A Tyr346→Cys substitution in the interdomain acidic region a1 of factor VIII in an individual with factor VIII:C assay discrepancy

Authors


Dr G. Kemball-Cook, Haemostasis Research, MRC Clinical Sciences Centre, Imperial College Medical School, Hammersmith Hospital, Du Cane Road, London, W12 0NN, UK. E-mail: geoffrey.kemball-cook@csc.mrc.ac.uk

Abstract

Summary. The interdomain acidic region a1 is a unique structural feature of coagulation factor VIII (FVIII) and may mediate the proteolytic activation of FVIII and the inactivation of FVIIIa. We report an individual with a Tyr346→Cys substitution within region a1, who presented with a one-stage FVIII activity (FVIII:C) of 0·34 iu/ml (normal range 0·5–2·0) but normal two-stage FVIII:C and FVIII antigen values. In a factor Xa (FXa)-generation assay for FVIII in which the activation time with thrombin was varied, the variant plasma showed normal FVIII:C at both short and long activation times. However, at intermediate activation times the FXa generation of the variant plasma was less than that of normal pooled plasma. In a modified one-stage FVIII:C assay in which partially purified FVIII was activated with thrombin at low concentrations, the variant FVIII showed less activation than wild-type FVIII, although this defect corrected with increasing concentrations of thrombin. When partially purified variant FVIII was activated with a large molar excess of thrombin, the subsequent rate of decay of FVIII:C was greater for variant FVIII. The complex defects in activation and inactivation displayed by FVIII Tyr346→Cys support the hypothesis that the a1 sequence is a key regulator of FVIII activity.

Coagulation factor VIII (FVIII) is a multidomain protein related through domain homology to several other proteins both associated with coagulation, such as factor V (FV), and not associated (caeruloplasmin, hephaestin, and various lectins). However, a unique feature of FVIII is the presence of three short sequences (30–40 residues) of mainly acidic residues separating the A1 and A2 domains, the A2 and B domains, and the B and A3 domains (Fig 1). These acidic interdomain regions, termed a1, a2 and a3, respectively, have no known homology with any other protein, are close to important proteolytic cleavage sites within FVIII (Eaton et al, 1986) and together contain six highly conserved tyrosine residues that are sulphated during post-translational processing (Pittman et al, 1992). It has been proposed that these tyrosine residues and the surrounding acidic residues constitute important sites through which FVIII interacts with other components of the coagulation pathway in ways which are unique and essential to FVIII procoagulant function (reviewed in Lenting et al, 1998).

Figure 1.

Diagram showing the domain structure of FVIII. The three homologous A domains are separated by unique acidic sequences a1, a2 and a3 (in grey) and the B domain, then followed by the two homologous C domains. On thrombin activation, there is a cleavage between a1 and the A2 domain, and the B domain and a3 sequence are excised, forming a heterotrimeric FVIIIa molecule. The A2 domain spontaneously dissociates with loss of FVIIIa functional activity.

Sulphated tyrosine residue 346 lies within region a1 (approximately residues 337–372) and is functionally associated with a series of proteolytic cleavage sites. Substitution of Tyr346 with phenylalanine by in vitro mutagenesis yielded a FVIII variant which was cleaved less efficiently by thrombin at Arg372 and therefore showed delayed activation (Michnick et al, 1994). Tyr346 also lies close in sequence to Arg336 which is a cleavage site for factors IXa, Xa (FIXa, FXa) and activated protein C (Eaton et al, 1986). Cleavage of activated FVIII (FVIIIa) at Arg336 results in the release of the a1 region which in turn promotes dissociation of the A2 domain and inactivation of the FVIIIa heterotrimer (reviewed in Saenko et al, 1999). The a1 sequence may therefore also be a key regulator of this inactivation process (Fay et al, 1993). Finally, Tyr346 and the surrounding a1 region may also comprise a binding site for the tenase-complex substrate factor X (FX) (Lapan & Fay, 1997). Thus, the consequences of individual amino acid substitutions within this sequence are difficult to predict, as they might affect either FVIII activation or inactivation, or conceivably both. To date, no naturally occurring mutations in the a1 sequence resulting in amino acid change have been reported on the FVIII mutation database (http://europium.csc.mrc.ac.uk).

