Báo cáo khoa học: Probing plasma clearance of the thrombin–antithrombin complex with a monoclonal antibody against the putative serpin–enzyme complex receptor-binding site docx

This region overlaps, by two residues, the putative binding site of antithrombin for the serpin–enzyme complex receptor.. Studies in rats and with HepG2 cells in culture indicated that t

Probing plasma clearance of the thrombin–antithrombin complex with a monoclonal antibody against the putative serpin–enzyme complex receptor-binding site

George L Long1,*, Margareta Kjellberg2, Bruno O Villoutreix3and Johan Stenflo2

1

Department of Biochemistry, University of Vermont, Burlington, VT, USA;2Department of Clinical Chemistry, Lund University, University Hospital Malmo¨, Malmo¨, Sweden;3INSERMU428, University of Paris V, France

A high-affinity monoclonal antibody (M27), raised against

the human thrombin–antithrombin complex, has been

identified and characterized The epitope recognized by M27

was located to the linear sequence FIREVP (residues 411–

416), located in the C-terminal cleavage peptide of

anti-thrombin This region overlaps, by two residues, the putative

binding site of antithrombin for the serpin–enzyme complex

receptor Studies in rats and with HepG2 cells in culture

indicated that the Fab fragment of M27 does not block

binding and uptake of the thrombin–antithrombin complex,

suggesting that this region does not play a major role in the recognition and clearance of the thrombin–antithrombin complex M27 blocked the ability of antithrombin to inhibit thrombin as well as antithrombin cleavage, both in the presence and absence of heparin

Keywords: antithrombin; thrombin; thrombin–antithrom-bin complex; monoclonal antibody; serpin–enzyme complex receptor

Antithrombin (AT), a member of the serine protease

inhibitor family (serpin), is a 58-kDa molecular mass

glycoprotein that circulates in human plasma at a

concen-tration of 5 lM [1–5] AT modulates blood coagulation

by inhibiting thrombin, active factor X (factor Xa) and

active factor IX (factor IXa), and thereby prevents

inappropriate clot formation and thrombosis The rate of

AT-mediated inhibition of thrombin and factor Xa is

increased several thousand-fold by binding of the sulfated

polysaccharide heparin or heparin-like molecules

Individ-uals with AT deficiency are at a significantly increased risk

of venous thrombosis [6,7]

AT and other serpins inhibit their serine protease

cognates by the formation of a long-lived, covalent acyl

intermediate upon specific protease cleavage [1–5] In AT,

cleavage is at Arg393 in a so-called reactive center loop

(RCL) with the formation of a C-terminal, 39 amino-acid

residues long, disulfide-bonded (Cys247 to Cys430) peptide

[8] Prompt insertion of residues P1–P17of the RCL, with

attached protease, as an additional strand into b-sheet A of the inhibitor, causes a dramatic conformational change in the serpin [9] Recent X-ray crystallographic diffraction analysis of the trypsin–antitrypsin covalent complex has shown that insertion of the RCL also leads to a critical distortion of the structure of the protease [10] As a result, the canonical active site Ser in position 195 is reoriented at a distance of more than 6 A˚ from His57, which is too far to form the critical hydrogen bond of the catalytic triad – a bond that is a prerequisite for cleavage of the acyl intermediate that links the protease to the serpin Moreover, the distortion of the complexed thrombin molecule renders

it susceptible to proteolytic degradation A further conse-quence of the complexation-induced conformational change

in the serpin is exposure of structure(s) that are recognized

by serpin–enzyme complex (SEC) receptors on the surface

of hepatocytes Receptor binding followed by endocytosis results in rapid clearance of the protease–serpin complexes from the circulatory system [11]

In addition to native and complexed AT, two forms of

AT have been characterized: a cleaved uncomplexed form with the RCL inserted into b-sheet A, and a so-called latent uncleaved, inserted form The cleaved, loop-inserted inhibitor is formed if loop insertion, with the acyl-linked protease, is not sufficiently rapid to compete with deacylation, which leads to release of the active protease Human elastase cleaves AT at Ile390 without complex formation [12] The elastase-cleaved form has properties that are indistinguishable from those of the thrombin-cleaved inhibitor The latent form is a conform-ational isomer of the native form in which the RCL has been inserted into b-sheet A without prior complex formation/cleavage Neither the cleaved uncomplexed nor the uncleaved latent form exhibit inhibitory activity

Correspondence to J Stenflo, Department of Clinical Chemistry,

Lund University, University Hospital, Malmo¨, S-205 02 Malmo¨,

Sweden Fax: + 46 40 929023, Tel.: + 46 40 331421,

E-mail: johan.stenflo@klkemi.mas.lu.se

Abbreviations: AT, antithrombin; LRP, low-density lipoprotein

receptor-related protein; PVDF, poly(vinylidene difluoride); RCL,

reactive center loop; SEC, serpin–enzyme complex; SECR, serpin–

enzyme complex receptor; T–AT, covalent thrombin-antithrombin

complex.

*This work was carried out during the sabbatical of G L Long

to the Department of Clinical Chemistry, Lund University,

University Hospital Malmo¨, S-20502 Malmo¨, Sweden.

(Received 3 July 2003, revised 30 July 2003, accepted 15 August 2003)

[9,13,14] Latent AT forms spontaneously under mild

conditions [13,14], including storage of plasma at 37C

[15] A naturally occurring genetic variant, AT Rouen-VI

(Asn187fi Asp), readily forms the latent form and is

associated with fever-induced thromboembolic disease

[16] In commercial concentrates of AT, up to 40% exists

as the latent form [17] Recently, it was reported that AT

has potent antiangiogenic activity and inhibits tumor

growth [18,19] Native AT has little effect, cleaved AT has

an intermediate effect, and latent AT is the most potent

antiangiogenic agent reported to date [19]

