Yeast avatars for drug repurposing
Using yeast cells engineered to express variants that cause ultra-rare metabolic diseases, a collaboration between Perlara, UCSF and motivated families has yielded actionable repurposing hits.
A year after placing Perlara in stasis and on the eve of COVID-19, a professor at UCSF reached out to see if the PerlQuest multi-species drug repurposing approach we created at Perlara 1.0 could be resurrected in an academic lab as a “yeast to fibroblast to kiddo” pipeline.
With funding and community support from the MEPAN Foundation, we’re excited to announce that promising hits from yeast MEPAN avatar screens will soon be validated in MEPAN patient fibroblasts, and hopefully quickly thereafter a compelling clinical candidate will enter single-patient pioneer trials.
This post will summarize all the hard work that got us here and then go on to describe how we’re compressing and accelerating a process that took two years during a pandemic to several months in a decentralized pop-up lab.
Professor Ophir Klein contacted me in January 2020 asking if there was a way to replicate the PMM2-CDG drug repurposing workflow at Perlara 1.0 that led to the discovery of epalrestat, which will soon be tested in a Phase 3 clinical trial at Mayo Clinic. I presented the epalrestat story to Prof Klein and his colleagues in April 2020 over Zoom, and a commitment was made to move forward with the project. We selected two ultra-rare metabolic diseases for disease modeling in yeast followed by drug repurposing screens. We began working with two motivated affected families who were both located at that time in the Bay Area. I will present the results of one of those projects today; the other project supported by an ALG11-CDG family is still in progress.
MEPAN syndrome is caused by recessive mutations in the MECR gene, which encodes an enzyme called trans-2-enoyl-CoA reductase, which is subunit of a multi-protein complex called the mitochondrial fatty acid synthase, or mtFAS. Dr. Tom Wald led the project from the outset and was joined by Kelly Pan in December 2020. Dr. Jessica Lao, who led the yeast team at Perlara 1.0, hopped aboard as a consultant in early 2021.
By the end of last summer, Tom and Kelly had reproduced the results of the seminal Heimer et al., paper, which describes MEPAN for the first time as a new neurometabolic disease and characterizes six pathogenic variants in MECR.
Here are the data that Heimer et al published back in 2016 when they created two MEPAN yeast avatars:
Heimer et al modeled two MECR mutations in yeast cells in simple but elegant growth experiments: c.695G>A, which results in the G232E missense variant; c855T>G, which results in the Y285* nonsense variant. Notice above that in the original Heimer et al paper, c.695G>A/G232E exhibits modest growth at the highest cell concentration on lactate and scant growth on glycerol. At the same time, c855T>G/Y285* does not show any sign of growth at the highest cell concentration on either lactate or glycerol.
Tom and Kelly independently replicated that human MECR (HsMECR) restores the ability of yeast cells lacking ETR1 (∆ETR1), the yeast version of HsMECR, to grow under respiratory conditions. And as was previously shown by Heimer et al., HsMERC rescues growth on lactate better than growth on glycerol:
Tom and Kelly also independently replicated the Heimer et al results with G232E and Y285*, and added three additional MEPAN yeast avatars: two missense variants Y285C and R258W; the premature stop Asn83Hisfs∗4 (AsnHis4) truncated variant. In total, five MEPAN yeast avatars were assessed for growth on lactate or glycerol, resulting in the following order from least to most severe:
Y285C > G232E > R258W > Y285* = AsnHis4 = ∆ETR1
Assay optimization took some twists and turns until Jez Revalde from the UCSF Small Molecule Discovery Center had the insight to implement BacTiter-Glo, a more sensitive luminescence-based readout of yeast growth in contrast to the conventional absorbance-based optical density readout (e.g., OD600) of yeast growth that gave variable results due to the extreme slow growth in liquid media with glycerol as the sole carbon source — up to 5 days of incubation which resulted in inconsistent evaporation across a 384-well plate and data-distorting edge effects. Switching from an absorbance readout to a luminescence readout not only increased assay sensitivity by several orders of magnitude but also preserved the rank order of growth-defect severity that was observed above on solid media, as shown in these plots of relative luminescence units (RLU) over time from MEPAN yeast avatars grown in liquid media in 384-well plates:
Jez ran high-throughput screening optimization experiments to determine the incubation time and other variables for the drug repurposing screen. The same rank order of MECR variant growth-defect severity was recapitulated. Y285C is less severe than G232E, which is less severe than R258W, which is nearly indistinguishable from Y285*, AnsHis4 and a whole-gene knockout (KO) of ETR1:
After several rounds of screening earlier this year, 26 compounds were reordered as fresh powder stocks and prepared for dose-response experiments. One top class of hits emerged from the hit validation stage: echinocandins. Echinocandins are a well-known and safe class of antifungal drugs. Because they are cyclic peptides, they are not orally bioavailable and must administered by IV injection.
G232E and Y285C are two protein-destabilizing missense mutations. We hypothesize that echinocandins interact with and insert their acyl tails into mitochondrial membranes and thereby exert a pharmacological chaperone effect on mtFAS. Echinocandins have also been shown to affect the activity of mitochondrial electron transport chain complexes.
With help from Danny Miller, MEPAN researchers are preparing to test echinocandins on MEPAN patient fibroblasts and assess rescue of respiratory function. Stay tuned for the results and check out MEPAN Foundation for updates as well.