Human Factor XI1 (Hageman Factor) Autoactivation by Dextran Sulfate CIRCULAR DICHROISM, FLUORESCENCE, AND ULTRAVIOLET DIFFERENCE SPECTROSCOPIC STUDIES*

The first event leading to the activation of the plasma kallikrein-kinin system is the surface-dependent con-version of factor XI1 to an active enzyme. Factor XI1 autoactivation was investigated using dextran sulfate as a soluble activating surface, and the significance of aggregation and the nature of the conformational change were examined by ultraviolet difference spectroscopy, fluorescence and circular dichroism. Results indicate that DS500 (500-kDa dextran sulfate) induces aggregation of factor XII. Analysis of the binding data suggests that 165-192 factor XI1 molecules can bind to one DS500 chain, while a 1:l stoichiometry is observed with 5-kDa dextran sulfate. The interaction of factor XI1 and dextran sulfate is a biphasic process. It is initiated by a fast contraction of the molecule upon binding, as revealed by an apparent increase in orga- nized secondary structures, and then followed by a slow relaxation process during cleavage and subse- quent activation. Overall, the results are consistent with a model in which factor XI1 undergoes confor- mational changes upon binding to the activating surface. The rapidity of autoactivation in the presence of DSSOO, as opposed to 5-kDa dextran sulfate, implies that aggregation provides a special mechanism whereby proteolytic cleavage is accomplished effi- ciently when factor XI1 molecules are bound side by side on the

The first event leading to the activation of the plasma kallikrein-kinin system is the surface-dependent conversion of factor XI1 to an active enzyme. Factor XI1 autoactivation was investigated using dextran sulfate as a soluble activating surface, and the significance of aggregation and the nature of the conformational change were examined by ultraviolet difference spectroscopy, fluorescence and circular dichroism. Results indicate that DS500 (500-kDa dextran sulfate) induces aggregation of factor XII. Analysis of the binding data suggests that 165-192 factor XI1 molecules can bind to one DS500 chain, while a 1:l stoichiometry is observed with 5-kDa dextran sulfate. The interaction of factor XI1 and dextran sulfate is a biphasic process. It is initiated by a fast contraction of the molecule upon binding, as revealed by an apparent increase in organized secondary structures, and then followed by a slow relaxation process during cleavage and subsequent activation. Overall, the results are consistent with a model in which factor XI1 undergoes conformational changes upon binding to the activating surface. The rapidity of autoactivation in the presence of DSSOO, as opposed to 5-kDa dextran sulfate, implies that aggregation provides a special mechanism whereby proteolytic cleavage is accomplished efficiently when factor XI1 molecules are bound side by side on the DS500 molecule.
Human blood coagulation factor XI1 (Hageman factor) is a single chain 80-kDa plasma glycoprotein. Concomitant with cleavage of a single peptide bond, factor XI1 is converted to a serine protease, factor XIIa, which initiates the intrinsic pathway of blood coagulation. The enzyme plays a role in the activation of the fibrinolytic system, in the production of kinins, in the initiation of cell-mediated inflammatory responses, and in the activation of the classical complement pathway (1)(2)(3)(4)(5)(6). In vitro activation of factor XI1 occurs when the zymogen becomes bound to negatively charged "surfaces" * This work was supported by National Institutes of Health Grant HL 43252 (to G.B.V), a Specialized Center for Research in Thrombosis Grant HL45486 (to R. W. C), and Grant-In-Aid 891231 from the American Heart Association National program (to R. A. P). Parts of this work were presented at the XIIIth International Conference on Thrombosis and Hemostasis, June 30 to July 6, 1991, Amsterdam, The Netherlands. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In uiuo, contact of plasma with anionic components of the subendothelial basement membrane or cell surface may be responsible for the activation of factor XI1 (11, 12) but the specific component(s) have not been identified.
