Brain-specific deletion of histone variant H2A.z results in cortical neurogenesis defects and neurodevelopmental disorder.

Defects in neurogenesis alter brain circuit formations and may lead to neurodevelopmental disorders such as autism and schizophrenia. Histone H2A.z, a variant of histone H2A, plays critical roles in chromatin structure and epigenetic regulation, but its function and mechanism in brain development remain largely unknown. Here, we find that the deletion of H2A.z results in enhanced proliferation of neural progenitors but reduced neuronal differentiation. In addition, neurons in H2A.z knockout mice exhibit abnormal dendrites during brain development. Furthermore, H2A.zcKO mice exhibit serial behavioral deficits, such as decreased exploratory activity and impaired learning and memory. Mechanistically, H2A.z regulates embryonic neurogenesis by targeting Nkx2-4 through interaction with Setd2, thereby promoting H3K36me3 modification to activate the transcription of Nkx2-4. Furthermore, enforced expression of Nkx2-4 can rescue the defective neurogenesis in the H2A.z-knockdown embryonic brain. Together, our findings implicate the epigenetic regulation by H2A.z in embryonic neurogenesis and provide a framework for understanding how disruption in the H2A.z gene may contribute to neurological disorders.

INTRODUCTION

During embryonic neurogenesis, neurons are generated from neural progenitor cells (NPCs) directly or indirectly in a spatially and temporally regulated manner (1,2). These postmitotic neurons form functional neuronal circuits after migrating over a long distance to their final locations (3). Any disturbance in this process of neurogenesis might result in neurodevelopmental disorders, including schizophrenia and autism (4–6). Besides transcription factors, epigenetic molecules also play a crucial role in the spatial and temporal control of cortical neurogenesis (7–10). Posttranslational modifications of histone proteins affect the access of transcription factors to target sites, contributing to the complex control of gene expression (11,12). An additional regulatory mechanism is the replacement of conventional histones with their specific variants within the chromatin (13,14).

Histone variants are structural components of chromatin and have evolved crucial roles in genome integrity, transcriptional regulation, DNA repair, and other processes (15). Among those variants, H2A.z is of particular interest with respect to its function (16–18). H2A.z belongs to the H2A family but has only 60% identity to canonical H2A (19). Recent studies in various cellular systems show that H2A.z also has multiple functions in the transcription process, including transcriptional activation, repression, and elongation (17,20–23). Moreover, deficiency of H2A.z leads to embryonic lethality at the pre-implantation stage of the blastocyst (24). Interestingly, H2A.z has apparently contradictory roles in gene activation and silencing (18,25,26). Some of these contradictions might be explained by local histone modification patterns (25). In addition, a genome-wide linkage study indicated that the H2A.z gene may be a susceptibility factor for schizophrenia (27). Schizophrenia comprises a range of neurodevelopmental disorders characterized by mental disorder, delusions and cognitive deficits, with highly heritability (28). Therefore, it is particularly important to illustrate how H2A.z incorporation alters chromatin states to regulate gene expression during embryonic neurogenesis.

Although there is a wide consensus on the importance of H2A.z, the function of H2A.z in embryonic neurogenesis and animal behavior remains unknown. Here, we report that the deletion of H2A.z not only promotes the proliferation of progenitors but also inhibits neuronal differentiation. Neurons in H2A.z knockout mice exhibit abnormal dendrites with cortical development. Mechanistically, H2A.z regulates the expression of Nkx2–4 by interaction with Setd2, thereby promoting the H3K36me3 levels of the promoter. The combination of H2A.z and histone H3K36me3 modification by Setd2 promotes transcriptionally permissive chromatin and transcriptional activation in the developing brain. In addition, H2A.z mice exhibit decreased exploratory activity and social behavior and impaired learning and memory, and they are prone to anxiety and depression. Together, our study reveals the functions of H2A.z in progenitor proliferation and differentiation, neuronal morphology, and mice behavior, thus providing new insights into the epigenetic regulation by H2A.z in the developing brain.

MATERIALS AND METHODS

Animals

H2A.z1/2 floxed mice were provided by the RIKEN BRC through the National Bio-Resource Project of the MEXT. To generate H2A.z mice, H2A.z1/2 floxed mice were crossed with C57BL/6J mice, and then the F1 offspring were crossed with each other to obtain H2A.z1 mice. Nestin-Cre mice (B6. Cg-Tg(Nes-cre)1Kln/J) were obtained from the Jackson Laboratory. To generate H2A.z mice (H2A.z1 is regarded as H2A.z in our study), H2A.z1 mice were crossed with Nestin-Cre mice. Genotyping for H2A.z1 and Cre alleles was performed by PCR analysis of tail DNA. All animal experiments were conducted in accordance with the guidelines of the Animal Care and Use Committee of the Institute of Zoology, Chinese Academy of Sciences.

Plasmid constructs

ShRNAs for the target genes were cloned into the pSicoR-GFP vector. The sequences were as follows: H2A.z-shRNA1: 5′-GGTAAGGCTGGAAAGGACT-3′ (25); H2A.z-shRNA2: 5′-CGGGAAGAAAGGACAACAGAA-3′ (26); SETD2-shRNA: 5′-GAAACCGTCTCCAGTCTGT-3′ (29); and NKX2–4-shRNA: 5′-CCAGAACCATCGCTACAAGAT-3′; H2A.v-sh1: 5′-GTGTTGGAGTTAGCAGGTAAT-3′. The cDNA of H2A.z, siRNA-resistant H2A.z mutant, H2A, and NKX2–4 were amplified by polymerase chain reaction (PCR) and subcloned into the pCDH-3xFLAG vector. SETD2-SET domain cDNA was amplified and subcloned into pCDH-3xHA. The H2A.z-shRNA1-resistant H2A.z cDNA mutant: forward 5′-TAGAGCTGGCGGCAAAGCCGGGAAAGATTCCGGAAAGGC-3′; reverse 5′-GCCTTTCCGGAATCTTTCCCGGCTTTGCCGCCAGCTCTA-3′; SETD2-SET domain cDNA: forward 5′- GGAATTCGGGCAAGTACCAGATTCTCTAACA-3′; reverse 5′- CGGGATCCCTCTTCCACAGTGACAGTCTCTAA-3′.

Cell cultures

Both the cell lines and the primary neural progenitor cells were cultured as previously described (30). Briefly, mouse neuroblastoma N2a cells and the human embryonic kidney (HEK) 293FT cell line were cultured in DMEM (Life Technologies) supplemented with 10% FBS, GlutaMAX (Invitrogen, 0.5%), nonessential amino acids (Invitrogen, 1%), and antibiotics. Neural progenitor cells were isolated from the dorsal brains of E12 embryos and then cultured in proliferation or differentiation media for subsequent treatment. For primary neurosphere assays, neural progenitor cells were cultured in non-adherent conditions at a clonal density of 50 cells per ml in six-well plates, and neurospheres were counted 3 days later.

In-utero electroporation

In-utero electroporation was performed as previously described (31). Briefly, pregnant mice at E13 were anesthetized, and plasmid DNA and 0.02% Fast Green (Sigma) were microinjected into the lateral ventricle with a glass micropipette. The electroporated brains were removed by dissection at the indicated days and were cut into sections for further analysis.

BrdU labeling

Generally, BrdU was injected intraperitoneally into E15 or E16 pregnant mice at a dose of 100 mg/kg. For cell proliferation analysis, BrdU was injected 2 h before pregnant mice were sacrificed at E16. For cell cycle exit analysis, BrdU was administered at E15 before the pregnant mice were harvested at E16.

