Murine esBAF chromatin remodeling complex subunits BAF250a and Brg1 are necessary to maintain and reprogram pluripotency-specific replication timing of select replication domains

Background Cellular differentiation and reprogramming are accompanied by changes in replication timing and 3D organization of large-scale (400 to 800 Kb) chromosomal domains (‘replication domains’), but few gene products have been identified whose disruption affects these properties. Results Here we show that deletion of esBAF chromatin-remodeling complex components BAF250a and Brg1, but not BAF53a, disrupts replication timing at specific replication domains. Also, BAF250a-deficient fibroblasts reprogrammed to a pluripotency-like state failed to reprogram replication timing in many of these same domains. About half of the replication domains affected by Brg1 loss were also affected by BAF250a loss, but a much larger set of domains was affected by BAF250a loss. esBAF binding in the affected replication domains was dependent upon BAF250a but, most affected domains did not contain genes whose transcription was affected by loss of esBAF. Conclusions Loss of specific esBAF complex subunits alters replication timing of select replication domains in pluripotent cells.


Background
Developmental changes in chromosome structure can occur at the level of large, often megabase-sized chromosome domains [1][2][3][4][5]. This cell type-specific chromosomal domain structure is thought to be important for coordinating expression of genes, thereby ensuring proper development of embryos. However, the mechanisms regulating large-scale changes in chromosome structure during development are poorly understood. In particular, very few gene products have been found to be necessary to maintain structure and function of chromosomes at this level of organization.
The temporal order of replication (replication timing) is linked to many basic cellular processes that are regulated both during the cell cycle and development. We have developed a simple and robust assay to measure replication timing genome-wide [6,7]. We found that 400 to 800 Kb-sized replication domains are spatiotemporally reorganized genome-wide during embryonic stem (ES) cell differentiation into various cell lineages [6,8]. Similar sized replication domains are also misregulated in leukemia [9]. Cell type specific reorganization of replication domains is generally coordinated with transcriptional changes and is conserved between mouse and human [10][11][12]. Replication domain reorganization is also observed during iPSC generation in which somatic cell specific replication domain structure is erased and ESC-specific replication domain structure is reestablished [8]. Considering that replication domains are regulated in the context of development and disease, it is presumed that epigenetic mechanisms play an important role in the formation of replication domain structure. However, in mammals, little or no effect on replication timing regulation has been reported for many chromatin modifier mutants, while these mutations significantly affect gene expression patterns [13][14][15]. Recently the first gene products with widespread effects on global replication timing in yeast (Fkh1/2 and Rif1) and mammals (Rif1) were identified [16][17][18][19]. Other gene products have been shown to have small effects on pericentric heterochromatin replication (Sub39h1/2 and G9a) [13,14]. Finally, replication timing of rDNA was shown to be affected by mutations in the rDNA-specific chromatin remodeling complex NoRC [20]. Together, these results suggest that specific gene products should eventually be identified that regulate cell type and domain-specific affects. Inspired by the specific and dramatic effect of NoRC on regulation of rDNA replication timing, we investigated the role of cell type specific chromatin remodeling complexes in replication timing changes during embryonic stem cell differentiation.
Brahma-associated factor (BAF) complexes are members of SWI/SNF ATP-dependent chromatin-remodeling family and regulate access of transcription factors by modulating chromatin structure. Of particular interest is that BAF subunits undergo compositional and stoichiometric change during mammalian development, which confers unique and essential roles to the complexes in cell fate determination [21][22][23][24]. For example, BAF155, BAF250a, and Brg1 are highly expressed in ESCs and their expression decreases significantly when ESCs differentiate, suggesting that these components may be essential for keeping ESCs in the undifferentiated 'ground state' [25]. In fact, Brg1 and BAF155 significantly promote reprogramming of mouse embryonic fibroblasts (MEFs) in combination with Yamanaka factors (Oct4, Sox2, Klf4, and c-Myc) [26]. BAF components are also instrumental for tissue-specific differentiation. The proper switch of neuron-specific BAF53 and BAF45 isoforms determines either the self-renewal or differentiation of neuron progenitor cells [27] and can convert fibroblasts to neurons [28]. Ectopic expression of BAF60c, a cardiac-enriched subunit, along with transcription factors GATA4 and TBX5, can convert non-cardiogenic mesoderm into beating cardiomyocytes [29]. These studies suggest that tissue-specific BAF complexes create chromatin environments favorable for transcription factor access.
In this study, we found that the embryonic stem cellspecific BAF complex (esBAF) complex deficiency leads to alterations of replication timing both in ESCs and during cellular reprogramming. Loss of DNA binding of the complex, but not transcriptional changes, correlated with changes in replication timing. These findings demonstrate the importance of chromatin remodeling complexes for maintaining replication-timing programs and, by proxy, large-scale chromatin reorganization.

