ATP and adenosine in the local regulation of water transport and homeostasis by the kidney

Abstract

Regulation of body water homeostasis is critically dependent on the kidney and under the control of AVP, which is released from the neurohypophysis. In the collecting duct (CD) of the kidney, AVP activates adenylyl cyclase via vasopressin V₂ receptors. cAMP-dependent activation of protein kinase A phosphorylates the water channel aquaporin-2 and increases water permeability by insertion of aquaporin-2 into the apical cell membrane. However, local factors modulate the effects of AVP to fine tune its effects, accelerate responses, and potentially protect the integrity of CD cells. Nucleotides like ATP belong to these local factors and act in an autocrine and paracrine way to activate P2Y₂ receptors on CD cells. Extracellular breakdown of ATP and cAMP forms adenosine, the latter also induces specific effects on the CD by activation of adenosine A₁ receptors. Activation of both receptor types can inhibit the cAMP-triggered activation of protein kinase A and reduce water permeability and transport. This review focuses on the role and potential interactions of the ATP and adenosine system with regard to the regulation of water transport in the CD. We address the potential stimuli and mechanisms involved in nucleotide release and adenosine formation, and discuss the corresponding signaling cascades that are activated. Potential interactions between the ATP and adenosine system, as well as other factors involved in the regulation of CD function, are outlined. Data from pharmacological studies and gene-targeted mouse models are presented to demonstrate the in vivo relevance to water transport and homeostasis.

water is the most abundant molecule in the human body. The maintenance of water balance is dependent on water intake, initiated by the sensation of thirst, and the regulation of water excretion by the kidneys. Whereas all nephron segments contribute to various extents to water homeostasis, it is primarily the collecting duct (CD) system in which the reabsorption of ∼5–10% of the filtered water is regulated and adjusted to meet bodily needs. The antidiuretic hormone AVP is the primary regulator of water reabsorption in the CD system and critically involved in the regulation of water balance and stabilization of plasma osmolality (52). AVP is released from the neurohypophysis in response to small increases in plasma osmolality or greater reductions in circulating volume (52). As illustrated in Fig. 1, AVP acts on the CD via the G_s protein-coupled vasopressin V₂ receptor (V₂R) to stimulate adenylyl cyclase (AC) and thus the synthesis of cAMP. Increases in cAMP activate PKA, which phosphorylates the water channel aquaporin-2 (AQP2), with subsequent insertion of the channel into the apical plasma membrane. This allows water to be reabsorbed from the CD lumen into the cell and via basolateral AQP3 and AQP4 into the interstitium down its osmotic gradients. In addition, PKA-mediated phosphorylation of a cAMP-response element binding protein (CREB protein) promotes its binding to DNA and increases the transcription of the AQP2 gene in the longer term (for a review, see Ref. 50).

Fig. 1.

Proposed model for signaling mechanisms involved in aquaporin-2 (AQP-2) regulation in inner medullary collecting duct principal cells. See text for details. A₁R, adenosine A₁ receptor; AC, adenylyl cyclase; COX-1, cyclooxygenase 1; CREB, cAMP responsive element-binding protein; DAG, diacylglycerol; EP₃, PGE₂ receptor subtype 3; ET-1, endothelin-1; ET_B, endothelin B receptor; G_i, inhibitory G protein; G_q, phospholipase C stimulatory G protein; G_s, stimulatory G protein; IP, inositol triphosphate; βγ, G protein βγ subunit; P, phosphorylation; PKA, protein kinase A; PKC, protein kinase C; PLC, phospholipase C; V₂R, vasopressin V₂ receptor.

Importantly, the AVP-mediated effects on water reabsorption are modulated by local factors. The functional importance of local factors may, in part, relate to the relatively long AVP half-life of 10–35 min (15). Since AVP-mediated free water reabsorption in the kidney is an important determinant of plasma osmolality, and thus the water distribution between extracellular and intracellular spaces, rapid adjustments in renal free water handling in response to changes in plasma osmolality are necessary. Local mechanisms may thus serve to tune CD water transport to bodily needs until responses in AVP levels can take over, as well as to fine adjust the effects induced by AVP. In addition, local factors may modulate AVP actions to protect CD cells from excessive perturbations in cell volume as a consequence of rapid changes in extracellular osmolality in the inner renal medulla. Increases in CD cell volume may release local factors and thus serve as a sensor of changes in plasma osmolality to accelerate AVP responses as well to maintain cell volume and integrity (see below).

