Antigenic and functional properties of the human red blood cell urea transporter hUT-B1.

The Kidd (JK) blood group locus encodes the urea transporter hUT-B1, which is expressed on human red blood cells and other tissues. The common JK*A/JK*B blood group polymorphism is caused by a single nucleotide transition G838A changing Asp-280 to Asn-280 on the polypeptide, and transfection of erythroleukemic K562 cells with hUT-B1 cDNAs carrying either the G838 or the A838 nucleotide substitutions resulted in the isolation of stable clones that expressed the Jk(a) or Jk(b) antigens, respectively, thus providing the first direct demonstration that the hUT-B1 gene encodes the Kidd blood group antigens. In addition, immunochemical analysis of red blood cells demonstrated that hUT-B1 also exhibits ABO determinants attached to the single N-linked sugar chain at Asn-211. Moreover, immunoadsorption studies, using inside-out and right-side-out red cell membrane vesicles as competing antigen, demonstrated that the C- and N-terminal ends of hUT-B1 are oriented intracellularly. Mutagenesis and functional studies by expression in Xenopus oocytes revealed that both cysteines Cys-25 and Cys-30 (but not alone) are essential for plasma membrane addressing. Conversely, the transport function was not affected by the JK*A/JK*B polymorphism, C-terminal deletion (residues 360-389), or mutation of the extracellular N-glycosylation consensus site and remains poorly para-chloromercuribenzene sulfonate (pCMBS)-sensitive. However, transport studies by stopped flow light scattering using Jk-K562 transfectants demonstrated that the hUT-B1-mediated urea transport is pCMBS-sensitive in an erythroid context, as reported previously for the transporter of human red blood cells. Mutagenesis analysis also indicated that Cys-151 and Cys-236, at least alone, are not involved in pCMBS inhibition. Altogether, these antigenic, topologic, and functional properties might have implications into the physiology of hUT-B1 and other members of the urea transporter family.

In the last 10 years, facilitated urea transporters (UT), 1 which play a major role in urinary concentration mechanism, have been molecularly characterized in different animal species (1) following the cloning by functional expression in Xenopus oocytes of the rabbit urea transporter (2). Currently, two types of mammalian UT can be distinguished, those that are encoded by the Slc14a2 gene (type UT-A) and those that are encoded by the Slc14a1 gene (type UT-B) (3)(4)(5)(6). In humans and mouse, these two UT genes occur in tandem on chromosome 18q12 (for humans, see GenBank TM accession number AC023421) (7)(8)(9). The Slc14a2 gene encodes five alternative spliced isoforms named UT-A1 to -A5 mainly expressed in the renal tubules, except for UT-A5, which is expressed only in testis (6,10). The Slc14a1 gene only encodes the UT-B1 protein expressed on the red blood cells (RBCs) (8,11) and in endothelium of the descending vasa recta irrigating renal medulla (12,13). UT-B1 is also expressed in various organs, as shown in the rat model (14 -16).
Recently, it was reported that the urea transport function of human RBCs and the Kidd (JK) blood group are carried by the same protein, hUT-B1/Jk (8). The two major codominant alleles of the JK gene, JK*A and JK*B, have a similar frequency in Caucasian populations (0.51 and 0.49, respectively) and define the three common phenotypes Jk(aϩb-), Jk(a-bϩ), and Jk(aϩbϩ) (17,18). The genetic basis of the JK*A/JK*B blood group polymorphism is a single nucleotide transition G838A changing Asp-280 to Asn-280 in the Jk a and Jk b polypeptide, respectively (19). Because of hUT-B1, RBCs do not undergo excessive cell volume changes during their transit in the vasa recta irrigating the hypertonic renal medulla. In addition, they participate in medulla urea sequestration by taking up urea when flowing in the descending vasa recta and subsequently releasing urea when flowing in the ascending vasa recta (20). The presence of hUT-B1 in the endothelium descending vasa recta enables a countercurrent exchange of urea with the ascending vasa recta, a process that also contributes to the urea medulla gradient required for water reabsorption. More recently, renal rUT-B1 has been shown to be regulated by antidiuretic hormone independently from medulla hypertonicity (16). Despite these functional properties, hUT-B1 