Role of contraction duration in inducing fast‐to‐slow contractile and metabolic protein and functional changes in engineered muscle

The role of factors such as frequency, contraction duration and active time in the adaptation to chronic low‐frequency electrical stimulation (CLFS) is widely disputed. In this study we explore the ability of contraction duration (0.6, 6, 60, and 600 sec) to induce a fast‐to‐slow shift in engineered muscle while using a stimulation frequency of 10 Hz and keeping active time constant at 60%. We found that all contraction durations induced similar slowing of time‐to‐peak tension. Despite similar increases in total myosin heavy (MHC) levels with stimulation, increasing contraction duration resulted in progressive decreases in total fast myosin. With contraction durations of 60 and 600 sec, MHC IIx levels decreased and MHC IIa levels increased. All contraction durations resulted in fast‐to‐slow shifts in TnT and TnC but increased both fast and slow TnI levels. Half‐relaxation slowed to a greater extent with contraction durations of 60 and 600 sec despite similar changes in the calcium sequestering proteins calsequestrin and parvalbumin and the calcium uptake protein SERCA. All CLFS groups resulted in greater fatigue resistance than control. Similar increases in GLUT4, mitochondrial enzymes (SDH and ATPsynthase), the fatty acid transporter CPT‐1, and the metabolic regulators PGC‐1α and MEF2 were found with all contraction durations. However, the mitochondrial enzymes cytochrome C and citrate synthase were increased to greater levels with contraction durations of 60 and 600 sec. These results demonstrate that contraction duration plays a pivotal role in dictating the level of CLFS‐induced contractile and metabolic adaptations in tissue‐engineered skeletal muscle. J. Cell. Physiol. 230: 2489–2497, 2015. © 2015 The Authors. Journal of Cellular Physiology Published by Wiley Periodicals, Inc.

Chronic low-frequency stimulation (CLFS) has been shown to induce fast-to-slow fiber type transitions both in vivo (Salmons and Sreter, 1976) and in vitro (Wehrle et al., 1994), as seen in co-ordinated changes in, calcium handling (Green et al., 1984;Hamalainen and Pette, 1997), metabolic (Kernell et al., 1987;Takahashi and Hood, 1993) and myofibrillar proteins (Wehrle et al., 1994;Windisch et al., 1998). Electrical stimulation patterns can differ in firing frequency, contraction duration, rest between contractions and the amount of time the muscle is active (active time) (Westgaard and Lomo, 1988). The role that each of these factors play in inducing a fast or slow phenotype is disputed (Westgaard and Lomo, 1988;Ashley et al., 2008). This is due in part to the differential response to CLFS in different muscle groups (Eken and Gundersen, 1988;Huang et al., 2006) and animal species (Salmons and Sreter, 1976;Eken and Gundersen, 1988;Ashley et al., 2008), showing that no single CLFS pattern exists to transform all muscles into a slow muscle fiber type.
CLFS patterns are based off electromyography (EMG) recordings that have shown that slow muscles have a natural tonic firing frequency of 10-20 Hz and can fire 300,000 impulses per day (Hennig and Lomo, 1985;Eken et al., 2008). During the development of the rat soleus muscle, neural input to the muscle not only increases over time (i.e., an increase in active time towards 60%) but the duration of single contractions also increases (Eken et al., 2008). Despite CLFS patterns attempting to mimic the natural neural input to slow fiber type muscles, some muscles, such as the rat extensor digitorum longus muscle, require a continuous 20 Hz electrical stimulation pattern for more than 60 days to induce myosin shifts (Windisch et al., 1998).
We have previously found that engineered muscle can adapt to an adult-like 60% active time when using 0.6 sec long contractions and that frequency dictates the level of slowing (Khodabukus and Baar, 2014b). However, despite observing functional slow shifts we did not observe changes in MHC isoform. In this paper we sought to determine whether we could engineer a more complete slow phenotype in vitro by increasing the duration of each contraction and rest period while maintaining the 60% active time. We hypothesized that mimicking the neural patterns experienced by developing slow muscles at different stages of development would enable us to find an electrical stimulation protocol that produced fast-to-slow shifts at both the functional and protein level. By varying contraction durations from short, 0.6 sec long embryonic-like, contractions to adult-like, 600 sec long, contractions we found that contraction duration plays a key role in the development of a more complete slow phenotype muscle in vitro.

