CTCF and octamer sites regulate ICR DNA methylation in F9 cells

The mouse Igf2/H19 ICR has four CTCF sites and a closely spaced pair of Sox-Oct motifs (Sox-OctA and Sox-OctB) that lie about 200 bases from CTCF site 2 (Fig. 1A). Previous work indicated that CTCF and octamer sites regulate CpG methylation of the ICR, but whether these sites collaborate is unknown [9]–[11], [13], [16]–[19]. To more fully address octamer functions in the ICR, we searched for similar sequences and found an additional consensus octamer site 450 bp upstream of CTCF site 1 (Fig. 1A). To determine whether the octamers and CTCF sites cooperate in regulating ICR methylation, we mutated the three octamers and four CTCF sites in ICR-containing plasmids and assessed the methylation status of the transgenes after their stable incorporation into F9, C2C12 or 3T3 cells. F9 embryonic carcinoma cells were chosen, as they express Sox2 and Oct4, and have demonstrated both de novo methylation and demethylation of transgenes. C2C12 and 3T3 cells, on the other hand, do not express Sox2 and Oct4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Analysis of ICR transgene methylation in F9 cells.

(A) Top: Structure of mouse ICR transgene depicting consensus octamer, O (black oval); four CTCF sites, C1–C4 (white ovals); two Sox-Oct motifs (striped oval). Middle: ICR sequence encompassing both Sox-Oct motifs (A and B) and a downstream Sox site (SoxC). Nucleotides matching the respective consensus sequences are underlined. Bottom: HhaI, and HpaII restriction maps for the 2.4 kb BglII,(Bg) ICR fragment. (B–E) Southern blot analysis of F9 genomic DNA prepared from polyclonal populations with stably incorporated WT or mutant ICR transgenes. DNA was digested with BglII followed by digestion with HhaI (H), HpaII (P), or MspI (M). The 2.4kb BglII fragment was used as a probe. The endogenous ICR is visible only as a faint signal compared to that generated by the multicopy transgene. Each blot is representative of the data obtained from three independent polyclonal populations. (B) Stable transfectants obtained from unmethylated ICR constructs containing either a WT sequence, (wt ICR); mutations in all four CTCF sites, (mut ctcf); mutation in the three octamer sites, (mut oct); or in all octamer and CTCF sites (mut oct/ctcf). (C–E) All plasmids were in vitro methylated at HhaI and HpaII sites prior to transfection. (C) Stable transfectants obtained from the same WT and mutant constructs as in (B) but with in vitro methylation (D) Stable transfectants obtained from methylated constructs with OctA or OctB mutations combined with either WT or mutant CTCF sites. (E) Stable transfectants obtained from methylated constructs with separate mutations in SoxA, SoxB or SoxC. Both the mutant and WT (wt sox) constructs contain two modified SpeI cloning sites flanking the respective Sox-Oct elements. See Table S1 for WT and mutant sequences.

https://doi.org/10.1371/journal.pone.0081962.g001

Using methylation sensitive Southern analysis, we found that integrated transgenes containing the wild type ICR and H19 promoter did not acquire methylation in F9 cells, while mutation of the octamers resulted in a small amount of de novo ICR methylation, which is shown by the incomplete digestion of the BglII fragment with HpaII (Fig. 1A and B). In contrast to results seen in mice, mutation of the four CTCF sites did not lead to increased methylation of the ICR (Fig. 1B). The protective activity of CTCF was apparent only when both the CTCF and octamer sites were mutated, as the double mutant ICR showed less digestion by HhaI and HpaII than the octamer mutant alone (Fig. 1B).

In addition to protection from de novo methylation, binding by certain transcription factors can drive demethylation of partially methylated sequences. We tested the demethylation activity of the CTCF and octamer sites using stably integrated plasmids that had been treated with HhaI and HpaII methyltransferases, and found that both WT and CTCF mutant ICRs were demethylated in F9 cells (Fig. 1C). By contrast, ICR transgenes with mutant octamers retained most of their partial methylation. Moreover, ICR constructs with single mutations in either OctA or OctB retained methylation to the same degree as the double octamer mutant (Fig. 1D). Combining CTCF and octamer mutations resulted in a modest additional decrease in demethylation activity (Fig. 1C and D). Transgenes containing mutations in the distal octamer showed wild type levels of demethylation regardless of CTCF site functionality, suggesting that it is not involved in regulating ICR methylation (Fig. S1A).

