The Mechanical Properties of fast and Slow Skeletal Muscles of the Mouse in Relation to their Locomotory Function

Comparative studies of the properties of mammalian muscle in vitro are common, but little has been done to relate these findings to in vivo function. Extensor digitorum longus (EDL) and soleus muscles have often been used experimentally as representative fast-twitch and slow-twitch muscles respectively. Histochemical fibre-typing, using myosin ATPase staining techniques, has typically shown mouse EDL to consist wholly of type IIA and type IIB fibres in roughly equal proportions (Haida et al. 1989; Klueber et al. 1989). In contrast, mouse soleus has been shown to consist of only type I and type IIA fibres in varying proportions from 31 to 41 % type I fibres (Dribin and Simpson, 1977; Lewis et al. 1982; Carnwath and Shotton, 1987; Haida et al. 1989). This variation in percentage of fibre types found in soleus is likely to be caused by the use of different ages, sexes and strains of mice.

Energetic studies have complemented these findings and shown that mouse EDL has far higher basal rates of glycolysis and glyconeogenesis (Bonen et al. 1984, 1990), but a lower basal rate of glycogenesis (Bonen et al. 1990) than soleus. Measurements of heat production by mouse EDL and soleus during and after short tetani have given further indications of the differences between these muscles (Leijendekker and Elzinga, 1990). The results suggested that, during recovery after a tetanic contraction, soleus tends to use fat as a substrate rather than glycogen, whilst EDL uses oxidative phosphorylation supplemented by a low level of glycogenolysis.

Mechanical studies have also highlighted differences in the properties of EDL and soleus. Mouse EDL has a faster twitch and a lower fatigue resistance than soleus (Brooks and Faulkner, 1991). During isovelocity contractions, shortening by 10 % of L₀ (the length at which the muscle produces maximal force in vitro) at a velocity optimal for power generation, EDL produced a higher power output than soleus during the initial shortening phase, 164 W kg^-1 compared with 62 W kg^-1.

Electromyography (EMG) studies on the rat have demonstrated differences between the EDL and soleus in both the total daily time of activation and the activation per stride. EDL motor units are active for 5–22 % of every 24 h period but soleus motor units are active for 22–35 % of the time (Hennig and Lømo, 1985). During locomotion, EMG activity in rat soleus (an ankle extensor) begins just prior to foot contact with the ground and continues until immediately before the foot lifts off again (Nicolopoulos-Stournaras and Iles, 1984). In contrast, EDL (a toe extensor) has an unusual pattern of EMG activity, unlike the general patterns observed in flexors or other extensors. EDL EMG activity in the rat begins after the foot has left the ground and continues until just after the foot touches down again.

Measurements of glycogen depletion have been used to estimate the proportion of a muscle and the proportion of each fibre population within a muscle that are active at different locomotory speeds. Such studies have indicated that, in superficial extensor muscles of the cat, most of the force produced during walking is due to the activity of slow oxidative (SO) fibres, with a minimal contribution from fast oxidative glycolytic (FOG) fibres (Smith et al. 1977). As speed increases to a slow trot, in the cat and the rat, SO and FOG fibres appear to provide the majority of the force produced by the muscle, with a small contribution by fast glycolytic (FG) fibres (Armstrong et al. 1977; Sullivan and Armstrong, 1978). This trend continues with further increases in speed throughout the trot, resulting in FG fibres becoming a more important contributor to the force produced by the muscle. At a slow gallop, SO and FOG fibres again become responsible for the majority of force production, with FG fibres becoming increasingly important as speed increases (Armstrong et al. 1977). Force production by the gastrocnemius muscle has consequently been found to vary greatly in the cat, by up to threefold, over a large range of running speeds (Smith et al. 1977; Walmsley et al. 1978). The soleus appears to be activated to a high degree in the cat, for postural maintenance and throughout all locomotory paces, with little variability in force production (Smith et al. 1977; Walmsley et al. 1978).

In the present study, we compared the isometric properties of mouse EDL and soleus and measured power output using the work loop technique (Josephson, 1985a).

