Oncofusion

start / graph / localization-paf1-nucleoplasmic-puncta-versus-a-nucleolar-pool

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q57-localization2

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

also called PAF1 localization · hPaf1 PD2 · PAF1 nucleolar · PAF1c subnuclear · intergenic rDNA PAF1

The standard database entry for PAF1 says it is scattered in dots through the nucleus but not in the nucleolus. A 2024 study that mapped what sits next to RNA polymerase II inside nucleoli found PAF1 there, and showed that removing it wrecks nucleolar organisation. Both statements cannot be right, and nobody has noticed the clash.

The same thing, for a scientist

UniProt Q8N7H5 annotates PAF1 as `Nucleus` on two experimental references, with a subnuclear note - *"Punctuate distribution throughout the nucleus except in nucleoli and the perinuclear chromatin"* - that carries **no evidence code at all** [E1]. The note traces to the paper that characterised PD2/hPaf1 as a nuclear 80 kDa protein interacting with RNA polymerase II in a pancreatic-cancer amplicon study [E2]. Against it: compartment-enriched proximity-dependent biotin identification of the **nucleolar** Pol II interactome recovers PAF1, places it at intergenic rDNA spacers where it promotes Pol II elongation and prevents unscheduled R-loops, and shows that **PAF1 deficiency disrupts nucleolar organisation and rRNA biogenesis**, rescued by repressing those R-loops [E3]. So the graph now holds two incompatible subnuclear statements for a protein that is a documented component of the KMT2A::AFF1 machinery [E4]. **Neither statement has been tested in a leukaemic cell** [E5].

Go deeper
The clash, in the two sources' own words

UniProt, fetched 2026-09-20 (Q8N7H5), complete note text:

"Punctuate distribution throughout the nucleus except in nucleoli and the perinuclear chromatin"

The note carries no evidences array. The Nucleus location itself carries ECO:0000269 for PubMed 16491129 and 30367041 - and the second of those is a glioblastoma paper about a circRNA-encoded peptide that binds PAF1c, not a localisation study, which shows how loosely a location reference can attach.

Khosraviani et al. 2024 3: compBioID with nucleolar Pol II recovers PAF1; PAF1 acts at intergenic rDNA spacers; PAF1 deficiency disrupts nucleolar organisation and rRNA biogenesis; repressing unscheduled IGS R-loops in PAF1-deficient cells rescues nucleolar organisation and rRNA production.

A protein whose loss disorganises the nucleolus and which is recovered by a nucleolus-restricted proximity map is not plausibly absent from nucleoli.

How to reconcile them honestly. The two are not necessarily in flat contradiction: proximity labelling detects transient and substoichiometric pools that an immunofluorescence exclusion would miss, and the 2006 observation may be accurate about the bulk distribution while the 2024 result is about a functional minority pool. But the database sentence says 'except in nucleoli' without qualification, and that is the sentence a graph copies. What would settle it: immunofluorescence of endogenous PAF1 co-stained with fibrillarin and nucleophosmin, at exposure adequate to see a minority pool, in the same cells used for the compBioID.

Why PAF1's compartment is load-bearing here

PAF1c is not a peripheral factor in this disease. It binds the KMT2A N-terminus, it is retained by the fusion, and MLL-AF4 cooperates with PAF1 and FACT to drive high-density enhancer interactions in leukaemia 4. The graph already carries that paper in several entries.

If PAF1 also has a nucleolar job in rRNA biogenesis, then:

  • a PAF1 dependency in a KMT2A-rearranged screen is not attributable to the fusion complex without further evidence - the same problem this beat found for DPY30 and for WDR5;
  • targeting the PAF1c-KMT2A interface would leave the nucleolar pool intact, which is good for the therapeutic index and bad for interpreting genetic data as a proxy for the pharmacology;
  • and nucleolar stress is already in play in this disease from the WDR5 WIN-site work, so two of the core interactome's subunits converge on ribosome biogenesis from different directions.
What has never been measured

No source located on 2026-09-20 images PAF1 in a KMT2A-rearranged cell 5. The PAF1 literature in this disease is entirely genomic - ChIP, degron, occupancy, enhancer contacts. That is the same pattern as DOT1L, PSIP1 and MEIS1 in this beat: the genomic address is well mapped and the nuclear address is unmeasured.

