ELP6 (TMEM103/C3orf75) - an Elongator subunit named after a membrane protein it is not, whose entire localisation record comes from a Drosophila paper its own symbol cannot retrieve

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• ELP6 is not a therapeutic target and this entry must not be read as proposing one: it has no catalytic activity, no annotated domain, and it is required in nearly every DepMap line at -0.981 across 1,208 lines.
• The direction of effect for the complex is tumour-suppressive in the two best-evidenced systems - germline ELP1 loss predisposes to paediatric medulloblastoma, conditional Elp3 loss drives p53-mutant leukaemia/lymphoma in mice - so inhibiting Elongator in a leukaemia is predicted to point the wrong way. The one pro-tumour result, ELP6 knockdown reducing melanoma tumorigenicity, is cell-line RNAi with no rescue reported.
• The fly ortholog's loss is lethal at the larval stage and the human complex's loss of function causes familial dysautonomia and neurodevelopmental disease, so systemic inhibition has a known and severe developmental phenotype.
• The identifier hazards are live reasoning hazards. An agent expanding ELP6's UniProt synonyms automatically picks up 'ATP1' and can assign Na+/K+-ATPase literature to this gene, which is the CD15/FUT4 failure this project's rule 69 documents; and an agent reasoning from 'Protein TMEM103' would place a soluble complex subunit in a membrane.
• No ELP6-directed chemical matter exists, and the absence is recorded as UNMEASURED rather than zero: the ChEMBL activity endpoint returned HTTP 500 on three attempts on 2026-09-22 and no row count was obtained.

1. Does human ELP6 really move into the nucleus on insulin stimulation? It is the only stimulus-conditional localisation in this beat, it is UniProt's only localisation evidence for the protein, and it rests on a single clause of a Drosophila paper's abstract with no human cell type, antibody or quantification named - and nothing has followed it up in fourteen years. One insulin time-course with a fractionation would confirm or delete a field on a reviewed UniProt record.
2. Is Elongator's nuclear pool metabolically gated, and does that matter in a leukaemia stem cell? ELP6 is reported to enter the nucleus on insulin, the fly ortholog sits in an InR-TOR axis with an altered triglyceride-to-protein ratio, and this graph already holds a fatty-acid-oxidation dependency in CML leukaemia stem cells through SMYD3/CPT1A. Nothing connects them and the question is cheap: does insulin or nutrient state change Elongator's nuclear occupancy in a haematopoietic cell?
3. Why do three subunits of one obligate hetero-hexamer span 0.75 Chronos units in DepMap? ELP5 -1.209, ELP6 -0.981, ELP4 -0.459. For an assembly in which each subunit is reported to be required, that spread is unexplained and is a better question than any selectivity figure from the same data.
4. Should UniProt's 'Protein TMEM103' alternative name and 'ATP1' gene synonym be reported for correction? The first asserts a membrane topology the record's own localisation comment contradicts, and the second collides with the Na+/K+-ATPase family in a way that breaks automated alias expansion. FLAGGED FOR THE PI rather than actioned, alongside the ELP4 and TRIM27 submissions already on the coordinator's list.
5. Does any structure place ELP6 near chromatin? There is no human ELP6 structure at all, and the assembled-complex structures are yeast and mouse.

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

   ELP6 (Q0PNE2, ELP6_HUMAN) is a 266-residue Elongator subunit whose UniProt feature table is COMPLETELY EMPTY - no domain, motif, repeat, zinc finger, coiled coil, DNA-binding feature, active site or binding site anywhere - with no PDB entry and only Pfam PF09807; its alternative names are Angiotonin-transactivated protein 1 and Protein TMEM103, its gene synonyms are ATP1, C3orf75 and TMEM103, it carries no TISSUE SPECIFICITY comment, and its ENTIRE subcellular-location record is Cytoplasm plus Nucleus plus the note 'Concentrates in the nucleus upon insulin stimulation', all three resting on the single citation PMID 22645656.

