The effects of muscle starting length on work loop power output of isolated mouse soleus and extensor digitorum longus muscle Free

The development and use of isolated muscle models have not only been integral in advancing the understanding of skeletal muscle mechanics but also been at the forefront of research exploring the effects of disease (Ciapaite et al., 2015; Hessel and Nishikawa, 2017; Tallis et al., 2018), nutra/pharmaceutical agents (Tallis et al., 2015; Hayes et al., 2019; Debruin et al., 2020; Fletcher et al., 2020), climate change and pollution (James et al., 2023), and species diversity and evolution (Altringham and Johnston, 1986; Padilla et al., 2020; Stoehr et al., 2020). Over the years the methodological approaches for the small number of techniques utilised [e.g. isometric studies, force–velocity experiments and the work loop (WL) technique] to assess isolated skeletal muscle contractility have been refined and are typically utilised in a standardised manner (Josephson, 1993; Caiozzo, 2002). Techniques capable of determining skeletal muscle power are considered more relevant for in vivo activities compared with techniques which provide information on skeletal muscle force only (Josephson, 1993; James et al., 1996; Caiozzo, 2002; Rice et al., 2023); it is rare for skeletal muscle–tendon units to act completely isometrically with muscle shortening phases required to perform work and to produce power (Rome, 2002). Traditionally, experiments which utilise techniques to determine muscle power output (PO) will implement the muscle length which evokes peak isometric force as the starting muscle length for dynamic contractions (Askew et al., 1997; Hessel and Nishikawa, 2017; Seebacher et al., 2017; Hill et al., 2019; Padilla et al., 2020; Stoehr et al., 2020; Hinks et al., 2022; Shelley et al., 2022; Bemis and Nishikawa, 2023; Rice et al., 2023). However, contractile parameters that elicit peak force may not be entirely applicable for achieving peak power and the practice of determining peak power from a starting length that reflects maximal isometric force may underestimate the maximal power-producing capacity of the muscle. Despite this understanding, the optimal muscle length for eliciting maximal skeletal muscle PO is rarely determined. This would provide important insight into the application of force–length characteristics to power producing contractions, which could be influenced by fibre-type composition, sarcomere uniformity across the whole tissue and the specific functional range of motion required by the tissue during in vivo activity. Furthermore, understanding the optimal muscle length for power production is essential in the refinement of methods to evaluate isolated skeletal muscle contractile function.

The WL technique is recognised as the in vitro method which comes closest to emulating the dynamic characteristics of in vivo muscle mechanics (Josephson, 1993; James et al., 1995; Caiozzo, 2002) and, as such, has become an increasingly utilised method to evaluate and understand muscle function. The WL technique involves phasic external electrical stimuli being provided to activate the muscle for the onset of shortening (Shadwick and Syme, 2008; Hurst et al., 2019; Padilla et al., 2020), or lengthening (Hessel et al., 2017; Hill et al., 2018), thereby generating concentric (positive) or eccentric (negative) forces during a series of typically symmetrical length change cycles. As muscles do not activate instantaneously, in order to maximise WL power, it is common to implement a muscle-specific phasing of electrical stimulation prior to the muscle reaching its greatest length in the WL cycle so that work is maximised during shortening (Swoap et al., 1997). By incorporating length change cycles and phasic electrical stimuli, the technique ensures that activation and relaxation kinetics are considered, factors which mediate net work production in vivo (Josephson, 1985; James et al., 1995). In contrast, usage of performance data from traditional power assessments such as isotonic or isovelocity experiments assumes that activation and relaxation are instantaneous, and that force during shortening is constant, and does not consider work required for lengthening, leading to overestimates of likely in vivo work and power production (James et al., 1995; 1996). From the outcomes of a WL experiment, a force–length loop can be generated, illustrating the mechanical work performed. Muscle net power (J s⁻¹) can then be derived from the WL by multiplying net work in joules (J; area within the loop) by cycle frequency, which represents the number of length change cycles per second (s⁻¹) (James et al., 1996). Although the WL technique is highly regarded as a method for studying isolated skeletal muscle function, it is important to acknowledge that the protocols utilised are not without limitation (Shelley et al., 2022; Bemis and Nishikawa, 2023; Rice et al., 2023). Thus, refinement to the methodological protocols implemented may improve the in vivo applicability of the data obtained. One important issue is that it is standard practice to utilise the muscle length that yields peak isometric force, referred to as L₀, for all contractile assessments. However, this approach may be suboptimal as the force–length relationship observed during isometric activity may not be directly applicable to dynamic conditions (Rassier et al., 1999).

