Oncofusion

start / graph / gene-hdgf

Geneweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q132-interactome-tail-promotion

HDGF - the third PWWP paralogue, which pulls down LEDGF but has the reader domain WITHOUT the domain that grips MLL

also called hepatoma-derived growth factor · HMG1L2 · HATH domain protein · Hepatoma-derived growth factor

HDGF is the founding member of a small family of proteins that read a chemical mark on packaged DNA; two of its relatives, LEDGF and HRP2, are already central to this project because they hold the leukaemia fusion onto chromatin. HDGF itself pulls LEDGF out of cells in a 2011 experiment, but it is missing the second domain its two relatives use to grip MLL - which makes it the family's built-in negative control and nobody has run it.

The same thing, for a scientist

HDGF (HGNC:4856, ENSG00000143321, UniProt P51858, 240 aa, 1q23.1) is the founding member of the HDGF-related protein (HRP) family, which also contains **PSIP1/LEDGF** and **HDGFL2/HRP2** - the two proteins this project already treats as chromatin tethers for KMT2A and its fusions. HDGF enters the core-complex interactome as a **bait** whose purification recovers PSIP1 as prey: two IntAct rows, IM-17230, tandem affinity purification, association, MI-score 0.35, host HEK293T, from Yu et al.'s 106-protein HDGF interactome [E1][E2]. **The structural comparison is the point of this entry.** UniProt annotates HDGF as PWWP(12-69) followed by 171 residues of disorder and nothing else; PSIP1 carries PWWP(1-64) **and an integrase-binding domain (IBD, 340-417)**; HDGFL2 carries PWWP(7-64) and a C-terminal interaction region [E3]. The IBD is the module through which LEDGF/p75 and HRP-2 engage the MLL IBM [E4]. **HDGF has the reader and not the grip.** That makes it a principled negative control for the LEDGF tether hypothesis - and this beat could find no experiment that has ever used it as one.

What it looks like
HDGF - the third PWWP paralogue, which pulls down LEDGF but has the reader domain WITHOUT the domain that grips MLL240 aa
PWWP (HATH)PWWP (HATH) · residues 12–690100200

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.

Go deeper
What the interaction record actually is

Two IntAct rows, IM-17230, deposited 2011-09-20, last updated 2026-01-11. Detection MI:0676 tandem affinity purification; interaction type MI:0914 association; expansion MI:1060 spoke; experimental roles MI:0496 bait (HDGF) / MI:0498 prey (PSIP1); host organism taxid:9606(Homo sapiens HEK293T embryonic kidney cell); curator annotation figure legend: Table S1, with IntAct dataset tags Cancer and Apoptosis 2.

Yu et al. purified HDGF complexes with a streptavidin-binding-peptide / Flag tandem affinity tag and LC-MS/MS, recovering 106 proteins, confirmed some by co-IP and RT-PCR, and showed that the HATH domain (the family name for the PWWP fold) is essential for HDGF's protein-protein and protein-RNA interactions 1. PSIP1 is one row in that 106 and the paper's abstract does not name it; this entry does not claim the authors validated this specific pair.

A caution specific to this family. All five HRP proteins share the same PWWP/HATH fold and considerable sequence similarity. A co-purification between two members of one paralogue family in an over-expression pull-down is exactly the situation in which a shared epitope, a shared nucleic-acid contact or a shared chromatin fragment produces a row that is not a protein-protein contact. This pair should be read with that discount applied.

The domain argument, and why it makes HDGF a control rather than a candidate
proteinlengthPWWPIBD / C-terminal interaction module
HDGF240 aa12-69none annotated - residues 69-240 are disordered
PSIP1 / LEDGF530 aa1-64IBD 340-417
HDGFL2 / HRP2671 aa7-64C-terminal interaction region 470-552

3

This graph already holds interface-psip1-ibd-kmt2a, the LEDGF IBD-to-MLL interface, and interface-psip1-pwwp-h3k36me-nucleosome, the PWWP-to-H3K36me nucleosome contact. Van Belle et al. showed that MLL and HRP-2 interact through a conserved interface, although this interaction proved less dependent on menin than the MLL-LEDGF/p75 interaction, and that HRP-2 knockout cells are still efficiently transformed by MLL-ENL while HRP-2 depletion reduces colony formation independently of MLL rearrangement 4.

