Proteolytic activity of contact factor zymogens

Abstract

Contact activation is triggered when blood is exposed to compounds or “surfaces” that promote conversion of the plasma zymogens factor XII (FXII) and prekallikrein to the active proteases FXIIa and kallikrein. FXIIa promotes blood coagulation by converting zymogen factor XI (FXI) to the protease FXIa. Contact activation appears to represent an enhancement of the propensity for FXII and prekallikrein to reciprocally activate each other by surface-independent limited proteolysis. The nature of the activities that perpetuate this process, and that trigger contact activation, are debated. FXII and prekallikrein, like most members of the chymotrypsin/trypsin protease family, are synthesized as single polypeptides that are presumed to be in an inactive state. Internal cleavage leads to conformational changes in the protease domain that convert the enzyme active site from a closed conformation to an open conformation accessible to substrates. We observed that FXII expresses a low level of activity as a single-chain zymogen that catalyzes prekallikrein activation in solution, as well as surface-dependent activation of prekallikrein, FXI, and FXII (autoactivation). Prekallikrein also expresses activity that promotes cleavage of kininogen to release bradykinin, and surface-dependent FXII activation. Modeling suggests that a glutamine residue at position 156 in the FXII and prekallikrein protease domains stabilizes an open active site conformation by forming hydrogen bonds with Asp194. The activity inherent in FXII and prekallikrein suggests a mechanism for sustaining reciprocal activation of the proteins and for initiating contact activation, and supports the premise that zymogens of some trypsin-like enzymes are active proteases.

1 |. INTRODUCTION

In 1964, R. Gwyn Macfarlane¹ and Earl Davie and Oscar Ratnoff² independently proposed mechanisms for blood plasma coagulation based on sequential proteolytic activation of trypsin-like enzymes. A version of their “cascade/waterfall” hypothesis used in clinical laboratories is shown in Figure 1A. In each step, an inactive protease precursor, also referred to as a proenzyme or zymogen, is converted to an active protease by internal cleavage catalyzed by another protease. The exception to this recurring theme involves activation of factor XII (FXII) in the first step of the cascade, where no activating protease or process was initially assigned.

FIGURE 1.

The cascade-waterfall model of coagulation and contact activation. (A) Cascade/waterfall model of coagulation. Coagulation is initiated by factor (F) XIIa through the intrinsic pathway (yellow arrows) or by FVIIa through the extrinsic pathway (green arrow). Coagulation proteases and their precursors are indicated in black type, with a lower case ‘a’ indicating the active form. Cofactors are indicated in red ovals. The question mark at the top of the cascade indicates uncertainty regarding the process that converts FXII to FXIIa. (B) Substances that induce contact activation in plasma. (C) FXII autoactivation. Reducing SDS-polyacrylamide gels of time courses of FXII incubated in the absence (top) or presence of 70 μM polyphosphate (60–100 phosphate units in length). The positions of standards for FXII and the heavy chain (HC) and light chain (LC) of α-FXIIa are shown on the right. The graph shows generation of FXIIa activity as measured by cleavage of a chromogenic substrate in the presence (Δ) or absence (♦) of polyphosphate. (D and E) Reciprocal activation of FXII and PK. The schematic diagrams show reciprocal FXII and PK activation in the (D) absence or (E) presence of a surface. In the presence of a surface, HK facilitates PK binding to the surface. The western blots show conversion of FXII to FXIIa (top row) and PK to PKa (bottom row) in mixtures of FXII and PK (200 nM each) either in the (D) absence or (E) presence of 70 μM polyphosphate. Positions of standards for FXII and PK, and the heavy chains (HC) and light chains (LC) of FXIIa and PKa are shown on the right of each blot. From reference 26. (F) FXI and FXII mixtures. The western blots show conversion of FXII to FXIIa and FXI to FXIa in mixtures of FXII (200 nM) and FXI (50 nM) either in the absence (no surface) or presence (with surface) of 70 μM polyphosphate. Positions of standards for FXII and FXI, and the HC and LC of FXIIa and FXIa are shown on the right of each blot. This research was originally published in Blood (Ivanov I et al¹⁴, © the American Society of Hematology)

