Characterization of Anopheles gambiae Transglutaminase 3 (AgTG3) and Its Native Substrate Plugin*

Background: A. gambiae transglutaminase AgTG3 cross-links Plugin within seminal fluids. Results: AgTG3 and Plugin were purified and their structure and activity characterized in vitro. Conclusion: AgTG3 forms a dimer in solution, analogous to human FXIII. Plugin is nonglobular in solution. AgTG3 is Ca2+-dependent and prefers Plugin as a substrate. Significance: Inhibition of AgTG3 cross-linking of Plugin is a possible method to chemosterilize male Anopheles mosquitoes. Male Anopheles mosquitoes coagulate their seminal fluids via cross-linking of a substrate, called Plugin, by the seminal transglutaminase AgTG3. Formation of the “mating plug” by cross-linking Plugin is necessary for efficient sperm storage by females. AgTG3 has a similar degree of sequence identity (∼30%) to both human Factor XIII (FXIII) and tissue transglutaminase 2 (hTG2). Here we report the solution structure and in vitro activity for the cross-linking reaction of AgTG3 and Plugin. AgTG3 is a dimer in solution and exhibits Ca2+-dependent nonproteolytic activation analogous to cytoplasmic FXIII. The C-terminal domain of Plugin is predominantly α-helical with extended tertiary structure and oligomerizes in solution. The specific activity of AgTG3 was measured as 4.25 × 10−2 units mg−1. AgTG3 is less active than hTG2 assayed using the general substrate TVQQEL but has 8–10× higher relative activity when Plugin is the substrate. Mass spectrometric analysis of cross-linked Plugin detects specific peptides including a predicted consensus motif for cross-linking by AgTG3. These results support the development of AgTG3 inhibitors as specific and effective chemosterilants for A. gambiae.

Transglutaminases (TGs) 2 are ubiquitous enzymes that catalyze the deamidation and transamidation of glutamine and the cross-linking of proteins by formation of ⑀-(␥-glutamyl)-lysine isopeptide bonds (1). Mammals possess multiple TGs that are involved in processes such as blood clotting by Factor XIII (FXIII), regulating cellular responses to stress by tissue transglutaminase (TG2), formation of the epithelial barrier (TG3), and coagulation of seminal plasma (TG4) (2). Similar roles for TGs have been identified in insects yet with a reduced gene set. The single TG gene present in Drosophila is involved in cuticle morphogenesis (3) and coagulation of hemolymph in response to septic injury (4). This TG is conserved in mosquitoes (Culicidae) including the malaria vector Anopheles gambiae (AGAP009100, or AgTG1). A second TG is specific to Culicidae (A. gambiae AGAP009098, or AgTG2), whereas a third TG is specific to Anopheles: A. gambiae AGAP009099, hereafter termed AgTG3.
Although AgTG1 and AgTG2 have yet to be functionally characterized, AgTG3 has recently been found to play a role in male Anopheles fertility (5). In many animals the coagulation of male seminal fluids results in the formation of a "copulatory plug" or "mating plug" that completely occludes the female reproductive tract. A recent study of A. gambiae seminal fluid proteins and male accessory glands showed that the plug is formed by cross-linking of a substrate protein called Plugin (AGAP009368) by AgTG3 (5). Furthermore, dsRNAi-mediated knockdown of AgTG3 inhibited proper sperm storage by females, indicating that correct formation of the mating plug is required for reproductive success of Anopheles males. Within Culicidae, formation of a mating plug is specific to Anopheles (6). TG-mediated cross-linking of seminal fluid proteins occurs in rodents (7), but other invertebrates use different mechanisms, such as secretion of O-glycosylated mucins in Caenorhabditis elegans (8) and homopolymerization of PEBII in Drosophila melanogaster (9).
Formation of A. gambiae mating plug is interesting because it represents not only the independent evolution of a seminal TG in an insect but also a target for control of Anopheles, which are vectors for malaria, the world's most devastating parasitic disease. Targeting fertility of insect populations, the sterile insect technique, is a classic method of pest control first used to con-trol the cattle screwworm fly (10). Field tests with radiationsterilized, chemosterilized, or genetically modified sterile mosquitoes have demonstrated feasibility of the sterile insect technique for public health purposes (11)(12)(13), provided that logistical and regulatory issues are considered (14 -16). A potent and specific inhibitor of AgTG3 could serve as a safe and effective chemosterilant for use alone or in combination with radiation and/or genetically modified mosquitoes to advance the application of sterile insect technique for Anopheles mosquitoes.
