Versatile and efficient chromatin pull-down methodology based on DNA triple helix formation

Abstract

The goal of present paper is to develop a reliable DNA-based method for isolation of protein complexes bound to DNA (Isolation of DNA Associated Proteins: IDAP). We describe a robust and versatile procedure to pull-down chromatinized DNA sequences-of-interest by formation of a triple helix between a sequence tag present in the DNA and a complementary triple helix forming oligonucleotide (TFO) coupled to a desthiobiotin residue. Following optimization to insure efficient recovery of native plasmids via TFO probe in vitro, the procedure is shown to work under various experimental situations. For instance, it allows capture proteins associated to plasmids hosted in E. coli, and is also successfully applied to recovering nucleosomes in vitro opening many possibilities to study post translational modifications of histones in a genuine nucleosome context. Incubation in human nuclear extracts of a plasmid carrying a NF-κB model promoter is shown to pull-down a specific transcription factor. Finally, isolation of a specific locus from human genomic chromatin has been successfully achieved (Chromatin-of-Interest Fragment Isolation: CoIFI). In conclusion, the methodology can be implemented for capturing proteins that specifically bind to any sequence-of-interest, DNA adduct or secondary structure provided a short sequence tag for triple helix formation is located nearby.

Figure 1

Structure of the TFO probes and the TFT cassettes containing constructs: (a) TFO probe structure: the specificity head (underlined) is formed by a 22-mer oligonucleotide composed of a mixture of LNA and DNA residues. A psoralen residue is grafted at the 5′-end of the oligonucleotide. Its 3′-end is modified with a desthiobiotin residue attached to spacer composed of a linear chain of 124 atoms. (b) Construction of plasmids containing a Triplex Forming Tag (TFT) cassette: pAS03 is derived from pcDNA3.1(+)CAT. pAS03.1 is derived from pAS03 by inserting a TFT cassette containing two different TFO-target sites (TFT-1 and TFT-2). pAS03.2 is derived from pAS03.1 by inserting an additional TFT cassette. pAS04 is derived from pAS03.2 by inserting a third TFT cassette. pUC ori: high copy number origin in E. coli. SV40 ori: replication origin in primate cells expressing SV40 large T antigen. Ap: ampicillin resistance gene. Neo: neomycin resistance gene. CAT: chloramphenicol resistance gene. P_CMV: human cytomegalovirus immediate-early promoter/enhancer.

Figure 2

Plasmid capture in vitro via TFO-mediated triplex formation: 600 ng (≈140 fmol) of each plasmid (pAS03, pAS03.1, pAS03.2, pAS04) are mixed with 8 pmol of TFO-1 for 24 hr, followed by mixing with streptavidin magnetic beads. The plasmid fraction not bound to beads (unbound: UB) is recovered using a magnetic stand (lanes 1–6). After the washing steps, plasmid bound to the beads is released either by boiling (lanes 7–11) or by competitive elution using a biotin-containing buffer (lane 12). Following competitive elution, the beads are re-suspended in a buffer and further eluted by boiling (lane 13). All plasmids are supercoiled circular except for lanes 5 and 11 where the plasmid is linearized. The recovered samples are analysed by agarose gel electrophoresis. Similar results were obtained in three independent experiments and a representative gel picture is shown here. In addition, TFO-3 was shown to perform similarly (data not shown). The loading amounts are 15 ng (lanes 1–6), 30 ng (lanes 7–12), and 60 ng (lane 13) plasmid equivalent. Estimation of plasmid recovery compared to input is: pAS03, <1% (lane 7); pAS03.1, ≈50% (lane 8); pAS03.2, ≈60% (lane 9); pAS04 (lanes 10–12), ≈70%, as quantified by NIH ImageJ software. CC: closed circular. OC: open circular. Dimer: dimerized plasmid.

