|Home | About | Journals | Submit | Contact Us | Français|
The activation-induced cytidine deaminase enzyme (AID) is required for somatic hyper-mutation and class switch recombination at the immunoglobulin locus1. In GC-B cells, AID is highly expressed, with inherent mutator activity that helps generate antibody diversity2. However, AID may also regulate gene expression epigenetically by directly deaminating 5-methylcytosine (5mC) in concert with base excision repair to exchange cytosine3. This pathway promotes gene demethylation, thereby removing epigenetic memory. For example, AID promotes active demethylation of the genome in primordial germ cells4. However, different studies have suggested either a requirement5 or a lack of function6 for AID promoting pluripotency in somatic nuclei following fusion with embryonic stem cells (ESCs). We tested directly whether AID regulates epigenetic memory, by comparing the relative ability of cells lacking AID to reprogram from a differentiated cell type to an induced pluripotent stem cell (iPSC). We show that AID-null cells are transiently hyper-responsive to the reprogramming process. Although they initiate expression of pluripotency genes, they fail to stabilize the pluripotent state. The genome of AID-null cells remains hyper-methylated in reprogramming cells, and hyper-methylated genes associated with pluripotency fail to be stably up-regulated, including many MYC target genes. Recent studies identified a late step of reprogramming associated with methylation status7, and implicated a secondary set of pluripotency network components8. AID regulates this late step, removing epigenetic memory to stabilize the pluripotent state.
The path to pluripotency involves multiple steps that can be more or less efficient, and this can be modeled in transcription factor (TF)-induced reprogramming9. Reprogrammed iPSC clones initially retain methylation patterns that may reflect an epigenetic memory of the source lineage10,11. To test directly the function of AID in reprogramming, we prepared tail-tip fibroblasts (TTFs) from AID knockout or control sibling mice. The starting fibroblasts appeared morphologically identical (Fig.S1a), and were transduced equivalently with lentivirus encoding the four human “Yamanaka” reprogramming TFs12: OCT4, KLF4, SOX2 and cMYC (OKSM, Fig.S1b). Expression levels 72 hours post-infection of exogenous human OCT4 (prior to induction of the endogenous murine gene) were equivalent, showing that lack of AID does not impact expression vector function (Fig.S1c). Exogenous TF expression was subsequently equivalently repressed in wildtype and AID-null transduced cells (Fig.S1d). Using qRT-PCR assays, transcript levels for AID were not reliably detected above background in wildtype TTFs. However, after one week of reprogramming, transcripts were readily measured in wildtype cells, increasing as much as 10-fold during the reprogramming process (Fig.S2).
Initial reprogramming steps include induction of proliferation and change in the morphology of fibroblasts to smaller and rounder cells13. We discovered that cells lacking AID are initially hyper-responsive to reprogramming factors. Change in morphology was more rapid in AID−/− cells, beginning at 2 days post-transduction. After 4 days, AID−/− cells are rounded and smaller than AID+/+ cells (Fig.1a). A higher fraction of AID−/− cells stain positive for SSEA1, an early marker for pluripotency14 (Fig.1b). At day 7, more AID-null cells express NANOG compared to controls (Fig.1c), correlating with modestly higher transcript levels for several pluripotency genes prior to the first week of reprogramming (Fig.1d). Consistent with the molecular data, the AID-null cells consistently generated at 2 weeks more early colonies than wildtype cells (6-fold more colonies on average over six independent experiments). Thus, AID-null fibroblasts are hyper-responsive to reprogramming, even though growth curves for uninfected wildtype and AID-null cells were indistinguishable (Fig.S3a). Rather, the enhanced expression of pluripotency markers in AID−/− cells suggests that AID normally helps to stabilize the differentiated state, creating a barrier to the initial process of reprogramming. When the AID−/− fibroblasts were passaged prior to transduction, this hyper-responsiveness was no longer seen (Fig.S3b), suggesting that passive removal of epigenetic marks through DNA replication can normalize the initial response to OKSM.
