Muscle Power Output During Escape Responses in an Antarctic Fish

Antarctic fish live in contact with ice and show a range of adaptations to avoid freezing (Clarke and Johnston, 1996). Muscle fibres in Antarctic fish remain excitable at an ambient temperature of −1.86 °C, but how does their performance compare with that of muscle from warm-water species? Some insight into this question can be obtained from measurements of steady-state isometric and isotonic contractions. Muscle fibres from Antarctic, temperate and tropical fish produce similar maximum isometric stresses (F_max) (180–250 kN m⁻²) at their respective normal body temperatures (Johnston and Brill, 1984). In contrast, unloaded shortening speed (V_max) is no faster at low temperatures in cold-water than in warm-water species. The maximum shortening velocity (V_max) for fast fibres is only 1.5–2.5 FL s⁻¹, where FL is fibre length, at 0 °C in Antarctic species compared with 15–20 FL s⁻¹ at 25 °C in Indo-West Pacific species (Johnston and Brill, 1984; Johnson and Johnston, 1991). Studies of the force–velocity relationship of fast muscle fibres indicate that, as a result, muscle power output during isotonic shortening is significantly lower in Antarctic than in tropical fish at their respective body temperatures (Johnston and Altringham, 1985; Johnson and Johnston, 1991). Isometric and isotonic contractions are, however, rarely used during swimming, and it is important to consider muscle properties measured under conditions that are more relevant to natural behaviours.

In order to escape predators or to capture prey, fish use fast-starts. Typically, in an escape response, the body is bent into a C-shape and the subsequent tailbeat(s) results in rapid acceleration from a resting position (Eaton et al. 1977; Webb, 1978). These escape manoeuvres involve large-amplitude body movements, rapid changes in direction and the recruitment of all the fast muscle fibres in the trunk. As a result, muscle strain and activation patterns are complex, varying along the body and from tailbeat to tailbeat (Johnston et al. 1995; van Leeuwen, 1995).

In order to investigate muscle performance during fast-starts in the Antarctic rock cod (Notothenia coriiceps Nybelin), a high-speed ciné camera (500 Hz) was used to film the initial C-bend and the subsequent contralateral contraction. The activation and strain patterns of the superficial layers of fast myotomal muscle fibres were measured simultaneously using sonomicrometry and electromyography, respectively. To investigate muscle power output during the fast-start, we adapted the work loop technique of Josephson (1985) using information from the in vivo recordings. Representative strain waveforms were digitised, and segments corresponding to the initial C-bend and the contralateral contraction were abstracted to create the cyclical events necessary for work loop experiments. This is the first of a series of such studies on polar, temperate and tropical fish designed to test the hypothesis that it is muscle power output, rather than hydromechanical constraints or other factors, that limits fast-start performance, as suggested by Frith and Blake (1995).

Notothenia coriiceps Nybelin (body mass 154±24 g, total length 20.5±3.5 cm; mean ± S.D., N=15) were caught around Signy Island, South Orkneys, Antarctica (60°43′ S, 45°36′ W), transported to St Andrews and maintained in a custom-built cold aquarium at 0±0.5 °C. Fish were fed daily on krill and chopped squid.

Escape responses to tactile stimuli were filmed in silhouette using a 16 mm NAC E10 high-speed ciné camera at 500 frames s⁻¹via a mirror suspended at 45 ° above a glass arena (850 mm×850 mm×100 mm; length×width×depth) illuminated from below. The arena was maintained at 0 °C with a water-jacket containing antifreeze. The film (Ilford HP5) was over-exposed by one stop, developed ‘open trap’, and then viewed and analysed using a motion-analysis system (MOVIAS, NAC, Japan). The positions of the snout and centre of gravity of the fish were used to determine the kinematics of the startle response. Velocities and accelerations were calculated from the raw position data which had been smoothed using a moving average filter with 10 passes (MOVIAS, NAC, Japan).

