Pathways of purine metabolism in human adipocytes. Further evidence against a role of adenosine as an endogenous regulator of human fat cell function.

Previous results demonstrated that the adenosine that accumulates in human fat cell suspensions is derived from extracellular sources (Kather, H. (1988) J. Biol. Chem. 263, 8803-8809). To get insight into the mechanisms responsible for the lack of adenosine release, extracellular adenine nucleotide catabolism was minimized by 10 mmol/liter beta-glycerophosphate and 10 mumol/liter alpha,beta-methyleneadenosine 5'-diphosphate. Intracellular adenine nucleotide catabolism resulted in a release of inosine and hypoxanthine under these conditions that was increased markedly by isoproterenol. Experiments with inhibitors of adenosine deaminase and adenosine kinase indicated that the production of inosine and hypoxanthine proceeded via AMP deamination. Consistently, IMP levels were increased transiently in the presence of isoproterenol. In addition, the cells possessed a nucleotide phosphomonoesterase that was resistant to the inhibitory actions of ATP and alpha,beta-methyleneadenosine 5'-diphosphate and showed preference for IMP over AMP. Adenosine (approximately 1 nmol/10(6) cells/h) was also produced inside the cells. However, adenosine production was unrelated to ATP turnover via adenylate cyclase, and any adenosine formed was immediately reconverted to adenine nucleotides in the absence and presence of isoproterenol. It was concluded that adenosine is not released by intact human adipocytes, because the alternative routes of intracellular AMP catabolism are compartmentalized (at least in functional terms), and adenosine kinase is not saturated with substrate in the absence and presence of isoproterenol.

To get insight into the mechanisms responsible for the lack of adenosine release, extracellular adenine nucleotide catabolism was minimized by 10 mmol/liter @-glycerophosphate and 10 pmol/liter cr,&methyleneadenosine 5'-diphosphate. Intracellular adenine nucleotide catabolism resulted in a release of inosine and hypoxanthine under these conditions that was increased markedly by isoproterenol. Experiments with inhibitors of adenosine deaminase and adenosine kinase indicated that the production of inosine and hypoxanthine proceeded via AMP deamination.
Consistently, IMP levels were increased transiently in the presence of isoproterenol.
In addition, the cells possessed a nucleotide phosphomonoesterase that was resistant to the inhibitory actions of ATP and c&methyleneadenosine 5'-diphosphate and showed preference for IMP over AMP. Adenosine (-1 nmol/106 cells/h) was also produced inside the cells. However, adenosine production was unrelated to ATP turnover via adenylate cyclase, and any adenosine formed was immediately reconverted to adenine nucleotides in the absence and presence of isoproterenol.
It was concluded that adenosine is not released by intact human adipocytes, because the alternative routes of intracellular AMP catabolism are compartmentalized (at least in functional terms), and adenosine kinase is not saturated with substrate in the absence and presence of isoproterenol.
Adenosine is a degradation product of adenine nucleotides postulated to act as an endogenous (feedback)-inhibitor of hormone-activated lipolysis (l-3). Its potent antilipolytic and cyclic AMP-lowering effects are well known and the processes mediating these events have been defined (l-7). The nucleoside is inevitably present in fat cell suspensions and is released by intact canine adipose tissue during prolonged sympathetic stimulation (l-7). However, the sources of this adenosine and the factors controlling its availability have remained unclear. It has been proposed that the nucleoside is produced inside the adipocytes from cyclic AMP via the sequence cyclic *This work was supported by grants from the Deutsche Forschungsgemeinschaft and Grant Ftirderungskennzeichen 0704721 AZ from the Bundesministerium fiir Forschung und Technologie, Bonn, West Germany. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

AMP+AMP+adenosine
(l-3). Once formed, adenosine may be deaminated, released into the pericellular space, or reconverted to adenine nucleotides by adenosine kinase (Fig. 1). Adenosine kinase has a low K, for adenosine (1-5 pmol/liter), whereas that for the deaminase is perhaps SO-fold higher (1). If adenosine kinase was saturated with substrate under basal conditions, a hormone-induced increase in cyclic AMP turnover could therefore result in intracellular adenosine accumulation and release.
