Dr Makoto Handa Blood Centre, School of Medicine, Keio University, 35 Shinanomachi, Shinjuku-ku, Tokyo 160, Japan.
We analysed the molecular basis of Glanzmann thrombasthenia (GT) in four Japanese patients with type I or type II disease. Polymerase chain reaction (PCR) and subsequent direct sequencing of platelet RNA and genomic DNA revealed three single nucleotide substitutions of the αIIb gene, which were confirmed by allele-specific PCR or restriction analysis. One patient with type I GT had a T to C base substitution in exon 11 resulting in a Phe (TTT)-289 to Ser (TCT) mutation (F289S) of the subunit. Another type I patient had a G to A base substitution in exon 12 resulting in a Glu (GAA)-324 to Lys (AAA) mutation (E324K). Interestingly, two unrelated patients with type II GT shared an A to C base substitution in exon 23, a region previously not associated with GT, resulting in a Gln (CAA)-747 to Pro (CCA) mutation (Q747P). To analyse the effects of these mutations on αIIbβ3 surface expression, the wild-type αIIb cDNA or mutant αIIb cDNAs were transfected into Chinese hamster ovary (CHO) cells together with a wild-type β3 cDNA. Flow cytometric analysis using an anti-αIIbβ3 complex antibody revealed that 50.6% of CHO cells with wild-type αIIbβ3 expressed complexes, whereas only 1.6%, 7.7% and 31.3% of cells, with αIIb(F289S)β3, αIIb(E324K)β3 and αIIb(Q747P)β3 expressed complexes, respectively. Our data indicate that these three novel point mutations in the αIIb subunit may hamper surface expression of the αIIbβ3 complex, thus resulting in the quantitative GT phenotypes of platelets from these patients.
The platelet membrane glycoprotein (GP) IIb–IIIa complex plays an essential role in primary haemostasis at sites of vascular wall injury as a receptor for at least four RGD-bearing adhesive ligands including fibrinogen, von Willebrand factor, fibronectin and vitronectin. The complex, known as αIIbβ3, is a member of the integrin family and consists of two distinct membrane glycoproteins, GPIIb (αIIb) and GPIIIa (β3), which are non-covalently bound in a divalent-cation manner. Biosynthetic processes, including maturation and subsequent surface expression of the complex molecule, are regulated in a coordinated fashion by the αβ subunit association.
Glanzmann thrombasthenia (GT) is an autosomal recessive bleeding disorder that arises from quantitative or qualitative abnormalities of either subunit of the complex. GT is characterized by the failure of platelets to bind fibrinogen and form aggregates in response to physiologic agonists such as ADP, thrombin, epinephrine or collagen. This disorder has been subclassified into three categories: (a) type I patients with a severe αIIbβ3 deficiency (< 5% of normal), (b) type II patients with a moderate deficiency (5–20% of normal), and (c) variants with half normal to normal amounts of dysfunctional αIIbβ3 complexes ( George et al, 1990 ).
Following the elucidation of the structural organization of the αIIb and β3 genes ( Poncz et al, 1987 ; Heidenreich et al, 1990 ; Zimrin et al, 1988 , 1990), a number of genetic defects associated with GT have been identified and include rearrangements, major or minor deletions, insertions, abnormal mRNA splicing, and nonsense or missense mutations (for review see Bray, 1994; French & Coller, 1997). Among these, the missense mutations in particular have provided important information about the biology of αIIbβ3. Three missense mutations in the αIIb gene corresponding to amino acid substitutions of Gly-273 to Asp (G273D), Gly-418 to Asp (G418D), and Arg-327 to His (R327H) prevent surface expression of the αIIbβ3 complex, thus defining regions involved in the intracellular trafficking of the complex molecule including maturation and subsequent translocation to the plasma membrane ( Poncz et al, 1994 ; Wilcox et al, 1994 , 1995) (These mutations were numbered using different methods: for Gly-273 the numbering commenced at the first amino terminal amino acid, a methionine, of αIIb; for Gly-418 and Arg-327 at the first amino terminal amino acid, a leucine, of the matured αIIb.) In addition, four separate missense mutations which result in variant GT have been described in the β3 gene ( Loftus et al, 1990 ; Bajt et al, 1992 ; Lanza et al, 1992 ; Chen et al, 1992). The Asp-119 to Tyr (D119Y), Arg-214 to Gln (R214Q) and Arg-214 to Trp (R214W) mutations define regions involved in ligand binding, and a Ser-752 to Pro (S752P) mutation in the cytoplasmic domain affects signalling necessary for receptor activation.
