A Point Mutation Leads to an Unpaired Cysteine Residue and a Molecular Weight Polymorphism of a Functional Platelet p3 Integrin Subunit THE Sr" ALLOANTIGEN SYSTEM OF GPIIIa"'

Recently, we described a low frequency platelet al- loantigen on human platelet membrane glycoprotein (GP) IIIa, termed Sr", that was involved in neonatal alloimmune thrombocytopenia. To identify the molecular nature of the Sr" alloantigen, we analyzed the nucleotide sequence of polymerase chain reaction-amplified GPIIIa mRNA, and found a CZoo4+T substitution in seven Sr" positive individuals which results in an Arg636 + Cys polymorphism within the cysteine-rich region of GPIIIa. Analysis of allele-specific recombinant forms of GPIIIa that differed only at amino acid residue 636 showed that anti-Sr" alloantibodies reacted with the but not the recombinant form of GPIIIa. Interestingly, under reducing conditions, the form of GPIIIa migrated with a slightly increased apparent molecular weight compared with the form. Following treatment with Endoglycosidase H, both allelic

Recently, we described a low frequency platelet alloantigen on human platelet membrane glycoprotein (GP) IIIa, termed Sr", that was involved in neonatal alloimmune thrombocytopenia. To identify the molecular nature of the Sr" alloantigen, we analyzed the nucleotide sequence of polymerase chain reaction-amplified GPIIIa mRNA, and found a CZoo4+T substitution in seven Sr" positive individuals which results in an A r g 6 3 6 + Cys polymorphism within the cysteine-rich region of GPIIIa. Analysis of allele-specific recombinant forms of GPIIIa that differed only at amino acid residue 636 showed that anti-Sr" alloantibodies reacted with the but not the recombinant form of GPIIIa. Interestingly, under reducing conditions, the form of GPIIIa migrated with a slightly increased apparent molecular weight compared with the form. Following treatment with Endoglycosidase H, both allelic forms of GPIIIa exhibited the same mobility, however the Sr" epitope was lost. Sr" positive platelets express the same number of GPIIb-IIIa complexes on their surface as wild-type Sra negative platelets, and also aggregate normally in response to a variety of platelet agonists. Based upon these results, we conclude that 1) GPIIIa residue 636 specifically controls the formation and expression of the Sr" alloantigenic determinant, and 2) an unpaired cysteine residue alters the N-linked glycosylation pattern of the extracellular domain of GPIIIa, but affects neither the degree of surface expression nor the adhesive function of the GPIIb-IIIa complex.
Integrins constitute a large family of cell surface aP heterodimers that are widely distributed on many cells and are involved in cell-matrix or cell-cell interactions. Human platelet glycoprotein IIIa (GPIIIa)' is the common /3 subunit of the P3 subfamily of integrins and associates with either cqlb (GPIIb) to * This work was supported in part by Grant HL-44612 (to P. J. N.) from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. form the major platelet fibrinogen receptor, GPIIbLIIa (integrin aIIbP3) on human platelets ( l ) , or with a, to form the vitronectin receptor, a&, which is more widely distributed. Mature GPIIIa protein consists of 762 amino acids, including six potential N-glycosylation sites in the extracellular domain that have been shown to be post-translationally modified with high-mannose carbohydrate residues (2)(3)(4). GPIIIa also contains 56 cysteine residues in highly conserved locations within the extracellular domain of the molecule, all of which are normally disulfide-linked. Thirty-one of these cysteine residues comprise a cysteine-rich, protease-resistant core of the molecule and are clustered between residues 433 and 655 into four tandemly repeated segments of about 40 amino acids each (4). Another 7 cysteines are concentrated within the first 50 aminoterminal amino acids (4,5 ) . The cysteine pairing pattern of GPIIIa has been largely established by Calvete et al. (6), who showed that a number of long range disulfide bonds form within the molecule, bringing together at least two linearly distant polypeptide segments. Overall, these cysteine residues likely serve an important role in preserving the overall threedimensional structure of the integrin complex, as perturbation or absence of 1 or more of these residues has been shown to severely affect stability and/or ligand binding function (7).
