The trans-histone crosstalk H2Bub → DOT1L → H3K79me: modulatory for me1, effectively required for me2/me3
also called trans-histone crosstalk · H2Bub-H3K79 crosstalk · trans-tail regulation · H2BK120ub to H3K79me
The ubiquitin tag on histone H2B does not just sit next to DOT1L's target — it physically reshapes the nucleosome so that lysine 79 of histone H3 swings out into DOT1L's active site. Without the tag DOT1L can still add one methyl group, but it largely cannot add the second and third, and it is the second and third that KMT2A-fusion leukaemia runs on.
The same thing, for a scientist
H2BK120ub (K123 in yeast) stimulates DOT1L in cis within the same nucleosome. The degree of requirement is the question this entry answers, because the graph's 35 H3K79me2 entries implicitly assume DOT1L acts autonomously. The evidence says otherwise but not absolutely: in yeast, Rad6-mediated H2B ubiquitination is required for *efficient* H3K79 methylation [E1], H3K4 and H3K79 **trimethylation** are solely dependent on H2BK123 monoubiquitination when confounders are removed [E2] (PRECISION NOTE 2026-09-20: H2BK123 is the YEAST residue and that experiment is a yeast histone-shuffle strain; the human equivalent is partial rather than absolute, and later in vivo work reports flexibility in this crosstalk), and in a 2026 genome-wide map the loss of Rad6 causes **complete loss of the H3K79me3 zone**, converting it into an H3K79me1-enriched region [E6]. So the stoichiometry is the discriminator: **me1 survives without H2Bub, me2/me3 largely do not.** Structurally, cryo-EM of DOT1L on ubiquitinated nucleosomes shows ubiquitin plus the histone H4 tail position DOT1L and induce a conformational change in the globular core of H3 that reorients K79 from an inaccessible position into the active site [E4]. Whether that conformational change is what does the work is **contested**: Zhou et al., using photocrosslinking in semisynthetic designer chromatin with the human enzyme, place the functional hotspot on ubiquitin against the N-terminus of histone H2A and describe H2B-Ub as *corralling* hDot1L into a productive orientation (PMID 26830124, counter-evidence on [E4]). Allosteric reorientation and corralling both predict stimulation and are not yet separated. In leukaemia cells this is load-bearing: structure-guided DOT1L mutants that cannot bind ubiquitin, or cannot bind the nucleosome acidic patch, fail to support proliferation of MV4;11 [E5]. The crosstalk is also **bidirectional**: Dot1 promotes H2B ubiquitination by a methyltransferase-independent mechanism [E7], which matters directly for how a catalytic DOT1L inhibitor such as pinometostat should be expected to behave.
Absolute or modulatory? The answer depends on which methylation state you mean
This beat was dispatched to settle whether the requirement is absolute. It is not a single answer, and the distinction is the useful finding.
| Methylation state | Dependence on H2Bub | Source |
|---|---|---|
| H3K79me1 | Not dependent. Survives Rad6 loss; the H3K79me3 zone is converted into an me1-enriched region when Rad6 is lost, and the pre-existing me1 zone is only diminished | 6 |
| H3K79me2 | Strongly reduced; in vitro stimulation of DOT1L by ubiquitylated H2B is direct and intranucleosomal | 13 |
| H3K79me3 | Effectively required IN YEAST. Trimethylation is solely dependent on H2BK123ub once background and epitope-tag confounders are removed; the human dependency is partial, so this row states the yeast result and not a human one; the genome-wide me3 zone disappears entirely on Rad6 loss | 26 |
The practical translation for this disease: KMT2A-fusion biology is read out almost entirely in H3K79me2/me3 (the graph's own DOT1L entry, mechanism-dot1l-h3k79-methylation, cites H3K79me2 profiling as the field's best mechanistic biomarker). Those are precisely the states that do not survive without H2Bub. So for the states that matter here, the requirement behaves as close to absolute; for total H3K79 methylation it is modulatory.
The qualifier that keeps this honest is 8: in engineered yeast, ubiquitin tethered to every nucleosome promotes H3K79 and H3K4 methylation from a proximal and a more distal site provided the orientation is correct. The exact attachment site, the native ubiquitin-lysine linkage and ubiquitination cycling are therefore not critical for the crosstalk in vivo. What DOT1L requires is a ubiquitin in roughly the right place and pose, not H2BK120 specifically — which is also why a second site, H2BK34, can substitute (see mark-h2bk34ub).
