Cdk12 Regulates Neurogenesis and Late-Arising Neuronal Migration in the Developing Cerebral Cortex

Abstract

DNA damage response (DDR) pathways are critical for ensuring that replication stress and various types of DNA lesion do not perturb production of neural cells during development. Cdk12 maintains genomic stability by regulating expression of DDR genes. Mutant mice in which Cdk12 is conditionally deleted in neural progenitor cells (NPCs) die after birth and exhibit microcephaly with a thinner cortical plate and an aberrant corpus callosum. We show that NPCs of mutant mice accumulate at G2 and M phase, and have lower expression of DDR genes, more DNA double-strand breaks and increased apoptosis. In addition to there being fewer neurons, there is misalignment of layers IV–II neurons and the presence of abnormal axonal tracts of these neurons, suggesting that Cdk12 is also required for the migration of late-arising cortical neurons. Using in utero electroporation, we demonstrate that the migrating mutant cells remain within the intermediate zone and fail to adopt a bipolar morphology. Overexpression of Cdk5 brings about a partially restoration of the neurons reaching layers IV–II in the mutant mice. Thus, Cdk12 is crucial to the repair of DNA damage during the proliferation of NPCs and is also central to the proper migration of late-arising neurons.

Cdk5, Cdk12, DNA damage response, late-arising neuron, neuronal migration

Introduction

The development of the mammalian cerebral cortex occurs during embryonic neurogenesis, which is a period when neurons are generated from neural progenitor cells in the ventricular (VZ) and subventricular zones (SVZ) (Bystron et al. 2008; Geschwind and Rakic 2013). Defects in neurogenesis lead to neurodevelopmental disorders such as microcephaly (a reduced brain size) (Haydar et al. 1999; Kuan et al. 2000; Rakic 2005). Neurogenesis requires intensive DNA replication by the neural progenitor cells to generate the large number of neurons that are required and as a result the DNA damage that occurs during DNA replication is constantly surveyed and repaired throughout the process. A failure to do so reduces neuronal viability and perturbs neuronal functioning (Barzilai 2007; O'Driscoll and Jeggo 2008; McKinnon 2009, 2013; Gilmore and Walsh 2013; Rulten and Caldecott 2013; Pan et al. 2014).

Double-strand breaks (DSBs) as a result of DNA replication are one of the most deleterious forms of DNA damage (McKinnon 2013; Reynolds and Stewart 2013). Unrepaired DSBs may activate cell death responses or lead to genome instability within the damaged cells (Barzilai 2010; McKinnon 2013; Reynolds and Stewart 2013). Several neurodegenerative disorders are caused by the presence of inefficient DSB repair (Lee and McKinnon 2007; Katyal and McKinnon 2008; Wilson et al. 2008; Jeppesen et al. 2011). DSBs are detected by DNA damage response (DDR) genes, which then activate several downstream targets, including cell cycle checkpoint proteins, chromatin-remodeling factors and other DNA repair components; these bring about an arrest of ongoing replication together with expression of genes that promote processes that resolve DSBs. Thus, the proper expression of DDR genes is critical to correct neural development (Carlessi et al. 2009; McKinnon 2009; Lee et al. 2012; Zhou et al. 2013; Pao et al. 2014).

When newly generated neurons exit the cell cycle and leave the VZ, they migrate through the SVZ and the intermediate zone (IZ), and finally reach the cortical plate (CP) (Nadarajah et al. 2003; Tan and Shi 2013). In the CP, late-arising neurons (E15.5–E16.5) migrate past earlier-arising neurons (E12.5–E14.5) and settle in more superficial positions, resulting in the inside-first, outside-last six-layered pattern of the cerebral cortex (Hatanaka et al. 2004; Hashimoto-Torii et al. 2008; Tan and Shi 2013). Radial migration plays a key role in establishing the layered structures of the cerebral cortex (Rakic 1972; Cooper 2008). There are at least 2 distinct migratory mechanisms involving cortical neurons. Nuclear translocation takes place involving the early-arising neurons to create/split-off the preplate (PP). In contrast, movement along the radial glia is adopted by late-arising neurons and this is known to require Cyclin-dependent kinase 5 (Cdk5) activity (Gupta et al. 2002; Hatanaka et al. 2004).

Cyclin-dependent kinase 12 (Cdk12) together with its activator, Cyclin K, phosphorylates Ser2 in the carboxyl terminal domain of RNA polymerase II, which is then responsible for the productive transcriptional elongation and synthesis of full-length mature mRNA (Bartkowiak et al. 2010; Liang et al. 2015). The Cdk12/Cyclin L1 complex is also involved in regulation of alternative splicing (Chen et al. 2006). In addition, Cdk12 has been implicated in the cancer predisposition via the regulation of DDR gene transcription (Blazek et al. 2011; Cancer Genome Atlas Research N 2011; Bajrami et al. 2014; Joshi et al. 2014; Natrajan et al. 2014; Ekumi et al. 2015; Geyer et al. 2015). We have previously reported that Cdk12 is essential for the proper development of the inner cell mass through the regulation of DDR genes (Juan et al. 2015) and is also required for the axonal elongation of cultured neurons via the regulation the transcription of Cdk5 (Chen et al. 2014). However, the roles of Cdk12 during neural development in vivo remain elusive. Here, we show that genetic ablation of Cdk12 in the developing nervous system of Nestin-Cre-driven Cdk12 conditional knockout (N-cKO) mice causes microcephaly and defective neurogenesis. At the onset of neurogenesis, the loss of Cdk12 prolongs the cell cycle and increases the apoptosis of proliferating neural progenitors, which consequently reduces neuronal production. By lineage tracing and in utero electroporation, we also have demonstrated that Cdk12 deletion perturbs the migration of late-arising neurons by downregulating Cdk5.

