Novel anti-thrombotic agent for modulation of protein disulfide isomerase family member ERp57 for prophylactic therapy

Protein disulfide isomerase (PDI) family members including PDI and ERp57 emerge as novel targets for anti-thrombotic treatments, but chemical agents with selectivity remain to be explored. We previously reported a novel derivative of danshensu (DSS), known as ADTM, displayed strong cardioprotective effects against oxidative stress-induced cellular injury in vitro and acute myocardial infarct in vivo. Herein, using chemical proteomics approach, we identified ERp57 as a major target of ADTM. ADTM displayed potent inhibitory effects on the redox activity of ERp57, inhibited the adenosine diphosphate (ADP)-induced expressions of P-selectin and αIIbβ3 integrin, and disrupted the interaction between ERp57 and αIIbβ3. In addition, ADTM inhibited both arachidonic acid (AA)-induced and ADP-induced platelet aggregation in vitro. Furthermore, ADTM significantly inhibited rat platelet aggregation and thrombus formation in vivo. Taken together, ADTM represents a promising candidate for anti-thrombotic therapy targeting ERp57.

Scientific RepoRts | 5:10353 | DOi: 10.1038/srep10353 the anti-oxidative stress Nrf2/heme oxygenase-1 (HO-1) pathway 7 . Several registered pharmaceutical products containing danshen and/or chuanxiong such as "Danshen Dripping Pill" and "Guanxinning Injection" have been widely used in China to treat CVDs. The "Danshen Dripping Pill" was approved by the US Food and Drug Administration (FDA) to undergo Phase III clinical trials in the treatment of CVDs 8 . Clinical studies indicated that these pharmaceutical products are effective in treating angina pectoris, a symptom of ischemic heart disease 9 . It is believed that active ingredients in danshen and chuanxiong induce beneficial therapeutic effects by the promotion of better blood circulation 10,11 . Therefore, we postulated that ADTM might also display anti-platelet and anti-thrombotic properties.
Several members of the protein disulfide isomerases have been shown as important players in the regulation of platelet functions and thrombus formation including PDI, ERp72, ERp57 and ERp5 [12][13][14] . However, which member in the family represents the best anti-thrombotic drug target is largely unknown 13 . The prototypical member PDI is a multifunctional redox chaperone protein that plays critical roles in a variety of diseases including platelet aggregation and thrombus formation in vivo and in vitro [15][16][17] . More recently, it is shown that blood platelets also secreted ERp57 (also known as GRP58 and PDIA3) in response to vascular injury to trigger platelet activation and aggregation and emerges as a novel target for anti-platelet aggregation and anti-thrombosis [18][19][20] . Human ERp57 has 32% amino acid sequence homology with PDI, and exhibits 40% identity with the a-b-b´-a´domains of ERp72 21 . However, selective chemical agents for ERp57 are not yet available.
In this study, we further identified the drug target of ADTM as ERp57 by using a chemical proteomics approach analyzing liquid chromatography linked to tandem mass spectrometry (LC-MS/MS) profile of pull-down proteins bound to a biotinylated-ADTM analogue (BAA). The anti-platelet aggregation and anti-thrombosis activities of ADTM were determined both in vitro and in vivo. The results provided promising evidences for the potential of developing novel anti-CVD and anti-thrombotic agents with selectivity at the protein disulfide isomerases by using ADTM.

