Abstract
DNA repair is crucial for genome stability in the developing cortex, as somatic de novo mutations cause neurological disorders. However, how DNA repair contributes to neuronal development is largely unknown. To address this issue, we studied the spatiotemporal roles of DNA polymerase β (Polβ), a key enzyme in DNA base excision repair pathway, in the developing cortex using distinct forebrain-specific conditional knock-out mice, Emx1-Cre/Polβfl/fl and Nex-Cre/Polβfl/fl mice. Polβ expression was absent in both neural progenitors and postmitotic neurons in Emx1-Cre/Polβfl/fl mice, whereas only postmitotic neurons lacked Polβ expression in Nex-Cre/Polβfl/fl mice. We found that DNA double-strand breaks (DSBs) were frequently detected during replication in cortical progenitors of Emx1-Cre/Polβfl/fl mice. Increased DSBs remained in postmitotic cells, which resulted in p53-mediated neuronal apoptosis. This neuronal apoptosis caused thinning of the cortical plate, although laminar structure was normal. In addition, accumulated DSBs also affected growth of corticofugal axons but not commissural axons. These phenotypes were not observed in Nex-Cre/Polβfl/fl mice. Moreover, cultured Polβ-deficient neural progenitors exhibited higher sensitivity to the base-damaging agent methylmethanesulfonate, resulting in enhanced DSB formation. Similar damage was found by vitamin C treatment, which induces TET1-mediated DNA demethylation via 5-hydroxymethylcytosine. Together, genome stability mediated by Polβ-dependent base excision repair is crucial for the competence of neural progenitors, thereby contributing to neuronal differentiation in cortical development.
SIGNIFICANCE STATEMENT DNA repair is crucial for development of the nervous system. However, how DNA polymerase β (Polβ)-dependent DNA base excision repair pathway contributes to the process is still unknown. We found that loss of Polβ in cortical progenitors rather than postmitotic neurons led to catastrophic DNA double-strand breaks (DSBs) during replication and p53-mediated neuronal apoptosis, which resulted in thinning of the cortical plate. The DSBs also affected corticofugal axon growth in surviving neurons. Moreover, induction of base damage and DNA demethylation intermediates in the genome increased DSBs in cultured Polβ-deficient neural progenitors. Thus, genome stability by Polβ-dependent base excision repair in neural progenitors is required for the viability and differentiation of daughter neurons in the developing nervous system.
- cortical development
- DNA demethylation
- DNA double-strand break
- DNA repair
- neuronal apoptosis
- neuronal differentiation
Introduction
Cortical developmental disorders, such as autism spectrum disorders, are attributable to multiple gene mutations (Poduri et al., 2013). Recent whole-genome sequence technology revealed that CNS neurons have somatic de novo mutations involving single nucleotide variation, copy number variation, and retrotransposition (Poduri et al., 2012; McConnell et al., 2013; Bundo et al., 2014). Further, increasing evidence suggests that even a very small number of cortical neurons with mutations can disrupt the function of cortical circuits (Poduri et al., 2013). Thus, genome stability is essential for normal cortical development and function.
DNA repair is necessary to maintain genome stability. Human genetic diseases caused by DNA repair defects include immune dysfunction and cancer predisposition (Lindahl and Wood, 1999). In the nervous system, DNA repair deficiency can result in microcephaly, ataxia, mental retardation, and neurodegeneration (Caldecott, 2008; McKinnon, 2013). It is likely that these disorders are attributable to neuronal cells with extensive DNA damage; indeed, mice deficient in several DNA repair enzymes exhibit neuronal apoptosis and structural abnormalities in the developing brain (Gao et al., 1998; Deans et al., 2000; Gu et al., 2000; Sugo et al., 2000). An intriguing question is how these DNA repair enzymes function in normal brain development.
DNA polymerase β (Polβ), a core component of the base excision repair (BER) pathway, has been shown to be involved in development of the nervous system rather than that of other organs (Sobol et al., 1996; Sugo et al., 2000; Wilson et al., 2000; Niimi et al., 2005). The most characteristic aspect has been shown in the developing cortex of Polβ-deficient mice (Polβ−/−); extensive apoptosis occurs in the intermediate zone (IZ) and cortical plate (CP) rather than the ventricular zone (VZ) (Sugo et al., 2000, 2004). Because it is known that Polβ-dependent BER can remove spontaneous apurinic/apyrimidic sites and reactive oxygen species-induced oxidative DNA base damage in aerobic cells (Lindahl, 1993; Sobol et al., 1996), such DNA damage likely increases susceptibility to genome instability in Polβ-deficient cortical cells. In addition to the accidental and pathological damage, physiological damage may also be repaired by the Polβ-dependent BER machinery. Indeed, the BER pathway is involved in active DNA demethylation process, which is necessary for epigenetic regulation of gene expression (Bhutani et al., 2011). However, how Polβ-dependent BER contributes to the development of the nervous system remains unknown.
In the present study, we investigated the roles of Polβ in the developing cortex using forebrain-specific conditional knock-out mice. We found that Polβ deficiency led to increased DNA double-strand breaks (DSBs) during S phase in neural progenitors, and subsequently p53-mediated neuronal apoptosis. The DSBs remained unrepaired in surviving postmitotic neurons, causing corticofugal axon growth defects. Induction of pathological base damage and epigenetic base modification markedly enhanced DSB formation in cultured Polβ-deficient neural progenitors. These findings demonstrate that genome stability mediated by Polβ-dependent BER suppresses DSB formation in neural progenitors and underlies neuronal differentiation in cortical development.
Materials and Methods
Mice.
To generate Polβdel/del mice, Polβfl/fl mice (Gu et al., 1994) were crossed with Sycp1-Cre mice (Noguchi et al., 2009), and then Sycp1-Cre/Polβfl/+ male mice were crossed with C57BL/6J background female mice. p53-deficient mice (Tsukada et al., 1993) were also crossed with Polβdel/+ mice. To generate Emx1-Cre/Polβfl/fl mice and Nex-Cre/Polβfl/fl mice, we crossed Polβfl/fl mice with Emx1-Cre (Iwasato et al., 2000) and Nex-Cre mice (Goebbels et al., 2006), respectively. Noon of the day on which the vaginal plug was detected was designated embryonic day 0.5 (E0.5). All experiments were conducted under the guidelines for laboratory animals of the Graduate School of Frontier Biosciences, Osaka University. The protocol was approved by the Animal Care and Use Committee of the Graduate School of Frontier Biosciences, Osaka University.
In situ hybridization.
cDNA fragment encoding Polβ (NM_011130.2, nucleotides 31-1079) and Pax6 (NM_013627.1, nucleotides 161-1559) was subcloned into pGEM-T Easy vector (Promega). The preparation of RNA probes and in situ hybridization were performed as described previously (Zhong et al., 2004).
In utero electroporation.
pCAGGS-EGFP plasmid DNA was purified with the PureLink HiPure Plasmid Maxiprep Kit (Invitrogen), then dissolved in PBS. In utero electroporation was performed on E13.5 embryos (Fukuchi-Shimogori and Grove, 2001). Pregnant mice were deeply anesthetized with isoflurane (Wako) by using inhalation anesthesia equipment (KN-1071-1, Natume). A total of 1 μg of plasmid solution was injected to the lateral ventricle with a glass micropipette connected to an injector (IM-30, Narishige). Electric pulses were delivered with tweezers electrodes (LF650P1, BEX) connected to an electroporator (CUY21, BEX). Five 35 V pulses of 50 ms duration were applied at intervals of 100 ms.
