Key Points
-
•
A high-throughput screening of 2300 chemical compounds identified adapalene as a procoagulant in severe HA plasma.
-
•
The mechanism of action of adapalene involves FXII activation.
Visual Abstract
Abstract
Hemophilia is a rare bleeding disorder due to factor VIII (FVIII) or FIX deficiency involved in hemophilia A (HA) or HB, respectively. Treatment has long relied on invasive IV infusions of the missing factor for prophylaxis or bleeding treatments. Despite significant progress in patient care, there is still no oral procoagulant drug available for people with hemophilia. Such an oral procoagulant could provide benefits to other inherited and acquired bleeding disorders, as well as anticoagulant-induced bleeding. To find chemical compounds that could become potential future orally administered procoagulants, we designed a hierarchical high-throughput screening protocol combining 2 successive screening filters: a miniaturized fibrin formation assay, followed by a thrombin generation assay, both on severe HA plasma. We screened 3 chemical collections totaling > 2300 chemical compounds; we identified adapalene, a commercialized antiacneic compound (Differin), which is strongly hydrophobic. To design a drug that could be orally administered, we developed a series of chemical analogs of adapalene, and 3 of them, with similar procoagulant activities in FVIII-deficient plasma, showed improved solubilities. The mechanism of action was thoroughly investigated by a series of thrombin generation and enzymatic assays. These studies conclude that the procoagulant activity of the chemical compounds in FVIII-deficient plasma is due to the activation of FXII.
Introduction
Hemophilia is a rare inherited bleeding disorder characterized by repetitive spontaneous or post-trauma bleeding in joints and muscles. A deficiency in coagulation factor VIII or IX (FVIII or FIX) is responsible for hemophilia A or B (HA or HB), respectively. These factors are both involved in the FX activation step of the intrinsic pathway of the coagulation cascade.
Major progress has been achieved in the last decade in the treatment of hemophilia,1 and contemporary therapies were recently reviewed.2 Recently approved gene therapies in HA and HB open up new treatment perspectives.3, 4, 5, 6, 7
Routinely, hemophilia is treated by replacement therapies of missing coagulation factors by IV infusions. Recent innovative subcutaneous treatments include nonfactor therapies8 based on compounds that either mimic activated FVIII (emicizumab, a humanized bispecific monoclonal antibody)9 or block the major physiological anticoagulants, namely antithrombin and tissue factor pathway inhibitor (TFPI). Fitusiran, a small interfering RNA that knocks down antithrombin expression, was approved by the US Food and Drug Administration in 2025,10 whereas the antibody concizumab, targeting TFPI, was shown to be safe and effective in phase 2 and 3 clinical trials for the prophylactic treatment of patients with HA and HB with or without alloantibodies.11, 12, 13
Despite significant progress in patient care, there is still no available oral procoagulant drug. Such an oral procoagulant could provide benefits to other inherited and acquired bleeding disorders, as well as anticoagulant-induced bleeding. In this study, we describe an automated high-throughput screening (HTS) of chemical libraries to identify small chemical compounds that are able to restore coagulation in severe HA plasma, with the aim of developing orally administered procoagulants.
A hierarchical screening procedure involving 2 successive and orthogonal coagulation assays on severe HA plasma was developed, combining first a miniaturized fibrin formation assay, followed by a thrombin generation assay (TGA). Three chemical libraries were screened: the Prestwick Chemical Library (1280 off-patent drugs with high chemical and pharmacological diversity, as well as known bioavailability and safety), the chemoinformatics-based protein-protein inhibitor–focused subset of the Life Chemicals library (“Rule of Four”: 800 compounds), and an in-house library of natural product–like compounds (240 compounds) representative of the laboratory’s expertise.
The screening procedure allowed identification of adapalene, which restores coagulation in severe HA plasma. Further coagulation assays showed that it also increases thrombin generation (TG) in HB plasma. Adapalene is commercialized as an antiacneic compound (Differin).14, 15, 16 It is strongly hydrophobic, which makes it inappropriate for an oral drug; therefore, we undertook a pharmacomodulation study. A series of chemical analogs of adapalene were synthesized, and 3 of them showed similar coagulation profiles and improved solubilities. The mechanism of action of adapalene was investigated by a series of TGAs on different factor-deficient plasmas, compound spiking experiments, and chromogenic substrate hydrolysis assays, evidencing that its procoagulant activity results from activation of FXII.
Materials and methods
Primary screen
We developed a primary screen on a robotic integrated system (CMBA platform, Grenoble, France) for use in a 384-well plate format (reference: Greiner 781101; D. Dutscher). A single batch of severe HA plasma (Cryopep) was used throughout the assay. Fibrin formation was followed by the absorbance at 600 nm. Each plate contained 32 negative controls (with 0.5% dimethylsulfoxide [DMSO]) and 32 positive controls. All compounds were tested in duplicate.
