Báo cáo khoa học: High molecular weight kininogen binds to laminin – characterization and kinetic analysis pot

However, at low but not at high concentrations of albumin, Zn2+ increased the affinity for the binding by abolishing an inhibitory effect of Ca2+.. Attempts to reduce the binding to the n

High molecular weight kininogen binds to

laminin – characterization and kinetic analysis

Inger Schousboe and Birthe Nystrøm

Department of Biomedical Sciences, The Panum Institute, University of Copenhagen, Denmark

Introduction

The extracellular matrix (ECM) controls a variety of

cellular functions by interacting with a vast array of

macromolecules, exhibiting a wide range of activities [1] Recent in vitro investigations have indicated that

Keywords

extracellular matrix; high molecular

weight kininogen; kininostatin; laminin;

Zn 2+ -independent

Correspondence

I Schousboe, Department of Biomedical

Sciences, The Panum Institute, University of

Copenhagen, Blegdamsvej 3C, DK-2200

Copenhagen, Denmark

Fax: +45 3536 7980

Tel: +45 3532 7800

E-mail: schousboe@sund.ku.dk

(Received 11 March 2009, revised 8 July

2009, accepted 16 July 2009)

doi:10.1111/j.1742-4658.2009.07218.x

High molecular weight kininogen (HK) is an abundant plasma protein that plays a central role for the function of the kallikrein⁄ kinin ⁄ kininogen sys-tem Thus, cleavage of HK by kallikrein liberates bradykinin, which stimu-lates vascular repair and a two-chain protein, activated HK (HKa), which induces apoptosis in proliferating endothelial cells The localization of these events remains obscure, although the basement membrane may be of importance Analyzing the interaction between HK and HKa and selected basement membrane proteins, we observed that they bound to the major noncollageneous proteins laminin, but not to vitronectin or fibronectin coated on microtiter plates The binding to laminin was Zn2+independent However, at low but not at high concentrations of albumin, Zn2+ increased the affinity for the binding by abolishing an inhibitory effect of

Ca2+ Recombinant human kininostatin encompassing the amino acid sequence, Arg439-Ser532 but not the endothelial cell binding peptide sequence (His479-His498; HKH20) within kininostatin inhibited the bind-ing of HKa to laminin This established that the amino acid sequence Arg439-Lys478 in domain 5 of HK is of importance for its binding to lami-nin Extensive proteolytic cleavage of HK and HKa with kallikrein abol-ished the binding to laminin, releasing a 12 kDa anti-kininostatin reacting peptide On the basis of these results, we propose that the binding of HK

to laminin is a primary event, which secures proper localization of the cleavage products for subsequent interaction with the endothelium to promote inflammatory and pro- and anti-angiogenic activities

Structured digital abstract

l MINT-7218019: Laminin alpha 5 (uniprotkb:Q61001), Laminin beta 1 (uniprotkb:P02469), Laminin gamma 1 (uniprotkb:P02468), Laminin alpha 5 (uniprotkb:Q61001), Laminin beta 2 (uniprotkb:Q61292) and Laminin gamma 1 (uniprotkb:P02468) physically interact (MI:0915) with HK (uniprotkb:P01042) by solid phase assay (MI:0892)

l MINT-7219326: Laminin alpha 1 (uniprotkb:P19137), Laminin beta 1 (uniprotkb:P02469) and Laminin gamma 1 (uniprotkb:P02468) physically interact (MI:0915) with HK (uniprotkb: P01042) by solid phase assay (MI:0892)

Abbreviations

ECM, extracellular matrix; FN, fibronectin; HK, high molecular weight kininogen; HKa, activated HK; HRP, horseradish peroxidase; HUVEC, human umbilical vein endothelial cells; LM, laminin; uPAR, urokinase plasminogen activator receptor; VN, vitronectin.

this array of macromolecules includes the surface

bind-ing proteins in the contact activation system of the

blood coagulation system This system consists of

factor XII, high molecular weight kininogen (HK),

prekallikrein and factor XI, but only factor XII and

HK interact directly with the ECM However, in the

plasma, prekallikrein and factor XI form complexes

with HK, enabling the assembly of the entire contact

activation system on the vascular wall [2–5] After

being assembled, the system functions locally

depen-dent on the demand for activation Thus, as a result of

factor XI activation by activated factor XII, the rate

of fibrin formation becomes enhanced [6], whereas

activation of prekallikrein by activated factor XII

enhances the proteolytic cleavage of HK by plasma

kallikrein [7] However, the localization of the

activa-tion remains to be revealed, although factor XII

recently was shown to bind to fibronectin (FN) [8],

and several endothelial cell membrane proteins have

been suggested as receptors for HK, including the

globular C1q receptor [9,10] and cytokeratin-1 [10–12]

The binding of HK to these receptors is strictly

depen-dent upon the free Zn2+ concentration Cleavage of

HK releases a short-lived strong inflammatory

nona-peptide, bradykinin, leaving behind a two-chain

acti-vated HK (HKa), which Zn2+-dependently binds to

the urokinase plasminogen activator receptor (uPAR)

[13] This binding is considered to inhibit angiogenesis

[14]

