Urea and urine concentrating ability: new insights from studies in mice

Baoxue Yang, Lise Bankir


Urea is the most abundant solute in the urine in humans (on a Western-type diet) and laboratory rodents. It is far more concentrated in the urine than in plasma and extracellular fluids. This concentration depends on the accumulation of urea in the renal medulla, permitted by an intrarenal recycling of urea among collecting ducts, vasa recta and thin descending limbs, all equipped with specialized, facilitated urea transporters (UTs) (UT-A1 and 3, UT-B, and UT-A2, respectively). UT-B null mice have been recently generated by targeted gene deletion. This review describes 1) the renal handling of urea by the mammalian kidney; 2) the consequences of UT-B deletion on urinary concentrating ability; and 3) species differences among mice, rats, and humans related to their very different body size and metabolic rate, leading to considerably larger needs to excrete and to concentrate urea in smaller species (urea excretion per unit body weight in mice is 5 times that in rats and 23 times that in humans). UT-B null mice have a normal glomerular filtration rate but moderately reduced urea clearance. They exhibit a 30% reduction in urine concentrating ability with a more severe defect in the capacity to concentrate urea (50%) than other solutes, despite a twofold enhanced expression of UT-A2. The urea content of the medulla is reduced by half, whereas that of chloride is almost normal. When given an acute urea load, UT-B null mice are unable to raise their urinary osmolality, urine urea concentration (Uurea), and the concentration of non-urea solutes, as do wild-type mice. When fed diets with progressively increasing protein content (10, 20, and 40%), they cannot prevent a much larger increase in plasma urea than wild-type mice because they cannot raise Uurea. In both wild-type and UT-B null mice, urea clearance was higher than creatinine clearance, suggesting the possibility that urea could be secreted in the mouse kidney, thus allowing more efficient excretion of the disproportionately high urea load. On the whole, studies in UT-B null mice suggest that recycling of urea by countercurrent exchange in medullary vessels plays a more crucial role in the overall capacity to concentrate urine than its recycling in the loops of Henle.

  • urea transport
  • water transport
  • vasa recta
  • urea clearance
  • knockout mouse
  • renal medulla
  • protein intake

it has long been known that urea plays an important role in the urinary concentrating mechanism and that complex urea movements occur in the kidney. The permeability to urea of the epithelium of specific nephron segments and of the endothelium in some vessels is greatly increased by the insertion in their cell membrane of specialized proteins that allow “facilitated” urea transport when a favorable transmembrane or transepithelial urea concentration difference is present. In the last 12 years, at least seven facilitated urea transporters (UTs) have been cloned, five of which are expressed in the kidney. They belong to two different subfamilies encoded by two different genes, UT-A and UT-B, which share a high homology (for a review, see Refs. 5, 82, 83, and 90). There is also abundant functional evidence for active or secondary active “uphill” urea transport in some vertebrate tissues, including amphibian kidney (32, 57, 88), elasmobranch gills (30), ventral skin of some toads (25, 27, 76), mammalian inner medullary collecting duct (IMCD) (4, 41, 42, 44, 85), but none of these active transporters has yet been cloned.

Rat UT-B protein has 62% identity to rat UT-A2 and has a similar membrane topology (based on hydropathic analysis). It was first cloned from human bone marrow cells and initially called hUT-11 (70, 110). In subsequent studies in the rat, it was called UT11 (73, 74, 101) and UT3 (103, 113), but there is now a general agreement for the nomenclature proposed by Sands et al. in 1997 (5, 86, 90). UT-B is highly expressed in the membrane of erythrocytes, in which facilitated urea transport was known for a long time. It is also abundantly expressed in the endothelium of the arterial vasa recta throughout the renal medulla, where it contributes to the urinary concentrating mechanism. The UT-B protein has been shown to carry the Kidd blood group antigen in humans (60, 69), and mutations that induce a loss of function of UT-B result in the lack of this antigen (a very rare condition) (59, 93), a dramatically reduced urea permeability of red blood cells (RBCs) (33), and a mild urinary concentrating defect (84). UT-B is also expressed in a number of other organs or tissues including bone marrow, spleen, fetal but not adult liver, mouse gastrointestinal tract, human colon, urothelium of the renal pelvis, ureter and bladder, astrocytes in the brain, Sertoli cells of the testis, prostate, and endothelial cells of some vascular beds. The existence of carrier-mediated urea transport in all organs other than RBCs and the kidney was not suspected before the molecular identification of urea transporters, and the function it serves in these organs remains to be elucidated.

We recently generated UT-B null mice, representing the first model of a urea transporter deletion. These mice allowed a better characterization of the role of UT-B in kidney function and in the urinary concentrating mechanism (8, 111). In addition, the generation of double null mice for both UT-B and aquaporin-1 (AQP1) allowed a precise evaluation of the capacity of UT-B to transport water. Another group has recently generated UT-A1/3 null mice (28). This review will summarize the information that has been gained by the molecular identification and precise localization of UTs and by the phenotypic analysis of UT-B null mice and their adaptation to various interventions.


Urea Handling in Mammals

The catabolism of carbohydrates and lipids generates carbon dioxide and water that can be excreted easily by the lung and kidney, respectively. In contrast, protein catabolism generates several waste products, the most abundant of which is nitrogen. In mammals, most of the nitrogenous wastes are excreted in the urine in the form of urea. Because of the proportion of proteins in their diet relative to that of other nutrients and minerals, urea represents ∼40–50% of all urinary solutes in humans on a Western-type diet and about the same or even more in laboratory rats and mice, depending on the composition of their diet. Table 1 illustrates in numbers the special features of urea handling compared with that of sodium in different mammals. It also shows the wide differences among humans, rats, and mice in urine osmolality and the relative “concentrating effort” of the kidney, best illustrated by the amount of “solute-free” water (TcH2O) reabsorbed from the glomerular filtrate compared with the urine output. In contrast to that of sodium and chloride, the blood urea level is relatively low (5–10 mmol/l for a normal-protein diet), and the urea concentration in the urine may be 20–100 times higher than in the blood in humans, and up to 250 times in rodents. NaCl is usually not, or only modestly, concentrated in the urine (up to twice the plasma level). Potassium, which is secreted actively in the collecting duct (CD) lumen, is usually concentrated 5–30 times above plasma level. Only ammonium and protons that have very low plasma levels and are known to be secreted actively can be concentrated much more than urea (up to 1,000-fold the plasma level), but they represent a far smaller contribution to the total urinary osmoles. Thus a very large fraction of the concentrating effort of the kidney is devoted to the concentration of urea (7, 10).

View this table:
Table 1.

Comparison of water, urea, and sodium handling in mice, rats, and humans

Most rodents concentrate their urine more than humans and, among rodents, mice more than rats (Table 1). Actually, the mouse kidney exhibits special anatomic-functional adaptations that improve its concentrating ability (9, 52). Another feature that deserves attention is the scaling of food intake/metabolic rate according to size (17, 95). The smaller the size, the higher are these parameters and the higher the excretory needs in relation to body weight (Table 1).

