Allosteric Inhibition of Factor XIIIa. Non-Saccharide Glycosaminoglycan Mimetics, but Not Glycosaminoglycans, Exhibit Promising Inhibition Profile

Abstract

Factor XIIIa (FXIIIa) is a transglutaminase that catalyzes the last step in the coagulation process. Orthostery is the only approach that has been exploited to design FXIIIa inhibitors. Yet, allosteric inhibition of FXIIIa is a paradigm that may offer a key advantage of controlled inhibition over orthosteric inhibition. Such an approach is likely to lead to novel FXIIIa inhibitors that do not carry bleeding risks. We reasoned that targeting a collection of basic amino acid residues distant from FXIIIa's active site by using sulfated glycosaminoglycans (GAGs) or non-saccharide GAG mimetics (NSGMs) would lead to the discovery of the first allosteric FXIIIa inhibitors. We tested a library of 22 variably sulfated GAGs and NSGMs against human FXIIIa to discover promising hits. Interestingly, although some GAGs bound to FXIIIa better than NSGMs, no GAG displayed any inhibition. An undecasulfated quercetin analog was found to inhibit FXIIIa with reasonable potency (efficacy of 98%). Michaelis-Menten kinetic studies revealed an allosteric mechanism of inhibition. Fluorescence studies confirmed close correspondence between binding affinity and inhibition potency, as expected for an allosteric process. The inhibitor was reversible and at least 9-fold- and 26-fold selective over two GAG-binding proteins factor Xa (efficacy of 71%) and thrombin, respectively, and at least 27-fold selective over a cysteine protease papain. The inhibitor also inhibited the FXIIIa-mediated polymerization of fibrin in vitro. Overall, our work presents the proof-of-principle that FXIIIa can be allosterically modulated by sulfated non-saccharide agents much smaller than GAGs, which should enable the design of selective and safe anticoagulants.

Fig 1. The library of sulfated GAGs and NSGMs.

Molecules in this library possessed structural diversity at multiple levels, e.g., different core scaffolds, number of sulfate groups, and position of sulfate groups. NSGMs included five distinct chemical classes of pentasulfated monomeric flavonoid scaffolds (1–2), hexa- and octa- sulfated dimeric flavones (3–9), pentasulfated dimeric flavone-quinazolinones (10–12), undecasulfated trimeric flavones (13 and 14), and dodecasulfated hexabenzoyl inositol derivative (15). GAGs comprised of unfractionated heparin (UFH, 16), enoxaparin (17), and chondroitin sulfate A-C (19–21) in addition to the octasulfated pentasaccharide fondaparinux (18) and the octasulfated disaccharide sucrose octasulfate (22).

Fig 2. The putative anion-binding allosteric site of human FXIIIa.

(A) The electrostatic potential of the surface exposed anion-binding site of FXIII (PDB ID: 1GGU). (B) The basic residues in the site are shown as spheres. The residues matching the heparin-binding site of transglutaminase are K61, K73, R303, and K678. (C) The electrostatic potential of human thrombin is shown (PDB ID: 1XMN). (D) The basic residues of thrombin’s exosite 2 are shown in spheres. Positive and negative potentials are colored in blue and red, respectively.

Fig 3. Interaction of human FXIIIa and α-thrombin (α-Th) with NSGM 13 and UFH.

(A) The inhibition of FXIIIa (●) and α-Th (○) by NSGM 13 was measured spectrofluorometrically through a bisubstrate, fluorescence-based transglutamination assay (FXIIIa) or chromogenic substrate assay (α-Th) at pH 7.4/8.0 and 37°C. Solid lines represent sigmoidal fits to the data to obtain IC₅₀, HS, Y_M, and Y_O using Eq 1. (B) Spectrofluorometric measurement of the affinity of human FXIIIa for inhibitor 13 at pH 8.0 and 37°C using the intrinsic tryptophan fluorescence (λ_EM = 348 nm, λ_EX = 280 nm). Solid lines represent nonlinear regressional fits using quadratic Eq 2. (C) Spectrofluorimetric measurement of the affinity of human FXIIIa for UFH at pH 8.0 and 37°C using the intrinsic tryptophan fluorescence (λ_EM = 348 nm, λ_EX = 280 nm). Solid lines represent nonlinear regressional fits using the standard Hill Eq 3. See details in Materials and Methods.

Fig 4. Michaelis−Menten kinetics of dansylcadaverine and N,N-dimethylcasein conjugation by human FXIIIa in the presence of NSGM 13.

The initial rate of conjugation at (A) various dansylcadaverine concentrations (0–750 μM) and fixed dimethylcasein concentration (5 mg/mL) or (B) various dimethylcasein concentrations (0–5 mg/mL) and fixed dansylcadaverine concentration (250 μM) was measured spectrofluorometrically in pH 8.0 buffer at 37°C. Solid lines represent nonlinear regressional fits to the data by the standard Michaelis− Menten Eq 4 to yield K_M and V_MAX. See details in Materials and Methods.

Fig 5. Relative inhibition potency and reversibility of NSGM 13 toward human FXIIIa and related enzymes.

(A) Residual activity (%) of four enzymes in the presence of inhibitor 13. The inhibition of FXIIIa (●), factor Xa (□), thrombin (♦), and papain (Δ) by inhibitor 13 was studied by the corresponding assays at pH 7.4 and 37 °C as described in the Materials and Methods. Solid lines represent the sigmoidal dose−response fits (Eq 1) to the data to obtain the values of IC₅₀, ΔY, and HS. (B) Reversibility assay was performed at inhibitor 13 concentration of 100 μM (●) by adding increasing concentration of polybrene (0–3 mg/mL). Shown also the residual FXIIIa activity in the presence of only polybrene (○) or only inhibitor 13 (100 μM) (□). Solid lines represent the sigmoidal fits to the data to obtain IC₅₀ (or EC₅₀), HS, Y_M, and Y_O using Eqs 1 or 5, as described in the Materials and Methods.

Fig 6. Effect of NSGM 13 on FXIIIa-mediated fibrin crosslinking.

(A) The formation of crosslinked fibrin upon addition of aqueous solution of thrombin (2.5 μg/mL) to aqueous solution of fibrinogen (10 mg/mL) and human FXIIIa (2.0 μg/mL) was measured spectrophotometrically at 405 nm and 25°C. The reaction was carried out in 50 mM TrisHCl buffer of pH 7.4 containing 10 mM CaCl₂ and in the presence or absence of thrombin, FXIIIa, or NSGM 13 (5, 50, and 500 μM). (B) Relative fibrin crosslinking (%) of a solution containing fibrinogen (10 mg/mL), FXIIIa (2.0 μg/mL), thrombin (2.5 μg/mL), and 10 mM CaCl₂ in the presence of various concentrations of NSGM 13 was determined by measuring the absorbance of each well at 405 nm, pH 7.4, and 25°C. Solid lines represent sigmoidal fits to the data to obtain IC₅₀, HS, Y_M, and Y_O using Eq 1. See details in Materials and Methods.