R to Q Amino Acid Substitution in the GFFKR Sequence of the Cytoplasmic Domain of the Integrin _IIb Subunit in a Patient With a Glanzmann’s Thrombasthenia-Like Syndrome

GLANZMANN’S THROMBASTHENIA is an inherited bleeding disorder in which quantitative or qualitative defects of the fibrinogen receptor, α_IIbβ₃(the platelet glycoprotein [GP] IIb-IIIa complex), result in the absence of the aggregation response after platelet activation.1-3 A member of the integrin family, α_IIbβ₃ is formed through the noncovalent association of two subunits encoded by different genes, both of which are localized on chromosome 17.4 Binding studies with monoclonal antibodies (MoAbs) showed that about 50,000 copies of this receptor are present at the surface of resting platelets.5An internal pool of about the same size is associated with membranes of the surface-connected canalicular system, α-granules, and dense granules.6,7 After activation of platelets by physiological agonists such as adenosine diphosphate, a conformational change in the surface receptor enables binding of fibrinogen (or other adhesive proteins) and aggregation.8 With strong agonists such as thrombin, the internal pool is also mobilized, providing more α_IIbβ₃ complexes to participate in platelet-platelet interactions.9 

Transcription of the α_IIb and β₃ genes occurs in megakaryocytes and the two proteins are assembled and then glycosylated in the endoplasmic reticulum. Endoproteolytic cleavage of α_IIb in the Golgi apparatus leads to the separation of heavy and light chains which remain disulfide bonded.10 It is the light chain that carries the transmembrane and cytoplasmic domains of α_IIb. Post-translational modifications, including glycosylations and the association of the two glycoproteins into a heterodimeric complex, are necessary for cell-surface expression of the receptor.10,11Molecular abnormalities described in patients with Glanzmann’s thrombasthenia have provided information on structural domains essential for α_IIbβ₃ complex formation and expression. For example, two amino acid substitutions adjacent to the first and the fourth calcium binding sites of α_IIbaltered the conformation of the complex and prevented its transport to the golgi apparatus,12,13 whereas a mutation located between the second and the third calcium binding site greatly slowed the maturation kinetics.14 Further information on the functional domains of the complex have been provided by studies on variant forms of the disease. Thus, mutations in the extracellular domain of β₃ are associated with a failure of the constituted complex to bind fibrinogen when stimulated and a mutation in the cytoplasmic domain of β₃ leads to defective “inside-out” signaling.15-17 

We now report the molecular analysis of α_IIbβ₃ from a patient with an atypical form of Glanzmann’s thrombasthenia. As we previously reported, the patient’s platelets were characterized by (1) a total platelet α_IIbβ₃ content of less than 50% of normal levels and (2) an altered partition of the residual α_IIbβ₃ receptor within the different platelet membrane systems.18 Only approximately 15% of the normal levels of this receptor was found at the platelet surface, although the internal pool remained substantial. The residual receptor was functional and aggregation was observed after in vitro activation of platelets by physiological agonists, although the kinetics were slow and only small aggregates formed. The patient’s platelets did not bind fibrinogen spontaneously. Among the abnormalities that we have found for this patient is a heterozygous point mutation, G to A located at nucleotide 3078 of the cDNA (numbering according to Frachet et al19), leading to an R⁹⁹⁵ to Q amino acid substitution in the α_IIb cytoplasmic domain. Site-directed mutagenesis and transient expression in Cos-7 cells showed that the mutation was likely to be directly involved in the low surface expression on his platelets. In view of the purported role of the cytoplasmic domains of α_IIb and β₃ in controlling the activation state of the complex,20,21 we also prepared a stable transfected Chinese hamster ovary (CHO) cell line expressing the mutated receptor. Flow cytometry showed that the cells only bound appreciable amounts of PAC-1, a marker of the activated complex,22 in the presence of an α_IIbβ₃-activating antibody.

This is the first time that a mutation in the cytoplasmic domain of α_IIb has been reported in Glanzmann’s thrombasthenia. The interest of the R⁹⁹⁵ to Q substitution and the reduced cell-surface expression of α_IIbβ₃ is reinforced by its location in the GFFKR sequence which is conserved in the cytoplasmic domain of all integrin α subunits.

