A Novel Mammalian Complex Containing Sin3B Mitigates Histone Acetylation and RNA Polymerase II Progression within Transcribed Loci^▿

Cell culture, plasmids, and transfections.

Human embryonic kidney HEK293T cells and human cervical carcinoma HeLa cells were cultured in Dulbecco's modification of Eagle's medium (DMEM; catalog number 10-013-CV; Cellgro) supplemented with 10% newborn calf serum. Cells were maintained at 37°C in an atmosphere containing 5% CO₂ and 100% humidity. Sin3B and Pf1 wild type (WT) and mutants used in coimmunoprecipitation (co-IP) assays were cloned in a pBabe Puro vector. For complex reconstitution assays, Sin3B and Pf1 WT and mutants were cloned in a pcDNA3.1(+) vector. Detailed cloning and mutational strategies are available upon request. Transient-transfection assays with Ca-phosphate were performed using a standard procedure. The amount of plasmid used in the transfection mixtures was 2 μg unless otherwise stated. Stable transfections with shRNAs in HeLa cells were performed according to the manufacturer's recommendations (OpenBiosystems).

Protein-protein interaction assays, micrococcal nucleosome extraction, and antibodies.

Endogenous coimmunoprecipitation was performed as follows. The nuclear extracts were prepared from HEK293T cells. Per co-IP mixture, one 10-cm dish was used. The cells were washed twice in ice-cold phosphate-buffered saline (PBS). One milliliter of reticulocyte standard buffer (10 mM Tris [pH 7.5], 10 mM KCl, 3 mM MgCl₂; with fresh protease cocktail inhibitor [PIC] added) (RSB) was added, and cells were scraped into an Eppendorf tube and spun at 2,000 × g, and 100 μl RSB-G40 (10 mM Tris [pH 7.5], 10 mM KCl, 3 mM MgCl₂, 10% glycerol, 0.5% NP-40; with fresh PICs added) was added. The samples were spun at 2,000 × g to pellet the nuclei. The nuclei were washed with 100 μl RSB-G (10 mM Tris [pH 7.5], 10 mM KCl, 3 mM MgCl₂, 10% glycerol), and nuclei were lyzed with 50 μl NE buffer (20 mM HEPES [pH 7.9], 420 mM NaCl, 1.5 MgCl₂, 0.2 mM EDTA, 25% glycerol; with fresh 0.5 mM dithiothreitol [DTT] and PICs added). After 30 min of incubation on ice the samples were spun at maximum speed and the supernatant was placed in a separate tube. One milliliter of PBS was supplemented with 0.5% NP-40 and PICs, and 2 μg of immunoprecipitating antibody was added and incubated at 4°C overnight. The next day, 10 μl prewashed protein A/G Plus-agarose (sc-2003) was added, and the reaction was incubated for 4 h. After three washes in PBS supplemented with 0.5% NP-40, the samples were boiled in SDS-protein loading buffer. The proteins were then separated on SDS-PAGE and detected with Western blotting with the corresponding antibodies.

Nucleosome extraction was done using a standard procedure. The cells were washed twice in PBS followed by adding 300 μl buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 1.5 mM MgCl₂, 0.34 M sucros, 10% glycerol, 0.1% Triton X-100, with freshly added 1 mM DTT and PICs). After 8 min of incubation on ice, the samples were spun for 5 min at 2,000 × g, and pellets were washed once with buffer A without Triton X-100. Pellets were then resuspended in 100 μl buffer B (3 mM EDTA, 0.2 mM EGTA, with freshly added 1 mM DTT and PICs) and incubated on ice for 20 min. The samples were spun for 5 min at 2,500 × g, and 50 μl buffer C was added (10 mM Tris, 10 mM KCl, 1 mM CaCl₂). This was followed by addition of 1 U micrococcal nuclease for 2 min at 37°C. The reaction was put on ice, and 1 mM EGTA was added to neutralize the enzyme. The sample was spun at 3,000 × g, and the supernatant with nucleosomes was placed in a separate tube. Five microliters was used for checking on a 2% DNA agarose gel.

