INTRODUCTION

Factor XI (FXI)¹ is a member of the contact activation system of coagulation (1) and, at least in vitro, constitutes the link between contact activation and clotting via the intrinsic pathway of blood coagulation (1, 2). In addition to FXI, the contact system consists of factor XII (FXII), prekallikrein, and high molecular weight kininogen. FXII and prekallikrein reciprocally activate each other upon contact with negatively charged surfaces such as kaolin, glass, celite, or dextran sulfate (3, 4). Activated FXII (FXIIa) in turn cleaves FXI into the active form factor XIa (FXIa) (5, 6). High molecular weight kininogen serves as a nonenzymatic cofactor in these reactions. The contact system participates in the surface-dependent activation of blood coagulation, fibrinolysis, kinin generation, and inflammatory reactions (3, 4).

In plasma FXI is present at a concentration of 3-6 µg/ml (5, 7, 8, 9). It is a dimeric glycoprotein consisting of two identical polypeptide chains held together by a disulfide bond. Upon activation, each polypeptide chain can be cleaved at an internal peptide bond giving rise to disulfide-linked heavy and light chains, the latter each containing one active site (5, 10, 11). The activity of each active site of FXIa is regulated by plasma protease inhibitors (6, 12), including ₁-antitrypsin (13), antithrombin III (14), C1 inhibitor (15), and ₂-antiplasmin (16), each a member of the superfamily of serine protease inhibitors (serpins).

Although initial studies suggested ₁-antitrypsin to be the main inhibitor of FXIa (17), we recently demonstrated, utilizing enzyme-linked immunosorbent assays (ELISAs) to quantitate complexes between FXIa and its inhibitors, C1 inhibitor to be a major inhibitor of FXIa, followed by ₁-antitrypsin, ₂-antiplasmin, and antithrombin III (18).

Dextran sulfate, heparin, heparan sulfate, and dermatan sulfate belong to the so-called glycosaminoglycans (19). Dextran sulfate is a synthetic polyanion, whereas heparin, heparan sulfate, and dermatan sulfate are physiological compounds, which occur in the human body (20, 21). For example, heparan sulfate is the predominant cell-associated glycosaminoglycan in the vascular bed (22, 23).

It has been reported that the inactivation of FXIa by antithrombin III is accelerated by heparin (12, 14, 24, 25), suggesting that the presence of glycosaminoglycans may alter the inhibition of FXIa. In contrast, no effect of glycosaminoglycans or a minimal protection from inactivation was found for the interaction of FXIIa with C1 inhibitor (26, 27). These observations raise the possibility that glycosaminoglycans selectively may affect the activity of contact system proteases. In the present report we investigated the influence of various physiological and nonphysiological glycosaminoglycans on the kinetics of the inactivation of FXIa by its plasma inhibitors, that is C1 inhibitor, ₁- antitrypsin, ₂-antiplasmin, and antithrombin III, and on the relative contribution of these inhibitors to the inactivation of FXIa in plasma. Similarly, we studied the effects of glycosaminoglycans on the kinetics of the inactivation of -FXIIa, -FXIIa, and kallikrein by their main inhibitor in plasma, C1 inhibitor. Our results indicate that glycosaminoglycans differentially affect the inhibition of contact system proteases, i.e. they selectively enhance inhibition of FXIa without affecting the activity of FXIIa or kallikrein.

EXPERIMENTAL PROCEDURES

Dextran sulfate (M_r 500,000, sulfur content 17%) was obtained from Pharmacia Fine Chemicals, Uppsala, Sweden; unfractionated heparin (1 unit/ml corresponding to 7 µg/ml) was from Kabi Vitrum, Stockholm, Sweden; dermatan sulfate (chondroitin sulfate B), heparan sulfate (from bovine intestinal mucosa), and soybean-trypsin inhibitor (SBTI, type I-S) were from Sigma. Hexadimethrine bromide (Polybrene) was from Janssen Chimica, Beerse, Belgium; Tween 20 from J. T. Baker, Inc. The chromogenic substrates Glu-Pro-Arg-p-nitroanilide (S-2366) and H-D-Pro-Phe-Arg-p-nitroanilide (S-2302) were from Chromogenix, Mölndal, Sweden.

