At Physiological Expression Levels the Kidd Blood Group/Urea Transporter Protein Is Not a Water Channel*

The Kidd (JK) blood group locus encodes a urea transporter that is expressed on human red cells and on endothelial cells of the vasa recta in the kidney. Here, we report the identification in human erythroblasts of a novel cDNA, designated HUT11A, which encodes a protein identical to the previously reported erythroid HUT11 urea transporter, except for a Lys44→ Glu substitution and a Val-Gly dipeptide deletion after proline 227, which leads to a polypeptide of 389 residues versus391 in HUT11. Genomic typing by polymerase chain reaction and transcript analysis by ribonuclease protection assay demonstrated that HUT11A encodes the true Kidd blood group/urea transporter protein, which carries only 2 Val-Gly motifs. Upon expression at high levels inXenopus oocytes, the physiological Kidd/urea transporter HUT11A conferred a rapid transfer of urea (which was insensitive top-chloromercuribenzene sulfonate or phloretin), a high water permeability, and a selective uptake of small solutes including amides and diols, but not glycerol and meso-erythritol. However, at plasma membrane expression levels close to the level observed in the red cell membrane, HUT11A-mediated water transport and small solutes uptake were absent and the urea transport was poorly inhibited byp-chloromercuribenzene sulfonate, but strongly inhibited by phloretin. These findings show that, at physiological expression levels, the HUT11A transporter confers urea permeability but not water permeability, and that the observed water permeability is a feature of the red cell urea transporter when expressed at unphysiological high levels.

In mammals, urea is the chief end product of nitrogen catabolism and is produced in the liver by the urea cycle during the conversion for arginine to ornithine. Additionally, urea is a key component in the urinary concentrating mechanism, in which it is essential for renal water retention and prevention of dehydration. In this later process, urea transporters in red cells and the kidney have been shown to play a pivotal role (reviewed in Refs. 1 and 2). In the last 5 years, two types of facilitated urea transporters have been molecularly characterized in different animal species: (i) transporters encoded by the UT-A gene, only present in the kidney, and (ii) more ubiquitous transporters encoded by the UT-B gene, present in red cells, kidney, and brain (3)(4)(5).
The first human erythroid urea transporter (HUT11) was identified by homology cloning and was later shown to be encoded by the Kidd (JK) 1 blood group locus (6,7). The HUT11 cDNA encodes a membrane glycoprotein of 391 amino acids, which facilitates urea transport. Expression studies in Xenopus oocytes have shown that HUT11 urea transport was inhibitable by phloretin and para-chloromercuribenzene sulfonate (pC-MBS) (6), as expected for the transporter of human erythrocytes (8). Immunohistochemical and in situ hybridization studies have shown that HUT11 is also expressed on endothelial cells of the vasa recta in the inner and outer medulla of the kidney, but not on the epithelial cells of the renal tubules, interstitial cells, and glomerular cells (9). This distribution of expression of HUT11 fully accounted for studies in which a physiological urea transport has been described (10), as well as for the model of experimental hydronephrotic rat kidneys showing a strong staining of the urea transporter on preserved intact vasculature despite a complete loss of the tubular epithelium (11). These findings suggested that, in the renal circulation, the Kidd/urea transporter is involved in countercurrent exchange between the ascending and descending vasa recta to prevent loss of urea from the medulla and to enhance the cortico-papillary osmolality gradient, which is critical in the urinary concentrating mechanism (12).
Recently, we have reported that the JK gene, which encodes the human urea transporter, is composed of 11 exons extending over 30 kilobase pairs of DNA (13), and we have identified the molecular basis of the JK*A/JK*B polymorphisms (14). We also found that different splice site mutations in two unrelated Jk null individuals provided a rational explanation for the lack of Kidd/urea transporter protein at the red cell surface (13). Thus in Jk null individuals, the absence of Kidd/urea transporter protein on red cells and probably on endothelial cells of the vasa recta should alter the urea recycling mechanism, thus explaining the reduced capacity to concentrate urine of these individuals (15). No erythrocyte hemolysis has been reported in Jk null individuals who do not suffer from a clinical syndrome except for the urine concentrating defect.
Interestingly, another urea transporter has been characterized, which is only expressed in human kidney (16,17). This transporter, which is called HUT2 (hUT-A2), is 62% identical to HUT11 (hUT-B2). Both human urea transporters are encoded by genes localized on chromosome 18q12, suggesting that they evolved from duplication of a common ancestor.