Materials and Methods

Experimental Animals

The production and genotyping of Cdk12^fx/fx mice has been described previously (Chen et al. 2014). Adult C57BL/6J mice were obtained from the National Laboratory Animal Center (Taipei, Taiwan). The Nestin-Cre transgenic mice were obtained from Dr Fu-Chin Liu (Tronche et al. 1999). The mice were handled according to the University guidelines and all experiments were approved by the National Yang-Ming University Animal Care and Use Committee. For the timed pregnancies, mice were set up in the late afternoon and when plugs were detected next morning, the stage of the mouse embryo was designated as E0.5.

Protein Extraction and Western Blotting

Previously described procedures were followed for these experiments (Wong et al. 2010).

In Vitro Magnetic Resonance Imaging Analysis for Volumetry

Brains from N-cKO and control littermates were fixed using 4% paraformaldehyde (PFA) with gentle agitation. Magnetic resonance imaging (MRI) measurements were carried out by embedding the fixed tissue in 1% agarose gel in a 15-mL falcon tube for analysis, which was performed by the Taiwan Mouse Clinic (Institute of Biomedical Sciences, Academia Sinica, Taiwan). Images were acquired using a Bruker Biospec spectrometer (Bruker) 4.7-T 40-cm bore horizontal-magnet (Magnex) MRI system. The MRI scanning used a spin-echo sequence (TR = 6000 ms, TE = 60 ms, NEX = 10, FOV = 2 × 2 cm, slice thickness = 0.5 mm, and matrix size = 256 × 256, scanning time = 32 min). Measurement of the image volume from the region of interest was carried out by Avizo 6.1 software (Zuse Institute).

Nissl Staining

Brains from N-cKO and control littermates were fixed by 4% PFA with gentle agitation. The fixed tissue was dehydrated through graded alcohols (70%, 90%, 95%), placed in xylene, and then embedded in molten paraffin wax. The paraffin embedded brains were sectioned at 15 µm. After overnight drying, slides were de-waxed in xylene (3 changes with 3-min incubations), rehydrated in 100% alcohol (2 changes with 3-min incubation), and stained in 0.1% cresyl violet for 15 min. Sections were then rinsed in tap water to remove excess stain and washed in 70% ethanol for 3 min. Next, the sections were dehydrated in absolute ethanol for 3 min with 2 changes; after this, the sections were cleared in xylene for 3 min with 2 changes. Finally, cover glasses were then mounted on the sections using Permount histology mounting medium (Millipore).

Immunohistochemistry

Brains from N-cKO and control littermates were fixed using 4% PFA at 4°C overnight with gentle agitation, cryopreserved in 30% sucrose, frozen, and finally stored at 80°C until use. For staining, 20-µm cryosections were made and incubated in blocking/permeabilization solution containing 3% normal goat serum (NGS) and 0.2% Triton-X in phosphate-buffered saline (PBS). The sections were treated overnight with appropriate primary antibodies diluted in 1% NGS/0.2% Triton X-100/PBS followed by the corresponding secondary antibodies for 3 h at room temperature. The following primary antibodies were used: rabbit anti-cleaved Caspase 3 (1:1000, Cell Signaling); rat anti-Ctip2 (1 µg/mL, Abcam); rabbit anti-Cux1 (0.4 µg/mL, Santa Cruz Biotechnology); rabbit anti-γ-H₂AX (2 ϖg/mL, Abcam); rabbit anti-Ki67 (1 µg/mL, Abcam); rabbit anti-Sox2 (1 µg/mL, Millipore); rabbit anti-Tbr1 (1 µg/mL, Abcam); mouse anti-TuJ1 (1 µg/mL, Covance); anti-β-tubulin (1:10 000, Developmental Studies Hybridoma Bank); and p-histone H3 (1:1000, Millipore). The following secondary antibodies were used: rhodamine (TRITC)-conjugated anti-mouse (1:1000, Jackson ImmunoResearch Labs); rhodamine-conjugated anti-rabbit (1:1000, Millipore); and DyLight™ 549 anti-rat (1:1000, Jackson ImmunoResearch Labs).

DiI Labeling

A small crystal of 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI, Sigma) was placed in the cerebral wall of P0 N-cKO and control littermates using an insect needle pin. The brains were incubated for 3 weeks in 4% PFA/PBS at 37°C, and then sectioned coronally at 100 µm thickness using a vibratome. The sections were collected as floating sections, which was followed by staining with DAPI (5 µg/mL, Roche) for 30 min at room temperature. Finally, the sections were mounted before analysis.

In Vivo EdU Labeling

Proliferation and migration analysis of the neurons of the embryos were detected by intraperitoneal injection of EdU into the pregnant mice (120 mg/kg) at E12.5, E14.5, E15.5, E16.5, and E17.5. Embryos were dissected out after 30-min or 24-h incubation and then the brains were harvested and fixed in 4% PFA/PBS overnight at 4°C with gentle agitation. Next, the brains were washed with PBS, immersed in 30% sucrose/PBS, and cryosectioned (coronal; 20 µm). EdU was detected using a Click-iT EdU Assay kits (Invitrogen). The sections were mounted in mounting medium before analysis. For the proliferation analysis, the number of EdU⁺ cells was manually counted within the E12.5, E14.5, E15.5, E16.5, and E17.5 cortex, and this number divided by the total number of EdU⁺ cells in control cortex for each image. For the migration analysis, the numbers of EdU⁺ cells in the various cortical layers of the E15–E17 cortex were manually counted and this number was divided by the total number of EdU⁺ cells in each image.

RNA Extraction, Reverse Transcription, and Real-Time PCR

The total RNA of the indicated samples was extracted using an RNeasy Plus Mini kit (Qiagen). Next, 2–3 µg RNA was used for cDNA synthesis using SuperScript reverse transcriptase III (Invitrogen) according to the manufacturer's instructions. The expression levels of various mRNAs were measured using 50-ng cDNA and TaqMan-based real-time PCR by an ABI StepOne™ real-time PCR machine. The sequences of the real-time PCR primers are shown in Supplementary Table 1. The samples were analyzed in triplicate and normalized against the expression level of the TATA-box binding protein.

