miR-206 enforces a slow muscle phenotype

Skeletal muscle is a highly organized contractile tissue that comprises 40% of human body mass (Janssen et al., 2000). Muscle subgroups are further specialized to perform a diverse array of functions, such as chewing, focusing and moving the eye, breathing, maintaining body posture, and both burst and sustained movements. To meet these various demands, different muscle types are characterized by different metabolic and contractile machinery. Oxidative myofibers are more fatigue resistant and have higher mitochondrial content, to facilitate β-oxidation. They can be further subdivided into slow oxidative and fast oxidative myofibers. Slow oxidative fibers are also called type I fibers due to expression of type I (β)-myosin heavy chain (MyHC; MYH7), a member of the sarcomeric MyHC protein family, which ultimately confers contractile properties to the muscle. Fast oxidative fibers express type IIa (MYH2), IIx (MYH1), or a mixture of those two faster MyHCs. They also are more oxidative than type I fibers, presumably to accommodate the greater ATP demands of types IIa and IIx MyHCs compared to type I (Chemello et al., 2019). Fast glycolytic myofibers fatigue more easily and rely on glycolysis for energy production. Smaller mammals express the fastest MyHC (type IIb; MYH4) or a IIx/IIb mixture to promote higher contraction speed, while larger mammals do not express the type IIb myosin protein (Pereira Sant'Ana et al., 1997).

Although fiber types are established shortly after birth (Gokhin et al., 2008), adult skeletal muscle retains remarkable plasticity in order to adapt to changing demands. In addition to tuning performance, muscle is a metabolically demanding tissue that is critical to maintaining body-wide homeostasis due to the balance of energy substrate utilization. Physiological stimuli, such as short sprint versus endurance exercise training, shifts muscles to a more glycolytic or more oxidative phenotype, respectively (Allen et al., 2001; Andersen and Henriksson, 1977; Ross and Leveritt, 2001). Muscle unloading due to a variety of stimuli typically results in a more glycolytic profile (Caiozzo et al., 1998; Harrison et al., 2003; McCarthy et al., 1997). Certain pathological states, including muscular dystrophy, result in a shift to more oxidative/slow fibers, in part because fast fibers are more susceptible to damage (Webster et al., 1988). Finally, although understudied, evidence from rodents to humans suggests that many female muscles are slower than their male counterparts (Haizlip et al., 2015; Janssen et al., 2000).

Cardiomyocytes, the contractile cells in cardiac muscle, express two MyHCs in mammals: the same β-MyHC found in type I skeletal muscle fibers as well as the faster, cardiac-restricted α-MyHC (MYH6). There is also evidence that female cardiac muscle is slower than male muscle (Patrizio et al., 2013; Trexler et al., 2017). While most mammalian species express both MyHCs in the heart, the α:β ratio varies between species but is tightly controlled within a species. Mice typically express >99% α-MyHC while humans express >90% β-MyHC (Izumo et al., 1987; Krenz and Robbins, 2004; Miyata et al., 2000; Sadayappan et al., 2009). However, β-MyHC levels increase in most types of heart disease, regardless of species (Miyata et al., 2000; Nadal-Ginard and Mahdavi, 1989; Nakao et al., 1997). This is presumably a compensatory response as the slower, more energetically efficient β-MyHC may be more suited to the metabolic stress that typifies many forms of heart disease (Hoyer et al., 2007). However, the accompanying reduction in contractility ultimately renders this a maladaptive long-term response.

Tuning the gene expression profiles in these muscle types occurs through a coordinated transcriptional and post-transcriptional network. miRNA-mediated post-transcriptional regulation is a key event in muscle development: skeletal muscle-specific knockout of Dicer, which prevents all miRNA maturation in this tissue, is embryonic lethal (O'Rourke et al., 2007). Several miRNAs have been shown to modulate muscle fiber type. For example, double knockout of the two miR-133a genes doubles the percentage of type I fibers at the expense of type II fibers in the soleus, suggesting it enforces a fast program (Liu et al., 2011). In addition, miR-27a promotes a fast oxidative (type IIa) phenotype, while miR-499 and miR-208b promote the type I slow phenotype (Chemello et al., 2019; van Rooij et al., 2009). Along with the miR-208 and -499, and the miR-133 family, the miR-1/206 family is also predominantly expressed in muscle (Ma et al., 2015; McCarthy, 2008; Mitchelson, 2015). Interestingly, although miR-206 expression is low in non-muscle tissues, its levels decrease in numerous cancers. Moreover, ectopic miR-206 expression can inhibit tumor cell growth and induce apoptosis, suggesting functional relevance (Nohata et al., 2012). In the basal state, miR-206 is most highly enriched in slow muscle fibers (Boettger et al., 2014; Williams et al., 2009). miR-206 is also upregulated in fast muscle in several pathological settings, including the chronic, progressive damage associated with amyotrophic lateral sclerosis (ALS) and Duchenne muscular dystrophy (DMD), as well as acute chemical injury from cardiotoxin injection (Liu et al., 2012; Williams et al., 2009; Yuasa et al., 2008). In conditions such as spaceflight and aging, which are associated with mixed fiber type muscle shifting towards a faster profile, miR-206 levels decrease (Allen et al., 2009; Hamrick et al., 2010; Kim et al., 2014). Although miR-206 is induced after myocardial infarction (Dong et al., 2009; Limana et al., 2011; Shan et al., 2009), it is unclear whether it is important in the heart at baseline. Notably, despite a general sex difference in muscle fiber types, it is not known whether biological sex influences miR-206 expression. Thus, we sought to determine in both sexes whether the miR-206 slow muscle enrichment is an outcome or a driver of the oxidative phenotype.

