Physiological Reviews

Mouse Models and the Urinary Concentrating Mechanism in the New Millennium

Robert A. Fenton, Mark A. Knepper


Our understanding of urinary concentrating and diluting mechanisms at the end of the 20th century was based largely on data from renal micropuncture studies, isolated perfused tubule studies, tissue analysis studies and anatomical studies, combined with mathematical modeling. Despite extensive data, several key questions remained to be answered. With the advent of the 21st century, a new approach, transgenic and knockout mouse technology, is providing critical new information about urinary concentrating processes. The central goal of this review is to summarize findings in transgenic and knockout mice pertinent to our understanding of the urinary concentrating mechanism, focusing chiefly on mice in which expression of specific renal transporters or receptors has been deleted. These include the major renal water channels (aquaporins), urea transporters, ion transporters and channels (NHE3, NKCC2, NCC, ENaC, ROMK, ClC-K1), G protein-coupled receptors (type 2 vasopressin receptor, prostaglandin receptors, endothelin receptors, angiotensin II receptors), and signaling molecules. These studies shed new light on several key questions concerning the urinary concentrating mechanism including: 1) elucidation of the role of water absorption from the descending limb of Henle in countercurrent multiplication, 2) an evaluation of the feasibility of the passive model of Kokko-Rector and Stephenson, 3) explication of the role of inner medullary collecting duct urea transport in water conservation, 4) an evaluation of the role of tubuloglomerular feedback in maintenance of appropriate distal delivery rates for effective regulation of urinary water excretion, and 5) elucidation of the importance of water reabsorption in the connecting tubule versus the collecting duct for maintenance of water balance.


The mammalian kidney tightly controls systemic water balance by varying water excretion over a broad range, chiefly in response to changes in the circulating concentration of the antidiuretic hormone vasopressin (70). These broad adjustments in water excretion can be achieved without major effects on solute excretion because of the kidney's ability to dilute urinary solutes below plasma osmolality and to concentrate solutes in the urine to much greater than plasma osmolality (Fig. 1). The diluting mechanism is well understood as being chiefly the function of the cortical thick ascending limb (CTAL) of Henle's loop, sometimes called the diluting segment. The CTAL actively transports NaCl and NH4Cl from the lumen to the peritubular interstitial space, while (owing to its low water permeability) leaving water behind. In contrast, the urinary concentrating process is much more complex. The central goal of this review is to summarize the new knowledge that has accrued about the urinary concentrating process as a result of the development and application of gene deletion techniques, both conventional gene knockouts and cell specific gene knockouts. However, before addressing these mouse models, we provide a brief, basic summary of our current knowledge about the urinary concentrating mechanism prior to the advent of knockout mouse technology, with emphasis on extant controversies.

FIG. 1.

Steady-state renal response to varying rates of vasopressin infusion in conscious rats (70). A water load (4% of body weight) was maintained throughout the experiments to suppress endogenous vasopressin secretion. Although the urine flow rate was markedly reduced at higher vasopressin infusion rates, the osmolar clearance changed little. [Data adapted from Atherton et al. (6).]

A pedagogically useful, but oversimplified, starting point is the idea that the concentrating process can be divided into two separate elements: 1) transport processes residing in Henle's loops that concentrate solutes in the medullary interstitium and 2) vasopressin-regulated water transport in the collecting ducts, which allows the tubular lumen contents to equilibrate with the hypertonic medullary interstitium. Although providing a useful starting point, the reader should be mindful that in reality these two processes are not truly separable, particularly with regard to urea, the major solute responsible for waste nitrogen elimination in mammals (see below).

A. Transport Processes in Henle's Loops That Concentrate Solutes in the Medullary Interstitium

It is generally accepted that NaCl is concentrated in the renal medullary interstitium through the process of countercurrent multiplication, originally proposed by Wirz, Kuhn, and colleagues more than 50 years ago (139, 282). This process requires special properties in both the ascending limb and descending limb of Henle (124). The ascending limb in the outer medulla (MTAL) is like the CTAL (described above); it actively transports NaCl, but water remains in the lumen owing to a lack of aquaporin expression and thus low water permeability (186). This dilutional effect generates a small transepithelial osmolality difference (the “single effect” or Einzeleffekt) that is “multiplied” by the counterflow between the two limbs of Henle's loops, resulting in an axially aligned osmotic gradient that is much larger than the single effect (see Ref. 70 for details). The properties of the descending limb of Henle are critical because, for countercurrent multiplication to work effectively, the luminal fluid must be close to osmotic equilibrium with the surrounding interstitium at every point along the descending limb. How this osmotic equilibration occurs is one point of controversy (see Fig. 2). Theoretically, the equilibration could occur by rapid water efflux from the descending limb, by rapid solute entry into the descending limb, or by a combination of both processes. Studies of isolated perfused thin descending limbs from the outer medulla revealed that they have high water permeabilities (43, 130), which are thought to be due to the presence of high levels of aquaporin-1 in the apical and basolateral plasma membranes of thin descending limb cells (188). However, micropuncture studies addressing the mechanism of osmotic equilibration in the descending limb concluded that a substantial element of osmotic equilibration is due to solute entry (106). This controversy has been addressed using aquaporin-1 (AQP-1) knockout mice (see below).

FIG. 2.

Two modes of countercurrent multiplication. Figure depicts two loops of Henle showing NaCl recycling mode (left) and water short-circuiting mode (right). NaCl recycling mode concentrates NaCl in medulla by increasing the mean residence time of Na+ and Cl in medulla. Water short-circuiting mode (right) concentrates the medulla by reducing the mean residence time for water molecules in medulla. Water short-circuiting mode concentrates all solutes present in input to descending limb of loop.

The countercurrent multiplier mechanism described in the previous paragraph only functions in the renal outer medulla and medullary rays of the cortex. In contrast, the inner medulla lacks a thick ascending limb of Henle and the ascending limb of Henle's loop in the inner medulla is of “thin-limb” morphology. Repeated studies of the inner medullary thin limb have not been able to uncover evidence of net active NaCl transport (97, 124, 132). However, a substantial axial osmolality gradient has been demonstrated in the inner medulla of a variety of mammalian species. A major point of controversy has been regarding the question of how the inner medullary NaCl gradient is generated. Three hypotheses have been presented to address this question: 1) the “passive” hypothesis of Kokko and Rector (131) and Stephenson (246), in which energy for concentration of solutes (chiefly NaCl) in the inner medullary interstitium is derived from passive urea efflux from the inner medullary collecting duct (IMCD), which concentrates NaCl in the descending limb lumen via water efflux from the thin descending limb (225); 2) the “lactate” hypothesis of Thomas and co-workers (89, 261), in which lactate generation from glucose by glycolysis in inner medullary cells results in net generation of osmotically active particles (one glucose yields two lactates); and 3) the “mechanico-osmotic induction hypothesis” (124, 211), in which energy from the peristaltic contractions of the renal pelvic wall is stored by compression of hyaluronan in the inner medullary interstitium and that this energy is used to concentrate solutes in the descending limbs and collecting ducts by water withdrawal when the hyaluronan springs back to its extended state after passage of the peristaltic wave. Recent results in mice with genetic deletion of the IMCD urea transporters, UT-A1 and UT-A3, have addressed the first hypothesis (see below).

B. Osmotic Equilibration of Collecting Duct Fluid by Vasopressin-Regulated Water Transport

It is generally accepted that vasopressin regulates water permeability in the renal collecting duct system by stimulating the insertion of aquaporin-2 (AQP-2) water channels into the apical plasma membrane and by regulating AQP-2 gene expression (186). Furthermore, there has been evidence in the older physiological literature that the kidney may be capable of urine concentration in the absence of vasopressin's actions in the renal collecting duct (21), which, if true, raises hopes for more effective treatment of patients with X-linked nephrogenic diabetes insipidus (NDI). The development of mice in which the renal vasopressin V2 receptor (V2R) has been deleted have begun to address this possibility (see below). Another point of controversy derived from the “micropuncture era” (105) has been the relative importance of water transport in the connecting tubule and various subsegments of the renal collecting duct in the urinary concentrating mechanism, with regard to renal water conservation. Mouse models in which the V2R or AQP-2 is selectively deleted from the collecting ducts and not the connecting tubule are addressing this issue (see below).

C. Urea Recycling

Urea accumulation in the renal inner medulla depends on passive exit of urea from the IMCDs to the inner medullary interstitium. The retention of urea in the inner medullary interstitium depends on recycling processes, which return urea that would otherwise be lost to the general circulation to the inner medulla. One recycling pathway is classical countercurrent exchange, in which urea absorbed from the ascending vasa recta in the inner medulla enters the descending vasa recta (22). Additional recycling pathways have been subsequently proposed (Fig. 3) (123): 1) direct transfer of urea to the long loops of Henle in the inner medulla, 2) uptake of urea in the ascending vasa recta with transfer to the descending limbs of short loops of Henle in the vascular bundles of the renal outer medulla, and 3) transfer of urea from the thick ascending limbs of the outer stripe of the outer medulla and cortical medullary rays to the neighboring proximal straight tubules. Some aspects of these recycling pathways have been addressed recently in mice in which genes encoding the urea transporters UT-A2 and UT-B have been deleted (see below).

FIG. 3.

Pathways of urea recycling in renal medulla. Solid blue lines represent a short-looped nephron (left) and a long-looped nephron (right). Transfer of urea between nephron segments is indicated by dashed red arrows labeled a, b, and c corresponding to recycling pathways described in the text. tAL, thin ascending limb; CD, collecting duct; DCT, distal convoluted tubule; DL, descending limb; PST, proximal straight tubule; TAL, thick ascending limb; vr, vasa recta. [Adapted from Knepper and Roch-Ramel (123).]


A. Transport Proteins Involved in Urinary Concentration and Dilution

Our understanding of renal physiology has accelerated in recent years as a result of the advent of the era of “molecular physiology,” characterized by the development and application of tools to study proteins that can mediate renal function. These techniques have led to the cloning of multiple cDNAs and genes that are thought to be involved in urinary concentration and dilution. Figure 4 summarizes the renal tubule sites with abundant expression of aquaporins, urea transporters, and ion transporters/channels that are important to the urinary concentrating process. Several of these transporters are molecular targets for vasopressin action and are also expressed in low abundance in other renal tubule segments (Fig. 5).

FIG. 4.

Major aquaporins, urea transporters, and ion transporters/channels that are important to the urinary concentrating and diluting process. Figure depicts a schematic overview of a mammalian kidney tubule, showing the solute and water transport pathways in the proximal tubule (PT), thin descending limb of Henle's loop (tDL), thick ascending limb (TAL), distal convoluted tubule (DCT), cortical collecting duct (CCD), and inner medullary collecting duct (IMCD). Tubule lumen side is always on the left side of the cell, whereas the interstitium is on the right side. Arrows represent direction of movement.

FIG. 5.

Grid showing sites of expression of water channels, urea transporters, and ion transporters important to the urinary concentrating process and their regulation by vasopressin.

Thus now that the proteins thought to be responsible for the urinary concentrating mechanism have been identified (Fig. 4) and we are in the “gene-knockout era,” several important questions remain to be answered in renal physiology. This review summarizes the features/phenotype of reported mouse models that are relevant to the urinary concentrating mechanism with emphasis on the controversies described above. These models can be divided into two groups: 1) those in which water balance appears to be selectively compromised as a result of disruption of expression of proteins known or suspected to be involved in the urinary concentrating mechanism, and 2) those that result in water balance abnormalities that are due in part to a destructive effect of the gene deletion on renal structure or that are associated with broader systemic effects.

B. Sodium Transporters and Channels

1. NHE3

The Na+-H+ exchanger (NHE3) is the major Na absorptive pathway in the proximal tubule. In addition to the proximal tubule, NHE3 has also been immunolocalized to the outer medullary thin descending limb of Henle and the TAL (219, 220). In the MTAL, NHE3 activity is increased by hypotonicity via a phosphatidylinositol 3-kinase (PI-3-K)-dependent pathway, which is inhibited by vasopressin working though cAMP (81, 82). Thus vasopressin has the net effect to inhibit NHE3 activity in the MTAL.

