A novel single‐chain enzyme complex with chain reaction properties rapidly producing thromboxane A2 and exhibiting powerful anti‐bleeding functions

Abstract Uncontrollable bleeding is still a worldwide killer. In this study, we aimed to investigate a novel approach to exhibit effective haemostatic properties, which could possibly save lives in various bleeding emergencies. According to the structure‐based enzymatic design, we have engineered a novel single‐chain hybrid enzyme complex (SCHEC), COX‐1‐10aa‐TXAS. We linked the C‐terminus of cyclooxygenase‐1 (COX‐1) to the N‐terminus of the thromboxane A2 (TXA2) synthase (TXAS), through a 10‐amino acid residue linker. This recombinant COX‐1‐10aa‐TXAS can effectively pass COX‐1–derived intermediate prostaglandin (PG) H2 (PGH2) to the active site of TXAS, resulting in an effective chain reaction property to produce the haemostatic prostanoid, TXA2, rapidly. Advantageously, COX‐1‐10aa‐TXAS constrains the production of other pro‐bleeding prostanoids, such as prostacyclin (PGI2) and prostaglandin E2 (PGE2), through reducing the common substrate, PGH2 being passed to synthases which produce aforementioned prostanoids. Therefore, based on these multiple properties, this novel COX‐1‐10aa‐TXAS indicated a powerful anti‐bleeding ability, which could be used to treat a variety of bleeding situations and could even be useful for bleeding prone situations, including nonsteroidal anti‐inflammatory drugs (NSAIDs)‐resulted TXA2‐deficient and PGI2‐mediated bleeding disorders. This novel SCHEC has a great potential to be developed into a biological haemostatic agent to treat severe haemorrhage emergencies, which will prevent the complications of blood loss and save lives.

Generally, many bleeding emergencies can be very dangerous, and even life-threatening. For example, arterial haemorrhage, one of the most dangerous bleeding emergencies, is always difficult to control and can result in massive blood loss in a short time. Another example is the application of aspirin, and other nonsteroidal anti-inflammatory drugs (NSAIDs) in surgical operations or medical treatment, which strongly inhibits the COX-1 activity, shutting down the biosynthesis of TXA 2 in platelets, and causing dangerous bleeding situations. 7 Aspirin, especially, can chemically modify COX-1 and irreversibly inhibit the COX-1 activity, which results in permanent damages to the platelet function. Fully rescuing the aspirin-resulted TXA 2 -deficient bleeding may take up to 7-10 days, until the newly produced functional platelets are released from the bone marrow. 8 Therefore, it is essential to develop a method which could be beneficial for saving lives in various bleeding emergencies.
Here, we proposed one possible effective approach to instantly handle a variety of bleeding situations and even be able to overcome aspirin-resulted TXA 2 -deficient bleeding disorder or PGI 2 -mediated bleeding disorder. This novel approach was aimed to isomerize the AA (released in the bleeding site) into more TXA 2 and simultaneously restrict the production of PGI 2 and PGE 2 . A biological reagent with these multiple effects has not been developed yet. One of the major challenges is that the prostaglandin synthases, TXAS, PGIS and PGES, almost have equal affinities to share PGH 2 as their common substrate. 9 Therefore, a change in the distribution of PGH 2 to the particular isozyme is the key to control the metabolism of AA into the specific prostanoid. In recent years, using an enzymatic engineering approach to control the distribution of PGH 2 has been focused by our group to address this issue. [10][11][12][13][14][15][16][17] In our previous studies, we have successfully created a single-chain hybrid enzyme complex (SCHEC), COX-1-10aa-PGIS, through the enzymatic engineering approach, which can force AA to be isomerized into PGI 2 , in order to rescue the deficiency of PGI 2 and to study the vascular protection effects of PGI 2 in cellular and animal models. [10][11][12][13] Another SCHEC, COX-2-10aa-mPGES-1, which can effectively pass PGH 2 to mPGES-1, to convert AA to PGE 2 , has also been created as a model for understanding how PGE 2 is biosynthesized during inflammation. [14][15][16][17][18] In this study, we created a novel SCHEC, linking the C-terminus of COX-1 to the N-terminus of the TXAS, through a 10-residue amino acid linker, to directly guide the metabolism of AA to TXA 2 by effectively passing the COX-1 produced PGH 2 to TXAS. COX-1-10aa-TXAS demonstrated triple properties, including increasing PGH 2 to be metabolized into the anti-bleeding TXA 2 and simultaneously reducing PGH 2 to be isomerized into the pro-bleeding PGI 2 and PGE 2 .
These triple properties could rapidly rescue the cellular deficiency of TXA 2 and rebalance the AA metabolites in platelets to meet the needs to terminate bleeding. Therefore, this novel SCHEC has great potential to be developed into an efficient enzymatic haemostatic agent, which is able to prevent severe complications or even deaths caused by severe blood loss in various bleeding emergencies.

