Structure activity analysis of semisynthetic nucleosomes: Mechanistic insights into the stimulation of Dot1L by ubiquitylated histone H2B

Abstract

Post-translational modification of histones plays an integral role in regulation of genomic expression through modulation of chromatin structure and function. Chemical preparations of histones bearing these modifications allows for comprehensive in vitro mechanistic investigation into their action to deconvolute observations from genome-wide studies in vivo. Previously, we reported the semisynthesis of ubiquitylated histone H2B (uH2B) using two orthogonal expressed protein ligation (EPL) reactions. Semisynthetic uH2B, when incorporated into nucleosomes, directly stimulates methylation of histone H3 lysine 79 (K79) by the methyltransferase, disruptor of telomeric silencing-like (Dot1L). Although recruitment of Dot1L to the nucleosomal surface by uH2B could be excluded, comprehensive mechanistic analysis was precluded by systematic limitations in the ability to generate uH2B in large scale. Here we report a highly optimized synthesis of ubiquitylated H2B bearing a Gly76Ala point mutation (uH2BG76A), yielding tens of milligrams of ubiquitylated protein. uH2BG76A is indistinguishable from the native uH2B by Dot1L, allowing for detailed studies of the resultant trans-histone crosstalk. Kinetic and structure activity relationship analyses using uH2BG76A suggest a non-canonical role for ubiquitin in the enhancement of the chemical step of H3K79 methylation. Furthermore, titration of the level of uH2B within the nucleosome revealed a 1:1 stoichiometry of Dot1L activation.

Histones harbor an extraordinary density of post-translational modifications, including acetylation, methylation, phosphorylation, and ubiquitylation (1, 2). Acting directly through enhancement or abbrogation of internucleosomal contacts (3) or indirectly through the recruitment of position-specific binding modules (4), these modifications direct the complex expression of the genome. Monoubiquitylation of H2B on K120 (5) is implicated in diverse nuclear processes, ranging from DNA damage repair and cell-cycle checkpoint regulation, to transcriptional elongation and stimulation of histone lysine methyltransferase activity (6–8). However, the mechanistic role of ubiquitin in these contexts remains elusive. In organisms ranging from yeast to humans, ubiquitylation of H2B is a prerequisite for efficient methylation of H3K79 by Dot1 (9–12). H3K79 methylation maintains transcriptional silencing by demarcating euchromatin-heterochromatin boundaries (13), and much like uH2B, is integral to DNA damage repair pathways (14). Aberations of H3K79 methylation are pathogenic in a subset of mixed lineage leukemia (MLL) fusion leukemias (15). Investigation of the regulation of H3K79 methylation by H2B ubiquitylation is critical to understanding the role of these modifications in normal development and disease states.

In order to elucidate the mechanism of stimulation of Dot1L by uH2B, it is necessary to purify or generate significant quantities of homogenously ubiquitylated H2B. Isolation from lysates is complicated by the coexistence of multiple post-translational modifications of H2B, while in vitro reconstitution of the ubiquitin conjugating enzymes leads only to modest yields (9, 16). Moreover, both of these approaches restrict inquiry to native ubiquitin and H2B sequences. Previously, we described the semisynthesis of native uH2B using two orthogonal EPL reactions, ensuring chemical homogeneity and bypassing the cellular ubiquitylation machinery (17). EPL allows for the formation of an amide bond between two polypeptides of recombinant and synthetic origins, one containing a C-terminal thioester, and the other an N-terminal cysteine (18). The absence of cysteines in uH2B necessitates the combination of three polypeptides using two traceless protein ligation strategies: one employing a photolytically removable ligation auxiliary and the other, a chemical desulfurization (Figure 1, panel a and Supplementary Figure 1). We used this semisynthetic uH2B to demonstrate a direct stimulation of Dot1L-mediated intranucleosomal methylation of H3K79 (17). Surprisingly, the stimulatory effect of uH2B is not a result of recruitment of Dot1L to the nucleosomal surface, as Dot1L is able to bind to nucleosomes in the absence of ubiquitylation, suggesting that uH2B may stimulate Dot1L activity through allosteric regulation of Dot1L or the nucleosome itself. At the time, detailed biochemical studies were precluded by limitations in scalability of our semisynthetic strategy. Here we report a highly optimized semisynthesis of uH2B bearing the single G76A point mutation at the C-terminus of ubiquitin, allowing the rapid preparation of tens of milligrams of ubiquitylated protein. Nucleosomes containing uH2BG76A were indistinguishable by Dot1L from those containing uH2B, permitting both a comprehensive kinetic analysis of the role of uH2B in H3K79 methylation and the first structure activity relationship investigation of this complex system.