Immunostaining

The procedure for immunostaining brain sections or cultured cells was described previously (32). The following antibodies were used: rat anti-BrdU (1:1000; Abcam; ab6326), rabbit anti-Ki67 (1:1000; Abcam; ab15580), rabbit anti-phospho-Histone H3 (pH3) (1:1000; Cell Signaling Technology; 3377S), mouse anti-Nestin (1: 200; Millipore; MAB353), rabbit anti-Pax6 (1:1000; Abcam; ab5790), mouse anti-Pax6 (1:100; DSHB), rabbit anti-Tbr2 (1:1000; Abcam; ab23345), mouse anti-Sox2 (1:1000; R&D; MAB2018), rabbit anti-Tuj1 (1:1000; Sigma; T2200), mouse anti-Tuj1 (1:2000; Millipore; MAB1637), rabbit anti-H2A.z (1:1000; Active Motif; 39943), rabbit anti-NeuN (1:1000; Abcam; ab177487), rabbit anti-Nkx2–4 (1:1000; Abcam; ab189202), mouse anti-β-actin (1:2000, Proteintech; 60008-1-Ig), rabbit anti-Flag (1:1000, Sigma; F1804), rabbit anti-IgG (1:1000, Bioss; bs-0295p), mouse anti-HA (1:1000; Abmart; M20003L), rabbit anti-H3K36me3 (1:1000; Cell Signaling Technology; 4909S), rabbit anti-H3K36me3 (Active Motif; 61101), rabbit anti-H3K27me3 (1:1000; Millipore; 07-449), rabbit anti-H3K9me3 (1:1000; Millipore; 07-442), rabbit anti-H3K4me3 (1:1000; Millipore; 07-473), and rabbit anti-H2A (1:500; Proteintech; 10445-1-AP). The secondary florescence antibodies used are listed below: Alexa Fluor 488, Cy3, or Cy5 (1:1000; Jackson ImmunoResearch). DAPI (2 μg/μl; Sigma; D9542) was used for nuclear counterstaining.

Western blotting and immunoprecipitation

Western blot analysis and immunoprecipitation were performed as previously described (30). Briefly, protein extracts were obtained by lysing cells/tissue in RIPA (Solarbio) buffer supplemented with protease inhibitors. For western blotting, protein extracts were separated by SDS-PAGE and transferred onto PVDF or NC membranes, followed by blotting with various antibodies. The bands were visualized using 800CW Donkey-Anti-Rabbit or Donkey-Anti-Mouse IgG as well as 680LT Donkey-Anti-Rabbit IgG or Donkey-Anti-Mouse IgG (LI-COR Biosciences) and the software Odyssey V3.0. For immunoprecipitation studies, protein extracts were incubated with anti-HA-tag magnetic beads (MBL) or anti-Flag-tag magnetic beads (MBL) overnight at 4°C. After washing six times with cold wash buffer, the bound proteins were subjected to western blot analysis.

Lentivirus production and infection

The detailed procedure for lentivirus preparation was described previously (31). For lentivirus infection, MOI (multiplicity of infection) was set to 5, and the incubation time was 12 h. After infection, neural progenitor cells were cultured continually and used for further analysis.

RNA-sequencing analysis

Total RNA was extracted from E13 telencephalic tissue of H2A.z and H2A.z mice. Then, total RNA was quality controlled and quantified using an Agilent 2100 Bioanalyzer. After converting to cDNA and building the library, high-throughput sequencing was performed using the Illumina HiSeq 2500 platform in Annoroad Genomics. The data have been deposited in NCBI’s GEO and are accessible through GEO Series accession number GSE99347.

MNase digestion assay

The assay was performed as previously described (33). In brief, cell nuclei were isolated using hypotonic buffer (10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 0.34 M sucrose, 10% glycerol, and 1 mM dithiothreitol (DTT)) and 0.1% Triton X-100. After washing in hypotonic buffer without Triton X-100, the cell nuclei were resuspended in reaction buffer (15 mM Tris–HCl, pH 7.4, 60 mM KCl, 0.25 M sucrose, 1 mM CaCl2, and 0.5 mM DTT). The digestion was carried out with MNase (New England BioLabs) at 4 U MNase per 200 μl of reaction buffer at 25°C for the indicated period of time. The reaction was terminated by adding an equal volume of 2 × TNESK buffer (20 mM Tris–HCl, pH 7.4, 0.2 M NaCl, 2 mM EDTA and 2% SDS) with freshly added proteinase K (0.2 mg/ml). The samples were then incubated overnight at 37°C, and genomic DNA was purified and separated by agarose gel electrophoresis.

Real-time polymerase chain reaction analysis

The extraction of total RNA and reverse-transcription of complementary DNA were performed as previously described (30). Quantitative real-time PCR (qRT-PCR) was performed in a 20 μl reaction mixture using SYBR qPCR master mix (Tiangen) and the 7500 real-time PCR system (Applied Biosystems). The primers used for real-time PCR were as follows:

qRT-PCR primers

ChIP-qPCR primers

Apoptosis assay

Apoptotic cells were detected by TUNEL assay using the In Situ Cell Death Detection Kit-TMR Red (Roche) according to the protocol.

ChIP–qPCR and ChIP-sequencing

ChIP analysis was performed essentially as previously described (34,35). Briefly, E13 neural progenitor cells were crosslinked with 1% formaldehyde solution (50 mM, HEPES–KOH, pH 7.5, 100 mM NaCl, 1 mM EDTA, 0.5 mM EGTA and 1% formaldehyde) for 10 min at room temperature, and the reaction was quenched with glycine. After washing twice with ice-cold PBS, cells were resuspended in lysis buffer 1 (50 mM HEPES–KOH, pH 7.5, 140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP-40, 0.25% Triton X-100, 1× protease inhibitors) for 10 min at 4°C. Nuclei were isolated at 4000 rpm for 5 min at 4°C and resuspended in lysis buffer 2 (10 mM Tris–HCl, pH 8.0, 200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 1× protease inhibitors) and then rocked gently at room temperature for 10 min. Nuclei were sonicated in lysis buffer 3 (1% SDS, 10 mM EDTA, 50 mM Tris–HCl pH 8.1, 1× Protease Inhibitor) using a Scientz-IID sonicator. Chromatin was immunoprecipitated overnight at 4°C with 50 μl Protein A beads (Dynabeads, Invitrogen) pre-coupled with 2 μg appropriate antibodies. Beads were washed three times with low-salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl, Tris–HCl, pH 8.1) and three times with high-salt buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 500 mM NaCl, Tris–HCl, pH 8.1). After reverse crosslinking was performed, DNA was eluted and purified using a DNA purification kit (Tiangen) and subjected to qRT-PCR on the ABI 7500 real-time PCR system (Applied Biosystems). ChIP-seq was performed essentially as described above. Samples were sequenced using the IlluminaHiseq™ 4000. The raw reads were mapped to the mouse reference genome (mm9) by the Clean data processing pipeline, allowing up to two mismatches. The genome ChIP-seq profiles were generated using MACS2 with only unique mapped reads. The data for H2A.z has been deposited in NCBI’s GEO and are accessible through GEO Series accession number GSE99348. The ChIP-seq raw data for H3K36me3 in E13 WT and H2A.z forebrain has been deposited in NCBI’s GEO are accessible through GEO Series accession number GSE106679.

Open field test

Male mice (8–9 weeks old) were gently placed into an open field arena (40 cm width × 40 cm length × 40 cm height) and allowed to explore freely for 5 min (36). The locomotor activity and time spent in the center and margin areas were monitored and analyzed by the Topscan behavioral analysis software (Clever Sys Inc., Reston, VA, USA). Testing occurred in a dimly lit room maintained at 30–40 lux.

