible upon reloading and correlated with normalization of PGC-1α expression. These authors furthermore suggested that the slower rate of recovery of force, as also seen by us, is related to suboptimal mitochondrial function as a result of reduced PGC-1α activity. Interestingly, earlier studies on muscle-specific PGC-1α knockout mice showed a similar, albeit more pronounced, shift in type II fiber types in soleus as well as a reduction of endurance (82).

Studies directly comparing gene expression patterns in soleus from D2KO and hypothyroid WT mice confirmed the functional analyses that indicated that the D2KO soleus does not appear to be hypothyroid. Furthermore, there was only a loose correlation between gene expression levels in adult muscle and our immunohistochemical staining, suggesting that posttranscriptional mechanisms of regulation of the MHC isoforms may also come into play. Our data are in agreement with other studies showing that mice with a skeletal muscle–specific deletion of Dio2 also had patterns of gene expression that did not match those of hypothyroid mice (91). Our results further indicate that the complex changes observed in D2KO muscle may be developmental in origin and are not consistent with what would be expected with simple hypothyroidism during adulthood. Motor neuron activity also contributes to muscle fiber phenotype; however, our studies do not address whether a loss of Dio2 either directly or indirectly changes the motor neuron input to muscle (8). Previous work comparing WT hypothyroid and D2KO mice found that despite a similar reduction in brain tissue T3 content, gene expression patterns were significantly different (86, 93). Tissue T₄ levels were increased in D2KO but not WT hypothyroid brain, suggesting that this could play a role in the differential gene expression. In our studies we also cannot rule out that an increased T₄ content in D2KO vs WT hypothyroid soleus may contribute to the differences observed.

The T₃-dependent transcriptional coactivator PGC-1α can significantly influence muscle fiber type. Overexpression of PGC-1α in mice leads to an increase in oxidative type I fibers, whereas muscle-specific deletion, or partial reduction due to unloading, results in a shift from type IIA to type IIX and type IIB fibers, similar to what was observed in our D2KO mice (81, 82, 92, 94). Previous studies have established a link between Dio2 and PGC-1α expression because mice with a skeletal muscle–specific deletion of the Dio2 gene have impaired exercise-mediated induction of PGC-1α in soleus, and D2KO mice had less induction of PGC-1α in their lungs following bleomycin injury (32, 95). We also noted that PGC-1α expression trended toward decreased levels in the soleus of D2KO neonates by P15, and was significantly decreased by ∼50% in the soleus of D2KO adults. Along with the shift from type IIA to the more glycolytic type IIB fibers, D2KO mice and mice with a muscle-specific deletion of PGC-1α also display similar phenotypical changes in muscle function with respect to fatigability ex vivo and in vivo (36, 82).

In vitro, we find that a disruption of the Dio2 gene in C2C12 cells also leads to a decrease in PGC-1α levels, as well as a more glycolytic phenotype with differentiation, with increased expression of MHC IIa and a reduction of MHC I. However, increasing PGC-1α expression in these cells did not lead to a reversion to a more oxidative phenotype. These results suggest that a loss of other, yet to be identified, factors upstream of PGC-1α expression drive the D2KO phenotype and that the resulting decrease in PGC-1α is not the primary cause of the observed phenotypic changes. Alternatively, the failure of PGC-1α expression to revert the glycolytic conversion caused by inactivation of Dio2 may mean that the C2C12 cells simply do not faithfully recapitulate some aspects of muscle in vivo. In this regard, primary cultures of Dio2-lacking myoblasts showed increased expression of MHC I and decreased expression of MHC IIa, whereas PGC-1α expression was unchanged (91, 96), indicating that some intrinsic differences between primary cells in culture vs the C2C12 cell line can occur.

It is unknown why PGC-1α expression is lower in D2KO soleus and in D2KO C2C12 cells. T₃ levels have been reported to be decreased by about one-third in the skeletal muscle of D2KO mice as a result of impaired T₄ to T₃ conversion (35, 36). Because PGC-1α is a T₃-responsive gene, one possibility is that PGC-1α expression requires higher T₃ concentrations than do other T₃-dependent muscle genes (25, 97). Thus, PGC-1α expression might be impaired at the lower T₃ concentrations found in D2KO skeletal muscle, whereas other genes that require less T₃ for activation would be unaffected. However, when C2C12 cells were differentiated in media containing TH-free serum with increasing amounts of T₃ added, PGC-1α induction did not require higher T₃ concentrations than did other genes such as SERCA1 and MHC IIa [Fig. 7(c)]. However, interpretation of these experiments is not straightforward, because the relationships between T₃, PGC-1α, and muscle differentiation are complex and represent a balance between the direct effects of T₃ itself and the indirect effects of T₃ potentiation of differentiation. Additionally, PGC-1α expression is also regulated by many other factors, including Ca²⁺ signaling, nutrient availability, β-adrenergic signaling, reactive oxygen species, and hypoxia, which were not evaluated in our system (98).

The D2KO mouse phenotype also contrasts with other genetic models of impaired TH action. In this regard, mice that are null for TRα and TRβ contain significantly more type I and less type IIA fibers in soleus and have decreased contraction and relaxation rates along with increased fatigue resistance, similar to hypothyroid rodents (15, 38, 39, 73). Mice with a knock-in of a dominant-negative TRα also contain more type I fibers and fewer type II fibers (99). Because TRα should be downstream of D2 in terms of TH action, these differences are puzzling. However, these dissimilarities may indicate that the effects of a total loss of TR, or a TRα that is unable to bind ligand, are different from those in which TR is present with slightly less receptor occupancy due to decreased intracellular ligand concentrations.

