A blood group-related polymorphism of CD44 abolishes a hyaluronan-binding consensus sequence without preventing hyaluronan binding.

CD44 is a widely expressed integral membrane protein that acts as a receptor for hyaluronan (HA) and is proposed to be important to cell-extracellular matrix interaction. The Indian (In) blood group antigens reside on CD44, and most individuals express the Inb antigen. Homozygosity for the Ina allele occurs as a rare event and is associated with production of alloantibody to the common Inb antigen after transfusion or pregnancy. The present study demonstrates that a single point mutation (G252 --> C) causes an Arg46 --> Pro substitution, which is responsible for the Inb/Ina polymorphism. Additional mutations were found in In(a+b-) cDNA but were not necessary to the antigenic phenotype as determined in site-directed mutagenesis studies. In studies using CD44 chimeric constructs, Arg46 has previously been shown to be crucial for maintenance of HA-binding ability to a CD44 peptide. However, the present study demonstrates that the Arg46 --> Pro substitution does not reduce HA binding to the intact CD44 protein, which contains two proposed extracellular HA-binding motifs. Down-regulation of HA binding to In(a+b-) CD44 by anti-CD44 monoclonal antibody (mAb) ligands, however, was weakened, although all mAbs tested bound In(a+b-) and In(a-b+) CD44 equally well. Competitive inhibition studies using human anti-Inb also showed that some mAbs that inhibit HA binding to CD44 may do so by interacting with a domain separate from, but affecting the structure of, the Inb epitope.

Investigation of the biochemical and molecular basis of human erythrocyte blood group antigens has begun to contribute to our understanding of structure/function relationships of some biologically active red cell membrane proteins (1). CD44, also previously identified as In(Lu)-related p80 (2,3), the Hermes antigen (4,5), Pgp-1 (6 -8), and ECMRIII (9,10), is a widely expressed integral membrane protein that bears Indian (In) blood group antigens (11,12). CD44 has been shown capable of binding a number of extracellular matrix components, including hyaluronan (HA), fibronectin, heparan sulfate, and collagen types I and VI (10,13). In addition, studies using monoclonal antibodies to this protein have suggested that CD44 plays a role in lymphocyte homing to mucosal lymphoid tissue (14), in leukocyte activation (15), and in tumor metastasis (16). CD44 also appears important to lymphopoiesis (17) and to progenitor cell-ECM interactions during erythropoiesis (18).
Several studies have attempted to identify the sites at which CD44H binds hyaluronan. Resting lymphocytes express CD44H but only bind hyaluronan poorly. Jurkat cells transfected with CD44H cDNA will bind HA when stimulated with phorbol 12-myristate 13-acetate (PMA) 1 (26). However, PMAtreated cells transfected with CD44E cDNA did not bind HA unless first incubated with CD44 monoclonal antibodies known to enhance HA binding (26). Different monoclonal antibodies to CD44 may enhance, block, or not affect HA binding to CD44H (26,27); the mechanisms that effect alteration of HA-binding ability have not been defined. The cytoplasmic domain of CD44H is also known to be necessary for activation of HAbinding ability (26).
In the extracellular portion of CD44H, two proposed HAbinding domains containing positively charged amino acids (amino acids 29 -46 and 150 -162) have been shown to be important for maximal HA binding, using deletion and sitedirected mutants (28,29). However, the first domain is the only one that resides within the portion of CD44 homologous to cartilage link protein. Peach et al. (28) have shown that mutation of Arg 41 3 Ala in a construct containing amino acids 1-220 of CD44H almost abolished HA binding altogether. Yang and colleagues (29), using different chimeric constructs containing only small CD44 peptides, showed that an Arg 46 3 Gly substi-tution abolished binding of HA to a chimeric protein containing CD44 amino acids 38 -46 and proposed that HA binds to a minimal consensus motif BX 7 B, where B represents a basically charged amino acid (Arg or Lys) and X 7 represents seven nonacidic residues. HA binding was enhanced if additional basic residues were clustered around the ends of the motif or if a basic residue occurred within the X 7 domain (29). To date, no studies showing the effect of mutations within the first HAbinding motif (amino acids 38 -46) on HA binding by intact CD44H have been reported.