Figure 1. Semisynthesis of uH2BG76A.

a, Retrosynthetic comparison of uH2B (top) and uH2BG76A (bottom) syntheses. Both were generated via a 3-piece ligation strategy with the following polypeptides: synthetic peptide containing residues 117–125 of H2B and bearing an A117C mutation, H2B-C, 1a and 1b; recombinant ubiquitin(1-75)-α-thioester 2; and recombinant H2B(1-116)-α-thioester 5. For the semisynthesis of uH2BG76A, the ligation auxiliary was replaced with a cysteine (blue) and the photolytically removable cysteine protecting group was replaced with a thiazolidine (green). The resultant ubiquitylated proteins differ only at position 76 of ubiquitin (yellow). Dashed lines indicate junctions formed by EPL reactions. b, Synthetic scheme for the generation of uH2BG76A. i) EPL was used to ligate peptide 1b to protein 2, forming branched protein 3. ii) Ligation product 3 was treated with methoxylamine at pH 5, affording 4. iii) Ligation of protein 4 to protein 5, forming uH2BA117C/G76A, 6. iv) Raney nickel or radical-initiated desulfurization of protein 6, forming uH2BG76A, 7. R = CH₂CH₂CH₂C(O)NHCH₃; R′ = CH₃; R″ = CH₂CH₂SO₃H.

RESULTS and DISCUSSION

Highly optimized uH2BG76A semisynthesis

We first sought to design a new uH2B semisynthesis, optimizing for efficiency and scalability, while minimizing changes to the native sequence. Central to the inefficiency and low yield of the uH2B semisynthesis is the use of a photolytically removable ligation auxiliary (Figure 1, panel a and Supplementary Figure 1). Firstly, the preparation of the auxiliary requires a complex nine-step solution phase synthesis, ultimately limiting the quantity that can be incorporated into a peptide to be used in EPL (19–21). Furthermore, the auxiliary-mediated ligation reaction requires a minimum of five days to reach 60% completion, owing from the sterically hindered ligation onto a disubstituted amine (17, 20). Finally, photolytic removal of the ligation auxiliary, concomitant with the cysteine-protecting group, is poorly scalable and cannot be performed in parallel without multiple irradiation sources. We reasoned that a traditional cysteine-mediated ligation could be substituted for the auxiliary-mediated ligation to attach ubiquitin to an H2B C-terminal peptide, thus alleviating the costly constraints imposed by the ligation auxiliary (Figure 1, panel a, blue). Further replacement of the photolytically removable S-(o-nitrobenzyl) cysteine protecting group in 1a with a chemically-labile group would eliminate the need for an irradiation step entirely (Figure 1, panel a, green). Following orthogonal ligation to the N-terminal sequence of H2B, the ligation site cysteines could be reduced to alanines, leaving a G76A substitution at the C-terminus of ubiquitin as the single sacrifice of the optimized semisynthetic strategy (Figure 1, panel a, yellow).

To this end, peptide 1b corresponding to residues 117–125 of H2B, bearing an A117C substitution, was synthesized (Supplementary Figure 2). Orthogonal side-chain protection of K120 allowed the selective installment of a cysteine through an isopeptide bond with the ε-NH₂ of K120. This cysteine facilitates ligation to ubiquitin and following desulfurization results in the G76A mutation. Transient protection of the N-terminal cysteine in 1b as a thiazolidine restricts ubiquitin ligation to the desired site (Figure 1, panel a).

In the first step of the synthesis, an excess of peptide 1b was ligated to recombinant ubiquitin(1-75)-α-thioester, 2 (Supplementary Figure 3), generated by thiolysis of the corresponding intein fusion protein (Figure 1, panel b, step i). Thioester 2 was quantitatively converted into ubiquitylated peptide 3 within two hours (Supplementary Figure 4). Treatment with methoxylamine in situ at pH 5 for 12 h led to complete conversion of the thiazolidine to a cysteine, affording branched protein 4 (Figure 1, panel b, step ii, and Supplementary Figure 4). Notably, purification was unnecessary between the first two synthetic steps, further increasing the overall efficiency of the semisynthetic strategy and limiting losses. Protein 4 was then ligated to an excess of H2B(1-116)-α-thioester, 5 (Supplementary Figure 3), yielding full-length ubiquitylated H2B bearing a cysteine at each of the ligation junctions, 6 (Figure 1, panel b, step iii, and Supplementary Figure 4). Raney nickel (22) or radical-initiated (23) desulfurization smoothly converted both cysteines in 6 to alanines, leaving the native A117 in H2B and a G76A mutation in ubiquitin, thus generating uH2BG76A, 7 (Figure 1, panel b, step iv, and Figure 2, panels a and b).

Figure 2. Validation of uH2BG76A as a surrogate for uH2B.

a, Reversed phase high performance liquid chromatography (RP-HPLC) chromatogram of purified uH2BG76A, 7. b, Electrospray ionization mass spectrometry (ESI-MS) spectrum of purified uH2BG76A, 7. Charge states are labelled. [(M+H)⁺ observed = 22,380 ± 4 Da (s.d.). (M+H)⁺ expected = 22,379 Da.] c, UCH-L3-mediated hydrolysis of uH2B and uH2BG76A. Coomassie stained gel of assay samples quenched after indicated times. + indicates addition of UCH-L3 at 0 min; ++ indicates addition of UCH-L3 at 0 and 10 min. d, Western blot of uH2B and uH2BG76A with linkage specific α-uH2B antibody (top panel). Western blot with α-ubiquitin antibody (middle panel) and ponceau stain (bottom panel) represent loading controls. e, Dot1L methyltransferase assay on uH2B and uH2BG76A containing nucleosomes. Nucleosomes methylated with ³H S-adenosyl methionine (SAM) were separated on native gels and stained with ethidium bromide (middle panel) prior to probing for ³H methyl incorporation by fluorography (top panel). Quantification of methyltransferase activity was performed by filter-binding followed by liquid scintillation counting (bottom panel). Error bars represent one s.d. (n = 3).

