Quantitative Proteomics of the Mitotic Chromosome Scaffold Reveals the Association of BAZ1B with Chromosomal Axes*

Our quantitative proteomics study determined the protein composition of the mitotic chromosome scaffold. MS results revealed a novel component of the chromosome scaffold, BAZ1B, which was localized to the mitotic chromosome axis. Our results using BAZ1A/B double-KO cells suggest that BAZ1 proteins are essential for timely chromosome condensation at mitosis entry. Graphical Abstract Highlights Quantitative proteomics of mitotic chromosome scaffold isolated from chicken DT40 cells. BAZ1B identified in the isolated mitotic chromosome scaffold localizes to mitotic chromosome axes. BAZ1B knockout caused prophase delay because of altered chromosome condensation timing and impaired mitosis progression. BAZ1B knockout did not affect prometaphase chromosome structure. In mitosis, chromosomes achieve their characteristic shape through condensation, an essential process for proper segregation of the genome during cell division. A classical model for mitotic chromosome condensation proposes that non-histone proteins act as a structural framework called the chromosome scaffold. The components of the chromosome scaffold, such as DNA topoisomerase IIα (TOP2A) and structural maintenance of chromosomes protein 2 (SMC2), are necessary to generate stable mitotic chromosomes; however, the existence of this scaffold remains controversial. The aim of this study was to determine the protein composition of the chromosome scaffold. We used the DT40 chicken cell line to isolate mitotic chromosomes and extract the associated protein fraction, which could contain the chromosome scaffold. MS revealed a novel component of the chromosome scaffold, bromodomain adjacent to zinc finger 1B (BAZ1B), which was localized to the mitotic chromosome axis. Knocking out BAZ1B caused prophase delay because of altered chromosome condensation timing and mitosis progression errors, and the effect was aggravated if BAZ1A, a BAZ1B homolog, was simultaneously knocked out; however, protein composition of prometaphase chromosomes was normal. Our results suggest that BAZ1 proteins are essential for timely chromosome condensation at mitosis entry. Further characterization of the functional role of BAZ1 proteins would provide new insights into the timing of chromosome condensation.

In mitosis, chromosomes achieve their characteristic shape through condensation, an essential process for proper segregation of the genome during cell division. A classical model for mitotic chromosome condensation proposes that non-histone proteins act as a structural framework called the chromosome scaffold. The components of the chromosome scaffold, such as DNA topoisomerase II␣ (TOP2A) and structural maintenance of chromosomes protein 2 (SMC2), are necessary to generate stable mitotic chromosomes; however, the existence of this scaffold remains controversial. The aim of this study was to determine the protein composition of the chromosome scaffold. We used the DT40 chicken cell line to isolate mitotic chromosomes and extract the associated protein fraction, which could contain the chromosome scaffold. MS revealed a novel component of the chromosome scaffold, bromodomain adjacent to zinc finger 1B (BAZ1B), which was localized to the mitotic chromosome axis. Knocking out BAZ1B caused prophase delay because of altered chromosome condensation timing and mitosis progression errors, and the effect was aggravated if BAZ1A, a BAZ1B homolog, was simultaneously knocked out; however, protein composition of prometaphase chromosomes was normal. Our results suggest that BAZ1 proteins are essential for timely chromosome condensation at mitosis entry. A major challenge in the field of mitosis research is elucidation of the mechanism through which chromosomes achieve the necessary 10,000-fold shortening of DNA and thus acquire the characteristic structure essential for proper mitosis. A classical model proposes that non-histone proteins act as a structural framework during the formation of mitotic chromosomes (1). This "chromosome scaffold" was hypothesized to be an insoluble biochemical fraction that could be obtained from chromosomes after most of the DNA and proteins were solubilized (2). The major components of this fraction are structural maintenance of chromosomes protein 2 (SMC2) 1 and DNA topoisomerase II␣ (TOP2A) (3), which is thought to affect chromosome condensation during late mitosis (4,5).
A breakthrough in understanding of the chromosome condensation mechanism was the identification of condensin, a pentameric protein complex including SMC2 (6 -8). The purified chromatin-associated protein complex from Xenopus egg extracts, which contains chromosome-associated protein (CAP)-E/SMC2, CAP-C/SMC4, CAP-D2, CAP-G, and CAP-H, showed chromosome condensation activity in vitro (6,9) by introducing positive supercoils into relaxed plasmid DNA in the presence of topoisomerase I (9). Furthermore, it has been shown that the condensin complex localizes on the mitotic chromosome axis in many vertebrate species (10,11).
