Alboserpin, a Factor Xa Inhibitor from the Mosquito Vector of Yellow Fever, Binds Heparin and Membrane Phospholipids and Exhibits Antithrombotic Activity*

The molecular mechanism of factor Xa (FXa) inhibition by Alboserpin, the major salivary gland anticoagulant from the mosquito and yellow fever vector Aedes albopictus, has been characterized. cDNA of Alboserpin predicts a 45-kDa protein that belongs to the serpin family of protease inhibitors. Recombinant Alboserpin displays stoichiometric, competitive, reversible and tight binding to FXa (picomolar range). Binding is highly specific and is not detectable for FX, catalytic site-blocked FXa, thrombin, and 12 other enzymes. Alboserpin displays high affinity binding to heparin (KD ∼ 20 nm), but no change in FXa inhibition was observed in the presence of the cofactor, implying that bridging mechanisms did not take place. Notably, Alboserpin was also found to interact with phosphatidylcholine and phosphatidylethanolamine but not with phosphatidylserine. Further, annexin V (in the absence of Ca2+) or heparin outcompetes Alboserpin for binding to phospholipid vesicles, suggesting a common binding site. Consistent with its activity, Alboserpin blocks prothrombinase activity and increases both prothrombin time and activated partial thromboplastin time in vitro or ex vivo. Furthermore, Alboserpin prevents thrombus formation provoked by ferric chloride injury of the carotid artery and increases bleeding in a dose-dependent manner. Alboserpin emerges as an atypical serpin that targets FXa and displays unique phospholipid specificity. It conceivably uses heparin and phosphatidylcholine/phosphatidylethanolamine as anchors to increase protein localization and effective concentration at sites of injury, cell activation, or inflammation.

The molecular mechanism of factor Xa (FXa) inhibition by Alboserpin, the major salivary gland anticoagulant from the mosquito and yellow fever vector Aedes albopictus, has been characterized. cDNA of Alboserpin predicts a 45-kDa protein that belongs to the serpin family of protease inhibitors. Recombinant Alboserpin displays stoichiometric, competitive, reversible and tight binding to FXa (picomolar range). Binding is highly specific and is not detectable for FX, catalytic siteblocked FXa, thrombin, and 12 other enzymes. Alboserpin displays high affinity binding to heparin (K D ϳ 20 nM), but no change in FXa inhibition was observed in the presence of the cofactor, implying that bridging mechanisms did not take place. Notably, Alboserpin was also found to interact with phosphatidylcholine and phosphatidylethanolamine but not with phosphatidylserine. Further, annexin V (in the absence of Ca 2؉ ) or heparin outcompetes Alboserpin for binding to phospholipid vesicles, suggesting a common binding site. Consistent with its activity, Alboserpin blocks prothrombinase activity and increases both prothrombin time and activated partial thromboplastin time in vitro or ex vivo. Furthermore, Alboserpin prevents thrombus formation provoked by ferric chloride injury of the carotid artery and increases bleeding in a dose-dependent manner. Alboserpin emerges as an atypical serpin that targets FXa and displays unique phospholipid specificity. It conceivably uses heparin and phosphatidylcholine/phosphatidylethanolamine as anchors to increase protein localization and effective concentration at sites of injury, cell activation, or inflammation.
To acquire a blood meal, hematophagous arthropods usually break the host's skin stratum corneum (1,2). As a result, wounded tissues and injured blood vessels trigger the host hemostatic process, which is a complex and redundant, inter-linked biological system consisting of platelet aggregation, blood clotting, and vasoconstriction (3)(4)(5). The clotting cascade can be activated by the intrinsic or extrinsic pathways, both converging to the activation of factor X (FX) 2 to FXa, which converts prothrombin to thrombin; the latter cleaves fibrinogen to produce fibrin (6). FXa also plays a pivotal role in inflammation through activation of protease-activated receptors (7). In order to counteract host response to injury, salivary glands (SGs) of blood-sucking arthropods express a number of inhibitors of blood coagulation, platelet aggregation, host immunity, inflammation, angiogenesis, neutrophil function, wound healing, and vasodilation (1, 8 -16). Among anticoagulants, inhibitors targeting FVIIa/tissue factor, thrombin, FXa, FIXa, FXIIa, and high molecular weight kininogen have been reported (17)(18)(19)(20). Of note, members of different protein families, including Kunitz-, ascaris-, and antistasin-like, and SALP have been identified as FXa inhibitors (17,21). Notably, the only serine protease inhibitor (serpin) described so far and found to target FXa was identified in the SG of the mosquito Aedes aegypti (22).
Serpins are a large and growing superfamily of structurally homologous yet functionally diverse proteins. Serpins have been identified in species of all kingdoms. These protease inhibitors act to control proteolytic activity in a myriad of contexts, including coagulation, fibrinolysis, and complement activation. They can also be found in the intracellular compartment and perform biologic functions that do not necessarily require their protease-inhibitory function. Approximately 500 serpins are currently known, consisting of between 350 and 400 amino acids with molecular weight ranging from 40 to 55 kDa (23)(24)(25). In this study, we describe the main anticoagulant from Aedes albopictus SGs. We provide experimental evidence indicating that the molecule responsible for this salivary activity belongs to the serpin superfamily (hereafter named Alboserpin). We show that recombinant Alboserpin is a highly specific, tight inhibitor of FXa. Both recombinant Alboserpin and A. albopictus saliva do not bind to FX or DEGR-FXa (active site-blocked factor Xa containing the fluorescent inhibitor dansyl-Gly-Gly-Arg chloromethyl ketone dihydrochloride). Moreover, Alboserpin binds heparin; notably, it also interacts with phosphatidylcholine (PC) and phosphatidylethanolamine (PE) but not with phosphatidylserine (PS). In addition, it displays potent antithrombotic properties in vivo.

