Apolipoprotein H (apoH), also known as [beta]2-glycoprotein-I, is considered to be a cofactor for the binding of certain antiphospholipid autoantibodies to negatively charged phospholipids. Genetically determined structural abnormalities in the lipid binding domain(s) of apoH can affect its ability to bind lipid and consequently the production of the autoantibodies. In this study we have identified two common structural mutations at codons 316 and 306 in the fifth domain of apoH which rendered apoH unable to bind to negatively charged phosphatidylserine (PS). The missense mutation at codon 316 (TGG -> TCG) replaces Trp316 with Ser316 and disrupts the integrity of four highly conserved hydrophobic amino acids sequence at positions 313-316, which is a potential protein-lipid hydrophobic interaction site. The missense mutation at codon 306 (TGC -> GGC) involves the substitution of Cys306 by Gly306 which causes the disruption of a disulfide bond between Cys281 and Cys306 and affects the normal configuration of the fifth domain of apoH that appears to be critical for clustering positively charged amino acids along with the four hydrophobic amino acids sequence. ApoH from the two homozygotes (Ser316/Ser316) and all seven compound heterozygotes (Ser316/Gly306) failed to bind to PS; all heterozygotes at one or the other sites and wild type showed normal PS binding. These data indicate that the fifth domain of apoH harbors the lipid binding region. An estimated 2 million Caucasians in the United States, who are compound heterozygotes for the two mutations, may be precluded from producing apoH-dependent antiphospholipid autoantibodies.
Apolipoprotein H (apoH, protein; APOH, gene), also referred to as [beta]2-glycoprotein I is a single chain glycoprotein of 326 amino acids as determined directly from purified protein (1 ) and subsequently confirmed by deduced amino acid sequence by cDNA cloning and sequencing (2 -4 ). The cDNA sequence predicts 345 amino acids which include 19 hydrophobic signal peptide residues not present in the mature protein. The apoH protein shows extensive internal homology with five consecutive homologous segments of ~60 amino acids each. These segments are referred to variously as GP-I domains (because they were first found in [beta]2-glycoprotein I (5 ), Sushi domains (6 ), SCRs (Short consensus repeats) or CCP (complement control protein) repeats (7 ). Such domains or repeats are commonly found in a number of complement component proteins, as well as in non-complement proteins (5 ). Based upon the predicted structure of the apoH protein from cDNA sequence, there are 22 cysteine residues in the human apoH. These positions are conserved in bovine (8 ), rat (9 ), mouse (10 ) and dog (11 ) apoH proteins which also consist of five GP-I domains. In contrast to bovine apoH where all 11 disulfide bonds have been mapped (8 ,12 ), the positions of only six disulfide bonds have been determined in human apoH (1 ,13 ). However, considering the homology of amino acid sequences and the positions of cysteine residues between the two species, concluding that the positions of all 11 disulfide bonds in human apoH are identical to those seen in bovine apoH is reasonable. In contrast to the first four GP-I domains which consist of four Cys residues each, the fifth and most C-terminal domain has six Cys residues and the resulting three disulfide bonds have been mapped (13 ).
ApoH has been implicated in a variety of physiologic pathways including lipoprotein metabolism (14 ), coagulation (15 ) and more recently in the production of antiphospholipid autoantibodies (aPA) (16 ). ApoH is considered to be a required cofactor for anionic phospholipid binding by the aPA found in sera of many patients with lupus and primary antiphospholipid syndrome (APS) (17 -19 ) but it does not seem to be required for the reactivity of aPA associated with infections (20 ). These studies suggest that the apoH-phospholipid complex forms the antigen to which aPA are directed (21 -22 ). Recently, however, the presence of autoantibodies to phospholipid-free apoH has been shown in patients with primary APS (23 -25 ), indicating that there may be two epitopes on apoH against which these autoantibodies are directed. Although the structural domains of apoH which bind to anionic phospholipids are unknown, studies have indicated that the expressed fifth domain is in itself capable of binding anionic phospholipids (13 ) and it may be critical for lipid-protein interaction (26 -27 ).