Here, we report the isolation of a cDNA clone encoding a polypeptide called HUT11A, which is slightly different from HUT11. Genomic and transcript analysis demonstrated that HUT11A, and not HUT11, is the physiological product of the Kidd blood group locus. Additionally, functional studies in Xenopus oocytes showed that HUT11A is only a urea transporter at physiological expression levels, but a water and small solute transport activity at high, unphysiological, expression levels.

MATERIALS AND METHODS
Blood Samples and Reagents-Blood samples from individuals of common Jk phenotypes were collected from anticoagulated blood and used for total reticulocyte RNA and leukocyte genomic DNA extraction. Restriction endonucleases and modifying enzymes were from New England Biolabs (Hertfordshire, United Kingdom (UK)). Radiolabeled nucleotides and [ 14 C]urea (1.96 GBq/mmol) were from Amersham Pharmacia Biotech (Bucks, UK), and [ 3 H]raffinose (188.7 GBq/mmol) was from NEN Life Science Products. The Expand High Fidelity PCR system from Roche Molecular Biochemicals (Mannheim, Germany) was used for PCR amplification. Nucleotide sequences were determined on both strands by the dideoxy chain termination method (Sanger) with ThermoSequenase fluorescently labeled primer cycle sequencing kit from Amersham Pharmacia Biotech using 5Ј-(Cy5) primers (Genset, Paris, France).
Reverse Transcription-PCR of the Jk cDNA and cDNA Constructs-Full-length cDNA encoding the human Kidd blood group/urea transporter protein was amplified by reverse transcription-polymerase chain reaction (RT-PCR) using total reticulocyte RNAs extracted by the acidphenol-ammonium method (18) from blood samples using a sense primer (position Ϫ21 to Ϫ1) and an antisense primer (position 1234 -1211) as described previously (11). AQP-1 cDNA (accession no. L07268) was PCR-amplified from a gt11 human bone marrow cDNA library (CLONTECH) using sense (position Ϫ15 to Ϫ1) and antisense (position 829 to 849) primers. All cDNAs were subcloned into the EcoRV-digested pT7TS plasmid (kindly provided by P. Krieg, Institute for Cell and Molecular Biology, Austin, TX) and sequenced on both strands, using an automated Alf-Express sequencer (Amersham Pharmacia Biotech, Uppsala, Sweden). For primer designation, nucleotide ϩ1 was taken as the first nucleotide of the HUT11 initiation codon (5).
PCR Genomic Typing of the Sequence Encoding the Val-Gly Repeats-To determine whether the genomic DNA encoded for a transporter with 2 or 3 Val-Gly dipeptide repeats, a hemi-nested PCR amplification was carried out under stringent conditions (94°C for 30 s, 62°C for 30 s, 72°C for 10 s, for 30 cycles) using 200 ng of leukocyte DNA from 126 independent donors. The first amplification was done between SP-A (sense primer, 5Ј-gcagTTGTTGAAATCTATACCAG-3Ј, exonic position 664 -682) and AS-B (antisense primer, position 797-774) and the second amplification between the primers SP-A and AS-C (antisense primer, position 728 -705), under the same experimental conditions. The PCR products were digested with 5 units of BglII restriction enzyme and electrophoresed on a 15% (w/v) polyacrylamide gel stained with ethidium bromide.
Ribonuclease Protection Assays-Using a described experimental strategy (19), an antisense RNA probe (ASP-1) encompassing 406 nucleotides between positions 426 and 831 of the HUT11 cDNA was constructed, which encodes 3 Val-Gly dipeptide repeats. For this, the corresponding cDNA fragment was amplified between primers SP (sense primer, position 426 -447) and AS (antisense primer, position 831-811) and cloned into the pCR2.1 vector (Invitrogen, SA). From HindIII-linearized recombinant plasmid, the antisense-RNA probe ASP-1 was in vitro synthesized using the Riboprobe Core System from Promega (Madison, WI) in the presence of [␣-32 P]UTP (800 Ci/mmol, NEN Life Science Products). After purification on a 5% (w/v) acrylamide, 8 M urea gel, the ASP-1 probe (2.0 ϫ 10 5 cpm) was hybridized overnight at 50°C with 25 ng of in vitro synthesized sense transcripts (cRNAs) encoding the 2Val-Gly or 3Val-Gly motifs (mCAP mRNA capping kit; Stratagene, La Jolla, CA) as controls, 2 g of human bone marrow poly(A ϩ ) RNAs (CLONTECH), and 20 g of total reticulocyte RNAs from two unrelated individuals, and then digested with a RNase A/T1 mixture, according to the manufacturer's instructions (RPAII; Ambion, Austin, TX). Subsequently, the protected RNA fragments were separated on a 5% (w/v) denaturing polyacrylamide gel (8 M urea) and exposed to a Biomax MS film with intensifying screens at Ϫ80°C.
