The MLL-protected Hoxa9 CpG sites bind the CXXC domain 4- to 9-fold tighter than the unprotected ones, in one ITC series at fixed everything else
also called Erfurth ITC table · CpG1 versus CpG3A · Hoxa9 CpG cluster affinity
In one clean experiment, the fusion's CpG-gripper was titrated against five short DNAs taken from the Hoxa9 gene. The pieces that MLL actually keeps unmethylated in living cells bound several times more tightly than the pieces it does not, which is the only published hint that the module can tell one CpG site from another.
The same thing, for a scientist
Erfurth 2008 measured the KMT2A CXXC domain (Q03164 1147-1203, tagless after thrombin cleavage) against five Hoxa9-derived oligonucleotides by ITC at 300 mM NaCl and 27 C. The two sites that show MLL-dependent protection from methylation in cells, CpG1-I and CpG1-II, bind at 3.5 +/- 0.4 uM and 7.4 +/- 1.3 uM. The two matched sites from CpG3A that show no MLL-dependent protection bind at 32 +/- 4.2 uM and 15 +/- 3.5 uM. An oligonucleotide with no CG motif gives no detectable binding. Construct, tag state, buffer, salt, temperature and method are constant across all five, so the 9-fold spread is a property of the DNA alone. This is the graph's only quantitative evidence that CXXC-DNA affinity is not flat across CpG-containing sequences, and it sits in tension with Cierpicki 2010's later statement from the same laboratory that the domain reads nothing beyond the CpG dinucleotide.
The table
All measurements: Q03164 1147-1203, GST cleaved with thrombin, ITC, 25 mM phosphate pH 7.2 / 300 mM NaCl / 1 mM DTT, 27 C 1.
| oligonucleotide | sequence | MLL-protected in cells? | Kd (uM) | N |
|---|---|---|---|---|
| CpG1-I | GGGTCGCGGGAG | yes | 3.5 +/- 0.4 | 1.19 |
| CpG1-II | GAGCGCGCGCCT | yes | 7.4 +/- 1.3 | 0.91 |
| CpG1-III | CAGCGGGCGGGCGCCT | yes with wild-type MLL, not with MLL-AF4 | 23.6 +/- 7.2 | 1.93 |
| CpG3A-I | TGCCGGGCGGAC | no | 32 +/- 4.2 | 1.20 |
| CpG3A-II | TGACGCGCGTGG | no | 15 +/- 3.5 | 1.10 |
| no-CG control | - | - | no binding detectable | - |
Three things are in this table that nothing else in the graph supplies.
- A 9-fold affinity range at fixed protein, salt, temperature and method. Whatever else is true, the choice of oligonucleotide is a first-order variable in any published CXXC Kd. Any entry comparing a CXXC affinity to anything must name the DNA.
- The direction is the functionally interesting one. Protected 3.5 and 7.4 uM; unprotected 32 and 15 uM. The authors chose the CpG3A sequences deliberately to have similar CpG spacing to CpG1, which is the control that makes the contrast worth something.
- A stoichiometry result that is itself mechanistic. CpG1-III has three CpGs and gives N = 1.93, not 3. Erfurth's reading is that one CXXC domain sterically blocks binding at an adjacent site, so a CpG cluster is not a set of independent sites - the domain's footprint extends past the dinucleotide it reads. That is the structural basis for the modest 2-4 fold multiple-CpG gain Cierpicki later reported, and it constrains any avidity model built on CpG-island density.
The tension with the later paper from the same laboratory
Cierpicki 2010, two years later and with overlapping authorship, concludes from its own ITC panel that "the CXXC domain does not recognize DNA bases beyond the CpG dinucleotide and therefore is not solely responsible for specific localization to the Hoxa9 locus" 2. Erfurth's five-oligonucleotide spread is 9-fold with CpG spacing controlled.
Both can be true and the honest statement is narrow: there is no evidence for base-specific readout outside the CpG - no hydrogen bonds to flanking bases are seen in 2KKF, and Q1187 contacts the CpG guanine only - but there is a measured 9-fold spread across CpG-containing 12-mers that the base-readout question does not explain. Candidate explanations Erfurth does not separate are CpG number, CpG spacing, duplex stability, and the electrostatic potential of the flanking sequence. The 9-fold is real and its mechanism is not established.