AgTG3 contains four canonical domains of the ␣/␤ TG fold and residues of the catalytic triad (Fig. 1A). The substrate Plugin has an unusual sequence with two distinct, repetitive domains. An N-terminal 30-kDa region (Fig. 1B) has a primary sequence dominated by Gln, His, and Pro, with multiple repeats of the consensus sequence PQIP(X) 3-6 QIPQ(X) 8 -9 K and no predicted secondary structure. In contrast, the C-terminal domain is rich in both Gln and Glu (Fig. 1C), consisting of six ␣-helices commencing with a consensus sequence EQE(K/R)QIGREML. Thus, the N-and C-terminal domains of Plugin are structurally distinct, yet both are rich in the TG substrate motif QI(P/G) and accompanying Lys residues.
AgTG3 is homologous at ϳ30% sequence identity to all TGs of known structure: FXIII, hTG2, hTG3, and Pagrus major TG (17)(18)(19)(20)(21). FXIII is a dimer in solution whereas other TGs are monomeric. All TGs are activated by calcium (Ca 2ϩ ), but only one structure, hTG2, bound by a peptidomimetic inhibitor, exhibits the presumed active conformation (21); other structures adopt a conformation in which the catalytic site is occluded by the ␤-barrel domains even in the presence of Ca 2ϩ . Other mechanisms of regulation exist for TGs besides Ca 2ϩ . Serum FXIII requires proteolysis by thrombin for activation, although cytoplasmic FXIII can be activated without proteolysis (22). The activation of hTG2 is inhibited by the binding of GTP (23). Hence, some initial questions regarding the mechanism of AgTG3-mediated cross-linking of Plugin are the oligomeric states of both enzyme and substrate in solution, the specific activity of AgTG3, and the requirements for its activity.
Here we report the expression, purification, and biophysical characterization of AgTG3 and its substrate Plugin. The in vitro reconstitution of Plugin cross-linking by AgTG3 was performed, and specific cross-linking sites within Plugin were identified. We demonstrate that AgTG3 can exist as both a monomer and dimer whereas Plugin has an extended or partially unfolded structure and tendency to oligomerize in solution. We measured the specific activity of purified AgTG3 by a fluorescent assay both in solution and in a high throughput plate-based format and confirm that the activity of AgTG3 is dependent on Ca 2ϩ but not GTP. Our results provide the basis for further molecular characterization of the AgTG3-Plugin mechanism of seminal fluid coagulation and screening for specific AgTG3 inhibitors.
For Plugin-C purification, cells were lysed by sonication in 20 mM HEPES, pH 7.5, 500 mM NaCl, 20 mM imidazole, and the lysate was clarified by centrifugation. The supernatant was loaded onto a HisTrap HP column, eluted with a 0 -600 mM imidazole gradient and dialyzed against a tobacco etch virus cleavage buffer (25 mM Tris-HCl, pH 8.0, 150 mM NaCl). Except in the case of plate-based assays, the His 6 tag was removed by treatment with tobacco etch virus protease overnight at room temperature and reapplication to a HisTrap column. Plugin-C was further purified by Superdex 200 SEC in 25 mM HEPES, pH 7.5, 0.1 M NaCl.
Peaks were detected with an in-line UV detector (Jasco UV975) at 280 nm, a light scattering detector (DAWN EOS; Wyatt Technology Corp.) at 690 nm, and a refractive index detector (Optilab; Wyatt Technology Corp.). Protein concentration was based on differential refractive index (dn/dc). The molecular mass was determined from the Debye plot of light scattering intensity versus scattering angle (Astra software; Wyatt Technology Corp.).
FITC Incorporation Assay-Plugin-C cross-linking was performed in 25 mM HEPES, pH 7.5, 10 mM CaCl 2 , 10 mM DTT, 0.5 mM FITC-cadaverine (FITC-CAD) at room temperature. After 2 h, the reaction was quenched by heat denaturation of proteins in the presence of Laemmli buffer and subjected to sodium  dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The gel was scanned using Alpha Imager 2200 software.
