Altered Calcium Homeostasis Does Not Explain the Contractile Deficit of Diabetic Cardiomyopathy

All animal use was approved by the University of Auckland animal ethics committee. Chemical reagents were obtained from Sigma unless otherwise stated.

Diabetes was induced by a single injection of STZ (60 mg/kg dissolved in 0.9% saline administered intravenously) in adult male Wistar rats (200–250 g). Control rats received injections of 0.9% saline alone. Blood glucose concentrations were measured 48 h after STZ injection (Advantage II; Roche) and weekly thereafter. Diabetes was diagnosed by a sustained glucose concentration >11 mmol/l. Insulin was not administered, and all animals had free access to food and water.

Blood pressure measurements were carried out on conscious animals using a tail-cuff method (MC4000 BP; Hatteras Instruments, Cary, NC) after first habituating animals to the procedure. Eight weeks post-injection, animals were anesthetized (5% halothane initially, maintained at 2% concentration in O₂) and the electrocardiogram (ECG) recorded (PowerLab data acquisition system; ADInstruments, Bella Vista, Australia). QT interval was measured from the beginning of the QRS complex to the end of the T wave, averaged over 6–10 cardiac cycles, and normalized to the square root of RR interval (25).

The day after ECG measurements, the animals were again anesthetized, the hearts removed and immersed in 4°C Krebs-Henseleit buffer (KHB) containing, in millimoles per liter: NaCl (118), KCl (4.75), KH₂PO₄ (1.18), MgSO₄ (1.18), NaHCO₃ (24.8), and D-glucose (10) with 2,3-butanedione monoxime (5) and CaCl₂ (0.5). Hearts were subsequently perfused at room temperature (KHB, gassed with 95% O₂ and 5% CO₂). Spontaneous action potentials were recorded from LV free wall using glass microelectrodes (tip resistance 4–7 MΩ filled with 3 mol/l KCl) and a Neuroprobe amplifier (Model 1600; A-M Systems, Carlsborg, WA) digitized at 10 kHz and acquired using a PowerLab. Action potential duration (APD) was analyzed using IDL software (IDL research Systems, Boulder, CO). APD₅₀ and APD₉₀ were measured as the time taken for 50 and 90% repolarization of the action potential, respectively. A small correction for varying heart rate was carried out using the square root of the corresponding RR interval, by means similar to those used for the measurement of the QT interval described above.

An unbranched cylindrical trabecula was dissected from the left ventricle while the heart was perfused with oxygenated KHB containing 20 mmol/l l 2,3-butanedione monoxime and 0.25 mmol/l CaCl₂. The trabecula was then transferred to a Perspex bath on the stage of an inverted microscope (Nikon Diaphot 300; Nikon, Tokyo, Japan) and continuously superfused with oxygenated KHB. One end of the preparation was mounted in a hook extending from a force transducer (AE801; SensorOne, Sausalito, CA), with the other end held in a fine snare protruding from a stainless-steel tube. Following a 30-min stabilization period, the trabecula was field stimulated at 0.1 Hz with 5 ms square-wave pulses (20% above threshold) delivered by a Digitimer D100 (Digitimer, Welwyn Garden City, Hertfordshire, U.K.) and extracellular Ca²⁺ ([Ca²⁺]_o) gradually increased to 1.0 mmol/l. Isometric force was normalized to the trabecula cross-sectional area and expressed as stress. LV trabeculae were not different in either length (diabetic 1.4 ± 0.3 mm, n = 9; control 1.5 ± 0.2 mm, n = 9) or cross-sectional area (diabetic 0.023 ± 0.008 mm², n = 9; control 0.017 ± 0.003 mm², n = 9) between groups.

Trabeculae were loaded with the fluorescent Ca²⁺ indicator fura-2/AM (20 μmol/l) at room temperature for 2 h. On completion of loading, the indicator was washed out and sarcomere length adjusted to 2.1–2.2 μm. Trabeculae were illuminated at 340-, 360-, and 380-nm excitation wavelengths using a spectrophotometric system (Cairn, Faversham, U.K.) and the ratio of the emitted fluorescence with 340- and 380-nm excitation was used as a measurement of [Ca²⁺]_i.