For nucleosome coimmunoprecipitation assay purposes, prior to nucleosome extraction as described above, HeLa cells were cross-linked with 1% formaldehyde for 10 min at room temperature. Fifty microliters of nucleosomes was then diluted in 1 ml IP buffer (100 mM Tris [pH 7.5], 150 mM NaCl, 0.1% NP-40, 5% glycerol, with freshly added 1 mM DTT and PICs), and 2 μg of corresponding antibody was added. The reaction mixture was then incubated over night at 4°C. The following day 10 μl prewashed protein A/G plus-agarose (sc-2003) was added, and the reaction mixture was further incubated for 2 h at 4°C. This was followed by five washes in immunoprecipitation (IP) buffer and boiling in SDS-protein loading buffer. The proteins were then separated on SDS-PAGE and detected with Western blotting with the corresponding antibodies. Antibodies used for IPs and Western blotting were the following: Sin3B (Santa Cruz Biotechnology AK-12), Pf1 (Bethyl Labs BL6317), Sin3A (Santa Cruz Biotechnology AK-12), FLAG (m2 monoclonal Sigma F3165).

Complex reconstitution assay and interaction with nucleosomes.

A complex reconstitution assay was preformed as follows. Each component of the complex was coexpressed in HEK293T cells by transient Ca-phosphate transfection. Only immunoprecipitating protein was FLAG tagged. Two days after transfection, proteins were extracted with 1 ml KCl lysis buffer (20 mM HEPES [pH 7.9], 150 mM KCl, 5% glycerol, 1 mM DTT, 100 μM Zn-acetate, 2 mM MgCl₂, 2 mM EDTA, 0.2% NP-40, with freshly added PICs) per 10-cm dish. One-half hour after incubation in lysis buffer, the samples were spun for 5 min at maximum speed. The supernatant was then placed in a separate tube, and the extracts were incubated overnight with prewashed FLAG-Sepharose beads (Sigma F2426). The following day the complexes were washed five times with high-salt buffer (20 mM HEPES [pH 7.9], 300 mM KCl, 5% glycerol, 1 mM DTT, 100 μM Zn-acetate, 2 mM MgCl₂, 2 mM EDTA, 0.1% NP-40), boiled in SDS-protein loading buffer, separated by SDS-PAGE, and stained with Coomassie brilliant blue. The complex-nucleosome interaction assay was performed as follows. The complex was purified using the standard procedure described above. After the washing step, the complex was additionally washed once in nucleosome interaction buffer (25 mM HEPES [pH 7.5], 100 mM KCl, 10% glycerol, 100 μg/ml bovine serum albumin, 0.1% NP-40, 0.5 mM DTT, with freshly added PICs) and subsequently incubated overnight in the same buffer supplemented with 10 μg (measured by Bradford assay) of nucleosomes extracted from HEK293T cells as described above. The nucleosomes were then detected by Western blotting with histone H3 antibodies.

RNA extraction, cDNA synthesis, and Northern blotting.

RNA extraction was done as previously described (5). For cDNA synthesis, 500 ng RNA was brought to a volume of 8 μl diethyl pyrocarbonate-treated water. One microliter of RNase-free DNase was added, and the reaction mixture was incubated for 30 min at 37°C. After adding 1μl STOP solution, the samples were further incubated at 65°C for 10 min and subsequently left on ice for 2 min. Then, oligo(dT) mix (Promega) was added, and cDNA synthesis was performed according to the manufacturer's instructions. Northern blot studies were done according to a standard procedure (27). The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and RPL13a probes were labeled using [α-³²P]dCTP according to the manufacturer's instructions (Amersham RPN1633).

ChIP assays.

