Histone H2A Monoubiquitination Represses Transcription by Inhibiting RNA Polymerase II Transcriptional Elongation

Identification of a RING Finger Protein 2A-HUB that Functions as a Transcriptional Repressor

We identified a histone H2A monoubiquitination ligase, the hypothetical product of the KIAA0675 clone [Genbank Accession Number AB014575], which we refer to as 2A-HUB. 2A-HUB/KIAA0675 was initially identified from a yeast two-hybrid screen that we conducted using Pitx2 (Amendt et al., 1998; Baek et al., 2003; Kioussi et al., 2002; Lu et al., 1999) as a bait, modulating Pitx2-mediated repression function (data not shown). The human KIAA0675 has been previously identified as an RNA-binding RING-dependent ubiquitin protein ligase (hRUL138) (Kreft and Nassal, 2003). The 2A-HUB protein contains an intact RING (‘really interesting new gene’) motif of the RING-H2 type defined by the consensus sequence CX₂CX_9-39CX_1-3HX_2-3C/HX₂CX_4-48CX₂C with eight cysteines and two histidines that coordinate two zinc ions (Borden, 2000), and lysine rich domain in the middle (Figure 1A). Two major bands reflecting the alternatively spliced 2A-HUB, as estimated by molecular weight determination, were detected by Western blot analysis using specific antibodies against 2A-HUB from nuclear extracts of C2C12 cells (Figure S1A). 2A-HUB exhibited a primarily nuclear localization by immunostaining with specific anti-2A-HUB IgG in C2C12 cells (Figure S1B).

To investigate the possibility that 2A-HUB is a component of a co-repressor complex, we performed gel filtration assays using nuclear extract from C2C12 cells and demonstrated that 2A-HUB eluted in a ~600 kDa fractions (Figure 1B). Intriguingly, analysis of candidate co-repressor molecules that might be present in this complex demonstrated that 2A-HUB coimmunoprecipitated with N-CoR, which also eluted in a ~600 kDa fractions (Figure 1B and 1C), but not with SMRT (Figure 1C). It also coimmunoprecipitated with HDAC1 and HDAC3, but not with HDAC2 (Figure 1C). Interestingly, an enzymatic domain-deficient mutation, 2A-HUB_mut (Kreft and Nassal, 2003) was still able to interact with N-CoR and HDAC1 (Figure 1D). To further understand the mechanism of 2A-HUB-mediated transcriptional repression, we fused full length 2A-HUB to the DNA binding domain (DBD) of Gal4. While Gal4-2A-HUB functioned as a repressor on Gal4RE luciferase reporter (Figure 1E), this 2A-HUB-dependent repression could be partially overcome by knockdown of N-CoR, HDAC1 or HDAC3, but not HDAC2 (Figure 1E and Figure S2). While Gal4-2A-HUB repression activity was only partially attenuated in cells treated with the HDAC inhibitor trichostatin A (TSA), the repression activity of Gal4-2A-HUB_mut was completely abolished, suggesting that both HDAC and ubiquitination activities are involved in 2A-HUB dependant-repression (Figure 1F). To further distinguish the ubiquitination activity of 2A-HUB mediated repression function, we performed a Gal4-2A-HUB reporter assays by knocking down the E2 enzyme UbCH5b, which exhibits specific interactions with 2A-HUB in vivo (Figure 1G and Figure S2E), observing that the repressive activity of Gal4-2A-HUB was markedly decreased (Figure 1H). More strikingly, the repressive activity of 2A-HUB was totally lost upon depletion both UbcH5b and N-CoR (Figure 1H). Taken together, these data suggest that the monoubiquitination activity of 2A-HUB exerts a combinatoric function with the NCoR/HDAC1/3 complex in mediating repression.

2A-HUB Monoubiquitinates Histone H2A in Vitro and in Vitro

RING domain-containing proteins have recently been shown to be integral parts of a second major class, in addition to HECT domain proteins, of ubiquitin ligase (or E3 enzyme) (Borden, 2000). We next investigated whether 2A-HUB can be an E3 ligase by testing its capability of self-ubiquitination. In vitro translated ³⁵Met 2A-HUB with a reaction mix containing ATP, His-tagged ubiquitin, recombinant E1 and E2 ubiquitin-conjugating enzyme UbcH5 exhibited self-ubiquitination activities (Figure2A, lane2), which was dependent on the RING finger (Figure 2A, lane 4).

