Selective inhibition of ERO1α with M6766, a novel small-molecule inhibitor, prevents arterial thrombosis and ischemic stroke in mice

Abstract

Using endoplasmic reticulum oxidoreductase 1α (ERO1α) conditional knockout (CKO) mice, a recent study underscores the crucial role of ERO1α in platelet activation under thrombotic conditions. Through a high-throughput screen of 39,901 compounds, we identify M6766 as a selective inhibitor of ERO1α with an IC₅₀ of 1.4 μM and a K_D of 1.1 μM. A docking model and biochemical studies reveal that M6766 binds to the flavin adenine dinucleotide-binding pocket in ERO1α and exhibits >70-fold selectivity over other tested enzymes, except ERO1β, which it inhibits with an IC₅₀ of 7.2 μM. M6766 concentration-dependently inhibits granule secretion, αIIbβ3 integrin activation, Ca²⁺ mobilization, and platelet aggregation induced by various agonists, but it does not affect agonist-induced production of reactive oxygen species. Pretreatment of ERO1α with M6766 reduces its binding to the Ca²⁺ sensor stromal interaction molecule 1. To validate whether these inhibitory effects result from the inhibition of ERO1α and ERO1β, we generate megakaryocyte-specific Ero1β or Ero1α/β CKO mice. Deletion of platelet Ero1α/β impairs platelet activation and aggregation, whereas deletion of Ero1β has no effect. While EN460 markedly inhibits the function of Ero1α/β-null platelets, M6766 does not, highlighting its specificity. M6766 treatment diminishes platelet accumulation on collagen-coated surfaces under arterial shear conditions. Moreover, intravenous injection of M6766 into mice decreases arterial thrombosis and infarct volume during ischemic stroke without prolonging tail bleeding times. Although eptifibatide, an αIIbβ3 antagonist, effectively blocks arterial thrombosis, it prolongs bleeding times at therapeutic doses. Our findings suggest that ERO1α inhibition is a promising anti-thrombotic strategy with potential advantages over current therapies.

INTRODUCTION

Endoplasmic reticulum oxidoreductase 1 (ERO1), a flavin adenine dinucleotide (FAD)-dependent enzyme, catalyzes the oxidation of protein disulfide isomerase (PDI) to facilitate protein folding in the ER.¹ While yeast has a single isoform of ERO1 required for disulfide bond formation, mammals possess two isoforms, ERO1α and ERO1β, whose function is compensated by alternative pathways that can oxidize PDI.^2–5 ERO1α is ubiquitously expressed, while ERO1β is predominantly found in pancreatic β cells and gastric chief cells.^4,6 A previous study has shown that mice deficient in both Ero1a and Ero1b are viable and exhibit no discernible phenotypes.⁷ ERO1α and ERO1β are regulated by the unfolded protein response during ER stress, but only ERO1α is upregulated by hypoxia.^4,8 Dysregulation of ERO1 expression or activity has been implicated in various diseases, including cancer and metabolic disorders.^9–11 In addition to the well-known role in oxidizing PDI, our recent study using megakaryocyte-specific Ero1α conditional knockout (CKO) mice and non-specific inhibitors identified platelet ERO1α as a novel Ca²⁺ regulator that enhances platelet activation and aggregation in arterial thrombosis and ischemic stroke.¹² These findings suggest that targeting ERO1α may be a novel therapeutic strategy for preventing or treating thrombotic conditions.

Arterial thrombotic diseases, such as myocardial infarction and ischemic stroke, are driven by excessive platelet activation and adhesion and remain the leading causes of death in the United States.^13,14

Following arterial injury, platelets adhere to von Willebrand factor and collagen through the glycoprotein (GP)Ib-IX-V complex and GPVI, respectively, inducing platelet aggregation through the interaction of fibrinogen with activated αIIbβ3 integrin.¹⁵ Ischemic stroke, resulting from the occlusion of cerebral blood vessels, accounts for approximately 85% of all strokes.^16,17 Currently, treatment options are restricted to thrombolysis or thrombectomy, depending on the time window (the first 4.5 h) of the onset of stroke symptoms.^18,19 Existing antiplatelet agents, which target platelet surface receptors or intracellular molecules, have reduced morbidity and mortality in patients with thrombotic diseases, albeit with an increased risk of major bleeding at therapeutically effective doses.^20,21 Thus, there remains an unmet need for safer and more effective anti-thrombotic therapies.

In the present study, we conducted high-throughput screening (HTS) of 39,901 drug-like compounds and their derivatives, identifying M6766 as a potent and selective inhibitor of ERO1α. The compound inhibits the activities of ERO1α and ERO1β with half-maximal inhibitory concentration (IC₅₀) values of 1.4 and 7.2 μM, respectively, and directly binds to ERO1α with a K_D of 1.1 μM. Treatment of platelets with 1–5 μM M6766 reduces platelet activation and aggregation induced by various agonists, including thrombin or collagen-related peptide (CRP). Importantly, unlike EN460, a non-specific ERO1α inhibitor, M6766 exhibits selective inhibitory activity at effective concentrations without affecting the function of ERO1α/β-null platelets. Intravital microscopy (IVM) and murine disease model studies demonstrate that pretreatment with M6766 diminishes platelet thrombus formation in arterial thrombosis and reduces brain infarct in ischemic stroke without prolonging tail bleeding times. These results provide strong evidence that pharmacological inhibition of ERO1α is a promising anti-thrombotic strategy.

RESULTS

HTS identifies M6766 as a selective ERO1α inhibitor

Although previous studies have identified ERO1α inhibitors, most are non-specific.^12,22–24 To identify ERO1α-specific inhibitors, we screened 39,901 drug-like compounds using a fluorescence-based ERO1α activity assay, along with various counter-screening assays (Figures 1A and 1B). The ERO1α activity assay demonstrated a Z-factor of 0.79, a signal-to-noise ratio of 15:1, and a coefficient of variation of 4.0%. We selected 365 compounds that inhibit ERO1α activity by greater than 80% at a concentration of 25 μM (Figure 1C). After evaluating the compound structures, we excluded those with identical scaffolds and tested 108 compounds using an Amplex Red assay to identify and eliminate those exhibiting antioxidant activity greater than 10% at the same concentration (Figure 1D). Then, the selected 72 compounds were retested in the ERO1α activity assay at a concentration of 10 μM (Figure S1A).

Figure 1. Identification of M6766 as a selective ERO1α inhibitor.

(A) A fluorescence-based ERO1α activity assay. (B) The screening funnel from primary HTS of 39,901 compounds. (C) Selection of 365 compounds in the ERO1α activity assay. (D) The effect of 108 compounds on the Amplex Red assay. (E) The 19 derivatives of W9418 were tested at 5 μM in the ERO1α activity assay. (F) The structures of M6766, M2873, M7095, M3008, and M4634. (G) The concentration-dependent inhibition of M6766. (H) Biolayer interferometry was performed using a biotinylated ERO1α biosensor. After incubation with various concentrations of M6766, the specific interaction between ERO1α and the compound was measured by subtracting the nonspecific binding. The dissociation constant, K_D, was calculated from the K_on and K_off. (I) A docking model of ERO1α-M6766 (purple) and ERO1α-FAD (orange) complexes following LigPrep and Jaguar electrostatic potential optimization. (J) The effect of FAD on ERO1α activity. (K) The effect of M6766 on FAD-enhanced ERO1α activity. (L–P) Various concentrations of M6766 were used in the activity assays of ERO1β, H₂O₂, Mao-A, and PDI and the assay measuring thiol reactivity. The data represent the mean ± SD (n = 4–6 for E and n = 3 for G–P). *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 vs. vehicle control after Student’s t test (E) or ANOVA and Dunnett’s test (J–P).

