Voltar
Artigo Acesso aberto Revisado por pares
BIBTEX RIS

Association of XK and Kell Blood Group Proteins

1998; Elsevier BV; Volume: 273; Issue: 22 Linguagem: Inglês

10.1074/jbc.273.22.13950

ISSN

1083-351X

Autores

David Russo, Colvin M. Redman, Soohee Lee,

Tópico(s)

Pancreatic function and diabetes

Resumo

A disulfide bond links Kell and XK red cell membrane proteins. Kell, a type II membrane glycoprotein, carries over 20 blood group antigens, and XK, which spans the membrane 10 times, is lacking in rare individuals with the McLeod syndrome. Kell is classified in the neprilysin family of zinc endopeptidases, and XK has structural features that suggest it is a transport protein. Kell has 15 extracellular cysteines, and XK has one in its fifth extracellular loop. Five of the extracellular cysteine residues in Kell are not conserved in the other members of the neprilysin family, and based on the hypothesis that one of the nonconserved cysteines is linked to XK, cysteines 72 and 319 were mutated to serine. The single extracellular cysteine 347 of XK was also mutated. Co-expression of combinations of wild-type and mutant proteins in transfected COS-1 cells showed that Kell C72S did not form a Kell-XK complex with wild-type XK, while wild-type Kell and Kell C319S did. XK C347S was also unable to form a complex with wild-type Kell, indicating that Kell cysteine 72 is linked to XK cysteine 347. Kell C72S was transported to the cell surface, indicating that linkage to XK is not required. In addition, chemical cross-linking of red cell membranes with dithiobispropionimidate indicated that glyceraldehyde-3-phosphate dehydrogenase is a near neighbor of Kell. A disulfide bond links Kell and XK red cell membrane proteins. Kell, a type II membrane glycoprotein, carries over 20 blood group antigens, and XK, which spans the membrane 10 times, is lacking in rare individuals with the McLeod syndrome. Kell is classified in the neprilysin family of zinc endopeptidases, and XK has structural features that suggest it is a transport protein. Kell has 15 extracellular cysteines, and XK has one in its fifth extracellular loop. Five of the extracellular cysteine residues in Kell are not conserved in the other members of the neprilysin family, and based on the hypothesis that one of the nonconserved cysteines is linked to XK, cysteines 72 and 319 were mutated to serine. The single extracellular cysteine 347 of XK was also mutated. Co-expression of combinations of wild-type and mutant proteins in transfected COS-1 cells showed that Kell C72S did not form a Kell-XK complex with wild-type XK, while wild-type Kell and Kell C319S did. XK C347S was also unable to form a complex with wild-type Kell, indicating that Kell cysteine 72 is linked to XK cysteine 347. Kell C72S was transported to the cell surface, indicating that linkage to XK is not required. In addition, chemical cross-linking of red cell membranes with dithiobispropionimidate indicated that glyceraldehyde-3-phosphate dehydrogenase is a near neighbor of Kell. The association of Kell and XK proteins on red cell membranes was predicted from early serological studies on two rare Kell-related phenotypes that noticed a relationship between antigens residing on the Kell blood group protein and Kx antigen, which is carried on a separate XK protein. In the McLeod phenotype red cell, there is an absence of Kx, an otherwise ubiquitous red cell antigen, and this is accompanied with a depression of all Kell antigens. On the other hand, in the Ko (null) phenotype, in which there is no detectable Kell surface antigens, there is an enhanced level of Kx antigens (1Redman C.M. Marsh W.L. Semin. Hematol. 1993; 30: 209-218PubMed Google Scholar, 2Marsh W.L. Redman C.M. Transfusion. 1990; 30: 158-167Crossref PubMed Scopus (73) Google Scholar). Also, treatment of normal red cells with reducing reagents, which inactivate the Kell antigens, showed a marked increase in Kx activity, indicating a sulfhydryl involvement in the presentation of Kell and Kx antigens on the red cells (3Branch D.R. Sy Siok Hian A.L. Petz L.D. Br. J. Haematol. 1985; 59: 505-512Crossref PubMed Scopus (19) Google Scholar). Biochemical studies in which Kell protein was isolated from red cells in nonreduced conditions, showed that Kell protein was associated with itself as an oligomer and was also complexed with other red cell membrane proteins (4Redman C.M. Avellino G. Pfeffer S.R. Mukherjee T.K. Nichols M. Rubinstein P. Marsh W.L. J. Biol. Chem. 1986; 261: 9521-9525Abstract Full Text PDF PubMed Google Scholar). Later, immunoprecipitation of Kell protein in nonreduced conditions co-isolated XK, demonstrating that the two proteins exist as a disulfide-bonded complex on the red cell membrane (5Khamlichi S. Bailly P. Blanchard D. Goossens D. Cartron J.P. Bertrand O. Eur. J. Biochem. 