Folic Acid Remodels Chromatin on Hes1 and Neurog2 Promoters during Caudal Neural Tube Development

Abstract

The mechanism(s) behind folate rescue of neural tube closure are not well understood. In this study we show that maternal intake of folate prior to conception reverses the proliferation potential of neural crest stem cells in homozygous Splotch embryos (Sp^−/−) via epigenetic mechanisms. It is also shown that the pattern of differentiation seen in these cells is similar to wild-type (WT). Cells from open caudal neural tubes of Sp^−/− embryos exhibit increased H3K27 methylation and decreased expression of KDM6B possibly due to up-regulation of KDM6B targeting micro-RNAs such as miR-138, miR-148a, miR-185, and miR-339-5p. In our model, folate reversed these epigenetic marks in folate-rescued Sp^−/− embryos. Using tissue from caudal neural tubes of murine embryos we also examined H3K27me2 and KDM6B association with Hes1 and Neurog2 promoters at embryonic day E10.5, the proliferative stage, and E12.5, when neural differentiation begins. In Sp^−/− embryos compared with WT, levels of H3K27me2 associated with the Hes1 promoter were increased at E10.5, and levels associated with the Neurog2 promoter were increased at E12.5. KDM6B association with Hes1 and Neurog2 promoters was inversely related to H3K27me2 levels. These epigenetic changes were reversed in folate-rescued Sp^−/− embryos. Thus, one of the mechanisms by which folate may rescue the Sp^−/− phenotype is by increasing the expression of KDM6B, which in turn decreases H3K27 methylation marks on Hes1 and Neurog2 promoters thereby affecting gene transcription.

Introduction

Maternal folic acid (FA)² supplementation prior to conception and during neural tube (NT) development rescues neural tube closure defects in 60% of human cases. Studies to elucidate mechanisms in FA-dependent neural tube defect (NTD) prevention have been done in NTD-prone mutant mouse models including Splotch, which has a loss of function mutation in the Pax3 gene (1–3). Pax3 mutation leads to deficits in neural crest-derived cells, with homozygous Splotch (Sp^−/−) mice exhibiting exencephaly or spina bifida (4, 5). Sp^−/− embryos have an impaired ability to synthesize thymidylate. Addition of either FA or thymidine prevents NTDs and restores metabolic blockade in Sp^−/− embryos. Addition of methionine induces NTDs in 50% of Sp^+/− embryos, which normally do not exhibit this phenotype (1). Thus, rescue of Sp^−/− mutants from NTDs is associated with that segment of the folate/homocysteine metabolic pathway that supports nucleotide biosynthesis. In contrast, increasing methylation exacerbates embryopathy (3). These observations raise the question, “does FA affect chromatin structures on key developmental genes?”

Epigenetic processes are at the center of stem cell function and neural tube development (6). Currently, there are no studies examining FA effects on chromatin remodeling during neural tube development. In this study, we used a FA responsive mouse model, Splotch (Sp^−/−), to examine this hypothesis. Histone modifications, such as acetylation and methylation, are epigenetic marks that regulate diverse biological processes. It has been reported that demethylation of lysine residue 27 of histone H3 (H3K27) by enzymes UTX and KDM6B (formerly JMJD3) regulates animal posterior development (7). We have examined levels of H3K27 methylation, and the demethylase enzyme KDM6B, in wild-type (WT), Sp^−/−, and FA-rescued Sp^−/− embryos. H3K27me2 methylation increased in Sp^−/− embryos compared with WT. The increase was reversed by maternal supplementation of FA prior to conception, further strengthening the hypothesis that demethylation of histone H3K27me2 may play a crucial role in posterior neural tube development. Concomitantly, levels of the histone demethylase KDM6B decreased in Sp^−/− embryos and this decrease was reversed with FA supplementation.

One of the mechanisms for regulating KDM6B levels could be via micro-RNAs (miRNAs). MiRNAs act as post-transcriptional regulators of gene expression and are involved in neuronal differentiation (8, 9). Emerging evidence suggests that miRNAs also play prominent roles in human diseases (10). In this study we have identified miRNAs up-regulated in Sp^−/− embryos compared with WT. At least four of these (miR-138, miR-148a, miR-185, and miR-339-5p) have binding sites on the 3′ UTR of KDM6B. FA administration reversed this up-regulation, indicating that these miRNAs may be important in neural tube development.

Hes1 and Neurog2 are essential for neural tube development, and therefore likely candidates for epigenetic modifications by FA supplementation. Hes1 prevents premature neurogenesis (11) and Neurog2 is critical in sensory neurogenesis (12). Absence of functional Pax3 in Sp^−/− embryos impairs stem cell proliferation and causes premature neurogenesis by affecting Hes1 and Neurog2 promoter activity (13). Using tissue from caudal neural tube, we have examined the association of H3K27me2 and KDM6B with Hes1 and Neurog2 promoters at embryonic days E10.5, proliferative stage, and E12.5, when neural differentiation begins. In comparison to WT, levels of H3K27me2 associated with the Hes1 promoter increased in E10.5 Sp^−/− embryos, whereas levels associated with the Neurog2 promoter increased in E12.5 Sp^−/− embryos. KDM6B association with Hes1 and Neurog2 promoters was inversely correlated to H3K27me2 levels. These epigenetic changes were reversed in FA-rescued Sp^−/− embryos suggesting that one of the mechanisms by which FA rescues the Sp^−/− phenotype is by increasing KDM6B expression, which in turn decreases the expression of H3K27 methylation on Hes1 and Neurog2 promoters, thereby affecting gene transcription.