Anaesthesia was initiated with a 1:5000 (m/v) solution of bicarbonate-buffered MS222 and maintained by irrigation of the gills during surgery, which was performed in a constant-temperature room at 0–1 °C. Piezoelectric crystals (1 mm in diameter) were attached to Perspex implants and calibrated in dissected blocks of muscle at 0 °C. Each pair of crystals was inserted approximately 8 mm apart in the superficial layers of fast muscle (2–2.5 mm deep) and anchored by silk sutures to the skin. Sets of crystals were positioned at 0.35L (rostral myotomes) and 0.65L (caudal myotomes) as illustrated in Fig. 1. The orientation of the muscle fibres at these positions had been established previously by dissection of a dead fish (Fig. 1), enabling the crystals to be correctly aligned along the longitudinal axis of the muscle fibres. An oscilloscope connected to the sonomicrometer (Triton Technology Inc, model 120-1000) was used to check the alignment of the crystals. A correction of 5 ms was used for the lag in the sonomicrometry readings due to the data filter in the sonomicrometer circuit. The velocity of sound through fish muscle at 0 °C (1507 ms⁻¹; C. E. Franklin, unpublished results) was used to calibrate the sonomicrometer, and errors in readings due to the change in the velocity of sound through the crystal lenses and to changes in muscle stiffness were calculated to be less than 4 % of the measured strain. A pair of Teflon-coated, silver electromyography (EMG) wires (150 μm in diameter), with the insulation removed from the tips, was implanted near to each pair of sonomicrometry crystals. A differential amplifier (A-M Systems, Everett, USA) with low-and high-pass filter settings of 100 Hz and 1000 Hz, respectively, amplified the muscle potentials, which were recorded together with the sonomicrometry signals on a thermal array recorder (Summagraphics Ltd). At the end of the experiments, the location of the sonomicrometry crystals was determined by X-ray photography and dissection. Only those results from fish with crystals aligned along the longitudinal axis of the muscle fibres were used (see Fig. 1).

Muscle preparations were obtained from the dorsal surface of the caudal myotomes. Bundles of 5–12 fast muscle fibres were isolated on an ice-cooled dissection stage with the aid of a binocular microscope. The Ringer’s solution used for dissection had the following composition (in mmol l⁻¹): NaCl, 132.2; sodium pyruvate 10.0; KCl, 2.6; MgCl₂, 1.0; CaCl₂, 2.7; NaHCO₃, 18.5; NaH₂PO₄, 3.2; pH 7.5 at 0 °C). Aluminium T-clips were wrapped around remnants of myosepta as close to the insertion of the muscle fibres as possible (illustrated in Altringham and Johnston, 1994). Preparations were suspended via the T-clips between two hooks in a stainless-steel water-jacketed chamber with a glass bottom through which Ringer’s solution was circulated at 0±0.2 °C. One hook was attached to a strain gauge (AME 801, Senso-Nor, Horten, Norway) mounted on a micromanipulator and the other to a computer-controlled servomotor. The length of the preparation was adjusted to give a maximum isometric twitch, corresponding to a sarcomere length of 2.3–2.4 μm as measured by laser diffraction.

In work loop experiments, muscle fibres are subjected to cyclical length changes and are phasically stimulated during each cycle. The net work done by the muscle is obtained from plots of fibre length against the force generated (Josephson, 1985; Altringham and Johnston, 1990). In our experiments, information on the in vivo strain fluctuations during escape responses was used to define the cyclical length changes used for the work loop experiments. Muscle strain was recorded by sonomicrometry from the rostral and caudal positions for a minimum of six different escape responses from each of seven fish. Representative strain waveforms were selected and digitised. In order to create cyclical events, segments were abstracted corresponding to the initial C-bend and the subsequent contralateral contraction. Each 360 ° cycle started and ended at close to the resting sarcomere length. The resulting cycles were used as an input to the servomotor. In addition, the computer was programmed to stimulate the muscle at the same point in the strain cycle and for the same duration as was determined from the electromyogram (EMG) recordings in the representative in vivo experiments. Stimulus amplitude and pulse duration (1–2 ms) were optimised to give a maximal isometric twitch. The stimulation frequency was set within the range required to give a maximal fused isometric tetanus (50–70 Hz), and the number of pulses was adjusted to give the required stimulus duration. In each series of experiments, the servomotor was programmed with either the abstracted strain waveform for the C-bend or that for the subsequent contralateral contraction, and the muscle was stimulated. The effects of the abstraction procedure used were small owing to the negligible activation of fibres outside the period of the abstracted waveform. To allow the muscle to recover fully, 10 min was allowed between each contraction. The work done on unstimulated muscle was typically less than 2 % of the work done by active muscle for the C-bend contraction and less than 4 % of the work done for the contralateral contraction. Each experiment typically lasted 8 h, during which the maximum isometric tension usually declined by less than 5 %. The instantaneous power output during work loop experiments was calculated over 1 ms intervals.

In order to determine whether the simulated in vivo parameters used in work loop experiments gave the maximum possible mean power output, we systematically varied the strain cycle duration, strain amplitude, stimulus phase and stimulus duration. Only one parameter was varied at a time, the remaining three parameters were kept constant at the values recorded in vivo.

where V is velocity in FL s ⁻¹, F is stress and A, B and C are constants. Data for individual preparations were fitted using a non-linear curve-fitting programme (Regression, Blackwell Scientific Software, Oxford, England).

