Oncofusion

start / graph / gene-hdgfl2

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.

HDGFL2 / HRP-2 - the second IBD protein, which binds KMT2A at the same site with the same affinity and does NOT need menin to do it

also called HDGFL2 · HRP-2 · HRP2 · HDGFRP2 · HDGF2 · Q7Z4V5 · hepatoma-derived growth factor-related protein 2

LEDGF is the protein that clips the leukaemia machine onto DNA, and it has a near-twin called HRP-2 that grips the same spot just as tightly. The difference matters: menin controls which of the two gets the grip, so a drug that removes menin may simply hand the job to the twin.

The same thing, for a scientist

HDGFL2 (Q7Z4V5, 671 aa) is, with PSIP1/LEDGF, **one of only two human proteins carrying an integrase-binding domain (IBD)** [E1]. It reached this graph not from the literature but from `analyses/unwritten-interactors-dependency-triage`, which found it selectively required in KMT2A-rearranged DepMap lines and unwritten [E5] - and the literature then **contradicted the selectivity reading and supplied something more interesting instead** [E2]. **The numbers are near-identical to LEDGF's.** The HRP-2 IBD solution structure (residues **469-549**) is compact and conserved with LEDGF's, and the dissociation constant for MLL 123-160 is **54.4 +/- 2.2 uM**, against **64.0 +/- 6.5 uM** for LEDGF/p75 measured the same way by the same group [E2]. **The difference is menin, and it runs the wrong way for the drug.** An MLL mutant that completely abolishes menin binding *still* co-precipitates HRP-2 while failing to co-precipitate LEDGF/p75 even with menin overexpressed; and **overexpressing menin increases LEDGF/p75 binding while REDUCING HRP-2 binding**. The authors read this as *"a competition between HRP-2 and LEDGF/p75 for binding to MLL under control of menin"* [E2]. **HRP-2 is dispensable for transformation and required for survival**, and the requirement is not fusion-specific: HRP-2-knockout HSCs are efficiently transformed by MLL-ENL, while HRP-2 knockdown reduces colony formation in MLL-rearranged **and** MLL-germline lines alike [E2].

Go deeper
Menin as a switch between two tethers, and what that does to the drug argument

This graph's account of menin is that it stabilises the LEDGF-KMT2A interface, and that removing menin therefore loosens the fusion's grip on chromatin. The Q33 beat sharpened that: LEDGF is a licence, not a tether for the fusion, and menin blockade reduces LEDGF chromatin binding (PMID 41279447).

This paper adds a second protein to the same site and makes menin a selector rather than just a stabiliser. Three observations, all from co-immunoprecipitation:

  1. A menin-binding-dead MLL mutant keeps HRP-2 and loses LEDGF/p75.
  2. Selective MLL-menin interaction inhibitors reproduce the asymmetry.
  3. Menin overexpression raises LEDGF binding and lowers HRP-2 binding.

So the predicted effect of a menin inhibitor is not simply less IBD protein at the fusion - it is a swap. Take menin out and the competition shifts toward the paralogue that does not need it. Whether the swap is functionally silent is exactly what is not known: HRP-2 and LEDGF bind MLL at the same site with the same affinity and the same structural mode, but their PWWP domains, their expression and their other partners are not identical.

This is a menin-independent route to the same interface, and it is a resistance hypothesis the pocket cannot be engineered around, because the escape is not a mutation in menin - it is a protein that was never in the complex the drug targets.

What would test it: measure HRP-2 occupancy at fusion target genes under revumenib, alongside LEDGF, in the same CUT&RUN experiment that already exists for LEDGF. The prediction is that LEDGF falls and HRP-2 rises. Nobody has done it - the published occupancy work profiles LEDGF alone.

The selectivity I found does not survive the paper that measured it

analyses/unwritten-interactors-dependency-triage put HDGFL2 among five unwritten interactors that are also selective dependencies: difference -0.161, index mean -0.239 against -0.079 in other leukaemia lines, pan-essentiality -0.119 5.

The experiment says otherwise, and it is a better instrument for this question. HRP-2 knockdown reduced colony formation in the MLL-rearranged lines THP-1 (32.8% +/- 2.1%) and SEM (51.5% +/- 7.0%) - and also in the MLL-germline lines K562 (44.4% +/- 27.4%), Kasumi-1 (38.3% +/- 14.2%) and Nalm-6 (19.3% +/- 8.8%). The authors' conclusion is that "loss of HRP-2 generally impairs growth of human and murine leukemic cells even in the absence of MLL fusions" 2.

