Developmental Cell Short Article

Loss of Tet Enzymes Compromises Proper Differentiation of Embryonic Stem Cells

Meelad M. Dawlaty,1 Achim Breiling,2 Thuc Le,3,4,5,6,7 M. Inmaculada Barrasa,1 Gu¨ nter Raddatz,2 Qing Gao,1 Benjamin E. Powell,1 Albert W. Cheng,1,8 Kym F. Faull,3,4,5 Frank Lyko,2 and Rudolf Jaenisch1,9,* 1Whitehead Institute for Biomedical Research, Cambridge, MA 02142, USA 2Division of , DKFZ-ZMBH Alliance, German Cancer Research Center, 69120 Heidelberg, Germany 3Pasarow Mass Spectrometry Laboratory 4Department of Psychiatry and Biobehavioral Sciences 5Semel Institute for Neuroscience and Human Behavior David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA 90095, USA 6Department of Molecular and Medical Pharmacology 7Ahmanson Translational Imaging Division University of California, Los Angeles, Los Angeles, CA 90095, USA 8Computational and Systems Biology Program 9Department of Biology Massachusetts Institute of Technology, Cambridge, MA 02142, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.devcel.2014.03.003

SUMMARY maintenance of DNA mediated by DNA methyltrans- ferases are achieved, the mechanisms for removal of this modi- Tet enzymes (Tet1/2/3) convert 5-methylcytosine fication are poorly understood (Wu and Zhang, 2010, 2014). The (5mC) to 5-hydroxymethylcytosine (5hmC) and Ten eleven translocation (Tet) family of enzymes (Tet1/2/3) has are dynamically expressed during development. been implicated in DNA (Tahiliani et al., 2009; Whereas loss of individual Tet enzymes or combined Ito et al., 2010). These enzymes contain a C-terminal catalytic deficiency of Tet1/2 allows for embryogenesis, the domain that has dioxygenase activity and converts 5-methylcy- effect of complete loss of Tet activity and 5hmC tosine (5mC) sequentially to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) (He marks in development is not established. We have et al., 2011; Ito et al., 2011). Given that DNMT1 recognizes generated Tet1/2/3 triple-knockout (TKO) mouse 5hmC poorly, it is believed that this modification promotes pas- embryonic stem cells (ESCs) and examined their sive DNA demethylation (Wu and Zhang, 2010). However, it has developmental potential. Combined deficiency of all also been proposed that these modified bases function as inter- three Tets depleted 5hmC and impaired ESC differ- mediates in the process of active DNA demethylation, where entiation, as seen in poorly differentiated TKO they are enzymatically recognized and removed from the embryoid bodies (EBs) and teratomas. Consistent by components of the DNA repair machinery. Because with impaired differentiation, TKO ESCs contributed 5hmC has also been found as a stable base in the genome that poorly to chimeric embryos, a defect rescued by reaches significant abundance in several cell types, it may also Tet1 reexpression, and could not support embryonic serve as a distinct epigenetic mark and not just as a DNA deme- development. Global gene-expression and methyl- thylation intermediate (Wu and Zhang, 2010, 2014). Genomic studies have identified 5hmC as highly abundant in CpG-rich ome analyses of TKO EBs revealed promoter hyper- promoters and gene bodies in mouse embryonic stem cells methylation and deregulation of genes implicated in (ESCs) (Pastor et al., 2011; Williams et al., 2011) and in enhancer and differentiation. These elements in human ESCs (Stroud et al., 2011), where it may have findings suggest a requirement for Tet- and 5hmC- critical regulatory roles in orchestrating . mediated DNA demethylation in proper regulation Tet enzymes and 5hmC are present in various embryonic and of gene expression during ESC differentiation and adult cell types, including the zygote (Wossidlo et al., 2011), em- development. bryonic stem cells (Ito et al., 2010; Koh et al., 2011), primordial germ cells (Hajkova et al., 2010), and (Kriaucionis and Heintz, 2009). Despite their roles in epigenetic regulation of the INTRODUCTION genome, notably DNA demethylation, and their presence in various stages of embryonic and adult development, the physio- DNA methylation is a prominent epigenetic modification in the logical relevance of these enzymes has not been well estab- eukaryotic genome, and dynamic changes in the DNA methyl- lished. The recent generation of Tet1, Tet2, and Tet3 single ation landscape are essential for normal regulation of genes dur- mutant (Dawlaty et al., 2011; Gu et al., 2011; Li et al., 2011; ing development (Smith et al., 2012; Gifford et al., 2013; Xie et al., Moran-Crusio et al., 2011) and Tet1/2 double-knockout (DKO) 2013). Although it is well defined how the establishment and mice and ESCs (Dawlaty et al., 2013) has shed light on the roles

102 Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. Developmental Cell Tet Enzymes and 5hmC in Differentiation

