Skip to main content
NCBI home page
Stem Cell Reports logoLink to Stem Cell Reports
. 2025 Jul 31;20(9):102595. doi: 10.1016/j.stemcr.2025.102595

Establishment of neuronal and glial competence of neural stem cells requires distinct enzymatic activities of TET enzymes

Blake C Ebert 1,2,3, Ian C MacArthur 1,2,3, Harmony C Ketchum 1,2,3, Michael Musheev 4, Christof Niehrs 4,5, Masako Suzuki 6, Meelad M Dawlaty 1,2,3,7,∗
PMCID: PMC12447321  PMID: 40749667

Summary

Ten-eleven translocation (TET1/2/3) enzymes are expressed in neural stem cells (NSCs). They iteratively oxidize 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC). The significance of hydroxymethylation (i.e., 5hmC) versus formylation and carboxylation (i.e., active demethylation) is undefined. We generated NSCs lacking only TET formylation and carboxylation activities (Tet-TFoCa) and compared them to NSCs lacking all three TET activities (Tet-TMut). Tet-TFoCa NSCs could differentiate into neurons but not into glial cells, while Tet-TMut NSCs could not form either cell type. Mechanistically, neuronal genes retained 5hmC at their enhancers in Tet-TFoCa NSCs and were expressed normally, consistent with the ability of these cells to form neurons. In contrast, enhancers of glial genes were hypermethylated in both Tet-TFoCa and Tet-TMut NSCs underpinning downregulation of these genes and the glial block in these cells. Our findings implicate TET-driven hydroxymethylation in establishing NSC neuronal competence and formylation and carboxylation in defining NSC glial competence.

Keywords: DNA demethylation, TET enzymes, 5hmC, 5fC, 5caC, Neural stem cells, neurons, glial

Highlights

  • •

    5hmC is required for establishing the neuronal competence of NSCs

  • •

    Presence of 5hmC at neuronal genes is sufficient for their proper expression in NSCs

  • •

    5fC and 5caC are essential for establishing the glial competence of NSCs

  • •

    Loss of 5fC and 5caC leads to hypermethylation and downregulation of glial genes

In this study, Dawlaty and colleagues have identified a key requirement for Tet-mediated DNA hydroxymethylation in establishing the neuronal competence of neural stem cells and a unique role for Tet-mediated DNA formylation and carboxylation in establishing the gliogenic competence of neural stem cells.

Introduction

Mammalian nervous system development is tightly regulated by epigenetic mechanisms that ensure proper activation of gene expression programs (Lister et al., 2013; Szulwach et al., 2011). Methylation of the 5-position-carbon of cytosines primarily within CpG dinucleotides is a predominant epigenetic modality in the eukaryotic genome (Smith and Meissner, 2013). 5-methylcytosine (5mC) is a repressive epigenetic mark deposited by the de novo DNA methyltransferases DNMT3A/B and maintained by the maintenance DNA methyltransferase DNMT1 (Smith and Meissner, 2013). Conversely, DNA demethylation promotes gene activation and is facilitated by the ten-eleven translocation (TET) family of dioxygenases (TET1, TET2, and TET3) (Tahiliani et al., 2009; Wu and Zhang, 2010). TET enzymes iteratively oxidize 5mC to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) (He et al., 2011; Ito et al., 2011; Tahiliani et al., 2009). The bases 5fC and 5caC are removed from DNA by thymine DNA glycosylase (TDG) followed by base excision repair resulting in active DNA demethylation (He et al., 2011). The base 5hmC evades recognition by UHRF1 and thereby prevents recruitment of DNMT1 during replication, leading to passive DNA demethylation as cells divide (Wu and Zhang, 2010). Because 5hmC is a more abundant and stable base than 5fC and 5caC in the mammalian genome, it is believed that 5hmC may have gene regulatory functions beyond promoting DNA demethylation.

Both TET enzymes and 5hmC are highly enriched in the nervous system and in neural progenitor cells (MacArthur et al., 2024; Ruzov et al., 2011), and previous work supports crucial neurodevelopmental roles for TET enzymes. NSCs or mice deficient in a single Tet gene demonstrate mild defects in proliferation but no deficits in differentiation or gross neuroanatomy (Li et al., 2017; Montalban-Loro et al., 2019; Rudenko et al., 2013). Tet1/2 and Tet1/3 double-knockout mice develop cranial neural tube defects (Dawlaty et al., 2013; Kang et al., 2015). Mouse embryos deficient for all three TETs fail to complete gastrulation (Dai et al., 2016), which has limited our ability to study the effects of loss of all three TET enzymes in neurodevelopment. Recently, by differentiating Tet1/2/3 triple-knockout mouse embryonic stem cells (ESCs) into NSCs, we demonstrated that TET-mediated developmental demethylation of neural enhancers is required for specification of NSCs with proper identity and gliogenic competence (MacArthur et al., 2024). We also found that in vivo deletion of all three Tet genes in the mouse dorsal telencephalon led to severe neuroanatomic abnormalities and postnatal lethality by 1 month of age (MacArthur et al., 2024). While these findings established a requirement for TET enzymes in commissioning NSC glial competence, they did not distinguish the role of DNA hydroxymethylation vs. formylation and carboxylation in NSC biology. The abundance and stability of 5hmC and its association with active genes in NSCs (MacArthur and Dawlaty, 2021; Pastor et al., 2013; Ruzov et al., 2011) suggest that these processes may have unique contributions to the biology of NSCs and thus warrants uncoupling these activities to define their biological significance in NSCs.

A point mutation in a conserved site within the catalytic domain of TET2 (TET2T1372E) was reported to allow conversion of 5mC to 5hmC but stalls its further oxidation, hence preventing the generation of 5fC and 5caC (Liu et al., 2017). Importantly, molecular modeling indicated that the mutants are not structurally disruptive to the TET protein structure (Torabifard and Cisneros, 2018). This mutation was used to show that DNA formylation and carboxylation, and thus active DNA demethylation, is essential for generation of induced pluripotent stem cells (iPSCs), as TET2-null mouse embryonic fibroblasts (MEFs) expressing TET2T1372E failed to reprogram into iPSCs (Caldwell et al., 2021). Likewise, this conserved mutation in TET1 was utilized to demonstrate that erasure of imprints during germ cell reprogramming requires TET1-mediated DNA formylation and carboxylation (Prasasya et al., 2024). These studies suggest that the distinct enzymatic activities of DNA hydroxymethylation vs. formylation and carboxylation may have unique functions in various developmental processes, in particular in NSCs where 5hmC is highly abundant and stable. Here, we have introduced the conserved TET2T1372E mutation in all three Tet genes in mouse ESCs to generate Tet1/2/3 triple formylation and carboxylation mutant (Tet-TFoCa) ESCs and have used them in conjunction with ESCs that are catalytically deficient for all Tet enzymatic activities (Tet1/2/3 triple mutant or Tet-TMut) (Chrysanthou et al., 2022; Flores et al., 2023; Ketchum et al., 2024). By differentiating the Tet-TFoCa and Tet-TMut ESCs to NSCs and subjecting them to comparative developmental, transcriptomic, and epigenomic analyses, we have identified a key requirement for Tet-mediated DNA hydroxymethylation in establishing NSC neuronal competence and a unique role for Tet-mediated DNA formylation and carboxylation in establishing NSC gliogenic competence. Given that TET enzymes are mutated or dysregulated and 5hmC levels are perturbed in neurodevelopmental disorders and neurodegenerative diseases, understanding the roles of each Tet enzymatic activity in NSC biology has implications for their effective therapeutic targeting.

Results

Generation of Tet-TFoCa and Tet-TMut ESCs

We applied CRISPR-Cas9-mediated gene editing to introduce the mouse equivalent of the previously reported human TET2 T1372E mutation (Liu et al., 2017) in Tet1 (T1610E), Tet2 (T1285E), and Tet3 (T940E) in mouse ESCs to generate Tet1/2/3 triple formylation-carboxylation mutant (Tet-TFoCa) ESCs (Figure S1A). This threonine (T) residue is conserved among all TET enzymes, and when mutated to a glutamic acid (E), it abrogates the ability of TET2 to catalyze the oxidation of 5hmC to 5fC and 5caC (Liu et al., 2017). To generate ESCs that are catalytically deficient for all Tet enzymatic activities (Tet-TMut), we introduced the H950Y and D952A mutations (Ketchum et al., 2024) into the Tet3 catalytic domain in our previously published Tet1/2 double catalytic mutant mouse ESCs (Flores et al., 2023). Properly targeted clones were screened by restriction fragment-length polymorphism (RFLP) using unique restrictions sites (Figure S1B) and further validated by Sanger sequencing (Figure S1C). We confirmed the wild-type (WT), TFoCa, and TMut Tet1/2/3 mRNA levels were properly and comparably expressed in WT, Tet-TFoCa, and Tet-TMut ESCs (Figure S1D). All genotypes were maintained for several passages in culture and retained normal morphology (Figure S1E). Quantification of the global levels of 5mC, 5hmC, 5fC, and 5caC by liquid chromatography-mass spectrometry (LC-MS) revealed that, consistent with the loss of all catalytic activities of the three Tet enzymes, Tet-TMut ESCs were depleted of 5hmC, 5fC, and 5caC (Figure S1F). In contrast, Tet-TFoCa ESCs were depleted of 5fC and 5caC but retained 5hmC, albeit to ∼40% of WT levels. This suggests that the FoCa mutation, while effectively blocking formylation and carboxylation, retains mild hydroxymethylation activity unlike originally reported (Liu et al., 2017). Nonetheless, given that Tet-TFoCa ESCs, but not Tet-TMut ESCs, retain 5hmC and that both Tet-TFoCa and Tet-TMut ESCs are devoid of 5fC and 5caC (Figure 1A), these lines can be used as a viable platform to dissect functional requirements of Tet-mediated DNA formylation and carboxylation from hydroxymethylation.

