Impaired intracellular transport produced by a subset of type IIA von Willebrand disease mutations.

Type IIA von Willebrand disease (vWD) results from abnormalities in von Willebrand factor (vWF) characterized by absence of plasma high molecular weight (HMW) vWF multimers. In this report, 5 distinct point mutations were identified in 6 Type IIA vWD families. A total of 7 mutations, all clustered within a 124-amino acid segment of the vWF A2 domain, now account for 9 of a panel of 11 Type IIA families. In COS-7 cells, 3 single amino acid substitutions, Val844----Asp, Ser743----Leu, and Gly742----Arg, impaired the transport of vWF multimers between the endoplasmic reticulum and the Golgi complex, with more profound effects on the secretion of HMW multimers than lower molecular weight forms. In contrast, 2 substitutions, Arg834----Trp and Gly742----Glu, resulted in secretion of HMW multimers similar to wild-type vWF. The vWF structure observed within patient platelets correlated closely with the synthesis pattern seen for the corresponding mutants in COS-7 cells. These findings demonstrate that structural alterations within the A2 domain of vWF can produce the characteristic phenotype of Type IIA vWD via two distinct molecular mechanisms.

disease (vWD) is the most common inherited bleeding disorder in humans. Over 20 distinct vWD phenotypes have been described, all manifesting as either quantitative (Type I) or qualitative (Type 11) abnormalities in plasma vWF (5).
vWF is synthesized as a 2813-amino acid prepropolypeptide in endothelial cells and megakaryocytes] where it subsequently undergoes a complex series of processing steps prior to storage or secretion. Cleavage of the signal sequence and propeptide results in a mature 2050-amino acid vWF monomer. Biosynthesis and processing of vWF has been characterized in cultured endothelial cells and includes dimerization, glycosylation, sulfation, propeptide cleavage, and multimerization to form molecular species consisting of up to 100 subunits. vWF molecules dimerize within the endoplasmic reticulum (ER) via disulfide bond formation at the carboxylterminal end. In the Golgi and post-Golgi compartments, the prosequence is cleaved and disulfide linkages are formed at the amino terminus to generate high molecular weight (HMW) multimers (6,7). The propeptide has been shown to be required for multimer assembly, although cleavage is not necessary for this function (8,9). Wise et al. (8) have postulated that some Type I1 vWD variants could be due to mutations in the propeptide resulting in failure to form HMW multimers. These large polymers are thought to be most critical to the function of vWF as mediator of platelet adhesion (1). Type IIA vWD is autosomal-dominant in inheritance and is characterized by the absence of HMW vWF multimers in plasma. There is evidence that extracellular proteolysis of vWF may be responsible for the loss of HMW multimers in a subset of Type IIA vWD patients (10-12)1 whereas the mechanism in others is unknown.
Several single base missense mutations potentially responsible for Type IIA vWD have been identified (13)(14)(15). In this study, we extend our analysis to an additional 8 Type IIA vWD families and report the identification of apparent missense mutations in 6. To investigate the molecular mechanism responsible for the absence of HMW multimers in Type IIA vWD plasma, we have characterized the synthesis and assembly of vWF containing each of 5 Type IIA vWD mutations, Val844 + Asp(V844D), A r P + Trp(R834W), Ser743 + Leu(S743L), Gly742 + Glu(G742E), and Gly742 + Arg(G742R).

EXPERIMENTAL PROCEDURES
Patient Material-Nine patients with Type IIA vWD from 8 unrelated families were studied (A3-11). A10 has been reported previously (10). Patients were classified by referring physicians as having Type IIA vWD based on a moderately severe clinical bleeding disorder, prolonged bleeding time, decreased Factor VI11 procoagulant activity, decreased vWF antigen, decreased ristocetin co-factor activity, and a vWF multimer pattern consistent with Type IIA vWD.