Specific Activity Assays-All assays are the average of three measurements. For the solution assay, Plugin-C was reductively methylated with dimethylamine borane using standard procedures (28). Assay conditions were 1.25 ml of 150 mM NaCl, 55 mM HEPES, pH 7.5, 10 mM DTT, 100 M FITC-CAD, 1 mg/ml Plugin-C, 9.6 g/ml AgTG3, and 5 mM CaCl 2 . Solutions were prewarmed to 30°C before addition of CaCl 2 (6.25 l of 1 M stock) to initiate reactions. At the indicated time points, 200-l aliquots were removed, mixed with 22 l of 0.5 M iodoacetamide, and incubated for 30 min at room temperature. Aliquots were diluted to 0.5 ml with PBS and loaded onto a PD-10 desalting column (GE Healthcare) followed by 2 ml of PBS, eluted with 1.5 ml, and diluted to 3.5 ml with PBS. Fluorescence (485/ 515) was measured on a JY Horiba Fluorolog-3 and fluorescence quantified relative to a FITC-CAD standard curve.
ELISAs for specific activity relative to guinea pig TG2 were performed with a standard kit (Sigma CS1070). For Plugin plat-ed-based assay, Plugin-C (25 g/well) was incubated in black 96-well Ni-coated plates (Pierce 15342) for 2 h at room temperature or overnight at 4°C. Plates were washed three times with 200 l of TBS and loaded with 50 l of AgTG3 (diluted to given concentration with water), then 50 l of 2ϫ assay buffer (2ϫ TBS, 20 mM CaCl 2 , 100 M FITC-CAD, 2 mM DTT). Wells were incubated for 60 min at room temperature, washed three times with 200 l of TBS, and filled to 100-l final volume for fluorescence measurement in a microplate reader (Biotek Synergy 2). Inhibition was performed by 30-min preincubation with 50 mM iodoacetamide (Sigma A3221).
Mass Spectrometry-Cross-linked Plugin-C was separated on SDS-PAGE, excised, and subjected to in-gel tryptic digest. Peptides were separated on a Waters nanoACQUITY (75 m ϫ 250 mm eluted at 300 nl/min) and analyzed on a LTQ Orbitrap    FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7 JOURNAL OF BIOLOGICAL CHEMISTRY 4847 mass spectrometer. Results were analyzed using MassMatrix (29). Identities of crosslinked peptides were confirmed by Collision-induced Dissociation tandem mass spectrometry (CID-MS/MS).

RESULTS
AgTG3 was expressed and purified to homogeneity in E. coli ( Fig. 2A). Purification and SDS-PAGE were equivalent in reducing and nonreducing conditions; hence, AgTG3 contains no disulfide bonds. The AgTG3 CD spectrum (Fig. 2B) corresponds to 40 -50% ␤-sheet and 5-15% ␣-helix. In the presence of 10 mM CaCl 2 AgTG3 elutes from Superdex 200 column with an apparent molecular mass of ϳ160 kDa suggesting a dimeric species (Fig. 2C). In the presence of 1 mM EDTA AgTG3 appears more heterogeneous, with the appearance of a shoulder peak at the apparent molecular mass of a monomer. AUC sedimentation velocity analysis of Ca 2ϩ -AgTG3 (Fig. 2D) gives a molecular mass estimate of ϳ170 kDa (s 20,w ϭ 7.0, f/f 0 ϭ 1.4), consistent with the estimate from SEC.
We used SAXS to further analyze the structure of Ca 2ϩ -AgTG3 in solution. In 1 M glycerol Ca 2ϩ -AgTG3 gave consistent scattering curves in the concentration range 1.0 -5.0 mg/ml (Fig. 2E). Guinier analysis (Fig. 2E, inset) at 1 mg/ml determined a radius of gyration R G ϭ 42.7 Å. Scattering at zero angle (I 0 ) was directly proportional to concentration, although a systematic increase in R G suggests some nonideality of the solute. Calibration with a 1.0 mg/ml solution of BSA gave an estimated molecular mass of 173 kDa, consistent with that of a dimer. The radial distribution function p(r) calculated by GNOM (30) requires D max ϭ 160 Å (Fig. 2F), almost 4ϫ the R G , implying that the Ca 2ϩ -AgTG3 dimer is a prolate ellipsoid. Ab initio shape determination was performed using 20 independent models calculated by DAM-MIF (31) with P2 symmetry and normalized spatial discrepancy (32) NSD ϭ 0.67 Ϯ 0.18.