Data were acquired using LabView software (National Instruments, Austin, TX) and analyzed offline. Unbiased measurement of Ca²⁺ transient and force was made from data averaged over 7–10 cardiac cycles, using a custom analysis program written in IDL. A five-parameter function was used to define the properties of the averaged Ca²⁺ transient, as previously described (21).

Frozen LV tissue was homogenized in ice-cold lysis buffer (50 mmol/l Tris-HCl, pH 8; 150 mmol/l NaCl; 1% NP-40; 0.5% sodium deoxycholate; and 0.1% sodium dodecylsulphate) with a proteinase inhibitor cocktail (Roche, Indianapolis, IN) and then centrifuged at 13,000g for 1 h at 4°C. Protein concentration in the supernatant was determined with a bicinchoninic acid protein assay (Pierce, Rockford, IL). A total of 20 μg protein was used for Western blots using rabbit anti-NCX antibody (Ab) (1:500; Swiss Antibodies, Swant, Switzerland) and mouse anti–sarco(endo)plasmic reticulum Ca²⁺-ase (SERCA)2a Ab (1:1,000; Affinity BioReagents, Golden, CO). Specific binding was detected with a donkey anti-rabbit or anti-mouse IgG–horseradish peroxidase conjugate (GE Healthcare, Buckinghamshire, U.K.) and ECL Plus Western blotting detection reagents (GE Healthcare) according to the manufacturer's instructions. An LA 3,000 image reader (Fuji Photo Film, Tokyo, Japan) was used to detect signals and take images of the membranes. Images were analyzed using Multi Gauge software (Fuji Photo Film).

Tissue for immunohistochemistry, frozen at −80°C, was sectioned at 30 μm. The sections were dried (1 h), fixed (ice-cold acetone, 10 min), and rehydrated (PBS, 5 min). Sections were then blocked (60 min, 10% goat serum in PBS) and immunolabeled (4°C) with rabbit anti-rat type I and type III collagen Ab (1:100 in 5% goat serum/1% BSA/PBS; Chemicon, Boronia, Australia). After washing (PBS), slides were stained for 2 h with secondary Abs (1:200) (Alexa Fluor 488 goat anti-rabbit Ab; Invitrogen) and phalloidin-rhodamine (1:50–100, Invitrogen) diluted in 5% goat serum/1% BSA/PBS at room temperature. Slides were washed (PBS), mounted (antifade; Dako), and imaged with a confocal microscope (Zeiss LSM410). Three transverse optical sections from the LV wall adjacent to the endocardium were analyzed from each animal (n = 7, per group). F-actin and collagen (types I and III) were quantified using Image J (http://rsb.info.nih.gov/ij) and a custom-written IDL program based on signal thresholding. Results (means ± SEM) were expressed as percentage of cross-sectional area. Differences between means were analyzed using either ANOVA or Student's t test, as appropriate, and were considered significant at P < 0.05.

The general characteristics of diabetic rats compared with their age-matched control counterparts are shown in Table 1. Diabetes was confirmed by a significant (approximately sixfold) elevation in blood glucose, 3 days after STZ injection, that was maintained. Eight weeks post-injection, diabetic rats had decreased body weight, tibial length, and heart weight, but their heart weight–to–body weight ratio was increased. In contrast, heart weight normalized to tibial length was not different between groups (Table 1).

Diabetic rats had a slower heart rate under anesthesia and a prolonged QT interval both before and after normalization for heart rate (diabetic 192 ± 4 ms, n = 18; control 177 ± 4 ms, n = 11; P < 0.05). Consistent with the longer QT interval, the LV action potential was also prolonged in diabetic rats. These results are summarized in Table 1.

