Role of Aquaporin-2 Water Channels in Urinary Concentration and Dilution Defects

The following article is presented with the permission of the publisher for educational use within the NDI community. No part of this article may be reproduced in any way without permission in writing from the publisher.

Urinary concentration defects are observed in many clinical situations. The concentrating defect can be extreme and be associated with serious clinical consequences, such as in central diabetes insipidus. Alternatively, the concentration defect may be mild as occurs in the nephrotic syndrome. Whatever its severity, this concentrating defect reflects a perturbation in renal solute-free water handling. Arginine vasopressin (AVP), or antidiuretic hormone, plays a central role in the regulation of water balance. Recently, the action of AVP on the permeability of the apical membrane of the principal cells in the collecting duct has been demonstrated to be mediated by a water channel protein, named originally aquaporin-CD (that is, collecting duct), and now aquaporin-2 (AQP-2). The cloning of this water channel and the studies demonstrating its functional capacity in response to vasopressin have been hallmarks in the understanding of the water regulation by the kidneys. This review will first briefly describe the family of aquaporins. The major mechanisms of regulation of water balance by the AVP-mediated AQP-2 pathway and the different clinical situations wherein alterations in AQP-2 occur will then be discussed.

THE AQUAPORINS

The aquaporin family is a group of selective water channels, which is a part of the larger family of MIP intrinsic membrane proteins. To date, eight different aquaporins have been cloned and aquaporins-1 through -4 have been identified within the kidney and play a role in the water reabsorption of the glomerular filtrate along the nephron. Aquaporin-1 (AQP-1) has been localized to both the apical and basolateral membranes of the proximal tubules and descending limb of Henle, nephron segments that are constitutively water-permeable [1]. It is postulated to be the most important water channel in these segments, but not the only one, since individuals who lack functional AQP1 do not have clinical disturbances of water conservation [2]. This interesting observation raised the possibility of other membrane proteins able to transport water. The recent evidence of water transport by the glucose cotransporters, which are abundant in the proximal tubule may be an alternative way for the tubules to transport water [3]. AQP3 and AQP4 are localized to the basolateral membrane of principal cells of the collecting ducts, AQP3 being located predominantly in the cortical and outer medulla, whereas AQP4 predominates in the inner medullary collecting duct (IMCD) [4, 5]. These water channels are constantly expressed on the membranes, thus providing a pathway for water movement from the principal cells to the hypertonic medullary interstitium. The expression of AQP3, but not AQP4, has also been shown to be modulated by AVP [6]. The clinical importance of these two water channels in the renal water conservation has yet to be determined. There is [sic] no known natural mutations of AQP3 and AQP4. Recently, a transgenic mice [sic] deficient in AQP4 has been generated [7]. Interestingly, the phenotype abnormalities were very subtle, with a small defect in maximal urinary concentration, consistent with the predominant localization of AQP4 in the lower portion of the IMCD.

ARGININE VASOPRESSIN-AQUAPORIN-2 PATHWAY

Arginine vasopressin release

Arginine vasopressin is released into the systemic circulation from the neurohypophysis, an extension of the ventral hypothalamus located in the dorsal and caudal surfaces of the adenohypophysis. The neurohypophysis is mainly composed of a capillary network and neurosecretory axons that originate in the magnocellular neurons of the supraoptic and paraventricular nuclei of the hypothalamus. AVP is synthesized as a preprohormone composed of a signal peptide AVP, neurophysin and a glycoprotein. The preprohormone is synthesized in neuronal bodies of the supraoptic and paraventricular neuclei, packaged in secretory granules, and transported down the axons. During this process, AVP and neurophysin are removed from the preprohormone by proteolytic cleavage and stored in the nerve terminals in the neurohypophysis until secreted by a calcium-dependent exocytic process.

