FRET biosensor–based RNAi screen for Aurora B–counteracting phosphatases. (A) Live-cell assay for Aurora B phosphorylation. Live HeLa Kyoto cells stably expressing a histone 2B (H2B)–targeted Aurora B FRET biosensor (Fuller et al., 2008) were excited with 426–440-nm light. Pseudocolored YPet/CyPet emission ratios indicate phosphorylation of the biosensor (low ratios indicate high phosphorylation). Mitotic stages were classified by supervised machine learning (Held et al., 2010), as indicated by colored contours. Nontargeting control siRNA (siControl) was transfected 42 h before imaging. Bar, 50 µm. (B) Enlarged images of mitotic cells. Bar, 10 µm. (C) Cells imaged, as in A, 42 h after transfection of Aurora B–targeting siRNA (siAurora B). Bar, 50 µm. (D and E) Statistical analysis of Aurora B biosensor phosphorylation. YPet/CyPet emission ratios were calculated for individual cells, and all measurements were then normalized (norm.) to the median of interphase control cells. Median (line), quartiles (boxes), and 10–90% data range (error bars) indicate YPet/CyPet ratios at different mitotic stages. (D) Biosensor phosphorylation of control cells transfected with nontargeting siRNA. (E) Aurora B RNAi cells. (F) Screen of an siRNA library targeting a genome-wide set of human phosphatases. FRET biosensor phosphorylation in late anaphase cells was assayed as in A–E. Data points represent mean z scores of YPet/CyPet emission ratios in late anaphase for three different siRNAs targeting per gene and two independent experimental replicates. Low z scores indicate hyperphosphorylation of the FRET biosensor.