Figure 1.

Figure 1

Open in a new tab

Differentiation of Tet-TFoCa and Tet-TMut ESCs into NSCs

(A) Schematic of catalytic activities of WT, Tet-FoCa, and Tet-Mut proteins.

(B) Overview of differentiation protocol for specification of NSCs from embryonic stem cells (ESCs). LIF, leukemia inhibitory factor; ITSFn, insulin, transferrin, sodium selenite, fibronectin media (left). Bright-field images of WT, Tet-TFoCa, and Tet-TMut NSCs at days 3, 9, and 21 in culture (days after specification from ESCs). N.D., none detected. Scale bars: 500 μm.

(C) Growth curve of WT, Tet-TFoCa, and Tet-TMut NSCs over 3 weeks in culture. Cell numbers for each time point were plotted relative to cell numbers at day 0. Data represent mean ± SD. Data representative of three independent experiments.

(D) Quantification of 5mC, 5hmC, 5fC, and 5caC by liquid chromatography-mass spectrometry (LC-MS) in WT, Tet-TFoCa, and Tet-TMut neural stem cells (NSCs).

(E) Expression of NSC markers in WT, Tet-TFoCa, and Tet-TMut NSCs assessed by immunofluorescence at day 4 (passage 1). NSCs are stained with anti-SOX2 and anti-NESTIN antibodies. Nuclei are stained by 4′,6-diamidino-2-phenylindole (DAPI). Scale bars: 100 μm.

(F) Expression of neural markers quantified by RTqPCR in NSCs at day 4 (passage 1). Data normalized to Gapdh. Data represent mean ± SD. In all experiments, three WT, two Tet-TFoCa, and two Tet-TMut NSC biological replicates were used.

NSCs derived from Tet-TFoCa ESCs have longer self-renewal than NSCs derived from Tet-TMut ESCs

To dissect the biological requirements of DNA hydroxymethylation vs. DNA formylation and carboxylation in NSC biology, we differentiated WT, Tet-TFoCa, and Tet-TMut ESCs into NSCs following a well-established protocol (Lee et al., 2000; Okabe et al., 1996). As part of this differentiation protocol, ESCs were first differentiated to embryoid bodies (EBs), which showed no significant differences in expression of endodermal, mesodermal, ectodermal, and trophectodermal markers in EBs (Figure S2A), suggesting no major early block in specification of any specific germ layer. Both Tet-TFoCa and Tet-TMut ESCs yielded NSCs that were morphologically comparable to WT NSCs at day 3 (passage 1) in culture (Figure 1B). However, Tet-TMut NSCs could only be maintained in culture for 9 days (3 passages) in contrast to Tet-TFoCa NSCs that could be maintained for up to 21 days (7 passages) (Figure 1C). As expected, 5hmC, 5fC, and 5caC levels were completely depleted in Tet-TMut NSCs. In contrast, only 5fC and 5caC levels were depleted in Tet-TFoCa NSCs while 5hmC levels were retained to ∼40% levels of WT NSCs (Figure 1D). Both Tet-TFoCa and Tet-TMut NSCs stained positive for the neural marker SOX2, which confirmed the acquisition of a neural state (Figure 1E). However, staining of an additional NSC marker NESTIN (Figure 1E), as well as mRNA expression of other core neural transcription factors Sox1, Pax6, and Otx1, was severely reduced only in Tet-TMut NSCs (Figure 1F). These findings suggest that while loss of all three enzymatic activities of TET enzymes impairs the identity and self-renewal of NSCs, the retention of 5hmC in Tet-TFoCa NSCs supports better establishment of NSC identity and maintenance.

Tet-TFoCa NSCs can form neurons but not glial cells while Tet-TMut NSCs can form neither cell type

Next, we examined how TET-mediated DNA hydroxymethylation vs. formylation and carboxylation contributes to the multipotency of NSCs (a key feature of NSCs to form neurons and the glial cell types, astrocytes and oligodendrocytes). We differentiated WT, Tet-TFoCa, and Tet-TMut NSCs into neurons, astrocytes, and oligodendrocytes using established protocols (Belenguer et al., 2016; Gouti et al., 2014) (Figure 2A). For these experiments, we used day 3 NSCs to capture a state prior to loss of self-renewal. We found that Tet-TFoCa NSCs could form TUJ1+ neurons (Figure 2B, top) and expressed normal levels of the neuronal markers Dcx and Tuj1 (Figure 2C). However, Tet-TMut NSCs failed to differentiate into neurons with most cells dying by day 3 of differentiation (Figure 2B, top). The remaining live cells had significantly reduced expression of Dcx and Tuj1 (Figure 2C), indicating a block in differentiation into neurons. This suggests that TET-mediated DNA hydroxymethylation is essential for the ability of NSCs to form neurons while formylation and carboxylation are dispensable. Regarding their glial differentiation potential, we found that neither Tet-TFoCa nor Tet-TMut NSCs could form GFAP+ astrocytes (Figure 2B, middle). Consistently, these cells had severely reduced expression of astrocytic markers Aqp4 and Gfap (Figure 2D), supporting that DNA formylation and carboxylation is required for the differentiation of NSCs into astrocytes. Similarly, neither Tet-TFoCa nor Tet-TMut NSCs could form O4+ oligodendrocytes (Figure 2B, bottom), and both failed to induce expression of oligodendrocyte marker genes Mrf and Sox10 during differentiation (Figure 2E). These findings reveal a complete block in the ability of Tet-TFoCa and Tet-TMut NSCs to differentiate toward the glial lineage and highlight the important role of DNA formylation and carboxylation in establishing glial competence in NSCs.

Figure 2.

Figure 2

Open in a new tab

Differentiation of Tet-TFoCa and Tet-TMut NSCs into neurons and glial cell types

(A) Schematic of NSC differentiation into neurons (top), astrocytes (middle), and oligodendrocytes (bottom). EGF, epidermal growth factor; bFGF, basic fibroblast growth factor; FBS, fetal bovine serum; T3, triiodothyronine.

(B) Expression of lineage markers in WT, Tet-TFoCa, and Tet-TMut neurons (top), astrocytes (middle), and oligodendrocytes (bottom) assessed by immunofluorescence at day 4 of differentiation. Neurons are stained with anti-TUJ1, astrocytes are stained with anti-GFAP, oligodendrocytes are stained with anti-O4, and nuclei are stained with DAPI. Scale bars: 100 μm.

(C) Quantification of mRNA levels of neuronal markers in day 4 neuron cultures by RTqPCR.

(D) Quantification of mRNA levels of astrocyte markers in day 4 astrocyte cultures by RTqPCR.

(E) Quantification of mRNA levels of oligodendrocyte markers in day 4 oligodendrocyte cultures by RTqPCR.

For all experiments, three independent WT NSC lines and two independent Tet-TFoCa and Tet-TMut NSC lines were used. All lineage-specific differentiations were performed using passage 1 NSCs. All RTqPCR data were normalized to Gapdh. Data represent mean ± SD.