Cell Cycle Analysis by Flow Cytometry

The brains of N-cKO and control littermates were dissected, titrated with a 1-mL pipette and then passed through a 70-µm cell strainer (BD Biosciences) to give single cells. These cells were then centrifuged at 1500 rpm for 10 min and the pellet of cells resuspended. This was followed by the addition of 600-µL red blood cell (RBC) lysis buffer (Invitrogen) to remove any RBCs that were present. Cells were centrifuged again, which was followed by fixing in 3 mL 70% ethanol at 4°C for overnight. Next, the cells were rehydrated with 5 mL PBS, and dissociated in 1 mL PBS. Finally, 100 µL of 200 µg/mL DNAase-free RNase and 100 µL of 1 mg/mL propidium iodide were added into samples for 30 min at room temperature and the cells analyzed using a flow cytometer (FACS Canto, BD Biosciences).

In Utero Electroporation

A laparotomy was performed and the embryos present within the uterine horn exposed; these were then gently placed on a sterile and irrigated gauze pad. After the embryos were visualized with a bright fiber optic lamp, a glass capillary tube was pulled to a fine point by high heat (Sutter Instrument, Co.) and filled with a mixture of fast green (Sigma) and appropriate plasmid (1 µg/µL). Plasmid had been purified using an Endofree Plasmid Maxi Kit (Qiagen) and was dissolved in endotoxin-free TE buffer. The pipette tip was then inserted through the uterus into one of the lateral ventricles and 1 µL of the plasmid mixture injected. Electroporation was conducted using the following parameters: five 40-V pulses, P (on) 50 ms, P (off) 450 ms. Successful injection was confirmed visually by observing the ventricle being filled with fast green. The embryos within the uterus were returned to the body cavity and the incision was closed with sterile silk sutures. To determine the percentage of electroporated-GFP⁺ and/or electroporated-tdTomato⁺ cells that were present in the different regions of the cerebral cortex, we divide the cortex equally into 5 divisions according to distance from the pial surface to the ventricle. Next, the GFP⁺ cells were counted in each division and the number was divided by the total number of GFP⁺ cells per 100 µm². The measurements were collected from 3 randomly selected squares in each division.

Results

Genetic Ablation of Cdk12 Causes Microcephaly with a Petite Cerebral Cortex and a Defective Corpus Callosum

To identify the roles of Cdk12 during neural development in vivo, we conditionally targeted Cdk12 in the nervous system by mating Cdk12^fx/del mice (Chen et al. 2014) with a Nestin-Cre^+/0 mouse line expressing Cre recombinase in the neural tube beginning at E10.5 (Supplementary Fig. 1A) (Tronche et al. 1999). The mice with the Nestin-Cre-mediated deletion of Cdk12 (Nestin-Cre^+/0;Cdk12^fx/del) are referred henceforth as N-cKO mice. The Cdk12^fx/+ mice were used as control mice. Cdk12 protein could be barely detected in the postnatal day 0 (P0) brain of the N-cKO mice as revealed by western blot analysis (Fig. 1A). Interestingly, the amount of Cdk5 protein present is also lower in the N-cKO mice, which is consistent with our previous observation in the E14.5 brain (Chen et al. 2014) (Fig. 1A). N-cKO mice were born at the expected Mendelian ratio (Table 1), but these mutant mice do not survive beyond P1. These mice exhibited a limited capacity for movement with drooping forelimbs; they died within a few hours of birth. The N-cKO mice at P0 show a reduced body weights in comparison to the control mice (Fig. 1B,C). The brains of the P0 N-cKO mice show a 32% reduction in weight compared with the control mice (Fig. 1E). MRI analysis of P0 N-cKO mice showed that the volume of the whole brain is 30% smaller and the volume of the cerebral cortex showed a 52% reduction (Fig. 1F,G). The area, length, and width of the cerebral cortex are proportionally downscaled in the N-cKO mice (Fig. 1D, Supplementary Fig. 1D). Furthermore, the olfactory bulb and the cerebellum were also smaller in the N-cKO mice compared with the control mice (Supplementary Fig. 1B,C). Histological analysis of coronal sections of the P0 brains showed that the cerebral cortex, hippocampus, striatum, and midbrain were reduced in size and that the thickness of the cortical layers were reduced in the N-cKO mice compared with the control mice (Fig. 1H–J). The P0 mutant mice also exhibit an abnormal corpus callosum with axons not crossing into the contralateral hemisphere; this was revealed by the DiI labeling (Fig. 1K). Additionally, several axonal tracts, such as the anterior commissure, the stria medullaris, the mammilothalamic tract, the internal capsule, and the posterior commissure, were found to be markedly diminished in the N-cKO mice compared with those present in their control littermates (Fig. 1H).

Table 1

Survival of progeny after mating between Cdk12^fx/fx and Nestin-Cre^+/0; Cdk12^del/+ is shown

Stage	Cdk12 del/fx Expecting: 25%	Cdk12 +/fx Expecting: 25%	Nestin-Cre+/o; Cdk12fx/+ Expecting: 25%	Nestin-Cre+/o;Cdk12fx/del Expecting: 25%
E12.5	6/25	5/25	8/25	6/25
E13.5	7/23	6/23	5/23	5/23
E14.5	20/83	22/83	20/83	21/83
E15.5	8/27	8/27	3/27	8/27
E16.5	9/27	7/27	4/27	7/27
E17.5	11/36	7/36	11/36	7/36
E18.5	3/15	3/15	4/15	5/15
P0	26/134	29/134	43/134	36/134
P1–1 week	8/31	9/31	14/31	0/31
3 weeks	2/14	5/14	7/14	0/14

Figure 1.