1995; 228: 931-934Crossref PubMed Scopus (80) Google Scholar). Further studies on the Kell-XK complex showed its absence in McLeod red cells and also demonstrated that although Ko (null) red cells have high Kx activity they contain less XK protein, indicating that lack of Kell protein may expose more Kx antigens on the cell surface (6Carbonnet F. Hattab C. Collec E. Le Van Kim C. Cartron J.P. Bertrand O. Br. J. Haematol. 1997; 96: 857-863Crossref PubMed Scopus (40) Google Scholar). The McLeod phenotype has an X-linked mode of inheritance, and since some McLeod patients also have chronic granulomatous disease and/or Duchenne muscular dystrophy, XK was located close to those genes at the Xp21 region of the X chromosome (7Bertelson C.J. Pogo A.O. Chaudhuri A. Marsh W.L. Redman C.M. Banerjee D. Symmans W.A. Simon T. Frey D. Kunkel L.M. Am. J. Hum. Genet. 1988; 42: 703-711PubMed Google Scholar). The red cell membrane protein (XK) that carries Kx antigen migrates on SDS-PAGE as a 37-kDA polypeptide (8Redman C.M. Marsh W.L. Scarborough A. Johnson C.L. Rabin B.I. Overbeeke M. Br. J. Haematol. 1988; 68: 131-136Crossref PubMed Scopus (48) Google Scholar). Positional cloning isolated the XK gene and predicted that it encodes a 444-amino acid protein that spans the red cell membrane 10 times and has structural characteristics of a membrane transporter (9Ho M. Chelly J. Carter N. Danek A. Crocker P. Monaco A.P. Cell. 1994; 77: 869-880Abstract Full Text PDF PubMed Scopus (238) Google Scholar). Since XK protein may only have one extracellular cysteine residue in the fifth extracellular loop, it was predicted that this residue may be involved in disulfide linkage with Kell protein, which has 15 extracellular cysteine residues and one cysteine in the transmembrane domain (5Khamlichi S. Bailly P. Blanchard D. Goossens D. Cartron J.P. Bertrand O. Eur. J. Biochem. 1995; 228: 931-934Crossref PubMed Scopus (80) Google Scholar, 10Lee S. Vox Sang. 1997; 73: 1-11Crossref PubMed Scopus (103) Google Scholar). McLeod red cells, which lack XK protein, have abnormal cell shape with a high proportion of acanthocytes. Another striking aspect of the McLeod phenotype is its association with a late onset form of muscular dystrophy, although high serum levels of the muscle enzyme, creatine phosphokinase are noted at earlier ages (11Marsh W.L. Marsh N.J. Moore A. Symmans W.A. Johnson C.L. Redman C.M. Vox Sang. 1981; 40: 403-411Crossref PubMed Scopus (53) Google Scholar). Some McLeod patients also develop neurological symptoms (1Redman C.M. Marsh W.L. Semin. Hematol. 1993; 30: 209-218PubMed Google Scholar, 2Marsh W.L. Redman C.M. Transfusion. 1990; 30: 158-167Crossref PubMed Scopus (73) Google Scholar, 12Marsh W.L. Redman C.M. Transfus. Med. Rev. 1987; 1: 4-20Crossref PubMed Scopus (70) Google Scholar). In accordance with the clinical studies, XK transcripts are present not only in erythroid tissues but also in skeletal muscle, brain, heart, and pancreas (9Ho M. Chelly J. Carter N. Danek A. Crocker P. Monaco A.P. Cell. 1994; 77: 869-880Abstract Full Text PDF PubMed Scopus (238) Google Scholar). This is in contrast to KEL, which appears to be mostly restricted to erythroid tissues (13Lee S. Zambas E.D. Marsh W.L. Redman C.M. Blood. 1993; 81: 2804-2809Crossref PubMed Google Scholar). Molecular cloning ofKEL (14Lee S. Zambas E.D. Marsh W.L. Redman C.M. Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 6353-6357Crossref PubMed Scopus (206) Google Scholar) showed that the Kell protein is a 731-amino acid type II membrane glycoprotein with close structural and sequence similarities to a subfamily of zinc endopeptidases that includes neutral endopeptidase 24.11, endothelin-converting enzyme, and the product of the PEX gene (10Lee S. Vox Sang. 1997; 73: 1-11Crossref PubMed Scopus (103) Google Scholar, 15Turner A.J. Tanzawa K. FASEB J. 1997; 11: 355-364Crossref PubMed Scopus (385) Google Scholar, 16Xu D. Emoto N. Giaid A. Slaughter C. Kaw S. deWit D. Yanagisawa M. Cell. 1994; 78: 473-485Abstract Full Text PDF PubMed Scopus (863) Google Scholar, 17Shimada K. Takahashi M. Tanzawa K. J. Biol. Chem. 1994; 269: 18275-18278Abstract Full Text PDF PubMed Google Scholar, 18Valdenaire O. Rohrbacher E. Mattei M.-G. J. Biol. Chem. 1995; 270: 29794-29798Abstract Full Text Full Text PDF PubMed Scopus (208) Google Scholar, 19Rowe P.S. Oudet C.L. Francis F. Sinding C. Pannetier S. Econs M.J. Strom T.M. Meitinger T. Garabedian M. David A. Macher M.A. Questiaux E. Popowska E. Pronicka E. Read A.P. Mokrzycki A. Glorieux F.H. Drezner M.K. Hanauer A. Lehrach H. Goulding J.N. O'Riordan J.L. Hum. Mol. Genet. 