INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

WITHIN THE PAST YEARS, the use of hepatocytes isolated from goldfish liver has evolved to a cellular model for studies of metabolic depression, in particular when these cells are challenged by anoxia (13, 26) or hypothermia (12). A number of mechanisms that allow viability to be maintained in the face of a severe reduction of intracellular ATP turnover were described. These mechanisms include the simultaneous inhibition of K⁺ influx and efflux across plasma membranes, the redirection of the available ATP production according to metabolic demands (15), and the prevention of anoxia-induced increase in cytosolic free Ca²⁺ (13). Among these and other adaptive responses that cells may employ, the common view is that intracellular ATP (ATP_i) should be guarded at all costs and that ATP does not permeate the plasma membrane of viable cells. However, evidence has accumulated that cellular ATP can indeed be released from viable cells through different mechanisms, such as exocytosis and transport through intrinsic plasma membrane proteins (for a review, see Refs. 8 and 24). Once released, ATP or its hydrolysis products can influence several biological processes, such as excitation, contraction, metabolism, and secretion. The extracellular metabolism of ATP is usually mediated by a family of enzymes called ecto-ATPases (24). These may act in a sequence together with ecto-ADPase (activity of which could be due to the same ecto-ATPase) and ecto-5'-nucleotidase to achieve complete dephosphorylation of extracellular ATP (ATP_e) to adenosine. The rate of nucleotide to nucleoside conversion is relevant, since it determines the effective time a given nucleotide is available to interact with surface receptors before being locally hydrolyzed by ectoenzymes. In mammalian liver, ecto-ATPase activity is largely restricted to the bile canalicular membrane (apical domain), where it is assumed to contribute to the degradation of nucleotides secreted into the canalicular lumen (22). Even though the source of ATP_e in the liver remains controversial, if the environment of the hepatocytes were to accumulate ATP, regardless of its origin, then the complete dephosphorylation of the nucleotide may serve for adenosine production. Adenosine, in turn, if not degraded extracellularly, could interact with specific purinoceptors (such as subtype 1) and/or be taken up by specific transporters for the resynthesis of nucleotides (4, 22). This process has systemic consequences, since, at least in mammals, the liver acts as a purine source for tissues that lack biosynthesis of purines (4).

Given the action that ATP_e and its hydrolysis products exert on cellular metabolism, the present study is aimed at 1) identifying ecto-ATPase activity in viable hepatocytes from goldfish and 2) studying the effects of ecto-ATPase activity on metabolism and vice versa, particularly under conditions of limited availability of metabolic energy. Goldfish hepatocytes are an adequate cellular system for this purpose, since they have shown great tolerance to pharmacological manipulation, including the use of metabolic inhibitors and toxic agents.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. [-³²P]ATP (10 Ci/mmol), [-³²P]ATP (30 Ci/mmol), [2,8-³H]ATP (32 Ci/mmol), adenosine 5'-O-(1-thiotriphosphate), and [-³⁵S]ATP (1,250 Ci/mmol) were purchased from DuPont. Suramin was purchased from Calbiochem. All other chemicals were obtained from Sigma.

Animals. Goldfish (Carassius auratus L.) were obtained from commercial suppliers and were kept in 200-liter aquaria at 20°C for 3 wk before being used.

Isolation of hepatocytes. The fish were killed, and hepatocytes were isolated from freshly excised livers as described in detail elsewhere (14). The final cell pellet was resuspended in an assay medium containing (in mM) 135 NaCl, 5 KCl, 1.3 CaCl₂, 1.2 NaH₂PO₄, 1.2 MgSO₄, 10 NaHCO₃, and 10 HEPES (pH 7.5 at 20°C), including 2% BSA, and kept on ice. Cells were counted in a Neubauer cytometer. Cellular viability assessed by trypan blue exclusion was >97% and remained constant until the end of the incubation period. Sixty minutes before measurements, the cell suspension (0.3 × 10⁶ to 3 ×10⁶ cells/ml) was preincubated at 20°C. Throughout the text, n represents the number of independent preparations used.

Conversion factors in hepatocytes. The ratio of fresh weight to cell number was 4.52 ± 0.13 mg fresh wt/10⁶ cells (n = 9), and the ratio of total protein content to cell number was 0.68 ± 0.09 mg protein/10⁶ cells (n = 13). Cell diameter assayed by a videomicroscope system was 12.83 ± 0.26 µm (51 cells measured from 3 independent preparations), and the percentage of dry weight was 24.7 ± 0.9 (n = 19).