What this does not do is rescue the CXXC as an address. Nine-fold is small against the tens of millions of CpG dinucleotides and roughly thirty thousand CpG islands in the genome, and Cierpicki's conclusion follows regardless.
| partner a | gene-kmt2a |
|---|---|
| partner b | Hoxa9 CpG-cluster DNA |
| motif a | KMT2A CXXC 1147-1203, tagless |
| motif b | 12- and 16-bp duplexes from the murine Hoxa9 CpG1 and CpG3A clusters |
| uniprot | Q03164 |
| method | isothermal titration calorimetry |
| conditions | 25 mM phosphate pH 7.2, 300 mM NaCl, 1 mM DTT, 27 C, VP-ITC |
| kd NM | 3500 |
| protected mean kd NM | 5450 |
| unprotected mean kd NM | 23500 |
| fold spread across panel | 9.1 |
| binding is endothermic note | All five titrations are entropically driven - dH between -1069 and -1776 cal/mol, dS between 15.9 and 19.5 cal/mol/K - consistent with a counterion-release-dominated protein-DNA contact, which is exactly the thermodynamic signature that predicts strong ionic-strength dependence and therefore predicts the salt sensitivity seen across the affinity ladder. |
| shown for fusion | PARTIAL - the CpG1-III site is reported as protected by wild-type MLL but NOT by MLL-AF4, which is the only fusion-specific datum in the panel; the affinities themselves are of the isolated domain with no fusion present. |
kd measurements — the numbers behind the chart
[
{
"oligo": "CpG1-I",
"sequence": "GGGTCGCGGGAG",
"mll_protected": true,
"kd_nM": 3500,
"kd_sd_nM": 400,
"stoichiometry_N": 1.19,
"dH_cal_per_mol": -1631,
"dS_cal_per_mol_K": 19.5,
"source_pmid": "18483194"
},
{
"oligo": "CpG1-II",
"sequence": "GAGCGCGCGCCT",
"mll_protected": true,
"kd_nM": 7400,
"kd_sd_nM": 1300,
"stoichiometry_N": 0.91,
"dH_cal_per_mol": -1187,
"dS_cal_per_mol_K": 19.5,
"source_pmid": "18483194"
},
{
"oligo": "CpG1-III",
"sequence": "CAGCGGGCGGGCGCCT",
"mll_protected": "with wild-type MLL but NOT with MLL-AF4",
"kd_nM": 23600,
"kd_sd_nM": 7200,
"stoichiometry_N": 1.93,
"dH_cal_per_mol": -1069,
"dS_cal_per_mol_K": 17.7,
"source_pmid": "18483194"
},
{
"oligo": "CpG3A-I",
"sequence": "TGCCGGGCGGAC",
"mll_protected": false,
"kd_nM": 32000,
"kd_sd_nM": 4200,
"stoichiometry_N": 1.2,
"dH_cal_per_mol": -1387,
"dS_cal_per_mol_K": 15.9,
"source_pmid": "18483194"
},
{
"oligo": "CpG3A-II",
"sequence": "TGACGCGCGTGG",
"mll_protected": false,
"kd_nM": 15000,
"kd_sd_nM": 3500,
"stoichiometry_N": 1.1,
"dH_cal_per_mol": -1776,
"dS_cal_per_mol_K": 16.1,
"source_pmid": "18483194"
},
{
"oligo": "no-CG control",
"sequence": "not stated",
"mll_protected": null,
"kd_nM": null,
"note": "no binding detectable",
"source_pmid": "18483194"
}
]structures
[ "2KKF" ]
- If affinity varies 9-fold across CpG-containing 12-mers, a DNA-competitive ligand's apparent potency will vary with the genomic site it is competing at, and a cell assay measuring displacement at one locus will not predict displacement at another.
- The entropic signature means the interaction is driven by counterion release rather than by complementary surface chemistry, which is the worst possible thermodynamic profile for a small-molecule competitor: there is no enthalpic handle to copy.
