Ectonucleotidases in the kidney

Abstract

Members of all four families of ectonucleotidases, namely ectonucleoside triphosphate diphosphohydrolases (NTPDases), ectonucleotide pyrophosphatase/phosphodiesterases (NPPs), ecto-5′-nucleotidase and alkaline phosphatases, have been identified in the renal vasculature and/or tubular structures. In rats and mice, NTPDase1, which hydrolyses ATP through to AMP, is prominent throughout most of the renal vasculature and is also present in the thin ascending limb of Henle and medullary collecting duct. NTPDase2 and NTPDase3, which both prefer ATP over ADP as a substrate, are found in most nephron segments beyond the proximal tubule. NPPs catalyse not only the hydrolysis of ATP and ADP, but also of diadenosine polyphosphates. NPP1 has been identified in proximal and distal tubules of the mouse, while NPP3 is expressed in the rat glomerulus and pars recta, but not in more distal segments. Ecto-5′-nucleotidase, which catalyses the conversion of AMP to adenosine, is found in apical membranes of rat proximal convoluted tubule and intercalated cells of the distal nephron, as well as in the peritubular space. Finally, an alkaline phosphatase, which can theoretically catalyse the entire hydrolysis chain from nucleoside triphosphate to nucleoside, has been identified in apical membranes of rat proximal tubules; however, this enzyme exhibits relatively high K_m values for adenine nucleotides. Although information on renal ectonucleotidases is still incomplete, the enzymes’ varied distribution in the vasculature and along the nephron suggests that they can profoundly influence purinoceptor activity through the hydrolysis, and generation, of agonists of the various purinoceptor subtypes. This review provides an update on renal ectonucleotidases and speculates on the functional significance of these enzymes in terms of glomerular and tubular physiology and pathophysiology.

Introduction

Adenosine 5′-triphosphate (ATP) can be released from renal epithelial cells across the apical membrane into the tubular lumen and also, to a lesser extent, across the basolateral membrane [1, 2]. As described elsewhere in this Special Issue, these nucleotides have the potential to activate renal P2 purinoceptors located along the nephron and thereby elicit a variety of autocrine/paracrine effects on tubular transport processes. Aside from their cellular distribution and expression levels, the extent of activation of the various renal P2 receptor subtypes by extracellular nucleotides will depend on the rate of nucleotide release, on the rate of generation of nucleotides by phosphorylating enzymes, and on how rapidly the nucleotides are hydrolyzed by nucleotidases.

The phosphorylating enzymes that can synthesise nucleotides include nucleoside diphosphate kinases (EC 2.7.4.6) and adenylate kinases (EC 2.7.4.3) [3, 4]. The former catalyse the transfer of the terminal phosphate of nucleoside 5′-triphosphates to nucleoside 5′-diphosphates. For example, ATP can donate a phosphate to GDP to produce GTP; i.e., ATP + GDP ⇆ ADP + GTP. Although initially believed to be restricted to the cell cytosol, there is evidence that nucleoside diphosphate kinases are also present in the cell membrane, where they could potentially generate extracellular nucleoside triphosphates. Both mRNA and enzyme protein for nucleoside diphosphate kinases have been identified in rat kidney [5], but their distribution along the nephron is unknown. The other phosphorylating enzymes, adenylate kinases, will either catalyse the production of ADP from ATP and AMP or vice versa, depending on the concentrations of the respective nucleotides; thus: ATP + AMP ⇆ 2ADP. Adenylate kinase activity has been documented in the glomerulus, proximal convoluted tubule, thick ascending limb of Henle and distal tubule [6], though it was not possible to distinguish between intracellular and extracellular activity.

Ectonucleotidases have been detected in virtually all tissues, including kidney. They are a heterogeneous group of enzymes with differing, though partly overlapping, catalytic properties (Table 1). Four families exist: the ectonucleoside triphosphate diphosphohydrolase (E-NTPDase) family, the ectonucleotide pyrophosphatase/phosphodiesterase (E-NPP) family, ecto-5′-nucleotidase and alkaline phosphatases. Their activities are dependent on the presence of divalent cations (Ca²⁺, Mg²⁺ or Zn²⁺). Ectonucleotidases located in the vasculature and in specific segments of the nephron will have a profound influence on the stimulation of purinoceptors, not only because the availability of nucleotide agonists is controlled by their hydrolysis but also because the generation of different nucleotides (e.g. ADP) may preferentially target different P2 receptor subtypes, while the nucleoside derivative adenosine can either target P1 (adenosine) receptors or re-enter cells via adenosine transporters (Fig. 1). In this short review, we will examine the characteristics of the four families of ectonucleotidases, describe current knowledge about their intrarenal distribution, and speculate on their functional significance.

