Experiment: a topology-controlled binding survey of the MLL AT-hook region, plus a triple AT-hook alanine mutant in an endogenously tagged fusion
also called cruciform versus AT binding survey · AT-hook triple mutant · GRP core alanine substitution
Two experiments that settle a thirty-two-year-old ambiguity. First, purify the hooks and offer them DNA of the same base composition in different shapes, to see whether they care about the letters or the shape. Second, mutate all three hooks in a real leukemia cell and see whether the fusion still sticks to the nuclear skeleton and still finds its genes.
The same thing, for a scientist
**Part A — binding survey with topology and composition decoupled.** Express and purify the MLL residues ~150-320 fragment spanning all three AT-hooks (UniProt Q03164: 169-180, 217-227, 301-309), tag-cleaved, alongside HMGA1 as a positive control for a canonical AT-sequence reader. Offer a substrate panel in which base composition and topology are varied independently: (i) relaxed linear 70% AT duplex; (ii) the same sequence as a supercoiled plasmid; (iii) a designed inverted repeat that extrudes a cruciform under negative supercoiling, and a point-mutated non-extruding control of identical base composition; (iv) an annotated SAR fragment from the KMT2A bcr and a matched-composition scramble; (v) GC-rich and CpG-island duplex as a CXXC-free negative. Measure by EMSA and by surface plasmon resonance or MST for affinities, and add competition with netropsin, distamycin and Hoechst 33258, which should collapse HMGA1 binding and — if the cruciform result is right — should not collapse MLL's. **Part B — cellular separation of function.** In the endogenously tagged KMT2A::AFF1 line from `exp-fusion-fate-imaging-menin-timecourse`, substitute the GRP cores of all three hooks with alanine. Score, separately: nuclear-matrix retention by biochemical fractionation and by imaging after in-situ extraction; CpG-island occupancy by CUT&RUN; subnuclear distribution; target-gene expression; and proliferation. The hypothesis predicts loss of matrix retention with largely preserved CpG-island occupancy — a dissociation exactly mirroring, and complementary to, the published CXXC result in which locus selection is lost while subnuclear position is preserved.
Why the composition-versus-topology design is the crux
The 1994 result and the motif literature are not in contradiction until composition and topology are decoupled, because AT-rich sequences are also the sequences that bend, unpair and extrude cruciforms most readily. Every naive substrate confounds the two. The inverted-repeat pair in panel (iii) is the decisive comparison: identical base composition, one extrudes a cruciform under supercoiling and one does not. If MLL prefers the extruding one and HMGA1 does not discriminate, the hypothesis is supported; if neither discriminates, the motif-class view wins.
The minor-groove ligand competition is the orthogonal check. Netropsin, distamycin and Hoechst 33258 are documented competitors of the canonical AT-hook minor-groove interaction. A tether that binds junction geometry rather than the minor groove of a straight duplex should be substantially less sensitive to them. This is a cheap, well-precedented discriminator that nobody has applied to MLL.
Part B and the symmetry with the CXXC experiment
The published CXXC mutagenesis showed point mutants that abolish CpG binding, transactivation and transformation while leaving subnuclear localisation intact. The reciprocal experiment — abolish the hooks, keep CXXC — has never been done in a modern system. If it produces the mirror-image result (matrix and compartment association lost, CpG-island occupancy preserved), then the graph can state, with two complementary separation-of-function alleles, that compartment address and locus address are encoded by different modules of the same retained fragment. That would be a clean, publishable structural statement about fusion biology and would also tell the therapeutics programme which module to target for which effect.