Although there are more of them, no obvious differences were observed at 2 weeks in morphologies of iPSC-like colonies derived from AID−/− cells compared with AID+/+ cells, and both stain positive for the pluripotency marker NANOG (Fig.S4). However, by four weeks the colonies derived from AID−/− fibroblasts are flattened with less defined edges (Fig.2a). We tracked individual iPSC-like colonies and observed many AID−/− colonies that appeared pluripotent at 3 weeks but showed a differentiated morphology at 4 weeks (Fig.2b). At 3 weeks, the AID−/− colonies showed a “patchy” NANOG pattern, and by four weeks most colonies differentiated with few NANOG+ cells (Fig.2c). This highly reproducible phenomenon was never observed for colonies derived from AID+/+ cells, which retained iPSC morphology and NANOG expression through 4 weeks (Fig.2b–c). The colonies derived from AID−/− cells showed a progressive decline in the frequency of cells expressing SSEA1 (Fig.2d) and OCT4 (Fig.2e). The same results were observed using mouse embryonic fibroblasts (MEFs), as the AID-null cells failed to maintain a pluripotent iPSC phenotype (Fig.S5). The stabilization of pluripotency was equally effective in wildtype cells (and ineffective in AID-null cells) regardless of whether constitutively expressed or doxycycline-inducible OSKM cassettes were used (Fig.S6a).
Using either wildtype or AID-null cells, expression of OKSM using a doxycycline-inducible vector for only 6 days was insufficient to generate pluripotent colonies. Expression for 9, 12, 21, or 28 days was sufficient to generate pluripotent colonies, showing a “dose-response” with the wildtype fibroblasts, but was equally ineffective for generating stable pluripotent cells from AID-null cells at the 4 week time-point (Fig.S6b,c). When AID was expressed along with OKSM during initial reprogramming stages, this failed to rescue pluripotency (not shown). However, when OKSM-transduced AID-null cells were secondarily infected with an AID-expressing retrovirus after one week (Fig.S7a,b), there was a partial but significant rescue of pluripotency markers stabilized at the 4 week time-point. Since the rescue was not complete we cannot rule out additional molecular mechanisms working with AID for stabilization of pluripotency. Expression of AID in wildtype cells further enhanced the stable expression of pluripotency markers at 4 weeks (Fig.S7c). Since Nanog is syntenic with the AID locus, an independent set of experiments used fibroblasts derived from an AID knockout strain in which the neoR gene was deleted15. These “clean” knockout cells also failed to derive stable iPSC colonies (not shown).
It was previously reported that iPSCs derived from different cell types retain epigenetic memory of their somatic phenotype10,11. We tested if genome replication through passaging could stabilize a pluripotent phenotype even in the absence of AID. We generated iPSC-like colonies and isolated at three weeks 13 AID+/+ and 12 AID−/− colonies (clones) that appeared pluripotent by morphology. After three passages (p3) all the clones, regardless of genotype, retained iPSC-like morphology and stained positive for OCT4 and NANOG (Fig.3a). However, between p7 and p10, 5 out of 12 AID−/− clones failed to retain pluripotency and differentiated (Fig.3b). All 13 AID+/+ colonies and the other 7 AID−/− clones that were stable beyond p10 retained pluripotency through p50. All iPSC clones that retained morphology beyond p10 formed embryoid bodies equivalently and could differentiate into cells expressing smooth muscle actin (mesoderm), βIII tubulin (ectoderm) and alpha-fetoprotein (endoderm), regardless of AID genotype (Fig.3c). In an independent set of experiments, wildtype and AID-null clones were picked and either maintained in culture as colonies or passaged. The non-passaged wildtype-derived clones maintained a pluripotent morphology, while the AID-null cells differentiated. With passaging, 60% of the AID-null clones were able to maintain a pluripotent morphology (Fig.S8).
