PIP4K2A - the lipid kinase at a genome-wide childhood-ALL susceptibility locus, whose product is the ligand a PHD finger reads

No three-dimensional model is available for this protein, so the map above is the whole picture. That is itself informative: AlphaFold does not model chains this long, and the parts that have been crystallised are held under the entries for the individual domains.

• A germline susceptibility allele is not a therapeutic target. The 10p12 association is an inherited risk factor measured in cohorts, the effect size is OR ~1.3, and there is no evidence that PIP4K2A protein or activity is elevated or required in an established blast.
• PIP4K2A CONSUMES PtdIns5P, so inhibiting it RAISES the lipid. If PtdIns5P competes for PHD-finger occupancy, a PIP4K2A inhibitor would perturb chromatin reading in a direction nobody has measured, and the direction cannot be predicted from a knockout screen in which the gene has no phenotype at all.
• PIP4K2A is not a dependency in any leukaemia line (mean gene effect +0.0526 across 1,208), and it has two close paralogues, one of which controls its nuclear pool, so single-agent inhibition is likely to be buffered and a single-gene screen cannot see the pathway.
• PtdIns(4,5)P2 is central to plasma-membrane signalling, endocytosis and cytoskeletal regulation in every cell, so a systemic PIP4K inhibitor carries an essential-function liability far outside the nucleus - and the only tissue with documented high expression is brain, which is the compartment infant ALL therapy already has to protect.

1. Does PIP4K2A depletion raise nuclear PtdIns5P and change ING2 or PHD-finger occupancy at KMT2A::AFF1 target loci? The lipid arm and the chromatin arm of this mechanism have never been measured in the same cell, let alone in a leukaemia cell.
2. Is the 10p12 childhood-ALL risk allele acting through PIP4K2A or through BMI1? The locus names both, BMI1 is a PRC1 core subunit already in this graph, and no fine-mapping was found - so the graph currently holds a susceptibility locus whose causal gene is unassigned.
3. Is the PIP4K2A risk allele present at the expected frequency in infant t(4;11) ALL? Infant KMT2A-rearranged ALL carries the lowest burden of common susceptibility alleles of any ALL subtype, and no published stratum for this locus exists.
4. Does KMT2A's own PHD3 bind a phosphoinositide? ING2's PHD finger does; nobody has tested the PHD finger this project's menin-interface reasoning depends on, and the PHD fingers are lost in the fusion, so a positive result would matter for full-length KMT2A rather than for the chimera.
5. Is the five-bait pattern one SEC contact or five? A PIP4K2A pulldown after AFF4 depletion would collapse five rows to zero if it is one contact through AFF4, and leave ELL and MLLT1 standing if it is not.

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

   UniProt P48426 is the single reviewed human PIP4K2A record at 406 aa (46,225 Da) with one PIPK domain spanning 33-405, catalyses PtdIns5P plus ATP or GTP to PtdIns(4,5)P2, carries no chromatin writer, eraser or reader module of any kind, and HGNC:8997 records PIP5K2A as a genuine PREVIOUS OFFICIAL SYMBOL with a previous name that names the opposite substrate and position.

   UniProtKB P48426 (PI42A_HUMAN) features, catalytic activity, subcellular location with evidence codes and tissue specificity; HGNC:8997 prev_symbol and prev_name via rest.genenames.org; retrieved 2026-09-22 · UniProt Consortium / HGNC 2026
   ▸What this evidence cannot show

   Retrieved 2026-09-22. Absence of an annotated chromatin module is absence of an ANNOTATION, not proof of absence. Two isoforms are annotated (P48426-1, P48426-2) and 406 aa is the canonical isoform; a RefSeq length may differ and was not checked, so any coordinate quoted here is on the UniProt canonical frame. UniProt attributes the GTP-dependent activity to PubMed:26774281, which was fetched separately and is Sumita K et al., Mol Cell 2016, on PI5P4K-beta rather than PIP4K2A - so the GTP claim is paralogue-imported in its primary source even though UniProt files it under PIP4K2A.

   ▸Evidence that argues against this (1)

   • UniProt's cited experimental annotations frequently point at a paper that is not about the fact being annotated, and filtering on evidence code keeps the least informative comment while discarding isoform-scoped ones, so a UniProt feature table is a pointer to be checked rather than a source. UniProt Consortium 2025
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   PIP4K2A has an experimentally established nuclear and cytoplasmic pool, and its nuclear localisation is modulated by interaction with its paralogue PIP4K2B - the single ECO:0000269 source behind both of UniProt's experimental location comments for this protein.