High-speed video recordings of running mice were used as the basis of in vivo muscle and sarcomere length strain determinations. As EMG data for the mouse were unavailable, rat EMG data (Nicolopoulos-Stournaras and Iles, 1984; Roy et al. 1991) were used to determine locomotory muscle activation patterns. The conditions required to produce optimum power output over a range of cycle frequencies in vitro were compared with those believed to be operating in vivo.

Female LACA white mice, between 6 and 8 weeks of age, were killed by cervical dislocation. Extensor digitorum longus (EDL) and soleus muscles were rapidly dissected out in frequently changed, oxygenated Ringer’s solution (composition in mmol l^-1: NaCl, 144; sodium pyruvate, 10; KCl, 6; MgCl₂, 1; NaH₂PO₄, 1; MgSO₄, 1; Hepes, 10; CaCl₂, 2; pH 7.4 at 21 °C) then immersed in fresh oxygenated Ringer’s solution. For mechanical experimentation, muscles were placed in a flow-through chamber containing oxygenated Ringer maintained at 35±0.5 °C (pH 7.2). The muscles were attached using small aluminium foil clips to an isometric force transducer (AME 801, SensoNor, Horten, Norway) at one end and to a servomotor at the other end.

Muscles were supramaximally stimulated with 1 ms pulses via parallel platinum wire electrodes. Muscle length was adjusted to obtain maximal isometric twitch height; this length was defined as L₀ and corresponded to the length yielding maximum tetanic force. Twitch kinetics and the stimulation frequency yielding maximum tetanic force were determined under isometric conditions. The muscle preparation was then subjected to sinusoidal length changes using a constant strain amplitude of ±5 % fibre length, symmetrical about L₀, whilst being phasically stimulated. The stimulation phase shift (phase shift is expressed in degrees, where 0 ° denotes muscle at L₀ and lengthening, and a complete strain cycle is 360 °), and the number of stimuli used was systematically altered at each cycle frequency until the conditions required for maximal power output were established. The muscle was subjected to four oscillatory work cycles per run with a rest of 7 min between runs. A set of control stimulation and oscillation variables, corresponding to those yielding maximal power output from the muscle, were used for every third or fourth run to monitor and allow correction for deterioration in the ability of the muscle to produce power, assuming a linear decline in power output with each run. At low cycle frequencies, where decline in power output was greatest, every other run was used as a control. Changes in force production and displacement of the muscle were controlled, monitored and analysed on-line by a PC microcomputer using in-house software. Work loops were generated to allow determination of muscle power output (Josephson, 1985a; Altringham and Johnston, 1990a).

Preliminary experiments revealed that, for both muscles, a strain amplitude of ±5.0 %L₀ was nearly optimal for power output over the range of cycle frequencies yielding peak power output. This constant strain amplitude was then used whilst the number of stimuli and their phase shift with respect to the strain cycle were varied to determine the conditions required for maximal power output over a range of cycle frequencies. The effects of optimising strain amplitude at different cycle frequencies and the effects of changing starting length (defined as L_s) were also investigated.

These mechanical experiments were then repeated on muscle without the use of electrical stimulation, to determine the importance of work done against passive elements in the muscle.

Muscle length to fibre length ratios were determined for EDL and soleus after fixation, using 4 % formalin (v/v) in mammalian Ringer, at the length which yielded maximal twitch height.

Video recordings were made of running mice, at 200 frames s^-1, to determine the range of stride frequencies and the associated ranges of limb positions used during locomotion. As a comparison, the limb positions used during locomotion in the rat were determined from high-speed cine film (provided by R. McN. Alexander). As the limb positions used during locomotion by the rat and mouse were essentially the same, and as EMG data for the mouse were unavailable, EMG data from the rat (Nicolopoulos-Stournaras and Iles, 1984) were used to determine when, during the stride, the soleus and EDL are likely to be active. This assumes geometric scaling between mouse and rat, a reasonable assumption (Alexander, 1989).