   UniProtKB Q0PNE2 (ELP6_HUMAN) - reviewed human record, length 266 aa · UniProt Consortium · UniProtKB 2026
   ▸What this evidence cannot show

   Accession resolved 2026-09-22 by gene_exact:ELP6 AND organism_id:9606 AND reviewed:true with gene_primary and length read back (rules 62, 65, 66): exactly one reviewed human row, primary gene name ELP6, entry ELP6_HUMAN, 266 aa. Evidence codes: the tRNA-modification clauses are ECO:0000303 (author statement) and ECO:0000269:22854966; the cell-migration clause is ECO:0000250 (by similarity, mouse Q8BK75); the SUBUNIT clause is ECO:0000269:22854966; and ALL THREE localisation statements are ECO:0000269:22645656, a single paper - see E4. The alternative name 'Protein TMEM103' asserts a membrane topology that the record's own subcellular-location comment contradicts: it gives Cytoplasm and Nucleus and NO membrane and NO topology. The gene synonym 'ATP1' collides with the Na+/K+-ATPase subunit family and would produce a confident wrong assignment under any automated alias expansion - the failure mode this project's rule 69 documents for CD15/FUT4.

   ▸Evidence that argues against this (1)

   • An empty feature table is partly a statement about annotation effort and must not be read as 'no fold'. There is no experimental structure of human ELP6 alone, and the yeast Elp456 subcomplex is a RecA-like ATPase-fold hetero-hexamer - a fold UniProt does not annotate on this record. The defensible statement is 'no annotated module'. Jaciuk M et al. 2023
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   DERP6 and C3ORF75 encode human ELP5 and ELP6, both are essential for Elongator function, and knockdown of C3ORF75/ELP6 in melanoma-derived cells decreases migration, invasion and tumorigenicity, phenocopying depletion of Elongator through ELP1 or ELP3.

   DERP6 (ELP5) and C3ORF75 (ELP6) regulate tumorigenicity and migration of melanoma cells as subunits of Elongator. · Close P, Gillard M, et al. · J Biol Chem 2012
   PMID 22854966DOI
   ▸What this evidence cannot show

   The paper that established the identity of human ELP6, combining biochemistry with RNAi in melanoma-derived cell lines; before it, human ELP5 and ELP6 were uncharacterised. Abstract read; full text not read. The phenotypic arm is knockdown with migration, invasion and tumorigenicity endpoints and no rescue described in the abstract. Note that the paper's own title uses C3ORF75 rather than ELP6, and C3orf75[tiab] returns 2 records - so the identity paper is reachable from both symbols only because it names both.

   ▸Evidence that argues against this (1)

   • This is the only result in which an Elongator subunit is REQUIRED for a cancer phenotype, and it is the weakest of the three lineage results: human germline genetics with replication for medulloblastoma, mouse conditional knockout for haematopoiesis, RNAi in melanoma cell lines here. Waszak SM et al. 2020
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   ELP6's own gene symbols cannot retrieve the paper that is the sole source of its entire UniProt localisation record: measured 2026-09-22 with one [tiab] tag per term and querytranslation read back, ELP6[tiab] = 43, C3orf75[tiab] = 2, TMEM103[tiab] = 0, ELP6[tiab] AND insulin[tiab] = 0 and ELP6[tiab] AND Drosophila[tiab] = 0, while the source paper (PMID 22645656) is retrievable as poly[tiab] AND Elongator[tiab] = 9.

   NCBI E-utilities esearch over PubMed and GEO DataSets, run 2026-09-22 by Q98 through a wrapper that refuses any response with an empty querytranslation · Oncofusion knowledge-graph project (Q98-elongator-smyd3) · NCBI E-utilities 2026
   ▸What this evidence cannot show

   The two zeros are the load-bearing measurements and both were taken through the guarded wrapper with querytranslation asserted non-empty, which matters especially here because ELP6[tiab] itself returned 0 with an EMPTY querytranslation on the first pass. TMEM103[tiab] = 0 returns querytranslation 'TMEM103[tiab]' - the term echoed back unparsed, the signature of a term with no index entry - which is VISIBLY DIFFERENT from the empty-string failure and is how the two can be told apart. The 'poly AND Elongator' route is a demonstration that the paper is findable by SOMETHING, not a recommendation: 'poly' is a contaminated token and that query's 9 records were not all title-checked. The ATP1 synonym was deliberately NOT run as a query, because it would measure Na+/K+-ATPase contamination rather than ELP6.

