Purinergic Inhibition of Glucose Transport in Cardiomyocytes*

ATP is known to act as an extracellular signal in many organs. In the heart, extracellular ATP modulates ionic processes and contractile function. This study describes a novel, metabolic effect of exogenous ATP in isolated rat cardiomyocytes. In these quiescent (i.e. noncontracting) cells, micromolar concentrations of ATP depressed the rate of basal, catecholamine-stimulated, or insulin-stimulated glucose transport by up to 60% (IC50 for inhibition of insulin-dependent glucose transport, 4 μm). ATP decreased the amount of glucose transporters (GLUT1 and GLUT4) in the plasma membrane, with a concomitant increase in intracellular microsomal membranes. A similar glucose transport inhibition was produced by P2 purinergic agonists with the following rank of potencies: ATP ≈ ATPγS ≈ 2-methylthio-ATP (P2Y-selective) > ADP > α,βmeATP (P2X-selective), whereas the P1 purinoceptor agonist adenosine was ineffective. The effect of ATP was suppressed by the poorly subtype-selective P2 antagonist pyridoxal-phosphate-6-azophenyl-2′,4′-disulfonic acid but, surprisingly, not by the nonselective antagonist suramin nor by the P2Y-specific Reactive Blue 2. Glucose transport inhibition by ATP was not affected by a drastic reduction of the extracellular concentrations of calcium (down to 10−9 m) or sodium (down to 0 mm), and it was not mimicked by a potassium-induced depolarization, indicating that purinoceptors of the P2X family (which are nonselective cation channels whose activation leads to a depolarizing sodium and calcium influx) are not involved. Inhibition was specific for the transmembrane transport of glucose because ATP did not inhibit (i) the rate of glycolysis under conditions where the transport step is no longer rate-limiting nor (ii) the rate of [1-14C]pyruvate decarboxylation. In conclusion, extracellular ATP markedly inhibits glucose transport in rat cardiomyocytes by promoting a redistribution of glucose transporters from the cell surface to an intracellular compartment. This effect of ATP is mediated by P2 purinoceptors, possibly by a yet unknown subtype of the P2Y purinoceptor family.

It is well established that ATP acts as an extracellular signal in many tissues and is involved in a variety of regulatory processes including the control of vascular tone, muscle contraction, pain, or neuronal communication (for review see Ref. 1). In particular, ATP serves as a neurotransmitter or cotransmitter in central, as well as in peripheral neurons (1,2). For instance, ATP is co-released with noradrenaline or acetylcholine from sympathetic or parasympathetic nerve endings, respectively, and even from neuromuscular synapses (1). An-other source of extracellular ATP is that released from parenchymal cells under hypoxic or ischemic conditions (3).
Many biological responses to ATP are mediated via P 2 purinoceptors, which belong to either of two structurally and functionally distinct families of receptors: ligand-gated, nonselective cation channels (P 2X -type), or G-protein-coupled receptors linked to the phospholipase C signaling cascade (P 2Y -type) (1,4). Activation of P 2X receptors leads to sodium and calcium influx and to cell depolarization and is therefore stimulatory in nature (e.g. it causes smooth muscle cell contraction). Although P 2Y purinoceptor activation also results in an increased cytosolic calcium concentration (at least in part via inositoltrisphosphate-mediated calcium release), it induces both stimulatory and inhibitory effects (the latter through the opening of potassium channels or indirectly through nitric oxide production). In the heart, extracellular ATP and ATP analogues were found to exert pronounced although relatively complex effects on coronary tone and mechanical activity, depending on the type of preparation studied and on the purine concentrations used (5)(6)(7)(8).
By contrast, it is not known whether ATP may also impinge on cardiomyocyte metabolism, whose regulation often closely matches the changes in energy demand and contractile status. The aim of the present investigation was therefore to investigate possible effects of ATP and other purinergic agents on the trans-sarcolemmal transport and utilization of glucose that, besides fatty acids and lactate, represents an essential substrate in cardiac myocytes. In these studies, we observed a pronounced inhibitory action of purinergic agonists on the rate of glucose transport and explored the underlying mechanisms.
