Fig 1.
A model of the proposed mechanism of binding of CTCF to DNA.
This cartoon is based on the X-Ray crystal structure of CTCF bound to DNA [33]. It portrays the approximate positions of ZFs 3–11 aligned with the most common variation of the CTCF consensus that typically has a non-specific 5 bp linker separating the 2 separate CTCF motifs. ZF domains coloured orange bound to the core consensus, and ZF domains coloured purple bound to the upstream sequence, each bind along the major groove and spiral around the DNA helix. ZF8, coloured green, lies away from the major groove and binds to phosphates across the linker region. The core consensus often has a conserved TGG-like sequence upstream of the 15 bp core but the role of this element in DNA binding is unclear as ZF7 binds to the adjacent CCA element. In practice the number of bases separating the 2 motifs can vary from 2 to 7 bp.
Fig 2.
Conservation among insulator and boundary CTCF sites.
(A) Comparison of sequence homology between previously characterized CTCF binding sites defined as insulators, repressors or chromatin boundaries [1,11,13,14,16,21,24,28,37,40,45,47]. The three lines of sequence in the global CTCF consensus above the table represent the predominant alternative bases allowed at each position with the preferred bases shown above in uppercase [41–43]. Regions protected in published DNase I or DMS footprinting assays are highlighted in bold. Underlined bases represent bases where methylation or carboxyethylation blocks binding of CTCF in interference assays. Above the table are the predicted binding sites of Zn finger domains 3–7 and 9–11 (indicated by brackets and numbers) [33]. Zn Finger 8 is not shown because this domain sits away from the DNA bases above the variable length spacer region. Zn fingers 3–7 bind the core CTCF consensus sequence common to all CTCF sites (primary motif) and Zn fingers 9–11 bind the secondary upstream consensus sequence which is present in ~13% of CTCF sites. Underneath the table are shown a web logo [48] derived from the sequences in the table, and the published global consensus sequence for CTCF sites that contain the upstream binding site [42,43]. (B) Locations of the core and upstream CTCF motifs within CTCF sites found within the human IL-3 gene insulator [11]. The boxes represent regions protected from DNase I in footprinting assays. G bases protected from modification by DMS in in vivo footprinting assays are indicated by open circles. Regions hypersensitive to DNase I (arrows) or modification by DMS (filled circles) are also indicated.
Fig 3.
EMSAs of CTCF binding to the IL3 insulator CTCF site.
(A) Sequences of CTCF EMSA probes used in this study. Mutated residues are shown in bold lowercase and the two consensus sequences are underlined. (B and C) EMSAs using intact or mutated IL3 + 2.9 kb CTCF sites. EMSAs included 4 µg of Jurkat nuclear extract with either the + 2.9, Δ1°, Δ2°(1), Δ2°(2), Δ1°/Δ2°(2), HS4, or ΔHS4 probes (B). Specificity of CTCF binding was confirmed by incubating 4 µg of nuclear extract with 2 µl of anti-CTCF or control IgG antibodies, and by comparison to the chicken beta globin HS4 CTCF site (B). The vertical black lines indicated parts of the gel where 2 irrelevant lanes were deleted. Binding data was quantified by densitometry (C). (D-F) Competition assays of CTCF binding. EMSAs included 4 µg of Jurkat nuclear extract in the presence of increasing amounts unlabelled DNA competitor as indicated above. (F) Densitometric quantification of the EMSAs in panels D and E depicting the efficiency of the mutated +2.9 CTCF sites to compete with wild-type +2.9 site. Values are expressed as the fold decrease in the presence of competitor.
Fig 4.
Enhancer blocking assay of a CTCF site.
Linearised luciferase reporter gene plasmids containing the human IL-3 gene promoter (pIL3) were assayed in Jurkat T cells. As depicted in the cartoon at the right, derivatives of pIL3H contained either just the CSF2 enhancer (pIL3-GME) or the enhancer plus a CTCF site inserted between the promoter and the enhancer (pIL3-C-GME). CTCF site plasmids contained either the intact IL3 + 2.9 kb CTCF site, or this CTCF with either the primary or the secondary consensus motif mutated, as indicated below. The CTCF sites are inserted into the plasmids in the opposite orientation to that shown which means that the NH2 terminus of bound CTCF faces towards the enhancer. Luciferase activities were measured 40 h after transfection plus an additional 8 h after stimulating transfected Jurkat cells with PMA and calcium ionophore. Luciferase activities were expressed relative to pIL3-GME having a value of 1, and represent the mean of 6 transfections, using two independently derived DNA clones. Error bars indicate Standard Deviation. Data was normalised by comparison with the cotransfected Renilla luciferase control plasmid pRL-TK.