Computer Design of Novel Pesticides Against the Olive Fruit Fly (Bactrocera oleae)

The olive fruit fly Bactrocera oleae targets the fruits of olive trees in the Mediterranean and has been causing increasing economic harm to Californian growers. Past efforts to combat Bactrocera oleae by limiting the spread of its population have included traps, bait sprays, barrier films, and the release of natural predators, but these methods do little against large populations and many fruit fly generations are now resistant to the chemical compounds used. This study attempts to improve the current pesticides targeting the ecdysone receptor of Bactrocera oleae by comparing current pesticides and then using computer software to build upon them. After several generations of molecules were docked in the enzyme 1R20 (ecdysone receptor for a similar insect, Heliothis virenscens), the novel molecule “icon11conf1c0m2” was found to have a significantly better docking score, based on structural matches and hydrogen bonding, than the best original pesticide RH-2485. If further research is done docking “icon11conf1c0m2” into other insect ecdysone receptors, it could have the potential to serve as another more efficient Insect Growth Regulator. Although the enzyme for Bactrocera oleae was not accessible, the pesticides found by this study can be tailored to it when it becomes available. Introduction The state of California is currently the largest producer of domestic olive oil, and production is only growing. While most olive oil consumed in the United States is imported, there is increasing awareness of locally pressed oil, which is less likely to have been adulterated with cheaper refined olive oils.1,2 California’s olive oil industry, however, is now sharing the same pest that has been affecting European olive trees since the third century BC.3 The olive fruit fly (Bactrocera oleae) was discovered in Los Angeles in 1998, and has since spread throughout groves in California.4 Bactrocera oleae is devastating to both table olive and olive oil production. The standards for table olives produced in California have zero tolerance for infestation, so 100% of the crop will be destroyed if a single olive is shown to have fruit fly damage. Because the burrowing of larvae can introduce fungal and bacterial infections inside the fruit and drastically reduce the yield, up to 80% of the olive oil value can be lost.3,4 If the spread of Bactrocera oleae continues unchecked, the Californian olive industry will become increasingly unstable. Currently, most methods to control the olive fruit fly include various traps and sprays to prevent Bactrocera oleae from reaching the olive fruit.5 However, some substances being used as baits or toxins are no longer effective against new resistant generations of Bactrocera oleae. Another proposal to control fruit fly infestation Computer Design of Novel Pesticides Against the Olive Fruit Fly (Bactrocera oleae) Irene Lin1*, Jill Tucker2, Ryan Pemberton3, Dean J. Tantillo3, Sharlene Chin3, and Selina Wang3 Student1, Teacher2: Northview High School, 10625 Parsons Rd. Johns Creek, GA 30097 Mentor3: UC Davis, One Shields Ave. Davis, CA 95616 *Corresponding Author: [email protected] INTERNSHIP ARTICLE is one in which natural predators of Bactrocera oleae from SubSaharan Africa are imported to olive groves in California to prey on the flies.6 This method, however, is at once extremely costly and impractical for large infestations. The latest research on the control of Bactrocera oleae focuses on a receptor that controls the life cycle of the fly population. Insect Growth Regulators work by binding to an ecdysone receptor and causing premature molting in larvae as well as decreased egg production. So far, the best examples of these have been tebufenozide (RH-5992), methoxyfenozide (RH-2485), and RH-5849. None of these, however, are perfect matches for the ecdysone receptor of Bactrocera oleae.7 The purpose of this study, therefore, is to use computer software to build and test molecules that will bind more effectively to the ecdysone receptor 25 Materials and Methods This study was performed entirely on software from the Gaussian and OpenEye suites. The hosts of programs allowed molecules to be built and then tested for structural and electrostatic matches with the enzyme as well as toxicity to humans and animals. Before any matching could be done, however, the enzyme used for comparison of pesticides was found. The PDB library was searched for ecdysone receptors, and the closest match based on the structure of the binding site was found in the ligand for the tobacco budworm Heliothis virescens (PDB ID 1R20). This enzyme was then run through