Results and discussion
BAF250a is required to maintain replication timing at specific domains in embryonic stem cells We first examined the effect of acute BAF250a loss on replication timing. BAF250a is essential for early embryogenesis and has shown to be involved in the recruitment of esBAF to its target sites [30,31]. We generated a cell line in which both homologues of BAF250a undergo simultaneous conditional deletion. In these cell lines, exon 8 of the BAF250a gene is flanked by 2 loxp sites and Cre recombinase (Mer-Cre-Mer) is induced upon addition of 4-hydroxytamoxfen (OHT), resulting in frameshift mutation followed by non-sense mediated decay. BAF250a protein level was rapidly and homogeneously diminished within 24 h and was undetectable 72 h after OHT treatment [see Additional file 1A].
Genome-wide replication timing analysis ( Figure 1A, [7]) identified a set of genomic regions that changed replication timing either from early to late (EtoL) or from late to early (LtoE) in response to BAF250a loss after 72 h but not after 24 h (Figure 1B-D and [see Additional file 1B-C]). Observed changes in replication timing were highly reproducible between replicates [see Additional file 1D]. Since the changes were not as extensive as developmental changes [6,8], we calculated the P values for replication timing changes of 10,974 200-Kb segments, and applied a False Discovery Rate (FDR). Using this method, 691 and 1,667 200-Kb segments were identified as significantly changing replication timing at a 1% and 5% FDR, respectively ( Figure 1C and [see Additional file 1E]). All affected segments examined aligned in register to domains of differential replication in one or more tissues during normal development ( Figure 1E) and encompassed 400 to 800 Kb genomic segments ( Figure 1F), consistent with domains whose replication timing is normally regulated during development [8][9][10]. We conclude that BAF250a is required to maintain normal developmental control of replication timing for a fraction of the ESC genome.
Several of the EtoL regions recapitulated developmental changes that occur during ES cell differentiation, raising the possibility that the observed changes might be an indirect result of cell differentiation after BAF250a loss. Indeed, it is known that BAF complex deficiency induces cell differentiation toward the primitive endoderm lineage after several rounds of cell division [30]. However, during the considerably shorter 72-h induction period, BAF250a-disrupted ESCs retained a higher genome-wide correlation in replication timing profile with pluripotent cell types than differentiated cell types ( Figure 1G). For example, pluripotency-associated Dppa2/4 and Rex1 domains, which rapidly become late replicating during differentiation to every germ layer [8], retained ESC-specific early replication [see Additional file 2]. Moreover, we did not observe significant changes in the expression level of pluripotency-associated genes [see Additional file 3]. Together, we conclude that mutant ESCs still globally maintain an overall pluripotent cell replication timing program at least 72 h after OHT treatment, while specific domains require esBAF to maintain their replication time.
BAF250a is required to re-establish replication timing of an overlapping set of select domains during somatic cell reprogramming MEFs were treated with OHT at Days 3 to 5 after virus-mediated transduction of reprogramming factors (Oct4, Sox2, Klf4, and c-Myc; OSKM). We observed a significant decrease in the number of alkaline phosphatase (AP)-positive colonies derived from cells expressing Mer-Cre-Mer compared to control cells ( Figure 2A and B), demonstrating that BAF250a is required for efficient somatic cell reprogramming, or for growth and viability of the latter. Also, colonies from OHT-treated BAF250a flox/flox Mer-Cre-Mer cells have an irregular unsmoothed shape ( Figure 2B), which is reminiscent of the morphology of BAF250adeficient ESCs [30]. Several iPSC-like colonies from BAF250a flox/flox and BAF250a flox/flox Mer-Cre-Mer

