Binding of heparin to human antithrombin III as studied by measurements of tryptophan fluorescence

Biochimica et Biophysica Acta, 490 (1977) 104-111

© Elsevier/North-Holland Biomedical Press BBA 37550 B I N D I N G OF H E P A R I N TO H U M A N A N T I T H R O M B I N 111 AS S T U D I E D BY M E A S U R E M E N T S OF T R Y P T O P H A N F L U O R E S C E N C E

ROLAND EINARSSON and LARS-OLOV ANDERSSON A B Kabi, Research Department, Fack, S-104 25 StockhohTz (Sweden)

(Received June 16th, 1976)

SUMMARY Corrected fluorescence excitation and emission spectra of human antithrombin Ill have been determined. The fluorescence observed originates almost entirely from tryptophan residues. Reduction of the disulfide bonds followed by carboxymethylation did not change the fluorometric properties of the protein. The binding of heparin to antithrombin !II caused a marked fluorescence enhancement by about 30~o of the intrinsic protein emission intensity. Various samples of heparin yielded different binding curves. Heparin fractionated by gel filtration seemed to be bound to two sites on antithrombin III with association constants of 0.6.106 M -1 and 0.2.106 M -1 respectively. Heparin, prepared by affinity chromatography on matrix-bound antithrombin ill appeared to be bound to only one site with an association constant of 2.3.106 M - L Under similar conditions heparin caused no increase of the intrinsic protein emission intensity when added to reduced and carboxymethylated antithrombin 111. The implications of these findings are discussed.

INTRODUCTION Antithrombin Ill is the main coagulation inhibitor in blood and is of great importance for the regulation of the haemostatic balance. It is identical to the heparin cofactor [1-3], and the well-known strong anticoagulant effect of heparin is mediated through this factor. Antithrombin ill, or heparin cofactor, inhibits thrombin and the activated forms of the coagulation factors IX, X and X1 [2, 4, 5]. in the absence of heparin the process of inhibition is slow, and therefore antithrombin III has been classified as a progressively acting inhibitor. In the pre~ence of heparin the inhibition is very rapid, usually taking just a few seconds for completion. The mechanism whereby heparin affects the rate of the inhibition reaction is not known, but several suggestions have been made [2, 6], It is usually assumed [2] that heparin is bound to antithrombin Ill and alters its conformation so as to make it a more rapidly reacting inhibitor. However, it has also been suggested that heparin binding would occur to thrombin [6]. Antithrombin Ill is a single-chain glycoprotein with a molecular weight of 65 000 [2, 3]. It has been purified by various methods [1-3], including affinity chro-