DNA samples from two individuals carrying the APOH*3W allele were used to PCR amplify the coding region of the entire fifth domain of apoH (amino acids 243-326). The observed PCR product was more than twice the expected size of 401 bp fragment, which indicated the presence of an intron. Direct DNA sequencing from both directions identified an intron which interrupted the codon 309 sequence, confirming that the fifth domain of apoH is encoded by the last two exons 7 and 8 (data not shown). Subsequently, new reverse (APOH 9) and forward (APOH 10) primers were designed in intron 7 to amplify the exons 7 and 8, respectively, in combination with originally designed primers (APOH 2, APOH 5) for the fifth domain sequence. A point mutation was identified at the second position of codon 316 in exon 8 (TGG -> TCG) which replaced Trp316 by Ser316 (Fig. 1 A). This point mutation created a restriction site for the BstBI enzyme and, therefore, a PCR based restriction analysis was performed to screen the remaining individuals who were carriers of the APOH*3 allele. The uncut fragment size of 148 bp corresponded to the Trp316 wild type and 86 and 62 bp corresponded to the Ser316 mutant type (Fig. 1 B). All samples which reacted with mAb 3D11 (APOH*3W allele) had the Ser316 mutation and none of the mAb 3D11 negative samples had this mutation, strongly confirming that a missense mutation at codon 316 is responsible for the APOH*3W reactivity with mAb 3D11. Only two of the 12 samples, which showed reactivity with mAb 3D11 and originally classified as APOH 3-3 by polyclonal antibody, were homozygous for the Ser316 mutation; the remaining samples were heterozygous having both the wild and mutant types (Trp/Ser). Direct DNA sequencing of the remaining seven APOH exons from an individual homozygous for the Ser316 allele did not identify any other mutation (data not shown).
Since mAb 3G9 recognized the products of all APOH alleles in our previous study (30 ), 3G9 was used to detect PS binding by apoH in the ELISA. The two individuals homozygous for the Ser316 mutation showed no binding with PS (Fig. 2 ) but apoH binding to PS was detected in eight individuals who were heterozygous for this mutation (Fig. 2 ), indicating that homozygosity of the Ser316 mutation is essential to negate lipid binding. However, apoH from seven additional individuals who were heterozygous for the Ser316 mutation also showed no PS binding (see Fig. 2 ) which suggested that the presence of an additional mutation(s) is required to interact with the Ser316 mutation in its heterozygous form to negate PS binding.
DNA from one of the seven heterozygotes with the Ser316 mutation which revealed no apoH binding with PS was subjected to sequencing to identify the second putative functional mutation. A missense mutation at codon 306 (TGC -> GGC) was found in exon 7 which replaced Cys306 with Gly306 (Fig. 3 A). This point mutation also created a restriction site for the CviJI enzyme and, therefore, a PCR based screening method was devised to screen all the remaining individuals who were heterozygous for the Ser316 mutation (Fig. 3 B). While none of the eight heterozygotes for the Ser316 mutation who showed PS binding had the Gly306 mutation, all seven heterozygotes with the Ser316 mutation who showed no PS binding did have the Gly306 mutation (Fig. 2 ). These data show that the seven individuals with the Ser316 and Gly306 mutations are compound heterozygotes. To determine if the Gly306 mutation is widely distributed in the general population or is present only in combination with the Ser316 mutation we examined 331 DNA samples. We found that the Gly306 mutation was present in ~8% of the general US Caucasian population irrespective of the presence or absence of the Ser316 mutation. Altogether, we identified 19 individuals who were heterozygous for the Gly306 mutation and did not carry the Ser316 mutation; apoH from all of these 19 samples showed binding to PS (Fig. 2 ), suggesting that one copy of this mutation is not sufficient to prevent apoH binding to PS. No individual homozygous for the Gly306 mutation was observed in this investigation. This is not a surprising observation because the expected frequency of the Gly306/Gly306 homozygous in the 331 samples examined was only 0.5 or 0.15%.
The fifth domain of human apoH consists of six Cys residue and the resulting three disulfide bonds have been mapped (13 ) to show linkage between Cys245 and Cys296, Cys281 and Cys306 and between Cys288 and Cys326 (Fig. 4 A). We have used this information to construct GP-I domain of the fifth repeat (Fig. 4 B). The formation of the disulfide bonds between Cys281 and Cys306 brings five additional positively charged amino acids (four Lys and one His) at positions 305, 308, 310, 317 and 324 close to the highly positively charged sequence of Lys-Asn-Lys-Glu-Lys-Lys at positions 282-287. This model clusters nine positively charged amino acids (eight Lys at positions 282, 284, 286, 287, 305, 308, 317 and 324, and one His at position 310) in the fifth domain and makes it the most plausible configuration for binding with negatively charged phospholipids. In addition to the negatively charged amino acids in this loop, there is a hydrophobic sequence at positions 313-316 (Leu, Ala, Phe, Trp) in this loop and that may serve as an interaction site with the fatty acid chains of the phospholipid aggregate (13 ). The substitution of Trp316 by Ser316 appears to affect the hydrophobic bonding between apoH and PS. On the other hand, the substitution of Cys306 by Gly306 would result in the breakage of the second disulfide bond between Cys281 and Cys306 and will disrupt the normal configuration of the inner loop (Fig. 4 C) which appears to be necessary to cluster positively charged amino acids along with the four hydrophobic amino acids at positions 313-316.
