Effects of α₁- or β-adrenoceptor stimulation on work-loop and isometric contractions of isolated rat cardiac trabeculae

Abstract

1. We studied the effects of α₁-or β-adrenoceptor stimulation on the contractility of isolated rat ventricular trabeculae at 24 °C using the work-loop technique, which simulates the cyclical changes in length and force that occur during the cardiac cycle. Some muscles were injected with fura-2 to monitor the intracellular Ca²⁺ transient.

2. Comparison of twitch records revealed that peak force was greater and was reached earlier in work-loop contractions than in corresponding isometric contractions. This was attributed to the changes in muscle length and velocity during work-loop contractions, since the Ca²⁺ transients were largely unaffected by the length changes.

3. Stimulation of α₁-adrenoceptors (with 100 μm phenylephrine) increased net work, power production, the frequency for maximum work, and the frequency for maximum power production (f_opt). The increase in net work was due to the positive inotropic effect of phenylephrine, which was similar at all frequencies investigated (0.33–4.5 Hz). The increase in f_opt was attributed to an abbreviation of twitch duration induced by α₁-stimulation at higher frequencies (> 1 Hz), even though the twitch became longer at 0.33 Hz.

4. β-Adrenoceptor stimulation (with 5 μm isoprenaline) produced marked increases in net work, power output, the frequency for net work, and f_opt. These effects were attributed both to the positive inotropic effect of β-stimulation, which was greater at higher frequencies, and to the reduction in twitch duration. β-Stimulation also abolished the frequency-dependent acceleration of twitch duration.

5. The increase in power output and f_opt with α₁- as well as β-adrenoceptor stimulation suggested that both receptor types may contribute to the effects of catecholamines, released during stress or exercise, although the greater effects of β-stimulation are likely to predominate.

The effects of inotropic interventions on the mechanical performance of isolated cardiac muscle are generally assessed using either isometric or isotonic techniques. However, neither technique accurately reflects the dynamics of the heart in vivo, which undergoes periods of isovolumic contraction and relaxation (analogous to isometric contraction and relaxation, respectively), interspersed with ventricular ejection and refilling phases (shortening and lengthening, respectively). In fact, in situ measurements of the strain in ventricular muscle have shown that its cyclical length changes are approximately sinusoidal (e.g. Semafuko & Bowie, 1975; Delhaas et al. 1993a). A sinusoidal length trajectory can therefore be applied to isolated muscles to approximately simulate physiological length changes. From this, the work and power output of the muscle can be assessed. A major advantage of this procedure, compared with purely isometric or isotonic techniques, is that it provides a direct, physiologically relevant measure of myocardial performance, because the main function of the ventricles is to perform work, i.e. to develop force while undergoing shortening to eject a volume of blood.

This is the approach taken in the work-loop technique (Josephson, 1985), which combines sinusoidal length changes with phasic electrical stimulation to investigate the sustainable mechanical power output of isolated muscles during repetitive, cyclical contractions as occur in vivo. It incorporates the changes in shortening velocity and level of activation that occur during cyclical contractions and takes into account both the work done by the muscle during shortening and the work required to re-extend the muscle during lengthening. Average power output is calculated as the product of work per beat and beat frequency. This technique has previously been applied to isolated cardiac muscle preparations (Syme, 1993, 1994; Layland et al. 1995) and is useful in assessing the effects of inotropic agents on cardiac muscle contractility since it is sensitive to changes in force, velocity, activation rate, relaxation rate and frequency.

In a previous study, Layland et al. (1997) demonstrated that adrenaline increased the power production of rat papillary muscles during work-loop contractions and raised the frequency for maximum power output (f_opt) from 6 to 8 Hz at 37°C. Since adrenaline acts on both α₁- and β-adrenoceptors in the rat myocardium (Skomedal et al. 1988) the relative contribution of each receptor system to the overall response is unknown. Therefore, in the present study we investigated the effects of separate α₁- and β-adrenoceptor stimulation on the mechanical performance of isolated rat trabeculae muscles during work-loop contractions. The results led us to examine the individual effects of α₁- and β-stimulation on the isometric force-frequency relationships and on twitch duration at different frequencies.

The use of trabeculae also allowed us to monitor both force and cytosolic [Ca²⁺] (after micro-injection of fura-2) during work-loop contractions. This was of interest because it is known that step decreases in muscle length during the relaxation phase cause a rise in cytosolic [Ca²⁺] as a result of the release of Ca²⁺ from troponin C (e.g. Allen & Kurihara, 1982; Jiang et al. 1988). We investigated whether similar alterations in the shape of the Ca²⁺ transient were produced by the sinusoidal changes in length that would be experienced by cardiac muscle in intact hearts.

Preliminary accounts of some of this work have been published in abstract form (Layland & Kentish, 1997; Layland & Kentish, 1999b).

METHODS

Muscle preparation

The muscle preparation and experimental apparatus were similar in most respects to those described in detail previously (Layland & Kentish, 1999a). In brief, Wistar or LBNF₁ rats (either sex, ∼250 g) were stunned and then killed by cervical dislocation (Schedule 1 procedure in accordance with UK Home Office regulations). The hearts were removed and rinsed free of blood in modified Krebs-Henseleit solution (Krebs solution) which contained (mM): NaCl, 93; NaHCO₃, 20; Na₂HPO₄, 1; MgSO₄, 1; KCl, 5; CaCl₂, 1; glucose, 10; and sodium acetate, 20; with insulin, 5 U l⁻¹; bubbled with 95 % O₂-5 % CO₂; pH 7.4 at 24°C. EGTA (10 μM) was added to the Krebs solution to inhibit oxidation of catecholamines catalysed by heavy metal ions. Suitable trabeculae (strongly beating, free running, unbranched, diameter < 250 μm) were dissected from the right ventricle. Muscle damage during dissection was minimised by the inclusion of 25 mM 2,3-butandenione-monoxime (BDM) in the Krebs solution. Trabeculae were attached horizontally between a force transducer element (SensoNor, Horten, Norway) and servomotor (300B, Cambridge Technology Inc., Watertown, MA, USA) and immersed in a muscle bath mounted on the stage of a Nikon Diaphot inverted microscope. The muscle was maintained in normal Krebs solution and stimulated via platinum field electrodes at 0.33 Hz for > 90 min. This allowed time for any spontaneous oscillations, resulting from damage during the isolation procedure, to disappear and for force to stabilize. During this stabilization period, muscle length was increased to approximately 95 % of the length for maximum active force generation (L_max), at which length resting force was about 10 % of active force. Previous work-loop experiments (Layland et al. 1995) showed that 95 %L_max was the optimal length for net work production in rat papillary muscles, and we found that this was also true for rat cardiac trabeculae (results not shown). Muscle length and width were measured and cross sectional area estimated, assuming that the muscle was cylindrical. Muscle wet weight was calculated from muscle volume, assuming that muscle density was 1.06 mg mm⁻³.