EXPERIMENTAL PROCEDURES

Animal Breeding, Genotyping, and FA Administration

C57BL/6J-Pax3^Sp/J male and female breeding pairs were obtained from The Jackson Laboratory. Embryos were obtained from heterozygote mating and genotyped by isolating genomic DNA from embryonic membranes and performing PCR with WT (Sp^+/+), heterozygous (Sp^+/−), and homozygous (Sp^−/−) specific reverse primers and a common forward primer as previously described (14). Administration of FA (25 mg/kg/day; F7876, Sigma) to the heterozygous female was initiated 2 weeks prior to the first attempted mating and continued until the female was sacrificed and embryos were obtained at E10.5. Normal closed neural tube phenotype in Sp^−/− embryos as ascertained by genotyping was considered as “rescued” by FA.

Immunostaining

FA-treated colonies (neurospheres) were stained for the neural stem cell marker Nestin (Millipore MAB353), and pluripotent stem cell markers CD133 (Abcam ab19898–100), Sox2 (Chemicon AB5603), Oct4 (Santa Cruz Biotechnology sc-8629), and alkaline phosphatase (Millipore SCR004). Medium without EGF and bFGF was used for differentiation. After differentiation, migrated cells were immunostained with antibodies for Brn3a (sensory neuron marker; Millipore AB5945), TuJ1 (neuronal marker; Covance MMS-435p), GFAP (astrocyte marker; Millipore AB5804), O4 (oligodendrocyte marker; Millipore MAB345), and α-SMA (smooth muscle actin; Abcam ab5694).

RNA Isolation

50 μl of total RNA was eluted from 5 mg (size: 2 mm × 2 mm × 2 mm) of WT, Sp^−/−, and folate-rescued Sp^−/− caudal neural tube tissue by using the mirVana^TM miRNA Isolation Kit (Ambion AM1560), as per the manufacturer's instructions. RNA concentration was measured by NanoDrop (Fischer Scientific).

Quantification of MiRNAs by TaqMan Low Density Arrays and Bioinformatic Analyses

Quantification of miRNAs by real time PCR was done with the TaqMan miRNA Assay System from Applied Biosystems as described (15, 16). Expression of miRNAs was characterized after RNA extraction with TRIzol® (Invitrogen), from caudal neural tubes of WT (n = 3), Splotch (Sp^−/−) with lumbar open NTD (n = 3), and Splotch NTD rescued by folic acid (n = 3) E10.5 embryos. The TaqMan® Rodent MicroRNA Array A version 2.0 (Applied Biosystems) was used to investigate miRNA expression. Each reverse transcriptase (RT) reaction contained purified total RNA and components from the TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems). Threshold cycle (C_t) values with more than 12 C_t cycles above the average value of the endogenous control (∼35 C_t), represented a very low copy number. These miRNAs were excluded from our analyses as previously described (17). Differences between this cutoff and the remaining C_t values were calculated; the values were normalized and comparisons were made between samples. The minimum miRNA C_t differences (ΔC_t) between the WT and experimental samples (three embryos of each genotype) were computed to identify miRNAs that were differentially expressed between groups. ∼1 cycle change in ΔC_t value is equivalent to a 2-fold change in gene expression. The data are represented as fold-change in miRNA expression ± S.E. of three separate experiments. Based on miRNA array results, targeted genes for each miRNA candidate, which showed tri-phased expression, i.e. low expression in WT, high in Sp^−/−, and reversed to Wt levels in folate-rescued Sp^−/− were validated using the following online algorithms: TargetScan4.2 and micro-RNA.

Preparation of cDNA and Quantitative Real Time PCR

Two-step RT-PCR was performed. First cDNA was synthesized from RNA and then real time PCR was done using 5 ng/well of cDNA and 4 μl of qScript cDNA SuperMix (5 times) in a total volume of 20 μl. A PCR cycle was run for 5 min at 25 °C, 30 min at 42 °C, and 5 min at 85 and 4 °C (Applied Biosystem 7500 Fast Real-Time Thermal Cycler). Each 20-μl RT-PCR included 1.25 μl of forward primer, 1.25 μl of reverse primer, 2.5 μl of water, 5 μl of cDNA (1 ng/μl), and 10 μl of the double-stranded DNA binding dye SYBR Green I to detect PCR product (PerfeCTa SYBR Green FastMix, Low ROX (95074-250) Quanta Biosciences). RT-PCR cycle parameters were 95 °C for 10 min followed by 40 cycles at 95 °C for 10 s, 58 °C for 5 s, and 72 °C for 32 s. The following murine primers were used: Hes1 forward, 5′-ggtcctaacgcagtgtcacctt-3′ and reverse, 5′-cagtggcctgaggctctca-3′; Neurog2 (accession number AF303001) forward, 5′-agaggtggcccttgcaatc-3′ and reverse, 5′-cacacgccatagtcctctttga-3′; β-actin forward, 5′-acggccaggtcatcactattg-3′ and reverse, 5′-tggatgccacaggattcca-3′; and KDM6B forward, 5′-tgcctactactgcaacgaatgc-3′ and reverse, 5′-cgtgcggtactgctctagca-3′. Primers were designed using Primer Express software (PerkinElmer Life Sciences) and synthesized by Eurofins MWG Operon. For RT-PCR we used reverse transcription and amplification product qScript cDNA SuperMix (95048-025) from Quanta Biosciences.

Extraction of Histones from Embryonic Caudal Neural Tube Tissue

Histone proteins were extracted from genotyped embryos using a Histone Purification Mini Kit (Active Motif, Inc.) and quantified by measuring absorbance at 230 nm. Equal amounts of histone protein were subjected to 15% SDS-PAGE, transferred to PVDF membrane, and probed with anti-dimethyl H3K27 antibody (ab24684, Abcam).