Thus, although AID is not essential for reprogramming, an important transition occurs around 3 weeks that is assisted by AID to stabilize the pluripotent phenotype. We hypothesized that the previously described DNA demethylating role of AID might be at least partially responsible for this phenotypic stabilization. We profiled the epigenome by carrying out reduced-representation bisulfite sequencing (RRBS) in AID-null and control cells after 3 weeks of reprogramming. We observed global DNA hyper-methylation in the AID-null cells (Fig.S9a), considering either total CpGs or differentially methylated regions (DMRs). DNA hyper-methylation occurs preferentially near RGYW motifs (Fig.S9b), the characteristic DNA-targeting sites for AID (45% of the hyper-methylated CpGs). Strikingly, as AID targets gene bodies, the hyper-methylated regions are enriched in gene bodies, even though RRBS is biased toward capturing CpG-rich promoter and enhancer regions (Fig.S9c).
Transcript profiles comparing starting fibroblasts from AID+/+ and AID−/− embryos are closely matched (Fig.S10a). In contrast, many genes fail to be up-regulated in the mutant cells during reprogramming (Fig.4a,b;Fig.S10b). The genes that fail to be up-regulated during reprogramming are highly enriched (p<1e-10) in the gene-set displaying hyper-methylation in AID-null cells (Fig.4c). Of note, genes that are induced in wildtype and AID-null cells are also enriched in hyper-methylated genes (p<1e-14), suggesting that many genes overcome the loss of AID during reprogramming. Focused on the hyper-methylated genes, there is a failure to up-regulate in AID-null cells a set of secondary pluripotency genes, including Rex1, Gdf3, Dnmt3L, Cbx7, Zfp296, Dnmt1, Apobec1, and Tet3 (Fig.4d). This expression data correlates well with the RBBS data, which was validated by MassArray bisulfite sequencing (Fig.S11), and is consistent with a failure of this downstream network to be demethylated at late stages of reprogramming. Comparing the gene-set of hyper-methylated under-expressed genes to public ESC ChIP-seq datasets revealed a dramatic enrichment for cMYC target genes in the reprogramming cells lacking AID (Fig.S12). According to qPCR data there was no difference in cMyc transcript levels in wildtype or AID-null cells at any point of the reprogramming process (not shown). Furthermore, enhanced levels of cMYC by retroviral transduction starting at one week of reprogramming failed to stabilize pluripotency in AID-null cells (Fig.S13), suggesting that cMYC access to binding key target pluripotency genes (due to hyper-methylation) rather than cMYC levels per se may be a limiting step for stabilization of the network.
Individual iPSC-like clones were isolated at 2 weeks and expanded to evaluate global methylation patterns for each individual clone. Hyper-methylation patterns in the AID-null clones were highly consistent with data obtained from bulk colony analysis. Based on genome-wide DNA methylation profiles (with the genome binned for 100 kb regions), AID-null or wildtype clones cluster according to genotypes (Fig.S14a). Regions that were hyper-methylated in the AID-null bulk analysis tend to be hyper-methylated in the AID-null clones (Fig.S14b). In fact, 66% of bulk hyper-methylated DMRs were also hyper-methylated in DNA derived from the isolated clones (p<1e-151). Methylation differences found previously for the secondary pluripotency genes were largely validated by MassArray bisulfite sequencing of DNA (Fig.S14c), and most of the hyper-methylated genes were expressed in the AID-null clones at significantly lower levels (Fig.S14d).
Altogether, our results show that active demethylation through AID-dependent cytosine deamination is important for stabilizing a genetic network controlling stem cell phenotype. AID is not essential for reprogramming, since cells lacking AID can form stable pluripotent iPSCs, either through a passive demethylation process facilitated by DNA replication, or perhaps through compensation from related members of the APOBEC family16. A recent report suggested that loss of AID impacts TF-induced reprogramming only by an early acute shRNA-dependent depletion17, but did not report the “iPSC” phenotype beyond three weeks, when we show AID-null cells eventually fail to stabilize pluripotency. Unless passaged, AID-null cells always fail to reprogram. RNA-seq profiles of AID-null cells after 4 weeks of reprogramming cluster with fibroblast samples (not shown), and qPCR analysis validates significant up-regulation of fibroblast-associated genes in the AID-null samples (Fig.S15), consistent with an epigenetic memory for fibroblast fate that fails to be fully removed during reprogramming in the absence of AID. The function of AID for transition to a stable pluripotent phenotype may be relevant during embryogenesis, since AID knockout strains have small litters. We found that ESC lines could be established from AID-null blastocysts, and these appeared by morphology and staining patterns to be normal (Fig.S16). However, the efficiency of ESC line derivation was markedly reduced (17/28 embryos, 60.7%) compared with congenic wildtype embryos (21/24, 87.5%,), p<0.03 by chi-squared analysis.