Pooled normal human plasma containing 10 mmol/liter EDTA (EDTA-plasma) was obtained and prepared as described (28). Purified human FXIa was obtained from Kordia Laboratory Supplies, Leiden, The Netherlands, and was stored at 70 °C in 0.1 mol/liter, Tris-HCl, pH 7.4, 0.14 mol/liter NaCl, 0.1% (w/v) Tween 20. This preparation was made by incubating FXI with factor XIIa, after which factor XIIa was removed by absorption onto a corn trypsin inhibitor column. This FXIa preparation migrated as a single band at 160 kDa on nonreducing, and as two bands at 50 and 30 kDa, respectively, on reducing SDS-10-15% polyacrylamide gel electrophoresis (w/v). Monoclonal antibody OT-2, which is directed against the light chain of activated FXII and blocks its catalytic activity (29), was added to the FXIa preparation (80 µg/ml final concentration) to block traces of contaminating FXIIa. FXIa concentrations were expressed as the molar concentration of the 80-kDa subunits. Purified human -FXIIa was obtained from Kordia Laboratory Supplies, Leiden, The Netherlands. Kallikrein and -FXIIa were purified as described (30). All protease inhibitors were from human origin. C1 inhibitor was obtained from Behringwerke AG (Marburg, Germany), ₁-antitrypsin, ₂-antiplasmin, and antithrombin III were from Calbiochem. The concentration of functional antithrombin III was measured by titrations against active-site titrated human -thrombin (kindly provided by Dr. K. Mertens, Department of Blood Coagulation, Central Laboratory of the Netherlands Red Cross Blood Transfusion Service, Amsterdam).

Amidolytic activity of FXIa was determined in wells of microtiter plates (Greiner GmbH, Frickenhausen, Germany) by using the chromogenic substrate S-2366 at a final concentration of 0.4 mmol/liter in a buffer containing 0.1 mol/liter Tris-HCl, pH 7.4, 0.14 mol/liter NaCl, and 0.1% (w/v) Tween 20 (total volume of 200 µl). The initial change in absorbance at 405 nm (A) was measured at 37 °C using a Titertek twinreader (Flow Laboratories, Irvine, UK). The kinetic parameters of the hydrolysis of the chromogenic substrate S-2366 by FXIa were determined to be K_m = 0.34 mmol/liter and k_cat = 262 s¹.

All incubations were performed in 0.5-ml polypropylene tubes at 37 °C. Buffer (0.1 mol/liter Tris-HCl, pH 7.4, 0.14 NaCl, 0.1% (w/v) Tween 20) containing protease inhibitor with or without glycosaminoglycans was prewarmed at 37 °C for 5 min. After addition of prewarmed FXIa (final concentrations 3-8 nmol/liter) to the reaction mixtures, 10-µl aliquots were removed at various times, and residual amidolytic activity of FXIa was assessed by diluting in 190 µl of buffer and substrate as described above. The observed A/min, which was constant during the time of measurement, was converted to percentage of maximum activity by comparison with the A/min of the sample containing FXIa and glycosaminoglycan but no protease inhibitor. The kinetics of the inhibition was studied under pseudo first-order conditions with the inhibitors in 13-210-fold molar excess over FXIa. Under these conditions, the equation ln(E/E_o) = k × t, where E_o is the initial concentration of FXIa, and E the concentration of remaining FXIa at time t, describes the inhibitory kinetics (12) and was used to determine k, the pseudo first-order rate constant for the reaction. The data from each experiment were analyzed by linear regression analysis, and the second-order rate constant was calculated from plots of the pseudo first-order rate constant versus the inhibitor concentration (for C1 inhibitor) or from the equation k = second-order rate constant × inhibitor concentration (for the other serpins). Each rate constant was determined at least twice, the variation between the different determinations was 9.6 ± 0.5% (mean ± S.E.).