Neuronal differentiation programs are downregulated only in Tet-TMut NSCs while glial differentiation programs are downregulated in both Tet-TFoCa and Tet-TMut NSCs

To investigate the impact of DNA hydroxymethylation vs. formylation and carboxylation on gene expression of NSCs, we analyzed the transcriptome of WT, Tet-TFoCa and Tet-TMut NSCs by RNA sequencing (RNA-seq) (Figure S2B). Principal component analysis revealed strong correlation between the biological replicates for each genotype (Figure S2C). The bulk of the variance, PC1, described the differences in the transcriptomes and revealed that Tet-TFoCa NSCs were transcriptionally more similar to WT NSCs than to Tet-TMut NSCs. Likewise, sample clustering by Euclidean distance revealed that biological replicates of each genotype clustered together and that the transcriptomes of Tet-TFoCa NSCs clustered closer to WT NSCs, while the transcriptome of Tet-TMut NSCs clustered farther away (Figure S2D). A total of 5,151 genes were differentially expressed between the three genotypes (Figure S2E). We identified 1,355 differentially expressed genes (DEGs) in Tet-TFoCa NSCs (455 down and 900 up) and 3,796 DEGs in Tet-TMut NSCs (2256 down and 1540 up) when compared to WT NSCs (Figure 3A). Next, we defined two categories of genes: (1) genes that were uniquely deregulated in Tet-TMut NSCs, to gain insight into the neuronal differentiation deficits observed in Tet-TMut NSCs and (2) genes that were commonly deregulated between Tet-TMut and Tet-TFoCa NSCs, to explain the glial block observed in both of these cell lines. Of the 3,796 genes that were deregulated in Tet-TMut cells, 3,114 were deregulated uniquely in Tet-TMut NSCs while 682 genes were deregulated in both Tet-TMut and Tet-TFoCa NSCs (Figure 3B). Gene Ontology (GO) analysis of the 3,114 genes uniquely deregulated in the Tet-TMut NSCs was highly enriched for neuronal processes (Figure 3C). Most of these genes were downregulated and included key neuronal transcription factor genes including Neurog1/2, Sox11, and Ascl1 (Figure 3D). The downregulation of these genes may underlie the neuronal differentiation deficit observed in Tet-TMut NSCs. It supports that DNA hydroxymethylation (i.e., 5hmC) regulates NSC commitment toward the neuronal lineage, and therefore, loss of hydroxymethylation activity in Tet-TMut NSCs blocks their potential to form neurons. On the other hand, GO analysis of the 682 commonly deregulated genes in Tet-TMut and Tet-TFoCa NSCs revealed an enrichment for glial processes (Figure 3E). We found many critical glial transcription factor genes, including Zic3/4/5, Olig1/2, and Nkx6-2, to be downregulated in both Tet-TFoCa and Tet-TMut NSCs (Figure 3F). The downregulation of these genes may underlie the glial differentiation deficit observed in Tet-TFoCa and Tet-TMut NSCs. This suggests that Tet-mediated DNA formylation and carboxylation (i.e., active DNA demethylation) could play a role in activation of glial lineage programs in NSCs, and therefore, loss of these activities in Tet-TFoCa and Tet-TMut NSCs blocks their gliogenic competence.

Figure 3.

Figure 3

Open in a new tab

Transcriptomic analysis of Tet-TFoCa and Tet-TMut NSCs by RNA-seq

(A) Volcano plots of gene expression changes in Tet-TFoCa vs. WT NSCs (left) and Tet-TMut vs. WT NSCs (right). Upregulated differentially expressed genes (DEGs) are shown in red and downregulated DEGs are shown in blue.

(B) Total DEGs in Tet-TMut NSCs (3,796 genes) include 682 DEGs commonly deregulated in Tet-TMut and Tet-TFoCa NSCs and 3,114 DEGs uniquely deregulated in Tet-TMut NSCs.

(C) Representative gene ontology (GO) terms enriched in the 3,114 DEGs unique to Tet-TMut NSCs. The −log of the p value is plotted.

(D) Heatmap of expression of selected neuronal genes that are downregulated only in Tet-TMut NSCs.

(E) Representative GO terms enriched in the 682 common DEGs between Tet-TFoCa and Tet-TMut NSCs. The −log of the p value is plotted.

(F) Heatmap of expression of selected glial genes that are downregulated in both Tet-TFoCa and Tet-TMut NSCs.

Regions that retained 5hmC in Tet-TFoCa NSCs are associated to enhancers of neuronal genes, consistent with the ability of these cells to differentiate into neurons

To examine how DNA methylation and hydroxymethylation influences the glial and neuronal programs of NSCs, we assessed the genomic landscapes of 5mC and 5hmC in WT, Tet-TFoCa, and Tet-TMut NSCs by enzymatic methyl sequencing (EM-seq) and enzymatic-hydroxymethylation sequencing (E5hmC-seq), respectively. Our genome-wide mapping of 5mC and 5hmC revealed increased 5mC levels in both Tet-TFoCa and Tet-TMut NSCs compared to WT NSCs (Figure S3A). Consistent with our LC-MS data, we found that 5hmC levels were completely depleted in Tet-TMut NSCs but retained in Tet-TFoCa NSCs (1.2%) albeit to ∼40% of levels present in WT NSCs (3%) (Figures S3A–S3C). Comparison of the methylome and hydroxymethylome of Tet-TMut and Tet-TFoCa NSCs identified differentially methylated regions (DMRs) and differentially hydroxymethylated regions (DhMRs) between the two genotypes. Tet-TMut NSCs were predominantly hypermethylated (3,310 hyper-DMRs) and hypo-hydroxymethylated (940 hypo-DhMRs) compared to Tet-TFoCa NSCs (Figure S3D). Hyper-DMRs were largely found at promoters, whereas hypo-DhMRs were mainly found at intronic and intergenic regions (Figure S3E).

To evaluate the effects of TET enzymatic activities on the neuronal block uniquely seen in Tet-TMut NSCs, we focused on identifying the epigenetic deficits exclusive to the Tet-TMut NSCs. To this end, we focused on the 940 unique hypo-DhMRs in Tet-TMut NSCs. These regions maintained normal levels of 5hmC in Tet-TFoCa NSCs but lost 5hmC in Tet-TMut NSCs. (Figure 4A). They were associated to 874 genes of which 121 genes (14%) were downregulated (p < 5.69 × 10−180) uniquely in Tet-TMut NSCs (Figure 4B). Interestingly, this included many neuronal differentiation factors including Sox11, Rfx4, Npas3, Arid1a, Sema5b, Sema6a, Syn3, and Zfp521 (Figures 4C and 4D). This suggests that 5hmC may play a role in regulating the expression of these genes. Because the majority of DhMRs were in distal intergenic regions, we examined if they could be influencing enhancer elements. We overlapped a set of 804 TET-dependent neural active enhancers that we had previously identified (MacArthur et al., 2024) with the 940 regions that retained 5hmC in Tet-TFoCa NSCs. We found that 254/940 regions were located within active enhancers. These enhancers were in the vicinity of 228 genes that were differentially expressed in Tet-TMut NSCs (Figure 4E) and included several neuronal transcriptional factor genes including Zfp521, Npas3, Tox3, and Rbfox3, which were downregulated in Tet-TMut but not in Tet-TFoCa NSCs (Figure 4F). These genes were near enhancers that were hydroxymethylated in Tet-TFoCa but not in Tet-TMut NSCs (Figure 4G). Profiling of H3K27ac, an active enhancer mark, in Tet-TFoCa and Tet-TMut NSCs (Figures S4A–S4C) revealed that these enhancers were active in Tet-TFoCa but not in Tet-TMut NSCs (Figure 4G). Therefore, the proper expression of these neuronal genes may be controlled by 5hmC deposition at their enhancers. Globally, H3K27ac levels were reduced in both Tet-TFoCa and Tet-TMut NSCs compared to Tet-WT, with more pronounced reduction in Tet-TMut NSCs (Figures S4B and S4C). A similar pattern was observed at the 4,332 developmental DMRs—regions normally demethylated during differentiation of ESCs into NSCs and including many neural enhancers—where H3K27ac levels were markedly lower in Tet-TMut NSCs (Figure S4D). Importantly, at the 804 Tet-dependent neural enhancers, H3K27ac levels were comparable in Tet-WT and Tet-TFoCa but diminished in Tet-TMut NSCs (Figure S4E). This trend was particularly evident at enhancers of several neuronal genes that retained 5hmC in Tet-TFoCa NSCs but not in Tet-TMut NSCs (Figure 4G). Together, these data suggest that Tet-mediated DNA hydroxymethylation promotes proper expression of neuronal differentiation factors.

Figure 4.

Figure 4

Open in a new tab

Analysis of 5hmC levels at neuronal transcription factor genes in Tet-TFoCa NSCs

(A) Profile plot displaying average 5hmC levels ± 2 kb around the center of the 940 hypo-DhMRs in Tet-TMut NSCs.

(B) Number of genes associated to the 940 hypo-DhMRs unique to Tet-TMut NSCs. Fraction of genes that are up or downregulated in Tet-TMut NSCs are shown. Statistical enrichment determined by hypergeometric test.

(C) Heatmap of expression of genes that are uniquely hypo-hydroxymethylated and downregulated in Tet-TMut NSCs. Heatmap displays Z scores of normalized expression. Several neuronal genes are noted.

(D) Genome browser tracks displaying 5hmC levels at two neuronal genes. Hypo-DhMRs in Tet-TMut NSCs are highlighted.

(E) 254 regions that lose 5hmC uniquely in Tet-TMut NSCs are at active enhancers. H3K4me1 and H3K27ac data from MacArthur et al., Sci. Adv 2024 were used to define enhancers.

(F) Plot of normalized RNA-seq counts (i.e., expression) of key neuronal genes that are in proximity to hypo-DhMR-containing active enhancers. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Data represent mean ± SD.

(G) Genome browser tracks displaying 5hmC and H3K27ac levels at enhancers of neuronal transcription factor genes. DhMRs are highlighted. Note that these enhancers lack 5hmC and H3K27ac in Tet-TMut NSCs.