Herein, we report the molecular defects in four Japanese patients with type I or type II GT. By means of direct sequencing of the αIIb and β3 genes, we identified three novel point mutations in the αIIb gene. To analyse the effects of these three mutations on surface expression of the αIIbβ3 complex, a wild-type αIIb cDNA or mutant αIIb cDNAs were transfected into Chinese hamster ovary (CHO) cells together with a wild-type β3 cDNA. The surface expression of these mutant αIIbβ3 was found to be significantly reduced when compared with that of the wild-type. Hence, these three novel missense mutations appear to be responsible for the quantitative GT phenotypes of our four patients.
* Examined by the method of Castaldi et al (1996) using PRP, normal range 50–95% (n = 10). † 51Cr-labelled platelet adhesion to fibrinogen by the method of Nagai et al (1993 ), data presented are the percentage of adherent platelets from patients compared to a normal control.‡ Reactivity of platelets with anti-αIIbβ3 complex antibodies (AP-2 and LJ-P9) was analysed by flow cytometry, data presented are the percentage of mean fluorescence intensity obtained from patient's platelets compared to a normal control, after subtracting the value obtained with a subtype (IgG1)-specific control antibody IS11-1 (Takara Shuzo, Otsu, Japan).
Isolation and purification of platelet RNA and genomic DNA
PRP from patients and normal controls were prepared by collecting venous blood into a one-fifth final volume of 3.14% sodium citrate, followed by incubation with prostaglandin E1 (final concentration of 50 ng/ml) for 10 min at room temperature (RT) and then centrifuged at 120 g for 15 min. Platelet pellets were obtained by further centrifuging the PRP at 1200 g for 10 min at RT. Platelet RNA was prepared from the platelet pellet by the modified method of Newman et al (1988 ). Briefly, each platelet pellet was solubilized in a GIP buffer consisting of 4 M guanidinium isothiocyanate, 25 m M sodium citrate, 0.1 M 2-mercaptoethanol, 0.5% N-lauroylsarcosine (Sarkosyl) and 0.1% antifoam-A in siliconized and diethylpyrocarbonate (DEPC)-treated Eppendorf tubes. Total RNA was extracted from the solubilized platelet pellet with chloroform/isoamyl alcohol, and precipitated with isopropanol. The platelet RNA was dissolved in 15 μl DEPC-treated water and stored at −20°C until use.
Genomic DNA was isolated from peripheral blood mononuclear cells using a Qiagen kit (Qiagen Inc., Chatsworth, Calif.) according to the manufacturer's instructions.
Table 2. Table II. Sequences and locations of primers.
cDNA and genomic DNA of αIIb are numbered according to Poncz et al (1987 ) and Heidenreich et al (1990 ), respectively. cDNA and genomic DNA of β3 are numbered according to Zimrin et al (1988 , 1990). +: sense primer; −: anti-sense primer. S2, S3, S4, S6, S13 and S15 are sense primers; S1, S5, S7, S8, S9, S10, S11, S12 and S14 are anti-sense primers.
Genomic DNA was also amplified by PCR. The amplified products were checked on a 1.4% agarose gel. Amplified DNAs were purified and recovered from Seakem GTG gels (FMC Bio Products, Rockland, Maine) by the glass-milk procedure (Bio 101, La Jolla, Calif.). Nucleotide sequences were determined by the direct sequencing method with a fmol DNA sequencing system kit (Promega, Madison, Wis.) according to the manufacturer's instructions. The following primers were used for cDNA sequencing: A+, S1, S2, S3, S4, S5, S6, C−, S7, S8, S9, F−, S10, S11, S12, S13, S14 and S15. The sequencing reaction products were separated on 6% acrylamide/7 M urea gels and assayed by autoradiography.
