Characterization of the Gene Encoding the Human Kidd Blood Group/Urea Transporter Protein

The Kidd (JK) blood group is carried by an integral membrane glycoprotein which transports urea through the red cell membrane and is also present on endothelial cells of the vasa recta in the kidney. The exon-intron structure of the human blood group Kidd/urea transporter gene has been determined. It is organized into 11 exons distributed over 30 kilobase pairs. The mature protein is encoded by exons 4–11. The transcription initiation site was identified by 5′-rapid amplification of cDNA ends-polymerase chain reaction at 335 base pairs upstream of the translation start point located in exon 4. The 5′-flanking region, from nucleotide −837 to −336, contains TATA and inverted CAAT boxes as well as GATA-1/SP1 erythroid-specificcis-acting regulatory elements. Analysis of the 3′-untranslated region reveals that the two equally abundant erythroid transcripts of 4.4 and 2.0 kilobase pairs arise from usage of different alternative polyadenylation signals. No obvious abnormality of the Kidd/urea transporter gene, including the 5′- and 3′-untranslated regions, has been detected by Southern blot analysis of the blood of two unrelated Jknull individuals (B.S. and L.P.), which lacks all Jk antigens and Jk proteins on red cells, but was genotyped as homozygous for a “silent”Jk b allele. Further analysis indicated that different splice site mutations occurred in each variant. The first mutation affected the invariant G residue of the 3′-acceptor splice site of intron 5 (variant B.S.), while the second mutation affected the invariant G residue of the 5′-donor splice site of intron 7 (variant L.P.). These mutations caused the skipping of exon 6 and 7, respectively, as seen by sequence analysis of the Jk transcripts present in reticulocytes. Expression studies in Xenopusoocytes demonstrated that the truncated proteins encoded by the spliced transcripts did not mediate a facilitated urea transport compared with the wild type Kidd/urea transporter protein and were not expressed on the oocyte’s plasma membrane. These findings provide a rational explanation for the lack of Kidd/urea transporter protein and defect in urea transport of Jknull cells.

The Kidd blood group system (JK) is defined by two codominant alleles, Jk a and Jk b , of similar frequency (0.51 and 0.49, respectively) in the Caucasian population, but showing large differences in other ethnic groups (1,2). Alloantibodies against the Jk antigens may occasionally be involved in severe transfusion incompatibilities and newborn hemolytic disease. There are three common phenotypes Jk(aϩbϪ), Jk(aϪbϩ), and Jk(aϩbϩ) and a rare null phenotype, Jk(aϪbϪ), first described by Pinkerton et al. (3), also called Jk null . The frequency of this phenotype is increased in certain populations (Asian, Polynesian, or Indian extraction). The Jk null phenotype results from two different genetic backgrounds: (i) homozygous inheritance of a "silent" allele Jk at the JK locus and (ii) inheritance of a dominant inhibitor gene In(Jk), unlinked to the JK locus (1,2). Following immunization by transfusion or pregnancy, Jk null individuals may produce an antibody called anti-Jk 3 (or anti-Jk ab ), which reacts with all common red cells carrying the Jk a and/or Jk b antigens, but is unreactive with Jk null cells themselves.
The discovery that red cells from Jk null individuals exhibited an increased resistance to lysis in aqueous 2 M urea (4) led to the suspicion that the Jk antigens might be related to the urea transporter of the human erythrocytes (for a review, see Moulds (5)). This prediction was fully confirmed by molecular cloning of the human erythroid urea transporter (clone HUT11) (6), by cross-hybridization with a rabbit cDNA transporter (7), and the demonstration that the HUT11 urea transporter and the Kidd blood group are carried by the same protein (8). This is based on the following findings: (i) in coupled transcriptiontranslation assays, the HUT11 cDNA directed the synthesis of a 36-kDa protein, which was immunoprecipitated by a human anti-Jk 3 antibody; (ii) the anti-Jk 3 immunoprecipitated also a protein material of similar mass from all red cell membranes (after N-glycanase treatment), except those from Jk null cells; (iii) a rabbit antibody against the HUT11-protein reacted on immunoblots with all human erythrocytes except those from Jk null ; and (iv) the structural gene encoding HUT11 was assigned to chromosome 18q12-q21 by in situ hybridization, like the Kidd blood group locus. More recently, the Kidd blood group Jk a /Jk b polymorphism (D280N) was determined and used to demonstrate its lack of association with type 1 diabetes mellitus (9).