Neurosphere Generation

Neurospheres were grown from lower lumbar regions of E10.5 embryos as described earlier (13), using 5 × 10⁴ cells/ml in 4 ml of medium. A group of cells was counted on the 7th day of culture as a neurosphere when it was 50 μm or larger. Each data point represents one independently prepared culture derived from one embryonic caudal neural tube. For treatment groups FA (200 μg/ml) was added to the medium.

Neurosphere Immunostaining

After 7 days in culture, neurospheres were transferred to 8-well chamber slides pre-coated with laminin (Sigma L2020) and fixed in 4% paraformaldehyde in PBS. Alkaline Phosphatase Detection Kit (Millipore, SCR004) was used to detect pluripotency, as per the manufacturer's protocol. Cells were also immunostained with antibodies for pluripotency markers, i.e. Sox2, Oct3/4, CD133, and neural progenitor marker Nestin after pre-blocking with 10% normal donkey serum, 0.1% Triton X-100, 0.01% sodium azide in PBS. Secondary antibodies included Cy3-conjugated donkey anti-rabbit and anti-mouse IgG (Jackson ImmunoResearch) and Alexa Fluor 488-labeled donkey anti-rabbit and goat IgG (Invitrogen). Nuclear staining was done using DAPI.

EdU Incorporation

To measure the ability of cells in the neurospheres to proliferate, neurospheres taken at 7 days in culture were subjected to EdU incorporation using the Click-iT EdU Imaging Kit (Invitrogen). The signal was detected with a Leica DMIRB Inverted Microscope.

Neural Tube Explant Cultures

Embryos were dissected at E10.5 (30 somite) with closed caudal neural tubes of WT and open caudal neural tube of Sp^−/−. Lower lumbar portions were plated and cultured on Matrigel® (BD Biosciences) in Neurobasal medium (NB Plus) with EGF (20 ng/ml) and bFGF (20 ng/ml). For treatment, FA (200 μg/ml) was added to the culture medium of Sp^−/− embryos. Neural crest cells were allowed to migrate for 48 h. Medium without EGF and bFGF was used for differentiation. One week after differentiation, migrated neural crest cells were immunostained with antibodies (Abcam) for H3K27me2, H3K27me3, and KDM6B.

Pre-miRNA Precursor Transfection, Western Blots, and Immunofluorescence

MiRNA knockdown experiments were performed in ND7 cells via transfection with precursors for the following micro-RNAs: pre-miR-339-5p, pre-miR-185, pre-miR-148a, and pre-miR-138 in various combinations or a pre-miR negative control using siPORT NeoFX transfection reagent, as described by Ambion. Concentrations of 10 to 30 nm of each pre-miRNA precursor gave the same results, hence 10 nm of each pre-miRNA precursor and negative controls were used in all experiments. ND7 cells (a generous gift from Dr. M. Calissano, Institute of Child Health London) were obtained by fusing neonatal rat (E12) dorsal root ganglion neurons with mouse neuroblastoma cells (18). Proliferating ND7 cells were cultured in DMEM containing 10% FBS and 10% donor horse serum. Forty-eight h after transfection, cells were harvested or fixed in 4% paraformaldehyde. KDM6B protein expression was assayed by Western blot analysis and immunofluorescence. For Western analysis the primary antibody was KDM6B rabbit polyclonal antibody (1:400, Abcam). The secondary antibody was anti-rabbit HRP-conjugated antibody (1:5000, Santa Cruz). For immunostaining, KDM6B rabbit polyclonal antibody was diluted 1:100 and incubated at 4 °C overnight. Cells were incubated with a donkey anti-rabbit IgG-Cy3 (1:100), secondary antibody (Jackson ImmunoResearch) for 1 h at room temperature, and counterstained with DAPI (20 ng/ml).

KDM6B 3′ UTR-luciferase Reporter Assays

Single strand oligos with or without putative miRNA binding sequences for miR-138, miR-148a, miR-185, and miR-339-5p were synthesized as described under the supplemental data. The oligos were designed using online algorithms TargetScan4.2 and micro-RNA (supplemental Fig. S2). Oligos were annealed to make double-stranded constructs and ligated with T4 kinase into SpeI/HindIII sites of the pMIR-reporter vector (Ambion), which contains a multiple cloning site on the 3′ UTR of the luciferase gene for insertion of predicted miRNA binding targets. The 8 constructs produced included (a) four KDM6B 3′ UTR-luciferase reporter constructs with intact miRNA binding sites: pMIR-138, pMIR-148a, pMIR-185, and pMIR-339-5p; and (b) four KDM6B 3′ UTR-luciferase reporter constructs in which the sequence for the miRNA binding sites was altered by deletion: pMIR-138c, pMIR-148ac, pMIR-185c, and pMIR-339-5pc. The ligated vectors were sequenced to confirm that they had the proper insertion. KDM6B 3′ UTR-luciferase reporter constructs and pre-miRNA precursors were co-transfected into ND7 cells using the MegaTran 1.0 Transfection Reagent (OriGene). Renilla luciferase plasmid, pRL-null (5 ng/well), was used as an insertional control for transfection efficiency. Pre-miRNA precursor negative control (Ambion, AM17111) was used for a negative control transfection. The final concentration of each pre-miR precursor was 10 nm. Luciferase assays were performed using the Dual Luciferase kit (Promega).