| Engineering of the SCHEC, COX-1-10aa-TXAS cDNA and subcloning
The sequence of COX-1 linked to TXAS through a 10-aa linker (COX-1-10aa-TXAS) was produced by PCR and subcloning procedures using similar methods as previously described. [19][20][21][22] Through the PCR cloning method, the cDNA was successfully subcloned into the pcDNA 3.1(+) vectors between the two BamHl sites containing a cytomegalovirus early promoter. The correct inserted size of cDNA was confirmed by restriction enzyme digestion analysis and DNA sequencing.

| Cell culture and expressing the SCHEC in HEK293 cells
The human embryonic kidney cells 293 (HEK293) were purchased from ATCC. HEK293 cells were cultured in a 100-mm cell culture dish and grown in a 37°C humidified incubator with 5% CO 2 supply.
The medium used for the culture was the high glucose Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 1% antibiotic and antimycotic.

| Expression of the SCHEC in HEK293 cells
Stable expression of COX-1-10aa-TXAS and control enzymes in HEK293 cells was performed using the previously established methods. 23,24 Briefly, the cells were cultured and transfected with a purified plasmid contained the corresponding cDNA using the Lipofectamine 2000 approach following the instructions provided by the manufacturer (Invitrogen). After 48 hours of transfection, G418 (400 μg/mL) was added to the medium for screening HEK293 cell line with stable expression ability for the recombinant proteins.
The whole screening process took 4-5 weeks.

| Immunofluorescence staining
The immunostaining procedures were performed as previously described. 25 Briefly, the cells cultured on cover slides were fixed and then incubated with the primary antibody (10 µg/mL COX-1 or TXAS antibody) in the presence of 0.25% saponin. After 1-hour incubation (room temperature), the unbound primary antibodies were washed away with PBS and then incubated with the secondary antibodies labelled with rhodamine or FITC. The stained slides were examined under the Zeiss Axioplan 2 epifluorescence microscope.

| Examination of the enzymatic activities of SCHEC using HPLC-scintillation analyzer method
The method was followed as previously described. 25 Briefly, the transfected cells were washed three times and then suspended in PBS. Then, [ 14 C]-AA (3 µmol/L) was added and PBS was used to balance the total volume to 100 µL. After 5-minute incubation, the reaction was terminated by the addition of buffer A (50 µL of 0.1% acetic acid containing 35% acetonitrile) and centrifuged at 16 500 g for 5 minutes. The C18 column (4.5 × 250 mm) was used to separate the mixture, using buffer A with a gradient of 35%-100% of acetonitrile for 45 minutes at a 1.0 mL/min flow rate. The metabolites of

| Transgenic mice
The transgenic mice were generated as previously described. 13,26

| Tail-cut arterial bleeding assay
A 0.5 cm tip of mouse tail was cut with a scissor. The arterial bleeding was blotted on a filter paper with a 10 seconds interval. The total bleeding time was calculated as the following formula:

| Molecular modelling for TXAS and Coordination between upstream COX-1 and downstream TXAS, PGIS and mPGES-1 in the bleeding site
To perform structure-based enzyme complex engineering, analysing 3D structures of the enzymes is a key step. So far, the 3D crystal structures for COX-1, 18,27,28 PGIS 29 and mPGES 30 have been solved. However, the crystal structure of TXAS is not available yet. In this study, we created a 3D structure model for human TXAS through the homology modelling approach, using the crystal structure of human PGIS as a template, which has highest similarity and identity with human TXAS (Figures 1 and 2). It shall also be indicated that because of the lack of crystal structure for the N-  F I G U R E 1 Schematic presentation of COX-1 co-ordinated with downstream three enzymes to maintain the balance of haemostasis. 3D crystal structures of human COX-1 (PDB ID: 3N8Z 28 ), PGIS (PDB ID: 3B6H 25 ) and mPGES-1 (PDB ID: 5T37 29 ) used were downloaded from PDB. 3D structural model of human TXAS was created by homology modelling using the crystal structure of PGIS as a template. The co-ordination of the upstream COX-1 and downstream TXAS, PGIS and mPGES-1 on ER membrane was schematically presented. A model demonstrating the three-step catalytic activities: AA converted to PGG 2 , then PGH 2 and finally to biologically active anti-bleeding mediator TXA 2 , and the bleeding mediators PGE 2 and PGI 2 were shown These hydrophobic domains also form an entrance channel for the substrate AA 25,27,28 ( Figure 1). On the other hand, the results of our topological and structural studies performed through immunostaining and homology modelling led to the suggestion that the TXAS was anchored to the cytoplasmic side of the ER towards However, in these vascular cells, the PGH 2 could also be isomerized into the bleeding contributors, PGI 2 and PGE 2 , by PGIS and PGES in the ER environment, respectively ( Figure 1, blue rectangle), thus increasing the production of TXA 2 , while decreasing the production of PGI 2 and PGE 2 should be able to increase the anti-bleeding effect ( Figure 1). This has led to the consideration of engineering an enzyme complex to control onsite AA metabolism in favour of TXA 2 , while disfavouring PGI 2 and PGE 2 .

| Design of a SCHEC, COX-1-10aa-TXAS, performing three-step reactions continually to control cellular AA metabolism into TXA 2 specifically
In considering that the substrate channels of COX-1 and TXAS open on opposite sides of the ER, it is safe to propose that the two channels held a very short distance from each other. In respect to the spatial structure and distribution of the substrate channels of COX-1 and TXAS, a single-chain hybrid enzyme complex (SCHEC), COX-1-10aa-TXAS, which contained the two enzyme domains, was created as shown in and then anchoring to the N-terminus of TXAS. This design was aimed to shorten the travelling pathway of PGH 2 , from COX-1 to TXAS, compared with that being passed to other free PGIS and PGES.

| Creating a novel cDNA of COX-1-10aa-TXAS using PCR approach and subcloning of the cDNA into an expression vector
A cDNA encoded the protein sequence of the SCHEC, COX-1-10aa-TXAS, was prepared by PCR and subcloning approaches. First, the full cDNA of human TXAS was obtained by PCR using the previously F I G U R E 2 Design of a novel SCHEC. COX-1-10aa-TXAS was created by linking the C-terminus of COX-1 to N-terminus of TXAS through a linker, which contains 10-amino acid residues (−10aa). The complete structural model of the COX-1-10aa-TXAS with triple catalytically biological activities to convert AA to PGG 2 , then PGH 2 and a final anti-bleeding TXA 2 within a single polypeptide chain were demonstrated F I G U R E 3 Construction and subcloning of the cDNA of COX-1-10aa-TXAS into a mammalian expression vector, pcDNA3.1(+). The starting plasmid of TXAS and pcDNA3.1(+)-COX-1-10aa-PGIS vector was prepared previously. 13 Full cDNA sequence of human TXAS was prepared by PCR after inactivated the BamHI site of the TXAS plasmid via site-directed mutagenesis. On the other hand, the PGIS cDNA fragment was removed from the cDNA of COX-1-10aa-PGIS, which was inserted in a pcDNA 3.1(+) vector. Next, the isolated TXAS cDNA was linked to the cDNA of the COX-1-10aa and then ligated to form a complete expression plasmid containing COX-1-10aa-TXAS cDNA