The Kidd/HUT11 polypeptide is expressed on human red cells as well as on the endothelial cells of vasa recta in the inner and outer medulla of the kidney (10,11). Rapid urea transport may help to preserve the osmotic stability and deformability of the red cells and to stabilize osmotic gradients in the renal medulla (12,13). In the kidney, this transport system contributes to the urinary concentrating mechanism (14,15) involved in water preservation. Recently, a new urea transporter specific for the human kidney (clone HUT2) was cloned (16) and functionally compared with the erythroid transporter (17).
These transporters are analogous to those found in other species, and their characterization, probably as a new family of transporters, will contribute to the clarification of their critical role in the renal water conservation mechanism (see Hediger et al. (18) for review) (19,20). In this report we have further characterized the gene encoding the erythroid Kidd/HUT11 polypeptide, and we have analyzed the molecular basis of the Jk null phenotype from two unrelated individuals.

MATERIALS AND METHODS
Blood Samples and Reagents-Blood samples from individuals of common Jk phenotypes came from the Institut National de la Transfusion Sanguine (Paris, France). Two Jk null blood samples were investigated; donor L.P. was from the Center National de Référence sur les Groupes Sanguins in Paris, and donor B.S. was from the Irwin Memorial Blood Center (San Francisco, CA). Restriction endonucleases, modifying enzymes, pUC vectors, and N-glycosidase F (PNGase-F purified from Flavobacterium meningosepticum) came from New England Biolabs (UK). Radiolabeled nucleotides, [ 14 C]urea (1.96 GBq/mmol), and [ 3 H]raffinose (188.7 GBq/mmol) were purchased from Amersham (Bucks, UK). Expand High Fidelity and Expand Long Template PCR 1 systems from Boehringer Mannheim (Germany) were used for PCR amplification. Nucleotide sequences were determined on both strands by the dideoxy chain termination method (Sanger) with Sequenase version 2.0 (U. S Biochemical Corp., Cleveland, OH) or with Thermo-Sequenase fluorescent labeled primer cycle sequencing kit from Amersham (Bucks, UK) using 5Ј(Cy5)-primers (Genset, France).
Isolation of Human JK Gene-Approximately 2.5 ϫ 10 6 phages from a human leukocyte genomic library constructed in EMBL3 Sp6/T7 (CLONTECH Laboratories, Inc., Palo Alto, CA) were plated and hybridized under standard procedures with a 32 P-labeled full-length HUT11 cDNA probe (1-1528) using the random primed DNA labeling kit (Boehringer Mannheim, Germany). The positive clones were initially mapped by SacI digestion followed by Southern blot hybridization using HUT11 and the 5Ј-end region (see below) as probes. The positive individual gene fragments were subcloned into pUC vector and sequenced. The exons, identified from the HUT11 cDNA sequence, and their flanking regions, were fully sequenced as were some short introns. The introns were sized by PCR using primers from flanking regions of the intron. At least, all sizes were confirmed by PCR using human genomic DNA as template and Expand Long Template PCR system. 5Ј-and 3Ј-End Region Determination by cDNA Cloning-The 5Ј-UT sequence was cloned by 5Ј-rapid amplification of the cDNA ends (RACE) (16), using the human fetal liver 5Ј-RACE Ready cDNA kit with Advantage KlenTaq Polymerase Mix (CLONTECH). Briefly, a heminested PCR amplification under stringent conditions (94°C for 45 s, 60°C for 45 s, 72°C for 1 min, for 30 cycles) was carried out between the primer complementary to 5Ј-end anchor and two antisense primers from the 5Ј-coding sequence of HUT11 cDNA, AS-1 (antisense primer, position 109 -86) and AS-2 (antisense primer, position 44 -21). After agarose gel analysis, transfer, and Southern blot hybridization using 32 P-labeled probe Pr.2* (nt Ϫ194 to Ϫ174), 5Ј-CTACCTAAAATAAA-GATTATA-3Ј, deduced from the 5Ј-end of an unpublished human bone marrow cDNA clone, the positive bands were subcloned and sequenced. Similarly, the 3Ј-UT sequence was amplified using human bone marrow Marathon-Ready cDNA (CLONTECH) between the primers complementary to 3Ј-end adaptor and two sense primers from the 3Ј-coding sequence of HUT11 cDNA, SP-1 (sense primer, position 1219 -1242) and SP-2 (sense primer, position 1304 -1327. For primer designation, nt ϩ1 was taken as the first nucleotide of the HUT11 initiation codon (6).
