|Home | About | Journals | Submit | Contact Us | Français|
Like that of many protein-coding genes, expression of the p21CIP1 cell cycle inhibitor is regulated at the level of transcription elongation. While many transcriptional activators have been shown to stimulate elongation, the mechanisms by which promoter-specific repressors regulate pausing and elongation by RNA polymerase II (RNA PolII) are not well described. Here we report that the transcription factor Sp3 inhibits basal p21CIP1 gene expression by promoter-bound RNA PolII. Knockdown of Sp3 led to increased p21CIP1 mRNA levels and reduced occupancy of the negative elongation factor (NELF) at the p21CIP1 promoter, although the level of binding of the positive transcription elongation factor b (P-TEFb) kinase was not increased. Sp3 depletion correlated with increased H3K36me3 and H2Bub1, two histone modifications associated with transcription elongation. Further, Sp3 was shown to promote the binding of protein phosphatase 1 (PP1) to the p21CIP1 promoter, leading to reduced H3S10 phosphorylation, a finding consistent with Sp3-dependent regulation of the local balance between kinase and phosphatase activities. Analysis of other targets of Sp3-mediated repression suggests that, in addition to previously described SUMO modification-dependent chromatin-silencing mechanisms, inhibition of the transition of paused RNA PolII to productive elongation, described here for p21CIP1, is a general mechanism by which transcription factor Sp3 fine-tunes gene expression.
A large number of protein-coding genes are regulated at the level of the transition of initiated, paused RNA polymerase II (RNA PolII) to productive elongation. Recent studies that measured the distribution of RNA PolII across Drosophila and human genomes suggest that regulated elongation is a major determinant of the pattern of gene expression (1, 2). This mode of regulation is prevalent among inducible genes activated by developmental cues, various cell signaling pathways, and stress stimuli. While pausing may be an intrinsic feature of RNA PolII that is influenced by features of the template, including the DNA sequence and the positioning of nucleosomes, it is also regulated both positively and negatively by a large number of trans-acting factors (3, 4). Following initiation, RNA PolII falls under the control of negative factors, including DRB (5,6-dichlorobenzimidazole riboside) sensitivity-inducing factor (DSIF) and the negative elongation factor complex (NELF). Together, these factors inhibit elongation in vitro by increasing the duration of intrinsic pauses (5–7). Positive transcription elongation factor b (P-TEFb), composed of CDK9 and cyclin T, functions to antagonize the negative elongation factors DSIF and NELF. The kinase activity of P-TEFb phosphorylates subunits of DSIF and NELF as well as serine 2 of the RNA PolII C-terminal domain (CTD), thereby relieving repression and promoting the transition to productive elongation (7–10). In addition to relieving the pause, P-TEFb plays important roles in the synthesis, processing, and transport of mRNA (4).
The transition of RNA PolII to productive elongation involves multiple steps and factors, which may differ among different genes and cellular contexts, providing an opportunity for gene-specific regulation. The best-described mechanism for gene-specific and signal-dependent regulation of pausing and elongation is through the recruitment of P-TEFb. Several mechanisms of P-TEFb recruitment have been reported, including interaction with transcriptional activators and specific chromatin-binding factors (4). However, recruitment of P-TEFb may not be sufficient for the transition of paused RNA PolII to productive elongation. In agreement with such a possibility, posttranslational modifications of P-TEFb may play a role in the regulation of its activity (10, 11). Furthermore, it is likely that P-TEFb-dependent phosphorylation may be antagonized by the action of opposing phosphatases. Many transcriptional activators, including Myc, NF-κB, and p53, have been shown to stimulate elongation by enhanced recruitment of P-TEFb, thereby antagonizing negative elongation factors (12–14). In contrast, the negative regulation of elongation mediated by promoter-specific factors is less well described.
Sp3 is a broadly expressed zinc finger transcription factor that is required for the postnatal survival and differentiation of bone, tooth, and hematopoietic lineages in mice (15, 16). Sp3 is highly related to Sp1, and binding sites for Sp3 and Sp1 are common promoter-proximal elements that control the expression of genes implicated in diverse processes, including cell cycle, hormone response, and housekeeping functions (17). Sp3 has two glutamine-rich transactivation domains that promote transcription activation, likely through interactions with components of the general transcriptional machinery and other cofactors, as has been described for Sp1 (18–20). When Sp3 was first characterized, a major distinguishing feature was its ability to both activate and repress transcription depending on the promoter context (21). Sp3 is posttranslationally modified by SUMO, and this modification has been shown to play an important role in the repressor activity of Sp3 (22, 23). Recent studies have demonstrated that SUMOylation of Sp3 promotes the recruitment of corepressors, including the chromatin remodeler Mi2, chromatin-associated proteins L3MBTL1 and L3MBTL2, and heterochromatin protein HP1, as well as histone methyltransferases SETDB1/ESET and SUV4-20H. Sp3-SUMO-mediated recruitment of these factors correlated with the establishment of a repressive chromatin structure characterized by the modified histones H3K9me3 (histone H3 trimethylated at Lys9) and H4K20me3 (24–26). Interactions of Sp3 with other corepressors, such as histone deacetylase 1 (HDAC1) and HDAC2, occur independently of SUMO modification (17, 27), suggesting that Sp3 can repress transcription in both SUMO-dependent and non-SUMO-dependent pathways.