Plasmid Preparation and Oocyte Expression-The pT7TS-cDNA constructs were linearized with SmaI restriction enzyme, and capped sense RNAs were synthesized using T7 RNA polymerase from the mCAP mRNA capping kit (Stratagene). Expression studies were carried out by microinjection of cRNAs in collagenase-treated Xenopus laevis oocytes as described previously (20). Briefly, stage V and VI oocytes from mature female X. laevis (CNRS, Nantes, France) were removed, defolliculated with 1.4 units/ml collagenase type A (Roche Molecular Biochemicals) in OR2 solution (82.5 mM NaCl, 2 mM KCl, 5 mM Hepes, 1 mM MgCl 2 , pH 7.5), and microinjected with 50 nl of water or cRNA solution (0.05-40 ng/oocyte in 50 nl). The microinjected oocytes were then kept at 18°C for 2 or 3 days in Barth solution (200 mosM) containing 50 g/ml Geneticin with daily changes of Barth solution until functional tests.
Oocyte Flux Measurements-After injection of 40 ng cRNA/oocyte, urea transport activity was measured by [ 14 C]urea uptake as described previously (6), or under conditions as indicated in the text. In all these experiments, the Barth incubation solution contained 8 Ci/ml [ 14 C] urea (145 M) and 5 Ci/ml [ 3 H]raffinose as a control of oocyte plasma membrane integrity. Incubation time was 90 s, except for time-course experiments where the incubation time varied from 0 to 60 min. After washing and solubilization, the samples were subjected to liquid scintillation in a counter, and urea permeability (P urea) was calculated from the oocyte-associated amount of [ 14 C]urea at each time point, corrected for the optically determined oocyte surface area.
Oocyte water permeability (P f ) was measured by a swelling assay 3 days after the injection (24). Oocyte swelling was performed at 18°C after the transfer from Barth solution (200 mosM) to 40 mosM at t ϭ 0. Permeability measurements were made by a microscopy technique using a Nikon Eclipse TE300 microscope (Nikon, Paris, France) (2ϫ objective) coupled to a Biocom computer system of image integration (Biocom, Les Ulis, France). The osmotic water permeability P f (cm/s) was calculated from the initial osmotic cell volume increase between t ϭ 0 and t ϭ 90 s by the relation where S is the oocyte surface area and w the molar volume of water (18 cm 3 /mol) (25). In order to determine the influence of small solutes on the water movement across injected oocyte plasma membranes, oocytes were placed at t ϭ 0 in 40 mosM Barth solution adjusted to 200 mosM with the tested solute including formamide, acetamide, propionamide, ethylene glycol, propylene glycol, glycerol, or meso-erythritol, as described previously (17). Oocyte swelling was performed for 240 s. Osmolarity was checked with a Roebling osmometer just before the experiments.
Electron Microscopy-The same batches used to measure the urea permeability were prepared to determine the particle density in the P-face plasma membrane of oocytes. Before fixation, the oocytes were rapidly emptied of their cytoplasm by aspiration with a pipette (diameter Ͼ100 m). H 2 O-and cRNA-injected oocytes were fixed in 2.5% glutaraldehyde in Barth solution for 2 or 3 h at 18°C, then washed in Barth solution. Emptied oocytes were incubated in Barth solution supplemented with 30% glycerol for 1 h at room temperature and placed between two copper sample holders and then frozen in melting freon. Samples were fractured in a Balzers 300 apparatus at Ϫ150°C under 10 Ϫ7 torr vacuum. Fractured surfaces were coated with platinum at 45°C and carbon at 90°C under the conditions described by the manufacturer. Replicas were cleaned in bleach, washed in distilled water, and observed in a Philips EM400 microscope at 80 kV. Representative series of images of P-face fractures were enlarged at 95,000ϫ final magnification in order to determine the number of intramembrane particles (IMP) in known areas as described by Zampighi et al. (26). Some images were amplified by 800,000 to estimate the size of the particles according to Eskandary et al. (27).

