Assigning mode of action to phenotypically discovered anticancer leads.

ABSTRACT. Assigning the mode of action to bioactive compounds is an essential step in drug discovery and a major challenge in chemical biology. This problem is particularly acute for drug discovery from nature. Natural products (NP) provide unique scaffolds not found in synthetic libraries, and the abundant NP collections are a vast diversity reservoir for drug development. However, the lack of mechanistic understanding is a major hindrance to preclinical development. Existing HTS techniques permit efficacious screening of NP libraries, but the follow-up purification, chemical structure identification, and MOA assessment of purified metabolites are lengthy, costly, and tedious. Worse, as the MOA is determined only at the end, much effort is wasted on isolating redundant and irrelevant hits. Since the purification of active constituents requires significant work, it should be performed only for high-value molecules with pharmacological novelty. Currently used MOA assessment approaches employ various platforms, including panels of cell-based and biochemical assays and systems biology techniques, but none provide a satisfactory solution for the MOA problem. Here, we describe an alternative MOA evaluation technique based on a systems biology approach developed at Attagene. Under this approach, cell response is characterized by the activity of transcription factors (TF) that link cellular signaling pathways to genes. The enabling technology is the FACTORIAL, a proprietary Attagene platform for quantitative TF activity profiling (TFAP). We demonstrated that TFAP signatures enable a straightforward MOA assessment of chemicals by pinpointing perturbed bioprocesses and cell systems. Most importantly, this approach does not involve complex bioinformatic inferences. Here, we will extend the TFAP approach to ascribe the MOA to anticancer drug leads from nature. In pilot studies, we examined TFAP signatures of approved anticancer drugs and anticancer fungal metabolites. We found that (i) major classes of approved anticancer drugs have specific TFAP signatures; (ii) anticancer fungal metabolites, too, have distinct TFAP signatures. Moreover, these signatures allowed correct identification of metabolites' MOA; (iii) most unexpectedly, crude fungal extracts and purified active metabolites showed identical TFAP signatures. These data suggest a new approach to the mechanistic evaluation of nature-derived anticancer leads. We will develop this approach with a UNC-Greensboro team with over 700 purified anticancer fungal metabolites with established structures. First, we will obtain TFAP signatures for all FDA-approved drugs and a large fraction of the UNCG library (SA1). Then, we will analyze these TFAP datasets and compare the 'MOA spaces' for the anticancer fungal metabolites and approved drugs to identify metabolites with novel MOA (SA2). Finally, we will validate the approach to identify the MOA in crude fungal extracts, allowing prioritizing high-value strains for purification (SA3). Implementing this proposal will establish a new approach that uses a single instrumental platform, does not involve bioinformatic analyses, and allows ascribing the MOA to unpurified NP extracts.