When the second condensin complex (condensin II: SMC2, SMC4, CAP-D3, CAP-G2, and CAP-H2) was discovered, the canonical condensin complex was retroactively named condensin I (12). Both condensin complexes localize to the mitotic chromosome axis but show alternate distribution (12,13). Thus, condensin II exists predominantly in the nucleus during interphase, whereas condensin I is sequestered in the cytoplasm and gains access to chromosomes only after nuclear envelope breakdown (NEB) in prometaphase (13). These findings suggest that the two condensin complexes act sequentially to initiate the assembly of mitotic chromosomes (14,15). Condensin II is involved in DNA repair during interphase through association with several chromosomal proteins and chromosome condensation during mitotic entry (13,16). Moreover, TOP2A and kinesin family member 4A (KIF4A), both included in the chromosome scaffold fraction, show alternate localization on the mitotic chromosome axis (17,18). However, the existence and functional significance of such a chromosome scaffold is highly controversial, and another widely accepted model proposes that chromosomes are formed solely through a hierarchy of chromatin coiling events (19).
Several studies have demonstrated the involvement of bromodomain adjacent to zinc finger 1B (BAZ1B) in heterochromatin remodeling (20,21). The gene encoding BAZ1B is also known as the William syndrome transcription factor (WSTF) because of its initial identification as a hemizygously deleted gene in patients with the disease (22). BAZ1B may form a complex with the nucleosome-dependent ATPase, imitation switch (ISWI)/sucrose non-fermenting protein (SNF)-related matrix-associated actin-dependent regulator of chromatin subfamily A member 5 (SMARCA5) (20,21). BAZ1B depletion was reported to affect the localization of heterochromatin protein 1 (HP1) and histone H3 with trimethylated lysine-9 (HH3-K9me3) (23). Furthermore, BAZ1B exhibits tyrosine-protein kinase activity during DNA double-strand break (DSB) repair by phosphorylating Tyr-142 on histone H2A.X (HH2A.X-pY142 or ␥-H2A.X), a protein that recruits the MRN complex, including Mre11, Rad50, and Nbs1, during initial DSB processing (24,25). Moreover, it has now become clear that BAZ1B forms a complex with topoisomerase I and SMARCA5 during the S phase and is associated with the progression of DNA replication forks (26).
In addition, the human genome contains the BAZ1A gene (also named ATP-utilizing chromatin assembly factor 1, ACF1), which encodes a protein showing 21% amino acid sequence homology with BAZ1B. Similar to BAZ1B, BAZ1A interacts with SMARCA5 to form the chromatin accessibility complex (CHRAC), which is required for DSB repair through recruitment of Ku proteins in human cells (27). Moreover, BAZ1A and BAZ1B were identified in mitotic chromosomes assembled in Xenopus egg extracts (14,28). However, the functions of both BAZ1A and BAZ1B in mitosis remain unclear.
In this study, we used MS to determine the protein composition of the chromosome scaffold in chicken DT40 cells. To our knowledge, this is the first quantitative proteomic analysis showing that BAZ1B is present in the mitotic chromosome scaffold along with previously identified components such as TOP2A, SMC2, and KIF4A. Our results suggest that BAZ1B and its homolog BAZ1A co-regulate the timing of chromosomal condensation prior to mitotic entry.
U2OS or HeLa (Kyoto) cells in the exponential growth phase were seeded onto coverslips and grown overnight at 37°C and 5% CO 2 in Dulbecco's Modified Eagle medium (DMEM, Wako Pure Chemical Industries Ltd.) supplemented with 10% FBS, 100 U/ml penicillin, and 100 g/ml streptomycin.
Mitotic chromosomes from three independent preparations were pooled together and solubilized in phase-transfer surfactant (PTS) buffer (31).
Isolation of Histone-depleted Mitotic Chromosomes-Histone-depleted mitotic chromosomes were prepared as previously described (2,29) and transferred to non-EDTA polyamine buffer containing 0.375 mM spermidine and 0.03% N,N-dimethyldodecylamine Noxide. DNA was digested by adding 40 g/ml micrococcal nuclease and 2 mM calcium chloride for 20 min. Chromosomes were treated with 0.1 mM copper sulfate under nitrogen gas flow for 10 min and diluted with an equal volume of 1ϫ TEE buffer (1 mM triethanolamine-HCl, pH 8.5, and 0.2 mM Na-EDTA). An equal volume of 2ϫ lysis buffer (20 mM Tris-HCl, pH 9.0, 20 mM Na-EDTA, 0.06% N,N-dimethyldodecylamine N-oxide, and 4 M NaCl) was immediately added, and the mixture was incubated for 20 min. The histone-depleted mitotic fraction was obtained in the pellet after centrifugation at 10,000 ϫ g for 5 min.
ing in the purification of the background fraction (2,29). This procedure enabled us to perform quantitative comparison between the histone-depleted mitotic chromosome fraction and the background because it removed many contaminants from the list of potential chromosome scaffold proteins (supplemental Fig. S1A).