Down-regulation of KDM6B with shRNA

Single cells were prepared by triturating WT neurospheres at day 7 and transfecting the cells with HuSH-29 shRNA constructs against KDM6B in pRFP-C-RS vector (OriGene) by MegaTran transfection reagent (OriGene). Scrambled non-effective shRNA (TR30015, OriGene) in pRFP-C-RS plasmid was used as a negative control. Cells were incubated for 72 h with KDM6B and H3K27me3 antibodies. Secondary antibodies conjugated with Alexa Fluor 488 (Green) distinguished the red fluorescent protein (RFP) signal of shRNA.

Chromatin Immunoprecipitation (ChIP)

ChIP assays using neural progenitors were performed as described by Abcam (X-ChIP protocol). Rabbit polyclonal antibodies were used: acetylated H3K9 and H3K18 (9671 and 9675, Cell Signaling Technology), dimethylated H3K27 (ab24684, Abcam), UTX (KDM6A) (ab36938, Abcam), and JMJD3 (KDM6B) (ab38113, Abcam). PCR was performed with murine Hes1 and Neurog2 primers: Hes1 forward (bp 98–118) 5′-ttggctgaaagttactgtgg-3′ and reverse (bp 279–299) 5′-tcttagggctacttagtgat-3′, and Ngn2 forward (bp 6881–6900) 5′-ggcagatctgattgttttct-3′ and reverse (bp 7061–7080) 5′-gcagctcgcccgagtctcgt-3′. All ChIP samples were tested for false-positive PCR amplification by sequencing the 200-bp amplified product to ascertain the specificity of binding to cis-regulatory elements by histones or histone modifying enzymes. False-positive PCR amplification was ruled out by amplifying the sequence from the murine β-actin gene; forward primer, 5′-acggccaggtcatcactattg-3′, and reverse primer, 5′-tggatgccacaggattcca-3′.

Statistical Analysis

p values were determined by Student's t test using GraphPad Prism version 5.0.

RESULTS

FA Rescued Neural Stem Cells from Splotch Homozygous Embryos Show Proliferation and Differentiation

Neural precursors from the caudal region of Sp^−/− embryos generally do not form neurospheres in culture. When they do, the spheres are small (<50 μm in size) and do not survive beyond 7 days (13). To test the hypothesis that FA rescues neural stem cell proliferation in the caudal neural tube of Sp^−/− embryos, neural precursors were grown from Sp^−/− embryos for up to 7 days in FA-supplemented Neurobasal medium containing FGF and bFGF (Fig. 1, A and B). The results showed that FA rescued neurosphere formation in cells taken from the caudal neural tube of Sp^−/− embryos. EdU incorporation experiments (Fig. 1C) indicated that FA rescued cell proliferation, although aggregation and fusion of proliferating neurospheres cannot be completely ruled out. FA-rescued neurospheres that stained positive for stem cell markers were: CD133, Oct4, Sox2, and alkaline phosphatase and the neural progenitor marker nestin (Fig. 2A), suggesting that the rescued cells are neural stem cells similar to WT cells not treated with FA. To determine whether FA-rescued neural progenitors from Sp^−/− embryos can differentiate into cells of different lineages, neurospheres were allowed to differentiate on growth factor-depleted Matrigel in Neurobasal medium in the absence of EGF and bFGF for 7 days. Upon differentiation, the cells were immunostained with TuJ1 (neuron marker), O4 (oligodendrocyte marker), GFAP (astrocytes marker), α-SMA (smooth muscle marker), and Brn3a (sensory neuron marker). The number of immuno-positive cells per overall cell count was highest for GFAP. TuJ1, O4, and α-SMA immunoreactivity in WT were less than 10% (Fig. 2B). Sp^−/− cells treated with FA showed a similar pattern of differentiation as WT. A subset of TuJ1 positive cells were also Brn3a positive (supplemental Fig. S1a), suggesting that the sensory neurogenesis was also rescued in Sp^−/− embryos by FA treatment. Differentiated cells were also stained with pluripotency markers; both WT- and FA-treated Sp^−/− cells showed some staining for Sox2 and nestin (supplemental Fig. S1b). This may reflect that some cells retain pluripotency stem cell and neural progenitor markers. Differentiation studies were not done on Sp^−/− cells not treated with FA, because in the absence of FA, Sp^−/− cells form fewer neurospheres and more critically these neurospheres die within a week in culture. Overall this data indicates that FA rescues stem cell proliferation potential and that the rescued cells differentiate similar to WT.

FIGURE 1.

FA rescues proliferation and differentiation potential of neural progenitors from Sp^−/− embryos. A, neurospheres were grown from closed neural tubes of WT, open neural tubes of Sp^−/−, and open neural tubes of Sp^−/− with 200 μg/ml of FA (E 10.5) in 4 ml of media containing EGF and bFGF. To initiate the colonies 5 × 10⁴ cells/ml were used. A group of cells was considered to be a neurosphere at 50 μm or larger. Neurospheres were counted on day 7. B, each data point represents one (independently prepared) culture and each culture was derived from one embryonic caudal neural tube. In the absence of FA, colony forming units were significantly lower from Sp^−/− embryos (*, p < 0.05) compared with WT. FA significantly increased the number of neurospheres in Sp^−/− but not the WT (**, p < 0.01) sphere culture. C, EdU incorporation into the neurospheres indicates that FA treatment reversed the reduced number of neurospheres from Sp^−/− mutants to near WT levels via cell proliferation. Experiments were performed in triplicate with each data point in duplicate.

FIGURE 2.