In this communication we report the characterization of

a murine mAb, M27, against human AT that binds to the

native, complexed, cleaved, and latent forms of AT M27

blocks the thrombin-inhibiting capacity of AT and

protects it from cleavage by thrombin M27 binds to a

linear epitope (residues 411–416; FIREVP) that partly

overlaps a region (residues 408–412; FLVFI) implicated in

the recognition of certain serpin–protease complexes by

the SEC receptor (SECR), first identified on the surface of

human hepatoma HepG2 cells [20] M27 binds the

thrombin–AT (T–AT) complex with a picomolar

dissoci-ation constant The rate of clearance of the ternary

T–AT–Fab complex in rats was only slightly slower than

the rate of clearance of T–AT Studies with cultured

HepG2 cells indicated that M27 does not block the

binding of T–AT in an epitope-dependent manner These

findings are consistent with recent results that cast doubt

on the notion that the sequence FLVFI in AT, and

homologous regions in certain other serpins, is crucial for

receptor-mediated elimination of protease–serpin

com-plexes from blood plasma [21,22]

Materials and methods

Preparation of proteins

Native AT was purified from human, bovine, rabbit, and

mouse plasma, as previously described [23], and stored at

)20 C Cleaved human AT was produced essentially as

described previously [24] Native AT was incubated with

porcine elastase (Sigma-Aldrich, Stockholm, Sweden), at a

100 : 1 molar ratio, in 50 mMTris/HCl, 0.15MNaCl and

0.1% (w/v) PEG 8000 for 4 h at 37C After addition of

phenylmethanesulfonyl fluoride to a final concentration of

1 mM, the solution was dialyzed against 50 mMTris/HCl,

0.1M NaCl (pH 7.5) and purified by heparin–Sepharose

chromatography Elution with a NaCl gradient (0.1–1.0M)

gave one peak at 0.3M NaCl Sequence analyses revealed

cleavage at positions 389, 390 and 393 The material was

aliquoted and frozen at)70 C

Latent human AT was prepared as described

previ-ously [13] Native AT was incubated in 10 mM Tris/HCl,

0.25M sodium citrate, pH 7.4, for 70 h at 60C and

purified on a heparin–Sepharose column, as described

above A portion of the AT did not bind to the heparin–

Sepharose, and a second major peak, which eluted at

0.3M NaCl, showed a sequence corresponding to native

AT, but migration by SDS/PAGE corresponding to that

of latent AT The material eluting at 0.3M NaCl was

concentrated, stored at )70 C, and used in further

experiments

Human prothrombin was purified by slight modification

of a standard procedure, including precipitation with barium chloride and ammonium sulfate, followed by DEAE Sephacel chromatography [25] Prothrombin was activated with venom from Oxyuranus scutellatus (ICN Biomedicals Inc., Irvine, CA) and purified employing Q-Sepharose followed by SP–Sepharose column chromato-graphy [26]

Production of mAb M27 Murine mAbs were produced as described previously [27] The mice were immunized with human T–AT complexes These complexes were also used to test the clones in an ELISA Conditioned media were collected from cloned mouse myeloma cells cultured in a TECNOMOUSE (Integra Biosciences, Wallisellen, Switzerland) hollow fiber chamber, and stored at)20 C until used Thawed media were centrifuged at low speed to remove cellular debris and the supernatant was diluted with an equal volume of column equilibration buffer (1M glycine, 150 mM NaCl,

pH 8.0) and purified by Protein A–Sepharose chromato-graphy [28] mAb M27 was stored in aliquots at)20 C Its concentration was estimated from the absorbance at

280 nm, assuming an absorbance of 1.34 for a 1 mgÆmL)1 solution [28] M27 was determined to be of the IgG2b isotype by a standard procedure using a commercial kit (Miles Laboratories, Elkhart, IN)

Production of the M27 Fab fragment The general procedure for generating Fab was that described by Parham [29] Purified M27 was dialyzed against 100 mM sodium acetate, pH 5.5, andD,L-cysteine and EDTA were added to final concentrations of 45 and 0.9 mM, respectively After 5 min of preincubation at 37C, digestion was performed for 30 min with 0.5% (w/w) fresh papain (Sigma) After incubation with iodoacetamide (final concentration 70 mM) for 30 min at room temperature, the digest was dialyzed at 4C against 1M glycine, 150 mM NaCl, pH 8.0 The digest was then purified by protein A– Sepharose chromatography to remove traces of undigested IgG and Fc fragment The material in the flow-through peak (unbound Fab fragment) was dialyzed against 20 mM Tris/HCl, pH 8.5, and subjected to chromatography on a Q-Sepharose Fast Flow column that was eluted with a linear NaCl gradient (0–250 mM) The Fab fragment eluted

at 70 mMNaCl It was homogenous, as judged by SDS/ PAGE, and was able to bind Eu3+chelate-labeled native antithrombin in a DELPHIA assay [30] The Fab fragment was dialyzed against NaCl/Tris buffer, pH 7.5, aliquoted and stored at)20 C About 20% of the material eluted at

 160 mM NaCl; it possessed no ability to bind AT and presumAbly consisted of Fab fragments with an aberrant j chain [31]

SDS/PAGE and Western blotting SDS/PAGE and Western blotting were performed using standard methods For Western blotting, poly(vinylidene difluoride) (PVDF; Immobilon P, 0.45-lm pore size) membranes (Millipore, Bedford, MA) were used Proteins

