In this study, we present the first results of STZ-induced diabetes on the energetics of isolated left-ventricular (LV) trabeculae, acknowledging that diabetes prolongs twitch duration. We measured mechanical work output and heat production. Several additional first results from diabetic cardiac preparations are also presented: (1) twitch duration as a function of stress or afterload, (2) heat-stress and heat-STI (stress-time integral) relations obtained under isometric contractions, (3) dynamic modulus as a function of isometric twitch stress, and (4) heat-afterload, work-afterload and efficiency-afterload relations derived from work-loop contractions. Given these results, we are in a position to reconcile the performance of the isolated cardiac muscle with that of the intact heart.
Relation of trabecula performance to that of the whole heart
Diabetes prolongs twitch duration in both isolated trabeculae and the heart, but contractile dysfunction prevails only in the heart. In our recent study of isolated, working, hearts , we showed that the maximal afterload achievable by the diabetic heart (i.e., when its aortic outflow is near zero) is substantially lower than that of Control hearts. Thus, at high afterloads, the work output of the diabetic heart is compromised. In contrast, diabetes does not affect the stress development of isolated trabeculae (Figure 2C) nor any other index of mechano-energetic performance, with the single exception of twitch duration, which is prolonged in diabetic preparations under both isometric (Figure 3) and work-loop (Figure 9) protocols. Because of the comparability of mechanical performance, the efficiency-afterload curve of the diabetic trabeculae is the same as that of the Control group (Figure 11F). Note that the null effect of diabetes on the absolute stress production of trabeculae allows us to express efficiency as a function of relative afterload.
When a trabecula is isolated from a diabetic heart, it is freed from the effects of insufficient ventricular filling. It can now be stretched to L
, where its stress development is found to be as high as that of healthy Control trabeculae (Figure 2C). This finding gives us confidence that sub-optimal ventricular filling, as a consequence of prolonged twitch duration and attendant abbreviation of diastolic interval, is sufficient to explain LV contractile dysfunction in the heart. Hence, twitch prolongation alone is sufficient to develop LV contractile dysfunction at high afterloads. That is, prolonged twitch duration reduces the period, and hence the extent, of diastolic refilling, leading to failure to generate the high pressures required to overcome high afterloads. This, however, does not compromise peak stress development at the level of isolated myocardial tissue. Consequently, diabetes does not affect the efficiency-afterload relation of isolated trabeculae, as hypothesised.
To explain further the apparent contradiction between the contractile performance of the heart and isolated trabeculae, it is necessary to consider four facts: (1) twitch duration increases with increasing muscle length (Figure 3), as well as with afterload (Figure 9), (2) for a given pacing rate, prolongation of twitch duration is associated with a reduction of the diastolic period (Figure 2A inset and Figure 8A), i.e., reducing the time for diastolic filling of the ventricles or re-lengthening of isolated trabeculae, (3) the extent of reduction of the diastolic period increases with increasing stimulus frequency (Figure 6A) or heart rate, and (4) the reduced intrinsic heart rate of the diabetic animal [6, 32, 40, 42, 44, 50].
Consideration of these facts requires comparison of twitch duration between the diabetic heart and the Control heart at the same rate of stimulation. This can be achieved by externally pacing the heart. In our heart study, we paced the hearts to beat at 4 Hz, which is above their intrinsic rate at 32°C: 2.6 Hz and 2.9 Hz for the STZ and Control groups, respectively . We did not measure twitch duration of the heart and are unaware of any study that has paced the diabetic heart at the same rate as the Control heart in order to quantify the diabetes-induced prolongation of twitch duration. However, we can make use of the trabecula results of the current study in order to draw an inference concerning the contractile dysfunction of the diabetic heart at high afterloads. We electrically stimulated diabetic trabeculae to contract at the same rate as their controls (3 Hz). By doing so, we saw a diabetes-induced prolongation of twitch duration (Figure 3), consistent with the results of others in isolated papillary muscles [2, 3, 28, 30, 34–37], ventricular trabeculae [6, 12], and single myocytes [13, 31, 38, 39]. We infer that the same behaviour occurs in the diabetic heart.
Collectively, our results allow us to infer that the diastolic filling time of the diabetic heart is reduced, and is disproportionately reduced as the afterload challenge is increased. Hence, at sufficiently high afterloads, the diabetic heart suffers inadequate ventricular filling and, consequently, reduced aortic outflow. The healthy heart, in contrast, can pump to a higher afterload given its longer period of diastolic filling.