One volume of pooled EDTA-plasma was incubated at 37 °C with 1 volume of phosphate-buffered saline (PBS, 10 mmol/liter sodium phosphate, pH 7.4, 0.14 mmol/liter NaCl) containing dextran sulfate (0.025, 0.1, 0.5, or 1 mg/ml final concentration) for 30 min or 120 min. Activation was then stopped by adding 3 volumes of PBS containing 0.1 mg/ml SBTI and 0.05% (w/v) Polybrene (stop solution), and the amount of FXIa-FXIa inhibitor complexes generated in the mixtures was assessed.

For glass activation, EDTA-plasma was diluted 1:1 either with PBS alone or with PBS containing dextran sulfate (final concentration 10 µg/ml), heparin (50 unit/ml), heparan sulfate (1 mg/ml), or dermatan sulfate (1 mg/ml), respectively, and incubated for 0.5 h at 37 °C in glass tubes. Reaction was stopped either by adding stop solution as described above or by transferring the plasma mixtures into polypropylene tubes and freezing them at 70 °C.

mAb OT-2, which inhibits the catalytic activity of factor XIIa (29), was added to EDTA-plasma (final concentration 80 µg/ml) to prevent endogenous contact activation during the incubation period. This plasma was then mixed with an equal volume of PBS containing FXIa (final concentration 15 nmol/liter) in the presence or absence of dextran sulfate (final concentration 10 µg/ml), heparin (50 units/ml), heparan sulfate (1 mg/ml), or dermatan sulfate (1 mg/ml). After an incubation period of 1 h at 37 °C in polypropylene tubes, dilutions of the mixtures were tested in the ELISAs for FXIa-FXIa inhibitor complexes (see below). Control experiments showed no measurable generation of FXIa-FXIa inhibitor complexes in the absence of exogenous FXIa.

Complexes between FXIa and C1 inhibitor, ₁-antitrypsin, ₂-antiplasmin, antithrombin III, respectively, were determined using ELISAs as described (18). Molar concentrations of generated FXIa-FXIa inhibitor complexes were expressed based on a molecular mass of FXIa of 80 kDa, thus calculating the active sites of FXIa. The total amount of FXIa-FXIa inhibitor complexes was calculated as the sum of the measured complexes in the four ELISAs.

Kinetic Studies of the Inhibition of -FXIIa, -FXIIa, and Kallikrein by C1 Inhibitor

The kinetics of inhibition of -FXIIa, -FXIIa, and kallikrein were studied in analogous experiments as described above for inhibition of FXIa by ₁-antitrypsin, ₂-antiplasmin, and antithrombin III. The proteases were used at 5-40 nmol/liter final concentrations and C1 inhibitor at 100-500 nmol/liter final concentrations in PBS containing 0.1% (w/v) Tween 20. Residual enzymatic activity was measured as described above using the chromogenic substrate S-2302 (0.5 mmol/liter, final concentration).

RESULTS

Activation of the contact system in EDTA-plasma by dextran sulfate resulted in generation of a total amount of FXIa-FXIa inhibitor complexes of 4.6 ± 0.7 nmol/liter. In a previous study we showed that in plasma activated with kaolin or glass FXIa is inhibited by C1 inhibitor, ₁-antitrypsin, and ₂-antiplasmin each for about 30%, and to a lesser extent by antithrombin III (Fig. 1) (18). However, to our surprise, upon dextran sulfate activation, 90% of the complexes generated appeared to be FXIa-C1 inhibitor complexes, whereas complexes of FXIa with ₁-antitrypsin or ₂-antiplasmin were virtually absent (Fig. 1). A similar distribution pattern of FXIa between its inhibitors was obtained after prolongation of the incubation time up to 120 min or upon testing various concentrations (from 1 to 1000 µg/ml, final concentrations) of dextran sulfate (data not shown). The absence of FXIa-FXIa inhibitor complexes other than FXIa-C1 inhibitor complexes was not due to the inability to detect these complexes in the presence of dextran sulfate. In control experiments, the recovery of purified complexes added to dextran sulfate-activated FXI-deficient plasma was 83, 83, 104, and 97% for FXIa-C1 inhibitor complexes, FXIa-₁-antitrypsin complexes, FXIa-₂-antiplasmin complexes, and FXIa-antithrombin III complexes, respectively. These observations also ruled out the possibility that dextran sulfate in some way would stimulate the degradation of FXIa-FXIa-inhibitor complexes in plasma. We, therefore, hypothesized that the observed changes in the distribution pattern of FXIa among its inhibitors might be due to an influence of dextran sulfate on the inactivation kinetics of FXIa by its inhibitors. Therefore, we decided to investigate the effect of dextran sulfate and other naturally occurring glycosaminoglycans on the kinetic parameters of FXIa inactivation by C1 inhibitor and other plasma protease inhibitors.

