Identification of a mutation in a GATA binding site of the platelet glycoprotein Ibbeta promoter resulting in the Bernard-Soulier syndrome.

Bernard-Soulier Syndrome (BSS) is a rare congenital bleeding disorder due to absent or decreased expression of the glycoprotein Ib-IX-V (GpIb-IX-V) receptor complex on the platelet surface. To date, only mutations in GpIbalpha or GpIX have been reported in patients with BSS. GpIbbeta differs from the other proteins in this receptor in that the gene is more complex, and an alternative form is expressed in cells of non-megakaryocytic lineage, including endothelial cells. It appears that the megakaryocytic and endothelial cell mRNA species are transcribed from different start sites and have different proximal promoter regions. We have identified a patient with BSS who has a deletion on one chromosome 22, resulting in velocardiofacial syndrome. The GpIbbeta gene has been mapped to this deleted (22q11.2) region of chromosome 22. The patient has greatly reduced levels of GpIbbeta mRNA and no detectable platelet GpIbbeta protein, suggesting that his BSS results from a mutation in his remaining GpIbbeta allele. Sequence analysis revealed that the coding region of GpIbbeta is normal, but the 5'-upstream region contains a C to G transversion at base -133 from the transcription start site used in megakaryocytes. The mutation changes a GATA consensus binding site, disrupts GATA-1 binding to the mutated site, and decreases promoter activity by 84%. Thus, in this patient, Bernard-Soulier syndrome results from a deletion of one copy of GpIbbeta and a mutated GATA binding site in the promoter of the remaining allele, resulting in decreased promoter function and GpIbbeta gene transcription.

the membrane with GpIb in a 1:1 and 1:0.5 stoichiometry, respectively (1). The GpIb␤ gene is alternatively processed in megakaryocytes versus endothelial and other cells. It appears that different transcription start sites are used to produce platelet versus endothelial cell mRNA species (2,3). GpIb␣, GpIb␤, GpIX, and GpV all belong to a family of proteins known as the leucine-rich glycoprotein (LRG) family. The proteins in this family contain highly conserved leucine-rich repeats, which are thought to mediate protein-protein interactions (1).
Abnormalities in the GpIb-IX-V complex result in the moderate to severe bleeding disorder, Bernard-Soulier Syndrome (BSS). BSS is usually inherited in an autosomal recessive manner and is characterized by thrombocytopenia, enlarged platelets, and decreased or absent GpIb-IX-V on the platelet surface. Except in the one family with an apparent autosomal dominant variant of the disorder (4), heterozygotes for the disorder do not have a bleeding diathesis (5). A number of the mutations responsible for BSS have been elucidated. To date, nine published cases have been analyzed at the molecular level. Seven cases have identified mutations in GpIb␣. Three patients were identified as having a nonsense mutation producing a truncated GpIb␣ protein (6 -8), and three patients have mutations that changed a conserved amino acid within a LRG repeat (4,9,10). One patient was identified as having a mutation that changed a cysteine residue in GpIb␣ involved in disulfide bonding (11). Two of the nine cases have identified mutations in GpIX, both of which changed an amino acid in the LRG repeat or its flanking region (12,13). No mutations have been reported in GpIb␤ in a BSS patient.
The GpIb␤ gene has been localized to chromosome 22q11.2 (2,14). This is within a region in 22q11 that is deleted in 90% of patients with DiGeorge syndrome, 85% of patients with velocardiofacial syndrome and 20% of patients with conotruncal heart defects (15,16). DiGeorge syndrome is characterized by both thymic and parathyroid hypoplasia or aplasia, as well as conotruncal heart defects. Patients with velocardiofacial syndrome have heart defects, palatal abnormalities, learning disabilities, and atypical facies. This region of 22q11 has been termed the "DiGeorge chromosomal region." The majority of patients have large deletions (Ͼ1.5 megabases). The defect responsible for these syndromes is thought to cause abnormal migration of the cephalic neural crest cells. The cephalic neural crest cells contribute to the development of thymus, parathyroid, and conotruncus of the heart. The DiGeorge chromosomal region is a "gene-rich" region under intensive investigation. The GpIb␤ gene is located in the middle of this region (17).
