Role of high-molecular-weight kininogen in surface-binding and activation of coagulation Factor XI and prekallikrein.

In the contact phase of activation of the kinin-forming, intrinsic clotting, and fibrinolytic systems, high-molecular-weight kininogen acts as a cofactor for the activation of Factor XI, prekallikrein, and Hageman factor. One mechanism by which high-molecular-weight kininogen acts as a cofactor has been studied by using 125I-labeled Factor XI and prekallikrein in kaolin-activated normal human plasma and plasmas deficient in high-molecular-weight kininogen and Hageman factor. High-molecular-weight kininogen was found to be essential for normal binding and cleavage of both Factor XI and prekallikrein on the kaolin surface. Hageman factor was essential for cleavage but not for binding of Factor XI and prekallikrein to kaolin. In normal plasma 80% of the activated Factor XI remained surface-bound, whereas 80% of the kallikrein was not surface-bound. These findings are consistent with the hypothesis that, in the initial phase of contact activation, high-molecular-weight kininogen links both Factor XI and prekallikrein to the exposed surface where they are activated by surface-bound activated Hageman factor. Once activated, the Factor XI molecules remain localized at the site of activation, in contrast to the kallikrein molecules which are found largely in the surrounding plasma.     

Fibrinogen blocks the autoactivation and thrombin-mediated activation of factor XI on dextran sulfate.

The intrinsic pathway of blood coagulation is activated when factor XIa, one of the three contact-system enzymes, is generated and then activates factor IX. Factor XI has been shown to be efficiently activated in vitro by surface-bound factor XIIa after factor XI is transported to the surface by its cofactor, high molecular weight kininogen (HK). However, individuals lacking any of the three contact-system proteins--namely, factor XII, prekallikrein, and HK--do not suffer from bleeding abnormalities. This mystery has led several investigators to search for an "alternate" activation pathway for factor XI. Recently, factor XI has been reported to be autoactivated on the soluble "surface" dextran sulfate, and thrombin was shown to accelerate the autoactivation. However, it was also reported that HK, the cofactor for factor XIIa-mediated activation of factor XI, actually diminishes the thrombin-catalyzed activation rate of factor XI. Nonetheless, it was suggested that thrombin was a more efficient activator than factor XIIa. In this report we investigated the effect of fibrinogen, the major coagulation protein in plasma, on the activation rate of factor XI. Fibrinogen, the preferred substrate for thrombin in plasma, virtually prevented autoactivation of factor XI as well as the thrombin-mediated activation of factor XI, while having no effect on factor XIIa-catalyzed activation. HK dramatically curtailed the autoactivation of factor XI in addition to the thrombin-mediated activation. These data indicate that factor XI would not be autoactivated in a plasma environment, and thrombin would, therefore, be unlikely to potentiate the activation. We believe that the "missing pathway" for factor XI activation remains an enigma that warrants further investigation.     

Monoclonal antibody to human high-molecular-weight kininogen recognizes its prekallikrein binding site and inhibits its coagulant activity

We developed a mouse monoclonal antibody (MoAb 115-21) to human high- molecular-weight kininogen (HK) that recognizes its prekallikrein binding site (residues 565 through 595 of HK). The corresponding synthesized 31-amino acid peptide (peptide IV) was recently shown to retain native HK's prekallikrein binding property. The same peptide bound factor XI also, although less avidly. Our MoAb recognizes purified HK, peptide IV, and the light chain moiety of HK (where the peptide IV resides), as shown by enzyme-linked immunosorbent assay (ELISA) and Western blotting experiments. The apparent dissociation constant for the HK and MoAb 115-21 interaction was 2.2 nmol/L. It does not recognize low-molecular-weight kininogen (LK) with which HK shares its heavy chain moiety or any antigens in human plasma congenitally deficient in kininogens. The binding of MoAb 115-21 to purified light chain of HK was competitively inhibited by peptide IV. In addition, the antibody inhibits HK-dependent clotting activity of normal human plasma and dextran sulfate-mediated activation of prekallikrein in plasma and retards cleavage of HK in normal plasma after contact activation with dextran sulfate. Also, purified Fab fragments of MoAb 115-21 inhibited the HK-dependent coagulant activity and dextran sulfate-mediated prekallikrein activation in normal plasma. Since the kd for HK-MoAb 115- 21 interaction is ten times lower than that of HK-prekallikrein, our data suggest that binding of MoAb 115-21 to HK's peptide IV site increases the free prekallikrein concentration in plasma and thus results in the decreased efficiency of factor XIIa-mediated activation of prekallikrein. Decreased levels of kallikrein thus formed may be responsible for the inhibition of HK-dependent clotting activity and the decrease in rate and extent of HK cleavage in normal plasma on contact activation with dextran sulfate. MoAb 115-21 may thus prove very useful, especially with its high affinity for HK, in further delineation of the role of HK and prekallikrein in contact activation and kinin-related human pathology.     