The DepMap difference is small and the direct experiment is not selective, so the honest reading is that HRP-2 is a general leukaemic-survival gene, not a KMT2A-rearranged one. The entry is written anyway, and the reason has nothing to do with selectivity: it is the menin-controlled competition above.

Worth recording as a method note, because it cuts against an instrument this project leans on: a -0.16 DepMap difference at a pan-essentiality of -0.12 was contradicted by a five-line colony assay. The DepMap number was not wrong; it was too small to mean what ranking it implied.

A Kd discrepancy this graph should carry rather than resolve

interface-psip1-ibd-kmt2a records LEDGF IBD-KMT2A at 14.7 +/- 0.7 uM binary and 1.4 uM against the preformed menin-MLL complex, from PMID 25305204 4.

This paper reports LEDGF/p75 at 64.0 +/- 6.5 uM for the same class of measurement 2 - about 4.4-fold weaker, from a different group.

They are not strictly the same experiment. Murai and colleagues titrated MLL 1-160; Van Belle and colleagues titrated the MLL 123-160 peptide against LEDGF/p75 345-426. A shorter ligand binding more weakly is unsurprising, and the graph already records that MLL 1-46 binds menin about 8x tighter than the MBM1 peptide every competition assay uses - the same kind of construct-length effect, measured for the other partner.

So this is a scope difference and not a contradiction, and it is recorded here because the graph quotes 14.7 uM as though it were the affinity of the interface. It is the affinity of one construct pair. Any argument that turns on whether 14.7 uM is strong enough to hold the fusion at cellular concentrations - and this graph has such an argument open - has to survive the number being 64 uM for a shorter peptide.

