A Synthetic Heparin Mimetic That Allosterically Inhibits Factor XIa and Reduces Thrombosis In Vivo Without Enhanced Risk of Bleeding

Abstract

Background:

Human factor XIa (FXIa) is an actively pursued target for development of safer anticoagulants. Our long-standing hypothesis has been that allosterism originating from heparin-binding site(s) on coagulation enzymes is a promising approach to yield safer agents.

Objectives:

To develop a synthetic heparin mimetic as an inhibitor of FXIa so as to reduce clot formation in vivo but not carry high bleeding risk.

Methods:

We employed a gamut of methods involving synthetic chemistry, biophysical biochemistry, enzyme assays, blood and plasma coagulation assays, and in vivo thrombosis models in this work.

Results:

SCI, a non-saccharide mimetic of heparin, was synthesized in three steps in overall yields of >50%. SCI inhibited FXIa with potency of 280 nM and preferentially engaged FXIa’s heparin-binding site to conformationally alter its active site. SCI inhibition of FXIa could be rapidly reversed by common antidotes, such as protamine. SCI preferentially prolonged plasma clotting initiated with recalcification, rather than thromboplastin, alluding to its intrinsic pathway-based mechanism. Human blood thromboelastography indicated good ex vivo anticoagulation properties of SCI. Rat tail bleeding and maximum-dose-tolerated studies indicated that no major bleeding or toxicity concerns for SCI suggesting a potentially safer anticoagulation outcome. FeCl₃-induced arterial and thromboplastin-induced venous thrombosis model studies in the rat showed reduced thrombus formation by SCI at 250 micrograms/animal, which matched enoxaparin at 2500 microgram/animal.

Conclusions:

Overall, SCI is a highly promising, allosteric inhibitor of FXIa that induces potent anticoagulation in vivo. Further studies are necessary to assess SCI in animal models mimicking human clinical indications.

Introduction

Thrombosis or thrombosis-related clinical disorders continue to be the major cause of health-related mortality in developed and under-developed countries. Since the time preceding the Food and Drug Administration (FDA), the drug that has offered relief from thrombotic episodes is heparin, a highly sulfated, polymeric natural product obtained from pigs. Nowadays, a handful of anticoagulants are clinical used including warfarin, low molecular weight heparins, hirudin and its analogs, fondaparinux and direct oral anticoagulants (DOACs) such as dabigatran, rivaroxaban, apixaban and edoxaban. The DOACs offer excellent options for treatment of thrombosis-related indications in cancer as well as non-cancer patients [1]. In fact, in combination with heparins and vitamin K antagonists (VKAs), i.e., warfarin, the anticoagulation therapy has achieved considerable success in addressing the scourge of thrombotic episodes.

Yet, bleeding averse anticoagulation continues to be a significant challenge. While some reports suggest improvement in bleeding consequences with DOACs as compared to VKAs [2,3], others suggest no major improvement [4,5]. Another challenge with DOACs is the difficulty of achieving rapid reversal in incidences of excessive bleeding. Although two agent-specific reversal options (idarucizumab for dabigatran; andexanet alfa for oral factor Xa inhibitors) have become available, both are biologics that may carry their own risks and expenses. Non-specific antidote strategies, although available, do not appear to be exhibit consistency meriting a major recommendation [6]. Additionally, DOACs and VKAs are challenging to use in surgical procedures that require rapid induction and, if necessary, reversal of anticoagulation intervention. Thus, despite the recent advances in anticoagulation therapy, better agents that exhibit fewer bleeding complications, more consistent reversibility, faster onset of action, etc. are desirable.

A growth concept in anticoagulant drug discovery is that it may be better to target coagulation factors other than factor Xa (FXa) and thrombin, the two proteases inhibited by all FDA-approved anticoagulants. In fact, active site inhibitors (also called orthosteric inhibitors) are being actively pursued against several coagulation factors including factor XIa (FXIa), factor VIIa (FVIIa), factor IXa (FIXa), etc [7–9]. Of special note is the massive effort that is currently underway to discover active site inhibitors of FXIa, which has realized promising agents [10–14]. For example, two FXIa inhibitors (BMS986177 and EP-7014) are currently in clinical trials [15]. It is likely additional FXIa active site inhibitors will reach clinical trials in the years to come.

Human FXIa is a particular attractive target because it is structurally unique among all coagulation proteases. It is a disulfide-linked symmetric homodimer, wherein each subunit is comprised of four apple domains (A1→A4) and a C-terminal catalytic domain of 238 residues [16–18], which bears structural similarity to other proteases including thrombin and FXa. The catalytic domain sits as a “cup” in the ~60 Å wide “saucer” formed by the four apple domains, of which the A4 domain is connected through a symmetric Cys321 bridge to the second subunit [16]. The role of the dimeric structure in hemostasis/thrombosis is not fully understood but there is evidence to suggest its importance in vivo [19].