Intracellular [Ca²⁺]_i (fura-2 340/380 ratio) and isometric force were measured in LV trabeculae under experimental conditions that mimicked those in vivo (1.5 mmol/l [Ca²⁺]_o, 37°C, and 5 Hz). Figure 1A shows five consecutive cardiac cycles recorded from representative control (solid lines) and diabetic (dotted lines) trabeculae, superimposed for comparison. Parameters describing these data are summarized in Table 2. No difference was found in peak or resting [Ca²⁺]_i between groups, but the kinetics of the Ca²⁺ transient were slower for diabetic rats (increased time to peak [Ca²⁺]_i and time constant of Ca²⁺ transient decay; P < 0.05). In comparison with their control counterparts, diabetic rats had decreased peak and developed stress with a slower time course (increased time to peak stress and time to 90% relaxation of stress; P < 0.05). Figure 1B shows superimposed Ca²⁺ transients (i), twitches (ii), and phase plots (iii), obtained by combining averaged data from seven trabeculae per group. The [Ca²⁺]_i-stress relationship (Fig. 1Biii) shows a right-shifted relaxation phase for diabetic rats with decreased slope (diabetic 29 ± 7 mN mm⁻²/ratio unit, n = 8; control 67 ± 17 mN mm⁻²/ratio unit, n = 8; P < 0.05), suggesting that myofibrillar Ca²⁺ sensitivity was reduced.

Increased cytoplasmic Ca²⁺ can paradoxically reduce sarcoplasmic reticulum (SR) Ca²⁺ content when it increases the probability and frequency of spontaneous Ca²⁺ release from the SR during diastole (26,27). This is an important factor for in vitro studies in the rat, as it has been reported that the rat is highly dependent on SR Ca²⁺ stores for contraction (28) and susceptible to spontaneous SR diastolic Ca²⁺ release (27). To examine the possibility that this effect might be altering Ca²⁺ release, trabeculae were exposed to a range of extracellular Ca²⁺ concentrations from 0.5 to 5 mmol/l. There was no difference between groups for either resting [Ca²⁺]_i (Fig. 2A) or resting stress (Fig. 2B) at all [Ca²⁺]_o. Although peak [Ca²⁺]_i and peak stress increased with increasing [Ca²⁺]_o in both groups, peak stress was reduced in diabetic animals at all [Ca²⁺]_o (P < 0.05; Fig. 2A and B). In contrast, there was no difference in peak [Ca²⁺] between groups, consistent with the lack of a difference in resting [Ca²⁺]_i.

Figure 2C examines the relationship between peak [Ca²⁺]_i and peak stress and clearly shows that, regardless of the peak [Ca²⁺]_i, diabetic rats had decreased responsiveness to [Ca²⁺]_i. It is well known that the time course of the Ca²⁺ transient affects force responses and that a decrease in the duration of the transient decreases force. Figure 2C shows that regardless of the peak [Ca²⁺]_i reached, diabetic trabeculae developed less stress, whereas Fig. 2D shows that they also had a slower Ca²⁺ transient (which would be expected to increase force production [29]). The reduced slope of the relationship between peak [Ca²⁺]_i and peak force (Fig. 2C) is consistent with diabetic rats having a diminished myofibrillar Ca²⁺ sensitivity.

In cardiac muscle, force is also modulated by stimulation frequency, and as noted above, most in vitro investigations have been carried out at low stimulation frequencies. We examined the response of trabeculae to stimulation frequencies in the range 0.1–7 Hz (Fig. 3). Trabeculae from control rats (n = 6) showed a biphasic response for developed stress and Ca²⁺transient amplitude with increasing stimulation frequency (Figs. 3A and 3B). There were no differences in peak or resting [Ca²⁺]_i between diabetic and control rats, except at 7 Hz, where resting [Ca²⁺]_i was increased in diabetic animals (Fig. 3A). The positive phase of the force-frequency response was diminished in diabetic rats (n = 6; Fig. 3B) such that, despite the increase in the Ca²⁺ transient with frequency, force did not follow. Both diabetic and control rats showed frequency-dependent acceleration in the decay of the Ca²⁺ transient (Fig. 3C) and force relaxation (Fig. 3D), but these parameters were always slower for diabetic rats (P < 0.05).