Cell type, cell state, localisation, epigenetics, and the knockout
  • Knockout mouse. Systemic HRP-2 deletion causes increased postnatal mortality and, in weaned survivors, a significant increase in neutrophils (p = 0.042) with no change in total white or red cell counts. In vitro colony formation of haematopoietic stem cells is decreased 2. This is a worse tolerability signal than LEDGF's, whose knockout is dispensable for haematopoiesis - relevant because the LEDGF-directed strategy in this graph leans on that dispensability.
  • Cell type. Broadly expressed; in this disease, studied in THP-1, SEM and murine lin-negative bone marrow 2. SEM is the index KMT2A::AFF1 line, and it showed the largest colony reduction of the two MLL-r lines tested.
  • Cell state. Not characterised. No differentiation, cell-cycle or quiescence readout is reported.
  • Subcellular / subnuclear. Nuclear and chromatin-associated by domain architecture (PWWP reading methylated H3 tails); no subnuclear compartment is reported for HRP-2 in this work, which is a gap given that this graph has a subnuclear beat and a LEDGF condensate argument.
  • Epigenetics. The HATH/PWWP region binds methylated lysines in histone tails 2; the fragment-screening paper is explicit that the LEDGF PWWP reads H3K36me2/3 and that HRP-2's PWWP was used as the crystallographic surrogate because the LEDGF crystals were unsuitable 3.
  • Chemical matter. 68 crystallographic fragment hits against the HRP-family PWWP pocket, expanded in three directions to series with double-digit micromolar affinity and ligand efficiency above 0.5 kcal/mol per heavy atom 3. Raised on HRP-2 crystals, so they are at least as much HRP-2 ligands as LEDGF ligands.
Facts and figures
uniprotQ7Z4V5
gene symbolHDGFL2
length AA671
ibd residues469-549
subnuclear localizationheterochromatin, by association with HP1alpha, MPP8 and the HUSH complex - which is the OPPOSITE compartment from the one its paralog LEDGF/p75 occupies on this graph's axis, where LEDGF tethers KMT2A-menin to ACTIVE transcribed chromatin through H3K36me2/3.
shown for fusionNO for the localisation, YES for the genetics, and the split is the finding. HRP-2 interacts with MLL through a conserved interface - less menin-dependent than the MLL-LEDGF/p75 interaction - and a systemic HRP-2 knockout mouse shows HRP-2 is DISPENSABLE for MLL-ENL transformation while being important for leukaemic cell survival; that is the index-adjacent genetics. But no imaging, fractionation or subnuclear measurement of HRP2 has been made in any KMT2A-rearranged cell: the heterochromatin work is in MDA-MB-231 breast cancer and the HUSH work is a chemical-probe pulldown. So the graph can say HRP2 binds MLL and that HRP2 sits in heterochromatin, and cannot say whether the MLL-bound pool is the heterochromatic one. Zero in titles and abstracts; supporting figures not searched.
identifier trapNEW AND SEVERE, verified live 2026-09-21. HDGFL2's synonym HRP2 is one of the most heavily used strings in a completely different field: HRP2[tiab] returns 639 PubMed records and the overwhelming majority are Plasmodium falciparum HISTIDINE-RICH PROTEIN 2, the antigen behind malaria rapid diagnostic tests, plus pfhrp2/pfhrp3 deletion surveillance. 'HRP' is separately the standard abbreviation for horseradish peroxidase, a reagent in most western blots. Meanwhile the current symbol HDGFL2 is recent: the older chromatin and HIV-integration literature publishes as HDGF2, HDGFRP2 and HRP-2, so a symbol-only search splits this literature by era (rule 69 part 1). AND the readback defence fails: gene_exact:HDGFL2 returns entry name HDGR2_HUMAN with protein name 'Hepatoma-derived growth factor-related protein 2', neither containing 'HDGFL2' - the DPF2/LMO2 class. Length is a usable defence against the family: HDGFL2 671 aa, HDGF 240 aa, HDGFL1 251 aa, HDGFL3 203 aa.
localization
{
  "subcellular": "Nucleus, chromatin-associated. HRP2 co-purifies and co-sediments with heterochromatin-associated proteins and alters chromatin structure and stability when knocked out. UniProt Q7Z4V5 was read back to confirm the identifier and it is a readback trap: entry name HDGR2_HUMAN, recommended protein name 'Hepatoma-derived growth factor-related protein 2', gene synonyms HDGF2, HDGFRP2 and HRP2 - the string 'HDGFL2' appears nowhere in the protein name.",
  "subnuclear": "HETEROCHROMATIN. HRP2 copurifies and cosediments with heterochromatin protein 1 alpha (HP1alpha) and with M-phase phosphoprotein 8 (MPP8); in the Chromatin in vivo Assay across multiple cell types, HP1alpha-mediated gene repression dynamics are altered by the presence of HRP2, and HRP2 knockout in MDA-MB-231 cells produces significant changes to chromatin structure and stability. Independently, a peptidomimetic ligand against the MPP8 chromodomain pulled out HRP2 as an associate of the HUSH complex, the human silencing hub - a second, chemically-derived route to the same heterochromatin assignment. The PWWP domain that reads methylated histone tails is the presumed tether and is the same module its paralog LEDGF/PSIP1 uses.",
  "how_shown": "biochemical copurification and cosedimentation with HP1alpha and MPP8, Chromatin in vivo Assay in multiple cell types, and CRISPR knockout with chromatin structure and stability readouts (PMID 36672260); chemical-probe pulldown from the MPP8 chromodomain (PMID 34415726)",
  "systems": "MDA-MB-231 breast cancer and other non-haematopoietic lines for the heterochromatin work; MLL-ENL mouse transformation and human leukaemic lines for the leukaemia genetics - the two were never done in the same cell",