Another distinguishing feature of FXIa is the presence of two anion-binding sites (ABSs) that interact with polyanions such as polyphosphate [20,21], heparin [22–24] and nucleic acids [25]. ABS1 has been identified on the A3 domain in the Arg250-Ile-Lys-Lys-Ser-Lys255 sequence, whereas ABS2 is present in the catalytic domain and involves residues Lys529-Arg-Tyr-Arg532. Both these sequences are classic Cardin-Weintraub sequences known to bind to heparin with high affinity [26]. Interestingly, ABS1 is also engaged by the extracellular domain of platelet glycoprotein Ibα [27], which implies a possible role in cross-talk with platelets. Although the exact reason for the in vivo role of the two ABSs in FXIa remains unclear, both have been shown to contribute to the regulation of FXIa activity. Engagement of either ABS modulates FXI autoactivation and FXIa inhibition by serpins [28–30]. The rates for both processes – autoactivation and serpin inhibition – are enhanced several-fold by heparin. Also both processes depend on the polymeric chain of heparin, which alludes to a template-mediated mechanism to bridge the two interacting protein partners. Yet, the ABSs, especially ABS2, may also support charge neutralization and/or allosteric mechanisms in mediating their functional role [30].

Our long standing hypothesis has been that allosteric modulation of coagulation proteases through their heparin-bindings sites offers novel opportunity of developing new anticoagulants with potentially reduced adverse effects [31–39]. Allosteric inhibition offers advantages over orthosteric inhibition because of the possibility of controlled modulation of protease activity, as demonstrated recently for thrombin [31,32]. Whereas active site inhibitors offer only one parameter (dose or potency) as the modulator of protease activity, allosteric inhibitors offer two independent parameters (potency and efficacy). This mechanistic opportunity coupled with the observation genetic deficiency of functional FXI (hemophilia C) results only in mild bleeding consequences [40] supports the notion that allosteric inhibition of FXIa is likely to be a better therapeutic approach than the traditional active site-mediated thrombin/FXa inhibition.

In this report, we present sulfated chiro-inositol (SCI) as an allosteric inhibitor of FXIa. SCI is a synthetic, homogeneous agent that exhibits characteristics of high potency (~280 nM), excellent selectivity (>100-fold against related factors) and good reversibility with protamine (>50% reversible). SCI preferentially engaged heparin-binding site on FXIa to conformationally alter its active site. Rat tail bleeding and maximum-dose-tolerated studies indicated that SCI exhibits no major bleeding or toxicity concerns suggesting a potentially safer anticoagulation regimen, while FeCl₃-induced arterial and thromboplastin-induced venous thrombosis model studies in the rat indicated that SCI at 250 micrograms/animal dose reduces thrombus formation almost equal to enoxaparin at 2500 microgram/animal. Overall, SCI is a highly promising novel allosteric inhibitor of FXIa that induces potent anticoagulation in vivo. Further studies are necessary to assess SCI in animal models mimicking human clinical indications.

Results

Rational design and chemical synthesis of SCI.

We have recently shown that the ABSs of coagulation proteases offer promising opportunities of developing novel allosteric anticoagulants [31,32,37,38]. Specifically, our drug discovery efforts led to three different types of heparin mimetics as promising inhibitors of human FXIa [33–35,38,39]. Of the three, sulfated pentagalloyl glucoside (SPGG) demonstrated the most promise with an in vitro IC₅₀ of 1.1 μg/ml [34,38]. Yet, SPGG is clinically not viable because it is a heterogeneous mixture of a large number of species with varying levels of sulfation (Fig. 1A and Supplementary Fig. S1). More importantly, later studies indicated that SPGG’s high heterogeneity may actually be reducing selectivity of binding [34].

Figure 1.

(A) Structure of SPGG showing variation in level of sulfation arising from incomplete sulfation of the 3,4,5-trihydroxy phenyl ring. (B) Structure of SCI, which is fully sulfated at all available –OH groups on the polyphenolic precursor. (C) SCI was synthesized in three steps (benzylation, deprotection and sulfation) followed by exchange of quaternary amines for Na⁺ on a resin. (D) The RP-IP UPLC-ESI-MS profile shows only 1687.86 amu species corresponding to [M plus 12 ion-pairing molecules]²⁺, which implies the presence of 12 sulfate groups in SCI (see panel C of Fig. S1). Conditions for resolution and detection of mass peaks are described in Methods. (E) The 254 nm UV profile of the UPLC chromatogram of SCI also shows a single peak further confirming homogeneous nature.

Thus, we reasoned that deriving a chemically distinct, homogeneous agent would be important for establishing the clinical viability of the concept of heparin mimetic-based allosteric inhibition of FXIa. We hypothesized that the 4-position of the 3,4,5-trihydroxy phenyl ring of SPGG is the reason for its heterogeneity. Because chemical per-sulfation of all three adjacent –OH groups is likely to be difficult, SPGG species may tend to carry sulfation at 3 and 5 positions only. Thus, a simple approach to derive a homogeneous SPGG analog would be to use a substituent without the 4-OH group.

