NELF — the brake that has to come off for elongation, absent from the KMT2A literature, and less required in KMT2A-rearranged lines than in any other leukaemia
also called NELF · negative elongation factor · NELF-A/B/C-D/E · NELFA NELFB NELFCD NELFE · WHSC2 COBRA1 TH1L RDBP
NELF is the four-protein brake that holds RNA polymerase still at the start of a gene until a signal releases it. In leukaemia cells driven by a KMT2A fusion, two of its four parts are among the least-needed genes in the entire genome relative to other leukaemias — which is what you would expect if these cells are already running with the brake off.
The same thing, for a scientist
NELF is the four-subunit complex — **NELFA (Q9H3P2, 528 aa), NELFB (Q8WX92, 628 aa), NELFCD (Q8IXH7, 590 aa) and NELFE (P18615, 380 aa)** — that, with DSIF, enforces promoter-proximal pausing. The cryo-EM structure of Pol II-DSIF-NELF at 3.2 A shows NELF binding the polymerase funnel, bridging two mobile polymerase modules, contacting the trigger loop to restrain the mobility required for pause release, blocking TFIIS, and extending two flexible tentacles toward DSIF and the exiting RNA [E1]. Pause release proceeds by **PAF displacing NELF from the funnel**, so NELF — unlike DSIF — is the factor whose *removal* marks the transition [E2]. **NELF is absent from the KMT2A literature**: `(MLL OR KMT2A) AND NELF` returns 1 record, a CDK9 review; the broader alias-and-fusion-phrase query returns 2, of which the second is a proteomics methods paper benchmarking NELF, P-TEFb, PAF, SWI/SNF and MLL complexes side by side with no leukaemia result [E3][E4]. The unbiased dependency data are the only quantitative statement available, and they are striking: **NELFB ranks 17,671 of 17,728 (99.678th percentile, difference +0.2650) and NELFA ranks 17,650 (99.560th percentile, +0.2403)** — two of the sixty least KMT2A-selective genes in the genome — while NELFE (+0.0583, 86.5th) and NELFCD (-0.0208, 36.2nd) are unremarkable [E5].
The four subunits, side by side
| Subunit | UniProt | aa | Role in the paused complex | KMT2A selectivity difference | Rank / 17,728 | Pan-essentiality | Separate entry? |
|---|---|---|---|---|---|---|---|
| NELFA | Q9H3P2 | 528 | HDAg domain aa 89-248; NELF-C/D-binding aa 125-188; RNAPII-binding aa 189-248 | +0.2403 | 17,650 (99.56 pct) | -1.0384 | no - see below |
| NELFB | Q8WX92 | 628 | scaffold; largely without annotated domains | +0.2650 | 17,671 (99.68 pct) | -1.4051 | yes, gene-nelfb |
| NELFCD | Q8IXH7 | 590 | NELF-C and NELF-D are splice forms of one gene | -0.0208 | 6,425 (36.24 pct) | -0.6208 | no - see below |
| NELFE | P18615 | 380 | RRM aa 262-332 (RNA binding); coiled coil aa 7-36; 30 approximate RD dipeptide repeats aa 184-243 | +0.0583 | 15,340 (86.53 pct) | -0.4482 | yes, gene-nelfe |
Domain coordinates from UniProt records fetched for this beat 6; dependency numbers from the genome-wide scan 5.
Why only two subunits get their own entries. The granularity rule in this project says one entry per thing, and it is a good rule. It is applied here with one documented exception: NELFA and NELFCD have no distinguishing evidence of any kind in this disease — identical counted absence, no structure-function result specific to them in a leukaemia context, and dependency numbers that are either an extreme with no biological interpretation (NELFA) or entirely unremarkable (NELFCD). Writing two further entries whose every field would read 'not established' would add rows, not knowledge. Their data are recorded in the table above and in this node's data.subunits, and the decision is logged in ana-interactome4-enumeration so it can be reversed by whoever disagrees.
