Body weight-dependent troponin T alternative splicing is evolutionarily conserved from insects to mammals and is partially impaired in skeletal muscle of obese rats

Skeletal muscles must counteract gravity and adjust their size and/or performance to accommodate variation in body weight. Aside from hypertrophy, the molecular and biochemical mechanisms that muscles use to adjust to changes in body weight are poorly understood, but recent studies in mammals suggest a central role for regulation of sarcomere gene expression (Goldsmith et al., 2010; Kemp et al., 2009). Sarcomere genes encoding proteins that function at the interface between thin and thick filaments are of particular interest, as these play a large role in regulating muscle force output and energy consumption.

One such protein is skeletal muscle troponin T, a component of the troponin complex that regulates muscle contraction and force output (Brotto, 2005; Gomes et al., 2004; Ogut et al., 1999). Mammalian skeletal muscle troponin T is encoded by a slow (Tnnt1) and fast (Tnnt3) gene (Perry, 1998), the latter of which undergoes extensive alternative splicing. Tnnt3 pre-mRNA comprises 18 exons, including a cassette of five alternatively spliced exons near the 5′ end and a mutually exclusive pair of exons near the 3′ end, allowing a possible 128 unique splice forms. Changes in the relative abundance of different Tnnt3 alternative splice forms is likely to be an important component of muscle plasticity, as it affects the calcium sensitivity of contraction in isolated striated muscle fibers (Briggs and Schachat, 1996; Brotto, 2005; Gomes et al., 2004; Ogut et al., 1999; Pan and Potter, 1992).

Alternative splicing of troponin T in mammals has been studied primarily in regard to qualitative shifts in protein isoform expression during early development, and in response to altered muscle use (Medford et al., 1984; Perry, 1998; Stefancsik et al., 2003; Stevens et al., 2003; Yu et al., 2006). A recent study of insect flight muscles demonstrated that there is precise quantitative variation in the relative abundance of alternatively spliced troponin T mRNA transcripts and protein isoforms in response to experimental manipulation of body weight and nutritional state (Marden et al., 2008). Here, we used rats to test the hypothesis that in a mammalian load-bearing (gastrocnemius) muscle, quantitative alternative pre-mRNA splicing of Tnnt3 is similarly regulated in response to natural and experimentally imposed variation in body weight.

In addition to responding to body weight, the troponin T mRNA splice form profile in insect flight muscle is associated with metabolic rate during activity (Marden et al., 2008). This is consistent with the known effects of troponin T isoforms on muscle fiber force output in both insects and mammals (Chandra et al., 2006; MacFarland et al., 2002; Marden et al., 2001; Marden et al., 1999; Nassar et al., 2005), which presumably affect the rate of ATP consumption (Greaser et al., 1988). How variation in troponin T alternative splicing relates to mammalian muscle energetics in vivo is unknown. Hence, we tested the hypothesis that quantitative variation in Tnnt3 splice form abundance in rat gastrocnemius muscle is associated with differences in in vivo energy expenditure.

Mammalian obesity involves extreme weight gain and a possible failure of body weight homeostasis. There are deficits in skeletal muscle function in obesity (Hulens et al., 2001; Jackson et al., 2009; Lafortuna et al., 2005), accompanied by reduced physical activity and energy expenditure (Galgani and Ravussin, 2008; Goldsmith et al., 2010; Larson et al., 1995; Rosenkilde et al., 2010), phenotypes which may further exacerbate the obese etiology. Root causes of such deficits may originate from within the contractile apparatus, but little is known about the way skeletal muscle composition changes as body weight increases during the development of obesity. We therefore tested the hypothesis that the Tnnt3 alternative splicing response to changing body weight is impaired during the development of obesity in the Zucker rat, a genetic model of mammalian obesity.

Male Sprague–Dawley (strain 400) and Zucker rats (Crl:ZUC-Lepr^fa/fa, strain 185 and Crl:ZUC-Lepr^Fa/x, strain 186) were purchased from Charles River Laboratories International, Inc. (Horsham, PA, USA). In all experiments, animals were given ad libitum access to water and a Teklad 8604 Rodent Diet (Harlan Inc., Indianapolis, IN, USA), containing 24% protein and 4.5% fat per total mass. Animals were housed individually in flat-bottomed cages containing soft bedding. To ensure that the animals had full access to food and water, food pellets were placed on the bottom of the cage and extra-long sipper tubes were attached to water bottles. Individual food consumption was monitored throughout these experiments and after recovery from anesthesia, and showed no lasting abnormalities associated with the treatments.

