The SWI SNF complex consists

The SWI/SNF complex consists of 15 core subunits whose unique composition is critical for its distinct functions in different types of cells (Kadoch et al., 2013; Wu et al., 2009). In our study, we demonstrated that both BAF170 and BAF155 existed in the human BRG1-SWI/SNF complex and were enriched with BRG1, BAF53A, and BAF47. Depletion of BAF170 instead of BAF155 led to the loss of stem cell properties in hESCs. This pattern is different from that observed in their rodent counterparts (i.e., mESCs and mEpiSCs), suggesting an intrinsic functional divergence of BAF components between human and mouse. Although BAF155 and BAF170 share more than 60% homology in their protein sequences, their functions are not interchangeable. BAF155 is unable to correct the differentiated morphology caused by BAF170 depletion in hESCs. Similarly, knockdown of BAF53A, but not BAF53B, led to obvious differentiation of hESCs (Figure S2D). It is plausible that the BAF170- and BAF53A-containing SWI/SNF complex interacts with different partners in hESCs from BAF155 and in turn regulates a distinct set of target genes that are important for pluripotency. OCT4, SOX2, and NANOG are core pluripotency-related transcription factors that maintain undifferentiated phenotypes of ESCs (Boyer et al., 2005). Our study demonstrated that BRG1 deficiency did not lead to any immediate alteration in expression of OCT4, SOX2, or NANOG. Changes in the protein levels of these pluripotency markers became apparent only after prolonged depletion of BRG1 (Figures 1C and 1D). By contrast, significantly increased expression of genes that are important for trophectodermal and mesendodermal lineage development was observed at the early stage of BRG1 knockdown. Several of those genes (e.g., EOMES, GSC, FOXA2, NODAL, LEFTY1/2, and WNT3) were direct targets regulated by BRG1 through modulation of H3K27ac levels, thus providing insights into epigenetic regulation during human embryogenesis. In addition, our data demonstrate that BRG1 participates in a broad range of biological functions of hESCs, including cell cycle, cell-to-cell interactions, and metabolism. Therefore, it is plausible that reduced expression of these pluripotency-related factors is secondary to prolonged cellular abnormalities and differentiation. Nevertheless, we cannot exclude the possibility that BRG1 coregulates target genes of OCT4 or NANOG. We interrogated the published ChIP-seq data using histone methyltransferase inhibitor against BRG1, OCT4, or NANOG (Figure S6A). About 40% of OCT4-bound loci and 18%–38% of NANOG-occupied regions were enriched with BRG1 using ±250 bp as a cutoff distance between BRG1 and OCT4- or NANOG-enriched peak centers (Figure S5A). Although BRG1 colocalizes extensively with OCT4 and NANOG in hESCs, the overlap is far less than that in mESCs (Ho et al., 2009a), highlighting the different regulatory networks between human and mouse cells. A balance between gene activation and repression is crucial for multiple biological or pathological processes. This balance is mainly regulated through changes in chromatin structure imparted by DNA methylation, chromatin remodeling, and histone modifications. DNA methylation, histone deacetylation, and certain histone methylations (such as H3K27me3) are often associated with transcriptional repression (Perissi et al., 2010). In mESCs, BRG1 and PcG components such as EZH and SUZ12 co-occupy many of the OCT4/SOX2 target genes, and BRG1 opposes PcG activity by altering H3K27me3 levels (Ho et al., 2011). In hESCs, however, BRG1 depletion did not result in any significant changes of H3K27me3 levels in either the promoter or enhancer regions of upregulated lineage-specific genes such as EOMES and FOXA2 (Figures S5B–S5D). In contrast, we observed a significant increase of genome-wide H3K27ac levels upon BRG1 knockdown in the enhancer regions that become active in early development. This increase was detected in the majority of these enhancer regions (such as those for EOMES, FOXA2, LEFTY2, and NODAL) after transient BRG1 depletion for 3 days (Figure S6A). These data thus suggest that the elevation of H3K27ac levels is BRG1 dependent and likely a direct consequence of BRG1 depletion. We also observed an upregulation of the acetyl-H3 level in the promoter regions for about 60% of the genes we examined (Figure S6B), which was correlated with their activated transcription. However, the increase for H3K9ac was much less obvious (Figure S6C), indicating a specificity for H3K27ac in the enhancer regions regulated by BRG1. In addition, no obvious alteration in the levels of acetyl-H3, acetyl-H4, or H3K27ac modification was detected upon BRG1 depletion by western blots (Figure S6D), suggesting a regulatory specificity of BRG1 on H3K27ac in enhancer regions. Collectively, our data support the notion that BRG1 inhibits the transcription of lineage-specific genes by specifically modulating H3K27ac levels in hESCs. In this context, it is noteworthy that in several previous reports, BRG1 was shown to be associated with histone acetyltransferases (such as p300/CBP) (Huang et al., 2003) or histone deacetylases containing corepressor complexes (such as SIN3A) (Liang et al., 2008; Pal et al., 2003; Underhill et al., 2000). It will be of great interest to systematically dissect which coactivator or corepressor complex associates with BRG1 to regulate histone acetylation during early development. Nevertheless, the present study suggests a mechanism by which the BRG1-SWI/SNF complex negatively regulates gene transcription, and clearly illustrates the functional differences between the epigenetic machineries present in hESCs and mESCs that are important for pluripotency.