Trypsin Digestion-Proteins associated with mitotic chromosomes were extracted using PTS (32), and 100-g aliquots were digested with 1 g endoproteinase Lys-C (Wako Pure Chemical Industries Ltd.) and 0.4 g trypsin (Promega, Fitchburg, WI) in 2 mM DTT and 10 mM IAA for 16 h. Digestion was stopped with 1% trifluoroacetic acid followed by desalting with an SDB-XC StageTip (33).
Mass Spectrometry-The digested peptides or enriched phosphopeptides were analyzed on a Q Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometer or an Orbitrap Fusion Tribrid Mass Spectrometer (Thermo Fisher Scientific, Bremen, Germany) using the acquisition software for the fusion. MS2 spectra were obtained using the Orbitrap analyzer on the Orbitrap Fusion Mass Spectrometer. For LC-MS/MS analysis, we used a Dionex -Precolumn (5 m particle size, 300 m inner diameter ϫ 5 mm length) and an Acclaim PepMap100 C18 separation column (3 m particle size, 75 m inner diameter ϫ 150 mm length) as described in a previous report (34). Peptides were eluted with a nonlinear gradient of 4 -30% buffer B (0.2% formic acid [FA], 5% DMSO in acetonitrile) in buffer A (0.2% FA, 5% DMSO in double-distilled water) applied at a flow rate of 250 nL/min over 123 min. After each elution, the column was washed with 96% buffer B and re-equilibrated with buffer A. All parameters used in MS analysis are shown in supplemental Table S1.
Peptide Identification and Quantification-For identification of proteins and phosphorylation sites, peak lists were created using Andromeda (35) based on recorded fragmentation spectra searched against the Gallus gallus protein database in UniProt (release 2013_07) and our in-house chicken database. Spectra were also searched against the Homo sapiens protein database in UniProt (release 2016_11). Precursor mass tolerance was 10 ppm, fragment ion mass tolerance was 0.02 Da, and trypsin specificity was strict (C-terminal cleavage of Lys or Arg but not before Pro), allowing up to four missed cleavages. The results of stable isotope labeling with amino acids in cell culture (SILAC) were quantified using MaxQuant 1.4.0.3 or 1.5.6.5. Details of the parameters are given in supplemental Table S2. MS proteomic data and all other parameters used in Max-Quant have been deposited in the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the JPOST partner repository (data set identifiers: PXD007269, PXD008429, and PXD008696) (36 -38). All protein and phosphopeptide quantification data are also presented in supplemental Tables S3-S7.
Plasmid Construction-The ORFs of human BAZ1A and BAZ1B were amplified by PCR from total cell RNA using Superscript II and KOD polymerase and inserted into the entry vector (pENTR™ 4 Dual Selection Vector; Thermo Fisher Scientific) between XmnI and NotI sites. Vectors for overexpression in human cells were generated via LR clonase-mediated recombination between the entry vector and pDEST131NGFP (39). The pTORA14BAZ1A plasmid for CRISPR-mediated mutagenesis was obtained by inserting the double-strand oligo DNA fragment into the AgeI site (blunted by Mung Bean Nuclease) of pTORA14 constructed from the U6gRNA-Cas9 -2A-GFP plasmid (CAS9GFPP-1EA, Sigma-Aldrich). The guide RNA target sequence in the DNA fragment was 5Ј-TCTGTCTCACAAACGGCTTT-3Ј.
Mitotic Chromosome Spread-The cytospin technique was used to make chromosome spreads with preserved chromosome structure. Briefly, mitotic cells were synchronized by 0.1 g/ml nocodazole treatment for 12 h and collected by tapping culture dishes. Cells were then treated with 75 mM KCl at 37°C for 15 min, and 10,000 cells were deposited onto coverslips by centrifugation at 800 rpm for 3 min.
