The BAF complex is required for the basal and induced levels of expression of the majority of the IFN-α-inducible genes.
In order to determine the requirement for the BAF complexes in the IFN signaling pathways, we specifically knocked down BAF47/INI1/hSNF5, a critical subunit of the BRG1- and hBRM-containing complexes, by RNA interference. Transfection of HeLa cells with an episomal BAF47 RNA interference construct for 2 days resulted in complete inhibition of the BAF47 protein (Fig. ). Furthermore, both BRG1 and BRM were also significantly down-regulated (Fig. ), possibly by the decreased stability in the absence of BAF47. These experiments suggest that the function of the BRG1- and BRM-containing complexes could be severely compromised through the expression of the siRNA targeting BAF47 in the cell. Therefore, mRNA samples were isolated from HeLa cells transfected with the RNA interference construct and were analyzed by DNA microarrays. The analysis revealed that approximately 2.5% of the genes were repressed and that 0.5% were activated more than threefold among the 44,000 cDNA and expressed sequence tag sequences (data not shown). In sharp contrast to the low percentage of affected genes in the whole genome, the expression of more than 90% of the IFN-α-inducible genes was significantly down-regulated in cells expressing the BAF47 siRNA (Fig. , compare lanes 2 and 4, and data not shown). Moreover, upon IFN-α stimulation, the induction of these inducible genes was either dramatically reduced or completely inhibited (Fig. , compare lanes 1 and 3, and data not shown). The strongly inhibited genes included those encoding the well-characterized antiviral proteins such as the double-stranded RNA-dependent protein kinase (PKR), the 2-5A system (OAS1, OAS2, and OAS3), IFITM1, and the Mx proteins (36
). The inhibition by the BAF47 siRNA was not caused by nonspecific activity of antisense RNA, since another siRNA construct, which targeted a different region of the BAF47 sequence but was not able to knock down the BAF47 mRNA, failed to inhibit the cellular response to IFN-α stimulation (data not shown). These data strongly suggest the BAF complex is required for the normal function of the IFN signaling pathways.
Knockdown of the BAF complex significantly inhibited the cellular response to poly(I)/poly(C).
To mimic the cellular response to viral infection, control and BAF-inhibited (siBAF47) HeLa cells were treated with double-stranded RNA poly(I)/poly(C) and examined for the expression of IFN-inducible genes. These experiments were of interest, since the above-described microarray analysis found toll-like receptor 3 (TLR3) to be constitutively reduced in BAF-inhibited cells. TLR3 specifically recognizes double-stranded RNA and is required for the poly(I)/poly(C) induction of IFN target genes (4
). As shown in Fig. , poly(I)/poly(C) induced the expression of many antiviral genes, such as the Mx1, PKR, OAS3, and IFITM1 genes, in control cells. However, the induction of these genes was dramatically reduced or abolished in cells in which the BAF complex was inhibited. These results indicate that the establishment of antiviral activities is severely impaired when the BAF complex is inhibited.
The BAF complex is required for the elicitation of cellular antiviral activities upon viral infection.