With this revised semisynthetic route, tens of milligrams of ubiquitylated protein were routinely generated – an increase in excess of ten-fold over our previous methodology. Moreover, by eliminating the sluggish auxiliary-mediated ligation and subsequent purification, the time required to generate ubiquitylated H2B (steps i–iv in Figure 1, panel b) was halved. Eliminating the need for photolytic removal of protecting groups not only further enhanced the scalability of the semisynthesis, but also negated the need for specialized optics. Most importantly, all materials required for the solid phase synthesis of peptide 1b are commercially available, obviating time intensive solution phase organic synthesis and significantly increasing the accessibility of this methodology.

uH2BG76A is indistinguishable from uH2B

We envisioned that a G76A mutation in ubiquitin would be tolerated by any biochemical processes that recognize elements of ubiquitin and its target protein without specifically engaging the isopeptide bond between them (24). In these cases, the ubiquitylated protein with the G76A mutation would likely be indistinguishable from the native protein. Conversely, we reasoned that processes recognizing the isopeptide bond directly might be disrupted by the mutation. Indeed, the ubiquitin G76A mutant is efficiently mobilized and transferred to target proteins in vivo, but the conjugated product is resistant to many deubiquitylating enzymes (DUBs) (25, 26). Surprisingly, uH2BG76A was recognized and deubiquitylated by ubiquitin C-terminal hydrolase L3 (UCH-L3), to a similar extent as the native control (Figure 2, panel c). Cleaved ubiquitin and ubiquitin G76A were identified by liquid chromatography-mass spectrometry, verifying hydrolysis of the isopeptide bond (Supplementary Figure 5). Furthermore, recognition of uH2B by a linkage-specific anti-uH2B antibody was only slightly decreased by introduction of the G76A mutation (Figure 2, panel d). Key to our goals was the effect of the uH2BG76A mutation on the stimulation of Dot1L. In order to test this, uH2B (17) and uH2BG76A were reconstituted into nucleosomes with recombinant histone proteins (Supplementary Figure 6) and 147 bp of the 601 positioning sequence (27) and submitted to radioactive methyltransferase assays with full-length Dot1L, isolated from an insect cell over expression system. uH2BG76A stimulated methyltransferase acitivity to levels comparable to uH2B (Figure 2, panel e), validating this mutant as an equally acceptable surrogate for the native sequence in biochemical analysis.

Kinetic analysis of Dot1L stimulation

Disruption of H2B ubiquitylation in organisms from yeast to humans decreases levels of higher order methylation of H3K79 without altering levels of monomethylation (10, 28). However, in vitro ubiquitylation of H2B drastically increases the level of Dot1L-mediated monomethylation and dimethylation in nucleosomes (17). This pattern is observed with full-length Dot1L as well as the catalytic domain alone. We predicted that analysis of steady-state kinetics would provide insight into the effect of uH2B on the ability of Dot1L to perform the initial methylation of H3K79. Accordingly, the catalytic domain of Dot1L, containing the first 416 residues of the methyltransferase, was isolated following overexpression in Escherichia coli as a GST fusion protein (Dot1L_cat). Nucleosomes containing H2B or uH2BG76A were prepared with the 146 bp α-satellite palindromic sequence (29). Initial rates of nucleosome methylation by Dot1L_cat were obtained by measuring transfer of a radioactive methyl group from S-adenosyl methionine, illustrating a vast rate enhancement due to ubiquitylation (Figure 3, panel a). At early time points, only monomethylation of H3K79 was detected, allowing for the direct steady-state velocity analysis of the first methylation event (Figure 3, panel b). Under steady-state conditions (Figure 3, panel c), Dot1L_cat clearly followed Michaelis-Menten kinetics in the monomethylation of ubiquitylated nucleosomes (K_m of 3.3 ± 0.6 μM and a k_cat of 1.16 ± 0.09 s⁻¹). However, under all reasonable conditions tested, Dot1L_cat failed to show any concentration dependence toward unmodified nucleosomes. This could result from an exorbitantly high K_m or an immeasurably low k_cat. The latter could reflect an extremely slow chemical step and/or a decreased rate of product release. To probe this further, we performed Dot1L_cat methyltransferase reactions under single turnover-type conditions (Figure 3, panel d). Even at saturating levels, Dot1L_cat was far more efficient in methylation of ubiquitylated nucleosomes, suggesting an increase in the velocity of the chemical step of the first methylation. While these experiments do not prevent dimethylation, early time points are likely restricted to a single methylation, mimicking a single turnover paradigm. This suggests a conformational change in the active site complex of Dot1L bound to H3K79.