were stained with GelCode Blue Stain Reagent (Pierce,

Rockford, IL) according to the supplier’s instructions

Peptide synthesis and epitope mapping

Peptides were synthesized on a Milligen 9050 Plus

instru-ment (Perkin-Elmer, Stockholm, Sweden), using Fmoc

chemistry, and purified by reverse-phase HPLC followed by

lyophilization and storage at)20 C until required for use

All peptides were readily dissolved in deionized water, to a

nominal concentration of 0.5 mM, and stored frozen The

exact concentrations were determined by amino acid

analysis after acid hydrolysis

Binding of the synthetic peptides to M27 was studied

by an ELISA-based method The peptides and native AT,

1.5 nmol and 15 pmol, respectively, in 100 lL of coating

buffer (100 mM sodium bicarbonate, pH 9.6), were

deliv-ered to wells of high-binding, polystyrene microtitre plates

(Costar, Corning, NY) After incubation for 15 h at 4C,

the wells were rinsed several times with 10 mM sodium

phosphate buffer containing 0.5MNaCl and 0.1% Tween

20, pH 8.0 (NaCl/Pi/Tween), followed by blocking for

15 min with 1% (w/v) BSA (Sigma Fraction V) in NaCl/

Pi/Tween Wells were rinsed again, and different amounts

of M27, diluted in NaCl/Pi/Tween containing 0.1% BSA,

were added to the wells followed by incubation on a

platform shaker for 1 h at room temperature Wells were

rinsed again and then incubated, as described above, with

a horseradish peroxidase-conjugated rabbit anti-mouse

IgG (DAKOPATTS AB, Alvsjo, Sweden) diluted 1 : 1000

in NaCl/Pi After washing, the chromogenic enzyme

substrate 2,2¢-azinobis(3-ethylbenzo-6-thiazolinesulfonic

acid) (ABTS) was added and the absorbance at 405 nm

was recorded as a function of time Binding of M27 to the

peptides (all containing a single cysteine residue, see

Fig 3A) was also determined after coating the peptides,

dissolved in NaCl/Pi, pH 7.4, to a maleimide-activated

microtitre plate (Pierce) for 15 h at room temperature

Following coating and subsequent rinsing with NaCl/Pi,

pH 7.4, wells were blocked by incubation for 90 min with

150 lL of D,L-cysteine in 10 lgÆmL)1 NaCl/Pi, pH 7.4

Wells were then rinsed with NaCl/Pi/Tween, followed by

antibody binding and enzymatic color development, as

described above

Effect of AT and M27 on thrombin activity

A two-stage assay was used to determine the effect of M27

on the inhibition of thrombin by AT The first stage consists

of a-thrombin incubation with native AT at 37C for

different lengths of time in the presence or absence of M27

Freshly diluted thrombin (0.5 lg in buffer comprising

50 lL of 50 mM Tris/HCl, 150 mM NaCl, 0.1% BSA,

pH 7.5; Tris/HCl/NaCl/BSA) was delivered into wells of a

low-affinity microtitre plate (Bibby Sterilin, Ltd, Staffs.,

UK) and allowed to equilibrate for 5 min at 37C Native

AT (0.8 lg in 10 lL of Tris/HCl/NaCl/BSA), M27 (4 lg in

10 lL of Tris/HCl/NaCl/BSA) or AT preincubated with

M27 (same concentrations and volume) were added to the

thrombin solution, mixed and incubated at 37C At

different time-points, 10-lL aliquots were removed for

SDS/PAGE or mixed into 90 lL of Tris/HCl/NaCl/BSA

(at room temperature) Duplicate aliquots (10 lL) of the latter were immediately transferred into clean wells con-taining 190 lL of freshly prepared thrombin substrate (400 lM S-2238; Chromogenix, Gothenburg, Sweden) in

50 mM Tris/HCl, 0.1% BSA, pH 8.4 The samples were briefly mixed, and the increase in absorbance at 405 nm was recorded as a function of time

A similar procedure was used to measure the effect of heparin on the above system Native AT (90 lgÆmL)1), M27 (435 lgÆmL)1), or AT plus M27 (same concentra-tions) were preincubated (for 1 h at 37C) in Tris/HCl/ NaCl/BSA containing 100 UÆmL)1heparin (average MW,

15 kDa; Lo¨vens Kemiske Fabrik, Ballerup, Denmark) Aliquots (10 lL) were then added to an equal volume

of Tris/HCl/NaCl/BSA containing 20 lgÆmL)1 thrombin, followed by brief mixing and incubation at 25C for

5 min The interaction with thrombin was stopped by the addition of 180 lL of quench solution: 110 lg of protamine sulfate (Lo¨vens Kemiske Fabrik) per ml of Tris/HCl/NaCl/BSA Duplicate 10-lL quenched aliquots were then used in the measurement of thrombin amido-lytic activity, as described above The experiments, with and without heparin, were performed on three separate occasions, with essentially identical results They were also performed once with a molar equivalent (anti-gen-binding sites) of purified M27 Fab fragment, and gave results identical to those obtained with the intact mAb

Surface plasmon resonance spectroscopy Binding of M27 IgG and the Fab fragment to different forms of AT was studied using a BIACORE 2000 biosensor (Biacore AB, Uppsala, Sweden) Purified M27

or Fab fragments were diluted into 10 mM Hepes,

150 mM NaCl, 3 mM EDTA, 0.005% Polysorbate 20,

pH 7.4 (Hepes/NaCl/EDTA/Polysorbate 20) They were immobilized with NH2-coupling to a CM5 sensor chip (Biacore) to levels of 1700 and 740 response units (RU) for the intact mAb and the corresponding Fab fragment, respectively Analytes were diluted into Hepes/NaCl/ EDTA/Polysorbate 20 and used to measure binding to the immobilized IgG and Fab using a programmed protocol with 20 s preinjection delay, 180 s association time and 600 s dissociation time The experiments were performed at room temperature and the proteins were pumped at 30 lLÆmin)1 The chip was regenerated with two pulses of 5 lL 0.1M glycine/HCl, 0.5M NaCl,

pH 2.75, at a flow rate of 5 lLÆmin)1 The analyte concentrations ranged from 0.1 to 100 nM (based on amino acid analysis) Runs were performed three times and with five different concentrations for each analyte Data were analyzed using the BIAEVALUATION 3.0 soft-ware package (Biacore), assuming noninteracting anti-body-binding sites and a 1 : 1 stoichiometry of binding