MATERIALS AND METHODS
Unless otherwise indicated, the protocols followed standard procedures (26), and all experiments were performed at room temperature (25 Ϯ 1°C). All water used was of 18 megaohm quality, produced by a MilliQ apparatus (Millipore, Bedford, MA). Heparin (4 -6 kDa average) from porcine intestinal mucosa was purchased from Sigma. Egg L-␣-PC and brain L-␣-PS were obtained from Avanti Polar Lipids (Birmingham, AL). PE from bovine brain was purchased from Sigma.
Phylogenetic Analysis of the Serpin Family-BLASTP analysis was performed with Alboserpin (gi:56417456) against the nonredundant data base, and all blood-feeding insect sequences were retrieved. Sequences were cleaned up to obtain a nonredundant set (proteins with Ͼ95% identity in the core domain were treated as identical) and aligned with ClustalX (27,28). Alignments were manually checked, adjusted, and trimmed to include the conserved serpin core. Phylogenetic analysis was performed using neighbor-joining analysis (29). Gapped positions were treated by pairwise deletion. Poisson correction was used as a substitution model to determine pairwise distances. Confidence was determined using bootstrap analysis (10,000 replicates) with 346 informative sites.
Modeling and Electrostatic Surface Calculations-The structure of Alboserpin was modeled using the automatic modeling mode in SWISS-MODEL (30). Alboserpin coordinates were loaded on PyMOL, and the electrostatic surface was calculated using the Adaptive Poisson Bolzmann Solver (APBS) tools. The template structure was the antithrombin-S195A factor Xa-pentasaccharide complex (Protein Data Bank code 2GD4). Secondary structure was obtained based on the Alboserpin model.
A. albopicitus Collection of Saliva and SG Dissection-Samples of saliva from female A. albopictus mosquitoes were collected by oil-induced salivation. After saliva collection, the sample was spun down at 14,000 ϫ g in a bench top centrifuge, and the lower phase, containing the saliva, was transferred to a clean Eppendorf tube. SGs were dissected as indicated (31). The protein concentration from the collected saliva and SG extracts was estimated spectrophotometrically in an ND1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE).
Expression of Alboserpin in Escherichia coli-For bacterial expression of recombinant Alboserpin, a synthetic gene was designed coding for the mature protein, which also contains NdeI and XhoI restriction sites. The synthetic Alboserpin gene was subcloned into pET-17b (Biobasic Inc., Markham, Canada) for expression in E. coli (BL21pLYS) cells. Recombinant protein production and inclusion body preparation were carried out as indicated (9). The inclusion bodies were solubilized in 20 mM Tris-HCl, pH 7.4, 6 M guanidinium hydrochloride, 15 mM dithiothreitol, 1 mM EDTA. The solubilized material was diluted in 4 liters of 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.2 mM GSSG, 1 mM GSH, and 200 mM arginine monohydrochloride and incubated overnight. Refolded Alboserpin was concentrated and purified as indicated (9). The purified recombinant protein was submitted to automated Edman degradation for N-terminal sequencing. Concentration of purified Alboserpin (corrected for ⑀ 280 nm ϭ 44,410) (calculated using software from DNAStar Inc., Madison, WI) was estimated by its absorbance at 280 nm using a NanoDrop ND1000 spectrophotometer (NanoDrop Technologies).
Anticoagulant Assays-Anticlotting measurements were performed either by measuring the recalcification time as described before (32) or by prolongation of the activated partial thromboplastin time (aPTT) and prothrombin time (PT). aPTT was carried out as supplied by the aPTT reagent kit (Helena Laboratories, Beaumont, TX). Briefly, 30 l of recombinant Alboserpin or SG extracts at different concentrations and 30 l of normal reference plasma (American Diagnostica, Greenwich, CT) were incubated for 10 min at room temperature before adding 30 l of ALEXIN LS (diluted 1:3 in 20 mM HEPES, 120 mM NaCl, pH 7.4). After 5 min, clotting was induced with 30 l of 20 mM CaCl 2 , 20 mM HEPES, 120 mM NaCl, pH 7.4, and measured at 650 nm every 11 s for 30 min. PT was measured under the same conditions described above, replacing ALEXIN reagent with Thromboplastin reagent (Helena Laboratories) diluted 1:2 in 20 mM HEPES, 120 mM NaCl, pH 7.4. All readings were performed in a Thermomax microplate reader (Molecular Devices, Menlo Park, CA).
Kinetics or FXa Inhibition by Alboserpin-All reactions were carried out at 37°C. Five SG pairs from adult female mosquitoes (2-4 days old, non-blood-fed) were dissected under a stereoscopic microscope in 20 l of PBS (0.02 M sodium phosphate, 0.15 M NaCl, pH 7.4) and kept at Ϫ80°C until use. Factors X and Xa were obtained from Hematologic Technologies Inc. (Essex Junction, VT), and chromogenic substrate N-benzoyl-L-isoleucyl-L-glutamyl-glycyl-L-arginine-p-nitroaniline hydrochloride (S2222) was purchased from DiaPharma Group Inc. (Westchester, OH). Recombinant Alboserpin at concentrations ranging from 0.1 to 14.5 nM or SG extract (0.0023-0.3 g/l) was incubated with either 1, 3, or 6 nM FXa in 50 mM Tris, 120 mM NaCl, 5 mM CaCl 2 , 0.1% (w/v) BSA, pH 7.4, for 20 min at room temperature in a final volume of 100 l. The reaction was initiated by adding 375 M of S2222, and the FXa activity was measured spectrophotometrically at 405 nm for 1 h in a plate reader (Thermomax microplate reader, Molecular Devices).
