Two DOT1 enzymes cooperatively mediate efficient ubiquitin-independent histone H3 lysine 76 tri-methylation in kinetoplastids

In higher eukaryotes, a single DOT1 histone H3 lysine 79 (H3K79) methyltransferase processively produces H3K79me2/me3 through histone H2B mono-ubiquitin interaction, while the kinetoplastid Trypanosoma brucei di-methyltransferase DOT1A and tri-methyltransferase DOT1B efficiently methylate the homologous H3K76 without H2B mono-ubiquitination. Based on structural and biochemical analyses of DOT1A, we identify key residues in the methyltransferase motifs VI and X for efficient ubiquitin-independent H3K76 methylation in kinetoplastids. Substitution of a basic to an acidic residue within motif VI (Gx6K) is essential to stabilize the DOT1A enzyme-substrate complex, while substitution of the motif X sequence VYGE by CAKS renders a rigid active-site loop flexible, implying a distinct mechanism of substrate recognition. We further reveal distinct methylation kinetics and substrate preferences of DOT1A (H3K76me0) and DOT1B (DOT1A products H3K76me1/me2) in vitro, determined by a Ser and Ala residue within motif IV, respectively, enabling DOT1A and DOT1B to mediate efficient H3K76 tri-methylation non-processively but cooperatively, and suggesting why kinetoplastids have evolved two DOT1 enzymes.

2. Introduction, the end of first paragraph, besides of independent of H2B ubiquitination, are H3K76me1/2/3 in Trypanosoma brucei known to associate with any other histone modifications (H3K4me3, H4K16 acetylation)?3. page 5, line 92, change "lack of cross-talk with ubiquitination" to "lack of cross-talk with H2B ubiquitination".Also, when the lysine is substituted to histidine, which is an amino acid that could also be modified post-translationally. 4. page 5, line 104, expand/explain the last sentence on why me3 generated by DOT1B cannot compensate for DOT1A defects in vivo.Does this suggest additional function of DOT1A in addition to its enzymatic activity? 5. page 7, lines 160-161, explain how the radioactive assay can tell DOT1A has a di-methyltransferase activity.6. Discussion section contains results that should be moved to the Results section, e.g., DOT1A Ser218/DOT1B Ala196 swap mutations and DOT1L Asn241.Throughout the manuscript, the authors almost always have DOT1L as a comparison, which is good, but also it created difficulty to know which data panels are for DOT1L.Please consider organize all DNOT1L data into a single figure.
6. Figure 3i, the labels are unclear: the WT and deltaN refer to nucleosome and H4 tail-less deletion.However, the same WT and deltaN could refer to the enzyme and N-terminal deletion.For the No enzyme control, what is the substrate present?Please go through every panel, make sure the labels are clear and meaningful enough without the need to read the text or legends.
Figure 4a, the motif alignment panels are truncated.
Figure 4c, some of the data points seem to have a single dot, no repeats?Figure 5a, again, the sequence alignment is truncated, and what are the zinc coordinates?Is the same zinc atom bound by the N-terminal beta hairpin? Figure 5b and 5e, the activity for the "WT" has a wide variation in panel e, this is also true in other figure panels.Please explain why.
Figure 6c, what is the meaning of the protection ratio?What are the numbers for the complete protection or no protection?
Reviewer #2 (Remarks to the Author): The manuscript "Two DOT1 enzymes cooperatively 1 mediate efficient ubiquitin2 independent histone H3 lysine 76 tri-methylation in kinetoplastids" by Frisbie et al. addresses the mechanistic features of Dot1a and Dot1b's methyltransferase activity and substrate specificity.The authors show by kinetic analysis that Dot1a and Dot1b are both need to efficiently generate H3K79 trimethylation on in vitro substrates.In their investigation, they also show that unlike Dot1L and Dot1 from S. cerevisiae Dot1a and Dot1b do not require H2B ubiquitination or H4 tail for H3K79 methylation to occur which is unique.Interestingly, they do go on to show that and acidic patch-arginine anchor interactions are used for substrate recognition for Dot1 methyltransferases.Although the in vitro data is interesting and well done, it would be interesting to tie some of their observations to new biological impact or insight.