We identified 14 compounds representing 6 distinct scaffolds, most of which demonstrated equal or greater inhibition of ERO1α activity compared to 5 μM EN460 (Figures S1B and S1C).

While the other 5 scaffolds had a limited number of available derivatives, W9418 featured 19 compounds based on the tetrahydrobenzimidazoloisoquinoline nitrile scaffold or similar structures. Among them, M6766, M2873, M7095, M3008, and M4634 at a concentration of 5 μM significantly inhibited ERO1α activity, which was comparable to or greater than EN460 (Figures 1E and 1F; Table S1). The IC₅₀ values of M6766, M2873, M7095, M3008, M4634, and EN460 were 1.4 ± 0.2, 5.2 ± 1.2, 9.6 ± 1.9, 16.8 ± 3.3, 23.7 ± 4.5, and 5.5 ± 1.6 μM, respectively (Figures 1G and S2). Since M6766 was the most potent ERO1α inhibitor, we further performed biolayer interferometry using a biosensor conjugated with biotinylated ERO1α. M6766 directly bound to ERO1α with a K_D of 1.1 ± 0.4 μM (Figure 1H).

To gain additional insights into the binding property of M6766 to ERO1α, compound-protein docking of the X-ray model (PDB: 3AHQ) was conducted.²⁵ The tetrahydrobenzimidazoloisoquinoline nitrile region of M6766 occupied the same space as the isoalloxazine ring of FAD based on molecular docking results based on LigPrep and Jaguar electrostatic potential optimization (docking score −6.8 kcal/mol) (Figure 1I). The phenyl region of the four-ring core of M6766 formed a π-π stacking interaction with Trp200, a feature not observed with FAD binding. In addition, a 2.7-Å hydrogen bond was observed between His255 and a triazole nitrogen of M6766. We found that exogenous addition of 0.01–1 μM FAD significantly increased ERO1α activity (Figure 1J) and that M6766 inhibited this FAD-enhanced activity in a concentration-dependent manner (Figure 1K). These results suggest that M6766 competes with FAD for binding to ERO1α, thereby inhibiting its activity.

In various counter-screening assays, we found that M6766 inhibited ERO1β activity with an IC₅₀ of 7.2 ± 1.3 μM (Figure 1L), likely due to the 65% sequence homology with ERO1α. In contrast, 100 μM M6766 did not inhibit the activities of H₂O₂, monoamine oxidase A (Mao-A, an FAD-binding enzyme), or PDI, nor did it react with free thiol groups (Figures 1M–1P). Our data suggest that M6766 is a selective ERO1 inhibitor, with 5-fold greater potency for ERO1α than ERO1β.

M6766 inhibits platelet activation and aggregation following agonist stimulation

Our recent study demonstrates that platelet ERO1α promotes platelet activation and aggregation by enhancing Ca²⁺ mobilization.¹² Thus, we examined the effect of M6766 on platelet function. Compared to the vehicle control, M6766 and EN460 inhibited P-selectin exposure (α-granule secretion) and αIIbβ3 integrin activation in mouse platelets in a concentration-dependent manner following stimulation with low concentrations of various agonists, such as 0.01 U/mL thrombin, 0.025 μg/mL CRP, 0.5 μM U46619, and 0.5 μM A23187 (Figures 2A, 2B, and S3). Similar results were obtained with human platelets (Figure S4). The high concentrations of M6766 (5 μM) and EN460 (2.5 or 5 μM) appeared to enhance both events in resting platelets. Although weaker, the inhibitory effects of 2.5 or 5 μM M6766 and EN460 were still observed in P-selectin exposure and αIIbβ3 integrin activation in response to high concentrations of thrombin (0.1 U/mL) or CRP (0.2 μg/mL) (Figure S5). Minimal inhibition was observed at 0.5 μM M6766 (Figure S6).

Figure 2. M6766 inhibits granule secretion, αIIbβ3 integrin activation, platelet aggregation, and Ca²⁺ mobilization.

(A–F) C57BL/6 mouse platelets were pretreated with vehicle (0.1% DMSO) or various concentrations of M6766 or EN460. (A and B) After incubation with 0.01 U/mL thrombin (Thr) or 0.025 μg/mL collagen-related peptide (CRP), P-selectin exposure and αIIbβ3 integrin activation were measured by flow cytometry. (C–F) After incubation with 0.015 U/mL thrombin or 0.05 μg/mL CRP, platelet aggregation and ATP secretion were measured using an aggregometer. (C and E) Representative traces of platelet aggregation. (G–I) Platelets in C57BL/6 mouse blood were labeled with DiOC6. After pretreatment with vehicle (0.1% DMSO) or 5 μM M6766 or EN460, blood was perfused under 50 dyne/cm² into a microfluidic chamber coated with type 1 collagen. Adherent and aggregated platelets were captured under an epifluorescence microscope. (J–N) Mouse platelets were pretreated with vehicle (0.1% DMSO) or various concentrations of M6766 or EN460. Ca²⁺ release and influx were assessed in response to 0.02 U/mL thrombin, 0.5 μM A23187, or 5 μM thapsigargin, followed by the addition of 2 mM CaCl₂. (M and N) The Ca²⁺ signal was quantified by the area under the curve (AUC). (O) Biolayer interferometry was performed using a biotinylated STIM1 biosensor. After incubation of 2.5 μM ERO1α with 5 μM M6766, the specific interaction between STIM1 and ERO1α was measured by subtracting the nonspecific binding. The flow cytometric data are presented as the geometric mean fluorescence intensity (MFI). The data represent the mean ± SD (n = 4–6 for A and B, n = 3 for C–F and O, and n = 4 for G–N). *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 vs. vehicle control after ANOVA and Dunnett’s test (A, B, and H–N) or Student’s t test (C–F and O).

We next assessed the effect of M6766 on platelet aggregation and ATP secretion. Compared to the vehicle control, 1 or 2.5 μM M6766 or EN460 significantly inhibited aggregation but had no effect on ATP secretion in mouse platelets in response to low concentrations of thrombin or CRP (Figures 2C–2F and S7A–S7D). Both compounds at 5 μM markedly inhibited platelet aggregation and ATP secretion (Figures S7E–S7H). M6766 at 5 μM but not at 1 or 2.5 μM inhibited serotonin secretion, a marker of dense granule release, after thrombin or CRP stimulation (Figure S8). EN460 at 2.5 or 5 μM enhanced serotonin secretion from resting platelets but showed a distinct effect on activated platelets, either enhancing or inhibiting secretion. These results suggest that M6766 exhibits a weaker inhibitory effect on dense granule secretion compared to its effect on α-granule secretion.

The inhibitory effects of M6766 on platelet aggregation and ATP secretion were minimal, even at 5 μM, in response to a high concentration of thrombin but remained potent at 2.5 or 5 μM in response to a high concentration of CRP (Figure S9). In contrast, EN460 significantly inhibited both responses. M6766 inhibited platelet aggregation and ATP secretion in a concentration-dependent manner following U46619 stimulation (Figures S10A and S10B). Notably, EN460, even at 1 μM, completely inhibited U46619-induced platelet aggregation without affecting ATP secretion, whereas 5 μM EN460 abolished both responses (Figures S10C and S10D). M6766 and EN460 also inhibited platelet aggregation and ATP secretion induced by A23187 (Figures S10E–S10H). Similar inhibitory effects were observed with both compounds in human platelets, with 1 μM M6766 significantly inhibiting platelet aggregation and ATP secretion, whereas the same concentration of EN460 had no effect (Figure S11). These results indicate that M6766 effectively inhibits granule secretion, αIIbβ3 integrin activation, and platelet aggregation.