Table 1.

Hydrolysis pathways of the ectonucleotidases expressed in the kidney

Major pathways are shown in black, others in grey. N.B. For most of these enzymes, nucleotides derived from other bases (UTP, GTP, TTP, CTP, UDP, GDP, TDP and CDP) can also act as substrates

Fig. 1.

Potential effects of renal ectonucleotidases and consequences for activation of purinoceptor subtypes. The major enzymes involved in each degradative pathway are shown in bold print; for more detail, see text. The information given in the figure indicates the relative potencies of ATP and ADP with respect to P2X and P2Y receptor subtypes. At sufficiently high concentrations, ATP can activate all P2 receptors other than P2Y₆ and P2Y₁₄. It is important to note that nucleotides derived from other bases are also hydrolysed/synthesised by these enzymes, but have been omitted for clarity. Uracil-based nucleotides are particularly significant: UTP is a potent agonist of P2Y₂ and P2Y₄ subtypes, and its dinucleotide derivative UDP is the major naturally occurring agonist of the P2Y₆ subtype. The mechanism(s) of ATP exit from renal cells has/have not been defined (see article by Leipziger [7] in this Special Issue). NDP, nucleoside diphosphate; NTP, nucleoside triphosphate

Ectonucleoside triphosphate diphosphohydrolases (NTPDases; EC 3.6.1.5)

The E-NTPDase family comprises eight members. Four of these, NTPDases 1, 2, 3 and 8, hydrolyse extracellular nucleotides. As NTPDases 4–7 are mainly intracellular enzymes, they will not be considered here. NTPDases 1–3 and 8 are anchored to the cell membrane via two transmembrane domains. These enzymes hydrolyse adenine-based nucleotides, as well as other nucleotides such as UTP, GTP, ITP, CTP and their respective nucleoside diphosphates, leading ultimately to the generation of the corresponding nucleoside monophosphates as final products.

NTPDase1 (CD39) hydrolyses ATP and ADP with almost equal preference [8]. Although ATP is dephosphorylated one phosphate at a time, the processing step is favoured, leading to the rapid generation of AMP with little ADP accumulation in the milieu [9–11]. Interestingly, and of possible physiological importance, the kinetics are different when UTP is used as substrate, as UDP accumulates transiently, being completely hydrolyzed only after the concentration of UTP has decreased [10]. NTPDase2 has a much greater preference for the hydrolysis of triphosphonucleosides (e.g., ATP) than the diphosphate derivatives (e.g., ADP), and therefore causes accumulation of the latter; NTPDase3 and NTPDase8 also have a preference for the hydrolysis of tri- over diphosphonucleosides, though in these cases the preference is less marked [8, 12]. Thus, NTPDases 3 and 8 generate only a transient accumulation of diphosphonucleosides [10, 13, 14].

Ectonucleotide pyrophosphatase/phosphodiesterases (NPPs; EC 3.1.4.1; EC 3.6.1.9)

The E-NPP family comprises seven members, but only NPPs 1–3 are able to hydrolyse nucleotides. NPP1 and NPP3 are anchored to the cell membrane by a single transmembrane domain, although these enzymes may be proteolytically cleaved and released in soluble form; whilst NPP2 exists only in secreted form [15].

NPPs 1–3 have a wide range of enzymatic activity. They are able to catalyse the hydrolysis of ATP and ADP to AMP, but, in contrast to NTPDases, they always release the monophosphate derivative (here AMP) along with the remaining moiety (pyrophosphate, when using ATP as a substrate). Importantly, NPPs can also hydrolyse diadenosine polyphosphates (again to AMP) and other dinucleoside polyphosphates, as well as UDP-glucose (an agonist of P2Y₁₄ receptors). NPP1 and NPP3 hydrolyze nucleotides more effectively than does NPP2 [16], the latter preferring phospholipids as substrates (as do NPPs 4-7). It is noteworthy that NPPs prefer alkaline pHs for activity. In our hands, recombinant human NPP1 and NPP3 had only low ATPase activities (compared with those of NTPDases) [17], suggesting that the hydrolysis of dinucleoside polyphosphates may be a more important physiological function of NPPs.