Using this as the basis for 2^nd generation designs, we reviewed several core scaffolds (e.g., saccharide and pseudo-saccharide), stereochemistry of substituents on scaffolds (glucose, galactose, inositol, chiro-inositol, myo-inositol, etc.) and level of sulfation (4 to 15 sulfates per molecule). Based on three-dimensional similarity with parent lead agent (SPGG), we identified SCI (Fig. 1B) as a potentially equivalent analog that could remove most of the faults of SPGG, while also possibly maintaining good anticoagulant potency.

SCI was synthesized in high yield (>65% overall) from readily available chiro-inositol using a 3-step protocol involving esterification, hydrogenation and sulfation (Fig. 1C) [41]. The RP-IP UPLC-ESI MS profile showed the SCI is homogeneous species (1688 m/z; Figs. 1D and 1E) made up of only dodecasulfated species, which was further confirmed by ¹H and ¹³C NMR (see Supplementary Figs. S1 – S3). This implied that our hypothesis on the origin of heterogeneity in SPGG was likely to be correct. Thus, if SCI was at least as promising as SPGG, a major heterogeneity bottleneck besetting SPGG as a drug was likely to be overcome.

SCI is a potent inhibitor of human FXIa cleavage of peptide and macromolecular substrates.

SCI was evaluated for its potential to inhibit full length, wild type, purified human FXIa using S-2366, a chromogenic substrate, at pH 7.4 and 37 °C. SCI reduced FXIa activity in a classic dose–dependent manner, which yielded in an IC₅₀ of 0.28±0.01 μM with an efficacy of 100% (Fig. 2A). Interestingly, this potency is ~2-fold better than that of SPGG, which is an added advantage. To ensure that the observed inhibition of FXIa is not specific to S-2366 substrate, we utilized a sub-optimal chromogenic substrate (Spectrozyme TH) and observed potent inhibition of FXIa (see Fig. S4). Likewise, the presence of SCI dose-dependently inhibited FXIa cleavage of its in vivo macromolecular substrate FIX to FIXa (see Fig. S5).

Figure 2.

(A) Direct inhibition of full-length FXIa by SCI and SPGG. The inhibition of FXIa by SCI (●) and SPGG (◊) was studied at pH 7.4 and 37 °C, as described in Methods. Solid lines represent sigmoidal dose–response fits using equation 1 to the data to calculate the IC₅₀, ΔY, and HS. (B) Michaelis–Menten kinetics of S-2366 hydrolysis by full length, wild type FXIa in the presence of SCI. The initial rate of hydrolysis at various substrate concentrations was measured in pH 7.4 Tris-HCl buffer, as described in Methods. Solid lines represent nonlinear regressional fits to the data using the standard Michaelis–Menten equation to calculate the K_M and V_MAX. (C) Spectrofluorimetric measurement of the affinity of full-length FXIa and DEGR-FXIa for SCI at pH 7.4 and 37 °C using the intrinsic tryptophan fluorescence (FXIa, λ_EM=330 nm, λ_EX=280 nm) or dansyl fluorescence (DEGR-FXIa, λ_EM=544 nm, λ_EX=345 nm). Solid lines represent nonlinear regressional fits. (D) Quenching of dansyl fluorescence of DEGR-FXIa by acrylamide in the absence (□) and presence of 2 μM SCI (●). Fluorescence intensity at 544 (λ_EX=345 nm) was recorded following sequential addition of acrylamide. Solid lines represent fits to the data using the Lakowicz equations.

SCI is a selective, non-competitve inhibitor of human FXIa.

The selectivity of SCI for FXIa was studied by assessing its influence on human FIIa, FVIIa/TF, FIXa, FXa, FXIIa, FXIIIa, plasma kallikrein, plasmin, trypsin and chymotrypsin. None of these were inhibited at the concentrations ~140-fold higher than the IC₅₀ against FXIa (Table 1). We also studied SCI inhibition of thrombin using a sub-optimal chromogenic substrate (S-2366) and observed essentially no inhibition (see Fig. S6). This implied that SCI was more selective in targeting human FXIa than SPGG, which appeared to be an additional advantage of homogeneity over heterogeneity. To assess this selectivity further, we studied the effect of SCI on plasma clotting times in the presence of added FXI neutralizing antibody. We selected a FXI neutralizing antibody dose that induced ~2-fold increase in APTT (9 μg per 300 μL) and then varied SCI concentrations over a wide range. We found that SCI did not significantly impact plasma clotting times even at concentrations as high as 100 μM (see Fig. S7), which implied that SCI’s primary effect arises from targeting FXIa.

Table 1.