- Is the 9-fold spread CpG number, CpG spacing, flanking electrostatics, or duplex stability? A designed panel varying one at a time would settle it and costs five oligonucleotides.
- CpG1-III is protected by wild-type MLL and NOT by MLL-AF4, and it is also the only site with N = 1.93. Is the fusion's failure to protect it a consequence of the two-domain occupancy, and does that generalise to dense CpG clusters?
- Every one of these titrations was run at 300 mM NaCl. What does the same panel look like at 150 mM, where the spread could compress or expand depending on whether the sequence effect is electrostatic?
- The binding is entropically driven with small favourable enthalpies. Does that thermodynamic signature hold for the longer MT/BCB construct, or does the basic tail change the mechanism as well as the magnitude?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
ITC of tagless KMT2A CXXC 1147-1203 at 300 mM NaCl, 27 C against five Hoxa9-derived oligonucleotides gives Kd = 3.5, 7.4, 23.6, 32 and 15 uM, with the two MLL-protected CpG1 sites tighter than the two unprotected CpG3A sites chosen for similar CpG spacing, N = 1.93 for the three-CpG 16-mer, and no detectable binding to an oligonucleotide devoid of CG motifs.
MLL protects CpG clusters from methylation within the Hoxa9 gene, maintaining transcript expression. · Erfurth FE, Popovic R, Grembecka J, et al. · Proc Natl Acad Sci U S A 2008What this evidence cannot show
Human MLL CXXC domain against MURINE Hoxa9-derived sequences, in vitro, no fusion protein, no nucleosome, no cell. Five oligonucleotides is a small panel and the protected/unprotected assignment comes from the same paper's own bisulphite data in MEFs, so the correlation is internal rather than independent. The two groups differ in more than protection status - CpG1-I has two CpGs and CpG3A-I has two but in a different context - and the authors make no formal statistical comparison between the groups. Standard deviations are from at least two independent ITC measurements. Full text read via the PMC HTML route (PMC2396713).
Evidence that argues against this (1)
- A 9-fold spread over five oligonucleotides with n of at least 2 each and SDs up to 30% of the mean is suggestive rather than decisive, and with two sequences per group no test can separate 'protected sites bind tighter' from 'these two particular 12-mers bind tighter'. The comparison needs a designed panel, which nobody has run. Cierpicki T et al. 2010
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Cierpicki et al. conclude from their own ITC panel that the CXXC domain does not recognise DNA bases beyond the CpG dinucleotide and is therefore not solely responsible for specific localisation to the Hoxa9 locus, and that sites with multiple CpG motifs are bound only 2- to 4-fold more tightly.
Structure of the MLL CXXC domain-DNA complex and its functional role in MLL-AF9 leukemia. · Cierpicki T, Risner LE, Grembecka J, et al. · Nat Struct Mol Biol 2010What this evidence cannot show
The supporting numbers are in Supplementary Table 1, which is behind the Europe PMC open-access gate and was NOT read this cycle; only the authors' prose summary of it could be verified. The structural claim - that all base-specific contacts come from the 1182-1188 loop and reach the CpG only - is from the main text and was read.
Evidence that argues against this (1)
- The same laboratory's earlier ITC series, at the same salt and on the same construct, shows a 9-fold affinity range across five CpG-containing oligonucleotides with CpG spacing deliberately matched between groups. 'No base-specific readout beyond the CpG' does not by itself predict a flat affinity landscape, and the 2010 paper does not reconcile the two datasets. Erfurth FE et al. 2008
Five of the nine rungs are the Erfurth ITC panel, whose internal 9-fold spread at fixed construct, salt and method is the ladder's cleanest single-variable result.
KMT2A CXXC 1147-1203 binds Hoxa9-derived CpG duplexes at 3.5 to 32 uM by ITC.
Supplies the sequence dependence the structural entry's single-oligonucleotide affinity cannot show.
The Hoxa9 sites MLL protects from methylation bind the CXXC domain 4- to 9-fold more tightly than matched unprotected sites.
A measured 9-fold affinity spread across CpG-containing 12-mers sits in tension with, and does not overturn, the no-readout-beyond-the-CpG conclusion.
The DNA-identity term of the concept, priced at 9-fold.