Materials and Methods 2D Cell culture
The C2C12 myoblast cell line (ATCC) was grown in growth media consisting of high glucose DMEM supplemented with 10% fetal bovine serum (FBS) and 100 Units/ml penicillin until 70% confluent. C2C12 cells were used between passages 6 and 10.

3D Cell culture
Muscles were engineered using fibrin casting as reported previously (Khodabukus and Baar, 2009). Briefly, the muscle constructs were engineered between two 6 mm long silk sutures set 12 mm apart on Sylgard (PDMS)-coated dishes. 500 ml of growth media containing 10 U/ml thrombin, 0.2 mg/ml genipin, and 0.5 mg/ml aprotinin was added to the plate and agitated until it covered the entire surface. Two hundred microliters of 20 mg/ml fibrinogen was added dropwise and the gels were left to polymerize for 1 h before addition of 100,000 cells. Two days after plating cells, the constructs were switched to differentiation media consisting of high-glucose DMEM supplemented with 10% horse serum and penicillin (100 U/ml) for 2 days. Following the second day in differentiation media, the constructs were moved to high-glucose DMEM with 7% FBS and penicillin (100 U/ml) for the remainder of the experiment. We have previously shown that the shift back to 7% FBS maximizes force production compared to 10% horse serum (Khodabukus and Baar, 2009).

Electrical stimulation
Electrical stimulation was performed using a custom made electrical stimulator previously described in detail before (Khodabukus and Baar, 2012). Constructs were differentiated for 7 days and then were initially electrically stimulated for 24 h with an electrical stimulation protocol consisting of a continuous 0.4 sec 10 Hz train followed by a 3.6 sec rest. The constructs were then electrically stimulated for 14 days with the appropriate electrical stimulation protocol (Fig. 1). Once electrical stimulation was started, both the non-stimulated and electrically stimulated constructs were fed every 24 h.

Contractile testing
Functional testing of the C2C12 constructs was performed 14 days after the onset of electrical stimulation as described previously (Dennis et al., 2001). To determine both passive tension and contractile properties of the engineered tissue, one of the anchors was freed from the Sylgard substrate and attached to a custom-made force transducer via one of the minutien pins. After 2 min equilibration, the length of the engineered tissue (L) was set to a baseline length (L 0 ) that generated zero passive tension. Rheobase (R 50 ) and chronaxie (C 50 ) were then determined as described previously (Dennis et al., 2001). Rheobase was calculated as the electric field strength (V/mm) eliciting 50% peak twitch force (P t ) with a 4 ms pulse width. Chronaxie was calculated as the pulse width required to elicit 50% peak force at twice rheobase. Once excitability had been determined all impulses were delivered with a 4 ms pulse width at 4 Â R 50 as described previously (Khodabukus and Baar, 2012). Time-to-peak tension (TPT) and half-relaxation time (1/2RT) were measured in each individual construct a minimum of three times following a single impulse (twitch). Force-frequency (1, 5, 10, 20, and 40 Hz) was then determined with a 1-second train duration and the maximum force recorded designated as peak force. Fatigue was determined by stimulating for 0.75 sec with 0.75 sec rest at 50 Hz for 3 min at four times rheobase with a 4 ms pulse width. A 50 Hz stimulation was used to induce maximal force in both unstimulated and stimulated muscles. If a 10 Hz stimulus were used, the stimulated muscles would be contracting at 93-98% peak force and non-stimulated muscles at 80% peak force. At 50 Hz all fibers are recruited in both groups.
Cross-sectional area was calculated from the measured width of each construct (at its narrowest point), assuming a rectangular cross section and a depth of 500 mm. Specific force was calculated as kilonewtons per square meter: the force generated by the construct (kN) divided by its cross-sectional area (m 2 ).