This study was undertaken to identify unique structural mutations in the APOH gene which affect the binding of apoH to anionic phospholipids. We sought functional mutations in the fifth domain of APOH as it seems important for ionic interaction with anionic phospholipids (13 ,26 -27 ). Indeed we have identified two naturally occurring structural mutations at codons 306 and 316 in the fifth domain which appear to be functionally related with the apoH ability to bind lipid. Both mutations reside in the C-terminal of apoH only nine amino acids apart. The mutation at codon 316 involves the substitution of one of the four hydrophobic amino acids at position 313-316 (Leu, Ala, Phe, Trp) which are highly conserved in primates including bovine (8 ), rat (9 ), mouse (10 ), dog (11 ) and human (1 -4 ). This hydrophobic sequence is suggested to be involved in hydrophobic bonding with the fatty acid chains of the anionic phospholipid aggregate (13 ). Our data indicate that the integrity of this hydrophobic sequence is critical for binding to anionic phospholipids because serum apoH from two individuals who were homozygous for the Ser316 mutation did not bind to PS (see Fig. 2 ). Apparently, the homozygosity of the Ser316 mutation is necessary to prevent apoH binding to anionic phospholipids because individuals heterozygous for this mutation showed complete lipid binding (Fig. 2 ). However, when the Ser316 mutation occurs in conjunction with the Gly306 mutation it also prevents the binding of apoH to PS (Fig. 2 ). Taken together, these data suggest that these mutations operate in a recessive fashion where both chromosomes must carry either the same mutation at one site (homozygous for the Ser316 mutation) or different mutations at two different sites (compound heterozygous for the Ser316 and Gly306 mutations). Although no example of the Gly306/Gly306 was observed in this investigation, we predict apoH from such homozygous individuals will not bind to anionic phospholipids.
The Gly306 mutation does not seem to have direct impact on protein-lipid interaction because this does not involve any charged amino acid. However, the substitution of Cys306 by Gly306 disrupts the critical disulfide bond which seems important in clustering several positively charged amino acids (see Fig. 4 B and C). By molecular modeling of the fifth GP-I domain and highlighting the lysine residues Steinkasserer et al. (13 ) have shown that lysine at positions 282, 284, 286, 287 and 324 may form a lipid binding region since in the model they were clustered at the distal end. The disruption of the disulfide bond between Cys306 and Cys281 would affect the positioning of Lys324 and also move Lys308, Lys305 and His310 away from the positively charged cluster along with the four hydrophobic amino acids sequence at positions 313-316 (see Fig. 4 C). Recent findings that a clipped apoH molecule, which was cleaved between Lys317 and Thr318, failed to bind to cardiolipin and lost its cofactor activity (26 ) and a synthetic peptide containing Cys287-Cys288 sequence inhibited the binding of apoH to cardiolipin (27 ) provide further evidence that the proposed configuration of the fifth domain is essential for lipid binding and structural changes in this region, as shown here by the Gly306 and Trp316 mutations, would affect the normal configuration of apoH (Fig. 4 ).
Notably, the lack of apoH binding to PS associated with the codons 306 and 316 mutations is confined to individuals who are APOH 3-3 homozygous on IEF-immunoblot gels, using polyclonal antiserum, and also showed reactivity with mAb 3D11. In the general Caucasian population of the United States the frequency of the APOH 3-3 phenotype is 1.66% (calculated in 661 individuals, manuscript in preparation) and most of them (1.36%) show reactivity with mAb 3D11 and are also carriers of the Ser316 mutation. Of those with the APOH 3-3 phenotype, 0.91% are either compound heterozygous for the Ser316/Gly306 mutations or homozygous for the Ser316 mutation and show no binding with lipid (see Fig. 2 ). If we assume that there are 220 million Caucasians in the United States then 2 million (0.91%) may be precluded from the production of apoH-dependent aPA and, thus, be protected from apoH-dependent thrombosis and related disorders.
In summary, we have identified two naturally occurring structural mutations in the fifth GP-I domain of apoH which affect the binding of apoH to anionic phospholipds. Our data show that the normal configuration of the fifth GP-I domain is critical for protein-lipid interaction.
Initial screening of APOH protein polymorphism with polyclonal anti-apoH was carried out as described originally (28 ) and modified for the mAb 3D11 (30 ).
