TRIM27 - the E3 ligase inside PRC1.1 that USP7 spends its time undoing, and the worst UniProt localisation record this graph has audited

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.

• The only available pharmacology on this axis lowers PRC1.1 function, and in KMT2A-rearranged leukaemia under a menin inhibitor that is the direction of the documented resistance lesion. Drugging this node as currently possible is therapeutically backwards for this disease.
• TRIM27 knockdown rescuing USP7-inhibitor sensitivity means that TRIM27-deficient cells would be resistant to the one PRC1.1-directed drug class that exists, which is a resistance mechanism for an intervention nobody has yet taken into the clinic.
• The complex membership rests on one laboratory; building a therapeutic argument on it before independent reproduction would repeat a pattern this project has been caught by before.
• TRIM27 has substrates outside Polycomb, so any TRIM27-directed intervention would not be a clean PRC1.1 intervention any more than a USP7 inhibitor is.

1. Has anyone reproduced TRIM27 as a PRC1.1 component outside the laboratory that reported it? No independent confirmation was found, and the two other groups working on USP7-PRC1.1 did not name it.
2. Would a TRIM27 inhibitor stabilise PRC1.1 and raise H2AK119ub? That is the only pharmacological direction that would help in KMT2A-rearranged disease under a menin inhibitor, and no such compound is known.
3. Which PRC1.1 subunit does TRIM27 ubiquitinate? The model requires a substrate inside the complex and the abstract does not name one; without it the mechanism is an epistasis rather than a biochemistry.
4. Does TRIM27 status vary in KMT2A-rearranged patients? It is not on routine sequencing panels and its loss would be predicted to stabilise PRC1.1 - potentially a sensitising rather than resistance lesion, which would be the first of that kind in this axis.

1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   USP7 and TRIM27 are integral components of PRC1.1; the two are reciprocally required for incorporation into the complex; TRIM27 knockdown partially rescues USP7-inhibitor sensitivity; and the proposed model is that USP7 counteracts TRIM27 E3 ligase activity, thereby maintaining PRC1.1 integrity and function.

   The USP7-TRIM27 axis mediates non-canonical PRC1.1 function and is a druggable target in leukemia. · Maat H, Atsma TJ, Hogeling SM, Rodriguez Lopez A, Jaques J, Olthuis M, et al. · iScience 2021
   PMID 34113809DOI
   ▸What this evidence cannot show

   Single laboratory, not independently reproduced. The rescue of USP7-inhibitor sensitivity by TRIM27 knockdown is explicitly PARTIAL, which is consistent with TRIM27 being one of several routes by which USP7 inhibition acts - and USP7 has many substrates outside Polycomb. The model of USP7 counteracting TRIM27 is stated by the authors as a proposal.

   ▸Evidence that argues against this (1)

   • An independent genome-wide CRISPRi screen that recovered multiple ncPRC1 components - USP7, KDM2B, RING1 and PCGF1 - did not name TRIM27 among them. That is an absence in a hit list rather than a refutation, and screens miss genes for library and redundancy reasons, but the screen that found the partner did not find this subunit. Gobel C et al. 2026
2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   An independent EZH2-inhibition-anchored genome-wide CRISPR interference screen in GCB-DLBCL identified non-canonical PRC1 components USP7, KDM2B, RING1 and PCGF1 as modifiers of EZH2-inhibitor sensitivity.

   DOT1L Shapes ncPRC1-Target Gene Repression to Maintain Germinal Center B Cell Identity of Diffuse Large B cell Lymphoma. · Gobel C, Niccolai R, Gregoricchio S, de Groot MHP, Kneefel SFK, Kreft M, et al. · Blood 2026
   PMID 42213647DOI
   ▸What this evidence cannot show

   GCB-DLBCL, EZH2-anchored, abstract-level. Cited here for what the hit list contains and does not contain, which is a weak form of evidence about TRIM27 and is stated as such. A screen's hit list is not a complex roster.

   ▸Evidence that argues against this (1)

   • The original PRC1.1 proteomics places TRIM27 in the complex by co-precipitation, which is a direct physical measurement and is not contradicted by a functional screen failing to call the gene. Physical membership and functional modifier status are different questions. Maat H et al. 2021
3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   In neuroblastoma, global proteomics and ubiquitinomics show that USP7 inhibition disrupts non-canonical PRC1 complex assembly, resulting in destabilisation of PRC1.1 and subsequent proteolysis, with reduced H2AK119Ub deposition on chromatin.

