The whole plant of GJ was obtained from the Natural Product Central Bank (Ochang-eup, Cheongju-si, Chungcheongbuk-do, Republic of Korea) registered with voucher number KPM032-039 (https://www.kobis.re.kr/npcb/uss/main.do). Dried whole plant of GJ was extracted using 50% ethanol for 2 h at 80°C at a ratio of 1:20 (dried plant: solvent) in a reflux setup to minimize evaporation. After extraction, the extract was filtered with Grade 1 Whatman^® filter paper (GE Healthcare, PA, United States) and concentrated using a rotary evaporator (Rotavapor^® R-100; BUCHI Labortechnik, Switzerland) to remove excess solvent. The extract was frozen and lyophilized to obtain the solvent-free powdered extract that was stored at −30°C until use. Powdered extract of GJ was weighed and dissolved in dimethylsulfoxide (DMSO) for subsequent experiments.

UPLC–QTOF-MS analysis was performed using an Acquity H-Class Plus UPLC system (Waters Corporation, Milford, MA, United States) interfaced with an Acquity RDa QTOF-MS detector (Waters). Chromatographic separation was achieved using an Acquity UPLC HSS C18 reverse-phase column (2.1 × 100 mm, 1.8 µm particle size; Waters). The mobile phase consisted of solvent A (distilled water containing 0.1% formic acid) and solvent B (acetonitrile containing 0.1% formic acid). A linear gradient elution was performed, starting with 10% solvent B for 2 min, increasing to 100% over 25 min, and held at 100% for an additional 5 min. The flow rate was maintained at 0.2 mL/min, and the injection volume was 1 µL. Mass spectrometric detection was carried out in negative electrospray ionization (ESI) mode, with a desolvation temperature of 550°C, a capillary voltage of 0.8 kV, and a cone voltage of 30 V.

Collagen, ADP and thrombin were obtained from Chrono-Log (Havertown, PA, United States). The adenosine triphosphate (ATP) assay kit was from the Biomedical Research Service (Buffalo, NY, United States). Enzyme-linked immunoassay (ELISA) kits for thromboxane B2 (TXB2; item no. 501020), cyclic adenosine monophosphate (cAMP; item no. 501040), and cGMP (item no. 581021) were obtained from Cayman Chemicals (Ann Arbor, MI, United States). The ELISA kit for serotonin was obtained from LDN (BA E-5900R; Nordhorn, Germany). FITC-conjugated fibrinogen and Fura-2AM were from Molecular Probes (Eugene, OR, United States). Antibodies for protein detection: phospho-Akt, total-Akt, phospho-PI3K, total-PI3K, phospho-ERK, total-ERK, phospho-JNK, total-JNK, phospho-p38, total-p38, VASP ser²³⁹, VASP ser¹⁵⁷, total-VASP, and β-Actin were acquired from Cell Signaling Technologies (Beverly, MA, United States). ERK inhibitor (PD98059), p38 inhibitor (SB203580), JNK inhibitor (SP600125), PI3K inhibitor (LY294002), and Rho kinase inhibitor (Y-27632) were acquired from Tocris Bioscience (Bristol, United Kingdom).

Seven-week-old male Sprague-Dawley rats and 6-week-old ICR mice were acquired from Orient Bio (Gyeonggi-do, Republic of Korea). The animals acclimated for 1 week in a pathogen-free facility (humidity of 50% ± 10%, temperature of 23°C ± 2°C) with a 12 h light-dark cycle. The experiments adhered to IACUC guidelines and received approval from the Ethics Committee of the College of Veterinary Medicine at Kyungpook National University (approval number: 2012-125). Animal procedures were conducted in compliance with humane practices.

Blood was collected from rats via cardiac puncture using acid citrate dextrose solution as an anticoagulant. The collected blood was then centrifuged at 1,000 rpm for 7 min to obtain platelet-rich plasma (PRP). Next, the PRP was then centrifuged at 2,000 rpm for 7 min. The resulting pelleted platelets were resuspended in Tyrode’s buffer (137 mM NaCl, 12 mM NaHCO₃, 5.5 mM glucose, 2 mM KCl, 1 mM MgCl₂, and 0.3 mM NaHPO₄, pH 7.4) to obtain washed platelets.