Results
Protein disulfide isomerases are identified as targets of ADTM by chemical proteomics. In order to isolate the protein targets that bound to ADTM, a biotin-conjugated ADTM analogue (BAA, Fig. 1) was designed and synthesized in our laboratory as described in Supplementary Methods (Supplemental Fig. 3). BAA (300 μ M) was incubated with rat blood platelet lysates and the BAA-protein complexes were pulled down with NeutrAvidin-agarose followed by protein profiling using LC-MS/MS. It was observed that BAA could bind to various proteins involved in platelet function, including glycoprotein 5 (a potential marker of thrombotic platelet activation 22,23 ), superoxide dismutase and xanthine oxidase (regulators of vascular redox signaling 24 ). In particular, four platelet aggregation-associated proteins were identified with > 95% protein identification probability (Supplementary Table S1), including ERp72, ERp57 ERp5 and PDI, which are glycoprotein-specific members of the PDI family related to platelet function and redox homeostasis 25,26 . We confirmed the identity of the pulled-down proteins as ERp72, ERp57 and ERp5 with their respective antibodies by using immunoblot analysis (Fig. 2b). Furthermore, the specificity of the binding between BAA and these targets (ERp72, ERp57 and ERp5) could be demonstrated by the competition with untagged ADTM in excess (Fig. 2a). The molecular docking analysis demonstrated that the free binding energy to the active cavity of ERp57 (PDB: 3F8U) is − 6.5 kcal/mol. ADTM may form hydrogen bonds with side chains of ARG47, ILE453 and PHE450, respectively, in the catalytic center (CYS406/CYS409) of ERp57 (Supplemental Fig. 2). ERp72, ERp57, ERp5 and PDI are identified as important new targets in redox homeostasis of blood platelet aggregation 12 , therefore we further evaluated the effects of ADTM on the reducing activities of the PDI family members. As shown in Fig. 2c, ADTM exhibited potent inhibition on the redox activity of ERp57 in a concentration-dependent manner (IC 50 = 100 to 300 μ M). ADTM inhibited the activities of ERp72, ERp5 and PDI to a much lesser extent (18%, 38% and 30% of inhibition at 300 μ M, respectively), displaying differential inhibitory activity at ERp57 (order of potency as follows: ERp57 > ERp5 > PDI > ERp72). ERp57 inhibition by ADTM was irreversible after dilution of the ERp57-ADTM complex (Supplemental Figure 1), indicating that ADTM may covalently bind ERp57.

ADTM inhibits platelet aggregation in vitro.
It was observed that ADTM exhibited concentration-dependent inhibition on ADP-induced platelet aggregation (Fig. 3a) as well as AA-induced platelet aggregation (Fig. 3b) in vitro and its effects were comparable to first-line clinical anti-platelet drugs, aspirin and clopidogrel. The parent compounds DSS and TMP, alone or in combination, could only inhibit AA-induced platelet aggregation but not ADP-induced platelet aggregation. These results demonstrated that ADTM exhibited broad-spectrum anti-platelet activities.
ADTM inhibits the expression of P-selectin. To further investigate the underlying mechanism of ADTM anti-platelet activity, we examined the effect of ADTM on the expression of P-selectin (also Proteins isolated from rat platelets were pretreated with DMSO or excess amount of untagged ADTM for 2 h, followed by treatment with BAA for another 2 h, and compound-protein-complexes were enriched with NeutrAvidin-agarose resin. The precipitates were detected by immunoblot for indicated proteins of PDI family. (c) ADTM selectively inhibits ERp57. The recombinant PDI, ERp57 and ERp72 proteins were incubated with different concentrations of ADTM for 100 min, and its catalytic activity was determined in insulin reduction assay. known as CD62P), which is an adhesion molecule expressed on the surface of platelet and is a critical marker of platelet activation, by using immunoreactivity and flow cytometry analysis. We observed that ADTM concentration-dependently inhibited ADP-induced expression of P-selectin and attenuated the expression of P-selectin by more than 65% at 100 μ M ( Fig. 4a and 4b).
ADTM inhibits activation of αIIbβ3 and disrupted ERp57/αIIbβ3 interaction. Recently,ERp57 has been reported to directly interact with α IIbβ 3 integrin and regulate its function in platelet aggregation 20 . In line with this, we observed that ADTM could modulate α IIbβ 3 and concentration-dependently inhibited ADP-induced α IIbβ 3 activation, by using flow cytometry analysis with PAC-1, a monoclonal antibody that binds only the activated form of α IIbβ 3 ( Fig. 5a,b). We further examined the action of ADTM on the ERp57/α IIbβ 3 interaction by immunoprecipitation with anti-α IIbβ 3 antibody, followed by detection of ERp57 immunoreactivity using immunoblotting. It was found that ADTM disrupted the interaction of ERp57 with α IIbβ 3, both in the presence and absence of ADP (Fig. 5c).
ADTM increases vasodilator-stimulated phosphoprotein (VASP) phosphorylation and HO-1 protein level. In platelets, the phosphorylation of VASP is involved in the down-regulation of platelet  Whole blood was treated with DMSO (0.1%) as a control or 100 μ M ADTM for 5 min at 37 °C. After that the platelets were activated using ADP (10 μ M). Platelet lysates were prepared for immunoprecipitation (IP) and probed with anti-ERp57 antibody for immunoblot analysis (IB). The precipitates were also blotted with anti-α IIbβ 3 antibody to ensure that equal amounts of α IIbβ 3 were pulled down under each condition. # P < 0.01 compared with the control group, * P < 0.01 compared with the ADP group. Data are representative of three independent experiments. aggregation accompanied by inhibition of platelet integrin α IIbβ 3 activation 27,28 . As shown in Fig. 6a, pretreatment with ADTM (10 and 100 μ M) stimulated approximately three-fold increase in the phosphorylation of VASP. In addition, we also observed that ADTM increased the protein expression of HO-1, an anti-oxidative stress microsomal enzyme which when up-regulated could inhibit platelet-dependent thrombus formation (Fig. 6b) 29 .