deficiency, which occurs in the rare human phenotype called Jk null (21), is not associated with any obvious clinical syndrome, except for mild urinary concentration defect, which is, however, more severe in transgenic UT-B null mice (22,23). The absence of UT-B1 prevents Jk null RBCs from releasing urea as they traverse the ascending vasa recta, thus decreasing the efficiency of the countercurrent exchange and the renal concentration ability. Although a very rare form of Jk null may be inherited as a dominant character, most result from homozygous inheritance of a silent allele at the JK locus (18) and may arise by at least four distinct molecular mechanisms: (i) splice site mutations causing the skipping of either exon 6 or exon 7 (5), (ii) missense mutation resulting in a S291P substitution (24), (iii) nonsense * This investigation was supported in part by the Institut National de la Santé et de la Recherche Médicale (INSERM). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
hUT-B1 is a hydrophobic polypeptide of 389 amino acid residues with a 42.5 kDa apparent molecular mass, which is reduced to 36 kDa following deglycosylation. Of the 10 cysteine residues present in hUT-B1 and hUT-A2 proteins, 7 are conserved and aligned at equivalent positions, but 3 (Cys-25, Cys-30, and Cys-151 on hUT-B1) are present at distinct positions ( Fig. 1) (11). The predicted membrane topology based on hydropathy plots consists of two repeated hydrophobic domains spanning the membrane five times each, linked by a large extracellular loop carrying a single N-glycan chain attached to Asn-211 ( Fig. 1) (11,24). Each repeat contains a conserved sequence motif (LPXXTXPF), suggesting an internal duplication (27). According to the predicted model, both the N-terminal stretch of 60 amino acids and the C-terminal domain of 35 amino acids are on the cytoplasmic side of the cell membrane, and the common JK*A/JK*B polymorphism resulting from the D280N substitution (19) is located in the fourth extracellular loop (Fig. 1). With this unusual pattern of hydrophobicity, about 70% of hUT-B1 is embedded in the cell membrane, a feature also observed with the isoforms of UT-A gene (3). Upon expression in Xenopus oocytes, hUT-B1 confers a high urea permeability, which is strongly inhibited by phloretin but poorly inhibited by para-chloromercuribenzene sulfonate (pC-MBS) (28). This result is surprising in view of the fact that urea permeability of human RBCs is decreased by both inhibitors (29). This discrepancy remains to be clarified.
The present report has four purposes: (i) to provide direct evidence that hUT-B1 carries the Jk blood group antigens by flow cytometry analysis of Jk-K562 transfectants, (ii) to analyze the membrane topology of hUT-B1 in the RBC membrane, (iii) to characterize the urea transport mediated by hUT-B1 in an erythroid context by expressing the protein in the K562 erythroleukemic cell line, and (iv) to examine the structurefunction relationship with respect to plasma membrane targeting and pCMBS sensitivity of the hUT-B1 protein.

MATERIALS AND METHODS
Blood Samples and Reagents-RBC samples from individuals of common and rare Jk phenotypes were obtained from the Centre National de Référence sur les Groupes Sanguins (CNRGS, Paris, France). Restriction endonucleases and modifying enzymes were from New England Biolabs (Hertfordshire, UK). The [ 14 C]urea (1.96 GBq/mmol) and the [ 3 H]raffinose (188.7 GBq/mmol) came from Amersham Biosciences and PerkinElmer Life Sciences, respectively. Pwo DNA polymerase from Roche Molecular Biochemicals was used for PCR amplification. Nucleotide sequences were determined on both strands with ThermoSequenase fluorescently labeled primer cycle sequencing kit from Amersham Biosciences using 5Ј(Cy5) primers (Genset, Paris, France) and an automated Alf-Express sequencer (Amersham Biosciences). The human antiserum containing an alloanti-Jk3 was obtained from an immunized Jk(a-b-) individual also called Jk null . The human monoclonal antibodies (mAbs) anti-Jk a (IgM, MS15) or anti-Jk b (IgM, MS8) and the human polyclonal antisera anti-Jk a or anti-Jk b were from Biotest AG (Dreieich, Germany). Rabbit polyclonal antisera and affinity purified antibodies directed against the N-terminal region (anti-Nter, residues 8 -22) or the C-terminal region (anti-Cter, residues 377-389) of the hUT-B1/Jk protein were described previously (8,12).
