Yeast screen to kiddo's first dose in 48 hours
The same day we got the hit list from the yeast PGAP3 drug repurposing screen, we received independent confirmation that Lucy's brain cells are dying by a mechanism targeted by the top yeast hits.
Last month, we launched Perlara’s yeast drug repurposing pop-up lab located at SFBiolabs, a community driven incubator space in the heart of San Francisco for early-stage biotech companies to get building on day one. The inaugural project is with our partner Moonshots for Unicorns. Over the course of eight weeks — from August 1 to September 30 — we validated the known temperature-sensitive growth defect of a PGAP3-deficient yeast mutant, optimized a 384-well-plate, luminescence-based growth assay, and screened several thousand generics and GRAS compounds for rescue of cell growth. Remarkably, the top yeast hits were so compelling from a mechanism of action standpoint that Lucy’s parents started dosing her with four of them within days of completing the yeast screen! Keep reading to learn why..
We left off last time having shown that a homozygous diploid per1∆/per1∆ (PGAP3-deficient) mutant strain exhibits a statistically robust and reproducible growth defect at 37˚C in liquid media in 96-well plates. The next step in our distributed high-throughput screening workflow is assay optimization in 384-well plates. Working with Jez Revalde from the UCSF Small Molecule Discovery Center (SMDC), we had to establish assay conditions for a pilot drug screen of the Pharmakon collection.
In the plots below, we were able to determine the appropriate cell seeding density and incubation time that maximizes the delta between the wild-type (WT) yeast strain and the per1∆/per1∆ homozygous diploid mutant (Mut75). BTG stands for Bac-Titer Glo, which is a reagent used to quantify the amount of ATP in a sample. Typically, yeast cell growth can be measured by a plate reader using absorbance/optical density at a particular wavelength of light, either 595nm or 600nm. However, ATP levels correlate with cell number, and a luminescence readout (albeit a destructive terminal measurement) is much more sensitive than an absorbance readout.
Once we had a screenable window, we went ahead with a single replicate screen of the Pharmakon collection. We were very pleased that the screen worked on the first attempt. The distribution of hits is shown below. Consistent with previous model-organism-based high-throughput screens, the hit rate is between 0.1%-0.5% depending on where you draw the cutoff. Pay attention to the left-hand side of the curve where the suppressors (in genetics parlance) or rescuers (in lay terms) live. The top 4 yeast hits have an “intensity b-score” (equivalent to a Z-score) greater than or equal to 15. Encouragingly, we had not observed those four hits at the top of the pile for any other yeast drug repurposing screen, so this hit signature is specific for PGAP3.
The entire dataset is replotted below with thresholds for statistical significance. Two of the top hits, floating in the cloud of positive control wild-type yeast, completely rescued the growth defect of the per1∆/per1∆ mutant! And here’s where the story got really interesting.
The very same day two weeks ago on Monday October 3 that we got the yeast hit list, we received this email from Lucy’s parents, Drs Zach and Geri Landman: “Our research at _________ is showing that PGAP3 (at least Lucy's cells) demonstrate a severe mitochondrial phenotype with astrocyte-induced neuron cell death and could be a good target for nutritional and repurposed medicines.”
It turns out that two of the top four yeast hits are known mitochondrial cocktail components. What’s more, the other two top hits not only target a pathway that ameliorates astrocyte-induced neuron death, but the other lab also showed that another compound with the same mechanism as these top two hits rescued Lucy’s brain cells in dish. Buoyed and emboldened by this cross-species pharmacological convergence, Lucy’s parents started dosing her with all four yeast hits. A few days ago we received this video of Lucy taking her first (baby) step:
https://youtube.com/shorts/4j9DCbFUtFI?feature=share
In the next week or so, we’ll learn whether the Lucy yeast avatar has a growth defect comparable to the homozygous diploid per1∆/per1∆ mutant. In parallel, we’re gearing up to screen the homozygous diploid per1∆/per1∆ mutant against a much bigger drug repurposing library with 10,000+ compounds.