Western blot
Tissues were washed in ice-cold PBS, and then blot-dried before freezing in liquid nitrogen and storing at À80°C. At the time of processing, samples were powdered in a 1.5 ml microcentrifuge tube on dry ice, suspended in 200 ml ice-cold sucrose lysis buffer (50 mM Tris pH 7.5, 250 mM sucrose, 1 mM EGTA, 1 mM EDTA, 1 mM sodium orthovanadate, 50 mM sodium fluoride, 5 mM Na 2 (PO4) 2 , and 0.1% DTT) and shaken at 1,400 rpm for 1 h at 4°C in an Eppendorf thermomixer (Eppendorf, Hauppauge, NY). The samples were then centrifuged at 4°C for 1 min at 10,000 g to remove insoluble material. The supernatant was transferred to a new tube, and protein concentration was determined using the DC protein assay (Bio-Rad, Berkeley, CA). Equal aliquots of protein in 1Â Laemmli sample buffer were boiled for 5 min before separation on a 10% acrylamide gel by SDS-polyacrylamide gel electrophoresis. After electrophoresis, proteins were transferred to a nitrocellulose membrane (Protran, Whatman, Piscataway, NJ) at 100 V for 1 h. The membrane was blocked for 1 h in 3% milk in Tris-buffered saline þ 0.1% Tween (TBST). Membranes were incubated overnight at 4°C with the appropriate primary antibody in TBST at 1:1,000. The membrane was then washed three times in TBST before incubation for 1 h at room temperature with the  appropriate peroxidase-coupled secondary antibody in TBST at 1:10,000 (Pierce, Rockford, IL). Antibody binding was detected using an enhanced chemiluminescence horseradish peroxidase substrate detection kit (Millipore, Billerica, MA). Imaging and band quantification were carried out using a Chemi Genius Bioimaging Gel Doc System (Syngene, Cambridge, UK). Each of the target protein band was then normalized to the level of tubulin in the same gel. We have previously shown that tubulin levels are not changed following stimulation (Khodabukus and Baar, 2014b). The primary antibodies used in this study were MF20 (detects all MHC isoforms), F59 (detects all fast MHC isoforms), CaF2-5D2 (Fast SERCA), E7 (b-Tubulin), MANDYS13B7 (Dystrophin), N3.36 (Neonatal MHC) (Hybridoma Bank, Iowa), total eEF2, SERCA 2a,

Statistical analysis
Data is presented as means AE S.E.M. The experiments were repeated three times, with an n ¼ 6 for each experiment.  Differences in mean values were compared within groups and significant differences were determined by ANOVA with post-hoc Tukey-Kramer HSD test using Brightstat.com (Stricker, 2008). The significance level was set at (P < 0.05).

Results
To study the role of contraction duration on the response of engineered muscle to chronic low-frequency electrical stimulation (CLFS) we electrically stimulated constructs with a frequency of 10 Hz and an active time of 60% for 2 weeks. We utilized contraction durations of 0.6, 6, 60, and 600 sec which were followed by rest periods of 0.4, 4, 40, and 400 sec respectively, to maintain an active time of 60%. As would be expected, at the onset of stimulation longer contractions resulted in greater levels of fatigue at the end of each contraction (Fig. 1).
Role of contraction duration on active force production and MHC content following 2wk electrical stimulation All contraction durations induced a similar increase in force production compared to non-stimulated constructs (CTL ¼ 0.0008 AE 0.00005 kN/m 2 ; 0.6 sec ¼ 0.0031 AE 0.00015 kN/ m 2 ; 6 sec ¼ 0.0030 AE 0.00022 kN/m 2 ; 60 sec ¼ 0.0031 AE 0.00017 kN/m 2 , 60 sec ¼ 0.0028 AE 0.00023 kN/ m 2 ) ( Fig. 2A). Force-frequency curves for all electrically stimulated muscles were significantly shifted up and to the left compared to control muscles (Fig. 2B). Peak force was achieved at 20 Hz and 40 Hz for electrically stimulated and non-stimulated controls, respectively. Muscles electrically stimulated at 60 and 600 sec produced significantly higher force at 5 and 10 Hz than other electrically stimulated groups (P < 0.05).
CLFS led to a significant increase in both total and neonatal myosin heavy chain (MHC) protein and the myogenic regulatory factor (MRF) proteins Myf-5 and MyoD but not myogenin compared to non-stimulated muscles ( Fig. 2C and  D).