Surfacebound factor XI1 is activated to factor XIIa and acquires enzymatic activity toward its protein substrates, prekallikrein and factor XI (5, 13), which are complexed in uiuo, with the contact activation procofactor high molecular weight kininogen (14). Plasma kallikrein produced by the activation of prekallikrein by factor XIIa, in turn, cleaves more factor XI1 to factor XIIa. This reciprocal activation accounts for the rapid and amplified activation of the intrinsic pathway. To render factor XI1 susceptible to proteolytic cleavage by autoactivation or by enzyme activation, it has been hypothesized that a conformational change of the zymogen upon binding to a negatively charged activating substance must occur (15). Other investigators have also proposed a model in which a conformational change occurs upon activation (16)(17)(18)(19). Earlier studies also indicated that autoactivation of factor XI1 in the presence of an insoluble surface, ellagic acid, was accompanied by aggregation (16). Although experimental verification of the conformational change has been reported, the exact nature of this change was not clearly evaluated, and the relevance of aggregation to autoactivation has not been investigated. In this report a comparative study of DS500' and DS5 was carried out since, unlike DS500, DS5 is a weak surface activator and does not cause aggregation of factor XII. Ultraviolet difference spectroscopy, fluorescence, and native electrophoretic analysis were used to determine the binding parameters of factor XII-DS5OO and factor XII-DS5 interactions. By analysis of the spectral perturbations of aromatic amino acid residues, these techniques also provided new information on the nature of the surface binding site and the aggregation site in factor XII. The circular dichroism studies have identified some of the detailed features of the conformational change that accompanies the surface activation of factor XII. from Calbiochem Behring Corp. Dextran sulfates (DS5 and DS500), hexadimethrine bromide (polybrene), and the siliconizing agent Sigmacote were purchased from Sigma. Buffer strips and separation media used in PhastSystem Electrophoresis Unit were purchased from Pharmacia LKB Biotechnology Inc. Buffers and salts of analytical quality were purchased from Fisher. Pure factor XI1 was obtained either by immunopurification as described elsewhere (20) or purchased from Enzyme Research Laboratories (South Bend, IN). Prior to use, the purity of the purchased factor XI1 was verified by gel electrophoresis and was found to be single band on reduced SDS-PAGE. The specific activity ranged from 60 to 80 units/mg based on amidolytic assay.

Methods
Amidolytic Assay-The factor XI1 samples to be assayed were incubated at 25 "C in the buffer used in spectroscopic studies (50 mM NaCl, 1 mM sodium phosphate, pH 7.5), and amidolytic activity was measured using the chromogenic substrate, S-2302. Ninety pl of assay buffer (140 mM NaC1; 50 mM Tris-HC1, pH 7.8, 1 mM EDTA) containing a final substrate concentration of 0.7 mM was pipetted to a siliconized ultramicro cell. Ten pl of the sample to be assayed after necessary incubation was then added to the cuvette, and the absorbance change at 405 nm was recorded immediately using a 14 DS UV Visible spectrophotometer (AVIV Associates) to determine the initial velocity of the reaction.
Gel Electrophoresis-The "band shift" assay to demonstrate factor XI1 binding to DS500 was done under native conditions using a precast 12.5% homogeneous polyacrylamide gel (4.5% stacking gel) on a PhastSystem Electrophoresis Unit (Pharmacia). In these experiments, the samples are applied as in the PhastSystem Guide (Technique File No. 120) using the eight-toothed 1-p1 sample applicator comb. Electrophoresis on reduced SDS gel was performed similarly (Separation Technique File No. 110) on a precast 10-15% gradient polyacrylamide gel. The gels were silver stained using the automated development unit (Separation Technique File No. 210). Densitometric analysis was performed using "IMAGE-PRO PLUS" image processing system software (Media Cybernetics, Silver Spring, MD). Ultraviolet Difference Spectroscopy-All spectroscopic measurements were carried out on acid-cleaned (9:1, H,S04-HN03) quartz cuvettes which have been coated with either polybrene (2 mg/ml) or Sigmacote. The UV difference spectra were recorded on a 14 DS UV Visible spectrophotometer using ultramicro cells (Hellma Cells). In order to investigate the extent of light scattering contribution to the UV difference spectrum of factor XII-DS500, three methods were employed. 1) Using a "LOGGEN" light scattering correction software (AVIV Associates): this program automatically subtracts the light scattering contribution based on the slope of the spectrum between 320 and 350 nm. 2) Using a modified 1-mm pathlength jacketed cylindrical fluorscat cell (21,22) where the sample is surrounded by a solution of 0.1 M sodium salicylate. The difference between the spectrum taken in the fluorscat cell and standard cell is a direct measure of differential light scattering. 3) By comparing the UV difference spectra of factor XII-DS500 and factor XII-DS5: DS500 is found to induce light scattering on factor XI1 while DS5 does not.