RESULTS
Identification of the Kidd/Urea Transporter-Using two primers located in the 5Ј-and 3Ј-untranslated regions of the previously reported cDNA clone HUT11 (11) in a RT-PCR reaction, a new cDNA clone called HUT11A was isolated from human reticulocyte RNAs. Comparison of HUT11 and HUT11A cDNA sequences, showed that both clones derived from a JK*A allele (G838) but differed in two ways: (i) an A130G transition resulting in a Lys 44 3 Glu substitution, and (ii) a hexanucleotide (GTG GGA) deletion in HUT11A that leads to a Val-Gly dipeptide deletion after the Pro-227 (Fig. 1). Because of this deletion, HUT11A encoded a polypeptide chain of 389 amino acids versus 391 for HUT11. Sequence alignments of several urea transporters indicated that only HUT11 contained an additional Val-Gly dipeptide ( Fig. 1), which is located within the third external loop and is in close proximity of the Nglycosylation site (Asn-211).
Since the initially reported HUT11 clone was derived from a human bone marrow cDNA library (CLONTECH, catalog HL 1058b, lot 1911), constructed from RNAs isolated from a single adult individual, we developed a hemi-nested PCR assay on genomic DNA to determine whether the repeat of 2 or 3 Val-Gly motifs is a common polymorphism in the human population. Accordingly, a hemi-nested PCR was carried out on the genomic DNA from 126 unrelated blood donors of different Jk phenotypes to amplify a 63/69-bp fragment encompassing the sequence encoding the repeated dipeptides, and the PCR product was digested by BglII. Analysis on a 15% polyacrylamide gel revealed only the 36-bp fragment, which encodes the variant with two Val-Gly dipeptide repeats ( Fig. 2A). The 42-bp fragment encoding a 3 Val-Gly repeat was obtained only in the HUT11 cDNA control. These data show that the presence of 2 or 3 Val-Gly motifs is not allelic and that the JK locus only encodes 2 Val-Gly motifs. Allele-specific PCR analysis also indicated that the Lys 44 3 Glu polymorphism was not allelic to the JK locus (data not shown).
To further confirm these results, Kidd/urea transporter transcripts (bone marrow poly(A ϩ ) RNA and two total RNA reticulocyte preparations) were analyzed by ribonuclease protection assay using a labeled antisense RNA probe (ASP-1) encoding 3 Val-Gly repeats (Fig. 2B). With cRNA encoding the 3Val-Gly, a protected RNA fragment of 406 bp was obtained. Due to the non-complementary 6-nucleotide stretch, two protected fragments of 261 and 139 bp were obtained with a cRNA encoding the 2Val-Gly repeat as a template. With poly(A ϩ ) RNA as well as total RNAs from unrelated individuals, only hybridization signals of 261 and 139 nt were detected, which indicated that the JK transcripts have a nucleotide sequence that encoded a polypeptide carrying only 2 Val-Gly motifs. Our extensive analysis of genomic DNAs and RNAs from several individuals thus indicated that the physiological urea transporter was encoded by the sequence found in clone HUT11A and not, as initially reported (6), in clone HUT11.
Transport Studies-To analyze the functional features of HUT11 (Lys 44 /3Val-Gly) and HUT11A (Glu 44 /2Val-Gly), 40 ng of cRNAs encoding these proteins were injected into Xenopus oocytes. Immunocytochemical analysis using affinity-purified anti-N-terminal antibodies (7,22) revealed that both proteins were expressed at the plasma membrane of oocytes, which rendered these cells suitable for functional analysis (Fig. 3A). Nevertheless, for the same amount of cRNA injected, the HUT11A signal at the oocyte plasma membrane was clearly stronger than the HUT11 signal. Water-injected controls were not stained, showing the specificity of the antibody labeling (Fig. 3A).
Next, the time course of urea uptake by HUT11 and HUT11A expressing Xenopus oocytes was explored (Fig. 3B). HUT11 and HUT11A expressing oocytes showed a rapid initial urea uptake that was 15 and 45 times higher than in water-injected oocyte controls, respectively. The urea permeabilities (P urea ) of HUT11-and HUT11A-injected oocytes, determined at 90 s, were, respectively, 15.7 Ϯ 0.82 ϫ 10 Ϫ6 cm/s (n ϭ 56) and 46.6 Ϯ 1.86 ϫ 10 Ϫ6 cm/s (n ϭ 72) versus 1.02 Ϯ 0.11 ϫ 10 Ϫ6 cm/s (n ϭ 68) for water-injected oocytes (p Ͻ 0.001). Consequently, the plateau corresponding to the equilibration of urea was more rapidly reached with oocytes expressing HUT11A than with HUT11. These data suggest that both HUT11 and HUT11A function as efficient urea transporters. In all experiments, the raffinose permeability was not increased, which indicate that the plasma membrane integrity was intact (data not shown).
