Inhibition of ecto-5'-nucleotidase by nitric oxide donors. Implications in renal epithelial cells.

We evaluated, in renal epithelial cells with a proximal tubule phenotype, the effect of nitric oxide (NO) on ecto-5 -nucleotidase (5'-N U), the underlying mechanism and its functional consequence. Sodium nitroprusside (SNP, 1-1000 microM), a NO donor, inhibited 5'-NU activity in a time- and concentration-dependent manner. Consequently, NO blunted the inhibition by extracellular cyclic AMP (cAMP, 10-1000 microM) of sodium-phosphate cotransport, a pathway which involves degradation of adenosine monophosphate (AMP) by 5'-NU. SNP-induced inhibition of 5'-NU was not mediated by cyclic GMP, since it was not mimicked by atrial natriuretic peptide, and was reproduced by isosorbide dinitrate and sodium nitrate, two NO donors. SNP and genuine NO decreased the activity of 5'-NU in renal homogenates, and the effect of SNP was potentiated by dithiothreitol and glutathione, but not by nicotinamide adenine dinucleotide. In vivo in rats, kidney ischemia/reperfusion, which activates inducible NO-synthase, inhibited renal 5'-NU. This inhibition was prevented by Nomega-nitro-L-arginine methyl ester, a NO-synthase inhibitor. These results indicate that: (i) NO-related activity inhibited the activity of an ecto-enzyme, 5'-NU, most likely through S-nitrosylation of the enzyme; (ii) inhibition of 5'-NU activity by NOx, which can occur in vivo under pathophysiological conditions, affected the extent to which extracellular cAMP inhibited sodium-Pi cotransport.

i.e. adenosine triphosphate, adenosine diphosphate, adenosine monophosphate (AMP), and cyclic AMP (cAMP), into adenosine (2,3). Released adenosine can be taken up by proximal tubular cells through dipyridamole-sensitive carriers (4 -7) and phosphorylated into adenine nucleotides. This cascade of events was shown to account for the protective effect of extracellular adenine nucleotides on tubular function during and after anoxia (8). Our previous studies have evidenced that degradation of extracellular cAMP in the tubular lumen followed by adenosine uptake were mandatory steps in the well known inhibitory effect of extracellular cAMP on renal proximal phosphate (P i ) reabsorption (9). Through this pathway, luminal cAMP (nephrogenous cAMP), added to the tubular fluid under the influence of parathyroid hormone, is not only a marker of the activity of parathyroid hormone but also participates in the overall phosphaturic effect of the hormone (9 -11). We have recently reported that parathyroid hormone-stimulated 5Ј-NU activity via a mechanism which involved protein kinase C activation and de novo protein synthesis (12).
Nitric oxide (NO) is a local mediator which is synthesized from L-arginine by numerous cell types including endothelial cells, activated macrophages, and renal tubular cells under physiological or pathological conditions (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23). Cellular targets of NO and signaling pathways involved in its pleiotropic effects have been extensively studied in the past decade. Stimulation of soluble guanylate cyclase (24 -26), ADP-ribosylationlike reaction with proteins (27)(28)(29) and, more recently, S-nitrosylation of proteins (30 -37) have been reported as biochemical events which accounted for the actions of NO. S-Nitrosylation results from direct or indirect (via intermediate S-nitrosothiols) transfer of NO ϩ to thiol groups of proteins (34,35,37). Such reactions were recently shown to affect the activity of nuclear, cytosolic, and membrane-bound proteins (34) including heterotrimeric G proteins (33), p21 ras (36), and glyceraldehyde-3-phosphate dehydrogenase, a key enzyme of glycolysis (32). However, a direct interaction between NO and an ecto-enzyme has not been reported.
The aim of the present study was: (i) to evaluate whether NO-related activity affected 5Ј-NU activity of proximal tubular cells and the extent to which such a modulation might influence inhibition of sodium-P i cotransport by extracellular cAMP; (ii) to elucidate the mechanism involved in the effect of NO; (iii) to identify possible conditions in which NO inhibits 5Ј-NU in vivo. We show that NO x inhibits 5Ј-NU activity in a cyclic GMP (cGMP)-and protein synthesis-independent manner, most likely through S-nitrosylation of the enzyme, and that renal ischemia/reperfusion results in NO x -dependent inhibition of 5Ј-NU.