| Further identification of the stable expression and subcellular localization of the COX-1-10aa-TXAS in HEK293 cell line using immunocytochemistry approach
To further verify the stable expression of the recombinant COX-1-10aa-TXAS in HEK293 cells, the stable cell line was subjected to be analysed using fluorescent immunocytochemistry. In addition, the high-resolution immunostaining will be able to determine

| Establishing a reliable triple-catalytic enzyme assay for the COX-1-10aa-TXAS by using [ 14 C]-labelled arachidonic acid ([ 14 C]-AA) as the initial substrate
As we mentioned above, directly converting AA into anti-bleeding TXA 2 while avoiding the side metabolites, such as the bleeding contributors, PGI 2 and PGE 2 (Figure 1), was the main characteristic of the novel COX-1-10aa-TXAS based on our expectation. We tried to F I G U R E 7 Determination of the activities of the HEK293 cells stably expressing COX-1-10aa-TXAS and comparison with that of co-expressing individual COX-1 and TXAS. HEK293 cells were transfected with COX-1-10aa-TXAS cDNA (B), or cotransfected with the individual COX-1 cDNA and TXAS cDNA (A). After G418 screening for 2 mo, the cells were stored in the liquid nitrogen. After re-cultured the cells, a serial of tests of the enzymatic activities of the cells were designed for P1, P4 and P8 cells. The cells firstly reaching 100% confluency were marked as P1 cells; 1 week later, after three passages, the cells with 100% confluency were marked as P4 cells; and another week for the P8 cells. The procedures described in Figure 6

| Identification of the biological activities of COX-1-10aa-TXAS to promote platelet aggregation in normal platelets and to rescue the aggregative functions of NSAIDs-treated platelets
First, the COX-1-10aa-TXAS dramatically promoting platelet aggregation was observed in AA-induced platelet aggregation assay using normal platelet-rich plasma (PRP, Figure 8A). In comparison with that of the HEK cells control, the maximal aggregation rate was increased from 38% to 60%, and the ½ time for the maximal aggregation was reduced from 3.2 to 1.8 minutes by the HEK cells expressing COX-1-10aa-TXAS ( Figure 8A). The results are also supported by the observation that the aggregated platelets, in the presence of COX-1-10aa-TXAS, exhibited a much heavier and solid form ( Figure 8B, right) than that of the control (Figure 8B, left).
Administration of NSAIDs, such as aspirin, is one of the major causes of emergency bleeding situations. To test how the engineered SCHEC can rescue the NSAIDs-resulted platelet dysfunction, the platelets were treated with aspirin, and then, the excessive as-   Figure 8D). The results suggested that the COX-1-10aa-TXAS has great potential to be developed into a novel type of bio-enzymatic treatment for the platelet-deficient bleeding.

| Restoring aggregative functions of the expired PRP by COX-1-10aa-TXAS
In general, the aggregative activities of platelets can only last for a couple of weeks under 4°C storage. To test the potential application to extend and/or restore the aggregative functions of platelets by using the SCHEC, the human PRP were tested after 45 days at 4°C storage. The responses of the expired platelets to AA-stimulated platelet aggregation were very weak ( Figure 8E, triangles). However, in the presence of the SCHEC, the platelet aggregation functions were significantly restored ( Figure 8E, circles). This suggested that the SCHEC might be used to extend the reasonable expiration days for functional PRP.