NHE3 knockout mice have been developed and are viable (233) (in contrast to NKCC2 knockout mice, see below). Their renal phenotype is predominantly associated with the fact that NHE3 is the major Na entry pathway in the proximal tubule (Fig. 4). NHE3 null mice have a marked reduction in proximal tubule fluid absorption and a compensatory decrease in glomerular filtration rate (GFR). This decrease in GFR is the result of activation of the tubuloglomerular feedback (TGF) mechanism (153).

Metabolic cage studies revealed that with free access to water, NHE3 null mice manifest a moderate increase in water intake associated with lower urinary osmolalities (average of 1,737 mosmol/kgH2O), although maximal urinary osmolality was not evaluated (4). In addition, NHE3 knockout mice have a marked decrease in renal NKCC2 expression, despite elevated plasma vasopressin levels (36). In conclusion, NHE3 null mice have a mild urinary concentrating defect that may be associated with a reduction in NKCC2 expression.

2. NKCC2

The renal Na+-K+-2Cl cotransporter (NKCC2; also known as “BSC1”) is expressed in the TAL, where it is localized to the apical membrane of epithelial cells (110, 187). NKCC2 is also expressed in the macula densa (187). In rat, long-term increases in the circulating vasopressin concentration result in an increase in NKCC2 protein abundance in the TAL (118), which is associated with an increase in the maximal urinary concentrating capacity (119). In addition, NaCl absorption in the MTAL can be increased by acute vasopressin administration (86). This is thought to occur, in part, due to increased apical membrane expression of NKCC2 in association with phosphorylation of the NH2-terminal tail (76, 198).

NKCC2 knockout mice have been developed by standard gene targeting techniques (252). However, due to perinatal fluid wasting and dehydration, the animals are not viable and die prior to weaning. Treatment of these mice with indomethacin and the administration of fluid allowed some mice to survive until adulthood, although the extreme polyuria, hydronephrosis, and growth retardation could not be abrogated. Studies of NKCC2 heterozygous animals showed essentially no difference from wild-type mice (251).

Why does deletion of NKCC2 result in such a severe phenotype, whilst deletion of NHE3, a transporter responsible for the majority of Na reabsorption in the kidney, results in a viable mouse capable of maintaining extracellular fluid volume? The answer appears to be in the special role that NKCC2 plays in the macula densa in the mediation of TGF. An intact TGF system in NHE3 knockout mice allows them to maintain a relatively normal distal fluid delivery through a reduction in GFR. In contrast, NKCC2 mice cannot compensate in this manner since the transporter is necessary for the TGF response to occur. Thus, in NKCC2 knockout mice, the distal nephron will be exposed to a NaCl load that drastically exceeds its absorptive capacity, leading to massive salt wasting and osmotic diuresis.

3. NCC and ENaC

The thiazide-sensitive Na+-Cl cotransporter (NCC) and the amiloride-sensitive Na+ channel (ENaC) are important targets for the action of aldosterone in the regulation of sodium excretion. Due to their interdependent role, we discuss them together in this section. NCC has been immunolocalized to the distal convoluted tubule (DCT) (58, 206), whereas ENaC is predominantly expressed in the connecting tubule, initial collecting tubule, and the cortical collecting duct (CCD) (85, 151). An increase in circulating vasopressin levels results in increases in abundance of both NCC and the β- and γ-subunits of ENaC (53, 184). In addition, vasopressin acutely increases Na absorption in the rat CCD by increasing apical Na entry via ENaC (214, 227, 266), which is proposed to be due to vasopressin-induced trafficking of ENaC-containing vesicles from intracellular stores to the apical plasma membrane (240).

NCC knockout mice have a mild phenotype, with a small decrease in blood pressure (234). On a normal diet, they appear to have a normal urinary concentrating ability, but upon dietary potassium restriction, the NCC null mice develop hypokalemia and a consequent polyuria that is associated with an apparent central defect in the regulation of vasopressin secretion (172). It is only after prolonged hypokalemia that the mice develop evidence of NDI, resulting from a suppression of AQP-2 expression in the collecting ducts.

Knockout of any of the three ENaC subunits results in a severe phenotype, with the mice suffering from neonatal death (15, 95, 165). In the α-ENaC knockout mice (95), early death appears to be due to failure to adequately clear fluid from the pulmonary alveoli after birth, whereas knockout of the β- and γ-subunits of ENaC results in mice that die from hyperkalemia and sodium chloride wasting (15, 165). Interestingly, when α-ENaC expression was deleted selectively from the collecting ducts, leaving intact ENaC expression in the renal connecting tubule and nonrenal tissues, the mice were viable and exhibited only a very mild phenotype with little or no inability to maintain fluid homeostasis in the face of salt or water restriction (217). In these mice, after water restriction, urine osmolality was not different from wild-type controls; thus it appears that Na absorption from the renal collecting duct via ENaC does not appear to be necessary for urinary concentration.

Taken together, NCC deleted only from the DCT or ENaC deleted only from the collecting duct results in a very mild phenotype, presumably because one can compensate for the other with regard to sodium balance. In this respect, double knockout animals could be informative. At present, it remains unclear whether the severity of the phenotype seen when any ENaC subunit is deleted globally is due to the importance of ENaC in nonrenal tissues or is related to the role of ENaC in the connecting tubule, which is conserved in the collecting duct specific α-ENaC knockout mice (217).

C. Potassium Channels


The ATP-sensitive, inward rectifier ROMK potassium channel (Kir1.1) is expressed in the TAL, DCT, connecting tubule, and collecting duct system, predominantly in the apical plasma membrane (147, 166, 287). In the TAL, ROMK plays a critical role in the process of active NaCl transport and thus the urinary concentrating and diluting mechanism. The abundance of ROMK in the TAL is increased with chronic vasopressin administration (52). In the connecting tubule and collecting duct, ROMK is responsible for the process of potassium secretion, thus regulating urinary potassium excretion and systemic potassium balance. The later process is strongly regulated by vasopressin (266). This secretory process may be an indirect consequence of vasopressin's action to increase Na entry via ENaC, which results in apical plasma membrane depolarization and an increase in the electrochemical driving force for K+ movement through ROMK (226). Alternatively, vasopressin may regulate the open probability of ROMK, in a process mediated by the cystic fibrosis transmembrane conductance regulator (CFTR), a cAMP responsive protein (136, 154).

ROMK knockout mice, developed by Lorenz et al. (152), manifest early death associated with hydronephrosis and severe dehydration, consistent with the known role of ROMK in active NaCl absorption in the TAL. Approximately 5% of these mice survive the perinatal period, but suffer from metabolic acidosis, hypernatremia, reduced blood pressure, polydipsia, polyuria, and an impaired urinary concentrating ability. Furthermore, whole kidney GFR is reduced, apparently as a result of hydronephrosis, and the fractional excretion of electrolytes is elevated. Micropuncture analysis revealed that the single-nephron GFR was relatively normal, absorption of NaCl in the TAL was reduced and the TGF mechanism was severely impaired. From these animals, a line of mice has been derived that has a greater survival rate and no hydronephrosis, although adults have higher water excretion rates (155). Interestingly, these mice do not exhibit hyperkalemia, indicating that the connecting tubule and/or collecting duct principal cells must be capable of secreting K via some other pathway, presumably flow-dependent, Ca2+-activated K channels referred to as “maxi-K” channels (283).

D. Chloride Channels

1. ClC-K1

ClC-K1 is a kidney-specific chloride channel that is localized to the apical and basolateral plasma membranes of the thin ascending limb of Henle's loop (tAL) (271). ClC-K1 expression in both the apical and basolateral plasma membranes, and examination of its transport properties, could explain why the tAL possesses an extremely high transepithelial chloride permeability compared with the TAL.

In 1999, Matsumura et al. (163) generated ClC-K1 null mice (Clcnk1−/−) and have made use of this model to examine the role of ClC-K1 in the urinary concentrating mechanism. Microperfusion studies determined that there was drastically reduced transepithelial chloride transport in the tAL of knockout mice. Importantly however, Clcnk1−/− mice had no significant differences in the plasma concentrations of Na+, K+, Cl, and HCO3 and pH values compared with controls, indicating that a loss-of-function mutation of ClC-K1 does not result in hypokalemic alkalosis (unlike CLCNKB mutations, type III Bartter syndrome). Physiological studies revealed that Clcnk1−/− mice had significantly greater urine volume and lower urine osmolality compared with controls and that even after a 24-h water deprivation, the knockout mice were unable to concentrate their urine. This observed polyuria was insensitive to [deamino-Cys1, d-Arg8]-vasopressin (dDAVP) administration, indicating NDI. Clearance studies have shown that the fractional excretion of sodium, chloride, and urea are not different in knockout mice. Taken together, these data indicate that the polyuria observed in ClC-K1 null mice is water diuresis and not osmotic diuresis. Solute analysis of the inner medulla of Clcnk1−/− mice determined that the concentrations of urea, Na+, and Cl were approximately half those of controls, resulting in a significantly reduced osmolality of the papilla (2). Unlike wild-type mice, the accumulation of these solutes was not increased by water deprivation.

In conclusion, Clcnk1−/− mice showed that rapid chloride exit from the tAL in the inner medulla is essential for generating a hypertonic inner medullary interstitium. As is pointed out by Knepper et al. (124), rapid solute exit from the tAL is theoretically critical to any process that concentrates solutes in the inner medulla because of a need for avoidance of dissipation of solutes from the inner medulla. Similarly, rapid Na+ and urea exit from the descending limb is a necessity for any inner medullary concentrating process (124), and the mechanisms involved in the exit of these entities will be an important area of focus for future research.

E. Aquaporins

In the kidney, to date, eight aquaporins have been localized to various segments of the renal tubule. In this review we focus only on the mouse models where gene deletion results in either a urinary concentrating defect or a reduction in transepithelial water transport (Fig. 6).

FIG. 6.

Schematic overview of a mammalian kidney tubule, showing the location of the major aquaporins involved in the urinary concentrating mechanism. See text for description.

1. AQP-1

AQP-1 is localized to the apical and basolateral plasma membrane of epithelial cells in the proximal tubule (189), where the majority of fluid filtered by the glomerulus is reabsorbed by an active near-isosmolar transport mechanism. AQP-1 is also expressed in the thin descending limb of Henle's loop (tDL) and the epithelium of the descending vasa recta (DVR) (186, 188), nephron segments thought to be involved in countercurrent multiplication and exchange. The constitutively high water permeability of nephron segments expressing AQP-1 is consistent with a lack of regulation of AQP-1 by vasopressin (259).

Verkman and colleagues (158) generated a knockout mouse model of AQP-1 by targeted gene deletion. Compared with wild-type littermates, AQP-1 knockout mice have a reduced urinary osmolality that is not increased in response to water deprivation. Indeed, the urinary concentrating defect is so severe in these mice that after 36 h of water deprivation, the average body weight decreased by 35% and serum osmolality increased to greater that 500 mosmol/kgH2O. The urinary concentrating defect observed in these mice is likely due to a combination of different mechanisms. Schnermann et al. (229) used a combination of isolated tubule microperfusion and free-flow micropuncture to define the role of AQP-1 in proximal tubule water transport and fluid reabsorption. These studies determined that the transepithelial osmotic water permeability (Pf) of isolated microperfused proximal tubule S2 segments was fivefold less in AQP-1 knockout mice, indicating that the major pathway for osmotically driven transepithelial water transport is through AQP-1. In addition, the proximal fluid reabsorption rate in AQP-1 knockout mice was approximately half that observed in control animals. Additional studies in AQP-1 knockout mice (273) demonstrated that active transport in the absence of a high water permeability increased the transepithelial osmotic gradient in the proximal tubule to ∼40 mosmol/kgH2O in contrast to the usual gradient of ∼5 mosmol/kgH2O. However, micropuncture of distal nephron segments revealed that the single-nephron glomerular filtration rate (snGFR) is reduced in AQP-1 mice, thus reducing distal fluid delivery to approximately the same level as in wild-type mice (229).