The cheap experiment is the same one three times over: endogenous protein, two compartment markers, one KMT2A-rearranged line, before and after menin inhibition.

Facts and figures
genegene-paf1
uniprotQ8N7H5
length AA531
compartment subcellularnucleus
compartment subnuclearCONTESTED. UniProt's uncited note says punctate throughout the nucleus EXCEPT in nucleoli and perinuclear chromatin; a 2024 nucleolus-restricted proximity map recovers PAF1 with nucleolar Pol II at intergenic rDNA, and PAF1 loss disrupts nucleolar organisation and rRNA biogenesis.
contradiction recordednucleolar exclusion (UniProt note, uncited, traceable to PMID 16491129) versus a functional nucleolar pool (PMID 39505901, compBioID plus an R-loop rescue). Possible reconciliation: bulk distribution versus a transient functional minority pool. Resolving experiment: endogenous PAF1 immunofluorescence with fibrillarin and nucleophosmin at exposure adequate for a minority pool, in the cells used for the compBioID.
identifier trap hitgene_exact:PAF1 AND organism_id:9606 AND reviewed:true returns TWO reviewed human records - Q8N7H5 (PAF1, 531 aa) and P28328 (PEX2, 305 aa, peroxisomal, carrying PAF1 as a gene synonym). Rule 62 applies; the row was selected on gene_primary and the length checked.
acronym trapPD2, the older name for this protein, collides with Platycodin D2 and several unrelated abbreviations; every PD2 hit in a literature count must be title-checked
shown for fusionNO for localisation. PAF1 is load-bearing in KMT2A::AFF1 biology genomically (MLL-AF4 cooperates with PAF1 and FACT at high-density enhancer interactions, PMID 37626123), but its position in a leukaemic nucleus has never been measured.
patient material datafalse
bounded absence date2026-09-20
uniprot evidence codes
{
  "Nucleus": "ECO:0000269|PubMed:16491129, 30367041",
  "subnuclear note (punctate, not in nucleoli)": "NO evidence code attached"
}
methods
[
  "immunofluorescence (2006, stringency unknown)",
  "compartment-enriched proximity-dependent biotin identification (compBioID) with nucleolar Pol II",
  "R-loop repression rescue of nucleolar organisation"
]
species and systems
[
  "human pancreatic cancer cells and NIH 3T3 (2006)",
  "human cell lines (2024 nucleolar study)"
]
How this could bite us back
  • If PAF1 has a nucleolar requirement for rRNA biogenesis, any PAF1-directed agent carries a ribosome-biogenesis toxicity in every proliferating tissue, and nucleolar stress responses are p53-dependent - so efficacy and toxicity would both shift in TP53-altered relapse disease.
  • Genetic PAF1 dependency data cannot be used as a proxy for the pharmacology of a PAF1c-KMT2A interface inhibitor, because the genetic perturbation removes the nucleolar pool as well.
  • A pharmacodynamic readout based on fusion-target gene expression would miss a nucleolar effect entirely, and vice versa.
  • The PD2 acronym collision means any automated literature or safety sweep on this target will return large amounts of irrelevant material, and a counted absence built on it without title-checking would be worthless.
Still open
  1. Is PAF1 in the nucleolus or excluded from it? A single co-stain with fibrillarin and nucleophosmin at adequate exposure settles a clash that currently sits unremarked between a database sentence and a 2024 proximity map.
  2. How much of a PAF1 dependency in KMT2A-rearranged leukaemia is fusion-complex and how much is rRNA biogenesis? A separation-of-function allele, or an rDNA-restricted degron, would separate them; neither exists.
  3. Two core interactome subunits - PAF1 and WDR5 - now converge on ribosome biogenesis from different directions. Is nucleolar stress a shared mechanism of the chromatin-directed agents in this disease rather than a side effect?
  4. Does menin inhibition change PAF1's nuclear distribution? Nobody has looked, and the reagents and cells are standard.
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt Q8N7H5 (human PAF1, 531 aa) annotates Nucleus with ECO:0000269 for PubMed 16491129 and 30367041, and carries a subnuclear note - 'Punctuate distribution throughout the nucleus except in nucleoli and the perinuclear chromatin' - with no evidence code attached to the note.

