A predicted AnCYPOR homology model generated using rat CYPOR as a template (pdb code: 1JA1, [19]) was chosen based on a consensus judgment of discrete optimized protein energy (DOPE) and residue specific all-atom probability discriminatory function (RAPDF) scores. Figure 1 shows an oval, bowl-like 3D structure of the AnCYPOR homolog in the presence of NADP⁺, FAD, and FMN. The overall AnCYPOR structure contains three structural domains including FMN-binding, FAD/NADP⁺-binding, and connecting domains. The FMN, FAD and NADP(H) are positions in the middle of the structure as found in rat CYPOR [10,19]. The model has an overall ProSA-z score of −11.01. Ramachandran plot analysis revealed 88.7% and 0.6% of residues in most favorable and disallowed regions, respectively (Figure S1). Three residues (V256, N503, and E506) residing in disallowed region locate in the connecting domain which functions to bring the two flavins together and modulate electron transfer between them [10,19,20]. The crystal structure of rat CYPOR lacks 63 N-terminal amino acids and starts with Val64, so the corresponding first residue in AnCYPOR is Thr64. Although structure of the membrane-binding region of AnCYPOR could not be determined in the model, sequence alignment indicated a clear difference in membrane-binding sequence from rat CYPOR (Figure S2).

Superimposition of the wildtype- and L86F/L219F-AnCYPOR with rat CYPOR showed topology differences in all domains (Figure 1). Deviations are found in one location in FMN-binding domain (⁷⁷QSSGRR, the connecting loop between helix A and β-sheet 1), one in FAD-binding region (²⁹⁷HKAGGR at the tip of β-sheet 8), one loop in NAD(P)H binding region (⁵⁸⁸GIINLRVA), and ⁴¹¹STAPE at the tip of helix K in connecting domain. These differences may reflect enzyme properties and conformational change during AnCYPOR catalysis compared to rat CYPOR.

Although replacement of two leucine residues successfully increases FMN binding, possibly through altered topology of the loop at ⁷⁷QSSGRR that superimposes well with rat structure than wild-type AnCYPOR, the L86F/L219F-Δ55AnCYPOR remains loosely bound to FAD cofactor. Therefore, a topology change in ²⁹⁷HKAGGR of L86F/L219F from wild-type AnCYPOR to that of rat CYPOR may not affect FAD binding (Figure 1). Relative to that of rat CYPOR structure, the re-side and the si-face of FAD ring are similarly stacked by the indole ring of W678 and the si-face by Y459 in AnCYPOR. The interaction of FAD isoalloxazine ring with side chains of S460, T475, A476 and V474 of AnCYPOR is also similar to rat CYPOR [10]. The rest of the FAD molecule lies at the interface between the FAD-binding domain, but polypeptide chains surrounding FAD show variable degrees of homology with rat CYPOR (Figure 1 and supplemental Figure 2). In addition, residues participating in stabilization of FAD binding in AnCYPOR are notably different from rat CYPOR. Residues observed in rat CYPOR are R424, R454, T491, Y478 and V489 [10], while in AnCYPOR they are R457, T494, Y481 and V492. The rat R424, which is important for interaction with FAD, is missing from AnCYPOR. Instead, a Cys residue was observed at position 427 (Figure S2). The short chain sulfhydryl-group of C427 might drastically abolish interaction with FAD, leading to a loose binding of FAD to the enzyme (Figure 2). Recently, disruption of local H bonding with FAD pyrophosphate moiety leading to weaker FAD binding, unstable protein, and loss of catalytic activity has been reported in V492E and R457H two naturally occurring missense mutations of human CYPOR [21]. Thus we replaced Cys with Arg together with L86F/L219F and generated a soluble triple mutant (L86F/L219F/C427R-Δ55AnCYPOR). The triple mutant could increase FAD binding about 1.5 folds of double mutant L86F/L219F-Δ55AnCYPOR, and about 1.3 folds compared to the wild-type-Δ55AnCYPOR (Table 1). The NADPH-dependent cytochrome c reduction activity in C427R triple mutant without supplementation of FAD was about two and four folds higher than L86F/L219F double mutant and wild-type-Δ55AnCYPORs, respectively (Figure 3). Supplementation with FAD reduced the difference in the cytochrome c reduction activity between the C427R triple mutant and the double mutant. The activity of this triple mutant enzyme is about 0.5 fold of rat CYPOR activity (51.5 μmol/min/mg) [17]. The results suggest a role of C427 in FAD binding and as a consequence increased AnCYPOR catalysis. Supplementation of FAD could further elevate the activity of C427R triple soluble mutant, indicating another factor might also influence FAD binding. The absence of cytochrome c reduction activity in the reaction omitted AnCYPOR enzyme excludes the possibility of an external electron transfer pathway to cytochrome c by exogenous flavins (unreported data). This increase in enzymatic activity of mutant AnCYPORs upon FAD supplement during enzyme catalysis had been reported in Y459A and V492E human mutant CYPOR enzymes that are loosely bound to FAD [22,23]. In addition, the C427R triple mutant had similar K_m values for NADPH and cytochrome c substrates to those of L86F/L219F double mutant, but its catalytic electron transfer rate (V_max) was approximately two folds higher (data not shown). Nonetheless K_m values of L86F/L219F double and C427R triple mutants were not substantially different from wild-type-Δ55AnCYPOR indicating that the mutations did not significantly alter structure or substrate binding mode of enzyme (Table 2). The K_m values for cytochrome c and NADPH of wild-type-Δ55AnCYPOR in this study is noticeably lower than that previously reported [14] due to enzymatic assays performed in this study were in low ionic strength (0.1 M Tris pH 7.5) condition, while in previous report the assays were at high ionic strength (0.3 M Potassium phosphate pH 7.7). Effect of ionic strength on substrate binding of CYPOR has also been reported [11].