In vivo clearance of the T–AT complex in rats Thrombin, native AT, and elastase-cleaved AT were labeled with 125I (Amersham Pharmacia AB, Uppsala, Sweden) using the chloramine T method, according to the supplier’s instructions 125I-labeled thrombin was covalently

complexed with unlabeled AT at a 1 : 2 molar ratio for 1 h

at room temperature, then applied to a heparin–Sepharose

column and eluted The concentrations of two fractions

containing the T–AT complex were estimated by comparing

the absorbance at 280 nm with a nonradioactive T–AT

complex, for which the concentration had been determined

by amino acid analysis The proteins were stored in aliquots

at)70 C

Sprague-Dawley rats (350 g) were anesthesized

(2 mLÆkg)1) with a 1 : 1 : 2 (v/v/v) mixture of Hypnorm

(JANSSEN-CILAG Ltd, High Wycombe, UK),

Dormi-cum (Pharma hameln GmbH, Hamelm, Germany) and

deionized water 125I-labeled T–AT (73 pmol in 500 lL),

with and without 2.9 nmol M27 Fab, were injected in a tail

vein, and blood samples (200 lL) were drawn from a

jugular vein into tubes containing 10 lL of 0.5MEDTA

after 1, 3, 5, 10, 15 and 20 min Radioactivity was measured

in 50 lL aliquots of the plasma Blood samples drawn after

1 min were considered to represent the amount of injected

radioactivity after equilibration in the circulation 125

I-labeled thrombin, native AT, and cleaved AT alone were

also injected into control animals

Binding of125I-labeled T–AT complex to HepG2 cells

The binding studies were performed, as previously

described, with minor modifications [21] HepG2 cells

were cultured at 37C, 5% CO2 in 75-cm2 flasks (Nunc,

Roskilde, Denmark) containing Dulbecco’s modified

Eagle’s medium (DMEM) (Invitrogen Corp.)

supplemen-ted with 10% (v/v) fetal bovine serum (HyClone),

2 mM L-glutamine (Invitrogen Corp.), and penicillin/

streptomycin (100 unitsÆmL)1/100 lgÆmL)1) (Invitrogen

Corp.) Cells were transferred to 24-well plates, at a

concentration of 2–3· 105 cells per well, and grown for

2 days before the experiments The cells were washed

twice with DMEM containing 0.2% BSA, 0.5 lM

PPACK and 10 mM Hepes, pH 7.4 (binding buffer)

Radiolabeled T–AT complex (20 nM) in binding buffer,

with or without unlabelled T–AT (2 lM) or M27

(0.2 lM) or M38 (0.2 lM), or cleaved AT (1 lM), were

each added to four wells After incubation at 4C for

2 h, the wells were washed three times with 10 mM

Hepes, pH 7.4, containing 0.15M NaCl, 1 mM CaCl2,

2 mMMgCl2 and 0.2% BSA Cells were lysed in 0.4 mL

of 2M NaOH overnight at room temperature and the

radioactivity connected to the cells was measured The

results are expressed as the mean of triplicate samples

Molecular modeling

X-ray structures of human native AT [32], latent AT [9],

heparin–AT complex [33], cleaved bovine AT [34], and

trypsin–antichymotrypsin covalent complex [10], were

analyzed using theACCELRYSmolecular modeling package

(San Diego, CA, USA), running on a Silicon Graphics

workstation 02 or Fuel Solvent accessibility was computed

using the method of Lee & Richards [35] The packing

density was calculated with the method of Kurochking &

Privalov [36], with theMOLEmolecular graphics software

package kindly provided by R Tarr (Applied

Thermo-dynamics, Inc., Hunt Valley, MD, USA)

Results

Identification of the M27 epitope Western blotting was used in the initial characterization of the epitope of mAb M27 (Fig 1) Bands corresponding to all forms of nonreduced AT were rapidly visible However, when the inhibitors were analyzed after reduction of the disulfide bonds, only the native and latent forms of AT (i.e not cleaved at Arg393) were observed The 39-residue C-terminal peptide (residues 394–432), formed upon com-plex formation with thrombin, is linked by means of Cys430

to Cys247 in the main body of the inhibitor The results suggest that this peptide, which is not visible owing to its small size, harbors the eptiope that is recognized by M27 Binding of the mAb to reduced and nonreduced native and latent AT indicates that the mAb recognizes a linear sequence

These results warranted a test of the reactivity of mAb M27 in Western blotting with AT from mouse, rabbit and bovine blood plasma ATs from these species all differ from their human counterpart at three positions in the C-terminal peptide: residues 411, 416 and 432 (Fig 2A) [37] M27 did not react with AT from any of the three species (Fig 2B), whereas a control rabbit polyclonal antiserum against human AT gave positive results As mAb M27 is not sensitive to reduction of the Cys residue at position 430 of

AT, it is unlikely that the epitope is at the very C-terminus of the peptide The results therefore suggested that residues 411 and 416 are part of the epitope of M27

Synthetic peptides were used to localize the epitope of M27 more precisely (Fig 3A) Wells of microtitre plates were coated with the peptides for ELISA-binding studies A peptide including residues 404–420 (P-74) was found to bind M27 (Fig 3B) In contrast, a corresponding peptide, with substitutions at positions 411 (P-80), 416 (P-81), or both (P-79), showed a very weak reaction with M27 Native AT competed less well with the synthetic peptides than with native immobilized AT for binding to M27 A control peptide with the same composition as the 404–420 peptide, but with the sequence scrambled, did not react with M27 Identical results were obtained when the peptides were covalently linked to the wells of microtitre plates by a maleimide reaction with C-terminal Cys residues (data not

Fig 1 Western blotting of AT with mAb M27 Different forms of AT were electrophoresed by SDS/PAGE (12% gels), transferred to poly(vinylidene difluoride) (PVDF) membrane, and detected with M27 Nonreduced native, latent, elastase-cleaved, or thrombin–AT complex (1.7 pmol each) were electrophoresed in lanes 2–5, respect-ively The same amounts of dithiothreitol-reduced proteins were elec-trophoresed in lanes 6–9 The arrow points to the position of the

62 kDa molecular mass protein marker in lane 1.

shown) The results of the ELISA-binding studies are

consistent with the results of the Western blotting

experi-ments and establish that the sequence including FIREVP

(residues 411–416) constitutes a critical part of the linear

epitope of mAb M27

Effect of M27 on inhibition of thrombin by AT

The effect of mAb M27 on the formation of covalent

complexes between thrombin and AT was studied in a

two-stage assay First, thrombin and AT were incubated in

microtitre plate wells, with and without M27 Aliquots were

then removed at different time-points for SDS/PAGE and

for measurements of the amidolytic activity M27 inhibited

AT both in the presence and absence of heparin (Table 1)