Chromatin immunoprecipitation assays (ChIPs) were performed as previously described (34). The primers used in this study are available upon request. Antibodies used in the ChIPs were as follows: Sin3B (Santa Cruz Biotechnology AK-12), Pf1 (Bethyl Labs BL6317), Sin3A (Santa Cruz Biotechnology AK-12), H4Ac (Upstate Biotechnology 06-866); H3 (ab1791), RNAP II (Sigma 8wg16). Normal rabbit or mouse serum was used as an IgG control.

Sin3B, but not Sin3A, interacts with Pf1.

To investigate whether mammalian Sin3 proteins, like their yeast counterparts, participate in the restoration of the repressive chromatin structure at transcribed loci, we first sought to identify the mammalian homologues of each subunit of the yeast Rpd3S complex. By sequence comparison and limited functional characterization, the homologues of Rpd3, Sin3, and Eaf3 had previously been identified as HDAC1/2, Sin3A/B, and Mrg15, respectively (1, 28, 36, 45). In contrast, the identity of the mammalian homologue of Rco1, the fourth component of the yeast Rpd3S complex, remains elusive. Recently, the purification of the Mrg15-interacting proteins led to the isolation of Sin3B, HDAC1, and most notably a PHD-containing protein, Pf1, also known as Phf12 (13). Pf1, like Rco1, contains two PHD fingers in its N terminus, both of which share the highest homology with Rco1 PHD fingers (13, 43) (data not shown). Along with Pf1's reported ability to interact with the mammalian Sin3 proteins HDAC1/2 and Mrg15, this observation suggests that Pf1 is likely to represent the mammalian homologue of Rco1. While Pf1 was originally reported to associate with Sin3A (43), surprisingly only Sin3B was identified in Mrg15-interacting proteins along with Pf1 (13). To clarify whether Pf1 interacts with Sin3A, Sin3B, or both, we performed co-IP assays (Fig. 1). As shown in Fig. 1a, ectopically expressed Pf1 associates with endogenous Sin3B but not endogenous Sin3A. The specificity of the Pf1-Sin3B interaction was confirmed in a co-IP assay with endogenous Pf1, where endogenous Pf1 was detected in Sin3B immunoprecipitate but not in Sin3A immunoprecipitate, indicating a specific interaction between Sin3B and Pf1 (Fig. 1b). Since Sin3B is known to associate with chromatin (11), we then asked whether the association between Sin3B and Pf1 occurs on chromatin. Briefly, HeLa cells were subjected to chemical cross-linking to ensure that protein interactions remained intact, and the insoluble nuclear fraction containing nucleosomes was digested with micrococcal nuclease. Using antibodies raised against the Sin3B, Pf1, or Sin3A proteins, we then immunoprecipitated the chromatin fraction associated with each protein. The presence of nucleosomes in the different immunoprecipitates was confirmed by Western blotting using a histone H3 antibody (data not shown). As shown in Fig. 1c, Pf1-associated chromatin contains Sin3B specifically, indicating that the interaction between Pf1 and Sin3B occurs on chromatin.

To further understand the nature of the Sin3B-Pf1 association and define which particular protein interaction domains mediate the interaction, we generated mutants of Sin3B and Pf1 and tested their abilities to interact by co-IP (Fig. 1d and e). Sin3B has several conserved protein interaction domains, including four PAHs (paired amphipathic helices) and one HID (HDAC interaction domain), consistent with Sin3B's putative function as a scaffold protein for complex formation (28). Within each PAH, a double point mutation was introduced to convert two leucines, previously reported to be essential for the PAH tertiary structure, into two prolines (26). Since the identity of the amino acids important for HID tertiary structure is unknown, we elected to generate a mutant deleted for the HID region, Sin3BΔHID. Following overexpression of each mutant, IP experiments were performed with an anti-Pf1 antibody, and the Pf1-interacting Sin3B proteins were detected with anti-Sin3B antibody (Fig. 1d). Of note, all mutants were stably expressed and detected by the Sin3B antibody (Fig. 1d, left panel). These experiments indicate that the interaction between Sin3B and Pf1 requires multiple domains within Sin3B, namely, PAH1, PAH3, and HID, suggesting that the interaction with Pf1 may necessitate a specific conformation of the Sin3B protein.

