INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE CLASSIC IDEA on energy homeostasis in muscle was that ADP and P_i, the breakdown products of ATP hydrolyzed in myofibrils and at ion pumps, diffuse quickly to the mitochondria and stimulate oxidative phosphorylation (OxPhos) (7). This would form a metabolic feedback loop keeping mitochondrial ATP synthesis in balance with ATP hydrolysis. Initially, no decrease in phosphocreatine (PCr) or increase in P_i was found with increased workload in cardiac muscle (3), contradicting the importance of feedback control. As a consequence, it was proposed that cytosolic calcium not only stimulates muscle contraction but also enters the mitochondria, activating NADH-producing dehydrogenases in the mitochondrial matrix (2), implying feedforward control. Increases in cytosolic calcium indeed seem to lead to increased mitochondrial NADH fluorescence, but this occurs at a later stage after NADH has initially been oxidized during workload jumps (4). OxPhos is activated with a time constant of 5-15 s (39) during fast transitions in cytoplasmic ATP hydrolysis, which is too fast to be explained by the entry of calcium into the mitochondria (4, 29). Furthermore, when the work performance of the heart in situ is increased to high levels, a decrease in PCr and increase in P_i is often found with ³¹P NMR spectroscopy not only in animal studies (11, 25) but also in human volunteers (23). Determining the cardiac mitochondrial aerobic capacity, Mootha et al. (28) concluded that stimulation by ADP and P_i plays a role in the mammalian heart, at least when maximal O₂ consumption is approached. The mechanism of regulating cardiac OxPhos, especially during the first phase of the dynamic response in the first half-minute after a work jump, is therefore still unresolved, and the metabolic feedback system has not been ruled out.

From the classic model of mitochondrial stimulation via ADP and P_i, the prediction follows that the response time of O₂ consumption during steps in ATP hydrolysis (t_mito, see below) is inversely proportional to the mitochondrial aerobic capacity (MAC). Indeed, it is a general property of proportional feedback loop systems that the time constant of the step response is inversely related to the feedback loop gain. The feedback loop gain of the ADP-P_i feedback system is proportional to MAC. This property of proportional feedback systems is specifically shown by a linearized model of muscle respiration (27) where stimulation of the mitochondria is assumed to occur via ADP and P_i, which diffuse to the mitochondria with negligible delay. In this model mitochondrial oxidative phosphorylation is linearly related to changes in the free energy of ATP hydrolysis, and the creatine kinase reaction is assumed to be in equilibrium. From the analysis of this model (27) follows that the response time of oxidative phosphorylation should be inversely proportional to MAC. It is for instance predicted that halving MAC leads to doubling of the response time. Such a relation has indeed been found experimentally in skeletal muscle (32). Other enzyme kinetic or nonequilibrium thermodynamic models of feedback control in muscle (20, 39), which have not been simplified by linearization, predict that the response time increases even more than inversely with a decrease in MAC.

Methods have been developed to derive the response time of mitochondrial O₂ consumption (t_mito) to heart rate steps in the intact heart from measurements of O₂ uptake at the whole organ level (40). The time course of coronary venous O₂ tension is measured during steps in heart rate and corrected for the O₂ transport delay between mitochondria and the venous O₂ electrode. However, it is not possible to derive the full time course of O₂ consumption at the level of the mitochondria because the mitochondria are spatially distributed in heterogeneous tissue, along an arterial-to-venous O₂ concentration gradient. The solution to this problem was to derive mean transport and metabolic response times, similarly defined as the mean transit time, i.e., the average time taken by intravascular tracer molecules injected into an artery entering an organ to reach the venous exit of that organ (37). Thus the concept of mean transit time has been generalized to metabolic response times (37). The response time of O₂ uptake at the whole organ level is calculated from the time course of venous O₂ tension. This venous response time (t_v) is corrected for the mean transit time of O₂ between the mitochondria and the venous O₂ electrode (t_trans), which is calculated from the venous response to steps in arterial O₂ concentration or in perfusion flow (40). Finally, t_mito equals t_v  t_trans. This complex method has been extensively described and reviewed (37, 39, 40).

The aim of the present study was to test the hypothesis that cardiac oxidative phosphorylation is regulated by a simple feedback control mechanism via ADP and P_i. The prediction derived from this hypothesis is that t_mito increases inversely (or even more) with decreases in MAC. The cardiac mitochondria were inhibited to various degrees with nonsaturating doses of oligomycin, and the effect of such reductions of MAC on t_mito were determined. If the dependence of t_mito on MAC is less than inversely proportional, then the hypothesis of simple feedback regulation is refuted. An advantage of this test is that it does not rely on assumptions on the compartmentation of metabolites measured biochemically or with NMR spectroscopy nor does it assume near equilibrium of enzyme reactions for calculation of metabolite concentrations.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Overview of experimental design. The response of venous [O₂] to heart rate steps was determined in isolated rabbit hearts before and after graded infusion of oligomycin. The venous response to steps in arterial O₂ concentration and to a step decrease in perfusion flow was determined in the same heart to characterize the O₂ transport time. The t_mito is then obtained by subtracting the O₂ transport time from the measured venous response time. Immediately after the O₂ time course measurements in the isolated heart were completed, mitochondria were isolated to assess the degree of inhibition of mitochondrial capacity by oligomycin, because this inhibition is variable and unpredictable. The relation between mitochondrial inhibition and changes in t_mito is then examined.