Isolation of erythrocytes. Goldfish were killed by a blow on the head. Blood was withdrawn by cardiac puncture using heparinized syringes and subsequently spun for 4 s at 10,000 g to remove the plasma and the buffy coat. Erythrocytes were then incubated as described for isolated hepatocytes.

Assay of ecto-ATPase activity. Ecto-ATPase activity was determined at 20°C by following the release of -³²P_i from [-³²P]ATP as described elsewhere (25). In brief, the reaction was started by adding [-³²P]ATP to a hepatocytes suspension under continuous stirring. At different times, a 50- to 100-µl aliquot of the suspension was withdrawn and poured into 750 µl of a stop solution containing 4.05 mM Mo₇O₂₄(NH₄)₆ and 0.83 mM HClO₄. The ammonium molybdate solution formed a complex with the released phosphate, which was then extracted by adding 0.6 ml of isobutyl alcohol under vigorous stirring. After the phases were separated by centrifugation for 5 min at 1,000 g, 200-µl aliquots of the organic phase containing -³²P_i were transferred to vials containing 2.5 ml of 0.48 M NaOH. Radioactivity was measured by Cerenkov effect.

1

1

Production and uptake of adenosine. Production of adenosine from ATP was estimated at 20°C by following the release of -³²P_i from [-³²P]ATP using a method similar to that described for measuring ecto-ATPase activity. Under these conditions, one adenosine is formed for every -³²P_i produced.

Determination of cytosolic free Ca²⁺. Cytosolic free Ca²⁺ (Ca²⁺_i) was measured fluorometrically as previously described (13). In short, hepatocytes were loaded with 10 µM of fura 2-AM for 60 min in wrapped reaction tubes incubated at 20°C. After loading, cell suspensions were washed and resuspended in BSA-free incubation medium and placed in a thermostated cuvette equipped with a stirring device. Fluorescence measurements were performed at 20°C with a Hitachi F-2000 Fluorescence Spectrophotometer with excitation wavelength set to 340 nm and emittance wavelength set to 480 nm. Calibrations were performed on each sample applying the method described by Moon et al. (21) for a single excitation instrument. Ca²⁺_i concentrations were calculated according to the formula given by Grynkiewicz et al. (9) using the dissociation constant of 135 nM given therein.

Determination of ATP. ATP contents of the incubation medium were measured by incubating cells (70 × 10⁶cells/ml) in the presence of 5 mM EGTA. After 0, 60, and 120 min of incubation, duplicate samples of the cell suspension were withdrawn and rapidly centrifuged in small reaction tubes (5 s at 6,000 g), and an aliquot of the cell-free supernatant was removed and frozen for later analysis. ATP was measured in a luminometer using the luciferin-luciferase method as described by Brown (2).

Protein determination. This was carried out by the method of Lowry et al. (17).

Mathematical analysis. Paired t-test (2 sided) was used to determine differences in mean ecto-ATPase activities and Ca²⁺_i values among the different treatments used. For experiments with IAA plus CN, analysis of variance was used to test whether cellular viability and ecto-ATPase activity decreased with time. A P value of 0.05 was considered significant. Results of ecto-ATPase activity vs. [ATP] were analyzed by means of nonlinear regression. Each point of the curve represents the result of a single experiment in which the value of the slope (±SE) was obtained by linear regression of the time course of -³²P_i release with at least 10 experimental points.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Identification of ecto-ATPase activity in intact hepatocytes of goldfish. Figure 1 shows four experiments in which ecto-ATPase activity was measured as the rate of release of -³²P_i from labeled ATP. The reaction was linear in time from 5 to 5,000 µM ATP. No further production of -³²P_i was detectable after removal of the cells from the suspension. Stoichiometric chelation of both extracellular Mg²⁺ and Ca²⁺ with EDTA led to ~100% inhibition of ecto-ATPase activity.

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Production of -32Pi assayed at different ATP concentrations: 5 (A), 50 (B), 500 (C), and 5,000 µM (D). Open symbols represent -32Pi production after addition of 5 mM EDTA () or after removal of cells from suspension ().

1

1

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Ecto-ATPase activity vs. ATP (0.01-10 mM) in hepatocytes (0.8 × 106 to 3 × 106 cells/ml) of goldfish. Each point represents slope (±SE) obtained by linear regression of a time course of 32Pi release with at least 10 experimental points. Line represents fit by nonlinear regression of a single hyperbola to data. Different symbols represent 2 independent preparations.

