Factor XIIIa Inhibitors as Potential Novel Drugs for Venous Thromboembolism

Abstract

Human factor XIIIa (FXIIIa) is a multifunctional transglutaminase with a significant role in hemostasis. FXIIIa catalyzes the last step in the coagulation process. It stabilizes the blood clot by cross-linking the α- and γ-chains of fibrin. It also protects the newly formed clot from plasmin-mediated fibrinolysis, primarily by cross-linking α₂-antiplasmin to fibrin. Furthermore, FXIIIa is a major determinant of clot size and clot’s red blood cells content. Therefore, inhibitors targeting FXIIIa have been considered to develop a new generation of anticoagulants to prevent and/or treat venous thromboembolism. Several inhibitors of FXIIIa have been discovered or designed including active site and allosteric site small molecule inhibitors as well as natural and modified polypeptides. This work reviews the structural, biochemical, and pharmacological aspects of FXIIIa inhibitors so as to advance their molecular design to become more clinically relevant.

Graphical abstract

1. Factor XIIIa (FXIIIa): Structure and Function

Human FXIIIa is a transglutaminase (TG) that has multiple extracellular and intracellular biological roles. In addition to FXIIIa, the TG family includes seven other members (TG1 – TG7), many of which have been identified with a potential enzyme catalytic activity. Interestingly, only a limited sequential similarity can be found among TGs’ primary structures, yet a significant similarity exists among their secondary structures [1–8]. Importantly, FXIIIa and its precursor factor XIII (FXIII) have been the subject of extensive research over the last two decades.

Structurally, the cellular form of FXIII is a homodimer of two A subunits (FXIII-A₂) (Figure 1A). However, the plasma form of FXIII is a tetramer of two potentially active A subunits and two inhibitory/carrier B subunits (FXIII-A₂B₂). In one hand, the A subunit is a single, non-glycosylated polypeptide chain with a molecular weight of 83 kDa. Without the initiator methionine, it is of 731 amino acids with nine cysteine residues, none of them is involved in the formation of disulfide bonds. Cys314 represents the active site residue. The A subunit contains a central core domain (residues 184 – 515) with the catalytic triad, an N-terminal β-sandwich domain (residues 38 – 183), and two β-barrel domains (residues 516−627 and 628−730). Besides, it also possesses an N-terminal activation peptide (residues 1 – 37), which is cleaved off during the activation process. In the other hand, the B subunit is a glycosylated peptide of 10 sushi domains held together by disulfide bonds. It is of 641 amino acids and a molecular weight of 73 kD, without its carbohydrate content. The structural elements of the A subunit involved in the non-covalent interaction with the B subunit do not appear to be fully understood, nevertheless, the first sushi domain of the B subunit appears to be important for complex formation. The A subunit is predominantly biosynthesized by monocytes/macrophages, megakaryocytes/platelets, chondrocytes, and osteoblast/osteocytes. However, hepatocytes have also been reported to express low levels of the A subunit. The B subunit is biosynthesized and secreted by hepatocytes. In the circulatory system, the tetramer FXIII-A₂B₂ is bound to the fibrinogen ɣ-chain through its B subunits [1–8].

Figure 1.

A) A cartoon representation of the crystal structure of recombinant human cellular FXIIIA₂ (PDB ID: 1F13). FXIII-A contains a central core domain (residues 184 – 515) (green) with Cys314 and the catalytic triad (yellow), an N-terminal β-sandwich domain (residues 38 – 183) (raspberry), and two β-barrel domains (residues 516−627 and 628−730) (orange and cyan). It also possesses an N-terminal activation peptide (residues 1 – 37) (blue), which is cleaved off during the activation process. B) A cartoon representation of the proteolytic activation of the tetramer form of FXIII i.e. FXIII-A₂B₂ by thrombin and then low concentration of Ca²⁺. Thrombin facilitates the removal of the activation peptide from the inactive tetramer (FXIII-A₂B₂) to form FXIII-A₂*B₂, followed by Ca²⁺ which facilitates the removal of B subunits resulting in the formation of the active dimer FXIII-A₂** (or simply FXIIIa).

The activation of plasma and cellular forms of FXIII to their activated forms can proceed proteolytically or non-proteolytically. For the plasma form, the activation process involves the conversion of the tetramer form of FXIII i.e. FXIII-A₂B₂ to the dimer form of FXIII-A₂** (presented throughout this paper as FXIIIa) (Figure 1B). During the proteolytic activation process, thrombin first cleaves the activation peptide from the A subunits (the peptide bond between Arg37 and Gly38) resulting in the formation of the intermediate form of FXIII-A₂*B₂. Then in the presence of Ca²⁺, several conformational changes take place leading to the B subunits dissociating and resulting in the intermediate form being transformed into the active form of FXIII-A₂** i.e. FXIIIa. Importantly, the rate of activation is highly enhanced in the presence of fibrin. Although it appears that the proteolytic activation of FXIII-A₂B₂ is well established, non-proteolytic activation may also occur in the presence of a high concentration of Ca²⁺. Likewise, thrombin and Ca²⁺ are important elements in activating the cellular form of FXIII [2].

The catalytic activity of FXIIIa is attributed to its active site which contains the catalytic triad of Cys314, His373 and Asp396. FXIIIa exploits a double displacement mechanism for cross-linking proteins via the formation of an ε-(γ-glutamyl) lysine iso-peptide bond (Figure 2A). The mechanism has also been reported as a modified ping-pong mechanism which is shared among TGs. In the first stage, the thiol group of Cys314 residue attacks the glutamine γ-carboxyamide group (also labelled as the acyl-donor or the amine-acceptor) of a specific substrate protein, displacing ammonia molecule and forming a thioester intermediate (Figure 2B). In the second stage, the reactive thioester intermediate is attacked by the ε-amino group of a lysine residue (also labelled as the acyl-acceptor or the amine-donor) of another specific substrate protein. This attack displaces the Cys314 residue and leads to the formation of an iso-peptide bond between the two substrate proteins as well as to the release of FXIIIa. In the absence of lysine residues, water reacts with the thioester intermediate which converts glutamine into glutamic acid and releases the free enzyme [9, 10].

Figure 2.

A) A depiction of the ping-pong mechanism or the double displacement mechanism of FXIIIa-mediated crosslinking mechanism. B) The catalytic cycle of FXIIIa involving Cy314, His373, and Asp396, as described before in reference [9]. Colors and labels are to further illustrate the process of forming iso-peptide product as a result of FXIIIa action.