FXIIa initiates coagulation through a series of enzymatic reactions called the intrinsic pathway (Figure 1A, yellow arrows). Coagulation through this pathway starts when plasma is exposed to a “surface.”^3–5 A variety of inorganic and organic substances, many of which carry a net negative charge, function as pro-coagulant surfaces (Figure 1B).^3–11 FXII undergoes autocatalytic conversion to FXIIa in the presence of such substances (Figure 1C),^12–14 suggesting a triggering mechanism for the cascade. However, the question of how autocatalytic activity arises from an inactive state has been a point of controversy. Early suggestions that unidentified plasma proteases catalyze initial conversion of FXII to FXIIa were countered by experiments demonstrating autoactivation of highly purified FXII. It was postulated that plasma always contains traces of FXIIa that promote autoactivation upon surface exposure.^13,15,16 This idea has parallels with the mechanism for initiating coagulation through the extrinsic pathway by the FVIIa/tissue factor complex (Figure 1A, green arrow), with 1% to 4% of plasma FVII circulating in the activated form FVIIa.^17,18 Alternatively, FXII may undergo conformational changes after surface-binding, without internal cleavage, that facilitate enzyme activity.^12,19–21 These possibilities are not mutually exclusive, and it was conceivable that different mechanisms could predominate depending on conditions.

Subsequent work showed that optimal surface-initiated FXII-dependent coagulation requires the zymogen prekallikrein (PK) and the cofactor high molecular weight kininogen (HK).^3–5,22 FXII, PK, and HK comprise the plasma kallikrein-kinin system (KKS).^4,23,24 When FXII and PK are mixed in a buffer in the absence of a surface, they reciprocally convert each other by limited proteolysis to the activated forms FXIIa and plasma kallikrein (PKa) (Figure 1D).^14,25,26 FXII autoactivation is limited when a surface is not available, leaving us with a chicken-or-egg dilemma regarding what initiates the reciprocal process. Conceivably, traces of either FXIIa or PKa present at the start of the reaction could serve as a trigger, but in our experience attempts to eliminate active protease from FXII and PK preparations have little impact on the activation rates.

In plasma, reciprocal activation of FXII and PK is restrained by C1 inhibitor (C1-INH), a serpin that inhibits both FXIIa and PKa.^27–29 Addition of a surface overcomes C1-INH inhibition, allowing generation of sufficient FXIIa to initiate coagulation through activation of FXI.^30,31 HK enhances this process by facilitating PK and FXI binding to the surface (Figure 1E). The surface-dependent reactions that trigger the coagulation cascade are collectively referred to as contact activation.^3–5,22 Models of contact activation explain the importance of FXII, PK, HK, and FXI to plasma coagulation. However, as with the original cascade/waterfall scheme, they leave open the question of how FXIIa or PKa activity first arise in plasma under initial conditions where neither activity is presumed to be present. Here, we discuss recent work showing that the zymogens FXII and PK express proteolytic activity that could sustain basal reciprocal activation of FXII and PK in the absence of a surface, and trigger contact activation during surface-induced plasma coagulation.^14,26,32,33

2 |. ACTIVATION OF TRYPSIN-LIKE PROTEASES

FXII and PK, like other enzymes in the coagulation cascade, are members of a large class of serine proteases with homology to the pancreatic degradative proteins chymotrypsin and trypsin.^34,35 Proteases with chymotrypsin/trypsin folds are usually secreted as single polypeptides, with their catalytic active sites in a closed conformation, as shown in Figure 2A for the coagulation zymogen prothrombin.^35–38 Active sites for this class of protease typically contain a catalytic triad consisting of His57, Asp102, and Ser195 (chymotrypsin numbering) at the entrance to a substrate binding pocket located at the interface between two β-barrels. In 1978, Robert Huber and Wolfram Bode proposed that zymogens of chymotrypsin/trypsin-like proteases are irreversibly converted to active enzymes by cleavage after an internal arginine, usually Arg15.³⁶ The structural changes that follow this cleavage are illustrated in Figure 2B, which shows the active site structure for α-thrombin, the activated form of prothrombin. Cleavage after Arg15 creates a new N-terminus (Ile16 in the case of α-thrombin) for the protease domain that forms a salt bridge with the side-chain carboxylate group of the conserved Asp194. This allows formation of the protease oxyanion hole (indicated in Figure 2B by juxtaposed blue spheres representing the nitrogen atoms of Gly193 and Ser195) and the S1 substrate specificity pocket (indicated in cyan by the side [R] group of the P1 residue of a hypothetical substrate).^34,36,37 This iconic model elegantly explains the onset of protease activity, and the behavior of proteases in cascades. However, it does not account for certain behaviors, such as the capacity of some zymogens, including FXII, to autoactivate. Furthermore, the model does not readily explain how a series of protease reactions is initiated if all constituents start as inactive zymogens.

