|Home | About | Journals | Submit | Contact Us | Français|
During development of the cerebral cortex, neural stem cells divide to expand the progenitor pool and generate basal progenitors, outer radial glia and cortical neurons. As these newly born neurons differentiate, they must properly migrate toward their final destination in the cortical plate, project axons to appropriate targets, and develop dendrites. However, a complete understanding of the precise genetic mechanisms regulating these steps is lacking. Here we show that a member of the nuclear factor one (NFI) family of transcription factors, NFIB, is essential for many of these processes in mice. We performed a detailed analysis of NFIB expression during cortical development, and investigated defects in cortical neurogenesis, neuronal migration and differentiation in NfiB−/− brains. We found that NFIB is strongly expressed in radial glia and corticofugal neurons throughout cortical development. However, in NfiB−/− cortices, radial glia failed to generate outer radial glia, subsequently resulting in a loss of late basal progenitors. In addition, corticofugal neurons showed a severe loss of axonal projections, while late-born cortical neurons displayed defects in migration and ectopically expressed the early-born neuronal marker, CTIP2. Furthermore, gene expression analysis, by RNA-sequencing, revealed a misexpression of genes that regulate the cell cycle, neuronal differentiation and migration in NfiB−/− brains. Together these results demonstrate the critical functions of NFIB in regulating cortical development.
The mammalian cerebral cortex is a six-layered structure, responsible for higher cognitive function and sensory perception. It is composed of many different neuronal subtypes, each with unique characteristics including laminar position, gene expression profile and axonal targets. The cortex is generated in an inside-out fashion such that cortical projection neurons generated early in development occupy deep layers (layers 5 and 6), while those that reside in upper layers (layers 2–4) are generated later (McConnell et al., 1995; Hevner et al., 2006; Molyneaux et al., 2007).
All cortical projection neurons are generated by neural progenitors in the proliferative region of the cortex, encompassing both the ventricular and subventricular zones (VZ and SVZ) (Haubensak et al., 2004; Farkas and Huttner, 2008). To date, three different types of neural progenitors have been identified: radial glia, outer radial glia and basal progenitors. They each have distinct roles during neurogenesis and can be distinguished by characteristics including morphology, location of cell body and expression of molecular markers (Farkas and Huttner, 2008; Wang et al., 2011a; Shitamukai and Matsuzaki, 2012).
At the onset of neurogenesis, neuroepithelial cells differentiate into radial glia (Shitamukai and Matsuzaki, 2012), which can be genetically defined by their expression of transcription factors, PAX6 and SOX2 (Gotz et al., 1998; Tarabykin et al., 2001; Bani-Yaghoub et al., 2006). Highly polarized, radial glia have apical and basal processes that attach to ventricular and pial surfaces of the cortex, respectively (Shitamukai and Matsuzaki, 2012). Their cell bodies reside in the VZ and they can undergo multiple rounds of cell division, which take place at the ventricular surface (Pontious et al., 2008). Radial glia can self-renew and also divide to generate basal progenitors, outer radial glia and cortical neurons (Shitamukai et al., 2011; Shitamukai and Matsuzaki, 2012). However, the main function of radial glia is likely to expand the progenitor pool (Kowalczyk et al., 2009; Lui et al., 2011; Wang et al., 2011b).
Basal progenitors, the second major class of neural progenitors, uniquely express TBR2, a T-box transcription factor (Arnold et al., 2008; Sessa et al., 2008; Kowalczyk et al., 2009). Unlike radial glia, basal progenitors are multipolar, lack apico-basal processes and their cell bodies reside throughout the SVZ. Although some have reported the potential for few rounds of self-renewing divisions, basal progenitors predominantly undergo neurogenic symmetric division (Farkas and Huttner, 2008; Noctor et al., 2008; Shitamukai and Matsuzaki, 2012). Thus, the main function of basal progenitors is the generation of cortical neurons (Noctor et al., 2004; Pontious et al., 2008; Kowalczyk et al, 2009).
Outer radial glia, the third subclass of neural progenitor, are similar to radial glia in that they have a basal process and express PAX6 and SOX2. However, they lack an apical process and their cell bodies reside in the superficial (outer) region of the SVZ (oSVZ). They are capable of asymmetric self-renewal and also generate neurons or basal progenitors (Reillo et al., 2011; Shitamukai et al., 2011; Wang et al., 2011a, 2011b; Shitamukai and Matsuzaki, 2012). Previous work showed that outer radial glia complete many rounds of self-renewal, exhibit enhanced transit amplification and generate basal progenitors, likely contributing to neocortical expansion in human brains (Hansen et al, 2010; Lui et al., 2011; Reillo et al., 2011). However, there are significantly fewer outer radial glia in the rodent brain and their role in neurogenesis is not fully understood.
Despite well-defined characteristics unique to neural progenitor subtypes, molecular mechanisms regulating their proliferation and differentiation remain undetermined. One genetic candidate in neural progenitor cell maintenance is NFIB, a member of the Nuclear Factor One (NFI) family of transcription factors that is essential for normal brain development. Mice with an NfiB gene mutation (NfiB−/−) have enlarged lateral ventricles, agenesis of the corpus callosum, defects in the formation of the hippocampus, basilar pons and midline structures and die at birth due to lung defects (Plachez et al., 2008; Kumbasar et al., 2009; Piper et al., 2009). NFIB regulates the migration and axonal projection of cerebellar granule neurons (Wang et al., 2007; Mason et al., 2009; Heng et al., 2012). Furthermore, NFIB is expressed in the proliferative zone and in subcerebral and corticothalamic projection neurons (Plachez et al., 2008; Mason et al., 2009; McKenna et al., 2011), but its role in regulating neural progenitor maintenance and development of these neurons remains to be elucidated.
In this study, we investigated the function of NfiB in differentiation and expansion of distinct neural progenitor subpopulations. Additionally, we characterized its role in cortical neuron development. Similar to previous studies, we found that NFIB was expressed in corticofugal neurons and the proliferative zone throughout development (Plachez et al., 2008; Mason et al., 2009; McKenna et al., 2011). Specifically, we identified high expression of NFIB in radial glia and observed defects in both neurogenesis and cortical neuron differentiation in NfiB−/− mice. Radial glia failed to generate outer radial glia and basal progenitors during late corticogenesis, corticothalamic and subcerebral axons were severely diminished, and late-born neurons ectopically expressed early-born neuronal marker, CTIP2, and displayed migration defects. Additionally, genes that regulate cell cycle progression, neuronal differentiation and axon projection were mis-regulated in NfiB−/− cortices, as revealed by gene expression analysis. Our study clearly demonstrates that NFIB is essential in regulating differentiation of radial glia, migration of cortical projection neurons and development of corticofugal axons.
Abbreviations used are listed in Table 1.
The generation of the following mouse strains were described previously: NfiB+/− (Steele-Perkins et al., 2005), Golli-τ-GFP+ (Jacobs et al., 2007), and Fezf2+/PLAP (Chen et al., 2005a). NfiB+/− mice were time-mated to generate NfiB+/+ and NfiB−/− embryos. NfiB+/− and Fezf2+/PLAP mice were mated to generate NfiB+/−; Fezf2+/PLAP mice. These mice were then time-mated with NfiB+/− mice to generate NfiB+/+;Fezf2+/PLAP, NfiB+/−;Fezf2+/PLAP and NfiB−/−;Fezf2+/PLAP embryos for PLAP staining studies of axonal projections. NfiB+/− and Golli-τ-GFP+ mice were mated to generate NfiB+/−;Golli-τ-GFP+ mice. These mice were then time-mated to NfiB+/− to generate NfiB+/+;Golli-τ-GFP+ and NfiB−/−;Golli-τ-GFP+ embryos for GFP immunostaining of axonal projections. To acquire timed-pregnant mice, male and female mice were put together overnight. The following morning, females were inspected for a vaginal plug; observation of a plug day was defined as embryonic day (E)0.5. Postnatal day (P)0 was designated as the day of birth. Genders of embryonic mice were not determined.
