Heterologous expression reveals distinct enzymatic activities of two DOT1 histone methyltransferases of Trypanosoma brucei

Post-translational modifications of histone proteins have crucial roles in DNA-related processes, such as gene regulation, DNA repair and replication (Campos and Reinberg, 2009; Dinant et al., 2008; Li et al., 2007). One highly conserved core modification is methylation of lysine 79 of histone H3 (H3K79) by the Dot1 family of histone methyltransferases. In contrast to yeast, flies, and humans, the protozoan parasite Trypanosoma brucei expresses two Dot1-like proteins that methylate histone H3K76, the T. brucei counterpart of H3K79. DOT1A is an essential enzyme, whereas DOT1B is dispensable for normal growth but is essential for the differentiation of bloodstream form to procyclic (insect-stage) trypanosomes (Janzen et al., 2006). Furthermore, DOT1B is involved in the tight repression and switching of telomeric variant surface glycoprotein (VSG) gene expression site promoters (Figueiredo et al., 2008; Stockdale et al., 2008), indicating a role for DOT1B in antigenic variation of this parasite (Taylor and Rudenko, 2006; Verstrepen and Fink, 2009).

DOT1A and DOT1B are most likely to exert their different functions by differentially affecting the methylation state of H3K76. Di- and trimethylation (H3K76me2 and K76me3, respectively) appear to be the predominant H3K76 methylation states in T. brucei (Janzen et al., 2006). Immunostaining experiments revealed that H3K76me3 is constitutively detectable throughout the cell cycle, whereas H3K76me2 is restricted to mitosis (Janzen et al., 2006). Genetic studies suggest that DOT1B is responsible for H3K76 trimethylation because inactivation of DOT1B results in loss of H3K76me3 and an increase in H3K76me2 (Janzen et al., 2006). RNAi-mediated depletion of the essential DOT1A protein results in a partial decrease in H3K76me2 (Janzen et al., 2006), indicating that DOT1A might be responsible for the H3K76me2 peak in M phase.

To determine how the H3K76 methylation pattern is controlled in vivo, it is important to understand the substrate specificities of the two DOT1 enzymes and how they can be regulated. Dot1 enzymes recognize a part of H3 on the nucleosome surface that is in close proximity to histone H4 and DNA. They, therefore, are most likely to interact with a very complex binding surface. In-vitro studies on Dot1 proteins to determine their enzymatic properties are complicated by the lack of activity of Dot1 towards free histone H3 and short peptides, and its weak activity towards reconstituted nucleosome substrates (Feng et al., 2002; Janzen et al., 2006; McGinty et al., 2008; Sawada et al., 2004). Interestingly, DOT1B efficiently trimethylates H3K76-dimethylated native chromatin isolated from DOT1B knockout trypanosomes in vitro (Janzen et al., 2006). However, whether DOT1B requires prior dimethylation of H3K76 by DOT1A remained unknown. Since DOT1A is an essential enzyme it is currently not possible to purify native chromatin substrates that lack H3K76 methylation from T. brucei for in-vitro studies. To circumvent these problems, we used a heterologous expression system in Saccharomyces cerevisiae to investigate the activities and substrate specificities of DOT1A and DOT1B.

To investigate the activity of DOT1A and DOT1B, they were expressed individually in dot1Δ yeast strains that lack endogenous H3K79 methylation. Quantitative mass spectrometry and immunoblot analysis using antibodies specific for the TAP tag and mono-, di-, and trimethylated H3K79 showed that DOT1A and DOT1B were expressed (supplementary material Fig. S1) and that they very efficiently methylated H3K79 in yeast (Fig. 1A-D). The highly conserved GxGxG motif (Fig. 1A-B) is essential for catalytic activity of Dot1 enzymes (Cheng et al., 2005; Cheng and Zhang, 2007; Frederiks et al., 2008). Catalytic mutants of DOT1A and DOT1B in which the central glycine of the GxGxG motif was mutated to arginine (DOT1A-G138R and DOT1B-G121R) showed no H3K79 methylation (Fig. 1D). We noticed that the expression of DOT1B-G121R was somewhat reduced (supplementary material Fig. S1). Interestingly, although DOT1A and DOT1B both methylated the vast majority of H3 in yeast, they showed a very distinct methylation pattern. DOT1A expression resulted in abundant H3K79me2 (Fig. 1C), which is in agreement with the observations in T. brucei (Janzen et al., 2006). Expression of DOT1B resulted in abundant H3K79me3 (Fig. 1C), confirming that DOT1B is responsible for trimethylation of H3K76 in T. brucei (Janzen et al., 2006). The almost complete absence of H3K79me1 and H3K79me2 suggests that DOT1B more efficiently converts the lower methylation states into higher methylation states than ScDot1. Importantly, our results show that DOT1B was able to trimethylate H3K79 without prior dimethylation of H3K79 by DOT1A. How this impinges on the regulation of H3K76 methylation in T. brucei is discussed below.