The muscle length change (strain) waveform used during locomotion was initially estimated by placing limbs into the positions observed during video analysis and measuring passive muscle length. Further analysis showed that the muscle length change waveform obtained showed a high degree of correlation (r² values of 0.9 and 0.8 for soleus and EDL respectively) with the change in ankle angle. Ankle angles were determined for at least five pairs of adjacent strides, from two or three different mice, for each gait. Unless stated otherwise, the results are presented as mean ± S.E.M. (number of observations). The ankle angles measured from video analysis were subsequently used to relate sarcomere length to the extreme limb positions used during locomotion, to enable the sarcomere length changes for the limb cycle to be predicted from ankle angle.

The isometric studies demonstrated marked differences in contraction kinetics between soleus and EDL. EDL had much shorter activation and relaxation times than soleus, with both muscles producing a stress of about 230 kN m^-2 (Table 1). Stimulation frequencies for maximum force were 130 Hz for soleus and 200 Hz for EDL, and all stimuli used in the work loop studies were delivered at these rates. Using the work loop technique, at all but the lowest cycle frequencies, work decreased from the first to the second cycle and then remained constant throughout more than 12 successive cycles for EDL and 20 successive cycles for soleus. The third cycle per run was taken to be representative of the muscle’s ability to produce power. Work per cycle produced by soleus at 1 Hz cycle frequency typically decreased by 3.5 % from the first to the second cycle and decreased by 7.0 % from the first to the fourth cycle. At 2 Hz and 4 Hz cycle frequencies, the work per cycle produced by EDL typically decreased by 6.5 % from the first to the second cycle and then remained stable throughout successive cycles. At these low cycle frequencies, the first loop per run was taken as representative of the real capacity of the muscle to produce work prior to the onset of fatigue due to these long tetani.

Muscle preparations were discarded if power output decreased to below 80 % of its starting value. In those experiments where power did decline appreciably, the power output and cycle frequency relationship, loop shape, optimum strain and stimulation variables did not change. The only variables affected were stress and power output. Deterioration of muscle was found to be dependent on the level of work performed in a run, such that runs at low cycle frequency caused a more rapid decrease in the ability of a muscle to produce power.

The number of stimuli for optimum power output decreased with increasing frequency, from 70 at 1 Hz to 2 at 12 Hz for soleus and from 59 at 2 Hz to 2 at 24 Hz for EDL. Phase also decreased over the same frequency ranges: for soleus, from 40 to 25 °; for EDL, from 50 to 30 °. These results are summarised in Fig. 1.

Over a broad range of cycle frequencies using constant strain amplitude (±5 %L₀), EDL consistently produced a higher maximum power than soleus (Fig. 2). Power output was found to be strongly dependent on cycle frequency: with increasing cycle frequency, power ouput rose to a broad peak before it decreased again. Peak power output for EDL was at cycle frequencies of 10–12 Hz, whereas peak power output for soleus was at 4–6 Hz.

Optimal strain amplitude decreased with increasing cycle frequency. At cycle frequencies corresponding to maximal power output, a broad range of strain amplitudes yielded near maximal power output (Fig. 3). At higher cycle frequencies, variation in strain amplitude had a more marked effect on muscle power output, with optimum strain amplitude decreasing to approximately ±4 %L₀ at cycle frequencies of 15 Hz and 10 Hz for EDL and soleus respectively. Strain optimisation resulted in a slight flattening of the power output against cycle frequency curve, relative to that produced using constant strain, but peak power output was produced at the same cycle frequency (Fig. 2).

At low cycle frequencies, power output increased with increasing strain amplitude; however, a point was reached where the strain was so great that power output could not be sustained, owing to muscle damage. This typically occurred at strains above ±10 %. Power output of EDL at 2 Hz cycle frequency was highest at a strain amplitude of ±12 %L₀ (10 % greater than at ±9 %L₀), but this resulted in damage to the muscle and a subsequent drop in power output.

Starting lengths (L_s) greater or less than L₀ resulted in a marked reduction in power output in both EDL and soleus (Fig. 4).

