Automated quantification of lipophagy in Saccharomyces cerevisiae from fluorescence and cryo-soft X-ray microscopy data using deep learning

Abstract

Lipophagy is a form of autophagy by which lipid droplets (LDs) become digested to provide nutrients as a cellular response to starvation. Lipophagy is often studied in yeast, Saccharomyces cerevisiae, in which LDs become internalized into the vacuole. There is a lack of tools to quantitatively assess lipophagy in intact cells with high resolution and throughput. Here, we combine soft X-ray tomography (SXT) with fluorescence microscopy and use a deep learning computational approach to visualize and quantify lipophagy in yeast. We focus on yeast homologs of mammalian Niemann Pick type C proteins, whose dysfunction leads to Niemann Pick type C disease in humans, i.e., NPC1 (named NCR1 in yeast) and NPC2. We developed a convolutional neural network (CNN) model which classifies ring-shaped versus lipid-filled or fragmented vacuoles containing ingested LDs in fluorescence images from wild-type yeast and from cells lacking NCR1 ({Delta}ncr1 cells) or NPC2 ({Delta}npc2 cells). Using a second CNN model, which performs automated segmentation of LDs and vacuoles from high-resolution reconstructions of X-ray tomograms, we can obtain 3D renderings of LDs inside and outside of the vacuole in a fully automated manner and additionally measure droplet volume, number, and distribution. We find that cells lacking functional NPC proteins can ingest LDs into vacuoles normally but show compromised degradation of LDs and accumulation of lipid vesicles inside vacuoles. This phenotype is most severe in{Delta} npc2 cells. Our new method is versatile and allows for automated high-throughput 3D visualization and quantification of lipophagy in intact cells.