Of note, we also observed 3,310 hyper-DMRs unique to Tet-TMut NSCs, which were largely enriched at gene promoters (Figure S3E). They were associated to 2,239 genes of which 592 (26%) were downregulated (p < 3.10 × 10−116) uniquely in the Tet-TMut NSCs (Figure S5A). Many of them included the same neuronal differentiation factors whose enhancers were hydroxymethylated in Tet-TFoCa NSCs such as Sox11, Rfx4, Sema5b, Sema6a, and Syn3 (Figures S5B and S5C). This suggests that demethylation of promoters of these neuronal genes is likely another modality of gene regulation in addition to hydroxymethylation of their enhancers. Finally, we also found that promoters of neural genes important for NSC identity, such as Nestin, Sox1, Otx1, and Pax6, were hypermethylated in Tet-TMut NSCs but not in Tet-TFoCa NSCs (Figure S6). This is consistent with the downregulation of these genes only in Tet-TMut NSCs (Figure 1F) and may underpin the limited self-renewal potential of Tet-TMut NSCs compared to Tet-TFoCa NSCs.

Glial genes are hypermethylated and downregulated in both Tet-TMut and Tet-TFoCa NSCs consistent with the glial block in these cells

To evaluate the effects of Tet enzymatic activities on the glial block exhibited by both Tet-TFoCa and Tet-TMut NSCs, we focused on the common epigenetic defects between these two cell types. To this end, we identified regions that are commonly hypermethylated between Tet-TMut and Tet-TFoCa NSCs. We compared the methylome of Tet-TFoCa and Tet-TMut NSCs to that of WT NSCs and identified 5,321 hyper-DMRs in Tet-TFoCa and 7,386 hyper-DMRs in Tet-TMut NSCs, with 2,626 hyper-DMRs common between them (Figures 5A and 5B). 5mC levels were sharply increased at the center of the 2,626 common peaks in both Tet-TFoCa and Tet-TMut NSCs compared to WT NSCs (Figure 5C). We found that of the 682 genes that are deregulated in both Tet-TFoCa and Tet-TMut NSCs (Figure 3B), 155 genes (23%) contained hyper-DMRs in NSCs of both genotypes (Figure 5D). They included several glial transcription factor genes that were downregulated and hypermethylated in both Tet-TFoCa and Tet-TMut NSCs, including members of the Zic and Olig transcription factor families (Figures 5E and 5F). Because we have previously implicated Tet enzymes in demethylation of 804 neural active enhancers in NSCs (MacArthur et al., 2024), we examined if the 2,626 common hyper-DMRs in Tet-TFoCa and Tet-TMut NSCs impacted any of the neural enhancers. We found that 128 active neural enhancers contained a hyper-DMR (i.e., were hypermethylated in Tet-TFoCa and Tet-TMut NSCs) (Figure 5G). They were associated to 121 differentially expressed genes. Some of these genes were important glial transcription factors including Zic4, Gli2, and Sall2 that were downregulated in both Tet-TFoCa and Tet-TMut NSCs (Figure 5H). H3K27ac levels were reduced at enhancers of these glial genes in both Tet-TMut and Tet-TFoCa NSCs, and consistently these enhancers were hypermethylated in both lines (Figure 5I). This supports that TET-mediated formylation and carboxylation, leading to active DNA demethylation of enhancers of glial genes, is required for proper activation of these enhancers and expression of glial programs, leading to establishing the gliogenic competence of NSCs. Collectively, our findings establish that TET-driven DNA hydroxymethylation regulates the neuronal competence of NSCs and DNA formylation and carboxylation regulates the glial competence of NSCs (Figure 6).

Figure 5.

Figure 5

Open in a new tab

Analysis of 5mC levels at glial transcription factor genes in Tet-TFoCa and Tet-TMut NSCs

(A) Number of differentially methylated regions (DMRs) identified by EM-seq in Tet-TFoCa vs. WT NSCs and Tet-TMut vs. WT NSCs. DMRs were defined as regions with a minimum of 5 CpGs, average methylation difference of at least 20%, and FDR ≤0.05. Passage 1 NSCs were used for EM-seq analyses.

(B) Overlap of hyper-DMRs in Tet-TFoCa and Tet-TMut NSCs reveals the common hyper-DMRs between them.

(C) Profile plot displaying average 5mC levels ± 5 kb around the center of 2,626 common hyper-DMRs in Tet-TFoCa and Tet-TMut NSCs.

(D) Common DEGs in Tet-TFoCa and Tet-TMut NSCs that are hypermethylated in both cell types.

(E) Heatmap of expression of genes that are downregulated and hypermethylated in both Tet-TFoCa and Tet-TMut NSCs. Heatmap displays Z scores of normalized expression. Several glial genes are noted.

(F) Genome browser tracks displaying 5mC levels at three glial transcription factor genes. Hyper-DMRs in Tet-TFoCa and Tet-TMut NSCs are highlighted.

(G) Comparison of the 2,626 common hyper-DMRs in Tet-TFoCa and Tet-TMut NSCs with 804 neural enhancers in WT NSCs (from MacArthur et al., Sci. Adv, 2024) reveals 128 neural active enhancers that are hypermethylated in both Tet-TFoCa and Tet-TMut NSCs.

(H) Plot of normalized RNA-seq counts (i.e., expression) of several glial genes that are in proximity to hyper-DMR-containing active enhancers. Data represent mean ± SD.

(I) Genome browser tracks displaying 5mC and H3K27ac levels at enhancers of glial transcription factor genes. DMRs are highlighted. Note that these enhancers are hypermethylated and inactive (low H3K27ac) in both Tet-TFoCa and Tet-TMut NSCs.

Figure 6.

Figure 6

Open in a new tab

TET-driven DNA hydroxymethylation promotes neuronal competence of NSCs while Tet-driven DNA formylation and carboxylation—mediating active demethylation—promotes the glial competence of NSCs

Discussion

TET enzymes oxidize 5mC to 5hmC, 5fC, and 5caC in a stepwise manner (Ito et al., 2011; Tahiliani et al., 2009; Wu and Zhang, 2010). However, the significance of hydroxymethylation versus formylation and carboxylation in NSC biology is not well defined. In this study, by comparing the self-renewal, multipotency, transcriptome, and epigenome of Tet-TFoCa and Tet-TMut NSCs, we provide five lines of evidence supporting distinct requirements for both TET-mediated DNA hydroxymethylation and formylation and carboxylation in NSC biology. (1) Tet-TFoCa NSCs had extended self-renewal capacity compared to Tet-TMut NSCs. (2) Tet-TFoCa NSCs could differentiate to neurons, but not to glial cells, compared to Tet-TMut NSCs, which were unable to form cells of either lineage. (3) Glial differentiation genes were downregulated in both Tet-TFoCa and Tet-TMut NSCs while neuronal differentiation genes were downregulated only in Tet-TMut NSCs. (4) Neuronal transcription factor genes retained 5hmC at their enhancers in Tet-TFoCa NSCs, which remained active, leading to normal expression of these genes, consistent with the ability of these cells to form neurons. (5) Glial transcription factor genes and their enhancers were hypermethylated and inactive in both Tet-TFoCa and Tet-TMut NSCs, consistent with the inability of these cells to form glial cells. These findings implicate TET-driven DNA hydroxymethylation in establishing the neuronal competence of NSCs and DNA formylation and carboxylation in establishing the glial competence of NSCs.

TET enzymes and 5hmC have been well studied in the context of neural progenitor cells and throughout neurodevelopment (Jin et al., 2011; Khare et al., 2012; Szulwach et al., 2011). Their levels increase significantly from the early postnatal period to adulthood, particularly within post-mitotic neurons (Lister et al., 2013). Interestingly, the increase in 5hmC is not always associated with DNA demethylation (Onodera et al., 2021), suggesting that hydroxymethylation could possess independent regulatory functions particularly at genes critical for neurodevelopment. However, a lack of suitable models has prevented dissecting the role of 5hmC from the two less abundant bases 5fC and 5caC, which are removed from the genome to promote active demethylation. The identification of conserved amino acids within the catalytic pocket of TET2 that are required for interaction with 5hmC, thereby allowing DNA hydroxymethylation but not formylation and carboxylation (i.e., FoCa mutant), has created opportunities for uncoupling the biological requirements of TET-modified cytosine bases (Liu et al., 2017). However, very few studies have utilized these mutations to investigate the distinct functions of each TET enzymatic activity in gene regulation and in biological processes. For example, it has been shown that Tet2-FoCa mutant fibroblasts, like Tet2 knockout fibroblasts, cannot form iPSCs, demonstrating that the mere loss of 5fC and 5caC blocks iPSC formation and therefore Tet-driven active demethylation is critical for reprogramming into iPSCs (Caldwell et al., 2021). Likewise, Tet1-FoCa mutant mice fail to erase imprints in primordial germ cells during germ cell reprogramming, suggesting that 5fC- and 5caC-mediated active demethylation is critical for imprint erasure (Prasasya et al., 2024). However, no exclusive roles for 5hmC have been attributed to any specific biological process. Moreover, given the redundancy among TET enzymes, it is probable that loss of enzymatic activities of individual Tet genes is compensated by other Tet paralogs. Therefore, it is imperative to use a system with FoCa mutations in all three TETs to establish the absolute requirements of 5hmC vs. 5fC and 5caC. Our generation of Tet-TFoCa ESCs revealed that the FoCa mutations, while completely blocking formylation and carboxylation, are not fully permissive of DNA hydroxymethylation as expected and previously reported (Liu et al., 2017), as 5hmC levels are reduced to ∼40% of WT levels. Nevertheless, both Tet-TFoCa ESCs and NSCs retained a substantial amount of 5hmC, and using them in conjunction with Tet-TMut ESCs and NSCs provided a viable platform for distinguishing 5hmC vs. 5fC and 5caC requirements in NSC biology.