Physiologically, a recent proteomic approach combined with TG-specific labeling identified 147 potential substrates for FXIIIa [7]. The physiological substrates include coagulation substrates, fibrinolytic proteins, adhesive/matrix proteins, and cytoskeletal proteins, among others [1–7]. Accordingly, several studies have presented results to significantly link FXIIIa to pregnancy maintenance [11], wound healing [4,12], immunity [13], adipogenesis [14,15], and angiogenesis [16]. Studies have also indicated its cardioprotective effect [17,18] as well as its potential contribution to vascular permeability [19,20] and stabilization and mineralization of extracellular matrix in cartilage and bone [21–23]. Very recently, a study showed that FXIIIa-mediated cross-linking may contribute to the formation of amyloid β deposits in cerebral amyloid angiopathy and Alzheimer’s disease [24]. Importantly, the most recognized function of FXIIIa is as a blood coagulation factor. Fibrin α- and γ-chains, factor V, α₂-antiplasmin, plasminogen, plasminogen activator inhibitor-2, and thrombin-activable fibrinolysis inhibitor are all substrates for FXIIIa [2,4]. Therefore, FXIIIa is very important for hemostasis and is found to be a critical determinant of clot properties. In fact, it is well reported that FXIIIa is essential for maintaining hemostasis by stabilizing the fibrin clot and protecting it from fibrinolytic degradation [25–30]. Accordingly, FXIII(a) has been implicated in the risk of atherothrombotic diseases and venous thromboembolism (VTE) [31–33]. Its deficiency results in bleeding diathesis and patients with FXIII deficiency usually need substitution therapy [34,35].

Given its multiple biological roles, FXIIIa has become a potential target to design and develop new drugs for pharmacological intervention. This review focuses on the design and development of FXIIIa inhibitors as anticoagulants for the treatment of thrombosis, particularly VTE.

2. FXIIIa: An emerging drug target for new anticoagulants

Generally, thromboembolic conditions have been estimated to attribute to 25% of deaths worldwide. Thromboembolic conditions can be either arterial or venous conditions. Ischemic heart disease and ischemic stroke are the major types of arterial thrombosis. Deep vein thrombosis and pulmonary embolism are the main conditions of VTE. In fact, data suggests that VTE continues to be the third most common cardiovascular disease after coronary heart disease and ischemic stroke. Estimates also suggest that VTE is responsible for >500,000 annual deaths in Western countries [36, 37]. In many incidents, up to 50% of patients develop recurrent VTE or suffer long-term complications [38–40]. Not only that but a 2-way link between VTE and cancer has also been confirmed. Cancer patients have at least 4-fold elevated risk for VTE in comparison with non-cancer patients. Cancer patients constitute 15–20% of all patients diagnosed with VTE [41]. Moreover, the prevalence of VTE significantly varies among ethnic and racial groups. In the US, African Americans are more likely to be diagnosed with VTE than any other group [42–45] which further complicates the issue of health disparities.

VTE is intravascularly triggered by activating the coagulation factors in a cascade fashion. Ultimately, the activation results in thrombin generation, deposition of fibrin, incorporation and consolidation of red blood cells (RBCs), and eventually formation of fibrin-rich, “red” thrombi [46–48]. Anticoagulants are the mainstay of VTE prevention and treatment. Current anticoagulants inhibit the plasma procoagulant activity via indirect inhibition (traditional or conventional therapy: heparins and coumarins) or direct inhibition (newer or novel therapy: peptidomimetics) of thrombin and/or factor Xa [49, 50]. However, traditional and newer anticoagulants are plagued with a number of drawbacks, and importantly, carry a significant risk of life-threatening internal bleeding [49, 50]. Thus, efforts to develop new approaches to safely prevent and/or treat VTE are of substantial clinical and economical impacts. Along these lines, multiple coagulation enzymes are being targeted to develop new anticoagulants. These protein targets include factors XIa [51, 52], XIIa [53,54], and XIIIa [55,56]. In this review, we will focus on drug discovery efforts targeting human FXIIIa.

FXIIIa is a thiol-containing TG that catalyzes the last step in the coagulation process. FXIIIa mechanically stabilizes the blood clot by cross-linking the α- and γ-chains of fibrin. FXIIIa also protects the newly formed clot from plasmin-mediated fibrinolysis, primarily by cross-linking α₂-antiplasmin to fibrin [30, 47]. Furthermore, FXIIIa appears to be a major determinant of the clot size and the clot’s RBC content. In a stasis-mediated venous thrombosis model, thrombi from the inferior vena cava of FXIII-deficient mice were found to have lower RBC content, resulting in 50% reduction in thrombus weight. The thrombi were also smaller than those from wild type mice indicating that FXIII(a) contributes to venous thrombus formation in vivo [57–59]. Interestingly, mice carrying mutations in the fibrinogen residues of γ390–396 exhibited a diminished FXIII binding to fibrinogen as well as a lagged FXIII activation and fibrin crosslinking. Those mice were also similar to FXIII-deficient mice in producing smaller venous thrombi with diminished RBC content [57, 60]. Moreover, several data analyses suggested that a specific FXIIIa polymorphism provides a substantial protection against VTE as well as coronary artery disease [60]. Furthermore, neither FXIII-heterozygous mice nor Fibγ390–396A mice exhibited signs of severe bleeding [57, 61, 62]. This has suggested that partial inhibition of FXIIIa (<100%) potentially does not significantly increase the bleeding risk. Overall, human FXIIIa appears to be a promising drug target to design inhibitors that may prevent and/or treat VTE without a significant increase in the risk of bleeding.

3. FXIIIa inhibitors under development

Given the above studies, inhibiting FXIIIa directly or indirectly leads to smaller clots with fewer entrapped RBCs and relatively more fragile clots that are more susceptible to hydrolysis by the action of proteolytic enzymes such as plasmin (Figure 3). Generally, the inhibitors that have been reported, thus far, to target human FXIIIa are either small molecules or polypeptides. Considering the small molecules, several FXIIIa inhibitors have been identified or designed intuitively by introducing 1) structural elements similar to endogenous substrates i.e. primary amines and/or amide-like characteristics; 2) electrophilic structural domains i.e. warheads such as α,β-unsaturated carbonyl, α-halomethyl carbonyl, imidazolium and others that covalently reacts with the key Cys314 residue; or 3) functional groups that steer binding towards potential allosteric site(s) such as sulfate or sulfonate groups. In addition, few peptidomimetic FXIIIa inhibitors have been rationally designed by considering the crystal structures of the zymogen FXIII [63–65] as well as the enzyme FXIIIa [66]. Regarding the polypeptides, a number of natural and synthetic molecules has also been reported. In the following sections, we describe the chemical, biochemical, and pharmacological aspects of FXIIIa inhibitors reported in peer-reviewed literature up to the time of this review publication.

Figure 3.

A) FXIIIa mediates the crosslinking of fibrin chains, which eventually increases the clot size, rigidity, and RBCs retention. It also catalyzes the crosslinking of α2-antiplasmin to the growing thrombi which makes them less susceptible to hydrolysis by proteolytic enzymes such as human plasmin. B) The effect of reduced/inhibited FXIIIa activity. Reducing FXIIIa activity, particularly by inhibitors, may decrease the clot size as well as the RBCs retention. It also makes the clot more susceptible to hydrolysis by proteolytic enzymes. This is particularly more important in preventing/treating venous thrombosis.

3.1. Small molecules

Alkylamines

The first small molecules that were investigated for their inhibitory effects on the catalytic activity of human FXIIIa were polyamines [67]. A series of SDS-PAGE experiments demonstrated that 1 mM spermidine 1 (or spermine 2) (Figure 4) inhibits FXIIIa-mediated crosslinking between fibronectin and collagen with approximate IC₅₀ value of 100 μM. Likewise, 0.7 mM spermidine inhibited fibronectin-collagen and fibronectin-fibrin but not fibronectin-fibronectin and fibrin-fibrin cross-linking. In contrast, 0.7 mM mono-dansylcadaverine (MDC) 3 (Figure 4) inhibited the FXIIIa-mediated cross linking of fibronectin-collagen, fibronectin-fibrin, fibronectin-fibronectin, and fibrin-fibrin [68].

