Oncofusion

start / graph / gene-epb41l3

Proteinmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q99-interactome-read-queue

EPB41L3 / DAL-1 - a membrane-skeleton tumour suppressor that binds and modulates two arginine methyltransferases, and whose own promoter methylation is a clinical test

also called DAL1 · DAL-1 · 4.1B · KIAA0987 · Band 4.1-like protein 3 · Differentially expressed in adenocarcinoma of the lung 1 · E41L3_HUMAN

EPB41L3 is a cytoskeletal scaffold that also acts as a tumour suppressor, and it grips two enzymes that put methyl groups on proteins - turning one of them down. Its own gene is switched off by DNA methylation in many cancers, which is why a methylation test for it is used in cervical-cancer screening.

The same thing, for a scientist

EPB41L3 (Q9Y2J2, 1,087 aa, HGNC:3380, 18p11.31) is a protein 4.1 superfamily member with a FERM domain (110-391), a spectrin-actin-binding region (514-860) and a C-terminal domain (861-1083), annotated by UniProt as a tumour suppressor that inhibits proliferation, promotes apoptosis and **modulates the activity of protein arginine N-methyltransferases including PRMT3 and PRMT5** [E1]. Those two claims are experimental and specific: the DAL-1/4.1B FERM domain binds PRMT3 directly and **inhibits its methyltransferase activity in vitro and in vivo** [E2], and DAL-1/4.1B modulates PRMT5 **substrate-specifically** - inhibiting methylation of Sm proteins while enhancing methylation of myelin basic protein [E3]. **No histone substrate was assayed in either paper**, which is exactly where the reopening condition sits. Its own locus is a heavily used epigenetic biomarker: `EPB41L3[tiab] AND methylation[tiab]` returns 60 records and `AND cervical[tiab]` returns 43 [E4]. Its interactome evidence is thin and bait-side: five IntAct rows over two publications (BioPlex 3.0 in HEK293T, and a 2013 Hippo-pathway interaction proteome in HEK293), recovering AFF1, AFF4, MLLT1 and MLLT3 - four subunits of one complex - with EPB41L3 the bait every time [E5]. It is not a DepMap dependency in any line (mean gene effect +0.0873) [E6].

What it looks like
EPB41L3 / DAL-1 - a membrane-skeleton tumour suppressor that binds and modulates two arginine methyltransferases, and whose own promoter methylation is a clinical test1,087 aa
FERM (4.1/Ezrin/Radixin/Moesin)FERM (4.1/Ezrin/Radixin/Moesin) · residues 110–391Spectrin-actin-binding regionSpectrin-actin-binding region · residues 514–860CTD (C-terminal domain)CTD (C-terminal domain) · residues 861–108302004006008001,000

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 curated evidence actually is - and it is the thinnest arm of this tranche

Five IntAct binary rows, two interaction records, two publications, fetched from IntAct REST 2026-09-22 5:

publicationwhat it iscell linebaits recovered
PMID 33961781BioPlex 3.0, AP-MSHEK293TAFF1, AFF4, MLLT1, MLLT3
PMID 24366813Hippo-pathway interaction proteome, AP-MSHEK293MLLT3

Two publications, two platforms, and the second is not about elongation at all. PMID 24366813 is an interaction-proteome map of human Hippo signalling built around the co-activator YAP1 and cell-polarity complexes 5. EPB41L3 belongs in that study because it is a FERM-domain polarity protein; MLLT3/AF9 appearing in its pulldown there is a single row in a 480-interaction network, and the paper makes no claim about it.

AFF1, AFF4, MLLT1 and MLLT3 are four subunits of one complex. One contact with the super elongation complex reported four times is the parsimonious reading, and EPB41L3 is the bait in all five rows - nobody has pulled on a SEC component and recovered EPB41L3 5.

All five rows are MI:0007 anti-tag co-immunoprecipitation typed MI:0914 association, spoke-expanded, MI-scores 0.35-0.53. None is a binding event. And EPB41L3 is a cytoplasmic, cytoskeletal, cell-junction protein with no reported nuclear pool at all 1 - which makes co-purification with nuclear elongation factors in an overexpression AP-MS the least credible arm of this tranche.

Why it is written anyway, when CAMKV and EPB41L1 were not: the arginine-methyltransferase arm 23 and the promoter-methylation arm 4 are both real, both epigenetic, and both independent of the AP-MS rows. The entry exists for that biology, and it says plainly that the interactome row is weak.