As illustrated in Fig. 1, local AVP-counterregulatory factors in the CD principal cell include endothelin-1 (ET-1) (44) and PGE₂ (10). ET-1 is formed and released from CD cells in response to changes in osmolality (45). However, the exact mechanisms are still under investigation. ET-1 activates ET_B receptors in an autocrine/paracrine fashion to reduce PKA activation (see Fig. 1). Supporting this concept, inner medullary CD (IMCD) suspensions of CD-specific ET-1 knockout mice have enhanced AVP- and forskolin-stimulated cAMP formation and reduced plasma AVP levels (26). ET_B receptor activation also stimulates cyclooxygenase-1 (COX-1) and PGE₂ formation and release (46). The most abundant PGE₂ receptor in the CD is the EP₃ receptor subtype (38), which is a G_i protein-coupled receptor that inhibits AVP-induced cAMP formation in the CD (9, 10). In addition, EP₃ receptor activation inhibits PKA via PLC-mediated activation of PKC (30). Of note, ET_B receptors were found to be mainly localized to the basolateral membrane (46), whereas V₂R (51) and EP₃ receptors (65) were localized to both the luminal and basolateral membrane of CD cells, which is expected to further enhance the regulatory versatility.

Over the last years, evidence is accumulating that the renal tubular and CD system releases nucleotides like ATP in response to physiological stimuli and that these nucleotides act locally to modulate transport processes (for review, see Refs. 48, 74, 75). ATP has a short extracellular half-life of <5 min and can be broken down to form adenosine, another well-established local regulator of renal vascular and tubular function (for a review, see Ref. 76). The regulation of CD water transport by ATP and adenosine is the focus of this review.

P2Y Receptors and CD Water Transport

P2Y receptors are heptahelical receptors, which can be subdivided into five G_q-coupled subtypes (P2Y₁, P2Y₂, P2Y₄, P2Y₆, P2Y₁₁) and three G_i-coupled subtypes (P2Y₁₂, P2Y₁₃, P2Y₁₄) (1). Whereas the G_q-coupled receptors activate PLC and increase [Ca²⁺]_i, the G_i-coupled subtypes inhibit AC and lower cAMP levels. With different techniques P2Y₁, P2Y₂, P2Y₄, and P2Y₆ have been identified in the CD (4, 5, 43, 72, 73). Depending on the receptor subtype, P2Y receptor can be activated by ADP, ATP, UDP, UTP, or UDP sugars. P2Y₂ receptors are typically activated nearly equipotent by ATP and UTP (1). Several groups have described an inhibitory effect of ATP on AVP-stimulated osmotic water permeability (P_f) in the cortical CD (CCD) and IMCD. On the basis of a pharmacological approach, many of these studies proposed a role of a P2Y₂-like receptor as outlined below.

Rouse et al. (64) (Table 1) were the first to report that ATP can modulate the AVP-stimulated P_f in rabbit CCD by activation of the PLC/Ca²⁺ signaling pathway, consistent with the expression of a functional purinergic receptor. Ecelbarger et al. (19) provided evidence for the existence of a P2Y₂-like receptor in rat terminal IMCD using ATP, UTP, and ATPγS (a stable analog of ATP), which all increased [Ca²⁺]_i. Cha et al. (13) showed in isolated rat outer medullary CD the possible involvement of P2Y₁ and/or P2Y₂ receptors using 2-methylthio-ATP (2-MeS-ATP, P2Y receptor agonist), suramin and reactive blue 2 (both P2Y receptor antagonists). Kishore et al. (42) demonstrated in the isolated perfused terminal IMCD of the rat that ATP decreases AVP-stimulated cAMP formation by ∼30% in a PKC-dependent manner, most likely resulting from activation of the phosphoinositide signaling pathway; this was associated with inhibition of AVP-stimulated P_f by ∼40%. Supporting this concept, in rat IMCD cells, activation of PKC resulted in inhibition of AVP-stimulated cAMP formation (69, 70) (Fig. 1).

Table 1.