If AQP-1 knockout mice have relatively normal distal fluid delivery (despite defective proximal tubule water reabsorption), why do they have much greater urinary flow? It is likely that the answer to this lies in the role of AQP-1 in countercurrent multiplication. As described earlier, AQP-1 is abundantly expressed in the tDL. The osmotic water permeability of tDLs from AQP-1 knockout mice is ∼10-fold reduced compared with control animals (44), thus confirming the hypothesis that AQP-1 is the major water channel in tDL and pointing to the possibility that the countercurrent multiplier mechanism is impaired as a result of deletion of AQP-1 and diminished water absorption in the tDL. In addition, in isolated microperfused outer medullary DVRs from AQP-1 knockout mice, osmotic water permeability was reduced by greater than 50-fold compared with controls (201). The impairment of water transport in the vasa recta presumably results in a defect in countercurrent exchange, with concomitant depletion of medullary solutes. Taken together, these results indicate that the diminished urinary concentrating ability observed in AQP-1 knockout mice is likely due to a reduced ability to generate and maintain a hypertonic medullary interstitium. This conclusion is supported by the finding in AQP-1 null mice, based on tissue slice analysis, that there is a profound decrease in inner medullary solute accumulation (M. A. Knepper and T. Pisitkun, unpublished data). Also consistent with this conclusion is the finding that there is an almost complete lack of an increase in urine osmolality in AQP-1 null mice after the administration of vasopressin (158).

In humans, AQP-1 encodes the Colton blood group antigen (239). Consistent with the finding in knockout mice, it was observed that Colton-null individuals are unable to concentrate their urine as effectively as “normal” subjects when challenged by water deprivation (120).

F. Collecting Duct Aquaporins

The production of concentrated urine requires high collecting duct water permeability, allowing for the osmotically driven movement of water from the lumen to the interstitium. Transepithelial water transport across the collecting duct epithelium is generally believed to occur by a transcellular route with serial passage across the apical and basolateral plasma membranes. The water channels responsible for this transport appear to be AQP-2 in the apical plasma membrane and a combination of aquaporin-3 (AQP-3) and aquaporin-4 (AQP-4) in the basolateral plasma membrane (186). Knockout models have been developed to investigate the roles of these aquaporins in the urinary concentrating mechanism.

1. AQP-2

AQP-2 is abundantly expressed in all renal tubule segments beyond the DCT, including the connecting tubule (connecting tubule cells), the cortical and outer medullary collecting duct (principal cells), and the IMCD (IMCD cells) (185, 186). In these cell types, AQP-2 is found in the apical plasma membrane, subapical vesicles, and (especially in IMCD cells) in the basolateral plasma membrane. AQP-2 can be regulated by vasopressin (short-term and long-term) both by changes in transporter abundance (long-term regulation) and trafficking (short-term regulation) (reviewed in Ref. 186). The mechanisms behind arginine vasopressin (AVP)-mediated trafficking of AQP-2 have been discussed in detail elsewhere (reviewed in Refs. 37, 180), but for the purposes of this review are summarized as follows (see Fig. 7). Vasopressin binds to the V2R in the basolateral plasma membrane, resulting in activation of adenylate cyclase through V2R-coupled GTP-binding protein Gs. This increases intracellular cAMP levels and leads to activation of protein kinase A (PKA), intracellular Ca2+ oscillations (297), activation of myosin light-chain kinase (42) as well as cAMP-dependent, PKA-independent activation of the Rap-GEF Epac (298). Through multiple (largely unknown) mechanisms, V2R-mediated signaling leads to the exocytic insertion of AQP-2-bearing intracellular vesicles into the apical plasma membrane. These vesicles appear to be recycling endosomes rather than secretory vesicles (14). In addition, long-term exposure to vasopressin leads to an increase in AQP-2 synthesis via increased gene transcription, increasing the total abundance of AQP-2 in the cells. These mechanisms ultimately allow an increase in water reabsorption in the collecting duct.

FIG. 7.

An overview of multiple pathways within a kidney collecting duct cell that are involved in the urinary concentrating process. Vasopressin (AVP) binds to the type 2 vasopressin receptor (V2R) in the basolateral plasma membrane. The GTP-binding protein Gs activates the synthesis of cAMP via an adenylate cyclase (AC). Binding of cAMP to the regulatory subunit of protein kinase A (PKA) then causes activation of the catalytic subunit of PKA, resulting in release of calcium from intracellular stores. This Ca2+ can further regulate other proteins, including calmodulin, protein kinase C (PKC-α), cytosolic phospholipase A2 (cPLA2), and nitric oxide synthase (NOS). Genetic deletion of individual components of these pathways in mice can result in a defective urinary concentrating mechanism. See text for description.

AQP-2 plays an essential role in the urinary concentrating mechanism. This is underlined in clinical disorders that result in defects in body water balance (232). AQP-2 is dysregulated in acquired NDI resulting from, for example, lithium treatment or hypercalemia-induced NDI (68, 141, 161). Furthermore, several mutations in the AQP-2 gene can cause the rare disorder of hereditary NDI (autosomal recessive or dominant NDI) (reviewed in Ref. 69). These clinical disorders have been discussed in detail elsewhere (223).

Despite the strong evidence implicating an essential role of AQP-2 in the urinary concentrating mechanism, a number of suitable mouse models to examine its function have only recently been developed/discovered. In 2001, a mouse knock-in model of AQP-2-dependent NDI was generated by inserting a T126M mutation into the mouse AQP-2 gene by a Cre-loxP strategy (292). This mutation has been shown to result in a failure of delivery of mature AQP-2 protein to the apical plasma membrane. Although the mutant mice appeared normal at birth, they failed to thrive and generally died within 1 wk. Analysis of the urine and serum revealed serum hyperosmolality and low urine osmolality, trademark characteristics of a defective urinary concentrating mechanism. Forward genetic screening of ethylnitrosourea-mutagenized mice isolated another mouse model of NDI with a F204V mutation (150). These mice survive beyond the neonatal period and have a much milder form of NDI. Examination of these mice could result in the identification of molecular chaperones that can correct similar forms of human NDI.

Two other mouse models have been developed that allow the role of AQP-2 in the adult mouse to be examined. One model, developed by Rojek et al. (216), makes use of the Cre-loxP system of gene disruption to create a collecting duct specific deletion of AQP-2, leaving relatively normal levels of expression in the connecting tubule. Another model, developed by Yang et al. (294), with inducible AQP-2 protein deletion in the kidney, was accomplished by tamoxifen-inducible Cre-recombinase expression in homozygous mice in which loxP sites were introduced in introns of the mouse AQP-2 gene. The major phenotype in both of these mice is severe polyuria, with average basal daily urine volumes approximately equivalent to body weight (216, 294). The high urine output observed resulted in very low urine osmolality, which was not increased after water restriction. However, despite the polyuria, with free access to water, plasma concentrations of electrolytes, urea, and creatinine are not different in knockout mice compared with controls, and neither was the estimated GFR. Thus, despite having normal renal function (presumably normal active Na+ transport), there is a major defect in the urinary concentrating mechanism in these mice. This defect confirms that AQP-2 is responsible for the majority of transcellular water reabsorption at the rate-limiting luminal membrane of the collecting duct.

A mouse model has also been “discovered” that implicates an essential role of AQP-2 phosphorylation on transporter function (164). This mouse model has a single base change in codon 256 of AQP-2, resulting in a serine to leucine amino acid substitution and loss of AQP-2 phosphorylation at amino acid 256. This mutation results in an absence of AQP-2 accumulation in the apical plasma membrane. Phenotypically, the mutant mice have no response to vasopressin and produce large quantities of hypotonic urine, characteristic of NDI. This mouse model provides direct genetic evidence that phosphorylation of AQP-2 at S256 is essential for its apical membrane accumulation and maximal water reabsorption in the collecting duct.

The latest mouse model to study the role of AQP-2 has helped us to understand the molecular mechanisms behind AQP-2-dependent autosomal-dominant nephrogenic diabetes (AD-NDI) insipidus (AD-NDI). Using a “knockin” strategy, and based on previous studies identifying three frameshift mutations in the COOH terminus of AQP-2 that result in AD-NDI (140), Sohara et al. (242) have created a mouse model that expresses 76 amino acids of the COOH terminus of a human AD-NDI mutant AQP-2 (763–772del), fused to the “wild-type” 254 NH2-terminal amino acids of mouse AQP-2. Mice that are heterozygous for the mutation exhibited a severely impaired urinary concentrating ability, but after dehydration were able to moderately increase their urine osmolality, a milder phenotype indicative of AD-NDI compared with autosomal-recessive NDI. Furthermore, the mutant AQP-2 was missorted to the basolateral plasma membrane and formed heteroligomers with wild-type AQP2, resulting in a dominant-negative effect on the normal apical sorting of wild-type AQP-2. Additional studies in this mouse model determined that the phosphodiesterase 4 inhibitor rolipram was partially able to restore concentrating ability, indicating that phosphodiesterase inhibitors may be useful drugs for the treatment of AD-NDI.

2. AQP-3

Xenopus oocyte expression studies have shown that AQP-3 is not only permeable to water but functions efficiently as a glycerol transporter, thus making it a member of the so-called aquaglyceroporins (56, 103, 293). In the kidney, AQP-3 is localized to the basolateral plasma membranes of the connecting tubule cells and collecting duct principal cells in cortex and outer medulla (48, 54, 104). AQP-3 is thought to mediate basolateral exit of water that enters via AQP-2. Interestingly, AQP-3 is not abundant in the cytoplasm, and there is no evidence for the short-term regulation of AQP-3 by vasopressin-induced trafficking. However, a marked increase in the abundance of AQP-3 mRNA and protein is observed during long-term vasopressin stimulation, such as seen during water deprivation or vasopressin infusion in Brattleboro rats (54, 104, 176, 259).

AQP-3 knockout mice have been generated by targeted gene deletion and found to have a greater than threefold reduction in osmotic water permeability of the basolateral membrane of the CCD compared with wild-type control mice (156). AQP-3 null mice are markedly polyuric (10-fold greater daily urine volume than controls), with an average urine osmolality of <300 mosmol/kgH2O. However, unlike AQP-1 or AQP-2 null mice, AQP-3 knockout mice are able to partially raise their urine osmolality after either water deprivation or the administration of dDAVP. Serum electrolyte concentrations from AQP-3 null mice are not significantly different from controls, although plasma osmolality is mildly elevated. In contrast to AQP-1 null mice, with a defective countercurrent exchange mechanism, the countercurrent exchange in AQP-3 knockout mice is virtually intact.

What is the basis for the defective urinary concentrating mechanism in these mice? It is likely that when AQP-3 is deleted, the reduced osmotic water permeability of the basolateral membrane results in a decrease in transepithelial water permeability. Since the majority of vasopressin-dependent fluid reabsorption in the antidiuretic kidney is in the connecting tubule and collecting duct, reduced water permeability is predicted to result in a hyposmolar urine. However, AQP-3 null mice are able to partially raise their urine osmolality after water deprivation or vasopressin stimulation despite a relatively water-impermeable collecting duct (156). This response is likely due to the fact that AQP-4 and not AQP-3 is the predominant basolateral water channel in the medullary parts of the collecting duct system, allowing a normal response to vasopressin in the most distal portions of the collecting duct system.

Another factor leading to increased urinary flow rates in adult AQP-3 knockout mice is the striking distortion of medullary architecture that progressively destroys the inner medulla and replaces it with a dilated renal pelvis (hydronephrosis) (156). The mechanism of hydronephrosis in this model is unknown.

