Regulation of Energy Metabolism of the Heart during Acute Increase in Heart Work*

We determined the contribution of all major energy substrates (glucose, glycogen, lactate, oleate, and triglycerides) during an acute increase in heart work (1 μm epinephrine, afterload increased by 40%) and the involvement of key regulatory enzymes, using isolated working rat hearts exhibiting physiologic values for contractile performance and oxygen consumption. We accounted for oxygen consumption quantitatively from the rates of substrate oxidation, measured on a minute-to-minute basis. Total β-oxidation (but not exogenous oleate oxidation) was increased by the work jump, consistent with a decrease in the level of malonyl-CoA. Glycogen and lactate were important buffers for carbon substrate when heart work was acutely increased. Three mechanisms contributed to high respiration from glycogen: 1) carbohydrate oxidation was increased selectively; 2) stimulation of glucose oxidation was delayed at glucose uptake; and 3) glycogen-derived pyruvate behaved differently from pyruvate derived from extracellular glucose. Despite delayed activation of pyruvate dehydrogenase relative to phosphorylase, glycogen-derived pyruvate was more tightly coupled to oxidation. Also, glycogen-derived lactate plus pyruvate contributed to an increase in the relative efflux of lactate versuspyruvate, thereby regulating the redox. Glycogen synthesis resulted from activation of glycogen synthase late in the protocol but was timed to minimize futile cycling, since phosphorylase a became inhibited by high intracellular glucose.

Increased heart work, usually elicited by catecholamines, increases carbohydrate oxidation because of activation of the pyruvate dehydrogenase complex (PDC) 1 by dephosphorylation (1); PDC phosphatase is stimulated by increased mitochondrial Ca 2ϩ (2). Activation of PDC under conditions of high workload could be advantageous, tending to increase carbohydrate oxidation selectively. Under the assumption that the ATP yield per O 2 consumed is higher for oxidation of carbohydrate versus lipid, a selective increase in carbohydrate oxidation could provide increased ATP synthesis despite maximum oxygen extraction.
Whether increased substrate oxidation with workload is selective for carbohydrate remains unresolved. Studies by us (3) and by Collins-Nakai et al. (4), using isolated working rat hearts, indicated that increased substrate oxidation is selective for carbohydrate, with little or no increase in exogenous fatty acid oxidation. In contrast, early studies by Neely et al. (5,6) and Crass et al. (7) suggested that fatty acid and carbohydrate oxidation increase in parallel. Hall et al. (8) showed that swine heart exhibit increased fatty acid uptake in vivo in response to dobutamine, presumably from stimulated ␤-oxidation, since levels of malonyl-CoA were reduced. Increased systemic nonesterified fatty acids resulting from peripheral lipolysis could also have promoted fatty acid uptake in that study. Awan and Saggerson (9) found reduced malonyl-CoA in isolated hearts and stimulated palmitate oxidation by isolated (nonworking) heart myocytes upon adrenergic stimulation. Therefore, the potential exists for direct adrenergic stimulation of total ␤-oxidation in the heart resulting from lowered malonyl-CoA, which is a potent inhibitor of a limiting enzyme for ␤-oxidation, carnitine palmitoyltransferase I.
The problem with existing studies is the failure to consider all relevant exogenous and endogenous substrates. The question of whether carbohydrate oxidation is increased selectively with the workload cannot be answered based on measures of glucose and exogenous fatty acid oxidation alone. Regulation of total ␤-oxidation should, ideally, consider both exogenous and endogenous lipids. Therefore, we extended existing studies in the following important ways. We examined oxidation of all carbohydrates (glucose, glycogen, and lactate) that together contribute virtually all PDC flux. We also considered total ␤-oxidation of exogenous and endogenous lipids. We used a pulse-chase technique to measure glycogen turnover continuously. Using this method, we find that the degree of coupling of glycogen utilization to oxidation varies depending on the relation between glycogenolysis and the capacity for carbohydrate oxidation (i.e. phosphorylase activation versus PDC activation). To explain the tight coupling of glycogen to subsequent oxidation, we examined the hypothesis that the burst of glycogenolysis is coordinated with activation of PDC. Surprisingly, it was not.
Lactate and pyruvate are released or consumed by tissues at different rates, since their systemic concentrations differ. We hypothesize that heart, which, at times, displays large fluxes for lactate or pyruvate across the plasma membrane, uses differential efflux of lactate versus pyruvate (a redox pair) as a supplemental mechanism for regulation of the cytosolic redox.
We previously found small amounts of simultaneous glycogen synthesis and degradation in heart, especially during glycogen depletion (10). Compared with total glycolytic flux, the contribution of flux through glycogen in the absence of cyclical changes in glycogen content ("glycogen cycling" (11)) was minor. In the present study, we examined activities for synthase and phosphorylase and their allosteric effectors in relation to fluxes for glycogen synthesis and glycogenolysis. We postulate that the well described reciprocal regulation of glycogen synthase and phosphorylase, apparently designed to prevent futile cycling of glycogen, is robust, preventing more than minor glycogen cycling in heart.