FIGURE 2.

Structures for trypsin-like serine proteases. Shown are stick diagrams of the active sites of select trypsin-like proteases showing the positions of key amino acids, the oxyanion holes, and S1 specificity pockets. Amino acids are indicated with the numbering system used for chymotrypsin. Key hydrogen bonds are indicated by dotted lines (magenta), with salt bridges indicated by “+” and “−” symbols. The positions of the nitrogen atoms of the conserved residues Gly193 and Ser195 that form parts of the oxyanion hole are indicated by blue spheres. In structures with open active site conformations the P1 residues of a substrate or inhibitor are shown in cyan. All figures were prepared with PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC). This research was originally published in Blood (Ivanov I et al¹⁴, and Emsley J⁶⁰, © the American Society of Hematology). (A) Prothrombin and (B) α-thrombin. A comparison of the structures shows the salt bridge that forms between Ile16 and Asp194 after cleavage of the peptide bond between Arg15 and Ile16. Formation of the protease oxyanion hole is indicated in panel B by juxtaposition of the spheres representing Gly193 and Ser195. (C) Single-chain tPA active S1 pocket crystal structure (pdb:1BDA) is shown with Asp194 stabilized by a salt bridge formed with Lys156, and by hydrogen bonds with the main-chain nitrogens of Gly142 and Cys191. The cyan stick figure represents the side-chain of the P1 arginine of the tPA inhibitor dansyl-Glu-Gly-Arg-chloromethylketone. (D) Crystal structure (pdb:4XDEP) showing the inactive zymogen conformation of FXII catalytic domain. Note that the oxyanion hole is not formed. (E) Homology model (SWISS-MODEL) of the S1 pocket of FXII-T based on the tPA crystal structure (pdb:1BDA) with Gln156 forming a hydrogen bond with the Asp194 carboxylate group. A superposition of the FXII-T model with the tPA structure was performed with PyMOL and the calculated root-mean-square deviation (RMSD) was 0.092 Å. The side chain in cyan represents the P1 arginine of a substrate (PK or FXI). (F) Homology model of the S1 pocket of PK-R371A based on the tPA crystal structure (pdb:1BDA), with Gln156 forming a hydrogen bond with the Asp194 carboxylate group (RMSD 0.160 Å). The side chain shown in cyan represents the P1 arginine of a substrate (HK or FXII)

Chakraborty and colleagues have proposed an alternative interpretation for activity of trypsin-like proteases based on their work with prothrombin and α-thrombin.³⁸ They noted that the protease domain active sites in crystal structures of these proteins may adopt an “ensemble” of conformations with varying active site accessibility. They suggest that zymogen and protease forms, rather than representing distinct states with or without activity, are points on a continuum of active site conformations. In support of this, structures for both prothrombin and thrombin may display open active site conformations (E form) accessible to a substrate or closed inaccessible conformations (E*).^39,40 This implies that equilibria exist between E and E* forms in both zymogen and active forms of a protease. For most zymogens, the catalytic site is primarily in the E* form. Structural changes accompanying cleavage after Arg15 result in the E form predominating, and correspond to full activity. However, depending on the protease, a zymogen may express significant activity if a sufficient fraction adopts the E form, either in an unbound state or after substrate binding. Two proteins closely related to FXII, tissue plasminogen activator (tPA) and urokinase plasminogen activator (uPA), express significant activity in their single-chain forms,^41–43 a property referred to as “low zymogenicity.” Based on the behaviors of FXII and PK, we hypothesized that these zymogens also express activity that could promote reciprocal activation and trigger contact activation.

