The MT/BCB domain, Q03164 1147-1244: the CXXC fold plus a 41-residue basic tail nobody has measured on its own
also called MT domain · BCB domain · basic-cysteine-basic domain · methyltransferase-homology domain of MLL · GST-MT
The famous CpG-gripping module of the fusion is usually drawn as a 57-residue zinc finger. The original 2002 work used a longer 98-residue piece that adds a tail loaded with positively charged residues, and nobody has ever measured what that tail contributes to gripping DNA.
The same thing, for a scientist
The region Birke 2002 calls the MT (methyltransferase-homology) or BCB (basic-cysteine-basic) domain is Q03164 1147-1244, 98 residues, stated as such in print by Erfurth 2008. This entry establishes the boundaries independently: the PCR primers given in Birke's Methods translate exactly onto the live UniProt Q03164 sequence at K1147 (forward, KKGRRSRR) and P1244 (reverse, NVVDSSQKP), in frame with pGEX-3X. The minimal CXXC used for every structure and for the 44 uM figure is 1147-1203 - the same N terminus, 41 residues shorter at the C terminus. Those 41 residues (AYLQKQAKAVKKKEKKSKTSEKKDSKESSVVKNVVDSSQKP) carry 13 lysines against 5 acidic residues, net charge about +8, and fall inside a UniProt-annotated disordered region (1200-1375). Every published affinity for the longer construct is GST-tagged and was measured on an SPR chip; every tagless solution measurement stops at 1203 or 1214. The basic tail's contribution to DNA binding is therefore unmeasured, and it is the single most consequential unknown in this beat because it is the part of the address the CXXC-centric literature has been ignoring.
How the boundaries were established, since the paper gives primers and not residue numbers
Birke et al. give the construct only as PCR primers 1:
BamMTfw 5'-caagg GGATCC t aaagaaaggacgtcgatcgaggcgg-3'
EcoMTrev 5'-acac GAATTC ccaggtttctgactagagtccaccacgttc-3'
Translating the forward primer downstream of the BamHI site, in the frame pGEX-3X imposes (its BamHI site spans a codon boundary, which is why the primer carries a single filler t), gives aag aaa gga cgt cga tcg agg cgg = K K G R R S R R. Against the Q03164 sequence fetched live from UniProt on 2026-09-21 (3,969 aa), KKGRRSRR occurs at 1147-1154, uniquely.
Reverse-complementing the reverse primer gives g|aac gtg gtg gac tct agt cag aaa cct|gg = N V V D S S Q K P, which occurs at 1236-1244, uniquely.
So GST-MT = Q03164 1147-1244, 98 residues. The paper's own prose independently calls it "this 100 amino acid domain" 1, and Erfurth et al. state it outright as "MLL amino acids 1147-1244" 2. Three routes, one answer.
This matters because the N termini are identical. Birke's construct and Cierpicki's both begin at K1147. The entire construct difference between the 33 nM measurement and the 44 uM measurement is 41 residues at the C terminus, and the graph has been describing them as two different proteins when they are one protein truncated in two places.
What the 41 residues are
Q03164 1204-1244:
A Y L Q K Q A K A V K K K E K K S K T S E K K D S K E S S V V K N V V D S S Q K P
- 13 lysines, 0 arginines, 5 acidic residues (3 E, 2 D), net charge roughly +8 at neutral pH.
- No aromatic cluster, no hydrophobic core, no predicted secondary structure. UniProt annotates Q03164 1200-1375 as a disordered region (fetched 2026-09-21).
- This is the C-terminal 'B' of BCB - basic-cysteine-basic - the alternative name Birke gives the domain precisely because of the two basic flanks around the cysteine-rich core 1. The N-terminal 'B' is
KKGRRSRRat 1147-1154, which is inside the minimal CXXC construct. The C-terminal one is not.
A disordered, net +8 extension immediately C-terminal to a DNA-binding fold is the textbook architecture of an electrostatic affinity enhancer: it contributes non-specifically to the association rate and to the salt dependence without adding sequence readout. If that is what it does here, then it raises affinity for any DNA, which makes the domain a stronger gripper and a worse address-reader at the same time - and it is exactly the modification that would make a 300 mM NaCl measurement of the short construct maximally unrepresentative.
Where the 1242 figure in this graph came from, and what UniProt says today
Two entries in this graph record the ZF-CXXC feature as extending "to roughly 1242" (interface-kmt2a-cxxc-cpg-dna E3 caveat and data.motif_a; target-kmt2a-cxxc-dna-interface data.domain_boundaries, which says 1147-1242). Live UniProt on 2026-09-21 gives the Zinc finger CXXC-type feature as 1147-1195 - forty-seven residues shorter than the graph states, and shorter even than the 1147-1203 construct that was crystallised.