Figure 4.

The chemical structures of alkylamines (1 – 8), which were reported to demonstrate substantial inhibitory activity toward human FXIIIa under various experimental conditions.

Similarly, a series of phenylthiourea alkylamine derivatives (Figure 4) were chemically synthesized and biochemically evaluated as inhibitors of TGs. The derivatives had 2, 3, 4, 5, or 6-atom spacer between the thiourea group and the terminal amine group. The derivatives were tested for the inhibition of guinea pig liver TG-catalyzed methylamine incorporation into glutamine-containing substrates (Z-Gln-Gly and the B chain of oxidized insulin) as well as for the inhibition of FXIIIa-catalyzed amine incorporation into fibrin and fibrin cross-linking. In this exercise, it was found that the inhibitory activity of phenylthiourea derivatives toward both the liver TG and FXIIIa increases with increasing the spacer length in the side chain up to 5-atom spacer with 1-(5-amino-pentyl)-3-phenylthiourea (PPTU) 4 (Figure 4) being the most potent [68]. A further increase in the linker length from 5-atom to 6-atom decreased the activity of the resulting derivative. The K_i value of PPTU for the inhibition of guinea pig TG-catalyzed amine incorporation into the B chain of oxidized insulin was found to be about 49 μM. According to Michaelis-Menten kinetics, the inhibition mechanism was competitive in nature [68]. Under similar conditions, MDC exhibited a K_i value of 25 μM.

It appears that phenythiourea alkylamines were designed so as to have a primary amino group which competes with the amine-containing substrate of FXIIIa. The primary amine-containing inhibitors afford a nucleophilic attack at the acyl-enzyme intermediate during the catalysis cycle resulting in FXIIIa inhibition. Along these lines, multiple endogenous monoamines including serotonin (5-HT) 5, dopamine 6, and histamine 7 (Figure 4) were shown to competitively inhibit FXIIIa- mediated crosslinking reactions [69]. Particularly, 5-HT was shown to act as a competitive inhibitor of FXIIIa-mediated crosslinking of plasma fibronectin in both an in vitro trans-glutamination activity assay and in MC3T3-E1 osteoblast cultures over the concentration range of 20 – 500 μM [69]. The inhibition impaired plasma fibronectin assembly and reduced matrix lysyl oxidase activity, alkaline phosphatase activity, type I collagen deposition, and osteoblast culture mineralization. The inhibition of FXIIIa-mediated matrix assembly was proposed as a potential mechanism by which 5-HT may adversely affect bone quality and mass and weaken bone matrix. Lastly, 100 mM cystamine 8 (Figure 4), a disulfide-containing diamine, fully blocked FXIIIa-mediated fibrin crosslinking in SDS-PAGE experiments, partially accounting to the anticoagulant activity of its preparations [70].

Given their potential, the chemical scaffolds of the aforementioned alkylamines have been exploited as a lead platform to subsequently develop more effective FXIIIa inhibitors with potential anticoagulant activity.

α-Halomethyl carbonyls

Mechanistically, it increasingly became evident that the primary amino group in alkylamine-based inhibitors (Figure 4) competes with the lysine residue during the aminolysis of the thioester intermediate, during the FXIIIA-mediated iso-peptide linkage formation. Consequently, these “competitive substrates” terminate the polycondensation reaction chain by rendering the crosslinking sites in the fibrin monomers inaccessible to further reaction, which subsequently prevents the formation of polymeric fibrin. Generally, the most efficient substrate competitors are structurally related to the lysine residue, and therefore, MDC and PPTU (Figure 4) are amongst the most potent competitive inhibitors. Nonetheless, high concentrations must be available from these inhibitory substrates to exhibit a significant enzyme inhibition, given that they have to compete with high affinity, large macromolecular substrates such as fibrin. Therefore, to produce more effective inhibitors of FXIIIa-mediated crosslinking of fibrin, the high specificity of the substrate-like amine structure of MDC was combined with the high affinity (reactivity) of α-halomethyl carbonyl to the sulfhydryl group of Cys314 residue in the active site of FXIIIa [71]. As a result, about 36 molecules with the general structural in figure 5 were synthesized and tested for fibrin crosslinking inhibition. Structure-activity relationship studies indicated that the most optimal number of methylene units (n) is 4 – 6 units. Extending the linker beyond this length significantly compromised the activity. Furthermore, Y was either divalent oxygen, divalent nitrogen, or methylene. Derivatives with Y=CH₂ (as in inhibitor 11) are 300-fold more potent than those with divalent nitrogen (as in inhibitor 10), which subsequently 3-fold more potent than derivatives with divalent oxygen (as in inhibitor 9). Moreover, unsubstituted sulfonamide i.e. R=H appeared to be the optimal structural feature for this position. X group is halogen and it was either chlorine or bromine atom. The impact of the nature of the halogen atom on activity was variable. Lastly, the sulfonamide group (as in inhibitor 11) was favored over the carbonyl group (as in inhibitor 15). The tosyl group-containing derivatives (as in inhibitor 11) also demonstrated better potency over p-chlorophenyl-containing derivatives (as in inhibitor 14). The potency of these molecules was expressed as the threshold concentration (μg/mL) of the inhibitor, which is the concentration at which the ratio of insoluble fibrin to the total fibrin starts to drop. Further studies suggested that the α-halomethyl carbonyl-based inhibitors are irreversible, active site inhibitors of FXIIIa. The nucleophilic thiol group of Cys314 in the active site of FXIIIa covalently reacts with the electrophilic α-carbon between the carbonyl and the halide substituent in the inhibitor [71].

Figure 5.

The chemical structures of α-halomethyl carbonyls (9 – 15), which were reported to demonstrate substantial inhibitory activity toward human FXIIIa via covalent interaction with the sulfhydryl group of active site Cys314. The potency of these inhibitors was expressed as the threshold concentration (μg/mL) of the inhibitor, which is the concentration at which the ratio of insoluble fibrin to the total fibrin starts to drop.

Imidazolium derivatives

A series of 2- [(2-oxopropyl)thio]imidazolium derivatives (16 – 19) (Figure 6A) were synthesized and tested for their inhibition potential and inhibition mechanism toward human FXIIIa [72]. The chemical structures and the second-order rate constants for FXIIIa inactivation and glutathione reaction are provided in figure 6A. Particularly, molecule 16 inhibited FXIIIa with an apparent second-order rate constant of 6.3 × 10⁴ M⁻¹ s⁻¹ which was 3.9 × l0⁷ times greater than its reaction rate with glutathione suggesting a great margin of specificity. Mechanistic enzyme studies of FXIIIa inhibition by these molecules demonstrated that acetonylation of the active site Cys314 residue of FXIIIa occurred along with the stoichiometric release of the complementary fragment of the inhibitor as the corresponding thione (Figure 6D). Kinetic analysis of the inactivation of FXIIIa by the non-quaternary derivative 20 (Figure 6B) indicated the formation of a reversible complex between the inhibitor and FXIIIa prior to irreversible inhibition of FXIIIa. Interestingly, inhibitor 16 exhibited no effect on prothrombin time or partial thromboplastin time which is indicative of lack of its effect on coagulation serine proteases at the highest tested concentration of 1 mM. Furthermore, the inhibitor had no effect on multiple thiol reagent-sensitive enzymes including papain, calpain, fatty acid synthetase, HMG CoA synthetase, and HMG CoA reductase at the highest concentration tested of 1 mM. Using concentrations 1 – 10 μM, the imidazolium derivatives and the related molecule 2-(1-acetonylthio)-5-methylthiazolo-[2,3]-1,3,4-thiadiazolium (L-722151) 19 increased the rates of tissue plasminogen activator (t-PA)-catalyzed clot lysis in vitro. These inhibitors prevented the FXIIIa-catalyzed covalent crosslinking of α₂-antiplasmin to the α-chain of fibrin and the formation of high molecular weight α-chain-based fibrin polymers, which justified the increased rates of clot lysis.