Immunostaining.
The brains were fixed with 2 or 4% PFA in 0.1 m sodium phosphate buffer, pH 7.4, equilibrated with 25% sucrose-PBS, frozen in OCT compound (Sakura Finetech), and cut into 4, 10, or 20 μm sections using a cryostat (CM1850, Leica). The sections were permeabilized and blocked for 1 h at room temperature in buffer G (0.1% or 1.0% Triton X-100, 5% normal goat serum, Vector Laboratories in PBS) or buffer D (0.1% or 1.0% Triton X-100, 2% normal donkey serum, Jackson ImmunoResearch Laboratories in PBS). The sections were incubated overnight at 4°C with the following antibodies in buffer G or buffer D: rabbit polyclonal anticleaved caspase-3 (Asp175) (9661S, Cell Signaling Technology) at 1:250, mouse monoclonal anti–neuron-specific β-III tubulin (Tuj1) (MAB1195, R&D Systems) at 1:1000, goat polyclonal anti-Nurr1 (AF2156, R&D Systems) at 1:50, rabbit polyclonal anti-Tbr1 (ab31940, Abcam) at 1:500, rat monoclonal anti-Ctip2 (ab18465, Abcam) at 1:800, rabbit polyclonal anti-CDP (Cux1) (sc-13024, Santa Cruz Biotechnology) at 1:200, rat monoclonal anti-GFP (04404-84, Nacalai Tesque) at 1:1000, goat polyclonal anti-Contaction-2 (TAG-1) (AF4439, R&D Systems) at 1:200, rabbit polyclonal anti-histone H2AX phosphor Ser139 (γH2AX) (39117, Active Motif) at 1:200, rabbit monoclonal anti-Ki67 (RM-9106-S0, Thermo Scientific) at 1:200, or rabbit polyclonal anti-53BP1 (GTX102595, GeneTex) at 1:100. For Polβ immunostaining, the sections were fixed with 2% PFA in PBS for 5 min at room temperature, then washed with PBS for 10 min three times. The sections were treated with 1.0% SDS in PBS for 10 min at room temperature, then washed with PBS for 10 min three times. The sections were incubated with 10% normal goat serum in PBS for 1 h at room temperature, then incubated overnight at 4°C with rabbit polyclonal anti-DNA polymerase β antibody (ab26343, Abcam) diluted in Can Get Signal immunostain solution A (Toyobo) at 1:500. Primary antibodies were detected by incubation with Alexa488-conjugated antimouse IgG (A-11029, Invitrogen), Alexa488-conjugated antirabbit IgG (A-11034, Invitrogen), Alexa488-conjugated antirat IgG (A-11006, Invitrogen), Cy3-conjugated antimouse IgG (AP192C, Millipore), Cy3-conjugated antirabbit IgG (AP182C, Millipore), Cy3-conjugated antirat IgG (AP136C, Millipore), or Cy3-conjugated antigoat IgG (705-165-147, Jackson ImmunoResearch Laboratories), in all cases at 1:400 in buffer G or buffer D for 2 h at room temperature. The sections were washed with PBS for 10 min three times. Nuclei were stained with 0.1% DAPI (Sigma) in a mounting medium containing 50% glycerol and 2.3% 1,4-diazobicyclo[2.2.2]octane (Sigma) in 50 mm Tris-HCl, pH 8.0.
Cultured cells were fixed with 4% PFA in PBS for 10 min at room temperature. The cells were permeabilized and blocked for 1 h in buffer G at room temperature. They were incubated with rabbit polyclonal anti-γH2AX antibody at 1:200, rabbit polyclonal anti-53BP1 antibody at 1:100, or mouse monoclonal anti-nestin antibody (MAB353, Millipore) at 1:200 in buffer G overnight at 4°C. Primary antibodies were visualized by incubation with Alexa488-conjugated antimouse IgG, Alexa488-conjugated antirabbit IgG, and Cy3-conjugated antimouse IgG at 1:400 in buffer G for 2 h at room temperature. The cells were mounted with DAPI-containing medium as above.
5-Ethynyl-2′-deoxyuridine (EdU) labeling.
Pregnant mice were injected intraperitoneally with 150 μl of EdU (2.5 mg/ml in PBS, Invitrogen), and embryos were harvested at 30 min after the injection or at E18.5. EdU was visualized using a Click-iT EdU AlexaFluor-555 imaging kit (Invitrogen) according to the manufacturer's recommendations.
Quantitative RT-PCR.
Total RNA was extracted from E14.5 Polβfl/fl and Emx1-Cre/Polβfl/fl cortices using RNeasy Plus Mini Kit (QIAGEN) following the manufacturer's procedure. cDNA synthesis was performed using Transcriptor First Strand cDNA Synthesis Kit (Roche). Quantitative RT-PCR was performed with Universal ProbeLibrary (UPL, Roche). Specific probes and intron-spanning primer pairs were designed using Universal ProbeLibrary Assay Design Center (https://qpcr.probefinder.com/organism.jsp). The following probes and primer pairs were used in each gene. Gap43, UPL probe 63, 5′-cggagactgcagaaagcag-3′ and 5′-ggtttggcttcgtctacagc-3′; Apc, UPL probe 70, 5′-gtgcgccagcttttacagt-3′ and 5′-catgcctgctctgagatgac-3′; Tubb6, UPL probe 55, 5′-ctacgtgggcgactcagc-3′ and 5′-agccctgggtacgtacttctt-3′; and Rplp1, UPL probe 67, 5′-ttctggcctggcttgttt-3′ and 5′-atggagcagcaccaccag-3′. GAPD gene was used as an endogenous control (UPL Mouse GAPD Gene Assay, Roche).
Cell culture.
Pregnant mice were deeply anesthetized with pentobarbital (50 mg/kg, i.p.). E14.5 cortices were collected from the embryos in ice-cold Hanks' solution and then minced with fine scissors in PBS. The minced tissues were incubated with 0.125% trypsin and 0.02% EDTA in PBS for 5 min at 37°C, and then dissociated by pipetting. After centrifugation, the cells were resuspended in a mixture of DMEM and F12 (Invitrogen), 10% FBS (Hyclone), 10 ng/ml mouse FGF (Peprotech), and 20 ng/ml mouse EGF (Peprotech). A suspension containing 1.5–2.0 × 105 cells was plated with culture medium on a 12 mm microcoverglass (Matsunami) in a Nunc 4-well dish (Thermo Scientific); alternatively, 6.5–7.5 × 105 cells were plated on 35 mm cell culture dishes (Falcon) coated with 0.1 mg/ml poly-L-ornithine (Sigma). The cultures were maintained at 37°C in an environment of 5% CO2 and humidified 95% air.
To label DNA with EdU, the cells were incubated with culture medium containing 20 μm EdU (Invitrogen) for 15 min. To induce base damage, the cells were treated with culture medium containing 1 mm methylmethanesulfonate (MMS) (Sigma) for 1 h, then fixed for 10 min.
To induce DNA demethylation, the cells were incubated with culture medium containing 100 μg/ml l-ascorbic acid 2-phosphate (vitamin C, Sigma) for 24 h, and then fixed for 10 min.