105 matography on heparin-agarose [3]. No direct measurements of the binding of heparin to antithrombin III have been performed previously, but the effect of heparin on the reactivity of antithrombin III and its ready adsorption to matrix-bound heparin [3] demonstrates that binding does occur. In this investigation the fluorescence properties of antithrombin III and its complex with heparin have been studied. The binding of heparin to antithrombin III induces a marked increase in the fluorescence emission spectrum and this effect has been utilized to determine the stoichiometry of the reaction and the binding constants. MATERIALS AND METHODS Chemicals: All chemicals used were of reagent grade. Protein: Human antithrombin III was prepared according to a previously described method [3]. The preparation was homogeneous as judged by disc gel electrophoresis and immunoelectrophoresis. A fresh stock protein solution was prepared prior to use. The protein concentration was determined by absorbance measurements, the absorptivity being "al°/o 6.1 [3], or by the protein detenni1280 n m : nation method developed by Lowry using human serum albumin as a standard [7]. The antithrombin III antiserum used was obtained from Behringwerke AG, Marburg, G.F.R. Reduced and alkylated antithrombin III: To an 0.5 per cent antithrombin III solution in 0.05 M Tris/chloride buffer was added dithiotreitol (Sigma) to a concentration of 10 mM and the pH of the solution was adjusted to 8.0. The solution was allowed to stand at room temperature in the dark for two hours followed by alkylation with excess iodoacetamide (Sigma). The excess of reduction and alkylation reagents was removed by gel filtration on Sephadex G-25 (Pharmacia, Sweden). The content of the carboxymethyl-cysteine of the reduced and alkylated antithrombin llI was determined by amino acid analysis using a Durrum D-500 amino acid analyzer. Heparin: The heparin used was of three different types. One type of heparin (pig intestine) was obtained as a gift from Drs. I. BjSrk and M. H86k, Uppsala, and had a molecular weight of 15 000. The specific activity of this preparation was 240 units/rag. Another type of heparin was lyophilized commercial heparin (Vitrum, Sweden) from pig intestine (mucosa). It had a specific activity of 138 units/mg. The heterogeneity of the commercial material made it unsuitable for direct use in fluorometric binding studies. Therefore, prior to use this material was fractionated on Sephadex G-200 (Pharmacia, Sweden) with 0.2 M ammonium bicarbonate buffer as an eluant. The top fraction was freeze-dried and this pooled fraction appeared to be homogeneous on ultracentrifugation and had a molecular weight of 10 700 [8]. The specific activity of the gel-filtrated material was 178 units/mg. The third heparin sample was obtained by affinity chromatography (Johnson, E. A., Holmer, E., Barrowcliff, T. and Andersson, L.-O., unpublished) of the commercial heparin (Vitrum, Sweden) on matrix-bound antithrombin III. It had an average molecular weight of 11 200 and a specific activity of 260 units/mg. Fluorometric measurements Fluorescence measurements were performed with an Aminco Bowman spectrophotofluorometer. The excitation and emission spectra were recorded with an Aminco

106 Bowman SPF equipped with a corrected spectra accessory. The excitation light came from a high-pressure xenon lamp. All experiments were carried out at 25 °C in 0.05 M Tris/chloride buffer, pH 7.4, containing 0.2 M glycine and 0.03 M sodium chloride. To minimize photochemical damage during titration, the protein was exposed to light only when the fluorescence was being measured. In a typical experiment the antithrombin solution was contained in a total volume of 2.0 ml buffer and the change of emission intensity at 333 nm by adding heparin was followed. The relative fluorescence intensity of antithrombin III saturated with heparin (F~) was extrapolated from a doublereciprocal plot using the experimental data by plotting [F0 - - F ] - 1 versus [heparin]- 1 where F is the observed fluorescence intensity of the antithrombin ill. heparin complex and F0 is the emission intensity of antithrombin Ill alone. RES U LTS

Protein fluorescence The use of fluorescence spectroscopy to study the intrinsic fluorescence of proteins and the binding of various ligands to protein is welt established. Since proteins vary in their tyrosine and tryptophan content, they would be expected to vary in both excitation and emission spectra. Accordingly, the fluorescence intensity varies markedly from protein to protein and also with different experimental conditions. Accessible tryptophan groups in antithrombin Ill were determined with reduced and alkylated protein as starting material. This material contained six equivalents of carboxymethyl-cysteine and no cysteine as determined by amino acid analysis. Two different techniques were applied for the tryptophan determination. One was the simplified colorimetric procedure by Karkhanis et al. [9], which makes use of the tryptophan reagent 2-hydroxy-5-nitrobenzyl bromide. This technique indicated 5.5 tryptophan groups in antithrombin Ill. The other technique was amino acid analysis after alkaline hydrolysis of the alkylated antithrombin IlI [10]. The amino acid determination of tryptophan groups revealed six residues. Evidently, both methods indicated an average value of six tryptophan groups in antithrombin IIl, which is in agreement with the value reported by Kurachi et al. [11 ]. On excitation of antithrombin llI at different wavelenghts in the 270-305 nm range the wavelength of the emission maximum of the protein was essentially unchanged, indicating that the contribution of tyrosine residues to the emission spectrum was of minor significance. However, the emission intensity varied appreciably depending on the selected excitation wavelength. The excitation maximum is located around 280 nm (uncorrected 285 nm), which coincides rather well with the absorption maximum of the indole chromophores. Illumination of antithrombin II! at the excitation maximum (280 nm) produced an emission spectrum with a maximum intensity around 333 nm (uncorrected 341 rim). In general, various proteins show different protein fluorescence maxima, and this can be explained in terms of different ratios of two classes of tryptophan groups. One class includes tryptophan residues inside the protein in a low-polar hydrophobic microenvironment, the other class includes tryptophans on the surface of the protein in a high-polar aqueous microenvironment. Antithrombin III, which has its spectral