The process of angiogenesis is a complex event

requiring signals from both plasma and the

extracellu-lar basement membrane, and adhesive interactions of

endothelial cells with the underlying basement

mem-brane are instrumental in regulating the development

and maintenance of the vascular wall The basement

membrane contains a network of collagen and laminin

(LM) [15] Formation of new vessels involves the

migration and proliferation of cells To assist the cells

in their migration, the extravascular matrix provides

an environment consisting of hyaluronic acid,

vitronec-tin (VN) and FN [16] LM plays an important role in

cell adhesion to the basement membrane by interacting

in the endothelium with a series of integrins, including

a3b1, a6b1, avb1, avb3 and avb5 [16–18] The latter

three of these integrins are activated and engaged by

VN [19] when VN interacts with uPAR [20–22] The

induction of apoptosis in proliferating endothelial cells

by HKa [23] has been suggested to be the result of the

binding of HKa not only to uPAR, but also to VN

[24–26], preventing the interaction between VN and

uPAR However, the apoptotic effect of HKa is

appar-ently regulated by several ECM proteins [11,24] Thus,

Guo et al [24] observed that the adhesion of

endothe-lial cells cultured on VN, but not that of cells cultured

on FN, was inhibited by HKa, whereas Sun and McC-rae [14] demonstrated that HKa induced apoptosis of endothelial cells cultured on not only VN, but also on

FN and LM Whether this inhibition is the result of the binding of HKa to the ECM proteins remains to

be revealed

LMs are a family of glycoprotein heterotrimers com-posed of an a, b and c chain To date, five a, four b and three c LM chains have been identified that can combine to form 15 different isoforms [15,27] The prototype of LMs is LM 1 The LM 1 isoform is char-acterized by the presence of one LM a1 chain, which combines with one LM b1 chain and one LM c1 chain LMs expressed in endothelial cells are character-ized by the presence of LM a4 and a5 chains, which combine with LM b1 and c1 chains to form LM 8 and

LM 10, respectively In LM 11, which is also present

in the endothelium, one a5 chain combines with b2 and c1 chains [15]

In the present study, using a solid phase binding assay, we analyzed the binding of HK and HKa to

LM, FN and VN, and showed that both HK and HKa bind with high affinity to LM The LMs used in the study comprised LM 1 and LM 10⁄ 11, the latter of which are characteristic of the endothelium

Results

HK and HKa binding to proteins of the extracellular membrane

To analyze the ability of HK to bind to selected pro-teins present in the extracellular membrane, freshly drawn citrate anti-coagulated plasma was incubated on

LM, VN or FN coated on microtiter plates The amount of HK absorbed from the plasma by these matrix proteins was subsequently analyzed by immu-noreactions using a monoclonal antibody to the heavy chain of HK (mAb 2B5) as the primary antibody This demonstrated that HK apparently could be extracted from plasma by binding to all three matrix proteins However, only the amount extracted by LM was higher than the amount extracted by the noncoated surface (Fig 1A) Analyzing the binding of purified

HK and HKa to LM, VN and FN showed that, over-all, a larger amount of HKa than of HK bound when incubated at the same concentration However, the amount of HKa bound to VN was identical to the amount bound to the noncoated plate, whereas the amount bound to LM and FN was lower By contrast, relative to the amount bound to the noncoated plate, more HK bound to LM and VN than to FN (Fig 1B)

Attempts to reduce the binding to the noncoated plate

by increasing the BSA concentration from the standard

concentration of 3.5 mgÆmL)1 to as much as

75 mgÆmL)1 in the block buffer gradually decreased

the amount of HKa bound nonspecifically, but had no

influence on the amount of HKa bound to LM, VN

and FN relative to the amount bound to the

noncoat-ed surface (results not shown) Because a considerable

amount of HKa bound to the noncoated microtiter

plate, we next analyzed whether the concentration of

LM and VN was sufficiently high to saturate the

sur-face of the microtiter plate, thereby preventing the

problem of nonspecific binding This was achieved by increasing the concentration of the LM and VN coated

on the microtiter plate As indicated above, HK did not bind to the noncoated plate With increasing LM concentrations up to 5 lgÆmL)1, increasing amounts of

HK bound to LM By contrast, the amount of HKa that bound to the noncoated plate decreased as the concentration of LM increased to 5 lgÆmL)1 A stant amount of HK and HKa bound to LM at con-centrations higher than 5 mgÆmL)1 (Fig 2A) The binding of HKa to LM was inhibited by the presence

of soluble LM, but not by the presence of soluble VN

or FN (data not shown)

Zn2+has been shown to play a determining role for the interaction of HK and HKa with all previously identified receptors and ligands The presence of Zn2+ increased the amount of HK as well as HKa bound to the noncoated plate but, analogous to the binding in

1.2 1.4 1.6

1 0.8 0.6 0.4 0.2

0

0 0.2 0.4 0.6 0.8 1 1.2

None

LM

FN

VN

None

LM

FN

VN

HK/HKa bound (absorbance units)

HK/HKa bound (absorbance units)

Fig 1 Extraction of HK from plasma (A) and of HK and HKa (B)

from solutions of purified proteins by immobilized LM, VN and FN.

(A) Blood was drawn by the syringe method using one volume of

3.8% (w ⁄ v) sodium citrate to nine volumes of blood and

cen-trifuged at 1180 g for 5 min Aliquots of 100 lL of the supernatant

(citrate anti-coagulated plasma) were then added to wells coated

overnight with LM (10 lgÆmL)1), VN (5 lgÆmL)1) and FN

(10 lgÆmL)1), respectively, and blocked with Locke’s buffer

contain-ing 1% (w ⁄ v) BSA (B) Alternatively, the wells were incubated with

purified solutions of 20 n M HK (hatched columns) or 20 n M HKa

(grey columns) in Locke’s buffer containing 0.35% (w ⁄ v) BSA After

1 h of incubation, the content in the wells was removed and the

plates were washed and incubated with primary and secondary

antibodies, as described in the Experimental procedures Binding to

the surface coated with buffer alone was used as a measure

of nonspecific binding The results are the mean ± SD (n = 3) as

indicated by vertical bars.