In this study, we report an individual with a Tyr346→Cys substitution within the a1 region who presented with a reduced one-stage FVIII activity (FVIII:C) assay but normal two-stage FVIII:C and FVIII antigen (FVIII:Ag) assays. This relationship between functional assay results is the converse of recently reported FVIII variants in which the one-stage FVIII:C exceeds the two-stage FVIII:C (Pipe et al, 2001). We demonstrate that the variant FVIII shows complex abnormalities in activation and inactivation which are likely to explain the discrepancy.

Materials and methods

Clotting screens and standard one- and two-stage FVIII:C assays.  Clotting screens were performed according to published methods (Mercuri et al, 2001). One-stage factor VIII:C assays were performed by mixing 50 μl dilutions of test or reference plasmas (Technoclone, Dorking, UK) with 50 μl of FVIII-deficient plasma (Technoclone), 50 μl of Actin FS activated partial thromboplastin time (APTT) reagent (Dade/Sysmex, Milton Keynes, UK). Clotting times were measured after the addition of 50 μl of 25 mmol/l CaCl2. Two-stage FVIII:C assays were performed with a factor VIII chromogenic assay kit (Dade-Behring, Marburg, Germany) in which 50 μl dilutions of test or reference plasmas were mixed with 50 μl of the FX reagent and 50 μl of the FIXa/thrombin/CaCl2/phospholipid reagent for 150 s. Stopping buffer/chromogenic substrate (500 μl) was added and the reaction mixture was incubated for 60 s. FXa generation was determined by an end-point measurement of absorbance at 405 nm. Both one- and two-stage assays were performed on the Sysmex CA6000. FVIII:Ag was determined on plasma and partially purified FVIII by enzyme-linked immunosorbent assay (ELISA, Immunozym FVIII:Ag kit; Immuno, Sevenoaks, UK). The FVIII concentration of the partially purified preparations was calculated on the basis that 1 iu/ml FVIII:Ag was equivalent to 0·5 nmol/l.

FVIII mutation detection.  The promoter and the entire coding sequence of the FVIII gene were polymerase chain reaction (PCR)-amplified from genomic DNA using primers and experimental conditions described previously (Vidal et al, 2001). DNA sequencing was performed on a model 3700 DNA analyser using an ABI PRISM Big Dye V2 terminator ready reaction kit (Applied Biosystems, Foster City, CA, USA).

Plasma FVIII activity in a FXa-generation assay.  Aliquots of 50 μl of variant and pooled normal plasmas, diluted 30-fold in 50 mmol/l Tris, 150 mmol/l NaCl, pH 7·4 containing 1 mg/ml human albumin (Zenalb 20; Bio Products Laboratory, Elstree, UK; TBSA) were mixed with 100 μl of the FIXa/FX/phospholipid reagent from a Coatest VIII:C kit (Chromogenix/Quadratech, Epsom, UK) to which thrombin (Haematologic Technologies, Essex Junction, USA) had been added to a final concentration of 1·5 nmol/l. The reaction mixture was incubated at 37°C for the time intervals indicated to allow FVIIIa generation. CaCl2 (50 μl 25 mmol/l) containing 1·25 μmol/l recombinant desulphatohirudin (Giba-Geigy/Novartis, Camberley, UK) was then added and the mixture was incubated for a further 5 min at 37°C. FXa generation was then determined by adding 100 μl of the Coatest VIII:C S-2222/I-2581 chromogenic substrate reagent and measuring colour generation kinetically with a Thermomax plate reader (Molecular Devices, Wokingham, UK).

Analysis of FVIII activation and decay by Western blot. Partially purified FVIII was derived from variant (FVIII-Y346C) and normal pooled plasma (FVIII-WT) by cryoprecipitation, reduction and chromatography according to a previously published method (O'Brien & Tuddenham, 1989). A reaction mixture containing a 1 nmol/l dilution of this material was treated with 30 nmol/l thrombin (Haematologic Technologies) and incubated at 37°C. At intervals, 25 μl aliquots were precipitated in ice-cold acetone, resuspended in reducing sodium dodecyl sulphate (SDS) sample buffer and resolved on a 10% Bis-Tris SDS-PAGE (polyacrylamide gel electrophoresis) precast gel (Novex, Frankfurt, Germany). Following electroblotting transfer, the blots were probed with monoclonal anti-A2 (a gift from Dr A. Goodall, Leicester, UK) or monoclonal anti-a3 (C7F7, Dr G. Vehar, Genentech, San Francisco, USA). Aliquots of FVIII-Y346C that had been partially purified by cryoprecipitation alone without subsequent reduction were studied before and after thrombin treatment by Western blot using both non-reducing and reducing sample buffers.