All embryos were genotyped by PCR. Genotyping of NfiB alleles was accomplished using two sets of primers. The wild type allele was genotyped by using p1 (GCTGAGTTGGGAGATTGTGTC) and p2 (TTCTGCTTGATTTTCGGGCTTC) with an expected PCR product of about 300bp. The PCR conditions were 94°C for 2 min, followed by 30 cycles of 94°C for 30 sec, 64°C for 1 min and 72°C for 1 min. The mutant allele was genotyped using primers, p3 (TTTCCATGTTGCCACTCGC) and p4 (AACGGCTTGCCGTTCAGCA). This set of primers detects the LacZ gene, yielding a product of about 400bp. The PCR conditions were 94°C for 2 min, followed by 30 cycles of 94°C for 30 sec, 55°C for 1 min and 72°C for 1 min. Genotyping of Fezf2 alleles was previously reported (Chen et al., 2005a). To determine whether embryos contained a copy of a Golli-τ-GFP allele, we used one set of primers, p5 (CCTACGGCGTGCCAGTGCTTCAGC) and p6 (CGGCGAGCTGCACGCTGCGTCCTC), yielding an expected product of about 300bp. PCR conditions were 94°C for 5 min, followed by 30 cycles of 94°C for 20 sec, 55°C for 30 sec and 72°C for 1 min. All experiments and animal husbandry were executed in accordance with protocols approved by the Institutional Animal Care and Use Committee at University of California, Santa Cruz and institutional and federal guidelines.
Antibodies, sources and dilutions that they were used at are listed in Table 2.
The anti-βGal antibody detects the full length native β Galactosidase, a protein encoded by the lacZ gene in the lac operon of E. coli. In the NfiB mutant allele, the lacZ gene and a nuclear localization sequence were inserted into the NfiB locus resulting in the expression of β-Galactosidase within cells expressing the NfiB mutant allele (Plachez et al. 2008). Using an NfiB+/− brain, we use the anti-βGal antibody to show that β-Galactosidase expression recapitulates that of NFIB protein (Figure 1).
The anti-BETA3 antibody specifically detects a single band at 55 kDa on western blots of rat lung and brain extracts (manufacturer’s information). We used the anti-BETA3 antibody to label cortical projection neurons of layers 2–5 of the cortex. BETA3 expression in NfiB+/+ mice is consistent with that of previously published results (Joshi et al., 2008).
The anti-BrdU antibody specifically detects bromodeoxyuridine (BrdU), a haplopyrimidine that labels proliferating cells in S phase (Struikmans et al., 1997). This was confirmed by testing the antibody’s precipitation against 5-Methyl Cysteine (5-MeC), BrdU or no antigen (as a control). There was an 8-fold increase of reactivity with BrdU over 5-MeC (manufacturer’s information). We, and others, have shown that use of this anti-BrdU antibody allows for analysis of neurogenesis (Taupin 2007; McKenna et al., 2011). In this study, we inject BrdU into mice and then use the anti-BrdU antibody to identify neural progenitors in S-phase and, based on signal intensity, neurons born on the day of BrdU injection (see Material and Methods for BrdU incorporation analysis). Note: antigen retrieval is needed for this antibody (10mM Citrate Buffer in PBS; 10 min at 95°C).
The anti-Cleaved Caspase-3 (Asp175), or anti-CC3, antibody specifically detects bands at 17 or 19 kDa on western blots of HeLa, NIH/3T3 and C6 cell extracts representing the large fragment of activated caspase-3 resulting from cleavage of this protein adjacent to Asp175 (manufacturer’s information). Immunostaining with this antibody detects apoptotic cell death in vitro and on brain tissue. Staining was evident in HT-29 cells treated with cell-death-inducing Staurosporine (Cell Signaling #9953), but not in untreated cells (manufacturer’s results). In newborn rat brain tissue, the antibody detected apoptotic cells in transient cerebral ischemic tissue versus control (Dr. Bingren Hu, University of Miami School of Medicine, Florida) and in the vomeronasal organ of Fezf2−/− mice (Eckler et al., 2011). In this study we used the anti-CC3 antibody to detect and quantify apoptotic cells within the proliferative zones in NfiB+/+ and NfiB−/− mice.
The rat monoclonal anti-CTIP2 antibody (clone 25B6) specifically detects two bands at about 95–100 kDa on a western blot of Jurkat cells immunoprecipitated with anti-Sir2 antibody and mouse brain tissue lysate. Immunohistochemical specificity was confirmed by staining on mouse cortical, spinal cord and hippocampal tissue (manufacturer’s information). CTIP2 is expressed in postmitotic neurons residing in layers 5 and 6 of the mouse cortex and is critical for development of corticospinal motor neurons (Chen et al., 2005a; McKenna et al., 2011). In this study we use the anti-CTIP2 antibody to detect layer 5 and 6 neurons throughout cortical development.
The mouse monoclonal anti-DARPP32 (clone EP27Y) antibody specifically detects a band of approximately 32 kDa on western blots of rat brain lysate (manufacturer’s information). DARPP32 expression is cytoplasmic and was reported in layer 6 neurons (Molyneaux et al., 2005); we also observed the same staining pattern (McKenna et al., 2011). In our study, we use the anti-DARPP32 antibody to identify layer 6 postmitotic neuronal cell bodies and processes in NfiB+/− and NfiB−/− brains at E18.5. We use DP32 as nomenclature for this antibody (see Table 1).
The anti-GFP antibody specificity was confirmed by both western blot and immunohistochemical analyses (manufacturer’s information) using transgenic mice expressing the green fluorescent protein (GFP) gene product. In this study, we use the anti-GFP antibody to detect the expression of GFP protein in NfiB+/+;Golli-τ-GFP+, NfiB+/−;Golli-τ-GFP+ and NfiB−/−;Golli-τ-GFP+ brains. In these mice, the Golli-τ-GFP allele directs expression of GFP in cell bodies and axons of deep layer cortical neurons (Jacobs et al., 2007).
The mouse monoclonal Ki67 antibody (clone B56) specifically detects a double band at 345 and 395 kDa on western blots of proliferating cells. Ki67 is a nuclear cell-proliferation associated antigen expressed in all active stages of the cell cycle (manufacturer’s information). Antibody specificity is validated in various ways: In flow cytometric analysis, reactivity to the Ki67 immunogen by the Ki67-B56 antibody is blocked by a second Ki67 antibody (clone MIB). Immunostaining with the antibody (clone B56) on human tonsil, successfully labels proliferating cells (manufacturer’s results). In this study, we used the Ki67 antibody to label neural progenitors and perform quit fraction analyses at different stages throughout development, as we have done previously (McKenna et al., 2011). This antibody requires antigen retrieval.
The anti-NFIA antibody specifically detects a single band at about 50 kDa on western blots of MCF-7, HeLa (manufacturer’s information) and JEG-3 cells (Plachez et al., 2008). In an electromobility shift assay, addition of the NFIA probe and antibody results in a supershift. However, in the presence of a neutralizing peptide, there is no shift, indicating antibody specificity (manufacturer’s information). Additionally, the antibody did not cross-react against other NFI proteins nor did it detect any signal in NfiA−/− mice (Plachez et al., 2008). Here, we use the anti-NFIA antibody, as others have, to label postmitotic neurons of layers 5/6 and neural progenitors in E18.5 NfiB+/− and NfiB−/− brains (Plachez et al., 2008).
The anti-NFIB antibody specifically detects a band at about 50 kDa on western blots of rat liver nuclear extract. Addition of the immunizing peptide eliminates the band, thereby confirming specificity of the antibody (manufacturer’s information). Immunostaining showed specificity, such that the antibody did not cross-react with other NFI proteins nor was any staining detected in NfiB−/− mice (Plachez et al., 2008). Similar to previous protocols (Palchez et al., 2008; McKenna et al., 2011), we use the anti-NFIB antibody to label layer 5/6 postmitotic neurons and neural progenitors and analyze its co-expression with other factors critical for cortical development.