For the inhibition constant determination of Alboserpin for FXa, the SpectroFluor FXa fluorogenic substrate (American Diagnostica) was used. Briefly, 25 pM factor Xa was allowed to interact with increasing inhibitor concentrations (0 -525 pM) in the presence of varying concentrations of substrate (100 -500 M), in 50 mM Tris, 120 mM NaCl, 5 mM CaCl 2 , 0.1% BSA, pH 7.4. Assays were performed in the 96-well plate format and were initiated by the addition of the enzyme to a mixture containing substrate and Alboserpin; reactions were followed for 30 min. Enzymatic activity was measured from the increase in absorbance of the free chromophore generated by substrate hydrolysis ( ex ϭ 360 nM, em ϭ 440 nM) using a SpectraMax Gemini XPS plate reader linked to SOFTmax Pro 3.0 software (Molecular Devices). For kinetics, the percentage inhibition of the reaction at different substrate concentrations was calculated from a control reaction containing only vehicle, and data were fitted using the Morrison equation (Equation 1) as reported (19), where K i * is the apparent dissociation constant for the enzymeinhibitor complex, V s is the inhibited steady-state velocity, V o is the control (uninhibited) velocity, [I t ] is the total inhibitor concentration, and [E t ] is the total FXa concentration.
Affinity Chromatography on a Heparin-Sepharose Column-Two M FXa was incubated for 2 min in the absence or presence of Alboserpin (5 M) in 50 mM Tris-HCl, pH 7.4, buffer containing 5 mM CaCl 2 . The sample (0.5 ml) was applied at 0.5 ml/min on a 1-ml HiTrap heparin-Sepharose (GE Healthcare) column pre-equilibrated with the same buffer. The column was washed with 8 ml followed by elution with a 30-ml gradient of 0 -1.0 M NaCl prepared in the same buffer. Fractions (0.5 ml) were collected, and their activity toward S2222 was determined as described above.
Prothrombinase Assembly-Activation of prothrombin by human FXa was performed in TBS-Ca 2ϩ (20 mM Tris-HCl, 150 mM NaCl, 5 mM CaCl 2 , 0.3% (w/v) BSA, pH 7.5), using a discontinuous assay. FXa (10 pM; final concentrations are given) was incubated with Alboserpin (0 -500 pM) for 20 min at room temperature. Human FVa (1 nM) and PC/PS (10 M) were added and incubated for 5 min. Reactions were initiated by the addition of human prothrombin (1.4 M). Aliquots of 25 l were removed every 1 min into microplate wells containing 50 l of TBS-EDTA (20 mM Tris-HCl, 150 mM NaCl, 20 mM EDTA, 0.1% BSA, pH 7.5) to stop reactions. After the addition of 25 l of S2238 (312.5 M), absorbance at 405 nm was recorded at 37°C for 15 min at 11-s intervals using a Thermomax microplate reader (Molecular Devices). Initial velocities (millioptical densities/min) obtained were used to calculate the amount of thrombin formed, using a standard curve (33). The absence of one of the components in the prothrombinase showed no thrombin formation.
Surface Plasmon Resonance (SPR)-Binding of recombinant Alboserpin to FXa, FX, and DEGR-FXa was analyzed at 25°C by SPR spectrometry using a BIAcore T100 instrument (BIAcore AB, Uppsala, Sweden) as described (34). Factors X, Xa, and DEGR-Xa (30 g/ml) were covalently immobilized on the surface of a CM5 sensor chip using the amine coupling kit supplied by BIAcore at a flow rate of 10 l/min (aiming to reach 1500 resonance units (RU) in 10 mM acetate buffer, pH 4.8), resulting in a final immobilization of 1470.8 RU for FXa, 1764.4 RU for FX, and 1807 RU for DEGR-FXa. Blank flow cells were used to subtract the buffer effect on the sensograms. For kinetic experiments, recombinant Alboserpin in concentrations ranging from 0.04 to 25 nM was run for 120 s at 30 l/min in HBS-P buffer (10 mM HEPES, 150 mM NaCl, 0.005% surfactant P20, pH 7.4). The Alboserpin-coagulation factor complex dissociation was monitored for 1800 s, and the active cell was regenerated with a 3-s pulse (50 l/min) of 10 mM glycine-HCl, pH 1.5. Kinetic experiments were performed in triplicate on different days. The same experimental design was utilized for kinetic analysis of bacterially expressed Alboserpin. BIAcore T100 evaluation software was used for kinetic evaluation, and sensograms were fitted using the 1:1 binding model (Langmuir binding model). Alternatively, an experiment was designed to test the presence of salivary Alboserpin in the mosquito's saliva. A new sensor chip was used so as to have the FX, Xa, and DEGR-FXa on the same sensor chip. The flow cell 1 was used as a blank to subtract the buffer effect on the sensograms. The immobilization procedure was the same as described above. Four different concentrations of A. albopictus saliva (15,25,50, and 90 g/ml in HBS-P) were manually injected over the four flow cells in the sensor chip for 90 s at a flow rate of 20 l/min. The complex dissociation was monitored for 500 s, and the sensor surface was regenerated by a pulse of 5 s of 10 mM glycine-HCl, pH 1.5, at 40 l/min. These experiments were carried out in duplicate.