Each well contained 3 μL of each compound diluted in DMSO (to reach a target concentration of 50μM) and 24 μL of a mixture of PPP reagent low (Diagnostica Stago) as a source of tissue factor (0.82pM final) and phospholipids (2.44μM) diluted in citrated HA plasma and and preincubated for 10 minutes at 37°C. Careful mixing was conducted by aspiration/dispense. After 10 minutes of incubation at 37°C, fibrin generation was induced by addition of 5 μL of 20mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid) at pH 7.3, 100mM CaCl2, and 0.25 % bovine serum albumin (final CaCl2 concentration of 16mM). After centrifugation for 1 minute at 3000g to remove any bubbles, the absorbance at 600 nm was read every 5 minutes at 37°C for 90 minutes using the Tecan Infinite M1000 microplate reader. Fresh plasma-PPP was renewed every 2 screening plates. At each time point, the Z’ factor was calculated to analyze data from the time point giving the highest Z’ factor.17 We analyzed the data at 600 seconds, which showed the highest Z’ factor value, using in-house software. A hit was considered positive when the absorbance value at this selected time was >3. The compounds were ranked according to their responses in duplicate wells.
Secondary screen
TGAs were performed as a secondary screen on the 60 selected compounds at 37°C using the new fully automated ST Genesia analyzer (Diagnostica Stago), a benchtop analyzer using STG-Bleedscreen reagent (Diagnostica Stago) corresponding to a low tissue factor concentration (exact concentration not provided by the manufacturer). TG is based on continuous measurement of the fluorogenic signal using a fluorogenic substrate (ZGGR-AMC) cleaved by thrombin, generated after addition of a coagulation trigger (tissue factor + phospholipids + CaCl2). The calibration was performed with the STG-Cal&Fluo kit (Diagnostica Stago) and with STG-FluoSet (Diagnostica Stago) according to the manufacturer’s instructions. Briefly, 475 μL of plasma from 1 patient with HA obtained from Cryopep was spiked with 25 μL of a molecule to reach a target concentration of 50μM or with DMSO (negative control, 0.4% DMSO). Each sample was then tested with the STG-Bleedscreen reagent on the ST Genesia analyzer. Positive control was the same plasma spiked with plasmatic FVIII (Factane; LFB, Les Ulis, France) at a concentration of 1 IU/mL.
Adapalene and derivatives
Stock solutions of adapalene or derivatives were at a concentration of 12.5mM in 100% DMSO. These were extemporally diluted in a buffer (18mM HEPES, 135mM NaCl, pH 7.35) to reach a target concentration of 1mM. A vehicle preparation was produced with 8% DMSO in the buffer (18mM HEPES, 135mM NaCl, pH 7.35) for the negative control of each experiment. For all TGAs, the 1mM adapalene or derivatives solutions and the vehicle were diluted 1:20 in plasma (0.4% DMSO at final concentration).
TGAs by CAT
TGAs were performed by calibrated automated thrombinography (CAT) (Diagnostica Stago) on an automated fluorometer (Fluoroskan Ascent, ThermoLab Systems, Franklin, MA). Coagulation was triggered by 1pM tissue factor and 4μM phospholipids (PPP Reagent Low; Diagnostica Stago). All TGAs were run in triplicate in 96-well plates in standard conditions: 80 μL of plasma were mixed with 20 μL of PPP Reagent Low. Plates were incubated for 10 minutes at 37°C in the automated fluorimeter before adding the fluorogenic substrate (ZGGR-AMC) and calcium (FluCa kit, Diagnostica Stago). Raw data were analyzed using the Thrombinoscope version 5.0 software (Thrombinoscope BV). The experimental conditions specific to each TGA are detailed below.
CAT in hemophiliac and normal plasma spiked with FVIII-neutralizing antibodies
Severe HA plasma (reference: 6-PPD08C-S; Cryopep) and severe HB plasma (reference: 6-PPD09C-S; Cryopep) were used. FIX spiked in HB plasma (reference: 9-HCIX-0040; Cryopep) was used as a positive control. Plasma from 1 healthy individual was spiked and incubated for 2 hours at 37°C with an anti-FVIII neutralizing monoclonal antibody (40μM, clone ESH4; Cryopep) for the dose-response experiment with a concentration of adapalene ranging between 2μΜ and 50μM.
CAT in plasma spiked with DOACs
Apixaban, rivaroxaban, and dabigatran were obtained from Alsachim (Illkirch Graffenstaden). Plasmas with direct oral anticoagulants (DOACs) were obtained by spiking plasmas from different healthy donors (HD) with DOACs. Apixaban and rivaroxaban were dissolved in DMSO to prepare a 400 μg/mL stock solution that was diluted on the day of the experiment in a buffer (18mM HEPES, 135mM NaCl, pH 7.35) to obtain a 4 μg/mL DOAC working solution; this latter solution was diluted at 1:20 in an aliquot of plasma to obtain a theoretical DOAC concentration of 200 ng/mL in plasma (0.05% DMSO). The plasma was spiked with dabigatran; the latter was dissolved in 1N HCl to prepare a 100 μg/mL stock solution. This solution was subsequently diluted at 1:25 in buffer (18mM HEPES, 135mM NaCl, pH 7.35) on the day of the experiment to obtain a concentration of 4 μg/mL working solution; this latter was then diluted at 1:20 in an aliquot of plasma to obtain a theoretical DOAC concentration of 200 ng/mL in plasma. These plasmas were then spiked with adapalene in a ratio of 1:20 (50μM target concentration).