Effects of P2Y and adenosine A₁ receptor activation in native renal tissue

Receptor	Segment	Effect	Method (Site of Effect) Criterion for Receptor Subtype
P2Y2	CD	AVP-stimulated cAMP formation and water reabsorption↓	Evidence in P2Y2−/− mice (59)
P2Y2-like	CD, diverse	PLC↑, [Ca2+]i↑, PKC↑, cAMP↓, AVP-stimulated Pf↓	Isolated, perfused CCD, OMCD, or IMCD of rabbit or rat (B) agonist profile (13, 19, 21, 42, 64)
P2Y2-like	CCD	PLC↑, [Ca2+]i↑	Isolated, perfused rat or rabbit CCD, principal and intercalated cells (A and B) agonist profile (17, 64, 84)
P2Y2-like	IMCD	COX-1-dependent PGE2 release↑, hydrated > dehydrated rats, endothelin-1 release↓	Rat IMCD suspension (B) agonist profile (33, 67, 81)
A1R	CD	Basal cAMP formation↓	Isolated, cultured rabbit CCD (2, 3), rat IMCD primary culture (85, 86) and freshly isolated IMCD of mice with A1R agonist (60)
A1R	CD	AVP-stimulated cAMP formation↓	Evidence in A1R−/− mice (60)
A1R	CCD, IMCD	AVP-stimulated cAMP formation↓ dependent on a functional Gi protein	Cultured rabbit CCD with A1R agonist (2, 3), rat IMCD primary culture with A1R agonist (85)
A1R	IMCD	AVP-induced increases in Pf and cAMP formation↓	Rat IMCD (B) A1R agonist (20)
A1R	IMCD	AVP-stimulated cAMP formation↓	Rat IMCD primary culture with A1R agonist (A, B) (85, 86)
A1R	IMCD	IP3↑, [Ca2+]i↑	Isolated, superfused terminal IMCD of rat (19) and cultured rabbit CCD cells with A1R agonist (2)

A, apical; AVP, arginine-vasopressin; B, basolateral; [Ca²⁺]_i, intracellular Ca²⁺; IP, inositol phosphates; P_f, osmotic water permeability; PLC, phospholipase C; PKC, protein kinase C; COX-1, cyclooxygenase 1; CD, collecting duct; CCD, cortical collecting duct; IMCD, inner medullary collecting duct, OMCD, outer medullary collecting duct.

Ecelbarger et al. (19) demonstrated that prior exposure of rat IMCD to indomethacin, an unselective COX inhibitor, attenuates [Ca²⁺]_i responses to ATP, suggesting that PGE₂ is facilitating or mediating the response in [Ca²⁺]_i. Moreover, ATPγS stimulated PGE₂ release in freshly isolated IMCD preparations of hydrated rats, whereas the response was blunted in dehydrated rats (67). Finally, hypervolemic conditions are associated with greater P2Y₂ receptor abundance in the renal medulla than hypovolemia (68). The purinergic-prostanoid system was speculated to represent a vasopressin-independent regulatory mechanism of IMCD function (68).

To further substantiate the above pharmacological evidence, we studied aspects of renal water transport in mice lacking P2Y₂ receptors (P2Y₂^−/−) (59). Studies in freshly isolated IMCD showed that ATPγS enhanced the EC₅₀ for the stimulation of cAMP formation by the V₂R agonist, 1-desamino-8-d-arginine vasopressin (dDAVP) in wild-type (WT) mice, but not in P2Y₂^−/− mice (59). To determine the ambient contribution of V₂R activation on water transport in WT and P2Y₂^−/− mice, acute responses to the V₂R antagonist SR121463 were assessed. In WT animals, SR121463 increased urinary flow rate and electrolyte free water clearance (Cl^e-H₂O) (Fig. 2, A and C). These changes in WT were associated with an increase in urinary PGE₂ (Fig. 2B) but a reduction in urinary ATP (Fig. 3A), implying a tonic inhibition of PGE₂ formation but stimulation of ATP release by V₂R activation (the regulation of ATP release in CD is discussed in more detail below). In P2Y₂^−/−, SR121463 elicited a significantly greater diuresis and Cl^e-H₂O compared with WT (Fig. 2, A and C), indicating greater basal reabsorption of fluid in the CD of P2Y₂^−/−, which was associated with greater renal AQP2 expression in the latter animals. SR121463 induced similar increases in urinary PGE₂ in P2Y₂^−/− and WT (Fig. 3A). Basal urinary excretion of fluid and Cl^e-H₂O were not different between P2Y₂^−/− and WT mice. This is proposed to be the consequence of a greater delivery of a more hypotonic fluid to the distal nephron segments in P2Y₂^−/−, which is the consequence of the integrated renal and blood pressure phenotype of these mice (59, 75). Under basal conditions, the hyperactivity of the V₂R-AQP2 system reabsorbs the excess water resulting in normal net fluid excretion in P2Y₂^−/−. Together, the data are consistent with the concept that P2Y₂ receptor activation inhibits the cAMP-mediated effects of AVP on water transport in the CD (Fig. 1) (59).