The anti-PAX6 antibody specifically detects PAX6 protein expression by immunohistochemical analysis of mouse retina, brain tissue and human epididymis tissue (manufacturer’s information). PAX6 expression has been reported in both radial glia and outer radial glia in the developing cortex (Davis and Reed, 1996; Gomez-Lopez et al., 2011). Here, we use the anti-PAX6 antibody to identify radial glia and outer radial glia in NfiB+/+ and NfiB−/− brains throughout cortical development.
The mouse monoclonal anti-phosphorylated histone H3 (Ser10) antibody (clone 6G3), or anti-PHH3 antibody, recognizes endogenous histone H3 only when it is phosphorylated at serine 10, a hallmark of cells in M phase (manufacturer’s information). It specifically detects a band at 16.5 kDa on western blots of NIH/3T3 whole cell lysate treated with serum plus calyculin A (which induces phosphorylation) but not on untreated lysate. Flow cytometric analysis of pacilitaxel-treated THP1 cells using this antibody revealed higher DNA content in cells stained by the antibody, corresponding to cells that progressed through S phase and into M phase of the cell cycle. Specificity was also confirmed by co-immunostaining NIH-3T3 cells with anti-PHH3 and anti-tubulin antibodies, and by staining in mouse subventricular zone and third instar Drosophila larval neuroblasts. PHH3 staining is evident in prophase, metaphase and anaphase of mitosis but is absent in non-mitotic cells and in cells pre-treated with phosphatase or phosphohistone H3 (Ser10) blocking peptide (Cell Signaling, #1000) (manufacturer’s results). We use the anti-PHH3 antibody to detect and quantify mitotic cells in NfiB+/+ and NfiB−/− brains throughout cortical development.
The anti-SATB2 antibody detects a band at approximately 85 kDa on western blots of mouse brain tissue lysate and immunofluorescent staining revealed expression in human osteosarcoma 2 cell nuclei (manufacturer’s information). SATB2 is an established marker of callosal projecting neurons (Alcamo et. al, 2008; Britanova et. al, 2008; Chen et al., 2008; McKenna et al., 2011). In this study, we use the anti-SATB2 antibody in combination with deep layer markers to identify layer 2–4 neurons and define upper cortical layers in NfiB+/+ and NfiB−/− mice.
The anti-SOX2 antibody specifically detects a band at 34 kDa on western blots of human and mouse embryonic stem cell lysates. Immunofluorescence staining is evident in the mouse embryonic stem cell nuclei (Dr. Nobuaki Kikyo, Stem Cell Institute, University of Minnesota). It was reported that SOX2 is expressed in radial glia and outer radial glia in developing mouse cortex (Wang et al., 2011a). In this study, we use the anti-SOX2 antibody to label and quantify these two cell types in NfiB+/+ and NfiB−/− mice, throughout cortical development.
The anti-TBR1 antibody specifically detects a band at 74 kDa on western blots of mouse and rat hippocampal whole cell lysates (manufacturer’s results). Immunostaining with this antibody revealed nuclear staining in human cortical tissue (manufacturer’s results), pyramidal neurons of rat brains (Ruma Raha-Chowdhury), rat cortical neurons in culture (Dr. Ioana Carcea), and layer 6 neurons of embryonic and adult mouse cortical tissue (manufacturer’s results and Dr. Carlos Perez-Garcia), all which confirmed its specificity. McKenna et al., additionally showed specificity in layer 6 neurons of the developing mouse cortex. In this study, we use the anti-TBR1 antibody to identify layer 6 neurons in NfiB+/+ and NfiB−/− brains at E18.5.
The anti-TBR2 antibody detects a 72 kDa band on western blots of human mesoderm lysate, mouse brain tissue lysate and EL4 cells expressing a V5 tagged Eomesodermin. However, no band was detected when using the TBR2 antibody on human mesoderm whole cell lysate incubated with TBR2 blocking peptide (Abcam, #ab25698), or on EL4 cells expressing an empty vector or V5 tagged vector, thus validating antibody specificity (manufacturer’s results). Immunostaining results showed specificity to developing mouse cerebral cortical tissue (Guillermo Estivill-Torrus), human embryonic stem cells that differentiated into mesoderm (Ludovic Vallier, University of Cambridge), and in adult mouse hippocampal tissue (Hongjun Song Lab). As others have done (Arnold et al., 2008), we use the anti-TBR2 antibody to detect and analyze basal progenitors in NfiB+/+ and NfiB−/− brains throughout cortical development.
The mouse Tuj1 monoclonal antibody has been used to detect the neuronal class III β-tubulin protein (officially known as TUBB3) in axons and dendrites of postmitotic neurons (Casanovas et al., 2008). It was produced from the Tuj1 hybridoma mouse clone developed from immunizing mice with rat brain microtubules. The Tuj1 antibody does not react with β-tubulin in glial cells and staining on rat hippocampal tissue revealed immunohistochemical specificity (manufacturer’s results). In this study we use laminar position of Tuj1 positive cells to determine location of postmitotic neurons within the cortex on NfiB+/+ and NfIB−/− mice.
To obtain embryonic tissue, pregnant mice were sacrificed by cervical dislocation and embryos were removed. E12.5 heads were drop-fixed in 4% paraformaldehyde in 1x phosphate buffered saline (PBS; pH 7.4) at 4°C. E13.5–E16.5 brains were dissected from skull then drop-fixed in 4% paraformaldehyde (PFA) in 1x PBS at 4°C. Older embryos were transcardially perfused with PBS followed by 4% PFA in 1x PBS before drop-fixation as above. All embryonic brains were cryoprotected in 30% sucrose, frozen in Optimal Cutting Temperature Compound (Tissue-Tek) and cut using a cryostat or sliding microtome. Coronal or sagittal sections were cut at 20μm via cryostat (Figures 1–4, ,66–11) or 50μm (Figure 5) on sliding microtome.
For immunofluorescence, sections were blocked in 5% horse serum and 0.1%Triton X-100 in 1x PBS for 1 hour at room temperature. Sections were incubated with primary antibodies at appropriate dilutions (Table 2) overnight at 4°C. The following day, slides were washed with 0.1%Triton X-100 in 1x PBS three times for five minutes, incubated with Alexa Fluor-conjugated secondary antibodies for one hour at room temperature, then washed with 1x PBS for three times at 5 minutes before mounting with Fluoromount (Southern Biotech). Secondary antibodies used in this study were Donkey Anti-Mouse IgG Alexa Fluor 547 or 647, Donkey Anti-Rabbit IgG Alexa Fluor Cy3, Cy5 or 488, Donkey Anti-Sheep IgG Alexa Fluor 488 or 647, Donkey Anti-Chicken IgG Alexa Fluor Cy3 or 488, Donkey Anti-Rat IgG Alexa Fluor 488 or 647 and Donkey Anti-Goat IgG Alexa Fluor Cy3, 488 or 647 (all at 1:1000; Jackson ImmunoResearch Laboratories, West Grove, PA).
Nuclei were visualized using DAPI (4′,6-Diamidino-2 Phenylindole; at 1:10,000; Molecular Probes) (Figure 9) or DRAQ5 (1,5-bis((2-(dimethylamino) ethyl)amino)-4, 8- dihydroxyanthracene-9, 10-dione; at 1:5000; Cell Signaling Technology) (Figure 6). At least 3 brains/genotype were analyzed for all experiments; all tissue sections were collected onto Superfrost slides (Fisher Scientific).
Human placental alkaline phosphotase (PLAP) staining (Figure 5A–H) was performed as previously described (Chen et al., 2005a). Brain tissue was collected and sectioned at 50μm using a sliding microtome, washed in 1x PBS and post-fixed in 4% PFA in 1x PBS at 4°C for 1 hour. Sections were again washed with 1x PBS, then incubated at 70°C for 1 hour in 1x PBS. Sections were incubated with NBT/BCIP (1 mg/ml nitroblue tetrazolium/0.1 mg/ml 5-bromo-4-chloro-3-indolyl phosphate in 100 mM Tris·Hcl, pH 9.5/100 mM NaCl) at 37°C for 3 hours, then washed with 0.3% Triton X-100 in 1x PBS and mounted in Fluoromount.