EXPERIMENTAL PROCEDURES
Materials-Isotopes were from ICN (Costa Mesa, CA). Enzymes were from Boehringer Mannheim. Other chemicals were from Sigma. Fatty acid synthase was purified from rat liver after inducing the enzyme; rats were fasted for 2 days and then fed bread (8% protein, no fat) for 3 days. Enzyme was purified from the cytosol of four rat livers (53 g) as described by Awan and Saggerson (9) except that chromatography was on DEAE-cellulose, as described by Linn (12). We obtained 68 units (1.1 unit/mg of protein), assayed by the method of Carey and Dils (13).
Heart Perfusions-Hearts from chow-fed male Harlan Sprague-Dawley rats (349 Ϯ 8 g, n ϭ 25) were perfused using the working heart apparatus (14) in a gas-tight configuration described previously (10). The filling pressure was 15 cm H 2 O, and the afterload/perfusion pressure was initially 100 cm H 2 O, increased to 140 cm H 2 O at the time of adrenergic stimulation. The initial perfusate was Krebs-Henseleit buffer containing 1.4 mM free Ca 2ϩ (1.5 mM CaCl 2 plus 0.1 mM EDTA), equilibrated with 95% O 2 , 5% CO 2 . Hearts were perfused in the working mode using 200 ml of recirculated perfusate, lacking carbon substrate initially. The initial part of this protocol ( Fig. 1) was designed to deplete and then resynthesize glycogen, as we previously described (3). After 20 min, the perfusate was supplemented to 5 mM glucose, 40 microunits/ml regular insulin (Lilly), 5 mM sodium D-␤-hydroxybutyrate, and 0.5 mM sodium L-lactate, and perfusion was continued for 25 min to allow glycogen resynthesis while contractile function returned to base line. Hearts were then switched to a nonrecirculating mode as described previously (3). The perfusate during this period was Krebs-Henseleit buffer containing 5 mM glucose, 40 microunits/ml insulin, 0.5 mM sodium L-lactate, 0.4 mM sodium oleate prebound to 3% (w/v) bovine serum albumin (fraction V, fatty acid free, Intergen, Purchase, NY), with 1.4 mM free Ca 2ϩ (the preparation was dialyzed against a large volume of albumin-free perfusate containing 1.5 mM CaCl 2 and 0.1 mM EDTA). At 55 min, epinephrine bitartrate was added to 1 M, and the height of the aortic overflow was raised to 140 cm above the heart. Hearts were freeze-clamped on their cannulae with aluminum tongs cooled in liquid N 2 .
Oxygen Consumption (MVO 2 ) and Contractile Performance-A portion of the coronary flow (10 ml/min) was pumped through a stirred, thermostated 3-ml chamber fitted with a Clark electrode (Yellow Springs Instrument Co.) and then returned to the heart chamber. A second electrode was fitted to the bottom of the oxygenator, and the two electrode currents were displayed continuously. MVO 2 was calculated from the A Ϫ V difference times the coronary flow, using 1.06 mM for the concentration of dissolved O 2 at 100% saturation (15). Electrodes were calibrated with air-saturated water (19.6% O 2 saturation after correction for water vapor, 47 mm Hg at 37°C). Hydraulic power (watts) is the product of cardiac output (coronary plus aortic flow, m 3 /s) times the afterload (pascals). Aortic pressure was recorded continuously with a Millar pressure transducer (Millar Instruments, Houston, TX) at the side arm of the aortic cannula, interfaced to a Gould physiologic recorder (Gould model 2400S; Cleveland, OH).
Analytical Methods-14 CO 2 and 3 H 2 O were determined in fresh perfusate as described previously (3). We verified that 14 CO 2 is recovered quantitatively by injecting 2 Ci of [ 14 C]NaHCO 3 into the apparatus. The sum of 14 C-lactate plus 14 C-pyruvate was determined in deproteinized perfusate by paper chromatography (16). Lactate and pyruvate migrate together in this system. The recovery of lactate from perfusate spiked with authentic [U-14 C]lactate was 88 Ϯ 2% (n ϭ 5) and was completely separated from [U-14 C]glucose. Other metabolites (lactate, pyruvate, glucose, glycerol) were measured in deproteinized samples by established spectrophotometric enzymatic assays (17). Metabolic rates were calculated as described previously (3). Apparent rates from glycogen were calculated based on the specific activity of glucose used to prelabel glycogen and then converted to true rates by assuming uniform isotopic enrichment of 54.7% (unstimulated treatment group in Table  I).