Fig. 2.

Aspects of renal water transport in P2Y₂ knockout mice (P2Y₂^−/−) under basal conditions (basal), in response to acute vasopressin V₂ receptor inhibition (V₂-I), acute water loading (WL) or WL plus pretreatment with indomethacin (WL+Indo). A, C: compared with basal measurements, V₂-I increased urinary flow rate and electrolyte free-water clearance (Cl^e-H₂O) to a greater extent in P2Y₂^−/− vs. wild-type (WT) mice (determined over a 2-h period after intraperitoneal application of the receptor inhibitor). These findings indicated greater water reabsorption but also the basal delivery of greater amounts of more hypotonic fluid to the distal nephron in P2Y₂^−/−. This could explain the facilitated Cl^e-H₂O in response to water loading in P2Y₂^−/−, i.e., less suppression of vasopressin and/or cAMP is sufficient to increase Cl^e-H₂O to the same extent as in WT (not shown). B: Increases in urinary PGE₂ after oral water loading (3% of body weight) were much greater in P2Y₂^−/−; treatment with indomethacin reduced urinary PGE₂ excretion in water-loaded P2Y₂^−/− to levels of WT mice without indomethacin treatment; urinary flow rate and electrolyte-free water clearance (Cl^e-H₂O) were unaffected in P2Y₂^−/− mice compared with untreated. *P < 0.05 vs. WT; data for WL+Indo are unpublished; the other data are from Ref. 59.

Fig. 3.

A: relationship between urinary ATP excretion and urinary flow rate under different maneuvers. Urinary ATP was reduced by acute vasopressin V₂ receptor inhibition (V₂-I) but increased by acute WL. Increasing urinary flow rate alone is not a good predictor of urinary ATP excretion. For further details, see text. B: a proposed positive relationship between the cell volume, manipulated by different maneuvers in WT and P2Y₂^−/− mice and urinary ATP excretion. We propose that feedback regulation of cell volume via cell volume-regulated ATP release is absent/reduced in P2Y₂^−/− mice resulting in greater ATP release. The influence is minimized by reducing water uptake by V₂-I and maximized by water loading-induced cell swelling. All data are from Ref. 59.

How does the absence of P2Y₂ receptors affect the response to acute water loading? Acute water loading induced similar increases in urine flow rate and Cl^e-H₂0 in WT mice as observed in response to V₂R blockade (Fig. 2, A and C). Likewise, water loading increased urinary PGE₂ excretion (Fig. 2B). In contrast to V₂R blockade, however, water loading enhanced urinary ATP excretion (Fig. 3A). In P2Y₂^−/−, the water load-induced increase in Cl^e-H₂O was similar to that in WT. However, a lesser suppression of the vasopressin and cAMP system was sufficient in P2Y₂^−/− to increase Cl^e-H₂O to the same extent as in WT, indicating that free water excretion was actually facilitated in the P2Y₂^−/− (59). Whereas this may relate to the delivery of greater amounts of hypotonic fluid to the distal nephron in these mice, they also presented greater urinary excretion of both PGE₂ (Fig. 2A) and ATP in response to acute water loading (Fig. 3A).

Our unpublished work showed that treatment of WT mice with indomethacin (5 mg/kg ip) 30 min before an acute oral water load (3% of body weight) almost completely inhibited urinary PGE₂ excretion and blunted the urine flow response in the 2-h experimental period (Fig. 2, A and B). Remarkably, urinary PGE₂ excretion in P2Y₂^−/− was reduced to levels observed in WT mice without indomethacin treatment, and the responses in urinary flow rate and Cl^e-H₂O were unaffected compared with untreated P2Y₂^−/− mice (Fig. 2, A and C). Whether this is due to incomplete COX inhibition in P2Y₂^−/− or involves indomethacin-insensitive PGE₂ formation, needs to be determined, but it stresses potential differences in the quantitative or qualitative role of PGE₂ in P2Y₂^−/−.

