Involvement of Cytidine Deaminases in DNA Demethylation in Zebrafish

(A) Schematic of the M-DNA (Methylated DNA) fragment injected into fertilized embryos at the single-cell stage. The four HpaII/MspI sites are indicated. HpaII-resistant (methylated C^meCGG) and HpaII-cleaved species (unmethylated) run at ∼750 bp and ∼250 bp, respectively (B).

(B) M-DNA methylation status was assessed by HpaII susceptibility, with cutting observed on unmethylated (U) (lane 2) but not methylated (Me) DNA (lane 2). MspI-digested (lanes 3 and 6) and uncut DNA (lanes 1 and 4) served as controls. Fragments were detected by Southern blot analysis probed with full-length M-DNA probe.

(C) M-DNA methylation status during development. Total DNA was isolated at the time points shown from embryos injected at the single-cell stage with M-DNA (5 pg or 200 pg), and treated as in (B). M-DNA induced demethylation peaks at ∼13 hpf (Compare lanes 2, 5, 8, and 11).

(D) LC-MS quantitation of 5-meC content of the bulk genome (normalized to total deoxyguanosine content) in genomic DNA isolated from embryos (hpf indicated) injected with methylated M-DNA or unmethylated U-DNA at the single-cell stage.

(E and F) qRT-PCR determinations from embryos injected with M-DNA (E), and at different fragment concentrations (F).

(G and H) Methylation status of M-DNA assessed by HpaII digestion and Southern blotting (G), or LC-MS quantitation of total 5-MeC (H) in total genomic DNA isolated from embryos at 13 hpf, injected at the single-cell stage with M-DNA (200 pg) and morpholinos as indicated. Lanes 1, 7, and 13 correspond to wild-type sample.

AAAmm refers to a set of three control morpholinos against AID (4 pg), Apobec2a (4 pg), and Apobec2b (2 pg) (AAA), which each contain five mismatched (mm) bases (of 25 total to prevent binding) relative to the efficacious morpholino (same amount as controls). For HpaII/MspI susceptibility, one representative of at least three biological repeats is shown. LC-MS measurements; two biological replicates. Asterisks (^*) depict statistical significance (p < 0.05). Error bars: +/- one standard deviation.

Overexpression of a Deaminase/Glycosylase Pair Elicits DNA Demethylation

(A–C) Methylation status assessed by HpaII digestion of total genomic DNA (A), LC-MS quantitation ([B] upper panel), HpaII digestion of M-DNA (Southern analysis) ([B] lower panel), and bisulphite sequencing of M-DNA (C). Lanes 1, 7, and 13 in (A) and lane 1 in (B) correspond to wild-type sample. For (B), M-DNA was injected at 5 pg, below the threshold level for eliciting demethylation on its own (See Figures 1C and 1D). For (C), twenty clones were subjected to bisulphite sequencing, and the methylation status of each HpaII/MspI (CCGG) site reported as a percentage of total sites tested.

(D) Repeat elements from DNA isolated from embryos (13 hpf) injected at the single-cell stage with RNA encoding wild-type AID, along with MBD4 wild-type mRNA.

For each experiment, one representative of at least three biological repeats is shown except in LC-MS measurement where graph is prepared from values of two biological replicates. Asterisks (^*) depict statistical significance (p < 0.05). Error bars are ± one standard deviation.

A PCR Strategy to Detect a G:T Intermediate

(A) Schematic of the PCR reaction for thymine (C^meCGG > CTGG) detection at M-DNA HpaII/MspI sites using an A-tailed primer (only 3 of the ∼22 bases shown) with an adenosine at the 3′ end.

(B) Detection of a G:T mismatch on M-DNA by PCR. M-DNA, AID mRNA, and RNA encoding either wild-type or catalytically inactive hMbd4 (D560A) was injected at the single-cell stage and assessed at 13 hpf.

Gadd45 Proteins Promote Demethylation and Selectively Upregulate Deaminases

(A) Gadd45 family members are upregulated by M-DNA, assessed by RT-PCR.

(B) Gadd45α induces moderate demethylation of M-DNA as detected by HpaII digestion and subsequent Southern blotting. M-DNA is injected at 5 pg which does not induce demethylation on its own. Lanes 1, 4, and 7 correspond to wild-type sample.

(C and D) Lowering the levels of Gadd45 family members via morpholino injection attenuates demethylation. Demethylation of 5-methylcytosine as assessed by HpaII digestion and Southern blotting (C), or LC-MS quantitation of 5-meC (D) in total genomic DNA isolated from 13 hpf old embryos injected at the single-cell stage with M-DNA alone (200 pgs) or along with morpholinos as shown. Lanes 1, 7 and 13 correspond to wild-type sample. Combined Gadd45 Mo refers to the combination of morpholinos to all four Gadd45 family members tested (α,α -like, β, and γ; 2 pg each).

(E and F) Synergy among AID, hMbd4, and Gadd45 for demethylation. Methylation status of total genomic DNA from embryos (13 hpf) was assessed by HpaII susceptibility (E) or by LC-MS quantitation of global 5-methylcytosine levels (F) injected at single-cell stage with mRNAs encoding factors as indicated. Lanes 1, 5, and 9 correspond to wild-type sample. Note: AID, MBD4, and Gadd45α were injected at subthreshold amounts (at 25 pgs each), levels which are not sufficient to induce demethylation alone.

