Smart thrombosis inhibitors without bleeding side effects via charge tunable ligand design

The study addresses the need for antithrombotic therapies that prevent thrombosis without increasing bleeding risk. Conventional anticoagulants and antiplatelet agents disrupt essential hemostatic pathways and are associated with bleeding complications. Polyphosphate (polyP), a highly anionic linear polymer released by bacteria and platelets, accelerates coagulation via the contact pathway and modulates multiple steps in coagulation (e.g., factor V activation, fibrin structure, factor XI activation), yet the contact pathway is not required for normal hemostasis. Thus, selectively inhibiting polyP offers a strategy to attenuate thrombosis with minimal impact on hemostasis. The research goal is to design and validate macromolecular polyanion inhibitors (MPIs) that maintain low cationic charge at physiological pH to minimize off-target effects, but increase charge density upon binding to polyP to achieve high-affinity, selective neutralization, thereby providing antithrombotic efficacy without bleeding.

Prior approaches to inhibit polyP include enzymatic degradation using alkaline phosphatase or bacterial exopolyphosphatase (PPX), which can be slow and risk off-target dephosphorylation of vital molecules like ADP. Cationic polymers such as polyethylenimine (PEI) and polyamidoamine (PAMAM) dendrimers bind polyP electrostatically and reduced thrombosis in mice but required toxic doses, limiting clinical utility. Earlier universal heparin reversal agents (UHRAs) from the authors’ laboratories could inhibit polyP but caused bleeding at therapeutic levels. These limitations underscore the need for polyP inhibitors with high selectivity and biocompatibility. Foundational work on polyamine pKa tuning and microenvironment-dependent protonation informed the present charge-switchable ligand design on a biocompatible polymer scaffold.

Design and library construction: Two cationic binding groups (CBGs), inspired by PMDETA-like alkylamines with tunable protonation, were selected: CBG I (two-carbon linker) and CBG II (three-carbon linker). These were conjugated at varying densities to hyperbranched polyglycerol (HPG) cores bearing a PEG corona (HPG-PEG) to reduce non-specific interactions. Two scaffold sizes (∼10 and 20 kDa) were used to create a library (MPI 1–9), alongside UHRA controls (UHRA-8, UHRA-10). Polymer characterization included NMR, conductometric and potentiometric titrations, and GPC-MALLS. Protonation and binding characterization: Potentiometric titration determined average protonation states per CBG on each MPI at pH 7.4. Surface plasmon resonance (SPR) measured apparent affinity to short-chain polyP (SC polyP) under physiological buffer. Isothermal titration calorimetry (ITC) with defined polyP chain lengths quantified binding thermodynamics (apparent K_D/K_a, ΔH, ΔS, stoichiometry) and assessed proton recruitment by measuring binding enthalpy in buffers of differing heats of ionization. In vitro coagulation assays: Human plasma assays included recalcification clotting with long-chain polyP (LC polyP) to assess inhibitory activity, and calibrated automated thrombography (TGA) to evaluate effects on thrombin generation with LC and SC polyP. TF-initiated clotting in FXII-deficient plasma (without polyP) assessed hemocompatibility and off-target effects; dose-response curves yielded IC50 values (Supplementary Table 4). Whole blood clotting was assessed by rotational thromboelastometry (ROTEM). Fibrin clot morphology and fiber diameter were examined by SEM with/without polyP and MPI. In vivo thrombosis models: Antithrombotic efficacy was evaluated in mice using: (1) laser-induced cremaster arteriole thrombosis (intravital microscopy of platelet and fibrin accumulation after i.v. dosing of MPI candidates); (2) FeCl3-induced carotid artery thrombosis (time to occlusion/patency compared with UHRA-10); and (3) inferior vena cava (IVC) stenosis (partial ligation) model (thrombus weight and weight/length at 48 h). Bleeding and safety: Tail-clip bleeding time and hemoglobin loss in mice compared MPI 8 with saline and unfractionated heparin (UFH). Saphenous vein hemostasis laser ablation model examined hemostatic clot formation. Acute and chronic toxicity studies in mice assessed body weight, serum enzymes (LDH, AST, ALT), and histopathology (H&E) after i.v. administration up to 500 mg/kg.