To investigate the potential substrates of 2A-HUB, we examined the 2A-HUB interacting proteins. We did not observe any changes of the protein levels of N-CoR, HDAC1, and HDAC3 by knockdown of 2A-HUB in HEK293T cells (Figure 2B). In light of the ability of 2A-HUB to function as a transcriptional repressor, it appeared plausible that 2A-HUB might affect histone modifications (Jenuwein and Allis, 2001; Zhang, 2003). We therefore examined 2A-HUB and histone interactions in vivo by employing transfected Flag-tagged human H2A or H2B in HEK293T cells. Substantial amounts of 2A-HUB was coimmunoprecipitated with Flag-H2A or Flag-H2B, but was not precipitated with control IgG (Figure 2C). Furthermore, interactions at physiological levels suggested by the finding that the endogenous 2A-HUB specifically brought down endogenous H2A or H2B from HEK293T in coimmunoprecipitation assays (Figure 2D). Together, these observations imply that 2A-HUB can interact with H2A and H2B in vivo. To test whether this factor could promote H2A or H2B ubiquitination, purified histones H2A or H2B were incubated with bacterially-expressed recombinant GST-2A-HUB or 2A-HUBR_mut and His-tagged ubiquitin, under conditions appropriate for in vitro ubiquitination. As seen in Figure 2E lane 2, a His-tagged band migrating at ~25 kDa was obtained only for histone H2A, but not for histone H2B (lane 4). The size is consistent with mono His-ubiquitinated H2A. Monoubiquitination was not observed in the presence of recombinant GST-2A-HUB carrying a RING domain mutation, which abolishes E3 activity (Figure 2A). Thus, 2A-HUB is capable of promoting histone H2A monoubiquitination in vitro in a RING domain-dependent fashion.

To confirm that 2A-HUB can monoubiquitinate histone H2A in vivo, we ectopically expressed 2A-HUB in HEK293T cells, finding that histone H2A ubiquitination level was increased significantly, as detected by western blot using a specific antibody recognizing the monoubiquitinated form of H2A (uH2A) (Wang et al., 2004) (Figure 3A). In contrast, over-expression of 2A-HUBR_mut did not affect H2A ubiquitination (Figure 3A). To further investigate 2A-HUB, an ubiquitin E3 ligase that catalyzes monoubiquitination of H2A in vivo, cell extracts from HEK293T cells transfected with Flag-H2A, together with His₆-ubiquitin and 2A-HUB or 2A-HUBR_mut, expression plasmids were subjected to immunoprecipitation using Ni-NTA beads under denaturing conditions, and immunoblotted with anti-Flag antibodies to detect all forms of Flag-H2A in the precipitate. While cells transfected only with Flag-H2A and his-ubiquitin displayed barely detectable levels of monoubiquitinated Flag-H2A, presumably reflecting activity of endogenous E3 ligases, the expression of 2A-HUB resulted in a pronounced increase in ubiquitinated Flag-H2A both in whole cell lysates and in precipitates pulled down by Ni-NTA beads (Figure 3B, lane 3). In order to test whether K119 is the only 2A-HUB-dependent ubiquitination site in mammalian H2A, 2A-HUB FL or 2A-HUBR_mut expression vectors were co-transfected with Flag-H2A or Flag-H2A (K119R) (H2A harboring a mutation in the major monoubiquitinated residue) and a His₆-ubiquitin expression plasmid into HEK293T cells. While wt Flag-H2A can be monoubiquitinted by the expression of 2A-HUB (Figure 3C, lane 3), the ubiquitinated form of Flag-H2A (K119R) could not be detected in the same condition (Figure 3C, lane 7). This result suggests that 2A-HUB was not able to promote H2A monoubiquitination at sites other than the K119 residue (Figure 3C). We also examined whether 2A-HUB might promote histone H2B ubiquitination in vivo. His₆-ubiquitin plasmid and 2A-HUB or 2A-HUBR_mut were transfected into HEK293T cells and cell extracts were subjected to immunoprecipitation of his-tagged ubiquitin using Ni-NTA beads under denaturing conditions. Western blot analysis with anti-H2B antibodies was employed to detect all forms of H2B in the precipitates. Cells co-transfected with His₆-ubiquitin, and 2A-HUB or 2A-HUBR_mut displayed highly comparable levels of monoubiquitinated H2B (Figure 3D). Together, these data indicate that 2A-HUB can specifically promote H2A, but not H2B, monoubiquitination both in vitro and vivo.