| Identification of anti-bleeding effect of the COX-1-10aa-TXAS in vivo
The tail-cut bleeding assay was used to test the anti-bleeding activity of the COX-1-10aa-TXAS in vivo. The mice were divided into two groups: one group for the treatment, using HEK293 cells stably expressing COX-1-10aa-TXAS, and another group for the control, using HEK293 cells transfected with vector only (HEK control, Figure 9A). An arterial bleeding site was created by cutting the tail arteries (0.5 cm from the end of tail) of the mice ( Figure 9B

| Further identification of the anti-arterial bleeding of SCHEC in vivo by using a transgenic mouse model with a bleeding tendency
Recently, we have successfully created a transgenic mouse model by overexpressing another engineered SCHEC designed by our group, COX-1-10aa-PGIS in vivo. 13 Among these transgenic mice, the overexpressed COX-1-10aa-PGIS could directly convert AA into PGI 2 , which is a bleeding contributor with the effects of antiplatelet aggregation and vasodilation. The transgenic mice were created to prove that the COX-1-10aa-PGIS is an effective enzyme complex to be against thrombotic stroke and ischaemia in vivo. The detailed procedures for creating and characterization of the transgenic mice were described previously. 13 Briefly, a single-chain cDNA of the COX-1-10aa-PGIS was created and injected into the embryo to generate the transgenic mice ( Figure 10A). Based on our previous studies, the transgenic mice demonstrated extended bleeding time as a result of effectively converting endogenous AA to PGI 2 , which inhibits platelet aggregation and promotes vasodilation; furthermore, the production of TXA 2 was dramatically reduced in circulation in these transgenic mice. The arterial bleeding time was extended around twofold in the transgenic mouse model (an average of 28 minutes, Figure 10C, left), compared with that of the wild-type mice (an average of 12 minutes, Figure 9C, left). After applying the microsomes of F I G U R E 9 Examination of the anti-bleeding effect of the expressed COX-1-10aa-TXAs on bleeding site using tail-cut arterial bleeding assay for normal mice in vivo. A, A 0.5 cm tip of mouse tail was cut by scissors to create an arterial bleeding site. The artery and vein displayed on the cutting site were shown in (B). Immediately, the bleeding tail was placed into the solution containing isolated microsomes of the HEK293 cells with (C, right) or without (C, left) expressing COX-1-10aa-TXAS. The blood was blotted on a clean filter paper every 10 s until no detectable bleeding (A). The average and standard deviation of the bleeding time for each group (n = 6) were plotted in (C) HEK293 cells expressing COX-1-10aa-TXAS, the bleeding time of the transgenic mice was reduced to around 3 minutes ( Figure 10C, right).
These results demonstrated that the recombinant COX-1-10aa-TXAS has the ability to overcome the TXA 2 deficiency and rebalance the ratio of the TXA 2 to PGI 2 to stop bleeding in vivo.