Non-radioactive in situ hybridization was essentially performed as previously described (Schaeren-Wiemers and Gerfin-Moser, 1993). Slides with fixed brain sections were post-fixed in 4% PFA for 10 minutes, treated with proteinase K for 10 minutes, fixed again in 4%PFA for 10 minutes and acetylated for 10 minutes, all at room temperature, with DEPC PBS washes in between each step. Slides were prehybridized for 2 hours at 65°C, hybridized with Digoxigenin (DIG) labeled cRNA and sealed with coverslips (HybriSlip, Grace Bio-Labs). The cDNA templates for cRNA probes for Notch1, Hes1 and Hes5 genes were generous gifts from Drs. Toshiyuki Ohtsuka and Ryoichiro Kageyama from Kyoto University, Kyoto, Japan. Nucleotides 4270–5611 of mouse Notch1 cDNA was used to prepare the Notch1 probe. Full-length cDNA sequences for Hes1 and Hes5 were used to prepare Hes1 and Hes5 probes, respectively. Slides were then placed in a humidity chamber overnight at 65°C. The following day, slides were washed with SSC, incubated with RNase A at 37°C for 30 minutes and washed with the following: SSC, 0.1MTris (pH 7.4)/0.15M NaCl (Buffer 1) for 5 minutes, Buffer 1 containing 1% heat inactivated goat serum for one hour at room temperature. Slides were incubated with an anti-DIG antibody in a humidity chamber overnight at 4°C. On the third day, slides were washed with Buffer 1 and equilibrated with 0.1M Tris (pH9.5)/0.1M NaCl/50mM MgCl2 (Buffer 2) for 5 minutes at room temperature. To visualize probe-tissue interaction, slides were incubated for 6 hours with NBT/BCIP substrate in Buffer 2. Slides were washed with PBS, then with PBS containing 0.3% Triton X-100. Lastly, slides were fixed with 4% PFA and mounted with Fluormount-G.
Images for quantitative analyses were captured on a Leica SP5 confocal microscope. The detector gain was set such that <1% of pixels were saturated. Additional fluorescent and bright field images were captured on an Olympus BX51 microscope and Q-imaging Retiga camera. Adjustments of contrast and brightness and image cropping for presentation were done using Adobe Photoshop to the same degree on entire images of NfiB+/+ and NfiB−/− matched cortical sections.
For all measurements and cell-counting, statistical analyses were performed by an unpaired t-test; significance was defined as *p ≤ 0.05, **p ≤ 0.005, ***p ≤0.0005; error bars represent standard error of means (SEM). Specific numbers of mice used in each quantitative experiment can be found in figure legends. Briefly, at least nine matched sections/age from anterior, middle and posterior regions from at least three mice/genotype were used in each quantitative comparison. Matched sections were defined as brain sections from approximately the same anterior-posterior axial position in NfiB+/+ and NfiB−/− littermates. Unless otherwise noted, data presented in figures are from sections midway along the anterior-posterior axis of the brain.
To measure independent thicknesses of upper layers (UL-S) or deep layers (DL-C), coronal sections of NfiB+/+ and NfiB−/− mice were immunostained as described above with antibodies against SATB2 and CTIP2. DL-C was defined as the region along the dorsal-ventral (D–V) axis containing CTIP2-expressing cells; UL-S was defined as the region containing SATB2-expressing cells dorsal to DL-C, along the D–V axis (schematic Figure 4Q). For each cortical image, three measurements were taken along 300μm of the mediolateral axis and averaged. To measure the proliferative zone, the region containing the VZ, SVZ and oSVZ (VZ + o/SVZ), NfiB+/+ and NfiB−/− mice were immunostained as described above with antibodies against Ki67 and TBR2. The VZ + o/SVZ was defined as the region along the D–V axis of cells expressing Ki67 and/or TBR2 (schematic Figure 6C).
Cell counting was performed on individual Z-slices from confocal microscopic images. For each brain analyzed, the cortex was divided into anterior, middle and posterior regions. Cells on 300μm wide sections of the mediolateral axis of cortical sections expressing indicated markers were counted.
For analysis of cells in S-phase, BrdU (Bromodeoxyuridine at 100μg/g body weight; US biologicals, Swampscott, MA) was intraperitoneally (i.p.) injected into timed pregnant mice at E18.5; embryos were collected 2 hours post-injection and processed for BrdU immunostaining and quantification. To analyze the distribution of BrdU-labeled cells, cortices were divided into ten bins of equal thickness and BrdU-labeled cells in each bin were quantified (schematic Figure 6N). For birthdating analysis, BrdU (100 μg/g body weight) was i.p. injected into timed pregnant mice at E15.5; embryos were collected at E18.5 and processed for BrdU immunostaining and quantification. Only strong BrdU-labeled cells, cells with ≥75% BrdU nuclei saturation, were counted. The distribution of strong BrdU-labeled cells was analyzed by binning, as described above.
BrdU (100 μg/g body weight) was i.p. injected into timed pregnant mice at E11.5, E12.5, E13.5, E14.5 or E15.5. Embryos were collected 24 hours later and processed for BrdU and Ki67 immunostaining, as described above. The numbers of all cells stained for BrdU (all BrdU+) and cells stained for BrdU but not for Ki67 (BrdU+Ki67−) were counted. The quit fraction (QF%) was calculated as the percentage of BrdU+Ki67− among all BrdU+ cells.
E15.5 NfiB+/+ and NfiB−/− cortices were dissected (n=3/genotype), total RNAs were isolated from each cortex (Qiagen RNAeasy kit) and used to prepare RNA-seq libraries (Illumina RNA Truseq Library Prep protocol). Libraries were paired-end (50 nucleotides per end) sequenced on Illumina Hiseq2000 platform. Between 77 and 88 million read pairs were obtained for each sample. After filtering out read pairs that matched elements in the mouse RepeatMasker library (Smit et al, 1996–2010) (less than 2% per sample), reads were mapped to the UCSC (mm9) mouse assembly with TopHat (Trapnell et al., 2009), using Bowtie (Langmead et al., 2009) as the underlying aligner. Only uniquely-mapping, fully-paired reads were kept and potential PCR duplicates were removed using SamTools (Li et al., 2009). Between 49 and 57 million mapped read pairs remained per sample. From these mappings, the count of reads was determined for the canonical transcript of each gene in the set of UCSC Known Genes (Hsu et al., 2006) for the UCSC mm9 mouse assembly. These counts were analyzed for differential expression of genes between NfiB+/+ and NfiB−/− samples using DESeq (Anders and Hubers, 2010), which models the expression variation within replicates as a function of total gene expression, to determine the significance of expression differences between conditions. A threshold for significance was set at the DESeq adjusted p-value of 0.1.
To understand the biological meaning of the output gene list, we used the Database for Annotation, Visualization and Integrated Discovery (DAVID) (Huang et al., 2009a, 2009b). The list of differentially expressed genes was uploaded into DAVID then analyzed by functional annotation clusters. A summary of gene clusters exhibiting the highest enrichment scores are in Table 3 and and44.
NFIB is expressed in both deep layer neurons and neural progenitor cells (Mason et al., 2009; McKenna et al., 2011), but its expression profile in individual subtypes of neural progenitors is not known. To determine which progenitor cells expressed NFIB during cortical development, we used specific antibodies on E12.5–E18.5 brain sections to co-label cells expressing NFIB against known neural progenitor and layer-specific neuronal markers. Taking advantage of the LacZ insertion into the NfiB locus of NfiB+/− mice (Steele-Perkins et al., 2005), in addition to NfiB+/+ mice, we used NfiB+/− mice to maximize detection of NFIB and cell-type specific marker co-expression. To ensure βGAL was an appropriate marker of NFIB-expressing cells, we performed immunohistochemistry with antibodies against βGAL and NFIB and found that βGAL expression recapitulates that of NFIB in NfiB+/− brains (Figure 1).