Circular Dichroism-The CD measurements were conducted on a Jasco J-5OOC Spectropolarimeter, and the CD difference spectra were obtained using the ADALAB-PC hardware and "ADAPT" computer software (Interactive Microware, Inc., State College, PA). A solvent blank containing dextran sulfate (25 pglml), which gave a negligible CD (<5%) in the spectral region of interest, was subtracted from the CD spectra of the factor XII-DS500 complex. Sodium chloride was replaced by sodium fluoride in the solvent in order to improve the sensitivity and signal to noise ratio at lower wavelengths. The use of sodium fluoride did not affect the measured functional activities of factor XIIa. The mean residue molecular weight (MJ of 106 (19) for factor XI1 is used in the ellipticity calculations.
In evaluating the contribution of light scattering to the CD spectrum of factor XII-DSBOO, the first method used involves varying the distance of the sample cell to the photomultiplier detector. Differential light scattering is indicated if the CD spectrum at 340 nm is a function of the distance from the detector. In addition, any deviation from the base line at 340 nm would indicate light scattering because proteins and polysaccharides do not have CD bands at this wavelength. The second method involves comparing the CD spectra of factor XII-DS5OO at 340 nm with factor XII-DS5 and factor XI1 alone.
Fluorescence Spectroscopy-A Perkin-Elmer Cetus LS-5B luminescence spectrofluorometer is used to measure the protein intrinsic fluorescence. All measurements are performed at 25 "C in a 3-mm pathlength quartz microcuvette. The excitation wavelength is set at 280 nm, and the emission spectra are scanned from 300 to 500 nm. In all measurements, 200 pg/ml of factor XI1 is used in the presence of 50 mM NaCl, 1 mM phosphate, pH 7.5. The sensitivity scale is adjusted for an initial fluorescence above 80% full scale.
Analysis of Binding Parameters-In the derivation below, the following terms are used [Po], initial factor XI1 concentration; a, fraction of free [Po]; (1-a), fraction of bound [Po]; [D,], initial DS500 concentration; p, fraction of free binding sites on DS500; (1-p), fraction of bound binding sites on DS500; n, number of binding sites/ DS500 molecule, and K", dissociation constant. (1) where From Equation 1 Equation 5 is particularly useful for the evaluation of binding parameters in which the concentration of free ligand is not measured directly but rather expressed in terms of the total ligand present in the system. The value of a is calculated for every desired point on the titration curve of the molar difference absorption, ACM uersus total DS500 concentration, using the relationship a = Atmax -ACM Atme, ACM represents the molar difference absorption at a certain DS500 concentration and Aemax represents the maximum molar difference absorption. It is not possible to titrate DS500 with factor XI1 because the intensity of the corrected protein UV difference spectrum due to DS500 is not very large at low DS500 concentration. However, we find that similar binding parameters can be determined by titrating factor XI1 with DS500. Since the spectral parameter that is monitored originates from factor XI1 the saturation of binding still corresponds to the saturation of the spectral change. (For a review of binding studies using this technique, see Ref. 23.) In the evaluation of the binding parameters of factor XII-DS5 by fluorescence spectroscopy, the value of a is calculated for every desired point in the titration curve using the relationship where F represents the fluorescence intensity at a certain DS5 concentration, F,i, represents maximum quenching of fluorescence upon saturation, and F, is the initial fluorescence intensity.

Studies of Dextran Sulfate-induced Factor X I I Aggregation and Autoactivation Using
Ultraviolet Difference Spectroscopy-The conformational aspects of factor XI1 autoactivation by insoluble surfaces such as glass and ellagic acid have been reported (19). Since it is not known whether the true surface activator in vivo is soluble or not the present study was conducted in order to characterize the conformational and structural aspects of factor XI1 autoactivation by a soluble surface, dextran sulfate. We first determined if aggregation occurs in the presence of dextran sulfate. To this end, the technique of ultraviolet difference spectroscopy is used. When the UV difference spectrum of factor XI1 is examined in the presence of DS5 or DS500, striking differences are found. The factor XI1 spectrum in the presence of DS500 becomes positive immediately at 340 nm and thereafter rises exponentially at lower wavelength (data not shown). Since no sugars or amino acid residues absorb above 310 nm, this spectrum of factor XI1 in the presence of DS500 must contain contribution from light scattering due to aggregation. In order to interpret this spectrum, it is necessary to separate the light scattering contribution. This is accomplished by procedures outlined under "Experimental Procedures." The light scattering-corrected difference spectra (Fig. 1A) are obtained using the "LOGGEN" light scattering correction software and confirmed using the fluorscat cell technique. In the presence of 25 pg/ml DS500 (0.05 pM), factor XI1 (2.4 p M ) exhibits a positive difference spectrum with a A~M of +2,500 f 250 a t 288 nm (dotted line). At higher protein concentration (5.1 p M factor XII), this spectral maximum is more "red-shifted to 290 nm (dashed line). In the presence of 25 pg/ml DS5 (5 p~) , no light scattering is observed, but the factor XI1 spectrum is negative (ACM = -3,800 5 300 at 280 nm) and independent of protein concentration (Fig. 1B). Fig. 1C shows the time-dependent changes in the protein UV difference spectra of factor XI1 in the presence of DS5 and DS500. Upon addition of DS500, an immediate positive absorption difference spectrum is observed followed by a gradual decrease in amplitude (open and closed squares). In the presence of DS5 (open and closed triangles), the initial change is also immediate, and the difference spectrum becomes more negative with time. The differences in both sign and magnitude of the spectral changes indicate that different perturbations of aromatic residues occur in each case.