The role of HF cofactor in the Hageman factor-dependent fibrinolytic mechanism

In 1969, Ogston et al. reported that the normal activation of fibrinolysis by surface contact requires, in addition to Hageman factor and plasminogen, a HF cofactor which is present in the euglobulin fraction and other factor(s) present in the supernatant. It has also been suggested that the glass-treated plasma is deficient in HF cofactor, In our laboratory the glass-treated plasma was found not to be deficient in HF or in a streptokinase-activated proactivator or in plasminogen. The glass-treated plasma was found deficient in prekallikrein in kininogen and in clotting factors XI, IX, VIII and V. The results presented indicate that HF cofactor activity is not different from that of kallikrein and that HF cofactor does not act as a plasminogen proactivator. Furthermore, the results indicate that the "other factors' present in the supernatant are not involved in contact-activated fibrinolysis.     

Pseudo-factor-XI deficiency: effect of an inhibitor of factor XI adsorption to surface

Recently we have described a normal plasma activity that modulates contact activation by inhibiting adsorption of factor XI to activating surfaces. Here we report the first identified case in which a patient has abnormal clotting tests due to an excess of a similar activity. The patient's plasma had a prolonged partial thromboplastin time and low apparent factor XI assay. His plasma prolonged the partial thromboplastin time of normal plasma and partially neutralized normal factor XI activity in vivo and in vitro. Analysis in dilute plasma revealed normal amounts of factor XI activity and antigen. Factor XI adsorption from plasma to activating surfaces was tested by adding a small amount of 125I-labeled purified factor XI to plasma, exposing the mixture to a glass tube or kaolin, and determining the amount of factor XI adsorbed to the surface. Whereas normal plasma and plasmas deficient in factor XII, factor XI, or Fletcher factor yielded about 4% adsorption to glass, factor XI adsorption from patient's plasma was less than 1%, indicating the presence of an adsorption inhibitor. This inhibitor did not affect factor XI activation or the activity of preformed factor XIa. It was not adsorbed by AI(OH)3 and was present in serum and the macroglobulin peak on gel filtration of the plasma through Sephadex G-200. The patient's history does not allow a definitive conclusion as to whether this inhibitor was associated with abnormal bleeding.     

Interaction of high molecular weight kininogen, factor XII, and fibrinogen in plasma at interfaces

Using ellipsometry, anodized tantalum interference color, and Coomassie blue staining in conjunction with immunologic identification of proteins adsorbed at interfaces, we have previously found that fibrinogen is the main constituent deposited by plasma onto many man- made surfaces. However, the fibrinogen deposited from normal plasma onto glass and similar wettable materials is rapidly modified during contact activation until it can no longer be identified antigenically. In earlier publications, we have called this modification of the fibrinogen layer "conversion," to indicate a process of unknown nature. Conversion of adsorbed fibrinogen by the plasma was not accompanied by marked change in film thickness, so that we presumed that this fibrinogen was not covered but replaced by other protein. Conversion is now showen to be markedly delayed in plasma lacking high molecular weight kininogen, slightly delayed in plasma lacking factor XII, and normal in plasma that lack factor XI or prekallikrein. We conclude that intact plasma will quickly replace the fibrinogen it has deposited on glass-like surfaces by high molecular weight kininogen and, to a smaller extent, by factor XII. Platelets adhere preferentially to fibrinogen-coated surfaces; human platelets adhere to hydrophobic nonactivating surfaces, since on these, adsorbed firbinogen is not exchanged by the plasma. The adsorbed fibrinogen will be replaced on glass-like surfaces during surface activation of clotting, and platelets failing to find fibrinogen will not adhere.     

Changes in high molecular weight kininogen levels during and after cardiopulmonary bypass surgery measured using a chromogenic peptide substrate assay.