Experimental Design and Statistical Rationale-Mitotic chromosomes were isolated from chicken cells or human cells treated with nocodazole after six cell divisions to incorporate labeled arginine and lysine. Each sample for MS analysis was generated by combining three individual preparations of isolated mitotic chromosomes or mitotic chromosome scaffolds; two biological replicates for each cell type were analyzed. LC-MS/MS data were searched against UniProt database using MaxQuant 1.5.6.5 and statistically evaluated using one sample test in Perseus 1.6.1.3 (43). Continuous variables were expressed as the mean Ϯ S.D., variance was evaluated by F-test, and the significance of the results was analyzed by Welch's t test in R. Differences were considered significant at p Յ 0.05.

Protein Composition of the Mitotic Chromosome Scaffold-
Previously, we reported that SILAC ratios were useful for distinguishing chromosomal protein classes (30). Here, we demonstrated their utility for identifying chromosome scaffold proteins. It was shown that mitotic chromosome scaffold proteins such as ScI (TOP2A) and ScII (SMC2) were accumulated in the fraction of histone-depleted mitotic chromosomes (29) and that the addition of 1% ␤-mercaptoethanol dissociated the chromosome scaffold (Fig. 1A) (2,29). Therefore, to obtain accurate results, we compared proteins in the isolated scaffold with those in the background obtained by chromosome scaffold isolation with 1% ␤-mercaptoethanol. This procedure allowed removal of many contaminating proteins from the list of putative chromosome scaffold proteins (forward [FW] label: lysine-8 and arginine-10 [heavy] -chromosome scaffold; lysine-0 and arginine-0 [light] -background; supplemental Fig. S1A-S1D).
To examine reproducibility of the results, we used the same strategy with alternative labeling (reverse [RV] label: heavybackground, light -chromosome scaffold). Accordingly, 786 proteins could be quantified more than once in the FW and RV labeling experiments. Labeling revealed that TOP2A, KIF4A, condensin subunits, PLK1 (a condensation regulation factor localized to the chromosome axis) (44), and BAZ1B were highly concentrated in the mitotic chromosome scaffold fraction (Fig. 1B, supplemental Table S3).
Proteins Enriched in the Mitotic Chromosome Scaffold Versus Mitotic Chromosomes-We expected that core components of the chromosome scaffold, such as TOP2A and SMC2, should be more concentrated in the mitotic chromosome scaffold (insoluble) fraction compared with the soluble fraction from which the chromosome scaffold was removed. Therefore, we combined histone-depleted mitotic chromosomes from DT40 cells grown in heavy SILAC medium with an equal protein amount of the soluble fraction from a parallel culture grown in light SILAC medium (FW label; RV labeling was performed to investigate reproducibility; Fig. 1A, supplemental Fig. S1E). As a result, 739 proteins could be quantified more than once in the FW and RV experiments. Ranking these proteins according to their enrichment in the chromosome scaffold fraction revealed that not only condensin subunits and TOP2A but also several centromere/kinetochore proteins were enriched in the chromosome scaffold, whereas condensin II subunits were less abundant (Fig. 1C, supplemental Table S3). Furthermore, BAZ1B was as enriched as other scaffold components in histone-depleted mitotic chromosomes (Fig. 1C). Interestingly, even core histones were not observed in the histone-depleted mitotic chromosome fraction (Fig. 1A, 1C); however, we detected anomalous levels of histone macroH2A (Fig. 1C, supplemental Fig. S1F), which could possibly be involved in linking histones and chromosome scaffolds during chromosome condensation.
Protein Composition in Histone-depleted Mitotic Chromosomes of SMC2 ON and SMC2 OFF Cells-SMC2 is essential for the formation of the mitotic chromosome scaffold (11). Therefore, we used the histone-depleted mitotic chromosome fraction isolated from SMC2 KO cells as a possible second background fraction. Previously, we compared protein composition in mitotic chromosomes of SMC2 ON and SMC2 OFF cells (45). Here, we assessed protein abundance in histonedepleted mitotic chromosomes obtained from cells expressing SMC2 or not. For this, we combined equal amounts of histone-depleted mitotic chromosomes isolated from SMC2 ON cells in heavy SILAC medium and SMC2 OFF cells in light SILAC medium (supplemental Fig. S2A). Consistent with the results of the previous experiment, TOP2A, KIF4A, and condensin subunits were not concentrated in histone-depleted mitotic chromosomes from SMC2 OFF cells; however, BAZ1B was concentrated (supplemental Fig. S2B, S2C). This finding suggests that, unlike other scaffold proteins, BAZ1B associates with the chromosome scaffold independently of condensin.