The work done on passive muscle under imposed cyclical length changes was only 1–2.5 % in EDL and less than 4 % in soleus of the work done by active muscle. Passive work done on EDL and soleus only became substantially higher when L_s was increased, the passive work then becoming about 20 % of the work done by active muscle at an L_s 20 % above L₀, or when the strain amplitude was far higher than that producing maximum power output (Fig. 5). The size of the passive loops increased with length change velocity, i.e. with increases in both cycle frequency and strain amplitude.

A stride frequency range of 2.87±0.09 Hz (N=5) for a slow walk to 8.23±0.07 Hz (N=3) for a flat-out gallop was determined for running mice of 23.7±0.76 g body mass (N=9). The trot–gallop transition occurred at a stride frequency of 5.89±0.10 Hz (N=7).

Soleus and EDL were found to have slightly different sarcomere lengths at L₀ of 2.58±0.07 and 2.71±0.04 μm respectively (Fig. 6). The sarcomere length ranges determined for soleus and EDL when active during locomotion were found to be close to the values determined for L₀, 2.28±0.04 to 2.57±0.03 μm and 2.49±0.07 to 2.88±0.09 μm respectively (Fig. 6). The strain waveforms varied slightly with changes in gait, but were essentially the same as those shown for the trot (Fig. 7).

EDL and soleus fibre lengths were determined to be 75 % and 85 %, respectively, of their muscle length (tendon to tendon) measurement, values similar to those reported by Crow and Kushmerick (1982) and Leijendekker and Elzinga (1990).

Soleus and EDL both produced similar maximal isometric stress values of 224 and 233 kN m^-2 respectively; these values are in close agreement with the values of 215 and 219 kN m^-2 found for mice by Leijendekker and Elzinga (1990) and 245 and 237 kN m^-2 determined by Brooks and Faulkner (1991). The twitch kinetics determined for EDL and soleus are also similar to kinematics quoted in the literature. The maximal power outputs determined using the work loop technique (107 and 34 W kg^-1 for EDL and soleus respectively) are lower than those determined from isovelocity releases by Brooks and Faulkner, 1991; 164 and 67 W kg^-1 for EDL and soleus respectively). However, in vivo power output is likely to be overestimated by isovelocity contractions of maximally activated muscle. In vivo, muscles are only active for part of each locomotory cycle and spend about half of the cycle being stretched during the periods of inactivity between the power-generating shortening phases. Muscle power output is determined by the level of work per cycle that a muscle can produce and is dependent on cycle frequency, maximal stress and the force–velocity relationship. EDL would therefore be expected to produce a higher power output than soleus as its faster activation and relaxation times enable it to produce similar levels of work to soleus but at higher cycle frequencies. Fig. 8 shows the maximum power output for EDL and soleus in relation to reported values of maximum power output derived using the work loop technique. The power output values were determined at the normal working temperature of the muscle and at the cycle frequency for which power output was maximal. Three-dimensional plots have been used to illustrate the marked interdependence between temperature and cycle frequency, with a trend towards increased power output with increased temperature and cycle frequency.

The power output of both EDL and soleus was influenced by a number of factors. The number of stimuli for maximum power output decreased with increasing cycle frequency, and the phase shift between length change and onset of stimulation also decreased. Strain amplitude decreased with increasing cycle frequency, but the effects were small at all but the highest and lowest frequencies used: a near optimum power output against cycle frequency curve was obtained at a constant ±5 % strain. The effects of these variables on power output were similar to those observed and discussed in previous studies (e.g. Josephson, 1985a; Altringham and Johnston, 1990a,b).

At high cycle frequencies, the ability of EDL and soleus to maintain force during shortening was reduced as a result of constraints imposed by the force–velocity relationship, resulting in a decrease in the work done per cycle. Shortening deactivation, which is thought to result from a transient decrease in the affinity of troponin C for calcium (Ekelund and Edman, 1982), may also be a contributing factor. Increased cycle frequency therefore results in increased power output until an optimum is reached; thereafter, the decrease in the work done per cycle counteracts the gain made by the increased frequency.