Immunoblot analysis.
Genomic DNA was extracted using DNeasy Blood and Tissue kit (QIAGEN) or MagExtractor-Genome-kit (Toyobo). Isolated DNA was denatured in 0.1 m NaOH for 5 min at 95°C and then chilled on ice. For immunoblot analysis of 5-methylcytosine (5mC), serially diluted DNA samples (200 μl) were loaded on a positively charged nylon membrane (Millipore) using Bio-Dot SF Microfiltration Apparatus (Bio-Rad). For immunoblot analysis of 5-hydroxymethylcytosine (5hmC), aliquots (1 μl) of serially diluted DNA samples were spotted on a positively charged nylon membrane (Millipore). The blotted membrane was air-dried and UV cross-linked using a CL-1000 Ultraviolet Crosslinker. The membrane was washed with Tris-buffered saline containing 0.1% Tween 20 (TBS-T) for 30 min and then blocked in TBS-T containing 5% skim milk (Cell Signaling Technology) for 1 h at room temperature. The membrane was incubated with mouse monoclonal anti-5mC antibody (39649, Active Motif) at 1:2000 or rabbit polyclonal anti-5hmC antibody (39769, Active Motif) at 1:5000 in TBS-T containing 5% skim milk overnight at 4°C. The membrane was washed with TBS-T for 10 min three times and then incubated with antimouse IgG HRP conjugate (115-035-146, Jackson ImmunoResearch Laboratories) or -rabbit IgG HRP conjugate (711-035-152, Jackson ImmunoResearch Laboratories) at 1:20,000 in TBS-T containing 5% skim milk for 2 h at room temperature. The membrane was washed with TBS-T for 10 min three times and visualized by chemiluminescence with ECL Select Western Blotting Detection Reagent (GE Healthcare), and images were captured with an LAS-3000 UV mini (Fujifilm).
Image analysis.
Digital images were obtained by confocal microscopy (ECLIPSE FN with EZ-C1; Nikon) with 10 ×/0.3, 20 ×/0.75, 40 ×/0.95, and 100 ×/1.40 objective lenses (Nikon). The images were imported into Photoshop 6.0 (Adobe Systems) and ImageJ to adjust brightness and contrast.
For the quantitative analysis of Polβ protein levels, fluorescence intensities of Polβ in the nucleus were quantified using ImageJ. The position of the nucleus was judged by DAPI staining. The fluorescence intensities of Polβ in the CP neurons of Emx1-Cre/Polβfl/fl mice were measured as background intensity. The average intensities were calculated from 10 cells in a section. The ratio was calculated from three independent experiments from two embryonic mice, respectively.
Experimental design and statistical analysis.
Mice in both sexes were used for all experiments. In statistical analysis, the analyzed number of samples was described in each experiment. Significant differences were determined with Student's t test or Mann–Whitney U test. All statistical values are presented as mean value ± SD or mean value ± SEM. All data were analyzed using Excel 2013 (Microsoft).
Results
Polβ deficiency in neural progenitors induces neuronal apoptosis in the developing cortex
To dissect Polβ functions in neural progenitors and postmitotic neurons, Polβfl/fl mice (Gu et al., 1994) were crossed with Cre driver lines Emx1-Cre and Nex-Cre. Emx1-Cre-mediated recombination occurs in neural progenitors restricted to the dorsal telencephalon (Iwasato et al., 2000), whereas Nex-Cre-mediated recombination occurs in postmitotic excitatory neurons (Goebbels et al., 2006). In addition, to generate mice lacking Polβ in the whole-body (Polβdel/del), which correspond to conventional Polβ−/− knock-out mice, Polβfl/fl mice were crossed with mice from a testis-specific Cre driver line, Sycp1-Cre (Noguchi et al., 2009). To confirm CRE-mediated recombination of the floxed Polβ allele, in situ hybridization were performed on these mutant cortices (Fig. 1A). As expected, Polβ expression disappeared in both neural progenitors in the VZ and postmitotic neurons in the CP of Emx1-Cre/Polβfl/fl mice, whereas only postmitotic neurons lacked expression of Polβ in Nex-Cre/Polβfl/fl mice. Consistent with this result, immunohistochemistry further showed that Polβ signal in Nex-Cre/Polβfl/fl mice was almost absent in CP cells (the signal intensity in CP cells was 14 ± 3% of that in Polβfl/fl mice) (Fig. 1B). Thus, these mutant mice allowed us to study the spatiotemporal role of Polβ in the developing cortex, focusing on proliferative neural progenitors and postmitotic neurons.
Forebrain-specific CRE-mediated recombination occurs in Polβ conditional mutant mice. A, In situ hybridization of Pax6 and Polβ was performed in E14.5 Polβfl/fl, Emx1-Cre/Polβfl/fl, and Nex-Cre/Polβfl/fl forebrain. Scale bar, 500 μm. B, Immunohistochemistry was performed with anti-Polβ antibody in E14.5 wild-type and Nex-Cre/Polβfl/fl cortex. Nuclei were stained with DAPI. Scale bar, 100 μm.
Consistent with a previous result (Sugo et al., 2000), Polβdel/del mice died after birth (Table 1). Meanwhile, both Emx1-Cre/Polβfl/fl and Nex-Cre/Polβfl/fl mice were obtained at the expected Mendelian ratios at both E14.5 and the weaning stage (Tables 2, 3), indicating that the dorsal telencephalon-specific Polβ deficiency does not lead to lethality. According to the previous study (Sugo et al., 2000), we examined apoptosis in the developing cortex using immunohistochemical analysis with anticleaved caspase-3 antibody. Cleaved caspase-3-positive apoptotic cells were rarely found in E14.5 wild-type mice (Fig. 2A–C). In accordance with the result in Polβ−/− mice, cleaved caspase-3-positive apoptotic cells were remarkably observed in Tuj1-positive postmitotic neurons in the IZ and CP, but not in neural progenitors in the VZ, of Polβdel/del mice at E14.5 (Fig. 2D–F). Emx1-Cre/Polβfl/fl mice displayed a similar distribution of cleaved caspase-3-positive cells to Polβdel/del mice (Fig. 2G–I). In contrast, the apoptotic cells were absent in Nex-Cre/Polβfl/fl mice, which is comparable with wild-type mice (Fig. 2J–L). To evaluate whether the level of apoptosis correlates with genotype, the laminar distribution of cleaved caspase-3-positive cells was quantified in the developing cortex of each mouse line. The cortical depth was divided into 10 equal bins from the ventricle to the pial surface (Fig. 2M). Cleaved caspase-3-positive cells were abundant in the CP in both Polβdel/del and Emx1-Cre/Polβfl/fl mice (bins 7–10, 58% in Polβdel/del mice and 59% in Emx1-Cre/Polβfl/fl mice), but were less numerous in the VZ (bins 1–5, 19% in Polβdel/del mice and 13% in Emx1-Cre/Polβfl/fl mice) (Fig. 2M; Polβdel/del mice vs wild-type: **p < 0.01, t test, Emx1-Cre/Polβfl/fl mice vs wild-type: *p < 0.05, **p < 0.01, t test). In contrast, very few apoptotic cells were detected in every layer of either wild-type or Nex-Cre/Polβfl/fl cortex (Fig. 2M). These results indicate that, although apoptosis is predominantly observed in postmitotic neurons in the Polβ-deficient cortex, loss of Polβ activity in neural progenitors in the VZ leads to this phenotype.