   Inhibition of USP7 Destabilizes the Noncanonical PRC1.1 Complex and Induces Neuroblastoma Differentiation. · Cmarik EA, Wahi A, Chandekar SS, Otchere SK, Sturgis JE, Salhotra S, et al. · Mol Cancer Res 2026
   PMID 41855552DOI
   ▸What this evidence cannot show

   A different disease and lineage, and the abstract does not name the E3 ligase responsible for the proteolysis. It independently confirms the USP7-stability-of-PRC1.1 relationship by an orthogonal method and leaves the TRIM27 attribution untested. Abstract-level.

   ▸Evidence that argues against this (1)

   • That an orthogonal proteomic and ubiquitinomic analysis reached the same destabilisation phenotype without identifying TRIM27 is a mild argument that the ligase responsible may differ by context, or that several ligases contribute - the partial nature of the published TRIM27 rescue is consistent with either. Maat H et al. 2021
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProt reviewed human record: TRIM27 is P14373, entry name TRI27_HUMAN, primary gene name TRIM27, synonyms RFP and RNF76, 513 aa, recommended protein name 'Zinc finger protein RFP' with alternative names 'Ret finger protein', 'RING finger protein 76' and 'Tripartite motif-containing protein 27'.

   UniProtKB reviewed human record for TRIM27 (P14373), resolved by gene_exact:TRIM27 AND organism_id:9606 AND reviewed:true with gene_primary and length read back · UniProt Consortium · UniProtKB 2026
   ▸What this evidence cannot show

   Database fact fetched 2026-09-20; exactly one reviewed row, gene_primary matched. Note the rule-65 hazard in its benign form: the recommended protein name is 'Zinc finger protein RFP' and contains neither 'TRIM27' nor 'tripartite motif' in the primary position, so a reader checking the name back sees no confirmation of the symbol asked for. The retired-style alias RFP is also a common abbreviation for red fluorescent protein, so it must never be used as a literature search term.

   ▸Evidence that argues against this (1)

   • The alias RFP is a precision hazard of exactly the class brief rule 62's septin discussion describes in its second half - a short alias that another field uses heavily will inflate any count built on it. A TRIM27 literature search over aliases must title-check, and this record's alias list is the reason to expect noise rather than a licence to search on it. Maat H et al. 2021
5. E_q95_trim27_loc1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   UniProt P14373's location-to-citation mapping is INVERTED in the sharpest way this project has recorded. Read back from the raw JSON on 2026-09-22: 'Early endosome' carries ECO:0000269 PubMed 1437549, whose title is 'RFP is a DNA binding protein associated with the nuclear matrix'; 'Nucleus' carries a single ECO:0000269, PubMed 17156811, a TRIM5alpha comparative virology paper whose abstract states that the TRIM proteins studied co-localized in the CYTOPLASM with TRIM5alpha and names no nuclear compartment; and 'Nucleus, PML body' - the only subnuclear body in the record - carries a bare ECO:0000250 with NO SOURCE AT ALL, although a 1998 paper titled for exactly that finding exists and is not cited.

   UniProtKB P14373 (TRI27_HUMAN), SUBCELLULAR LOCATION comment with every location-to-citation mapping read back from the raw JSON, fetched live 2026-09-22 · UniProt Consortium · Databases 2026
   ▸What this evidence cannot show

   This is a statement about the database record. The full text of PubMed 17156811 is unreachable - Europe PMC returns pmcid null and isOpenAccess N so routes 2 to 4 are structurally unavailable, and DOI resolution returns an 11-character Elsevier redirect stub - so this entry does NOT claim that paper contains no TRIM27 observation, only that no reachable text supports a nuclear one and that its abstract reports the opposite compartment. PubMed 1437549 by contrast was read in full (PMC334335) and is unambiguous about the nuclear matrix.