3. AQP-4

AQP-4 is localized to the basolateral plasma membrane of outer medullary collecting duct (OMCD) principal cells and IMCD cells (258). AQP-4 is most abundant in the inner medulla, with a gradual decrease in expression towards the cortex (258). In mice, AQP-4 has also been reported to be expressed in the basolateral plasma membrane of proximal tubule S3 segments (274). Renal medullary AQP-4 mRNA abundance was found to be increased in response to water restriction (176) or vasopressin infusion (34), whereas there was no change in protein expression in response to either manipulation (259).

AQP-4 null mice have been generated by standard gene deletion methods (157). Isolated perfused tubule studies determined that there is a fourfold decrease in IMCD osmotic water permeability in AQP-4 null mice relative to wild-type mice, indicating that AQP-4 is responsible for most of the basolateral membrane water movement in this segment (45). Despite this reduced water permeability in the IMCD, in hydrated mice, there was no difference in urine osmolality compared with controls and no difference in serum electrolyte concentrations. However, there was a significant reduction in maximal urine osmolality in AQP-4 null mice after a 36-h water deprivation, and this reduced urine osmolality could not be further increased by vasopressin administration, indicating a mild urinary concentrating defect (157). Why does deletion of AQP-4 have such a profound effect on water permeability of the IMCD, with only a modest decrease in urinary concentrating ability? The answer to this is based on the normal distribution of water transport along the collecting duct. Micropuncture studies performed under antidiuretic conditions demonstrated that the amount of water reabsorbed osmotically in the late distal tubule (connecting tubule plus initial collecting tubule) is much greater than that absorbed in the medullary nephron segments (145). Thus, since AQP-4 in mouse is expressed predominantly in the medullary collecting ducts, deletion of AQP-4 results in only a mild defect in kidney water absorption.

4. AQP-7

AQP-7 is a member of the aquaglyceroporins and effectively transports both glycerol and water (102). In the kidney, AQP-7 is abundantly expressed on the apical plasma membrane of the proximal straight tubules (101, 181). AQP-7 knockout mice have reduced water permeability in the proximal tubule brush-border membrane (241). However, AQP-7 null mice do not exhibit a urinary concentrating defect or water balance abnormality. Because AQP-1 has been shown to be the major water transport pathway in the proximal tubule, it is not surprising that there is no abnormality in water excretion in AQP-7 knockout mice. However, AQP-1/AQP-7 double-knockout mice have been generated and showed a significantly greater urine output compared with AQP-1 null mice, leading the authors to speculate that the amount of water reabsorbed through AQP-7 in the proximal straight tubules is physiologically substantial, although direct measurements of proximal tubule water transport were not reported (241). The greater water excretion observed in the AQP-1/AQP-7 double-knockout mice compared with AQP-1 solo-knockout mice was accompanied by a proportional decrease in urine osmolality, suggesting that the total excretion of osmolar substances was not altered in these mice.

G. Urea Transporters

In mammals, ∼90% of waste nitrogen is normally excreted by the kidney as urea, the balance being attributable to ammonium and uric acid. The majority of this urea is generated in the liver via the urea-ornithine cycle. In humans and animals, under most circumstances, dietary protein intake greatly exceeds that necessary for the support of anabolic processes; thus excess quantities of urea are generated that need to be excreted. This urea constitutes a large osmotic load to the kidney. Most solutes excreted in such large amounts, for example, mannitol (7), would obligate large amounts of water excretion by causing an osmotic diuresis. However, under normal circumstances, urea does not induce an osmotic diuresis. Studies in the 1930s by Gamble et al. (71) demonstrated “An economy of water in renal function referable to urea,” which provided early evidence for a unique role of urea in the urinary concentrating mechanism.

The process of urea accumulation in the medulla has been thoroughly studied, and it is generally accepted that urea accumulation is dependent on facilitated urea transport across the epithelium of the IMCD (125). This urea transport process and regulation have been reviewed extensively (65, 221, 222) and thus are not covered here. Central to urea transport within the kidney are UT-A and UT-B facilitative urea transporters. Recently, three mouse models have been created with selective deletion of different urea transporter isoforms. These mouse models have shed new light on the role of urea in the urinary concentrating mechanism.

1. UT-A1 and UT-A3

UT-A1 and UT-A3 are expressed exclusively in IMCD cells. Immunochemical methods have localized UT-A1 to the cytoplasm and apical region of the IMCD (66, 190), whereas UT-A3 is localized both intracellularly and in the basolateral membrane (248, 260) (Fig. 8). In 2004, we developed a mouse model that allowed us to specifically assess the role of inner medullary urea transport in kidney function (63) by deleting both UT-A1 and UT-A3 by standard gene targeting techniques (UT-A1/3−/− mice).

FIG. 8.

Localization of UT-A urea transporters in the mouse renal tubule. A schematic representation of the mouse nephron is shown. UT-A1 is localized to the terminal portion of the IMCD and is both intracellular and in the apical domain. UT-A2 is localized to the thin descending limbs of Henle's loop in both the outer medulla and inner medulla. UT-A3 is localized to the terminal portion of the IMCD and is both intracellular and in the basolateral domains.


Isolated perfused tubule studies demonstrated a complete absence of phloretin-sensitive and vasopressin-regulated urea transport in IMCD segments from UT-A1/3−/− mice (63), thus providing confirmation of the role of these transport proteins in IMCD urea transport. The prime conceptual framework outlining the proposed contribution of collecting duct urea transport to the urinary concentrating mechanism is derived largely from a model of urea handling proposed in the 1950s by Berliner et al. (22). They hypothesized that luminal urea in the IMCD is osmotically ineffective because of a high IMCD urea permeability that, abetted by countercurrent exchange processes, allows urea to accumulate to high concentrations in the inner medullary interstitium. Thus, although urea accumulates to 1 M or more in the lumen, it accumulates to a similar concentration in the inner medullary interstitium, thereby preventing an osmotic diuresis. Indeed, UT-A1/3−/− mice on either a normal protein (20% protein by weight) or high-protein (40%) diet had a significantly greater fluid intake and urine flow than wild-type animals, resulting in a decreased urine osmolality (63, 64). However, UT-A1/3−/− mice on a low-protein diet did not show a substantial degree of polyuria. In this latter condition, hepatic urea production is low and urea delivery to the IMCD is predicted to be low, thus rendering collecting duct urea transport immaterial with regard to water balance. Studies investigating the maximal urinary concentrating capacity of UT-A1/3−/− mice showed that after an 18-h water restriction, mice on a 20 or 40% protein intake are unable to reduce their urine flow to levels below those observed under basal conditions, resulting in volume depletion and loss of body weight (64). In contrast, UT-A1/3−/− mice on a 4% protein diet were able to maintain fluid balance. Thus the concentrating defect in UT-A1/3−/− mice is caused by a urea-dependent osmotic diuresis; greater urea delivery to the IMCD results in greater levels of water excretion. Overall, these results demonstrate that the primary role of IMCD urea transporters in the urinary concentrating mechanism is in their ability to prevent a urea-induced osmotic diuresis.


In 1959, Kuhn and Ramel (139) proposed the classical countercurrent multiplier model, the basis for the urinary concentrating mechanism (see sect. i). The gradient that drives the countercurrent multiplier in the outer medulla, thus concentrating the urine, is dependent on the active reabsorption of NaCl in the water-impermeable TAL. However, the mechanism that concentrates NaCl in the inner medulla interstitium and thus water reabsorption from the collecting ducts remains controversial, as the tAL is apparently incapable of measurable rates of active NaCl transport (97, 132). Various mechanisms have been offered to explain NaCl accumulation in the inner medulla (121, 124, 228, 261), with the most influential model independently proposed by Stephenson (246) and by Kokko and Rector in 1972. In this mechanism, known as the “passive countercurrent multiplier mechanism,” the rapid reabsorption of urea from the IMCD generates a high urea concentration in the inner medullary interstitium, resulting in urea-induced water absorption in the descending limb of Henle generating a transepithelial gradient for the passive reabsorption of NaCl from the tAL. In addition, if tAL urea permeability is extremely low (almost zero), then NaCl that has been reabsorbed from the tAL will not be replaced by urea and the ascending limb fluid will be dilute relative to other nephron segments. This dilution process, analogous to that observed in the outer medulla, is proposed to constitute a “single effect” that can be multiplied by the counterflow between the ascending and descending limbs of Henle's loops.

If the passive model of NaCl accumulation in the inner medulla functions as proposed, abolition of rapid passive urea absorption in the IMCD as seen in UT-A1/3−/− mice would be expected to eliminate NaCl accumulation in the inner medulla. However, two independent experiments in UT-A1/3−/− mice failed to corroborate the view that inner medullary NaCl accumulation depends on facilitated urea transport in the IMCD. In an initial experiment, the mean urea, Na+, Cl, and K+ concentrations were measured in whole inner medulla tissue isolated from water-restricted UT-A1/3−/− mice and wild-type littermates (63). In UT-A1/3−/− mice there was a significantly reduced inner medullary urea concentration. However, there were no reductions in the mean Na+, Cl, or K+ concentrations. In a second study, osmolality, urea, and Na+ concentrations were measured in the cortex, outer medulla, and two levels of the inner medulla from UT-A1/3−/− and wild-type mice (64) fed either a low- (4%) or high-protein (40%) diet. In wild-type mice, changing the dietary protein intake from 4 to 40% resulted in a greater tissue osmolality that was attributable solely to a greater accumulation of urea in the inner medulla. However, at all levels of the corticomedullary axis, sodium concentrations were unaffected by changes in dietary protein intake. Furthermore, UT-A1/3−/− mice had a substantially attenuated corticomedullary osmolality gradient and no urea gradient on either diet. Despite this difference in urea concentration, the corticomedullary sodium gradients of UT-A1/3−/− mice were virtually identical to wild-type mice. Thus marked medullary urea depletion resulting from either dietary protein restriction or deletion of collecting duct urea transporters does not affect the ability of the kidney to form a corticomedullary sodium chloride gradient. We conclude from these studies in UT-A1/3−/− mice that NaCl accumulation in the inner medulla is not reliant on either IMCD urea transport or the accumulation of urea in the IMCD interstitium. Thus the passive concentrating model in the form originally proposed by Stephenson and by Kokko and Rector, where NaCl reabsorption from Henle's loop depends on a high IMCD urea permeability, is apparently not the mechanism by which NaCl is concentrated in the inner medulla.

2. UT-A2

Under normal conditions, UT-A2 mRNA and protein are expressed in the inner stripe of the outer medulla (Fig. 8), where it is localized to the lower portions of the tDL of short loops of Henle (66, 237, 278). Under prolonged vasopressin action, UT-A2 mRNA and protein can also be detected in the inner medulla, being localized to the tDL of long loops of Henle. Multiple studies have shown that UT-A2 mRNA and protein can be regulated in response to changes in circulating vasopressin concentration (34, 66, 278), and recent studies have determined that UT-A2-mediated urea transport can be acutely regulated by cAMP (207). The presence of vasopressin receptors has not been reported in the descending limb of Henle's loop, raising the possibility that the effects of vasopressin are indirect, e.g., mediated by autacoids released by vasopressin-responsive cells.