The force–length relationship of skeletal muscle during isometric force production was originally established over a century ago in frog striated muscles (Blix, 1892) and has subsequently been firmly established (Ramsey and Street, 1940; Gordon et al., 1966). Briefly, as skeletal muscle length increases, isometric force production exhibits an ascending limb, where force increases until peak force is attained and maintained in the plateau region. Subsequent lengthening beyond the plateau range leads to a gradual decline in force production along the descending limb. The length corresponding to peak force production indicates optimal filament overlap, binding site availability and lattice spacing (Rassier et al., 1999; Williams et al., 2013). The principle of the force–length relationship is applicable to dynamic contractions. Whilst shortening velocity will influence muscle force-producing capacity, a length change cycle that is optimised around the plateau region of the force–length relationship will maximise power. However, unlike the force–velocity relationship, which has been systematically evaluated, the force–length relationship during dynamic muscle activity has not been well explored. Measuring PO from the length that elicits maximal isometric force is probably flawed given that this length may not be optimal for maximal power production (James et al., 1995), indicating there may be a shift in the force–length curve for dynamic contractions (Hou, 2018). Furthermore, it could be expected that any muscle-specific variation in the force–length curve (i.e. extended/shortened plateau region and steepness of ascending and descending arms), coupled with functional range of motion of the muscle during in vivo muscle activity may result in differing optimal starting lengths needed to achieve maximal power during dynamic contractions. In support of this, data from experiments utilising cardiac tissue demonstrate starting lengths lower than L₀ are needed to elicit maximum PO (Layland et al., 1995). Despite this strong theoretical basis, there are few research studies which have systematically examined the force–length relationship of skeletal muscle during dynamic contractions. A study by James et al. (1995) suggests that the force–length curve for dynamic contractions is not uniform across muscles varying in phenotype or function, and that the muscle length needed to achieve maximal isometric force may not result in maximal power. However, inferences were made based on small sample sizes (N=3 and 4) across large manipulations of staring length (±10% or ±20% L₀). Thus, whether L₀ corresponds to maximal PO and whether any relationship between skeletal muscle length for force and power production is muscle specific has not yet been systematically established.

The present work aimed to determine the effect of starting length on WL PO of whole isolated mouse soleus (SOL) and extensor digitorum longus (EDL) muscle by adjusting muscle length in small increments (to ±5% and ±10% L₀) across multiple cycle frequencies. Such work systematically investigates the effect of starting length on both fast and slow twitch skeletal muscle power-producing capacity. The present work is not only important in providing insight for furthering the understanding of skeletal muscle mechanics but also essential to the refinement of the WL technique, which has broad application in the understanding of health and disease. We hypothesised that the optimal muscle length for maximal isometric force would not be optimal for power-producing contractions, and that starting length would have significant effects on the WL PO of both the SOL and EDL.

Coventry University approved animal use for this study (P108131). Female C57BL/6J mice (University of Warwick, Warwick, UK), ∼8 weeks of age (mean±s.e.m. body mass 18.5±0.2 g; N=10) were culled via Schedule 1 procedure (cervical dislocation) in accordance with the British Home Office Animals (Scientific Procedures) Act 1986. Whole intact SOL (N=10; muscle mass 6.4±0.2 mg, fibre length 8.2±0.2 mm) and EDL (N=8; muscle mass 7.2±0.2 mg, fibre length 7.1±0.1 mm) muscles were rapidly isolated in cold oxygenated (95% O₂:5% CO₂) Krebs–Henseleit solution (in mmol l⁻¹: NaCl 118; KCl 4.75; MgSO₄ 1.18; NaHCO₃ 24.8; KH₂PO₄ 1.18; glucose 10; CaCl₂ 2.54; pH 7.55 at room temperature). The SOL and EDL were selected to represent muscles of predominantly slow and fast fibre type composition, respectively (Agbulut et al., 2003; Messa et al., 2020), and are frequently used muscles in studies of mammalian isolated skeletal mechanics (Askew et al., 1997; Hessel et al., 2021; Hill et al., 2021). For each muscle preparation, the tendon at the distal end was wrapped in an aluminium T foil clip as close to the muscle as possible, and bone at the proximal end was left intact so that each muscle could be anchored to points of attachment inside an organ bath.