3 |. FACTOR XII ACTIVITY

F12, the gene encoding FXII, arose from a duplication of HGFAC, the gene for the repair protease pro-hepatocyte growth factor activator (pro-HGFA).^44–46 From an evolutionary perspective, the catalytic domains of FXII and pro-HGFA are closely related to those of tPA and uPA.^45–48 In single-chain tPA and uPA, Lys156 (chymotrypsin numbering) forms a salt bridge with Asp194, and hydrogen bonds with Gly142 and Cys191, maintaining the active site in an open (E) conformation (Figure 2C),^42,43 and explaining the low zymogenicity of the single-chain forms of these proteases. The corresponding residue in FXII, Gln156 (chymotrypsin numbering), cannot form a salt bridge with Asp194 and, therefore, should be less effective than Lys156 for stabilizing the E conformation. Indeed, a crystal structure for the FXII catalytic domain indicates a closed (E*) conformation (Figure 2D).⁴⁸ However, modeling indicates that the carboxylate group of Asp194 in FXII can form the same hydrogen bonds as Asp194 in tPA, and an additional hydrogen bond with the side-chain of Gln156, removing impediments to oxyanion hole formation (Figure 2E).¹⁴ This could permit the FXII active site to adopt an open conformation and express activity.

During FXII activation by PKa or by autoactivation, the protein is cleaved after Arg353 (FXII numbering, corresponding to Arg15 in chymotrypsin) to form α-FXIIa, a protease containing a 50-kDa heavy chain and a 30-kDa light chain (protease domain) connected by a disulfide bond (Figures 1C–E and 3A).^14,45,49,50 An additional cleavage of unknown significance occurs after Arg343 and, with time, cleavage after Arg334 releases the heavy chain to form the truncated species β-FXIIa (Figure 3A).^14,49,50 A problem with assessing activity intrinsic to FXII derived from plasma is that contamination with α-FXIIa or β-FXIIa may give the false impression the zymogen has activity. To address this, we replaced Arg334, Arg343, and Arg353 in FXII with alanine residues.¹⁴ The resulting protein, designated FXII-T (for triple mutant, Figure 3A), unlike wild-type FXII (FXII-WT), cannot be converted to FXIIa by autoactivation or through catalysis by PKa (Figure 4A, middle row). We also prepared FXII with the active site serine replaced by alanine (FXII-S544A, Figure 3A). FXII-S544A is cleaved after Arg353 by PKa allowing it to assume a FXIIa-like conformation (Figure 4A, bottom row), but it lacks activity because of the absence of the catalytic triad serine residue (Figure 4B).

FIGURE 3.

Factor XII and prekallikrein constructs. Schematic diagrams showing noncatalytic (white) and catalytic (gray) domains of FXII and PK. Positions of active site serine residues are indicated by green bars. Sites of proteolysis during activation are indicated by green arrows, with solid arrows indicating sites of cleavage required for conversion to FXIIa and PKa. (A) FXII may be cleaved at three locations. Cleavage after Arg353 converts FXII to αFXIIa, whereas cleavage of α-FXIIa after Arg334 forms β-FXIIa. The FXII noncatalytic domains are the fibronectin type 2 (F2), epidermal growth factor (EGF), fibronectin type 1 (F1), and kringle (K) domains, and a proline-rich region (PRR). FXII-T, a form of FXII that cannot be converted to αFXIIa or βFXIIa was created by replacing Arg334, Arg343, and Arg353 with alanine residues (indicated by red arrows). FXII-S544A was created by replacing the active site residue Ser544 with alanine (red bar). (B) PK is cleaved after Arg371 to form PKa. The noncatalytic portion of PK contains four apple domains, designated A1 to A4. PK-R371A is a form of PK that cannot be converted to PKa. It was created by replacing Arg371, with alanine (indicated by red arrow). PK-S559A was created by replacing the active site residue Ser559 with alanine (red bar). Images originally published in Blood (Ivanov I et al¹⁴, and Ivanov I, et al³³, © the American Society of Hematology)

FIGURE 4.