0 1 2 2.5

1.5

0.5

3 3.5 4 4.5

HK HKa

HK + Zn HKa + Zn

HK HKa

HK + Zn HKa + Zn

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0

10 8

6 4 2

LM concentration (µg·mL–1)

VN concentration (µg·mL –1 )

A

B

Fig 2 Binding of HK and HKa to wells coated with increasing con-centrations of LM (A) and VN (B) Wells were coated overnight with increasing concentrations of LM (A) and VN (B) in coating buffer as indicated and subsequently blocked with Locke’s buffer containing 1% (w ⁄ v) BSA Next, the wells were incubated for 1 h with 20 n M

HK or 20 n M HKa in Locke’s buffer containing 0.35% (w ⁄ v) BSA in the presence (open symbols) or absence of 50 l M Zn 2+ (closed symbols) The amount of HK (squares) and HKa (circles) bound after extensive washing was determined by incubation with mAb 2B5 The results are the mean ± SD (n = 3) as indicated by vertical bars when extending beyond the symbols.

the absence of Zn2+, the binding to LM decreased

with increasing concentrations of LM and became

constant at concentrations higher than 5 lgÆmL)1

(Fig 2A) Therefore, a standard concentration of

10 lgÆmL)1LM was used throughout the study

A similar analysis, performed with VN as the

immo-bilized ligand, showed no binding of HK at any

con-centration of VN in the absence of Zn2+ (Fig 2B)

The high amount of HKa that bound to the noncoated

plate (Fig 1B), particularly in the presence of Zn2+,

decreased exponentially with increasing VN

concen-trations over the whole VN concentration range

(0–20 lgÆmL)1) (Fig 2B) This indicated that any

concentration of VN lower than 20 lgÆmL)1 was

insufficient to saturate completely the surface in the

microtiter plate Thus, the binding of HK and HKa to

VN shown in Fig 1B was most likely nonspecific

The effect of Zn2+on the binding of HK and HKa

to LM

Further investigations of the effect of Zn2+ on the

binding of HKa to LM showed that the amount of

bound HKa increased with increasing Zn2+

concentra-tions up to 15–20 lm and then decreased (Fig 3)

Because the enhancement was more pronounced at

lower than at higher HKa concentrations, this indi-cated that the effect of Zn2+ might be caused by a change in the affinity for the binding of HKa to LM Titration of LM with increasing concentrations of HKa in the presence and absence of 20 lm Zn2+ showed that the presence of Zn2+ enhanced the affin-ity of the binding, but apparently had no or only little effect on the maximum amount of bound HKa (Fig 4) Moreover, in the presence of Zn2+, the amount of HKa bound at high HKa concentrations was constant, indicating that HKa did not bind non-specifically to LM

The LM isoform used in the above measurements was the prototype, LM 1 To determine whether HK and HKa would bind also to LM 10⁄ 11, the dissocia-tion constants, KD, for the binding were determined in

a series of identical experiments using LM 1 as well as

LM 10⁄ 11 This revealed that, in the absence of Zn2+,

HK and HKa bound with the same affinity, regardless

of whether LM 1 or LM 10⁄ 11 had been coated on the microtiter plate (Table 1) The presence of Zn2+ affected more the affinity of the binding of HKa than

HK Thus, a five- to seven-fold increase was observed

in the affinity for the binding of HKa to both LM 1 and LM 10⁄ 11, whereas only a three-fold increase was observed for the binding of HK to LM 1 (Table 1) The presence of Zn2+had no or only a minimal effect

on the maximal amount bound in each individual experiment (data not shown)

0

50

100

150

200

250

Concentration of Zn 2+ (µ M )

Fig 3 Binding of HKa to LM as a function of the concentration of

Zn2+ A microtiter plate was coated overnight with LM (10 lgÆmL)1)

and blocked in Locke’s buffer containing 1% (w ⁄ v) BSA Next, it

was incubated with either 10 n M HKa (open squares) or 30 n M HKa

(open circles) diluted in Locke’s buffer containing 0.35% (w ⁄ v) BSA

and supplemented with increasing concentrations of Zn 2+ The

amount of HKa bound to the LM after incubation for 1 h was

deter-mined using mAb 2B5, as described in the Experimental

proce-dures Using the amount of HKa bound to LM in the absence of

Zn 2+ as the reference (100%), the enhancement of the binding at

the varying Zn 2+ concentrations is shown as a percentage The

results are the mean ± SD (n = 3) as indicated by vertical bars

when extending beyond the symbols.

3.000 2.500 2.000 1.500 1.000 0.500 0.000 0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0

Concentration of HKa (n M )

Fig 4 Concentration dependent binding of HKa to LM in the pres-ence or abspres-ence of Zn 2+ A microtiter plate was coated overnight with LM (10 lgÆmL)1) and blocked in Locke’s buffer containing 1% (w ⁄ v) BSA, as described in the legend to Fig 2 Next, it was incu-bated with increasing concentrations of HKa diluted in Locke’s buf-fer containing 0.35% (w ⁄ v) BSA in the presence (open circles) or absence (closed circles) of 20 l M Zn 2+ The amount of HKa bound

to the LM after incubation for 1 h was subsequently determined using mAb 2B5, as described in the Experimental procedures The results are the mean ± SD (n = 3) as indicated by vertical bars when extending beyond the symbols.

