A genuine small-molecule disruptor of an oncogenic oligomer DOES exist - 7.44 against the RUNX1::RUNX1T1 NHR2 tetramer - which corrects this graph's record; and its eighteen-year potency curve says the class is a chemistry problem, which is what this graph concluded for the wrong reason
also called 7.44 · M23 · NHR2 tetramerisation inhibitor · RUNX1/ETO oligomer disruptor · the oligomer-interface chemistry precedent
This project has been asking whether anyone has ever made a small molecule that pulls apart the glue holding a fusion protein's copies together. The answer is yes - a compound called 7.44 does it to the RUNX1-ETO fusion, and newer versions of it are about ten times better. But even the best of them only works at concentrations tens of thousands of times higher than a real drug, which tells you the idea is sound and the chemistry is not there yet.
The same thing, for a scientist
`hyp-oligomerisation-interfaces-as-mutation-resistant-targets` records, as the answer to its own druggability question, that **exactly one oligomerisation-interface small molecule is approved anywhere in medicine - tafamidis, which STABILISES the transthyretin tetramer - and none disrupts an oncogenic oligomer.** The approval half is right. **The disruption half is wrong, and the counter-example is the very family that hypothesis is named after.** Compound **7.44** interferes with NHR2-domain tetramerisation of RUNX1/ETO, restores gene expression downregulated by the fusion, inhibits proliferation of RUNX1/ETO-dependent cells and reduces fusion-driven tumour growth in a mouse model; biophysically it binds NHR2 with **Kd 3.75 +/- 1.22 uM**, **shifts the population from tetramer toward dimer** rather than to monomer, engages NHR2 with both heteroaromatic moieties by NMR with molecular-dynamics support, and has favourable physicochemical, pharmacokinetic and toxicological properties [E1]. A 2025 follow-up designed **non-charged analogues** M23, M27 and M10 to fix 7.44's negative charge at physiological pH and predicted low-to-medium permeability; the analogues bind NHR2 with **Kd 39-114 uM by microscale thermophoresis**, give **cell-viability IC50 of 33-77 uM in RUNX1/ETO-positive Kasumi cells**, are **5 to 10-fold more potent than 7.44**, and M23 induces apoptosis and promotes differentiation in fusion-positive cells [E2]. **Read those two numbers together and the strategic conclusion survives the factual correction.** The best compound in the series, after more than a decade of optimisation by the group that invented it, has a cellular IC50 in the tens of micromolar. **So the class is chemically real and pharmacologically unarrived**, and this graph's verdict - the problem is chemistry, not biology - is correct while the evidence it cited for that verdict was not.
The correction, stated precisely so it can be applied
Entry: hyp-oligomerisation-interfaces-as-mutation-resistant-targets, in summary_expert and in the section 'Counter-pass (cycle 33)'.
What it says: 'exactly one oligomerisation-interface small molecule is approved anywhere in medicine -- tafamidis, which STABILISES the transthyretin tetramer -- and none disrupts an oncogenic oligomer.' Its open question asks: 'Is there any oncogenic oligomerisation interface anywhere with a genuine steric-disruptor small molecule, as opposed to a peptide, a metalloid or a stabiliser? If the answer after a full search is no, the class's problem is chemistry and the hypothesis should be reframed as a chemistry programme, not a biology one.'
What it should say: the APPROVAL claim is correct and should be kept - no oligomerisation-interface disruptor is approved. The clause 'none disrupts an oncogenic oligomer' is wrong as a statement about existence. 7.44 is a small molecule, not a peptide, not a metalloid and not a stabiliser, and it disrupts the NHR2 tetramer of RUNX1/ETO 1, with a designed non-charged analogue series following it 2.
Citation that settles it: PMID 35986043, doi 10.1038/s41598-022-17913-6, abstract read in full on 2026-09-23; and PMID 40399488, doi 10.1038/s41598-025-02422-z, same date. A bounded search NHR2[tiab] AND inhibitor[tiab] returns 5 records and NHR2[tiab] returns 45, so the literature here is small enough to have been enumerated.