Figure 3. Kinetic analysis of Dot1L-mediated methylation of uH2BG76A nucleosomes.

a, Representative time course of Dot1L_cat-mediated methylation of 0.9 μM nucleosome bearing H2B (squares) or uH2BG76A (diamonds). Linear regression models of the data are shown. Error bars represent one s. d. (n = 2). b, LC-MS/MS chromatogram showing K79 monomethylation (K79me1) in the absence of dimethylation for uH2BG76A nucleosomes at 1.8 μM. [M+2H]²⁺ charge states are labelled. An asterisk marks a contaminant that is not dimethylated K79. c, Steady-state kinetic analysis of Dot1L_cat activity of nucleosomes bearing H2B (squares) or uH2BG76A (diamonds). Fit of uH2BG76A data to the Michaelis-Menten model is shown. Error bars represent one s. d. (n = 6). d, Kinetic analysis of methylation of nucleosomes bearing H2B (squares) or uH2BG76A (diamonds) by excess Dot1L_cat. X-axis is shown in log₁₀ scale to emphasize early time points. Single exponential modelling of the data is shown. Error bars represent one s. d. (n = 3).

The calculated K_m for Dot1L-mediated monomethylation of ubiquitylated nucleosomes is similar to values calculated for other histone lysine methyltransferases acting on peptide and protein substrates (30–35); however, the k_cat of Dot1L on ubiquitylated nucleosomes is 10 to 100 times greater than reported values for these other reactions, suggesting a rate enhancement on nucleosomal substrates. Similar to yeast Dot1 (36), Dot1L appears to be distributive enzyme, as evidenced by its ability to monomethylate an excess of substrate without measurable accumulation of dimethylation (Figure 3, panel b). Although a direct comparison of kinetic parameters of Dot1L on ubiquitylated and unmodified nucleosomes was not possible, it is clear that Dot1L is able to monomethylate uH2B bearing nucleosomes to a much greater extent than unmodified nucleosomes. This seems contradictory to in vivo experiments where levels of H3K79 monomethylation are apparently unaffected by genetic disruption of H2B ubiquitylation (10, 28). We propose that accumulation over time or parallel pathways may compensate specifically for any deficit in monomethylation of K79 in the absence of H2B ubiquitylation in vivo.

Structure activity analysis of uH2B nucleosomes

A myriad of small chemical modifications to histones, such as acetylation, phosphorylation, and methylation, have evolved to control accessibility of genomic loci and to tune the recruitment and activity of effectors of genomic expression (1, 4). Given these established mechanisms, why would nature select such a large modification in ubiquitin, when a smaller modification might suffice? To delve into this question, we performed a structure activity relationship analysis of the role of ubiquitin in Dot1L stimulation.

We first wondered whether the ubiquitin fold was essential to stimulate methyltransferase activity. Could the mere acylation of H2BK120 be sufficient? We synthesized a truncated uH2B (tr-uH2B) lacking the globular domain of ubiquitin (Figure 4, panels a and b). First, peptide 1c was synthesized corresponding to residues 117–125 of H2BA117C, with residues 72–76 of ubiquitin attached to the ε-NH₂ of K120 (Supplementary Figure 7). Similar to peptide 1b, orthogonal protection of K120 allowed the C-terminal residues of ubiquitin to be installed through an isopeptide bond with K120. The N-terminus of the ubiquitin branch was acetylated to prevent interference during EPL. Ligation of 1c to H2B(1-116)-α-thioester, 5, followed by Raney nickel-mediated desulfurization afforded tr-uH2B (Supplementary Figure 8). Nucleosomes assembled with 147 bp of 601 sequence containing tr-uH2B were assayed against full-length Dot1L. However, the presence of tr-uH2B did not lead to an increase in methyltransferase activity over that observed for unmodified nucleosomes (Figure 4, panel d), suggesting that the ubiquitin fold is necessary for Dot1L stimulation.

Figure 4. Structure activity relationship analysis of uH2B.

a, Schematic of semisynthetic uH2B structural variants. Top panel: H2B bearing the terminal five residues of ubiquitin attached to K120 (tr-uH2B); middle panel: H2B modified with HA-Smt3 with A117C/G98C mutations (sH2B(cys)); bottom panel: uH2BG76A with L8A/I44A double mutation (mut-uH2B). b, Ubiquitin structure (1UBQ) (48) represented by superimposed ribbon and mesh diagrams. The surface of the five residues of tr-uH2B are shown in red. Surfaces of L8 and I44 mutated to alanines in mut-uH2B are shown in blue. c, Structural alignment of ubiquitin (1UBQ) and Smt3 (1EUV) (37) shown in ribbon diagram. Structures and alignment rendered with PyMol. d, Dot1L methyltransferase assay on uH2B structural variants. Nucleosomes containing uH2B(cys), 6, sH2B(cys), H2B A117C, or tr-uH2B methylated with ³H SAM were separated on native gels and stained with ethidium bromide (middle panel) prior to probing for ³H methyl incorporation by fluorography (top panel). Quantification of methyltransferase activity was performed by filter-binding followed by liquid scintillation counting (bottom panel). e, Dot1L assay as in d, comparing uH2BG76A and mut-uH2B. Error bars represent one s.d. (n = 3–4).