Having established an expression system for DOT1A and DOT1B that recapitulates the properties of the enzymes in T. brucei, we took advantage of yeast genetics to investigate the possible modes of regulation of H3K76 methylation. In yeast and humans, ubiquitylation of a C-terminal lysine on H2B (H2Bub) by the ubiquitin ligase Bre1 stimulates the activity of Dot1 (Briggs et al., 2002; Frederiks et al., 2008; Krogan et al., 2003; McGinty et al., 2008; Nakanishi et al., 2009; Ng et al., 2002; Schulze et al., 2009; Shahbazian et al., 2005; Wood et al., 2003; Zhu et al., 2005). The molecular mechanism of this activation is unknown but the positive effect of H2Bub on the activity of human Dot1 does not require additional factors (McGinty et al., 2008). Therefore, we first asked whether methylation by TbDOT1 in yeast is affected by ubiquitylation of H2BK123 (H2BK123ub). For that purpose, DOT1A and DOT1B were expressed in dot1Δbre1Δ cells, which lack endogenous H3K79 methylation and H2BK123ub. In this bre1Δ background, ScDot1 showed less H3K79me3 and more H3K79me1 than in wild-type cells (Fig. 1E), confirming the slow accumulation of methylation in the absence of H2BK123ub (Frederiks et al., 2008). The DOT1A activity was similarly affected by H2BK123ub, albeit somewhat less than ScDot1. By contrast, DOT1B showed only a subtle increase in H3K79me2 and H3K79me1 in bre1Δ cells, indicating a minor effect of H2BK123ub on this enzyme (Fig. 1E). Thus, chromatin ubiquitylation can differentially affect the activity of T. brucei DOT1 enzymes in yeast.

The negatively charged C-terminus of ScDot1 directly binds to a positively charged patch on the N-terminal tail of histone H4 in vitro, and this interaction is required for full activity of ScDot1 in vivo (Altaf et al., 2007; Fingerman et al., 2007). Many of the positively charged residues on the H4 tail are conserved between yeast and trypanosomes, although the overall sequence of the tail is less conserved than that of the core (Fig. 2A). ScDot1 also binds to nucleosomes via its N-terminal domain in vitro (Altaf et al., 2007; Sawada et al., 2004). Interestingly, DOT1A and DOT1B lack the known chromatin-binding domains (Fig. 1A), indicating that they have different targeting mechanisms or, perhaps, lack specific targeting mechanisms, the latter of which would be in line with their global activity in T. brucei (Janzen et al., 2006). Each canonical histone protein in yeast is encoded by only two genes, which offers the possibility to delete one copy and make mutations in the remaining copy. Using this possibility, we investigated the activity of DOT1A and DOT1B in yeast cells that lack the tail of histone H4 (H4Δ4-19; Fig. 2A). In agreement with previous findings (Altaf et al., 2007; Fingerman et al., 2007), in these cells the activity of ScDot1 was severely reduced compared with wild-type cells, as indicated by the loss of H3K79me3 and H3K79me2 (Fig. 2B). However, H3K79 methylation by DOT1A and DOT1B was not or only slightly reduced in cells expressing H4Δ4-19 as compared with wild type cells (Fig. 2B). Therefore, DOT1A and DOT1B did not require interactions with the H4-tail for efficient methylation in yeast.