Ecto-5′-nucleotidase (CD73; EC 3.1.3.5)

Ecto-5′-nucleotidase catalyses the hydrolysis of nucleoside monophosphates to nucleosides (e.g., AMP → adenosine). Interestingly, this enzyme is inhibited by ATP and ADP, suggesting that the rate of adenosine production may depend on which other ectonucleotidase(s) is/are present in the proximity of ecto-5′-nucleotidase. For example, NTPDase1 will convert ATP through to AMP and thereby supply substrate for ecto-5′-nucleotidase and at the same time deplete the enzyme inhibitors, whereas NTPDase2 preferentially converts ATP to ADP, so provides little substrate for ecto-5′-nucleotidase, while the enzyme inhibitor ADP accumulates. Another feature of ecto-5′-nucleotidase is that it is attached to the cell membrane via a glycosylphosphatidyl inositol (GPI) anchor, which may be cleaved to release a soluble form.

Alkaline phosphatases (EC 3.1.3.1)

This family of enzymes has a broad substrate specificity, capable of catabolising nucleoside tri-, di- or monophosphates down to the related nucleoside. Thus, one enzyme can, in theory, catalyse the entire hydrolysis chain [8, 18]. However, the K_m values (i.e., the concentration of substrate required to attain half-maximal velocity for the reaction) for all these substrates are high, i.e. in the low millimolar range. Together with the fact that these enzymes prefer alkaline conditions, this raises questions about their physiological significance in relation to nucleotide degradation. Indeed, ATP is a rather poor substrate for these enzymes [18]. Nevertheless, there are reports showing that tissue-nonspecific alkaline phosphatase, together with ecto-5′-nucleotidase, is involved in AMP hydrolysis [19, 20]. Like ecto-5′-nucleotidase, alkaline phosphatases are GPI-anchored proteins, suggesting that they can also exist as soluble forms.

Intrarenal distribution of ectonucleotidases

NTPDases 1–3 and 8

NTPDases 1–3 and 8 have all been found in the kidney at the protein and/or mRNA level [12, 21–24].

Immunohistochemical studies show that NTPDase1 is prominent in the renal vasculature of rats and mice: it is present in interlobular arteries (endothelial layer and smooth muscle cells), afferent (but not efferent) arterioles and glomerular capillaries and mesangial cells [22, 23; J. Sévigny, unpublished observations]. Kishore et al. [22] also located NTPDase1 in peritubular capillaries (Fig. 2). A degree of interspecies agreement exists in that Lemmens et al. identified the porcine homologue of NTPDase1 in glomerular and peritubular capillaries of the pig kidney [21, 24]. As far as tubular structures are concerned, the only region of the kidney staining for NTPDase1 is the inner medulla, where it has been found in the rat thin ascending limb of Henle (identified as thin segments of tubule that did not stain for AQP1) [22] and (very-low-level staining only) collecting duct (identified using AQP2 as a marker) [23].

Fig. 2.

Immunohistochemical labelling of NTPDase1 in murine kidneys. a, b Labelling of medium-sized blood vessels (denoted by arrows) and peritubular capillaries (denoted by arrowheads) in mouse kidney cortex. c Labelling of glomerular mesangial cells and/or capillary membrane (denoted by arrow) and peritubular capillaries (denoted by arrowheads) in mouse kidney cortex. d Labelling of small blood vessels (denoted by arrows) and glomerular capillaries (denoted by arrowheads) in rat kidney cortex. e Labelling of peritubular capillaries in mouse kidney cortex (denoted by arrows), which was not seen when the specific antibody was substituted with pre-immune serum (f). Reproduced, with permission, from Kishore et al. [22]

NTPDase2 protein has been immunolocalised in the adventitial layer of blood vessels and in Bowman’s capsules in mice and rats [22] and in rat thick ascending limb of Henle (TALH; using Tamm–Horsfall protein as a marker) and distal tubules (using calbindin-D_28k as a marker), with again some low-level expression in the inner medullary collecting duct [23].

The intrarenal expression of NTPDase3 has been investigated only in the rat, where it was found in all post-proximal nephron segments examined: the TALH, the distal tubule and the entire collecting duct [23] (Fig. 3).

Fig. 3.