TET enzymes have been implicated in mammalian neurodevelopment (Lister et al., 2013) and primarily exert their effects at promoters and enhancers of neural genes (Hon et al., 2014). Deficiency of TETs affects differentiation and developmental potential of NSCs (Montalban-Loro et al., 2019; Pastor et al., 2013). Recently, we have demonstrated that TET enzymes are required for specification and multipotency of NSCs, where they demethylate and commission neural and glial enhancers to activate crucial neural programs that establish the gliogenic competence in NSCs (MacArthur et al., 2024). Our findings in this study that TET-mediated DNA hydroxymethylation promotes the neuronal while TET-mediated DNA formylation and carboxylation promotes the glial competence of NSCs expand on our previous work. (1) While we had shown that glial enhancers are demethylated and activated by TET enzymes in NSCs (MacArthur et al., 2024), implicating DNA formylation and carboxylation in proper expression of glial genes demonstrates that glial enhancer demethylation and activation specifically require active demethylation. (2) Implicating DNA hydroxymethylation in proper expression of neuronal genes in NSCs reveals a specific requirement for 5hmC in the formation of neurons. This explains why Tet triple knockout (Tet-TKO) (MacArthur et al., 2024) and Tet-TMut NSCs have compromised neuronal differentiation while Tet-TFoCa NSCs (which have sufficient 5hmC retention at neuronal enhancers and genes) can differentiate into neurons. This specific requirement for 5hmC in neuron formation may explain the high abundance of 5hmC in neurons compared to any other brain cell type (Diotel et al., 2017; MacArthur et al., 2024). (3) Finally, our findings that Tet-TKO (MacArthur et al., 2024) and Tet-TMut NSCs have limited self-renewal of up to 3 passages in contrast to Tet-TFoCa NSC, which can be maintained for up to 7 passages, suggest that both 5hmC and 5fC/5caC contribute to NSC self-renewal and maintenance. The downregulation and hypermethylation of promoters of key NSC transcription factors (i.e., Sox1, Pax6, and Otx1) in Tet-TMut NSCs, but not in Tet-TFoCa NSCs, support this. It is likely that their proper expression in Tet-TFoCa NSCs is due to hydroxymethylation of their promoters that leads to passive demethylation.

Previously, we have shown that loss of all three Tet enzymes in mouse frontal cortex during embryogenesis leads to lethality by 4 weeks of age with severe neuroanatomical defects including underdeveloped hippocampus and agenesis of the corpus callosum (MacArthur et al., 2024). It would be interesting to examine how Tet-FoCa NSCs, which have extended self-renewal and the potential to form neurons, contribute to neurodevelopment. Future studies involving intracranial transplantation of Tet-FoCa and Tet-TMut NSCs or generation of conditional Tet-TFoCa mice may provide more insights on the neurodevelopmental contributions of 5hmC vs. 5fC and 5caC. Likewise, additional studies are warranted to assess the electrophysiological and other functional properties of neurons derived from Tet-FoCa NSCs. Such investigations may identify more unique and common requirements of 5hmC vs. 5fC/5caC in neurogenesis and neurodevelopment.

While we implicate 5hmC in the regulation of neuronal genes in NSCs, further studies are warranted to expand upon the mechanisms involved. 5hmC is a relatively stable base in NSCs and may serve as an independent epigenetic mark and control the recruitment of specific factors to gene regulatory elements. For example, it has been reported that the methyl CpG binding protein 2 has a lower binding affinity for 5hmC, while methyl DNA-binding domain 3 has a stronger 5hmC binding affinity (Mellen et al., 2012; Yildirim et al., 2011). Likewise, other mechanisms may include 5hmC evading recognition by UHRF1 and thereby preventing recruitment of DNMT1 during replication, leading to passive DNA demethylation (Pastor et al., 2013). Future mapping of readers and evaders of 5hmC will certainly provide more mechanistic insights into how 5hmC regulates neuronal genes in NSCs. Moreover, recent findings that 5fC can act as an activating mark during zygotic genome activation in Xenopus embryos, where it stimulates RNA polymerase III transcription (Parasyraki et al., 2024), raise the possibility of DNA demethylation-independent roles of 5fC in other cell types. Although 5fC is less abundant in NSCs, future advancement in technologies to properly map low levels of 5fC may facilitate investigating potential 5fC regulatory roles in NSCs.

In conclusion, our findings implicate specific TET enzymatic activities to neuronal vs. glial lineage potentiation of NSCs. This will have implications for targeting specific TET activities or reconfiguring the landscape of TET-modified cytosines for improving differentiation of NSCs into glial or neuronal cell types, as well as in neurodevelopmental disorders and neurovegetative diseases where TET enzymes are dysregulated or mutated and 5hmC levels are perturbed. Tet1 is linked to familial intellectual disability and Parkinson’s. Tet2 loss is linked to intellectual disability, Alzheimer’s, and frontotemporal dementia. Tet3 mutations are associated with intellectual disability and craniofacial abnormalities. 5hmC levels are altered in autism and neurodevelopmental disorders (MacArthur and Dawlaty, 2021). Our findings will open future opportunities to establish how each TET enzymatic activity contributes to the aforementioned disease etiologies and to identify relevant TET-modified bases with clinical diagnostic or therapeutic potential.

Limitations of the study

The substantial retention of 5hmC in Tet-TFoCa NSCs along with our focus on the 940 regions that retained normal levels of 5hmC in these cells has allowed us to identify unique functions of this base in neuronal competence of NSCs. We acknowledge that while the Tet-FoCa mutation fully blocks DNA formylation and carboxylation, its inability to generate WT levels of 5hmC may have limited the discovery of all functions of 5hmC in NSCs and should be taken into consideration in data interpretation. Future identification and testing of other mutations within the catalytic pocket of TET enzymes that would allow normal DNA hydroxymethylation but block DNA formylation and carboxylation may serve as more stringent models for comprehensively understanding all roles and requirements of 5hmC vs. 5fC and 5caC. Likewise, we did not map the genomic distribution of 5fC and 5caC as they were completely depleted in Tet-TFoCa and Tet-TMut NSCs and are difficult to map even in WT cells because they are readily removed by TDG (He et al., 2011). Current technologies require inhibiting TDG to map these bases, but that can confound interpretation of data due to TDG’s own critical roles in development (Cortazar et al., 2011). Advances in technologies to better map 5fC and 5caC may further solidify their roles in active DNA demethylation at gene regulatory elements. Finally, it is possible that in Tet-TMut NSCs, the mutant TET proteins occupy their target sites preventing other compensatory proteins from accessing these regions, or that the mutant proteins have alerted binding affinity to their target regions. The lack of suitable Tet antibodies to map their genomic occupancy restricts our ability to properly investigate these possibilities and should be taken into consideration.

Methods

Generation of Tet-TFoCa and Tet-TMut ESC lines

Tet-TFoCa and Tet-TMut mouse ESCs (line V6.5, male, 129/B6 mixed background) were generated following our previously published CRISPR-Cas9 gene editing protocols (Flores et al., 2023; Wang et al., 2013). Tet-TFoCa ESCs: a gene block (Table S1) carrying nucleotide substitutions in Tet1 exon 10 to introduce T1610E mutation and silent mutations to establish a novel HaeIII site was synthesized by IDT and cloned into a pCR 2.1 TOPO vector following the kit protocols. This donor plasmid along with a pX330 plasmid expressing Cas9 and a guide RNA against Tet1 exon 10 (Table S1) was used for gene editing in v6.5 mouse ESCs following our previously published protocols (Wang et al., 2013). Targeted clones were screened by RFLP using the introduced HaeIII site, and properly edited homozygous clones were verified by Sanger sequencing. Next, the Tet2 locus was targeted in this line using the same strategy as described earlier as gene blocks and gRNAs listed in Table S1 to introduce T1285E mutation in Tet2 exon 9 and silent mutations to establish a novel XbaI restriction enzyme site. Properly targeted clones were validated, and a resulting clone was targeted to introduce T1610E mutation in Tet3 exon 9 and silent mutations to establish a novel HaeIII restriction enzyme site using gene blocks and gRNAs listed in Table S1. Targeted cones were screened and three independent clones carrying Tet1 T1610E, Tet2 T1285E, and Tet3 T1610E homozygous mutations (i.e., Tet-TFoCa) were validated and used for experiments. Tet-TMut mouse ESCs: in our previously published Tet1/2 double catalytic mutant ESCs, which carried Tet1 H1621Y and D1623A and Tet2 H1367Y and D1369A mutations (Flores et al., 2023), we targeted the Tet3 locus to introduce H950Y and D952A mutation in Tet3 exon 9 using a targeting strategy, donor vector, and gRNAs from our previously study (Ketchum et al., 2024). Three independent clones were validated by RFLP and Sanger sequencing and used in experiments. Proper expression of catalytic mutant and FoCa mutant Tet1/2/3 mRNAs was confirmed in all clones by quantitative real-time PCR (real-time qPCR). Sequences of gRNAs, donor vectors, and primers are shown in Table S1.