   PIP4Kbeta interacts with and modulates nuclear localization of the high-activity PtdIns5P-4-kinase isoform PIP4Kalpha. · Bultsma Y, Keune WJ, Divecha N (3 authors; senior author Divecha N) · Biochem J 2010
   PMID 20583997DOI
   ▸What this evidence cannot show

   A single paper carrying both of UniProt's experimental localisation calls for this protein, so the nuclear assignment rests on one laboratory. The system is not haematopoietic. The finding is explicitly CONDITIONAL - nuclear localisation is modulated by PIP4K2B - which means a bare 'nuclear' field for PIP4K2A loses the dependency that the experiment established. Read as abstract and title; the imaging panels were not read.

   ▸Evidence that argues against this (1)

   • A paralogue-dependent localisation is a claim about a heterodimer and not about PIP4K2A alone, and the same laboratory's framing of PIP4K2B as an intracellular GTP sensor with its own nuclear behaviour means the two isoforms' nuclear pools cannot be attributed separately from a co-expression experiment. Sumita K et al. 2016
3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   The 10p12.31-12.2 locus containing PIP4K2A is a genome-wide significant childhood acute lymphoblastic leukaemia susceptibility locus - index SNP rs7088318, P = 1.1e-11 in a multiethnic GWAS of 1,605 cases and 6,661 controls, independently replicated in European American (P = .001), African American (P = .009) and Hispanic American (P = .04) series - and the locus is annotated BMI1-PIP4K2A, naming both genes.

   Novel susceptibility variants at 10p12.31-12.2 for childhood acute lymphoblastic leukemia in ethnically diverse populations. · Xu H, Yang W, Perez-Andreu V, et al. (27 authors; senior author Yang JJ) · J Natl Cancer Inst 2013
   PMID 23512250DOI
   ▸What this evidence cannot show

   This is a germline association at a LOCUS, not a demonstration that PIP4K2A is the causal gene: the locus is named BMI1-PIP4K2A and BMI1 is a PRC1 core subunit, so a polycomb gene is an equally well-positioned candidate and no fine-mapping or functional assignment is reported here. The cohorts are childhood ALL overall; no KMT2A-rearranged or infant stratum is reported, and infant t(4;11) ALL is the ALL subtype with the lowest burden of common susceptibility alleles, so extension to the index fusion is untested.

   ▸Evidence that argues against this (1)

   • The same 10p12 region reappears in a Down-syndrome ALL genome-wide association study whose four genome-wide significant loci are reported at IKZF1, CDKN2A and elsewhere, which shows that susceptibility architecture at these loci is subtype-dependent and that a locus significant in one ALL population need not be significant in another. Brown AL et al. 2019
4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   A meta-analysis of six independent studies totalling 3,508 ALL cases and 12,446 controls across multiethnic populations confirms association of both top independent PIP4K2A-locus SNPs with ALL risk - rs7088318 OR 1.28 (95% CI 1.20-1.36) and rs4748793 OR 1.29 (95% CI 1.19-1.40).

   Association Between PIP4K2A Polymorphisms and Acute Lymphoblastic Leukemia Susceptibility. · Liao F, Yin D, Zhang Y, et al. (9 authors; senior author Li Y) · Medicine (Baltimore) 2016
   PMID 27149463DOI
   ▸What this evidence cannot show

   A meta-analysis, so it inherits the discovery studies' inability to assign the signal to PIP4K2A rather than to BMI1, and the effect size is modest (OR ~1.3). The paper's own framing notes inconsistent replication in earlier studies. Multiethnic pooling can mask population-specific effects, and no fusion-subtype stratification is reported.

   ▸Evidence that argues against this (1)

   • The same authors' later replication in an all-age Chinese series finds the PIP4K2A signal restricted to childhood patients without a BCR-ABL1 fusion, so the pooled odds ratio averages over strata in which the effect is present and strata in which it is not, and the single pooled number overstates generality. Liao F et al. 2020
5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   In an all-age Chinese ALL replication series (n = 451), single nucleotide polymorphisms at ARID5B, IKZF1, CEBPE and PIP4K2A were significantly associated with ALL susceptibility only in CHILDHOOD patients with no BCR-ABL1 fusion, whereas the GATA3 signal held in adults irrespective of BCR-ABL1 status.