Isolated heart preparation. All experiments were approved by the local animal ethics and experimentation authority (DEC-Vrije Universiteit). New Zealand White rabbits of both sexes, weighing 2.7 ± 0.2 (means ± SD) kg, were anesthetized with 9 mg/kg fluanisone and 0.3 mg/kg fentanyl citrate (Hypnorm, Janssen Pharmaceutica), injected intramuscularly, and 10 mg/kg pentobarbital sodium, injected in an ear vein. The heart was isolated and perfused with Tyrode solution, pH 7.3, gassed with 95% O₂-5% CO₂, containing 11 mM glucose as exogenous substrate. The O₂ supply does not limit O₂ consumption of the isolated saline-perfused rabbit heart at 28°C, whereas at 37°C most hearts are O₂ limited (15, 40). Measurements on mitochondria are usually done at <30°C. Here temperature of the heart and isolated mitochondria have been kept the same at 28°C. Coronary flow was kept constant during the experiment. Adenosine (10⁵ M) was added to the perfusion medium to obtain maximal vasodilatation. The heart contracted against a water-filled latex balloon in the left ventricle (LV), the pressure of which was measured. The volume of the balloon was kept constant during the experiment, and changes in systolic pressure therefore reflect changes in contractility. After we destroyed the atrioventricular node by crushing the tissue, resulting in spontaneous heart rates below 60 beats/min, hearts were electrically paced at 60 beats/min via two electrodes attached to the right ventricle. The hearts could be paced up to 120 beats/min without irreversible depression of cardiac function (10). O₂ tensions in the arterial inflow and venous outflow were continuously monitored by two fast-responding O₂ electrodes. O₂ consumption was calculated by multiplying the arterial-venous O₂ concentration difference with coronary flow. The perfusion and measurement procedures for the isolated heart have been described in detail before (10, 15, 40).

Calculation of t_mito. The t_mito, whose meaning is discussed in the introduction, is mathematically defined as the first statistical moment of the impulse response function, which is a generalized version of the time constant, also applicable when the response is not monoexponential (39, 40). From the venous PO₂ transient after a step in heart rate, the t_v is calculated by numerical integration. The t_trans takes diffusion between capillaries and mitochondria and subsequent vascular transport to the site of the O₂ electrode into account and is calculated (using a transport model) from the measured venous PO₂ transient either after a downward step in perfusion flow, by 11.4 ± 1.9% (mean ± SD), or after a downward step in arterial PO₂, by 6.2 ± 1.8% (37, 39, 40). After coronary venous PO₂ had reached a steady state, the flow or arterial PO₂ was returned in a single step to the baseline value. Detailed descriptions of the techniques including the O₂ transport model and the equations, assumptions, and correction values have been given previously (15, 39, 40).

Experimental protocol. The effect of oligomycin was determined by comparing t_mito in the same heart before and after oligomycin infusion. The infused amount of oligomycin was graded: 4 hearts received 0.21 mg/l oligomycin for 2 min, 4 hearts received the same oligomycin concentration for 4 min, and 4 hearts for 6 min. As control, 4 hearts received no oligomycin. After equilibration of the heart to its isolated condition, the baseline values of t_mito to different step sizes in heart rate were measured. The heart rate was stepped from 60 to 70 beats/min and back. Steps were also made from 60 to 80, 100, and 120 beats/min, respectively, and in each case back to 60 beats/min (10). For the determination of the t_trans, steps in arterial PO₂ and in perfusion flow were applied after each heart rate step series as described above.

After we assessed the response to heart rate and O₂ steps in the baseline state, oligomycin (O-4876, Sigma) dissolved in 100% ethanol at a concentration of 0.125 g/l was added for 2-6 min to the perfusion buffer to obtain a final concentration of 0.21 ± 0.03 mg/l. After the heart was equilibrated for O₂ consumption to reach a steady state with oligomycin in the tissue, the heart rate steps and O₂ transport measurements were repeated to determine t_mito in the presence of oligomycin, lasting ~25 min. Arterial and venous samples were taken at 60 beats/min during a steady state of O₂ consumption to determine lactate concentration (see Biochemical assays). Directly after the last measurements in the isolated heart, 0.3-0.4 g of the apex was cut off and frozen in isopentane cooled by liquid nitrogen. After the perfusion experiment about one-half of the left ventricle was used for isolation of mitochondria and the rest to determine the wet weight-to-dry weight ratio.

