Introduction

During development, neural progenitor cells (NPCs) have the ability to maintain self-renewal and differentiate into multiple neural cell types, including neurons, astrocytes and oligodendrocytes [1]. NPCs arise from the neuroepithelial cell layer of the neural tube, and their cell bodies reside in the ventricular zone in the developing cerebral cortex [2]. NPCs asymmetrically divide into one immature neuron and an intermediate progenitor or produce new NPCs by symmetric division [3, 4]. NPC daughter cells produce new neurons throughout cortical neurogenesis [5, 6]. Meanwhile, some NPCs differentiate into glial cells [7]. However, the molecular and mechanistic details controlling NPC self-renewal and the differentiation process are unclear at present. NPCs derived from embryonic stem cells (ESCs) have provided better tools for investigation of the molecular mechanisms of NPC self-renewal and neuronal generation.

Fibroblast growth factor (FGF) signaling retains a multipotent state capable of inducing NPCs to generate neuronal and glial cells [8, 9] and is required for neurogenesis and gliogenesis in the developing brain [10,11,12]. ETV5 is an ETS-related transcription factor that is defined by the conserved winged helix-turn-helix DNA-binding domain and mediates FGF signaling [13, 14]. A wealth of studies have revealed that ETV5 takes part in neural development events. During spontaneous differentiation of ESCs, ETV5 drivers the exit from naive pluripotency [15] and is involved in promotion of ESC proliferation and induction of ectoderm marker gene expression [16]. Etv5 is also necessary for NPC proliferation mediated by Capicua [17]. In peripheral ganglia, ETV5 is expressed in neural progenitors, neurons and satellite glia with the exception of Schwann cells [18]. ETV5 also mediates the axonal growth of dorsal root ganglion neurons in response to nerve growth factor and brain-derived neurotrophic factor [19, 20]. During hippocampal development, deletion of ETV5 causes deficits in hippocampal dendrite morphology and a reduction of hippocampal neurons [21]. ETV5 also controls the proliferation of hypothalamic serotoninergic precursors and induces hypothalamic serotoninergic neurons in zebrafish [22]. In addition, ETV5 plays a role in the initiation of gliogenesis and modulates propagation of cell production in neural crest stem cells and glioma [18, 23]. ETV5 contributes to gliogenesis during brain development and increased ETV5 leads to the perinatal lethality [24]. However, the molecular mechanisms by which ETV5 mediates these behaviors of NPCs are unknown.

Here, we describe the function of ETV5 in NPC self-renewal and neuronal generation during neural differentiation and provide insights into its molecular mechanism. We find that loss-of-function of ETV5 results in multipotency defects of NPCs and increases the formation of neuronal populations during the NPC differentiation process. Furthermore, we demonstrate that ETV5 represses transcription of the proneural gene NEUROG2. We find that ETV5 binds the NEUROG2 promoter and maintains its inactive chromatin markers. Additionally, ETV5 binds to the CoREST protein, which is essential for its function in NEUROG2 gene transcriptional regulation. Finally, we investigate the role of ETV5 in NPC-derived neural subtype specification. These data indicate that ETV5 functions to regulate both NPC self-renewal and neural subtype specification by modulating NEUROG2, which is a known proneural gene in NPCs.

Materials and Methods

Cell Culture

Human ESC H9 cells (WA09) were plated on irradiated mouse embryonic fibroblasts and cultured in DMEM/F12 medium (Gibco) supplemented with 20% Knockout Serum Replacer (Gibco), 1% MEM nonessential amino acids (Gibco), 1 mM L-glutamine (Gibco), 0.1 mM β-mercaptoethanol (Gibco) and 4 ng/ml of FGF2 (Invitrogen) as previously described [25]. Mouse embryonic fibroblasts were isolated from the CF-1 mouse strain (Slaccas). HEK293FT cells (Invitrogen) were grown in DMEM (Gibco) supplemented with 10% FBS. Human ESC H9 and HEK293FT cells were generous gifts from Dr. Xiaoqing Zhang.