   Validations of Top and Novel Susceptibility Variants in All-Age Chinese Patients With Acute Lymphoblastic Leukemia. · Liao F, Ye Y, Yin D, et al. (15 authors; senior author Lu X) · Front Genet 2020
   PMID 33193587DOI
   ▸What this evidence cannot show

   n = 451 in a single ancestry, so the age- and fusion-stratified subgroups are small and a negative in a stratum is weakly powered rather than a demonstrated absence. The stratification variable is BCR-ABL1 status, NOT KMT2A rearrangement, so this paper says nothing directly about t(4;11); it is cited here because it is the only source read that stratifies this locus by fusion at all.

   ▸Evidence that argues against this (1)

   • An absence of association in a stratum of a 451-patient series is an absence of testing power rather than a negative result, and the discovery GWAS at this locus reported consistent replication across three ancestries in childhood ALL as a whole, so the fusion-restricted reading rests on the weaker of the two studies. Xu H et al. 2013
6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   The PHD finger of ING2 is a nuclear phosphoinositide receptor: the PHD fingers of ING2 and other diverse nuclear proteins bind PtdInsPs in vitro including the rare species phosphatidylinositol 5-phosphate, the ING2 PHD finger interacts with PtdIns5P in vivo, and that interaction regulates ING2's ability to activate p53 and p53-dependent apoptosis - which places the substrate of PIP4K2A on the ligand side of a chromatin-reader module.

   The PHD finger of the chromatin-associated protein ING2 functions as a nuclear phosphoinositide receptor. · Gozani O, Karuman P, Jones DR, et al. (23 authors; senior author Yuan J) · Cell 2003
   PMID 12859901DOI
   ▸What this evidence cannot show

   This paper is about ING2 and PtdIns5P; it does not mention PIP4K2A, and the link to PIP4K2A is through the shared metabolite rather than through any reported PIP4K2A experiment. PIP4K2A CONSUMES PtdIns5P, so the predicted direction is that PIP4K2A activity lowers the competing ligand - a prediction, not a measurement. No PHD-finger phosphoinositide measurement has been made on KMT2A's PHD3 or on any fusion-driven leukaemia cell. The system is not haematopoietic.

   ▸Evidence that argues against this (1)

   • PHD fingers are characterised in the chromatin literature as histone-tail readers, and the subset that binds phosphoinositides is not the same as the subset that reads H3K4me3: the module was shown the same year to couple H3K4me3 recognition to chromatin remodelling in NURF. A lipid-binding property demonstrated on ING2's PHD finger cannot be assumed for KMT2A's PHD3, whose H3K4me3 recognition is the characterised function of that module. Wysocka J et al. 2006
7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   IntAct REST serves twelve binary PIP4K2A rows against core elongation proteins from THREE publications - BioPlex 2.0 (PMID 28514442, HEK293T), BioPlex 3.0 (PMID 33961781, HEK293T) and a 2025 U2OS multimodal map (PMID 40205054) - all MI:0007 anti-tag co-immunoprecipitation typed MI:0914 association and spoke-expanded, with PIP4K2A the BAIT in every row, and the five apparent baits (AFF1, AFF4, ELL, MLLT1, MLLT3) are five subunits of the super elongation complex.

   IntAct REST findInteractions/P48426 (320 indexed interactions), filtered to Q03164/O00255/O75475/Q9UHB7/Q8TEK3/P51825/Q03111/P42568/P55199/P55197, with experimentalRole, detectionMethod, hostOrganism and publicationPubmedIdentifier read per row; retrieved 2026-09-22 · IntAct molecular interaction database (EMBL-EBI) 2026
   ▸What this evidence cannot show

   Retrieved 2026-09-22 and the number will move. The comparison against the read queue is not like-for-like in ONE direction that must be stated: the queue is built on PSICQUIC, which for AFF4 serves 179 rows over 23 PMIDs with a maximum PMID of 39251607, so PSICQUIC carries no PMID 40205054 rows at all - the third publication is a REST-only record and the queue's omission is an instrument limit, not a curation gap. Counting five SEC subunits as five baits double-counts one complex; counting them as one contact assumes the complex was intact in the lysate, which AP-MS cannot show. Every row is an overexpression AP-MS in a non-haematopoietic line.