High molecular weight kininogen (HK) is a co-factor in the blood-contact activation system. A chromogenic peptide substrate assay for HK (HKcs) has been developed in which test plasmas are mixed with diluted HK-deficient plasma and incubated with a soluble contact system activator that activates prekallikrein and factor XII. Calcium chloride, a synthetic thrombin inhibitor and a chromogenic peptide substrate for activated factor X (FXa) are then added. The FXa generated cleaves the FXa substrate releasing p-nitroanaline, which is measured photometrically. Test plasma HK values were calculated from a standard curve generated using a pooled normal plasma. Acceptable intra-assay and inter-assay precision values were obtained and levels of HK up to 200% were measurable. The assay measured HK in plasmas deficient in factor XII, prekallikrein and factor XI, was not affected by antiphospholipid antibodies and gave an acceptable correlation (r = 0.95) when normal plasmas and mixtures of HK-deficient and normal pooled plasma, calculated to give HK levels of 25 and 50%, were compared using HKcs and a HK one-stage clotting assay. The HKcs was used to measure HK levels in seven patients undergoing cardiopulmonary bypass (CPB). HK levels fell significantly during CPB (P = 0.0014) and were significantly higher (P = 0.016) 6 days after CPB, suggesting that HK may be a positive acute-phase reacting protein.     

Regulation of the procoagulant activity within the bronchoalveolar compartment of normal human lung

The nature of the procoagulant activity of normal bronchoalveolar fluid was examined both qualitatively and quantitatively. Unconcentrated, cell-free lavage freshly obtained from normal volunteers clotted normal plasma in a mean of 84 +/- 20 s. The procoagulant activity was initiated by Factor VII-tissue factor complexes as judged by differential activity in various plasmas genetically deficient in single clotting factors, by neutralization of the procoagulant activity with antibodies to either Factor VII or tissue factor, and by a Factor X activation assay. Preincubation of the lavage with calcium was required to demonstrate Factor VII activity in unconcentrated samples. The cell-free fluid contained about 8,500 thromboplastin units/mg protein, equivalent to a third of the thromboplastin standard and indicating high amounts of cofactor. Quantitation of Factor VII was estimated by functional analysis in coagulation and amidolytic assays with reference to dilutions of normal plasma of known Factor VII concentration. When lavage and diluted plasma were adjusted to yield equivalent amidolytic activities, the average ratio of the Factor VII-clotting activity of the alveolar fluid relative to plasma Factor VII was 19 +/- 7, suggesting the presence of Factor VIIa in lavage. In contrast to previous reports with serum or activated plasma, immunoblots of concentrated lavage revealed only single-chain Factor VII, and 125I-Factor VII added to the fluid was not converted to 125I-Factor VIIa, suggesting a unique control mechanism in the lung compartment which differs from plasma. When equivalent Factor VII amidolytic activities in diluted plasma and cell-free lavage were compared, the rates of Factor Xa formation were very similar.(ABSTRACT TRUNCATED AT 250 WORDS)     

Polyphosphate is a cofactor for the activation of factor XI by thrombin

Factor XI deficiency is associated with a bleeding diathesis, but factor XII deficiency is not, indicating that, in normal hemostasis, factor XI must be activated in vivo by a protease other than factor XIIa. Several groups have identified thrombin as the most likely activator of factor XI, although this reaction is slow in solution. Although certain nonphysiologic anionic polymers and surfaces have been shown to enhance factor XI activation by thrombin, the physiologic cofactor for this reaction is uncertain. Activated platelets secrete the highly anionic polymer polyphosphate, and our previous studies have shown that polyphosphate has potent procoagulant activity. We now report that polyphosphate potently accelerates factor XI activation by α-thrombin, β-thrombin, and factor XIa and that these reactions are supported by polyphosphate polymers of the size secreted by activated human platelets. We therefore propose that polyphosphate is a natural cofactor for factor XI activation in plasma that may help explain the role of factor XI in hemostasis and thrombosis.     