The effects of α₁- or β-adrenoceptor stimulation on the mechanical performance of the trabeculae were assessed in separate experiments. Propranolol (10 μM) was routinely included in the Krebs solution used for control and washout conditions. This blocked any possible β-adrenergic effects resulting from endogenous catecholamine release. Selective stimulation of the α₁-adrenoceptors was achieved using phenylephrine (100 μM), plus propranolol (10 μM) to block any β-adrenergic effects. Stimulation of the β-receptors was achieved using isoprenaline (5 μM), a non-selective β-adrenoceptor agonist, applied after a 20–30 min period of propranolol washout. Agonists were applied during isometric contractions at 0.33 Hz. Experiments were conducted 20–30 min after the start of drug application, when a new steady state had been achieved. All experiments were performed at 24°C, since some experiments required the use of fura-2 and higher temperatures greatly accelerate the loss of fura-2 from the cell (Thomas & Delaville, 1991; Layland & Kentish, 1999a).

Work-loop protocol

The work-loop technique was used to measure net work and power output at different frequencies during control conditions, during stimulation of either the α₁- or β-adrenoceptors and following a 30 min period of drug washout. Muscles were subjected to sinusoidal length changes at various frequencies and were stimulated to contract at a particular time (phase shift) in the sine wave. The range of frequencies investigated was dependent on the experimental condition, i.e. 1–4.5 Hz for control, phenylephrine and washout, or 1–6.5 Hz for isoprenaline. Muscle length changes were produced by the lever arm servo-system. A function generator (FG601, Feedback Instruments, Crowborough, East Sussex, UK) was used to control the amplitude and frequency of the sinusoidal length change and to synchronise the electrical stimulus with the sinusoidal waveform. The amplitude of the length change was ±5 % muscle length (10 % peak to peak), which is similar to physiological length changes of cardiac muscles (e.g. Semafuko & Bowie, 1975). For each cycle, a plot of force against length on an oscilloscope produced an anti-clockwise work loop (see Figs 3 and 4), the area of which represented the net work done during that cycle (Josephson, 1985). Timing of the stimulation was adjusted to maximise the area of the work loop. Cyclical contractions at each frequency were continued until a steady state was reached (∼2–3 min). The average force and length records of 10 or 20 consecutive work-loop cycles were then recorded using pCLAMP software (Axon Instruments). In order to monitor, and correct for, any decline in net work with time, every third test frequency was followed by a repeat of the initial test frequency (usually 1.5 Hz). A plot of work done at each repeated frequency against time indicated an approximately linear deterioration of muscle performance and allowed us to derive correction factors for each test frequency. Multiplication of the appropriate correction factor by the value of net work obtained for each test frequency allowed the prediction of net work had muscle performance been maintained. Muscles were discarded if, during the course of an experiment, net work at this repeat frequency declined to less than 70 % of its initial value. Justification of this approach has been detailed previously (e.g. see Layland et al. 1995).

Figure 3. Effects of α₁-stimulation on work-loop contractions of a typical muscle.

Left, force records during control conditions (thin lines) and during α₁-adrenoceptor stimulation (thick lines) at the cycle frequencies indicated. Dashed lines represent the sinusoidal length trajectories (± 5 % muscle length) at each frequency. Right, corresponding work-loops at each frequency, derived from the force and length records illustrated on the left. Force axes are the same scale as on the left. Arrows drawn onto loops indicate that all work-loops are anti-clockwise. Vertical arrows indicate the onset of electrical stimulation in the work-loop. Selective α₁-adrenoceptor stimulation was achieved using 100 μM phenylephrine in the presence of 10 μM propranolol. Averages of 10 traces. Note the varying time scales.

Figure 4. Effects of β-stimulation on work-loop contractions of a typical muscle.

Force and length records (left) and corresponding work-loops (right) of a typical muscle during work-loop contractions in control conditions (thin lines) and during β-adrenoceptor stimulation (thick lines) at the cycle frequencies indicated. Other details are as for Fig. 3. Selective stimulation of the β-adrenoceptors was achieved using 5 μM isoprenaline. Note the varying time scales.

Measurement of intracellular [Ca²⁺] using fura-2

In some muscles, intracellular Ca²⁺ transients during isometric and work-loop contractions were monitored using fura-2 fluorescence. Fura-2 is suitable for assessing Ca²⁺ transients during length changes since it uses a ratiometric method and is therefore insensitive to changes in the amount of dye present in the field of view. Our fura-2 fluorescence procedure was a modification of the technique established by Backx & ter Keurs (1993). In brief, trabeculae were loaded with fura-2 (K⁺ salt) by iontophoresis of fura-2 into 1–3 cells and the fluorescence in response to excitation at 340 and 380 nm was monitored using a spectrophotometer system (Cairn Research, Faversham, UK). The 340 nm/380 nm fluorescence ratio, a measure of intracellular [Ca²⁺], was calculated on-line after subtraction of the initial autofluorescence components at 340 and 380 nm (determined at 0.33 Hz with the muscle contracting isometrically). A potential problem was that changes in muscle length and operating frequency during work-loop contractions could alter the autofluorescence levels, such that the subtraction of the isometric values of autofluorescence before calculation of the 340 nm/380 nm ratio would give erroneous results. To address this problem, in pilot experiments we compared the ‘Cairn calculated’ 340 nm/380 nm fluorescence ratio (which subtracted the isometric autofluorescence at 0.33 Hz) with the ‘true’ 340 nm/380 nm ratio (which subtracted the autofluorescence recorded during work-loop cycles at different frequencies before injection of fura-2). At low frequencies (1 and 2 Hz) it was found that there was no difference between the ‘Cairn calculated’ and ‘true’ fluorescence ratios. However, at higher frequencies (3 and 4 Hz) the ‘Cairn calculated’ ratio tended to underestimate the peak of the ‘true’ Ca²⁺ transient by 3–10 %, which is similar to the error estimated previously for isometric contractions at comparable frequencies (Layland & Kentish, 1999a). These observations suggest that the differences in autofluorescence were largely due to frequency per se, rather than to the length change, which was the same amplitude (± 5 %) at all frequencies. Since the main focus of our investigation was to compare Ca²⁺ transients during isometric and work-loop contractions at a particular frequency, small errors in the fura-2 340 nm/380 nm ratio arising from frequency-dependent changes in autofluorescence would not influence the conclusions drawn from this comparison.

Chemicals and solutions

All chemicals used for Krebs solution were analytical grade and were obtained from B.D.H. (Poole, Dorset, UK). BDM, phenylephrine and propranolol were obtained from Sigma (Poole, Dorset). Isoprenaline was obtained in the form of a stabilised pharmaceutical solution containing ascorbic acid and disodium EDTA (Saventrine I.V., Pharmax Ltd, Bexley, UK). Phenylephrine was made up as a stock solution in de-ionised H₂O and stored at −20°C. Fura-2 (K⁺ salt) was obtained from Molecular Probes Inc. (Eugene, USA).