The expression of the cell cycle inhibitor p21CIP1 is controlled predominantly at the transcriptional level (28). p21CIP1 is transcriptionally activated by p53, which acts on distal regulatory elements in the p21 promoter (29). p53-mediated induction of p21 plays a crucial role in mediating cell growth arrest upon exposure to DNA-damaging agents (29, 30). Other transcription factors, in addition to p53 (28), have been reported to regulate p21CIP1 in response to diverse stimuli. In particular, Sp1/Sp3 binding sites within the p21 proximal promoter have been found to contribute to transcriptional activation or repression in response to a variety of stimuli and stress signals, including nerve growth factor (NGF), butyrate, transforming growth factor β (TGF-β), and anticancer drugs such as oxaliplatin (31, 32). The p21CIP1 gene contains high levels of paused RNA PolII at the promoter in the absence of DNA damage, allowing for the rapid induction of this gene following p53 activation (12, 30, 33). However, no role for promoter-specific transcription factors in limiting RNA PolII release and elongation in the absence of an inducer has been described.
In the present study, we report that Sp3 represses p21CIP1 transcription, in a SUMO-independent manner, by inhibiting the transition of RNA PolII to productive elongation. We observed stalled RNA PolII at the promoter of p21CIP1, as well as at the promoters of other Sp3-repressed genes, and chromatin immunoprecipitation (ChIP) analyses across the p21 transcription unit revealed Sp3-dependent inhibition of the chromatin elongation marks H3K36me3 and H2Bub1 (monoubiquitination of H2B) in the body of the gene. At p21CIP1, Sp3 promoted NELF occupancy, in the absence of any increase in P-TEFb binding. We also observed Sp3-dependent inhibition of H3S10 phosphorylation at the promoters of p21CIP1 and other Sp3-repressed genes. Sp3 was found to recruit protein phosphatase 1 (PP1) to the p21CIP1 promoter, and phosphatase activity contributed to the inhibition of the p21CIP1 gene. These findings support the view that the transition of paused RNA PolII to productive elongation is an important regulatory step in the fine-tuning of p21CIP1 mRNA expression, and they further suggest that local antagonism of kinases that promote productive elongation may be one of several mechanisms by which Sp3 acts to inhibit transcription.
Short hairpin RNAs (shRNAs) were expressed from the U6 promoter in the pBluescript vector (34). The sequences targeted by the short hairpin RNAs were GGGCCATGGCACGTACGGCAA (green fluorescent protein [GFP] control), GGGAGGTTTTGTCAGCCACAC (Sp3-1), and GGGACCAACAACATCAAGAAG (Sp3-2). Wild-type (wt) Flag-Sp3 or Flag-Sp3 bearing the K539R (KR) mutation at the major SUMO acceptor site was expressed from the cytomegalovirus (CMV) promoter in pRC-Flag-Sp3. Flag-Sp3 wt and Flag-Sp3 KR were mutated to become resistant to Sp3-1 interfering RNA (RNAi) (GGGAGGTgTTaTCAGCCACAC [lowercase letters indicate mutations]). HeLa cells were grown in Dulbecco's modified Eagle's medium containing 10% synthetic fetal serum (Fetal Clone II) at 37°C. HeLa cells were cotransfected with 0.75 μg shRNA and 0.2 μg pBABEpuro plasmids/2 × 105 cells by using Lipofectamine 2000 (Invitrogen). Total protein or RNA was isolated after 60 h of selection with 1 μg/ml of puromycin (Calbiochem). For RNAi rescue, cotransfection with 0.5 to 1 μg of Flag-Sp3 RNAi-resistant constructs or an empty vector was performed. Where indicated, either 50 μM 5,6-dichlorobenzimidazole riboside (DRB; Sigma) or dimethyl sulfoxide (DMSO) was added 8 h before cells were harvested for ChIP or 8 h to 20 h before harvesting for mRNA analysis.
For flow cytometry analysis, cells transfected with shRNAs as described above were harvested by trypsinization, washed with phosphate-buffered saline (PBS), and fixed overnight in 70% ethanol. Before cell sorting, cells were washed with PBS, treated with RNase A, and stained with 20 μg/ml of propidium iodide. Cells were analyzed on a Becton Dickinson FACSCalibur instrument.
