To identify the kinase s

To identify the kinase(s) mediating H3.3S31 phosphorylation, we devised a 96-well-plate-compatible immunostaining assay, with PQ 401 specific for H3.3S31ph, and utilized it to screen both a Qiagen siRNA library covering 720 human kinases and a kinase inhibitor library (Selleck) containing 194 compounds targeting various human kinases. Both screenings were performed in triplicate, and 80ng/ml nocodazole supplementation was used to synchronize HEK293F cells in the G2/M phase 58h after siRNA transfection or after the addition of kinase inhibitors; immunostaining of H3.3S31ph was performed 14h after the addition of nocodazole (a). Images were collected with an Opera high content imaging system (PerkinElmer) and then analyzed with a Columbus image processing server (PerkinElmer) to quantify the percentage of H3.3S31ph positive cells in each well. Cells that did not undergo nocodazole treatment displayed a low percentage of H3.3S31ph positivity (2.7%) in contrast to nocodazole-treated samples (49% positive;Fig. S1a and b). Treatment with siRNA targeting one of the five kinases (AURKB, WEE1, BUB1B, EPHB2, or TTK) reduced the proportion of H3.3S31ph positive cells to less than 30% (b). Among these, siRNAs targeting AURKB and BUB1B displayed the best effects and reduced the proportion of H3.3S31ph positive cells to less than 5% (b). These results were confirmed by Western blot analysis (Fig. S1c). From the kinase inhibitor library screen, several compounds displayed positive effects (c). Interestingly, several AURKB inhibitors, including Hesperadin, AZD1152-HQPA (Barasertib), and ZM447439, were identified as the top hits and robustly inhibited H3.3S31ph in nocodazole-treated cells (c and S2). Notably, of all the potential hits identified in the abovementioned two independent screens, AURKB was the only shared target. Given that siRNA screening often has a high ratio of false positives due to off-target events and that kinase inhibitors are often cross-reactive with other kinases, we decided to focus our study on AURKB. In addition, because the inhibitor library that we used did not contain a BUB1B inhibitor, we also performed additional experiments on BUB1B, another top hit identified in the siRNA screen. In the Qiagen siRNA library used for the primary RNAi screening, four pairs of siRNAs were pooled for each target kinase. To eliminate potential off-target events, we first synthesized another four pairs of siRNAs targeting AURKB or BUB1B and tested their knockdown efficiency and effect on H3.3S31ph individually. All four pairs of siRNAs targeting AURKB were effective (d), and three pairs of siRNAs that targeted BUB1B effectively also had a role in regulating H3.3S31ph (e). We then established cell lines that stably express siRNA-resistant Turbo-GFP-tagged AURKB or BUB1B; H3.3S31ph signals in these cells became resistant to the corresponding RNAi treatment (f and S3), further confirming the roles of AURKB and BUB1B in positively regulating H3.3S31ph. Given that AURKB and BUB1B downregulation might trigger cell cycle arrest and indirectly impact H3.3S31 phosphorylation, we measured the percentage of cells arrested at the G2/M phase with or without RNAi treatment. Indeed, the downregulation of AURKB reduced the number of cells at the G2/M phase (4N) and induced the accumulation of 8N cells (Fig. S4). These results were consistent with previous observations and suggested cell cycle deregulation upon AURKB knockdown. However, H3.3S31ph positive 4N cells were greatly reduced upon AURKB or BUB1B knockdown (Fig. S4), indicating that indirect events caused by cell cycle alteration cannot be the sole explanation. The results above indicated that AURKB and BUB1B exhibited a causal effect on H3.3S31ph . To test whether they have the corresponding enzymatic activity , we tested recombinant oligo-nucleosomes containing wild-type H3.3 (WT), an H3.3S31A mutant, an H3.3 S10A/S28A double mutant, or an H3.3S10A/S28A/S31A triple mutant as substrates for these enzymes. Western blot analysis using antibodies against H3S10ph and H3.3S31ph indicated that AURKB could phosphorylate nucleosomal H3.3 at both S10 and S31 (g). By contrast, we did not detect H3.3S31 phosphorylation activity for BUB1B (g). Furthermore, BUB1B knockdown also reduced mitotic cells that were positive for H3S10ph, another well-recognized product of AURKB (Fig. S5). These results suggest that BUB1B might function upstream of AURKB, which is consistent with a previous report that showed that BUB1 kinase activity is required for AURKB function .