Genotyping of Polβdel/+ intercrosses
Genotyping of Emx1-Cre/Polβfl/fl mice
Genotyping of Nex-Cre/Polβfl/fl mice
Polβ is essential for neural progenitors rather than postmitotic neurons. Immunohistochemistry was performed with anti-Tuj1 (A, C, D, F, G, I, J, L) and anticleaved caspase-3 (B, C, E, F, H, I, K, L) antibodies in E14.5 wild-type (A–C), Polβdel/del (D–F), Emx1-Cre/Polβfl/fl (G–I), and Nex-Cre/Polβfl/fl (J–L) cortex. Nuclei were stained with DAPI (C, F, I, L). Scale bar, 100 μm. M, Distribution histogram of cleaved caspase-3-positive cells in the cortical column of E14.5 wild-type, Polβdel/del, Emx1-Cre/Polβfl/fl, and Nex-Cre/Polβfl/fl mice. The cortical column was divided into 10 equal bins from the ventricle to the pial surface. VZ, bins 1–4; IZ, bins 5–7; CP, bins 8–10 in wild-type and Nex-Cre/Polβfl/fl cortex. VZ, bins 1–5; IZ, bin 6; CP, bins 7–10 in Polβdel/del and Emx1-Cre/Polβfl/fl cortex. Values in parentheses are the number of cells analyzed. Data are mean ± SEM from at least three different brains. Significant difference from wild-type mice: *p < 0.05 (Student's t test); **p < 0.01 (Student's t test).
Polβ deficiency increases DSB formation in neural progenitors, thereby inducing p53-mediated apoptosis
To investigate the mechanism by which Polβ deficiency in neural progenitors triggers neuronal apoptosis, we considered the possibility that cytotoxic DSBs accumulate in postmitotic neurons (Lee et al., 2001). DSB formation was examined in the conditional mutant cortex by immunohistochemistry with anti-γH2AX (Ser 139-phoshorylated histone H2AX) antibody. γH2AX forms foci at DSB sites in the nucleus and is involved in DSB repair (Rogakou et al., 1998, 1999). Even in control Polβfl/fl mice, a few γH2AX foci formed in a small fraction of Ki67-positive proliferative neural progenitors in the VZ and subventricular zone, but not in Tuj1-positive postmitotic neurons in the IZ or the CP (Fig. 3A–D,F–I). Notably, an increased number of γH2AX foci formed in both the VZ and CP in Emx1-Cre/Polβfl/fl mice (Fig. 3J–M). Conversely, Nex-Cre/Polβfl/fl mice displayed γH2AX foci formation at a level similar to control Polβfl/fl mice (Fig. 3N–Q). Immunohistochemical analysis with an antibody to 53BP1, another marker protein of DSB formation (Schultz et al., 2000), confirmed these results (Fig. 3S–Z,A′–D′). To quantify the distribution of damaged cells in the developing cortex, γH2AX foci-possessing cells were counted in each bin as described above (Fig. 3E). γH2AX foci-positive cells in Polβfl/fl mice were more numerous in the VZ and subventricular zone (bins 1–4, 24%) than in the CP (bins 8–10, 4%) (Fig. 3E). In Emx1-Cre/Polβfl/fl mice, the γH2AX foci-possessing cells were abundant in all layers, particularly in the CP (VZ: bins 1–5, 53%, CP: bins 7–10, 69%, Emx1-Cre/Polβfl/fl mice vs wild-type: p < 0.01, t test). In contrast, the distribution of DNA-damaged cells in Nex-Cre/Polβfl/fl mice (VZ: bins 1–4, 20%, CP: bins 8–10, 6%) was comparable with that in Polβfl/fl mice (Fig. 3E). These results indicate that Polβ deficiency in neural progenitors leads to excessive DSB formation.
Polβ deficiency increases DSBs in neural progenitors, which remain in postmitotic neurons. Immunohistochemistry was performed with anti-γH2AX (A, B, F–Q), anti-Tuj1 (C), and anti-Ki67 (D) antibodies in E14.5 wild-type (A–D), E14.5 Polβfl/fl (F–I), Emx1-Cre/Polβfl/fl (J–M), and Nex-Cre/Polβfl/fl (N–Q) cortex. Nuclei were stained with DAPI (B–D, G, I, K, M, O, Q). Arrows indicate γH2AX foci. Scale bars: D, 50 μm; Q, 10 μm. E, Distribution histogram of γH2AX foci-positive cells in the cortical column in E14.5 Polβfl/fl, Emx1-Cre/Polβfl/fl, and Nex-Cre/Polβfl/fl mice. The cortical column was divided into 10 equal bins from the ventricle to the pial surface. Values in parentheses are the number of cells analyzed. Data are mean ± SEM from at least three different brains. Significant difference from Polβfl/fl mice: **p < 0.01 (Student's t test). R, Histogram showing the number of γH2AX foci in individual VZ and CP cells of E14.5 Polβfl/fl, Emx1-Cre/Polβfl/fl, and Nex-Cre/Polβfl/fl cortex. Data are the mean ± SEM from 24 cells in each of three different brains. Significant difference from Polβfl/fl mice: *p < 0.05 (Student's t test); **p < 0.01 (Student's t test). Immunohistochemistry was performed with anti-53BP1 antibody (S–Z, A′–D′) in E14.5 Polβfl/fl (S–V), Emx1-Cre/Polβfl/fl (W–Z), and Nex-Cre/Polβfl/fl (A′–D′) cortex. Nuclei were stained with DAPI (T, V, X, Z, B′, D′). Scale bar, 10 μm.
To characterize the DNA damage level in individual VZ and CP cells, we counted γH2AX foci in the nuclei. The number of γH2AX foci in both VZ and CP cells was significantly larger in Emx1-Cre/Polβfl/fl mice compared with Polβfl/fl mice (Fig. 3R; Polβfl/fl vs Emx1-Cre/Polβfl/fl, VZ: 1.13 ± 0.25, 2.21 ± 0.31, p = 0.021, CP: 0.13 ± 0.069, 2.58 ± 0.29, p = 0.00045, t test). In contrast, the number of γH2AX foci in Nex-Cre/Polβfl/fl mice was similar to that in Polβfl/fl mice (Fig. 3R). Moreover, the number of γH2AX foci was comparable between VZ and CP cells in Emx1-Cre/Polβfl/fl mice, although γH2AX foci were rare in the CP cells in Polβfl/fl and Nex-Cre/Polβfl/fl mice. These results suggest that the majority of DSBs by Polβ deficiency in neural progenitors remain unrepaired in postmitotic neurons.