Data analysis

Results were analysed off-line using Clampfit (Axon) and Origin (Microcal) software. Active force was measured as the difference between total force and resting force. Components of twitch and Ca²⁺ transient duration were assessed by measuring the time to peak force or peak [Ca²⁺] (T_pk), time from peak force (or [Ca²⁺]) to 50 % relaxation (RT₅₀, a measure of relaxation rate) and time from stimulation to 90 % relaxation (T₉₀, an index of twitch or Ca²⁺ transient duration). Averaged force traces (n = 10 or 20) recorded from work-loop contractions at different frequencies were plotted against their corresponding averaged length records to produce anti-clockwise work loops (Figs 3 and 4). The area under the lengthening portion of the loop represented the work done by the apparatus to stretch the muscle (lengthening work). The area under the shortening portion of the loop represented the work done by the muscle during shortening (shortening work). Net work (μJ) was therefore the difference between shortening work and lengthening work, i.e. the area enclosed within the work loop, and was measured by integration (for further details see Josephson, 1985). Net work values (μJ) at each frequency were divided by the estimated wet weight (mg) to give net work in microjoules per milligram (equivalent to J kg⁻¹). Average power output (W kg⁻¹) at each frequency was calculated by multiplying net work (J kg⁻¹) by cycle frequency (Hz). Statistical analyses were performed using Student's paired t test (Microcal Origin) or 2-way repeated-measures ANOVA (Jandel SigmaStat) as appropriate.

RESULTS

Force and intracellular Ca²⁺ transients during isometric and work-loop contractions

Figure 1 illustrates force records and corresponding Ca²⁺ transients measured from a typical muscle performing isometric and work-loop contractions at different frequencies. In work-loop contractions (thick lines, Fig. 1), changes in passive force, resulting from changes in length, have been subtracted, so that only the active components of force are compared. Averaged data comparing twitches and Ca²⁺ transients during isometric and work-loop contractions at each frequency are shown in Fig. 2. Note that ‘frequency’ refers to both the stimulation and length oscillation frequencies. Each of the parameters measured to compare isometric and work-loop contractions at different frequencies (Fig. 2) were analysed using 2-way ANOVA with repeated measures (2-way RM ANOVA). For both isometric and work-loop contractions, an increase in frequency produced an increase in active force and peak systolic [Ca²⁺], which appeared to reach their maximum values at 2–3 Hz (Figs 1 and 2). This was coupled with a frequency-dependent shortening of both the twitch and Ca²⁺ transient, as reported previously for isometric twitches (e.g. Schouten, 1990; Bassani et al. 1995b; Layland & Kentish, 1999a).

Figure 1. Comparison of isometric and work-loop contractions of a typical muscle.

Active force records and intracellular Ca²⁺ transients (fura-2 fluorescence ratio) measured during isometric contractions (thin lines) and work-loop contractions (thick lines) at different frequencies (A–D). The active force component during work-loop contractions was calculated by subtracting the passive component measured in identical work-loop contractions but without electrical stimulation. The time of stimulus application is indicated by a vertical arrow. Dashed lines represent the sinusoidal length trajectories (± 5 % muscle length) of the work-loop contraction. Averages of 20 traces. Note the varying time scales.

Figure 2. Averaged data comparing isometric and work-loop contractions.

Comparison of the amplitudes and time course of twitches (left, 9 muscles) and Ca²⁺ transients (right, 4 muscles) during work-loop (▪) and isometric (○) contractions at different frequencies. Ca²⁺ transients are expressed as the fura-2 340 nm/380 nm ratio. Left: A, active stress; B, time to peak force (T_pk force); C, time from peak force to 50 % relaxation (RT₅₀ force); and D, time from stimulus to 90 % relaxation (T₉₀ force). Right: E, peak systolic and diastolic ratio; F, time to peak Ca²⁺ (T_pk Ca²⁺); G, time from peak Ca²⁺ to 50 % decline (RT₅₀ Ca²⁺); and H, time from stimulus to 90 % Ca²⁺ transient decline (T₉₀ Ca²⁺). Symbols represent means ±s.e.m. For clarity, some error bars are shown in one direction only. Error bars are omitted from diastolic ratio data since they are less than the size of the symbol. * Significant differences between work-loop and isometric parameters at the same frequency (paired t test, significance at P < 0.05).

Active force was slightly greater (Figs 1 and 2A, P < 0.05, n = 9) and T_pk force was significantly faster (Fig. 2B, 2-way RM ANOVA, P < 0.05, n = 9) in work-loop contractions than in corresponding isometric contractions. However, small differences in RT₅₀ and T₉₀ of the force record between work-loop and corresponding isometric contractions (Figs 1 and 2C and D) were not statistically significant (2-way RM ANOVA, P > 0.05, n = 9). There were very few detectable differences in Ca²⁺ transients measured during isometric and work-loop contractions (Figs 1 and 2E–H). At 3 and 4 Hz, Ca²⁺ transient decline appeared to be faster in work-loop contractions than in isometric contractions (see Fig. 1C and D) but the differences in RT₅₀ were not significant (2-way RM ANOVA, P > 0.05).

The reductions in T_pk force, RT₅₀ and T₉₀ during work-loop contractions compared with corresponding isometric contractions (Fig. 2B, C and D) were most pronounced at the higher stimulation frequencies (> 1 Hz). Indeed, comparison of T_pk, RT₅₀ and T₉₀ for each individual frequency using paired t tests revealed that the differences in T_pk force were significant at 2, 3 and 4 Hz (Fig. 2B, P < 0.05); differences in RT₅₀ were significant at 4 Hz only (Fig. 2C, P < 0.05) and differences in T₉₀ were significant at 2 and 3 Hz (Fig. 2D, P < 0.05). Interestingly, 2-way RM ANOVA indicated that there were statistically significant interactions between frequency and type of contraction for T_pk force and RT₅₀ force (P < 0.05, n = 9) suggesting that the differences in T_pk and RT₅₀ between work-loop and isometric contractions were indeed frequency dependent. This would be expected since dynamic processes occuring during shortening may be velocity dependent (see Discussion).

Effects of α₁- or β-stimulation on isometric twitch force

The concentrations of the α₁- and β-agonists used were those found to produce maximal inotropic responses in rat cardiac muscles during preliminary experiments. As observed in previous studies (Endoh & Blinks, 1988), both α₁- and β-adrenoceptor stimulation had positive inotropic effects. During isometric twitches at 0.33 Hz, α₁-adrenoceptor stimulation (100 μM phenylephrine + 10 μM propranolol) increased active stress from 29.1 ± 9.1 to 48.2 ± 8.4 mN mm⁻² (P < 0.05, n = 6). Similarly, stimulation of the β-adrenoceptors with isoprenaline (5 μM) increased isometric active stress at 0.33 Hz from 37.1 ± 9.3 to 57.0 ± 10.6 mN mm⁻² (P < 0.05, n = 6).