Neurosphere Cultures

The neural differentiation method was described previously [26]. Briefly, H9 cells were dissociated with Dispase (Gibco) and resuspended in ESC growth medium without FGF2. After 6 days, ESC aggregates were plated on 6-well plates coated with laminin (Sigma) in neural induction medium containing DMEM/F12 medium, 1% N2 supplement (Gibco), 2 μg/ml heparin (Sigma) for 12 days and then resuspended cell pellets with neural induction medium supplemented with B27 (Gibco) for next 8 days. After that, medium was changed to neurosphere growth medium containing DMEM/F12 medium, 1% N2 supplement, 2% B27, 2 μg/ml heparin, 10 ng/ml EGF (Invitrogen) and 10 ng/ml FGF2. After 5–7 days, the neurospheres were passaged with Accutase (Gibco), and the dissociated single cells were suspended in fresh growth medium. After 3–5 passages, the numbers and diameters of the newly formed spheres were counted and measured, respectively. For the differentiation assay, the neurospheres were dissociated into individual cells and plated on poly-L-ornithine- (100 μg/ml, Sigma) and Matrigel-coated glass coverslips. The plated cells were allowed to differentiate in DMEM/F12 medium with 2% FBS for 7 days.

Neural Differentiation

Neurospheres were dissociated with Accutase and replated on 6-well plates coated with 1 μg/ml laminin in neural differentiation medium containing neurobasal medium (Gibco) supplemented with 1% N2 supplement, 2% B27, 1 μg/ml laminin, 100 nmol/l cAMP (Sigma), 200 ng/ml ascorbic acid (Sigma), BDNF, GDNF and IGF-1 (all 10 ng/ml, Pepro Tech). Cells were changed the medium every 2 days and cultured 25 days to induce differentiation of neurons.

Plasmid Construction

The ETV5 and CoREST cDNA-encoding regions were cloned into the FUGW plasmid via the BamHI and EcoRI sites using PCR primers and ETV5 was subsequently cloned into the pLVX-Tight-Puro plasmid (Clontech). A Flag-tag was fused at the N-terminus of ETV5, and an HA tag was fused at the C-terminus of CoREST. The ETV5 and CoREST mutant fragment plasmids were generated by PCR amplification of the FUGW-Flag-ETV5 and FUGW-CoREST-HA plasmids, respectively. To construct the luciferase reporter plasmids, NEUROG2 promoter fragments were amplified by PCR from genomic DNA and cloned into the pGL4 plasmid. Two shRNAs against CoREST (CoREST shRNA 1: TRCN0000128570, targeting 5'-GATGGTGGAATAGAACCATAT-3', and CoREST shRNA 2: TRCN0000147486, targeting 5'-GTTGGATGAATACATTGCCAT-3') were inserted into the pLKO.1 plasmid. The pLKO.1 scrambled shRNA target sequence was 5'-CGTACGCGGAATACTTCGAAA-3'.

Knockout ESC Line Construction

The ETV5 targeting donor plasmid carried the EF1α promoter-driven blasticidin resistance gene followed by polyadenylation sequence flanked by the ETV5 homology arms. ESCs were cultured in 1 mmol/l Rho kinase inhibitor Y27632 (Calbiochem) 24 h before electroporation. ESCs were dissociated to single cells using trypsin solution and washed with phosphate buffered saline. Cell pellets were electroporated with 40 μg of ETV5 targeting donor plasmid, 5 μg of Cas9 plasmid and 5 μg of guide RNA (gRNA) targeting plasmid using Gene Pulser Xcell System (Bio-Rad). Individual colonies were picked after 10 to 14 days of blasticidin selection (1 mg/ml) after electroporation. The primers used to amplify the homology arms were as follows: 5’ homology arm, forward 5'-TGCGTCcgtctcGGATCTTTGTTTAGCTAGTTCTACCTCGT-3', reverse 5'-GCTGACcgtctcCGTTCGCTCTCAGCAAGAAATACTCAA-3' and 3’ homology arm, forward 5'- TGCGTCcgtctcTGACATAAGAATATCCAGGCCCCAG-3', reverse 5'- GCTGACcgtctcCGGCCTGCTGTGGTCCAGGCTCT-3'. The gRNA targeting sequence was 5'-TGCAGCTCCCGTTTGATCT-3'.

Lentivirus-Mediated Stable ESC Line Construction

Lentiviral particles were generated by transfecting HEK293FT cells with pLVX-Tight-ETV5-Puro, psPAX2, and VSVG plasmids at a ratio of 4:3:2. Lentiviral supernatants were collected after 48–72 h after transfection and concentrated by ultracentrifugation. rtTA overexpression H9 cells, described previously [27], were infected with the concentrated virus and individual colonies were picked after selected using puromycin. To induce the ETV5 expression, cells were treated with 1 μg/ml doxycycline (DOX).