2A-HUB and N-CoR Complex are Required for Repression of a Subset of Chemokine Genes Expression

N-CoR acts as an active transcriptional repressor for many nuclear receptors, NF-κB and AP-1 _target genes that regulate diverse biological processes including inflammation, cell migration, and collagen catabolism, with loss of N-CoR in macrophages, resulting in a partially activated-phenotype (Ogawa et al., 2004; Pascual et al., 2005; Perissi et al., 2004). To identify a potential cohort of genes co-regulated by N-CoR and 2A-HUB, we performed an RNA profiling analysis in RAW 264.7 (a murine leukaemic monocyte/macrophage cell line) cells treated with either control siRNA or 2A-HUB siRNA that specifically knocked down 2A-HUB protein (Figure S3A). Knockdown of endogenous 2A-HUB expression revealed a number of genes that were directly or indirectly regulated by 2A-HUB. Analysis of biological process annotations defined by the GO Consortium (Gene Ontology Consortium, 2001) associated with each differentially expressed gene, indicated statistically-significant over-representation of genes involved in several categories, including immune response, cell cycle, and pattern specification. A subset of the chemokine genes that were selectively derepressed by knockdown of 2A-HUB in RAW 264.7 cells included those previously identified as N-CoR _target genes (Figure S3B) (Ogawa et al., 2004). To verify that these identified candidate genes were differentially expressed in RAW 264.7 cells treated by control siRNA when compared to those treated by siRNA to 2A-HUB or N-CoR, quantitative real-time RT-PCR (qRT-PCR) analysis was performed. qRT-PCR from different biological samples revealed, for example, that Cxcl2/MIP2(macrophage inflammatory protein-2), Ccl5/RANTE, (regulated on activation normal T cell expressed and secreted) and Cxcl10/IP-10(interferon-inducible protein-10) were indeed derepressed in RAW 264.7 cells by knockdown of 2A-HUB using different siRNA oligo nucleotides specific for 2A-HUB, as well as by knockdown of N-CoR (Figures 4A, B and Figure S4A). These genes were also depressed in thioglycollate-elicited macrophages after siRNA knockdown of 2A-HUB (Figure S4B). In contrast, the Cxcl1/KC gene expression was increased in cells with knockdown of N-CoR, but remained unchanged in cells with knockdown of 2A-HUB (Figure 4B).

