Development and Application of a Yeast Screening Platform to Investigate Heat Shock Networks
The main function of heat shock proteins (HSPs) is to counterbalance the harmful effects of stress experienced by the cell. They do this by folding de novo and nonnative polypeptide chains that have exposed hydrophobic amino acid residues which in turn helps to maintain cell homeostasis. In diseased states where cells are constantly under stress, regulating the expression and function of heat shock proteins would be extremely advantageous. To date, thousands of heat shock modulators have since been developed. However, less than 1% has made it to clinical trials (with moderate success) prompting the identification of novel heat shock modulators. In this study, we seek to develop a platform via which we can identify both allosteric and non-allosteric regulators of heat shock proteins. We use a liquid assay platform in a 384 plate format that uses turbidity to evaluate phenotype. This has the advantage of capturing quantitative phenotypic information and the statistical robustness to provide a basis for scoring the degrees of drug effects. Using a reverse pathway to drug approach, we first determined predictor strains that when screened against a compound library would identify potential heat shock modulators. For this gene screen, we used quadratic form distance (QFD) to measure distances between the growth phenotype of haploid deletion mutants of Saccharomyces Cerevisiae grown in Geldanamycin (GA) and Radicicol (RAD) at 15 µM and 5 µM and 1% DMSO. These measured distances were used to calculate z and z' values which were defined as strains showing sensitivity relative to DMSO controls and strains showing differential sensitivity between GA and RAD respectively. We incorporated a large number of repeated treatments per strain (4-16 repeats) in each 384 well plate which allowed us to approximate p-values using an asymptotic Monte-Carlo permutation test. Repeated rounds of screening and application of our z, z' and p-value cutoffs, eventually yielded sixteen predictor strains which we called HSP90 Antagonist Sensitive (HAS) strains. Three of our HAS strains (hsp82Δ, ydj1Δ, sst2Δ) along with wild type (WT), were then used to screen the NCI Set II compound library of ~2400 compounds. For the compound library screen, QFD was replaced with time warping distance since time warping distance generated results similar to QFD and increased processing time without loss of data integrity. In order to delineate compound hits, we established 4 new metrics: %change, sum index, diversity index and v-values. We defined a hit as a compound demonstrating strong selective toxicity towards one HAS strain relative to WT. Extensive filters using the new metrics aforementioned resulted in the identification of 9 potential heat shock modulators including NSC#330500 (Macbecin II) a known HSP90 inhibitor. Since heat shock modulators have the ability to activate the heat shock response and hence induce expression of heat shock proteins, we tested the utility of the platform to identify growth enhancers via screening our hits in a &agr;-synuclein Parkinson's disease yeast model. Macbecin II and NSC#42199 (Benztropine Mesylate) showed strong growth recovery of the &agr;-synuclein GFP/YFP double mutant. We then screened a small library (44 compounds) of substituted pyrazoles and benzofurans, against the &agr;-synuclein GFP/YFP double mutant and found that approximately 10 compounds protected against &agr;-synuclein toxicity with compounds S4, S5 and S9 demonstrating the most potential at both 25 µM and 5 µM.
Davisson, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our