The case report and specific features that characterize the platelets of patient A.P., a young Italian man, were given by Hardisty et al.18 His father (F.P.) and mother (M.P.) were also studied. There is no evidence of consanguinity and no family history of bleeding. However, the patient has suffered from occasional bleeding since early childhood although blood transfusion has never been performed. Bleeding mostly occurred after knocks or small trauma and was not spontaneous. As detailed earlier, the main features of A.P.’s platelets are a total α_IIbβ₃ content that is about half that observed with normal platelets yet with a surface expression that is sparse compared with the size of the internal pool.18 The platelet aggregation response is poor but not absent and this is why he was originally referred to as having a “thrombasthenia-like” syndrome. His unactivated platelets did not bind fibrinogen spontaneously. Platelets from his parents aggregated normally, although the number of α_IIbβ₃ receptors, as determined in MoAb-binding studies, was lower than is usually seen: approximately 28,000 for his father’s platelets and approximately 20,000 for his mother’s platelets.18 Control donors were adult members of our hospital staff. Informed consent was obtained.

Genomic DNA from the propositus, both parents and healthy volunteers, was isolated from whole blood using the QIAmp blood kit (Qiagen, Chatsworth, CA) according to the manufacturer’s instructions. A total of 43 pairs of oligonucleotides allowed the amplification of each exon together with the intronic splicing signals. After 30 cycles of amplification, an aliquot of each product was loaded onto a 2% agarose gel.23 After electrophoresis, bands were visualized under ultraviolet light after ethidium bromide coloration.

SSCP analysis was performed using the Pharmacia Phast System and precast minigels (Pharmacia-Biotech, Saint-Quentin en Yvelines, France) as we previously described.23-25 The oligonucleotide primers for amplifying exon 3 of the β₃ gene and exons 21 and 26 of the α_IIb gene are given elsewhere.23,24 Genotyping for the HPA-1 and HPA-3 alloantigen systems was performed using an Allele-Specific Restriction Assay (ASRA) as previously described (also see below).23 24The primers used to amplify exon 30 of the α_IIb gene were^5′CAGCAAATCATCTGTATACCCT^3′ (sense, hybridizing in intron 29) and5′CCCAAAGCTTGGAGGCAACT^3′ (antisense, hybridizing downstream of exon 30). DNA amplification products exhibiting an altered migration pattern in the SSCP analysis were directly sequenced using a fentomol DNA sequencing kit (Promega-France, Lyon, France). To ensure that those mutations detected in heterozygous form were not artifacts of the sequencing methodology, we cloned the corresponding amplification products into the pGem-T vector (Promega-France) and sequenced 12 independent clones using the T7 Sequencing kit (Pharmacia-Biotech).

Site-directed mutagenesis was performed as before,25 using the pALTER-1 Vector system (Promega-France) in which we had previously cloned wild-type cDNA for α_IIb. A CTTCAAGCAGAACCGGC oligonucleotide was used to generate the α_IIb mutation (the mutated base is indicated in bold case). After selection, full-length cDNAs were entirely sequenced to be certain that only the desired mutations were present. Wild-type or mutated cDNAs for α_IIb and β₃ were subcloned between EcoRI and HindIII sites of the eukaryotic expression vector pSM as we previously constructed.25 Cos-7 cells, cultivated in Dulbecco’s modified Eagle’s medium (Life Technologies, Cergy-Pontoise, France) supplemented with 7% fetal calf serum, were transfected as follows: 3 × 10⁵ cells per 3.5-cm petri dish were seeded 24 hours before transfection which was performed during 4 hours with 3 μg of one plasmid or 1.5 μg of each plasmid (one carrying α_IIb cDNA and the other β3 cDNA) and Transfectam (Promega-France) according to the manufacturer’s conditions. Surface membrane expression of α_IIbβ₃ was analyzed by flow cytometry and total cell expression by Western blotting, both after 72 hours of culture.

In some experiments, Cos-7 cells were also cotransfected with pSV-β plasmid (Promega) coding for β-galactosidase. Transfection efficiency was measured on a fraction of the cells using an anti–β-galactosidase MoAb (Promega) after cell permeabilization using the Fix and Perm Cell Permeabilization Kit (Caltag Laboratories, Burlingame, CA) and flow cytometry.