Unlike FMN- and FAD-binding regions, there is no substantial alteration among residues in their roles in NADPH binding (R301, S597, R598, K603, Y605, T607, N636 and M637) and catalysis (S459, C631 and D676) in AnCYPOR model structure. However, one minor difference in rotameric conformation from rat CYPOR structure was found in NAD(P)H binding domain at the end of β-sheet 21 (⁶⁷⁸WS), located at the nicotinamide-binding site and covering the isoalloxazine ring of FAD cofactor. This conserved tryptophan (Trp) residue has been proposed to flip and facilitate nicotinamide binding and thus hydride transfer [10,19]. To investigate the contribution of such difference in rotameric conformation in nicotinamide binding and catalysis of AnCYPOR, we replaced W678 with alanine and histidine as both replacements play critical role in catalytic activity and nicotinamide binding of human and rat CYPORs [24,25].

In contrast to C427R construct, the Ala and His replacements at W678, (L86F/L219F/W678A-Δ55AnCYPOR and L86F/L219F/W678H-Δ55AnCYPOR) neither increased FAD binding nor cytochrome c reduction activity (Table 1 and Figure 3). However, cytochrome c reduction activity was decreased compared to L86F/L219F double mutant and C427R triple mutant. Mutations with aliphatic (W678A) and planar (W678H) residue substitutions decreased catalytic activity to 77% and 55% of that of L86F/L219F double mutant, respectively. The decrease in cytochrome c activity has been reported in the W676A and W676H human CYPOR mutants that retain 5 and 22% of cytochrome c activity, respectively [24,25]. Moreover, non-linear regression steady-state kinetic analysis using cytochrome c as an electron acceptor in this study revealed that removal of bulky Trp residue in W678A and W678H increased NADPH binding affinity by two- to three-fold compared to wild-type and L86F/L219F mutant AnCYPOR enzymes, but not that to NADH (Table 2). This differs from human CYPOR mutants in which both W676A and W676H mutants lower both NADPH K_m, and NADH K_m[24,25]. Specific charge interactions of amino acid residues with 2′-phosphate of NADPH are responsible for discrimination against binding of NADH [24]. Thus, it is conceivable that W678 is only crucial for catalysis and NADPH binding in AnCYPOR, and not nicotinamde selectivity of NAD(P)H and NADH substrates. Whether deviation we observed at ⁵⁸⁸GIINLRVA (Figure 1) might involve in 2′-phosphate binding of NADPH is not known. This deviation might explain low binding affinity of A. minimus CYPOR to 2′,5′-ADP column compared to human and rat CYPOR [18,26]. Values of 2′AMP K_i for AnCYPOR are four-fold higher than those for rat CYPOR, suggesting significant differences in NAD(P)H binding property between rat CYPOR and AnCYPOR [14,15]. Low binding affinity to 2′,5′ADP column has also been observed for the mosquito AgCYPOR [18]. A two-fold higher IC₅₀ in suppression of AgCYPOR by 2′AMP has also been reported compared to human CYPOR [18]. It is thus speculated that W678 could affect electron transfer rate and electron transfer activity, while dramatic change in polypeptide chain that specifically interact with NAD(P)H such as that of ⁵⁸⁸GIINLRVA might result in low nucleotide (2′,5′-ADP, 2′-AMP and NAD(P)H) binding affinity. However, the amino acids in the ⁵⁸⁸GIINLRVA polypeptide chain are distant from those of rat CYPOR.

Further mutational investigation of this polypeptide on nicotinamide selectivity may provide important information that could help to understand the catalytic basis of AnCYPOR.