The finding: the pause axis is asymmetric in KMT2A-rearranged lines
Put this entry's numbers next to the DSIF entries' and a pattern appears that neither alone shows 5:
| Factor | Role at the pause | KMT2A selectivity difference | Direction |
|---|---|---|---|
| SUPT4H1 (DSIF) | retained through pause AND elongation | -0.2642 (rank 138) | more required |
| SUPT5H (DSIF) | retained through pause AND elongation | -0.1215 (rank 767) | more required |
| NELFCD | ejected at pause release | -0.0208 (rank 6,425) | neutral |
| NELFE | ejected at pause release | +0.0583 (rank 15,340) | less required |
| NELFA | ejected at pause release | +0.2403 (rank 17,650) | less required |
| NELFB | ejected at pause release | +0.2650 (rank 17,671) | less required |
The factor that stays with the polymerase into productive elongation is needed more; the factor that has to be thrown off is needed less. That is the shape an elongation-hijacked cell should have, and as far as this beat can determine nobody has reported it. It is recorded here as an observation and carried as a falsifiable hypothesis in hyp-pause-axis-asymmetry-kmt2a-r, not as a conclusion — six DepMap rows across eleven cell lines is a pattern, not a result, and the NELF subunits do not agree among themselves.
Cell type, cell state, localisation and epigenetic readout
| Axis | Documented | Where |
|---|---|---|
| Cell type | No KMT2A-rearranged system studied. Reconstituted Sus scrofa Pol II with human NELF. | 1 |
| Cell state | Not established in leukaemia. | — |
| Subcellular | Nucleus | 6 |
| Subnuclear | Not established. | — |
| Epigenetic readout | None. NELF has no histone-modifying activity. Its readout is the pausing index measured by PRO-seq or NET-seq, not a chromatin mark - and no such measurement exists for KMT2A::AFF1 target genes with NELF perturbation. | 1 |
| Position in the mechanism | Present only in the paused state; displaced by PAF at pause release | 12 |
The blank rows here are the honest answer to the beat's standing request for cell type, cell state, subnuclear localisation and epigenetic readout on every interactor. For NELF in this disease, none of them exists.
| complex name | NELF (negative elongation factor) |
|---|---|
| classification | NOT a documented interactor of KMT2A::AFF1; the pausing machinery, with a counted absence in this disease and an unexplained anti-dependency signal |
| position in mechanism | present only in the paused state; displaced from the Pol II funnel by PAF at pause release |
| epigenetic readout | none - NELF has no histone-modifying activity; the relevant readout is a pausing index from PRO-seq or NET-seq, which has never been measured with NELF perturbation in a KMT2A-rearranged cell |
| granularity exception | NELFA and NELFCD are recorded here rather than as separate nodes because every field of such a node would read 'not established'. Logged in ana-interactome4-enumeration; reversible. |
subunits — the numbers behind the chart
[
{
"gene_symbol": "NELFA",
"uniprot": "Q9H3P2",
"length_aa": 528,
"aliases": [
"WHSC2",
"NELF-A",
"P/OKcl.15"
],
"separate_entry": false,
"reason_no_entry": "no distinguishing evidence in this disease; 0 PubMed records with MLL/KMT2A; dependency extreme is floor-compressed and uninterpretable on its own",
"depmap": {
"difference": 0.2403,
"rank": 17650,
"percentile": 99.56,
"mean_kmt2a_r": -0.8725,
"mean_other": -1.1127,
"pan": -1.0384
}
},
{
"gene_symbol": "NELFB",
"uniprot": "Q8WX92",
"length_aa": 628,
"aliases": [
"COBRA1",
"NELF-B",
"KIAA1182"
],
"separate_entry": true,
"node": "gene-nelfb",
"depmap": {
"difference": 0.265,
"rank": 17671,
"percentile": 99.678,
"mean_kmt2a_r": -1.2954,
"mean_other": -1.5604,
"pan": -1.4051
}
},
{
"gene_symbol": "NELFCD",
"uniprot": "Q8IXH7",
"length_aa": 590,
"aliases": [
"TH1L",
"NELF-C",
"NELF-D",
"TH1"