Cos-7 cells were harvested after a 10-minute incubation at room temperature after the addition of phosphate-buffered saline (PBS) containing 5 mmol/L EDTA. The cells were washed twice with PBS-0.1% (wt/vol) bovine serum albumin (BSA) and counted. Volumes (0.5 mL) containing 400,000 cells in PBS-0.1% BSA was then incubated for 30 minutes under saturating conditions with EDU-3, an MoAb specific for the α_IIbβ₃ complex (a gift from Dr Villela, Barcelona, Spain); SZ22 (Immunotech, Marseille, France), specific for α_IIb; or XIIF9 (prepared by our laboratory), specific for β₃. Cells were washed twice with PBS-0.1% BSA before being incubated for 30 minutes in 0.5 mL PBS-BSA containing fluorescein isothiocyanate (FITC)-conjugated F(ab)₂ antibody to mouse IgG (Silenius Laboratories, Hawthorn, Australia) and analyzed after another round of centrifugation using a Becton-Dickinson FACscan (Becton-Dickinson, Le Pont de Claix, France).25 In each experiment 10,000 cells were counted across the laser. The limit of positivity was defined as the maximum observed fluorescence when Cos-7 cells were transfected by the expression vector alone.

Transfected Cos-7 cells were obtained as described above, sedimented, and resuspended in 50 μL of 10 mmol/L Tris-HCl, pH 7, containing 150 mmol/L NaCl, 3 mmol/L EDTA, and 2% (wt/vol) sodium dodecyl sulfate (SDS). Samples (50 μg) were loaded onto a 7% polyacrylamide gel and the proteins separated by electrophoresis before transfer to nitrocellulose membrane as described.26 The membranes were saturated for 2 hours with 5% (wt/vol) nonfat dry milk in 20 mmol/L Tris-HCl, 150 mmol/L NaCl, 0.05% (vol/vol) Tween 20, pH 7. After five washes in washing buffer (0.5% nonfat dry milk, 20 mmol/L Tris-HCl, 150 mmol/L NaCl, 0.05% Tween 20, pH 7), membranes were incubated for 2 hours with the murine MoAb XIIF9, specific for β₃, or SZ 22, specific for α_IIb. After five washes, membranes were incubated for 1 hour with peroxydase-coupled goat antibody to murine IgG (Jackson ImmunoResearch, West Grove, PA) and bound MoAb revealed using a chemiluminescence procedure (Amersham-France, Courtaboeuf, France). In each experiment, 1 μg of SDS-soluble proteins from platelets of a normal donor were electrophoresed as a control.

Wild-type and mutated α_IIb cDNAs were subcloned into the pCDN(−) expression vector (Invitrogen, San Diego, CA) betweenEcoRI and HindIII restriction sites. A previously characterized CHO cell line stably expressing the β₃subunit was transfected using pCDN wild-type or mutated α_IIb cDNA using lipofectamine as previously described.27 Briefly, after neomycin selection, cell membrane expression of α_IIbβ₃ was assessed using MoAbs P2 and EDU-3, both specific for the α_IIbβ₃ complex. A stable cell line maximally expressing the mutated α_IIbβ₃ was subcloned. The affinity state of the complex was analyzed in these cells using FITC-labeled PAC-1 (Becton-Dickinson), a murine IgM specific for the activated complex.22 GP IIb-IIIa expression was controlled using PerCP-conjugated anti-β₃(Becton-Dickinson) detected with an FL3 threshold gate as recommended by the manufacturer. In each experiment, 4 × 10⁵cells were incubated for 45 minutes in the presence of FITC–PAC-1 (15 μL of the solution provided by the manufacturer) with or without 2 μmol/L of the MoAb anti-LIBS6 (a generous gift of Dr Mark Ginsberg, Scripps Research Institute, La Jolla, CA) as described.28Anti-LIBS6 is an activating MoAb that binds directly to the complex. Binding of PAC-1 was then assessed by flow cytometry, and 5,000 cells were analyzed. Results for the subcloned cell line expressing the mutated complex were compared with those obtained for CHO cells transfected with wild-type α_IIbβ_3. The specificity of PAC-1 binding was also assessed in the presence of 1 mmol/L RGDS peptide (Sigma Chemical Co, St Louis, MO). On occasion, results were expressed as the activation index (AI) defined as 100 × (F_o − F_r)/(F_oLIBS6 − F_r LIBS6), where F_o is the median fluorescence intensity (MFI) of PAC-1 binding and F_r is the MFI of PAC-1 binding in the presence of RGDS. F_o LIBS6 is the MFI of PAC-1 binding in the presence of 2 μmol/L of this antibody and F_r is the MFI of PAC-1 binding in the presence of anti-LIBS6 and the competitive inhibitor.20 