DNA samples from individuals with known protein typings were subjected to PCR to amplify the entire coding sequence of the fifth domain using a forward primer (APOH 2) in the 5' flanking intron, 5'-GTGTAGGTGTACTCATCTACTGTG-3' (31 ) and a reverse primer (APOH 5) in the non-coding 3' flanking region, 5'-TGGATGAACAAGAAACAAGTG-3'. DNA sequencing from both directions identified an intron (termed intron 7) and intron-exon boundaries corresponding to exons 7 and 8. Subsequently, exon 7 was amplified using APOH 2 (forward primer) in conjunction with APOH 9 (reverse primer), 5'-CAAGTGGGAGTCCTAGCTAA-3'. Similarly, exon 8 was amplified using APOH 10 (forward primer), 5'-TTGTTCCCTTAGAATGTTTAT-3' in conjunction with APOH 5 (reverse primer). Genomic DNA (1 µg) was subjected to PCR amplification using forward and reverse primers specific for each exon in 50 µl of reaction mixture containing 0.3 M of each primer, 200 µM of each dNTP (Pharmacia), 5 µl of 10* reaction buffer (100 mM Tris-HCl pH 9.0, 500 mM KCl, and 1% Triton X-100 pH 9.0), 3.5% DMSO and 1.25 U of Taq DNA polymerase. After initial denaturing DNA for 5 min at 95oC, reaction mixture was subjected to 30 cycles of denaturation for 1 min at 95oC; 1.5 min annealing at 57oC (for exon 7) and at 49oC (for exon 8) and 2 min extension at 72oC. DNA sequencing of purified PCR product was carried out directly on double-stranded DNA by dideoxynucleotide chain termination method using PCR Product Sequencing Kit and sequenase version 2.0 (US Biochemicals) with 35S-labeled dATP (Dupont). Sequenced products of all fragments were migrated on 6% denaturing (7 M urea) glycerol-tolerant sequencing gel (US Biochemicals). Vacuum dried gels were autoradiographed on Kodak X-ray film for 24 h.
Population screenings of two newly identified mutations in exon 7 (codon 306) and exon 8 (codon 316) were carried out by restriction enzyme digestion of PCR products with CviJI (Molecular Biology Resources) and BstBI (New England Biolabs) enzymes, respectively, followed by electrophoresis on either 9% polyacrylamide (codon 306 mutation) or 2% Nusieve agarose (codon 316 mutation) gels.
Phospholipid binding by apoH was measured by ELISA as previously described (29 ). Briefly, flat bottom Titertek microtiter plates IICN, Horsham, PA) were coated with 30 µl of a 50 µg/ml solution of cardiolipin or PS (Sigma, St. Louis, MO) diluted in methanol:chloroform (3:1) and dried under a stream of nitrogen. The plates were given three 2 min washes with TRIS-buffered NaCl (TBS; 0.02 M TRIS, 0.15 M NaCl, pH 7.3) after antigen coating, blocking, sera, antibody and conjugate incubations. The plates were blocked with 10% BSA (Sigma). Each serum was diluted (1:80 in 1% BSA/TBS) and incubated in triplicate wells (50 µl/well) for 60 min. Then 50 µl of mAb 3G9 (0.5 µg/ml) was incubated in each well for 30 min followed by a 30 min incubation with an alkaline phosphatase conjugated rabbit antimouse IgG. Development substrate (p-nitrophenylphosphate tablets in 10% w/v diethanolamine, 5mM MgCl2, pH 9.8) was added to each well (50 µl), and the plates were incubated in the dark at 37oC for 45 min. Color development was stopped by the addition of 75 µl of 3 M NaOH, then the optical density of each well was measured at 405 nm.
We greatly appreciate the clerical assistance of Ms Kimberley Smithwick. This work was funded by a National Institutes of Health grant HL54900.
APOH, apolipoprotein H; aPA, antiphospholipid autoantibodies; PS, phosphatidylserine; IEF, isoelectric focusing; mAb, monoclonal antibody.
Human Molecular Genetics Pages
©
Introduction
Results
Molecular basis of the APOH*3W allele
Effect of the codon 316 mutation on lipid binding
Identification of codon 306 mutation and its role in lipid binding
Consequence of codons 306 and 316 mutations on lipid binding
Discussion
Material And Methods
IEF/immunoblotting
PCR and DNA sequencing
DNA polymorphisms
Phospholipid binding of apoH by ELISA
Acknowledgements
Abbreviations
References
REFERENCES
This page is maintained by OUP admin. Last updated Fri Jan 31 11:45:11 GMT 1997. Part of the OUP Journals World Wide Web service.Copyright Oxford University Press, 1996