RTF1 — the PAF1C subunit the graph was missing, on a complex that does bind the retained KMT2A fragment

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• PAF1C subunits are required in 80% of cell lines; blocking the complex would be broadly toxic.
• If the PAF1C contact and the menin contact both sit on the retained KMT2A N-terminus, a menin inhibitor may leave PAF1C-mediated recruitment intact, which is a candidate route to incomplete response that has not been examined.

1. Which PAF1C subunit contacts KMT2A? The complex-level result has stood since 2010 without subunit resolution, and it determines whether the interface is druggable.
2. Does the PAF1C contact region overlap or abut the menin-binding motifs at the KMT2A N-terminus? Two interactions competing for the same retained fragment would be a strong reason to map both.
3. Has PAF1C dependence been tested in KMT2A::AFF1 B-ALL, as opposed to MLL-AF9 myeloid models? Every functional result here is AF9.
4. Is RTF1 a stoichiometric PAF1C subunit in human leukemia cells, or a loosely associated one as in yeast?
5. Where is human RTF1 in any cell? There is no imaging, no fractionation and no subnuclear assignment for this protein in any organism's literature that a localisation-scoped search reaches - and it is a PAF1C subunit whose contact sits on the retained KMT2A fragment of every clinical fusion.
6. Does the human RTF1 HMD stimulate H2BK120ub in a KMT2A-rearranged cell, and does removing it change H3K4me3 at HOXA? The HMD is a point-mutatable module, so this is a single-allele experiment and nobody has run it in leukaemia.
7. Which PAF1C subunit does the H2Bub work at MLL-fusion target genes? The biochemistry says Ctr9 docks Rad6 and the Rtf1 HMD accelerates discharge; in human cells RTF1 is only loosely complex-associated, so the answer may differ from yeast and decides which subunit is the tractable node.
8. Is the H3K4 mono- versus tri-methylation asymmetry (PMID 16185711) preserved in human COMPASS, where KMT2A is one of six H3K4 methyltransferases? If it is, an RTF1 perturbation would move H3K4me1 and H3K4me3 in different directions and a single bulk western would hide it.

1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   MLL interacts directly with the polymerase-associated factor complex through sequences flanking the CxxC domain; PAFc interacts with RNA polymerase II and stimulates post-translational histone modifications; PAFc augments MLL- and MLL-AF9-mediated transcriptional activation of Hoxa9; PAFc knockdown disrupts MLL fusion protein-mediated transcriptional activation and MLL recruitment to target loci; and MLL deletions that abolish PAFc interaction also eliminate MLL-AF9-mediated immortalisation.

   The PAF complex synergizes with MLL fusion proteins at HOX loci to promote leukemogenesis. · Muntean AG, Tan J, Sitwala K, Huang Y, Bronstein J, Connelly JA, Basrur V, Elenitoba-Johnson KSJ, Hess JL · Cancer Cell 2010
   PMID 20541477DOI
   ▸What this evidence cannot show

   Abstract read in full; not full text. The fusion tested is **MLL-AF9**, not KMT2A::AFF1, and the immortalisation assay is murine. The interaction is established for PAFc as a complex; the abstract does not say which subunit contacts MLL, so this entry cannot attribute the contact to RTF1 specifically. 'Sequences flanking the CxxC domain' is the resolution given — no residue numbers.

   ▸Evidence that argues against this (2)