CAT in plasma of individuals under VKA
The effect of adapalene on TG was tested in plasmas from 2 individuals under vitamin K antagonists (VKA) (international normalized ratio ≈2.5).
CAT in plasma spiked with asundexian
Asundexian, an orally bioavailable FXIa inhibitor18 was purchased from MedChemExpress. HD plasma was spiked with asundexian at a concentration of 10 mg/L (which corresponds to a supratherapeutic plasma level), before the TGA. A stock solution of 1 mg/mL in DMSO was diluted in a buffer solution (135mM NaCl, 18mM HEPES, pH 7.35) to obtain a final concentration of 500 μg/mL, which was used to spike the plasma
CAT in FII-, FV-, FVII-, FX-, FXI-, FXII-, and TFPI-deficient plasmas
Lyophilized FII- and FV-deficient plasmas and FVII-, FX, FXI-, and FXII-immunodepleted plasmas were purchased from Diagnostica Stago. Frozen plasma of a patient with an FXII deficiency was supplied by Cryopep. The vehicle was 0.4% DMSO (final concentration). Plasmas were spiked with adapalene (0.4% DMSO at final concentration). TGAs were performed with PPP Reagent Low. TFPI-depleted plasma (Cryopep) was spiked with either adapalene or vehicle (0.4% DMSO at final concentration).
CAT in PK-deficient plasma
The vehicle was 1.35% DMSO (final concentration). Plasmas were spiked with adapalene (50μM target concentration). For some experiments, prekallikrein (PK)-deficient plasma was spiked with 37.5 μg/mL of PK (corresponding to the normal plasma level) and/or with 10 mg/mL of asundexian to inhibit FXIa before the addition of adapalene.
PK-deficient plasma and PK were purchased from Cryopep. Asundexian was obtained from MedChemExpress.
Thromboelastometry
A thromboelastometric assay using a ROTEM Delta analyzer (Pentapharm GmbH, Munich, Germany) was triggered by calcium chloride (12.5mM final concentration) without any activator. Whole blood from different healthy individuals collected in trisodium citrate tubes (0.109M) was preincubated for 2 hours at 37°C with either 40 μg/mL of anti-FVIII inhibitor (monoclonal antibody, clone ESH-4; Cryopep) or vehicle. Blood was then spiked with adapalene or vehicle at a ratio of 1:20 (50μM target concentration of adapalene), before triggering with CaCl2 at 12.5mM (final concentration).
FXII enzymatic activity assays
FXII activity was measured in an in-house enzymatic assay using a chromogenic substrate (Pefachrome FXIIa/TH5253; Cryopep). Either the vehicle or the chemical compounds (adapalene or compounds 1, 2, or 3) (50μM target concentration) were added to 375nM FXII (Cryopep) in a buffer (18mM HEPES, 135mM NaCl, pH 7.35) with or without PK (0.5 μg/mL, Cryopep). The reaction was initiated with 0.5mM FXII substrate (Pefachrome TH5253; Cryopep). Absorbance at 405 nm was monitored on a Multiskan Sky multiplate reader (Thermo Fisher Scientific) at 37°C in a 96-well plate for 1 hour. This experiment was performed in triplicate.
Inhibition assays of FXa by antithrombin
FXa (125nM or 250nM) was incubated with antithrombin (0nM to 250nM) at 37°C in a buffer (18mM HEPES, 135mM NaCl, pH 7.35) and adapalene (50μM target concentration) or vehicle. Every 10 minutes, 25 μL of the substrate was added (PNAPEP 1025 at 1.2mM final concentration, Cryopep) to 100 μL of the mixture. Absorbance was measured on a Multiskan Sky multiplate reader (Thermo Fisher Scientific) at 405 nm, and the velocity of hydrolysis was expressed in optical density per minute. These assays were performed in triplicate.
Results
Searching for procoagulant chemical compounds with the aim of identifying an oral treatment for hemophilia, 3 chemical libraries totaling 2300 chemical compounds were submitted to an HTS. The first screening filter was based on a miniaturized fibrin formation assay performed with commercial plasma from patients with severe HA in 384-well microplates.
This sensitive primary screen allowed us to quickly select 60 chemical compounds that were then submitted to a secondary screen consisting of a TGA on a ST Genesia platform (Diagnostica Stago) with STG-Bleedscreen reagent. Surprisingly, only 1 compound, namely adapalene, significantly increased TG in severe HA plasma, whereas the other hits only showed background TG.
Adapalene, represented in Figure 1A, is a US Food and Drug Administration–approved retinoid prescribed for the treatment of acne14 (Differin [adapalene 0.1% gel]). Its strong lipophilicity (logPo/w = 8.6),15 partially due to the presence of an adamantyl moiety, is ideal for a cutaneous administration.
Figure 1.