Epigenetics: two arginine methyltransferases modulated, and no histone substrate tested

This is the case the dispatch brief warned against defaulting to not-applicable, and the default would have been wrong twice over.

Arm 1 - EPB41L3 modulates arginine methyltransferase ACTIVITY.

enzymewhat DAL-1/4.1B doesevidence
PRMT3binds directly via the FERM domain; inhibits methyltransferase activity in vitro and in vivoyeast two-hybrid with FERM as bait, co-IP in lung and breast cancer lines, in vitro binding and methylation assays 2
PRMT5modulates substrate-specifically: inhibits methylation of Sm proteins, enhances methylation of myelin basic proteinin vitro and in cells 3
PRMT6interacts (UniProt SUBUNIT, citing the PRMT3 paper)12

The limit is exact and it is where the next experiment goes: no HISTONE substrate was assayed in either paper. PRMT5 is a genuine histone arginine methyltransferase - H4R3me2s and H3R8me2s - and PRMT3's principal characterised substrate is a ribosomal protein, not a histone. So EPB41L3 is established as a regulator of arginine methyltransferases and is not established as affecting any histone mark. Writing it either way without that distinction would be wrong.

Arm 2 - EPB41L3's own locus is an epigenetic READOUT, and a clinical one. EPB41L3[tiab] AND methylation[tiab] = 60 and EPB41L3[tiab] AND cervical[tiab] = 43 4: DNA methylation of the EPB41L3 promoter is a widely replicated cancer biomarker and is used in cervical-cancer triage panels. That is the gene as a target of DNA methylation, not the protein as a writer - the same cause-or-consequence distinction this graph already draws carefully for CLEC12A.

Domain architecture, fetched live, confirming no direct activity. Q9Y2J2 has a FERM domain at 110-391, a hydrophilic region 394-513, a spectrin-actin-binding region 514-860 and a C-terminal domain 861-1083. There is no SET, JmjC, HAT, HDAC, bromo, chromo, Tudor, PHD, CXXC, YEATS or PWWP module and no annotated DNA-binding region 1. EPB41L3 catalyses no modification itself.

Reopening condition, and it is one experiment: an in vitro PRMT5 methylation assay on nucleosomes or histone H4 with and without recombinant DAL-1/4.1B FERM, reading H4R3me2s. Given that DAL-1/4.1B already inhibits PRMT5 on one substrate and enhances it on another, the histone answer cannot be predicted and has never been measured.

Localization: no nuclear pool, and that is the strongest argument against the interactome row

Q9Y2J2 carries four subcellular-location comments 1:

locationevidence codesource
Cell junctionECO:0000269PubMed 9892180
Cell membrane, peripheralECO:0000269PubMed 9892180
CytoplasmECO:0000269PubMed 9892180
Cytoplasm, cytoskeletonECO:0000250by similarity

There is no nucleus and no nucleoplasm comment, coded or uncoded. The location note adds only that it is "detected in the cytoplasm of actively dividing cells" 1. And the literature agrees: EPB41L3[tiab] AND nucleus[tiab] = 0 4.

So the AP-MS rows ask a reader to accept co-purification of a peripheral-membrane, junctional, cytoskeletal protein with nuclear transcription-elongation factors. In an overexpression pulldown in HEK293T that is exactly the geometry in which abundant, sticky, coiled-coil-rich nuclear complexes produce false positives.

Where the PRMT interaction happens is therefore the open question, not a detail. PRMT3 is predominantly cytoplasmic, which makes a cytoplasmic FERM-PRMT3 contact coherent. PRMT5 shuttles. If EPB41L3 never enters the nucleus, it can regulate an arginine methyltransferase's pool and activity without ever touching chromatin - which is a real mechanism and a different one from chromatin regulation.

Subnuclear: not applicable rather than unpublished, because no nuclear pool is reported. Stated as such rather than left blank.

Cell type and state. Tissue specificity is high in brain, lower in kidney, intestine and testis, detected in lung 1. The PRMT work is in lung and breast cancer lines 23; the interaction rows are HEK293T and HEK293 5; the methylation-biomarker literature is cervical, lung and other epithelial cancers 4. EPB41L3[tiab] AND leukemia[tiab] = 2 and DAL1[tiab] AND leukemia[tiab] = 0 4. No measurement in any KMT2A-rearranged or B-lineage cell.