In the absence of heparin the M27-mediated inhibition was

complete, whereas in the presence of heparin it was partial

The inhibition in the presence of heparin was not influenced

by increasing the mAb concentration from a four- to a

40-fold molar excess over the AT concentration, suggesting

that the heparin and mAb binding sites on AT are

independent of one another (data not presented) Solution

binding studies with Eu3+chelate-labeled native AT have

also indicated that there is no competition between heparin

and M27 for binding to AT (data not presented) As shown

in Fig 4, Western blotting of aliquots removed after the first stage of the assay also revealed that the presence of M27 blocked the T–AT formation Moreover, at most, a minute amount of AT could have been cleaved by thrombin when M27 was bound to AT A very weak band above the bands

of native AT in lanes 2 and 4 is probably a small amount of cleaved AT, which is an impurity of our native AT The bands correspond to the mobility of the cleaved inhibitor during SDS/PAGE

Measurement of AT binding to M27 by surface plasmon resonance

The binding of different forms of AT to immobilized intact IgG and the Fab fragment of mAb M27 was studied in real time by surface plasmon resonance using a BIACORE biosensor Representative binding curves and binding

Fig 2 Western blotting of AT from different species with M27 (A)

C-terminal sequences of AT from different species are compared

(human AT numbering) Amino acids identical to that of the human

are shown with dashes (–) The arrow indicates the thrombin cleavage

site in AT that forms the C-terminal 39 residue peptide Cys430, which

is disulfide bonded to Cys247, is underlined (B) Nonreduced native

human AT (0.34 pmol) was included in lanes 2 and 6 Equal amounts

of purified, nonreduced native mouse, rabbit or bovine AT were

included in lanes 3–5 and 7–9, respectively Protein markers were run

in lane 1 Membrane-bound proteins from lanes 1–5 were incubated

with M27, followed by incubation with rabbit anti-mouse IgG alkaline

phosphatase conjugate and enzyme color development

Membrane-bound proteins from lanes 6–9 were incubated in the same manner,

except with a primary polyclonal rabbit anti-(human AT) Ig followed

by secondary goat anti-rabbit IgG alkaline phosphatase conjugate

(DAKOPATTS).

Fig 3 Binding of M27 to immobilized synthetic peptides (A) Synthetic peptide sequences The vertical arrow designates the thrombin-clea-vage site in antithrombin (AT) Peptide P-78 corresponds to the 42 C-terminal residues of human AT Bold, underlined residues are changes from the wild-type sequence Peptide P-82 is a scrambled sequence, whose composition is the same as that of the wild-type plus a C-terminal cysteine residue (B) Peptides were immobilized on the surface

of polystyrene microtitre plate wells Binding of M27 was determined using an ELISA, with a final development of 405 nm absorption vs time Bars represent average values of duplicates at the 10-min time-point Error bars indicate the range Color development in the assay is still linear, with respect to time, at the 10-min time-point Key: black bars, 200 ng of M27 added; white bars, 1000 ng of M27 added; stip-pled bars, 200 ng of M27 + 2 lg of native AT added.

constants are presented in Fig 5 and Table 2 Results

obtained with immobilized intact mAb and the

correspond-ing Fab fragment were identical within experimental error

The derived dissociation constants ranged from 20 pMto

8 nM, i.e indicating that M27 has a high affinity for all

forms of AT The Kdfor native and elastase-cleaved AT

were almost identical The highest Kd, for latent AT, was the

result of both a decrease in the association rate constant (ka)

and an increase in the dissociation rate constant (kd) relative

to native AT, each about one order of magnitude M27 had

the highest affinity for the T–AT complex (KD

 2 · 10)11M) This can be attributed almost totally to a

very slow dissociation rate (Fig 5C)

Clearance of the T–AT complex in rats SECs are removed from the circulation by cellular receptors that recognize receptor-binding site(s) on the complex [11]

A pentapeptide (residues 408–412; FLVFI in AT) in the C-terminal fragment of the cleaved inhibitor has been implicated in receptor binding and internalization [38] As the M27-binding site on complexed AT (residues 411–416; FIREVP) partially overlaps the proposed SECR-binding site, a clearance study in rats was performed to determine whether the antibody would inhibit removal of the complex from the circulation The results obtained for the complex,

in the absence of the Fab fragment, and for the native and cleaved AT, were similar to that reported previously [39] Although the Fab fragment of M27 caused a small, but significant (P < 0.0001, two-factorialANOVA), reduction in the rate of clearance of the complex, from a half-life (t½) of

 7 min in the absence of the Fab fragment to a t½ of

 9 min in its presence, it did not block the uptake to the liver (Fig 6) Considering the very high affinity of the M27 Fab fragment for the T–AT complex, this result is not consistent with a model where the putative SECR-binding site on the T–AT complexed AT is important for complex clearance [11]

To determine whether the M27 could inhibit the binding

of T–AT to the SEC receptor in a less complex system, HepG2 cells were incubated with radiolabeled T–AT in the presence and absence of the antibody The binding of radiolabeled T–AT was reduced to 33 and 16% when 50-fold (not shown) and 100-50-fold molar excess of unlabeled T–

AT complex was added, respectively, which is in accordance with a previous study [38] The binding was 69 and 79%

in the presence of a 10-fold molar excess of M27 and Fab M27, respectively The high affinity of M27 to T–AT

Table 1 Effect of M27 on the inhibition of thrombin byAT Concentration and molar ratios of proteins in the second stage amidolytic assay are shown in parentheses V max is the rate of substrate hydrolysis, as measured by the change in absorbance at 405 nm over a period of 20 min In all cases the change was linear during this time-period Values represent the average and range of duplicate measurements In part A, first-stage components were combined and incubated for 60 min at 37 C prior to addition to the second stage In part B, first-stage components were combined and incubated for 5 min at 25 C prior to addition to the second stage Other details of the assay are described in the Materials and methods.