We measured miR-206 levels in three mouse lower hindlimb muscles with different fiber type compositions. In both males and females, we found a stepwise increase in miR-206 expression from the fastest muscle [tibialis anterior (TA)] to a mixed fiber type muscle [gastrocnemius and plantaris (GP)] to the slowest muscle [soleus (SOL)]. As shown in Fig. 1A, there was an ∼30-fold enrichment in SOL compared to TA. Interestingly, there were significantly higher miR-206 levels in female TA compared to male TA (Fig. 1A, right panel), which supports our previous observation that the female TA contains a greater fraction of slow fibers than the male TA (Haizlip et al., 2015). miR-206 expression was not sexually dimorphic in either the GP or SOL. There was also no sex difference in the expression of the other myomiRs (miR-1, miR-133b or miR-133a) in the TA (Fig. S1A). Since miR-206 levels can differentially respond to ligands binding to different estrogen receptors in breast cancer cells (Adams et al., 2007), we assessed a potential hormonal basis for the sex difference in the TA by examining miR-206 levels in female aromatase (also known as CYP19A1) knockout (ArKO) mice, which do not produce any estrogen. In ArKO TA, we observed a significant 1.3-fold reduction in miR-206 levels (Fig. 1B), while miR-1, -133b, and -133a were unaffected (Fig. S1B), suggesting a miR-206-specific phenomenon. miR-206 levels were not altered in either the ArKO GP or SOL (Fig. S1C). In light of the lower miR-206 levels in the female ArKO TA, we considered whether there might be a corresponding shift in myosin heavy chain (MyHC) expression to suggest a shift towards a faster twitch phenotype. The adult MyHCs can be ranked from slowest to fastest with respect to ATPase activity: β (type I slow), IIa, IIx and IIb (Resnicow et al., 2010). While β-MyHC mRNA is not abundant enough in the TA to be reliably interpreted, we found 4-fold and 1.7-fold reductions in MyHC IIa and IIx expression, respectively, with no change in MyHC IIb mRNA levels (Fig. 1C). Thus, the slower the type II myosin, the greater the decrease in the female ArKO TA.

As it is well known that endurance exercise induces a shift towards a slow twitch phenotype (Allen et al., 2001; Andersen and Henriksson, 1977), we examined miR-206 expression in the TAs of male WT mice during adaptation to voluntary wheel running. miR-206 levels increased 5.3-fold after 3 days of exercise but were not different from sedentary levels after 7 or 14 days of running (Fig. 1D). Hindlimb suspension, a model of muscle disuse that can induce a fast-twitch shift, did not result in changes in miR-206 expression (Fig. 1D).

We next evaluated whether miR-206 expression responds to pathological muscle stimuli that shift typical fast muscles to a slower phenotype. One characteristic of Duchenne muscular dystrophy (DMD) is selective sparing of slow muscle and disproportionate loss of fast muscle (Webster et al., 1988). In addition, the damage-induced regeneration in the earlier phases of the disease is accompanied by re-expression of embryonic MyHC, one of the slowest sarcomeric myosins (Ciciliot and Schiaffino, 2010; Resnicow et al., 2010). In agreement with both of these observations, in the mdx4cv mouse model of DMD, we saw induction of miR-206 in both TA and GP with no change in the SOL (Fig. 2A), which is in agreement with previous reports (Guess et al., 2015; Liu et al., 2012). Consistent with this, miR-206 was induced in human DMD vastus lateralis biopsies, a normally mixed fiber type thigh muscle (Fig. 2B). Muscle regeneration in response to acute injury also includes a transient period of slow embryonic MyHC (MYH3) re-expression before re-establishing the normal adult muscle phenotype. Fig. S1D,E illustrate this early, marked increase in embryonic MyHC mRNA as well as the initial depletion of the adult fast MyHC, IIx, followed by recovery of both of their levels by day 14.