Procoagulant effect of adapalene in TGA. (A) Two-dimensional structure of adapalene (6-[3-(1-adamantyl)-4-methoxyphenyl]naphthalen-2-carboxylic acid). (B) TGA performed on the ST Genesia platform with STG-Bleedscreen reagent in commercial plasma from a patient with severe HA. FVIII denotes plasma-derived FVIII (Factane; LFB). HA plasma was spiked with adapalene (50μM) in 0.4% DMSO (final concentration). The values of the TG variables (lag time, time to peak, peak height, and endogenous thrombin potential) for a representative TGA among the 3 replicates are given in supplemental Table 1. Adap, adapalene; NPP, normal pool plasma.
In the secondary screen performed on a ST Genesia platform, the addition of adapalene increased the TG similarly to 1 IU/mL of FVIII (Figure 1B).
We confirmed the procoagulant activity of adapalene by conducting a TGA with a CAT system under different conditions using platelet-poor plasma from HD plasmas preincubated with an anti-FVIII inhibitor (ESH-4, 40 μg/mL), plasma from a patient with HB, plasma from patients under VKA, or plasma spiked with asundexian, a small-molecule FXIa inhibitor.
Adapalene was able to restore coagulation in all these conditions. The obtained calibrated automated thrombograms are represented in Figure 2.
Figure 2.
Procoagulant effect of adapalene in CAT experiments. (A) Dose-dependent effect of adapalene on TG in HD plasma preincubated with an anti-FVIII inhibitor (F8i). Plasmas were spiked with adapalene in 0.4% DMSO at different ratios (1:500 to 1:20; ie, a target concentration of 2μM to 50μM). The vehicle was DMSO at 0.4% (final concentration). (B) Effect of adapalene on TG in plasma from a patient with severe HB. FIX denotes plasma-derived FIX. HB plasma was spiked with adapalene in 0.4% DMSO (final concentration). (C) Effect of adapalene on TGA in plasmas from 2 individuals under VKA (VKA1 and VKA2) (international normalized ratio ≈2.5) and 2 HD (HD1 and HD2). (D) Effect of adapalene on TG in HD plasma spiked with asundexian (10 mg/L). (E) Effect of adapalene on TG in HD plasma spiked with 200 ng/mL of either rivaroxaban, apixaban, or dabigatran. The TG variables (lag time, time to peak, peak height, and ETP) for a representative TGA among the 3 replicates are indicated in supplemental Tables 2-6 associated with panels A-E, respectively. Apix, apixaban; Dabi, dabigatran; Riva, rivaroxaban.
Furthermore, as illustrated in Figure 2A, adapalene restored TG in a dose-dependent manner, increasing the thrombin peak and endogenous potential (ETP) and shortening the lag time in an in vitro hemophilic A model (plasma from a healthy individual preincubated with an FVIII-neutralizing antibody). In HB plasma (Figure 2B), adapalene was less effective, failing to restore a normal thrombin peak. In VKA plasmas (Figure 2C), adapalene fully corrected the lag time, although the thrombin peak and ETP were only partially restored. In HD plasma spiked with asundexian at a high concentration, adapalene significantly restored TG (Figure 2D).
In addition, adapalene was able to promote TG in normal plasma and plasma spiked with DOACs, such as dabigatran, rivaroxaban, and apixaban and (Figure 2E).
As observed in Figure 2, the HD condition resulted in significant differences among different TG experiments, likely because the experiments were performed in plasma from different HD.
To confirm the procoagulant effect of adapalene, thromboelastometry experiments were performed in normal blood preincubated with an anti-FVIII inhibitor (Figure 3). These experiments showed a dramatic decrease in clotting time and a slight increase in maximum clot firmness in the presence of adapalene.
Figure 3.
Procoagulant effect of adapalene in the thromboelastometry experiment. Rotational thromboelastometry curves in native conditions; whole blood from HD was preincubated with an anti-FVIII inhibitor (F8i) and spiked with adapalene (50μM) before triggering with CaCl2 (12.5mM). The numerical values of the variables clotting time, clotting formation time, and maximal clot firmness are included in supplemental Table 7. NWB, normal whole blood.
However, the low solubility of adapalene, which is necessary for topical administration in the case of acne, is inappropriate for oral treatment of bleeding episodes that require passage through the intestinal barrier. Furthermore, the known teratogenicity of adapalene precludes its use in chronic treatment.16
Therefore, a pharmacomodulation strategy guided by TGA led to the synthesis of 30 adapalene derivatives, allowing the identification of the adamantyl moiety, the naphthyl cycle, and the negative charge present in the para position of this cycle as essential pharmacophoric features. Notably, compounds 1, 2, and 3 (schematized in Figure 4A) showed procoagulant activities similar to adapalene (Figure 4B) and slightly improved solubilities (supplemental Table 9). A complete description of the synthetic pathway and characterization of the derivatives is available in supplemental Figure 3.
Figure 4.
Identification of the mechanism of action of adapalene and its derivatives. (A) Structure of the adapalene derivatives. (B) TGA of adapalene and its derivatives in plasma from HD preincubated with an anti-FVIII inhibitor (F8i). The vehicle was DMSO (0.4% at final concentration). The values of the TG variables (lag time, time to peak, peak height, and ETP) for a representative TGA are indicated in supplemental Table 8.