So the family splits cleanly: two members have the grip and one does not. If the tether model is right, HDGF should bind H3K36me2/3 chromatin like its relatives and should not bind the MLL IBM - and should therefore not rescue LEDGF loss in a KMT2A-rearranged cell. Nobody has tested this. It is a one-week co-IP plus a rescue experiment and it would discriminate the tether model from a generic PWWP-recruitment model.

Localization and cell type

UniProt P51858 gives experimental locations nucleus, cytoplasm and secreted, extracellular exosome 5. No subnuclear compartment is assigned and none is asserted here. The secreted pool is real and is the reason HDGF is named a growth factor, and it is also a reason to treat a nuclear-complex co-purification with care.

Cell type of the observation: HEK293T embryonic kidney. Not haematopoietic, not leukaemic, not KMT2A-rearranged. HDGF's cancer literature is dominated by solid tumours, particularly hepatocellular carcinoma.

Family redundancy, and the one in vivo experiment that exists

Wang et al. knocked out Hdgfrp2 (HRP2) in normal and Psip1 knockout mice: Hdgfrp2 single knockouts developed normally and were fertile, while the double-deficient embryos died around E13.5 with ventricular septal defects 6. That is an in vivo demonstration of functional redundancy between two members of this family, and it is why a negative result for HDGF alone would not be conclusive - the family compensates. It is also the strongest argument for testing HDGF and HRP2 together rather than separately.

Facts and figures
hgnc IDHGNC:4856
symbolHDGF
ensembl gene IDENSG00000143321
uniprotP51858
length AA240
chromosomal location1q23.1
identifier provenanceHGNC REST fetch/symbol/HDGF and UniProtKB P51858, O75475, Q7Z4V5 all retrieved live 2026-09-23. The symbol-collision registry was checked first and returned the existing HRP2 row, which is about HDGFL2, not about HDGF.
subnuclear locationnone assigned in UniProt; none asserted here
cell type of the observationHEK293T embryonic kidney - not haematopoietic, not KMT2A-rearranged
epigenetic rolePWWP/HATH chromatin reader by family homology; UniProt records HDGF acting as a transcriptional repressor. No H3K36me binding measurement for HDGF itself was found by this beat.
shown for fusionNO for HDGF - nobody has looked in any fusion. For the PARALOGUES: LEDGF/p75 is essential for MLL-rearranged leukaemia and HRP-2 was tested against MLL-ENL in primary murine cells and found dispensable for transformation but important for leukaemic cell survival (PMID 33477970). Neither result is a KMT2A::AFF1 result.
evidence classsingle publication, single method, spoke-expanded TAP-MS between two members of one paralogue family - the discount for shared epitopes applies
prev and alias symbols
[
  "HMG1L2"
]
interaction with project baits
[
  {
    "bait": "PSIP1",
    "bait_uniprot": "O75475",
    "partner_experimental_role": "bait",
    "scaffold_experimental_role": "prey",
    "method": "MI:0676 tandem affinity purification",
    "interaction_type": "MI:0914 association",
    "expansion": "MI:1060 spoke",
    "mi_score": 0.35,
    "imex": "IM-17230",
    "pmid": "21907836",
    "host_cell": "HEK293T embryonic kidney",
    "n_rows": 2,
    "n_publications": 1
  }
]
domains
[
  {
    "name": "PWWP (HATH)",
    "start": 12,
    "end": 69,
    "source": "UniProt P51858 feature table"
  }
]
family
{
  "name": "HDGF-related protein (HRP) family",
  "members_in_this_graph": [
    "gene-psip1",
    "gene-hdgfl2",
    "gene-hdgf"
  ],
  "architecture": "HDGF PWWP only (240 aa); PSIP1 PWWP + IBD 340-417 (530 aa); HDGFL2 PWWP + C-terminal interaction region 470-552 (671 aa)"
}
subcellular location uniprot
[
  "Nucleus (ECO:0000269)",
  "Cytoplasm (ECO:0000269)",
  "Secreted, extracellular exosome (ECO:0000269)"
]
How this could bite us back
  • If HDGF or another family member can substitute for LEDGF at PWWP-marked chromatin, an intervention aimed at the LEDGF-MLL interface has a built-in compensation route that no LEDGF-only experiment would reveal.
  • The PSIP1/HDGFL2 double knockout is embryonic lethal with cardiac defects, so any strategy that hits more than one HRP member carries a developmental-toxicity signal already documented in mouse - relevant in an infant population.
  • HDGF has a secreted, exosomal pool and a solid-tumour growth-factor literature; an agent against the family would not be acting on one compartment or one disease.
Still open
  1. Does HDGF bind the MLL IBM? The prediction from domain architecture is NO, and a co-IP against the MLL IBM with LEDGF as positive and HDGF as the paralogue negative would be one of the cheapest discriminating experiments available to this project.
  2. Can HDGF over-expression rescue LEDGF depletion in a KMT2A-rearranged line? If it can, the tether model is wrong and PWWP recruitment alone is sufficient; if it cannot, the IBD is load-bearing and the interface is the target.
  3. Does HDGF bind H3K36me2/3 nucleosomes with the affinity its relatives do? The PWWP folds are homologous but the affinities are not assumed equal, and this graph has a cxxc-affinity beat that already knows how much affinity differences matter.
  4. Is HDGF expressed at all in infant KMT2A::AFF1 blasts? If it is absent, the negative-control experiment is cleaner but the biological question evaporates.
  5. Would a triple HDGF/PSIP1/HDGFL2 perturbation be needed to see a phenotype, given that the PSIP1/HDGFL2 double knockout is embryonic lethal while each single knockout is viable?
Every source behind this entry (6)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Yu et al. used streptavidin-binding-peptide and Flag tandem affinity purification coupled with LC-MS/MS to show that 106 proteins form complexes with HDGF, found RNAs in the HDGF complex by SBP-tag RNA co-immunoprecipitation, confirmed some interactions by co-IP and RT-PCR, and found that the HATH domain was essential for HDGF's protein-protein and protein-RNA interactions.