The structural mechanism, which is more interesting than 'ubiquitin recruits DOT1L'
Four independent 2019 cryo-EM studies converged on the same picture 4. What they disagree about with the earlier chemical-biology work is not that ubiquitin stimulates hDot1L but how.
- DOT1L docks on the nucleosome using the H2A/H2B acidic patch and the conjugated ubiquitin, which together fix its orientation.
- The histone H4 N-terminal tail contributes to positioning DOT1L.
- Those contacts induce a conformational change in the globular core of histone H3 that swings K79 out of its buried, inaccessible rotamer and into the DOT1L active site in a catalysis-primed position 4.
That last point is the substantive one. H3K79 lies on the nucleosome disc face and is not on a flexible tail; without the reorientation it is not a usable substrate. The ubiquitin is therefore doing conformational work on a different histone — the literal meaning of trans-histone crosstalk.
Single-molecule and bulk biophysics added the kinetic half: ubiquitin and cofactor binding drive conformational changes in DOT1L itself that promote productive nucleosome binding, rather than increasing the amount of binding 5.
The competing model, which this entry previously did not state. Zhou et al. installed a site-specific photocrosslinker on the ubiquitin of semisynthetic H2B-Ub chromatin and mapped the functional hotspot on ubiquitin to the N-terminus of histone H2A; that contact is required for stimulation and repositions hDot1L on the nucleosomal surface. Their reading is that H2B-Ub physically corrals the enzyme into a productive orientation - a restraint and repositioning model, closer to the 'adhesive' account this entry had dismissed, and obtained with the human enzyme (counter-evidence on 4). The two models are not mutually exclusive and neither has been excluded; anyone trying to drug the surface needs to know which contact carries the energy, and nobody does.
Why this is a druggable claim and not just a mechanism
Spangler et al. made structure-guided DOT1L mutants that disrupt (i) the DOT1L–ubiquitin interface and (ii) the DOT1L–acidic-patch interface, and showed both are required for cell proliferation in the MV4;11 leukaemia model 5. MV4;11 carries KMT2A::AFF1.
That is a genuinely different therapeutic proposition from the one that failed in the clinic. Pinometostat is a SAM-competitive catalytic inhibitor (mechanism-dot1l-h3k79-methylation 6 there), and its phase 1 result was a disappointment. The interfaces identified here are protein–protein/protein–nucleosome surfaces outside the catalytic site. An agent that blocked the ubiquitin-binding surface would prevent the stimulated state without competing with SAM, which is a distinct chemical problem with a distinct resistance profile. This is written up separately as interface-dot1l-h2bub-ubiquitin.
The honest caution is that 5 is a genetic proof of principle in one line, with no compound, and the surfaces in question are large and shallow.
The one measurement in a human leukaemia line that points the other way
Every result above is from yeast, from reconstituted nucleosomes, or from structural work. There is one quantitative measurement in a human leukaemia cell line, and it disagrees.
Darwanto et al. built an LC-MS/MS multiple-reaction-monitoring assay quantifying ~20 histone modification sites simultaneously and applied it to U937 human leukaemia cells and U937 derivatives overexpressing AF10 (MLLT10, a KMT2A fusion partner this graph carries as gene-mllt10) and an AF10 mutant lacking the hDot1-binding OM-LZ domain. They found H2B ubiquitination inversely correlated with H3K79 methylation, and proposed a catalytic-and-inhibitory-loop model rather than a simple positive crosstalk 9.
This entry does not treat that as refuting the crosstalk, for three stated reasons: the measurement is of global bulk levels rather than of the mark at specific loci, where all the positive evidence lives; the perturbation is AF10 overexpression, which forces DOT1L recruitment and could deplete the H2Bub pool at highly methylated loci as a consequence rather than a cause; and an inverse global correlation is compatible with a positive local dependency if the two marks turn over on different timescales.
But it is the only human-leukaemia measurement of both marks together that this beat found, it is orthogonal in method to everything else here, and it should be reported rather than buried. A reader who wants the strongest reason to doubt the chain in this graph should read 9 first.