| D ISCUSS I ON
The engineered SCHEC, COX-1-10aa-TXAS, is capable of mimicing the triple-catalytic activities of wild-type COX-1 and TXAS in converting the cellular AA to TXA 2 . Thus, this enzyme complex has provided very valuable information to understand the active ER configuration of the wild-type COX-1 and TXAS in the biosynthesis of TXA 2 . It has allowed us to predict that the physical distance between the native COX-1 and TXAS on ER membrane is very similarly close to that of an enzyme complex, which has not been proposed yet. Crystallization of COX-1-10aa-TXAS will be helpful to uncover the detailed structural and functional relationship between COX-1 and TXAS in the PGH 2 movement during the biosynthesis of the key molecule, TXA 2 , which is directly involved in the control of haemostasis as a blood clotting mediator.
There are several medical applications for the SCHEC, COX-1-10aa-TXAS. First, it could be used to treat bleeding emergencies. One of the major advantages is that the engineered SCHEC is able to use endogenous cellular AA as a substrate. As a result of three-step enzymatic reactions that occur continually and instantly to convert the released AA into TXA 2 in the bleeding site, the SCHEC should be able to stop the bleeding on site effectively.
Thus, the COX-10aa-TXAS has great potential to be developed into a biological reagent to treat bleeding in various bleeding emergencies.
Second, the biological functions of TXA 2 in mediating platelet aggregation and vasoconstriction in haemostasis are well characterized.
However, its roles on other cells, such as neuronal and cancer cells, are poorly understood. Recent studies have reported that TXA 2 demonstrates effects on promoting cancer cell proliferation, 21,22 and is also involved in post-stroke-related neuronal cell damages. 33 The success of engineering the active SCHEC, which specifically directed COX-1-produced PGH 2 to be passed to TXAS, was the first to make it possible to control cellular AA metabolism in favour to TXA 2 and disfavouring other prostanoids. Transfection of the cDNA of COX-1-10aa-TXAS to different cells, such as neuronal and cancer cells, could be used as models to uncover the roles of the TXA 2 biosynthesis on the related diseases, such as neurodegeneration, and cancer development and metastasis.
Finally, overproduction of TXA 2 by COX-1 co-ordinating TXAS is one of the major causes of thrombosis and vasoconstriction in ischaemic diseases, such as stroke, heart arrest, pulmonary arterial hypertension and deep vein thrombosis. Specifically, suppressing the production of TXA 2 is an important step to rescue TXA 2 -mediated ischaemia. One of the most common ways to prevent blood clotting is to apply low dose aspirin to reduce the production of TXA 2 .
However, aspirin is a non-selective COX inhibitor, which could also reduce the production of other important prostanoids, such as PGI 2 , which is a very important vascular protector that prevents damages from excess TXA 2 . It becomes more and more clear that aspirin may also promote cardiovascular diseases by inhibiting the generation of PGI 2 . However, the drug specifically inhibiting TXA 2 production and maintaining normal PGI 2 level is not available yet.
With the combination of our previously developed COX-1-10aa-PGIS with this newly engineered COX-1-10aa-TXAS, we could use these complexes as targets for the screening of specific drugs, which only inhibit TXAS without affecting PGIS and COX-1. Thus, this newly created COX-1-10aa-TXAS could have a great impact if being used as a drug target for discovering the next generation of NSAIDs, which have fewer side effects on cardiovascular diseases.
In conclusion, the engineered SCHEC, COX-1-10aa-TXAS, demonstrated the integrated triple-catalytic reactions within just F I G U R E 1 0 Examination of the anti-bleeding effect of the expressed COX-1-10aa-TXAs on bleeding site using tail-cut arterial bleeding assay for transgenic mice with the bleeding tendency in vivo. A, The transgenic mice with bleeding tendency were created by injection of the cDNA of COX-1-10aa-PGIS into the embryos of the mice (13). The arterial bleeding conditions were created by tail-cut approach and then treated by the isolated microsomes of the HEK293 cells with (B, open bar) and without (B, closed bar) expressing COX-1-10aa-TXAS. The averages and standard deviations of the bleeding times for each group (n = 6) were plotted (B) a single polypeptide, which could effectively convert endogenous AA into TXA 2 . These properties make it possible to control cellular AA metabolism to be in favour of TXA 2 and disfavour of other prostanoids, such as PGI 2 and PGE 2 . This controlled AA metabolism activity has exhibited effective anti-bleeding properties.
Furthermore, the SCHEC could be used as a specific target for screening anti-stroke drugs, as well as a cellular model for understanding the relationship between TXA 2 and other diseases, such as cancer and neuronal degeneration diseases. This SCHEC could be further used for understanding the topology, structure and functional relationships between the two enzymes, COX-1 and TXAS, anchored to the ER membrane, during the biosynthesis of TXA 2 (proposed in Figure 1).

ACK N OWLED G M ENTS
The authors would like to thank Mr Imran Siddiqui for assistance with language proofreading. This work was supported by NIH Grants (RO1 HL56712 and HL79389 for KHR) and American Heart Association grants (10GRNT4470042 and 14GRNT20380687 for KHR).

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.