Before evaluating the significance of these contrasting spectral changes to factor XI1 binding and autoactivation, we first need to demonstrate that they are not due to artifacts in the experimental conditions. The possibility that factor XIIa is released from dextran sulfate upon autoactivation can be ruled out because it has been demonstrated by previous studies that factor XIIa binds more strongly to dextran sulfate than factor XI1 (24). The light scattering is not due to a decrease in protein solubility because an absorbance measurement a t 410 nm shows that the solution of factor XII-DS500 is more transparent than factor XII-DS5 (data not shown). We observed these contrasting difference spectra in polybrenecoated as well as in siliconized cuvettes (Fig. 2). This finding rules out the possibility of artifacts due to aggregation between the polybrene (polycation) present on the cuvette wall and dextran sulfate (polyanion).
The solvent condition used throughout these studies was 50 mM NaCl or 50 mM NaF, 1 mM sodium phosphate, pH 7.5. Low ionic strength could affect the binding and autoactivation reactions of factor XII. Therefore, we examined the ionic strength dependence of factor XI1 autoactivation under this condition using reduced SDS-PAGE. The formation of the 52-and 28-kDa fragments resulting from factor XI1 autoactivation is found to be maximal at ionic strength around 50 mM (data not shown), and very little fragmentation occurs at high ionic strength. When the factor XII-DS5OO complex is allowed to form at low ionic strength (50 mM) and then the ionic strength is raised to 0.5 M, the positive UV difference spectrum is reduced to the base line (Fig. 2, dashed-dotted  line). This is presumably due to disruption of the protein- polyanion complex a t high ionic strength.
We also examined the time-dependent cleavage patterns of factor XI1 from the uncleaved 80-kDa molecule to the cleaved activated molecule (52 and 28 kDa) by SDS-PAGE under reducing condition and the generation of amidolytic activity against the factor XIIa chromogenic peptide substrate S-2302. Both of these structural and functional tests showed that the time-dependent UV spectral changes (Fig. 1C) were accompanied by proteolytic cleavage of factor XI1 and the hydrolysis of S-2302 (data not shown). The data also confirmed the previous observation (25) that factor XI1 autoactivation is about 5-fold faster in the presence of DS500 as compared to DS5. We also observed slightly enhanced difference spectrum when factor XI1 is mixed with DS500 in the presence of factor of Factor X I I Autoactiuation  (-) to inhibit proteolysis and autoactivation. The spectra were all corrected for the respective base lines. In all cases, the final concentration of factor XI1 and DS500 were 2.5 p~ and 25 pg/ ml, respectively. The vertical lines represent the range of error based on three to four determinations.
XIIa inhibitor, H-D-Phe-Phe-Arg-CMK (Fig. 2, solid lines). This indicates that the UV difference spectrum may contain small components attributed to proteolytic cleavage and further suggests that the spectral contribution due to cleavage is a negative spectrum. It is now clear that the binding and cleavage of factor XI1 in the presence of dextran sulfate is accompanied by a negative UV difference spectrum as seen in the case of the low molecular weight DS5. In the case of DS500, the positive difference spectrum must therefore be the net effect of binding (which is negative) and aggregation. Therefore, it is concluded that aggregation of factor XI1 on DS500 is responsible for the positive difference spectrum.