Immunohistochemical analysis of E12.5 and E13.5 brains revealed coexpression of NFIB and Ki67, confirming NFIB expression in neural progenitors (Figure 2A, E, P and T) (Plachez et al., 2008; Mason et al., 2009). Starting at E12.5, two major progenitor cell types were present: radial glia and basal progenitors. Radial glia in the ventricular zone (VZ) were SOX2+PAX6+ (Figure 2F–H and J). Basal progenitors in the SVZ were TBR2+SOX2low (Figure 2K–M and O) (Arnold et al., 2008). βGAL was expressed at high levels in all SOX2+PAX6+ radial glia (Figure 2F–J), but expression was generally low or completely absent in TBR2+ basal progenitors, despite the presence of a few TBR2+βGAL+ cells (Figure 2K–O). It is likely that these TBR2+βGAL+ cells were in transition from radial glia to basal progenitors, or from basal progenitors to neurons. Similarly, very few TBR2+ basal progenitors expressed βGAL at E15.5 or E18.5 (Figure 2EE, 2II and and3E3E).
Beginning at E15.5, the presence of outer radial glia, identified by SOX2 expression in the outer subventricular zone (oSVZ), became apparent (Figure 2W, and FF). Both SOX2+ and PAX6+ radial glia expressed NFIB (or βGAL) at E15.5 and E18.5, however, only low-level expression was apparent in outer radial glia (Figure 2U, Y, Z, DD, EE, FF, II and 3A, C). Together, these data demonstrate that during corticogenesis, radial glia express NFIB, but expression is almost entirely excluded from basal progenitors and outer radial glia.
We next characterized NFIB expression in the developing cortical plate. Immunostaining analyses revealed co-expression of NFIB and known markers of corticofugal projection neurons. Subcerebral marker, CTIP2, and NFIB were co-expressed in deep layers of the cortical plate at E12.5 and E13.5 (Figure 2A, E, P and T), the time when these neurons are generated (Leone et al., 2008; McKenna et al., 2011). CTIP2+ neurons also expressed NFIB (or βGAL) at E15.5 (Figure 2U, Y, Z, DD).
To further characterize the molecular identity of NFIB-expressing cells, we analyzed co-expression with NFIA, DARPP32 and SATB2 (Figure 3G, I, K). NFIA, another transcription factor in the Nfi gene family, is expressed in proliferative zones and deep layer cortical neurons (Plachez et al., 2008; Mason et al., 2009), but whether NFIA and NFIB are expressed in the same cells is unknown. We observed βGAL and NFIA co-expression in the VZ and layer 6 (Figure 3G). βGAL was also co-expressed with layer 6 corticothalamic marker, DARPP32 (Molyneaux et al., 2005) (Figure 3I). Conversely, there was little co-expression with SATB2, a known marker of callosal projection neurons (Alcamo et. al, 2008; Britanova et. al, 2008; Chen et al., 2008) (Figure 3K). However, of the few SATB2+βGAL+ neurons, most were located in layer 6 (Figure 3K). Additionally, NFIB and βGAL expression was noticeably stronger in postmitotic neurons compared to that in neural progenitors (Figure 2). Collectively, analysis of neuronal marker expression indicates that NFIB is expressed in corticofugal neurons throughout corticogenesis and may also be expressed in a small subset of deep layer callosal projection neurons. Overall, the expression profile of NFIB throughout cortical development suggests it may regulate neurogenesis and differentiation of radial glia and corticofugal neurons.
To determine the function of NFIB in cortical development, we investigated what effects the loss of NFIB had on overall cortical organization in NfiB−/− mice. We first determined the molecular identity of βGAL+ cells in E18.5 NfiB−/− mice by immunostaining with antibodies against βGAL and markers of neural progenitors or postmitotic neurons. Similar to NfIB+/+ and NFiB+/− brains, we found that in NfiB−/− mice, βGAL was expressed in SOX2+ and PAX6+ radial glia, but expression was weak or absent in SOX2+ outer radial glia and TBR2+ basal progenitors (Figure 3A–F). We observed βGAL and NFIA co-expression in layer 6 and the proliferative zone (Figure 3H), and βGAL expression in DARPP32+ corticothalamic projection neurons (Figure 3J). The majority of SATB2+ callosal projection neurons did not express βGAL (Figure 3L) and those that did were predominantly in layer 6, similar to NfiB+/− cortices (Figure 3K–L). Together, these results indicated that the molecular identity of βGAL+ NFIB-mutant cells was similar to NFB-expressing cells in NfiB+/+ and NfiB+/− cortices.
While the overall expression of cell-type specific markers was comparable in control (NfiB+/+ or NfiB+/−) and NfiB−/− cortices, thickness of upper layers 2–4 appeared smaller in NfiB−/− cortices (Figure 3). To further investigate this, we compared laminar organization of control (NfiB+/+) and NfiB−/− cortices. CTIP2 expression was higher in layer 5 than in layer 6 in both control (Figure 4A, C, D) and mutant (Figure 4E, G, H) littermates. Co-immunostaining with CTIP2 and SATB2 clearly revealed a reduction in upper layer (2–4) thickness as well as a modest expansion of deep layers 5–6 (Figures 4A–H). Immunostaining with additional antibodies against layer specific markers, BETA3 and TBR1, revealed a similar outcome. In control mice, BETA3 is strongly expressed in layer 5 and to a lesser extent in upper layers (Figure 4I, K, L). However, in NfiB−/− cortices, we observed an increase in BETA3+ cell density within layer 5 (Figure 4M, O, P), ectopic BETA3+ neurons in layer 6 (arrows, Figure 4P) and a reduction in upper layer thickness (Figures 4I, K, L, M, O, and P). In NfiB+/+ mice, TBR1 was predominantly expressed in layer 6 corticothalamic neurons, while some TBR1+ neurons were also observed in upper layers (Figure 4J–L). However, NfiB−/− mice displayed an absence of TBR1+ cells in upper layers (Figure 4N–P).
To quantitatively assess the loss of NfiB on cortical lamination, we measured and compared thicknesses of regions containing upper layer SATB2+ neurons (UL-S) or deep layer CTIP2+ neurons (DL-C) in NfiB+/+ and NfiB−/− brains. Subsequently, this enabled comparison of overall cortical plate thickness. For measurements, we performed immunostaining with antibodies against CTIP2 and SATB2 at E16.5 and E18.5. We defined (UL-S), layers 2–4, as the cortical region containing SATB2+ but not CTIP2+ neurons. DL-C, layers 5 and 6, were defined as the region containing CTIP2+ cells, regardless of SATB2 expression. Cortical plate thickness was defined as the sum of upper and deep layer thicknesses (schematic in Figure 4Q). At E16.5, the overall thickness of the cortical plate was significantly reduced in NfiB−/− mice (171.8 ± 10.9μm) compared to NfiB+/+ (215.5 ± 15.1μm, p=0.0313), predominantly due to a reduction in UL-S thickness. In NfiB+/+ mice, UL-S (41.8 ± 3.1μm) comprised approximately 19.5% of the cortical plate but in NfiB−/− littermates, UL-S (13.5 ± 2.7μm, p<0.0001) was only 7.7% of the cortical plate.
Likewise, at E18.5, UL-S was significantly thinner in NfiB−/− brains compared to NfiB+/+ mice (NfiB+/+ 90.6 ± 4.1μm vs. NfiB−/− 57.2 ± 1.6μm, p=<0.0001). Regardless of an overall reduction in cortex thickness (from apical to basal surfaces) (Figure 3, ,4),4), the thickness of the cortical plate was comparable between mutant and control littermates, a result explained by an expansion of the DL-C (Figure R). Collectively, immunostaining and quantitative analysis reveal that in NfiB−/− mice, the ratio of deep to upper layers in the cortical plate shifted more toward a deep layer profile. Observed changes in cortical layer thickness in NfiB−/− mice and NFIB expression in radial glia and corticofugal neurons, together suggest that NFIB is required for proper neurogenesis and neuronal differentiation.