And the open question should be marked ANSWERED, in the direction its own conditional specifies. The conditional was 'if no such molecule exists, the problem is chemistry'. One exists and its cellular IC50 is 33-77 uM. The conclusion holds for a better reason than the one offered: the problem is not that nobody has built a disruptor, it is that a disruptor of a buried four-helix oligomerisation interface has so far cost four to five orders of magnitude of potency relative to a pocket binder like revumenib.
What the NHR2 programme is a template for, and what it costs
This graph asks, in hyp-oligomerisation-interfaces-as-mutation-resistant-targets and again in concept-why-bcr-abl1-became-curable, whether KMT2A::AFF1 has an obligate self-association interface. hyp-aff1-chd-dimerisation-in-kmt2a-fusion and exp-aff1-chd-dimerisation-separation-of-function state the candidate and the separation-of-function experiment. The NHR2 story is what the whole road looks like when the answer is yes:
- A structurally defined self-association module with a solved structure (NHR2 tetramer, PDB 1WQ6).
- Genetic proof that oligomerisation is required for oncogenic activity.
- A dominant-negative peptide - and note that a CBFA2T3/ETO2 NHR2-mimicking peptide inhibits the RUNX1-CBFA2T3 interaction and reduces B-cell-precursor ALL blast proliferation, which is the same module doing the same job in the correct lineage for this project 3.
- Computational identification of a small-molecule disruptor, 7.44 1.
- Biophysical validation: Kd, NMR contacts, oligomer-state shift, ADMET 1.
- Ligand-based virtual screening for improved analogues, a decade later, reaching 33-77 uM cellular IC50 2.
Step 6 is the number a project should look at before starting at step 1. Every earlier step in this list is a success. The programme is textbook. And what it has produced is a tool compound.
The one genuinely encouraging detail is that 7.44 shifts the tetramer toward dimer, not monomer 1 - partial disassembly was enough to restore fusion-repressed gene expression and reduce tumour growth in vivo. If the same is true of an AFF1 self-association interface, a partial disruptor would be therapeutically meaningful, which lowers the potency bar somewhat. It does not lower it by four orders of magnitude.
Where the analogy breaks for KMT2A::AFF1
It breaks at step 2, and that is the step this project has not passed. NHR2's requirement for RUNX1/ETO oncogenic activity is established genetics. For KMT2A::AFF1, the corresponding claim - that the fusion requires AFF1-mediated self-association or AFF1-AFF4 heteromerisation - is stated in this graph as a hypothesis with an unbuilt separation-of-function experiment. hyp-oligomerisation-interfaces-as-mutation-resistant-targets asks the question in exactly those words and it is still open.
It also breaks on architecture. NHR2 is a compact, folded, four-helix bundle with a crystal structure and a definable groove. The AFF1 C-terminal homology domain sits in a protein that is substantially intrinsically disordered and whose interactions with AFF4, ELL2 and P-TEFb are multivalent. A disruptor's job is much better defined against a folded bundle than against a disordered network, and the Ewing case in analogy-tk216-is-the-interface-drug-that-reached-the-clinic-and-turned-out-to-be-a-tubulin-agent is what happens when it is attempted against the disordered kind.
And it breaks on the resistance argument that motivated the hypothesis. The claim was that an oligomerisation-interface drug cannot be escaped, because the drug surface and the function surface are the same. At 33-77 uM cellular IC50 2 that question is untestable - resistance is only meaningful for a compound that exerts selection pressure at achievable exposures.