We and others have reported that reactive oxygen species (ROS) produced by NADPH oxidases 1 and 2 enhance platelet activation during arterial thrombosis.^26–28 Since ERO1α oxidizes reduced PDI, generating H₂O₂ as a by-product during protein folding in the ER,²⁴ we examined the effect of ERO1α inhibitors on ROS production in platelets. As assessed by 2′,7′-dichlorodihydrofluorescein diacetate (DCF) fluorescence, treatment with 5 μM M6766 or EN460 moderately increased ROS production in resting platelets (Figure S12A). While 1–5 μM M6766 did not affect the agonist-induced DCF signal, 5 μM EN460 reduced the signal after stimulation with thrombin but not CRP. These results suggest that platelet ERO1α is unlikely to contribute to H₂O₂ production.

Recent studies have shown that ERO1α regulates cancer and neuronal cell death.^29,30 To test whether ERO1α inhibition affects platelet viability, we examined annexin V binding. Compared to the vehicle control, incubation with 5 μM EN460 markedly increased annexin V binding within 20 min, with a progressive increase over time (Figure S12B). The same concentration of M6766 also enhanced annexin V binding, but after a 4-h incubation. Prolonged incubation (24 h) led to increased annexin V binding even in vehicle-treated platelets. These results imply that M6766 may influence the viability of platelets and other cell types following prolonged treatment.

Previous studies have reported that CCL5 (RANTES) is released from α-granules of activated platelets, contributing to inflammatory conditions.³¹ We found that treatment of M6766 or EN460 inhibits CCL5 secretion from activated platelets compared to the vehicle control (Figure S13). However, 5 μM EN460 appeared to increase CCL5 secretion in resting and activated platelets. Consistent with reduced P-selectin exposure, this result suggests that M6766 may attenuate inflammatory conditions by reducing the release of proinflammatory cytokines and chemokines from activated platelets.

To examine the effect of M6766 on platelet thrombus formation, we performed a flow chamber assay. When mouse blood was pretreated with 5 μM M6766 or EN460 and perfused over collagen-coated surfaces under arterial shear stress (50 dyne/cm²), surface coverage and volume of platelet thrombi were significantly reduced (Figures 2G–2I; Videos S1, S2, and S3). Similar results were observed with human blood (Figure S14; Videos S4, S5, and S6). Since platelet ERO1α promotes agonist-induced Ca²⁺ release and influx by regulating the function of stromal interaction molecule 1 (STIM1) or sarcoplasmic/ER Ca²⁺ ATPase 2 (SERCA2),¹² we investigated whether M6766 affects Ca²⁺ mobilization during platelet activation. Pretreatment with 1 or 2.5 μM M6766 or EN460 significantly reduced both Ca²⁺ release and influx following stimulation of mouse platelets with thrombin, A23187, or thapsigargin (TG) (Figures 2J–2N). Lower concentrations of M6766 and EN460 (0.25 or 0.5 μM) exhibited no or minimal inhibition of Ca²⁺ release and influx (Figure S15). We found that 2 or 5 μM M6766 and EN460 significantly inhibited both Ca²⁺ release and influx in activated human platelets, and both inhibitors even at 0.5 μM reduced thrombin-induced Ca²⁺ influx (Figure S16). Since we reported that ERO1α directly binds to STIM1, promoting Ca²⁺ mobilization during platelet activation,¹² we investigated whether M6766 affects this interaction. Using biolayer interferometry (BLI), we found that M6766 significantly reduced the interaction between ERO1α and STIM1 (Figure 2O). These results suggest that M6766 impairs Ca²⁺ mobilization during platelet activation by inhibiting ERO1α binding to STIM1.

The effects of M6766 result from the selective inhibition of ERO1α

Although ERO1β is predominantly detected in pancreatic β cells,^4,6 our recent mass spectrometric analysis revealed that human platelets express ERO1β.¹² To investigate the role of ERO1β in platelet function, Ero1b^flox/flox mice were generated and bred with Pf4-cre mice to delete Ero1β in megakaryocytes and platelets (Ero1b^{flox/flox;Pf4−cre+/−}) (Figures S17A and S17B). Although Ero1β CKO mice were confirmed by PCR (Figure S17B), we could not detect the protein (around 50–55 kDa) in human and mouse platelets by an anti-ERO1β antibody (Figure S17C). There were no differences in complete blood counts (CBCs) between wild-type (WT) control and Ero1β CKO mice (Table S2). We found that deletion of platelet Ero1β did not impair P-selectin exposure, αIIbβ3 integrin activation, platelet aggregation, and ATP secretion after stimulation with thrombin or CRP (Figures 3A–3D). These results suggest that even if expressed, platelet Ero1β is unlikely to regulate platelet activation and aggregation.

Figure 3. Treatment with M6766 recapitulates the defect in Ero1α/β-null platelets.

(A–J) P-selectin exposure, integrin activation, aggregation, and ATP secretion of WT control. (A–D) Ero1β-null or (E–J) Ero1α/β-null platelets were induced by various concentrations of thrombin (Thr) or CRP. (K–N) WT control or Ero1α/β-null platelets were pretreated with vehicle (0.1% DMSO) or various concentrations of M6766 or EN460. (K and L) After incubation of Ero1α/β-null platelets with 0.01 U/mL thrombin and 0.025 μ/mL CRP, P-selectin exposure and αIIbβ3 integrin activation were measured by flow cytometry. (M and N) After incubation of WT control or Ero1α/β-null platelets with 0.015 U/mL thrombin, platelet aggregation and ATP secretion were measured using an aggregometer. The data represent the mean ± SD (n = 3 for A–D, G–J, M, and N and n = 5 for E, F, K, and L). *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001 vs. vehicle control after Student’s t test (A–J) or ANOVA and either Dunnett’s test (K and L) or Tukey’s test (M and N).

Since M6766 is likely to inhibit the activity of both ERO1 isoforms due to its low selectivity, we used Ero1α/β-null platelets to rule out any unknown role of Ero1β in platelet function and to specifically assess whether the effects of M6766 are Ero1 dependent. Thus, we generated Ero1a/b^flox/flox CKO mice by crossing Ero1a/b^flox/flox mice with Pf4-cre mice. The double CKO mice were confirmed by PCR and showed platelet-specific deletion of Ero1α (Figures S18A and S18B). Loss of platelet Ero1α/β did not alter the expression of P-selectin or β3 integrin, nor did it affect the expression of αIIbβ3 integrin, GPIbα, or GPVI (Figures S18C–S18E). CBCs were not different between WT control and Ero1α/β CKO mice (Table S3). Compared to WT control platelets, Ero1α/β-null platelets exhibited a significant reduction in P-selectin exposure and αIIbβ3 integrin activation in response to thrombin or CRP (Figures 3E and 3F). In addition, deletion of Ero1α/β significantly impaired thrombin- or CRP-induced platelet aggregation and ATP secretion, with a more pronounced inhibition following CRP stimulation (Figures 3G–3J). However, platelets from Ero1α/β CKO mice exhibited normal ROS production and did not show increased annexin V binding until 8 h (Figure S19). This result suggests that the enhanced annexin V binding after treatment with 5 μM M6766 (Figure S12B) is likely due to an off-target effect.

