Thrombus-targeting therapy with apyrase-annexin V fusion attenuates thrombosis without bleeding


 Antithrombotic therapy is essential to prevent thrombotic reocclusion in patients with acute myocardial infarction, stroke, or venous thromboembolism. However, current antithrombotic drugs cause bleeding that limits dose and clinical effectiveness. Previously, we reported that APT102 (AZD3366), a human apyrase optimized to scavenge extracellular ADP and ATP, exhibited potent antiplatelet efficacy without bleeding in experimental myocardial infarction and stroke. Here we describe APT402, an optimized fusion of APT102 and annexin V that uniquely inhibits both platelet and coagulation components of thrombus growth. APT402 preferentially bound to injured vessels and thrombus and prevented thrombotic occlusion in arterial and venous thrombosis models, without increasing bleeding. Thus, APT402 breaks the confounding link between antithrombotic efficacy and bleeding, pioneering a safe and effective approach to prevent and treat a broad range of thrombotic diseases, and may be particularly useful in clinical conditions associated with high bleeding risks.

Acute myocardial infarction (AMI) and acute ischemic stroke (AIS) result from lifethreatening, thrombus-driven blockages of the coronary or cerebral arteries in which activated platelets and thrombin both play critical roles (1,2). The goals of treatment are to expedite restoration of blood flow and thereby maximize salvage of ischemic myocardium or brain penumbra (1)(2)(3)(4). Percutaneous coronary intervention (PCI) is the preferred reperfusion approach after AMI, while fibrinolysis with recombinant tissue plasminogen activator (rt-PA) and/or thrombectomy are the approved treatments for AIS (1,4,5,6). Adjunctive antithrombotic treatment with dual antiplatelet therapy (aspirin and a P2Y12 antagonist) plus an anticoagulant (heparin or bivalirudin, Angiomax®) to inhibit thrombin activity is the standard-of-care for PCI patients (7), but none of the currently available antithrombotic drugs can be given during AIS treatment with rt-PA due to the high risk of devastating intracranial hemorrhage (4).
Despite aggressive antithrombotic therapy after AMI, reocclusion from recurrent thrombosis and dose-limiting bleeding occur in a significant number of patients (7,8). Attempts to further improve clinical outcomes have led to the development of more potent platelet P2Y12 inhibitors including prasugrel (Effient®) and ticagrelor (Brilinta®,(8)(9)(10)(11), as well as direct factor Xa (FXa) inhibitors, rivaroxaban (Xarelto®) and apixaban (Eliquis®) (12,13, not approved for PCI ), but these agents also increase bleeding. Currently, net adverse composite endpoints of death, coronary reocclusion, or secondary stroke remain as high as 7-12% for PCI and 10-12% for fibrinolysis, with rates of major bleeding of 5-11% (9)(10)(11). Importantly, most adverse events occur within the first 6-9 hours after intervention. Major bleeding within 48 hours of PCI is associated with a one-year mortality of 7.2% compared to 2.1% in patients who do not have periprocedural major bleeding (14). Due to its narrow time-window of up to 4.5 hours postsymptoms and 6-to 7-fold increased risk of intracranial hemorrhage, only 3-5% of AIS patients receive rt-PA therapy (4,15).
Deep vein thrombosis (DVT) is associated with a high risk of pulmonary embolism that affects up to 900,000 patients annually, with greater than 30% mortality in the US (16). An historical pharmacological standard-of-care for both treatment and prophylaxis of DVT has included initial administration of low-molecular-weight heparin (enoxaparin, Lovenox®) with subsequent transition to long-term oral vitamin K antagonist therapy (warfarin, Coumadin®) (17). Over the past few years, several new direct, selective oral drugs have become available, including dabigatran etexilate (Pradaxa®) that inhibits thrombin; and rivaroxaban (Xarelto®) and apixaban (Eliquis®) that inhibit FXa (13). However, within 11-15 days after knee replacement surgery, net adverse outcomes of bleeding remain as high as 18.9% for enoxaparin and 9.6% for rivaroxaban (12,13,18). Moreover, all of these drugs induce hypocoagulability with major bleeding event rates as high as 10.8% (13,18).
APT102 is a homolog of human apyrase (CD39), the physiologic antiplatelet enzyme on the endothelium that acts to maintain blood fluidity and flow. APT102 has been optimized to inhibit platelet activation and aggregation, and to limit vascular inflammation by scavenging excess extracellular ADP and ATP (19). APT102 does not interfere with platelet function per se; thereby maintaining vascular integrity with no increased bleeding (19)(20)(21)(22)(23). The metabolism of ATP and ADP increases AMP, which is further metabolized to vasodilatory and cardioprotective adenosine by endothelial CD73 (19). These features have translated into the capability of APT102 to attenuate thrombosis and ischemia-reperfusion injury in experimental AMI without increased bleeding (19)(20)(21)(22)(23). Strikingly, APT102 also extended the therapeutic window for rt-PA-mediated reperfusion following experimental stroke while attenuating hemorrhagic transformation (20).
Annexin V is a physiologic anticoagulant protein in the placenta that prevents thrombosis during delivery of the fetus (24). The protein binds to anionic phospholipids, specifically phosphatidylserine (PS) (25). PS moves to the cell surface upon activation of platelets or damage to endothelial cells where it attracts leukocytes and also tethers assembly of prothrombinase and tenase complexes. The competitive binding of annexin V to PS blocks assembly or displaces the procoagulant complexes and thereby attenuates generation of thrombin, the most powerful agonist for platelet activation and catalyst for fibrin formation. Annexin V binding requires a threshold of 2.5-8% PS exposure; thereby allowing generation of minute quantities of thrombin needed for normal homeostasis (26). Thus, annexin V inhibits up to 95% of platelet-associated prothrombinase activity and up to 95% of thrombin generation. Indeed, annexin V was shown to be an effective inhibitor of both venous and arterial thrombosis in vivo without increasing bleeding (27)(28)(29). Moreover, annexin V preferentially binds to thrombi or injured vessels, as compared to non-injured vessels, and thereby targets antithrombotic activity to the vascular injury site (27,29).
Annexin V also provides a potent anti-inflammatory effect by blocking the PS binding site for leukocytes on activated endothelial cell membranes (30). Despite these potential benefits, the clinical utility of annexin V as a drug has been limited by its small size (37kD) resulting in rapid renal clearance with a distribution half-life of 5 min and an elimination half-life of 20 min (27).
We report here the design and characterization of APT402, an optimized fusion protein of APT102 and annexin V, and compare the pharmacologic effects of APT402 to APT102 and to current FDA-approved antithrombotic therapies during induction of arterial and venous thrombosis in animals.