Anti-Sp3 (catalog no. sc-644), anti-N-terminal RNA PolII (H-224) (sc-9001), anti-NELF-A (sc-23599), anti-Cdk9 (sc-8338), anti-PP1 (sc-6109), anti-PP2A (sc-166034), and anti-PP2C (sc-50854) were obtained from Santa Cruz Biotechnology. Antibodies against RNA PolII (4H8) (ab5408), RNA PolII with serine 2 phosphorylated (phospho-S2) (ab5095), RNA PolII(phospho-S5) (ab5131), histone H3 (ab1791), histone H2B (ab1790), and histone H3 trimethylated at Lys36 (H3K36me3) (ab9050) were obtained from Abcam. Antibodies against histone H3 trimethylated at Lys4 (H3K4me3) (04-745), panacetylated histone H4 (06-866), monoubiquitinated histone H2B (H2Bub1) (05-1312), and histone H3 phosphorylated at Ser10 (H3S10ph) (06-570) were obtained from Millipore.
For microarray analysis, HeLa cells were cotransfected in duplicate with shRNA targeting either GFP or Sp3-1 and pBABEpuro by using Lipofectamine 2000 (Invitrogen). Total RNA was isolated after 60 h of selection with 1 μg/ml of puromycin (Calbiochem) using the RNeasy kit (Qiagen). Biotinylated cRNA was prepared from 2 μg total RNA using a one-cycle cDNA synthesis kit and a HT-IVT labeling kit (Affymetrix). Following fragmentation, 12 μg of cRNA was hybridized to the GeneChip Human Genome U133 Plus 2.0 array (Affymetrix), according to the manufacturer's protocol. For quantitative real-time reverse-transcription PCR (RT-qPCR), RNA was harvested using TRIzol lysis buffer (Invitrogen), followed by DNase I treatment, and first-strand cDNA synthesis was performed using 1 μg of total RNA and oligo(dT) for reverse priming with SuperScript III Supermix (Invitrogen). Samples were analyzed by quantitative real-time PCR with gene-specific primer pairs on a CFX96 real-time PCR detection system (Bio-Rad) using the ΔΔCT method. In each case, multiple reactions were performed using two to six independent biological replicates. Values were normalized to those of the hprt1 gene. (The sequences of all primers used are available upon request.)
Cells were lysed using radioimmunoprecipitation assay (RIPA) buffer with 1× protease inhibitor cocktail (Roche), 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM Na3VO4, 1 mM NaF, and 10 mM N-ethylmaleimide (NEM). Lysates were centrifuged to remove cellular debris and were subjected to immunoblot analysis using the antibodies indicated in the figures. Nuclear extracts were prepared as described previously (35). Briefly, 1 × 108 cells were suspended in 10 packed-cell volumes (pcv) of the hypotonic buffer (10 mM Tris-HCl [pH 7.3], 1.5 mM MgCl2, 10 mM KCl, 10 mM β-mercaptoethanol, 0.2 mM PMSF, and 10 mM NEM) and were disrupted by a Dounce homogenizer. Nuclei were collected by centrifugation at 4,000 × g for 15 min at 4°C and were resuspended in 0.5 packed-nucleus volume (pnv) of low-salt buffer (20 mM Tris-HCl [pH 7.3], 25% glycerol, 1.5 mM MgCl2, 0.2 mM EDTA, 20 mM KCl, 10 mM β-mercaptoethanol, and 0.2 mM PMSF), and 0.5 pnv of high-salt buffer (20 mM Tris-HCl [pH 7.3], 25% glycerol, 1.5 mM MgCl2, 0.2 mM EDTA, 1.2 M KCl, 10 mM β-mercaptoethanol, and 0.2 mM PMSF) was carefully added. After high-speed centrifugation for 20 min, the nuclear soluble fraction was dialyzed and used for immunoprecipitation with 3 μg of an anti-Sp3 antibody (Santa Cruz Biotechnology) and protein A-agarose beads.
Chromatin immunoprecipitation was performed by following the Upstate Biotechnology protocol with modifications. Approximately 1 × 107 HeLa cells transfected with either GFP or Sp3 shRNA were cross-linked with 1% formaldehyde for 15 min at room temperature. Cells were lysed with 1% sodium dodecyl sulfate (SDS)-lysis buffer, according to the manufacturers' protocol, in the presence of a protease inhibitor cocktail (Roche), 1 mM PMSF, and 10 mM NEM. Chromatin was sonicated to generate DNA fragments of 200 to 500 bp. One to 10 μg of serum was used to immunoprecipitate 250 μg of chromatin fraction, depending on the antibody used. Samples were incubated overnight at 4°C. During the last 4 h of incubation, 40 μl of either protein A- or protein G-agarose beads with salmon sperm DNA (Millipore) was added. After washing, cross-linking was reversed at 65°C overnight. DNA was isolated by phenol-chloroform extraction and ethanol precipitation. ChIP-enriched DNA was analyzed by real-time PCR in reaction mixtures containing 1× SYBR green mix (Bio-Rad), a 1/200 fraction of the ChIP-enriched DNA, and 200 nM primers in a total volume of 20 μl.