3).* Primers were engineered to create synthetic restriction sites at DNA Sequence Analysis-PCR products were cloned into M13mp18 and M13mp19 for sequence analysis as previously described (13,16) or directly sequenced. For the latter, asymmetric PCR was performed as previously described (18,19) using each PCR primer alone for a second round of PCR with 1 pl of the first round product as template. Single-stranded DNA template was isolated by purification over a Quiagen column according to the manufacturer's protocol (Quiagen). DNA sequence analysis was performed on the purified single-stranded material using Sequenase (U. S. Biochemicals).
Expression Vector Construction-Identified Type IIA vWD mutations were inserted into full length vWF cDNA in the expression vector pMT2 (20). The expression vector containing wild-type full length vWF is designated pWTvWF, and expression vectors contain-pS743L, pG742E, andpG742R. Construction of pV844D and pR834W ing each mutation are designated as follows: pV844D, pR834W, has been described previously (13). For the remaining mutations, a segment of DNA was amplified by PCR from patient genomic DNA using primers A and B (Fig. 1). An NcoI-KpnI fragment within the PCR product was assembled into full length vWF cDNA in pMT2 using standard methods. All segments derived by PCR were completely sequenced to ensure that final constructs were free of Taq polymerase errors and contained the desired mutations. The pJ3@-Gal vector, which contains the SV40 origin, enhancer, and promoter and a LacZ cDNA, was kindly provided by L. Spain (Whitehead Institute).
Cell Culture and Transfection-COS-7 cells were maintained in Dulbecco's modified Eagle's medium (JRH Biosciences), supplemented with 10% fetal bovine semm (Gibco Laboratories). Transfections were performed as previously described (21). Cell medium and cell lysates were harvested between 48 and 60 h post-transfection. Cell medium was collected into a final concentration of 5 mM EDTA and 2 mM phenylmethylsulfonyl fluoride. For pulse-chase experiments, cells were lysed directly from plates. Cell lysates were prepared in 3-5 ml of Nonidet P-40 lysis buffer (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 0.5% SDS, 1% Nonidet P-40, 5 mM EDTA, 2 mM iodoacetic acid, and 2 mM N-ethylmaleimide) per 100-mm plate. To examine steady state levels of vWF, cells were first released from plates with 0.1% (w/v) trypsin followed by two washes with phosphate-buffered saline to remove extracellular vWF adhering to cells. Cell lysates were then prepared from cell pellets in 1.5-3 ml of Nonidet P-40 lysis buffer. Phenylmethylsulfonyl fluoride at a final concentration of 2 mM was added to all cell lysates.
ELZSA-vWF was quantitated by a sandwich ELISA using 1:lOOO anti-vWF (immunoglobulin fraction, Dakopatts) as the coating antibody and 1:4000 peroxidase-conjugated anti-vWF (immunoglobulin fraction, Dakopatts) as the detecting antibody, according to manufacturer's instructions. @-Galactosidase (,%Gal) was quantitated by an ELISA utilizing 1:500 anti-D-Gal antibody (Cappel) as the coating antibody and a 1:4000 dilution of anti-P-Gal conjugated to biotin, followed by 5 milliunits of streptavidin-peroxidase conjugate (Boehringer Mannheim) as the detecting system. To conjugate biotin to anti-D-Gal, anti-B-Gal was incubated with D-biotinyl-e-aminocaproic acid N-hydroxysuccinimide ester (Boehringer Mannheim) at a weight/weight ratio of 1O:l 1gG:biotin at room temperature for 4 h followed by dialysis against 50 mM Tris-HC1, pH 7.5. ELISAs were developed with o-phenylenediamine as the colorimetric substrate and quantitated at Adg2 on a Dynatech MR650 ELISA reader.
Pulse-Chase Labeling-48 h post-transfection, COS-7 cells were washed twice with phosphate-buffered saline and placed in minimum essential medium lacking methionine and cysteine (Gibco) for 1 h. Cells were then pulsed for 15 min with 250 pCi each of [35S]methionine and [36S]cysteine per 100-mmplate in 2 ml of minimum essential * D. Ginsburg and M. E. Bruck, unpublished results. medium minus methionine and cysteine with 0.5% bovine serum albumin. Labeled medium was then removed, 4 ml of complete Dulbecco's modified Eagle's medium containing 0.5% bovine serum albumin and 1% insulin/transferrin/selenium supplement (Collaborative Research, Inc.) was added for various chase periods, and cell medium and cell lysates were collected.
