Endonuclease G preferentially cleaves 5-hydroxymethylcytosine-modified DNA creating a substrate for recombination

Abstract

5-hydroxymethylcytosine (5hmC) has been suggested to be involved in various nucleic acid transactions and cellular processes, including transcriptional regulation, demethylation of 5-methylcytosine and stem cell pluripotency. We have identified an activity that preferentially catalyzes the cleavage of double-stranded 5hmC-modified DNA. Using biochemical methods we purified this activity from mouse liver extracts and demonstrate that the enzyme responsible for the cleavage of 5hmC-modified DNA is Endonuclease G (EndoG). We show that recombinant EndoG preferentially recognizes and cleaves a core sequence when one specific cytosine within that core sequence is hydroxymethylated. Additionally, we provide in vivo evidence that EndoG catalyzes the formation of double-stranded DNA breaks and that this cleavage is dependent upon the core sequence, EndoG and 5hmC. Finally, we demonstrate that the 5hmC modification can promote conservative recombination in an EndoG-dependent manner.

Figure 1.

A component of LiNE preferentially cleaves 5hmC-modified DNA. 4.5 ng (2.6 fmol DNA molecules) of either cytosine or the fully 5hmC-modified 2.7 kbp substrates (A) were incubated with 25 ng LiNE with an increasing molar excess (0.1- to 10-fold) of poly(dIdC) (B). *, radioactive label; S, substrate; P, products.

Figure 2.

Purification and identification of EndoG that preferentially cleaves 5hmC-modified DNA. Purification scheme, using mouse livers as a starting material, for the endonuclease that preferentially cleaves 5hmC-modified DNA (A). 4.5 ng (2.6 fmol DNA molecules) of either unmodified or 5hmC-modified DNA substrates were incubated with each successive column pool (B). Full column activity fractions are shown in Supplementary Figures S4–S7. 4.5 ng (2.6 fmol DNA molecules) of either cytosine or 5hmC-modified substrates were incubated with 50 ng LiNE from wild-type mice or 50 ng LiNE derived from Endog^−/− mice (C). 50 ng LiNE, 6.5 nM Purified EndoG or 6.5 nM Purified EndoG/H128A were incubated with 4.5 ng (2.6 fmol DNA molecules) unmodified or 5hmC-modified 2.7 kbp DNA substrates (D). Quantification of the specific cleavage products resulting from a titration of EndoG and EndoG/H128A (E). S, substrate; P, products.

Figure 3.

The 5hmC-cleavage activity cannot cleave 5meC modified DNA and EndoG preferentially cleaves the sequence 5′-GGGG^5hmCCAG-3′. 4.5 ng each short (130 bp) cytosine, fully 5meC and fully 5hmC substrates (5hmC), ³²P labeled at both ends, were incubated with a titration of LiNE ranging from 37.0 to 333 ng (A). dsOligonucleotides (45 bp) that were singly 5hmC-modified at the indicated position were incubated with purified 12.5 nM EndoG (B). S, substrate; P, products; P₁, product 1; P₂, product 2.

Figure 4.

EndoG both cleaves cytosine hydroxymethylated DNA within the sequence context 5′-GGGGCCAG-3′ and promotes the formation of γ-H2AX foci in vivo. Schematic (not to scale) of the substrate design for the assay, each 4.1 kbp substrate contained either cytosine or 5hmC in place of all the cytosines. The core substrate contains the EndoG recognition sequence while the core mutated substrate has the EndoG recognition sequence mutated (A). 0.5 μg of each of the four DNA substrates was transfected into HeLa cells that were treated with a control or EndoG siRNA 1. The total DNA was recovered 20 h after transfection. Bars on the graph represent the total amount of the core EndoG recognition region remaining (amplicon I) as a fraction of DNA transfected (amplicon II) quantified by qPCR (B). Data are presented as the mean ± standard deviation, n = 7; statistical significance was assigned using a two-tailed Student's t-test. Representative images of γ-H2AX foci and 53BP1 foci observed in HeLa cells stably transfected with either a control plasmid or a plasmid that overexpresses the Tet2 CD. Two different siRNAs targeting EndoG were transfected into both the control and the Tet2 CD cell line and incubated for 48 h (C). γ-H2AX foci were counted and grouped according to the number of foci per cell in each of the six combinations of cell lines evaluated (D).

Figure 5.

EndoG promotes conservative recombination when the DNA is 5hmC-modified. Substrate A lacks the 3′ region of an amplicon containing the EndoG recognition sequence (blue primers) and Substrate B lacks the 5′ region of this same amplicon. If recombination occurs the complete amplicon is restored resulting in a quantifiable product. Recombination is quantified as the recombination amplicon quantity (blue primers) as a percent of the total DNA quantity recovered (orange amplicon). Substrates A and B were either unmodified or 5hmC-modified at every cytosine residues (A). 100 ng of Substrate A and 100 ng of Substrate B were incubated with 15.3 μg LiNE or μg LiNE and 3.2 nM of a specific inhibitor of EndoG (EndoGI) from Drosophila melanogaster as described in the ‘Materials and Methods’ section (B). Data are presented as the mean ± standard deviation, n = 3; statistical significance was assigned using a two-tailed Student's t-test. Recombinant amplicons from the recombination assays in panel B were cloned and sequenced, representative sequences are shown aligned to the expected sequence for a conservative recombinant molecule (top sequence), the relevant sequence of substrate A (sequence second from top) and the relevant sequence of substrate B (sequence third from top) (C). Full sequences are shown in Supplementary Figure S21. Model for the initiation of recombination at 5′-GGGGCCAG-3′ sequences mediated by EndoG and cytosine hydroxymethylation (D).