RT-PCR was performed on platelet lysates from the patient, his father, and his mother prepared according to a procedure we recently described.29 Genotyping of HPA-1 (PlA) and HPA-3 (Bak) systems was performed on PCR products amplified as already described30,31 and using HpaII and FokI restriction enzymes, respectively. A TaqI polymorphism was determined using the TaqI restriction enzyme.32Mutations located in exon 30 of the α_IIb gene were determined both after direct sequencing of amplification products and after cloning and sequencing amplification products from cDNAs. In this case, 12 independent clones were sequenced for each family member using a T₇ sequencing kit (Pharmacia-Biotech).

We designed 43 pairs of oligonucleotides allowing the specific amplification of all exons including splice sites of both the α_IIb and β₃ genes, and developed a nonradioactive SSCP procedure to analyze each amplification product. For the patient (A.P.), 4 out of 43 DNA amplification products showed a SSCP profile different to that observed for the control. Three of the observed differences in the SSCP patterns were caused by known polymorphisms (Fig 1A). The first of these involved exon 3 of the β₃ gene and was due to the HPA-1 polymorphism.23 As summarized in Table 1, the SSCP patterns for this exon showed that the patient and his father were heterozygous HPA-1a/HPA-1b, whereas his mother was homozygous HPA-1a/HPA-1a. ASRA performed usingHpaII confirmed these genotypes. In the same manner, the differences observed for the SSCP patterns for amplification products carrying exons 21 and 26 of the α_IIb gene (Fig 1A) were caused by two polymorphisms that we have previously described as being bilaterally linked: HPA-3 and a 9-bp deletion (Del) in intron 21.24 SSCP analysis showed that both A.P. and his mother were heterozygous HPA-3a/HPA-3b and Del⁻/Del⁺,whereas his father was homozygous for the HPA-3a and Del⁻ polymorphisms (Table 1). Genotyping for HPA-3 was confirmed using the FokI restriction enzyme.

In contrast to these results, the pattern observed for the amplification product carrying exon 30 of the α_IIb gene of the propositus was new to us (Fig 1B). Up to now, no polymorphism has been reported in this region of the α_IIb gene. As is shown in Fig 1B, not only was the SSCP pattern for this amplification product different for the patient, but it was also different for both parents. In summary, band a was present for the control and for both parents but absent for the patient (A.P.), band b was present only for the propositus and his mother, and band c was restricted to the propositus and his father (F.P.). Direct DNA sequencing of PCR products showed that the differences between A.P. and the control were in fact due to two different base pair substitutions. The patient was heterozygous for both substitutions, the first, a C to T change in position 3064 of the cDNA, and the second, a G to A change in position 3078 of the cDNA (Fig 2). These mutations were detected in a heterozygous state in his parent’s DNA; the first was possessed by his mother and the second by his father (data not shown). Although the C to T transition did not induce an amino acid change, the G to A transition inherited from his father led to an amino acid substitution, R to Q in position 995 of the α_IIbprotein. Significantly, this amino acid substitution was located in the GFFKR sequence conserved in all cytoplasmic tails of integrin α subunits (see Discussion).