To increase skeletal muscle loading, rats were briefly anesthetized using isoflurane, and fitted with a custom-made vest held in place by Tygon tubing placed around the shoulders and lower abdomen. The vest consisted of a narrow, semi-flexible plate, worn dorsally, to which two elastic pouches, one each side of the spine, were firmly attached by means of Velcro. Control rats carried a vest without weights, or no vest at all. In randomly selected individuals, the pouches were filled with an equal number of lead spheres (∼1 g each) and secured by the Tygon tubing. Weight loads varied from 5 (vest only) to 90 g. Body mass was defined as the post-experiment body mass of animals without weight loads. Total mass was defined as the native body mass plus vest mass. Rats wore vests for 5 days, after which the vest and body mass were measured. The body mass of obese and lean Zucker rats was not experimentally manipulated, and is therefore equivalent to total mass. In all, we collected samples from 50 male Sprague–Dawley, nine lean Zucker and nine obese Zucker rats varying in total mass from approximately 90 to 400 g, with experimental loads comprising 28–36% of body mass. Rats were anesthetized using isoflurane, after which gastrocnemius muscles were dissected and flash frozen in liquid nitrogen until further use. Rats were killed post-dissection. This protocol was approved by the Institutional Animal Care and Use Committee at The Pennsylvania State University College of Medicine.

Weight-loaded and control rats were placed in metabolic chambers for 5 days, under 12 h light/dark regimes. Oxygen consumption () and respiratory exchange ratio (RER) throughout the 5 day period were obtained using an Oxymax open circuit indirect calorimetry system (Columbus Instruments, Columbus, OH, USA), with airflow of 2.5 l min^–1. Data were collected from each chamber for 1 min at 15 min intervals. Ambulatory activity was determined using an Opto M3 system (Columbus Instruments) that measured optical beam breaks in both the horizontal and vertical direction. Mean ambulatory activity was defined as the mean total sequential horizontal beam breaks per 15 min measuring interval. Body composition was measured in a subset of the live Sprague–Dawley and Zucker rats with a Bruker Minispec LF90 NMR Analyzer (Bruker Optics, Inc., Billerica, MA, USA) immediately before gastrocnemius muscle dissection. Briefly, rats were weighed, immobilized in a Plexiglas cylinder, and placed inside the Minispec for approximately 1 min. During this time measurements of whole body fat and lean (muscle) mass content were obtained using manufacturer-recommended rat-specific acquisition parameters.

Total RNA was extracted from gastrocnemius muscle using Trizol reagent (Invitrogen, Carlsbad, CA, USA), and precipitated in isopropanol, according to the manufacturer's instructions. Total RNA was reverse transcribed using a High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA, USA). Tnnt3 amplicons were amplified by PCR using fluorescein (FAM)-labeled forward primer, fTnt_F1 (5′-FAM-CCCCCAACCTTCTCAGACT-3′), and two unlabeled reverse primers, fTnt_R2 (5′-CCTTCTTGCTGTGCTTCTGG-3′) and fTnt_R4 (5′-CGGACAGTCATGATATCGTATTT-3′). Forward primer fTnt_F1 hybridizes to the Tnnt3 5′ UTR, 28 nucleotides upstream of the start codon. The 5′ alternative exon cassette starts 27 nucleotides downstream of the start codon. Reverse primer fTnt_R2 sequence spans constitutive exon 18 and 3′ alternative exon 17, whereas fTnt_R4 spans 3′ alternative exon 16 and constitutive exon 15 (Fig. 1). Using this amplification strategy, all of the possible alternative Tnnt3 mRNA splice form amplicons had a unique size, with a minimum length difference of three nucleotides. PCR was performed using HotStart GoTaq polymerase (Promega, Madison, WI, USA) under the following cycling conditions: 5 min at 95°C, followed by 4 cycles of 30 s at 94°C, 30 s at 65°C (–1.0°C/cycle), followed by 1 min, 15 s at 72°C. This was followed by 29 cycles of 30 s at 94°C, 30 s at 60°C, 1 min and 15 s at 72°C, with a final 15 min at 72°C to end.

