Fibrinogen Naples I (Bβ A68T) Nonsubstrate Thrombin-Binding Capacities

Thrombosis Research 103 (2001) 63 ± 73

REGULAR ARTICLE

Fibrinogen Naples I (Bb A68T) Nonsubstrate Thrombin-Binding Capacities

David A. Meh1, Michael W. Mosesson1, Kevin R. Siebenlist2, Patricia J. Simpson-Haidaris3, Stephen O. Brennan4, James P. DiOrio5, Kevin Thompson1 and Giovanni Di Minno6 1 Blood Research Institute, The Blood Center of Southeastern Wisconsin, Milwaukee, WI, USA; 2Department of Biomedical Sciences, College of Health Sciences, Marquette University, Milwaukee, WI, USA; 3Department of Medicine Ð Vascular Medicine, University of Rochester School of Medicine and Dentistry, Rochester, NY, USA; 4Molecular Pathology Laboratory, Christchurch Hospital, Christchurch, New Zealand; 5Baxter Health Care, Round Lake, IL, USA; and 6Instituto di Medicina Interna e Geriatria, UniversitaÁ di Palermo, Palermo, Italy (Received 9 October 2000 by Editor L. Medved; revised/accepted 19 March 2001)

Abstract Fibrinogen Naples I (Bb A68T) is characterized by defective thrombin binding and fibrinopeptide cleavage at the fibrinogen substrate site in the E domain. We evaluated the fibrinogen of three homozygotic members of this kindred (II.1, II.2, II.3) who have displayed thrombophilic phenotypes and two heterozygotic subjects (I.1, I.2) who were asymptomatic. Electron microscopy of Naples I fibrin networks showed relatively wide fiber bundles, probably due to slowed fibrin assembly secondary to delayed fibrinopeptide release. We evaluated 125I-thrombin binding to the fibrin from subjects I.1, I.2, II.1, and II.2 by Scatchard analysis with emphasis on the high-affinity site in the D domain of fibrin(ogen) molecules containing a g chain variant termed g0. Homozygotic subjects II.1 and II.2 showed virtually absent low-affinity binding, consistent with the Bb A68T mutation, whereas heterozygotes I.1 and I.2 showed only moderately reduced low-affinity binding. The homozygotes also showed impaired high-affinCorresponding author: Michael W. Mosesson, Fibrinogen Research, The Blood Center of Southeastern Wisconsin, PO Box 2178, Milwaukee, WI 53201-2178, USA. Tel: +1 (414) 937 3811; Fax: +1 (414) 937 6284; E-mail: .

ity thrombin binding, whereas that of the heterozygotes was nearly the same as normal. Genomic sequencing of the g0 coding sequence (I.2, II.2), ELISA measurements of two g0 chain epitopes (L2B, g0409±412, and IF10, g0417±427) (I.2, II.1, II.2, II.3), and mass spectrometry of Naples I fibrinogen (II.2) showed no differences from normal, thus indicating that there were no abnormal structural modifications of the g0 chain residues in Naples I fibrinogen. However, thrombin reportedly utilizes both of its available exosites for binding to high- and low-affinity sites on normal fibrin, suggesting that binding is cooperative. Thus, reduced high-affinity thrombin binding to homozygotic Naples I fibrin may be related to the absence of lowaffinity binding sites. D 2001 Elsevier Science Ltd. All rights reserved. Key Words: Fibrinogen; Dysfibrinogen; Thrombin; Thrombosis

T

hrombin binding to fibrinogen at its substrate recognition sites in the central E domain of the molecule leads to cleavage and release of fibrinopeptides A and B concomitant with fibrin formation [1]. Fibrin itself has an appreciable thrombin-binding potential that was identified more than 55 years ago by Seegers et

0049-3848/01/$ ± see front matter D 2001 Elsevier Science Ltd. All rights reserved. PII S0049-3848(01)00273-0