   • This entry's first sentence - 'RTF1 is a subunit of the polymerase-associated factor complex (PAF1C)' - is the step on which its whole claim to the fusion interactome rests, and it is contested for human cells. Cao et al. state that 'Rtf1, however, is not stably associated with the PAF1C in most species except Saccharomyces cerevisiae', show that human Rtf1 is a transcription elongation factor that may function independently of PAF1C and requires a separate coactivator activity unrelated to PAF1C or DSIF, and report from RNA-seq and ChIP in HeLa cells that Rtf1 and PAF1C regulate distinct gene subsets and that PAF1C was recruited to the genes examined in an Rtf1-INDEPENDENT manner. The transitive inference 'PAF1C binds the retained CxxC-flanking region, therefore RTF1 is on the fusion' fails at step one. The entry should say RTF1 is a PAF1C-associated factor whose presence in the human complex is sub-stoichiometric and disputed. Read: abstract in full, not full text. Cao QF et al. 2015
   • The separation is not new and was mapped genetically in yeast, where Rtf1 IS a stable Paf1C subunit. Warner et al. made sequential deletions across Rtf1 and found discrete non-overlapping segments for Chd1 interaction, for promoting H2B and H3 modification, for recruitment to active ORFs, and for association with other Paf1 complex subunits - and, critically, that 'disruption of the physical association between Rtf1 and other Paf1 complex subunits caused only subtle mutant phenotypes' while the histone-modification and recruitment regions were required. Rtf1's functions are therefore separable from its complex membership even where the membership is real, which removes the last reason to treat 'PAF1C binds KMT2A' as a statement about RTF1. Read: abstract in full, not full text. Warner MH et al. 2007
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

   RTF1 has DepMap mean gene effect -0.6659 across 1,208 lines with 80.1% of lines depleted, mean -0.7026 in 11 KMT2A-rearranged leukemia lines against -0.6201 in 43 other leukemia lines (difference -0.083); SKIC8 (formerly WDR61) has mean -0.7458 with 81.0% depleted and a difference of -0.016; MEN1 in the same lines is -0.866 on an all-lines mean of -0.363.

   DepMap CRISPR (Chronos) gene effect via the Open Targets Platform GraphQL API, with KMT2A-rearrangement status from Cellosaurus; per-gene recomputation for this beat · Broad Institute DepMap; Cellosaurus; Open Targets Platform; analysis run by this project · Database and derived analysis 2026
   ▸What this evidence cannot show

   DepMap release version not stated by the API. Eleven KMT2A-rearranged lines. A gene depleted in 80% of lines has limited dependency variance, which compresses any difference statistic computed on it.

   ▸Evidence that argues against this (2)

   • Beyond whether RTF1 is in the complex, the direction of any PAF1C-directed intervention in a KMT2A-rearranged cell is unsettled. Chen et al. show that reducing PAF1 releases paused Pol II into gene bodies at thousands of genes, raising nascent and mature transcript levels and Ser2 phosphorylation, and that the effect is mediated by recruitment of the super elongation complex - the very machinery KMT2A::AFF1 uses. Depleting PAF1C could therefore increase output at fusion targets. Read: abstract in full, not full text. Chen FX et al. 2015
   • And the opposite arrow was published four months later in Science: Yu et al. report PAF1 as 'a critical regulator of paused Pol II release', with P-TEFb directly regulating PAF1C recruitment and PAF1C in turn required for CDK12 recruitment and CTD Ser2 phosphorylation. Two 2015 papers from major laboratories assign PAF1 opposite roles in pause release. Any entry proposing PAF1C as the second handle on the retained KMT2A fragment has to carry that disagreement, because it determines the sign of the intended effect. Read: abstract in full, not full text. Yu M et al. 2015
3. E_q65_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Human RTF1 is resolved within the complete Pol II-DSIF-PAF-SPT6 transcription complex by cryo-EM, where it acts as an allosteric activator - placing RTF1 on elongating RNA polymerase II as a structural fact, which is the only positional information available for the human protein.

   Structure of complete Pol II-DSIF-PAF-SPT6 transcription complex reveals RTF1 allosteric activation. · Vos SM, et al. · Nat Struct Mol Biol 2020;27:668-677 2020
   PMID 32541898DOI
   ▸What this evidence cannot show

   A reconstituted complex, not a cell: this establishes molecular position within an assembly and says nothing about subnuclear compartment, concentration or dynamics. It is recorded as the boundary of what is known rather than as a localisation. Abstract only.

