- Email: [email protected]
EFFECT OF DIVALENT METAL IONS ON THE BINDING OF TISSUE FACTOR AND ACTIVATED FACTOR VII
Thmmbais Research,Vol. 85, No. 4, pp. 327-339,1997 Copyright@ 1997 Elsevier SeierrceLtd Printedirrthe USA. ARrightsm..served 0049-3848/97 $17.CQ+ .00
Pergamon
PII S0049-3848(97)00018-2
EFFECT OF DIVALENT METAL IONS ON THE BINDING OF TISSUE FACTOR AND ACTIVATED FACTOR VII DeniseM. Headl, km T.W. Matthewsl and MichaelA. TonesLz 1 Ciba
ph~aceutic@
wi~bleh~~st
Road,
Horsh~,
west
Sussex,
RJ312
4AB,
and zPfi=r centralResearch,Sandwich,Kent,CT13 9NJ, UK
(Received 19 November 1996 by Editor FtJ. Gaffney; revised/accepted 13 January 1997)
Abstract Surface plasmon resonance technology was used to study the binding of activatedfactor VII to tissuefactor,in the presenceof divalentmetal ions. The binding was calcium-dependent,with the optimumcalcium concentrationat 5 mM. Only very minimal binding was observed in the absence of added calcium ions and presence of 10 mM EGTA. The effect of the calcium concentrationon the apparentdissociationconstantsfor the TF/VIIa complex formation was also investigated. In the absence of calcium ions but in the presence of either magnesiumor zinc ions, no significantbinding of activated factor VII to tissuefactor was detected. Howeversome binding was observed with manganeseions, indicatingthat manganeseions could partially replace calcium ions to supportthe formationof the TF/VHa complex. These results are consistent with studies of the effect of divalent metal ions on the amidolyticactivityof the TF/VIIacomplex. Copyright @ 1997 Elsevier Science Ltd
The initiation of the blood coagulationcascade occurs when factor VII, a vitamin-K dependentserine protease circulatingin the blood comes into contact with tissue factor (TF), its integral membrane-boundreceptor. Factor VII in complex with TF rapidly becomes activatedand the resultingcomplexof TF and activatedfactor VII (VIIa) has a Key words:FactorVII, Tissuefactor,calcium,divalentmetalions. Correspondingauthor:DeniseM. Head. CibaPharmaceuticals,WimblehurstRoad, Horsham,West Sussex,RH124AB, UK. Tel: +1403 272827; Fax: +1403 323307. E-Mailaddress:[email protected]
327
328
TF AND FVlla
vol. 85, No. 4
much increased catalytic activitytowardsthe natural substratesfactors IX and X. This results in the formation of activatedfactor X and eventuallyleads to the production of thrombininvolvedin the blood clot. The formationof the TF/VIIa complex is therefore a crucial step in this amplification process. The dissociation constant for complex formation has been reported within the picomolar to nanomolar range (1,2). The resolutions of the crystal structures of the extracellulardomain of TF, alone or in a complex with VIIa have revealed that the binding sites between TF and VIIa were not linear sequencesbut were constitutedfrom residues from discontinuousregions (3,4). Studies of the effects of substitutingdomainsof VII with the homologousregions from factor IX have shown a direct role of the first EGF domain and the catalytic domain of VIIa in bindingto TF (5). In additionthe Gla domainis believedto play a role indirectly by conferring a calcium-dependentconformationon factor VII which is required for interactionwith VIIa (6). Whether or not the bindingof TF and VIIa dependscritically on the presence of calcium ions has been a subject of debate. Gel electrophoresisand sedimentationvelocitystudiesalso show complexformationbetween TF and VIIa in the presenceof EDTA (7). SolubleTF is still capableof bindingto VIIa or Gla domain-less VIIa in the absence of added calcium and in the presence of excess divalent metal-ion chelator (8). Several high and low-affinitycalciumbinding sites have been reported on VIIa: TF can interactwith VIIa in the absenceof calcium,which howeverwould enhance this interaction 150,000fold (9). It appearsthat TF and VIIa can form complexesin the absenceof calciumor other divalentions, suggestingthe existenceof calcium-dependent and calcium-independentbinding interactionsbetween them. In contrast, it has been reported that binding of TF depends critically on the presence of calcium ions (10). Surface plasmon resonance(SPR) is a new technologyallowinga direct visualizationof the binding process of 2 molecules, with relatively easy calculation of apparent dissociationconstants(11). In this study we used SPR to investigatethe divalentcation dependenceof the bindingbetweenTF and VIIa: apparentrate and equilibriumconstants were obtainedand evidencefor metal-independentinteractionbetweenTF and VIIa. MATERIALAND METHODS Reagents
Human recombinant TF was obtained from American Diagnostic (USA). Human recombinantVIIa was purchasedfrom Novo-Nordisk(Sweden),in a lyophilisedform in 10 rnM glylcylglycine,50 rnM NaCl, 10 mM CaC12,30 mg/ml mannitol,0.01 YO Tween 80, pH 5.5 buffer. Biacore system, sensorchip CM5, amine couplingkit contaning Nhydroxysuccinimide(NHS), (N-ethyl-N’-(3-diethylamino-propyl-)carbodiimide (EDC) and ethanolarnine-hydrochloridewere obtained from Pharmacia Biosensor (Sweden). All otherreagentswere of analyticalgrade. Su#ace plasmon resonance measurements Coupling of human recombinant TF to the chip
Vol. 85,
No.
4
TF AND FVlla
The sensor chip was equilibrated with HEPES buffer (10 rnM HEPES, 150 mM NaCl, 0.1% triton X-1OO,pH 7.2) + 5 mM CaCl~, at a flow rate of 2 pl/min and then activated by injecting 20 @ of 0.2 M EDC + 0.05 M NHS. TF (66 pg/ml) was diluted 1;2 (v/v) in 0.1% triton X-1OOand then further diluted in 10 mM sodium acetate buffer at pH 4.0 to the requil-ed concentration. 40 W1of TF was injected , followed by 20 ~1 of I M ethanolamine-HCl pH &5 to deactivate the excess NHS-esters on the chip surface. The amount of TF coated corresponded to various resonance units, depending on the concentration of TF. Effect of cukium ions on the TFNlla binding Chips with different coating densities (357 to 2167 R[Js) were prepared by performing the coupling procedure with 1.8-5.5 vg/ml TF. VIIa (1 pM in H13PESbuffer + 5 mM CaC12) was diluted between 3 I and 250 nM in the same buffer, 10 U1 of VIIa was injected on the chip at a flow rate of 5 @/min. The binding of VIIa is followed during the association phase, after which its dissociation from TF occurs. The amount of binding is measured by the number of RU at the beginning of the dissociation phase. Dissociation in HEPES buffer was monitored for several minutes. The chip was regenerated by the injection of 5 pl of 10 mM glyciue pH 1.9. The apparent kinetic constants were calculated by linear or nonlinear regression analysis using software supplied by the manufacturer. Thereafter, for all the kinetic experiments, the chips were coated with less than 1000 RU of TF. The running buffer was made of HEPES buffer + calcium chloride concentrations ranging from 0.25 to 50 mM, 01-HEPES buffer alone or HEPES buffer+ 10 mM EGTA. VIIa (1 ~M) was diluted between 3 I und 250 nM in the relevant running buffer. The injections of VIIa and the regeneration were the same as described above. The apparent kinetic constants were calculated at each CaC12 concentration. Efiect of metal ions on the rTF/rVIIa binding VIIa (1 PM in water) was diluted between 31 and 250 nM in running buffer containing either magnesium, zinc or manganese chloride and then injected on a chip coated with TF. Concentrations of 10, 5 and 2.5 mM of divalent metal ions were used and the binding was expressed as a percentage of the binding in the presence of 5 mM CaC12.