Examination of the [ 14 C]urea uptake also indicated that, for a given level of cRNA injected (40 ng), the urea transport activity was greater for HUT11A-injected oocytes than for HUT11-injected oocytes (Fig. 3B). This could result from either a faster turnover of HUT11A with comparable levels of expression or a higher level of HUT11A with a comparable turnover. The latter hypothesis was most likely since immunostaining of the same oocytes clearly indicated a higher expression level of HUT11A as compared with HUT11 (Fig. 3A).
To test whether HUT11A and HUT11 also confer water permeability, oocytes expressing these proteins were tested in a standard swelling assay. Surprisingly, oocytes expressing HUT11A swelled significantly faster than oocytes expressing AQP1, which was taken as a positive control (Fig. 3C). As already reported (7), swelling of oocytes expressing HUT11 was not different from water-injected controls in these conditions. Urea transport analysis in the presence of pharmacological inhibitors revealed that the HUT11A-mediated urea flux was poorly inhibited by pCMBS, phloretin, and HgCl 2 (Fig. 4A). However, as previously reported (6), the HUT11-mediated urea flux was strongly inhibited by pCMBS (74%) and phloretin (85%) but only slightly inhibited by HgCl 2 (30%). Incubation with inhibitors showed that the HUT11A-mediated osmotic water permeability (P f ) of about 150 m/s was strongly inhibited by pCMBS and phloretin, but not by HgCl 2 (Fig. 4B).
We speculated that the unusual properties of HUT11A as compared with HUT11 might have resulted from a difference in the expression level of the two transporters in oocytes (see above). To test this hypothesis, the urea and water transport and HUT11A plasma membrane expression levels were determined in oocytes injected with different amounts of HUT11A cRNA. As shown in Fig. 5A, the urea transport rate increased steadily with injections of between 0.005 and 0.05 ng of HUT11A cRNA and sustained at a plateau rate of about 20 pmol/90 s for oocytes injected with 0.05-40 ng of cRNA. Accordingly, the urea uptake between 10 and 360 s of oocytes injected with 0.05, 0.1, or 40 ng/oocyte HUT11A cRNA, was identical (Fig. 5B). The corresponding urea permeabilities calculated at 10 s were 124 Ϯ 11 ϫ 10 Ϫ6 cm/s (n ϭ 20), 119 Ϯ 12 ϫ 10 Ϫ6 cm/s (n ϭ 20), and 133 Ϯ 7 ϫ 10 Ϫ6 cm/s (n ϭ 20), respectively, versus 1.6 Ϯ 1 ϫ 10 Ϫ6 cm/s (n ϭ 25) for waterinjected controls. In contrast, using oocytes from the same batch, a water permeability was detected only when at least 0.5 ng of HUT11A cRNA was injected and steadily increased up to FIG. 3. Immunocytochemical analysis and characterization of urea transport and water permeability of oocytes injected with cRNAs encoding HUT11 and HUT11A proteins. A, Xenopus oocytes were injected with 40 ng of cRNA encoding HUT11 or HUT11A. Waterinjected oocytes were used as negative control. After 3 days, oocytes were fixed and sections of injected oocytes were stained with an affinitypurified antibody against the N terminus of the Kidd/urea transporter protein and visualized with fluorescein isothiocyanate-conjugated antirabbit IgG as described under "Materials and Methods" and then imaged using a Nikon Eclipse TE300 microscope (Nikon, Paris, France) (20ϫ objective). Images were recorded with epifluorescence illumination and treated with a Biocom computer system of image integration (Biocom, Les Ulis, France).  (Fig. 5A).