Octamer mediated demethylation of the ICR was both specific to embryonic carcinoma cells and limited to ICR constructs with low density methylation. Partially methylated WT and mutant ICR transgenes stably integrated into either mouse C2C12 or NIH3T3 cells remained methylated (Fig. S1B), which is consistent with a requirement for Sox2 and Oct4 for demethylation. In addition, these results suggest that Oct1, which is expressed in F9, C2C12 and NIH3T3 cells and can bind the ICR octamers (Fig. 2A, 2B and data not shown), is not sufficient to drive ICR demethylation. In contrast to partially methylated plasmids, WT ICR transgenes fully methylated in vitro with SssI methyltransferase remained highly methylated in F9 cells (Supplementary Material, Fig. S1C). Thus, the presence Sox2/Oct4 is not sufficient to drive demethylation of a completely methylated ICR in F9 cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Sox2 and Oct4 binding to the ICR.

EMSAs were performed using WT or mutant ICR probes as indicated with either F9 NE (A, B) or recombinant Oct4 and Sox2 (C, D, F). (A) The effect of octamer specific point mutations on Oct4 and Sox2 binding was assessed by addition of nonspecific (N) and specific unlabeled DNA competitors - oct (O), sox (S), ICR Sox-OctA and B (A and B), or mutant ICR Sox-OctA and B (mA and mB). (B) Oct4 and Sox2 supershift assays. Sox2 and Oct4 protein-DNA complexes (white and black triangles respectively) were identified by their respective antibodies (Ab). Presumptive Oct1 complexes are indicated by a grey triangle. (C) Recombinant Sox2 binding to Sox-OctA, B and SoxC. ICR probes with mutations to the Sox elements (mSA, mSB, and mSC) were used to confirm Sox2 binding sites. (D) Methylation-independent binding of Oct4 and Sox2 to the ICR Sox-Oct motifs Wild type ICR Sox-OctA and B probes (A* and B*) were methylated at all CpGs. (E) Oct4 and Sox2 ChIP assays. Enrichment of endogenous F9 or WT and mutant transgenic ICR octamers by Oct4 and Sox2 ChIP was determined by qRT-PCR and normalized to serum controls. Error bars represent standard deviation of the mean (n = 3). Statistical significance of Oct4 and Sox2 enrichments was shown using Student's t-test (*p<0.05). (F) Cooperative binding of Sox2 to Oct4 complexes. Cooperative binding characteristics of Sox2 to the ICR were assessed by measuring the amount of binary and ternary protein-DNA complexes formed using a fixed amount of Oct4 and increasing amounts of Sox2. See Table S1 for a complete list of probe and competitor sequences.

https://doi.org/10.1371/journal.pone.0081962.g002

Sox binding sites regulate ICR methylation in F9 cells

Octamers are often found adjacent to Sox sites, with binding of both Sox and POU factors required to form a functional unit. We tested the importance of the Sox binding sites in the Sox-Oct motifs using the F9 demethylation assay and partially methylated ICR constructs containing mutations that disrupted each Sox site (Fig. 1E). The Sox site mutations prevented formation of a Sox2-Oct4 ternary complex but did not alter Oct4 binding to the adjacent octamer (Fig. 2C). Southern analysis demonstrated a modest reduction in ICR demethylation activity when SoxA was mutated. However, mutation of the SoxB exhibited a greater loss in demethylation activity, which was comparable to the single octamer mutations. We also identified a sequence (SoxC) 22 bps downstream from Sox-OctB that bound Sox2 in vitro (Fig. 1A and 2B). Mutation of this element, however, did not affect ICR demethylation alone or in combination with the Sox-OctA mutant (Fig. 1E and Fig. S1D).

Oct4 and Sox2 bind cooperatively to the ICR Sox-Oct motifs

The limited demethylation of WT transgenes in NIH3T3 and C2C12 cells suggested that Sox2 and Oct4 binding mediates ICR demethylation. Consistent with this idea, chromatin immunoprecipitations (ChIP) from F9 cells showed Oct4 and Sox2 enrichment at both the endogenous and wild type transgenic ICRs, but not at the octamer mutant ICR (Fig. 2D). Analysis of published data from genome-wide Oct4 and Sox2 ChIPs also revealed that the pluripotency factors are localized to the Sox-Oct motifs in mouse embryonic stem (ES) cells (data not shown) [20]. Furthermore, EMSAs performed with F9 nuclear extract showed that both motifs formed heterodimeric complexes with Oct4 and Sox2, and the same octamer mutations that disrupted ICR demethylation also prevented Oct4 binding in vitro (Fig. 2A and 2B). A lower mobility complex, which is likely to contain the POU factor Oct1, also was present in EMSAs with F9 nuclear extract (Fig 2A). Thus, Oct1 or other POU factors could contribute to or inhibit ICR demethylation. Because the Sox-Oct-dependent demethylation in F9 cells was limited to partially methylated transgenes, we performed EMSAs with ICR probes methylated at all CpGs and found that the methylation did not inhibit formation of Sox2/Oct4 complexes (Fig. 2C). SoxC, which lacked demethylation activity, produced a clear complex with Sox2 but was not able to form a Sox2-Oct4 complex (Fig. 2B and C).

