Distinct roles for recombinant cytosolic 5'-nucleotidase-I and -II in AMP and IMP catabolism in COS-7 and H9c2 rat myoblast cell lines.

Catabolism of AMP during ATP breakdown produces adenosine, which restores energy balance. Catabolism of IMP may be a key step regulating purine nucleotide pools. Two, cloned cytosolic 5'-nucleotidases (cN-I and cN-II) have been implicated in AMP and IMP breakdown. To evaluate their roles directly, we expressed recombinant pigeon cN-I or human cN-II at similar activities in COS-7 or H9c2 cells. During rapid (more than 90% in 10 min) or slower (30-40% in 10 min) ATP catabolism, cN-I-transfected COS-7 and H9c2 cells produced significantly more adenosine than cN-II-transfected cells, which were similar to control-transfected cells. Inosine and hypoxanthine concentrations increased only during slower ATP catabolism. In COS-7 cells, 5'-nucleotidase activity was not rate-limiting for inosine and hypoxanthine production, which was therefore unaffected by cN-II- and actually reduced by cN-I- overexpression. In H9c2 cells, in which 5'-nucleotidase activity was rate-limiting, only cN-II overexpression accelerated inosine and hypoxanthine formation. Guanosine formation from GMP was also increased by cN-II. Our results imply distinct roles for cN-I and cN-II. Under the conditions tested in these cells, only cN-I plays a significant role in AMP breakdown to adenosine, whereas only cN-II breaks down IMP to inosine and GMP to guanosine.

(cN-I and cN-II) 1 (EC 3.1.3.5) have been implicated. The cN-I was very recently cloned and shown to catalyze adenosine formation during ATP breakdown (9). The role of cN-II is less clear. Levels of expression of cN-II are usually higher in cultured cells compared with the corresponding tissue. It has therefore been suggested that cN-II is particularly important in cells with high DNA synthesis or turnover (10). The cN-II was cloned from chicken, human, and ox (11,12) and was later shown to have activity in vitro (12,13). Stable cell lines overexpressing cN-II were recently shown to have slightly reduced total nucleoside triphosphate pools (14). However, the function of cN-II in cells undergoing ATP breakdown has remained untested.
From enzyme kinetic data, the cN-I displays a preference for AMP over IMP and is stimulated by ADP but not ATP (15,16), implying that it is active against AMP during ATP catabolism. Furthermore, its pH regulation (17) and studies with selective inhibitors (18) have implicated cN-I in adenosine formation in the heart, adding weight to the cloning and expression data (9). The cN-II displays selectivity for IMP and GMP over AMP, is stimulated by both ATP and ADP, and is inhibited by inorganic phosphate (19). These data suggest a selective role for cN-II in catabolism of IMP and cast doubt on whether it is active during ATP catabolism. On the other hand, the K m for AMP of cN-II measured in the presence of optimal activator, ATP, is similar to that of cN-I (approximately 5 mM) (19), and studies with selective inhibitors have also implicated cN-II in adenosine formation (20) during ATP breakdown.
The availability of cDNA clones for cN-I and cN-II has allowed us for the first time to compare their ability to catalyze AMP and IMP degradation directly in the same cell types. The overexpression strategy employed here allowed us to increase the activity of cN-I or cN-II selectively, without altering the levels of the other enzymes in the pathway. We chose to study two different cell types, COS-7 and H9c2 cells. COS-7 cells were chosen because of their ease of transfection and because flux through AMP deaminase was low compared with cytosolic 5Јnucleotidase (see Fig. 1). Hence, IMP did not accumulate, and this was an ideal cell line in which to compare the activity of cN-I and cN-II in adenosine formation. In order to compare the effect of cN-I and cN-II overexpression on IMP degradation, it was necessary to use a cell line in which AMP deaminase was greater than cytosolic 5Ј-nucleotidase activity and hence IMP accumulated. Skeletal muscle contains the highest concentration of AMP deaminase (21); here, AMP deaminase plays a role in the initiation of the purine nucleotide cycle (22). We therefore investigated the H9c2 cell line, a subclone derived from embryonic rat heart that has morphological and biochemical characteristics of embryonic cardiac and skeletal myocytes (23,24). Cells transfected to express ␤-galactosidase (␤-gal) or green fluorescent protein (GFP) were used as controls.