of these proteins in pluripotency as well as embryonic and adult ure 1A) in ESCs and in differentiated cell types such as EBs or development. Both Tet1 and Tet2 are expressed in mouse ESCs, retinoic-acid-treated ESCs. To investigate how loss of Tet en- and depletion of either of these proteins reduces 5hmC levels but zymes affects global 5hmC levels, we applied mass spectrom- does not affect pluripotency (Dawlaty et al., 2011; Koh et al., etry to measure levels of 5mC and 5hmC in genomic DNA 2011). Tet1 knockout ESCs remain pluripotent and Tet1 isolated from TKO, THet, DKO, and WT EBs (Figure 1B). knockout mice are viable and fertile despite displaying reduced Whereas THet and DKO EBs had 50% and 80% reduction body and litter sizes, suggesting a subtle role for Tet1 in embry- in 5hmC levels, respectively, TKO EBs were completely depleted onic development and gametogenesis (Dawlaty et al., 2011). of 5hmC, suggesting that Tet1, Tet2, and Tet3 collaborate in Like Tet1, Tet2 is also dispensable for embryonic development, establishing and maintaining 5hmC marks in the genome. and adult mice are viable and fertile. However, Tet2 deficiency Concomitant with loss of 5hmC, TKO EBs had a subtle increase promotes chronic myelomonocytic leukemia formation in mice in global 5mC levels, suggesting that depletion of 5hmC levels and humans (Li et al., 2011; Moran-Crusio et al., 2011). In leads to increased global hypermethylation during ESC differen- contrast to Tet1 and Tet2, Tet3 is not expressed in ESCs and tiation, as observed previously (Dawlaty et al., 2013). is only induced upon differentiation, consistent with its presence in various differentiated cell types (Dawlaty et al., 2013). In addi- Combined Loss of All Three Tet Enzymes Compromises tion, Tet3 is expressed in the and zygote, where it par- Differentiation in Embryoid Body Formation and ticipates in hydroxylating the paternal pronucleus and promotes Teratoma Assays DNA demethylation (Gu et al., 2011; Wossidlo et al., 2011). To assess how the combined loss of all three Tet proteins affects Conditional deletion of Tet3 in the leads to delayed deme- differentiation of ESCs, we differentiated TKO and WT or THet thylation of the paternal genome and increased developmental control ESCs to EBs. Although both TKO and control ESCs failure, suggesting an essential role for Tet3 during preimplanta- formed EBs, histologic examination of TKO EBs revealed poorly tion development. Tet3 homozygous mutant mice can develop differentiated tissues with substantially fewer differentiated to term but die at birth (Gu et al., 2011), suggesting that loss of structures compared to control EBs (Figure 1C). Moreover, Tet3 alone does not block differentiation and embryonic devel- TKO EBs expressed reduced levels of mesodermal and endo- opment and is possibly compensated for, at least in part, by dermal markers (Figure 1D). The expression levels of these Tet1 and Tet2. markers also remained low in late-stage day 15 TKO EBs (Fig- Recently, we have shown that combined loss of Tet1 and ure 1E), suggesting that the poor differentiation of TKO ESCs is Tet2 depletes 5hmC levels in ESCs and germ cells but is not due to a delay in differentiation and rather is likely due to compatible with embryonic development and allows develop- restricted developmental potency of ESCs. Consistent with ment of viable and overtly healthy adult mice (Dawlaty et al., these observations, teratomas derived from WT, THet, and 2013). However, a large fraction of double-mutant embryos ex- DKO ESCs contained multiple tissue types of all three germ hibited midgestation defects, perinatal lethality, and compro- layers, whereas TKO teratomas lacked endodermal and mised imprinting and had reduced 5hmC and increased 5mC selected mesodermal structures and did not contain more levels. This prompted us to investigate whether Tet3 compen- advanced ectodermal structures such as pigmented neural sates for loss of Tet1 and Tet2 during development and whether epithelium (Figures 1F and 1G). These findings suggest that DNA hydroxymethylation is critical for proper development. To Tet enzymes are critical for proper differentiation of ESCs. address this, we have generated Tet1/2/3 triple-knockout (TKO) mouse ESCs and examined their differentiation and TKO ESCs Contribute Poorly to Chimeric Embryos and developmental potential. Our results identify Tet-mediated Cannot Support Development epigenetic regulation of developmental genes during differenti- To more stringently assess the differentiation potential of TKO ation of ESCs and establish a critical role of these enzymes in ESCs during embryogenesis, we injected Rosa26-enhanced differentiation and development. green fluorescent protein (EGFP)-targeted TKO and THet control ESCs, which ubiquitously express EGFP (Figures S2A and S2B), RESULTS into blastocysts. After transplantation into foster mothers, em- bryos were dissected at embryonic day (E)13.5, revealing wide- Generation and Characterization of Tet Triple-Knockout spread contribution of THet-R26-EGFP cells to embryos, with Mouse Embryonic Stem Cells nearly 60% being chimeric and most having high and medium To study the effect of combined deficiency of Tet1, Tet2, and GFP signals. In contrast, two independent TKO ESC clones Tet3 on ESC pluripotency and differentiation potential, we gener- (TKO#26-R26-EGFP and TKO#29-R26-EGFP) exhibited very ated Tet TKO ESCs by intercrossing Tet1/2/3 triple-heterozygote poor contribution to developing embryos (Figure 2A), with only (THet) mice and deriving THet and Tet1–/–jTet2–/–jTet3+/– ESC 15% being chimeric and displaying an extremely low GFP lines. The latter cell lines were subsequently targeted to delete signal, mostly in tail, appendages, and dorsolateral areas (Fig- the wild-type (WT) Tet3 allele and obtain Tet TKO ESCs. Geno- ures 2A–2C). We also inspected chimeric embryos at E9.5 for typing by Southern blot confirmed the loss of the WT allele of TKO ESC contribution and observed, similar to more advanced all three Tet genes (see Figure S1A available online). All TKO embryos, a significantly lower incidence of chimeric embryos ESC lines maintained normal ESC morphology, expressed the (35%) as compared to control THet ESCs (92%), with the pluripotency markers Oct4 and Nanog, and could form embryoid majority of TKO chimeric embryos displaying very poor contribu- bodies (EBs) (Figure S1B). Further experiments confirmed deple- tion (Figure 2D). The few embryos that had slightly higher contri- tion of all three Tet transcripts (Figure S1C) and proteins (Fig- butions were growth retarded or morphologically defective.

Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. 103 Developmental Cell Tet Enzymes and 5hmC in Differentiation

A B

C DE

FG

Figure 1. Loss of 5hmC and Restricted Differentiation Potential of Tet TKO ESCs in Embryoid Body and Teratoma Formation Assays (A) Confirmation of loss of Tet enzymes in Tet TKO ESCs and differentiated cells by western blot. Asterisks indicate nonspecific bands. (B) Quantification of 5hmC and 5mC in ESCs and EBs of the indicated genotypes by mass spectrometry. (C) Sections of paraffin-embedded EBs derived from ESC lines of the indicated genotypes stained with H&E. Insets show higher-magnification images. (D and E) Quantitative RT-PCR for markers of embryonic germ layers in day 10 EBs (D) and day 15 EBs (E) of the indicated genotypes. Data are normalized to Gapdh. (F) Representative images of each embryonic germ layer from H&E-stained sections of teratomas derived from ESCs of the indicated genotypes. N.D., none detected. (G) Summary table of the various germ layer tissues detected in teratomas of the indicated genotypes. Error bars indicate standard deviation. See also Figure S1.

104 Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. Developmental Cell Tet Enzymes and 5hmC in Differentiation

A TKO#29 T.Het#2 TKO#29 TKO#26 B TKO R26-EGFP R26-EGFP R26-EGFP R26-EGFP (E13.5) R26-EGFP (E12.5) Bright Field GFP Bright Field Tail and Limbs Tail GFP GFP

C No. of No. of No. of Chimeric Embryos Total No of Blastocysts Embryos Chimeric Cell Lines Tested Very high High Medium Low Very low Extremely weak Injected (E13.5) (>90%) (60-90%) (30-60%) (20-30%) (10-20%) (<5%) Embryos (%) THet#2-R26-EGFP 40 16 1 4 4 0 0 0 9 (57%) TKO#29-R26-EGFP 80 37 0 0 0 0 1 4 5 (14%) TKO#26-R26-EGFP 80 41 0 0 0 0 1 3 4 (10%)

D THet#2-R26-EGFP TKO#26-R26-EGFP TKO#29-R26-EGFP

GFP GFP GFP No. of No. of No. of Chimeric Embryos Total No of Blastocysts Embryos Chimeric Cell Lines Tested Very high High Medium Low Very low Extremely weak Injected (E9.5) (>90%) (60-90%) (30-60%) (20-30%) (10-20%) (<5%) Embryos (%) THet#2-R26-EGFP 40 12 5 3 2 1 0 0 11 (92%) TKO#29-R26-EGFP 40 20 0 0 1* 1 3 3 8 (40%) TKO#26-R26-EGFP 40 24 0 0 1* 1 4 1 7 (29%) E F TKO-R26-EGFP ESC +EV +Tet1 TKO#26 R26-EGFP TKO#26 R26-EGFP Embryos Embryos THet R26-EGFP THet R26-EGFP (Embryo) TKO29 R26-EGFP (Embryo) TKO26 R26-EGFP (ESC)

WT (14kb) Bright Field

Rosa-EGFP (5kb) GFP GFP

Rosa26 Probe (EcoRV Digest) 18% (14/78) of 58% (29/50) of G 4N E9.5 embryos embryos chimeric embryos chimeric THet TKO Transferred 4N Implanted Developed ESCs blastocysts embryos E9.5 embryos

Ctrl-R26-EGFP 80 37 3 GFP GFP TKO-R26-EGFP 180 86 0

Figure 2. Limited Contribution of Tet TKO ESCs to Developing Embryos in a Chimera Assay (A and B) Bright-field and fluorescence images of E13.5 chimeric embryos generated by injecting Rosa26-EGFP-targeted ESCs of the indicated genotypes into WT blastocysts. The very poor and weaker GFP signal in TKO ESC chimeras is highlighted in the images in (B). (C) Summary table for all cell lines tested in the chimera assay. (D) Top: bright-field and fluorescence images of E9.5 chimeric embryos generated by injecting Rosa26-EGFP-targeted ESCs of the indicated genotypes into WT blastocysts. Bottom: summary table for all cell lines tested in this chimera assay. Asterisks indicate growth-retarded and defective embryos. (E) Southern blot confirming the negligible-to-no detection of the TKO-Rosa26-EGFP allele in DNA extracted from E13.5 TKO chimeric embryos. (F) E14.5 chimeric embryos generated by injecting Rosa26-EGFP-targeted ESCs of the indicated genotypes into WT blastocysts. EV, empty vector. (G) Left: bright-field and fluorescence images of E9.5 4N embryos generated by injecting Rosa26-EGFP-targeted ESCs of the indicated genotypes into 4N WT blastocysts. Right: summary table for all cell lines tested in the tetraploid complementation assay. Arrowheads point to poor GFP signal. See also Figure S2.

Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. 105 Developmental Cell Tet Enzymes and 5hmC in Differentiation

A BC

D

E

FG

Figure 3. Aberrant Promoter Hypermethylation and Deregulation of Developmental Genes in Tet TKO Embryoid Bodies (A) Schematic of the MeDIP and gene-expression analyses identifying hypermethylated and downregulated genes during differentiation of TKO ESCs. (B) Scatter plots showing differentially expressed genes in red (upregulated) or green (downregulated) across a panel of TKO EBs when compared to WT EBs. FC, fold change. (C) Gene ontology analysis of differentially expressed genes in TKO EBs. (D) MeDIP-seq profile of a 3 Mb region from mouse chromosome 6 with the Hoxa cluster in the center in two independent TKO and WT EB clones as an example of the general hypermethylation observed in TKO EBs. Enrichments are indicated as normalized read counts. (legend continued on next page)

106 Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. Developmental Cell Tet Enzymes and 5hmC in Differentiation

To exclude the possibility that lack of contribution was due ure 3A). We analyzed mRNA levels from WT and TKO EBs by mi- to silencing of the GFP transgene in the 5hmC-deficient cells, croarray (Figure 3B) and found that the majority (1,072/1,801) of we assayed for the presence of the donor cells by Southern deregulated genes in TKO EBs were downregulated, in contrast blot analysis on whole-embryo DNA using a Rosa26 locus to 729 upregulated genes. Gene ontology analysis revealed that probe. As shown in Figure 2E, no Rosa-EGFP-targeted DNA both up- and downregulated genes in TKO EBs were implicated was detected, confirming the lack or at most very low presence in various developmental processes, including embryonic devel- of TKO cells in chimeric embryos. Consistent with these opment and differentiation (Figure 3C). Nevertheless, the spe- observations, examination of selected organs from postnatal cific enrichment for developmental categories was much more mice derived from embryos injected with TKO-R26-EGFP cells significant (lower p values) in the set of downregulated genes revealed no detectable incorporation of EGFP-positive cells than in the set of upregulated genes. The latter set was associ- (Figure S2C). ated with a broader variety of biological functions, which Both independent clones of TKO ESCs tested in this assay ex- included development but also cell-cycle replication, DNA main- hibited a similar compromised differentiation potential and poor tenance, and proliferation. contribution to chimeras, strongly suggesting that the observed Because loss of Tet enzymes promotes global hypermethyla- phenotypes are due to the engineered Tet enzyme mutations tion in TKO EBs (Figure 1B), we analyzed the methylome of TKO rather than any unlinked mutations. To further support this, we EBs by 5-methylcytosine DNA immunoprecipitation using spe- overexpressed Tet1 in TKO-R26-EGFP ESCs and found that cific antibodies against 5mC followed by massive parallel ectopic expression of Tet1 in TKO ESCs rescued their differenti- sequencing (MeDIP-seq). Consistent with the mass spectrom- ation defects and restored contribution to chimeras. Fifty-eight etry analysis (Figure 1B), we found a significant increase in total percent of embryos injected with TKO-R26-EGFP+Tet1 were 5mC reads across all chromosomes in TKO EBs compared to chimeric, with the majority of the chimeric embryos having a WT EBs. Also, baseline 5mC levels were substantially increased medium-to-high degree of contribution and comparable to the across the genome, as exemplified by a 3 Mb region from mouse chimeric embryos derived from THet-R26-EGFP ESCs. In chromosome 6 containing the Hoxa locus (Figure 3D). To further contrast, only 18% of embryos injected with TKO-R26-EGFP characterize the TKO methylation landscape, we analyzed the cells transduced with an empty vector were chimeric, with all position-wise coverage of 5mC peaks in gene bodies and pro- embryos having a very poor level of contribution (Figure 2F). moters (±1 kb of the start site [TSS]) of deregulated To assess whether the poor contribution of TKO ESCs to a genes in TKO EBs (Figure 3E). We found a weak correlation developing embryo is due to reduced proliferation of these cells, between increased gene body methylation and higher gene we quantified the growth rate of TKO ESCs, which was indistin- expression and a strong enrichment for promoter hypermethyla- guishable from that of control THet ESCs (Figure S2D). This sug- tion among downregulated genes. Thirty-nine percent of genes gests that TKO ESCs are not outcompeted by host inner cell with reduced expression in TKO EBs had higher promoter mass cells due to a proliferation defect. We also tested the ability methylation levels than WT EBs. Gene ontology analyses of TKO ESCs to support embryonic development in a tetraploid revealed that the majority of genes with hypermethylated pro- (4N) complementation assay, where the absolute differentiation moter regions in TKO EBs were implicated in developmental potential of TKO ESCs and their ability to develop into an embryo processes, including embryonic development and cellular can be assessed in the absence of competition of WT host cells. differentiation (Figure 3F). Similarly, downregulated genes with When analyzed at E9.5, the TKO 4N-injected blastocysts dis- hypermethylated promoters were enriched for developmental played only rudimentary structures and failed to support devel- categories (Figure 3G). opment of an embryo proper, in contrast to the three normally To validate the hypermethylation of deregulated develop- developed THet embryos (Figure 2G). These findings further mental regulators during differentiation of TKO EBs, we applied support the notion that Tet deficiency compromises the differen- locus-specific 454 bisulfite sequencing to analyze promoter tiation potential of ESCs and restricts their developmental poten- methylation patterns of representative hypermethylated genes tial during embryogenesis. from our MeDIP analysis (Figure 4A). Analysis of 200–300 bp pro- moter regions of the downregulated and hypermethylated genes Deficiency of Tet Enzymes Leads to Aberrant Promoter Emid2, Mall, Gja5, and Tal1 in DNA from TKO EBs and ESCs Hypermethylation and Deregulation of Developmental confirmed robust hypermethylation of these regions (2- to Genes during Differentiation 5-fold more than controls) during differentiation to EBs but not To determine the effects of loss of Tet enzymes and 5hmC deple- in ESCs (Figures 4B and 4C; Figure S3). Two other genes, tion on DNA methylation and gene expression during differentia- Lhx9 and Fgf20, which are hypermethylated but not expressed tion, we differentiated TKO and WT ESCs to EBs and subjected in EBs, exhibited increased hypermethylation in both EBs and them to gene-expression profiling and methylation analyses (Fig- ESCs (Figure 4C; Figure S3). Our findings indicate that the

(E) Analysis of deregulated genes in TKO EBs correlating their expression to the methylation status of their promoters or gene bodies. The pie charts show the percentage of deregulated genes in TKO EBs that show hypermethylation compared to WT (green), hypomethylation compared to WT (yellow), or do not change (gray) in the gene body or the promoter (±1,000 bp from the TSS). (F) Gene ontology analysis of all genes with hypermethylated promoter regions (normalized average read counts >4 in the region ±1,000 bp from the TSS) in TKO EBs. (G) Gene ontology analysis of all genes downregulated in TKO EBs that also have differentially hypermethylated promoters (TKO versus WT). See also Table S1.

Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. 107 Developmental Cell Tet Enzymes and 5hmC in Differentiation

A

B

C

Figure 4. Gene-Specific Bisulfite Sequencing for Validation of Hypermethylation of Deregulated Developmental Genes in Tet TKO Embryoid Bodies (A) MeDIP-seq profiles of representative developmental genes in TKO EBs. Enrichments are indicated as normalized read counts. Red boxes indicate the genomic region analyzed by gene-specific 454 bisulfite sequencing in (B). The green bar indicates a CpG island (as defined by the UCSC Genome Browser; http:// genome.ucsc.edu). (B) Validation of MeDIP data by gene-specific 454 bisulfite sequencing. Results for a 200–300 bp region (red boxes in A) at the promoters of the indicated genes are shown as heatmaps in which each row represents one sequence read and each column represents an individual CpG site within the analyzed region.

(legend continued on next page)

108 Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. Developmental Cell Tet Enzymes and 5hmC in Differentiation

promoter regions of many developmental genes become hyper- opment (Li et al., 1992; Okano et al., 1999; Sakaue et al., 2010) methylated in Tet-deficient EBs. A subset of these differentia- and propose that both DNA methyltransferases and Tet dioxyge- tion-specific genes is active in WT EBs, with their promoter nases work in concert to establish and maintain the appropriate hypermethylated in TKO EBs, which correlates with reduced pattern of DNA methylation during ESC differentiation and expression and likely causes the observed developmental development. Recent work has shown differential genomic phenotypes. localization of Tet1 and Tet2 in ESCs (Huang et al., 2014). It will be interesting to further investigate the genomic occupancy of DISCUSSION Tet enzymes upon differentiation to establish whether Tet1, Tet2, and Tet3 bind and activate similar or different classes of Defining the biological significance of Tet proteins and 5hmC target genes and can substitute for one another. In addition, marks during development has been a primary focus of recent further studies are needed to define whether demethylation-in- investigations. Here we have established that the loss of all three dependent functions of 5hmC as a stable epigenetic mark are Tet enzymes impairs differentiation of ESCs. This phenotype is also implicated in the epigenetic regulation of development. less severe than that observed with Dnmt1 mutant ESCs, where This should also entail examining how loss of 5hmC impairs the loss of methyltransferase activity and severely decreased localization of its readers or binding partners. As this study pri- 5mC levels block differentiation and contribution to embryoid marily underpins the requirements of Tet enzymes in differentia- bodies in chimera assays (Lei et al., 1996; Panning and Jaenisch, tion of embryonic stem cells during embryogenesis, future work 1996). In contrast to Dnmt1 mutant cells, Tet TKO ESCs can using conditional deletion systems will be critical in defining the form most germ layers in teratoma assays and contribute, exact role of Tet enzymes in regulating the homeostasis of many albeit very poorly, to a developing embryo in chimera assays. adult tissues. This will help to further define physiological impli- This indicates that Tet activity and 5hmC modifications are cations of Tet-mediated epigenetic mechanisms of gene regula- essential for proper differentiation during development. Expres- tion in development and diseases including cancer, where Tet sion profiles and genome-wide methylation maps of WT and enzymes are mutated or downregulated and 5hmC levels are TKO EBs show that the majority of deregulated genes in TKO significantly reduced. EBs are downregulated, show promoter hypermethylation, and include developmental regulators that are repressed by DNA EXPERIMENTAL PROCEDURES methylation in ESCs (Gifford et al., 2013; Xie et al., 2013). A subset of these affected genes are also potential targets of Derivation and Culture of TKO ESCs DNA methylation during early embryonic development, where Tet1 and Tet2 knockout alleles have been described previously (Dawlaty et al., 2013). The Tet3 null allele was generated by deleting exon 4 of Tet3 in V6.5 transition from the blastocyst to epiblast stage of a developing ESCs (M.M.D. and R.J., unpublished data). Tet1/2/3 triple-heterozygote and embryo is associated with modulation of DNA methylation levels Tet1–/–jTet2–/–jTet3+/– mouse embryonic stem cells were derived by inter- for proper orchestration of gene expression (Smith et al., 2012). crossing Tet1/2/3 triple-heterozygote mice. Due to the very low expected These observations suggest that Tet enzymes confer their bio- Mendelian frequency of TKO embryos from this cross, we did not derive –/– –/– +/– logically critical effects during ESC differentiation by regulating TKO ESCs from blastocysts but subjected Tet1 jTet2 jTet3 ESCs to promoter methylation levels of a subset of developmental one round of gene targeting to delete the Tet3 WT allele and generate TKO ESCs. All ESC lines were expanded on feeders using regular ESC media con- regulators and lineage commitment genes and thus enabling taining leukemia inhibitory factor (LIF), and were genotyped by PCR and their activation by differentiation-induced and lineage-specific Southern blot using the protocols and probes previously outlined (Dawlaty demethylation. et al., 2013). For the labeling of ESCs with EGFP, an EGFP-pgk-puro cassette Tet3 is not expressed in ESCs, and therefore the epigenetic was targeted to the Rosa26 locus and positive clones were screened by landscape of TKO ESCs is likely to be similar to that of Tet1/2 Southern blot. DKO ESCs. Our previous work as well as a recent methylation analysis of Tet1/2 DKO ESCs (Hackett et al., 2013) have Chimera and Tetraploid Complementation Assays and Analysis confirmed increased methylation in these cells. Postnatal sur- of Midgestation Embryos To generate chimeric embryos, 10–12 Rosa26-EGFP-labeled ESCs were vival of some Tet1/2 double-mutant mice (Dawlaty et al., 2013) injected into B6D2F1 x B6D2F1 E3.5 WT blastocysts and surgically im- supports the notion that Tet3 activity partially rescues Tet1/2 planted into 2.5 days postcoitum pseudopregnant Swiss Webster female deficiency. However, as we find here, complete loss of Tet mice following standard procedures. E9.5, E13.5, or E14.5 embryos were activity impairs proper promoter demethylation and leads to harvested, dissected, and imaged under a fluorescence dissecting scope silencing of target genes, resulting in reduced developmental and scored for contribution of TKO ESCs to the developing embryo potential of ESCs. based on EGFP signal. The tetraploid complementation assay was per- formed as described (Wernig et al., 2007) and is briefly explained in The experiments described here address a long-standing the Supplemental Experimental Procedures. Animal care was in accordance question in the field regarding the requirements of Tet proteins with institutional guidelines, and was approved by the Committee on Animal and the 5hmC mark in development. Our findings complement Care, Department of Comparative Medicine, Massachusetts Institute of previous studies on the role of DNA methyltransferases in devel- Technology.