Elevated-plus maze

Testing was performed as previously described (37). Mice were placed in the center of the maze and allowed to explore freely for 5 minutes. The elevated-plus maze has two open arms and two closed arms (each arm 40 cm × 9.5 cm; height of closed arms 9.5 cm). The maze stood 40 cm above the ground. The percentage of time mice spent in the open and closed arms was analyzed.

Forced swim test

Testing was performed as previously described (38). Briefly, the forced swim test was carried out in a 20-cm diameter and 35 cm high cylindrical container filled with water (20 cm high) at 25°C. Behavior over a 6-min period was recorded, and only the last 4 minutes were analyzed.

Fear conditioning

Testing was performed as previously described (39). Briefly, mice were placed in a fear conditioning apparatus chamber for 2 min to habituate them to the environment and trained with three repeats of a 28.5 s, 80 dB tone that co-terminated with a 0.6 mA, 1.5 s shock (US). Mice were then returned to their home cage. One day after conditioning, the mice underwent both contextual fear conditioning and cued tests. For the contextual fear conditioning test, the mice were exposed to the original contextual chamber for 5 min. After 2 h of the contextual test, the cued test was performed. Mice were exposed to a novel context and permitted to explore freely for 2 min, and freezing behavior in response to a CS sound lasting 60 s was scored by the SMART v2.5.21 software.

Y-Maze

Spontaneous alternation testing was carried out in the Y-maze test as previously described (40,41). The procedure consisted of two phases: the training phase and the testing phase. During the training phase, mice were allowed to explore two arms of the Y-maze with the third arm (the ‘novel arm’) blocked for 5 min. After a 15-min interval, the testing phase was performed, and mice were allowed to explore all three arms freely for 5 min. The sequence of arm choices was recorded by the Topscan behavioral analysis software.

Three-chamber social interaction test

The three-chamber apparatus consisted of a transparent acrylic box with three equally sized chambers (25 cm × 25 cm), which could be closed or opened with a 5-cm door between each dividing wall. The three-chamber social test for sociability and social recognition/preference was performed as previously described with minor modifications (37). Briefly, the social interaction behavior test comprised two parts: a habituation session and two trial sessions. For the habituation session, two cylindrical wire cages were placed in the left and right chambers, and the test mice were habituated to all three chambers of the apparatus for 5 minutes.

After the habituation period, a wire cage containing an unfamiliar mouse (Stranger 1) was placed on one side of the chamber, and an empty mesh cage was placed on the other side. The test mice were moved into the middle chamber and allowed to freely explore all three chambers for 5 min. Following this period, a novel stranger mouse (Stranger 2) was placed in the empty cage, and again the test animal was left to explore for another 5 min. The time spent in each chamber, the total duration of active contacts such as sniffing and close huddling, and the trajectory of the test mice were calculated using the Topscan behavioral analysis software.

Statistical analysis

All data represent means ± S.E.M. Statistical comparisons in this study were performed using the unpaired t-test, Kolmogorov–Smirnov test and one-way ANOVA. Differences were considered significant with a P value of P < 0.05 (*), P < 0.01(**) or P < 0.001(***). n.s., not significant. All statistical analyses were performed using Excel 2016 (Microsoft) or GraphPad Prism 6.0.

RESULTS

H2A.z is expressed in the developing brain

To explore the role of H2A.z in brain development, we first detected the expression pattern of H2A.z in the brain from E11 to P1. Western blotting showed that H2A.z was highly expressed at E11 and gradually decreased with the progression of corticogenesis (Figure 1A), indicating that H2A.z might have important functions in brain development. H2A.z and H2A.v, differing by only three amino acids at the protein level, are encoded by distinct nucleotide sequences (42). We further performed quantitative RT-PCR (qRT-PCR) using H2A.z and H2A.v specific primers. The temporal expression pattern of H2A.v in the developing cortex was similar to that of H2A.z (Supplementary Figure S1A), with a reduction in expression throughout early development. Immunostaining revealed that H2A.z was ubiquitously expressed in the developing cortex and dominantly localized to the nucleus in the VZ/SVZ at E13 (Figure 1B and C). In addition, we found that H2A.z was not only co-labeled with Nestin, Sox2 and Pax6 but also co-expressed with Tuj1 (Figure 1D; Supplementary Figure S1B and C), indicating the expression of H2A.z in neural progenitor cells and neurons in the developmental cortex. To confirm the expression of H2A.z in vitro, neural progenitor cells isolated from E12 brains were used for analysis. Immunostaining assays revealed that H2A.z was also expressed in dissociated neural progenitors. In line with the findings in vivo, the immunoreactivity of H2A.z was weaker in Tuj1-positive cells (Supplementary Figure S1C).

Figure 1.

H2A.z is expressed in the embryonic cerebral cortex. (A) Western blot analysis of H2A.z in the development of the cerebral cortex. Brains in different developmental stages were isolated and lysed for western blot analysis with the antibodies indicated. (B) An overview of E13 brain sections labeled for H2A.z and DAPI. (C) Images of brain sections of the developing cerebral cortex labeled for H2A.z and DAPI. (D) H2A.z is abundantly expressed in Nestin positive neural progenitors and co labeled with Tuj1-positive neurons in the embryonic cortex. The E15 brain sections were immunostained with anti-H2A.z and anti-Nestin or anti-Tuj1 antibodies. (E) Western blot analysis reveals that the expression of H2A.z is effectively suppressed by H2A.z-shRNAs in primary neural progenitors. (F) Graph shows that the amount of H2A.z is obviously decreased in H2A.z-shRNA lentivirus-infected primary neural progenitors. n = 5 for all samples. (G) Electroporation of H2A.z-shRNAs results in abnormal cell distribution in the developing neocortex. The electroporation was performed at E13, and the mouse was sacrificed at E16. (H) Graph shows the percentage of GFP-positive cells distributed in the VZ/SVZ, IZ and CP. n = 5 for all samples. Representative images from at least three independent experiments. Error bars represent means ± S.E.M.; Two-tailed unpaired t-test, P < 0.05 (*), P < 0.01(**) or P < 0.001(***). n.s., not significant. CP, cortical plate; IZ, intermediate zone; VZ/SVZ, ventricular/subventricular zone. Scale bar represents 100 μm (B); 20 μm (C and D); 50 μm (G).

To identify the functions of H2A.z in embryonic neurogenesis, two shRNAs were used to knock down the expression of H2A.z. Western blotting showed that endogenous H2A.z was obviously reduced in H2A.z-shRNA lentivirus-infected primary neural progenitor cells (Figure 1E and F). Next, two H2A.z-shRNAs and the control were delivered to E13 brains through in utero electroporation, and the brains were harvested at E16. H2A.z was also decreased in shRNA-electroporated cells in vivo (Supplementary Figure S1D and E). Simultaneously, exogenous Flag-H2A.z was also efficiently reduced in H2A.z-shRNA-transfected N2a cells (Supplementary Figure S1F). Upon H2A.z knockdown, we observed a consistent overall increase in the number of GFP+ cells in the VZ/SVZ where neural progenitors reside, suggesting a profound increase in proliferation (Figure 1G and H). Consistently, the number of GFP+ cells in the cortical plate (CP) was obviously decreased (Figure 1G and H). Taken together, these results suggest that H2A.z is required for cortical neurogenesis.