The UV spectral changes observed as factor XI1 aggregates on DS500 are quantitative measure of factor XI1 binding and can be utilized to determine binding parameters. Since the UV difference spectrum of factor XII-DS500 complex has some negative spectral contribution due to cleavage the spectral titration to study the binding behavior is monitored in the presence of the factor XIIa inhibitor H-D-Phe-Phe-Arg-CMK. In the presence of this inhibitor, the spectrum is only due to binding and aggregation because catalysis cannot occur, and subsequent cleavage and autoactivation are suppressed. T o fixed amounts of factor XI1 (97.5 pl, 200 pg/ml) are added 2.5 pl of different stock solutions of DS500 to obtain the desired concentrations (0-120 pg/ml). The results shown in Fig. 3A (closed circles) indicate that the DS500-induced spectral change attain maximum between 6-10 pg/ml DS500, and the titration curve exhibits a hyperbolic behavior. However, as more DS500 is added, the spectrum diminishes and is abolished a t high DS500 concentration (ACM = 0 f 100 a t approximately 47 pg/ml DS500). Upon further titration, the spectrum becomes negative and is similar to that obtained in the DS5 titration. The abolition of the positive spectrum suggests that at high DS500 concentration the number of adjacent factor XI1 molecules bound to DS500 decreases as they redistribute among large number of DS500 chains. Under this condition, aggregation is diminished because the intermolecular interaction among factor XI1 molecules is decreased and so is the positive UV difference spectrum. To determine the binding parameters of the factor XII-DS500 interaction, the observed ACM values from 0 pg/ml DS500 to 20 pg/ml DS500 are plotted against micromolar concentration of DS500, and the data are analyzed as shown in number of factor XII-binding sites/DS500 chain is n = 192 * 20 sites.
The titration of factor XI1 by DS500 was also examined by native gel electrophoresis (inset, Fig. 3A). This band shift assay was devised to visualize the binding of factor XI1 to DS500. When preincubated in varying amounts of DS500, the factor XII-DS500 complex migrates only a short distance within the stacking gel and does not enter the separation gel, presumably because of increase in size due to complex formation and aggregation. The unbound factor XI1 migrates to the separation gel as a light staining, diffused band. However, the amount of factor XI1 that is retained in the stacking gel depends on the DS500 concentration. Under the same conditions used in the spectral titration, all of factor XI1 is retained in the stacking gel (factor XI1 band disappears in the separation gel) when the concentration of DS500 reaches 10 pg/ml (lane 5 ) , which corresponds to the maximum ACM in the spectral studies. The reappearance of protein staining at high DS500 concentration (lane 8) is possibly due to partial dissociation of the factor XII-DS500 complex as suggested by the loss of the positive UV difference spectra. When the binding parameters are evaluated from the band shift assay (Fig. 3A, open circles), the values obtained are KD = 1.7 X 10" M (S.E. = 0.4 x W 7 , n = 5) and n = 165 30 sites.
It is difficult to evaluate the binding parameters of factor XII-DS5 complex from the UV spectral data in Fig. 3A because the spectra at 280 nm did not show saturation and continue to become negative as more DS5 is added. To circumvent this problem, fluorescence spectroscopy is used by monitoring the quenching of the intrinsic fluorescence of factor XI1 as small amounts of DS5 are added. Fig. 4 shows the relative changes in the intrinsic tryptophan fluorescence of factor XI1 at 340 nm (excitation at 280 nm) as 2.5 pM of the protein is titrated with small increments of DS5. The fluorescence intensity decreases as DS5 is added and levels off when about 15% of the initial fluorescence is quenched at 2.4 p~ DS5. Addition of more DS5 after the initial saturation is reached results in further quenching of fluorescence. When the data points from 0 to 2.4 pg/ml are plotted (inset, Fig. 4 scattering of aggregated factor XII.