Considering that NFIB is robustly expressed in corticofugal neurons, we asked whether NFIB is required for proper differentiation of corticofugal neurons, specifically, for axon development. To answer this, we used transgenic methods to label and analyze axonal projections. Fezf2 (also known as Fezl, Zfp312 and Znf312) is expressed in both subcerebral and corticothalamic projection neurons (Chen et al., 2005a; Chen et al., 2008; Leone et al., 2008; McKenna et al., 2011). In the Fezf2 mouse, an ires-PLAP (internal ribosomal entry site-Placental Alkaline Phosphatase) was inserted into the Fezf2 open reading frame (Chen et al., 2005a), thus cell bodies and axons of subcerebral and corticothalamic projection neurons are made visible by PLAP staining. Similarly, the Golli-τ-GFP+ allele directs expression of GFP in corticofugal projection neurons (Jacobs et al., 2007), allowing direct observation of GFP-labeled axons.
We analyzed corticofugal axons in NfiB+/+;Fezf2PLAP/+, NfiB+/−;Fezf2PLAP/+ and NfiB−/−;Fezf2PLAP/+ mice at E18.5 by PLAP staining. As expected, in NfiB+/+; Fezf2PLAP/+ mice, bundles of PLAP+ axons were observed projecting through the internal capsule (Figure 5B, asterisk), toward the thalamus (Figure 5B, arrowhead) and projecting through the cerebral peduncle toward the spinal cord (Figure 5A, C, D, arrows,). There was a modest reduction of PLAP+ corticofugal axons in NfiB+/−;Fezf2PLAP/+ mice (data not shown). Strikingly, in NfiB−/−; Fezf2PLAP/+ mice, very few PLAP+ axons projected through the internal capsule (Figure 5F, asterisk), and axons projecting toward the thalamus (Figure 5F, arrowhead) or through the cerebral peduncle (Figure 5E, G, H, arrows) were not detected. In addition, no ectopic axons were observed. These results suggest that in the absence of NfiB, corticofugal neurons failed to project axons out of the cerebral cortex.
We also analyzed GFP-labeled corticofugal axons in NfiB+/+; Golli-τ-GFP+, NfiB+/−; Golli-τ-GFP+ and NfiB Golli-τ-GFP+ mice at E18.5. In NfiB+/+; Golli-τ-GFP+ animals, GFP+ axons projected through the internal capsule (Figure 5K, L, asterisks) toward the thalamus (Figure 5K–M, open arrows), pons (Figure 5I and J, white arrows) and cerebral peduncle (Figure 5N, white arrowheads). There was an intermediate loss of GFP+ corticofugal axons in NfiB+/−; Golli-τ-GFP+ mice (data not shown). However, in NfiB−/−; Golli-τ-GFP+ mice, very few GFP+ axons were detected in the internal capsule (Figure 5Q, R, asterisks) and thalamus (Figure 5Q–S, open arrows). A drastic reduction of GFP+ axons projecting through the pons (Figure 5O and P, white arrows) and cerebral peduncle (Figure 5T, white arrowheads) was also observed. Collectively, these results showed that in NfiB−/− mice, corticofugal neurons failed to project axons to expected targets, indicating that NFIB is required for proper differentiation of corticofugal neurons.
To determine the function of NFIB in neural progenitors, we first compared proliferative zone organization of NfiB+/+ and NfiB−/− mice from E12.5 to E18.5 by analyzing the expression of TBR2 and Ki67. Together, these markers labeled the VZ, SVZ and oSVZ, thus defining the boundaries of the proliferative zone (VZ + o/SVZ) (Figure 6A–C). At E12.5 and E14.5, the thickness of the proliferative zone, as both an absolute measurement and as a percentage of overall cortical thickness, was similar in control and mutant mice (Figure 6D). At E16.5, NfiB−/− brains displayed a significant decrease in proliferative zone thickness compared to NfiB+/+ brains (Figure 6D). As a percentage of overall cortical thickness, the proliferative zone in NfiB+/+ animals made up 47.6% ± 2.2% of the cortex, while in NfiB−/−, it was only 39.0% ± 5.1%, p=0.0004. Most notable was the striking absence of TBR2+ cells in the oSVZ and compaction of these cells near the ventricular surface in NfiB−/− cortices (Figure 6A–B). A significant decrease was also observed at E18.5 (NfiB+/+ 41.6% ± 3.4% versus NfiB−/− 32.0% ± 3.6%, p<0.0001) (Figure 6D), indicating that proliferative zone thickness was specifically affected at later stages of cortical development.
To further examine defects in the proliferative zone, we analyzed neural progenitors during different phases of the cell cycle. First, we compared the number and cortical location of cells in E12.5–E18.5 NfiB+/+ and NfiB−/− mice that expressed phosphorylated histone H3 (PHH3), a marker for cells in M phase (Figure 6E–I). At E12.5 and E14.5, the number of PHH3+ cells and their cortical position were similar in NfiB+/+ and NfiB−/− brains (Figure 6I). However, a significant loss of total number of PHH3+ cells (NfiB+/+ 77.1 ± 6.6 cells/mm vs. NfiB−/− 41.1 ± 3.0 cells/mm, p=<0.0001) was observed in E16.5 mutant brains (Figure 6E–I). The loss was due to significantly fewer PHH3+ cells in both the o/SVZ (NfiB+/+ 37.1 ± 4.0 cells/mm vs. NfiB−/− 16.1 ± 2.8 cells/mm, p=0.0006) and VZ (NfiB+/+ 40.2 ± 3.9 cells/mm vs. NfiB−/− 25.1 ± 2.3 cells/mm, p=0.0044) (Figure 6E–I). Interestingly, despite a significant loss of total number of PHH3+ cells in E18.5 NfiB−/− brains (NfiB+/+ 24.1 ± 2.5 cells/mm versus NfiB−/− 16.4 ± 1.9 cells/mm, p=0.0263), a comparable number of PHH3+ cells were observed in the VZ, indicating that the loss was specifically due to significantly fewer PHH3+ cells within the o/SVZ (Figure 6I). These results suggest a loss of actively dividing neural progenitors in NfiB−/− mice during late corticogenesis, an effect more striking in the SVZ and oSVZ.
Next, we compared number and cortical distribution of neural progenitors in S phase (Figure 6J–O). BrdU was administered to pregnant mice carrying E18.5 embryos; brains were collected 2 hours later and processed for BrdU immunohistochemistry. BrdU+ cells were quantified by cortical binning (10 bins). When individual cortices were divided into bins, there was a trend such that the marginal zone was in bin 1, the cortical plate was in bins 2–6, the SVZ and oSVZ (o/SVZ) was in bins 7–9 and the VZ was predominantly in bin 10 (Figure 6N, schematic). We determined the distribution of S phase progenitors by quantification of BrdU+ cells in each of the following regions: the cortical plate (including the marginal zone), o/SVZ, and VZ. There was a significant loss of total number of BrdU+ cells in NfiB−/− mice (54.6 ± 2.2 cells/100 μm, p=0.0316) compared to littermate controls (70.5 ± 6.5 cells/100 μm) (Figure 6J–O). While the number of BrdU+ cells in bin 10 was similar in NfiB+/+ and NfiB−/− animals, loss of total number was due to significantly fewer BrdU+ cells in bins 7–9 (NfiB+/+ 33.3 ± 2.8 cells/100 μm vs. NfiB−/− 18.5 ± 1.7 cells/100 μm, p<0.0001) and 1–6 (NfiB+/+ 8.9 ± 1.2 cells/100 μm vs. NfiB−/− 2.0 ± 0.4 cells/100 μm, p<0.0001) (Figure 6O), clearly indicating a severe loss of S phase progenitors within the SVZ and oSVZ in NfiB−/− mice. Taken together with the observed reduction in proliferative zone thickness in NfiB−/− mice, these results suggest that NFIB is required for the generation or survival of neural progenitors located specifically within the SVZ and oSVZ, but not VZ.