Renal action of arginine vasopressin

The cellular actions of AVP are mediated by at least three types of G-protein-coupled receptors: the V1a (vascular and hepatic), the V1b (anterior pituitary) and the V2, which is the collecting duct receptor responsible for the renal action of AVP on water excretion. The V2 receptors are located on the basolateral membrane of the collecting duct cells. AQP2 is a selective, mercurial-sensitive, water channel cloned and characterized in rats [8] and in humans [9]. This water channel is expressed almost exclusively within the kidney principal cells of the collecting duct. A quantitative analysis using a single tubule ELISA method demonstrated that AQP2 was extremely abundant throughout the collecting duct, thus accounting for the rapid osmotic water transport measured in these nephron segments [10]. The connecting tubules from the renal arcade segment have also been demonstrated to contain AQP2 in the rat [11]. When inserted in the apical membrane, AQP2 allows the selective passage of water, which then crosses the basolateral membrane through the constitutively present AQP3 and AQP4 water channels. The hypertonic interstitium provides the osmotic driving forces. AQP2, like the other aquaporins, is a six membrane-spanning region with both the amino-terminal and the carboxy terminal tails located intracellularly. The three-dimensional conformation of AQP2 is not known but different models have been provided. The most attractive being the hourglass model proposed for AQP1 in which two hydrophilic loops containing Asn-Prol-Ala sequences are folded into the lipid bilayer, forming a single aqueous pore [12].

Short term action. The binding of AVP activates 3',5' cyclic adenosine monophosphate (cAMP) formation, via the stimulatory Gsa protein, which then activates a process whereby the cytosolic AQP2 moves to the apical membrane, thereby increasing the water permeability. There is now a strong body of evidence that the plasma membrane expression of AQP2 in principal cells is mediated by AVP [13-15]. The increased apical membrane content of AQP2 occurs via the exocytotic insertion of transport vesicles enriched in AQP2 protein that contains the necessary information to target the membrane [16]. It has been also shown in LLC-PK₁ cells that this vasopressin-induced trafficking of AQP2 was dependent upon PKA-induced phosphorylation rather than the water permeability of AQP2 molecule [17]. This may provide an explanation for the controversial results concerning the role of phosphorylation in the action of AVP [18, 19]. Decreased plasma AVP concentration down-regulates the AQP2 membrane expression by promoting an endocytic retrieval of the transport vesicle enriched in AQP2, thus decreasing the water permeability of the apical membrane. This translocation from the apical membrane to the cytoplasmic vesicles and vice versa is microtubule-dependent [20]. This trafficking-regulated pathway is the basis of the ?shuttle hypothesis,? it is dependent on the short term action of AVP, is fast and is rapidly reversible [21]. This shuttle hypothesis has been supported by the recycling of the AQP2 proteins which has been very nicely demonstrated in vitro in LLC-PK₁ cells transfected with AQP2 [22].

Long term action. In addition to this short term action, the abundance of AQP2 can also be regulated, thus influencing the basal water permeability as well as the magnitude of the permeability effect following exocytosis or endocytosis of AQP2 [14, 23]. For instance, during chronic fluid deprivation the collecting duct can attain a higher water permeability because of the larger number of AQP2 available to be translocated to the apical membrane. An increased amount of AQP2 mRNA and protein occurs in this setting [24]. In the collecting ducts of the rat model of central insipidus diabetes (Brattelboro rats), the lack of AVP is associated with a very low basal water permeability, a low content of AQP2 and a moderate response to acute administration of AVP. Chronic administration of AVP (7 days), however, increases the basal and maximal permeability of the Brattelboro collecting ducts, and this is associated with an increase in AQP2 protein content [23]. This long term effect is therefore mediated by AVP and implies new synthesis of AQP2 protein and mRNA [24]. The molecular basis of this long-term regulation by AVP is explained by the presence in the promoter region of the human AQP2 gene of a cAMP response element (CRE), which promotes the transcription of AQP2 mRNA [9]. Moreover, it has been recently shown that cAMP activation by the V2 receptor has a dual effect on CRE and the immediate early gene expression, c-Fos/c-Jun, which activates the AP1 promoter site [25]. AVP-independent regulation of AQP2 expression has also to be considered, and oxytocin has been demonstrated to up-regulate the expression of AQP2 via the V2 receptor [26]. In addition, inner medullary tonicity may also contribute to this long term regulation of AQP2 as suggested by the results of in vivo studies [27].