A major problem in testing this hypothesis resides in the fact that adenosine can be produced via intra-and extracellular pathways (Fig. 1). Extracellular ATP catabolism proceeds exclusively via AMP dephosphorylation resulting in adenosine as the end product (8)(9)(10)(11)(12)(13)(14). In contrast, intracellular ATP catabolism appears to be mainly via deamination at the nucleotide level (AMP-*IMP+inosine).
Under physiological conditions, any adenosine formed inside the cells is immediately reconverted to AMP in virtually all cell types that have been studied in detail, e.g. hepatocytes (9), leukocytes (lo), lymphoblasts (ll), erythrocytes (12), cardiomyocytes (13), and probably also vascular endothelia (14). Consistently, human adipocytes released inosine and hypoxanthine, and their export was markedly increased by isoproterenol (8). By contrast, adenosine appeared to be derived from extracellular sources, and its concentration was not influenced by @-adrenergic catecholamines (8). The current studies are detailed investigations into the pathways of purine metabolism in human adipocytes. As purine metabolism can proceed via intracellular and extracellular pathways and involves cyclic routes, analytical data alone are not sufficient to indicate the sources of individual purines (Fig. 1). Therefore, a pharmacological approach was used. The properties of the inhibitors used are listed in Table  I. EXPERIMENTAL PROCEDURES

Subjects
Subcutaneous adipose tissue was from surgical subjects undergoing elective abdominal or cosmetic breast surgery. Operations were performed after an overnight fast. Anesthesia was initiated with a shortacting barbiturate and maintained with oxygen, nitrous oxide, and halothane. The tissue specimens were obtained at the start of the operations.  selective inhibitor of ecto-5'-nucleotidase' ( Table I). None of the inhibitors used had a detectable influence on cellular adenine nucleotide contents or lactate dehydrogenase release (not shown). ATP levels dropped by approximately one-third in the course of incubations (Table II). In contrast to conventional incubations, the decrease in ATP was counterbalanced by a corresponding accumulation of ADP and AMP in the media upon inhibition of ectophosphatase activities. The medium concentrations of these latter nucleotides were related to lactate dehydrogenase release, indicating that the extracellular catabolism of adenine nucleotides had in fact been blocked effectively at the level of AMP under the conditions used (Table II).
i Despite its specificity as an inhibitor of ecto-5'-nucleotidase, a,@methyleneadenosine 5'-diphosphate should not be used uncritically, as at higher concentrations than those required to inhibit the ectoenzyme (230 rmol/liter), it can also suppress cyclic AMP levels in the presence of isoproterenol. Unlike the cyclic AMP-lowering action of 5'-deoxy-5-iodotubercidin, the latter effect was not reversed by adenosine deaminase.
The issue whether this inhibitory action is mediated via Ai-adenosine receptors (7), Pp-purinergic receptors (38), or other mechanisms remains to be clarified.
Inhibition of ectophosphatase activities was specific, in as much, as cY,p-methyleneadenosine 5'diphosphate and p-glycerophosphate failed to suppress the activity of a distinct nucleotide phosphomonoesterase recovered in fat cell lysates (Fig. 2). Similar to the soluble 5'-nucleotidases from chicken and rat liver (29, 30), the latter enzyme displayed marked preference for IMP over AMP. With half-saturation occurring at -0.3 mmoljiter (IMP) and -3 mmol/liter (AMP), its substrate affinity was low, however, in comparison to the ectoenzyme exhibiting an apparent K, value for AMP of -5 rmol/liter, and this explains why the newly discovered enzyme did not contribute measurably to extracellular adenine nucleotide catabolism, even though one-third of the cells were damaged in the course of incubations (Table II). Fig. 3 shows the effects of 2'-deoxycoformycin and/or 5'deoxy-5-iodotubercidin on basal and hormone-activated rates of lipolysis and cyclic AMP levels in the presence of isoproterenol. A treatment of the cells with 2'-deoxycoformycin had no effect on lipolysis and cyclic AMP accumulation (Fig. 3, A  and B). By contrast, 5'-deoxy-5-iodotubercidin caused a marked inhibition of nonstimulated glycerol release (Fig. 3A 1 markedly reduced, e.g. from 6.2 + 1.3 nmol/106 cells (controls) and 5.1 f 0.7 nmol/106 cells (2'-deoxycoformycin-treated cells) to 2.2 + 0.7 and 2.3 f 0.7 nmol/106 cells, respectively. However, the lipolytic effect of the P-adrenergic agonist was unimpaired, consistent with the frequent notion that the increases in cyclic AMP caused by activating hormones may be orders of magnitude higher than those required for a maximal stimulation of lipolysis (31). The inhibitory effects of 5'-deoxy-5-iodotubercidin could be overcome by adenosine deaminase (0.5 unit/ml), indicating that its antilipolytic and cyclic AMP-lowering effects were caused by the adenosine that accumulated in the media upon inhibition of adenosine kinase (see below), rather than reflecting an agonist activity at A,-adenosine receptors (Fig. 3C, Table I).
