As discussed in a previous post, we are currently working on a GNAO1 PerlQuest collaboration with the Undiagnosed Diseases Network (UDN) and Harvard Medical School. GNAO1 is a gene that encodes for the alpha subunit of a G protein, and mutations in this gene lead a range of neurodevelopmental disorders. In yeast, the GNAO1 ortholog is GPA1, and in our last update, we characterized five variants to find a strain that we will use to screen our drug repurposing library in the second phase of our PerlQuest (Table 1).

 

GNAO1 variant: DNA base change

GNAO1 variant: amino acid change

Yeast GPA1 model
c.662C>A A221D A339D
c.736G>A E246K E364K
c.625G>A R209G R327G
c.134G>A G45E G53E
c.118G>T G40W G48W

Table 1. GNAO1 patient alleles and the equivalent variants that we generated in yeast GPA1

 

Troubleshooting A339D

Developing a screenable yeast GNAO1 model came with a few challenges. We initially generated five variants in GPA1 and analyzed three biological replicates each: G48W, G53E, R327G, E364K, and A339D. Four of the variants grew consistently: G48W and G53E cells failed to grow, while R327G and E364K cells grew at the same rate as wild type. Biological replicates of A339D grew at different rates. Without a dependable growth pattern, we were hesitant to move forward with a GNAO1 yeast screen.

Therefore, we remade A339D strains to get more biological replicates. After numerous growth assays, we narrowed it down to one replicate in particular– A339D #9. This replicate grew consistently slower than wild type GPA1 cells while still viable (Figure 1). We chose to move forward with this A339D replicate for our screen.

 

Figure 1. Growth assay of technical replicates. This is representative data of growth assays done with six technical replicates of GPA1 A339D #9. We measured cell density (absorbance at 600 nm) over time (hours)

 

GNAO1 yeast screen set-up

The order of operations is important in conducting a thorough and reproducible drug screen. Our repurposing compound library contains about 2500 compounds distributed across eight 384-well plates with columns 1, 2, 23, and 24 reserved for positive and negative controls (Figure 2). The positive and negative control wells contain DMSO because the drugs in our compound libraries are dissolved in DMSO. We performed the screen in triplicates as described in Figure 3.

 

Figure 2. Plate Map. The positive controls contain DMSO and wild type cells (yellow), the negative controls contain DMSO and A339D-9 (blue), and the sample wells contain drugs and A339D-9 (purple)

 

Figure 3. Workflow. Our workflow followed a three step process. First, using the Echo, we dispense plates with drugs from our repurposing library at 25 µM or with DMSO-only in our control wells. Second, we dispense a solution containing yeast cells into the drug plates using the Multiflo, and incubate the plates at 30˚C. After 32 hours, we determine the optical density using the SpectraMax

 

After completing the three replicates, we moved on to data analysis. Figure 4 shows the separation of our positive and negative controls from one replicate of our screen. As you can see, there is no overlap between the positive and negative controls. We compared our sample wells to the negative control wells to determine if there is rescue. In this case, rescue is defined as growth (optical density) that is 2.5 standard deviations higher than the mean of the negative controls (z-score ≥ 2.5). We identified thirty compounds that meet this criteria. These compounds include natural products, generic drugs, and FDA-approved drugs. We plan to validate them in future experiments.

 

Figure 4. Comparison of z-scores of controls. Shown are the z-scores of the positive and negative controls of the first replicate of our screen. We considered hits to be compounds with a z-score ≥  2.5 across all replicates

 

Moving Forward

We used Synthetic Complete media containing 5-FOA during our drug screen. As described in my last blog post, GPA1 is an essential gene, so we had used a conditional rescue plasmid that express wild type GPA1 in these cells. Because of the variable phenotype of the A339D variants, we kept the wild type plasmid in these cells until the start of the screen. Then, the plasmid was selected against during the screening process by the presence of 5-FOA. In future experiments, we plan to also validate our hits without 5-FOA in the media so that we can be sure that the drug is specifically rescuing the GNAO1 mutation instead of deactivating 5-FOA. Stay tuned for more updates!

 

Feature image modified from the original on wallpapersafari.com