DOT1L's localization record is a mouse kidney paper wearing a human label - and the one image of it in a leukaemia cell is a supporting figure a title search cannot see

A database record, not an experiment. Resolution followed the brief's identifier rules: gene_exact:DOT1L AND organism_id:9606 AND reviewed:true returned exactly one reviewed record whose gene_primary is DOT1L, accession Q8TEK3, length 1537, matching the length already asserted in gene-dot1l. Evidence codes were read from the JSON evidences array rather than from the flat text, because the flat text renders both codes identically to a casual reader.

• A by-similarity annotation is not wrong, it is unmeasured. Mouse and human DOT1L are 1540 and 1537 residues and the AF17-binding region is conserved, so the transfer is reasonable. The failure mode this entry guards against is not the annotation - it is an automated import writing 'DOT1L is cytoplasmic' into a human leukaemia entry with no marker that the observation is murine, renal and overexpression-driven.

Mouse renal collecting-duct epithelium, not human and not haematopoietic. The cytoplasmic pool is produced by OVEREXPRESSING AF17, so it reports what AF17 can do to Dot1a, not where endogenous Dot1a sits at native stoichiometry. Leptomycin B is a covalent CRM1 inhibitor with pleiotropic effects on nucleocytoplasmic traffic. Read: abstract only (PubMed efetch, 2026-09-20).

• The converse experiment - whether AF17 loss changes DOT1L localisation in a leukaemia cell, where AF10/MLLT10 rather than AF17/MLLT6 is the dominant DOT1L partner - has not been done. MLLT6 and MLLT10 are paralogous octapeptide-motif/leucine-zipper DOT1L binders, and it is not established that the export behaviour is a property of AF17 specifically rather than of that binding mode, which would make it relevant to AF10-containing fusions as well.

Mouse renal epithelium and an aldosterone-signalling question; the colocalisation is nuclear-versus-cytoplasmic resolution, not subnuclear. AF9 here is acting as a transcriptional repressor partner in kidney, the opposite of its role in the leukaemic SEC, so this system should not be used to reason about the leukaemic complex. Read: abstract only (PubMed efetch, 2026-09-20).

• The same Dot1a-AF9 pairing that this paper shows repressing ENaCalpha in kidney is, in leukaemia, read as an activating recruitment module. A protein pair can colocalise in the nucleus of two cell types and do opposite things; nuclear colocalisation is therefore weak evidence about mechanism even when it is strong evidence about compartment.

Mouse N2a neuroblastoma, transient transfection, DOT1L OVEREXPRESSED and FLAG-tagged - so the imaged species is not endogenous DOT1L and the compartment assignment is of an interaction, not of the protein alone. Antibodies used (rule 74): FLAG goat ab95045, NPM1 mouse ab10530, fibrillarin rabbit ab184817, lamin B1 rabbit ab133741, H3K79me2 rabbit ab3594. A PLA signal reports proximity within roughly 40 nm, not binding, and its compartment assignment inherits the resolution of confocal microscopy. Read: full text via Europe PMC PMC10537513, 2026-09-20.

• The paper's own negative bounds the claim: DOT1L overexpression and DOT1L catalytic inhibition both left NPM1 localisation unchanged. So whatever DOT1L is doing at the peri-nucleolar compartment, it is not organising NPM1, and the heterochromatin reorganisation runs through H3K27me3 rather than through a DOT1L-dependent relocation of the nucleolar scaffold. A reader should not take 'DOT1L is peri-nucleolar' as 'DOT1L builds the peri-nucleolar compartment'.

KNOWN BUT UNREADABLE, recorded per the brief's third case. Routes tried 2026-09-20: (1) Europe PMC core for EXT_ID:15851025 returns no pmcid; (2) cross-check of efetch db=pubmed for <ArticleId IdType="pmc"> also returns none, so routes 2-4 cannot be attempted; (5) curl -sL https://doi.org/10.1016/j.cell.2005.02.020 with a browser User-Agent returned 11 stripped characters ('Redirecting'), the documented Elsevier stub. This entry therefore does NOT assert what the paper's localisation data show; it asserts that the annotation has a source and that the source was not read here.

• The unread source could well contain a clean human immunofluorescence result that would make the nuclear annotation stronger than this entry treats it. The honest position is symmetric: not reading a paper is not evidence against it. What the entry can say is that the nuclear annotation cannot be audited from open sources, and that a graph relying on it should say so.

QUERY: `(DOT1L[tiab] OR DOT1-like[tiab] OR hDOT1L[tiab] OR Dot1a[tiab]) AND (MLL[tiab] OR KMT2A[tiab] OR leukemia[tiab] OR leukaemia[tiab]) AND (immunofluorescence[tiab] OR confocal[tiab] OR "subcellular localization"[tiab] OR "subnuclear"[tiab] OR "nuclear localization"[tiab])` = 1 record, PMID 38374003. RULE 67 POSITIVE CONTROL through the same shape: `DOT1L[tiab] AND (immunofluorescence[tiab] OR confocal[tiab])` = 12, so the instrument returns ones. NAME LIST searched: DOT1L, hDOT1L, DOT1-like, Dot1a; KMT4 not searched. RULE 80, AND THIS IS WHY THE FIRST VERSION OF THIS ITEM WAS WRONG: the single hit was title-checked and dismissed as off-topic, and its FULL TEXT contains the measurement. A PubMed count is bounded by abstracts, not by papers, and a localisation panel is almost always a supporting figure in a paper about something else. The narrowed claim above is bounded by abstracts PLUS one full text; other supporting figures in the wider DOT1L literature were not searched, so even the narrowed claim should be read as 'not announced anywhere and not found in the one paper I opened'.

• E7 is the direct counter to the original version of this item and is recorded as evidence rather than as a note, because it is a positive finding. The general lesson is the transferable part: this graph's absence claims about IMAGING are systematically more fragile than its absence claims about other things, because imaging is rarely the subject of a title.

This is a SUPPORTING figure (Fig. 4h) in a paper about an lncRNA and autophagic degradation of MLL fusion proteins; localisation is not its subject and it is not announced in the title or abstract, which is why an abstract-bounded search missed it. RULE 74: DOT1L primary antibody CST #90878S (rabbit), detected with DIG-FITC (Abcam ab119349) and goat anti-rabbit DyLight 647 (Abcam ab150115), imaged on a Zeiss LSM7 DUO NLO confocal; no DOT1L-knockout or knockdown specificity control for the IF is described. The readout is co-localisation with an RNA FISH signal at one locus, so it establishes that DOT1L is nuclear and present at LAMP5 - it is NOT a compartment survey, there is no nucleolar, speckle, PML or lamina marker, and it says nothing about a cytoplasmic pool. RULE 73: THP-1 is a human MONOCYTIC line carrying KMT2A::MLLT3, not infant B-lineage KMT2A::AFF1; the paper's other lines are MV4-11 (KMT2A::AFF1) and MOLM-13 (KMT2A::MLLT3), but the FISH/IF is stated as THP-1. Read: full text via Europe PMC PMC10877858, 2026-09-20.

• A co-localisation panel designed to show an RNA-protein-locus relationship is optimised for that signal, not for reporting the protein's distribution: exposure and thresholding chosen to display a focal overlap routinely suppress diffuse and low-abundance pools, including a cytoplasmic one. So this figure cannot be used to argue that DOT1L is ONLY nuclear in a KMT2A-rearranged cell, which is the question the by-similarity cytoplasmic annotation actually raises.