The G to A change responsible for the R to Q substitution in position 995 was introduced into wild-type α_IIb cDNA by site-directed mutagenesis. After DNA sequencing to ensure that only this alteration was present, and subcloning into the eukaryotic expression vector pSM, Cos-7 cells were cotransfected with wild-type β₃ cDNA and wild-type or mutated α_IIb cDNA. Cell-surface expression of the complex was analyzed by flow cytometry 72 hours later. Using EDU3, an MoAb specific for the α_IIbβ₃ complex, the percentage of positive Cos-7 cells was 13% when cotransfected with the two wild-type cDNAs and 5.6% for cells cotransfected with wild-type β₃ cDNA and mutated α_IIb cDNA (Fig3). In the illustrated experiment, the MFI for positive cells expressing complexes composed of mutated α_IIb and wild-type β₃ was 192, a value that was considerably lower than the MFI of 422 observed for cells expressing the wild-type complex (these results were typical of those obtained in three separate experiments). Similar findings were also obtained using XIIF9, an MoAb specific for the β₃ subunit (data not shown). Nevertheless, because the definition of the limit of positivity was established as the maximum observed fluorescence when Cos-7 cells were transfected by the expression vector alone, we cannot exclude that some cells expressing a low surface α_IIbβ₃density in the transfection experiments remained in this zone. Results were comparable when using preparations of DNA from three different clones, one of them from an independent construction. To eliminate the possibility of a different transfection efficiency, we used an internal standard to control this parameter. Using pSV-β coding for β-galactosidase and MoAb against this protein, flow cytometry showed similar transfection efficiencies in permeabilized cells expressing wild-type or mutated complexes. Therefore, surface α_IIbβ₃ expression seems to be directly influenced by the R995 to Q amino acid substitution in the α_IIb subunit.

Western blotting was performed on lysates from transfected Cos-7 cells to further ensure that the lower expression of the mutated complex at the cell surface was not caused by a lower synthesis of one of the two proteins. As shown in Fig 4, α_IIb and β₃ subunits were present at approximately the same level in Cos-7 cells after cotransfection of wild-type β₃ cDNA and wild-type or mutated α_IIb cDNA. The low molecular weight product detected with SZ22 for mutated α_IIb was not observed in all experiments and was not observed when Western blotting was performed on the patient’s platelets.18 

A CHO cell line stably expressing the β₃ subunit was transfected using pCDN wild-type or mutated α_IIb cDNA. After neomycin selection, evaluation of the surface expression of α_IIbβ₃ again showed a low surface expression of the mutated complex (data not shown). Replating under conditions of limiting dilution was followed by the selection of a clone that expressed the mutated α_IIbβ₃complex in readily measurable levels. Binding of PAC-1 to these cells and to a stable CHO cell line containing wild-type α_IIband β₃ is shown in Fig 5. Little or no binding was observed to either cell line, indicating that the mutated complex was not in a high-affinity state. At the same time, greater than 95% of the cells bound PerCP-conjugated anti-β₃ or either of the MoAbs P2 or EDU-3 (anti–GP IIb-IIIa complex, data not shown). In contrast, incubation of cells expressing both the wild-type and the mutated complex with the activating MoAb, anti-LIBS6, induced the binding of PAC-1 (Fig 5). In other words, the mutated complex was only able to bind PAC-1 after activation. This result is in agreement with the previously reported observation that platelets of patient A.P. did not spontaneously bind fibrinogen but bound this adhesive protein after activation.18 The anti–LIBS6-induced binding of PAC-1 was not observed in the additional presence of 1 mmol/L RGDS, a competitive inhibitor of PAC-1 for α_IIbβ₃ (not illustrated). Hughes et al20 have expressed the activation state of α_IIbβ₃ as the AI (see Materials and Methods). Typical results for the CHO line expressing α_IIbβ₃ with α_IIb (R995Q) showed the AI to be 23, whereas that obtained for the wild-type α_IIbβ₃ was 8. The ratio for AI 995Q/AI wild-type was therefore of the order of 3, whereas the corresponding ratio given by Hughes et al20 for 995A/wild-type was of the order of 10 (see Discussion).