Tissue Analysis-Frozen hearts were weighed and ground to a fine powder under liquid N 2 , and a portion was taken for dry weight determination. High energy phosphates (ATP, ADP, AMP, glucose 6-phosphate, phosphocreatine), P i , pyruvate, and malonyl-CoA were measured in freshly prepared 6% perchloric acid extracts, adjusted to pH 5 with buffered KOH. High energy phosphates and pyruvate were measured using established enzymatic assays (17). Inorganic phosphate was measured with the acid molybdate reaction (18). Malonyl-CoA was measured radiochemically with purified fatty acid synthase (19). For analysis of triglycerides, a total lipid extract was prepared (20), and the triglyceride fraction was isolated by thin layer chromatography on silica gel plates (Whatman No. 4860-320) using CHCl 3 as the mobile phase. Lipid in the triglyceride band (R F ϭ 0.5) was extracted from the silica (20), hydrolyzed with alcoholic KOH, neutralized with perchloric acid, and analyzed for glycerol (17). Values were referred to acyl content (glycerol/3). This procedure had a recovery of 80%, using tripalmitin as an internal standard. To determine de novo synthesis, a portion of the neutralized hydrolysate was taken for scintillation counting, and the radioactivity referred to new acyl group incorporation based on the specific activity of extracellular [9,10-3 H]oleate. Glycogen was measured as described previously (3). The isolated glycogen was free from contamination by glucose and extracellular isotopes. To measure the tissue distribution of 14 C in glycogen, heart powder (50 mg) was dispersed into 1 ml of glacial acetic acid and then evaporated to dryness to remove 14 CO 2 . The residue was dissolved by heating with 1 ml of Solvable TM (Packard Instrument Co.) and then counted after adding 10 ml of scintillation mixture (Ultima Gold; Packard). In hearts subjected to 14 C-pulse labeling (i.e. 14 C-glycogen and unstimulated treatment groups), 85 Ϯ 3 and 90 Ϯ 3% of 14 C was found in glycogen, respectively. This indicates that glycogen recovery from the isolation procedure is at least 90% (otherwise there would be more 14 C-glycogen than total radioactivity) and that glycogen was virtually the sole source of radioactivity during the chase.
Enzyme Assays-Glycogen synthase was measured as described by Skurat et al. (21) using the filter binding assay of Thomas et al. (22). The glycogen phosphorylase assay was based on the method of Gilboe et al. (23). We determined the enzyme in the reverse direction by release of P i measured with the acid molybdate reaction (18). Phosphorylase a was measured in the presence of 0.5 mM caffeine to prevent activation by endogenous AMP (24), and total phosphorylase was measured in the presence of 3 mM AMP. Pyruvate dehydrogenase was measured by 14 CO 2 production from 2 mM [1-14 C]pyruvate, as described by Harris et al. (25). The active form was measured directly, and the total activity was measured following incubation for 30 min at 30°C with 1 mM CaCl 2 plus 5 mM MgCl 2 , to allow dephosphorylation by endogenous PDC phosphatase. This procedure produced stable, maximal values for the total activity of PDC. We established that all of the enzyme assays were linear with respect to time and amount of tissue extract used in the assays.
Data are expressed as mean Ϯ S.E. Statistical comparison was by analysis of variance with post hoc comparison by Newman-Keuls multisample test. p Ͻ 0.05 was considered significant.

RESULTS
Physiological Performance (Fig. 2 and Appendix Table  IA)-We performed five sets of matched perfusions, three of which were used to measure oxidation of three different 14 Clabeled substrates ( 14 CO 2 from glucose, glycogen, or lactate). Physiologic performance of these three groups is shown in Fig.  2. Fig. 2A shows contractile activity (hydraulic power, watts), and the bottom panel shows oxygen consumption (MVO 2 ). Two other treatment groups (unstimulated, and acute stimulation) were omitted from the figure for clarity, because the data points overlapped the values presented. These two groups were freeze-clamped at earlier times of the protocol, for tissue analysis. Various measures of performance for all the groups are given in the Appendix. The five groups were well matched for all values of performance measured. Performance is comparable with values measured in vivo for resting and exercising rats (26).
During the first 20 min of the protocol (Fig. 1), the absence of exogenous substrates resulted in diminishing contractile function. The protocol was designed to deplete endogenous substrates, evidenced by continued oxygen consumption. The addition of substrates at 20 min (glucose plus insulin, lactate, and D-␤-hydroxybutyrate) restored performance (Fig. 2). We used D-␤-hydroxybutyrate as a co-substrate to maintain incorporation of glucose into glycogen. Perfusions were switched to a nonrecirculating mode starting at 45 min (beginning of the "chase" for glycogen labeling), and the only change in substrate availability at that time was the replacement of ␤-hydroxybutyrate for a physiologic long-chain fatty acid, oleate prebound to albumin. Following a 10-min equilibration period, hearts were stimulated with epinephrine (1 M), and at the same time, we raised the afterload by 40%. There was an immediate 95% increase in contractile performance and 103% increase in MVO 2 . As occurs in vivo, oxygen extraction was slightly increased by this intervention, and O 2 was almost completely extracted, but most of the increase in MVO 2 resulted from an increase in coronary flow (see Appendix). Immediately after the work jump, there was no change in the supply/demand ratio for oxygen (MVO 2 /power), suggesting that hearts did not experience acute demand ischemia (this issue is examined further below). There was a 14% reduction in the ratio MVO 2 /power after prolonged stimulation, consistent with the well known reduction in cardiac efficiency during adrenergic stimulation.
Rates for Substrate Oxidation- Fig. 3 shows rates of substrate oxidation during the chase period for glycogen. Rates were measured by the Fick principle (V Ϫ A difference times coronary flow, but the A side was fresh perfusate or endogenous substrate), based on 14 CO 2 production for carbohydrate oxidation, or 3 H 2 O production from exogenous [9,10-3 H]oleate. The results are similar to those we reported previously (3), determined in the absence of insulin and lactate and without increasing the perfusion pressure. The new finding is that, like glycogen, lactate oxidation is rapidly increased, but unlike glycogen, increased lactate oxidation is sustained.