Effects of α₁- or β-stimulation on work-loop contractions

Figures 3 and 4 illustrate the force and length records, and resulting work-loops, for typical muscles undergoing work-loop contractions at different frequencies in control conditions and during stimulation of either the α₁- (Fig. 3) or β-adrenoceptors (Fig. 4). The length change waveform is illustrated by the dashed sinusoidal line, and represents a strain amplitude of ±5 % muscle length. Stimulation phase shift was optimised for net work production at each frequency. Mean data are presented in Fig. 5.

Figure 5. Effects of α₁- or β-stimulation on net work and power output at different frequencies.

Work-frequency relationships (A and B) and power-frequency relationships (C and D) derived during stimulation of the α₁-adrenoceptors (A and C, ○, 6 muscles) or β-adrenoceptors (B and D, ▵, 6 muscles) compared with control relationships (▪) from corresponding muscles. Net work and power output are expressed per unit wet weight. Symbols represent the means ±s.e.m. Curves were fitted using 4th order polynomial regressions.

In control conditions (thin lines, Figs 3 and 4) the shape and area of the work loop varied with frequency. This is expected, as the shape and area of the work loop are determined by factors such as: (i) the force developed, (ii) the rate of force development, (iii) the rate of force decline, (iv) the shortening velocity, (v) the fraction of the shortening and lengthening periods occupied by force generation, and (vi) other dynamic processes such as shortening deactivation and force enhancement (for further details see James et al. 1996). At low frequencies, net work (work done by the muscle during shortening minus work done on the muscle during lengthening; equal to the area of the work loop) was largely determined by the fraction of the shortening period occupied by force generation. Thus net work was lower at 1 Hz than at 2 Hz (Figs 3-5), even though peak force was similar, because contraction was a smaller fraction of the shortening period. However, at the higher frequencies (e.g. 3 and 4 Hz), the twitch lasted longer than the shortening period. This incomplete relaxation at the onset of lengthening increased lengthening work and so tended to decrease net work. This was offset partly, during the optimisation of loop area, by stimulating the muscle earlier (during lengthening), so allowing more time for relaxation before re-lengthening began. However, stimulation during lengthening tended to increase the area under the rising phase of the loop (i.e. from 0 to +5 % muscle length), so net work decreased. At the highest frequencies examined (e.g. 5 Hz, Fig. 4), both incomplete relaxation before re-lengthening and activation during lengthening decreased further the area under the loop. The shape of the work-loop twitches, particularly during relaxation, was also influenced by frequency, with a noticeable inflexion during the relaxation phase at the end of muscle shortening (e.g. at 4 Hz, Fig. 3).

α₁-Stimulation with phenylephrine during work-loop contractions produced an increase in force at each frequency (Fig. 3, left) and the effects of phenylephrine on the work loops (Fig. 3, right) were largely attributed to this positive inotropic effect. However, at frequencies above 0.33 Hz, phenylephrine also reduced twitch duration slightly (see below), which would tend to increase net work by facilitating relaxation before re-lengthening occurred, so limiting the work required to re-stretch the muscle.

β-Adrenoceptor stimulation during work-loop contractions increased force but also shortened the time course of the twitch (Fig. 4, left). The positive inotropic effect of β-stimulation undoubtedly played a major role in increasing net work output, particularly at higher frequencies (see below). The β-mediated shortening of twitch duration also influenced net work output. For example, for work loops at 1 and 2 Hz (Fig. 4, right), the effects of isoprenaline to reduce twitch duration meant that force production, and thus shortening work, was terminated earlier than under control conditions. Hence, although isoprenaline increased force at these frequencies, the accompanying reduction in twitch duration meant that isoprenaline actually had no significant effect on net work and power output at 1 and 2 Hz (paired t tests, P > 0.05). At higher cycle frequencies (e.g. 5 Hz) the areas of the work loops under control conditions were limited by incomplete relaxation of the muscle before re-lengthening, such that more work was required to stretch the muscle, thereby reducing net work. Incomplete relaxation between twitches was illustrated as an increase in control force above the zero baseline in force records and work loops (e.g. at 5 Hz, Fig. 4). The effect of isoprenaline to accelerate relaxation therefore contributed to the increase in net work at higher frequencies (4 and 5 Hz) by allowing more complete relaxation before re-lengthening occurred.

Figure 5 shows the work-frequency (Fig. 5A and B) and power-frequency (Fig. 5C and D) relationships derived from work-loop contractions in control conditions and during stimulation of either the α₁-adrenoceptors (left) or β-adrenoceptors (right). The work-frequency and power- frequency curves were found to be best fitted by 4th order polynomials. Therefore, for each muscle, maximum net work or power output and the frequencies for maximum net work or power output (f_opt) were determined by fitting the data with a 4th order polynomial regression and finding the peak of the fit. Stimulation of the α₁-adrenoceptors with phenylephrine (Fig. 5A) increased maximum net work from 3.3 ± 0.7 to 4.2 ± 0.7 J kg⁻¹ (P < 0.01, n = 6) with an increase in the frequency for maximum work from 2.2 ± 0.3 to 2.6 ± 0.3 Hz (P < 0.01, n = 6). Stimulation of the β-adrenoceptors (Fig. 5B) increased maximum net work from 3.4 ± 0.6 to 4.7 ± 0.6 J kg⁻¹ (P < 0.05, n = 6) and raised the frequency for maximum net work from 2.1 ± 0.2 to 3.7 ± 0.1 Hz (P < 0.01, n = 6).

Average power output at each cycle frequency was calculated by multiplying net work by cycle frequency. The effects of α₁- and β-adrenoceptor stimulation on the resulting power-frequency relationships (Fig. 5C and D) were more pronounced than on the work-frequency relationships. Stimulation of the α₁-adrenoceptors (Fig. 5C) increased the maximum power produced by the muscles from 7.8 ± 1.1 to 12.4 ± 1.6 W kg⁻¹ (P < 0.01, n = 6), with a rise in the frequency for maximum power output (f_opt) from 3.1 ± 0.3 to 3.5 ± 0.2 Hz (P < 0.05, n = 6). β-Adrenoceptor stimulation (Fig. 5D) dramatically increased maximum power output from 8.2 ± 1.1 to 18.8 ± 2.4 W kg⁻¹ (P < 0.01, n = 6) and produced a large increase in f_opt from 2.9 ± 0.3 to 4.4 ± 0.2 Hz (P < 0.01, n = 6). The effect of β-adrenergic stimulation on power output was greater than its effect on net work since β-adrenergic stimulation increased the range of operating frequencies over which net work was produced (as explained for Fig. 4) and power output is calculated as the product of net work and cycle frequency.