Ionic Gradients-In the experiments examining the relevance of ionic gradients (see Fig. 3), cardiomyocytes were incubated in media whose composition was identical to medium A, except that the concentration of the ion to be tested was varied as indicated in the figure. All the media used were isotonic; thus sodium was replaced by an equimolar amount of choline (see Fig. 3A), potassium was replaced by sodium (see Fig. 3C), or chloride was replaced by HEPES (see Fig. 3D). The calcium concentration was adjusted according to the method of Fabiato and Fabiato (22) (see Fig. 3B). The experiments in which the sodium concentration was lowered (see Fig. 3A) were performed at a calcium concentration of 10 Ϫ7 M to avoid a calcium overload of the cells due to impairment of the sodium-dependent calcium extrusion via the sarcolemmal Na ϩ /Ca 2ϩ exchanger.
Glycolytic Flux-The flux through the 6-phosphofructo-1-kinase reaction was estimated by measuring the rate of 3 H 2 O formation from [3-3 H]glucose with a protocol adapted from Ref. 23. In brief, cardiomyocytes were treated (with or without ATP and insulin) as described in the legend of Fig. 5 before [3-3 H]glucose (0.75 Ci/sample) was added for 30 min at 37°C. The incubation was then stopped with perchloric acid (0.3 M). The extracts were neutralized with KOH/KHCO 3 , and the tritiated water glycolytically formed was separated on Dowex-borate columns (Bio-Rad, Munich, Germany) as described (23). The amount of nonspecific 3 H 2 O (i.e. that which is not related to the glycolytic activity of the myocytes) was determined by adding perchloric acid to a parallel set of cell samples prior to the incubation with [3-3 H]glucose.
Measurement of 14 CO 2 Production from [1-14 C]Pyruvate-Flux through the pyruvate dehydrogenase complex was determined by monitoring the production of 14 CO 2 from [1-14 C]pyruvate as described (24). In brief, cardiomyocytes were exposed to the experimental conditions to be investigated (ATP and/or insulin treatment) at 37°C in 20-ml vials sealed with rubber stoppers (under the same conditions as for the glucose transport assays, i.e. in a total volume of 1.5 ml and with ϳ1.5 mg cell protein/ sample). The oxidation reaction was then started by injecting 0.1 Ci of [1-14 C]pyruvate (final concentration, 0.1 mM) into the cell suspension. After another 10 min at 37°C, the incubation was terminated by injecting 300 l of 0.3 N perchloric acid. The released 14 CO 2 was trapped (over 18 h at 4°C) in 0.3 ml ethanolamine/ethylene glycol (1:1, v/v) contained in a polypropylene center well (placed in the vials before the incubations). Nonspecific values (determined by injecting perchloric acid in parallel cell samples before adding [1-14 C]pyruvate) were subtracted from all samples. The quenching caused by the ethanolamine/ethylene glycol mixture was determined in each experiment and taken into account for the calculation of pyruvate oxidation rates.
Preparation of Purified Membrane Fractions from Cardiomyocytes and Western Blot Analysis-Cardiomyocytes were treated as detailed in the legend of Fig. 4 and then rapidly washed once with TES buffer (20 mM Tris, 1 mM EDTA, 250 mM sucrose, 0.1 mM (phenylmethylsulfonyl fluoride, pH 7.4) and immediately frozen in liquid nitrogen in a ratio of 10 7 cells/2.7 ml of TES. Preparation of purified plasma and intracellular membrane fractions (25) and semi-quantitative immunodetection and quantitation of GLUT4 and GLUT1 (26) were performed as described.
Calculations-Cell protein was measured by the biuret method. The curves shown in Fig. 1, as well as the IC 50 values of glucose transport inhibition, were obtained from the dGlc transport data using a computer program (Prism) from GraphPad Software, Inc. (San Diego, CA), according to the following equation.
where X is the logarithm of agonist concentration, Y min and Y max are the minimum and maximum dGlc transport rates, IC 50 is the agonist concentration producing a half-maximal inhibition, and S the Hill slope of the curves. For the statistical tests used to assess the difference of data sets, see the figure legends.