The BSS patient reported in this study has a microdeletion in the DiGeorge chromosomal region of one allele of chromo-* This work was supported in part by Grants HL44956 (to B. A. K.), DC02027 (to B. L. D. and D. A. D.), HL51533 (to B. L. D. and D. A. D.), HL51481 (to B. P. S.) from the National Institutes of Health and from a grant-in-aid from the American Heart Association (to B. A. K.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom correspondence should be addressed: Cardeza Foundation for Hematologic Research, Thomas Jefferson University, 1015 Walnut St., Philadelphia, PA 19107. Tel.: 215-955-5598; Fax: 215-923-3836. 1 The abbreviations used are: GpIb␤, glycoprotein Ib␤; GpIb␣, glycoprotein Ib␣; LRG, leucine-rich glycoprotein; BSS, Bernard-Soulier Syndrome; HEL, human erythroleukemia; PCR, polymerase chain reaction; CAT, chloramphenicol acetyltransferase. some 22, which is presumably responsible for his developmental abnormalities (17). Because of this deletion, he also is lacking one allele of the GpIb␤ gene. As reported previously (17), he was diagnosed with BSS after presenting with congenital thrombocytopenia, large platelets, markedly decreased ristocetin-induced platelet aggregation, and markedly reduced GpIb␣ on the platelet surface by flow cytometry. Analysis of patient platelet lysates by Western blotting revealed that GpIb␣ protein was decreased compared with normal platelets, and GpIb␤ protein was not detected. In addition, GpIb␣ and GpIX mRNA were readily amplified from the patient's platelet RNA by reverse transcription PCR, but GpIb␤ mRNA was only detected at very low levels (17). We report that in this patient BSS results from a mutation in the promoter region of the remaining GpIb␤ allele that alters a GATA binding consensus site. In vitro, the mutation disrupts GATA-1 binding to this site and results in a 6-fold decrease in promoter function.

MATERIALS AND METHODS
Genomic DNA Isolation and Southern Analysis-Genomic DNA was isolated from 5 ml of citrated blood as described previously (18). Samples of 10 g of DNA were cut with various restriction enzymes: SacI, SalI, and PstI (Promega Corp., Madison, WI). After running the digests on a 1% agarose (Life Technologies, Inc.) gel, the gel was exposed to 312-nm UV light for 5 min. The gel was then soaked in 1 N NaOH for 30 min followed by 2.5 M NaCl, 0.5 M Tris-HCl, pH 7.5, for 2 ϫ 30 min. The DNA was transferred to a nylon membrane (Hybond-N ϩ , Amersham Corp.) by Southern blotting in 10 ϫ SSC overnight. The DNA was fixed to the membrane with 312-nm UV irradiation. The blots were prehybridized for at least 1 h at 65°C in 1 M NaCl, 0.1% SDS, 1.5 mg/ml herring sperm DNA, 10 g/ml human placental DNA, and 10% dextran sulfate and then hybridized with a radiolabeled GpIb␤ cDNA probe (19) or a radiolabeled GpIX cDNA probe (20) overnight at 65°C in prehybridization solution with an additional 1.5 mg/ml herring sperm DNA. The blots were washed to high stringency (0.1 ϫ SSC, (SSC: 2.25 M NaCl, 0.225 M sodium citrate), 0.1% SDS, 1 mM EDTA, 10 mM sodium phosphate, pH 6.8, at 68°C and analyzed by autoradiography.
Construction of CAT-expressing Vectors-The CAT-containing vector used was pCAT Basic (Promega Corp.). A 384-base pair fragment of the GpIb␤ promoter was amplified from genomic DNA using the PCR conditions described above. The GpIb␤ primers used were 677-694 and 1059 -1040 with linkers for SalI on the forward primer and XbaI on the reverse primer. The 384-base pair product was digested with XbaI and SalI and cloned into pCAT Basic linearized with the same enzymes. The A4 construct contains a 623-base pair piece of the GpIb␣ promoter, nucleotides Ϫ573 to 50 in the pCAT Basic vector (22).