Differentiation of ESCs into NSCs and culture of NSCs

All ESCs were cultured on irradiated feeders using our previously published protocols (Chrysanthou et al., 2022). Differentiation of ESCs into NSCs was performed using the method of Okabe et al. (1996), which was modified in the study by Lee et al. (2000) and recently used by us in the study by MacArthur et al. (2024). Briefly, ESCs were differentiated in the absence of leukemia inhibitory factor to form EBs over 4 days. EBs were seeded to attach on tissue culture plates and cultured in insulin, transferrin, sodium selenite, fibronectin medium (DMEM/F-12 with 5 μg/mL insulin, 50 μg/mL apotransferrin, 30 nM sodium selenite, 5 μg/mL fibronectin, and 100 U/mL penicillin with 100 g/mL streptomycin). After 8 days, cells were dissociated and seeded onto poly-l-ornithine hydrobromide and laminin-coated plates at a density of 1.0 × 105 to 1.5 × 105 cells/cm2 in N2 medium (DMEM/F-12 with 25 μg/mL insulin, 50 μg/mL apotransferrin, 20 nM progesterone, 100 μM putrescine, 30 nM sodium selenite, 100 U/mL penicillin, and 100 g/mL streptomycin) supplemented with 20 ng/mL epidermal growth factor, 10 ng/mL basic fibroblast growth factor (bFGF), and 1 μg/mL laminin. NSCs were passaged every 2 to 3 days after dissociation with Accutase. NSC growth over several passages was assessed following our previous protocol (MacArthur et al., 2024). For all experiments, NSCs were harvested or analyzed after one or two passages (between 3 and 6 days of NSC culture).

NSC differentiation into neurons, astrocytes, and oligodendrocytes

For lineage-specific differentiations, NSCs at passage 1 (day 3) were seeded into Matrigel growth factor reduced-coated wells. For wells used for immunofluorescence staining, coverslips were placed in wells before coating. Neuron differentiation was adapted from Gouti et al. (2014) and recently used by us and described in detail in the study by MacArthur et al. (2024). Astrocyte differentiation was adapted from Belenguer et al. (2016) and recently used by us and described in detail in the study by MacArthur et al. (2024). Oligodendrocyte differentiation was according to the study by MacArthur et al. (2024) where NSCs were cultured in N2/B27 medium with 10 ng/mL bFGF for 24 h and then in N2/B27 medium with 30 ng/mL triiodo-L-thyronine for 72 h. For all three differentiations, at 96 h, cells in wells for RNA extraction were lysed directly in wells, and RNA was isolated. Cells in wells for immunofluorescence staining were fixed and stained using a protocol adapted from Belenguer et al. (2016) and recently used by us and described in detail in the study by MacArthur et al. (2024). Primary antibodies (anti-NESTIN [1:200], anti-SOX2 [1:200], anti-TUJ1 [1:150], anti-GFAP [1:300], and anti-O4 [5 μg/mL]) and secondary antibodies (anti-mouse Alexa Fluor 488 [1:500], anti-mouse Alexa Fluor 594 [1:500], and anti-rabbit Alexa Fluor 594 [1:500]) were used. Cells were stained with 4′,6-diamidino-2-phenylindole. Coverslips were mounted with VECTASHIELD Antifade Mounting Medium onto Superfrost Plus slides and imaged with a Zeiss Apotome.2 fluorescence microscope.

RTqPCR

RTqPCR was performed as previously described (Chrysanthou et al., 2022). Briefly, RNA was extracted using the Omega EZNA Total RNA Kit following manufacturer’s protocols and subjected to cDNA synthesis using the Superscript IV kit, according to the manufacturer’s guidelines. 10 ng of cDNA per reaction was subjected to real-time qPCR using Fast SYBR Green Master Mix and primers listed in Table S1 (MacArthur et al., 2024) on a BD Applied Biosystems StepOne Real-Time PCR System. For quantification of lineage markers in EBs, we used previously published primers (Chrysanthou et al., 2022). Data were normalized to Gapdh.

Quantification of 5mC, 5hmC, 5fC, and 5caC by LC-MS

WT, Tet-TFoCa, and Tet-TMut NSCs were harvested at passage 2 (day 6) of culture and lysed with proteinase K overnight. Genomic DNA was extracted by ethanol precipitation with ammonium acetate and treated with RNase A. Quantification of 5mC, 5hmC, 5fC, and 5caC was carried out as previously described (Schomacher et al., 2016).

RNA-seq and data analysis

Three independent biological replicates of WT, Tet-TFoCa, and Tet-TMut NSCs were subjected to RNA-seq as previously described (Chrysanthou et al., 2022). Briefly, NSCs were harvested at passage 2 (day 6) of culture and RNA was extracted with the Omega EZNA Total RNA Kit I. Library preparation and 150 base pair (bp) paired-end sequencing were performed at Novogene. Libraries were sequenced using the Illumina NovoSeq 6000 platform. Adaptor and low-quality trimming were performed with trim galore (v.0.6.5, www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Clean reads were mapped to Mus musculus reference genome mm10 using STAR (v.2.7.3a) (Dobin et al., 2013) with default parameters. Counts were extracted from mapped reads with featureCounts using the largest Overlap parameter. DESeq2 (v.1.40.2) (Love et al., 2014) was used to identify DEGs (false discovery rate [FDR] < 0.05, fold change >1.5) from raw counts using the standard package documentation. GO terms were identified on selected sets of DEGs using DAVID (Sherman et al., 2022). Statistical significance of gene set enrichments was performed using the hypergeometric test.

EM-seq, E5hmC-seq, and data analysis

One clone each of WT, Tet-TFoCa, and Tet-TMut NSCs was harvested at passage 2 (day 6) of culture. DNA was extracted with the Quick-DNA Miniprep Kit (Zymo, D3024) according to the manufacturer’s instructions. 200 ng of DNA per sample was used for EM-seq or E5hmC-seq. For EM-seq, DNA was oxidized, deaminated, and amplified using the NEBNext EM-seq kit (NEB E7120). For E5hmC-seq, DNA was glucosylated, deaminated, and amplified using the NEBNext E5hmC-seq kit (NEB E3350). Both EM-seq and E5hmC-seq library sets were subjected to 300-bp paired-end sequencing at Novogene on the Novaseq X Plus (PE150-25B) platform. Spike-in lambda (negative control), pUC19 (5mC positive control), and T4 (5hmC positive control) DNA were mapped to their respective genomes using Bismark (v.0.24.2) (Krueger and Andrews, 2011) to assess reaction efficacy. Sample reads were trimmed using trim_galore (v.0.6.7) and mapped to mm10 using Bismark with default parameters. Read deduplication was performed using deduplicate_bismark, and cytosine methylation levels were extracted with bismark_methylation_extractor. BAM files were balanced (i.e., downsampled) to the sample with the lowest number of reads with SAMtools (v.1.9) (Danecek et al., 2021). BAM files were pre-processed for MethPipe (v.3.4.3) (Song et al., 2013) using format_reads and sorted using SAMtools. Single-site methylation levels were computed using methcounts, and symmetric CpG methylation was extracted using symmetric-cpgs. DMRs and DhMRs were identified between Tet-TFoCa and Tet-TMut, Tet-FoCa and WT, and Tet-TMut and WT using MethPipe with standard parameters (regions with >5 CpGs, methylation difference >20%, and FDR < 0.05). DMRs and DhMRs were annotated to genomic regions with the R package ChIPseeker (v.1.30.3) (Yu et al., 2015). DMRs and DhMRs were assigned to genes based on proximity to the closest transcription start site. Line plots were generated with the plotProfile function of deepTools (v.3.5.1) (Ramirez et al., 2016). The values used in the profile plot represent absolute or percentage methylation levels at each CpG site, calculated as the ratio of methylated reads to total coverage at that position (i.e., methylated/total reads). These ratios are scaled between 0 and 1 (or 0–100%). For visualization on Integrative Genome Browser, bedGraph files were converted to bigwigs using bedtools bedGraphToBigWig (v.2.30.0) (Kent et al., 2010). Bed files were intersected using bedtools intersect with default parameters. Enhancers were identified by mapping the genomic distribution of the enhancer mark H3K27ac by CUT&Tag following our previously used protocols and data analysis pipelines (MacArthur et al., 2024).