For quantitative analyses of Tnnt3 splice form relative abundance in individual muscle samples, FAM-labeled PCR products were diluted 1:25 and 1 μl of this dilution was analyzed by capillary electrophoresis (ABI DNA Analyzer, Applied Biosystems). Any samples with a fragment peak height exceeding the linear detection range of the instrument were further diluted and run again. Relative abundance of each amplicon in the PCR reaction was determined by dividing its peak height by the total of all peak heights (see Fig. 1B). Amplicon fragment size was determined using an internal size standard and Genemapper® (Applied Biosystems) fragment analysis software.

Nucleotide sequences were confirmed by cloning and sequencing each of the uniquely sized amplicons from PCR amplification of cDNA pooled from gastrocnemius muscles of a number of different-sized Sprague–Dawley rats. This amplicon pool was extracted from an agarose gel using a QiaQuick gel extraction kit (Qiagen, Valencia, CA, USA), cloned using a TOPO-TA cloning kit (Invitrogen) and sequenced (ABI Hitachi 3730XL DNA Analyzer, Applied Biosystems).

Myofibrillar protein homogenates were prepared from flash-frozen gastrocnemius muscle samples as described previously (Hartner et al., 1989). The protein concentration of homogenates was determined (Bradford reagent, Bio-Rad, Hercules, CA, USA) and 15 μg of total protein per sample was loaded onto a 14% SDS-PAGE gel (acrylamide/bisacrylamide ratio 30:0.19). Electrophoretically separated proteins were transferred onto PVDF membranes, which were subsequently probed with a polyclonal anti-fast skeletal muscle troponin T antibody (C-18, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). TNNT3 protein bands were visualized using chemiluminescence (ECL Plus™, GE Healthcare, Chalfont St Giles, Bucks, UK) and a GeneGnome imaging system (Syngene, Frederick, MD, USA). Blots were stripped for 30 min at 55°C, in 0.063 mol l^–1 Tris-Hcl, 0.07 mol l^–1 sodium dodecyl sulfate (SDS) and 0.11 mol l^–1 β-mercaptoethanol, pH 6.7, and reprobed with a monoclonal β-actin antibody (Sigma A5060, St Louis, MO, USA). Background-corrected, integrated optical band densities were quantified with Genetools (Syngene) software. Individual TNNT3 protein band abundance was expressed relative to the total abundance of all detected TNNT3 protein bands per sample. We also quantified β-actin protein abundance; this quantity was included in a statistical analysis to verify that covariance between relative abundance of TNNT3 isoforms and body mass occurred independently of variation in other sarcomere proteins and was not an artefact of differences in gel loading (see supplementary material Table S1).

We detected 12 mature Tnnt3 splice forms expressed in rat skeletal muscle (see Fig. 1A for the exon structure of each splice form) by means of RT-PCR screening of total RNA. Sequencing of cDNA clones showed that 9 of the 12 distinct mRNA splice forms contained exon 17 (henceforth referred to as Tnnt3 β1–9), and three contained exon 16 (henceforth referred to as Tnnt3 α1–3; Fig. 1A). These results are consistent with previous findings regarding the Tnnt3 splice forms expressed in adult mammalian skeletal muscle (Briggs and Schachat, 1996), but extend considerably the known number of Tnnt3 mRNA splice forms expressed in adult rodent muscle. To our knowledge, only seven adult Tnnt3 mRNA splice forms have been reported previously (Wang and Jin, 1997).

Weight loads caused pronounced effects on the Tnnt3 mRNA splice form relative abundance, whereas wearing the unweighted vest had little if any effect (Table 1). The first principal component of variation in Tnnt3 splice form relative abundance, characterizing the overall splice form mixture, was significantly related to body mass (F=101.2, P<0.0001) and weight load (F=22.3, P<0.0001). Individually, 8 out of 12 Tnnt3 mRNA splice forms responded to total mass, with body mass and weight load showing highly significant additive effects (Table 1). For example, the Tnnt3 α1 relative abundance in weight-loaded rats was indistinguishable from that in unloaded rats of the same total mass (Fig. 2A). Not all splice forms responded to weight in the same fashion (Fig. 2B), as certain ones (e.g. Tnnt3 β2; Fig. 2B) were completely non-responsive to weight. Tnnt3 β3 showed a mixed response, increasing in relative abundance in rats weighing ∼100–250 g, but with no further increase at higher body mass.