64

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

al. [2,3] and was termed antithrombin I. In present day terms, antithrombin I is defined by two classes of nonsubstrate thrombin-binding sites, one of low affinity in the E domain and the other of high affinity in D domains of fibrin(ogen) molecules that contain a g chain variant termed g0 (g01±427L) [4,5]. The g0 chains arise through alternative processing of the primary mRNA transcript [6], and contain a unique 20 residue anionic, tyrosine-sulfated, C-terminal sequence (408V-R-P-E-H-P-A-E-T-E-YSO3-D-S-LYSO3-P-E-D-D-L427) [7,8]. They amount to  8% of the total g chain population [9] and differ from the more common gA chain (g1±411V), which terminates as 408A-G-D-V411 and contains a Cterminal platelet-binding sequence comprised by gA400 ± 411V [10]. The low-affinity thrombinbinding site in fibrin evidently is a residual feature of fibrinogen substrate binding, in which the N-terminal portions of a chains (a27±50) and b chains (b15±42) contribute to their formation [8,11 ± 13]. The high-affinity thrombin-binding site is situated between residues 414 and 427 of the g0 chains, and sulfated tyrosines at g0418 and g0422 significantly increase its thrombin-binding potential [8]. In addition to the full-length g01± 427L chain described above, a shortened version of the g0 chain (termed g55 or g01±423P) has been identified in plasma fibrinogen preparations [14±16]. g01±423P probably arises by posttranslational, proteolytic modification of the g01±427L chain and accounts for  20% of the total g0 chain population and  3% of all g chains [14]. The existence of this truncated form of g0 chain is relevant to the thrombin-binding theme of this report since Meh et al. [8] have demonstrated that the ultimate C-terminal portion of the g01±427L chain is required for effective thrombin binding. Thromboembolic disease occurs in about 20% of dysfibrinogenemic families [17,18], and this is associated with several different types of structural anomalies in the affected fibrinogen molecules [19]. Pertinent to the subject of our present report are thrombophilic dysfibrinogenemic phenotypes that manifest defective nonsubstrate thrombin binding, most notably fibrinogens New York I (des Bb9±72) [20,21] and Naples I (Bb A68T) [22±24], the latter of which is the subject of this present report.

Fibrinogen New York I lacks a major portion of the thrombin recognition site on fibrinogen and was associated with recurrent deep venous thrombosis (DVT) and fatal pulmonary embolism [21,22]. Two homozygous members of the Naples I kindred experienced occlusive arterial strokes at an early age, while a third homozygous family member suffered DVT following abdominal surgery [22]. Naples I fibrinogen is associated functionally with delayed thrombinmediated fibrinopeptide release [22 ± 24] and moderately reduced [22] or nearly absent [23] nonsubstrate thrombin binding in homozygotes (Table 1). Fibrin formed from recombinant homodimeric Bb A68T fibrinogen treated with ancrod to release only fibrinopeptide A has decreased lateral association and decreased thrombin binding, resulting in fibers that are thinner and more branched than normal fibrin clots [25]. Thrombin-binding studies of fibrinogen Naples I were reported prior to recognition that there was more than one type of nonsubstrate thrombin-binding site in fibrin. Whereas there is little doubt that the low-affinity thrombinbinding sites are adversely affected in the Naples I kindred, it seemed curious to us in light of the discovery of a distinct high-affinity nonsubstrate thrombin-binding site in the g0 chain, that overall thrombin binding in one of three reported homozygous members was virtually absent. That is to say, the g0 chain site, which is remote from the affected thrombinbinding sites in the Naples I E domain, accounts for much of the measured nonsubstrate thrombin binding. For this reason, we reevaluated the nonsubstrate thrombin-binding sites in fibrin from five members of the Naples I kindred. We also documented the effect of defective thrombin binding on Naples I fibrin clot structure.

1. Materials and Methods 1.1. Preparation of Plasma Proteins Citrated blood was collected from five family members (Table 1), and the plasma fibrinogen was purified as described by Martinez et al.

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

65

Table 1. Reported features of the Naples I kindred Family member I.1 I.2 II.1 II.2

Gender

Genotype

Thrombotic events

Thrombin time

Thrombin-binding, % normal

female male female male

heterozygous heterozygous homozygous homozygous

normal normal " "

not reported not reported 25 ± 30%a 25 ± 30%a

male

homozygous

none none stroke stroke, aortic occlusion DVT

"

< 10%b

(AZ) (FL) (VL) (PL)

II.3 (AL) a

Di Minno et al. [22]. b Koopman et al. [23].

[26]. Naples I fibrinogen was equivalent in composition to normal fraction I-2 [27], whose g0 chains amount to 7% to 8% of the total g chain population [28 ± 30]. Normal fibrinogen fraction I-2 was isolated by a glycine precipitation and subfractionation procedure [27]. For certain experiments, fraction I-2 was chromatographically separated into fibrinogen 1 (`peak 1' fibrinogen) and fibrinogen 2 (`peak 2' fibrinogen) [31]. Plasma factor XIII (2000 Loewy units/ mg) [32] was prepared from pooled human plasma [33]. 1.2. Genomic Sequencing of the g0 Chain Coding Region of Fibrinogen Naples To establish the molecular sequence of exons IX and X of the Naples I g chain gene, genomic DNA from subjects I.2, and II.2 was isolated from the buffy coat of blood. Oligonucleotides GC3, 50-CTTCATAGACTTGCAGAG-30 (sense), and GC4, 50-TTGCAAGTCCATTGTCCAATAGGAAAAATA-30 (antisense) (Bio-Synthesis, Lewisville, TX) were used to amplify exons IX and exon X from positions 9276 to 10158 of the fibrinogen g chain by PCR [34] (numbered according to Chung and Davie [6]). PCR products were purified using the Geneclean Kit (BIO 101, La Jolla, CA) and sequenced using GC8, 50-CGGTGGTATTCCATGAAG-30 (sense), a n d G C 6 , 5 0- T C C A T T G A A G G C TAAATGTCCTTA-30 (antisense) as sequencing primers using the Sequenase Version 2.0 kit, and [a35S]-dATP (United States Biochemical, Cleveland, OH). Sequence extending from 9333 through 9827 could be read using this combination of primers.