Figure 6.

The chemical structures of imidazolium derivatives (16 – 21), all of which reported as FXIIIa inhibitors, except derivative 20, which was used to study the mechanism of enzyme inhibition. Provided are the second-rate constants and the potential selectivity ratios over the sulfhydryl-containing glutathione as well as the IC₅₀ values of inhibitors 19 and 21 (panels A-C). Provided also is a depiction of the covalent reaction between the inhibitors and a key sulfhydryl group of a key Cys residue in the active site of FXIIIa (panel D).

The effect of FXIIIa inhibition by the perchlorate salt of molecule 19 (L-722151) on coronary thrombolysis and re-occlusion was studied in an acute model of electrically induced coronary thrombosis in dogs [73]. The inhibitor was intravenously administered (0.1 mg/kg/min) 15 min before current initiation, and then, the administration continued throughout the experiment for 270 min. About 15 min after thrombus formation, heparin (300 U/kg) was intravenously administered. After that by 45 min, recombinant t-PA was intravenously administered (10 pg/kg/min) for 90 minutes. Placebo-treated animals developed blood clots at 48.9 ± 8.1 min and re-perfused in response to t-PA at 49.1 ± 9.3 min. Animals, which were pre-treated with inhibitor 19, developed blood clots at 44.4 ± 9.7 min and re-perfused in response to t-PA at 16.4 ± 2.8 min (p<0.05) [3-fold faster reperfusion]. Furthermore, residual thrombus mass was reduced by inhibitor 19 from 6.9 ± 1.9 mg in placebo-treated animals to 1.7 ± 0.6 mg in the treated animals (p<0.05) [4-fold smaller mass]. In fact, the incidence of acute reocclusion was less in animals treated with inhibitor 19 (75% vs 86%), in comparison with that in placebo-treated animals. Moreover, the incidence of t-PA-induced reperfusion was high in animals treated with inhibitor 19 (100% vs 70%), in comparison to that in the placebo-treated animals. When inhibitor 19 was administered 15 minutes after thrombus formation in a separate group of animals, there was no beneficial effect on thrombolysis time or thrombus mass. Therefore, the mechanism for the improvement in reperfusion rates was proposed to result from either the inhibition of FXIIIa-mediated fibrin α-polymer formation and/or the inhibition of FXIIIa-mediated α₂-antiplasmin incorporation into the clot. Importantly, these studies were first to suggest that FXIIIa is an important target for therapeutic control of thrombosis and that inhibitor 19 may serve as a prototype for the development of effective FXIIIa inhibitors [73].

In a different set of experiments, 200 μM of 1,3-dimethyl-4,5-diphenyl-2[2(oxopropyl)thio]-imidazolium trifluoromethylsulfonate 21 (Figure 6C) prevented the formation of nearly all of the ɣ-dimers and higher cross-linked polymers [74]. The potency of inhibitor 21 was further demonstrated by its ability to significantly reduce the stiffness of clots over a wide range of physiological fibrinogen concentrations (3 – 12 μM). About 3-fold reduction in clot rigidity was produced at all fibrinogen concentrations in the presence of 2.5 μM of the inhibitor [74].

Along these lines, a series of water-soluble, imidazolium (22 – 27) and alkylsulfonium (28 – 31) peptidomimetics (Figure 7) was also synthesized and evaluated for FXIIIa inhibition and selectivity against recombinant TG2 [75]. Some of the inhibitors possessed glycine moiety (22, 24, 26, 28 and 30) and the others had phenylalanine moiety (23, 25, 27, 29, and 31). The enzyme assays in which these inhibitors were evaluated for their IC₅₀ were based on the Ca⁺²-mediated incorporation of N-(5-aminopentyl)-biotinamide into N,N′-dimethylcasein by FXIIIa or recombinant human TG2. In one hand, there were only six imidazolium peptidomimetics that inhibited FXIIIa, yet with no significant selectivity over TG2 (Figure 7A). Several derivatives in this series were reported to be irreversible, active site inhibitors of TGs, albeit with moderate to weak potency. Moreover, the structure-activity relationship studies revealed that the overall inhibition trends of FXIIIa by the imidazolium derivatives were similar in both the glycine series (inhibitors 22, 24, and 26) and the phenylalanine series (inhibitors 23, 25, and 27). Importantly, their FXIIIa inhibitory potencies appeared to be very sensitive to modifications of the substituents on the imidazolium moiety. The inhibition potency was gradually lost with the increase in the size and/or the number of substituents. The most potent among imidazolium derivatives were those with methyl substituents as in inhibitors 22 and 24. In the other hand, the alkylsulfonium peptidomimetics were more selective to recombinant TG2 with molecules having small substituents on the sulfur atom demonstrating better inhibition profiles (Figure 7B). Molecular modeling studies established the binding of these derivatives into the active site of the two enzymes in a way that the electrophilic carbon (the carbon between the carbonyl moiety and the sulfur atom) in the inhibitor comes into close proximity to the nucleophilic thiol group of the cysteine residue in the active site. Thus, this class of inhibitors was considered as irreversible, active site inhibitors of FXIIIa. Overall, inhibitor 24 demonstrated the best profile in this series for selective inhibition of human FXIIIa (FXIIIa IC₅₀= 30 μM; TG2 IC₅₀ >200 μM), yet no studies have reported any further development of this inhibitor.

Figure 7.

The chemical structures of a series of water-soluble, A) imidazolium (22 – 27) and B) alkylsulfonium (28 – 31) peptidomimetics. Provided are the corresponding IC₅₀ values for FXIIIa inhibition as well as TG2 inhibition.