Part A

Thrombin (50 ngÆmL)1) + AT (1 : 1.04) 3.52 ± 0.12 Thrombin (50 ngÆmL)1) + AT (1 : 1.04) + M27 (1 : 1.04 : 3.7) 35.88 ± 1.30 Thrombin (50 ngÆmL)1) + M27 (1 : 3.7) 35.19 ± 1.53 Thrombin (50 ngÆmL)1) + AT (1 : 1.04) + Fab (1 : 1.04 : 3.7) 33.30 ± 1.47 Thrombin (50 ngÆmL)1) + Fab (1 : 3.7) 35.22 ± 1.36 Part B

Thrombin (50 ngÆmL)1) + Heparin (0.25 unitsÆmL)1) 31.28 ± 1.20 Thrombin (50 ngÆmL)1) + Heparin (0.25 unitsÆmL)1) + AT (1 : 2.6) 0.13 ± 0.00 Thrombin (50 ngÆmL)1) + Heparin (0.25 unitsÆmL)1) + AT (1 : 2.6) + M27 (1 : 2.6 : 9.8) 19.90 ± 0.77 Thrombin (50 ngÆmL)1) + Heparin (0.25 unitsÆmL)1) + M27 (1 : 9.8) 32.04 ± 1.28 Thrombin (50 ngÆmL)1) + Heparin (0.25 unitsÆmL)1) + AT (1 : 2.6) + Fab (1 : 2.6 : 9.8) 19.44 ± 0.83 Thrombin (50 ngÆmL)1) + Heparin (0.25 unitsÆmL)1) + Fab (1 : 9.8) 32.19 ± 0.86

Fig 4 M27 protection of antithrombin (AT) from thrombin cleavage.

Samples of native AT incubated for 40 min in the presence of

thrombin, with or without M27, were submitted to SDS/PAGE and

Western blotting with M27, as described in the legend to Fig 1 Lanes

1–5 contain nonreduced samples, and lanes 6–9 contain reduced

samples Lane 1, size markers; lane 2, 80 ng of AT before incubation;

lane 3, 60 ng of AT + thrombin; lane 4, 60 ng of AT + thrombin +

M27; lane 5, thrombin + M27; lane 6, 64 ng of AT before incubation;

lane 7, 48 ng of AT + thrombin; lane 8, 48 ng of AT + thrombin +

M27; lane 9, thrombin + M27 The intense bands at the top of lanes 4

and 5 are caused by reaction of M27 in the loaded samples with the

secondary antibody–enzyme conjugate The arrow denotes the

posi-tion of the 62 kDa molecular mass marker protein.

(KD3.7· 10)11M), ensured that > 99.9% of the T–AT was

complexed with M27 in these experiments Similar results

were obtained with a 50-fold molar excess of the antibodies

(not shown) As a control, HepG2 cells were incubated with

labeled T–AT in the presence of M38, which is an anti-AT

mAb that does not compete with M27 for binding to AT

The binding of labeled T–AT was decreased to 54% A

50-fold molar excess of cleaved AT decreased the binding to

79% These results also argue against a major role for the

FIREVP sequence in receptor-mediated complex binding

Discussion

We have generated and characterized a murine mAb, M27, possessing high affinity for human AT This antibody reacts with all naturally occurring forms of AT, with KDvalues ranging from 8 to 0.02 nM The epitope for M27 was identified by comparison of Western blotting results for reduced with nonreduced cleaved and noncleaved forms of

AT The results indicate that the epitope resides in the C-terminal 39-amino acid residue peptide generated by T–AT complex formation and cleavage Synthetic peptides helped define the epitope to the linear sequence, FIREVP (residues 411–416), although native AT could not compete with the synthetic immobilized peptides for the binding to M27, as well as with the immobilized AT One explanation could be that the antibody only binds with of one of its binding sites to AT and therefore could bind to a small immobilized peptide with the other It is clear that the binding of M27 to immobilized peptides is not saturated [100-fold more peptides (1.5 nmol) than AT (15 pmol) were added to the wells] Yet, the same amount of antibody gives only 50% of the signal in peptide-coated wells compared with AT-coated wells We consider that only a small fraction of the peptides immobilized on the plastic surface (without spacer arm) bind the mAb Another possibility is that our antibody is, to some extent, conformation dependent (Fig 1), but with its crucial binding pointing at the hexapeptide FIREVP

Several mAbs against human AT have been reported that map to the C-terminal region of the molecule Asakura

et al described a murine mAb (JITAT-16) raised to human

AT that recognizes the T–AT complex as well as cleaved

AT, but not native AT [40] The epitope of JITAT-16 (AAAST; residues 382–386) is just upstream of the Arg393 cleavage site [41] JITAT-16 destroyed the ability of AT to inhibit thrombin, as does M27 However, the mechanism of inhibition is different for the two antibodies JITAT-16 acts

by enhancing the hydrolysis of the T–AT acyl intermediate

to free, cleaved AT and active thrombin (normally a slow process) relative to the formation of a stable covalent complex Presumably, this results from delayed insertion of the RCL into b-sheet A In contrast, M27 seems to inhibit formation of the acyl intermediate quantitatively and hence complex formation and subsequent hydrolysis to the cleaved form of AT (Fig 4) Picard et al described a mAb (12A5) recognizing the linear sequence, DAFHK (residues 366–370), in the C-terminal region of AT [24] Antibody 12A5 also differs from M27 in that the former recognizes AT when it exists as a binary complex with thrombin, factor Xa and, to some exent, the P14-P9synthetic peptide, but not native, latent or cleaved forms of AT Dawes et al have reported a conformationally sensitive mAb (ESAH 1) that recognizes native, thrombin-com-plexed, and cleaved AT [42] The authors also observed that binding of heparin to AT counteracts the ability of ESAH 1

to neutralize AT inhibition of thrombin, but heparin binding had no effect on ESAH 1 binding to AT These observations are different from those (seen by us) for M27, where the effects of heparin and antibody on AT inhibition

of thrombin are independent of one another The epitope recognized by ESAH 1, involves residues 402–407 and 429 (FKANRP/P), all in the C-terminal cleavage peptide, but

Fig 5 BIACORE binding curves for M27 Fab interaction with AT.