Assessment of inhibition of ATP synthetic capacity by oligomycin in isolated mitochondria. To quantify the extent of inhibition of mitochondrial O₂ consumption by oligomycin, the mitochondria were isolated according to Mela and Seitz (26) immediately after the perfusion experiment, and their respiration was measured. The heart tissue was immersed and homogenized in ice-cold isolation solution: 0.225 M mannitol, 0.070 M sucrose, 1 mM EDTA, 10 mM 4-morpholinopropanesulfonic acid (MOPS; Merck), 10 mM phosphate buffer (3 mM KH₂PO₄ and 7 mM K₂HPO₄) and 5 g/l albumin (A-7030, Sigma) with 0.2 mg/ml protease (Nagarse; Sigma). The pH of the isolation solution was set to 7.1 with KOH. To remove large cell fragments, the homogenate was centrifuged at 850 g during 5 min. The supernatant containing the mitochondria was then centrifuged at 6,950 g during 10 min. To wash the preparation, the pellet containing the mitochondria was resuspended in 20 ml of cold isolation solution without Nagarse and again centrifuged at 6,950 g during 10 min. Then 10 ml of cold isolation medium without albumin was used for resuspending the pellet of mitochondria to obtain uncontaminated determination of the mitochondrial protein content with the biuret method (Merckotest, Merck), calibrated with bovine serum albumin. After protein determination albumin was again added to the mitochondrial suspension at a final concentration 0.5 g/l. The final mitochondrial protein concentration in the reaction vessel was 1 mg/l. Glutamate (5.0 mM) and malate (5.0 mM) were given as mitochondrial substrates. To stimulate mitochondrial respiration, ADP was added to the reaction mixture, final concentration 400 µM. To obtain maximal respiration, bypassing the oligomycin inhibition, mitochondria were uncoupled with carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (FCCP; Sigma). Mitochondrial respiration attained a maximum when uncoupled with FCCP concentrations between 6 and 11 nmol/mg mitochondrial protein and decreased slightly after further increase of the FCCP concentration. The fraction of the MAC remaining after inhibition by oligomycin was characterized by dividing state 3 respiration (resulting from ADP stimulation) by maximal respiration uncoupled with FCCP.

In eight separate experiments on isolated mitochondria without oligomycin, FCCP-stimulated respiration was compared with state 3 respiration stimulated by ADP. Furthermore, a calibration line for the relation between FCCP- and ADP-stimulated respiration was determined in isolated mitochondria by comparing mitochondria from the same batch with a range of oligomycin concentrations and without oligomycin.

We also investigated whether the isolation procedure of the mitochondria from the hearts might affect the inhibition of the ATP synthase complex because of possible washout of oligomycin. We therefore examined the stability of the oligomycin inhibition in separate experiments on isolated mitochondria. Oligomycin was added to the isolated mitochondria to obtain a 45 ± 6% reduction of ADP-stimulated state 3 respiration. Thereafter, oligomycin-inhibited mitochondria and control mitochondria without oligomycin were centrifuged and resuspended again twice in fresh medium, and mitochondrial respiration was measured to determine the amount of inhibition. A separate sample of the oligomycin-inhibited mitochondria was also centrifuged and resuspended twice in its own supernatant, containing oligomycin that had leaked out of the mitochondria. Furthermore, separate hearts were perfused with a saturating dose of oligomycin, and ADP stimulation of respiration was tested after isolation of the mitochondria.

Biochemical assays. Lactate was measured in the perfusate samples as described previously (16). Briefly, the lactate was stoichiometrically converted to NADH, and the low NADH level was determined from the densitometrically determined rate of production of formazan in an enzymatic cycling reaction. Lactate efflux (µmol · min¹ · g dry wt¹) was calculated from the venous-arterial lactate concentration difference multiplied by flow.

After homogenization of the frozen tissue samples in ice-cold perchloric acid solution followed by neutralization, PCr, Cr, and ATP were determined in a single run of reversed-phase HPLC. The metabolites were separated on a 3-µm Microspher C18 column (Chrompack) employing elution with a gradient of mobile phase A consisting of 0.2 M KH₂PO₄ adjusted to pH 5.0 with KOH and mobile phase B consisting of water, acetonitrile, and methanol (50/25/25, vol/vol/vol). The gradient increased linearly from 1 to 35% mobile phase B in 3 min. Ultraviolet detection of the metabolites at 210 nm was used. The HPLC method has been described in detail elsewhere (36).

Statistical analysis. Data are presented as means ± SD unless indicated otherwise. The O₂ consumption under the various experimental conditions was compared by means of analysis of variance for repeated measurements. Trends in LV systolic pressure and O₂ consumption with MAC were tested by means of linear regression analysis. The relation between t_mito and MAC was analyzed for all heart rates simultaneously, applying General Linear Model hypothesis testing (30, 35) using the SYSTAT statistical computer package. The model incorporated a term accounting for the possible dependence of the slope of the relation between t_mito and MAC on the heart rate. Using the General Linear Model method, we tested separately for each heart rate whether the slope of the relation between MAC and t_mito differed significantly from zero. To this end we constructed matrices of linear weights on the model coefficients across the independent variables. The significance level for multiple tests was adjusted according to Bonferroni (30). Null hypotheses were rejected when P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Contractile and metabolic steady-state measurements on isolated hearts. To characterize the isolated heart preparation and effects of oligomycin, first some steady-state results are given. Wet weight of the hearts just after the experiment was 10.1 ± 0.8 g and dry weight after 2 days of dehydration at 60°C was 1.40 ± 0.12 g. Flow, kept constant during the experiment, was 7.5 ± 1.1 ml · min¹ · g wet wt¹. The perfusion pressure increased slightly from 84.0 ± 6.5 mmHg during baseline to 86.4 ± 6.5, 86.0 ± 7.9, and 93.9 ± 6.8 mmHg after 2, 4, and 6 min oligomycin infusion, respectively, indicating a slight effect of oligomycin on coronary resistance.