Thrombin activation of partially purified plasma FVIII.  Sensitivity of variant and wild-type FVIII to low thrombin concentrations, and time courses of activation and decay at a higher thrombin level were performed and activities assessed by a modified one-stage FVIII:C assay.

For the study of thrombin sensitivity, partially purified FVIII-WT and FVIII-Y346C were diluted to 1·5 nmol/l in TBSA and 10 μl incubated with 10 μl human thrombin (Haematologic Technologies) at final thrombin concentrations in the range 0–2·75 nmol/l for 7 min at room temperature, followed by assay in a modified one-stage assay. Each 20 μl reaction mixture was diluted further (× 30–× 100) and 100 μl of the diluted mixture added to a mixture of 100 μl each of FVIII-deficient plasma (Diagnostic Reagents, Thame, UK) and activator/PL reagent (Instrumentation Laboratory, Warrington, UK) which had been preincubated together at 37°C for 5 min. Clotting times were measured after the immediate addition of 100 μl of 25 mmol/l CaCl2 using a Coag-a-Mate XC coagulometer (Organon Teknika, NC, USA). FVIII:C following thrombin treatment was determined from the mean of triplicated experiments using a standard curve derived from normal pooled plasma (designated 1·0 iu/ml).

Secondly, the time course of FVIII activation and decay at a higher thrombin concentration (25 nmol/l) was studied: partially purified variant and wild-type FVIII preparations were diluted with TBSA to 1·5 nmol/l and, to ensure rapid and complete activation of FVIII followed by measurable decay, 50 μl diluted FVIII samples were mixed with 5 μl human thrombin at a final thrombin concentration of 25 nmol/l. The reaction mixture was then subsampled at the indicated time points and FVIII:C in the aliquots determined by modified one-stage assay after TBSA dilution as above. FVIII:C following thrombin treatment was again determined from the mean of triplicated experiments using a standard curve derived from normal pooled plasma.

Results

Patient details and plasma clotting assays

The index case was a 71-year-old genotypic male who was investigated for a prolonged APTT. A reduced one-stage FVIII:C was demonstrated and a provisional diagnosis of mild haemophilia A was made. However, the patient had previously undergone successful gender reassignment surgery without FVIII replacement and there was no family history of abnormal bleeding. During further investigation, two-stage FVIII:C and FVIII:Ag assays were performed and found to be normal: the discrepancy in FVIII:C-values was consistent across three samples collected at different times (Table I).

Table I.  APTT, FVIII:C and FVIII:Ag values reported on samples collected from the patient on three separate occasions.
 Sample 1Sample 2Sample 3MeanNormal range
  1. ND, not determined.

APTT (s)3332343324–32
1-stage FVIII:C (iu/ml)0·350·350·330·340·5–2·0
2-stage FVIII:C (iu/ml)1·060·921·321·100·5–2·0
FVIII:Ag (iu/ml)NDND1·18[1·18]0·5–2·0

DNA sequence analysis

The patient was hemizygous for an A to G transversion in codon 346 predictive of a Tyr to Cys substitution. The remaining coding sequence was normal.

Plasma FVIII activity in the FXa-generation assay

As Tyr346 has been provisionally identified by site directed mutagenesis as a residue important for the interaction of FVIII with thrombin during the cofactor's activation, we investigated the relationship between incubation time with thrombin and FVIII:C in plasma using a modified two-stage FXa-generation assay. FVIII activity in variant and pooled normal plasmas was determined by the ability of FVIIIa to generate FXa in the presence of FIXa and phospholipid. Theresults of duplicate experiments are shown in Fig 2. After incubation times of 15 and 30 min the activities of FVIII-WT and FVIII-Y346C were similar, however, after incubation times of between 0·5 and 5 min inclusively, FVIII-Y346C showed significantly lower activity (P < 0·05).

Figure 2.