In a previous report (31) glycosaminoglycans such as dextran sulfate were shown to affect directly the amidolytic activity of kallikrein. We, therefore, tested the effect of glycosaminoglycans on the activity of FXIa. Heparin, heparan sulfate, or dermatan sulfate had no measurable effect on the amidolytic activity of FXIa, whereas dextran sulfate dose-dependently inhibited this activity up to 50% (Fig. 2). In further experiments, results obtained with dextran sulfate were corrected for this direct effect of dextran sulfate on FXIa.

Inactivation of FXIa by C1 inhibitor was studied by measuring the disappearance of the amidolytic activity of FXIa using the chromogenic substrate S-2366. This reaction followed pseudo first-order kinetics, as was concluded from the straight lines obtained when the natural logarithm of residual FXIa amidolytic activity was plotted against time (Fig. 3A). The values of the apparent first-order rate constants, k, were calculated from the slopes of these lines and were found to be directly proportional to the C1 inhibitor concentrations (Fig. 3B). Therefore, inhibition was found to be second-order, in agreement with previous studies (32), and the calculated rate constant was 1.8 × 10³ M¹ s¹.

1

The reactions in the presence of various amounts of dextran sulfate, heparin, heparan sulfate, or dermatan sulfate also appeared to be pseudo first-order; however, the rate constants increased with increasing amounts of the glycosaminoglycans (Fig. 4). The apparent second-order rate constants at the highest concentrations of these glycosaminoglycans tested were increased considerably compared with the value in the absence of any glycosaminoglycan (Table I).

Table I. Second-order rate constants for the inactivation of FXIa by C1 inhibitor (C1 lnh), 1-antitrypsin (a1AT), 2-antiplasmin (a2AP), and antithrombin III (ATIII) in the presence of various glycosaminoglycans (GAGs) No GAG DXS,a (10 µg/ml)b Hep,a (50 units/ml)b HS,a (1 mg/ml)b DS,a (1 mg/ml)b 103 M1 s1 C1 lnh 1.8 210 85 42 6 A1AT 0.1 0.02 0.06 0.08 0.06 a2AP 0.43 0.19 0.51 0.57 0.65 ATIII 0.32 1.54 4.4 1.27 1.24 a DXS, dextran sulfate; Hep, heparin; HS, heparan sulfate; DS, dermatan sulfate. b Final concentration.

Inhibition of FXIa by antithrombin III, ₁-antitrypsin, and ₂-antiplasmin was studied using 0.96 µmol/liter inhibitor and 6 nmol/liter FXIa in the presence or absence of dextran sulfate (final concentration 10 µg/ml), heparin (50 units/ml), heparan sulfate (1 mg/ml), or dermatan sulfate (1 mg/ml), respectively. Again, straight lines were obtained in a semilogarithmic plot of the residual FXIa amidolytic activity against time demonstrating that the reaction was pseudo first-order. The values of the apparent second-order rate constants were calculated and are given in Table I.