The In blood group system comprises two antigens presumed to be encoded by CD44 alleles (11,30,32). The In b antigen is of extremely high frequency, but the low incidence antigen In a is somewhat more common among peoples of the Middle East and Indian subcontinent. This study identifies the molecular basis of the In a /In b polymorphism and demonstrates that although this polymorphism abolishes the BX 7 B motif within the part of CD44 homologous to cartilage link protein, it does not inhibit HA binding by CD44H and reduces the effect of monoclonal antibody ligands on HA binding to CD44.

MATERIALS AND METHODS
Cells and Sera-Erythrocytes were collected in acid citrate-dextrose and washed in phosphate-buffered saline (PBS). For radioimmunoassays, a 5% (v/v) erythrocyte suspension in PBS was assayed according to previously published methods (33,34). EBV-transformed B lymphocytes from the In(aϩbϪ) donors Dh, Ra, and Bi were produced and propagated by standard methodology. Cell lines Ra and Bi were the generous gift of Drs. F. Spring and D. Anstee (Bristol, United Kingdom).
All monoclonal antibodies (mAbs) used in this study have been previously described or have been evaluated by the Fifth International Workshop on Leucocyte Differentiation Antigens (2,35). Human anti-In b were identified using routine blood bank methodology (agglutination). Anti-In b -Dh is from the patient previously described by Ferguson and Gaal (36). Anti-In b -Ra was supplied by the International Blood Group Reference Laboratory (Drs. F. Spring, J. Smythe, and D. Anstee, Bristol, United Kingdom). The specificity of both antisera has been confirmed by several laboratories.
cDNA Clones and Sequencing-The wild-type CD44 (CD44WT) cDNA clone was provided in the pCDM8 vector as a gift of Dr. Brian Seed (Boston, MA) (37); the insert was then cloned into the pcDNA3 expression vector (Invitrogen, San Diego, CA). In(aϩbϪ) Dh and Ra cDNA clones were obtained by reverse transcription-polymerase chain reaction (RT-PCR) of mRNA obtained from EBV-transformed lymphocytes using the FastTrack mRNA isolation system (Invitrogen) or standard guanidinium isothiocyanate-cesium chloride isolation techniques (38). Synthesis of first-strand cDNA was accomplished using the GeneAmp RT-PCR kit (Perkin-Elmer), using oligonucleotide MT2 (Table I). Amplification by PCR was performed using oligonucleotides MT1 and MT2 (Table I). Amplification conditions were as follows: denaturation for 1 min at 95°C, annealing for 2 min at 48°C, and extension for 3 min at 72°C for 40 cycles, followed by a 10-min final extension period. The phosphorylated PCR product was cloned into calf intestinal alkaline phosphatase-treated pTZ18U (Bio-Rad) linearized with SmaI. Sequencing was accomplished by standard dideoxy methodology using Sequenase TM 2.0 (U.S. Biochemical Corp.).
PCR of genomic DNA sequence from In(aϩbϪ) propositus Bi was accomplished as described above, using the primers MT11 and MT12 (Table I), 1 g of genomic DNA, and an annealing temperature of 55°C.
Site-directed Mutagenesis-CD44WT in pTZ18U was used as template for synthesis of mutated cDNA fragments, using the mutagenizing primers listed in Table I in conjunction with either MT1 (for the noncoding primer ⌬252n) or MT2 (for the coding-sense primer ⌬252s), in order to synthesize overlapping mutated CD44 cDNA fragments. These fragments were gel-purified (GeneClean, Bio101, La Jolla, CA) and coincubated in PCR reaction buffer, including nucleotides and Taq polymerase (Perkin-Elmer) for three amplification cycles (denaturation for 1 min at 95°C, annealing for 2 min at 65°C, and extension for 3 min at 72°C). Oligonucleotide primers MT1 and MT2 were then added, and the reaction was continued for an additional 35 cycles, using 48°C as annealing temperature. Products were then cloned into pTZ18U and sequenced to verify the complete coding sequence.