Effects of Purinergic Agonists and
Antagonists on Glucose Transport in Cardiomyocytes-Glucose transport is the first step and under many physiological conditions the rate-limiting step of myocardial glucose utilization and is known to be the target of a host of regulatory factors in heart muscle cells, such as insulin, catecholamines, contraction, anoxia, ischemia, or alternative substrates (24,(27)(28)(29)(30).
In an initial series of experiments, we therefore investigated the action of purinergic agonists on cardiomyocyte glucose transport. As documented in Table I, ATP produced a substantial inhibition of glucose transport in basal (nonstimulated) myocytes, as well as in cells stimulated with insulin or with the ␣ 1 -adrenergic agent phenylephrine. The effect of ATP was rapid; thus the extent of inhibition of insulin-stimulated glucose transport was the same, regardless of whether ATP was added to the cells 30, 20, 10, or 0 min before or even 2 min after insulin addition (not shown).
Similarly, other purine nucleotides known to mainly interact with P 2 purinoceptors (2, 31) efficiently depressed glucose transport; thus addition of ATP and nonhydrolyzable ATP analogues (ATP␥S 1 , 2MeSATP, and ␣,␤meATP) at micromolar concentrations resulted in a decrease in insulin-dependent glucose transport by up to ϳ50%. Inhibitory effects were also observed with other purine nucleotides such as ADP, NADH, and NAD ϩ (Table I); NADPH and NADP ϩ also tended to inhibit glucose transport, although these effects did not quite reach statistical significance (Table I). By contrast, adenosine (100 M), which preferentially activates members of the P 1 purinoceptor family, had no effect on the basal or on the insulin-stimulated rate of glucose transport (Table I).
In an attempt to discriminate the potential P 2 receptor type involved, we examined the concentration dependence of the TABLE I Effects of adenine nucleotides on dGlc uptake in cardiomyocytes Cardiomyocytes were incubated for 10 min (at 37°C in medium A) with or without nucleotides at the indicated concentrations before insulin (12 nM) or phenylephrine (120 M) were added for another 30 min. In another series of samples, cells were treated for 60 min with oligomycin B (0.6 M) in the presence or absence of ATP. At the end of these incubations, the rate of dGlc uptake was determined as described under "Experimental Procedures." Data are the mean values expressed in pmol/h/mg protein Ϯ S.E. The numbers of independent experiments are indicated in parentheses. The percentage values given in the third column in relation to the control values in the corresponding experiments. The p values given are in relation to the corresponding uptake rates measured in the absence of nucleotide (analysis of variance, except for the comparison done with basal and phenylephrine-stimulated values that were paired in each experiment and were therefore analyzed with a paired Student's t test).

Additions
Rate of dGlc uptake Percentage p effects of some of the above agonists. As illustrated in Fig. 1, the P 2 -subtype nonselective agonists ATP, ATP␥S, and ADP were effective in the low micromolar range (IC 50 ϭ 4.0 Ϯ 0.7, 5.4 Ϯ 1.5, and 7.8 Ϯ 3.4 M, respectively). The P 2Y -selective analogue 2MeSATP (31) also produced a significant inhibition (p Ͻ 0.001) at 1-10 M, but higher concentrations than with ATP or ATP␥S were required to reach a maximal degree of inhibition (Fig. 1); in this case the dose-response relationship could not be fitted to a simple sigmoidal function. The selective P 2X purinoceptor agonist ␣,␤meATP (31) was clearly less potent (with an apparent IC 50 of ϳ100 M) (Fig. 1). UTP, which specifically activates so-called P 2U purinoceptors (which belong to the P 2Y family and are also expressed in the myocardium; Ref. 32), was without effect at a concentration as high as 100 M (Table I).
A second approach aimed to address the question of the P 2 -type mediating the inhibiting effect of ATP was to use purinoceptor antagonists. Surprisingly, neither did the P 2Y -specific antagonist Reactive Blue (31) nor suramin, which is known to antagonize the binding of agonists to and the activation of both P 2X and P 2Y purinoceptors (31), affect the decrease in glucose transport brought about by ATP (Fig. 2). By contrast, the poorly subtype-selective antagonist pyridoxal-phosphate-6-azophenyl-2Ј,4Ј-disulfonic acid (PPADS, 100 M; Ref. 33) suppressed the ATP-dependent inhibition of glucose transport (Fig. 2). It should also be mentioned that the effect of ATP (30 M) was not influenced by the P 1 receptor antagonists 1,3diethyl-8-phenylxanthine, or 3,7-dimethyl-1-propargylxanthine (100 M each) (not shown).
