Deep Learning-Driven Image Analysis for Tracking Nucleolar Morphology

Recent work has discovered that two platinum chemotherapeutics currently in widespread use operate by different cell death mechanisms. Cisplatin, particularly effective against testicular cancer, activates the DNA damage response. In contrast, oxaliplatin, a component of frontline colon cancer treatments, induces nucleolar stress and has also been linked to immunogenic cell death. The nucleolus is highly sensitive to chemotherapeutic stress by select agents, but its amorphous structure makes high-throughput analysis difficult and traditionally dependent on manual processing. We have developed expansion microscopy of the nucleolus to enable high-resolution structure analysis and monitor different states of reorganization upon drug treatments. For high-throughput screening analyses using conventional confocal microscopy, we leveraged Thermo Fisher Scientific’s Invitrogen™ SYTO™ RNASelect™ Red, a bright RNA selective dye that provides clear nucleolar staining. Using this dye, we developed deep learning models that accurately classify stressed versus unstressed cells and quantify nucleolar stress dependent morphological changes. This platform enabled us to track the rapid and irreversible onset of nucleolar stress induced by platinum chemotherapeutics, and to monitor the reversal of stress following treatment with the organic compound Actinomycin D.

Learning Objectives:

1. Understand how platinum chemotherapeutics effect nucleolar stress.

2. Use of an RNA specific dye to monitor nucleolar stress onset and reversal through expansion microscopy.

3. Understand deep learning models developed to describe the stress on cells when treated with platinum chemotherapies.