In some muscles, power-frequency relationships were derived following a 30 min period of drug washout with control Krebs solution (containing 10 μM propranolol). Our results suggested that the effects of α₁- and β-stimulation were not fully reversible after this time (data not shown). Maximum power output measured after phenylephrine washout (11.3 ± 2.8 W kg⁻¹, n = 4) was higher than paired control data (7.6 ± 1.7 W kg⁻¹, P < 0.05) and indeed was not significantly different from maximum power output during α₁-stimulation (12.1 ± 2.4 W kg⁻¹, n = 4). After phenylephrine washout, f_opt was 3.5 ± 0.2 Hz (n = 4), which was slightly higher than in control conditions (2.9 ± 0.4 Hz), although this difference was not statistically significant.

Maximum power output after isoprenaline washout was 12.7 ± 1.0 W kg⁻¹ (n = 4) which, although lower than during β-stimulation (21.3 ± 2.9 W kg⁻¹, n = 4, P < 0.05) was still greater than in control conditions (9.2 ± 1.5 W kg⁻¹, n = 4, P < 0.05). Furthermore, the isoprenaline-induced shift in f_opt, from 2.7 ± 0.4 to 4.4 ± 0.3 Hz (P < 0.01, n = 4) was only partially reversed (to 3.5 ± 0.3 Hz) after washout. This was lower than during β-stimulation (P < 0.01) but higher than control values (P < 0.01).

Effects of α₁- or β-stimulation on the force-frequency relationship

An increase in frequency per se can influence the amount of force developed and hence influence the net work produced during work-loop contractions. In order to investigate the contribution of force-frequency properties to the work- frequency relationships, we measured the active force generated during work-loop contractions at different frequencies, for control conditions and during stimulation of either α₁- or β-adrenoceptors. The resulting force-frequency relationships are illustrated in Fig. 6. Phenylephrine (Fig. 6A) increased force by a similar amount at all frequencies investigated (percentage increase ranged from 33 to 66 %, mean increase 49 ± 5 %, n = 6 muscles, over nine frequencies from 0.33 to 4.5 Hz). The frequency for maximum force production (determined by fitting the individual data with 3rd order polyomials) in control conditions was 2.3 ± 0.3 Hz (n = 6) and was not significantly altered by phenylephrine, i.e. 2.4 ± 0.3 Hz (n = 6). As a result phenylephrine did not alter the peak of the force- frequency relationship but merely shifted the relationship upwards on the force axis (Fig. 6A). This is contrary to previous reports that the positive inotropic effect of α₁-adrenoceptor stimulation becomes less pronounced as stimulation frequency is increased (Endoh & Schümann, 1975; Fedida et al. 1993; Terzic et al. 1993). In contrast, the positive inotropic effect of β-stimulation was found to be greatest at higher frequencies (Fig. 6B), so altering the nature of the force-frequency relationship. For example, the percentage increase in force ranged from 2 % at 1.5 Hz to 106 % at 4.5 Hz (mean percentage increase 48 ± 12 % for n = 6 muscles over nine frequencies from 0.33 to 4.5 Hz). Furthermore, the frequency for maximum force production in control conditions was 2.1 ± 0.1 Hz (n = 6) and was significantly increased to 3.7 ± 0.3 Hz with isoprenaline (paired t test, P < 0.01, n = 6). Additional experiments confirmed that α₁- and β-stimulation had comparable effects on force-frequency relationships derived during isometric contractions. Hence, phenylephrine increased maximum force from 65.4 ± 14.4 to 85.4 ± 15.2 mN mm⁻² (P < 0.05, n = 6) but had no significant effect on the frequency for maximum force production (i.e. 2.0 ± 0.3 Hz in control conditions and 2.2 ± 0.2 Hz with phenylephrine, n = 6). Stimulation of the β-adrenoceptors during isometric contractions increased maximum force from 63.1 ± 14.3 to 93.0 ± 19.6 mN mm⁻² (P < 0.05, n = 4). This was accompanied by a significant increase in the frequency for maximum force production from 2.1 ± 0.3 to 3.7 ± 0.1 Hz (P < 0.05, n = 4).

Figure 6. Effects of α₁- or β-stimulation on the force-frequency relationship.

Force-frequency relationships derived during stimulation of the α₁-adrenoceptors (A, •, 6 muscles) or β-adrenoceptors (B, ▴, 6 muscles) compared with corresponding control relationships (▪). Active force was measured from work-loop contractions at each frequency except for 0.33 Hz, where active isometric force was measured (shown as open symbols). Stimulation of the α₁-adrenoceptors was achieved with 100 μM phenylephrine, plus 10 μM propranolol to block any β-mediated effects. β-Adrenoceptor stimulation was achieved using 5 μM isoprenaline. Force is expressed per unit cross sectional area, i.e. as stress (mN mm⁻²). Symbols represent means ±s.e.m. Curves were fitted using 3rd order polynomial regressions.

Effects of α₁- or β-stimulation on twitch duration

One unexpected finding of our study was the shortening of the twitch produced by α₁-stimulation at higher stimulation frequencies (e.g. see twitch and loop at 2 Hz, Fig. 3). Previous work at lower stimulation frequencies has generally found that α₁-stimulation prolongs the twitch or has little effect on its timecourse (e.g. Endoh & Blinks, 1988; Terzic et al. 1993). We therefore investigated the effects of phenylephrine and isoprenaline on twitch duration during both isometric and work-loop contractions (Fig. 7). For these experiments, we also studied isometric twitches at 0.33 Hz to compare with previous studies. As stated above, an increase in stimulation frequency per se reduced twitch duration in both isometric and work-loop contractions (see Figs 1 and 2 and control data, Fig. 7). Stimulation of the α₁-adrenoceptors at 0.33 Hz prolonged isometric twitch duration (Fig. 7A), increasing T₉₀ from 372 ± 41 to 403 ± 33 ms (paired t test, P < 0.05, n = 7). However, over a range of higher stimulation frequencies (1–3.5 Hz) phenylephrine significantly reduced T₉₀ for both isometric (Fig. 7A) and work-loop (Fig. 7B) contractions. At the highest frequencies (3.5–4.5 Hz), phenylephrine did not alter T₉₀.

Figure 7. Effects of α₁- or β-stimulation on twitch duration.

Left, the effects of α₁-adrenoceptor stimulation on T₉₀ (time from stimulus to 90 % relaxation) during isometric (A, 7 muscles) and work-loop (B, 5 muscles) contractions at different frequencies. Right, the effects of β-stimulation on T₉₀ during isometric (C, 7 muscles) and work-loop (D, 3 muscles) contractions. In all, symbols represent means ±s.e.m. Control values, ▪; phenylephrine, ○; and isoprenaline, ▵. *Significant differences between T₉₀ measured in control conditions and during α₁- or β-stimulation (comparing corresponding muscles at corresponding frequencies, paired t test, significance at P < 0.05).

β-Stimulation with isoprenaline significantly reduced T₉₀ for the isometric twitch at 0.33 Hz from 427 ± 53 to 294 ± 17 ms (Fig. 7C). Similarly, T₉₀ was reduced at all stimulation frequencies except the very highest, for both isometric (Fig. 7C) and work-loop (Fig. 7D) contractions. Interestingly, the frequency-dependent shortening of the twitch that was observed in control conditions, or in the presence of phenylephrine, was largely abolished by isoprenaline, at least over the range 1–4.5 Hz. Hence, during β-stimulation, T₉₀ values measured at frequencies from 1.5 to 4.5 Hz were not significantly different from T₉₀ measured at 1 Hz.

