Two entries record zero chemical matter against the CXXC-DNA interface. There is one: dimethyl fumarate, IC50 about 1 uM, and it is already an approved drug
also called DMF and the CXXC domain · Tecfidera against MLL1 CXXC · the chemical matter that was not zero
This project's drug dossiers say nobody has ever made a molecule that blocks the fusion's DNA grip. One has been published: dimethyl fumarate, a drug already given to people with multiple sclerosis, blocks the grip at about one micromolar in a single-molecule experiment. It is a sticky, unselective molecule rather than a designed drug, but it is not nothing.
The same thing, for a scientist
target-phase2-cxxc-dna-dossier records 'no CXXC-directed series known to this review' and target-kmt2a-cxxc-dna-interface records 'no reported small molecule against any ZF-CxxC domain' and 'no bioactivity series in ChEMBL'. Liang 2021 reports that dimethyl fumarate inhibits the MLL1 CXXC-DNA complex with a relative IC50 of 0.91 +/- 0.09 uM, measured as the dose-dependent loss of single-molecule hairpin unfolding time in the presence of 1 uM CXXC domain, fitted by a four-parameter logistic (RMSE 0.012, R-squared 0.997). The prodrug's active metabolite monomethyl fumarate is inactive at 10 uM, so the activity belongs to the electrophilic parent. DMF is an approved oral therapy for relapsing multiple sclerosis and psoriasis. Three things follow and they point in different directions: the dossiers' zero is wrong and must be qualified; DMF covalently modifies more than 2,400 cysteines, so sub-micromolar inhibition of a four-cysteine zinc-coordinating domain is the expected generic result rather than evidence of tractability; and the authors note that chemical proteomics has never identified CXXC-family proteins as DMF targets, which makes the cellular relevance unknown and the clinical question - whether DMF exposure does anything in KMT2A-rearranged disease - entirely unasked.
What was measured
Liang et al. formed MLL1 CXXC-DNA complexes on a DNA hairpin with well-isolated CpG sites in an AT-only background, then measured the time required to mechanically unfold the hairpin at a 12.5 pN probing force - protein binding lengthens that time, so the unfolding time is a label-free readout of complex integrity 1.
| compound | dimethyl fumarate (DMF) |
| dose range | 1, 8, 64, 512 nM and 1, 4, 32, 128, 512 uM |
| protein | 1 uM MLL1 CXXC domain, chemically synthesised, untagged |
| pre-incubation | at least 15 min before the force protocol |
| fit | four-parameter logistic, RMSE 0.012, R-squared 0.997 |
| relative IC50 | 0.91 +/- 0.09 uM (mean +/- SE), N > 20 traces from > 3 molecules per concentration |
| negative control | monomethyl fumarate, the active metabolite, inactive at 10 uM |
The MMF control is the informative half. DMF is a prodrug rapidly metabolised to MMF after oral dosing, and MMF is the species that reaches tissue as a GPR109A ligand. MMF does nothing to this complex, so whatever DMF does here it does as the intact electrophile, which is largely consumed on first pass. That is a sharp limit on any clinical reading.
What it means for the two dossiers, honestly and in both directions
The zero is wrong and must be qualified. target-phase2-cxxc-dna-dossier says 'To this review's knowledge no published series engages the CXXC domain itself' and target-kmt2a-cxxc-dna-interface says 'There is no reported small molecule against any ZF-CxxC domain'. Both should now read: one compound is reported, it is an approved drug, it is a promiscuous covalent electrophile, and there is no series. 'No series' survives; 'no molecule' does not.
And the finding argues AGAINST tractability at least as much as for it. DMF covalently modifies over 2,400 cysteines 1. The CXXC domain coordinates two zinc ions through eight cysteines. A cysteine-reactive electrophile inhibiting it at 1 uM is the generic zinc-finger liability this graph already records in target-kmt2a-cxxc-dna-interface's bite-back, now demonstrated rather than predicted. It does not show that the interface has a ligandable pocket; it shows the domain has reactive cysteines, which every member of the twelve-protein family also has. If anything, the selectivity gate the phase-2 dossier sets - 30-fold against CXXC1, KDM2B and TET1 - is harder to pass for a covalent cysteine chemotype than for anything else.