Imidazo-thiadiazole-containing inhibitors

A series of 3-substituted imidazo[1,2-d][1,2,4]-thiadiazole derivatives (32 – 38) (Figure 8A) were synthesized and evaluated for FXIIIa inhibition [76]. Initially, molecule 32 was found to inhibit FXIIIa with an IC₅₀ value of 11 μM. Six derivatives were subsequently synthesized to have the amino group substituted with different phenylsulfonyl groups. The new derivatives inhibited FXIIIa with IC₅₀ values of < 3 μM with the most potent being inhibitor 36 with IC₅₀ of 0.13 μM. The two molecules 39 and 40 (Figure 8B), which lacked the imidazo-thiadiazole moiety, were either weakly active or inactive. This outcome highlighted the significance of this moiety to the overall activity of this class of FXIIIa inhibitors. The active site cysteine residue of FXIIIa is thought to attack the sulfur atom in the imidazo-thiadiazole heterocyclic system resulting in covalent inhibition of the enzyme activity through the formation of the inhibitor-enzyme adduct (Figure 8C). Further modification of inhibitor 36 by introducing methyl, acetamide, or acetic acid moieties led to three inhibitors 41– 43 (Figure 9) with IC₅₀ values of 0.11, 0.42, and 0.62 μM, respectively. Inhibitors 41–43 exhibited 27–76-fold selectivity index to inhibit FXIIIa over guinea pig liver TG. Next, inhibitor 36 (40–160 μM) was evaluated for its effect on the composition of platelet-depleted, human plasma clots which was generated and prepared following spiking the plasma with ¹²⁵I-fibrinogen. In relative to the control, the FXIIIa-mediated formation of γ-γ dimers and higher molecular weight fibrins was significantly decreased. The formation of α-polymers was fully inhibited at 160 μM. At the same inhibitor concentration, the amount of γ-γ dimers was halved in relative to the control. Over the same concentration range, the inhibitor 36 eventually accelerated the t-PA-mediated fibrinolysis of platelet-depleted and platelet-rich human plasma clots [76], which highlighted the biological relevance of this molecule. Although the pharmacokinetic aspects of inhibitor 36 was subsequently studied in rabbits, its development was not further pursued, perhaps because of its short half-life (T_0.5 = 0.13 hr) and high total clearance (Total Cl > 3.8 L/h/kg) which made the target plasma concentration of ~80 μM, as demonstrated by in vitro study, not achievable [77].

Figure 8.

A) The chemical structures of imidazo-thiadiazole-containing FXIIIa inhibitors (32 – 38) and nitro-arylsulfonamide-based inhibitors (39 and 40). Provided are the corresponding IC₅₀ values. B) The covalent reaction between the imidazo-thiadiazole moiety and a key sulfhydryl group of a key Cys residue in the active site of FXIIIa.

Figure 9.

The chemical structures of FXIIIa inhibitors (41 – 43), which were designed based on combining the structural aspects of imidazo-thiadiazole-containing FXIIIa inhibitors and those of nitro-arylsulfonamide-based inhibitors. Their structure-activity relationship is described considering the corresponding IC₅₀ values toward FXIIIa and guinea pig TG.

Natural products and their analogs

Several natural products were reported to inhibit human FXIIIa. In this arena, cerulenin 44 (Figure 10), a fatty acid-like epoxide carboxamide isolated from the culture broth of Cephalosporium caerulens, was reported to inhibit FXIIIa at concentrations as low as 10 μM [78]. In a similar study, cerulenin was found to inhibit FXIIIa with an IC₅₀ value of 4 μM and this potency was unaffected by adding 20 μM of glutathione and only 2.5-fold decreased upon the addition of 1 mM of glutathione suggesting a significant potency and selectivity [79]. Moreover, computational studies revealed that the inhibition involves a nucleophilic attack by the active site Cys314 on position-2 of the epoxide ring that bears the primary carboxamide i.e. covalent inhibition, and that the resulting oxyanion is stabilized via hydrogen bonding with backbone N-H moieties within the ‘oxyanion hole’ of the enzyme. Furthermore, the computational exercise suggested that replacing the 3-keto unit with an amide group may enhance binding to FXIIIa in the initial complex via formation of an additional hydrogen bond from this amide N-H to the backbone carbonyl of Tyr372. In addition, replacement of the alkenyl chain in cerulenin with an aromatic ring was predicted to result in a potential π-stacking interaction with Tyr372, and subsequently, to potentially enhance the initial binding of the inhibitor. Inspired by the molecular modeling studies, 11 racemic mixtures of eleven cis-bisamido epoxides were synthesized and tested against human FXIIIa and TG2 [79]. Most of the resulting molecules inhibited FXIIIa with sub-micromolar IC₅₀ values, yet with no significant selectivity over TG2. The best two molecules in targeting FXIIIa were 45 and 46 (Figure 10) as they inhibited FXIIIa with IC₅₀ of 4 nM. Importantly, all inhibitors showed negligible inhibition of cathepsin S at concentrations as high as 100 μM, which is equivalent to ~10,000-fold selectivity for FXIIIa. In molecular modeling studies, inhibitor 46 was found to recognize amino acids Trp279, Gln313, Cys314, Tyr372, and His373 in the active site of FXIIIa.

Figure 10.

The chemical structures of natural products and their analogs (44 – 52), which have been reported with inhibitory activity against FXIIIa. Provided are the corresponding IC₅₀ values in μM units.

Along these lines, two cyclopropenone fungal metabolites, alutacenoic acids A 47 and B 48 (Figure 10), from Eupenicillium alutaceum were also found to be potent and specific inhibitors of FXIIIa. In one hand, alutacenoic acid A 47 has 5-methylene linker separating the cyclopopenone ring and the terminal carboxylic acid group and inhibited human FXIIIa with an IC₅₀ value of 1.9 μM. In the other hand, alutacenoic acid B 48 has 7-methylene linker separating the cyclopopenone ring and the terminal carboxylic acid group and inhibited human FXIIIa with an IC₅₀ value of 0.61 μM [80, 81]. Interestingly, alutacenoic acid B exhibited 20-fold selectivity index over TG from guinea pig liver. Not only that but alutacenoic acid B did not inhibit other thiol-containing enzymes including papaine, calpain from porcine erythrocyte, cathepsin B from human liver, and IL-1β converting enzyme at the highest concentration tested which was 50 μM. In a structure-activity relationship study, alutacenoic acid derivatives with 6-, 8-, 9-, 10-, and 11-methylene linkers inhibited human FXIIIa with IC₅₀ values of 0.93, 0.52, 0.58, 0.31 (inhibitor 49, Figure 10), and 0.34 μM, respectively. Transforming alutacenoic acid B to the corresponding phenylethyl amide derivative 50 (Figure 10) resulted in 23.5-fold increase in the FXIIIa inhibition potency. Molecular modeling studies revealed that inhibitor 50 binds to FXIIIa with the cyclopropenone ring fits highly complementarily into the active site residues located at the base of the binding cavity. The ketone oxygen of the ring forms a hydrogen bond with the indole NH group of Trp279 residue, and the terminal carbon is spatially close to the thiol group of Cys314 residue [80].

Another natural product with inhibitory activity towards human FXIIIa is cis-R-(−)-resorcylide 51 (Figure 10). It is a fungal polyketide secondary metabolite that is defined by a 1,3-benzenediol moiety. Positions-5 and −6 are bridged by a macrocyclic lactone ring [82]. In initial screening, the natural lactone inhibited FXIIIa and a tissue TG with IC₅₀ values of 5.7 and 3.1 μM, respectively. Cis-R-(−)-resorcylide did not inhibit thrombin, factor Xa, factor IXa, factor VIIa, or papain at the highest concentration tested of 100 μM [83]. At a similar concentration, cis-R-(−)-resorcylide inhibited the fibrin cross-linking and the formation of higher fibrin oligomers during the clot formation process and rendered the formed clots more susceptible to hydrolysis by plasmin and urokinase. Lastly, N-acetyl-tyramine 52 (Figure 10), which is a tyramine alkaloid isolated from fungi and actinomyces, was reported to inhibit human FXIIIa with an IC₅₀ range of 0.2–12.8 mM [84].