Both numbers can be defended if they are labelled: 1147-1195 is the UniProt zinc-finger feature, 1147-1203 is the structural construct, 1147-1244 is the MT/BCB domain. What cannot be defended is attributing 1242 or 1244 to the UniProt ZF-CXXC feature, which is what those entries do. Corrected in interface-kmt2a-cxxc-cpg-dna this cycle; reported for target-kmt2a-cxxc-dna-interface.
| partner a | gene-kmt2a |
|---|---|
| partner b | unmethylated CpG DNA |
| motif a | KMT2A MT/BCB domain, Q03164 1147-1244 (98 aa): basic patch 1147-1154, CXXC zinc-finger core, basic patch 1204-1244 |
| motif b | CpG-containing duplex DNA; six CpG sites on a 52-bp duplex in the only measurement of this construct |
| uniprot | Q03164 |
| length AA | 3969 |
| kd NM | 33 |
| fate in fusion | retained in full - the whole 1147-1244 region lies about 200 residues N-terminal to the breakpoint cluster region and is present in every leukaemogenic KMT2A fusion |
| structures note | NO structure of 1147-1244 exists. 2KKF, 2J2S, 2JYI and 4NW3 all cover the minimal CXXC only; the 41-residue basic tail is absent from every deposit. |
| shown for fusion | NO - nobody has looked in any fusion. The only measurement of this construct is a GST fusion of the isolated domain on naked oligonucleotide DNA. |
construct boundaries
{
"mt_bcb_domain": "1147-1244 (98 aa) - Birke 2002 GST-MT; stated in print by Erfurth 2008",
"minimal_cxxc_structural": "1147-1203 (57 aa) - PDB 2KKF, Cierpicki 2010, Erfurth 2008 ITC, Risner 2013 FP",
"allen_nmr_itc": "1146-1214 (69 aa)",
"allen_gelshift": "1136-1208 (73 aa)",
"uniprot_zf_cxxc_feature": "1147-1195 (49 aa), fetched live 2026-09-21",
"uniprot_disordered_region_overlapping": "1200-1375"
}post cxxc basic patch
{
"span": "1204-1244",
"length_aa": 41,
"sequence": "AYLQKQAKAVKKKEKKSKTSEKKDSKESSVVKNVVDSSQKP",
"lysines": 13,
"arginines": 0,
"acidic": 5,
"approx_net_charge": 8,
"structural_status": "inside UniProt disordered region 1200-1375; no structure of this segment exists",
"affinity_contribution": "NEVER MEASURED IN ISOLATION OR BY DIFFERENCE"
}primer derivation
{
"forward_primer": "caagggatcctaaagaaaggacgtcgatcgaggcgg",
"forward_translated": "KKGRRSRR",
"forward_maps_to": "Q03164 1147-1154 (unique match)",
"reverse_primer": "acacgaattcccaggtttctgactagagtccaccacgttc",
"reverse_complement_translated": "NVVDSSQKP",
"reverse_maps_to": "Q03164 1236-1244 (unique match)",
"vector": "pGEX-3X (GST, obligate homodimer)",
"sequence_source": "rest.uniprot.org/uniprotkb/Q03164.fasta, fetched 2026-09-21, length 3969"
}kd measurements
[
{
"ligand": "GST-MT (1147-1244) + CpG52 52-bp six-CpG duplex, DNA immobilised",
"kd_nM": 33,
"method": "SPR, k_d/k_a from initial rates, n = 8",
"source_pmid": "11842107"
},
{
"ligand": "GST-MT (1147-1244) + fully methylated CpG52",
"kd_nM": ">1000000",
"method": "SPR, rough estimate only - binding too weak to fit",
"source_pmid": "11842107"
}
]structures
[]
- If the basic tail carries most of the affinity, a ligand designed against the folded CXXC surface from 2KKF or 4NW3 is designed against the minority of the interaction, and would be expected to fail in cells for reasons no in vitro assay on the short construct would reveal.
- A disordered, highly basic DNA-affinity extension is the least druggable object in structural biology: no pocket, no fixed conformation, and a ligand would have to be polyanionic.
- Conversely, if the tail matters, the CXXC-DNA interaction is more nearly a two-part address than the graph assumes, and a fusion could in principle lose CXXC function while retaining tail-mediated chromatin association - which would make a CXXC-directed agent look inactive for a reason unrelated to target engagement.
- What does 1204-1244 contribute to DNA affinity? A tagless 1147-1203 versus 1147-1244 titration at 150 mM salt on one oligonucleotide answers it in an afternoon and has never been done.
- Does the basic tail contribute sequence-independent affinity only, or does it change the salt dependence, the association rate, or the residence time? Those have different consequences for a chromatin scanning model.
- Is 1244 a real boundary? The disordered region runs to 1375 by UniProt, so a 1147-1300 construct is the obvious next test and nobody has made one.