Immunohistochemical labelling of NTPDase3 in rat kidney. a Labelling of thick ascending limbs. b Same section as a, stained for Tamm–Horsfall protein. Scale bar 100 µm. c Labelling of distal tubules. d Same section as c, stained for calbindin-D_28k. Scale bar 100 µm. e Labelling of outer medullary collecting ducts. f Same section as e, stained for aquaporin-2. g Merged image of e and f, stained for NTPDase3 (green) and aquaporin-2 (red); arrows indicate presumed intercalated cells. Scale bar 50 µm. Reproduced, with permission, from Vekaria et al [23]

Firm information on NTPDase8 localisation in the kidney is scant. Using a monoclonal antibody raised against a liver ATP diphosphohydrolase, since identified as NTPDase8 [14], the enzyme was immunolocalised in porcine renal tubules, on brush-border membranes (presumably, therefore, proximal tubules) [21], but its exact distribution remains unknown. NTPDase8 was not detected in the renal vasculature [21].

It is worth mentioning that the study of Lemmens et al., which documented the immunolocalisation of NTPDase1 in the vasculature of porcine kidneys, also used a polyclonal antibody (named ‘Ringo’) that must have recognised additional members of the E-NTPDase family because its epitope is part of the apyrase conserved region IV of porcine NTPDase1, although this was unknown at the time [J. Sévigny, unpublished observation]. This polyclonal antibody stained a number of tubular structures: Bowman’s capsule, proximal and distal tubules, ascending segments of Henle’s loop and papillary collecting ducts [24]. Furthermore, Lemmens et al. used an enzyme histochemical method to detect ATPase and ADPase activities along the nephron. Electron microscopy revealed strong enzyme activity in proximal tubular brush-border membranes, with some activity also on the basolateral membrane [24]. At the light microscopic level, the authors also showed strong ATPase activity in the proximal convoluted tubules and blood vessels, lower ATPase activity in the glomeruli and no activity in the distal tubules.

NPPs 1–3

Only limited information is available concerning the presence and distribution of NPPs in the kidney. In the mouse, NPP1 protein has been identified in proximal and distal tubules but was absent from glomeruli [25]. Staining was much stronger in the distal tubules, where it was located in the basolateral membrane. Absence of suitable antibodies to rat NPP1 or NPP2 has thus far precluded an investigation of enzyme protein along the nephron in this species. However, a recent study in our laboratory has used a specific antibody to rat NPP3 to examine this enzyme’s intrarenal distribution [23]. Prominent staining for NPP3 was found in glomeruli and in the brush-border membrane of proximal straight tubules, identified using neutral endopeptidase antibody as a marker of the rat S3 segment (Fig. 4). NPP3 was absent from more distal regions of the nephron.

Fig. 4.

Immunohistochemical labelling of NPP3 in rat kidney. a Labelling of glomerulus. Scale bar 50 µm. b Labelling of proximal tubules. Arrows show positively stained proximal tubules lacking NPP3 expression. c Same section as b, stained for Phaseolus vulgaris erythroagglutinin, a marker of proximal tubules. Scale bar 100 µm. d Labelling of pars recta. e Same section as d, stained for neutral endopeptidase, a marker of rat proximal tubular S3 segment. Scale bar 100 µm. Reproduced, with permission, from Vekaria et al [23]

Ecto-5′-nucleotidase

Ecto-5′-nucleotidase is present in all tissues studied. It has a high level of expression in the kidney [26], where it is the most documented ectonucleotidase. Cole et al [6] were the first to report high ecto-5′-nucleotidase activities in rat proximal tubule. Subsequently, a comprehensive mapping of the enzyme protein in the kidney has been performed by Kaissling’s group, and more recently by Vekaria et al. , using immunohistochemical techniques. The enzyme has been detected in the brush-border membrane of the rat proximal tubule, mainly in the S1 and, to a lesser extent, S2 segments, and in the apical membrane of intercalated cells in the connecting tubule and collecting duct [23, 27–29]. It is also found outside the tubule, in the peritubular space—probably in interstitial fibroblasts [28]. In mice (but not rats) it is also present in glomerular mesangial cells [29].

Alkaline phosphatases

Alkaline phosphatases have not been adequately studied in the kidney. One iso-enzyme has been identified in the brush-border membrane of proximal tubules in the rat kidney [30], although the enzyme’s high K_m values for adenine nucleotides (vide supra) suggests that its function there may be unconnected with nucleotide degradation.

Figure 5 summarises existing knowledge of the distribution of ectonucleotidases along the rat nephron.

Fig. 5.