Design, optimization, and production of APT102-annexin V fusion proteins.
We hypothesized that the fusion of APT102 with annexin V would modulate both the platelet and coagulation axes of thrombosis, leading to synergistic antithrombotic efficacy in arterial and venous thrombosis (Fig. 1). In addition, the fusion protein would target sites of platelet activation and/or endothelial cell injury, minimizing systemic bleeding risk. It was also predicted that the fusion protein would prolong the pharmacologic half-life of annexin-V and improve the clinical utility of this physiologic anticoagulant.
The gene encoding human annexin V was fused to the C-terminus of APT102 with linkers of various lengths to optimize expression and function. HEK 293T cells were stably transfected with the linearized expression plasmids and the proteins were purified to homogeneity. The first version of the fusion protein designed with a flexible linker of 20 AA was 50% degraded in the linker region by protease activity present in the culture medium ( fig. S1). Further study showed that low pH (e.g. 5.0) accelerated degradation in the linker sequence compared to pH 7.4. Version 2 was designed with substitution of threonine for glycine at the cleavage site of P1, but was still susceptible to protease degradation when the pH of the clarified supernatant was reduced to 5 with sodium citrate. However, when this mutation was combined with replacement of serine by alanine or glutamine at the cleavage site P1 (as shown in V3 and V4), the degradation was not visible (fig .   S2). Expression levels of V3 were compared with flexible linkers of 5, 10, 15, and 20 amino acid residues or a rigid linker of 9 amino acid residues. The 20AA flexible linker resulted in the highest expression ( fig. S3) and no degraded products were detected. The apyrase-annexin V fusion with the 20 AA and protease-resistant linker (V3) was designated as APT402 and chosen for further validation.
The APT402 gene was subcloned into the GPEx vector and transfected into Chinese Hamster Ovary (CHO-S) cells (31). A purification process was successfully developed, leading to 98% purity. The host cell protein was <25 ng/mg ( fig. S4) and the endotoxin in the formulated bulk was < 0.3 EU/mg, which was suitable for the subsequent studies in rabbits.
APT402 exhibits ex vivo antiplatelet efficacy comparable to APT102 while providing synergistic inhibition of activated platelet-associated thrombin generation compared to APT102 or annexin V.
APT402 was designed to maintain the enzymatic and biological activity of both APT102 and the annexin V moieties. Using a malachite green assay (19) The inhibitory activity on lipopolysaccharide (LPS)-induced factor X activation was assayed using peripheral blood mononuclear cells (PBMC) purified from normal human donors and anesthetized rabbits. There was a similar dose-dependent reduction in factor X activation by 30-50% for both APT402 and human annexin V in rabbit and in human peripheral blood mononuclear cells (PBMCs, fig. S7). As anticipated, APT102 had no effect on LPS-induced procoagulant activity.
The effect on thrombin generation was determined in human platelet-poor and plateletrich plasmas (PPP, PRP) (32). APT102 had a negligible effect on thrombin generation in PPP ( fig. S8A). In contrast, both annexin V and APT402 dose-dependently delayed the time to peak and decreased peak thrombin concentration with comparable potency at equimolar concentrations in PPP (figs. S8, B to D). APT102 slightly delayed time to peak and modestly decreased the peak thrombin concentration in 20 M ADP-activated PRP ( Fig. 2A), while annexin V and APT402 dose-dependently delayed time to peak thrombin concentration and decreased the peak thrombin concentration ( Fig. 2B and 2C). When the reagents were added in equimolar concentrations, APT102 plus annexin V prolonged time to peak thrombin concentration by 6-fold and inhibited the peak thrombin concentration by 60% compared to no effect and 23% for APT102 and 2.7fold and 25% for annexin V, respectively (Fig. 2D). APT402 accurately mimicked the thrombin generation inhibition of APT102 in combination with annexin V.