Abnormalities of the cerebral cortex in N-cKO mutant mice. (A) P0 whole brain lysates of control and Nestin-Cre;Cdk12-cKO (N-cKO) littermates were subjected to western blot analysis using anti-Cdk12 or anti-Cdk5 antibody. β-Tubulin (β-tub) was used as loading control. (B) The sizes of the P0 control and N-cKO littermates are compared. (C) Measurement of body weights of P0 control (n = 25) and N-cKO littermates (n = 24). (D) Morphologies of the brains dissected from the P0 control and N-cKO littermates are shown. Scale bars: 1 mm. (E) Measurement of brain weights (Control, n = 8; N-cKO, n = 5). (F) Magnetic resonance imaging analysis of the P0 control and N-cKO littermates. (G) Quantitative results for the whole brain volume from (F) are showed (Control, n = 5; N-cKO, n = 5). (H) Coronal sections of the P0 brains along the different rostral-caudal axes are labeled with cresyl violet. Arrowheads indicate axonal tracts. Scale bars: 1 mm. (I) Thinner cortical layers are present in the P0 N-cKO mice. Scale bar: 100 µm. (J) Quantitative results from (I) are shown (Control, n = 3; N-cKO, n = 3). (K) Axonal tracts in the corpus callosum as revealed by anterograde DiI labeling. Lower panels are magnified photos of the squares shown in the upper panels. Scale bar: 300 µm. C, cortex; cc, corpus callosum; Cb, cerebellum; H, hippocampus; ic, internal capsule; M, midbrain; O, olfactory bulb; S, striatum; str, stria medullaris. Statistics: data from at least 3 embryos were analyzed by Student's t-test and are presented as mean ± SEM; *P < 0.05; **P < 0.01.

Deletion of Cdk12 Decreases the Proliferation of Neural Progenitor Cells via a Prolongation of the Cell Cycle and This Leads to Defective Neurogenesis

Given the dramatic reduction in the size of the cerebral cortex in the N-cKO mice, we focused our investigation on the functions of Cdk12 within the developing cerebral cortex. The reduced brain size of the mutant mice begins to be detectable at E14.5 (Supplementary Fig. 1F). We reasoned that the microcephaly phenotype of the mutant mice may result from defective neurogenesis and that this could be due to either a reduced proliferative capacity and/or an increase in the death of neural progenitor cells. To detect the presence of neural progenitor cell proliferation, EdU (5-ethynyl-2′-deoxyuridine) was administrated half an hour before sacrificing the animals. The majority of EdU-positive cells in the developing cortex at E12.5 were found to be positioned in the VZ and SVZ (zone 1, Z1) with only a few EdU-positive cells in the preplate (Z2) (Fig. 2A). No difference in the number of EdU-positive cells between the mutant and control mice could be detected at E12.5. However, the total number of proliferating cells in the mutant mice had decreased to 92.8 ± 1.74% (mean ± SEM) of those in the control mice at E14.5 and this reduction continued resulting in the total number be 73.5 ± 5.61% of that of the control mice at E15.5 (Fig. 2A,B). To reveal why there was a reduction in the proliferation capacity at E14.5, we assessed whether defects in cell cycle progression were present by examining DNA content of the cells using flow cytometry. It was found that the tetraploid subpopulation in N-cKO mice had increased, which presumably is caused by arrest G2 and M phases (16.4 ± 0.46% vs. 10.9 ± 0.73% in the control mice) (Fig. 3A,B). Furthermore, there was presumably more cells undergoing cell death in the brains of the mutant mice because many of the cells had a DNA content of less than 2n (17.4 ± 0.96% vs. 5.3 ± 0.93% in the control mice). To clarify whether there are more progenitors retained at M phase, coronal sections of the cerebral cortex were stained with phospho-histone H3 (PH3) antibody, which is able to mark M phase cells. A 1.3- to 4.6-fold increase in PH3-positive mitotic cells located within the VZ and CP in N-cKO mice compared with the control mouse brain between E14.5 and E17.5 was noted (Fig. 3C,D), which suggests that the neural progenitor cells have a longer cell cycle time when Cdk12 has been deleted.

Figure 2.

Deletion of Cdk12 decreases the proliferation capacity of progenitor cells in the developing cerebral cortex. (A) Control and N-cKO embryos (littermates) at various developmental stages were subjected to a 30-min EdU labeling and were sacrificed immediately. Then 20-µm-coronal sections were developed for EdU fluorescence and stained with DAPI. The cortex has been divided into zone 1 (Z1), zone 2 (Z2), zone 3 (Z3) from the ventricle to the pia surface. Scale bar: 100 µm. (B) The quantitative results of (A) are shown (Control, n = 3; N-cKO, n = 3, at each developmental stage). Statistics: data from 3 embryos of each developmental stage were analyzed by Student's t-test and are presented as mean ± SEM; *P < 0.05; **P < 0.01.

Figure 3.

Cdk12 deficiency leads to neural progenitors accumulating in G2 and M phases. (A) Flow cytometric cell cycle analysis of the cerebral cortex. The E14.5 cortices were dissociated and the DNA content of each cell was analyzed by a cell sorter after labeling with propidium iodide. Cells are grouped to sub-G1 (genome content <2n), G1 (genome content = 2n), and S/G2/M (genome content >2n) phases. (B) Quantitative results of the ratios of cells at different cell cycle phases as displayed in (A) are shown. Control, n = 3; N-cKO, n = 3. (C) Cerebral cortices from the control and N-cKO embryos were stained with anti-phospho histone 3 (PH3) antibody to reveal mitotic cells. (D) Quantification of (C) is shown. Control, n = 3; N-cKO, n = 3. Statistics: data from 3 embryos of each developmental stage were analyzed by Student's t-test and are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001.