1997; 6: 539-549Crossref PubMed Scopus (173) Google Scholar). In this study, the association of Kell and XK on red cell membranes is explored with emphasis on the topology of XK, on the near neighbor relationship of Kell to other red cell membrane proteins, and on the cysteine residues that link Kell and XK proteins. A 42-amino acid synthetic peptide, corresponding to the second extracellular loop domain of XK, was prepared by Synpep (Deblin, CA). A 30-mer peptide derived from an intracellular domain of Kell (amino acids 2–31) was synthesized by the Microchemistry Laboratory of the New York Blood Center. Polyclonal antibodies were raised in rabbits and were affinity-purified. The antibody to the XK peptide reacted with several red cell membrane proteins on Western immunoblots, but it isolated XK from a nonionic detergent extract from normal, but not from McLeod, red cells. The antibody to the Kell peptide was specific both by Western immunoblotting and the ability to isolate Kell from a red cell detergent extract. Polyclonal antibody to human glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 1The abbreviations used are: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PAGE, polyacrylamide electrophoresis; PCR, polymerase chain reaction; COS-1, monkey kidney fibroblast cells; CMV, cytomegalovirus; HPLC, high pressure liquid chromatography; kb, kilobase pair; bp, base pair. purchased from Sigma, was raised in mice. A mouse monoclonal antibody to a Kell surface antigen (KEL14) was a gift from Dr. Pablo Rubinstein of the New York Blood Center. Surface-exposed Kell was isolated from red cells using a monoclonal antibody to KEL14. Red cells were coated overnight with 5 volumes of 300 μg/ml anti-KEL14 at 4 °C. The cells were washed with phosphate-buffered saline, and membrane "ghosts" were prepared as described previously (4Redman C.M. Avellino G. Pfeffer S.R. Mukherjee T.K. Nichols M. Rubinstein P. Marsh W.L. J. Biol. Chem. 1986; 261: 9521-9525Abstract Full Text PDF PubMed Google Scholar). The membrane were solubilized with 1% Triton X-100, 0.5% sodium deoxycholate in phosphate-buffered saline containing protease inhibitors, and the solution was cleared by centrifugation at 27,000 rpm for 20 min in a Beckman 50Ti rotor. The following protease inhibitors were used: 0.1 mml-1-tosylamido-2-phenylethyl chloromethyl ketone, 0.1 mm phenylmethylsulfonyl fluoride, and 10 units/ml aprotinin (Sigma). The Kell-antibody complex was isolated with protein G coupled to Sepharose (Pierce) as described previously (4Redman C.M. Avellino G. Pfeffer S.R. Mukherjee T.K. Nichols M. Rubinstein P. Marsh W.L. J. Biol. Chem. 1986; 261: 9521-9525Abstract Full Text PDF PubMed Google Scholar, 20Redman C.M. Marsh W.L. Mueller K.A. Avellino G.P. Johnson C.L. Transfusion. 1984; 24: 176-178Crossref PubMed Scopus (41) Google Scholar). XK and XK-Kell complexes were also isolated from detergent-soluble extracts of red cell membranes using rabbit antibody to a peptide representing the second loop of XK. Membrane ghosts were prepared and solubilized as described above and treated overnight at 4 °C with 3.5 μg/ml affinity-purified antibody. The immune complex was isolated with protein A-Sepharose. (4Redman C.M. Avellino G. Pfeffer S.R. Mukherjee T.K. Nichols M. Rubinstein P. Marsh W.L. J. Biol. Chem. 1986; 261: 9521-9525Abstract Full Text PDF PubMed Google Scholar, 20Redman C.M. Marsh W.L. Mueller K.A. Avellino G.P. Johnson C.L. Transfusion. 1984; 24: 176-178Crossref PubMed Scopus (41) Google Scholar) Confluent, transiently transfected, COS cells were incubated withl-[35S]methionine (NEN Life Science Products; specific activity, 1110 Ci/mmol) in 100-mm Petri dishes. After removal of the incubation medium, the cells were lysed with 1 ml of 0.5% sodium deoxycholate, 1%n-dodecyl-β-d-maltoside (Sigma), and a mixture of protease inhibitors described above. The extract was centrifuged in a microcentrifuge at 12,000 rpm for 10 min at 4 °C. Antibody to the second loop of XK (3.5 μg/ml) or to the COOH-terminal peptide of Kell (3.5 μg/ml) was added and incubated overnight at 4 °C. The immune complex was isolated with protein A-Sepharose. To isolate surface-exposed Kell and its complexes from transfected COS cells, the cells were labeled overnight withl-[35S]methionine, and the cells, attached to the bottom of the 100-mm Petri dishes, were washed with phosphate-buffered saline and then treated for 30 min over ice with 3 ml of 300 μg/ml mouse monoclonal antibody to KEL14. The cells were again washed with phosphate-buffered saline and lysed with detergent as described above. The immune complexes present in the detergent-soluble extract were isolated with protein G-Sepharose. The immune complexes on protein A or protein G-Sepharose were eluted with SDS-loading buffer (0.25 m Tris-HCl, pH 6.8, 1% SDS, 5% glycerol or 4 m urea with or without 10 mm dithiothreitol or 300 mm 2-mercaptoethanol, 0.005% bromphenol blue) and separated by SDS-PAGE using 9% polyacrylamide gels. Red cells were coated with monoclonal antibody to KEL14, washed with normal saline, and lysed with 5 mm phosphate buffer, pH 7.4, and membrane "ghosts" were prepared as described previously (4Redman C.M. Avellino