Canonical versus non-canonical NF-kB in KMT2A-rearranged leukaemia — the two arms have opposite signs, and the two papers usually read as contradicting each other actually agree
also called NF-kB two-arm problem · canonical vs non-canonical MLL · the NF-kB sign problem
NF-kB is two different circuits sharing a name. In MLL leukaemia the fast one keeps the disease going and the slow one holds it back, so saying 'target NF-kB' means nothing until you say which. The two famous papers on this look like they disagree and, read properly, they do not.
The same thing, for a scientist
**The claim usually made.** One 2013 paper says suppressing NF-kB kills MLL leukaemia; one 2018 paper says activating NF-kB (via NIK) suppresses MLL leukaemia. Read as headlines these contradict. **They do not, and it matters that the graph says so.** The 2018 paper measures RelA overexpression in the same model and finds that it **accelerates** AML, and states explicitly that this 'is consistent with RelA overexpression, which accelerates AML with MLL mutations (Kuo et al., 2013)' [E1]. Both papers therefore agree on the canonical arm: **more RELA is worse.** What the 2018 paper adds is that a **second, differently-wired arm exists and runs the other way**: RelB overexpression represses AML, RelB deletion abolishes NIK's protective effect, and RelB binds the Mef2c and Dnmt3a promoters directly [E1][E2]. **So the structure is not a contradiction, it is a two-arm system with opposite signs.** Three consequences follow, and they are the reason this entry exists. **(1)** 'NF-kB inhibitor' is an unusable label here. **(2)** IKKa belongs to both arms, so pan-IKK chemistry blocks the protective arm too, and the anti-leukaemic effect reported in 2013 was obtained *despite* that [E3]. **(3)** The tension does not fully dissolve: overexpressing RelA **failed to rescue** the NIK phenotype, so reduced canonical output is not sufficient to explain the non-canonical arm's benefit [E1]. The arms are opposed in sign and are not the same axis measured twice.
The table the field needs and does not have
| manipulation | arm moved | model | effect on leukaemia | source |
|---|---|---|---|---|
| IKK inhibitors III/IV/VII | both (IKKa serves both) | mouse and human MLL AML | suppressed; fusion evicted from chromatin | 3 |
| Rela knockdown | canonical down | MLL-AF10/AF9 mouse, MV4;11 and ML-2 human | suppressed | 3 |
| Rela overexpression | canonical up | MLL-AF6 mouse | accelerated, shorter latency | 3 |
| Rela-/- fetal liver progenitors | canonical absent | MLL-AF9, MLL-ENL | colonies fail past second plating; non-MLL oncogenes unaffected | 3 |
| RelA overexpression | canonical up | MLL-AF9 mouse | accelerated | 1 |
| NIK stabilisation | non-canonical up, canonical down | MLL-AF9 mouse | suppressed | 1 |
| RelB overexpression | non-canonical up | MLL-AF9 mouse | repressed | 1 |
| RelB knockdown / deletion | non-canonical down | NIK-stabilised MLL-AF9 | reverses the protection | 12 |
| RelA overexpression in NIK-stabilised cells | canonical restored | MLL-AF9 mouse | fails to rescue | 1 |
Every row is mouse or human myeloid. There is no row for any lymphoid or KMT2A::AFF1 system, in either arm.
The residual tension, stated rather than resolved
If NIK's benefit were simply canonical repression, restoring RelA should have restored the leukaemia. It did not 1. Two readings survive:
- The arms are independent. Canonical output is permissive and the non-canonical arm's own target programme (Dnmt3a up, Mef2c down, TIFAB down, HOXA9 down) is separately suppressive. The failure to rescue then says the RelB programme carries the effect.
- RelA overexpression is not the same as restored canonical signalling. Nuclear translocation of overexpressed RelA was blocked when NIK was stabilised 1, so the rescue may have failed because the protein never reached chromatin. This is a technical explanation and the paper itself raises it.
The graph should hold both. Do not write 'canonical NF-kB is required and non-canonical opposes it' as a resolved mechanism — the first half is well supported and the second half's route is not settled.
What this forbids downstream
Any entry proposing an NF-kB-directed therapy in this disease must name the arm and the node. Specifically:
- Proteasome inhibition is not an arm-selective intervention; it blocks p100 processing and IkBa degradation together, and in MLL pro-B leukaemia its documented mechanism is neither (
mechanism-proteasome-inhibition-splits-by-lineage-in-mll-leukaemia). - Pan-IKK inhibition blocks both arms via IKKa.
- NEMO disruption is canonical-selective and untested here (
gene-ikbkg). - NIK stabilisation is non-canonical-selective, has no chemical matter, and also damages normal HSPCs (
gene-map3k14).
