PARP1 - the target whose inhibitors this disease was specifically shown to resist, and which three later papers combine anyway

A naming fact. gene_exact:PARP1 AND organism_id:9606 AND reviewed:true returned exactly one row with gene_primary PARP1. The alternative name ARTD1 is used in part of the literature and is NOT in this record's gene-synonym field, so a symbol-keyed search misses it - one of the 8 records returned by (MLL OR KMT2A) AND PARP1 uses ARTD1 in its title.

• The paralogue is the standing hazard: clinical PARP inhibitors are not PARP1-selective. Olaparib, niraparib, rucaparib and talazoparib all inhibit PARP1 and PARP2, so every result in this entry that comes from a compound is a PARP1/PARP2 result. Q9UGN5 is 583 aa against P09874's 1014, which makes an accession swap detectable but does nothing about the pharmacological ambiguity. UniProt Consortium 2026

Mouse and human AML models. 'MLL fusions with dominant transactivation ability' is a functional class; the infant lymphoid index disease KMT2A::AFF1 is not the setting tested, and the HOXA9 dependency that carries the mechanism is variable across KMT2A-r subsets - this graph holds HOXA-low cases as a distinct group.

• The same group subsequently reported that olaparib potentiates conventional chemotherapy and DNMT inhibitors in MLL leukaemia cells, with single-agent treatments having little impact. Single-agent insensitivity and combination potentiation are compatible, and the later paper does not withdraw the resistance finding - but an entry citing only one of the two would mislead in either direction. Zhao L et al. 2017

The PubMed abstract for this record consists only of the two bullet-point 'key points' and carries no methods, no models and no numbers; the full text was not retrieved in this beat. Everything asserted here is therefore at the level of the key points. MLL-AF9 is myeloid and is not the index fusion.

• A maintenance requirement for PARP1 in a leukaemia is not evidence of a synthetic-lethal vulnerability and does not contradict the finding that the same disease class is insensitive to PARP inhibition as monotherapy. The claim in this record is explicitly about combination with cytotoxic drugs. Esposito MT et al. 2015

Colony and growth assays in mouse primary cells and one human MLL-AF9 line; no in vivo arm described in the abstract; myeloid. The authors' own framing is potentiation, and they state that single agents had little impact - which is a restatement of the monotherapy insensitivity rather than a contradiction of it.

• Gamma-H2AX focus counts are the mechanistic evidence, and gamma-H2AX is induced by replication stress and apoptotic fragmentation as well as by double-strand breaks. A DNMT inhibitor plus a PARP inhibitor perturbs replication directly, so the focus increase is a weaker inference about break burden than it reads. Kosik P et al. 2021

HEK293 with an AAVS1 knock-in; not haematopoietic. It is a protein-level and functional-capacity negative, not a measurement of PARP trapping or of sensitivity to a PARP inhibitor, which the study did not test.

• Unchanged PARP1 protein level says nothing about PARP-inhibitor sensitivity, because the dominant cytotoxic mechanism of clinical PARP inhibitors is trapping of PARP1 on DNA rather than catalytic inhibition, and trapping depends on the lesion context rather than on enzyme abundance. The negative is about repair capacity and is being used here only for that. Esposito MT et al. 2015

Recorded from the PubMed title only; neither the abstract nor the full text was read in this beat, and no numbers, models or mechanisms are asserted. Confidence weak accordingly. It is included because it is the only record found treating PARP1 as a regulated node rather than as a drug target in this disease, and a future beat should read it rather than re-find it.

• A microRNA-to-acetylation-to-drug-sensitivity chain has three inferential steps and single-microRNA mechanism papers in leukaemia have a poor replication record. Nothing in this graph should be built on it until the full text is read. Esposito MT et al. 2015

A curated composite drawn from more than twenty papers across four decades and many cell systems, none haematopoietic in the rows examined here. The evidence codes say a curator saw experiments; rule 91 and E_q89_parp1_loc3 below show that for at least the nucleolus row the cited paper is not about the fact.