To investigate the specificity of M6766 for Ero1, we examined its effects on the activation and aggregation of Ero1α/β-null platelets. EN460 significantly reduced agonist-induced P-selectin exposure and αIIbβ3 integrin activation, whereas M6766 had no additional effect (Figures 3K and 3L). M6766 at 5 μM and EN460 at 2.5 μM appeared to increase the levels of these surface markers in resting Ero1α/β-null platelets. Treatment of Ero1α/β-null platelets with 2.5 μM EN460 markedly inhibited thrombin-induced aggregation, but not ATP secretion (Figures 3M and 3N). In contrast, 2.5 μM M6766 had no effect on Ero1α/β-null platelet aggregation. A higher concentration of M6766 (5 μM) also did not significantly affect thrombin-stimulated aggregation and ATP secretion in Ero1α/β-null platelets (Figures S20A and S20B). Conversely, 5 μM EN460 significantly inhibited both responses, indicating its off-target effects. Similar results were observed with 5 μM M6766 and EN460 for P-selectin exposure, αIIbβ3 integrin activation, aggregation, and ATP secretion in Ero1α-null platelets (Figures S20C–S20F). Overall, these results suggest that the inhibitory effects of M6766 on platelet function, at concentrations up to 5 μM, are mediated by Ero1α inhibition.

Since ERO1β contributes to insulin biogenesis in the pancreas,³² we further examined whether M6766 inhibits Ero1β-mediated cellular function. Treatment with 1 or 5 μM M6766 or EN460 did not reduce insulin secretion or content in pancreatic islet cells under high-glucose conditions (Figure S21), suggesting that at concentrations that inhibit platelet function, M6766 and EN460 do not impair Ero1β function.

Treatment of M6766 attenuates platelet thrombus formation in arterial thrombosis and reduces brain damage in ischemic stroke without prolonging tail bleeding times in mice

We reported that intravenous injection of B12-5, a non-specific ERO1α inhibitor, reduces arterial thrombogenesis and infarct volume in ischemic stroke.¹² To evaluate the in vivo effects of M6766 in thrombotic conditions, we first examined its ex vivo effects. When we tested its solubility, M6766 was dissolved in DMSO up to 12.5 mM (data not shown), allowing for injection into mice at a maximum dose of 0.3 μg/g body weight (BW). Mouse blood was collected 10 min after injecting the dose of M6766 or EN460, followed by flow cytometry. We found that injection of M6766 or EN460 exhibited a moderate but significant reduction in P-selectin exposure and αIIbβ3 integrin activation following thrombin stimulation (Figures 4A and 4B). As measured by liquid chromatography-tandem mass spectrometry (LC-MS/MS), plasma concentrations of M6766 were 55.5, 6.3, 1.0, and 0.1 ng/mL at 5, 15, 30, and 60 min, respectively, after intravenous injection of 0.3 μg/g BW of M6766 (Figure 4C). Treatment with M6766 or EN460 did not alter CBCs in mice compared to the vehicle control, based on blood samples collected at 10 min and 8 h after intravenous injection (Tables S4 and S5). When 0.3 μg/g BW of M6766 or EN460 was intravenously injected into C57BL/6 mice 10 min before IVM, both compounds significantly inhibited platelet accumulation without affecting initial platelet adhesion at sites of laser injury compared to the vehicle control (Figures 4D and 4E; Videos S7, S8, and S9). The inhibitory effect was not significantly different between M6766 and EN460. We further assessed the inhibitory effects of M6766 and EN460 in a thrombosis model induced by FeCl₃-mediated injury to the large carotid artery. Intravenous injection of 0.3 μg/g BW of M6766 or EN460 significantly prolonged the time to occlusion (TTO) following 7% FeCl₃ but did not affect TTO after 10% FeCl₃ application (Figures 4F and 4G). However, treatment with eptifibatide (5 μg/g BW, an antagonist of αIIbβ3 integrin) significantly prolonged TTO irrespective of the FeCl₃ concentration. Administration of a lower dose of M6766 or EN460 (0.1 μg/g BW) did not inhibit carotid arterial thrombosis (Figure S22).

Figure 4. M6766 exhibits anti-thrombotic effects without prolonging tail bleeding times in mice.

(A and B) C57BL/6 mice were treated with intravenous injection of vehicle (1% DMSO in saline), M6766, or EN460 (0.3 μg/g body weight [BW]). Ten minutes later, blood was collected, and platelets were isolated. P-selectin exposure and αIIbβ3 integrin activation were assessed in flow cytometry. (C) Plasma concentrations of M6766 were analyzed by LC-MS/MS after intravenous injection of the compound (0.3 μg/g BW) into C57BL/6 mice and quantified by comparison with a standard curve of M6766 (mean ± SD, n = 3). (D and E) C57BL/6 mice were pretreated with intravenous injection of vehicle, M6766, or EN460 (0.3 μg/g BW), followed by injection of a DyLight 649-conjugated anti-CD42c antibody. Ten minutes later, intravital microscopy was performed to quantify the median integrated fluorescence intensities of an anti-CD42c antibody (n = 30–32 arterioles in 5 mice per group). (D) Representative images. (E) Quantification of the antibody signal at various time points after laser injury. (F and G) C57BL/6 mice were pretreated with intravenous injection of vehicle, M6766, EN460 (0.3 μg/g BW), or eptifibatide (5 μg/g BW) 10 min before applying a (F) 7% or (G) 10% FeCl₃-soaked filter paper to a carotid artery. The TTO was measured using a Doppler flow meter. (H and I) C57BL/6 mice were treated with intravenous injection of vehicle, M6766, EN460 (0.3 μg/g BW), or eptifibatide (5 μg/g BW). Ten minutes later, tail bleeding times and hemoglobin contents were measured after amputation of the tail tip. (J–M) C57BL/6 mice were subjected to transient middle cerebral artery occlusion for 1 h, followed by intravenous injection of vehicle, M6766, or EN460 (0.3 μg/g BW). Twenty-three hours later, neurological deficits were assessed by the Bederson score and grip strength test. Infarct volume was measured as described in materials and methods. The data represent the mean ± SD (n = 3 for A–C and n = 7 for M). The bar indicates the median (n = 5 for E and n = 7–8 for F–I). *p < 0.05, **p < 0.01, and ***p < 0.001 vs. vehicle control after Student’s t test (A, B, and M) or Mann-Whitney U test (E–K).

We then investigated whether treatment with M6766 impairs hemostasis following tail tip amputation. Compared to the vehicle control, intravenous injection of either M6766 or EN460 (0.3 μg/g BW) did not prolong tail bleeding times or increase blood loss at the site, whereas eptifibatide markedly increased both (Figures 4H and 4I). Although higher doses of M6766 could not be tested in bleeding time assays due to limited solubility, these results suggest that targeting ERO1α may be a promising therapeutic strategy for preventing arterial thrombosis.

Platelet thrombus-mediated occlusion of a cerebral artery leads to ischemic stroke, which remains a major cause of morbidity and mortality.¹⁷ Our recent study demonstrated that compared to WT control mice, Ero1α global KO but not megakaryocyte-specific CKO mice exhibited a significant reduction in infarct volume following transient middle cerebral artery occlusion (tMCAO) and subsequent reperfusion.¹² Thus, we investigated whether M6766 exerts a protective effect on brain damage in stroke. C57BL/6 mice were intravenously injected with either M6766 or EN460 (0.3 μg/g BW) immediately after 1 h of tMCAO. Following 23 h of reperfusion, both compounds decreased the Bederson score and increased grip strength, suggesting improved neurological function (Figures 4J and 4K). Consistently, treatment with M6766 or EN460 significantly reduced infarct volume following focal brain ischemia (Figures 4L and 4M). These results provide evidence that selective ERO1α inhibitors may be a promising therapeutic approach for preventing or treating ischemic stroke.