Optimum strain amplitude for both muscles was found to decrease as cycle frequency increased. For a given cycle frequency, decreasing the strain amplitude results in a corresponding reduction in shortening velocity. The deleterious effects of increasing velocity, with increasing cycle frequency, can therefore be counteracted by reducing strain amplitude. However, reducing strain amplitude also reduces work, so there will be an optimum strain at each cycle frequency. These effects of strain are similar to those reported for crab scaphoganite levator muscle (Josephson and Stokes, 1989).

Optimisation of strain amplitude had little effect on the overall shape of the power output against cycle frequency curves, relative to using a constant ±5 %L₀, with the frequency for peak power remaining unchanged. Strain optimisation did, however, increase power output at cycle frequencies above and below those that produced peak power output, causing the curve to become slightly more flattened.

The low level of work done against passive elements was of a similar magnitude to that found in rat gastrocnemius and diaphragm muscles (Heerkens et al. 1987; Syme, 1990). Passive viscoelastic work only became substantial when the starting length of the muscle was increased, work absorbed becoming about 20 % of the work done at an L_s 20 % above L₀, or when the strain amplitude was far higher than that optimal for power output. Work done against passive viscoelastic elements was consequently ignored in any determinations of muscle power output at L₀.

Variation in the starting length (L_s) of soleus and EDL caused a marked reduction in power output (Fig. 4). As L_s was increased above L₀, the passive forces in the muscle increased, as would be predicted from previous reports (Wells, 1965; Altringham and Bottinelli, 1985). With increased passive forces in muscle there is a resultant decrease in power output due to the greater resistance to muscle lengthening (Heerkens et al. 1987). The increases in passive forces were also accompanied by an increase in the contraction time of the muscle, a phenomenon previously observed in crab muscle (Josephson and Stokes, 1989), and a consequent reduction in the work done per cycle.

The maximum work produced by mouse EDL in a single cycle was 32 J kg^-1 using an optimised strain amplitude of ±12 %. Why is there a discrepancy between the observed and the predicted work done per cycle? First, the muscle does not activate and relax instantaneously, so that for part of the work-producing phase of the cycle it is only partially active. Second, at a cycle frequency of 2 Hz and a strain amplitude of ±12 % (the measured optimum strain in our experiments, and the maximum strain Alexander predicts to be used in normal movement), force declines during shortening as a result of the effects of the force–velocity relationship and shortening deactivation. At cycle frequencies lower than 2 Hz, EDL rapidly fatigues, and at lower strain amplitudes work output drops. Finally, maximum stress is only achieved under isometric conditions. Stress decreases with increasing cycle frequency from 87 % of the maximal experimental isometric stress at 2 Hz to 52 % at 24 Hz. The measured highest work value of 32 J kg^-1 can be increased to 42 J kg^-1, by assuming that activation is instantaneous and that force does not decrease during shortening, and to 48 J kg^-1 if isometric stresses are used. If Alexander’s maximal stress value of 300 kN m^-2 is also included, rather than the measured 233 kN m^-2, the theoretical maximal work value of 70 J kg^-1 is reached.

These experiments were carried out on whole soleus and EDL muscles. Whilst these can be described as slow and fast twitch muscles, respectively, neither is homogeneous with respect to fibre type (EDL consisting of type IIA and IIB fibres in roughly equal proportions; soleus consisting of 31–41 % type I fibres, the remainder being type IIA). Preparations of a single fibre type have yet to be isolated from these muscles, and practical considerations make the whole muscle a more appropriate preparation.

Martin et al. (1988) studied the variation in activity of marker enzymes for oxidative (succinate dehydrogenase, SDH) and glycolytic (α-glycerophosphate dehydrogenase, α-GDP) metabolism in the cat tibialis anterior muscle. A comparison was made between whole muscle and single motor unit variation. Interfibre coefficients of variation were calculated, based on the mean and standard deviation, for single motor units; these were 55 % and 81 % (SDH and α-GDP respectively) of those for whole muscle, indicating substantial fibre type heterogeneity within motor units (but see Nemeth et al. 1986, using different enzymes). This observation implies that caution should be exercised in the interpretation of histochemical and biochemical results to determine the functional properties of a muscle. These findings certainly question the need for preparations of a single fibre type in experiments aimed at studying in vivo function.