The nonhyperbolic shape of the Zn2+-dependency

curve (Fig 3) indicated further that the effect of Zn2+

might be explained by the presence of other divalent

cations in the incubation mixture To verify this, the

standard constituents, Ca2+ and Mg2+ in the

incuba-tion buffer (Locke’s buffer) were excluded This

enhanced the amount of HKa bound to LM in the

absence, but not in the presence, of Zn2+ and

indi-cated that the effect of Zn2+ was to abolish an

inhibi-tory effect of Mg2+ or Ca2+ or both Excluding one

of these ions at a time showed that the inhibition was

mainly the result of the presence of Ca2+ The amount

of HKa bound to LM in the presence of Zn2+ was

not or only slightly affected by the depletion of Ca2+

and Mg2+, as long as the BSA concentration was low

(3.5 mgÆmL)1) At a higher BSA concentration, the

presence of 20 lm Zn2+ was unable to abolish the

inhibitory effect of Mg2+and Ca2+(Table 2)

Identification of the binding region in HKa

The amino acid sequence Arg439-Ser532 in domain 5

of HK (Fig 5) [28,29], and particularly the

His479-His498 sequence, have been mapped as the endothelial

cell-, ECM- and surface-binding site [11,13,30] To

define the binding site within HK for the binding to

LM, we determined whether a recombinant human

kininostatin peptide (rhkininostatin; Arg439-Ser532) and

a synthetic dodeca-peptide (His479-His498; HKH20)

competitively inhibited the binding of HKa to LM

We also tested whether a synthetic sequence of amino

acids copying the Ser372-Arg419 sequence

N-termi-nally to kininostatin affected the binding This

sequence is cleaved off secondary to bradykinin Only

rhkininostatin inhibited the binding and, both in the

presence and absence of Zn2+, a 50% inhibition was

observed at a 500 molar excess of rhkininostatin

com-pared to the concentration of HKa (Fig 6) However,

the presence of HKH20, even at a 5000 molar excess

of HKa, had no effect This indicates that the N-terminal region of kininostatin encompassing the Arg439- Lys478 sequence might be of importance for the binding of HKa to LM Further investigations reveled that the binding to LM was abolished if HK and HKa had been pre-incubated at increasing lengths

of time up to 60 min with kallikrein at a 1 : 1 molar concentration ratio (Fig 7A) This was not the result

of a time-dependent binding of kallikrein to HK and HKa because no inhibition was observed when HK and HKa were incubated with kallikrein for the same length of time in the presence of protease inhibitors (0 time data point) The use of western blotting at reduced conditions to follow the progression of the kallikrein catalyzed cleavage of HK revealed the gener-ation of a 55 kDa heavy chain fragment as visualized

by the mAb 2B5 antibody Complete cleavage was seen only at equimolar concentrations of HK and kallikrein (Fig 7B) Visualization of the cleavage products using anti-rhkininostatin IgG revealed that the 45 kDa light chain generated after 60 min of incubation of HK with

a 1 : 10 molar concentration of kallikrein became par-tially cleaved, generating a 12 kDa anti-rhkininostatin reacting peptide, when HK was incubated for 60 min with an equimolar concentration of kallikrein (Fig 7C)

Table 1 The effect of Zn 2+ on K D for the binding of HK and HKa

to LM 1 and LM 10 ⁄ 11 The concentration of Zn 2+

was 20 l M Data are the mean ± SD (n).

KD(n M )

Hka

HK

Table 2 The effect of Ca 2+ , Mg 2+ and Zn 2+ on the binding of HKa

to LM The complete binding buffer (Locke’s buffer) contained physiological concentrations of Ca 2+ (2.3 m M ) and Mg 2+ (1 m M ) The experiment was performed as described using the buffer com-position of Locke’s buffer, but lacking the divalent cations indi-cated Some of the experiments were performed at a 10-fold higher BSA concentration, as indicated The results are representa-tive of one of three experiments performed in triplicate and are shown as the mean ± SD A statistical analysis was performed using one-way analysis of variance followed by the Bonferoni post-hoc test.

Locke’s buffer

[BSA]

(mgÆmL)1)

HKa bound in the absence

of Zn 2+

(absorbance units)

HKa bound in the presence

of Zn 2+ (20 l M ) (absorbance units) Complete (control) 3.5 0.806 ± 0.014 2.133 ± 0.005 Lacking 2.3 m M Ca2+

and 1 m M Mg 2+

3.5 1.580 ± 0.044* 1.929 ± 0.032** Lacking 1 m M Mg 2+ 3.5 0.963 ± 0.017 1.795 ± 0.058** Lacking 2.3 m M Ca 2+ 3.5 1.538 ± 0.027* 2.190 ± 0.042 Complete (control) 35.0 0.835 ± 0.036 0.879 ± 0.024 Lacking 2.3 m M Ca 2+

and 1 m M Mg 2+

35.0 1.470 ± 0.018* 1.675 ± 0.072*

Statistically significant difference from respective control:

*P < 0.001; **P < 0.005.

Binding of HK to ECM

Although HK has been shown to bind Zn2+

-depen-dently to the ECM generated during the growth of

human umbilical vein endothelial cells (HUVEC) and

ECV304 cells of a carcinoma cell line [5], the target for

the binding was not identified Because LMs are being

deposited in vivo in the basement membrane by

prolif-erating cells, it was next determined whether LM was

present in ECM generated during growth of HUVEC

Immunostaining of ECM with anti-LM IgG verified

the presence of LM in this matrix (data not shown)

Analysing the binding of HK to ECM, the influence

from binding to the cell-free surface in the culture dish

had to be taken into account because a confluent layer

of cells covers only the bottom of the cell culture dish

Thus, the cell-free surface would be expected to

account for approximately two-thirds of the surface in

a 96-well cell culture plate incubated with 100 lL per well, and hence be accessible for nonspecific binding Because a considerably higher amount of HKa than