Cell Culture, Transfection, and Flow Cytometry Analysis-Human erythroleukemic K562 cells were obtained from the American Type Culture Collection (Manassas, VA) and were grown in Iscove's modified Dulbecco's medium with Glutamax-1 (Invitrogen) supplemented with penicillin-streptomycin and 10% fetal calf serum. To establish a stable cell line expressing the Kidd/urea transporter, hUT-B1 cDNAs (nucleotides Ϫ9 to 1182) carrying either a G838 (JK*A allele) or an A838 (JK*B allele) were amplified by PCR from the Jk a /Jk b -pT7TS constructs (19) using SP-1 and AS-1 primers (see Table I), subcloned into the pCEP4 episomal expression vector (Invitrogen), and transfected into K562 cells using the Lipofectin reagent according to the manufacturer's instructions (Invitrogen). Stable transfectants resistant to hygromycin (300 g/ml) were selected for Jk antigen expression by immunomagnetic separation using a human anti-Jk3 antiserum and Biomag Goat antihuman-IgG (PerSeptive Biosystems, Connecticut, MA). Stable clones were isolated, and Jk a , Jk b , and Jk3 antigens expression was analyzed by flow cytometry. Briefly, K562 transfectants (3-5 ϫ 10 5 ) were incubated for 60 min at 22°C with appropriate antibodies used at saturating concentration and then washed and stained with 100 l of phycoerythrinconjugated F(abЈ) 2 fragments of goat anti-human IgG diluted 1:40 (Coulter/Immunotech, Marseille, France). After another washing step, 20 nM TO-PRO-1, Molecular Probes (Interchim, Montluçon, France) was added to the cell suspension 15 min before FACScan analysis (BD Phar-Mingen) to exclude dead cells (TO-PRO positive-cells).
Immunoadsorption Assays-Right-side-out (ROVs) and inside-out (IOVs) vesicles were prepared from Jk-positive RBCs and purified through a barrier of 8% (w/v) dextran T70, as described previously (32). One volume of each rabbit polyclonal serum anti-Nter or anti-Cter of hUTB1 was mixed for 2 h at 37°C with 4 volumes of sealed vesicles at the protein concentration of 2.5 mg/ml and subsequently centrifuged at 20,000 ϫ g for 15 min. Free antibodies in the supernatants and the bound antibodies eluted from the vesicles by the digitonin method (33) were analyzed by Western blot on Jk-positive RBC membrane proteins separated by SDS-PAGE and transferred to nitrocellulose sheets (Schleicher and Schuell, Keene, NH; 0.1 m). Bound antibodies were detected with alkaline phosphatase-labeled goat anti-rabbit IgG (1:800 dilution) and the alkaline phosphatase substrate kit (Bio-Rad).
Site-directed Mutagenesis and Truncation of the hUT-B1/Jk cDNA-All primers used are given in Table I. Single point mutations C25S, C30S, C151S, N211I, and C236S were introduced into the Jk a allelic cDNA using a two-step PCR approach with the common SP-1 and AS-1 primers, in addition to sequence-specific mutagenic primers overlapping each other, AS-2 to 6 and SP-2 to 6, respectively. The second PCR was performed with 1/100 of each two first PCR reactions. The double mutation C25S,C30S was generated similarly using the Jk a cDNA carrying the single mutation C25S as initial PCR matrix. Deletions of nucleotide sequences encoding the first 59 (⌬N) or the last 30 (⌬C) amino acids of the Jk a polypeptide were generated by PCR amplification between SP-⌬N and AS-1 or SP-1 and AS-⌬C primers, respectively. The double Nand -C-terminal deletion (⌬Nϩ⌬C) construct was obtained using SP-⌬N and AS-⌬C primers. For efficient translation, artificial ATG codon and Stop codon were inserted at the 5Ј end of the SP-⌬N and AS-⌬C primers, respectively. All PCR amplifications were done under stringent conditions: 94°C for 45 s (1 cycle); 94°C for 45 s, 58 -72°C depending of primer pairs used for 30 s, 72°C for 60 s (30 cycles); 72°C for 1 min (1 cycle). All mutant Jk a cDNAs were subcloned into the EcoRV-digested pT7TS plasmid (kindly provided by P. Krieg, Austin, TX) except for the C151S and C236S mutants, which were subcloned into KpnI-HindIII-digested pCEP4 vector and transfected into K562 cells as described above. All constructs were sequenced on both strands using an automated Alf-Express sequencer (Amersham Biosciences) or an ABI-Prism 310 Genetic Analyser (Applied Biosystems, Foster city, CA) to confirm that the correct junctions/mutations were obtained.