G. Pfeffer S.R. Mukherjee T.K. Nichols M. Rubinstein P. Marsh W.L. J. Biol. Chem. 1986; 261: 9521-9525Abstract Full Text PDF PubMed Google Scholar). The washed membranes, suspended in 5 volumes of 25 mmsodium phosphate, pH 8, containing 1 mm MgCl2, were cross-linked at room temperature for 2 h with a final concentration of 4 μg/ml freshly prepared 3,3′-dithiobispropionimidate (DTBP) purchased from Pierce. The membranes were solubilized with 0.5% sodium deoxycholate and 1% Triton X-100 and the proteins, cross-linked to the Kell antibody complex, were isolated with protein G-Sepharose. Since DTBP is a thiol-cleavable cross-linking reagent, the isolated complexes were disassembled with DTT, separated on SDS-PAGE, detected by silver staining, or analyzed by Western immunoblotting using polyclonal antibody to GAPDH or to Kell protein. Proteins, separated by SDS-PAGE, were transferred to ProBlotTM polyvinylidene difluoride membranes (purchased from Applied Biosystems, Foster City, CA) and digested with trypsin (21Fernandez J. Andrews L. Mische S.M. Crabb J.W. Techniques in Protein Chemistry V. Academic Press, Inc., San Diego1994: 215-223Google Scholar). The released peptides were separated by HPLC, and well separated peptides, were sequenced using an automated Applied Biosystems, model 477A, amino acid sequencer. All site-directed mutations were carried out by PCR, changing a single base in the cysteine codon of one of the primer sets to encode serine. The primers containing the mutation site included a restriction enzyme site that, used together with another nearby unique restriction enzyme site in the cDNAs of Kell or XK, were used to replace the respective region of the wild-type DNA. The wild-type cDNAs were placed in the expression vector pRc/CMV (Invitrogen Inc.) and were linearized before being used as template DNA in the PCR procedure. When there was more than a single restriction enzyme site in the construct, limited digestions were performed. Positive clones were selected with the appropriate restriction enzymes or selected by sequencing if it was not possible to select with restriction enzymes. The mutated regions in the plasmids were confirmed by sequencing. An expression vector, pRc/CMV, containing XK and Kell cDNAs, either wild-type or mutant, in tandem, each with its own CMV promoter and polyadenylation signals, was constructed as summarized in Fig.1. Kell cDNA was modified to contain a Kozak sequence (22Kozak M. J. Biol. Chem. 1991; 266: 19867-19870Abstract Full Text PDF PubMed Google Scholar) before the ATG initiation codon as follows. A Kozak sequence (underlined) was placed at the 5′-end of a sense PCR primer, and an antisense primer was modified to contain a BamHI site (also underlined). These primers were used to amplify full-length Kell cDNA.Sense primer:5′­GCCGCCACC¯ATGGAAGGTGGGGACCAA­3′Anti­sense primer:5′­GGGGTGGCATCTTTGGGATCC¯AAGTTACC­3′ The 2.2-kb PCR product was placed in pBC SK(+) vector (Stratagene, La Jolla, CA) using EcoRV and BamHI cloning sites. A 650-bp fragment was released by cutting withHindIII and PpuMI (at nucleotide 753 of Kell cDNA) and ligated to a 7-kb Kell cDNA in pRc/CMV expression vector that had been cut with HindIII andPpuMI. The following denaturation, annealing, and polymerization steps were performed in an automated thermocycler (Minicycler; MJ Research Inc, Watertown, MA). The initial cycle was 94 °C for 3 min, 60 °C for 1 min, and 72 °C for 30 s; in cycles 2 through 30, the conditions were 94 °C for 30 s, 60 °C for 30 s, and 72 °C for 30 s. In the last cycle, the polymerization step at 72 °C was extended to 7 min to complete copies. The PCR procedure described earlier was used to create a G335C mutation in Kell cDNA that encodes serine instead of cysteine at Kell amino acid 72. The PCR product was inserted into wild-type Kell cDNA that had been modified to contain a Kozak sequence (GCCGCCACC) before the ATG initiation codon, as described above, and had been placed in pRc/CMV vector using HindIII and XbaI cloning sites. Wild-type Kell in pRc/CMV was amplified using the following primers to yield a 512-bp product. Anti­sense primer:5′­CTTGAGGGGAACATCAAACTCTGGC­3′ The sense primer contained a G to C mutation that is conveniently cut with BstXI at the mutation site (underlined). The PCR product also contains a PpuMI site that is unique in Kell cDNA. Kell cDNA in pRc/CMV, however, has two BstXI sites (nucleotides 165 and 327). First, pRc/CMV Kell was cut with PpuMI followed by limited digestion withBstXI. A 7.5-kb band that was produced was isolated by 0.8% low melting agarose gel electrophoresis. The 512-bp PCR product was also cut with BstXI and PpuMI, and the isolated 427-bp product was ligated to the 7.5-kb Kell cDNA fragment. The ligation products were transformed into Top10F1 Escherichia coli competent cells (Invitrogen). Miniplasmid preparations were