Differences are also apparent between the TR-null and TRα dominant-negative mice and D2KO in terms of fiber CSA. Whereas both the TR-null and TRα dominant-negative mice have a decrease in CSA of type I and type II fibers (38, 99), the average fiber CSA in the soleus muscles of D2KO mice is increased. Furthermore, analysis of type I, type IIA, and type IIX fibers showed that the increase in CSA in D2KO mice is restricted to both type II fibers. Interestingly, when FACS-purified D2KO satellite cells were differentiated in vitro, we found that they have increased myonuclear accretion with a greater number of nuclei per myotube, although the fusion index was unchanged. This result is in agreement with what has been found in many other systems where changes in myonuclear accretion have been positively correlated with corresponding changes in CSA (100–105). Other cell types in the muscle stem cell niche, such as the fibro-adipogenic progenitors, can also potentiate differentiation, fusion, and myonuclear accretion of satellite cells during the repair process after injury (106). However, our data using FACS-purified muscle stem cells suggest that the change in myonuclear accretion of D2KO satellite cells is a cell-autonomous property, and not a result of changes in factors derived from other cell types in the muscle stem cell niche.

With respect to embryonic and fetal development, our data on the composition and CSA of the different fiber types suggest that D2 is particularly relevant after embryonic day 16, during the wave of myoblast proliferation and fusion that gives rise to secondary fast myofibers, and perhaps less so for the early development and maintenance of the primary type I fibers (83). Surprisingly, there are minimal differences in gene expression patterns between WT and D2KO soleus during the early neonatal period, with the exception of a peak of Dio3 expression at P5 that is present in WT mice but not in D2KO neonates. The lack of Dio3 induction may compensate for the loss of T₄ to T₃ conversion by D2 in this tissue, providing a mechanism by which further downstream abnormalities are prevented. At present it is unknown whether the increase in Dio3 expression is in a specific cell type, and what is driving this increase. Serum T₃ and T₄ levels steadily increase in neonates, peaking at ∼P15 (76, 77). Furthermore, Dio3 is a T₃-responsive gene (107, 108). Thus, it is tempting to speculate that the increased T₃ provided by D2 induces a compensatory increase in Dio3 expression in the soleus of WT mice, but not in D2KO neonates. Interestingly, during the early neonatal period serum T₃ levels are slightly higher in D2KO neonates, suggesting that this may compensate for the lack of intratissue conversion of T₄ to T₃ by D2 to some extent (76). By P15 differences in gene expression become apparent between WT and D2KO soleus, with the expression of MHC I and IIa and MyoD being significantly decreased, whereas the expression of PGC-1α and Crym trend toward being lower. However, in adults, only the changes in PGC-1α, MyoD, and Crym persist (44). Of note, Crym knockout mice have hypertrophy of type IIB fibers and an upregulation of fast glycolytic gene expression that can be rescued by the inhibition of TH synthesis (109). Thus, it is possible that the decrease in Crym expression found in adult D2KO mice could also contribute to the increased fast characteristics of their soleus and increased myofiber CSA.

In summary, the phenotypical changes found in D2KO soleus are surprisingly complex (Fig. 8) and do not concur with the increased oxidative phenotype that would be expected based on previous simple hypothyroid models or models where mice were null for TRα and TRβ, or a dominant-negative TRα was knocked in. The soleus of the D2KO mouse has increased contraction and relaxation rates ex vivo, along with decreased fatigue resistance, which correlates with a shift of type IIA to type IIB fibers. Although we cannot conclude direct causality, it is notable that D2KO mice have lower levels of PGC-1α, a known regulator of muscle fiber type. Furthermore, the muscle phenotype of these mice appears to be similar to that of mice with a muscle-specific knockout or reduction of PGC-1α. In vitro, we found that a loss of Dio2 leads to reduced expression of PGC-1α expression, and it further reduces expression of the oxidative MHC I isoform while increasing the expression of the faster MHC IIa. However, our studies found that a simple replacement of PGC-1α could not ameliorate this phenotype, suggesting that changes in other upstream factors might also come into play. Overall these data suggest that effects of T₃ on muscle differentiation are complex, with the end result reflecting the sum of interactions between many different factors, and that surprisingly, the loss of Dio2 does not lead to a phenotype that is consistent with a slower, more oxidative phenotype in soleus muscle.

Figure 8.

Summary of the changes in D2KO mice in vivo, and in differentiated D2KO C2C12 cells in vitro, relative to WT mice and cells, respectively. The changes in MHC gene and protein expression in D2KO C2C12 relative to WT C2C12 are shown in the right column. The changes in MHC gene expression, fiber type composition and CSA, contractile function, and serum TH levels in D2KO relative to WT mice are shown in the leftmost column. For comparison, the central column shows the changes in MHC gene expression and serum TH levels in WT hypothyroid (HYPO) mice relative to WT mice (this study), as well as fiber type composition, CSA, and contractile function of WT hypothyroid mice and rats relative to WT animals reported previously (8, 38). Parameters that were significantly increased compared are highlighted in blue, whereas those that were decreased are highlighted in pink, and those that were unchanged are not highlighted.

Glossary

Abbreviations:

CSA

cross-sectional area

D1

type 1 deiodinase

D2

type 2 deiodinase

D2KO

type 2 deiodinase knockout

D3

type 3 deiodinase

FACS

fluorescence-activated cell sorting

MHC

myosin heavy chain

P

postnatal day

PTU

propylthiouracil

qPCR

quantitative real-time PCR

SERCA

sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase

TH

thyroid hormone

TR

thyroid hormone receptor

WT

wild-type