Oocyte Expression, Flux Measurements, and Immunocytochemistry-Capped sense RNAs were transcribed in vitro from the pT7TS-cDNA constructs linearized with SmaI restriction enzyme using T7 polymerase and the mCAP mRNA capping kit from Stratagene (La Jolla, CA). Expression studies were carried out by microinjection of each cRNA (0.1 ng/oocyte in 50 nl) in collagenase-treated Xenopus laevis oocytes, and functional tests were performed 3 days after injection as described previously (24,28). In urea transport inhibition experiments, oocytes were incubated in 1 mM pCMBS (Sigma) or 1 mM phloretin (Sigma) for 20 and 10 min, respectively, before and during the assay at 18°C. From the same oocyte batches, groups of three to six oocytes without chorionic membrane were embedded in paraffin as described (34). Sections (7 m thick) were stained overnight at 4°C with 10 g/ml of affinity-purified antibodies anti-Nter or anti-Cter and visualized with fluoresceinconjugated goat anti-rabbit IgG (1:100 dilution) for 1 h at room temperature (35).
Stopped Flow Experiments-Water and solute transport analysis were performed at 15°C by following the 90°light scattering variations ( exc : 530 nm) using a stopped flow spectrophotometer (SFM3, Biologic, Claix, France) as described previously (36). 60 l of phosphate-buffered saline washed K562 cells resuspended at the density of 0.5-1.0 ϫ 10 7 cells/ml were mixed to an equal volume of hypertonic solution containing solutes (sorbitol or urea) to create an inwardly directed osmotic gradient of 150 mosM/kg of H 2 O, and the scattered light intensity variation was followed.
The first part of the curve corresponded to a water efflux to osmotically equilibrate cells with the external medium and allowed to calculate the osmotic water permeability coefficient (P f ). Data averaged from 5 to 10 time courses were fitted to a double exponential function by using the simplex procedure of the Biokine software (Biologic), and the apparent PЈ f was calculated according to the equation: where K exp is the first exponential rate constant; V (t) is the relative volume of the K562 cells at time, t; V osm /S is the ratio of cell osmolyte volume to cell surface area (7.94 10 Ϫ5 cm); V w is the molar volume of water (18 cm 3 /mol), and C in and C out are the initial concentrations of total solute inside and outside the cells, respectively.
The second part of the curve, which was only observed with Jk transfectants in urea condition, resulted from cell swelling as water accompanied urea influx to maintain the osmotic equilibrium. The analysis of this part of the curve was fitted to a single exponential function using the same software as indicated above, and the apparent PЈ urea was calculated using the following equation: PЈ urea ϭ k exp ⅐(Imp out ϩPerm out )/S⅐((IЈ o ⅐VЈ o /V (t) )Ϫ Imp out ) (37), where IЈ o and VЈ o represent the initial concentration of the impermeant solute and the cell volume, respectively. Imp out and Perm out were the external concentrations of impermeant and permeable solutes, respectively. In some experiments, pCMBS (0.1-3.0 mM) was added to the cells and hypertonic media 15 min before the shrinkage-swelling measurement. Reversibility was carried out after addition of 5 mM ␤-mercaptoethanol (Sigma).

RESULTS
The hUT-B1 Polypeptide Carries Jk and ABO Blood Group Specificities-As Northern blot analysis revealed that the K562 erythroleukemic cell line lacked hUT-B1 transcripts (not shown), stable transfectants expressing the hUT-B1/Jk a and hUT-B1/Jk b allelic cDNAs were established in these cells. Flow cytometry analysis showed that polyclonal antiserum anti-Jk a strongly reacted with K562-Jk a but not K562-Jk b transfec-  tants, whereas antiserum anti-Jk b strongly reacted with K562-Jk b but not K562-Jk a transfectants (Fig. 2). Identical results were obtained with human mAbs directed against Jk a and Jk b antigens (not shown). The geometric mean of fluorescence was higher with Jk b as compared with Jk a transfectants (Fig. 2), presumably because of a better transfection efficiency. As expected, both transfectants reacted with the human alloanti-Jk3, indicating that both the JK*A and JK*B alleles also encode the Jk3 antigenic specificity. In contrast, parental K562 cells did not react with antibodies directed against Jk a , Jk b , or Jk3 antigens (Fig. 2).