There are several amino acid differences between trypanosome and yeast histone H3, in particular in the N-terminal tail and the sequence surrounding H3K76 and H3K79 (Fig. 3A). Our results show that DOT1A and DOT1B methylate H3K79 despite the divergent flanking sequence (Fig. 1). To more closely mimic the T. brucei H3 protein substrate, we replaced the yeast H3K79 flanking sequence (see the box in Fig. 3A) with the trypanosome H3K76 flanking sequence (Fig. 3A,B). The activity of DOT1A, DOT1B and ScDot1 towards the chimeric histone H3 in vivo was determined by immunoblot analysis using antibodies specific for TbH3K76me2 and TbH3K76me3, which were raised against methylated TbH3K76 peptides that are identical to the chimeric protein (Janzen et al., 2006), although the lysine residue is still at position 79 in yeast. Antibodies against H3K76me1 are currently not available. DOT1A generated H3K76me2 and low levels of H3K76me3 (Fig. 3C). This is similar to the activity of DOT1A towards wild-type yeast H3K79 (Fig. 1), even though yeast H3K79me and T. brucei H3K76me blots cannot be directly compared in a quantitative manner. By contrast, DOT1B expression resulted in undetectable H3K76me3 and low H3K76me2 levels (Fig. 3C). ScDot1 was unable to di-, and tri-methylate the chimeric substrate (Fig. 3C). To determine whether DOT1B requires prior methylation of the chimeric H3 protein, DOT1A and DOT1B were co-expressed. This, however, did not result in detectable H3K76me3. Therefore, DOT1B was unable to efficiently methylate the chimeric H3 protein, even when H3K76me2 was present (Fig. 3D). We expect that in the context of the full yeast nucleosome, the T. brucei K76 region may adopt an unusual structure, even though the flanks of the loop are unchanged, and that DOT1B is more sensitive to the structural change.

To understand the cell-cycle related functions of DOT1A and the role of DOT1B in control of VSG expression sites, it is important to determine the enzymatic properties and modes of regulation of these enzymes. Unfortunately, methods that have been used for other Dot1 proteins can not be applied to the TbDOT1 enzymes. Human Dot1L has been well characterized by analysis of the activity of recombinant Dot1L towards defined reconstituted recombinant chromatin substrates or purified yeast chromatin templates (Chatterjee et al., 2010; McGinty et al., 2008; McGinty et al., 2009; Mohan et al., 2010). Attempts to reconstitute trypanosome chromatin have failed so far, and one of the two DOT1 proteins is essential, precluding the possibility to purify nucleosomes with unmodified H3K76 from DOT1-deficient parasites (Janzen et al., 2006). The enzymatic properties of yeast Dot1 have been best characterized by in vivo manipulation of the single yeast Dot1 enzyme using genetic tools (Frederiks et al., 2008; van Leeuwen et al., 2002). Since DOT1A is essential, a similar genetic approach is not possible for trypanosomes (Janzen et al., 2006). To circumvent these restrictions, we developed a heterologous expression system. Heterologous expression has previously been successfully applied for characterization of many other proteins (e.g. Jeong et al., 2007; Kunz et al., 2004; Luo et al., 2002; Marini et al., 2008). One possible limitation of heterologous expression is that the T. brucei DOT1 proteins expressed in yeast might lack their putative natural binding partners or might interact with unknown yeast proteins. Other approaches have similar intrinsic limitations, however. The activity of DOT1 proteins in their native environment in trypanosomes might also be influenced by putative interactions with additional factors and in in-vitro studies (if developed in the future) it is possible that crucial components or conditions are missing. However, the TbDOT1 proteins are mainly composed of a conserved methyltransferase core domain and lack the domains that are found in yeast or human Dot1 and that are known to be involved in protein- and DNA-interactions (Fig. 1). Furthermore, the observed activities of DOT1A and DOT1B in yeast are in agreement with our previous observations in T. brucei (Janzen et al., 2006).