DISCUSSION

Our recent study using Ero1α CKO and KO mice and non-specific ERO1 inhibitors demonstrated that ERO1α promotes Ca²⁺ mobilization and enhances platelet activation and aggregation, contributing to arterial thrombosis. In the present study, we conducted HTS of 39,901 drug-like compounds and their derivatives and identified M6766 as a selective ERO1α inhibitor. In vitro studies show that M6766 inhibits ERO1α activity with an IC₅₀ of 1.4 μM, exhibiting 5-fold selectivity over ERO1β and at least 70-fold selectivity over PDI and Mao-A. Pretreatment with M6766 at concentrations of 1–2.5 μM inhibits Ca²⁺ mobilization, granule secretion, αIIbβ3 integrin activation, aggregation, and ATP secretion in activated platelets. Using platelets from Ero1α/β or Ero1α CKO mice, we observe that unlike EN460, the inhibitory effects of M6766 are specifically attributed to Ero1α inhibition. IVM and murine disease model studies demonstrate that intravenous injection of M6766 reduces platelet thrombus formation in arterial thrombosis and mitigates brain damage in ischemic stroke, while preserving normal tail bleeding times at the therapeutically effective dose. These findings provide strong evidence that targeting ERO1α may serve as a novel therapeutic strategy for preventing or treating thrombotic conditions.

Using megakaryocyte-specific PDI or Ero1α CKO mice, we reported that deletion of platelet PDI or Ero1α results in a moderate but significant reduction in platelet aggregation upon stimulation with various agonists.^12,33–35 Platelet-released PDI promotes the ligand-binding function of platelet surface receptors, such as αIIbβ3 integrin and GPIbα, by modifying allosteric disulfide bonds.^33–35 In contrast, platelet ERO1α enhances Ca²⁺ mobilization by regulating STIM1 or SERCA2 function, facilitating platelet activation.¹² Our BLI data revealed that M6766 inhibits ERO1α-STIM1 interaction and Ca²⁺ mobilization during platelet activation (Figures 2J–2O). While various small-molecule PDI inhibitors have been reported,²⁴ most lack specificity for PDI, exhibit off-target effects, such as antioxidant activity, or inhibit the essential function of intracellular PDI during protein folding after penetrating cells. Indeed, mice deficient in P4hb (Pdi) exhibit embryonic lethality.²⁴ This result suggests that, despite the presence of 21 PDI family member oxidoreductases,³⁶ PDI function in the ER cannot be replaced by other enzymes. In contrast, mice with loss-of-function mutations in both Ero1α and Ero1β are viable and fertile, exhibiting only a modest delay in disulfide bond formation.³² Furthermore, in the absence of ERO1, peroxiredoxin 4 can oxidize PDI.^5,37 Given the non-essential role of ERO1 and the presence of only two isoforms, ERO1 may represent a more favorable therapeutic target than PDI.

Johnson and colleagues reported that T151742 is a selective ERO1α inhibitor, with IC₅₀ and K_D values of 8.7 and 31.4 μM, respectively, and does not bind to the FAD-containing enzyme lysine-specific demethylase 1.²³ However, this compound is known to inhibit the activity of two other FAD-binding enzymes, Mao-A and Mao-B, with IC₅₀ values of 1.0 and 0.2 μM, respectively.³⁸ We reported that B12-5 (IC₅₀ = 8 μM) inhibits ERO1α activity, reducing platelet activation and aggregation in vitro and platelet thrombus formation in vivo.¹² Although the phenothiazine ring of B12-5 aligns with the isoalloxazine ring of FAD in ERO1α, the compound also diminishes Mao-A activity,¹² suggesting that the ring moiety likely binds to the FAD-binding pocket in other FAD-binding enzymes. In contrast, the tetrahydro-benzimidazoloisoquinoline nitrile region of M6766 binds to the FAD-binding pocket in ERO1α (Figure 1I) but is unlikely to bind to the same pocket in Mao-A as the compound does not impair its activity (Figure 1N). Nevertheless, M6766 has limited solubility. To address this, we are designing and synthesizing derivatives, such as those bearing an N,N-dimethylaminoethyl group to replace the N-methyl substituent, to enable formation of a water-soluble hydrochloride salt, a well-established strategy in drug development.³⁹ Our ongoing studies will optimize these derivatives for improved selectivity, solubility, and binding affinity to ERO1α.

The transcripts of ERO1a and ERO1b display both distinct and overlapping tissue distributions. ERO1β is predominantly expressed in pancreatic β cells, but is also found in the stomach, duodenum, testis, liver, appendix, thyroid, and pituitary gland.⁴ Our recent MS analysis revealed that ERO1β is present in human platelets.¹² However, we were unable to detect ERO1β in human and mouse platelets by immunoblotting, likely due to its low abundance or the limited sensitivity of the antibodies used. Our study using megakaryocyte-specific Ero1β CKO mice suggests that regardless of its expression, Ero1β is unlikely to play a role in platelet activation and aggregation (Figures 3A–3D). Given that the expression or activity of ERO1α and ERO1β is regulated during ER stress, with ERO1α being upregulated under hypoxic conditions,^4,8 their roles in modulating platelet function and thrombosis may hold greater significance in metabolic diseases or cancer.^40–42 However, targeting ERO1 in non-platelet cells may lead to unexpected side effects, warranting careful evaluation.

EN460 was identified as an irreversible ERO1α inhibitor through HTS and has been used in cell-based assays and animal model studies.^{12,22,43–45} Nevertheless, EN460 exhibits thiol reactivity and off-target effects, including the inhibition of other FAD-containing enzymes, at concentrations required to reduce ERO1α activity.^22,43 In addition to these known off-target effects, we found that EN460 inhibits H₂O₂ activity and thrombin-induced ROS production (Figures 1M and S12A). Since antioxidants exert anti-thrombotic effects by inhibiting platelet function,⁴⁶ the anti-platelet and anti-thrombotic effects of EN460 may be attributed to its off-target effects. In contrast, M6766 has higher selectivity over other tested enzymes and does not exhibit antioxidant effects or thiol reactivity up to 100 μM (Figures 1M–1P and S12A). Furthermore, 1–5 μM EN460 significantly impairs the function of Ero1α/β-null platelets, whereas the same concentrations of M6766 do not exhibit such effects (Figures 3K–3N). Therefore, our study using Ero1α/β- or Ero1α-null platelets demonstrates that the anti-platelet activity of 1–5 μM M6766 results from Ero1α inhibition. However, since the deletion of Ero1α/β does not alter the basal level of platelet activation (Figures 3E and 3F), further studies are needed to investigate how high concentrations of M6766 enhance Ca²⁺ release, P-selectin exposure, and αIIbβ3 integrin activation in resting platelets.

In conclusion, our study demonstrates that selective inhibition of ERO1α represents a promising therapeutic strategy for the prevention or treatment of thrombotic diseases.