Influence of Ionic Gradients on the Inhibitory Effect of ATP on Glucose Transport-Because the pharmacological data presented above were not sufficient to clearly define or exclude either P 2X or P 2Y receptors as the mediator of the effect of ATP on glucose transport, we considered the possible involvement of the former purinoceptor type by using a functional approach. P 2X purinoceptors are nonselective cation channels whose activation produces a depolarizing inward sodium and calcium current (1, 2, 4). We therefore examined the influence of trans-sarcolemmal ionic gradients on the inhibition of glucose transport by ATP. To this end, we modified the composition of the incubation medium with respect to sodium, calcium, and potassium over a wide range of concentrations (while keeping the osmolarity constant).
The results of this series of experiments are summarized in Fig. 3. Even a reduction of the extracellular sodium concentration to nominally 0 mM (Fig. 3A) or of the free calcium concentration down to 10 Ϫ9 M (Fig. 3B) did not suppress the ATPinduced glucose transport inhibition. It is noteworthy that the basal rate of glucose transport (in the absence of ATP) significantly rose when the calcium level was lowered from the physiological value of 10 Ϫ3 M (Fig. 3); conversely, we had previously found that an increase in calcium from 1 to 5 mM depresses the rate of glucose transport in these cells (34). Hence it appears that calcium exerts an inhibiting influence of the glucose transport system of cardiomyocytes. However, our present data (Fig.  3, A and B) show that the action of ATP on glucose transport is not dependent on extracellular calcium and sodium, thus ruling out that a depolarizing influx of these ions could mediate the action of the nucleotide. In addition, depolarization with 30 mM potassium chloride (in the absence of ATP) did not produce an inhibition of glucose transport (Fig. 3C, open bars).
Since purinoceptor activation was also shown to activate cardiac potassium channels (18,35), we also examined the possible influence of the potassium gradient across the plasma membrane. As shown in Fig. 3C (closed bars), ATP significantly reduced the rate of glucose transport at all tested potassium concentrations, although the percentage of inhibition tended to be smaller at 0.6 mM potassium. However, such a low potassium concentration may limit the function of the Na ϩ /K ϩ -ATPase, i.e. of Na ϩ extrusion, thus impairing Ca 2ϩ efflux or even promoting Ca 2ϩ influx via the sarcolemmal Na ϩ /Ca 2ϩ exchanger, which would eventually result in an increased cytosolic Ca 2ϩ concentration. We therefore also studied the effect of ATP at a very low extracellular Ca 2ϩ concentration (10 Ϫ7 M) (at which the Na ϩ / Ca 2ϩ exchanger is not required to maintain a low cytosolic calcium concentration). We found that ATP inhibited glucose transport by 57 Ϯ 7% even in the complete absence of extracellular potassium (two independent experiments; not shown). Moreover, the effect of ATP (as tested at 6 mM potassium and 1 mM Ca 2ϩ ) was not affected by the potassium channel blockers Tedisamil, Apamin, or Glibenclamide (3-10 M): thus, in the presence of these inhibitors, the percentage of inhibition by ATP was 48 Ϯ 1%, 37 Ϯ 16%, and 49 Ϯ 2%, respectively (n ϭ 2-3). In conclusion, we found no indication that the trans-sarcolemmal potassium gradient or potassium channels may be relevant to the action of ATP on glucose uptake.
Finally, it was reported that in cardiomyocytes ATP activates trans-sarcolemmal bicarbonate/chloride exchange (10,11). We therefore also investigated the influence of the extracellular chloride concentration on the effect of the nucleotide on glucose transport. As illustrated in Fig. 3D, this parameter did not affect the action of ATP; neither did the inhibitor of bicarbonate/chloride exchange DIDS (50 M) (not shown).