Cold-induced contact surface activation of the prothrombin time in whole blood

Studies of the prothrombin time (PT) have revealed that contact with borosilicate or commercial siliconized borosilicate markedly shortens the PT. This shortening is related to the activation of the contact phase of blood coagulation. This shortening of the PT occurs in both normal whole blood and plasma when stored in borosilicate or siliconized borosilicate tubes at 4 degree C and to a lesser degree at room temperature. Studies indicated the importance of several coagulation factors in decreasing the PT. The PT did not change in blood deficient in factor XII or in plasma deficient in Fletcher factor or high molecular weight kininogen, while blood deficient in CI esterase inhibitor (CI INH) had the most profound shortening. Shortening of the PT correlated directly with increased levels of factor VII. When purified CI INH was added to normal blood, it markedly reduced the activation of factor VII and the shortening of the PT in a dose-related manner. These studies indicate the pivotal roles of the contact phase of coagulation in initiating activation of the PT and of CI INH in inhibiting the activation of the coagulation factor(s) responsible for the cold-promoted activation of factor VII.     

Structural and functional characterization of factor XII

In this article we have reviewed the current knowledge regarding the involvement of Factor XII in contact activation. Clearly in the past decade an overwhelming amount of data and hypotheses have been published regarding the central role of this zymogen in the initiation and further propagation of contact activation reactions. Therefore we feel that it will be helpful to conclude this article with a figure that summarizes those interactions and reactions that are generally believed to reflect the major molecular events occurring during surface-dependent contact activation. The contact factors are capable of very efficient interation with each other, provided a suitable negatively charged surface is present. Such surfaces are thought to stimulate the interactions between the contact factors through binding of the proteins and thus bringing the proteins together. Factor XII readily binds to the negatively charged surface, but for the binding of prekallikrein and Factor XI, the cofactor HMW kininogen is likely to be necessary. Bound at the surface, the zymogens Factor XII and prekallikrein are thought to be involved in a so-called reciprocal activation mechanism in which Factor XIIa activates prekallikrein to kallikrein, which in turn converts Factor XII to Factor XIIa. The formation of Factor XIIa is further promoted by the fact that surface-bound Factor XII is likely more susceptible to proteolytic cleavage and by the fact that the activated Factor XIIa is capable of auto-activating its own zymogen Factor XII. However, the latter effect, although undoubtedly contributing to the formation of Factor XIIa at the surface, seems to be of less importance than the reciprocal activation mechanism. This is underscored by the fact that Factor XII activation is rather slow in prekallikrein-deficient plasma. Surface-bound Factor XIIa is then responsible for the activation of Factor XI to Factor XIa, thereby propagating the initial trigger. Presumably, Factor XIa must leave the surface in order to be able to become involved in the activation of blood coagulation Factor IX.     

Assembly of contact-phase factors on the surface of the human neutrophil membrane

H-kininogen (HK), a major factor involved in contact-phase activation, was recently immunolocalized on the external surface of human neutrophils. Experiments were, therefore, designed to consider the question of whether the complete assembly of contact factors occurs on the outer surface of the neutrophil membrane. By immunolocalization techniques, and using specific antibodies directed against the various contact factors, we now demonstrate that plasma prekallikrein (PK), factor XI (FXI), and factor XII (FXII) are present on the exterior face of the human neutrophil. Failure to localize HK, PK, or FXI by monoclonal antibodies directed to their reciprocal binding sites, and displacement of PK/FXI by peptide HK31, which mimics the relevant binding site(s) of HK, suggested that prekallikrein and FXI are anchored to the neutrophil membrane through attachment to the kininogen molecule. Probing of the kinin moiety by a specific antibody showed that kininogen molecules bound to the neutrophil cell membrane contain the kinin sequence, which can be released by plasma kallikrein or by tissue kallikrein. Our results led us to the novel conclusion that neutrophils provide a circulating platform for the components of the contact-phase system.     

A unique factor XIII inhibitor to a fibrin-binding site on factor XIIIA.

An 81-year-old woman, who presented with sudden episodes of spontaneous bleeding, was found to have a specific inhibitor of factor XIII. Her fibrin clots had approximately 70% gamma-gamma and no alpha polymer formation, under conditions where normal fibrin was fully cross-linked; the patient's clots were soluble in urea or monochloroacetic acid. Factor XIII activity in her plasma was 24%, measured by the dansylcadaverine incorporation assay. When mixed with normal plasma, the patient's plasma inhibited fibrin cross-linking; however, in mixtures of patient and normal plasma, there was no inhibition of factor XIII activity when assayed by the incorporation of dansylcadaverine into casein. Thus, this inhibitor was active against fibrin cross-linking but not against ligation of small molecules to casein. Consequently, gel electrophoresis of reduced, sodium dodecyl sulfate-solubilized fibrin clots was a simple, quantitative method that was used to measure inhibitor activity. This inhibitor is unique and has been designated inhibitor New Haven. It was neutralized by anti-IgG and anti-kappa. It did not inhibit the activation of factor XIII but did inhibit fibrin cross-linking. There was complex formation between the inhibitor and activated factor XIII (A', A*) but not between A2 or fibrinogen. Only A', A* and the 56-Kd fragment bound to affinity columns made with this IgG. The inhibitor significantly decreased the binding of A', A* to fibrin clots. These data indicate that the epitope for this inhibitor is in a fibrin binding site. It is hidden in the zymogen and expressed on A' and A*, indicating that the conformational change occurring with the cleavage of the activation peptide is sufficient to expose the fibrin binding site.     