Single-chain FXII (FXII-T) activity. (A) Western blots of time courses of recombinant FXII species (200 nM) incubated with vehicle (control, left), 70 μM polyphosphate (60–100 orthophosphate units in length, center), or 50 nM α-kallikrein (right). A polyclonal goat anti-human FXII IgG was used to develop the blots. Positions of standards for FXII and the heavy chain (HC) and light chain (LC) of αFXIIa are indicated on the right. (B) FXII (200 nM) was incubated with 50 nM PKa for 120 min at 37°C. Cleavage of the substrate S-2302 (200 μM) was measured on a spectrophotometer. (C) Western blots of times courses of recombinant FXII species (200 nM) and PK (200 nM) incubated at 37°C. Polyclonal goat anti-human FXII IgG (top) or sheep anti-human PK (bottom) were used to develop the blots. Positions of standards for FXII and PK and the HC and LC of αFXIIa and PKa are indicated on the right. (D) Kinetic parameters for activation of PK by αFXIIa or FXII-T. (E) Conversion of PK (200 nM) to PKa in the presence or absence 70 μM polyphosphate, and in the presence or absence 200 nM FXII-T (T) was followed by continuous monitoring of S-2302 (200 μM) cleavage. (F) Western blots of a mixture of FXII-T (200 nM) and FXII-S544A (200 nM) in the absence of a surface (left), or in the presence of 70 μM polyphosphate (right). The middle panels show results for FXII-S544A and FXII-T incubated separately with 70 μM polyphosphate. Blots were developed using IgG D06 which recognizes the FXIIa active site. (G) FXI-S557A (30 nM) incubated with 200 nM FXII-WT, FXII-T, or FXII-S544A in the absence (top) or presence (bottom) of 70 μM polyphosphate. Blots were developed with goat anti-human FXI polyclonal IgG. Positions of standards for FXI and the HC and LC of FXIa and PKa are indicated on the right. This research was originally published in Blood (Ivanov I et al¹⁴, © the American Society of Hematology)

Western blots in the left-hand column of Figure 4C show reciprocal conversion of FXII and PK to FXIIa and PKa in the absence of a surface. It is reasonable to assume that FXIIa catalyzes PK activation during this process and, indeed, PK cleavage does not occur if FXII-WT is replaced with FXII-S544A (Figure 4C, right-hand column). Interestingly, FXII-S544A is also not cleaved during the reaction, suggesting FXII(a)-mediated conversion of PK to PKa, and not PK(a) activation of FXII, initiates reciprocal activation in solution (more on this in the following section). In contrast to reactions with FXII-S544A, PK is converted to PKa by FXII-T (Figure 4C, middle column), despite the inability of FXII-T to form FXIIa (note FXII-T is not cleaved during the reaction). Similar results were obtained with FXII in which Arg334, Arg343, and Arg353 are replaced with glutamine. FXII-T activates PK with a catalytic efficiency that is ~ 50,000-fold lower than for WT FXIIa (Figure 4D), consistent with the impression from modeling that Gln156 is less effective than tPA Lys156 for stabilizing an open active site.⁴² Nevertheless, the ability of FXII-T to convert PK to PKa provides a plausible mechanism for initiation of reciprocal activation.

During surface-induced contact activation, FXIIa catalyzes (a) rapid conversion of PK to PKa, (b) conversion of FXII to FXIIa (autoactivation), and (c) conversion of FXI to FXIa.^3–5,30,31 In the following experiments orthophosphate polymers (polyphosphate) were used as a contact surface.^10,11,51 Specifically, we used relatively short chain polyphosphates containing 60 to 100 phosphate units, which are similar to those in platelet dense granules that are released upon platelet activation.^10,51 Polyphosphate accelerated PK activation by FXII-T (Figure 4E),¹⁴ indicating surface binding alters the conformation of PK, FXII-T, or both to facilitate PK conversion to PKa. FXII-Locarno, a naturally occurring FXII variant in which Arg353 is replaced with proline, also displays activity in this assay, although the activity is considerably lower than that of FXII-T, probably because the proline substitution causes alterations in the activation loop structure that affect active site conformation.⁵²

When assessing FXII-T activation of FXII, we cannot use FXII-WT as the substrate because it will autoactivate in the presence of polyphosphate (Figure 4A, top row).^14,32 Instead, we used FXII-S544A because it possesses no activity once cleaved after Arg353 and therefore cannot autoactivate. In the western blots in Figure 4F, the presence of cleaved FXII-S544A is detected with an antibody specific for FXIIa.²⁵ No FXIIa signal is generated when FXII-S544A is incubated with FXII-T in the absence of a surface, consistent with the impression that autoactivation is limited without a surface. Incubation of FXII-S544A or FXII-T separately with polyphosphate does not generate a signal because the former lacks activity and the latter cannot form FXIIa. However, when FXII-S544A and FXII-T are incubated together with polyphosphate, a FXIIa signal is detected, indicating the proteolytic activity of FXII-T can convert FXII-S544A to FXIIa-S544A. Similar results were obtained using silica as a surface.¹⁴