Effect of ATP on the Subcellular Distribution of the Glucose Transporters GLUT4 and GLUT1-We next addressed the mechanism by which ATP decreases the rate of glucose transport. In particular, we investigated the possibility that this effect might be explained by a decrease in the amount of the glucose transporters GLUT4 and GLUT1 (which mediate the uptake of glucose in cardiomyocytes; Ref. 26) in the plasma membrane of these cells. As previously reported (25,26), insulin (in the absence of ATP) produced an increase in the GLUT4 and GLUT1 content in the plasma membrane fraction, which was paralleled by a decreased of these transporters in the FIG. 1. Concentration dependence of the inhibiting effects of P 2 purinoceptor agonists on insulin-stimulated glucose transport. Cardiomyocytes were exposed for 10 min to ATP, ATP␥S, ADP (P 2 -subtype-nonselective agonists), ␣,␤meATP (P 2X -selective), or 2Me-SATP (P 2Y -selective) at the indicated concentrations before insulin (12 nM) was added for another 30 min. Following this incubation the rate of dGlc transport was measured as described under "Experimental Procedures." Each point represents the mean from three to eight independent experiments; S.E. values were less than 10% in all cases. The curves corresponding to the effects of ATP, ATP␥S, ADP, and ␣,␤meATP were calculated using Equation 1; R 2 was Ͼ0.99 in all cases (with the 2Me-SATP data, the fit did not converge; the points were therefore connected by a simple line). Significant statistical difference (p Ͻ 0.05) versus the corresponding control (in the absence of agonist) was reached at 1 M 2MeSATP, 3 M ATP, and 10 M ATP␥S, ADP, or ␣,␤meATP (paired Student's t test). intracellular membrane fraction (LDM) (Fig. 4). In the presence of ATP (30 M) the insulin-dependent effect on plasma membrane GLUT4 and GLUT1 was largely suppressed (Fig. 4). Concomitantly, ATP treatment resulted in a significant increase in the amount of GLUT4 and GLUT1 detected in LDM (Fig. 4). In other words, ATP caused a redistribution of GLUT1 and GLUT4 from the plasma membrane to an intracellular site, thus offsetting the insulin-induced recruitment of these transporters to the cell surface.
Effects of ATP on the Rates of Glycolysis, Pyruvate, and Palmitate Oxidation-In view of the large effects produced by purinergic agents on the uptake of glucose, the question arose FIG. 2. Influence of P 2 purinoceptor antagonists on the inhibitory action of ATP on glucose transport. Cardiomyocytes were incubated for 10 min in the presence of the antagonists Reactive Blue (300 M; P 2Y -selective), PPADS (100 or 300 M; poorly P 2X -selective), or suramin (300 M; nonselective P 2X and P 2Y antagonist), before ATP (50 M) or ␣,␤meATP (100 M) was added for 20 min. Subsequently, the cells were exposed to insulin (12 nM) for another 30 min, and the rate of dGlc transport was measured as described. Values are means from three to five independent experiments Ϯ S.E. * denotes statistical significance at the p Ͻ 0.05, and ** denotes statistical significance at the p Ͻ 0.01 level (analysis of variance).

FIG. 3. Influence of the extracellular concentrations of sodium, calcium, potassium, and chloride on the inhibition of insulin-dependent glucose transport by ATP.
Cardiac myocytes were incubated in isotonic media corresponding to medium A, except that the concentration of one of the ions to be tested was varied as indicated (for details see "Experimental Procedures"). Note that physiological concentrations of sodium, potassium, chloride, and calcium are ϳ145, 4, 120, and 1.5 mM, respectively. After a 10-min preincubation, ATP (50 M) was added for 20 min. The myocytes were then treated for another 30 min with insulin (12 nM) before the rate of dGlc was measured as described. Data are means of two to five independent experiments (Ϯ S.E. or range). In the data sets shown in panels A, C, and D, all values obtained in the presence of ATP were significantly lower than those measured in the corresponding ATP-free controls at the same ion concentration (p Ͻ 0.05 or less; paired Student's t test). Statistical analysis of data in panel C: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001 versus rate of insulin-stimulated glucose transport in ATP-free samples at the corresponding Ca 2ϩ level (paired t test); ¶, p Ͻ 0.05; ‡, p Ͻ 0.01 versus basal rate of glucose transport measured at [Ca 2ϩ ] ϭ 10 Ϫ3 M (analysis of variance). as to whether extracellular ATP might regulate other steps of glucose metabolism such as glycolysis and pyruvate oxidation. The glycolytic flux in cardiomyocytes was estimated from the rate of detritiation of [3-3 H]glucose, which is a measure of the flux through the 6-phosphofructo-1-kinase reaction (23,36). When the measurements were performed at a glucose concentration at which glucose transport represents the limiting step in the utilization of the sugar in these cells (i.e. 0.5 mM; Ref. 37), ATP caused a significant decrease in the rate of basal and insulin-stimulated [3-3 H]glucose detritiation, as expected (Fig.  5A). On the other hand, when the cells were incubated at a glucose concentration of 5 mM, a condition where the transport step no longer limits the uptake and metabolism of glucose (at least in the presence of insulin; Ref. 37), ATP failed to decrease the rate of glycolysis (Fig. 5B), which indicates that 6-phosphofructo-1-kinase reaction is not a direct target of purinergic action in these cells.