Luciferase Reporter Assays

Human ESC-derived neurospheres were dissociated with Accutase, plated into laminin-coated 24-well plates and allowed to attach overnight. Then, the cells were transfected with the pGL4 luciferase reporter plasmid, Renilla plasmid and expression plasmid using the Lipofectamine 3000 reagent (Life Technologies). After 48 h, the cells were harvested and prepared with the dual luciferase assay kit (Promega). The luciferase activities were determined using the Infinite 200 Pro Luminometer (Tecan).

Real-Time Quantitative PCR

Total RNA was extracted from the cells using the TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. cDNA was synthesized with 1 μg of RNA sample using the Superscript III reverse transcriptase (Invitrogen). Real-time quantitative PCR analysis was performed using SYBR Premix Ex Taq II (TaKaRa) and the CFX96 real-time PCR detection system (Bio-Rad). The primers used for the real-time quantitative PCR are listed in Table S1.

Chromatin Immunoprecipitation (ChIP)

ChIP was performed using an EZ-Magna ChIP A/G chromatin immunoprecipitation kit (Millipore) according to the manufacturer’s instructions. Chromatin was immunoprecipitated using anti-ETV5 (sc-22,807, Santa Cruz Biotechnology), anti-CoREST (07–455, Millipore), anti-H3K4me3 (04–745, Millipore), anti-H3K27me3 (ab6002, Abcam), anti-acetylated H3 (06–599, Millipore) and anti-acetylated H4 (06–598, Millipore) antibodies. Normal mouse IgG was used as a negative control. Real-time quantitative PCR was performed using the following primers spanning the NEUROG2 promoter region: NEUROG2 promoter primer 1, forward 5'-AGGAGGAGGATGGATGGC-3', reverse 5'-CCTCTGGCTTATTCTTTTCATTG-3' and NEUROG2 promoter primer 2, forward 5'-CAGGAGCTTTAGAACCATTTTATG-3', reverse 5'-CCTCTTTCTTCACCGCCTTT-3'.

Co-Immunoprecipitation and Western Blotting Analysis

HEK293FT cells were cotransfected with the Flag-tagged ETV5 and the HA-tagged CoREST WT/mutant plasmids. After 48 h, the cells were lysed with ice-cold immunoprecipitation buffer (40 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 20% glycerol and 0.3 mM EDTA) in the presence of protease inhibitor cocktail and 0.1 mM PMSF. The lysates were incubated with an Anti-Flag M2 Affinity Gel (Sigma) or EZview Red Anti-HA Affinity Gel (Sigma) for 2 h at 4 °C. The affinity gels were harvested and washed 3 times with immunoprecipitation buffer and subjected to Western blotting analysis. To detect ETV5-CoREST binding in the NPCs, NPC lysates were immunoprecipitated with control IgG or anti-CoREST antibodies and probed with an anti-ETV5 antibody. For the Western blotting analysis, cell lysates were prepared with RIPA buffer, and the protein concentrations were determined using the BCA method (Thermo Scientific). The primary antibodies are anti-ETV5 (1:500), anti-CoREST (1:1000), anti-Flag (1:1000; F1804, Sigma), anti-HA (1:1000; ab9110, Abcam) and anti-β-Actin (1:2000; A5316, Sigma).

Immunofluorescence Staining

Cells plated on glass coverslips were fixed in 4% paraformaldehyde for 10 min. The cells were washed in PBS and then blocked with 10% normal donkey serum and 0.2% Triton X-100 in PBS for 1 h. The cells were incubated with primary antibodies, including anti-TUJ1 (1:1000; T8660, Sigma and PRB-435P, Covance), anti-GFAP (1:1000; Z0344, Dako), anti-VGLUT1 (1:1000; AB5905, Millipore), anti-GABA (1:1000; A2052, Sigma), anti-TH (1:1000; T2928, Sigma) and anti-NEUROG2 (1:500; AB5682, Millipore), for 12 h at 4 °C, followed by the Alexa Fluor 488- (1:1000; Jackson ImmunoResearch) and Cy3-conjugated secondary antibodies (1:1000; Jackson ImmunoResearch) for 1 h at room temperature. The nuclei were labeled with Hoechst 33342. Images were captured on a Leica TSC SP5 confocal microscope (Leica).