Distribution of ectonucleotidases along the rat nephron. Information compiled from Cole et al. [6], Kishore et al. [22], Vekaria et al. [23], Gandhi et al. [27], and Le Hir and Kaissling [28, 29]. Adapted, with permission, from Vekaria et al. [23]. N.B. No information is yet available concerning NPPs 1 and 2 in the rat. NTPDase8 has been detected in porcine tubules [21], but its exact location is not yet known

Functional significance of renal ectonucleotidases

The existence of a variety of P2, as well as P1, purinoceptors in the renal vasculature and along the nephron suggests that enzymatic modification of secreted or synthesised extracellular nucleotides will profoundly influence renal physiological and/or pathophysiological events. At present, since our knowledge of ectonucleotidase location is incomplete, and since there are uncertainties as to the cellular polarity (apical vs. basolateral) of some of the enzymes, it is not possible to describe the precise consequences of ectonucleotidase action. It seems likely, however, that the hydrolysis of extracellular nucleotides will result in a constantly shifting activation of different purinoceptor subtypes. Based on existing knowledge, we can make a number of—sometimes speculative—inferences.

Glomerular and pre-glomerular function

The NTPDase1 located in pre-glomerular vessels, in glomeruli and in peritubular capillaries is likely to fulfil a function of NTPDase1 found elsewhere in the vasculature: termination of the platelet aggregation response to extracellular ADP [31, 32]. In addition, NTPDase1 may specifically affect renal blood flow and glomerular filtration. Stimulation of P2X₁ receptors by ATP in afferent arterioles causes vasoconstriction (see article by Inscho [33] in this Special Issue); hydrolysis of locally produced ATP may therefore be important in maintaining renal blood flow. In this context, circulating or locally produced diadenosine polyphosphates can also cause vasoconstriction of afferent arterioles [34], and NPPs can hydrolyse these compounds. However, although NPP3 has been detected in glomeruli, thus far it has not been described in pre-glomerular vessels.

In the glomerulus, ATP induces relaxation of pre-contracted mesangial cells [35]. NTPDase1 and NPP3 may modulate this activity, thereby influencing the capillary surface area available for filtration (K_f). Additionally, these enzymes may influence other glomerular purinoceptor responses such as P2Y-dependent cell proliferation of mesangial cells, or may serve in the protection of these cells by preventing ATP levels reaching concentrations that activate the apoptotic P2X₇ receptor [36].

Tubuloglomerular feedback

As described elsewhere in this Special Issue (see articles by Inscho [33] and Bell et al. [37]), there is evidence that ectonucleotidases play a pivotal role in tubuloglomerular feedback (TGF), whereby changes in renal perfusion pressure, or other causes of altered NaCl delivery to the macula densa, ultimately cause compensatory changes in afferent arteriolar resistance so that glomerular filtration rate is regulated [38]. ATP concentrations in renal interstitial fluid increase in response to elevations in renal arterial perfusion pressure [39], and, as indicated above, ATP can act directly on P2X₁ receptors on the afferent arteriole to induce constriction [40]. However, strong evidence also exists for adenosine (acting via A₁ receptors) being the chemical mediator in this response [41, 42]. In this context, it is possible that NTPDase1, expressed in the peritubular space, and/or NPP1, expressed basolaterally in the distal nephron of mice [25], might convert any unbound or excess ATP into AMP. The latter, being a suitable substrate for ecto-5′-nucleotidase, also expressed in the peritubular space, could then be converted to adenosine to cause or augment the vasoconstrictive response in TGF. Strong support for this scheme comes from the observation that ecto-5′-nucleotidase ‘knockout’ mice display a blunted TGF response [43, 44]. Moreover, the response in wild-type mice is blocked by adenosine (A₁) receptor blockade [44]. As yet, no information on TGF is available from NTPDase or NPP ‘knockout’ mice.