The crosstalk runs backwards too, and that has a clinical corollary
Dot1 promotes H2B ubiquitination in a dose-dependent, methyltransferase-independent manner, through its N-terminus, and this is exacerbated by loss of SAGA-DUB activity 7.
The corollary for this disease is specific. A catalytic DOT1L inhibitor occupies the SAM site; it does not remove the DOT1L protein. If DOT1L protein sustains H2Bub independently of catalysis, then pinometostat leaves the H2Bub layer — and therefore the H3K4me2/3 arm of the crosstalk — substantially intact, while a DOT1L degrader would not. That is a testable difference between inhibition and degradation in KMT2A-rearranged cells that nobody appears to have measured, and it is recorded as an open question below.
Correction (coverage repair, 2026-09-19)
This entry originally asserted that the crosstalk mechanism is "a genuine allosteric mechanism, not a tethering effect", and that it is "allosteric rather than adhesive". Both statements have been withdrawn as overstated. The cryo-EM work 4 supports an allosteric reorientation of H3K79; a photocrosslinking study on human hDot1L supports a corralling/repositioning model in which the functional hotspot on ubiquitin contacts the H2A N-terminus (PMID 26830124). The disagreement is live, and it is now recorded as counter-evidence on 4 and in data.mechanism_class.
Two further pieces of opposition were added at the same time and are stated here so a reader of the summary reaches them: the only human-leukaemia measurement of H2Bub and H3K79 methylation together found them inversely correlated (9, PMID 20442396, now also attached as counter-evidence to 1, 3 and 6); and the therapeutic reading of 5 is opposed both by this project's DepMap scan, in which the upstream writer RNF20 ranks 16,229 of 17,728 with a positive selectivity difference, and by Shema et al., who report RNF20 behaving as a putative tumour suppressor whose depletion elicited transformation and tumorigenesis (PMID 18832071).
| crosstalk | H2BK120ub (in trans) -> DOT1L -> H3K79me |
|---|---|
| direction | bidirectional; DOT1L protein also promotes H2Bub independently of its catalytic activity |
| requirement verdict | state-dependent: modulatory for H3K79me1, effectively required for H3K79me2/me3 |
| mechanism class | contested. Cryo-EM [E4] describes an allosteric mechanism in which ubiquitin plus the H4 tail reposition DOT1L and reorient H3K79 out of the nucleosome core into the active site; photocrosslinking on human hDot1L (PMID 26830124) describes H2B-Ub corralling the enzyme into a productive orientation via a ubiquitin-H2A N-terminus contact. Both predict stimulation; neither has been excluded. |
| functional test in leukaemia | structure-guided DOT1L ubiquitin-binding and acidic-patch mutants fail to support MV4;11 proliferation |
| second ubiquitination site that also works | mark-h2bk34ub |
| competing mechanism model | PMID 26830124 (Zhou 2016, Nat Commun) - ubiquitin hotspot contacts the H2A N-terminus and corrals hDot1L |
| counter evidence added | 2026-09-19 coverage repair: counter-evidence attached to E1, E3, E4, E5 (x2) and E6 |
interfaces
[ "DOT1L-ubiquitin", "DOT1L-H2A/H2B acidic patch", "H4 N-terminal tail" ]
corroborating structures not separately cited
[ "PMID 31173720", "PMID 30981630", "PMID 30923167", "PMID 30759380" ]
- Because the crosstalk is bidirectional and partly catalysis-independent [E7], catalytic DOT1L inhibition may leave the upstream ubiquitin layer intact and therefore leave part of the programme running - a mechanistic candidate explanation for pinometostat's modest clinical effect that this graph has not previously offered.
- Ubiquitin position is flexible [E8], so an intervention aimed at H2BK120 specifically can be bypassed by ubiquitination at another site, and H2BK34 by MSL1/2 is a documented alternative that also stimulates DOT1L.
- The crosstalk also feeds H3K4me2/3 by SET1/COMPASS, so anything upstream of it cannot be selective for the H3K79 arm; the selectivity that made DOT1L an attractive target is lost as soon as you move one step upstream.
- H3K79me1 survives without H2Bub, so a strategy that removes the ubiquitin layer leaves a residual methylation state whose transcriptional function is repressive in yeast [E6] and unknown in human leukaemia - the outcome of partial removal is not simply 'less of the same signal'.