H2A.z regulates brain neural progenitor cell proliferation

Given that more GFP+ cells were observed in the VZ/SVZ, where neural progenitors are enriched, we next examined whether the increase in GFP+ cells was due to enhanced neural progenitor proliferation. We next used pH3 to label mitotically active cells and observed an increase in mitotic activity in H2A.z-knockdown embryos (Supplementary Figure S2A and B). We further labeled the cycling neural progenitors with BrdU for 2h before analysis. The brains immunostaining with antibody against BrdU showed that H2A.z shRNAs caused an increase in BrdU incorporation in the GFP+ cell population (Supplementary Figure S2C and D). To further explore the role of H2A.z in brain development, H2A.z mice were crossed with Nestin-Cre mice to generate H2A.z conditional knockout (H2A.z) mice. Immunostaining showed that the expression of H2A.z was depleted in E13 H2A.z brains (Supplementary Figure S2E). Of note, there was no significant difference in the number of apoptotic cells between H2A.z and H2A.z mice (Supplementary Figure S2F and G), indicating that the abnormal cell distribution is not caused by cell death. In addition to this, we observed that H2A.z knockout caused an increase in the number of pH3-positive cells in the rostral, central, and caudal areas of E16 mouse cortices (Figure 2A and B), indicating that mitotic activity was increased when H2A.z was absent. Consistently, the immunolabeling of BrdU on coronal sections from H2A.z and H2A.z embryos showed that more BrdU-incorporating cells were detected in H2A.z embryos (Figure 2C and D), suggesting that progenitor cells in S phase were also increased when H2A.z was ablated.

Figure 2.

H2A.z regulates the proliferation of NPCs during brain development. (A) Brain sections of H2A.z mice and H2A.z mice at E16 were immunostained with mitotic marker pH3 and DAPI. (B) Statistics of pH3-positive cells in rostral, central, and caudal areas of the cortex. n = 5 for all samples. (C) Representative images of E16 coronal brain sections were immunostained for Tbr2 and BrdU. BrdU was injected intraperitoneally to pregnant mice at E16 for 2 h of pulse labeling. (D) Statistic of BrdU-positive cells per 100 μm2 surface of VZ/SVZ. n = 7 for all samples. (E) Statistics of Tbr2-positive cells per 100 μm2 surface of VZ/SVZ. n = 6 for all samples. (F) Percentage of Tbr2+BrdU+ cells among all Tbr2+ cells. n = 7 for all samples. (G) Neural progenitor cells isolated from H2A.z mice form larger neurospheres. Primary neural progenitors were dissociated from E12 H2A.zor H2A.z mice and seeded at low density (50 cells/ml), then cultured in proliferated medium (suspension culture) for 3 days. (H) Statistic of the diameters of neurospheres. n = 8–9 for all samples. (I) Abnormal cell distribution is observed in H2A.z ablated neocortex. The GFP plasmid was electroporated into E13 mouse brains of H2A.z mice and H2A.z mice, and the mice were sacrificed at E16. (J) Graphs of the percentage of GFP-positive cells in the VZ/SVZ, IZ and CP. n = 8–10 for all samples. Representative images from at least three independent experiments. H2A.z, WT; H2A.z, cKO. CP, cortical plate; IZ, intermediate zone; VZ/SVZ, ventricular/subventricular zone. Error bars represent means ± S.E.M.; Two-tailed unpaired t-test, P < 0.05 (*), P < 0.01(**) or P < 0.001(***). n.s., not significant. Scale bar represents 15 μm (A); 20 μm (C); 80 μm (G); 50 μm (I).

Radial glial cells (RGs) and intermediate progenitors (IPs) are two major types of progenitors in the VZ/SVZ (43). Our in vivo data raised the question of whether H2A.z knockout affected the lineage progression from VZ to SVZ progenitors. For this purpose, we compared the proportions of Pax6+ cells and Tbr2+ cells in the H2A.z and H2A.z embryos. Although the number of Pax6+ cells was not significantly affected (Supplementary Figure S2H and I), Tbr2+ cells were increased in the VZ/SVZ regions of H2A.z knockout mice (Figure 2E). To clarify whether the increase in Tbr2+ cells is due to the proliferation of IPs, we labeled the proliferated IPs by injecting BrdU for 2h before analysis. As expected, the amplification of IPs, which were Tbr2+BrdU+, was increased, while the proliferation of RGs showed no obvious alteration after H2A.z was deleted (Figure 2F; Supplementary Figure S2J). Consistently, the expression of Tbr2 was increased at the protein and mRNA levels, while no significant variation was found in Pax6 when H2A.z was lost (Supplementary Figure S3A–C). To test whether the promoted proliferation of H2A.z knockout NPCs was cell autonomous or influenced by external cues, we isolated primary NPCs from E12 forebrain in order to form neurospheres in culture. Compared with H2A.z, the neurosphere size formed from H2A.z NPCs was obviously increased (Figure 2G and H). Then, we electroporated GFP plasmid into the embryonic brains of H2A.z mice and H2A.z mice at E13, and we harvested the brains at E16. Compared to H2A.z mice, H2A.z mice also showed an abnormal distribution of GFP-positive cells (Figure 2I and J), which was similar to the results of H2A.z knockdown. Quantification confirmed that a higher proportion of GFP-positive cells resided in the VZ/SVZ in H2A.z mice, suggesting that the loss of H2A.z promoted the ability of neural progenitor cells to maintain their proliferative potential.

In the following experiments, we electroporated the plasmid of H2A.z overexpression into WT or H2A.z brains (Supplementary Figure S3D). First, we found that overexpression of H2A.z caused decreased in the percentage of GFP-positive cells in the VZ/SVZ compared to controls. This was accompanied by a concomitant increase proportion of GFP-positive cells in the cortical plate and decreased BrdU incorporation (Supplementary Figure S3E–G). Overexpression of H2A.z in H2A.z-knockdout embryos are able to normalize the distribution of GFP+ cells to control levels as well as restore levels of BrdU incorporation (Supplementary Figure S3E–G). Hence, these data demonstrated an early intrinsic control of H2A.z on the proliferation of NPCs.

Loss of H2A.z expression impairs neurogenesis

To examine if the decreased proportion of GFP-positive cells in the cortical plate leads to decreased neuronal differentiation. we stained electroporated brains with antibody against Tuj1(Figure 3A). Quantification of GFP+Tuj1+ cells revealed that the knockdown of H2A.z led to a decrease in the percentage of GFP+Tuj1+ cells, indicating that H2A.z knockdown caused a reduction in neuronal differentiation in the developing mouse cortex (Figure 3B). Interestingly, the mRNA level of H2A.v remained unaltered upon loss of H2A.z in E13 H2A.z cortex (Supplementary Figure S4A). To confirm these variant-specific functions, we further performed in utero electroporation with sequence-specific shRNAs for H2A.z and H2A.v individually or combinationally at E13 and analyzed brains at E16. It is worth noting that loss of H2A.z, but not H2A.v, disrupts the distribution of GFP-positive cells in the developing cortex (Supplementary Figure S4B–D). Besides the distribution change, we noticed that knockdown of H2A.z and H2A.v combination did not additively increase the percentage of cells positive for neuronal marker Tuj1 (Supplementary Figure S4E). In addition to this, immunofluorescence staining of Tuj1 revealed a decrease in the spontaneous neuronal differentiation of Pax6+ neural progenitors in H2A.z-knockout embryonic brains at E16 (Figure 3C and D; Supplementary Figure S4F and G). Moreover, we performed RT-PCR and western blotting analysis of neurospheres derived from the cortex. This approach ensured that our analysis included differentiation capacity of NPCs only in a cell-autonomous manner. We quantified the mRNA or protein levels of neuronal differentiation-related markers such as NeuroD1, Tuj1 and NeuN in H2A.z-deleted neural progenitors and found that the loss of H2A.z expression led to a decrease in neuronal differentiation (Figure 3E–G), in accordance with our Tuj1 immunochemistry results. To test this directly, we injected BrdU into pregnant mice for 24 h prior to brain analysis at E16, followed by immunofluorescence staining of BrdU and Ki67. Using this analysis, we found that the percentage of cells exiting the cell cycle (the percentage of BrdU+Ki67−/BrdU+Ki67+cells) was reduced when H2A.z was deleted (Figure 3H and I), indicating that the increase in cell proliferation was accompanied by decreased cell cycle exit.