To begin to test whether 2A-HUB directly regulates RANTES and IP-10 gene expression at the promoter level, we performed transient transfection assays using luciferase reporter constructs under control of the regulatory promoter sequences from either RANTES or IP-10. The basal activity of the 220bp RANTES promoter reporter was increased by two-fold in RAW 264.7 cells treated with siRNA against 2A-HUB, in comparison to the control/scramble siRNA (Figure 4C, left panel). Induction of the RANTES promoter reporter was also increased significantly after stimulation with LPS in RAW 264.7 cells treated with siRNA against 2A-HUB (Figure 4C, left panel). In contrast, induction of the RANTES promoter reporter was significantly reduced in RAW 264.7 cells, in which full length 2A-HUB was overexpressed, but not in which the 2A-HUBR_mut was overexpressed (Figure 4C, right panel). We next examined the effect of knockdown or over-expression of 2A-HUB on the IP-10 promoter reporter activity. The activity of this reporter gene was increased both at the basal level and upon LPS treatment in response to siRNA depletion of 2A-HUB (Figures 4D, left panel). In contrast, induction of the IP-10 promoter reporter was reduced significantly in the presence of over-expressed full length 2A-HUB, while the mutant 2A-HUBR_mut had no effect (Figure 4D, right panel). Consistent with these data, the induction of endogenous RANTES or IP-10 gene upon LPS treatment was also significantly increased in RAW 264.7 cells after knockdown of 2A-HUB (Figure 4E and data not shown). Collectively, these results suggested that 2A-HUB can mediate active repression of RANTES, IP-10 and other chemokine gene expression in the presence and absence of activation stimuli. IP-10 has been shown to be a chemoattractant for activated T-lymphocytes (Rollins, 1997). RANTES, plays key role in inflammation by chemo-attracting various cell types, including T cells, macrophages, and eosinophils to inflammatory sites (Zlotnik and Yoshie, 2000). Therefore, we were interested in determining whether the increased expression of RANTES and other _targets by siRNA depletion of 2A-HUB would affect monocyte or macrophage migration. We performed cell migration experiments using supernatants taken from LPS treated RAW 264.7 cells either transfected with control siRNA or 2A-HUB siRNA. RAW 264.7 cells were strongly attracted by the source of supernatant taken from 2A-HUB knockdown macrophage stimulated with LPS compared to control siRNA treated cells (Figure 4F). These results indicated that 2A-HUB is a critical regulator of chemokine expression and chemokine-mediated cell migration.

2A-HUB negatively regulates AP-1-dependant or NF-κB dependant transcription units, consistent with the presence of common elements of AP-1 and NF-κB consensus sequences within the promoter regions of Cxcl2, Cxcl5 and Cxcl10 genes (Doyle et al., 2002). Both AP-1-luc or NF-κB-luc reporter gene activities was increased in the basal level as well as in response to TPA or TNF-α treatment in cells depleted of 2A-HUB by siRNA treatment (Figure 4G & 4H). We also examined the functional role of 2A-HUB in other N-CoR complex-regulated transcriptional events, finding that 2A-HUB negatively regulates retinoic acid receptor (RA) and estrogen receptor (ER)-dependent transcription, as revealed by reporter assays and by analysis of mRNA expression of endogenous gene _targets (Figure S5A-S5B). In contrast, PSA, a known _target gene of androgen receptor (AR), was not affected by 2A-HUB (Figure S5C). Collectively, these results reveal that 2A-HUB, along with the nuclear receptor corepressor (N-CoR), play important combinatoric roles in active repression of a cohort of repressed _target genes.

ChIP experiments revealed that both 2A-HUB and uH2A were present on the promoter regions of the endogenous MIP2, RANTES and IP-10 transcription units, but not on the KC promoter region (Figure 5A). Both 2A-HUB and uH2A were dismissed from the promoter region of MIP2, RANTES and IP-10 upon LPS stimulation (Figure 5A). We next performed a two-step chromatin immunoprecipitation experiment, first using α-N-CoR IgG and subsequently α-2A-HUB IgG. We were able to observe 2A-HUB recruitment on the N-CoR⁺ RANTES promoter, but not on the N-CoR⁺ KC promoter (Figure 5B and 5C). These data suggest that 2A-HUB is selectively required for repression a specific cohort of N-CoR- repressed chemokine genes. To determine how 2A-HUB is recruited to its _target gene promoter, we performed ChIP analysis in cells treated N-CoR siRNA. Recruitment of 2A-HUB on the RANTES promoter was significantly decreased following knockdown of N-CoR (Figure 5D), suggesting that 2A-HUB recruitment is dependent on the presence of N-CoR. Together these results indicate that H2A monoubiquitination, mediated by 2A-HUB, along with N-CoR complexes, directly regulates RANTES, IP-10 and other chemokine gene expression, and plays a role in maintaining these inflammatory response genes in a repressed state in the absence of an inflammatory stimulus.

