The sequence homology of Cyt3Aa1 was determined using the NCBI protein-protein BLAST program (http://www.ncbi.nlm.nih.gov/BLAST). The alignment of the Cyt3Aa1, Cyt1Aa1 and Cyt2Aa1 sequences was performed using Clustal W software (European Bioinformatics Institute, http://www.ebi.ac.uk/clustalw/). Phylogenetic analysis of the Cyt proteins was performed using MEGA 6.0 software [11].

The cyt3Aa1 gene was ligated into the cloning vector pGEM-T (Promega), which was digested with Kpn I and Sal I and inserted into the same sites in the pET32a expression vector (Novagen) to obtain a recombinant plasmid (pET-Cyt3A). Subsequently, pET-Cyt3A was transformed into E. coli Arctic Express cells (Zoon Bio). The expression of the cyt gene was assessed as described by Cohen et al. [2]. The protein preparation was analyzed by SDS-PAGE as described by Green et al. [12].

The activity of Cyt3Aa1 was tested against larvae of A. aegypti (Diptera) as previously described [13]. The hemolytic activity of the Cyt3Aa1 protein was determined using human red blood cells as described by Ragni et al. and Juárez-Pérez et al. [14] [15]. Protein was added at six different concentrations (0.1 to 50 μg/ml), and the hemolytic rate (%) was examined after 4 h. Each bioassay was repeated at least five times, the 50% lethal concentrations were determined by Probit analysis, and the hemolytic rate of Cyt3Aa1 was estimated as follows: hemolytic rate (%) = (supernatant Hb * [1 − Hct%])/total Hb * 100.

An alignment of Cyt3Aa1 and Cyt2Aa1 was produced using the Clustal W program, and the alignment was submitted to Swiss-Model (http://www.expasy.ch/spdbv/). The preliminary model of Cyt3Aa1 was predicted using the server SWISS-MODEL (http://swissmodel.expasy.org/) [16].

The minimized model structure of Cyt3Aa1 was then determined to perform energy minimization using the PMEMD module of the AMBER 14.0 software to eliminate unreasonable contacts. Additionally, the minimized model was assessed and validated using the PROCHECK validation server for molecular dynamics simulations [17]. Using the structure of Cyt3Aa1 lacking unfavorable contacts, the entire system was gradually heated to 310 K using Langevin dynamics. A 10-Å cutoff was used for the non-bonded interactions. The Particle Mesh Ewald (PME) method was used to treat long-range electrostatic interactions. In addition, the ProSA program (https://prosa.services.came.sbg.ac.at/prosa.php) was used to evaluate the computational model.

The gene cyt3Aa1 (GenBank accession: HM596591) cloned from BtTD516 encoded a 259-amino acid protein with a molecular mass of 29.5 kDa. The currently known subfamily members of Cyt2-type and Cyt3Aa1 share a high level of sequence identity (Figure 1), the figure of which was created using ESPript [18], and Cyt3Aa1 has a maximum 39.04% amino acid sequence homology with Cyt2Aa.

Phylogenetic analysis of Cyt proteins was conducted using MEGA 6.0 [11]. Most of the known Cyt proteins are specific against the larvae of Diptera insects,

such as Cyt1A, Cyt2B, and Cyt2Aa [2] [7] [8]. The phylogenetic analysis of the three families of Cyt proteins showed that they may have evolved from a common ancestor (Figure 2), suggesting that Cyt3Aa1 may have potential mosquitocidal activity.

The SDS-PAGE results showed that cry3Aa1 was highly expressed in E. coli as inclusion bodies (47-kDa protein as well as the approximately 19-kDa Trx-tagin pET32a) (Figure 3). The approximately 30-kDa protein purified (Cyt3Aa1) was used to assay the biological activity against A. aegypti and human red blood cells. The bioassay results showed that Cyt3Aa1 was not toxic to A. aegypti, but causing a teratogenic effect, which appears to be a new way for A. aegypti control. In contrast to Cyt1Aa and Cyt2Ba, the hemolytic activity of Cyt3Aa1 was very weak (0.0024% - 0.034%) when it was solubilized, a finding that was similar

to the hemolytic activity of Cyt2Bc2 [15].

In the absence of an experimentally determined structure, computational homology modeling has an added advantage due to the use of template information from known biological samples [19]. In this study, 1CBY (PDB Code of Cyt2Aa1) was employed to be the most suitable template for homology modeling of Cyt3Aa1, which has 43.13% similarity to Cyt3Aa1 [20], and the structural models of the Cyt3Aa1 toxic core were obtained comprising 223 of the 259 amino acid primary structure.

The minimized model structure of Cyt3Aa1 constructed by the PMEMD module of AMBER 14.0 was assessed and validated using the PROCHECK [17]. The result showed that all the residues were found in the allowed region (Figure 4),

which indicated that the Cyt3Aa1 model was of good quality and considered fit for molecular dynamics simulations. Subsequently, the model was optimized by the force field FF99SB module of AMBER 14.0. Then the final frame structure of Cyt3Aa1 was simulated and generated by PyMOL (http://www.pymol.org/) (Figure 5). The most covalent/non-covalent bonds, bond angles, and torsion angles of the structure have the expected values for a naturally folded protein assessed by Protein Structure Evaluation Suite and Server (http://www.prosess.ca/index.php). Additionally, the z-score (−2.82) of the model was within the range of similar protein model derived by NMR and X-ray crystallography evaluated using ProSA program (https://prosa.services.came.sbg.ac.at/prosa.php) [21], indicating that the model was in the native conformation (Figure 6). All these results indicated that the generated model was valid with good stereochemical quality.

The structure of Cyt3Aa1 resembles the crystal structure of Cyt2Aa1 [3], which also has a single α/β domain (Figure 5). The β-sheet of Cyt3Aa1 consists of four main β-strands arranged in space as β2↓, β6↑, β5↓, and β4↑ and was flanked by two α-helical layers (α2 - α3 on one side, and α1 - α4 on the other). The mechanism of action of Cyt1Aa1, Cyt2Ba1, Cyt2Aa1, and VVA2 (PDB code:

1VCY) is thought to be that the two surrounding α-helical layers swing away from the β-sheet upon membrane contact to expose the hydrophobic β-sheet, which can insert into the membrane to form a β-barrel pore, resulting in colloid osmotic lysis [22] [23]. The Cyt3Aa1 also has two α-helical layers with considerable variation that surround the β-sheets, and all β-sheets adopt a cytolysin fold, which revealed that Cyt3Aa1 toxin might be a membrane-perforation toxin.

Cyt1Aa could synergize the activities of Cry11Aa to A.aegypti larvae [24]. The three single residues (K198, E204, and K225) of two binding epitopes (₁₉₆EIKVSAVKE₂₀₄ and ₂₂₀NIQSLKFAQ₂₂₈) of Cyt1Aa were shown to be mainly involved in the interaction between Cyt1Aa1 and Cry11Aa, Cry4Ba, respectively [25] [26]. Interestingly, the three single residues are charged in most of the Cyt family members, and the corresponding residues of Cyt3Aa1 (P198, L204, and A225) are also charged. Mapping of the three charged residues on the Cyt3Aa1 structure revealed that the residues are exposed to the surface of the toxin (Figure 7), which might introduce binding sites and induce synergism with Cry protein.