While there is much promise for the use of iPSCs for disease modeling and cellular therapies18, there remains concern whether iPSC genomes are damaged through the process of reprogramming19–21. AID with natural mutator activity is activated during reprogramming, and mediates demethylation in gene bodies including secondary pluripotency genes that encode proto-oncogenes. Although inefficient, we showed that iPSCs could be generated in the absence of AID, removing epigenetic memory marks through an AID-independent mechanism. Retaining epigenetic memory, in the absence of AID, might be useful for promoting efficiency of differentiation toward parental lineage fate. Furthermore, if AID-independent reprogramming lessens the mutation load, it could provide a safer strategy to generatie iPSCs for cellular therapies.
TTFs were prepared using 0.2% collagenase; MEFs were prepared from E13.5 embryos. Intracellular immuno-staining was performed after fixing the cells with 4% paraformaldehyde. Surface staining was done without fixing the cells. cDNA was prepared using Superscript III and qPCR performed (primers listed in Table ST1) using SYBR-green and a Roche LightCycler 480 II. All data were from at least three independent experiments. Enhanced RRBS was performed as described22 and bisulfite reads aligned to the bisulfite-converted mm9 using Bismark23. We achieved very high coverage by sequencing one full Illumina lane (~200M 51bp reads) per sample. For bulk colonies, the wildtype and AID-null samples (at 3 weeks) were analyzed with RRBS that covered 1.96M and 1.82M CpGs, respectively, with at least 10X coverage (with >1.62M CpGs in common between the two runs). 1.52M and 1.59M CpGs had ≥20X coverage in wildtype and AID−/−, respectively. 1.17M and 1.32M CpGs had ≥50X coverage, respectively. Differentially methylated CpGs were identified using the Fisher exact test with correction for Benjamini-Hochberg multiple testing. DMRs were defined as containing at least 5 differentially methylated CpGs and total methylation difference more than 10%. Paired-end RNA-seq libraries were constructed as described24 and sequenced using an Illumina HiSeq2000. Reads were aligned to mm9 using TopHat and gene expression levels were quantified using Cufflinks, using upper-quartile and GC-content normalizations. 1.5-fold changes with FPKM>5 in at least one condition were used to derive differentially expressed genes. MassArray EpiTyper analysis was performed on selected regions identified as hyper-methylated by RRBS. PCR primers (Table ST2) were designed to probe amplicons using EpiDesigner (http://www.epidesigner.com/). Bisulfite conversion and MassArray analysis were performed as previously described22. Metadata for all RRBS are given in Table ST3. Methylation differences were calculated at each CpG and boxplot analysis performed at each interrogated region. All statistical analyses were performed using the R software package. All RNA-seq and RRBS data is available at GEO, accession GSE46700.
AID-deficient (AID−/−) mice were a gift from T. Honjo. Some of the reprogramming experiments were performed on the AID−/− and AID+/+ littermates derived by crossing AID+/− mice and for some of the experiments wild-type BALB/c mice were purchased from the Jackson Laboratory. Reprogramming experiments were performed on tail fibroblasts obtained from 7 AID−/− and 7 AID+/+ age-matched mice (4 littermates and 3 non-littermates, each). We observed similar results when comparing littermates or the purchased mice. Mice or embryos used for MEFs were genotyped by PCR. All animals were maintained according to the guidelines for animal welfare of the Memorial Sloan-Kettering Research Animal Resource Center.