To determine which subclasses of neural progenitors were affected by loss of NfiB, we immunostained E12.5–E18.5 NfiB+/+ and NfiB−/− brains with PAX6, SOX2 and TBR2 antibodies. Tuj1 (Figure 7A–L) or CTIP2 (Figure 7M–V) antibodies were used to label cortical neurons. PAX6 and Tuj1 expression was similar in NfiB+/+ and NfiB−/− mice at E12.5 (Figure 7A–F). At E14.5, expression of PAX6 was unaffected in NfiB−/− mice compared to controls, as was the expression of SOX2 and TBR2 (Figure 7G, I, J, L and 8A, B, D, E, F, H). However, Tuj1 staining revealed fewer axons at E14.5, a time when most cortical neurons are of corticofugal identity (Figure 7H, I, K, L and Figure 8C, D, G, H). These results indicate that axon defects in corticofugal neurons due to a loss of NFIB begin around mid-corticogenesis (E14.5), yet all subclasses of neural progenitors were unaffected up to this timepoint.
Defects in the expression of markers for neural progenitor subtypes in NfiB−/− mice were observed during late (E16.5–E18.5) corticogenesis. In E16.5 NfiB−/− cortices, despite a similar total number, there was significantly more PAX6+ cells in the VZ (NfiB−/− 153.9 ± 9.1 cells/100μm vs. NfiB+/+ 128.9 ± 7.6 cells/100μm, p=0.0095), yet markedly fewer in the o/SVZ (NfiB−/− 12.9 ± 1.0 cells/100μm vs. NfiB+/+ 40.4 ± 3.1 cells/100μm, p<0.0001) (Figure 7M, O, Q, R, T, V and EE), indicating more radial glia but a loss of outer radial glia. TBR2+ cells were present in both the SVZ and oSVZ in NfiB+/+ mice (Figure 8I and J). However, the total number of TBR2+ basal progenitors was significantly reduced in NfiB−/− littermates (NfiB+/+ 156.2 ± 10.0 cells/100μm vs. NfiB−/− 99.9 ± 3.6 cells/100μm, p<0.0001), and although there were significantly fewer TBR2+ basal progenitors in the deeper SVZ, more dramatic was the loss in the oSVZ (NfiB+/+ 47.2 ± 4.0 cells/100 μm vs. NfiB−/− 13.0 ± 1.2 cells/100 μm, p<0.0001) (Figure 8I–L and Q).
At E18.5, NfiB−/− mice displayed a loss of PAX6+/SOX2+ outer radial glia and TBR2+ basal progenitors (Figure 7W–EE and 8M–Q). Although there was approximately the same total number of PAX6+ cells in NfiB+/+ and NfiB−/− cortices, there were significantly fewer PAX6+ cells in the NfiB−/− o/SVZ (Figure 7W, Y, Z, AA, CC, DD and EE). Similar to analysis at E16.5, we observed a significant decrease in the number of TBR2+ basal progenitors in both the SVZ and oSVZ (Figure 8M–Q), Taken together, these results reveal that NfiB−/− mice have more radial glia but display a striking loss of outer radial glia and basal progenitors at E16.5 and E18.5. This suggests that NFIB may regulate the differentiation of radial glia into outer radial glia and basal progenitors during late corticogenesis.
To explain the loss of these two neural progenitor subtypes, we asked whether the absence of NFIB resulted in an increase in apoptosis or a decrease of their generation. We assayed the number of apoptotic cells in NfiB+/+ and NfiB−/− cortices by immunostaining with an antibody against activated caspase 3 (CC3) at E15.5, one day prior to the observed loss of neural progenitors (Figure 9). No statistical difference in CC3+ cell number was observed (Figure 9J), indicating that loss of outer radial glia and basal progenitors in NfiB−/− cortices was not due to cell death, but instead, due to the lack of their generation. Furthermore, compared to the SVZ, the more severe loss of TBR2+, SOX2+ and PAX6+ cells in the oSVZ strongly suggests that the fewer number of outer radial glia contributes to the reduction of basal progenitor generation in NfiB−/− mice.
To further investigate the function of NFIB in regulating neural progenitor proliferation and differentiation, we compared the quit fractions in NfiB+/+ and NfiB−/− cortices at several timepoints throughout neurogenesis. BrdU was administered to pregnant mice at E11.5, E12.5, E13.5, E14.5 and E15.5; embryonic brains were collected 24 hours later and immunostained with Ki67 and BrdU antibodies. The quit fraction was calculated as the percentage of BrdU+Ki67− cells among all BrdU+ cells, representing the fraction of cells that exited the cell cycle on a given day. Compared to NfiB+/+ littermates, the quit fraction was unaffected in NfiB−/− cortices at E11.5, E12.5, E13.5 and E14.5. However, a significant increase was observed at E15.5 (NfiB+/+ 64.5% ± 1.7% versus NfiB−/− 76.8% ± 1.8%, p=0.0045) (Figure 10A).
The increased quit fraction in E15.5 NfiB−/− brains suggested more neurons were generated at this time. To test this, we performed birthdating analysis by injecting BrdU into pregnant mice carrying E15.5 embryos, collecting brains at E18.5 and immunostaining with antibodies against BrdU, CTIP2, and SATB2 (Figure 10B–I). The number of E15.5 BrdU+ cells (see Material and Methods) was compared between NfiB+/+ and NfiB−/− cortices. Consistent with results from quit fraction analyses (Figure 10A), the number of E15.5 BrdU+ cells was significantly higher in NfiB−/− brains (70.0 ± 3.1 cells/100 μm, p=0.0001) than controls (49.4 ± 2.8 cells/100 μm) (Figure 10B, E, F, I, J).
To address whether NFIB plays a role in migration of neurons generated at E15.5, we compared the cortical distribution of E15.5 BrdU+ cells by binning (refer to schematic, Figure 6N). E15.5 BrdU+ cells per bin were quantified as a percentage of total E15.5 BrdU+ cells. In NfiB+/+ cortices, almost all E15.5 BrdU+ cells were located in the cortical plate, with many of them in upper layers (Figure 10B, E, K). Conversely, in NfiB−/− mice, a significantly higher percentage of E15.5 BrdU+ cells were positioned below the cortical plate while a significantly lower percentage were in upper cortical layers (Figure 10F, I, K). This indicates a delay or improper termination of neuronal migration, which may lead to the observed thinning of upper cortical layers (Figure 3, ,44 and 10E, I).
Additionally, we investigated the identity of E15.5 BrdU+ cells by their expression of CTIP2 and SATB2. There was no significant difference between the percentage of E15.5 BrdU+ cells expressing SATB2 in NfiB+/+ and NfiB−/− cortices. However, the percentage of those expressing CTIP2 was significantly higher in NfiB−/− brains (NfiB+/+ 7.9% ± 1.4% versus NfiB−/− 21.1% ± 1.9%, p<0.0001) (Figure 10E, I, L). Similar to BrdU staining, many more SATB2+ and CTIP2+ cells were located below the cortical plate in NfiB−/− cortices (Figure 10B–I), suggesting a migration defect. This possibly explains our observations of ectopic CTIP2+ cells located below the cortical plate and morphological abnormalities of CTIP2+ cells within the cortical plate of E16.5 NfiB−/− cortices (Figure 7N–P and S–V). In NfiB+/+ mice, CTIP2+ neurons in the cortical plate were round in shape, a morphology typical of neurons that have completed migration (Figure 7N–P). In NfiB−/− mice, however, CTIP2+ neurons displayed an elongated morphology that is associated with neurons that are currently migrating (Figure 7S–U). Collectively, these results suggest that at E15.5, more cells exited the cell cycle and differentiated into neurons in NfiB−/− mice. They ectopically expressed early-born neuron marker, CTIP2, and their migration was defective. Thus, NFIB is required for proper neurogenesis and migration of late-born neurons.
To investigate the function of NFIB in regulating cortical neurogenesis and neuronal differentiation, we performed gene expression analysis of E15.5 NfiB+/+ and NfiB−/− cortices by RNA-seq (see Material and Methods). We used the DESeq program to compare gene expression of NfiB−/− samples against controls (Anders et al., 2010) and found that expression levels of 1844 genes in E15.5 NfiB−/− mice were significantly changed. Among them, 929 were increased by at least 1.17-fold and 915 were decreased by at least 0.85-fold.