BAF250a f/f;Mer-Cre-Mer
Next we performed replication-timing analysis of BAF250a flox/flox OSKM and BAF250a −/− OSKM cells. Despite the fact that loss of BAF250a significantly reduced the efficiency of AP-positive colony production, the genome-wide replication timing profiles of three independent AP-positive BAF250a −/− OSKM clones were almost identical to that of control BAF250a flox/flox OSKM or other ESC lines and were clearly more similar to pluripotent cells than to partially reprogrammed iPSCs (piPSCs; Figure 2C and D). This result suggests that BAF250a −/− OSKM have passed the common epigenetic block experienced by piPSCs [8]. Nonetheless, BAF250a −/− OSKM cells display distinct replication timing differences from ESCs or control OSKM cells ( Figure 2E-H). When replication timing differences in OSKM cells are compared to those in ESCs, we observed a conservation of BAF250a-affected domains between ESCs and OSKM cells ( Figure 2E). Indeed, we identified a set of chromosomal domains that undergo replication timing switching in BAF250a-deficient OSKM cells ( Figure 2F) and found that significant fraction of these switching domains overlap with those identified in BAF250a-deficient ESCs ( Figure 2G and H). These results confirm a role for BAF250a in replication timing regulation of specific chromosomal domains in the pluripotent state.

Loss of Brg1, but not BAF53a, affects an overlapping set of replication domains
Since BAF250a is a subunit of the esBAF complex in ESCs [32], we next examined the role of two other esBAF subunits, Brg1 (catalytic ATPase subunit) and BAF53a (another ESC-specific subunit), in the regulation of replication domain structures. To this end, we performed genome-wide replication timing analysis in Brg1 and BAF53a conditional knockout ESC lines [33,34]. Similar to BAF250a knockouts, the overall replication timing profiles of OHT-treated Brg1 flox/flox and BAF53a flox/-ESCs are almost identical to that of parental (mock-treated) ESCs, suggesting an overall maintenance of stem-cell identity in these cells ( Figure 3A and B). This was further confirmed by the finding that expression of Oct4 did not change in these mutant cells [see Additional file 4]. However, we found that loss of Brg1 induced altered replication timing profiles in a subset of chromosomal domains with a bias for EtoL switching (Figure 3C and D) and that these Brg1-sensitive domains significantly overlap with BAF250a-sensitive domains ( Figure 3C and [see Additional file 5]). On the other hand, loss of BAF53a did not induce any significant changes in replication timing ( Figure 3C and D). This latter result, coupled with our previously published results with other Cre-inducible deletions [15], confirms that the changes in replication timing are not due to the activation of Cre recombinase by OHT treatment.
The fraction of chromosome domains that displayed EtoL switching in response to BAF250a loss (commonly misregulated in BAF250a-deficient ESCs and OSKM cells), showed a very similar tendency of replication timing switching in Brg1 but not BAF53a mutant ESCs ( Figure 3E-F). For example, at chromosome 4 (104.5-105.0 Mb) and chromosome 7 (82.5-83.0 Mb) domains where the BAF250a is required for early replication in both ESCs and OSKM cells, these domains are late replicating after Brg1 loss, while they remain early replicating in the absence of BAF53a ( Figure 3F). Together, these results demonstrate a BAF53a-independent function of the esBAF complex is required for proper regulation of replication timing at specific replication domains. However, the partial overlap in affected regions between Brg1 and BAF250a suggests the potential for independent roles of each subunit or gain of function effects of each subunit in the absence of the other.