At the end of the experiments, muscle fibres were pinned out at their resting lengths and immersed in liquid isopentane cooled to −159 °C using liquid nitrogen. Frozen transverse sections (10 μm) were cut at several points along the length of the preparation, transferred to coverslips and stained for myosin ATPase activity (Johnston et al. 1974). Cross-sectional areas of intact fibres were determined using digital planimetry and image-analysis software (VideoPlan, Kontron, Basel, Switzerland). The total cross-sectional area of the fibres together with measurements of fibre length were used to calculate mass-specific power outputs.

Results are presented as means ± S.E.M. Seven fish were used for the swimming experiments, nine for steady-state isometric contractile properties and five for work loop experiments. Significant differences were determined using Wilcoxon signed-rank tests.

Escape responses consisted of an initial C-bend in which the head and tail of the fish bend in the same direction away from the centre of mass (half tailbeat), followed by one or more complete tailbeats and an unpowered glide of variable duration (Fig. 2). The escape responses were highly variable, especially with respect to the final direction of travel. Only in a few cases did N. coriiceps swim in a straight line (relative to the starting position of the fish): there was usually an angular component dictated by the initial C-bend and subsequent contralateral contraction. The presence of the sonomicrometry and electromyography (EMG) wires had no significant effect on swimming velocity and acceleration relative to unwired fish (signed-rank test) (Table 1). In 300 ms, N. coriiceps travelled 16.3±0.3 cm, reaching a maximum velocity and acceleration (as recorded from the centre of gravity of the fish) of 0.71±0.03 m s⁻¹ and 17.1±1.4 m s⁻², respectively (Table 1).

Sonomicrometry and EMG recordings were made from the superficial white muscle of N. coriiceps during escape responses (Figs 3, 4). There was considerable variation in muscle strain (Fig. 4) and activation patterns, reflecting the variation seen in the kinematics of the escape response. Fig. 3 presents results for a typical escape response showing a C-start initiated by muscles on the right-hand side resulting in fibres in the rostral myotomes (0.35L) shortening by 9 % of their resting length (FL) (Fig. 3A). The strain ranged between 5 % and 13 %FL (Fig. 4), with the magnitude of the strain being dependent upon the final direction of travel. For example, sonomicrometry recordings of strain taken from the right myotomes tended to be larger if the fish was moving to the right during the escape response than if it was going straight ahead or veering off to the left. The modal strain recorded from the rostral myotomes during the C-bend was 10 %, with a mean strain of 9.5±1.5 % (mean ± S.D., seven fish). The mean fibre-length-specific speed of muscle shortening for the largest strain recorded for each fish was 1.68±0.21 FL s⁻¹ (N=7). In some cases, there was a small pre-stretch (never more than 0.01FL), presumably due to the shortening of more anterior muscles and/or changes in muscle stiffness (Figs 3A, 5A). Electrical activity (EMGs) was recorded in rostral myotomes 10–20 ms before shortening and continued for 100–150 ms (Fig. 3A,C). The muscle fibre strain waveforms recorded from the caudal myotomes (0.65L) on the same side of the fish were of similar amplitude (5–12 %FL; modal value 9.5 %) and resembled those of the rostral myotomes in shape. At 0.65L, electrical activity was recorded 60–90 ms after the initiation of the fast-start whilst the muscle fibres were still lengthening (Fig. 3D). The pre-stretch recorded from the caudal myotomes was always greater than that recorded from the rostral myotomes. The C-bend caused muscle fibres on the contralateral side of the body (Fig. 3B) to be stretched by 5.0 %FL and 7.0 %FL at 0.35L and 0.65L, respectively, producing nearly sinusoidal strain waveforms (Fig. 4B). During the contralateral contraction, muscle fibres in both rostral and caudal positions shortened by up to 12 %FL (peak-to-peak strain) (Fig. 4B) at a maximum contraction speed of 1.62±0.22 FL s⁻¹ (N=7). In the rostral position, muscle strain during the contralateral contraction varied from 5 to 12 %FL, with a mean of 10.0±1.0 %FL, whereas in the caudal location muscle strain was 6–12 %FL, with a mean of 9.5±1.0 %FL (Figs 3, 4). Again, this variation reflected the final direction of travel of the fish. EMGs recorded from rostral and caudal positions started 40–65 ms after the beginning of fibre lengthening and finished 75–100 ms later, just after the onset of shortening (Fig. 3B).