Variation in starting length, L_s, will change the degree of overlap between thick and thin filaments, altering the range of sarcomere lengths over which the muscle is operating and affecting the ability of the muscle to produce force (Gordon et al. 1966; Altringham and Bottinelli, 1985). The thick filament length of 1.6 μm taken from rat muscle (Page and Huxley, 1963) is likely to be correct because lengths have been found to vary little between frog (Page and Huxley, 1963), tortoise (Page, 1968), rabbit psoas (Page and Huxley, 1963), rhesus monkey and human leg muscles (Walker and Schrodt, 1973). Thin filament lengths are more variable, ranging from 2.64 μm in human EDL, sartorius and gastrocnemius to 2.41 μm in rhesus monkey EDL, sartorius and gastrocnemius (Walker and Schrodt, 1973) and 2.24 μm in rabbit psoas (Page and Huxley, 1963). Page and Huxley (1963) obtained a value of 2.30 μm for rat leg muscle, assuming that actin periodicity is the same in all animals. For the purpose of this study, this value from rat leg muscle was used as it was thought to be the best estimate of thin filament length in mouse EDL and soleus.

The sarcomere lengths measured in this experiment for soleus and EDL at L₀ (Fig. 6) indicate that, during these in vitro studies, EDL is operating at lengths slightly above the plateau of the rat muscle sarcomere length–force curve, but soleus is operating at lengths corresponding to the plateau of the rat muscle sarcomere length–force curve. The length at which EDL produces maximal force corresponds to the length for maximal power output (Fig. 4) and the length at the middle of the in vivo range (Fig. 6). However, although the length at the middle of the in vivo range in soleus does correspond to the length found in vitro for maximal power production, it is 95 % of the in vitro length for maximal force production.

Other studies have also shown that the ranges of sarcomere lengths used during locomotion correspond to the plateau region of the relevant length–force curves; for example, carp muscle during swimming (Rome and Sosnicki, 1991), rabbit leg muscles during the gallop (Dimery, 1985), human thigh muscles during the walk (Cutts, 1989) and bird wing muscles during flight (Cutts, 1986).

It seems highly unlikely that the sarcomere lengths for EDL, when optimised for maximal isometric force production, would correspond to anything other than points on the plateau of the sarcomere length–force curve. It is therefore possible that the actin filaments in the mouse EDL are longer than in rat leg muscle.

In determining in vivo sarcomere length changes, we had to consider the effects of tendon elasticity. A comparison of the dimensions of kangaroo rat tendons with those of kangaroos and wallabies (Biewener et al. 1981) suggests that leg tendon elasticity in small mammals due to the stresses involved in locomotion is low, resulting in tendon length changes of less than 5 %. However, in the present study, precautions were still taken to minimise errors due to tendon extension. For work loop experiments, the aluminium clips were attached as near to the muscle as possible; therefore, any measurements of strain amplitude would involve negligible tendon elasticity error.

James et al. (1995) demonstrate that, when tendon compliance is accounted for in in vitro measurements of sarcomere length, shorter sarcomere lengths are predicted when the tendon is under stress. However, the muscle length change pattern remains largely unaltered and the greatest error in sarcomere length of 0.01 μm (less than 0.13 % of mean sarcomere length) occurs briefly at high stress, late in the muscle shortening phase. Tendon compliance would therefore have no significant effect on the results or interpretation.

The 2.87–8.23 Hz stride frequency range observed in the mouse is comparable to the range of 3.0–8.1 Hz previously reported by Heglund et al. (1974) for a 30 g mouse. Their value of 5.8 Hz for the trot–gallop transition is similar to our observed value of 5.89 Hz.