Proximal tubular function

Although NPP3 is present in the proximal straight tubule of the rat, thus facilitating degradation of ATP or ADP to AMP in this segment, the enzyme’s presence throughout the proximal convoluted tubule (PCT) could not be confirmed [23]. As indicated above, we do not yet know whether NPP1 or NPP2 is present, but a liver ATP diphosphohydrolase, now known to be NTPDase8, was detected in porcine tubules (most likely to be PCTs), predominantly on the luminal side [21]. In accord with this, ATPase and ADPase activities were described in the PCT brush-border and, to some extent, basolateral membranes [24]. It is uncertain whether proximal tubular alkaline phosphatase, with its high K_m values for ATP and ADP, could be responsible for degrading ATP. However, it is clear that any AMP produced in the proximal tubule should be converted to adenosine by the ecto-5′-nucleotidase present throughout this segment. The adenosine would be expected to activate proximal tubular A₁ receptors [45], causing increased sodium and water reabsorption [46]. Additionally or alternatively, adenosine may be taken up by epithelial cells for re-use: nucleoside transporters are present in both apical and basolateral membranes of proximal tubular cells [47]. It is worth noting that free-flow micropuncture studies in ecto-5′-nucleotidase ‘knockout’ mice failed to observe any disruption of PCT function under standard conditions [44], although the possibility of chronic compensatory mechanisms in these genetically engineered animals cannot be excluded.

Distal nephron function

Extracellular nucleotides have a number of major (largely inhibitory) effects on solute and water transport in the distal nephron, particularly in the collecting duct, acting from both apical and basolateral sides (see articles by Bailey and Shirley [48] and Kishore et al [49] in this Special Issue). Furthermore, activation of adenosine A₁ receptors, found in medullary collecting ducts [45], reduces sodium reabsorption [50]. Therefore, the presence of NTPDases and ecto-5′-nucleotidase throughout the distal nephron (see Fig. 5) is likely to be of considerable functional significance. Indeed, adenosine formation in rat collecting ducts was reported to be reduced when ecto-5′-nucleotidase was inhibited with α,βMeADP [51], indicating that this enzyme (and, by implication, at least one AMP-producing enzyme) is tonically active. It remains to be seen how inhibition/deletion of this and other enzymes affects transport processes.

Renal ectonucleotidases under pathological conditions

Given that adenine nucleotides are in general pro-inflammatory [52], it is unsurprising to find that ectonucleotidases can have a renal protective function in a number of pathological situations. For example, in both type 1 and type 2 diabetes, there is evidence that patients with a polymorphism of the gene encoding NPP1 have more severe reductions in renal function than those seen in matched patients with a normal gene [53, 54]. NTPDase1 also appears to be important in this condition: NTPDase1 ‘knockout’ mice subjected to streptozotocin-induced diabetes exhibit increased proteinuria and more severe glomerular sclerosis compared with their wild-type counterparts [55]. This enzyme is also implicated in offsetting organ rejection processes [56]. It has been reported that the glomerular expression of ‘ecto-ATPase’ (a term used for enzymes that we now believe to belong mainly to NTPDase family members) is reduced in patients with chronic allograft nephropathy [57] and also in rats with experimental renal transplant failure [58]. Thus, down-regulation of NTPDases may be an unwanted consequence of the rejection process, contributing to its progression. A similar conclusion has been drawn with regard to ischaemia–reperfusion injury [59]. Set against this is the recent study of Grenz et al [60] in which mice were subjected to short periods of ischaemic preconditioning, which helped to protect renal function against a subsequent period of ischaemia. It was found that ischaemic preconditioning induced an increase in renal NTPDase1 (but not in NTPDases 2, 3 or 8) mRNA and protein; moreover, renal protection was absent in NTPDase1 ‘knockout’ mice.

As already indicated, the chief end-product of NTPDase activity is AMP, the substrate for ecto-5′-nucleotidase; this enzyme can also be up-regulated in pathological situations. Bakker and colleagues have reported increased glomerular expression of ecto-5′-nucleotidase in patients with chronic allograft nephropathy [57] and in those with glomerular ischaemia due to malignant hypertension [61]; increased activity of ecto-5′-nucleotidase in fibroblasts in the peritubular space of chronically hypoxic rats has also been reported [62]. A recent detailed study of the importance of ecto-5′-nucleotidase in protecting against renal ischaemia in mice demonstrated that short periods of ischaemic preconditioning caused increases in renal ecto-5′-nucleotidase mRNA and protein, and in renal tissue adenosine [63]. That these events were critical to the protective mechanisms was indicated by the findings that renal function after a subsequent period of renal ischaemia was severely compromised in mice subjected to either pharmacological inhibition of ecto-5′-nucleotidase or deletion of the gene encoding the enzyme [63].