What is the predicted role of UT-A2 in the urinary concentrating mechanism and why is the thin limb localization of UT-A2 important for the kidneys concentrating ability? The large amount of urea that is reabsorbed from the IMCD (via UT-A1 and UT-A3) is trapped in the inner medullary interstitium by countercurrent exchange processes. Any urea that “escapes” the inner medulla via the ascending vasa recta (AVR) can exit via the fenestrated endothelium of the AVR. However, the close proximity of the AVR and short-loop descending limbs in the vascular bundles of the inner stripe theoretically can permit transfer of this urea from the AVR into the descending limbs via UT-A2 (Fig. 3). This mechanism is predicted to be important for the concentrating mechanism by ensuring efficient recycling of urea, thus making urea available to distal sites of short-looped nephrons. Evidence for urea recycling was obtained from the micropuncture studies of Lassiter et al. (145, 146), showing that there is a large amount of net urea secretion into the short loops of Henle, resulting in the delivery of large amounts of urea to the superficial distal tubule that often exceeds the filtered load. UT-A2 in the inner medullary portion of the tDL may also recycle urea absorbed from the ascending thin limb of Henle [which is highly permeable to urea (43) despite the absence of any known urea transporter] and from the AVR. The enhanced countercurrent exchange between ascending and descending structures would help maintain a high urea concentration in the inner medullary interstitium and thus the driving force for the osmotic extraction of water.

UT-A2 knockout mice have recently been developed, and some aspects of their renal phenotype have been described (272). On a normal level of protein intake (20% protein), the UT-A2 null mice do not have significant differences in daily urine output compared with control mice and, even after a 36-h period of water deprivation, differences in urine output and urine osmolality are not observed. Furthermore, UT-A2 knockout mice do not have an impairment of urea or chloride accumulation in the inner medulla (272). These results are surprising, considering the role that UT-A2 has been proposed to play in maintaining a high inner medullary urea concentration. However, on a low-protein diet (4% protein), UT-A2 null mice have a mildly reduced maximal urinary concentrating capacity compared with wild-type controls and a significant reduction in urea accumulation in the inner medulla.

The findings from these initial studies in UT-A2 knockout mice are surprising, especially considering the strongly induced upregulation of UT-A2 under antidiuretic conditions. Clearly from these experiments, under basal conditions, UT-A2 makes a minimal contribution to urea accumulation in the inner medullary interstitium and does not play a major role in the formation of concentrated urine. However, when urea supply to the kidney is limited, UT-A2 may be important for maintaining a high concentration of urea in the inner medulla and thus maximal urinary concentrating ability.

3. UT-B

In contrast to the multiple UT-A isoforms described above, the mouse UT-B gene encodes only a single protein. UT-B is expressed exclusively throughout the kidney medulla in the basolateral and apical (luminal) regions of the DVR endothelial cells (209, 268, 288). In humans, UT-B protein carries the Kidd blood group antigen, and subjects lacking this antigen [JK (a,b) individuals] have a dramatically reduced urea permeability of red blood cells and a mild urinary concentrating defect (224). UT-B is thought to contribute to the urinary concentrating mechanism by allowing any urea that “escapes” the inner medullary interstitium via the venous AVR, to be recycled back into the DVR, thus not allowing this urea to return to the general circulation. This complex intrarenal urea recycling between AVR and DVR, in combination with recycling of urea between UT-B and UT-A2 (in thin descending limbs of Henle's loop), is thought to maintain the high inner medullary interstitium urea concentration and thus the driving force for passive water reabsorption.

In 2002, Yang et al. (291) developed a mouse model with genetic deletion of UT-B that has enabled the role of UT-B in the urinary concentrating mechanism to be thoroughly investigated. However, since the physiology of the UT-B knockout mice has been recently discussed very extensively elsewhere (290), for the purposes of this article only a brief overview is given. Erythrocytes from UT-B knockout mice have an ∼45-fold lower urea permeability compared with those from controls. Under basal conditions (normal protein diet), daily urine output in UT-B null mice is significantly higher and urine osmolality significantly lower compared with wild-type mice. However, when UT-B knockout mice are subjected to water deprivation for 36 h, they are able to concentrate their urine, although to a lesser extent than controls. Knockout mice have a significantly lower urine urea concentration (although in same proportion to other solutes) and significantly higher plasma urea; thus their urine-to-plasma urea ratio (U/P) is more severely reduced than that of other solutes. This reduced capacity to concentrate urea compared with other solutes indicates that the UT-B null mice have a “urea-selective” urinary concentrating defect (13). This diminished ability to concentrate urea is highlighted by a lower inner medullary urea concentration compared with other solutes. Taken together, these data indicate that the urinary concentrating defect observed in UT-B null mice is due to impaired urea recycling in the vasa recta, although at present, it is unclear whether the loss of urea transport in red blood cells also contributes to this phenomenon. The latter could occur if urea can enter erythrocytes via non-UT-B pathways as they enter the medulla, but could not exit fast enough to equilibrate with the surrounding interstitium as the erythrocytes are carried back to the general circulation.

H. Receptors and Signaling Molecules

1. V2R

AVP is essential in the regulation of body fluid homeostasis. In response to small increases in plasma osmolality or a reduction in the effective circulating blood volume, AVP is released by the posterior pituitary gland and promotes water reabsorption in the kidney collecting duct, enhanced urinary concentration in the TAL, and vasoconstriction via four subtypes of receptors (93). The antidiuretic effects of AVP result from a cascade of events, initialized by the binding of AVP to the V2R in the collecting duct and TAL (reviewed in Refs. 37, 186). In the collecting duct, this binding causes activation of G-coupled proteins, a rise in intracellular cAMP, and eventually, through several mechanisms, the exocytic insertion of AQP-2 water channels into the apical plasma membrane (Fig. 7). This dramatically increases water permeability (typically 8- to 10-fold) and allows the osmotically driven movement of water from the kidney tubule lumen into the kidney interstitium, promoting water retention and thereby lowering plasma osmolality. Physiological studies, receptor binding, and radioactive tracer studies have shown that V2R is subject to internalization in response to binding of vasopressin (28, 92, 98100). Internalization may also be involved in the vasopressin escape phenomenon, a physiological adaptation to prevent water intoxication with prolonged vasopressin action (262).

Inactivating V2R mutations in humans cause a rare kidney disease known as X-linked nephrogenic diabetes insipidus (XNDI), an AVP-insensitive form of diabetes insipidus that is inherited in an X-linked manner (223). XNDI patients produce large volumes of dilute urine, are polydipsic, and in the case of an inadequate water supply, can become severely hypernatremic. Failure to recognize the disease in affected boys soon after birth can result in abnormalities of the central nervous system, owing to severe dehydration.

In 2000, Yun et al. (300) created a mouse model of XNDI by introducing a nonsense mutation (Glu242stop) known to cause XNDI in humans into the mouse genome. This particular mutation was chosen as it has been shown that the encoded mutant receptor is retained intracellularly and completely lacks functional activity (231), thus mimicking the functional properties of many other disease-causing V2R mutants.

Male V2R mutant mice (V2R−/y) typically died within 7 days after birth (300). Urine osmolalities, collected from the bladders of 3-day-old pups, were significantly lower than controls. Serum electrolyte analysis revealed that V2R−/y pups have increased Na+ and Cl levels, indicative of a severe state of hypernatremia. In control mice, an intraperitoneal injection of dDAVP resulted in a significant increase in urine osmolality, whereas no effect was observed in V2R−/y mice. Analysis of adult female V2R+/− mice revealed that the mice have polyuria, polydipsia, and a reduced urinary concentrating ability, clear symptoms of XNDI, although milder than in hemizygous males. Furthermore, females have an ∼50% decrease in total AVP binding capacity, resulting in an ∼50% decrease in vasopressin-induced intracellular cAMP levels. Taken together, the results obtained from this loss-of-function mutation in the V2R clearly show that the V2R is necessary for normal regulation of water excretion.

2. Bradykinin B2 receptor

Kinins are produced by proteolytic cleavage of a protein precursor, kininogen, by the enzyme kallikrein. They act as “local hormones” by activating the release of endothelium-derived relaxing factor and prostaglandins (39). Kinins act through a family of G protein-coupled receptors, the B1 and B2 receptors, with the B2 receptor mediating vasodilation, diuresis, and natriuresis (213). In the kidney, kallikrein is localized to the distal nephron including the DCT, connecting tubule, and initial collecting tubule (210). It has been proposed that kinins are formed chiefly in the lumen of the collecting ducts (276). However, in isolated perfused tubule experiments, bradykinin is only effective in decreasing vasopressin-stimulated NaCl and water transport when added to the peritubular environment and not when added to the lumen (265, 266).

It has been proposed that renal kinins counteract the hydrosmotic effect of AVP, and tubule perfusion studies have shown that kinins added to the peritubular bath can decrease AVP-stimulated water reabsorption in the cortical collecting duct (235). Furthermore, in Brattleboro rats, the kallikrein inhibitor aprotinin augments the renal response to AVP (113, 114).

Bradykinin B2 receptor knockout mice have been developed and their renal phenotype partially characterized (3, 27). During basal conditions there is no significant difference in either urine volume or urine osmolality between knockout mice and controls, suggesting that renal kinins play little role in regulation of water excretion under normal conditions. However, after a 24-h fluid restriction, urine volume was significantly lower and urine osmolality significantly higher in knockout mice. Furthermore, subcutaneous administration of dDAVP results in a significantly greater increase in urine osmolality in knockout mice compared with controls. Taken together, these results suggest that water restriction or V2-receptor stimulation has a greater urinary concentrating effect in B2 receptor knockout mice than in controls, suggesting that endogenous kinins acting through B2 receptors oppose the antidiuretic effect of AVP in vivo (33).

3. Integrin α1β1

Integrins are transmembrane receptors for extracellular matrix components and are thought to play an important role in regulating cytoskeletal organization (208), as well as in the src-dependent activation of mitogen-activated protein (MAP) kinases (84). One member of the integrin family, integrin α1β1, acts as a collagen binding receptor and is highly expressed in the glomerulus and along the renal tubule (137, 277). Studies of integrin α1 null mice have shown that integrin α1β1 plays a regulatory role in the hypertonicity-mediated accumulation of osmolytes in the kidney inner medulla (168). Integrin α1 knockout mice have reduced osmolyte accumulation under control, diuretic, and antidiuretic conditions. These changes are probably caused by a reduced expression of TonEBP, a result of impaired signaling pathways associated with reduced p38 activation and increased ERK activation. In addition, under basal conditions, urine osmolality is significantly reduced in integrin α1 null mice compared with controls, although maximal urinary concentrating ability is not affected (168). The mechanism behind this concentrating defect is not understood. We speculate that it may be due, at least in part, to a reduced expression of other proteins involved in the urinary concentrating mechanism (see sect. iiK1).

4. Heterotrimeric G protein subunit, Gsα

Heterotrimeric G proteins function in signal transduction pathways as molecular switches. Each G protein is composed of α-, β-, and γ-subunits, the products of separate genes. The α-subunit binds guanine nucleotides and interacts with specific receptors and effectors (243, 244). The α-subunit of Gs (Gsα) is ubiquitously expressed and is known to couple multiple receptors to the stimulation of adenylyl cyclase and specific ion channels (see Fig. 7). For example, in the kidney, vasopressin binds to the V2R in the basolateral plasma membrane, resulting in the activation of adenylyl cyclase through the V2R-coupled Gs. Adenylyl cyclase activation increases intracellular cAMP levels and ultimately increases water reabsorption in the collecting duct (see above and Ref. 37).

Gsα knockout (GSKO) mice have been created by targeted disruption of the Gnas gene (299). Homozygote knockout mice do not develop beyond the embryonic stage, but heterozygotes survive and manifest an interesting phenotype that is dependent on whether the mutant gene is derived from the mother (m/p+) or father (m+/p). This phenotypic variability is due to genomic imprinting (281). Heterozygotic mice have a significantly reduced expression of Gsα protein abundance in the outer medulla, resulting in reduced glucagon-stimulated cAMP production (55). In addition, NKCC2 protein abundance is significantly less in the outer medulla of GSKO mice, as is the abundance of type VI adenyl cyclase. In response to a single intramuscular injection of dDAVP, GSKO mice have a significantly lower urine osmolality compared with controls. However, after thirsting for 48 h, there was no significant difference in outer medullary NKCC2 abundance or urine osmolality compared with wild-type mice, suggesting that the impairment of outer medullary NKCC2 expression in GSKO mice could be overcome by a chronic increase in circulating vasopressin (55). Taken together, these results provide additional evidence for the importance of NKCC2 abundance in the TAL of the outer medulla of rodents as a determinant of urine concentrating capacity, due to its involvement in countercurrent multiplication by the loop of Henle.