   ▸Evidence that argues against this (1)

   • THE CHARITABLE READING, WHICH IS PARTLY RIGHT AND DOES NOT RESCUE THE RECORD. PubMed 1437549 does contain a plausible route to an endosomal-adjacent call - it reports RFP in four human and rodent cell lines and a unique 68 kDa species in testis, so a curator reading the fractionation could in principle have taken something cytoplasmic from it. And PubMed 9247190, co-cited on the same endosome line, is a genuine subcellular-compartmentalisation paper: mutations in the RING finger and B box change where rfp sits. So the endosome line is not fabricated. What is not defensible is the ordering: the paper whose TITLE is a subnuclear assignment is filed under a cytoplasmic organelle, the only subnuclear term in the record has no citation, and the nuclear term points at a paper about a different TRIM protein in the cytoplasm. Brief rule 91 says an evidence code tells you a curator saw an experiment; this record shows it does not even tell you the compartment matched. Cao T et al. 1997
6. E_q95_trim27_loc2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   TRIM27/RFP's subnuclear address is the PML NUCLEAR BODY, established directly and in a leukaemic system: rfp co-localizes with PML in a subset of PML nuclear bodies and interacts directly with PML through its B-box and the distal two coils, the association is altered by mutations that disrupt the rfp/PML interaction AND in NB4 cells derived from acute promyelocytic leukaemia patients, retinoic acid treatment makes rfp re-associate with the bodies in a non-APL pattern, and rfp co-localizes with the PML-RARA fusion protein itself.

   Ret finger protein is a normal component of PML nuclear bodies and interacts directly with PML. · Cao T, Duprez E, Borden KL, Freemont PS, Etkin LD · J Cell Sci 1998
   PMID 9570750DOI
   ▸What this evidence cannot show

   1998 immunofluorescence with overexpressed and mutant constructs alongside endogenous staining; the abstract does not separate which panels are endogenous. The compartment is partial by the authors' own words - a SUBSET of PML bodies - so this is a minority-pool assignment, not a wholesale address. Independent corroboration exists in the same compartment from a different angle: rfp co-localises into nuclear bodies with the Int-6/EIF3E gene product (PMID 10504338), and PIAS-mediated SUMO-1 modification changes its intracellular translocation and repressive activity (PMID 15907835).

   ▸Evidence that argues against this (1)

   • THE RESULT CUTS AGAINST THE EASY READING OF ITS OWN RELEVANCE HERE. It is tempting for this graph to treat a PML-body address as a leukaemia result because the system is APL - but what the paper shows is that a FUSION ONCOPROTEIN REORGANISES THE COMPARTMENT: in NB4 cells rfp's association with the bodies is altered, and retinoic acid restores it. That is a statement about PML-RARA's effect on PML bodies, which is textbook, and it says nothing about KMT2A fusions, which do not touch PML. So the transferable content is narrower than 'TRIM27 is a leukaemic nuclear-body protein': it is that TRIM27's subnuclear residence is CONTINGENT on the integrity of a body another oncoprotein can dismantle. Whether a KMT2A fusion perturbs PML bodies at all is unmeasured, and the PML-RARA precedent is the reason it is worth asking rather than a reason to assume it. Morris-Desbois C et al. 1999
7. E_q95_trim27_loc3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   TRIM27/RFP is also strongly associated with the NUCLEAR MATRIX, and the measurement was made in a human B-lineage cell: using AH7974 rat ascites hepatoma and Raji human Burkitt lymphoma cells, RFP - identified as a 58 kDa protein in four human and rodent cell lines - shows strong association with the nuclear matrix, and RFP solubilised from the nuclear matrix has DNA-binding activity with a preference for double-stranded over single-stranded DNA.

   RFP is a DNA binding protein associated with the nuclear matrix. · Isomura T, Tamiya-Koizumi K, Suzuki M, Yoshida S, Taniguchi M, Matsuyama M, Ishigaki T, Sakuma S, Takahashi M · Nucleic Acids Res 1992
   PMID 1437549DOI
   ▸What this evidence cannot show

   1992 biochemical fractionation with a polyclonal antibody raised against a bacterially expressed cDNA fragment, and nuclear-matrix preparations are operationally defined by extraction conditions rather than by a microscopically visible structure - this graph's own localization-nuclear-matrix-mitotic-scaffold-kmt2a entry exists because that definitional problem recurs. No fluorescence image, no marker co-stain, no modern replication: TRIM27 or RFP against 'nuclear matrix' returns only three records in total, of which this is the only fractionation study.