Weight-dependent changes in Tnnt3 expression were also evident at the protein level, albeit less well resolved given the complexity of the isoform mixture. The relationship between body mass and the relative abundance of the two most abundant TNNT3 bands (bands 2 and 4 in Fig. 3A) followed the same pattern as the two most abundant mRNA splice forms (i.e. Tnnt3 β3 and Tnnt3 α1; compare Fig 2 and Fig. 3B). Thus, although we did not fully characterize these protein bands, there were clear body weight-associated shifts in relative TNNT3 band abundance that correlated with the more readily measured changes at the mRNA level.

Changes in Tnnt3 splice form abundance in response to weight loading were not associated with gross alterations in activity or energy expenditure (Fig. 4), as weight-loaded and control rats showed no differences in ambulatory activity, energy expenditure () or RER (i.e. ). Moreover, a multiple regression model showed that ambulatory activity was not significantly associated with either rat size (body mass; F=2.21, P=0.16) or the amount of weight load (F=0.11, P=0.75).

To more closely examine the effect of the Tnnt3 splice form profile on energy expenditure, we controlled for individual differences that would otherwise inflate the error variance and reduce statistical power. In this analysis, the response variable was night-time , the period when rats were most active. We included as independent variables the day-time , to account for individual variation in size and resting metabolic rate, and night-time ambulatory activity to control for activity level (Table 2). The relative abundance of Tnnt3 splice forms had significant independent effects on night-time independently of weight load (Table 2). Rats with muscles containing more Tnnt3 α1 and β3 at a given daytime and weight load expended energy at a higher rate (see directionality of effects in Table 2), as expected if the functional effect of expressing proportionately more of these splice forms increases force and power. Tnnt3 α1 and β3 are among the smaller Tnnt3 splice forms and have identical 5′ exon composition (Fig. 1A), and hence these energetic results are consistent with a previous qualitative finding that unloaded soleus muscles [hindlimb-suspended rats (Yu et al., 2006)] had reduced expression of smaller Tnnt3 isoforms and produced less force (i.e. opposite to the effects we observed in response to weight loading).

Fat loads of obese Zucker rats in relation to lean mass were similar to the exogenously applied loads in the foregoing experiment (ANCOVA post hoc Student's t-test: α=0.05, t=0.96, P=0.34; Fig. 5B) and hence the skeletal muscles were loaded in a comparable fashion across the two experiments (body composition data are presented in Fig. 5A,B, and statistical comparisons in supplementary material Table S2).

Tnnt3 expression in lean Zucker rats showed the same relationship to total mass as we observed in Sprague–Dawley rats (Fig. 5C). In contrast, obese Zucker rats showed an impaired body weight-dependent response in the β-splice forms (Fig. 5C and Table 3), i.e. those that contain the 3′ exon 17, whereas the α-splice forms behaved in the normal fashion. For some of the Tnnt3 β-splice forms, including the most abundant transcript, Tnnt3 β3, obese rats had a splice form relative abundance normally observed in rats of much lower body mass (e.g. 350 g obese rats had a Tnnt3 β3 relative abundance matching that of 150 g lean rats; Fig. 5C). The mismatch between Tnnt3 splice form profile and body mass in obese rats increased as body mass increased (e.g. Tnnt3 β3; Fig. 5C and significant interaction effects in Table 3).

The mismatch between Tnnt3 mRNA splice form profile and body mass in obese rats was accompanied by deviations from the normal troponin T pattern at the protein level (Fig. 6A,B). Large obese Zucker rats showed TNNT3 band patterns that qualitatively resembled those of much smaller Sprague–Dawley and lean Zucker rats (Fig. 6A). Large obese Zucker rats showed an altered relative abundance of TNNT3 bands 2 and 3 compared with large Sprague–Dawley (see also Fig. 3B) and lean Zucker rats (Fig. 6B). For TNNT3 band 1, the pattern shown in Fig. 3B for Sprague–Dawley rats (i.e. decrease with increasing body mass) was not replicated in a second set of Sprague–Dawley and Zucker rats (Fig. 6). The high variability of this band across the two western blots (Fig. 3B, Fig. 6B) may reflect expression of a mRNA splice form(s) that is not consistently related to body mass (e.g. as we saw at the mRNA level for Tnnt3 β2 or Tnnt3 β4, Table 1).