| domain | engineering |
|---|---|
| where the analogy breaks | RUNX1::RUNX1T1 has established genetics showing oligomerisation is required; KMT2A::AFF1 does not, and that is the step the analogy cannot supply. NHR2 is a compact folded bundle with a structure and a groove; the AFF1 self-association candidate sits in a largely disordered protein with multivalent partners, which is the architecture that defeated the Ewing programme. And the resistance-resistance argument that motivated the whole target class cannot be tested at 33-77 uM, because a compound that exerts no selection pressure generates no escape mutants. |
| what it predicts here | If an AFF1 self-association requirement is demonstrated genetically, the NHR2 programme predicts the road: structure, dominant-negative peptide, virtual screen, micromolar hit, a decade of analogues, and a compound still in the tens of micromolar. A project deciding whether to start should price that in at the outset rather than after the structure. |
| relationship to existing entries | fusion-runx1-runx1t1 describes NHR2 as a tetramer whose integrity is critical for AML1-ETO activity and cites PDB 1WQ6, but contains no chemical matter against it. hyp-oligomerisation-interfaces-as-mutation-resistant-targets asks the druggability question and answers it incorrectly. Neither mentions 7.44; a corpus search for '7.44' on 2026-09-23 found it in neither entry, and PMIDs 35986043 and 40399488 are cited in zero node files. |
mapping
[
{
"bio": "RUNX1::RUNX1T1 requires NHR2-mediated tetramerisation for oncogenic activity; the module is a folded four-helix bundle with a solved structure",
"analog": "KMT2A::AFF1 is HYPOTHESISED to require AFF1 self-association or AFF1-AFF4 heteromerisation; the separation-of-function experiment is stated and unbuilt"
},
{
"bio": "7.44 disrupts the NHR2 tetramer, Kd 3.75 uM to the isolated domain, shifting tetramer toward DIMER rather than monomer",
"analog": "partial disassembly sufficed for gene-expression restoration and in vivo tumour reduction - if the same holds for an AFF1 interface, the potency bar is lower than full disruption would require"
},
{
"bio": "non-charged analogues M23/M27/M10, a decade later: Kd 39-114 uM, cell IC50 33-77 uM, 5-10x better than 7.44",
"analog": "the realistic cost of drugging a buried oligomerisation interface - four to five orders of magnitude worse than a pocket binder like revumenib"
},
{
"bio": "an NHR2-mimicking peptide reduces BCP-ALL blast proliferation",
"analog": "the module is biologically targetable in B-lineage ALL; the deficit is small-molecule chemistry, not interface biology"
},
{
"bio": "this graph recorded that NO small molecule disrupts an oncogenic oligomer",
"analog": "CORRECTION - one does, in the family the hypothesis is named after; the strategic conclusion survives, the factual premise does not"
}
]generated hypotheses
[]
correction reported
{
"entry": "hyp-oligomerisation-interfaces-as-mutation-resistant-targets",
"field": "summary_expert, and the section 'Counter-pass (cycle 33): the resistance argument has a clinical counterexample, and the druggability question now has an answer'",
"wrong_text": "exactly one oligomerisation-interface small molecule is approved anywhere in medicine -- tafamidis, which STABILISES the transthyretin tetramer -- and none disrupts an oncogenic oligomer",
"should_say": "no oligomerisation-interface disruptor is APPROVED (tafamidis, a stabiliser, is the only approved oligomer-interface small molecule); but a small-molecule disruptor of an oncogenic oligomer does exist - 7.44 against the RUNX1/ETO NHR2 tetramer - and its potency, not its absence, is what makes the class a chemistry problem",
"citation_pmids": [
"35986043",
"40399488"
],
"read_level": "abstract in full from the PubMed record, both papers, 2026-09-23; the founding 7.44 paper itself was not read",
"open_question_to_mark_answered": "Is there any oncogenic oligomerisation interface anywhere with a genuine steric-disruptor small molecule, as opposed to a peptide, a metalloid or a stabiliser? ANSWERED YES - compound 7.44 and its analogues M23/M27/M10 against NHR2. The conditional's conclusion nevertheless holds: the class is a chemistry programme, because the best analogue is at 33-77 uM cellular IC50."
}- The honest headline for this project is discouraging: the oligomerisation-interface class has had a small molecule for over a decade and the best optimised analogue is at 33-77 uM in cells. Any proposal to pursue an AFF1 self-association interface should state this number in its first paragraph.
- The resistance-resistance argument that makes this target class attractive is currently untestable, because no compound in it reaches exposures that would select for escape. An untested advantage is not an advantage.
- Essential-function toxicity: NHR2-directed disruption also hits wild-type CBFA2T3/ETO2 complexes with RUNX1 [E3], which are required in normal haematopoiesis. A disruptor of a self-association module shared between a fusion and its wild-type parent has no fusion selectivity built in, and this is the same problem the KMT2A field has with anything that hits wild-type KMT2A.