Sp3 is a dual-function transcription factor that can both activate and repress transcription (21). In order to further understand the mechanisms underlying the context-dependent activities of Sp3, we used RNAi to knock down Sp3, followed by microarray analysis to identify a set of endogenous Sp3-regulated genes. Immunoblot analysis demonstrated the almost-complete knockdown of Sp3 protein in HeLa cells transfected with Sp3 shRNA (Fig. 1A and data not shown). The expression of the transcription factor Sp1 was not reduced, demonstrating the specificity of the shRNA.
Changes in a large number of transcripts were observed upon Sp3 knockdown. We found that levels of 3,338 different mRNA transcripts were altered ≥1.5-fold (P, <0.01) (Affymetrix GeneChip Human Genome U133 Plus 2.0 array); of these, 1,699 (50.9%) were upregulated and 1,639 (49.1%) were downregulated after Sp3 knockdown (see Table S1 in the supplemental material). We validated our microarray data by RT-qPCR analysis of 29 candidate Sp3 target genes selected on the basis of gene set enrichment analysis (GSEA) (36) or previous studies. In our RT-qPCR analyses, 26 of the 29 candidate Sp3 target genes analyzed showed expression changes in the same direction as in the microarray study, and most showed changes of similar magnitude as well (Fig. 1B and data not shown). Similar results were obtained when a second shRNA targeting a different sequence of the Sp3 mRNA was used (data not shown).
ChIP analysis of both Sp3-repressed (mRNA level increased upon knockdown) and Sp3-activated (mRNA level decreased upon knockdown) genes demonstrated that Sp3 was bound to the promoter (Fig. 1C). As expected, occupancy of Sp3 at these promoters was significantly reduced after knockdown (Fig. 1C). Binding by the highly related transcription factor Sp1 at the promoters analyzed was not increased after Sp3 knockdown (data not shown).
Glutamine-rich motifs present in the amino-terminal region of Sp3 are likely to interact with components of the general transcription factor machinery, as has been shown for Sp1 (19, 20). At Sp3 activation targets, ChIP studies revealed that the amount of RNA PolII bound near the transcription start site was significantly reduced upon Sp3 knockdown (Fig. 1D, Activated). These data suggest that Sp3-dependent activation stabilizes the recruitment of RNA PolII to these promoters.
Interestingly, ChIP analysis showed an overall higher occupancy of RNA PolII at the promoters of Sp3-repressed genes than at those of Sp3-activated genes (Fig. 1D). Notably, when we examined targets of Sp3-dependent repression using antibodies against either the C-terminal domain (4H8, 8WG16) or the globular N-terminal region (H-224) of RNA PolII, in no case did we observe an increase in RNA PolII occupancy under Sp3 knockdown conditions (Fig. 1D and data not shown). For those repression targets with the highest levels of RNA PolII occupancy, Sp3 knockdown did not reduce RNA PolII binding to the promoter, either. These data indicate that at repressed promoters, Sp3 neither blocks nor generally promotes RNA PolII binding, and therefore, they support a postrecruitment mechanism of Sp3-dependent repression.
We identified the gene encoding the cyclin-dependent kinase inhibitor p21CIP1 (referred to below as p21) as an Sp3-repressed target gene (Fig. 1). p21 is an important negative regulator of the G0/G1-to-S-phase transition in the cell cycle, and its expression can be positively and negatively regulated by Sp1/Sp3 binding sites at the promoter (32, 37). Gene ontology (GO) classification of Sp3 target transcripts revealed a significant enrichment in genes that regulate cell cycle progression (201 genes; P, 3.4E−8) and cell cycle arrest (34 genes; P, 4.6E−4). In agreement with this, cell cycle analysis showed an increase in the fraction of cells at the G0/G1 phase of the cell cycle upon Sp3 knockdown (Fig. 2A).
p21 expression is well established to be regulated at the level of RNA PolII transcription elongation (30). In order to understand the repressive role of Sp3 in postrecruitment steps of RNA PolII, we focused on the p21 gene for a deeper analysis. ChIP analysis of Sp3 binding at several regions along the p21 promoter demonstrated enrichment of Sp3 at proximal-promoter positions correlating with previously described Sp1/Sp3 binding sites (31, 38) (Fig. 2B).