107 maximum between the two discrete model values, appears to have contributions from both classes, even if the contribution from buried tryptophans appears to preponderate. The corrected excitation and emission spectra of reduced and alkylated antithrombin III have an appearance similar to that of the native protein, suggesting that the microenvironment of tryptophan groups is not drastically altered as a result of the reduction and carboxymethylation of the disulfide bonds. The addition of fractionated heparin to a solution of native antithrombin III caused a distinct increase in the fluorescence intensity of the protein emission. At the same time a shift in the wavelength of the fluorescence emission maximum towards a shorter wavelength of about 3 nm was observed, indicating that at least some of the tryptophan residues are in a more hydrophobic environment in the protein-ligand complex compared to the native free protein. In general, the fluorescence maximum of a protein indole group shifts to a shorter wavelength as its environment becomes less polar [12]. Consequently, the effect of heparin on antithrombin III suggests a conformational change of the protein molecule [2] and/or a change of the polarity of the local microenvironment around tryptophan groups. The titration of antithrombin III with heparin produced an increase in the fluorescence intensity of the protein, ligand complex until the antithrombin III molecule appeared to be saturated. The magnitude of this increase in fluorescence of the native antithrombin III was about 30 per cent. In contrast, the addition of heparin to reduced and S-carboxymethylated cysteine residues in antithrombin III produced no change in the intrinsic fluorescence property of the protein. This suggests that heparin does not bind to the reduced and carboxymethylated antithrombin III. Alternatively, the binding of heparin may occur at quite different sites without inducing an alteration of the microenvironment around the tryptophyl groups. This latter assumption does not appear to be very likely. however.

Binding of heparin to antithrombin III The increase in intrinsic fluorescence of antithrombin III was used to monitor the binding of heparin and to evaluate the ligand binding data. Preliminary experiments with the unfractionated commercial heparin show that the heterogeneity with respect to the molecular weight of this preparation made it impossible to obtain relevant binding data. Therefore, prior to use this material was fractionated by gel filtration as described under Methods. The fraction obtained was studied with respect to its binding to antithrombin 1II. Fig. la shows the Scatchard [13] plot for the fluorometric titration. The amount of free and bound heparin was calculated from the ratio ( F - - F o)/(Fo~--F) [14], where Fo, F and Foo are, respectively, the fluorescence intensities of antithrombin III alone, of the protein in the presence of heparin and, finally, of antithrombin III saturated with heparin. The shape of the Scatchard plot indicates the presence of two binding sites with different affinities, one having an association constant of 0.6.106 M -1 and the other 0.2. l 0 6 M -1. These values are based on the assumption that the increase in tryptophan fluorescence on binding of heparin to antithrombin III is the same for both sites. This may not be entirely true, and therefore the values for the association constants must be regarded as somewhat

