Histone Crosstalk: H2Bub and H3K4 Methylation

Abstract

Two new studies in this issue of Molecular Cell (Kim et al., 2013 and Wu et al., 2013) provide new insights and reignite debate over how histone H2B ubiquitination promotes methylation of histone H3 lysine 4.

Eukaryotic genomic DNA wraps around histones to form the nucleosome units of chromatin. Nucleosomes allow for regulated access of nuclear factors to DNA and provide a platform for recruitment of factors that do not interact directly with nucleic acids. Posttranslational modifications of histone tails mediate selective recruitment of factors with roles in gene transcription, DNA damage response, and other nuclear processes. Combinatorial and interdependent patterns of histone modifications create a multilayered system of chromatin domains and gene expression control (Suganuma and Workman, 2011). One intensively studied example of histone crosstalk is how monoubiquitination at a specific lysine on histone H2B (H2Bub) promotes histone H3 lysine 4 di- and trimethylation (H3K4me2 and H3K4me3) at actively transcribing genes (Shilatifard, 2012). In this issue, two studies describe in vitro systems that recapitulate this trans-tail dependence. Kim et al. (2013) and Wu et al. (2013) identify (in yeast and humans, respectively) the minimal requirements for H2Bub-dependent H3K4 methylation and propose new molecular mechanisms for the crosstalk.

H3K4 methylation is carried out by multicomponent histone methyltransferase (HMT) complexes such as the budding yeast Set1 complex (COMPASS or Set1C), which consists of Set1 methylase and seven additional proteins (Bre2, Sdc1, Swd1, Swd2, Swd3, Spp1, and Shg1) (Shilatifard, 2012). Humans have six homologous complexes, each containing a Set1-like catalytic subunit (Set1A/B, MLL1–MLL4), a core of common factors (Ash2, hDPY30, RBBP5, WDR5), and often some complex-specific accessory subunits. A single complex can perform mono-, di-, and trimethylation, yet the different methylation states have different locations and functions. H3K4me3 is highest near active promoters and recruits multiple factors that promote transcription, including TFIID, histone acetyltransferases, and chromatin remodelers. H3K4me2 peaks slightly further downstream and recruits a histone deacetylase. Proper H3K4 methylation is important for normal gene induction in yeast and for developmental regulation of HOX genes in higher eukaryotes, while translocations of the HMT can cause human leukemias (Shilatifard, 2012). The basis for differential placement of H3K4me2 and H3K4me3 in the genome remains unclear, but these higher levels of methylation require H2Bub while monomethylation apparently does not.

At least three nonexclusive models could explain the cross-tail connection (Figure 1). First, H2Bub could alter the configuration of the nucleosome, making the H3 tail more accessible or amenable to methylation. A second simple model is that H2Bub may bind one of the HMT subunits, tethering the complex to promote higher-level methylation by increasing the residence time of the enzyme. Alternatively, HMT binding to H2Bub could alter the catalytic properties or accessibility of the enzyme active site through an allosteric mechanism. Some experimental evidence exists for each of these possibilities, and all three mechanisms may contribute.

Figure 1. Three Models for Histone Crosstalk.

Model 1: H2Bub may change the configuration of the nucleosome H3 tail to make it more accessible to the HMT. Model 2: H2Bub may tether the HMT to the nucleosome to increase its residency time. Model 3: H2Bub binding to the HMT may trigger a conformation change that exposes or activates the catalytic site.

The first model is suggested by the observation that H2B ubiquitylation inhibits compaction of chromatin in vitro (Fierz et al., 2011). Furthermore, H2Bub is also required for methylation of H3K79 by Dot1, an HMT completely unrelated to the Set1 class of enzymes (Shahbazian etal., 2005). Importantly, H2Bub promotes H3K79 methylation on mononucleosomes in vitro without changing the affinity for Dot1, arguing the effect is direct, independent of higher order structure, and not due to a simple recruitment model (McGinty et al., 2008). For H3K79, H2Bub may instead affect the H3 tail configuration or somehow allosterically activate Dot1.

While Swd1, Swd3, and Set1 are essential for all H3K4 methylation, the other COMPASS subunits specifically affect only higher levels (Shilatifard, 2012), making them candidates for mediating the effect of H2Bub. Two previous studies implicate Swd2, although by very different mechanisms. Lee et al. (2007) proposed that Swd2 binds to H2Bub-modified nucleosomes (albeit indirectly because no in vitro interaction could be detected), independently of COMPASS. The tethering of this subunit and the recruitment of the remaining components of the HMT by the RNA polymerase II would be essential for the full activity of COMPASS. The idea that tethering the HMT promotes higher-level H3K4 methylations is consistent with the observation that COMPASS crosslinking to DNA peaks at the promoters where H3K4me3 is strongest. A more indirect model incorporating recruitment and complex structure was proposed by Vitaliano-Prunier et al. (2008), who found that H2Bub was connected to ubiquitylation of Swd2, which in turn affected association of the Spp1 subunit to promote H3K4me3. Spp1 contains a PHD finger, a domain often involved in recognizing methylated lysines in histone tails.

The two new studies reconstitute yeast COMPASS (Kim et al., 2013) and human MLL1 (Wu et al., 2013) complexes in vitro and recapitulate the H2Bub requirement for H3K4 methylation, providing an exceptional platform for addressing mechanisms of crosstalk. Neither paper sees any increase in HMT recruitment by H2Bub, instead arguing for enhancement of methyltransferase activity. Both studies observe a strong dependence on H2Bub for HMT activity, including monomethylation, on nucleosomes. This finding suggests intrinsically similar mechanisms for mono- versus di- and trimethylation and argues against multiple or radically rearranging HMT active sites as an explanation for the different methylation levels. Interestingly, both studies argue against an essential role for Swd2 in the crosstalk. The MLL complex lacks a homolog of this subunit, while COMPASS without Swd2 is actually more active for H2Bub-dependent methylation in vitro. Despite these agreements, the papers identify different requirements for COMPASS and MLL1 response to H2Bub.

Mapping the interactions of COMPASS subunits with Set1, Kim et al. find that the region N-terminal to the catalytic SET domain (n-SET) is critical for binding Spp1 and for H2Bub-dependent H3K4 methylation. An experiment reconstituting mammalian Set1A complex also finds that the Spp1 homolog CFP1 is also required for H2Bub stimulation of H3K4 methylation. The authors propose an interdomain interaction within Set1, by which H2Bub somehow causes n-SET/Spp1 to enhance activity of the SET domain. Despite this attractive model, they have so far been unable to observe any direct binding of ubiquitin or H2Bub to this module.

Wu et al. instead show that ubiquitin binds directly to the winged helix domain of Ash2, the mammalian homolog of yeast Bre2, to activate methylation by both the MLL1 and Set1A complexes. Remarkably, they show that a ubiquitin moiety can stimulate H3K4 methylation when attached directly to Ash2 and at least some other components of MLL. This result strongly argues against a primary role for H2Bub in modifying nucleosome accessibility or in recruiting the HMT. Surprisingly, the closely related MLL3 complex, which also contains Ash2, does not require or respond to H2Bub. The authors postulate that the Ash2-ubiquitin interaction induces allosteric changes in MLL1 to increase methylase activity.

The new in vitro systems for histone crosstalk will greatly facilitate mechanistic studies, but should also inspire structural studies to illuminate the proposed allosteric activation. Significant work remains to reconcile the yeast and human results with each other and with the earlier in vivo studies. However, whether there are one or multiple mechanisms, an obvious and overarching question remains: what is the biological rationale, the purpose, for having this complicated pathway?