Southern Blot Analysis-Total genomic DNA from B lymphoid cell lines (Epstein-Barr virus-transformed) or from leukocytes was digested with SacI restriction enzyme according to the supplier (10 units/g of DNA), resolved by electrophoresis in 0.8% (w/v) agarose gel and transferred to a nylon membrane (Hybond N ϩ , Amersham, UK). The blot was prehybridized in 0.25 M Na 2 HPO 4 (pH 7.2), 7% SDS for 5 min at 65°C and then hybridized using a full-length 32 P-labeled cDNA probe (exon 1-5Ј to end exon 11) in the same buffer. Hybridization was carried out overnight at 65°C, and the last washing was in 0.02 M Na 2 HPO 4 (pH 7.2), 3% SDS for 20 min at 65°C.
Amplification by Reverse Transcription-PCR of Jk cDNAs-Five micrograms of total reticulocyte RNA extracted by the acid-phenol-ammonium method (22) were used to produce the first cDNA strands using the first strand cDNA synthesis kit (Pharmacia, Uppsala, Sweden). One sixth of the cDNA products was used to perform a hemi-nested PCR (94°C for 30 s, 60°C for 30 s, 72°C for 1 min 15 s, for 30 cycles) in a first step between primers SP-A (sense primer, position Ϫ21 to Ϫ1) and AS-B (antisense primer, position 1260 -1237). The second PCR was performed with 1/25 of the first PCR products in the same conditions, using primers SP-A and AS-C (antisense primer, position 1234 -1211). Final PCR products were identified by Southern blot analysis, subcloned, and sequenced on both strands, using an automated Alf-Express sequencer (Pharmacia, Uppsala, Sweden).
Analysis of Splice Sites-Direct PCR amplification was carried on genomic DNA (100 ng) under stringent conditions (94°C for 30 s, 60°C for 30 s, 72°C for 30 s, for 30 cycles) between primers designed from intronic sequences flanking each exon. The PCR products were subcloned and sequenced. To avoid PCR artifacts, sequencing was performed on both strands using independent PCR reactions.
Transcription-Translation and Immunoprecipitation Assays-Fulllength (Jk ϩ ) and spliceoforms (Jk Ϫ (⌬6) and Jk Ϫ (⌬7)) cDNAs were subcloned into the EcoRV-digested pT7TS plasmid (kindly provided by P. Krieg, Austin, TX) and placed under the control of the T7 promoter. The corresponding proteins were synthesized in vitro in the transcriptiontranslation coupled reticulocyte lysate kit from Promega (Madison, WI) in the presence of L-[ 35 S]methionine (1.85 Gbq/mmol, Amersham, Bucks, UK) and immunoprecipitated with the human anti-Jk 3 antiserum obtained from an immunized Jk(aϪbϪ) individual and with an affinity-purified polyclonal antibody raised against the N-terminal region (residues 8 -22) of the Kidd/urea transporter polypeptide (anti-HUT11), as described previously (8). The immunoprecipitates were analyzed by SDS-polyacrylamide gel electrophoresis (15% separating gel) on a discontinuous buffer system (23), followed by enlightening treatment (NEN Life Science Products) and autoradiography.
Oocyte Urea Flux Measurements and Immunoblotting Analysis-After linearization of the pT7TS-cDNA constructs with SmaI restriction enzymes, capped sense RNAs were synthesized using T7 RNA polymerase from the mCAP mRNA capping kit (Stratagene, La Jolla, CA). Expression studies were carried out by microinjection of cRNAs (40 ng/oocyte) in collagenase-treated Xenopus oocytes (24) and functional tests were realized 3 days after injection as described previously (6). To determine the expression of Jk ϩ , Jk Ϫ (⌬6), and Jk Ϫ (⌬7) isoforms, fractions enriched for oocyte plasma membranes were prepared from 25 oocytes, as described by Wall and Patel (25). N-Glycosidase F treatments were performed according to the manufacturer, and a control reaction was carried out in enzyme-free buffer in otherwise identical conditions. Untreated and N-glycosidase F-treated plasma membranes, equivalent to six oocytes, were separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and incubated with the affinity-purified antibody anti-HUT11 (2 g/ml). Specifically bound antibodies were detected with affinity-purified goat antirabbit IgG conjugated to horseradish peroxidase (Sigma) (1:15,000 dilution) and using Luminol/Enhancer (Pierce) according to the manufacturer's protocole.