• It is tempting to read twelve ECO:0000269 attributions for 'Chromosome' as overwhelming evidence, and the number is doing less work than it looks. Nine of those twelve are cited in the note specifically for localisation TO SITES OF DNA DAMAGE - an induced, sub-second, lesion-proximal recruitment - and only one (PubMed:15607977) is cited for binding normal undamaged chromatin. So a count that reads as 'twelve independent demonstrations that PARP1 lives on chromatin' is largely twelve demonstrations that PARP1 GOES to broken chromatin, which is a different claim and the one this entry is at pains to separate. The constitutive chromatin association rests on far fewer citations than the row's evidence count suggests, and a reader grading confidence by attribution count would get this backwards. Rancourt A et al. 2009

Abstract read, full text not read; the cell system is not named in the abstract and is therefore not named in the claim (rule 73). The 40% figure is quoted from the authors' own framing of their earlier work rather than re-derived here, and no error bar or method is given for it in the abstract. Not haematopoietic, no KMT2A fusion, and camptothecin/MNNG are pharmacological rather than physiological stresses.

• The entry's own framing - constitutive nucleolar pool, damage-induced exit - may have the causality inverted, and the paper cannot exclude it. Camptothecin is a topoisomerase I poison that generates replication-associated breaks AND is well known to disrupt nucleolar organisation directly; the nucleolus is famously the first structure to reorganise under almost any stress. So 'PARP-1 leaves the nucleolus in response to damage' and 'the nucleolus dissolves and PARP-1 goes with it' predict the same observation, and the second does not require PARP-1 to be a stress sensor at all. The finding that resists this reading is the functional one: prior delocalisation changed the CONSEQUENCE of later activation - more automodification, more apoptosis, with repair activity and NAD+ consumption unchanged - which is a specific, non-trivial result that bulk nucleolar disintegration does not obviously predict. The compartment claim and the sensing claim should therefore be graded separately, and this entry grades only the compartment. UniProt Consortium 2026

Bounded by what is readable. The abstract was read in full. The Europe PMC fullTextXML route returned HTTP 500 for PMC147507 and the PMC HTML page returns 18,257 characters that contain ZERO occurrences of 'nucleol' and only one each of 'periphery' and 'confocal' - i.e. the 1998 article body is served as a scanned PDF and the text is not reachable, so a nucleolar observation could exist in the figures. The correct statement is therefore 'the accessible text of the cited paper reports nuclear periphery and not nucleolus', graded moderate for that reason. This does NOT challenge PARP1's nucleolar pool, which is independently established (E_q89_parp1_loc2).

• The obvious defence is that the nucleolus abuts the nucleoplasm and a 1998 confocal image of 'nuclear periphery' might have meant the perinucleolar region, so the curator may have read the figures correctly and this beat, unable to see them, is second-guessing. That defence is plausible and cannot be refuted from outside the paper - which is why this item is moderate rather than strong. But the alternative reading is worse for the annotation, not better: 'nuclear periphery' in confocal usage overwhelmingly means the nuclear ENVELOPE/lamina region, which is close to the opposite of the nucleolus in both position and biology, and this graph maintains a separate localization-nuclear-envelope-proximity-kmt2a-amino-terminus entry on the strength of that distinction. If the curator did mean the envelope and filed it as nucleolus, a wrong compartment has been propagated; if the curator meant perinucleolar, the annotation is right and its wording is misleading. Either way a reader cannot tell from the record, and that is the defect. UniProt Consortium 2026

The fusions studied are MLL fusions with dominant transactivation ability - the paper's own phrase - and the comparator leukaemias are AML1-ETO/RUNX1-RUNX1T1 and PML-RARA. This is myeloid and largely murine-model work; it is NOT a dedicated human KMT2A::AFF1 experiment, and this project's standing finding is that 0 of 24 sampled load-bearing experiments are. This item re-states the entry's existing E2 with re-fetched identifiers, and is added here because the cell-state field needs a citation of its own; the identifiers were taken from a fresh PubMed efetch rather than copied from the sibling block.

• This entry's own later evidence argues against reading the negative as final, and the tension should be visible rather than resolved by preferring the first paper. PARP1 is reported REQUIRED for maintenance of MLL-AF9 leukaemias with inhibitors enhancing cytotoxic drugs (this entry's E3), and olaparib plus DNMT inhibitors or chemotherapy reduced colony formation in primary mouse leukaemia and a human MLL-AF9 line (E4). Those are not contradictions of Esposito if read carefully - they are COMBINATION results and a maintenance requirement, against a MONOTHERAPY sensitivity claim - but they do mean 'PARP inhibition does not work in MLL leukaemia' is too broad, and the entry's title already concedes this by saying three later papers combine anyway. The localisation dimension adds a specific reason the question may not be closed: nobody has asked whether PARP1's nucleolar fraction, or its chromatin residence time, differs in a KMT2A-fusion blast, and a DDR-proficient cell with a differently partitioned PARP1 could respond differently to trapping agents than to catalytic inhibitors. Rancourt A et al. 2009