Cell Culture-Opossum kidney (OK) cells (passages 80 -100) were grown to confluence in 24-well trays in a medium consisting of a 1:1 (v/v) mixture of Ham's F-12 and Dulbecco's modified Eagle's medium containing 15 mM Hepes, 21.5 mM HCO 3 , 1 mM sodium pyruvate, 4 mM L-glutamine, 50 units/ml penicillin, 50 g/ml streptomycin, 50 nM sodium selenite, 5 g/ml insulin, 5 g/ml transferrin, 50 nM hydrocortisone, and 2.5% fetal calf serum. Medium was changed on alternate days. Monolayers of OK cells reached confluence after 4 days, and they were used for experiments 2 or 3 days after confluence was achieved. Cells were subcultured weekly by trypsinization. The splitting ratio was 1 to 5.
On the day prior to experiments, culture medium was changed to hormone-free and serum-free medium, and, on the day of experiment, preincubations were usually performed in the same medium to which drugs were added as concentrated aliquots.
Determination of Cyclic GMP Content-After removal of culture medium, cells were washed with 1 ml/well of HBS supplemented with 15 mM Hepes and 2 mM L-glutamine (HBS-Hepes), and were then preincubated for 15 min at 37°C in HBS-Hepes (500 l/well) containing 0.5 mM 3-isobutyl-1-methylxanthine, a nonspecific phosphodiesterase inhibitor. Medium was then removed and cells were incubated usually for 5 min in a similar solution to which hormones or drugs were added. At the end of incubation, intracellular cGMP content was measured by radioimmunoassay as described previously (39).
Determination of 5Ј-NU Activity-The activity of ecto-5Ј-NU was determined on intact OK cells using a method adapted from Gentry and Olsson (40) as described previously (12). Briefly, after removal of the culture medium, cells were rinsed twice with HBS-Hepes, pH 7.4, and then incubated in the same solution (500 l/well) in the presence of [ 14 C]5ЈAMP (0.02 Ci/ml), unlabeled 5Ј-AMP (usually 10 M except for determination of kinetic parameters), and 20 M dipyridamole which abolishes cellular uptake of generated adenosine (6,7,9). At the end of incubation, 400 l/well of incubation medium were mixed with 100 l of 0.15 M ZnSO 4 and 100 l of 0.15 M Ba(OH) 2 . After vigorous shaking, the mixture was centrifuged (10,000 ϫ g; 5 min) in order to separate adenosine from 5Ј-AMP and 400 l of supernatant, which contains adenosine, were counted by liquid scintillation. Blank values were measured in the absence of cells and represented less than 0.5% of added radioactivity and, in any case, less than 10% of the signal attributed to genuine 5Ј-NU. In each experiment, recovery of adenosine was evaluated with [ 3 H]adenosine and was found to average 75-80%.
Activity of 5Ј-NU from renal homogenates was determined by incubating aliquots of the homogenate (2-4 g of protein/ml) in HBS-Hepes solution in the presence of labeled and unlabeled 5Ј-AMP as described above during 20 min at 37°C. The reaction was terminated as for OK cells.
Uptake Studies-Uptake of P i was performed as described previously with minor modifications (9). Briefly, uptake studies were performed at 37°C in a buffered solution with the following composition (millimoles/ liter): 137 NaCl, 5.4 KCl, 1 CaCl 2 , 1.2 MgSO 4 , 15 Hepes (pH 7.4). The sodium-free solution was made isoosmotic by replacing sodium chloride with N-methyl-D-glucamine. After removal of culture medium, cells were washed with 1 ml/well of the uptake solution, and were incubated for various periods of time in the presence of K 2 H 32 PO 4 (0.5 Ci/ml) and 100 M KH 2 PO 4 . All these steps were performed at 37°C. At the end of incubation, the uptake was stopped by washing the cells three times with 1 ml/well of ice-cold solution (137 mM NaCl, 15 mM Hepes, pH 7.4). Cells were then solubilized in 0.5% Triton X-100 (250 l/well) and aliquots were counted by liquid scintillation.