Analogous biology in other fusion families
  • The arginine-methyltransferase axis is shared with the fusion literature, from the other direction. PRMT5 is an established dependency and drug target in haematological malignancy, and arginine methylation of spliceosomal Sm proteins - the exact substrate DAL-1/4.1B inhibits PRMT5 on 3 - is the mechanism linking PRMT5 inhibition to splicing collapse. So EPB41L3 is a natural, unstudied modulator of a pathway this field is already drugging. No PRMT5 or PRMT3 entry exists in this graph, which this entry names as a gap rather than fills.
  • Promoter hypermethylation as a tumour-suppressor mechanism is the epigenetic route that fusion-driven leukaemias share with epithelial cancers, and EPB41L3 is one of the best-documented examples of it in any gene 4. Whether the EPB41L3 locus is methylated in KMT2A-rearranged leukaemia has never been asked, and the answer is sitting in existing methylation-array data - a measured-not-analysed absence, closable by compute alone.
  • The paralogue family is a trap rather than an analogy. EPB41, EPB41L1, EPB41L2, EPB41L3, EPB41L4A/B and EPB41L5 are six FERM-domain membrane-skeleton paralogues with overlapping aliases (4.1R, 4.1N, 4.1G, 4.1B). EPB41L1 was in the same read-queue tranche and was rejected; the two are not interchangeable and a claim about one is not a claim about the other.
The identifier trap that would have cost this entry its epigenetics

A search on the HGNC-approved symbol finds NONE of the arginine-methyltransferase work. Measured 2026-09-22, each querytranslation read back 4:

queryrecords
EPB41L3[tiab] AND PRMT[tiab]0
EPB41L3[tiab] AND arginine methyltransferase[tiab]0
4.1B[tiab] AND PRMT5[tiab]1 (PMID 15737618)
EPB41L3[tiab]120
EPB41L3[tiab] AND methylation[tiab]60
EPB41L3[tiab] AND histone[tiab]1

The approved symbol returns zero and the alias returns the paper. Both founding PRMT papers call this protein DAL-1/4.1B and never EPB41L3, because they predate the symbol's adoption 23. An agent that had run only the approved symbol - which is what the instruction to use HGNC symbols produces - would have recorded epigenetic_readout: not applicable for a protein that binds and inhibits an arginine methyltransferase. That is the failure mode the brief names, caught by running the alias.

And note HGNC records prev_symbol: null for this gene: DAL1, 4.1B and KIAA0987 are alias symbols, not previous official symbols 1. So the standing instruction to run the previous symbol as a control would not have fired here either - the control that mattered was the ALIAS control, on a gene with no previous symbol at all.

Two more, recorded for the registry:

  • 4.1B is a fragile query token: it is a decimal-containing string that also matches figure and table labels ("Figure 4.1B", "Table 4.1B") in full-text indexes. It is safe in [tiab] on PubMed and should not be used against a full-text index without inspection.
  • DAL1 also names unrelated things and returns zero in a leukaemia-scoped query (DAL1[tiab] AND leukemia[tiab] = 0), so a negative on that alias is uninformative rather than reassuring 4.
Facts and figures
uniprotQ9Y2J2
length AA1087
molecular weight da120678
uniprot IDE41L3_HUMAN
gene primaryEPB41L3
hgnc IDHGNC:3380
cytoband18p11.31
ensembl gene IDENSG00000082397
alias control runCRITICAL AND IT CHANGED THE ENTRY. EPB41L3 has NO previous official symbol, so the standing previous-symbol control does not apply; the control that mattered was the ALIAS control. EPB41L3[tiab] AND PRMT[tiab] = 0 and EPB41L3[tiab] AND arginine methyltransferase[tiab] = 0, while 4.1B[tiab] AND PRMT5[tiab] = 1 (PMID 15737618). Both founding arginine-methyltransferase papers call this protein DAL-1/4.1B and are wholly unreachable from the approved symbol. An agent running only the HGNC symbol would have written epigenetic_readout: not applicable for a protein that binds and inhibits an arginine methyltransferase.
paralogue trapSix FERM-domain membrane-skeleton paralogues with overlapping numeric aliases: EPB41 (4.1R), EPB41L1 (4.1N, Q9H4G0, 881 aa), EPB41L2 (4.1G), EPB41L3 (4.1B, Q9Y2J2, 1,087 aa), EPB41L4A/B and EPB41L5. In this beat an IntAct query was first run with Q9Y2J2 mislabelled as EPB41L1 from recall; the accessions were then resolved programmatically from UniProt gene_exact and HGNC. The 206-residue length difference between Q9H4G0 and Q9Y2J2 would have been caught by a length check, but the ALIAS confusion (4.1N versus 4.1B) would not.
domain noteAlso a hydrophilic region at 394-513. Coordinates are on the UniProt canonical frame (Q9Y2J2-1, 1,087 aa); four isoforms are annotated and RefSeq was not checked.
localizationCell junction (ECO:0000269, PubMed 9892180), cell membrane as a peripheral membrane protein (ECO:0000269, same source), cytoplasm (ECO:0000269, same source) and cytoplasm-cytoskeleton (ECO:0000250, by similarity). Location note: detected in the cytoplasm of actively dividing cells. THERE IS NO NUCLEUS COMMENT, CODED OR UNCODED, and EPB41L3[tiab] AND nucleus[tiab] returns 0.
localization evidence codeECO:0000269 x3, all from one 1999 source (PubMed 9892180); cytoskeleton ECO:0000250 only
subnuclear localizationNOT APPLICABLE rather than unpublished - no nuclear pool is reported for this protein in UniProt or in the literature searched, so there is no subnuclear compartment to assign. Stated explicitly rather than left blank. This is also the strongest argument against the AP-MS rows that placed EPB41L3 with nuclear elongation factors.
localization shown for fusionNO - nobody has looked in any fusion, and the protein has no reported nuclear pool in which a fusion could be encountered.
tissue specificityHigh in brain, lower in kidney, intestine and testis, detected in lung (UniProt Q9Y2J2, fetched 2026-09-22).
cell type absenceNo EPB41L3 measurement in any KMT2A-rearranged or B-lineage cell. EPB41L3[tiab] AND leukemia[tiab] = 2 (neither in a fusion context), DAL1[tiab] AND leukemia[tiab] = 0.
epigenetic readoutNOT A WRITER, ERASER OR READER ITSELF - AND EMPHATICALLY NOT NOT-APPLICABLE, ON TWO SEPARATE ARMS. Domain architecture fetched live 2026-09-22: UniProt Q9Y2J2 is 1,087 aa with a FERM domain (110-391), a spectrin-actin-binding region (514-860) and a CTD (861-1083), and NO SET, JmjC, HAT, HDAC, bromo, chromo, Tudor, PHD, CXXC, YEATS or PWWP module and no annotated DNA-binding region, so it catalyses no modification. ARM 1 - IT MODULATES TWO ARGININE METHYLTRANSFERASES. The DAL-1/4.1B FERM domain binds PRMT3 directly and INHIBITS its methyltransferase activity in vitro and in vivo (PMID 15334060), and DAL-1/4.1B modulates PRMT5 SUBSTRATE-SPECIFICALLY - inhibiting methylation of Sm proteins and ENHANCING methylation of myelin basic protein (PMID 15737618); UniProt also lists a PRMT6 interaction. THE EXACT LIMIT: NO HISTONE SUBSTRATE WAS ASSAYED IN EITHER PAPER. PRMT5 is a genuine histone arginine methyltransferase (H4R3me2s, H3R8me2s) and PRMT3's principal characterised substrate is a ribosomal protein, so EPB41L3 is established as a REGULATOR OF ARGININE METHYLTRANSFERASES and is NOT established as affecting any histone mark. Because the PRMT5 effect reverses between substrates, the histone answer cannot be extrapolated in either direction. ARM 2 - ITS OWN LOCUS IS A CLINICAL EPIGENETIC READOUT. EPB41L3 promoter DNA methylation is a widely replicated cancer biomarker used in cervical-cancer triage: EPB41L3[tiab] AND methylation[tiab] = 60 of 120 total records, AND cervical[tiab] = 43. That is the gene as a TARGET of DNA methylation, not the protein as a writer - the same cause-or-consequence distinction this graph draws for CLEC12A. BOUNDED, AND THE APPROVED SYMBOL IS THE WRONG INDEX: EPB41L3[tiab] AND PRMT[tiab] = 0, AND arginine methyltransferase[tiab] = 0, AND histone[tiab] = 1, AND KMT2A[tiab] = 0, AND MLL[tiab] = 0, AND nucleus[tiab] = 0, against 4.1B[tiab] AND PRMT5[tiab] = 1 and EPB41L3[tiab] = 120; control H3K79me2[tiab] = 224. REOPENING CONDITION, ONE EXPERIMENT: an in vitro PRMT5 methylation assay on nucleosomes or histone H4, plus or minus recombinant DAL-1/4.1B FERM, reading H4R3me2s. Never measured.