performed using RPMR Miniplasmid preparation kit (BIO 101, Inc., Vista, CA). Positive clones were selected, and plasmid DNA was obtained by the polyethylene glycol procedure. A similar procedure was followed to create the Kell C319S mutant. The following primers were used to create a 224-bp PCR product containing the G1076C mutation. Anti­sense primer:5′­GAGCTTTCTGCGTGCCTCCTGGAATTGAC­3′ An AhdI site in the sense primers is underlined, as is the G to C mutation. Template pRc/CMV-Kell was used, and the PCR product was cut with AhdI and BstEII and purified in 1% low melting agarose electrophoresis. Kell in pRc/CMV was cut with BstEII, followed by limited digestion withAhdI. A 7.6-kb fragment was isolated and ligated with the 176-bp PCR product. Wild-type XK cDNA, placed in pRc/CMV in HindIII and XbaI cloning sites, was linearized by digestion withNruI, prior to its use as a template in the PCR reaction. XK cDNA was a generous gift from Drs. M. Ho and A. P. Monaco (Imperial Cancer Research Fund Laboratories, Institute of Molecular Medicine, John Radcliffe Hospital, Oxford, United Kingdom). The following phosphorylated primers were used to generate a 698-bp product.Sense primer:5′­pCAGAGGAGATTGAGAAGGAGGTGG­3′Anti­sense primer:5′­pGCT¯CACATACATATAGATGTCAGT­3′The antisense primer contains a T1121 Amutation (underlined), which encodes serine instead of cysteine at XK amino acid 347. About 2 × 107 cells were released from Petri plates with trypsin-EDTA, washed twice with phosphate-buffered saline, and suspended for 15 min with 10 μg of the circular plasmid DNA prior to electroporation. Electroporation was performed in 1-ml disposable electrochambers (Life Technologies, Inc.) to which an electrical pulse (250 V, 330 microfarads) was applied in a Life Technologies cell porator. The cells were allowed to recover for 10 min in ice and were then seeded at 5 × 106 cells on 100-mm Petri dishes with 12 ml of growth medium (RPMI 1640 supplemented with 10% fetal bovine serum) and grown to confluence in 2 days. Two days after transfection, confluent COS cells (100-mm plates) were washed with phosphate-buffered saline and incubated at 37 °C for 30 min or overnight in 1 ml of l-methionine-free medium with 0.5 mCi of l-[35S]methionine. The cells were then washed twice with phosphate-buffered saline and "chase" incubated with normal growth media for 5 h. Previous studies showed that Kell and XK are surface-exposed on red cells, and both can be labeled by radioactive iodine (4Redman C.M. Avellino G. Pfeffer S.R. Mukherjee T.K. Nichols M. Rubinstein P. Marsh W.L. J. Biol. Chem. 1986; 261: 9521-9525Abstract Full Text PDF PubMed Google Scholar, 8Redman C.M. Marsh W.L. Scarborough A. Johnson C.L. Rabin B.I. Overbeeke M. Br. J. Haematol. 1988; 68: 131-136Crossref PubMed Scopus (48) Google Scholar). Normal and McLeod red cells were surface-labeled with125I by the lactoperoxidase-catalyzed method (4Redman C.M. Avellino G. Pfeffer S.R. Mukherjee T.K. Nichols M. Rubinstein P. Marsh W.L. J. Biol. Chem. 1986; 261: 9521-9525Abstract Full Text PDF PubMed Google Scholar, 23Reichstein E. Blostein R. J. Biol. Chem. 1974; 250: 6256-6263Abstract Full Text PDF Google Scholar) and lysed, and the washed membranes were solubilized with detergents. Rabbit antibody to a synthetic peptide, which represents the predicted second extracellular loop of XK protein, was used to immunoprecipitate radioactive XK and any other associated surface-labeled proteins. The isolated proteins were separated by SDS-PAGE in reduced conditions. The patterns of total surface-labeled proteins in normal and McLeod red cells were similar. The majority of radioactivity corresponded with the predominant surface-exposed proteins, band 3 and the glycophorins (Fig.2, lanes 1 and 3). Antibody to the second loop of XK isolated radioactive proteins from normal red cells (Fig. 2, lane 4) but not from McLeod red cells that lack this protein (Fig. 2, lane 2). From normal red cells, radioactive proteins were noted at the top of the SDS-PAGE gel and at two other locations that correspond to proteins of approximately 93 and 40 kDa. The electrophoretic migration of the 93- and 40-kDa proteins correspond to the expected sizes of Kell and XK proteins. Failure to isolate the 40-kDa protein from extracts of McLeod red cells and its presence in normal cells indicated that the 40-kDa protein is surface-labeled XK protein. Western immunoblot using a rabbit antibody specific for Kell protein, reacted with the 93-kDa protein (data not shown). The 93-kDa protein was transferred from the SDS-polyacrylamide gels to polyvinylidene difluoride membranes and treated with trypsin, the peptides were separated by HPLC, and a well separated peptide was sequenced. The sequence VSPWDYNAYYSVSDXVWF was obtained, which corresponds to amino acid residues 535–552 of Kell protein (14Lee S. Zambas E.D. Marsh W.L. Redman C.M. Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 6353-6357Crossref PubMed Scopus (206) Google Scholar). To determine if other proteins are associated with or are near neighbors of Kell, red cells were treated with a mouse monoclonal antibody to KEL14, which specifically reacts with a surface-exposed Kell epitope. The cells were lysed, and membranes were isolated and were incubated with or without a thiol-cleavable cross-linking reagent, dithiobispropionimidate. The membranes were then dissolved in detergents (sodium deoxycholate and Triton X-100), and the immune complexes were isolated, reduced, and separated by SDS-PAGE. The isolated proteins were detected by silver staining (Fig.3) or by Western immunoblotting using a polyclonal antibody to GAPDH or to Kell protein. The silver-stained protein patterns are shown in Fig. 3. In dithiobispropionimidate-cross-linked membranes (Fig. 3, lane 1), but not in untreated membranes (Fig. 3, lane 2), a 36-kDa protein co-precipitated with Kell protein. In these experiments, very little XK was detected, and the predominant silver-stained proteins were Kell and the heavy and light chains of IgG. The bands of approximately 30 and 24 kDa, that appear to be more prominent in the cross-linked sample in this experiment, were variable and were not characterized. The 36-kDa protein was unique in the cross-linked samples and was identified as GAPDH by amino acid sequencing of two peptides isolated from a trypsin digest. Two peptides, separated by HPLC, had the amino acid sequences LVSWYDNEFGYSNR and GALQXIIPASTGAAK. These two sequences correspond to amino acid residues 19–32 and 200–214 of human GAPDH. The 36-kDa protein also reacted on Western immunoblotting with a rabbit antibody to human GAPDH (data not shown). Hydropathy plots predict that XK has a single extracellular cysteine residue in its small, 11-amino acid, fifth loop (9Ho M. Chelly J. Carter N. Danek A. Crocker P. Monaco A.P. Cell. 1994; 77: 869-880Abstract Full Text PDF PubMed Scopus (238) Google Scholar) and that Kell has 15 extracellular cysteine residues (14Lee S. Zambas E.D. Marsh W.L. Redman C.M. Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 6353-6357Crossref PubMed Scopus (206) Google Scholar). Any one of the 15 extracellular cysteine residues of Kell could form a disulfide bond with the single extracellular cysteine of XK. To determine which of the cysteine residues are involved in disulfide linkage, the single cysteine extracellular residue in XK and two different cysteines in Kell were mutated to serine by site-directed mutagenesis. The wild-type or the mutated proteins were co-expressed in COS cells, and their ability to form Kell-XK disulfide-bonded complexes was determined. Kell is a member of a subfamily of membrane zinc metalloendopeptidases, which includes neutral endopeptidase 24.11, endothelin-converting enzyme, and the product of the PEX gene (14Lee S. Zambas E.D. Marsh W.L. Redman C.M. Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 6353-6357Crossref PubMed Scopus (206) Google Scholar, 15Turner A.J. Tanzawa K. FASEB J. 1997; 11: 355-364Crossref PubMed Scopus (385) Google Scholar, 16Xu D. Emoto N. Giaid A. Slaughter C. Kaw S. deWit D. Yanagisawa M. Cell. 1994; 78: 473-485Abstract Full Text PDF PubMed Scopus (863) Google Scholar, 17Shimada K. Takahashi M. Tanzawa K. J. Biol. Chem. 1994; 269: 18275-18278Abstract Full Text PDF PubMed Google Scholar, 18Valdenaire O. Rohrbacher E. Mattei M.-G. J. Biol. Chem. 1995; 270: 29794-29798Abstract Full Text Full Text PDF PubMed Scopus (208) Google Scholar, 19Rowe P.S. Oudet C.L. Francis F. Sinding C. Pannetier S. Econs M.J. Strom T.M. Meitinger T. Garabedian M. David A. Macher M.A. Questiaux E. Popowska E. Pronicka E. Read A.P. Mokrzycki A. Glorieux F.H. Drezner M.K. Hanauer A. Lehrach H. Goulding J.N. O'Riordan J.L. Hum. Mol. Genet. 1997; 6: 539-549Crossref PubMed Scopus (173) Google Scholar, 24Rawlings N.D. Barrett N.J. Methods Enzymol. 1995; 248: 183-228Crossref PubMed Scopus (699) Google Scholar). Members of this subfamily of zinc endopeptidases conserve 10 cysteine residues in their extracellular domains. We hypothesized that one of the five nonconserved cysteine residues of Kell is likely to be involved in linkage to XK. A likely candidate was Kell cysteine 72, since it is close to the membrane-spanning domain, as is the single cysteine of XK. As a control, we also mutated cysteine 319 of Kell to serine, since this cysteine residue is the next nonconserved cysteine from the transmembrane domain. COS cells were transiently transfected with an expression vector containing Kell and XK cDNAs in tandem and the recombinant Kell-XK complex isolated with an antibody to the second loop of XK. On analysis, by SDS-PAGE under reduced conditions, both Kell (93 kDa) and XK (40 kDa) were detected when wild-type XK and Kell proteins were co-expressed (Fig. 4 A,lane 