H2A Ubiquitination Mediated by 2A-HUB Negatively Regulates Transcriptional Elongation by Preventing FACT Recruitment

FACT (for facilitates chromatin transcription), a dimeric protein that comprises human SPT16 and SSRP1 (Orphanides et al., 1998; Orphanides et al., 1999), acts to relieve chromatin-mediated inhibition of transcriptional elongation by binding and displacing the H2A/H2B dimer from the core nucleosomes (reviewed in Reinberg and Sims, 2006). Histone H2B monoubiquitination occurs at both promoters and ORFs, facilitating FACT function, thereby enhances the rate of transcription elongation on chromatin templates (Kao et al., 2004; Kim et al., 2005a; Pavri et al., 2006; Zhu et al., 2005). In contrast, histone H2A monoubiquitination primarily occurs at promoters region, repressing _target gene expression (Cao et al., 2005; Wang et al., 2004). To test the hypothesis that H2A ubiquitination might prevent FACT function, in direct opposition to the ubiquitination function of H2B, we examined the potential interaction between one of the FACT subunits, Spt16, with histone H2A or uH2A, as well as with H2B or uH2B. We observed that Spt16 specifically interacted to H2A but not with uH2A, in HEK293 cells stably expressing Flag-H2A (Figure 6A, middle panel). In contrast, Spt16 bound both histone H2B and uH2B with the same efficiency in HEK293 cells stably expressing Flag-H2B (Figure 6A, bottom panel).

In accord with this finding, RANTES gene induction by LPS was found to be impaired after depletion of Spt16 in RAW 246.7 cell (Figure 6B), suggesting that Spt16/FACT, is functionally required for RANTES gene transcription. ChIP analysis of the endogenous RANTES promoter and exon 3 regions in RAW 264.7 cells revealed that N-CoR, 2A-HUB and uH2A were all present on promoter under basal conditions and dismissed following LPS treatment (Figure 6C, left panel). Conversely, FACT and IRF3 were recruited to the promoter only after LPS treatment (Figure 6C, left panel). Importantly, FACT recruitment increased both in the promoter and in the coding region upon LPS treatment (Figure 6C). Specificity of the ChIP was demonstrated by the inability to detect N-CoR, 2A-HUB or IRF3 within the transcribed region (Figure 6C, right panel). We did not observe dramatic alternation on histone H2A recruitment under these conditions.

These studies suggest that histone H2A ubiquitination might regulate Spt16/FACT recruitment and influence RNA Pol II transcriptional elongation events. In order to investigate this possibility, we performed ChIP analysis with siRNA knockdown of 2A-HUB in RAW 264.7 in combination with LPS treatment. Under basal conditions, recruitment of 2A-HUB and uH2A at the RANTES gene promoter was decreased significantly in cells treated with 2A-HUB but not with control siRNA, indicating that 2A-HUB is crucial in maintaining the H2A ubiquitination status (Figure 6D, left panel). Consequently, recruitment of FACT to the RANTES promoter and exon 3 region was increased in cells treated with 2A-HUB siRNA (Figure 6D). Following dephosphorylation of Ser5 and phosphorylation of Ser2 of the CTD of RNA PolII by CtK1 kinase (Cdk9/P-TEFb in mammals), the Pol II enzyme is converted into a fully productive, elongation-engaged form and transits to the 3′ end (Komarnitsky et al., 2000). We therefore investigated RNA Pol II phosphorylation status in RANTES gene promoter and exon 3 region upon LPS treatment after depletion of 2A-HUB by siRNA. ChIP analysis using antibodies against Ser5- and Ser2-phosphorylated CTD (S5P-CTD and S2P-CTD) revealed significantly increased S5P signals in response to LPS at the promoter region, but not in the exonic region (Figure 6D). In contrast, S2P-CTD signal in response to LPS increased at exonic region (Figure 6D). While S5P-CTD signals in response to LPS exhibited little alternation in cells previously treated with 2A-HUB siRNA, FACT and Pol II signal, in particular, S2P-CTD signals in response to LPS in the exon region increased significantly in cells in which 2A-HUB was depleted (Figure 6D, right panel). These results suggest that H2A ubiquitination mediated by 2A-HUB plays an important role in blocking FACT recruitment, inhibiting RNA Pol II-dependent transcriptional elongation and repressing those chemokine genes expression (Figure 6E).