Adenosine production and uptake. In experiments of Fig. 3, hepatocytes were incubated for up to 10 min with 5 µM of different labeled ATP compounds with the following results. Addition of [-³²P]ATP resulted in the release of -³²P_i and adenosine at 8.2 ± 0.3 pmol · (10⁶ cells¹) · min¹. After incubation of cells with [2, 8-³H]ATP, the time course of [2,8-³H]adenosine uptake showed a lag phase, after which it reached a rate of 7.9 ± 0.4 pmol · (10⁶ cells¹) · min¹. This was fully inhibited by 0.5 mM formycin B plus 0.125 mM dipyridamole. On the other hand, exposure of cells to the ATP analog, [-³⁵S]ATP, showed no significant uptake of label. Figure 3 (inset) compares the same data with a measurement of ecto-ATPase activity, which amounted to 48 pmol · (10⁶ cells¹) · min¹. Each set of data consisted of four independent measurements.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Time course of -32Pi release () from 5 µM [-32P]ATP, uptake of [2,8-3H]adenosine from 5 µM [2,8-3H]ATP in presence () and absence () of both 0.5 mM formycin B + 0.125 mM dipyridamole, and uptake of 5 µM [-35S]ATP (). Inset: same data are shown (except ), together with rate of -32Pi production () from 5 µM [-32P]ATP. Results are means ± SE of 4 independent preparations.

Effects of metabolic inhibition on ecto-ATPase activity. With the aim of inhibiting the two main sources of ATP production of the cell, hepatocyte suspensions were preincubated for up to 120 min in the presence of 0.5 mM IAA and 2 mM CN, inhibitors of glycolysis and oxidative phosphorylation, respectively. After 120 min, ATP_i decreased to 20% of the control value, whereas Ca²⁺_i increased about six times. At the same time ecto-ATPase activity (at 5 µM ATP) and cellular viability remained constant. In Fig. 4, results are shown as percentage of the respective control value at time 0. Control values were as follows: ATP_i 2.37 ± 0.25 nmol/10⁶ cells (n = 6), Ca²⁺_i 134 ± 47 nM (n = 4), and ecto-ATPase activity 0.018 ± 0.002 nmol · (10⁶ cells¹) · min¹ (n = 4).

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Intracellular ATP (ATPi; ), ecto-ATPase activity (), viability (), and concentration of free cytosolic Ca2+ (Ca2+i; ) after incubating hepatocytes with 2 mM NaCN + 0.5 mM iodoacetic acid. Results are means ± SE of 4-6 independent preparations.

Effects of nucleosides on Ca²⁺_i. In the absence or presence of 1.3 mM extracellular Ca²⁺ (Ca²⁺_e), addition of 5 µM ATP to the cell suspension increased Ca²⁺_i ~2.5 times (Table 2). As illustrated in Fig. 5, in the presence of Ca²⁺_e, the effect of ATP_e on Ca²⁺_i was dose dependent.

Leakage of ATP. Cell suspensions (70 × 10⁶ cells/ml) were incubated for 120 min in the presence of 5 mM EGTA to inhibit ecto-ATPase activity. Throughout the experiment, ATP_e remained constant at 12.22 ± 3.9 pmol/10⁶ cells (n = 9), i.e., 0.85 µM.

Ecto-ATPase activity of goldfish erythrocytes. Ecto-ATPase activity was 1.8 ± 0.1 pmol · (10⁶ cells)¹ · min¹ (n = 4) at 5 µM ATP and 10.3 ± 0.7 pmol · (10⁶ cells)¹ · min¹ (n = 4) at 50 µM ATP.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

This paper reports unequivocal evidence for the presence of ecto-ATPase activity in suspensions of goldfish hepatocytes. The external localization of the nucleotide-hydrolyzing site is supported by the observation that ATP added to the extracellular medium was hydrolyzed in intact cells, whereas hydrolysis of ATP did not continue when cells were removed from the suspension (Fig. 1). Suramin, one of the few substances known to inhibit ecto-ATPase activity in many systems (27), inhibited 70% of ecto-ATPase activity. A distinctive characteristic of the enzyme is the lack of sensitivity to specific inhibitors of P-, F-, and V-type ATPases (Table 1) and the absolute requirement of divalent cations for activity (Fig. 1). The lack of effect of vanadate points to the absence of a phosphorylated intermediate. Ecto-ATPase activity was measured over a wide range of hepatocyte (0.3 × 10⁶ to 10 × 10⁶ cells/ml) and ATP (5-10,000 µM) concentrations. When millimolar concentrations of extracellular Ca²⁺ and Mg²⁺ were used, the apparent V_max as well as the K_1/2 were in the range of those reported in many other systems (see Ref. 24), including rat hepatocytes (16).