- Does a DOT1L degrader deplete H2Bub in KMT2A-rearranged cells where pinometostat does not? [E7] predicts yes, and it would be a clean mechanistic distinction between inhibition and degradation with direct bearing on why the catalytic inhibitor underperformed.
- Is the H3K79me1-to-me3 zone architecture reported in yeast [E6] reproduced in human KMT2A-rearranged cells, and do fusion target genes sit in the me3 zone? If they do, they are the genes most exposed to H2Bub loss.
- Does H2Bub loss phenocopy DOT1L inhibition at the KMT2A::AFF1 enhancer elements (KEEs) described in the graph's DOT1L entry, or only in gene bodies?
- Can the DOT1L ubiquitin-binding surface be drugged at all? [E5] establishes it is required but there is no chemical matter, and the surface is a shallow protein-protein interface.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Rad6-mediated ubiquitination of H2B lysine 123 is important for efficient methylation of lysine 79, but not lysine 36, of histone H3; conversely lysine 79 methylation of H3 is not required for ubiquitination of H2B.
Ubiquitination of histone H2B by Rad6 is required for efficient Dot1-mediated methylation of histone H3 lysine 79. · Ng HH, Xu RM, Zhang Y, Struhl K · J Biol Chem 2002What this evidence cannot show
Abstract read in full; full text not read. Saccharomyces cerevisiae. The word the authors chose is 'efficient', not 'required' - this is the original statement of a modulatory rather than absolute requirement, and it is the reading that later work refined by methylation state.
Evidence that argues against this (1)
- The only measurement of both marks together in a human leukaemia line points the other way. Darwanto et al. quantified ~20 histone modification sites by LC-MS/MS MRM in U937 and in U937 overexpressing AF10 or an AF10 mutant lacking the hDot1-binding OM-LZ domain, and found H2B ubiquitination inversely correlated with H3K79 methylation, proposing a catalytic-and-inhibitory loop rather than simple positive crosstalk. This entry carries that study as [E9] and describes it in its own section, but it is opposition to this founding requirement claim and is recorded here as such. It is bulk-level, one line, and PICALM::MLLT10 rather than KMT2A - and it is still the only human-leukaemia measurement of the pair that this beat found. Darwanto A et al. 2010
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
H3K4 and H3K79 trimethylation is solely dependent on H2B monoubiquitination regardless of additional alteration to the H2B sequence or genome; a widely used yeast histone shuffle strain (Y131) was found to carry wild-type HTA2 and HTB2 under the GAL1/10 promoter, which explains earlier reports that challenged the dependency.
Histone H2BK123 monoubiquitination is the critical determinant for H3K4 and H3K79 trimethylation by COMPASS and Dot1. · Nakanishi S, Lee JS, Gardner KE, Gardner JM, Takahashi YH, Chandrasekharan MB, Sun ZW, Osley MA, Strahl BD, Jaspersen SL, Shilatifard A · J Cell Biol 2009What this evidence cannot show
Abstract read in full; full text not read. Yeast. The claim is specifically about **trimethylation**; the abstract does not make the same claim for mono- or dimethylation, and this entry does not extend it to them. This paper is itself a correction of prior literature, which is a reminder that the strain background can manufacture a negative result.
Evidence that argues against this (1)
- The absolute form of this dependency is a YEAST result on H2BK123, obtained in a histone-shuffle strain, and it does not transfer unchanged to human chromatin where the requirement is partial. Later in vivo work reports FLEXIBILITY in the crosstalk rather than strict dependence — ubiquitin tethered at a distal nucleosome still promotes methylation. The entry body and its summary table were qualified on 2026-09-20 to state the species rather than leaving the absolute wording to read as a human claim. EMBO Reports 2014
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Chemically ubiquitylated H2B in chemically defined nucleosomes directly stimulates hDot1L-mediated intranucleosomal methylation of H3K79 through the hDot1L catalytic domain, most likely allosterically, and the enhancement in asymmetric dinucleosomes is limited to the nucleosome bearing ubiquitylated H2B.