Data obtained by surface plasmon resonance experiments were

ana-lyzed using the BIAEVALUATION 3.0 software package Symbols

repre-sent actual data points and solid lines are simulation curves, assuming

univalent, noninteracting binding sites (A) Concentrations of latent

AT are 24.1, 40.2, 56.2, 64.3, and 80.3 n M (B) Concentrations of

elastase-cleaved AT are 0.9, 2.3, 4.6, 9.2 and 18.4 n M (C)

Concen-trations of the T–AT complex are 0.9, 2.3, 4.3, 8.7 and 17.4 n M

(D) Concentrations of native AT are 0.8, 2.0, 4.0, 8.0 and 16.0 n M

distinct from the region recognized by M27 (residues 411–

416).Three-dimensional molecular models for native AT are

presented in Fig 7 Examination of the model reveals that

the epitope recognized by M27 resides in strand 4 of b-sheet

B of AT, on the face opposite b-sheet A into which the

reactive-center loop inserts Analyses of X-ray-derived

structures for the latent and cleaved forms of AT suggest

that the epitope is in the same general location for all forms

of AT and at least partially surface exposed (Fig 7) The

model derived from the crystal structures indicates that only

the side-chains of residues 413–416 are exposed to the

surface, in agreement with residue 416 playing an important

role in M27 binding However, X-ray structures showing

that residue 411 is buried is in apparent contradiction with

our peptide epitope mapping that implicates residue 411 in

M27 binding Examination of molecular models for native,

cleaved, and latent AT, based upon X-ray crystallography, does not lead to a clear explanation of the differences in the equilibrium-binding constants of M27 for the different forms

The antibody-binding site shown in Fig 7 is sufficiently close to the RCL of AT (40–50 A˚) to allow antibody-mediated blocking of the formation of the Michaelis-like complex with thrombin, even in the case of the Fab fragment, which typically has a length from the antigen-binding site to the papain cleavage site of 45 A˚ However, this explanation for the neutralization of AT by M27 would require that the bound antibody has very little free movement relative to the RCL, and is always in an orientation that blocks the binding of thrombin to AT

An alternative explanation is that binding of M27 distorts the conformation of the RCL in a manner that prevents

Table 2 Binding constants for M27 IgG and Fab interaction with AT Measurements were made with surface plasmon resonance on a BIAcore instrument and analyzed using the software package BIAEVALUATION 3.0 See the Materials and methods for details Values represent the mean (± standard deviation) of three independent determinations, except for AT native/Fab where only two determinations were made.

Analyte/ligand k a ( M )1 Æs)1) k Da (s)1) K D ( M )

AT native/IgG 3.85 · 10 5 (± 0.04) 5.64 · 10)5(± 0.20) 1.46 · 10)10(± 0.07)

AT native/Fab 4.64 · 105(± 0.04) 6.28 · 10)5(± 0.03) 1.36 · 10)10(± 0.02)

AT latent/IgG 4.20 · 10 4

(± 0.36) 2.87 · 10)4(± 0.11) 6.83 · 10)9(± 0.85)

AT latent/Fab 3.81 · 10 4 (± 0.46) 3.01 · 10)4(± 0.10) 8.02 · 10)9(± 1.06)

AT cleaved/IgG 3.14 · 10 5 (± 0.00) 5.18 · 10)5(± 0.16) 1.65 · 10)10(± 0.05)

AT cleaved/Fab 3.45 · 10 5

(± 0.02) 5.23 · 10)5(± 0.17) 1.51 · 10)10(± 0.06) T–AT complex/IgG 1.80 · 10 5 (± 0.11) 6.53 · 10)6(± 4.37) 3.73 · 10)11(± 2.57) T–AT complex/Fab 2.28 · 10 5 (± 0.14) 4.37 · 10)6(± 3.51) 1.98 · 10)11(± 1.66)

Fig 6 Effect of mAb M27 on T–AT clearance in vivo and in vitro (A)125I-T–AT complex (73 pmol in 500 lL) was injected into the tail vein of rats

in the absence (h) and presence (d) of M27 Fab fragment (2.9 nmol) Blood samples were collected at the time-points indicated and the radioactivity was measured As controls,125I-labeled diisopropylphosphoryl (DIP)-thrombin (j), native AT (n) and elastase-cleaved AT (s) were injected in the same manner as the complex The complexes with and without M27 Fab fragment were injected into two rats each The error bars represent the range (B) HepG2 cells were incubated with 200 lL of 20 n M125I-T–AT in the absence or presence of 2 l M unlabeled T–AT, 0.2 l M

Fab M27, 0.2 l M mAb M27, 0.2 l M mAb M38, and 1 l M cleaved AT Each bar represents the percentage of binding ± 2SD Binding of radiolabeled T–AT without competitor was set to 100%.

recognition by thrombin A precedent for such a subtle, yet

significant, conformational change is offered by heparin

binding to AT, resulting in a conformational change in the

RCL [33] Also consistent with this proposal is the ease with

which AT can adopt alternative conformations (e.g the

latent form) This explanation may also, in part, explain the

only partial neutralization by M27 observed in the presence

of heparin

In the presence of high-molecular-weight heparin, the M27-mediated block of AT inhibition was not complete and was not influenced by a 10-fold increase in antibody concentration (data not presented) This suggests that the binding sites on AT for antibody and heparin are independent of one another Second-order association rate constants of 8.9· 10)3 and 3.7· 10)7ÆM )1Æs)1 have been reported for T–AT in the absence and presence of heparin,

Fig 7 Molecular models of the M27 epitope on antithrombin.

respectively [43] Dissociation of M27 from AT (the Kd

value for the M27–AT complex is 5.6· 10)5Æs)1; t½, 3.4 h)

cannot account for the T–AT complex formation in the

presence of heparin We propose that heparin induces a

conformational change in the RCL of AT that allows slow

complex formation with thrombin, even with M27 bound

Lollar & Owen were the first to demonstrate that the

AT–125I-thrombin complex is rapidly cleared by the liver in

rabbits [44] Subsequent studies by others, involving

com-petitive clearance studies, indicated that a common pathway

exists for several SECs, including AT–thrombin, heparin

cofactor II–thrombin, a1-antitrypsin–trypsin, and a1

-anti-trypsin–elastase [45,46] Mast et al demonstrated that

the rate of SEC clearance is 10–50 times faster than that

of the corresponding free inhibitor [47] In the case of T–AT,

the t½ for the elimination of the complex from the

circulation is in the order of several minutes [45,47] The

first receptor identified as being involved in SEC binding

and clearance, termed SECR, was implicated by Perlmutter

and co-workers based on in vitro studies of HepG2 and

monocyte stimulation of a1-antitrypsin biosynthesis [20]