  "node": "localization-psip1-ledgf-chromatin-tether-subnuclear"
}
localization counted absence
{
  "query": "(HDGFL2[tiab] OR HDGF2[tiab] OR HRP2[tiab] OR HDGFRP2[tiab]) AND (localiz* OR nuclear OR chromatin OR speckle* OR nucleolar)[tiab]",
  "records": 34,
  "records_in_a_kmt2a_rearranged_cell": 0,
  "operand_controls": {
    "HRP2[tiab] (bare, to show the collision)": 639
  },
  "names_searched": [
    "HDGFL2",
    "HDGF2",
    "HDGFRP2",
    "HRP2",
    "HRP-2"
  ],
  "date": "2026-09-21",
  "tool": "PubMed E-utilities esearch",
  "bound": "zero in titles and abstracts; supporting figures not searched. The bare HRP2 operand is dominated by malaria rapid-diagnostic-test papers and every hit must be title-checked."
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.2394,
  "chronos_mean_other_leukaemia": -0.0789,
  "difference": -0.1605,
  "pan_essentiality_all_lines": -0.1192,
  "selectivity_rank": 415,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
cell type
{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "identifier_note": "HDGFL2 was HDGF2 / HRP-2 / HDGFRP2. Three separate traps, all live on 2026-09-21: (1) HRP2[tiab] returns 639 records dominated by Plasmodium falciparum histidine-rich protein 2, the malaria rapid-test antigen; (2) the CURRENT symbol is no better - HDGFL2[tiab] returns 13 records and ALL THIRTEEN are TDP-43 cryptic-exon papers in ALS, FTD, inclusion body myositis and Alzheimer's, because the HDGFL2 cryptic exon is now a leading TDP-43 proteinopathy biomarker and that field has taken the token; (3) the mouse gene is Hdgfrp2, so the knockout literature is unreachable from either human spelling. The chromatin/KMT2A literature is reached only through HRP-2 paired with MLL, LEDGF or leukaemia. Logged to data/symbol-collisions.json by this beat.",
  "expressed_in": {
    "class": 1,
    "summary": "broadly expressed; HRP-2 is the paralogue of LEDGF/p75 and shares its PWWP-plus-IBD architecture. No cell-type-restricted expression claim is made here."
  },
  "required_in": {
    "class": 2,
    "claim": "NO, in two independent settings, and the double negative is the finding. (a) In MICE, Hdgfrp2 knockout animals DEVELOP NORMALLY AND ARE FERTILE; it is only the Psip1/Hdgfrp2 DOUBLE knockout that dies, at approximately E13.5, with a ventricular septal defect and deregulated TGF-beta signalling [E_q69_ct1]. (b) In MLL-rearranged leukaemia, HRP-2 is DISPENSABLE for leukaemogenesis - unlike its paralogue LEDGF/p75 - but is important for leukaemic cell survival [E_q69_ct2].",
    "pmids": [
      "26367869",
      "33477970"
    ],
    "system_named_in_the_claim": "murine germline knockouts for (a); MLL-rearranged leukaemia models for (b)",
    "the_cell_type_where_it_IS_required": "the only cell type in which loss of HDGFL2 produces a phenotype is the EMBRYONIC CARDIAC VENTRICLE, and only when PSIP1/LEDGF is also absent. That is a paralogue-buffered requirement in a non-haematopoietic tissue, and it is the cleanest available statement that HDGFL2's cell-type biology is not a blood biology."
  },
  "nodes": [
    "cell-type-cardiomyocyte"
  ],
  "no_cell_type_node_reused": {
    "RESOLVED_2026_09_21_Q73": "cell-type-cardiomyocyte now exists and this entry is edged to it with type `regulates`, labelled as the paralogue-buffered requirement it is - neither Psip1 nor Hdgfrp2 single knockout gives the cardiac phenotype. The reason text below is kept as the record of why the approximation was refused. The new node also carries the more consequential cardiac case it surfaced: cardiac-specific Dot1L knockout phenocopies dilated cardiomyopathy through Dystrophin (PMID 21289070), and DOT1L is a drugged target in this graph.",
    "reason": "the graph's cell vocabulary has no cardiac or cardiomyocyte node, and the leukaemic result is a survival requirement that is not tied to any cell type or state in the existing 51. Rather than force a wrong edge or create a node outside this beat's remit, the answer is recorded in this field and linked to gene-psip1, which is the biologically load-bearing relation.",
    "vocabulary_gap_reported": true
  },
  "context": "murine embryonic heart for the developmental requirement; MLL-rearranged leukaemia models for the survival requirement"
}
cell state
{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "maintains_state": {
    "class": 3,
    "states": [],
    "claim": "NO. No cell state has been shown to depend on HDGFL2 in any system. The closest published statement is that HRP-2 is important for leukaemic cell SURVIVAL in MLL-rearranged models [E_q69_ct2], which is a viability requirement rather than a state, and that TGF-beta signalling is the only pathway significantly deregulated in the Psip1/Hdgfrp2 double-knockout ventricle [E_q69_ct1]. This absence is UNPUBLISHED rather than not-applicable: the experiments have not been done, not that the question is ill-posed.",
    "pmids": [
      "33477970",
      "26367869"
    ]
  },
  "bounded_query": {
    "date": "2026-09-21",
    "tool": "NCBI esearch db=pubmed",
    "query": "HRP2[tiab] AND (MLL OR KMT2A OR LEDGF OR leukemia)[tiab]",
    "count": 15,
    "on_topic_after_title_check": 3,
    "names_searched": [
      "HDGFL2",
      "HDGF2",
      "HRP-2",
      "HRP2",
      "HDGF-related protein 2",
      "Hdgfrp2"
    ],
    "why_the_name_list_is_the_instrument_here": "HDGFL2[tiab] alone returns 13 records, none of them about chromatin or leukaemia - every one is TDP-43 cryptic-exon neurodegeneration. (HDGFL2 OR HDGF2 OR 'HDGF-related protein 2' OR HRP-2)[tiab] paired with developmental terms returns 48, and (HDGFL2 OR HRP-2 OR HDGF2)[ti] alone returns 577, almost all malaria. A reader cannot judge this absence without the name list, which is why it is written out (brief rule 70).",