We next asked whether a specific surface of ubiquitin was required, or whether a protein of similar size and shape could substitute. We replaced ubiquitin with Smt3, the yeast homolog of human Small Ubiquitin-like Modifier 1 (SUMO1) (Figure 4, panel a). Smt3 has a similar overall fold to that of ubiquitin (Figure 4, panel c), but only a 17% sequence identity (37). An HA-Smt3(2-97)-α-thioester (Supplementary Figure 3) was substituted for the ubiquitin thioester, 2, in the semisynthetic strategy. Steps xv-xvii proceeded smoothly (Supplementary Figure 8), but no protein was recovered after incubation with Raney nickel. Therefore, the pre-desulfurization product, sH2B(cys), bearing an A117C mutation in H2B and a G98C mutation in Smt3, was incorporated into nucleosomes. Unlike nucleosomes containing the non-desulfurized uH2B(cys), 6, sH2B(cys) nucleosomes failed to stimulate Dot1L to significant levels (Figure 4, panel d). This result suggests that a surface of ubiquitin is recognized specifically.

We turned our attention to the interaction of ubiquitin with ubiquitin binding proteins. In almost all cases reported, ubiquitin interactions involve a hydrophobic patch including leucine 8 (L8) and isoleucine 44 (I44) (Figure 3, panel b) (24). Mutation of one or both residues to alanine typically abrogates binding (38–40). We generated mut-uH2B containing the triple alanine mutant L8A/I44A/G76A by replacing thioester 2, with the corresponding thioester bearing the L8A/I44A double mutation (Supplementary Figure 3) and proceeding with the two EPL reactions and desulfurization (Figure 3, panel a and Supplementary Figure 8). Dot1L was able to methylate nucleosomes bearing mut-uH2B to a similar degree as those bearing uH2BG76A (Figure 4, panel e). Therefore, a non-canonical interaction surface of ubiquitin seems to be involved in stimulation of Dot1L, likely by binding to the nucleosome, Dot1L, or both. This raises the possibility that multiple surfaces of ubiquitin are interpreted simultaneously by parallel pathways, thereby, necessitating the recognition of non-canonical surfaces. This could explain the evolutionary need for such a large modification. Further investigation of the role of the surfaces of uH2B in H3K79 methylation and its myriad of other downstream processes will put this hypothesis to the test.

If ubiquitin binds Dot1L leading to allosteric activation, ubiquitin added in trans might be able to increase methylation of unmodified nucleosomes. However, at 1 mM exogenous ubiquitin, no stimulation of Dot1L was observed (Supplementary Figure 9). In contrast, if ubiquitin binds to the nucleosome, mutations on the nucleosomal surface surrounding the site of ubiquitin attachment might prevent stimulation of Dot1L activity by decoupling ubiquitylation with changes to the nucleosomal structure. Previously, alanine scanning of the yeast nucleosome revealed two mutants in the acidic patch of H2A that prevent H3K4 methylation, which is also directly stimulated by uH2B (16), without an observable effect on ubiquitylation of H2B (41). We reasoned that if H3K4 methylation is coupled to uH2B through a similar mechanism as H3K79 methylation, these mutations might disrupt Dot1L stimulation in our system. We prepared nucleosomes containing unmodified H2B or uH2BG76A in combination with each of the mutations, H2AE64A and H2AN68A (Figure 5, panel a and Supplementary Figure 6). Neither mutation altered Dot1L stimulation by uH2B (Figure 5, panel b), indicating divergent roles for uH2B in methylation of H3K4 and K79.

Figure 5. Dot1L assays on mutant nucleosomes.

a, Surface representation of nucleosome structure (1KX5) (49) with H2A, H2B, H3, and H4 shown in green, blue, red, and orange, respectively. Amino acids in each histone are emphasized by darker shades. b, Dot1L methyltransferase assay on mutant nucleosomes containing H2B or uH2BG76A. Nucleosomes methylated with ³H SAM were separated on native gels and stained with ethidium bromide (middle panel) prior to probing for ³H methyl incorporation by fluorography (top panel). Quantification of methyltransferase activity was performed by filter-binding followed by liquid scintillation counting (bottom panel). Error bars represent one s. d. (n = 3). c, Representation of liquid scintillation counting data in c, illustrating the effect of the H4R17/19A on Dot1L-mediated methylation of unmodified nucleosomes

The basic patch of H4 is critical to the action of yeast Dot1 in telomeric silencing (42). Simultaneous mutation of two arginines in H4 at positions 17 and 19 to alanines (R17/19A) nearly abolishes H3K79 methylation in yeast (43). We wondered if the same pattern would hold true for human Dot1L and if ubiquitylation could overcome any deficit in methylation due to the mutations. Nucleosomes were prepared containing H4R17/19A (Figure 5 and Supplementary Figure 6) and either unmodified H2B or uH2BG76A. The H4R17/19A mutation decreased Dot1L activity on unmodified nucleosomes (Figure 5, panel c). A similar decrease was observed with nucleosomes containing uH2BG76A (Figure 5, panel b). Therefore, while conserved from yeast to humans, mutation of the basic patch seems to disrupt Dot1 activity independent of ubiquitylation.