We further investigated whether DSB formation is associated with DNA replication in the VZ by observing DSB and replication sites simultaneously in the nucleus. To visualize DNA replication, EdU was injected into E14.5 pregnant mice, and its incorporation was examined after 30 min. This short-term labeling should allow us to identify active replication in the nucleus (O'Keefe et al., 1992). The majority of the EdU-positive cells in the VZ contained γH2AX foci in both Polβfl/fl (70 ± 4.4%) and Emx1-Cre/Polβfl/fl (71 ± 3.5%) cortex (Fig. 4A–I), although the number of γH2AX foci in individual cells was much larger in Emx1-Cre/Polβfl/fl mice (see above). In contrast, EdU-negative VZ cells more frequently showed γH2AX foci in Emx1-Cre/Polβfl/fl mice (65 ± 3.2%) than in Polβfl/fl mice (19 ± 1.5%) (Fig. 4I; EdU-negative cells in Polβfl/fl mice vs EdU-negative cells in Emx1-Cre/Polβfl/fl mice: p = 0.0002, t test, EdU-positive vs EdU-negative cells in Polβfl/fl mice: p = 0.00015, t test). Similar foci formation of 53BP1 was also observed in the VZ (Fig. 4J–Q). Thus, although DSBs likely formed at DNA replication sites during S phase in both wild-type and Polβ-deficient neural progenitors, it seems that most of these DSBs remain unrepaired until the next cell cycle in the Polβ-deficient neural progenitors or postmitotic neurons.
DSBs arise during S phase and lead to p53-mediated neuronal apoptosis in Polβ-deficient cortex. EdU staining (B, D, F, H) and immunohistochemistry for γH2AX (A, D, E, H) were performed in E14.5 Polβfl/fl (A–D) and Emx1-Cre/Polβfl/fl (E–H) cortex. Nuclei were stained with DAPI (C, D, G, H). Arrows and arrowheads indicate EdU-positive and -negative cells, respectively. Scale bar, 10 μm. I, Quantitative analysis of γH2AX foci-positive cells in either EdU-positive or -negative cells in the VZ of Polβfl/fl and Emx1-Cre/Polβfl/fl cortex. Values in parentheses are the number of neural progenitors analyzed. Data are mean ± SEM from three different brains. Significant difference: **p < 0.01 (Student's t test). EdU staining (K, M, O, Q) and immunohistochemistry for 53BP1 (J, M, N, Q) were performed in E14.5 Polβfl/fl (J–M) and Emx1-Cre/Polβfl/fl cortex (N–Q). Nuclei were stained with DAPI (L, M, P, Q). Arrows and arrowheads indicate EdU-positive and -negative cells, respectively. Scale bar, 10 μm. Immunohistochemistry was performed with anticleaved caspase-3 (R, S) and anti-γH2AX antibodies (T, U) in E14.5 Polβdel/delp53−/− cortex. Nuclei were stained with DAPI (S, U). Scale bars: S, 100 μm; U, 10 μm.
Because it is known that a genome guardian p53 mediates apoptosis induced by DSBs in cortical neurons (Frank et al., 2000; Gao et al., 2000), we examined γH2AX foci formation in Polβdel/delp53−/− mice. As the loss of p53 completely suppresses neuronal apoptosis in Polβ−/− mice (Sugo et al., 2004), apoptosis could be inhibited in Polβdel/delp53−/− mice if DNA damage is due to a lack of Polβ induced apoptosis. We found that, indeed, apoptosis was completely attenuated in the CP of Polβdel/delp53−/− cortex, despite extensive DSB formation (Fig. 4R–U). Together, these results indicate that Polβ deficiency leads to increased DSB formation in neural progenitors and the subsequent activation of the p53-mediated apoptotic pathway in postmitotic neurons.
Polβ deficiency differently affects cell viability in each cortical layer
Although cortical neurons show extensive apoptosis in Emx1-Cre/Polβfl/fl mice, a substantial number of neurons are still viable in the CP (Fig. 2). To investigate the impact of Polβ deficiency on cortical cytoarchitecture, immunohistochemistry was performed with antibodies for layer-specific molecules in Polβfl/fl and Emx1-Cre/Polβfl/fl mice at E18.5. Nurr1 (Fig. 5A–D), Tbr1 (Fig. 5E,G,H,J), Ctip2 (Fig. 5F,G,I,J), and Cux1 (Fig. 5K–N) antibodies were used for subplate, layer VI, layer V, and layer II-IV markers, respectively (Hevner et al., 2001; Arimatsu et al., 2003; Nieto et al., 2004; Arlotta et al., 2005). Although the laminar structure formed normally in Emx1-Cre/Polβfl/fl mice, the thickness and the number of cells in each layer was significantly lower in Emx1-Cre/Polβfl/fl (Fig. 5C,D,H–J,M,N) than in Polβfl/fl cortex (Fig. 5A,B,E–G,K,L). The extent of reduction was more prominent in the Nurr1-, Tbr1-, and Ctip2-positive deep layers than in the Cux1-positive upper layers (Fig. 5O; Nurr1: 0.4 ± 0.011, p = 0.0094, Tbr1: 0.56 ± 0.04, p = 0.001, Ctip2: 0.54 ± 0.038, p = 0.008, Cux1: 0.81 ± 0.021, p = 0.023, t test; Fig. 5P; Nurr1: 0.41 ± 0.052, p = 0.00014, Tbr1: 0.54 ± 0.043, p = 0.0008, Ctip2: 0.56 ± 0.032, p = 0.0061, Cux1: 0.77 ± 0.023, p = 0.031, t test). Furthermore, to examine the possibility that Polβ may affect cell cycle and neuronal laminar fate, we performed birth date labeling with EdU. In accordance with previous reports (Angevine and Sidman, 1961; Takahashi et al., 1995), early-born neurons labeled with EdU at E12.5 were mainly located in Ctip2-positive deep layers (Fig. 5Q–V), and late-born neurons labeled at E14.5 were present in the upper layers at E18.5 in both Polβfl/fl and Emx1-Cre/Polβfl/fl mice (Fig. 5W–Z,A′,B′). Together, these results suggested that deep-layer neurons undergo more extensive apoptosis compared with upper-layer neurons, which resulted in deep layers that were thinner than upper layers in Polβ-deficient cortex.
Polβ deficiency preferentially reduces the survival of deep-layer neurons in the developing cortex. Immunohistochemistry was performed with anti-Nurr1 (A–D), anti-Tbr1 (E, G, H, J), anti-Ctip2 (F, G, I, J, L, N), and anti-Cux1 (K–N) antibodies in E18.5 Polβfl/fl (A, B, E–G, K, L) and Emx1-Cre/Polβfl/fl (C, D, H–J, M, N) cortex. Nuclei were stained with DAPI (B, D, G, J, L, N). Scale bars, 100 μm. O, The thickness of each marker-positive layer in Emx1-Cre/Polβfl/fl mice relative to Polβfl/fl mice. Data are mean ± SEM from at least three different brains. Significant difference from Polβfl/fl mice: *p < 0.05 (Student's t test); **p < 0.01 (Student's t test). P, The number of layer marker-positive cells in each cortical layer of Emx1-Cre/Polβfl/fl mice relative to Polβfl/fl mice. The columns analyzed were 270 μm wide. Data are mean ± SEM from at least three different brains. Significant difference from Polβfl/fl mice: *p < 0.05 (Student's t test); **p < 0.01 (Student's t test). EdU staining (Q–Z, A′, B′) and immunohistochemistry for Ctip2 (R, S, U, V, X, Y, A′, B′) were performed in E18.5 Polβfl/fl (Q–S, W–Y) and Emx1-Cre/Polβfl/fl (T–V, Z, A′, B′) cortex. Nuclei were stained with DAPI (S, V, Y, B′). Scale bar, 100 μm.
