Apoptolidin family glycomacrolides target leukemia through inhibition of ATP synthase

Cancer cells have long been recognized to exhibit unique bioenergetic requirements. The apoptolidin family of glycomacrolides are distinguished by their selective cytotoxicity towards oncogene transformed cells, yet their molecular mechanism remains uncertain. We used photoaffinity analogs of the apoptolidins to identify the F1 subcomplex of mitochondrial ATP synthase as the target of apoptolidin A. CryoEM of apoptolidin and ammocidin-ATP synthase complexes revealed a novel shared mode of inhibition that was confirmed by deep mutational scanning of the binding interface to reveal resistance mutations which were confirmed using CRISPR-Cas9. Ammocidin A was found to suppress leukemia progression in vivo at doses that were tolerated with minimal toxicity. These studies reveal that OXPHOS dependent cancers are vulnerable to ATP synthase inhibition by apoptolidin family glycomacrolides compounds. Overall design: The goal of this experiment was to identify mutations within the glycomacrolide binding site on the human ATP synthase that conferred resistance to apoptolidin and ammocidin. A library of ATP synthase mutants (420 total genes: ATP5F1A, ATP5F1B, ATP5F1C) was generated in vitro using nicking mutagenesis on a attB containing plasmid and transferred into HEK293 cells expressing a Bxb1 attP 'landing pad,' such that each cell can express one protein variant. After selection for cells which had undergone successful recombination, each pool of mutants was split into 6 wells and treated with different doses of ammocidin A, apoptolidin A, or vehicle for 4 days on, 3 days off, for two weeks. Resistant mutants were expanded, genomic DNA extracted, and the transgene was amplified by PCR and subjected to next generation sequencing. The mutant library was initially validated by amplifying the transgene from the plasmid to verify the presence of the targeted mutations. A wildtype library (WT) was included as a negative control at both the library validation stage and in cells. A single pooled library was transfected into three separate dishes containing HEK293T-LP-neg (empty landing pad) cells to create three separate biological replicates. After selection for successful recombination each pool was treated with each of the drug conditions, for a total of three biological replicates for each condition. Each sample was compared to its own 'parent' representing the library after selection for recombined cells prior to drug selection. A vehicle control was passaged for 1 week after selection for recombined cells as a negative control. Cells were also transfected with WT library and treated with compounds, though no cells were recovered from the drug treated conditions except for apoptolidin A, 100 nM.