Cdk12 Deficiency Downregulates DDR Genes, Leading to Inefficient DNA Repair by Neural Progenitor Cells

Previous studies of Cdk12 functions using tumor cell lines have shown that Cdk12 controls DNA repair by regulating various DDR genes, such as Atr, Atm, Brca1, Fancd2, and Fanci (Blazek et al. 2011). Results from our study on the development of the early embryo showed that loss of Cdk12 reduces expression of Atr, Brca1, Fancd2, and Fanci (Juan et al. 2015). Based on the above results, we examined expression of these DDR genes in the brain by quantitative RT-PCR. At E12.5, only Fanci expression is downregulated to 42.3 ± 0.17% of that of the control littermates (Fig. 4A). However, expression of Atm, Atr, and Fanci were found to be lowered in the E14.5 N-cKO cerebral cortex, with their levels ranging from 30.3 to 50.5% of those in the control mouse cortex (Fig. 4A). Finally, Cdk5 was also downregulated in E14.5 N-cKO cerebral cortex. The level of Cyclin K mRNA in the cerebral cortex was used as the control for the RT-PCR (Fig. 4A).

Figure 4.

Cdk12 deficiency downregulates DDR genes, which leads to DNA double-strain breaks in the neural progenitor cells. (A) The cerebral cortices of E12.5 and E14.5 embryos were dissected out and subjected to quantitative-PCR (Q-PCR) analysis. The expression levels of the various RNAs of interest were adjusted based on the level of TATA binding protein in each sample. The graphs present the mRNA expression levels of the indicated DDR genes and of Cdk5 in the control and N-cKO embryos. The Q-PCR was performed in triplicate for each embryo. Data are presented from 3 individual embryos in each group. Cyclin K was used as Q-PCR control. (B) The cerebral cortices of E12.5, E3.5, and E14.5 embryos were stained with anti-γ-H2AX and Sox2 antibodies that label DNA double-strand breaks and neural progenitor cells, respectively. Statistics: data from 3 embryos of each developmental stage were analyzed by Student's t-test and are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001.

To confirm the functional consequences of the downregulation of Atm, Atr, and Fanci, we analyzed whether there is an increase in the formation of DSBs in the N-cKO neural progenitor cells; this was done by staining with anti-γ-H₂AX antibodies at E12.5, E13.5, and E14.5 (Fig. 4B). There are more γ-H₂AX-positive cells among the neural progenitor cells (labeled by SOX2 in adjacent sections) (Fig. 4B). These findings indicate that deficiency in Cdk12 reduces the expression of DDR genes in the neural progenitor cells and this causes an increase in DSBs in the neural progenitor cells.

Deletion of Cdk12 Leads to Increased Cell Death among Neural Progenitor Cells and Postmitotic Neurons

When DNA damage is not repaired, cells usually undergo senescence or apoptosis (D'Sa-Eipper et al. 2001; Roos and Kaina 2006; d'Adda di Fagagna 2008). Based on this, we assessed whether there is an increase in apoptosis of neural progenitor cells and/or postmitotic neurons. Neural progenitors were labeled with a 30-min short-pulse of EdU at E14.5 and E17.5 and then subjected to immunostaining with antibody against cleaved caspase 3 (cCasp). More cCasp-positive cells are present in the VZ and SVZ of the mutant cortex than in control cortex at E14.5 (Fig. 5A, closed arrowheads, 28 ± 4.7 in N-cKO mice vs. 5 ± 3.5 in control mice, Fig. 5D). Furthermore, many proliferating cells were found to undergo apoptosis (cCasp/EdU double-positive cells, 8 ± 1.5 in N-cKO mice vs. 2 ± 1 in control mice, red bars in Fig. 5D). A similar pattern was also detected in the E17.5 mutant cortex (Fig. 5B,D). These results suggest that knock-out of Cdk12 leads to apoptosis of neural progenitor cells beginning at E14.5. Interestingly, we observed the presence of many cCasp-positive cells in the IZ and CP of E17.5 and P0 mutant brains, suggesting that loss of Cdk12 also triggers apoptosis of postmitotic neurons (Fig. 5A–C open arrowheads).

Figure 5.

An increase in the level of apoptosis in neural progenitor cells can be detected in N-cKO mice. (A and B) Control and N-cKO embryos (littermates) at E14.5 and E17.5 were labeled with EdU for half an hour and sacrificed afterward. About 20-µm-coronal sections were developed for EdU fluorescence (green) and stained with antibody against cleaved caspase 3 (cCasp, red) and with DAPI (blue). Magnified areas of squares (a) and (b) are shown in upper right panels. The white arrowheads indicate neural progenitor cells and the open arrowheads indicate postmitotic neurons. Scale bar: 100 µm. (C) Apoptotic cells in P0 cortex were detected by anti-cCasp antibody. Higher magnifications of the square areas in upper panels are shown in the lower panels. The white arrowheads point to progenitor cells and the open arrowheads point to postmitotic neurons. Scale bar: 100 µm. (D) Quantification of the numbers of EdU⁺cCasp⁺ and EdU⁻cCasp⁺ cells per section in the ventricular zone and in the subventricular zone, as shown in (A) and (B), are presented. Control, n = 3; N-cKO, n = 3. (E) Quantitative results of the number of cCasp⁺ cells per section in the cortical plate from (C) are presented. Control, n = 3; N-cKO, n = 3. Statistics: data from 3 embryos of each developmental stage were analyzed by Student's t-test and are presented as mean ± SEM; *P < 0.05.