MATERIALS AND METHODS

M6766 was synthesized by WuXi AppTec (Tianjin, China). FeCl₃, A23187, D-Phe-Pro-Arg-chloromethyl ketone (PPACK), bovine insulin, human Mao-A enzyme, a Mao-A assay kit, glutathione (GSH), dithiothreitol (DTT), EN460, 2,3,5-triphenyltetrazolium chloride, rabbit immunoglobulin G, and prostaglandin E1 (PGE1) were obtained from Sigma-Aldrich (St. Louis, MO). Recombinant human ERO1β was from Biomatik (Wilmington, DE). TG, U46619, and isoquercetin were purchased from Cayman Chemical (Ann Arbor, MI). ClearColi BL21 (DE3) bacteria was from LGC Biosearch Technologies (Hoddesdon, UK). Isopropyl-β-d-thiogalactoside was purchased from Millipore (Burlington, MA). A bicinchoninic acid (BCA) assay kit and 3-(N-maleimidopropionyl)-biocytin (MPB) were obtained from Pierce (Waltham, MA). Catalase was purchased from MP Biomedicals (Irvine, CA). Horseradish peroxidase (HRP) was from Worthington Biochemical (Lakewood, NJ). A rabbit polyclonal anti-ERO1β antibody, a serotonin competitive ELISA kit, cell-permeant DCF, and Amplex Ultra Red (AUR) reagents were purchased from Invitrogen (Waltham, MA). A rabbit monoclonal anti-β3 integrin antibody (clone: D7X3P) was obtained from Cell Signaling Technology (Danvers, MA). A goat polyclonal anti-mouse P-selectin antibody and mouse CCL5/RANTES DuoSet ELISA kit were purchased from R&D Systems (Minneapolis, MN). Human thrombin, fibrillar type 1 collagen, and Chrono-lume were from Chronolog (Havertown, PA). CRP was purchased from Cambcol Laboratories (Ely, UK). 5,5-Dithio-bis-2-nitrobenzoic acid (DTNB), 3,3′-dihexyloxacarbocyanine iodide (DiOC6), HisPur Ni-NTA resin, and a biotinylation kit were purchased from Thermo Fisher Scientific (Waltham, MA). An FLIPR Ca²⁺ assay kit was obtained from Molecular Devices (San Jose, CA). Phycoerythrin (PE)- or fluorescein isothiocyanate (FITC)-conjugated antibodies against mouse P-selectin (clone: Wug.E9), GPVI, αIIbβ3 integrin (clone: MwReg30), or activated αIIbβ3 integrin (clone: JON/A), and DyLight 649-conjugated anti-mouse CD42c (GPIbβ) and anti-CD42b (GPIbα) antibodies were purchased from Emfret Analytics (Eibelstadt, Germany). A mouse monoclonal anti-β-actin antibody was obtained from LI-COR Biosciences (Lincoln, NE). PE-conjugated anti-human P-selectin (clone: AK-4) and FITC-conjugated mouse anti-activated αIIbβ3 integrin antibodies (clone: PAC-1) were from BD Biosciences (San Jose, CA). Eptifibatide was obtained from Schering Plough (Kenilworth, NJ). FITC-annexin V, cell staining buffer, and cell-binding buffer were purchased from BioLegend (San Diego, CA).

Mice

C57BL/6 and Pf4-cre mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Ero1a^flox/flox mice and megakaryocyte-specific CKO mice were reported previously.¹² Ero1b^flox/flox mice were generated by inserting loxP sites flanking exons 3, 4, and 5 using the CRISPR-Cas9 system at the Genome Engineering and iPSC Center at Washington University in St. Louis. Male and female Ero1b^flox/+ mice were bred to generate Ero1b^flox/flox mice. The following PCR primers were used for genotyping: 5′-GCATGAGGAGACCCAAGGACAAC-3′ and 5′-CCCATCTGATGGCTGACGAAAC-3′ for Ero1a WT (375 bp) and flox (457 bp); 5′-TGCTGCTGTCAAGAGTGAGG-3′ and 5′-GCCCAAGGGTCTGCAATAGT-3′ for Ero1b WT (304 bp) and flox (344 bp); and 5′-CCCATACAGCACACCTTTTG-3′ and 5′-TGCACAGTCAGCAGGTT-3′ for Pf4-cre (450 bp). WT littermate control (Ero1b^flox/flox) and megakaryocyte-specific Ero1β CKO mice (Ero1b^{flox/flox; Pf4−cre+/−}) were generated by breeding Ero1b^flox/flox mice with Pf4-cre mice. To generate Ero1a/b^flox/flox mice, Ero1a^flox/flox mice were crossed with Ero1b^flox/flox mice. WT control (Ero1a/b^flox/flox) and megakaryocyte-specific Ero1α/β double CKO mice (Ero1a/b^{flox/flox;Pf4−cre+/−}) were generated by breeding Ero1a/b^flox/flox mice with crossing Pf4-cre mice. Age-matched (8–12 weeks old) male and female mice were used in all studies except IVM and ischemic stroke, in which male mice were used. All animal care and experimental procedures were approved by the Institutional Animal Care and Use Committee at the Washington University School of Medicine. Animals were assigned randomly to the different experimental groups.

Recombinant human ERO1α and PDI purification

Recombinant 6× His-tagged human ERO1α and PDI were expressed in the ClearColi BL21 (DE3) bacteria (LGC Biosearch Technologies) as we described.^12,33 After growth at 37°C to an optical density at 600 nm (OD₆₀₀) of 0.6–0.8, the protein expression was induced by 0.1 mM isopropyl-β-d-thiogalactoside for an overnight induction. Bacteria were harvested by centrifugation, resuspended in a lysis buffer (phosphate-buffered saline [PBS], pH 7.4, containing 1% Triton X-100, 1% NP-40, 1 mM EDTA, 200 μM PMSF, and 1 mg/mL pepstatin A), and lysed by sonication. After centrifugation for 30 min at 12,000 rpm, the supernatant was subjected to Ni²⁺ affinity chromatography. The concentration of purified ERO1α or PDI was determined by a BCA assay. PDI was further incubated with 10 mM of DTT overnight at 4°C to cleave a disulfide bond in the CysGlyHisCys active sites. The excess DTT was removed by size exclusion chromatography using PD-10 gel filtration (GE Healthcare).

ERO1α activity assay

ERO1α activity assay was performed as we reported.¹² Recombinant ERO1α was incubated with vehicle (1% DMSO) or an ERO1α inhibitor in 20 mM sodium phosphate buffer, pH 7.4, containing 65 mM NaCl and 1 mM EDTA for 20 min at room temperature (RT) in wells of a 384-well black clear-bottom plate (Greiner Bio-One, Monroe, NC). In some experiments, 1 μM FAD was added and incubated for 10 min. After additional incubation with 900 nM reduced PDI and 0.1 U/mL HRP, the reaction was initiated by the addition of 5 μM AUR. The fluorescence signal was measured at 535 ± 20 nm excitation and 590 ± 20 nm emission on a FlexStation 3 microplate reader (Molecular Devices).

HTS

Recombinant human ERO1α (200 nM) in a buffer (20 mM sodium phosphate buffer, pH 7.4, 65 mM NaCl, 1 mM EDTA, 0.01% BSA) was added to wells of a 384-well black clear-bottom plate that had been pre-dispensed with 50 nL of DMSO control or a library compound. After a 30 min incubation, the reaction was started by the addition of reduced human PDI (900 nM) combined with AUR (5 μM) and HRP (0.1 U/mL) in the assay buffer without BSA. The fluorescence signal was measured at 530 nm excitation and 590 nm emission on a platelet reader (FlexStation 3, Molecular Devices). A chemical library containing 39,901 drug-like compounds was provided by the Center for Drug Discovery at Washington University in St. Louis and screened at a final concentration of 25 μM.

Amplex Red assay

H₂O₂, 2 μM, was incubated with vehicle (1% DMSO) or various concentrations of inhibitors in 20 mM sodium phosphate buffer, pH 7.4, containing 65 mM NaCl and 1 mM EDTA in a well of a 384-well black clear-bottom plate for 15 min. The Amplex Red assay was performed as described above. Catalase was used as a positive control.

BLI

The BLI assay was performed on the Octet K2 platform (ForteBio, Ann Arbor, MI) using high-density streptavidin biosensors (Sartorius, Ann Arbor, MI) at 25°C with an orbital shake speed of 1,000 rpm. Human ERO1α and STIM1 were biotinylated according to the manufacturer’s instructions (Thermo Fisher Scientific) and captured on super-streptavidin and streptavidin biosensors, respectively. The vehicle control (0.5% DMSO) or various concentrations of M6766 were incubated with ERO1α-captured biosensors or 2.5 μM ERO1α in a binding buffer (Octet Kinetics Buffer, Sartorius). All data were analyzed using Octet Analysis Studio 12.2. The specific interaction between ERO1α and M6766 or between STIM1 and ERO1α was determined by subtracting the non-specific binding.