(G and H) Quantitative RT-PCR for deaminase family members (assessed at 13 hpf) injected at single-cell stage with mRNA encoding Gadd45α or Gadd45β (F) and M-DNA (200 pg) alone or with morpholinos as shown (G). For each experiment, one representative of at least three biological repeats is shown except in LC-MS measurement where graph is prepared from values of two biological replicates.

Asterisks (^*) depict statistical significance (p < 0.05). Error bars: are ± one standard deviation.

Methylated plasmid induces demethylation in zebrafish
(A) pCMV-Luc was in vitro-methylated using HpaII methylase which converts CCGG sites to C^meCGG sites. Full methylation was verified by HpaII resistance and MspI susceptibility. Unmethylated (U) and methylated (Me) plasmids were subjected to southern blotting using a 736 bp probe against the luciferase gene cDNA coded on the plasmid (same as M-DNA; see Figure 1). (B) Fertilized zebrafish embryos were injected with 150 pg of in vitro-methylated pCMV-Luc (prepared as in panel A), and genomic DNA was isolated at indicated time points. Genome-wide methylation status was verified by the same procedure, with ethidium bromide staining revealing cleavage, and on the plasmid itself by Southern blotting using the same probe as in panel A. Note that both genome-wide and plasmid demethylation (steady state measurement) is first detected at ∼8 hpf and peaks at 13 hpf, whereas remethylation occurs by 28 hpf.

Efficiency of morpholino knockdown of deaminase proteins and MBD4
(A-B) Morpholino knockdown efficiency of Apobec2a (4pg) (A) and Apobec2b (2pg) (B) was assessed at protein level by western blotting using whole cell extracts derived from wild type or morpholino-injected embryos. Anti-Apobec2a and anti-Apobec2b were specific rabbit antibodies generated from recombinant zebrafish proteins. Anti-Vinculin and anti-panH3 antibodies were used as loading controls. In panel B arrow indicates the band for Apobec2b whereas asterisk indicates a non-specific band. (C-D) Morpholino knockdown of AID and MBD4 was assessed at 80% epiboly in cDNA made from total RNA obtained from wild type embryos or embryos injected with AID Mo (4 pg) or MBD4 Mo (4pg). Spliced and unspliced DNA length is shown. 28S was used as a loading control.

AID morphants, zMbd4 morphants, and Gadd45α morphants show neurogenesis defects
(A) Gross morphological defects are shown by a bright field image of AID, Gadd45α and zMbd4 morphants at 24 hpf. These defects can be reversed to a large extent by injection of mRNAs encoding the indicated proteins that are refractory to the morpholino. AID and Gadd45α mRNA were of zebrafish origin, whereas hMbd4 mRNA was of human origin. AID Scr morpholino and MBD4 Scr morpholino injected embryos showed wild type morphology (not shown). (B-C) Whole mount in situ hybridizations were performed for neurogenin-1 (B; a proneural marker) and sox-2 (C; a gene expressed in neural progenitors) on AID, Gadd45α and zMbd4 morpholino (4pg each) injected embryos or wild type embryos at 80% epiboly (a stage when wild type and morphant embryos look relatively similar in gross morphology). Note the complete absence of neurogenin-1 expression and drastic reduction in sox-2 expression in these embryos. These defects can be compensated by injection of corresponding mRNAs (refractory to the morpholino) as indicated in panel A. zAID and zGadd45a mRNA were injected at 25pgs each whereas hMBD4 mRNA was injected at 50pgs for rescue of respective morphants. AID Scr morpholino and MBD4 Scr morpholino injected embryos showed wild type expression for ngn-1 and sox-2 (not shown).

Protein expression levels of catalytic mutant derivatives
Expression levels of wild type and catalytically-inactive derivatives of AID (A), Apobec2a (B), Apobec2b (C) and Mbd4 (D) were compared by transfection in RKO cells and detection in total cellular extracts by antibodies against the epitope tags as indicated.

Morpholino knockdown efficacy of Gadd45 family members
Knockdown efficiencies of morpholinos against Gadd45α (4pg) (A), Gadd45αlike (4pg) (B), Gadd45β (2pg) (C), and Gadd45γ (2pg) (D) was assessed at 80% epiboly in cDNA made from total RNA obtained from wild type embryos or ones injected with respective morpholinos. Spliced and unspliced DNA length is shown. 28S (A), Gadd45γ (B-C) and Gadd45β (D) were used as a loading controls. Panels A, B, and D contain lanes cut and pasted next to each other from the same gel.

Physical interactions between AID and Mbd4 and enhancement by Gadd45α
Western blot showing the co-immunoprecipitation of V5-AID and Myc-hMbd4 (A-B), V5-Apobec2a and Myc-hMbd4 (C-D) and HA-Apobec2b and Myc-MBD4 (E-F) in the absence (A, C, E) or presence (B,D,F) of Gadd45α when overexpressed in RKO cells. (G-J) Interaction between Myc-hMbd4 and HA-tagged Gadd45α (G), V5-AID and HAGadd45α (H), V5-Apobec2a and HA-Gadd45α (I) and HA-Apobec2b and His-Gadd45α (J) were detected in RKO cell extracts overexpressing the two proteins, suggesting that Gadd45α can individually interact with deaminases or hMbd4. Immune complexes were pulled down using the IP antibody as shown, and interacting proteins were detected in western blot format (WB) antibody as shown.