SCI inhibition of coagulation factors related to human FXIa.^a

Human Enzyme	IC50 (μM)
FXIa	0.28 ± 0.01b
Thrombin	>40c
FXa	>40
FVIIa/TF	>50
FIXa	>65
FXIIa	>50
FXIIIa	>90
Plasmin	>40
Plasma Kallikrein	>65
Trypsin	>50
Chymotrypsin	>40
rFXIa‒FLWTd	4.2 ± 0.6
rFXIa‒FLABS1d	3.0 ± 0.2
rFXIa‒FLABS2d	20.7 ± 3.6

^aIC₅₀s were obtained following non-linear regression analysis of direct inhibition of human enzymes in optimal conditions reported in the literature. Inhibition was followed using spectrophotometric measurement of hydrolysis of appropriate substrate by the enzyme.

^bErrors represent ± 1 S.E.

^cLess than 10 % inhibition was observed at the highest concentration of SCI studied.

^dFL_WT: Full length FXIa wild type, FL_ABS1: Full length FXIa missing A3 domain’s HBS, and FL_ABS2: Full length FXIa missing catalytic domain’s HBS.

We then studied the kinetics of S-2366 hydrolysis in the presence of SCI at pH 7.4 and 37 °C (Fig. 2B and Supplementary Table S1), which showed that S-2366’s K_M remained essentially constant (~0.3 mM), while V_MAX decreased ~4-fold (52.1 to 14.2 mAU/min). This means that SCI does not compete with S-2366 for binding to the active site of FXIa, while inducing dysfunction in its catalytic apparatus. This implies that SCI inhibits human, wild-type, full-length FXIa through a non-competitive mechanism.

SCI binds directly to human FXIa.

To confirm that inhibition arises from direct binding of SCI to FXIa, we measured its affinity (K_D) using fluorescence spectroscopy, a tool that has been routinely used to measure the affinity of sulfated ligands for coagulation factors [42]. We measured SCI’s affinity for FXIa and active site-modified FXIa containing the dansyl group (DEGR-FXIa) at pH 7.4 and 37 °C. The interaction of SCI with FXIa resulted in a saturating increase of 133% in the intrinsic fluorescence, which corresponded to a K_D of 63±11 nM (Fig. 2C, Table S2). A comparable 152% increase in dansyl fluorescence of DEGR-FXIa was observed, which yielded an affinity of 25 nM (Table S2). These K_D values are nearly 4-fold lower than the measured IC₅₀. For allosteric inhibitors, K_I and IC₅₀ are expected to be essentially similar, which has been generally observed for allosteric ligands binding to thrombin, a monomeric protein [31,32]. Yet, the dimeric form of FXIa with two possible binding sites for SCI (ABS1 and ABS2) may make the system different from thrombin. To assess whether dimeric FXIa binds one or two molecules of SCI, we measured stoichiometry using spectrofluorometric titration and found it to be 1.83±0.08 (not shown). Considering that the IC₅₀ reflects overall inhibition of the dimeric enzyme, whereas fluorometric titration reports on earliest binding events with monomeric units, some difference between IC₅₀ and K_D is to be expected.

Allosteric modulation of the active site.

To assess whether SCI remotely affects the conformation of the active site and surrounding region of FXIa, we studied quenching of dansyl fluorescence of DEGR-FXIa by acrylamide, a small physical quencher. Considering that the dansyl group is located at the P4 position in DEGR-FXIa, the experiment is designed to sense changes in electrostatics and/or sterics near the active site with and without SCI. The fluorescence of dansyl group in SCI–DEGR-FXIa complex remained invariant irrespective of the concentration of acrylamide, whereas significant quenching was observed for FXIa alone (Fig. 2D). In fact, the non-linear Stern–Volmer profile observed for FXIa alone probably implies two slightly different fluorophores, possibly from the dimeric nature of FXIa, which are being differentiated by the quencher. Indeed, it is possible to isolate FXIa with only half-functional unit [16,17]. This suggests that the binding of SCI at one or more site reduces steric access of the dansyl fluorophore to acrylamide. These results suggest that SCI allosterically inhibits human FXIa.

SCI preferentially engages residues of the ABS2.

Our hypothesis for pursuing heparin mimetics required interaction with either or both of the ABSs on FXIa. To test this expectation, we studied competition between SCI and UFH for binding to FXIa using the S-2366 hydrolysis assay at pH 7.4 and 37 °C. The IC₅₀ of FXIa inhibition increased from ~6-fold as UFH increased from 0 to 150 μM (Fig. 3A, Table S3). To further identify the preferred ABS, we utilized two full-length, recombinant FXIa block mutants, rFXIa-FL_ABS1 and rFXIa-FL_ABS2 (Fig. 3B). The two mutants have previously yielded key insights into heparin binding to FXIa [16–28]. In comparison to the recombinant wild-type enzyme (rFXIa-FL_WT), the ABS2 mutant (rFXIa-FL_ABS2) was inhibited by SCI ~5-fold weaker, while also displaying a reduction of ~20% in efficacy (Fig. 3C, Tables 1 and S4). In contrast, SCI inhibited rFXIa–FL_ABS1 with a potency essentially identical to that of rFXIa-FL_WT. Considering that Arg/Lys mutations were introduced at only select positions in both ABSs, the results indicate that SCI prefers ABS2, which is present in the catalytic domain of FXIa. Thus, of the several basic residues of ABS2, at least three (Arg529, Arg530, and Arg532) appear to be important for SCI binding.