FXIIa drives thrombin production by activating FXI.^30,31 FXI and PK are homologs,⁴⁴ and their polypeptides are similarly organized.^31,53–58 However, unlike mixtures of FXII and PK, mixtures of FXII and FXI do not undergo reciprocal activation without a surface (Figure 1F). Consistent with this, neither FXII-WT nor FXII-T convert FXI to FXIa in the absence of polyphosphate (Figure 4G, top row). Because FXI autoactivates on surfaces, FXI lacking the active site serine residue (FXI-S557A) was used as a substrate for experiments with polyphosphate. In these reactions, FXII-WT converts all FXI-S557A to FXIa-S557A after a lag phase, most consistent with autoactivation of FXII to FXIIa, followed by FXIIa cleavage of FXI. In contrast, FXII-T generates a small amount of FXIa in a time-dependent manner, consistent with the intrinsic activity noted in experiments using PK and FXII-S544A as substrates. Taken as a whole, these data demonstrate that zymogen FXII expresses sufficient activity to convert PK to PKa in the absence of a surface; and FXII to FXIIa and FXI to FXIa in the presence of a surface. The intrinsic activities are several orders of magnitude lower than for FXIIa, suggesting that they are most likely relevant early in the course of contact activation when the FXIIa concentration is low.

4 |. PREKALLIKREIN ACTIVITY

During contact activation, FXIIa cleaves PK after Arg371 (PK numbering, corresponding to chymotrypsin Arg15) to form PKa (Figure 3B).^53,55 PKa has two major activities in plasma: (a) cleavage of HK to liberate bradykinin and (b) conversion of FXII to FXIIa.^3–5 Joseph and coworkers first suggested that PK expresses proteolytic activity independent of conversion to PKa.⁵⁹ They observed that plasma PK cleaves HK by a process that is inhibited by C1-INH. As in FXII, residue 156 (chymotrypsin numbering) in the PK protease domain is glutamine, and modeling based on the tPA structure indicates that PK Gln156 can coordinate with the Asp194 carboxylate group through hydrogen bonding to stabilize an open active site conformation (Figure 2F).⁶⁰ As in experiments with FXII, a consideration when studying activity in plasma-derived PK is that it may be contaminated with PKa that contributes to the observed activity. Using a similar strategy to that used for FXII, we prepared a nonactivatable form of PK by replacing Arg371 with alanine (PK-R371A, Figure 3B), and PK in which the active site serine is replaced with alanine (PK-S559A, Figure 3B), and tested their capacities to cleave HK and FXII.³³

PKa cleaves HK at two locations, releasing the nine amino acid peptide bradykinin, and forming the heavy and light chains of cleaved kininogen (HKa, Figure 5A).^{3–5,49,61,62} A similar cleavage pattern occurs when HK is incubated with WT PK (PK-WT) or PK-R371A, but not with PK-S559A (Figure 5B).^33,59 The reactions are inhibited by C1-INH, consistent with the notion that the protease domain of PK can adopt an open conformation. The catalytic efficiency of HK cleavage by PK-R371A is ~ 1500-fold lower than for PKa (Figure 5C).³³

FIGURE 5.