Zmmunoprecipitution-vWF and 0-Gal were immunoprecipitated from 500 pl of cell lysate or 900 pl of cell medium in immunoprecipitation buffer (10 mM Tris-HC1, pH 7.5, 1 mM EDTA, 150 mM NaC1, 0.5% Triton X-100,0.5% sodium deoxycholate) with 500 ng of rabbit anti-human vWF (IgG fraction, Dakopatts) and 3 pg of rabbit anti-0-Gal (IgG fraction, Cappel), followed by incubation with 15 pl of protein A-agarose beads (Bethesda Research Laboratories). Beads were washed twice in immunoprecipitation buffer and twice in phosphate-buffered saline prior to addition of gel loading buffer. The quantity of immunoprecipitated @-Gal served as an internal control both for transfection efficiency and for recovery of immunoprecipitated material. Duplicate immunoprecipitations of each sample indicated that recoveries of vWF and @-Gal were reproducible and reprecipitations of samples showed that the immunoprecipitations were complete. Co-immunoprecipitation of vWF and binding protein, BiP were performed in phosphate-buffered saline with 100 p1 of rat anti-BiP antisera kindly provided by Dr. D. Bole (University of Michigan).
Endoglycosidase H Digestion-Endoglycosidase H (Endo H) was a gift of Dr. I. Goldstein (University of Michigan). Cell medium or cell lysate samples were heated to 100 "C in 90 mM sodium acetate, pH 5.2, 0.2% SDS. Endo H digestion was performed with 2 pg of Endo H in 70 D M sodium acetate, pH 5.2,0.15% SDS, 0.15% bovine serum albumin, 0.8% Triton X-100, 1 mM phenylmethylsulfonyl fluoride overnight at 37 "C. Undigested samples were treated identically without the addition of Endo H.
Protein Gel Electrophoresis and Electroblotting-SDS-Polyacrylamide gel electrophoresis was performed as described by Laemmli (22) using 6% gels with a modified acrylamide to bisacrylamide ratio of 1101. Samples were placed in a 100 "C water bath for 3 min immediately prior to loading on gels. Polyacrylamide gels containing %-labeled samples were fixed and treated with ENHANCE (Du Pont-New England Nuclear). Gels were then dried and exposed to Kodak XAR-5 film. Quantitation of 35S-labeled bands was performed on a Betascope 603 blot analyzer (Betagen). SDS-polyacrylamide gels that were transferred to nitrocellulose were electroblotted in 50 mM Trizma base, 380 mM glycine, 0.05% SDS, 20% methanol for 2 h.
Plasma and platelet samples were prepared and analyzed on nonreducing SDS-agarose gels at the Mayo Clinic, Rochester, MN. Patient platelet lysates were prepared from blood by a modification of Ref. 24. Platelet pellets were resuspended in 0.5 ml of 2.5 mM EDTA, pH 7.2, and subjected to 5 cycles of freeze, thaw, and vortex mixing, followed by the addition of 0.2 ml of 10% SDS and 0.2 ml of 20% Triton X-100 and heating to 60 "C for 15 min. Plasma and platelet multimer analysis was performed as previously described (25).