P2X Receptors and CD Water Transport

P2X receptors are ligand-activated ion channels with permeability to Na⁺, K⁺, Ca²⁺ and in a few cases Cl⁻, and P2X₁, P2X₂, P2X₃, P2X₄, P2X₅, P2X₆, and P2X₇ channels have been described (40, 53). Studies in native rat tissue have immunolocalized P2X₄, P2X₅, and P2X₆ in CD principal cells (74) and protein expression of P2X₁, P2X₄, and P2X₆ was confirmed by mRNA expression of these receptors in microdissected CD (72). Preliminary studies in the Xenopus oocyte expression system proposed a functional interaction between AQP2 and P2X₂ (82, 83), but the in vivo relevance remains to be determined. In conclusion, very little is known about the physiological relevance of P2X receptors for water transport in the kidney. The use of knockout mouse models should be helpful to unravel their functions.

Nucleotide Release in CD

Cellular ATP release is an essential and physiological relevant part of purinergic signaling in the kidney with the released ATP acting on P2 receptors in an autocrine/paracrine way (47, 59). Possible sources of extracellular ATP in the kidney include perivascular and peritubular nerve terminals, circulating erythrocytes, aggregating platelets, and renal endothelial and epithelial cells (8, 14). In human renal cortex, adrenergic stimulation was shown to release ATP from neuronal and nonneuronal sources (79). Under basal conditions, intracellular ATP concentrations are in the range of 3–5 mM (27). Thus, there is a big pool of intracellular ATP and just a fraction (resulting in values 100-fold lower than those inside the cell) needs to be released to mediate purinergic autocrine/paracrine signaling (1, 58). Once outside the cell, ATP has a half-life of minutes due to ecto-nucleotidases and other hydrolytic activities (23).

With regard to ATP release mechanisms in the kidney, evidence has been provided for a role of a maxi-anion channel for the basolateral release of ATP from the macula densa, but the molecular nature of this channel is unknown (7). The released ATP is converted by ecto-ATPase (54) and ecto-5′-nucleotidase (12, 32) to adenosine, the final mediator of the tubuloglomerular feedback (71), illustrating the close link between the ATP and adenosine system. The mechanisms and proteins involved in ATP release in the tubule and CD system are also largely unknown. Connexin (Cx) hemichannels have been proposed to contribute to ATP release, and recent studies by McCulloch et al. (49) identified continuous Cx30 hemichannel expression from the medullary thick ascending limb to the CD system. In the mouse CCD, the expression of Cx30.3 appeared to be restricted to the cytosol and the apical membrane of intercalated cells (29). A physiological role in ATP release remains to be established.

The presence of ATP in tubular fluid and its release by epithelial cells in the rat was described by means of micropuncture (77). In this study, the half-life of ATP in proximal tubular fluid was ∼3.4 min with concentrations between 100 and 300 nM. Concentrations closer to the plasma membrane are expected to be significantly greater. Measurement of ATP in distal tubules showed 3.5-fold lower ATP values compared with the proximal tubule. However, ATP concentrations in the distal tubule may have been underestimated because of the need for longer collection times and the presence of soluble nucleotidase. ATP in tubular fluid of CD has not been measured yet.

ATP release in the CD is triggered by changes in tubular flow rate, as well as cell volume (31, 39). In the kidney, flow-induced ATP release was shown to be at least partially dependent on the presence of the primary cilium (57). Cilia have important roles in sensory physiology in response to flow and osmotic stimuli (reviewed in Refs. 57 and 66). Supporting this concept, Hovater et al. (31) showed that ATP release is impaired when the cilium is malformed. This conclusion was drawn from studies in CD principal cells derived from an Oak Ridge polycystic kidney (Tg737orpk) mouse model of autosomal recessive polycystic kidney disease, which lack a well-formed apical cilium: principal cell monolayers with normally formed apical cilia responded with 3- to 5-fold greater ATP release to hypotonicity than mutant monolayers lacking cilia. The observed cilium-derived Ca²⁺ transient required an underlying paracrine/autocrine ATP signal that is likely transduced by P2X and P2Y on or near the cilium (31). Notably, Woda et al. (84) showed that a flow-induced rise in [Ca²⁺]_i can also be triggered in intercalated cells from perfused rabbit CD, which do not have cilia.