Glycogen Content and Enrichment- Table I gives values for the content of total glycogen, 14 C-glycogen, and enrichment. In hearts clamped at 55 min (unstimulated), the glycogen enrichment achieved by the labeling protocol (pulse portion of the pulse-chase) was 55 Ϯ 10%. With acute adrenergic stimulation, the decrease in glycogen content was not large, since the duration of stimulation was short (3 min). Following prolonged stimulation, the total glycogen content decreased by half, as did the content for 14 C-glycogen. The enrichment of residual glycogen remained the same (49 Ϯ 9%, 14 C-glycogen group in Table I). A similar value for enrichment (49%) was calculated for the glycogen that was broken down, correcting for the small amount of de novo synthesis ([ 14 C]glucose group in Table I).
These findings indicate that glycogen was degraded randomly, without discretion for new [ 14 C]glycosyl residues and preexisting [ 12 C]glycogen. As expected, there was no glycogen synthesis from lactate (Table I, last line).
We next considered the possibility that the initial glycogen degraded was enriched to a greater extent than the rest (i.e. a hybrid of random and ordered degradation). The distinction  (purely random versus hybrid model) is important for interpretation of the present data, as well as data from 13 C NMR spectroscopy studies. To do this, we examined the time course of glycogen enrichment by taking advantage of oxygen consumption measurements, but it was first necessary to account for oxygen consumption from all the substrates.
Triglyceride Turnover and Pyruvate Release-Total triglyceride content of the hearts increased slightly (not significantly) during the 20 min of adrenergic stimulation (from 56.9 Ϯ 6.2 (n ϭ 5) to 59.3 Ϯ 6.3 (n ϭ 15) mol of acyl/g, dry weight for unstimulated and prolonged stimulation treatment groups, respectively). During the same interval, de novo synthesis, based on incorporation of [9,10-3 H]oleate into triglycerides, increased (p Ͻ 0.05) from 1.38 Ϯ 0.34 (n ϭ 3) to 6.38 Ϯ 1.40 (n ϭ 15) mol of acyl/g, dry weight. Therefore, the total triglyceride pool size expanded by 2.4 mol acyl/g, dry weight, and de novo synthesis was 5.0 mol of acyl/g, dry weight. Using the relation degradation ϭ synthesis Ϫ change in pool size, the value for degradation is 5.0 -2.4 ϭ 2.6 mol/g, dry weight, and the corresponding rate over 20 min is 0.13 mol/min/g, dry weight (8% of exogenous oleate oxidation). To calculate oxygen consumption, we assumed that triglycerides were oxidized using a value of 25.5 mol of O 2 /mol for the average acyl group (i.e. like oleate).
Release of pyruvate contributes slightly to oxygen consumption (0.5 mol of O 2 /mol of pyruvate release) because pyruvate is more oxidized than its precursors (glucose, glycogen, or lactate). Rates of pyruvate release are given in Appendix Table  IIA. Fig.  4 shows predicted and measured rates of oxygen consumption. Predicted rates are based on the sum for the measured rates of oxidation of every major substrate (glucose, glycogen, lactate, oleate, triglycerides, and release of pyruvate). To calculate rates of oxygen consumption resulting from glycogen oxidation, we assumed that there is uniform isotopic dilution (i.e. that a purely random pattern of synthesis and degradation applies) and used the values given in Fig. 3. Two conclusions can be drawn from the close agreement between measured and predicted MVO 2 . First, we accounted for every major oxidizable substrate for the heart quantitatively. Second, the assumption of uniform isotopic dilution is valid. The second conclusion is shown more clearly in Fig. 5, which depicts oxygen consumption resulting from glycogen oxidation over time. The upper curve is oxygen consumption from oxidation of all glycogen, based on the difference between MVO 2 and oxygen consump-tion from every substrate except glycogen. The lower curve is oxygen consumption from oxidation of that portion of glycogen (55% of the total) that was labeled with 14 C (i.e. the apparent rate of glycogen oxidation ϫ 6). The relationship between the two curves is precisely the pattern expected based on purely random synthesis and/or degradation. It is distinguished from the pattern expected based on the last on-first off model or from a hybrid model.

Oxygen Consumption from Total Substrate Oxidation and the Time Course of Glycogen Enrichment (Figs. 4 and 5)-
Glycogen Synthase Activity in Relation to Glycogen Turnover (Table II)-There was a complete absence of incorporation of [5-3 H]glucose into glycogen (n ϭ 3) when the isotope was included between 45 and 55 min of perfusions in the unstimulated treatment group. Therefore, neither synthesis nor degradation occurred during the 10 min prior to adrenergic  Fig. 3 and Appendix Table IIA. FIG. 5. Oxygen consumption resulting from total glycogen oxidation compared with [ 14 C]glycogen oxidation. Oxygen consumption from glycogen oxidation during the chase is depicted. Open symbols, oxygen consumption resulting from oxidation of all glycogen, calculated by taking the difference between MVO 2 (shown in Fig. 4) and oxygen consumption resulting from oxidation of every substrate except glycogen. Closed symbols, oxygen consumption resulting from oxidation of that portion of glycogen (55% of the total) that was labeled with 14 C (mean Ϯ S.E., n ϭ 5).
stimulation. There was turnover following prolonged stimulation, since, as described above, there was robust glycogen oxidation (Fig. 3), and de novo synthesis occurred during the same period (Table I, [ 14 C]glucose treatment group). Table II shows activity states for glycogen synthase at different times of the protocol, along with the content of glucose 6-phosphate (activator of the glucose 6-phosphate-independent form), which account for the observed changes in de novo synthesis. The activity state of synthase was increased by 65% following prolonged stimulation. However, the enzyme was not activated acutely. The content of glucose 6-phosphate was not significantly changed during the protocol. These data indicate that a small amount of glycogen synthesis occurred late in the protocol, since synthase did not become activated until after prolonged stimulation. The potential for futile cycling is therefore limited, since most of glycogenolysis occurred early (Fig. 3).