Frequency-dependence of peak active stress
The null effect of diabetes on peak active stress of isolated trabeculae renders our hypothesis valid. Many previous studies have also reported a null effect of diabetes on peak active stress (i.e., when the muscle is held at L
) at stimulus frequencies ≤ 1 Hz in isolated papillary muscles [2, 3, 28, 30, 34–37] and in isolated single myocytes [46, 47, 49]. But, there are several studies showing lower contractility of diabetic myocytes at stimulus frequencies ≤ 1 Hz [31, 38, 57] as well as at 2 Hz . The reason for these discrepant literature findings is unclear given that these experiments were performed at comparable temperatures (30°C – 37°C). Given these ambiguous literature reports, we compared the results of four independent studies which examined the effect of stimulus frequency at a fixed temperature: Cameron et al.  and Nobe et al.  on isolated papillary muscles, Zhang et al.  on isolated LV trabeculae, and Ren and Davidoff  on isolated single myocytes. The first study showed no effect of diabetes on peak active stress at 30°C and at a range of stimulus frequency between 0.1 Hz and 4 Hz. The second reported no effect of diabetes on peak active stress at 36°C between 0.2 Hz and 5 Hz. The third study, at 37°C, showed no effect of diabetes at 1 Hz or 2.5 Hz, but lower values at 5 Hz and 7 Hz. Lastly, the fourth study, also at 37°C, found the differential effect of diabetes to disappear at 5 Hz, but not at frequencies below 2 Hz. We have no explanation for the discrepant findings between the latter studies. We conclude that the effect of diabetes on peak active stress production appears to be dependent on stimulus frequency. Our results, at 32°C, show no effect of diabetes on the peak active stress production at 3 Hz stimulation, but a negative effect at 6 Hz (Figure 7B).
Frequency-dependence of diastolic stress
At 6 Hz, relaxation of the twitch is incomplete in both Control and diabetic trabeculae (Figure 6A), resulting in the elevation of diastolic stress between successive twitches. We have previously shown  that, in healthy trabeculae, incomplete relaxation is initiated by elevation of diastolic intracellular Ca2+ and a subsequent decreased myofilament sensitivity to Ca2+, and is not due to inadequate supply of either glucose or oxygen. Compared with the Control trabeculae, the diabetic trabeculae experience greater diastolic stress (Figure 7A), indicating that they fail to relax between twitches as fully as that of the Control. The inability of the diabetic trabeculae to achieve complete relaxation at 6 Hz is exacerbated by their prolonged twitch duration. The diabetic trabeculae have less time to relax before the next twitch commences, and consequently, they experience a greater extent of incomplete twitch relaxation. This result implies that the diastolic intracellular Ca2+ at 6 Hz is greater in the diabetic trabeculae, and the decrease of sensitivity of myofilaments to Ca2+ is more severe in the diabetic trabeculae. The latter implication is consistent with the findings of Zhang et al.  and Op Den Buijs et al.  showing a diabetes-induced decrease of Ca2+ responsiveness.
Dynamic stiffness as a probe of cross-bridge function
Because of the well-documented negative effect of diabetes on actomyosin ATPase activity (see above), we invoked the technique of high-frequency, low-amplitude oscillation of muscle length in order to interrogate crossbridge function. We measured dynamic modulus (dynamic stiffness normalised to muscle dimensions) to quantify the status of crossbridges (i.e., the number attached and their individual stiffness) throughout the time course of twitch stress production.
We find no effect of diabetes on dynamic modulus (Figure 6). The linear relation between dynamic modulus and active stress (which implies that the net number of attached crossbridges changes linearly with stress production) is unaffected by diabetes. The maximal and minimal values of dynamic modulus, obtained at the peak stress and diastolic stress, respectively, are also unaffected by diabetes. Our modulus result, at high perturbation frequency (100 Hz), is consistent with that of Metzger et al. , who demonstrated that the modulus-tension relations are similar between control and β-MHC-expressing ventricular myocytes of the hypothyroid rat. These authors inferred that “… force production per strong crossbridge interaction, or the distribution of force-generating crossbridge states, is not cardiac MHC isoform dependent”. Thus, our results suggest that, although diabetes prolongs twitch duration, it does not affect the net number of crossbridges attached or their individual stiffness. This inference further suggests that the contractile dysfunction at the whole heart level is predominantly due to insufficient LV diastolic filling, and not to crossbridges status per se.
During a train of isometric contractions, trabeculae perform negligible external work, and hence the metabolic change of enthalpy (heat plus work) consists almost entirely of heat. We plotted isometric heat as functions of both developed stress and STI (Figure 5). Both relations have previously been used for studying the mechano-energetic of healthy, non-diabetic, isolated papillary muscles [60, 61] and trabeculae [53, 62] under a variety of experimental conditions. In the present study, we observed the heat-stress relation to be slightly curvilinear but the heat-STI relation to be linear. The extrapolated y-intercepts of these relations, which did not differ and were unaffected by diabetes, are presumed to estimate ‘activation heat’, i.e., the energy expenditure associated with Ca2+ cycling by the sarcoplasmic reticulum Ca2+-ATPase and Na+ extrusion by the sarcolemmal Na+-K+ ATPase. Note that the null effect of diabetes on activation heat obtained in this study is not at odds with reports showing decreased activities of both the sarcoplasmic reticular Ca2+-ATPase [17–22] and the sarcolemmal Na+-K+-ATPase [14–16] in diabetic cardiac preparations. This is because a decreased activity of an ATPase does not imply decreased metabolic energy expenditure, since the same total amount of heat could be produced independent of the rate of ATP hydrolysis.