The rate of 14 CO 2 formation from [1-14 C]pyruvate by cardiac myocytes was used to monitor the activity of the pyruvate dehydrogenase complex of the cells in the presence and in the absence of ATP. To investigate the pyruvate dehydrogenase complex reaction in two different states of the enzyme complex, cells were either used in basal, nonstimulated conditions, as well as in the presence of the pyruvate dehydrogenase complex stimulator dichloroacetate (2 mM). As shown in Fig. 5C, extracellular ATP had no inhibitory action on the basal and dichloroacetate-stimulated rate of [1-14 C]pyruvate oxidation. On the contrary, ATP even tended to increase pyruvate oxidation in dichloroacetate-stimulated cells (Fig. 5C). DISCUSSION Extracellular ATP acting through purinergic receptors has been shown to exert pronounced effects on ionic currents and contractility in hearts and isolated cardiomyocytes (5)(6)(7)(8)(9)(10)(11)(12)(13)(14)17). In this report, we present first evidence that in heart muscle cells ATP is also endowed with metabolic properties, in terms of a marked inhibition of glucose transport, independently of the electrical or mechanical activity of these cells. Thus we found that in isolated, noncontracting cardiac myocytes, addition of micromolar concentrations of ATP and other purinergic agents brings about a large inhibition of glucose transport (Table I), which is independent of trans-sarcolemmal ionic gradients ( Fig. 3; see also below).
Purinergic Receptors Involved-Our pharmacological data clearly point to the involvement of P 2 purinoceptors in the inhibitory action of ATP on glucose transport: (i) absence of effect of the P 1 purinoceptor agonist adenosine (Table I), (ii) failure of P 1 antagonists to influence the effect of ATP (not shown), (iii) inhibition of glucose transport by P 2 -selective agonists in the micromolar range (Fig. 1), and (iv) suppression of this inhibitory effect by the P 2 antagonist PPADS (Fig. 2).
Furthermore, P 2Y receptors seem more likely to mediate the action of ATP than receptors of the P 2X -type. Thus, the selective P 2X agonist ␣,␤meATP was clearly less potent that the P 2Y agonist 2MeSATP (Fig. 1). Furthermore, ADP produced a marked inhibition of glucose transport (Table I), whereas this nucleotide is known to have only very weak agonist properties or even to be completely inefficient at P 2X sites (2, 38 -41). More importantly, P 2X receptors are ligand-gated cation chan- FIG. 4. Effect of ATP on the subcellular distribution of the glucose transporters GLUT4 and GLUT1. Cardiomyocytes were treated for 10 min with or without ATP (30 M) and then without (Bas) or with 12 nM insulin (Ins) for another 30 min. Subsequently, the cells were fractionated to obtain plasma membranes and intracellular membranes (LDM), and quantitation of GLUT4 and GLUT1 content was performed by Western blot analysis. Upper panels, representative autoradiograms. Lower panels, quantitative analysis of data from three to four independent experiments (means Ϯ S.E.). **, p Ͻ 0.01; ***, p Ͻ 0.001 versus insulin alone (paired t test). nels, and their effects are mediated by a depolarizing influx of sodium and especially of calcium (1,2,4). Therefore the insensitivity of the effect of ATP on glucose transport toward drastic changes in transmembrane gradients for sodium and calcium (Fig. 3) would be inconsistent with a receptor of the P 2X family being involved here.