Production of factor VIII deficient plasma by immunodepletion using three monoclonal antibodies

Factor VIII deficient plasma was made from pooled, HIV antibody and hepatitis B antigen screened, normal human plasma by cryoprecipitation and immuno-depletion, using three different monoclonal antibodies bound to Sepharose columns, in series. These monoclonal antibodies are specific respectively for von Willebrand factor, factor VIII heavy chain and factor VIII light chain. The immunodepleted plasma contained less than 0.002 u/ml factor VIII coagulation activity (VIII:C) less than 0.0001 u/ml von Willebrand factor antigen and 1-2 g/l fibrinogen, while the levels of other clotting factors were unchanged. This immunodepleted plasma was compared with commercial factor VIII deficient plasma obtained from a severe haemophilia A patient as substrate in the one-stage factor VIII assay. Plasmas obtained from 20 normal subjects and 28 patients with von Willebrand's disease or haemophilia A were assayed for VIII:C using the two substrates. The results were very highly correlated (r = 0.96). The columns have high capacity and can be regenerated at least 10 times. Large-scale production of a substrate for factor VIII assays free of virus contamination is now feasible.     

Distribution of plasma fibronectin (cold-insoluble globulin) and components of the factor VIII complex after heparin-induced precipitation of plasma

Factor VIII procoagulant (VIII:C) activity, factor VIII coagulant antigen (VIII:CAg), von Willebrand ristocetin cofactor (VIIIR:RC) activity, factor VIII-related antigen (VIIIR:Ag), and plasma fibronectin (CIg; cold-insoluble globulin) were measured in the heparin precipitable fraction (HPF) and heparin supernatant fraction (HS) of normal human plasma. Following heparin induced precipitation, most measurable VIII:C activity (77% +/- 24%) was recovered in the HS. Although there was little VIII:C activity (less than 1%) in the HPF, 20% +/- 6.5% VIII:CAg was present as well as CIg (81% +/- 5.6%). VIIIR:RC activity (72% +/- 12%), and VIIIR:Ag (34 +/- 5.2%). As assessed by Na dodecyl SO4 glyoxyl agarose electrophoresis, the multimeric forms of plasma VIIIR:Ag could be resolved into a series of bands. Larger multimers tended to precipitate with the HPF whereas the smaller multimers tended to remain supernatant. Plasma from a subject with congenital afibrinogenemia was also studied. Although the afibrinogenemic HPF contained CIg, neither VIIIR:RC activity nor VIIIR:Ag was precipitated. However, both were present in the HPF from afibrinogenemic plasma to which fibrinogen had been added, suggesting that they are incorporated in this precipitate because of an affinity for fibrinogen. The ability of heparin to induce precipitation of CIg while leaving most VIII:C activity in the supernatant plasma may be useful in the preparation of procoagulant-rich plasma subfractions, since VIII:C can subsequently be recovered in good yield by cryoprecipitation.     

Activation of factor XI in plasma is dependent on factor XII [see comments]