   ▸Evidence that argues against this (1)

   • Complex membership is not a localisation, and this graph already has the worked example: PAF1, RTF1's own complex partner, turns out to have nucleoplasmic puncta AND a nucleolar pool - an address no structural model of PAF1C predicts. Inheriting PAF1's compartment for RTF1 would be exactly the inference this entry refuses to make. UniProt Knowledgebase, reviewed (Swiss-Prot) 2026
4. E_q69_ct1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   RTF1, a gene associated with increased risk of ulcerative colitis, is required for Th17 cell differentiation: Rtf1 deficiency disrupts the differentiation of interleukin-17A-expressing T cells while leaving regulatory T cells unaffected; mechanistically RTF1 facilitates histone H2B monoubiquitination, which requires its histone modification domain, and impaired Th17 differentiation is also observed in cells lacking the H2Bub1 E3 ligase subunit RNF40, an enzyme known to physically interact with RTF1.

   RTF1 mediates epigenetic control of Th17 cell differentiation via H2B monoubiquitination. · Galan C, Lu G, Gill R, Li D, Liu Y, Huh JR, Hang S · J Immunol 2025
   PMID 40073106DOI
   ▸What this evidence cannot show

   Abstract read in full; full text not read. T-cell differentiation in an inflammatory-disease context - not leukaemia, not B-lineage, not KMT2A-rearranged (brief rule 73). A single 2025 study. The Treg-unaffected control is the strongest part of it and is why the result is recorded as a cell-type answer at all.

   ▸Evidence that argues against this (1)

   • RNF40 loss phenocopies RTF1 loss in the same assay, so the measurement may be of an H2Bub1 requirement in Th17 cells rather than of an RTF1-specific function. On that reading RTF1 is a route to the mark, not a lineage factor, and the cell-type specificity belongs to the mark's targets rather than to RTF1. Galan C et al. 2025
5. E_q97_epi1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   PAF1C is required for Rad6/Bre1-mediated monoubiquitination of histone H2B, and the contributions are SUBUNIT-RESOLVED: Ctr9 acts as a hub and interacts with the carboxyl-terminal acidic tail of Rad6, which is required for PAF1C-induced stimulation of H2Bub; the Ras-like domain of Cdc73 accelerates ubiquitin discharge from Rad6; and the Rtf1 histone modification domain stimulates H2Bub, probably by accelerating ubiquitin discharge from Rad6 alone or in cooperation with Cdc73.

   Biochemical insights into Paf1 complex-induced stimulation of Rad6/Bre1-mediated H2B monoubiquitination. · Proc Natl Acad Sci U S A 2021
   PMID 34385316
   ▸What this evidence cannot show

   Abstract fetched and read in full 2026-09-21; full text not read. RECONSTITUTED BIOCHEMISTRY on the YEAST complex (Rad6/Bre1 are the yeast enzymes; the human orthologues are UBE2A/B and RNF20/RNF40, both of which this graph holds as gene-rnf20 and gene-rnf40). The authors state the Cdc73 mechanism MIGHT be conserved from yeast to humans, which is the paper's own hedge and is carried here. No mammalian cell, no haematopoietic system, no KMT2A fusion. This is cited for ONE thing: it is the source that assigns the H2Bub contribution to named PAF1C subunits rather than to the complex, which is what rule 64 demands before a subunit entry may claim the complex's mark.

   ▸Evidence that argues against this (1)

   • Subunit resolution obtained in a reconstituted yeast system is not subunit resolution in a human blast. Human PAF1C composition differs - notably RTF1 is only loosely associated with human PAF1C rather than being a stable core subunit, which this graph's gene-rtf1 entry already records from Warner's deletion series - so a contribution measured in a yeast complex where Rtf1 IS a stable subunit may not transfer. The direction of the possible error is that the human Rtf1 contribution could be larger (a separable, diffusible stimulatory module) or smaller (not present in the complex at the relevant time), and this abstract cannot distinguish them. Genetics 2011
6. E_q97_epi2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   Single amino-acid substitutions within a region of Rtf1 conserved from yeast to humans, termed the histone modification domain, result in loss of histone H2B ubiquitylation and impaired histone H3 methylation, with further defects in telomeric silencing, transcription elongation and prevention of cryptic initiation, and they disrupt 3'-end formation of snoRNA transcripts, identifying a regulatory role for the histone H2B K123 ubiquitylation mark in that process.