Localization of UT-B and Other UTs in the Kidney

In the kidney of all species that have been investigated so far, i.e., rats, mice, rabbits, humans, monkeys, UT-B has been found exclusively in the endothelium of the descending (arterial) vasa recta (DVR) throughout the renal medulla (Fig. 1A) (58, 74, 99, 102, 103, 110). It is not expressed in the ascending (venous) vasa recta (AVR) (which have a fenestrated endothelium) nor in the cortical capillaries or in the capillary plexuses supplying blood to tubules in the inner stripe (IS) of the outer medulla (Fig. 1, AC). UT-B is not expressed in epithelial cells of the nephron or in interstitial cells either. But it is expressed in the papillary surface epithelium, the pelvic epithelium (Fig. 1A) and the urothelium of the ureter and bladder. In DVR, UT-B protein is inserted in the basolateral and luminal membranes of endothelial cells (Fig. 1, B and C).

Fig. 1.

Localization of urea transporter UT-B protein in rat kidney. A. in situ hybridization of UT-B mRNA in kidney. Adapted from Ref. 103. C, cortex; OS and IS, outer stripe and inner stripe of the outer medulla, respectively; IM, inner medulla. Note the lack of UT-B in C and OS, the typical bundles of radially oriented labeled structures within the IS vascular bundles, and the more dispersed, but still radial, labeling in the IM. UT-B expression disappears in the very tip of the papilla (the region in which UT-A1 is expressed along the collecting ducts). Punctuated labeling seen in the upper IM is due to the fact that the kidney section was not fully parallel to the axis of the vessels so that they are cut more tranversally in this region. Note also that the papillary surface epithelium and the urothelium of the pelvic wall express UT-B mRNA. B and C: immunofluorescence localization of UT-B protein in IS seen in cross section (B) and oblique section (C) of a vascular bundle, respectively. Adapted from Ref. 38. Note that only some vessels are labeled corresponding to arterial (descending) vasa recta. Venous (ascending) vasa recta that surround the arterial vasa recta do not express UT-B. Nephron segments around the vascular bundle also do not express this urea transporter. The small labeled round figures in C correspond to red blood cells trapped in capillaries. Scale bars = 20 μM.

Several other UTs are present in the kidney and localized in very specific parts of the nephron but not in the vasculature (Fig. 2). They all belong to the UT-A family. UT-A1, UT-A3, and possibly UT-A4 are expressed in the most distal part of the CD located in the deepest part of the inner medulla, and UT-A2 is expressed exclusively in a limited portion of the thin descending limbs of Henle's loops (TDL). UT-A1 is the vasopressin-regulated urea transporter that accounts for the increase in urea permeability of the luminal membrane of terminal IMCD cells (66, 92) as confirmed by recent studies in UT-A1/3 null mice (28). UT-A2 message and protein have been observed in the lower half of TDLs of short loops of Henle in the IS of the outer medulla (66, 106). Interestingly, AQP1 is restricted to the upper half of this structure (65, 106). This spatial dissociation of water and urea transport favors a more efficient accumulation of solutes in the inner medulla in two ways. It prevents the dilution of solutes in deeper regions of the medulla by allowing a water short-circuit in the upper IS, and it favors the return of urea to the inner medulla by recycling this solute in the deep IS (101). UT-A2 message has also been localized in the inner medulla (73), lining a short portion of the long-loop TDLs (91, 101). However, UT-A2 protein is not expressed in these long TDLs under normal conditions (65). Some expression can, however, be induced in long-loop TDLs in the upper part of the inner medulla by chronic infusion of dDAVP (a potent antidiuretic agonist of vasopressin) (106).

Fig. 2.

Diagram depicting the vascular and tubular routes of urea recycling within the kidney. A short and a long loop of Henle are depicted within the 4 kidney zones, along with an arterial (descending) and a venous (ascending) vasa recta (DVR and AVR, respectively). C, OS, IS and IM are defined as in Fig. 1. Urea delivery to the IM and urea transit in AVR are shown in green. The pathways allowing urea to return to the IM are indicated by red arrows for the vascular route (V) and by blue arrows for the tubular route (T). In normal mice, concentrated urea is delivered to the tip of the papilla by the terminal part of the collecting ducts expressing the vasopressin-regulated urea transporters UT-A1/3 and possibly UT-A4. Urea is taken up by ascending blood (plasma and red blood cells) in AVR, and a significant fraction of it is returned to the IM by being reintroduced either in the DVR (expressing UT-B) or in the descending thin limbs of Henle's loops (expressing UT-A2).

It may be interesting to underline that the vascular and tubular structures that drive blood and tubular fluid down into the medulla, the arterial vasa recta and the thin limbs, exhibit the same water channel (AQP1) but different UTs (UT-B and UT-A2, respectively) encoded by two different genes. In contrast, RBCs exhibit the same water channel and the same urea transporter as the vasa recta. On the other hand, all facilitated UTs expressed along the nephron (UT-A1 and UT-A3 in the CD, UT-A2 in TDL) are encoded by the same gene and produced by alternative splicing or distinct promoters. These differences may originate from successive evolutionary steps.

Role of Urea in the Urine Concentrating Mechanism

In the 1970s-1980s, a number of clearance, micropuncture, and microperfusion studies as well as anatomic-functional correlations have brought a good understanding of the role of urea in the urinary concentrating mechanism (16, 87) (see reviews in Refs. 6 and 50). It has been understood that urea is accumulated and somehow “sequestrated” (104) in the renal medulla at a concentration increasing from the outer medulla to the tip of the papilla. This accumulation results from three associated processes (Fig. 2). 1) Urea becomes progressively concentrated along the CD because of vasopressin-dependent water reabsorption in a segment poorly permeable to urea, thus bringing a highly concentrated urea solution to the terminal CD. 2) A vasopressin-dependent increase in urea permeability of the terminal IMCD (due to UT-A1/3) enables this concentrated urea to be transported into the interstitial tissue of the deep inner medulla. 3) Medullary urea, which continuously tends to escape the inner medulla via the ascending venous vasa recta blood, is continuously returned to the inner medulla by a complex intrarenal urea recycling process involving reentry of urea along the DVR (through UT-B) and specialized sections of the loops of Henle (through UT-A2). Thus urea molecules are not really sequestrated in the medulla but are continuously cycling back to the inner medulla by vascular and tubular routes (10). The discovery of the molecular structure of UTs and subsequent in situ hybridization and immunolabeling of kidney sections allowed a more precise localization of the facilitated UTs along the nephron and renal vessels than that deduced from micropuncture and microperfusion studies but did not suggest any yet unsuspected intrarenal urea movements. Evidence for intrarenal urea recycling initially came from the demonstration that, in the antidiuretic kidney, more urea flows through the early distal tubule (at the exit of the loop of Henle) than in the late proximal tubule (the last accessible segment before the loop of Henle) (7, 10). The close association of DVR and AVR and paucity of interstitial tissue in the vascular bundles of the IS of the outer medulla suggested intense countercurrent exchange of water and solutes (including urea) between arterial and venous blood (going to and coming from the inner medulla). Such countercurrent exchange also occurs in the inner medulla, where all ascending and descending structures (vessels and loops of Henle) run in parallel. The close proximity of the TDLs of short-looped nephrons to the vascular bundles in the IS in the rat kidney suggests the importance of countercurrent exchange of urea between AVR and TDLs in the IS. Actually, in vitro studies of isolated perfused rat vasa recta suggest that urea permeability of DVR is much higher in the outer than in the inner medulla (72). In the mouse and in some desert-adapted rodents, TDLs are even intermingled among DVR and AVR within “complex” vascular bundles, making these exchanges even more efficient (9, 51). A positive correlation has been observed among different mammalian species between the intensity of urea recycling in short-looped nephrons (evaluated by urea delivery in the early distal tubule) and the proximity of short-loop thin limbs to the vascular bundles (105).