    "bound": "titles and abstracts only; supporting figures not searched (brief rule 80)"
  },
  "depmap_reading": "Chronos -0.2394 in the KMT2A-rearranged arm against -0.0789 in other leukaemias, difference -0.1605, rank 415 of 17,728, pan-essentiality -0.1192 - the SELECTIVE half, and one of only four genes in this tranche that are. That is consistent with 'dispensable for leukaemogenesis but important for leukaemic cell survival': a proliferation screen measures survival of an established line, which is exactly the arm the paper says HRP-2 matters in, and cannot see initiation, which is the arm it says HRP-2 does not matter in. The two instruments agree here, which is worth saying because in most of this tranche they do not.",
  "confidence": "moderate",
  "caveats": "The cardiac result is a murine germline double knockout - a developmental, non-haematopoietic phenotype, and single Hdgfrp2 loss has none. The leukaemia result is in MLL-rearranged models rather than infant KMT2A::AFF1 specifically, and 'important for leukaemic cell survival' is the abstract's phrasing, with the magnitude not stated there. No cell state has been measured for this protein in any system, and the entry says so rather than inferring one from the localisation data it already holds.",
  "counter_evidence": {
    "note": "The DepMap agreement above should not be over-read. A Chronos difference of -0.1605 on a pan-essentiality of -0.1192 is a small absolute effect, and rank 415 of 17,728 is a rank on a statistic this project has itself shown correlates at Spearman -0.92 with comparator-arm narrowness among depleted genes - i.e. the rank partly measures how few other leukaemias care about the gene. The raw difference, not the rank, is the number to quote.",
    "citation": {
      "title": "DepMap KMT2A-selective dependency scan (this project, genome-wide, 11 KMT2A-rearranged vs 43 other leukaemia lines)",
      "url": "analyses/depmap-kmt2a-selective-scan-full/results.json",
      "type": "analysis"
    }
  }
}
How this could bite us back
  • A LEDGF-directed agent may be answered by the paralogue. HRP-2 binds the same MLL site with comparable affinity and needs menin less, so removing LEDGF - or removing menin - may shift the interface to HRP-2 rather than vacate it. Any LEDGF-removal experiment that leaves HRP-2 intact may be measuring paralogue buffering rather than LEDGF dispensability.
  • Hitting both paralogues is worse, not better, on the tolerability evidence available. LEDGF knockout is dispensable for haematopoiesis; the systemic HRP-2 knockout shows increased postnatal mortality and a neutrophil shift. The PWWP fragment series binds both, so the one chemical strategy that exists is the one with the worse genetic prediction.
  • The requirement is not fusion-specific. HRP-2 knockdown cuts colony formation in K562, Kasumi-1 and Nalm-6 as much as in THP-1 and SEM, so an HRP-2-directed agent should be expected to hit non-KMT2A leukaemia and normal progenitors, and the therapeutic window has not been measured in any system.
  • Nothing here is in an infant, in KMT2A::AFF1 in vivo, or in a human primary cell. The transformation arm is murine MLL-ENL and the only index-fusion datum is a knockdown in one cell line.
  • a paralog that does not obey the menin rule - HRP-2 binds MLL through a conserved interface that is LESS menin-dependent than the MLL-LEDGF/p75 interaction, and HRP-2 is dispensable for MLL-ENL transformation while mattering for leukaemic cell survival. A menin inhibitor would not be expected to displace it, so HRP2 is a candidate residual-binding route that menin-inhibitor resistance screens looking for MEN1 mutation would not detect.
Still open
  1. Measure HRP-2 occupancy at fusion target genes under revumenib, in the same CUT&RUN experiment that already profiles LEDGF. The competition model predicts LEDGF falls and HRP-2 rises; nobody has looked, and it is an antibody away.
  2. Does KMT2A::AFF1 - rather than MLL-ENL or wild-type MLL - bind HRP-2? Every co-IP in the source paper uses MLL or MLL-ELL constructs. The fusion retains the IBD-binding region, so the prediction is yes, but it is a prediction.
  3. Is the HRP-2 requirement in leukaemic cells separable from its role in normal progenitors? The knockout mouse has decreased HSC colony formation and the knockdown reduces leukaemic colonies, which is the shape of a gene with no window.
  4. What is HRP-2's subnuclear distribution, and does it share LEDGF's condensate behaviour? The graph has a subnuclear beat and a condensate-evidence standard, and neither has been applied to this protein.
  5. Resolve or scope the LEDGF Kd: 14.7 uM with MLL 1-160 against 64.0 uM with MLL 123-160. Until someone measures both constructs side by side, the interface has no single number.
  6. Is the MLL-bound pool of HRP2 the same pool that sits with HP1alpha, MPP8 and HUSH? The paralog pair LEDGF/HRP2 currently occupies opposite chromatin compartments in this graph, and a single ChIP-seq of HRP2 in a KMT2A-rearranged line would say whether that split is real or an artefact of which cell type each was studied in.
Every source behind this entry (10)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    HDGFL2 is UniProt Q7Z4V5, 671 amino acids, with gene synonyms HDGF2, HDGFRP2 and HRP2, named hepatoma-derived growth factor-related protein 2. With PSIP1/LEDGF it is one of only two members of the HDGF-related protein family carrying an integrase-binding domain at its C-terminus.

