Experimental system and constructs used in this study. (A) Structure of the human β-globin locus with the locations of the two replicators (Rep-P and Rep-I) and the asymmetric purine:pyrimidine (AG) region marked. LCR, locus control region; IR, initiation region. (B) Transgene constructs. LCR, a mini-LCR sequence that includes the DNase I-hypersensitive sites 4, 3, and 2 from the human β-globin locus control region; Pro, human β-globin promoter; EGFP, enhanced green fluorescence protein gene. (C) All transgene constructs flanked by a pair of inverted loxP sites, L1 and 1L, were inserted into a late-replicating RL4 site of the murine chromosome 15 by recombinase-mediated cassette exchange (RMCE). The insertion replaces the hygromycin-thymidine kinase (Hyg-TK) hybrid selection marker. (D) Wild-type and mutant sequences of the AG region of Rep-P used in this study. Nucleotides shown on a white background are mutations. The CG (red) below the sequences indicates the CpG dinucleotide created in the Rep-PAG2 mutation.

MeCP1, SWI/SNF, and hnRNP C1/C2 interactions with Oligo AG. (A) ChIP analysis measuring the binding of hnRNP C1/C2 to the asymmetric Rep-P region in K562 cells and Jurkat cells. Primers and probes include the following: LB2, lamin B2 origin; for the globin IR, see the illustration; bG59.8, Rep-P 5′-end sequence; AG, AG region of Rep-P; and Rep-I, another replicator in the β-globin locus. (B) Binding of the AG region by SWI/SNF and MeCP1 components (Brg1 and Mi2β, respectively) in K562 cells. (C) ChIP analysis for hnRNP C1/C2 binding at AG region and hypersensitive site 2 (HS2) region of LCR in RL4 cells carrying wild-type (Rep-P) or mutant (Rep-PAG2) transgene cassettes. (D) ChIP analysis for LCR-associated remodeling complex (LARC) components binding at the AG region in RL4 cells carrying Rep-P or Rep-PAG2 transgene cassettes. (E) ChIP analysis for LARC components binding at the HS2 region of LCR in RL4 cells carrying Rep-P or Rep-PAG2 transgene cassettes. (Primer and probe sets used in this study are listed in Table 1.)

hnRNP C1/C2 antisilencing activity. (A) The percentages of GFP-positive cells of RL4 cells at 1 month and 3 months after transfection with the indicated transgene cassette. The GFP expression of RL4 cells was analyzed using FACS. Viable cells were gated on the basis of morphological parameters (forward scatter [FSC] and side scatter [SSC]). RL4 cells without a transgene were treated as a negative control. (B) Partial silencing of hnRNP C1/C2 expression in K562 cells. Cells (two independent clones) harboring a doxycycline-inducible shRNA targeting hnRNP C1/C2 were exposed to doxycycline (+ Dox) for 10 days, and the levels of β-globin mRNA were measured. The relative expression levels were normalized according to the control levels (no doxycycline [− Dox]). (Inset) Levels of hnRNP C1/C2 and PCNA in clone 1 with and without doxycycline. This experiment was repeated in 3 independent clones with similar results; error bars were too small and not visible on the histogram. hnRNP, hnRNP C1/C2; PCNA, proliferating cell nuclear antigen. (C) Outcome of zebularine exposure (the results of 1 representative experiment of 3 experiments is shown). RL4 cells containing the indicated constructs were treated continuously with zebularine (Zeb). The GFP expression of RL4 cells carrying no Rep-P or Rep-PAG2 transgene cassette were analyzed using FACS. Viable cells were gated on the basis of morphological parameters (FSC and SSC). The percentage of cells expressing the GFP transgene was estimated using untreated cells as a negative control. Data points presented are day 5 (before GFP expression), day 7 (expressing GFP), and day 26 (GFP expression resilenced after prolonged zebularine treatment). (D) ChIP analysis of the binding of hnRNP C1/C2 to the indicated constructs before and during exposure to zebularine. GFP +, GFP expression; GFP −, GFP expression resilenced.

Effects of zeburaline treatment on Dnmt protein levels, Dnmt1 activity, and CpG methylation. (A) Western blot analysis of Dnmt proteins in RL4 cells that were exposed to zebularine for 3 days (d3), 18 days (d18), or 28 days (d28). RL4 cells carrying an LCR–Pro–EGFP transgene cassette (No Rep-P) and RL4 cells carrying an LCR–Pro–EGFP–Rep-PAG2 transgene cassette (Rep-PAG2) were exposed to zebularine (Zeb). (B) Results of bisulfite sequencing of genomic DNA from RL4 cells exposed to zebularine. Me, methylated.