For preparation of tail fibroblasts either whole tails or tail tips were collected from age matched adult AID+/+ and AID−/− mice (Age range, 3–8 weeks). Tails or tail tips were minced using a razor blade/scalpel after washing with PBS. For preparation of mouse embryonic fibroblasts (MEFs) AID+/− males and AID+/− females were time mated and E13.5 embryos were collected. Embryos were washed several times and extra care was taken to prevent maternal cross contamination. After removing the head and all internal organs embryos were minced using a razor blade/scalpel. Minced tails or embryos were washed twice with DMEM (Cellgro) containing 10% FBS (Gemini Bio Products), 1 mM L-glutamine (Cellgro), 100 U/ml penicillin (Cellgro) and 10 µg/ml streptomycin (Cellgro) and dissociated using 0.2% collagenase from Clostridium histolyticum (Sigma) at 37°C for 4 hours (tails) or 2 hours (embryos) with continuous shaking. Dissociated cells were filtered through a 70 µm cell strainer (BD falcon) and washed two times with DMEM containing 10% FBS and plated on 0.1% gelatin (Cellgro) coated tissue culture plates. Fibroblasts were cultured in DMEM containing 10% FBS, 1 mM L-Glutamine, 100 U/ml penicillin and 100 µg/ml streptomycin and media was replaced daily. After passaging two times, fibroblasts were frozen in media containing 10% DMSO, 40% FBS and 50% DMEM. Generally, passage two (P2) fibroblasts were used for the reprogramming experiments, unless indicated otherwise. Genotype was confirmed by PCR. Primers for wildtype, F: 5’-GGTCCCAGTCTGAGATGTA, R: 5’-CAACGTGGCGTCCAAACAGGC; for knockout, F: 5’-CTGCCAAACCTGATGTCTTGA, R: 5’-AACCAAGCCTATGCCTACAGC.
Reprogramming virus was produced by transfection of a single lentiviral stem cell cassette (STEMCCA) in HEK293T cells using polyethylenimine (PEI) (Polysciences). STEMCCA (constitutive) expresses all four human factors (OCT4, KLF4, SOX2, cMYC), using a single Elongation Factor 1α (EF1α) promoter within a lentiviral vector using a combination of 2A peptide and internal ribosome entry site (IRES) strategies. Both constitutive and doxycycline-inducible STEMCCA vectors were used for the reprogramming studies. For the doxycycline-inducible system, fibroblasts were treated with doxycycline (1 µg/ml) starting one day after transduction for 2 weeks. Lentiviruses were produced using a five-plasmid transfection system with STEMCCA transfected together with four expression vectors encoding the packaging proteins Gag-Pol, Rev, Tat, and the G protein of the vesicular stomatitis virus (VSV). Five vectors were incubated in DMEM with 0.06% PEI for 20 minutes and then transferred to the plates containing 80% confluent HEK293T cells. Transfection was performed in FBS (10%) containing media. After 5 hours media was replaced with fresh media. Viral supernatant was collected after 48 hours and spun at 3000 rpm to remove dead cells and filtered through a sterile syringe filter with 0.45 µm polyethersulfone membrane (VWR international). Ectopic expression of wildtype AID or catalytically mutant AID was achieved using retroviral vectors as described previously25. Retroviruses were produced by co-transfecting the vector with the packaging plasmid pCL-ECO in HEK293T cells. The cMYC overexpression experiments were achieved using retroviral vector pMXs-c-Myc (Addgene, 13375). pMXs-c-Myc was co-transfected with vector encoding VSVg in GP2-293 cells to collect functional virus.