Considering FXIIIa as a drug target, none of the above natural products or their analogs were subsequently developed as antithrombotics.

Synthetic Michael acceptor-containing inhibitors: 4-(acrylamido)phenyl derivatives and peptidomimetics

A series of 4-(acrylamido)phenyl derivatives (53 – 62) and a related molecule 63 (Figure 11) were synthesized and evaluated for their inhibitory activities toward TG2 in the pursuit of treatment for Huntington’s disease [85]. Yet, some of these derivatives also exhibited significant inhibition potency toward FXIIIa. In fact, inhibitors 53 – 63 inhibited human FXIIIa with IC₅₀ values of 0.039–6.8 μM. The most selective among these molecules was inhibitor 63, which was 13-fold more selective to FXIIIa over TG2. Yet, this molecule inhibited TG1 with IC₅₀ value of 8.9 μM. Similar to most of other FXIIIa inhibitors, 4-(acrylamido)phenyl derivatives are covalent inhibitors.

Figure 11.

The chemical structures of synthetic Michael acceptor-containing inhibitors: 4-(acrylamido)phenyl derivatives (53 – 62) and a related molecule 63. These inhibitors were originally developed for TG2 inhibition, yet they also showed significant inhibition potential toward FXIIIa. Provided also is the IC₅₀ for each inhibitor. The β-carbon of Michael acceptor domain in each of the presented inhibitors is to covalently bind to Cys314 in the active site of FXIIIa.

Along these lines, two inhibitors i.e. inhibitor 64 (ZED1301) [66] and 65 (ZED3197) [56] (Figure 12) were also reported recently. In particular, the molecule 64 inhibited FXIIIa with an IC₅₀ value of 110 nM and demonstrated a selectivity index of 26.4-fold over TG2. Inhibitor 64, an octapeptide irreversible inhibitor, was used to obtain the first high-resolution crystal structure (1.98 Å) of FXIII in an active state (a recombinant human cellular FXIII which was activated by a high concentration of Ca²⁺). The inhibitor was identified following a phage display screening [86] and carries an α,β-unsaturated methyl ester i.e. Michael acceptor instead of the substrate glutamine side chain. The inhibitor ZED1301 binds to the surface of the catalytic domain of FXIIIa in a region that can be split into three interaction sites: a) the catalytic site (black color in figure 12); b) the proximal “α-space” (green color), and c) the more distant hydrophobic pocket (orange color). Within the active site of FXIIIa, the Michael acceptor β-carbon atom (warhead group) of inhibitor 64 forms a covalent bond with the Cys314 residue, which is protected from hydrolysis by the hydrophobic tunnel of Trp279 and Trp370 residues. The carbonyl oxygen atom of the terminal methyl ester establishes hydrogen bonds to the oxyanion hole formed by the indole nitrogen atom of Trp279 and the backbone NH group of Cys314 residue. In the α-space, the N-acetylated aspartate moiety of the inhibitor interacts via seven hydrogen bonds with FXIIIa. These bonds are formed with Tyr283, Gln313, Asn371, Tyr372, and Arg223 residues. Lastly, the indole ring of the inhibitor fits into the hydrophobic pocket of FXIIIa. Overall, the reported co-crystal structure presented multiple structural insights that guided subsequent structure-based drug design efforts of more clinically relevant FXIIIa inhibitors.

Figure 12.

The chemical structures of synthetic Michael acceptor-containing inhibitors: peptidomimetics (64–65). Provided also is the IC₅₀ for each inhibitor. Depiction of inhibitor 64 shows three structural domains along with the corresponding binding pockets/cavities which they occupy/bind to in the active site of FXIIIa. The β-carbon of Michael acceptor domain in each of the presented inhibitors is to covalently bind to Cys314 in the active site of FXIIIa. Inhibitor 65 arguably is the most advanced among all reported FXIIIa inhibitors as far as in vivo testing.

Given the peptidic nature of inhibitor 64 (ZED1301), it was well anticipated that the inhibitor will lack an adequate oral bioavailability and will suffer from short half-life. Therefore, structural modifications were further implemented by replacing the N-acetylated aspartate moiety with substituted thiazole or benzimidazole moieties (64a – 64c). However, no significant improvement in drug-like properties was achieved.

Importantly, the anticoagulant potential of inhibitor 65 (ZED3197) was reported very recently [56]. It is a hexapeptide-like molecule which was synthesized following the standard solid-phase peptide chemistry. The inhibitor possesses a Michael acceptor warhead which covalently reacts with the active site’s Cys314 residue. The inhibitor appeared to demonstrate good solubility and stability in the corresponding biological fluids. In the iso-peptidase assay, ZED3197 inhibited human plasma derived FXIII-A₂B₂ as well as the recombinant cellular form FXIII-A₂ with similar IC₅₀ values of 10 and 16 nM, respectively. The inhibitor also showed similar potency targeting FXIII-A₂ from various animal species (IC₅₀ values =8–365 nM). In the trans-amidation assay, ZED3197 inhibited the recombinant human form of FXIII-A₂ with an IC₅₀ value of 24 nM. Likewise, ZED3197 inhibited recombinant FXIII-A₂ from other animal species (IC₅₀ values = 7–24 nM). Importantly, the inhibitor demonstrated selectivity indices of 463-, 19-, 2791-, and 56-fold over TG1, TG2 (most abundant), TG3, TG4, and TG7, yet it was not selective against TG6 (neuronal transglutaminase). Neither thrombin nor plasmin was inhibited by ZED3197 at the highest concentration tested of 100 μM.

Interestingly, 10 μM of ZED3197 inhibited FXIII-A₂-mediated crosslinking of thrombin-initiated fibrin clots resulting in readily dissolvable clots i.e. readily hydrolyzed by human plasmin, as revealed in size exclusion chromatography, SDS-PAGE, and Western-blotting experiments. Rotational thrombo-elastometry in whole human blood revealed that ZED3197 dose-dependently (0.08–20 μM) prolonged clot formation, reduced clot firmness (calculated EC₅₀= 1.7 μM), and facilitated clot lysis at 60 minutes in the presence of 0.02 μg/ml t-PA (calculated EC₅₀= 0.7 μM), without affecting the clotting time. Furthermore, the in vivo effect of the inhibitor was confirmed using the Wessler rabbit model of venous stasis and reperfusion. The inhibitor was admisntred via IV bolus injection (24 mg/kg) followed by IV infusion (6 mg/kg + 54 mg/kg) to maintain a steady concentration of ~13 μM throughout the study, a concentration at which FXIIIa inhibition was more than 95% as determined by the iso-peptidase assay. Considering the vessel patency after termination of the stasis, the blood flow in the inhibitor-treated animals continued to increase over 35–135 min following the removal of the clamps, however it was continuously declining over the same period in the control-treated animals. During the reperfusion period, the inhibitor resulted in an average baseline blood flow recovery of 22%, in relative to only 6% in the control group. In addition, the thrombus weight in the inhibitor-treated animals was 2-fold less than that in the control animals (about 29 vs 65 mg). The thrombi of rabbits infused with the inhibitor were also softer than those exposed only to the vehicle, and less adhered to the suture thread. Remarkably, a template bleeding time assay showed no significant difference between ZED3197-treated animals and control animals (about 135 vs 134 sec). Moreover, thrombo-elastography analysis confirmed that the coagulation index obtained for ZED3197-treated rabbits was significantly lower than that obtained for control rabbits. Overall, the above studies indicated that ZED3197 (inhibitor 65) facilitated fibrinolysis and exhibited no effect on the bleeding time. Although the lack of significant selectivity over other TGs, the less than optimal oral bioavailability of this inhibitor, and its short half-life may hinder its further development, yet the reported studies further advance the concept of inhibiting FXIIIa as a viable approach for treating/preventing VTE.