Cdk12 Controls the Migration of Late-Arising Neurons

The above experiments point out the importance of Cdk12 to the repair of the DNA damage present in neural progenitor cells during their proliferation. While conducting the 30-min EdU-labeling experiments, we noticed that there are significant numbers of EdU-labeled cells entering the IZ of the control mice, but very few EdU-labeled cells were present in the IZ of the mutant mice (Fig. 2A,B), which raises the possibility that Cdk12 is also involved in neuronal migration. To delineate whether a deficiency in Cdk12 leads to a delay in neuronal migration, we traced cells during the various developmental stages of the embryos after incorporation of EdU for 24 h. When cortical cells are examined at E15.5 (labeled at E14.5), there is no significant difference in the distribution of EdU-positive cells across the VZ/SVZ, IZ, and CP between the control mice and their N-cKO littermates (Fig. 6A,B). At E16.5 (labeled at E15.5), more EdU-positive cells were found to be stagnating within at the VZ/SVZ (83.33 ± 0.67% in mutant mice vs. 69.33 ± 4.81% in control mice) and fewer EdU-positive cells were found to be located within the IZ and CP (14.67 ± 0.33% in IZ and 2.00 ± 0.58% in CP of the mutant mice vs. 24.33 ± 2.96% in IZ and 6.33 ± 1.86% in CP of the control mice, Fig. 6A,B). A similar distribution was also found for the E17.5 mutant mice (89.33 ± 1.45% in VZ/SVZ, 9.33 ± 1.2% in IZ, 1.33 ± 0.33% in CP of the mutant mice vs. 75 ± 1.53% in VZ/SVZ, 20.67 ± 1.20% in IZ, 4.33 ± 0.33% in CP of the control mice). These results suggest that Cdk12 facilitates neuronal migration of late-arising neurons, namely those originating after E15.5.

Figure 6.

Delayed migration and ectopic localization of late-arising neurons in N-cKO cerebral cortex. (A) Birth dating and cell tracing by a 24-h EdU labeling. EdU was injected peritoneally into E14.5, E15.5, and E16.5 embryos. Embryos were then sacrificed 1 day later and 20-µm-coronal sections of cerebral cortex were developed for EdU fluorescence and were also stained with DAPI. Scale bar: 100 µm. (B) Quantitative results of distribution of EdU⁺ cells from (A) are showed. 3 embryos from each stage are analyzed. (C) P0 cerebral cortex of control and N-cKO mice were stained with anti-Tbr1, anti-Ctip2, and anti-Cux1⁺ antibodies that identify cells in layers VI, VI–V, IV–II, respectively. (D) Distribution of Cux1⁺ cells as revealed in (C) is quantified. Control, n = 3; N-cKO, n = 3. Scale bar: 100 µm. Statistics: data from 3 embryos of each developmental stage are analyzed by Student's t-test and are presented as mean ± SEM; *P < 0.05, **P < 0.01, ***P < 0.001.

We next investigated whether loss of Cdk12 changes the layer stratification of the cortical plates at P0. The immunofluorescence staining patterns of the neurons labeled with Tbr1, which delineates layer VI, and Ctip2, which delineates layers VI–V, (Dekimoto et al. 2010; Dudok et al. 2013) are similar when the control and mutant mice are compared (Fig. 6C). However, strikingly, Cux1-positive neurons, which in control mice are mainly locate in layers IV–II, are rather present in layers VI–IV of the mutant mice (Fig. 6C,D), which confirms that the loss of Cdk12 leads to the abnormal migration of late-arising neurons that destined for layers IV–II. There is also a 12% reduction in Cux1-positive cells in the P0 N-cKO cortex (data not shown), a result that is likely also to be due to the effects of Cdk12 on neurogenesis.

As Cdk12 is involved in maintenance of the neural progenitor pool, the observed effects of Cdk12 on migration of late-arising neurons may be a consequence of the early effects of Cdk12 on neural progenitor cells. To circumvent this confounding factor, we performed in utero electroporation of Cre-expression plasmids (pCre-GFP) into lateral ventricles of Cdk12^fx/fx mice at E13.5, E14.5, and E15.5 to delete Cdk12 after the peak of neurogenesis. To quantify the distribution of transfected cell in terms of their location at P0, we divide the cortex evenly into 5 zones from the ventricle to pial surface (Fig. 7A). When electroporation was performed at either E13.5 or E14.5, the distribution of GFP-positive cells in the P0 cerebral cortex was similar between Cdk12^fx/fx (transfected with GFP) and Cdk12^del/del cells (transfected with pCre-GFP) (Fig. 7A). However, if the electroporation was conducted at E15.5, many Cdk12^del/del cells remained within zone 2 (50 ± 2.80% vs. 15.67 ± 2.33% in the control), and fewer GFP-positive Cdk12^del/del cells reached zone 5 (12.33 ± 1.45% vs. 54 ± 1.00% in the control) (Fig. 7A). To verify that a similar proportion of cells exits from cell cycle in these in utero electroporation experiments, embryos were harvested 24 h after electroporation and immunostained with the anti-Ki67 antibody that labels proliferating cells. No significant difference in Ki67⁻/GFP⁺ cells could be detected between the GFP- electroporated and Cre-electroporated groups (Fig. 7B,C).

Figure 7.

The effects of deleting Cdk12 on migrating late-arising neurons are partially rescued by Cdk5 overexpression. (A) The indicated vectors were electroporated into developing the cerebral cortex of E13.5, E14.5, or E15.5 Cdk12^fx/fx embryos. At P0, the ratios of electroporated cells in indicated regions (zone 1–5) were quantified. GFP group, n = 3; Cre/GFP, n = 3; Cre/GFP + Cdk5, n = 4. Scale bar: 100 µm. (B) GFP vector or Cre/GFP vector was introduced into the developing cerebral cortex of E15.5 Cdk12^fx/fx embryos. Embryos were fixed 1 day later and coronal sections of cerebral cortex were stained with Ki67 antibody (red). Lower panels are magnified images of squares shown in the upper panel. Scale bar: 100 µm. (C) Quantification results from (B) are shown. No significant difference in the numbers of Ki67⁻GFP⁺ cells could be detected when the control and Cdk12-deleted embryos were compared (GFP, n = 3; Cre/GFP, n = 3). (D) Quantification results from E15.5 in utero electroporation experiments as shown in (A) are compared between the different groups. (E) High-magnification views of the cellular morphology of the VZ/SVZ, IZ, and CP of E15.5 embryos electroporated with different vectors. Multipolar cells (MCs) are marked with white arrowheads and bipolar cells (BCs) are marked with open arrowheads. Scale bar: 100 µm. (F) Percentage of cells with multipolar or bipolar morphologies quantified from (E). (G) The percentages of BCs in the various different cortical areas were quantified and are shown in (E). Statistics: data from at least 3 embryos for each condition were analyzed by Student's t-test and are presented as mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001.

