Improving Measures of Chemical Structural Similarity Using Machine Learning on Chemical-Genetic Interactions.

A common strategy for identifying molecules likely to possess a desired biological activity is to search large databases of compounds for high structural similarity to a query molecule that demonstrates this activity, under the assumption that structural similarity is predictive of similar biological activity. However, efforts to systematically benchmark the diverse array of available molecular fingerprints and similarity coefficients have been limited by a lack of large-scale datasets that reflect biological similarities of compounds. To elucidate the relative performance of these alternatives, we systematically benchmarked 11 different molecular fingerprint encodings, each combined with 13 different similarity coefficients, using a large set of chemical-genetic interaction data from the yeast Saccharomyces cerevisiae as a systematic proxy for biological activity. We found that the performance of different molecular fingerprints and similarity coefficients varied substantially and that the all-shortest path fingerprints paired with the Braun-Blanquet similarity coefficient provided superior performance that was robust across several compound collections. We further proposed a machine learning pipeline based on support vector machines that offered a fivefold improvement relative to the best unsupervised approach. Our results generally suggest that using high-dimensional chemical-genetic data as a basis for refining molecular fingerprints can be a powerful approach for improving prediction of biological functions from chemical structures.

Resistance against Sclerotinia sclerotiorum in soybean involves a reprogramming of the phenylpropanoid pathway and up-regulation of antifungal activity targeting ergosterol biosynthesis.

Sclerotinia sclerotiorum, a predominately necrotrophic fungal pathogen with a broad host range, causes a significant yield-limiting disease of soybean called Sclerotinia stem rot. Resistance mechanisms against this pathogen in soybean are poorly understood, thus hindering the commercial deployment of resistant varieties. We used a multiomic approach utilizing RNA-sequencing, gas chromatography-mass spectrometry-based metabolomics and chemical genomics in yeast to decipher the molecular mechanisms governing resistance to S. sclerotiorum in soybean. Transcripts and metabolites of two soybean recombinant inbred lines, one resistant and one susceptible to S. sclerotiorum were analysed in a time course experiment. The combined results show that resistance to S. sclerotiorum in soybean is associated in part with an early accumulation of JA-Ile ((+)-7-iso-jasmonoyl-L-isoleucine), a bioactive jasmonate, increased ability to scavenge reactive oxygen species, and importantly, a reprogramming of the phenylpropanoid pathway leading to increased antifungal activities. Indeed, we noted that phenylpropanoid pathway intermediates, such as 4-hydroxybenzoate, cinnamic acid, ferulic acid and caffeic acid, were highly accumulated in the resistant line. In vitro assays show that these metabolites and total stem extracts from the resistant line clearly affect S. sclerotiorum growth and development. Using chemical genomics in yeast, we further show that this antifungal activity targets ergosterol biosynthesis in the fungus, by disrupting enzymes involved in lipid and sterol biosynthesis. Overall, our results are consistent with a model where resistance to S. sclerotiorum in soybean coincides with an early recognition of the pathogen, leading to the modulation of the redox capacity of the host and the production of antifungal metabolites.

MOSAIC: a chemical-genetic interaction data repository and web resource for exploring chemical modes of action.

Summary: Chemical-genomic approaches that map interactions between small molecules and genetic perturbations offer a promising strategy for functional annotation of uncharacterized bioactive compounds. We recently developed a new high-throughput platform for mapping chemical-genetic (CG) interactions in yeast that can be scaled to screen large compound collections, and we applied this system to generate CG interaction profiles for more than 13 000 compounds. When integrated with the existing global yeast genetic interaction network, CG interaction profiles can enable mode-of-action prediction for previously uncharacterized compounds as well as discover unexpected secondary effects for known drugs. To facilitate future analysis of these valuable data, we developed a public database and web interface named MOSAIC. The website provides a convenient interface for querying compounds, bioprocesses (Gene Ontology terms) and genes for CG information including direct CG interactions, bioprocesses and gene-level target predictions. MOSAIC also provides access to chemical structure information of screened molecules, chemical-genomic profiles and the ability to search for compounds sharing structural and functional similarity. This resource will be of interest to chemical biologists for discovering new small molecule probes with specific modes-of-action as well as computational biologists interested in analysing CG interaction networks. Availability and implementation: MOSAIC is available at http://mosaic.cs.umn.edu. Contact: hisyo@riken.jp, yoshidam@riken.jp, charlie.boone@utoronto.ca or chadm@umn.edu. Supplementary information: Supplementary data are available at Bioinformatics online.

Functional annotation of chemical libraries across diverse biological processes.

Chemical-genetic approaches offer the potential for unbiased functional annotation of chemical libraries. Mutations can alter the response of cells in the presence of a compound, revealing chemical-genetic interactions that can elucidate a compound's mode of action. We developed a highly parallel, unbiased yeast chemical-genetic screening system involving three key components. First, in a drug-sensitive genetic background, we constructed an optimized diagnostic mutant collection that is predictive for all major yeast biological processes. Second, we implemented a multiplexed (768-plex) barcode-sequencing protocol, enabling the assembly of thousands of chemical-genetic profiles. Finally, based on comparison of the chemical-genetic profiles with a compendium of genome-wide genetic interaction profiles, we predicted compound functionality. Applying this high-throughput approach, we screened seven different compound libraries and annotated their functional diversity. We further validated biological process predictions, prioritized a diverse set of compounds, and identified compounds that appear to have dual modes of action.

Errata: Functional annotation of chemical libraries across diverse biological processes.

This corrects the article DOI: 10.1038/nchembio.2436.

Errata: Functional annotation of chemical libraries across diverse biological processes.

This corrects the article DOI: 10.1038/nchembio.2436.