Sulfated molecules

Following the double-displacement mechanism, the catalytic cycle of FXIIIa requires two endogenous substrates to engage with the active site, as illustrated in figure 2. Therefore, this cycle is susceptible to any minor conformational change in the active site that can be brought about by potential allosteric modulators. At fundamental level, allosteric inhibition may offer two major advantages over orthosteric inhibition including: 1) higher specificity of interaction by targeting less conserved sites; and 2) better regulation of activity by achieving partial inhibition [87–92]. The first advantage is particularly important considering the lack of significant specificity of reported FXIIIa over TGs. The latter advantage is also substantial because complete inhibition of FXIIIa activity leads to bleeding diathesis, whereas partial reduction does not result in excessive bleeding [11, 34, 35,93].

An anion-binding site on TG2 was previously identified as the heparin-binding site and was found to be important for enzyme function [94, 95]. Binding of heparin to this site is mediated via electrostatic bonds between the negatively charged sulfate groups of heparin and the positively charged, protonated Arg and Lys residues of the binding site. Likewise, the A subunit of FXIIIa has a substantial number of Arg and Lys residues forming a potential anion-binding allosteric site (Figure 13A). In contrast to the conserved active sites, the anion-binding sites on TG2 and FXIIIa differ in the number, orientation, and location of Arg and Lys residues as well as in their hydrophobic sub-sites. This has led to the hypothesis that sulfated or sulfonated molecules, which complements the potential anion-binding site on FXIIIa, can potentially be designed as potent and selective allosteric modulators of FXIIIa [55].

Figure 13.

A) The putative anion-binding site which can serve as an allosteric site on FXIIIa. The site includes a cluster of Arg and Lys amino acids (blue spheres). The catalytic triad is also depicted in yellow spheres. B) The chemical structures of sulfated nonsaccharide heparin mimetics 66 and 67 that were reported to inhibit FXIIIa (trans-amidation and crosslinking) via allosteric mechanism of inhibition. Provided also is the IC₅₀ for each inhibitor.

To put this hypothesis into action, 22 variably sulfated molecules were screened for their inhibitory potential against FXIIIa using the trans-glutamination assay under physiological conditions [55]. As a result, sulfated molecules 66 and 67 (Figure 13B) were identified to have moderate inhibition potency. Sulfated flavonoid trimer 66 inhibited FXIIIa with an IC₅₀ value of 36.2 μM and efficacy of 98%, whereas the structurally related trimer 67 inhibited FXIIIa with an IC₅₀ value of 118 μM and efficacy of 93%. The sulfated molecule 66 was found to bind to FXIIIa with a K_D value of about 25.3 μM. Furthermore, Michaelis-Menten kinetics suggested that the sulfated molecule 66 is allosteric modulator of FXIIIa. Further enzyme inhibition studies indicated that sulfated molecule 66 is at least 26-, 9-, and 27-fold selective to FXIIIa over thrombin, factor Xa, and papain, respectively. Moreover, 85% of the activity of FXIIIa was restored by about 0.80 mg/mL polybrene, an Arg-rich polypeptide, following inhibition with 100 μM of molecule 66, proving the reversibility of the inhibition. Interestingly, inhibitor 66 also inhibited FXIIIa-mediated cross-linking of fibrin monomers with an IC₅₀ value of 76.3 μM and efficacy of ~100%. Overall, although inhibitor 66 was not tested in animal models, yet it has presented a proof-of-concept that FXIIIa activity can be modulated through allosterism, an alternative approach that can be recruited to design antithrombotics with reduced bleeding risks [55].

Nitric oxide donors

Few alkylamines that can donate nitric oxide such as S-nitroso-N-acetyl-penicillamine 68, spermine-nitric oxide 69, 3-morpholinosydnonimine 70, and S-nitroso-glutathione 71 (Figure 14) were also reported to dose-dependently inhibit FXIIIa in vitro and in vivo, albeit at very high concentrations (>1 mM). Inhibition was reported to take place by S-nitrosylation of a highly reactive cysteine residue [96]. No further development of this group of inhibitors was subsequently reported.

Figure 14.

The chemical structures of nitric oxide donors/carriers (68 – 71) that in vitro and in vivo inhibit FXIIIa at high mM concentrations. The inhibition occurs via S-nitrosylation of a highly reactive cysteine residue, presumably Cys314.

3.2. Polypeptides

Polypeptides have also been reported to target FXIIIa for antithrombotic activity. Specifically, tridegin is a small peptide of 66 amino acids and apparent molecular weight of 7.3 kDa [97]. It was first obtained from the salivary gland extract of the gigantic Amazon leech, Haementeria ghilianii. In initial biochemical studies, the effect of tridegin on plasma FXIIIa activity was estimated by the rate of ammonia release during FXIIIa-mediated formation of covalent cross-links between ethylamine and β-casein. Results demonstrated that the rate of ammonia production was reduced in the presence of tridegin in a concentration-dependent manner. The reduction was attributed to the inhibition of FXIIIa by tridegin for which the IC₅₀ was found to be ~0.046 μg/mL. Tridegin also dose dependently inhibited the incorporation of 5`-(biotinamido)pentylamine into N,N`-dimethylcasein with an IC₅₀ value of ~0.07 μg/mL (~9.2 nM), efficacy of 91%, and approximate stoichiometry of 1. At a concentration of 8 μg/mL, tridegin inhibited the cross-linking of both the α- and the γ-chains of fibrin in SDS/PAGE experiment. Importantly, tridegin exhibited no fibrinogenolytic effect at concentrations as high as 35 μg/mL. Furthermore, tridegin did not affect clotting times of human plasma at 4.6 μg/mL. Tridegin did not affect the amidolytic activity of thrombin or factor Xa at the highest concentration tested of 2.2 μg/mL. Likewise, tridegin did not affect several thiol proteases (bromelain, papain or cathepsin C) at the highest concentration tested of 2.2 μg/mL. However, tridegin inhibited guinea pig TG with IC₅₀ values of 1.61–1.83 μg/mL. In related studies, it was demonstrated that tridegin accelerates plasmin-mediated fibrin degradation and 50% reduces the time for fibrinolysis at concentration of 1.2 U/mL [98–100]. It was also shown that plasma clots formed in the presence of tridegin (2.6–5.0 U/mL) did undergo faster lysis when one of the fibrinolytic enzymes (hementin, streptokinase, or t-PA) was added 2 hrs following clot formation. Importantly, the impact of tridegin was substantially increased when the clots were from platelet-rich plasma, perhaps due to its effect on platelet FXIIIa. Platelet-rich plasma clots were found to lyse with slow rate by the fibrinolytic enzymes, yet the rate of lysis returned to that of platelet-free clots upon the addition of tridegin, suggesting that the polypeptide can be a great adjunct therapy to be administered with thrombolytic agents in case of arterial thrombosis [98–100]. These results were further confirmed by measuring the in vitro effect of purified tridegin on the storage modulus parameter during thrombin-induced clot formation. The addition of purified tridegin was found to dose-dependently decrease/inhibit the development of the storage modulus on application of shear force i.e. to decrease/inhibit the rigidity of the clots, with prominent effect on aged clots. The IC₅₀ of this inhibition was found to be 138 ng/mL. Once again, the in vitro study showed that clots formed in the presence of purified tridegin (0.8 μg/mL) lysed faster by the leech’s hementin as the 50% lysis time dropped from ~22.3 hrs in the absence of purified tridegin to ~16.0 hrs in its absence [98–100]. At this concentration, tridegin inhibited fibrin cross-linking as determined by the absence of γ-γ dimers and α-polymers in SDS/PAGE gel experiments.