],
"separate_entry": false,
"reason_no_entry": "no distinguishing evidence in this disease; 0 PubMed records with MLL/KMT2A; dependency unremarkable (36th percentile)",
"depmap": {
"difference": -0.0208,
"rank": 6425,
"percentile": 36.242,
"mean_kmt2a_r": -0.7037,
"mean_other": -0.6828,
"pan": -0.6208
}
},
{
"gene_symbol": "NELFE",
"uniprot": "P18615",
"length_aa": 380,
"aliases": [
"RDBP",
"NELF-E",
"RD"
],
"separate_entry": true,
"node": "gene-nelfe",
"depmap": {
"difference": 0.0583,
"rank": 15340,
"percentile": 86.53,
"mean_kmt2a_r": -0.405,
"mean_other": -0.4633,
"pan": -0.4482
}
}
]localization
{
"subcellular": "nucleus",
"subnuclear": "not established"
}enumeration record counts — the numbers behind the chart
[
{
"query": "(MLL OR KMT2A) AND NELF",
"records": 1
},
{
"query": "(MLL OR KMT2A) AND (NELFA OR WHSC2 OR NELF-A)",
"records": 0
},
{
"query": "(MLL OR KMT2A) AND (NELFB OR COBRA1 OR NELF-B)",
"records": 2
},
{
"query": "(MLL OR KMT2A) AND (NELFCD OR TH1L OR NELF-C OR NELF-D)",
"records": 0
},
{
"query": "(MLL OR KMT2A) AND (NELFE OR RDBP OR NELF-E)",
"records": 2
},
{
"query": "(NELF OR NELFA OR NELFB OR NELFCD OR NELFE OR COBRA1) AND (MLL-AF9 OR MLL-AF4 OR MLL-ENL OR MLL-rearranged OR mixed lineage leukemia)",
"records": 2
},
{
"date": "2026-09-19",
"discarded_query_note": "including the bare NELFE alias 'RD' returned 115 records, nearly all spurious; discarded and re-run without it"
}
]- NELFA and NELFB have pan-essentialities of -1.04 and -1.41. Inhibiting NELF would be a general transcriptional intervention with no KMT2A rationale and no window.
- The dependency data point the wrong way for a target: KMT2A-rearranged lines need NELF LESS than other leukaemias, so removing NELF is if anything a smaller insult to them than to comparator cells.
- Relieving pausing pharmacologically - the only direction this biology offers - would increase transcription of the fusion's targets, which is the opposite of what any menin or DOT1L inhibitor is trying to achieve.
- The most likely practical harm from this entry is over-reading six DepMap rows. It is filed as weak confidence for that reason.
- What is the pausing index at KMT2A::AFF1 target genes relative to matched non-target genes? A single PRO-seq dataset would convert this entry from inference to measurement and none exists for the fusion.
- Is the NELFA/NELFB anti-dependency real, or a floor artefact? Permutation testing and line-normalisation on these two genes specifically would settle it cheaply.
- Why do NELFA and NELFB behave one way and NELFCD the other, when NELF is an obligate four-subunit complex?
- If PAF1 displaces NELF at pause release, and the KMT2A fusion binds PAF1, does the fusion increase NELF displacement at its target genes? That is a testable chain and nobody has tested it.
- Does NELF depletion phenocopy or antagonise menin inhibition in a KMT2A-rearranged line?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The cryo-EM structure of a paused transcription elongation complex containing Sus scrofa Pol II and Homo sapiens DSIF and NELF at 3.2 A reveals a tilted DNA-RNA hybrid that impairs NTP substrate binding; NELF binds the polymerase funnel, bridges two mobile polymerase modules and contacts the trigger loop, thereby restraining the Pol II mobility required for pause release; NELF prevents binding of the anti-pausing factor TFIIS; and NELF possesses two flexible tentacles that can contact DSIF and exiting RNA.
Structure of paused transcription complex Pol II-DSIF-NELF · Vos SM, Farnung L, Urlaub H, Cramer P · Nature 560(7720):601-606 2018What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. Reconstituted complex, pig Pol II, human factors, no chromatin and no nucleosome. The abstract does not attribute specific contacts to individual NELF subunits, so this entry does not either.