An important aspect not fully addressed by the authors is why kinetoplastids developed two distinct Dot1 enzymes with differing substrate preference and product specificity?Some additional insight into this would be helpful, impactful and interesting.For example, the authors could do some detail protein phylogenetic analysis across many organisms/species to see if additional evolutionary insight can be obtained to help add biological reasoning behind their observed in vitro findings.
Additional issues: Fig. 2b The authors should evaluate if there is any statistical difference between FL-Dot1A and the Delta42-Dot1A and Delta27-DOT1B.In addition, Fig 2b looks like the same data as shown in Figure 4C first three bar graphs.If so, the authors need to mention this either in the text for figure legend.Fig 2c .The authors show that FL-DOT1A binds to DNA but show that it isn't required for methyltransferase (Fig. 2b).However, Fig 2b might suggest that DNA binding and/or the N-terminus (1-42) is inhibitory for activity.The authors could determine if this is the case.In addition, the authors could also investigate if the FL-DOT1A binds to RNA or RNA/DNA hybrids and if this alters chromatin association or activity.Thus, RNA binding or recognition of R-loops may also contribute to it associate to chromatin or function as it does other chromatin associated factors.I don't understand why the results stopped at figure 6 when there clearly are 7 figures?Or Why Figure 7 seemed to be part of the discussion?Was this intentional?Because of this, the discussion seemed disjointed where the result section is placed in the middle of the discussion.In addition, the histone H3V section needs further development and a bit strange to end on this note.For example, the authors should talk more about H3V and is it similar to other H3 variants in other species, etc.Furthermore, additional experimentation could make this section more relevant and interesting for the manuscript.

Re: Nature Communica0ons manuscript NCOMMS-23-36050
We would like to thank the two referees for their posi0ve comments about our work and appreciate their construc0ve sugges0ons for improving our manuscript.We have addressed all raised issues by carefully revising the text and the figures.As requested by both reviewers, we have moved Figure 7 to the Results part, and we have substan0ally expanded the Discussion in response to the reviewers' comments and ques0ons to increase the biological insight and impact of our findings.Our detailed response to the reviewers' comments is outlined below and highlighted in red.

Point-by-point response:
Reviewer #1 (Remarks to the Author): The manuscript by Frisbie and Hashimoto et al. well characterized two related histone H3 lysine methyltransferases from Trypanosoma brucei (Tb).These enzymes are eukaryo0c homologs of yeast Dot1p and human DOT1L.However, Tb DOT1A and DOT1B are shorter in protein size (< 300 residues), and unlike Dot1p/DOT1L, their enzyma0c ac0vi0es are independent of H2B ubiqui0na0on.
In the study, the authors used elegant quan0ta0ve mass spectrometry to monitor TbDOT1 ac0vi0es on nucleosome substrates.They discovered that TbDOT1A prefers H3K76me0 as substrate, and generate me1/me2 products, while TbDOT1B prefers me1 as substrate and generates me2/me3 products.They quan0ta0vely measured kine0c rate for each reac0on step.
Next, they determine two crystal structures of TbDOT1A in complex with SAH, and found that the N-terminal domain folds into a Zn-coordina0ng beta-hairpin with extensive interac0ons with the methyltransferase domain.The structure allowed the authors to model a nucleosome-bound conforma0on and, using a series of muta0onal and biophysical method of X-ray footprin0ng, they found that (1) histone H4 N-terminal tail is not required for TbDOT1 ac0vity, (2) TbDOT1 enzymes use unique mo0f X residues for ac0vity, (3) DOT1-nucleosome interac0ng residues includes a conserved Arg finger for interac0on with an acidic patch of nucleosome, (4) a unique K-to-E/D subs0tu0on in mo0f VI enhanced TbDOT1-nucleosome interac0on, and (5) important residues in TbDOT1A and 1B for substrate specificity are iden0fied.
Along the way, the authors also carried out parallel study on human DOT1L for comparison.
Overall, this is a very completed study.Some minor sugges0ons are below.