Allele-specific PCRs and restriction analysis were used to confirm the sequence results. Genomic DNAs from case 1 and a control were amplified using a common anti-sense oligonucleotide primer J− and either of two sense oligonucleotide primers, a wild-type primer Jw+ (5′-GCAGATGGCGTCGTATTT-3′, position 943–960 in αIIb cDNA, position 5687–5704 in αIIb genomic DNA) ending in the normal T or a mutant primer Jm+ (5′-GCAGATGGCGTCGTATTC-3′) ending in the mutated C. Meanwhile, cDNAs from case 1 and a control were amplified using a common anti-sense oligonucleotide primer F and either of Jw+ or Jm+. Genomic DNAs from case 2 and a control were amplified by PCR using a sense oligonucleotide primer K+ and an anti-sense oligonucleotide primer J−. The amplified products were digested with MscI (TGG/CCG; Toyobo, Osaka, Japan) and analysed on a 1.4% agarose gel. Genomic DNAs from case 3, case 4, and a control were amplified by PCR using a common anti-sense oligonucleotide primer I− and either of two sense oligonucleotide primers, a wild-type primer Iw+ (5′-GGTCCGGGCAGAGGCCCA-3′, position 13316–13334 in αIIb genomic DNA) ending in the normal A or a mutant primer Im− (5′-GGTCCGGGCAGAGGCCCC-3′) ending in the mutated C.
Expression of recombinant αIIbβ3
Wild-type human αIIb and human β3 cDNAs were subcloned into a pBJ-1 vector as previously described ( Kamata et al, 1996 ). F289S, E324K and Q747P mutations were introduced into αIIb cDNA using an unique site elimination method ( Deng & Nickoloff, 1992). The presence of a mutation was verified by DNA sequencing.
For transient expression of αIIbβ3, 50 μg of wild-type or mutant αIIb cDNA construct was transfected into CHO-K1 cells (107 cells) together with 50 μg of wild-type β3 cDNA by electroporation. In separate experiments αIIbβ3 cDNAs were co-transfected with 20 μg of green fluorescent protein (GFP) expression vector pEGFP-C1 (Clontech, Palo Alto, Calif.). Transfected cells were maintained in Dulbecco's modified eagle's medium supplemented with 10% fetal calf serum at 37°C in 6% CO2 for 48 h. The surface expression of wild-type or mutant αIIb was examined by flow cytometry using FACScan (Becton Dickinson) as described previously ( Takada et al, 1992 ).
Clonal cell line stably expressing αIIb(Q747P)β3 were obtained as previously described ( Kamata et al, 1996 ). Briefly, mutant αIIb cDNA and wild-type β3 cDNA were transfected together with a neomycin-resistance gene by electroporation. After 48 h, the cells were transferred to medium containing 700 μg/ml of G418 (Gibco BRL, Gaithersburg, Md.). After 10–14 d, the G418-resistant colonies were harvested and cells expressing αIIbβ3 mutants were cloned by cell sorting in FACStar (Becton Dickinson). The anti-β3 antibody mAb 15 ( Frelinger et al, 1990 ) was the generous gift of M. H. Ginsberg (the Scripps Research Institute, La Jolla, Calif.), the anti-αIIbβ3 antibody 2G12 ( Woods et al, 1984 ) that of V. Woods (University of California San Diego, Calif.) and an anti-αIIb antibody PL98DF6 ( Ylänne et al, 1988 ) that of J. Ylänne (University of Helsinki, Helsinki, Finland).
αIIbβ3 cDNA and genomic DNA sequence analysis
To analyse the molecular defects in four Japanese patients with type I or II GT, the nucleotide sequences of the αIIb and β3 coding regions were determined by direct sequencing of RT-PCR products from platelet RNA. The sequences in the coding regions were identical to those of the published cDNA sequences of αIIb ( Poncz et al, 1987 ) and β3 ( Zimrin et al, 1988 ) except for three nucleotide substitutions in αIIb.
As shown in Fig 1A, the first mutation was a T to C base substitution at position 960 of αIIb cDNA (cDNA numbering based on Poncz et al, 1987 ) in case 1. This substitution results in a Phe-289 to Ser amino acid substitution (F289S). As shown in Fig 2A, the second mutation was a G to A base substitution at position 1064 of αIIb in case 2. This substitution results in a Glu-324 to Lys amino acid substitution (E324K). The third change was an A to C base substitution at position 2334 of αIIb cDNA in case 3 (Fig 3A). Interestingly, the same mutation was seen in case 4 (data not shown). This substitution results in a Gln-747 to Pro amino acid substitution (Q747P).