Although hUT-B1 is known to carry a single N-glycan attached to Asn-211 ( Fig. 1) (24), it has not been determined whether this oligosaccharide chain may carry some blood group determinants. To address this issue, the hUT-B1 protein was immunopurified from RBCs of known ABO blood group phenotypes with the affinity-purified anti-Nter antibody to hUT-B1, and the immunoprecipitate was submitted to SDS-PAGE and immunoblotted with murine mAbs anti-A or anti-B. hUT-B1 from blood group A individuals was detected with mAb anti-A only, and hUT-B1 from B individuals was detected with mAB anti-B only, whereas hUTB1 from AB individuals was detected with both mAbs anti-A or anti-B (Fig. 3). In contrast, hUT-B1 prepared from group O individuals were unreactive with anti-A or anti-B antibodies (Fig. 3). As a control, all preparations, except that prepared from a Jk null individual, gave a strong signal with the affinity-purified anti-Cter antibody to hUT-B1. Altogether, these results clearly demonstrated that the hUT-B1 polypeptide carries Jk blood group antigens, and in addition, exhibits ABO antigens attached to the N-glycan at Asn-211 (Fig. 1).
Topology of the N-and C-terminal Ends of hUT-B1-To provide experimental evidence that the N and C terminus of hUT-B1 are intracytoplasmic, immunoadsorption assays of anti-Nter or anti-Cter antibodies to hUT-B1 onto ROVs and IOVs prepared from human RBCs were performed. Antibodies remaining unabsorbed and those acid-eluted from each type of vesicles were immunoblotted on RBC membrane proteins from Jk-positive individuals as compared with the starting serum (anti-Nter or anti-Cter). These studies revealed that the anti-Nter or anti-Cter antibodies could not be recovered from acid eluate from ROVs (Fig. 4) and remained in the supernatant. In contrast, the antibodies were recovered in the acid eluate but not in the supernatant from IOVs. As a control, both starting sera reacted with hUT-B1 as a diffuse band of 46 -69 kDa, as expected. These data are in agreement with the hydrophobicity profile of hUT-B1 predicting an intracellular orientation of the N and C terminus of the polypeptide (Fig. 1).
Membrane Expression and Urea Transport of hUT-B1 in Oocytes-Membrane expression and urea transport activity of wild type and selected mutants of hUT-B1 were investigated in the Xenopus expression system. Urea uptake at 90 s was performed from oocytes injected with the corresponding cRNAs, and membrane expression was analyzed by immunohistochemistry of oocyte sections stained with affinity-purified anti-Nter or anti-Cter antibodies to hUT-B1.
First, immunocytochemical analysis revealed that the Jk a and Jk b allelic forms of hUT-B1 were expressed at the plasma membrane of cRNA-injected oocytes (Fig. 5, B and C), whereas waterinjected oocytes remain unstained with both the anti-Nter or anti-Cter antibodies (Fig. 5A). Transport studies showed that both proteins mediated a urea flux of the same rate inhibited strongly by phloretin as reported previously (Fig. 6) (28), but not (or very mildly in other experiments) by pCMBS.
Next, we examined the effect of mutations of the functional N-glycosylation consensus site and of cysteine residues Cys-25 and Cys-30 specific to hUT-B1 (Fig. 1). Oocytes injected with cRNAs encoding mutant forms of the JK*A allele carrying the substitutions N211I, C30S (Fig. 5, G and H), and C25S (not shown) normally expressed the mutant proteins at the plasma membrane and transported urea similarly to the wild type Jk a protein (Fig. 6). When a cRNA preparation encoding the Jk a polypeptide truncated of the last 30 Cterminal residues was injected into oocytes, a weak decrease of the plasma membrane expression was noted without significant change in urea transport activity (Figs. 5F and 6). Conversely, the deletion of the first 59 N-terminal residues completely abolished expression at the oocyte membrane, therefore precluding any transport function study (Figs. 5E and 6). As this N-terminal domain contains two cysteines Cys-25 and Cys-30, which could play some role in the UT function, the protein-protein interactions, and the addressing to the plasma membrane, these residues were mutated to serine. Interestingly, neither the C25S nor the C30S mutations alone affected membrane expression and transport activity (Figs. 5H and 6), but double mutation of the two cysteines (C25S,C30S) completely abolished membrane expression in oocytes (Fig. 5I). In agreement with these observations, no urea transport activity was detected when oocyte membrane expression was absent (Figs. 5 and 6). In all instances, when a urea transport occurs, phloretin behaved as a potent inhibitor, whereas pCMBS did not or behaved very poorly (Fig. 6).