Individual blue boxes indicate methylated and yellow boxes indicate unmethylated CpG dinucleotides. Panels below the heatmaps show the average methylation of each interrogated CpG for the analyzed DNA fragment. For a color bar, see (C). Sequencing coverage (reads) are indicated. (C) Heatmaps for four additional promoter regions of the genes analyzed. Panels show the average methylation of each interrogated CpG according to the color bar shown on the right. Numbers indicate the sequencing coverage. See also Figure S3.

Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. 109 Developmental Cell Tet Enzymes and 5hmC in Differentiation

Embryoid Body Formation and Teratoma Assays SUPPLEMENTAL INFORMATION Embryoid bodies were formed from ESCs following standard hanging- drop methods as explained previously (Dawlaty et al., 2011). EBs were main- Supplemental Information includes three figures, Supplemental Experimental tained in fetal bovine serum media –LIF for 10 days (or 15 days) before use Procedures, and one table and can be found with this article online at http:// for analysis. Teratoma assays and histological analyses of tissues and EBs dx.doi.org/10.1016/j.devcel.2014.03.003. were performed exactly as explained before (Dawlaty et al., 2013). Hematox- ylin and eosin (H&E)-stained sections of teratomas were examined and ACKNOWLEDGMENTS scored for the presence or absence of tissue types from the three germ layers. We thank Kibibi Ganz, Dongdong Fu, Ruth Flannery, Linyu Shi, Hui Yang, Ping Xu, and Raaji Alagappan for help with animal husbandry, histology, and blas- Quantification of 5hmC and 5mC tocyst injections. We are also grateful to Tayyeb Adil, John Cassady, Thorold Combined liquid chromatography-tandem mass spectrometry with multiple Theunissen, and Jianlong Wang for plasmids, Alex Yoon for help with mass reaction monitoring (LC-MS/MS-MRM) was applied to quantify 5hmC and spectrometry, Tanja Musch for 454 sequencing, and the DKFZ Genomics 5mC levels in DNA extracted from ESCs or EBs as described previously (Le and Proteomics Core Facility for Illumina sequencing services. M.M.D. is a et al., 2011; Dawlaty et al., 2013). Damon Runyon Postdoctoral Fellow. B.E.P. is supported by a PhD fellowship from the Boehringer Ingelheim Fonds. A.W.C. is supported by a Croucher scholarship. T.L. is supported by a UCLA Molecular, Cellular and Neurobiology Microarray Analysis training grant, UCLA Mental Retardation training grant, and Eugene V. Cota- RNA was extracted from day 10 embryoid bodies using a QIAGEN RNeasy Kit, Robles fellowship. Work in F.L.’s lab was supported by a grant from the labeled with Cy3, and hybridized to Agilent arrays. Two technical replicas of WT Deutsche Forschungsgemeinschaft (SPP 1463). R.J. is funded by NIH grants EBs (two samples in total) and two technical replicas of two independent TKO 5-R01-HDO45022 and 5-R37-CA084198 and the Simons Foundation. R.J. is EBs (four TKO samples in total) were hybridized to a Mouse GE 8x60K an advisor to Stemgent and cofounder of Fate Therapeutics. Microarray. Arrays were normalized by quantile normalization. All probes for the same gene were summarized by mean before assaying for differential Received: November 25, 2013 expression by a moderated t test corrected for false discovery rate, as imple- Revised: February 23, 2014 mented by the limma package in Bioconductor (http://www.bioconductor.org). Accepted: March 7, 2014 The cutoff used to select differentially expressed genes was fold change >2 Published: April 14, 2014 and adjusted p value <0.01. Gene ontology analyses were performed using In- genuity Pathway Analysis software (http://www.ingenuity.com/products/ipa). REFERENCES