Lactate/Pyruvate Ratios, High Energy Phosphates, and Regulation of the Redox (Appendix Tables IIA and IIIA)-We measured high energy phosphates and the intracellular ratio of lactate/pyruvate to determine if hearts experienced demand ischemia upon adrenergic stimulation, which would exaggerate glycogenolysis. The ratio of lactate/pyruvate, which rises dramatically during ischemia, reflects the cytosolic redox potential (NADH/NAD ϩ ), assuming lactate dehydrogenase is near equilibrium. We did not observe an acute increase in this ratio (Appendix Table IIA). Therefore, hearts were not ischemic during acute stimulation, when glycogenolysis occurred. The largest change in high energy phosphates was a 21% decrease in phosphocreatine during acute stimulation, in exchange with P i for the most part (Appendix Table IIIA). The ATP content was not decreased during acute stimulation. That acute changes in high energy phosphates were small or absent further supports the conclusion of adequate oxygen supply.
Prior to adrenergic stimulation, there was small net lactate extraction resulting from a small lactate gradient across the plasma membrane (0.5 mM extracellular versus 0.45 mM in the cytosol). With adrenergic stimulation, extraction switched to net release, and there was a large transient burst of release (Appendix Table IIA). Subsequently, lactate release quickly decreased to a new steady state, which was maintained for the remainder of the protocol. There was simultaneous oxidation of exogenous lactate during net release (Fig. 3). This phenomenon has been observed previously (27), and could result from mixing of intracellular and extracellular lactate pools or from cellular heterogeneity. Not unexpectedly, a pyruvate gradient (roughly equal to the intracellular pyruvate concentration, Appendix Table IIA) produced pyruvate efflux. Although pyruvate release followed its concentration gradient throughout the protocol, the burst of lactate release immediately after adrenergic stimulation (derived primarily from glycogen) was out of proportion to the lactate gradient (0.5 mM extracellular versus 0.87 mM in the cytosol). The resulting large increase in the ratio for lactate efflux relative to pyruvate efflux (from Ϫ0.49 Ϯ 0.62 to ϩ4.42 Ϯ 1.14, p Ͻ 0.05) prevented an acute increase in the lactate/pyruvate ratio (Appendix Table IIA). The ratio of lactate/pyruvate did eventually increase (after glycogenolysis had ceased) despite persistent increase in the ratio for lactate efflux relative to pyruvate efflux (3.85 Ϯ 1.53, p Ͻ 0.05 versus the unstimulated state). We explain the result by a switch from glycogen to glucose as the glycolytic substrate and tighter coupling of glycogen to oxidation (see below), since 70% of released lactate plus pyruvate was derived from glycogen during acute stimulation, but it was entirely from exogenous glucose later in the protocol.
Coupling of Glycolysis to Pyruvate Oxidation- Fig. 6 shows cumulative values for metabolism of glycogen (A) or glucose (B) by glycolysis, oxidation, and the percentage of glycolysis that was subsequently oxidized. We expressed the results as cumulative values (the slope is equal to the rate), because different TABLE II Glycogen synthase activity and the content of glucose 6-phosphate Values are the mean Ϯ S.E. The activity state of glycogen synthase is the activity for the glucose 6-phosphate-independent form Ϭ the dependent form ϫ 100%. Glycogen synthase D did not differ between the different groups and was 6.9 Ϯ 0.2 mol/min/g, dry weight (n ϭ 25).  metabolites ( 14 CO 2 , [ 14 C]lactate, etc.) produced at the same time may not appear in the coronary effluent at the same rate. Differences in diffusion are averaged out over a cumulative plot. As we reported previously (16), there is tighter coupling of glycolysis to oxidation from glycogen compared with glucose. By the end of the protocol, 31.4 Ϯ 2.1% of cumulative glucose utilization was subjected to oxidation, while the corresponding value from glycogen was 45.3 Ϯ 5.3% (44% higher, p Ͻ 0.05). (Table III)-The burst of release of glycogen-derived lactate plus pyruvate immediately after adrenergic stimulation arose, first, because PDC was slow to become activated relative to phosphorylase (Table  III) and, second, because of competition between extracellular lactate with glycogen as sources of pyruvate for PDC (Fig. 3).