Role of contraction duration on time-to-peak tension following 2wk electrical stimulation
We next looked at the change in time-to-peak tension (TPT) and half-relaxation time (1/2RT), classical markers used to help determine muscle phenotype. Electrical stimulation significantly slowed TPT (Fig. 3A) regardless of contraction duration (CTL ¼ 55.8 AE 1.4 ms; 0.6 sec ¼ 81.3 AE 2.7 ms; 6 sec ¼ 81.0 AE 4.6 ms; 60 sec ¼ 82.8 AE 4.5 ms; 600 sec ¼ 90.6 AE 5.4 ms). In adult muscle, myosin isoform is the main determinant of TPT so we analyzed the relative levels of slow (sMHC) and fast (fMHC) MHC. Relative to total MHC protein content we found a significant decrease in fMHC ( Fig. 3B) with contraction durations of 6 sec and longer (CTL ¼ 1.00 AE 0.04AU; 0.6 sec ¼ 0.96 AE 0.06AU; 6 sec ¼ 0.066 AE 0.02AU; 60 sec ¼ 0.024 AE 0.02 AU; 600 sec ¼ 0.025 AE 0.007 AU). Additionally, relative to total MHC proteins we found a significant increase in IIa MHC and a decrease in IIx MHC protein with contraction lengths of 60 and 600 sec (P < 0.05).

Role of contraction duration on troponin isoform following 2wk electrical stimulation
We next analyzed changes in the slow and fast isoforms of the three troponin proteins, TnC, TnI, and TnT (Fig. 4). Electrical stimulation resulted in an increase in slow and decrease in fast isoforms on TnC and TnT ( Fig. 4A and B). Slow TnT increased to significantly greater levels with contraction durations of 6, 60, and 600 sec compared to 0.6 sec (Fig. 4B). Both the slow and fast isoforms of TnI increased following electrical stimulation, with the fast TnI isoform increasing to significantly greater levels with contraction durations of 6, 60, and 600 sec compared to 0.6 sec ( Fig. 4A and C).
To determine the cause of the shift in 1/2RT we looked at the protein content of the fast specific calcium sequestering protein parvalbumin (Parv) and both the slow and fast isoforms of the calcium sequestering proteins SERCA and calsequestrin (CSQ) (Fig. 5B). Following CLFS the fast CSQ and SERCA isoforms significantly decreased and slow CSQ and SERCA isoforms significantly increased compared to non-stimulated controls. Parvalbumin protein was not detected following electrical stimulation at any contraction duration.
Role of contraction duration on fatigue resistance and metabolic proteins following 2wk electrical stimulation Following 2wk CLFS, fatigue resistance increased greatly in stimulated muscles compared to non-stimulated constructs (CTL ¼ À58.4 AE 3.6 ms; 0.6 sec ¼ À27.8 AE 1.7 ms; 6 sec ¼ À28.7 AE 1.4 ms; 60 sec ¼ À32.1 AE 1.6 ms; 600 sec ¼ À31.9 AE 1.0 ms) (Fig. 6A) with no difference between different contraction durations. To determine whether the change in fatigue resistance was due to changes in metabolic protein content we looked at the levels of glycolytic, mitochondrial, and fatty acid transport and oxidation proteins ( Fig. 6B and C). Phosphofructokinase (PFK) protein was not detected in electrically stimulated constructs but was found in CTL constructs. GLUT4 protein was found to increase with CLFS in a contraction duration independent manner. Citrate synthase (CS), part of the Krebs Cycle, was significantly increased compared to CTL only with contraction durations of 60 and 600 sec. Cytochrome C (Cyt C), part of complex III and IV of the electron transport chain, was significantly increased with all stimulation groups but increased to significantly greater levels with contraction durations of 60 and 600 sec compared to 0.6 sec. Succinate dehydrogenase (SDH), part of the Krebs Cycle and complex II of the electron transport chain and the complex V enzyme ATP synthase (ATPsyn) were found to be significantly increased following electrical stimulation, independent of contraction duration.
The fatty acid transport protein carnitine palmitoyltransferase-1 (CPT-1) and the b-oxidation enzyme b-hydroxyacyl-CoA dehydrogenase (b-HAD) increased following CLFS independent of contraction duration. The expression of the medium (MCAD) and very-long chain Acyl-CoA dehydrogenase (VLCAD) were unchanged following stimulation but the long chain (LCAD) isoform decreased with electrical stimulation.
We also looked at the levels of the metabolic regulators peroxisome proliferator-activated receptor-g coactivator (PGC-1a), myocyte enhancing factor 2 (MEF2) and sirtuin 1 (SIRT1) proteins levels ( Fig. 7A and B). PGC-1a and MEF2 were increased by electrical stimulation regardless of contraction duration, whereas SIRT1 levels were unchanged.