454 DNA Bisulfite Sequencing Bocker, M.T., Tuorto, F., Raddatz, G.U.N., Musch, T., Yang, F.-C., Xu, M., Deep DNA bisulfite sequencing was performed as described previously (Gro¨ n- Lyko, F., and Breiling, A. (2012). of 5-methylcytosine by niger et al., 2010). Genomic DNA from WT and TKO ESCs and WT and TKO TET2 maintains the active state of the mammalian HOXA cluster. Nat. embryoid bodies was bisulfite treated using the EpiTect Bisulfite Kit (QIAGEN) Commun. 3, 818. according to the manufacturer’s instructions. Treated DNA was amplified with sequence-specific primers (described in Supplemental Experimental Proce- Dawlaty, M.M., Ganz, K., Powell, B.E., Hu, Y.-C., Markoulaki, S., Cheng, A.W., dures) containing cell-type-specific barcodes and standard 454 linker se- Gao, Q., Kim, J., Choi, S.-W., Page, D.C., and Jaenisch, R. (2011). Tet1 is quences. PCR products of 200–300 bp were gel extracted using the peqGOLD dispensable for maintaining pluripotency and its loss is compatible with em- Extraction Kit (Peqlab). For sequencing, equimolar amounts of all amplicons bryonic and postnatal development. Cell Stem Cell 9, 166–175. were combined in a single tube and processed on a GS Junior sequencer Dawlaty, M.M., Breiling, A., Le, T., Raddatz, G., Barrasa, M.I., Cheng, A.W., (Roche) according to the manufacturer’s instructions. Sequence reads were Gao, Q., Powell, B.E., Li, Z., Xu, M., et al. (2013). Combined deficiency of aligned and are displayed as color-coded heatmaps. Tet1 and Tet2 causes epigenetic abnormalities but is compatible with post- natal development. Dev. Cell 24, 310–323. MeDIP-Seq, Mapping of Sequencing Data, and Calculation of Gifford, C.A., Ziller, M.J., Gu, H., Trapnell, C., Donaghey, J., Tsankov, A., Methylation Levels Shalek, A.K., Kelley, D.R., Shishkin, A.A., Issner, R., et al. (2013). MeDIP-seq was performed as described (Bocker et al., 2012) but significantly Transcriptional and epigenetic dynamics during specification of human em- adapted and scaled down to smaller starting amounts according to the proto- bryonic stem cells. Cell 153, 1149–1163. col by Taiwo et al. (2012). Two micrograms of genomic DNA isolated from day Gro¨ nniger, E., Weber, B., Heil, O., Peters, N., Sta¨ b, F., Wenck, H., Korn, B., 10 EBs was sonicated to 300 bp fragments using a Covaris S220 ultrasonica- Winnefeld, M., and Lyko, F. (2010). Aging and chronic sun exposure cause tor. One microgram of purified adaptor-ligated DNA was used for pooled distinct epigenetic changes in human skin. PLoS Genet. 6, e1000971. immunoprecipitations using a polyclonal 5mC-specific antibody (Active Motif; Gu, T.-P., Guo, F., Yang, H., Wu, H.-P., Xu, G.-F., Liu, W., Xie, Z.-G., Shi, L., He, 39791). Details of MeDIP, library preparation, and data analysis are outlined in X., Jin, S.-G., et al. (2011). The role of Tet3 DNA dioxygenase in epigenetic re- Supplemental Experimental Procedures. Briefly, in order to quantitatively programming by oocytes. Nature 477, 606–610. analyze sequencing reads, position-wise read counts were normalized according to the 5mC levels of each sample DNA, as determined by mass Hackett, J.A., Dietmann, S., Murakami, K., Down, T.A., Leitch, H.G., and spectroscopy. Differences in methylation levels for each gene in different sam- Surani, M.A. (2013). Synergistic mechanisms of DNA demethylation during ples were calculated by subtracting the normalized average read counts in transition to ground-state pluripotency. Stem Cell Reports 1, 518–531. TKO samples from the respective WT values. If the difference was equal or Hajkova, P., Jeffries, S.J., Lee, C., Miller, N., Jackson, S.P., and Surani, M.A. larger than 0.25, the region was considered hypermethylated in the TKO. If (2010). Genome-wide in the mouse germ line entails the the difference was equal or smaller than 0.25, the region was considered pathway. Science 329, 78–82. hypomethylated. He, Y.-F., Li, B.-Z., Li, Z., Liu, P., Wang, Y., Tang, Q., Ding, J., Jia, Y., Chen, Z., Li, L., et al. (2011). Tet-mediated formation of 5-carboxylcytosine and its exci- ACCESSION NUMBERS sion by TDG in mammalian DNA. Science 333, 1303–1307. Huang, Y., Chavez, L., Chang, X., Wang, X., Pastor, W.A., Kang, J., Zepeda- The MeDIP sequencing data and gene-expression array data sets have been Martı´nez, J.A., Pape, U.J., Jacobsen, S.E., Peters, B., and Rao, A. (2014). deposited in the Gene Expression Omnibus database under accession Distinct roles of the methylcytosine oxidases Tet1 and Tet2 in mouse embry- numbers GSE55049 and GSE55574, respectively. onic stem cells. Proc. Natl. Acad. Sci. USA 111, 1361–1366.