To further examine whether the extensive apoptosis in the deep-layer neurons is related to the extent of DSBs, the number of γH2AX foci-positive cells was counted in Emx1-Cre/Polβfl/fl mice at E16.5 (Fig. 6A–G), when only a minor amount of cleaved caspase-3-positive apoptosis was observed in the CP (Fig. 6H,I). We found that γH2AX foci-positive cells were densely distributed in the Ctip2-positive deep layers (Fig. 6A–F). The ratio of γH2AX foci-positive cells was significantly higher in Ctip2-positive deep-layer neurons (58 ± 4.9%) than in Ctip2-negative upper-layer neurons (41 ± 4.5%) (Fig. 6G; p = 0.0284, t test), suggesting the possibility that extensive apoptosis in the CP accompanies increased DSBs.
DSBs are relatively accumulated in deep-layer neurons compared with upper-layer neurons in Polβ-deficient cortex. Immunohistochemistry was performed with anti-γH2AX (A, C, D, F) and Ctip2 (B, C, E, F) antibodies in E16.5 Emx1-Cre/Polβfl/fl cortex. Nuclei were stained with DAPI (C, F). Scale bar, 100 μm. D–F, Magnified images of the boxed area in C. Scale bar, 10 μm. G, Quantitative analysis of γH2AX foci-positive cells in the Ctip2-positive deep and -negative upper layers of E16.5 Emx1-Cre/Polβfl/fl cortex. Values in parentheses are the number of neurons analyzed. Error bars indicate the mean ± SEM from three different brains. Significant difference between the layers: *p < 0.05 (Student's t test). Immunohistochemistry was performed with anticleaved caspase-3 antibody (H, I) in E16.5 Emx1-Cre/Polβfl/fl cortex. Nuclei were stained with DAPI (I). Scale bar, 100 μm.
Polβ deficiency affects axonal growth in a subset of deep-layer neurons
In Emx1-Cre/Polβfl/fl mice, DSBs were frequently observed, even in surviving neurons in the CP (Fig. 3). Such DSB may also affect cortical cell differentiation. We asked whether the DSBs affect axon formation in damaged deep-layer neurons (approximately layer V) in Emx1-Cre/Polβfl/fl mice. To observe axons originating from the deep-layer cells, EGFP plasmid was transfected into VZ cells by in utero electroporation at E13.5, when these cells are mostly destined to become deep-layer neurons. Then, axonal projection was examined at E16.5, when initial axonal projections have been formed. In Polβfl/fl mice, the majority of EGFP-labeled neurons were localized in deep layer and extended their axons toward the corpus callosum or internal capsule (Fig. 7A–C) (Custo Greig et al., 2013). Similarly, callosal and subcortical projections were observed in Emx1-Cre/Polβfl/fl mice, but the length of corticofugal axons was significantly shorter than in Polβfl/fl mice (Fig. 7D–G; Polβfl/fl vs Emx1-Cre/Polβfl/fl: 704 ± 41 μm, 520 ± 27 μm, p = 0.008, U test). In contrast, the commissural axonal growth was indistinguishable between Polβfl/fl and Emx1-Cre/Polβfl/fl mice (Fig. 7D–G). Consistent with this analysis, immunohistochemistry with anti-TAG1 antibody also demonstrated corticofugal axonal growth defects in Emx1-Cre/Polβfl/fl mice (Fig. 7H–O) (Denaxa et al., 2001). TAG1-positive axons were clearly observed in the internal capsule of E16.5 Polβfl/fl mice, but few were observable in Emx1-Cre/Polβfl/fl mice (Fig. 7L,N). In accordance with this result, qRT-PCR analysis of the embryonic cortex demonstrated that expression of axon growth-related genes, such as Gap43 and Apc, was downregulated in Emx1-Cre/Polβfl/fl compared with Polβfl/fl mice (Fig. 7P; Gap43: 0.47 ± 0.058, p = 0.0015, Apc: 0.35 ± 0.075, p = 0.00074, t test) (Kruger et al., 1998; Shi et al., 2004). These results demonstrate that axonal extension of subcortically projecting deep-layer neurons is severely affected in Polβ deficient cortex, suggesting that the genome instability leads to neural defects even in the subset of cortical cells that escape from cell death.
Polβ deficiency affects corticofugal axonal growth. Cortical layer IV neurons were labeled with EGFP by in utero electroporation at E13.5 and analyzed at E16.5. Immunohistochemical analyses were performed with anti-GFP antibody (A, C, D, F) in Polβfl/fl (A–C) and Emx1-Cre/Polβfl/fl mice (D–F). Nuclei were stained with DAPI (B, C, E, F). Scale bar, 500 μm. Frames indicate electroporated area. Asterisks indicate the center of electroporated area. Arrow and arrowhead indicate tips of corticofugal axons and commissural axons, respectively. G, Quantitative analysis of the length of axons was calculated from the center of each electroporated area to tips of axons in Polβfl/fl and Emx1-Cre/Polβfl/fl mice. **p < 0.01 (Mann–Whitney U test). Values indicate the mean ± SEM from 11 sections of 4 Polβfl/fl brains or 16 sections of 5 Emx1-Cre/Polβfl/fl brains. Immunohistochemistry was performed with anti-TAG-1 antibody (H–O) in E16.5 Polβfl/fl (H–K) and Emx1-Cre/Polβfl/fl (L–O) forebrain. Nuclei were stained with DAPI (I, K, M, O). Scale bar, 500 μm. J, K, N, O, Magnified images of the boxed area in I and M, respectively. Scale bar, 300 μm. IC, Internal capsule. Arrows indicate TAG-1-positive axons. P, Relative gene expression of Gap43, Apc, Tubb6, and Rplp1 was examined by qRT-PCR in E14.5 Polβfl/fl and Emx1-Cre/Polβfl/fl cortices. Gap43 and Apc are known to relate to axonal growth. Tubb6 and Rplp1 are ubiquitously expressed genes. Significant difference from Polβfl/fl cortices: **p < 0.01 (Student's t test).
Polβ deficiency increases DSBs arising from base damage and DNA demethylation intermediates
Finally, we investigated the mechanistic basis for DSB induction in neural progenitors. Polβ deficiency is generally thought to increase nicks and gaps as single-strand breaks (SSBs) (Wilson et al., 2000). However, we found that DSBs arose during replication in neural progenitors (see above). SSBs may be converted to DSBs due to replication fork collapse in S phase (Kuzminov, 2001). To test this possibility, cultured cortical progenitor cells were treated with the alkylating agent MMS for 1 h. MMS treatment produces base damage that yields substrates for BER (Beranek, 1990). After the treatment, we observed DSB formation using immunocytochemistry with anti-γH2AX antibody (Fig. 8A–P). Similar to in vivo analysis, the control culture revealed that the number of γH2AX foci was greater in EdU-positive Emx1-Cre/Polβfl/fl cells (Fig. 8I–L,Q; mean number of γH2AX foci: 2.3 ± 0.39) than in Polβfl/fl cells (Fig. 8A–D,Q; 0.8 ± 0.24). Importantly, the MMS treatment significantly increased the number of γH2AX foci in Emx1-Cre/Polβfl/fl cells (Fig. 8M–Q; 6.0 ± 0.79) relative to Polβfl/fl cells (Fig. 8E–H,Q; 3.4 ± 0.89, control Polβfl/fl cells vs control Emx1-Cre/Polβfl/fl cells: p = 0.0024, t test, MMS treated Polβfl/fl cells vs MMS treated Emx1-Cre/Polβfl/fl cells: p = 0.023, t test, control vs MMS treated Polβfl/fl cells: p = 0.0025, t test, control vs MMS treated Emx1-Cre/Polβfl/fl cells: p = 0.00029, t test). This result indicates that base damage is frequently converted to DSBs during replication in Polβ-deficient neural progenitors.