- Does the tail remain accessible in the fusion protein, or is it occluded by AFF1, menin or LEDGF? No structure covers it, so this is unknown even in principle at present.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Birke et al. expressed the MLL MT domain as a GST fusion from pGEX-3X using primers whose translated product maps onto Q03164 1147-1244 (98 residues), describe it in their own text as a 100-amino-acid domain, and give it the alternative name BCB (basic-cysteine-basic) for its two basic flanks around a cysteine-rich core.
The MT domain of the proto-oncoprotein MLL binds to CpG-containing DNA and discriminates against methylation. · Birke M, Schreiner S, García-Cuéllar MP, et al. · Nucleic Acids Res 2002What this evidence cannot show
The residue numbers here are this cycle's derivation from the primer sequences printed in the paper's Methods, translated against the Q03164 sequence fetched from rest.uniprot.org on 2026-09-21; the paper itself states no residue numbers, and it describes MLL as 3,968 aa where the current UniProt entry is 3,969, so any residue number quoted in that era is potentially off by one. The derivation is corroborated independently by Erfurth 2008 stating 1147-1244 in print and by the paper's own 100-residue description. Full text read via the PMC HTML route (PMC100340) after Europe PMC fullTextXML returned HTTP 500.
Evidence that argues against this (1)
- The construct boundary does not tell you the domain boundary. Nothing establishes that 1244 is a structural or functional edge rather than a convenient PCR endpoint; the region is predicted disordered and extends to at least 1375 by UniProt, so a construct ending at 1300 or 1350 might behave differently again. The MT/BCB domain is defined by homology to DNMT1's regulatory region, not by a solved structure. Allen MD et al. 2006
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Erfurth et al. state in print that the MLL methyltransferase-homology (MT) domain is MLL amino acids 1147-1244 and includes a CXXC domain, and that their own ITC work used the shorter 1147-1203 construct.
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
A statement in a discussion paragraph, not a mapped domain boundary from a structure or an alignment figure. It agrees with this cycle's primer derivation to the residue, which is the reason to trust it.
Evidence that argues against this (1)
- Allen et al. used V1146-K1214 for NMR and calorimetry and T1136-K1208 for gel shifts, two further boundaries that agree with neither 1203 nor 1244. There is no consensus construct for this domain and every group has chosen its own, which is itself part of why the affinity literature does not compare. Allen MD et al. 2006
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProt Q03164 (3,969 aa) annotates a single CXXC-type zinc finger at 1147-1195 and a disordered region at 1200-1375, so the 41-residue basic tail that distinguishes the MT construct from the minimal CXXC construct lies entirely within predicted disorder.
UniProtKB Q03164 (KMT2A_HUMAN), feature table and sequence · UniProt Consortium 2026What this evidence cannot show
Fetched live 2026-09-21 by REST, reading ft_zn_fing, ft_domain, ft_region, sequence and length in one call; length 3,969 confirms the sequence version used for the primer mapping. A UniProt disorder region is a prediction and a feature boundary is a curator's call, not a measurement (brief rule 91) - no experiment has shown 1204-1244 to be disordered in isolation or in the full-length protein.
Evidence that argues against this (1)
- Two entries in this graph state the UniProt ZF-CXXC feature reaches to roughly 1242, which the live record contradicts. Because the same 1242 figure appears in two independently written entries it may come from an older UniProt release or from a Pfam/InterPro boundary rather than from an error of transcription, in which case the live value is the correction and the provenance of the old one is still unexplained. UniProt Consortium 2026
independently places the adjacent 1147-1244 region ~200 residues N-terminal to the cluster
The 33 nM end of the ladder is the only measurement of the 1147-1244 construct, and this entry establishes its boundaries.
- part of →KMT2A / MLL1 — the gene that breaks
Q03164 1147-1244, the MT/BCB domain, derived from Birke's primers against a live UniProt fetch.
The structural entry describes 1147-1203; this entry describes the 98-residue construct the functional literature has always used, and corrects the UniProt ZF-CXXC boundary from 'roughly 1242' to the live 1147-1195.
The 41-residue basic tail this entry delimits is the subject of the hypothesis.
The whole 1147-1244 region lies about 200 residues N-terminal to the breakpoint cluster region and is retained in every leukaemogenic KMT2A fusion.
The construct that produced the 33 nM figure cited in the CpG-address entry is 41 residues longer than the one that produced 44 uM, with the same N terminus.
The MT/BCB name is attached to the longer construct and the CXXC name to the shorter one, so the naming split and the construct split are the same split.
- ← related toA published Kd is a property of five experimental choices, and this beat measured how much each one is worth
The construct-span term of the concept, which is the one term in this literature that has never been priced.
The hypothesis is about the 41 residues this interface entry delimits.
Prices the 1204-1244 basic tail directly.