Model for two different gene silencing mechanisms. (A) Transgene cassette without Rep-P cannot establish LCR–Rep-P interaction. The transgene is silenced by a DNA methylation-dependent maintenance gene silencing mechanism. (B) Transgene cassette with wild-type Rep-P. The LCR-AG interaction mediated by the LARC complex prevents gene silencing. (C and D) In the Rep-PAG2 cassette, GFP expression is silenced (C), but after about 1 week of zebularine treatment (D), GFP expression is activated upon DNA demethylation. (E) Gene expression is resilenced by the primary/de novo silencing mechanism after prolonged DNA demethylation.

Protein-DNA interactions at the asymmetric purine:pyrimidine region of Rep-P. (A) Electrophoretic mobility shift assays and supershift assays with K562 cell nuclear extract and Oligo AG (wild-type oligonucleotide) or Oligo AG2 (mutant oligonucleotide) with their sequences. The protein-binding sites are underlined, and the bold letters show base alterations. The black arrowhead in panel (a) points to the position of the slower-migrating complex. The black arrowhead in panel (b) indicates the supershift caused by anti-hnRNP C1/C2 antibody. The asterisks in panel (a) and (d) indicate the nonspecific band that cannot be competed with the specific competitor. (B) DNase I footprinting indicating protection at the AG2 site. The figure shows the automated DNA sequence analysis of ligation-mediated PCR-amplified products from DNase I-treated permeabilized asynchronous K562 cells. In the histogram, high-intensity peaks represent DNase I-hypersensitive sequences, whereas no peak or low-intensity peaks represent protected regions. Genomic DNA was used as the standard for sequence alignment. The asymmetric AG region is in bold font. As shown in the figure, the AG2 site in the AG region is protected from DNase, indicating protein binding at the AG2 site (marked by the black bar with solid black diamonds on the histogram and the underlined sequence below).

Chromatin immunoprecipitation-chromosome conformation capture (ChIP-3C) experiments demonstrating direct interactions between HS2 and Rep-P in hnRNP C1/C2-containing chromatin. (A) Schematic illustration of the experiment. Fixed K562 cells were lysed and digested with HindIII (H3). Cross-linked chromatin fragments were immunoprecipitated with hnRNP C1/C2 antibody and ligated with T4 ligase. The hnRNP C1/C2-containing protein complexes that bound to both sequences brought remote sequences together. The locations of primers marked on the illustration are as follows: P1, HS2 outer primer; P2, HS2 inner primer; P3, Rep-P outer primer; P4, Rep-P inner primer. All primers are listed in Materials and Methods. (B) An agarose gel showing the nested PCR product of the ChIP-3C sample (black arrow). The PCR product was cloned and sequenced. In the control sample, no antibody was added in the ChIP step. No nested PCR products were generated from the control. K562 genomic DNA was used as a negative control for the experiment. Without HindIII digestion and religation, the nested PCR primers are unable to generate PCR products from the genomic DNA template. The positions of molecular size standards (in base pairs) are shown to the left of the gel.

Replication timing of cells transfected with the LCR–Pro–EGFP–Rep-PAG2 cassette before zebularine treatment (A) and after zebularine treatment (B). We quantified newly replicated transgene (Rep-P) DNA and host sequences (mBG and mAmylase) at different cell cycle phases by real-time PCR. mBG, mouse β-globin locus, an early replicating region in mouse MEL cells; mAmylase, mouse amylase gene, a late replicating region; Rep-P, the Rep-PAG2 region of the inserted transgene cassette.

Analysis of histone H3 and H4 modifications in the transgene cassette. ChIP analysis of acetylated histone H3 (H3Ac) (A), histone H3 with trimethylated lysine 9 (H3K9me3) (B), H3K4me2 (C), H3K27me3 (D), and H4K20me3 (E) in RL4 cells carrying the indicated constructs with or without zebularine treatment. The primers and probes used included mAmy (murine amylase), GFP (the EGFP gene), AG (the AG region of Rep-P), mBG (a murine β-globin sequence), and hBG (human β-globin promoter). GFP, AG, and hBG are sequences from the transgene. Zeb GFP +, cells treated for 1 to 2 weeks with zebularine and exhibiting expression of the EGFP marker; Zeb GFP −, cells treated for 4 to 5 weeks with zebularine and exhibiting resilencing. The results of 1 representative experiment of 3 experiments are shown.