We also found a stepwise increase in miR-206 expression during a 2-week regeneration time course in BaCl₂-injured TA (Fig. 2C). This increase seems likely due to transcriptional activation, as we observed a robust induction of the primary miR-206 transcript that preceded peak levels of the mature miRNA and then decreased towards basal levels as regeneration proceeded (Fig. 2D). We assessed transcriptional induction directly and in vivo by injecting a luciferase reporter gene controlled by a 200-bp upstream miR-206 enhancer and a minimal promoter (Fig. S2A) into GP that we subsequently injured with BaCl₂. During the regeneration time course, we saw a peak in luciferase activity at 3 days post-injury, consistent with the peak in endogenous pri-miR-206 levels (Fig. 2E). This behavior was dependent on the miR-206 enhancer, as there was no injury response from a construct with just a basal promoter driving luciferase (Fig. 2E; Fig. S2D). We also saw that this enhancer region responded to endogenous myogenic cues in differentiating C2C12 mouse myoblasts and in 10T1/2 mouse fibroblasts transfected with each of the four muscle regulatory factors (MyoD, myogenin, MRF4 and Myf5; Fig. S2B,C). MyoD is known to bind this region (Rao et al., 2006; Sweetman et al., 2008), and there are five E-boxes with the generic consensus CANNTG (Fig. S2A). A more refined muscle consensus site has been proposed that has a more specific core hexamer [CA(C/G)(C/G)TG] and extended flanking sequence (Yutzey and Konieczny, 1992). We found that the first, second and fifth E-boxes conform well to this muscle consensus but the third and fourth do not (Fig. S2E). In agreement, only the muscle consensus E-boxes are conserved in the human genome, and mutation of only these E-boxes abolished the reporter response to C2C12 differentiation (Fig. S2F). Moreover, a reporter construct with the orthologous human region is also induced with mouse C2C12 differentiation cues, buttressing the assertion that only the three conserved E-boxes are bona fide sites of transcriptional activation (data not shown). This region has also been reported to function as a transcriptional enhancer dependent on the same E-boxes in neonatal rat cardiomyocytes (Yang et al., 2015).

We next investigated whether miR-206 expression is an outcome or a driver of the slow muscle phenotype by examining the miR-206 knockout mouse (206KO) (Williams et al., 2009). In this mouse, the miR-206 precursor stem loop is specifically deleted without impacting the downstream miRNA, miR-133b, or exons of the long noncoding RNA, linc-MD1. In the original work from the Olson group, young 206KO mice were found to be grossly normal, so we examined 6–7-month-old mice to assess whether age would reveal any phenotypes. We found a male-specific increase in SOL mass while the faster TA and GP muscles were unaffected (Fig. 3A; Fig. S3A,B). This effect appears to be specific to lack of miR-206, as we did not observe any compensatory change in the expression of the closely related miR-1 nor any perturbation in the expression of the genetically linked miR-133b (Fig. S3D) or in linc-MD1, the noncoding RNA also generated from this locus (Fig. S3E).

We further characterized the phenotype of the 206KO SOL through cross-sectional area (CSA) analysis and MyHC-based fiber typing. We stained cross-sections with antibodies specific to slow type I (β-MyHC), intermediate type IIa and fast type IIx, as well as laminin, to reveal fiber borders (representative images are shown in Fig. 3B). We measured fiber type-specific CSA and calculated the percentage of fibers that were distributed in 200 µm² bins for both WT and 206KO males. Heatmaps display these data in Fig. 3C. We observed that 206KO type I fibers are smaller while 206KO type IIA fibers have a broader size distribution when compared to WT. While IIX fibers are a rarer fiber type in the SOL, there was a noticeable shift towards larger fibers in 206KO mice. In all fiber types, the largest size bins are only populated in the 206KO genotype. In analyzing proportions of fiber types, we calculated a 25% decrease in the proportion of type I fibers and a 14% increase in the proportion of type IIA fibers in 206KO mice. We also observed a striking increase in the fraction of fast IIX fibers in 206KO mice. While this was not statistically significant due to inter-animal variability, it is noteworthy (Fig. 3D). Corroborating these fiber typing data, we saw a significant 2.5-fold increase in MyHC IIa and a 3.4-fold increase in MyHC IIx mRNA (Fig. 3E). MyHC IIb was essentially undetectable in the SOL of either genotype (data not shown). Consistent with unchanged female muscle masses, we found no change in MyHC expression in female 206KO SOL (Fig. S3C). While we cannot rule out the possibility of increased total fiber number to account for the greater SOL mass in 206KO males, increased proportion and size of types IIA and IIX fibers could be contributing factors.

To extend our molecular profiling of 206KO soleus, we measured expression levels of RNAs typically enriched in either slow or fast skeletal muscle (Fig. 4). We considered transcription factors, noncoding RNAs, troponin isoforms and metabolic factors in addition to the MyHC signature described above. Amongst transcription factor-encoding genes, we found a 2-fold upregulation of the slow-associated Mef2c as well as the fast-associated Six1 and Eya1. We found no change in the expression of the slow-associated miRNAs miR-208b and miR-499 but a 2-fold upregulation of the fast-associated long noncoding RNA linc-MYH. Beyond the MyHC isoform content, contractile properties are also tuned by functionally distinct isoforms of other sarcomere components, including troponins. In skeletal muscle, the Tnni2 isoform of troponin I is fast-associated while Tnni1 is slow-associated (Sheng and Jin, 2016). However, we did not see changes in the expression of either isoform. Expression of the slow-associated Ca²⁺-handling factor Serca2a did not change but there was a modest but significant increase in the oxygen-binding protein myoglobin (Mb). When we consider these data in conjunction with the MyHC analysis, we see an overall shift towards inducing a fast oxidative muscle phenotype in the absence of miR-206, indicating it is likely to be an enforcer of the slow program.