In Vivo Studies-Sprague-Dawley rats, weighing 180 -200 g, were anesthesized and infused as described previously (9 -11). Two groups of four rats (n ϭ 4 in each group) were studied. In the first group, the pedicle of the left kidney was clamped during 15 min. At the end of that period, reperfusion was allowed during 60 min. In the second group, rats were treated similarly except that L-NAME infusion (50 g/min/ 100 g body weight), starting after a priming dose of 5 mg/100 g body weight, was initiated 15 min prior to renal ischemia.
At the end of the reperfusion period, both kidneys of each animal were then removed, decapsulated, and homogenized with a Teflon Potter-Elvehjem device in an ice-cold buffer (250 mM sucrose, 5 mM Tris-HCl, 3 mM MgCl 2 , 1 mM EDTA, 1 mM levamisole, pH 7.4). Homogenates were aliquoted and stored under liquid nitrogen until determination of 5Ј-NU activity as described above.
Presentation of Data-5Ј-NU activity and P i uptake were expressed as nanomoles/mg of protein (41). 5Ј-NU activity was calculated as 10 M AMP-PCP-sensitive 5Ј-AMP hydrolysis. Sodium-dependent uptakes were calculated by subtracting uptake values measured in the presence of N-methyl-D-glucamine from those measured in the presence of sodium. Intracellular cGMP content was expressed as picomoles/mg of protein. Results were presented as mean Ϯ S.E. of three to five different experiments (n) in which duplicates were obtained. One-way or twoway analyses of variance were performed and, when allowed by the F value, results were compared by the modified t test (42).

RESULTS
Effect of NO Donors on 5Ј-NU Activity in OK Cells-As previously reported, OK cells exhibited ecto-5Ј-NU activity (Fig. 1). This activity increased linearly with the time of incubation up to 120 min, both under control conditions and after preincubation with 1 mM SNP (Fig. 1, panel A). A 60-min incubation time was therefore chosen in subsequent experiments. SNP decreased 5Ј-NU in a time-and concentration-dependent manner (Fig. 1, panels B and C). The effect of SNP was already apparent after 30 min of incubation and reached significance after 60 min (Fig. 1, panel B). After 3 h of incubation, SNP was inhibitory at the lowest concentration of 10 M. The magnitude of this effect increased with SNP concentration up to 1 mM (Fig. 1, panel C). We next examined the effect of SNP on the kinetic parameters of 5Ј-NU. As shown in Fig. 1, panel D, SNP, at 1 mM, affected the V max value of 5Ј-NU which decreased significantly from 46 Ϯ 4.3 to 27 Ϯ 4.5 nmol/mg of protein/60 min (n ϭ 3, p Ͻ 0.05). In contrast, apparent K m values were not different whether cells were pretreated with SNP or vehicle (71 Ϯ 12 versus 57 Ϯ 8 M, n ϭ 3, NS).
We have previously reported that 5Ј-NU played a key role in the inhibitory effect of extracellular cAMP on sodium-P i cotransport (9,12). In order to evaluate the influence of NO on this inhibitory pathway, we measured sodium-dependent P i uptake after that OK cells had been preincubated with increasing concentrations of cAMP in the presence or absence of SNP or AMP-PCP. As expected, extracellular cAMP (10 -1000 M) inhibited P i uptake in a concentration-dependent manner (Table I). AMP-PCP, a potent inhibitor of 5Ј-NU, blunted significantly the inhibition by cAMP. The effect of AMP-PCP was mimicked by SNP, although to a lesser extent. It is noteworthy that neither AMP-PCP nor SNP affected P i uptake by themselves.
In order to confirm that NO x was indeed responsible for the inhibitory effect of SNP on 5Ј-NU, we evaluated: (i) the effect of another NO donor, isosorbide dinitrate; (ii) the effect of the ferricyanide and ferrocyanide moieties. As shown in Fig. 2, panel A, isosorbide dinitrate and SNP, each of them at 1 mM, inhibited 5Ј-NU to a similar extent. In contrast, neither K 3 Fe(CN) 6 nor K 4 Fe(CN) 6 , at the same concentration, affected significantly 5Ј-NU activity.