DISCUSSION

Relevance of the work-loop technique

In the present study we examined the effects of α₁- or β-adrenoceptor stimulation on the net work and power output of isolated rat ventricular trabeculae using the work-loop technique. This technique assesses the mechanical performance of isolated cardiac muscles under conditions approximately simulating the normal cardiac cycle, taking into account both the work done during shortening and the work required to re-extend the muscle during lengthening. Sustainable power output measured using the work-loop technique therefore represents a useful index with which to examine the effects of α₁- and β-stimulation on cardiac performance, with the muscle undergoing quasi-physiological, but well-controlled, length changes. Although the work-loop technique provides a better representation of contractions of heart muscles in situ than does either the isometric or isotonic contraction, there are some potential limitations of the technique we used. For example, although measurements of fibre strain in papillary muscles (Semafuko & Bowie, 1975) and sub-epicardial fibres (Delhaas et al. 1993a) suggest that the ‘physiological’ length change trajectory can be reasonably approximated by a sinusoidal waveform, there are still subtle differences between the true ‘physiological’ length change waveform and the sinusoidal change. However, it is unlikely that minor differences in waveform trajectory would substantially influence our measurements of net work and power output, since experiments in skeletal muscle have demonstrated that variations in strain waveform from a sinusoidal trajectory have little effect on the average power output (e.g. Marsh & Olson, 1994).

Another possible difference between our work-loop contractions and contractions of muscles in the intact heart is that we adjusted the timing of stimulation at each frequency in order to optimise the net work and power output at that particular frequency. In the intact heart, electrical activation of a muscle begins just before peak fibre length is reached, i.e. during the final stages of lengthening (Semafuko & Bowie, 1975; Delhaas et al. 1993b). In the present study, this was the case at intermediate frequencies (e.g. 2 and 3 Hz, Fig. 1B and C), for which net work and power output were maximal when the stimulus was applied just before peak length was reached, i.e. just prior to shortening. However, at 1 Hz the optimal stimulation phase generated contractions in which the muscle continued to shorten when it was relaxed (Fig. 1A), which would not occur during normal cardiac contraction. This arose because this frequency (which is below physiological for the rat) gave a length change that was too slow compared with the rapid twitch of the rat muscle. This would not, however, affect our measurement of work and power, because this phase of shortening occurred when force was zero. On the other hand, at the highest frequencies the muscle was activated slightly earlier in the lengthening phase than would normally occur in the heart (e.g. 5 Hz, Fig. 4). If we had applied the stimulation at the same point in the length oscillation (i.e. immediately prior to shortening) at each frequency, this would have reduced net work and power output at the extreme low and high frequencies but would not have altered our measurements at intermediate frequencies (since this phase was optimal for these frequencies anyway). Our optimisation procedure is therefore unlikely to have drastically altered the effects of α₁- and β-stimulation on maximum power output and f_opt.

Comparison of isometric and work-loop contractions

Comparison of twitch force during isometric and work-loop contractions at 1–3 Hz (Figs 1 and 2A–D) revealed that peak force was slightly greater and time to peak force was faster in work-loop than in corresponding isometric contractions. The difference in time to peak force was particularly marked at higher frequencies (Fig. 2B). Twitch relaxation time (RT₅₀) and twitch duration (T₉₀) also tended to be slightly faster in work-loop contractions than in isometric contractions at frequencies above 1 Hz, although several differences were not significant (Fig. 2C and D). There was also a difference in the pattern of relaxation, in that an ‘inflexion point’ observed in the relaxation of work-loop contractions at higher frequencies was absent in corresponding isometric contractions (Fig. 1C and D, Fig. 2). The earlier time to peak force in work-loop contractions was probably due to the onset of muscle shortening, which would tend to reduce force development as a consequence of the force-velocity relationship, or shortening deactivation, or both (Edman, 1980; Colomo et al. 1986). These dynamic effects of shortening would be greater at higher frequencies, which could explain why differences in time to peak force and RT₅₀ are most pronounced at 4 Hz (Fig. 2B and C). In addition, dynamic processes would be most marked during the rapid phase of shortening and would have less effect towards the end of the shortening period, when the muscle is decelerating and velocity is lower. This may explain the ‘inflexion point’ during the relaxation of the work-loop contractions.

This is the first study to measure the intracellular Ca²⁺ transients during work-loop contractions of isolated cardiac muscle. In general, Ca²⁺ transients were similar in the work-loop and isometric contractions (Figs 1 and 2E–H). Previous studies have shown that a sudden decrease in muscle length during relaxation increases cytosolic [Ca²⁺], due probably to Ca²⁺ release from the thin filaments as a result of a fall in the Ca²⁺ affinity of troponin C (e.g. Allen & Kurihara, 1982; Jiang et al. 1998). Although the shortening of the muscle during relaxation (10 %) was greater in our experiments than in these previous studies, we observed little difference in [Ca²⁺] between isometric and work-loop contractions; in fact, a bump of [Ca²⁺] during relaxation was seen in both types of twitch (e.g. Fig. 1B). This similarity suggests that the [Ca²⁺] bump results primarily from the detachment of cross-bridges due to de-activation of the thin filament during relaxation (which was similar for both types of contraction) rather than from detachment of cross-bridge due to muscle shortening. These results indicate that the length changes experienced by muscle in the working heart in vivo are not likely to, in themselves, significantly increase cytosolic [Ca²⁺] during relaxation. Probably the increase in [Ca²⁺] produced by a rapid decrease in length in previous studies (e.g. Allen & Kurihara, 1982; Jiang et al. 1998) was not apparent in our study because the physiological length changes that we used were much slower, thus allowing time for any extra Ca²⁺ released by troponin C to be taken up by the sarcoplasmic reticulum (SR).