Our findings show that quantitative variation in rat skeletal muscle Tnnt3 splice form abundance is regulated in response to natural and experimental variation in body weight. This response depends on weight per se, rather than on other hidden correlates of body mass, as externally attached weight loads had the same effect as an equal change in body mass due to growth (Fig. 2A). Given the known effect of Tnnt3 protein isoforms on muscle force and sensitivity to activation by calcium (Brotto, 2005; Gomes et al., 2004; Ogut et al., 1999), the function of this molecular change may be to adjust skeletal muscle mechanical performance in response to load, with consequent effects on force output and energy expenditure during activity (e.g. Table 2). Indeed, the overall Tnnt3 splicing adjustments associated with body weight gain observed in this study consisted of increases in the relative abundance of Tnnt3 splice forms that exclude exon 4 (e.g. Tnnt3 α1 and β3), and decreases in those that include exon 4 (e.g. Tnnt3 β5 and β9). Reduced inclusion of exon 4 is known to correlate with increased muscle fiber calcium sensitivity (Briggs and Schachat, 1996; Schachat et al., 1987). Hence, we predict that the overall effect of natural and experimental increases in body weight is an increased calcium sensitivity and force output in gastrocnemius and perhaps other weight-bearing muscles. We are currently testing these predictions.

In obese Zucker rats, this response was impaired, causing a mismatch between body mass and certain components of skeletal muscle Tnnt3 composition that became more pronounced as obesity increased (Fig. 5C, Table 3). This mismatch includes Tnnt3 β3, β4, β5, β6 and β8 (Fig. 5C); transcripts that respond to native and/or applied weight loads (except Tnnt3 β4, see Table 1), and are associated with nocturnal energy expenditure rate (except Tnnt3 β6, see Table 2). For Tnnt3 β3–β5, there were significant interaction effects between body mass and obesity status, indicating that the weight dependency of the relative abundance of these Tnnt3 splice forms changed as body size increased and obesity progressed. Thus, obese Zucker rats appear to partially, yet progressively, lose the ability to homeostatically regulate their skeletal muscle molecular composition, starting at an early point in the development of obesity.

Moreover, from the previous paragraph it follows that the dramatic reduction of Tnnt3 β3 and increase in Tnnt3 β5 mRNA relative abundance (i.e. relatively more exon 4 inclusion in the overall Tnnt3 splice form pool) in larger obese Zucker rats would lead to decreased calcium sensitivity and hence force output by their gastrocnemius muscles.

The resulting mismatch between body weight and skeletal muscle molecular composition (and presumably performance) may be a component of pathologies common in human obesity, such as chronic muscle weakness, increased load-induced muscle injury and unwillingness to participate in exercise (Anandacoomarasamy et al., 2008; Levine et al., 2005).

Quantitative changes in the Tnnt3 isoform profile at the mRNA level were paralleled by qualitative changes at the protein level, in both Sprague–Dawley and Zucker rats, including obese Zucker rats. Achieving a more precise understanding of the relationship between transcript and protein expression of this gene is extremely problematic, as the Tnnt3 α- and β-splice forms differ at the mutually exclusive exon pair near the 3′ end, and each exists with varying combinations of alternative exons near the 5′ end of the gene. Hence, antibodies specific for α and β proteins would hybridize with a mixture of different TNNT3 isoforms. For this reason, we have restricted our protein-level analysis of this gene to qualitative and simple quantitative examination.

Zucker rats become obese at an early age because of a genetic impairment in leptin signaling (Bray, 1977), and therefore it remains to be determined whether the effects observed in this study are specific to this model system or are associated generally with the development of obesity per se. Investigating the Tnnt3 splicing response to the development of obesity caused by a high fat diet in otherwise healthy rats will help to clarify this. Nutritional regulation of pre-mRNA alternative splicing has not been examined in great detail (but see Salati and Amir-Ahmady, 2001; Salati et al., 2004) but, interestingly, in addition to body weight, alternative splicing of insect muscle troponin T is sensitive to nutrition (Marden et al., 2008). Moreover, alternative splicing of the vertebrate insulin receptor gene, a key component of nutrient signaling, is co-regulated with that of troponin T (Ho et al., 2004). Hence, it is possible that signals involving energy homeostasis commonly affect the regulation of troponin T splicing.