   Identification of a role for histone H2B ubiquitylation in noncoding RNA 3'-end formation through mutational analysis of Rtf1 in Saccharomyces cerevisiae. · Genetics 2011
   PMID 21441211
   ▸What this evidence cannot show

   Abstract fetched and read in full 2026-09-21; full text not read. SACCHAROMYCES CEREVISIAE throughout, and the ubiquitylated residue is H2B K123 - the YEAST site; the human equivalent is H2BK120 (this graph's mark-h2bk120ub). H3 methylation is reported as impaired without the state being named in the abstract, so H3K4 specifically is not quoted here. The conservation claim is the authors' own - the domain is conserved from yeast to humans - and is the only bridge to the human protein offered.

   ▸Evidence that argues against this (1)

   • The same mutational series shows the HMD is pleiotropic - telomeric silencing, elongation, cryptic initiation and snoRNA 3'-end formation all break together - so a phenotype after HMD disruption cannot be attributed to the H2Bub loss without an independent H2Bub-restoring control. In a human leukaemia context that matters directly: an RTF1-directed perturbation would move several chromatin and RNA-processing outputs at once. J Mol Biol 2005
7. E_q97_epi3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   H3K4 mono-methylation does not require H2B ubiquitination: loss of Paf1 or of H2B ubiquitination has only a modest effect on bulk H3K4me1; despite decreased Set1 recruitment, loss of Paf1 INCREASES H3K4me1 at the 5' coding region of active genes, while deleting RTF1 affects H3K4me1 at the 3' coding region and decreases global H3K4me1 - so the requirements for mono-methylation are distinct from those for di- and tri-methylation and PAF1C subunits differ from one another.

   Histone H3 lysine 4 mono-methylation does not require ubiquitination of histone H2B. · J Mol Biol 2005
   PMID 16185711
   ▸What this evidence cannot show

   Abstract fetched and read in full 2026-09-21; full text not read. YEAST, and the quantification is bulk N-terminal Edman sequencing plus locus-level ChIP on a small number of active genes - not genome-wide. It is a 2005 paper and the H3K4me1 antibody landscape has changed since. Cited for the SUBUNIT-ASYMMETRY and MARK-STATE-ASYMMETRY conclusions, which are the parts that refute the simple cascade, not for a quantitative effect size.

   ▸Evidence that argues against this (1)

   • A yeast result that separates H3K4me1 from H2Bub does not establish the same separation in human cells, where the H3K4 methyltransferases are a six-member COMPASS family including KMT2A itself rather than a single Set1. The counter-possibility is that the human trans-tail requirement is tighter than the yeast one, in which case the simple cascade would be closer to right for this graph's system - and no measurement in a KMT2A-rearranged cell exists either way. Proc Natl Acad Sci U S A 2021

• refutes →PAF1C couples transcription to H2B ubiquitination - but the RTF1 half of that coupling is yeast-specific and contested in human cells, and PAF1C is one of the few complexes that genuinely binds the retained KMT2A fragment

  Counter-evidence (Cao 2015, PMID 26217014): human Rtf1 is not stably associated with PAF1C, regulates a distinct subset of genes from PAF1C in HeLa, and - opposite to the yeast result - PAF1C recruitment to the genes examined was Rtf1-independent, so the RTF1 half of this coupling may not transfer to human cells

• related to →Separation-of-function RTF1 rescue in a KMT2A-fusion line: does the fusion need PAF1C for elongation, or for H2B ubiquitination?