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Representative traces of Ca2+i illustrating effect of ,- methyleneadenosine 5'-triphosphate (AMP-PCP) and extracellular ATP (ATPe) in presence of Ca2+e (A) and ATPe in absence of Ca2+e (B).

In addition to a characterization of enzyme activity as such, the present study tried to shed light on the potential cellular function of ecto-ATPase activity, in particular during situations of energy limitation.

For this purpose we incubated hepatocytes with CN and IAA, inhibitors of oxidative phosphorylation and glycolysis, respectively. As shown in Fig. 4, this treatment decreased ATP_i by ~80% and increased Ca²⁺_i up to sixfold. However, ecto-ATPase activity as well as cellular viability remained unchanged. If it is assumed that ecto-ATPase in goldfish hepatocytes is an integral membrane protein, this would imply that cytosolic aspects of the enzyme do not sense changes in the ATP_i concentration of this magnitude and that the severe metabolic consequences of cutting the most important pathways of ATP production are not relevant for ecto-ATPase function. The question arises as to how ecto-ATPase activity can be included in a general metabolic scheme. Isolated goldfish hepatocytes at 20°C showed an ATP_i of ~3.5 mM (estimated from Refs. 6 and 11, as described in Mathematical analysis).

In these cells, one of the most important promoters of ATP_i consumption is the Na⁺-K⁺-ATPase, whose activity in intact cells at 20°C is ~0.4 nmol P_i · (10⁶ cells)¹ · min¹ (11, 26). This is almost the same activity that ecto-ATPase activity displays at the same temperature in the presence of only 40 µM ATP_e. The difference between both enzyme activities is even more pronounced under metabolic inhibition caused by chemical anoxia or hypothermia: 90-min incubation of hepatocytes with CN decreases Na⁺-K⁺-ATPase activity to ~50%, representing 50-90% of total ATP_i turnover (14). However, incubation of hepatocytes with 2 mM CN for up to 120 min had no influence on ecto-ATPase activity (unpublished results). During cooling Na⁺-K⁺-ATPase activity displayed an apparent Q₁₀ ranging from 2 [Q_10(15-25°C)] to 8 [Q_10(8-15°C)] (12), whereas for ecto-ATPase activity apparent Q_10(5-20°C) was ~1.2-2 (7). Thus, due to the high specific activity displayed by ecto-ATPase and its relative independence from environmental and/or metabolic factors, a putative ATP liberated into the surroundings of liver cells will be rapidly hydrolyzed. Once ATP is hydrolyzed to ADP or even further to AMP (because some ecto-ATPases display ecto-ADPase activity as well), the ultimate fate of the nucleotide depends on the presence of 5'-nucleotidase.

Experiments of Fig. 3 were specifically designed to test for the existence of ectonucleotidases capable of completing the sequence leading to the total dephosphorylation of ATP_e. By incubation of cell suspensions with [-³²P]ATP, the liberation of -³²P_i may be expected only if ectoenzymes are present that liberate -, -, and -P_i of ATP at the surface of the cell. This is indeed what we observed: the total hydrolysis of the nucleotide was determined through the liberation of -³²P_i from [-³²P]ATP. Moreover, experiments using [2,8-³H]ATP showed that as soon as adenosine is released, it is taken up at almost the same rate (Fig. 3). That is, the rates of -³²P_i production and [³H]adenosine uptake were similar, supporting the idea of a mechanism for adenosine salvage. Because dipyridamole plus formycin B fully inhibited the uptake of [³H]adenosine, we postulate adenosine uptake to occur through facilitated diffusion.

Figure 3 also shows that the initial rate of -³²P_i release was about four times higher than adenosine production. This agrees well with results of other systems showing a delay in adenosine production due to inhibition of 5'-nucleotidase (which promotes conversion of AMP to adenosine) by ADP (20).

Concerning the use of [³H]ATP, it could be argued that these experiments rely on the assumption that nucleotides are impermeable to the membrane. However, when adenosine uptake was inhibited, there was no uptake of label from [³H]ATP. Moreover, [-³⁵S]ATP, a nonhydrolyzable ATP analog, did not permeate the plasma membrane (Fig. 3).