To assess the potential cooperative binding characteristics of Oct4 and Sox2 to the Sox-Oct motifs, we performed EMSAs with recombinant proteins and found that Oct4 bound to both elements equally well (Fig. 2C and E). Sox2, however, bound weakly to the WT and Sox site mutant Sox-OctB probes, and bound to the WT and Sox site mutant Sox-OctA probes almost equally well (Fig 2C and E). Sox2's binding to the mutant Sox-OctA probe is likely to be through the octamer, as it contains a Sox consensus site (data not shown). Although the ectopic binding by Sox2 complicated our ability to measure it accurately, we were able to demonstrate cooperative binding between Sox2 and the Oct4-Sox-OctA complex. First, we established EMSA conditions in which no more than 10–20% of the Sox-OctA probe was bound by either Sox2 or Oct4 when added separately. If the two factors bound independently, then only 1–4% of the probe would be expected to be in a ternary complex with Sox2 and Oct4. However, we found roughly equal amounts of binary and ternary complexes (Fig. 2E), which when quantified indicated a 6–10 fold enhanced binding of Sox2 to the Oct4-DNA complex (Supplementary Material, Fig. S2aA). For the Sox-OctB motif, the amount of ternary Sox-OctB complex was at least 50 fold more than that expected from independent binding of Oct4 and Sox2 (Fig. S2A). In this case, the cooperativity was much more pronounced as Sox2 alone bound to less than 1% of the probe (Fig. 2E). Conserved Sox-Oct motifs in the Human ICR A2 repeat also demonstrated cooperative binding of recombinant Sox2 and Oct4 in EMSAs (Fig. S2B).

Given the proximity of the two Sox-Oct motifs to one another, it was unclear if Sox2 would be occluded from binding between the two Oct4 proteins. Using probes containing both Sox-Oct motifs, we found that recombinant Oct4 and Sox2 were capable of forming several complexes with the probe, including one that was likely to represent full occupancy of both motifs (Fig. S2C). However, this complex represented a small fraction of observed complexes, making it difficult to confidently assign its composition. To simplify the EMSAs, we used a human ICR A2 probe with a mutant Sox site upstream of the first octamer. This probe showed the formation of an Oct4-Sox2-Oct4 complex, which indicated that there is sufficient space for Sox2 to bind between the two Oct4 molecules (Fig. S2C).

The ICR Sox-Oct motifs activate transcription in F9 cells

The capacity of Oct4 and Sox2 to initiate DNA demethylation along with transcriptional activation during cellular reprogramming has been well established [21]. A similar correlation between the ability to activate transcription and to drive DNA demethylation has been demonstrated for other transcription factors, suggesting a mechanistic link between the two events.To determine whether transcriptional activation by the ICR Sox-Oct motif correlated with the demethylation seen in the stable transfections, we placed ICR fragments containing the elements upstream of a luciferase gene driven by the H19 promoter (Fig. 3A). In transiently transfected F9 cells, the Sox-Oct regions activated the reporters more than five-fold compared to the H19 promoter alone. Mutation of one or both octamers abolished transactivation and produced a modest repression. Mutations disrupting the SoxA and B binding sites also abolished transcriptional activation. Mutating SoxC lead to a modest reduction in luciferase expression, suggesting that this element is functional. To determine if Oct4 was required for transactivation, we transfected F9 cells that had been depleted of Oct4 by shRNAs and found that relative luciferase expression dropped at least 90% compared to control shRNA expressing cells (Fig. 3B). Consistent with the Oct4 knock-down, mouse 3T3 cells lacking endogenous Oct4 and Sox2 expression were transiently transfected with ICR-reporter constructs and found to have luciferase levels equal to or less than that of H19 promoter controls (data not shown). Therefore, both transcriptional activation in transient assays and DNA demethylation of integrated ICR transgenes appear to share similar requirements for cooperative Oct4 and Sox2 binding to both Sox-Oct motifs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. ICR Sox-Oct motifs activate transcription in F9 cells.