Figure 3.

H2A.z regulates the differentiation of NPCs. (A) The number of neurons is decreased in H2A.z-shRNA-electroporated brains. Neurons were labeled with anti-Tuj1 in E16 brain sections. (B) Percentage of Tuj1+GFP+ cells among GFP+ cells. n = 4 for all samples. (C) Coronal brain slices of E16 H2A.z and H2A.z mice were immunostained using anti-Tuj1. (D) Relative Tuj1+ cells in the VZ of H2A.z and H2A.z mice. n = 8 for all samples. (E) The mRNA level of NeuroD1 and Tuj1 is decreased in H2A.z-deleted neurospheres. E12 primary neural progenitor cells (NPCs) were isolated from H2A.z and H2A.z mice, and cultured in proliferation medium (suspension culture) for 3 days, and then collected the neurospheres for further analysis (E and F). Actin mRNA served as an internal control. n = 4 for all samples. (F) Western blot analysis reveals that the expression levels of the neuron markers Tuj1 and NeuN are downregulated in H2A.z-loss neurospheres. (G) Statistics of the normalized density of H2A.z, Tuj1 and NeuN. n = 6 for all samples. (H) Cell cycle exit is decreased in H2A.z mice. E16 brain sections were stained with anti-BrdU and anti-Ki67. H2A.z mice and H2A.z mice were administered BrdU (100 mg/kg) for 24 h and were euthanized at E16. (I) Percentage of cell cycle exit (BrdU+Ki67−/Ki67+) in H2A.z or H2A.z mice. n = 6 for all samples. Representative images from at least three independent experiments. H2A.z, WT; H2A.z, cKO. VZ, ventricular zone. Error bars represent means ± S.E.M.; two-tailed unpaired t-test, P < 0.05 (*), P < 0.01(**) or P < 0.001(***). Scale bar represents 50 μm (A); 20 μm (C and H).

The reduced differentiation of progenitors prompted us to examine the neuronal cell-type specification. Staining with layer-specific markers at P0 revealed that H2A.z knockdout did not affect the positioning of neurons within the cortical plate (Supplementary Figure S5A, D and G). In H2A.z neocortex at P0, we found an increase in the number of Cux1+ neurons (layers II–III) (Supplementary Figure S5A and B), along with a concomitant reduction in Tbr1+ cells (layer VI) (Supplementary Figure S5G and H). To examine whether progenitor cell cycle kinetics altered the timing of neurogenesis we pulse-labeled progenitors with BrdU at E13 then analyzed with layer-specific markers and BrdU at P0. We observed a reduction of the production of Cux1+BrdU+ neurons (Supplementary Figure S5C) and an increase in the number of Tbr1+BrdU+ cells in the H2A.z cortex (Supplementary Figure S5I). In addition, the numbers of Ctip2+ (layer V) neurons and the proportion of Ctip2+ BrdU+ neurons were not altered between WT and H2A.zcKO mice (Supplementary Figure S5D–F).

H2A.z deletion results in abnormal neuronal morphology

To test whether the loss of H2A.z affects the morphology of neurons during brain development, primary NPCs isolated from GFP plasmid-electroporated brains were differentiated for 4 or 6 days in vitro. The results showed that not only the total length of the dendrites but also the number of primary dendrites was reduced in H2A.z-deleted neurons after four days of differentiation (Figure 4A–C). Similar results were observed in H2A.z-suppressed neurons (Supplementary Figure S6A and B). After 6 days of differentiation, the reductions in dendritic length and branching number in H2A.z-loss neurons were still observed (Figure 4D and E). Related to this, we performed in utero electroporation in E13 WT or H2A.zcKO embryos with GFP plasmid and analyzed brains at P0 or P5 (Figure 4F; Supplementary Figure S6C). We found that the H2A.z-loss neurons after their migration also exhibited abnormal dendritic morphology (Figure 4F and G; Supplementary Figure S6C–F). Furthermore, we found that abnormal GFP+ cell distribution upon loss of H2A.z is obvious in early brain development (Supplementary Figure S7A-D). Nevertheless, to confirm a cell-intrinsic function of H2A.z and rule out the possible contribution of other cell types in the brain affecting the dendritic development, we performed in utero electroporation in E13 embryos with shRNAs for H2A.z and H2A.v individually or combinational and analyzed at P17 (Supplementary Figure S7E). It is worth noting that loss of H2A.z, but not H2A.v, caused a dramatic reduction in branching (Supplementary Figure S7F). Knockdown of H2A.z and H2A.v combinational did not additively cause an obvious reduction in branch number (Supplementary Figure S7F), further demonstrating H2A.v has no additive effect on dendritic complexity. Altogether, these results demonstrate that H2A.z is essential for dendritic growth.

Figure 4.

The loss of H2A.z results in abnormal morphology of neurons. (A) Representative images of neurons after 4 days of culture in vitro. GFP-expressing plasmid was electroporated into the E13 cerebral cortices of H2A.z or H2A.zcKO mice to mark neural progenitor cells. After 24 h, the GFP-positive cells were isolated and cultured in differentiation medium for 4 days. (B) Graph shows that the total dendritic length of neurons is decreased when H2A.z is deleted. n = 8 cells from three samples. (C) Statistics show that the number of primary dendrites is reduced upon H2A.z deletion. n = 8 cells from three samples. (D) Images of neurons cultured in differentiation medium for 6 days. (E) Sholl graphs of dendrites of neurons electroporated with GFP alone in H2A.z or H2A.z mice. The GFP-positive cells were isolated and cultured in differentiation medium for 6 days. H2A.z, n = 6 cells; H2A.z = 6 cells. (F) High-magnification confocal images of upper layer show that H2A.z knockout results in abnormally branched processes compared to the control. GFP plasmid was electroporated into the H2A.z or H2A.z embryonic brain at E13 and harvested at P0. (G) Quantification of dendritic numbers in H2A.z and H2A.zcKO cortices. n = 6 for all samples. Representative images from at least three independent experiments. H2A.z, WT; H2A.z, cKO. Error bars represent means ± S.E.M., Kolmogorov–Smirnov test or two-tailed unpaired t-test, P < 0.05 (*), P < 0.01(**). Scale bar represents 15 μm (A and D); 10 μm (F).

Loss of H2A.z in developing brain results in behavioral abnormalities in adult mice