In rat hepatocytes, ATP_e binds to P_2Y receptors (10), causing a rapid mobilization of Ca²⁺_i from intracellular Ca²⁺ stores via inositol trisphosphate (23). Our results with goldfish hepatocytes are in agreement with these studies. It can be seen (Table 2, Fig. 5) that low ATP_e, independently of the presence of Ca²⁺_e, strongly increased Ca²⁺_i. ATP (5 µM) is sufficient to increase Ca²⁺_i more than two times without significantly changing the amount of P_i of the incubation medium. Thus the observed effect cannot be due to P_i activation of a megachannel at the mitochondrial membrane, as postulated for rat hepatocytes (28). An effect on P₁ receptors by ATP degradation products was ruled out, since 100 µM adenosine has little and 100 µM AMP no effect on Ca²⁺_i (Table 2). That AMP-PCP, a nonhydrolyzable ATP analog, has no effect on Ca²⁺_i is consistent with its postulated action on P_2X receptors, which are not linked to intracellular Ca²⁺ release. Typical P_2Z receptors can also be excluded, since addition of 1 mM ATP in the presence of extracellular Ca²⁺ leads to a rapid and reversible increase of Ca²⁺_i (Fig. 5). As reported for rat hepatocytes (29), P_2Z receptors mediate a late and sustained increase of Ca²⁺ that causes irreversible hepatocellular injury. Finally, the fact that 100 µM ATPS significantly increased Ca²⁺_i confirms that ATP hydrolysis is not necessary and that P₁ receptors are not involved.

These results, together with the experimental evidence showing that the increase of Ca²⁺_i at low [ATP]_e is entirely due to intracellular stores (Table 2), support the hypothesis that the Ca²⁺_i increase is exerted through binding of the nucleotide to specific P_2Y receptors. The rank order for the effect of the different nucleosides on Ca²⁺_i was ATP >> ATPS > ADP > adenosine.

According to the above-mentioned results, ectonucleotidases are expected to reduce the effective concentration of hydrolyzable purinoceptor agonists and thus restrain the magnitude and/or duration of cell responses. Coupling of ecto-ATPase and 5'-nucleotidase with a nucleoside transport system would then decrease the toxicity of ATP_e and promote adenosine conservation.

ATP has been identified in the fluid surrounding many cell types (24). However, in goldfish hepatocytes, as in many other systems, the source of ATP remains to be detected. As mentioned in the results, we were unable to detect a significant leakage of ATP from hepatocytes. The ultrastructure of fish liver differs from that of higher vertebrates in that there is a large blood-hepatocyte exchange area (3). This means that, in principle, a putative ATP being released from blood cells, endothelial cells, or any other cell type could easily interact with the basolateral membrane of goldfish hepatocytes. However, considering the high levels of ecto-ATPase activity of goldfish erythrocytes (as opposed to mammalian erythrocytes, see Ref. 5), our present data tend to rule out a significant role of these cells as a source of ATP. Alternatively, ATP could also arise as a consequence of the export from endothelial cells (27). However, as pointed out by Gordon (8), ATP could locally reach micromolar concentrations without being detectable. With respect to the goldfish, on the basis of the fact that this species is a good anaerobe whose cells may repeatedly experience prolonged periods of hypoxia and steep oxygen gradients, the following scenario may be proposed. If as a consequence of any conditions of extreme energy limitation a certain population of cells died, their nucleotides would be released into the interstitial space with the subsequent uptake of adenosine by the surviving cells. As seen in mammalian hepatocytes, this extra adenosine may improve the energetic state of cells by increasing the phosphorylation potential together with a decreased glycolytic flux (18, 19). In this way, dying cells would constitute part of a mechanism mediated by ectoenzymes that improve the survival chances of the rest of the cells.

When trying to generalize about the possible functions of ecto-ATPase, a last word of caution should be given with regard to the model of hepatocytes isolated by the use of collagenase, since their plasma membranes may have lost their functional heterogeneity. In the intact liver, ecto-ATPase activity may be localized in different membrane domains and probably have different functions. For example, the function of ecto-ATPase acting near purinoceptors may be different from a canalicular ecto-ATPase promoting adenosine salvage.

Perspectives

Another point that deserves clarification is the role of the adenosine liberated as a consequence of the extracellular dephosphorylation of nucleotides. Similar to the action of ectoenzymes terminating the effect of nucleotides on P₂ receptors, the observed uptake of adenosine could be a mechanism to terminate the effect of the nucleoside on a putative P₁ receptor (23). In this way, ectoenzymes and nucleoside uptake would act in concert to control the action of agonists on purinoceptors.