The concentrating activity of the kidney depends in large part on the accumulation of NaCl in the medulla to levels that are two- to fivefold higher than in peripheral blood. This accumulation is achieved by an energy-demanding countercurrent multiplication process (not to be confounded with passive countercurrent exchange). This multiplication results from active reabsorption of NaCl from the water-impermeable thick ascending limbs and the entry of NaCl in the nearby sodium-permeable TDLs (mostly of long loops). An additional contribution to the overall concentrating process is provided by the accumulation of urea, as described above. However, without NaCl, urea could not be accumulated in the medulla because there would be no osmotic driving force to remove water in the upper part of the medullary CD and to deliver concentrated urea to the terminal IMCD where the vasopressin-dependent UT is expressed (10).

Birds, which can concentrate urine, although to a lesser extent than mammals, possess only one kind of Henle's loop (and only in their deep nephrons) (15). Their TDL exhibits high sodium permeability like that of long loops in mammals and can accumulate only one solute, NaCl, in their medullary cones (6, 15, 67, 68). In mammals, there are two types of loops and two solutes accumulated in the medulla. The loops of Henle of the deep nephrons have lengthened to form an inner medulla, and the superficial nephrons have also developed loops (“short loops”) that play an important role in urea accumulation because of the high permeability to urea of their thin limbs (due to UT-A2). Thus evolution in mammals has probably favored concomitantly the development of Henle's loops with high permeability to urea in superficial nephrons and the development of a special terminal CD epithelium with vasopressin-regulated urea permeability. This allows mammals to keep a low urea concentration in their internal milieu and to excrete urea without excessive water needs, by concentrating it in the urine.

Immunolabeling studies provide additional support for the concept that short loops may indeed play a crucial role in intrarenal urea recycling (and thus in water economy) because they revealed that UT-A2 protein is strongly expressed in these loops but not detectable in long loops. In addition, the permeability to urea of the short-loop TDLs may be higher than what was measured by in vitro microperfusion (40) because it was not known at that time that UT-A2 was expressed only in the lower half of this segment (66). Because the tubules, during the microdissection, are usually identified by their attachment to the terminal pars recta (which is easily recognized) and broken at various lengths before the bend of the loop, the TDLs studied by microperfusion probably included all of the UT-free early portion and probably only a fraction of the terminal portion expressing the UT protein. Finally, like in all rodents, short loops largely outnumber long loops in the mouse (9, 52), contrary to what is sometimes believed. Consequently, it may be assumed that, within the kidney as a whole, urea recycling through the short loops is largely predominant over that through the long loops.

Function of Facilitated Urea Transport in RBCs

The high permeability of RBCs to urea, now known to depend on intense UT-B expression, may serve two different functions. First, it could protect these cells from repeated osmotic stresses when they flow through the vasa recta. Second, it may contribute to intrarenal urea recycling and thus to more efficient urine concentration (61).

Because of their high (AQP1-dependent) water permeability, RBCs would undergo severe shrinkage when exposed to the high urea concentration of the inner medulla, were their membrane not also highly permeable to urea. A mathematical simulation by Macey and Yousef (62) predicted that RBCs should shrink during their transit in the hyperosmotic medulla but that they would subsequently swell on leaving the medulla if they did not possess a high urea permeability. However, one should consider that, statistically speaking, only a very small fraction of whole body RBCs are exposed to this risk at a time. Inner medullary blood flow is probably about one-quarter of the total medullary flow, which is only ∼10% of total renal blood flow, and this renal blood flow itself is ∼20% of cardiac output. Thus only ∼0.5% of whole body RBCs traverse the inner medulla at any time. This theoretical result is in keeping with results obtained in vivo in rats. Red cell survival was not altered significantly when urine concentrating activity was chronically enhanced or reduced over a large range (55). On the whole, UT-B expression in RBCs is certainly a favorable adaptation, but its beneficial effect on red cell life span does not seem to be crucial.

Another, probably more important role of UT-B in RBCs might be to increase the mass of urea involved in the vascular route of urea recycling. The mathematical simulation of Macey and Yousef (61, 62) suggests that rapid urea uptake by RBCs in the inner medulla and subsequent urea release in more superficial medullary regions during their transit in the medullary circulation contribute to the maintenance of the urea gradient in the medulla. Rapid equilibration of urea between plasma and RBCs in the vasa recta lumen allows a larger mass of urea to be transported by the medullary blood and to be available for rapid countercurrent exchange between AVR and DVR. As a consequence, whole blood volume, not just plasma volume, is able to participate in the vascular urea recycling pathway. The hematocrit is only ∼35% in renal medullary vessels. Thus RBCs should contribute to about one-third of the total urea transfer between AVR and DVR.

Of note, the transit time of blood is much longer in AVR than in DVR because AVR are more numerous and show wider lumens, thus exhibiting a much larger total cross-sectional area at any level of the medulla. In addition, studies in human RBCs showed that urea efflux is unaffected by intracellular urea concentration, contrary to urea influx, which decreases with decreasing urea concentration (13). This should allow efficient exit of urea from RBCs even when intracellular urea concentration declines, as cells progressively ascend in AVR. As a result, the amount of urea returning to the renal vein is minimized, and a greater fraction of the medullary urea can be recycled within the kidney instead of returning to the general circulation.

A significant role for RBCs in the urine concentrating mechanism is supported by experiments that have been carried in isolated, perfused rat kidneys (56). With dDAVP and 5 mmol/l urea in the perfusate, kidneys were unable to produce hyperosmotic urine and exhibited necrosis of the thick ascending limbs when perfused with erythrocyte-free medium. With erythrocytes at a hematocrit of 40–45%, urine osmolality rose to about twice that of the perfusate. The restoration of a significant urinary concentrating capacity in vitro probably resulted not only from improved oxygenation of the energy-demanding thick ascending limbs but also from a contribution of RBCs to medullary urea recycling.