(A) F9 cells were co-transfected with a β-galactosidase reporter for normalization and an unmethylated luciferase reporter construct driven by the H19 promoter alone or in combination with either a 352 bp or an 83 bp ICR fragment which included WT or mutant Sox-Oct motifs without CTCF sites. Normalized luciferase activities are shown relative to the activity of the H19 promoter alone. (B) The effect of Oct4 shRNA knock-down on relative luciferase activity is shown for constructs containing either two copies of the wt 83 bp ICR fragment (wt 83×2) or the wt 352 bp fragment. S1 and S3 indicate two different Oct4 shRNAs. SCR indicates scrambled control shRNA. ‘O4 shRNA’ includes both S1 and S3 shRNA knockdown samples. Western blot of F9 cell extract (CE) from S1, S3 or SCR treated cells probed with α-Oct-4 and α-NPM antibodies.

https://doi.org/10.1371/journal.pone.0081962.g003

Octamers are required for maintenance of the unmethylated maternal ICR in somatic cells

To determine if the octamers participated in establishing or maintaining the ICR's differentially methylated state in vivo, we used homologous recombination and ES cells to create mice with mutations in the three octamer sites (Fig. 4A and B). Southern analysis of DNA from neonatal liver and muscle (Fig. 4C) showed that the loss of octamer binding had no effect on the methylation of paternally transmitted mutant ICRs. However, methylation of maternally transmitted mutant ICRs increased significantly, as indicated by reduced HpaII and HhaI digestion compared to wild-type maternal ICR DNA. The asymmetric distribution of methyl-sensitive restriction sites in the probed region (with HpaII sites being more 5′ and HhaI more 3′ to the octamers) also provided a means for assessing the relative location of methylated CpG downstream of Sox-OctB. Analysis of the banding patterns from HpaII and HhaI digests suggests that the majority of mutant allele methylation was concentrated near the octamers while the region including the promoter proximal CTCF sites was largely unmethylated.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Octamer mutations increase maternal ICR methylation in neonatal somatic tissues.

(A) Restriction map for the H19/Igf2 locus, targeting vector insertion and the targeted locus following excision of the selection cassette. The neomycin resistance (neo^r) and Cre recombinase (cre) genes were removed in the germline of male chimeras by testes-specific expression of Cre. Restriction sites are AseI (A), BamHI (B), EcorV (E), NdeI (N), and SalI (S). (B) Southern blot analysis to confirm targeting of the mutant ICR (M.ICR) was carried out on EcorV or NdeI digested WT and targeted ES cell DNA hybridized to 5′ and 3′ probes, respectively. Correct Cre-mediated excision of the selection cassette was confirmed by BamHI and AseI-digestion of tail DNA from F1 mice. (C) Methylation sensitive Southern blot analysis was performed on DNA from neonatal liver and muscle from WT and heterozygous mutant mice. Genomic DNA was digested with BamHI and BspHI (Bs) and then HhaI (H), HpaII (P), or MspI (M). Restriction maps of ICR are shown. The mutation in OctB created an additional BspHI (Bs*) restriction site. The 1.4 kb mutant ICR BspHI-BamHI fragment was used for the probe. The results shown are representative of methylation sensitive Southern analysis performed on six pups from four different mutant females.

https://doi.org/10.1371/journal.pone.0081962.g004

To more accurately characterize the increased methylation, we performed bisulfite sequencing of the region including CTCF site 2 and the octamers on DNA from neonatal liver and found increased CpG methylation flanking the mutant Sox-Oct motifs consistent with the Southern analysis. On average, CpGs within 35 bps of the Sox-Oct motifs from maternally inherited mutant alleles were 84% methylated and more than 53% methylated for all CpGs in the fragment. In WT ICRs, CpGs within 35 bps of the Sox-Oct motifs were 20% methylated compared to a 14% average methylation for all CpGs in the amplified fragment (Fig. 5A). In contrast, the three CpGs within CTCF site 2 showed a modest 31% average methylation for the mutant alleles compared to a 19% average methylation for the maternal wild-type alleles. A comparison of WT and mutant maternal alleles from neonatal muscle yielded similar results (Fig. S3). Additional bisulfite analysis of the ICR region surrounding CTCF site 4 from neonatal liver showed virtually no methylation of maternal mutant alleles (Fig. 5A). Paternal methylation imprints were likewise unaffected by the octamer mutations (Fig. 5A and data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Methylation profiles of WT and mutant ICRs. Each horizontal line of circles represents a unique sequence.