HK bound nonspecifically to the noncoated surface (Figs 1 and 2), and this could not be prevented by increasing the concentration of BSA, only the binding

of HK was used in this part of the study However, at conditions in which cell-free areas were blocked with a 0.2% (w⁄ v) gelatin or a 0.35% (w ⁄ v) BSA, no HK bound to ECM in the absence of Zn2+ and, in its presence, a higher amount of HK bound to the cell-free surface than to ECM (Fig 8) Increasing the BSA concentration increased the optimal concentration of

Zn2+for the binding of HK without blocking the non-specific binding (data not shown) This excludes the possibility of measuring the binding of HK to ECM and suggests that HK might bind to one of the com-pounds in the cell culture medium that was absorbed

on the surface of the well when the cells were growing

Discussion

There are numerous investigations showing that

HK⁄ HKa binds Zn2+-dependently to different recep-tors on the surface of endothelial cells, and only a few analyzing the possibility of the interaction of the kini-nogens with the proteins in the basal membrane, which

is of equal importance when explaining the in vivo effect of HKa Therefore, the binding of HK and HKa

to selected noncollageneous proteins could be impor-tant for the function of the kallikrein⁄ kinin ⁄ kininogen system on the vascular wall During the present study,

it was shown that both HK and HKa bind to LM, which is the most abundant noncollageneous protein

in the basal membrane [15] The binding was inhibited

by rhkininostatin, but not by the surface binding peptide sequence (His479-His498; HKH20) within kininostatin, which has been identified as the sequence that is responsible for the Zn2+-dependent binding of

HK⁄ HKa to endothelial cell [11] Equimolar concen-trations of HK⁄ HKa and kallikrein cleaved off an anti-rhkininostatin reacting peptide from HKa,

abol-n i n i k d r

B S 2 – R 9 K 0 – Q 8 R 9 – K 8 N 9 – S 2

Domain 5

n i e r k il a K

H 9 – H 8

n i t a t s o i n i K h r

e c n u e s g i d i b c a f r u S

LM binding sequence

Fig 5 Functionally identified fragments of

domain 5 Numbering of the N- and C-

ter-minal amino acids of indicated fragments

and the generally accepted cleavage sites of

kallikrein are based on the previously

reported sequence [28].

0.5

None

HKa bound (absorbance units)

Plus zink Minus zink

Fig 6 Specificity of HKa binding to LM in the presence or absence

of Zn 2+ A LM coated microtiter plate (Fig 2) was incubated with

HKa diluted in Locke’s buffer containing 0.35% (w⁄ v) BSA in the

presence (closed columns) or absence (open columns) of 20 l M

Zn 2+ In the experiments indicated, binding was measured in the

presence of rhkininostatin (1 and 5 l M ), the amino terminal

sequence of the light chain (Ser372-Arg419; 10 l M ) and the surface

binding peptide, HKH20 (50 l M ) The amount of HKa bound to LM

after incubation for 1 h in the presence or absence of the effectors

was determined using mAb 2B5, as described in the Experimental

procedures The antibodies did not react with any of the effectors.

The results are the mean ± SD (n = 3) as indicated by vertical bars.

ishing the binding of HK⁄ HKa to LM This, when

combined with the inhibition of the binding by

rhkini-nostatin, indicates that the binding of HK⁄ HKa to

LM is mediated via the N-terminal region in

kinino-statin encompassing the amino acid sequence

Arg439-Lys478 (Fig 5)

Zn2+ has been assumed to have a decisive role with

respect to the binding and function of HK and HKa

In the present study, we show that Zn2+ is not

required for the binding of either HK or HKa to LM

However, it enhanced the affinity of the binding of

both HK and HKa to LM, although the total amount

bound was not affected However, this was without

significance because the binding affinity, even in the

absence of Zn2+, was approximately 15-fold higher

than the plasma concentration of HK, demonstrating

no obvious need for the presence of Zn2+ Further-more, the effect of Zn2+ on the binding to LM was only evident at low concentrations of BSA, at which it abolished the inhibitory effect of Ca2+ At a higher BSA concentration, the effect of Ca2+ remained, whereas the effect of Zn2+was eliminated by reducing its free concentration by binding to BSA Because the total plasma concentration of Zn2+ is 25 lm [31], the free Zn2+ concentration in vivo may never be suffi-ciently high to influence either the function of HKa or the binding of HK and HKa to LM Therefore, the present study, showing that HK and HKa bind Zn2+ -independently to LM, must be physiologically relevant Experimentally, there was a considerable difference

in the minimal amount of matrix protein required to cover the microtiter plate The concentration of LM

0 0.5 1 1.5 2 2.5

10

Incubation period (min)

HK: PK (10 : 1) HK: PK (1 : 1) HKa: PK (10 : 1) HKa: PK (1 : 1)

A

c/60 b/60 HK

b/60 b/30

120 kDa

120 kDa

45 kDa

12 kDa

55 kDa

C

B

Fig 7 Incubation with kallikrein abolished the binding of HKa and HK to LM (A) by cleaving the light chain of HKa (B, C) In Locke’s buffer containing 0.35% (w ⁄ v) BSA, one volume of HK and HKa, respectively, at a concentration of 60 n M was incubated for varying lengths of time with one volume of either 6 n M (10 : 1) or 60 n M (1 : 1) kallikrein (A) At the times indicated, cleavage was stopped by adding one vol-ume of a protease inhibitor cocktail consisting of 40 lgÆmL)1each of leupeptin, aprotenin, benzamidin and soy bean trypsin inhibitor, giving

a final concentration of 20 n M of HK and HKa, respectively The binding of the proteolysed HK (squares) and HKa (circles) to LM coated on a microtiter plate at a concentration of 10 lgÆmL)1(Fig 2) are shown after incubation at high (closed symbols) and low (open symbols) con-centrations of kallikrein Western blots of incubation mixtures of HK and kallikrein followed the proteolysis (B) Using the mAb 2B5 towards the heavy chain showed that, at a low kallikrein concentration, a higher amount of HK was cleaved after 60 min (b ⁄ 60) than after 30 min (b ⁄ 30) of incubation, whereas, at equimolar concentrations of HK and kallikrein, HK was completely cleaved after 30 min (c ⁄ 30) incubation and a considerable amount was already cleaved after 0.5 min (c ⁄ 0.5) (C) Using the anti-rhkininostatin IgG specific for the light chain, it was found that 60 min of incubation of equimolar concentrations of HK with kallikrein (c ⁄ 60) apparently cleaved the light chain generating a

