Mouse embryonic stem cells (mESC) lines used in this study included Dnmt1/3a/3b triple knockout  on a J1 (129S4/SvJae) background, as well as WT J1 and V6.5 (C57BL/6 × 129S4/SvJae) lines. Cells were cultured in Knockout-DMEM (Gibco) with 15% ES qualified FBS (Gibco), ESGRO mouse LIF 1000U/mL (Millipore), 1X PenStrep (Invitrogen), 1X Glutamax (Invitrogen), 55 µM b-mercaptoethanol (Invitrogen), 1X non-essential amino acids (Invitrogen) and 1X Primocin (Invivogen). Cells were maintained at 37 °C and 5% CO2 on dishes coated with 0.1% gelatin (Millipore) unless otherwise stated. Experiments in which mESCs were plated on a layer of mitomycin C-treated mouse embryonic fibroblasts (MEFs) (ThermoFisher Scientific) are explicitly stated.
Determining GSK-3484862 toxicity
An assay to determine the optimal concentration and toxicity of GSK-3484862 (ChemieTek) was performed using J1 WT and DNMT TKO mESCs. 30,000 cells were seeded in 24-well plates pre-coated with 0.1% gelatin. The next day, medium was changed to fresh mESC medium or medium containing DMSO (0.1% or 1%) for the following concentrations of GSK-3484862: 2 pM, 20 pM, 200 pM, 2 nM, 20 nM, 200 nM, 2 µM, 20 µM (in 0.1% DMSO) and 200 µM (in 1% DMSO). The medium was refreshed every day for the next six days, after which cell morphology was assessed, followed by cell dissociation with 0.05% Trypsin–EDTA (Gibco) for cell counting.
Next, the demethylation efficacy and long-term cytotoxicity of GSK-3484862 from two companies, ChemieTek and MedChemExpress, was evaluated in duplicate experiments. To improve solubility, after resuspension in DMSO, GSK-3484862 was subjected to ultrasonication. GSK-3484862 from both companies were sonicated for 6 min at 42 kHz in an ultrasonic water bath (Sper Scientific). Still, drug precipitation was observed for concentrations at or above 20 µM in media, and therefore an upper concentration of 10 µM was chosen. WT and DNMT TKO cells were seeded in 12-well plates pre-coated with 0.1% gelatin and had 0.1% DMSO, 2 µM or 10 µM GSK3484862 added in medium from day zero. The medium was refreshed every day and cells were counted using the Countess II FL instrument and passaged every 2–3 days for the next 14 days.
We found that WT mESCs died if treated with 5-azacytidine immediately after plating (data not shown). Therefore, we plated cells at low density (1.4 × 104 cells/ cm2) and added 0.1% DMSO, 0.1 µM or 0.3 µM of the drug 24 h later. WT and DNMT TKO cells were exposed to 5-azacytidine for 48 h, with a media change at the 24 h timepoint to replenish 5-azacytidine. The cultures were then maintained in media without the drug for 46 h. Cells were then dissociated with 0.05% Trypsin–EDTA, counted, and collected for subsequent analysis. Because of their higher survival and consequent higher density, DNMT TKO cells were harvested after only 24 h recovery time.
Concentrations of decitabine as low as 0.1 µM proved lethal to WT mESCs plated on gelatin (data not shown), so mESCs were seeded at 1.4 × 104 cells/ cm2 on a monolayer of MEF feeder cells to enhance survival. 24 h later, 0.1% DMSO or decitabine at concentrations of 0.1 µM, 0.3 µM or 1 µM was added and cells were treated for 48 h, with a media change at the 24 h timepoint. Cells were allowed to recover for another 24 h and then harvested. Because of the low viability of the WT J1 line mESCs, the experiment was repeated with V6.5 mESCs, a robust line on a hybrid genetic background (C57BL/6 × 129S4/SvJae).
Cell imaging and counting
All microscopy images were acquired with an EVOS M5000 (Invitrogen) and cell numbers were counted using Countess II FL Automated Cell Counter (Invitrogen). Images were processed using Adobe Photoshop and ImageJ, and graphs were created using GraphPad Prism software.