Expression of CD44 cDNA in Mammalian Cells-Naturally occurring and mutated CD44 cDNA sequences were cloned into pcDNA3 between the EcoRI and HindIII sites of the expression vector, in an orientation that would allow expression of CD44 cDNA via the cytomegalovirus promoter. Jurkat cells (250 l of 1 ϫ 10 7 cells/ml), which have previously been shown not to express CD44 mRNA or protein, were transfected with 20 -25 g of plasmid DNA containing insert by electroporation (0.28 kV, capacitance at 960 microfarads, Bio-Rad Gene Pulser TM ) and then grown in 5 ml of RPMI 1640 medium with 10% fetal calf serum. For Western blots, cells were harvested 24 h post-transfection. Stably transfected cell lines were produced by long term culture in similar media with 1 mg/ml G418 (Life Technologies, Inc.) followed by selection of strongly CD44-positive cells by sterile cell sorting using mAb A3D8, fluorescein isothiocyanate (FITC)-labeled anti-mouse Ig (Tago, Burlingame, CA), and a Becton-Dickinson FACStar plus flow cytometer (Mountain View, CA).
Western Blot Analysis-Transfected Jurkat cells (derived from 2.5 ϫ 10 6 cells transfected 24 h previously) were harvested, washed in PBS, and lysed in 100 l of PBS, 1% Nonidet P-40 (Calbiochem), 1 mM phenylmethylsulfonyl fluoride for 30 min at 4°C. Nonsolubilized material was removed by centrifugation, and an equal volume of 2 ϫ nonreducing sample buffer (0.125 M Tris, pH 6.8, 2.5% SDS, 10% glycerol, 0.008% bromphenol blue) was added to the protein solution. Solubilized cellular proteins were then separated by 10% SDS-polyacrylamide discontinuous gel electrophoresis (39) and transferred to nitrocellulose (Costar, Cambridge, MA) (40). Detection of binding of mAb A3D8 and human anti-In b was accomplished using the Protoblot alkaline phosphatase detection system (Promega, Madison, WI) according to the manufacturer's directions.
Immunofluorescence, Hyaluronan Binding, and Radioimmunoassays-Surface expression of CD44 by stably transfected Jurkat cells was assayed by indirect immunofluorescence, using mAb A3D8, as described previously (35). Rooster comb HA (Sigma) was conjugated with fluorescein isothiocyanate (Sigma) as described (26,41,42). CD44ϩ and nontransfected Jurkat cells (1 ϫ 10 7 cells/ml) were cultured in the presence or absence of 16 nM PMA for 16 h prior to HA-binding assay. HA binding was measured by incubation of 50 l of CD44ϩ or nontransfected Jurkat cells resuspended at 3 ϫ 10 6 cells/ml in PBS-BSA with an equal volume of FITC-HA (0.1 mg/ml); negative controls included incubation of FITC-HA with nontransfected Jurkat cells, immunofluorescent staining with P3 and FITC-labeled anti-mouse Ig, and blocking of FITC-HA binding by 10 g/ml unlabeled HA (26). In some experiments, cells were preincubated with saturating amounts of anti-CD44 monoclonal antibodies, washed in PBS-BSA, and then incubated with FITC-HA. All incubations with HA and wash steps were performed at 4°C. After washing, cells were resuspended in 0.5 ml of PBS containing 0.3% paraformaldehyde prior to analysis by flow cytometry. All flow cytometry analyses were performed using an Ortho Cytoron flow cytometer (Ortho Diagnostics, Raritan, NJ) and repeated at least three times in separate experiments.