O₂ consumption and LV systolic pressure did not change (P > 0.05) during the step decrease in perfusion flow or arterial PO₂, independent of oligomycin dose, proving sufficient O₂ supply of the isolated heart. O₂ consumption increased with heart rate and decreased with oligomycin dose (Fig. 1, left). There was also a decrease in LV systolic pressure with oligomycin (Fig. 1, right). Diastolic pressure increased with higher heart rate because of incomplete relaxation but was independent of oligomycin dose. The PCr-to-Cr and PCr-to-ATP ratios at the end of the experiments are given in Table 1. The decrease in both ratios with oligomycin is corroborated by the decrease in PCr and the fourfold increase in P_i found after 6 min of infusion of oligomycin in a pilot experiment using ³¹P NMR spectroscopy. The PCr-to-ATP ratio in the frozen samples was similar as the ratio, 1.69 ± 0.08 (n = 7), found by Eijgelshoven et al. (11) with ³¹P NMR spectroscopy in the same type of preparation at the same temperature. Others have measured a PCr-to-ATP ratio of 1.85 with ³¹P NMR spectroscopy in the isolated rabbit heart (13). Thus the tissue sample from the apex appears representative for the LV. Lactate efflux from the isolated heart increased with oligomycin infusion (Table 1).

View larger version (29K): [in this window] [in a new window]   Fig. 1.   O2 consumption (left, in µmol · min1 · g dry wt1) and left ventricular systolic (Psys) and diastolic pressures (Pdia) (right, in mmHg) in isolated rabbit heart at five heart rates (60, 70, 80, 100, and 120 beats/min) without oligomycin administration (0 min) and after 2, 4, and 6 min infusion of 0.21 mg/l oligomycin (see text). Bars indicate means ± SE.

Response time measurements in isolated heart. In Fig. 2, registrations of changes in venous O₂ tension to obtain the t_mito during heart rate steps after 0, 2, and 4 min of accumulated oligomycin infusion in one heart are shown. Because of the decrease in O₂ consumption with oligomycin (Fig. 1), venous PO₂ increases. The baseline t_v before oligomycin was 13.5 ± 0.5, 15.1 ± 0.8, 16.2 ± 0.5, and 19.1 ± 0.6 (SE) s for heart rate steps from 60 to 70, 80, 100, and 120 beats/min, respectively. The measured O₂ response time at the venous level during the arterial O₂ concentration steps was 14.3 ± 2.1 s, whereas the arterial change took 3.6 ± 1.6 s. The venous response time to the perfusion flow step was 8.9 ± 1.6 s. The average t_trans (see METHODS) calculated from these O₂ transients was 7.9 ± 1.1 s and did not depend on the dose of oligomycin (P > 0.05). This t_trans is subtracted from the t_v to obtain t_mito (see METHODS). Consequently, the baseline t_mito before oligomycin was 5.6 ± 0.6, 7.2 ± 0.8, 8.3 ± 0.5, and 11.2 ± 0.6 (SE) s for the heart rate steps to 70, 80, 100, and 120 beats/min, respectively. As reported earlier (10), t_mito increases with heart rate at 28°C (P < 0.05). To express the effect of oligomycin, t_mito after oligomycin is divided by t_mito before oligomycin for each heart (see below).

View larger version (9K): [in this window] [in a new window]   Fig. 2.   Venous O2 tension transients in a heart after steps in heart rate from 60 to 80 beats/min after (from left to right) 0, 2, and 4 min oligomycin infusion (in min), respectively. Heart rate was increased at time 0, and O2 tension was steady before the heart rate jump. Measured venous response time (tv), which is proportional to shaded areas, was 14.8, 16.2, and 17.6 s, respectively. Decrease in O2 consumption at 60 beats/min is reflected by the increase in venous O2 tension at beginning of each trace, because flow is constant.

Isolated mitochondria measurements. In Fig. 3 a registration of respiration of isolated mitochondria partially inhibited with oligomycin is shown. The O₂ consumption during uncoupling with FCCP is larger than that during ADP stimulation, showing that the ATP synthase activity is partially inhibited. FCCP-stimulated O₂ consumption was 140 ± 20 nmol · mg mitochondrial protein¹ · min¹ (n = 16). In experiments on isolated mitochondria not receiving oligomycin, the maximum of the FCCP-stimulated respiration correlated strongly (r = 0.98, n = 8) with the state 3 respiration obtained using excess ADP. The maximal FCCP-stimulated respiration was on average 12.5 ± 2.1% higher than ADP-stimulated state 3 respiration. From this it follows that the ratio of ADP-stimulated respiration to FCCP-stimulated respiration in hearts that did not receive oligomycin is expected to be ~0.9.

View larger version (10K): [in this window] [in a new window]   Fig. 3.   Registration of O2 tension (PO2) in reaction vessel containing isolated rabbit heart mitochondria incubated with a nonsaturating amount of oligomycin. O2 consumption is proportional to slope of the curve. Mitochondrial O2 consumption was stimulated by adding ADP (arrow) to reach state 3 O2 consumption. After return to state 4, where free ADP concentration and O2 consumption had reached a minimal level again, mitochondria were uncoupled with FCCP. Apparent response time of isolated mitochondria to ADP injection is mainly determined by response time of O2 electrode in respirometer. O2 tension decreased much faster after maximal uncoupling with FCCP than during maximal ADP stimulation, indicating that ATP synthase complex was partially inhibited by oligomycin. FCCP-stimulated respiration diminishes sharply when approaching PO2 = 0 mmHg.