identical over a wide range of adenosine concentrations (30 nmol/liter-I pmol/liter) and were not influenced by isoproterenol (not shown), suggesting that adenosine transport was slow relative to intracellular metabolism. At low concentrations (30-100 nmol/liter), exogenously supplied adenosine had no effect on inosine and hypoxanthine production. Even at the highest concentration used (1 pmol/liter), two-thirds of the adenosine that had been removed from the media were incorporated into adenine nucleotides, whereas only one-third was converted to inosine and hypoxanthine (Fig. 4, A and C; Table III). In controls, adenine nucleotide concentrations displayed a steady increase up to 2.5 h of incubation in the presence of 1 pmol/liter of added adenosine and then declined, indicating that the increase in adenine nucleotide synthesis was in part counterbalanced by an enhancement of adenine nucleotide catabolism that became predominant at the end of incubations (Fig. 40). 2'-Deoxycoformycin had no perceptible effect on the conversion of exogenously supplied adenosine to adenine nucleotides, as expected (Fig. 40). It was surprising, however, that the production of inosine and hypoxanthine was reduced by only 30% in 2'-deoxycoformycin-treated cells (Fig. 4C, Table  III). Consistently, adenosine uptake was decreased only marginally by the inhibitor of adenosine deaminase (Fig. 4A). Adenosine Uptake- Fig.  4 illustrates the effects of 2'-deox-5'-Deoxy-5-iodotubercidin prevented the conversion of exycoformycin and/or 5'-deoxy-5-iodotubercidin on the removal ogenously supplied adenosine to adenine nucleotides (Fig. and metabolic fate of exogenously supplied adenosine. More 40). As adenosine deaminase displays low substrate affinity than 80% of the adenosine that had disappeared from the in comparison to the kinase, the formation of inosine and media could be accounted for by a formation of inosine, hypoxanthine was increased only moderately. Therefore, hypoxanthine, and adenine nucleotides in all experimental adenosine removal was decreased by 30% upon inhibition of settings (Table III). Adenosine uptake displayed pseudo firstadenosine kinase (Fig. 4A, Table III). order kinetics (Fig. 4A). The fractional rates of removal were The metabolism of exogenously supplied adenosine was  Values are those obtained in the experiments shown in Fig. 4 after 2.5 h of incubation (when adenine nucleotide concentrations peaked). Adenosine uptake refers to the amount of adenosine that had disappeared from the media. The net synthesis of inosine and hypoxanthine was calculated by subtracting the spontaneous release of these latter purines from the total amount produced in the presence of exogenously supplied adenosine. The flux through adenosine deaminase was calculated from the difference in inosine and hypoxanthine release between untreated and 2'-deoxycoformycin-treated cells. The rate of phosphorylation was calculated from the difference between adenosine uptake and deamination.