EXPERIMENTAL PROCEDURES
Isolation of Pigeon cN-I and Human cN-II cDNA-The full-length cDNA clone coding for pigeon cN-I has already been described (9). The human cN-II cDNA clone was obtained by polymerase chain reaction from total placental cDNA using oligonucleotide primers designed from the published sequence (11) and a proofreading thermostable DNA polymerase (BIO-X-ACT, Bioline, London, United Kingdom). The polymerase chain reaction product was subcloned into the TA cloning vector pTargeT (Promega, Southampton, United Kingdom). Several clones were sequenced in both directions using the dideoxy chain termination method and an ABI (Perkin-Elmer, Applied Biosystems Division) automated sequencer. The clone used for these studies contained an open reading frame coding for a 561-amino acid, 65-kDa protein with a single difference from the published sequence, D265G, but use of this clone did not affect any kinetic parameter measured (see below).
Expression of cN-I and cN-II cDNAs in COS-7 and H9c2 Cells-Cells were routinely cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mM glutamine, 100 g/ml streptomycin, 100 units/ml penicillin. Cells were transfected according to the manufacturer's instructions in 6-well plates with 10 l of Lipo  (25). Cells were then incubated for up to 10 min in fresh medium containing EHNA and 50 M 5Ј-amino-5Ј-deoxyadenosine to inhibit adenosine kinase (EC 2.7.1.20) (25). Metabolic inhibition was triggered by inhibition of glycolysis with 5 mM 2-deoxy-D-glucose (DOG) or with 10 mM DOG plus the mitochondrial uncoupler FCCP (1 M). FCCP did not inhibit the activity of AMP deaminase, cN-I, or cN-II measured in homogenates. Allopurinol (100 M) was added to inhibit xanthine oxidase. Cell incubations were terminated, and metabolite and protein concentrations were determined as described below. The cell viability was assessed by measurement of lactate dehydrogenase in cell supernatants at the end of the incubations (Roche Molecular Biochemicals cytotoxicity test) or by fluorescence microscopy after addition of propidium iodide (1 g/ml). Nonviable cells were expressed as a percentage of the total lactate dehydrogenase present in parallel wells extracted in KRH containing 0.1% Triton X-100 or total cell counts, respectively. Expression of ␤-gal was measured histochemically as described previously (26), as a measure of transfection efficiency.
Measurement of Nucleoside, Nucleotide, and Protein Concentrations-Addition of 0.75 volumes of 1.6 M HClO 4 was used to terminate reactions followed by neutralization of supernatants with 0.125 volumes of 3 M K 3 PO 4 . Nucleosides and nucleotides were measured by reverse phase high pressure liquid chromatography as described previously (27), but performed at pH 6.7 and using a modified gradient of B (15% acetonitrile in Buffer A containing 150 mM KCl and 150 mM KH 2 PO 4 ) changed linearly as follows: 0 min, 0% B; 0.1 min, 3% B; 3 min, 9% B; 7.5 min, 100% B; 8.5 min, 0% solution B for 3.5 min. Protein concentrations were measured in the HClO 4 pellets resuspended in 1 M NaOH using a micro-BCA assay (Pierce).

Characterization of Recombinant cN-I and -II in Cell
Extracts-To measure cN-I activity, control-and cN-I-transfected cells were extracted with a buffer containing 40 mM N-dimethylglutarate, 40 mM ␤-glycerophosphate, 200 mM KCl, 0.2 mM dithiothreitol, 0.2% Triton X-100, pH 6.9. Aliquots of extract were incubated for 0 and 5 min at 37°C with an equal volume of the substrate mix to yield final concentrations of 2 M pentostatin, 10 M 5Ј-deoxy-5Ј-aminoadenosine, 50 M ␣, ␤-methylene-ADP, and 1 mM free MgCl 2 with 4.8 mM ATP, 2.8 mM ADP, 0.66 mM AMP or 4.8 mM ATP, 4.9 mM ADP, 2 mM AMP (28). The Hanes-Woolf equation was used to linearize the data, and the slope allowed the calculation of V max . To measure cN-II activity, cells were extracted with a buffer containing 10 mM TES, 120 mM KCl, 1 mM EDTA, 0.1 mM dithiothreitol, 0.1% Triton X-100, pH 6.5. Aliquots of cell extract were incubated for 0 and 5 min at 37°C with an equal volume of the substrate mix to yield final concentrations of 70 mM Na-␤glycerophosphate (pH 6.5), 35 mM MgCl 2 , 5 mM ATP, and 2 mM IMP to measure V max or a with a range of IMP concentrations for K m determinations. The effect of phosphate (at 1 mM ATP), ATP, and FCCP were investigated at 0.25 mM IMP, and the substrate preference was investigated at 2 mM AMP or GMP. Adenosine and inosine concentrations were measured by high pressure liquid chromatography and protein concentration by BCA assay, as described above.