A comparison of different vertebrate groups brings further support to the concept that expression of UT-B in RBCs is related to the capacity to concentrate urea in the urine. Birds, which concentrate urine only modestly but do not concentrate urea (they excrete most of their nitrogen end products as uric acid), express facilitated water transport but no facilitated urea transport in their RBCs (14). Two unusual fish, the gulf toadfish (Opsanus beta) and the tilapia (Alcolapia grahami), which excrete most of their nitrogenous waste as urea but have no urine concentrating mechanism, also do not express facilitated urea transport in their RBCs (107, 108).


Until the availability of knockout models for urea transporters, the extent to which intrarenal urea recycling contributed to the urinary concentrating process could not be precisely evaluated. The relative importance of UT-A1-dependent delivery of urea to the tip of the papilla and of the two intrarenal recycling pathways, depending on UT-A2 in the loops of Henle and on UT-B in the vasa recta and RBCs, was also unknown.

Generation of the UT-B Null Mouse and Main Features of Its Phenotype

The human UT-B gene (gene name Slc14a1) contains 10 exons (gene ID 6563). The translation start codon ATG is located in exon 3. The rat UT-B gene contains nine exons (gene ID 54301). Its translation start codon is in exon 6. In the mouse, this gene is located at chromosome 18E3 (size = ∼40.5 kb) and contains nine exons (gene ID 108052). The translation start codon ATG is located in exon 2. UT-B knockout mice were generated by targeted gene deletion of exons 4–7 shown in Fig. 3A (111). Heterozygous founder mice containing the disrupted UT-B gene were bred to produce wild-type, heterozygous, and UT-B null mice. Northern blot analysis shows only truncated UT-B transcripts in the kidneys of knockout mice (Fig. 3B). Immunoblot analysis revealed UT-B protein in the kidneys of wild-type but not in those of UT-B null mice (Fig. 3C). A 98-kDa band is detected in the kidney by some (29, 99) but not all UT-B antibodies (102). This band can be competed away by the immunizing peptide, indicating that the antibody recognition is specific. However, this band is present in kidneys from UT-B null mice (47), indicating that the 98-kDa band is not UT-B. Thus the molecular identity of this band is still unknown (47). Immunohistochemistry shows UT-B protein expression in medullary vasa recta of wild-type mice but no staining in the medulla of UT-B null mice (111). Urea permeability in erythrocytes from null mice is 45-fold lower than those from wild-type mice, indicating that UT-B was functionally deleted and that RBCs do not express to a significant extent other proteins facilitating urea transport (Fig. 3D). Measurements of urea permeability in vasa recta of UT-B null-mice are not yet available, but it may be assumed that, like AQP1 for water transport, there is only one urea transporter in these vessels. Thus their permeability to urea is probably reduced to the level observed in cells not expressing facilitated urea transport.

Fig. 3.

Generation of UT-B null mice. A: targeting strategy for UT-B gene deletion. Rectangles indicate exon segments. Homologous recombination results in replacement of the indicated segments (solid horizontal bar) of the UT-B gene by a 1.8-kb pol II-neo selectable marker. B: Northern blot of mouse kidney probed with the mouse UT-B coding sequence. C: immunoblot of mouse kidney inner medulla with UT-B polyclonal serum. D: erythrocyte urea permeability. Urea permeability was measured from the time course of erythrocyte volume as determined by light scattering in response to a 250 mM inwardly directed urea gradient. Adapted from Ref. 111.

UT-B null mice exhibit a grossly normal appearance, activity, and behavior. UT-B null mice have body weights similar to those of wild-type mice up to age 4 wk but seem to gain significantly less weight thereafter. Their kidneys have a normal appearance, and the kidney weight-to-body weight ratio is not different from that of wild-type mice at any age (47). Plasma sodium, potassium, chloride, bicarbonate, and creatinine concentrations are unaltered compared with wild-type mice, but plasma urea is ∼30% higher and urine concentrating ability is reduced ∼30% (see below) (111).

Transport Properties of UT-B

When expressed heterologously in Xenopus laevis oocytes, UT-B significantly increases urea permeability by 12-fold, thus showing its capacity to facilitate urea transport through cell membranes (113). Interestingly, UT-B (but not UT-A isoforms) was also found to increase osmotically driven water transport. The measured low urea reflection coefficient of UT-B and the fact that this transport was blocked by urea transport inhibitors suggest a common water/urea pathway (113). These findings are in agreement with previous observations showing that urea and water may share, at least to a small extent, a common pathway in RBCs (18, 100). Increased water permeability in X. laevis oocytes expressing UT-B was subsequently confirmed by Sidoux-Walter et al. (94). However, after studying oocytes expressing different levels of UT-B, these authors concluded that UT-B-facilitated water transport occurs to a significant extent only when UT-B is expressed at nonphysiologically high levels (94). Studies in genetically engineered mice have clearly shown that AQP1 provides the major pathway for water transport (112). Nonetheless, some studies suggest that there is also a significant, presumably AQP1-independent, protein-mediated mercurial-insensitive water permeability in erythrocytes (23) and in outer medullary DVR (71). Because these two tissues express UT-B constitutively in addition to AQP1, these observations suggest that UT-B might be permeable to water even in physiological conditions.

Studies in double knockout mice lacking both UT-B and AQP1 have allowed us to settle the issue by quantifying UT-B-mediated water transport in erythrocytes lacking the major water transporter AQP1 (114). The osmotic water permeability in erythrocytes from these double null mice was 4.2-fold lower than that in erythrocytes from mice lacking AQP1 alone (Fig. 4A). A similarly low water permeability was found in erythrocytes from AQP1 null mice after UT-B inhibition by phloretin and in erythrocytes from UT-B null mice after inhibition of AQP1 by HgCl2. These studies thus confirm that UT-B may be responsible for a small fraction of water transport in RBCs. Figure 4B summarizes the respective contributions of protein and lipid pathways for water and urea transport in mouse erythrocytes at 10°C. At 37°C, ∼79% of water is transported through AQP1, 6% through UT-B, and 15% through the lipid membrane. The vast majority of urea is transported through UT-B (112).

Fig. 4.

Erythrocyte water permeability measured by stop-flow light scattering. A: osmotic water permeability (Pf) measurements based on the time course of changes in erythrocyte volume in response to a 250 mM inwardly directed sucrose gradient. These data were obtained in erythrocytes from mice of indicated genotype at 10°C, under control conditions and in the presence of 0.3 mM HgCl2 or 0.7 mM phloretin. Values are means ± SE; n = 3 mice/genotype. *P < 0.01 compared with no inhibitor. #P < 0.01 compared with wild-type in control conditions Pf. B: relative contributions of aquaporin-1 (AQP1), UT-B, and the lipid bilayer to erythrocyte water and urea permeability at 10°C. Reproduced from Ref. 114.

These data concern water and urea transport in whole erythrocytes. However, these cells express AQP1 and UT-B at much different levels (∼200,000 copies/cell for AQP1 vs. only 14,000 copies for UT-B, a 14-fold difference). These numbers allowed us to calculate the intrinsic (single channel) water permeability of AQP1 and UT-B proteins. At 37°C, AQP1 contributes 13 times more than UT-B to erythrocyte water permeability (79 vs. 6%; see above). Thus the single-channel water permeabilities of individual UT-B and AQP1 molecules in the configuration they take when inserted in the erythrocyte membrane are very close (114).