BAZ1B Localizes to the Mitotic Chromosome Axis-To validate our quantitative proteomic data, we isolated histonedepleted mitotic chromosomes (chromosome scaffolds) from HeLa cells and assessed the levels of BAZ1B by immunoblotting (Fig. 1D). Furthermore, to ensure that important protein mechanisms were conserved, we expressed GFP-tagged human BAZ1B in human epithelial U2OS cells and assessed its localization during mitosis. A previous study showed localization of GFP-tagged condensin subunits SMC2-GFP, CAP-H: GFP, and CAP-D:GFP to the chromosome axis (18). In the control experiment, GFP-␤-glucuronidase did not show any specific localization on mitotic chromosomes (Fig. 1E), whereas GFP-BAZ1B was confirmed to be localized to the chromosome axis (Fig. 1F), strongly suggesting that BAZ1B is associated with the chromosome scaffold during mitosis. Moreover, GFP signals were stronger in the centromeric and telomeric regions on the chromosome axis (Fig. 1F), indicating that BAZ1B localizes preferentially to the heterochromatin region.
BAZ1 Deficiency Alters Progression Through Mitosis-To investigate whether BAZ1B plays a functional role in mitotic chromosomes, we generated mutant BAZ1B KO HeLa cells using CRISPR/Cas9; the lack of endogenous BAZ1B was confirmed by immunoblotting and DNA sequencing ( Fig. 2A,  supplemental Fig. S3A). Compared with WT cells, BAZ1B KO cells grew only slightly slower (Fig. 2B, supplemental Fig.  S3B), which prompted us to investigate why BAZ1B KO cells were not more affected despite the known localization of BAZ1B to the chromosome axis. For this, we examined the role of BAZ1B homolog BAZ1A by generating BAZ1A KO cells and BAZ1A/BAZ1B double-KO cells lacking endogenous BAZ1A or both BAZ1 proteins, respectively ( Fig. 2A, supplemental Fig. S3C). BAZ1 and BAZ1B share the same structure and interact with each other and with SMARCA5 (21, 46); however, BAZ1A has two instead of three WSTF-HB1-Itc1p-MBD9 (WHIM) DNA-binding domains (Fig. 2C). Like BAZ1B, GFP-BAZ1A was confirmed to be present in the nuclei in interphase (supplemental Fig. S4A, (Fig. S4B), but it was difficult to detect its localization to the chromosomes and chromosomal axis in mitosis (supplemental Fig. S4C). None-FIG. 1. Proteomics of proteins concentrated in the mitotic chromosome scaffold. A, Chromosomes, histone-depleted mitotic chromosomes, or histone-enriched (soluble) fraction isolated with (ϩ) or without (Ϫ) 1% ␤-mercaptoethanol (␤ME) from DT40 cells were analyzed by immunoblotting using antibodies against SMC2, Top2A, or INCENP and Coomassie Brilliant Blue (CBB) staining for core histones. Each sample corresponded to the same number of isolated chromosomes. B, Identification of 511 overlapped proteins in SILAC forward versus reverse labeling experiments comparing histone-depleted mitotic chromosomes isolated with and without ␤ME addition. C, Identification of 352 overlapped proteins in forward versus reverse labeling experiments comparing histone-depleted mitotic chromosomes and total mitotic chromosomes. D, Chromosomes, histone-depleted mitotic chromosomes, or soluble fraction isolated with (ϩ) or without (Ϫ) 1% ␤ME from HeLa cells were analyzed by immunoblotting using antibodies against TOP2A, BAZ1B, or BAZ1A and by CBB staining. Each sample corresponded to the same number of isolated chromosomes. E, F, GFP-␤-glucuronidase (GFP-Gus, E) and GFP-BAZ1B (F) were expressed in human epithelial U2OS cells and analyzed for intracellular localization; (a, b) metaphase and (c) anaphase. GFP-BAZ1B and GFP-Gus are shown in green, and DAPI-stained nuclei are red. Scale bars, 5 m. High-magnification immunofluorescence images of white boxes are on the right. theless, immunoblotting clearly showed that BAZ1A existed in isolated mitotic chromosomes but not in chromosome scaffolds (Fig. 1D). BAZ1A KO and WT cells exhibited similar growth; however, BAZ1A/B double-KO cells grew significantly slower than WT cells, but GFP-BAZ1B overexpression partially restored normal growth in double-KO cells (Fig. 2B,  (supplemental Fig. S3B). Mitotic profiles revealed that the percentage of cells at prometaphase tended to be higher in the double-KO group (47%) compared with the WT group (37%); at the same time, the proportion of cells at anaphase and telophase decreased in the double-KO group compared with the WT group (7% versus 13% in anaphase and 9% versus 15% in telophase, respectively; Fig. 