In addition to its role in mediating cell adhesive interactions, GPIIIa is the most polymorphic of the integrin subunits, with six currently recognized alleles known to exist in the human gene pool. Differences in these allelic isoforms have been shown to be responsible for eliciting an alloimmune response leading to platelet destruction in two clinically significant pathologic syndromes, post transfusion purpura and neonatal alloimmune thrombocytopenia (for recent reviews, see Refs. 5,8,9). The advent of platelet mRNAPCR technology (10) has made it possible to elucidate the molecular basis for a number of these platelet membrane glycoprotein polymorphisms. To date, all of these have been found to result from single nucleotide, and consequently single amino acid substitutions (11-171, which are in turn thought to subtly affect the conformation of the protein, leading to expression of the offending antigenic determinant. Recently, we described a new low frequency platelet alloantigen on GPIIIa, termed Sra, that was responsible for a case of neonatal alloimmune thrombocytopenia (18). Since that time several other private platelet alloantigens (CA, Va, Mol that reside on GPIIIa have also been reported (16,(19)(20)(21). In this study, we have further characterized the biochemical and molecular properties that underlie the polymorphism of human platelet GPIIIa that is responsible for the immunogenicity of the Sr" alloantigen, as well as their effects on expression and function of this unique integrin isoform.

8439
The Sr" Alloantigen System of GPIIIa

MATERIALS AND METHODS
Serological Studies-Typing of human platelets for the presence of P1" and Sr" alloantigens was performed using the glycoprotein-specific immunoassay, MAIPA, using anti-PIA1-and anti-Sr"-specific alloantisera and the monoclonal antibody Gi5 directed against GPIIb/IIIa a s capture antibody (22). The platelet-specific alloantibodies PIA1 and Sr;' were obtained from mothers of children with neonatal alloimmune thrombocytopenia, a s previously described (18). Monoclonal antibodies (mAbs) Gi5 and Gi22 directed against GPIIb/IIIa and GPIb,, respectively, were produced in our laboratory (23, 24).
Labeling ofPlatelets-Platelets were surface labeled with 5 m M NHS-LC-Biotin (Paesel, Frankfurt, Germany) as previously described (25). For thiol-specific labeling, washed platelets were radiolabeled with tritiated N-ethylmaleimide (NEM) as described by Kalomiris and Coller by boiling for 10 min in SDS buffer. In some instances, the immune complexes were treated with Endoglycosidase H before elution from the beads (see below). Immunoprecipitates were analyzed on 7.5% SDS-PAGE under reduced conditions. Gels containing biotinylated proteins were transferred to nitrocellulose by immunoblotting and visualized using streptavidin-horseradish peroxidase and chemiluminescent substrate as recommended by the manufacturer (ECL Western Blotting Detection System; Amersham Buchler, Braunschweig, Germany). Gels containing radiolabeled proteins were dried and fluorographed or autoradiographed on Kodak X-Omatt film using an intensifying screen (Cronex Hi-Plus; DuPont, Frankfurt, Germany). "GMethylated rainbow protein mixture (Amersham Buchler, Braunschweig, Germany) was run as molecular weight markers in parallel.
Analysis of N-Linked Glycosylation-One-hundred-pl aliquots of biotinylated platelet detergent lysates were incubated with 50 milliunits of Endoglycosidase H (Boehringer Mannheim, Germany) overnight at 37 "C and immunoprecipitated with human sera, and analyzed by SDS-PAGE as described above. For comparison, washed immunoprecipitates were divided into two 50-pl aliquots and resuspended in 50 pl of acetate buffer (0.15 M sodium acetate, 0.1 M phenylmethylsulfonyl fluoride, 0.170 SDS, pH 5.5) in the absence or presence of 15 milliunits of Endoglycosidase H. After an overnight incubation a t 37 "C, the reaction was stopped by adding an equal volume of two times SDS buffer.
Isolation and Amplification of Platelet RNA-Human platelet mRNA was isolated from EDTA-anticoagulated blood as previously described (12). Twenty-nine-p1 aliquots of platelet mRNA were heated to 68 "C for 10 min and quickly cooled on ice water before reverse transcription. The GPIIIa-specific PCR primers used in these studies were constructed based on the published numbering system of Mannheim, Germany) was added a t 85 "C and "hot-start" PCR was performed in a DNA thermal cycler (Biometra, Gottingen, Germany). Amplification proceeded for 30 cycles, with denaturation for 1.5 min at 94 "C, annealing of primers for 1.5 min a t 50 "C, and extension for 3 min at 72 "C. Isolation and Amplification of Genomic DNA-Genomic DNA was isolated from peripheral blood leukocytes using the proteinaseisalting out procedure of Miller et al. (27). A 335-bp region of the GPIIIa gene encompassing the Sr polymorphic nucleotide was amplified using the PCR primer pair shown in Table I. PCR was performed using 1-2 pg of DNA and 0.5 p~ of each primer using 2.5 units of Taq polymerase in PCR buffer in a total volume of 100 pl as described above. Thirty cycles of 1.5 min at 96 "C, 1.5 min a t 57 "C, and 3 min at 72 "C were performed.