Our characterization of the T. brucei Dot1 enzymes in yeast showed that DOT1A and DOT1B have distinct catalytic properties. Therefore, regulation of the degree of histone methylation can be achieved by differential regulation of the two enzymes, something that is not possible in yeast or humans, where a single Dot1 enzyme is present. That DOT1B can di- and trimethylate H3K79 in the absence of prior H3K76 methylation by DOT1A (Fig. 1) has implications for the regulation of DOT1B to establish the appearance of H3K76me2 during M phase. Using immunostaining techniques, H3K76me3 is detectable throughout the T. brucei cell cycle, whereas H3K76me2 is not detectable from G1 until the end of G2, and restricted to M phase and cytokinesis (Janzen et al., 2006). Our new findings suggest that DOT1B is most likely to be inactive during S and M phases towards the newly deposited unmodified histones to allow dimethylation of unmethylated H3K76 by DOT1A in M phase (Fig. 4). Briefly, our findings, together with our previous observations (Janzen et al., 2006) suggest (1) no or low activity of DOTA and DOT1B in S phase and G2, when new unmodified histones are present in the chromatin, (2) subsequent activation of DOT1A to dimethylate H3K76 on the new histones in M phase and, (3) activation of DOT1B after exit from M phase to convert lower H3K76 methylation states including H3K76me2 to H3K76me3 (Fig. 4). Alternatively, the H3K76 methylation state might be regulated by cell cycle-specific demethylases. However, no H3K76 or H3K79 demethylases have been described to date, and this model does not explain why DOT1A is present and why it is required for normal cell cycle progression.

Although the TbDOT1 proteins were not expressed in their native environment, our results offer some suggestions for possible mechanisms of regulation of H3K76 methylation in trypanosomes. For example, DOT1A activity was higher in yeast cells that contain ubiquitylated H2B (Fig. 1E). However, the specific histone crosstalk between H2B and H3K79 or H3K76 does probably not exist in T. brucei because the canonical H2B of T. brucei does not contain a lysine at the C-terminal tail, and K129 of the histone variant H2BV is not required for H3K76 methylation and most likely not ubiquitylated (Mandava et al., 2008). Therefore, the mechanism by which H2Bub affects Dot1 enzymes might be rather non-specific; for example, by changing the chromatin structure and, thereby, altering access to the substrate lysine on the nucleosome core. Another possibility is that ubiquitylation of another chromatin protein complements the role of H2Bub in T. brucei. This is not unlikely because it has recently been show that the exact structure of the ubiquitin moiety and its position on the nucleosome core are not crucial for the activation of human Dot1 (Chatterjee et al., 2010; McGinty et al., 2009). Finally, our finding that the ScH4-tail is not required for activity of DOT1A or DOT1B suggests that H3K76 methylation by DOT1A and DOT1B is not directly modulated by interactions of other chromatin proteins with the histone H4 tail, a mechanism that has been suggested for the yeast heterochromatin protein Sir3 (Altaf et al., 2007; Fingerman et al., 2007; Norris and Boeke, 2010). For a full understanding of the possible mechanisms of regulation of DOT1A and DOT1B it will be important to develop new genetic and biochemical tools to determine the behavior of the two proteins in their native environment and in defined in-vitro assays.

Yeast cells were grown as described previously (Frederiks et al., 2008). Strains and plasmids are described in the supplementary material Tables 1 and 2.

Histone methylation was analyzed by mass spectrometry and immunoblots as described previously (Frederiks et al., 2008). Primary antibodies used for immunodetections: H3, H3K79me1, H3K79me2 and H3K79me3 (Frederiks et al., 2008), Pgk1 (Invitrogen, A-6457), TAP (Open Biosystems, CAB1001), HA (12CA5), H3K76me2 and H3K76me3 (Janzen et al., 2006).

We thank Saara Vainio and Pankaj Tripathi for whole-cell lysates of T. brucei and Leishmania tarentolae. We thank members of the van Leeuwen and Janzen laboratories, and Piet Borst and Michael Boshart for critical reading of the manuscript. This work was supported by the EU 6th framework program (NOE ‘The Epigenome’ LSHG-CT-2004-503433), The Netherlands Organization for Scientific Research (NWO), The Netherlands Genomics Initiative and by the SFB TR5 of the LMU Munich. The authors declare that they have no competing financial interest.