Histone ubiquitinylation

A.2.3 Histone ubiquitinylation

Ubiquitinylation of histone molecules was found for histones H2A, H2B, H3, H4, H2A.Z, macroH2A and H1 (Zhang 2003; Kinyamu et al. 2005; Osley 2006). In most cases only a single ubiquitin moiety is attached to histones, which is not sufficient for targeting proteins for degradation via the 26S proteasome. During mitosis the core histones are globally

deubiquitinated at metaphase and reubiquitinylated as cells enter anaphase (Goldknopf et al.

1980), indicating a continuous turnover of the ubiquitin mark during cell growth.

Histone H1 was shown to be ubiquitinylated by TAFII250, the largest subunit of TFIID comprising both E1 ubiquitin activating and E2 ubiquitin conjugating activities (Pham and Sauer 2000). H2A was the first protein found to be ubiquitinylated (Goldknopf and Busch 1975). The ubiquitinylation site was mapped to lysine K119 within the C-terminal histone fold domain. Interestingly, although H2Aub1 accounts for 5-15% of total H2A molecules it is restricted to higher eukaryotes and absent from S. cerevisiae (Osley 2006). In contrast, ubiquitinylated H2B (ubH2B) was also found in yeast at lysine K123 and K120 in higher eukaryotes and constitutes 1-2% of the total H2B pool (Osley 2006). H2A and H2B ubiquitinylation will be discussed in detail in the following sections.

H3 is ubiquitinylated in rat spermatids, although the site is unknown (Osley 2006). The ubiquitin E3 ligase UHRF1 (A.2.2) specific for H3 was identified (Citterio et al. 2004;

Karagianni et al. 2008), but the function of H3 ubiquitinylation remains elusive. Recently, histone H4 was found to be mono-ubiquitinylated at lysines K31 and K91 in both human and yeast (Osley 2006) and the complex EEUC responsible for ubiquitinylation has been identified.

A.2.3.1 Histone H2A ubiquitinylation

Ubiquitinylated H2A (ubH2A) was long thought to play a role in activation and repression of genes due to its association with active and inactive regions of chromatin respectively.

Nowadays, ubH2A is recognized as an epigenetic mark involved in silencing. First indications came from immuno-fluorescence studies with mouse spermatocytes that showed ubH2A being present in the heterochromatin XY body of meiotic prophase cells (Baarends et al.

1999). Additional cytological studies revealed that ubH2A is also present at the inactive X-chromosome in female mammals, at unpaired autosomal regions in male meiosis and at unpaired X and Y chromosomes in female meiosis ((Baarends et al. 2005). In proliferating cells, ubH2A is present throughout the cell cycle, but it is down-regulated during G2/M transition and is not present on condensed chromosomes (Mueller et al. 1985; Joo et al.

2007).

The RING finger containing E3 ubiquitin ligases, namely mouse Ring1B, human Ring2, fly dRing, that target H2A for ubiquitinylation in vitro have been identified in both vertebrates and flies as components of the Polycomb group (PcG) complex PRC1 (A.1.3) (de Napoles et al.

2004; Wang et al. 2004a; Cao et al. 2005). In addition, PRC1 and ubH2A are enriched in regions of the genome that are subject to PcG-dependent transcriptional silencing such as the inactive X-chromosome and homeotic genes and the presence of the Ring E3 is required for H2A ubiquitinylation at these locations (de Napoles et al. 2004; Wang et al. 2004a).

Interestingly, PRC1 contains two additional RING domain - containing proteins, Ring1A and Bmi1, but unlike Ring1B, neither possesses a functional ubiquitin ligase. However, the two proteins stimulate the activity of Ring1B in vitro, and regulate the H2AK119ub1 levels at silenced Hox genes in vivo (Cao et al. 2005). Further evidence of PRC1 complexes and ubiquitinylated H2A playing direct roles in the control of transcriptional silencing came from

RNAi experiments directed against the RING E3’s. In Drosophila, RNAi-mediated knockdown of dRing resulted in derepression of the homeotic gene Ubx (Wang et al. 2004a). Similarly, deprivation of cells of Bmi1 showed expression of a number of Hox genes that had been silenced before (Cao et al. 2005). In contrast, reactivation of the silenced X-chromosome did not occur upon depletion of ubH2a through double knockout of Ring1A and Ring1B in ES cells (de Napoles et al. 2004).

Besides silencing in heterochromatic regions, ubiquitinylated H2A is also involved in the transcriptional repression in euchromatin (Osley 2006). In addition, ubiquitinylation of H2A was found to be highly dynamic, as suggested from varying global changes of ubH2A during the cell cycle, indicating the involvement in the regulation of cell cycle progression (Joo et al.

2007). Accordingly, the existence of several mammalian H2A deubiquitinating enzymes belonging to the JAMM/MPN+ family (2A-DUB) and to the ubiquitin-specific protease family (USP3, USP16/Up-M, USP22) was reported (Joo et al. 2007; Nicassio et al. 2007; Zhu et al.

2007b; Nakagawa et al. 2008; Zhang et al. 2008b).