Docking model

Computational studies were completed in Maestro (version 14.0.136), where the ERO1α protein structure was imported (PDB: 3AHQ) and prepared via Maestro Protein Preparation Workflow under default settings. A receptor grid was generated from the prepared 3AHQ structure centered at the centroid of FAD (Maestro Receptor Grid Generation; default). M6766 and FAD were prepared via Maestro LigPrep, while generating possible states at a target pH of 7.00 ± 0.5, and further optimized with Jaguar Single Point Optimization (density functional theory; B3LYP-D3/6-31G** functional) to generate atomic electrostatic potential charges. Docking studies were conducted via Glide XP Docking.

Mao-A activity assay

Mao-A activity was measured using a Mao-A assay kit according to the manufacturer’s instructions. The fluorescence signal was measured at 530 nm excitation and 585 nm emission. Clorgyline was used as a positive control.

PDI activity assay

PDI activity was measured using an insulin turbidity assay as we described elsewhere.³⁴ PDI, 8 μg/mL, was incubated with vehicle (1% DMSO) or various concentrations of inhibitors for 15 min in sodium phosphate buffer, pH 7.4, in a well of a 384-well black clear-bottom plate. Bovine insulin, 1 mg/mL, in 50 mM Tris-HCl buffer, pH 7.5, was added to the reaction. After the addition of 1 mM DTT, the reaction was monitored at OD₆₅₀ nm over 60 min. Isoquercetin was used as a positive control.

Ellman’s assay

GSH, 100 μM, was incubated with vehicle (1% DMSO) or various concentrations of inhibitors in 100 mM sodium phosphate buffer, pH 8.0, containing 1 mM EDTA in a well of a 384-well black clear-bottom plate for 10 min. After the addition of 0.5 mM DTNB, the absorbance was read at 412 nm. MPB was used as a positive control.

Isolation of platelets, neutrophils, and monocytes

Mouse platelets were isolated as we described elsewhere.¹² Citrate-dextrose solution (ACD)-treated blood from male and female mice (8–12 weeks old) was centrifuged at 300 × g for 20 min. The plasma and the top one-third of red blood cells were collected and re-centrifuged at 700 × g for 4 min. The plasma-rich platelets (PRPs) were collected and mixed with 1/9 volume of HEPES-Tyrode buffer (20 mM HEPES, pH 7.3, 136 mM NaCl, 12 mM NaHCO₃, 2.7 mM KCl, 1 mM MgCl₂, and 5 mM glucose) containing 10% ACD and 0.5 μM PGE1, followed by centrifugation at 700 × g for 4 min. The pellet was washed once with HEPES-Tyrode buffer containing 10% ACD and re-suspended in HEPES-Tyrode buffer to a density of 2 × 10⁸ cells/mL unless stated otherwise. For human platelets, ACD-treated blood was centrifuged at 200 × g for 20 min at RT. PRPs were collected and mixed with 10% ACD and 0.5 μM PGE1. After centrifugation at 1,000 × g for 10 min, the pellet was washed with HEPES-Tyrode buffer containing 10% ACD and 0.1 μM PGE1. After centrifugation at 750 × g for 5 min, platelets were re-suspended with HEPES-Tyrode buffer as described above. Neutrophils and monocytes were isolated from WT control and Ero1α/β double CKO mice as we described elsewhere.⁴⁷ All healthy human donors provided informed consent. The collection and use of human blood samples for laboratory analysis were approved by the institutional review board of the Washington University School of Medicine.

Flow cytometry

Platelets were pretreated with vehicle (0.1% DMSO) or various concentrations of M6766 or EN460 for 20 min at 37°C and then incubated with various concentrations of an agonist for 5 min at 37°C. Cells were incubated with fluorescently labeled antibodies against P-selectin or activated αIIbβ3 integrin (JON/A for mouse and PAC-1 for human) for 15 min, followed by flow cytometry (CytoFlex, Beckman Coulter, Brea, CA). Data were obtained using CytExpert (version 2.2) and presented as the geometric mean fluorescence intensity.

Platelet aggregation assay

Platelets were pretreated with vehicle (0.1% DMSO) or various concentrations of M6766 or EN460 for 20 min at 37°C. After incubation with various concentrations of an agonist, platelet aggregation and ATP secretion were measured in a lumi-aggregometer (Chronolog) at 37°C with stirring (1,000 rpm).

Detection of intracellular ROS

Platelets from C57BL/6 mice, WT control, or Ero1α/β CKO mice were pretreated with or without vehicle (0.1% DMSO) or various concentrations of M6766 or EN460 for 20 min at 37°C, followed by incubation with 1 μM DCF for 10 min. After stimulation with thrombin or CRP for 5 min, DCF fluorescence was measured by flow cytometry.

Analysis of cell viability

Platelets from C57BL/6 mice, WT control, or Ero1α/β CKO mice were incubated with or without vehicle (0.1% DMSO), or various concentrations of M6766 or EN460 for various time points at 37°C. At various time points, platelets were labeled with FITC-annexin V for 20 min according to the manufacturer’s instructions, followed by flow cytometry.

Measurement of CCL5/RANTES in platelet releasates

Mouse platelets (5 × 10⁸ cells/mL) were pretreated with vehicle (0.1% DMSO) or various concentrations of M6766 or EN460 for 20 min at 37°C. After treatment with 0.01 U/mL thrombin or 0.05 μg/mL CRP for 10 min, the supernatant was collected, and CCL5/RANTES was measured by a CCL5/RANTES ELISA kit according to the manufacturer’s instructions. Samples were read using a FlexStation 3 microplate reader at 450 nm.

Isolation of pancreatic islet cells

Ice-cold type V collagenase (0.45 mg/mL, Sigma-Aldrich) in Hank’s balanced salt solution (HBSS) was perfused into the bile duct. After digestion at 37°C and washing with HBSS, islets were handpicked under a stereomicroscope and incubated overnight in RPMI media supplemented with 10% fetal calf serum, 100 U/mL penicillin and 100 μg/mL streptomycin (Thermo Fisher Scientific).

Glucose-stimulated insulin secretion and insulin content measurements

After overnight incubation, islets of similar size across all samples were incubated for 3 h in Krebs-Ringer Bicarbonate-HEPES buffer, pH 7.4, in the following conditions: (1) low glucose (LG, 2.8 mM), (2) high glucose (HG, 16.7 mM), (3) HG + vehicle (0.1% DMSO), (4) HG + 1 μM EN460, (5) HG + 5 μM EN460, (6) HG + 1 μM M6766, and (7) HG + 5 μM M6766. Ten islets from each group were washed, pre-incubated in LG for 1 h for stabilization, and then incubated under the seven conditions described above for 1 h at 37°C, with two technical replicates per sample. Samples were centrifuged, and the supernatant was collected for insulin secretion. Islet cells were lysed in 0.2 N/80% acid-ethanol for measurement of total insulin content. Insulin concentrations were determined using a commercially available ELISA kit (Crystal Chem, Elk Grove Village, IL).

Flow chamber assay

A microfluidic system (BioFlux200, Fluxion Biosciences, Alameda, CA) was described previously.⁴⁸ Briefly, BioFlux 1000z 48-well plate 0–200 dyne/cm² (Fluxion Biosciences) was coated with 100 μg/mL fibrillar type 1 collagen for 1 h at 37°C. Plate wells were washed with PBS containing 0.05% BSA. ACD-treated mouse or human blood was treated with vehicle (0.1% DMSO), 5 μM M6766, or EN460 for 15 min at 37°C and then 1 μM DiOC6 for 5 min. After the addition of 6 mM CaCl₂ and 3 mM MgCl₂, mouse blood was perfused over collagen-coated surfaces under 50 dyne/cm² for 2 min. For human blood, 18 mM CaCl₂ and 18 mM MgCl₂ were added before perfusion for 5 min. Fluorescence and bright-field images were recorded using a Zeiss Axio Observer 7 microscope system equipped with a Colibri LED fluorescence light source and four filter sets (385, 475, 555, and 630 nm wavelengths). Images were collected using an objective (A-Plan 10×, Zeiss) and a high-speed, high-resolution camera (2,304 × 2,304 pixel format; ORCA-Fusion BT, Digital Camera C15440, Hamamatsu, Shizuoka, Japan). Thrombus formation was assessed by measuring the percentage of surface coverage area and volume of the thrombi. The data were analyzed using BioFlux software.