Statistical Analysis

All results are presented as the mean ± SD. The statistical analysis was performed using PRISM 5.0 (GraphPad Software). The data were analyzed with an unpaired Student’s t test or one-way ANOVA. A two-tailed p < 0.05 was considered statistically significant. * and ** indicate p < 0.05 and p < 0.01, respectively. N.S. means not significant.

Results

ETV5 is Essential for NPC Self-Renewal and Inhibition of Neuronal Differentiation

The subventricular zones, containing higher concentration of NPCs, were composed of intensity of major neurogenic events during mid-fetal period (15–22 postconceptional weeks, PCW) with an increase in thickness in human fetal brain [28]. From 10 to 17 PCW, human fetal brain appeared expansion of the cortical plate and differentiation of the subventricular zone, suggesting NPCs started to differentiate [29]. At 16 PCW, neuropeptide Y (NPY) began to express in subplate neurons, revealed the earliest differentiation of GABAergic neurons [30]. We hypothesized ETS family members were involved in maintenance and/or early neuronal fate of NPCs during these PCW in the human. To evaluate the expression of ETS family members in the human fetal brain, we analyzed gene array data for ETS family proteins at mid-fetal period (15 PCW) from the Prenatal Laser Microdissection Microarray of BrainSpan Atlas (http://www.brainspan.org/lcm/search/index.html) [31]. We found that ETS translocation variant 5 (ETV5) was typically restricted to the ventricular and subventricular zones within the human fetal brain. Because embryonic NPCs were observed in the same cortical layers, we hypothesized that ETV5 expression might be related to NPC identity and function.

To determine the role of the ETV5 gene, we performed targeted deletion of ETV5 in the human ESC line H9 using the Cas9/CRISPR genome editing tool (Fig. 1a). Indeed, the ETV5 mRNA and protein were absent in the ETV5 knockout (KO) human ESCs, as validated by real-time quantitative PCR and Western blotting assays (Figure S1A and S1B). Then, we generated ETV5 KO NPCs derived from the ETV5 KO ESC lineage following the neural differentiation method.

Fig. 1
figure 1

ETV5 KO affects NPCs self-renewal and neuronal differentiation. a Schematic diagram showing the homologous recombination strategy used to disrupt the ETV5 gene. EF1α, EF1α promoter; BSD, blasticidin resistance gene; PA, polyadenylation sequence. b NPCs were generated from WT and ETV5 KO ESCs and formed free-floating neurospheres in neural induction medium containing EGF and bFGF. c The diameters of the neurospheres from the WT and ETV5 KO NPCs were measured at passage 3. d The numbers of neurospheres were counted at passage 3. e The NPC markers NESTIN, SOX1, Musashi1, TLX, HES5 and PAX6 were quantified by real-time quantitative PCR. f The proneural markers NEUROG2 and NEUROD1 were quantified by real-time quantitative PCR. g Immunostaining of neurons (TUJ1) and astrocytes (GFAP) derived from WT and ETV5 KO NPCs. h, i Quantification of TUJ1-positive cells and GFAP-positive cells derived from the WT and ETV5 KO NPCs. The scale bar represents 100 μm 

Full size image

Next, we cultured and passaged neurospheres from the ETV5 KO NPCs. Compared to the wild-type neurospheres, we observed that the ETV5 KO neurospheres were noticeably smaller (Fig. 1b). The effects on proliferation were confirmed by measuring the sphere diameters and counting the sphere numbers (Fig. 1c, d). The sizes and numbers of the neurospheres were obviously decreased in the ETV5 KO NPCs. Real-time quantitative PCR for self-renewal markers of NPCs showed that NESTIN, SOX1, Musashi1, TLX, HES5 and PAX6 failed to accumulate in the ETV5 KO NPCs (Fig. 1e). Conversely, expression of the proneural genes NEUROG2 and NEUROD1 was significantly increased in the ETV5 KO NPCs (Fig. 1f). Our results demonstrated that the proliferation and self-renewal functions of ETV5 KO NPCs were significantly decreased. We sought to determine whether the ETV5 KO NPCs displayed the capacity to differentiate into neurons and glial cells under given conditions. Immunocytostaining analysis showed that TUJ1-positive cells with neuronal morphologies were increased with a reduction in the number of GFAP-positive cells in the ETV5 KO NPCs after differentiation (Fig. 1g). The quantitative data showed that 70% and 14% of the ETV5 KO NPCs had differentiated into TUJ1-positive cells and GFAP-positive cells, respectively, whereas 45% and 34% of the WT NPCs became TUJ1-positive cells and GFAP-positive cells, respectively (Fig. 1h, i). Taken together, these observations suggest that ETV5 plays a critical role in NPC self-renewal and repression of neuronal differentiation genes.