    What this evidence cannot show

    Accession, length and synonyms fetched from UniProt on 2026-09-20 with `gene_exact` and the primary gene name read back, after rule 62 - a `gene_exact` query can return several reviewed records and the first may be the wrong protein. The two-IBD-proteins claim is from the Van Belle paper's introduction [E2], not from UniProt.

    Evidence that argues against this (1)
    • Domain architecture is not function. HRP-2 and LEDGF share the PWWP and the IBD and diverge elsewhere, and this graph has been wrong before by carrying a structural similarity into a functional claim - the clamp that holds wild-type KMT2A was written as holding the fusion. Van Belle S et al. 2021
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.journal-article

    The HRP-2 IBD solution structure (residues 469-549) is compact and conserved with that of LEDGF/p75, and the NMR dissociation constant for MLL 123-160 binding HRP-2-IBD is 54.4 +/- 2.2 uM against 64.0 +/- 6.5 uM for LEDGF/p75. An MLL mutant that abolishes menin binding still co-precipitates HRP-2 while failing to co-precipitate LEDGF/p75 despite overexpressed menin, and menin overexpression increases LEDGF/p75 co-precipitation while reducing HRP-2, which the authors read as competition between the two under the control of menin. Systemic HRP-2 knockout mice show increased postnatal mortality and a significant increase in neutrophils (p = 0.042) with normal total white and red counts; HRP-2-knockout HSCs are efficiently transformed by MLL-ENL; and HRP-2 knockdown reduces colony formation in THP-1 (32.8%), SEM (51.5%), K562 (44.4%), Kasumi-1 (38.3%) and Nalm-6 (19.3%).

    PMID 33477970DOI
    What this evidence cannot show

    Read in full on 2026-09-20 (PMC7835958, 31,833 characters after stripping) and the numbers read back from the text. **Established as results within the reported study.** The transformation arm is **MLL-ENL in murine HSCs, not KMT2A::AFF1 and not human**; the only index-fusion material is the SEM knockdown. The co-immunoprecipitation work is in overexpression systems, so the menin-controlled competition is a result about tagged proteins at non-physiological levels. The K562 colony reduction carries a standard deviation of 27.4% on a 44.4% effect.

    Evidence that argues against this (1)
    • The paper's own colony data are the strongest argument against reading HRP-2 as a KMT2A-rearranged dependency: three MLL-germline lines fall as much as the two MLL-rearranged ones, and the authors say so. Any therapeutic case for HRP-2 in this disease has to explain why it would not equally be a case in K562 and Nalm-6. Oncofusion knowledge graph (internal analysis) 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.journal-article

    X-ray crystallography-based fragment screening against the PWWP domain of the HRP family identified 68 diverse fragment hits sampling the H3K36me2/3 pocket, expanded by structure-guided design in three directions into compound series with double-digit micromolar affinity and Kd-based ligand efficiency exceeding 0.5 kcal/mol per non-hydrogen atom. The screen used crystals of the HRP-2 PWWP domain because the LEDGF PWWP crystals were not suitable.