DNA and RNA extraction
Cell pellets were snap frozen and kept at –80 °C until extraction. RNA and DNA from GSK-3484862 and 5-azacytidine treated samples were isolated simultaneously using the AllPrep DNA/RNA kit (Qiagen), whereas only RNA was isolated from decitabine treated samples using the RNAzol-RT total RNA protocol (Sigma). DNA and RNA concentrations were measured with the Qubit dsDNA and RNA HS Assay Kits, respectively (ThermoFisher).
Western blot analysis of DNMT1 protein levels in cells treated with GSK inhibitor
To validate the specificity of the DNMT1 antibody and DNMT1 deletion in the DNMT TKO, both J1 WT and DNMT TKO cells were cultured in a 6-well plate pre-coated with 0.1% gelatin. Medium was refreshed every day until cells were 70–80% confluent. Cells were then dissociated with 0.05% Trypsin–EDTA and pellets were snap frozen for protein extraction.
To determine the effect of GSK-3484862 on DNMT1 protein levels, J1 WT cells were seeded in 12-well plates pre-coated with 0.1% gelatin and treated with medium supplemented with 0.1% DMSO, 2 µM or 10 µM of GSK-3484862 (ChemieTek). The medium was refreshed every day for the next four days, with cells being passaged two days post-plating. After four days of treatment, cells were dissociated with 0.05% Trypsin–EDTA and pellets were snap frozen and stored at -80 °C for subsequent protein extraction.
Protein was extracted in ice cold RIPA lysis buffer, supplemented with protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 10mM sodium fluoride and 1 mM sodium orthovanadate). Cell samples were then exposed to 5 cycles of freeze–thaw using liquid nitrogen to ensure complete lysis. Protein lysate concentrations were measured using a Bradford Assay and 30 μg of protein were subjected to a 6%-12% gradient SDS-PAGE. The resolved proteins were transferred to a polyvinylidene fluoride (PVDF) membrane and blocked using 5 mL of LI-COR Odyssey Blocking Buffer for 1 h at room temperature. Primary antibodies (DNMT1 (Santa-Cruz H-300) 1:1000 dilution and Histone H3 (Abcam ab10799) 1:10,000 dilution) were diluted in Odyssey Blocking Buffer supplemented with 0.15% Tween-20 and incubated overnight at 4 °C. Membranes were then washed three times for 5 min each in PBS supplemented with 0.1% Tween-20 and incubated in secondary antibodies (LI-COR IRDye 680RD and LI-COR IRDye 800CW, 1:20,000 dilution) for 1 h at room temperature. Membranes were then washed three times for 5 min each in PBS 0.1% Tween and left in PBS before imaging on the LI-COR imaging system. Fiji was used to quantify fluorescent band intensities of DNMT1, which were normalized to band intensities of Histone H3.
500 ng of RNA was used for cDNA synthesis using Froggabio SensiFAST cDNA synthesis kit and following the manufacturer's instructions. The qPCR reaction was performed using PowerUp SYBR green mix (Invitrogen) containing cDNA generated from the equivalent of 5 ng RNA and 0.5 μM of primer mix (forward and reverse) in a final reaction of 6 µL per duplicate. Quantification and analysis were performed with the QuantStudio5 instrument (Applied Biosystems), and the cycling conditions were: (50 °C 2 min, 95 °C 20 s, 55x (95 °C 3 s, 60 °C 30 s), 95 °C 1 s). The following primers were used:
Actin: F:5′-ACTGGGACGACATGGAGAAG-3 ′ R:5′-GGGGTGTTGAAGGTCTCAAA-3′, Gapdh: F:5 ′-CATCAAGAAGGTGGTGAAGC-3′ R:5′-GGGAGTTGCTGTTGTAAGTCG 3′, Tuba3b: F:5′-AGGAAGATGCAGCCAACAATTA-3′ R:5′-TGCACAGATCGGCCAGTTT-3′, Ctcfl: F:5′-GCCTTCAGCATTGCGTGAC-3′ R:5′-AGCAGGTGAAAATGTATCCGC-3′, Magea4: F:5′-GTCTCTGGCATTGGCATGATAG-3′ R:5′-GCTTACTCTGAACATCAGTCAGC-3′, Rhox1: F:5′-CCGGTTTTCTGGAGTATGAGAGA-3′ R:5′-CCAGCCGTTTTCTGTCTTGTG-3′, Hormad1: F:5′-TGAAAACTCTGGAGCTTCTGAAA-3′ R:5′-ACTGACTAACTGTTCAACCTGACA-3′, Sohlh2: F:5′-CCATCGAGCTGTTCCTTCCA-3′ R:5′-GGAATACACGTTCAGGCCCC-3′, Dazl: F:5′-GTGGCTTCTGCTCCACCTTCG-3′ R:5′- CCTTGACTTGTGGTTGCTGA-3′, Gln: F:5′-CGTAAGGACCCTAGTGGCTG-3′ R:5′- GCACTCACTCTTCTTCACTCTG-3′, IAP-Ez: F:5′-AAGCAGCAATCACCCACTTTGG-3’ R:5’- CAATCATTAGATGYGGCTGCCAAG-3’. The Dazl, Gln and IAP-Ez primers were taken from published sources .