The most probable volume for the Ca 2ϩ -AgTG3 dimer (Fig.  2G) suggests a structure similar to that of the Ca 2ϩ -bound FXIII dimer (18), yet with an open conformation for each AgTG3 monomer, analogous to that observed for inhibitor-bound hTG2 (21). A homology model for Ca 2ϩ -AgTG3 based on hTG2 (Protein Data Bank ID 2Q3Z) was superimposed on the FXIII dimer (Protein Data Bank ID 1GGU) and aligned with the most probable volume; imposition of P2 symmetry gives a good fit to the model.
Full-length Plugin is susceptible to proteolysis and aggregates in solution. Plugin-C (residues 345-557) was purified to homogeneity (Fig. 3A). The CD spectrum (Fig. 3B) has minima at 208 nm and 222 nm consistent with predominant ␣-helical secondary structure. At Յ0.5 mg/ml Plugin-C exhibits an apparent molecular mass of 90 kDa on SEC, but static light scattering (Fig. 3C) is consistent with a monomeric molecular mass of ϳ30 kDa, suggesting that the protein has a nonglobular conformation in solution. AUC sedimentation velocity analysis (Fig. 3D) also yielded a molecular mass estimate of ϳ30 kDa for Plugin-C (s 20,w ϭ 2.1 S, f/f 0 ϭ 1.4).
We analyzed Plugin-C (0.5 mg/ml) by SAXS to confirm its extended structure in solution. The scattering curve (Fig. 3E) resembles that of an unfolded or heterogeneous sample, precluding calculation of a p(r) function. Guinier analysis (Fig.  3F) yields an estimated molecular mass of ϳ30 kDa based on I 0 (calibrated with BSA) but an unusually large radius of gyration (R G ϭ 43 Å). In such cases the Kratky plot provides a qualitative measure of flexibility in a scattering particle (33,34). Both absolute (Fig. 3G) and dimensionless (Fig. 3H) Kratky plots for Plugin-C suggest a partly unfolded tertiary structure. At concentrations Ͼ0.5 mg/ml Plugin-C forms oligomers or soluble aggregates that interfere with particle analysis.
Cross-linking of Plugin by AgTG3 was previously demonstrated using endogenous extracts of male accessory glands, monodansylcadaverine, Ca 2ϩ , and GTP (5). We therefore reconstituted AgTG3 cross-linking of Plugin-C in vitro. Combining AgTG3 and Plugin-C in the presence of Ca 2ϩ produced a ladder of cross-linked Plugin-C bands of increasing molecular mass (Fig. 4A). The presence of FITC-CAD led to fluorescent labeling of both the monomer and cross-linked bands (Fig. 4B), confirming that cross-linking was the result of transamidation. The requirement for Ca 2ϩ was confirmed by the inhibitory effect of the chelating agent EGTA. Whereas Ca 2ϩ binding sites are conserved between AgTG3 and other TGs, the GTP binding site found in tissue TG (TG2) is not: only two binding site res-idues are conservative substitutions (S171A, F174G, R476E, R478K, S482E, R580K). Consistent with this observation, no effect on AgTG3 cross-linking of Plugin-C was detected in the presence of GTP.
Next, we used the conjugation of FITC-CAD to Plugin-C to determine the specific activity of transamidation for AgTG3. Reductive methylation of Plugin-C prevents the transamidation of lysine residues. Fluorescence was measured after incubation for 0 -60 min at 30°C and buffer exchange (PD-10 column) to remove unbound FITC-CAD (Fig. 5A). In a 1.25-ml total volume, 1 mg/ml (40 M) Plugin-C, 100 M FITC-CAD, 1.2 g of AgTG3 cross-linked FITC-CAD to Plugin-C at 0.408 M min Ϫ1 , yielding a specific activity of 4.25 ϫ 10 Ϫ2 unit mg Ϫ1 or 3.5 unit mol Ϫ1 AgTG3 (1 unit ϭ 1 mol min Ϫ1 FITC-CAD).
The specific activity of AgTG3 determined above is 100ϫ lower than is typically observed for standard TGs such as guinea pig TG2 (35). To confirm this result we performed a standard ELISA for transamidation of a biotinylated peptide (Btn-TVQQEL-OH) to polylysine-coated plates, with guinea pig TG2 (1.5 unit mg Ϫ1 ) as a standard (Fig. 5B). The relative specific activity of AgTG3 was 0.04 Ϯ 0.01 unit mg Ϫ1 , significantly less than is measured for human Factor XII or TG2 (1 unit ϭ 1 mol min Ϫ1 hydroxamate from N␣-Z-Gln-Gly, NH 2 OH, pH 6.0 at 37°C).