Predicting bioprocess targets of chemical compounds through integration of chemical-genetic and genetic interactions.

Chemical-genetic interactions-observed when the treatment of mutant cells with chemical compounds reveals unexpected phenotypes-contain rich functional information linking compounds to their cellular modes of action. To systematically identify these interactions, an array of mutants is challenged with a compound and monitored for fitness defects, generating a chemical-genetic interaction profile that provides a quantitative, unbiased description of the cellular function(s) perturbed by the compound. Genetic interactions, obtained from genome-wide double-mutant screens, provide a key for interpreting the functional information contained in chemical-genetic interaction profiles. Despite the utility of this approach, integrative analyses of genetic and chemical-genetic interaction networks have not been systematically evaluated. We developed a method, called CG-TARGET (Chemical Genetic Translation via A Reference Genetic nETwork), that integrates large-scale chemical-genetic interaction screening data with a genetic interaction network to predict the biological processes perturbed by compounds. In a recent publication, we applied CG-TARGET to a screen of nearly 14,000 chemical compounds in Saccharomyces cerevisiae, integrating this dataset with the global S. cerevisiae genetic interaction network to prioritize over 1500 compounds with high-confidence biological process predictions for further study. We present here a formal description and rigorous benchmarking of the CG-TARGET method, showing that, compared to alternative enrichment-based approaches, it achieves similar or better accuracy while substantially improving the ability to control the false discovery rate of biological process predictions. Additional investigation of the compatibility of chemical-genetic and genetic interaction profiles revealed that one-third of observed chemical-genetic interactions contributed to the highest-confidence biological process predictions and that negative chemical-genetic interactions overwhelmingly formed the basis of these predictions. We also present experimental validations of CG-TARGET-predicted tubulin polymerization and cell cycle progression inhibitors. Our approach successfully demonstrates the use of genetic interaction networks in the high-throughput functional annotation of compounds to biological processes.

Correction: Directed Evolution Reveals Unexpected Epistatic Interactions That Alter Metabolic Regulation and Enable Anaerobic Xylose Use by Saccharomyces cerevisiae.

[This corrects the article DOI: 10.1371/journal.pgen.1006372.].

Chemical genomic guided engineering of gamma-valerolactone tolerant yeast.

BACKGROUND: Gamma valerolactone (GVL) treatment of lignocellulosic bomass is a promising technology for degradation of biomass for biofuel production; however, GVL is toxic to fermentative microbes. Using a combination of chemical genomics with the yeast (Saccharomyces cerevisiae) deletion collection to identify sensitive and resistant mutants, and chemical proteomics to monitor protein abundance in the presence of GVL, we sought to understand the mechanism toxicity and resistance to GVL with the goal of engineering a GVL-tolerant, xylose-fermenting yeast. RESULTS: Chemical genomic profiling of GVL predicted that this chemical affects membranes and membrane-bound processes. We show that GVL causes rapid, dose-dependent cell permeability, and is synergistic with ethanol. Chemical genomic profiling of GVL revealed that deletion of the functionally related enzymes Pad1p and Fdc1p, which act together to decarboxylate cinnamic acid and its derivatives to vinyl forms, increases yeast tolerance to GVL. Further, overexpression of Pad1p sensitizes cells to GVL toxicity. To improve GVL tolerance, we deleted PAD1 and FDC1 in a xylose-fermenting yeast strain. The modified strain exhibited increased anaerobic growth, sugar utilization, and ethanol production in synthetic hydrolysate with 1.5% GVL, and under other conditions. Chemical proteomic profiling of the engineered strain revealed that enzymes involved in ergosterol biosynthesis were more abundant in the presence of GVL compared to the background strain. The engineered GVL strain contained greater amounts of ergosterol than the background strain. CONCLUSIONS: We found that GVL exerts toxicity to yeast by compromising cellular membranes, and that this toxicity is synergistic with ethanol. Deletion of PAD1 and FDC1 conferred GVL resistance to a xylose-fermenting yeast strain by increasing ergosterol accumulation in aerobically grown cells. The GVL-tolerant strain fermented sugars in the presence of GVL levels that were inhibitory to the unmodified strain. This strain represents a xylose fermenting yeast specifically tailored to GVL produced hydrolysates.

Plant-derived antifungal agent poacic acid targets ß-1,3-glucan.

A rise in resistance to current antifungals necessitates strategies to identify alternative sources of effective fungicides. We report the discovery of poacic acid, a potent antifungal compound found in lignocellulosic hydrolysates of grasses. Chemical genomics using Saccharomyces cerevisiae showed that loss of cell wall synthesis and maintenance genes conferred increased sensitivity to poacic acid. Morphological analysis revealed that cells treated with poacic acid behaved similarly to cells treated with other cell wall-targeting drugs and mutants with deletions in genes involved in processes related to cell wall biogenesis. Poacic acid causes rapid cell lysis and is synergistic with caspofungin and fluconazole. The cellular target was identified; poacic acid localized to the cell wall and inhibited ß-1,3-glucan synthesis in vivo and in vitro, apparently by directly binding ß-1,3-glucan. Through its activity on the glucan layer, poacic acid inhibits growth of the fungi Sclerotinia sclerotiorum and Alternaria solani as well as the oomycete Phytophthora sojae. A single application of poacic acid to leaves infected with the broad-range fungal pathogen S. sclerotiorum substantially reduced lesion development. The discovery of poacic acid as a natural antifungal agent targeting ß-1,3-glucan highlights the potential side use of products generated in the processing of renewable biomass toward biofuels as a source of valuable bioactive compounds and further clarifies the nature and mechanism of fermentation inhibitors found in lignocellulosic hydrolysates.