- Charge-switching behavior: Potentiometry showed each CBG on MPIs bears ~1 positive charge at pH 7.4, with additional amines (pKa ~6–7) available to protonate upon binding. ITC buffer enthalpy analysis indicated significant proton recruitment upon MPI 3 binding to polyP P75: slope ≈ −13.6 (95% CI 7.0–20.3) protons per binding event, increasing cationic charge by ~32% to ~2.6 ± 0.6 charges/kDa in the bound state. - Binding affinity and thermodynamics: SPR demonstrated sub-micromolar apparent affinities to SC polyP for all MPIs, comparable to UHRA controls despite lower unbound charge density. ITC with P45 showed enthalpically driven binding: e.g., MPI 3 apparent K_D ~0.74 µM with ΔH ≈ −107 kcal/mol and ΔS ≈ −98.7 kcal/mol; MPI 5 ~0.95 µM (ΔH ≈ −75 kcal/mol); MPI 7 ~1.27 µM (ΔH ≈ −37 kcal/mol); MPI 9 ~1.2 µM (ΔH ≈ −50.9 kcal/mol). - In vitro inhibition of polyP: MPIs normalized LC polyP-induced acceleration of coagulation in plasma clotting and TGA. Several candidates (e.g., MPI 5, MPI 9) normalized clot times at ≥12.5 µg/mL in LC polyP-accelerated recalcification. Against SC polyP (with low TF), MPIs dose-dependently normalized TGA parameters. IC50 values for MPIs against LC polyP were generally lower than UHRA controls of comparable scaffold size. - Hemocompatibility: In TF-initiated clotting (FXII-deficient plasma) without polyP, MPIs showed clot times similar to buffer; UHRA-8 markedly prolonged clotting (>550 s at higher doses). In TGA without polyP, select MPIs (MPI 1, 2, 4, 6, 8) had minimal effect across 10–50 µg/mL, whereas MPI 3 and MPI 9 affected parameters at higher concentrations. ROTEM of whole blood showed MPI 1, 6, 8 did not alter clot time or maximum clot firmness versus buffer; UHRA-8/10 did. SEM showed MPI 8 prevented polyP-induced increases in fibrin fiber diameter, restoring morphology to control without altering baseline clot structure. - In vivo antithrombotic efficacy: In laser-induced cremaster arteriole injury, 100 mg/kg of MPI 1, 6, or 8 reduced platelet accumulation rates; MPI 8 also reduced fibrin accumulation (n=3 mice/group, ~8 thrombi/mouse). In FeCl3 carotid injury, 100 mg/kg MPI 8 significantly delayed occlusion and improved patency versus control and outperformed UHRA-10; at 200 mg/kg both reached high patency (likely maximal for this model). In IVC stenosis, continuous delivery of MPI 8 significantly reduced thrombus weight (p=0.0014) and weight/length (p=0.0201) compared with vehicle; alternative analysis reported p=0.0003 after outlier removal. - Bleeding and safety: Tail-clip bleeding times and hemoglobin loss with MPI 8 up to 300 mg/kg were similar to saline; UFH significantly prolonged bleeding and increased hemoglobin loss. Acute (24 h) and chronic (15 days) toxicity studies showed no significant changes in body weight, LDH/AST/ALT (acute), or LDH (chronic), and normal histology (heart, lungs, liver, kidneys) at 500 mg/kg i.v., indicating high tolerability.

The findings validate a charge-switchable macromolecular inhibitor strategy that targets polyP, an accelerant of coagulation not essential for hemostasis, to achieve thromboprotection with minimal bleeding risk. By engineering weakly basic CBGs on a biocompatible HPG-PEG scaffold, MPIs maintain low cationic charge under physiological conditions to limit nonspecific interactions, yet recruit additional protons upon engaging the highly anionic polyP to strengthen binding. Biophysical data (SPR, ITC) support strong, enthalpically driven binding with proton recruitment, consistent with ion-pairing and microenvironment-driven protonation. In plasma, MPIs inhibited both long- and short-chain polyP-mediated procoagulant effects without perturbing TF-driven hemostasis, and demonstrated superior hemocompatibility to prior polycations (e.g., UHRAs). In vivo, the lead MPI 8 reduced thrombus growth in arterial and venous models, delayed carotid occlusion, and decreased IVC thrombus burden, all without prolonging bleeding, supporting the central hypothesis that selective polyP neutralization can prevent thrombosis while preserving hemostasis. The platform’s modularity (ligand pKa tuning, scaffold size, ligand density) enables optimization of selectivity and efficacy.

This work introduces macromolecular polyanion inhibitors (MPIs) with charge-tunable ligands that selectively neutralize polyP, achieving potent antithrombotic effects without bleeding in mouse models. The lead candidate, MPI 8, binds polyP with sub-micromolar affinity, exhibits protonation switching upon binding, inhibits polyP-driven coagulation across chain lengths, preserves normal hemostasis in vitro and ex vivo, reduces thrombosis in multiple in vivo models, and is well tolerated at high systemic doses. The design strategy—minimizing baseline charge while enabling target-induced charge increase on a sterically shielded scaffold—offers a path to safer antithrombotics. Future work should optimize pharmacokinetics and dosing, evaluate efficacy across broader thrombosis indications and comorbid conditions, and advance translational studies toward human safety and efficacy.

- Quantification of proton recruitment during MPI–polyP binding is limited by the dispersity of both polymeric partners, preventing precise determination of protons per binding event (ITC slope provides an estimate with broad confidence intervals). - While MPIs improved patency in the FeCl3 carotid model, complete patency is not expected with polyP inhibitors since polyP is an accelerant and not required for clotting, which may cap observable efficacy in this model. - Some MPI candidates (e.g., MPI 3 and MPI 9) exhibited off-target effects in thrombin generation at higher concentrations, indicating a need for careful candidate selection and dosing. - Efficacy and safety were demonstrated in murine models; human pharmacology, dosing, and long-term outcomes were not assessed in this study.