The mutation leading to the R to Q amino acid substitution was present in a heterozygous state in the patient’s genome and was unlikely on its own to account for a surface expression of 15% of the complex on his platelets.18 Because we failed to find any other abnormality using SSCP, we next analyzed platelet mRNA. Our strategy was to look for the genetic markers of α_IIbβ₃ (as revealed using genomic DNA) in mRNA isolated from platelets of the patient and his parents. By using RT-PCR and ASRA, we were able to determine if both alleles of each gene were effectively expressed or at least present at a detectable level. Using random oligonucleotides, we reverse transcribed platelet RNA and with specific oligonucleotides for α_IIb and for β₃ we amplified the appropriate regions of both cDNAs. Figure 6 shows that mRNA carrying the genetic determinant for HPA-1b and encoded by the β₃allele inherited from his father was undetectable in platelets from A.P. although present in his father’s platelets. Analysis of theTaqI polymorphism carried on the β₃ gene confirmed this result (Table 1). On its own, this would explain the lower total content of α_IIbβ₃ complexes in the patient’s platelets as compared with those of his father.18 However, mRNA carrying the HPA-3b determinant and coded by the α_IIb allele inherited from his mother’s genome was also not detected in RNA isolated from platelets of the patient. In this case, the corresponding mRNA was absent from his mother’s platelets. Study of a second polymorphism carried by the same allele, a silent C to T mutation in exon 30, confirmed these findings (Table 1). Sequencing also showed that mRNA coded by the mutated α_IIb allele from his father was exclusively present in the platelets of A.P. Thus, the data which are summarized in Table 1suggest that A.P. may have inherited from his mother an unexpressed allele for α_IIb, the other α_IIb allele coming from his father being mutated and giving rise to a R to Q substitution in the highly conserved cytoplasmic GFFKR domain. The low total α_IIbβ₃ complex content in A.P.’s platelets are therefore due to a combination of factors, one or more of which remain to be elucidated.

We have described the molecular analysis of α_IIb and β₃ from a patient with an atypical form of Glanzmann’s thrombasthenia. In their initial report on this patient, Hardisty et al18 showed that the density of α_IIbβ₃ complexes on his platelets was strongly decreased (approximately 15% of the normal level). However, thrombin stimulation was followed by the appearance on the surface of a substantial internal pool of α_IIbβ₃. Although a residual aggregation response was observed with physiological agonists, the kinetics were slower and the size of the aggregates much smaller than usual. Fibrinogen-binding experiments confirmed that the residual complexes were functional. This point is perhaps reinforced by the presence of fibrinogen in the α-granules of the patient’s platelets, a presence presumed to be caused by the α_IIbβ₃-dependent internalization of plasma fibrinogen.33 To define the genetic defect responsible for the disease, we performed an extensive analysis of the exons of the α_IIb and β₃ genes using nonradioactive PCR-SSCP methodology developed in our laboratory. We found that a heterozygous G to A mutation at nucleotide 3078 of α_IIbcDNA, leading to an R⁹⁹⁵ to Q amino acid substitution in α_IIb, was sufficient to decrease surface α_IIbβ₃ expression by approximately 50% when mutated α_IIb was coexpressed with wild-type β₃ in Cos-7 cells. Because this mutation was heterozygous in the patient, intriguing questions were raised relating to the extent of the observed decrease of α_IIbβ₃ on the surface of his platelets (approximately 15% of normal levels) and the altered repartition of the integrin within the surface and internal membrane systems.

In view of this unusual phenotype, we performed RT-PCR on platelet extracts for the patient and both parents to determine if both alleles of the two genes were correctly expressed. First, we found that mRNA corresponding to the α_IIb allele carrying the HPA-3b determinant and coded for by DNA from his mother and the propositus was not detected in the platelets of either subject. Secondly, we failed to find mRNA for the HPA-1b–containing β₃ allele in the platelets of the patient despite the fact that SSCP and genomic analysis indicated that mRNA for this allele was present in platelets from his father. It would be of interest to serologically type the patient’s platelets for the expression of the HPA-1a and HPA-1b alloantigens using specific alloantibodies, but the very low level of α_IIbβ₃ at the cell surface makes this difficult. At the present time, we tentatively conclude that additional and so far unidentified DNA abnormalities contribute to the low or absent expression of two of the alleles. Such abnormalities could affect noncoding regions of the α_IIb or β₃genes (promotor sequences, for example). Alternatively, they may have been missed by the SSCP procedure which is known to have less than a 100% detection rate, particularly when the size of the amplified products reaches 300 bp or more.34 We also cannot exclude the possibility of exon skipping within one or more alleles, an example of which has been recently described.35 Whatever the cause, it should be emphasized that only the mRNA coding for the mutated α_IIb was detected in A.P. platelets, suggesting that Q⁹⁹⁵ predominated in the α_IIbβ₃complexes that were present.