In a healthy rodent heart, miR-206 levels are typically quite low (Fig. S4A) (Boettger et al., 2014; Kim et al., 2006). However, under pathological conditions, including myocardial infarction (MI), hyperglycemia and genetic dilated cardiomyopathy, miR-206 is induced (Dong et al., 2009; Limana et al., 2011; Shan et al., 2009, 2010; Westendorp et al., 2012). Two MyHC isoforms are expressed in the heart: β-MyHC and the faster α-MyHC. One characteristic of multiple forms of heart disease is a shift towards a greater proportion of β-MyHC. To evaluate whether miR-206 induction extends to other pathological conditions in the heart, we treated wild-type male mice with increasing doses of the β-adrenergic receptor agonist isoproterenol, which chemically models pressure overload. Confirming the pathologic impact of the treatment, we saw increases in heart rate, normalized left ventricular (LV) mass, and the mRNAs encoding the pathologic markers atrial and B-type natriuretic factors (ANF and BNP; also known as NPPA and NPPB, respectively) (Fig. S4B–D). We saw a significant dose-responsive increase in β-MyHC mRNA (Fig. 5A) and in miR-206, the latter of which was induced 4-fold in 45 mg kg⁻¹-treated animals compared to vehicle controls (Fig. 5B).

Therefore, we proceeded to morphologically, functionally and molecularly characterize hearts from 206KO animals. In males, we found a significant 11% increase in normalized LV mass (Fig. 6A) with no change in female LVs (Fig. S5A). Similar to what we found with the SOL, we saw no dysregulation of other myomiRs, including miRs-133b, -1, -208a and -208b (Fig. S6A). Using M-mode echocardiography, we observed significant LV anterior wall thinning with no change in LV posterior wall thickness, as well as chamber dilation during systole in male 206KO mice compared to wild-type controls (Fig. 6B). Consistent with this, M-mode calculations revealed a 59% increase in male 206KO LV systolic volume and an accompanying 17% decrease in ejection fraction and 25% decrease in percent fractional shortening, which were all statistically significant (Fig. 6C). Again, we saw no change in female LV dimensions or function (Fig. S5D,E).

We molecularly assessed the male 206KO LVs by quantitative (q)PCR-based analysis of a suite of mRNAs and miRNAs whose expression is frequently perturbed in a variety of cardiac pathologies (Taegtmeyer et al., 2010; Van Rooij et al., 2006). These include mRNAs encoding ANF and BNP, the skeletal muscle isoform of sarcomeric actin (Acta1), the Ca²⁺-handling factors Serca2a and Pln, the fetal cardiac transcription factors Gata4 and Nkx2.5, the Nfat target myocyte-enriched calcineurin-interacting protein (mCip1.4), the pro-fibrotic collagen 1 (Col1a1), and the anti-hypertrophic muscle cytokine myostatin (Mstn). Surprisingly, the only changes we observed were a 1.7-fold increase in Acta1 and a 1.6-fold decrease in Mstn, consistent with cardiac pathology and an increase in LV mass, respectively (Fig. S6B). We also observed no significant dysregulation in ten miRNAs previously identified as part of a signature set of miRNAs whose expression levels respond to three different pathologic cardiac stimuli (Fig. S6C) (Van Rooij et al., 2006). However, when we measured MyHC mRNA levels, we saw a significant 1.4-fold induction of the faster cardiac isoform α-MyHC with a 1.3-fold downward trend in the slower β-MyHC (Fig. 6D). Consistent with unchanged LV size and preserved function, we found no change in α- or β-MyHC expression in female 206KO LV (Fig. S5C). This suggests a shift towards a faster cardiac phenotype with pathological functional consequences in the absence of miR-206 expression.

We show that miR-206 is essential to enforce a slow muscle program in both skeletal muscle and the heart. In healthy skeletal muscle, miR-206 expression is strongly associated with a slow oxidative phenotype and is significantly higher in the female compared to male TA (Fig. 1A). This aligns with our previous observations that the female TA contains a greater proportion of slow fibers, and that females have a greater, but hormone-dependent, capacity for endurance exercise (Haines et al., 2012; Haizlip et al., 2015). There is likely to be a hormonal component to this expression pattern as aromatase-null females express less miR-206 in the TA than wild-type females (Fig. 1B). Interestingly, estrogen can regulate miR-206 expression levels in breast cancer cells, further reinforcing this hormonal link (Adams et al., 2007). We did not see the same sex bias or response to aromatase deletion in expression of the miR-206 family member miR-1 nor in the genetically linked miR-133b, suggesting a specialized role for miR-206 in promoting the slower female phenotype.