Statistical Methods-Each experiment consisted of at least duplicate observations. Averages for at least three independent experiments are expressed as mean Ϯ S.E. When only two experiments were carried out, the range is given. Comparisons are made by Student's t test using paired data if available.

RESULTS
Characterization of the cN-I and cN-II cDNA Clones-As described previously (9), the pigeon heart cN-I expressed in COS-7 cells had characteristics similar to the enzyme purified from pigeon, rat, rabbit, and human hearts. It had a preference for AMP over IMP, a K m for AMP of approximately 3 mM. Human recombinant cN-II expressed in COS-7 cells had a K m value measured in the presence of a maximum activating concentration of 5 mM ATP of 0.2 and 0.06 mM IMP. It preferred GMP to IMP to AMP (1.33:1:0.33 or 1.33:1:0.27 in two separate experiments). Removal of ATP from the assay reduced activity by 68 and 64%, whereas addition of 5 mM phosphate inhibited activity in the presence of 1 mM ATP by 98 and 99%. All of these properties are consistent with those of the purified enzymes from humans and other species (19,29).
ATP Catabolism and Adenosine Formation in COS-7 Cells Overexpressing cN-I or cN-II-We diluted the cN-I and cN-II plasmids with control pSV␤-gal plasmid in order to achieve similar levels of 5Ј-nucleotidase activity in transfected cells (Table I). The specific activity of the enzymes was compared with that in control cells that had been transfected to express ␤-gal alone. Homogenate cN-I activity was increased 7-fold, and cN-II activity was increased 10-fold, respectively (Table I). Between 24 and 30% of the cells were shown to express ␤-gal by histochemical staining.
We measured the breakdown of ATP to ADP, AMP, and its further metabolites adenosine, inosine, and hypoxanthine (Fig.  1). Incubation of the cells with inhibitors of adenosine deaminase and adenosine kinase allowed the measurement of the rates of adenosine formation without further metabolism back to AMP or to inosine and hypoxanthine. To control for the effectiveness of inhibition, we measured metabolism of 2 M exogenous adenosine. With inhibitors present, 105 Ϯ 6% was Divergent Roles of Cytosolic 5Ј-Nucleotidases recovered after a 10 min incubation at 37°C (n ϭ 3). In the experiments with metabolic inhibitors (see below), inosine and hypoxanthine did not arise by deamination of adenosine but from metabolism of AMP to IMP by AMP deaminase followed by catabolism of IMP (Fig. 1). COS-7 cells did not contain detectable xanthine oxidase (results not shown), and hence, allopurinol was not used to inhibit further metabolism of hypoxanthine to uric acid. We investigated ATP breakdown and adenosine formation under two conditions of metabolic inhibition. The amount of lactate dehydrogenase present in the supernatants remained unaffected and below 5% under all conditions tested. Treatment of control and cN-II-expressing COS-7 cells with 10 mM DOG ϩ 1 M FCCP for 10 min caused 91 Ϯ 1 and 91 Ϯ 1% falls in the ATP concentration after 10 min, respectively (Fig. 2a). After 10 min, 74 Ϯ 1 and 74 Ϯ 1%, respectively, of the fall in ATP concentration was accounted for by increased AMP and 8.6 Ϯ 1.7 and 9.3 Ϯ 2.1% by adenosine. Although cN-I-transfected cells showed a similar 91 Ϯ 2% fall in ATP concentration (Fig. 2a), only 48 Ϯ 2% was accounted for by increased AMP (p Ͻ 0.01 versus ␤-gal or cN-II), whereas 31 Ϯ 3% was accounted for by increased adenosine (p Ͻ 0.02 versus ␤-gal or cN-II). Measurements of absolute adenosine concentration (Fig. 3a) show that adenosine accumulation in cN-I-transfected cells was nonlinear but significantly greater than in ␤-galor cN-II-transfected cells at 5 and 10 min. There was no difference in adenosine accumulation between ␤-galand cN-II-transfected cells (Fig. 3a). The concentration of IMP (the preferred substrate of cN-II) remained below 0.25 nmol⅐mg of protein Ϫ1 under all conditions of incubation and did not accumulate during ATP breakdown (Fig. 2a). This demonstrates that the activity of adenosine deaminase was low under these conditions. Indeed inosine and hypoxanthine, the ultimate products of AMP deamination (Fig. 1), did not accumulate either under these conditions (Fig. 3b). These data indicate that even when the competing substrate, IMP, remained at very low concen- trations, only cN-I and not cN-II overexpression increased adenosine production.