Taken together, these results show that the reduced contractility of diabetic rat hearts cannot be explained on the basis of an altered [Ca²⁺]_i, although there were changes in Ca²⁺ homeostasis that slowed the Ca²⁺ transient.

The rate of decline of the Ca²⁺ transient is dominated by the rate of Ca²⁺ uptake by the SR and, to a lesser extent, sarcolemmal Ca²⁺ extrusion by the Na⁺/Ca²⁺ exchanger (NCX), which contributes ∼15% (30). Figure 4 shows Western blots of SERCA2a and NCX protein expression. Diabetic rats had lower levels of SERCA2a expression (P < 0.05), whereas NCX was not significantly altered. This result suggests that the origin of the slowed Ca²⁺ transient in diabetic hearts resides in a decrease in the SR uptake capacity.

Because cardiac remodeling may contribute to diabetic cardiomyopathy (23,24), we examined the extracellular collagen content and the organization of sarcomeric thin filaments (F-actin). Figure 5 shows images of typical antibody-labeling patterns for type I collagen (green) and phalloidin-labeled F-actin (red) in diabetic and control rats. Diabetic hearts showed increased type I collagen per endocardial cross-sectional area (diabetic 7.1 ± 0.1% of area, control 6.2 ± 0.2% of area; P < 0.01) without significant change in type III collagen (diabetic 6.5 ± 0.1% of area, control: 6.2 ± 0.1% of area). Diabetic rats also showed a more disorganized F-actin structure with decreased F-actin content (diabetic 56.9 ± 0.6% of area, control 61.7 ± 0.4% of area; P < 0.0001).

Diabetic cardiomyopathy or “diabetic heart muscle disease” starts as an asymptomatic slowing in relaxation kinetics (diastolic dysfunction) (31). As the syndrome progresses, systolic function also becomes compromised, which increases morbidity and mortality (32). In the present study, STZ-injected rats were used as a model of diabetic cardiomyopathy (8) and cardiac function, evaluated in relation to cellular [Ca²⁺]_i homeostasis, electrical activity, and remodeling of the left ventricle.

Diabetic rats had a decreased heart rate with depressed SR Ca²⁺-uptake capacity and no detectable change in NCX. Although the cause of the decreased SA node discharge rate was not examined, our data suggest a possible explanation. SA nodal cell action potential generation is promoted by spontaneous SR Ca²⁺ release, which produces inward current by NCX (Maltsev et al. [33]). In our model, a reduced ability of the SR to take up Ca²⁺ might delay the time taken for the SR to reach the level of Ca²⁺ that promotes spontaneous release. This would therefore reduce SA nodal cell discharge rate, as observed.

At 1.5 mmol/l [Ca²⁺]_o, 37°C, and 5 Hz, which are close to physiological conditions, trabeculae from diabetic rats showed depressed contractility with prolonged time for contraction and relaxation (Table 2). This result is consistent with previous studies from our group using the isolated working heart (34) and from other research groups, albeit under different conditions (8,17,35). Altered intracellular Ca²⁺ homeostasis has previously been suggested as underlying diabetic cardiac dysfunction (Pierce and Russell [7] and Cesario et al. [36]) although, as noted, results are inconsistent. Although some of these discrepancies might be attributable to the extent of disease progression (diabetic stage) and experimental conditions, few studies have examined [Ca²⁺]_i and the control of contractility under near-physiological temperatures and rates of stimulation. This problem is illustrated in Fig. 3; at 1 and 2.5 Hz, there was no detectable difference in developed stress between groups, despite contractility being very much reduced at all other frequencies tested in diabetic rats. In the present study, experiments were performed from a base of 37°C and 5 Hz, and our data showed that diabetic rats had an unchanged resting [Ca²⁺]_i level, maximum rate of rise in fluorescence, and Ca²⁺ transient amplitude, despite showing reduced contractility. These results are consistent with some previous studies (14,15) performed on isolated cardiomyocytes at 35–37°C and 1 Hz. In contrast, results from other studies showed a reduced Ca²⁺ transient amplitude or peak [Ca²⁺]_i (11,16–18), but it should be noted that those experiments were performed at low temperature (21–25°C) and low stimulation frequencies (0.2–0.5 Hz).