Recently, another mouse model has been developed that further implicates an essential role of Gsα in the vasopressin signaling cascade and ultimately the urinary concentrating mechanism. Mice that express Cre recombinase in the kidney medulla (Cre recombinase driven by the renin promoter) were crossed with mice in which exon 1 of the Gnas gene was “floxed,” resulting in mice with deletion of Gsα in the medullary collecting duct (41). These mice had reduced basal concentrating ability that did not respond to vasopressin, indicative of NDI.

Interestingly, human patients with heterozygous inactivating mutations of the Gsα gene (pseudohypoparathyroidism type Ia) do not have a decreased urinary concentrating capacity after overnight fluid restriction or in response to exogenous vasopressin (173). This suggests that the impairment of countercurrent multiplication in humans may be less severe than in mice with analogous mutations.

5. Protein kinase C

Protein kinase C (PKC) is a member of a family of closely related kinases that phosphorylate serine or threonine residues in proteins (183). It acts on numerous intracellular proteins and influences multiple cellular functions such as protein expression, ion transport, hormone release, and regulation of receptor proteins (183). The PKC family consists of 10 different isozymes that are classified according to their primary structure: conventional (α, γ, βI, βII), novel (δ, ε, η/L, θ), and atypical (ζ, ι/λ). Several of these isoforms are expressed in the kidney (50, 59, 109, 111, 204, 212), where they have been proposed to play a role in vascular and glomerular function, renin gene transcription, and tubular transport (5, 17, 23, 46, 88, 296).

PKC-α is predominantly expressed in the kidney, where it is localized to the glomerulus, CCD intercalated cells, and the medullary collecting duct (212). In the IMCD, PKC-α is not directly activated by vasopressin as judged by translocation assays (46), but is regulated indirectly by intracellular calcium (Fig. 7). Insights into the role of PKC-α in kidney function have been uncovered by the use of PKC-α knockout mice (PKCα-KO) (148, 296). Under basal conditions (free access to water), PKCα-KO mice have a greater urinary flow rate and lower urinary osmolality, associated with a greater urinary vasopressin-to-creatinine ratio. In addition, the significantly lower urinary osmolality in PKCα-KO mice persists after a 36-h water deprivation. Taken together, these results indicate that the lower urinary concentrating ability in PKCα-KO mice is independent of water intake. Despite the obvious phenotype, the precise role of PKC-α on urinary concentrating ability remains undetermined. However, it seems likely that the effect is in the collecting duct or thick ascending limb, the segments most centrally involved in the urinary concentrating process. Studies directly assessing abundance, localization, and phosphorylation state of the major transporters involved in the concentrating mechanism in the knockout mice are needed to determine the mechanism of the effect of deletion of PKC-α on water excretion.

6. Serine/threonine phosphatase, calcineurin Aα

The calcium- and calmodulin-dependent serine/threonine protein phosphatase calcineurin has been implicated in numerous cellular processes and Ca2+-dependent signal transduction pathways (218). Calcineurin is a heterodimeric protein composed of the catalytic subunit calcineurin A and a Ca2+-binding regulatory subunit, calcineurin B. The A subunit can exist as three closely related forms: α, β, and γ. In the kidney, the proximal tubules and collecting ducts express predominantly the α- and β-isoforms, with the distal tubules containing the highest calcineurin activity corresponding to the α-isoform (269). Calcineurin Aα (CN-Aα) colocalizes with AQP-2 in the collecting duct and has been proposed to be involved in the vasopressin-stimulated AQP-2 trafficking mechanism (80), further emphasized by recent studies in CN-Aα knockout mice.

Compared with wild-type controls, CN-Aα knockout mice (79) have a higher serum osmolality, a reduced urine osmolality, and an impaired response to concentrate their urine after vasopressin administration. Furthermore, abundance of a phosphorylated form of AQP-2 (S256), known to be required for release of AQP-2 from the endoplasmic reticulum (ER) (186), is decreased compared with controls, resulting in minimal accumulation of AQP-2 in the apical plasma membrane. Taken together, these results show that calcineurin is required for normal phosphorylation of AQP2, and loss of calcineurin activity disrupts AQP-2 trafficking. The mechanism behind these seemingly paradoxical results is presently unclear, but calcineurin may be involved in dephosphorylating AQP-2 during transit from the ER to the Golgi as previously proposed.

7. Nitric oxide synthase

Nitric oxide (NO) is a short-lived gaseous free radical that passes freely through plasma membranes and produces multiple intracellular actions, either directly by nitrosylation of proteins or by means of a soluble guanylate cyclase cGMP-mediated pathway (29, 78, 96, 169) (Fig. 7). NO is formed from l-arginine by three distinct members of the NO synthase (NOS) family: neuronal (nNOS; also known as NOS1), inducible (iNOS; also known as NOS2), and endothelial NOS (eNOS; also known as NOS3). All three NOS isoforms are expressed in the kidney; nNOS and eNOS are localized mainly in renal tubules, collecting ducts, and glomeruli [nNOS is abundantly expressed in macula densa cells, and its expression is stimulated in conditions with high levels of renin (174, 200)], whereas iNOS is found predominantly in renal tubules and collecting ducts (133135).

Mouse models with deletion of single or double NOS isoforms have been extensively studied, but because of the substantial compensatory interactions among the NOS isoforms, the role of NO in the kidney has been difficult to assess (reviewed in Refs. 94, 162, 175). However, elegant studies by Morishita et al. (171) using triple NOS-deleted animals have proven to be extremely informative in our understanding of the role of NO in water homeostasis. These so-called “n/i/eNOS−/−” mice have markedly reduced survival and fertility rates. In addition, they are polydipsic and excrete a large volume of hypotonic urine. Although plasma vasopressin levels are not different from controls, n/i/eNOS−/− mice have a reduced antidiuretic response to exogenous vasopressin and reduced vasopressin-induced cAMP production in the collecting ducts, leading to decreased amounts of AQP-2 in the plasma membrane of collecting duct cells. Over a period of time, the n/i/eNOS−/− mice develop structural abnormalities in glomeruli and renal tubules. Taken together, the phenotype of the n/i/eNOS−/− mice closely resembles NDI. At present, the cause of the defective urinary concentrating mechanism in the n/i/eNOS−/− mice is unknown. Importantly, however, the occurrence of the functional abnormalities preceded that of structural abnormalities. Previous studies have shown that (consistent with the knockout mice results) NO stimulates cAMP production by means of cGMP-dependent activation of adenylate cyclase in isolated rat kidney, presumably in the proximal tubule (90). In addition, it is also conceivable that NO increases cAMP accumulation by interfering with cAMP degradation through cGMP-regulated cAMP phosphodiesterase activity (18). In the collecting duct, there is morphological evidence that NO can promote the membrane insertion of AQP-2 into the apical membrane (reviewed in Ref. 37); however, in isolated perfused CCD tubules, NO has been demonstrated to inhibit vasopressin-stimulated osmotic water permeability (72). Furthermore, in isolated perfused IMCD tubules, cGMP reduces vasopressin-stimulated water permeability (191), and in the TAL, NO has a strong effect to increase the expression of NKCC2 (270). In summary, we speculate that the decreased concentrating ability in the n/i/eNOS−/− mice could be due to the loss of the stimulatory effect of NO on NKCC2 expression and the consequent reduction in countercurrent multiplication in the outer medulla.

8. Endothelin-1

Within the kidney, the collecting duct is the major site of endothelin-1 (ET-1) synthesis (127), where it is thought to be an autocrine inhibitor of AVP-stimulated water reabsorption (similar to bradykinin). Several studies, performed in vitro, have shown that exogenous ET-1, through activation of the ETB receptor, reduces AVP-stimulated water permeability in the CCD (264) and can inhibit AVP-stimulated cAMP accumulation (57, 129) and osmotic water permeability in the IMCD (177, 193). Due to the effects of ET-1 on the renal vasculature (alteration of renal plasma flow and GFR), it has proven to be difficult to assess the role of ET-1 in regulating CD water permeability in vivo. However, low doses of systematically administered ET-1, which do not significantly alter renal hemodynamics, have been shown to increase water excretion (77, 230).

Recently, studies in mice with CD-specific knockout of ET-1 (CD ET-1 KO) have provided direct evidence for a role of ET-1 in regulating renal water excretion (74). Under basal conditions, CD ET-1 KO mice have a normal urine volume, urine osmolality, and plasma Na+ concentration and osmolality. However, plasma AVP concentration is reduced by 35% in CD ET-1 KO mice. Knockout mice also have an impaired ability to excrete an acute but not a chronic water load. Administration of dDAVP results in greater increases in urine osmolality, V2R and AQP-2 mRNA, and AQP-2 phosphorylation in CD ET-1 KO mice compared with controls. Furthermore, IMCD suspensions isolated from CD ET-1 KO mice have enhanced AVP and forskolin-stimulated cAMP accumulation. Taken together, these results show that the CD ET-1 KO has an increased renal sensitivity to the urinary concentrating effects of AVP, indicating that ET-1 functions as an autocrine regulator of AVP action in the CD. Why absence of ET-1 results in these defects in the urinary concentrating mechanism is unknown; however, several possible explanations have been suggested. One explanation is that a reduction in ET-1 results in reduced activation of the ETB receptor and a lack of an ETB receptor-mediated downregulation of cAMP production, thus increasing water permeability. Another explanation is that a lack of ET-1 results in decreased levels of collecting duct NO production, which in itself has been shown to decrease AVP-stimulated water permeability in isolated cortical CD (see above and Ref. 72). However, recent data suggest that neither ET-1-mediated NO production or NO donors inhibit AVP-stimulated cAMP accumulation, suggesting that NO does not mediate ET-1 inhibition of IMCD cAMP production (249).

9. Endothelin A receptor

The mechanism by which ET-1 exerts its effects on the urinary concentrating mechanism (described above) is not fully understood, but it is thought to result from an initial interaction with either of the two ET receptors expressed in the kidney collecting duct, ETA and ETB (47, 128, 197, 257). In vitro, exogenous ET-1 inhibition of AVP is mediated through the ETB receptor (57, 129, 264). However, whether the ETA receptor plays a role in regulation of water excretion in vivo is a subject of controversy. This controversy, in part, is augmented by a relative lack of ETA receptors compared with ETB receptors in collecting duct cells (∼1:4, ETA;ETB in CD) (57). A mouse model with selective deletion of the ETA receptor in the CD (CD ETA KO mice) has been utilized to address this controversy (75).

CD ETA KO mice have no differences in urinary sodium excretion, plasma aldosterone, plasma renin activity, or systemic blood pressure (75). Under basal conditions, CD ETA KO mice have a normal water intake, urine volume, and urine osmolality, but their plasma AVP concentration is increased greater than twofold compared with controls. In contrast to CD ET-1 KO mice (74), CD ETA KO mice have a greater ability to excrete an acute, but not a chronic, water load, and IMCD suspensions from CD ETA KO mice have a reduced AVP- and forskolin-stimulated cAMP accumulation. Taken together, these data suggest that activation of the CD ETA receptor opposes the diuretic effect of ET-1, most likely through increased sensitivity to AVP-induced cAMP accumulation. Furthermore, these results suggest that ET-1 exerts its diuretic actions through either activation of the CD ETB receptor or through paracrine effects.