Although the single-channel water permeability of UT-B is relatively high, water movements through UT-B in vivo should not be physiologically significant because of its much lower level of expression compared with that of AQP1 (at least in RBCs), as also predicted by a recent model study (116).

Consequences of UT-B Deletion on the Expression of AQPs and Other UTs in the Kidney

Because of the complex movements of urea and water in the kidney and of the role played by urea in water conservation, it was interesting to determine whether the lack of one of the UTs could affect the expression of other UTs and possibly also that of AQPs. mRNA and protein abundances of UT-A1, -2, and -3, and of AQP1, -2, -3, and -4 were evaluated in UT-B null mice (47). These experiments revealed AQP- and UT-selective changes. No change was observed in mRNA and protein abundance of AQP1, the aquaporin coexpressed with UT-B in vasa recta and RBCs. In contrast, AQP2 and AQP3 mRNAs were both upregulated (Fig. 5A). With regard to UTs, mice of the two genotypes had similar UT-A1 or UT-A3 abundance. In contrast, UT-A2 in the medulla was upregulated at both the mRNA (Fig. 5A) and protein levels (Fig. 5B) (47). If the abundance of the urea transport proteins is a limiting factor for transepithelial urea transport, the above observations suggest that urea delivery to the tip of the papilla is not increased in UT-B null mice but that the recycling of urea in the TDLs may be enhanced when that in the DVR is compromised by the lack of UT-B.

Fig. 5.

Expression of AQPs and other UTs in kidney of UT-B null (−/−) and wild-type mice (+/+). A: transcript expression of aquaporins and urea transporters. Real-time PCR was carried using a LightCycler and with LightCycler FastStart DNA MasterPLUS SYBR Green I kit. mRNA expression level for each sample is expressed relative to wild-type kidney, which is arbitrarily considered equal to 1.0 in each individual comparison. Values are means ± SE; n = 6 mice/genotype. *P < 0.01. B: UT-A proteins in kidney of wild-type and UT-B-null mice. Each lane was loaded with protein (10 μg) extracted from the kidneys of a single mouse. The arrows indicate positions of the glycosylated group (45–60 kDa) comprising UT-A2 in the OM, and of the 117- and 97-kDa UT-A1 in the IM. Adapted from Ref. 47.

The mechanism responsible for the upregulation of UT-A2, AQP2 and AQP3 is not known, but elevated vasopressin levels may play a role. In relation to their urine concentrating defect, UT-B null mice have higher plasma osmolality than wild-type mice (see below) and thus probably increased vasopressin secretion. Vasopressin has been shown to increase the transcription and the protein abundance of AQP2 and AQP3 (1, 26, 46, 81, 98) and to increase UT-A2 mRNA (73) and protein (106) in the outer medulla and upper inner medulla in rats. These effects on AQPs in the CD may be direct, but because V2 receptors are not expresssed in thin limbs, the upregulation of UT-A2 probably depends on local physicochemical stimuli secondary to vasopressin action on the CD, as discussed elsewhere (73).

Urea Handling in the UT-B Null Mouse in Basal Conditions

Several physiological variables were measured in plasma and urine of wild-type and UT-B null mice under basal conditions (food intake was equal in the 2 genotypes) (8). As previously mentioned, UT-B null mice have normal renal function except for a moderate defect in their urinary concentrating ability. This defect is probably responsible for their slightly higher plasma osmolality (+6 mosmol/kgH2O). Plasma creatinine and creatinine clearance (an index of glomerular filtration rate) were similar in both groups (Table 2), suggesting that UT-B deletion has no influence on glomerular hemodynamics. In agreement with this finding, kidney weight was also not influenced by UT-B deletion. The clearance and fractional excretion of sodium and potassium were also similar in both groups. In contrast, urea clearance was significantly lower in UT-B null mice than in wild-type mice by ∼25% (Table 2) (111).

View this table:
Table 2.

Urea handling in wild-type and UT-B knockout mice in normal conditions

In basal conditions, urine osmolality of UT-B null mice was significantly lower than in wild-type mice, amounting to 1,530 vs. 2,060 mosmol/kgH2O in a first series of mice (111) and 1,780 vs. 2,650 mosmol/kgH2O in another series (8). Thus UT-B deficiency results in a 30% decline in whole urine concentrating ability (Table 2). Urea concentration in urine was reduced in the same proportion as that of other osmoles (as obviously expected when the same amount of each solute is simultaneously excreted in a larger volume of urine). However, because UT-B null mice exhibit a significant elevation of their plasma urea level (Table 2), the concentration of urea in their urine with respect to plasma, the urine-to-plasma ratio (U/P), was more severely reduced than that of other solutes or whole urine osmolality. The U/P urea ratio was reduced to less than half that in wild-type whereas the U/P ratio for total osmoles was reduced by only about 30%. Thus UT-B null mice exhibit a “urea-selective” urinary concentrating defect. In response to 24-h water deprivation, urine osmolality rose significantly less in UT-B null mice than in wild-type mice (by 870 mosmol/kgH2O vs. 1,380 in wild-type) (Fig. 6). However, the ratio for urine osmolality in the two genotypes (UT-B/wild-type) remained similar to that in basal conditions (∼0.70).

Fig. 6.

Urinary concentrating defect in UT-B null mice. Osmolality measured in 24-h urine of mice of the 2 indicated genotypes given free access to food and water (basal) and the next day during 24-h water deprivation (no water bottle but free access to food). Each solid line represents 1 mouse. Values are means ± SE of 10 mice in each group. *P < 0.001 compared with wild-type mice during the same condition. The rise in urine osmolality in response to water deprivation was significantly smaller in null mice than in wild-type (P < 0.02). Reproduced from Ref. 8.

The opposite changes in plasma and urine urea concentrations suggest that plasma urea increased because the kidney was less able to accumulate urea in the medulla and thus to concentrate it in the urine. Measurements of inner medullary concentration of total osmoles, chloride, and urea confirmed this conclusion (111). Total osmolality in medullary tissue was 36% less in UT-B null mice than in wild-type mice. However, this reduction was mainly due to a deficit in urea accumulation, which was reduced by 55%, whereas chloride concentration (reflecting NaCl accumulation) was reduced only by 19% (not significant) (Fig. 7).

Fig. 7.

Concentration of osmoles, urea, and chloride in the inner medulla of UT-B null and wild-type mice in basal conditions. *P < 0.001 compared with wild-type mice. Tissue osmolality in UT-B null mice is reduced by ∼37%, but this is mainly due to a deficit in urea (−53%), with chloride (and likely sodium) accumulation being almost normal. Adapted from Ref. 111.

Comparison of Urea and Creatinine Clearances: Evidence for Urea Secretion

During the study of wild-type and UT-B knockout mice, we observed that urea clearance was higher than creatinine clearance (Table 2). This suggests that some of the excreted urea originates from tubular secretion (i.e., uphill transport from blood to lumen) in addition to filtration. This observation raises two questions. First, is creatinine clearance in this study a reasonable estimate of GFR? Second, how important is this secretion and why is it apparent in mice when the contemporary thinking is that urea does not undergo significant active secretion in mammals?