1.3. Thrombin-Binding Experiments Human a-thrombin (specific activity, 3.04 U/mg) was obtained from Enzyme Research Laboratories (South Bend, IN). D-Phe-Pro-Arg chloromethyl ketone-thrombin (PPACK-thrombin) was prepared by adding a fivefold molar excess of PPACK (Sigma, St. Louis, MO), dialyzed, and then labeled with 125I [35]. 125I-thrombin was separated from unbound iodine by affinity chromatography on fibrin 2-Sepharose CL-4B in 50 mM HEPES, 100 mM NaCl, pH 7 buffer containing 0.01% PEG 8000. Thrombin elution was achieved with the HEPES buffer containing 500 mM NaCl. Thrombin±fibrin binding experiments were performed as described [5] but the method was scaled down to accommodate the small quantities of fibrinogen that were available for study. Briefly, fibrin monomer solutions were prepared from fibrinogen in 60 mM of sodium phosphate buffer, pH 6.4, to which thrombin (1 U/ml, final) had been added. Clots were synerized and dissolved in 20 mM of acetic acid to a stock concentration of 10 mg/ml. Clots containing 75 mg fibrin in 150 ml (0.22 nmol) were formed by adding a fibrin monomer solution to a 100 mM NaCl, 50 mM HEPES, 0.01% PEG 8000, pH 7.4 buffer containing varying amounts of added 125I-thrombin (67 nM to 6.7 mM; 0.01 to 1.0 nmol), and incubated for 2 h at room temperature. Clot-bound thrombin was separated from free thrombin by syneresis, and both soluble and fibrin-bound thrombin determined by radioactivity counting. Binding data were determined from duplicate samples and graphed as Scatchard plots [4,36], and data indicating a two-component system deconvoluted as described by Klotz and Hunston [37].

66

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

1.4. ELISA Measurements Two mouse monoclonal antibodies (MoAbs) were used to determine the g0 content of plasma fibrinogen: (1) L2B is an IgG1 kappa light-chain antibody raised against a g0 peptide immunogen, g0408±416. Its epitope is located within g0409±412 [38,39]. (2) IF10 is an IgG1 kappa light-chain antibody raised against a g0 peptide immunogen, g0417 ± 427. This antibody recognizes g01 ± 427L but not g1 ± 423P (Simpson-Haidaris, unpublished data). We have localized its epitope more precisely (see below). ELISA was carried out in 96-well plates as follows: fibrinogen antigen, 100 ml at 2 mg/ml, was placed in a well and incubated overnight at 4°C. Residual liquid was removed and residual binding sites on the plates blocked by adding 200 ml 2% bovine serum albumin (BSA) in 100 mM NaCl, 20 mM sodium phosphate buffer, pH 7.3 (PBS), and incubating at room temperature for 60 min. Each well was then exchanged four times with 1% BSA in PBS containing 0.1% Tween 20. The primary MoAb, 100 ml at 1 mg/ml, was then added to each well, the plates incubated for 2 h at 37°C, and then washed four times with 1% BSA in PBS containing 0.1% Tween 20. The secondary antibody, goat anti-mouse F(ab)2-alkaline phosphatase (Zymed, San Francisco, CA), was then added (100 ml of a 1:1500 dilution for L2B and 1:3000 dilution for IF10 in PBS and 0.1% Tween 20), the mixture incubated for 1 h at 37°C, and color developed for 20 min by adding 100 ml of p-nitrophenyl phosphate (PNPP; Zymed), which had been diluted 1:100 to 1 mg/ml with 0.75 M 2-amino-2-methyl-propandiol, pH 10.3. The reaction was stopped by the addition of 50 ml 2 N NaOH, and the plates read at 405 nm in an ELISA plate reader. A standard calibration curve ranging from 100% to 0% fibrinogen 2 was prepared by mixing various proportions of fibrinogen 1 (100% gA) and fibrinogen 2 and diluting to a final fibrinogen concentration of 2 mg/ml.