In some experiments, PMA-treated nontransfected, and CD44-transfected Jurkat cells were incubated in a final concentration of 0.005 M dithiothreitol for 1 h at 4°C to denature cell surface disulfide bonds. Cells were then washed and resuspended in PBS-BSA. Viability of treated and sham-treated cells was determined by trypan blue exclusion and was uniformly Ͼ95% for both dithiothreitol-treated and shamtreated cells.
Radioimmunoassays were performed as described previously (33,34). For competitive inhibition assays, red cells were incubated for 1 h at room temperature with blocking antibody, washed twice with PBS-BSA, and incubated with second antibody for 1 h at room temperature. After repeated washing steps, cells were resuspended in a fixed volume, incubated with radiolabeled goat anti-mouse or anti-human IgG, and centrifuged through phthalate oils. Centrifuge tube tips containing cell pellets were analyzed in a ␥ scintillation counter (LKB/Wallac Inc., Gaithersburg, MD) to quantitate radiolabel bound to cells.

Identification of DNA Polymorphisms in Three Unrelated
In(aϩbϪ) Propositi-During the course of these and previous studies, two In(aϩbϪ) individuals, Dh and Ra, were identified, and their B lymphocytes were transformed using EBV. Cloning and sequencing of RT-PCR products of propositus Dh mRNA led to identification of five mutations in comparison with the CD44H sequence published by Stamenkovic et al. (37) (Table II and Fig. 1). However, the mutations at nucleotides 322 and 370 were silent and were present in only 50% of all clones, suggesting that Dh was either heterozygous for these mutations or that these mutations were the result of errors during amplification. This latter possibility was made less likely by the fact that these silent mutations were present in multiple clones obtained from several independent PCR amplification reactions. We then performed RT-PCR on cDNA extracted from a second EBV-transformed cell line from an unrelated In(aϩbϪ) propositus (Ra). Partial sequencing of cDNA clones obtained from this donor demonstrated the same mutations at nucleotides 252, 370, and 441 as did Dh. However, the silent mutation at nucleotide 322 was not found, and sequencing was not extended to nucleotide 831. Comparison of the Dh sequence with that published by Goldstein et al. (43) further showed that the nucleotides found at positions 441 and 831 (codons 109 and 239, respectively) in Dh matched both that sequence as well as the murine CD44 sequence. Thus, the G 252 3 C substitution  Expression of In b Antigen by Cells Transfected with CD44 cDNA Constructs-In order to demonstrate which mutation(s) was responsible for the In b /In a alteration of phenotype, Jurkat cells were transfected with wild-type CD44H cDNA (CD44WT, as described by Stamenkovic et al. (37)), cDNA amplified from the In(aϩbϪ) propositus Dh (CD44Dh), or a mutated cDNA (CD44(R46P)) that differed from CD44WT only at nucleotide 252 (Table II). Transiently transfected cells were tested for CD44 and In b antigen expression using mAb A3D8 and two human anti-In b sera, respectively (Fig. 2). All cells except nontransfected cells expressed easily detectable levels of CD44 migrating at approximately 80 kDa on SDS-polyacrylamide gel electrophoresis (Fig. 2, left panel). However, only CD44WT reacted with human anti-In b , whereas CD44Dh and CD44(R46P) did not (Fig. 2, right panel). These results demonstrate that the Arg 3 Pro substitution at amino acid 46 is sufficient to effect the In b /In a antigenic polymorphism.

Effect of In b 3 a -associated Alteration of the First Hyaluronan-binding Motif on Hyaluronan
Binding-The first HAbinding motif identified by the sequence BX 7 B is amino acids 38 -46. The Arg 46 3 Pro substitution of In(aϩbϪ) cells thus removes the basically charged amino acid at the C terminus of this motif. In previous studies, substitution of this amino acid by a glycine has been shown to abolish HA binding by this sequence expressed as part of a chimeric protein (29). There-fore, we examined the ability of the intact CD44 molecule with a mutation at this site to bind HA. Previous work has demonstrated that there is little HA binding to Jurkat cells expressing transfected CD44 cDNA, while incubation with PMA increases the ability of transfected CD44H to bind HA, and PMAtreated nontransfected Jurkat still does not bind HA (42).