Ca²⁺ mobilization

Mouse platelets were preincubated with vehicle (0.1% DMSO), M6766, or EN460 for 20 min and incubated with a Ca²⁺ dye (FLIPR Ca²⁺ 5 assay kit) for 30 min at 37°C in the dark, followed by incubation with 0.02 U/mL thrombin, 0.5 μM A23187, or 5 μM TG. After measurement of Ca²⁺ release, 2 mM CaCl₂ was added to evaluate Ca²⁺ influx. Cytosolic Ca²⁺ levels were measured using a FlexStation 3 microplate reader with an excitation at 485 nm and emission at 525 nm. Ca²⁺ mobilization was quantified by calculating the area under the curve and expressed as a relative fluorescence unit.

Measurement of plasma concentrations of M6766

M6766 (0.3 μg/g BW) was administered to C57BL/6 mice via tail vein injection. Blood samples were collected from the saphenous vein at 5, 15, 30, 60, 120, 240, and 480 min post-injection. Plasma concentrations of M6766 were measured by LC-MS/MS using an Applied Biosystems Sciex Instruments Triple Quad 5500 (Framingham, MA). Concentrations were quantified by comparison with a standard curve of M6766.

IVM

IVM was performed in a mouse model of laser-induced cremaster arteriolar injury as we described elsewhere.¹² Male mice (8–12 weeks old) were anesthetized with intraperitoneal injection of ketamine and xylazine. The cremaster muscle was exteriorized and superfused with 37°C bicarbonate-buffered saline throughout the experiment. Arteriolar wall injury was induced by laser ablation (Ablate, Intelligent Imaging Innovations). Multiple thrombi were generated in 3–4 different arterioles of one mouse. Platelets were visualized by infusion of a DyLight 649-conjugated anti-CD42c (0.1 μg/g BW) antibody. Mice were intravenously injected with vehicle (1% DMSO), M6766, or EN460 (0.3 μg/g BW) 10 min before treatment with the anti-CD42c antibody. The experiments were performed in a single-blind fashion in which the investigator did not know the identity of the sample. Fluorescence and bright-field images were captured in cremaster arterioles with a diameter of 30–45 μm in each mouse and recorded using a Zeiss Axio Examiner Z1 microscope system with a Yokogawa confocal spinning disk (CSU-W1) equipped with four-stack laser system (405, 488, 561, and 637 nm wavelengths). Images were collected using an objective (W Plan-Apochromat 63×, Zeiss) and a high-speed, high-resolution camera (2,304 × 2,304 pixel format; ORCA-Fusion BT, Digital Camera C15440, Hamamatsu). Time 0 was set to when image capture began on each vessel. The data were analyzed using Slidebook 2024 (version 1.0, Intelligent Imaging Innovations, Denver, CO). Due to the variation between arterioles in the same mouse, the result from each vessel was counted as an individual value. Representative images were chosen that most closely resemble the median in the quantifications.

FeCl₃-induced carotid arterial thrombosis

Mice (8–12 weeks old male and female) were anesthetized with intraperitoneal injection of ketamine and xylazine mixture. Mice were intravenously injected with vehicle (1% DMSO), M6766, EN460 (0.3 μg/g BW), or eptifibatide (5 μg/g BW). The right carotid artery was isolated from surrounding tissues. A nanoprobe (MA-0.5PSB) was hooked to the artery, and blood flow was monitored with a TS420 flowmeter (Transonic Systems, Ithaca, NY). After stabilization, 1.2 μL 7 or 10% FeCl₃ was applied to a filter paper disc (2 × 2 mm²), and the paper was placed on the top of the artery for 2 min. After removing the filter paper, blood flow was monitored continuously until 5 min after occlusion. The TTO was measured as a difference in time between the removal of the filter paper and stable occlusion (no blood flow for 5 min).

Bleeding time and hemoglobin content

Mice (8–12 weeks old male and female) were anesthetized with intraperitoneal injection of ketamine and xylazine mixture. Mice were intravenously injected with vehicle (1% DMSO), M6766, EN460 (0.3 μg/g BW), or eptifibatide (5 μg/g BW). Using a sharp razor blade, 5 mm of the tail tip was cut, and the tail was held in a 15 mL tube containing 13 mL of PBS prewarmed to 37°C. Tail bleeding was monitored, and the time to cessation of blood flow was measured. Blood loss was quantified by measuring the hemoglobin content of blood collected into 13 mL PBS. After centrifugation, the red blood cells were lysed with 5 mL lysis buffer (8.3 mg/mL NH₄Cl, 1.0 mg/mL KHCO₃, and 0.037 mg/mL EDTA). The absorbance of the samples was measured at 575 nm.

tMCAO-induced ischemic stroke

tMCAO surgery was performed as we described.⁴⁸ Male mice (10–12 weeks old) were anesthetized with 1.2%–1.5% isoflurane in an oxygen mixture and treated by intraperitoneal injection of buprenorphine-ER (0.1 μg/g mouse). The body temperature was maintained at 37°C throughout surgery using a warming pad. After exposure of the left common carotid artery, external carotid artery, and internal carotid artery, the MCA was occluded for 1 h using silicon rubber-coated 6.0 nylon monofilament (Doccol Corporation, Sharon, MA). Cerebral blood flow was monitored using a laser Doppler perfusion system (PeriFlux System 5000, Perimed, Järfälla, Sweden). Mice with a reduction in the regional cerebral blood flow to less than 10% of the baseline were counted as a successful occlusion, and only those mice were included in the study. After 1 h of MCA occlusion, the filament was removed, and blood flow was restored to the baseline. Vehicle (1% DMSO), M6766, or EN460 (0.3 μg/g BW) was intravenously injected into mice. Following 23 h of reperfusion, neurological function was assessed by Bederson scores and grip strength. The Bederson scores were measured according to the following system: 0, no observable deficit; 1, forelimb flexion; 2, forelimb flexion and circling; 3, forelimb flexion, circling, and decreased resistance to lateral push; 4, death. For the grip strength test, tMCAO-challenged mice were placed on the iron grid connected to a strength meter (GT3, Bioseb). Maximal peak force (gram) was recorded as the mice pulled away from the grid, serving as an indicator of neuromuscular function post-tMCAO. The test was repeated 10 times, with intervals of 3–5 min between each measurement, and each maximal peak force was normalized to BW. Upon completion of neurological behavioral assessments, mouse brains were collected, and the 2-mm brain slices were stained with a solution of 2% 2,3,5-triphenyltetrazolium chloride for 20 min, followed by fixation in 4% paraformaldehyde. Sections were scanned, and infarct volume was quantified using ImageJ version1.54f. This experiment was conducted in a blinded manner.

Statistical analysis

Statistical significance was assessed by unpaired Student’s t test or Mann-Whitney U test for comparisons between two groups. For multiple group comparisons, one-way or two-way ANOVA and either Dunnett’s test or Tukey’s test were used. All analyses were performed using GraphPad Prism (version 10.2.3), and p < 0.05 was considered statistically significant.

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.ymthe.2025.07.033.

DATA AVAILABILITY

All data are available from the corresponding author upon request.