epigenetic mark namedH4R3me2s and H3R8me2s are the marks AT RISK (PRMT5's histone products) and neither has been measured under DAL-1/4.1B. The measured substrates are Sm proteins and myelin basic protein, which are not histones. Separately, DNA methylation of the EPB41L3 promoter is a measured readout ON this gene.
shown for fusionNO - nobody has looked in any fusion, and the protein has no reported nuclear pool.
absence classmeasured-not-analysed
absence class notemeasured-not-analysed applies specifically to the open question of whether the EPB41L3 locus is methylated in KMT2A-rearranged leukaemia: methylation-array data on KMT2A-rearranged cohorts is public, the probe set covers this promoter, and the analysis has never been run. The PRMT5-on-histone question is never-measured.
cell typeMeasured in lung and breast cancer cell lines for the PRMT3/PRMT5 modulation, HEK293T and HEK293 for the AP-MS rows, and cervical, lung and other epithelial cancers for the promoter-methylation biomarker literature. No measurement in any KMT2A-rearranged or B-lineage cell.
cell stateState-dependent in two senses: UniProt records EPB41L3 'detected in the cytoplasm of actively dividing cells', and the gene is transcriptionally SILENCED by promoter DNA methylation in a large fraction of epithelial cancers - so its protein-level presence is a function of the cell's methylation state. Neither has been assessed in a KMT2A-rearranged cell.
domains
FERM (4.1/Ezrin/Radixin/M…110.0FERM (4.1/Ezrin/Radixin/Moesin): 110Spectrin-actin-binding re…514.0Spectrin-actin-binding region: 514CTD (C-terminal domain)861.0CTD (C-terminal domain): 861
hgnc alias symbols
[
  "DAL1",
  "KIAA0987",
  "4.1B"
]
query token hazards
[
  "4.1B is a decimal-containing token that also matches figure and table labels such as 'Figure 4.1B' in full-text indexes; it is safe in [tiab] on PubMed and must be inspected before use against a full-text index.",
  "DAL1 names unrelated things and returns 0 in a leukaemia-scoped query, so a negative on that alias is uninformative rather than reassuring."
]
domains — the numbers behind the chart
[
  {
    "name": "FERM (4.1/Ezrin/Radixin/Moesin)",
    "start": 110,
    "end": 391
  },
  {
    "name": "Spectrin-actin-binding region",
    "start": 514,
    "end": 860
  },
  {
    "name": "CTD (C-terminal domain)",
    "start": 861,
    "end": 1083
  }
]
cell types measured
[
  "Lung and breast cancer cell lines - the PRMT3 and PRMT5 modulation work",
  "HEK293T - BioPlex 3.0 interaction rows",
  "HEK293 - the Hippo-pathway interaction proteome row",
  "Cervical, lung and other epithelial cancers - the promoter-methylation biomarker literature",
  "DepMap pan-cancer panel, 1,208 lines"
]
prmt axis
{
  "PRMT3": "Direct FERM-domain binding; INHIBITS methyltransferase activity in vitro and in vivo (PMID 15334060). PRMT3 is predominantly cytoplasmic and its principal substrate is a ribosomal protein, coherent with EPB41L3 having no nuclear pool.",
  "PRMT5": "Substrate-specific modulation: INHIBITS methylation of Sm proteins, ENHANCES methylation of myelin basic protein (PMID 15737618). PRMT5 is a histone arginine methyltransferase and an established haematological drug target; no histone substrate was tested.",
  "PRMT6": "Interaction listed in UniProt SUBUNIT citing PMID 15334060; no activity effect reported.",
  "graph_gap": "Neither gene-prmt3 nor gene-prmt5 nor gene-prmt6 exists in this graph as of 2026-09-22. This entry names that gap and does not fill it."
}
intact evidence
{
  "n_binary_rows": 5,
  "n_interactions": 2,
  "n_publications": 2,
  "publications": [
    "33961781",
    "24366813"
  ],
  "platform_note": "33961781 is BioPlex 3.0 (HEK293T). 24366813 is Hauri S et al., Mol Syst Biol 2013, an interaction proteome of human Hippo signalling (480 interactions, 270 components) in HEK293, in which EPB41L3 belongs as a FERM-domain polarity protein; MLLT3 is one row with no claim attached. Two platforms, both overexpression AP-MS in embryonic-kidney lines.",
  "baits_recovered": [
    "AFF1",
    "AFF4",
    "MLLT1",
    "MLLT3"
  ],
  "baits_are_one_complex": "AFF1, AFF4, MLLT1 and MLLT3 are four subunits of the super elongation complex; counting them as four partners double-counts one complex.",