AMP deaminase was active under these conditions, and AMP breakdown proceeded mainly to inosine and hypoxanthine (Figs. 2b and 3d). Inosine and hypoxanthine accumulated to a similar extent in ␤-galand cN-II-transfected cells (Fig. 3d). As expected, cN-I overexpression caused a converse decrease in the accumulation of inosine and hypoxanthine because catabolism was diverted to adenosine (Figs. 2b and 3d). The concentration of IMP remained below 0.4 nmol⅐mg of protein Ϫ1 in all transfected cells under these conditions. Hence, even under these conditions, the activity of AMP deaminase was ratelimiting, and the endogenous activity of 5Ј-nucleotidase in all cells was therefore sufficient for production of inosine, and hypoxanthine. It was therefore impossible to measure the effect of cN-I or cN-II on catabolism of IMP in COS-7 cells.
ATP Catabolism and Adenosine Formation in H9c2 Cells Overexpressing cN-I or cN-II-In order to investigate the effect of cN-I or cN-II overexpression on IMP breakdown, it was necessary to find a cell line in which AMP deaminase was not rate-limiting. Skeletal muscle contains the highest concentration of AMP deaminase (21). We therefore chose to reexamine the effect of cN-I and cN-II overexpression in H9c2 cells, which have a mixed heart/skeletal muscle phenotype.
H9c2 cells were cotransfected with a plasmid coding for enhanced GFP, together with one for cN-I or cN-II, so as to express similar activities of 5Ј-nucleotidase (Table I) or with GFP alone, as control. The proportion of cells transfected, as determined by GFP fluorescence, was 20 -25%. Activities of cN-I and cN-II were some 5-6-fold lower than in COS-7 cells (Table I). ATP concentrations were similar in GFP-, cN-I-, and cN-II-transfected cells (103 Ϯ 4, 94 Ϯ 4, and 101 Ϯ 9 nmol⅐mg of protein Ϫ1 , respectively) in the absence of metabolic inhibitors. ATP breakdown was induced either by DOG ϩ FCCP or by DOG alone, and the same metabolites were measured as for COS-7 cells. The concentration of guanosine was also measured, as it is a potential product of 5Ј-nucleotidase activity. Cell viability by release of lactate dehydrogenase and propidium iodide staining was above 95% under all conditions used.
The purine metabolite concentrations are summarized in Fig. 4 and Table II. The major difference between H9c2 and COS-7 cells was that IMP clearly accumulated during ATP breakdown, especially with DOG plus FCCP, indicating that AMP deaminase activity was not rate-limiting (Fig. 4). ATP breakdown was similar in GFP-, cN-I-, and cN-II-transfected cells. Adenosine accumulation occurred in cN-I-overexpressing cells even in the absence of metabolic poisons (Table II), as previously shown in COS-7 cells (9). Under both conditions of ATP breakdown, cN-I overexpression significantly accelerated adenosine formation, as demonstrated above for COS-7 cells. However, cN-II overexpression did not increase adenosine formation under either condition of ATP breakdown (Table II). Overexpression of cN-II increased inosine and hypoxanthine accumulation in the presence of 5 mM DOG (Table II). The concentration of guanosine was also significantly increased (Table II). Both measures clearly indicate that cN-II is active under these conditions and is capable of IMP and GMP catabolism. By contrast, overexpression of cN-I did not increase inosine, hypoxanthine, or guanosine concentrations relative to GFP-transfected cells, demonstrating that it does not contribute to the metabolism of IMP under these conditions.