The following calibration line (r = 0.94, n = 15) was found in separate experiments by comparing isolated mitochondria from the same preparation without (Oli) and with (+Oli) oligomycin: (ADP-stimulated respiration + Oli)/(ADP-stimulated respiration  Oli) = 1.36 × (ADP-stimulated respiration + Oli)/(FCCP-stimulated respiration + Oli)  0.24.

The ratio of ADP-stimulated to FCCP-stimulated respiration thus characterizes the degree of inhibition of MAC by oligomycin.

Preservation of oligomycin inhibition during mitochondrial isolation. Mitochondria isolated from hearts that had received a saturating dose of oligomycin showed no stimulation of respiration by ADP but showed FCCP-stimulated respiration, in agreement with findings by Rouslin et al. (33). Washing isolated cardiac mitochondria did not change the level of inhibition of the ATP synthase complex by oligomycin (Table 2). The respiratory control ratio (RCR), i.e., state 3 respiration divided by state 4 respiration, of mitochondria without oligomycin was reduced after washing. The RCR of the oligomycin-inhibited mitochondria was lower but was reduced after washing by the same fraction as for mitochondria without oligomycin. Mitochondria inhibited with oligomycin showed a similar RCR whether washed with fresh isolation medium that did not contain oligomycin or with their own supernatant, possibly containing oligomycin leaked out of the mitochondria. State 4 respiration was not affected by oligomycin. These observations, together with our finding that mitochondria isolated from hearts saturated with oligomycin cannot be stimulated by ADP, prove that the extent of inhibition by oligomycin can be quantified in mitochondria isolated at the end of the perfusion experiment, and that washing the mitochondria with fresh medium does not diminish oligomycin inhibition.

Relating intact heart measurements to isolated mitochondrial measurements. The reduction of O₂ consumption in intact hearts after oligomycin was correlated (r = 0.70, P < 0.05) with the reduction in ATP synthetic capacity found in the mitochondria isolated from those hearts (Fig. 4, left). The LV systolic pressure at a heart rate of 60 beats/min was also significantly reduced (r = 0.80, P < 0.05) with the reduction in ATP synthetic capacity (Fig. 4, right). Because the LV balloon volume was kept constant, the decrease in systolic pressure directly indicates a decrease in cardiac contractility. The reduction of LV systolic pressure with MAC was not different (P > 0.05) at heart rates of 70-120 beats/min. LV systolic pressure decreased by 2.4 ± 0.5% (SE) per 10% decrease in MAC. The O₂ consumption decreased by 2.6 ± 0.7% (SE) per 10% decrease in MAC. This is derived from the slopes in Fig. 4 divided by 1.36 to take the calibration line for isolated mitochondria (see above) into account. Without intervention the reduction in systolic pressure during perfusion is minimal (<10% in 2 h) at 28°C, and O₂ consumption even increases (44) so that the effects of oligomycin are clearly larger than control.

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Effect of reduction of mitochondrial aerobic capacity on cardiac O2 consumption and contractility. Myocardial respiration after oligomycin administration normalized to baseline value before oligomycin (left), and left ventricular systolic pressure after oligomycin administration normalized to baseline value (right) determined at 60 beats/min in the intact heart. Normalized mitochondrial aerobic capacity is defined as ADP-stimulated state 3 respiration divided by maximal respiration of the same isolated mitochondria uncoupled with FCCP, which essentially bypasses the oligomycin inhibition. This ratio is determined in mitochondria isolated from hearts after perfusion experiment (see METHODS). Left: slope of the line was 0.35 ± 0.10 (means ± SE); right: slope was 0.32 ± 0.07. Both slopes differed significantly from zero, P < 0.05. Each symbol represents a heart (n = 16).

Response time and MAC reduction. The t_mito after oligomycin divided by the baseline t_mito before oligomycin infusion is shown in Fig. 5 as a function of the normalized ATP synthetic capacity, indicating MAC remaining after oligomycin. The slope of the relation between MAC and normalized t_mito depended on the heart rate (P = 0.042). For heart rate steps from 60 to 70 or 80 beats/min, t_mito increased significantly with decreasing MAC (slope < 0, P = 0.0003 and 0.0008, respectively). The change is significant, despite the fact that the scatter in t_mito was large for these small heart rate steps due to the small amplitude of the O₂ response. For heart rate steps from 60 to 100 or 120 beats/min, no significant dependence of the response time on MAC was found (P = 0.15 and 0.74, respectively), although the scatter was smaller than for the low heart rates. The slopes of the lines were 6.8 ± 3.5% and 7.9 ± 2.9% (SE) increase in t_mito per 10% decrease in MAC for the steps in heart rate from 60 to 70 and 80 beats/min, respectively, accounting for the mitochondrial calibration line. According to the feedback hypothesis there should have been an inversely proportional relationship between mitochondrial aerobic capacity and t_mito so that at normalized mitochondrial aerobic capacity 0.5 and 0.25 the t_mito should for instance have increased to 2 and 4, respectively. Figure 5 shows that this is clearly not the case, in particular for heart rates 100 and 120 beats/min, but also at the lower heart rates the normalized t_mito at normalized capacity 0.25 is clearly smaller than 4, disproving the feedback hypothesis.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Effect of reduction of mitochondrial aerobic capacity on mean response time of O2 consumption at level of mitochondria (tmito) to heart rate steps. Response time after oligomycin administration normalized to baseline value before oligomycin plotted versus normalized mitochondrial ATP synthetic capacity (see legend to Fig. 4). Four step sizes in heart rate, from 60 beats/min to 70 (A), 80 (B), 100 (C), and 120 beats/min (D), respectively, are given separately. Slopes of linear regression lines were 0.93 ± 0.48 (means ± SE), 1.08 ± 0.39, 0.27 ± 0.27, and 0.04 ± 0.22, respectively. Each symbol represents a heart (n = 16). Feedback hypothesis predicts an increase in normalized tmito to a value of 2 at normalized MAC 0.5, and normalized tmito should have reached a value of 4 at normalized MAC 0.25.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Methodological considerations. The experiments were done at a lower than physiological temperature to ensure adequate O₂ supply. However, current experiments performed at 37°C to study the effect of pressure-overload hypertrophy corroborate the present findings (8). Reduction of MAC by partial inhibition of the mitochondria with oligomycin, the same protocol as applied here, shows that t_mito does not decrease from 6.1 ± 0.5 s (SE, n = 10, sham control group of hypertrophy study) with a 20-75% reduction in MAC. In this 37°C study the risk of tissue hypoxia is increased, but the results for high heart rates (120-220 beats/min) are quite similar to the present results for heart rates of 100-120 beats/min at 28°C, where O₂ supply is nonlimiting. Without intervention a small increase in t_mito, by 1.2 ± 0.3 s (SE, P < 0.05), is observed during perfusion at 28°C (44), which makes the refutation of the feedback hypothesis even stronger, because this time effect would be added to the effect of oligomycin.