Pharmacokinetic and pharmacodynamic profiles of APT402 in vivo.
APT402 was injected IV as a single bolus (0.4 mg/kg) in anesthetized rabbits. Pharmacokinetic modeling showed best fit to a biphasic exponential curve for ADPase activity ( fig. S9). The maximal activity was detected in the plasma 30 min after administration. The distribution phase and elimination half-life (t½) of APT402 were 30 min and 6 h, respectively, which were significantly extended by 6-and 18-fold compared to the distribution half-life of 5 min and elimination half-life of 20 min for annexin V (27). Preliminary studies indicated that administration of APT402 displayed a fast onset of action, inhibiting 95% of 20 µM ADP-induced ex vivo platelet aggregation by 10 min and returning to baseline 60 min after single bolus IV dosing. These data suggested that to maximize the therapeutic potential of APT402, it should be administered as a bolus followed by continuous IV infusion to ensure consistent attenuation of thrombosis.
To investigate the pharmacokinetic and pharmacodynamics of APT402, a single IV bolus (0.2 mg/kg) followed by IV infusion at 12 or 24 µg/kg/min for 120 min was administered to anesthetized rabbits. ELISA data showed that APT402 was detectable in plasma by 15 min and reached a peak and plateau level 30 min after administration. The steady-state concentration for 24 µg/kg/min ranged between 1.6-1.7 mg/L, and was approximately double the range of 0.6-0.7 mg/L observed during the 12 µg/kg/min dose. Protein concentrations were negligible at 180 min (60 min after discontinuation of the infusion, Fig. 3A). The ex vivo ADPase activity, inhibition of ADP-induced platelet aggregation, and inhibition of thrombin generation were consistent with the plasma concentrations of APT402 during infusion, then returned to baseline levels within 60 min after discontinuation of the infusion (Fig. 3, B to D). When the steady-state level was reached with administration of APT402 at 24 µg/kg/min, the time to peak thrombin concentration in platelet-poor plasma was prolonged by 2-fold and the peak thrombin concentration was decreased by 77-82% compared to the baseline (Fig. 3D). These data show that the optimal inhibitory level of APT402 was achieved within 30 min of the start of infusion and could be achieved by pretreatment to ensure attenuation of both local and systemic thrombosis as blood flow in the affected artery is restored.