The miR-206 KO mouse model studied here was previously shown to have delayed muscle regeneration after cardiotoxin injury, impaired satellite cell differentiation, and delayed neuromuscular reinnervation (Liu et al., 2012; Williams et al., 2009). Curiously, a separate knockout mouse that is lacking not only miR-206 but also miR-133b and a portion of the muscle regulatory long noncoding RNA linc-MD1 (including the sequence encoding a miR-133 sponge) displays no overt phenotype at baseline, after acute muscle injury, or in the context of muscular dystrophy (Boettger et al., 2014). However, miR-206 and miR-133 family members have been shown to play opposing roles, where the former is pro-differentiation and anti-proliferative while the latter is pro-proliferative (Chen et al., 2006; Kim et al., 2006). Moreover, double knockout of miR-133 family members miR-133a-1 and miR-133a-2 (which, collectively, differ only at the terminal 3′ position and so are presumed to target overlapping sets of transcripts) results in an increase in oxidative myofibers (Liu et al., 2011). This also supports opposing roles for miR-206 and the miR-133 family when considered with the increased proportion of fast fiber types in male 206KO SOL that we observed (Fig. 3), potentially complicating interpretation of the miR-206/miR-133b double-knockout (dKO) mouse. Supporting these observations, in Fig. 4 we show higher levels of the key fast muscle transcription factors Six1 and Eya1 (Grifone et al., 2004) as well as the fast-associated lncRNA linc-MYH, which is encoded in the fast MyHC genomic locus, shares a Six1/Eya1-dependent enhancer with these MyHCs and promotes fast-type gene expression (Sakakibara et al., 2014). The Six1 3′ UTR harbors a predicted target site for another slow muscle-specific miRNA family, miR-208/499 (TargetScan.org). Transgenic miR-499 overexpression has been shown to convert all fast fibers in the mouse SOL into slow fibers (van Rooij et al., 2009). We did not observe differential expression of either miRNA in the miR-206 KO SOL, suggesting that miR-206 could be acting downstream of miR-499 and miR-208b in promoting the slow fiber program. We also saw increased Mef2c and myoglobin mRNA levels in the miR-206 KO SOL (Fig. 4), which may initially seem to contradict a fast-type phenotypic switch as expression of these genes is associated with an oxidative profile. However, a recent gene expression analysis of single SOL myofibers revealed that fast oxidative fibers (types IIa and IIx) are characterized by a stronger mitochondrial and electron transport chain signature than Type I, which is consistent with our overall change in gene expression and fiber type profiles (Chemello et al., 2019). In recently published work, a mouse in which miR-206 and both miR-1a loci were knocked out resulted in partial embryonic lethality, but miR-206 single KO mice and two surviving triple KO (tKO) individuals exhibited decreased physical performance, which was supported by impaired mitochondrial function in a tKO cell culture model (Przanowska et al., 2020). Although slow skeletal muscles were not examined in that study, it would be interesting to perform a metabolic analysis of slow oxidative muscles in single, double and triple miR-1/206 family KO mice.

miR-206 has been well-studied in the context of the early stages of myogenesis, particularly in immortalized and primary myoblast cell culture models where it inhibits proliferation and promotes the transition to differentiation by targeting transcripts such as Pola1, Pax7 and Pax3 (reviewed in McCarthy, 2008; Mitchelson, 2015). Although the targeting network through which it enforces a slow muscle phenotype is not yet clear, it notably targets the class II histone deacetylase (HDAC) Hdac4 through translational inhibition (Williams et al., 2009; Winbanks et al., 2011). Hdac4 inhibits slow muscle gene expression and, when knocked out in mice along with other class II HDACs, results in an increased proportion of type I myofibers (Potthoff et al., 2007). This targeting relationship was established in the same miR-206 KO mice studied here when they were examined in the context of ALS (Williams et al., 2009). Interestingly, certain targets, such as Pax7 and Pax3, are subject to alternative polyadenylation in different muscles, leading to different 3′ UTR sequences with differential presentation of miR-206 target sites. Thus the miR-206 targeting network is complex and will need to be considered through the lens of different muscle groups (Boutet et al., 2012; de Morree et al., 2019).

In the heart, we also found that miR-206 promotes a slow phenotype. Although adult rodents express very little β-MyHC in the ventricles, these levels are not inconsequential, and there are distinct regions where β-MyHC predominates (Krenz et al., 2007). It would be interesting to determine whether miR-206 follows the cardiac β-MyHC expression pattern. In addition, the α-MyHC:β-MyHC ratio is strictly regulated. Interestingly, β-MyHC expression is reportedly 4-fold higher in the left ventricles of sexually mature female mice compared to males, and this difference is hormone dependent (Patrizio et al., 2013). We have also previously reported that isolated female rat cardiomyocytes are slower to reach both peak shortening and relaxation (Trexler et al., 2017). These observations are consistent with our comparison of female and male TA muscle. As miR-206 levels are very low in the healthy ventricle, we could not reliably compare male and female expression levels. However, we and others have detected miR-206 in primary isolated rat ventricular myocytes (Yang et al., 2015). We found 5-fold higher expression level in adult rat ventricular myocytes compared to neonatal rat ventricular myocytes (NRVMs) and 15-fold higher expression in NRVMs compared to neonatal rat ventricular fibroblasts, supporting the notion that cardiac miR-206 is expressed specifically in myocytes (data not shown).