Apart from the blocking effect of oligomycin on ATP synthase, of which we make use in the present study, few side effects of oligomycin have been reported. Oligomycin is only toxic to aerobic organisms (24). Oligomycin does not affect the Ca²⁺-stimulated respiration of isolated mitochondria (6). The diastolic pressure increases with heart rate in this preparation at 28°C, showing that relaxation is easily compromised and that examining the end-diastolic pressure after oligomycin infusion forms a sensitive test for effects on relaxation and calcium cycling. Because oligomycin does not affect end-diastolic pressure (Fig. 1), oligomycin apparently does not affect relaxation and calcium homeostasis. Oligomycin does not influence Ca²⁺-Mg²⁺-ATPase activity (42), and pilot assays in our laboratory showed it does not inhibit sarcoplasmic reticulum Ca²⁺ ATP activity in the rabbit heart. However, oligomycin inhibits the Na⁺-K⁺-ATPase (24) but only at much higher concentrations than we used to inhibit the ATP synthase complex (12). Inhibition of Na⁺-K⁺-ATPase, e.g., with ouabain, increases contractility, an effect opposite to that found with oligomycin in the present experiments (Figs. 1 and 4). The lack of significant side effects of oligomycin suggests that the depression of cardiac contractility is caused by the inhibition by oligomycin of the ATP synthetic capacity. When cardiac contracility was influenced via other pathways, with dobutamine, no effect on t_mito was found (39), suggesting that changed contractility does not affect the dynamic response of oxidative phosphorylation.

The present results appear to refute the feedback hypothesis. The discussion in the preceding paragraph provides no way to save the feedback hypothesis, because we could not identify a side effect that counteracts the predicted effect of MAC reduction by oligomycin. On the other hand, the inhibition of the mitochondria by oligomycin has been clearly established in the present experiments and does not appear to influence t_mito in a way compatible with the feedback hypothesis.

The infusion of oligomycin did not lead to cell membrane damage, because creatine kinase leakage from the heart was not increased (De Groot, unpublished results, n = 5). Despite the increase in lactate efflux during oligomycin infusion, the P_i NMR peak was not split due to regions with low intracellular pH, as would have been expected if oligomycin had been distributed heterogeneously. In pilot experiments no change in t_mito or LV pressure development was found after infusion of the vehicle ethanol without oligomycin with the same duration and concentration (29 mM) as during oligomycin administration.

The apparent response time to ADP of the isolated mitochondria in Fig. 3 is determined almost exclusively by the slow response of the respirometer O₂ electrode, which was slower than the electrode used for the isolated heart. The true response of mitochondria to ADP has a half-time of 70 ms at 26°C (6). The mitochondrial response time is therefore much smaller than the response time measured in the intact cell, suggesting that cytoplasmic events determine the slow time course of the response.

Response time of oxidative phosphorylation. The experiments show unequivocally that the response time of mitochondrial O₂ consumption during heart rate steps is not inversely proportional to mitochondrial aerobic capacity. This refutes the hypothesis that oxidative phosphorylation is regulated by a simple feedback system where ADP and P_i diffuse quickly from the sites of ATP hydrolysis to the mitochondria to stimulate aerobic ATP synthesis.

This refutation is straightforward for simple models of metabolism that only contain feedback regulation by ADP and P_i. However, feedback control may be incorporated in models of metabolic regulation containing other regulators such as intramitochondrial calcium or limiting factors such as carbon substrate supply, NADH production, or O₂ supply. Not all of these models will necessarily be refuted by the present data. An advantage of feedforward control is that it would be faster than feedback control. Quick stimulation of oxidative phosporylation by Ca²⁺, which simultaneously causes ATP hydrolysis to increase, may prevent depletion of high-energy phosphates. However, comparison of NMR data with t_mito (see below) indicates that the feedback control metabolites change faster than oxidative phosphorylation, so that feedforward control mechanisms are apparently not very effective. Using the t_mito approach, Harrison et al. (19) have recently also refuted the hypothesis that the obligatory creatine kinase shuttle regulates oxidative phosphorylation, contradicting this popular model. The quick changes in PCr and P_i during heart rate steps found in our experimental model (11) suggest that phosphate metabolites play a role in stimulation of OxPhos but in a different way than in simple feedback models.