As P450 catalysis requires electron supplement from CYPOR, therefore, such a mutation that affects cytochrome c reduction activity could affect the P450-mediated oxygenase reaction in vitro[27]. This is supported by an increased efficiency of L86F/L219F-flAnCYPOR and L86F/L219F/C427R-flAnCYPOR mutants in supporting BROD mediated by mosquito CYP6AA3 enzyme by 2.4 and 3 times, respectively compared to wild-type (Table 3). Moreover rat CYPOR which has higher cytochrome c reduction activity than wild-type AnCYPOR and all AnCYPOR mutants could increase mosquito P450 activity more than wild-type flAnCYPOR by 5.5 times. These increases in velocity by L86F/L219F, C427R mutants, and rat CYPOR had no effect on substrate binding value (K_m). The results suggest that in vitro mosquito P450 activity is an electron transfer rate dependent reaction. In human CYPOR, the loss of FAD and FMN from its binding site caused by mutation is the major cause of diminished P450 activity. For example, the R457H and V492E mutations at the FAD-binding site in human CYPOR result in an impaired function of CYP17A1 (17a-hydroxylase/17, 20-lyase), CYP19A1 (aromatase) and CYP1A2 in vitro[23,28–30]. Recently, a study on heme oxygenase I (HO-I) activity also indicated a decrease in CYPOR activity affected HO-I catalytic rate, and rate of bilirubin formation by HO-I enzyme is CYPOR-activity dependent [31]. Detailed saturated kinetic studies reveal that a decrease in HO-I activity is caused by a decrease in catalytic rate (V_max), not substrate binding (K_m) of HO-I [31]. Thus, we hypothesize the high catalytic activity of the mosquito CYP6AA3-mediated enzymatic activity found in the present study originates from a high electron transfer rate of rat CYPOR, C427R and L86F/L219F-flAnCYPOR activity compared to that of wildtype-flAnCYPOR enzyme. This, to our knowledge, is the first mosquito P450-mammalian CYPOR reconstitution that is more efficient than the native mosquito P450-CYPOR complex.

Flavin mono-nucleotide (FMN), flavin-adenosine di-nucleotide (FAD), cytochrome c, nicotinamide adenosine diphosphate (NADP⁺), nicotinamide adenosine diphosphate reduced form (NADPH) and phenylmethylsulphonyl fluoride (PMSF) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Isopropyl-β-d-thiogalactopyranoside (IPTG) was obtained from USB (Cleveland, OH, USA), Ni²⁺-NTA affinity column from Qiagen (Valencia, CA, USA), and Bio-Rad protein assay kit from Bio-Rad (Hercules, CA, USA). Quickchange site-directed Polymerase Chain Reaction (PCR) mutagenesis kit was purchased from Stratagene (LaJolla, CA, USA) and was used according to the manufacturer’s instructions.

The amino acid sequence of wild-type A. minimus CYPOR was obtained from GenBank (ABL75156.1) and was used to search for template using PSI-BLAST against protein structures deposited in Brookhaven Protein Data Bank [32] for comparative modeling. Rat CYPOR-triple mutant (PDB:1JA1, [19]), exhibiting high resolved power of 1.80 Å and sharing 55% sequence identity to A. minimus CYPOR, was selected as the template. The first 63 residues were omitted from modeling since the N-terminus is associated with membrane anchor and membrane portion is absent in the template structure. Sequence alignment derived from ClustalW program of various CYPOR enzymes with some manual adjustment is shown in supplemental Figure 2. Alignments of AnCYPOR wild-type and mutants against the template were subsequently subjected to homology modeling using MODELLER9v6 [33]. A set of 1000 models for wild-type and mutant AnCYPORs was independently constructed. The cofactor FAD, FMN, and NADPH were positioned in target models as in the 1JA1 template. The most promising models of wild-type and mutant CYPORs were discriminated from incorrectly-folded structures using DOPE [34] and RAPDF [35] scoring functions. Candidate models were further energy minimized using AMBER ff03 all atom force field implemented in Amber10 [36] to remove bad van der Waals contacts. Model refinement was performed using steepest descent (SD) for 3000 iterations and followed by conjugated gradient (CG) minimizations until the energy gradient was less than 0.05 kcal/mol. The final energy minimized models were evaluated for conformation qualities using ProSAII [37,38] and Procheck [39]. Models were visualized and displayed using PyMOL (Schrödinger LLC, NY, USA).

Three amino acids were separately introduced into pET28a-L86F/L219FΔ55AnCYPOR plasmid DNA by Quickchange PCR mutagenesis kit as described in manufacturer’s instructions to generate triple mutants of the following substitutions: C427R, W678H and W678A. The sequences of primers used in this experiment are listed in Table 4. The same primer set was employed to generate C427R membrane-bound triple mutant using pTrc-L86F/L219F-flAnCYPOR plasmid DNA as template. All of the mutations were verified by DNA sequencing and the resultant plasmids were transformed into BL21 (DE3) E. coli cells.