RESULTS
Effect of calcium ions on the TFNIIa binding In the presence of buffer containing 5 rnM CaC12, the apparent kinetic constants (Kd) for the binding of VIIa to a TF coated chip remained essentially constant between 1.0 and 1.3 nM for amounts of TF immobilised up to 2000 RU (Table I). These values were indicative of a tight binding between TF and VIIa. Increasing the flow rate (from 5 to 20 plhnin) did not significantly change the Kd (data not shown). For subsequent kinetic experiments, in order to avoid mass transport limitation and rebinding of VIIa, a low ligand density (less than 1000 RU of TF) and a flow rate of 5 @/rein were used.
329
TF AND FVlla
330
Vol.85, No. 4
TABLE I Effect of the Amount of TF (RU) Coated on the Chip on the ApparentDissociationConstants(Kd) of VIIa Bindingto it. RU of TF coatedon chip 357 632 868 1051 1536 1855 2167
ApparentKd (nM) for TF/VIIabinding 1.23+/- 0.28 (n=4) 0.97 +/- 0.23 (n=3) 1.35+/- 0.09 (n=3) 0.90 (n=l) 1.30 (n=l) 1.33+/- 0.17 (n=3) 1.30 (n=2)
The running buffer contained5 mM CaC12. The results were the MEAN +/- SEM of n separateexperiments. As shown in Table II, the non-specificbinding was minimal (maximum 1.2 % of VIIa bound to a blank chip, when comparedto a TF chip). TABLE II Specificityof the TF/VIIaBinding. VIIa (nM) 31 62 125 250
% of non-specificbinding 1.2+/- 0.2 0.8 +/- 0.3 0.7 +/- 0.2 0.8 +/-0.4
VIIa (31, 62, 125 and 250 nM) was injectedfor 2 minutes,either on a TF coated chip (2167 to 3324 RU) or on a blank chip, in the HEPES buffer containing5 mM CaC12.The RU of VIIa bound to the blank chip were expressedas a percentageof the RU of VIIa bound to the TF chip. The resultswere the MEAN +/- SEM of 4 separateexperiments.
TF AND FVlla
Vol.85, No. 4
331
The calcium dependence of the TF/VIIa binding was studied by varying the CaC12 concentrationin the HEPES buffer from 50 to OmM, and also by adding 10 mM EGTA. To check whetherin the absenceof added CaC12the buffer containedsignificantcalcium ions, the calcium concentrationwas measuredby atomic absorption spectrophotometry (Figure 1). The running buffer without added calcium ions did not contain any measurable amounts of calcium (Table HI). However the rVIIa vial itself contained calcium (15.19 mM which was quite comparableto the theoreticalvalue of 16.66 mM). A fraction of VIIa put on a Chelex 100 column was found to be devoid of any measurableamountsof calcium and used specificallyfor kineticexperimentsin running buffer without added calcium +/- 10 mM EGTA. Figure 2 showed that the optimum CaC12concentrationwas 5 mM, as the amount of binding plateaued from this value onwardsand that 250 nM VIIa was approachingthe saturatinglevel on the TF chip.
1.00 g 0.75 ~ g
a
0.50 0.25 0.00 0
5
10
15
20
25
CaC12(mM) FIG. 1.
Standard curve of calcium concentrationsmeasured by atomic absorption spectrophotometry on a Pye Unicam SP9 AA spectrophotometer. The equipmentwas set up as followed: 5 cm burner, air/acetyleneflame, burner height = 4, lamp at 6 mAmp, wavelength= 422.7 nm. The detectionlimit was 0.002 absorbance units.
TF AND FVlla
Vol.85, No. 4
TABLE III Measurementsof CalciumConcentrationsby AtomicAbsorptionSpectrophotometry Sample HEPES buffer HEPES buffer+ 10mM EGTA rVIIa (2 @) rVIIa/Chelex100
Absorbance
Calcium(mM)
O.000
0
0.001
0
0.092 0.004
15.19 0
100
75
% 50
o~
0.01
I 0.1
1
10
100
CaCl , (mM)
FIG. 2. Effect of calciumchlorideon the TF/VIIabinding. VIIa was diluteddown to 31 (0) ,62 (0), 125(A)or 250 nM (V), in running buffer containingbetweenOand 50 mM CaC12and injectedfor 2 minuteson a chip coatedwith TF. Bindingwas plottedas a percentageof the RU boundfor 250 nM VIIa at 5 mM CaC12. The resultswere the MEAN +/- SEM of 3 separateexperiments.
333
TF AND FVlla
Vo[.85, No. 4
TABLE IV Effect of CalciumChlorideon the Apparent Kinetic Constantsof VIIa Bindingto a TF CoatedChip.
CaC12 (~) 50 10 5 2.5 1.0 0.5 0.25
Apparentka~~ (X105M-1 ~-i) 3.24 +/- 0.27 3.63 +/- 0.18 3.41 +/- 0.14 3.50 +/- 0.17 2.59 +1-0.07 1.86+/- 0.14 1.75+/-o.11
Apparentkdiss (X1O-4s-1) ‘7.33+/- 0.46 3.90 +1-0.86 3.70 +1-0.79 2.51+/-0.80 5.44 +/- 1.45 5.86 +1-0.47 7.04 +/- 3.29
CaC12 (mM) O(n=l) 0 + EGTA (n=l)
ApparentKd 1 (wM) 0.21 0.75
ApparentKd 2 (nM) 27.60 31.75
ApparentKd (~) 2.26 +1-0.25 1.07+/- 0.25 1.08+/- 0.19 0.72 +/- 0.23 2.10 +/- 0.58 3.15 +/- 0.16 4.02 +1-1.26
The results were the MEAN +/- SEM of 3 separate experiments, exceptwhen statedotherwise. There was a small amount of binding in the absence of added CaC12,especially at the higher VIIa concentrationswhich could be attributedto the amountof residual CaC12in the VIIa vial. Table IV summarisedthe effect of Ca2+ on the apparentkineticconstants. At sub-optimalconcentrationsof CaC12the apparentdissociationrate constants (kdiss) increasedand the apparentassociationrate constants(kms) decreased,resultingin higher apparentKd values. To determinethe apparentkineticconstantsin HEPES buffer +/- 10 mM EGTA, the VIIa treated on Chelex 100 was used at higher concentrations(0.125, 0.25, 0.5, 1, 1.5, 2 and 2.5 PM). Reducedand unreducedSDS gels confirmed that the Chelex treatment had not affected the integrity of the VIIa molecule, as there was no apparent difference with an untreated VIIa (data not shown). The association and dissociation phases were biphasic, leading to the obtention of 2 Kd values, in the micromolarand nanomolarrangesrespectively(TableIV).