ETV5 Negatively Regulates NEUROG2 Transcription via the NEUROG2 Promoter

The proneural gene NEUROG2 is a key regulator that promotes neuronal differentiation during neural development [32]. We found that NEUROG2 was increased markedly in the ETV5 KO NPCs. Next, we investigated whether ETV5 played a role in regulating NEUROG2 in NPCs. To explore the relationship between ETV5 and NEUROG2, we performed luciferase reporter assays in ETV5 KO NPCs using NEUROG2 promoter reporter constructs or an empty reporter plasmid as a control. Compared to that of the WT NPCs, NEUROG2 gene transcription was significantly induced in the ETV5 KO NPCs (Fig. 2a). Furthermore, we introduced an ETV5 expression plasmid into NPCs together with NEUROG2 promoter reporter constructs and found that ETV5 strongly suppressed NEUROG2 gene transcription in the NPCs (Fig. 2b). To provide further evidence that ETV5 regulated the NEUROG2 promoter, we generated various deletion constructs to define which region of ETV5 caused the transcriptional repression (Fig. 2c). This deletion analysis identified amino acids 352–455 of ETV5 as containing the ETS domain that was indispensable for repression of NEUROG2 promoter activity in NPCs (Fig. 2d).

Fig. 2
figure 2

ETV5 negatively regulates NEUROG2 transcription in NPCs. a WT and ETV5 KO NPCs were transfected with the pGL4-empty plasmid or the pGL4-NEUROG2 promoter plasmid. Luciferase activity was measured 48 h after transfection. b Luciferase activity was measured after transfection of increasing amounts of the ETV5 expression plasmid into WT NPCs with the pGL4-NEUROG2 promoter plasmid. c Schematic of ETV5 deletion mutants used for the mapping experiments. d Analysis of ETV5 domains that regulate NEUROG2 transcription in NPCs. e Extended deletion analysis of the NEUROG2 promoter using luciferase reporter assays was performed as described with deletions from −1331 to +128. f Schematic representation of two putative ETS transcription factor binding sites in the NEUROG2 promoter region using MatInspector

Full size image

To determine the functional regions of the NEUROG2 promoter regulated by ETV5, we generated a series of luciferase constructs driven by different lengths of the NEUROG2 promoter. The luciferase reporter gene assays showed that the NEUROG2 promoter harboring sequences from −1331 to −340 exhibited a decrease in luciferase activity in the presence of ETV5 in NPCs. However, ETV5 was unable to downregulate luciferase activity following deletion of the −340 to −146 fragment of the NEUROG2 promoter (Fig. 2e). These results indicated that the −340 to −146 fragment of the NEUROG2 promoter was required for the ETV5-dependent downregulation of luciferase activity. Bioinformatics analysis of the −340 to −146 fragment of the NEUROG2 promoter showed two putative ETS binding sites using MatInspector (Genomatix) (Fig. 2f). Together, these results demonstrate that ETV5 negatively regulates activation of the NEUROG2 promoter in NPCs.

ETV5 Binds the NEUROG2 Promoter and Removes its Active Chromatin Markers

We extended our study to include a detailed analysis of ETV5 binding in the transcriptional unit of NEUROG2. We performed ChIP assays with an ETV5 antibody in NPCs using 2 pairs of real-time quantitative PCR primers that flanked the −340 to −146 fragment of the NEUROG2 promoter (Fig. 3a). We found that the NEUROG2 promoter was specifically enriched with the ETV5 antibody relative to the IgG control antibody as measured by real-time quantitative PCR (Fig. 3b). Next, we analyzed chromatin marks in the NEUROG2 promoter using the ChIP assay. The ChIP experiment demonstrated that the NEUROG2 promoter in the ETV5 KO NPCs triggered a significant accumulation of histone H3 trimethylated at Lys 4 (H3K4me3) and reduced methylation levels of histone H3 trimethylated at Lys 27 (H3K27me3) (Fig. 3c). Total acetylation of histones H3 and H4, which are linked to active transcription, was enriched in the NEUROG2 promoter in the absence of ETV5 (Fig. 3d). Our data reveal that the NEUROG2 promoter is associated with activating chromatin modifications in ETV5 KO NPCs.