Five hundred thousand fibroblasts were plated in one well of a gelatin coated 6 well tissue culture plate. After six hours, 1 ml of viral supernatant containing 8 µg/ml polybrene (Millipore) was added to the fibroblast containing wells. For every experiment, AID−/− and AID+/+ fibroblasts were infected at the same time with the identical titer of virus, prepared for each experiment from the same batch. Viral supernatant was removed after 12 hours of infection and fresh media containing DMEM, 10% FBS, 1 mM L-glutamine, 100 U/ml penicillin and 10 µg/ml streptomycin was added. After one day media was replaced by mouse embryonic stem cell (MES) media containing DMEM, 20% ES cell compatible FBS (Gemini Bio Products), LIF (2% conditioned medium) and 1.5 × 10−4 M monothioglycerol (MTG; Sigma), 1 mM L-Glutamine, 100 U/ml penicillin and 10 µg/ml streptomycin. Cells were maintained in the same media. In some of the experiments cells were transferred to mitotically inactivated mouse embryonic fibroblast feeders after two days of transduction. Approximately 30,000 to 50,000 cells were transferred to one well of a feeder-containing 6 well plate. Isolated induced pluripotent stem cell (iPSC) colonies were always cultured on feeders. All the cells were kept at 37°C in a humidified environment at 5% CO2.
Immuno-staining was performed on fixed cells (4% PFA in BBS with 1 mM CaCl2, 15 min) washed and blocked for 30 min in BBT-BSA buffer (BBS with 0.5% BSA, 0.1% Triton, and 1 mM CaCl2). Cells with primary antibodies were incubated overnight at 4C at the following dilutions: anti-Nanog (eBiosciences 14-5761, 1:100, 5µg/ml) and Anti-Oct4 (Santa Cruz 5279, 1:100, 2 µg/ml). Cells were washed and blocked in BBT-BSA and then incubated with Alexa-conjugated secondary antibodies (1:500, from Molecular Probes). Vectashield-DAPI was used as mounting medium. Images were acquired using a Zeiss LSM 510-Meta confocal microscope or a Zeiss epifluorescence microscope with AxioVision software. For flow cytometric analysis cells were trypsinized, fixed with 4% PFA for 20 min, blocked for 1 hour and then stained in suspension. SSEA1 (Santa Cruz, 1:100) staining was performed on unfixed cells. Cells were analyzed on a BD-Accuri C6 flow cytometer (BD Biosciences) using CFlow Plus software.
Cells were trypsinized and collected in Trizol reagent (Life technologies). Total RNA extraction was done using the RNeasy Mini Kit (Qiagen). The cDNA was synthesized using the SuperScript III First-Strand Synthesis System (Invitrogen). Quantitative RT-PCR (qPCR) was performed on three experimental replicates using SYBR green Master I (Roche). Data was generated on a LightCycler 480 II (Roche) and analyzed using LightCycler 480 software. qPCR data was calculated based on the mean of three experimental replicates. All quantifications were normalized to an endogenous Gapdh control. The relative quantification value for each target gene was compared to the calibrator for that target gene. All the primers used for qPCR spanned introns (Supplementary Table ST1).
For embryoid body (EB) formation iPSCs were passaged two times on feeder free gelatin coated culture dishes. 5 × 105 iPSCs were plated on a 10 cm low attachment dish. EBs were cultured for 6 days in MES media without LIF, and media was changed every other day. For mesodermal differentiation EBs were replated on gelatin-coated dishes and cultured for another 7 days in the same media. For neural differentiation, EBs were cultured in media containing DMEM F-12 (Cellgro), 0.5% N2 (Gibco) and 0.5% B27 (Gibco) supplements, 1 mM L-glutamine, 1% non essential amino acids (Gibco) and 1.5 × 10−4 M monothioglycerol for another 3 days, after which EBs were seeded onto gelatin coated dishes with the same media, plus 10 µM retinoic acid (Sigma) for another 4 days. For endoderm differentiation, EBs were replated onto gelatin-coated dishes in MES media lacking LIF but containing 0.5% FBS and 50 ng/ml Activin (R&D) for seven days. For the all differentiation cultures, media was changed daily.