12 kDa anti-rhkininostatin recognizable peptide.

was 5 lgÆmL)1, whereas more than 20 lgÆmL)1 was

required for VN, and almost no HKa bound to VN at

this concentration, even in the presence of Zn2+ A

previous study reported that HKa binds Zn2+

-depen-dently to VN coated at a concentration of 5 lgÆmL)1

[26] At present, this discrepancy cannot be explained

because the use of a considerably higher BSA

concen-tration for blocking did not prevent Zn2+-dependent

nonspecific binding However, the nonspecific binding,

particularly that of HKa, in the presence of Zn2+may

explain the controversial results reported on the effect

of matrix proteins on regulation of the apoptotic effect

of HKa on HUVEC grown in culture HKa induced

Zn2+-dependent apoptosis of the cells, regardless of

the cells being cultured on 2 lgÆmL)1 VN or LM

[14,24,26,32], although not on 10 lgÆmL)1 FN

[24,26,32]

It was not possible to measure binding of HK to

LM in ECM generated during the growth of HUVEC

This may likewise be the result of too low a

concentra-tion of the LM deposited during the 2–4 days of cell

growth The addition of Zn2+ enhanced the

nonspe-cific binding and indicated that the previously shown

Zn2+-dependent binding of HK to ECM [5] might

have been nonspecific The strongest evidence for this

was that a lower amount of HK bound Zn2+

-depen-dently to ECM than to the cell-free wells, implying that ECM blocked the binding of HK to the cell-free area of the well Furthermore, the Zn2+ optimum for the binding of HK to the surface of the microtiter plate was exactly the same as those reported for the binding of HK to ECM and HUVEC [4,5,33,34], regardless of whether the binding assay was performed

in buffer supplemented with gelatin or albumin Accordingly, and because the correction for nonspe-cific binding in these previous studies was performed

by the subtraction of binding in the absence of Zn2+,

a series of investigations investigating the significance

of Zn2+for the interaction between HK and endothe-lial cells in cell culture systems [4,5,13,23,34,35] needs

to be revisited

In vivo, LM might be a matrix protein of signifi-cance with respect to the function of HKa on the activity of the endothelial cells in as much as LM plays

an important role in cell adhesion to the basement membrane [16–18] Furthermore, both HK and HKa bind to LM, suggesting that this may be a prerequisite for the kallikrein⁄ kinin ⁄ kininogen system to be assem-bled and activated to enable the local liberation of the short-lived bradykinin and its participation in inflam-matory and pro- and anti-angiogenic reactions [25,36,37] Thus, the binding of HK to LM in the basement membrane may be followed by activation of prekallikrein bound in a 1 : 1 molecular complex to

HK The activation of prekallikrein is accomplished either by factor XIIa, heat shock protein 90 [38] or a prolylcarboxypeptidase [39] The observation that kallikrein cleaves off a 12 kDa anti-rhkininostatin reacting peptide shows that kallikrein, secondary to bradykinin, releases kininostatin from HK, as previ-ously anticipated [36] This fits neatly with the observa-tion that the HKa signaling of an apoptotic effect is promoted as a result of extensive cleavage by kallikrein [37,40]

Experimental procedures

Materials

One-chain HK, two-chain HKa and kallikrein delivered lyophilized from Enzyme Research Laboratories Ltd (Swansea, UK) were dissolved, aliquoted and stored in sili-conized test tubes at –80C All dilutions of HK, HKa and kallikrein were likewise performed in siliconized test tubes and excess dilutions were discarded VN, FN, LM from Engelbreth-Holm-Swarm murine sarcoma basement mem-brane (LM 1) and LM from placenta (LM 10⁄ 11) were obtained from Sigma Chemicals (St Louis, MO, USA) The surface binding peptide sequence within the light chain of

0

0.2

0.4

0.6

0.8

1

1.2

Concentrarion of Zn 2+ (µ M )

Fig 8 Binding of HK to ECM and cell-free wells Cell-free wells,

which had been exposed to growth medium, exactly as for the

cul-tures of HUVEC were, similarly to a confluent layer of the cells,

rinsed with NaCl ⁄ P i and incubated with EDTA Next, the wells

(ECM and cell-free) were blocked for 30 min with Locke’s buffer

containing 0.2% (w⁄ v) or 0.35% (w ⁄ v) BSA (blocking solutions) and

subsequently incubated for 1 h with 20 n M HK diluted in the

respective blocking solutions containing varying concentrations of

Zn2+ The amount of HK bound was measured by

immunoreac-tions, as described in the Experimental procedures A higher

amount of HK bound Zn 2+ -dependently to cell-free wells (open

symbols) than to ECM (closed symbols), regardless of whether the

surface had been blocked with gelatin (circles) or BSA (triangles).

The results are shown as the mean ± SD (n = 3) when extending

beyond the symbols.