The one genuinely new question it opens is clinical, not chemical. DMF has been taken by hundreds of thousands of people. The authors observe that CXXC-family proteins have never been identified as DMF targets by chemical proteomics, and that DMF's effect on MLL1-dependent processes is unexplored. Nobody has asked whether DMF exposure does anything in KMT2A-rearranged disease, in either direction - and given the enasidenib liability this graph already records (a drug that plausibly helps the fusion by expanding its unmethylated-CpG substrate), the direction is not obvious. It is a registry question before it is an experiment.
| script | none - a reading of one primary paper |
|---|---|
| status | run |
| result summary | One published inhibitor of the KMT2A/MLL1 CXXC-DNA interface exists: dimethyl fumarate, relative IC50 0.91 +/- 0.09 uM in a single-molecule assay, with the metabolite MMF inactive at 10 uM. Two entries in this graph record zero chemical matter for this interface and must be qualified. The finding is not evidence of tractability - DMF modifies more than 2,400 cysteines and the CXXC fold coordinates zinc through eight of them - and it cannot pass the phase-2 selectivity gate, against which it was never counter-screened. |
| unasked clinical question | Has anyone looked at KMT2A-rearranged leukaemia incidence, or outcome, in the very large population exposed to oral DMF for multiple sclerosis and psoriasis? The authors say the clinical effect is unexplored. Given that DMF is largely converted to the inactive metabolite MMF on first pass, the prior on any effect is low - but it is a registry question, not an experiment, and nobody has asked it. |
| shown for fusion | NO - nobody has looked in any fusion. The assay uses an isolated synthesised CXXC domain on naked DNA; no fusion protein, no cell, no leukaemia model. |
inputs
[ "PMID 34280370 full text (PMC8392069)" ]
compound
{
"name": "dimethyl fumarate",
"aliases": [
"DMF",
"Tecfidera",
"Fumaderm component"
],
"approved_for": "relapsing multiple sclerosis; psoriasis",
"mechanism_here": "inferred covalent cysteine modification; not demonstrated on this domain",
"relative_IC50_uM": 0.91,
"IC50_SE_uM": 0.09,
"assay": "single-molecule hairpin unfolding time at 12.5 pN, 4PL fit, RMSE 0.012, R-squared 0.997",
"protein_conc_uM": 1,
"dose_range": "1 nM to 512 uM",
"metabolite_control": "monomethyl fumarate inactive at 10 uM",
"counter_screened_against_other_cxxc_domains": false,
"cellular_data": false,
"pmid": "34280370"
}entries to qualify
[
{
"id": "target-phase2-cxxc-dna-dossier",
"field": "data.chemical_matter.cxxc_directed_series and the 577-rows section",
"current": "none known to this review",
"should_read": "no SERIES; one compound reported - dimethyl fumarate, relative IC50 about 1 uM in a single-molecule assay (PMID 34280370), a promiscuous covalent electrophile that was not counter-screened against any other CXXC domain and would not pass gate G1"
},
{
"id": "target-kmt2a-cxxc-dna-interface",
"field": "the chemistry-is-bad section and data.chemical_matter",
"current": "There is no reported small molecule against any ZF-CxxC domain, and ChEMBL contains no bioactivity series for the KMT2A CXXC domain",
"should_read": "ChEMBL still contains no series; one compound is reported in the primary literature - dimethyl fumarate (PMID 34280370) - and it is an approved drug acting as a promiscuous covalent electrophile, which demonstrates rather than refutes the zinc-finger polypharmacology liability this entry already lists in bite_back"
}
]- A covalent electrophile that modifies more than 2,400 cysteines and inhibits a zinc-coordinating domain at 1 uM is the definition of the polypharmacology this project's own dossiers warn about. Treating it as a starting point would mean starting from the least selective possible chemotype against a family whose defining feature is reactive cysteines.
- Disrupting the CXXC zinc coordination does not merely block DNA binding - it unfolds the domain, as every cysteine mutant in Allen 2006 shows. A covalent CXXC agent is therefore closer to a domain denaturant than to a competitive inhibitor, with whatever that implies for the rest of the 3,969-residue protein.
- If DMF really does engage MLL1 CXXC in cells, the exposed population is enormous and the absence of any reported haematological signal is itself weak evidence that the engagement does not happen at clinical exposure - which would be a rare case of pharmacovigilance bounding a biochemical claim.
- Does DMF inhibit the other eleven human CXXC domains at similar concentrations? A counter-screen in the same assay is the single most decisive missing experiment and would almost certainly settle the selectivity question negatively.
- Does DMF modify the MLL1 CXXC cysteines covalently, and which ones? The mechanism here is inferred from DMF's general chemistry and was not demonstrated on this domain; intact-mass spectrometry would show it in an afternoon.