Water and Urea Permeabilities of K562 Transfectants-To understand why urea transport mediated by the cloned hUT-B1 expressed in Xenopus oocytes and the physiological UT of human RBCs differ in their sensitivity to pCMBS, the urea transport function of hUT-B1 was analyzed in an erythroid context following expression in the erythroleukemic cell line K562. Accordingly, cell suspensions of K562-Jk a or K562-Jk b transfectants were rapidly mixed with a hyperosmolar solution to drive osmotic water efflux and cell shrinking. The rate of increase in scattered light intensity corresponding to the cell shrinking was measured in K562 cells with (0.15 M) sorbitol as osmolyte (Fig. 7A). The apparent PЈ f values in cm s Ϫ1 Ϯ S.D. calculated were 1.50 Ϯ 0.47 ϫ 10 Ϫ2 , 1.20 Ϯ 0.65 ϫ 10 Ϫ2 , and 1.48 Ϯ 0.28 ϫ 10 Ϫ2 for wild type K562 cells, K562-Jk a , and K562-Jk b transfectants, respectively (Table II). When the hyperosmolarity was performed with (0.15 M) urea, the light scattering increase in wild type K562 cells (PЈ f ϭ 1.71 Ϯ 0.53 ϫ 10 Ϫ2 cm s Ϫ1 , n ϭ 5) was identical to the rate increase observed with sorbitol excluding an endogenous urea transport activity (Fig. 7A). However, only the K562-Jk a and -Jk b transfectants presented, after a brief scattered light intensity increase due to the efflux of water, a significant decrease of light scattering due to the cell swelling directly related to the osmotic influx of water accompanying urea uptake to ensure osmotic equilibrium (Fig. 7B). The PЈ urea values calculated were 1.78 Ϯ 0.34 ϫ 10 Ϫ5 and 3.35 Ϯ 0.9 ϫ 10 Ϫ5 cm s Ϫ1 for K562-Jk a and Jk b , respectively (Table II).
The urea permeability of K562-Jk a was measured in the presence of 0.3 mM pCMBS. We found that the urea flux was strongly reduced to 20%, a value close to that reported for RBCs (29). Moreover, pCMBS inhibition was reversed by 5 mM ␤-mercaptoethanol (Fig. 7B). Identical results were obtained with K562-Jk b (not shown). Next, to identify the cysteine residues involved in the pCMBS sensitivity, the unique extracellular cysteine Cys-236 and the intramembranous cysteine Cys-151 specific of hUT-B1, near the cell surface, were mutated to FIG. 3. hUT-B1 carries ABO antigens on the N-linked sugar chain. Immune precipitates obtained with the affinity-purified antibody directed against the N terminus of hUT-B1 using Triton X-100 solubilized RBC membranes from defined ABO phenotypes were separated by SDS-PAGE, transferred to nitrocellulose sheets, and incubated with the mAbs anti-A or anti-B. RBC membranes from Jk null donor and the affinity-purified antibody directed against the C terminus of hUT-B1 were used as negative and positive controls, respectively. After washings, bound antibodies were revealed as described under ''Materials and Methods.''

FIG. 4. Western blot analysis of antibodies in supernatants and eluates from ROVs and IOVs.