If instead the triple mutant loses both, the modules are coupled and the whole decomposition in this beat needs rewriting.
| design | Part A: purified MLL ~150-320 AT-hook fragment versus HMGA1 against a substrate panel decoupling base composition from DNA topology (relaxed AT duplex, supercoiled same sequence, cruciform-extruding inverted repeat versus non-extruding composition-matched control, KMT2A bcr SAR fragment versus scramble, GC/CpG duplex), by EMSA and SPR/MST, with netropsin/distamycin/Hoechst competition. Part B: triple GRP-core alanine substitution of all three AT-hooks in the endogenously tagged KMT2A::AFF1 line, scoring nuclear-matrix retention, CUT&RUN CpG-island occupancy, subnuclear distribution, target-gene expression and proliferation separately. |
|---|---|
| readout | dissociation constants per substrate and per topology state; ligand-competition IC50s; nuclear-matrix retained fraction; CUT&RUN occupancy at CpG-island targets; compartment fractions; target-gene expression; growth |
| controls | HMGA1 as canonical AT-sequence reader positive control; composition-matched non-extruding inverted repeat; scrambled SAR; CXXC point mutant as the reciprocal separation-of-function allele; parental tagged line; single-hook mutants to test redundancy |
| cost tier | med |
| timeline | Part A 3-4 months; Part B 6-9 months, sharing the knock-in line with exp-fusion-fate-imaging-menin-timecourse |
| tests hypothesis | hyp-kmt2a-address-is-dna-structure-not-at-richness |
- If the hooks prove to be conventional AT-sequence readers after all, the entries in this beat that lean on the cruciform result — including the nuclear-matrix explanation and the chromocentre-absence explanation — lose their mechanism and must be downgraded. The experiment is designed to be able to embarrass its own hypothesis.
- A triple mutant that is simply dead tells you nothing about which function was lost; the design depends on the milder GRP-core substitution retaining enough protein integrity to give a dissociated phenotype, and that is the main technical risk.
- Are the three hooks redundant? Single and double mutants will say whether avidity or a specific hook carries the function, which matters for whether a competitive ligand could ever work.
- Does the isolated fragment behave like the hooks in the context of the full 240-kDa fusion? Part A is a reductionist assay by necessity and its answer must be checked against Part B rather than treated as decisive.
- Should the substrate panel include G-quadruplex and R-loop substrates? The graph already carries hyp-r-loop-landscape-kmt2a-r, and a topology reader is a natural candidate for R-loop association, which would link two open entries at no extra cost.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Human KMT2A carries three A.T hook motifs at residues 169-180, 217-227 and 301-309, defining the fragment boundaries and the residues to be substituted.
UniProtKB Q03164 (KMT2A_HUMAN) — Histone-lysine N-methyltransferase 2A · UniProt Consortium · UniProt Knowledgebase 2026What this evidence cannot show
Motif-matched annotation; the construct boundaries (~150-320) are chosen to include flanking context and are not themselves an annotated domain.
Evidence that argues against this (1)
- The three annotated hooks are treated as one interchangeable set, and the founding structural work on this motif says they are not. Huth et al. solved the solution structure of a two-domain HMG-I(Y) fragment bound to the PRDII site and found a stoichiometry of one protein to TWO DNA molecules; binding involves a central Arg-Gly-Arg motif plus two further modules making extensive hydrophobic and polar contacts; and the absence of one of those modules in the THIRD DNA-binding domain accounts for its roughly 100-fold reduced affinity relative to the second. A triple mutant that removes all three KMT2A hooks at once therefore cannot report which hook carries the binding, and the fragment boundaries chosen here (~150-320) assume an equivalence that would be surprising if KMT2A behaves like the protein the motif was defined in. Huth JR et al. 1997
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The canonical AT-hook peptide binds the minor groove of A.T-rich DNA and its binding is competed by netropsin, distamycin and Hoechst 33258, providing an established pharmacological discriminator for minor-groove versus non-minor-groove binding modes.
The A.T-DNA-binding domain of mammalian high mobility group I chromosomal proteins. A novel peptide motif for recognizing DNA structure · Reeves R, Nissen MS · Journal of Biological Chemistry 1990What this evidence cannot show
The competition was demonstrated for HMG-I, which is why HMGA1 is included as the positive control rather than assumed.