Unmethylated and methylated CpGs are presented as open and closed circles, respectively. Multiples of the same methylation pattern are indicated to the right of each sequence. Lines above the circles indicate relative location of CTCF sites 2 and 4 (C2, C4) and the Sox-Oct motifs (SO). The mutant status and the allelic parent of origin are indicated to the upper left of each block of sequences. (A) Bisulfite analysis was performed on neonatal liver from three heterozygous mutant mice and (B) the maternally transmitted mutant ICRs from one 12 dpc embryonic liver. (C) Both WT and mutant bisulfite PCR products from oocytes were the result of four separate reactions each using the DNA from 10–20 oocytes. Oocytes were collected from a combination of homozygous and heterozygous mutant and WT females after superovulation. Each PCR product was cloned separately, and 3–6 clones of each were sequenced. Identical sequences from each product were counted only once. (D) Bisulfite analysis of maternally derived mutant alleles from heterozygous blastocysts was the result of two separate reactions of 6–10 blastocysts each. PCR products from each reaction were cloned separately.

https://doi.org/10.1371/journal.pone.0081962.g005

Octamer mutations increase methylation in oocytes but not blastocysts

To determine if the ectopic methylation of the maternal mutant ICR occurred during oogenesis or after fertilization, we conducted additional methylation analysis of mutant embryos, oocytes and blastocysts (Fig. 5B–D). In somatic tissue from 12 dpc embryos, methylation of the mutant maternal ICR appeared relatively unchanged from methylation patterns present in neonatal tissues (Fig. 5B). In oocytes, bisulfite sequencing of the octamer region revealed that roughly half of the mutant ICRs exhibited methylation patterns similar to those obtained from somatic tissue, while the other half remained unmethylated (Fig. 5C). However, we found the octamer region from mutant blastocysts to be almost completely unmethylated (Fig. 5D). Although we cannot exclude maintenance of the oocyte-derived methylation in some animals, its absence in blastocysts suggests that maternal ICR methylation found in later stages occurred after implantation.

The octamers are not required for establishing an unmethylated ICR in the primordial germline

We assessed mutant ICR methylation in germ cells during different stages of embryogenesis to determine if the ectopic methylation in oocytes resulted from a failure to erase paternal methylation imprints or from inadequate protection from de novo methylation during normal imprint establishment in oocytes [22], [23]. Primordial germ cells (PGCs) were isolated from 14, 16 and 18 dpc ovaries and testes by crossing ICR mutants with a Pou5f1-EGFP strain and FACS sorting the respective dissociated tissues. 94–99% of GFP+ FACS isolated cells tested also stained positive for alkaline phosphatase (data not shown). Bisulfite analysis of PGCs from 14 dpc ovary and testes showed normal hypomethylation of maternally inherited mutant ICRs (Fig. 6). The ICR octamers also were not required for erasure of paternal methylation in female PGCs as evidenced by the absence of methylation on paternally derived mutant alleles from 16 dpc ovaries. Moreover, no de novo methylation of the mutant octamer region was seen in the female germline up to and including 18 dpc development. In embryonic testes, mutant alleles acquired paternal specific methylation imprints irrespective of parental origin (Fig. 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Methylation profiles of WT and mutant ICRs from female and male embryonic germ cells.

Bisulfite sequencing of the CTCF site 2 and Sox-OctA and B region are shown for germ cells isolated from 14, 16, and 18 dpc ovaries and testes from heterozygous mutant embryos. WT and mutant alleles are denoted by white and grey backgrounds respectively. The parental origin of alleles is indicated by the symbol to the left of sequence graphical representations. Each block of bisulfite sequences was the result of 1–2 separate PCR reactions performed with DNA from 1×10³–1×10⁴ GFP-sorted germ cells obtained from 4–16 pooled gonads. DNAs from separate PCR reactions were cloned separately.

https://doi.org/10.1371/journal.pone.0081962.g006

The ICR octamers have little influence on mono-allelic expression of H19 and Igf2

We performed a single nucleotide primer extension assay (SNuPE) in order to determine if the methylation observed on the maternal mutant ICR was sufficient to alter the normal allelic expression ratios of H19 and Igf2 in liver, muscle, tongue, kidney, and brain from neonatal wild type and heterozygous mutant mice. Despite the partial methylation of mutant ICRs in liver and muscle, no significant difference in the ratio of maternal/paternal H19 allelic expression was observed for the maternally transmitted mutant alleles (Fig. 7A). The ratio of paternal to maternal Igf2 expression was similarly unchanged in all tissues assayed except for kidney where we observed an apparent 2.5 fold increase in maternal Igf2 expression (Fig. 7B). We observed no difference in size between neonatal mice that inherited the mutant maternal allele and their WT littermates, which was consistent with normal repression of maternal Igf2 in most tissues (data not shown).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Allelic H19 and Igf2 mRNA expression ratios in neonatal mice.