    Rational fragment-based design of compounds targeting the PWWP domain of the HRP family · Vantieghem T, ..., Strelkov SV · European Journal of Medicinal Chemistry 2024
    PMID 39461037DOI
    What this evidence cannot show

    Read from the abstract on 2026-09-20; no PMCID and the full text was not retrieved. Established as what the paper reports. **Double-digit micromolar is not a drug** and the authors state the affinity plainly rather than claiming otherwise. Because the crystals are HRP-2's, cross-reactivity with LEDGF is the design goal in one reading and the liability in another, and the abstract does not resolve which.

    Evidence that argues against this (1)
    • A PWWP-directed compound blocks chromatin reading by BOTH paralogues, so it cannot be used to test the menin-controlled switch between them - the experiment that matters here needs a selective tool or a genetic one, and neither exists. PLoS Pathogens 2012
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.journal-article

    This graph records the LEDGF IBD-KMT2A dissociation constant as 14.7 +/- 0.7 uM binary and 1.4 uM against the preformed menin-MLL complex, measured with MLL 1-160; the HRP-2 paper reports 64.0 +/- 6.5 uM for LEDGF/p75 with the MLL 123-160 peptide against LEDGF/p75 345-426.

    PMID 25305204DOI
    What this evidence cannot show

    Both numbers were read from their sources. **This is a scope difference, not a contradiction**: the ligands differ (MLL 1-160 against MLL 123-160) and a shorter peptide binding more weakly is expected - the same graph already records MLL 1-46 binding menin about 8-fold tighter than the MBM1 peptide used in competition assays. Graded `strong` rather than `established` because it is a comparison across two studies that did not set out to be compared.

    Evidence that argues against this (1)
    • Recording it as a scope difference is the charitable reading and it may be too charitable. Two groups measuring the same interface by NMR and differing 4.4-fold is also what a systematic difference in construct, buffer or fitting looks like, and nothing here distinguishes those. The actionable consequence is only that 14.7 uM should not be quoted as *the* affinity of the interface. Van Belle S et al. 2021
  5. E5weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.analysis

    In a DepMap triage of 262 curated interactors with no entry in this graph, HDGFL2 showed a mean gene-effect difference of -0.161 between 11 KMT2A-rearranged lines (-0.239) and 43 other leukaemia lines (-0.079), at a pan-essentiality of -0.119 across 1,208 lines.

    What this evidence cannot show

    Run 2026-09-20. **Moderate, and deliberately downgraded**: the effect is small, the index arm is 11 lines, no permutation null was computed, and **the direct experiment in [E2] contradicts the selectivity reading**. Recorded because it is how this protein entered the graph and because the disagreement between a screen-derived ranking and a colony assay is itself worth having on the record. **Null computed 2026-09-20 and this item does not survive it** (`ana-the-null-kills-my-own-shortlist`): 50,000 label permutations give p = 0.0187, and Bonferroni over the 254 genes the shortlist was drawn from leaves nothing below 0.05. This agrees with the published colony data already cited here, which show the requirement is not fusion-specific. The entry's reason for existing - menin selecting between two IBD paralogues - never rested on selectivity.

    Evidence that argues against this (1)
    • HRP-2's myeloid/lymphoid split in that triage (-0.35 against -0.05) runs opposite to FANCD2's (-0.48 against -0.70), so the two selective hits from the same run cannot both be evidence for a lymphoid-biased mechanism, and neither should be used as if it were. Van Belle S et al. 2021
  6. E_q65_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    HRP2/HDGFL2 copurifies and cosediments with the heterochromatin-associated proteins HP1alpha and MPP8; in the Chromatin in vivo Assay across multiple cell types HP1alpha-mediated gene repression dynamics are altered by the presence of HRP2, and HRP2 knockout in MDA-MB-231 cells produces significant changes to chromatin structure and stability that alter gene expression.

    PMID 36672260DOI
    What this evidence cannot show

    MDA-MB-231 breast cancer and other non-haematopoietic cells; nothing here is in leukaemia (brief rule 73). Cosedimentation and copurification place HRP2 in a heterochromatin protein community, which is a biochemical compartment assignment rather than an image of HRP2 in a heterochromatic domain. Abstract only.