Figure 3.

(A) Competitive direct inhibition of FXIa by SCI in the presence of UFH. The FXIa inhibition was measured spectrophotometrically at pH 7.4 and 37 °C. Solid lines represent fits by the dose–response equation 1 to calculate the IC₅₀, ΔY, and HS, as described in Methods. The concentrations of UFH selected for the study are 0 (●), 5 (□), 50 (▲), 150 (◊) μM. (B) A space filling model of a FXI monomer (PDB ID: 2F83) showing the positions of Lys and Arg residues in blue color, which are involved in anions binding: an anion-binding site in the A3 apple domain (ABS1) and an anion-binding site in the catalytic domain (ABS2). Active site is highlighted in orange color. Note: Full-length wild-thpe FXI is a homodimer, of which only the monomer is shown for clarity. (C) Direct inhibition of full length, recombinant FXIa (rFXIa-FL) by SCI in Tris-HCl buffer of pH 7.4 at 37 °C. Shown is the inhibition of rFXIa-FL_WT (wild type) (●), rFXIa-FL_ABS1 (apple domain block mutant) (□), and rFXIa–FL_ABS2 (catalytic domain block mutant) (♦). Solid lines represent dose-response fits of equation 1 to the data to derive the IC₅₀, HS, and ΔY.

SCI inhibition of FXIa is amenable to rapid reversal.

Rapid reversibility through the use of an antidote is a highly desirable aspect of anticoagulation therapy. To assess whether FXIa inhibition by SCI can be rapidly reversed, we studied two polycationic polymers (polybrene and protamine) and one anionic small competitive agent (sucrose octasulfate, SOS). First, FXIa’s activity was 90% suppressed with a saturating concentration of SCI. Then, varying antidote levels were introduced and restoration of activity was studied using S-2366 hydrolysis assay. Figure 4A shows the dose–response profiles from which restoration yield (ΔY) were calculated. SCI inhibition of FXIa could be reversed ~88% by polybrene, ~69% by protamine sulfate and ~39% by SOS (Table S5). The restoration potencies (EC₅₀) were calculated to be 0.4, 14.0, and 750 μg/mL for polybrene, protamine, and SOS, respectively. Thus, SCI inhibition of FXIa is amenable to rapid reversal by more than one agents.

Figure 4.

(A) In vitro reversibility of SCI inhibition of human FXIa by potential antidotes. Shown is the restored FXIa activity (%) (inhibited by 2 μM of SCI) in the presence of increasing concentrations of polybrene (♦), protamine sulfate (□), and SOS (●) at pH 7.4 and 37 °C. Solid lines represent fits by the dose-response equation to obtain the EC₅₀, as described in Methods. (B) Effect of SCI on activated partial thromboplastin time assay (APTT, ■) or prothrombin time assay (PT, □) in human plasma. Error bars represent ±1 SE (many smaller than the size of the symbol). Clotting assays were performed as described in Methods.

SCI preferentially prolongs the APTT of human plasma.

Traditionally, FXIa is considered to be part of the intrinsic pathway of coagulation. Thus, we expected a preferential effect on the APTT in comparison to the PT. Figure 4B shows the dose-dependent variations in the APTT and PT of human plasma. A 2-fold increase in APTT required 32.2 μM, whereas SCI’s impact on the PT was minimal even at 150 μM. Likewise, SCI was found to preferentially enhance APTT of rat plasma in a manner similar to enoxaparin (see Fig. S8). We also measured SCI effect on human plasma deficient of antithrombin and heparin cofactor II, two serpins known to be important for mediating UFH’s anticoagulant function. SCI preferentially prolonged the APTT of plasma deficient in either serpins, as expected of a FXIa inhibitor (see Fig. S9).

Human blood thromboelastography (TEG) and hemostasis analysis (HAS) of SCI.

TEG, an ex vivo technique that measures thrombodynamic properties of blood under a low–shear conditions, has been approved by the FDA for diagnosis and management of coagulation disorders. It is also used during surgery [43,44]. In contrast, HAS is an ex vivo technique that elucidates forces generated by platelets within a clot [31,45]. To assess SCI’s promise, we utilized both TEG and HAS. The effects of SCI on human whole blood with respect to TEG parameters (time to initial fibrin formation, clot formation angle, maximum amplitude, and modulus strength) and HAS parameters (force onset time, platelet contractile force, and clot elastic modulus) were studied. SCI was found to increase time to initial fibrin formation, while decrease all other parameters, as expected for an anticoagulant that significantly reduced fibrin polymerization and clot integrity (Fig. S10 and Table S6). In fact, these changes were similar to those observed in TEG analysis of blood from FXI-deficient patients [46,47]. Likewise, in the HAS study SCI reduced platelet contractile force and clot elastic modulus, whereas prolonged force onset time, as one would expect for a promising anticoagulant (see Fig. S11 and Table S7). These ex vivo results suggest that SCI is expected to be a good anticoagulant in human whole blood.