The mechanism of action of adapalene and derivatives was then thoroughly investigated. Initially, we investigated whether adapalene could interfere with the physiological inhibitors TFPI and antithrombin. To that end, TGA on TFPI-depleted samples were performed, showing an increase in TG upon addition of adapalene (Figure 5A). An enzymatic inhibition assay was conducted and showed that adapalene has no effect on the FXa inhibition by antithrombin (Figure 5B). Therefore, the procoagulant mechanism of adapalene is independent of both TFPI and antithrombin.
Figure 5.
Effect of adapalene on the activities of physiological inhibitors. (A) TGA on TFPI-depleted plasma spiked with adapalene (50μM) or vehicle (0.4% DMSO final concentration). The values of the TG variables (lag time, time to peak, peak height, and ETP) for a representative TGA are indicated in supplemental Table 10. (B) Effect of adapalene on inhibition of FXa by antithrombin, followed by absorbance related to the hydrolysis of the chromogenic peptide substrate of FXa (PNAPEP 1025) at 2 antithrombin concentrations, 125nM (in green) and 250nM (in red). AT, antithrombin; Def TFPI, TFPI-depleted plasma.
To investigate whether one of the clotting factors was essential for adapalene’s procoagulant activity, TGAs with and without adapalene were performed in FII-, FV-, FVII-, FX-, and FXII-deficient plasmas and in plasmas immunodepleted in FXI or FXII.
As expected, the TGA in FII-depleted plasma and FV- and FX-deficient human plasmas showed no TG at baseline, with no effect of adapalene (data not shown).
In FII-deficient plasma, in vitro spiking with purified human FII to reach a level of 0.10 IU/mL of FII produced no TG at baseline, and adapalene did not increase TG under these conditions (data not shown). In FV-deficient plasma, in vitro spiking with normal plasma to reach an FV level of 0.01 IU/mL produced TG at baseline that increased in the presence of adapalene (supplemental Figure 1).
In FX-deficient plasma, in vitro spiking with human FX to reach a level of 0.10 IU/mL of FX produced no TG at baseline; TG increased in the presence of adapalene (supplemental Figure 1), indicating that FX is necessary for adapalene’s procoagulant action.
Adapalene strongly increased TG in FVII-depleted plasma (Figure 6A); this result supports the hypothesis that adapalene does not affect the extrinsic pathway, which is consistent with the increase in TG observed in TFPI-deficient plasma (Figure 5A). The enzymatic assays on FXa inhibition by TFPI indicate that adapalene slightly mitigates the TFPI inhibition of FXa (supplemental Figure 2).
Figure 6.
Search for the mechanism of action of adapalene and its derivatives. (A) TGA with and without adapalene in FVII-depleted plasma, FXI-depleted plasma, FXI-depleted plasma with addition of asundexian, and FXII-depleted plasma. (B) TGA with and without adapalene in plasma from a patient with FXII deficiency that was spiked with FXII or buffer. (C) Monitoring of the FXII activity for 1 hour, with or without PK (0.5 μg/mL, Cryopep) and/or adapalene. (D) Monitoring of the FXII activity with/without adapalene or the chemical compounds 1, 2, and 3 (each at 50μM). (E) TGA in PK-deficient plasma in the presence or absence of PK and/or asundexian. The TG variables associated with panels (A-B,E) are given in supplemental Tables 11-13, respectively. Def VII, FVII-depleted plasma; Def XI, FXI-deficient plasma; Def XII, FXII-depleted plasma; DXIIP, plasma with FXII deficiency (Cryopep); PKDP, PK-deficient plasma.
Experiments with FXII-immunodepleted plasma (Figure 6A-B) showed a constant baseline TG upon addition of adapalene, indicating that FXII is essential for the effectiveness of adapalene. This suggests an interaction between this clotting factor and the chemical compound.
To validate this hypothesis, adapalene was spiked into commercial plasma from a patient with severe FXII deficiency (Figure 6B). An extremely small increase in TG was observed, in line with our previous results in lyophilized FXII-immunodepleted plasma. After spiking FXII-deficient plasma with either 0.1 IU/mL or 1.0 IU/mL of purified human FXII, adapalene strongly increased TG. In a chromogenic assay using an FXII-specific substrate, we demonstrated that adapalene activates FXII, with greater efficacy in the presence of PK (Figure 6C). The optimized derivatives 1, 2, and 3 exhibited similar FXII-activation profiles (Figure 6D).
Figure 6A shows that TG in FXI-deficient plasma is increased in the presence of adapalene. This finding is surprising if adapalene acts as an activator of FXII because the latter would activate FXI. It suggests that FXI only plays a minor role in adapalene’s action. However, when asundexian was added to FXI-depleted plasma to neutralize potential residual FXI (Figure 6A), adapalene showed only a poor procoagulant effect, suggesting that FXI plays an important role. The overall result is consistent with the hypothesis that adapalene induces FXII activation upstream of the coagulation cascade.
To investigate the role of PK, we performed TGAs in PK-deficient plasmas and observed that the addition of adapalene increased TG (Figure 6E); we concluded that PK was not essential for adapalene’s procoagulant activity. However, the addition of both purified human PK at a concentration of 37.5 μg/mL (normal plasma level) to these PK-deficient plasmas and adapalene resulted in a significantly higher peak of thrombin and a shorter lag time (Figure 6E), suggesting that PK was also involved in the procoagulant effect of adapalene. This finding is consistent with the hypothesized mechanism that adapalene activates FXII.