One uH2B stimulates methylation of one H3K79

To date, Dot1L activity has only been studied in the context of homogenously ubiquitylated nucleosomes, where each copy of H2B is ubiquitylated. Given a 1% prevalence of uH2B in vivo (5), statistically, it is likely that heterogeneously ubiquitylated nucleosomes are predominant. This raises the question: can a nucleosome bearing one ubiquitylated H2B stimulate Dot1L? If so, will the methyltransferase activity be confined to one copy of H3, or is one uH2B sufficient to drive methylation of H3 on both sides of the nucleosome? To address these questions, we reconstituted histone octamers, varying the ratio of H2B to uH2BG76A. Electrophoretic analysis of the purified octamers confirmed incorporation of appropriate levels of unmodified and ubiquitylated H2B (Figure 6, panel a). Separation of nucleosomes formed with each octamer sample by native gel electrophoresis verified independent assortment of the two forms of H2B, with singly ubiquitylated nucleosomes exhibiting an intermediate electrophoretic mobility when compared to doubly ubiquitylated and unmodified nucleosomes (Figure 6, panel b). Methyltransferase assays showed a linear relationship between the level of uH2BG76A in a nucleosome and the activity of Dot1L, suggesting that one ubiquitylation stimulates methylation of only one H3K79 (Figure 6, panels b and c). However, whether uH2B stimulates methylation of the same nucleosomal surface or the opposite surface cannot be determined from this set of experiments. As further validation of this result, a pool of nucleosomes reconstituted from octamers containing an equal mixture of unmodified and ubiquitylated H2B, have two singly ubiquitylated nucleosomes for every one doubly ubiquitylated nucleosome; yet, half of the methyltransferase activity observed by fluorography occurred on the doubly ubiquitylated nucleosomes.

Figure 6. Investigation of ubiquitylation/methylation stoichiometry.

a, Octamers formed with varying ratios or uH2BG76A and H2B were separated on a 15% SDS-polyacrylamide gel and stained with Coomassie. b, Nucleosomes formed with octamers in a, were methylated with Dot1L and ³H SAM. Assay samples were separated by native electrophoresis and stained with ethidium bromide (bottom panel) followed by probing for ³H methyl incorporation by fluorography (top panel). c, Quantification of assay samples by filter-binding followed by liquid scintillation counting, plotted relative to the fully ubiquitylated sample. The blue line (rel. CPM = (f_ub); where f_ub is the fraction of uH2BG76A) represents the case where each uH2BG76A stimulates methylation of only one H3K79 side chain in the same nucleosome. The red line (rel. CPM = −(f_ub)² + 2(f_ub)) represents a case where each uH2BG76A stimulates methylation of both H3K79 side chains in the same nucleosome. Error bars represent one s.d. (n = 4).

Conclusion

In summary, we have developed a robust, scalable semisynthesis of ubiquitylated proteins to overcome shortcomings of our previously reported strategy (17). This methodology allowed us to perform detailed kinetic and structure activity relationship analyses of the stimulation of Dot1L-mediated methylation of H3K79 by ubiquitylated H2B. These studies suggest that a non-canonical interaction surface of the ubiquitin fold leads to activation of the chemical step of Dot1L-mediated H3K79 monomethylation. Moreover, titration of the level of uH2B within a nucleosome allowed the demonstration of a 1:1 stimulation of H3K79 methylation by ubiquitylation. As both modifications are integral to regulation of expression and maintenance of integrity of the genome (44), a full illumination of their mechanisms is key to understanding their roles in development and disease.

This greatly improved semisynthetic strategy will also allow the facile and rapid preparation of other ubiquitylated histones, and more broadly, other ubiquitylated proteins. This is especially important in systems where the natural abundance of ubiquitylated proteins is low or where ubiquitin conjugating enzymes are unknown or unavailable. Additionally, as we have illustrated here, chemical control of both ubiquitin and the target protein permits detailed analysis of the mechanisms underlying signaling through monoubiquitylation.

METHODS

Peptide synthesis

Peptide 1a was synthesized as previously described (17). For peptide 1b, the sequence corresponding to residues 117–125 of Xenopus H2B with an A117C substitution was synthesized on pre-loaded Wang resin using automatic solid-phase peptide synthesis with a 9-fluorenylmethoxycarbonyl (Fmoc) N^α protection strategy and using 2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU) for amino acid activation. Standard ^tbutyl side-chain protection was used throughout with the following exceptions; the ε-amino group of K120 was protected with the 4-methyltrityl (Mtt) group, and the thiol group of C117 was installed as an N^α-(^tbutoxycarbonyl)-thiazolidine (Boc-thiazolidine). The Mtt group on K120 was deprotected by successive incubations of the peptidyl-resin with 1% TFA in DCM containing 1% triisopropylsilane (TIS) for 10 min intervals, until no yellow color evolved. Boc-S-trityl-cysteine was coupled to the ε-NH₂ of K120. Following cleavage from the resin with TFA:TIS:H₂O (95:2.5:2.5) for 3 h, peptide 1b was purified by RP-HPLC on a process scale using a 6–18% B gradient over 60 min (A: 0.1% trifluoroacetic acid (TFA) in water; B: 90% acetronitrile, 0.1% TFA in water), yielding 56 mg peptide from 0.25 mmol resin. Peptide 1b was characterized by ESI-MS [(M+H)⁺ observed: 1,115.9 Da; expected: 1,115.3 Da].