- Delivery: 7.44's charge at physiological pH was a permeability problem that took a dedicated medicinal-chemistry campaign to address [E2]. Interfaces in disordered proteins will tend to select for exactly this kind of polar, charged chemical matter.
- Does KMT2A::AFF1 actually require AFF1-mediated self-association? This entry supplies the whole downstream roadmap and cannot supply the genetics; hyp-aff1-chd-dimerisation-in-kmt2a-fusion and exp-aff1-chd-dimerisation-separation-of-function state the experiment and it has not been run.
- Is partial oligomer disassembly therapeutically sufficient in general? 7.44 shifts tetramer to dimer rather than monomer and that was enough for gene-expression restoration and in vivo effect [E1]. If partial disruption suffices generally, every oligomer-interface programme's potency requirement falls, and nobody has tested the principle in a second family.
- Why has no NHR2-directed compound entered a clinical trial in the years since 7.44? The hypothesis entry already records that the closest replica of the menin precedent, CBFbeta-SMMHC/RUNX1, has no registered trial eleven years after AI-10-49. Two independent interface programmes in core-binding-factor AML have stalled at the same point, and whether the obstacle is potency, pharmacokinetics or commercial is not documented anywhere this project can see.
- Would a degrader beat a disruptor at an oligomerisation interface? A compound that binds a buried four-helix bundle with 3.75 uM affinity is a poor occupancy drug and might be an adequate degrader ligand, which is the same reframing hyp-obligate-partner-interface-generalises reaches for the WDR5 WIN site.
- Does the NHR2-mimicking peptide result in BCP-ALL [E3] extend to KMT2A-rearranged B-ALL, where RUNX1 and CBFA2T3 are also expressed? That is a single experiment in an available cell panel and it would tell this project whether an NHR2-directed agent has any relevance to its own disease.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Compound 7.44 is a first-in-class small-molecule inhibitor of NHR2-domain tetramer formation in the RUNX1/ETO fusion protein: previously shown to interfere specifically with NHR2, restore gene expression downregulated by RUNX1/ETO, inhibit proliferation of RUNX1/ETO-dependent SKNO-1 cells and reduce RUNX1/ETO-related tumour growth in a mouse model; biophysical characterisation shows it interferes with NHR2 tetramer stability and increases the dimer population, binds NHR2 with an affinity of 3.75 +/- 1.22 uM, engages NHR2 with both heteroaromatic moieties by NMR with molecular-dynamics support including conformational changes at the N-termini of the NHR2 tetramer, and has favourable physicochemical, pharmacokinetic and toxicological properties.
Biophysical and pharmacokinetic characterization of a small-molecule inhibitor of RUNX1/ETO tetramerization with anti-leukemic effects. · Gopalswamy M, Kroeger T, Bickel D, Frieg B, Akter S, Schott-Verdugo S, Viegas A, Pauly T, Mayer M, Przibilla J, Reiners J, Nagel-Steger L, Smits SHJ, Groth G, Etzkorn M, Gohlke H · Sci Rep 2022What this evidence cannot show
Abstract read in full from the PubMed record on 2026-09-23. The cellular and in vivo results described in the abstract are attributed to the group's own earlier work and are restated here at second hand within the same authors' abstract; this entry has not read the founding 7.44 paper and says so. The disease is t(8;21) AML, not a KMT2A rearrangement. A 3.75 uM binding affinity to an isolated domain is a starting point, not a drug-like number, and the compound is negatively charged at physiological pH with predicted low-to-medium membrane permeability, as the follow-up paper states [E2].
Evidence that argues against this (1)
- The improved non-charged analogues reported three years later bind NHR2 with Kd of 39-114 uM and give cell-viability IC50 of 33-77 uM while being 5-10 fold MORE potent than 7.44 in cells - which places 7.44's own cellular potency in the hundreds of micromolar. The existence claim is unaffected; any claim that this is a viable therapeutic lead is not. Gopalswamy M et al. 2025
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Because 7.44 is negatively charged at physiological pH and was predicted to have low-to-medium membrane permeability, ligand-based virtual screening plus biophysical and cellular validation identified non-charged analogues M23, M27 and M10 that interact with the NHR2 domain with Kd values of 39-114 uM by microscale thermophoresis and IC50 values of 33-77 uM for cell viability in RUNX1/ETO-positive Kasumi cells, i.e. approximately 5 to 10-fold more potent than 7.44; M23 is about 10-fold more potent than 7.44 in inhibiting proliferation of RUNX1/ETO-positive cells and induces apoptosis and promotes differentiation in RUNX1/ETO-positive AML cells, and the non-protonated species are predicted to be highly membrane-permeable.