Analysis of PCR Products-Five pl of PCR-amplified products were analyzed on 1.4% agarose gel containing ethidium bromide (Dianova, Hamburg, Germany). Selected amplified cDNAs were purified by Geneclean (Dianova, Hamburg, Germany) and subcloned into the plasmid vector pGEM-5Zf (Promega Biotech, Madison, WI). Plasmids from positive clones were sequenced by the dideoxy termination method using a Sequenase 2.0 (United States Biochemical, Bad Homburg, Germany) as recommended by the manufacturer. Computer analyses of protein and nucleic acid sequences were performed using the program PC/GENE (Intelligenetics Inc and Genofit SA) on an IBM PC-compatible computer.
Allele-specific Oligonucleotide Hybridization Analysis-Ten p1 of amplified DNA was applied to nylon membranes ( N Y 13N, Schleicher & Schuell, Dassel, Germany) and hybridized at 54 "C with 17-mer oligonucleotide probes having specificity for either the SP or Srh sequence, differing only in the central nucleotide (shown in Table I). The oligonucleotide probes were end-labeled with digoxigenin-ll-2',3'-dideoxyuridine-triphosphate (DIG-11-ddUTP, Boehringer Mannheim, Germany) as previously described (28) and were immunologically detected using alkaline phosphatase-conjugated polyclonal sheep anti-digoxigenin Fab fragment antibody and AMPPDTM (Serva, Heidelberg, Germany) as a chemiluminescent substrate. Positive reactions were visualized by exposing the nylon membrane to Kodak X-Omatt X-ray film (Rochester, N Y ) .
Construction and Expression ofAllele-specific Recombinant Forms of GPZIZa-Allele-specific recombinant forms of GPIIIa were created using cartridge mutagenesis. A full-length GPIIIa cDNA, the internal EcoRI restriction site of which had been removed by site-directed mutagenesis (kindly provided by Dr. Gilbert C. White 11, University of North Carolina, Chapel Hill, NC) was used as a host vector for construction of allelic GPIIIa isoforms. Plasmids (pGEM-BZf, Promega Biotech, Madison, WI) from two clones containing GPIIIa nucleotides 1666-2237 of platelet GPIIIa, and having either a T or a C at position 2004 were digested with MluI and AflII (New England Biolabs, Bad Schwalbach, Germany). The resulting 249-bp fragment encoding the polymorphic base was gel-purified and ligated to full-length GPIIIaA3luescript cDNA which had been digested with the same enzymes. After subcloning, the resulting plasmid construct inserts were removed with EcoRI, flushed using Klenow DNA polymerase, and finally subcloned into the EcoRV site of the mammalian expression vector pcDNA N E 0 (ITC, Heidelberg, Germany). Purified plasmids used for subsequent transfection were validated by nucleotide sequence analysis.
COS cells were transfected with both allele-specific recombinant forms of GPIIIa using the DEAE-dextran method, as previously described (29).

RESULTS
Amplification a n d Analysis of the NH2and COOH-terminal Regions of GPIIIa Platelet mRNA-Previous immunochemical studies have shown that anti-SP alloantibodies react with the 68-kDa fragment of GPIIIa generated by chymotryptic treatment of intact platelets (18). Since this fragment has been shown to lack residues 121-348 of the large disulfide-bonded loop formed by disulfide bonding of Cys5 with Cys4"" (6,31-33), we were able to predict that the remaining region formed by amino acid residues 1-120 and 349-762 (encompassing nucleotides 99-470 and 1143-2384, respectively) of GPIIIa is likely to carry the Sr" epitope. Thus, we amplified this region as two separate segments of 642 (nucleotides 56-698) and 571 (nucleotides 1666-2237) base pairs. Nucleotide sequence analysis of the resulting 571-bp COOH-terminal fragment derived from an Sr" positive individual revealed a single C -> T nucleotide substitution at base 2004 ( Fig. 1) in five of seven subclones examined, consistent with the fact that all Sr" positive individuals examined to date have been serologically heterozygous for this low frequency allelic form of GPIIIa. In contrast, all clones from an Sr" negative individual encoded a C a t this position (not shown). No other nucleotide differences between Sr" positive versus negative individuals were found.