Expressed protein ligation reaction 1

The ligation reaction between peptide, 1b, and ubiquitin(1-75)-MES, 2, was performed using conditions similar to those previously optimized for the corresponding auxiliary-mediated reaction (17). In a typical reaction, purified peptide 1b, (19.9 mg, 17.9 μmol) and protein 2 (45.7 mg, 5.3 μmol) were combined in 2.5 mL of ligation buffer (6 M guanidinium chloride, 300 mM sodium phosphate, 100 mM MESNa, 50 mM tris(2-carboxyethyl)phosphine (TCEP), pH 7.8). The pH was adjusted to 7.8 using 5 N NaOH and the reaction was allowed to proceed for 4 h at room temperature. After 2 h, reaction completion was verified by RP-HPLC and ESI-MS. [(M+H)⁺ observed = 9,609 ± 2 Da; (M+H)⁺ expected = 9,606 Da]

Cysteine deprotection

To crude ligation product 3 was added 2.5 mL 50% HPLC buffer B and 0.72 mL 4 M methoxylamine, to reach a final concentration of 0.5 M. The pH of the resulting solution was increased to 5 with 5 N NaOH. Following incubation for 12 h at room temperature to allow complete deprotection of the cysteine, product, 4, was purified by preparative RP-HPLC using a 25–55% B gradient over 45 min, yielding 27.3 mg purified protein. The identity of the purified protein 4 was verified by ESI-MS. [(M+H)⁺ observed = 9,593 ± 2 Da; (M+H)⁺ expected = 9,594 Da]

Expressed Protein Ligation reaction 2

In a typical reaction, protein 4 (29.9 mg, 3.1 μmol), and H2B(1-116)-MES, 5 (52.6 mg, 4.0 μmol), were dissolved in ligation buffer to a final volume of 1.6 mL. The pH of the resulting solution was increased to 7.8 with 5 N NaOH and the reaction was allowed to proceed for 48 h. Fresh TCEP was then added to a final concentration of 50 mM. The ligation product, 6, was purified using semi-preparative RP-HPLC with a 35–52% B gradient over 45 min, yielding 38.4 mg of protein 6. The identity of the ligation product was verified by ESI-MS. [(M+H)⁺ observed = 22,444 ± 5 Da; (M+H)⁺ expected = 22,443 Da]

Desulfurization

The 38.4 mg of protein 6 was converted to 30 mg uH2BG76A, 7, using using two complementary desulfurization methodologies. Raney nickel reduction was used to convert C117 and C76 of protein 6 to alanines (17, 22, 45). In a typical reaction, protein 6, (8.0 mg, 0.36 μmol) was dissolved in 3 mL desulfurization buffer 1 (6 M guanidinium chloride, 200 mM sodium phosphate, 35 mM TCEP, pH 7.0). Raney nickel was prepared as previously described (17) and added at 0 h, 6.5 h, 10 h, and 13.5 h. The reaction progress was followed by RP-HPLC and ESI-MS. After 22 h, the Raney nickel was pelleted by centrifugation and washed. The reaction supernatant and washes were combined, added to an equivalent volume of 50% HPLC buffer B, and purified using semi-preparative RP-HPLC with a 42–52% B gradient over 45 min, yielding 4.0 mg of uH2BG76A, 7. Protein 7 was characterized by ESI-MS.

Protein 6 was also desulfurized by the previously reported radical-initiated methodology (23, 46). In a typical reaction, protein 6, (9.0 mg, 0.40 μmol) was dissolved in 500 μL desulfurization buffer 2 (6 M guanidinium chloride, 200 mM sodium phosphate, pH 7.0). To this solution was added 10 μl ethane thiol, 750 μl 0.5 M TCEP (dissolved in desulfurization buffer followed by pH adjustment to 7.0), 50 μl 2-methylpropane thiol, and 12.5 μl 0.2 M VA-061 (Wako Chemicals) (dissolved in methanol), and the mixture was incubated in a 37 °C water bath. The reaction progress was followed by RP-HPLC and ESI-MS. After 24 h, the product was purified using semi-preparative RP-HPLC with a 42–52% B gradient over 45 min, yielding 7.6 mg of uH2BG76A, 7. Protein 7 was characterized by ESI-MS. [(M+H)⁺ observed = 22,380 ± 4 Da; (M+H)⁺ expected = 22,379 Da]