We thank the reviewer for the posi0ve and cri0cal feedback.We have addressed and incorporated the sugges0ons into our revised manuscript.
1. Abstract, the last sentence, "… coopera0ve ac0on of two DOT1 enzymes in vivo" -are there any data showing these two enzymes have similar expression paherns in vivo?While DOT1B mRNA is about double the amount of DOT1A mRNA, both mRNAs are present at roughly constant levels throughout the cell cycle with a slight increase at the G2 phase.However, DOT1A/B protein abundance is currently not known due to the lack of specific an0bodies, and it is unknown whether DOT1A and DOT1B are ac0ve throughout the cell cycle.Indeed, the existence of a regulatory cell cycle-dependent DOT1A/B methyltransferase ac0vity mechanism has been proposed based on the observa0on that the DOT1A/B products H3K76me1 and me2 are only detected in G2/M but not in S phase (Janzen et al., Mol Cell 2006 PMID: 16916638, and Gassen et al., NAR 2012 PMID: 22941659).In the revised Discussion, we have included a sec0on on DOT1A/B abundance and men0on the recently published RNase H2-DOT1A/B interac0on that may regulate DOT1A/B ac0vity (see also response to point 4 below for more details).and Siegel et al., Genes Dev.2009 PMID: 19369410).So far, no modifica0on has been reported or suggested to colocalize or be excluded with H3K76me1/2/3.In the revised manuscript, we have added this informa0on in the Introduc0on to address T. brucei H3K76 methyla0on and other epigene0c modifica0ons.

Done.
Also, when the lysine is subs0tuted to his0dine, which is an amino acid that could also be modified post-transla0onally.
Previous analyses of histone PTMs in T. brucei (Janzen et al., FEBS Le@er 2006 PMID: 16580668, Zhang et al., iScience.2020 PMID: 32403088) have not iden0fied any modifica0on of H2B His108 (corresponding to human H2BK120).Unlike most eukaryote histone sequences, which are highly conserved, the T. brucei histone sequences have highly diverged.The T. brucei histone H2A amino acid sequence shares only 49%, H2B 41%, H3 56%, and H4 60% iden0ty with human histones.As such, histone modifica0ons are quite different from histones of most eukaryotes (Janzen et al., ).Therefore, it is impossible at present to assess how the Lys-to-His amino acid subs0tu0on in the T. brucei H2B sequence impacts its modifica0on and its poten0al cross-talk with other poshransla0onal modifica0ons.
In the revised manuscript, we have emphasized histone sequence divergence and expand on T. brucei epigene0c histone modifica0ons in the Introduc0on and Discussion.4. page 5, line 104, expand/explain the last sentence on why me3 generated by DOT1B cannot compensate for DOT1A defects in vivo. [redacted] This conundrum has been discussed since the discovery of DOT1A and DOT1B almost 20 years ago (Janzen et al., Mol Cell 2006 PMID: 16916638).Based on our DOT1A/B kine0cs data, we can now provide an explana0on for this finding.In short, in vivo data suggest that H3K76me1/me2 is important for cell cycle progression.Our analysis confirmed that DOT1B can methylate H3K76 from me0 to me3, but importantly revealed that me1 and me2 intermediates do not accumulate.Thus, if DOT1B can generate H3K76me3 from me0 in vivo, DOT1A knock-down cells would not accumulate H3K76me1/me2 and lead to cell cycle defects, even if DOT1B was ac0ve.Our in vitro data show that combina0on of DOT1A and DOT1B -concomitantly or sequen0ally -yields significant amounts of H3K76me1 and me2 and accelerates H3K76me3 genera0on by up to eight 0mes.We have included this explana0on in the revised Discussion.Does this suggest addi0onal func0on of DOT1A in addi0on to its enzyma0c ac0vity?
The DOT1A overexpression phenotype, in which cells contain >4n DNA content, is indeed dependent on methyltransferase ac0vity, as overexpression of an AdoMet binding residue mutant (G138R) did not cause such a phenotype.By contrast, it is not known if the DOT1A knockdown, in which cells contain a 1n DNA content, is dependent on enzyma0c ac0vity (Gassen et.al, 2012 Nuc.Acids Res.PMID: 22941659).