IMCD were shown to respond to AVP-stimulation with an increase in cell volume (25). Cells of various organs respond to changes in cell volume with a release of ATP (23, 80). Measuring urinary ATP excretion, our studies in P2Y₂^−/− mice provided indirect evidence for a volume-dependent ATP release in CD in vivo (59). Urinary ATP excretion was similar in P2Y₂^−/− compared with WT during inhibition of water transport by V₂R blockade but modestly greater under basal conditions and much greater in response to acute water loading in P2Y₂^−/− compared with WT. A summary of these findings is illustrated in Fig. 3A. V₂R blockade and acute water loading both increased urinary flow rate but had opposite effects on ATP excretion. We propose that 1) urinary flow rate alone is not a good predictor of urinary ATP excretion, 2) acute water loading increases ATP release, which may reflect increases in CD cell volume due to a reduction in extracellular tonicity, 3) acute pharmacological blockade of V₂R reduces cell volume by blocking the apical water entry, thus offsetting the basal, AVP- and cell volume-induced ATP release, and 4) urothelial cells of the lower urinary tract, which were exposed to similar increases in flow rate and therefore similar distension, and hypotonicity may not play a dominant role for urinary ATP excretion. A graph illustrating the proposed relationship between CD cell volume and ATP release and thus urinary ATP excretion is shown in Fig. 3B.

Adenosine Receptors and CD Water Transport

Consistent with a prominent role of adenosine A₁ receptor (A₁R) in the regulation of CD function, studies in rats and mice revealed a strong corticomedullary gradient for the A₁R expression, with the highest density in CD and, in particular, IMCD (56, 73, 76, 78, 87, 90). In fact, stimulation of A₁R by adenosine and other agonists inhibits AVP-induced cAMP formation in cultured rabbit CCD cells (2, 3), as well as rat medullary and IMCD cells (85, 86). Basal, nonstimulated, cAMP formation is reduced in rabbits, rats, and mice by selective A₁R activation (3, 60, 85). Activation of A₁R inhibits AC activity through activation of pertussis toxin-sensitive G_i proteins, as well as through activation of PLC via G_βγ subunits (55) (Fig. 1). The selective A₁R agonist, N⁶-2-phenylethyladenosine (NPEA), mobilized [Ca²⁺]_i in IMCD, a response that was significantly inhibited by the selective A₁R antagonist, 8-phenyltheophylline (8-PT) (19). This would suggest that A₁R, like P2Y₂ receptors, are linked to a signaling pathway capable of mobilizing [Ca²⁺]_i in the IMCD (Fig. 1).

Studies by Yagil (85, 86) showed the dose dependence of AVP-stimulated cAMP in primary cultures of rat IMCD. Interestingly, AVP applied from either side increased cAMP formation, although lower concentrations (100 pM and 1 nM) were more effective when applied from the basolateral side. When adenosine was applied from the basolateral side, 1 μM was sufficient to inhibit AVP-stimulated cAMP formation, whereas 100 μM were necessary to inhibit AVP-stimulated cAMP formation from the apical side. In this regard, it may be relevant that concentrative nucleotide transporters (CNT) expressed in the apical membrane (16, 28) could reduce adenosine availability on the apical surface and increase basolateral adenosine because of passive adenosine efflux via equilibrative nucleoside transporters (ENT1 and ENT2) expressed in the basolateral membrane (16, 28, 63) (Fig. 4).

Fig. 4.

Proposed integrated model for local water transport inhibition by activation of P2Y₂ and adenosine A₁ receptors (A₁R) in inner medullary collecting duct principal cell. Activation of vasopressin V₂ receptors (V₂R) by arginine-vasopressin (AVP) stimulates aquaporin-2 (AQP2)-mediated water flux, which increases cell volume. The latter increases basolateral and apical release of ATP (and potentially UTP?) by unknown mechanism. Basolateral water exit is mediated by constitutively expressed aquaporin-4 (AQP4) and AVP-regulated aquaporin-3 (AQP3) water channels. Urea transporters UT-A1 and UT-A3 play key roles in the urinary concentrating mechanism and fluid homeostasis; both are regulated by AVP. P2Y₂ receptor activation may inhibit water reabsorption by (I) stimulation of volume-regulatory K⁺ channels, and (II) inhibition of protein kinase A (PKA). ATP may also be broken down to adenosine (Ado). Alternative sources of extracellular Ado are the extracellular cAMP-Ado pathway (conversion by extracellular phosphodiesterase, ePDE, and ecto-5′-nucleotidase) or transport mediated by equilibrative nucleoside transporter (ENT). To which extent extracellular ATP is degraded by ectonucleotidases to ADP, AMP, and Ado at the luminal (not shown) and basolateral surface is unclear. Activation of A₁R and P2Y₂ receptors inhibits AVP-stimulated water flux, which helps to normalize cell volume and accelerates the excretion of free water in response to water loading before circulating AVP levels fall. AC, adenylyl cyclase; AQP-3, -4, aquaporin-3, -4; CNT, concentrative nucleoside transporter; COX-1, cyclooxygenase 1; DAG, diacylglycerol; P, phosphorylation. See also text for details.