Urinary aquaporin-2

AQP2 protein can be measured in the urine either by radioimmunoassay [28] or by Western blot [29] and appears to be a good index of the action of AVP on the collecting duct. With respect to the "shuttle hypothesis" and the recycling of AQP2 protein, this finding means that a percentage of the AQP2 protein anchored in the membrane is not recycled, but lost in the lumen and then washed out by the urine. Thus, the more AQP2 present within the membrane, the more AQP2 that is lost in the urine. Thus, urinary AQP2 provides a potential biologic marker of AVP action in humans. The studies measuring urinary AQP2 will be discussed below depending on the pathophysiological situation.

CONCENTRATING DEFECTS

Hereditary

Central diabetes insipidus. In the Brattleboro rats as well as in humans with familial autosomal-dominant central diabetes insipidus, the inability to release AVP is due to a mutated peptide that cannot be processed at the level of the endoplasmic reticulum into the neurosecretory vesicles. The majority of the mutations, but not all, have been shown to reside primarily in the neurophysin coding region [30]. The importance of neurophysin in the processing and secretion of AVP has been recently demonstrated in COS cells tranfected [sic] with different constructs of the AVP gene precursor [31].

Absence of circulating AVP is associated in Brattelboro rats with a very low content of AQP2 protein in the collecting duct cytosol, and AVP administration significantly increases this AQP2 content [23]. The radioimmunoassay of urinary excretion of AQP2 was very low during water restriction in five patients with central diabetes insipidus and was increased five times by the injection of the V2 analog desmopressin (DDAVP) [28]. Moreover, the urinary AQP2 response to a hypertonic saline infusion test and to AVP infusion has been shown to be a useful tool for the diagnosis of central diabetes insipidus [32].

Nephrogenic diabetes insipidus. In most families, nephrogenic diabetes insipidus is caused by a mutation in the gene for the vasopressin V2 receptor on the chromosome X (X-linked recessive), but approximately 10% of the families show an autosomal recessive pattern of inheritance. Among this autosomal recessive nephrogenic diabetes insipidus, an AQP2 mutation with intact V2 receptor was shown recently to account for the insensitivity of the collecting duct to the action of AVP [33]. This observation in humans established the link between AQP2 and AVP in the regulation of water renal excretion. Since then, additional cases of AQP2 mutations have been reported. It has been demonstrated in vitro that at least one cause underlying the defect in water permeability is a misrouting of the AQP2 mutant protein, rather than a dysfunction of the water channel [34]. By expressing six missense AQP2 proteins in Xenopus oocytes, the authors demonstrated nicely that the transport to the plasma membrane was impaired, the AQP2 mutant proteins apparently being retained in the endoplasmic reticulum.

Therefore, based on these observations, it appears that two out of three forms of inherited diabetes insipidus are caused by a gene mutation altering the protein trafficking.

Acquired nephrogenic diabetes insipidus

Table 1. Clinical situations where upregulation of aquaporin-2 (AQP2) has been demonstrated Syndrome of inappropriate secretion of antidiuretic hormone (SIADH) Chronic cardiac failure Cirrhosis Pregnancy Table 2. Clinical situations where downregulation of aquaporin-2 (AQP2) participates in the urinary concentrating defect Autosomal-recessive nephrogenic diabetes insipidus Hypokalemia Lithium Nephrotic syndrome Unilateral and bilateral ureteral obstructions Low proteins diet Hypercalcemia

Hypokalemia. Potassium depletion is commonly associated with a mild to moderate impairment of urinary concentrating ability (Table 1). It was known that the accompanying polyuria was AVP-resistant, but the mechanism was unclear. It was postulated that the cAMP sensitivity to AVP was decreased, while in the toad bladder it was [sic] been demonstrated that a defect distal to cAMP might occur [35]. Down-regulation of AQP2 in potassium-depleted rats provided a further insight into the cause of the collecting duct resistance to AVP [36]. The decreased expression of AQP2 in the hypokalemic rats correlated with the lower urinary osmolality and was similar in both the cortex and the medulla, thus suggesting that the AQP2 down-regulation was, at least partially, independent of the medullary tonicity. The cortical effect was important to demonstrate because of the known inhibitory effect of hypokalemia on the NaCl reabsorption of the thick ascending limb (TAL), which would decrease medullary, but not cortical, tonicity. In addition, furosemide was administered in another group of control rats to decrease the medullary gradient [35]. Furosemide, although decreasing urine osmolality did not down-regulate AQP2 expression, thus further suggesting that medullary tonicity is not involved in the down-regulation of AQP2. Repletion of potassium corrected the AQP2 down-regulation and re-established a normal urinary concentration (Fig. 1).