Vol.85, No. 4
TF AND FVlla
334
4
T
v
O
I
I
I
I
2
4
6
8
10
MgC12(mM)
() I
I
ZnC12L(mM) FIG. 3. Effect of magnesiumchloride(A) or zinc chloride(B) on the TF/VIIabinding. VIIa was diluteddown to 31 (0), 62 (0), 125(A ) and 250 nM (V) in runningbuffer containingOto 10mM MgC12 (n=4) or ZnC12(n=3) and injectedon a TF coatedchip. For each VIIa concentration,the resultswere expressedas the percentageof bindingin the presenceof 5 mM CaC12.The resultswere the MEAN +/- SEM of n separateexperiments.
TF AND FVlla
vol. 85, No. 4
335
E#ect of other divalent metal ions on the TFNIIa binding
The binding of VIIa to the TF chip was small in the presence of Mg2+: the amount of binding for each VIIa concentrationwas not significantlydifferent from the binding in the absence of added Ca2+ (Figure 3A). In the presence of Zn2+, some non-specific binding was observed,possiblydue to the tendencyto precipitateof the running buffer containingZn2+ and thereforeall sampleswere run in parallel on a TF coated or blank chips. We observed that the additionof dextran (1 mg/ml) to this buffer could entirely reduce the non-specificbinding. Figure 3B was plotted using the corrected values (RU of binding on TF chip - RU of bindingon blank chip). Specificbinding was minimal in the absenceof Ca2+ and presenceof Zn2+ and was not significantlydifferent from the bindingin the absenceof addedCa2+. Theseresultsshowedthat neitherMg2+ nor Zn2+ were able to substitutefor Ca2+ to supportthe formation of the TF/VIIa complex. In contrast in the absence of added Ca2+ and in the presence of Mn2+, VIIa was able to bind to the TF chip (Figure4). Non-specificbindingto a blank chip was negligible(data not shown). At the optimumMnC12concentrationof 5 mM, the TF/VIIa binding was still markedlyless than in the presenceof 5 mM CaC12and could not be explainedby the calciumresidualin the sample. The apparentdissociationconstants(Kd) were one order of magnitude higher in the presence of Mn2+ than in the presence of 5 mM CaC12, mainly due to decreasedkass values and increasedkdiss values (Table V). Therefore it appeared that Mn2+ could only partially replace Ca2+ to support the formation of the TF/VIIa complexand the resultingcomplexwas not as stableand tight as in the presence of Ca2+.
TABLE V Apparent Kinetic Constants of the TF/VIIa Binding in the Presenceof MnC12or 5 mM CaC12. Divalentmetal ion 2.5 mM MnC12 5 mM MnC12 10mM MnC12 5 mM CaC19
Apparentka,s (X105M-15-]) 1.65+/- 0.30 1.86+/- 0.26 1.53+/- 0.13 5.16 +/- 1.17
Apparent iss ?, (X1O-4s26.15+1-2.84 22.63 +/- 0.13 20.53 +/- 1.04 4.97 +1-0.27
ApparentKd (nM) 15.84+/- 4.49 12.16+/- 1.43 13.41+/- 1.89 0.96 +/- 0.15
The resultswere the MEAN +/- SEM of 4 separateexperiments. DISCUSSION In the presenceof 5 mM CaC12,the apparentdissociationconstant(Kd) of VIIa binding to full-lengthTF was about 1 nM and agreedwell with otherpublishedSPR results:Kd
336
TF AND FVlla
I
I
0
5
Vol.85, No. 4
10
MnC12(mM) FIG. 4. Effect of manganesechloride on the TF/VIIa binding. VIIa was diluted down to 31 (0), 62 (0), 125 (A) and 250 nM (V) in runningbuffer containingOto 10rnMMnC12and injectedon a TF coated chip. For each VIIa concentration, the results were expressed as the percentageof binding in the presence of 5 rnM CaC12. The results were were the MEAN +/- SEM of 4 separate experiments. values of recombinanthuman VIIa bindingto a chip coated with soluble TF (1-219) of 5.47 nM and 2.3 nM respectively(2,12)or with solubleTF (1-218)of 3.0 nM or 3.2 UM respectively(13,14). The binding of VIIa to the TF coated chip was very specific and clearly calcium dependent. Though sub-optimumcalcium concentrationsadded to the buffer had an effect on the apparent kms and kdis~ values, the binding data fitted a simple 1:1model,resultingin higherapparentKd values. Somebindingwas observedin the absenceof calcium and presenceof 10 rnM EGTA, when rnicromolarconcentrations of VIIa were used. This supportedthe fact that a very low affinity interaction could occur between VIIa and TF in the absence of divalentmetal ions. The Kd values
vol. 85, No. 4
TF AND FVlla
337
reported so far for the TF/VIIa interactionin the absenceof calcium have been: 3.2 f.LM (8), 1.5 PM (9) or 3.9-34 nM (7). Interestinglythe 2 apparentKd values obtainedusing SPR in this study were in the micromolarand nanomolarranges. The binding data did not fit a simple 1:1 model any more and the associationand dissociationphases were biphasic. The reasonswhy the kineticswere more complex,could be that once VIIa has weakly bound to TF via a calcium-independent site, VIIa could undergo a confirmational change leading to a tighter binding and resulting in a lower Kd2 value. Alternatively, despite the fact that the calcium concentrationin the Chelex-VIIa was below the limit of detectionof the equipmentused, we could not rule out that a minimal number of calcium ions was still bound to the VIIa and both calcium-independentor dependentinteractionswere measured. However at physiologicalCa2+ concentration, the high affinityinteractionwould clearlybe the only relevantform of TF/VIIa complex. In the presence of added calcium ions, there might be stabilisationand consolidationof the TF/VIIa complexwhichrenderedany calcium-independentinteractionsnegligible. By SPR, no significantbinding of VIIa to TF in the absence of Ca2+ and presence of Mg2+ or Zn2+ ions was detectable,whereasMn2+ could support the TF/VIIa complex formationvery effectively. The TF/VIIa amidolyticactivitywas known to be influenced by divalent metal ions: Ca2+ had the greatest enhancing effect on substrate cleavage, Mn2+ a moderate effect and Mg2+ an even lower one; but the amidolyticactivity was still higher than in experimentswith EDTA (7). In this study, the same ranking order was found for the effect of these ions on the TF/VIIa binding. Ca2+ could induce the exposure of hydrophobicside chains of Gla-containingcoagulationproteins which was crucial for membrane binding (15) and could directly mediate protein-proteincontacts (16). Mn2+ referred to as a “mimic/antagonistof Ca2+” could compete with Ca2+ for binding sites and had an antithromboticeffect (17). Ca2+ binding to the Gla-domainof VIIa resulted in at least 2 distinctstructuraltransitionswith different functionaleffects: stimulationof the amidolyticactivity and induction of expression of the phospholipid interactiveform (18). As Mg2+ was only able to replaceCa2+ for the first effect and not for the secondone, Mg2+ mightinfluencethe VIIa activesiteby bindingto the same site as Ca2+, but could not bind to a second Ca2+ binding site directly involved in phospholipidor TF binding. At physiologicalCa2+ concentrations,Zn2+ inhibit the amidolytic and proteolytic activity of the TF/VIIa complex (19). Although Mg2+ or Zn2+ could not substitutefor Ca2+ in the TF/VIIa binding, we could not exclude the possibilitythat they could bind to a Ca2+bindingsite involvedin the TF/VIIa interaction but without inducing the necessary confirmational change for binding to occur. Alternatively the inhibitory effect of Zn2+ on the TF/VIIa enzyme activity could be mediatedby anothersite. In bovine factor IX, Ca2+ and Mn2+ had 2 different binding sitesbut Ca2+ and Mn2+ competedfor the samebindingsite on bovineprothrombin. As Mn2+ could replace Ca2+ for the TF/VIIa binding, Ca2+ and Mn2+ might bind to the same site on VIIa. The observationthat the TF/VIIa complexformed in the presence of Mn2+ has a different Kd compared to the Ca2+-dependentcomplex suggests that the Mn2+-boundconfirmationalstateof VIIa is not identicalto the Ca2+-boundstate.
TF AND FVlla
338
vol. 85, No. 4
REFERENCES 1. KRISHNASWAMYS. The interactionof human factor VIIa with tissue factor. J. Biol. Chem 267,23696-23706, 1992. 2. O’BRIEN DP, KEMBALL-COOK G, HUTCHINSON AM, MARTIN DMA, JOHNSON DJD, BYFIELD PGH, TAKAMIYAO, TUDDENHAMEGD and McVEY JH. Surface plasmon resonancestudiesof the interactionbetween factor VII and tissue factor. Demonstrationof defective tissue factor binding in a variant FVII molecule (FVII-R79Q). Biochemistry33, 14162-14169,1994. 3. MULLER Y.A, ULTSCH H and de VOS A.M. The crystal structure of the extracellulardomainof humantissuefactorrefinedto 1.7A resolution. J. Mol. Biol 256, 144-159,1996. 4. BANNER D.W, D’ARCYA, CHENE C, WINKLERF.K, GUHA A, KOENIGSBERG W.H, NEMERSON Y and KIRCHHOFERD. The crystal structure of the complex of blood coagulationfactor VIIa with solubletissuefactor. Nature380,41-46, 1996. 5. CHANG J-Y, STAFFORD DW and STRAIGHT DL. The roles of factor VII’s structuraldomainsin tissuefactorbinding. Biochemistry34, 12227-12232,1995. 6. TOOMEY JR, SMITH KJ and STAFFORDDW. Localizationof the human tissue factor recognitiondeterminantof human factor VIIa. J. Biol. Chem 266, 19198-19202, 1991. BUTENAS S, LAWSON JH, KALAFATIS M and MANN KG. Cooperative interaction of divalent metal ions, substrate and tissue factor with factor VIIa. Biochemistry33,3449-3456, 1994.
7.
8. NEUENSCHWANDER PF and MORRISSEY JH. Roles of the membraneinteractiveregionsof factor VIIa and tissuefactor. J. Biol. Chem269, 8007-8013,1994. 9. SABHARWALAK, BIRKTO~ JJ, GORKA J, WILDGOOSE P, PETERSEN LC and BAJAJ SP. High affinityCa2+-bindingsite in the serineproteasedomain of human factor VIIa and its role in tissue factor binding and developmentof catalyticactivity. J. Biol. Chem 270, 15523-15530,1995. 10. PETERSEN LC, SCHI@DTJ, CHRISTENSENU. Involvementof the hydrophobic stack residues 39-44 of factor VIIa in tissue factor interactions. FEBS Lett 347, 73-79, 1994.
Vol. 85, No. 4
TF AND FVlla
339
11. KARLSSON R, MICHAELSSON A and MATTSON L. Kinetic anaIysis of monoclinal antiody-antigeninteractionswith a new biosensorbased analytical system. J. Immunol.Methods145,229-240, 1991. 12. KELLEY RF. Thermodynamicsof protein-proteininteractionstudiedby using BIAcore and single-sitemutagenesis. Methods:A Companionto Methodsin Enzymology6, 111-120,1994. 13. PERSSONE. Influenceof the ~-carboxyglutamicacid-richdomain and hydrophobicstack of factorVIIa on tissuefactorbinding. Haemostasis26,31-34, 1996.
14. PERSSONE, NIELSENL.S. Site-directedmutagenesisbut not ~-carboxylationof GIu-35in factor VIIa affectsthe association with tissuefactor. FEBS Letters 385,241243, 1996. 15. SUNNERHAGENM, FORS4NS, HOFFR6NA, DRAKENBERGT, TELEMAN O and STENFLOJ. Structureof the Ca2+-freeGLA domainshedslight on membrane bindingof blood coagulationproteins. Nature structuralbiology2,504-509, 1995. 16. RAO Z, HANDFORDP, MAYHEWM, KNOTTV, BROWNLEEGG and STUARTD. The structureof a Ca2+-bindingEpidermalGrowthFactor-likedomain:its role in protein-proteininteractions. Cell 82, 131-141,1995. 17. HARDY E, GLENN J, HEPTINSTALLS, RUBIN PC, HORN EH. Magnesium modifiesthe responsesof plateletsto inhibitoryagentswhich act via cAMP. Thromb Haemost74, 1132-1137,1995. 18. PERSSONE, PETERSENLC. Structurallyand functionallydistinctCa2+ binding sites in the y-carboxyglutamicacid-containingdomainof factor VIIa. Eur. J. Bioch 234, 293-300, 1995. 19. PEDERSEN AH, LUND-HANSEN T, KOMIYAMA Y, PETERSEN LC, OESTERGARDPB and KISIELW. Inhibitionof recombinanthuman blood coagulation factor VIIa amidolyticand proteolyticactivityby zinc ions. Thromb. Haemost 65, 528534, 1991.