2D). However, GFP-BAZ1B overexpression partially restored segregation in double-KO cells (Fig. 2D). On the other hand, BAZ1A KO or BAZ1B KO cells did not show significant changes in the mitotic profile compared with WT cells (Fig. 2D). Therefore, to focus on the delay of chromosome alignment in double-KO cells, we next analyzed prometaphase arrest by synchronizing cells at the prometaphase stage by monastrol treatment for 4 h and then placing them into monastrol-free medium supplemented with MG132. Monastrol treatment led to the arrest of ϳ15% of total cells that were in mitosis, 80% of which were in early prometaphase (Fig. 2E). However, after 1 h in monastrol-free medium, the percentages of BAZ1B KO and BAZ1A/B double-KO cells in early prometaphase were significantly higher compared with that of WT cells (42 and 56% versus 21%, respectively), whereas the percentages of cells in metaphase were significantly lower (31 and 17% versus 63%, respectively; Fig. 2E). At the same time, no significant differences were observed between BAZ1A KO and WT cells (Fig.  2E). These results suggest that although BAZ1B is important for mitosis progression, endogenous BAZ1A alleviates the BAZ1B KO phenotype, whereas BAZ1A/B double-KO aggravates it.
Segregation errors were detected in less than 6% of WT, BAZ1A KO, and BAZ1B KO cells but in 15% of double-KO cells (Fig. 2F, 2G); however, GFP-BAZ1B overexpression partially restored segregation in double-KO cells (Fig. 2F). It has been shown that segregation errors are frequently observed in cells deficient in condensin I or II, which are localized to the mitotic chromosome axis (13,47,48), suggesting that BAZ1 proteins may perform a similar function. Furthermore, timelapse imaging of histone H2B-mRFP-expressing cells showed a slight increase in the time between estimated NEB and anaphase onset in double-KO cells (33.8 min) compared with WT, BAZ1A KO, and BAZ1B KO cells (28.6, 29.8, and 29.4 min, respectively; Fig. 2H, 2I and supplemental Movie S1-S5). Furthermore, double-KO cells showed a significant increase in segregation errors at anaphase compared with WT cells (22.6% versus 1.5%, respectively; Fig. 2I, 2J).
Interestingly, the percentage of double-KO cells in prophase increased significantly following monastrol treatment, which was also observed in nonsynchronized cells (Fig. 2D,  2E), although the difference was not significant.
BAZ1 Deficiency Causes Mitotic Entry Delay-We further investigated the prophase delay in BAZ1A/B double-KO cells by immunofluorescence using the anti-histone H3 phosphoserine 10 (H3pS10) antibody and DAPI to visualize chromosome morphology before NEB. However, phosphorylation of histone H3 occurs and is completed in the preceding prophase (49,50). Therefore, it should be noted that two types of chromosome morphology could be present in prophase in this study: relaxed chromosomes marking early prophase and condensed chromosomes marking late prophase (Fig. 3A-3C). Late prophase, in which individual condensed chromosomes could be clearly recognized, was distinguishable from early prophase; however, about 34% of cells had chromosomes of both types, marking the third type: middle prophase (Fig. 3A-3C). The percentage of WT cells in early prophase (36%) was lower, whereas that in late prophase (30%) was significantly higher compared with those of BAZ1A/B double-KO cells (58 and 18%, respectively), and a similar tendency was observed for BAZ1B KO cells (Fig. 3D).
We then investigated whether prophase delay occurred in synchronized cells. WT or KO cells were arrested in the G2 phase using the Cdk 1 inhibitor RO-3306 (51), and prophase profiles were observed 10, 20, or 30 min after switching to RO-3306-free medium. At 10 min, 90% of all cells were in early prophase; however, at 20 and 30 min, there were significantly more cells at early prophase in BAZ1B KO or BAZ1A/B KO groups compared with WT ( Fig. 3E). At the same time, there were significantly fewer cells in middle prophase for the BAZ1A/B KO group and in late prophase for both BAZ1A/B KO and BAZ1B KO groups compared with WT at 30 min after RO-3306 release (Fig. 3E). These results indicate that the absence of BAZ1B delays chromosome condensation onset, and the effect is further aggravated if BAZ1A is absent as well, suggesting that BAZ1A and BAZ1B might  co-regulate the chromosomal condensation timing at early prophase.