    What this evidence cannot show

    IDENTIFIER TRAP HIT AND RECORDED (rule 62): `gene_exact:PAF1 AND organism_id:9606 AND reviewed:true` returns TWO reviewed human records - Q8N7H5 (gene_primary PAF1, 531 aa) and P28328 (gene_primary PEX2, 305 aa, a peroxisomal membrane protein that carries PAF1 as a gene synonym). A script taking the first data row could have written a peroxisome-membrane accession into a transcription-elongation entry. The row was selected on gene_primary and the length checked. This is the same shape as the TRAF2/TANK case in the standing brief.

    Evidence that argues against this (1)
    • An uncited note in UniProt is not necessarily wrong - it may simply predate the evidence-tagging system. But it is the only subnuclear statement UniProt makes about PAF1, it is the statement a graph would copy, and a 2024 proximity map contradicts it. An entry citing UniProt for 'PAF1 is excluded from nucleoli' would be citing an unattributed sentence.
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    PD2, the human homologue of RNA polymerase II-associated factor 1 (hPaf1), located on the 19q13 amplicon in pancreatic cancer cells, is a nuclear 80 kDa protein that interacts with RNA polymerase II, and its overexpression in NIH 3T3 cells enhances growth rate and produces tumour formation in vivo.

    The human homologue of the RNA polymerase II-associated factor 1 (hPaf1), localized on the 19q13 amplicon, is associated with tumorigenesis. · Moniaux N, Nemos C, Schmied BM, Chauhan SC, Deb S, Morikane K, Choudhury A, Vanlith M, Sutherlin M, Sikela JM, Hollingsworth MA, Batra SK · Oncogene 2006
    PMID 16491129DOI
    What this evidence cannot show

    RULE 73: pancreatic cancer cells and NIH 3T3 fibroblasts - not haematopoietic and not leukaemic. The abstract states only that PD2 is a nuclear 80 kDa protein; the punctate/nucleolar-exclusion detail that UniProt carries is not in the abstract, and this entry did not read the full text, so the stringency of that observation (exposure, antibody, marker co-stain) is unknown. The paper's framing is oncogene discovery in a 19q13 amplicon, not cell biology. Read: abstract only (PubMed efetch, 2026-09-20).

    Evidence that argues against this (1)
    • An immunofluorescence exclusion from a compartment is one of the weakest positive-looking observations in cell biology: a minority pool, a masked epitope in a dense compartment, or a short exposure all produce an apparent hole. Nucleoli in particular are notorious for epitope masking and for being read as negative space in low-exposure images.
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Compartment-enriched proximity-dependent biotin identification of the nucleolar RNA polymerase II interactome recovers PAF1 together with TBPL1; PAF1 promotes Pol II elongation at intergenic rDNA spacers and prevents unscheduled R-loops that hyper-restrain Pol I-associated non-coding RNAs; PAF1 deficiency disrupts nucleolar organisation and rRNA biogenesis, and repressing the unscheduled intergenic-spacer R-loops in PAF1-deficient cells rescues nucleolar organisation and rRNA production.