Chemically ubiquitylated histone H2B stimulates hDot1L-mediated intranucleosomal methylation. · McGinty RK, Kim J, Chatterjee C, Roeder RG, Muir TW · Nature 2008What this evidence cannot show
Abstract read in full; full text not read. Reconstituted system with semisynthetic histones; the isopeptide bond is formed chemically rather than enzymatically. In vitro stimulation of a purified enzyme is not the same as a requirement in chromatin.
Evidence that argues against this (1)
- Same study, applied to the in vitro stimulation claim: a reconstituted, chemically ubiquitylated nucleosome establishes that ubiquitin can stimulate hDot1L, not that the stimulation sets the methylation level in a cell. In the one human leukaemia line where both marks were measured together the correlation was negative (PMID 20442396), which is what an in vitro sufficiency result cannot exclude. Darwanto A et al. 2010
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Cryo-EM structures of Dot1L bound to ubiquitinated nucleosome show how H2B-Ub stimulates Dot1L activity and reveal a role for the histone H4 tail in positioning Dot1L; contacts mediated by Dot1L and the H4 tail induce a conformational change in the globular core of histone H3 that reorients K79 from an inaccessible position so the side chain inserts into the active site primed for catalysis.
Mechanism of Cross-talk between H2B Ubiquitination and H3 Methylation by Dot1L. · Worden EJ, Hoffmann NA, Hicks CW, Wolberger C · Cell 2019What this evidence cannot show
Abstract read in full; full text not read. Cryo-EM of a reconstituted complex; the structures are of a catalytically trapped or mimic state and the resolution and exact construct boundaries are not given in the abstract. Three further 2019 studies (PMIDs 31173720, 30981630, 30923167, 30759380) reached broadly concordant conclusions independently, which is unusually strong corroboration for a structural claim, but this entry cites only the one whose abstract was read.
Evidence that argues against this (1)
- The mechanism is contested, not settled. Zhou et al. used semisynthetic designer chromatin carrying a site-specific photocrosslinker on H2B-Ub and found the functional hotspot on ubiquitin contacts the N-terminus of histone H2A; they concluded that H2B-Ub corrals hDot1L into a productive binding orientation, i.e. a repositioning/restraint model rather than a conformational change in the globular core of H3. Their enzyme is human hDot1L, where [E4]'s cryo-EM used a reconstituted trapped complex. Both models predict stimulation; they disagree about what does the work, so this entry's flat statement that the mechanism is allosteric 'and not a tethering effect' overstated a live disagreement and has been corrected in the body. Zhou L et al. 2016
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Bulk and single-molecule biophysics show that ubiquitin and cofactor binding drive conformational changes that stimulate DOT1L activity, and structure-guided mutations demonstrate that ubiquitin binding and nucleosome acidic-patch binding by DOT1L are each required for cell proliferation in the MV4;11 leukaemia model.
DOT1L activity in leukemia cells requires interaction with ubiquitylated H2B that promotes productive nucleosome binding. · Spangler CJ, Yadav SP, Li D, Geil CN, Smith CB, Wang GG, Lee TH, McGinty RK · Cell Rep 2022What this evidence cannot show
Abstract read in full; full text not read. One cell line (MV4;11, KMT2A::AFF1, AML-like biphenotypic); a separation-of-function mutant rescue experiment can fail for reasons other than the intended interface, such as protein misfolding or altered stability, and the abstract does not state what controls excluded that. No compound exists against either interface.
Evidence that argues against this (2)
- The therapeutic framing built on this result does not survive a selectivity comparator. In this project's genome-wide DepMap scan the upstream writer RNF20 has a KMT2A-selectivity difference of +0.076 and ranks 16,229 of 17,728 - marginally LESS required in KMT2A-rearranged lines - on a pan-essentiality of -0.734; RNF40 is +0.006 (rank 9,953). A DOT1L-ubiquitin-interface agent would act one step below a layer that shows no KMT2A selectivity at all, so the 'distinct chemical problem with a distinct resistance profile' argument does not come with a distinct therapeutic window. Broad Institute DepMap; Cellosaurus; Open Targets Platform; scan run by this project et al. 2026
- Removing the H2B ubiquitin layer may be actively harmful rather than merely unselective. Shema et al. depleted RNF20, the major H2B-specific E3, in human cells: RNF20 suppresses several proto-oncogenes, and its depletion augmented EGF-driven transcription, increased cell migration and elicited transformation and tumorigenesis, with frequent RNF20 promoter hypermethylation in tumours. That makes the writer a putative tumour suppressor and is direct opposition to reading this crosstalk as a place to intervene. Not a leukaemia model and not KMT2A-rearranged. Shema E et al. 2008
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Genome-wide ChIP-seq mapping of H3K79me1, me2 and me3 in yeast identified three state-specific methylation zones; loss of Rad6 leads to the complete loss of the H3K79me3 zone, converting it into an H3K79me1-enriched region while simultaneously diminishing the H3K79me1 zone; loss of H4K16 acetylation also disrupts the H3K79me1 zone, albeit weakly; and Dot1 occupancy does not always correlate with H3K79me3 level.