   ▸Evidence that argues against this (1)

   • Recovery in an affinity-purification or proximity map is dominated by subcellular compartment occupancy rather than by binding: the largest published proximity map of a human cell shows a prey's bait profile is organised by where it lives, so a protein with a nuclear pool recovered by nuclear elongation baits is the expected result of shared residency alone. Go CD et al. 2021
8. E8established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   PMIDs 28514442 and 33961781 are BioPlex 2.0 and BioPlex 3.0 - one platform, one collaboration, two releases, with Huttlin EL first author on both - whereas PMID 40205054 is an independent 5,100-protein joint AP-MS and immunofluorescence map of U2OS osteosarcoma cells from a different laboratory, so PIP4K2A's support is two platforms rather than one.

   Multimodal cell maps as a foundation for structural and functional genomics. · Schaffer LV, Hu M, Qian G, Moon KM, et al. (34 authors; senior author Ideker T) · Nature 2025
   PMID 40205054DOI
   ▸What this evidence cannot show

   The independence claimed here is platform and laboratory independence, not biological replication: all three publications use affinity purification with mass spectrometry on tagged, overexpressed constructs, so a systematic artefact of tagged-bait AP-MS would appear in all three. U2OS is an osteosarcoma line and HEK293T an embryonic kidney line; neither is haematopoietic and neither carries a KMT2A rearrangement. The BioPlex identifications were verified by fetching PMIDs 28514442 (Nature 2017, BioPlex 2.0, Harper JW senior) and 33961781 (Cell 2021, BioPlex 3.0 plus HCT116, Gygi SP senior).

   ▸Evidence that argues against this (1)

   • A large proteome-scale map assigns unexpected functions to hundreds of proteins by integration rather than by direct test - this one reports new functional assignments for 975 proteins - so a single row inside it is a hypothesis generated by the map's model, and independence of platform does not confer independence of inferential method. Schaffer LV et al. 2025
9. E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

   PIP4K2A is not a dependency in any leukaemia line: mean DepMap gene effect is +0.0526 across 1,208 lines, +0.1305 in 11 KMT2A-rearranged leukaemia lines and +0.1157 in 43 other leukaemia lines, ranking 11,086 of 17,728 on the KMT2A-selectivity difference - the anti-selective half.

   analyses/depmap-kmt2a-selective-scan-full/results.json - 17,728 protein-coding genes, 11 Cellosaurus-annotated KMT2A-rearranged versus 43 other leukaemia lines, computed 2026-09-19 · DepMap (Broad Institute), via this repository's scan 2026
   ▸What this evidence cannot show

   The gene effects here are POSITIVE and close to zero, which means the measurement is floor-compressed: in a gene nothing depends on, the between-group difference is a difference between two nulls and its rank carries no information. Reporting rank 11,086 as 'anti-selective' would be reporting the base rate - about half of any gene set lies above the median by construction and this project's own base rate is 45.3% - so the informative statement is the absolute value, not the rank. The stored scan records a compute date and NOT a DepMap release, so it cannot be told from quarterly drift.

   ▸Evidence that argues against this (1)

   • A CRISPR knockout screen cannot detect a dependency buffered by a paralogue, and PIP4K2A has two close paralogues, PIP4K2B and PIP4K2C, with PIP4K2B independently reported to interact with and control PIP4K2A's nuclear pool - so single-gene knockout is the wrong instrument for this family and a null here is uninformative about the pathway. Bultsma Y et al. 2010
10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    PIP4K2A has no histone, no ING2 and no KMT2A literature under its own symbol - PIP4K2A[tiab] AND histone[tiab] = 0, AND ING2[tiab] = 0, AND KMT2A[tiab] = 0, AND MLL[tiab] = 1, AND chromatin[tiab] = 2, AND epigenetic[tiab] = 5 - while the metabolite vocabulary reaches the mechanism (PtdIns5P[tiab] AND ING2[tiab] = 5; ING2[tiab] AND H3K4me3[tiab] = 23), and the previous-official-symbol control is a no-op (PIP5K2A[tiab] AND leukemia[tiab] = 0 against PIP4K2A[tiab] AND leukemia[tiab] = 24).