Subsequently, SECR was reported to recognize a minimal

pentapeptide sequence (FVFLM) in a1-antitrypsin, based

upon synthetic peptide competitive-binding studies [38] The

authors also proposed that the corresponding pentapeptide

(FLVFI in antithrombin) in the homologous serpin portion

of SEC complexes, is similarly and competitively recognized

by SECR In contrast to Perlmutter and co-workers,

Maekawa et al showed that five recombinant heparin

cofactor II–thrombin complexes, each with different single

mutations in the proposed receptor-binding pentapeptide,

were not prevented from binding and uptake to HepG2 cells

[22] In recent years, conjugates of poly Lys-peptides,

containing the FVFLM sequence, have been used to

effectively transfer DNA into hepatic cells both in vitro

and in vivo [48–50] Kounnas et al demonstrated the

importance of the low-density lipoprotein receptor-related

protein (LRP) in the clearance of SEC complexes in vivo and

in vitro[39] Their conclusions were based on the ability of

the receptor-associated protein, an inhibitor of LRP

activity, to prevent uptake of SEC complexes to liver in

rats and to HepG2 cells

We have demonstrated that mAb M27 recognizes a

hexapeptide segment (FIREVP), which overlaps the last

two residues of the pentapeptide (FLVFI) reported to be

recognized by the SECR We found that clearance of125

I-T–AT from the blood circulation of rats was only slightly

reduced by bound M27 Fab fragment Furthermore, the

Fab fragment only marginally reduced binding of the

complex to HepG2 cells The effect was similar to that

obtained with a control antibody against AT The limited

effect obtained with Fab M27 bound to T–AT is

presum-ably nonspecific and caused by to an increase in size and/or

change of charge Our interpretation is that the

penta-peptide site is, at most, marginally involved in complex

clearance

Acknowledgements

We acknowledge the technical assistance of Bjorn Hambe in purifying

antithrombin from bovine, rabbit and mouse plasma, and of Ulla

Persson in production of conditioned media containing the mAb, M27.

Financial support to G L L., while on sabbatical stay at the Department of Clinical Chemistry, Lund University Hospital, Malmo¨, was provided, in part, by the Wenner-Gren Foundation This work was supported by grants from the Swedish Medical Research Council (B96-03X-04487-22B and B96-03X-10825-03A), the Swedish Foundation of Strategic Research, the Kock Foundation, the Pa˚hlsson Foundations, and the Foundation of University Hospital, Malmo¨.

References

1 Carrell, R.W & Stein, P.E (1996) The biostructural pathology of the serpins: critical function of sheet opening mechanism Biol Chem Hoppe–Seyler 377, 1–17.

2 Gettins, P.G.W., Patston, P.A & Olsen, S.T (1996) Serpins: Structure, Function and Biology R G Landes, Austin, TX.

3 Gils, A & Declerck, P.J (1998) Structure–function relationships

in serpins: current concepts and controversies Thromb Haemost.

80, 531–541.

4 Schulze, A.J., Huber, R., Bode, W & Engh, R.A (1994) Struc-tural aspects of serpin inhibition FEBS Lett 334, 117–124.

5 Gettins, P.G.W (2002) Serpin structure, mechanism, and func-tion Chem Rev 102, 4751–4803.

6 Miletich, J.P., Prescott, S.M., White, R., Majerus, P.W & Bovill, E.G (1993) Inherited predisposition to thrombosis Cell 72, 477–480.

7 Florell, S.R & Rodgers, G.M (1997) Inherited thrombotic dis-orders: an update Am J Hematol 54, 53–60.

8 Fish, W.W., Orre, K & Bjo¨rk, I (1979) The production of an inactive form of antithrombin through limited proteolysis by thrombin FEBS Lett 98, 103–106.

9 Carrell, R.W., Stein, P.E., Fermi, G & Wardell, M.R (1994) Biological implications of a 3A˚ structure for dimeric antithrombin Structure 2, 257–270.

10 Huntington, J.A., Read, R.J & Carrell, R.W (2000) Structure of a serpin–protease complex shows inhibition by deformation Nature

407, 923–926.

11 Wells, M.J., Sheffield, W.P & Blajckman, M.A (1999) The clearance of thrombin–antithrombin and related serpin–enzyme complexes from the circulation: role of various hepatocyte recep-tors Thromb Haemost 81, 325–337.

12 Chang, W.-S.W., Wardell, M.R., Lomas, D.A & Carrell, R.W (1996) Probing serpin reactive-loop conformations by proteolytic cleavage Biochem J 314, 647–653.

13 Wardell, M.R., Chang, M.-S.W., Bruce, D., Skinner, R., Lesk, A.M & Carrell, R.W (1997) Preparative induction and chracterization of L-antithrombin: a structural homologue of la-tent plasminogen activator inhibitor-1 Biochemistry 36, 13133– 13142.

14 Carrell, R.W., Huntington, J.A., Mushunje, A & Zhou, A (2001) The conformational basis of thrombosis Thromb Haemost 86, 14–22.

15 Zhou, A., Huntington, J.A & Carrell, R.W (1999) Formation of antithrombin heterodimer in vivo and the onset of thrombosis Blood 94, 3388–3396.

16 Bruce, D., Perry, D.J., Borg, J.-Y., Carrell, R.W & Wardell, M.R (1994) Thromboembolic disease due to thermolabile conforma-tional changes of antithrombin Rouen-VI (187 Asn fi Asp).

J Clin Invest 94, 2265–2274.

17 Chang, W.-S.W & Harper, P.L (1997) Commercial antithrombin concentrate contains inactive L-forms of antithrombin Thromb Haemost 77, 323–328.

18 O’Reilly, M.S., Pirie-Shepherd, S., Lane, W.S & Folkman, J (1999) Antiangiogenic activity of the cleaved conformation of the serpin antithrombin Science 285, 1926–1928.

19 Larsson, H., Sjo¨blom, T., Dixelius, J., O¨stman, A., Ylinenja¨rvi, K., Bjo¨rk, I & Claesson-Welsh, L (2000) Antiangiogenic effects