Trans-histone crosstalk establishes distinct H3K79 methylation zones with differential transcriptional functions. · Park NH, Kim HR, Kim HY, Lee SY, Cho YJ, Roe JS, Kim T · Nucleic Acids Res 2026What this evidence cannot show
Abstract read in full; full text not read. Saccharomyces cerevisiae, not human and not leukaemia. Antibody specificity across the three methylation states is the central technical risk in any such map and the abstract does not describe the validation. The authors' own phrasing is that the zones 'may be' established via H2B ubiquitination, which is weaker than this entry's use of the result; the strong part of the claim is the observed conversion of the me3 zone to me1 on Rad6 loss.
Evidence that argues against this (1)
- Species limit, stated as opposition rather than as a caveat. The zone architecture, the me3-to-me1 conversion and the Rad6 dependence are all Saccharomyces cerevisiae. The only human-leukaemia measurement of H2Bub and H3K79 methylation together reports the opposite sign of association (PMID 20442396), so the transfer of the yeast zone result to human KMT2A-rearranged cells is not supported by the one relevant human dataset that exists. Darwanto A et al. 2010
- E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Increased Dot1 dosage promotes H2B ubiquitination in a dose-dependent manner, exacerbated by loss of SAGA-DUB activity, and the stimulatory effect is mediated by the N-terminus of Dot1 independently of methyltransferase activity, showing that Dot1 and H2Bub1 are subject to bi-directional crosstalk.
Dot1 promotes H2B ubiquitination by a methyltransferase-independent mechanism. · van Welsem T, Korthout T, Ekkebus R, Morais D, Molenaar TM, van Harten K, Poramba-Liyanage DW, Sun SM, Lenstra TL, Srivas R, Ideker T, Holstege FCP, van Attikum H, El Oualid F, Ovaa H, van Leeuwen F · Nucleic Acids Res 2018What this evidence cannot show
Abstract read in full; full text not read. Yeast, and the key experiments use **increased-dosage** DOT1 alleles, which is an overexpression regime; the physiological relevance at normal Dot1 levels is not established by the abstract. Human DOT1L has a very different domain architecture outside the catalytic core, so the transfer of the N-terminus result to DOT1L is an inference.
Evidence that argues against this (1)
- Increased Dot1 dosage promoting H2B ubiquitination reverses the arrow this entry draws: it makes the methyltransferase an upstream influence on the ubiquitin mark rather than only a reader of it, and it does so by a methyltransferase-INDEPENDENT mechanism, which means catalytic inhibition would not reproduce it. Nucleic Acids Research 2018
- E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In engineered yeast strains, ubiquitins tethered to every nucleosome promote H3K79 and H3K4 methylation from a proximal as well as a more distal site, but only if in a correct orientation, indicating that the exact location of the attachment site, the native ubiquitin-lysine linkage and ubiquitination cycles are not critical for trans-histone crosstalk in vivo.
Flexibility in crosstalk between H2B ubiquitination and H3 methylation in vivo. · Vlaming H, van Welsem T, de Graaf EL, Ontoso D, Altelaar AF, San-Segundo PA, Heck AJ, van Leeuwen F · EMBO Rep 2014What this evidence cannot show
Abstract read in full; full text not read. Yeast, and a synthetic tethering system in which every nucleosome carries a ubiquitin - the opposite of the sparse, dynamic, transcription-coupled distribution of the natural mark. The result argues against site specificity, not against the requirement for ubiquitin.