For lack of a better calibration standard at the time the analyses were conducted, we used fibrinogen 2 and assumed that 100% of its g0 chains were equally reactive with IF10 and L2B. Since the g0423P chain reportedly amounts to  20% of the g0 chain population [14], our use of this material for calibration may result in overestimation of the content of g0427L and the IF10:L2B ratio. Epitope mapping of IF10 and L2B was carried out using peptide competition in the ELISA format described above. MoAbs at a concentration of 2 mg/ml were further diluted with an equal volume of a 100- or 50-mM solution of a g0 peptide, and the mixture incubated for 1 h at room temperature. The MoAb±peptide mixture was then added to a fibrinogen-coated well, and further processed as described above to develop a color reaction with PNPP. Antibody binding was expressed as the `percent inhibition,' the level of antibody response in the presence of inhibitory peptide [(1 response with peptide/ response without peptide)  100]. Marked inhibition of IF10 binding was observed with all peptides containing residue 427, whereas there was poor binding inhibition with any peptide lacking g0427 (Table 2). Thus, the epitope for IF10 lies between 417 and 427 and includes C-terminal Leu427. Good inhibition of L2B binding was observed with g0407±427, but not with g0410± 427 or g0410±418 (data not shown), thus reinforcing the earlier finding that the L2B epitope is located between g0409 and g0412 [38,39]. 1.5. Fibrin Cross-Linking Experiments Fibrinogen isolated from normal pooled plasma or from Naples I subjects' plasma was diluted to 1.0 mg/ml in 100 mM NaCl, 50 mM Tris, 10 mM CaCl2, pH 7.4 buffer, factor XIII (35 Loewy units/ ml) added, the mixture clotted with thrombin (1 U/ml) and allowed to cross-link for 15 min at 37°C. The clot was then dissolved in Laemmli

Table 2. Competitive inhibition by g0 peptides for IF10 binding to fibrinogen 2 (% inhibition) ;0 Peptide (mM)

408 ± 427SO3 (%)

407 ± 427 (%)

414 ± 427SO3 (%)

417 ± 427 (%)

414 ± 426 (%)

414 ± 425 (%)

414 ± 424 (%)

50 25

96 87

91 81

86 77

96 87

2.5 0

9 7

7 7

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

buffer containing 5% b-mercaptoethanol, and subjected to 9% SDS-PAGE [40]. 1.6. Electron Microscopic Analysis of Fibrin Fibrin for low-voltage scanning electron microscopy (SEM) was formed directly on carbon ± formvar-coated gold grids in a 100 mM NaCl, 50 mM Tris, pH 7.4 buffer at a fibrinogen concentration of 200 mg/ml and a thrombin concentration of 0.1 U/ml. Clots were incubated for 3 h in a humidity chamber and then fixed with 2.5% glutaraldehyde in 100 mM HEPES, pH 7 buffer containing 0.2% tannic acid, washed with buffer, dehydrated, critical-point-dried, sputter-coated with gold± palladium, and imaged in a JOEL 6300F SEM at 5 kV.

67

C = 12.011, H = 1.00794, N = 14.00674, O = 15.9994, P = 30.9737, and S = 32.006.

2. Results 2.1. Thrombin-Binding Sites in Naples I Fibrin We carried out studies of 125I-thrombin binding to Naples I fibrins and graphed the data as a

1.7. Mass Spectrometric Analysis Fibrinogen from subject II.2 was dissolved (5 mg/ ml, final) in 8 M urea, 0.10 M Tris/HCl, pH 8.0, and reduced in 15 mM dithiothreitol for 4 h at 37°C. Seventy-five microliters of the reduced protein was then injected onto a Phenomenex (25  0.45 cm) C-4 column equilibrated in 64% solvent B, where B was 0.05% TFA in 60% acetonitrile and A was 0.05% TFA. Individual fibrinogen chains were then eluted with a linear gradient to 85% B over 20 min at a flow rate of 0.75 ml/min, with the effluent monitored at 215 nm. Peak crests spanning a volume of 200 ml were collected and analyzed directly by electrospray ionization mass spectrometry on a VG Platform quadropole analyzer operating in positive ion mode. Ten to twenty microliters of each peak was injected into the source at a flow rate of 5 ml/min. The probe was charged at + 3500 V and the source maintained at 60°C. The mass range of 700±1400 m/z was scanned every 2 s and a cone voltage ramp of 30±60 V was applied over this range. A maximum of 120 scans was averaged in acquiring the data. Calibration was made over this same m/z range using the charge series generated by human globin (0.5 mg/ml in 0.1% formic acid, 50% acetonitrile). Data were acquired and processed using MassLynx software and transformed on a true molecular mass scale using maximum entropy (MaxEnt) software supplied with the instrument. Molecular masses are based on atomic weights of

Fig. 1. Scatchard analyses of 125I-PPACK-thrombin binding to Naples I fibrin and to normal fibrin. (A) II.1, II.2, and normal fibrin. (B) I.1 and I.2. The solid lines in (A) indicate the high- and low-affinity binding components for fraction I-2 fibrin from eight normal samples. In all cases, the high-affinity component was corrected for the contribution from low-affinity binding [37]. We drew only one line for subjects II.1 and II.2 (A) and for subjects I.1 and I.2 (B) because the individual values were very close to one another in each case. The long dashed lines indicate the high-affinity binding component and the short dashed lines indicate the low-affinity binding component for the Naples I fibrin.

68

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

Table 3. Binding parameters for thrombin-binding sites in normal and Naples I fibrins High-affinity site a

6

Low-affinity site 5

Fibrin sample

n

Ka  10 ‹ S.D.