RESULTS
Structure of the JK Gene-Of the seven positive clones isolated from partially Sau3A-digested human leukocyte genomic DNA library, three Jk2, Jk10, and Jk12 were used to characterize the exon organization and to define the intron/exon junction sequences (Fig. 1). SacI restriction mapping was established by Southern blot hybridization with the HUT11 cDNA and the 5Ј-and 3Ј-end regions (see below) as probes. Sequence analysis of the SacI fragments from the three genomic clones revealed that the human JK gene was composed of 11 exons distributed over 30 kb of DNA (Fig. 1A). The exon sizes ranged from 50 bp (exon 10) to 551 bp (refer to the first used polyadenylation signal). The three first exons include the 5Ј-UT part of the gene (see below) and the remaining eight exons the open reading frame. The translation initiation codon was located in exon 4 and the stop codon in exon 11. Thus, exons 4 -11 corresponded to amino acids 1-50, 51-113, 114 -156, 157-221, 222-270, 271-315, 316 -332, and 333-389, respectively (Fig. 1B). The intron sizes determined by PCR, ranged from 0.6 kb to approximately 8.6 kb. After partial sequence of the introns, all exon/intron junctions were found to contain the canonical 5Ј-donor gt and the 3Ј-acceptor ag sequences (Fig. 1B). Each class of intron-exon boundary was found in the JK gene. When all exons were sequenced and compared with the published HUT11 cDNA isolated from a human bone marrow library, two differences, one amino acid change (K44E) and one dipeptide deletion were noted in exon 4 and 8, respectively. These changes correspond to nonallelic differences previously reported between HUT11 and the Jk a / Jk b alleles (9).
Characterization of the 5Ј-End of the Jk Transcript-The 5Ј-end of the Jk transcript was cloned by a modified rapid amplification of cDNA ends technique, using a fetal liver cDNA as matrix, and hemi-nested PCR between the 5Ј-end anchor (sens primer, 49 bp) and the AS-1 and AS-2 antisense primers (see "Materials and Methods"). When the amplification products were analyzed by Southern blot hybridization with the probe Pr.2* (nt Ϫ194 to Ϫ174), two bands of ϳ270 and ϳ430 bp were detected, indicating the presence of an alternative splicing event (Fig. 2). These bands, which corresponded to the first ϳ220 and ϳ380 bp, respectively, at the 5Ј end of the primer AS-2 (after deduction of the 49 bp from the 5Ј-anchor primer), were subcloned, and five independent clones were sequenced. The results indicated that the ϳ430-bp fragment included the full-length transcript and located the transcription initiation site at 335 nt upstream from the translation initiation codon (Fig. 2). The ϳ270-bp fragment, however, which was generated by alternative splicing lacked 157 nt corresponding to exon 3 ( Fig. 2). This was confirmed by Southern blot with an exon 3-specific probe which hybridized with the ϳ430-bp but not the ϳ270-bp fragments (not shown). All of this information was used to deduce the structure organization of the three first exons of the JK gene described above. Further studies (reverse transcription-PCR) of the transcripts isolated from normal human reticulocytes indicated the presence of alternatively spliced transcripts lacking sequences encoded by exons 8 and 9, and exons 7-9 (data not shown).
Next, 500 bp upstream of the erythroid transcription initiation site from the Jk2 genomic clone was sequenced and analyzed to identify putative cis-acting regulatory elements. This region (nt Ϫ837 to Ϫ336) contains the consensus motifs AP-3, NF-ATp, and GATA-1 in reverse orientation at position Ϫ529/Ϫ522, Ϫ586/Ϫ581, and Ϫ806/Ϫ801, respectively. Sequence deviating by one nucleotide from the consensus motifs Ets-1, AP-2, reverse Ets-1, CTCF, and a reverse SP-1 were identified between positions Ϫ587 and Ϫ794 (Fig. 2). A typical TATA box was also present at nt Ϫ362/Ϫ358, which is in the expected range of 25 to 30 nucleotides upstream of the erythroid transcription initiation site (nt Ϫ335). Although two inverted CAAT boxes (ATTG) were identified at nt Ϫ95 and Ϫ220, only the closest of transcription initiation site might be functional.