Pergamon
PII S0049-3848(97)00018-2
EFFECT OF DIVALENT METAL IONS ON THE BINDING OF TISSUE FACTOR AND ACTIVATED FACTOR VII DeniseM. Headl, km T.W. Matthewsl and MichaelA. TonesLz 1 Ciba
ph~aceutic@
wi~bleh~~st
Road,
Horsh~,
west
Sussex,
RJ312
4AB,
and zPfi=r centralResearch,Sandwich,Kent,CT13 9NJ, UK
(Received 19 November 1996 by Editor FtJ. Gaffney; revised/accepted 13 January 1997)
Abstract Surface plasmon resonance technology was used to study the binding of activatedfactor VII to tissuefactor,in the presenceof divalentmetal ions. The binding was calcium-dependent,with the optimumcalcium concentrationat 5 mM. Only very minimal binding was observed in the absence of added calcium ions and presence of 10 mM EGTA. The effect of the calcium concentrationon the apparentdissociationconstantsfor the TF/VIIa complex formation was also investigated. In the absence of calcium ions but in the presence of either magnesiumor zinc ions, no significantbinding of activated factor VII to tissuefactor was detected. Howeversome binding was observed with manganeseions, indicatingthat manganeseions could partially replace calcium ions to supportthe formationof the TF/VHa complex. These results are consistent with studies of the effect of divalent metal ions on the amidolyticactivityof the TF/VIIacomplex. Copyright @ 1997 Elsevier Science Ltd
The initiation of the blood coagulationcascade occurs when factor VII, a vitamin-K dependentserine protease circulatingin the blood comes into contact with tissue factor (TF), its integral membrane-boundreceptor. Factor VII in complex with TF rapidly becomes activatedand the resultingcomplexof TF and activatedfactor VII (VIIa) has a Key words:FactorVII, Tissuefactor,calcium,divalentmetalions. Correspondingauthor:DeniseM. Head. CibaPharmaceuticals,WimblehurstRoad, Horsham,West Sussex,RH124AB, UK. Tel: +1403 272827; Fax: +1403 323307. E-Mailaddress:[email protected]
327
328
TF AND FVlla
vol. 85, No. 4
much increased catalytic activitytowardsthe natural substratesfactors IX and X. This results in the formation of activatedfactor X and eventuallyleads to the production of thrombininvolvedin the blood clot. The formationof the TF/VIIa complex is therefore a crucial step in this amplification process. The dissociation constant for complex formation has been reported within the picomolar to nanomolar range (1,2). The resolutions of the crystal structures of the extracellulardomain of TF, alone or in a complex with VIIa have revealed that the binding sites between TF and VIIa were not linear sequencesbut were constitutedfrom residues from discontinuousregions (3,4). Studies of the effects of substitutingdomainsof VII with the homologousregions from factor IX have shown a direct role of the first EGF domain and the catalytic domain of VIIa in bindingto TF (5). In additionthe Gla domainis believedto play a role indirectly by conferring a calcium-dependentconformationon factor VII which is required for interactionwith VIIa (6). Whether or not the bindingof TF and VIIa dependscritically on the presence of calcium ions has been a subject of debate. Gel electrophoresisand sedimentationvelocitystudiesalso show complexformationbetween TF and VIIa in the presenceof EDTA (7). SolubleTF is still capableof bindingto VIIa or Gla domain-less VIIa in the absence of added calcium and in the presence of excess divalent metal-ion chelator (8). Several high and low-affinitycalciumbinding sites have been reported on VIIa: TF can interactwith VIIa in the absenceof calcium,which howeverwould enhance this interaction 150,000fold (9). It appearsthat TF and VIIa can form complexesin the absenceof calciumor other divalentions, suggestingthe existenceof calcium-dependent and calcium-independentbinding interactionsbetween them. In contrast, it has been reported that binding of TF depends critically on the presence of calcium ions (10). Surface plasmon resonance(SPR) is a new technologyallowinga direct visualizationof the binding process of 2 molecules, with relatively easy calculation of apparent dissociationconstants(11). In this study we used SPR to investigatethe divalentcation dependenceof the bindingbetweenTF and VIIa: apparentrate and equilibriumconstants were obtainedand evidencefor metal-independentinteractionbetweenTF and VIIa. MATERIALAND METHODS Reagents
Human recombinant TF was obtained from American Diagnostic (USA). Human recombinantVIIa was purchasedfrom Novo-Nordisk(Sweden),in a lyophilisedform in 10 rnM glylcylglycine,50 rnM NaCl, 10 mM CaC12,30 mg/ml mannitol,0.01 YO Tween 80, pH 5.5 buffer. Biacore system, sensorchip CM5, amine couplingkit contaning Nhydroxysuccinimide(NHS), (N-ethyl-N’-(3-diethylamino-propyl-)carbodiimide (EDC) and ethanolarnine-hydrochloridewere obtained from Pharmacia Biosensor (Sweden). All otherreagentswere of analyticalgrade. Su#ace plasmon resonance measurements Coupling of human recombinant TF to the chip
Vol. 85,
No.
4
TF AND FVlla
The sensor chip was equilibrated with HEPES buffer (10 rnM HEPES, 150 mM NaCl, 0.1% triton X-1OO,pH 7.2) + 5 mM CaCl~, at a flow rate of 2 pl/min and then activated by injecting 20 @ of 0.2 M EDC + 0.05 M NHS. TF (66 pg/ml) was diluted 1;2 (v/v) in 0.1% triton X-1OOand then further diluted in 10 mM sodium acetate buffer at pH 4.0 to the requil-ed concentration. 40 W1of TF was injected , followed by 20 ~1 of I M ethanolamine-HCl pH &5 to deactivate the excess NHS-esters on the chip surface. The amount of TF coated corresponded to various resonance units, depending on the concentration of TF. Effect of cukium ions on the TFNlla binding Chips with different coating densities (357 to 2167 R[Js) were prepared by performing the coupling procedure with 1.8-5.5 vg/ml TF. VIIa (1 pM in H13PESbuffer + 5 mM CaC12) was diluted between 3 I and 250 nM in the same buffer, 10 U1 of VIIa was injected on the chip at a flow rate of 5 @/min. The binding of VIIa is followed during the association phase, after which its dissociation from TF occurs. The amount of binding is measured by the number of RU at the beginning of the dissociation phase. Dissociation in HEPES buffer was monitored for several minutes. The chip was regenerated by the injection of 5 pl of 10 mM glyciue pH 1.9. The apparent kinetic constants were calculated by linear or nonlinear regression analysis using software supplied by the manufacturer. Thereafter, for all the kinetic experiments, the chips were coated with less than 1000 RU of TF. The running buffer was made of HEPES buffer + calcium chloride concentrations ranging from 0.25 to 50 mM, 01-HEPES buffer alone or HEPES buffer+ 10 mM EGTA. VIIa (1 ~M) was diluted between 3 I und 250 nM in the relevant running buffer. The injections of VIIa and the regeneration were the same as described above. The apparent kinetic constants were calculated at each CaC12 concentration. Efiect of metal ions on the rTF/rVIIa binding VIIa (1 PM in water) was diluted between 31 and 250 nM in running buffer containing either magnesium, zinc or manganese chloride and then injected on a chip coated with TF. Concentrations of 10, 5 and 2.5 mM of divalent metal ions were used and the binding was expressed as a percentage of the binding in the presence of 5 mM CaC12.