Studies of Factor XII Conformation by Circular Dichroism-
The CD spectra of factor XI1 and factor XI1 in the presence of dextran sulfate were investigated in order to determine whether the anionic mucopolysaccharide has an effect on the secondary structure of the protein. To accomplish this we determined the possible contribution of light scattering to the CD spectrum of factor XII-DS500 as described under "Experimental Procedures." First, we examined the effect on the CD of factor XII-DS500 of varying the distance of the sample cell from the photomultiplier detector in the Jasco J-500C spectropolarimeter. The CD scans between 195 and 350 nm were compared when the sample was placed 18.5 and 5.2 cm from the detector. It is found that the CD of factor XII-DS500 is independent of the distance from the detector and does not deviate from the base line at 340 nm (data not shown). It is also found that the CD spectrum of factor XII-DS500 is superimposable with the spectra of factor XI1 alone and factor XII-DS5 at 340 nm (data not shown). The same solution of factor XII-DS500 shows a large light scattering absorption when recorded in a regular spectrometer. Additionally, if light scattering contributed to the CD signal of factor XII-DS500, deviation from the base line at 340 nm would be observed as we examined the CD at increasing protein concentrations. No concentration dependence was observed for the CD of factor XI1 in the presence of either DS5 or DS500 over a factor XI1 concentration range of 0.3 to 5.1 p~ (data not shown). These experiments indicate that although factor XII-DS500 has a large unpolarized scattering absorption, there is no detectable differential light scattering as determined by CD. From these data, we conclude that the aggregation of factor XI1 in the presence of DS500 must be scattering the left and right circularly polarized light with essentially equal efficiency so that the CD spectrum is solely due to differential CD absorption. This finding means that it is not necessary to correct the CD signal of factor XII-DS500 for light scattering.
The CD of factor XI1 in the absence of dextran sulfate is similar to previously reported studies (19)  = +960 degrees. cm2. dmol"). Overall, the CD spectra suggest that factor XI1 conformation is mostly of the random structure with very little or no a-helices or @-sheets. The fit of the CD spectrum of factor XI1 using reference proteins (26) is poor. This finding is consistent with a low percentage of organized secondary structure. A better fit was obtained using the data on synthetic polypeptides (27). The best fit corresponds to 0% a-helix, 26.5% @-sheet, and 73.5% random coil.
We examined closely the effect of dextran sulfate on the CD spectrum of factor XI1 in the far ultraviolet region between 225 and 195 nm. This region of the CD spectrum corresponds to the amide transitions and is very sensitive to changes in protein secondary structures. Fig. 5A shows representative scans of the time-dependent changes in direct CD spectra of factor XI1 in the presence of DS500. When DS500 is added to factor XI1 there is an immediate spectral change characterized by a 3 nm red-shift of the trough at 204 nm (curue b) to 207 nm (curue c ) . This is followed by a gradual "blue-shift," and in 60 min (curue e ) the trough shifts back to 204 nm and the amplitude is increased by 25%. The initial red-shift is the same for both DS5 and DS500 but the blueshift occurs at a longer time frame in the case of DS5 (data not shown). These spectral shifts are also clearly demonstrated as CD difference spectra (Fig. 5B). Immediate enhancement of the difference CD signal at zero time (<3 min) should be zero if there is no conformational change during the initial binding of DS500 to factor XII. The deviation from zero ellipticity at zero time as shown in Fig. 5C is a clear indication of the instant conformational change. Although the relative CD changes are small, the data certainly indicates that changes occur in the protein secondary structure when factor XI1 binds to dextran sulfate. The slow spectral changes after 3 min demonstrate the conformational change that occurs upon cleavage of factor XI1 subsequent to binding. Interestingly, these CD spectral changes occur concurrently with the observed UV difference spectral changes.

DISCUSSION
The origin of UV difference spectra of proteins have been well-documented (28)(29)(30)(31)(32). The spectral perturbation resulting in red-shift, indicated by a positive difference spectrum, arises when the aromatic chromophores ( i e . Trp, Tyr, Phe) experience more nonpolar environment and occurs when the chromophores become buried within the protein fold as a result of intermolecular/intramolecular interactions or masked from the solvent environment by ligands or substrate molecules. A blue-shift that is indicated by a negative spectrum is generally interpreted as due to increased polarity of the chromophoric environment and can be due to increases in exposure of the chromophores to the solvent as a result of a conformational change or by direct interaction with molecules more polar than the solvent environment.
The negative UV difference spectrum of factor XI1 in DS5 suggests that the chromophores are in a more polar environment when factor XI1 is bound to DS5. Since no aggregation is observed in the presence of DS5, this negative spectrum must be the result of direct interaction of factor XI1 with highly anionic DS5. The positive UV difference spectrum of factor XI1 in DS500 suggests increased hydrophobicity of the aromatic environment when factor XI1 is bound to DS500. Since DS500 causes aggregation of factor XI1 we interpret this to mean that, in addition to direct interaction with DS500, part of the contribution to the spectrum arises from perturbation of aromatic chromophores located at the aggregation site, the interface between adjacent factor XI1 molecules bound to neighboring sites in the DS500 chain. This is possible only if factor XI1 molecules aggregate on the negative surface, and this accounts for the positive difference spectrum. The longer red-shift at higher concentrations of factor XI1 (5.1 ~L M factor XI1 in 25 pg/ml DS500, Fig. 1A) is an additional indication of larger contributions from this type of perturbation.