BAF250a-dependent binding of esBAF complexes to affected domains independent of transcriptional regulation
Since disruption of BAF complexes have a profound effect on genome-wide transcriptional regulation in ESCs [30,33,35], we wished to determine whether the effect of BAF250a loss on replication timing was linked to altered gene expression in these domains. Brg1 ChIP-seq data in mESCs [35] revealed significant enrichment of Brg1 proteins in early replicating regions of the ESC genome ( Figure 4A). We observed significant enrichment of Brg1 in Brg1-sensitive EtoL domains compared to LtoE domains, though this enrichment level is comparable to that seen in unaffected early replicating (EtoE) domains ( Figure 4B). Unfortunately, several anti-BAF250a antibodies were not of sufficient quality for chromatin immunoprecipitation (ChIP). However, Brg1 ChIP at two of the BAF250asensitive early replicating domains, 500-Kb genomic segments on chromosome 4 (104.5-105.0 Mb) and chromosome 7 (82.5-83.0 Mb), revealed BAF250adependent Brg1 enrichment at multiple sites within these domains ( Figure 4C). It should be noted that loss of BAF250a also impaired Brg1 enrichment at the Oct4 gene region that retains its time of replication (EtoE) regardless of BAF250a availability ( Figure 4C and [see Additional file 6]). Thus, esBAF is generally enriched in early replicating domains, although loss of esBAF is not sufficient to elicit a change in replication timing at all domains.
Finally, we examined the effect of esBAF disruption on transcription of genes within the domains affected for replication timing [35]. None of the esBAF target genes in switching regions (0/19) was more than two-fold up-or downregulated upon Brg1 knockdown (Brg1 KD). Genes in switching regions, regardless of whether they were targets of esBAF binding, have no coordination between EtoL/LtoE changes and  . Replication timing data were averaged into 200-Kb windows and dRTs (that is, Brg1 OHT ratio -Brg1 mock ratio) were calculated for comparisons between different groups. A substantially lower conservation of replication timing affected regions is seen between BAF53a ESCs and BAF250a or Brg1 embryonic stem cells (ESCs) than between BAF250a and Brg1 ESCs. (D) Summary of significant early to late (EtoL) and late to early (LtoE) switching segments in OHT-treated cells as determined in Figure 1C. (E) Box plots show the replication timing of segments that are sensitive to BAF250a loss (false discovery rate (FDR) = 1% domains identified in Figure 2G) in various esBAF subunit mutants. Replication timing in mutants (red) is compared to that in control counterparts (green). P values were calculated using the Wilcoxon rank sum test. transcriptional downregulation/upregulation, respectively (overall R = 0.02) ( Figure 5A and B). This is consistent with the hypothesis that replication timing is associated more with transcriptional competence than transcription per se [36] and suggests that the role of esBAF in regulating replication timing is not a direct transcriptional role for this complex. Taken together, these results demonstrate that BAF250a-dependent Brg1-containing esBAF complexes are recruited to regions that require BAF250a and Brg1 for early replication in ESCs, but most of these regions do not contain esBAF-regulated genes.

Conclusions
In summary, our data presented here reveal an unanticipated effect of esBAF complex disruption on replication timing and, by proxy, higher-order chromatin folding [10,37,38]. Yeast transcription factors Fkh1 and Fkh2 are thought to modulate replication timing by bringing early replication origins in close proximity in the nuclear space independent of their transcriptional activity [16]. It is possible that the BAF complexes play a similar role in mammalian cells, thereby promoting the formation of an early replication domain. Indeed it has been shown that Brg1 is involved in cell type-specific chromatin loop formation at the beta-globin locus [39]. Interestingly, esBAF complexes are known to interact with the nuclear matrix protein Rif1 which has recently been identified as global replication timing regulators [18,19,32]. Currently, it is unclear why only a small subset of esBAF-enriched replication domains is sensitive to esBAF complex deficiency. For example, other early replicating domains harboring genes such as Oct4 have multiple Brg1 binding sites but maintain their early replication in the absence of BAF250a or Brg1 [see Additional file 6]. This suggests that there are additional mechanisms maintaining early replication of these domains, whereas we have identified a subset of domains at which esBAF presence has a major effect on replication timing. This may be related to whether or not the affected domains are capable of switching replication timing, as none of the affected domains were constitutively early replicating ( Figure 1E). Future studies are warranted to uncover the mechanism by which BAF complexes influence replication timing during stem cell self-renewal and differentiation.