Fig. 3
figure 3

ETV5 binds the NEUROG2 promoter and changes H3K4me3, H3K27me3, AcH3 and AcH4 occupancy in NPCs. a Diagram of the NEUROG2 promoter showing the two predicated ETS binding sites. The positions of the ChIP amplicons are specified as P1 and P2. b ChIP assays from WT NPCs using an ETV5 antibody. c ChIP real-time quantitative PCR analysis for altered H3K4me3 and H3K27me3 occupancy of the NEUROG2 promoter in the WT and ETV5 KO NPCs. d ChIP real-time quantitative PCR analysis for altered AcH3 and AcH4 occupancy of the NEUROG2 promoter in the WT and ETV5 KO NPCs

Full size image

CoREST is Necessary for ETV5-Medicated NEUROG2 Transcriptional Repression

To study how ETV5 affected NEUROG2 gene transcription, we speculated that CoREST acted as a regulator of ETV5-medicated transcriptional repression in NPCs. Expression of CoREST was insusceptible to ETV5 KO in NPCs (Figure S2A). ChIP in NPCs showed that CoREST was bound to the NEUROG2 promoter near the ETV5-binding sites (Fig. 4a). Next, we explored the possibility of an ETV5-CoREST interaction in NPCs. ETV5 was coimmunoprecipitated from the NPC extract with the CoREST antibody but not with an unrelated IgG antibody (Fig. 4b). ETV5 was also coimmunoprecipitated from HEK293FT cells cotransfected with ETV5 and CoREST expression plasmids (Figure S2B). Furthermore, the low NEUROG2 transcription activity level caused by ETV5 overexpression was dramatically enhanced by shRNA knockdown of CoREST. This result indicated that ETV5 repression in the NEUROG2 promoter relied on CoREST in NPCs (Fig. 4c).

Fig. 4
figure 4

CoREST is required for ETV5-mediated transcriptional repression of the NEUROG2 promoter. a ChIP assays from WT NPCs using a CoREST antibody. P1 and P2 are two ChIP amplicons on the NEUROG2 promoter, as shown on Fig. 3A. b Coimmunoprecipitation analysis of interactions between ETV5 and CoREST in NPC lysates. c The reduction in luciferase activity caused by ETV5 in NPCs is rescued by CoREST knockdown. d Deletion mutation constructs of ETV5 employed to analyze interactions with CoREST. e Deletion mutation constructs of CoREST employed to analyze interactions with ETV5 

Full size image

To better understand the interactions between ETV5 and CoREST, we conducted in vitro binding experiments to identify domains involved in the ETV5-CoREST interactions. Amino acids 200–352 of ETV5 are important for the interaction with CoREST (Fig. 4d). We also investigated the domains within CoREST that bound to ETV5. Amino acids 189–241 of CoREST appear to interact with ETV5 (Fig. 4e). Taken together, these findings indicate that the function of ETV5 is dependent on CoREST for transcriptional repression of NEUROG2 promoter activity.