Characterization of the 3Ј-UT Region of the Jk Transcripts-The transcription start site of the JK gene is short (335 bp upstream of the translation initiation site) and cannot account for the size difference between the two major transcripts (ϳ2.0 and ϳ4.4 kb) detected in erythroid tissues by Northern blot (16). Therefore, we examined whether these differences arose from the 3Ј end of the Jk transcripts. At first, two unconsensus polyadenylation signals (AATTAAA and TATAAA) and four consensus sites (AATAAA) have been located by sequence analysis of 3Ј-RACE-PCR amplifications performed with reticulocyte transcripts (see "Materials and Methods") and by sequence analysis of the 3Ј-end from the Jk12 genomic clone (Fig. 3). Then, Northern blot analysis using the HUT11 probe as well as probes p1, p2, and p3 located in the 3Ј-UT region of the JK gene clearly indicated that the two erythroid transcripts of 4.4 and 2.0 kb arose from differential usage of two distinct polyadenylation signals which are separated by 2.0 kb. Indeed, the HUT11 probe located 5Ј to the proximal signal AATTAAA hybridized both with the 4.4-and 2.2-kb transcripts, while probes p1, p2, and p3, located downstream this signal hybridize only with the 4.0-kb species (Fig. 3). The colinearity of the full 3Ј-end region (exon 11) was confirmed by hemi-nested reverse transcription-PCR and hybridization of the 2.7-kb reaction product with the HUT11 probe (see Fig. 3).
The JK Gene Is Not Rearranged in Jk null Cells-In preliminary investigations, we confirmed by Western blot analysis with the affinity-purified anti-HUT11 antibody (nonglycosylation dependent) (10), that the Jk null cells under study (B.S. and L.P.) lack the Jk erythrocyte membrane protein (45-69 kDa), as seen in Fig. 4A, as well as the Jk a and Jk b antigens (data not shown). Next, we demonstrated by Southern blot hybridization of genomic DNA an identical SacI digestion pattern of the Jk null samples as compared with common Jk(aϩbϪ) and Jk(aϪbϩ) phenotypes (Fig. 4B), indicating that a gross rearrangement of JK gene did not occur in these variants. The five detected bands (ranging from 2 to 7 kb) are fully concordant with the known restriction sites in the JK gene (see Fig. 1) and altogether contain all 11 exons (Fig. 4A). Moreover, restriction analysis with a number of other enzymes showed a simple pattern consistent with a single copy gene (data not shown). Further analysis by DNA genotyping (9) indicated that both Jk null individuals were homozygous for a silent Jk b allele (data not shown).
Molecular Analysis of the Jk null Mutations-To determine the molecular basis of the Jk null phenotypes, total reticulocyte RNA from donors B.S. and L.P. were prepared and used to amplify full-length Jk transcripts. Three amplification products of different size (not shown) were obtained from the B.S. sample, and several independent clones of each size were se-   kb were obtained by hemi-nested PCR with primers p4 (exon 11), p5, and p6 (3Ј end). Horizontal arrows localize the HUT11 cDNA and the PCR primers used to prepare probes p1, p2, and p3, which were hybridized to the Northern blots shown below. Poly(A ϩ ) RNA (2 g/lane) from human fetal liver was separated by agarose gel electrophoresis and hybridized with the indicated 32 P-labeled PCR probes as described under "Materials and Methods." Autoradiography was for 16 h at Ϫ80°C. Size markers (kb) are indicated on the left. Southern blot of the full 3Ј-end reverse transcription-PCR product detected with the HUT11 probe is shown on the right. quenced. We identified spliceoforms that resulted from the alternative splicing of exon 6 (Jk Ϫ (⌬6)), exons 6, 8, and 9 (Jk Ϫ (⌬6, 8, and 9)) and exons 6 -9 (Jk Ϫ (⌬6 -9)). The nucleotide sequence of the 5/7 exon junction found in the Jk Ϫ (⌬6) transcript is shown in Fig. 5. All of these spliced transcripts corresponded to the spliceoforms found in normal Jk ϩ individuals (see above), which in addition, all lacked exon 6 sequences. The Jk Ϫ (⌬6) clone carried the nt 838A polymorphism typical of the Jk b allele (9), which confirmed the results from the DNA genotyping (see above). No other alteration of the nucleotide sequence was detected. Since exon 6 was missing in all transcripts, the intron/exon junctions surrounding this exon were analyzed on the genomic DNA prepared from donor B.S. We found that the skipping of exon 6 was most likely caused by a G 3 A transition that affected the invariant G residue of the 3Ј-acceptor splice site of intron 5 (Fig. 4C). The transcripts from B.S. potentially encode truncated polypeptides; the predicted Jk Ϫ (⌬6) polypeptide would lack amino acid residues 114 -156 (encoded by exon 6) of the third and fourth transmembrane domains of the Kidd/urea transporter. The other spliceoforms (Jk Ϫ (⌬6, 8, and 9) and Jk Ϫ (⌬6 -9)) would encode much smaller peptides with a new C-terminal extension generated by a frameshift and premature termination.