110 Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. Developmental Cell Tet Enzymes and 5hmC in Differentiation

Ito, S., D’Alessio, A.C., Taranova, O.V., Hong, K., Sowers, L.C., and Zhang, Y. Sakaue, M., Ohta, H., Kumaki, Y., Oda, M., Sakaide, Y., Matsuoka, C., (2010). Role of Tet proteins in 5mC to 5hmC conversion, ES-cell self-renewal Yamagiwa, A., Niwa, H., Wakayama, T., and Okano, M. (2010). DNA methyl- and inner cell mass specification. Nature 466, 1129–1133. ation is dispensable for the growth and survival of the extraembryonic line- Ito, S., Shen, L., Dai, Q., Wu, S.C., Collins, L.B., Swenberg, J.A., He, C., and ages. Curr. Biol. 20, 1452–1457. Zhang, Y. (2011). Tet proteins can convert 5-methylcytosine to 5-formylcyto- Smith, Z.D., Chan, M.M., Mikkelsen, T.S., Gu, H., Gnirke, A., Regev, A., and sine and 5-carboxylcytosine. Science 333, 1300–1303. Meissner, A. (2012). A unique regulatory phase of DNA methylation in the early Koh, K.P., Yabuuchi, A., Rao, S., Huang, Y., Cunniff, K., Nardone, J., Laiho, A., mammalian embryo. Nature 484, 339–344. Tahiliani, M., Sommer, C.A., Mostoslavsky, G., et al. (2011). Tet1 and Tet2 Stroud, H., Feng, S., Morey Kinney, S., Pradhan, S., and Jacobsen, S.E. (2011). regulate 5-hydroxymethylcytosine production and cell lineage specification 5-hydroxymethylcytosine is associated with enhancers and gene bodies in hu- in mouse embryonic stem cells. Cell Stem Cell 8, 200–213. man embryonic stem cells. Genome Biol. 12, R54. Kriaucionis, S., and Heintz, N. (2009). The nuclear DNA base 5-hydroxymethyl- is present in Purkinje neurons and the brain. Science 324, 929–930. Tahiliani, M., Koh, K.P., Shen, Y., Pastor, W.A., Bandukwala, H., Brudno, Y., Agarwal, S., Iyer, L.M., Liu, D.R., Aravind, L., and Rao, A. (2009). Conversion Le, T., Kim, K.-P., Fan, G., and Faull, K.F. (2011). A sensitive mass spectrom- of 5-methylcytosine to 5-hydroxymethylcytosine in mammalian DNA by MLL etry method for simultaneous quantification of DNA methylation and hydroxy- partner TET1. Science 324, 930–935. methylation levels in biological samples. Anal. Biochem. 412, 203–209. Lei, H., Oh, S.P., Okano, M., Ju¨ ttermann, R., Goss, K.A., Jaenisch, R., and Li, Taiwo, O., Wilson, G.A., Morris, T., Seisenberger, S., Reik, W., Pearce, D., E. (1996). De novo DNA cytosine methyltransferase activities in mouse embry- Beck, S., and Butcher, L.M. (2012). Methylome analysis using MeDIP-seq onic stem cells. Development 122, 3195–3205. with low DNA concentrations. Nat. Protoc. 7, 617–636. Li, E., Bestor, T.H., and Jaenisch, R. (1992). Targeted mutation of the DNA Wernig, M., Meissner, A., Foreman, R., Brambrink, T., Ku, M., Hochedlinger, methyltransferase gene results in embryonic lethality. Cell 69, 915–926. K., Bernstein, B.E., and Jaenisch, R. (2007). In vitro reprogramming of fibro- Li, Z., Cai, X., Cai, C.L., Wang, J., Zhang, W., Petersen, B.E., Yang, F.C., and blasts into a pluripotent ES-cell-like state. Nature 448, 318–324. Xu, M. (2011). Deletion of Tet2 in mice leads to dysregulated hematopoietic Williams, K., Christensen, J., Pedersen, M.T., Johansen, J.V., Cloos, P.A.C., stem cells and subsequent development of myeloid malignancies. Blood Rappsilber, J., and Helin, K. (2011). TET1 and hydroxymethylcytosine in tran- 118, 4509–4518. scription and DNA methylation fidelity. Nature 473, 343–348. Moran-Crusio, K., Reavie, L., Shih, A., Abdel-Wahab, O., Ndiaye-Lobry, D., Wossidlo, M., Nakamura, T., Lepikhov, K., Marques, C.J., Zakhartchenko, V., Lobry, C., Figueroa, M.E., Vasanthakumar, A., Patel, J., Zhao, X., et al. Boiani, M., Arand, J., Nakano, T., Reik, W., and Walter, J.O.R. (2011). 5-hy- (2011). Tet2 loss leads to increased hematopoietic stem cell self-renewal droxymethylcytosine in the mammalian zygote is linked with epigenetic re- and myeloid transformation. Cancer Cell 20, 11–24. programming. Nat. Commun. 2, 241. Okano, M., Bell, D.W., Haber, D.A., and Li, E. (1999). DNA methyltransferases Wu, S.C., and Zhang, Y. (2010). Active DNA demethylation: many roads lead to Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian Rome. Nat. Rev. Mol. Cell Biol. 11, 607–620. development. Cell 99, 247–257. Panning, B., and Jaenisch, R. (1996). DNA hypomethylation can activate Xist Wu, H., and Zhang, Y. (2014). Reversing DNA methylation: mechanisms, geno- expression and silence X-linked genes. Genes Dev. 10, 1991–2002. mics, and biological functions. Cell 156, 45–68. Pastor, W.A., Pape, U.J., Huang, Y., Henderson, H.R., Lister, R., Ko, M., Xie, W., Schultz, M.D., Lister, R., Hou, Z., Rajagopal, N., Ray, P., Whitaker, McLoughlin, E.M., Brudno, Y., Mahapatra, S., Kapranov, P., et al. (2011). J.W., Tian, S., Hawkins, R.D., Leung, D., et al. (2013). Epigenomic analysis Genome-wide mapping of 5-hydroxymethylcytosine in embryonic stem cells. of multilineage differentiation of human embryonic stem cells. Cell 153, Nature 473, 394–397. 1134–1148.

Developmental Cell 29, 102–111, April 14, 2014 ª2014 Elsevier Inc. 111