Recently, both DOT1A/B were found to form a complex with RNase H2 and PCNA in vivo (Eisenhuth et al., mBio 2021 PMID: 34749530 and Staneva et al., Genome Res.2021 PMID: 34407985), sugges0ng a link of DOT1A/B with the replica0on machinery, as RNase H2 and PCNA are found at DNA replica0on foci in S phase.However, the DOT1A/B products H3K76me1/me2 are not detected in S phase, only in G2/M phase (Gassen et al., NAR 2012 PMID: 22941659).These findings indicate that RNase H2 may suppress or inac0vate DOT1A/B H3K76 methyltransferase ac0vity, which we confirmed in vitro (unpublished results).While further biochemical studies are required to address how DOT1A/B and RNase H2 interac0on impact DOT1 methyltransferase ac0vity and RNase H2 ac0vity and how DOT1A-RNAi impacts DOT1A-RNase H2-PCNA complex and cell cycle progression, these studies are beyond the scope of this manuscript.In the revised manuscript, we have included the DOT1A/B and RNase H2 interac0on in the discussion and discuss a poten0al regulatory role of RNase H2 for DOT1A/B ac0vity.
The radioac0ve assay can indeed not dis0nguish between mono-, di-, or tri-methyltransferase ac0vity, and we have fixed this in the revised manuscript.
6. Discussion sec0on contains results that should be moved to the Results sec0on, e.g., DOT1A Ser218/DOT1B Ala196 swap muta0ons and DOT1L Asn241.Throughout the manuscript, the authors almost always have DOT1L as a comparison, which is good, but also it created difficulty to know which data panels are for DOT1L.Please consider organize all DOT1L data into a single figure.
We have revised the labelling in each panel for clarifica0on.
7. Figure 3i, the labels are unclear: the WT and deltaN refer to nucleosome and H4 tail-less dele0on.However, the same WT and deltaN could refer to the enzyme and N-terminal dele0on.For the No enzyme control, what is the substrate present?Please go through every panel, make sure the labels are clear and meaningful enough without the need to read the text or legends.
We have revised the labelling for clarifica0on.8. Figure 4a, the mo0f alignment panels are truncated.Fixed.9. Figure 4c, some of the data points seem to have a single dot, no repeats?
We have done the repeat experiments, which further confirmed the results.Figure 4c with added sta0s0cal analysis was updated.10. Figure 5a, again, the sequence alignment is truncated, and what are the zinc coordinates?Is the same zinc atom bound by the N-terminal beta hairpin?
We have updated the figure to clarify the Zn-coordina0ng cysteine residues.11. Figure 5b and 5e, the ac0vity for the "WT" has a wide varia0on in panel e, this is also true in other figure panels.Please explain why.
The rela0vely wide varia0on in Figure 5b and 5e (radioac0vity-based assay) with respect to the smaller varia0on of the mass spec-based methyltransferase data can be explained by the different condi0ons of these two assays.In the radioac0vity-based assay, we used 5.6 μM 3 H-AdoMet, 0.125 μM enzyme and 1 μM nucleosome, which corresponds to 2 µM of H3K76 residues.The total poten0al substrate amount in this assay is 4 μM (me0 and me1) for the di-methyltransferase DOT1A, and 6 μM (me0, me1, and me2) for the tri-methyltransferase DOT1B in a 20 μl reac0on volume.Because the 3 H-AdoMet concentra0on 5.6 μM is around or below the KM value, not all of the substrate can be consumed, unlike in the MS-based assay, which contains 160 µM AdoMet.In addi0on, product inhibi0on by AdoHcy entails that not all 3 H-AdoMet is used.By conver0ng dpm to nanomolar concentra0on of transferred methyl groups in our 20 μL reac0on condi0on, 533 dpm is equivalent to 1 nM of AdoMet.Therefore, 400k dpm is equivalent to 0.8 μM methyl groups being transferred from AdoMet to the nucleosome substrate in our 20 μL assay condi0on.While the absolute dpm numbers seem within a wide range (200-500k dpm), the concentra0on range of methyl groups transferred is only ~0.4-0.9 μΜ.This analysis shows that the radioac0vitybased assay was conducted under co-factor-limited condi0ons and merely serves to assess if an enzyme is ac0ve or not.If ac0ve, the enzyme was subjected to the more quan0ta0ve mass specbased methyltransferase assay for further characteriza0on, where co-factor was not a limi0ng factor.We have pointed this out in the revised Methods sec0on.