SCI does not exhibit significantly enhanced tail bleeding risk.

In preparation for thrombosis studies, we first performed tail bleeding and maximum tolerated dose (MTD) studies. At 250 μg/rat dose of SCI, a blood loss of 119±41 mg (n=15) was observed, which was not much different from that observed with PBS (147±29 mg; n=15) (Fig. 5A). Likewise, the tail bleeding times for the two treatments were also not much different (SCI = 14±4 min and PBS 13±4). For enoxaparin, a dose of 500 μg/rat induced nearly 6–7-fold higher bleeding, while higher doses, which are more relevant for anticoagulant effect, exhibited further increases in tail bleeding (see Fig. S12). Similar results were observed in mouse tail bleeding studies, wherein both blood loss and bleeding time induced by SCI were much lower than that induced by enoxaparin (Fig. 5B).

Figure 5.

Box and whisker plots showing the effect of SCI on rat ((A) and (B)) and mouse ((C) and (D)) tail bleeding. Following IV administration of 0.25 mg/rat SCI (n=15) or saline (n=15), the blood loss (in mg, (A)) and the optical density of blood samples (B) after the tail transection were measured. Likewise, following IV administration of 35 μg/mouse SCI (n=4 – 7) or 300 μg enoxaparin (n= 4 – 7), the blood loss (in mg, (C)) and the time to cessation of bleeding (D) after the tail transection were measured. Statistical analysis was performed using two-tailed t-tests with unequal variance (p<0.01). *The time to cessation of bleeding in nearly each mouse exceeded 20 min.

We then performed MTD studies using three escalating doses of SCI (5, 10, and 25 mg per animal) injected IV to rats (n=3/group). None of the animals displayed any abnormal mobility, weight loss or obvious bleeding over an observation period of 7 days suggesting that high levels of SCI are well tolerated by rats (not shown). SCI’s cellular toxicity was also assessed using HT-29 colorectal cell line. No significant adverse effect, e.g., cell death, was observed in MTT assays at SCI concentrations as high as 100 μM [41] confirming that SCI should be studied further in rat thrombosis models.

SCI is an effective anticoagulant in rat arterial thrombosis model.

We utilized the well-characterized FeCl₃-induced arterial thrombosis model in rats [48,49]. Although several FeCl₃ concentrations (5–15% w/v) have been reported in the literature, we chose 10% FeCl₃, which seems to have been more often used for heparins. Exposure of the aorta in wild-type rats to a 10% FeCl₃ patch resulted in the formation of an occlusive platelet-rich thrombus within 15 min. Pre-injecting SCI ~5 min before the FeCl₃ patch resulted in a distinct and visible reduction in the extent of thrombus formation (Figs. 6A and 6B; Table S8). At 250 μg SCI per rat (n=14), the wet weight of thrombus was found to be 28.85±2.12 mg, which was ~33% lower than that for the negative control (PBS, n=14, 43.35±2.63 mg). In comparison, a 2500 μg/rat dose of enoxaparin gave a wet weight of 18.42±1.82 mg (n=14). Similar results were noted by using the dry thrombus weight for both groups. Escalating the SCI dose to 500 or 2000 μg/animal gave further reductions in extents of thrombus formed, further confirming SCI’s anticoagulation potential (Fig. S13). The average wet and dry thrombus weights decreased by 53–55% and 55–63%, respectively, in comparison to that of the PBS group.

Figure 6.

Box and whisker plots showing the effect of SCI on rat arterial thrombosis. The effects of SCI (0.25 mg/rat) and enoxaparin (2.5 mg/rat) in 10% FeCl₃-induced arterial thrombosis model was followed by measuring wet (A) and dry (B) weights of isolated thrombi (n=14 animals). Statistical analysis was performed using two-tailed t-tests with unequal variance (p<0.01).

SCI is an effective anticoagulant in rat venous thrombosis model.

We also studied SCI in the thromboplastin-induced venous thrombosis model [48,50], which attempts to mimic deep vein thrombosis, the primary indication for use of heparins. Exposure of the vena cava in wild-type rat to thromboplastin resulted in the formation of an occlusive thrombus within 15 min. The clot could be readily prevented by pre-injection of SCI ~5 min before thromboplastin (Fig. 7A and 7B; Table S8). At 250 μg SCI per rat, the wet weight of thrombus was found to be 22.46±1.14 mg (n=13), which was ~42% less than that for the PBS group (n=13, 38.61±3.56 mg). In comparison, a 2500 μg/rat dose of enoxaparin reduced the average wet weight to 20.61±1.89 mg (n=13) representing about 47% reduction relative to that the negative control. The average dry weights of thrombus conveyed the same information (48% and 52% reductions, respectively). This implied that SCI at 250 μg per animal was equivalent to enoxaparin at 2500 μg SCI per animal.

Figure 7.