Lithium. Polyuria is a frequent side effect in bipolar depressive patients treated with lithium.

Fig 1. Aquaporin-2(AQP2) protein expression, as illustrated by densitometry measurements in control animals (100%), after 11 days of low potassium diet, and after a 7-day recovery period on a normal diet. Hypokalemia induces a substantial decrease in AQP2 protein expression which recover completely within one week of normal potassium diet. Adapted with permission from Marples et al [36].

This lithium-induced polyuria is not responsive to the administration of vasopressin. Although not all patients are polyuric, the majority of the patients treated long-term with lithium have a persistent impairment of their maximal urinary concentrating capacity. Among the different hypotheses tested to account for this defect, it was proposed more than twenty years ago that the effect of lithium was post cyclic AMP [37]. In a rat model of chronic lithium treatment, a dramatic down-regulation of AQP2, associated with a severe urinary concentration defect, has recently been demonstrated [38]. This down-regulation was observed after 10 days of lithium treatment and was more marked after 35 and 45 days. As with hypokalemia, the AQP2 down-regulation presented a similar pattern in both cortical and medullary principal cells. Water deprivation increased expression of AQP2 in association with an increase in urinary osmolality, but DDAVP administration did not, thus suggesting that other factors than AVP can also regulate AQP2 expression during fluid deprivation. Nevertheless, DDAVP increased urinary osmolality, thus showing an intact short term action of AVP on AQP2 (that is, translocation from the cytosol to the apical membrane). Cessation of lithium only partially reestablished the urine concentrating capacity, thus suggesting a prolonged inhibitory effect of lithium on AQP2.

Hypercalcemia. Hypercalcemia and hypercalciuria produce a renal concentrating defect, which is resistant to AVP. The mechanism of this cellular resistance to AVP is not known but may be due to luminal rather than extracellular calcium concentration. Recently, a microperfusion study demonstrated a role for the apical extracellular calcium sensing receptor (CaSR), which has already been incriminated in the NaCl and calcium reabsorption of the thick ascending limb (TAL) [39] and in the regulation of water permeability of the IMCD [40]. This CaSR specifically reduces AVP-elicited osmotic water permeability when luminal calcium increases. Teleologically, this may be to prevent a potential risk of precipitation of calcium oxalate or phosphate. The CaSR protein is located on the apical membrane of the principal cells within the same endosomes, which contain AQP2 proteins, as well as stimulatory and inhibitory GTP binding proteins, and two isoforms of protein kinase C (PKC) [40]. The authors postulate the existence of a unique apical IMCD membrane signaling mechanism linking calcium and water metabolism, thus providing a first line control of alterations in luminal calcium concentrations. Thus, hypercalciuria may downregulate AQP2 expression or interfere with the AQP2 trafficking. The in vivo demonstration of a down-regulation of AQP2 during hypercalcemia and/or hypercalciuria, which could account for the nephrogenic diabetes insipidus during hypercalcemia, has not yet been reported. Finally, as already stated in the hypokalemia-induced down-regulation, the effect of hypercalcemia along the nephron, such as the decreased NaCl reabsorption in TAL through its basolateral CaSR, may also participate in the urinary concentrating defect by decreasing the medullary osmotic driving force for passive water movement.