Since three nucleotide substitutions were found in four of the patients' αIIb cDNAs, direct sequencing of PCR products derived from genomic DNAs from these patients was performed for αIIb gene analysis. In case 1 both the mutated allele C and the wild-type allele T were detected at the corresponding position 5705 (genomic DNA numbering based on Heidenreich et al, 1990 ) of genomic αIIb exon 11 (Fig 1B), although only the mutated allele was detected in αIIb cDNA (Fig 1A). The G to A substitution in case 2 was also detected at the corresponding position 6000 of genomic αIIb exon 12 (Fig 2B). No wild-type allele G was detected at position 6000. The A to C substitution in case 3 (Fig 3B) and case 4 (data not shown) was also detected at the corresponding position 13 334 of genomic αIIb exon 23. No wild-type allele A was detected at position 13 334 in either case.
Allele-specific PCR and restriction analysis
To confirm the sequence results, allele-specific PCR and restriction analysis were performed. As shown in Fig 1C, a DNA fragment of the expected size (809 bp) was amplified from control cDNA by the wild-type primer (Jw+), but not by the mutant primer (Jm+) (left gel). A DNA fragment of the expected size was amplified from case 1 cDNA by the mutant primer (Jm+), but only a trace amount of product was seen with the wild-type primer (Jw+). A DNA fragment of the expected size (481 bp) was amplified from control genomic DNA by Jw+, but not by Jm+ (right gel). However, from case 1 genomic DNA, DNA fragments of identical size were amplified by Jm+ and Jw+. These data confirmed the sequencing results and showed that case 1 has the mutated allele C and the wild-type allele T at position 5705 of the αIIb gene. However, only the mutated allele was clearly detected in the amplified αIIb cDNA because the αIIb mRNA from the allele with the wild-type allele T at position 5705 of the gene could not be amplified easily under the present study conditions.
Since the G to A substitution at position 6000 of the case 2 αIIb gene creates a new MscI restriction site, genomic DNA from case 2 and a control was amplified using the J− and K+ oligonucleotide primers and digested with MscI. As shown in Fig 2C, the 507 bp PCR product from the control remained uncut, while the same-size product of case 2 containing an A at position 6000 of the αIIb gene was cut into a 336 bp and 171 bp product with MscI. These data demonstrated case 2 to be homozygous for an A at position 6000 of the αIIb gene.
As shown in Fig 3C, a DNA fragment of the expected size (532 bp) was amplified from control genomic DNA by the wild-type oligonucleotide primer (Iw+), but not by the mutant primer (Im+). Conversely, DNA fragments of identical size were amplified from genomic DNA of cases 3 and 4 by the mutant oligonucleotide primer (Im+), but not by the wild-type primer (Iw+). These data confirmed the sequencing results showing that cases 3 and 4 were homozygous for a C at position 13 334 of the αIIb gene. In order to determine whether this mutation was pathologic or merely a polymorphism, allele-specific PCR was performed in normal individuals. No mutations were detected in a total of 106 alleles from the normal Japanese population, suggesting that this was unlikely to be a polymorphism (data not shown).
Expression of wild-type and mutant αIIb in CHO cells
To analyse the effects of the three mutations detected in αIIb (F289S, E324K, Q747P) on surface expression of the αIIbβ3 complex, wild-type or mutant αIIb cDNAs were transfected into Chinese hamster ovary (CHO) cells together with wild-type β3 cDNA. Wild-type and mutant αIIbβ3 complexes were transiently expressed on CHO cells and the reactivities of the mutant αIIbβ3 complex to the anti-αIIb monoclonal antibody PL98D6, the anti-β3 monoclonal antibody mAb15, and the anti-αIIbβ3 complex monoclonal antibody 2G12 were examined by flow cytometry (Fig 4). Non-transfected parental CHO cells were 0.91%, 0.75%, 0.57% and 0.51% positive for control mouse IgG, PL98DF6, mAb15 and 2G12, respectively. 42.4% and 50.6% of CHO cells expressing the wild-type αIIbβ3 were positive for PL98D6 and 2G12, respectively. Of the CHO cells expressing αIIb(F289S)β3, αIIb(E324K)β3 and αIIb(Q747P)β3 complexes, 4.1%, 15.6% and 29.5% were positive for PL98D6 and 1.6%, 7.7% and 31.3% for 2G12, respectively. To monitor transfection efficiency, we co-transfected a GFP expression vector together with αIIbβ3 cDNAs into CHO cells in separate experiments. The mutant αIIbβ3-transfected CHO cells showed comparable control GFP expression to wild-type αIIbβ3-transfected cells, showing equivalent transfection efficiency (data not shown). These results suggest that the difference in the surface expression of mutant αIIbβ3 was a direct result of the mutations in the αIIb cDNAs rather than of any difference in transfection efficiency.
