2.2 Modeling Aedes aegypti JHAMT

the molecular model of AeJHAMT was built by homology modeling using Modeller9v4 (Eswar et al., 2008; Sali and Blundell, 1993). The O-methyltransferases from Anabaena variabilis (PDB ID: 3CCF) and Clarkia breweri (PDB ID: 1M6E) (Zubieta et al., 2003) were chosen as templates on the basis that they had the highest identity to AeJHAMT (20.6% and 16.9%, respectively) among methyl transferases that use a similar catalytic mechanism to transfer the methyl group to a carboxylic acid moiety in their substrates, including the absence of a requirement for a metal cation (e.g. Fe) in the active site for enzymatic activity (Mayoral et al., 2009b). The alignment of the three sequences was done with T-COFFEE (Poirot et al., 2003). Supplementary Figure 1A shows the alignment between the sequences of the two selected templates and the modeled AeJHAMT. We manually corrected the alignment of AeJHAMT to the templates based on the PSIPRED predictions of the template secondary structures (McGuffin et al., 2000). We added constraints to the amino acid residues that interact with SAM when we generated the model, since these key residues in Clarkia are conserved in mosquitoes. In addition, we also checked the secondary structure prediction and FOLD assignments made by the program PHYRE, which is a valuable tool for template selection. In the case of AeJHAMT, the secondary structure predictions were in good agreement with the 3CCF and 1M6E secondary structures. Supplementary Figure 2 displays the structural alignment between the overall folds of the two selected templates and AeJHAMT. The models are colored showing the Root Mean Standard Deviation (RMSD) of the structurally aligned parts (Leach, 2001).

The model with the lowest Discrete Optimized Protein Energy (DOPE) assessment score was subjected to loop refinement using the “loop module” of Modeller9v4. The SCRWL program was then used to regenerate side chains based on a backbone-dependent rotamer library (Bower et al., 1997). The model was then minimized using the AMBER program (Case et al., 2005), with implicit solvent and the ff99sb force field (Hornak et al., 2006) and 10 kcal/molŲ restraints on the protein atoms. The final structure was evaluated using WHAT_CHECK (Hooft et al., 1996; the WHAT_CHECK analysis is included as Supplementary Information).

We evaluated an additional Anabaena variabilis SAM-dependent methyltransferase (PDB ID: 3GGD) with 17% sequence identity to AeJHAMT. This protein was not selected as a template because its secondary structure is very different to that predicted for AeJHAMT and critical residues for substrate interaction, such as Q14 and W120, are missing (Supplementary Figure 1B).

2.3 Modeling A. gambiae, B. mori, D. melanogaster and T. castaneum JHAMTs

To build additional insect JHAMT models, we had the choice of using the already optimized AeJHAMT model as template, or start all over using the original Clarkia and Anabaena structures. We chose the former option because the sequence identities between Clarkia and Anabaena and the additional insect JHAMTs to be modeled were often lower than those with AeJHAMT. Using them as template would have added uncertainty to the new models. In contrast, the amino acid identities between AeJHAMT and these four proteins varied between 62.6% (A. gambiae) and 40.6% (T. castaneum); furthermore, the important residues for catalysis and cofactor binding are well conserved.

Using the Modeller9v4 program and the AeJHAMT homology model as template we generated models for A. gambiae, B. mori, D. melanogaster and T. castaneum JHAMTs (Eswar et al., 2008). The alignment of the sequences was performed with ClustalX 2.0.12. (Larkin et al., 2007). The best models were chosen using the DOPE lowest score. Structural alignment and visualization were done with the program VMD (Eargle et al., 2006; Humphrey et al., 1996). Images were rendered with Tachyon (Stone, 1998).