Sites/molecule

Ka  10

Sites/molecule

I.1 (AZ) I.2 (FL) II.1 (VL) II.2 (PL) Normal

2 2 2 2 8

7.71 ‹ 0.09 8.27 ‹ 0.38 2.82 ‹ 0.52 3.37 ‹ 0.10 6.18 ‹ 1.50

0.13 0.11 0.08 0.08 0.10

0.85 0.85 0.29 0.27 1.6

2 2 2 2 1.8

a

Number of independent binding experiments.

Scatchard plot (Fig. 1; Table 3). Each fibrin displayed a high- and low-affinity thrombin-binding component, but there was an insufficient number of replicate determinations on II.3 to carry out a formal Scatchard analysis (data not shown). Homozygotic subjects II.1 and II.2 (panel A), each showed a markedly reduced Ka for the low-affinity component, a finding that is entirely consistent with previous studies on this dysfibrinogen [22,23]. This finding is very similar to the reduced low-affinity Ka that was found for des Bb1±42 fibrin prepared from normal fibrinogen [5]. The extrapolated value for the number of low-affinity binding sites approached two, but because of the low binding affinity, this value could not be determined with good precision. The high-affinity binding component for subjects II.1 and II.2 (panel A) was considerably lower than that for normal. Scatchard analysis of heterozygous Naples I fibrin samples (I.1, I.2)

(Fig. 1B) revealed in both cases a reduced lowaffinity Ka with values midway between the normal and homozygotic samples (Table 3), consistent with one of the two low-affinity thrombinbinding sites in their respective fibrins being normal. The extrapolated value for low-affinity thrombin-binding sites was consistent with two per molecule. The Ka for the high-affinity component was slightly higher than normal. Owing to lack of material, we were not able to carry out enough replicate determinations to ascertain whether the increased values found for subjects I.1 and I.2 were significantly higher than the normal control. 2.2. The Naples I Fibrinogen g0 Sequence There are several potential structural explanations relating to the g0 site for the reduced highaffinity thrombin binding in subjects II.1 and

Table 4. Quantification of g0 chain epitope binding in fibrinogen fraction I-2 from Naples 1 subjects and from normals Fibrinogen sample

na

L2B (;0409± 412) fibrinogen 2 content (% ‹ S.D.)b

IF10 (;0 ± 427) fibrinogen 2 content (% ‹ S.D.)b

Ratio IF10:L2B

I.2 (FL) II.1 (VL) II.2 (PL) II.3 (AL) Normal donor 1 Normal donor 2 Normal donor 3 Normal donor 4 Normal pool 1 Normal pool 2 All normalsc

9 5 9 6 9 9 9 9 8 6 6

17.1 ‹ 4.7 14.7 ‹ 1.6 16.5 ‹ 3.5 14.2 ‹ 4.4 11.6 ‹ 2.1 14.0 ‹ 3.0 16.2 ‹ 5.7 12.4 ‹ 4.0 14.8 ‹ 2.7 13.1 ‹ 3.5 13.6 ‹ 1.7

17.9 ‹ 3.2 13.4 ‹ 2.4 17.2 ‹ 2.1 14.0 ‹ 5.3 11.8 ‹ 2.7 11.7 ‹ 2.2 12.4 ‹ 2.9 11.4 ‹ 2.5 15.0 ‹ 2.9 14.2 ‹ 4.7 12.8 ‹ 1.5

1.05 0.91 1.04 0.98 1.02 0.84 0.77 0.92 1.01 1.08 0.94 (range, 0.77 ± 1.08)

a

Number of experiments. b Fibrinogen 2 content in fraction I-2 specimen. c Includes the mean ‹ S.D. of four single-donor and the two pooled-donor specimens.

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

II.2: (1) an altered amino acid sequence for g0 chains; (2) an increased proportion of g01±423P chains relative to g01±427L; (3) reduced levels of g0 chains; (4) a reduced level of sulfation of g0 chain tyrosine residues. We addressed each of these possibilities. Genomic sequencing of the coding region between nucleotides 9333 and 9827 covering exon IX (amino acid g0351 through g0427) revealed a normal coding sequence for I.2 and II.2, thus eliminating the possibility that there was an abnormal primary sequence in the g0 region. ELISA measurements were carried out using either L2B, whose epitope is located at g0409± 412, or IF10, whose epitope (g0417±427) overlaps the C-terminus. Several normal fibrinogen preparations were analyzed to establish the range of values for the L2B and IF10 epitopes. Determinations were also made for subjects I.2, II.1, II.2, and II.3 (Table 4). The L2B epitope, which reflects the total g0 chain population, was slightly to moderately higher in Naples I family members than the mean level found for the normal fibrinogen, indicating that the g0 chain content in Naples I fibrinogen samples were not reduced relative to that in normal fibrinogen. Furthermore, the ratio of the IF10 to L2B epitope in Naples I fibrinogen samples were well within the range of values found for normal. This finding suggested that the reduction in high-affinity thrombin binding for subjects II.1 and II.2 were not attributable to decreases in the absolute level of g0 chains or increases in the levels of g01 ± 423P at the expense of g01±427L. Reinforcement for the conclusion that the relative g0 chain content was normal was obtained from SDS-PAGE analyses of crosslinked fibrins (Fig. 2). Visual inspection indicated that the distribution of g0 ±gA dimer bands relative to gA±gA bands, and containing more than 93% of available g0 chains1, did not differ from one another or from normal fibrin, thus suggesting that the level of g0 chains in homozygous Naples I fibrinogens were in the same