RESULTS
Identification of Type IIA vWD Missense Mutations-Patient DNAs were first amplified by PCR and screened by restriction enzyme digestion for three previously identified mutations (13,14). Three patients (A8, A9, A10) showed loss of the BstEII site associated with the A2 mutation (designated patient 2 in Ref. 13), a C + T substitution at nucleotide 4789 which results in a single amino acid substitution, + Trp (R834W) (Fig. 1). Direct sequence analysis of A8, A9, and A10 PCR products confirmed the same mutation in these Impaired Transport in von Willebrand Disease FIG. 1. Schematic diagram of Type IIA vWD mutations. Primers A, B, and C (indicated by arrows) were used to amplify Exon 28 sequence without amplification of the pseudogene sequence. Exon 28, which encodes A1 and A2 vWF repeats, is depicted as a black bar, and the surrounding intron sequence as black lines. The full length vWF coding sequence is depicted as a gray bar, including the homologous A, B, C, and D repeats (4,27). The A2 homologous repeat is expanded below including the 7 Type IIA vWD missense mutations with their corresponding amino acid substitutions which are shown schematically. Amino acids are numbered according to Bonthron et al. (17). Sites of cleavage of the signal peptide (sp) and vWF antigen I1 propeptide ( p r o ) are illustrated. The 176-kDa carboxyl-terminal vWF fragment associated with Type IIA vWD is shown as a black bar, and the site of cleavage (Y842-M843) is indicated by an asterisk (35). The five mutations analyzed by transfection are highlighted by boxes.
patients. A9 is the son of A8. Using a frequent neutral polymorphism, C4641 + T (26)) located 148 bp upstream of the A2 mutation, individual alleles could be distinguished. The A2 mutation was observed to be on a (+)-chromosome (4641C), whereas the same mutation in A8 was found to be on a (-)-chromosome (4641T)) suggesting at least two independent genetic origins.
PCR and sequence analysis of Exon 28 from patients A3, A4, and A6 identified three distinct mutations in two adjacent codons (Fig. 1). A G + A transition at nucleotide 4514 (A4) and a G + C transversion at nucleotide 4513 (A6) result in substitutions of Glu and Arg, respectively, for Gly742. In the adjacent codon, a C + T transition at nucleotide 4517 (A3) substitutes Leu for Ser743. A T + C substitution at nucleotide 4619 was identified in A7, altering amino acid 777 from Leu + Pro. In an additional two patients (A5 and All), no abnormalities in Exon 28 sequence were found. All 5 mutations were observed to be heterozygous with the normal sequence, and 149 normal alleles screened by PCR for each mutation were negative. Including the three previously reported mutations (13,14), we have now identified 7 distinct mutations accounting for 9 of a total of 11 unrelated Type IIA vWD families studied. The locations of these mutations within the vWF molecule are shown schematically in Fig. 1. All 7 mutations are clustered within a 124-amino acid segment within the A2 repeat of vWF.

Point Mutations Have Heterogeneous Effects on Secretion
of uWF-In order to examine the biosynthesis and secretion of vWF containing the Type IIA mutations, COS cells were transfected with a vWF expression plasmid (pMT2vWF) containing each of five mutations indicated by boxes in Fig. 1. A &galactosidase expression vector (pJ3P-Gal) was co-transfected with each construct to provide an internal control for transfection efficiency and immunoprecipitation. Quantitative analysis of cell media and lysates by ELISA showed that the amount of vWF present in the cell medium was similar or mildly decreased compared to wild-type for R834W and G742E, but was greatly reduced for the mutants V844D, S743L, and G742R (Fig. 2). The V844D mutation was associated with the most pronounced defect, with extracellular vWF levels at only 3.5% of WTvWF, while the S743L and G742R mutations resulted in extracellular vWF levels at 14% and 22% of wild-type, respectively. Steady state levels of intracellular vWF were elevated compared to WTvWF for the mutant V844D, while the other mutants had levels similar to WTvWF.