Binding of recombinant Alboserpin to heparin was carried out by SPR using a BIAcore 3000 instrument. Heparin (4 -6 kDa average molecular mass) was biotinylated at the reducing end and injected over an SA-sensor chip (GE Healthcare) in HBS-P (10 mM HEPES, pH 7.4, 150 mM NaCl, 0.005% P20 surfactant) at a flow rate of 10 l/min for 2 min. The resulting immobilization level obtained was 230.6 RU. The kinetics assay was performed at 25°C in HBS-P. Alboserpin concentrations ranging from 3.9 to 62.25 nM were allowed to interact with immobilized heparin for 180 s at 30 l/min. Alboserpin-heparin complex dissociations were monitored for 600 s, and the sensor was regenerated with a 1-min injection of 2.5 M NaCl at 30 l/min. Kinetic parameters (k on , k off , and K D ) were determined by global fitting of the sensograms using the 1:1 Langmuir binding model in the BIAcore Evaluation 3.1 software.
Isothermal Titration Calorimetry-Calorimetric assays for measuring FXa binding to Alboserpin were performed using a VPITC microcalorimeter (Microcal, Northampton, MA) at 30°C. Titration experiments were performed by successive injections of 10 l each of 17 M FXa into the sample cell containing 2 M Alboserpin. Prior to the run, the proteins were dialyzed against 50 mM Tris-HCl, 0.15 M NaCl, pH 7.4, 5 mM CaCl 2 for binding experiments. All solutions were degassed under vacuum for 5 min before use.
The calorimetric enthalpy (⌬H cal ) for each injection was calculated, and binding isotherms were fitted according to a model for a single binding site by nonlinear squares analysis using Microcal Origin software. The enthalpy change (⌬H), and stoichiometry (n) were determined according to Equation 2, where Q is the total heat content of the solution contained in the cell volume (V o ) at fractional saturation , ⌬H is the molar heat of ligand binding, n is the number of sites, and Mt is the bulk concentration of macromolecule in V o . The binding constant, K a , is described as follows, where [X] is the free concentration of ligand.
Binding of Alboserpin to FXa by Size Exclusion Chromatography-Analysis of Alboserpin-FXa complex formation was carried out using a Superdex 75 PC 3.2/30 column (3.2 ϫ 300 mm; GE Healthcare), 2.4-ml bed volume, assembled in Akta Purifier equipment (Amersham Biosciences). The column optimal separation range is 300 -70,000 Da, with an exclusion limit of 100,000 Da. The column was equilibrated in HEPES-buffered saline, pH 7.4, at 40 l/min. For the experiment runs, recombinant Alboserpin and FXa were loaded independently on the column. For the Alboserpin-FXa complex, equimolar concentrations of both proteins were incubated in HEPES-buffered saline, pH 7.4, at room temperature. After 30 min of incubation, the mixture was loaded on the column, and the protein-protein complex was monitored at 280 nm.
Phospholipid Vesicle Formation and Protein Binding-Large unilamellar vesicles were formed by drying the lipids in a chloroform solution under a stream of nitrogen in a glass vial and then resuspended in 20 mM Tris-HCl, pH 7.4, vortexed, and extruded using a LiposoFast (Avestin Inc., Ottawa, Canada) extrusion device. Extrusion was performed through two (stacked) polycarbonated filters (19-mm diameter) of 0.1-m pore size (Whatman, Clifton, NJ). The final concentration of the vesicle suspensions was 1 mg/ml, and nitrogen was bubbled through the mixture during preparation. Binding was measured at various salt concentrations by diluting the vesicle suspension in TBS containing the appropriate concentration of NaCl and 3 M Alboserpin in a final volume of 50 l. A second set of competition experiments was carried out where Alboserpin and phospholipids were incubated in the presence of 30 g/ml heparin or 10 M annexin V. The mixtures were incubated at room temperature for 30 min and centrifuged at 100,000 ϫ g for 30 min. The supernatant was removed, and the pellet was resuspended in 40 l of 1ϫ LDS loading buffer supplemented with 1ϫ NuPAGE reducing agent. Both the pellet and supernatant fractions were analyzed by SDS-PAGE, and the fraction of bound protein was determined by densitometry of Coomassie Blue-stained gels using ImageJ software (public domain, open source; National Institutes of Health).
Animals and Tail Bleeding Assay; ex vivo PT and aPTT Assays-Female C57BL/6 mice (6 -10 weeks old) were purchased from Charles River Laboratories (Wilmington, MA). Mice were maintained at an American Association of Laboratory Animal Care-accredited facility at NIAID, National Institutes of Health. All experiments with mice were evaluated and approved by the NIAID Animal Care and Use Committee of the National Institutes of Health (Rockville, MD). For the tail vein bleeding assay, mice were injected intravenously with PBS or two doses of Alboserpin (10 and 100 g/kg) in a 100-l volume. After 30 min, mice were anesthetized and placed on a warming pad. A transverse incision was performed using a scalpel over a lateral vein at a position where the diameter of the tail is 2.5 mm. The tails were left hanging over the edge of the bench and immersed in a 12 ϫ 75-mm tube containing 4 ml of saline buffer for 30 min. The tubes were gently rotated from time to time to prevent the shed blood from obscuring the incision. Although only mild bleeding was expected to occur in the assay, mice were observed for the next 6 h for signs of delayed bleeding. The samples were properly diluted, and the absorbance was determined at 540 nm to estimate hemoglobin content. Results represent the mean Ϯ S.E. of three animals per group, and the assay was repeated twice.