The H2A.z mice can survive to adulthood and show a reduced bodyweight in comparison to the control mice (Supplementary Figure S8A–D). Furthermore, the brain weights of P0 H2A.z mice were indistinguishable from those of control mice (Supplementary Figure S8E and F). However, significant reductions in brain weights (by ∼15%) were observed in adult brain (Supplementary Figure S8G and H). To test whether the deletion of H2A.z during cortical development results in behavioral deficits in adult mice, H2A.z mice and their littermate H2A.zfl/fl mice were first subjected to the open field test. Compared with the controls, H2A.z mice spent less time in the central area, although they showed no apparent difference in motility (Figure 5A–C). These results suggested that exploratory behavior was reduced in H2A.z mice. Next, we employed the elevated-plus maze to measure whether the H2A.z mice showed anxiety-like behavior. Compared with the control, H2A.z mice spent less time in the open arms, while they spent more time in the closed arms of the maze (Figure 5D–F). These data further supported that the loss of H2A.z resulted in reduced exploratory behavior in adult mice. To test whether H2A.z mice were abnormal in social interaction, three-chamber social arena assays were performed (Supplementary Figure S9A–D). In these assays, H2A.zmice spent more time in the chamber where a novel stranger mouse (Stranger 2) was placed in the cage, whereas H2A.z mice did not display an apparent preference (Figure 5G). Furthermore, H2A.zmice tended to interact with the mouse in the novel cage, while the H2A.z mice were inactive in terms of contact with the novel mouse (Figure 5H). These data suggest that the loss of H2A.z results in deficiencies in social interaction. Next, the forced swim test was used to detect whether H2A.z mice had depressive-like states. Compared with the control, the immobile time of H2A.z mice was increased, indicating that H2A.z mice were prone to depression (Figure 5I). To evaluate the learning and memory of H2A.z mice, fear conditioning tests and Y-maze tests were performed. The results revealed that the freezing percentage of the H2A.z mice was decreased compared with the control mice, indicating that the loss of H2A.z impaired the associative memory (Figure 5J and K). The Y-maze test showed that the H2A.z mice were less prone to enter new arms and spent less time in the new arms, indicating that the working memory was affected (Figure 5L and M; Supplementary Figure S9E). Taken together, the results from this range of behavior tests indicate that H2A.z mice exhibit behavioral deficits in exploratory activity, social behavior and spatial learning and memory. In addition, H2A.z mice are prone to anxiety and depression.

Figure 5.

The loss of H2A.z results in behavioral deficits. (A) Representative tracing pathway of H2A.z and H2A.z mice in the open field test. (B) The total traveling distance was not different between H2A.z and H2A.z mice. (C) Time spent in the center is reduced in H2A.z mice. (D) Representative tracing pathway of H2A.z and H2A.z mice in the elevated-plus maze test. (E) H2A.z mice spent less time in the open arms. (F) H2A.z mice spent more time in the closed arms. (G) Social interaction behavior is impaired in H2A.z-null mice. H2A.z mice spent similar time in the chamber with the stranger 1 or stranger 2 mouse, while H2A.z mice spent more time in the chamber with stranger 2 than with the stranger 1 mouse. (H) Close social interaction is affected by H2A.z deletion. H2A.z mice did not spend more time in close interaction with stranger 2 than with stranger 1. (I) The immobile time of H2A.z mice was more than that of H2A.z mice during the forced swim test. (J) The percentage of freezing in H2A.z mice was less than in H2A.z mice, although they were explored in the same background. (K) Freezing response was also reduced in H2A.z mice in the cued test. (L) Compared with H2A.z mice, the entries of H2A.z mice into new arms were decreased, while the entries into old arms were increased. (M) Compared with H2A.z mice, H2A.z mice spent less time in the new arms and more time in the old arms. All mice were male, 8–10 weeks old. H2A.z, WT, n = 11–14; H2A.z, cKO, n = 9–16. Error bars represent means ± S.E.M.; one-way ANOVA, P < 0.05 (*), P < 0.01(**) or P < 0.001(***). n.s., not significant. Analysis was performed by an experimenter blinded to experimental conditions.

H2A.z regulates neurogenesis by targeting Nkx2-4

To gain deeper insight into how H2A.z regulates embryonic neurogenesis, RNA sequencing (RNA-seq) was performed to analyze the genome-wide changes resulting from H2A.z deletion at E13 (Supplementary Figure S10A). Gene ontology analysis showed that the downregulated genes were enriched in terms related to transcription and cortex development, such as regulation of transcription, cerebellar cortex formation, forebrain development, and cerebellum development (Figure 6A). Upregulated genes showed a significant enrichment of terms involved in the negative regulation of cell differentiation, neuron migration, and cognition (Figure 6B). These results reflected the importance of H2A.z in cortical development. Because the genomic localization of H2A.z might reveal its essential role during early neurogenesis, ChIP-seq was performed on developing mouse E13 forebrain as previously described (35). As expected, H2A.z was prone to bind the promoter (32.4%) and the intergenic regions (32.1%) (Figure 6C). The analysis of biological processes of H2A.z-bound genes indicate H2A.z play roles in pathways relating to cell cycle, DNA replication and chromatin disassembly (Figure 6D). In addition, we examined the overlap between H2A.z ChIP-seq peaks with differentially expressed genes in mRNA-seq data of H2A.z deletion and subdivided GO terms for H2A.z-bound genes into the 2 classes: upregulated or downregulated group upon deletion of H2A.z (Figure 6E). Downregulated genes enriched in H2A.z ChIP-seq peaks were involved in processes relating to microtubule cytoskeleton organization, dendritic spine morphogenesis and growth (Figure 6F). Upregulated genes are involved in regulation of glucose metabolic process, reproductive process, stem cell population maintenance and mitotic division (Figure 6G). We next investigated how H2A.z deletion affected the molecular program controlling cortical neurogenesis. We focused on studying Nkx2–4, because it showed the reduction in E13 H2A.z-deficient forebrain (Figure 6H; Supplementary Figure S10B–D; Supplementary Table S1). Furthermore, it was reported Nkx2–4 may be involved in the regionalization of the forebrain and may affect embryonic neurogenesis (44).

Figure 6.

H2A.z regulates neurogenesis by targeting Nkx2–4. (A) GO analysis for downregulated genes from E13 forebrain RNA-seq data sets in H2A.z-loss mice. (B) Analysis of gene ontology of biological functions for upregulated genes in H2A.z-depleted E13 forebrain. (C) Genome-wide distribution of H2A.z occupancies in E13 forebrain. (D) GO analyses of H2A.z binding genes. (E) Venn diagram showing the overlap between H2A.z bound genes by ChIP-seq with differentially expressed genes following H2A.z knockout (RNA-seq). (F) GO analyses for H2A.z-bound genes which are downregulated upon deletion of H2A.z. (G) GO analyses for H2A.z-bound genes which are upregulated upon deletion of H2A.z. (H) Volcano plots illustrate differentially expressed genes. Red dots are upregulated genes, and green dots are downregulated genes compared to the control. Nkx2–4 is one of the significantly differentially expressed genes. (I) H3K36me3 occupancy for all genes in E13 WT and H2A.z forebrain. (J) RPKM values for all genes in E13 WT and H2A.z forebrain.

A key aspect of the regulation of gene expression during neurogenesis occurs at the level of chromatin states (8). We consider that the coordination of transcription with histone modifications is critical for gene regulation. We first detected the methylations of histone upon H2A.z knockout. Notably, the level of H3K36me3 was reduced, while no obvious variation was detected in the levels of H3K27me3, H3K9me3, and H3K4me3 in the E13 brain, suggesting that H2A.z incorporation can affect global H3K36me3 modification patterns (Supplementary Figure S10E and F). In accordance with previous observations (34), H3K36me3 ChIP-seq signals were distributed in both the body and the promoter of genes (Supplementary Figure S10G). We further evaluated the patterns of H3K36me3 enrichment in the E13 H2A.zcKO forebrain and found that H3K36me3 is decreased throughout the whole genome (Figure 6I, Supplementary Figure S10H). Furthermore, we found the amount of H3K36me3 in the gene body and the promoter regions of Nkx2–4 was decreased in response to H2A.z loss (Supplementary Figure S10I). These data indicate that H2A.z mediates transcriptional initiation of Nkx2–4.