Sp^−/− progenitor stem cells treated with folate exhibit pluripotency and differentiation. A, folate-rescued neurospheres from Sp^−/− embryos were stained for Nestin, CD133, Sox2, Oct4, and alkaline phosphatase (n = 8). These neurospheres demonstrated stem cell pluripotency, as seen in neurospheres from WT embryos. B, neurospheres from WT and FA-treated Sp^−/− embryos were allowed to differentiate for 7 days in the absence of growth factors and immunostained for TuJ1, GFAP, O4, and α-SMA. DAPI was used as a nuclear stain. Stained cells were counted from 5 regions per well (n = 4 wells per antibody) and data were expressed as percentage of the average of DAPI positive cells for each stain. Cells in both groups were capable of differentiation. The insets in TuJ1, O4, and SMA staining shows an enlarged differentiated cell in WT and FA-treated Sp^−/− cells.

FA Reverses H3K27me2 Methylation and Expression of KDM6B in FA-rescued Sp^−/− Embryos as well as FA-treated Sp^−/− Explants

H3K27 demethylation by KDM6B is important in animal posterior development (7). We hypothesized that in Sp^−/− embryos, which show decreased stem cell proliferation, H3K27 methylation would increase and FA would reverse this increase. To test this, caudal neural tubes from E10.5 WT, Sp^−/−, and FA-treated Sp^−/− embryos were used for explant cultures as described earlier (13). Upon differentiation, neural crest cells that migrated from the explants were immunostained with H3K27me2 and H3K27me3 antibodies. H3K27 methylation levels increased in Sp^−/− embryos compared with WT. This increase was reversed by FA addition to the culture medium (Fig. 3A). To confirm these results, histones were acid extracted from caudal neural tubes of WT, Sp^−/−, and FA-rescued Sp^−/− embryos and immunoblotted with H3K27me2 antibody. Fig. 3B shows H3K27me2 up-regulation in Sp^−/− embryos, and reversal of this up-regulation in FA-rescued Sp^−/− embryos.

FIGURE 3.

H3K27 methylation and KDM6B expression in FA-treated Sp^−/− embryos. A, neural tube explants from E10.5 WT and Sp^−/− with/without FA treatment were initially grown in the presence of EGF and bFGF. After 140 h neural crest cells were allowed to differentiate in the absence of growth factors for 48 h. Differentiated cells were stained with DAPI, H3K27me2, H3K27me3, and KDM6B antibodies. Cells that stained positive for each antibody from five different explant cultures were counted and expressed as a percentage of DAPI positive cells. H3K27me2 (p < 0.001) and H3K27me3 (p < 0.05) staining was significantly increased and KDM6B staining was significantly decreased (p < 0.001) in Sp^−/− embryo neural tubes explants. B, H3K27me2 immunoblots of acid-extracted histones from neural tubes of WT, Sp^−/−, and FA-rescued Sp^−/− embryos (E10.5). Protein loading was ascertained with Ponseau S staining; C, murine KDM6B expression was analyzed by real time RT-PCR. The data shows ΔΔC_t values for KDM6B transcript levels in Sp^−/− and FA-rescued Sp^−/− embryos compared with WT littermates. Data were normalized to β-actin. ΔΔC_t values were obtained by subtracting the C_t value of WT KDM6B from Sp^−/− or FA-rescued Sp^−/− embryos after β-actin normalization. Experiments were performed in quadruplicate with each data point in duplicate.

To test whether decreased KDM6B expression is involved in increased H3K27 methylation in Sp^−/− embryos, migrating cells from explants were immunostained with KDM6B antibody. Decreased KDM6B immunostaining was observed in Sp^−/− neural tube explants compared with WT. This decrease was reversed to near WT levels in FA-rescued Sp^−/− embryos (Fig. 3A). Quantitative real time RT-PCR with total RNA isolated from caudal neural tubes of Sp^−/− embryos showed decreased KDM6B transcript levels compared with WT. Sp^−/− embryos rescued by maternal intake of FA showed a 3-fold increase in KDM6B transcript levels, when compared with WT (Fig. 3C). Thus the increase in H3K27 methylation in Sp^−/− embryos could be due to decreased KDM6B expression in agreement with earlier reports (7). Overall neural tube explant cultures from FA-rescued embryos reversed H3K27me2/3 methylation patterns seen in Sp^−/− embryos back to near wild type levels. Similarly decreased KDM6B levels seen in Sp^−/− embryos were reversed in Sp^−/− NTD embryos rescued by maternal intake of FA.

FA Reverses Levels of KDM6B Targeting miRNAs That Are Up-regulated in Sp^−/− Embryos

To test the hypothesis that miRNAs play a role in regulating KDM6B levels we performed miRNA screening using total RNA isolated from caudal neural tubes of WT, Sp^−/−, and FA-rescued Sp^−/− embryos. Increased expression of several miRNAs was observed in Sp^−/− embryos compared with WT (supplemental Tables S1a and S1b). We searched for miRNA binding sequences on KDM6B by bioinformatics. Using TargetScan and microRNA.org online programs, we identified four miRNAs, miR-138, miR-148a, miR-185, and miR-339-5p, with predicted binding sequences to KDM6B 3′ UTR (supplemental Fig. S2). Pre-miRNAs were transiently transfected into ND7 cells individually or in combinations of two, three, or four. Two negative controls, no pre-miRNA and non-targeting negative control pre-miRNA from Ambion, were also used (Fig. 4). Immunoblot data showed that individual pre-miRNAs, pre-miR-339-5p, and pre-miR-138 minimally (∼15%) silenced KDM6B (Fig. 4A). Combinations of two pre-miRNAs, notably pre-miR-339-5p + pre-miR-185, pre-miR-339-5p + pre-miR148a, and pre-miR-339-5p + pre-miR-138 were effective in silencing KDM6B expression more than 50% (Fig. 4B). Combinations of pre-miR-339-5p + pre-miR-185 along with either pre-miR-138 or pre-miR-148 were the most effective, silencing KDM6B expression over 90% (Fig. 4C). Immunochemical analysis of ND7 cells transfected with pre-micro-RNAs and immunostained with KDM6B antibody (supplemental Figs. S3 and S4) confirmed the immunoblot data.