FXa generation rate of FVIII-Y346C and FVIII-WT after thrombin treatment in which incubation time with thrombin was varied. The data are plotted as the mean and SEM of results derived from two independent experiments (*P < 0·05).

Analysis of FVIII activation by Western blotting

To identify any gross abnormalities in the activation of FVIII-Y346C, we treated partially purified preparations of FVIII with a large molar excess of thrombin and followed the subsequent proteolysis by Western blotting. Both FVIII-Y346C and FVIII-WT showed progressive proteolysis of FVIII and liberation of the B-domain with increasing incubation periods with thrombin. With an approximately 10-fold molar excess of thrombin, there was no detectable difference in B domain release between FVIII-Y346C and FVIII-WT (data not shown).

As mutation in secreted proteins, resulting in novel cysteine residues, can disrupt protein folding or function through the formation of illegitimate disulphide bonds with other plasma proteins or within the molecule, we also examined intact and thrombin-treated FVIII-Y346C that had undergone cryo-precipitation without subsequent reduction. There were no differences in the apparent molecular weights of intact or thrombin-treated FVIII-Y346C on Western blots performed in reducing and non-reducing conditions (data not shown).

Thrombin activation of partially purified variant FVIII

As there was no demonstrable difference by SDS-PAGE in FVIIII cleavage with a large molar excess of thrombin, we also studied functional activation at much lower thrombin concentrations. Functional activation of FVIIIa increased over the range of thrombin concentrations studied but at thrombin concentrations below 1 nmol/l, activation of FVIII-Y346C was less than that of FVIII-WT, although both were used at the same FVIII:Ag concentration (P < 0·05). At thrombin concentrations of 1–2·75 nmol/l, FVIII-Y346C and FVIII-WT behaved similarly (Fig 3).

Figure 3.

Activation of partially purified FVIII-Y346C and FVIII-WT (diluted to the same FVIII:Ag level) following activation with suboptimal concentrations of thrombin. FVIII:C levels are plotted as the mean and SEM of three independent experiments. Below 1 nmol/l thrombin, significant differences are seen (*P < 0·05).

As the interdomain region a1 has been implicated in maintaining stability of the FVIIIa heterotrimer, we further studied the decay of FVIIIa where FVIII-WT and FVIII-Y346C had been completely activated with a large molar excess of thrombin. After 5 min incubation with thrombin, FVIII:C for FVIII-WT was 13·0 ± 0·7 iu/ml (mean ± SEM) and for FVIII-Y346C was 2·0 ± 0·2 iu/ml. Thereafter, both FVIII-WT and FVIII-Y346C showed a logarithmic (single-order) decline in activity over time. The decay rate of FVIIIa-Y346C was faster (t1/2 = 185 s) than for FVIIIa-WT (t1/2 = 370 s) (Fig 4).

Figure 4.

Decay of FVIII:C following treatment of FVIII-Y346C and FVIII-WT with 25 nmol/l thrombin. The data are plotted as the mean and SEM of results from three independent experiments.

Discussion

The Tyr346→Cys substitution is of particular interest because it lies in the interdomain acidic region a1 of FVIII which mediates the proteolytic activation and degradation of FVIII and may promote stability of the active heterotrimer against non-proteolytic dissociation. This substitution is associated with the phenotype of normal two-stage FVIII:C and FVIII:Ag, but reduced one-stage FVIII:C (approximately 35% of normal levels). To date, discrepancy of this type has not been reported to the haemophilia A mutation database (Kemball-Cook et al, 1998; http://europium.csc.mrc.ac.uk), although FVIII Tyr346→Cys has been described in abstract form by us (Mumford et al, 2001) and others (Goodeve et al, 2001): the individuals described in the latter report showed a similar assay discrepancy to the subject of the current study. As two-stage FVIII:C assays are not commonly performed in many haemophilia centres in the assessment of individuals with suspected haemophilia, other reported FVIII variants may display a similar assay discrepancy but remain unrecognized.

When we studied FXa generation in our subject's plasma (Fig 2), although the rate of FXa generation was normal after 10–15 min incubation with thrombin, at shorter times the rate was as low as 40% of that found in normal plasma. This result was strikingly similar to the result of the one-stage FVIII:C assay. This indicated that in this plasma-based assay, our subject's apparent FVIII:C was dependent on experimental variables such as thrombin incubation time, and any single time-point assay may therefore yield an apparent FVIII:C value that is normal or abnormal, depending on the assay conditions. Failure of the variant FVIII to generate the same peak rate of FXa generation as WT FVIII may reflect the enhanced decay of variant FVIIIa as seen with the partially purified material (Fig 4).