And the menin question inherits the same problem. Menin binds p50, p52 and p65 4 — one subunit from each arm plus the shared precursor family — so a menin inhibitor does not release a single well-signed species, and hyp-menin-inhibition-liberates-rela-and-reinforces-the-fusion-programme should be read with that ambiguity in front of it.
| mechanism class | pathway-architecture |
|---|---|
| shared component | gene-chuk (IKKa) belongs to both arms |
| tension status | partially resolved: both papers agree the canonical arm is pro-leukaemic; the non-canonical arm's suppressive route is NOT explained by canonical repression, because RelA overexpression fails to rescue |
| lineage coverage | myeloid only in every row |
arms
[
{
"name": "canonical",
"kinases": [
"gene-chuk",
"gene-ikbkb",
"gene-ikbkg"
],
"inhibitor": "gene-nfkbia",
"subunits": [
"gene-rela",
"gene-nfkb1",
"gene-rel"
],
"direction_in_mll": "permissive / pro-leukaemic"
},
{
"name": "non-canonical",
"kinases": [
"gene-map3k14",
"gene-chuk"
],
"precursor": "gene-nfkb2",
"subunits": [
"gene-relb"
],
"direction_in_mll": "suppressive / anti-leukaemic"
}
]- Any pan-NF-kB or proteasome-based intervention moves both arms at once, in opposite directions, and the net sign in a patient is not predictable from any published experiment.
- An intervention that succeeds by activating the non-canonical arm is activating the arm that is oncogenic in mature B-cell malignancies, in a B-lineage disease.
- An intervention that succeeds by blocking the canonical arm removes the principal stress-survival pathway of normal haematopoietic cells and of the whole immune system.
- Because the arms cross-repress, partial or intermittent inhibition of one may be equivalent to activating the other, which makes dose and schedule mechanistically load-bearing rather than merely pharmacokinetic.
- Does the non-canonical arm suppress this leukaemia through its own target programme or through repressing the canonical arm? The one experiment designed to answer it - RelA rescue - was confounded by blocked nuclear translocation.
- Do the signs invert in B lineage? NIK and non-canonical signalling are oncogenic in mature B-cell malignancies, and the index disease is a B-precursor leukaemia. Nothing in this table is lymphoid.
- Which arm does a menin inhibitor release? Menin binds p50, p52 and p65, so it restrains both arms and the net released species is unknown.
- Is there a node that is canonical-selective and druggable? NEMO is the only structurally clean candidate and it has never been tested in this disease.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In MLL-AF9 AML, overexpression of RelB significantly repressed and overexpression of RelA significantly accelerated leukaemia, the latter described by the authors as consistent with RelA overexpression accelerating AML with MLL mutations in Kuo et al. 2013. NIK stabilisation activated non-canonical signalling (NfkB2 and Relb up, nuclear RelB and p52 up, p100 processing on) and repressed canonical signalling (RelA protein and nuclear RelA down, Nfkbia and Tnfaip3 up), and suppressed AML. Overexpression of RelA failed to restore the reduced colony formation or the delay in leukaemogenesis caused by NIK stabilisation, and when NIK was stabilised the nuclear translocation of overexpressed RelA was blocked.
Stabilization of NF-κB-Inducing Kinase Suppresses MLL-AF9-Induced Acute Myeloid Leukemia. · Xiu Y, Dong Q, Li Q, Li F, Borcherding N, Zhang W, Boyce B, Xue HH, Zhao C · Cell Reports 2018What this evidence cannot show
Mouse MLL-AF9 only, one laboratory, genetic stabilisation of a non-degradable NIK mutant rather than pharmacology, RelA and RelB manipulated by overexpression. The failed RelA rescue is confounded by blocked nuclear translocation of the overexpressed protein, which the paper reports and which makes the rescue uninterpretable as a clean test of sufficiency. Full text read this session (PMC5810947).
Evidence that argues against this (1)
- Overexpression experiments establish sufficiency in a forced system, not what the endogenous proteins do at physiological levels, and RelA and RelB overexpression also change dimer availability for every other NF-kB partner. The opposite directions are the useful part; the magnitudes are not interpretable from overexpression alone. Zhao J et al. 2021
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Deletion of RelB prevents the inhibitory effect of NIK stabilisation in MLL-AF9 AML; RelB suppresses TIFAB, and forced TIFAB or forced HOXA9 expression reverses NIK-induced impairment of AML by downregulating RelB, with forced HOXA9 additionally upregulating RelA.