Overexpression of cN-II did not increase inosine, hypoxanthine, or guanosine concentrations during severe ATP depletion with DOG plus FCCP, which, together with the greater accumulation of IMP (Fig. 4), indicates that cN-II is inactive under these conditions. Overexpression of cN-I was also without effect on IMP and GMP metabolites under these conditions. DISCUSSION Kinetic, molecular, and inhibitor studies have previously implicated cN-I in adenosine formation (9,15,16,18). However, the role of cN-II in adenosine formation has remained unclear. Ours is the first study to compare directly the roles of cN-I and cN-II in AMP catabolism to adenosine in the same cell type during ATP breakdown. The major conclusion is that only cN-I and not cN-II catalyzes significant formation of adenosine under rapid or more progressive conditions of ATP depletion. This conclusion applied equally to two cell types of widely different origin, namely COS-7 or H9c2 cells. However, the lack of AMP catalysis through cN-II is all the more remarkable in COS-7 cells, given that IMP concentrations remained very low, owing to the relatively low activity of AMP deaminase, even when ATP levels were less severely depleted. Hence competition between IMP and AMP for metabolism by cN-II could not explain its inability to produce adenosine.
Our studies could not clearly establish whether cN-I or cN-II is capable of IMP catabolism in COS-7 cells, because AMP deaminase was rate-limiting under all the conditions investigated. We therefore addressed this issue in H9c2 cells, in which IMP was shown to accumulate during ATP breakdown and hence 5Ј-nucleotidase activity rather than AMP deaminase was rate-limiting for the pathway. In H9c2 cells, it was clear that cN-II could contribute to IMP catabolism under conditions of progressive ATP breakdown but not when ATP pools were rapidly depleted. These observations are consistent with previous in vitro and in vivo data (30,31) showing that cN-II activity is biphasically related to energy charge. Activity of cN-II is increased during moderate falls in energy charge but strongly inhibited by profound falls. The failure of cN-I to catalyze IMP breakdown is consistent with its preference for AMP in vitro. The kinetic properties of the enzymes from pigeon, rabbit, rat, and human are remarkably similar (15,16,32). It is therefore unlikely that our results arise from the use of a pigeon enzyme in mammalian cells. Putative human (Gen-Bank TM accession number AA446194) and mouse (GenBank TM accession number AI429147) homologues exist for pigeon cN-I and show considerable protein sequence homology.
Our present results further support a physiological role for cN-I in adenosine formation, which opens up the possibility of varying adenosine formation rates by manipulating the enzyme activity. Increasing cN-I activity would be expected to promote a variety of cytoprotective actions.
In contrast to cN-I, the role of cN-II appears to be restricted to IMP rather than AMP catabolism. IMP can arise by action of AMP deaminase on AMP, during ATP breakdown, but is also the product of de novo purine synthesis, under conditions of abundant ATP. Activity of cN-II may therefore be a key step controlling the balance of purine accumulation within cells and hence the upper limit of the purine nucleotide pool (33). Interestingly, work from Bianchi and co-workers (14) shows that stable cell lines overexpressing cN-II have somewhat lower total ATP and other purine trinucleotide levels. It would be interesting, in future studies, to examine the responses of these cell lines to mild and severe metabolic stresses. In the present experiments, we did not observe any effects of cN-I or cN-II overexpression on baseline ATP levels in COS-7 or H9c2. However, transient transfection occurred in only 20 -30% of cells, which may confound these measurements. Overexpression of cN-I at 10-fold higher levels than used here did produce a decrease in baseline ATP concentration in COS-7 cells (9). In conclusion, our present results clearly demonstrate that cN-I and cN-II play distinct roles in AMP and IMP catabolism, respectively.