    PMID 21907836DOI
    What this evidence cannot show

    ABSTRACT ONLY - not open access at Europe PMC (isOpenAccess=N, no PMCID served), so Table S1, the tagging construct, the wash stringency and the control purification were not read. The abstract does NOT name PSIP1/LEDGF; the pair is reachable only through the IMEx deposition [E2]. The author list was not retrieved beyond the PubMed record and is therefore given as the citation title only. A 2011 over-expression TAP-MS with 106 recovered proteins is a permissive experiment by current standards.

    Evidence that argues against this (1)
    • HDGF and PSIP1 are paralogues sharing the PWWP/HATH fold. A co-purification between two members of one small paralogue family in an over-expression pull-down can arise from a shared epitope, shared nucleic acid or a shared chromatin fragment rather than a protein-protein contact, and no orthogonal assay for this pair was found. source 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    IntAct serves two curated human HDGF-PSIP1 rows: MI:0676 tandem affinity purification, MI:0914 association, MI:1060 spoke expansion, HDGF experimental role MI:0496 bait and PSIP1 MI:0498 prey, MI-score 0.35, host taxid:9606 Homo sapiens HEK293T embryonic kidney cell, IMEx IM-17230, publication PMID 21907836, curator annotation 'Table S1'.

    What this evidence cannot show

    Retrieved live 2026-09-23. Two rows from ONE publication is not two independent observations. Rule 93: PSICQUIC is a floor relative to IntAct REST.

    Evidence that argues against this (1)
    • TWO ROWS FROM ONE PUBLICATION ARE ONE OBSERVATION. IntAct deposits a row per experiment within a publication, so a reader counting rows as replication would double-count this pair. The correct reading is n=1 paper, n=1 method, and the row count is a curation artefact rather than evidence weight. source 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt annotates HDGF (P51858, 240 aa) with a single PWWP domain at 12-69 followed by a disordered region 69-240 and no further domain; PSIP1 (O75475, 530 aa) with PWWP 1-64 and an integrase-binding domain (IBD) at 340-417; HDGFL2 (Q7Z4V5, 671 aa) with PWWP 7-64 and a C-terminal interaction region at 470-552. All three are annotated 'Belongs to the HDGF family'.