Under the assumption that the activation heat is independent of developed stress, the monotonic increase of heat output with increasing stress reflects the metabolic energy expenditure of the contractile apparatus by the actin-activated myosin-ATPase. The inverse of the slope of the heat-stress relation (Figure 5B) is hence an index of crossbridge economy. The absence of a difference in the magnitudes of this index between the healthy and STZ-treated trabeculae implies that diabetes does not render the contractile apparatus less ‘economic’, despite reports showing reduced rate of ATP hydrolysis by myofibrillar-ATPases [23–28]. Since twitch duration is prolonged but twitch stress is unaffected, the area under the time-course of the twitch (i.e., its stress-time integral, STI) is increased in the diabetic preparations (Figure 4D). For a given value of active stress, heat production is unchanged in the diabetic preparations (Figure 5B). Given the effects of diabetes on twitch duration, twitch stress and twitch heat, the slope of the heat-STI relation is destined to be lower in the intact diabetic trabeculae, as confirmed in Figure 5D.
Our results, showing the effect of diabetes on the heat-stress relation of intact LV trabeculae at 32°C, are not comparable with those of Rundell et al.  who used skinned RV trabeculae at 20°C. Those authors reported decreased tension cost (indexed by the slope of the linear ATPase-tension relation) in the diabetic group, which they attributed to reduced expression of the α-MHC (fast) isoform. Studies by Holubarsch et al. [64, 65], using intact LV papillary muscles at 21°C, showed that hypothyroid rats (expressing β-MHC) have a lower slope of the heat-STI relation, in agreement with our result (Figure 5D). Thus, the use of intact versus skinned preparations may be responsible for the inconsistency of our heat-stress relation with that of the ATPase-tension relation of Rundell et al. .
Null effect of diabetes on muscle shortening
The heart does not ever perform purely isovolumic contractions. Rather, it reduces volume in the process of ejecting blood. During the ejection period, when the outflow valves are open, pressure and volume change continuously. We approximated the pressure-volume loops of the heart by subjecting each trabecula to a series of isotonic stress-length loops (Figure 8). Using the data obtained during the isotonic shortening phases of the stress-length loops, we quantified three parameters related to muscle shortening: (i) the peak velocity of shortening, computed as the maximal slope of the length-time trace during the isotonic shortening phase of the work-loop (Figure 8B), (ii) the power of shortening, which is the product of maximal velocity of shortening and active afterload, and (iii) the peak extent of shortening, calculated as the relative length at which the trabecula transitioned from isotonic shortening to isometric relaxation (Figure 8B and C). The latter (end-systolic) length corresponds to the end-systolic volume of the heart. We found that the peak extent of shortening, as well as its peak velocity (extrapolated to the y-intercept), was comparable between control and diabetic trabeculae (Figure 10B and D). Shortening power also did not differ between the two groups (Figure 10F). These results imply that shortening is unaffected by diabetes, and are thus in accord with the findings of many [13, 46–49] but not all [31, 38, 57] single-myocyte studies.
Our finding of an absence of effect of diabetes on shortening velocity is not consistent with those studies showing a shift of myosin heavy-chain expression from the α to the β isoform [20, 24, 26, 29, 30]. We note that the velocity of shortening of diabetic preparations is dependent on extracellular Ca2+ concentration. Fein et al.  found that muscle shortening, both its velocity and its extent, were similar between control and diabetic rat papillary muscles at a bath Ca2+ concentration of 0.6 mM. But at an elevated (and non-physiological) Ca2+ concentration (2.4 mM), the shortening velocities of the diabetic preparations were lower than those of the control group. Similarly, Siri et al.  showed the peak extent of shortening to be unchanged, but shortening velocity reduced, in diabetic-hypertensive rat papillary muscles tested with 2.4 mM Ca2+. Joseph et al.  also reported decreased peak shortening velocity in the diabetic papillary muscle at 2.5 mM Ca2+. Whether the activity of the MHC is Ca2+-dependent requires future experiments. Our results, at physiological Ca2+ (1.5 mM), show no effect of diabetes on muscle shortening.
Lastly, crossbridge efficiency, revealed by subtracting activation heat (extrapolated from the heat-relative afterload relation shown in Figure 11B) from the denominator of the expression for mechanical efficiency, is also indifferent to diabetic status (Figure 11H). The null effect of diabetes on crossbridge efficiency is consistent with that reported by Joseph et al. . These authors calculated, using their experimental force and velocity data, together with several assumptions about crossbridge characteristics and energetics, a value of crossbridge efficiency of 30%, in agreement with our experimentally measured values (Figure 11H).