Cdk5 Partially Restores the Neuronal Migration Defect in Cdk12-Depleted Late-Arising Neurons

During cortical development, migrating neurons undergo a multipolar-to-bipolar transition in the IZ and this initiates the radial migration into the CP (Tan and Shi 2013) (Fig. 7E, left panels). We observed that many stagnant GFP-positive Cdk12^del/del cells are multipolar in VZ/SVZ and IZ (Fig. 7E, middle panels). Furthermore, although some cells make the transition to a bipolar morphology, they do not align radially and fewer bipolar cells are thus able to reach the CP compared with the control mice (Fig. 7A,E). Thus, our results suggest that Cdk12 is required for late-arising neurons to make the morphological transition from the multipolar to the bipolar shape, to rotate from the horizontal axis to the radial axis and then to migrate into the proper CP layers (Fig. 7E). As Cdk5 is reduced at both the protein and mRNA level on loss of Cdk12 (Figs 1A and 4A) (Chen et al. 2014) and is an important factor that is involved in the regulation of the migration of late-arising neurons (Gupta et al. 2002; Hatanaka et al. 2004), we examined whether the observed effects of Cdk12 on neuronal migration are mediated via Cdk5. A Cdk5 expression vector (pCAG-Cdk5-tdTomato) and pCre-GFP were co-electroporated into Cdk12^fx/fx mouse brains at E15.5 (Fig. 7A, right panels). Introducing Cdk5 increases the percentage of neurons reaching zone 5 (38.33 ± 7.26% vs. 12 ± 1.15% in Cdk12^del/del cells, Fig. 7A). Cdk5 also was able to rescue the multipolar-to-bipolar transition defect and reduces the percentage of stagnant bipolar cells in the IZ (Fig. 7E–G). These results indicate that the neuronal migration defects of Cdk12^del/del cells are partially rescued by overexpression of Cdk5.

Discussion

Cdk12 is Essential for Overcoming Replication Stress and Does This via the Regulation of DDR Genes, and Controls Late-Arising Neuronal Migration in a Cdk5-Dependent Manner

In this study, we report on the functions of Cdk12 during neural development in vivo (Fig. 8A). Disruption of Cdk12 at E10.5 in neural progenitor cells increases DNA DSBs at E12.5. The underlying mechanism is that Cdk12 downregulates DDR genes, which results in progenitor cells spending longer period in the G2 and M phase. The accumulation of a large amount of unrepaired DNA damage then triggers apoptosis of the neural progenitor cells at E14.5 (Fig. 8B). At E15.5, in addition to the effect of Cdk12 on neural progenitor cells, Cdk12 also play a role in the migration of late-arising neurons (Fig. 8A). In the absence of Cdk12, many late-arising neurons stall within the IZ. These Cdk12 mutant cells fail to undergo the multipolar-to-bipolar transition, and there is a failure of the cell's axis to align with the processes of the radial glia calls (Fig. 8C). Even if these Cdk12-depleted migrating neurons are successful at becoming bipolar, they are unable to reach the layers IV–II of the developing cerebral cortex. A reduced expression of Cdk5 in the Cdk12-deficient neurons seems to be responsible, at least in part, for these defects in morphological maturation and in the neuron migration machinery (Fig. 8C). Thus, we propose a dual role for Cdk12 in neurogenesis and late stage neuronal migration via 2 distinct pathways.

Figure 8.

A proposed model for the functioning of Cdk12 in developing cerebral cortex. (A) Schematic presentation of the cellular defects in the Cdk12 knockout cortex at various different developmental stages. These defects include DNA damage (marked with yellow +) and apoptosis (hollow cells) in proliferating neural progenitor cells, and flawed migration of late-arising neurons. Apoptosis also occurs in differentiating neurons. (B, C) Detailed mechanisms of Cdk12 functioning in the developing cerebral cortex. (B) Action of Cdk12 regarding DNA repair in the proliferating progenitor cells of the developing cerebral cortex. Cdk12 regulates the DDR repair system of the neural progenitor cells to overcome the replication stress during proliferation (left panel). Disruption of Cdk12 in the neural progenitor cells at E10.5 causes an increase in DNA double strand breaks, which in turn leads to an accumulation of neural progenitor cells at G2 and M phases. The longer cell cycle and subsequent apoptosis of the neural progenitor cells results in defective neurogenesis (right panel). (C) Action of Cdk12 regarding the migration of late-arising neurons in the developing cerebral cortex. When late-arising migrating neurons reach the IZ, they undergo a multipolar-to-bipolar transition. The bipolar cells (BCs) then migrate to layers IV–II (left panel). In the absence of Cdk12, there is a lower expression of Cdk5 in late-arising neurons, which then results in a failure of the neurons to undergo the appropriate morphological change from a multipolar shape to a bipolar shape within the IZ. Furthermore, these cells also fail to reach layer IV–II, even if they make the multipolar-to-bipolar transition (right panel).