RESULTS
Effect of calcium ions on the TFNIIa binding In the presence of buffer containing 5 rnM CaC12, the apparent kinetic constants (Kd) for the binding of VIIa to a TF coated chip remained essentially constant between 1.0 and 1.3 nM for amounts of TF immobilised up to 2000 RU (Table I). These values were indicative of a tight binding between TF and VIIa. Increasing the flow rate (from 5 to 20 plhnin) did not significantly change the Kd (data not shown). For subsequent kinetic experiments, in order to avoid mass transport limitation and rebinding of VIIa, a low ligand density (less than 1000 RU of TF) and a flow rate of 5 @/rein were used.
329
TF AND FVlla
330
Vol.85, No. 4
TABLE I Effect of the Amount of TF (RU) Coated on the Chip on the ApparentDissociationConstants(Kd) of VIIa Bindingto it. RU of TF coatedon chip 357 632 868 1051 1536 1855 2167
ApparentKd (nM) for TF/VIIabinding 1.23+/- 0.28 (n=4) 0.97 +/- 0.23 (n=3) 1.35+/- 0.09 (n=3) 0.90 (n=l) 1.30 (n=l) 1.33+/- 0.17 (n=3) 1.30 (n=2)
The running buffer contained5 mM CaC12. The results were the MEAN +/- SEM of n separateexperiments. As shown in Table II, the non-specificbinding was minimal (maximum 1.2 % of VIIa bound to a blank chip, when comparedto a TF chip). TABLE II Specificityof the TF/VIIaBinding. VIIa (nM) 31 62 125 250
% of non-specificbinding 1.2+/- 0.2 0.8 +/- 0.3 0.7 +/- 0.2 0.8 +/-0.4
VIIa (31, 62, 125 and 250 nM) was injectedfor 2 minutes,either on a TF coated chip (2167 to 3324 RU) or on a blank chip, in the HEPES buffer containing5 mM CaC12.The RU of VIIa bound to the blank chip were expressedas a percentageof the RU of VIIa bound to the TF chip. The resultswere the MEAN +/- SEM of 4 separateexperiments.
TF AND FVlla
Vol.85, No. 4
331
The calcium dependence of the TF/VIIa binding was studied by varying the CaC12 concentrationin the HEPES buffer from 50 to OmM, and also by adding 10 mM EGTA. To check whetherin the absenceof added CaC12the buffer containedsignificantcalcium ions, the calcium concentrationwas measuredby atomic absorption spectrophotometry (Figure 1). The running buffer without added calcium ions did not contain any measurable amounts of calcium (Table HI). However the rVIIa vial itself contained calcium (15.19 mM which was quite comparableto the theoreticalvalue of 16.66 mM). A fraction of VIIa put on a Chelex 100 column was found to be devoid of any measurableamountsof calcium and used specificallyfor kineticexperimentsin running buffer without added calcium +/- 10 mM EGTA. Figure 2 showed that the optimum CaC12concentrationwas 5 mM, as the amount of binding plateaued from this value onwardsand that 250 nM VIIa was approachingthe saturatinglevel on the TF chip.
1.00 g 0.75 ~ g
a
0.50 0.25 0.00 0
5
10
15
20
25
CaC12(mM) FIG. 1.
Standard curve of calcium concentrationsmeasured by atomic absorption spectrophotometry on a Pye Unicam SP9 AA spectrophotometer. The equipmentwas set up as followed: 5 cm burner, air/acetyleneflame, burner height = 4, lamp at 6 mAmp, wavelength= 422.7 nm. The detectionlimit was 0.002 absorbance units.
TF AND FVlla
Vol.85, No. 4
TABLE III Measurementsof CalciumConcentrationsby AtomicAbsorptionSpectrophotometry Sample HEPES buffer HEPES buffer+ 10mM EGTA rVIIa (2 @) rVIIa/Chelex100
Absorbance
Calcium(mM)
O.000
0
0.001
0
0.092 0.004
15.19 0
100
75
% 50
o~
0.01
I 0.1
1
10
100
CaCl , (mM)
FIG. 2. Effect of calciumchlorideon the TF/VIIabinding. VIIa was diluteddown to 31 (0) ,62 (0), 125(A)or 250 nM (V), in running buffer containingbetweenOand 50 mM CaC12and injectedfor 2 minuteson a chip coatedwith TF. Bindingwas plottedas a percentageof the RU boundfor 250 nM VIIa at 5 mM CaC12. The resultswere the MEAN +/- SEM of 3 separateexperiments.
333
TF AND FVlla
Vo[.85, No. 4
TABLE IV Effect of CalciumChlorideon the Apparent Kinetic Constantsof VIIa Bindingto a TF CoatedChip.
CaC12 (~) 50 10 5 2.5 1.0 0.5 0.25
Apparentka~~ (X105M-1 ~-i) 3.24 +/- 0.27 3.63 +/- 0.18 3.41 +/- 0.14 3.50 +/- 0.17 2.59 +1-0.07 1.86+/- 0.14 1.75+/-o.11
Apparentkdiss (X1O-4s-1) ‘7.33+/- 0.46 3.90 +1-0.86 3.70 +1-0.79 2.51+/-0.80 5.44 +/- 1.45 5.86 +1-0.47 7.04 +/- 3.29
CaC12 (mM) O(n=l) 0 + EGTA (n=l)
ApparentKd 1 (wM) 0.21 0.75
ApparentKd 2 (nM) 27.60 31.75
ApparentKd (~) 2.26 +1-0.25 1.07+/- 0.25 1.08+/- 0.19 0.72 +/- 0.23 2.10 +/- 0.58 3.15 +/- 0.16 4.02 +1-1.26
The results were the MEAN +/- SEM of 3 separate experiments, exceptwhen statedotherwise. There was a small amount of binding in the absence of added CaC12,especially at the higher VIIa concentrationswhich could be attributedto the amountof residual CaC12in the VIIa vial. Table IV summarisedthe effect of Ca2+ on the apparentkineticconstants. At sub-optimalconcentrationsof CaC12the apparentdissociationrate constants (kdiss) increasedand the apparentassociationrate constants(kms) decreased,resultingin higher apparentKd values. To determinethe apparentkineticconstantsin HEPES buffer +/- 10 mM EGTA, the VIIa treated on Chelex 100 was used at higher concentrations(0.125, 0.25, 0.5, 1, 1.5, 2 and 2.5 PM). Reducedand unreducedSDS gels confirmed that the Chelex treatment had not affected the integrity of the VIIa molecule, as there was no apparent difference with an untreated VIIa (data not shown). The association and dissociation phases were biphasic, leading to the obtention of 2 Kd values, in the micromolarand nanomolarrangesrespectively(TableIV).