Methods
Embryonic stem cell culture BAF250a flox/flox; Mer-Cre-Mer ESC lines were established from day 3.5 blastocysts obtained by crossing BAF250a flox/+; Mer-Cre-Mer with BAF250a flox/flox and maintained on feeder MEFs in the presence of leukemia inhibitory factor (LIF) as described previously [30]. Mer-Cre-Mer mice were purchased from the Jackson Laboratory; Bar Harbor, ME USA (stock number: 008463). Brg1 flox/flox; Actin-CreER and BAF53a flox/-; Actin-CreER ESC lines were maintained as described previously [33,34]. To generate mutant ESCs, these ESC lines were treated with 1 μM 4-hydroxytamoxifen (OHT) for 24 h and harvested 48 h later, unless otherwise indicated. As a control, cells were treated with ethanol. OSKM) [40]. Early passage fibroblasts (less than passage 5) were cultured in 6-well dishes and about 4 × 10 4 cells in each well were infected overnight with viral supernatants freshly prepared by transfection of the retroviral packaging Plat-E cell line (Lipofectaine 2000, Invitrogen, Life Technologies, Carlsbad, CA, USA) containing the cDNAs of the mouse reprogramming factors. Three days after infection, cells were passaged into new wells and tamoxifen was added for three days (Days 3 to 5) or other time windows to ablate BAF250a. Control iPSClike colonies (BAF250a +/+, OHT treatment or BAF250a flox/flox; Mer-Cre-Mer, no OHT treatment) were typically picked 21 days after infection and iPSC-like colonies from BAF250a flox/flox; Mer-Cre-Mer, OHT treated fibroblast culture were typically picked 30 days post infection. Genotyping of BAF250a was performed by PCR. We used the primer sequences 5′-GTAATGGGAAAGCGACTAC TGGAG-3′ and 5′-TGTTCATTTTTGTGGCGGGAG-3′, which amplify a 632-bp fragment from the WT locus, an 812-bp fragment from the floxed locus and a 298-bp fragment from the knockout locus, respectively. PCR reactions were carried out with 40 cycles (30 sec at 94°C, 30 sec at 59°C, 1 min at 72°C). For alkaline phosphatase (AP) staining, culture wells containing iPSC-like colonies were washed with PBS and cells were fixed with 4% paraformaldehyde in PBS for 2 min at 20°C. Fixed cells were then rinsed twice with 0.5 ml of TBST (TBS plus 0.05% Tween-20) and incubated with fresh AP staining solution (4.5 μl 50 mg/ml nitro blue tetrazolium, 3.5 μl 50 mg/ml 5bromo-4-chloro-3-indolyl phosphate in 100 mM Tris-HCl, pH 9.5, 100 mM NaCl, 50 mM MgCl 2 ) in the dark room at 25°C for about 15 min. Stained cells were rinsed with PBS and kept at 4°C.

Chromatin Immunoprecipitation
ChIP was performed as previously described [41]. Two million cells were harvested and fixed in 1% formaldehyde for 10 min at 25°C, then stop fixation in 0.125 M glycine. Fixed cells were sonicated to produce chromatin fragments 300 to 700 bp in length. Chromatin fragments were then immunoprecipitated with anti-Brg1 antibody [42]. The precipitated DNAs were then purified by ethanol precipitation after phenol-chloroform extraction. Quantitative PCR reactions were performed to detect the occupancy of Brg1 at multiple sites within the chromosome 4 and 7 EtoL domains. Quantitative PCR reactions included the following: 4 μl of ChIP product (200 μl per ChIP assay), 10 μl of 2X SYBR green PCR master mix (Applied Biosystem, Carlsbad, CA, USA, 4309155) and 25 nM of each primer. QPCR reactions were tripled and performed in ABI StepOnePlus system through 50 cycles (15 sec at 95°C, 45 sec at 60°C). Ct values were generated by ABI software. Standard errors in Figure 4C

RNA FISH
RNA FISH was performed as described previously [43]. To generate RNA FISH probes, Rex1 genomic DNA fragments were amplified, cloned into pBluscript, and labeled by nick translation. Cells were treated with 0.5% Triton X-100 in CSK buffer (100 mM NaCl, 300 mM sucrose, 10 mM Pipes, pH 6.8, 3 mM MgCl 2 , 1 mM EGTA) for 30 sec at 4°C, fixed with 4% paraformaldehyde, and then immersed in 70% ethanol for 5 min at −20°C, dehydrated through a 90% and 100% ethanol series, and the denatured FISH probe mixture was hybridized to slides at 37°C for 16 h in a moist chamber. Slides were washed three times with 50% formamide in 2X SSC at 43°C and three times with 0.8X SSC at 60°C. Slides were then incubated for 30 min in a blocking solution (3% BSA, 0.1% Tween 20 in 2X SSC) at 37°C and incubated in a detection solution (in 1% BSA, 0.1% Tween 20 in 2X SSC) containing anti-digoxigeninconjugated rhodamine (Roche, Nutley, New Jersey, USA, 11207750910) for 30 min at 37°C. Then slides were washed three times with 4X SSC, 0.1% Tween 20 for 5 min at 43°C. Before imaging, the slides were counterstained with DAPI (200 ng/ml), washed with 4X SSC, and then mounted in 90% glycerol containing antifade reagent.