Temporal Dissociation of Activation of Glycogen Phosphorylase from PDC and Regulation of Phosphorylase
Adrenergic stimulation predictably converted a larger portion of phosphorylase to the a form (Table III), since phosphorylase kinase is stimulated by cAMP and Ca 2ϩ . However, because intracellular glucose was acutely increased, augmented phosphorylase flux must have resulted largely from elevated AMP, a potent allosteric activator of phosphorylase b. Increased P i (derived mostly from phosphocreatine; see Appendix  Table IIIA), being a substrate for phosphorylase, probably also contributed to increased phosphorylase flux. Because of accumulation of intracellular glucose, which inhibits phosphorylase a allosterically, phosphorylation of phosphorylase to the a form may have paradoxically inhibited flux through the enzyme. Indeed, the mechanism for cessation of phosphorylase flux, despite 50% residual glycogen, appears to be the return of AMP to control levels, since phosphorylase a and intracellular glucose remained elevated. We conclude that the transient burst of phosphorylase flux reflects the status of high energy phosphates and not the activation state of phosphorylase kinase.
Coordination of Glycolysis from Glucose Versus Glycogen-During the first minutes of adrenergic stimulation, there was a gradual switch from glycogen to glucose as the glycolytic substrate (Fig. 3). The coordination between glucose and glycogen during the switch resulted from inhibition of glucose transport by accumulation of intracellular glucose during rapid glycogen breakdown (Table III), partially derived from glycogen itself (i.e. the 8% or so released as glucose instead of glucose 1-phosphate). The majority of glycolytic flux from glycogen bypasses hexokinase. Therefore, one would expect that if a bottleneck were at phosphofructo-1-kinase, then glucose 6-phosphate would increase during a burst of glycogenolysis, but this was not the case (Table II). We conclude that coordination between glucose and glycogen occurs at glucose transport. Table IV gives values for exogenous oleate oxidation and total ␤-oxidation (exogenous plus endogenous lipids). We measured exogenous fatty acid oxidation directly ( 3 H 2 O from [9, 10-3 H]oleate). Total ␤-oxidation was measured indirectly by a completely independent method based on oxygen consumption not resulting from carbohydrate oxidation. To check the validity of this approach, we calculated total ␤-oxidation based on the sum for oleate oxidation plus the value for triglyceride oxidation measured during adrenergic stimulation (0.13 mol/ min/g, dry weight) that was described above (see "Triglyceride Turnover and Pyruvate Release"). Values for total ␤-oxidation calculated from oleate plus triglycerides during acute stimulation (1.71 mol/min/g, dry weight) and prolonged stimulation (1.74 mol/min/g, dry weight) are slightly less (not significant) than values for total ␤-oxidation based on oxygen consumption measurements (Table IV). This is the expected result, since most of endogenous lipid oxidation is from oxidation of triglycerides. Prior to adrenergic stimulation, total ␤-oxidation was in excellent agreement with oleate oxidation (Table IV). This result is not surprising, because we expect to find diminished rates of triglyceride oxidation in the unstimulated state. We used the values for total ␤-oxidation based on oxygen consumption measurements as the best estimate of total ␤-oxidation. With adrenergic stimulation, the increase in oleate oxidation (20%) was small and did not reach statistical significance. The increase in total ␤-oxidation (40%) was larger and was significant. The level of malonyl-CoA was decreased by 33% following prolonged stimulation (Table IV), which is consistent with the increase in total ␤-oxidation.

DISCUSSION
The present study provides a comprehensive analysis of fuel selection during acute transition from low to high workload of isolated working rat heart. We confirm preliminary studies (3) by showing that glycogen and, to a lesser extent, lactate are important energy substrates for aerobic heart when the workload is acutely increased (Fig. 3). Although exogenous fatty acid oxidation was not significantly changed, total ␤-oxidation was increased by 40%, and the increase was associated with a decrease in the level of malonyl-CoA. This is expected based on the regulation of total ␤-oxidation by malonyl-CoA inhibition of carnitine palmitoyltransferase I (28). The increase in total ␤-oxidation was not in proportion to energy demand, however, so increased carbohydrate oxidation, facilitated by activation of PDC, became relatively more important. The pattern of substrate oxidation that developed with more prolonged adrenergic stimulation (Fig. 3) is in keeping with the observation in vivo that nonesterified fatty acids and lactate are major respiratory substrates for the heart in the exercising state (29,30). of phosphorylase, AMP, and glucose Values are the mean Ϯ S.E. The activity state of PDC is the active form/total activity ϫ 100%. The activity state of phosphorylase is phosphorylase a/total phosphorylase ϫ 100%. The total activity of PDC did not differ between the groups and was 21.3 Ϯ 1.4 mol/min/g, dry weight (n ϭ 25). Total phosphorylase activity did not differ between the groups and was 221 Ϯ 6 mol/min/g, dry weight (n ϭ 25). Intracellular glucose was calculated using the formula given in Appendix Table IIA  Triglyceride oxidation resulted from turnover of the triglyceride pool. Our result agrees with the previous estimate that triglyceride turnover contributes about 10% to total ␤-oxidation in the steady state (31). The capacity for mobilization of triglycerides, reflecting maximum triglyceride lipase activity, is unknown. However, lipolytic capacity is probably less than the tremendous capacity for glycogen mobilization in heart, since triglycerides were not an important endogenous energy substrate compared with glycogen in the short term. Glycogen may serve to prevent transient supply-demand mismatch for carbon substrate during acute increase in heart work, which would impair contractile function in the short term.