    What this evidence cannot show

    Fetched live 2026-09-23 with fields=ft_domain,ft_region. ABSENCE OF AN ANNOTATION IS NOT ABSENCE OF A DOMAIN: UniProt not annotating an IBD in HDGF's 171-residue disordered tail is strong but not conclusive evidence that no MLL-binding module exists there. A sequence alignment of the three tails against the LEDGF IBD would settle it and was not run by this beat.

    Evidence that argues against this (1)
    • HDGF's C-terminal 171 residues are annotated as disordered, and disordered regions are exactly where short linear interaction motifs hide from domain-based annotation. The structural argument in this entry is a prediction to be tested, not an established negative. source 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Van Belle et al. demonstrated that MLL and HRP-2 interact through a conserved interface, although this interaction proved less dependent on menin than the MLL-LEDGF/p75 interaction; systemic HRP-2 knockout mice showed only an increase in peripheral-blood neutrophils, depletion of HRP-2 in leukaemic cell lines and transformed primary murine cells reduced colony formation independently of MLL rearrangement, and primary murine HRP-2 knockout cells were efficiently transformed by the MLL-ENL fusion - so HRP-2, unlike LEDGF/p75, is dispensable for MLL-ENL leukaemogenesis but important for leukaemic cell survival.

    PMID 33477970DOI
    What this evidence cannot show

    Abstract read, not full text. The transformation experiments are MLL-ENL in primary murine cells, NOT KMT2A::AFF1 and not human - rule 86 applies and the fusion is named here rather than collapsed to a yes/no. This citation is used for the FAMILY architecture argument; it says nothing about HDGF itself, which the paper does not test.

    Evidence that argues against this (1)
    • This paper is about HDGFL2/HRP-2, not about HDGF. Extending its interface finding to HDGF would be exactly the paralogue-substitution error this graph has recorded elsewhere; the entry uses it only to establish that the IBD-bearing members engage MLL. source 2026
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt P51858 is 240 aa, records HDGF as acting as a transcriptional repressor with contested mitogenic and heparin-binding activity, and assigns experimental (ECO:0000269) subcellular locations nucleus, cytoplasm and secreted/extracellular exosome.

    What this evidence cannot show

    Fetched live 2026-09-23. UniProt itself records CONTRADICTORY function comments for this protein - one isoform-scoped comment states it has mitogenic activity for fibroblasts and binds heparin while another states it does not (PubMed:26845719 on both sides). That contradiction is in the source record and is reproduced here rather than resolved. Rule 91 applies to the location codes.

    Evidence that argues against this (1)
    • UniProt's own record for HDGF contains directly CONTRADICTORY function comments sourced to the same publication (PubMed:26845719): one states HDGF has mitogenic activity for fibroblasts and binds heparin, another states it does not. The name 'hepatoma-derived GROWTH FACTOR' therefore rests on a contested activity, and any inference from the growth-factor label to HDGF's nuclear behaviour is unsafe. source 2026
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Wang et al. found that Hdgfrp2 (HRP2) knockout mice developed normally and were fertile, whereas Psip1/Hdgfrp2 double-deficient mice died at approximately embryonic day 13.5 with ventricular septal defects, indicating functional redundancy between these two HDGF-family members in development.

    PMID 26367869DOI
    What this evidence cannot show

    Abstract read, not full text. Mouse, developmental, cardiac - not haematopoietic and not leukaemic. HDGF itself was NOT part of this double knockout, so the redundancy demonstrated is between PSIP1 and HDGFL2 and is only suggestive for HDGF.

    Evidence that argues against this (1)
    • This double knockout removed PSIP1 and HDGFL2 and left HDGF intact, so it demonstrates redundancy between two family members and is silent about the third. Reading it as evidence that HDGF is redundant with PSIP1 would be a paralogue-substitution error. It is also mouse, developmental and cardiac, with no haematopoietic readout reported in the abstract. source 2026
Connected to