To further examine the cellular mechanism for these variable levels of secreted vWF, transfected cells were pulselabeled, and the vWF in cell media and lysates were immunoprecipitated and analyzed by electrophoresis and autoradiography. At each time point, quantities of vWF were normalized to the amount of immunoprecipitated P-Gal. WTvWF was detected in the medium after 2 h and was completely chased into the medium between 48 and 66 h (Figs. 3 and 4). For R834W or G742E, kinetics of secretion determined by pulse-chase analyses were similar to WTvWF (Figs. 3 and 4). In contrast, the mutants V844D, S743L) and G742R were in uon Willebrand Disease  Fig. 3) was quantitated on a Betascope blot analyzer. The counts obtained in proand mature vWF were normalized for differences in cysteine and methionine content. The quantity of P-Gal immunoprecipitated at each time point was used to normalize for transfection efficiency. Within each pulse-chase experiment, the number of counts is expressed in arbitrary units.  (-) Endo H and were electrophoresed through 6% reducing SDS-polyacylamide gels, followed by electroblotting and detection by chemiluminescence. secreted at decreased rates compared to WTvWF. The three mutations impaired secretion to varying degrees. V844D was not detected in the cell medium by pulse-chase, even after 66 h. While V844D resulted in a nearly complete loss of vWF secretion in COS cells, S743L and G742R caused less severe transport defects. With these three mutants, intracellular pulse-labeled vWF accumulated and then decreased (Figs. 3C and 4A), indicating that the mutant V844D, S743L, and G742Rproteins are degraded in an intracellular compartment.

u WF Secretory Mutants Are Blocked in Transport from the Endoplasmic Reticulum
to the Golgi Complex-COS cells transfected with pWTvWF secreted both pro-and mature vWF (Fig. 3A) consistent with previous studies (27). Cell lysates contained a single form of vWF which appeared slightly smaller than the pro-vWF seen in cell medium and was Endo H-sensitive, indicating it had not undergone the complete carbohydate processing that occurs in the Golgi complex (Fig. 5). vWF forms that had passed through the ER to the Golgi compartments were not detectable in cell lysates, indicating that exit of pro-vWF from the ER is the ratelimiting step in the synthetic process. Thus, after exit from the ER, vWF travels rapidly through the Golgi compartments to be secreted.
All of the mutants contained a single vWF species in cell lysates that appeared identical in size with WTvWF on reducing gels. For all mutants, intracellular vWF was observed to be Endo H-sensitive, indicating that these proteins were retained in a compartment prior to the medial Golgi. These results suggest that intracellular transport of the secretory mutants, V844D, S743L, and G742R is blocked between the ER and the Golgi complex. vWF present in the cell medium from all transfections was Endo H-resistant, indicating that it represented secreted vWF rather than the release of intracellular vWF.
Many proteins that are retained within the ER have been found to be associated with resident ER binding proteins (28). The binding protein BiP (GRP78) associates with partially assembled complexes and misfolded or incorrectly glycosylated proteins in the ER (29-31). vWF has been shown to be transiently associated with BiP (29). Therefore, mutants blocked in ER to Golgi transport were examined for binding to BiP as a mechanism for altered secretion. Cell lysates were collected after either a 15-min labeled pulse or a pulse followed by a 16-h cold chase, and immunoprecipitated with antibodies to vWF or BiP. A small amount of both wild-type and mutant vWF proteins were co-immunoprecipitated with BiP at these time points, but there was no significant difference in the quantities of BiP-associated vWF between WTvWF and any of the mutant forms (Fig. 6). Thus, an increased association with BiP does not appear to occur in conjunction with retarded transport of these mutant forms of vWF.

Severity of the Secretory Impairment Correlates with Decreased Secretion of High Molecular Weight Multimers-Syn-
thesis of multimers was examined by analysis on nonreducing 1.5% SDS-agarose gels. In cell lysates from pulse-chase experiments, WTvWF was assembled into multimers over a 24h chase period (data not shown). Since vWF in cell lysates was completely Endo H-sensitive, these data indicate that multimer assembly begins in the ER in COS cells. Analysis of intracellular steady state multimers for each mutant revealed a pattern identical with WTvWF (Fig. 7A), consisting of up to four lower molecular weight bands. Failure to observe fully assembled multimers in these constitutively secreting cells is probably due to the rapid transport of vWF to the extracellular compartment after exit from the ER.