We studied HA binding by cells transfected with CD44WT, CD44Dh (cDNA from the In(aϩbϪ) propositus Dh), and CD44(R46P). As shown in Fig. 3, when compared with PMAtreated CD44WT-transfected cells, PMA-treated CD44Dh transfectants showed equal ability to bind HA, as measured both by percentage of positive cells as well as by degree of binding per cell, indicated by mean fluorescence channel. When the ability of CD44(R46P) to bind HA was measured, it also was shown to bind HA as well (Fig. 4).
Therefore, despite alteration of the first BX 7 B motif of CD44, intact In(bϪ) CD44Dh and CD44(R46P) maintained their ability to bind HA.
In experiments in which PMA-treated transfected Jurkat cells were pretreated with dithiothreitol, neither CD44WT nor CD44Dh bound HA (data not shown).
Effect of Alteration of the HA-binding Site of CD44WT on the In b Epitope-We studied the ability of five antibodies that inhibited and one antibody each that enhanced or had no effect on HA binding to interfere with the ability of human anti-In b to bind to its epitope on In(aϪbϩ) erythrocytes. Reverse competitive inhibition experiments were also performed.
All antibodies were shown by Western blotting to bind equally well to In(aϩbϪ) and In(aϪbϩ) CD44 (data not shown). mAbs BRIC235, HP2/9, BU75, 5F12, and 3F12 had been previously shown to partially or completely inhibit HA binding to CD44-transfected Jurkat cells; mAb F10.44.2 had been shown to enhance HA binding 2-fold (42). mAbs BU75 and 3F12 had further been shown to inhibit 5F12 binding to red cells by 84 and 71%, respectively (27). Saturating dilutions of all CD44 monoclonal antibodies tested produced Ͼ10% inhibition of binding of subsaturating dilutions of human anti-In b . However, only three antibodies (Bric235, Bu75, and HP2/9) produced more than 25% competitive inhibition of anti-In b binding (Fig. 5). In reverse experiments, human anti-In b was unable to inhibit or enhance binding of subsaturating dilutions of anti-CD44 mAbs by greater than 25% (data not shown).
Comparison of Effect of Monoclonal Antibody Ligands on HA Binding to CD44WT, CD44Dh, and CD44(R46P)-Monoclonal antibody 5F12 has been previously described to inhibit markedly binding of HA to PMA-stimulated CD44-transfected Jurkat cells (26,42). As shown in Fig. 6, HA bound to 36.2% of CD44WT-transfected Jurkat cells in the absence of 5F12 but to only 3.0% of cells in the presence of 5F12. However, HA binding to CD44Dh and CD44(R46P) transfectants was reduced by incubation with 5F12 from 45.1 to 30.6% and from 50.0 to 17.1%, respectively. In fact, four other monoclonal antibodies (BRIC235, HP2/9, BU75, and 3F12) that inhibited HA binding to CD44WT-transfected cells showed reduced inhibitory effect on HA binding to CD44Dh and CD44(R46P) (data not shown). In all cases, antibodies had greater inhibitory effect on CD44(R46P) than on CD44Dh, which may be attributable to the other amino acid changes encoded for by CD44Dh cDNA. However, approximately similar degrees of enhancement of HA binding were seen when F10.44.2 was used to augment HA binding to all three variants of CD44 (data not shown).