H2A Ubiquitination and Gene Repression

Here, we have identified and characterized 2A-HUB, an ubiquitin E3 ligase that catalyzes monoubiquitination of histone H2A in vivo, as well as in vitro. 2A-HUB functions as a transcriptional corepressor and 2A-HUB-dependent transcriptional repression requires its E3 ubiquitin ligase activity. We documented that H2A monoubiquitination-mediated by 2A-HUB is involved in repression of specific gene cohorts, such as those involved in inflammatory response in macrophages. Because there is no significant change of general H2A ubiquitination after 2A-HUB depletion (data not shown), we suggest that the role of 2A-HUB is to modulate a specifically-localized histone modification, rather than a global effect. To begin to understand the functions related to the roles of other proteins capable of catalyzing monoubiquitination of H2A, we tested whether another well known H2A E3 ligase Ring1b might be involved in transcriptional regulation of those chemokine genes regulated by 2A-HUB. In contrast to the requirement for 2A-HUB, no effect on these chemokine genes expression was incurred by depletion of Ring1b by siRNA in RAW 246.7 cells (data not shown). These results suggest that regulation of H2A ubiquitination occurs in a gene- and enzyme-specific fashion, while RANTES gene promoter region is modified by 2A-HUB, reflecting recruitment by the N-CoR complex (Figure 6E).

The fact that the RING domain of 2A-HUB is suggested to elicit efficient transcriptional repression when tethered to a promoter region implicates that 2A-HUB may contribute to transcriptional repression in gene-specific manner via local induction of histone ubiquitination. Indeed, the activity of 2A-HUB may serve to dampen the transcriptional activity, of at least some _target genes, under nonstimulated conditions. Because 2A-HUB does not exist in the biochemically purified N-CoR core complexes (Li et al., 2000; Zhang et al., 2002), it can be suggested that 2A-HUB interacts with N-CoR complexes dependent upon the specific promoter context. Interestingly, 2A-HUB was recruited by the N-CoR complex to the _target gene promoters to tightly control gene expression at transcriptional elongation level, supporting the concept that histone ubiquitination by 2A-HUB facilitates gene repression programs. While ubiquitination of other components of the transcriptional machinery by 2A-HUB is not excluded, we suggest that repression is due, at least in part, to enhanced histone H2A ubiquitination. The recruitment of at least three distinct classes of histone-modifying enzymes, histone deacetylases, E3 ubiquitin ligases and JmjC domain-containing histone demethylases by N-CoR suggests that the unique combination of epigenetic modifications is utilized to direct gene repression by N-CoR, implying that histone ubiquitination, deacetylation and demethylation are coordinated during gene transcriptional repression regulation (Klose et al., 2006; Whetstine et al., 2006; Zhang et al., 2005).

In contrast, another histone H2A E3 ligase Ring1b was found in polycomb repressive complex (PRC1), which ubiquitinates H2A, playing an important role in X-chromosome inactivation (de Napoles et al., 2004), polycomb _target gene silencing (Wang et al., 2004) and Hox gene expression (Cao et al., 2005). However, the mechanism of how H2A ubiquitination may lead to gene silencing or repression has been largely unknown. It has been proposed that H2A monoubiquitination may directly or indirectly block recruitment of the basal transcriptional machinery (Cao et al., 2005). Recent studies have suggested that PRC1 may repress transcription via direct interaction with components of the basal transcription machinery due to the presence of TFIID subunits in PRC1 (Breiling et al., 2001; Saurin et al., 2001). Furthermore, TBP and Pol II are occupied on promoters that were repressed by PRC1, suggesting that PRC1 selectively interferes with events downstream of Pol II recruitment (Breiling et al., 2001; Dellino et al., 2004). Conversely, the Drosophila trithorax group protein Kismet facilitates RNA polymerase II transcriptional elongation (Srinivasan et al., 2005). In addition, Ring1b was identified as the co-repressor BCoR complex and promoted monoubiquitinated H2A at BCL6 _target gene promoters, to direct gene repression (Gearhart et al., 2006). It is interesting that FBXL10, a JmjC-domain containing protein that has recently been identified as histone H3K36 demethylase (Klose et al., 2006) also co-purified with BCoR, suggesting that BCoR mediated _target gene repression may need a combination of histone ubiquitination and histone demethylation events. Because histone H3K36 methylation was mainly coupled with active gene elongation (Eissenberg and Shilatifard, 2006), it is possible that BCoR also represses _target gene expression by negatively regulating transcriptional elongation. Taken together, we suggest that distinct H2A E3 ligases, each recruited by different corepressor complexes, contribute to effective repression of distinct transcriptional programs.