The characteristics of the purinergic effect observed in this study, however, do not match the profile of known purinoceptors. In particular, this applies to the recently cloned P 2Y subtypes that were shown to be expressed in cardiac tissue and more specifically in adult cardiomyocytes of the rat, i.e. the P 2Y1 , P 2Y2 , and P 2Y6 receptors (32). Thus, in contrast to the glucose transport response to ATP (Table I and Figs. 1 and 2), activation of P 2Y1 purinoceptors was shown to be completely blocked by the P 2 antagonists suramin and Reactive Blue (42), and activation of P 2Y2 and P 2Y6 receptors was shown to be potently induced by UTP (43,44). Moreover, the agonist profiles of these receptors (P 2Y1 , P 2Y2 , and P 2Y6 ), as well as of P 2X receptors present in the adult heart (P 2X1 , P 2X4 , and P 2X5 ; Refs. 20, 40, and 45), are distinct from the rank of potencies found in the present study (ATP Ϸ ATP␥S Ϸ 2MeSATP Ͼ ADP Ͼ ␣,␤meATP; see Fig. 1) (38,40,(42)(43)(44)(45). On the other hand, it cannot be ruled out that the depression of cardiomyocyte glucose transport may involve more than one purinergic receptor.
Another point worth being highlighted is that purinergic inhibition of glucose transport is likely to be mediated by receptors and/or signals differing from those underlying the previously described effects of ATP on ionic processes and contractility in rat cardiomyocytes. In particular, many of these latter effects were shown to be dependent on extracellular calcium (9,(12)(13)(14) or chloride ions (9,11), in contrast to the response of the glucose transport system (Fig. 3). Moreover, ATP-dependent calcium influx was reportedly blocked by suramin (19), whereas the effect of the nucleotide on glucose transport was not influenced by this antagonist (Fig. 2). The same is true with respect to the efficiency of agonists. Thus, ADP was ineffective in inducing the cytosolic increase in calcium in cardiac cells (9,12), although it clearly depressed glucose transport (Table I and Fig. 1). Conversely, UTP was reported to increase the contraction amplitude as potently as ATP (14), whereas it had no effect on glucose transport (Table I).
Mechanism of Glucose Transport Inhibition-The rate of glucose transport is largely determined by the amount of glucose transporters (GLUTs) present in the plasma membrane, and this amount is in turn dependent on the trafficking of these transporters between the cell surface and intracellular membrane vesicles (46,47). As shown in Fig. 4, ATP causes a redistribution of GLUT4 and GLUT1 from the plasma membrane to intracellular vesicles (LDM) and thus influences the trafficking of the transporters between these compartments. There appears to be a discrepancy between the observation that ATP nearly abrogates the effect of insulin on the subcellular partition of GLUTs (Fig. 4) and the finding that the nucleotide only depresses insulin-dependent glucose transport by half (Table I). This may be explained by the fact that the type of methodology used, i.e. to fractionate different membrane compartments (plasma membrane and LDM), only allows a semi-quantitative evaluation of changes in the amount of transporters in these fractions. This is true, in particular, for the plasma membrane fraction, which is contaminated to some extend by intracellular membranes (26), so that the relative content of transporters is overestimated in samples from basal cells. An alternative explanation would be that insulin may, in addition to recruiting intracellular GLUTs to the cell surface, also increase the intrinsic activity of transporters and that this latter process is insensitive to ATP. Be that as it may, the results illustrated in Fig. 4 show that activation of P 2 purinergic receptors affects the GLUT trafficking machinery, and this is likely to represent the basis for the observed changes in the rate of glucose uptake.
Importantly, ATP influences the distribution of both GLUT4 and GLUT1. Because these transporter isoforms are confined to partly different intracellular locations (26), and display different trafficking kinetics (47), the target of the purinergic mechanism may be a general step of trafficking that is common to both GLUT4 and GLUT1, such as endocytosis, or alternatively, the recruitment of an intracellular pool containing both GLUT isoforms.