To test whether the specific activity of AgTG3 may be substrate-dependent we repeated our FITC-CAD assay in a plate-based format. His 6 -tagged Plugin-C was adsorbed to Ni-coated plates and incubated with FITC-CAD and either guinea pig TG2 or AgTG3 for 60 min at room temperature (Fig. 5C). The specific activity of AgTG3 was 0.3-0.4 unit mg Ϫ1 relative to guinea pig TG2, an 8 -10-fold increase compared with the previous assay. Iodoacetamide irreversibly inhibits transglutaminases by acetylating the active site cysteine. AgTG3 (10 g) was efficiently inhibited by 30-min preincubation with iodoacetamide in the plate-based Plugin/FITC-CAD assay (Fig. 5D).
To identify specific AgTG3 cross-linking sites within Plugin-C the high molecular mass band observed on SDS-PAGE (Fig. 4) was excised and subjected to in-gel tryptic digestion and tandem mass spectrometry. Mass spectrometry data were input into the MassMatrix (29) search engine, which evaluates the monoisotopic masses of peptides, looking for a mass shift of Ϫ17 Da due to release of ammonia during isopeptide formation. Three specific peaks: m ϭ 2947.5 ([MϩH] 5ϩ , m/z ϭ 590.5), m ϭ 2977.4 ([MϩH] 5ϩ , m/z ϭ 596.5), and m ϭ 3429.7 ([MϩH] 6ϩ , m/z ϭ 572.6) were detected. The first peak corresponds to Plugin peptides 432-445 (m ϭ 1729.9) and 500 -509 (m ϭ 1233.7) minus 17 Da resulting in a monoisotopic mass m ϭ 2946.6. The calculated mass is within error of m ϭ 2947.5 (Ϯ1 atomic mass unit, accuracy Ϯ0.02%). The identity of crosslinked species was confirmed by y and b ion fragments (Fig. 6A). On the basis of the fragments observed and the sequences of two tryptic peptides, we deduce that Lys-433 in peptide 432-445 and either Gln-504 or Gln-506 in peptide 500 -509 are linked by an ⑀-(␥-glutamyl)lysine isopeptide bond. By similar calculations, we found that the second peak (m ϭ 2977.4) corresponds to Plugin peptides 432-445 and 408 -417 with the isopeptide bond formed between Lys-433 and either Gln-411 or Gln-412. The third peak (m ϭ 3429.7) corresponds to Plugin peptides 409 -431 and either 442-445 or 496 -449 with the isopeptide bond formed between Lys-417 and either Gln-442 or Gln-496.
Interestingly, whereas Plugin-C contains 10 lysine residues, two of the three cross-linked peptides detected involve the same lysine residue Lys-433 cross-linked to Gln residues Gln-411 and Gln-506, in the preceding and subsequent ␣-helices, respectively (Fig. 6B). The third cross-link is between Lys-417 and the sequence QIGR, which occurs twice in Plugin-C within the consensus repeat near the N terminus of each predicted ␣-helix. The Gln residues involved in cross-linking are present in a majority of the six ␣-helices within the C-terminal domain of Plugin (Fig. 6C).

DISCUSSION
We have characterized A. gambiae seminal transglutaminase AgTG3 and the cross-linking of its native substrate Plugin in vitro. Intriguingly, we find that AgTG3 forms a dimer in solution, similar to the mammalian blood-clotting enzyme FXIII, yet distinct from other transglutaminases of known structure. AgTG3 has similar (ϳ30%) sequence identity to all of these enzymes. This raises the question: what features of the TG sequence are responsible for dimerization of AgTG3/FXIII versus tissue-type transglutaminases?
This question can be addressed by mapping the dimer interface of the Ca 2ϩ -bound FXIII reported by PISA (36) to the alignment of AgTG3 with TGs of known structure (Fig. 7). For the majority of this interface AgTG3 is no more similar to FXIII than other TGs with two exceptions. First, tissue-type TGs have an extended loop following the second ␣-helix of the core domain (hTG2 200 -206), absent from AgTG3 and FXIII, which would interrupt the FXIII dimer interface. Second, AgTG3/FXIII share a charged residue in the loop prior to the catalytic aspartate within the core domain, the ␤CD loop, at the center of the dimer interface. Specifically FXIII Arg-408 (AgTG3 Lys-417) is replaced by hTG2 Cys-370, which forms a vicinal disulfide with Cys-371 in the active conformation of hTG2 and is implicated in oxidative inhibition of tissue TGs (37).