The contribution of the four inhibitors to inactivation of FXIa was calculated by multiplying the determined second-order rate constant by the plasma concentration of the respective inhibitor (C1 inhibitor 2.5 µmol/liter, ₁-antitrypsin 45 µmol/liter, ₂-antiplasmin 1 µmol/liter, and antithrombin III 2 µmol/liter). The relative contribution of the respective inhibitors to FXIa inactivation was then obtained by converting the calculated values into the percentage of total (17). The results given in Table II predict C1 inhibitor to be the predominant inhibitor of FXIa in plasma in the presence and absence of glycosaminoglycans. The other protease inhibitors were predicted to contribute to a varying degree to inactivation of FXIa in plasma, dependent on the presence of glycosaminoglycans (Table II). To test this predicted relative contribution of the protease inhibitors to inactivation of FXIa in plasma, we studied the inhibition of exogenously added or endogenously generated FXIa in plasma (see below).

Table II. Observed (predicted) relative contributions of protease inhibitors in the presence of various glycosaminoglycans (GAGs) to inactivation of endogenous FXIa generated upon glass activation of plasma percentage of total No GAG DXSa (10 µg/ml)b Hepa (50 units/ml)b HSa (1 mg/ml)b DSa (1 mg/ml)b C1lnhc 48  (44)d 88  (99)d 76  (95)d 81  (94)d 59  (72)d a1ATc 21  (44) 5  (0) 2  (1) 7  (3) 5  (13) a2APc 27  (5) 6  (0) 4  (0) 8  (1) 8  (3) ATIIIc 4  (7) 1  (1) 18  (4) 4  (2) 28  (12) a DXS, dextran sulfate; Hep, heparin; HS, heparan sulfate; DS, dermatan sulfate. b final concentration. c C1lnh, C1 inhibitor; a1AT, 1-antitrypsin; a2AP, 2-antiplasmin; ATIII, antithrombin III. d Percentage of total.

Endogenous FXI in plasma is activated upon contact with artificial surfaces such as glass. We incubated EDTA-plasma in glass tubes with an equal volume of PBS containing dextran sulfate (final concentration 10 µg/ml), heparin (50 units/ml), heparan sulfate (1 mg/ml), or dermatan sulfate (1 mg/ml). Activation was stopped by removing the incubation mixtures from the glass tubes. The presence of glycosaminoglycans during activation of the mixtures significantly changed the distribution of endogenously generated FXIa between its inhibitors (Table II): the relative amount of endogenously generated FXIa complexed to C1 inhibitor was increased in the presence of all glycosaminoglycans, whereas the proportion of FXIa-₁-antitrypsin and FXIa-₂-antiplasmin complexes decreased in all incubation mixtures containing glycosaminoglycans. It should be noted that glass activation in the presence of dextran sulfate is in fact activation of plasma by two activators, namely by glass and with dextran sulfate. This is not the case for glass activation in the presence of the other glycosaminoglycans, control experiments showed them to be unable to induce contact activation.

Control experiments were performed to establish to what extent the generated FXIa remained bound to the glass surface and, therefore, escaped inactivation by the inhibitors and subsequent detection in the respective ELISAs. In these experiments, glass activation was terminated by adding stop buffer containing Polybrene to remove the contact system proteins from the glass surface. The total amount of FXIa-FXIa inhibitor complexes recovered under this condition was not different (99.5 ± 11%) from the total amount recovered without stop buffer containing Polybrene.

The relative contribution of inhibitors to inactivation of a protease may depend on whether the enzyme is generated endogenously or added to plasma (33). We, therefore, studied in addition the distribution pattern of exogenous FXIa between its inhibitors in plasma in the presence of glycosaminoglycans. FXIa was added to a mixture of 1 volume of EDTA-plasma (containing mAb OT-2 at a concentration sufficient to prevent activation of the endogenous contact proteins during incubation time) and 1 volume of PBS containing either no glycosaminoglycans or dextran sulfate (final concentration 10 µg/ml), heparin (50 units/ml), heparan sulfate (1 mg/ml), or dermatan sulfate (1 mg/ml), respectively. After incubation at 37 °C for 60 min, FXIa-FXIa inhibitor complexes were measured. The observed distribution of FXIa between its inhibitors (data not shown) was in agreement with the results from inactivation of endogenously generated FXIa (Table II). However, in the presence of dermatan sulfate, the contribution of C1 inhibitor increased from 59 (endogenously generated FXIa) to 75% (exogenously added FXIa), whereas the contribution of antithrombin III decreased from 28 to 18%, respectively.