Likewise, it is of interest that the inhibitory effect of ATP on glucose transport was observed both in basal, nonstimulated cells, as well as in the presence of different glucose transport stimuli such as insulin, and the ␣-adrenergic agent phenylephrine (Table I), i.e. under a variety of conditions that are likely to differently influence the GLUT trafficking machinery. Thus, the trafficking kinetics of glucose transporters differ between the basal and the insulin-stimulated state (48); moreover, other studies indicate that phenylephrine probably affects another step of the trafficking cycle or another pool of transporters than  A and B, glycolysis. After a 10-min preincubation at 37°C, cardiomyocytes were treated with or without ATP (30 M) for 20 min and subsequently with or without insulin (20 nM) for 30 min in medium A supplemented with either 0.5 or 5 mM D-glucose, as indicated. [3-3 H]Glucose was then added, and its rate of detritiation was determined over a period of 30 min, as described under "Experimental Procedures." Values are means from three to four independent experiments Ϯ S.E. *, denotes a significant difference at the p Ͻ 0.05 level (paired Student's t test). C, pyruvate decarboxylation. Cardiomyocytes were incubated for 10 min with or without ATP (50 M) and for an additional 20 min with or without dichloroacetate (2 mM) before the rate of 14 CO 2 formation from [1-14 C]pyruvate was measured over a period of 10 min as described under "Experimental Procedures." Shown are means Ϯ S.E. from three to four independent experiments. insulin (28). Hence, the property of ATP to depress glucose transport regardless of the experimental conditions suggests that the nucleotide acts on a basic process involved in glucose transporter trafficking in these cells.
In addition to the ATP-dependent GLUT redistribution, one may also consider the possibility of a direct interaction of adenine derivatives with glucose transporters, as it was described for ATP or adenosine in erythrocyte membranes (49,50). However, this mechanism is unlikely to contribute to the effect of micromolar concentrations of ATP in cardiomyocytes, because binding of ATP or adenosine to erythrocyte glucose transporters only decreased the rate of glucose transport when the adenine compounds were used at millimolar concentrations (49,50). In cardiomyocytes, the ATP concentration had to be raised above the millimolar range (i.e. well beyond the level required to saturate the effect described in Fig. 1) to produce an additional depressing effect on glucose transport (down to 10% of control values; not shown).
Physiological Significance-At this stage, one can only speculate on the physiological significance of the purinergic effect on glucose transport. A finding of interest in this context is that ATP does not depress the glycolytic utilization of exogenous [3-3 H]glucose when the transmembrane transport no longer limits the overall rate of uptake (i.e. in the presence of insulin at 5 mM glucose; (37,51) and Fig. 5). This observation suggests that the inhibitory purinergic mechanism would only be relevant in situations where the transport step does determine the utilization of the sugar, i.e. for instance under basal conditions or under the influence of stimuli other than insulin, such as an increased myocardial workload or ischemia (37,51,52).
It is noteworthy that the range of ATP concentrations (ϳ1-10 M) found to be active with respect to glucose transport inhibition corresponds to reported blood levels in rats and humans (53). Hypoxia or ischemia cause a substantial release of ATP from heart cells (3). Although cardiac tissue contains high ectonucleotidase activities, locally released ATP (and ADP consequently formed) may restrict the uptake of glucose in the affected areas. Whether partial inhibition of glucose transport (and thus utilization) is deleterious or not under these pathological conditions remains to be determined.
Alternatively, ATP that is co-released with noradrenaline from sympathetic nerve endings may modulate the utilization of exogenous glucose in the (normoxic) heart, e.g. in terms of preventing an excessive uptake of the sugar when the contractile activity is increased by the catecholamine. In whatever physiological condition the purinergic inhibition of glucose transport may be effective, its very large extent (Table I) makes it worth being investigated in further detail in future studies.
Concluding Remarks-The present study describes a novel, metabolic action of extracellular ATP and related purine nucleotides, namely a pronounced inhibition of glucose transport in isolated cardiomyocytes. The decrease in glucose transport is brought about by a redistribution of glucose transporters (GLUT1 and GLUT4) from the cell surface to an intracellular location. This purinergic effect is mediated by P 2 purinoceptors, presumably of the P 2Y family. However, the purinoceptor type involved appears to differ from previously characterized members of the P 2 class and in particular from those mediating ionic and contractile effects of ATP in these cells.