Box and whisker plots showing the effect of SCI on rat venous thrombosis. The effects of SCI (0.25 mg/rat) and enoxaparin (2.5 mg/rat) in thromboplastin-induced thrombosis in rat vena cava. Thrombosis was induced using 5 μg of rabbit brain thromboplastin in 0.9% saline. Wet (A) and dry (B) weights of isolated clots are shown (n=13 animals). Statistical analysis was performed using two-tailed t-tests with unequal variance (p<0.01). ‘NS’ represents not significant.

Discussion

In one sentence, this work presents reason-based design, synthesis, mechanistic characterization, and anticoagulation (in vitro, ex vivo and in vivo) study of a structurally novel direct, inhibitor of FXIa. SCI is a major improvement over SPGG, the agent identified earlier through screening studies [33,34,38]. Although SPGG displayed good activity, it was a heterogeneous mixture of thousands of species. In contrast, SCI is a homogeneous agent with improved potency (2-fold) and selectivity characteristics. In terms of drug discovery, this is a major point because the FDA is not expected to view heterogeneity in drug composition very favorably. Homogeneity also positions the heparin mimetic better in terms of PK/PD and other IND parameters. The fact that SCI can be synthesized in just three short steps while simultaneously displaying additional benefits is a major milestone toward IND application.

SCI is the first-in-class agent that exhibits good in vivo antithrombotic promise. No allosteric FXIa inhibitor, including SPGG, has displayed this level of promise in vivo. Additionally, SCI shows much less bleeding tendency in both mouse and rat tail bleeding models. SCI also appears to carry lower bleeding risk in comparison to enoxaparin. Finally, SCI inhibition of FXIa could be reasonably reversed with protamine, an antidote currently used in the clinic for heparin. This feature bodes well for its further development as anticoagulant for acute or surgical needs, where rapid induction as well as reversal of anticoagulation is highly desirable.

These highly promising results do not guarantee IND status. Much remains to be accomplished. One, advanced mechanism-related knowledge would be desirable. For example, does SCI bind and influence the precursor of FXIa, i.e., FXI? Both FXIa and FXI bind heparin, albeit with different affinities, and this may imply that SCI binds to FXI too. Will this be pharmacologically significant? Further, will this impact the high molecular weight kininogen–FXI–FXII system? Likewise, does SCI compete with polyphosphates, released from activated platelets, for binding to FXI/FXIa?

Two, detailed pharmacokinetic studies are in order. In the pre-clinical in vivo experiments performed so far, we have utilized time points that are within the 30 min window of infusion. This was important to ensure that 100% of the infused dose was being assayed. However, if SCI persists in circulation for longer time period, e.g., 4 hours or so, it might be advisable to repeat the pre-clinical studied being reported here. It would also be important to assess different modes of administration, e.g., SC and/or PO. Thus, PK studies would be important for further advancement of this potential drug.

Three, advanced toxicity studies would be desirable. Although we studied doses as high as 25 mg/animal (>100-fold effective dose) and found it to not induce bleeding or cellular toxicity, long term toxicity studies at organ level would be important for its clinical viability. This implies that it would be important to pursue chronic toxicity studies using 100-fold effective dose levels.

Overall, several promising FXI- or FXIa-targeting agents (antisense oligonucleotide [51], aptamers [52], and small molecule orthosteric agents [10–15]) are being pursued by several groups. We have presented a radically different approach to antithrombotic activity, which is heparin-binding site based allosteric inhibition of human FXIa. SCI represents a structurally and mechanistically unique entity from all of the agents being pursued so far. Realization of the concept of allosteric disruption of FXIa activity is likely to generate new enthusiasm and newer agents. This is important because although allosterism is nature’s major mechanism for regulation of multi-domain proteins [53], discovery of allosteric anticoagulants [31,32] has remained a challenge. SCI breaks that barrier by showing its in vivo relevance and its clinical promise. We expect that our contribution will help development of allosterism-based direct FXIa inhibitors that function as safer antithrombotics.

Materials and Methods

General.

Detailed materials and methods are described in the online supplementary material. Brief description of these is provided below.

Synthesis and characterization of SCI.

Inositol was esterified using DCC-mediated coupling with di-benzylated 3,5-dihydroxybenzoic acid, followed by debenzylation on palladized carbon to give the polyphenol precursor, which was sulfated under microwave conditions for 12 hr at 90 °C using SO₃–Me₃N complex in the presence of Et₃N to afford SCI–Et₃N in ~65% overall yield [41]. SCI was characterized using NMR (¹H and ¹³C) spectroscopy and RPIP-UPLC-ESI mass spectrometry (see Supplementary Material for details).

Direct inhibition of human FXIa and other proteases.