Vol.85, No. 4
TF AND FVlla
334
4
T
v
O
I
I
I
I
2
4
6
8
10
MgC12(mM)
() I
I
ZnC12L(mM) FIG. 3. Effect of magnesiumchloride(A) or zinc chloride(B) on the TF/VIIabinding. VIIa was diluteddown to 31 (0), 62 (0), 125(A ) and 250 nM (V) in runningbuffer containingOto 10mM MgC12 (n=4) or ZnC12(n=3) and injectedon a TF coatedchip. For each VIIa concentration,the resultswere expressedas the percentageof bindingin the presenceof 5 mM CaC12.The resultswere the MEAN +/- SEM of n separateexperiments.
TF AND FVlla
vol. 85, No. 4
335
E#ect of other divalent metal ions on the TFNIIa binding
The binding of VIIa to the TF chip was small in the presence of Mg2+: the amount of binding for each VIIa concentrationwas not significantlydifferent from the binding in the absence of added Ca2+ (Figure 3A). In the presence of Zn2+, some non-specific binding was observed,possiblydue to the tendencyto precipitateof the running buffer containingZn2+ and thereforeall sampleswere run in parallel on a TF coated or blank chips. We observed that the additionof dextran (1 mg/ml) to this buffer could entirely reduce the non-specificbinding. Figure 3B was plotted using the corrected values (RU of binding on TF chip - RU of bindingon blank chip). Specificbinding was minimal in the absenceof Ca2+ and presenceof Zn2+ and was not significantlydifferent from the bindingin the absenceof addedCa2+. Theseresultsshowedthat neitherMg2+ nor Zn2+ were able to substitutefor Ca2+ to supportthe formation of the TF/VIIa complex. In contrast in the absence of added Ca2+ and in the presence of Mn2+, VIIa was able to bind to the TF chip (Figure4). Non-specificbindingto a blank chip was negligible(data not shown). At the optimumMnC12concentrationof 5 mM, the TF/VIIa binding was still markedlyless than in the presenceof 5 mM CaC12and could not be explainedby the calciumresidualin the sample. The apparentdissociationconstants(Kd) were one order of magnitude higher in the presence of Mn2+ than in the presence of 5 mM CaC12, mainly due to decreasedkass values and increasedkdiss values (Table V). Therefore it appeared that Mn2+ could only partially replace Ca2+ to support the formation of the TF/VIIa complexand the resultingcomplexwas not as stableand tight as in the presence of Ca2+.
TABLE V Apparent Kinetic Constants of the TF/VIIa Binding in the Presenceof MnC12or 5 mM CaC12. Divalentmetal ion 2.5 mM MnC12 5 mM MnC12 10mM MnC12 5 mM CaC19
Apparentka,s (X105M-15-]) 1.65+/- 0.30 1.86+/- 0.26 1.53+/- 0.13 5.16 +/- 1.17
Apparent iss ?, (X1O-4s26.15+1-2.84 22.63 +/- 0.13 20.53 +/- 1.04 4.97 +1-0.27
ApparentKd (nM) 15.84+/- 4.49 12.16+/- 1.43 13.41+/- 1.89 0.96 +/- 0.15
The resultswere the MEAN +/- SEM of 4 separateexperiments. DISCUSSION In the presenceof 5 mM CaC12,the apparentdissociationconstant(Kd) of VIIa binding to full-lengthTF was about 1 nM and agreedwell with otherpublishedSPR results:Kd
336
TF AND FVlla
I
I
0
5
Vol.85, No. 4
10
MnC12(mM) FIG. 4. Effect of manganesechloride on the TF/VIIa binding. VIIa was diluted down to 31 (0), 62 (0), 125 (A) and 250 nM (V) in runningbuffer containingOto 10rnMMnC12and injectedon a TF coated chip. For each VIIa concentration, the results were expressed as the percentageof binding in the presence of 5 rnM CaC12. The results were were the MEAN +/- SEM of 4 separate experiments. values of recombinanthuman VIIa bindingto a chip coated with soluble TF (1-219) of 5.47 nM and 2.3 nM respectively(2,12)or with solubleTF (1-218)of 3.0 nM or 3.2 UM respectively(13,14). The binding of VIIa to the TF coated chip was very specific and clearly calcium dependent. Though sub-optimumcalcium concentrationsadded to the buffer had an effect on the apparent kms and kdis~ values, the binding data fitted a simple 1:1model,resultingin higherapparentKd values. Somebindingwas observedin the absenceof calcium and presenceof 10 rnM EGTA, when rnicromolarconcentrations of VIIa were used. This supportedthe fact that a very low affinity interaction could occur between VIIa and TF in the absence of divalentmetal ions. The Kd values
vol. 85, No. 4
TF AND FVlla
337
reported so far for the TF/VIIa interactionin the absenceof calcium have been: 3.2 f.LM (8), 1.5 PM (9) or 3.9-34 nM (7). Interestinglythe 2 apparentKd values obtainedusing SPR in this study were in the micromolarand nanomolarranges. The binding data did not fit a simple 1:1 model any more and the associationand dissociationphases were biphasic. The reasonswhy the kineticswere more complex,could be that once VIIa has weakly bound to TF via a calcium-independent site, VIIa could undergo a confirmational change leading to a tighter binding and resulting in a lower Kd2 value. Alternatively, despite the fact that the calcium concentrationin the Chelex-VIIa was below the limit of detectionof the equipmentused, we could not rule out that a minimal number of calcium ions was still bound to the VIIa and both calcium-independentor dependentinteractionswere measured. However at physiologicalCa2+ concentration, the high affinityinteractionwould clearlybe the only relevantform of TF/VIIa complex. In the presence of added calcium ions, there might be stabilisationand consolidationof the TF/VIIa complexwhichrenderedany calcium-independentinteractionsnegligible. By SPR, no significantbinding of VIIa to TF in the absence of Ca2+ and presence of Mg2+ or Zn2+ ions was detectable,whereasMn2+ could support the TF/VIIa complex formationvery effectively. The TF/VIIa amidolyticactivitywas known to be influenced by divalent metal ions: Ca2+ had the greatest enhancing effect on substrate cleavage, Mn2+ a moderate effect and Mg2+ an even lower one; but the amidolyticactivity was still higher than in experimentswith EDTA (7). In this study, the same ranking order was found for the effect of these ions on the TF/VIIa binding. Ca2+ could induce the exposure of hydrophobicside chains of Gla-containingcoagulationproteins which was crucial for membrane binding (15) and could directly mediate protein-proteincontacts (16). Mn2+ referred to as a “mimic/antagonistof Ca2+” could compete with Ca2+ for binding sites and had an antithromboticeffect (17). Ca2+ binding to the Gla-domainof VIIa resulted in at least 2 distinctstructuraltransitionswith different functionaleffects: stimulationof the amidolyticactivity and induction of expression of the phospholipid interactiveform (18). As Mg2+ was only able to replaceCa2+ for the first effect and not for the secondone, Mg2+ mightinfluencethe VIIa activesiteby bindingto the same site as Ca2+, but could not bind to a second Ca2+ binding site directly involved in phospholipidor TF binding. At physiologicalCa2+ concentrations,Zn2+ inhibit the amidolytic and proteolytic activity of the TF/VIIa complex (19). Although Mg2+ or Zn2+ could not substitutefor Ca2+ in the TF/VIIa binding, we could not exclude the possibilitythat they could bind to a Ca2+bindingsite involvedin the TF/VIIa interaction but without inducing the necessary confirmational change for binding to occur. Alternatively the inhibitory effect of Zn2+ on the TF/VIIa enzyme activity could be mediatedby anothersite. In bovine factor IX, Ca2+ and Mn2+ had 2 different binding sitesbut Ca2+ and Mn2+ competedfor the samebindingsite on bovineprothrombin. As Mn2+ could replace Ca2+ for the TF/VIIa binding, Ca2+ and Mn2+ might bind to the same site on VIIa. The observationthat the TF/VIIa complexformed in the presence of Mn2+ has a different Kd compared to the Ca2+-dependentcomplex suggests that the Mn2+-boundconfirmationalstateof VIIa is not identicalto the Ca2+-boundstate.
TF AND FVlla
338
vol. 85, No. 4
REFERENCES 1. KRISHNASWAMYS. The interactionof human factor VIIa with tissue factor. J. Biol. Chem 267,23696-23706, 1992. 2. O’BRIEN DP, KEMBALL-COOK G, HUTCHINSON AM, MARTIN DMA, JOHNSON DJD, BYFIELD PGH, TAKAMIYAO, TUDDENHAMEGD and McVEY JH. Surface plasmon resonancestudiesof the interactionbetween factor VII and tissue factor. Demonstrationof defective tissue factor binding in a variant FVII molecule (FVII-R79Q). Biochemistry33, 14162-14169,1994. 3. MULLER Y.A, ULTSCH H and de VOS A.M. The crystal structure of the extracellulardomainof humantissuefactorrefinedto 1.7A resolution. J. Mol. Biol 256, 144-159,1996. 4. BANNER D.W, D’ARCYA, CHENE C, WINKLERF.K, GUHA A, KOENIGSBERG W.H, NEMERSON Y and KIRCHHOFERD. The crystal structure of the complex of blood coagulationfactor VIIa with solubletissuefactor. Nature380,41-46, 1996. 5. CHANG J-Y, STAFFORD DW and STRAIGHT DL. The roles of factor VII’s structuraldomainsin tissuefactorbinding. Biochemistry34, 12227-12232,1995. 6. TOOMEY JR, SMITH KJ and STAFFORDDW. Localizationof the human tissue factor recognitiondeterminantof human factor VIIa. J. Biol. Chem 266, 19198-19202, 1991. BUTENAS S, LAWSON JH, KALAFATIS M and MANN KG. Cooperative interaction of divalent metal ions, substrate and tissue factor with factor VIIa. Biochemistry33,3449-3456, 1994.
7.
8. NEUENSCHWANDER PF and MORRISSEY JH. Roles of the membraneinteractiveregionsof factor VIIa and tissuefactor. J. Biol. Chem269, 8007-8013,1994. 9. SABHARWALAK, BIRKTO~ JJ, GORKA J, WILDGOOSE P, PETERSEN LC and BAJAJ SP. High affinityCa2+-bindingsite in the serineproteasedomain of human factor VIIa and its role in tissue factor binding and developmentof catalyticactivity. J. Biol. Chem 270, 15523-15530,1995. 10. PETERSEN LC, SCHI@DTJ, CHRISTENSENU. Involvementof the hydrophobic stack residues 39-44 of factor VIIa in tissue factor interactions. FEBS Lett 347, 73-79, 1994.
Vol. 85, No. 4
TF AND FVlla
339
11. KARLSSON R, MICHAELSSON A and MATTSON L. Kinetic anaIysis of monoclinal antiody-antigeninteractionswith a new biosensorbased analytical system. J. Immunol.Methods145,229-240, 1991. 12. KELLEY RF. Thermodynamicsof protein-proteininteractionstudiedby using BIAcore and single-sitemutagenesis. Methods:A Companionto Methodsin Enzymology6, 111-120,1994. 13. PERSSONE. Influenceof the ~-carboxyglutamicacid-richdomain and hydrophobicstack of factorVIIa on tissuefactorbinding. Haemostasis26,31-34, 1996.
14. PERSSONE, NIELSENL.S. Site-directedmutagenesisbut not ~-carboxylationof GIu-35in factor VIIa affectsthe association with tissuefactor. FEBS Letters 385,241243, 1996. 15. SUNNERHAGENM, FORS4NS, HOFFR6NA, DRAKENBERGT, TELEMAN O and STENFLOJ. Structureof the Ca2+-freeGLA domainshedslight on membrane bindingof blood coagulationproteins. Nature structuralbiology2,504-509, 1995. 16. RAO Z, HANDFORDP, MAYHEWM, KNOTTV, BROWNLEEGG and STUARTD. The structureof a Ca2+-bindingEpidermalGrowthFactor-likedomain:its role in protein-proteininteractions. Cell 82, 131-141,1995. 17. HARDY E, GLENN J, HEPTINSTALLS, RUBIN PC, HORN EH. Magnesium modifiesthe responsesof plateletsto inhibitoryagentswhich act via cAMP. Thromb Haemost74, 1132-1137,1995. 18. PERSSONE, PETERSENLC. Structurallyand functionallydistinctCa2+ binding sites in the y-carboxyglutamicacid-containingdomainof factor VIIa. Eur. J. Bioch 234, 293-300, 1995. 19. PEDERSEN AH, LUND-HANSEN T, KOMIYAMA Y, PETERSEN LC, OESTERGARDPB and KISIELW. Inhibitionof recombinanthuman blood coagulation factor VIIa amidolyticand proteolyticactivityby zinc ions. Thromb. Haemost 65, 528534, 1991.