1

About 7% of the cross-linked g0 chains would occur as g ± g dimers [29], and this dimer band would be too faint to be visualized. 0

0

69

Fig. 2. SDS-PAGE of cross-linked fibrin specimens II.1, II.2, II.3, I.2, I.1, and normal fibrin.

range as normal. These findings are consistent with the normal values for g0 chain content estimated from ELISA measurements. We had only enough material to determine the mass of the fibrinogen g0 chain population in subject II.2. The mass of the dominant monosialo form of the Naples I Bb chain was 54,220 Da compared to a mean of 54,200 Da for three normal controls (S.D. ‹ 3). The 20-Da increase was consistent with the Ala to Thr mutation. The mass of the monosialo g chain was 48,384 Da compared to 48,383 Da for three normal controls (S.D. ‹ 5). The normal g0 chain yielded a signal at 50,561 Da corresponding to a doubly sulfated g0 chain. The g0 chain of subject II.2 yielded a signal at 50,550 Da and a much smaller signal very near to the baseline at 50,532 Da. The first signal was within 8 Da of the doubly sulfated g0 sequence, and the second signal was probably too small to be significant. In any case, there was no suggestion of the presence of a g0 chain lacking one (50,478 Da) or two (50,398 Da) sulfate groups. 2.3. Fibrin Structure SEM analysis of Naples I fibrin (Fig. 3) indicated that fibers from homozygous subjects II.1, II.2, and II.3 tended to be wider than normal and were superimposed upon network fibers of more normal width. Heterozygotic fibrin specimens also appeared to contain somewhat wider fibers

70

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

Fig. 3. SEM of fibrin from members of the Naples I kindred. Bar: 1 mm.

than the normal, and this was less prominent than in the homozygotes. This result is attributable to slowed fibrin assembly secondary to defective thrombin binding at the fibrinogen substrate site and consequent delayed fibrinopeptide release.

3. Discussion We attempted to identify the basis for reduced high-affinity binding in subjects II.1 and II.2. Genomic sequencing of exon IX of their g chain genes revealed a normal coding sequence in both cases, thus eliminating the possibility that there was an altered primary sequence in the g0 region. ELISA measurements indicated

that there were no differences from normal in the g0 chain content, as assessed by measurements with two g0-directed MoAbs. This observation was supported by the results of SDSPAGE analyses of cross-linked fibrin, which showed no differences from normal in the distribution of g0 ±gA and gA±gA dimer bands. Furthermore, we found no differences from normal in the IF10 to L2B ratios of any Naples I family member, suggesting that reduced high-affinity binding in II.1 and II.2 was not due to increases in the level of non thrombinbinding g01±423P chains at the expense of g01± 427L. We also evaluated the degree of tyrosine sulfation by mass spectrometry of the fibrinogen from subject II.2, for whom we had a barely sufficient amount of fibrinogen to attempt such an analysis. The data indicated that there was a doubly sulfated g0 chain in subject II.2 and there was no suggestion of an undersulfated g0 chain sequence. Thus, on the basis of the available experimental evidence, we can provide no evidence for a defective or deficient g0 chain in subjects II.1 and II.2. However, thrombin has recently been reported to bind to high-affinity sites through its exosite 2 [41], whereas it binds to low-affinity sites through exosite 1 [5], suggesting that thrombin binding to fibrin may be cooperative. On this basis, the profoundly defective low-affinity nonsubstrate thrombin binding in homozygotic Naples I subjects appears to contribute to reduced thrombin binding at the high-affinity site. The availability of half the normal lowaffinity sites in fibrin from heterozygotic Naples I subjects (I.1, I.2) evidently provides a sufficient number of low-affinity sites that are capable of promoting effective cooperative high-affinity thrombin binding to the minor population of g0 chains [9]. As an aspect of these investigations, we characterized the organization of Naples I fibrin networks by SEM. Homozygous Naples I fibrin matrices contained wider fibers than the normal specimen superimposed upon a fiber network having the same widths as found in normal fibrin. Fibrins from heterozygous family members were intermediate in width compared to the normal specimens. We interpret these findings in homozygotes to be a reflection of defec-