Research on mechanisms controlling body weight homeostasis has focused primarily on neuroendocrine regulation of food intake, satiety and energy storage depot size (Levin, 2006; McMinn et al., 2000; Morton et al., 2006; Schwartz et al., 2000). In concert with such mechanisms, our results suggest that there may be an important role for the skeletal musculature in maintaining proper body weight homeostasis. Healthy skeletal muscles apparently receive signals (intracellular and/or extracellular) about the amount of load they experience, and respond by changing their Tnnt3 isoform composition. Elucidating the signaling pathways responsible for this regulatory response may improve our understanding of muscle function and identify targets for therapeutic intervention to reduce muscle pathologies in obesity.

Molecular signaling pathways mediating mechanical load-associated effects on muscle physiology are poorly understood, and have been examined primarily in terms of their involvement in skeletal muscle protein synthesis and growth (e.g. Atherton et al., 2009; Hornberger et al., 2005; Spangenburg, 2009). It is becoming apparent that many alternative splicing events require the activity of the same signaling pathways (Lynch, 2007). For example, the phosphatidylinositol 3-kinase (PI3K)/Akt/mTOR (mammalian target of rapamycin) signaling cascade mediates skeletal muscle protein synthesis in response to nutritional and mechanical stimuli (Anthony et al., 2000; Bolster et al., 2003; Kimball and Jefferson, 2006; Kimball et al., 2000; Sasai et al., 2010), and also activates and interacts with members of the serine/arginine (SR)-rich protein family (Blaustein et al., 2009; Karni et al., 2008; Lynch, 2007), a ubiquitous set of alternative splicing factors that affect exon inclusion/exclusion and facilitate translation of specific mRNAs (Blaustein et al., 2005; Sanford et al., 2005). Modulation of sarcomere gene alternative splicing by nutritional and/or mechanically induced signaling through PI3K/Akt/mTOR would add important functionality to the already complex nature of this pathway, and may mediate the dietary and weight-loading effects on troponin T alternative splicing observed in both insects and mammals.

Impairment of the normal Tnnt3 splicing response in obese rats is restricted to the β-splice forms (see Fig. 5C and Table 3, i.e. those that contain the 3′ exon 17) (Pan and Potter, 1992). This result implies that a particular aspect of the regulation of alternative splicing (i.e. 3′ end exon inclusion) is modified during obesity. This specificity may help to guide future examination of the signaling pathways responsible. Obesity is associated with a chronic inflammatory state mediated by the infiltration of adipose tissue by macrophages (Wisse, 2004). Importantly, inflammation in skeletal muscle impairs splice factor expression (Xiong et al., 2006), and attenuates Akt (Varma et al., 2009) and mTOR signaling (Lang et al., 2007). Indeed, skeletal muscle in rodent models of obesity (including Zucker rats) shows impaired Akt and mTOR signaling (Katta et al., 2009; Kim et al., 2000), which reinforces that idea that these pathways are prime candidates to examine in regard to the impairment in Tnnt3 splicing we observed.

More broadly, these results indicate that skeletal muscle composition is regulated to accommodate performance requirements associated with changing body weight in a fashion that is evolutionarily conserved in animals ranging from insects (Marden et al., 2008) to mammals (illustrated in Fig. 7). Such a response may be necessary because muscle force scales with cross-sectional area whereas gravitational load scales with body mass. For dimensionally similar animals (i.e. proportions of linear dimensions are invariant), increasing muscle mass alone will not maintain a constant ratio of force to body weight as animals grow larger (Biewener, 1989). The ability to modulate performance by changing muscle Tnnt3 isoform composition may at least partially circumvent such geometric constraints.

In summary, our findings provide a new mammalian model system to study the quantitative regulation of alternative splicing, and a quantitative marker for an evolutionarily conserved homeostatic mechanism by which animals respond to body weight. Identifying the body weight-sensitive pathways that regulate Tnnt3 alternative splicing, and the dysregulation thereof during obesity, may reveal new ways to approach the biomedicine of body weight regulation in health and disease.

The authors thank Dr Ronald P. Wilson for help in developing the weight-loading protocol, Dr Howard W. Fescemeyer for help with Tnnt3 splice form cloning and quantification, and Dan A. Brill for help with indirect calorimetry experiments. We also acknowledge the Penn State Genomics Core Facility at University Park, PA, USA, for services provided in Tnnt3 splice form sequencing and quantification.