APT402 is preferentially targeted to sites of arterial injury and thrombosis.
Electrolytic injury-induced thrombosis was initiated in a carotid artery of anesthetized rabbits 30 min after a bolus injection (0.2 mg/kg) and the start of an infusion of APT402 at 12 or 24 µg/kg/min for 120 min (33), with some receiving APT402 containing tracer quantities of near infrared (NIR)-labeled APT402 as described (34). Other rabbits given a bolus of APT102 (1.0 mg/kg) received a tracer quantity of NIR-labeled APT102 as a control (fig. S10). Fluorescence intensity at the injury site was not different from the baseline in the NIR-APT102-treated animals ( Fig. 4A). In contrast, fluorescence was approximately 6-7-fold higher than baseline in the NIR-APT402-treated animals (Fig. 4B). Fluorescence signal did not increase in the non-injured control artery. Similar to previous studies examining administration of annexin V in rabbits (27, 28), these data confirm that APT402 was preferentially targeted to the site of arterial injury and thrombosis ( Fig. 4C).
APT402 maintains patency and more effectively attenuates arterial thrombosis in rabbits without increasing bleeding risk compared to standard-of-care agents The rabbit electrolytic injury model of arterial thrombosis was used to define the dose-response of APT402 and to compare the antithrombotic efficacy and bleeding risk of APT402 to APT102 and to current FDA-approved antiplatelet (clopidogrel, ticagrelor) and anticoagulant (enoxaparin, bivalirudin) agents administerd alone or in combination (ticagrelor + bivalirudin) (33). In addition, hydroxypropyl methylcellulose (HPMC), the chemical agent needed for efficient ticagrelor absorption from the gut, and saline vehicle for other agents were included as separate control groups. Since strong antithrombotic effects were observed in the preliminary studies of APT402 with 0.2 mg/kg single IV bolus followed by 12 µg/kg/min, we examined the antithrombotic effect of this regimen as well as higher (24 µg/kg/min) and lower (4 µg/kg/min) infusion rates with the same initial bolus dose (0.2 mg/kg). Rabbits were randomized into eleven groups with the treatments initiated 30 min before electrolytic injury and continued for 120 min after the initiation of the electrolytic injury (n=10 and 6/group for the placebo control and treatment groups, respectively).