A second group of mice identically treated with Alboserpin or PBS was used for ex vivo aPTT and PT assays. Briefly, 1 h after Alboserpin or PBS administration, blood was collected by cardiac puncture in 3.8% trisodium citrate (9:1, v/v). Platelet-poor plasma was obtained by centrifugation at 2000 ϫ g for 15 min. aPTT and PT assays were carried out as described above.
FeCl 3 -induced Artery Thrombosis-Mice were anesthetized with intramuscular xylazine (16 mg/kg) followed by ketamine (100 mg/kg). The right common carotid artery was isolated through a midline cervical incision, and the blood flow was continuously monitored using a 0.5VB Doppler flow probe coupled to a TS420 flow meter (Transonic Systems, Ithaca, NY) as described previously (35). Alboserpin (at 50, 100, or 200 g/kg) or vehicle was infused into the tail vein 15 min before injury. Thrombus formation was induced by applying a piece of filter paper (1 ϫ 2 mm) saturated with 7.5% FeCl 3 solution to the adventitial surface of the artery for 3 min. After exposure, the filter paper was removed, and the vessel was washed with sterile normal saline. Carotid blood flow was continuously monitored for 60 min or until complete occlusion (0 flow for at least 10 s) occurred.
Serine Protease Inhibition Assays-All assays were performed at 30°C in triplicate. Protein (300 nM) was preincubated, with each enzyme tested for 10 min before the addition of the corresponding substrate as described (36).
Statistical Analysis-Statistical differences among the groups were analyzed by analysis of variance using Tukey as a multiple comparison post-test. A p value of 0.05 or less was considered statistically significant.

RESULTS AND DISCUSSION
The sialotranscriptome of female A. albopictus SG (37) revealed a cluster of sequences coding for a putative serpin protein similar to the A. aegypti FXa-directed anticoagulant factor (22). The Alboserpin cDNA has 1257 nucleotides, coding for a protein of 418 amino acids, including a signal peptide of 19 amino acids (gi:56417456). The mature protein has a calculated molecular mass of 45,986.97 Da, with a pI of 9.21.
Alboserpin Is an Atypical Serpin-Salivary FXa inhibitors have been classified as Kunitz-type, antistasin family, and ascaris-type protease inhibitors and members of the SALP family (17). Sequence analysis of Alboserpin shows high sequence similarities to several other serpins, including plasminogen activator inhibitor-2 (PAI-2), antithrombin, and protein C inhibitor (PCI). A Clustal alignment of Alboseprin and other relevant serpins is presented in Fig. 1. A comparison of the reactive site loop with Alboserpin and known serpin proteins illustrates that although there is remarkable conservation in amino acid sequences, there is also considerable divergence.
For example, the reactive site loop in Alboserpin appears to be truncated in comparison with other serpins. The predicted P1 for Alboserpin aligns as an arginine that would be expected for an FXa-specific serine protease inhibitor. Furthermore, P17 within the hinge region is typically a glutamic acid residue, whereas in Alboserpin, it is a histidine residue. This P17 has been identified as glutamic acid for 39 serpins analyzed (23)(24)(25). The hinge region of serpins is proposed to be essential for protease inhibition (mousetrap mechanism) by promoting a conformational change that allows a tight interaction with the active site of serine proteases followed by cleavage of the P1-P1Ј peptide bond of the serpin reactive center loop (RCL) (23)(24)(25). This forms an acyl-intermediate complex in which the bond is cleaved and the P1 residue becomes covalently linked to the enzyme. This leads to a large conformational change in which the RCL inserts into the center of ␤-sheet A, causing the tethered proteinase to be dragged to the opposite end of the protein and inactivated by conformational deformation and irreversible binding. Other serpins have less typical mechanisms (23)(24)(25). Thus, Alboserpin shares primary amino acid sequence in common with typical serpins, including the appropriate spacing of domains, but significant divergence is evident and makes it a unique protein.
Alboserpin and Phylogeny- Fig. 2A shows BLASTP analysis using the FXa inhibitor from A. albopictus-retrieved serpins from several blood-feeding insects that included the tsetse fly Glossina morsitans morsitans, the sand fly Lutzomyia longipalpis, the cat flea Ctenocephalides felis, and the mosquitoes Anopheles gambiae, A. aegypti, and A. albopictus. The recently sequenced genomes of A. gambiae and A. aegypti contributed the bulk of the serpins retrieved. Most of these grouped into well supported clades, indicating that they are orthologs and limited to the head and fat bodies, which suggests that they perform housekeeping functions conserved in the ancestor to all Culicidae. The proteins from the Aedes genus that has a confirmed localization to the SG group into a well supported clade to the exclusion of any Anopheline sequences. BLASTP analyses of the genome of A. gambiae using any of these sequences do not retrieve other serpins, which suggests that no orthologs exist for these proteins in Anopheles. The closest homologs are serpin 14 from fat bodies. This suggests that a gene duplication event in SG led to the evolution of a new class of SG-derived serpins in Aedes sp.
Charges and Model-The structure of Alboserpin was modeled using the fully automated structure homology-modeling server. The model coordinates obtained were based on the antithrombin-S195A factor Xa-pentasaccharide complex (Protein Data Bank code 2GD4). Modeled Alboserpin shows a typical three-dimensional structure of native inhibitory serpin with its RCL completely exposed (Fig. 2B). The electrostatic surface generated using APBS shows a highly positively charged region similar to the one found in the antithrombin-S195A factor Xapentasaccharide complex (Fig. 2C). This region might be the heparin binding site in Alboserpin. Further structural study is necessary to address this question.