The procedures utilised to measure the contractile properties of isolated mouse SOL and EDL in this study were based on well-established protocols. Experimental parameters for isometric and WL protocols were controlled using custom-written software (Testpoint, CEC) via a D/A board (KPCI3108, Keithley Instruments) on a standard desktop personal computer or laptop. Once tissue was isolated and prepared, it was mounted to a force transducer (UF1, Pioden Controls Ltd) and motor arm (DFG5.0, Solartron Metrology) via crocodile clips inside a custom-built water jacketed organ bath. The organ bath contained continuously circulating oxygenated Krebs–Henseleit solution maintained at 37.0±0.2°C via an external heater/cooler (Grant LTD6G, Grant Instruments, Shepreth, UK), monitored using a digital thermometer (Traceable, Fisherbrand, Fisher Scientific, Loughborough, UK). Once tissue was mounted, it was left to rest for 10 min (Tallis et al., 2022). Following this rest period, the tissue was externally stimulated using 2 ms square-wave pulses (PL320, Thurlby Instruments, Huntington, UK) via platinum electrodes running parallel to the muscle to produce a series of isometric twitch responses. Resting length, followed by stimulation amplitude (typically 12–18 V), was systematically adjusted until maximal twitch force was achieved, determined via a digital storage oscilloscope (2211 or 1002, Tektronix, Marlow, UK). The muscle length for maximal twitch performance was determined as L₀, as is common practice in previous research (Kissane et al., 2018; Debruin et al., 2020; Hill et al., 2020; Hessel et al., 2021). However, it is important to note that previous work has demonstrated that the optimal muscle length for twitch activations is longer than that needed for maximal tetanic activations, known as an activation-dependent shift (Holt and Azizi, 2014; Sundar et al., 2022). The muscle resting length corresponding to L₀ (i.e. the resting length of the muscle prior to activation, which subsequently brings the muscle to L₀) was used as the length that all other starting lengths were adjusted against during the WL protocol. Muscle length at the resting length corresponding to L₀ was measured using an eye piece graticule and 85% and 75% of the physical length recorded for the SOL and EDL, respectively, was used to determine estimated fibre length (James et al., 1995). Following this, stimulation frequency was set to 130 Hz and 230 Hz, and electrical stimuli duration to 350 ms and 250 ms for the SOL and EDL, respectively, to produce maximal tetanic force at L₀ (SOL stress: 232.7±6.8 kN m⁻²; EDL stress: 303.9±16.1 kN m⁻²). Five minutes rest was allowed after each tetanic activation or WL activation (Altringham and Young, 1991).

Once L₀ was determined and tetanic force recorded, PO was measured using the WL technique. Each tissue preparation underwent a series of symmetrical sinusoidal length change waveforms around the muscle's resting length. During the length change cycle, using the stimulation amplitude for maximal twitch activation, electrical stimulation was provided to the tissue prior to the muscle reaching its greatest length to ensure active force output throughout the muscle shortening phase. As activation is not instantaneous, a phase shift in stimulation (time at which stimulation starts prior to shortening) was implemented, which was fixed at −10 ms and −2 ms for the SOL and EDL, respectively; parameters utilised have previously been shown to be optimal for the range of cycle frequencies (number of length change cycles per second; 4, 5 and 6 Hz for SOL, and 8, 10 and 12 Hz for EDL) used in the present study (Hill et al., 2020). The length change was measured using a linear variable displacement transformer (DFG5.0, Solartron Metrology, Bognor Regis, UK). Instantaneous force and length were plotted against each other, forming a WL, to give a visual representation of performance. Instantaneous PO (instantaneous force×instantaneous velocity) values were averaged to generate an average PO for each length change cycle (Van Wassenbergh et al., 2007; Vanhooydonck et al., 2014).