Examination of the second Jk null sample (L.P.) revealed the presence of Jk b transcripts lacking exon 7 sequences (Jk Ϫ (⌬7) and Jk Ϫ (⌬7, 8, and 9)). The nucleotide sequence of the 6/8 exon junction found in the Jk Ϫ (⌬7) spliceoform is shown in Fig. 5. Further studies of the genomic DNA indicated that the skipping of exon 7 was most likely caused by a G 3 T transversion that affected the invariant G residue of the 5Ј-donor splice site of intron 7 (Fig. 4D). The protein isoform encoded by the Jk Ϫ (⌬7) transcript would lack amino acids 157-221 and include new C-ter residues generated by a frameshift and premature termination. In all instances, the protein isoforms potentially encoded by the spliced transcripts identified in B.S and L.P. were not detected on red cells by Western blot analysis with the anti-HUT11 antibody (see above).
Expression and Functional Analysis of Jk Ϫ (⌬6) and Jk Ϫ (⌬7) Protein Isoforms-To understand why Jk null red cells lack Jk polypeptides and exhibit a defective urea transport activity, we examined whether the Jk Ϫ (⌬6) and Jk Ϫ (⌬7) transcripts characteristic of these cells could be expressed in vitro and in vivo.
In the cell-free transcription-translation coupled system, the plasmids pT7TS-Jk ϩ , -Jk Ϫ (⌬6), and -Jk Ϫ (⌬7) directed the synthesis of 36-, 31-, and 17-kDa protein bands, respectively, which were immunoprecipitated with the affinity-purified anti-HUT11 antibody and with the human anti-Jk 3 antibody (Fig.  6). No radioactive material in these regions could be immunoprecipitated from transcription-translation of the luciferase peptide control vector.
To test whether these Jk Ϫ (⌬6) and Jk Ϫ (⌬7) proteins were expressed as functional urea transporters, cRNA transcripts were synthesized and injected into Xenopus oocytes. Three days later, we found that the [ 14 C]urea uptake of Jk Ϫ (⌬6) and Jk Ϫ (⌬7) cRNA-injected oocytes was not different from that of water-injected control oocytes, whereas the [ 14 C]urea uptake of Jk ϩ cRNA-injected oocytes reached 38 pmol/oocyte after 3 min (Fig. 7A). Northern blot analysis also revealed that all cRNA injected oocytes showed a stable specific signal between day 0 and 3 after injection (see Fig. 7B). Protein expression in oocyte plasma membranes was also analyzed by Western blot analysis with the affinity-purified anti-HUT11 antibody.
We found that a strongly reactive band of 46 -69 kDa was present in control oocytes expressing the functional Jk ϩ protein, the size of which was reduced to 36 kDa after N-glycosidase F treatment (Fig. 7C). However, using oocytes injected with the Jk Ϫ (⌬6) and Jk Ϫ (⌬7) cRNAs, no signal could be detected by Western blotting, neither in the total cell lysates (not shown) nor in the plasma membrane fraction (Fig. 7C). DISCUSSION The studies reported here define the structural organization of the JK gene which encodes the human Kidd blood group/urea transporter protein. This gene spans 30 kb of DNA and consisted of 11 exons, of which exons 4 -11 contained all the coding information for the mature protein. Exons 4 -11 appear as being distributed along the gene into two groups of two times two exons separated by a large intronic sequence (E4, E5 and E6, E7, then E8, E9 and E10, E11), evocative of an internal gene duplication. Indeed, this may parallel the topology of the Jk polypeptide which can be subdivided into two homologous hydrophobic parts, each carrying a LP box (LPXXTXPF) encoded by exons 7 and 11, respectively, and previously reported to be an internal duplicated signature sequence of urea transporters (26).