  Design objection (Cao 2015, PMID 26217014): if human Rtf1 is not stably in PAF1C and PAF1C recruitment is Rtf1-independent in human cells, an RTF1 degron tests RTF1 rather than the fusion's PAF1C contact; the separation-of-function precedent that does exist is in yeast RAD6 (PMID 37216505), pointing at a UBE2A-side allele or a catalytically dead RNF20 knock-in instead

• regulates →H2BK120ub (H2Bub1) — the ubiquitin tag on histone H2B that licenses H3K79 methylation, and the mark this graph was missing

  Rtf1 is required for H2B ubiquitination; its effects on H3K4 and H3K79 methylation are an indirect consequence

• refutes →PAF1 — pan-essential subunit of a complex the fusion needs, where the window is in the interface and not in the protein

  human RTF1 is not stably associated with PAF1C outside S. cerevisiae, regulates a distinct gene set, and PAF1C recruitment was RTF1-independent (Cao 2015)

• part of →PAF1 — pan-essential subunit of a complex the fusion needs, where the window is in the interface and not in the protein

  PAF1C subunit; completes the complex alongside CTR9, CDC73, LEO1 and SKIC8

• binds →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)

  PAF1C binds sequences flanking the CxxC domain, which the fusion retains (complex-level claim, shown for MLL-AF9)

• related to →SKIC8 (formerly WDR61) — the last PAF1C subunit, and a gene-symbol trap worth recording

  the two PAF1C subunits the graph was missing before this cycle

• regulates →HOXA9 — required for survival of established MLL-rearranged leukemia, and dispensable for starting it

  PAF1C augments MLL- and MLL-AF9-mediated activation of Hoxa9

• ← related toA quarter of this graph's gene entries have a UniProt localization that nobody measured - and the subnuclear vocabulary UniProt does carry avoids the genes we care about most

  core elongation machinery whose localisation is by-similarity only

• ← supportsCorrections owed to other beats: nine entries whose stated provenance is now resolvable, and one unflagged import

  the sibling entry that flags the same paper correctly

• related to →PAF1 is annotated as punctate everywhere in the nucleus EXCEPT the nucleolus - and a 2024 proximity map puts it in the nucleolus

  RTF1's compartment is unmeasured in any organism; its complex partner PAF1 has nucleoplasmic puncta and a nucleolar pool, and this entry deliberately does not inherit that address

• related to →Transcription factories and Pol II clusters - no KMT2A::AFF1 imaging, condensate evidence for SEC and MLL-ENL, and 3D-FISH that actively contradicts the factory story for t(4;11)

  cryo-EM places RTF1 on elongating Pol II inside the complete Pol II-DSIF-PAF-SPT6 complex - a molecular position, not a subnuclear one

• related to →RNF20 (BRE1A) — the H2B ubiquitin ligase that MLL-rearranged leukaemia needs, and that DepMap says every other cancer line needs just as much

  RTF1's Th17 requirement runs through H2B monoubiquitination and is phenocopied by loss of the E3 ligase subunit RNF40, the partner of RNF20 (PMID 40073106) - so the two entries describe one module

• supports →Expressed, required, maintains — three different claims about where a gene matters, and the graph must label which one it has

  a worked case where required_in is answerable (Th17, with Treg as an internal negative) while maintains_state is genuinely unpublished in every haematopoietic compartment this graph models

• regulates →Th17 cell - the mature CD4 T-helper subset whose differentiation and maintenance require MENIN, which makes it the clearest published on-target normal-tissue liability of the menin interface

  Rtf1 deficiency disrupts Th17 differentiation while leaving Treg unaffected, via H2B monoubiquitination through its histone modification domain; the Treg-spared control is what makes it programme-specific. The cell type Q69 named as missing vocabulary

• regulates →H3K4me3 - the mark KMT2A's own SET domain writes, that the fusion cannot write, that an interface in this graph already binds, and that switches PRC2 off

  RTF1 deletion decreases global H3K4me1 and shifts it 3'; the mono- versus di/tri requirements are distinct, so the H2Bub-to-H3K4 link is not a simple cascade

• ← part ofCTR9 — a PAF1C subunit the field assumes is a fusion partner, with no paper that tested it

  Ctr9/Paf1 is the PAF1C assembly scaffold and the Ctr9 TPR module provides the Rtf1 and Cdc73 interfaces; the two subunits stimulate H2Bub by different steps