Kinetic Studies of the Inhibition of -FXIIa, -FXIIa, and Kallikrein by C1 Inhibitor

Inhibition of -FXIIa, -FXIIa, and kallikrein by C1 inhibitor was studied in the presence or absence of dextran sulfate (final concentration 125 µg/ml), heparan sulfate (1 mg/ml), or dermatan sulfate (1 mg/ml), respectively. Again, straight lines were obtained in a semilogarithmic plot of the residual enzymatic activity against time demonstrating that the reaction was pseudo first-order. The values of the apparent second-order rate constants for -FXIIa, -FXIIa, and kallikrein are given in Table III. These results demonstrate no significant influence of the glycosaminoglycans on the inhibition rate of kallikrein by C1 inhibitor and show an about 2-fold protection of -FXIIa and -FXIIa from inhibition by C1 inhibitor in the presence of dextran sulfate.

Table III. Second-order rate constants for the inactivation of -FXIIa, -FXIIa, and kallikrein by C1 inhibitor in the presence of various glycosaminoglycans (GAGs) No GAG DXSa (125 µg/ml)b HSa (1 mg/ml)b DSa (1 mg/ml)b 103M1s1  -FXIIa 8.0 3.1 6.8 10.3  -FXIIa 9.8 5.4 8.2 11.9 Kallikrein 25.5 22.1 24.5 26.0 a DXS, dextran sulfate; HS, heparan sulfate; DS, dermatan sulfate. b Final concentration.

DISCUSSION

There is growing interest in the potential role of glycosaminoglycans as biological active compounds in a number of (patho)physiological processes (23, 34). Here we report on differential effects of various glycosaminoglycans on the rate of inactivation of FXIa, -FXIIa, -FXIIa, and kallikrein by C1 inhibitor and other serpins.

The apparent second-order rate constant for the inhibition of FXIa by C1 inhibitor in the absence of any glycosaminoglycans was determined to be 1.8 × 10³ M¹ s¹, in agreement with previous work (32). The pseudo first-order as well as the second-order rate constants increased in the presence of the various glycosaminoglycans (Fig. 4 and Table I), dextran sulfate being the most effective compound. Similar effects of heparin in particular have been described for the inhibition of C1s by C1 inhibitor (35, 36), although these effects are not so pronounced as those reported here for the inhibition of FXIa by C1 inhibitor.

Inactivation of FXIa by antithrombin III and its acceleration by heparin have both been studied in detail (12, 14, 24, 25). Our results, showing a second-order rate constant of 0.32 × 10³ M¹ s¹, which increased about 14-fold in the presence of heparin (50 units/ml), are in agreement with these reports. Moreover, our results demonstrate that antithrombin III-mediated inhibition of FXIa is also enhanced in the presence of dextran sulfate, heparan sulfate, and dermatan sulfate (Table I). It is indeed well known that the functional activity of antithrombin III can be enhanced by various glycosaminoglycans (37, 38). The findings reported here indicate that this also applies for inactivation of FXIa.

The mechanism by which glycosaminoglycans potentiate C1 inhibitor and antithrombin III toward inhibition of FXIa is not known. However, in analogy to what is known for heparin-accelerated inhibition of thrombin by antithrombin III (39, 40, 41), several mechanisms can be postulated for the accelerated inhibition of FXIa by antithrombin III and C1 inhibitor. (i) Glycosaminoglycans may induce a conformational change in the inhibitor, rendering it more active. (ii) Glycosaminoglycans may work as a template on which inhibitor and FXIa may assemble. (iii) Glycosaminoglycans may neutralize positive charges either on the inhibitor or on the protease or both, thereby allowing inhibitor and FXIa to interact more easily. Which one of these mechanism(s) may operate in the observed glycosaminoglycan-induced acceleration of inactivation of FXIa by C1 inhibitor and by antithrombin III remains to be shown in further studies.