108 ~I0

C

~

S

10-

•

®

10S

C

i

i

q

~ - , 106

20

@

" 015

1.0

9

Fig. 1. Scatchard plot of heparin binding to antithrombin 111 at 25 C in 0.05 M Tris/chloride buffer, pH 7.4, containing 0.2 M glycine and 0.03 M sodium chloride. The titration data was obtained with excitation wavelength at 285 nm and emission wavelength at 341 nm. (a) The protein concentration used was 1.1/~M. The heparin used had a molecular weight of 10 700. (b) The protein concentration was 3.9/~M. The heparin had a molecular weight of 15 000. (c) The protein concentration was 2.3 f~M. The heparin had a molecular weight of 11 200. uncertain. The Scatchard plot (Fig. la) shows a significant bend which was quite reproducible when p e r f o r m i n g several titrations of a n t i t h r o m b i n I l l with the gel filtered commercial heparin. This effect can be interpreted as being due to heterogeneity in the heparin preparation. A l t h o u g h h o m o g e n e o u s with respect to molecular weight, there may be heparin molecules with different chemical structures present in this gel-filtered sample. T i t r a t i o n of a n t i t h r o m b i n III with the heparin which had a molecular weight of 15 000 yielded a slightly different b i n d i n g curve. However, even in this case two b i n d i n g sites were indicated as can be seen from the Scatchard plot (Fig, l b). The association constants o b t a i n e d were 0.7.106 M -1 a n d 0.1 • 106 M -1,

109 respectively. In contrast to the heparin sample with a molecular weight of 10 700, this sample yields a binding curve with no deviations from the linearity, which may be indicative of a more homogeneous preparation. The third sample of heparin studied was prepared by affinity chromatography on matrix-bound antithrombin III using the commercial heparin as starting material. This type of fractionation results in highly efficient concentration of the heparin molecules having the highest affinity for antithrombin III. Fluorometric titration of antithrombin III with this heparin sample yielded the Scatchard plot shown in Fig. lc. In this case only one ligand binding site is seen and, moreover, the association constant is considerably higher: 2.3.106 M - 1. DISCUSSION The fluorometric technique can often be a valuable tool for the spectroscopic characterization of proteins and for measuring the binding of ligands to proteins. In certain cases this technique can produce information not obtainable by other techniques. In proteins that contain more than one tryptophan residue the fluorescence from each of the tryptophyls is probably not identical. Apparently, in many proteins studied, the buried and exposed tryptophans contribute in various degrees to the total emission intensity. According to data in the literature, the emission maxima of tryptophan fluorescence in native proteins have been observed to range from 332 nm to 342 nm [12]. However, these values are mostly uncorrected for energy variations in the light source and the photomultiplier. Unfolding of proteins with denaturants such as urea usually shifts the tryptophan emission to about 350 nm [12, 15, 16]. Reduction and alkylation of antithrombin 111 did not affect the excitation or emission spectra. The unfolding caused by the breaking of the disulfide bonds must therefore be very limited or localized in a part of the molecule, where there are no tryptophan residues. Cowgill [17] has shown that disulfides quench both indole and tyrosine fluorescence if the disulfide comes into contact with the chromophore. The binding of heparin to antithrombin III results in an increase in tryptophan fluorescence. There is also a small blue shift in the wavelength of the emission maximum. There are several possible explanations of these observations [18]. One is that some tryptophan residues are located in, or close to, the heparin binding sites and are thus directly affected by the binding of heparin, which makes their environment more hydrophobic. Another is that there is a general conformation change of the antithrombin III molecule on binding of heparin resulting in changes in the local microenvironment around certain tryptophan residues located some distance away from the heparin binding sites. The observation that breaking of the disulfide bonds affects the binding of heparin but not the tryptophan fluorescence may be taken as supporting evidence for the latter possibility. The Scatchard plots obtained from data on the binding of heparin to antithrombin vary with the type of heparin used. This is not unexpected, as commercial heparin is known to be a heterogeneous mixture of molecules differing both with respect to molecular size and probably also in chemical structure. The heparin prepared by affinity chromatography has the highest specific anticoagulant activity and is strongly bound only to one site on the antithrombin III molecule. The binding