Membrane proteins from Jk-positive RBCs (100 g/lane) were separated on SDS-PAGE, transferred onto nitrocellulose, and immunologically stained with supernatants (1:250 dilution) and acid eluates (1:250 dilution) from IOVs and ROVs preincubated with antibodies against the N or C terminus of hUT-B1. Staining with serum before immunoadsorption assay (1:500 dilution) was performed as a positive control. After extensive washings, antibodies bound to hUT-B1 (46 -69 kDa) were visualized as described under ''Materials and Methods.'' serine ( Fig. 1). Thus the pCMBS sensibility of K562-Jk a and of two K562 transfectants expressing Jk a mutant proteins carrying the C151S and C236S substitutions at the same level were compared. All exhibited comparable PЈ urea values found for K562-Jk a and Jk b (Table II). Fig. 7C shows that K562-Jk a and K562-C151S had identical dose-response inhibition by pCMBS, whereas the K562-C236S exhibited a higher sensitivity. At 0.3 mM pCMBS, the remaining transport activities were 55% for the former transfectants and only 20% for the latter. At 3 mM pCMBS, the urea transport activity of all samples was completely inhibited (Fig. 7C). Thus, the urea transport mediated by hUT-B1 is pCMBS-sensitive in an erythroid context, but Cys-151 and Cys-236, at least alone, are not involved in pC-MBS inhibition. D and E were stained with affinity-purified antibodies against the N or C terminus of hUT-B1, respectively, as described under ''Materials and Methods.'' Negative control for antibodies against the C terminus tested with water injected oocytes was identical to section A (not shown). Images were generated using a Nikon Eclipse TE300 microscope (Nikon, Paris, France) (ϫ40 objective) with epifluorescence illumination and treated with a Biocom informatic system of image integration (Biocom, les Ulis, France). allelic forms of the cloned hUT-B1 urea transporter, respectively. In addition, these studies showed that the N-glycan chain attached to Asn-211 of hUT-B1 carries ABO blood group determinants as found for the N-glycan chains of Band-3, Band-4.5 (glucose transporter), and AQP-1 (38 -40). Knowing that all hUT-B1 molecules appear glycosylated in RBCs (8), less than 1% of RBC ABO determinants is carried by the hUT-B1 protein, which is present at a low copy number (1.4 ϫ 10 4 molecules/cell), as compared with Band-3 and AQP-1 proteins, which are present at high copy numbers (10 6 and 2 ϫ 10 5 molecules/cell, respectively). Thus, the antigenic properties of hUT-B1 and its expression in renal tissues might have biological implications in kidney transplantation Topological and Functional Properties of hUT-B1-In accordance with the predicted membrane topology of hUT-B1 (11), expression studies reported here indirectly confirmed the extracellular exposure of the third and fourth loops carrying the N-glycosylation site (at Asn-211) and of the JK*A/JK*B allelic polymorphism, respectively (Fig. 1). Moreover, immunoadsorption studies provide evidence for the intracellular orientation of the C and N terminus of the protein in RBCs (Fig. 1). Similar membrane topology has been found for other membrane proteins such as the anion exchanger Band-3 (AE1) (41) and the water channel AQP-1 (42). Whether the N-and/or C-terminal domains play some role in the UT function and/or protein-protein interaction with other membrane components is presently not known. However, our preliminary results in Xenopus oocytes suggested that cysteines Cys-25 and Cys-30 together (not alone) within the 59 N-terminal amino acids domain are critical for correct addressing and insertion of hUT-B1 in the plasma membrane. Conversely, neither the deletion of the last 30 C-terminal nor the JK*A/JK*B polymorphism had any affect on the expression level and urea transport activity of hUT-B1. Although N-glycosylation often determines membrane expression level and addressing polypeptides to the membrane, (43)(44)(45), we found that unglycosylated hUT-B1 polypeptide is normally addressed and inserted in the oocyte plasma membrane and exhibited a normal urea transport activity as reported for AQP2 (46,47).