Recently, the deubiquitinase Ubp-M (also referred to as USP16) was identified to specifically target ubH2A but not ubH2B for deubiquitination in vitro and in vivo (Joo et al. 2007). Even more important, ablation of Ubp-M in Hela cells resulted in slow growth rates owing to defects in the mitotic phase of the cell cycle and showed that H2A deubiquitination is a prerequisite for H3S10ph and subsequent chromosome segregation. Furthermore, blocking the function of Ubp-M by injection of Ubp-M antibodies in Xenopus laevis embryos resulted in defects of posterior development indicating that Hox gene expression is regulated through H2A deubiquitination and counteracts the PcG mediated repression (Joo et al. 2007).

Moreover, it was shown that chromatin assembled in vitro with ubH2A, but not H2A, specifically repressed the di- and tri-methylation of H3K4 and thus transcriptional initiation, but not elongation (Nakagawa et al. 2008). Furthermore, USP21, an ubiquitin-specific protease, was found to catalyze the hydrolysis of ubH2A and to relieve ubH2A mediated repression. Over-expression of USP21 in the liver up-regulated a gene that was normally down-regulated during hepatocyte regeneration (Nakagawa et al. 2008).

A similar observation was made for the JAMM/MPN+ domain containing H2A deubiquitinase 2A-DUB that participated in the transcriptional regulation of the androgen receptor gene (Zhu et al. 2007a). 2A-DUB forms a protein complex with the HAT p300/CBP associated factor (p/CAF) thereby regulating transcriptional initiation by stepwise coordination of histone acetylation, H2A deubiquitination and linker histone H1 dissociation.

USP22 is the mammalian counterpart of the deubiquitinase Ubp8 in yeast specifically targeting ubH2B. As for Ubp8, USP22 is a component of the human co-activator complex SAGA (Zhang et al. 2008b; Zhao et al. 2008), but in addition to H2B, USP22 deubiquitinates H2A in vitro. Furthermore, USP22 was recruited to myc target genes through direct recruitment of the Myc transcription factor.

In agreement to the role of transcriptional regulation through ubH2A, it was shown that the N-CoR/HDAC1/3 repressor complex specifically recruited the E3 ubiquitin ligase 2A-HUB that catalyzed mono-ubiquitinylation of H2A and thus mediated a selective repression of a specific set of chemokine genes (Zhou et al. 2008b). In addition, H2A mono-ubiquitinylation

acted to prevent FACT recruitment at the transcriptional promoter region thereby blocking RNA polymerase II release at the early stage of elongation (Zhou et al. 2008b).

Latest data identified the existence of ‘bivalent promoters’ (Bernstein et al. 2006) that are associated with histone marks characteristic of both active (H3K4me3) and inactive (H2K27me3) chromatin. Low-level transcription of the 5’ region of the coding region of these genes was detectable in WT cells, despite the presence of Ring1A/B and ubH2A (Stock et al.

2007). In fact, S5pCTD RNA polymerase II was situated at these promoters at levels comparable to actively transcribed genes, indicating that ubH2A hinders transcription at early stages of elongation.

As a summary, histone H2A ubiquitinylation is important for X-chromosome inactivation and Hox gene silencing mediated by Polycomb group proteins (A.1.3). Accumulating evidence indicates that H2A ubiquitinylation/deubiquitination through repressive and activating complexes regulate transcriptional silencing and activation respectively. The emergence of

‘bivalent promoters’ renders the picture even more complex and suggests that ubH2A may act as a platform for recruitment of regulatory factors responsible for the fine-tuning of transcriptional initiation and elongation processes. Among transcriptional regulation, H2A ubiquitinylation was found to be involved in cell cycle progression.

A.2.3.2 Histone H2B ubiquitinylation

Histone H2B mono-ubiquitinylation (H2BK123 in yeast and H2BK120 in vertebrates) has in general been linked to active transcription, although there is evidence for a possible role in silencing. It was shown that H2BK123ub1 was a prerequisite for H3K4me3 and H3K79me3 (Sun and Allis 2002; Wood et al. 2003a), whereas deubiquitination of H2BK123ub1 was necessary for methylation of H3K36 (Henry et al. 2003), indicative of a positive role in transcription.

In S. cerevisiae, H2B mono-ubiquitinylation is catalyzed by the enzymes Rad6 (E2) and Bre1 (E3) (Wood et al. 2003a). Additionally, the PAF elongation complex was shown to be required for RAD6-Bre1 catalytic activity but not for recruitment to the promoter (Wood et al.

2003b).

The human homologs of yeast Bre1, RNF20 and RNF40, are two RING domain proteins that are 15% identical and 28% similar to yeast Bre1 (Hwang et al. 2003). The E2 UbcH6 has been shown to interact with RNF20/RNF40 and to mediate H2B mono-ubiquitinylation together with hPAF as a trimeric complex in vitro. Furthermore, knockdown of RNF20 or components of the hPAF complex decreased whereas RNF20 over-expression increased H2Bub1 levels in vivo (Zhu et al. 2005).