To study whether the inhibition of the BAF complex affects cells' ability to control viral infection, control and BAF-inhibited HeLa cells were infected with NDV. Infected cells were tested first for the induction of the IFN-β gene, an immediate early IFN gene important for establishing an antiviral state in the cells (36
). Real-time RT-PCR analysis (Fig. ) showed that IFN-β transcript levels were markedly elevated after viral infection in control cells but that transcript levels were much lower in siBAF47 cells at all three time points tested. The inhibition of IFN-β transcription was less severe at the 24-h point than the 6- and 12-h points. Therefore, we examined whether the extent of the BAF47 knockdown diminishes at the 24-h point. As shown in Fig. , no difference in BAF47 knockdown levels was observed during the course of stimulation. These experiments suggest that the BAF complex is required for the rapid and full induction of IFN-β. Consistent with the data for poly(I)/poly(C), the induction of all antiviral proteins shown in Fig. was markedly attenuated or completely blocked in siBAF47 HeLa cells after NDV infection, while these genes were robustly induced in control cells (data not shown). We next assessed whether inhibition of the BAF complex has an effect on viral replication. To this end, the levels of the NDV NP transcripts were measured by real-time RT-PCR at various time points following infection (Fig. ). In siBAF47 cells, levels of viral transcripts increased more than 70-fold by 24 h, while the transcripts in control cells increased less than 15-fold over the same period. These results indicate that inhibition of the BAF complex reduces cells' ability to control viral growth, most likely by inhibiting the expression of IFN-β and other antiviral proteins. To examine whether the inhibition of the BAF complex compromises IFN′s antiviral activities, we measured NDV NP transcript levels in cells that had been treated with IFN-α for 18 h prior to NDV infection for 12 h. As seen in Fig. , IFN treatment potently inhibited NDV nucleocapsid transcript expression in control HeLa cells at three doses of NDV, while the treatment did not have discernible inhibitory effects on siBAF47 cells at any viral dose tested. These results indicate that IFN fails to confer antiviral activities on cells in which the BAF complex is inhibited. Similar experiments were performed with vesicular stomatitis virus. Consistent with the data presented above for NDV infection, IFN fully protected control cells from the cytopathic effect of vesicular stomatitis virus. However, it had no protective effects on siBAF47 cells, leading to uncontrolled destruction of the cells (data not shown), confirming that the BAF complex has a critical role in establishing IFN′s antiviral activities.
The BAF complex controls the chromatin accessibility at the IFITM1 promoter.
To elucidate the mechanisms by which the BAF complex mediates the action of IFN-α, we asked whether the BAF complex directly regulates the promoter activity of one IFN-α target gene, the IFITM1 gene. We cloned 200 bp of the IFITM1 promoter region into the chromatin-forming episomal pREP4-luc vector. HeLa cells were cotransfected with the promoter reporter construct and the RNA interference construct targeting BAF47. As shown in Fig. , when cotransfected with a control vector into HeLa cells, the IFITM1 promoter was robustly stimulated by IFN-α. However, siRNA targeting BAF47 strongly inhibited the basal-level activity of the promoter and completely inhibited induction by IFN-α (Fig. ), indicating that the BAF complex is required for the activity of the IFITM1 promoter. The expression of siBAF47 had no detectable effect on the activity of the claudin promoter (Fig. ), indicating that the inhibition of the IFITM1 promoter was specific.
To clarify how the BAF complex controls the activity of the IFITM1 promoter, we examined the chromatin structure of the endogenous IFITM1 promoter. As shown in Fig. , microccocal nuclease digestion identified a nucleosome between positions −5 and −145 relative to the transcription start site. The ISRE (positions −21 to −35) is located within the nucleosome. In order to determine whether the BAF complex is required to remodel the nucleosomal structure, we probed the chromatin structure by restriction enzyme accessibility assay. Nuclei isolated from HeLa cells were briefly digested with AvaII, which recognizes a sequence beginning at position −68 upstream of the transcription start site (Fig. ). Following complete digestion of the purified DNA with HgiAI, the cleavage sites were detected by linker ligation-mediated PCR (28
). As shown in Fig. , stimulation of the control HeLa cells with IFN-α for 2 h strongly elevated the accessibility of the promoter to AvaII (compare lanes 3 and 4) about 5.2-fold, indicating that the chromatin was remodeled to a more open structure in response to IFN-α. The BAF47 siRNA abolished the increase in chromatin accessibility (compare lanes 1 and 2), indicating that the BAF complex is required for the chromatin remodeling induced by IFN-α signaling. To determine whether the knockdown of BAF47 changed the levels of histone acetylation at the IFITM1 promoter, we performed ChIP experiments with antibodies against tetra-acetylated histone H4 tail. As shown in Fig. , the IFITM1 promoter band intensity from the cells transfected with the siBAF47 construct was significantly weaker than that from the cells transfected with the control vector (compare lanes 5 and 10, upper panel), while the intensities for the CSF1 gene upstream control sequence were similar (compare lanes 5 and 10, lower panel). Quantification with the PhosphorImager indicates that the IFITM1 promoter signal was threefold higher in the control cells than in the siBAF47 cells. The acetylation levels in the regions 4 kb upstream and 4 kb downstream of the promoter region were not significantly changed by siBAF47 (data not shown), suggesting that the effect of the BAF complex on the acetylation of histone H4 is not global but is localized in the promoter region. These data indicate that the inhibition of BAF47 resulted in lower levels of histone H4 acetylation and a less open chromatin structure at the promoter.