Inhibition of platelet aggregation and thrombus formation by ADTM in vivo.
Various concentrations of ADTM were administered in rats in vivo (10 -20 mg/kg, 5 days, intravenous, IV) and the platelet aggregation activity and the thrombus formation were evaluated. As shown in Fig. 7, ADP-induced platelet aggregation was significantly compromised (> 40% reduction) in rats treated with ADTM (20 mg/kg). Similarly, ADTM also exhibited significant anti-thrombotic effect in vivo as shown in the ferric chloride (FeCl 3 )-induced venous thrombosis (Fig. 8a). The inhibitory effects of ADTM in platelet aggregation and thrombus formation were comparable to clopidogrel. Furthermore, FeCl 3 induced significant decrease in plasma 6-Keto-PGF 1α (an indicator of plasma PGI 2 , an intrinsic inhibitor  of platelet aggregation) and the treatment of ADTM (20 mg/kg, 5 days, IV) abolished the reduction of FeCl 3 -induced 6-Keto-PGF 1α in a concentration-dependent manner (Fig. 8b), giving more evidence of the anti-thrombotic properties of ADTM.

Discussion
Anti-thrombotic therapy plays an essential role in the course of treatment of various cardiovascular disorders (CVDs) because platelet dysfunction and thrombogenesis are common underlying factors in the development and progression in CVDs, particularly atherosclerosis, myocardial infarction and ischemic stroke. Aspirin and clopidogrel are most routinely used anti-platelet agents for the management of CVDs, but these agents irreversibly inhibit platelet activation with potentially devastating unwanted effects such as hemorrhages. In present study we reported that ADTM, a novel cardioprotective compound synthesized in our own laboratory 6 , displayed specific binding for several members of the PDI family. By using a biotin-conjugated ADTM analogue (BAA) as bait and chemical proteomic analysis of the bound protein complexes, we identified eight potential interacting protein targets of ADTM (Supplementary Table S1). With further validation using immunoblot analysis the identity of major targets were confirmed as members of the PDI family, including ERp72, ERp57 and ERp5. In particular, ADTM showed highest potency in inhibiting the redox activity of ERp57 in enzymatic assay in vitro. Furthermore, untagged-ADTM could compete with BAA for the binding of ERp57, providing evidence that ADTM is a competitive ligand at ERp57. We further demonstrated that ADTM displayed concentration-dependent inhibition in ADP-induced platelet aggregation and AA-induced platelet aggregation in vitro. This result is in line with the observation that ERp57 has been recently implicated to play crucial roles in mediating platelet aggregation, homeostasis and thrombosis [18][19][20] . It has also been reported that ERp57 present on platelet surface were activated by staphylococcal extracellular adhesive protein resulting in arteriosclerosis in Staphylococcus aureus infection, and the inhibition of ERp57 could suppressed platelet aggregation 30 .
We further demonstrated that ADTM concentration-dependently inhibited the ADP-induced expression of P-selectin and activation of α IIbβ 3 integrin in vitro. P-selectin is an adhesion molecule expressed on the surface of platelet and is an important marker for platelet activation. α IIbβ 3 integrin is the most abundant adhesion receptor on the surface of circulating platelets and has been reported to  31 . We further demonstrated that ERp57 bound to α IIbβ 3 directly by using co-immunoprecipitation, and ADTM disrupted ERp57/α IIbβ 3 interaction, as shown by the dissociation of the complex (Fig. 5c). In line with this, ERp57 has been reported to directly interact with α IIbβ 3 on platelet surface and regulate its function in platelet aggregation [18][19][20] . Recent study has shown that disulfide bond exchanges in α IIbβ 3 is required for activation and post-ligation signaling during clot retraction 32 , and the modulation of functional disulfide bonds is being explored for blood protein control 33 . Our results suggested that ADTM produced its anti-thrombotic action via the modulation of disulfide bond by interacting with ERp57 (Fig. 9).