For measures of WL PO, stimulation frequency was increased from that used for measures of tetanic force (130 and 230 Hz) to 160 and 260 Hz for the SOL and EDL, respectively, as previous work indicated that a greater stimulation frequency is required for maximal WL PO (Vassilakos et al., 2009; Shelley et al., 2022). To examine the effects of starting length on WL PO, a range of starting lengths was used (previously determined L₀ and ±5% or ±10% L₀). The order in which starting lengths were performed was randomised, except for the starting length corresponding to L₀, which was used as the control muscle length in all experiments to monitor change in performance during the protocol. Monitoring PO at the starting length corresponding to L₀ during the protocol allows for the correction of PO over time; a linear decline in PO of isolated mouse skeletal muscle has previously been observed (Tallis, 2013) and is proposed to occur through an incremental build-up of an anoxic core over time (Barclay, 2005). As such, all PO were corrected assuming a linear decline in performance between controls (Vassilakos et al., 2009; Hurst et al., 2019; Hill et al., 2020; Shelley et al., 2022). At the starting length corresponding to L₀, digital callipers (Traceable, Fisherbrand, Fisher Scientific) were used to determine the distance between the attachment points of the proximal and distal ends of the muscle. The distance between the attachment points at the starting length corresponding to L₀ was recorded and ±5% or ±10% L₀ was calculated. The distance between the attachment points was then adjusted accordingly for each starting length. At each starting length, cycle frequencies of 4, 5 and 6 Hz for the SOL, and 8, 10 and 12 Hz for the EDL were used, given that previous work indicates that this range of cycle frequencies is appropriate to capture maximal WL PO of these mouse muscles (James et al., 1995). Burst duration (duration of electrical stimuli) and strain (symmetrical length change around starting length, e.g. 0.1 strain results in the muscle lengthening by 5%, shortening by 10% before being re-lengthened by 5% back to starting length) parameters were adjusted to optimise net work produced during the WL. During each assessment of WL PO, muscles were subjected to four WL cycles and data were used from the WL that produced peak power. After WL PO was assessed at each cycle frequency at a specific starting length, the distance between the muscle attachment points was returned to the starting length corresponding to L₀ to monitor the change in performance over time. Every time starting length was adjusted, the estimated fibre length was changed accordingly. Furthermore, at each starting length, passive net work was assessed at 5 Hz and 10 Hz only for the SOL and EDL, respectively, to determine the influence of passive net work on maximal active net work irrespective of starting length.

Once the WL protocol was complete, the tissue was removed from the organ bath and all components (T foil clip, tendons and bone) other than the muscle tissue were discarded. The remaining muscle tissue was blotted to remove excess solution and wet muscle mass was determined (TL-64, Denver Instrument Company, Arvada, CO, USA). Mean muscle cross-sectional area (CSA) was calculated from L₀, muscle mass and an assumed density of 1060 kg m⁻³ (Méndez and Keys, 1960; Brooks and Faulkner, 1991; Tarpey et al., 2018). Isometric stress (kN m⁻²) was calculated by dividing peak tetanic force by mean muscle CSA. WL PO normalised to muscle mass (W kg⁻¹ muscle mass) was calculated as an indicative measure of muscle quality.

Statistical analysis and visualisation of data were performed using JASP (JASP TEAM, 2023; jasp-stats.org) and GraphPad Prism v.9 (GraphPad Software; graphpad.com); all data were normally distributed (checked via histogram using descriptive statistics) with acceptable limits of skewness of ≤±2 observed (Gravetter and Wallnau, 2014) other than passive work (J kg⁻¹ muscle mass) and passive net work as a percentage of maximal active net work of the SOL and EDL. Non-parametric analysis was performed on the skewed passive work datasets. Friedman tests with multiple comparisons were utilised to examine the effect of starting length on passive net work and passive net work as a percentage of maximal active net work of the SOL and EDL at 5 Hz and 10 Hz cycle frequency, respectively. A two-factor repeated measures analysis of variance (ANOVA), with cycle frequency and starting length as a percentage of L₀ as the factors, was used to determine the effect of starting length and cycle frequency on maximum WL PO of the SOL at 4, 5 and 6 Hz cycle frequency and of the EDL at 8, 10 and 12 Hz cycle frequency. Significant main effects and interactions observed for ANOVA were explored using the Holm post hoc test for multiple comparisons, which is considered more powerful than the commonly implemented Bonferroni post hoc test (Aickin and Gensler, 1996). Partial eta squared (η_p²) was calculated to estimate effect sizes for all significant main effects. Thresholds for η_p² effect size were classified as small (<0.05), moderate (0.06–0.137) or large (>0.138) (Cohen, 1988). Cohen's d was calculated to measure effect size and was then corrected for bias using Hedge's g according to the appropriate sample size (Hedges, 1981). Hedge's g effect size was interpreted as trivial (<0.2), small (0.2–0.6), moderate (0.6–1.2) or large (>1.2) (Hopkins et al., 2009). Pearson's rank correlation was performed to assess the relationship between passive net work and WL PO normalised to muscle mass for the EDL. Because of violation of normality, a non-parametric alternative (Spearman's rank correlation) was performed to assess the relationship between passive net work and WL PO normalised to muscle mass for the SOL. Correlations were used to determine whether, irrespective of starting length, the magnitude of passive net work was related to maximal WL PO. All data are presented as means±s.e.m. Statistical power from each ANOVA (β) is also reported. The level of significance was set at P≤0.05.