We next compared the effect of pharmacological inhibitors (pCMBS and phloretin, 1 mM) on the HUT11A-mediated urea transport in oocytes injected with small (0.1 ng) and large (40 ng) amounts of cRNA, since there was no difference in the urea permeability under these conditions (see Fig. 5). As shown from Fig. 6A, the urea flux mediated by oocytes injected with 0.1 ng of HUT11A cRNA was poorly inhibited by pCMBS (18%), but was strongly inhibited by phloretin (79%), thus partly exhibiting the properties of the urea transporter from human red cells (8). In contrast, when the amount of HUT11A cRNA injected was increased to 40 ng/oocyte, phloretin no longer inhibited the urea transport.
To analyze the influence of small solutes (non-electrolytes) on the water movement across oocytes injected with low and high levels of HUT11A cRNA, the initial rates of swelling, between t ϭ 0 and t ϭ 240 s, were measured in an iso-osmotic solution containing 160 mM small solutes adjusted to 200 mosM with diluted Barth solution (Fig. 6B). Oocytes injected with 20 ng of HUT11A cRNA swelled significantly in a solution containing either amides (formamide, acetamide, and propionamide) or diols (ethylene glycol and propylene glycol), but not in a solution containing glycerol (triol) or meso-erythritol (tetraol). However, when less 0.1 ng of cRNA was injected per oocyte, there was no difference with the controls.
Membrane Expression of HUT11A in Oocyte Plasma Membranes-Since the transport studies suggested functional differences in Xenopus oocytes injected with small and large amounts of HUT11A cRNA, paraffin-embedded sections of oocyte membranes were immunostained with a well characterized, affinity-purified antibody directed against the N terminus of the Kidd/urea transporter protein (7, 22) (Fig. 7A). Although oocytes injected with 0.05, 0.1, and 40 ng of HUT11A cRNA/ oocyte exhibited similar levels of urea transport (see Fig. 5A), immunocytochemical staining clearly revealed increasing levels of HUT11A expression on the plasma membrane (Fig. 7A). No staining was obtained with water-injected oocytes.
To determine whether the functional properties of HUT11A could be explained by the formation of large oligomers, oocytes from the same batches were examined by electron microscopy (Fig. 7B). No ultrastructure alteration of the oocyte plasma membrane were detected in these experiments. The density of IMP inserted in the P-face plasma membrane was measured, which was 289 Ϯ 28/m 2 for water injected oocytes (Fig. 7C). After subtraction of this endogenous background, the net minimal IMP density increased from 200 Ϯ 49/m 2 for oocytes injected with 0.05 ng to 458 Ϯ 88/m 2 for those injected with 40 ng of HUT11A cRNA (Fig. 7C). Apparently, an 800-fold increase in injected cRNA resulted only in a 2.3-fold increase in IMP density.
As a preliminary analysis of the membrane organization of the urea transporter, the diameter of P-face particles was determined in oocytes expressing HUT11A (40 ng of cRNA injected) and control oocytes. Control oocytes and oocytes expressing HUT11A appeared to express P particles with a mean diameter of 7 Ϯ 0.5 and 6.5 Ϯ 0.5 nm, respectively (Fig. 7D). Assuming that each membrane helix occupies 1.40 Ϯ 0.03 nm 2 (27) and a film thickness of 1.20 Ϯ 0.2 nm, our cross-sectional area data predict a membrane protein containing 9 Ϯ 3 helices. If the hydrophobicity profile, predicting 10 transmembrane domains, is assumed to be correct, these data suggest that the HUT11A urea transporter functions as a monomer. DISCUSSION HUT11A, and Not HUT11, Is the Physiological Urea Transporter-In the present study, we used RT-PCR amplification of human reticulocyte RNAs to identify a new cDNA clone, called HUT11A, encoding a polypeptide nearly identical to the Jk blood group/urea transporter characterized previously as the predicted product of the HUT11 clone (6, 7). However, HUT11A and HUT11 encode two distinct polypeptides of 389 and 391 residues, respectively, which differ by a Val-Gly dipeptide motif after Pro-227. The two polypeptides differ also by a Lys 44 3 Glu polymorphism, unrelated to the JK*A/JK*B alleles.
PCR analysis of genomic DNA from 126 unrelated donors and ribonuclease protection assay have unambiguously shown that the JK gene and the JK transcript encode a protein carrying 2 instead of 3 Val-Gly motifs. Therefore, the previously reported HUT11 clone (6) may result either from a rare JK allele encoding 3 Val-Gly motifs or from a cloning artifact in the cDNA library used.