Proteomics Analysis of Mitotic Chromosomes Lacking BAZ1 Proteins-To further elucidate the role of BAZ1 in mitotic chromosomes, we compared the association of all chromosomal proteins with mitotic chromosomes in cells with and without BAZ1B or BAZ1A/B using our previously developed quantitative proteomic approach (30,52,53). Mitotic chromosomes were isolated after mixing equal amounts of WT cells cultured with lysine-8 and arginine-10 (heavy) and BAZ1B KO or BAZ1A/B double-KO cells cultured with lysine-0 and arginine-0 (light) (supplemental Fig. S5A), and their proteomes were analyzed by MS. A total of 1243 and 1290 proteins were identified based on comparison between WT and BAZ1B KO or BAZ1A/B double-KO cells, respectively, in two separate experiments.
The lack of BAZ1B decreased the association of SMARCA5 with mitotic chromosomes in BAZ1B KO cells to ϳ30% of that in WT (Fig. 4A). Given that SMARCA5 interacts with BAZ1B in the S phase, our results indicate that the SMARCA5-chromosome association depends on BAZ1B. Moreover, although topoisomerase I is part of the BAZ1B/SMARCA5 complex in the S phase (26), its association with mitotic chromosomes was unchanged in BAZ1B KO or BAZ1A/B double-KO cells compared with that in WT cells, suggesting that topoisomerase I associates with chromosomes through the BAZ1B/ SMARCA5 complex independently of BAZ1B (Fig. 4A, 4B).
No dramatic changes were observed in the mitotic chromosome scaffold components (condensin I and II, TOP2A, and KIF4A) in BAZ1A/B double-KO cells (Fig. 4B), suggesting that their association with mitotic chromosomes does not depend on BAZ1 proteins. K i -67, a component of the mitotic chromosome periphery and one of the most abundant proteins in mitotic chromosomes (30) that regulates their organization via interaction with protein phosphatase 1 (PP1) (54), showed a tendency to decrease in BAZ1A/B double-KO cells compared with WT; however, the changes did not reach statistical significance in a volcano plot (Fig. 4B). Moreover, immunofluorescence staining with anti-K i -67 antibodies did not reveal significant differences between WT and BAZ1A/B double-KO cells (Fig. 4C, 4D).
In addition, we analyzed protein phosphorylation in mitotic chromosomes by enriching for phosphopeptides using TiO 2based HAMMOC (32,55) and subjecting them to LC-MS/MS (supplemental Fig. S5A). In SILAC experiments comparing WT and BAZ1B KO cells or WT and BAZ1A/B double-KO cells, we identified 1536 or 1446 phosphorylation sites in proteins associated with isolated mitotic chromosomes, respectively. However, we were unable to identify phosphorylation sites directly targeted by BAZ1B or BAZ1A in mitotic chro-mosomes, even though the phosphorylation of chromosomal proteins was mildly reduced in BAZ1A/B double-KO cells (supplemental Fig. S5B, S5C).
Mitotic Chromosome Axes Remain Intact in BAZ1A/B Double-KO Cells-Although the lack of BAZ1A/B resulted in a mild phenotype characterized by aberrant chromosome structure in prometaphase, we did not observe morphological abnormalities in mitotic KO cells after visualization of the chromosome axis and its components by immunofluorescence (Fig. 4E). Next, we quantified CAP-G, CAP-H2, and TOP2A in mitotic chromosome axes of KO cells relative to co-spread HeLa cells overexpressing histone H2B-mRFP fusion protein; the design of the experiment is shown in Fig. 4F. The presence of CAP-G and CAP-H2 was unchanged in all KO cells, whereas that of TOP2A tended to increase in double-KO cells compared with WT cells (Fig. 4G). These results suggest that BAZ1A and BAZ1B did not significantly affect chromosome axes in prometaphase.