  "direction": "EPB41L3 is the BAIT in all five rows. No SEC or core-scaffold protein has ever been used as a bait to find EPB41L3.",
  "method": "anti tag coimmunoprecipitation (MI:0007), typed association (MI:0914), spoke-expanded, MI-scores 0.35-0.53",
  "not_a_binding_event": "No MI:0407 direct-interaction record and no affinity constant.",
  "hosts": [
    "Homo sapiens HEK293T embryonic kidney cell",
    "Homo sapiens HEK293 embryonic kidney cell"
  ],
  "self_assessment": "THE WEAKEST ARM OF THIS ENTRY, and the entry is written for the PRMT and promoter-methylation biology rather than for these rows. A peripheral-membrane, junctional, cytoskeletal protein with no reported nuclear pool co-purifying with nuclear elongation factors in an overexpression pulldown is the classic AP-MS false-positive geometry."
}
depmap
{
  "scan": "analyses/depmap-kmt2a-selective-scan-full/results.json, computed 2026-09-19",
  "rank": 12731,
  "of_genes": 17728,
  "percentile": 71.813,
  "difference": 0.0287,
  "mean_kmt2a_rearranged": 0.0677,
  "mean_other_leukemia": 0.0389,
  "pan_essentiality_all_lines": 0.0873,
  "n_all_lines": 1208,
  "interpretation": "Not a dependency anywhere - all three values positive, which is the EXPECTED result for a tumour suppressor. The rank is the base rate and is not a finding. A knockout screen is also the wrong instrument for a gene silenced by promoter methylation in most lines.",
  "release_recorded": false,
  "release_note": "The stored scan records a compute date and NOT a DepMap release, so it cannot be told from quarterly drift."
}
How this could bite us back
  • EPB41L3 is a TUMOUR SUPPRESSOR and its DepMap gene effect is positive in every panel, so it is an entity to restore rather than inhibit - and there is no pharmacology for restoring a methylation-silenced tumour suppressor except broad demethylating agents, whose effects are genome-wide and unpredictable in an infant.
  • If EPB41L3 INHIBITS PRMT3 and inhibits PRMT5 on Sm proteins, then restoring EPB41L3 partially phenocopies PRMT5 inhibition on the splicing arm - which is a known mechanism of splicing collapse. A restoration strategy would carry the same on-target toxicity as a PRMT5 inhibitor without the ability to titrate it.
  • The PRMT5 effect REVERSES between substrates - inhibited on Sm proteins, enhanced on myelin basic protein. Any intervention aimed at this axis could raise arginine methylation on one substrate class while lowering it on another, and the net effect on chromatin has never been measured in any cell.
  • The interactome rationale for this protein's presence in the fusion axis is five overexpression AP-MS rows on a protein with no nuclear pool. If those rows are artefacts, the fusion connection disappears entirely and what remains is epithelial-cancer biology - the entry should be read with that possibility live.
Still open
  1. Does DAL-1/4.1B change PRMT5 activity on a HISTONE substrate? PRMT5 makes H4R3me2s and H3R8me2s, the effect reverses between the two substrates already tested, and an in vitro nucleosome methylation assay plus or minus the FERM domain would settle it in one experiment.
  2. Is the EPB41L3 promoter methylated in KMT2A-rearranged leukaemia? Public methylation-array data on KMT2A-rearranged cohorts covers this promoter and the query has never been run - a measured-not-analysed absence closable by compute alone.
  3. Where does the EPB41L3-PRMT contact happen? EPB41L3 has no reported nuclear pool, PRMT3 is predominantly cytoplasmic and PRMT5 shuttles, so a tumour suppressor may be regulating an arginine methyltransferase's activity entirely outside the nucleus - a different mechanism from chromatin regulation, and a more testable one.
  4. Do the four super-elongation-complex rows survive an endogenous pulldown? A reciprocal AP-MS with AFF4 as bait in any cell would either recover EPB41L3 or retire these rows; nobody has run the direction that matters.
  5. Given that PRMT5 inhibition is an active haematological therapeutic strategy and PRMT5's Sm-protein substrate is exactly the one DAL-1/4.1B inhibits, does EPB41L3 status modify PRMT5-inhibitor response? No PRMT5 entry exists in this graph to hang that question on.
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt Q9Y2J2 is the single reviewed human EPB41L3 record at 1,087 aa (120,678 Da) with a FERM domain (110-391), spectrin-actin-binding region (514-860) and C-terminal domain (861-1083), annotated as a tumour suppressor that inhibits proliferation, promotes apoptosis and modulates the activity of protein arginine N-methyltransferases including PRMT3 and PRMT5; it carries NO nucleus location comment of any kind, and HGNC:3380 records prev_symbol null with DAL1, 4.1B and KIAA0987 as aliases.