Anticlotting Activity in Saliva and SGs of A. albopictus-In an attempt to characterize the anticoagulant activity of Alboserpin, the corresponding cDNA was cloned in a Pet17b expression vector, and expression was induced by IPTG. Alboserpin was purified in a Mono-Q ion exchange column (not shown) followed by a last step in molecular sieving chromatography. It was eluted at 10 ml, which corresponds to a protein of ϳ45 kDa (Fig. 3A). This finding was consistent with the predicted molecular weight for Alboserpin and is indicative of high purity of the preparation as confirmed by SDS-PAGE (Fig. 3A, inset). Recombinant Alboserpin produced in E. coli was tested in coagulation assays. Fig. 3B shows that Alboserpin prolongs both PT and aPTT, whereas Fig. 3C shows that it also blocks prothrombinase assembly in a dose-dependent manner, consistent with FXa-directed anticoagulant. A. albopictus SGs also increase the recalcification time (Fig. 3D) and prolong both PT and aPTT (Fig. 3E). In addition, SPR experiments demonstrated unambiguously that saliva displays FXa-binding properties (Fig. 3F). A, phylogenetic analysis of serpins found in blood-feeding insects. Analysis was performed using neighbor-joining analysis. Bootstrap support (10,000 replicates) is indicated by gray (Ͼ70%) and black (Ͼ90%) dots. The red dots indicate proteins that tested positive for FXa inhibition, whereas blue dots indicate proteins that tested as negative. Serpins are labeled according to the annotation found for serpins from the genome of A. gambiae. Information on organ localization was obtained by tblastn analysis of the expressed sequence tag database. Sequences are indicated by the three-letter abbreviation of the species name followed by the GI accession number. B and C, models were constructed using the coordinates generated with Swiss-Model (automated mode). A surface map was generated using PyMOL APBS tools. B, ribbon diagram of the Alboserpin model with the reactive center loop side chains highlighted. C, electrostatic potential surfaces of Alboserpin showing the positively charged surface (blue) corresponding to the putative heparin-binding site. Table 1 shows that incubation of Alboserpin with a series of enzymes does not affect their catalytic activities, indicating that Alboserpin is a highly specific FXa inhibitor.

Specifiticy of Alboserpin-
Alboserpin Is a Tight Inhibitor of FXa-Experiments revealed that when FXa was added to the reaction buffer containing Alboserpin and chromogenic substrate, the progress curve displayed a downward concavity (Fig. 4A, curves b and c). These results resemble slow binding kinetics, as seen with many peptidic inhibitors of serine proteinases when small chromogenic or fluorogenic substrates are used to kinetically characterize the interaction. However, appropriate kinetics studies are needed to further address this question. In addition, Alboserpin significantly inhibits FXa at concentrations similar to that of the enzyme (Fig. 4B), indicating that Alboserpin is also a tightbinding inhibitor. Conventional Michaelis-Menten kinetics do not apply to the study of tight binding inhibitors, because they assume that the free inhibitor concentration is equal to the total inhibitor concentration, a reasonable approximation when the enzyme used is at a much lower concentration than the inhibitor. Therefore, Morrison's equation for tight binding inhibi-   (38) was used as described (19) to obtain apparent dissociation constants for Alboserpin. In these experiments, 25 pM FXa was allowed to interact with increasing inhibitor concentrations (0 -525 pM) for 30 min in the presence of varying concentrations of substrate (100 -500 M) before product rate formation for the following 30 min was recorded (not shown).
Resulting steady-state rates were fit by nonlinear regression to Equation 1 for several substrate concentrations as described (19). When the K i * was plotted against the substrate concentration, a linear regression line (r 2 ϭ 0.9923) indicated a y intercept of 68.2 Ϯ 5.4 pM, which is the K i value for the binding of Alboserpin to FXa (Fig. 4C).

Kinetics and Stoichiometry of Alboserpin-FXa Interactions-To investigate binding kinetics of Alboserpin-FXa interactions,
SPR experiments were performed. Typical sensograms are shown in Fig. 5A. Best fit was attained using a 1:1 model. Using this model, a k on of 1.005 ϫ 10 6 M Ϫ1 s Ϫ1 and k off of 2.685 ϫ 10 Ϫ5 s Ϫ1 were obtained . The fast k on for Alboserpin-FXa interaction determined by SPR suggests that enzyme activity is blocked very shortly after binding to the serpin. Accordingly, it is plausible that interactions of FXa with Alboserpin can be better described as fast when physiological macromolecules (FXa) instead of small chromogenic substrates (S2222) are used to determine the kinetics of the interaction involving enzyme and inhibitor. Similar findings have been reported before for tick salivary FVIIa/TF inhibitor, Ixolaris (20). Although the remarkably low k off makes it inaccurate to calculate the K D , it was estimated at 26.85 pM, which is in reasonable agreement with the affinity estimated by the Morrison equation (Fig. 4C). Notably, covalent binding of Alboserpin to FXa did not take place because acidic conditions (glycine, pH 1.5) dissociate Alboserpin from FXa immobilized in the sensor chip in contrast to typical serpins. In this respect, Alboserpin more likely behaves as a Kunitz-type inhibitor, for which binding to serine proteases is typically reversible (39). No binding was observed for FX or DEGR-FXa (Fig. 5B), which is an indication of the catalytic site dependence for binding; however, interaction of the serpin with FXa exosite cannot be excluded (40,41).