SNuPE analysis was carried out on (A) H19 and (B) Igf2 cDNA from 4 dpn heterozygous mice having a paternally inherited WT M. cast. allele and a maternally inherited WT or mutant B6 allele. The ratio of transcript for each tissue was determined by averaging the individual allelic ratios for 3–5 different mice. Error bars represent the standard deviation of the mean. The p values were generated using Student's t-test.

https://doi.org/10.1371/journal.pone.0081962.g007

Plasmid construction

Mutations in the ICR were created using overlapping PCR with mutant primers. Substitutions at OctB and the upstream octamer (−3248 and −4338 with respect to the H19 start site) created novel BspHI sites. The 1.5 kb mutant PCR products were ligated into the XhoI and BspEI sites of a pcDNA3-luciferase vector containing the mouse H19 upstream region (from −5547 to +695) with either WT or mutant CTCF sites. WT and mutant Sox constructs were prepared by replacing the octamer region from −3305 to −3209 in the same pcDNA3 vector containing the mouse H19 upstream region with a unique SpeI site via overlapping PCR. WT and mutant 83 bp sequences corresponding to the deleted region were constructed by ligating three double-strand oligonucleotides containing the appropriate mutations into a SpeI linearized vector. (See Table S1 for WT and mutant sequences.)

Luciferase reporter constructs were prepared by inserting either a 352 bp WT or mutant octamer PCR product (−3396 to −3044) or the 83 bp WT or mutant Sox sequence into the NheI site of the pH19-Luc vector [26]. All PCR-derived constructs were sequenced to ensure fidelity.

Oct4 and Sox2 recombinant protein expression vectors were prepared by cloning BamHI – SalI fragments containing the respective cDNAs from pMXS-Oct3/4 and pMXS-Sox2 (Addgene plasmids 13366 and 13367 [31]) into the BamHI site of pET-15b (Novagen).

The mutant octamer knock-in vector was prepared by ligating the mutant ICR contained within an XbaI fragment from −5.5 to −0.8 kb between the lox511 and loxP sites of a previously described targeting vector [9].

Cell culture and stable transfection

F9, NIH-3T3, C2C12, and HEK293T cells (all from ATCC) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and penicillin-streptomycin. F9 cells were cultured on 0.1% gelatin-coated dishes. For ICR demethylation analysis, constructs were treated with HhaI and HpaII methyltransferases for partial methylation and SssI methyltransferase for full CpG methylation. Complete methylation was verified by HhaI and HpaII restriction digest followed by agarose gel electrophoresis and ethidium bromide staining. Only DNA showing no digestion was used in transfections. Cells were co-transfected with test constructs and either a pcDNA3.1 or pBS-puromycin selection plasmid at a ratio of 5∶1 using either FuGENE 6 [30] or MirusTransIT-LT1 at a lipofectant to plasmid ratio of 3∶1. Cells underwent 10 days of selection with either G418 or puromycin and were then passaged onto 60 mm plates and harvested when confluent.

Luciferase assays

Transient transfections of F9 cells were performed in six well plates using 3 µL MirusTransIT-LT1 and 0.5 µg each of test construct and pSV-β-gal (Promega). After 48 hours, cells were harvested in Reporter Lysis Buffer (Promega), and β-galactosidase and luciferase activity were assayed as previously described [32]. Relative luciferase expression was calculated after normalizing for β-galactosidase activity. Student's t-test was used to determine statistical significance of at least three independent transfections.

Lentivirus-mediated Knockdown

Knock-down of Oct4 in F9 cells was achieved using lentiviral particles derived from Sigma-Aldrich MISSION plasmids pLKO.1, (S1 =  Clone ID: NM_013633.1-659s1c1, and S3 =  Clone ID: NM_013633.1-544s1c1). A pLKO.1 vector with a non-targeting shRNA (SCR =  Sigma-Aldrich SHC002) served as a negative control. Lentivirus was produced in 293T cells and used to infect F9 cells on a 6-well plate according to the Addgene pLKO.1 Protocol version 1.0 (December 2006). F9 cells underwent selection with puromycin from 25−48 hrs post-infection and then were passaged onto a 6-well plate. At 72 hrs post-infection, cells were transfected with luciferase vectors and pSV-β-gal and then collected for analysis after 48 hrs. Oct4 knock-down was verified by Western blotting extracts from cells infected in parallel and collected 5 days post-infection using rabbit α-Oct4 [Santa Cruz sc-9081 (1∶500 dilution)] and mouse α-nucleophosmin (NPM) [LabVision ms-1849 (1∶10,000)]. HRP-conjugated human α-rabbit IgG [Sigma-Aldrich A0545 (1∶5,000)] and goat α-mouse IgG [Jackson ImmunoResearch 63656 (1∶8,000)] secondary antibodies. Blots were visualized using Peirce SuperSignal West Pico Chemiluminescent Substrate.