The repressor activity of Sp3 has been associated with posttranslational modification by SUMO (22, 23). Recent studies have demonstrated that repression by SUMO-modified Sp3 is associated with increased recruitment of the histone methyltransferase SETDB1, the heterochromatin protein HP1, and H3K9 trimethylation (24–26). Although we have not found broad association of H3K9me3 with the repressor activity of Sp3 on p21 and other targets analyzed (data not shown), we investigated whether SUMO modification of Sp3 was required for inhibition of the p21 gene. We used an RNAi rescue approach to compare the activities of wild-type (wt) Sp3 and SUMOylation-defective mutant (K539R) Sp3 on the p21 gene (Fig. 2C). We consistently observed that repression was restored on endogenous p21 when Sp3 RNAi was cotransfected with RNAi-resistant wt or K539R Sp3 (Fig. 2D). Similar results were obtained with several other Sp3 repression targets tested (data not shown). This is in striking contrast to findings on reporter genes, where Sp3 K539R is a potent activator (22, 23). These findings suggest that SUMOylation of Sp3 does not play a significant role in the repression of p21 and some other Sp3 repression targets. Thus, in addition to previously described SUMO-dependent mechanisms, Sp3 is able to repress transcription independently of SUMO modification.
As expected, RNA PolII was highly enriched near the p21 promoter, in contrast to low levels across the transcribed region (Fig. 3A), a hallmark of polymerase stalling (2, 39, 40). While Sp3 knockdown led to increased expression, loss of Sp3-dependent repression did not eliminate the proximal-promoter pausing of RNA PolII, suggesting that this remains a rate-limiting step for the transcription of the p21 gene (Fig. 3A).
Phosphorylation of the RNA PolII CTD at Ser5 by the CDK7 subunit of transcription factor IIH (TFIIH) is required for the initiation and promoter clearance stages of transcription (41, 42). Upon Sp3 depletion, we observed similar levels of Ser5 phosphorylation (Ser5ph-CTD) at the promoters of p21 and other Sp3-repressed genes analyzed, suggesting that Sp3 does not affect the initiation step of preloaded RNA PolII (Fig. 3B). In contrast, at Sp3-activated genes, Ser5ph-CTD was significantly reduced upon Sp3 knockdown, further supporting an Sp3-dependent activation mechanism that enhances RNA PolII recruitment and initiation (Fig. 3B).
Phosphorylation of the RNA PolII CTD at Ser2 by P-TEFb and other kinases is associated with transcription elongation (40, 43). We therefore analyzed the levels of phosphorylated Ser2 in the RNA PolII CTD. Interestingly, in the absence of Sp3, the relative peak of RNA PolII CTD Ser2ph was shifted toward more 3′ positions of the p21 gene (Fig. 3C). The finding that upon Sp3 knockdown, the levels of PolII across the p21 locus were not increased (Fig. 3A and data not shown), but a higher fraction of CTD Ser2ph was found more toward the 3′ end, suggests that Sp3 acts to limit the competency of PolII complexes for productive elongation. These data further support the model that Sp3 inhibits postinitiation steps of the transcription process.
The negative elongation factor (NELF) complex induces stalling of RNA PolII during early transcription elongation and is known to be involved in RNA PolII pausing observed at the p21 locus (12). We found that NELF was enriched at the p21 promoter as well as at the promoters of other Sp3-repressed genes (Fig. 3D). Similar to total RNA PolII, the NELF complex was more strongly associated with Sp3-repressed target genes than with Sp3-activated genes (data not shown). After Sp3 depletion, significant reductions in the level of NELF binding were observed at p21 and the other three Sp3-repressed genes with the highest RNA PolII occupancy levels. Sp3 knockdown did not affect the expression of NELF-A, and we were unable to coimmunoprecipitate NELF with Sp3 (data not shown), suggesting that the effect of Sp3 in promoting NELF occupancy at these promoters was indirect.
In order to further investigate the model that Sp3 limits productive elongation by PolII, we analyzed the levels of histone modifications that have been associated with elongating RNA PolII (44, 45) across the p21 transcription unit. Specific histone modifications, such as methylation of H3 Lys4 (H3K4me) at 5′-end regions and of H3 Lys36 (H3K36me) toward 3′ ends, as well as monoubiquitylation of H2B (H2Bub1) and deacetylation of H4 throughout the coding regions of the gene, have been associated with active transcription elongation (1, 42, 44–48). ChIP analysis across p21 showed enrichment of H3K4me3 at 5′ positions (Fig. 4A). These levels were not altered upon transcription factor knockdown. In contrast, when Sp3 levels were reduced, H3K36me3 increased significantly throughout the body of the gene but not 3′ to the poly(A) addition site (Fig. 4B). Previous studies of the p21 locus have shown an inverse correlation between H3K36me3 and H4 acetylation levels during the RNA PolII elongation process (42). In agreement with this observation, Sp3 depletion promotes a reduction in H4 acetylation levels at positions within the body of the p21 gene (Fig. 4C). In addition, Sp3 depletion also increased H2Bub1 levels at positions within the transcribed region of the p21 gene, although not at the proximal-promoter region (Fig. 4D). Knockdown of Sp3 had no effect on total levels of histone H2B or H3 detected at the positions of the p21 gene analyzed (data not shown).