Frequency-dependence of isometric and work-loop contractions under control conditions

In agreement with a previous study (Layland & Kentish, 1999a), the active force generated by rat ventricular trabeculae under control conditions showed a positive force-frequency relationship for frequencies from 0.33 to 2 Hz and a negative relationship at higher frequencies (Fig. 6). The twitch relaxation also became shorter as frequency increased (Figs 2 and 7). These effects are largely responsible for the shape of the work- and power-frequency relationships (see later). Although it is generally believed that rat cardiac muscle possesses a negative force-frequency relationship, a number of studies have shown that rat muscles can exhibit a positive force-frequency relationship under conditions where the cells are not overloaded with Ca²⁺ (Layland & Kentish, 1999a and references therein). The positive force-frequency relationship can then be explained on the same basis as that for other mammalian species, as follows (e.g. Bers, 1992). An increase in stimulation frequency increases Ca²⁺ influx (via I_Ca) per unit time and increases Na⁺ entry (via I_Na), which raises cytosolic [Na⁺] and promotes Ca²⁺ influx and reduces Ca²⁺ efflux, via Na⁺-Ca²⁺ exchange. The resultant increase in intracellular [Ca²⁺] will promote SR Ca²⁺ loading. More SR Ca²⁺ will be available for release and the fraction of SR Ca²⁺ released per beat, which depends upon intra-SR Ca²⁺ load (e.g. Bassani et al. 1995a), may also increase. These mechanisms will thus tend to increase SR Ca²⁺ release. As the optimum frequency for force production is approached, the capacity of the SR to load with Ca²⁺ may become limiting (Layland & Kentish, 1999a). This could occur because the SR Ca²⁺ uptake reaches thermodynamic equilibrium (Shannon & Bers, 1997) or because there is reduced time available for uptake between beats. As frequency is increased further, mechanisms that tend to decrease SR Ca²⁺ release will predominate, leading to a decline in force. Such mechanisms include a reduction in SR fractional Ca²⁺ release as a result of insufficient time for the SR Ca²⁺ release channel (ryanodine receptor, RyR) to recover from its inactivated/adapted state (Fabiato, 1985; Györke & Fill, 1993; Jafri et al. 1998).

The acceleration of the decline of the twitch or Ca²⁺ transient as stimulation frequency is raised (e.g. Figs 2 and 7) is a phenomenon that is well established but poorly understood (discussed more fully in Layland & Kentish, 1999a). One possibility is that the frequency-dependent increase in time-averaged [Ca²⁺]_i may lead to activation of Ca²⁺/calmodulin-dependent protein kinase, CaMKII, which would enhance the rate of SR Ca²⁺ uptake by phosphorylation of phospholamban (Schouten, 1990; Bassani et al. 1995b). However, the frequency-dependent shortening of the Ca²⁺ transient does not appear to require phosphorylation of phospholamban (Hussain et al. 1997; Li et al. 1998). Another proposal is that the acceleration of Ca²⁺ transient decline may merely reflect the fact that the greater [Ca²⁺]_i during the twitch stimulates SR Ca²⁺ uptake in a non-linear fashion (Bers & Berlin, 1995). However, the relevance of this mechanism is also questionable, since acceleration of Ca²⁺ decline can still be seen at frequencies where force or systolic [Ca²⁺]_i decreases (e.g. Frampton et al. 1991).

Effects of α₁-adrenoceptor stimulation on work-loop and isometric contractions

In work-loop contractions (Figs 3 and 5), phenylephrine increased the net work and power output of ventricular trabeculae and produced small but significant increases in the frequencies for maximum work and power (f_opt). The increase in maximum net work with α₁-stimulation could largely be attributed to its positive inotropic effect, which was seen at all frequencies investigated (Fig. 6A). The increases in the frequency for maximum net work and power output were more surprising since (i) the optimal frequency for isometric force production was not altered by phenylephrine (Fig. 6) and (ii) α₁-stimulation has been reported to prolong twitch duration (see below), which would tend to decrease the frequency for net work (since it would exacerbate incomplete relaxation at higher frequencies, thereby limiting net work production by increasing the work required to re-lengthen the muscle). However, further analysis (Fig. 7A and B) revealed that the effects of phenylephrine on twitch duration were frequency dependent, with prolongation of the twitch at low frequencies (0.33 Hz) and abbreviation of the twitch at higher frequencies (> 1 Hz). The twitch abbreviation probably explains the observed increase in the frequency for maximum net work, since it increases the range of frequencies over which the muscle can generate work. The effects of α₁-stimulation on the power-frequency relationship (Fig. 5C), i.e. a large increase in power output and a significant increase in f_opt from 3.1 to 3.5 Hz, reflect the combined effects of α₁-stimulation to increase both net work itself and the frequency for net work (since average power output = net work × cycle frequency).

α₁-Adrenergic stimulation has been observed to have a positive inotropic action in most species, but the precise mechanisms remain to be clarified. Proposed mechanisms include: (i) an increase in the intracellular Ca²⁺ transient, caused either by an increase in the magnitude of the L-type Ca²⁺ current (Liu & Kennedy, 1998; Zhang et al. 1998) or by a reduction in the transient outward K⁺ current, I_to (Fedida et al. 1990), which would increase action potential duration and thereby prolong the time for Ca²⁺ influx; (ii) an increase in myofilament Ca²⁺ sensitivity, caused by an intracellular alkalosis that results from stimulation of the sarcolemmal Na⁺-H⁺ exchanger (for reviews, see Fedida et al. 1993; Terzic et al. 1993). Our finding that the positive inotropic effect of α₁-stimulation was largely independent of frequency (Fig. 5A) contrasts with previous reports that the positive inotropic effect decreases as frequency is increased, and is virtually abolished at 2–3 Hz (Endoh & Schümann, 1975; Fedida et al. 1993). Endoh & Schümann (1975) suggested that at high frequencies α₁-stimulation may be unable to increase [Ca²⁺]_i further as it has already been maximally elevated by increased frequency. It is possible that the intracellular Ca²⁺ loading capacity of the muscles under our experimental conditions was far from saturated, even at high frequencies, thus allowing a positive inotropic effect of α₁-stimulation (by increase in [Ca²⁺]_i or myofibrillar Ca²⁺ sensitivity) over the entire frequency range studied. An alternative mechanism proposed for a smaller inotropic effect of α₁-stimulation at high frequencies is that the action of α₁-stimulation to decrease I_to and hence prolong the action potential, is reduced at higher frequencies because I_to itself becomes smaller (Fedida et al. 1990). However, the maintained inotropy over the frequency range we used suggests that this mechanism cannot be a major component of the α₁-mediated potentiation of force in our experiments.

The positive inotropic effect of α₁-stimulation has generally been associated with a slight prolongation or no change in the duration of the twitch (see Terzic et al. 1993, for review). What causes the lengthening of the twitch is uncertain, but at low frequencies it could result from a prolongation of the action potential by inhibition of I_to (Fedida et al. 1990). Another possible mechanism is that α₁-stimulation may reduce the intrinsic rate of cross-bridge cycling (e.g Strang & Moss, 1995), although other studies have suggested that cross-bridge cycling rate is unchanged (Hoh et al. 1988) or even increased (Li & Rouleau, 1991) by α₁-stimulation. We confirmed that α₁-stimulation significantly prolongs the duration of the isometric twitch at 0.33 Hz (Fig. 7A). However, an unexpected finding was that α₁-stimulation actually reduced twitch duration (for both isometric and work-loop contractions) at higher frequencies (1–3 Hz, Fig. 7A and B). To our knowledge, this is the first study to investigate the effects of α₁-stimulation on twitch duration at higher frequencies, which are more appropriate for rat heart muscle. Although our results do not provide an explanation for the abbreviation of the twitch during α₁-stimulation, they suggest that inhibition of I_to by α₁-stimulation does not play a major role in determining twitch duration at higher frequencies. One possibility is that the combination of α₁-stimulation and high [Ca²⁺]_i at higher frequencies activates protein kinase C (PKC), which phosphorylates phospholamban and increases SR Ca²⁺ uptake. However, although PKC induced phosphorylation of phospholamban and has been shown to stimulate the rate of SR Ca²⁺ uptake in vitro (Movsesian et al. 1984), the contractile effects of α₁-stimulation in the isolated heart were found not to involve phospholamban phosphorylation (Talosi & Kranias, 1992). It could be that the abbreviation of twitch duration during α₁-stimulation may result from stimulation of SR Ca²⁺ uptake, independent of phospholamban phosphorylation, as suggested for the frequency-dependent abbreviation of twitch duration discussed above. It is clear that further experiments are required to determine the mechanisms responsible for the effects of α₁-stimulation on twitch duration.