If miR-206 is also an enforcer of slow gene expression in the heart, we predict that: (1) in the absence of miR-206, the heart would shift towards a faster phenotype, and (2) miR-206 expression would increase in pathological conditions when the heart shifts towards a slower phenotype. Accordingly, we observed a significant increase in α-MyHC expression and a downward trend in β-MyHC expression in miR-206 KO LVs (Fig. 6D). Supporting the assertion that the small amount of β-MyHC that is expressed in the mouse heart is still critical for normal function, we observed ventricular dilation and systolic dysfunction in miR-206 KO mice (Fig. 6B,C). As in skeletal muscle, this is a male-specific phenomenon, the mechanistic basis for which is an intriguing area for future investigation. Interestingly, transgenic mice overexpressing miR-206 experience cardiac hypertrophy, which, combined with our knockout data, suggests the importance of maintaining miR-206 levels within a narrow range at baseline (Yang et al., 2015). However, miR-1 levels also increased in these transgenic miR-206 mice, and miR-1 overexpression in the heart is known to be detrimental (Zhao et al., 2005).

In support of our second prediction, we observed a dose-dependent increase in miR-206 expression in the left ventricles of wild-type mice treated with the β-adrenergic receptor agonist, isoproterenol. Several reports have also noted increased miR-206 expression after MI in multiple animal models (Dong et al., 2009; Limana et al., 2011; Shan et al., 2009). Both MI and β-adrenergic stimulation are associated with increased β-MyHC expression at the expense of the faster α-MyHC (Harada et al., 1999; Fig. 5A). This may be a compensatory acute stress-adaptation mechanism as β-MyHC is a more energetically efficient motor. This is supported by the finding that adenoviral-mediated overexpression of miR-206 in rat hearts protects from ischemia–reperfusion injury (Zhai et al., 2017). We speculate that miR-206 KO mice would be less able to adapt to cardiac stress as they may not properly induce β-MyHC. Indeed, transgenic mice expressing a miR-206 sponge develop larger infarcts after ischemia–reperfusion injury than wild-type animals (Yang et al., 2015). As this miRNA sponge may also inhibit miR-1, which is the most abundant cardiac miRNA, it will be important to examine the response of miR-206 KO mice to cardiac injury in the future.

Collectively, our data indicate that miR-206 enforces slow muscle gene expression, which ultimately tunes both skeletal and cardiac muscle performance. Moreover, the consistent protective role of miR-206 under pathological conditions in both tissues makes it an attractive target for RNA-based therapeutics.

The minimal promoter (minTATA) and MyoG enhancer firefly luciferase reporter gene constructs were previously described (Cheung et al., 2007). We cloned the miR-206 enhancer (GRCm38/mm10 chr1:20,678,053-20,678,259) in the same manner as described for MyoG. We generated E-box point mutations (CANNTG→CANNTA) with the QuikChange II site-directed mutagenesis kit (Agilent, 200523, Santa Clara, CA) as per the manufacturer's instructions. We verified all clones by Sanger sequencing. MRF expression constructs were a kind gift from Dr Xuedong Liu (University of Colorado Boulder, Boulder, CO). Primer sequences are listed in Table S1.

We grew C2C12 myoblasts (a gift from Dr Helen Blau, Stanford University, Stanford, CA) in growth medium [GM; high-glucose DMEM (Invitrogen, 11960069, Waltham, MA) supplemented with 20% fetal bovine serum, 2 mM L-glutamine, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin, and 1 mM sodium pyruvate]. We differentiated them to myotubes by changing the medium to differentiation medium [DM; high-glucose DMEM supplemented with 5% adult horse serum, 2 mM L-glutamine, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin, and 1 mM sodium pyruvate]. When differentiating, we refreshed DM every day to prevent medium acidification. We grew 10T1/2 cells (ATCC, Manassas, VA) in GM but with 10% FBS. For luciferase assays, we plated cells in triplicate in six-well dishes at a density of 50,000 cells/well (C2C12) or 100,000 cells/well (10T1/2) 24 h before transfection. We repeated each experiment at least twice with independent batches of cells. We routinely tested cells for mycoplasma contamination. We transfected with TransIT-LT1 (Mirus Bio, MIR 2305, Madison, WI) according to the manufacturer's instructions. We co-transfected all firefly luciferase constructs with pRL-TK as a control. When differentiating C2C12 myotubes, we collected day 0 time points at 24 h post-transfection and initiated differentiation at the same time for later time points. We collected 10T1/2 cells at 24 h post transfection. At harvest, we washed cells twice with phosphate-buffered saline (PBS) solution, and stored them in an ultralow freezer in order to process all time points together at the end of the experiment.

We used a Dual-Luciferase Reporter Assay System (Promega, E1960, Madison, WI) according to the manufacturer's instructions with a Turner Designs TD-20/20 luminometer. Briefly, we lysed frozen cells in 1× passive lysis buffer and measured reporter gene activities first in LARII reagent (firefly luciferase) and second in Stop & Glo reagent (Renilla luciferase). Read times for both were 10 s. We compared ratios of Renilla:firefly luminescence.