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

tive thrombin substrate binding, slowed thrombin-mediated release of fibrinopeptide A, and slowed polymerization resulting from the reduced rate of fibrinopeptide cleavage. Previous studies have shown that lowering thrombin levels in normal fibrinogen results in slowed rates of assembly and network fibers of increased width superimposed upon a background network of thin, highly branched fibrils and fibers [42]. Another study of this mutation in recombinant fibrinogen found thinner fibers formed as compared to normal fibrin, but high thrombin levels or ancrod were used to release the fibrinopeptides [25]. In conclusion, these data confirm the profoundly defective thrombin binding in homozygotes that characterizes Naples I fibrinogen (Bb A68T), a fact that hardly needs reemphasis given the thorough investigations that have been reported on both native Naples I fibrinogen and recombinant homodimeric Bb A68T fibrinogen [22±25]. Our studies, however, have more fully characterized the global thrombinbinding potential of fibrin from four members of the Naples I kindred, two of whom were homozygous for the Bb A68T mutation. What we found, surprisingly, was a significantly reduced high-affinity thrombin-binding component in two homozygous members of the family (II.1, II.2). This was manifested primarily as a reduction in the Ka. In contrast, there was no evidence in the heterozygotes (I.1, I.2) to indicate that high-affinity binding in their fibrin was lower than normal. Thus, these results provide new insights into `antithrombin I' function by suggesting that a mutation at one site in the fibrinogen molecule (i.e., Bb A68T) that results in complete absence of low-affinity thrombin-binding sites in homozygotes, contributes to reduced thrombin affinity at another thrombin-binding site, namely the g0 sequence. The role of high-affinity thrombin binding in mitigating thrombin procoagulant effects in vivo remains to be fully explored, but we believe that it serves to down-regulate thrombin potential. In this connection, it is highly relevant to point out that Naples I subjects with an absence of low-affinity thrombin-binding sites (i.e., homozygotes II.1, II.2, II.3) displayed a marked thrombophilic profile, whereas heter-

71

ozygotes who possessed half the normal number of low-affinity sites were asymptomatic.

This investigation was supported by NHLBI Grant No. HL59507.

References 1. Mosesson MW, Siebenlist KR, Meh DA. The structure and biological features of fibrinogen and fibrin. Ann NY Acad Sci 2001;936: 11±30. 2. Seegers WH, Nieft M, Loomis EC. Note on the adsorption of thrombin on fibrin. Science 1945;101:520±1. 3. Seegers WH, Johnson JF, Fell C. An antithrombin reaction related to prothrombin activation. Am J Physiol 1954;176:97±103. 4. Liu CY, Nossel HY, Kaplan KL. The binding of thrombin by fibrin. J Biol Chem 1979;254: 10421±5. 5. Meh DA, Siebenlist KR, Mosesson MW. Identification and characterization of the thrombin binding sites on fibrin. J Biol Chem 1996;271: 23121±5. 6. Chung DW, Davie EW. g and g0 chains of human fibrinogen are produced by alternative mRNA processing. Biochemistry 1984;23: 4232±6. 7. Wolfenstein-Todel C, Mosesson MW. Carboxy-terminal amino acid sequence of a human fibrinogen g chain variant (g0). Biochemistry 1981;20:6146±9. 8. Meh DA, Siebenlist KR, Brennan SO, Holyst T, Mosesson MW. The amino acid sequences in fibrin responsible for high affinity thrombin binding. Thromb Haemostasis 2001;85:470±4. 9. Mosesson MW, Finlayson JS, Umfleet RA. Human fibrinogen heterogeneities: III. Identification of g chain variants. J Biol Chem 1972;247:5223±7. 10. Kloczewiak M, Timmons S, Lukas TJ, Hawiger J. Platelet receptor recognition site on human fibrinogen. Synthesis and structure ± function relationship of peptides corresponding to the C-terminal segment of the g chain. Biochemistry 1984;23:1767±74.

72

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

11. Vali Z, Scheraga HA. Localization of the binding site on fibrin for the secondary binding site of thrombin. Biochemistry 1988; 27:1956±63. 12. Siebenlist KR, DiOrio JP, Budzynski AZ, Mosesson MW. The polymerization and thrombin-binding properties of des-(B beta 1±42)fibrin. J Biol Chem 1990;265:18650±5. 13. Binnie CG, Lord ST. A synthetic analog of fibrinogen a27±50 is an inhibitor of thrombin. Thromb Haemostasis 1991;65:165±8. 14. Francis CW, Kraus DH, Marder VJ. Structural and chromatographic heterogeneity of normal plasma fibrinogen associated with the presence of three gamma-chain types with distinct molecular weights. Biochim Biophys Acta 1983;744:155±64. 15. Francis CW, Muller E, Henschen A, Simpson PJ, Marder VJ. Carboxy-terminal amino acid sequences of two large variant forms of the human plasma fibrinogen g chain. Proc Natl Acad Sci USA 1988;85:3358±62. 16. Francis CW, Keele EM, Marder VJ. Purification of three gamma-chains with different molecular weights from normal human plasma fibrinogen. Biochim Biophys Acta 1984;797:328±35. 17. Ebert RF. Index of variant human fibrinogens. Boca Raton (FL): CRC Press, 1994. 18. Haverkate F, Samama M. Familial dysfibrinogenemia and thrombophilia. Report on a study of the SSC Subcommittee on Fibrinogen. Thromb Haemostasis 1995;73:151±61. 19. Mosesson MW. Dysfibrinogenemia and thrombosis. Semin Thromb Hemostasis 1999;25:311±9. 20. Liu CY, Koehn JA, Morgan FJ. Characterization of fibrinogen New York I. J Biol Chem 1985;260:4390±6. 21. Liu CY, Wallen P, Handley DA. Fibrinogen New York I. In: Lane DA, Henschen A, Jasani MK, editors. Fibrinogen, fibrin formation and fibrinolysis. Berlin: Walter de Gruyter, 1986. pp. 79±90. 22. Di Minno G, Martinez J, Cirillo F, Cerbone AM, Silver MJ, Colucci M, Margaglione M, Tauro R, Semeraro N, Quattrone A, Mancini M. A role for platelets and thrombin in