Prepared washed platelets underwent platelet aggregation induced by agonists (collagen, thrombin or ADP). Platelet aggregation was detected using a light-transmission aggregometer from Chrono-log (Havertown, PA, United States). The platelets were treated with 1 mM of CaCl₂, with or without GJ treatment, and pre-incubated for 1 min at 37°C before adding collagen (2.5 μg/mL), thrombin (0.1 U/mL) and ADP (10 μM). Aggregation was halted after 5 min. The experiments were repeated in triplicate with platelets from three different rats.

For ATP assay, platelet aggregation was conducted as described above. The platelets were collected and subjected to centrifugation at 7,500 rpm for 5 min. The supernatant was then collected for the ATP assay, following the manufacturer’s instructions. Intracellular calcium ion mobilization was detected using Fura-2AM. PRP was incubated with Fura-2AM. Next, platelets were pre-treated with 1 mM CaCl₂, with or without GJ, and incubated for 1 min at 37°C before stimulation with collagen. Calcium mobilization was measured using a fluorescence spectrophotometer at 224 nm (F-2500; Hitachi, Tokyo, Japan). The experiments were repeated in triplicate from three different animals.

Platelets were incubated with GJ for 1 min before activation by collagen for an additional 5 min. We used FITC-conjugated fibrinogen to investigate the activation of integrin αIIbβ3. Afterward, the platelets were fixed with 0.5% paraformaldehyde and subjected to flow cytometry analysis using a BD FACS Aria^TMIII cell sorter (BD Biosciences, Franklin Lakes, NJ, United States). As a positive control, we treated platelets with 10 μM of EGTA. The experiments were repeated in triplicate from samples of three different animals.

Collagen-induced platelet aggregation was performed as mentioned above. Protein extraction from platelets utilized Pro-Prep (iNtRON Biotechnology, Gyeonggi-do, Republic of Korea). The protein concentration was determined using the Bradford method and normalized before loading onto a 10% SDS-PAGE. Subsequently, proteins were transferred to PVDF membranes and blocked with skim milk before incubation with the target primary antibodies (1: 3,000) at 4°C overnight. Membranes were incubated with secondary antibodies (1: 1,000) for 90 min at room temperature. Protein expressions were visualized using enhanced chemiluminescence in a developer (General Electrics, United States). The experiment was repeated in three biological replicates from three different rats.

To investigate the secretion of TXB2 and serotonin, the assays were performed as described earlier in accordance to the manufacturer’s instructions (Irfan et al., 2021). Briefly, platelets suspension was immediately placed on ice after platelet aggregation assay. The supernatant was collected after centrifugation at 2,000 × g for 10 min at 4°C and analysed using ELISA assay. Intracellular cAMP and cGMP level detection was conducted as previously reported (Irfan et al., 2022). After platelet aggregation, the platelet mixture was immediately incubated on ice and then froze in −80°C. The next day, the mixture was thawed and centrifuged at 2,000 × g for 10 min at 4°C. The supernatant was obtained for ELISA assays, and were conducted following the manufacturer’s instructions. The plates were read at 420 nm on a microplate reader (Versamax; Molecular Devices, San Jose, CA, United States), as suggested in the manufacturer’s instructions. Experiments were repeated in triplicates from platelets of three different animals.

Blood was collected from rats as described above and conducted as previously described (Irfan et al., 2022). Briefly, blood was centrifuged, and platelet-rich plasma (PRP) was separated. Extract of GJ and 10 μM of Y-27632 were added to the PRP before introducing 5 μL of RBCs. Thrombin was then added to make a final concentration of 1 U/mL and thoroughly mixed. The platelet mixture was then incubated for 2 h, and the thrombus weight was measured. The experiment was repeated using samples from three different animals.

Following the platelet aggregation procedure described earlier, the platelets were directly fixed with 0.5% paraformaldehyde. The platelets were then oxidized using osmium tetraoxide and lyophilized before visualization using a field emission scanning electron microscope (SU8220; Hitachi, Tokyo, Japan). For platelet-neutrophil aggregate assays, cells were pre-treated with or without GJ for 1 min and activated with 2 μM of ADP. After 5-min incubation at room temperature, the cells were fixed, processed, and visualized as previously mentioned (Irfan et al., 2021).