FIGURE 4.

KDM6B is targeted by miR-148a, miR-185, and miR-339-5p. Panels A–C represent single pre-miRNAs, combinations of two pre-miRNAs, and combinations of three or all four pre-miRNAs, transfected into ND7 cells. Two negative controls, a no pre-miRNA transfection control and a non-targeting negative control miRNA from Ambion, were used. GAPDH was the protein loading control. Experiments were performed in quadruplicate with each data point in duplicate. Single miRNAs did not silence KDM6B (A). Combinations of two miRNAs that were most effective in silencing KDM6B were miR-339-5p + miR-185 and miR-339–5p + miR-148a (B). For >90% silencing of KDM6B, combinations of at least 3 miRNAs were necessary: miR-339-5p (red dot) + miR-185 (red dot) plus either miR-148a (green dot) or miR-138 (green dot) (C). Experiments were performed in quadruplicate with each data point in duplicate.

To further confirm the effect of miR-138, miR-148a, miR-185, and miR339-5p in silencing KDM6B, we co-transfected pMIR luciferase reporter vector (pMIR-REPORT^TM System, Ambion) with intact (Fig. 5, A and B) or mutated (Fig. 5, D and E) miRNA binding sites on KDM6B 3′ UTR constructs and appropriate cognate pre-miRNA precursors into ND7 cells. The results showed that miRNA-148a silenced 3′ UTR-luciferase activity most effectively. MiR-138, miR-185, and miR-339-5p mediated silencing of 3′ UTR-luciferase activity between 25 and 35%. Constructs where the miRNA binding sequence was absent showed ∼15–20% reduction in activity. This could be due to nonspecific off target silencing by these miRNAs (Fig. 5, D and E).

FIGURE 5.

KDM6B 3′ UTR is targeted by miRNAs. KDM6B 3′ UTR-luciferase constructs (pMIR) including targeted binding sites for miR-138, miR-148a, miR-185, and miR-339-5p (A and B), and KDM6B 3′ UTR-luciferase constructs, lacking the miRNA binding sites, pMIRc, were made (D and E). These constructs were co-transfected with cognate pre-miRNA precursors into ND7 cells. pRL null (5 ng/well) was used as a transfection control. Experiments were performed in quadruplicate with each data point in duplicate. After 48 h luciferase activity was assayed using the Dual Luciferase kit from Promega in a 96-well format employing the Vector2 (PerkinElmer Life Sciences) chemiluminescence detector system. Ambion negative control 2, which does not have a miRNA binding sites was used as a control (C). Experiments were performed in triplicate with each data point in duplicate.

KDM6B Directly Regulates Histone H3K27 Methylation

To determine whether H3K27 methylation is directly regulated by KDM6B activity, we transfected the shRNA-RFP construct against KDM6B (OriGene) into the single cells triturated from neurosphere cultures from WT embryos (E10.5). KDM6B shRNA-RFP silenced KDM6B levels almost 100% in cells that showed red fluorescence (see cell indicated by arrow in Fig. 6B), and also increased H3K27 methylation (see cell indicated by arrow in Fig. 6D) as compared with adjacent cells not transfected with the shRNA-RFP construct. A scrambled negative control-shRNA-RFP from OriGene (TR30015) did not silence KDM6B levels (Fig. 6A) or increase H3K27 methylation (Fig. 6C). These results clearly demonstrate that in cultured neurospheres H3K27 methylation is directly regulated by KDM6B.

FIGURE 6.

KDM6B directly regulates histone H3K27 methylation in WT cultured neurospheres. A, transfection of scrambled negative control shRNA-RFP (TR30015, OriGene) into cells from WT neurospheres and immunostaining for KDM6B; B, transfection with KDM6B shRNA-RFP and immunostaining for KDM6B; C, transfection of scrambled negative control shRNA-RFP and immunostaining for H3K27me3; D, transfection with KDM6B shRNA-RFP and immunostaining for H3K27me3. Experiments were performed in quadruplicate. A representative of 4 separate experiments is shown. These results demonstrate that H3K27 methylation is directly regulated by KDM6B in cultured neurospheres.

FA Reverses Hes1 and Neurog2 Expression in FA-rescued Sp^−/− Embryos and Sp^−/− Embryos Show Premature Neurogenesis

Hes1 and Neurog2 are necessary for stem cell proliferation/maintenance and neural differentiation, respectively. Expression of these genes is decreased in the caudal region of Sp^−/− embryos compared with WT (13). Quantitative RT-PCR was used to determine whether FA rescues Hes1 and Neurog2 expression. Results from RNA isolated from E10.5 caudal neural tubes of WT, Sp^−/−, and FA-rescued Sp^−/− embryos showed that FA supplementation reversed expression levels of Hes1 and Neurog2 transcripts in FA-rescued Sp^−/− embryos compared with WT levels (Fig. 7). This data suggest that FA may cause epigenetic modifications at Hes1 and Neurog2 promoters.

FIGURE 7.

Quantitative real time RT-PCR of Hes1 and Neurog2 from E10.5 embryos. Expression of murine Hes1 and Neurog2 was analyzed by RT-PCR. The data (n = 4; mean ± S.E.) shows ΔΔC_t values for Hes1 and Neurog2 transcript levels in Sp^−/− and FA-rescued Sp^−/− embryos compared with WT littermates. Data were normalized to β-actin. ΔΔC_t values are obtained by subtracting the C_t value for the WT gene from Sp^−/− and FA-rescued Sp^−/− embryos after normalization. Experiments were performed in quadruplicate with each data point in duplicate.