3.2 Docking of potential substrates

Figures 3A and 3B show the lowest energy conformers for (10R)-JHA and FA, respectively, where the carboxyl group of the substrates are oriented towards the methyl group of SAM, allowing the catalytic transfer. The substrate carboxyl groups of (10R)-JHA and FA fit nicely to form hydrogen bonds with the Trp-120 indole nitrogen and the amide nitrogen of Gln-14, while their carbon chains reside in an hydrophobic pocket formed by Ile-151, Ile-154, Tyr 155, Leu-158, Val-221 and Val-224 (Fig. 4A). Docking results indicated a clear difference in the interaction of (10R)-JHA and (10S)-JHA with the hydrophobic pocket of the binding site of AeJHAMT. Specifically, while the epoxide ring of (10S)-JHA is facing the hydrophobic residues Ile-151 and Ile-154, the (10R)-JHA epoxide ring is facing the opposite side of the pocket, where it could interact more favorably with Ser-176 and Tyr-178 (Fig. 4B). The potential formation of hydrogen bonds between the OH groups of Ser-176 and Tyr-178 and the epoxide group of (10R)-JHA (Fig. 4A, black dotted line) may result in a higher binding affinity of this molecule for the enzyme. This may explain why 10R and not 10S is the substrate of choice for the recombinant AeJHAMT (Mayoral et al., 2009b).

Open in a separate window

Figure 4

Epoxide recognition pocket

(A) Close-up showing the docking of JHA(10R). The residues Ser-176 and Tyr-178 are shown with the surface colored in green, as they are both part of the pocket and have an active role in the recognition. Potential hydrogen bonds are shown in black dotted lines. The cover of the binding pocket, with the surface of the residues colored in white, is formed by the residues Ile-151, Ile-154, Leu-158, and Val-221. (B) Close-up showing the docking of JHA(10S) with the epoxide group oriented toward the hydrophobic face of the pocket (Ile-151 and Ile-154) instead of toward Ser-176 and Tyr-178.

Docking simulations of palmitic acid in AeJHAMT failed to find any favorable conformation of the acidic moiety facing SAM and interacting with Gln-14 or Trp-120, probably due to the large tail of palmitic acid that did not fit well into the hydrophobic pocket. On the other hand, we found a conformer for lauric acid that could interact with Gln-14 and Trp-120 but the tail was outside the hydrophobic pocket. Moreover, the energy due to internal constraints of lauric acid was high. Thus, the predicted binding affinity of lauric acid was low compared to those assessed for JHA and FA. These results may explain why AeJHMAT is not able to catalyze the methylation of these two substrates (Mayoral et al., 2009b).

4.2 JHAMT esterifies FA and JHA, confirmation by studies using recombinant enzymes

Shinoda and Itoyama (2003) were the first to describe the enzymatic properties of an insect recombinant JHAMT. In the presence of SAM, this enzyme from B. mori efficiently catalyzed the conversion of FA and JH acids I, II, and III to their cognate methyl esters, although it displayed greater activity towards JHA I and JHA II than JHA III or FA (Shinoda and Itoyama, 2003). Later, studies on the activity of three additional recombinant JHAMTs were published, two from Diptera (Niwa et al., 2008; Mayoral et al., 2009b) and one from Coleoptera (Minakuchi, et al., 2008). All of three enzymes efficiently esterified FA and JHA III (Table 2). In addition, all of them showed enantioselectivity favoring (10R)-JHA III, generating (10S)- and (10R)-JH III isomers at ratios of 2:98 (B. mori), 0.5:99.5 (A. aegypti), 13:87 (T. castaneum) and 20:80 (D. melanogaster), respectively.

4.3 Structural studies validate that insect’s JHAMTs esterify FA and JHA

the identification of critical active site residues through modeling and docking studies offers a structure-based approach to help both understand the existing JHAMT experimental results and to predict the substrate specificity of untested enzymes of this family. Our aim was to understand the principles of substrate recognition and the transmethylation reaction in insect JHAMTs. Biochemical studies with insect recombinant proteins provided a framework for understanding the catalytic mechanism that underlies substrate recognition and methyl group transfer.

In our modeling and docking studies with AeJHAMT we identified several critical amino acid residues involved in the interactions of the enzyme with the methyl donor (SAM) and with the substrates (FA and JHA). The comparison with the additional three JHAMTs that have been biochemically characterized confirmed that the modeled active sites look very similar among them and showed that all these critical residues are conserved among the four tested proteins, or have substitutions that do not affect the ability of the active site to methylate FA or JHA (Table 2). We predict that A. gambiae, C. pipiens, S. ricini and the nine additional Drosophila species analyzed should be able to utilize both FA and JHA as substrates because all the critical residues are conserved in identity or function with the residues of the four JHAMTs that have been experimentally tested. The same prediction applies to the H. armigera protein, which that has two semi-conservative changes in residues that form the hydrophobic pocket for the carbon chain that should not affect the enzyme specificity.