I. Prostaglandins

1. Group IV cytosolic phospholipase A2

Cytosolic phospholipase A2 (cPLA2) is a member of the phospholipase A2 (PLA2) family, a heterogeneous group of enzymes that are involved in lipid digestion, phospholipid membrane remodeling, and signal transduction (25). cPLA2 catalyzes the first step in eicosanoid synthesis (Fig. 7), the release of arachidonic acid from membrane phospholipids (149). Eicosanoids (e.g., prostaglandins) are thought to play key roles in the kidney, including regulation of sodium and water excretion, renin secretion, and the regulation of renal blood flow and GFR (73, 83, 178). cPLA2 is activated by increases in intracellular calcium, which results in its translocation to intracellular membranes including the nuclear membrane.

To evaluate the physiological role of cPLA2 in the kidney, physiological studies have been performed on cPLA2 knockout mice (cPLA2−/−) (26, 51). cPLA2−/− mice have reduced urinary prostaglandin E2 (PGE2) excretion under both basal conditions and after furosemide treatment, suggesting that a primary source of the arachidonic acid used for renal production of PGE2 is cPLA2. Despite evidence that prostaglandins produced in the renal cortex can modulate GFR and renal blood flow (73), cPLA2−/− mice have a normal GFR, and their fractional excretion of Na+ and K+ is not different from controls. Urine osmolality in cPLA2−/− mice is significantly lower than controls, both under basal conditions and after 48 h of water deprivation. Water deprivation also results in a significantly greater increase in plasma osmolality and a greater percentage loss in body weight in cPLA2−/− mice. Vasopressin does not correct this urinary concentrating defect. cPLA2−/− mice have normal expression of the collecting duct aquaporins AQP-2, AQP-3, and AQP-4, but immunofluorescence staining revealed a marked reduction in staining of apical membrane AQP-1 in proximal tubules. This reduced immunostaining may be due to abnormal trafficking of AQP-1 or abnormal protein folding or protein-protein interactions. However, AQP-1 expression appeared to be normal in the tDL. Thus the concentrating defect is most likely via a different mechanism than seen in AQP-1 knockout mice in which the concentrating defect is believed to be due to reduced AQP-1 expression/function in the descending limb of Henle (158). AQP-1 expression in vasa recta was not investigated in the cPLA2−/− mice, and we speculate that the concentrating mechanism could be due to reduced AQP-1 expression in the DVR.

2. PGE2 E-prostanoid receptors, EP1 and EP3

PGE2 is a locally acting autacoid produced from the enzymatic metabolism of arachidonic acid (see above and Ref. 25). In the kidney, PGE2 is the predominant cyclooxygenase metabolite and the major prostanoid product excreted in the urine. PGE2 has several known effects in the kidney, including inhibiting the actions of vasopressin. For example, in the CCD, PGE2 inhibits cAMP production and reduces vasopressin-stimulated water reabsorption (32). The effects of PGE2 are mediated by four classes of PGE2 E-prostanoid (EP) receptors (EP1 through EP4) (30); cell surface receptors that belong to a large family of G protein-coupled receptors. Each EP receptor not only selectively binds PGE2, but also preferentially couples to different signal transduction pathways, including activation of phosphatidylinositol hydrolysis (EP1 receptor), stimulation of cAMP generation, via Gq (EP2 and EP4 receptors), and inhibition of cAMP generation, via Gi (EP3 receptors).

In the kidney, the EP1 receptor is localized to the connecting tubule and throughout the collecting duct (16, 170), where it can modulate phosphatidylinositol and intracellular calcium concentrations via the G-coupled protein Gq. The role of EP1 receptors in urinary concentrating processes has been examined by the use of EP1 knockout mice (EP1−/−) (116). Despite reduced urinary osmolality and reduced medullary interstitial osmolality after water deprivation, EP1−/− mice responded equivalently to wild-type control mice in concentrating their urine following dDAVP administration. These data, coupled with reduced urinary AVP levels and equivalent AQP-2 trafficking compared with controls, suggest that PGE2 modulates urine concentration via the EP1 receptor, not in the kidney collecting duct, but by promoting AVP synthesis in the hypothalamus.

The EP3 receptor is highly expressed in the medullary and cortical thick ascending limbs (254256), as well as the cortical and medullary collecting duct (31, 250), where it couples to Gi proteins and inhibits adenylyl cyclase activity (Fig. 7). The role of EP3 receptors in renal function has been examined by the use of EP3 knockout mice (EP3−/−) (67). EP3 knockout mice have a normal GFR and renal blood flow, suggesting that EP3 receptors do not have a primary role in regulating hemodynamics in the whole kidney. Under basal conditions, with free access to water, EP3−/− mice have a similar water intake, urine volume, and urine osmolality compared with controls. However, after administration of the cyclooxygenase inhibitor indomethacin, wild-type mice significantly increased their urine osmolality, whereas EP3−/− mice did not, suggesting that under basal conditions, PGE2 modulates the urine osmolality through the EP3 receptor.

Despite evidence to suggest that EP3 receptors mediate the effects of PGE2 to inhibit the actions of vasopressin in the kidney, both water restriction and dDAVP administration had a similar effect on urinary concentrating function in knockout mice and controls. Taken together, these results suggest that the EP3 receptor is not essential for the normal regulation of urinary concentrating mechanisms in response to various physiological stimuli and that EP3-mediated effects of PGE2 on urinary concentration are not directly dependent on vasopressin.

J. Renin-Angiotensin-Aldosterone System

The renin-angiotensin-aldosterone system (RAAS) plays a critical role in maintaining fluid-electrolyte balance (107, 192). In overview, renin is secreted by the kidney and cleaves angiotensinogen to angiotensin I, which is converted by the angiotensin converting enzyme (ACE) to angiotensin II. Angiotensin II binds to AT1 and AT2 receptors in kidney to produce its physiological effects. The RAAS system is involved in regulating renal tubular functions, e.g., modulating sodium/water reabsorption, regulating the renal microcirculation, and regulating aldosterone production by the adrenal glands (275).

1. Renin 1C (Ren1c)

In several mouse strains often used for genetic manipulation studies (e.g., SV129), two renin genes lie in tandem on the same chromosome (Ren2 and Ren1d), complicating the interpretation of experimental data (238, 289). However, a mouse model has recently been developed with genetic deletion of the renin gene in the C57BL/6 strain of mice that, like humans, express only one renin gene (Ren1c). This model has allowed the function of renin in the urinary concentrating mechanism to be examined (253).

Ren1c knockout mice (Ren1c−/−) have no renin mRNA expression in the kidney, and plasma renin is undetectable (253). ANG I and ANG II are also undetectable in plasma, whereas daily urinary aldosterone excretion is <10% of the levels observed in wild-type mice. As predicted, Ren1c−/− mice have a lower blood pressure compared with controls. Structural analysis of kidneys from Ren1c−/− mice revealed thickening of renal arterial walls, fibrosis, and severe hydronephrosis with marked atrophy of the renal medulla. Furthermore, only 20% of homozygote pups live to adulthood, with the remainder dying of dehydration within a few days of birth. Under basal conditions, Ren1c−/− mice have a greater daily water intake, fivefold greater urine volume, and a lower urine osmolality compared with controls. After 12 h of water deprivation, knockout mice lose >20% of their body weight and are still unable to concentrate their urine. The vasopressin analog dDAVP has no effect on urine osmolality in Ren1c−/− mice, whereas it increases osmolality severalfold in wild-type mice. Minipump infusion of ANG II to Ren1c−/− mice restores blood pressure to wild-type levels, but does not restore the kidney's urinary concentrating ability.

Taken together, the NDI observed in the Ren1c−/− mice is likely due chiefly to the medullary atrophy and hydronephrosis that they develop. The cause of these structural abnormalities is unknown, but it seems possible that the renin-angiotensin system plays an important role in renal development.

2. Angiotensinogen

Angiotensinogen (Atg) null mice have been created independently in two different laboratories, and studies of both models have shown a similar phenotype. Thus an overview of the two models is presented (117, 194).

Knockout mice have no detectable angiotensinogen or angiotensin peptides, and therefore lack a functional RAAS. Atg−/− mice have chronic hypotension, increased renal renin expression, and an abnormal renal morphology with marked atrophy of the renal inner medulla. Atg−/− mice have a higher basal urine output and lower urine osmolality than controls, and although 24 h of water deprivation causes a small decrease in urine volume, urine osmolality is unchanged. Administration of dDAVP results in only a small increase in urine osmolality in Atg−/− mice. Furthermore, both urinary excretion and plasma levels of vasopressin are significantly higher in mutant mice than in controls. Taken together, the results indicate that the abnormality in Atg−/− mice is not due to the absence of an ANG II-mediated stimulatory effect on the synthesis and secretion of vasopressin in the hypothalamus, but a form of NDI. The cause of the NDI is likely to be multifactoral, similar to those of the Ren1c−/− mice (see above). The structural abnormality of the renal medulla provides further evidence for a possible role of the renal-angiotensin system on renal development.

3. ACE

Inactive ANG I is converted by the ACE to ANG II (275). ACE exists as a single polypeptide with two homologous catalytic domains that are active against ANG I. In addition, ACE is a membrane-bound ectoenzyme (49), i.e., both catalytic domains are found outside of cells, whereas the remainder of the protein is bound to the cell membrane. Thus, although significant ACE activity is found in plasma, the majority of the enzyme is bound to tissues such as the vascular endothelium (20). ACE is also found in urine bound to exosomes (205).

Two “forms” of ACE null mice have been generated. The so-called ACE.1 strain of mice has a complete absence of both circulating and tissue ACE (60, 138). Whereas ACE.2 mice express a form of ACE that lacks the COOH-terminal half of the molecule, and although catalytically active, the ACE is entirely secreted from cells, resulting in mice with significant plasma ACE activity but no tissue-bound enzyme (61). Both strains of ACE null mice have low systolic blood pressures and an inability to concentrate the urine, both before and after water deprivation. Similar to Ren1c and Agt null mice, ACE.1 mice have major abnormalities in renal structure, including marked medullary and papillary atrophy and greater hydronephrotic lesions (60). In contrast, ACE.2 mice have a relatively preserved renal medulla and papilla (61).

The ACE.2 mouse strain has provided insights into the role of the RAAS system in the urinary concentration mechanism, without the complication of interpreting results in mice with renal abnormalities (e.g., ACE.1, Ren1c, and Agt null mice). Indeed, the greater urine output and reduced urine osmolality in the ACE.2 strain of mice suggests that the RAAS plays a direct role in urinary concentration. It is likely that multiple factors are responsible. Reduced NaCl reabsorption and natriuresis presumably impair concentrating ability by causing an osmotic diuresis. Furthermore, the associated reduction of ANG II levels may alter medullary blood flow, which may affect osmolar gradients (62). Finally, ANG II has been shown to be involved in the regulation of collecting duct urea transport (112), which could be disrupted in the ACE.2 strain of mice.

4. Type 1 angiotensin receptors

Pharmacological studies suggest that the effects of ANG II on water homeostasis are mediated by the type 1 (AT1) angiotensin receptors (263, 275). In mice there are two AT1 receptors termed AT1A and AT1B. Mice with combined AT1A-AT1B receptor deficiency have a marked atrophy of the renal medulla, similar to that seen in ACE.1, Ren1c, and Agt null mice (see above). This is accompanied by a profound inability to concentrate the urine (196, 267). The renal function of mice lacking the AT1A receptor alone has been more thoroughly investigated (195). These mice have a defect in urinary concentration manifested by an inability to increase urinary osmolality to levels seen in controls after thirsting. Basal (free access to water) plasma vasopressin levels of AT1A receptor knockout mice are similar to controls and increase normally after water deprivation, suggesting that the defect in urine concentration is intrinsic to the kidney. However, after dDAVP administration, AT1A receptor knockout mice rapidly and significantly increase their urine osmolality, although the maximal urine osmolality observed is substantially lower than controls. Thus the absence of AT1A receptors does not eliminate the actions of vasopressin to augment water permeability in the distal nephron; instead, the defect appears to be related to the generation of a maximal osmolar gradient.

