Novel Thrombolytic Drug Based on Thrombin Cleavable Microplasminogen Coupled to a Single‐Chain Antibody Specific for Activated GPIIb/IIIa

Background Thrombolytic therapy for acute thrombosis is limited by life‐threatening side effects such as major bleeding and neurotoxicity. New treatment options with enhanced fibrinolytic potential are therefore required. Here, we report the development of a new thrombolytic molecule that exploits key features of thrombosis. We designed a recombinant microplasminogen modified to be activated by the prothrombotic serine‐protease thrombin (HtPlg), fused to an activation‐specific anti–glycoprotein IIb/IIIa single‐chain antibody (SCE5), thereby hijacking the coagulation system to initiate thrombolysis. Methods and Results The resulting fusion protein named SCE5‐HtPlg shows in vitro targeting towards the highly abundant activated form of the fibrinogen receptor glycoprotein IIb/IIIa expressed on activated human platelets. Following thrombin formation, SCE5‐HtPlg is activated to contain active microplasmin. We evaluate the effectiveness of our targeted thrombolytic construct in two models of thromboembolic disease. Administration of SCE5‐HtPlg (4 μg/g body weight) resulted in effective thrombolysis 20 minutes after injection in a ferric chloride–induced model of mesenteric thrombosis (48±3% versus 92±5% for saline control, P<0.01) and also reduced emboli formation in a model of pulmonary embolism (P<0.01 versus saline). Furthermore, at these effective therapeutic doses, the SCE5‐HtPlg did not prolong bleeding time compared with saline (P=0.99). Conclusions Our novel fusion molecule is a potent and effective treatment for thrombosis that enables in vivo thrombolysis without bleeding time prolongation. The activation of this construct by thrombin generated within the clot itself rather than by a plasminogen activator, which needs to be delivered systemically, provides a novel targeted approach to improve thrombolysis.


Generation, expression and purification of single-chain antibodies fused with human thrombin activatable plasminogen
The DNA sequence coding for the human thrombin activatable microplasminogen (HtPlg) was designed from the sequence of human microplasminogen 1 in which the sequence CCT GGA AGG GTT GTA GGG GGG (nucleotides 49 to 69) has been replaced by this sequence ACC ACC AAA ATT AAA CCG CGT ATT GTT GGT GGT, and obtained from GeneArt TM (ThermoFisher Scientific, US). The HtPlg construct was then fused with two different single-chain antibodies, the activated GPIIb/IIIa-targeted (SCE5) and non-targeted (Mut-scFv) previously described 2 , subcloned into the pSecTag vector system. After amplification by polymerase chain reaction (PCR), DNA fragments were digested using restriction enzymes NotI and XhoI (NEB, US) then ligated together using T4 ligase (New England BioLabs, UK) at 16°C overnight. The resulting plasmid constructs were then transformed into BL21 Star E.coli cells (Invitrogen, US). The DNA amplified by PCR and restriction digests was analysed by electrophoresis on a 0.8% agarose gel.
The fusion constructs SCE5-HtPlg and Mut-scFv-HtPlg were transfected with polyethylenimine (PEI, Polyscience Inc., Germany) for expression in human embryonic kidney cells (freeStyleHEK 293-Fcells, Life Technologies, US) suspension cells grown in a CO 2 incubator at 37ºC, shaking at 110 rpm. DNA plasmid for transfection was diluted with PEI and added to culture suspensions following the ratio of 1µg DNA:3µg PEI:1mL culture. H293F cells were adjusted at 2x10 6 cells/mL with Freestyle 293 expression medium (Invitrogen, US) with a viability greater than 95 %.
The culture was supplemented with 5g/L Lupin after 1 and 5 days. At day 3, 5 and 7 2 after transfection, the culture was supplemented with 2mM glutamine. The glucose level was maintained at a final concentration of 6g/L. The cells were harvested when viability was 50 %. The suspensions were centrifuged at 14,000g for 15 minutes and the supernatants were collected for protein purification.
Both SCE5-HtPlg and Mut-scFv-HtPlg proteins carry a 6x His-tag at the C-terminal end of their amino acid sequence for purification and a V5-tag from their single-chain antibody part (SCE5 and Mut-scFv). Proteins were purified by fast liquid protein chromatography with a nickel-based metal affinity column Ni-NTA (Invitrogen, US).
Protein concentration was determined with Direct Detect Infrared Spectrometer (Merck Millipore, US). Purity of the proteins was analysed by SDS-PAGE gel stained with Coomassie Brilliant Blue visualized with Odyssey imaging system (LI-COR Biosciences, US) in the 700 channel. Western blot analysis was performed to confirm the presence of the SCE5-HtPlg and Mut-scFv-HtPlg fusion proteins by revealing the 6xHis-tag and the V5-tag. After SDS-gel electrophoresis, the proteins were transferred on PVDF membranes which were blocked with 5% skimmed milk at 4°C overnight then incubated 1 hour with Anti-6xHis-tag antibody HRP (horse radish peroxidase) or anti-V5-tag antibody HRP. Secondary hybridization was performed with SuperSignal West Pico chemiluminescent (ECL) substrate (Thermo Scientific Inc, US) for the HRP enzyme. ECL signal on membranes were visualized using a BioRad Gel-Doc system.