To confirm whether H2A.z regulates embryonic neurogenesis through targeting Nkx2–4, we further investigate the role of Nkx2–4 in cortical development. Using a specific antibody against Nkx2–4, we examined Nkx2–4 protein expression in the developing forebrain (E12–E18). Nkx2–4 was localized to the nucleus in neural progenitors in the VZ/SVZ and newborn neurons in the CP. Furthermore, we found Nkx2–4 protein was broadly expressed in Tuj1+ neurons, Tbr2+ IPs and Sox2+ progenitors (Supplementary Figure S11A–C). Similarly, the GFP+ cells in the VZ/SVZ were increased, while the number of GFP+ cells in the cortical plate was obviously reduced when the expression of Nkx2–4 was downregulated (Supplementary Figure S11D-F). Immunostaining of pH3 showed that more pH3 signals were observed in the VZ/SVZ regions in the Nkx2–4-suppressed cortex (Supplementary Figure S11G). Consistently with the results for H2A.z, the percentage of Tuj1-positive cells was also decreased in Nkx2–4 knockdown samples (Supplementary Figure S11H and I).

Considering that Nkx2–4 is an important gene downstream of H2A.z, we next detected whether the overexpression of Nkx2–4 could rescue the developmental deficiency caused by H2A.z downregulation. For this purpose, H2A.z- or Nkx2–4-overexpression plasmids were co-electroporated with H2A.z shRNA into E13 brains, which were harvested at E16. We found that co-expression of Nkx2–4 with H2A.z-sh1 restored the distribution of GFP+ cells, the percentage of GFP+ Tbr2+ cells and mitotic activity, as well as the percentage of GFP+ Tuj1+ cells (Supplementary Figure S12A–H). To determine whether forced Nkx2–4 expression could also rescue the dendritic abnormalities associated with H2A.z loss-of-function, we performed in utero electroporation at E13 with control, H2A.z-sh1 or co-expressed either control empty vector or Nkx2–4 overexpression constructs, and analyzed brains at P17. The reduction in branching of upper layer neurons was partially rescued upon Nkx2–4 co-expression (Supplementary Figure S12I and J). Thus, Nkx2–4 is an important downstream gene of H2A.z during embryonic neurogenesis.

H2A.z and Setd2-mediated H3K36me3 modification regulate Nkx2-4

Having shown that H2A.z mediated transcriptional initiation of Nkx2–4 during early neurogenesis, we set out to address the mechanism by which H2A.z regulates this pathway. Setd2 is a histone methyltransferase that specifically trimethylates Lys-36 of histone H3 (Supplementary Figure S13A and B) (45). We further detected the relationship between H2A.z and Setd2. A co-immunoprecipitation (co-IP) experiment showed that H2A.z could interact with Setd2 (Figure 7A–C and Supplementary Figure S13C). Higher H3K36me3 levels were also detected in the overexpression of both H2A.z and Setd2 than in H2A.z or Setd2 overexpression alone, indicating that H2A.z promotes the trimethylation of Setd2 on Lys-36 of histone H3 (Figure 7B and C). Western blotting showed that H3K36me3 levels and Nkx2–4 expression were reduced when both H2A.z and Setd2 were knocked down (Figure 7D). Consistently, the expression of Nkx2–4 and the level of H3K36me3 were both increased when H2A.z and Setd2 were overexpressed, while no obvious variation was found when H2A was overexpressed, suggesting that H2A.z and Setd2 coordinately increase H3K36me3 levels and Nkx2–4 expression (Figure 7E). Simultaneously, H2A.z overexpression could rescue the reduced H3K36me3 and Nkx2–4 levels caused by H2A.z knockdown, while H2A could not, indicating that the regulatory effect of H2A.z on H3K36me3 modification and Nkx2–4 expression is specific (Figure 7F). In addition, the regulatory effect of H2A.z on neurogenesis is also specific (Supplementary Figure S13D and E). Consistently with the ChIP-seq results, the binding enrichment of H2A.z in the Nkx2–4 promoter is higher in 300 bp upstream of TSS (Supplementary Figure S13F). In addition, the chromatin accessibility is reduced upon H2A.z deletion (Supplementary Figure S13G). To show that the regulatory effect of H2A.z on the Nkx2–4 promoter is via interaction with Setd2, the binding enrichment of H2A.z and Setd2 on the Nkx2–4 promoter was detected by ChIP experiments. When H2A.z was knocked down, the binding enrichment of Setd2 in the Nkx2–4 promoter region was reduced (Figure 7G). When Setd2 was knocked down, the accumulation of H2A.z in Nkx2–4 promoter region was not altered (Figure 7H). In contrast, the accumulation of H2A.z in Nkx2–4 promoter region was increased when Setd2 was overexpression (Figure 7I). Furthermore, the accumulation of H3K36me3 in the Nkx2–4 promoter region was abolished in both H2A.z-silenced and Setd2-OE NPCs (Figure 7J). These results indicate that H2A.z plays essential roles in recruiting Setd2 onto the Nkx2–4 promoter.

Figure 7.

H2A.z and Setd2-mediated H3K36me3 on Nkx2–4 promoter regulate the expression of Nkx2–4. (A) The interaction between H2A.z and Setd2 is verified by co-immunoprecipitation experiments. HA-Setd2-expressing plasmids were transfected into N2A cells. Three days later, the transfected cells were subjected to immunoprecipitation using HA beads. Bound proteins were detected by western blotting with the antibodies indicated. (B) H2A.z promotes Setd2-mediated H3K36me3 modification. HA-Setd2 plasmids were co-transfected with Flag-H2A.z or control into N2A cells. The cell lysates were immunoprecipitated with anti-HA, and bound proteins were detected by western blotting with the antibodies indicated. (C) Higher H3K36me3 levels were pulled down in both the H2A.z and Setd2 overexpression groups. Flag-H2A.z plasmids were co-transfected with HA-Setd2 or control into N2A cells. The cell lysates were immunoprecipitated with anti-Flag and blotted with the antibodies indicated. (D) The levels of H3K36me3 and Nkx2–4 are decreased when both H2A.z and Setd2 are silenced. Primary neural progenitors were infected with lentivirus indicated (D, E and F). Three days later, the cell lysates were analyzed by western blotting with the antibodies indicated. Graph below depicts quantification for Nkx2–4 and H3K36me3, normalized to actin. (E) Higher levels of Nkx2–4 and H3K36me3 are detected when H2A.z and Setd2 are overexpressed, while no obvious increase is investigated when H2A is overexpressed. Graph below depicts quantification for Nkx2–4 and H3K36me3, normalized to actin. (F) Overexpression of H2A.z, but not H2A, can rescue the downregulation of Nkx2–4 and H3K36me3 caused by H2A.z-sh1. Graph below depicts quantification for Nkx2–4 and H3K36me3, normalized to actin. (G) ChIP in primary neural progenitors overexpressing Setd2 with or without H2A.z-sh1, as indicated, performed with antibodies indicated. (H) ChIP in primary neural progenitors overexpressing H2A.z with or without Setd2-shRNA, as indicated, performed with antibodies indicated. (I) ChIP in primary neural progenitors overexpressing H2A.z with or without Setd2, as indicated, performed with antibodies indicated. (J) ChIP analysis of enrichment of H3K36me3 on promoters of Nkx2–4 in primary neural progenitors infected with control, Setd2 or Setd2 together with H2A.z-sh1 lentivirus. A total of 5% of input was loaded for western blotting. Representative images from at least three independent experiments. ChIP enrichment is expressed as percentage of input. n = 3–4 for all samples. Error bars represent means ± S.E.M.; two-tailed unpaired t-test, P < 0.05 (*), P < 0.01(**) or P < 0.001(***). n.s., not significant.