1). Kell was not co-isolated with XK, indicating a lack of Kell-XK formation, when the Kell mutant (C72S) was co-expressed with wild-type XK (Fig. 4 A, lane 2) or when the XK mutant (C347S) was co-expressed with wild-type Kell (Fig.4 A, lane 4). However, co-expression of XK with another Kell mutant (C319S) did co-isolate Kell protein (Fig.4 A, lane 3), indicating that mutation of another nonconserved Kell cysteine residue does not affect Kell-XK formation and that Kell Cys72 specifically links Kell to XK. Analysis on nonreduced SDS-PAGE confirmed that a 134-kDa Kell-XK complex was formed when wild-type XK and Kell (Fig. 4 B,lane 1) and when wild-type XK and Kell mutant (C319S) (Fig.4 B, lane 3) were co-expressed. However no 134-kDa complex occurred when XK and Kell mutant (C72S) (Fig. 4 B,lane 2) or mutant XK (C347S) and wild-type Kell (Fig.4 B, lane 4) were co-expressed. Taken together, these results indicate that Kell Cys72 is linked to XK Cys347. Although COS cells assemble recombinant Kell and XK, not all of the expressed proteins were disulfide-linked. Free XK was detected with antibody to XK, in conditions that isolated Kell-XK complex (Fig.4 B, lanes 1 and 3), and when antibody to Kell was employed, free Kell was detected, in addition to the 134-kDa complex (Fig. 4 C, lane 1). To establish if Kell-XK formation is required for cell surface expression and also to determine the fate of mutant Kell (C72S), COS cells, co-expressing wild-type XK and Kell or wild-type XK and mutant Kell (C72S), were incubated overnight withl-[35S]methionine and treated with an antibody to Kell (anti-KEL14) that recognizes Kell surface antigens. The cells were washed and lysed with detergents, and the radioactive proteins associated with anti-KEL14 were analyzed by nonreduced SDS-PAGE and autoradiography. When wild-type XK and Kell proteins were co-expressed, both the 134-kDa XK-Kell complex and free Kell (93 kDa) were present on the cell surface (Fig. 5,lane 2). Upon co-expression of wild-type XK and mutant Kell (C72S), the 134-kDa complex was not detected on the cell surface, but mutant Kell (C72S) was present (Fig. 5, lane 1). This demonstrates that wild type Kell and XK occur as a complex on the cell surface of transfected COS cells but that Kell does not have to be complexed with XK to be transported to the cell surface. The Kell blood group glycoprotein is classified as a member of the neprilysin family (M13) of zinc metallopeptidases (24Rawlings N.D. Barrett N.J. Methods Enzymol. 1995; 248: 183-228Crossref PubMed Scopus (699) Google Scholar). Proteins in the M13 family are type II membrane glycoproteins and currently consist of neutral endopeptidase 24.11, Kell, endothelin-converting enzyme, and the product of the PEX gene. This topic was recently reviewed (15Turner A.J. Tanzawa K. FASEB J. 1997; 11: 355-364Crossref PubMed Scopus (385) Google Scholar). The physiological substrates of neutral endopeptidase 24.11 and endothelin-converting enzyme are known, but those of Kell and PEX have not yet been identified. Neutral endopeptidase 24.11 is widely distributed and has broad specificity, processing a large number of peptide hormones (15Turner A.J. Tanzawa K. FASEB J. 1997; 11: 355-364Crossref PubMed Scopus (385) Google Scholar, 25Roques B.P. Noble F. Dauge V. Fournie-Zaluski M.C. Beaumont A. Pharmacol. Rev. 1993; 45: 87-146PubMed Google Scholar). By contrast, endothelin-converting enzyme is more specific and activates a 38- or 39-amino acid "big" endothelin, by cleavage of a Trp21–Val22 bond, to an active 21-amino acid endothelin peptide (15Turner A.J. Tanzawa K. FASEB J. 1997; 11: 355-364Crossref PubMed Scopus (385) Google Scholar, 16Xu D. Emoto N. Giaid A. Slaughter C. Kaw S. deWit D. Yanagisawa M. Cell. 1994; 78: 473-485Abstract Full Text PDF PubMed Scopus (863) Google Scholar, 17Shimada K. Takahashi M. Tanzawa K. J. Biol. Chem. 1994; 269: 18275-18278Abstract Full Text PDF PubMed Google Scholar). Endothelin is a potent vasoconstrictor that plays an essential role in maintaining vascular tone. Endothelin-converting enzyme is a dimer consisting of two disulfide-linked 130-kDa subunits. Site-directed mutagenesis and expression in COS cells suggests that the endothelin-converting enzyme subunits are linked by Cys412, a residue not conserved in neutral endopeptidase 24.11, Kell, or PEX (26Shimada K. Takahashi M. Turner A.J. Tanzawa K. Biochem. J. 1996; 315: 863-867Crossref PubMed Scopus (87) Google Scholar). In this study, we demonstrate that Kell is covalently linked through a nonconserved Cys72 to Cys347 of XK protein (see Fig.6). By contrast, neutral endopeptidase 24.11 is not covalently associated with itself or with another protein. Therefore, Kell differs from the other members of the M13 family in that it is linked to another membrane protein. XK, to which Kell is linked, is an integral membrane protein, which probably spans the membrane 10 times and has both its NH2 terminus and COOH terminus within the cell. The function of XK is not known, although it has structural features that suggest a transport function. Thus, although the functions of both Kell and XK are unknown, their covalent association on the red cell membrane suggests cooperative activities. It should be noted, however, that XK is present in nonerythroid tissues, primarily in skeletal muscle, brain, and pancreas (9Ho M. Chelly J. Carter N. Danek A. Crocker P. Monaco A.P. Cell. 1994; 77: 869-880Abstract Full Text PDF PubMed Scopus (238) Google Scholar), while Kell may be restricted to erythroid tissues (13Lee S. Zambas E.D. Marsh W.L. Redman C.M. Blood. 