In some contexts, phosphorylation of H3S10 at the promoter has been associated with enhanced release of paused RNA PolII due to increased binding of P-TEFb (49, 50), while other studies suggest additional functions of H3S10ph in transcription, including promoting an open chromatin state by preventing the spreading of H3K9me2 and HP1 (51, 52). Upon Sp3 depletion, we observed a significant increase in the level of phosphorylated H3S10 at the promoter and the proximal 5′ region of the p21 gene (Fig. 5A). In fact, the promoters of most of the Sp3 repression target genes analyzed here showed increased H3S10ph upon Sp3 knockdown (Fig. 5B), suggesting that Sp3 acts broadly to downregulate levels of H3S10ph at repressed promoters.
Taken together with the high occupancy of RNA PolII at the 5′ end and the CTD phosphorylation state, our analysis of the chromatin state at different regions across the p21 locus strongly suggests that Sp3 inhibits transcription elongation at steps subsequent to the establishment of an open chromatin structure and RNA PolII recruitment.
P-TEFb is a major activator of transcription elongation. CDK9 plays an important role not only in regulating RNA PolII activity but also in guiding cotranscriptional histone modifications, including H2Bub1 and H3K36me3 (53–56). Addition of the CDK9 inhibitor DRB did not increase the expression of p21, or that of any other Sp3 repression target analyzed, in the p53-inactivated HeLa cell line (12; data not shown). In contrast, levels of mRNA induced by Sp3 RNAi were dramatically reduced upon DRB treatment (Fig. 6A), suggesting that derepression of Sp3 target genes requires P-TEFb activity. We therefore analyzed the presence of P-TEFb at p21 and several other Sp3-repressed target promoters by ChIP of the CDK9 subunit. CDK9 was bound under repressed conditions, and in no case did P-TEFb occupancy increase upon Sp3 depletion (Fig. 6B). Paradoxically, we observed reduced CDK9 binding at some promoters under conditions of increased transcription. This reduction was blocked by DRB (data not shown), suggesting that it requires pTEFb activity and may be due to association of active P-TEFb with the elongating polymerase (12). Sp3 knockdown did not affect the expression of CDK9, and we were unable to coimmunoprecipitate CDK9 with Sp3 (data not shown). Thus, although increased H3S10ph has been shown to promote P-TEFb recruitment in some contexts (49, 50), increased expression of genes upon Sp3 knockdown was not associated with increased CDK9 occupancy at these genes.
P-TEFb (CDK9) kinase activity regulates promoter occupancy and relieves transcriptional repression by NELF (5, 6, 40). We therefore examined whether NELF binding was regulated by CDK9 kinase activity at the p21 gene. We found that the reduction in NELF occupancy upon derepression by Sp3 knockdown was blocked by DRB treatment (Fig. 6C). Similar results were seen for other Sp3-repressed genes (data not shown). Taken together, these data suggest that Sp3 inhibits the P-TEFb-dependent release of NELF at the p21 promoter without blocking the binding of P-TEFb.
In vivo, kinase activity is held in check by the functions of opposing phosphatases. Both PP1 and PP2A, but not PP2C, coimmunoprecipitated with endogenous Sp3 (Fig. 7A and data not shown). Neither PP1 expression nor PP2A expression was affected by Sp3 RNAi (Fig. 7B). Treatment with the phosphatase inhibitor okadaic acid (OA) led a significant increase in p21 mRNA levels (57) (Fig. 7C). The magnitude of p21 expression induced by OA is likely due to pleiotropic effects of these phosphatases. ChIP analysis showed that PP1 bound the p21 promoter and that this association was Sp3 dependent (Fig. 7D). We also examined p21 promoter occupancy in the presence of OA. The addition of OA did not significantly alter the binding of Sp3 to the p21 promoter (Fig. 7E). We observed that inhibition of PP2A and PP1 by OA led to increased levels of H3S10ph at the p21 promoter, and more interestingly, in the absence of Sp3, no effect of OA on H3S10ph was observed (Fig. 7E). These data indicate that local dephosphorylation of H3S10 is regulated largely by Sp3-recruited phosphatases. Further, these data suggest that antagonism of kinase activities is one of the mechanisms by which Sp3 may inhibit productive transcriptional elongation.