   ▸Evidence that argues against this (1)

   • ARGUING THAT THE NUCLEAR-MATRIX CALL AND THE PML-BODY CALL MAY BE THE SAME OBSERVATION SEEN TWICE, WHICH WOULD WEAKEN BOTH AS INDEPENDENT FACTS. PML nuclear bodies are classically resistant to the nuclease-and-salt extraction that defines a nuclear-matrix preparation, so a protein docked in PML bodies is expected to partition with the matrix. If that is what happened in 1992, the two assignments are one address measured by two methods rather than two pools - and the entry should not be read as giving TRIM27 two subnuclear compartments. The experiment that separates them is trivial and has never been done: co-stain endogenous TRIM27 for PML in a matrix preparation and ask whether matrix-retained TRIM27 is PML-body TRIM27. Note the cell type is nonetheless worth keeping - Raji is a human B-lymphoid line, and in a graph about B-lineage disease it is the only B-lineage cell in this protein's entire localisation literature. Cao T et al. 1998
8. E_q95_trim27_loc4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   The cytoplasmic pools of TRIM27 are the better-evidenced ones and they are compartment-specific rather than diffuse: TRIM27 is recruited to retromer-containing endosomes through interaction with MAGEL2 where it regulates WASH-dependent actin polymerisation, it co-localises with p62/SQSTM1 and TBK1 in cytoplasmic structures closely associated with mitochondria, it is a mitochondrial protein by a separate ECO:0000269, and its distribution between compartments is set by the tripartite RBCC motif - mutations in the RING finger and the B box alter its subcellular compartmentalisation in several cell lines.

   Involvement of the rfp tripartite motif in protein-protein interactions and subcellular distribution. · Cao T, Borden KL, Freemont PS, Etkin LD · J Cell Sci 1997
   PMID 9247190DOI
   ▸What this evidence cannot show

   The RBCC-mutant result is 1997 work on transfected constructs and 'various cell lines' are not named in the abstract; the retromer and mitophagy results are modern and specific but in non-haematopoietic systems. What is NOT established anywhere is a quantitative partition - no source measures what fraction of endogenous TRIM27 is nuclear versus cytoplasmic in any single cell type, and UniProt's own note concedes the problem in the words 'Nuclear or cytoplasmic depending on the cell type (By similarity)'.

   ▸Evidence that argues against this (1)

   • THIS IS WHERE THE DISPATCH BRIEF FOR THIS TRANCHE WAS WRONG, AND THE LITERATURE SETTLES IT AGAINST THE BRIEF. The brief stated that gene-usp8 and gene-trim27 'are a DUB and a TRIM ligase that work together at endosomes'. Measured 2026-09-22: TRIM27[tiab] AND USP8[tiab] returns ZERO, and the same query over USP8 and UBPY against TRIM27 and the spelled-out phrase ret finger protein also returns ZERO, against an operand control of 16 for TRIM27[tiab] AND USP7[tiab] - so the instrument returns ones. TRIM27's deubiquitinase partner is USP7, not USP8: USP7 and TRIM27 are reciprocally required components of PRC1.1 and TRIM27 knockdown partially rescues USP7-inhibitor sensitivity, and the TRIM27-USP7 complex regulates TNF-alpha-induced apoptosis. This entry's own title and partner field already say USP7 correctly; the conflation was in the brief, and the two DUBs are one digit apart in a graph that contains entries for both. Zaman MM et al. 2013
9. E_q95_trim27_loc5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   Bounded 2026-09-22: no localisation measurement of TRIM27 exists in any KMT2A-fusion context, and the search is hazardous in a way that has to be reported with the count. The gene clause over TRIM27, 'ret finger protein' and RNF76 returns 265 records - PubMed drops RNF76 entirely as a phrase not found - and intersects the nine-term KMT2A/MLL fusion clause at 1; with a localisation clause the gene clause returns 45 and with the fusion clause added 0; with a subnuclear clause it returns 7 and with the fusion clause added 0. The retired symbol RFP is unusable: RFP[tiab] alone returns 4,037 records, 15-fold the current symbol, dominated by RED FLUORESCENT PROTEIN.

   NCBI E-utilities esearch db=pubmed; every count stored with the querytranslation read back, in data.localization_bounded_absence · NCBI E-utilities · Databases 2026
   ▸What this evidence cannot show

   Titles and abstracts only; supporting figures not searched (brief rule 80). The RFP measurement makes this a per-gene case of brief rule 69 part 1 rather than a rule: searching the retired symbol here would INFLATE a count fifteenfold with noise, which is the opposite failure from the SEPT9 case the brief opens with, and identical in shape to the NELFE/'RD' and EAF2/'U19' inflations already recorded. The correct handling is to search RFP only inside a tight AND with a specific compartment term, which is how the three nuclear-matrix records were found.