For SOL WL PO normalised to muscle mass, there were significant main effects of cycle frequency (P<0.001, η_p²=0.763, β=1.00) and starting length relative to L₀ (P<0.001, η_p²=0.744, β=0.999) (Fig. 1A–C). Multiple comparisons indicated that as cycle frequency increased, from 4 to 6 Hz, WL PO decreased (P<0.007). Furthermore, multiple comparisons indicated that WL PO was greatest at −5% L₀ with the PO at the other starting lengths reduced (P<0.015) except for PO achieved at −10% L₀ (P=0.135), but PO from −10% L₀ did not differ from that produced at L₀ (P=0.138). There was no cycle frequency×starting length interaction (P=0.136, η_p²=0.187, β=0.438).

For EDL WL PO normalised to muscle mass, there was a significant main effect of starting length relative to L₀ (P=0.030, η_p²=0.494, β=0.647; Fig. 1D–F). Multiple comparisons indicated that isolated EDL produced the lowest PO at −10% L₀ compared with all other starting lengths (P<0.032), except compared with −5% L₀ (P=0.124), where PO did not differ. The PO for the remaining starting lengths were not statistically different from each other (P>0.163). There was no main effect of cycle frequency (P=0.964, η_p²=0.005, β=0.054; Fig. 1D–F) or cycle frequency×length interaction (P=0.071, η_p²=0.218, β=0.418). Table 1 displays P-values and range of effect sizes for each length comparison.

There was a significant main effect of starting length on passive net work and passive net work as a percentage of maximal net work of the SOL and EDL at 5 Hz and 10 Hz cycle frequency, respectively (P<0.001; Fig. 2). Multiple comparisons indicated that as length increased, passive work and the percentage of passive work relative to maximal active net work increased (P<0.012, d>1.3).

There was a significant relationship between passive work and maximal PO for the SOL (Fig. 3A; Spearman's r=0.709, P<0.001), indicating that as passive work increases, PO decreases. The relationship between passive work and PO is probably due to increased passive work at longer starting lengths contributing to greater negative work during lengthening, thus reducing total net work, coupled with instances where muscle length in the WL cycle goes beyond that which enables optimal filament overlap for maximal force production. However, there was no relationship between passive work and PO for the EDL (Fig. 3B; Pearson's r=0.191, r²=0.04, P=0.184).

Fig. 4 illustrates representative WL shapes obtained for maximal PO from mouse SOL (Fig. 4A–E) and EDL (Fig. 4F–J) at each starting length. The area within the WL represents the net work done during the length change cycle. For the SOL, WL shapes were generally similar, although at starting lengths of L₀ and shorter there appeared to be a greater capacity to produce work during shortening; the area within the loop at −5% L₀ (Fig. 4B) is largest, which corresponds to the length where maximal PO was achieved. WL shapes for the EDL at −10% L₀ were notability different when compared with most starting lengths (except −5% L₀), displaying a reduced ability to produce force during shortening, which corresponds with a reduced PO at −10% L₀ compared with that at starting lengths of L₀ and longer (Fig. 1D–F). Furthermore, the force increase observed towards the end of active shortening in the EDL at lengths of −5% L₀ and longer can be attributed to a reduction in shortening velocity that occurs prior to the initiation of lengthening, resulting in a subsequent increase in force, as explained by the force–velocity relationship. This was not apparent at −10% L₀, which may be a result of the muscle operating further down the ascending limb of the force–length curve. Whilst there also appeared to be a reduced capacity to produce force during shortening at −5% L₀ compared with longer starting lengths, albeit the magnitude of change was much less when compared with that at −10% L₀, statistical analysis of WL PO did not differ between −5% L₀ and other starting lengths. However, it should be noted that despite not reaching statistical significance, moderate to large effect sizes were observed between −5% L₀ and most other starting lengths (expect for +10% L₀) for WL PO.

In line with our hypotheses, our systematic examination of starting length effects on PO of isolated skeletal muscles, slow twitch SOL and fast twitch EDL, revealed starting length had significant effects on WL PO, but effects were not consistent between muscles. The data presented indicate the force–length relationship derived from isometric activations may not be applicable to dynamic conditions in skeletal muscle and that there is muscle-specific variation in the force–length relationship during power-producing muscle actions. For the application of techniques which assess power of isolated skeletal muscle, the current data indicate that SOL starting length should be reduced from the length which evokes maximal isometric twitch force (L₀) in order to achieve maximal WL PO. However, for the EDL, L₀ appeared near-optimal length to achieve maximal PO.