In our original report, we showed that platelets from the patient’s father (F.P.) expressed approximately 28,000 α_IIbβ₃ molecules as measured using radiolabeled AP-2 (anti-α_IIbβ₃) and Tab (anti-α_IIb) in direct binding studies.18 This compared with a mean value of 37,000 molecules when α_IIbβ₃ expression was assessed on platelets from a series of 27 normal subjects using AP-2. Such a value for the father is compatible with the presence of a wild-type α_IIb allele which leads to a normal cell-surface expression and a Q⁹⁹⁵ α_IIb mutated allele leading to a reduced expression. The platelets of the mother expressed approximately 20,000 molecules of α_IIbβ₃,18 a value compatible with the presence of two alleles coding for β₃ and one allele coding for α_IIb. Nevertheless, comparing the surface α_IIbβ₃ expression of individuals within a large population of normal donors is made difficult by the fact that numbers differ widely and that overlap can occur with obligate heterozygotes for type I Glanzmann’s disease.36Because the only expressed allele for α_IIb in the platelets of the propositus was mutated, the abnormal surface expression compared with the internal compartment would suggest a role for the GFFKR domain in protein trafficking. Significantly, on the basis of our results, a subject with a homozygous expression of the Q⁹⁹⁵ α_IIb mutated allele and two normally expressed β₃ alleles would possess platelets with sufficient surface-expressed α_IIbβ₃ to support platelet aggregation and for a bleeding diathesis not to be observed (as is the situation for heterozygotes in classic Glanzmann’s thrombasthenia). Thus, the bleeding syndrome in A.P. may be related to the special combination of defects in his case.

Much of the interest in the described amino acid substitution resides in its location in the highly conserved GFFKR sequence found in the cytoplasmic tail of all α subunits of human integrins and in all steroid hormone receptors (consensus sequence K×FF[K/R]R). Burns et al37 showed that overexpression of calreticulin, a calcium-binding protein present in the endoplasmic reticulum and also in the nucleus, inhibited glucocorticoid response–mediated transcriptional activation of a glucocorticoid-sensitive reporter gene and concluded that the binding of this protein to the GFFKR sequence maintained the receptor in a low-affinity state. These observations are relevant, because O’Toole et al38 showed that a deletion mutant in which α_IIbβ₃lacked the bulk of the cytoplasmic tail of the α_ΙΙbsubunit (they retained two amino acids adjacent to the transmembrane domain) was in a high activation state able to bind ligand (fibrinogen) directly when expressed in a CHO cell line. O’Toole et al39 then showed that replacement of the α_IIbcytoplasmic tail by the α₅ cytoplasmic tail gave rise to a α_IIbβ₃ chimeric receptor in a high activation state. This suggested that factors other than the GFFKR sequence per se influenced the shift from a low to a high activation state. Nonetheless, it was suggested that the GFFKR sequence could influence the interaction between the α- and β-subunit transmembrane peptides, and ultimately, the conformation of the extracellular domains. Kassner et al40 showed that replacement of the cytoplasmic tail of the α₄ integrin by a cytoplasmic tail from other α subunits did not alter the affinity state and that the effects of any change were influenced by the cell type. In contrast, Utsumi et al41 showed that truncation of the α₄ integrin subunit before the GFFKR sequence abolished α₄β₁ cell-surface expression. Kassner et al42 then showed that a deletion just after this sequence did not prevent expression but induced a much less active receptor. It was suggested that the affinity state may be regulated by the five to seven amino acids after the GFFKR sequence.