All experiments were performed using muscle fibres isolated from the caudal myotomes as it proved impossible to dissect preparations from the rostral myotomes. In a previous study with this species, using skinned muscle fibres isolated from the rostral myotomes, we obtained similar values for maximum isometric stress (180 kN m⁻²) and unloaded contraction velocity (2.1 fibre lengths s⁻¹) as reported for caudal muscle fibres in the present study (Mutungi and Johnston 1988; G. Mutungi and I. A. Johnston, unpublished results). However, it should be noted that the properties of live muscle fibres may differ along the trunk.

Work loops were produced by ‘playing back’ in vivo strains and stimulation patterns to stimulated isolated muscle fibres (Fig. 5). Four abstracted (cyclical) events corresponding to the C-bend at 0.35L and 0.65L (Fig. 5A,C,E,G), and the subsequent contralateral contraction at rostral and caudal positions (Fig. 5B,D,F,H), were studied. In the rostral position, the C-bend produced a maximum stress of 115.0±7.3 kN m⁻² (N=5) and the maximum power reached 144.1± 16.8 W kg⁻¹ wet muscle mass (Fig. 5C,E; Table 2). In the caudal position, during the C-bend, the increased pre-stretch prior to shortening (Fig. 5A) contributed to a 24 % increase in the maximum stress and a more than twofold increase in the peak instantaneous power output relative to rostral myotomes (Table 2; P<0.05, signed-rank test). Although the maximum power outputs differed along the length of the trunk, the average power outputs over the abstracted cycles were not significantly different (Fig. 5G; Table 2).

During the subsequent contralateral contraction, the muscle fibres were initially active whilst lengthening, producing maximum stresses of 124.2±12.7 kN m⁻² and 168.0± 10.0 kN m⁻² at 0.35L and 0.65L, respectively (Fig. 5F; Table 2). This resulted in maximum power outputs of 269.6± 10.9 W kg⁻¹ wet muscle mass and 247.5±30.8 W kg⁻¹ wet muscle mass being generated in the rostral and caudal locations respectively (Fig. 5D; Table 2). The average power outputs over the abstracted cycles for the contralateral contractions were not significantly different at different points along the body and reached 15.8 and 19.6 W kg⁻¹ wet muscle mass for rostral and caudal positions respectively (Table 2).

The effect of varying, in turn, strain amplitude, cycle duration, stimulus duration and stimulus timing on muscle power output are shown in Figs 6 and 7. In all cases, the in vivo recorded ranges (indicated by the shaded boxes) produced between 90 and 100 % of the maximum in vitro power outputs.

The steady-state isometric contractile properties of the muscle fibres are shown in Table 3 and Fig. 8. V_max was calculated from the Marsh–Bennett (1985) equation and was 2.34±0.11 FL s⁻¹ (N=3 preparations). Values for the constants A, B and C in the Marsh-Bennett equation were 0.22±0.018, 0.027±0.0014 and 1.14±0.22 respectively. V_slack determined for the same three preparations was 2.39 FL s⁻¹.

During escape responses at 0 °C, the Antarctic rock cod (N. coriiceps) reached swimming velocities and accelerations comparable to escape responses recorded for some other fish species of similar size but at water temperatures 5–15 °C warmer (Beamish, 1978). Archer and Johnston (1989) recorded even faster velocities in N. coriiceps during burst swimming (up to 1.28 m s⁻¹); however, their measurements were made using trained fish and during the transition from labriform to subcarangiform swimming, and not from a standing start as in the present study. During prey capture, Beddow et al. (1995) recorded a maximum swimming velocity and acceleration of 0.85 m s⁻¹ and 17 m s⁻², respectively, for the short horn sculpin Myoxocephalus scorpius at 10 °C; the short horn sculpin is a temperate species which has a similar body form and locomotory behaviour to N. coriiceps. The maximum swimming performance in more-streamlined acceleration specialists such as northern pike Esox lucius is, however, considerably higher than for these other fish (Webb, 1978; Harper and Blake, 1990; Domenici and Blake, 1991).