ETV5 Suppresses the Glutamatergic and Promotes the GABAergic Neuronal Fates

Previous evidence indicated that NEUROG2 was important to specify cortical NPCs toward glutamatergic neurons [33, 34]. We assessed whether ETV5 could affect the capacity of NPCs to differentiate into different neuronal types. NPCs were differentiated into neurons for 14 days using neuron-specific medium and subsequently analyzed immunocytochemically. Immunostaining showed that ETV5 KO resulted in increased NEUROG2 expression in TUJ1-positive neurons. ETV5 KO NPC-derived neurons exhibited more TUJ1-positive cells that were vesicular glutamate transporter 1 (VGLUT1)-positive (Fig. 5a). This result suggested that glutamatergic neurons were increased in the ETV5 KO neurons. On the other hand, the neurons differentiated from the ETV5 KO NPCs appeared to have fewer GABAergic neurons labeled by gamma-aminobutyric acid (GABA) and no significant difference in dopaminergic neurons labeled by tyrosine hydroxylase (TH) compared with those of neurons differentiated from WT NPCs (Fig. 5a). Furthermore, we established ETV5 overexpressing NPCs using a tetracycline-inducible system and finally conducted neuronal differentiation. We found that ETV5 overexpression (DOX+) led to a significant decrease in the numbers of NEUROG2- and VGLUT-positive cells and promoted the generation of GABA-positive cells (Fig. 5b). Quantitative assays for the numbers of VGLUT1-positive, GABA-positive, and TH-positive cells among the TUJ1-positive cells in the ETV5 KO or ETV5 overexpressing neurons gave the same results (Fig. 5c, d, e, f). Together, these data indicated that ETV5 played a negative role in glutamatergic differentiation and positively mediated the formation of GABAergic neurons.

Fig. 5
figure 5

Characterization of ETV5 KO and overexpression NPC-derived neurons. a Immunostaining for NEUROG2, VGLUT1, GABA and TH in WT and ETV5 KO NPC-derived neurons. b Immunostaining for NEUROG2, VGLUT1, GABA and TH in control (DOX-) and ETV5 overexpressing (DOX+) NPC-derived neurons. The neurons were immunofluorescently labeled with TUJ1. c-f Quantitation of the percentages of NEUROG2, VGLUT1, GABA and TH cells among the TUJ1-positive cells. The scale bar represents 100 μm 

Full size image

Discussion

Dysfunctions of inhibitory neuronal subtypes are related to a group of neurological diseases, including epilepsy, schizophrenia, autism spectrum disorder, Alzheimer’s disease, and Parkinson’s diseases [35]. Several studies have used innovative GABAergic neural precursor-based therapeutic approaches to treat these diseases [36,37,38]. In these studies, GABAergic neural precursors derived from medial ganglionic eminence in mouse embryos were immediately grafted into the experimental mouse brains. However, limited cell expansion of GABAergic neural precursors limited their application without culture or any further manipulation in vitro. Pluripotent stem cells present a model system to generate GABAergic neurons [39]. However, the mechanism and application of GABAergic neural precursors need to be clarified. In our study, we demonstrated that ETV5 was necessary to preserve NPC identity, repress NEUROG2 expression and decrease NEUROG2-positive progenitor cells and conversely to promote the appearance of GABAergic neurons derived from NPCs in vitro.

NEUROG2 is a well-known proneural transcription factor, characterized by the presence of bHLH (basic helix-loop-helix) domain, which plays roles in selection of NPC are committed to a neural fate and specification of neural cell fate in neural development [40, 41]. NEUROG2 is sufficient to promote neurogenesis and regulates the fate of NPC to neurons, while inhibition of NEUROG2 in NPC promotes the formation of astrocytes [42,43,44]. Moreover, NEUROG2 specifies glutamatergic phenotype of neocortical neurons in developing mouse [33, 45]. NEUROG2 is required to specify glutamatergic neurotransmitter phenotype in neocortical neurons and governs neuronal subtype reprogramming by altering the chromatin landscapes [46]. NPC with high levels of NEUROG2 will generate the glutamatergic daughter neurons finally [47]. Adult NEUROG2-positive NPC generate glutamatergic neurons in subependymal zone of adult mouse [48]. NEUROG2-transduced NPC displays differentiation into glutamatergic neurons in vitro and in dentate gyrus of hippocampus after cell transplantation [49]. The glutamatergic properties of neurons derived from NEUROG2-transduced NPC are conformed using electrophysiology [34]. NEUROG2 is required to the expression of VGLUT1, which mediates glutamate uptake by synaptic vesicles in glutamatergic neurons [50].

The essential roles of NEUROG2 in neocortical development have been fully studied [47]. However, the regulation of NEUROG2 in NPCs is unclear. Gene KO in ESC and NPC is a powerful approach to revealing the functions of the interesting gene in embryonic development in vitro even if this gene is critical for survival in animal models [51,52,53]. Here we found that ETV5 can regulated NPC differentiation and specification of neural cell fate through NEUROG2 using CRISPR/Cas9-medicated KO. ETV5 KO NPCs undergo differentiate into neurons with low differentiation capacity into glial cells, these suggest ETV5 and NEUROG2 exert the opposite function in controlling NPC differentiation. Further study provides NEUROG2 expression is negatively regulated by ETV5 during NPCs differentiation. ETV5 KO resulted in progressive expression of NEUROG2 in NPCs, which correlated with an increase in NEUROG2 transcription. Forced expression of ETV5 represses the transcriptional activity of NEUROG2 in NPCs. ETV5-modified NEUROG2 transcription by binding its promoter and promoting the appearance of repression markers of chromatin modification. Furthermore, ETV5 interacted with the neural-specific corepressor CoREST and then exerted its transcriptional repression role.