Polβ deficiency in neural progenitors increases DSBs following MMS treatment. EdU staining (B, D, F, H, J, L, N, P) and immunocytochemistry for γH2AX (A, D, E, H, I, L, M, P) were performed in cultured neural progenitors from E14.5 Polβfl/fl (A–H) and Emx1-Cre/Polβfl/fl (I–P) cortex. Nuclei were stained with DAPI (C, D, G, H, K, L, O, P). The cells were untreated (A–D, I–L) or treated with 1 mm MMS (E–H, M–P). Scale bar, 5 μm. Q, Histogram shows the number of γH2AX foci in individual EdU-positive cells. Values indicate the mean ± SEM from 24 Polβfl/fl or Emx1-Cre/Polβfl/fl cells. Significant difference: *p < 0.05 (Student's t test); **p < 0.01 (Student's t test).
In mammalian active DNA demethylation, TET1 converts 5mC to specific base modifications, such as 5hmC, leading to its replacement by BER (Bhutani et al., 2011). Because it is suggested that Polβ is involved in this process (Weber et al., 2016), we investigated whether the epigenetic base modification is removed by BER in Polβ-deficient cultured cortical progenitor cells. For this, cultured cortical cells were treated with vitamin C for 24 h, following the previous observation that vitamin C increases the level of 5hmC by activating TET1 (Blaschke et al., 2013). As expected, immunoblot analysis with anti-5hmC antibody showed increased levels of 5hmC in genomic DNA after vitamin C treatment (Fig. 9B), whereas the levels of 5mC in genomic DNA were approximately similar between E14.5 Polβfl/fl and Emx1-Cre/Polβfl/fl cortex (Fig. 9A). In Polβfl/fl cells, the number of γH2AX foci in Nestin-positive cells was unchanged with or without the treatment (Fig. 9C–J,S). On the other hand, vitamin C treatment significantly increased the number of γH2AX foci relative to the control in Nestin-positive Emx1-Cre/Polβfl/fl cells (Fig. 9K–S; control vs vitamin C-treated Emx1-Cre/Polβfl/fl cells: 0.58 ± 0.2, 1.25 ± 0.24, p = 0.036, t test; vitamin C-treated Polβfl/fl vs Emx1-Cre/Polβfl/fl cells: 0.33 ± 0.12, 1.25 ± 0.24, p = 0.001, t test). This result suggests that Polβ-dependent BER is necessary for the DNA demethylation process in neural progenitors, thereby contributing to the competence of neural progenitors.
Accumulation of the DNA demethylation intermediate 5hmC increases DSBs in Polβ-deficient neural progenitors. A, Immunoblot analysis shows 5mC global levels in genomic DNA from E14.5 Polβfl/fl and Emx1-Cre/Polβfl/fl mouse cortex. B, Immunoblot analysis shows global 5hmC levels in genomic DNA. Dissociated cell cultures from E14.5 cortices were incubated with or without vitamin C for 24 h. Immunocytochemistry was performed with anti-γH2AX (C, F, G, J, K, N, O, R) and anti-Nestin (D, F, H, J, L, N, P, R) antibodies in dissociated cells from E14.5 Polβfl/fl (C–J) and Emx1-Cre/Polβfl/fl mouse cortex (K–R). Nuclei were stained with DAPI (E, F, I, J, M, N, Q, R). The cells were untreated (C–F, K–N) or treated with 100 μg/ml vitamin C (G–J, O–R). Scale bar, 5 μm. S, Histogram showing the number of γH2AX foci in individual Nestin-positive cells. Values indicate the mean ± SEM from 24 Polβfl/fl or Emx1-Cre/Polβfl/fl cells. Asterisks indicate a significant difference (*p < 0.05, **p < 0.01, Student's t test).
Discussion
In the present study, we demonstrated that the loss of Polβ led to DSB formation in neural progenitors, and subsequently to p53-mediated apoptosis in postmitotic neurons, resulting in abnormal cortical layer formation. The DSBs also caused axon growth defects in a subset of surviving neurons. In addition, treatment of either MMS or vitamin C markedly enhanced DSB formation in cultured Polβ-deficient neural progenitors. These results suggest that pathological base damage and epigenetic base modification in neural progenitors cause neuronal developmental defects. Thus, genome stability by Polβ-dependent BER is required for the competence of neural progenitors, thereby contributing to cortical development.
Polβ-dependent BER is required for neural progenitors rather than postmitotic neurons in the developing cortex
Polβ is thought to be required for genome stability in postmitotic neurons because the alternative proliferating cell nuclear antigen-dependent long patch BER pathway does not function in neurons (Wei and Englander, 2008; Akbari et al., 2009; Sykora et al., 2013). Indeed, the previous and present results demonstrate that massive neuronal apoptosis occurs in Polβ-deficient mice (Sugo et al., 2000). However, the present findings provide a new insight that genome stability by Polβ-dependent BER is more important in neural progenitors than in postmitotic neurons in the developing cortex.
The present study revealed that γH2AX foci-positive DSBs are localized at DNA replication sites and accumulate in Polβ-deficient neural progenitors, even though Polβ is expressed ubiquitously and independently in the cell cycle (Figs. 3, 4) (Zmudzka et al., 1988). Although Polβ is not involved in the DSB repair pathway, SSBs (BER intermediates) may give rise to DSBs during DNA replication in Polβ-deficient neural progenitors. This would not be surprising, as SSBs are known to be converted to DSBs by replication fork collapse (Kuzminov, 2001). Consistent with our findings in cultured neural progenitors (Fig. 8), Polβ-deficient MEF cells show increased γH2AX foci at S phase and chromatid breaks at M phase in response to MMS treatment (Sobol et al., 2003; Pascucci et al., 2005; Senejani et al., 2012; Ensminger et al., 2014). Moreover, similar to Polβ-deficient mice, neural progenitors deficient in other BER-related components (such as Xrcc1 and Pnkp), which are thought to accumulate SSBs, also show increased DSBs in the developing cortex (Lee et al., 2009; Shimada et al., 2015).
We also found a small number of DSBs in wild-type neural progenitors, but not in postmitotic neurons (Fig. 3). The DSBs were, furthermore, predominantly observed in S phase in wild-type, similar to Polβ-deficient neural progenitors (Fig. 4). Therefore, SSBs, perhaps due to insufficient BER, may promote genome instability even in wild-type neural progenitors.
Neuronal apoptosis in Polβ-deficient developing cortex
Previous studies have shown that apoptosis occurs in the developing cortex, although the extent observed varied between different detection methods and developmental stages (Ferrer et al., 1992; Blaschke et al., 1996; Yang et al., 2004; Depaepe et al., 2005). These findings raised the hypothesis that an overproduction of neurons and the subsequent elimination processes are required for normal cortical development (Kuan et al., 2000). Contrary to these observations, it is unlikely that Polβ-dependent BER is involved in developmental apoptosis. Indeed, the present study showed that this pathway is crucial to maintain cellular homeostasis by repairing pathological and physiological DNA damage in neural progenitors (Figs. 8, 9).