Southern blotting

To assess ICR methylation, genomic DNA was purified as previously described and digested with BglII followed by HhaI or HpaII methyl-sensitive restriction enzymes [9]. MspI, a methylation insensitive isoschizomer of HpaII, was used as a positive control for complete digestion. Digested DNA was separated by agarose gel electrophoresis followed by blotting to Hybond-XL (Amersham Biosciences). Probes were radiolabeled with the Takara Random Primer DNA Labeling Kit and hybridized to membranes in either Rapid-Hybe Buffer (GE Healthcare) supplemented with 25 µg/ml salmon sperm (SS) DNA at 65°C for 2 hrs or 3% Poly(acrylic acid) pH8.0, 6× SSPE, 1% SDS, 0.1% BSA, 25 ug/ml SS DNA at 65°C overnight. After hybridization membranes were washed in 2× SSC, 0.1% SDS at room temperature for 2 min, twice in 1× SSC, 0.1% SDS at 65°C for 15 min, and once in 0.1× SSC, 0.1% SDS at 65°C for 30 min.

Electrophoretic mobility shift assays

F9 nuclear extract (NE) was prepared as previously described [33]. Recombinant Oct4 and Sox2 were purified from IPTG induced Rosetta (DE3) pLys competent cells (Novagen) transformed with pET-15b-Oct4 or pET-15b-Sox2 expression vectors and cultured in LB supplemented with ampicillin and chloramphenicol. After lysis with BugBuster Protein Extraction Reagent (Novagen), Oct4 and Sox2 were purified from inclusion bodies by preparative SDS-polyacrylamide gel electrophoresis. Isolation and renaturation of the gel-isolated proteins were performed according to Hager and Burgess [34].

Oligonucleotide probes and competitors used in EMSA are listed in Table S1. For each reaction 25 fmoles of labeled double-stranded probe was incubated for 10 min at room temperature with either F9 NE or recombinant Oct4 and/or Sox2 proteins in a 16 µl reaction buffer containing 10 mM Tris pH8.0, 2 mM MgCl, 100 mM KCl, 1.25 mM DTT, 5% glycerol, 0.0625% BSA, and 19 µg/ml dGdC. Supershifts included an additional 10 min RT incubation with 1 µg of anti-Oct4 (Santa Cruz sc-9081), anti-Sox2 (Santa Cruz sc-20088), or IgG (Santa Cruz sc-2028) as a negative control. Reactions were separated on a 4% 0.25X TBE polyacrylamide gel at room temperature, which was dried before exposing to film. Quantitative analysis of DNA binding data was performed using a Storm 860 Phosphoimager and ImageQuant 5.2 software.

Chromatin immunoprecipitations (ChIP)

Chromatin from F9 and F9 stable transfectants was prepared and immunoprecipitated using the ChIP-IT Express kit (Active Motif). Chromatin was cross-linked at room temperature with either 1% formaldehyde for 10 min or 2 mM disuccinimidyl glutarate (DSG) 45 min followed by formaldehyde for 10 min (Sox2 endogenous ICR precipitations only) and sonicated using a Vibra Cell (VC 375) ultrasonic processor. Each ChIP reaction contained 25 µg of sheared chromatin precipitated with 6ug anti-Oct4 (Santa Cruz sc-9081), 6 µg anti-Sox2 (3 µg each of Santa Cruz sc-20088 and sc-17320) or 6 µg IgG (Santa Cruz sc-2028 and sc-2027) as a negative control. Precipitated DNA was analyzed using a two-step qRT-PCR reaction in SsoFast EvaGreen Supermix (Bio Rad) on a Bio Rad iCycler. The primers used to amplify the ICR octamer region are: 5′-ATGCAGACCCCACTAAGCAT-3′ and 5′-CGGAGATCATTAGCATCTGA-3′.