The University of Colorado Boulder and Colorado State University Institutional Animal Care and Use Committees approved all animal work under protocols 1301.05, 1002.07, and 1002.08. Mice (Mus musculus) were 4-month-old wild-type C57Bl/6J (Jackson Laboratories, 000664, Sacramento, CA), 6–7-month-old miR-206 KO (a kind gift from Dr Eric Olson, University of Texas Southwestern Medical Center and described in Williams et al., 2009), and 4-month-old mdx4cv (a kind gift from Dr Bradley Olwin, University of Colorado Boulder). Aromatase knockout (ArKO) mice were 4 months old, as this age precedes the age-dependent increase in testosterone observed in the ArKO females (Haines et al., 2012). Mice were housed in standard temperature and humidity-controlled environments with a 12-h-light–12-h-dark cycle and offered food and water ad libitum. We euthanized mice by cervical dislocation under inhaled isoflurane. We excised and flash froze all tissues used for molecular analyses in liquid nitrogen. For muscles used for sectioning and staining, we excised and embedded tissues in OCT compound and then froze them in melting isopentane chilled with liquid nitrogen.

We used animals from multiple litters to achieve the pre-determined group sizes and always included littermates. We randomly allocated animals to experimental groups within a genotype. For genetically modified animals, we bred heterozygotes together to ensure littermates of both WT and genetically modified genotypes. Within a strain, we co-housed animals of the same sex to provide equivalent environments between genotypes.

To model endurance exercise, we provided a cage wheel to mice and allowed them to run voluntarily as previously described (Allen et al., 2001). To model muscle unloading, we performed hindlimb suspension as previously described (Allen et al., 2009). We suspended the tail at a 30° angle by taping it to a plastic dowel attached to a swivel apparatus. The forelimbs were still in contact with the cage bottom. We attached the suspension apparatus to a guide wire so mice were able to access all areas of the cage. We euthanized mice and harvested tissues 3, 7 or 14 days after initiating exercise or hindlimb suspension. We compared samples from exercised or hindlimb suspension mice to age-matched normal cage activity controls.

We chronically administered isoproterenol in 1 μM ascorbic acid diluted in sterile saline to C57Bl/6J wild-type male mice via subcutaneous implantation of a miniosmotic pump (model 2001; Alzet, Cupertino, CA). The pumps delivered isoproterenol at a dose of 30 or 45 mg kg⁻¹ body weight per day. Vehicle control pumps contained 1 μM ascorbic acid diluted in sterile saline. We euthanized mice as described above 7 days after pump implantation.

We injured the TA or gastrocnemius with barium chloride as previously described (Guess et al., 2013). Briefly, we anesthetized wild-type C57Bl/6J male mice with 1–4% inhaled isoflurane and shaved the right hind limb. We cleaned the shaved area with 70% ethanol and used a 27-gauge insulin syringe to inject the muscle with 50 μl of 1.2% BaCl₂ dissolved in normal saline. We monitored mice for adverse effects while they recovered on a 37°C heat block.

We performed M-mode transthoracic echocardiography as previously described (Haines et al., 2012). Briefly, we mildly sedated each animal immediately before imaging with 2% inhaled isoflurane. We used a rodent-specific echocardiography machine (Visualsonics, Toronto, Canada) and a 15 MHz linear array probe to image at the midpapillary muscle level. We analyzed images with Echo-Pack software (GE Vingmed, Horten, Norway). Heart rate was the same in all groups of 206KO and WT mice (Fig. S5B). Each measurement per animal represents the average of three consecutive cardiac cycles.

We performed hydrodynamic limb vein (HLV) injections and subsequent in vivo reporter gene activity imaging as previously described (Guess et al., 2013). Briefly, we co-injected the miR-206 enhancer-driven luciferase plasmid or the basal promoter-only (minTATA) control plasmid with the normalization construct p-mKate, which expresses a far red fluorescent protein from a constitutively active promoter. To ensure the plasmids remained only in the hindlimb that was injected but were widely distributed across muscle groups in that limb, we applied a tourniquet around the upper thigh of the anesthetized mouse to restrict blood flow for 1 to 2 min and then injected a large volume {determined by [1+((body weight in grams−25)/25)×1/2] ml} at a rate of 7 ml min⁻¹ with a syringe pump. This method is highly effective for delivering plasmid DNA to muscles with minimal damage (Hagstrom et al., 2004).

Human DMD (kindly provided by Dr Eric Hoffman of George Washington University) and healthy control vastus lateralis (VL) biopsy samples were previously described (Guess et al., 2015). The University of Colorado Human Research Institutional Review Board (IRB) considered the DMD samples exempt from approval according to the University of Colorado Human Research IRB guidelines and approved collection of control biopsies under IRB Protocol 0402.11, August 2002.