As mentioned above, Pf1 contains two PHD fingers. Interestingly, PHDs can function as protein-protein interaction domains, and some PHDs have recently been identified as either modified or nonmodified histone-binding modules (25, 29, 41). We therefore generated Pf1 deletion mutants for each of the PHD fingers, as well as a mutant resulting from the deletion of both fingers (Fig. 1d). Co-IP experiments between endogenous Sin3B and overexpressed wild-type or mutant Pf1 indicated that the interaction with Sin3B requires PHD2^Pf1, while PHD1^Pf1 is largely dispensable. Altogether, these results suggest that PHD2^Pf1 mediates a specific interaction with Sin3B, but not with Sin3A. It is interesting that the disruption of PHD1^Pf1 does not affect the interaction with Sin3B, consistent with what was reported for Rco1 in yeast (21).

Sin3B, Pf1, Mrg15, and HDAC1 associate in a stable complex that binds H3K4me3/H3K36me3-enriched nucleosomes.

It has been shown that Rpd3, Sin3, Eaf3, and Rco1 associate in a stable tetrameric complex (4). We therefore tested the hypothesis that all four mammalian homologues of these proteins, namely, HDAC1, Sin3B, Pf1, and Mrg15, can also stably associate in a tetrameric complex. We overexpressed the different components to limit the dependence on endogenous proteins. For each experiment, a single component was FLAG tagged, and its associated proteins were affinity purified and visualized by staining with Coomassie brilliant blue. As shown in Fig. 2a, immunoprecipitation with a FLAG antibody of either Sin3B-, Pf1-, or Mrg15-associated proteins leads to the detection of the tetrameric complex composed of Sin3B, Pf1, Mrg15, and HDAC1 (Fig. 2a, lanes 3, 6, and 9). These results strongly suggest that all four proteins can form a stable complex. Notably, the complex formation between Pf1, Sin3B, and Mrg15 remained intact upon omission of HDAC1 (Fig. 2a, lanes 2, 5, and 8), suggesting that HDAC1 does not mediate the interactions detected between the other components of the complex. Although unlikely based on the high overexpression levels for each component in this assay, it remains possible that endogenous HDAC1/2 can compensate for the absence of exogenous HDAC1 in this assay. To better understand the interdependence between all components in the formation of the complex, we used the same assay but omitted either Pf1 or Mrg15 (Fig. 2b). To ensure that the observed bands corresponded to the overexpressed proteins, Pf1 and Sin3B were expressed and purified as single components (Fig. 2b, lanes 1 and 2). While Pf1 was able to bind to Sin3B/HDAC1 in the absence of Mrg15 overexpression (lane 4), Mrg15 did not associate with Sin3B/HDAC1 in the absence of Pf1 overexpression (lane 3 and lane 5), suggesting that Pf1 mediates the interaction between Mrg15 and Sin3B/HDAC1. Next, we coexpressed Pf1-FLAG with Mrg15 and HDAC1 (Fig. 2c) and either Sin3B WT (lane 1), Sin3BΔHID (lane 2), or Sin3BΔPAH4 (lane 3). Consistent with our previous data (Fig. 1d), the Sin3BΔHID mutant was unable to associate with Pf1, while Sin3B WT and Sin3BΔPAH4 efficiently associated with Pf1. Together, these results suggest that (i) all four components can associate within a stable complex, (ii) Mrg15 does not bind Sin3B directly, and (iii) Pf1 serves as a bridging partner between Sin3B/HDAC1 and Mg15 in a manner dependent on the Sin3B HID.