Using freshly isolated IMCD, we showed that the A₁R agonist N⁶-cyclohexyladenosine (CHA, 1 μM) reduced basal, as well as forskolin-stimulated cAMP formation in WT, but was without effect in A₁R^−/− (60). dDAVP-induced increases in cAMP formation were about twofold greater in A₁R^−/− compared with WT mice, and, in contrast to WT mice, A₁R^−/− mice were not inhibited by CHA. Thus, A₁R^−/− mice show an absence of CHA-mediated inhibition of cAMP formation in IMCD cells, as well as an enhanced response to dDAVP, consistent with the loss of a A₁R-mediated inhibition of this response.

With regard to functional consequences, Dillingham and Anderson (18) showed, using an in vitro CCD perfusion system, a stimulatory effect of high concentrations of adenosine on hydraulic conductivity and net volume flux. In rat IMCD, Edwards and Spielman (20) found that adenosine, acting from the basolateral side, inhibits AVP-induced increases in cAMP formation and P_f.

Little has been known about the net in vivo role of A₁R for water transport in the intact organism. We observed that A₁R^−/− mice have higher urinary flow rates and greater fluid intake compared with WT mice, which may relate to the proposed basal antidiuretic tone of A₁R activation in the proximal tubule (11, 60–62) but argues against a prominent role of A₁R to offset AVP effects on water reabsorption in the CD. Moreover, A₁R^−/− and WT mice had similar basal urinary excretion of AVP and expression of AQP2 protein in renal cortex and medulla. They also increased urinary flow rate and Cl^e-H₂O in response to the V₂R antagonist SR121463 or acute water loading to the same extent. Finally, the dose dependence of dDAVP-induced antidiuresis after acute water loading was not different between genotypes (60). The lack of a clear phenotype of A₁R in AVP control of renal water reabsorption in vivo was surprising considering the very high receptor abundance in the CD system and the demonstrated ability of A₁R to inhibit AVP-induced cAMP formation in isolated IMCD. These findings indicate effective compensation for the absence of A₁R in CD in vivo. This is discussed in more detail below.

Adenosine Generation in CD

In the CD, extracellular adenosine can be generated by different mechanisms. Some are more closely linked to ATP metabolism and, therefore, discussed below in the section on ATP-adenosine interactions. Another mechanism relates to the so-called extracellular cAMP-adenosine pathway (34, 35). The rate of cellular cAMP efflux is proportional to intracellular levels (6, 41) and is increased following stimulation of AC (22). In the CD, AVP-mediated V₂R activation triggers intracellular cAMP synthesis. The latter is released and extracellular cAMP converted to adenosine by the sequential actions of ecto-phosphodiesterase [in the kidney possibly involving PDE8 (37)] and ecto-5′-nucleotidase (36). As illustrated in Figs. 1 and 4, this autocrine/paracrine pathway may constitute a mechanism for feedback inhibition when water transport in the CD is stimulated by AVP-induced cAMP formation (20, 34, 85, 86). Further studies are necessary to define the quantitative contribution of this pathway in the CD, e.g., by studies on AVP-induced cAMP formation in IMCDs of mice lacking PDE8.

Interactions Between the ATP and Adenosine System in the CD

Extracellular adenosine can derive from extracellular breakdown of ATP, AMP, or cAMP. Vekaria et al. (77) provided evidence for the breakdown of exogenous and endogenous ATP in tubular fluid by soluble nucleotidases. In addition, various ecto-ATPases and ecto-5′-nucleotidase are differentially expressed along the tubular and collecting duct epithelia (for a review, see Ref. 75). As a consequence, the cellular release of ATP can inhibit water transport by activation of P2Y₂ receptors and via A₁R after ATP breakdown to adenosine (Figs. 1 and 4). Both pathways converge and suppress AVP-induced cAMP formation and, thus, reduce water transport and potentially cell volume.