Maximal isometric stress of isolated muscle is an indicative marker of the quality of the preparation. As such, achieving comparable stress values to previous work using the same muscles is integral for the present study in order to be certain that changes in PO of the tissue is due to experimental changes (i.e. manipulating starting length) and not influenced by suboptimal tissue preparations. Whole isolated SOL and EDL produced maximal isometric tetanic stress in the present study of 233±7 and 304±16 kN m⁻², respectively, comparable to previous values of ∼160–259 kN m⁻² in the SOL and ∼251–430 kN m⁻² in the EDL from mice of a similar age (Axell et al., 2006; Gehrig et al., 2008; Widrick et al., 2011; Ammar et al., 2015; Regan et al., 2017; Eshima et al., 2020). Furthermore, SOL and EDL maximal WL PO achieved at L₀ was 49±3 W kg⁻¹ and 138±9 W kg⁻¹, respectively. Whilst these values are greater than previous WL PO of  ∼30–40 W kg⁻¹ and ∼99–105 W kg⁻¹ for the SOL and EDL, respectively (Widrick et al., 2011; Hurst et al., 2019; Hill et al., 2020; Tallis et al., 2022), this is likely to be a result of utilising a more optimal stimulation frequency for maximal power; previous work utilised the stimulation frequency for maximal isometric force for power assessments, which underestimates true maximal WL PO (Vassilakos et al., 2009; Shelley et al., 2022).

Data from the present study demonstrate that the force–length relationship of the SOL derived from isometric conditions cannot be directly applied to dynamic conditions; the starting length needed for maximal isometric SOL force does not evoke peak power production. Whilst inferences have been made to suggest that the force–length relationship derived from isometric conditions may not be directly applicable to dynamic conditions (Rassier et al., 1999), and that shorter starting lengths than L₀ are needed for the SOL to produce maximal WL PO (James et al., 1995), these hypotheses had never been systematically examined. The present study confirms that starting lengths shorter than L₀ are needed to produce maximal WL PO of the SOL, similar to what has been reported for isolated cardiac muscle (Layland et al., 1995; Fletcher et al., 2020).

The reduced starting length (relative to L₀) needed to evoke peak PO could be attributed to several factors. One possible explanation relates to where the muscle is operating on the force–length curve during the length change cycle. At L₀ and longer, when the muscle is initially passively lengthened during the first part of the WL, the muscle may be operating further on the descending limb of the force–length curve for the initial period of activation and shortening. As such, WL PO would be suboptimal as initial active shortening is occurring outside the optimal region for force production (i.e. plateau region). Furthermore, when SOL length is increased to +10% L₀, the magnitude of passive work required to lengthen the muscle reduces total net work. The reduction in net work evoked by negative work during lengthening, coupled with a reduced capacity of the muscle to produce force during shortening as a result of it operating outside the optimal range for force production, is probably a contributing factor to the substantial reduction in PO observed at longer starting lengths. Further increases in starting length beyond +10% L₀ in the SOL would probably exacerbate the decline in PO, as a result of further increases in passive work and suboptimal filament overlap, binding site availability and lattice spacing, supported by previous observations made at +20% L₀ (James et al., 1995). At starting lengths shorter than L₀ (i.e. −5% and −10% L₀), when the muscle is activated, it appears to be operating in a region of the force–length curve that is more optimal for force production, which is largely maintained throughout shortening, supported by the WL shapes (Fig. 4A–E). Whilst the muscle is unlikely to operate in only the plateau region during a length change cycle, for the SOL at −5% L₀ where PO was greatest, the data and WL shapes (Fig. 4B) indicate that the muscle is likely to be operating in the near-optimal range (i.e. around the plateau region) for force production during shortening. Previous estimates, albeit indirect, for mouse SOL suggest that in vivo the muscle operates around the plateau region (James et al., 1995). However, data obtained from human SOL during walking and running suggest the muscle operates predominately on the ascending limb and plateau region (Rubenson et al., 2012). Whilst more direct measures are needed to determine where mouse SOL operates in vivo, if the SOL is operating predominantly on the ascending limb and plateau region, it could explain why a reduced starting length (relative to L₀) is more optimal for power production.

Unlike in the SOL, starting length, and indeed passive force, had little effect on the power-producing capacity of isolated mouse EDL. Muscle-specific differences could be related to differences in the variation of sarcomere length, elastic properties and muscle architecture (i.e. pennation angle and fascicle length) of whole EDL compared with whole SOL.