In S. cerevisiae H2B ubiquitinylation can be reversed by two deubiquitinating enzymes (USPs), namely Ubp8 and Ubp10. Ubp8 as a component of the Gcn5 (HAT) - containing co-activator complex SAGA efficiently deubiquitinates H2Bub1 in vitro and in vivo, therefore combining deubiquitination and acetylation activities (Daniel et al. 2004). In fact, both H2B ubiquitinylation and Ubp8-dependent deubiquitination are required for proper transactivation, owing to a correct balance of H3K4 and H3K36 methylation set by sequential addition and removal of ubiquitin (Henry et al. 2003) (Kao et al. 2004). USP22 was found as the human

homolog of Ubp8 as part of the hSAGA/TFTC/STAGA complex (Zhang et al. 2008a; Zhao et al. 2008).

In contrast, Ubp10 (Dot4) acts independently of SAGA and in contrast to Ubp8 that deubiquitinates H2B in transcriptionally active euchromatin, interacts with Sir4 to facilitate silencing of sub-telomeric regions and rDNA in S. cerevisiae. (Emre et al. 2005; Gardner et al. 2005). However, Ubp10 may also have a role in silencing of euchromatin as well, since Ubp10 deletion strains show derepression of a number of genes that are not in the vicinity of heterochromatin (Gardner et al. 2005). In relation, USP7 has been described as an H2Bub1 deubiquitinating enzyme in Drosophila that is associated with Polycomb group proteins at the ubx-PRE (Polycomb responsive element) and at telomeres and chromocenters (van der Knaap et al. 2005). In addition, mutations of USP7 enhanced developmental defects of Polycomb protein knockout, further suggesting a role in PRC mediated silencing. Like Ubp10, USP7 might help to maintain transcriptionally inactive chromatin by removing the potentially activating ubiquitin moiety from H2B.

Several lines of evidence further support a positive role for H2Bub1 in transcriptional regulation. Bre1/RNF20 are recruited to promoters by transactivators Gal4 in yeast and p53 in human respectively, and the following ubiquitinylation of H2B was indispensible for full gene activation (Henry et al. 2003; Hwang et al. 2003). Over-expression and knockdown of RNF20 enhanced and reduced the induction of p53 - regulated p21 and MDM2 target genes (Kim et al. 2005). Moreover, Csp35, a protein that interacts with H3K4 HMT COMPASS and H3K79 HMT Dot1, is recruited to promoters by H2Bub1 (Lee et al. 2007a).

CHIP-on-CHIP analysis of ubiquitinylated H2B in human cells revealed a predominant location of this modification in the intragenic region of highly expressed genes (Minsky et al.

2008). In agreement, Rad6/Bre1 mediated H2Bub1 was abolished by mutations in subunits of protein complexes involved in transcriptional elongation (Espinosa 2008). Furthermore, H2Bub1 is required for full recruitment of Spt16 (a subunit of FACT) to intragenic regions (Fleming et al. 2008) and cooperates with FACT to promote transcriptional elongation through nucleosomal arrays (Pavri et al. 2006). In addition, Ubp8

promotes phosphorylation of Ser2 at the C-terminal domain (CTD) of RNAPII (Wyce et al.

2007), which is required for transcription elongation and co-transcriptional recruitment of mRNA processing factors (Ahn et al. 2004). A more detailed analysis in Bre1 mutants revealed that H2Bub prevented the recruitment of the Ser2-kinase Ctk1.

MDM2 is a major negative regulator of the tumor suppressor p53 and uses its RING finger domain to ubiquitinylate p53 for proteasomal degradation (Michael and Oren 2003).

Interestingly, MDM2 ubiquitinylated H2A and H2B in vitro, but predominantly H2B in vivo. In addition MDM2 and H2bub were localized to the p53 responsive gene p21. MDM2 binding to a promoter, following H2B ubiquitinylation showed silencing in a luciferase reporter assay (Minsky and Oren 2004). On the one hand this is in contrast to the above-mentioned observation that RNF20 over-expression led to increased expression of p21 (Kim et al.

2005), but on the other hand it is an indication that H2BK120ub is either not necessarily the sole determinate of active transcription or that other E3 ligases and DUB’s might regulate transcription in a gene-specific context. Considering the necessity of deubiquitination on elongation, one could imagine that poised RNA Polymerase II is stalled at the p21 promoter

upon H2B ubiquitination by MDM2. Deubiquitination of H2B would allow phosphorylation of the CTD of RNAPII and subsequent promoter escape and transcriptional elongation. USP7, interacting with both p53 and MDM2 and regulating their stability via deubiquitination (Li et al.

2002; Cummins and Vogelstein 2004) could be a candidate fulfilling H2B deubiquitination (van der Knaap et al. 2005).

Taken together, ubiquitinylated H2B is required for different steps in the process of transcriptional activation and elongation suggesting a model that multiple rounds of ubiquitinylation and deubiquitination are required for full gene induction (Henry et al. 2003;

Wyce et al. 2007). Alternatively, this mode of action was assigned to a ‘checkpoint’ model allowing for RNAPII pausing and coordination in early elongation with RNA processing (Espinosa 2008). The function of ubH2B in gene silencing seems to be restricted to the action of deubiquitinating enzymes, keeping low levels of ubH2B which in turn maintain low levels of histone marks associated with active transcription.