Evidence that argues against this (1)
- A structure of the paused state establishes how pausing works, not that pausing is the controlling step in any disease. For MLL fusion-dependent transcription specifically, Okuda and Yokoyama name the standard pause-release hypothesis and then report that TBP loading through SL1, not elongation, is the rate-limiting activity for the oncogenic property of MLL-AEP fusions. Okuda H et al. 2016
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In the activated elongation complex Pol II-DSIF-PAF-SPT6 at 3.1 A, PAF displaces NELF from the Pol II funnel for pause release, P-TEFb phosphorylates the Pol II CTD linker, and SPT6 binds the phosphorylated linker and opens the DSIF RNA clamp; formation of the activated complex in vitro requires P-TEFb kinase function plus PAF and SPT6.
Structure of activated transcription complex Pol II-DSIF-PAF-SPT6 · Vos SM, Farnung L, Boehning M, Wigge C, Linden A, Urlaub H, Cramer P · Nature 560(7720):607-612 2018What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. Reconstituted and chromatin-free. PAF here is the general PAF1 complex; the paper makes no claim about the KMT2A-fusion-associated PAF1 interaction that this graph carries separately.
Evidence that argues against this (1)
- If PAF1 is the agent of NELF displacement, then the KMT2A field's model of PAF1 - a chromatin-recruitment contact for the fusion - and this structural role cannot both be the dominant function of PAF1 at a fusion target gene. Nobody has tested the two against each other, and the graph carries both without resolving them. Vos SM et al. 2018
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
As of 2026-09-19, PubMed returns 1 record for '(MLL OR KMT2A) AND NELF' and 2 records for '(NELF OR NELFA OR NELFB OR NELFCD OR NELFE OR COBRA1) AND (MLL-AF9 OR MLL-AF4 OR MLL-ENL OR MLL-rearranged OR mixed lineage leukemia)'. Per-subunit counts against '(MLL OR KMT2A)' are: NELFA and its alias WHSC2, 0; NELFB and its alias COBRA1, 2; NELFCD and its aliases TH1L, NELF-C, NELF-D, 0; NELFE and its aliases RDBP, NELF-E, 2.
PubMed E-utilities esearch record counts for NELF subunits and aliases against KMT2A, MLL and MLL-fusion terms, run for this beat · NCBI PubMed, queried 2026-09-19 2026What this evidence cannot show
Symbol sets searched: NELFA/WHSC2/NELF-A; NELFB/COBRA1/NELF-B; NELFCD/TH1L/NELF-C/NELF-D; NELFE/RDBP/NELF-E. **One query in this beat had to be discarded and re-run**: including the bare alias 'RD' for NELFE returned 115 records, almost all spurious, which is the same wildcard trap the project hit with SEPT*. The corrected NELFE query returns 2. Counted absences are floors, not measurements.
Evidence that argues against this (1)
- Alias breadth is the known failure mode for exactly this kind of claim, and this beat demonstrated it on itself: a two-character alias inflated one subunit's count fifty-fold before being caught. A reader should treat these zeros as 'not retrievable by these terms' rather than as evidence that no laboratory has ever looked at NELF in an MLL-rearranged cell. NCBI PubMed, queried 2026-09-19; finding originally made by this project's non-elongation-partners beat, cycle 40 2026
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The only record retrieved by the NELF-plus-MLL-fusion query that is not a CDK9 review is a proteomics methods paper: a mammalian affinity purification and lentiviral expression system benchmarked against human transcription complexes including the RNA polymerase II-associated factor, negative elongation factor, positive transcription elongation factor b, SWI/SNF and mixed lineage leukemia complexes, and used to identify a Klf4-SWI/SNF interaction in reprogramming. It contains no leukaemia result and no NELF-MLL interaction.
A lentiviral functional proteomics approach identifies chromatin remodeling complexes important for the induction of pluripotency · Mak AB, Ni Z, Hewel JA, Chen GI, Zhong G, Karamboulas K, Blakely K, Smiley S, Marcon E, Roudeva D, Li J, Olsen JB, Wan C, Punna T, Isserlin R, Chetyrkin S, Gingras AC, Emili A, Greenblatt J, Moffat J · Molecular & Cellular Proteomics 9(5):811-23 2010What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. This citation is used only to establish what the retrieved record is and therefore that the counted absence is real rather than a retrieval failure. NELF and MLL complexes appear in the same benchmarking sentence of the abstract, which is precisely the kind of co-occurrence that creates a false positive in an enumeration.