Another goal of this study was to analyze protein expression and urea transport activity of hUT-B1 into the erythroleukemic K562 cells to investigate the sensitivity to mercurial agents in an erythroid context. Indeed, recombinant hUT-B1 in oocytes confers a urea permeability, which is poorly inhibited by pC-MBS, whereas the native UT of human RBCs is strongly inhibited by mercurial agents (28,29). Accordingly, the rates of volume change of wild type K562 cells and K562-Jk a and -Jk b transfectants submitted to an osmotic sorbitol gradient were measured by light scattering stopped flow experiments. Under these conditions, parental and K562 transfectants exhibited relatively high apparent water permeabilities close to published values (48). The absence of a further increase in K562- FIG. 7. Urea transport analysis in K562 transfectants by stopped flow light scattering. A, stopped flow curves obtained with wild type K562 cells submitted to 150 mosM/kg H 2 O urea or sorbitol gradients. The light scattering increase was similar with both gradients. B, typical stopped flow curves obtained with K562-Jk a transfectant submitted to a 150 mosM/kg H 2 O urea gradient. The increase light intensity corresponding to initial water efflux and the decrease light intensity resulting from water associated with the solute influx were recorded during 18 s. The curve was fitted to single exponential, and PЈ urea values were calculated as described under ''Materials and Methods'' (as well as in Table II). For inhibition experiments, 0.3 mM pCMBS was added 15 min before and during the assays. To test the reversibility of pCMBS inhibition, subsequent incubation for 5 min in 5 mM ␤-mercaptoethanol (␤-me) was performed before the assays. C, pCMBS sensitivity of urea transport of wild type (Jk a ) and mutant Jk a (C151S) and Jk a (C236S) polypeptides expressed in K562 cells. The transfectants were incubated with increasing concentration of pCMBS (0 -3 mM) 15 min before the assay as described under ''Materials and Methods.'' The mean data from three experiments are shown. Jk a and -Jk b transfectants confirmed that the hUT-B1 is not a water channel as reported previously (28). When an osmotic urea gradient was imposed, the urea transport capacity of hUT-B1 was preserved in K562-Jk a and -Jk b transfectants, the calculated PЈ urea values being of 1.78 Ϯ 0.34 ϫ 10 Ϫ5 and 3.35 Ϯ 0.9 ϫ 10 Ϫ5 cm s Ϫ1 , respectively. The difference of PЈ urea values observed might be related to the difference in the transporter density in K562 transfectants, as suggested by the shift in fluorescence intensity detected by the alloanti-Jk3 antibody (Fig. 2). We further demonstrated that the urea flux mediated by K562-Jk a and -Jk b transfectants was strongly inhibited by a low concentration of pCMBS and was reversed by ␤-mercaptoethanol, as reported for the UT of human RBCs (49). These findings suggest that the low pCMBS sensitivity in Xenopus oocytes reported previously was probably related to membrane environmental factors (lipid and protein) specific to this heterologous expression system (50). Thus, it is possible that hUT-B1 adopts a different conformation in oocyte and red cell membranes, which could modify the accessibility of the pCMBS to the different cysteines. However, a strong pCMBS inhibition was observed when oocytes were injected with HUT11 cRNAs, a rare variant of the RBC urea transporter exhibiting three Val-Gly repeat motifs after proline 227 (third extracellular loop) versus 2 Val-Gly in the physiological hUT-B1 transporter ( Fig. 1) (11,28). As pCMBS inhibits urea/water transports by reacting with sulfhydryl groups of proteins, the unique cysteine Cys-236 in the third extracellular loop close to the site of the dipeptide insertion was mutated to serine (Fig. 1). In parallel, the intramembranous cysteine Cys-151 specific of hUT-B1, near the cell surface, was also mutated. Our results indicated that the urea flux measured in K562-Jk a (C151S) and K562-Jk a (C236S) transfectants was not affected by pCMBS but was even increased in the latter variant. As a single cysteine mutation cannot abolish pCMBS inhibition, as reported for AQP1 (51), we cannot exclude that the pCMBS sensibility results from a combination of several cysteines. Moreover, it can be stressed that the absence of the Val-Gly dipeptide is sufficient to strongly reduce pCMBS inhibition in the oocyte expression system, whereas the C236S substitution increases pCMBS sensitivity in K562 cells. Presumably, such opposite effects might be due to conformation changes of the third extracellular loop of hUT-B1 via amino acid sequence changes, which in turn modify pCMBS accessibility to cysteine residues. These observations suggest a key role of the third extracellular loop connecting the two hydrophobic half-parts of hUT-B1 in the urea transport function.
Our results therefore, indicate that functional assays of UT and their modulation by pharmacological agents should be considered with great caution when analyzed in heterologous expression systems that may not reflect the true physiological properties exhibited in physiological tissues. Altogether, these investigations provide new molecular insights in the membrane topology of hUT-B1 as well as structure-function relationships regarding hUT-B1 and presumably other members of the UTfamily, which might lead to the design of novel diuretic molecules.