Intracellular breakdown of cAMP and ATP can also generate adenosine. Adenosine itself can exit the cell by ENT1 and ENT2 primarily localized to basolateral membranes, where they mediate bidirectional facilitated diffusion of adenosine (16, 28, 63). In comparison, CNT1-CNT3 are mainly localized to the apical membrane, where they mediate unilateral, cellular uptake of nucleosides (Fig. 4) (36). The role of ecto-nucleotidases and of the asymmetric expression of these transport pathways for basolateral vs. luminal effects of ATP and adenosine remain to be determined.

The fact that A₁R^−/− presented no in vivo phenotype after V₂R activation or blockade prompted us to assess under basal conditions and after acute water loading the expression level of several local factors that may compensate. Acute water loading upregulated inner medullary mRNA expression of ET-1 in WT (60), confirming previous studies (26). Notably, CD-specific ET-1 knockout mice have a reduced ability to excrete an acute water load (26). Moreover, acute water loading also increased inner medullary expression of A₁R mRNA in WT (60). Although the mechanism(s) that contribute(s) to the upregulation of A₁R in WT under these conditions remain to be determined, this response could help to facilitate water excretion. Under basal conditions A₁R^−/− had greater inner medullary expression of COX-1, and acute water loading increased P2Y₂ and EP₃ receptor expression more in A₁R^−/− than in WT mice (60), responses that may have compensated for the absence of A₁R and support the functional interaction between these systems. Unpublished observations showed that urinary nitric oxide metabolite excretion (NO_x) was not significantly different between genotypes in response to acute water loading (WT: 8.3 ± 3 vs. A₁R^−/−: 9.2 ± 2 nmol·min⁻¹·g body weight⁻¹), consistent with similar endothelial nitric oxide synthase expression between genotypes under basal conditions or after acute water loading (60). The up-regulation of ET-1 on the mRNA level by water loading was not altered in A₁R^−/−.

Why, with regard to AVP-induced cAMP formation and antidiuresis, do A₁R^−/− mice have a phenotype in isolated IMCD, but not in vivo, whereas P2Y₂^−/−mice show a phenotype in vivo, but not in freshly isolated inner medulla. In experiments in isolated IMCD, application of the V₂R agonist is expected to trigger cAMP formation, perhaps followed by extracellular adenosine formation, activation of A₁R, and inhibition of AC activity, an effect that is blunted in mice lacking A₁R. In the absence of osmotic gradients, however, the in vitro application of the V₂R agonist may not alter cell volume and thus may not generate a stimulus for ATP release. To unmask a defect in isolated IMCD of mice lacking P2Y₂ receptors, we had to apply ATPγS. In vivo, lesser compensation in the absence of P2Y₂ receptors than of A₁R may indicate that the feedback inhibition via cell volume-regulated ATP release and activation of P2Y₂ receptors is a more potent system than the extracellular formation of adenosine (from cellular release of cAMP and extracellular breakdown of ATP) and the subsequent activation of A₁R.

Perspectives and Significance

Multiple P2X, P2Y, and adenosine receptors are expressed in renal epithelial cells, often in the same membrane domain. There is accumulating evidence for an important role for both ATP, via P2Y₂ receptors, and adenosine, via A₁R, to inhibit AVP-stimulated water transport in the CD. Interactions between ATP and the adenosine system support the concept of sequentially organized feedback loops. The complexity of the ATP-adenosine interaction is further enhanced by the existence of receptor heterodimers (1, 24, 40). Heterodimeric associations between P2Y₂ and A₁R were described in transfected cells (88, 89), but their relevance in the kidney is unknown. Important issues to be clarified in future studies also include the identification and regulation of transport pathways involved in ATP release, as well as the variation of ATP and adenosine concentrations close to the receptor under physiological conditions. The role of apical versus basolateral signaling deserves further consideration, including the conversion of ATP to adenosine, and transport of the latter by asymmetrically expressed concentrative and equilibrative nucleoside transporters. The use of gene knockout mice will be helpful, but compensatory changes have to be considered, as shown in our studies using the A₁R^−/−. Finally, a potential role of the ATP and adenosine system in the pathophysiology of CD water transport needs to be explored.

GRANTS

This research was supported by the National Institutes of Health Grants DK56248, DK28602, and GM66232, the O'Brien Kidney Center (Grant P30DK079337), the American Heart Association (Grant 0655232Y), and the Research Service of the Department of Veterans Affairs (all to V. Vallon). T. Rieg was supported by the German Research Foundation (RI 1535/3-1 and 3-2) and a National Kidney Foundation Fellowship.