Figure 3

Nucleosome capture depends upon the presence of tTFO: (a) Schematic view of reconstitution of mono-nucleosome. (b) A double-stranded (209 bp) DNA (RN01) is used for reconstituting mono-nucleosomes in vitro. RN01 is incubated with the component of the histone core (i.e., human histones H2A, H2B, H3, and H4). Gel shift assay is carried out to assess mono-nucleosome reconstitution: octamer / DNA ratio ranging from 0.5 to 1.5. The results show that >50% of DNA is incorporated into mono nucleosomes at octamer/DNA ratio equal to one or above. (c) Cross-linked RN01 nucleosomes are incubated with buffer (no TFO), scrambled TFO (sTFO: a mixture of Scr-1 and Scr-2) the sequences unable to form a triplex with the TFT cassette, or target TFO (tTFO: a mixture of TFO-1 and TFO-3) complementary to the TFT cassette for triplex formation. These samples are subsequently incubated with streptavidin-conjugated beads. Beads are washed and bound material is eluted, followed by heat to reverse the cross-links. A small fraction (0.001%) of each elution product is amplified by PCR for DNA detection, followed by quantification using Image Lab software (Bio-Rad). PCR quantification was performed as detailed in Fig. 6d. We performed three independent experiments that all led to results similar to those shown in the present figure. For western blotting (WB), 80% of the elution products are used for histone H3 detection. Full-length blots are presented in Supplementary Fig. S8.

Figure 4

Isolation of NF-κB proteins from human nuclear extracts via IDAP: (a) Schematic structure of the plasmids. A model NF-κB binding site is only present in plasmid pAS104, not in pAS102. ori: high copy number pUC origin in E. coli. Luc: Luciferase gene for a reporter assay. Ap: ampicillin resistance gene. TFT: Triplex Forming Tag. NF-κB: a binding site for NF-κB proteins. (b) Protein detection via silver staining. 80 ng of plasmids are incubated with ≈ 2.4 × 10⁶ cells equivalent of human nuclear extracts. 20% of the elution products are loaded. Lanes are pAS102 or pAS104 in absence or presence of TFO. Input (nuclear extracts) from ≈ 2.5 × 10⁴ cells is loaded. MW: molecular weight marker. *: streptavidin (c) Plasmid detection via PCR. PCR quantification was performed as detailed in Fig. 6d. 0.003% of the elution products are used as a template. (d) Protein detection via WB. For histone H3 detection: 50% of the elution products and Input (nuclear extracts) from ≈ 1 × 10⁵ cells are loaded; For p65 detection: 22% of the elution products and Input (nuclear extracts) from ≈ 2.5 × 10⁴ cells are loaded. The set of experiments (b ~ d) were independently performed four times with similar results. Full-length blots are presented in Supplementary Fig. S8.

Figure 5

Schematic outline of the CoIFI procedure.

Figure 6

Feasibility of the CoIFI approach using a stably transformed cell line: (a) Schematic structure of the plasmid used to obtain stable human cell lines. ori: high copy number pUC origin in E. coli. Luc: Luciferase gene for a reporter assay. Ap: ampicillin resistance gene. TFT: Triplex Forming Tag. SV40: SV40 early enhancer/promoter. hyg^r: hygromycin resistance gene. NF-κB: a binding site for NF-κB proteins. (b) Reporter assay based on Luciferase activity. The G69 cells are stimulated by exposure to 10 ng/ml of TNFα for 2 hr; luminescence is measured by a luminometer. The data, derived from three independent assays, are expressed as relative light unit for non-stimulated cells (−TNFα) and stimulated cells (+TNFα). (c) Experimental design: Inputs are prepared from ≈0.9 × 10⁶ cells equivalent of cell lines containing the TFT site (+TFT; clone G69) or not (−TFT; clone c2-1). Input samples are incubated with either sTFO (scrambled sequence) or tTFO (target sequence). Together, three experimental conditions designed as a, b, c are tested (see main text for additional explanations). (d) Plasmid detection via PCR. 2% of the elution products are used as a template to amplify a DNA segment in the luciferase gene. Lanes a–c correspond to the conditions outlined in Fig. 6c. The indicated amounts of control plasmid pAS104.2, were similarly amplified. Quantification of the bands by Image Lab software (Bio-Rad) produces the standard curve shown in Supplementary Fig. S7. This experiment was performed independently five times and led to similar results. (e) Histone H3 detection via WB. ≈83% of the elution products and input from ≈ 50 cells equivalent are loaded. Lanes a~c correspond to the conditions outlined in Fig. 6c. (f) p65 detection via WB. Inputs are prepared from ≈2.5 × 10⁷ cells equivalent of clone G69 cell line. Extracts are incubated with either sTFO or tTFO. Input lane is loaded ≈ 200 cells equivalent. Full-length blots are presented in Supplementary Fig. S8.