110 curves obtained with the other two heparin samples are more difficult to interpret, but it appears as if there would be two binding sites in the antithrombin III [19] having an affinity for components present in these samples. The absence o f signs of this second binding site in the experiments with heparin prepared by affinity c h r o m a t o g r a p h y indicates that this site does not bind the most active heparin components. It is therefore likely that the second binding site accomodates certain heparin-like molecules with low or no anticoagulant activities [20-22]. These molecules appear to be o f molecular sizes similar to those o f the active heparin molecules, but must differ with respect to chemical structure. The type of binding curves determined in this study may thus be used to study the heterogeneity o f heparin samples. It follows from what has been said above that with regard to the anticoagulant activity of heparin, it is probably only the first strong binding site that is o f any importance. The association constant of this site is found to be 2.3. l06 M - 1 in the experiments with heparin prepared by affinity chromatography, but about 7.105 M -1 with the other heparin samples. These latter values are likely to be low as these samples probably contain molecules with low or no affinity for antithrombin I I I so that the heparin concentration values used in the binding calculations do not correspond to concentrations o f a high affinity heparin". It has previously been shown that in order to have an anticoagulant effect, heparin must have a molecular weight above 7000 [23]. A possible explanation o f this would be that the strong binding site might be composed o f several subsites on the molecule with each binding to a certain repeating unit in the heparin. A similar model has previously been proposed for lysozyme to explain its saccharide binding data [24]. ACKNOWLEDGEMENTS The authors wish to thank Drs. 1. Bj/krk and M. H 6 6 k for gift o f heparin. The skilful technical assistance o f Mr. E. Holmer, Miss E. Jahr and Miss L. E n g m a n is gratefully acknowledged. REFERENCES 1 2 3 4 5 6 7

Abilgaard, U. (1968) Scand. J. Clin. Lab. Invest. 21, 89-91 Rosenberg, R. D. and Damus, P. S. (1973) J. Biol. Chem. 248, 6490-6505 Miller-Andersson, M., Borg, H. and Andersson, L.-O, (1974) Thromb. Res. 5, 439-452 Yin, E. T., Wessler, S. and Stoll, P. J. (1971) J. Biol. Chem. 246, 3712-3719 Damus, P. S., Hicks, M. and Rosenberg, R. D. (1973) Nature 246, 355-357 Machovich, R. (1975) Biochim. Biophys. Acta 412, 13-17 Lowry, O. H., Rosebrough, N. J., Farr, A. L. and Randall, R. S. (1951) J. Biol. Chem. 193, 265-275 8 Cbervenka, C. H. (1970) Anal. Biochem. 34, 24-29 9 Karkhanis, Y. D., Carlo~ D. J. and Zeltner, J. (1975) Anal. Biochem. 69, 55-60 10 Hugh, T. and Moore, S. (1972) J. Biol. Chem. 247, 2828-2834 11 Kurachi, K., Schmer, G., Hermodson, M. A., Teller, D. C. and Davie, E. W. (1976) Biochemistry 15, 368 373 12 Teale, F. W. J. (1960) Biochem. J. 76, 381-388 13 Scatchard, G. (1949) Ann. N.Y. Acad. Sci. 51, 660-672 14 Klotz, 1. M. (1946) J. Am. Chem. Soc. 68, 1486-1490 15 Teale, F. W. J. and Weber, G. (1957) Biochem. J. 65, 476-482

111 16 17 18 19 20 21 22 23

Kronman, M. J. and Holmes, L. G. (1971) Photochem. Photobiol. 14, 113-134 Cowgill, R. W. (1967) Biochim. Biophys. Acta 140, 37-44 Rosenberg, R. D. (1975) New England, J. Med. 292, 146-151 Einarsson, R. (1976) Biochem. Biophys. Acta, 446, 124-133 Stivala, S. S. and Liberti, P. A. (1967) Arch. Biochem. Biophys. 122, 40-54 Silva, M. E. and Dietrich, C. P. (1975) J. Biol. Chem. 250, 6841-6846 Dietrich, C. P. (1968) Biochem. J. 108, 647-654 Nader, H. B., McDuffie, N. M. and Dietrich, C. P. (1974) Biochem. Biophys. Res. Commun. 57, 488-493 24 Chipman, D. M., Grisaro, V. and Sharon, N. (1967) J. Biol. Chem. 242, 4388-4394