When in vitro power output is related to in vivo function, a number of important relationships emerge. Stride frequencies below the trot–gallop transition (2.87–5.89 Hz) correspond to the in vitro cycle frequencies over which soleus produces 85–100 % of maximal power output. Over the entire range of stride frequencies observed, 2.87–8.23 Hz, soleus produces between 60 and 100 % of its maximal in vitro power output. Peak muscle efficiency was found to occur in frog sartorius muscle over a range of V/V_max values of 0.23–0.35 (Kushmerick and Davies, 1969; Woledge et al. 1985). The maximum shortening velocities have been determined using after-loaded isotonic contractions to be 4.8 and 10.4 fibrelengths s^-1 for mouse soleus and EDL respectively (Brooks and Faulkner, 1988). Therefore, in vivo soleus operates within the above V/V_max range cited above (V determined as average velocity during shortening) at the trot (Table 2). Brooks and Faulkner (1988) determined that soleus produces maximal power output between the V/V_max values of 0.25 and 0.29, which are also within the range determined for the mouse at the trot.

As running speed increases, faster motor units are gradually recruited, and with increasing speed the pattern of muscle recruitment spreads through the muscle group from the deep soleus outwards, through the plantaris, to the superficial gastrocnemius (Armstrong, 1981). This pattern of activity is reflected by the fibre types of these three muscles in the rat, with a predominance of first fast oxidative glycolytic (FOG) then fast glycolytic (FG) fibres from soleus to gastrocnemius (Armstrong and Phelps, 1984). With increased running speed, the fraction of the stride for which the foot is on the ground decreases so that the feet must exert larger forces on the ground to maintain the average level of force needed to match the body weight (Alexander, 1985). Therefore, with increasing speed, a greater muscle mass would need to be active to produce the necessary force.

The integrated EMG (IEMG) activity of soleus, corresponding to the level of muscle activity, has been shown to decrease with increasing running speed in the rat, whereas the IEMG for the medial gastrocnemius remains relatively constant (Roy et al. 1991). However, when the severity of exercise is increased, by inclining the treadmill, the IEMG of gastrocnemius increases with running speed. It is therefore probable that, as running speed increases, the actions of the plantaris and gastrocnemius become more important because they are likely to produce maximal power at higher stride frequencies than the soleus.

Under conditions corresponding to the upper stride frequency range, EDL works at frequencies approaching those yielding maximal power output. However, the strain waveform determined for EDL in vivo is such that EDL effectively shortens at approximately twice the velocity experienced during a sinusoidal waveform (Fig. 7B). This rapid shortening enables EDL to shorten with a V/V_max range during the trot (0.24–0.39) that would correspond both to the range predicted for maximal efficiency 0.23–0.35 (Kushmerick and Davies, 1969; Woledge et al. 1985), and to the range of 0.24–0.32 determined for maximal in vitro power output for EDL (Brooks and Faulkner, 1988).

Limb positions measured in the mouse during locomotion, from analysis of video recordings, were found to be very similar to those determined for the rat from analysis of high-speed film. It was therefore assumed that rat EMG data would give a reasonable indication of the EMG activity in the mouse during the same gaits.

When the muscle strain waveforms for EDL and soleus during locomotion are related to EMG data (Nicolopoulos-Stournaras and Iles, 1984), both EDL and soleus are found to be active primarily during shortening (Fig. 7). It should be remembered that mechanical activity will be delayed slightly relative to EMG activity. In larger mammals, such as the dog, extensors have been shown typically to undergo active stretch prior to shortening (Tokuriki, 1974; Goslow et al. 1981). This active stretch is thought to be important in the role of elastic storage of energy. However, in small mammals, tendon elasticity has been shown to be low (Biewener et al. 1981), suggesting that elastic storage of energy would be minimal. The soleus, when active, initially shortens, then undergoes a nearly isometric phase before further shortening. This nearly isometric phase corresponds to the period just after touchdown of the limb, when the soleus is likely to be involved in limb stabilisation.

We would like to thank Chris Smith and John Oughton for technical assistance, Professor R. McN. Alexander and Dr I. S. Young for their helpful comments on the manuscript and the SERC for the loan of high-speed video equipment. R.S.J. was supported by an SERC/CASE studentship with the National Heart Research Fund.