In both the absence of PK (PK-deficient plasma) and inhibition of FXI by asundexian, adapalene lost its ability to increase TG (Figure 6E). This finding suggests that at least PK or FXI are required for adapalene’s procoagulant activity; this is consistent with the hypothesis that adapalene activates FXII to FXIIa, which subsequently activates FXI and PK.
Discussion
From a hierarchical HTS of 2300 chemical compounds, using 2 orthogonal coagulation assays, we identified adapalene as a procoagulant molecule in plasma from patients with severe HA. Surprisingly, it was the only compound to be selected from this screening procedure and the only one to possess an adamantyl moiety. It showed procoagulant activity in plasma from patients with HA and HB, HD plasma preincubated with an anti-FVIII inhibitor, plasma from patients under VKA, and HD plasma spiked with DOACs and with asundexian.
HTS for drug repurposing encompasses many techniques and reaches many therapeutic fields;19 to our knowledge, this HTS study is the first to search for chemical procoagulants in plasma from patients with severe HA and the third in human plasma.20,21 The HTS AstraZeneca study20 identified compounds that prevent the inhibition of endogenous inhibitors and bind to heparin-binding sites of different coagulation factors, whereas the chemical hemostatic compounds recently identified by Desvages et al21 restore TG in DOAC-spiked plasma and decrease bleeding in in vivo tests, although the exact mechanism was not described.
To overcome the poor solubility of adapalene, we synthesized analogs, and the 3 best showed slightly improved solubilities. These drug-like compounds22 contain the essential pharmacophoric features identified in this study for procoagulant activity, namely an adamantyl moiety, a negative charge, and an aromatic cycle. The chemical procoagulants identified in the 2 preceding HTS studies20,21 also possess a strong aromaticity, a hydrophobic moiety, a quasi-planar conformation, and a polar moiety able to be involved in hydrogen bonds.
To identify the procoagulant mechanism of action of adapalene, several experiments were conducted. The unmodified coagulation profile in FXII-deficient plasma showed that adapalene interacts with this factor. It was confirmed, with both a TGA of plasma from a patient with FXII deficiency and by monitoring FXII activity over time, with or without PK, that adapalene and its analogs activate FXII. This activation results in the subsequent activation of FXI, FIX, and consequently FX. The absence of a procoagulant effect of adapalene in PK-deficient plasma in the presence of asundexian supports this hypothesis.
The contact coagulation pathway states that FXIIa activates FXI, which in turn activates FIX that then leads to FX activation and consequently thrombin and fibrin formation. The observed TG in FIX-deficient plasma upon addition of adapalene and analogs can be attributed to the FXI activation of FV, which leads to an increase in FX activation, as reported by in vivo studies.23 Adapalene was less effective at increasing TG when FIX was deficient, in accordance with the principle that FIX can only be bypassed by FV activation by FXIa.
It is established that FXIIa activates plasma PK to kallikrein. The FXI-bypass coagulation pathway, in which plasma kallikrein activates FIX, was recently demonstrated.24,25 We hypothesize that the compounds developed in this study promote coagulation in FXI-deficient and HA plasma via this FXI-independent pathway, similarly to polyphosphates.26
Although FXII is the target of numerous studies because it drives proinflammatory pathways via generation of the peptide hormone bradykinin and procoagulant pathways,27 the detailed mechanism of its activation remains unknown.
Activation of FXII in the presence of the chemical compounds described herein might be deleterious due to activation of coagulation independent of tissue factor. Therefore, the repositioning of adapalene for systemic use in hemophilia appears to be compromised; however, to our knowledge, this is the first time adapalene was shown to restore coagulation.28 When used as a topical agent at an extremely low concentration, it has never been shown to modify the coagulation profile because no thromboembolic adverse events have been reported to date.
The reported exogenous molecules able to induce FXII activation, such as kaolin, dextran sulfate, and ellagic acid, as well as endogenous FXII activators such inorganic polymer polyphosphate (Heestermans et al29 and references therein), also exhibit polar moieties able to interact through hydrogen bonding and are negatively charged at physiological pH.
To conclude, in our search for a potential oral antihemophilic treatment, we have identified chemical procoagulants, validating our developed HTS strategy. This activity, along with the pharmaceutical availability of Differin, could lead to new therapies for skin ecchymoses for which there are currently no treatments.
The adapalene’s derivatives were serendipitously found to be FXII activators, which prevents their use for future oral antihemophilic treatment. The compounds could become leads and serve other repurposing studies, similarly to adapalene, which was recently shown to exert antiviral30 and anticancerous activities.31 Furthermore, FXII-driven coagulation was recently shown to enhance innate immunity by trapping pathogens and restricting bacterial infection in mice32; hopefully, these novel FXII activators might be relevant for diagnostic purposes or serve other therapeutic purposes requiring local administrations.
Conflict-of-interest disclosure: The authors declare no competing financial interests.
Acknowledgments
The authors are grateful to Thomas Lecompte (Vascular Medicine Department, University Hospital of Nancy, University of Lorraine, Nancy, France) for fruitful and stimulating scientific discussions.