HK,

His479-Lys-His-Gly-His-Gly-His-Gly-Lys-His-Lys-Asn-Lys-Gly-Lys-Lys-Asn-Gly-Lys-His498 (HKH20), was a

kind gift from Dr Alvin Schmaier (Case Western Reserve

University, Cleveland, OH, USA) The sequence of amino

acids copying the Ser372-Arg419 region [28] was

synthe-sized at Novo Nordisk (Ma˚løv, Denmark) Recombinant

human kininostatin (rhkininostatin) and goat anti-(human

rhkininostatin) were obtained from R&D Systems

(Abing-don, UK) The monoclonal antibody to the heavy chain of

human HK (mAb 2B5) was obtained from Abcam

(Cam-bridge, UK) Horseradish peroxidase (HRP) conjugated

rabbit anti-(mouse IgG) (P-0260), HRP-conjugated rabbit

anti-(goat IgG) (P-0449) and ortophenylenediamine were

obtained from Dako (Glostrup, Denmark) SuperSignal

West Femto Maximum Sensitivity Substrate was from

Pierce Biotechnology, Inc (Rockford, IL, USA) Essentially

fatty acid free BSA (A7030) was obtained from Sigma

Chemicals All other reagents were of the purest grade

commercially available

Solid phase binding assay

Maxisorp microtiter plates (high binding capacity; Nunc,

Roskilde, Denmark) were coated overnight at 4C with

150 lL of 5 lgÆmL)1VN or 10 lgÆmL)1of either FN, LM

1 or LM 10⁄ 11 in coating buffer (0.1 m potassium

phos-phate, pH 7.4), or at the concentrations indicated After

blocking for 30 min with 200 lL of 1% (w⁄ v) BSA in

Locke’s buffer (154 mm NaCl, 5.6 mm KCl, 3.6 mm

NaH-CO3, 2.3 mm CaCl2, 1 mm MgCl2, 5.6 mm glucose, 5 mm

Hepes, pH 7.4), 100 lL of HK or HKa (20 nm or as

indi-cated) diluted in Locke’s buffer containing 0.35% (w⁄ v)

BSA was added After incubation for 1 h at room

tempera-ture, the wells were washed three times with Locke’s buffer,

one time with ice-cold methanol and five times with wash

buffer [50 mm Tris, 0.15 mm NaCl, pH 8.0, containing

0.05% (v⁄ v) Tween 20] This was followed by 1 h of

incu-bation with mAb 2B5 diluted 5000-fold in wash buffer

con-taining 1% (w⁄ v) dry skim milk After a further washing

cycle with wash buffer (five times), the plate was incubated

for 1 h with HRP-conjugated rabbit anti-(mouse IgG)

diluted 5000-fold in wash buffer Finally, the plates were

incubated for 10–30 min with ortophenylenediamine,

dissolved in water according to the manufacturer’s

instruc-tions The peroxidase reaction was stopped by a two-fold

dilution with 0.5 m H2SO4and the relative amount of HK

bound to the wells determined as absorbance units (A) at

490 nm All experiments were performed in triplicates and

repeated at least twice To obtain estimates of affinity

con-stants, the data were analyzed according to the isotherm:

A= Amax· [B] ⁄ (KD+ [B])

where [B] is the molar concentration of the analyt, which is

either HK or HKa; A is the absorbance of the oxidized

HRP substrate, which is assumed to be proportional to the

amount of bound analyt; and Amax represents the absor-bance at saturating concentrations of the analyt

Western blotting

Proteolytic cleavage of HK was visualized by western blotting of aliquots of HK incubated with kallikrein in Locke’s buffer Reduced samples were separated on 4–12% SDS-PAGE simultaneously with a standard sample

of a mixture of Mr markers After blotting to a poly (vinylidene difluoride) membrane according to standard procedures, the membrane was incubated for 10 min in Tween-BSA block-buffer [50 mm Tris, 0.15 mm NaCl, pH 8.0, containing 0.1% (v⁄ v) Tween 20 and 0.1% (w ⁄ v) BSA], and subsequently overnight with the primary anti-body; either mAb 2B5 or goat anti-(human rhkinino-statin), diluted 5000-fold in 1% (w⁄ v) dry skim milk in the Tween-BSA block-buffer The positions of the light and the heavy chains of HK were visualized by incubation with HRP conjugated rabbit anti-(mouse IgG) (P-0260) and HRP-conjugated rabbit anti-(goat IgG) (P-0449), respectively, followed by SuperSignal West Femto Maximum Sensitivity Substrate as recommended by the manufacturer The results were monitored using a chemi-luminator

Endothelial cell culture

HUVEC (Clonetics, Cambrex Bio Science, Verviers, Belgium) were prepared as previously described [41] Briefly, the cells were grown to confluence under standard conditions, cryopreserved and sub-cultured in half of a 96-well microtiter plate (Nunc) The microtiter plate was not pre-coated The other half of the plate (without cells) was used as a control The wells in this half were incubated with growth medium in absence of cells, and the medium was changed in accordance with the exact same schedule as for the exchange of medium in the wells with cells

HK binding to ECM

At confluence, all wells including the cell-free wells were washed with NaCl⁄ Pi(50 mm phosphate, 0.1 m NaCl, pH 7.5), and incubated with 5 mm EDTA in NaCl⁄ Pi for

30 min at room temperature, as previously described [41] This treatment detached the cells from the surface of the wells as visualized microscopically and as analyzed by the absence of anti-actin binding components [ELISA using mouse-anti actin IgG (M-0635); Dako] The wells were next washed twice with Locke’s buffer and incubated for mini-mum of 30 min at room temperature with 200 lL per well

of 1% (w⁄ v) BSA or otherwise as noted LM deposited in the matrix during growth of the cells was detected using rabbit anti-LM (Z-0097; Dako) Binding of HK was deter-mined as described above

