It is multiple nuclei capture rate. They talk about it in the supplementary section near the bottom of the article. They diluted the processed sample to get an estimated 2000 nuclei per microlitre. They then want the machinery to process individual nuclei, but cant tell from the results if two or more nuclei are processed simultaneously (which would give a mixed signal). That's why they estimate the multiple nuclei rate using a separate experiment with nuclei from a mix of mouse and human cells.