Our results also showed that diabetic rats had prolonged time to peak [Ca²⁺]_i and prolonged time constant of Ca²⁺ transient decay (Table 2), consistent with some other studies (6,11,14–16). Although the time to peak [Ca²⁺]_i was slower, the rate of rise of the Ca²⁺ transient was not significantly altered. The latter observation suggests that the trigger for Ca²⁺ release (I_Ca) was not reduced, with the delayed peak of the Ca²⁺ transient likely due to the prolonged action potential duration. The reduction in the level of expression of SERCA2a (∼50%), as detected by Western blots (Fig. 4), appears sufficient to explain the reduction (∼20%) in the rate of decline of the Ca²⁺ transient. The slower kinetics of the Ca²⁺ transient would contribute to the prolonged time course of cardiac contraction and relaxation in diabetic rats, but it is unclear whether the reduced rate of the decay in the Ca²⁺ transient is sufficient to explain the slowed mechanical relaxation. By comparing contractions between normal and diabetic rats at different stimulation rates, we can select contractions that have similar rates of Ca²⁺ transient decline to isolate the effect on contraction. For example, at 5 Hz, the diabetic rat Ca²⁺ transient decreases slightly faster than that of the control rat at 2.5 Hz (Fig. 3C). However, the time to 90% mechanical relaxation of the diabetic rat at 5 Hz (Table 2) is about 80 ms, whereas that of the control rat takes about 60 ms at 2.5 Hz. Similarly, for 50% mechanical relaxation, the times were 39 ms and 32 ms (for diabetic rats at 5 Hz and control rats at 2.5 Hz, respectively). From these comparisons, we suggest that the mechanical relaxation is intrinsically slower in diabetic animals and exacerbated by the reduced rate of decrease of [Ca²⁺]_i. In support of this idea, we also observed an increase in the time to peak stress in diabetic rats over and above that expected from the delayed time to peak [Ca²⁺]_i.

The slower kinetics of the Ca²⁺ transient in diabetic rats may also be affected by the prolonged APD (Table 1), altered intracellular Ca²⁺ handling, or the combination of both. It is well known that maintained depolarization slows the return of [Ca²⁺]_i to resting levels (37), an effect that is partly due to inhibition of Ca²⁺ extrusion by NCX. On the other hand, slowed SR Ca²⁺ uptake should lead to prolonged depolarization due to inward current generation by forward-mode NCX. The latter explanation seems likely because we found decreased levels of SERCA2a expression in the diabetic rat. Decreased SERCA2a expression should alter the balance of Ca²⁺ homeostasis in favor of increased Ca²⁺ extrusion via NCX and might be expected to reduce SR Ca²⁺ content. However, Ca²⁺ extrusion via NCX during the decline of the Ca²⁺ transient will lead to Na⁺ entry, which, during the diastolic period, will increase resting [Ca²⁺]_i and help restore SR Ca²⁺ levels. This is consistent with our observation of a slightly increased resting [Ca²⁺]_i in diabetic rats (at least at 0.1 Hz and varied [Ca²⁺]_o), although detailed modeling will be required to determine whether the change in SR Ca²⁺ uptake is by itself sufficient to explain all the effects on the APD and resting [Ca²⁺]_i.