This work was supported by LabEx Arcane CBH-EUR-GS (European Graduate School in Chemistry Biology and Health) (ANR-17-EURE-0003) and the SATT Linksium (Société d'Accélération du Transfert de Technologies). Part of this work has been performed at the CMBA platform (IRIG-DS-BGE-Gen&Chem-CMBA) (Institut de Recherche Interdisciplinaire de Grenoble-Département Santé-Laboratoire Bioscience et bioinGénierie pour la SantE-Genomique & Chemogénomique, Criblage pour des Molécules BioActives), CEA (Commissariat à l'Energie Atomique) Grenoble, F-38054 Grenoble (a member of GIS IBiSA (Groupement d'Intérêt Scientifique Infrastructures en Biologie Santé et Agronomie) and ChemBioFrance National Research Infrastructure), which is supported by LabEx GRAL (Grenoble Alliance for Integrated Structural and Cell Biology), a program of the Chemistry Biology Health Graduate School of the University Grenoble Alpes (ANR-17-EURE-0003).
Authorship
Contribution: R.Z., R.N., L.S., R.M., H.C., M.-C.D., C.B., and M.J. performed and analyzed the experiments; R.M., M.J., L.S., A.T., B.P., and M.-C.D. designed and analyzed the experiments; A.T., M.-C.D., R.M., and B.P. designed the study; and A.T., M.-C.D., R.M., M.J., L.S., and R.Z. wrote the manuscript.
Footnotes
R.Z. and L.S. contributed equally to this study.
Original data are available from corresponding author Raphaël Marlu (rmarlu@chu-grenoble.fr) on request.
The full-text version of this article contains a data supplement.
Contributor Information
Raphaël Marlu, Email: rmarlu@chu-grenoble.fr.
Aline Thomas, Email: aline.thomas@univ-grenoble-alpes.fr.
Supplementary Material
References
- 1.Pilunni D, Montrasio L, Navarra P. When innovation goes fast. The case of hemophilia. Curr Opin Pharmacol. 2019;45:95–101. doi: 10.1016/j.coph.2019.06.005. [DOI] [PubMed] [Google Scholar]
- 2.Valentino LA, Santaella ME, Carlson SA, Recht M. Contemporary approaches to treat people with hemophilia: what's new and what's not? Res Pract Thromb Haemost. 2025;9(1):102696–102713. doi: 10.1016/j.rpth.2025.102696. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Madan B, Ozelo MC, Raheja P, et al. Three-year outcomes of valoctocogene roxaparvovec gene therapy for hemophilia A. J Thromb Haemost. 2024;22(7):1880–1893. doi: 10.1016/j.jtha.2024.04.001. [DOI] [PubMed] [Google Scholar]
- 4.Samelson-Jones BJ, Small JC, George LA. Roctavian gene therapy for hemophilia A. Blood Adv. 2024;8(19):5179–5189. doi: 10.1182/bloodadvances.2023011847. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lenting PJ, Fong S. Gene therapy for hemophilia. Blood Adv. 2024;8(23):6081. doi: 10.1182/bloodadvances.2024013864. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Herzog RW, Kaczmarek R, High KA. Gene therapy for hemophilia - from basic science to first approvals of "One-and-Done" therapies. Mol Ther. 2025;33(5):2015–2034. doi: 10.1016/j.ymthe.2025.03.043. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gualtierotti R, Giachi A, Bitto N, La Mura V, Peyvandi F. Gene therapy in hemophilia: the dawn of a new era. Res Pract Thromb Haemost. 2025;9(1):102640–102655. doi: 10.1016/j.rpth.2024.102640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Franchini M, Mannucci PM. Non-factor replacement therapy for haemophilia: a current update. Blood Transfus. 2018;16(5):457–461. doi: 10.2450/2018.0272-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Oldenburg J, Mahlangu JN, Kim B, et al. Emicizumab prophylaxis in hemophilia A with inhibitors. N Engl J Med. 2017;377(9):809–818. doi: 10.1056/NEJMoa1703068. [DOI] [PubMed] [Google Scholar]
- 10.Lu D, Dou F, Gao J. Fitusiran: The first approved siRNA therapy for hemophilia via reducing plasma antithrombin levels. Drug Discov Ther. 2025;19(2):131–132. doi: 10.5582/ddt.2025.01031. [DOI] [PubMed] [Google Scholar]
- 11.Shapiro AD, Angchaisuksiri P, Astermark J, et al. Subcutaneous concizumab prophylaxis in hemophilia A and hemophilia A/B with inhibitors: phase 2 trial results. Blood. 2019;134(22):1973–1982. doi: 10.1182/blood.2019001542. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Matsushita T, Shapiro A, Abraham A, et al. Phase 3 trial of concizumab in hemophilia with inhibitors. N Engl J Med. 2023;389(9):783–794. doi: 10.1056/NEJMoa2216455. [DOI] [PubMed] [Google Scholar]
- 13.Kahr Rasmussen N, Berg B, Christiansen ASL, et al. The concizumab pen-injector is easy to use and preferred by hemophilia patients and caregivers: a usability study assessing pen-injector handling and preference. Patient Prefer Adherence. 2024;18:1713–1727. doi: 10.2147/PPA.S470091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Irby CE, Yentzer BA, Feldman SR. A review of adapalene in the treatment of acne vulgaris. J Adolesc Health. 2008;43(5):421–424. doi: 10.1016/j.jadohealth.2008.06.005. [DOI] [PubMed] [Google Scholar]
