Selective Nonmethylated CpG DNA Recognition Mechanism of Cysteine Clamp Domains

Methylation
of DNA at CpG sites is a major mark for epigenetic
regulation, but how transcription factors are influenced by CpG methylation
is not well understood. Here, we report the molecular mechanisms of
how the TCF (T-cell factor) and GEF (glucose transporter 4 enhancer
factor) families of proteins selectively target unmethylated DNA sequences
with a C-clamp type zinc finger domain. The structure of the C-clamp
domain from human GEF family protein HDBP1 (C-clampHDBP1) in complex with DNA was determined using NMR spectroscopy, which
adopts a unique zinc finger fold and selectively binds RCCGG (R =
A/G) DNA sequences with an “Arg···Trp-Lys-Lys”
DNA recognition motif inserted in the major groove. The CpG base pairs
are central to the binding due to multiple hydrogen bonds formed with
the backbone carbonyl groups of Trp378 and Lys379, as well as the
side chain ε-amino groups of Lys379 and Lys380 from C-clampHDBP1. Consequently, methylation of the CpG dinucleotide almost
abolishes the binding. Homology modeling reveals that the C-clamp
domain from human TCF1E (C-clampTCF1E) binds DNA through
essentially the same mechanism, with a similar “Arg···Arg-Lys-Lys”
DNA recognition motif. The substitution of tryptophan by arginine
makes C-clampHDBP1 prefer RCCGC DNA sequences. The two
signature DNA recognition motifs are invariant in the GEF and TCF
families of proteins, respectively, from fly to human. The recognition
of the CpG dinucleotide through two consecutive backbone carbonyl
groups is the same as that of the CXXC type unmethylated CpG DNA binding
domains, suggesting a common mechanism shared by unmethylated CpG
binding proteins.