Since we demonstrated that the interaction between Sin3B and Pf1 occurs on chromatin (Fig. 1c), we then asked whether the tetrameric complex, reconstituted as presented above, associates with purified nucleosomes. Additionally, we also investigated whether specific histone modifications are found enriched in the complex-bound nucleosomes. Of note, in the experiments presented in Fig. 2a to c, no histones were visualized with Coomassie staining; however, association of the complex with exogenous nucleosomes was detected by Western blotting with anti-histone H3 antibody, in the presence of Mrg15, as shown in Fig. 2d. Consistent with this observation, and in view of the established ability of Mrg15 to interact with methylated H3K36-containing nucleosomes (45), we first tested the presence of H3K36me3-modified histones in the complex-interacting nucleosomes. As anticipated, the complex efficiently interacted with H3K36me3 nucleosomes (Fig. 2e, lane 3). This interaction was largely dependent on the presence of Mrg15, as a complex composed of Pf1, Sin3B, and HDAC1 only was unable to associate with H3K36me3 nucleosomes (Fig. 2e, lane 2). Surprisingly, nucleosomes associated with a complex devoid of Mrg15 (Fig. 2e), indicating that Mrg15 is unlikely to be the only component within the complex that is able to interact with nucleosomes. Importantly, while H3K4me1-containing nucleosomes did not interact with the complex (Fig. 2e), H3K4me3 nucleosomes coprecipitated efficiently with a full tetrameric complex. Strikingly, a complex devoid of Mrg15 was almost as equally able to bind to H3K4me3-containing nucleosomes as a full tetrameric complex (Fig. 2e). As neither Sin3B nor HDAC1 is known to possess histone interaction domains, we speculated that this Mrg15-independent binding may be due to the presence of Pf1. Notably, while Pf1 by itself was not able to bind histone H3 (Fig. 2d, lane 5), the presence of Sin3B and HDAC1 promoted the interaction between Pf1 and nucleosomes (Fig. 2d and f), the molecular basis of which remains unclear. While the full tetrameric complex associates with both H3K4me3- and H3K36me3-enriched nucleosomes (Fig. 2f, lane 5), the Sin3B/HDAC1 heterodimer does not bind to nucleosomes (lane 2). Importantly, the omission of Pf1 abolished the binding to any nucleosome tested in this assay, as revealed by the absence of H3, H3K4me3, or H3K36me3 in the pulled-down nucleosomes (Fig. 2f, compare lane 3 and lane 4). Of note, the lack of association between Sin3B and H3K36me3 nucleosomes in the absence of Pf1 is consistent with our results suggesting that Pf1 is required for Sin3B to associate with Mrg15. Collectively, these results indicate that Pf1 mediates the interaction between the tetrameric complex and H3K4me3-enriched nucleosomes in an Mrg15-independent manner. We cannot rule out the possibility that H3K4me3-enriched nucleosomes are found associated with the complex via an interaction between Pf1 and a yet-unknown nucleosome modification that segregates with H3K4 methylation.

The complex associates with transcribed loci.

Having established that HDAC1, Sin3B, Pf1, and Mrg15 form a stable complex that associates with H3K4me3- and H3K36me3-enriched chromatin, we sought to determine the in vivo localization of the complex on chromatin. Since both H3K4me3 and H3K36me3 are hallmarks of active transcription, we first tested the possibility that this complex associates with the GAPDH and RPL13a loci, two prototypical constitutively transcribed genes in HeLa cells. As controls, we included CD4 and INS (insulin), transcriptionally silent genes in HeLa cells (Fig. 3a). A ChIP assay with antibodies specific to endogenous Sin3B or Pf1 demonstrated that both proteins accumulate in the region immediately downstream of the TSS of the active genes GAPDH and RPL13a (Fig. 3b, top panel). In contrast, neither protein was significantly enriched throughout the CD4 or INS loci. Importantly, no accumulation of Sin3A within any regions of these transcribed genes was detected (Fig. 4b and data not shown), which was consistent with our previous observation that Sin3A is not a part of the Pf1-containing complex (Fig. 1). In agreement with previous reports investigating the distribution of histone modifications throughout the coding region of transcriptionally active genes (20, 38), H3K4me3 levels decline downstream of the TSS, while H3K36m3 levels increase along RPL13a (Fig. 3c). Notably, both Pf1 and Sin3B are found to be significantly enriched at the very point of transition from H3K4me3- to H3K36me3-enriched nucleosomes. Together with our nucleosome interaction data (Fig. 2e and f), these results indicate that the Pf1/Sin3B-containing complex is recruited at discrete sites within actively transcribed loci, likely through its interaction with H3K4me3/H3K36me3-enriched chromatin.