An active BAF complex is required for rapid and full induction of the IFITM1 promoter by IFN-α.
To confirm that the BAF complex prepares the chromatin structure of the IFITM1 promoter for rapid induction by IFN-α, we examined the expression of the IFITM1 gene in HeLa cells as well as in SW-13 cells, which do not have an active BAF complex due to the absence of BRG1 and hBRM subunits (10
). Transient expression of BRG1 reconstitutes the active BAF complex in SW-13 cells (45
). IFITM1 was induced to high levels in HeLa cells by treatment with IFN-α, with maximal induction occurring between 4 and 8 h (Fig. ). In contrast, the levels of induction of IFITM1 mRNA by IFN-α were much lower in SW-13 cells, and the induction did not reach its plateau until approximately 12 h. However, the kinetics and levels of induction of another IFN target gene, ISG15, in HeLa cells and SW-13 cells were similar (Fig. ), suggesting that the deficient induction of IFITM1 in SW-13 cells was not caused by a defect in IFN signaling.
FIG. 4. Expression of BRG1 results in more rapid kinetics and higher levels of the IFITM1 gene induction in SW-13 cells in response to IFN-α. (A) Analysis of IFITM1 mRNA expression induced by treatment with IFN-α. Total RNAs extracted from HeLa (more ...)
To determine whether the slower kinetics and lower levels of induction of the IFITM1 gene in SW-13 cells was caused by the absence of the BRG1 protein, we transiently expressed BRG1 in SW-13 cells before the treatment with IFN-α. As shown in Fig. , the expression of BRG1 alone induced low levels of IFITM1 expression and IFN-α treatment of BRG1-transfected cells resulted in synergistic activation of the gene. Induction with IFN-α alone reached a plateau at 12 h, but IFN-α treatment in the presence of BRG1 induced the expression of the gene at much higher levels, with an earlier plateau (at approximately 4 h) (Fig. ). The expression of BRG1 in SW-13 cells did not alter the induction of the ISG15 gene by IFN-α (data not shown). The expression of the ATPase-dead form of BRG1 did not have a significant effect on the induction of IFITM1 (data not shown), suggesting that the chromatin-remodeling activity of the BAF complex is required for rapid and high-level induction of this gene by IFN-α.
BRG1 increases the accessibility of the IFITM1 promoter.
In order to confirm the mechanism by which BRG1 facilitates the induction of the IFITM1 gene in response to IFN-α by remodeling its chromatin structure, we examined restriction enzyme accessibility at the IFITM1 promoter in the absence and presence of BRG1. SW-13 cells were transiently transfected with a BRG1 expression vector for 24 h, followed by stimulation with IFN-α. Nuclei were isolated and subjected to brief digestion with AvaII or HgiAI. Following complete digestion of the purified DNA with BclI, the cleavage sites were detected by linker ligation-mediated PCR as described above. Stimulation of SW-13 cells with IFN-α for 2 h slightly increased the accessibility of the promoter to AvaII (Fig. , compare lanes 1 and 3) and HgiAI (Fig. , lanes 1 to 3). Interestingly, IFN-α signaling in the presence of BRG1 dramatically increased the accessibility of the promoter to the restriction enzymes (Fig. , compare lanes 1 and 4, and C, lanes 4 and 5). BRG1 alone without IFN-α stimulation also significantly increased accessibility (Fig. , compare lanes 1 and 2, and C, compare lanes 1 and 4), suggesting that the recruitment of BRG1 to the IFITM1 promoter does not require IFN-α stimulation and that the BAF complex constitutively remodels the chromatin structure at the IFITM1 promoter to a more “open” conformation that is more accessible to transcription activators and RNA polymerase machinery.