FA Remodels Chromatin Structures on Hes1 and Neurog2 Promoters during Caudal Neural Tube Development

To ascertain if FA restores stem cell proliferation and neurogenesis via epigenetic mechanisms, we examined the association between acetylated H3K9 and H3K18, methylated H3K27, and the H3K27 demethylases KDM6B and UTX with Hes1 and Neurog2 promoters. Fig. 8 shows chromatin immunoprecipitation assays (ChIP) using neurospheres from caudal neural tubes from WT, Sp^−/−, and Sp^−/− embryos treated with FA at E10.5, the proliferative stage, and E12.5, the stage at which differentiation begins.

FIGURE 8.

ChIP assays showing chromatin remodeling on Hes1 and Neurog2 promoters. Individual ChIP assays were performed using neurospheres from caudal neural tube portions of WT, Sp^−/−, and Sp^−/−-treated with FA (E10.5 and E12.5). IgG was used as an IP negative control. β-Actin primers were used as negative controls. Amplified product was present only in the input and not in the control IgG or the immunoprecipitate. Immunoprecipitated DNA was subjected to quantitative PCR using murine primers for Hes1 (A) and Neurog2 (B) promoters. The data represents fold-enrichment of immunoprecipitated DNA compared with the input sample. FA rescued chromatin marks on Hes1 at E10.5, and on Neurog2 at E12.5. Each ChIP experiment was performed in triplicate using one lumbar neural tube region per ChIP assay with a total n = 4.

Hes1

At E10.5 H3K9Ac and H3K18Ac associated with the Hes1 promoter in WT embryos, consistent with the transcriptionally active histone code and stem cell maintenance function of Hes1 during this stage of embryogenesis (Fig. 8A). By E12.5, there was a decline in H3K9Ac and no significant change in H3K18Ac association with Hes1. H3K27me2 was not associated with the Hes1 promoter in E10.5 WT embryos, but was associated with the promoter at E12.5, indicating a gradual transition from transcriptionally active to repressive chromatin as development progresses. Consistent with levels of H3K27me2 on the promoter, association of histone demethylases KDM6B and UTX (KDM6A) was inversely co-related.

Chromatin marks on the Hes1 promoter in Sp^−/− embryos were opposite to those observed in WT. At E10.5, H3K9 and H3K18 acetylation significantly decreased, H3K27 methylation increased, and KDM6B and UTX (KDM6A) association on Hes1 promoter decreased. Overall, there appears to be a repressive chromatin structure on this promoter in E10.5 Sp^−/− embryos, consistent with previous results showing decreased Hes1 transcript levels in Sp^−/− embryos (13). This supports a decrease in cell proliferation and early neurogenesis in Sp^−/− embryos. Kobayashi et al. (19) has suggested that decreased Hes1 activity results in increased neurogenesis. At E12.5, continued association of H3K27me2 (although a drop from 4.5 to 2-fold) and a decline in the association of H3K9Ac and H3K18Ac with Hes1 indicates a repressive chromatin structure. FA treatment rescued levels of H3K27me2 and KDM6B, but not H3K9 or H3K18 acetylation, on the Hes1 promoter in E10.5 Sp^−/− neural tube progenitors. These observations suggest that FA works via removal of methyl groups from H3K27me2 by increasing the levels of KDM6B associated with the Hes1 promoter.

Neurog2

In E10.5 WT embryos, the Neurog2 promoter is associated with H3K9Ac, H3K18Ac, and H3K27me2 (Fig. 8B). The association of transcriptionally active and inactive components of chromatin simultaneously on the Neurog2 promoter is consistent with the hypothesis that stem cells are held transcriptionally inactive by H3K27me2 just prior to differentiation (20). KDM6B and UTX did not associate with this promoter. This is in accordance with the need for the cell to keep the proneural gene Neurog2 transcriptionally inactive until differentiation. H3K9Ac is not associated with the Neurog2 promoter in Sp^−/− embryos, although there is a significant association of H3K18Ac. Small amounts of H3K27me2, KDM6B, and UTX were associated with the Neurog2 promoter from E10.5 Sp^−/− embryos, suggesting that the dynamic balance between acetylation and methylation is altered in Sp^−/− embryos, thereby signaling the cells to differentiate prematurely. FA did not rescue any chromatin marks on the Neurog2 promoter at E10.5.

In E12.5 WT embryos, H3K9Ac and H3K18Ac association with the promoter increased, whereas H3K27me2 decreased, suggesting transcriptionally active chromatin, consistent with the onset of embryonic neurogenesis. Sp^−/− embryos exhibited increased levels of H3K9 and H3K18 acetylation along with KDM6B on the Neurog2 promoter compared with WT, suggesting that premature neurogenesis may be observed in Sp^−/− embryos due to increased transcriptional activity, coupled with the onset of a decrease in H3K27me2. The FA-treated Sp^−/− progenitor population showed decreased H3K9Ac and H3K18Ac association with Neurog2. KDM6B also decreased on this promoter, however, increased H3K27me2 was not observed, possibly due to the presence of UTX at this developmental time point. In summary, at E12.5, FA reversed H3K27me2, KDM6B, and UTX levels on the Neurog2 promoter, and abolished H3K9Ac and H3K18Ac binding. This suggests that FA favors transcriptional inactive chromatin on the Neurog2 promoter at the beginning of neural differentiation.