23.

24.

25.

26.

27.

28. 29.

30.

31.

32. 33.

the juvenile stroke of two siblings with defective thrombin-absorbing capacity of fibrin(ogen). Arterioscler Thromb 1991;11:785±96. Koopman J, Haverkate F, Lord ST, Grimbergen J, Mannucci PM. Molecular basis of fibrinogen Naples associated with defective thrombin binding and thrombophilia. J Clin Invest 1992;90:238±44. Lord ST, Strickland E, Jayjock E. Strategy for recombinant multichain protein synthesis: fibrinogen Bb-chain variants as thrombin substrates. Biochemistry 1996;35:2342±8. Mullin JL, Gorkun OV, Lord ST. Decreased lateral aggregation of a variant recombinant fibrinogen provides insight into the polymerization mechanism. Biochemistry 2000;39:9843±9. Martinez J, Holburn RR, Shapiro SS, Erslev AJ. Fibrinogen Philadelphia. A hereditary hypodysfibrinogenemia characterized by fibrinogen hypercatabolism. J Clin Invest 1974;53:600±11. Mosesson MW, Sherry S. The preparation and properties of human fibrinogen of relatively high solubility. Biochemistry 1966;5:2829±35. Mosesson MW. Fibrinogen heterogeneity. Ann NY Acad Sci 1983;408:97±113. Wolfenstein-Todel C, Mosesson MW. Human plasma fibrinogen heterogeneity: evidence for an extended carboxyl-terminal sequence in a normal gamma chain variant (gamma0). Proc Natl Acad Sci USA 1980;77: 5069±73. Mosesson MW, Finlayson JS. Biochemical and chromatographic studies of certain activities associated with human fibrinogen preparations. J Clin Invest 1963;42:747±55. Siebenlist KR, Meh DA, Mosesson MW. Plasma factor XIII binds specifically to fibrinogen molecules containing g0 chains. Biochemistry 1996;35:10448±53. Loewy AG, Dunathan K, Kriel R, Wolfinger I, Fibrinase I. Purification of substrate and enzyme. J Biol Chem 1961;236:2625±33. Lorand L, Gotoh T. Fibrinoligase. The fibrin stabilizing factor. Methods Enzymol 1970;19: 770±82.

D.A. Meh et al./Thrombosis Research 103 (2001) 63±73

34. Saiki RK, Gelfand DH, Stoffel S, Scharf SJ, Higuchi R, Horn GT, Mullis KB, Erlich HA. Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 1988;239:487±91. 35. Martin BE, Wasiewski WW, Fenton JW, Detwiler TC. Equilibrium binding of thrombin to platelets. Biochemistry 1976;15:4886±93. 36. Scatchard G. The attractions of proteins for small molecules and ions. Ann NY Acad Sci 1949;51:660±6. 37. Klotz IM, Hunston DL. Properties of graphical representations of multiple classes of binding sites. Biochemistry 1971;10: 3065±9. 38. Haidaris PJ, Peerschke EIB, Marder VJ, Francis CW. The C-terminal sequences of the g57.5 chain of human fibrinogen constitute a plasmin sensitive epitope that is ex-

39.

40. 41. 42.

73

posed in crosslinked fibrin. Blood 1989;74: 2437±44. Haidaris PJ, Francis CW, Sporn LA, Arvan DS, Collichio FA, Marder VJ. Megakaryocyte and hepatocyte origins of human fibrinogen biosynthesis exhibit hepatocytespecific expression of gamma chain-variant polypeptides. Blood 1989;74:743±50. Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970;227:680±5. Farrell DH, Lovely RS, Moaddel M, Stafford AR, Weitz JI. Fibrinogen g0 chain binds thrombin exosite II. Blood 2000;96:448a±9a. Mosesson MW, DiOrio JP, Muller MF, Shainoff JR, Siebenlist KR, Amrani DL, Homandberg GA, Soria J, Soria C, Samama M. Studies on the ultrastructure of fibrin lacking fibrinopeptide B (b-fibrin). Blood 1987;69:1073±81.