SCI inhibition of wild-type or recombinant of FXIa was measured using a S-2366 hydrolysis assay at 37 °C in 20 mM Tris-HCl buffer, pH 7.4, from the initial rate of increase in absorbance at 405 nm. SCI inhibition of thrombin, FXa, FIXa, FVIIa/TF, FXIIa, plasmin, plasma kallikrein, trypsin, and chymotrypsin was also studied using similar chromogenic substrate assay, as reported earlier [32,38,39]. SCI inhibition of FXIIIa was evaluated using fluorescence-based, bisubstrate assay, as reported earlier [54]. A logistic equation was used to analyze the dose-dependence of protease activity to calculate the potency (IC₅₀), efficacy (ΔY) and Hill slope (HS) of inhibition.

Affinity of sulfated molecules for FXIa.

Fluorescence titrations were performed in 20 mM Tris-HCl buffer, pH 7.4, at 37 °C by measuring the change in tryptophan fluorescence of FXIa (λ_EM=340 nm; λ_EX=280 nm) or dansyl fluorescence of FXI-DEGR (λ_EM=547 nm; λ_EX=345 nm) arising from the interaction with SCI. The fluorescence changes were analyzed using the traditional quadratic binding equation to calculate the K_D of the complex.

Acrylamide quenching of DEGR-FXIa fluorescence.

The fluorescence of DEGR-FXIa at 547 nm was measured in the presence of fixed concentrations of SCI following the addition of acrylamide (0–800 mM) in 20 mM Tris-HCl buffer, pH 7.4, at 37 °C [42]. The observed fluorescence was analyzed using the classic linear or quadratic Stern–Volmer equations, as described earlier [55], to elucidate the influence of SCI on FXIa.

Reversibility studies of SCI inhibition of FXIa.

In vitro reversibility of SCI inhibition of FXIa was studied by titrating protamine sulfate, polybrene, or SOS into SCI–FXIa complex and measuring restoration of FXIa activity using S-2366 hydrolysis assay at pH 7.4 and 37 °C. The relative restored FXIa activity at each concentration of the reversing agent was calculated from the ratio of FXIa activity in the presence and absence of the antidote and analyzed using the dose-response logistic equation to calculate the potency (EC₅₀) and efficacy (ΔY).

Clotting times of human plasma.

A standard one-stage recalcification assay was used to measure PT and aPTT of human plasma, as reported earlier [38,39]. Thromboplastin-D and CaCl₂ were used to initiate clotting at 37 °C and the times to clot (PT and aPTT) in the presence of varying levels of SCI were measured. The data were fit using a quadratic equation to calculate the concentration of SCI required to double the clotting times.

Thromboelastography (TEG) and hemostasis (HAS) study of human blood.

TEG and HAS assays were performed as reported in the literature. For both studies, CaCl₂ was used to initiate clotting of citrated human blood containing varying levels of SCI or enoxaparin. In TEG, a Haemoscope cup oscillates through a 4° 45ˋ angle at 0.1 Hz and the instrument measures all data [reaction time in min, angle in deg, maximum amplitude, and shear elastic modulus] in an automated manner. In HAS experiments, a cone is lowered into the recalcified blood and platelets attaching to surfaces generate tension within the fibrin meshwork, which is automatically measured in terms of platelet contractile force, thrombin generation time, and clot elastic modulus.

Tail bleeding time studies in rats.

All procedures involving rats were performed by an independent contract research organization (Noble Life Sciences, Gaithersburg, MD) following IACUC approval (Protocol #14-12-001PRI). Tail bleeding studies followed literature reports. Female Wistar rats (weight ~250 gm) were anesthetized with ketamine/xylazine and maintained on anesthesia with isoflurane. Either SCI, enoxaparin or PBS, was injected into the tail vein. After 2 min, the tail (4–5 mm from the end) was clipped and immersed in PBS at 37 °C. The blood lost and the optical density of PBS were recorded for each animal.

FeCl₃-induced arterial and thromboplastin-induced venous thrombosis studies in rats.

Arterial and venous thrombosis model studies have been previously reported. Briefly, female Wistar rats (weight ~250 gm) were anesthetized with ketamine/xylazine and maintained on anesthesia with isoflurane. SCI, or enoxaparin, in PBS was infused into the tail vein. Five minutes after infusion of the anticoagulant, the abdominal aorta was exposed and a 10% FeCl₃-soaked filter paper was applied between two sutures for 10 min. The paper was then removed and the clot allowed to mature for 15 min, the sutures tightened and the clot’s wet and dry weights were measured. For venous thrombosis, two sutures were wrapped around the exposed vena cava ~1 cm apart below the left renal vein. Five minutes after infusion of the anticoagulant, rabbit brain thromboplastin was injected into the tail vein and the two sutures were tightened. After 15 minutes, the vena cava between the two sutures removed and the wet and dry weights of thrombus were recorded.

Highlights.

• The paper reports the realization of the concept of heparin-binding site-dependent allosteric inhibition of human factor XIa for reducing clot formation in vivo

• The small molecule agent devised in this work was synthesized in just three steps from readily available natural product, which opens up a major route to novel class of anticoagulants

• Arterial and venous thrombosis in the rat was obviated by the allosteric inhibitor at 250 micrograms/animal dose without increase in tail bleeding