The cause of the concentrating defect observed in mice lacking AT1 receptors remains unclear. However, similarly to the effects observed in ACE knockout mice (see above), it is unlikely to be the result of renal tubule structural abnormalities, as the apparent inability to concentrate the urine is prevalent in both mice lacking both AT1A-AT1B receptors (major renal hypertrophy) or those lacking AT1A receptors alone (no major structural defects). However, since AT1 receptors are expressed in the proximal tubule (179), where they regulate solute and fluid reabsorption, interruption of the direct epithelial actions of ANG II could alter the distal delivery of solutes, thus urinary concentrating ability. AT1 receptors are also expressed in the renal collecting duct (87), where they are involved in regulation of ENaC activity (203), ENaC abundance (24, 35), AQP-2 expression (142), and the abundance of the UT-A1 and UT-A3 urea transporters (279).

5. Aldosterone synthase

Aldosterone is an important regulator of body fluid homeostasis and electrolyte balance. The role of aldosterone in the kidney has been discussed extensively elsewhere (see Refs. 108, 215), so for the purposes of this review only a basic oversight is given. Aldosterone-dependent regulation of Na+ reabsorption and K+ secretion takes place in distal portions of the nephron: the DCT, CNT, and CCD. Aldosterone regulates several major proteins involved in the urinary concentrating mechanism, including ROMK, ENaC, and the Na+-K+-ATPase (reviewed extensively in Refs. 8, 122, 215, 280).

To assess the effects of decreased amounts or the absence of aldosterone on kidney function, Makhanova et al. (159) disrupted the gene coding for aldosterone synthase (AS) expressed chiefly in the zona glomerulosa of the adrenal cortex. AS catalyzes an essential step in the synthesis of aldosterone, involving the 11β-hydroxylation of 11-deoxycortisol (11-DOC) to form corticosterone, which is further hydroxylated to form 18-hydroxycorticosterone, before oxidation at position C-18 to give aldosterone. AS−/− mice have low blood pressure and abnormal electrolyte homeostasis, including increased plasma concentrations of K+, Ca2+, and Mg2+ and decreased concentrations of HCO3 and Cl. Examination of renal function showed that AS−/− mice have a reduced diluting capacity, a higher urine output, decreased urine osmolality, and an impaired maximal urinary concentrating ability compared with controls (159). In addition, the absence of aldosterone results in several compensatory changes, including induction of the RAAS and elevated plasma glucocorticoid levels.

The impaired urinary concentrating mechanism observed in the AS−/− mice is likely to be the result of a combination of several different factors; however, it is unlikely to be the result of hydronephrosis, as the reduced concentrating capacity was also apparent in heterozygote animals that do not suffer from structural defects in the kidney. The most likely cause of the concentrating defect is that the absence of aldosterone impairs NaCl reabsorption in the CNT and CD (via ENaC), thus reducing the driving force for the osmotic reabsorption of water from the cortical portion of the collecting duct system. In the classic formulation of Stephenson (247), there are three major determinants of concentrating ability: active transport of NaCl from the TAL, water permeability of the medullary collecting ducts, and the rate of fluid delivery to the beginning of the medullary collecting ducts. Decreased activity of ENaC in the CNT and CCD would affect the third factor. In vitro microperfusion experiments have also shown that aldosterone can affect Na+ transport in the TAL, e.g., adrenalectomized rats have a 30% reduction in Na+ reabsorption in the TAL that can be restored by aldosterone treatment, but not glucocorticoids (245, 284). Therefore, lack of aldosterone would be expected to blunt the countercurrent multiplication process. Finally, a marked reduction in the abundance of AQP-3 in the CD has been demonstrated in aldosterone-deficient rats, which may impact on collecting duct water permeability (143).

K. Osmoprotective Genes


The kidney inner medullary tissue is extremely hypertonic, facilitating water reabsorption from the collecting ducts. Renal medullary cells protect themselves from this hypertonic stress by increasing the synthesis or import of several organic osmolytes, including sorbitol (synthesized by the enzyme aldose reductase, see below), betaine (imported by the betaine/γ-aminobutyric acid transporter), myo-inositol (imported by the Na+-dependent myo-inositol transporter), taurine (imported by the taurine transporter), and glycerylphosphorylcholine (reviewed in Refs. 38, 182). The tonicity responsive transcriptional regulation of these genes is modulated by the osmotic response element binding protein (OREBP)/tonicity responsive element binding protein (TonEBP) (126, 167). OREBP/TonEBP is highly expressed in the kidney inner medulla and the inner stripe of the outer medulla, and hypertonic stimulation results in the rapid translocation of OREBP/TonEBP to the nucleus, where it binds with osmotic response elements (OREs) in the promoter region of the osmoprotective genes and stimulates gene transcription (40) (Fig. 9).

FIG. 9.

Putative role of osmoprotective genes in the urinary concentrating mechanism. The tonicity responsive element binding protein (TonEBP) is highly expressed in the kidney inner medulla and the inner stripe of the outer medulla, and hypertonic stimulation (during antidiuresis) results in the activation of TonEBP. This activation requires nuclear redistribution, dimerization, and phosphorylation. In the nucleus, TonEBP binds to TonE enhancer elements and stimulates gene transcription of several osmoprotective genes, including the sodium-myo-inositol cotransporter (SMIT), aldose reductase (AR), and the sodium-chloride-betaine cotransporter (BGT1), allowing renal cells to adapt to the high osmolality by accumulating intracellular osmolytes (reducing intracellular ionic strength). TonEBP also stimulates transcription of the UT-A urea transporter gene, AQP-2, and heat-shock protein 70 (Hsp70). Genetic deletion of either TonEBP or AR results in defective urinary concentrating ability. See text for description.

OREBP/TonEBP knockout mice die during embryonic development; however, a transgenic mouse line that overexpresses a dominant negative form of OREBP/TonEBP (OREBPdn) in the collecting duct has uncovered a novel role of the OREBP/TonEBP in the urinary concentrating mechanism (144). OREBPdn transgenic mice had significantly higher daily urine output and lower urine osmolality compared with nontransgenic controls. However, when subjected to either water deprivation for 24 h, or vasopressin administration, OREBPdn transgenic mice are able to increase urine osmolality, although to a lesser extent than controls. Although OREBPdn transgenic mice have major structural damage to the kidney, it is unlikely that this is the sole cause of the defective urinary concentrating mechanism as measurements of concentrating ability performed before the onset of hydronephrosis also show a defect. OREBPdn transgenic mice have drastically reduced expression levels of UT-A1, UT-A2, and AQP-2 in the inner medulla. These transporters play essential roles in the urinary concentrating mechanism (see above), and it is likely that the polyuria observed in OREBPdn transgenic mice is due, at least in part, to their reduced expression.

2. Aldose reductase

Aldose reductase (AR) is abundantly expressed in the renal medulla (reviewed in Refs. 38, 182). It is a member of the “osmoresponsive genes” and can be regulated by the TonEBP/OREBP. AR catalyzes the conversion of glucose to sorbitol, a major organic osmolyte in the renal medulla. Renal medullary cells accumulate large amounts of sorbitol, and other organic osmolytes, to compensate for the large interstitial hypertonicity that cells are subjected to during osmotic stress. Thus AR is thought to have an osmoprotective function in the kidney. Furthermore, it is also thought to play a role in kidney development (236).

AR-deficient mice (1, 91) have a urinary concentrating defect, with polydipsia and polyuria, which is only partially overcome by vasopressin administration, thus representing a mild form of NDI. The mechanism behind these defects is not fully understood. One group has reported that AR null mice have hypercalciuria, hypercalcemia, and hypermagnesemia and that the hypercalcemia may be partially responsible for the urinary concentrating defect observed (1). However, another group failed to corroborate these findings (91). Recently, by the use of a bitransgenic mouse (295), it has been shown that AR deficiency in the kidney inner medulla alone is responsible for the concentrating defect observed in AR knockout mice, thus ruling out a more general effect of AR gene deletion (ubiquitously expressed in several other tissues including brain) on kidney function. AR null mice have significant renal cellular and structural abnormalities, including cell shrinkage, disorganized tubular and vascular structures, and segmental atrophy. These abnormal structural features in AR knockout mice are most likely the cause of the defective urinary concentrating mechanism.

L. Miscellaneous

1. Foxa1

Foxa1 is a member of the winged helix family of transcription factors and is expressed in the kidney collecting duct (19). Potential Foxa1 binding sites are evident in several genes expressed in the kidney, including the V2R and several subunits of the Na+-K+-ATPase (199, 202). Foxa1 null mice have a urinary concentrating defect, with reduced urine osmolality and volume depletion (19). In addition, administration of vasopressin has no effect on the urinary concentrating ability of Foxa1 knockout mice, indicating a mild form of NDI. Analysis of the mRNA expression levels for several major genes involved in the urinary concentrating mechanism failed to identify a direct downstream target for Foxa1 that could account for the observed phenotype. Thus Foxa1 null mice are a unique model of NDI and provide the first example of a mutation in a transcription factor gene that can lead to a defect in renal water homeostasis. However, further studies are required to assess whether the observed phenotype is due to a primary defect in the kidney or a result of the other metabolic abnormalities observed in these mice.

2. Urate oxidase

Urate oxidase (Uricase) is an enzyme found in liver peroxisomes of several mammalian species and catalyzes the conversion of uric acid to allantoin (286), which is more soluble and easily excreted by the kidney. Uricase-deficient mice have a urinary concentrating defect, with an approximately sixfold greater urine volume compared with controls (115, 285). Furthermore, even after a 12-h water deprivation, uricase-deficient mice cannot concentrate their urine above 900 mosmol/kgH2O. Except for moderate azotemia, adult uricase-deficient mice do not show signs of renal insufficiency, except for a mildly elevated serum urea concentration. It has been suggested that the renal concentrating defect in uricase-deficient mice results from renal damage induced by the excretion of uric acid in amounts that exceed its solubility in urine. However, it is also possible that the urinary-concentrating defect observed is due to osmotic diuresis.

3. Tamm-Horsfall protein

The Tamm-Horsfall protein (THP; uromodulin) is the most abundant protein in urine (9). THP is produced exclusively in the TAL, where it resides in the basolateral membrane, subapical vesicles, and the luminal plasma membrane (10, 11). Luminal secretion of THP into the tubular fluid occurs via cleavage of THP from a glycosylphosphatidylinositol (GPI) membrane anchor (160). THP-deficient mice (THP−/−) have anatomically normal kidneys, steady-state electrolyte balance, a normal kidney diluting capacity, but a reduced creatinine clearance (12). However, THP−/− mice do not concentrate their urine as effectively as wild-type mice when challenged by a 24-h water deprivation. The cause of this defect is unknown; however, the authors speculate that the observed concentrating defect could result from a lack of THP modulation on the function of other transporters expressed in the TAL. However, the finding that the majority of the major renal transporters involved in the urinary concentrating mechanism are upregulated in THP−/− mice argues against this hypothesis.


The Water and Salt Research Centre at the University of Aarhus is established and supported by the Danish National Research Foundation (Danmarks Grundforskningsfond). Support for R. A. Fenton is provided in part by a Marie Curie Fellowship, the Carlsberg Foundation (Carlsbergfondet), the Nordic Council (the Nordic Centre of Excellence Programme in Molecular Medicine), and the Danish National Research Foundation. Funding to M. A. Knepper was provided by the Intramural Budget of the National Heart, Lung, and Blood Institute (National Institutes of Health Project ZO1-HL-01285).


We are grateful to Professor Donald Kohan, Dr. Shinichi Uchida, and Dr. Jurgen Schnermann for critical reading of the manuscript and Ken Kragsfeldt for help with illustrations.

Address for reprint requests and other correspondence: R. A. Fenton, Water and Salt Research Center, Institute of Anatomy, Univ. of Aarhus, Aarhus, Denmark (e-mail: ROFE{at}


View Abstract