1993; 81: 2804-2809Crossref PubMed Google Scholar). However, an extensive set of tissues has not been monitored for the presence of Kell protein. Kell is one of the major antigenic systems in human red cells mainly due to its complex polymorphisms. At least 20 different antigens are associated with Kell glycoprotein, and because of its immunogenicity severe reactions can occur if incompatible blood is transfused (1Redman C.M. Marsh W.L. Semin. Hematol. 1993; 30: 209-218PubMed Google Scholar, 2Marsh W.L. Redman C.M. Transfusion. 1990; 30: 158-167Crossref PubMed Scopus (73) Google Scholar). The molecular basis of the different Kell antigens has been elucidated and is due to single base mutations leading to amino acid changes (10Lee S. Vox Sang. 1997; 73: 1-11Crossref PubMed Scopus (103) Google Scholar). The covalent linkage of Kell and XK is not needed to express Kell antigens on the cell surface, since Kell antigens are expressed in a recombinant system that only synthesizes Kell protein and not XK (27Russo D.C. Lee S. Reid M. Redman C.M. Blood. 1994; 84: 3518-3523Crossref PubMed Google Scholar) and Kell antigens can be detected in McLeod red cells that lack XK (1Redman C.M. Marsh W.L. Semin. Hematol. 1993; 30: 209-218PubMed Google Scholar,2Marsh W.L. Redman C.M. Transfusion. 1990; 30: 158-167Crossref PubMed Scopus (73) Google Scholar). In this study, we also show that a specific Kell antigen, KEL14, is expressed on the surface of transfected COS cells, in the absence of XK, demonstrating that Kell protein can be transported to the cell surface without being linked to XK. In addition, since Kell protein, by itself, expresses KEL14, a conformational epitope, these data suggest that Kell protein assumes normal folding in the absence of XK. Since mutant Kell (C72S) was transported to the cell surface of COS cells in the absence of XK and retained its ability to be recognized by a specific monoclonal antibody to the KEL14 conformational epitope, these results further indicate that mutation of Kell (C72S) does not markedly change the tertiary structure of the protein. Our results are in agreement with previous studies (5Khamlichi S. Bailly P. Blanchard D. Goossens D. Cartron J.P. Bertrand O. Eur. J. Biochem. 1995; 228: 931-934Crossref PubMed Scopus (80) Google Scholar, 6Carbonnet F. Hattab C. Collec E. Le Van Kim C. Cartron J.P. Bertrand O. Br. J. Haematol. 1997; 96: 857-863Crossref PubMed Scopus (40) Google Scholar) that show that Kell and XK are disulfide-linked in the native state on human red cells. There is no evidence, however, that Kell and XK are part of a larger membrane complex, since antibodies to Kell or XK did not isolate other proteins and since chemical cross-linking studies, using a reagent that detects near neighbors with free amino groups, only co-isolated a cytoplasmic protein, GADPH. Although the cross-linking studies may not be quantitative, it appears that not all of the Kell proteins on red cells are cross-linked to GADPH. GADPH is known to be linked to other red cell membrane proteins, notably band 3, the anion transporter (28Low P.S. Biochim. Biophys. Acta. 1986; 864: 145-167Crossref PubMed Scopus (360) Google Scholar). The hydropathy plot of XK predicts five extracellular loops and predicts that Cys347 resides on the small fifth extracellular loop (9Ho M. Chelly J. Carter N. Danek A. Crocker P. Monaco A.P. Cell. 1994; 77: 869-880Abstract Full Text PDF PubMed Scopus (238) Google Scholar). Our results demonstrate that Kell cysteine 72, which is known to be in the extracellular domain of Kell protein (27Russo D.C. Lee S. Reid M. Redman C.M. Blood. 1994; 84: 3518-3523Crossref PubMed Google Scholar), is linked to XK Cys347. This indicates that the disulfide linkage of Kell to XK lies close to the extracellular membrane surface and supports the predicted topology of XK. We thank Xu Wu and Ying Cao for technical assistance in site-directed mutations and construction of expression vectors; Tellervo Huima, Yelena Oskov, and Robert Ratner for illustrations; and Jim Farmar and members of the Microchemistry Laboratory at the New York Blood Center for peptide synthesis and DNA sequencing.

Referência(s)