As it is unlikely that a single amino acid substitution could destroy the PL98D6 or 2G12 epitope, the reduced reactivities of the CHO cells expressing mutant αIIbβ3 with the antibodies may reflect defective surface expression of the αIIb(F289S)β3, αIIb(E324K)β3 or αIIb(Q747P)β3 complex. Therefore these three mutations may affect the biosynthetic processing of the complex resulting in impaired surface expression of the molecule. Meanwhile, CHO cells expressing the mutated αIIb(F289S)β3, αIIb(E324K)β3 and αIIb(Q747P)β3 complexes had nearly the same reactivity with the anti-β3 antibody mAb15 as those cells which expressed the wild-type αIIbβ3 complex. This reactivity was attributable to the presence of wild-type β3 on the cell surface in association with an endogenous hamster αv subunit which reacted with the anti-β3 antibody mAb 15. Since 61% of CHO cells expressed the αIIb(Q747P)β3 complex, compared with those expressing the wild-type αIIbβ3 complex, the receptor function of the mutant complex was analysed. Fluorescence-labelled fibrinogen binding to the mutant protein induced by the activating antibody PT25-2 ( Tokuhira et al, 1996 ) appeared to be unaffected (data not shown), indicating that the mutation may cause a quantitative but not a qualitative abnormality of the receptor. This is consistent with the observation that the patients manifested the type II GT phenotype.
As summarized in 3 Table III, we identified three novel single point mutations in the αIIb subunit (F289S in case 1, E324K in case 2, and Q747P in cases 3 and 4), which may result in loss of surface expression of the αIIbβ3 complex, and produced the thrombasthenic phenotypes in these four Japanese patients.
Table 3. Table III. Summary of patient studies.
* Compound heterozygous, second mutation not determined.† The reactivity of the mutant αIIbβ3 complex, transiently expressed on CHO cells, with anti-complex antibody 2G12 was examined by flow cytometry and data presented are the percentage of positive cells, i.e. those expressing the mutant complex, as compared with those expressing the wild-type complex.
Missense mutations reported thus far in αIIb from GT patients were all located at or in proximity to the four high-affinity Ca2+ binding domains of the amino-terminal portion of the subunit (G273D, G418D and R327H; Fig 5). Transfection studies have clearly shown marked deterioration of the intracellular trafficking of these mutated complexes in CHO cells. Therefore these mutations may have a major influence on the conformation of the complex maintained by the αβ subunit association necessary for its normal in situ maturation and surface expression, thereby producing quantitative (type I or II) phenotypes of GT.
Of the three mutations described herein, F289S and E324K, derived from two type I patients (cases 1 and 2), were also found to be localized to these amino-terminal Ca2+ binding domains (Fig 5). In fact, consistent with the aforementioned hypothesis, CHO cells transfected with the mutant αIIb cDNA and a wild-type β3 cDNA were defective in expression of the mutant αIIbβ3 complex on the cell surface. Although we have not performed a pulse-chase analysis, other investigators have clearly demonstrated biosynthetic blockade during maturation processes in the endoplasmic reticulum induced by these three missense mutations of the αIIb subunit ( Poncz et al, 1994 ; Wilcox et al, 1994 , 1995). They concluded that G273D, G418D and R327H mutations prevented the intracellular transport of mutated αIIbβ3 complexes to the cell surface. Wilcox et al (1994 ) postulated that the G to D substitution, from a neutral amino acid (Gly) to a negatively charged residue (Asp), perturbs the conformation of αIIb and prevents the proper folding required for intracellular transport of the complex. Since the E324K mutation involved the substitution of a negatively charged amino acid (Glu) with a positively charged amino acid (Lys) and the F289S mutation caused a change from a hydrophobic residue (Phe) to a hydrophilic residue (Ser), these mutations may potentially lead to a drastic conformational change in the αIIb molecule. Moreover, it is of particular interest that the E324K mutation results in the type I GT phenotype while the R327H mutation results in the type II GT phenotype, although the two residues are positioned in proximity. In fact, the R327H mutation preserves an existing positive charge and produces only a partial biosynthetic blockade ( Wilcox et al, 1995 ). Therefore the magnitude of conformational change rendered by these amino acid substitutions may account for the phenotypic (quantitative) differences between these two amino acid substitutions which occur only two residues apart in the αIIb molecule.