HUT11A, but Not HUT11, Confers Water Permeability on Oocytes-When we compared the functional properties of the HUT11A and HUT11 transporters, we found that Xenopus oocytes injected with 40 ng of cRNAs encoding HUT11A or HUT11 both expressed proteins at the plasma membrane of oocytes and exhibited a detectable urea transport activity (Fig.  3, A and B). However, the facilitated urea transport mediated by HUT11, but not by HUT11A, was highly sensitive to pCMBS and phloretin. In addition, oocytes expressing HUT11A appeared to be highly permeable to water, whereas oocytes expressing HUT11 had the same water permeability as the water-injected controls (Fig. 3C). In fact, HUT11A-expressing oocytes swelled significantly faster than AQP1-expressing oocytes. These findings are in line with the results of Yang and Verkman (31), who reported an increase in water permeability of Xenopus oocytes expressing the rat homologue of HUT11A called UT3.
In HUT11A-expressing oocytes, urea transport was not inhibited by mercurials, whereas water transport could be inhibited by pCMBS but not by HgCl 2 (Fig. 4, A and B). In addition, the HUT11A urea transport could not be inhibited by phloretin, whereas the water permeability could. In this respect, it was unclear whether the human urea transporter was the functional equivalent of the rat UT3 protein, because one report indicated that HgCl 2 did not inhibit UT3-mediated urea and water transport (31), whereas others (32,33) showed inhibition of UT3 urea transport by pCMBS. Moreover, these reports showed a strong inhibition of UT3 urea transport, whereas our data showed no inhibition of HUT11A urea transport by phloretin. At this stage, these results indicated that, although our genomic analyses proved that HUT11A was the physiological urea transporter, HUT11, but not HUT11A, showed the inhibitory features found for the native red cell urea transporter (8).
HUT11A-mediated Water Permeability and Uptake of Small Solutes Are Caused by Overexpression-A possible explanation for the observed differences between HUT11 and HUT11A was the relatively high expression level of HUT11A in oocytes (Fig.  3A). Dose-response analyses revealed that low HUT11A expression levels conferred urea, but not water, permeability on oocytes (Fig. 5A). In combination with immunocytochemical analyses (Fig. 7, A-C), it can furthermore be concluded that an increase in HUT11A plasma membrane expression did result in an increase in water permeability, whereas the urea permeability was saturated. This may be due to a saturation of the oocyte cell machinery or to some intracellular degradation of the cRNA injected. The latter hypothesis is unlikely since Northern blot analysis showed that the cRNA injected in oocytes remained stable at least for 72 h (13).
These results indicate that above 0.1 ng of injections, the urea transport rate is not determined by the number of HUT11A transporters in the membrane, but by an unknown factor intrinsic to the oocytes. In line with this finding is that injection of oocytes with 3 or 10 ng of AQP2 cRNA does not result in increased water permeability, whereas the plasma membrane expression level is significantly increased. 2 Our results also explain the observed water permeability of UT3, as reported by Yang and Verkman. (31), because these authors injected 5 ng of the corresponding cRNA.
From the structure-function point of view, the water permeation of HUT11A is interesting. As AQP-1 and HUT11A do not share any sequence homology and as no electron crystallography data on the urea transporter are available, one explanation for this water transport activity might be that the HUT11A transporter takes another conformation at high density in the oocyte membrane allowing water transport to occur. This hypothesis is corroborated by the following data. Oocytes were injected with water as control, 0.05, 0.1, 20, or 40 ng of cRNA encoding HUT11-A. A, 3 days after injection, fixed oocyte sections were prepared and stained as described under "Materials and Methods," and the images were analyzed as described in Fig. 3A. B, P-faces obtained after freeze-fracture of the same batches of injected oocytes were replicated by unidirectional shadowing with platinum. 95,000ϫ final magnification representative images for each set of conditions are shown. C, histogram of particle density (particles/m 2 ) from images (n ϭ 4 -15) of different P surfaces of two or three oocytes for each set of conditions. D, size distribution of P-face particles in oocytes expressing HUT11A (ࡗ) and water-injected oocyte control (Ⅺ).The particle densities were 289 Ϯ 28/m 2 (n ϭ 294) and 747 Ϯ 88/m 2 (n ϭ 2035) for water-injected controls and 40 ng of cRNA injection, respectively. HUT11A particles had a mean diameter of 6.5 Ϯ 0.5 nm close to the IMP density of the water-injected oocyte control.