The activation of factor XI initiates the intrinsic coagulation pathway. Until recently it was believed that the main activator of factor XI is factor XIIa in conjunction with the cofactor high molecular weight kininogen on a negatively charged surface. Two recent reports have presented evidence that in a purified system factor XI is activatable by thrombin together with the soluble polyanion dextran sulfate. To assess the physiological relevance of these findings we studied the activation of factor XI in normal and factor XII-deficient plasma. We used either kaolin/cephalin or dextran sulfate as a surface for the intrinsic coagulation pathway, tissue factor to generate thrombin via the extrinsic pathway, or the addition of alpha-thrombin directly. 125I-factor XI, added to factor XI-deficient plasma at physiologic concentrations (35 nmol/L), is rapidly cleaved on incubation with kaolin. The kinetics appear to be exponential with half the maximum cleavage at 5 minutes. Similar kinetics of factor XI cleavage are seen when 40 nmol/L factor XIIa (equal to 10% of factor XII activation) is added to factor XII-deficient plasma if an activating surface is provided. Tissue factor (1:500) added to plasma did not induce cleavage of factor XI during a 90-minute incubation, although fibrin formation within 30 seconds indicated that thrombin was generated via the extrinsic pathway. Adding 1 mumol/L alpha-thrombin (equivalent to 50% prothrombin activation) directly to factor XII deficient or normal plasma (with or without kaolin/cephalin/Ca2+ or dextran sulfate) led to instantaneous fibrinogen cleavage, but again no cleavage of factor XI was observable. We conclude that in plasma surroundings factor XI is not activated by thrombin, and that proposals of thrombin initiation of the intrinsic coagulation cascade are not supportable.     

Clot Retraction in a Factor XIII Free System

The role of Factor XIII in clot retraction was studied using the plasma from Factor XIII-deficient patients. Time course experiments revealed no significant difference in clot retraction between a plasma deficient in Factor XIII and one to which purified Factor XIII had been added. Using Factor XIII-free fibrinogen and Factor XIII-deficient platelets, it is shown that there is no significant difference in clot retraction with or without added Factor XIII.     

A glycosaminoglycan inhibitor of thrombin: a new mechanism for abnormal hemostatic assays in cancer

The isolation and partial characterization of a novel anticoagulant from the plasma of a patient with metastatic prostate cancer is described. The patient had a prolonged activated partial thromboplastic time, prothrombin time and thrombin time which did not correct by mixing with normal plasma. The reptilase time was normal and the prolonged thrombin time was corrected with protamine sulfate suggesting a heparin-like anticoagulant. A glycosaminoglycan anticoagulant (GAC) was isolated from the patient's plasma. The inhibitory activity of the GAC was destroyed by treatment with chondroitinase ABC. The GAC migrated on agarose gel electrophoresis between keratin sulfate and heparan sulfate. Purified GAC possessed only 2% (W/W) of the antithrombin III cofactor activity of porcine heparin. In assays using purified fibrinogen, the GAC was shown to directly inhibit fibrinogen proteolysis by thrombin. It is concluded that this glycosaminoglycan anticoagulant directly inhibits thrombin clotting of fibrinogen and is a new mechanism for abnormal hemostatic assays in cancer.     

The Coagulant Activity of Platelets

When platelet-rich plasma is incubated for 16–20 hours at 37° C. and the platelets are then separated and washed, tissue-factor activity develops in these platelets. The active platelets accellerate the clotting of plasma samples deficient in Factor XII, XI, IX or VIII, and of normal plasma; the coagulant activity for samples deficient in Factors V, VII and X is much less marked. The activity developed will cause activation of Factor X in a serum eluate, whereas no such activity is evident in platelets separated from plasma immediately after collection from the donor.
The development of tissue-factor activity of platelets depends on the presence of Factor XII but not on the presence of any of the other factors tested. The relationship of tissue-factor activity to platelet factor 3 is discussed.     

Contact Activation of Factor XI

S ummary . Factor XI is a circulating trace plasma protein composed of two similar or identical chains of about 80 000 daltons which upon activation undergo proteolytic cleavage. Recently, we have shown that trypsin activation leads to an active factor XI (factor XI_a) which, on reduction, yields three chains of 46 000, 37 000 and 26 000 daltons. Herein, we re-evaluate the effect of contact activation of factor XI at an activating surface both in normal human plasma and in a mixture of purified factors XI, XII, and high molecular weight kininogen (HMWK). Mixtures were analysed by coagulant activity and by reduced sodium dodecyl sulphate polyacrylamide gel electrophoresis using [¹²⁵I] factor XI. In the purified system, fully activated factor XI on reduction yielded chains of 46 000, 37 000 and 23 000 daltons. In contrast, factor XI activated by surface contact in plasma yielded on reduction only chains of 46 000 and 37 000 daltons in addition to some uncleaved 80 000 chain. We propose that factor XI_a containing only 46 000 and 37 000 chains be designated factor XI_aα, and that factor XI_a containing the third chain of 23 000 daltons be designated factor XI_aβ. Sequential elution of contact activated plasma factor XI revealed that factor XI_a was attached to the glass surface through the 46 000 dalton chain.