Our in vitro results point to an Mrg15- and Pf1-dependent association between Sin3B and modified nucleosomes. Therefore, we asked whether Pf1 and Mrg15 also contribute to the specific localization of Sin3B within transcribed regions. Consistent with our hypothesis, the recruitment of Sin3B within the region downstream of the RPL13a TSS is strongly affected by shRNA-mediated knockdown of either Pf1 or Mrg15 levels (Fig. 4a and b). Importantly, decreasing either Mrg15 or Sin3B levels prevented Pf1 from associating downstream of the TSS, suggesting that an integral Sin3B/Pf1/Mrg15 complex may be required to allow the tethering to chromatin. Knockdown of Pf1, Sin3B, or Mrg15 did not significantly affect the expression levels of the other components of the complex (Fig. 4a). Notably, no significant change in Sin3A recruitment was observed upon the loss of Sin3B, suggesting that Sin3A cannot compensate for the loss of Sin3B in this context (Fig. 4b, bottom panel).

The tetrameric complex modulates levels of gene expression.

Having established that a Sin3B-containing complex is recruited downstream of the TSS at transcribed loci, we then asked whether this complex has any regulatory function in gene expression. As shown in Fig. 5a, Sin3B or Pf1 knockdown resulted in a significant increase in the abundance of GAPDH and RPL13a transcripts (2.6- and 2.2-fold increases, respectively) related to 18S transcripts. This increase was confirmed by Northern blotting (Fig. 5b). Notably, a similar increase in expression was detected for several other constitutively active genes, including GPI, STK19, Actin, CYC1, and CCND (Fig. 5c), suggesting that Sin3B and Pf1 may modulate the transcription of constitutively active genes in a general manner. Of note, no transcripts of aberrant size were detected upon Sin3B and Pf1 knockdown, as revealed by Northern blot analysis with probes corresponding to the 5′ and the 3′ portion of the GAPDH and RPL13a transcripts (data not shown), indicating the authenticity of the transcripts. No change in expression was observed for the constitutively silenced CD4 gene upon Sin3B and Pf1 knockdown (data not shown).

As we found that the histone deacetylase HDAC1 is an integral component of the Sin3B-containing complex, we hypothesized that the molecular mechanism by which the tetrameric complex affects gene expression might involve the modulation of histone acetylation levels within the region downstream of the TSS. Indeed, HDAC1 accumulates within the same region as Sin3B and Pf1, downstream of the TSS of RPL13a, and this accumulation is dependent on both Pf1 and Sin3B (Fig. 5d, top panel). Upon RNA interference-mediated depletion of Pf1 or Sin3B, the levels of histone H4 acetylation increased in the 5′ part of the coding region, consistent with the loss of HDAC1 recruitment (Fig. 5d, middle panel). Concurrently, RNAP II levels increased significantly upon Pf1 or Sin3B knockdown along the RPL13a transcribed region. Notably, this increase was not restricted to the 5′ end of the coding region but also occurred within and downstream of the TSS region (Fig. 5d, bottom panel). Altogether, these results strongly suggest that the Sin3B-containing complex is present downstream of the TSS of transcribed regions, preventing histone acetylation from spreading within the coding region and mitigating RNAP II progression downstream of the TSS.