Effect of Human Anti-In b on HA binding to CD44WT-Incubation of PMA-treated CD44WT transfectants with human anti-In b had no effect on binding of HA-FITC (data not shown). DISCUSSION The CD44 protein (Fig. 7) is known to be a major receptor for HA, and it may also have physiologic importance as a receptor for fibronectin and laminin (18,46). The N terminus of CD44 is homologous to cartilage link protein, but this domain only contains one of the two amino acid motifs (KNGRYSISR, residues 38 -46) thought to be active in cellular binding to HA (29). The more C-terminal portion of the extracellular domain of CD44 encompasses two overlapping BX 7 B motifs (RDG-TRYVQKGEYR, residues 150 -162). Thus far, there has been insufficient evidence to indicate whether one of these two amino acid motifs is the more important in HA binding. Both motifs, expressed either as peptides or as part of chimeric molecules, have shown HA-binding activity (28, 29), and Liao et al. have suggested that the two motifs can come into proximity to one another and can cooperate in HA binding (44).
We have shown that a naturally occurring polymorphism of CD44 coding sequence is responsible for the expression of In a / In b blood group antigens. Evidence that the basis of the In b /In a polymorphism is an Arg 46 3 Pro substitution includes both cDNA and genomic DNA sequence analysis as well as antigen expression by transfected cell lines expressing various CD44 cDNA constructs. The localization of this polymorphism to the region of CD44 homologous to cartilage link protein is consistent with previously described characteristics of In antigens, including sensitivity to degradation by proteases and reducing reagents. Analysis of additional propositi expressing the In a antigen would be required to ascertain whether either the silent mutations or variable residues at positions 109 and 239 are in linkage disequilibrium with the In a -type sequence at position 46.
Since the In(aϩbϪ) phenotype has not been associated with any clinical abnormalities, it was surprising to find that the In a -associated Arg 46 3 Pro mutation disrupted an HA consensus binding motif. Studies of transcripts from both the CD44Dh and CD44(R46P) constructs containing this amino acid substitution showed that altering the C-terminal basic amino acid of the proposed first HA-binding motif in intact CD44 did not inhibit HA binding to CD44 expressed at the cell surface. In contrast, previously described alteration of this residue in a model studied by Yang  . In both immunofluorescence and Western blot assays, mAb 5F12 as well as other mAbs bound equally well to all CD44 proteins expressed (data not shown). In the absence of 5F12, HA-FITC bound to CD44 transfectants in proportion to CD44 expression. However, saturating concentrations of 5F12 antibody inhibited HA binding to CD44WT by 92%, while HA binding to CD44Dh and CD44(R46P) was inhibited by only 32 and 66%, respectively.
FIG. 5. Inhibition of anti-In b reactivity with human red blood cells by anti-CD44 mAbs with various effects on HA binding. mAbs BRIC235, 3F12, BU75, HP2/9, and 5F12 have been previously shown to inhibit HA binding to CD44, while mAb NIH44 -1 had no effect on HA binding, and mAb F10.44.2 increased HA binding 2-fold (42). mAbs were individually incubated with red cells, which were then washed and exposed to human anti-In b and radiolabeled anti-human Ig.
Binding was compared with that achieved when cells were preincubated with P3 nonreactive murine myeloma protein. BRIC235 and HP2/9 showed marked but incomplete inhibition of anti-In b binding, while mAb 5F12, which inhibits HA binding equally well, showed little inhibition. No antibody enhanced binding of anti-In b .
HA. Their experimental system, however, did not utilize the intact protein; instead, amino acids 38 -46 of CD44 were expressed as part of a chimeric construct, and mutation of Arg 46 3 Gly abolished binding of biotinylated HA to the CD44 peptide/RHAMM chimeric protein immobilized on nitrocellulose.
The work of Liao and colleagues (44) has shown that mutation of Arg 154 3 His in the second putative HA-binding motif also did not abrogate HA binding to recombinant intact CD44 expressed in Jurkat cells, although other studies on the second HA-binding motif have shown that altering basic residues within this motif did reduce the ability of CD44 chimeric constructs to bind HA (28, 29). Clearly, conclusions from studies of HA binding to small peptides or chimeric constructs cannot necessarily be extrapolated to the full protein expressed by intact cells.