the program Fred Receptor to generate the active site (Figure 1) that all molecules were to be Figure 1. The active site generated by OpenEye Fred Receptor for enzyme 1R20 is shown. This active site was used to dock all the pesticide candidates in Fred to generate the structural score. Irene Lin, Jill Tucker, Ryan Pemberton, Dean J. Tantillo, Sharlene Chin, and Selina Wang Page 2 of 5 26 Results When RH-5992, RH-2485, and RH-5849 were compared for structural and electrostatic similarities with the active site of enzyme 1R20, RH-2485 was found to be the best match with a score of -105. Finally, the “icon11conf1clu0mem2” Brood-generated molecule was found to fit the ecdysone receptor 1R20 the most closely with a score of -130. Many Fred-docking and Brood-generating cycles were performed to produce 11 “icon” pesticides, the progression of which is shown in Figure 3. All of these molecules are named first for the number of the “icon” design drawn, then the conformer number “conf ” as generated by OpenEye Omega2, then the labeling of the changes by OpenEye Brood. For example, the final “icon11conf1c0m2” molecule was the Brood bioisostere labeled “clu0mem2” of the first conformer of the 11th “icon” design. The first pesticide shown in Figure 3, “icon1”, was created by adding a hydrogen bond acceptor to one of the benzene rings of RH-2485. The chain of this acceptor was then lengthened in “icon2” in an attempt to close the distance to the nearest hydrogen bond donor. When “icon2” failed to show any significant improvement in score, the Brood program was run on “icon1”. The top result “icon1conf1c9m2”, however, had an electrophilic benzylic chloride (C-Cl) bond, whose high reactivity made it an unstable candidate for a pesticide despite the more negative score. The second result, “icon1conf1c0m18”, made less reactive changes to “icon1” by removing a methoxy group and adding methyl and methoxymethyl groups. The methyl group was then moved to the opposite side of the benzene ring to make “icon9”, which showed a slightly worse docking score. At this point, a planar cavity in the 1R20 active site was observed near one of the benzene rings, so another benzene ring was attached that included a nitrogen atom to act as a hydrogen bond acceptor for “icon10”. Figure 2. A screenshot from OpenEye VIDA that displays the pesticide methoxyfenozide RH-2485 along with the hydrogen bonds (yellow) that it forms with 1R20’s nearby amino acid sidechains, along with the length of each bond in angstroms. docked into. The original three pesticides, RH-5992, RH-2485, and RH-5849, were built on GaussView before being optimized by Gaussian 09W. The OpenEye program Omega2 was then used to generate a library of conformers for each of the pesticides. Once the library of conformers had been generated, the OpenEye program Fred was used to “dock” each conformer into the active site previously generated and then to give each of the conformers a relative total score based on how well the structure filled the cavity of the docking site as well as the strength of hydrogen bonding with amino acid sidechains. The best molecules were decided upon by comparing the relative total scores of different candidates; a more negative score is indicative of a stronger molecule. The top-ranked conformers were found in the score file using Microsoft Excel and then visually examined in the OpenEye program VIDA for hydrogen bonds with enzyme’s active site (Figure 2). If a pesticide with a good score relative to previous attempts was found, OpenEye Brood was used to run a query on a selected portion of the molecule and generate bioisosteres with similar properties. If a member of this library was found to provide better docking scores, the conformer generation and visual examination process was begun again. This cycle was repeated as long as improvements continued to be made using the Brood program. After a favorable group of pesticides was generated, the OpenEye program FILTER was used to test the size, molecular weight, Lipinski violations, and other toxicities of each molecule. Irene Lin, Jill Tucker, Ryan Pemberton, Dean J. Tantillo, Sharlene Chin, and Selina Wang Page 3 of 5 Figure 3. The flowchart of molecules along with names and scores. Each molecule was docked in OpenEye Fred to generate a score and then run through Openeye Brood if the score was favorable relative to the original RH-2485 and to the previous pesticides built. Figure 4a. The above diagram (left) and picture (right) show the bonds RH-2485 forms with the amino acid sidechains of the ecdysone receptor 1R20. Hydrogen bonds are shown in the diagram in dashed lines and in the picture in yellow with all measurements in angstroms.