The formation of DSBs as a result of Polβ deficiency occurs in neural progenitors, and the consequent apoptosis was mainly observed in the CP (Figs. 2, 3). These results indicate that, although neural progenitors accumulate many DSBs due to Polβ deficiency, they do not undergo apoptosis. It is known that DSBs activate ataxia telangiectasia-mutated signaling, which induces neuronal apoptosis via phosphorylation of p53 in the CP (Lee et al., 2001). Indeed, a previous report and our current results showed that p53 deficiency completely rescued apoptotic phenotype in Polβ-deficient mice (Fig. 4) (Sugo et al., 2004). The difference in apoptosis may be related to the accumulated level of phosphorylated p53 in apoptotic cells and nonapoptotic cells. Supporting this notion, the reduction of p53 phosphorylation level attenuates neuronal apoptosis in Polβ−/−p53+/− mice (Sugo et al., 2004; Sugo et al., 2007). Thus, these data suggest that the accumulation level of phosphorylated p53 is insufficient to induce damage-related apoptosis in neural progenitors.
Similarly to Polβ-deficient mice, mice deficient in nonhomologous end-joining (NHEJ) also show aberrant apoptosis in postmitotic neurons and subsequent microcephaly that are identical to the corresponding human mutants (Barnes et al., 1998; Gao et al., 1998; O'Driscoll et al., 2001; Gatz et al., 2011). The phenotypic similarity between Polβ and NHEJ-deficient mice may be attributed to DSB-induced apoptosis in the IZ and CP, rather than less proliferation and neurogenesis in the VZ.
Polβ deficiency affects cortical layer formation and axonal growth
The present results demonstrate that DSBs lead to apoptosis in postmitotic cells in the CP and result in thinner cortical layers. On the other hand, the laminar profile of Polβ-deficient cortex seems to be normal (Fig. 5), suggesting that abundant DSBs may not influence laminar fate and cell cycle progression in Polβ-deficient neural progenitors (Fig. 3). However, DSB formation was different between layers: DSBs were found more abundantly in deep-layer neurons than upper-layer neurons in Polβ-deficient mice (Fig. 6). One possibility is that early neural progenitors may not have enough time to repair DSBs by NHEJ in G1 phase, because the cell cycle length of G1 phase is shorter in early neural progenitors than that in late progenitors (Caviness et al., 1995).
Axonal growth was also affected in Polβ-deficient cortex (Fig. 7). It is likely that DSB-containing neurons can survive but may have some deficiencies in axon growth property. Indeed, the expression of the axon growth-related genes Gap43 and Apc was downregulated (Fig. 7), suggesting a transcriptional change induced by DSBs affects axonal growth in Polβ-deficient cortex. Furthermore, subcortical, but not callosal, projection neurons in deep layers displayed severe axonal growth defects in Polβ-deficient cortex (Fig. 7), indicating that DSBs caused by Polβ deficiency affect axonal growth in a subset of neurons. In accordance with this view, a previous study demonstrated that only specific tracts, such as thalamocortical projections and anterior commissure, show abnormalities in Polβ−/−p53−/− mice (Sugo et al., 2004). Therefore, a cell-type-specific aspect may produce different extents of DSB formation in Polβ-deficient cortex, resulting in cell-type-specific phenotypes in developing neurons.
Polβ-dependent BER is necessary for DNA demethylation process in neural progenitors
DNA methylation level in the neuronal gene promoter and/or enhancer regions influences neuronal development (Fan et al., 2001; Tyssowski et al., 2014). Analyses of DNA methylation patterns reveal that active DNA demethylation via BER changes its patterns during neuronal development (Cortázar et al., 2011; Cortellino et al., 2011; Wheldon et al., 2014). In this study, we showed that DNA demethylation intermediates induced DSB formation in Polβ-deficient neural progenitors (Fig. 9). Our finding provides the genetic evidence that Polβ is an essential enzyme in these DNA demethylation processes. Although the endogenous source of DSBs in the developing cortex remains a controversial issue (Karanjawala et al., 2002), the physiological DNA demethylation process in neural progenitors may be susceptible to genome instability during cortical development. Moreover, if the DSBs are fixed by error-prone NHEJ, which is the only DSB repair pathway in neurons, insertions/deletions are frequently introduced in neuronal gene promoter and/or enhancer regions (Lu et al., 2004; Lieber, 2010). These DSBs and/or DSB-induced de novo mutations may perturb transcription of neuronal genes (Sharma et al., 2016). Together, our findings suggest that neuronal differentiation is dependent on DNA demethylation via Polβ-dependent BER in neural progenitors.
The impact of Polβ deficiency on cortical development and function
Together, these data suggest the following paradigm. DNA base damage is formed in neural progenitors by not only pathological damage but also epigenetic modification of the genome. In normal development, the DNA base damage can be repaired by the Polβ-dependent BER pathway. In particular, the BER completes DNA demethylation, promoting gene expression. But, in the absence of Polβ, DNA gaps may be retained and converted into cytotoxic DSBs by replication, so that DSBs accumulate in postmitotic neurons. Based on the extent of DSBs, various phenotypes will appear in cortical development. Some neurons with a large extent of DSBs undergo apoptosis, whereas other neurons with the lesser extent can survive but exhibit abnormal differentiation. Thus, our findings provide a new insight that regulation of genome stability by Polβ-dependent BER in neural progenitors is crucial for neuronal differentiation in cortical development.
Several developmental defects observed in Polβ-deficient cortex are suspected to lead to brain dysfunction. Individuals with NHEJ gene mutations demonstrate intellectual disability related to microcephaly (Rosin et al., 2015). It is likely that accumulation of DSBs and/or DSB-induced de novo mutations in Polβ-deficient cortex leads to cortical developmental disorders. Polβ polymorphisms have not been identified in the CNS, although several mutations in BER-related genes are identified to be associated with human brain dysfunction (Date et al., 2001; Takashima et al., 2002; Shen et al., 2010; Sobol, 2012; Hoch et al., 2017).
Footnotes
This work was supported by MEXT KAKENHI on Innovative Areas Mesoscopic Neurocircuitry 23115102 and Dynamic Regulation of Brain Function by Scrap and Build System 16H06460 to N.Y., Cross Talk Between Moving Cells and Microenvironment as a Basis of Emerging Order in Multicellular System 23111516, Grant 20200009 to N.S., Japan Society for the Promotion of Science KAKENHI Grants 23700447, 15K14350, and 17K07109 to N.S. and Grants 20300110 and 23300118 to N.Y., and AMED-Core Research for Evolutional Science and Technology to T.Y. We thank Drs. T. Iwasato and S. Itohara for Emx1-Cre mice; K.A. Nave for the Nex-Cre mice; K. Rajewsky for Polβ flox mice; Drs. S. Toyoda, K. Takiguchi-Hayashi, and N. Adachi for helpful discussion; and Dr. I. Smith and H. Gabriel for critical reading of the manuscript.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Noriyuki Sugo, Graduate School of Frontier Biosciences, Osaka University Yamadaoka 1-3, Suita, Osaka 565-0871, Japan. sugo{at}fbs.osaka-u.ac.jp
