Machine vision reveals micronucleus rupture as a potential driver of the transcriptomic response to aneuploidy.

Micronuclei are aberrant nuclear compartments that trap a portion of a cell's chromatin in a distinct organelle separate from the nucleus and are drivers of inflammation, DNA damage, chromosome instability, and chromothripsis. Many of the consequences of micronucleus formation stem from micronucleus rupture: the sudden loss of micronucleus compartmentalization, resulting in mislocalization of nuclear factors and the exposure of chromatin to the cytosol for the remainder of interphase. Micronuclei form primarily from segregation errors during mitosis, errors that also give rise to other, non-exclusive phenotypes, including aneuploidy and chromatin bridges. The stochastic formation of micronuclei and phenotypic overlap confounds the use of population-level assays or hypothesis discovery, requiring labor-intensive techniques to visually identify and follow micronucleated cells individually. In this study, we present a novel technique for automatically identifying and isolating micronucleated cells generally and cells with ruptured micronuclei specifically using a de novo neural net combined with Visual Cell Sorting. As a proof of concept, we compare the early transcriptomic responses to micronucleation and micronucleus rupture with previously published responses to aneuploidy, revealing micronucleus rupture to be a potential driver of the aneuploidy response.