To confirm a weak effect of BAZ1A and BAZ1B depletion on mitotic chromosome structure, we performed an intrinsic metaphase structure (IMS) assay (supplemental Fig. S6A-S6C), in which chromosome unfolding to 10-nm structures is induced by removal of divalent cations, and then refolding is induced by Mg 2ϩ addition (54,56). Consistent with previous reports that mitotic chromosomes lacking SMC2 or KIF4A were severely impaired in the IMS assay even though they appeared morphologically normal (11,18), we observed the same effect in chicken DT40 cells (supplemental Fig. S6B). Our experiment clearly showed that chromosomes in ϳ90% of BAZ1A or BAZ1B KO cells efficiently regained their normal morphology after the unfolding-refolding cycle (supplemental Fig. S6C), suggesting that BAZ1A and BAZ1B are not individually required for maintaining the intrinsic structure of mitotic chromosomes. Moreover, mitotic chromosomes lacking both BAZ1A and BAZ1B were slightly affected as evidenced by the fact that 80% of double-KO cells regained normal chromosome morphology, and GFP-BAZ1B transfection partially rescued this phenotype (supplemental Fig. S6C). The effect of BAZ1A/BAZ1B KO was significantly weaker compared with SMC2 KO in DT40 cells, suggesting that mitotic chromosomes retained their structure in synchronized prophase irrespectively of BAZ1A/B presence. DISCUSSION Components of the chromosome scaffold play an important role in the formation of structurally stable mitotic chromosomes. In this study, we used MS to determine the protein composition of chromosome scaffolds in DT40 cells. Using quantitative proteomics, we identified proteins in the histone- depleted mitotic chromosome fraction, which were associated with the mitotic chromosome scaffold, including those not previously reported as scaffold components. One of them, BAZ1B, has been shown to serve as a transcription factor or tyrosine kinase (24,25). Consistent with our findings, BAZ1B and BAZ1A were shown to be abundant in mitotic chromosomes assembled in Xenopus egg extracts (14,28).
BAZ1B localization to chromosomal axes in mitosis confirmed the results of quantitative proteomics obtained herein. Although BAZ1B KO was not fatal, it decreased cell growth, and the effect was stronger for cells with double KO of BAZ1A and BAZ1B, suggesting that both BAZ1 proteins have similar functions. The chromosomal localization of GFP-tagged BAZ1A in mitosis was difficult to determine, but immunoblotting using isolated mitotic chromosomes clearly demonstrated chromosomal association of endogenous BAZ1A in mitosis. However, BAZ1A/B double-KO cells did not exhibit a clear mitotic phenotype. In a cell-free system based on Xenopus egg extracts, depletion of SMARCA5/ISWI, which forms a complex with both BAZ1A and BAZ1B, affected nucleosome spacing but not chromosome assembly (28). This study suggested that the SMARCA5-BAZ1 complex enhances proper interactions between chromatin and proteins involved in mitotic chromosome structure by catalyzing their nucleosomal fluidity (28). Our results also revealed a longer duration between phosphorylation of H3pS10 in whole nuclei and complete chromosome condensation in BAZ1B KO or BAZ1A/B double-KO cells. Based on these findings, we propose a model in which BAZ1 proteins expeditiously enhance chromatin association of other chromosome scaffold proteins in early prophase by regulating nucleosome spacing. This model is supported by the localization of BAZ1A and BAZ1B to the nuclei before mitotic entry. However, our proteomic and immunofluorescence analyses suggest an independent association between BAZ1A/B and known chromosome scaffold proteins such as condensin, TOP2A, and KIF4. Thus, the lack of BAZ1A and BAZ1B caused chromosome condensation delay, which could be recovered until prometaphase. These findings implied that the function of BAZ1 proteins involves regulation of proper chromosome condensation timing after phosphorylation of histone H3S10 rather than organizing mitotic chromosome structure; however, chromosome missegregation was detectable in BAZ1A and BAZ1B double-KO cells at the end of anaphase. BAZ1A/B double-KO cells demonstrated segregation errors in anaphase, which were like those reported in cells lacking condensin I or II (11,13,57). Moreover, IMS assay showed a slight increase in partially abnormal mitotic chromosome structure in prophase. This discrepancy suggested the possibility of an abnormality in mitotic chromosome structure of BAZ1A/B KO cells that was undetectable in this study.
We were unable to identify tyrosine residues whose phosphorylation levels were decreased in chromosomes of BAZ1B or BAZ1A/B KO cells using quantitative phosphoproteomics with HAMMOC, which is likely because of difficulties in detecting histone modifications by chromatin proteomics based on trypsin digestion (58). In future studies, putative histone phospho-tyrosine residues other than pTyr142 in H2A.X should be investigated. Further research should focus on the mechanisms underlying phosphorylation-dependent regulation of mitotic chromosome segregation and BAZ1 control of chromosome condensation at mitotic entry through cooperation with other components in the mitotic chromosome scaffold.
Acknowledgments-We thank Iyo Toramoto and Fusako Kawasaki (Kochi University) for their technical assistance.