The effect of the various glycosaminoglycans on the inactivation of FXIa by ₁-antitrypsin and ₂-antiplasmin was in sharp contrast to that observed for C1 inhibitor. The kinetic constants for the inactivation rate of FXIa by ₁-antitrypsin and ₂-antiplasmin (Table I) were 5- and 2-fold lower, respectively, in the presence of dextran sulfate. The other glycosaminoglycans only had marginal effects, if any, on the inactivation rate of FXIa by the two protease inhibitors. An inhibiting effect of glycosaminoglycans on the interaction of ₁-antitrypsin with proteases has been reported by others, showing that heparin and other glycosaminoglycans impair the inactivation of neutrophil elastase (42) or thrombin (43) by ₁-antitrypsin.

Considering the influence of the various glycosaminoglycans on the kinetic parameters of the inhibition of FXIa by its inhibitors, as well as the relative concentrations of these inhibitors in plasma, significant changes in the distribution of FXIa between its inhibitors in plasma could be predicted (Table II). To verify these predictions, experiments with FXIa generated in or added to plasma were done. The relative contribution of the respective inhibitors to inactivation of FXIa in the presence of glycosaminoglycans was similar in both situations, indicating that they were independent of whether FXIa was added exogenously to plasma or whether it was generated endogenously. All glycosaminoglycans dramatically changed the distribution of exogenously added FXIa or endogenously generated FXIa between its inhibitors in plasma (Tables II). The observed relative contributions of the respective inhibitors to inactivation of FXIa in plasma agreed quite well with the predicted values based on the kinetic data described above (Table II). However, there are two discrepancies. In the absence of glycosaminoglycans, the contribution of ₁-antitrypsin was about half of the predicted value, whereas ₂-antiplasmin contributed substantially more than was predicted. Furthermore, the contribution of antithrombin III was distinctly higher in the presence of heparin and dermatan sulfate than was predicted based on the kinetic data. These differences between the observed and the predicted data indicate that in plasma other factors may influence the potency of the respective inhibitors, e.g. heparin-binding proteins that may partially neutralize or alter the function of glycosaminoglycans. These differences once again illustrate that one should be careful to extrapolate data obtained in purified systems to more physiological conditions, for example to plasma.

C1 inhibitor is not only the major inhibitor of FXIa but also the predominant inactivator of -FXIIa, -FXIIa, and kallikrein (44, 45). In agreement with previous work (44, 46, 47), we found the apparent second-order rate constant for inactivation of -FXIIa, -FXIIa, and kallikrein by C1 inhibitor to be 8.0, 9.8, and 25.5 × 10³ M¹ s¹, respectively. The rate of -FXIIa inactivation by C1 inhibitor was reported to be reduced 2-4-fold in the presence of dextran sulfate, kaolin, or sulfatides, whereas heparin had no effect (26, 27). Our data are in accordance herewith and, furthermore, demonstrate the rate of inactivation of -FXIIa, -FXIIa, and kallikrein by C1 inhibitor to be grossly unaffected by the presence of the physiologically occurring glycosaminoglycans heparan sulfate and dermatan sulfate.

In summary, our data indicate a profound influence of glycosaminoglycans on the inactivation of FXIa by its inhibitors, whereas such an influence was not observed on the inactivation of -FXIIa, -FXIIa, and kallikrein. Endothelial cell-associated heparin-like molecules have been postulated to modulate the activity of thrombin by enhancing the inhibition by antithrombin III (48). In analogy, although speculative at this point, we suggest that our findings also may implicate a potential physiological mechanism, i.e. that endothelial cell-associated glycosaminoglycans may modulate biological effects of contact system activation in vivo under physiological or pathophysiological conditions. It is notable that dextran sulfate stimulates activation of factor XII with subsequent formation of FXIa, whereas at the same time it enhances inactivation of FXIa by C1 inhibitor. A dextran sulfate-like contact activator in vivo consequently would activate factor XII and prekallikrein while concomitantly preventing clotting by potentiating C1 inhibitor-mediated inhibition of FXIa and not affecting other biological activities of kallikrein and activated FXII, e.g. the potential to activate the fibrinolytic system. It is interesting in this context that we recently found dextran sulfate to potentiate factor XII-dependent activation of plasminogen.²