Because NO was reported in many cellular systems to act through generation of cGMP, we evaluated whether this signaling pathway was involved in inhibition of 5Ј-NU. For that purpose, we compared the effects of SNP and ANP on cGMP accumulation and 5Ј-NU activity. SNP-induced increase in intracellular cGMP content was modest and did not reach signif-icance (Fig. 2, panel B). In contrast, ANP stimulated dramatically cGMP generation. However, ANP was without effect on 5Ј-NU activity (Fig. 2, panel C).
In previous studies from several groups including ours (12), modulation of 5Ј-NU activity was reported to depend on de novo protein synthesis. Regarding the effect of NO x , this possibility was evaluated by a pretreatment of OK cells with cycloheximide or actinomycin D at concentrations previously reported to abolish the modulation of the enzyme by protein kinase C activators (12). Cycloheximide or actinomycin did not prevent NO-induced inhibition of 5Ј-NU activity (Table II).
S-Nitrosylation of proteins with NO x was recently described (30 -37) and was reported in some instances to account for inhibition of enzymatic activities (32,34). The possibility that a similar mechanism was involved in NO x -induced inactivation of 5Ј-NU was investigated. The effect of NO donor SNP was potentiated by addition of the reducing agent glutathione (GSH) to the incubation medium (Fig. 3). Incubation of OK cells during 3 h with GSH alone, 0.01 to 1 mM, had no effect on 5Ј-NU activity. However, the presence of GSH together with SNP during the preincubation period increased markedly the effect of the NO donor: SNP, at 10 M, decreased 5Ј-NU activity by 7, 13, 39, and 40% in the presence of GSH at a concentration of 0, 0.01, 0.1, and 1 mM, respectively.
Effect of NO Solutions, NO Donors, and Renal Ischemia on Rat Renal 5Ј-NU Activity-Finally, we asked the question of whether inhibition of renal 5Ј-NU by NO x might occur under pathophysiological conditions. For that purpose, we induced an ischemia/reperfusion injury in rat kidneys because hypoxia/ reoxygenation injury was recently shown to increase NO production, likely through activation of inducible NO-synthase in proximal tubules (20). As shown in Fig. 4, panel A, 15-min ischemia followed by 60-min reperfusion resulted in a marked decrease of 5Ј-NU activity. This inhibition was entirely prevented by infusion of L-NAME at a rate which did not affect per se 5Ј-NU activity (Fig. 4, panel B).
In a last set of experiments, the in vitro effect of NO solutions and of two NO donors, SNP and SNAP, was evaluated on 5Ј-NU activity in homogenates prepared from control rat kidneys.  M NO, respectively (basal value was significantly different from each experimental condition, p Ͻ 0.01). Furthermore, the inhibitory effect of SNP was potentiated by GSH and dithiothreitol, but not by NAD (Table III). DISCUSSION The main results of the present study are that: (i) in renal epithelial cells, NO donors inhibited 5Ј-NU activity in a cGMPand protein synthesis-independent manner; (ii) this effect re-sulted in impairment of cAMP-induced inhibition of sodium-P i cotransport; (iii) S-nitrosylation of the enzyme, either direct or indirect, is likely to underlay enzymatic inactivation; and (iv) in vivo, NO overproduction during ischemia/reperfusion injury led to inhibition of 5Ј-NU activity. To our best knowledge, this is the first demonstration of a direct interaction between a nitrogen oxide and an ecto-enzyme.
The mechanism of 5Ј-NU inhibition by NO x differs from that involved in previously reported hormonal modulation of renal 5Ј-NU (12,(43)(44)(45). In cultured glomerular mesangial cells, cAMP-protein kinase A activating substances, such as dopamine, and tumor necrosis factor-␣ or interleukin-1␤ were shown to stimulate 5Ј-NU in a cycloheximide-dependent manner (43)(44)(45). In OK cells, we established that parathyroid hormone stimulation of 5Ј-NU through protein kinase C was dependent on cycloheximide and actinomycin D (12). In contrast, de novo protein synthesis was not involved in the effect of NO as evidenced by the short-term action of this compound (Fig. 1) and its persistence in the presence of cycloheximide and actinomycin D (Table II).