Platelets were treated with or without GJ, as mentioned above. Collagen was added to induce platelet aggregation for 30 s, followed by direct fixation using a solution of 2% glutaraldehyde and 2% paraformaldehyde in 0.2 M cacodylate buffer for 1 h at room temperature. The platelets were centrifuged at 5,000 rpm for 5 min and then washed with 0.1 M sodium cacodylate buffer. Secondary fixation involved 1% osmium tetraoxide in 0.1 M sodium cacodylate buffer, conducted at room temperature and shielded from light for 2 h. The fixed pellet was then washed with ultra-pure water and incubated overnight at 4°C with 3% UA-Zero EM Stain (a substitute for uranyl acetate from Agar Scientific, Stansted, ESS, United Kingdom) in 30% acetone. The fixed platelets were then dehydrated using increasing concentrations of acetone and epoxy (Sigma-Aldrich, St. Louis, MO, United States). Finally, the platelets were mixed with the epoxy mixture and dried in an oven for polymerization. The epoxy was then subjected to sectioning (EM UC7/FC7; Leica, Wetzlar, Germany) onto copper grids and visualized with a transmission electron microscope (HT 7700; Hitachi, Tokyo, Japan).

Whole blood was collected from 6-week-old ICR mice in Gyeonggi-do, Republic of Korea. These mice had acclimatized for a week in a pathogen-free facility with a 12-h light-dark cycle. Blood was harvested via cardiac puncture and stored in sodium citrate vacutainers (BD, Franklin Lakes, NJ, United States). Collected blood samples were pre-treated with or without GJ for 1 min before stimulation with 2 μM of ADP and were allowed to incubate for 5 min. The blood was then fixed with 1% paraformaldehyde before RBC lysis using ACK lysis buffer. Blood cells were stained with Alexa Fluor 488 conjugated CD41 (BioLegend, San Diego, CA, United States) and PE-conjugated Ly6G (BD, Franklin Lakes, NJ, United States). After staining, subpopulations of platelet-neutrophil aggregates were analyzed via flow cytometry (FACSAria^TMIII; BD Biosciences, Franklin Lakes, NJ, United States). Post-analysis was conducted using Flowlogic version 7 (Miltenyi Biotec, Bergisch Gladbach, Germany). The experiment was repeated in triplicates, from samples of three different mice.

We investigated the statistical significance of this study using GraphPad Prism 7.00 (San Diego, CA, United States). A one-way ANOVA, followed by a single-step Dunnett’s post hoc test was conducted. Normality of the dataset was tested using the D’Agostino and Pearson test. The data were presented as mean ± SD, and P < 0.05 was considered significant.

Our analysis identified eight peaks, which were identified based on reference data from Yang et al. (2025) (Figure 1). Peaks 1, 3 and 4 were identified as 23-Hydroxytormentic acid (retention time; RT: 11.94, 14.56, 15.36 min). This compound is determined to be the major metabolite of GJ. The second most abundant metabolite is tormentic acid (RT: 17.12 min), followed by geumonoid (RT: 19.79 min). Other major metabolites include (10E,12Z)-9-Hydroxy-10,12-octadecadienoic acid (RT: 20.63 min) and pomolic acid (RT: 20.92 min). Peak 2, corresponding to 2α,3β,19α-trihydroxyurs-12-en-23,28-dioic acid, exhibited a relatively low detector count and was the least abundant among the six principal compounds (Table 1).

Various doses of GJ treatment inhibited aggregation induced by collagen (Figures 2A, B), ADP (Figures 2C, D), and thrombin (Figures 2E, F) in a dose-dependent manner. Notably, GJ most effectively inhibited thrombin-induced platelet aggregation, followed by collagen and ADP.

Calcium ion mobilization is a marker of platelet activation, which has been detected to increase with stimulation by collagen. However, increasing doses of GJ effectively suppresses this response (Figure 3A). The activation of platelets causes the release of ATP, which has been increased with collagen. A 100 μg/mL dose of GJ significantly reduces ATP secretion (Figure 3B). Additionally, GJ suppresses serotonin secretion in stimulated platelets (Figure 3C). However, GJ does not impact the secretion of TXB2 (Figure 3D). Acetylsalicyclic acid (ASA) at 2.5 mg/mL was used as a positive control.