The increased production of adenosine following up-regulation of ecto-5′-nucleotidase under pathological conditions may counteract the effects of ischaemic damage by several mechanisms: adenosine inhibits platelet aggregation, promotes vasodilatation, and scavenges reactive oxygen species [57]. Okusa [64] reported that stimulation of A_2A receptors on renal endothelial cells significantly reduced the expression of endothelial intracellular adhesion molecule-1. This molecule is required for binding and activating neutrophils, a key mechanism for pathogenesis during ischaemia–reperfusion injury. Thus, under these conditions the increased levels of adenosine might serve to prevent inflammatory responses in the renal endothelium. Adenosine A₃ receptors have been identified on mesangial cells of the glomerulus, the activation of which results in apoptosis [65]. During pathophysiological conditions, it could be argued that apoptosis should be favoured over necrosis, as the latter may augment inflammatory responses and result in further tissue damage.

Inhibitors of ectonucleotidases

The use of inhibitors of ectonucleotidases in attempts to determine the physiological roles of the enzymes is fraught with difficulty, partly because of the absence of selective and potent inhibitors for each individual enzyme, and partly because most inhibitors have other effects on the purinoceptor system. Thus, most agents used as P2 receptor antagonists, such as suramin and its derivatives, Reactive Blue 2 (RB2), Coomassie Brilliant Blue, and pyridoxal phosphate-6-azophenyl-2′, 4′-disulphonic acid (PPADS), also inhibit ‘ecto-ATPase’ [9, 66]. Indeed, RB2 and the suramin derivative NF279 are potent inhibitors of NTPDases, almost completely abrogating the activity of these enzymes at a concentration of 100 µM [66]. RB2 and PPADS also inhibit NPP1 [67]. PPADS and suramin have little effect on ecto-5′-nucleotidase; as indicated earlier, this enzyme is inhibited by α,βMeADP, a compound with no reported P2 receptor activity. The ATP analogue 6-N, N-diethyl-d-β,γ-dibromomethylene ATP (ARL 67156) is another commercially available ecto-ATPase inhibitor that does not affect significantly purinoceptors. Recent reports demonstrated that ARL 67156 is a weak competitive inhibitor of NTPDase1, NTPDase3 and NPP1, with K_i for the human enzymes of 11 ± 3, 18 ± 4 and 12 ± 3 µM, respectively [17, 68]. Another ATP analogue, 8-thiobutyladenosine 5′-triphosphate (8-BuS-ATP), and 1-naphthol-3,6-disulfonic acid (BG0136), appear of interest, but so far they have only been characterised on a NTPDase-active fraction from bovine spleen membranes [69, 70]. In addition, recent evidence indicates that some polyoxometalate anionic complexes inhibit rat NTPDases [71], that a uridine-5′-carboxamide derivative selectively inhibits human NTPDase2 [72] and that a novel monoclonal antibody selectively inhibits human NTPDase3 [73].

Finally, ion chelators (e.g., EDTA), by reducing the availability of the divalent cations necessary for ectonucleotidase activity, are potent inhibitors of all these enzymes. However, in addition to their lack of specificity with regard to the various ectonucleotidase families, their use in vivo is compromised by disruption of normal physiological processes. Similarly, inhibition of ectonucleotidases by NaN₃ [74] is also inappropriate for in vivo studies, as azide is a more potent inhibitor of the essential enzyme mitochondrial ATPase [75].

Conclusions

Although our knowledge of the intrarenal location of the ectonucleotidase families is still incomplete, it is evident that their distribution varies significantly along the different segments of the nephron. It does not seem too far-fetched to propose that the enzymes are strategically distributed so as to influence the activity of P2 and P1 purinoceptors through the generation, or hydrolysis, of agonists such as ATP, ADP or adenosine. In addition, it is possible that ectonucleotidases prevent desensitisation of tubular P2 receptors, as has been demonstrated in other tissues [76–78]. Finally, it has been hypothesised that in conditions such as ischaemia, the ectonucleotidase-mediated modulation of the inhibitory effects of nucleotides on water and electrolyte reabsorption might be overwhelmed by excess ATP release, thereby reducing energy-consuming reabsorptive processes [79].

We emphasise that much remains to be learned. It is hoped that insights into the physiological/pathophysiological roles of individual ectonucleotidases will emerge from studying the phenotypes of genetically manipulated ‘knockout’ animals, although long-term compensatory changes often hamper interpretation of the results of such studies. An alternative/complementary strategy is the development of further iso-enzyme-specific inhibitors. A pessimistic view, however, is that the overlapping catalytic activities of the various family members may bedevil both approaches.