    NCBI E-utilities esearch, db=pubmed, querytranslation read back from every response, run 2026-09-22 · NCBI PubMed 2026
    ▸What this evidence cannot show

    All queries are one tag per term and never a parenthesised group; each querytranslation was read back and all were clean - for example "PIP4K2A"[Title/Abstract] AND "histone"[Title/Abstract] - with no MeSH explosion and no UID filter. Instrument controls through the identical code path returned H3K79me2[tiab] = 224 and PIP4K2A[tiab] AND acute lymphoblastic leukemia[tiab] = 17, so the zeros are not instrument failures. These zeros bound TITLES AND ABSTRACTS only: a PtdIns5P or PHD-finger lipid measurement inside a leukaemia paper's supporting figures would be invisible to all of them, which is the likeliest place such a measurement would sit. Absence class: never-measured for the fusion-context question.

    ▸Evidence that argues against this (1)

    • A bounded absence about a MEASUREMENT rather than an entity is bounded by abstracts, not by papers. This project's worked case is exactly this paper: its abstract never names a histone mark, and an H3K79me2 ChIP across five HOXA promoters sits in its supporting figures. A zero on a lipid or a mark query should therefore be read as an index limit, with the founding papers opened by hand. Van Vlierberghe P et al. 2008

• ← related toNone of the 166 leukaemia-family GWAS in the Catalog names KMT2A, MLL, 11q23 or infants, although ETV6::RUNX1, TCF3::PBX1 and hyperdiploid ALL each have their own scans - and a 987-case scan could only detect odds ratios of about 1.4-1.6 or more

  ALL risk gene already in the corpus

• regulates →ING2 - the PHD finger that reads H3K4me3 and turns it into REPRESSION, and whose founding co-crystal is filed under the MOUSE accession

  PIP4K2A consumes nuclear PtdIns5P, the ligand of the ING2 PHD finger

• binds →AFF1 / AF4 — the 3' half of the fusion, and a target with almost no structural footing

  PIP4K2A-bait AP-MS, MI:0914 association, HEK293T (BioPlex 2.0 + 3.0); not a binding event

• binds →AFF4 — the textbook SEC scaffold that KMT2A::AFF1 cells need LESS than other leukemias

  PIP4K2A-bait AP-MS in HEK293T (BioPlex 2.0/3.0) and U2OS (PMID 40205054) - two platforms

• binds →ELL — rank 7 on selectivity and the deepest common-essential in the top ten, with one NMR fragment to its name

  PIP4K2A-bait AP-MS, BioPlex 2.0 and 3.0, HEK293T

• binds →MLLT1 / ENL — the reader that beat its own paralogue, with a degrader already in hand

  PIP4K2A-bait AP-MS in HEK293T and U2OS - the bait set is five SEC subunits, so probably one contact

• binds →MLLT3 / AF9 — a famous fusion partner that is not a dependency, and a hub domain that is

  PIP4K2A-bait AP-MS, BioPlex 2.0 and 3.0, HEK293T

• co-occurs with →BMI1 - the self-renewal gene that KMT2A fusions can do without, because HOXA9 covers for it at the same locus

  the childhood-ALL susceptibility locus is annotated BMI1-PIP4K2A at 10p12.31-12.2; causal gene unassigned

• related to →KMT2A-rearranged acute lymphoblastic leukaemia

  germline 10p12 risk allele is a childhood-ALL finding with NO KMT2A-rearranged stratum published

• regulates →Plain-English: chromatin and epigenetics — the sticky notes on the DNA, and which ones matter here

  consumes nuclear PtdIns5P, the ligand of the ING2 PHD finger - a reader-module ligand, not a mark

• related to →Plain-English: what transcription elongation is, and why a traffic light explains this leukemia

  co-purifies with five super elongation complex subunits, always as bait

• related to →MED19 - the Mediator subunit curated against an engineered MLL-AFF4 fusion protein, and the one REST uses to deliver G9a

  both nominated by the same six-bait interactome read queue; MED19's support is one laboratory, PIP4K2A's is two platforms

• ← related toEPB41L3 / DAL-1 - a membrane-skeleton tumour suppressor that binds and modulates two arginine methyltransferases, and whose own promoter methylation is a clinical test

  two read-queue partners whose epigenetic connection is one enzyme step away and invisible to the approved gene symbol

• ← supportsIntAct curates 47 interactions against engineered MLL fusion proteins, and every accession-keyed query in this project is blind to all of them - while its PSICQUIC mirror stops at 2024

  PIP4K2A gains a third publication (PMID 40205054) that the PSICQUIC-based read queue cannot see