DISCUSSION

Lack of dietary FA produces profound retardation in embryonic growth and developmental progression, suggesting that cell proliferation is compromised (21). In our previous work (13), we found that neural crest cells from open neural tubes of Sp^−/− embryos failed to form neurospheres in culture. Here we showed that FA rescues the proliferation potential of neural precursors in Sp^−/− embryos. Cell proliferation may be compromised in Sp^−/− embryos due to significantly lowered Hes1 expression (13). It is possible that an epigenetic component of FA-mediated rescue could affect cell proliferation, including transcription of Hes1 and other genes involved in early stages of development, and thereby rescue posterior neural tube development.

If FA brings about its effects epigenetically, then epigenetic marks may differ between Sp^−/− and WT embryos. This study focused on H3K27 methylation, because it has been linked to the polycomb group protein-mediated suppression of Hox genes, animal body patterning, animal posterior development, and possibly maintenance of embryonic stem cell identity (7, 22). The data shows that the H3K27me2 mark is increased in caudal neural tube explants from E10.5 Sp^−/− embryos compared with WT controls, suggesting that down-regulation of H3K27me2 demethylase is involved in the open neural tube phenotype in Sp^−/− embryos. Levels of H3K27me2 reverse back toward WT in Sp^−/− neural tube explants treated with FA. This supports the hypothesis that increased methylation of lysine 27 of histone H3 is due to a decrease in KDM6B expression.

MiRNAs may be involved in decreased KDM6B expression. There is limited literature on how FA affects micro-RNA expression during development. Wang et al. (23) identified that FA supplementation suppressed ethanol-induced teratogenesis, in part by down-regulating miR-10a. FA deficiency causes a pronounced global increase in miRNA expression in cultured human lymphoblastoid cells (24). Returning cells to complete medium reversed miRNA expression profiles, suggesting that dietary modulation of miRNA expression is reversible (24).

In this study, several miRNAs were up-regulated in Sp^−/− embryos. Expression of some of these miRNAs was reversed by maternal FA supplementation prior to conception, suggesting that the absence of functional Pax3, as in Sp^−/− embryos, may play a role in miRNA biogenesis and maturation pathways. Further studies are needed to confirm this hypothesis. Other transcription factors, such as p53 and activated estrogen receptor α, have been shown to play a role in miRNA biogenesis and maturation (25, 26). It is plausible that during development, the Pax3 transcription factor can have a dual function of regulating target genes at transcriptional (13, 27–29) and post-transcriptional levels.

It is hypothesized that genes responsible for epigenetic regulation of stem cell maintenance and differentiation will be regulated by several miRNAs. We found four miRNAs, miR-138, miR-148a, miR-185, and miR-339-5p, with predicted recognition sites on KDM6B 3′ UTR (supplemental Fig. S2), thereby making KDM6B a putative target for these miRNAs. Transient transfection studies using pre-miRNA precursors support this hypothesis. Transfection of individual pre-miRNAs did not down-regulate KDM6B. Combinations of two or more pre-miRNAs were effective in silencing KDM6B expression. Immunochemical and immunoblot analysis also supported this hypothesis.

To identify potential epigenetically altered genes, we focused on Hes1, and Neurog2, because activity of these genes is altered in Sp^−/− embryos (13). Hes1 is responsible for progenitor cell proliferation (30) and maintenance of stem cell character (31), and Neurog2 is critical for sensory neurogenesis (12). If FA-mediated rescue of neural tube closure is an epigenetic event it is plausible that chromatin remodeling on Hes1 and Neurog2 promoters would be involved. This study suggests that H3K27me2 on the Hes1 promoter keeps cells from proliferating and low levels of this repressive chromatin mark on the Neurog2 promoter allows premature neurogenesis in Sp^−/− embryos at the proliferative stage as depicted in the hypothetical model proposed based on our studies (Fig. 9). FA may rescue this chromatin mark by decreasing the levels of miRNAs that target KDM6B. This may cause an increase in KDM6B, and a subsequent decrease in the repressive chromatin mark on the Hes1 promoter, resulting in increased stem cell proliferation for neuronal maturation in Sp^−/−. When the same occurs on Neurog2 promoter, it causes normal neurogenesis instead of premature neurogenesis. The association of H3K27me2 with the Hes1 promoter during early embryonic development and with the Neurog2 promoter during the onset of neurogenesis is gene and developmental time-dependent, suggesting a dynamic interplay of KDM6B demethylase in fine-tuning stem cell proliferation and differentiation. This study has significance for understanding how FA influences neural cell development. Because FA plays an important role in neuroplasticity and in the maintenance of neuronal integrity, this study may also shed light on one of the mechanisms by which low FA levels may affect adult neural disorders.

FIGURE 9.

Hypothetical model for FA-mediated rescue of stem cell proliferation and neurogenesis. Pax3 mutant “Splotch” homozygous embryos do not express functional Pax3, which leads to up-regulation of several miRNAs during early cell proliferation stages, some of which target H3K27 histone demethylase KDM6B. This results in decreased KDM6B expression and increased H3K27 methylation. The association between methylated H3K27 and the Hes1 promoter results in (a) decreased Hes1 expression and lowered stem cell proliferation. At the same time low association between methylated H3K27 and the Neurog2 promoter, but high association between acetylated H3K9 and H3K18 and the Neurog2 promoter results in (b) premature neurogenesis. FA-mediated rescue could be due to reversible down-regulation of KDM6B targeting micro-RNAs, resulting in up-regulation of KDM6B and subsequent lowering of H3K27 methylation. Demethylated H3K27 shows decreased binding to the Hes1 promoter and increased binding to the Neurog2 promoter, which results in (c) increased stem cell proliferation/maintenance and (d) normal neurogenesis.