To answer the question posed in the Introduction, when all relevant exogenous and endogenous substrates are examined, the increase in carbohydrate oxidation upon adrenergic stimulation is, in fact, selective. Carbohydrate oxidation is increased selectively because total ␤-oxidation, regulated mostly by malonyl-CoA levels in the cytosol, is independent of the activity state of PDC in the mitochondria (compare PDC activities in Table III to malonyl-CoA levels in Table IV). Our data do not support the suggestion that PDC activation, by providing substrate for the synthesis of malonyl-CoA in the cytosol (by acetyl-CoA carboxylase), is a mechanism for reciprocal regulation of carbohydrate and fatty acid oxidation (32), at least in the setting of adrenergic induced activation of PDC.
In very broad terms, we confirmed early studies of Neely et al. (5,6) and Crass et al. (7) by showing that stimulation of heart work increases both carbohydrate and fatty acid oxidation. However, these early studies examined only exogenous substrate oxidation. Presently, we did not find an increase in exogenous fatty acid oxidation (there was a 20% increase, but this was not statistically significant). This result agrees with our previous report (3) and a similar study by Collins-Nakai et al. (4). Further, the small increase in total ␤-oxidation that we did find (40%) was not in step with carbohydrate oxidation or the workloads, contrary to Neely et al. (6). We also confirmed the report by Awan and Saggerson that epinephrine decreases the level of malonyl-CoA in isolated hearts (9). We extend the results to include the relation between malonyl-CoA and total ␤-oxidation, which is consistent with stimulation of ␤-oxidation at carnitine palmitoyltransferase I and inhibition of carnitine palmitoyltransferase I by malonyl-CoA (28). Since the concentration of nonesterified fatty acids was fixed in our perfusions, increased ␤-oxidation must have resulted directly from adrenergic stimulation and/or increased heart work and is not secondary to changes in delivery of nonesterified fatty acids. Although Hall et al. (8) found increased exogenous fatty acid uptake by dobutamine in pig heart in vivo, the result does not contradict our findings, since, as we will show elsewhere, 2 the sensitivity to workload of ␤-oxidation and triglyceride turnover becomes enhanced under metabolic conditions (high lactate and nonesterified fat) that develop in vivo during exercise or adrenergic stimulation.
The Strategy for Substrate Selection Based on ATP Yield-We can now consider why the heart adopted the strategy of preferential increase in carbohydrate oxidation to deal with increased energy demand, at least under idealized conditions of fixed substrate availability. Using a revised estimate for the stoichiometry of oxidative phosphorylation (33), the ATP/O 2 ratio calculated for complete oxidation of glucose (5.2) is only 4% higher than for oleate (5.0), and the value for complete oxidation of lactate (4.8) is actually less than oleate. Based on these calculations, the advantage of selective increase in carbohydrate oxidation during high oxygen extraction seems marginal, or it is a slight disadvantage in the case of lactate. The analysis is flawed, because carbohydrates are not completely oxidized. The actual value for ATP/O 2 is higher for glucose or glycogen relative to oleate, partially because carbohydrates carry more of their own oxygen but also because a portion of carbohydrates are metabolized anaerobically. Like skeletal muscle, during high workload states, a portion of glucose taken up by heart is diverted to the release of lactate plus pyruvate. This occurs irrespective of whether there is net release or extraction of lactate (Ref. 27; and compare Fig. 3 with net lactate fluxes given in Appendix Table IIA). Using values for the distribution between oxidation and release of lactate or pyruvate that were obtained during adrenergic stimulation in this study, we calculate that the effective value of ATP/O 2 is 5.4 for glucose and 5.7 for glycogen. Therefore, the advantage of using glycogen during periods of high energy demand is pronounced.
Coordination of Glucose and Glycogen Utilization-Upon adrenergic stimulation, the increase in glucose oxidation was delayed relative to the burst of glycogenolysis (Fig. 3). We explain the delay by interaction between glucose and glycogen at the level of glucose transport. Therefore, phosphorylase effectively regulates glucose uptake in this setting. We also conclude that acute regulation of phosphorylase mostly reflects the status of high energy phosphates (particularly AMP) and not the activity state of phosphorylase kinase. It follows that glucose uptake, in the setting of adrenergic stimulation, is regulated indirectly by the status of high energy phosphates. Blocked glucose uptake upon acute reduction in energy charge, and the associated burst of glycogenolysis, obviates the investment of ATP to prime glucose for glycolysis until after glycogen has been exploited.
Redox Regulation during Glycogenolysis and the Coupling to Oxidation-We confirmed our earlier observation (16) that glycogen-derived pyruvate is more tightly coupled to oxidation than is pyruvate derived from extracellular glucose (Fig. 6). We hypothesized that tighter coupling would reflect coordination between phosphorylase activation and PDC activation. Contrary to our hypothesis, PDC activation was delayed relative to the burst of glycogenolysis. We would have predicted pronounced activation of PDC resulting from the combined effect of Ca 2ϩ stimulation of PDC phosphatase and feed-forward activation resulting from inhibition of PDC kinase by high levels of glycogen-derived pyruvate. Paradoxically, we found that glycogen-derived pyruvate was preferentially oxidized despite the fact that PDC activation was slow to develop. Further, the portion not oxidized was preferentially reduced to lactate and was exported as such. Since the relative efflux of lactate and pyruvate varied during the protocol (Appendix Table IIA), the process contributed to regulation of the cytosolic redox potential. During the burst of glycolytic flux from glycogen, differential efflux of lactate versus pyruvate prevented the occurrence of cytosolic reduction that results when glycolytic flux is stimulated in excess of the capacity for pyruvate oxidation. In so doing, inhibition of glyceraldehyde-3P dehydrogenase and phosphofructo-1-kinase from cytosolic reduction and H ϩ accumulation, which would otherwise have self-limited pathway flux, was obviated.