   ▸Evidence that argues against this (1)

   • This measures RETRIEVABILITY and not annotation quality. UniProt's ELP6 localisation call is defensible on its content - the source paper does report a human-cell observation of insulin-dependent nuclear translocation. The two facts together are the point: an ECO:0000269 can be simultaneously the only evidence for a field and invisible to anyone who searches the gene. Bolukbasi E et al. 2012
4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   The Drosophila gene poly - identified as an Elongator small subunit by homology and the sole source of human ELP6's UniProt localisation record - is an essential mediator of insulin-receptor/TOR signalling whose loss is lethal at the third larval instar after a stage of extreme larval longevity, whose product interacts with InR, whose mutants show decreased phosphorylation of Akt, S6K and 4E-BP and a decreased triglyceride-to-protein ratio, and whose cellular distribution is insulin-dependent in BOTH Drosophila AND HUMAN cells, moving to the nucleus with insulin treatment.

   Drosophila poly suggests a novel role for the Elongator complex in insulin receptor-target of rapamycin signalling. · Bolukbasi E, Vass S, et al. · Open Biol 2012
   PMID 22645656DOI
   ▸What this evidence cannot show

   RULE 91, THE SOLE-BASIS VARIANT. This paper is the only ECO:0000269 behind all three of ELP6's subcellular-location statements, and it is a Drosophila paper about a gene called 'poly' that never names ELP6 in its title or abstract. The annotation is defensible because the paper does report a human-cell observation, and the pointer is deeply misleading. The human arm is a single sentence in the abstract with no cell type, no antibody and no quantification given, so 'ELP6 concentrates in the nucleus upon insulin stimulation' in human cells is an abstract-level statement this beat could not evaluate further; full text not read. The InR-TOR mechanism is entirely fly. Reported, not edited - this is UniProt's record.

   ▸Evidence that argues against this (1)

   • A fly gene's InR-TOR function does not transfer to human ELP6, and the 2012 identity paper published the same year frames ELP6 purely as an Elongator subunit with a migration phenotype, not as an insulin-signalling protein. Nothing has followed up the insulin-dependent translocation in human cells in the fourteen years since, and the observation is a single clause. Close P et al. 2012
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

   No paper names ELP6 and a KMT2A fusion together: measured 2026-09-22 with one [tiab] tag per term and querytranslation asserted non-empty, ELP6[tiab] = 43, ELP6[tiab] AND KMT2A[tiab] = 0, AND MLL[tiab] = 0, AND leukemia[tiab] = 0, AND leukaemia[tiab] = 0, and the mandatory complex form Elongator[tiab] AND KMT2A[tiab] = 0 and AND MLL[tiab] = 0, against positive controls through the identical shape of AFF1[tiab] AND KMT2A[tiab] = 104, DOT1L[tiab] AND KMT2A[tiab] = 30 and KMT2A[tiab] AND histone[tiab] = 218.

   NCBI E-utilities esearch over PubMed and GEO DataSets, run 2026-09-22 by Q98 through a wrapper that refuses any response with an empty querytranslation · Oncofusion knowledge-graph project (Q98-elongator-smyd3) · NCBI E-utilities 2026
   ▸What this evidence cannot show

   ELP6 IS THE SECOND PROTEIN IN THIS BEAT ON WHICH THE INSTRUMENT FAILURE FIRED. INSTRUMENT ALARM RECORDED WITH THE RESULT. On the first pass of this beat's symbol sweep, ELP3[tiab] and ELP6[tiab] each returned a well-formed esearchresult with count 0, querytranslation EMPTY STRING, and errorlist and warninglist both {}. Re-run minutes later through the identical code path they return 202 and 43. Every count in this entry was therefore taken through a guarded wrapper that REFUSES any response whose querytranslation is empty and retries. A zero with an empty querytranslation is an instrument failure, not a result, and nothing else in the response says so. Zero in TITLES AND ABSTRACTS ONLY; supporting figures and supplementary tables not searched (rule 80). Name list searched and measured: ELP6 (43), C3orf75 (2), TMEM103 (0), and the complex form Elongator (608); the ATP1 synonym was deliberately not run because it would measure Na+/K+-ATPase contamination. With a total literature of 43 records this is a small-literature absence, and the correct reading is 'almost nobody has studied this protein' rather than 'people looked and found nothing'.

   ▸Evidence that argues against this (1)

   • Grepping this project's five KMT2A-fusion interactome analyses for ELP1 through ELP6, Elongator and SMYD3 returns nothing, which is the load-bearing half of the absence. What is NOT bounded: 24 PRIDE projects mention Elongator against a KMT2A control of 37, and none was opened project by project. IntAct/EMBL-EBI 2026
6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

   IntAct REST returns 45 interaction records for ELP6 (Q0PNE2) spanning 19 distinct molecules, which include ELP2, ELP3, ELP4, ELP5 and ELP6 itself, and which include no POLR2A and none of KMT2A, MEN1, AFF1, AFF4, MLLT1, MLLT3, MLLT10, DOT1L, PSIP1, CDK9, ELL, KAT7, WDR5 or HOXA9.

