Chapter 3: Discovery of small molecule modulators of gene expression . 80
3.3.11 Initial screens of a diverse set of 3647 small molecules142
Through communication with our Ti-3D collaborators, we decided to begin our high-throughput small molecule screening efforts with a diverse set of small molecules. The first chemical libraries we screened were the NIH Clinical Collections (Molecular Libraries Small Molecule Repository operated by Evotec).
This library consists of 727 small molecules that have been used in clinical trials, resulting in a collection with relatively high bioactivity and therapeutic potential.
The next library tested was an institutionally derived collection of 600 kinase inhibitors. Kinase inhibitors are a very common class of therapeutic and biologically useful compounds (Chen 2012; Harrison 2012; Fabbro et al. 2012).
The last and most diverse set of small molecules screened was The Spectrum Collection (MicroSource). This collection contained 2320 compounds and serves as a great primary screening tool due to its wide range of biological activity and structural diversity. 60% of this library is comprised of existing drugs, 25% are natural products, and the remaining 15% are “other bioactive compounds” such as non-drug toxins, enzyme inhibitors, membrane active compounds, and receptor blockers. Together, these three diverse collections of small molecules provided an excellent starting point for our screening efforts.
Our screening strategy is outlined in Illustration 3.2 and in our primary screen we analyzed the aforementioned libraries at 10 µM by plate reader analysis with the drug-sensitive yeast strain GM ABC-16. OD (A600), GFP fluorescence, and mCherry fluorescence were measured for all wells. OD (A600)
normalized fluorescence values were used to calculate % inhibition, relative to the negative control samples. Compounds were ranked based on the % inhibition of GFP and mCherry separately and compounds with inhibitions of more than 50% were chosen for a secondary screen. Additionally, cytostatic compounds, as seen by no increase in OD (A600), were selected, since they might reveal an interesting phenotype at lower concentrations. In all, 354 small molecules were selected to be screened in a secondary screen. The secondary screen was performed in the same manner as the primary screen, except that three concentrations were tested (10 µM, 1 µM, and 100 nM). Additionally, high-throughput flow cytometry analysis was used to examine reporter fluorescence in addition to plate reader analysis. The three concentrations served to determine if the compounds resulted in dose-dependent effects, as well as to gauge the effect of compounds that were toxic at the primary concentration 10 µM.
3.3.12 The secondary small molecule screen resulted in 16 high-confidence hits
Of the 354 small molecules that were analyzed in the secondary screen, we identified 16 compounds with significant dose-dependent effects on reporter expression. In Figure 3.14 I display the phenographs from the secondary screen for 10 of the 16 compounds. Across the board, the dose-dependent responses are apparent. It was also very encouraging that a number of the compounds still have a significant reporter phenotype at the low concentration of 100 nM.
Figure 3.14: Dose-dependent reporter effects by high-confidence hits from small molecule screen
Small molecule treated flow cytometry data (pseudocolor) are overlaid onto DMSO only data (grayscale) for the three concentrations tested in the secondary screen. Phenographs are as usual with GFP and mCherry displayed on the x-
TI3D_0099_E07 was an interesting hit because there were three structural derivatives that resulted in almost identical reporter phenotypes. These three analogs of TI3D_0099_E07 only differ by one R group and represent a family of inhibitors for a specific kinase. Identifying four very similar small molecules with analogous phenotypes was very encouraging and results in higher confidence in the validity of this hit. Another interesting group of small molecules were ones that result in a phenotype alike TI3D_0104_I03. There were a total of three molecules with a more heterogeneous green-shifted phenotype, with the bulk of cells separated from the non-fluorescent cells at the origin. When investigating the nature of these compounds, all three were nucleoside analogs.
Our Ti-3D collaborators then attempted to determine the half maximal effective concentration (EC50) for the hit compounds, which essentially is the concentration that elicits a response halfway between the negative control and the maximal response. EC50 values were calculated for each fluorescent signal separately, as well as OD (A600; Figure 3.15). To complete these experiments fresh compounds were purchased in order to confirm the validity of the hit, as well as make a range of stock concentrations. The concentrations tested varied between compounds based on the severity of the reporter phenotype and growth inhibition, but all spanned at least a 4-log dilution series. We also determined the EC50 for wild-type yeast (BY4742/SS4050). Although a wide range of concentrations was used for each compound, some EC50 values were not able to be determined.
When studying the EC50 values, we were very encouraged by the results of some compounds at very low concentrations, including a number in the low to sub-micromolar range. We also observed overall lower EC50 values for the drug-sensitive strain GM ABC-16 when compared to wild-type yeast, which was surprising. Moreover, it was very interesting to observe some compounds with significantly different EC50 values for the same compound when comparing the effect on GFP and mCherry. For example, the mCherry EC50 for TI3D_0098_014 is nearly 10-fold lower than that of GFP. On the other hand, the GFP EC50 for TI3D_0104_I03 is 20-times lower than the mCherry EC50. These differences reveal specific effects on reporter expression instead of an overall decrease in expression.
Wild-type GM ABC-16
Abs GFP mCherry Abs GFP mCherry
TI3D_0098_O14 - - - 4.35 4.58 0.501
TI3D_0099_D10 2.58 - 0.54 0.0365 0.843 0.0355
TI3D_0099_E07 0.126 0.0363 0.0209 0.0268 0.0589 0.0415
TI3D_0099_I03 12.9 13.3 14.3 12.9 24.0 32.6
TI3D_0099_M16 0.0538 0.0191 0.0255 0.0467 0.0101 0.0278
TI3D_0099_O08 - 179 48.9 27.4 6.93 3.57
TI3D_0100_H19 0.036 0.184 0.0227 0.0144 0.0371 0.0122
TI3D_0104_A05 25.6 8.3 7.3 17.4 9 9.12
TI3D_0104_B19 18.1 - - 6.14 - 9.77
TI3D_0104_I03 40.6 10.7 - 3.42 0.553 12.8
TI3D_0104_I09 13.7 55 7.15 8.9 5.4 3.70
TI3D_0105_C07 40.2 2.45 5.56 35 2.68 12.6
TI3D_0107_B13 4.1 1.35 1.26 0.486 0.372 0.101
TI3D_0219_C19 9.94 12.5 5.07 1.41 2.23 0.489
Figure 3.15: EC50 values for high-confidence small molecule hits
The EC50 values for high-confidence small molecule hits are shown. Our Ti-3D collaborators determined values by screening compounds across a concentration gradient of 6 concentrations covering at least 4-logs. All values are in µM and represent the concentration of compounds that elicits 50% of the maximal effect. Wild-type cells are strain SS4050 and GM ABC-16 are strain SS4127. Abs represents absorbance (OD (A600)) while GFP and mCherry refer to fluorescence. The “-“ means the EC50 was unable to be determined.
3.3.13 Identification of possible target pathways for small molecule hits