Examination of vWF in the cell medium demonstrated that cells transfected with the vWF mutants pV844D, pS743L, and pG742R secreted decreased quantities of vWF (Fig. 7B I, 3, 5, 7, 9, 11, and 13) or with monoclonal anti-BiP antiserum (lanes 2,4, 6, 8, IO, 12, and 14). Molecular weight markers are indicated on the right. WT, wild-type vWF. WTvWF ( l a n e 2), 28 pl of R834W ( l a n e 4 ) , 46 pl of G742E ( l a n e 5), 900 pl of V844D (24X WT) ( l a n e 3), 395 pl of S743L (11X WT) (lane 6), and 214 pl of G742R (6X WT) ( l a n e 7) were immunoprecipitated. vWF bands were quantitated by densitometry using film exposures determined to be in the linear range. Quantities of vWF in each lane were normalized to the fourth anodal band in order to compare quantities of vWF present in the lowest three bands of each lane to each other. while cells transfected with pR834W or pG742E secreted quantities of vWF comparable to WTvWF, consistent with ELISA data (Fig. 2). Both R834W and G742E secreted the full range of multimers. In order to visualize the low levels of secreted vWF for V844D, S743L, and G742R, equal quantities of vWF, measured by ELISA, were immunoprecipitated from cell medium for WTvWF and each mutant, electrophoresed, and analyzed by densitometry (Fig. 7C). V844D, the mutant with the most profound quantitative secretory impairment, also showed the greatest qualitative defect, a relative decrease in the amount of vWF in HMW compared to lower molecular weight forms. S743L and G742R also demonstrated relative decreases in the ratio of HMW to low molecular weight vWF and the severity of the qualitative change paralleled the degree of quantitative loss. V844D contained 7-fold more vWF in the first anodal band than WT, while S743L contained 6-fold and G742R 2-fold more vWF than WT in this band (Fig. 7C). V844D contained a small amount of the second band but only trace quantities of the larger multimers, while higher molecular weight forms were observed in S743L and G742R.

Impaired Transport in uon Willebrand Disease
Analysis of vWF multimers in platelet lysates from patients with the S743L, G742E, and G742R mutations permitted comparison of the results in transfected COS cell with the synthesis of these mutant forms of vWF in vivo. Platelet vWF is contained in a post-Golgi storage pool in the a granule. While plasma vWF from all three patients lacked the highest molecular weight multimers, as is characteristic of Type IIA vWD (Fig. 8, lane 2), examination of platelet lysates demonstrated a relative decrease in HMW vWF multimers for G742R and S743L, but a normal pattern for G742E (Fig. 8, . Also of note, vWF antigen levels measured in plasma were markedly lower than normal for G742R and S743L, but were in the normal range for G742E (see Fig. 7). These observations show a striking correlation with the in vitro transfection results. Platelet lysates from patient A10 (R834W) were previously reported (10) to contain HMW multimers, also consistent with the normal transport of this mutant seen in COS cells. The normal multimer patterns observed in COS cells and in platelets for R834W and G742E may be due to the absence from the transfection system of a factor(s) responsible for multimer loss and protection from this factor(s) within the platelet in vivo. Taken together, these observations suggest that the absence of plasma vWF HMW multimers associated with R834W and G742E occurs in plasma after secretion, while in the remaining mutants the defect results from aberrant intracellular transport.

DISCUSSION
The 7 Type IIA vWD amino acid substitutions identified to date (Fig. 1) are clustered within a small segment of the vWF molecule, but share no other obvious biochemical similarity. Three of the mutations are acidic, one basic, and three neutral. The mutations in patients A4 and A6 are in the same G742 codon, one substituting an acidic and the other a basic residue. No consistent alterations in computer-predicted secondary structure are evident. Of the seven substitutions, two represent C .--, T transitions at CpG dinucleotides, a proposed hot spot for mutation within the human genome (32). Significantly, one of these C .--, T transitions (R834W) was observed in three unrelated families as at least two independent mutational events.
The correct diagnosis and classification of vWD is an important clinical problem, complicated by the extensive phenotypic heterogeneity and the low sensitivity and specificity 4429 of available laboratory tests. Although it is apparent from these studies that a number of different mutations can give rise to Type IIA vWD, the tight clustering of mutations, as well as the occurrence of at least one of these mutations in multiple patients may eventually permit precise DNA-based diagnosis and classification of this disorder.