Sequence analysis and allele-specific PCR revealed case 1 to be heterozygous for a T and a C at position 5705 of genomic αIIb, but the αIIb mRNA from the allele having the wild-type T at this position could not be amplified under the current study conditions. One possible explanation is that the allele having the wild-type T at position 5705 of genomic αIIb may have deletions at this position of the sense or antisense oligonucleotide primers necessary for mRNA amplification. We amplified two fragments of αIIb genomic DNA containing a sequence in which a sense or antisense primer was located and confirmed the PCR products of case 1 and a control to be exactly the same size (data not shown). Therefore this possibility is rather unlikely. Another possibility is that the transcriptional level of the allele having the wild-type T at this position is decreased. In this situation the αIIb mRNA from the allele having the wild-type T at position 5705 of genomic αIIb would clearly not be amplifiable. It is generally assumed that nonsense mutations and frameshift mutations which induce a premature stop codon lead to dramatically reduced mRNA levels ( Roos et al, 1996 ). In patients with GT, the transcript levels of mutant alleles with large gene rearrangements of β3 ( Li & Bray, 1993) or a nonsense mutation in αIIb of exon 17 ( Kato et al, 1992 ; Tomiyama et al, 1995 ) were shown to be absent or decreased. Since our screening for mutations was performed by direct sequencing of platelet mRNA, any mutation resulting in reduced levels of transcripts could have been missed, as Bray (1994) pointed out. Therefore, further analysis of genomic DNA is necessary to identify another potential genetic defect causing GT in case 1.
As mentioned above, all missense mutations in αIIb reported thus far have been identified within or in proximity to the four Ca2+ binding domains in αIIb. In fact, of our three mutations, F289S is located at the second Ca2+ binding domain and E324K between the second and third Ca2+ binding domains (Fig 5). Therefore Q747P is unique in that it is located at a position topologically distinct from and unrelated to the four Ca2+ binding domains (Fig 5). Since Gln-747 cannot be assigned to known functional domains of αIIb and is not a highly-conserved residue among integrin a subunits ( Poncz & Newman, 1990), we had to consider the possibility that this substitution was merely a polymorphism. However, a transfection study clearly excluded this possibility by demonstrating that CHO cells expressing the mutated cDNA expressed 31.3% of the complex while cells with the wild-type cDNA expressed 50.5%, a finding which is consistent with this mutation being responsible for the type II GT phenotype. This conclusion was also supported by the observation that the mutant complex expressed on the CHO cell surface was functional. Moreover, allele-specific PCR did not detect this mutation in 53 normal Japanese individuals, supporting the assumption that the mutation was pathologic. It is of interest to note that the Q747P mutation occurred in two patients (cases 1 and 2) from different families. This may either be due to independent mutational events at the position 747 or to the patients and their families being distantly related. Although the CpG dinucleotide is well known to be a hot spot for mutation ( Tomiyama et al, 1995 ; Roos et al, 1996 ), dinucleotides CC, GG, GC, CA and TC have been shown to be more frequently mutated in certain genetic diseases ( Roos et al, 1996 ). This might explain two unrelated Japanese GT patients sharing the same mutation, since the Q(CAA)747P(CCA) mutation occurred at the CA site. Allele-specific PCR, which confirmed the A to C base substitution at position 13 334 of the αIIb gene, will determine the frequencies of the mutation in Japanese population.
In conclusion, three novel missense mutations of the αIIb gene resulting in single amino acid substitutions, F289S, E324K and Q747P, were identified in four Japanese patients with GT. Transfection studies revealed that these mutations appear to cause defective expression of the αIIbβ3 complex on the plasma membrane surface, thereby producing the quantitative thrombasthenic phenotypes.
We thank T. J. Kunicki for providing information on oligonucleotide primer sequences and Drs M. H. Ginsberg, V. Woods, Z. M. Ruggeri, J. Ylänne and T. J. Kunicki for providing monoclonal antibodies. We also thank Y. Miyazaki, H. Nagai, K. Sato, Y. Ujiie and M. Yamamoto for their technical assistance and Drs S. Kitaguchi and S. Watanabe for their ongoing encouragement and helpful discussions.