Statistical analysis

The results are shown as the mean ± SD and statistically

significant differences were estimated using analysis of

variance and the Bonferoni post-hoc test P < 0.05 was

considered statistically significant

Acknowledgements

The present study was supported by grant 2005-1-192

and 2007-01-0355 from the Carlsberg Foundation

References

1 Aumailley M & Gayraud B (1998) Structure and

bio-logical activity of the extracellular matrix J Mol Med

76, 253–265

2 Berrettini M, Schleef RR, Heeb MJ, Hopmeier P &

Griffin JH (1992) Assembly and expression of an

intrinsic factor IX activator complex on the surface of

cultured human endothelial cells J Biol Chem 267,

19833–19839

3 Colman RW & Schmaier AH (1997) Contact system: a

vascular biology modulator with anticoagulant,

profi-brinolytic, antiadhesive, and proinflammatory

attri-butes Blood 90, 3819–3843

4 Motta G, Rojkjaer R, Hasan AA, Cines DB &

Schmaier AH (1998) High molecular weight kininogen

regulates prekallikrein assembly and activation on

endo-thelial cells: a novel mechanism for contact activation

Blood 91, 516–528

5 Motta G, Shariat-Madar Z, Mahdi F, Sampaio CA &

Schmaier AH (2001) Assembly of high molecular weight

kininogen and activation of prekallikrein on cell matrix

Thromb Haemost 86, 840–847

6 Renne T, Nieswandt B & Gailani D (2006) The intrinsic

pathway of coagulation is essential for thrombus

stabil-ity in mice Blood Cells Mol Dis 36, 148–151

7 Iwaki T & Castellino FJ (2006) Plasma level of

bradyki-nin are suppressed in factor XII-deficient mice Thromb

Haemost 95, 1003–1010

8 Schousboe I, Nystrøm BT & Hansen GH (2008)

Differ-ential binding of factor XII and activated factor XII to

soluble and immobilized fibronectin Localization of the

Hep-1⁄ Fib-1 binding site for activated factor XII FEBS

J 275, 5161–5172

9 Joseph K, Ghebrehiwet B, Peerschke EI, Reid KB &

Kaplan AP (1996) Identification of the zinc-dependent

endothelial cell binding protein for high molecular

weight kininogen and factor XII: identity with the

receptor that binds to the globular ‘heads’ of C1q

(gC1q-R) Proc Natl Acad Sci USA 93, 8552–8557

10 Joseph K, Ghebrehiwet B & Kaplan AP (1999)

Cyto-keratin 1 and gC1qR mediate high molecular weight

kininogen binding to endothelial cells Clin Immunol 92, 246–255

11 Shariat-Madar Z, Mahdi F & Schmaier AH (1999) Mapping binding domains of kininogens on endothelial cell cytokeratin 1 J Biol Chem 274, 7137–7145

12 Hasan AA, Zisman T & Schmaier AH (1998) Identifica-tion of cytokeratin 1 as a binding protein and presenta-tion receptor for kininogens on endothelial cells Proc Natl Acad Sci USA 95, 3615–3620

13 Mahdi F, Shariat-Madar Z, Kuo A, Carinato M, Cines

DB & Schmaier AH (2004) Mapping the interaction between high molecular weight kininogen and the uroki-nase plasminogen activator receptor J Biol Chem 279, 16621–16628

14 Sun D & McCrae KR (2006) Endothelial-cell apoptosis induced by cleaved high-molecular-weight kininogen (HKa) is matrix dependent and requires the generation

of reactive oxygen species Blood 107, 4714–4720

15 Hallmann R, Horn N, Selg M, Wendler O, Pausch F & Sorokin LM (2005) Expression and function of laminins

in the embryonic and mature vasculature Physiol Rev

85, 979–1000

16 Grant DS & Kleinman HK (1997) Regulation of capil-lary formation by laminin and other components of the extracellular matrix EXS 79, 317–333

17 Belkin AM & Stepp MM (2000) Integrins as receptors for laminins Microsc Res Tech 51, 280–301

18 Kikkawa Y, Sanzen N & Sekiguchi K (1998) Isolation and characterization of laminin 10⁄ 11 secreated by human lung carcinoma cells Laminin 10⁄ 11 mediates cell adhesion through integrins alpha3beta1 J Biol Chem 273, 15854–15859

19 Madsen CD & Sidenius N (2008) The interaction between urokinase receptor and vitronectin in cell adhesion and signalling Eur J Cell Biol 87, 617–629

20 Wei Y, Waltz DA, Rao N, Drummond RJ, Rosenberg

S & Chapman HA (1994) Identification of the uroki-nase receptor as an adhesion receptor for vitronectin

J Biol Chem 269, 32380–32388

21 Kugler MC, Wei Y & Chapman HA (2003) Urokinase receptor and integrin interactions Curr Pharm Des 9, 1565–1574

22 Wei Y, Lukashev M, Simon DI, Bodary SC, Rosenberg

S, Doyle MV & Chapman HA (1996) Regulation of in-tegrin function by the urokinase receptor Science 273, 1551–1555

23 Zhang JC, Claffey K, Sakthivel R, Darzynkiewicz Z, Shaw DE, Leal J, Wang YC, Lu FM & McCrae KR (2000) Two-chain high molecular weight kininogen induces endothelial cell apoptosis and inhibits angiogen-esis: partial activity within domain 5 FASEB J 14, 2589–2600

24 Guo Y-L, Wang S, Cao DJ & Colman RW (2003) Apoptotic effect of cleaved high molecular weight