    Nucleolar Pol II interactome reveals TBPL1, PAF1, and Pol I at intergenic rDNA drive rRNA biogenesis. · Khosraviani N, Yerlici VT, St-Germain J, Hou YY, Cao SB, Ghali C, Bokros M, Krishnan R, Hakem R, Lee S, Raught B, Mekhail K · Nat Commun 2024
    PMID 39505901DOI
    What this evidence cannot show

    RULE 73: human cell lines in a nucleolar-biology study; no leukaemic or haematopoietic system. Proximity labelling reports proximity, not residence: a factor can be biotinylated by a nucleolar bait during transient contact at the nucleolar periphery. The compartment enrichment strengthens but does not eliminate this. The rescue experiment (repressing R-loops restores nucleolar organisation) is the strongest part and is functional rather than positional. Read: abstract only (PubMed efetch, 2026-09-20).

    Evidence that argues against this (1)
    • A functional requirement at rDNA does not by itself place the bulk of PAF1 in the nucleolus, and proximity labelling is biased toward detecting transient pools that imaging cannot see. The two sources may both be right about different pools - which is the reading this entry favours - but the database sentence as written is unqualified, and no published source reconciles them.
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MLL-AF4 cooperates with PAF1 and FACT to drive high-density enhancer interactions in leukaemia.

    MLL-AF4 cooperates with PAF1 and FACT to drive high-density enhancer interactions in leukemia. · Crump NT, Smith AL, Godfrey L, Dopico-Fernandez AM, Denny N, Harman JR, Hamley JC, Jackson NE, Chahrour C, Riva S, Rice S, Kim J, et al. · Nat Commun 2023
    PMID 37626123DOI
    What this evidence cannot show

    Cited here only to establish that PAF1 is load-bearing in KMT2A::AFF1 biology, which is why its compartment matters; the paper is a genomic study of enhancer contacts and contains no localisation measurement. Title and identifiers fetched from PubMed 2026-09-20; the paper itself was read only at abstract level for this entry, and it is already cited in several other entries in this graph - those citations were not reused, the ids were re-fetched.

    Evidence that argues against this (1)
    • That PAF1 is required for a fusion-driven enhancer programme does not establish that a PAF1 dependency in this disease is a fusion dependency; PAF1 has a separate nucleolar requirement for rRNA biogenesis, and a genetic screen cannot attribute a dependency between the two. This is the same attribution problem this beat records for DPY30 and WDR5.
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Bounded absence, measured 2026-09-20: no study located reports the subnuclear localisation of PAF1 protein in a KMT2A-rearranged cell.

    What this evidence cannot show

    QUERY: `(PAF1[tiab] OR "PAF1 complex"[tiab] OR hPaf1[tiab] OR PD2[tiab]) AND (immunofluorescence[tiab] OR "subcellular localization"[tiab] OR nucleolus[tiab] OR nucleolar[tiab] OR speckle[tiab])` = 29 records on 2026-09-20; the first 15 were title-checked and none is a leukaemia localisation study - the on-topic ones are the nucleolar Pol II interactome and a CTR9 neurodevelopmental paper, and roughly half are PD2 acronym noise (Platycodin D2, platelet-derived, pressure-overload). RULE 67 CONTROL: the shape returns 29, so it returns ones. RULE 70 NAME-LIST WARNING: 'PD2' is a four-character acronym that collides heavily - the same class of trap as BDME in the standing brief - and every PD2 hit had to be title-checked. NAME LIST: PAF1, PAF1 complex, hPaf1, PD2; NOT searched: PAF1C as a token, and the previous symbol PD2 in combination with a leukaemia term. RULE 80 (added mid-beat): this count is bounded by TITLES AND ABSTRACTS, not by papers. Localisation panels are very often supporting figures whose abstracts never mention them, so a PubMed zero here means 'no paper ANNOUNCES this measurement', not 'no such figure exists'. Supporting figures and supplementary material were NOT searched.

    Evidence that argues against this (1)
    • As elsewhere in this beat, the absence concerns imaging only. PAF1 occupancy in KMT2A-rearranged cells is well characterised genomically, including at fusion-bound enhancers, which constrains where PAF1 is on DNA without constraining where it is in the nucleus.
Connected to