We cut 10 µm sections with a cryostat chilled to −20°C from the mid-belly of the soleus and immunostained with standard techniques. Primary antibodies against different MyHC isoforms were produced in-house from hybridoma cell lines purchased from Developmental Studies Hybridoma Bank, Iowa City, IA [MyHC IIa (SC-71; used at 1:2 dilution), MyHC IIx (6H1; used at 1:5 dilution), and β-MyHC (BA-D5; used at 1:50 dilution)]. We used 1:100 dilutions of isotype-specific secondary antibodies [FITC-goat α-mouse IgM (Jackson 115-095-075), Alexa Fluor 568-goat α-mouse IgG1 (Invitrogen A-21124), and Alexa Fluor 647-goat α-mouse IgG2b (Jackson 15-605-207)]. Laminin primary antibody was from Sigma (L-9393, St Louis, MO; used at 1:100 dilution) and secondary was TR-donkey α-rabbit (Jackson 711-075-152; used at 1:100 dilution). We stained DNA with DAPI. We collected all images on a Nikon TE2000 inverted fluorescent microscope with a 20× objective connected to a Nikon DS-QiMc-U3 camera controlled through the NIS-Elements AR software version 4.00.03. We kept exposure times for a given channel constant for all slides. We processed raw images with the Image5D plugin in ImageJ using the same settings for a given channel across all images.

In ImageJ, we manually counted fiber types across entire soleus sections (7 WT and 4 206KO mice) based on MyHC immunostaining. This analysis was double blinded: before analysis, we provided images with non-descriptive, numbered labels to an investigator not involved in the study who then relabeled them and returned them for analysis. After scoring fiber types for each section, we asked for the labeling key in order to reveal genotypes.

We manually measured myofiber cross-sectional area across entire laminin- and MyHC-stained soleus sections (seven WT and four 206KO mice; 4924 total fibers) using ImageJ. This analysis was also double blinded as described in the fiber typing section.

We isolated total RNA with TRI Reagent (MRC, TR 118, Cincinnati, OH) according to the manufacturer's protocol. We homogenized frozen tissue samples directly into ice cold TRI reagent using a dispersion tool (IKA, T10 Basic S1, Wilmington, NC). For pri-miR-206 analysis, we DNase treated the RNA (TURBO DNA-free kit; Thermo Fisher, AM1907, Waltham, MA) as the qPCR primers could not be intron-spanning. We performed all qPCR on an ABI 7500 Fast Real-Time PCR System. We assessed miRNA expression by TaqMan-based qPCR (Thermo Fisher, Part Number 4427975). We reverse transcribed miRNAs with a TaqMan MicroRNA Reverse Transcription Kit (Thermo Fisher, 4366596) and performed qPCR with TaqMan Universal PCR Master Mix, No AmpErase UNG (Thermo Fisher, 4324018), all according to the manufacturer's instructions. We used the ΔΔC_T method to analyze relative expression. We normalized all mouse miRNAs to sno202 expression levels and human miR-206 to U6 RNA. We measured mRNA expression with SYBR Green-based qPCR. All primer sequences are listed in Table S1. We reverse transcribed total RNA with random hexamer primers using the Superscript II Reverse Transcriptase kit (Thermo Fisher, 18064-022). We performed qPCR with SYBR Green PCR Master Mix (Thermo Fisher, 4312704) according to the manufacturer's instructions. We measured relative expression with the Pfaffl standard curve method. In regenerating TA (Fig. 2D, Fig. S1D,E) and isoproterenol-treated heart samples (Fig. 5A; Fig. S4D), we normalized RNA levels to 18S rRNA. In 206KO and WT soleus samples (Figs 3 and 4), we normalized mRNA levels to Acta1. In the 206KO and WT heart samples (Fig. 6D; Figs S5C and S6B), we normalized mRNA levels to Gapdh. Per animal, each data point is the average of technical duplicates.

We graphed and analyzed all data for statistical significance with GraphPad Prism. We analyzed simple two-group comparisons with a Student's t-test. We analyzed multi-group comparisons with ANOVA followed by a Bonferroni post-test or Sidak's multiple comparisons test in the case of CSA analysis (Fig. 3C). P<0.05 was taken as the significance threshold. Other statistical cutoffs are as noted in the figure legends. Data are presented as means with error bars representing the s.e.m. For experiments in which any group size was smaller than five, we present individual points as well as the mean. The n values for each experimental group are noted in the figure legends. To determine sample size for animal experiments, we performed a power analysis based on our preliminary data using G*Power (Faul et al., 2007). We assessed the presence of outliers by determining interquartile ranges and excluded data points more than 1.5× below the first or above the third quartile.

We thank Dr Eric Olson (UT-Southwestern) for generously providing the miR-206 KO mice. We also thank Amy Perry for performing echocardiography measurements, Darrian Bugg, Chris Haines and Dr Alberto Rossi for technical assistance, Drs Emily Pugach, Christa Trexler and Massimo Buvoli for valuable conversations and insight, Dr Sarah Lehman for critical reading and feedback, and the University of Colorado Boulder BioFrontiers Advanced Light Microscopy Core facility for support with fluorescence microscopy.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.243162.reviewer-comments.pdf