Binding of Alboserpin to FXa was measured by isothermal titration calorimetry, and the results are shown in Fig. 5C. Fitting of the observed enthalpies to a single-site binding model revealed a K D of Ͻ1.0 nM for Alboserpin binding to FXa. Estimation of the dissociation constant is limited by the parameter c ϭ K a [P], where K a is the association equilibrium constant and [P] is the protein concentration in the calorimeter cell. Measurement of the equilibrium constant is considered unreliable when the value for c exceeds 1000. Binding was exothermic, with a favorable enthalpy (⌬H) of Ϫ34.77 kcal/mol and unfavorable entropy (⌬S ϭ Ϫ73.9 cal/mol K) for binding of the enzyme to the inhibitor. Stoichiometry of the binding (n ϭ 0.85 Ϯ 0.003) indicates that one FXa molecule binds to one Alboserpin molecule (Fig. 5C). Binding of Alboserpin to FXa was also confirmed by identification of complex formation that eluted at a molecular weight compatible with 1:1 interaction, as estimated by gel filtration chromatography (Fig. 5D). Fig. 5E demonstrates that Alboserpin-FXa complex formation dissociates when the sample is warmed at 70°C for 7 min, indicating that the complex is reversible in contrast to typical serpins. This result is consistent with reversible interaction also demonstrated by SPR. Although Alboserpin-FXa complex formation is kinetically reversible, it is conceivable that this tight interaction results in perhaps a functionally irreversible binding in nature. Finally, saliva was found to interact with FXa but not DEGR-FXa, consistent with the presence of Alboserpin in the native secretion (Fig. 5F). Alboserpin Interacts with Heparin-Heparin accelerates the rate of vertebrate plasma antithrombin (AT) inhibition of FXa by 3-4 orders of magnitude (40). In fact, AT that normally regulates the proteolytic activity of FXa is a weak inhibitor of coagulation proteinases unless it binds heparin-like glycosaminoglycans. Heparin binding to both protease and serpin enhances the rate of encounter complex formation between the two proteins (bridging effects) (42). To verify whether Alboserpin binds to heparin, SPR experiments were performed as described under "Materials and Methods." Fig. 6A shows sensograms of Alboserpin-heparin interaction. Best fit was achieved with a 1:1 Langmuir model, yielding a K D of 20.8 nM (k on of 1.13 ϫ 10 6 M Ϫ1 s Ϫ1 and k off of 2.36 ϫ 10 Ϫ2 s Ϫ1 ). In addition, Fig. 6B demonstrates that FXa binds to a heparin-Sepharose column and is eluted with a NaCl gradient. Next, we studied the effects of heparin in the inhibition of FXa by Alboserpin. FXa (1 nM) was used to start reactions containing a mixture of S2222 and Alboserpin, with and without heparin. Alboserpin was tested at low concentrations (Ͻ0.5 nM), producing partial inhibition of FXa catalytic activity, in order to observe any possible effects of heparin. Fig. 6C shows that progress curves for FXa inhibition by Alboserpin were the same in the absence or in the presence of heparin. The effects of heparin were also tested after incubation of FXa with Alboserpin, followed by the addition of S2222 to start reactions. Fig. 6D demonstrates that the curves were superimposed, indicating no change in the affinity. It is concluded that Alboserpin belongs to the subfamily of serpins, such as AT (40), heparin cofactor II (43), protease nexin I (44), protein Z inhibitor (45), PAI-1 (46), and PCI (47), that reportedly binds heparin and other glyco- saminoglycans. However, heparin does not operate as a bridge for Alboserpin to interact with proteinases. In other words, Alboserpin interacts with FXa irrespective of the presence of cofactor heparin. This is not surprising, taking into account the tight binding of FXa and Alboserpin, the affinity of which is in the range of thrombin-antithrombin complex when heparin is present. It is also important to recognize that heparan sulfatecontaining proteoglycans are found membrane-associated to the cell surface and can substitute for heparin. Plausibly, heparin functions as an anchor that targets Alboserpin to endothelial cells, localizing it at the site of injury, where it remains bound instead of free in solution, resulting in higher effective concentration. In this context, tissue factor pathway inhibitor and vascular endothelial growth factor binding to heparin has also been demonstrated and sought to contribute to localization in endothelial cells and other glycosaminoglycan-bearing cells (48,49).