TIFAB accelerates MLL-AF9-Induced acute myeloid leukemia through upregulation of HOXA9. · Zhao J, Xiu Y, ... Zhao C · iScience 2021What this evidence cannot show
Abstract read; full text not retrieved. Same group and model as E1, so extension rather than replication. It also reports TIFAB as a direct RelB target while the earlier paper's ChIP found no RelB binding at the Tifab promoter; that discrepancy is unresolved here.
Evidence that argues against this (1)
- That forced TIFAB or forced HOXA9 each reverse the NIK phenotype argues they are downstream of it, but two different forced rescues reversing one phenotype is also what happens when a phenotype is a general proliferation deficit. The TIFAB arm additionally rests on a direct-target claim that a ChIP in the companion study does not support. Xiu Y et al. 2018
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Suppression of NF-kB produced robust anti-leukaemia effects phenocopying loss of functional MLL oncoprotein: Rela knockdown reduced growth and clonogenicity in MLL-transformed and MLL leukaemia cells but not normal progenitors, forced Rela expression increased growth and clonogenicity and shortened leukaemia latency in vivo, and Rela-/- fetal liver myeloid progenitors could not sustain colonies induced by MLL-AF9 or MLL-ENL beyond the second plating while remaining fully capable of continuous replating with E2A-PBX1, NUP98-HOXA9 or E2A-HLF. The pharmacology used non-subunit-selective IKK inhibitors III, IV and VII, which also inhibit IKKa, the non-canonical arm's kinase.
Epigenetic roles of MLL oncoproteins are dependent on NF-κB. · Kuo HP, Wang Z, Lee DF, Iwasaki M, Duque-Afonso J, Wong SHK, Lin CH, Figueroa ME, Su J, Lemischka IR, Cleary ML · Cancer Cell 2013What this evidence cannot show
The Rela-/- experiment is the strongest genetic evidence in either paper and it is a fetal-liver myeloid progenitor assay, so it tests transformation initiation rather than maintenance of established leukaemia. Nothing in the paper distinguishes the two arms, because nothing in it measures p100 processing, RelB or p52. Full text read this session (PMC3816582).
Evidence that argues against this (1)
- The canonical arm's direction is not monotonic. IMPDH inhibition induces OVERACTIVATION of Toll-like receptor-TRAF6-NF-kB signalling with VCAM1 upregulation and SUPPRESSES MLL-fusion leukaemia, and a TLR1/2 agonist is an effective combination partner. Driving the same arm harder is therefore also anti-leukaemic, which no two-arm sign table accommodates. Liu X et al. 2023
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Menin interacts specifically and directly with p50, p52 and p65 - that is, with subunits of both the canonical and the non-canonical arm - via their N-termini, with menin residues 305-381 essential, and represses p65-mediated transactivation dose-dependently.
The tumor suppressor protein menin interacts with NF-kappaB proteins and inhibits NF-kappaB-mediated transactivation. · Heppner C, Bilimoria KY, ... Burns AL · Oncogene 2001What this evidence cannot show
Abstract read this session. 2001 methods, in vitro and transfected-reporter systems, non-haematopoietic backgrounds, and the p52 interaction is curated in IntAct from a GST pull-down alone. The subunit that matters for arm assignment here - p52's partner RelB - was not tested at all.
Evidence that argues against this (1)
- A curated interaction inherits every limitation of the single experiment behind it. Anti-bait co-immunoprecipitation and affinity-capture mass spectrometry cannot distinguish a direct contact from a shared-complex association, tagged-bait overexpression can create contacts that do not occur at endogenous stoichiometry, and a partner curated from one publication has by definition never been independently replicated. The enumeration is accurate about the database; the database is not a statement about the blast. Matthews GM et al. 2015
the non-canonical effector, with the opposite sign to RELA
- related to →RELA (p65) — menin binds it and represses it, and it is required to keep MLL oncoproteins on chromatin
both papers agree RELA is pro-leukaemic in MLL-rearranged models
menin binds p50, p52 and p65, so it restrains both arms and the released species is unsigned
IKKa belongs to both arms, so pan-IKK chemistry cannot separate them
third canonical subunit; transcript unchanged by NIK stabilisation and otherwise unstudied here
the shared component whose dual membership creates the sign problem
p105/p50, the canonical DNA-binding partner; the pathway the MLL screen names is p105-dependent
IMPDH inhibition OVERACTIVATES TLR-TRAF6-NF-kB and suppresses MLL-fusion AML, against a monotonic reading
the upstream node of the non-canonical arm; NIK stabilisation suppresses murine MLL-AF9 AML, so TRAF2 loss is predicted protective there
A LINK IN THE CHAIN THIS ENTRY DEPENDS ON: if RelB to TIFAB is indirect, the non-canonical arm's suppression of MLL-AF9 remains real but stops naming a target.