Generation of transgenic mice

ES cell lines and chimeric mice were created by the University of Illinois Transgenic Mouse Facility. CJ7 cells [35] were electroporated with SalI linearized knock-in vector and colonies were selected for G418-resistance. ES clones demonstrating proper targeting by Southern analysis were injected into C57BL/6J (B6) derived blastocysts. Genotyping and confirmation of Cre-mediated deletion of the neo cassette were performed by Southern analysis and PCR of tail DNA (MOPrimer, 5′- ATCATTGGGGCGTTCAGATAATC-3′ and OCR3, 5′-GACAGTGCAAAACAGGTGAA-3′). All experiments involving animal were conducted in accordance with protocols approved by the Animal Care and Use Committee at the University of Illinois at Urbana-Champaign for the purposes of minimizing animal suffering (IACUC Protocol #9250).

Oocyte and blastocyst isolation

Unfertilized oocytes were collected from the oviducts of superovulated mice approximately 18 h after hCG was administered. The oocytes were incubated in hyaluronidase to dissociate the cumulus cell-oocyte complexes and then visually sorted and inspected to eliminate cumulus cell contamination. Expanded blastocysts were collected from the uteri of superovulated females 84-90 h after hCG injection and mating. All blastocysts were sorted and visually inspected to eliminate contaminating cells and debris.

Germ cell isolation

Embryos for PGC isolation were obtained from matings between mutant ICR mice and B6;CBA-Tg (Pou5f1-EGFP) [36]. Embryos were sexed and staged according to morphology [37]. Embryonic gonads were separated from the mesonephros, digested with 150 µl of trypsin-EDTA [0.2% trypsin, 1 mM EDTA, 1× phosphate-buffered saline (PBS)] for 15 min at 37°C, washed with DMEM supplemented with 10% fetal bovine serum and manually triturated to a single cell suspension using a 1 ml syringe and 30G needle. GFP-positive cells were isolated by flow cytometry on a BD Biosciences LSR II. PGC purity was verified by assaying for alkaline phosphatase activity using the Vector Blue Phosphatase Substrate Kit III (Vector Laboratories).

Bisulfite sequencing

Genomic DNA from PGCs, oocytes, and blastocysts was prepared and bisulfite treated using the EZ DNA Methylation-Direct Kit (Zymo Research) according to the manufacturer's instructions. Genomic DNA from neonatal liver and muscle was prepared separately and added directly to the conversion reagent. Primers (Bi3000F and Bi3455R) and PCR conditions for amplification of CTCF site 4 were as previously described [9]. Amplification of CTCF site 2 and the octamer region were performed using the following primer set: Bi1825F, 5′-TTGTAAAGAATTTTTTGTGTGTAAAG-3′ and Bi2310R, 5′-ATACAATTTCAAAATTATTTACAACCC-3′. PCR was performed for 35−40 cycles under the following conditions: 95°C, 30 s; 52°C 40 s; 72°C, 100 s. For oocytes, a secondary PCR with 1−5 µl of the first reaction was performed for 35 cycles with the same parameters. PCR products were gel-purified and cloned following the manufacture's protocol (pGEM-T, Promega). Sequence analysis was performed using BISMA software [38].

RNA isolation and SNuPE analysis

Tissue was dissected from WT and mutant 4 dpn pups obtained from mating female mice heterozygous for the mutant ICR with males that were homozygous for the M. casteneus Igf2/H19 locus. RNA was purified using the RNeasy Mini Kit (Qiagen), and all samples were subjected to on-column DNase digestion (RNase-Free DNase Set, Qiagen). Tissue was homogenized for 3 min in RLT buffer using 0.1 mm Zirconia/Silica beads and a Mini-BeadBeater (Biospecs Products). Igf2 and H19 transcripts were amplified using the SuperScript First Strand Synthesis System (Invitrogen), and the following primer sets: Igf2-168F/U, 5′- ACGTGTCTACCTCTCAGGCCGTACT-3′ and Igf2-166R, 5′-GGGTTGTTTAGAGCCAATCAA-3′ [39]; H19-172F, 5′-GCACTAAGTCGATTGCACTGG-3′and H19-172R, 5′-GCCTCAAGCACACGGCCACA-3′ [40]. PCR was performed for 35 cycles under the following conditions: (Igf2) 95°C, 20 s; 53°C 20 s; 72°C, 30 s; and (H19) 95°C, 20 s; 56°C 20 s; 72°C, 20 s. Control reactions lacking RT did not yield a PCR product. SNuPE was performed as previously described [41] using the following primers: Igf2-169R/S, 5′-TCAAATTTGGTTTTTTAGAA-3′ [39]; H19-173R/S, 5′-GGCAGCATTGCCAAAGAGG-3′ [39]. Quantification of SNuPE reactions following electrophoresis was performed using a Storm 860 Phosphoimager and ImageQuant 5.2 software.