The inhibitory effect of NO x on 5Ј-NU did not result from   (15,16,24). Our data argue against the involvement of this pathway in 5Ј-NU inhibition since: (i) NO x had a modest effect on cGMP generation in OK cells; (ii) ANP, a well known agonist of particulate guanylate cyclase, which increased dramatically cGMP generation in OK cells ( Fig. 2 and Ref. 46), did not affect 5Ј-NU activity; (iii) the effect of NO x on the enzyme was also observed in homogenates of renal tissue. Nitrosylation was recently reported to affect the activity of a large number of membrane-bound, cytosolic and nuclear proteins (30 -37). This expression of a wide variety of effects is achieved through interaction of nitrogen oxides with targets via a complex redox signaling and additive chemistry (34,35,37). As regards inhibition of 5Ј-NU, it may result from interaction with NO or with congeners NO ϩ and NO Ϫ . Indeed, SNP, which inhibited 5Ј-NU in the two preparations used in the present study, is better regarded as an NO ϩ donor rather than an NO donor (34,37). The observation that the effect of SNP was potentiated in the presence of thiols such as dithiothreitol or glutathione (Fig. 3 and Table III) raises the possibility that these compounds first interact with NO ϩ and that RSNO compounds then inhibit 5Ј-NU, probably by S-nitrosylation. It is noteworthy that neither glutathione nor dithiothreitol alone affected 5Ј-NU activity, a feature which contrasts with the reported inhibition of bull seminal plasma 5Ј-NU by dithiothreitol (47). This apparent discrepancy can be attributed to the fact that dithiothreitol concentrations used in our study were 2 to 3 orders of magnitude lower than those reported to inhibit 5Ј-NU (47). Alternatively, our data showing that genuine NO solutions decreased the activity of 5Ј-NU from kidney homogenates is also consistent with the possibility of a direct interaction between NO ⅐ and the enzyme. It can be pointed out that NO x was active within a concentration range similar to that reported to affect the activity of heterotrimeric G proteins and p21 ras (33,36). Along the same line, SNAP, which can be regarded as an NO donor, inhibited renal 5Ј-NU as well. Finally, the possibility that decreased activity of the enzyme resulted from an interaction between a nitrogen oxide and the zinc moiety of 5Ј-NU, which is a zinc metalloprotein (48,49), is unlikely since nitric oxide does not react with zinc, an element which was shown to be crucial for 5Ј-NU activity in the mammalian membrane-bound form of the enzyme (48,49).
The interaction between NO x and 5Ј-NU differs from that described between NO and glyceraldehyde-3-phosphate dehydrogenase: in the latter case, S-nitrosylation of the protein preludes to covalent linkage of NAD, a cofactor of the enzyme (31). This reaction was first interpretated as ADP-ribosylation since NO was also reported to stimulate an ADP-ribosyltransferase (27)(28)(29). In our model, the observation that NAD, alone or in combination with SNP, had no effect on 5Ј-NU activity rules out such a possibility. It should be stressed that the ectoenzymatic situation of 5Ј-NU made unlikely an interaction with NAD.
In proximal tubular cells, 5Ј-NU was shown to be involved in modulation of P i reabsorption and in restoration of intracellular stock of ATP following ischemia (8,9,12). (i) The inhibitory effect of extracellular cAMP on sodium-P i cotransport was previously shown to require extracellular degradation of the nucleotide by phosphodiesterases and 5Ј-NU and subsequent uptake of adenosine (9); (ii) the protective effect of extracellular nucleotides on intracellular ATP content during ischemia also requires degradation of extracellular nucleotides followed by adenosine uptake (8). Our present finding that impairment of 5Ј-NU activity by NO x blunts the phosphaturic effect of extracellular cAMP demonstrates that 5Ј-NU inhibition has functional implications in terms of P i homeostasis: the relief of tonic Left kidneys were subjected to 15-min ischemia followed by 60-min reperfusion while right kidneys served as controls. All along the experiments, rats were infused either with saline (panel A) or with L-NAME (50 g/min/100 g body weight after a priming dose of 5 mg/100 g body weight) (panel B). 5Ј-NU activity was determined on renal homogenates. *, significantly different from the value of the right (control) kidneys, n ϭ 4, p Ͻ 0.05.