Effects of β-adrenoceptor stimulation on isometric and work-loop contractions

In both work-loop and isometric contractions (Figs 6 and 7), β-adrenoceptor stimulation with isoprenaline produced a dramatic increase in developed force and markedly reduced twitch duration, largely by accelerating relaxation. The cellular mechanisms responsible for the effects on the isometric twitch are well established (reviewed by Bers, 1992). The potentiation of force is thought to result from protein kinase A (PKA)-mediated phosphorylation of the L-type Ca²⁺ channel, which increases I_Ca, resulting in enhanced Ca²⁺ loading of the cell and of the SR and a greater trigger for Ca²⁺-induced Ca²⁺ release from the SR. The acceleration of twitch relaxation is probably due to PKA-mediated phosphorylation of (i) phospholamban, thereby increasing the rate of SR Ca²⁺ uptake, (ii) myofilament proteins such as troponin-I, leading to faster cross-bridge cycling (e.g. Hoh et al. 1988; Fentzke et al. 1999) and (iii) the RyR, shortening the period for Ca²⁺ release from the SR (Valdivia et al. 1997). In our experiments the positive inotropic effect of β-stimulation was greatest at the higher stimulation frequencies (> 2.5 Hz), i.e. the force-frequency relationship shifted to the right (Fig. 6B). Such a shift has been observed previously (e.g. reviewed by Ross et al. 1995). One explanation for this shift may be that the phosphorylation of phospholamban would be expected to increase the rate of SR Ca²⁺ uptake at all frequencies. As a result, the frequency at which the reduced time for SR Ca²⁺ uptake begins to limit the SR loading capacity as frequency is increased (see above) would be higher during β-stimulation than in control conditions. The increased rate of cross-bridge cycling could also contribute to the shift, as the faster relaxation would counteract a tendency for relaxation at high frequencies to be incomplete and hence decrease the calculated active force.

In work-loop experiments β-adrenoceptor stimulation increased the net work and power output of the trabeculae and markedly increased f_opt (from 2.9 to 4.4 Hz). The increase in maximum power output with β-stimulation (129 %) was much greater than that generated by α₁-stimulation (59 %). This larger increase in power output can be attributed to the greater increase of both net work (38 %) and frequency for net work (76 %) produced by β-stimulation (Fig. 5B and D) compared with α₁-stimulation (increases of 27 and 18 %, respectively; Fig. 5A and C), since power is determined by both components. In turn, the increased frequency for net work with β-stimulation can be attributed to the rightward shift of the force-frequency relationship (Fig. 6B) and to the shortening of the twitch (Fig. 4), which allows the muscle to operate effectively at higher frequencies. Both the increase in force at higher frequencies and the acceleration of the twitch were more marked for β-stimulation than for α₁-stimulation, so accounting for the differing effects of these drugs on the work- and power-frequency relationships. In a previous study (Layland et al. 1997), the α- and β-agonist adrenaline was found to increase the power output and f_opt in rat papillary muscles during work-loop contractions. From the present results we suggest that these effects of adrenaline were predominantly due to the effects of β-adrenoceptor stimulation, although there would also be some contribution from α₁-adrenoceptor stimulation.

Our washout experiments suggested that the effects of isoprenaline on the power-frequency relationship were only partially reversed after a washout of 30 min. The reasons for this are unknown but could involve maintained phosphorylation of myofilament proteins, since it has been shown that isoprenaline washout for 15 min results in dephosphorylation of phospholamban but no significant dephosphorylation of either C-protein or troponin-I (Garvey et al. 1988). It is not clear why the effects of α₁-stimulation were also only partly reversible.

It is interesting to note that the frequency-dependent reduction in twitch duration observed in control conditions was largely abolished during β-stimulation (Fig. 7C). As described above, the mechanism responsible for the frequency-dependent acceleration of the twitch and Ca²⁺ transient remains unclear. Data presented here implicate the stimulation of SR Ca²⁺ uptake, since the frequency-dependent effect may be abolished if the SR Ca²⁺-uptake rate was already maximally stimulated due to prior phosphorylation of phospholamban in response to β-stimulation.

Conclusion and physiological significance

From previous studies it would seem that the effects α₁- and β-adrenergic stimulation on cardiac contractility are often conflicting, for example: (i) α₁-stimulation may prolong twitch duration while β-stimulation shortens it, (ii) α₁-stimulation may reduce the cross-bridge cycling rate while β-stimulation increases it, and (iii) α₁-stimulation may have a reduced effect at higher stimulation frequencies whereas β-stimulation has a greater effect. In the physiological situation, catecholamine release is initiated under conditions of stress or exercise when there is a physiological demand to increase cardiac output. This is achieved by increases in the heart rate and the force of contraction. Since both α- and β-adrenoceptors are stimulated by catecholamines, it would seem detrimental to the survival of the animal if the effects of β-stimulation to accelerate relaxation and increase force at higher frequencies were partly negated by effects of α₁-stimulation to slow relaxation and reduce force at higher frequencies. Our findings with regard to the frequency-dependent effects of α₁- and β-stimulation on the force-, work- and power-frequency relationships offer an interesting new perspective. We provide evidence to suggest that the positive inotropic effect of α₁-stimulation may be preserved at higher frequencies and that, at these frequencies, α₁-stimulation may actually accelerate relaxation rather than prolonging it. This would be beneficial in the physiological situation where α₁- and β-adrenoceptors are stimulated simultaneously since both types of response would then contribute to increasing cardiac performance under conditions of increased demand. The overall contractile response to catecholamines is likely to be dominated by β-adrenergic effects, however, since the effects of β-adrenoceptor stimulation on force-, work- and power-frequency relationships were more pronounced than corresponding α₁-mediated effects.

The increase in f_opt produced by isoprenaline matches the increased heart rate of rats treated with β-agonists (e.g. Irlbeck & Zimmer, 1993). This will help to maintain the efficiency of the heart during sympathetic stimulation of the heart in vivo, in that activation of myocardial adrenoceptors (α₁ and β) by adrenaline or noradrenaline will not only increase heart rate but will provide a matching increase in the frequency for optimal power output of the myocardium.