12. Figure 6c, what is the meaning of the protec0on ra0o?What are the numbers for the complete protec0on or no protec0on?
The protec0on ra0os for each pep0de are calculated as the frac0on of modifica0on rate Kfree over Kcomplex, where "free" refers to the nucleosome, and "complex" refers to the complex of DOT1A and nucleosome.Upon complex forma0on, regions in the binding interface experience considerable loss in solvent accessibility (protec0on).Thus, the protec0on ra0o measures the extent of decrease in oxida0ve modifica0on in the pep0de/s within the binding interface.A protec0on ra0o value close to 1 indicates that the solvent accessibility of the region remains unchanged (no protec0on in the complex), while a value > 1 suggests that the corresponding region exhibits protec0on from the solvent as a func0on of the complex forma0on.Although the pep0de can be completely protected from oxida0ve labeling in the complex (Kcom = 0) against the free form of the protein, this event is rarely observed in footprin0ng experiments.Typically, in footprin0ng experiments, a decrease in modifica0on rate for a specific pep0de/s is observed upon complex forma0on with high protec0on ra0os ≥ 2 and moderate protec0on ra0os ≥ 1.5 < 2. We have expanded the Methods sec0on accordingly for clarifica0on.
Reviewer #2 (Remarks to the Author): The manuscript "Two DOT1 enzymes coopera0vely 1 mediate efficient ubiqui0n2 independent histone H3 lysine 76 tri-methyla0on in kinetoplas0ds" by Frisbie et al. addresses the mechanis0c features of Dot1a and Dot1b's methyltransferase ac0vity and substrate specificity.The authors show by kine0c analysis that Dot1a and Dot1b are both need to efficiently generate H3K79 trimethyla0on on in vitro substrates.In their inves0ga0on, they also show that unlike Dot1L and Dot1 from S. cerevisiae Dot1a and Dot1b do not require H2B ubiqui0na0on or H4 tail for H3K79 methyla0on to occur which is unique.Interes0ngly, they do go on to show that and acidic patch-arginine anchor interac0ons are used for substrate recogni0on for Dot1 methyltransferases.Although the in vitro data is interes0ng and well done, it would be interes0ng to 0e some of their observa0ons to new biological impact or insight.
We thank the reviewer for the posi0ve feedback and for poin0ng out that our in vitro results would expand and accelerate in vivo studies in T. brucei.
1.An important aspect not fully addressed by the authors is why kinetoplas0ds developed two dis0nct Dot1 enzymes with differing substrate preference and product specificity?Some addi0onal insight into this would be helpful, impac{ul and interes0ng.For example, the authors could do some detail protein phylogene0c analysis across many organisms/species to see if addi0onal evolu0onary insight can be obtained to help add biological reasoning behind their observed in vitro findings.
Aravind et al. (Aravind, et al., Prog Mol Biol Transl Sci 2011 PMID: 21507350) performed a phylogene0c analysis of DOT1 and the results implied a lateral transfer of DOT1A and DOT1B from the animal lineage into mammalian trypanosome parasites.The dis0nct features of the Nterminal domain, including the Zinc-coordina0ng cysteines, were likely acquired amer this transfer, followed by gene duplica0on that generated two func0onally dis0nct paralogs.Since we fully agree with their plausible hypothesis, we have included their analysis in the Discussion of the revised manuscript.
Addi0onal issues: 2. Fig. 2b The authors should evaluate if there is any sta0s0cal difference between FL-Dot1A and the Delta42-Dot1A and Delta27-DOT1B.