Histone H2A Ubiquitination Negative Regulates Transcriptional Elongation

During the transcription into a nucleosomal template, RNA Polymerase II pauses at certain sites which may relate to the strength or nature of the histone-DNA contacts (Bondarenko et al., 2006; Kireeva et al., 2005). Given the fact that the transcribing RNA Pol II stalls at the first nucleosome, this and the subsequent nucleosome barriers must be relieved by FACT (Pavri et al., 2006). Here, we have found that histone H2A monoubiquitination forms a physical barrier in the nucleosome to prevent FACT recruitment, and causes RNA Pol II to pause at the promoter-proximal region. Promoter-proximal pausing is a wide spread phenomenon at many genes in eukaryotes (Kim et al., 2005b) and constitutes an important step in the regulation of Pol II transcriptional elongation. The RNA Pol II can be released rapidly from the pause into the productive elongation phase, and this highly dynamic level of regulation enables the rapid alteration of transcriptional output (Peterlin and Price, 2006; Saunders et al., 2006). In this regard, we propose that removing ubiquitin from H2A substantially increases the interaction between H2A and FACT and facilitates RNA Pol II elongation upon gene activation (Figure 6E). Therefore, we suspect that the H2A deubiquitination process is required for transcription initiation or elongation by Pol II. In light of our recent finding that histone H2A deubiquitination mediated by newly identified MPN⁺/JAMM domain–containing histone H2A deubiquitinase (2A-DUB) occurs during transcriptional initiation and elongation and 2A-DUB participates in transcriptional initiation/elongation events in androgen receptor-dependant gene activation (Zhu et al., 2007), we speculate that various H2A ubiquitinases/deubiquitinases will be discovered to regulate distinct transcriptional programs (Figure 6E).

While there is considerable knowledge with regard to transcription initiation in a chromatinized context, little is known about the subsequent post initiation events associated with transcription elongation. Our findings suggest the model that RNA Pol II pausing by histone H2A monoubiquitination might provide a checkpoint to assess whether the Pol II is correctly prepared for productive elongation, and allows rapid regulation of gene expression. Biologically, 2A-HUB appears of particular, specific important in regulating a subset of chemokine genes, modulating macrophage response to TLR signaling. Because chemokines and their receptors play a central role in recruitment, activation, and mobilization of leukocytes to sites of infection, eliminating infectious pathogens while at the same time avoiding immuno-mediated tissue damages (Rollins, 1997; Zlotnik and Yoshie, 2000), it is intriguing that 2A-HUB, along with the N-CoR complex, have evolved to combinationally attenuate expression of a cohort of chemokine genes normally induced by TLR signaling, critical for maintaining intact innate immunity. The requirement for a histone H2A monoubiquitination-dependent checkpoint is suggested to permit the appropriate dynamic range of transcriptional responses to inflammatory stimuli that regulate diverse immune responses and homeostatic processes.

Materials and Reagents

A GST-tagged 2A-HUB fragment (residues 1006–1193) was expressed in bacteria, affinity purified, and used as an antigen to produce anti-2A-HUB polyclonal antibody (Harland). Commercially available cell lines, antibodies and siRNAs are described in the Supplemental Data.

In Vivo and in Vitro Histone Ubiquitination Assays

The in vivo and in vitro ubiquitination experimental procedures are described in the Supplemental Data.

Transient transfection assay

Transfections were carried out in RAW 264.7 or HEK293T cells using Lipofectamin™ 2000 (Invitrogen), following manufacture’s protocol.

ChIP Assays

ChIP assay was conducted as previously described (Baek et al., 2003). Further details are available in Supplemental Data.