Creatinine clearance is not a perfect marker of GFR because some creatinine may be reabsorbed along the nephron. But this reabsorption is usually modest, at least in rats, rabbits, dogs, and humans. The enzymatic technique used for measuring creatinine in our study is free of artifacts and gave a plasma creatinine concentration of 17.2 μmol/l (8), very similar to that recently reported by HPLC measurement in mice (17.7 μmol/l) (75). Dunn et al. (24) have recently validated the use of creatinine clearance as a measure of renal function in mice when creatinine is measured with appropriate techniques. Based on several studies using validated FITC-inulin or creatinine clearances, GFR in normal, conscious mice (measured over 24 h) ranges from 140 to 260 μl/min, i.e., 200 to 370 ml/day (24, 39, 75). In the study by Bankir et al. (8), urea clearance (417 ml/day) was distinctly higher than this level of GFR (Table 2). Even if creatinine clearance underestimated GFR to some extent in their study, fractional excretion of urea (FEurea = urea clearance × 100/GFR) appears to exceed 100% in wild-type mice (vs. only 30–70% in other mammals), thus supporting the concept of significant secretion of urea in this species.

Actually, urea clearance does not need to exceed GFR to indicate the existence of significant urea secretion. Because ∼50% of filtered urea is always reabsorbed in the proximal convoluted tubule, whatever the level of diuresis (3), and returns to the cortical blood and then to the renal vein, any figure of FEurea above 50% (and not 100% as often assumed) suggests that net tubular secretion occurs (19). Now, FEurea reaches 60–70% during water diuresis in rats, dogs and humans, and values above 100% have even occasionally been reported in rats and dogs (12, 20, 43, 48, 77, 78, 89, 96). The fact that FEurea is much higher in mice than in other mammals, even exceeding 100% in basal conditions, is probably related to the huge turnover of urea that imposes on the kidney a considerably larger urea load to excrete, due to the more intense metabolism and much greater food intake per unit body weight observed in smaller animals (see Table 1 and Ref. 8). Urea clearance in excess of creatinine clearance or GFR has also been observed in rats, but only in special experimental conditions: 1) during chronic high-protein or high-protein/high-salt feeding (12, 48); and 2) during chronic urea overload induced by supplying a 700 mmol/l urea solution as the only drinking fluid (77). Interestingly, this situation also induced an increase in the diameter of proximal tubules (35).

Uphill urea secretion has been well characterized in amphibians, in which it allows urea to be concentrated in the urine five- to sevenfold above the level in plasma despite the inability of their kidney to concentrate urine as a whole (31, 32, 57, 88). Several whole animal or human studies suggest that active urea secretion also takes place in the mammalian kidney (11) through a specialized transporter protein located in the pars recta of the proximal tubule (see review in Ref. 10). As suggested by Hays et al. (36), mutations in the corresponding gene are most probably responsible for the rare cases of “familial azotemia without renal failure” reported in the literature (2, 21, 37). In rats and various other rodents, several micropuncture studies have shown that a much higher rate of urea is delivered to the early distal tubule than that flowing in the late proximal tubule (3, 53, 54, 64, 79, 80). This was attributed to intrarenal recycling of urea from the CD, but, given the magnitude of the distal-proximal difference, urea secreted in the pars recta could also have contributed to urea addition in the loops.

Taken together, these observations suggest that urea may be secreted to a significant extent in the mammalian kidney, and more so in species of small size like the mouse. But no definitive conclusion can be drawn until this secretion is demonstrated in isolated perfused pars recta. To our knowledge, only two studies have attempted to identify possible active urea secretion in the pars recta by microperfusion in isolated tubules (45, 49). Both groups used rabbit tubules because this was the only species in which microdissection and microperfusion were possible in the early times. A very modest but significant active urea secretion was observed in the first study (using cortical and medullary pars recta) (45), but this was not confirmed in the second study (using only cortical pars recta) (49). Unfortunately, the rabbit was probably not a good species in which to document such active secretion because of its protein-poor diet. Moreover, for reasons explained elsewhere (10), this secretion is likely to occur mostly in the medullary pars recta. The unusually high urea clearance recently observed in mice (8) should stimulate new studies addressing active urea secretion in the pars recta in this species and identification of the possible transporter involved.


Renal Handling of Urea After an Acute Urea Load

In rats and other mammals, acute urea infusion or high protein intake have been shown to improve urinary concentrating ability and water conservation, a “urea-referable water economy” first recognized by Gamble (22, 34) more than 60 years ago. The contribution of UT-B to this phenomenon has been evaluated by comparing the response of UT-B null mice to that of wild-type mice after acute intraperitoneal administration of a urea load. Care was taken to bring wild-type mice and UT-B null mice to a similar level of urine concentrating activity before administration of the urea load. This was achieved by adding an appropriate amount of water to the food of wild-type mice for 2 days before the experiments (8). In this way, possible differences observed in response to the load can be unequivocally attributed to the difference in genotype and not to preexisting differences in urea concentration in the medulla or in the basal urine flow rate and water turnover. A special experimental setting was designed for performing short-term accurate urine collection in non-instrumented, conscious, freely moving mice (8).

After administration of the urea load, urea excretion rose to the same extent in the two groups, but this increase resulted from a very different pattern of changes in urine flow rate and urea concentration in the two genotypes (Fig. 8). Wild-type mice increased their urea concentration and urine osmolality after the first postload period and exhibited only a small rise in urine flow rate, whereas UT-B null mice almost doubled their urine flow rate but showed only a modest rise in urine osmolality and urine urea concentration. After urea excretion returned to the basal level (6 and 8 h after the load), urine osmolality and urea concentration in wild-type mice remained far higher than during the basal period. This most probably results from the sequestration of higher amounts of urea in the inner medulla, and their urine output declined significantly to less than half the basal level. In contrast, urine osmolality was only modestly elevated in UT-B null mice, and the urine flow rate was reduced to a lesser extent compared with the basal level (Fig. 8).

Fig. 8.

Effect of acute urea loading on urinary concentrating activity and renal handling of urea in wild-type and UT-B null mice. Values are means ± SE of 6 mice/group. A special technique was designed to allow short-term urine collection in conscious freely moving mice (8). Three hundred micromoles urea were injected intraperitoneally just after the first 2-h urine collection (time 0), and urine was then collected for 4 more 2-h periods. A: urine osmolality (Uosm). B: urine output. C: urine urea concentration (Uurea). D: urea excretion (excr.). E: concentration of non-urea solutes in urine (Unon-urea solutes). F: excretion of non-urea solutes. Analysis by 2-way ANOVA (genotype and 2-h periods) indicated significant effects for urine osmolality (P < 0.0001 for genotype and for period; P = 0.0007 for interaction), urine output (P < 0.0001 for genotype and for period; P = 0.0015 for interaction), and urea concentration (P < 0.0001 for genotype and for period; P = 0.0203 for interaction). *Significant differences between 2 genotypes in individual periods (Fisher post hoc test). Wild-type mice had been given some water in their food during the previous 24 h so as to bring their urine output and osmolality to the same level as that of UT-B null mice. This accounts for the fact that, during the basal period (0 h), the 2 groups showed no significant differences for any parameter. Reproduced from Ref. 8.