Activation of Synthase and Phosphorylase Were Timed to Minimize Glycogen Cycling-We postulated that there is robust reciprocal regulation of glycogen synthase and phosphorylase, preventing more than minor occurrence of glycogen futile cycling in heart. We did observe activation (dephosphorylation) of synthase late in the protocol (Table II), accounting for a small amount of de novo glycogen synthesis (Table I) during net glycogenolysis (Fig.  3). Synthase activation could serve to prime the system for glycogen repletion during recovery but was timed to minimize futile cycling, since most of glycogenolysis occurred early on. The lag phase for synthase activation following stimulation of glycogenolysis, as suggested by Stalmans et al. (34) in liver, could have resulted from the progressive release of phosphorylase phosphatase, which then becomes available to dephosphorylate/activate glycogen synthase.
Is Glycogenolysis a Physiologic Response to Epinephrine?-Myocardial glycogenolysis is a physiological response to exercise of the whole animal (35), mediated in large part by increased adrenergic tone and circulating catecholamines. However, the glycogenolytic response to epinephrine has not been observed consistently. There are at least three explanations for the inconsistency. First, hearts perfused with saline in vitro are very sensitive to epinephrine induced glycogenolysis (3,36). This could be an artifact resulting from low reserve for O 2 delivery and from the induction of demand ischemia when no provision is made to increase O 2 delivery (37). In the present study, we did make provision for increased O 2 delivery at the onset of adrenergic stimulation (we increased the perfusion pressure at the same time), and we verified by various methods (intracellular ratio for lactate to pyruvate, content of high energy phosphates, oxygen consumption relative to contractile performance) that hearts did not experience demand ischemia. The increased AMP and small decrease in phosphocreatine that we observed during acute stimulation are similar to changes in vivo in response to increased pressure development and/or dobutamine infusion (37,38) and therefore represent a physiologic response. Second, we find that myocardial glycogenolysis is blunted under conditions of high lactate and nonesterified fatty acids 2 (39). High systemic lactate and nonesterified fat arise during exercise or infusion of epinephrine (40). This may explain the results reported by Williams and Mayer (Ref. 41 and references therein). Third, studies of glycogenolysis detected by 13 C NMR based on pulse-chase labeling of glycogen with [1-13 C]glucose, which suggest that glycogenolysis is insensitive to epinephrine, may be flawed because of the assumption of the last on-first off pattern of glycogen synthesis and degradation (42)(43)(44).
The Molecular Order of Glycogen Synthesis and Degradation Is Purely Random in Heart-In the present study, we exam-ined this issue in greater detail than previously (3, 10) by taking advantage of oxygen consumption measurements, using the fact that total substrate oxidation should be equal to oxygen consumption if the correct model for glycogen dynamics is employed. Specifically, we could distinguish between purely random versus a mixed pattern for glycogen dynamics. We have now shown by several independent techniques based on analysis of both metabolites derived from glycogen and by analysis of glycogen itself that there is no molecular order in the synthesis and subsequent degradation of glycogen (Refs. 3 and 10 and the present study). The pattern is not even a hybrid but appears to be purely random (Fig. 5), and it is therefore different from the situation in the liver (45). By adopting the last on-first off model proposed by Brainard et al. (42), apparent glycogenolysis based on disappearance of the 13 C NMR signal from glycogen will underestimate true glycogenolysis by an amount determined by the degree of isotopic dilution of [ 13 C] glycosyl residues by preexisting [ 12 C]glycogen. The degree of 13 C isotopic enrichment has not generally been specified, but it is probably considerably less than the value of 55% 14 C enrichment achieved in the present study. For example, Laurent et al. (44) recently concluded that epinephrine does not stimulate glycogenolysis in skeletal muscle. If the degree of isotopic enrichment of glycogen in that study were as high as the value for extracellular glucose (13.9%), then glycogenolysis was underestimated by a factor of 7.
In conclusion, we characterized total substrate utilization during an acute low to high work transition of the heart, described the mechanism responsible for high respiratory rates from glycogen, and explained the inconsistency among reports of epinephrine-induced glycogenolysis. High glycogen respiration was one component of the overall pattern of selective increase in carbohydrate oxidation. The advantage of increased carbohydrate oxidation may relate to a higher effective value for the ratio of ATP synthesized/O 2 consumed for carbohydrate versus lipid, not to be confused with the ATP/O 2 ratio calculated for complete oxidation. where ICF is the intracellular fluid space, ECF is the extracellular fluid space, and ECF ϩ ICF ϩ D/W ϭ 1. D/W is the dry to wet weight ratio of each heart. ECF was determined in separate perfusions from the concentration of [U-14 C]sucrose in the whole heart (dpm/g, wet weight) over the concentration in the perfusate (dpm/ml); this value was not changed by adrenergic stimulation and was 0.456 Ϯ 0.020 (n ϭ 5).