   IntAct molecular interaction database, REST endpoint findInteractions (not PSICQUIC) · IntAct/EMBL-EBI · IntAct 2026
   ▸What this evidence cannot show

   Queried 2026-09-22 through IntAct REST findInteractions and NOT PSICQUIC (rule 93), with the documented control findInteractions/Q03164 returning 376 records. At 19 distinct partners this absence carries little information and is recorded as such - a bait with 19 curated partners has not been looked at. The recovery of five of six sibling subunits confirms the query reached the right protein. Partner names read from moleculeA/moleculeB gene-symbol fields across paged results. Note ELP1 is NOT among the recovered partners, which at this curation depth is not evidence of anything.

   ▸Evidence that argues against this (1)

   • An IntAct absence is 'measured-not-curated' for any partner appearing only in an uncurated proteome, and this project has established that the index fusion's own AP-MS interactome (PMID 29777171) has never been entered into IntAct. For a 19-partner bait that limitation dominates. IntAct/EMBL-EBI 2026
7. E8established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   Elongator comprises the catalytic Elp123 subcomplex and the associated Elp456 hetero-hexamer, whose fully assembled structure was solved by cryo-EM in yeast at 4.3 Angstrom and in mouse at 5.9 Angstrom with mutational validation, alongside biochemical characterisation of the assembled human complex.

   Cryo-EM structure of the fully assembled Elongator complex. · Jaciuk M, Scherf D, et al. · Nucleic Acids Res 2023
   PMID 36617428DOI
   ▸What this evidence cannot show

   Yeast and mouse structures; the human arm is biochemical, so this is not a human structure of ELP6 and none exists. Abstract read; full text not read. The substrate throughout is tRNA. Cited here for the hetero-hexamer architecture that is the only structural handle on a protein whose own feature table is empty.

   ▸Evidence that argues against this (1)

   • Architecture inferred for human ELP6 from yeast and mouse structures is architecture by homology, and this entry asserts no residue coordinate for ELP6 on that basis - it has none to assert. UniProt Consortium 2026
8. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

   The Elongator complex has cancer phenotypes in three lineages pointing in two directions: germline ELP1 loss-of-function predisposes to paediatric Sonic-Hedgehog medulloblastoma with biallelic inactivation, conditional Elp3 loss in mouse haematopoiesis causes marrow failure and then p53-mutated leukaemia/lymphoma, and ELP5/ELP6 knockdown reduces melanoma migration and tumorigenicity.

   Germline Elongator mutations in Sonic Hedgehog medulloblastoma. · Waszak SM, Robinson GW, et al. · Nature 2020
   PMID 32296180DOI
   ▸What this evidence cannot show

   A three-paper synthesis assembled by this beat, with the medulloblastoma paper cited as the anchor; the other two arms are PMIDs 33507234 and 22854966 and are cited in full in gene-elp3 and gene-elp5. The synthesis is the beat's own and is labelled as such: the three results are in different species, different lineages and different experimental classes, and no paper puts them together. The evidence QUALITY is not symmetric - human germline genetics with replication, mouse conditional knockout, and cell-line RNAi respectively - so 'two directions' should not be read as two equally supported directions.

   ▸Evidence that argues against this (1)

   • A synthesis across three lineages and three species is exactly the reasoning this project's rule 73 warns about, and it is recorded as inference rather than as a finding. The specific risk is that 'Elongator is tumour-suppressive' is being generalised from brain and mouse blood to human leukaemia, where nobody has measured it - and that generalisation is the beat's own recommendation for the next experiment, not a result. Close P et al. 2012
9. E12_HAT_YEASTestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

   In YEAST, intact Elongator is a histone H3 and H4 acetyltransferase whose predominant sites are H3K14 and H4K8, it acetylates core and nucleosomal substrates where isolated Elp3 is not specific, the three smallest subunits Elp4, Elp5 and Elp6 are REQUIRED for that HAT activity, and multiply-acetylated H3 and H4 are reduced in vivo in ELP3-null cells.