Low protein diet. Decreased protein intake diminishes urine concentrating capacity and has been shown to be partially due to an increase in urea reabsorption in the initial IMCD, thus resulting in a decrease urea delivery in the deeper IMCD. The resultant decrease in hypertonicity of the deep medullary interstitium may therefore decrease the osmotic driving force to reabsorb water. In addition, it has been shown that the terminal IMCD of low-protein fed rats have a localized down-regulation of AQP2, which is not present in the other part of the nephron [27]. This mechanism could therefore contribute to the urinary concentrating defect associated with a low protein diet. The persistent integrity of the cortical collecting duct and the initial IMCD may explain why the defect is mild.

Nephrotic syndrome. Defects in both diluting and concentrating capacity have been documented in nephrotic syndrome. Whereas the diluting ability defect has been explained by a non-osmotic release of AVP [41], the concentrating defect has not been elucidated. Using the puromycin aminonucleoside (PAN)-induced nephrotic syndrome model in rats, Apostol et al [sic] found an extensive down-regulation of AQP2 and AQP3 water channels in IMCD [42]. Interestingly, this down-regulation was observed in the presence [of] normal plasma sodium and elevated plasma AVP concentrations, thus suggesting an unresponsiveness of the IMCD cells to the action of AVP. In addition, the countercurrent multiplier mechanism was assessed by the simultaneaous [sic] measurement of inner medullary tissue osmolality. Although decreased in PAN-treated rats, the inner medullary tissue osmolality was not as profoundly decreased as the maximal urinary osmolality, thus implicating a failure of complete osmotic equilibration across the IMCD, compatible with the AQP2 down-regulation [42].

Bilateral and unilateral ureteral obstruction. Polyuria is common after reversal of urinary tract obstruction and the associated defect in urine concentrating capacity can last for several days. A down-regulation of AQP2 has been demontrated [sic] in rat models of both bilateral and unilateral ureteral obstructions. The unilateral ureteral obstruction [43] was studied to exclude a role in the solute diuresis inherent to the relief of bilateral ureteral obstruction [44]. These unilateral experiments demonstrated that intrarenal factors, such as intrarenal pressure or prostaglandin production, may play a role in this AQP2 down-regulation. Interestingly, a mild down-regulation of AQP2 was also present in the contralateral unobstructed kidney, thus suggesting that a circulating factor may also contribute to the adaptative mechanism.

Vasopressin escape. In situations where AVP is inappropriately high relative to plasma osmolality such as the syndrome of inappropriate antidiuretic hormone secretion (SIADH), chronic heart failure (CHF) and cirrhosis, the degree of hypoosmolality is limited, and a counterregulatory mechanism has been proposed and termed vasopressin escape. One explanation for this mechanism has been proposed through a down-regulation of the AVP-elicited enhanced AQP2 expression [45]. Rats were infused with constant doses of AVP, thus mimicking a situation of non-osmotic release of AVP, and then subjected to a liquid diet. The onset of vasopressin escape, indicated by an increase in urine flow, was observed by day 2 of water loading and coincided with a decrease in the AVP-induced AQP2 expression. This escape phenomenon was not associated with a change in the high membrane/cytosol ratio of AQP2 elicited by AVP, suggesting that the escape phenomenon may be a counterregulatory mechanism of the action of AVP on AQP2 synthesis, but not on AQP2 trafficking. It should be emphasized, however, that the plasma sodium necessary to stimulate this escape mechanism was lower than the majority of hypoosmolality encountered in clinical situations, such as cirrhosis and chronic heart failure (that is, < 110 mmol/liter).

Increased AQP2 expression

Syndrome of inappropriate secretion of antidiuretic hormone (SIADH). Chronic administration of AVP in rats ingesting a liquid diet, a model of SIADH, elicits an increase expression of AQP2 which mediates the increased solute-free water reabsorption responsible for the hypoosmolality. The increase in AQP2 mRNA appears very rapidly (less than 6 hr after the starting dose) and is followed by an increase in AQP2 protein content. This increase is reversed by the blockade of AVP action [46].

Chronic heart failure (CHF). CHF is a situation involving renal water and sodium retention. Pretreatment hyponatremia is common in severe CHF and has been shown to be an ominous prognostic factor. Non-osmotic release of AVP is the cause of the hypoosmolality and administration of a V2 receptor antagonists [sic] corrects the hypoosmolality in rats. Moreover, administration of a V2 receptor antagonist improved the long-term mortality of chronic heart failure rats [47].