In terms of these data, patient A.P., who represents the first example of a mutation in the GFFKR region being detected in an inherited disease, represents a natural model for studying the role of this sequence in controlling the activation state of α_IIbβ₃. Initial experiments with Cos-7 cells transfected with α_IIbβ₃ containing α_IIb (R995Q) confirmed that the mutation results in a lower cell-surface expression of the complex. Then, experiments with a stable CHO cell line transfected with α_IIbβ₃ containing α_IIb(R995Q) showed that the mutated complex was not in a high activation state but remained functional in that activation could be induced by the anti-LIBS6 antibody. Recently, two opposing studies were published on this subject. Low et al21 showed that transfection of Epstein-Barr virus–transformed B lymphocytes with wild-type β₃ and α_IIb truncated proximal to the GFFKR region led to a reduced expression of the mutated α_IIbβ₃ complex. These authors further showed (1) that the α_IIb tail was not required for α_IIbβ₃ function and (2) that substitution of the GFFKR sequence by AAAAA resulted in an almost total loss of α_IIbβ₃ cell-surface expression because of an impairment in the ability of α_IIb to associate with β_3. Nevertheless, replacement of GFFKR by AAAAA gave rise to a complex which enabled the transfected lymphocytes to more easily adhere to fibrinogen when expressed at the lymphocyte cell surface. In contrast, Hughes et al20 reported that substitution of F, F, or R in the GFFKR sequence of the α_IIb cytoplasmic tail by alanine resulted in a high-affinity receptor when transiently expressed in CHO cells. The same result was obtained when D723 of the β₃ subunit was substituted by an A. Thus, the authors suggested that a salt bridge between D723 and R995 participates in the regulation of the activation state of the complex although the literature suggests that the situation is more complicated. The amino acid substitution that we have found did not give rise to a high activation state receptor. This may be because of the nature of the amino acid change. Indeed, the glutamine residue, an uncharged residue having a polar nature, could induce a modified interaction between the α_IIb and β₃ cytoplasmic tails and/or between cellular factors giving rise to receptor not locked in a high activation state but more easily activatable. This would fit with the results that we have obtained when using stable CHO cell lines. Here, the activation index was higher for 995Q than for the wild-type but was much weaker than that observed by others when the 995R was muted to 995A.20 As a result, the patient’s platelets would not be expected to spontaneously bind fibrinogen, and this was the case.18 

In view of our results, it is reasonable to consider that cell-surface expression and a high activation state of α_IIbβ₃ are not independent processes. Natural mutations giving rise to a constitutive high activation state receptor may be lethal if fully expressed, and give rise to a permanent thrombotic state. Because cell-surface expression of one subunit is dependent on the presence of the other, mutations inducing modifications in subunit association, even in cytoplasmic domains,21 can affect the efficiency of membrane expression. In the case of α_IIbβ₃, this could be the result of natural selection to impair a normal cell membrane expression of high affinity receptors. It is of interest to compare our results with those reported for another variant of Glanzmann’s thrombasthenia in which a role for the cytoplasmic domain of the β₃ subunit was highlighted. Chen et al17 showed that in the variant termed Lariboisière I, substitution of Ser⁷⁵² to Pro in the cytoplasmic tail of the β subunit resulted in a receptor that was locked in a low affinity state. Significantly, Ylänne et al43 showed that this effect was mainly caused by the presence of the proline, which induced a β-turn in the protein. Other recent work has established that a novel 111-residue polypeptide, designated β₃-endonexin, interacts with the cytoplasmic tail of β₃ (although not when the S⁷⁵² → P mutation is present) and is a potential modulator of the affinity state of α_IIbβ₃.44 It would seem that both cytoplasmic domains have key roles.

Recently, Coppolino et al45 highlighted that calreticulin is essential for integrin-mediated calcium signaling and cell adhesion. Because calreticulin is also a calcium-binding protein, it is tempting to speculate that this protein can play a major role in the activation state of integrins by modifying the salt bridge between the two subunits. Another candidate protein reacting with the GFFKR domain of α_IIb has also been described.46 In a general context, Briesewitz et al47 showed that functional specificity was under the control of the α subunit of the α₁β₁ integrin and that the assembly of the two subunits was dependent on both cytoplasmic domains. Further experiments are planned to determine the effect of the mutation that we have described in the GFFKR region on its capacity (1) to bind potential regulatory proteins, (2) influence the stability of the interaction of α_IIb with β₃, and (3) to modify intracellular trafficking of α_IIbβ₃within cells. With this goal, transfection experiments are planned in cells showing megakaryocyte-like features but lacking endogenous α_IIb.

In conclusion, to the best of our knowledge, this is the first time that an amino acid substitution in the highly conserved cytoplasmic GFFKR sequence of an integrin α subunit has been reported in a human genetic disease.