Given that direct stimulation by quickly diffusing ADP and P_i as the sole or dominant regulatory mechanism of oxidative phosphorylation is contradicted by the present results, other regulatory mechanisms have to be considered. It has been proposed that calcium, which is often increased at higher work states, enters the mitochondria from the cytosol to stimulate mitochondrial dehydrogenases and oxidative phosphorylation by increasing NADH levels via a feedforward mechanism (2). Intramitochondrial calcium could stimulate the electron transport chain (41) or ATP synthase activity (18). The NADH levels in rat heart muscle are increased at higher work states but only if cytosolic calcium is increased (4). However, this occurs after an initial phase of reduction of NADH levels and is too slow to account for the first phase of activation of OxPhos, which is characterized by the t_mito of 5-11 s. Furthermore, the increase in NADH was not found in the isolated rabbit heart (21). We reported earlier that t_mito was the same whether cardiac workload was increased with or without increases in cytosolic Ca²⁺ levels (15). Partial blockade of the mitochondrial Ca²⁺ entry channels also did not lead to slowing of the activation of oxidative phosphorylation (14). We conclude that Ca²⁺ entry into the mitochondria is not responsible for the early phase of activation of oxidative phosphorylation, although it may play an important role in the later phases of the mitochondrial response (4, 18, 41).

For steps to 70-80 beats/min, t_mito depends significantly on MAC, but no such relation is found for steps to 100-120 beats/min. Comparison of isolated mitochondrial respiration with whole heart respiration, assuming 300 mg mitochondrial protein/g tissue dry wt, shows that the mitochondria operated at low-to-medium respiratory rates in the heart during the present experiments, considerably below the maximum of mitochondrial respiration. It has been found in isolated mitochondria that the dominant flux control (contribution to rate limitation) shifts with increasing respiration rates. Three stages of control exist: control by proton leak in state 4, control by ATP turnover at intermediate respiration rates, and control shared by carbon substrate utilization and ATP turnover near state 3 (31). At the respiratory rates found in the present experiments, one therefore predicts the ATP turnover to be dominating mitochondrial control, because increased mitochondrial proton leak is very unlikely to explain the increase in respiration rate during heart rate steps. Because feedback from ATP hydrolysis cannot explain our present findings, we conclude that this control scenario derived from isolated mitochondria does not explain the dynamic regulatory response observed in the intact heart.

The feedback model assumed free diffusion of ADP and P_i in <30 ms. However, it has been proposed that the transfer of energy-related regulatory signals across the cytoplasm may take much longer (39). A quick decrease in PCr is found by time-resolved ³¹P NMR spectroscopy during the first 6 s after an upward step in heart rate (11), which should enable quick activation of oxidative phosphorylation given that isolated mitochondria are activated within 0.1 s (6, 7, 39). However, O₂ consumption changed with a t_mito of 14 s at 28°C and 11 s at 37°C. The time constant of the change in PCr and P_i was 5.3 s at 28°C and ~2.5 s at 37°C (11). We compare the time constant of P_i and PCr with t_mito because P_i and PCr were directly measured. Of course, ADP rather than PCr activates oxidative phosphorylation, but ADP cannot be measured directly using NMR spectroscopy. In model analyses ADP and PCr are often assumed to equilibrate instantly in the creatine kinase reaction (27), and ADP is calculated from this equilibrium. In reality there is some delay and dysequilibrium in the creatine kinase reaction (1, 22) so that the ADP changes, which cause the PCr changes, precede the changes in PCr. This strengthens the argument that phosphate metabolites change before oxidative phosphorylation flux changes. Thus the change in the NMR-measurable phosphate metabolites, and by inference also the change in ADP, is much quicker than the change in oxidative phosporylation. This is compatible with the suggestion that phosphate metabolite signals that stimulate oxidative phosphorylation are delayed by several seconds between sites of ATP usage and the mitochondria (39, 43), reflecting the traveling of a metabolic wave through the cytoplasm (34). This delay may be due to buffering of ADP and P_i, for instance, by creatine kinase and glycolytic enzymes.

The creatine kinase reaction rephosphorylates ADP to ATP, and glycolytic enzymes remove both ADP and P_i. If creatine kinase is partially blocked, the activation time of oxidative phosphorylation is reduced by half (19). Inhibition of glycolytic ATP synthesis by itself also causes quicker activation of oxidative phosphorylation, and if activation has already been accelerated by inhibition of creatine kinase, additional inhibition of glycolysis causes additional acceleration of the activation of oxidative phosphorylation (19, 38). It is important to consider glycolysis in this scheme, because it buffers changes in P_i, whereas the creatine kinase reaction does not. Dzeja et al. (9) also very recently suggested that glycolytic phosphotransfer enzymes play an important role in the transfer of energetic phosphoryl groups in muscle (9).