DISCUSSION

Embryonic neurogenesis is a critical process for brain development, which is regulated in a spatially and temporally restricted manner (2). Dysregulation of any of these steps might result in abnormal brain functions, thereby contributing to the occurrence of neurodevelopmental disorders (6,46). H2A.z, an evolutionarily conserved histone variant, not only plays essential roles in chromatin organization and stability but also is critical for the epigenetic regulation of gene expression (23). Previous reports show that H2A.z has important functions in key biological processes (23,25). However, the role of H2A.z in embryonic neurogenesis remains largely unknown. Here, we take advantage of H2A.z conditional knockout mice to explore the functions of H2A.z in neural progenitor proliferation and differentiation, neuronal morphology, and mouse behavior, as well as the underlying mechanisms. We find that H2A.z interacts with Setd2 to increase the H3K36me3 levels of the Nkx2–4 promoter, thereby increasing the expression of Nkx2–4. Taken together, the epigenetic regulation of H2A.z is essential for brain development.

We find that H2A.z is expressed in the embryonic cerebral cortex and gradually decreased with brain development. Our finding demonstrates that knockdown or knockout of H2A.z promotes the proliferation of NPCs and inhibits neuronal differentiation. Current available data indicate that H2A.z acts as a critical factor that mediates the activation of differentiation-associated genes (18,47). The reason for these differences may be that H2A.z regulates cell proliferation in a cell type-specific manner (48,49).

Neuronal dendritic arborization is acquired soon after neurons have entered their destined cortical layer in the brain (50). Our study demonstrates that H2A.z is crucial for dendritic morphology. We show that the dendritic morphology of neurons is impacted by the embryonic removal of H2A.z. Dendritic morphogenesis is crucial for the functional integration of neurons from neural progenitors and the establishment of neuronal connectivity (51). Abnormal dendrite morphogenesis leads to neuronal dysfunctions associated with neurodevelopmental disorders such as schizophrenia and autism (52–54). Furthermore, using H2A.z mice as a model, we demonstrate that in addition to the neuronal differentiation defects, the loss of H2A.z causes behavioral defects resembling the characteristics of patients with autism and schizophrenia. Interestingly, the H2A.z gene has been suggested as a risk factor for schizophrenia. Thus, our study provides further evidence for the recognition of H2A.z as a risk gene for neurodevelopmental disorders. Earlier studies have implicated H2A.z as a negative regulator of the consolidation of recent and remote memory (55). In our study, freezing responses were found to be reduced in H2A.z mice. This discrepancy could be due to differences in the timing and level of H2A.z suppression. In addition to this, we cannot rule out the possibility of changes of other type of cells such as astrocytes and oligodendrocytes in H2A.z brain. It is fascinating to consider how loss of H2A.z function only in astrocytes or oligodendrocytes might directly or indirectly contribute to the altered behavior.

Next, we examined dysregulated genes in H2A.z via the RNA-sequencing analysis of E13 brain samples. We find that Nkx2–4 RNA expression is reduced in H2A.z relative to H2A.z. The homeodomain (HD) protein Nkx2–4 is a member of the Nkx2 family, but less is known about the specification of Nkx2–4 in the development of the brain. Herein, we show that the depletion of Nkx2–4 promotes the proliferation and inhibits the differentiation of neural progenitors. Further, increasing Nkx2–4 levels promotes the differentiation of neural progenitors and rescues the impaired neurogenesis resulting from H2A.z knockdown. Analogous to other members (56), our findings suggest that Nkx2–4 has key roles in brain development.

Previously, the genome-wide binding properties and activity of H2A.z in the developing brain were unknown. It is reported that H2A.z is deposited at enhancers and promoters that exhibit bivalent signatures with both activating (H3K4me3; H3K36me3) and repressive (H3K27me3; H3K9me3) histone modifications (23). Indeed, our data show the promoter enrichment of H2A.z with both up- and downregulation of gene expression in H2A.z, suggesting that H2A.z binding shows context-dependent activity. The transcriptional changes in Nkx2–4 might be due to direct regulation by H2A.z or might be mediated indirectly via the functions of H2A.z-regulated genes. We then examine the relationship between H2A.z binding and chromatin state. In this regard, we compared histone modifications results from H2A.z and H2A.z conditional knockout at E13 to reveal H2A.z as an epigenetic regulator. The level of H3K36me3 was reduced when H2A.z was depressed. Setd2 (SET-domain containing 2), also known as KMT3A and HYPB, is recognized as a histone methyltransferase responsible for the trimethylation of histone H3 at Lys 36 (H3K36me3) (45). It is now well established that H3K36me3 is an epigenetic marker associated with gene transcription (57). Local recruitment of Setd2 by H2A.z could explain the changes in H3K36me3 associated with Nkx2–4 in H2A.z. We found that the loss of H2A.z impairs the recruitment of Setd2 and is correlated with reduced levels of H3K36me3 and Nkx2–4 protein. Furthermore, H3K36me3 is decreased throughout the whole genome, as well as at the transcription start site of Nkx2–4 in response to H2A.z loss. Thus, H2A.z appears to potentiate chromatin accessibility by modulating nucleosome occupancy at regulatory regions. Consistent with these observations, knockout of H2A.z results in decreased chromatin accessibility by micrococcal nuclease (MNase). In contrast to H2A.z, H2A cannot rescue the deficiency in embryonic neurogenesis caused by H2A.z-shRNAs. In addition, H2A.z, but not H2A, can rescue the downregulation of H3K36me3 and Nkx2–4 resulting from H2A.z knockdown. These results indicate that the regulatory effect of H2A.z on embryonic neurogenesis is unique and cannot be replaced by H2A. Here, we elucidate the molecular functions of H2A.z by integrating its genome-wide binding, regulation of gene expression, and epigenetic state to delineate the mechanisms by which H2A.z instructs cortical development.

Taken together, our results emphasize the essential role of the epigenetic regulation of H2A.z in embryonic neurogenesis and highlight the important functions of H2A.z in brain development. We find that H2A.z regulates the recruitment of Setd2 to promote the H3K36me3 levels of the Nkx2–4 promoter, thereby increasing the expression of Nkx2–4, which ultimately controls neuron production and the expansion of the cerebral cortex. Furthermore, the behavioral abnormalities in H2A.z-null mice provide a possible explanation for the deficits found in patients with schizophrenia (27). In summary, our findings reveal the role of H2A.z in embryonic neurogenesis and dendrite morphogenesis, and how H2A.z loss may lead to various developmental defects and brain disorders (Supplementary Figure S14).

DATA AVAILABILITY

Data have been deposited in NCBI’s GEO and are accessible through GEO Series accession number GSE99347 (RNA-sequencing analysis), GSE99348 (H2A.z), GSE106679 (H3K36me3 in E13 WT and H2A.zcKO forebrain).

Click here for additional data file.

Gene	Foreword	Reverse
Tbr2	CCGGGACAACTACGATTCCA	ACCTCCAGGGACAATCTGATG
Tuj1	CCAGTGCGGCAACCAGATAGG	AAAGGCGCCAGACCGAACACT
Ccnd1	AGTGCGTGCAGAAGGAGATT	CACAACTTCTCGGCAGTCAA
NeuroD1	CTCAGCATCAATGGCAACTTCTC	GACTCGCTCATGATGCGAATGCC
H2A.z	GGCCGTATTCATCGACACCT	GACGAGGGGTGATACGCTTT
H2A.v	CTGGGCTCCAGTTTCCTGTG	AGGTACTCCAGAATTGCGGC
Nkx2–4	GCTACTCGTCAATCTCCAGG	CAGGTACTTCTGCTGCTTGAA
Actin	AGGTCATCACTATTGGCAACGA	CACTTCATGATGGAATTGAATGTAGTT

Target (Nkx2–4) site	Forward	Reverse
-2k	GCAGTTCCAGCATAAGTGACC	CATACTGTGGGACAAAATGCCAG
-1k	AGAGAGCTGTCGTTTTAACCTG	GTTTGTAATGGATCTTTTTGGTGGT
-300 bp	GCCTCCGTTATCTGGAAGGG	TCCAGTGCCAGGTTTACGAC