completely blocked by the combined action of 2'-deoxycoformycin and 5'-deoxy-5-iodotubercidin (Fig. 4, A and D). The latter finding demonstrated that adenosine deaminase and adenosine kinase are the sole routes of intracellular metabolism of exogenously supplied adenosine and implied that both enzymes were virtually completely blocked in situ at the inhibitor concentrations used. As a blockade of adenosine deaminase alone resulted in only 30% inhibition of inosine and hypoxanthine formation, the latter observation also indicated that two-thirds of the inosine and hypoxanthine accumulating in the media in the presence of 1 Fmol/liter of exogenously supplied adenosine had been produced via a pathway involving adenosine kinase (adenosine+AMP+ IMP-Gnosine), consistent with the notion that not only the synthesis, but also the breakdown of adenine nucleotides was increased under these latter conditions (Fig. 40, Table III).

Purine
Release--Summarized in Fig. 5 are the effects of 2'deoxycoformycin and/or 5'-deoxy-5-iodotubercidin on purine release in the absence and presence of isoproterenol.
As reported previously (8), adenosine concentrations were beyond detectable levels (~5 nmol/liter) in the presence of ectophosphatase inhibitors, regardless of whether ATP turnover via adenylate cyclase was increased by isoproterenol or not (Fig. 5A). By contrast, the cells released inosine and hypoxanthine (Fig. 5, B and C). In controls, their concentrations were 2.3 + 0.8 nmol/106 cells (inosine) and 0.75 + 0.2 nmol/106 cells (hypoxanthine) after 3 h of incubation. Isoproterenol caused a 3.5-fold increase in inosine output that was paralleled by a corresponding enhancement of hypoxanthine release (Fig. 5, B and C). 2'-Deoxycoformycin had no effect on purine output or the distribution of individual purines in the absence and presence of isoproterenol, indicating that the purines accumulating in the media had been produced via deamination at the nucleotide level (AMP+IMP+inosine). By contrast, an inhibition of adenosine kinase resulted in the appearance of measurable amounts of adenosine in the media. The rates of adenosine accumulation were linear with time and not influenced by isoproterenol (Fig. 5A). On the average, 1 nmol of adenosine was released by lo6 cells within 3 h in the presence of 5'-deoxy-5-iodotubercidin (Fig. 5A). In the absence of isoproterenol, the release of inosine and hypoxanthine was also increased (by 1.6 nmol/106 cells; Fig. 5, B and C). The latter effect could be reversed by 2'-deoxycoformycin, indicating that a major fraction of the adenosine that accumulated upon inhibition of adenosine kinase was deaminated.
Consistently, adenosine release was increased to 3 nmol/106 cells/3 h when its further metabolism was completely blocked by the combined action of 2'-deoxycoformycin The experiments (n = 6) were conducted as time course studies, and the values + S.E. obtained after 3 h of incubation in the absence (black bars) or presence (open bars) of 1 pmollliter isoproterenol are given. The term purines refers to the sum of all purines recovered in the media (adenosine + inosine + hypoxanthine). For other experimental details and abbreviations see legends to Figs. 3 and 4. and 5'-deoxy-5-iodotubercidin (Fig. 5A). The release of adenosine was reduced 60% by dipyridamole, an inhibitor of nucleoside transport, under these latter conditions, indicating that it was mainly derived from intracellular sources ( Table  IV).