APT402 attenuates venous thrombosis in mice without increasing bleeding or inducing hypocoagulability.
Healthy mice received two days of treatment with enoxaparin at 6 mg/kg SC, a dose proven to demonstrate antithrombotic efficacy in mouse models of thrombosis (33,34), or with APT402 at 0.2 mg/kg IP, followed by an SC infusion at 1.5 g/kg/min. Enoxaparin significantly prolonged BT and aPTT (fig. S16, tables S10 and S11). In contrast, APT402 had no effect on BT or aPTT.
Neither enoxaparin nor APT402 treatment affected whole blood cell counts compared to the placebo controls (table S12). An electrolytic injury model of venous thrombosis was used to assess antithrombotic efficacy of APT402 in mice (35,36). Thrombus weight, BT, aPTT, and Thrombin Clotting Time (TCT) were measured 48h after thrombus induction in the inferior vena cava. Venous thrombosis was consistently generated in all placebo-treated mice, with the mean thrombus weight of 26.0±1.3 mg ( Fig. 6A and table S13). Treatment with enoxaparin significantly reduced thrombus weight by 63% compared to controls, but was associated with a significant prolongation of BT by 5-fold, aPTT by 2.3-fold, and thrombin clotting time (TCT) by 2-fold ( Fig. 6, B-D, tables S14-16). In contrast, APT402 reduced thrombus weight by 44% and 65% at the doses of 0.5 and 1.5 g/kg/min, respectively, without significantly affecting BT or TCT and only modestly increasing aPTT at the higher dose, compared to controls (Fig. 6, B-D, tables S14-16).

Discussion.
Coronary artery disease is the leading cause of death worldwide with 3.8 million men and 3.4 million women dying of the disease each year (37). Furthermore, the number of chronically ill cardiovascular patients is increasing due to aging of the population. Every year, 15 million people suffer AIS, which remains the leading cause of long-term disability and the second leading cause of death worldwide. Likewise, venous thromboembolism (VTE) is the third most common cardiovascular disease in the world, with an estimated 150,000 deaths and 900,000 hospitalizations per year in the US and Europe (38). Incomplete reperfusion caused by thrombotic reocclusion at the site of the original vessel occlusion or by distal embolization represent the major causes of morbidity and mortality in these diseases. Recently, thrombotic and bleeding diatheses have also been identified in a significant number of hospitalized COVID-2019 patients who exhibit platelet hyperactivation, coagulopathy, and bleeding complications (39). Although there have been improvements in antithrombotic therapy in the past few years, bleeding complications remain an inevitable side effect of current antiplatelet and anticoagulant agents due to their intrinsic mechanisms of action, highlighting the importance of developing safer and more effective therapeutic alternatives (40).
In many clinical settings, antithrombotic therapy consists of a combination of antiplatelet and anticoagulant agents (41). In these patients, the individual agents carry separate bleeding risks that are accentuated by the combination such that the physician must balance the thrombotic and bleeding risks of each patient individually. Even more troubling is the challenge of medically managing these patients to reduce periprocedural risks of thrombosis and bleeding when they require urgent or elective surgery (42). The optimal agent for these indications would be a combined antiplatelet and anticoagulant inhibitor that does not increase bleeding risk so it can be given safely and provide the required antithrombotic efficacy in a variety of clinical situations.
Harnessing the potential of naturally occuring antithrombotic proteins, APT402 combines the antiplatelet properties of an optimized human apyrase, APT102, with the anticoagulant activity of annexin V to produce a potent antithrombotic agent with minimal bleeding risk. In addition, the therapeutic index may be further increased by the thrombus-targeting property of the fusion protein. In this study, APT402 inhibited thrombosis in a rabbit model of arterial thrombosis and in a murine model of venous thrombosis at doses that did not increase bleeding time or markers of systemic hypocoagulability. As expected, inhibition of thrombosis with the FDA-approved comparators used in these experiments was accompanied by increased bleeding time and markers of systemic hypocoagulability. These data indicate that APT402 represents a new paradigm for treatment or prevention of thrombotic diseases, potentially breaking the confounding link between antithrombotic potency and dose-limiting bleeding side effects seen with conventional therapy.
The safe and effective administration of APT402 will not require diagnostic testing of bleeding risk or assessment of the patient's state of coagulation, which will simplify treatment for patients with acute thrombotic occlusion or risk. Moreover, APT402 may be uniquely positioned as a lifesaving therapy for patients undergoing surgery or stroke treatment with rt-PA, for which no antithrombotic drugs have been indicated due to associated bleeding risks. Competing interests: R. C. and S.J. have equity interest in APT Therapeutics Inc, which held patents, patent applications, and commercial rights to APT102 and APT402. The other authors declare that they have no competing interests.