   ▸Evidence that argues against this (1)

   • THE ZERO IS HONEST BUT IT IS NOT THE INTERESTING ABSENCE, AND SAYING SO IS THE POINT. TRIM27 reaches this project through PRC1.1, and the one study that puts TRIM27 and a KMT2A fusion in the same experiment is a USP7-inhibitor study whose leukaemia arm is an MLL-AF9 xenograft - a requirement and complex-integrity measurement with no compartment in it. So what is missing is not a paper somebody forgot to write but an assay nobody has run: whether the PRC1.1 destabilisation caused by USP7 inhibition moves TRIM27 out of PML bodies, out of Polycomb bodies, or neither. That is one immunofluorescence experiment in a cell line this project already works in, and it would be the first subnuclear measurement of a PRC1.1 stability switch in any leukaemia. Maat H et al. 2021

• binds →USP7 - the enzyme that holds PRC1.1 together, and the only pharmacological handle on this complex that exists today

  reciprocally required for incorporation into PRC1.1; USP7 counteracts TRIM27 E3 activity

• part of →PRC1.1 — the complex this graph has been leaning on, now enumerated subunit by subunit, and the enumeration changes the conclusion

  reported integral component; single-laboratory result, not independently reproduced

• regulates →H2AK119ub - the mark every PRC1.1 claim in this graph rests on, named in fifteen entries and written in none until now

  destabilising arm of the PRC1.1 stability switch; its activity lowers complex integrity and therefore the mark

• ← supportsQ50 enumeration: every query this beat ran, every zero it recorded with its bound, and the papers it read and did not use

  named once in the graph before this beat, in the PRC1.1 roster, with no entry of its own

• related to →PML nuclear bodies and PML::RARA — the best-worked case of a fusion destroying a nuclear compartment, and what it actually predicts for KMT2A

  TRIM27/RFP is a NORMAL COMPONENT of PML nuclear bodies and binds PML directly through its B-box and distal two coils; the association is altered in APL-derived NB4 cells, restored by retinoic acid, and rfp co-localises with PML-RARA itself (PMID 9570750). UniProt names this compartment and cites NOTHING for it - bare ECO:0000250, no source (Q95 2026-09-22)

• related to →Nuclear matrix and mitotic chromosome scaffold — KMT2A sits with topoisomerase II, in the same place the gene breaks

  TRIM27/RFP shows strong nuclear-matrix association by fractionation in Raji human Burkitt lymphoma and rat hepatoma cells, and matrix-solubilised RFP binds double-stranded DNA (PMID 1437549). UniProt files this paper under EARLY ENDOSOME. Caveat carried on the entry: PML bodies survive matrix extraction, so this and the PML-body call may be one address measured twice (Q95 2026-09-22)

• related to →Polycomb bodies — indirect evidence via CBX8, undercut by the finding that CBX8-PRC1 condensates are dissolved rather than built by the rest of PRC1

  THE COMPARTMENT NOBODY HAS LOOKED IN. TRIM27 is the destabilising E3 inside PRC1.1, and tranche 4 established Polycomb bodies as the PRC1 address for RING1 by FRAP; no source has asked whether TRIM27 is in a Polycomb body, or whether USP7 inhibition moves it out of one. One immunofluorescence experiment in a cell line this project already uses (Q95 2026-09-22)

• related to →USP7 - the enzyme that holds PRC1.1 together, and the only pharmacological handle on this complex that exists today

  CORRECTION TO THIS TRANCHE'S DISPATCH BRIEF, measured 2026-09-22: TRIM27's DUB partner is USP7, not USP8. TRIM27[tiab] AND USP8[tiab] = 0 and (USP8 OR UBPY) AND (TRIM27 OR 'ret finger protein') = 0, against an operand control of 16 for TRIM27 AND USP7. The graph had it right; the brief conflated two DUBs one digit apart (Q95 2026-09-22)

• related to →PML::RARA, t(15;17)(q24;q21)

  The only fusion oncoprotein whose effect on TRIM27's compartment has been measured, and it is not a KMT2A fusion: in NB4 cells rfp's association with PML bodies is altered and retinoic acid restores it, and rfp co-localises with PML-RARA (PMID 9570750). Read as a precedent for asking the question of KMT2A fusions, not as evidence about them (Q95 2026-09-22)

• ← related toUSP7 - the enzyme that holds PRC1.1 together, and the only pharmacological handle on this complex that exists today

  partner in PRC1.1; its title once carried the gap-marker that produced the claim USP7 was never written here, corrected in commit 4db6b99