Constitutive association of BRG1 with the IFITM1 promoter allows rapid recruitment of ISGF3 complex and RNA polymerase II.
Using ChIP assays, we determined whether the BAF complex binds directly to the IFITM1 promoter (Fig. ). Compared to the 3′ untranslated region, the IFITM1 promoter sequence was reproducibly enriched about twofold by the BRG1 antibody from SW-13 cells transiently expressing BRG1 (Fig. , panel a). Interestingly, BRG1 binding was observed even in the absence of IFN-α stimulation (Fig. , compare lanes 1 to 3 and lanes 4 to 6 in panel a), consistent with the persistent open chromatin structure at the IFITM1 promoter in the presence of the active BAF complex. Furthermore, transient expression of BRG1 in SW-13 cells without IFN-α stimulation up-regulated the basal-level expression of the IFITM1 gene, as demonstrated by RT-PCR analysis (Fig. ) and by the activity of an IFITM1 luciferase reporter construct (Fig. ). The constitutive association of BRG1 with the IFITM1 promoter was not an artifact resulting from the overexpression of BRG1, since the endogenous BRG1 in HeLa cells was also associated with the IFITM1 promoter independent of IFN-α stimulation (Fig. ). In contrast, STAT2 association required stimulation with IFN-α (Fig. ). No BRG1 or STAT2 binding to the 3′ untranslated region of the IFITM2 gene was detected. Thus, the association of BRG1 with the IFITM1 promoter appears to be constitutive and does not require stimulation with IFN-α.
FIG. 6. BRG1 is constitutively associated with the IFITM1 promoter. (A) SW-13 cells transfected with pBJ5 or pBJ5-BRG1 for 24 h were treated with 500 U of IFN-α/ml for indicated periods of time. Chromatin lysates were prepared and immunoprecipitated as (more ...)
Next, we evaluated the levels of histone H4 acetylation at the IFITM1 promoter in the presence and absence of BRG1 and IFN-α. We found that the presence of BRG1 increased the acetylation of histone H4 about threefold at the IFITM1 promoter region compared to the level at the 3′ untranslated region (Fig. , compare lanes 1 and 4 in panel b), consistent with a more open chromatin structure at the promoter region. IFN-α stimulation for 2 h further increased the H4 acetylation at the promoter about 10-fold (Fig. , panel b), resulting in a more open chromatin structure, as demonstrated by the restriction enzyme accessibility data (Fig. ).
IFN-α signaling activates the ISGF3 complex consisting of STAT1, STAT2, and p48, which relocates to the nucleus and binds to its target promoters (7
). As shown in Fig. , panel c, only low levels of STAT2 were associated with the IFITM1 promoter without BRG1, while in the presence of BRG1, IFN-α induced a strong association of STAT2 with the promoter (compare lanes 1 to 3 and lanes 4 to 6). A similar observation was made for p48 (data not shown). As expected, RNA polymerase II binding to the IFITM1 promoter was also enhanced by the presence of BRG1 and stimulation with IFN-α (Fig. , panel d, compare lanes 1 to 3 and lanes 4 to 6). However, significant binding to the 3′ untranslated region (data not shown), but not to the 5′ far-upstream region of CSF1 promoter, was also detected in the presence of BRG1 and IFN-α. The data indicate that significant binding of RNA polymerase II does not occur at the IFITM1 promoter until after IFN-α stimulation. This finding shows that although the chromatin structure of the promoter is already remodeled by the prebound BAF complex, the transcription machinery is not present on the promoter in a poised state. Instead, its assembly is dependent on the cellular signaling cascade. These results show that constitutive binding of the BAF complex facilitated histone H4 acetylation, ISGF3 binding, and the transcription machinery assembly on the IFITM1 promoter upon stimulation with IFN-α.