Evidence that argues against this (1)
- Searched and empty, recorded explicitly. No independent replication of the MLL AT-hook cruciform-binding result exists: PubMed searches on 2026-09-18 for MLL/KMT2A with cruciform, four-way junction, Holliday junction, non-B DNA and inverted repeat returned the 1994 paper and downstream citations of it, no repeat experiment and no structure. The proposed binding assay is therefore a replication attempt as much as an extension, and should be powered and reported as such - including the possibility that the 1994 result does not reproduce, which would remove the structural premise from `hyp-kmt2a-address-is-dna-structure-not-at-richness` as well as from this experiment. No citation is attached because there is nothing to cite: this note is a bounded absence: the replication it asks for does not exist, not a sourced fact. Uncited and unciteable look identical in this field, so it is said here explicitly (Q42 citation-integrity repair, 2026-09-20).
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
CXXC point mutants of an MLL oncoprotein abolish non-methylated-CpG binding, transactivation and myeloid transformation while leaving subnuclear localisation intact, establishing the precedent and the design for a module-specific separation of function in this protein.
Binding to nonmethylated CpG DNA is essential for target recognition, transactivation, and myeloid transformation by an MLL oncoprotein · Ayton PM, Chen EH, Cleary ML · Molecular and Cellular Biology 2004What this evidence cannot show
The localisation readout in the original was fluorescence microscopy of a construct in COS cells; the proposed experiment upgrades this to endogenous tagging and CUT&RUN, so results are complementary rather than directly comparable.
Evidence that argues against this (1)
- The experiment uses the CXXC mutant as its calibration point for 'abolishes function'. Risner et al. complicate that calibration: swapping the MLL CXXC domain in MLL-AF9 for heterologous CXXC or MBD domains - including ones that do not bind the relevant DNA - still delivered the fusion to Hoxa9 in primary progenitors, and the variable that separated leukemogenic from non-leukemogenic swaps was protection of specific CpGs from methylation rather than targeting. So a CXXC point mutant's loss of transformation should not be interpreted as loss of locus targeting, and an AT-hook mutant compared against it inherits that ambiguity. Read: abstract only. Risner LE et al. 2013
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Deletion of the AT-hook motifs impairs myeloid transformation by an MLL fusion, so a triple AT-hook mutant is expected to have a measurable phenotype and the experiment is not predicted to be silent.
The oncogenic capacity of HRX-ENL requires the transcriptional transactivation activity of ENL and the DNA binding motifs of HRX · Slany RK, Lavau C, Cleary ML · Molecular and Cellular Biology 1998What this evidence cannot show
Deletion, not point substitution, and in HRX-ENL under retroviral overexpression; point substitution of the GRP cores is a milder perturbation that preserves protein length and may have a weaker phenotype.
Evidence that argues against this (1)
- The imaging arm of this experiment is likely to be flat, and for a reason that is already in the literature. Yano et al. dissected ALL-1 and mapped the determinants of punctate subnuclear distribution to two short sequences within the N-terminal approximately 1100 residues that are conserved with TRITHORAX (the elements later called SNL1 and SNL2), not to the AT hooks; enforced nuclear expression of the PHD fingers or the SET motif gave uniform, non-punctate patterns. A triple AT-hook mutant therefore has no published reason to redistribute, so 'the mutant looks the same' would not be evidence that AT-hook DNA binding is dispensable for positioning - it would be the predicted result. The experiment should score chromatin dwell time, occupancy and transformation, and should treat the imaging channel as a negative control rather than as a readout. Read: abstract only. Yano T et al. 1997
Counter-evidence reframes the experiment as a replication: the 1994 cruciform gel-shift has no independent repeat in 30 years, and the imaging channel is predicted flat because positioning maps to the SNLs, not the hooks
decouples base composition from DNA topology, plus a cellular triple mutant
the reciprocal separation-of-function allele to the published CXXC mutants