Cleavage of the thrombin activatable microplasminogen proteins into microplasmin
The cleavage of SCE5-HtPlg and Mut-scFv-HtPlg from thrombin incubation into microplasmin was studied in vitro with western blot analysis and spectrophotometry. A zero (t))/(A total (t)-A zero (t)). Replicates were obtained with thrombi made from the blood of 4 different donors. Mean percentage of degradations ± SEM are plotted over time.

Flow cytometry
Blood from 5 healthy adult volunteers was collected in sodium citrate 3.8 % (w/v).
Platelets-rich plasma (PRP) was obtained by centrifugation at 180g for 10 min and

All experiments involving animals were approved by the Alfred Medical Research and Education Precinct Animal Ethics Committee (E/1534/2015/B and E/1589/2015/B).
Six weeks old male C57BL/6 mice were anesthetized with ketamine (50 mg/kg; Parnell Laboratories, Australia) and xylazine (10 mg/kg, Troy Laboratories, Australia) and placed on a 37 °C heater mat to prevent hypothermia. Mouse bleeding time was measured by tail template method. Several group of drug were injected intravenously; Urokinase at 100 and 500 Unit per gram body weight (/g BW), SCE5-HtPlg at 2, 4, 8 µg/g BW, Mut-scFv-HtPlg at 2, 4, 8 µg/g BW and saline (n=3). 30 seconds after the drug administration, a longitudinal incision, 2 mm deep, 4 mm long, was made, starting 10 mm from the beginning of the tail. Care was taken to ensure incision was made over the superficial tail vein running along the left axis of the tail.
Bleeding time was recorded between the section and the arrest of bleeding. Results was expressed as mean values ± SEM (n=3).

Hemoglobin, albumin and plasma fibrinogen measurements
Six weeks old male C57BL/6 mice were injected IV with Urokinase at 500 U/g BW, SCE5-HtPlg at 4 µg/g BW and saline (n=3). 24h after drug administration, mice were anesthetized and 600 µL of blood was collected in 3.2% citrate and centrifuged 15 min at 2,000g to isolate plasma. The concentration of fibrinogen in plasma was determined with a mouse fibrinogen antigen ELISA kit (Molecular Innovations, US).
Mice were then gently perfused with 30 mL of saline then brain and intestine were harvested. Similar parts of each tissue were isolated, weighted and lysed in Triton X-100 solution (1% v/v in PBS). An additional group of 3 mice treated with saline but not perfused were used as a positive control. Hemoglobin and albumin levels were measured in brain and intestine lysates by spectrophotometry using a hemoglobin substrate (Quantichrom Hemoglobin, Bioassay Systems, US) and by a mouse Albumin ELISA test (Bethyl Laboratories, US) respectively, and were expressed per gram of protein in the lysate measured by bicinchoninic acid assay.

Cell permeability assay
This experiment was adapted from a previously described cell permeability assay which mimics in vitro blood-brain barrier function 3 . Primary human brain microvascular endothelial cells (hBEC; line ACBRI 376, Cell-System Corporation) were seeded in collagen-I-coated Transwell inserts (6.5mm, polyester membrane with 0.4um pores; Corning) at 20,000 cells per insert. Cells were grown to confluence over 3 days in MV2 endothelial cell medium (PromoCell) with 50ug/ml gentamicin.
Following one wash in serum-free medium, cells were stimulated in the luminal compartment with SCE5-HtPlg alone (100nM), bovine thrombin alone (2.5 U/ml; plasminogen-free; Merck) or their combination. 6 h post stimulation, permeability changes were assessed by measurement of fluorescein isothiocyanate-conjugated bovine serum albumin passage from the luminal to the abliminal compartment, as previously described 3,4 . Results were expressed as fold induction from untreated inserts.