Statistical analysis

GraphPad Prism (v.8.4.3), bedtools2 (v.2.28.0), and R software (v.4.0.3) were used for statistical analyses. Unpaired Student’s t test was used to compare two groups. One-way analysis of variance was used to compare more than two groups. Enrichment and depletion of groups for features were assessed with the hypergeometric test. p values were adjusted for multiple testing correction where applicable. Statistical analyses for genome-wide methods are explained in detail under the respective methods subsections.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Meelad M. Dawlaty (meelad.dawlaty@einsteinmed.edu).

Materials availability

All the materials generated and used in this study will be available upon request.

Data and code availability

The RNA-seq, EM-seq, EhmC-seq, and CUT&Tag data have been deposited in the Gene Expression Omnibus database accession (GSE287657) and are publicly available as of the date of publication.

Acknowledgments

We thank all members of the Dawlaty Lab for helpful discussions and critical reading of the manuscript. We are grateful to Shahina Maqbool and Robert Dubin from the Einstein Epigenomics Facility for help with library preparation and data processing. This work was in part supported by funds to M.M.D. from the Einstein Rose F Kennedy Intellectual and Developmental Disabilities Research Center (IDDRC) pilot grant and Einstein 2030 Seed Funds pilot grant. M.M.D. was also supported by funds from the National Institutes of Health 1R01HL172730 and 2R01HL148852 as well as Hirschle Trust Funds. B.C.E. was supported by NIH predoctoral fellowship F30HD113183. I.C.M. was supported by NIH predoctoral fellowship F30HD107921. B.C.E. and I.C.M. were supported by Medical Scientist Training Program training grant T32GM149364. H.C.K. was supported by NIH T32GM007491 training grant in cellular and molecular biology and genetics and the Paul S. Frenette Scholar Awards Program of the Ruth L. and David S. Gottesman Institute for Stem Cell Research and Regenerative Medicine at Albert Einstein College of Medicine.

Author contributions

B.C.E. performed the experiments, analyzed the data, and prepared the figures and the first draft of the manuscript. I.C.M. assisted with establishing the ESC to NSC differentiation platform. M.M. and C.N. quantified the global levels of modified cytosine bases. M.S. and H.C.K. assisted with next-generation sequencing data analyses. M.M.D. and B.C.E. designed the study and wrote the manuscript with input from all authors. M.M.D. supervised the study and secured funding.

Declaration of interests

The authors declare no competing interests.

Published: July 31, 2025

Footnotes

Supplemental information can be found online at https://doi.org/10.1016/j.stemcr.2025.102595.

Supplemental information

Document S1. Figures S1–S6 and Tables S1 and S2
mmc1.pdf (3MB, pdf)
Document S2. Article plus supplemental information
mmc2.pdf (36.4MB, pdf)