Mechanistically, our results showed at first glance that CoREST was required for ETV5-induced transcriptional repression of the NEUROG2 gene in NPCs. CoREST is a chromatin-modifying repressor that regulates neuronal gene expression and neuronal cell fate. CoREST is strongly expressed in neural regions of the developing brain and is downregulated after birth [54, 55]. The target genes of CoREST are involved in neuronal subtype specification and maintenance as well as glial lineage species [56, 57]. Loss of CoREST function causes a delay in the migration of newborn pyramidal neurons in the cerebral cortex [58]. CoREST is also regulated by microRNAs to control neuronal polarization and migration in the cerebral cortex through inhibition of doublecortin transcription [59]. CoREST negatively regulates Notch pathway activity during development of the cerebral cortex with defects in neuronal migration and increased numbers of SOX2-positive and TBR2-positive cells [60]. CoREST can be recruited directly by REST to bind repressor element 1 (RE1) sites that are present in several neuron-specific genes and to silence neuronal gene expression [61,62,63,64]. However, some studies suggest that CoREST relies on factors contributing to transcriptional regulation to mediate repression of neural-specific genes in a REST-independent manner [58, 65]. Our data show that in addition to REST, ETV5 can also bind CoREST and finally develop transcriptional repression in neuron-specific gene promoters, such as the NEUROG2 promoter.

We found that ETV5 could alter H3K4m3 and H3K27me3 methylation and H3 and H4 acetylation in the NEUROG2 promoter. However, ETV5 lacks enzyme-modifying activities in its protein structure. As a replacement for ETV5, its partner CoREST serves as a transcriptional repressor in the NEUROG2 promoter in NPCs, may as the translational repressor PUM2 does [66]. CoREST interacts with acetylation and methylation modifiers, including HDAC1, HDAC2 and LSD1, to form a CoREST complex and modify transcriptional repression [67,68,69]. CoREST modifies the deacetylation of lysine residues on histones 3 and 4, resulting in transcriptional repression through recruitment of HDAC1 by the N-terminal ELM2 and SANT1 domains of CoREST [67]. CoREST also directly interacts with LSD1 and is required for LSD1-mediated lysine-specific demethylation [69,70,71]. LSD1 has the ability to remove mono- and di-methyl groups from lysine 4 of histone H3 (H3K4) [72]. We indicate that CoREST plays a role in regulation of acetylation and methylation of the NEUROG2 gene, whereas ETV5 provides a binding site for CoREST in the NEUROG2 promoter.

NEUROG2 plays a role in the specification of neuronal glutamatergic identity during neocortical development [73,74,75,76] and contrarily represses the formation of GABAergic neurons [33, 77]. NEUROG2 is a reporter allowing the specific labeling of prenatal and postnatal glutamatergic neurons progenitors [78]. One of our interesting findings is that ETV5 can repress NEUROG2 at the protein level in NPC-derived neurons. Moreover, ETV5 can repress the glutamatergic neuronal marker VGLUT1 and promote the GABAergic neuronal marker GABA in TUJ1-positive NPC-derived neurons. These findings suggest that ETV5 affects the fate of the neuronal glutamatergic and GABAergic subtype identities in the patterning program of NPCs, at least through repression of NEUROG2. ETV5 acts as an inhibitor of glutamatergic fate and increases the number of GABAergic neurons in neural cell-type specification from NPCs.

Together, our study presented here suggest that ETV5 underlies the diminished transcriptional activity of NEUROG2 in NPCs through binding to the NEUROG2 promoter and maintaining a close association with CoREST. ETV5 represses NEUROG2-induced glutamatergic identity and provides insights into the phenotypic properties of increasing the percentage of GABAergic cells in neural cell pools derived from NPCs and represents a potent new candidate protein for GABAergic cell therapies to treat neurological diseases.