The stimulatory effect of isoproterenol on inosine and hypoxanthine release was reduced by 5'-deoxy-5-iodotubercidin, as expected because cyclic AMP levels were markedly suppressed by the adenosine that accumulated in the media under these conditions (Figs. 3 and 5, A-D). The decrease in Purine Metabolism in Human Adipocytes inosine and hypoxanthine release was moderate, however, and became only apparent in 2'-deoxycoformycin-treated cells. In untreated cells the 5'-deoxy-5-iodotubercidin-induced inhibition was probably masked, because inosine and hypoxanthine were not only produced from IMP, but also from aden-101 of 10 experiments (Fig. 7). Purine release became progressively slower within 20-30 min in the presence of isoproterenol. Concomitantly, IMP concentrations returned to control levels, indicating that the flux through adenylate deaminase was in fact transiently increased by the P-adrenergic agonist, consistent with the conclusions drawn from the inhibitor studies. Isoproterenol had no detectable influence on intracellular AMP. However the amount of AMP accumulating in the media was large in relation to the intracellular concentrations of the nucleotide. Therefore, minor increases in intracellular AMP cannot be excluded. osine under these latter conditions. Isoproterenol-activated rates of inosine and hypoxanthine release were markedly nonlinear with time (Fig. 6). Approximately one-third of the cyclic AMP present in human fat cell suspensions was released into the media in the course of incubations, a figure that closely corresponded to lactate dehydrogenase release, indicating that not only adenine nucleotides, but also the majority of cyclic AMP was released via nonspecific mechanisms (Fig. 6B, Table II). A comparison of the time courses of cyclic AMP accumulation inside the cells with purine production during each of the 30-min intervals revealed that the rates of net production of inosine and hypoxanthine paralleled the changes with time of intracellular cyclic AMP, suggesting that the stimulatory effect of isoproterenol on the release of these latter purines was related to its cyclic AMP-elevating property (Fig. 6, B and C), which is in fact the case (32). Accordingly, the net production of inosine and hypoxanthine displayed only a small initial rise in the absence of the fl-adrenergic agonist. The linear phase of purine release was accompanied by a transient 2-to &fold increase in cellular IMP contents that was maximal within 6-10 min and could be detected in 8 out Values are mean of duplicate determinations and refer to intracellular cyclic AMP concentrations representing the difference between cyclic AMP contents of media and suspensions (m), IMP levels (O), and release of inosine and hypoxanthine (0).

DISCUSSION
In the current studies, the pathways of adenine nucleotide catabolism and the metabolic fate of adenosine were analyzed in human adipocytes by means of specific inhibitors of purine production and metabolism. The potential pitfalls inherent in this approach were considered carefully, and the results of various control experiments are consistent with the view that the known side effects of these latter compounds played a minor role, if any, under the conditions used (Fig. 3, Table  II).' The observation that the extracellular concentrations of adenine nucleotides (and cyclic AMP) were related to lactate dehydrogenase release in the presence of ectophosphate inhibitors, whereas adenosine concentrations dropped beyond detectable levels (~5 nmol/liter), provides direct evidence in support of the previous conclusion that the production of adenosine, occurring extracellularly, reflects the catabolism of adenine nucleotides from broken cells, and confirms that inosine and hypoxanthine are the sole products of adenine nucleotide breakdown that are released by intact human adipocytes in absence and presence of isoproterenol (8). 2'-Deoxycoformycin had no influence on basal or isoproterenol-stimulated rates of inosine and hypoxanthine accumulation, even though adenosine deaminase was effectively blocked in situ (Fig. 4). The latter finding is consistent with the results obtained in other cell types under physiological conditions (9)(10)(11)(12)(13)(14) and indicates that the formation of inosine and hypoxanthine proceeded via deamination at the nucleotide level (AMP+IMP+inosine).
IMP levels were consistently increased transiently in the presence of isoproterenol (Fig. 7).
A release of adenosine could only be observed upon inhibition of adenosine kinase, indicating that some adenosine was produced and was immediately recycled into adenine nucleotides (Fig. 5). Along with the observation that exogenously supplied adenosine was mostly phosphorylated, that latter finding indicated that adenosine kinase constituted the primary route of adenosine metabolism, and was not saturated with substrate irrespective of whether ATP turnover via adenylate cyclase was increased by isoproterenol or not. Indeed, phosphorylation was so dominant over deamination that the conversion of exogenously supplied adenosine to inosine and hypoxanthine also took place mainly via a circuitous route bypassing adenosine deaminase, e.g. adenosine+AMP+ IMP+inosine, consistent with the conclusions drawn from similar observation in rat hepatocytes (9).
Unexpectedly, isoproterenol had no effect on adenosine production, although the output of inosine and hypoxanthine was increased markedly by the P-adrenergic agonist. The molecular mechanisms responsible for the apparent compartmentation of the alternative routes of AMP catabolism are presently unclear. The simplest solution to the problem would be that the accumulation of adenosine observed in the absence of its further metabolism reflected a residual breakdown of