Fig 2. Aquaporin-2 (AQP2) protein expression in inner medullary collecting duct (IMCD) of control rats (lanes 1 to 2), chronic heart failure rats (CHF; lanes 3 to 4), and CHF rats treated with OPC 31260 (lanes 5 and 6). Western blots hybridization shows a band at 29 kDa, corresponding to the molecular weight of AQP2. A wide band (35 to 45 kDa) represents the glycosylated AQP2 protein. CHF rats expressed a significant increase in AQP2 protein (1.85-fold as compared to the control rats) which was completely reversed by the administration of OPC 31260, a V2 vasopressin receptor antagonist. Similar results were found with AQP2 mRNA analysis. Adapted with permission from Xu et al [50].

It has also been shown recently that V2 receptor antagonists were able to dramatically initiate a diuresis in CHF patients and that it was associated with a normalization of the decreased plasma osmolality [48]. Expression of AQP2 has been shown to be up-regulated in the coronary-ligation rat model of low cardiac output CHF [49, 50]. The increase was observed in AQP2 mRNA and protein expression and was associated with an increased translocation of AQP2 to the membrane, as illustrated by a high membrane/cytosol ratio. This trafficking and upregulation was abolished by the administration of a V2 receptor antagonist to CHF rats and was accompanied by an increased diuresis and a normalization of the hyponatremia (Fig. 2) [50]. In patients with CHF, administration of a V2 receptor antagonist is also associated with a dramatic decrease in the urinary AQP2 expression which correlates with the solute-free water diuresis [51].

Cirrhosis. Renal salt and water retention is also a hallmark of advanced cirrhosis with ascites. Non-osmotic release of AVP has been demonstrated in cirrhosis and accounts for the high incidence of hyponatremia observed in hospitalized cirrhotic patients with ascites. This non-osmotic release of AVP is also observed in the CCl4-induced cirrhosis rat model and is related to the hyperdynamic circulation elicited by the peripheral arterial vasodilation [52]. AQP2 has been demonstrated to be increased in cirrhotic rats with ascites [46, 53] and this increase was reversed by the administration of a V2 receptor antagonist [46]. No immunocytochemical localization was however performed in these two studies to investigate an effect on AQP2 trafficking.

Physiological alterations of AQP2

Newborn. Newborn humans and most mammals cannot concentrate their urine and there is no correlation between urine osmolality and plasma AVP concentrations, which are usually high in the postnatal period. Using the rat model it has been demonstrated that newborn rats have a very low expression of AQP2 and that AQP2 increases from infancy to adulthood, thus paralleling the increase in urine concentrating capacity [54]. Administration of glucocorticoid hormones significantly increased AQP2 mRNA and protein expression as well as urinary concentrating capacity in infant rats, whereas it had no effect on AQP2 expression in adult rats. It should be emphasized that the increase in mRNA was detectable only after 24 hours of hormone administration, in contrast to the rapid effect of AVP on the AQP2 mRNA of adult rats [55]; thus, there may be an intermediary step for the action of the glucocorticoids on AQP2 mRNA. In any case, these findings demonstrate that AQP2 plays a major role in the concentrating defect of infancy and that glucocorticoids promote their expression.

Pregnancy. Pregnancy is characterized by an increase in body fluid volume. Expression of AQP2 protein and mRNA has been shown to be elevated as early as the end of the first trimester of pregnancy in rats and appears to contribute to the water retention of pregnancy [56]. Indeed, although plasma AVP concentration in pregnant rats was not significantly different from non-pregnant rats, there was a significant increase in AQP2 which coincided with the occurrence of hypoosmolality. The same study demonstrated that this increase in AQP2 was reversed by V2 receptor antagonism, thus implicating a V2-mediated effect on AQP2 synthesis. Since oxytocin can also activate the V2 receptor, a role of this hormone cannot be excluded.

CONCLUSION

The recent discovery of AQP2 water channel has provided a new tool to unveil the physiological and pathophysiological mechanisms of water homeostasis as regulated by AVP. The dual short and long term regulation by AVP