Historically, the force–length curve was derived from single-fibre experiments (Gordon et al., 1966). However, the sarcomere lengths and architecture (e.g. pennation angle) within regions of the muscle (i.e. proximal, distal and mid-belly) are not uniform (Ahn et al., 2018; De Souza Leite and Rassier, 2020; Tijs et al., 2021) and become more non-uniform during activation (Moo et al., 2017). Lack of sarcomere uniformity and muscle architecture observed in whole tissue probably contributes to deviation between single-fibre and whole-muscle force–length curves (Rassier et al., 1999; Sundar et al., 2022). As such, it could be expected that muscle-specific variation in sarcomere length and architecture could also result in slightly altered force–length curves for each whole muscle. One would expect greater variation in EDL sarcomere uniformity when compared with the SOL, given the EDL has multiple bellies within the muscle which are distinct in many characteristics that are likely to influence the force–length relationship; namely, pennation angle and fascicle length (Chleboun et al., 1997). The SOL has only one muscle belly and thus is comparatively more uniform across the whole muscle (Siedi et al., 2022). Therefore, despite both the EDL and SOL being unipennate with similar mean pennation angle (Burkholder et al., 1994; Charles et al., 2016; Lal et al., 2021), greater variation in EDL architecture and thus sarcomere uniformity could extend and shift the ‘plateau region’ on the force–length curve. It should be further noted that muscle pennation angle may only be similar between the EDL and SOL in juvenile and adult rodents. With increasing age, the SOL has been shown to increase its pennation angle but the EDL remains largely unchanged (Lal et al., 2021). As such, it could be expected that muscle-specific differences in the force–length curve, and indeed differences in optimal starting length, are affected by age. Smaller increments in starting length of the EDL could have little effect on PO as the muscle is producing force in a near-optimal region during the length change cycle as a result of the extended plateau region on the force–length curve. This narrative is supported by the WL shapes, which indicate little difference in the ability to produce force during active shortening until a substantial reduction in starting length (−10% L₀; Fig. 4F). Muscle-specific variation in sarcomere uniformity and architecture is supported by the passive work data, whereby the increase in passive work associated with increasing length is substantially greater in the SOL when compared with the EDL. However, differences in sarcomere uniformity and architecture are unlikely to be the only factors contributing to differences in passive work in response to increasing length, with factors such as myofilament stiffness and the in vivo function of the muscle expected to contribute. For example, SOL passive work may increase more substantially in response to increasing length as the SOL is subject to frequent eccentric actions, particularly during gait (Sasaki et al., 2006).

The present study identifies L₀ as the optimal length to elicit maximal twitch force, which is common across studies examining contractile properties of isolated skeletal muscle (Kissane et al., 2018; Debruin et al., 2020; Hill et al., 2020; Hessel et al., 2021). As such, one factor that could account for differences in optimal muscle length needed to achieve maximal contractile performance between static and dynamic conditions, as demonstrated in the present study, is the activation-dependent shift in optimal muscle length (Askew and Marsh, 1997; Sundar et al., 2022). Whilst the exact mechanisms are still to be elucidated, previous research demonstrates a leftward shift (i.e. reduction) in optimal length needed for maximal tetanic force when compared with twitch activations (Askew and Marsh, 1997; Holt and Azizi, 2014; Cox et al., 2019; Sundar et al., 2022). The activation-dependent leftward shift could aid in explaining the need to reduce starting length in the SOL to elicit maximal PO. However, a leftward activation-dependent shift seen in isometric actions for the EDL (Askew and Marsh, 1997) would not appear to account for results gained for the EDL where L₀ obtained from twitch performance appears near-optimal to achieve maximal WL PO. Given that the distal tendon remained intact for both the SOL and EDL in the present study, differences in the potential influence of the activation-dependent shifts in optimal muscle length between the SOL and EDL may be partly attributable to tendon compliance, which has been highlighted as an integral factor for optimal muscle length needed for force production when muscle is maximally stimulated (Cox et al., 2019). Furthermore, the force produced by muscle relative to its maximum force output will also affect optimal length (Holt and Azizi, 2014). In the present study, during WL cycles, parameters are optimised to produce maximal power. However, force output during WL cycles is lower than would be achieved during maximal tetanic activations, as per the force–velocity relationship. As such, it remains unclear whether the activation-dependent shift in optimal length between twitch and tetanus activations can be directly applied to dynamic contractions.