Cell Transfection and CAT Assay-Human erythroleukemia (HEL) cells were transfected by a modification of the method described by Narayanan et al. (23). 1 ϫ 10 7 HEL cells growing in log phase were pelleted and washed with 10 ml of transfection buffer (30.8 mM NaCl, 120.7 mM KCl, 8.1 mM Na 2 HPO 4 , 1.46 mM KH 2 PO 4 , 5 mM MgCl 2 ). The cells were resuspended in 350 l of transfection buffer and transferred to 0.4-cm Gene Pulser electroporation cuvettes (Bio-Rad). 60 g of plasmid DNA was dried under vacuum, resuspended in 40 l of transfection buffer, and added to the cells. After gentle mixing, the cuvette was placed on ice for 15 min. The cells were then electroporated at 960 microfarads and 0.25 kV in a Bio-Rad gene pulser (Bio-Rad). The cuvette was kept on ice for 12 min, then at room temperature for 15 min. The mixture was then added to 20 ml of RPMI 1640 (Life Technologies, Inc.) with 10% fetal bovine serum (Hyclone, Logan, UT) and 0.1 mg/ml gentamycin (Life Technologies, Inc.). The cells were incubated for 72 h at 37°C in 5% CO 2 before harvesting for the CAT assay. CAT assays were performed as described (24) and evaluated by ascending thin layer chromatography in chloroform:methanol (95:1) and autoradiographed overnight at Ϫ70°C. Quantitative analysis of the CAT assay was performed using a PhosphorImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA).
Electrophoretic The labeled probes were purified using the Xtreme spin column kit (Pierce). Before adding probe, the binding reactions containing 3-12 g of nuclear extract and 0.5 g of poly(dI-dC)⅐(dI-dC) (Pharmacia Biotech Inc.) in 10 mM Tris-HCl, pH 7.5, 50 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, 15% glycerol were preincubated for 10 min at room temperature as described by Uzan et al. (27). 4 ϫ 10 4 cpm of labeled oligonucleotide was added, and the reactions were incubated for 20 min at room temperature. For the supershift assays, 2 g of antibody was added after the probe and incubated for an additional 45 min at room temperature. The antibodies used were a monoclonal anti-GATA-1 (Santa Cruz Biotechnology, Santa Cruz, CA) and a polyclonal anti-GATA-2 antibody supplied by Dr. Stuart Orkin (Children's Hospital, Boston, MA). The samples were analyzed on 5% polyacrylamide, 2.5% glycerol gels that were run in 0.5 ϫ TBE (1 ϫ TBE ϭ 89 mM Tris-HCl, 89 mM boric acid, 2 mM EDTA) for 1 h at 200 V and 4°C. Gels were dried and then autoradiographed overnight at Ϫ80°C.

RESULTS
Southern Blot Analysis-The patient's remaining GpIb␤ allele was analyzed by Southern blot analysis to determine if the gene was grossly normal. Genomic DNA from both the patient and a control sample was digested with SacI, SalI, and PstI. The autoradiograph of the digests probed with radiolabeled GpIb␤ cDNA are shown in Fig. 1A. The bands seen in the digests of the patient's DNA are identical to the control sample, so the patient has no large deletions or rearrangements in his GpIb␤ gene. Although the same amount of genomic DNA was analyzed, there is a decrease in the intensity of the bands in the patient lanes of Fig. 1A compatible with the loss of one GpIb␤ allele. Fig. 1B shows the SacI digest reprobed with a radiolabeled GpIX cDNA. The GpIX gene maps to chromosome 3 (20), which is unaffected in this patient, so the intensity of the band in the patient lane of Fig. 1B is similar to the control, indicating that unlike GpIb␤, both alleles of GpIX are present.
Sequence Analysis of the GpIb␤ Gene-Overlapping frag-ments of the patient's GpIb␤ gene were amplified by PCR and sequenced in parallel with a normal gene. The entire coding region, including the LRG repeat, was normal. The only change found, shown in Fig. 2, was a C to G transversion in the 5Ј-upstream region of the platelet GpIb␤ gene. This change lies 133 base pairs upstream from the megakaryocytic transcription start site (14) and alters a reverse GATA consensus site from TATC to TATG (Fig. 3). It is located 9 base pairs downstream from another GATA binding consensus site beginning at Ϫ146. No studies have been published identifying the sequences necessary for efficient GpIb␤ promoter function.