Evidence that argues against this (1)
- Ubiquitin tethered to every nucleosome promoting methylation from a proximal AND a distal position is an engineered-yeast result that argues the crosstalk is permissive rather than strictly intranucleosomal — which conflicts with the intranucleosomal mechanism the reconstituted human work supports, and the entry carries both without noting the tension. EMBO Reports 2014
- E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Using an LC-MS/MS multiple-reaction-monitoring method quantifying nearly 20 histone modification sites simultaneously, relative levels of H2B ubiquitination and H3 Lys-79 methylation were measured in the U937 human leukaemia line and in U937 derivatives overexpressing AF10 or an AF10 mutant lacking the hDot1-binding OM-LZ domain; H2B ubiquitination was found to be inversely correlated with H3 Lys-79 methylation, leading the authors to propose a catalytic and inhibitory loop mechanism rather than simple positive crosstalk.
A modified "cross-talk" between histone H2B Lys-120 ubiquitination and H3 Lys-79 methylation. · Darwanto A, Curtis MP, Schrag M, Kirsch W, Liu P, Xu G, Neidigh JW, Zhang K · J Biol Chem 2010What this evidence cannot show
Abstract read in full; full text not read. This is primarily a methods paper; the biology is the application example. The measurement is of **global bulk** modification levels on acid-extracted histones, not of either mark at defined loci, and every positive result for the crosstalk is locus-level or in vitro. The perturbation is AF10 overexpression, which is not a physiological state and which forces DOT1L recruitment. U937 carries a PICALM::MLLT10 (CALM-AF10) fusion, not a KMT2A rearrangement - verified in Cellosaurus CVCL_0007 for this entry rather than recalled. One cell line. The inverse correlation is reported without a stated effect size or statistic in the abstract.
Evidence that argues against this (1)
- Mass-spectrometric quantification across ~20 sites measures relative abundances, and the entry uses it as evidence about a causal arm. MRM abundance is a snapshot of steady state; a correlation between two marks' levels is compatible with either direction of crosstalk or with a shared upstream cause. Journal of Biological Chemistry 2010
Counter-evidence (Zhou 2016, PMID 26830124): photocrosslinking in semisynthetic designer chromatin places the functional hotspot on ubiquitin against the H2A N-terminus and describes H2B-Ub corralling hDot1L into a productive orientation - a repositioning model competing with the allosteric H3-core reorientation this entry asserted, and obtained with the human enzyme
Counter-evidence (Darwanto 2010, PMID 20442396): the only human-leukaemia measurement of H2Bub and H3K79 methylation together - LC-MS/MS MRM in U937 plus or minus AF10 - found them inversely correlated, which is opposition to the licensing chain the graph's 35 H3K79me2 entries rest on
Counter-evidence (DepMap scan, this project): RNF20 ranks 16,229 of 17,728 with a positive selectivity difference (+0.076) on a pan-essentiality of -0.734, and RNF40 is +0.006 at rank 9,953, so the therapeutic framing built on the MV4;11 interface mutants sits one step below a layer with no KMT2A selectivity
Counter-evidence (Shema 2008, PMID 18832071): RNF20 depletion in human cells de-repressed proto-oncogenes, increased migration and elicited transformation and tumorigenesis, with frequent RNF20 promoter hypermethylation in tumours - the writer of this mark behaves as a putative tumour suppressor, which opposes intervening anywhere on this crosstalk
- ← depends onDOT1L methylates H3K79 non-processively - so the three states are one distribution, not three signals
H2Bub is the one real state boundary in an otherwise continuous rheostat
the crosstalk requires the H2B ubiquitin mark
modulatory for H3K79me1, effectively required for H3K79me2/me3
- supports →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
DOT1L ubiquitin-binding and acidic-patch mutants fail to support proliferation of the KMT2A::AFF1 line MV4;11
- related to →Pinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
bidirectional, catalysis-independent arm predicts a catalytic inhibitor leaves the ubiquitin layer intact where a degrader would not
THE UNJOINED LINK: if CDK9 licenses H2BK120ub and H2BK120ub is the trans-histone input to DOT1L, then a CDK9 inhibitor is an indirect H3K79 agent in KMT2A-rearranged cells. No H2Bub1 or H3K79me2 readout under any CDK9 inhibitor exists in this disease.