The magnitudes of the muscle strains developed during the escape response in N. coriiceps were comparable to those calculated for the short horn sculpin during prey capture (S-shaped fast-starts) (Johnston et al. 1995), although the maximum muscle contraction velocities were considerably slower in N. coriiceps. The maximum shortening velocities of fast muscle fibres in caudal myotomes was 3.8 FL s⁻¹ for M. scorpius at 15 °C (Johnston et al. 1995), compared with only 1.6 FL s⁻¹ for N. coriiceps of similar length at 0 °C. This is not surprising as the maximum unloaded isotonic shortening velocity (V_max) of fast muscle fibres in fish shows no evidence for evolutionary temperature compensation, ranging from approximately 2 FL s⁻¹ in Antarctic species at −1 °C to 15–20 FL s⁻¹ in various tropical fish at 25 °C (Johnston and Brill, 1984; Johnson and Johnston, 1991). This is also reflected in the kinematics of fast-starts, where length-specific tailbeat frequencies of polar fish are lower than those for temperate and tropical species (Archer and Johnston, 1989; Harper and Blake, 1990; Domenici and Blake, 1991).

The maximum isometric stress and the force–velocity relationship are intrinsic properties of muscle fibres, and we have used them as reference points for performance under non-steady-state conditions. During escape responses, power is initially absorbed by the muscle because the fibres are active whilst lengthening. This results in significant force enhancement due to stretch (Fig. 8), as has been reported for cyclical contractions in frog muscle (Stevens, 1993). The mechanism of force enhancement with active stretch is unknown, but it may involve altered cross-bridge interaction with the thin filament (Edman et al. 1978). It is unlikely to involve an increase in the number of cross bridges recruited, since force enhancement after stretch increases with decreasing overlap of thick and thin filaments (Edman et al. 1978). Whatever the mechanism, active pre-stretch enhances performance, resulting in a significant deviation in behaviour from the steady-state force–velocity relationship (Fig. 8). For the contralateral contraction, differences in the rate and duration of stretch produced marked differences in the shape of the work (Fig. 5H) and power loops at 0.35L and 0.65L (Fig. 8). At both positions, V exceeded that predicted from the force–velocity curve for considerable periods. The velocity increased dramatically during the early part of shortening, whilst the load was relatively constant and still at an appreciable fraction of F_max, 0.65 at 0.35L and 0.85 at 0.65L (Fig. 8). The particularly high levels of force generation at 0 °C during isometric (Johnston and Brill, 1984; Johnston and Altringham, 1985) and cyclical (Fig. 7) contractions may reflect the unusual structure of the myosin molecule in Antarctic fish. Skeletal muscle myosin from Antarctic fish is readily denatured and loses its ATPase activity on isolation (Johnston et al. 1975). The myofibrillar ATPase activity of fast muscle from several species of Antarctic fish at 0 °C was higher, and the free activation energies (ΔG‡) were lower, than for the homologous enzyme in warm-water fish (Johnston and Goldspink, 1975). Values of F/F_max during the early part of shortening in N. coriiceps are much higher than those in muscle fibres from temperate fish measured under comparable conditions at 12 °C (R. S. James and I. A. Johnston, in preparation). Our results also contrast with those for continuous swimming, where steady-state muscle properties are better predictors of muscle performance. Red muscle fibres in carp (Cyprinus carpio) and scup (Stenotomus chrysops) have been shown to shorten at values of V/V_max of 0.18–0.36, precisely the values predicted to generate maximum power from the P–V relationship (Rome et al. 1988, 1992).

During the escape response, N. coriiceps generated maximum and mean power outputs comparable to those for predation fast-starts in a temperate fish, Myoxocephalus scorpius, at 15 °C. M. scorpius fast muscle fibres produced maximum instantaneous and mean power outputs of 175–265 and 21–24 W kg⁻¹ wet muscle mass, respectively (Johnston et al. 1995). No other similar data are available for fish; however, maximum instantaneous power output, W_max, for jet propulsion swimming in scallop is approximately 200 W kg⁻¹ wet muscle mass at 10 °C (Marsh et al. 1992), whereas mouse fast muscle delivers 300 W kg⁻¹ wet muscle mass at 35 °C during sinusoidal contractions (James, 1994). The similarity of these values during locomotion in Antarctic fish, scallop and mouse muscles is striking given the large differences in operating temperature, morphology and phylogeny of the animals being compared. In contrast, maximum muscle power output in N. coriiceps calculated from the P–V relationship, is only 60–70 W kg⁻¹ wet muscle mass, significantly lower than for fast muscles in most temperate fish and in homeotherms measured at their normal body temperature (Josephson, 1993).

When muscle fibre strain, the duration of the abstracted cycle, stimulation duty cycle and stimulus timing were each systematically optimised to give maximum power whilst holding all other parameters constant, the average power per cycle was between 90 and 100 % of that recorded using the in vivo parameters. Thus, during fast-starts, the muscle fibres in Antarctic fish are operating at close to their maximal performance.

This work was supported by a grant from the Natural Environment Research Council of the UK. We are also grateful to the British Antarctic Survey for providing the Antarctic fish.