An important assumption in the analysis above is that both the monomeric structure and dimeric interface of AgTG3 are equivalent to those of FXIII. Further structural studies of AgTG3 are necessary to test this assumption. Hence, a caveat to our analysis is that AgTG3 diverges significantly from all mammalian TGs, including numerous amino acid substitutions and unique extensions, notably in two loops within the ␤-sandwich domain. A high resolution structure of AgTG3 would be required to address the molecular basis of AgTG3 dimerization. Whereas AgTG3 adopts a folded tertiary structure, Plugin-C has an extended or partially unfolded tertiary structure despite clear ␣-helical secondary structure and a propensity for oligomerization or aggregation in solution. Oligomerization of TG substrates is found for both FXIII and seminal TGs. Fibrin ␣C domains preassemble into polymers which are then reinforced by cross-linking with FXIIIa (38,39), and seminal vesicle secretion proteins in rodents form disulfide-bridged high molecular mass complexes prior to cross-linking by the seminal transglutaminase TG4 (40). We suggest that oligomerization of Plugin is mediated by coiled-coil interactions between the repetitive helical regions of the C-terminal domain that are subsequently reinforced by AgTG3 cross-linking. The N-terminal domain of Plugin has a high density of putative cross-linking sites and is expected to form a densely cross-linked polymer, so association of the C-terminal domain may have a significant effect on the density and mechanical properties of the Anopheles mating plug.
The specific activity of AgTG3 is 8 -10 times lower than that of other well characterized TGs. Although this may arise from the use of E. coli to express AgTG3, we note that both FXIII and hTG2 have been expressed in an active form from this host (37,41). From studies of other TGs we propose two explanations. First, AgTG3 includes an N-terminal extension of ϳ20 amino acids that may bind across the dimer interface, as is observed in FXIII. Activation of serum FXIII includes cleavage of this sequence by thrombin: if a similar mechanism held for AgTG3 the purified enzyme may exhibit autoinhibition resulting in suboptimal activity in vitro. Second, hTG2 has been shown to exhibit redox regulation of activity (37). AgTG3 contains a number of cysteine residues that are not conserved with mammalian TGs; despite the lack of evident disulfides in the purified protein these cysteines may be involved in oxidative inhibition of AgTG3.
AgTG3 has higher relative activity toward Plugin-C/FITC-CAD versus Btn-TVVQEL-OH and polylysine-coated plates. Hence, AgTG3 may have some preference for specific sequence motifs in Plugin. Among the cross-linking sites identified is the motif QIGR present in the majority of ␣-helices in the C-terminal domain (Fig. 6C). Its similarity to the motif present in the N-terminal domain of Plugin suggests that QI(G/P) is a preferred motif for cross-linking by AgTG3.
AgTG3 activity toward Plugin-C/FITC-CAD, however, may be due to a preference for (i) polypeptide or ␣-helical versus hexapeptide substrates, or (ii) soluble polyamine versus surface-bound lysine. Further experiments are necessary to discriminate between these hypotheses. Nevertheless, the current signal-to-noise achieved with the nonspecific inhibitor iodacetamide is sufficient for in vitro high throughput screening of prospective AgTG3 inhibitors (ZЈ ϭ 0.84).
The mechanism for formation of the copulatory plug in A. gambiae bears intriguing similarities to the mammalian FIGURE 7. Sequence alignment of AgTG3(22-748) with transglutaminases of known structure: FXIII, hTG2, hTG3, and P. major TG. Secondary structure elements colored and labeled by domain. Identical residues are shaded gray, similar residues are boxed in gray. Catalytic residues are shaded yellow, Ca 2ϩbinding residues (based upon hTG3) are shaded pink. FXIII residues at the dimer interface are shaded green. Numbering corresponds to AgTG3 full-length sequence. Figure was generated with ESPRIPT (42). blood-clotting cascade: a dimeric TG activated by calcium to act upon a specific substrate capable of preassembly of a oligomeric/polymeric state. Given their diverse functions, these similarities appear to be an instance of convergent evolution. Future investigation of AgTG3 will provide insight into the evolution of TGs and advance our basic understanding of fertility as well as the development of potential chemosteriliants for male Anopheles.