Allele-specific PCR-Allele-specific PCR was used to confirm the mutation identified in the patient's GpIb␤ promoter. This method is based on the finding that a mismatch between the most 3Ј nucleotide of the primer and the target sequence prevents amplification under certain conditions. Thus, by utilizing a primer in which the mutation is at the 3Ј end, one can differentiate normal from mutant sequence in genomic DNA. The results are shown in Fig. 4. When control DNA was amplified with both the normal primer set (primers 1 and 2, see "Material and Methods" for sequence) and the mutated primer set (primers 1A and 2), the 259-base pair product was amplified only with the normal primers. This indicates that the control DNA contains only the normal allele. However, when the patient's DNA was amplified with both the normal and mutated primer sets, the PCR product was detected only with the mutated primers, indicating that the patient has only the mutant allele. These results show that the patient's promoter mutation, which was initially identified from a PCR subclone, is present in the patient's genomic DNA.
GpIb␤ Promoter Studies-Studies were performed using a CAT assay to determine if the altered GATA binding site in the patient's promoter has any effect on gene transcription in HEL cells. HEL cells are a megakaryocytic cell line that is known to express GpIb (28) and has been used in studies of transcriptional regulation of genes expressed in megakaryocytes (27). The GpIb␤ promoter constructs, diagrammed in Fig. 5, encompass nucleotides Ϫ276 to ϩ108 of the gene with the normal having a C at nucleotide Ϫ133 and the patient having a G at this position. A promoter construct of the GpIb␣ promoter was used as a positive control and the CAT vector with no promoter was used as a negative control. The constructs were transfected into HEL cells, which were assayed for CAT activity 72 h post-transfection. The amount of CAT enzyme produced, which is noted as the percent conversion of chloramphenicol to its acetylated forms, reflects promoter activity. As shown in Fig. 5, the normal GpIb␤ construct had more activity than the patient's GpIb␤ promoter construct. In six independent experiments, the patient's GpIb␤ promoter activity was decreased by 84 Ϯ 13% (mean Ϯ S.D.) compared with that of the normal promoter.
GpIb␤ Promoter Binding Studies-To identify which nuclear proteins are binding to the GpIb␤ promoter in the region with the patient's mutation, we performed electrophoretic mobility shift assays. The promoter oligonucleotides used in these assays are noted in the blocked sequence in Fig. 6. The 24-mers include nucleotides Ϫ141 to Ϫ118 of the promoter with the Fragments of the patient's and a normal GpIb␤ gene were amplified by PCR and cloned for sequencing. The sequence was analyzed on a 6% denaturing polyacrylamide gel. The normal (N) gene contains a C at nucleotide Ϫ133 (sequence numbers from the megakaryocytic transcription start site, (14)). The patient's (P) gene is mutated to a G at this position.  3. Location of the patient's mutation. The promoter region of the GpIb␤ gene is shown schematically. The patient's mutation lies 133 base pairs upstream from the megakaryocytic transcription start site (0) and changes a GATA consensus binding site between nucleotides Ϫ136 and Ϫ133 from TATC to TATG. It is located nine base pairs downstream from another GATA consensus site starting at nucleotide Ϫ146 (14). The locations of a third potential GATA binding site and an Ets consensus binding motif are also shown in relation to the transcription start site.

FIG. 4. Confirmation of the patient's mutation using allelespecific PCR.