Correlation of C2004 Q T Polymorphism with Sr Allotype-In order to determine whether the C + ) T polymorphism seen in the one Sr" positive individual was associated with Sr phenotype, genomic DNA from six Sr" positive and 10 different Sr" negative individuals (four family members and six unrelated donors) were amplified using PCR to yield a 335-bp product (not shown). Allele-specific oligonucleotide typing was then performed using 17-base probes containing the putative genotypespecific nucleotide in the middle (Table I, right). As shown in the top panel of Fig. 2, the putative SF-specific probe containing a T in the middle hybridized with the PCR products derived from all six Sr" positive heterozygous individuals, but was negative with 10 SF negative individuals. In contrast, the wildtype "Srb"-specific probe, containing a C in the middle, hybridized with both Sr" heterozygous and Sr" negative individuals. Based upon these results, we conclude that the observed base change at nucleotide 2004 segregates directly with the phenotypic presence of the Sr" alloantigen. Importantly, the C -> T mutation changes a -CGT codon for arginine into a -TGT that codes for cysteine a t amino acid 636 of the mature GPIIIa polypeptide chain. The presence of an additional cysteine within the cysteine-rich region of GPIIIa is likely to be responsible for the formation of the Sr" epitope.
Biochemical Properties of Recombinant GPIIIa Allelic Forms-To examine directly whether the Arg"" --j Cys polymorphism actually controls the formation of the SP antigenic determinant, we constructed an allele-specific recombinant form of GPIIIa that differs only by the presence of T a t nucleotide 2004 and analyzed the Cys""" protein product in a mammalian cell transfection system. Whereas PIA' alloantibody bound equally well to either the Arg6"" or CYS"~" isoforms of the GPIIIa molecule (Fig. 3B, compare lanes 2 and 3 with lanes 4  and 5), anti-Sr" reacted only with the Cysfisfi form (Fig. 3A).
These results demonstrate that amino acid 636 is directly involved in the expression of the Sr epitopes. Interestingly, the recombinant isoform (lane 4 ) migrated slightly slower in SDS-PAGE than the recombinant wild-type Argfi3" form (lane 3 ) , even under reducing conditions. Identical results were obtained using two independently constructed clones (lanes 2 and 5 ).
It is possible that the presence of a n additional cysteine residue in the S P allelic form of GPIIIa could have altered the conformation of the molecule such that the glycosylation pattern would also be affected. To examine whether the altered mobility of the Cys6"" isoform was unique only to the recombinant COS-7 cell product, or might be intrinsically present in Sf' positive versus SF' negative platelets, biotin surface-labeled platelets were subjected to immunoprecipitation analysis. As shown in Fig. 4, GPIIIa molecules derived from the platelets of an Sr" positive (heterozygous) individual migrated as a doublet (lane 1 ), corresponding to the wild-type (lower band) and Sr" (upper band) allelic forms. As predicted, only the upper GPIIIa band was reactive with anti-SP alloantisera (lane 2 ) , confirming the results obtained above using recombinant isoforms produced in COS cells (Fig. 3). GPIIIa derived from wild-type Sr" negative platelets (lane 3 ), was present as a single band, as expected, and was not reactive with anti-Sr" antibodies (lane 4 ) . After deglycosylation with Endoglycosidase H, however, both GPIIIa allelic forms migrated with the same mobility (lane 5), indicating that the molecular size polymorphism observed in the S P positive allele of GPIIIa is due to a variable degree of N-linked glycosylation, most likely the presence of an additional single high mannose moiety. In contrast to continued reactivity with anti-PIA' alloantibody (lane 5, and Ref. 34), however, deglycosylated GPIIIa failed to react with anti-SF alloantibodies (lane 6). These data suggest that the Sra antigenic determinant is, at least in part, dependent on the presence of one or more of the high-mannose carbohydrate residues that are known to comprise approximately 15% of the molecular mass of the GPIIIa molecule. Whether or not the additional carbohydrate residue added as a result of the Arg""" -' Cys amino acid substitution forms part of the alloantibodycombining site remains to be determined.
Effect of a Free Thiol Group on GPIIIa Expression a n d Function in SF Positive Platelets-In order to verify biochemically the presence of a n additional, unpaired cysteine residue in GPIIIa derived from Sr" positive individuals, intact platelets were incubated with tritiated NEM to derivatize accessible free thiol groups, solubilized in Triton X-100, and then immunoprecipitated with the anti-GPIIIa murine monoclonal antibody, Gi5. Since the P-subunit of the GPIb complex has previously been shown to contain a free sulfhydryl group (26), aliquots of these same lysates were subjected to immunoprecipitation analysis using the anti-GPIbp-specific monoclonal antibody Gi22, which served to validate the specificity of the labeling reaction. As shown in Fig. 5 , GPIb, was visualized by autora- aggregation, platelets from a Sr" positive individual were compared with Sr" negative (control) platelets in standard platelet aggregation assay. Fig. 6 shows the platelet aggregation responses obtained using PRP from 2 Sr" positive individuals compared with a single Sr" negative healthy donor. As shown, addition of M ADP resulted in a completely normal aggregation response in platelets derived from both Sr" positive individuals. Similar findings were obtained using collagen (5 pgl ml) or ristocetin (1.5 mg/ml) as agonists (not shown). These results suggest that neither the expression nor function of Sr" positive platelets are adversely affected by the presence of an unpaired cysteine residue on the cell surface.