When one considers the mass balance of the high-energy phosphates, the delay in the time course of ATP synthesis by oxidative phosphorylation relative to PCr and P_i can only exist if there is transient activation of another ATP synthetic pathway: glycolytic ATP production is the very likely candidate (11). Quick activation of glycolytic ATP production explains that PCr is synthesized faster than oxidative phosphorylation is activated. It is worth considering that delay of the energy-related signal in the cytoplasm, caused by buffering of P_i and phosphate acceptors (creatine, ADP, and AMP), may dominate the activation time of the mitochondria. Such buffering may be performed by glycolytic enzymes, creatine kinase, adenylate kinase, and perhaps other phosphotransfer reactions. The glycolytic enzyme glyceraldehyde 3-phosphate dehydrogenase takes up one P_i, and the next glycolytic step, phosphoglycerate kinase, synthesizes ATP from ADP. In addition, the last enzyme of glycolysis, pyruvate kinase, also transfers a phosphate group during glycolysis, converting ADP to ATP. Given extensive buffering and delay of P_i and ADP in the cytoplasm, t_mito does not have to be inversely proportional to the mitochondrial aerobic capacity.

Substantial delay of the transcytoplasmic signal may appear inconsistent with the small distances between myofibril and mitochondrion in the heart. However, the speed of diffusion is not only determined by the diffusion distance but also by the molecular diffusivity. In the very dense sarcoplasm, binding to macromolecules and chemical transfer reactions may interfere with free diffusion. Phosphotransfer reactions buffering ADP and P_i close to the myofibrillar ATPase and ion pumps might be of functional significance for unhindered maximal performance of these ATP-consuming molecules. On the other hand, the phosphotransfer-buffering reactions delay the transfer of the metabolic signal to the mitochondria. Inhibition of these energy-buffering phosphotransfer reactions may accelerate the transfer of ADP and P_i by diffusion, because the chemical reactions taking away these molecules are not functioning. Such inhibition of the energy-buffering reaction may be at the expense of the performance of ATP-utilizing processes in the immediate vicinity, as experimentally seen during inhibition of creatine kinase (19) and of glycolytic enzymes (38). However, although the hypothesis of substantial delay of metabolite transfer by local energy-buffering reactions in the cytosol provides an explanation for the present and previous (11, 43) findings, alternative hypotheses invoking other regulatory factors than phosphate metabolites and calcium might also provide explanations in the future.

Comparison with skeletal muscle. Findings on the dynamic response of oxidative phosphorylation in skeletal muscle differ from the results on heart muscle reported here. The time constant for the decline and recovery of PCr during and after 8 min of stimulation in rat gastrocnemius muscle was 1.44 min at ~37°C (32), more than an order of magnitude slower than in the rabbit heart (11). We had predicted a sevenfold higher t_mito in this type of skeletal muscle compared with the heart (40) based on Meyer's (27) linear model of simple feedback control and the much lower mitochondrial content and the much higher creatine content of skeletal muscle.

In rat gastrocnemius (27) and cat soleus (17) muscles, the PCr time constant did not significantly depend on the stimulation rate. In the rabbit heart at 37°C, t_mito did also not depend on the stimulation rate (43), although at 28°C t_mito increases with heart rate in the present and previous studies (10, 44). Paganini et al. (32) reported for fast-twitch muscle that the rate constant for PCr recovery after contraction increased with mitochondrial enzyme content, which had been varied by exercise training and chemical thyroidectomy. This contrasts with the lack of effect of MAC on t_mito in the rabbit heart. We postulate that models of metabolic regulation applicable to skeletal muscle, with time constants ~90 s, are not applicable to heart muscle where aerobic energy fluxes are higher and t_mito is 5-11 s. Thus simple feedback control may dominate the dynamic response of aerobic metabolism in skeletal muscle, where an additional transcytoplasmic delay of several seconds would be hard to detect, whereas other factors become more important in the much faster system in cardiac muscle, especially at high workloads.

Steady-state effects of reduced mitochondrial aerobic capacity. O₂ consumption in the heart was reduced after oligomycin administration but still increased with heart rate, showing that the mitochondrial capacity was not fully used. Infusion of oligomycin leads to a linear decrease in systolic pressure at unchanged LV volume, suggesting a dependence of cardiac contractility on mitochondrial ATP synthetic capacity at submaximal workloads. The appreciable sensitivity of cardiac performance and energy turnover to reduction of mitochondrial capacity indicates that mitochondrial ATP synthesis shares in the flux control at submaximal workloads, contradicting that the ATP utilizing processes in myofibrils and at ion pumps constitute the sole rate-limiting step for myocardial energy turnover (5). There is very little excess mitochondrial aerobic capacity near maximal cardiac workloads (28), compatible with the appreciable rate limitation exerted by OxPhos at submaximal workloads found here.

In conclusion, LV systolic pressure and steady-state cardiac O₂ consumption decrease appreciably with a reduction in mitochondrial aerobic capacity at submaximal cardiac workloads. The cardiac ATP synthetic capacity determines the response time of cardiac O₂ consumption to quickly changing ATP hydrolysis but only at low heart rates. However, under all conditions t_mito increases far less than inversely proportional with a reduction in mitochondrial aerobic capacity, refuting the hypothesis that cardiac oxidative phosphorylation is exclusively regulated by quickly diffusing ADP and P_i in a simple feedback control system.