The inhibitory action of ADTM on platelet integrin α IIbβ 3 activation was furthered evidenced by the increase of the phosphorylation of VASP after ADTM pretreatment. The phosphorylation VASP is involved in the down-regulation of platelet aggregation accompanied by inhibition of platelet integrin α IIbβ 3 activation 27,28 . We also observed that ADTM increased the expression of HO-1 in platelets. This is coherent with our previous finding that HO-1 activation has a key role in ADTM cardioprotective effect in H9c2 cardiomyoblast cells in vitro and in the myocardium in vivo 7 . HO-1 is a potential therapeutic target in treating thrombotic disease and HO-1 knockout mice demonstrated acute thrombus formation in response to hypoxia 34 . Previous studies reported that dengue virus glycoprotein-1 (DV-NS1) antibody inhibited PDI activity and blocked platelet aggregation accompanied by an increase in HO-1 35,36 . Furthermore, redox reactions have been shown to play crucial roles in platelet aggregation in a glutathione peroxidase-3 (GPx-3)-deficient mouse model. GPx-3 is an important anti-oxidant enzyme and redox sensor in plasma and GPx-3-deficiency impaired normal platelet inhibition and promoted thrombosis due to enhanced reactive oxygen species flux and decreased nitric oxide bioavailability 37 . In our in vivo studies, we observed that ADP-induced platelet aggregation was significantly compromised (> 40% reduction) in rats treated with ADTM (20 mg/kg). The anti-thrombotic effect of ADTM was further evidenced by the reduction of platelet activity and thrombus formation in the ferric chloride (FeCl 3 )-induced venous thrombosis assay in the rat. The major metabolites of ADTM in vivo were 2-hydroxymethy-3, 5, 6-trimethylpyrazin (TMP-OH) and DSS 38 . We have checked the in vivo activity of DSS and TMP. Interestingly, the results showed that ADTM exhibited stronger antiplatelet and anti-thrombotic activities when compared to DSS and TMP, alone or in combination in ADP-induced platelet aggregation and FeCl 3 -ionduced thrombosis model, respectively (Figs. 7 and 8). The precise mechanisms of this action conferred by ADTM would be another interesting question that is worth investigating in the future. Recently there are accumulating reports that the inhibition of protein disulfide isomerases including PDI and ERp57 could block thrombus formation in various in vivo models and suggest that this protein family represents an important novel class of anti-thrombotic target [39][40][41][42][43] . Preclinical studies demonstrate that deficiency in platelet ERp57 resulted in the increased tail bleeding times and delayed thrombus formation 20 , while PDI is unable to compensate for the absence of platelet ERp57 20 and other study also showed that the blockade of ERp57 with specific antibody further inhibited platelet aggregation in PDI deficient platelets 16 . These independent studies further provided evidences that ERp57 and PDI have distinct roles in the mediation of platelet function.
Taken together, our present study reported the anti-thrombotic action of ADTM with both in vitro and in vivo data. ADTM exhibited comparable anti-thrombotic properties as aspirin and clopidogrel. The results suggested that the anti-thrombotic action of ADTM is mediated through disrupting the interaction between ERp57 and α IIbβ 3 possibly by blocking the action of ERp57 on disulfide bonding (Fig. 9). Our data provided a rationale for the further development of ADTM as anti-thrombotic agent targeting the underlying mechanism involving ERp57 and α IIbβ 3. Furthermore, as a competitive ligand at ERp57, ADTM presents as a promising compound for the development of versatile anti-thrombotic agents. Our data also provide insights in novel strategies for the development of drugs targeted ERp57 for anti-thrombotic.