The cooperation of both extracellular BX 7 B motifs in the binding of HA has been proposed by Liao et al. (44). This hypothesis is further supported by the complete abrogation of HA binding to PMA-treated CD44-transfected Jurkat cells pretreated with dithiothreitol. As depicted in Fig. 7, the conformation of CD44 is likely to be highly dependent on disulfide bonds. Interaction of the BX 7 B motifs would thus be likely to be disrupted by reduction of these bonds and subsequent alteration of CD44 conformation. However, it should be noted that other studies concerning HA binding to proteins containing BX 7 B binding motifs have shown that the motif may not be required for the ligand to bind the molecule. Specifically, mutants of the cartilage link protein in which the exon containing this motif had been deleted nevertheless bound HA (47).
Studies using monoclonal antibodies that can affect ligand binding have helped to identify possible recognition sites on receptors (48). The ability of some anti-CD44 mAbs that inhibit HA binding to also partially inhibit binding of anti-In b to red cells is in concordance with the hypothesis that the In b epitope is closely related to an HA-binding domain. Inability to achieve more complete inhibition may be due to the polyclonal nature of human anti-In b sera or may be due to the mechanism whereby these mAbs may adversely affect HA-binding ability. Since anti-In b binding was partially inhibited by preincubation of cells with CD44 mAbs, but anti-In b did not reduce mAb reac-tivity to CD44 on erythrocytes, these monoclonal antibodies likely inhibit HA binding via transmitted effects, i.e. alteration of conformation of the CD44 molecule. Such an alteration of the structure of CD44 would prevent binding of anti-In b without directly blockading the In b epitope, whereas the reverse incubations might have no effect on the binding of the CD44 mAbs that themselves affect HA binding. Similarly, the preincubation of anti-In b with PMA-treated CD44WT-transfected cells did not prevent binding of HA, as the antibody might not alter the critical configuration of the binding site(s).
Further, our data regarding the effects of CD44 mAbs on HA binding to CD44Dh and CD44(R46P) might then be interpreted as due to alteration of ligand effect on the interaction of the two HA-binding motifs. If mutating Arg 46 3 Pro alters the mobility of the first HA-binding motif of CD44, then alteration of the tertiary structure by CD44 ligands might be hampered. It is thus likely that the substitution of a basic residue at the carboxyl-terminal end of the first motif, although it does not inhibit HA binding, does influence the effect mediated by monoclonal antibodies to CD44. Studies by Zheng et al. (48) have shown that transmission of effects of CD44 mAbs on HA binding can be hampered when residues have been altered, although ligand binding is maintained. The conformational change of CD44 thought to be induced by the previously described mAbs (42) may be reduced when Pro is substituted for Arg 46 , as shown by maintenance of HA-binding ability. Our results may thus be considered consistent with the theory that various domains of CD44 interact to form a recognition site for HA (44,48).
It should be noted that, although the Tyr 109 3 Ser substitution is common in human CD44H and conserved among species, the presence of Tyr 109 does not inhibit CD44 binding to HA, as theorized by Dougherty et al. (31) Both CD44WT and CD44(R46P) proteins, both of which contain Tyr 109 , bind to HA.
In conclusion, demonstration of a blood group antigen-related polymorphism within a functional domain of a protein may no longer be considered unexpected (1). Another recently described example is the Diego blood group antigen polymorphism of the anion exchange protein AE1 (45). One may expect that future exploration of the molecular basis of blood group antigen variants will uncover further such polymorphisms, which can then be used to extend our understanding of structure/function relationships of other membrane proteins. Such polymorphisms are especially useful because, as they occur naturally, their impact on protein function and the importance of a particular protein in various physiologic processes can be assessed from clinical evaluation of individuals with the variant phenotype.