- 15.Latter G, Grice JE, Mohammed Y, Roberts MS, Benson HAE. Targeted topical delivery of retinoids in the management of acne vulgaris: current formulations and novel delivery systems. Pharmaceutics. 2019;11(10):490–515. doi: 10.3390/pharmaceutics11100490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.US Food and Drug Adminstration . Center for Drug Evaluation and Research; 2009. Pharmacology/Toxicology Review NDA Number 22-502. [Google Scholar]
- 17.Zhang JH, Chung TD, Oldenburg KR. A simple statistical parameter for use in evaluation and validation of high throughput screening assays. J Biomol Screen. 1999;4(2):67–73. doi: 10.1177/108705719900400206. [DOI] [PubMed] [Google Scholar]
- 18.Roehrig S, Ackerstaff J, Jiménez Núñez E, et al. Design and preclinical characterization program toward asundexian (BAY 2433334), an oral factor XIa inhibitor for the prevention and treatment of thromboembolic disorders. J Med Chem. 2023;66(17):12203–12224. doi: 10.1021/acs.jmedchem.3c00795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Khambhati K, Siruka D, Ramakrishna S, Singh V. Current progress in high-throughput screening for drug repurposing. Prog Mol Biol Transl Sci. 2024;205:247–257. doi: 10.1016/bs.pmbts.2024.03.013. [DOI] [PubMed] [Google Scholar]
- 20.Gustafsson D, Nielsen S, McPheat J, et al. A serendipitously identified novel small molecule procoagulant compound giving rise to a high-throughput screening assay based on human plasma. Thromb Res. 2013;132(2):248–255. doi: 10.1016/j.thromres.2013.05.023. [DOI] [PubMed] [Google Scholar]
- 21.Desvages M, Borgel D, Adam F, et al. A small-molecule hemostatic agent for the reversal of direct oral anticoagulant-induced bleeding. Res Pract Thromb Haemost. 2024;8(4):102426–102434. doi: 10.1016/j.rpth.2024.102426. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lipinski CA, Lombardo F, Dominy BW, Feeney PJ. Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev. 2001;46(1-3):3–26. doi: 10.1016/s0169-409x(00)00129-0. [DOI] [PubMed] [Google Scholar]
- 23.Matafonov A, Cheng Q, Geng Y, et al. Evidence for factor IX-independent roles for factor XIa in blood coagulation. J Thromb Haemost. 2013;11(12):2118–2127. doi: 10.1111/jth.12435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Visser M, van Oerle R, Ten Cate H, et al. Plasma kallikrein contributes to coagulation in the absence of factor XI by activating factor IX. Arterioscler Thromb Vasc Biol. 2020;40(1):103–111. doi: 10.1161/ATVBAHA.119.313503. [DOI] [PubMed] [Google Scholar]
- 25.Kearney KJ, Butler J, Posada OM, et al. Kallikrein directly interacts with and activates Factor IX, resulting in thrombin generation and fibrin formation independent of Factor XI. Proc Natl Acad Sci U S A. 2021;118(3) doi: 10.1073/pnas.2014810118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Puy C, Tucker EI, Wong ZC, et al. Factor XII promotes blood coagulation independent of factor XI in the presence of long-chain polyphosphates. J Thromb Haemost. 2013;11(7):1341–1352. doi: 10.1111/jth.12295. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Maas C, Renné T. Coagulation factor XII in thrombosis and inflammation. Blood. 2018;131(17):1903–1909. doi: 10.1182/blood-2017-04-569111. [DOI] [PubMed] [Google Scholar]
- 28.Thomas A, Jourdan M, Navarro R, et al. Patent PCT/EP2021/070656 Deposited 2021-24-07; 2021. Composés chimiques pour le traitement de l'hémophilie. [Google Scholar]
- 29.Heestermans M, Naudin C, Mailer RK, et al. Identification of the factor XII contact activation site enables sensitive coagulation diagnostics. Nat Commun. 2021;12(1):5596–5612. doi: 10.1038/s41467-021-25888-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Fidan O, Mujwar S, Kciuk M. Discovery of adapalene and dihydrotachysterol as antiviral agents for the Omicron variant of SARS-CoV-2 through computational drug repurposing. Mol Divers. 2023;27(1):463–475. doi: 10.1007/s11030-022-10440-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Boulos JC, Chatterjee M, Shan L, Efferth T. In silico, in vitro, and in vivo investigations on adapalene as repurposed third generation retinoid against multiple myeloma and leukemia. Cancers (Basel) 2023;15(16):4136–4162. doi: 10.3390/cancers15164136. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Nickel KF, Jamsa A, Konrath S, et al. Factor XII-driven coagulation traps bacterial infections. J Exp Med. 2025;222(7):e20250049. doi: 10.1084/jem.20250049. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.