The erythroid transcription initiation site of the JK gene was identified 335 bp upstream from the translation initiation site. Examination of nucleotide sequences that are immediately upstream revealed a typical TATA box, one inverted CAAT box at the expected position, and several putative cis-regulatory elements that may bind a variety of transcription factors (27), among which are those involved in erythroid/megakaryocytic expression (28,29). The presence of a potential binding site for a NF-ATp factor which regulates the inducible expression of several cytokine genes is intriguing (30,31). However, functional analysis will be required to determine which elements and which factors bind to this promoter in tissues were the Kidd/urea transporter is expressed.
We have also shown that the 4.4-and 2.0-kb erythroid Kidd/ urea transport mRNAs detected by Northern blot (16) arise from usage of two different polyadenylation signals, indicating that the two equally abundant transcripts encode the same 45-kDa polypeptide. This is at the opposite to UT1 and UT2 urea transport transcripts from rat kidney (4.0 and 2.9 kb, respectively), which are alternative splice products derived from a single gene by differential utilization of alternative 5Ј-exon groups (19). The rat homologue of HUT11, called UT3 (20), is encoded by a single 3.8-kb transcript, which is translated into a protein of 384 amino acids sharing 80% identity with HUT11. It is possible that the first polyadenylation signal used in the human primary transcript is absent or not used in the rat. The role of large 3Ј-UT sequences in some transcripts is not well understood, although some may play a role in regulation of expression (32,33).
We next examined blood from two rare unrelated Jk null individuals, one Caucasian (L.P.) and another of Chinese (B.S.) origin, that lacked Jk antigens and Jk protein expression on red cells. Genomic DNA analysis indicated that both donors were homozygotes for a Jk b allele that exhibited no alteration of the coding sequence. Therefore, although the JK genes were present in these variants, they were not phenotypically expressed. Sequence analysis of reticulocyte transcripts from B.S. and L.P. indicated that alternatively spliced transcripts lacking at least exon 6 and exon 7, respectively, were present. Examination of exon/intron junctions of the JK gene further revealed that B.S. and L.P. were homozygous for point mutations at conserved 3Ј-acceptor (ag 3 aa) and 5Ј donor (gt 3 tt) splice sites of introns 5 and 7, respectively. Splice site muta-tions lead to exon skipping and are well known to abolish or reduce normal splicing (see Maquat (34) and references therein). Since the Kidd/urea transporter protein is absent from Jk null cells, it is likely that the spliced transcripts are either unstable and not translated, or the corresponding truncated proteins are misrouted. At first, we found that the Jk Ϫ (⌬6) and Jk Ϫ (⌬7) transcripts typical of Jk null cells could be translated into polypeptides of 31 and 17 kDa, respectively, as seen by immunoprecipitation with specific antibodies in a cell-free transcription-translation coupled system (Fig. 6). Next, we found that when expressed in Xenopus oocytes these truncated proteins did not mediate a facilitated urea transport, in contrast to the full-length wild type Jk ϩ protein, although all injected cRNAs had a similar stability on a 3 days period, as seen by Northern blot analysis. Further analysis revealed that the plasma membrane fraction from oocytes expressing the functional urea transporter (encoded by the Jk ϩ cRNA) carried a 46 -69-kDa glycoprotein component, which could be deglycosylated into a 36-kDa protein, as expected for the Kidd/urea transporter protein (6,8). On the contrary, the Jk Ϫ (⌬6) and Jk Ϫ (⌬7) polypeptides were neither detected in total cell lysates nor in the enriched plasma membrane fraction from the oocytes, which could be explained by a rapid intracellular degradation. Indeed, the predicted truncated proteins are most likely misfolded by lack the transmembrane domains 3 and 4, and transmembrane domain 5, including the hydrophilic loop carrying the N-glycosylation site at Asn 221 , respectively. Therefore, these findings provide a rationale explanation for the lack of Kidd/urea transporter protein and defect in urea transport of Jk null cells.
As the Kidd/HUT11 transcript is distributed widely in various organs (16), it is surprising that Jk null individuals who have a urea transport deficiency (35) did not suffer a clinical syndrome, except for a reduced capability to concentrate urine (36), as was the unexpected finding that donors of the blood group Colton null phenotype, who lack the water channel aquaporin-1, did not produce a severe or lethal phenotype (37,38). It is postulated that mechanisms which compensate or reduplicate the function of the missing protein may exist. This should stimulate more studies of these transporters in red cell membranes and various organs. In addition, further investigations of other Jk null individuals from different ethnic groups (5), from both physological and fundamental aspects, may provide new information for addressing these issues.