    What this evidence cannot show

    Retrieved 2026-09-22. Four isoforms are annotated (Q9Y2J2-1 to -4) and 1,087 aa is the canonical, so all coordinates here are on the UniProt canonical frame; RefSeq was not checked and may differ. Absence of a nucleus comment is absence of an ANNOTATION - a small nuclear pool below detection is not excluded, only unrecorded. The cytoplasm-cytoskeleton comment is ECO:0000250 (by similarity) while the other three are ECO:0000269 from one paper (PubMed 9892180), so the experimental localisation rests on a single 1999 source.

    Evidence that argues against this (1)
    • UniProt's coded experimental annotations frequently point at a paper that is not about the annotated fact, and the three experimental location comments here all derive from one 1999 paper, which means the localisation is a single-source annotation dressed as three. UniProt Consortium 2025
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The DAL-1/4.1B FERM domain binds protein arginine N-methyltransferase 3 directly and inhibits its ability to methylate substrates in vitro and in vivo - identified by yeast two-hybrid interaction cloning using the FERM domain as bait, confirmed by co-immunoprecipitation in lung and breast cancer cell lines, with direct binding shown in in vitro binding assays.

    PMID 15334060DOI
    What this evidence cannot show

    Lung and breast cancer cell lines, not leukaemia and not a KMT2A fusion. The substrates in the methylation assays are PRMT3's characterised substrates, and NO HISTONE substrate is reported - so this establishes methyltransferase regulation and not a histone-mark effect. PRMT3 is predominantly cytoplasmic and its best-characterised substrate is a ribosomal protein, which is coherent with EPB41L3 having no reported nuclear pool but makes a chromatin interpretation unsupported. A yeast two-hybrid origin. Read as abstract and title. Note this paper is unreachable from the approved gene symbol: EPB41L3[tiab] AND PRMT[tiab] returns 0.

    Evidence that argues against this (1)
    • The same laboratory's follow-up shows the effect on a second family member is not simple inhibition but substrate-specific in both directions - inhibiting PRMT5 on Sm proteins and ENHANCING it on myelin basic protein - so 'DAL-1/4.1B inhibits arginine methylation' does not generalise beyond PRMT3 and beyond the substrates tested. Jiang W et al. 2005
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    DAL-1/4.1B binding modulates PRMT5 activity in a substrate-specific manner, both in vitro and in cells: it inhibits PRMT5-dependent methylation of Sm proteins while enhancing methylation of myelin basic protein - so the direction of the effect depends on the substrate, and no histone substrate was tested.

    The tumor suppressor DAL-1/4.1B modulates protein arginine N-methyltransferase 5 activity in a substrate-specific manner. · Jiang W, Roemer ME, Newsham IF (3 authors; senior author Newsham IF) · Biochem Biophys Res Commun 2005
    PMID 15737618DOI
    What this evidence cannot show

    A short-format paper from one laboratory, the same group that reported the PRMT3 interaction, so the two results are not independent at laboratory level. The substrates assayed are Sm proteins and myelin basic protein; NO histone and NO nucleosome was assayed, so any statement about a histone arginine mark under DAL-1/4.1B is untested. Because the effect reverses between substrates, the histone answer cannot be extrapolated in either direction. Read as abstract and title. Unreachable from the approved symbol: 4.1B[tiab] AND PRMT5[tiab] = 1, EPB41L3[tiab] AND PRMT[tiab] = 0.