The acute urea load (300 μmol), although a relatively small fraction of the daily urea excretion (3,700 μmol, Table 2), was large compared with the body's pool of urea i.e., ∼150 μmol, calculated as plasma urea × body water (60% body wt). Thus this load probably significantly increased plasma urea concentration and osmolality. Even if urea is a less efficient osmole than sodium for stimulating vasopressin secretion (115), a significant rise in plasma vasopressin can be expected to occur in response to this load. The combination of an increased availability of urea and of a more intense action of vasopressin most probably accounts for the marked elevation of urinary osmolality and urea concentration in the urine observed in wild-type mice.

Results observed in the last two periods of the experiment (after the load had been excreted) reflect the long-lasting consequences of urea on urine concentrating ability. Urine osmolality and urea concentration in wild-type mice remained much higher and urine output much lower than in basal conditions. Moreover, the concentration of non-urea solutes was also markedly improved (Fig. 8). This is most probably due to an increase in urea accumulation in the medullary interstitium that improved the capacity of the kidney to concentrate urine and to conserve water. In contrast, this delayed effect did not occur in UT-B null mice.

This urea-loading experiment illustrates the role of urea in the urinary concentrating mechanism in mice and reveals the absolute need for countercurrent exchange in vasa recta (involving both plasma and red cell urea) for building up the urea gradient in the medulla. It also shows the crucial role of the vascular recycling route in the dynamic phase of progressive urea accumulation in the medulla. The tubular recycling route should not be impaired in UT-B null mice. It should even be favored because of the increased expression of UT-A2 in UT-B null mice (see above). Nonetheless, urea accumulation in inner medullary tissue under basal conditions (Fig. 7) and the capacity of urea to improve urinary concentrating ability (Fig. 8) are severely compromised by the lack of UT-B expression in vasa recta and RBCs (8).

Urea Handling After Chronic Alterations in Urea Excretion

To evaluate how UT-B null mice adapt chronically to different levels of urea excretion, wild-type mice and UT-B null mice were fed for 1 wk a diet containing either 10, 20, or 40% protein (low protein, normal protein, or high protein, respectively) (8). The different levels of protein intake did not induce significant changes in urine osmolality in either group but induced marked differences in urine output. Urine osmolality was consistently lower and urine flow rate higher in UT-B null mice than in wild-type mice on each diet (Fig. 9).

Fig. 9.

Urea handling in wild-type and UT-B null mice after chronic alterations in urea excretion induced by changes in protein intake. Mice were fed diets containing 10, 20, or 40% casein during 3 successive wk (urine collected on the last day of each week). Values are means ± SE of 6 mice/group. A: urine osmolality (Uosm). B: daily urine output. C: plasma urea concentration (Purea). D: daily urea excretion. E: urine urea concentration (Uurea). F: urine-to-plasma ratio of urea concentration (Uurea/Purea). Analysis was by 2-way ANOVA. a, Effect of genotype; b, effect of diet; c, interaction between genotype and diet. NS, not significant. Adapted from Ref. 8.

In contrast to the stability seen in wild-type mice, the plasma urea level in UT-B null mice was very sensitive to protein intake, being 48 and 75% higher in mice on the normal- and high-protein diets, respectively, than in those on the low-protein diet (Fig. 9). As a result, the UT-B null mice/wild-type mice ratio for plasma urea rose from 1.30 with the low-protein diet to 1.55 with the normal- and 1.80 on the high-protien diet. As expected, urea excretion rose to the same extent in both groups of mice in response to graded increases in protein intake. However, in wild-type mice, part of this increase in urea excretion was achieved by a significant rise in urea concentration in the urine, whereas in UT-B null mice, it was only due to an increase in urine output with no change in urine urea concentration (Fig. 9).

This experiment shows that, even when allowed to reach a new steady state for 1 wk, UT-B null mice cannot make use of extra urea to improve their capacity to concentrate urea in the urine. Urea concentration remained almost unchanged at ∼1,000 mmol/l when protein intake varied over a fourfold range, a value slightly lower than that observed in wild-type mice on the lowest protein intake. In contrast, in wild-type mice, urea concentration in the urine increased to >2,000 mmol/l. This reveals the crucial role of UT-B and of the vascular route of urea recycling in the capacity of urea to raise urinary concentrating ability chronically (8).

New Understanding of the Contribution of Urea to the Urine Concentrating Mechanism Provided by the UT-B Null Mouse

As shown above, the urinary concentrating defect in UT-B null mice is urea-selective because it reduces the capacity of the animals to concentrate urea over their blood and extracellular fluid level more than for other solutes. The fact that plasma urea level in UT-B null mice on a normal protein diet was 44% higher than in wild-type mice suggests that a large fraction of the urea that was delivered by the CD to the inner medullary interstitium returned to the general circulation through the renal vein, instead of being recycled within the kidney by countercurrent exchange. When mice are in balance, their urea excretion meets the daily needs at the price of a much higher plasma level and lesser urea clearance than in wild-type mice. Studies in UT-B null mice also indicate that the enhancement of overall urine concentrating ability referable to urea is largely due to urea recycling through the vascular route in the medulla. Acute urea infusion or chronic high protein intake in UT-B null mice did not increase urea concentration in the urine and failed to raise the concentration of non-urea solutes as observed in many previous studies in rats and again here in wild-type mice.

In the aggregate, these results suggest that the recycling of urea in vasa recta might play a much greater role in the urinary concentrating process than its recycling in the loops of Henle because suppression of the former almost completely prevented the kidney to take advantage of urea for improving its concentrating activity. This preponderant influence of vascular recycling over tubular recycling was not anticipated from previous studies and needs to be confirmed by future studies in UT-A2 null mice when they become available. It may be accounted for by several factors. First, UT-A2 is expressed along only a fraction of the TDLs in the IS of the outer medulla (short loops) and is probably not or very slightly expressed in long-loop TDLs in the inner medulla (see above). In contrast, UT-B is expressed in DVR along the entire outer and inner medulla. Second, in the vascular bundles of the outer medulla, DVR are completely surrounded by AVR, thus offering a much larger area of direct contact for countercurrent exchange than between AVR and thin limbs. Such exchange continues in the inner medulla even if descending and ascending vessels are less closely associated. Third, the flow rate of blood in the medullary circulation should be significantly higher than the flow rate of tubular fluid in thin limbs.

As explained earlier, the high level of UT-B expression in RBCs is assumed to make them contribute significantly to the concentrating mechanism by increasing the mass of urea recycled through the vascular route within the medulla. Unfortunately, it has not yet been possible to evaluate separately the contribution of vasa recta and that of the RBCs they contain. This would require complete blood exchange or kidney transplantation between wild-type and UT-B null mice, very delicate protocols in small animals. Hopefully, such experiments will be performed in future studies.