Alboserpin Interacts with PC and PE but Not with PS-Among serpin family members, PCI was initially found to bind to PE, PS, and PC (50). However, PCI was later demonstrated to interact with oxidized PE or oxidized PS but not with PC (47,51). Therefore, binding of Alboserpin to phospholipids vesicles in solution was investigated. Alboserpin was incubated with PC, PS, PE, or PC/PS for 15 min, followed by centrifugation at 100,000 ϫ g. Phospholipid preparation was performed under N 2 in order to avoid oxidation. The pellet and the supernatant were collected, and bound (found in the pellet) or free (found in the supernatant) Alboserpin was detected by SDS-PAGE. Fig.  7A demonstrates that Alboserpin binds to PC and PE but, surprisingly, not to PS. This phospholipid specificity distinguishes Alboserpin and PCI, which is a PS-binding serpin (50). It also binds to a mixture of PC/PS (4:1), which mimics phospholipid composition of activated platelets. When salt concentration was increased above 0.2 M NaCl, Alboserpin interaction with PC/PS was gradually lost (Fig. 7B for gel; Fig. 7C for quantification). In addition, as a control, salivary protein (AeDL1) also expressed in E. coli did not bind to phospholipids (Fig. 7B). It is important to recognize that high salt concentrations may affect Alboserpin binding to vesicles not only through electrostatic interference but also by affecting vesicle structure. Although we cannot exclude the latter possibility, our most relevant finding is that Alboserpin binds phospholipids in solution, at physiologic ionic strength. Annexin V, a protein that binds PS in the presence of Ca 2ϩ but PC in its absence (52) was used to investigate whether it can compete with Alboserpin for PC/PS binding. Experiments performed in the absence of Ca 2ϩ demonstrate that annexin V prevents interaction of Alboserpin with PC/PS (Fig. 7D), suggesting competition for the same binding site (i.e. PC). In addition, incubation of Alboserpin with heparin completely abolished its interaction with PC/PS vesicles  AUGUST 12, 2011 • VOLUME 286 • NUMBER 32 in solution, suggesting that the heparin-binding site of Alboserpin is also required for phospholipid binding (Fig.  7E). It is plausible that PC and PE, but not PS, works as an anchor that targets the inhibitor to cells at sites of injury or cells that have been activated or present at sites of inflammation. For example, PE, in addition to PS, is the major phospholipid component (nearly 40%) of the outer leaflet of activated platelet membrane; accordingly, PE conceivably targets Alboserpin to sites of platelet activation. Studies have also demonstrated that PE enhances the PS-mediated sensitivity of factor VIIa-tissue factor activity (53) and induces high affinity binding sites for factor VIII on surfaces containing PS (54). Moreover, PE has been reported in tumor cell membranes (55) and is known to regulate the activation of blood coagulation by enhancing APC inhibition by PCI (50,51). It is possible that occupation of these sites by Alboserpin interferes with the inhibitory function of PCI toward APC, which displays particularly important anticoagulant and anti-inflammatory activities in vivo (7,56). In addition, PC in its oxidized form is found in atherosclerotic lesions, apoptotic cells, and oxidized LDL and stimulates endothelial cells to produce inflammatory cytokines, leukocyte chemoattractants, and coagulation factors (57,58). The remarkable mechanism of action of Alboserpin (which concentrates and localizes the inhibitor at sites of injury or cell activation) suggests that it has evolved to remain bound at sites of inflammation where the concentrations of PE or PC are presumably very high. It is also clear that the phospholipid specificity of Alboserpin is unique among family members of the serpin family described so far in the sense that, in contrast to PCI, it does not recognize PS but interacts with PC. On the other hand, a lipocalin from Rhodnius prolixus binds to PS but not other phospholipids and affects multimolecular coagulation complex formation (59). These molecular adap- tations are notable examples of how evolutionary pressure has dictated the function of salivary components from blood-sucking arthropods.

Factor Xa Inhibition by Alboserpin
Alboserpin Displays Antithrombotic and Anticoagulant Activity in Vivo-To demonstrate that the anticoagulant effects of Alboserpin translate into inhibition of thrombus formation, the inhibitor was injected in mice, which were then submitted to FeCl 3 -induced carotid artery injury. Application of FeCl 3 to the exterior of blood vessels causes severe endothelial damage and occlusion by platelet-rich thrombi. Times to occlusion were not significantly different between control and mice treated with 50 g/kg Alboserpin (16.11 Ϯ 1.47 versus 17.0 Ϯ 1.63 min) (Fig. 8A). In contrast, mice treated with 100 and 200 g/kg Alboserpin were resistant to arterial occlusion, and in these cases, occlusion did not take place before 60 min for most animals (Fig. 8A). These results indicate that Alboserpin inhibition of FXa effectively blocks thrombin generation at sites of vascular injury. Next, the effects of Alboserpin in bleeding were estimated using the tail transection method. Fig. 8B shows that Alboserpin (100 g/kg) produces significant bleeding, as would be expected for a potent anti-FXa inhibitor. Furthermore, injection of Alboserpin (100 g/kg) in mice prolongs PT and aPTT ex vivo ( Table 2). Whereas the concentrations used in the experimental thrombosis model are high when compared with the amount of Alboserpin injected at the site of mosquito bite, this is explained by the intensity and extension of the injury provoked by FeCl 3 . However, one may calculate the approximate concentration of the inhibitor injected by the mosquito, which causes a microinjury to the host. Accordingly, the volume of blood taken in each blood meal is ϳ4 l, and ϳ50% of the salivary gland content (ϳ1 g) is released during the bite, Alboserpin representing ϳ1% of the salivary proteins (ϳ0.01 g). This results in a concentration of ϳ40 nM, which is much above the Alboserpin K D for FXa. Therefore, it is evident that Alboserpin behaves as an anticoagulant and antithrombotic in vivo, which is consistent with its biologic activity characterized in vitro.
Conclusions-Unique binding behavior, shorter and distinct composition of the RCL, reversible interaction with the enzyme, and no requirement for cofactors places Alboserpin as a useful prototype to understand structural features of FXa and serpin mechanisms of protease inhibition. Structural studies of Alboserpin will define the precise mechanism by which this molecule specifically and reversibly blocks FXa, thus preventing its role in thrombosis and inflammation (7). Alboserpin may also be regarded as a prototype to develop anticoagulants targeting FXa, an important target of antithrombotic therapy, clinically illustrated by the effectiveness of oral FXa inhibitors rivaroxaban and apixaban (60).