Methyltransferase assays

Full-length Dot1L and Dot1L_cat were prepared as previously described (17). Methyltransferase assays with Dot1L were performed as previously described (17). Ubiquitin (Sigma) was added as indicated. Methyltransferase assays for steady-state measurements were performed as follows: Serial two-fold dilutions of nucleosomes from 7.5 μM to 230 nM were preformed in methyltransferase assay buffer (20 mM Tris HCl, 4 mM EDTA, 1 mM DTT, pH 7.9) supplemented with 1.6 μM ³H SAM at 4 °C. Each solution was aliquoted into individual PCR tubes. The reaction was initiated by the addition of Dot1L_cat (2 nM, 4 nM, and 6 nM) and subsequent transfer to a 30 °C water bath. For nucleosomes containing uH2BG76A, 5 μL assay samples were quenched at 3 min by the addition of 10 μL 0.2 % TFA and spotted on filter paper for liquid scintillation counting as previously described (17). For unmodified nucleosomes, 10 μL assay samples were quenched at 15 min and quantified as above. Each concentration of nucleosome was assayed against each concentration of Dot1L_cat in duplicate. Counts from reactions containing no Dot1L_cat were subtracted from each sample. Conversion of CPM to moles of methylation was performed using LC-MS with isoptopically-labelled synthetic standards. Unmodified and monomethylated peptides containing residues 73–83 of H3 and a d₅ isotopically-labelled Phe78 (Fmoc-Phe(d5)-OH from Cambridge Isotopes) were added into gel slices containing assay samples. Propionylation, trypsin digestion, and propionylation were performed as previously described (47). Relative quantities were determined by standardizing spectral counts of modified tryptic peptides to synthetic isotopically-labelled peptides following LC-MS/MS using a Dionex U3000 capillary/nano-HPLC system (Dionex) directly interfaced with the Thermo-Fisher LTQ-Orbitrap mass spectrometer (Thermo Fisher). In a typical experiment, a conversion factor (F) of 2.15*10⁻¹⁶ mol/CPM was calculated. Values of v_o/E_T were calculated using equation (1),

$$\frac{v_0}{{\text{E}}_{\text{T}}}=\frac{F({\text{CPM}}_{\mathit{obs}}-{\text{CPM}}_{\mathit{background}})}{t\ast {\text{E}}_{\text{T}}}$$ (1)

where v_o is the initial velocity, E_T is the molar quantity of Dot1L_cat in the volume quantified, and t is time in seconds. Kinetic parameters for the methylation of ubiquitylated nucleosomes were obtained by fitting to the non-linear Michaelis-Menten equation (2) using profit 6.1.1 with the Levenberg-Marquardt algorithm:

$$\frac{v_0}{{\text{E}}_{\text{T}}}=\frac{k_{\mathit{cat}}[\text{S}]}{K_m+[\text{S}]}$$ (2)

where [S] is the concentration of nucleosome. Time courses were fit to linear regression models to verify steady-state conditions.

Assays performed with an excess of Dot1L_cat were set up as follows: Nucleosome and Dot1L_cat were combined in methyltransferase assay buffer supplemented with 1 μM ³H SAM and 4 μM cold SAM, at 250 nM and 5 μM concentrations, respectively. The reaction was initiated by transferring to a 30 °C water bath. Samples were removed, quenched, and counted as described above. The observed CPM value at 0 sec was subtracted from the observed CPM value after 15, 30, 60, 120, 300, 600, and 6,000 sec. The corrected value was multiplied by 5 to account for dilution with cold SAM and divided by the number of microliters quantified. The appropriate conversion factor (F) was applied to calculate methyltransferase activity at each time point. The resultant curves were fit to a single exponential model (3) using profit 6.1.1.

$$\text{Activity}=a{e}^{-bt}+c$$ (3)

For mass spectrometric analysis of methylation levels, assays were submitted to trypsin digestion as previously described (17). Tryptic peptides were analyzed by LC-MS as described above.

Asymmetrically ubiquitylated nucleosomes

Octamers were prepared as described previously described (17) using mixtures of uH2BG76A and H2B. Incorporation of the appropriate levels of ubiquitylated and unmodified H2B was verified by separation on a Criterion 15% Tris HCl gel, followed by Coomassie staining. Nucleosomes were reconstituted on a small scale as previously described (17) and analyzed on a Criterion 5% TBE gel. Each nucleosome sample was subjected to methyltransferase assays as described above. Results, relative to homogenously ubiquitylated nucleosomes, were compared to theoretical values for each model tested, assuming negligible methylation of unmodified nuclesomes. If each uH2BG76A stimulates methylation of one H3K79 within the nucleosome, the simple linear model (4) is expected.

$$\text{Rel}.\text{CPM}={\text{f}}_{\text{ub}}$$ (4)

where f_ub is fraction of H2B that is ubiquitylated. Assuming independent assortment of uH2BG76A and H2B, if each uH2BG76A stimulates methylation of both H3K79 within the nucleosome, the non-linear model (5) is expected.

$$\text{Rel}.\text{CPM}={{\text{f}}_{\text{ub}}}^2+2({\text{f}}_{\text{ub}})(1-{\text{f}}_{\text{ub}})$$ (5)

which simplifies to equation (6).

$$\text{Rel}.\text{CPM}=-{{\text{f}}_{\text{ub}}}^2+2{\text{f}}_{\text{ub}}$$ (6)