expected a measurable P f of about 75 m/s for 0.1-ng injections. Second, with the injection of 40 ng of HUT11A cRNA, the urea transport could not be inhibited by pCMBS or phloretin, whereas with an injection of 0.1 ng of cRNA there was significant inhibition with phloretin (Fig. 6A). If the conformation of the urea transporter is indeed changed with high injections, the relevance of the single-channel water permeability for UT3 (1.4 ϫ 10 Ϫ14 cm 2 /s) as determined by Yang and Verkman (31) is doubtful. Alternatively, the water permeability noted with HUT11A might be a result of some undefined oocyte perturbation caused by injection of large amounts of cRNA. It is striking in this regard that injection of 800-fold more HUT11A cRNA (0.05-40 ng) only resulted in a 2.3-fold net increase membrane particles (Fig. 7C). However, this alternative is unlikely because, when large amounts of cRNAs for HUT11, HUT2, or the anion exchanger Band-3 (AE1) were injected, no permeability to water and small solutes was detected (data not shown).
A finding in line with the water transport activity mediated at high levels of HUT11A expression is that small non-electrolytes including amides and diols pass through the oocyte membrane via the same unknown mechanism (Fig. 6B). However, the glycerol, meso-erythritol, and raffinose exclusion was sizeselective. In any case, these data strongly suggest that, at high levels of HUT11A expression, there is a loss of transport specificity that is confirmed by the loss of phloretin sensitivity.
Since the HUT11 and HUT11A expression constructs differ only in the described coding sequence, the difference noted in expression of both proteins was likely to be caused by the amino acid differences between HUT11 and HUT11A. Substitution of Lys for Glu 44 in HUT11A followed by expression in oocytes revealed that this polymorphism did not modify the transport properties of HUT11A. 3 Therefore, the presence of only 2 Val-Gly motifs might be critical for the functional properties of HUT11A or, alternatively, the stability of the protein. Further studies will be required to define the role of this motif more clearly.
The unusual properties of HUT11A seen at high density level in oocytes were not related to the formation of oligomers since electron microscopic analysis revealed a HUT11A particle diameter of about 6.5 Ϯ 0.5 nm, which is consistent with a monomeric form of the Kidd/urea transporter (characterized by 9 Ϯ 3 transmembrane helices). This result is in conflict with a size of 469 Ϯ 36 kDa for the red cell urea transporter determined by radiation inactivation (34). As N-glycosylation of HUT11A glycoprotein in the red cells cannot clear up this size difference (7), further investigation will be required to address this issue.
Low Expression of HUT11A Confers Physiological Urea Transporter Characteristics-Upon injection of oocytes with low level (0.1 ng) of HUT11A cRNA, urea transport was facilitated and was slightly sensitive to pCMBS (18%) and strongly sensitive to phloretin (79%). No water or small solute transport occurred under these conditions. These data are all in line with the physiological characteristics from the red cell urea transporter (8), although we cannot explain why HUT11A-mediated urea transport was not strongly inhibited by pCMBS in a nonerythroid context. Whether this may be due to some difference in the membrane properties of these cells require further studies.
From our data, we calculate that the number of HUT11A molecules expressed in oocyte membrane on injection of 0.1 ng of cRNA is similar to that of the physiological urea transporter in red blood cells. Assuming a density of 14,000 -32,000 copies of Jk/urea transporter per red cell (35,36), the calculated surface density is 100 -200 molecules/m 2 , which corresponded to the particle density seen by electron microscopy in oocytes injected with 0.05-0.1 cRNA/oocyte (Fig. 7C). We also noted that the urea permeability of oocytes injected with 0.05 ng (1.24 ϫ 10 Ϫ4 cm/s measured at 10 s) is close to the urea permeability of the erythrocyte plasma membrane (2.70 ϫ 10 Ϫ4 cm/s; Ref. 37). Thus, upon injection of oocytes with 0.05-0.1 ng of HUT11A cRNA, physiological densities of urea transporter are expressed, which confer no detectable water permeability or small solute uptake. According to these results, it is unlikely that HUT11A confers the residual red cell water permeability in Colton-null erythrocytes, which lack AQP1 (38). More likely, this permeability can be accounted for by AQP3, which is also expressed in red cells (39).
In summary, two important conclusions can be deduced from these studies: (i) the oocyte expression system must be used under carefully controlled conditions to preserve urea transport specificity, and (ii) HUT11A is the red cell urea transporter, which accounts for urea permeability but not water permeability or small solute uptake of normal erythrocytes in physiological conditions.