In contrast, the two hymenopteran JHAMTs present a series of non-conserved substitutions. In A. mellifera and N. vitripennis Ser-176 is replaced by Pro, and Tyr-178 has a conserved substitution by Phe. These two changes suggest that Apis and Nasonia JHAMTs are not able to form a hydrogen bond with (10R)-JHA. In addition, the replacement in these two insects of Ile-151 for a smaller Gly, implies an enlargement of the hydrophobic pocket that could further change the specificity. Longer substrates, such as palmitic or lauric acids, might now be able to fit inside the hydrophobic pocket. Nevertheless, it is important to emphasize that these two hymenopteran proteins should still be able to utilize both FA and JHA as substrates because they conserve the SAM binding pocket, the Trp and Gln residues to recognize the carboxyl moiety of the substrates and a hydrophobic pocket that is able to accommodate FA and JHA.

We also found JHAMT orthologues in two hemimetabola, the hemipterans Acyrthosiphon pisum (pea aphid) and Rhodnius prolixus (kissing bug). Critical residues such as Asp-41, Asp-69, Gln-14 and Trp-120 are conserved. There are also conservative, semi- conservative and non-conservative substitutions (Table 1). The JHAMT sequence in R. prolixus was obtained from the genome project currently in progress; therefore it is a partial sequence that might have some errors (Supplementary Fig. 4). Nevertheless, the two hemimetabola sequences show a high degree of conservation that suggests similar substrate specificity.

We are aware that the structural models done by homology modeling in these studies are based on templates with low homology with AeJHAMT, and consequently, care must be taken in the conclusions derived from them. In the absence of a crystal structure of an insect JHAMT, homology modeling based on proteins with a sequence identity that falls within or below what Chung and Subbiah defined as the twilight zone of 20–25% could be challenging (Chung and Subbiah, 1996). However in this case, the robustness of the AeJHAMT model is supported by the congruence of the docking and biochemical results, both in the substrate specificity and stereospecificity, as well as by the high degree of conservation of the critical amino acids from the five different insect JHAMTs modeled. In summary, even though discrepancy in low conservation zones could be expected, we are confident that the active site region involved in ligand and cofactor binding has been properly characterized.

4.4 JH epoxidase shows narrow substrate specificity

Locusta migratoria JH epoxidase is a microsomal enzyme that specifically converts MF into (10R)-JH III (Feyereisen et al., 1981). Experiments using the cockroach Blaberus giganteus CA and CA homogenates showed that the JH epoxidase is more active on MF than on FA, and it showed high specificity for the 10,11 double bond (Hammock, 1975).

Diploptera punctata recombinant JH epoxidase (CYP15A1) converts methyl (2E,6E)-farnesoate to the natural (10R)-JH III with a very high degree of specificity (98% 10R: 2% 10S). Other terpenoids were not metabolized by CYP15A1, including farnesol, farnesal, FA, JH III, 3,7,11-trimethyl-dodecanol, geranylgeraniol, farnesyl methyl ether and geraniol, as well as the fatty acids lauric acid and arachidonic acid (Helvig et al., 2004). An NADPH-dependent JH epoxidase acting on FA was described from the CA of M. sexta (Reibstein et al, 1976). This enzyme did not efficiently epoxidize MF in the conditions in which the experiments were carried out; the authors argued that FA could be converted to JH III through both branches of the pathway and suggested that Lepidoptera have a different sequence of reactions than cockroaches and locusts (Reibstein et al, 1976). Preliminary experiments using CYP15C1, a recombinant JH epoxidase from B. mori, showed a high stereoselective epoxidation of FA (Shinoda et al., 2007), suggesting that in Lepidoptera substrate specificity of the epoxidase may be the critical factor in determining epoxidation/methylation reaction order. Nevertheless the higher catalytic activity of B. mori JHAMT towards JHA I and JHA II when compared to JHA III or FA might also contribute to the reaction order observed in Lepidoptera (Shinoda and Itoyama, 2003).

We thank two anonymous reviewers for comments that have significantly improved this manuscript. This work was supported by NIH grant AI 45545 to FGN. The authors acknowledge the NSF Teragrid resource through Grant No. TG-MCA05S010 for providing computational resources and support that have contributed to the results reported in this paper. This work was partially supported by grants from Universidad de Buenos Aires 08-X499 to AGT, ANPCYT 06-5203 to AGT and Conicet PIP 01207 to AGT .