Discussion
In this article, we looked at the effect that contraction duration has on the functional and phenotype shifts induced by CLFS in engineered muscle tissue. Functionally, electrical stimulation increased force production and induced a slow shift in contractile dynamics and greater fatigue resistance compared to control. Increasing contraction duration resulted in a greater slowing of half-relaxation time but not TPT. Myosin shifts required contraction durations of at least 6 sec and were greater with increasing contraction durations.
Increased force generation and fatigue resistance with electrical stimulation demonstrates that the engineered C2C12 constructs can adapt to an adult-like slow phenotype neural pattern. We utilized a 24 h pre-conditioning stimulus of a continuous 10 Hz train of 0.4 sec followed by a 3.6 sec rest that we have used previously (Donnelly et al., 2010;Khodabukus and Baar, 2012), but found that this is not required for the cells to adapt to the adult-like stimulus (data not shown). Myoblasts of different origins have different ranges of plasticity in response to electrical stimulation (Wehrle et al., 1994;Huang et al., 2006). C2C12s originate from mouse thigh muscle which is a mixed phenotype muscle (Blau et al., 1985). The functional changes combined with the changes in myofibrillar, calcium-sequestering, and metabolic proteins suggest that they are a good cell model to study transitions to a slow muscle phenotype.
Contraction durations of at least 6 seconds were required to induce myosin shifts and longer contraction durations resulted in greater myosin shifts. Using the F59 antibody that recognizes all isoforms of fast MHC (Miller et al., 1989) we found a progressive decrease in fast MHC with contractions lengths of 6, 60, and 600 sec. This suggests that longer contractions are needed to shut down the production of fast myosin heavy chain proteins. The loss of F59 signal may also reflect changes in embryonic fast myosin, which would be expected to be high in engineered muscle. However, the progressive decrease in IIx and increase in IIa MHC with increasing contraction duration ( Fig. 3B and C), indicates that adult fast MHC isoforms were shifted to a slower phenotype.
Fast-slow myosin shifts do not occur directly but through progressive shifts in fast MHC isoforms from IIb to IIx to IIa to slow Hamalainen and Pette, 1997). Smaller mammals such as mice and rats (Windisch et al., 1998) are less adaptive in fast-to-slow shifts than larger animals such as rabbits (Salmons and Sreter, 1976;Ashley et al., 2008), particularly when it comes to shifts in myosin isoform. Therefore the lack of a complete MHC shift is not that surprising seeing that it can take up to 60 days of chronic 20 Hz stimulation (i.e., 24 h a day) to see a shift in rat muscle (Windisch et al., 1998;Peuker et al., 1999). It is possible that the stimulation protocol utilized, cell source and/or the 2 week time period is not sufficient to drive a full myosin shift and this will require further study.
We found that TPT slowed to the same degree regardless of contraction duration. Contraction rate closely correlates to MHC isoform in vivo (Close, 1972) but we found no change in MHC isoform when utilizing a 0.6-second train duration, suggesting that MHC was not responsible for the shift in TPT. The troponin complex can also regulate contraction rate and we found clear shifts in TnC and TnT isoform from fast to slow. In contrast, both the slow and fast isoforms of TnI increased with electrical stimulation suggesting that TnI responds to electrical activity in a different manner to TnC and TnT. In vivo, it has been shown that TnI (H€ artner and Leeuw and Pette, 1993) and TNT (H€ artner and ) isoform changes occur more slowly than that of TnC and TnT in response to CLFS. Functionally, TnT is typically the dominant troponin isoform in dictating contractile dynamics (Yu et al., 2007) and our data suggest that TnT may underlie the slowing in TPT observed in this study as we have suggest previously in engineered muscle (Khodabukus and Baar, 2014a). However, changes in calcium uptake can also effect TPT (Schwaller et al., 1999). The specific role of troponin and calcium uptake in dictating contraction rate in engineered muscle requires further study.
