Center for High-Throughput Functional Annotation of Natural Products
Technology

Cytological Profiling
Cytological profiling (CP) is a powerful method for quantifying and comparing the phenotypic effects of small molecules on cells. Combining automated fluorescence microscopy with computer-aided image processing, CP provides an information-rich phenotypic profile for each tested sample. Rather than focusing on a single narrowly defined phenotype, CP uses multiple cytological probes to generate hundreds of quantifiable cytological features, giving rise to a phenotypic fingerprint of each compound. These fingerprints can be compared to fingerprints of a set of reference compounds, allowing unknown compounds to be classified based on their similarity to compounds with known targets and/or mechanisms of action. Citological Profiling

Schulze, C. J.; Bray, W. M., Woerhmann; M. H., Stuart, J.; Lokey, R. S.; and Linington, R. G. "Function-first" lead discovery: mode of action profiling of natural product libraries using image-based screening" Chem. Biol. 2013, 20, 285. (http://www.sciencedirect.com/science/article/pii/S1074552112004887)

Functional Signature Ontology (FuSiOn)
A platform to study MOA for small molecules that takes advantage of gene expression signatures. By probing the expression signatures of small molecules and genetic perturbations (siRNA, miRNA) we utilize pattern-matching tools that produce verifiable mode-of-action. We carry out this analysis using a minimal mRNA reporter cohort and technologies for high-throughput quantitative multi-analyte detection. The endogenous reporter gene signatures resulting from each perturbation were assembled into a similarity matrix using Euclidean and Mahalanobis distance distributions. In this way we produce FUSION maps that link bioactive agents to the molecular entities and/or biological processes they engage in cells. Functional Signature Ontology

Potts, M.B.; Kim, Y.; Fisher, K. W.; Hu, Y.; Carrasco, Y.; Ou, Y-H.; Herrera, M.; Cubillos, F.; Xiao, G.; Hofree, M.; Ideker, T.; Xie, Y.; Lewis, R. E.; MacMillan, J. B.*; White, M. A.* "Broad-scale mode-of-action annotation of natural product perturbations by functional signature ontology (FUSION)" Science Signaling 2013, 6, ra90. (http://stke.sciencemag.org/content/6/297/ra90)

Unbiased Metabolomics
Confident and accurate identification of chemical entities from complex mixtures remains the central challenge for untargeted metabolomics platforms. This is particularly important for natural products and botanicals extracts, where the chemical complexity of a given sample can be very high. The metabolomics workflow is typically divided into two stages:

  1. Determination of the chemical constitution of the mixture using LCMS or NMR features to identify unique constituents (termed 'basketing')
  2. Connecting these unique constituents to specific molecules from a library of potential matching structures.
Recent computational analyses by the MacMillan group on the diversity and distributions of chemical scaffolds in Nature have revealed a high degree of convergence for specific structural classes, with 90% of all published bacterial natural products belonging to one of just 1,000 structural classes. Rather than the random distribution of possible structures one might expect based on independent evolution of biosynthetic gene clusters, this analysis demonstrates that only a small number of possible carbon skeletons are actually prevalent in Nature. This observation is important, as it reduces the analytical challenge for natural products from an unbounded problem, where all possible scaffolds must be considered, to a bounded problem where the scaffolds of many compound classes can be predefined.
The central objective of the untargeted metabolomics platform are to:
  1. Correctly describe the chemical constitution within and between samples.
  2. Accurately connect these chemical constituents with known compounds from a database of published natural products or botanical constituents.
  3. Integrate these data with the cytological profiling and FUSION datasets to perform functional annotation of individual molecular species.
To accomplish this we will employ a combination of MS-based metabolite profiling and new informatics tools for both sample basketing and functional correlation. Unbiased Metabolomics

UT Southwestern Medical CenterSimon Fraser UniversityUniversity of California, Santa Cruz