Genomic DNA from both a normal individual (C) and the patient (P) were subject to PCR using primer with the normal (N) or mutant (M) sequence. The normal primers amplify when a C is present at nucleotide Ϫ133, and the mutant primers amplify when a G is present at this position. The 259-base pair product was analyzed on a 2% agarose gel. The size of the DNA markers in base pairs is noted to the left. The primer sequences and amplification conditions are outlined under "Materials and Methods." normal having a C at position Ϫ133 and the patient having a G at this position. Nuclear extracts from CHRF-288, a megakaryoblastic cell line, HEL, or HeLa cells, an epithelial cell line, were incubated with radiolabeled normal patient or Sp1 consensus oligonucleotides, then further incubated with an antibody to GATA-1 or GATA-2 when noted. The samples were analyzed by gel electrophoresis, and the results are shown in Fig. 6. A protein from either CHRF or HEL nuclear extracts bound to the normal oligonucleotide, but not to the patient oligonucleotide. In competition studies, it was demonstrated that the protein binding the normal oligonucleotide could not be competed off with a 35-fold excess of the mutated oligonucleotide, but the binding could be inhibited with only 1.5-fold of the unlabeled normal oligonucleotide (data not shown). As shown in Fig. 6, the binding complex between the nuclear protein and the normal oligonucleotide was shifted after incubation with a monoclonal antibody to GATA-1, but not with a polyclonal antibody to GATA-2, which is known to supershift. 2 In addition, HeLa nuclear protein extract, which contains GATA-2, but not GATA-1 (29), did not bind either the normal or patient oligonucleotide, but did bind to Sp1 as a control. These results suggest that the GATA-1 protein is binding to the normal GATA consensus site, but is unable to bind to the patient's mutated GATA site.

DISCUSSION
In this study, we have identified the first case of Bernard-Soulier syndrome that can be attributed to a mutation in the GpIb␤ gene. The patient has only one allele of GpIb␤, because of a large deletion at 22q11.2 containing the other allele. His remaining allele contains a C to G transversion at nucleotide Ϫ133 from the megakaryocytic transcription start site that mutates a GATA consensus binding site. In vitro, the mutation disrupts GATA-1 binding at this site and decreases transcrip-tion of the gene by 84%. We propose that the mutation identified in the promoter region of the platelet GpIb␤ gene is responsible for causing BSS in this patient.
In classic BSS, it has appeared that the absence of one of the components of the complex affects the expression of the other proteins. This concept is supported by co-transfection studies using GpIb␣, GpIb␤, and GpIX in heterologous cells (30), where the expression of all three components was necessary for stable, functional expression of the receptor in the membrane. The ability of a mutation in the gene encoding either GpIb␣ or GpIX has been demonstrated to affect the expression of other proteins of the complex in patients with BSS. We now demonstrate that this is also true with GpIb␤.
The GpIb␤ gene is more complex than the other genes encoding the proteins of this complex. An alternative form of the gene is expressed in non-megakaryocytic cells, including endothelial cells. Endothelial cells and other tissues express a larger GpIb␤ mRNA species that produces a protein larger (approximately 45 kDa) than the platelet protein (2). The function of this larger protein is unknown, but it has been shown to be present in human umbilical vein endothelial cells (2), a human glioma cell line, and mouse brain tissue 3 by Western blotting with a polyclonal antibody directed against the platelet GpIb␤ protein. The mutation identified in this patient, which is located in the promoter region of the platelet GpIb␤ gene, is located in the coding region of the endothelial cell GpIb␤ cDNA. However, the mutation results in a conservative amino acid change (Leu 62 3 Val), suggesting that the endothelial cell mRNA and protein species may still be synthesized in this patient. Studies are under way to determine if the endothelial cell mRNA is still transcribed in this patient, while the platelet GpIb␤ mRNA is affected by the mutation.