Replication timing profiling by microarray
Replication timing analysis was performed as described previously [6,7]. In brief, cells were labeled with 50 μM BrdU for 2 h, washed twice with ice-cold PBS, trypsinized, and then were fixed in 75% ethanol. These cells were resuspended in PBS containing 1% FBS, stained with propidium iodide (50 μg/ml) for 30 min in the presence of RNaseA (0.5 mg/ml), and then were sorted into early and late S phase fractions by flow cytometry. After phenol-chloroform extraction of DNA, immunoprecipitation with anti-BrdU mouse monoclonal antibody (BD Biosciences, San Jose, CA, USA, 555627) was performed in each fraction to enrich BrdU-substituted replicating DNA. Isolated early and late replicating DNA were amplified by whole-genome amplification (WGA) (Sigma-Aldrich, St Louis, MO, USA, GenomePlex), labeled with Cy3 and Cy5, and hybridized to a mouse wholegenome microarray (NimbleGen Symtems, Madison WIS, USA, 2006-07-26_MM8_WG_CGH or 100718_MM9_ WG_CGH_HX3). Sample labeling, hybridization and data extraction were performed according to standard NimbleGen Systems procedures. Data analyses were performed using R/Bioconductor (http://www.rproject.org; http://www.bioconductor.org). Obtained raw datasets were normalized using the limma package in R/Bioconductor and loess-smoothed over a 300-Kb window size. These smoothed datasets were used to generate replication-timing plots in figures. For some analyses, datasets were averaged into 200-Kb windows (fixed position) and replication timing differential (that is, OHT ratio -mock ratio) was determined for each 200-Kb segment. In order to determine the significant replication timing switching domains that are independent of changes between replicates, we determined Euclidian distance at 10,974 200-Kb segments between groups (that is, mock versus OHT) and within groups (that is, mock replicate-1 versus mock replicate-2), which was used to calculate P values at each 200-Kb genomic segment. Statistical significance was then calculated using the qvalue package in R/Bioconductor, which yields a q-value for each segment that reflects the proportion of false-positives (False Discovery Rate; FDR) among segments deemed to have significant replication timing (RT) changes. High confidence replication timing switching domains were selected with a q-value cutoff of 0.01, corresponding to an overall FDR of 1%. A q-value cutoff of 0.05 was also used to identify a set of lower confidence domains. To examine alignment of timing switching domains to developmental domains, replication timing data from 9 cell types (ESC/iPSC, EBM3/ EPL, EBM6/EpiSC, NPC, Mesoderm, Endoderm, partial iPSC, MEF, and Myoblast) were assembled from the Repli-cationDomain.org database [44] and plotted together with the data from BAF250a mock and OHT. Timing switching domains from chromosome 1 (largest-sized) and chromosome 10 (middle-sized) were selected and their alignment to developmental domains was judged by visual inspection in Figure 1E. Indeed, when we examined statistical significance of replication timing changes of BAF250a OHT compared to other cell lines, most domains examined in Figure 1E were not significantly different from at least one of nine cell types, even with a q-value cutoff of 0.2 (42/52 EtoL domains and 42/44 LtoE domains). The size of switching domains was determined using a segmentation algorithm in the DNAcopy package in R/Bioconductor as described previously [6]. Unsmoothed datasets consisting of replication timing (BAF250a OHT ratio -mock ratio) for all probes were processed for switching domain segmentation and the resultant EtoL and LtoE segment sizes were shown in Figure 1F. Replication timing datasets are downloadable from ReplicationDomain (http://www. replicationdomain.org).

Imaging system and measurement
Images were collected using a Nikon Ti-U Eclipse fluorescence microscope equipped with a 60x, 1.40 NA lens and a cooled charge-coupled device camera (C4742-95-12ER, Hamamatsu Photonics, Hamamatsu, Japan), controlled by a windows computer running the software program MetaMorph (Molecular Devices, Sunnyvale CA, USA).