Regulation of transcriptional elongation by RNA polymerase II is a widely used mechanism for fine-tuning gene expression (1, 2, 54, 58, 59). While progress has been made in understanding the mechanisms involved in RNA PolII elongation control, the role of promoter-specific factors in the repression of elongation is less well understood. In this study, we provide evidence that the zinc finger transcription factor Sp3 represses basal p21 expression, independently of SUMO modification, through a mechanism that limits the transition of stalled RNA PolII to productive elongation. Our studies suggest that modulation of the local balance of kinase and phosphatase activities is one mechanism by which Sp3 inhibits transcription by bound, stalled RNA PolII. These studies support the view that the transition of paused RNA PolII to productive elongation is an important step regulated by both promoter-specific activators and repressors to finely modulate mRNA expression levels.
Previous studies have shown that in the basal state, RNA PolII is paused at the promoter of the cyclin-dependent kinase inhibitor gene p21 and that in response to stimuli, including DNA damaging agents, p53 acts to stimulate elongation and increase the expression of this gene (12, 30). We observed that Sp3 contributes to keeping basal levels of p21 expression low. In agreement with the presence of an open chromatin structure and paused polymerase at p21, we detected high levels of the active histone mark H3K4me3 at the promoter, and we observed much higher occupancy of RNA PolII at the 5′ end than across the body of the gene (Fig. 3A and and4A).4A). Sp3 neither blocked nor generally enhanced RNA PolII binding at p21 or other Sp3-repressed promoters. These findings reveal the function of Sp3 to be distinct from that of GAGA factor in Drosophila melanogaster, which promotes recruitment of RNA PolII that subsequently stalls (60, 61), but are consistent with a previous study showing that an Sp3 fusion protein targeted to a promoter-proximal RNA sequence repressed gene expression (62). Furthermore, our studies have revealed a SUMO-independent mode of repression by Sp3, distinct from SUMO-Sp3-mediated heterochromatin silencing (22–26). We suggest that promoters with paused RNA PolII are potentially sensitive to Sp3-dependent inhibition of elongation and that therefore, in the context of an open chromatin environment, the presence of paused RNA PolII is a key feature distinguishing targets of Sp3 activation and repression. Additional studies should provide further insights into how chromatin structure and factor binding at the promoter, together with SUMO modification, determine the context for Sp3-dependent activation, inhibition of elongation, and silencing.
Pausing of RNA PolII is a highly dynamic state that allows fine-tuning of gene expression in response to signals and changing cellular environments. Recent genomewide studies have revealed that paused RNA PolII appears to represent an active and tunable mechanism and that it may remain as a rate-limiting step even for highly transcribed genes (63). Our finding that enrichment of RNA PolII at the 5′ end of the p21 locus was maintained regardless of Sp3 levels indicates that Sp3 does not modify RNA PolII pausing as a rate-limiting step for p21 transcription. The genomewide studies also showed that whereas the rate of RNA PolII escape can be up- and downregulated, paused PolII is rarely entirely blocked from transcribing the body of the gene (63). In agreement with this, we observed P-TEFb binding, enriched CTD Ser2ph at the 3′ end of the gene, and a low level of p21 transcription even under conditions of Sp3-dependent repression (Fig. 3C and and6B).6B). The relative enrichment of CTD Ser2ph at more 3′ positions of the p21 gene upon Sp3 knockdown, in addition to the increases in the transcription elongation marks H3K36me3 and H2Bub1 (Fig. 3C and and4B4B and andD),D), is consistent with the model that Sp3 acts to lower the rate of escape of paused RNA PolII into productive elongation. Although we have not observed Sp3-dependent effects on the splicing or mRNA stability of p21 or other repressed target genes (data not shown), cotranscriptional mRNA processing is coordinated with promoter-proximal pausing and is regulated by many common factors, including the negative elongation factor (NELF), P-TEFb, and CTD Ser2ph (53, 55, 64), raising the possibility that Sp3-regulated pathways impact mRNA processing at some target genes.
NELF is the major factor associated with the early pausing of RNA PolII, and it has been suggested previously that NELF may control RNA PolII pausing at the p21 promoter (12). We observed strong binding of NELF to the promoter-proximal regions of p21 and other Sp3-repressed genes, correlating with the density profile of RNA PolII and supporting the presence of paused polymerase at these promoters. After Sp3 knockdown, NELF occupancy was significantly reduced at p21 and several other promoters, and this likely contributed to the observed increase in the levels of transcription of these genes. NELF-independent pausing mechanisms have also been described (2), and we consider it possible that these play a more significant role at those genes where Sp3 depletion increased transcription without reducing NELF occupancy. Several studies suggest a surprising complexity of steps and factors that contribute to the regulation of elongation. Recent investigations have demonstrated that cohesin, a factor essential for sister chromatid cohesion but also implicated in DNA repair and transcription, selectively binds genes with paused polymerase and inhibits the transition of paused polymerase to elongation at a step distinct from those regulated by NELF and DSIF (65). Thus, while Sp3 may also regulate additional mechanisms, our data show that Sp3 promotes NELF binding at the p21 promoter.