Evidence that argues against this (1)
- Because this paper purified both NELF and MLL complexes with the same tagged-ORF system, its full supplementary interaction tables could in principle contain a NELF-MLL contact that the abstract does not mention. That would not have been retrievable by any query in this beat, and it is the most likely place a real interaction could be hiding. NCBI PubMed, queried 2026-09-19 2026
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
In the project's genome-wide KMT2A-rearranged selectivity scan of 17,728 genes across 1,208 lines, the NELF subunits score: NELFB difference +0.2650, rank 17,671 (99.678th percentile), KMT2A-rearranged mean -1.2954 against other-leukaemia -1.5604, pan-essentiality -1.4051; NELFA +0.2403, rank 17,650 (99.560th), -0.8725 against -1.1127, pan -1.0384; NELFE +0.0583, rank 15,340 (86.530th), -0.4050 against -0.4633, pan -0.4482; NELFCD -0.0208, rank 6,425 (36.242nd), -0.7037 against -0.6828, pan -0.6208. In the same scan the DSIF subunits score in the opposite direction: SUPT4H1 -0.2642 (rank 138) and SUPT5H -0.1215 (rank 767).
Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026What this evidence cannot show
All ten rows read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19 in a single pass; none typed from recall. Eleven KMT2A-rearranged against 43 other leukaemia lines, none infant-derived, AML-heavy. NELFA and NELFB are deeply essential everywhere (-1.04 and -1.41), so their positive differences are differences between two strongly negative numbers and are subject to floor compression that can distort magnitude and in principle sign.
Evidence that argues against this (1)
- The NELF subunits do not agree with each other. NELFA and NELFB sit in the top 0.5% at one extreme while NELFCD sits at the 36th percentile - and NELF is an obligate four-subunit complex, so a genuine complex-level effect should move all four. Half the complex behaving one way is at least as consistent with guide-level or copy-number artefact as with biology, and the asymmetry pattern this entry draws attention to rests on the two subunits that are most floor-compressed. Vos SM et al. 2018
- E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB records for the four NELF subunits, fetched 2026-09-19: NELFA is Q9H3P2, 528 aa, with an HDAg domain at aa 89-248, a NELF-C/D-binding region at aa 125-188 and an RNAPII-binding region at aa 189-248; NELFB is Q8WX92, 628 aa, with no annotated domains outside disordered regions at aa 1-40 and 596-628; NELFCD is Q8IXH7, 590 aa; NELFE is P18615, 380 aa, with an RRM at aa 262-332, a coiled coil at aa 7-36 and 30 approximate 2-residue R-[DSNE] tandem repeats at aa 184-243.
UniProtKB records Q9H3P2 (NELFA), Q8WX92 (NELFB), Q8IXH7 (NELFCD) and P18615 (NELFE) · UniProt Knowledgebase, reviewed (Swiss-Prot) 2026What this evidence cannot show
All four records fetched on 2026-09-19 by gene_exact query with primary gene name and sequence length read back from the same fetched records; every coordinate above copied from those records rather than typed from memory. NELFCD is annotated with no domains beyond a short disordered region, which reflects the state of curation and not an absence of structure - NELF-C and NELF-D are splice products of one gene and the structural literature treats them together.
Evidence that argues against this (1)
- The domain annotations are close to useless as a description of NELF, because the cryo-EM work describes the complex as a funnel-binding body with two flexible tentacles, and the tentacles are in regions UniProt annotates as disordered or does not annotate at all. A domain-map figure built from these coordinates would misrepresent how NELF engages polymerase. Vos SM et al. 2018
- ← related toSUPT4H1 (SPT4) — zero papers linking it to KMT2A, and the 138th most KMT2A-selective gene out of 17,728
DSIF is retained at pause release, NELF is displaced - the mechanistic distinction behind the dependency asymmetry
scaffold subunit; the most extreme anti-dependency in this beat
RNA-binding subunit; RRM aa 262-332
PAF displaces NELF from the Pol II funnel to release the pause
NELF tentacles contact DSIF in the paused complex Pol II-DSIF-NELF
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
NOT an interactor: 1-2 records across all subunit aliases, all reviews; no functional test in KMT2A-rearranged disease
documents the counted absence of NELF in KMT2A-rearranged literature and the granularity exception for NELFA and NELFCD