- Does DMF do anything to fusion occupancy or HOXA9 expression in a KMT2A-rearranged cell line? There is no cellular experiment of any kind in this paper.
- Has anyone examined KMT2A-rearranged disease outcomes in the DMF-exposed population? The authors call the clinical effect unexplored and it is a pharmacovigilance-registry question rather than a bench one.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Dimethyl fumarate inhibits the MLL1 CXXC-DNA complex with a relative IC50 of 0.91 +/- 0.09 uM by single-molecule hairpin-unfolding-time dose response at 1 uM CXXC domain, while monomethyl fumarate is inactive at 10 uM; DMF is an electrophilic approved drug that covalently modifies more than 2,400 cysteine residues, and CXXC-family proteins have not been identified as DMF targets by chemical proteomics.
Dynamics and inhibition of MLL1 CXXC domain on DNA revealed by single-molecule quantification. · Liang L, Ma K, Wang Z, et al. · Biophys J 2021What this evidence cannot show
A RELATIVE IC50 from a four-parameter logistic fit to single-molecule unfolding times, not a conventional biochemical IC50 against a defined enzymatic or binding readout; N greater than 20 traces from more than 3 molecules per concentration, which is a small number of molecules. The readout is mechanical complex stability under 12.5 pN of applied force, not occupancy. Isolated synthesised domain on naked DNA, no cell, no fusion protein, no counter-screen against any other CXXC domain, and no demonstration that DMF modifies this domain's cysteines (the covalent mechanism is inferred from DMF's known chemistry, not shown here). The authors present it explicitly as a proof of concept for the assay rather than as a drug-discovery result. Full text read 2026-09-21 via the PMC HTML route (PMC8392069).
Evidence that argues against this (1)
- This project's own standing caution applies with full force: a ChEMBL row count measures how much medicinal chemistry has been done, and a single hit from a promiscuous covalent electrophile measures how reactive the target's cysteines are. Neither measures tractability. The right comparison is that DMF would be expected to inhibit KDM2B, CXXC1, TET1 and DNMT1's CXXC domains at similar concentrations by the same mechanism, and none of those counter-screens was run - so this compound cannot pass the phase-2 dossier's selectivity gate and there is no evidence it would. Risner LE et al. 2013
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The KMT2A CXXC-DNA interface is a flat, electropositive, solvent-exposed protein-DNA surface shared across a family of twelve human proteins, which is the least tractable geometry for a selective small molecule, and the interface is genetically validated as a required function rather than as a uniquely required protein surface.
Functional specificity of CpG DNA-binding CXXC domains in mixed lineage leukemia. · Risner LE, Kuntimaddi A, Lokken AA, et al. · J Biol Chem 2013What this evidence cannot show
The tractability judgement is a general medicinal-chemistry expectation about protein-DNA interfaces, not a measurement; the genetic-validation half is murine MLL-AF9 and is measured. The family size of twelve is from this graph's own UniProt/Pfam/InterPro census, verified at the coordinator 2026-09-21.
Evidence that argues against this (1)
- A covalent approach changes the geometry argument, because a covalent warhead does not need a pocket - it needs a reactive residue in reach, and this domain has eight cysteines. So the flat-interface objection is weaker against covalent chemotypes than against reversible ones, and DMF's sub-micromolar activity is consistent with that. The selectivity objection is correspondingly stronger, because the reactive cysteines are the family's defining feature. Liang L et al. 2021
The dossier recorded no CXXC-directed chemical matter. One compound is published - dimethyl fumarate, relative IC50 0.91 uM - and it strengthens the STOP verdict because it is a promiscuous covalent electrophile that would not pass the selectivity gate.
Corrects 'there is no reported small molecule against any ZF-CxxC domain'. One is reported; there is still no series.
The only published chemical inhibition of the interface this entry maps, from the same paper that supplies the near-physiological 5.7 uM affinity.
Same source paper as row 6 of the affinity ladder; the drug result and the tagless 150 mM measurement arrived together.
- inhibits →KMT2A / MLL1 — the gene that breaks
Dimethyl fumarate inhibits the MLL1 CXXC-DNA complex at a relative IC50 of 0.91 uM in a single-molecule assay; isolated domain, naked DNA, no cell.
- related to →The human CXXC/CpG-reading family is exactly twelve proteins, and TET2 is not one of them
A cysteine-reactive electrophile would be expected to hit all twelve family members equally; none was counter-screened, so the census is the selectivity liability this result makes concrete.
The sixth paper in the ladder also carries the interface's only published inhibitor.