Single-chain PK (PK-R371A) activity. Western blots of time courses of cleavage of HK (200 nM) by (A) α-kallikrein (2 nM) or (B) recombinant PK species (200 nM). (B) “Vehicle” indicates a reaction without PK. Some reactions were run in the presence of 750 nM C1 esterase inhibitor (C1INH). Blots were developed with a polyclonal goat anti-human HK IgG. Positions of standards for uncleaved HK (HK) and heavy and light chains of cleaved HK (HKa) are indicated at the right of each image. (C) Kinetic parameters for cleavage of HK by α-kallikrein or PK-R371A derived from curves generated by measuring bradykinin release from HK by ELISA. (D) Western blots of time courses of cleavage of FXII-S544A (200 nM) by recombinant PK-WT, PK-R371A or PKS559A in the absence (top) or presence (bottom) of 70 μM polyphosphate (60–100 orthophosphate units in length). Blots were developed using IgG D06, which recognizes formation of the FXIIa active site. Positions of a standard for FXIIa are shown on the right. (D) Kinetic parameters of FXII-A544A cleavage by PKa or PK-R371A in the presence of polyphosphate were determined from curves of FXII-S544A cleavage measured by ELISA using the FXIIa-specific antibody D06 as a capture antibody. This research was originally published in Blood (Ivanov I et al³³, © the American Society of Hematology)

FXII activation by PK-R371A was studied using FXII-S544A as the substrate to avoid contributions from FXII autoactivation.³³ In contrast to FXII-T conversion of PK to PKa (Figure 4C), neither PK-WT nor PK-R371A converts FXII-S544A to FXIIa without a surface (Figure 5D, top row). However, both PK-WT and PK-R371A cleave FXII-S544A in the presence of polyphosphate (Figure 5D, bottom row), indicating binding to polyphosphate induces conformational changes in the substrate (FXII-S544A) and/or the protease (PK-R371A) that facilitate conversion of FXII to FXIIa. Work from de Maat and colleagues⁶³ and Ivanov and coworkers²⁶ indicates that the heavy chain of unbound FXII restricts access to the activation cleavage site at Arg353, and that surface binding relaxes this inhibition. This may explain the inability of single-chain PK to activate FXII without a surface. The specific activity for FXII activation by PK-R371A is ~5 orders of magnitude weaker than for PKa. As with FXII-T then, activity intrinsic to single-chain PK would probably be important early in contact activation when the PKa concentration is very low.

5 |. CONCLUSIONS AND FUTURE CONSIDERATIONS

The results discussed here show that FXII and PK are active proteases, suggesting a mechanism for initiating and maintaining reciprocal activation of FXII and PK in plasma, and for triggering surface-dependent contact activation.^14,32,33 To make an analogy with an automobile motor, FXII and PK activity allow the KKS to idle at a low rate. This turnover in low-gear facilitates rapid shifting to higher gears through contact activation under appropriate conditions.⁶⁰ Althoughy the activities inherent in FXII and PK imply that preexisting FXIIa or PKa are not required for initiating reciprocal activation either in solution or on a surface, it is likely that these proteases are continuously generated in plasma by the proposed mechanism. This hypothesis is supported by work from Revenko and colleagues, who noted that reducing the plasma concentration of either FXII or PK in healthy mice results in a decrease in the plasma concentration of the activated form of the other protein (PKa or FXIIa, respectively).⁶⁴

The specific activities of FXII and PK relevant to reciprocal activation are more than 10,000 times lower than for FXIIa and PKa, respectively, a difference much greater than for prototypical enzymes demonstrating low zymogenicity such as tPA and uPA. This may be due to low activity in the open forms of the zymogens, a relatively low fraction of the zymogens adopting open conformations, or a combination of these possibilities. FXII-S544A and PK-S559A, versions of FXII and PK with alanine substitutions for the active site serine residue, could be used to determine the fractions of the proteins that are in open conformations using rapid kinetics approaches as described by Chakraborty et al,³⁸ or by active site titration with fluorescently labeled inhibitors as proposed by Pozzi and colleagues.⁶⁵

The low specific activities of FXII and PK suggest that it is the activated forms, FXIIa and PKa, which are responsible for the proteins’ physiologic and pathologic effects. This is almost certainly the case for contact activation-induced coagulation, as demonstrated by the inability of FXII-T to restore thrombus formation in FXII-deficient mice in a carotid artery occlusion model,¹⁴ and the low capacities of FXII-T or PK-R371A to correct defects in surface-induced clotting in plasma lacking FXII or PK, respectively.^14,33 However, the situation may be different for PK generation of bradykinin. In normal health, BK is generated at a low but measurable rate.⁶⁶ The basal rate of PKa generation is likely to be relatively low as well,⁶⁴ whereas the ratio of PK to PKa is high. Under such conditions, PK cleavage of HK, which is ~1500-fold less efficient than the reaction catalyzed by PKa, may contribute significantly to basal bradykinin formation. The inefficiency of the reaction suggests that PKa is required for substantial increases in BK generation above baseline, consistent with the impression that elevated BK formation in patients with hereditary angioedema (HAE) ^29,67,68 or sepsis ^69,70 involves PK conversion to PKa.