The kinase subunit of P-TEFb, CDK9, phosphorylates multiple substrates, including DSIF, the NELF-E subunit of the negative elongation factor complex, and Ser2 of the RNA PolII CTD, to promote the release of paused RNA PolII. P-TEFb also plays an important role in the cotranscriptional regulation of histone modifications, such as H3K36me3 and H2Bub1 (56, 66). CDK9 was present at Sp3-repressed genes, and its binding did not increase upon Sp3 knockdown (Fig. 6B), indicating that Sp3 does not inhibit transcription by blocking P-TEFb recruitment. The finding that reduced occupancy of NELF upon Sp3 knockdown was blocked by inhibition of CDK9 activity (Fig. 6C) is consistent with the view that, in the presence of Sp3, the activity of promoter-bound CDK9 in releasing NELF is inhibited. Taken together, our data are most consistent with the view that Sp3 acts to downregulate the transition to productive elongation via mechanisms that include antagonizing P-TEFb activity on specific substrates.
Reduced levels of H3S10 phosphorylation showed a strong correlation with Sp3-dependent repression at the p21 promoter and at the promoters of many other Sp3-repressed genes. These findings are consistent with a previous study supporting a role for a dual H3S10ph/K14Ac mark in the activation of p21 transcription (57). H3S10ph has been reported to act through a cascade of histone modifications and protein interactions to promote Brd4-dependent recruitment of P-TEFb (49, 50). In our studies, increased H3S10ph upon Sp3 RNAi was not correlated with increased recruitment of P-TEFb (Fig. 6B), highlighting the context-dependent functions of this histone modification. Brd4 has been shown to regulate the recruitment of additional elongation factors (67), and other pathways for H3S10ph-mediated activation have been proposed, including a role for H3S10ph in maintaining the active state of a gene, in part by preventing the spreading of H3K9me2 and HP1 and thus counteracting heterochromatin formation (51, 52, 68). Thus, while we have demonstrated that Sp3 inhibits the phosphorylation of H3S10 at the p21 promoter and have shown a correlation between H3S10 phosphorylation and derepression, the function of this histone modification is currently unclear.
Levels of protein phosphorylation are determined by the balance of opposing kinase and phosphatase activities. We found that both PP1 and PP2A interact with Sp3 and, furthermore, that PP1 is recruited to the p21 promoter in an Sp3-dependent manner (Fig. 7D). PP1 is an important regulator of transcription and posttranscriptional events, and in some cases, levels of PP1 at the promoter have been shown to correlate inversely with gene expression (69–71). PP2A has also been shown to localize to the p21 promoter and to antagonize the phosphorylation of H3S10 mediated by MSK1 and MSK2 (57). In agreement with these reports, the addition of the phosphatase inhibitor okadaic acid increased p21 mRNA levels. Dephosphorylation of several substrates by PP1 and/or PP2A could contribute to inhibiting the expression of p21. Several studies support a role for PP1 and/or PP2A as the H3S10 phosphatase (69, 72, 73), and our data suggest that Sp3-dependent binding of PP1 contributes to keeping H3Ser10ph levels low at the p21 promoter. Furthermore, PP1 and/or PP2A at the promoter could antagonize other stimulatory phosphorylation events, such as P-TEFb-dependent phosphorylation of NELF and DSIF, or the phosphorylation of CTD Ser2 by P-TEFb or other kinases. In fact, CDK9 itself can be phosphorylated, and although the functions of this modification are not resolved, PP1 has been shown to dephosphorylate CDK9 in vivo (10, 11, 74). Thus, our data suggest that modulation of the balance of kinase and phosphatase activities is one of several mechanisms by which Sp3 inhibits productive transcriptional elongation.
The data presented here suggest that transcription factor Sp3 acts to reduce the expression of many genes with Sp3 binding sites in their promoters by inhibiting the transition of paused RNA PolII to productive elongation. Our data further suggest that recruitment of phosphatases by promoter-specific transcription factors may help maintain paused RNA polymerase. Thus, local antagonism of the activity of promoter-bound kinases such as P-TEFb may provide an additional level of regulation that contributes to the precise regulation of gene expression.
We thank Lakshmanan Iyer at the Computational Genomics Core in the Tufts Center for Neuroscience Research (P30 NS047243) for help with microarray analysis. We are indebted to Judit Villen for help with the identification of Sp3-associated proteins. We also thank Craig Kaplan, Ananda Roy, and members of the Gill laboratory for discussions and helpful comments on the manuscript.
A.V. was supported in part by a fellowship from the Spanish Ministerio de Educacion y Ciencias. This work was supported by a grant from the National Institutes of Health (R01 GM077689) to G.G.
Published ahead of print 11 February 2013
Supplemental material for this article may be found at http://dx.doi.org/10.1128/MCB.00323-12.