Ferric chloride induced thrombosis on mesenteric vessel
Targeting and thrombolytic capacities of the thrombin activatable microplasminogen fusion proteins were tested on a mouse model of thrombosis induced by Ferric chloride superfusion on mesenteric vessel performed as described previously 5 . 8 Briefly, six weeks old male C57BL/6 mice were anesthetized with ketamine (50 mg/kg) and xylazine (10 mg/kg) and placed on a 37 °C heater mat to prevent hypothermia. The mesentery was exteriorised through a midline abdominal incision. Rhodamine 6G (30 µL,0.3% w/v,Sigma) was injected IV to label leukocytes and platelets. A filter paper (1mm x 2mm) saturated with 6% ferric chloride was placed on an isolated mesenteric vessel for 2 minutes to induce vessel wall injury and subsequent thrombus formation. Real time formation of the thrombosis was monitored by intravital microscopy on an Olympus IX81 inverted microscope in the TRITC fluorescent channel to visualise the thrombus stained with Rhodamine 6G and in the Differential Interference Contrast (DIC) channel to identify the vessel wall.
When the thrombus reached over 50 % occlusion, 4 groups of drug were injected intravenously; SCE5-HtPlg at 4 µg/g BW, Mut-scFv-HtPlg at 4 µg/g BW, SCE5 at 1.7 µg/g BW (equimolar dose) and saline (n=3). Snapshots were taken in DIC and TRITC channel every 2.5 minutes from 0 to 20 min post injection then every five minutes up to 1 hour post injection. To avoid any photo bleaching of the fluorescently stained thrombus, exposure to light was fully prevented in between each snapshot.
Thrombus size was measured at each time point (Thr t ) from TRITC channel images converted to binary images with ImageJ software (NIH, US). For each experiment, the biggest size measured for the thrombus was identify (Thr max ). The relative thrombus size was obtained from the formula 100*Thr t /Thr max and the mean values ± SEM are plotted over the time post-injection (n=3).
An additional experiment is obtained with SCE5-HtPlg (4 µg/g BW) pre labelled with anti-6x His tag AF488 antibody (Penta His Alexa-488, Qiagen). The accumulation of the SCE5-HtPlg at the site of the thrombus could then be visualized in the FITC fluorescent channel of the intravital microscope. Snapshots were taken in DIC, FITC 9 and TRITC channel every 2.5 minutes from 0 to 20 min post injection then every five minutes up to 1 hour post injection.

Lung embolism model
Emboli were induced and fluorescently stained by IV injection (5 µL/g BW) of a mixture of Innovin (5% (v/v) from reconstitution prepared according to manufacturer's instruction) and fibrinogen (10 µg/mL, Sigma) pre-labelled with Cy7-NHS dye (Lumiprobe) at 1:15 molar ratio. This model is similar to thromboplastin induced lung embolism, widely described in the haematology literature 6,7 and more recently combined with co-injection of fibrinogen labelled with a near-infrared fluorophore to enable quantification of fibrin deposition by fluorescent analysis of the whole lung 8 .
10 minutes after the induction of the prothrombotic mixture, 4 groups of drug were injected intravenously; urokinase at 500 U/g BW, SCE5-HtPlg at 4 µg/g BW, Mut-scFv-HtPlg at 4 µg/g BW and saline (n=3). Mice were killed 50 minutes after the treatment administration and perfused with saline. Lungs, heart, liver, kidney, spleen are harvested and scanned with the Odyssey imaging system in the 700 channel to visualise the organs and in the 800 channel to measure the fluorescence emitted from the near infrared stained emboli. For each animal, signal within the lung was rationalized to signal within the kidney. Results were presented as mean values of fluorescence signal ratio ± SEM (n=3).

Statistical analysis
All results are expressed as mean values ± SEM. Statistical analysis was performed with GraphPad Prism V6 (GraphPad Software, San Diego, CA, USA). Multiple groups (Flow cytometry, tail bleeding, fibrinogen level in plasma, hemoglobin and albumin levels in brain and intestine and thrombus degradation in both in vivo models) were compared with one-way ANOVA and Tukey post-tests. Parameters from in vitro fibrinolysis assay of SCE5-HtPlg and Mut-scFv-HtPlg groups were compared with unpaired t tests. A difference of p<0.05 was considered significant. Figures 1, 2, 5b and S1 are representative observations with no statistical analysis.