To translate tridegin into clinically relevant FXIIIa inhibitor, chemical synthesis and structure-activity relationship studies were performed [101]. Tridegin and its derivatives were synthesized either by solid-phase assembly or by native chemical ligation, both of which were followed by oxidation in solution phase. Tridegin and truncated analogues were evaluated for FXIIIa inhibitory activity in chromogenic substrate hydrolysis assay which revealed the significance of the C-terminal segment of the polypeptide. In fact, a linear recombinant tridegin had an IC₅₀ value of ~0.13±0.06 μM in this assay, whereas a linear synthetic oxidized tridegin had an IC₅₀ value of 0.48±0.13 μM. Nevertheless, the most important region of tridegin sequence, that is likely to interact with the active site of FXIIIa, was found to be between Pro46 and Phe57 (PMDDIYQRPVEF) as this sequence alone exhibited an IC₅₀ value of 26.2±2.0 μM. Shorter or modified sequences were associated with IC₅₀ values of >138 μM. Considering the N-terminal segment, it was suggested that it may contribute to the overall inhibitory activity of tridegin by decelerating the protein modification by FXIIIa, providing conformational stability, and/or enhancing binding affinity [101]. Important to mention here that although tridegin was first reported in 1997, yet its folding pathway remains to be explored. Therefore, structural and functional studies focusing on its six cysteine residues and its three disulfide bonds continue to evolve so as to assess their impact on folding, stability, and inhibitory activity, and further, to guide its development as antithrombotic agent [102, 103].

Lastly, few antibodies were also reported to inhibit human FXIIIa activity or FXIII activation. These include the natural antibody of IgG New Haven [104], the monoclonal antibody MAb309 [105], and the monoclonal antibody 5A2 [106]. Yet, none of these antibodies was further developed beyond the initial report or discovery.

4. Conclusion and future directions

Plasma FXIIIa is intravascularly formed upon activation of the zymogen FXIII by thrombin and Ca²⁺. Human FXIIIa affects diverse physiological processes, of which its effect on hemostatsis is the most recognized. Several biochemical and pharmacological studies have revealed that FXIIIa may serve as a promising drug target to develop new breed of anticoagulants with a limited bleeding risk, which is known to complicate the therapeutic use of all currently available anticoagulants. Particularly, the new FXIIIa-targeting anticoagulants can therapeutically be used for the prevention and/or treatment of VTE.

Despite promises, no FXIII(a)-targeting molecule is approved for any therapeutic indication thus far, and we are yet to see a FXIIIa inhibitor being tested in clinical trials. In fact, very few FXIIIa inhibitors are under development as anticoagulants. The reported inhibitors comprise small molecules and polypeptides. They include reversible and irreversible inhibitors as well as competitive and noncompetitive inhibitors. Of these, the hexapeptide ZED3197 (inhibitor 65; figure 12) and the polypeptide tridegin and its derivatives appear to be the most advanced. It is important to emphasize here that the thiol-containing active site is the hallmark of all TGs as well as cysteine proteases. This structural feature introduces considerable challenge in developing selective active site FXIIIa inhibitors. In addition to that, the limited number of FXIIIa-ligand co-crystal structures has led to a significant lack of fundamental understanding of interactions at atomic level. Thus, the progress toward identifying specific key structural elements that improve selectivity as well as potency is significantly hampered.

An emerging trend in targeting FXIIIa is the use of non-saccharide, sulfated and/or sulfonated heparin mimetics. Sulfated flavonoid trimers structurally mimic heparins, yet functionally inhibit FXIIIa in an allosteric manner. In a conserved family of several proteins, allosteric inhibitors usually convey two advantages over orthosteric inhibitors. First, allosteric sites tend to be relatively less conserved among the proteins of the same family which may provide access to an enhanced specificity for inhibitors targeting these sites. Second, allosteric sites, in concept, afford the potential of less than complete reduction in enzyme function i.e. <100% inhibition, which may become a safety gauge in modulating enzymes involved in multiple physiological roles, such as FXIIIa. The discovery of the sulfated flavonoid-based trimers as allosteric inhibitors may eventually lead to the design of inhibitors that specifically target FXIIIa. This is a relatively new concept and is in rapid development. For example, tetra-sulfated tetrahydroisoquinolines [107] were rationally designed to allosterically target the heparin-binding site of antithrombin so as to produce anticoagulant activity. Likewise, deca-sulfated pentagalloyl glucosides [108] and dodeca-sulfated inositol derivatives [109] have been developed to allosterically target human factor XIa so as to promote effective and near bleeding-free anticoagulant effects. In addition to allostery, small sulfated heparin mimetics offer many other advantages including (1) the high water solubility, which is expected to facilitate anticoagulant use during surgeries; and (2) the low cellular, placental, and central nervous system toxicity because of their highly negatively charged nature. Yet, more efforts are needed to advance this class, and the other classes, of FXIIIa inhibitors towards clinical settings. For instance, the effect of FXIIIa inhibitors beyond hemostasis should be thoroughly evaluated. Furthermore, approaches to reverse their actions may be needed to address potential side effects. These approaches may involve the use of agent-specific antidotes or FXIII supplement.

Overall, given the fact that FXIIIa activity is a critical determinant of venous thrombus size and composition [110–113], FXIIIa inhibitors hold a significant promise as a primary or adjunct approach for anticoagulation in VTE.

Highlights.

• FXIIIa is a transglutaminase that catalyzes the last step in the coagulation process

• FXIIIa is a viable target to develop new anticoagulants for venous thromboembolism

• FXIIIa inhibitors are covalent or noncovalent, small molecules or polypeptides

• Selectivity over other transglutaminases and cysteine proteases is the bottleneck

• Chemical, biochemical, & pharmacological properties of FXIIIa inhibitors are reviewed

Abbreviations

FXIII

factor XIII

FXIIIa

factor XIIIa

5-HT

serotonin

MDC

mono-dansylcadaverine

PPTU

pentyl-3-phenyl-thiourea

TG

transglutaminase

t-PA

tissue-type plasminogen activator

RBCs

red blood cells

VTE

venous thromboembolism