HAE is characterized by recurrent episodes of bradykinin-induced soft-tissue edema that is most often the result of dysregulated PKa activity and/or PK activation.^29,67,68 Most cases are associated with reduced plasma levels of C1-INH, the main regulator of PKa and FXIIa. Although proteins involved in contact activation are clearly active in most forms of HAE, it is not clear that PK and FXII activation is surface-dependent. Indeed, in one form of the disorder associated with normal C1-INH levels, pathologic activation of PK and FXII is most likely not surface-dependent because the causative mutation results in a truncated gain-of-function FXII species that lacks the surface-binding heavy chain.^26,63 It seems reasonable to postulate that inability to maintain basal reciprocal activation of FXII and PK within an acceptable range (regardless of whether or not a surface is required) could lead to increased turnover of these proteins, lowering the threshold for pathologic bradykinin generation in HAE.

Modeling of FXII and PK supports earlier observations with tPA and uPA^42,43 that indicate residue 156 influences zymogen active site conformation in chymotrypsin/trypsin-like proteases.^14,32,33,60 Mutagenesis experiments have shown that the salt bridge formed between Lys156 and Asp194 is central to activity in single-chain tPA and uPA.^41–43 Residue 156 is glutamine in several protease zymogens including FXII, PK, FXI, factor IX, prothrombin, neutrophil elastase, and chymotrypsinogen itself.⁶⁰ Modeling of the FXII and PK protease domains based on the tPA structure suggests that Gln156 can coordinate with the Asp194 carboxylate to weakly stabilize an open active site conformation through hydrogen bonding.^14,60 In contrast, the large hydrophobic side chain of Met156 in factor VII and factor VII activating protease (an enzyme with some homology to FXII and pro-HGFA) would not be expected to stabilize an open active site conformation in the zymogen.^60,71 Indeed, Met156 may contribute to the zymogen-like behavior of the activated form of FVII (FVIIa) when it is not bound to its cofactor tissue factor. Petrovan and Ruf replaced Met156 in factor VIIa with glutamine and noted increased protease activity in the absence of tissue factor.⁷² Residue 156 is not likely to be the sole determinant of low zymogenicity. Factor X and pro-HGFA have not been shown to express zymogen activity, despite having a basic residue at position 156 (Lys156 and Arg156, respectively) similar to tPA and uPA. FXI, a protease homologous to PK, does not appear to express activity despite having glutamine at position 156.³³ Our models for the FXII and PK active sites need to be tested with mutagenesis-based experiments to verify our hypothesis that Gln156 is required for zymogen activity in FXII and PK. It will be interesting to see how predicted loss-of-function substitutions such as Ala156, and potential gain-of-function substitutions such Lys156, affect the activity of the two zymogens.

The observations that FXII and PK express protease activity have implications for developing therapeutics. Antibodies and small molecule inhibitors that reduce KKS activity for treatment or prevention of angioedema, inflammation, or thrombosis are typically designed to interact with the FXIIa or PKa active sites.^73–76 It will be interesting to test these agents for their abilities to neutralize the activities of zymogen FXII and PK. A compound that inhibits zymogen activity could down-regulate basal reciprocal activation of FXII and PK, limiting the generation of the activated protease forms that would need to be inhibited by the drug. To use another analogy, drugs that effectively inhibit the zymogen active sites of FXII and/or PK in their open conformations could prevent the reciprocal snowball from forming in the first place, rather than trying to stop it once it is rolling downhill. Such compounds may have an advantage over agents that interact with FXIIa and PKa only. We observed that the monoclonal antibodies 559C-X181-D06 directed against the active site of FXIIa²⁵ and 559A-M202-H03 directed against the active site of kallikrein⁷⁷ block activity intrinsic to FXII-T^14,26 and PK-R371A,³³ respectively. However, it is conceivable that some compounds may interact differently with the zymogen and activated forms of a protease, and this may need to be taken into consideration during drug design.