When we studied the thrombin activation of partially purified FVIII, immunoblotting indicated that there were no gross defects in thrombin activation and there was no illegitimate disulphide bond formation, resulting from the introduced cysteine residue, which might disrupt function. In addition, the normal FVIII:Ag value in the subject's plasma suggests that no folding or secretion defect results from the substitution.

However, when we studied the activation of partially purified FVIII by low concentrations of thrombin (< 1 nmol/l), FVIII-Y346C showed reduced activation compared with FVIII-WT (Fig 3). One explanation for this is that the Tyr346→Cys substitution causes a subtle reduction in the affinity of thrombin to FVIII and impairs activation. When the free thrombin concentration is raised, this binding defect may be overcome and FVIII-Y346C shows normal activation. This is consistent with a role for Tyr346 as a mediator of the interaction between thrombin and FVIII (Lenting et al, 1998). Our data support the findings of Michnick et al (1994) that a recombinant FVIII variant with a Tyr346→Phe substitution showed delayed thrombin activation.

In the high-dose (25 nmol/l) thrombin-activation experiment using partially purified FVIII, we expected both FVIII-WT and FVIII-Y346C to show similar complete activation at very early stages in the assay. The subsequent decline in FVIII:C measured in the assay mixture therefore reflected FVIIIa inactivation alone. It has been proposed that the FVIIIa heterotrimer is inactivated by dissociation of the A2 domain (reviewed in Saenko et al, 1999) and does not require proteolysis (Fay & Smudzin, 1992). The linear decline in activity (plotted on a logarithmic scale) of both FVIII-WT and FVIII-Y346C is consistent with the simple first order kinetics proposed in this model. The faster rate of decline in activity of FVIII-Y346C suggests that the Tyr346→Cys substitution reduces stability of the variant FVIIIa heterotrimer (Fig 4). It has been suggested that the interdomain region a1 minimizes FVIIIa dissociation by stabilizing the A2 domain (Fay et al, 1993). The Tyr346→ Cys substitution may disrupt this function.

We have now demonstrated that FVIII-Y346C shows a complex defect in which there is both impaired activation of the zymogen and faster decay of the activated heterotrimer. As FVIII:C at any time point in an assay is influenced both by the quantity of FVIII activated and the quantity of FVIIIa which has decayed, the behaviour of this variant in different FVIII:C assays is difficult to predict. This is particularly so because the observed defects in FVIII-Y346C are independent and are likely to be influenced in different ways by different assay conditions.

In the one-stage FVIII:C assay, the time between the first appearance of catalytic amounts of thrombin and a measurable clot may be as little as 15–20 s. Under these conditions, the assay will be extremely sensitive to the effect of defective thrombin binding or cleavage of FVIIIa. However the modest reduction of FVIIIa half-life seen here would be very unlikely to have any additional effect on the one-stage assay result. Conversely, in the two-stage assay, the period allowed for full FVIII activation is much longer and is therefore less likely to show the effect of a defect in thrombin interaction, especially with high thrombin concentrations. However, we might expect the two-stage assay result to be reduced by the enhanced dissociation of FVIIIa in our variant (Pipe et al, 2001), yet this assay was normal in our subject. A possible explanation for this is that we demonstrated enhanced dissociation in a purified system with a different kinetic environment than the plasma-based two-stage assay. In addition, the modestly enhanced decay shown for FVIII-Y346C in Fig 4 may be ‘rescued’ in the two-stage assay in plasma by complexation of the FVIIIa heterotrimer into the tenase complex (Fay et al, 2001).

The experimental conditions of routine laboratory assays have been chosen such that FVIII-WT activity is robustly and conveniently assessed in both types of assay. The same is unlikely to be true of some variant FVIII molecules. In the case of FVIII-Y346C, we suggest that the one-stage assay is most sensitive to the functional abnormality of the variant FVIII. The complex molecular defect displayed by FVIII-Y346C highlights the protean function of the surrounding interdomain acidic region a1.

Ancillary