Using C. elegans Forward and Reverse Genetics to Identify New Compounds with Anthelmintic Activity

Abstract

Background: The lack of new anthelmintic agents is of growing concern because it affects human health and our food supply, as both livestock and plants are affected. Two principal factors contribute to this problem. First, nematode resistance to anthelmintic drugs is increasing worldwide and second, many effective nematicides pose environmental hazards. In this paper we address this problem by deploying a high throughput screening platform for anthelmintic drug discovery using the nematode Caenorhabditis elegans as a surrogate for infectious nematodes. This method offers the possibility of identifying new anthelmintics in a cost-effective and timely manner.

Methods/principal findings: Using our high throughput screening platform we have identified 14 new potential anthelmintics by screening more than 26,000 compounds from the Chembridge and Maybridge chemical libraries. Using phylogenetic profiling we identified a subset of the 14 compounds as potential anthelmintics based on the relative sensitivity of C. elegans when compared to yeast and mammalian cells in culture. We showed that a subset of these compounds might employ mechanisms distinct from currently used anthelmintics by testing diverse drug resistant strains of C. elegans. One of these newly identified compounds targets mitochondrial complex II, and we used structural analysis of the target to suggest how differential binding of this compound may account for its different effects in nematodes versus mammalian cells.

Conclusions/significance: The challenge of anthelmintic drug discovery is exacerbated by several factors; including, 1) the biochemical similarity between host and parasite genomes, 2) the geographic location of parasitic nematodes and 3) the rapid development of resistance. Accordingly, an approach that can screen large compound collections rapidly is required. C. elegans as a surrogate parasite offers the ability to screen compounds rapidly and, equally importantly, with specificity, thus reducing the potential toxicity of these compounds to the host and the environment. We believe this approach will help to replenish the pipeline of potential nematicides.

Fig 1. WormScan Score Analysis.

(A) Two VC2010 L4 stage C. elegans in a 1% DMSO control well that contains no drug, two sequential scans are taken after five days of exposure, the difference image generated for the region of interest (black circle) gives a WormScan Score of 100. (B) Two VC2010 L4 C. elegans exposed to 43 μM of CID 6741218 for five days exposure resulted in reduced brood size, reduced behavioral response to light stimulus and increased mortality, giving a WormScan Score of 50. (C) Two VC2010 L4 stage C. elegans exposed to 50 μM of Ivermectin for five days of exposure, giving a WormScan Score of 0, which resulted in mortality. (D) A well that contained no C. elegans gives a WormScan Score of 0. The scale bar applies to all images.

Fig 2. WormScan Scores from initial compound screen and confirmation of hits.

(A) Two VC2010 L4 C. elegans were sorted into each well of a 96-well flat-bottom plate. C. elegans were exposed to 43 μM for each of the 26,000 compounds from the Maybridge and Chembridge libraries for 5 days and then screened and sorted by WormScan Score, which was normalized by percent of control wells. The 404 top anthelmintic candidates from the initial screen are highlighted by the black box. (B) The WormScan Scores of the 5,152 controls from the compound screen are shown in histogram with a bin width of 2. The mean WormScan Score for the controls is 100. (C) The top 404 compounds were re-pinned for two more biological replicates and displayed here are the top 184 active compounds. After a series of filters were applied to these 184 compounds, 14 compound candidates were retained for further testing.

Fig 3. Homology model of C. elegans complex II with associated amino acid changes conferring resistance to various anthelmintics.

In all four panels the complex II subunits are illustrated as follows, SDHA-1 (red), SDHB-1 (blue) or SDHC-1 (mev-1; green) and SDHD-1 (purple). The CID 2747322 molecule is colored in cyan and represented as ball and sticks. The chemistry informatics tool Screen3D version 2015 [57] is used to bind CID 2747322 to the crystal structure of complex II with a bound flutolanil analogue. Structural models and docked ligands visualized with pymol [60]. (A) The black sphere of the alpha carbon, T66 position of SDHC-1 (mev-1), is the amino acid alteration in the CID 2747322 resistant strain VC3631. (B) The spheres indicate the million mutation project strains mutations in complex II. Amino acid alterations found to confer resistance towards CID 2747322 exposure are colored in black and wild-type toxicity levels are colored in green. (C) Colored in black are the residues of the pathogenic fungus Mycosphaerella graminicola, which give resistance to complex II inhibitors [75]. These positions have been aligned to the C. elegans position. (D) The black spheres of the alpha carbon residues induce resistance towards exposure from the WACT-11 compound family [34].

Fig 4. A structure-activity relationship (SAR) analysis of CID 2747322.

(A) Hierarchical maps. Hierarchical maps were used to categorize 14 of the 50 compounds from the SAR screen, with variations in the right R-group 2-(trifluoromethyl)benzene, left R-group 2-(4-methoxyphenoxy)ethyl or both R-groups. Included in the hierarchical map is a Bayer patented compound (N-(2-((5-methylpyridin-2-yl)oxy)ethyl)-2-(trifluoromethyl)benzamide) [65], which has a 3D similarity score of 0.82. Fluopyram has a 3D similarity score of 0.58. Also included is the WACT-11 compound identified by Burns [34] with a 3D similarity score of 0.66, and the flutolanil analogue (CID 49852661) used for modeling which has a 3D similarity score of 0.60. (B) A SAR analysis for CID 2747322 against our internal library of 24,989 compounds from Maybridge and Chembridge libraries. The top 50 compounds with the highest 3D similarity scores to CID 2747322, calculated using Screen3D version 2015 [57] were re-pinned for two additional biological replicates. Compounds with variations on the left R-group 2-(trifluoromethyl)benzene or the right R-group 2-(4-methoxyphenoxy)ethyl or both R-groups are highlighted. The compound CID 2747279 (3D Similarity Score of 0.74), has variations on the right R-group was the only compound from the set of 50 that displayed strong anthelmintic activity.

Fig 5. A phylogenetic comparison of the quinone binding site in MEV-1.

Sequences for mev-1 of representative nematode species together with, Homo sapiens, Sus scrofa, and S. cerevisiae were obtained using BlastP and aligned by ClustalW. The key residues involved in quinone binding, are conserved amongst all nematode species examined. The red star indicates the T66I variant that results in resistance to CID 2747322 exposure in the VC3631 strain. Residues important for the left binding pocket of CID 2747322 are indicated with a black star. Image was generated using Geneious version 8.1.7.

Fig 6. Complex II specificity model of CID 2747322 surface representation of nematode and vertebrate homologue.

(A) The A. suum structure of succinate dehydrogenase in complex II of the mitochondria, illustrating how the CID 2747322 molecule can be accommodated in the quinone binding pocket. The molecule CID 2747322 was aligned to analogue flutolanil using Screen3D [76]. The right side of CID 2747322, 2-(trifluoromethyl)benzene is found binding in a hydrophobic pocket with a Cation–π interaction from Arg278. The left side of CID 2747322, 2-(4-methoxyphenoxy)ethyl is found to orientate within a hydrophobic pocket. (B) The porcine structure of succinate dehydrogenase in complex II of the mitochondria indicates a non-favorable binding of the CID 2747322 molecule. The left side of CID 2747322 is unable to orient within the smaller hydrophobic pocket. Highlighted in yellow dots is the larger pocket of the nematode binding pocket.