   Elongator is a histone H3 and H4 acetyltransferase important for normal histone acetylation levels in vivo. · Winkler GS, Kristjuhan A, et al. · Proc Natl Acad Sci U S A 2002
   PMID 11904415DOI
   ▸What this evidence cannot show

   Saccharomyces cerevisiae throughout - not human, not haematopoietic, not a fusion system (this project's rule that the system belongs in the claim, not only the caveat). Abstract read; full text not read. The in vivo arm is a chromatin immunoprecipitation of bulk acetylated H3/H4 in an ELP3 deletion, so it establishes that Elongator contributes to steady-state acetylation levels and not that Elongator is the writer at any named gene. The requirement for the small subunits is a requirement for HAT ACTIVITY OF THE COMPLEX and does not make ELP4, ELP5 or ELP6 catalytic.

   ▸Evidence that argues against this (1)

   • The same paper's own framing has been overtaken. UniProt's current human ELP3 record renames the protein 'tRNA uridine(34) acetyltransferase' and states that protein-lysine acetyltransferase activity is unclear in vivo; the 2018 structural review calls the ELP3 HAT domain putative. A 2002 yeast result is not withdrawn by a 2026 database opinion, but a beat that cites only this paper is quoting the high-water mark of a claim that has receded. Dalwadi U et al. 2018
10. E12_HAT_HUMANstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In HUMAN HeLa cells the purified Elongator complex exists in two forms and only the six-subunit holo-Elongator has histone acetyltransferase activity against H3 and H4: the three-subunit core form has NO HAT activity despite containing the catalytic ELP3 subunit, and Elongator is a component of early elongation complexes formed in HeLa nuclear extracts and interacts directly with RNA polymerase II in solution.

    Purification and characterization of the human elongator complex. · Hawkes NA, Otero G, et al. · J Biol Chem 2002
    PMID 11714725DOI
    ▸What this evidence cannot show

    Biochemical purification from HeLa; abstract read, full text not read. This is the paper that reassigned two proteins to Elongator - it identifies StIP1 (ELP2) and IKAP (ELP1) as subunits of the human complex - so it is simultaneously the source of the small-subunit requirement in human cells and the source of two of the identity traps recorded in this beat. It is ALSO the resolution of a disagreement this graph already records: gene-elp4 notes that a contemporaneous human purification recovered only three polypeptides WITH HAT activity (PMID 11818576) and leaves the disagreement open; this paper's three-subunit core has NO HAT activity. Both are 2002, both are human, and they disagree. Recorded as a live disagreement, not as a resolution in either direction.

    ▸Evidence that argues against this (1)

    • THE DIRECT CONTRADICTION, and it is not resolved. A human Elongator purified from HeLa the same year was composed of THREE polypeptides and DID have HAT activity specific to histone H3, facilitating Pol II transcription in a chromatin- and acetyl-CoA-dependent manner. Complex composition determined by purification depends on the tag, the salt and the cell type, so 'three subunits, no HAT' and 'three polypeptides, HAT' are both properties of preparations. No later human paper this beat could find repeats either HAT assay. Kim JH et al. 2002
11. E12_TRNA_NOWstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The current curated and structural position is that Elongator is a tRNA wobble-uridine modification enzyme: UniProt names human ELP3 'tRNA uridine(34) acetyltransferase' (EC 2.3.1.311) and states that its protein-lysine acetyltransferase activity is unclear in vivo, and every high-resolution structure of human Elongator is of the ELP123 subcomplex bound to tRNA and cofactors, resolving four stages of the U34 acetylation reaction with no histone in the series.

    Cryo-EM structures of the human Elongator complex at work. · Abbassi NE, Jaciuk M, et al. · Nat Commun 2024
    PMID 38750017DOI
    ▸What this evidence cannot show

    Cryo-EM to 2.9 Angstrom of human ELP123 with tRNA; abstract read, full text not read. An absence of histone in a structural series is not evidence that histones are not substrates - nobody put one in - so this bounds what the human structural work SHOWS and not what Elongator can do. The UniProt FUNCTION clause quoted here is coded ECO:0000269 to PMID 29415125 for the tRNA half, while the replacement catalytic activity and the EC number are ECO:0000250 BY SIMILARITY from a non-human ortholog: UniProt renamed the human protein on an inferred activity, which is worth knowing before quoting the new name as settled.