Protein expression and purification of AnCYPORs were performed as previously described [14,15], while that of rat CYPOR followed Shen et al. (1989; [17]) with a slight modification. The E. coli C43 (DE3) carrying pIN-rat CYPOR plasmid was grown at 37 °C in TB broth containing 100 μg/mL ampicillin until OD⁶⁰⁰ was about 0.8 and protein expression induced by addition of 0.5 mM IPTG. Cells were allowed to grow for an additional 48 h and harvested by centrifugation at 5000× g for 15 min and resuspended in binding buffer (50 mM Tris pH 7.7, 0.1 M NaCl, 10% glycerol, and 20 mM imidazole) containing 0.2% Triton X-100. Cells were lysed by sonication and the supernatant was applied to a Ni²⁺-NTA affinity column previously equilibrated with the binding buffer. The column was extensively washed with binding buffer followed by the same buffer containing 30 mM imidazole. The protein was eluted by increasing imidazole concentration to 100 mM. Purity of protein was assessed by SDS-PAGE. Pure fractions were pooled, concentrated and stored at −80 °C until use. Protein concentration was determined by the Bio-Rad protein assay using bovine serum albumin (BSA) as standard.

Recombinant baculovirus containing CYP6AA3 cDNA was used for protein expression in Spodoptera frugiperda (Sf9) insect cells as previously described [7]. Briefly, Sf9 cells were infected with the CYP6AA3 expressed virus (2.5 × 10⁸ plaque-forming units/mL) at the multiplicities of infection of 3. Infected cells were harvested at 70–80 h post infection and resuspended in sodium phosphate buffer pH 7.2 containing 1 mM EDTA, 0.5 mM PMSF, 5 μg/mL leupeptin, 0.1 mM DTT, and 20% glycerol, and subjected to microsome preparation using differential centrifugation. CYP6AA3 protein expression was observed by SDS-PAGE analysis. Total P450 content was measured from CO-different spectrum analysis [40].

Commercial FMN and FAD were further purified by high performance liquid chromatography (HPLC) and used for generation of a standard curve and for cofactor supplementation experiments. Concentrations of standard FAD and FMN solutions were determined spectrophotometrically at 450 nm using extinction coefficients of 11.3 and 12.2 mM⁻¹ cm⁻¹, respectively. FAD and FMN contents of each sample were measured using a fluorometric method as described previously [41]. The Bio-Rad protein assay was utilized to determine concentration of protein using rat CYPOR as standard. The rat CYPOR protein concentration was determined using spectrophotometric method [42]. The CYPOR-mediated cytochrome c reduction was carried out in 0.1 M Tris-HCl buffer, pH 7.5 as previously described [7] with minor modifications. After 1 min pre-incubation of enzyme in buffer with 40 μM cytochrome c at 25 °C, the reaction was initiated by addition of 50 μM NADPH. NADPH-dependent cytochrome c reduction was followed by a change in absorbance at 550 nm. Velocities are expressed as μmol/min/mg protein. One unit of enzyme was defined as the amount of enzyme catalyzing the reduction of 3 μmol of cytochrome c per minute under described conditions [43]. Substrate saturation steady-state kinetic studies of cytochrome c reduction were performed in 0.1 M Tris-Cl buffer, pH 7.5 at 25 °C in a final volume of 0.75 mL. Substrate saturation experiments were performed by varying NADPH concentration (2, 5, 7.5, 15, 25, 50 and 100 μM) and NADH concentration (0.3, 0.5, 1, 2, 5 and 10 mM). The reaction was started by addition of enzyme and the initial velocity data were analyzed by non-linear regression using the Grafit 6.0 software package.

CYP6AA3-Mediated benzyloxyresorufin O-dealkylation reaction (BROD) was performed in 50 mM Tris-HCl buffer pH 7.5, in a total volume of 500 μL. Microsomes containing CYP6AA3 (~25 pmole) were used and enzymatic assays were reconstituted with either the purified AnCYPOR or rat CYPOR in the ratio of 3:1 and carried out with BR substrate (0.5, 1, 2, 4, 8 μM) as described [7]. Enzyme activity was measured with a RF-5301 PC spectrofluorophotometer (Shimadzu, Kyoto, Japan) at λ_ex = 530 and λ_em = 590 nm. The amount of resorufin product was calculated referring to the resorufin standard curve as described in Duangkaew et al., 2011 [9]. Apparent K_m and V_max values were estimated by non-linear regression analysis using GraphPad Prism 5 software package.