Loss of Cdk12 Causes Microcephaly, Which is Similar to the Phenotype Observed in DDR Knockout Mice

Cdk12 N-cKO mice exhibit microcephaly, which is a feature common to several DDR knockout mice (O'Driscoll and Jeggo 2008; Lee et al. 2012; Li et al. 2012; Zhou et al. 2013; Pao et al. 2014). The mechanisms by which the phenotypic defects in mice with a deficiency in early DDR genes, such as Atr, Brca1, and Nbs1, are created has been proposed to be alterations in neural progenitor cells that affect the response signaling to DNA damage and DNA repair, which in turn ultimately lead to a loss of neurons (Lee et al. 2012; Li et al. 2012; Pao et al. 2014). These findings indicate that normal DDR activities are critical during neural development. Disruption of Atr, Brca1, or Nbs1 in mice also cause an increase in spontaneous apoptosis in the developing nervous system (Lee et al. 2012; Li et al. 2012; Pao et al. 2014), a phenotypes that is similar to what we have observed in the Cdk12 N-cKO mice studied here. The closely correlation between the mutant phenotypes of DDR genes and the mutant phenotype of Cdk12 gives strong support to the hypothesis that Cdk12 is a key factor that is able to orchestrate the transcription of DDR genes during cortical development.

An Inefficient DNA Repair System Leads to a Prolonged Cell Cycle, Which in Turn Leads to a Reduction in the Proliferation Capacity of Cdk12-Depleted Neural Progenitor Cells

DDR genes are involved in sensing and repairing DSBs via the phosphorylation of target proteins, which in turn has an effect on cell cycle progression and/or apoptosis (Lobrich and Jeggo 2007). Passage through a checkpoint from one cell cycle phase to the next requires a coordinated set of proteins that monitor cell growth and DNA integrity (Lobrich and Jeggo 2007). Uncontrolled cell division or the propagation of damaged DNA contributes to genomic instability and tumorigenesis (Wang et al. 2015). Thus, eukaryotic organisms frequently prefer to opt for cell death rather than risking the proliferation of damaged cells (Zhou et al. 2013). As there are more neural progenitor cells in the N-cKO cortex accumulating at G2 and M phases that have DSBs, we reason that the M phase cells are not able to finish mitosis and these cells then rather undergo apoptosis as was observed within the VZ/SVZ (Figs 3D and 5A). Our findings indicate that Cdk12 is essential for the proliferation of neural progenitor cells and perturbation of this pathway increases the sensitivity of these cells to naturally accumulating DNA damage during DNA replication.

Deletion of Cdk12 Perturbs Migration of Late-Arising Neurons but not Early-Born Neurons

Deletion of Cdk12 in neural progenitor cells in N-cKO mice results in the defective migration of cerebral cortical Cux1-positive neurons that are located within layers IV–II (Fig. 6C,D). Results from our in utero electroporation experiments demonstrated that Cdk12 acts on the migration of neurons arising at about E15.5, but not neurons arising before this time point (Fig. 7A). During corticogenesis, late-arising Cux1-positive callosal projection neurons (CPNs) acquire their laminar position at layers IV–II and project their axons to the contralateral hemisphere (Fame et al. 2011; Greig et al. 2013). As these CPNs are misplaced in N-cKO mice, it is likely that projection of their axons is disoriented and this leads to the absence of the corpus callosum as shown in Figure 1K. These findings further confirm that the loss of Cdk12 results in a delayed migration of neurons within the superficial layers. In contrast, projection from cortex to thalamus seems to be normal in N-cKO mice, albeit with fewer axons being projected (Supplementary Fig. 1F), which gives support to the hypothesis that early-arising neurons are able to arrive in layer V and also reside in a correct position within layer V.

Cdk5 Acts as one of the Downstream Targets of Cdk12 During the Regulation of Neuronal Migration

One of our most interesting findings in the current study is that there is downregulation of Cdk5 and defective neuronal migration in the developing cortex of Cdk12 N-cKO mice. We uncover the fact that Cdk12 has a previously unreported role in migration and act in this during neural development via the Cdk5 pathway. During radial migration in the developing cerebral cortex, neurons change their morphology from multipolar to bipolar in the IZ (Cooper 2014; Ohshima 2014). This process requires the functioning of cytoskeletal regulators (Nadarajah et al. 2003; Cooper 2014; Ohshima 2014). Bipolar cells then migrate along the radially oriented processes of radial glia to reach the CP (Noctor et al. 2004; Ohshima 2014). Cdk5 is a pivotal factor that regulates multiple steps during the radial migration of cortical neurons (Ohshima et al. 2007; Ohshima 2014). Cdk5 is essential for the proper development of the central nervous system, as is evident from the abnormal neuronal migration, cortical lamination and perinatal death of Cdk5 knockout mice (Ohshima et al. 1996). It has been demonstrated that substrates of Cdk5 are able to modulate cytoskeletal reorganization. For example, activation of β-catenin and RapGEF2 by Cdk5 regulates their association with the cadherins that are required for both neuronal migration and the multipolar–bipolar transition (Ohshima 2014; Ye et al. 2014). In this study, in addition to a lower expression of Cdk5 at the protein and RNA level that could be detected in the N-cKO mice, the phenotype of late-arising neurons that have a deficiency of Cdk12 fits well with what would be expected if Cdk5 expression is decreased. Furthermore, overexpression of Cdk5 is able to partially rescue the effects of Cdk12 loss on neuronal migration. As higher Cdk5 activity is observed in Alzheimer's patients, thus an inhibition of Cdk12, which in turn decreases Cdk5 expression, may be one approach to reducing progression of this disease.

Supplementary Material

Supplementary material can be found at: http://www.cercor.oxfordjournals.org/.

Authors’ Contributions

H.-R.C. and M.-J.F. designed the experiments and wrote the manuscript. H.-R.C. conducted most of experiments. H.-C.J. and Y.-H.W. contributed to data interpretation and revising the manuscript. J.-W.T. provided the in utero electroporation technique and material support.

Funding

This research was supported by the Ministry of Science and Technology (NSC101-2320-B-010-064, MOST 104-2320-B-010-022-MY3) and the Ministry of Education (Aim for the Top University Plan) to M.-J.F..

Notes

We thank M.-M. Poo and J.-Y. Yu for their discussion during the project. We would also thank the Taiwan Mouse Clinic Core for conducting the MRI analysis, and the Developmental Studies Hybridoma Bank for providing monoclonal antibodies. Conflicts of Interest: None declared.