Identification of non-charged 7.44 analogs interacting with the NHR2 domain of RUNX1-ETO with improved antiproliferative effect in RUNX-ETO positive cells. · Gopalswamy M, Bickel D, Dienstbier N, Tu JW, Vogt M, Schott-Verdugo S, Bhatia S, Etzkorn M, Gohlke H · Sci Rep 2025What this evidence cannot show
Abstract read in full from the PubMed record on 2026-09-23. These are tens-of-micromolar compounds in a single fusion-positive cell line (Kasumi) with fusion-negative comparators; the permeability advantage is predicted, not measured in vivo. The paper's own framing is that these 'might serve as lead structures', which is the honest description of the stage the class has reached after more than a decade.
Evidence that argues against this (1)
- A dominant-negative PEPTIDE approach against the same module reached a biological result in the lineage this project cares about - an NHR2-mimicking peptide reduced B-cell-precursor ALL blast proliferation - which suggests the module is a real target and that the limitation is specifically small-molecule chemistry rather than the biology of the interface. Jakobczyk H et al. 2021
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In B-cell precursor acute lymphoblastic leukaemia, RUNX1 and CBFA2T3/ETO2/MTG16 form part of the same complex in which CBFA2T3 strongly potentiates RUNX1 transcription-factor activity; a CBFA2T3-mimicking peptide based on the NHR2 subdomain inhibits the RUNX1-CBFA2T3 interaction, abrogates the activity of the complex and reduces BCP-ALL lymphoblast proliferation, and the authors propose therapeutic development based on the NHR2 subdomain.
Reduction of RUNX1 transcription factor activity by a CBFA2T3-mimicking peptide: application to B cell precursor acute lymphoblastic leukemia. · Jakobczyk H, Debaize L, Soubise B, Avner S, Rouger-Gaudichon J, Commet S, Jiang Y, Serandour AA, Rio AG, Carroll JS, Wichmann C, Lie-A-Ling M, Lacaud G, Corcos L, Salbert G, Galibert MD, Gandemer V, Troadec MB · J Hematol Oncol 2021What this evidence cannot show
Abstract read in full from the PubMed record on 2026-09-23. This is wild-type RUNX1 and wild-type CBFA2T3 in BCP-ALL, not the RUNX1::RUNX1T1 fusion and not a KMT2A rearrangement; the patient context the paper emphasises is ETV6::RUNX1. It is cited here for one narrow purpose - that NHR2-directed disruption produces a biological effect in B-lineage ALL cells - and not as evidence about the fusion.
Evidence that argues against this (1)
- A peptide is exactly the modality the hypothesis this entry corrects explicitly excluded when it asked for a 'genuine steric-disruptor small molecule, as opposed to a peptide'. This evidence item therefore does not itself answer that question; only E1 and E2 do. It is included because it is the only NHR2 result in B-lineage ALL and because it separates 'the interface is targetable' from 'a small molecule can target it'. Gopalswamy M et al. 2022
- related to →RUNX1::RUNX1T1 (AML1-ETO), t(8;21)(q22;q22)
the family whose NHR2 tetramer has the only small-molecule oncogenic-oligomer disruptor
corrects its claim that no small molecule disrupts an oncogenic oligomer, while confirming its strategic conclusion for a different reason
the KMT2A-side candidate the NHR2 roadmap would apply to, if the genetics were established
- related to →RUNX1T1 (ETO / MTG8) domain architecture
the NHR2 domain that the compound binds
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
the index fusion whose hypothesised self-association interface this roadmap prices
- supports →Separation-of-function test: does KMT2A::AFF1 need its C-terminal homology domain to dimerise?
the NHR2 programme is the template this experiment's positive result would commit the lab to