We investigated the intracellular concentration of cAMP and cGMP, essential in regulating platelet aggregation. Our findings showed that 100 μg/mL of GJ has significantly increased cGMP concentration (Figure 4A), while the concentration of cAMP remains unchanged (Figure 4B). This indicates that GJ exhibits its anti-platelet activity via the upregulation of cGMP rather than cAMP. The conformational change of integrin αIIbβ3 is vital for platelet aggregation and activation. Fibrinogen binds to αIIbβ3 on platelets, allowing them to form aggregates. Utilizing FITC-conjugated fibrinogen, we detected αIIbβ3 activation via inside-out signaling. We found that the percentage of fibrinogen that was bound increased in collagen-stimulated platelets, and this was decreased with 50 and 100 μg/mL of GJ. EGTA (10 μM) was a positive control (Figures 4C, D). The outside-in activation of αIIbβ3 activates the Rho kinase pathway, which is indicated by clot retraction. GJ significantly inhibited thrombin-induced clot retraction at 25 and 50 μg/mL (Figure 4E). The amount of trapped plasma indicates the rigidity of the fibrin mesh and is indicated by the weight of the thrombus (Figure 4F).

The PI3K/Akt and MAPK pathways are associated with platelet activation. Representative blots are shown in Figure 5A. Our findings show that the expression of p-PI3K and p-Akt were significantly suppressed by GJ (Figures 5B, C). Protein expressions of p-ERK and p-p38 were also significantly inhibited by GJ but not p-JNK (Figures 5D–F). The protein expressions of VASP ser239 and VASP ser157 were investigated to explore the role of cGMP and cAMP further. Our findings show that collagen-treatment increased expression of VASP ser¹⁵⁷ and was suppressed with GJ treatment. GJ treatment has also upregulated VASP ser²³⁹ expression (Figures 5G, H). The relative expressions of the blots were normalized against their total forms. β-Actin was used as a housekeeping gene. This confirms that GJ mainly prevents platelet aggregation via the cGMP pathway and inhibits proteins in the PI3K/Akt and MAPK pathways.

As platelet shape change is a vital step in the activation of platelets, scanning electron microscopy was used to observe the morphology of platelets (Figure 6A). With activation by collagen, the platelets appeared fibrous (Figure 6A[b]) as compared to platelets in the resting state (Figure 6A[a]). With treatment of increasing doses of GJ (25, 50 and 100 μg/mL), platelets appeared less activated as compared to when stimulated with collagen (Figure 6A[c]–[e]). Transmission electron microscopy was used to observe the intracellular structure of platelets (Figure 6B). From our findings, the activation of platelets with collagen caused degranulation of platelets, whereas treatment of GJ prevented this from happening. Alpha granules (α) and dense granules (δ) were identified. The open canalicular system (OCS) that transports granules from the platelets was visible in resting platelets but filled with granules in the collagen-stimulated group. The OCS was then observable after GJ treatment. This further confirms that the treatment of GJ has prevented the activation and degranulation of platelets.

A cGMP inhibitor (Rp-8-Br-PET-cGMPs) and agonist (200 μM of SNP) were used to elucidate the effects of GJ via the cGMP pathway. From our findings, 100 μg/mL of GJ has significantly inhibited platelet aggregation but treatment with the cGMP inhibitor (Rp-8-Br-PET-cGMPs) reduced the efficacy of GJ (Figures 7A, B). This indicates that GJ elevates cGMP to prevent platelet aggregation. In ADP-induced platelet aggregation, an inhibitor of cGMP slightly reduced GJ’s anti-platelet effect, which was not statistically significant (Figures 7C, D). Using an agonist of cGMP (SNP), a combination of GJ and SNP has demonstrated increased anti-platelet activity (Figures 7E, F). This further confirms that GJ acts via the elevation of cGMP to inhibit platelet aggregation. However, the inhibitor of cGMP did not wholly attenuate the ability of GJ to inhibit platelet aggregation. This indicates the anti-platelet mechanism of GJ involves upregulating cGMP and other pathways, such as the PI3K/Akt and MAPK pathways (Figure 5).

As the formation of PNAs contributes to thromboinflammatory diseases, we investigate whether GJ can inhibit the formation of PNAs. Our results show that ADP significantly increased the population of PNAs in the whole blood of mice, where it is decreased by treatment of GJ (Figure 8A), where the quantitative data of three technical replicates were presented in Figure 8B. We observed an increase in platelet population in the whole blood of mice (top left quadrant, CD41) treated with GJ compared to the control and ADP-treated groups. With SEM, we observed ADP-induced platelet activation with early signs of platelet aggregation. We observed that with 100 and 200 μg/mL of GJ treatment, platelets (indicated with red arrowheads) are at the early stages of platelet aggregation, identified by the early formation of pseudopodia. In contrast, leukocytes (marked with a red arrow) were not attached to platelets (Figure 8C).