Analysis of qRT-PCR data
To calculate which genes showed statistically significant upregulation relative to control cells, expression of each gene in the DNMT TKO or GSK-3484862 cells was normalized to the control or control lines in the same experimental replicate and time-course. Data from cells treated with GSK-3484862 from the two manufacturers (ChemieTek and MedChemExpress) were combined. A paired, one-tailed t-test was conducted, pairing control cells with treated cells for each time point and calculating one p-value for each GSK-3484862 concentration across all timepoints.
Whole‑genome bisulfite library preparation
Genomic DNA (500 ng) with spiked-in lambda DNA (1.25 ng) (NEB) was sheared using an M220 ultrasound sonicator (Covaris) to an average size of 350 bp. The size of the DNA fragments was confirmed by electrophoresis on a 1.5% agarose gel. 200 ng of sheared DNA was subjected to bisulfite conversion using EZ DNA Methylation-Gold Kit (Zymo Research) as per manufacturer’s instructions. Libraries were prepared using 50–100 ng of bisulfite-converted DNA and the Accel-NGS Methyl-Seq DNA Library prep kit (Swift Biosciences) according to manufacturer's instructions, with 8–11 cycles of PCR-amplification. The integrity of the libraries was assessed using agarose gel electrophoresis.
Libraries were sequenced on an Illumina HiSeq and NovaSeq 6000 instruments at the Michael Smith Genome Sciences Centre at BC Cancer and the Centre for Applied Genomics (SickKids), respectively.
Whole-genome bisulfite sequencing analysis
Raw reads were quality checked with FastQC and adapters at either end were trimmed with the Trim Galore! (v0.6.6) software. Reads were then aligned to the mm10 reference genome using BSMAPz (v1.1.3). In cases where a given read had multiple equally possible alignments, the read was aligned to at most 500 regions at random. The level of DNA methylation over a given cytosine was calculated using the methratio.py script, with the -r flag. Overall level of DNA methylation for a given condition was calculated as the arithmetic mean of all cytosine bases in CpG context. Methylation calls of all cytosines in CpG context were used to generate bigwig tracks. Briefly, the methylation call files from the methratio.py script were converted to bedGraph file format, with the last column reporting the level of methylation over that cytosine as a percentage; in cases where no methylation was observed for a given cytosine, the value was increased by 1% so that it would appear visible on IGV. Finally, the bedGraphToBigWig command was used to create bigwig tracks which were visualized on IGV.
Scatterplot and violin plot analysis
The mm10 reference genome was split into 10 kb non-overlapping bins and the average CpG methylation over each bin was calculated using bedops (v2.4.39). Pearson correlation coefficient between two conditions was calculated using the cor.test function in the ggpubr package. Annotation of transposable elements to the mm10 genome was obtained from a gtf file (http://labshare.cshl.edu/shares/mhammelllab/www-data/TEtranscripts/TE_GTF/—accessed 15th November 2021) curated from RepeatMasker. For analysis over IAPEz-int elements, we only considered full length transcripts which we defined as being at least 6 kb in length. CpG methylation over repeat elements is reported as the mean methylation of all CpGs over those regions.