We found that CLFS induced a slowing of 1/2RT and that greater slowing occurred with 60 and 600 sec long contractions. Both calsequestrin and SERCA showed a clear isoform shift from fast to slow, parvalbumin, a fast specific calcium sequestering protein (Schwaller et al., 1999), was not detected in any of the stimulation groups. The protein changes in the three calcium sequestering proteins replicate the protein changes seen following CLFS in vivo (Schwarz et al., 1983;Hamalainen and Pette, 1997;Ohlendieck et al., 1999). The shifts in SERCA isoform showed a trend to be greater at 60 and 600 sec but this is unlikely to cause the much greater slowing in half-relaxation time in those groups. Other modifications such as phosphorylation of TnI (Zhang et al., 1995;Roman et al., 2004;Layland et al., 2005) or other regulatory proteins may also play a role in the greater slowing of 1/2RT at 60 and 600 sec and warrants further study.
Electrical stimulation induced improvements in fatigue resistance and changes in metabolic proteins (Fig. 4). PFK, the rate-limiting step of glycolysis (Mor et al., 2011), was decreased following electrical stimulation. In vivo, GLUT4 expression increases relative to the time that muscle is active (Megeney et al., 1993) and we found GLUT4 increased to similar levels with all contraction durations. We found contraction duration-dependent increases in the mitochondrial proteins citrate synthase (CS) and cytochrome C (Cyt C) but not SDH and ATP synthase, which increased independent of contraction duration (Fig. 7). Fatigue resistance (Westgaard and Lomo, 1988) and mitochondrial enzyme activity and protein levels  have been shown to increase with increasing active time following electrical stimulation. As active time was consistent it is likely that the greater fatigue induced by longer contractions (Fig. 1) provided the signal for increased CS and Cyt C. However, the mRNA levels of nuclear respiratory factor 1 (NRF-1), a key regulator of mitochondrial biogenesis (Virbasius and Scarpulla, 1994), only increase during rest periods between stimulation bouts (Nguyen and Hood, 2011) and hint that not only the contraction duration but the rest period regulate metabolic adaptations to electrical stimulation.
Slow fibers have higher oxidative capacity than fast fibers and oxidative capacity increases following CLFS. We found no change in MCAD or VLCAD but surprisingly a decrease in LCAD following electrical stimulation. In contrast, b-HAD which catalyses the third step of b-oxidation was increased with electrical stimulation, suggesting that different mechanisms regulate the levels of different enzymes of the b-oxidation pathway. We have previously found that compared to high glucose (25 mM), culturing in low glucose (5.55 mM) results in higher levels of LCAD and VLCAD (Khodabukus and Baar, 2015). This suggests that either ACAD content does not increase in response to electrical stimulation or that the high glucose used in this study prevented changes in in ACAD content.
Exercise (Baar et al., 2002) and in vitro electrical stimulation-induced mitochondrial biogenesis (Atherton et al., 2005;Burch et al., 2010) is thought to be primarily regulated by PGC-1a. PGC-1a is deemed a master regulator of mitochondrial biogenesis due to its effect on peroxisome proliferator-activated receptors (PPARs) (Aubert et al., 2013), estrogen-related receptors (ERRs) (Willy et al., 2004), NRF-1 and 2 and other proteins involved in mitochondrial biogenesis and itself is thought to play a role in promoting the slow-fiber type program (Rasbach et al., 2010). Following electrical stimulation we found no change in its deacetylase SIRT1 but an approximate twofold increase in PGC-1a and MEF2, an upstream regulator of PGC-1a which is implicated in promoting slow-fiber type (Wu et al., 2000) and mitochondrial biogenesis (Naya et al., 2002) and likely induced some of the metabolic adaptations seen.
Using the C2C12 cell line we have been able to induce a functional contractile and metabolic shift to a slow phenotype using electrical stimulation. While keeping active time and the number of impulses received each day consistent, we have demonstrated that the duration of each contraction and/or rest period play a key role in inducing myofibrillar and metabolic protein shifts.