List of Supplementary Materials:
figs. S1 to S16; tables S1 to S16

Study Design
Predefined study components: Previous data indicated that six (6) animals were required in each group for the rabbit model of arterial thrombosis in order to detect a 50% difference in thrombus weight with 80% power and type I error probability of 0.05. Ten (10) animals per group resulted in 80% power to detect 30% differences in the mouse acute venous thrombosis model.
Rationale and design of study: The overall objective of the study was to determine whether APT402 more effectively attenuated injury-induced arterial thrombosis in rabbits and venous thrombosis in mice with minimal risk of bleeding compared with current antiplatelet agents and anticoagulants. The primary efficacy endpoint was thrombus weight.
Replication: All groups were sufficiently powered to meet the desired objectives. Assays of platelet function, thrombin generation, and blood analytes were done in duplicate or triplicate as indicated.

Sequence analysis and computer graphics
GenBank and Swiss-Prot databases were searched for amino acid sequence similarities using the BLAST and PSI-BLAST programs. X-ray crystallographic structures were visualized with Swiss PDB Viewer.

Gene cloning
The DNA sequences of APT102, linker, and human annexin V were synthesized and cloned into expression plasmid pSEQTAG2a (Invitrogen, Waltham, MA), a vector suitable for production of secreted recombinant protein in HEK 293T cells. To facilitate cloning, pSEQTAG2a was modified by site-directed mutagenesis (Quick Change, Stratagene, LaJolla, CA) to introduce an Srf I restriction site in frame with the Ig leader sequence. The Sma I fragment of the genes was then translationally fused to the Srf I site. The GPEX vector (Catalent, Madison, WI) was used for stable CHO cell expression, as described previously (31).

Site-Directed Mutagenesis
Point mutations were introduced using a mutagenesis kit (Stratagene) according to the manufacturer's instructions. The entire open reading frames of the mutants generated were verified by DNA sequence analysis.

Protein expression and purification
Expression plasmids of APT102-annexin V fusion proteins were transfected into 293T

Apyrase assays and kinetic analysis
Spectrophotometric assays for ADPase and ATPase activities of apyrases were performed using Malachite Green as described (43). Briefly, enzymatic reactions were conducted in 50 mM Tris-HCl (pH7.4)/8mM CaCl 2 at 37°C. The enzymatic reaction solution (50 µl) was mixed with 950 µl of malachite working solution at 25°C for 30 min. Inorganic phosphate release resulted in an increase of absorbance at 630 nm, monitored using an Agilent 8453 UV-Visible spectrophotometer (Agilent, Palo Alto, CA). Estimates of kinetic parameters were determined spectrophotometrically with equimolar concentrations of APT102 at 10 ng/ml and APT402 at 18ng/ml using unweighted nonlinear least-squares Newton-Raphson regressions to the Michaelis-Menten model.

In vitro platelet aggregation assays
Blood was collected via 23 ga. butterfly needle from a peripheral vein of healthy animals and consenting human volunteers who had not ingested aspirin for at least one week.
Anticoagulation was achieved with fresh acid citrate-dextrose (38 mM citric acid, 75 mM sodium citrate, 135 mM glucose, vol/ml of blood) or heparin (100 U/ml). Platelet-rich plasma was obtained by centrifuging the blood at low speed (160 x g) for 10 min at room temperature. After the platelet-rich plasma was pipetted to a second tube, the remaining blood was centrifuged at high-speed (1,500 x g) for 15 min at room temperature to yield platelet-poor plasma. An aliquot of platelet-rich plasma (containing 1.0-1.2 x 10 8 platelets) was preincubated for 3 min at 37°C in an aggregometer cuvette (light transmission aggregometer, Chrono-log, Haverton, PA) with added APT102, APT402, or isotonic Tris-buffered saline (Sigma, St. Louis, MO). Light absorption of plasma was controlled for by adding platelet-poor plasma and the total volume was adjusted to 300 µl with Tris-buffered saline. After a 3 min preincubation, ADP was added and the aggregation response was recorded for 4-6 min. Fisher Scientific, Waltham). Thrombograms (nM thrombin vs. time) were generated using the Thrombinoscope software, as were key thrombin generation parameters: lag time to onset of thrombin generation (min), peak thrombin concentration (nM), time to peak thrombin (min), and endogenous thrombin potential (ETP; integrated area under the thrombogram curve). Parameters were normalized to control samples.