The number of factor XII-binding sites in the DS500 chain was previously reported to be about 88-100, based on the assumption that the average size of a globular factor XI1 molecule is equivalent to the size of a linear 5-kDa dextran sulfate (25,33). This assumes a single linear chain of factor XI1 molecules along the DS500 chain. The result of the present titration studies indicates that about twice this number of sites could be present in the DS500 chain ( n = 165-192 sites). This suggests the possibility that aggregation of factor XI1 in DS500 may be more compact as depicted in the model below.
tainty based on five to six determinations. Panel B, difference circular dichroism obtained using the "ADAPT" computer software. The letters indicate the difference spectra of the corresponding spectra shown in panel A . The difference spectra are enlarged three times to emphasize the difference. Panel C, quantitation of time-dependent changes in the CD difference spectra based on six separate determinations. The ellipticity, [ O ] , is expressed as degrees.cm'.dmol". ionic glycosaminoglycans, acidic phospholipids, glycolipids, and subendothelial basement membrane components, but direct demonstration of their role in surface activation of factor XI1 has been elusive. Heparin and collagen do not even activate purified factor XI1 i n vitro (35,36), acidic phospholipids are located in the cytosolic face of cell membranes, and the concentration of sulfatides are below the levels at which they could facilitate factor XI1 activation i n uiuo. Our understanding of dextran sulfate autoactivation of factor XI1 should Factor XII MODEL I. Factor XII-DSBOO complex.
In this model, factor XI1 is visualized as aggregating on either side of DS500 such that two linear chains of factor XI1 wrap around the DS500 chain.2 The closed circles represent the aromatic chromophores which are located at the factor XI1 interface (aggregation site). The positive UV difference spectrum arises when these chromophores, which are exposed and accessible to the solvent in free factor XII, are masked by the hydrophobic environment at the factor XI1 interface in the factor XII-DS500 complex. In the case of DS5, a 1:1 stoichiometry with factor XI1 was demonstrated by fluorescence spectroscopy at low DS5 concentration (Fig. 4). However, both fluorescence and UV difference spectroscopic titrations show further spectral changes after the stoichiometric amount of DS5 is reached. We cannot provide an unequivocal interpretation of this behavior, but it is reasonable to speculate that factor XI1 may have multiple binding sites for DS5.
Overall the results of both UV difference spectroscopy and circular dichroism studies demonstrate that autoactivation of factor XI1 in the presence of dextran sulfate is a biphasic process. The first rapid phase is associated with binding and aggregation and occurs concurrently with small but significant changes in the protein secondary structures. The second slow phase is associated with proteolytic cleavage and is also accompanied by conformational alteration of the factor XI1 molecule. Our interpretation of this observation is that when factor XI1 binds to dextran sulfate, the protein molecule tightens up transiently resulting in formation of small amount of organized elements of protein secondary structure such as a-helix or @-sheets. After cleavage during subsequent autoactivation, the factor XI1 molecule gradually relaxes back to its mostly random conformation.
Since the biphasic behavior of factor XI1 autoactivation has been demonstrated also in factor XII-sulfatide complex (34) and aggregation of factor XI1 has been observed in factor XII-ellagic acid complex (IO), the same mechanism may mimic factor XI1 autoactivation i n uiuo. Whether or not the fast initial conformational change of factor XII, as demonstrated by the present studies, is the key step that triggers the first event in contact activation can only be proven when the natural surface activator is identified. There are numerous physiologically relevant activating surfaces, such as polyan-In these studies, two molecular sizes of dextran sulfate, DS500 (500-kDa dextran sulfate) and DS5 (5-kDa dextran sulfate) are used.
These molecular masses are approximate values because commercial preparations of dextran sulfates are known to be polydisperse and therefore may contain small quantities of high and low molecular mass components.
provide some information on the expected properties and possible mechanism of action of the natural activator. These studies will be extended to other known activators as we learn more about the nature of the biological surface responsible for the activation of factor XII. At any rate, the use of soluble DS5 and DS500 proved valuable in explaining previously puzzling features of factor XI1 autoactivation.