Although we found almost no effects on peak Ca²⁺ in a variety of conditions, the finding that there was a reduced rate of SR Ca²⁺ uptake and SERCA2a expression was surprising, since these changes might be expected to decrease SR Ca²⁺ content (and therefore the amplitude of the Ca²⁺ transient). Simple interpretation of changes in Ca²⁺ transient kinetics is complicated by the interaction of some Ca²⁺ transport mechanisms with changes in membrane potential during the action potential. To separate these two effects, a logarithmic analysis of the decline of the Ca²⁺ transient is shown in Fig. 6. It is clear that there is a delay before the decline of fluorescence becomes exponential. It is notable that this delay is longer in diabetic rats and that the duration corresponds closely to the time for 50% action potential repolarization. After the delay, the rate of decline of the Ca²⁺ transient was slower in diabetic rats, as described above. The slowed initial phase of the decrease in the Ca²⁺ transient must be due to either continued sarcolemmal Ca²⁺ influx or a failure for SR release to terminate (because dCa_i/dt = J_influx − J_efflux). The latter factor would also be expected to reduce SR Ca²⁺ content, but because the amplitude of the Ca²⁺ transient was not reduced, increased sarcolemmal Ca²⁺ influx during this period seems more likely. Because the 50% repolarization potential corresponds to a time when Ca²⁺ influx via l-type Ca²⁺ channels terminates (38), the increased duration of the diabetic action potential would be expected to increase Ca²⁺ influx and may therefore explain why the Ca²⁺ is not decreased, in spite of a decreased SR uptake capacity.

Phospholamban is an endogenous inhibitor of SERCA activity in cardiac muscle, especially at low [Ca²⁺]_i, and might contribute to the differences in the decay of the transient between groups. However, the inhibitory effects of phospholamban are calcium dependent; as [Ca²⁺]_i is increased, the effects are diminished. Figs. 2D and 3D show the time constant of Ca²⁺ decay as a function of the peak fluorescence for both diabetic and control trabeculae under experimental conditions that alter [Ca²⁺]_i. Linear regression analysis for both datasets (Figs. 2D and 3D) showed that the slopes of the regression lines were either steeper (Fig. 2D) or not different (Fig. 3D) for diabetic hearts, suggesting that decreased phospholamban does not explain the slower decay of [Ca²⁺]_i in diabetic hearts.

Diabetic rats showed depressed cardiac contractility, even when peak [Ca²⁺]_i was matched by altering stimulation rate (Fig. 3A and 3B), suggesting that altered [Ca²⁺]_i handling was not the primary cause of contractile dysfunction. Because depressed cardiac myofibrillar ATPase activity occurs in diabetic cardiomyopathy (39), it might seem obvious to attribute the contractile dysfunction to this effect. However, we also observed a ∼8% reduction in the fractional area of actin filament (F-actin) per unit cross-section that will also contribute to a reduction in active stress. The reduced F-actin labeling per cross-section may be explicable on the basis of decreased protein synthesis, increased protein degradation (40), and reduced α-actin mRNA expression (41). More importantly, F-actin no longer showed a highly organized sarcomeric distribution in diabetic hearts (Fig. 5D). Activation of cardiac muscle myofilaments involves a series of complex interactions between the troponin complex, tropomyosin, and F-actin. The contractile deficit might therefore arise, in part, from an overall reduction in the number of myosin-binding sites available for crossbridge formation. Additionally, changes to the sarcomeric organization of F-actin might also explain the reduced myofilament Ca²⁺ sensitivity in diabetic trabeculae.

Diabetic cardiomyopathy is also associated with increased stiffness in the left ventricle (42). Types I and III collagen are the most abundant collagen types in the heart, forming 85 and 11% of the total collagen content, respectively (24). Our results show an increase (14%) in type I collagen in diabetic rats but no change in type III. However, any diastolic dysfunction arising from increased collagen would be compounded by the reduced rate of decline of the Ca²⁺ transient and its effects on force production.

In summary, under physiological conditions, we find no evidence to support the idea that altered myocardial Ca²⁺ handling underlies the decrease in contractile force in diabetic cardiomyopathy. Although decreased SR Ca²⁺ uptake exists, it is offset by an increase in action-potential duration, thereby ensuring that SR Ca²⁺ content is maintained. Instead, our data indicate that the contractile function is compromised by reduced and disrupted F-actin in conjunction with the increased type I collagen. In addition, we found evidence for a decreased myofilament sensitivity and slowed relaxation independent of the changes in the Ca²⁺ transient. Although the contribution of the observed structural changes to the contractile deficit in the diabetic heart cannot be evaluated at this time, such changes deserve careful consideration in future studies.

This work was supported by the Health Research Council of New Zealand.