Although Type IIA vWD has generally been defined as a single subtype of vWD characterized by loss of plasma vWF HMW multimers, analysis of mutant vWF biosynthesis reveals significant heterogeneity. Interestingly, in 1983, Weiss and co-workers (24) proposed three subgroups of Type IIA vWD based on platelet vWF multimer pattern. Our data now provide a molecular basis for these early observations. We thus propose a reclassification of Type IIA vWD into two groups, each characterized by a distinct molecular mechanism. In Group I, single amino acid substitutions result in defects in intracellular transport and decreased vWF secretion. Three of the five Type IIA vWD mutations reported here were transport-defective, resulting in reduced secretion, most pronounced in the HMW uersuS low molecular weight vWF multimer species. In Group 11, synthesis and transport are indistinguishable from WTvWF, and the loss of plasma HMW vWF multimers may result from extracellular proteolysis after secretion. No obvious structural motifs define each group or appear to differentiate the two groups. Group I mutations can result from either acidic (V844D), basic (G742R), or neutral (S743L) substitutions. Of note, the Group I1 mutation G742E occurs in the same codon as the Group I defect G742R and immediately adjacent to a second Group I mutation, S743L.
The Group I1 mutations, M34W and G742E, were both transported and secreted normally from COS cells. In addition, platelet lysates from patients with these mutations contained intact HMW multimers. Interestingly, endothelial cells from a Type IIA vWD patient produced and secreted the full range of multimers, but larger multimers were unstable in culture medium (33). Additionally, in some Type IIA vWD patients, collection of Type IIA vWD blood directly into protease inhibitors has been reported to preserve large multimers (10, 11,34). The first patient reported to have this type of defect (10) (A10 in this study) was found to have the previously identified R834W mutation (13). This same mutation has been observed in at least three unrelated families on two different chromosomal backgrounds, indicating that it has arisen at least twice independently. Taken together, these data strongly suggest that this single amino acid change is an authentic Type IIA vWD mutation and that extracellular proteolysis is responsible for the loss of vWF multimers in these patients. The proteolytic cleavage site generating the 176-kDa carboxyl-terminal vWF fragment associated with some Type IIA vWD has been localized (35) to the TyrM2-Metff13 peptide bond in the vicinity of the clustered Type IIA vWD mutations (Fig. 1); however, the responsible protease has not yet been identified. The Group I mutations V844D, S743L, and G742R demonstrated defects in transport resulting in quantitative and qualitative defects in vWF. While other examples of genetic disease resulting from impaired intracellular transport have been described (36-38), they are typically characterized by purely quantitative defects and recessive inheritance. In a heterozygote, expression of the normal allele is generally unaffected by failure of the mutant protein to be transported. For vWD, the multimeric nature of vWF provides a mechanism for a mutation of this type to function in a "dominantnegative" fashion. The dominant inheritance and qualitative loss of HMW multimers characteristic of Type IIA vWD could result from selective retention of HMW forms containing greater numbers of mutant subunits. While an increased interaction of the Group I mutants with BiP was not evident, binding to other resident ER proteins cannot be excluded (28).
Both subgroups of Type IIA vWD may result from a spectrum of alterations to the same sensitive surface structure. While vWF has a very high cysteine content (8.3% of the total amino acids), there are no cysteines between residues 696 and residue 905 (17). Additionally, epitope mapping studies with a panel of monoclonal antibodies suggest that this region may contain an immunodominant epitope(s) (39). As noted above, this domain appears highly accessible to plasma protease(s) (35). Taken together, these data suggest that this region is highly exposed on the surface of vWF. Group I substitutions may disrupt this structure, increasing the interactions of this domain with ER proteins, and resulting in markedly delayed transport through the secretory pathway. Other substitutions (Group 11) may escape detection by resident ER proteins and thus may allow vWF secretion but result in increased sensitivity to extracellular protease(s1 within the same exposed region.