We performed gene annotation analysis using the online bioinformatics tool, DAVID (Database for Annotation, Visualization and Intergrated Discovery). Misregulated genes in NfiB−/− cortices were clustered with regards to their function; clusters were organized by enrichment score (Huang et al., 2009a; Huang et al., 2009b). Cluster 1 had the highest enrichment score and included 168 genes known to regulate cell cycle properties, such as M phase progression and cell division. There were 123 genes in Cluster 2, including transcription factors correlated with neuron differentiation and axon guidance molecules. Gene members of the Notch, Fgf, Shh and Wnt signaling pathways, all of which are important for cortical development (Gaiano et al., 2000; Cornell and Eisen, 2005; Louvi and Artavanis-Tsakonas, 2006; Dong et al., 2012; Tiberi et al., 2012), were also listed in Cluster 2. Cluster 3 included 100 genes, those of which regulate cell motion, neuron migration and cell localization (Table 3 and and4).4). Mis-regulation of these specific gene clusters in E15.5 NfiB−/− cortices provides further evidence that NFIB is essential for both proliferation and differentiation of cortical progenitor cells, and axonal projection of cortical neurons.
Notch signaling is essential for regulating radial glia and outer radial glia maintenance and differentiation (Cornell and Eisen, 2005; Yoon and Gaiano, 2005; Louvi and Artavanis-Tsakonas, 2006; Guillemot, 2007; Shitamukai et al., 2011; Saito, 2012). In E15.5 NfiB−/− cortices, many members of the Notch signaling pathway were mis-regulated, including Dll1, Notch3, Hes1, Hes5, Hey1, and Hey2 (Table 4 and data not shown). We performed in situ hybridization on E18.5 NfiB+/+ and NfiB−/− brains for several of these genes (Figure 11) and found that mis-regulation of Notch1, Hes1 and Hes5 persisted throughout the remainder of cortical development. Compared to controls, expression of Notch1, Hes1 and Hes5 mRNA was significantly upregulated in the VZ (Figure 11), suggesting an increase in Notch signaling, specifically in radial glia, resulting in observed defects in radial glia differentiation in NfiB−/− mice.
By carefully analyzing NFIB expression at embryonic stages and investigating the defects in NfiB−/− brains, our study revealed both novel and broad functions of NFIB in regulating the generation of cortical projection neurons, including radial glial differentiation, neuronal migration and axon projection. Multiple steps are involved in generating cortical projection neurons. First, radial glia expand the progenitor pool and generate outer radial glia, basal progenitors and neurons. Newly generated neurons migrate to their final laminar destinations, project axons and dendrites to appropriate targets to form functional neural circuits, and differentiate into mature neurons. Each of these processes is regulated by specific genetic pathways and embryonic environmental factors. A detailed comparison of NfiB+/+ and NfiB−/− mice throughout cortical development revealed defects in both cortical neurogenesis and neuronal differentiation in NfiB−/− brains. During corticogenesis, NFIB is highly expressed in radial glia and in deep layer corticofugal neurons, suggesting important functions in regulating cortical development. Indeed, in NfiB−/− mice, outer radial glia and basal progenitors were absent, corticofugal neurons failed to project axons to the thalamus and spinal cord, and migration of upper layer cortical neurons was defective. Additionally, RNA-sequencing revealed a misexpression of genes regulating cell cycle, neuronal differentiation, axonal projection, and cell migration in NfiB−/− cortices. Together, these results demonstrated the necessity of NFIB in neural stem and progenitor cell differentiation, corticofugal axon projection, and migration of late-born cortical projection neurons. Thus NFIB regulates the development of cortical projection neurons across many levels.
Compared to radial glia and basal progenitors, despite having been discovered over three decades earlier, outer radial glia have only recently been more thoroughly investigated in mice, ferrets and humans (Hevner and Haydar, 2012). In humans, outer radial glia and basal progenitors greatly amplify the neural progenitor pool for neuronal production, which in turn, is likely responsible for the expansion of upper layer cortical neurons (Fietz et al., 2010; Hansen et al., 2010; Reillo et al., 2011; Shitamukai et al., 2011; Wang et al., 2011a, 2011b; Hevner and Haydar, 2012). Other research indicates that the basal process of outer radial glia serve as migrating scaffolds for newly born neurons in the human cerebral cortex (Reillo et al., 2011). However, in rodents, there are significantly fewer outer radial glia; their roles in cortical neurogenesis and as migration guides are not fully understood.
Our results strongly indicate the importance of outer radial glia in both generation and migration of late born cortical projection neurons in mice. Outer radial glia were absent, cortical upper layers were thinner, and late born cortical neurons exhibited a delay in migration toward the cortical plate in NfiB−/− mice. Among these defects, delayed migration of late-born cortical neurons may have contributed to the reduction in upper cortical layer thickness. NFIB expression in late-born neurons before birth is weak or absent (Figure 1), indicating that migration defects of these neurons in NfiB−/− mice may not be a direct consequence of lacking NFIB in the upper-layer neurons. The lack of a migration-scaffold provided by basal processes of outer radial glia is one likely cause of delayed migration. Additionally, ectopic expression of CTIP2 in late-born neurons in NfiB−/− mice suggests that lack of NFIB expression in neural progenitors can effect gene expression and differentiation of cortical neurons, which may indirectly leads to the neuronal migration defect. Collectively, these results suggest that NFIB is critical for generation of outer radial glia, and in turn, is essential for proper differentiation and migration of upper layer neurons.
To date, several genes and molecular pathways that regulate neural progenitor proliferation and differentiation have been identified. For example, increase in activation of the Notch signaling pathway in radial glia has been associated with maintaining “stemness”, while low Notch signaling is correlated with neuronal differentiation (Gaiano et al., 2000; Yoon and Gaiano, 2005; Basak and Taylor, 2007; Imayoshi et al., 2011). Here we show that several genes in the Notch signaling pathway are upregulated in NfiB−/− cortices (Table 4 and Figure 11), supporting previous results indicating that NFI family members regulate Notch signaling (Deneen et al., 2006; Mason et al., 2009; Piper et al., 2010). However, molecular mechanisms that determine the outcome of radial glial progeny are not fully understood, accordingly, further analysis of differentially expressed genes in NfiB−/− mice will provide insight into this process.
In addition to its role in generation of outer radial glia, NFIB is also essential for differentiation of corticofugal neurons as evidenced by the severe reduction of corticofugal axons in NfiB−/− mice (Figure 5). Moreover, NFIB is highly expressed in corticofugal neurons (Figures 2–4) and RNA-sequencing revealed a misexpression of genes that regulate axonal projections and axonogenesis in NfiB−/− mice (Table 3 and and4).4). Together, these results strongly indicate an extensive role for NFIB in differentiation of corticofugal projection neurons. Generation of a neuron-specific NfiB mutant mouse or selective interference of NFIB function in corticofugal neurons may further facilitate our understanding of its role in regulating corticofugal neuron differentiation and axon projection.
We thank Drs. Richard Gronostajski at State University of New York at Buffalo for generously providing the NfiB+/− mice and Dr. Anthony T. Campagnoni from University of California, Los Angeles for providing the Golli-τ-GFP+ mice. We thank Matthew Eckler, William McKenna, Chao Guo and other members of the Chen lab for technical help and scientific discussions. This work was funded by R01MH082965 from National Institute of Health (to BC), a New Faculty Award RN1-00530-1 from California Institute of Regenerative Medicine (to BC), and a training grant R25GM058903 from National Institute of Health (to JB).
Conflict of interest statement
The authors declare no conflict of interest in the present study.
Role of authorsAll authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: JB and BC. Acquisition of data: JB and SK. Analysis and interpretation of data: JB, SK and BC. Drafting of the manuscript: JB, SK and BC. Critical revision of the manuscript for important intellectual content: BC. Statistical analysis: JB and SK. Obtained funding: JB and BC. Administrative, technical and material support: BC. Study supervision: BC.