References

  1. Belenguer G., Domingo-Muelas A., Ferrón S.R., Morante-Redolat J.M., Fariñas I. Isolation, culture and analysis of adult subependymal neural stem cells. Differentiation. 2016;91:28–41. doi: 10.1016/j.diff.2016.01.005. [DOI] [PubMed] [Google Scholar]
  2. Caldwell B.A., Liu M.Y., Prasasya R.D., Wang T., DeNizio J.E., Leu N.A., Amoh N.Y.A., Krapp C., Lan Y., Shields E.J., et al. Functionally distinct roles for TET-oxidized 5-methylcytosine bases in somatic reprogramming to pluripotency. Mol. Cell. 2021;81:859–869.e8. doi: 10.1016/j.molcel.2020.11.045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Chrysanthou S., Tang Q., Lee J., Taylor S.J., Zhao Y., Steidl U., Zheng D., Dawlaty M.M. The DNA dioxygenase Tet1 regulates H3K27 modification and embryonic stem cell biology independent of its catalytic activity. Nucleic Acids Res. 2022;50:3169–3189. doi: 10.1093/nar/gkac089. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cortazar D., Kunz C., Selfridge J., Lettieri T., Saito Y., MacDougall E., Wirz A., Schuermann D., Jacobs A.L., Siegrist F., et al. Embryonic lethal phenotype reveals a function of TDG in maintaining epigenetic stability. Nature. 2011;470:419–423. doi: 10.1038/nature09672. [DOI] [PubMed] [Google Scholar]
  5. Dai H.Q., Wang B.A., Yang L., Chen J.J., Zhu G.C., Sun M.L., Ge H., Wang R., Chapman D.L., Tang F., et al. TET-mediated DNA demethylation controls gastrulation by regulating Lefty-Nodal signalling. Nature. 2016;538:528–532. doi: 10.1038/nature20095. [DOI] [PubMed] [Google Scholar]
  6. Danecek P., Bonfield J.K., Liddle J., Marshall J., Ohan V., Pollard M.O., Whitwham A., Keane T., McCarthy S.A., Davies R.M., Li H. Twelve years of SAMtools and BCFtools. GigaScience. 2021;10 doi: 10.1093/gigascience/giab008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Dawlaty M.M., Breiling A., Le T., Raddatz G., Barrasa M.I., Cheng A.W., Gao Q., Powell B.E., Li Z., Xu M., et al. Combined deficiency of Tet1 and Tet2 causes epigenetic abnormalities but is compatible with postnatal development. Dev. Cell. 2013;24:310–323. doi: 10.1016/j.devcel.2012.12.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Diotel N., Mérot Y., Coumailleau P., Gueguen M.M., Sérandour A.A., Salbert G., Kah O. 5-hydroxymethylcytosine marks postmitotic neural cells in the adult and developing vertebrate central nervous system. J. Comp. Neurol. 2017;525:478–497. doi: 10.1002/cne.24077. [DOI] [PubMed] [Google Scholar]
  9. Dobin A., Davis C.A., Schlesinger F., Drenkow J., Zaleski C., Jha S., Batut P., Chaisson M., Gingeras T.R. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29:15–21. doi: 10.1093/bioinformatics/bts635. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Flores J.C., Sidoli S., Dawlaty M.M. Tet2 regulates Sin3a recruitment at active enhancers in embryonic stem cells. iScience. 2023;26 doi: 10.1016/j.isci.2023.107170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gouti M., Tsakiridis A., Wymeersch F.J., Huang Y., Kleinjung J., Wilson V., Briscoe J. In vitro generation of neuromesodermal progenitors reveals distinct roles for wnt signalling in the specification of spinal cord and paraxial mesoderm identity. PLoS Biol. 2014;12 doi: 10.1371/journal.pbio.1001937. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. He Y.F., Li B.Z., Li Z., Liu P., Wang Y., Tang Q., Ding J., Jia Y., Chen Z., Li L., et al. Tet-mediated formation of 5-carboxylcytosine and its excision by TDG in mammalian DNA. Science. 2011;333:1303–1307. doi: 10.1126/science.1210944. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hon G.C., Song C.X., Du T., Jin F., Selvaraj S., Lee A.Y., Yen C.A., Ye Z., Mao S.Q., Wang B.A., et al. 5mC oxidation by Tet2 modulates enhancer activity and timing of transcriptome reprogramming during differentiation. Mol. Cell. 2014;56:286–297. doi: 10.1016/j.molcel.2014.08.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Ito S., Shen L., Dai Q., Wu S.C., Collins L.B., Swenberg J.A., He C., Zhang Y. Tet proteins can convert 5-methylcytosine to 5-formylcytosine and 5-carboxylcytosine. Science. 2011;333:1300–1303. doi: 10.1126/science.1210597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Jin S.G., Wu X., Li A.X., Pfeifer G.P. Genomic mapping of 5-hydroxymethylcytosine in the human brain. Nucleic Acids Res. 2011;39:5015–5024. doi: 10.1093/nar/gkr120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Kang J., Lienhard M., Pastor W.A., Chawla A., Novotny M., Tsagaratou A., Lasken R.S., Thompson E.C., Surani M.A., Koralov S.B., et al. Simultaneous deletion of the methylcytosine oxidases Tet1 and Tet3 increases transcriptome variability in early embryogenesis. Proc. Natl. Acad. Sci. USA. 2015;112:E4236–E4245. doi: 10.1073/pnas.1510510112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kent W.J., Zweig A.S., Barber G., Hinrichs A.S., Karolchik D. BigWig and BigBed: enabling browsing of large distributed datasets. Bioinformatics. 2010;26:2204–2207. doi: 10.1093/bioinformatics/btq351. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ketchum H.C., Suzuki M., Dawlaty M.M. Catalytic-dependent and -independent roles of TET3 in the regulation of specific genetic programs during neuroectoderm specification. Commun. Biol. 2024;7:415. doi: 10.1038/s42003-024-06120-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Khare T., Pai S., Koncevicius K., Pal M., Kriukiene E., Liutkeviciute Z., Irimia M., Jia P., Ptak C., Xia M., et al. 5-hmC in the brain is abundant in synaptic genes and shows differences at the exon-intron boundary. Nat. Struct. Mol. Biol. 2012;19:1037–1043. doi: 10.1038/nsmb.2372. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Krueger F., Andrews S.R. Bismark: a flexible aligner and methylation caller for Bisulfite-Seq applications. Bioinformatics. 2011;27:1571–1572. doi: 10.1093/bioinformatics/btr167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Lee S.H., Lumelsky N., Studer L., Auerbach J.M., McKay R.D. Efficient generation of midbrain and hindbrain neurons from mouse embryonic stem cells. Nat. Biotechnol. 2000;18:675–679. doi: 10.1038/76536. [DOI] [PubMed] [Google Scholar]
  22. Li X., Yao B., Chen L., Kang Y., Li Y., Cheng Y., Li L., Lin L., Wang Z., Wang M., et al. Ten-eleven translocation 2 interacts with forkhead box O3 and regulates adult neurogenesis. Nat. Commun. 2017;8 doi: 10.1038/ncomms15903. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Lister R., Mukamel E.A., Nery J.R., Urich M., Puddifoot C.A., Johnson N.D., Lucero J., Huang Y., Dwork A.J., Schultz M.D., et al. Global epigenomic reconfiguration during mammalian brain development. Science. 2013;341 doi: 10.1126/science.1237905. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Liu M.Y., Torabifard H., Crawford D.J., DeNizio J.E., Cao X.J., Garcia B.A., Cisneros G.A., Kohli R.M. Mutations along a TET2 active site scaffold stall oxidation at 5-hydroxymethylcytosine. Nat. Chem. Biol. 2017;13:181–187. doi: 10.1038/nchembio.2250. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Love M.I., Huber W., Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15:550. doi: 10.1186/s13059-014-0550-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. MacArthur I.C., Dawlaty M.M. TET Enzymes and 5-Hydroxymethylcytosine in Neural Progenitor Cell Biology and Neurodevelopment. Front. Cell Dev. Biol. 2021;9 doi: 10.3389/fcell.2021.645335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. MacArthur I.C., Ma L., Huang C.Y., Bhavsar H., Suzuki M., Dawlaty M.M. Developmental DNA demethylation is a determinant of neural stem cell identity and gliogenic competence. Sci. Adv. 2024;10 doi: 10.1126/sciadv.ado5424. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mellen M., Ayata P., Dewell S., Kriaucionis S., Heintz N. MeCP2 binds to 5hmC enriched within active genes and accessible chromatin in the nervous system. Cell. 2012;151:1417–1430. doi: 10.1016/j.cell.2012.11.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Montalban-Loro R., Lozano-Urena A., Ito M., Krueger C., Reik W., Ferguson-Smith A.C., Ferron S.R. TET3 prevents terminal differentiation of adult NSCs by a non-catalytic action at Snrpn. Nat. Commun. 2019;10:1726. doi: 10.1038/s41467-019-09665-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Okabe S., Forsberg-Nilsson K., Spiro A.C., Segal M., McKay R.D. Development of neuronal precursor cells and functional postmitotic neurons from embryonic stem cells in vitro. Mech. Dev. 1996;59:89–102. doi: 10.1016/0925-4773(96)00572-2. [DOI] [PubMed] [Google Scholar]
  31. Onodera A., González-Avalos E., Lio C.W.J., Georges R.O., Bellacosa A., Nakayama T., Rao A. Roles of TET and TDG in DNA demethylation in proliferating and non-proliferating immune cells. Genome Biol. 2021;22:186. doi: 10.1186/s13059-021-02384-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Parasyraki E., Mallick M., Hatch V., Vastolo V., Musheev M.U., Karaulanov E., Gopanenko A., Moxon S., Méndez-Lago M., Han D., et al. 5-Formylcytosine is an activating epigenetic mark for RNA Pol III during zygotic reprogramming. Cell. 2024;187:6088–6103.e18. doi: 10.1016/j.cell.2024.08.011. [DOI] [PubMed] [Google Scholar]
  33. Pastor W.A., Aravind L., Rao A. TETonic shift: biological roles of TET proteins in DNA demethylation and transcription. Nat. Rev. Mol. Cell Biol. 2013;14:341–356. doi: 10.1038/nrm3589. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Prasasya R.D., Caldwell B.A., Liu Z., Wu S., Leu N.A., Fowler J.M., Cincotta S.A., Laird D.J., Kohli R.M., Bartolomei M.S. Iterative oxidation by TET1 is required for reprogramming of imprinting control regions and patterning of mouse sperm hypomethylated regions. Dev. Cell. 2024;59:1010–1027.e8. doi: 10.1016/j.devcel.2024.02.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Ramirez F., Ryan D.P., Gruning B., Bhardwaj V., Kilpert F., Richter A.S., Heyne S., Dundar F., Manke T. deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res. 2016;44:W160–W165. doi: 10.1093/nar/gkw257. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Rudenko A., Dawlaty M.M., Seo J., Cheng A.W., Meng J., Le T., Faull K.F., Jaenisch R., Tsai L.H. Tet1 is critical for neuronal activity-regulated gene expression and memory extinction. Neuron. 2013;79:1109–1122. doi: 10.1016/j.neuron.2013.08.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Ruzov A., Tsenkina Y., Serio A., Dudnakova T., Fletcher J., Bai Y., Chebotareva T., Pells S., Hannoun Z., Sullivan G., et al. Lineage-specific distribution of high levels of genomic 5-hydroxymethylcytosine in mammalian development. Cell Res. 2011;21:1332–1342. doi: 10.1038/cr.2011.113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Schomacher L., Han D., Musheev M.U., Arab K., Kienhöfer S., von Seggern A., Niehrs C. Neil DNA glycosylases promote substrate turnover by Tdg during DNA demethylation. Nat. Struct. Mol. Biol. 2016;23:116–124. doi: 10.1038/nsmb.3151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Sherman B.T., Hao M., Qiu J., Jiao X., Baseler M.W., Lane H.C., Imamichi T., Chang W. DAVID: a web server for functional enrichment analysis and functional annotation of gene lists (2021 update) Nucleic Acids Res. 2022;50:W216–W221. doi: 10.1093/nar/gkac194. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Smith Z.D., Meissner A. DNA methylation: roles in mammalian development. Nat. Rev. Genet. 2013;14:204–220. doi: 10.1038/nrg3354. [DOI] [PubMed] [Google Scholar]
  41. Song Q., Decato B., Hong E.E., Zhou M., Fang F., Qu J., Garvin T., Kessler M., Zhou J., Smith A.D. A reference methylome database and analysis pipeline to facilitate integrative and comparative epigenomics. PLoS One. 2013;8 doi: 10.1371/journal.pone.0081148. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Szulwach K.E., Li X., Li Y., Song C.X., Wu H., Dai Q., Irier H., Upadhyay A.K., Gearing M., Levey A.I., et al. 5-hmC-mediated epigenetic dynamics during postnatal neurodevelopment and aging. Nat. Neurosci. 2011;14:1607–1616. doi: 10.1038/nn.2959. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Tahiliani M., Koh K.P., Shen Y., Pastor W.A., Bandukwala H., Brudno Y., Agarwal S., Iyer L.M., Liu D.R., Aravind L., Rao A. Conversion of 5-methylcytosine to 5-hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science. 2009;324:930–935. doi: 10.1126/science.1170116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Torabifard H., Cisneros G.A. Insight into wild-type and T1372E TET2-mediated 5hmC oxidation using ab initio QM/MM calculations. Chem. Sci. 2018;9:8433–8445. doi: 10.1039/c8sc02961j. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Wang H., Yang H., Shivalila C.S., Dawlaty M.M., Cheng A.W., Zhang F., Jaenisch R. One-step generation of mice carrying mutations in multiple genes by CRISPR/Cas-mediated genome engineering. Cell. 2013;153:910–918. doi: 10.1016/j.cell.2013.04.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Wu S.C., Zhang Y. Active DNA demethylation: many roads lead to Rome. Nat. Rev. Mol. Cell Biol. 2010;11:607–620. doi: 10.1038/nrm2950. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Yildirim O., Li R., Hung J.H., Chen P.B., Dong X., Ee L.S., Weng Z., Rando O.J., Fazzio T.G. Mbd3/NURD complex regulates expression of 5-hydroxymethylcytosine marked genes in embryonic stem cells. Cell. 2011;147:1498–1510. doi: 10.1016/j.cell.2011.11.054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Yu G., Wang L.G., He Q.Y. ChIPseeker: an R/Bioconductor package for ChIP peak annotation, comparison and visualization. Bioinformatics. 2015;31:2382–2383. doi: 10.1093/bioinformatics/btv145. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Figures S1–S6 and Tables S1 and S2
mmc1.pdf (3MB, pdf)
Document S2. Article plus supplemental information
mmc2.pdf (36.4MB, pdf)

Data Availability Statement

The RNA-seq, EM-seq, EhmC-seq, and CUT&Tag data have been deposited in the Gene Expression Omnibus database accession (GSE287657) and are publicly available as of the date of publication.

Articles from Stem Cell Reports are provided here courtesy of Elsevier

RESOURCES