    Evidence that argues against this (1)
    • A substrate-specific modulation measured on Sm proteins is a splicing-machinery result, and PRMT5's symmetric dimethylation of Sm proteins is the mechanism through which PRMT5 inhibition collapses splicing in haematological malignancy - so the effect this paper reports could be read as acting on the spliceosome rather than on chromatin, which is the opposite of an epigenetic interpretation. Bai R et al. 2021
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    EPB41L3 is a heavily used DNA-methylation biomarker rather than a methylation writer - EPB41L3[tiab] AND methylation[tiab] = 60 of 120 total records and AND cervical[tiab] = 43 - while the approved symbol reaches none of the arginine-methyltransferase work (AND PRMT[tiab] = 0, AND arginine methyltransferase[tiab] = 0, against 4.1B[tiab] AND PRMT5[tiab] = 1), has no nuclear literature (AND nucleus[tiab] = 0), no KMT2A literature (AND KMT2A[tiab] = 0, AND MLL[tiab] = 0) and two leukaemia records (against DAL1[tiab] AND leukemia[tiab] = 0).

    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. Instrument controls through the identical code path: EPB41L3[tiab] = 120, H3K79me2[tiab] = 224, 4.1B[tiab] AND PRMT5[tiab] = 1 - so the zeros are not instrument failures. The 60 methylation records were not individually classified; the count is of records matching the word, and 'biomarker' is the dominant sense inferred from the 43-record cervical overlap rather than from reading all 60. These counts bound titles and abstracts only.

    Evidence that argues against this (1)
    • The zero on the approved symbol against a positive on the alias is itself the demonstration that these counts measure the index and not the biology: two founding papers establishing that this protein binds and inhibits an arginine methyltransferase are wholly invisible to EPB41L3[tiab] AND PRMT[tiab], so every other zero in this row carries the same risk. Singh V et al. 2004
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    IntAct REST serves five binary EPB41L3 rows against this graph's elongation proteins from two publications - BioPlex 3.0 (PMID 33961781, HEK293T; AFF1, AFF4, MLLT1, MLLT3) and a Hippo-pathway interaction proteome (PMID 24366813, HEK293; MLLT3) - all MI:0007 anti-tag co-immunoprecipitation typed MI:0914 association and spoke-expanded at MI-scores 0.35-0.53, with EPB41L3 the bait in every row; the four recovered proteins are four subunits of one complex.

    What this evidence cannot show

    Retrieved 2026-09-22. This is the weakest interactome arm in this tranche and is presented as such: overexpression AP-MS in two non-haematopoietic embryonic-kidney lines, recovering four subunits of one complex, with a bait that has no reported nuclear pool. PMID 24366813 is an interaction proteome of Hippo signalling in which EPB41L3 belongs as a FERM-domain polarity protein and MLLT3 is one row of a 480-interaction network with no claim attached. Counting four SEC subunits as four partners double-counts one complex. The accession was resolved from HGNC/UniProt rather than from recall, after an earlier query in this beat wrongly used Q9Y2J2 for EPB41L1.

    Evidence that argues against this (1)
    • A peripheral-membrane, cell-junction, cytoskeletal protein with no reported nuclear pool co-purifying with nuclear transcription-elongation factors in an overexpression pulldown is the classic false-positive geometry for AP-MS, and the largest proximity survey of a human cell shows recovery is organised by compartment - which here argues AGAINST the rows rather than explaining them. Go CD et al. 2021
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    EPB41L3 is not a dependency in any line: mean DepMap gene effect +0.0873 across 1,208 lines, +0.0677 in 11 KMT2A-rearranged leukaemia lines and +0.0389 in 43 other leukaemia lines - all positive, so knockout is if anything mildly favourable to fitness.

    What this evidence cannot show

    All three gene effects are positive and near zero, so the measurement is floor-compressed and the between-group difference (+0.0287, rank 12,731 of 17,728) is a difference between two nulls whose rank carries no information; the rank is not reported as a finding. A positive gene effect is also the EXPECTED result for a tumour suppressor, so this is consistent with UniProt's annotation rather than contradicting it. 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 is the wrong instrument for a gene silenced by promoter methylation in most cancer lines, because a gene already epigenetically off cannot produce a knockout phenotype - and EPB41L3 promoter hypermethylation is one of the best-replicated silencing events in epithelial cancer, so a null here may be measuring the lines' methylation state rather than the protein's dispensability. NCBI PubMed 2026
Connected to