    ▸Evidence that argues against this (1)

    • ELP3 does demonstrably acetylate at least one non-histone PROTEIN in vivo: in Drosophila neurons ELP3 is necessary and sufficient to acetylate the ELKS-family protein Bruchpilot at presynaptic densities, with a structural and a neurotransmitter-release phenotype. So 'primarily a tRNA enzyme' is a statement about the dominant activity, not a demonstration that the GNAT pocket cannot take a protein substrate. Note also that UniProt cites this same paper for 'Not a major acetyltransferase of tubulin', a clause its abstract does not contain - read the body before quoting that negative. Miśkiewicz K et al. 2011
12. E7weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.dataset

    ELP6 shows no KMT2A-selective dependency in this project's own DepMap scan: mean Chronos -1.0500 in 11 KMT2A-rearranged lines against -0.9655 in 43 other leukaemia lines, difference -0.0845, at a pan-essentiality of -0.9812 across 1208 lines and selectivity rank 1575 of 17,728.

    DepMap CRISPR (Chronos) gene-effect scan of 11 KMT2A-rearranged versus 43 other leukaemia lines (this project's own genome-wide scan, analyses/depmap-kmt2a-selective-scan-full) · Oncofusion knowledge-graph project · analyses/depmap-kmt2a-selective-scan-full/results.json 2026
    ▸What this evidence cannot show

    Joined 2026-09-22 from analyses/depmap-kmt2a-selective-scan-full/results.json (generated 2026-09-19T10:12:06Z). NOT re-fetched and NOT independent of that scan: every caveat it carries applies, including that the 11 KMT2A-rearranged lines come from nine individuals and that the arm is myeloid-enriched, so a difference there mixes fusion status with lineage. The scan records a generation timestamp and NOT a DepMap RELEASE, which is this project's rule-83 exposure carried forward rather than fixed here. Read the numbers by this project's own DepMap rules: a rank is not an effect size, a difference is not a dependency, the anti-selective half is 50 percent of any gene set by construction with this graph's base rate at 45.3 percent, and a gene depleted this deeply across all lines is FLOOR-COMPRESSED - there is almost no dynamic range left in which selectivity could appear, so a small arm-to-arm difference is uninterpretable rather than small. The screen is also leukaemia lines only.

    ▸Evidence that argues against this (2)

    • A difference of -0.0845 at a pan-essentiality of -0.981 across 1208 lines is floor-compressed and uninterpretable, not small: a gene required in nearly every line has no dynamic range left in which KMT2A selectivity could show. Reporting the rank would be reporting the null. Oncofusion knowledge-graph project 2026
    • Within the complex: ELP5 -1.209, ELP1 -1.022, ELP6 -0.981, ELP3 -0.965, ELP2 -0.627, ELP4 -0.459 across 1,208 lines. ELP6 is third-deepest, and the three accessory subunits span 0.75 Chronos units despite forming one obligate hetero-hexamer - an unexplained spread that is a better question than any selectivity figure derived from these data. Oncofusion knowledge-graph project 2026

• ← part ofELP5 (DERP6) - the bridge that joins ELP3 to ELP4, the most essential Elongator subunit in DepMap, and a gene whose cytoplasmic call rests on an overexpression experiment

  both in the accessory ELP4/5/6 hetero-hexamer; identified together as human ELP5 and ELP6 in one 2012 paper

• part of →ELP4 (PAXNEB) - a required subunit of Elongator, the one partner of the seven with a genuine histone-acetyltransferase connection, and the one whose fusion supplies only its last 43 residues

  both in the accessory ELP4/5/6 hetero-hexamer (Elp456), and both explicitly named as required for the complex HAT activity in yeast

• part of →ELP5 (DERP6) - the bridge that joins ELP3 to ELP4, the most essential Elongator subunit in DepMap, and a gene whose cytoplasmic call rests on an overexpression experiment

  identified together as human ELP5 and ELP6 in one 2012 paper; both depletions phenocopy Elongator loss in melanoma cells

• related to →KMT2A / MLL1 — the gene that breaks

  BOUNDED ABSENCE, small-literature, and the second protein in this beat on which the empty-querytranslation instrument failure fired (ELP6[tiab] returned 0 before returning 43)

• tests →Hypothesis: transfer-RNA and ribosome biology is a genuinely empty cell in KMT2A-rearranged leukemia, and translational capacity may be the unmeasured dependency

  subunit of the tRNA wobble-modification complex, and the one subunit whose nuclear pool is reported to be metabolically gated (insulin-dependent nuclear translocation)

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

  accessory subunit of the complex named for elongating Pol II; ELP6 has no curated POLR2A contact of its own and no annotated domain at all