DISCUSSION
Although the importance of integrins in mediating cell-cell and cell-extracellular matrix interactions is well recognized, the fact that a number of integrin a-and P-subunits may be encoded by multiple allelic forms is not well appreciated.

GPIIIa ( P 3 ) is the most polymorphic integrin subunit in man,
and is most frequently responsible for eliciting an alloimmune response leading to thrombocytopenia and perhaps more widespread damage within the vasculature. To date, six different genetic variants of the GPIIIa molecule have been identified at the serological and molecular biological level, and these are summarized in Table 111. The PIA1 allelic form is by far the most common, with a gene frequency of nearly 85% within the Caucasian population and differs from the second most common form of P3, PIA2, by a single Leu33 + Pro amino acid substitution. Other alleles of GPIIIa, including the Sr" isoform of this study, are much less frequently represented, but also appear to have arisen by point mutations of the PIA' ancestral allele.
The Sr" alloantigen was originally described serologically in a case of alloimmune thrombocytopenia, and subsequent immunochemical studies localized the Sr" antigen to the 68-kDa membrane-bound protease-resistant core of GPIIIa (18), either on the amino-terminal 121 amino acid residues or on the carboxyl-terminal segment bounded by residues 349-692. By sequencing cDNA derived from PCR-amplified platelet mRNA (lo), we have localized the polymorphism underlying the Sr" polymorphism to a C2004 + T substitution, which in turn results in an Arg ~> Cys polymorphism at amino acid 636, just proximal to the membrane from the cysteine-rich region of the molecule. Using mammalian transfection techniques, we were able t o demonstrate that the additional cysteine residue at position 636 of GPIIIa controls the formation of the Sr" alloantigenic epitope. Other three-dimensional structural features of GPIIIa appear to be required as well, however, since both disulfide bond reduction (18) as well as deglycosylation (Fig. 4) abolish the presentation of the Sr" epitope.
Interestingly, the (Sr") allelic form of GPIIIa was found to have a higher apparent molecular weight than the wild-type Arg636 form. This molecular weight polymorphism appears to be attributable to differential glycosylation of the altered conformer of GPIIIa induced by the presence of an positive individuals (A and B ) compared with a normal Sra negative unpaired cysteine residue, since deglycosylation, but not disulfide bond reduction, allowed the Sr" form of GPIIIa to co-migrate on SDS-PAGE with the wild-type allele. At the present time, we do not know whether the glycosylation differences observed are due to altered trimming of the high-mannose carbohydrate side chains, or to the presence of additional carbohydrate moieties to normally cryptic, unmodified sites. Following incubation with [3H1NEM, label became incorporated into GPIIIa from Sr" positive platelets, confirming at a biochemical level the presence of a free sulfhydryl group in this rare allelic form of GPIIIa. Parallel experiments using Sr" negative platelets, however, revealed no incorporation of [3H]NEM into platelet GPIIIa, confirming the structural analysis of Calvete et al. (6). The finding of a naturally occurring isoform of GPIIIa having 57 cysteine residues is remarkable, since both the number and the position of the 56 cysteine residues are absolutely conserved in the integrin P-subunit family ranging from Drosophila to man (36, 37). One might have expected the presence of an additional, unpaired cysteine to have a deleterious effect on either expression of this integrin P-subunit or on platelet function, resembling the effects of similar mutations that have been shown to be responsible for the congenital platelet functional disorder, Glanzmann thrombasthenia (5,7,38). However, not only was expression of the Sr" allelic form of GPIIIa quantitatively normal, Sr" positive platelets aggregated to the same extent as wild-type Sr" negative platelets. Moreover, individuals carrying the Sr" allele also show no obvious hemostatic, immunologic, or vascular abnormalities, suggesting that the adhesive functions of cells expressing the vitronectin receptor a& are also unaffected. Examination of the effects on expression and function of other molecular variations and alloantigenic forms of platelet surface receptors should continue to provide insights into the structural features of these molecules that influence biosynthesis, trafficking, and ligand binding capacity. cal assistance of Micaela Bohringer. We are also grateful to David Wil-