In summary, there is a preference for a reduced starting length (relative to L₀) to achieve maximal PO of the SOL. Increasing starting length beyond L₀ evokes a decline in PO in part due to elevated passive work and a suboptimal operating region on the force–length curve. However, the opposite is true for the EDL, with longer starting lengths having little effect on PO, probably as a result of the increased plateau region. Furthermore, there is a profound reduction in PO and in the ability to produce force during shortening at starting lengths −10% L₀ in the present study and at −20% L₀ in data presented by James et al. (1995), which indicates a steeper ascending limb in the force–length curve of whole EDL when compared with that of the SOL (Askew and Marsh, 1997). Additionally, where the tissue frequently operates on the force–length curve during dynamic in vivo activity will probably be a determining factor for muscle-specific variation in the force–length curve shape (Maganaris, 2001). For the specific application of the WL technique, the present results suggest that the muscle length for maximal EDL isometric twitch force is near optimal for assessment of WL PO, but for the SOL, starting length should be reduced by 5%.

One limitation of the present study, and indeed most studies utilising the WL technique, is the use of sinusoidal length change waveforms, which are an approximation of unperturbed in vivo length changes. This approach can be limited for a number of reasons. Firstly, dynamic muscle actions often result in complex length changes (Dickinson et al., 2000; Spanjaard et al., 2007; Konow et al., 2020; Rice et al., 2023); there are data to suggest that during running, the muscle length change waveform of mouse EDL, a muscle used in the present study, deviates from a sinusoidal length change waveform (James et al., 1995). Furthermore, sinusoidal length change waveforms are unperturbed length changes, yet during many tasks, across most species, perturbations are frequently encountered during dynamic muscle activity (Sponberg et al., 2023). However, there are a limited number of available technologies which can determine the versatile and complex nature of in vivo skeletal muscle mechanics for assessments of isolated muscle. As such, the use of sinusoidal length change waveforms remains a replicable and robust approach for the assessment of skeletal muscle function, particularly in muscles such as the EDL that are probably primarily utilised for power production during locomotion.

One future direction is to consider the effects of muscle starting length on isolated skeletal muscle fatigue mechanics. Current fatigue protocols often produce a pattern of fatigue atypical of in vivo fatigue mechanics where prolonged activation during muscle relengthening can evoke large negative work components, particularly during the later stages of fatigue protocols (Shelley et al., 2022). A large degree of negative work during in vivo fatigue is unlikely as length change waveforms and magnitude of fibre stimulation are likely to be manipulated to avoid potentially damaging negative work (Wakeling and Rozitis, 2005). Recent work has identified that utilising a contractile assessment-specific stimulation frequency, as opposed to using fixed stimulation parameters, results in a fatigue pattern more applicable to in vivo fatigue (Shelley et al., 2022). Based on the present results, which suggest that starting length should be manipulated, in a contractile mode- and muscle-specific manner, to achieve peak contractile performance, it could be that manipulating starting length for assessment of isolated skeletal muscle fatigue could play a part in producing a more in vivo replicable fatigue response. Previous research suggests that the effects of starting length on force output may be more substantial in submaximally activated muscle (Cox et al., 2019). As such, another area of future research is to extend the present study by examining the effects of starting length during dynamic conditions across a range of activation levels and shortening velocities. As force output is a determining factor of optimal starting length (Holt and Azizi, 2014), it is speculated that optimal length would change depending upon shortening velocity and activation level, factors that determine the force production relative to maximum force output.

In summary, starting length has significant effects on WL PO in a manner which is not consistent across muscles. For the EDL, variation in starting length only evoked a change (reduction) in WL PO when length was −10% L₀, indicating that utilising the starting length needed for maximal EDL isometric force is near optimal for assessments of WL PO. However, the effects of starting length on the SOL were more pronounced, indicating significant changes in WL PO across most starting lengths (except for −10% L₀ versus −5% L₀), despite the relatively small adjustments to starting length. For the SOL, peak maximal WL PO occurred at −5% L₀. As such, the present study identifies that the force–length relationship derived from isometric conditions cannot be directly applied to power-producing muscle actions. Furthermore, we establish that the current approach of implementing the starting length for peak twitch SOL force for the assessment of power may underestimate WL PO. Based on the present data, future work may wish to consider adjusting starting length for each contractile assessment, particularly for whole isolated SOL and other muscles with more complex architecture, in order to achieve maximal contractile performance and improve the quality of the data obtained.

The authors would like to thank the members of The University of Warwick's Biological Services Unit for their support and care of the animals throughout the duration of this study.