In the kidney itself, UT-B is restricted to the vascular compartment and thus the tubular fluid is not exposed to membranes expressing this transporter. In contrast, after urine exits the CDs, urine flows in structures that all express UT-B in the basolateral membrane of their urothelium, including the pelvic wall, ureter, and bladder (58, 97). The role of UT-B in these structures is not yet understood, and the experiments performed so far in UT-B null-mice have not contributed information on this matter.

It seems odd that a facilitated urea transporter is abundantly expressed in structures that need to maintain a very high transepithelial urea concentration difference for long periods of time (at least in the bladder), thus leaving the luminal membrane as the only barrier between urine and blood. Interestingly, a small but significant active Na-dependent, phloretin-insensitive urea secretion has been observed in the deepest subsegment of the IMCD (i.e., along a much shorter length than that expressing UT-A1/3) (44). Because of the very small number of CDs remaining at the tip of the papilla, and of the very short length of CD in which it occurs, this active transport can add only very small amounts of urea in the CD lumen. Moreover, the high UT-A1/3-mediated permeability of these ducts to urea and the direction of the transepithelial difference in urea concentration in this region should induce all the secreted urea to diffuse back into the medullary interstitium. The role of this active urea secretion in the very terminal portion of the IMCD is thus puzzling. However, if this active urea secretion would extend further down along the pelvic epithelium, ureter, and bladder, it might contribute to prevent the dissipation of the urine-to-blood urea gradient in the lower urinary tract by bringing urea back into the lumen. This is of course speculative and requires further investigation.


Respective Roles of UT-A1/3 and UT-B in the Urine Concentrating Mechanism

Recently, UT-A1/3 null mice have been generated and some aspects of their phenotype have been described (28). It is thus interesting to compare the consequences of UT-A1/3 vs. UT-B deletion. Not surprisingly, the urine concentrating defect is more severe in the former. The urinary concentrating ability of UT-A1/3 null mice in basal conditions is reduced to ∼35% of that in wild-type mice, and their urine flow rate is threefold higher. Moreover, UT-A1/3 null mice almost did not raise their urine osmolality after 5 days of severe restriction in fluid intake whereas UT-B knockout mice were able to raise their urine osmolality after 24 h of dehydration, even if less than did wild-type mice. Urea accumulation in the inner medulla of the former was also more severely reduced to about one-third of normal vs. only one-half in UT-B null mice (although this comparison is not strictly valid because inner medullary solutes have been measured in basal conditions in UT-B knockout mice but after severe chronic water restriction in UT-A1/3 knockout mice). In none of these two models was chloride accumulation in the inner medulla impaired. This shows that urea accumulation is not a prerequisite for NaCl accumulation in the inner medulla (28).

It may be interesting to mention that surgical ablation of the renal papilla in rats, which obviously prevents the kidney from recycling urea, results in a decline in urinary concentrating ability of about the same magnitude as that seen in UT-A1/3 null mice (63, 109). The urinary concentrating defect observed in UT-B and UT-A1/3 null mice is less apparent with low protein intake (8, 28) because much less urea needs to be excreted in this case, and urea becomes a much smaller fraction of total urinary solutes. The accumulation of NaCl in the inner medulla is not compromised in these two models, and the concentration of other solutes is less affected than that of urea. Thus it is not surprising to see a lesser difference in urine osmolality and volume between wild-type and null mice when urea excretion is markedly reduced.

Specificity of Mouse Urine Concentrating Mechanism Compared with Other Mammals

Findings in UT null mouse models show that the intrarenal movements of urea allowed by facilitated UTs contribute quite significantly to the capacity of the mouse kidney to concentrate urine. On a normal-protein diet, urine osmolality is lowered to ∼35–50% of values found in normal mice. Would similar results be observed in other species? Extrapolation to other species should take into account several features of mice vs. other mammals, namely, the large differences in size of the different species, the different levels of urinary concentrating ability, and the different kidney adaptations.

It is important in the present context to remember that mice are much smaller than rats and most other mammals. Because of their very small size, their metabolic rate and food consumption per unit body weight are markedly higher, thus inducing a disproportionately much higher urea load (and other solutes) to excrete daily (Table 1). The need to excrete so much more solute would obligate an extremely high fluid intake/output if mice were not, in addition, able to concentrate urine to a much higher level than many other mammals, or even other larger rodents like rats. This higher urine concentrating ability is due, at least in part, to special adaptations of the kidney that have been well described in several anatomic-functional studies (9, 52). These include a relatively much longer papilla (hence much longer “long loops” of Henle than in the rat); a reorganization of the outer medulla to form giant vascular bundles and incorporate TDLs of short loops in these bundles (note that 75% of the nephrons are short-looped in the mouse kidney); and formation of an “innermost stripe” in the outer medulla, in which the short-loop descending limbs already exhibit a typical “thick ascending limb” epithelium, etc. (9, 52). Unfortunately, these special adaptations are too often ignored (especially in mathematical modeling) (116), and this may weaken or even invalidate the conclusions of some recent mouse studies.

Because mice carry such an extreme urine (and urea) concentrating ability, it may be assumed that the role played by UTs in overall urinary concentrating ability is less intense in other species than revealed here in mice. This may explain why the urine concentrating defect is relatively smaller in humans with the UT-B null genotype (84) than in UT-B null mice.


The studies reviewed here have shown that UT-B deletion does not affect the overall phenotype of the mice and does not alter their GFR and kidney weight. It also does not alter the clearance and fractional excretion of most urinary solutes, but it induces a moderate urinary concentrating defect, a rise in plasma urea level, and a fall in urea clearance. These studies have confirmed the global pattern of intrarenal urea recycling that had been deduced decades ago from micropuncture and microperfusion experiments in rats, rabbits, and a few other species. In addition, they have revealed that countercurrent exchange of urea between vasa recta supplying and draining the renal medulla (and possibly additional urea movements by the RBCs) plays a critical role in the renal concentrating process. Mice with UT-B deletion are unable to take advantage of urea for improving their urine concentrating activity acutely. When challenged chronically with a high protein intake, they are unable to prevent an ample rise in plasma urea and to enhance urea concentration in urine, as do wild-type mice. Their urinary concentrating defect is, however, somewhat less intense than that resulting from invalidation of the CD urea transport system. Future studies in UT-A2 null mice and in double UT-A2/UT-B null mice are awaited to provide a more complete picture of the contribution of urea to the concentrating mechanism.


This work was supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases (DK-66194) and the American Heart Association (0365027Y) (to B. Yang) and by annual funding from INSERM (to L. Bankir).


The authors thank Alan S. Verkman (Dept. of Medicine, UCSF, San Francisco, CA), Christian de Rouffignac (CEA, Saclay, France), and Nadine Bouby (INSERM Unit 652, Paris, France) for a critical reading of the manuscript.


View Abstract