Methods
Materials. ADTM and BAA were synthesized at Jinan University, China. DSS and TMP were of analytical pure grade, and obtained from Xi'an Honson Biotechnology (China) and Shanghai Banghai Chemical Company (China), respectively. Human recombinant ERp57 was obtained from Abcam (Cambridge, UK), PDI and ERp72 were purchased from Enzo Life Sciences (Exeter, UK). Insulin, DTT, AA and ADP were obtained from Sigma Aldrich (St. Louis, MO, USA).
Preparation of platelet-rich plasma. All animal experiments were approved by the Animal Care and Experimentation Committee of Jinan University and were performed in accordance with the approved guidelines. Sprague-Dawley rats were anesthetized with 10% chloral hydrate, and blood was obtained by an arterial puncture. Whole blood was anticoagulated with citrate (3.8%; 1:9, v/v) and centrifuged at 200 g for 8 min at room temperature to obtain platelet-rich plasma (PRP). The residue was centrifuged at 550 g for 5 min to obtain platelet-poor plasma (PPP).
Protein preparation from platelets. Washed platelets were lysed using a dounce homogenizer in NP-40 lysis buffer (Beyotine, China) with 1 mM phenylmethylsulfonyl fluoride (PMSF). Platelet lysates were centrifuged at 12,500 g for 20 min at 4 °C, and the supernatant was collected and stored at − 80 °C until further analysis.
NeutrAvidin Agarose Resin pull-down with BAA. Platelet lysates (3 μ g/μ l) were exposed to NeutrAvidin Agarose Resin (Pierce Biotech., Rockford, IL, USA) for 2 h at 4 °C and then centrifuged for 3 min at 2,500 g. The resin was discarded to remove endogenous biotin. Aliquots of supernatants (300 μ l) were first treated with DMSO or ADTM (600 μ M) for 2 h, and then incubated with BAA (600 μ M) for another 2 h. The supernatant was combined with NeutrAvidin Agarose Resin and allowed to shake overnight at 4 °C as described previously 44 . To eliminate the nonspecific combination with compound or agarose resin before protein analysis, 0.1% DMSO was added to one aliquot of lysate as a negative control. Immunoblot analysis. Proteins eluted from the resin were subjected to SDS-PAGE and then transferred to a PVDF membrane for immunoblot (details in Supplementary Methods).
Insulin reduction assay. PDI activity was measured using an assay that measured the catalytic reduction of insulin, as described previously with minor modifications 39,45 . Briefly, insulin (1 mg/ml) was freshly prepared in assay buffer (100 mM potassium phosphate, 1 mM EDTA (pH 7.4) and 10 μ M dithiothreitol). 50 μ l of insulin was incubated in the presence of 10 μ l of enzyme and varying concentrations of compound, and then 10 μ l of 0.1 mM dithiothreitol was added. The increase in turbidity was monitored at 620 nm using a microplate reader (Perkin-Elmer, Singapore) at 25 °C for 60 min. PDI, ERp57, and ERp72 were all used at 0.04 mg/ml. PDI family enzyme inhibition in the presence of compound was determined by the following formula: enzyme inhibition (%) = Methods. PE conjugated anti-CD62P (P-selectin) antibody and FITC conjugated anti-PAC-1 (activated α IIbβ 3) antibody were used to detect platelet activation markers after exposure to ADP in the presence of DMSO (0.1%) or different concentrations of ADTM. An isotype-matched IgG1 PE/FITC was used as a negative control.

Immunoprecipitation.
To determine ERp57-bound α IIbβ 3 contents, whole platelet lysates were incubated with anti-α IIbβ 3 antibody overnight at 4 °C under agitation. The antigen-antibody complex was pulled down after incubation for 4 h at 4 °C with protein G-agarose (Cell Signaling Technology) as described previously 46 . Immune complexes were boiled for 5 min with 2 × loading buffer followed by immunoblot analysis with antibodies for ERp57 and α IIbβ 3.

Rat platelet aggregation-induced by ADP in vivo.
To study the effects of ADTM on platelet aggregation in vivo, ADTM (5-20 mg/kg) in comparison with DSS (10 mg/kg), TMP (6 mg/kg), DSS (10 mg/kg) + TMP (6 mg/kg) and clopidogrel (18 mg/kg) were administered daily by IV injection for 5 days, respectively (detailed methodology in Supplementary Methods). Platelet aggregation was monitored for 5 min and the maximum aggregation inhibition (%) was calculated.

FeCl 3 -induced inferior vena cava thrombosis in rat.
Rats weighing 180 g to 220 g were anesthetized by intraperitoneal administration of 10% chloral hydrate. One hour post-treatment, the medioventral line in the rats was cut and the inferior vena cava was surgically exposed by blunt dissection. 1 × 1 cm filter paper, soaked in 50% FeCl 3 solution, was placed under and above the exposed inferior vena cava for 15 min. Subsequently, the thrombosed inferior vena cava was carefully dissected, and the weight (mg) of wet thrombus harvested from the inferior vena cava was measured as described previously 47 .
Evaluation of plasma 6-Keto-PGF 1α levels in rat. The concentration of 6-keto-PGF 1α present in the supernatant was measured using a radioimmunoassay kit (New England Nuclear Dupont, Boston, MA) according to the manufacturer's instructions. Blood was collected into plastic tubes containing 10% EDTA-Na 2 . Plasma was prepared by centrifuging blood at 550 g for 10 min. Plasma concentrations of 6-keto-PGF 1α were determined by radioimmunoassay and expressed as pg/ml. Statistical analysis. All assays were performed at least three times. Data are presented as mean ± SD.
The groups were compared using one-way ANOVA followed by Tukey's multiple comparison tests using the statistics module of Graph Pad Prism 5.0. A value of P < 0.05 was considered statistically significant.