In vivo Methods
All procedures involving animals were approved by the Institutional Animal Care and Use Committees at Washington University or the University of Michigan.

Pharmacokinetics and pharmacodynamics in rabbits
The time course of ATP402 protein levels in plasma was determined by intravenous administration to rabbits (n=2-3). APT402 was given as a single IV bolus at 0.4 mg/kg or a single IV bolus at 0.2 mg/kg followed by IV infusion at 12 or 24 µg/kg/min for 2h. Blood samples were collected serially in heparinized tubes. Pharmacodynamics were assessed ex vivo by measurement of ADPase activity, ADP-induced platelet aggregation in platelet-rich plasma, and thrombin generation in platelet-poor plasma.

Rabbit model of arterial thrombosis
Electrolytic injury was induced in the carotid artery of rabbits as described previously (33). Electrolytic injury to the carotid was induced at t=0 by application of 250 µA of anodal current to the indwelling needle electrode with use of a 9V battery for 2h (36). Current was increased to 300 μA after 30 min and to 350 μA after 1h. Blood flow was monitored continuously and recorded at the time of blood sampling. Blood samples were obtained for assays and bleeding times were measured at baseline, 30 min, 1h and 2h after the onset of current. After 2h, the injured carotid artery was excised and the thrombus carefully removed and weighed.

In vivo imaging of fluorescent dye-labeled thrombi by fluorescence molecular tomography
A near infra-red (NIR) fluorescent dye (LS288, Ex/Em 773/793) in methanol was conjugated (~1:1) to the functional amines of APT402 (and APT102 as a control) and purified. A custom built, fiber-based, portable, video-rate system was used for in vivo imaging as described (34). Briefly, a flexible imaging pad (3 cm x 3 cm) containing 12 sources 680 nm 20 Hz, and 710 nm 17 Hz laser diodes for excitation and absorption, respectively, was taped perpendicular to the skin incision over the carotid arteries to provide dynamic concurrent acquisition of tomographic data from both the injured and contralateral non-injured carotid artery used as a reference. After a 30 min stabilization period, the NIR-labeled APT402 (or NIR labeled-APT102) together with a bolus dose of non-labeled APT402 (or APT102) was given IV and an IV infusion of unlabeled APT402 was started. Anodal current was applied to the carotid needle electrode to initiate thrombosis 30 min after the bolus of APT102 or APT102. Carotid blood flow was monitored continuously with the electromagnetic probe. Accumulation of the labeled drug was monitored serially with the FMT probe. After 2 h, the carotid thrombus was collected, weighed and frozen in OCT media to observe the bound NIR-labeled drug in 10 micron sections cut on a cryostat and mounted on glass slides.
Epifluorescence was imaged using a Cy7 filter with excitation at 710/75 and emission at 810/90 using a dichroic 750LP film and exposure time of 20s.

Murine model of venous thrombosis
In the mouse venous thrombosis model, endothelial damage promoted a thrombogenic environment and subsequent, predictable thrombus formation, as described previously (35,36).

Statistical analysis
All data are presented as the mean ± SEM or SD as indicated. Gaussian distribution was assessed by the D'Agostino-Pearson normality test. Two-group comparisons were performed with Student's two-tailed, t test. One-way ANOVA with Tukey's post hoc test was used to perform multiple group comparisons. Serial data were analyzed using a General Linear Model for Repeated Measures ANOVA (SigmaStat v.3.11 Systat Software Inc,). Statistical differences with two-tailed probability values of P<0.05 were considered significant. All data were analyzed using Excel (Microsoft).