As noted previously, obligate heterozygotes for BSS do not, in general, have bleeding manifestations. It is possible that clinical manifestations of heterozygotes for BSS could vary depending on the subunit affected. However, patients with syndromes associated with a microdeletion on one chromosome 22q11.2, FIG. 6. The effect of the patient's mutation on nuclear protein binding to the GpIb␤ promoter. The GpIb␤ promoter oligonucleotides used are shown above in the blocked sequence. They are 24-mers spanning nucleotides Ϫ141 to Ϫ118 with the normal (N) having a C at position Ϫ133 and the patient (P) having a G at this position. A consensus oligonucleotide for the Sp1 transcription factor was used as a positive control for the HeLa nuclear extract. Nuclear extracts from CHRF-288, HEL, or HeLa cells were incubated with radiolabeled normal (N), patient (P), or Sp1 oligonucleotide, then further incubated with an antibody to GATA-1 or GATA-2 when noted. Samples were analyzed on a 5% polyacrylamide, 2.5% glycerol gel. The gel shift procedure is detailed under "Materials and Methods." The CAT vector with no promoter (pCAT) was used as a negative control. HEL cells were transfected with each construct and assayed for CAT activity 72 h post-transfection. After six independent experiments, the average percent conversion of chloramphenicol to its acetylated products was calculated and is noted below each construct. Construction of the CAT-expressing vectors, the transfection procedure and the CAT assay were performed as outlined in "Materials and Methods". which would include GpIb␤, have not been noted to be thrombocytopenic or have bleeding disorders (31). We have evaluated platelet counts and GpIb␣ protein in 11 patients with a 22q11.2 microdeletion and found no consistent difference between the patients and normal controls (data not shown). Thus it appears that the loss of one GpIb␤ allele alone is not sufficient to produce a bleeding disorder, but that these patients may be at greater risk to develop BSS through a single mutation of the remaining GpIb␤ allele.
This case of BSS is unique because it is the first described natural occurring mutation of a GATA promoter element in a megakaryocytic gene, demonstrating the importance of this regulatory element in megakaryocyte gene expression. In vitro studies have identified the GATA consensus binding sequence as an important element for the transcription of megakaryocyte-specific genes. The genes for platelet factor 4, GpIb␣, and GpIIb all contain GATA binding sites in their promoter regions that are necessary for efficient transcription (27,(32)(33)(34). The two members of the GATA family of proteins expressed in the megakaryocytes are GATA-1 and GATA-2, both of which bind to the (A/T)GATA(A/G) consensus sequence (35). Although the roles of GATA-1 and GATA-2 in erythroid development have been, at least in part, elucidated, the roles of these factors in non-erythroid cells is less clear. From studies of GATA-2 gene disruption in mice, we know that GATA-2 is important for maintenance of early hematopoietic progenitors and possibly stem cells (36). Studies of GATA-1 gene disruption in mice have shown that GATA-1 is not required for megakaryocyte or platelet formation (37). However, the potential role of GATA-1 or GATA-2 in the modulation of megakaryocyte gene expression has not been elucidated. Our study suggests that GATA-1 binding is necessary for efficient GpIb␤ transcription.
The platelet GpIb␤ gene has three GATA consensus binding sites in the promoter region. The first and most proximal to the transcription start site is located between nucleotides Ϫ37 and Ϫ34. The second site, which is mutated in this BSS case, is located between nucleotides Ϫ136 and Ϫ133. The third and most distal site is located between nucleotides Ϫ146 and Ϫ143 (14). This BSS patient demonstrates that the second GATA binding site is important for GpIb␤ promoter function. A mutation at this GATA sequence, while the other two sites are normal, decreases promoter function 6-fold. Ongoing studies evaluating GATA binding to the GpIb␤ promoter suggest that the patient's mutation may also affect binding to the adjacent GATA site at nucleotide Ϫ146. It is possible that the GATA proteins bind these two adjacent sites cooperatively. Studies are continuing to determine which GATA proteins are binding to the other motifs and what effect a mutation at these sites has on promoter function. In addition, we intend to extend our promoter studies to include sequence further 5Ј to identify other positive or negative regulatory elements that are important for GpIb␤ promoter function.
This case of Bernard-Soulier syndrome is also unique, because it can be attributed to a GpIb␤ gene mutation in combination with a chromosome 22 deletion. Classical BSS is almost always inherited in an autosomal recessive manner. This patient's chromosomal abnormality results in haploinsufficiency, which contributes to his phenotype by unmasking a defect in the GpIb␤ gene on the non-deleted chromosome. Microdeletions in 22q11.2 are present in the majority of patients with DiGeorge and velocardiofacial syndrome (15,16). Although several candidate genes have been identified, the specific gene(s) contributing to the manifestations of these syndromes have not been elucidated. Although the loss of one GpIb␤ allele does not appear to result in a bleeding abnormality, an unknown function of the gene may contribute to these developmental syndromes.