    Evidence that argues against this (1)
    • The same protein's best-characterised chromatin function points the other way: HDGF2/HRP2 with LEDGF relieves the nucleosome-induced barrier to transcription in differentiated cells, which is an activity at TRANSCRIBED chromatin, not at heterochromatin. HRP2 may sit at both and no experiment has partitioned it. LeRoy G et al. 2019
  7. E_q65_loc2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A peptidomimetic chemical ligand targeting the chromodomain of MPP8, a member of the human silencing hub (HUSH) complex, was used to reveal HRP2's association with the HUSH complex - an independent, chemically-derived route to the same heterochromatin assignment.

    PMID 34415726DOI
    What this evidence cannot show

    A chemical-probe pulldown identifies association, not stoichiometry or residence; probe pulldowns collect proteins that bind the probe-bound partner under lysis conditions. Not a leukaemia system. Abstract only.

    Evidence that argues against this (1)
    • Two independent routes agreeing on a heterochromatin community still leave the question this graph needs unanswered - whether the MLL-bound HRP2 pool is the HUSH-associated one - and both routes were run in cells with no KMT2A rearrangement. Wistner SC et al. 2023
  8. E_q65_loc3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    HRP-2 interacts with MLL through a conserved interface that is less menin-dependent than the MLL-LEDGF/p75 interaction; a systemic HRP-2 knockout mouse shows HRP-2 is DISPENSABLE for MLL-ENL transformation and for normal haematopoiesis while remaining important for leukaemic cell survival - and this work contains no localisation measurement.

    PMID 33477970DOI
    What this evidence cannot show

    The localisation absence is why it is cited here. MLL-ENL mouse transformation plus lentiviral miRNA knockdown in human leukaemic lines - not KMT2A::AFF1 and not infant B-lineage (brief rule 73). 'Less dependent on menin' is a comparative statement against LEDGF/p75 and is the load-bearing detail for this graph's menin-interface hypothesis. Abstract only.

    Evidence that argues against this (1)
    • If HRP2 binds MLL through a menin-INDEPENDENT interface, then a menin inhibitor would not displace it, and the paralog pair splits: LEDGF/p75 is essential for MLL-r leukaemia and menin-dependent, HRP2 is dispensable for transformation and less menin-dependent. That is a constraint on the menin-interface hypothesis and it is derived from the paralog rather than from the target. Wistner SC et al. 2023
  9. E_q69_ct1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In MICE, Hdgfrp2 knockout animals develop normally and are fertile, whereas Psip1/Hdgfrp2 double-deficient mice die at approximately embryonic day 13.5 with ventricular septal defect at E14.5; RNA sequencing of ventricular tissue found TGF-beta signalling to be the only pathway significantly deregulated by the double knockout relative to control and Psip1-knockout samples, and among genes encoding LEDGF/p75 binding factors only Nova1 was significantly deregulated, with global RNA splicing not driving the pathology.

    PMID 26367869DOI
    What this evidence cannot show

    Abstract read in full; full text not read. Murine germline knockouts, cardiac development - not haematopoietic, not leukaemia, not human (brief rule 73). The single-knockout 'normal' phenotype is a steady-state observation with no challenge applied. The TGF-beta conclusion is the authors' own speculation, stated as such in their abstract ('we accordingly speculate').

    Evidence that argues against this (1)
    • A normal single knockout does not establish dispensability in a leukaemic context, and the directly relevant test points the other way for survival: in MLL-rearranged leukaemia HRP-2 is dispensable for leukaemogenesis but IS important for leukaemic cell survival. Germline redundancy with PSIP1 in development and a non-redundant role in an established leukaemia are compatible, and the entry records both rather than generalising either. Cells 2021
  10. E_q69_ct2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Unlike its paralogue LEDGF/p75, HRP-2 is dispensable for MLL-rearranged leukaemogenesis but is important for leukaemic cell survival.

    PMID 33477970DOI
    What this evidence cannot show

    Title-level and abstract-level claim; full text not read this session. MLL-rearranged leukaemia models rather than infant KMT2A::AFF1 specifically (brief rule 73). The magnitude of the survival effect is not given at abstract level. This PMID is already cited elsewhere in this entry; it is re-fetched here and cited for the CELL-TYPE claim, which is a different claim from the same paper.

    Evidence that argues against this (1)
    • The murine germline data give the opposite impression of importance: Hdgfrp2-null mice are normal and fertile, and only the Psip1/Hdgfrp2 double knockout has a phenotype. Anyone reading 'important for leukaemic cell survival' as evidence of a general HDGFL2 requirement should see the normal knockout mouse next to it. Wang H et al. 2015
Connected to