AID−/− males and AID−/− females (or congenic wildtype pairs as controls) were time mated and E3.5 blastocyst staged embryos were collected in M2 media according to standard protocols26. Embryos were placed on mitotically inactivated mouse embryo fibroblasts (MEFs27) in knockout DMEM (Gibco) containing 15% KSR (knockout serum replacement, Gibco), 2 mM L-Glutamine, 0.1 mM β-mercaptoethanol, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 100 U/ml penicillin and 100 µg/ml streptomycin, 1000 U/ml LIF, 1 µM PD0325901 (ERK inhibitor) and 3 µM CHIR99021 (inhibitor of Gsk3b). After embryos attached, media was replaced every other day. After 10 – 11 days an outgrowth could be observed and was dissociated in 0.25% trypsin/EDTA using a mouth-controlled drawn glass pipette. Trypsin was inactivated by adding knockout DMEM (Gibco) containing 15% ES cell compatible FBS, 2 mM L-Glutamine, 0.1 mM β-mercaptoethanol, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 100 U/ml penicillin and 100 µg/ml streptomycin, 1000 U/ml LIF, 1 µM PD0325901 and 3 µM CHIR99021. From this point onwards the culture was maintained in the same media on mitotically inactivated MEFs. Media was replaced every other day. After 3 – 5 days ESC-like colonies were observed and passaged at 70% confluency. Stable ESC lines were validated by passaging a minimum of 10 times without loss of pluripotent morphology or marker expression.
RRBS was performed as described22 and bisulfite reads were aligned to the bisulfite-converted mm9 using Bismark28. Differentially methylated CpGs were identified using the Fisher exact test with correction for Benjamini-Hochberg multiple testing. We defined differentially methylated regions (DMRs) as regions containing at least 5 differentially methylated CpGs, where contiguous differentially methylated CpGs are separated by 250 bp or less, and for which the total methylation change between wildtype and AID-null cells is 10% or more (calculated using all CpGs within the considered region including those that were not called as differentially methylated). The distribution of DMR lengths was as follows: For hypermethylated DMRs, average = 222 bp, median = 190 bp, minimum = 10 bp, maximum = 4081 bp. For the lower number of hypomethylated DMRs, we observed the following statistics: average = 229 bp, median = 188 bp, minimum = 19 bp, maximum = 1250 bp. Paired-end RNA-seq libraries were constructed as previously described29 and sequenced using an Illumina HiSeq2000. Reads were aligned to mm9 using TopHat and gene expression levels were quantified using Cufflinks, using upper-quartile and GC-content normalizations. 2-fold changes with FPKM>5 in at least one condition were used to derive differentially expressed genes. All statistical analyses were performed using the R software package. The meta-data for all RBBS data is provided in Supplemental Table ST3.
The level of DNA methylation for specific genes was measured using a MALDI-TOF mass spectrometry based method (Epityper, Sequenom, San Diego, CA) as previously described30. Briefly 1 µg of DNA was treated with sodium bisulfite using the EZ methylation kit (Zymo-Research, Irvine, CA). The treatment converts non-methylated cytosines into uracil, leaving methylated cytosines unchanged. PCR amplification, addition of SAP solution and Transcription/RNase A cocktails were performed according to the protocol provided by Sequenom and the mass spectra were quantified by the EpiTYPER analyzer. Amplicons probed are given in Supplemental Table ST2.
We thank Gustavo Mostoslavsky for the gift of reprogramming vectors. Ari Melnick provided advice and consultation. We are grateful to Tasuku Honjo for knockout AID mice, and Nina Papavasiliou for assistance in designing genotyping PCR primers. The Epigenomics Core Facility of Weill Cornell Medical College carried out the bisulfite sequencing and provided consultation. This study was supported by NIH grant HL056182 (TE), and NSF CAREER grant 1054964 (OE).
Author ContributionsRK conceived the study, carried out experiments, and wrote the manuscript. LD, NS, T-CL, PF, and SM-D carried out experiments. AAZ and A-KH provided essential reagents and expertise. JC conceived the study and wrote the manuscript. OE conceived the study, carried out computational and informatics analyses, and wrote the manuscript. TE conceived the study and wrote the manuscript.
The authors have no competing interests.