The mammalian forebrain contains three principle cell types: neurons, oligodendrocytes, and astrocytes. During embryogenesis, these cells are generated in distinct waves from progenitor cells located in different germinal zones of the telencephalic vesicle (Ross et al., 2003; Temple, 2001). The process of neurogenesis requires NPC proliferation, neural fate choice, migration, and differentiation. This delicate process is regulated by cell-extrinsic signals including growth factors, morphogens, and guidance molecules that function through cell surface receptors to regulate cell-intrinsic programs involving transcription factors (Anderson, 2001; Edlund and Jessell, 1999; Guillemot, 2007; McConnell, 1995; Ross et al., 2003). In the adult forebrain, neurogenesis continues in the SVZ of the lateral ventricle and the SGZ of the dentate gyrus, supplying granular neurons to the olfactory bulb and the dentate gyrus, respectively (Alvarez-Buylla and Lim, 2004; Ming and Song, 2005; Zhao et al., 2008).
The Sry-related box (Sox) genes encode a family of transcription factors with a high mobility group (HMG)-type DNA binding domain (Bowles et al., 2000; Kamachi et al., 2000; Schepers et al., 2002). The mouse and human genomes contain 20 orthologous pairs of Sox genes and these 20 genes are divided into 8 subgroups on the basis of sequence similarity and genomic organization (Bowles et al., 2000; Schepers et al., 2002). Sox proteins bind to the consensus sequence 5′-(A/T)(A/T)CAA(A/T)G-3′ in the minor groove of DNA and can function either as transcriptional activators or repressors (Bowles et al., 2000; Harley et al., 1994; Wegner, 1999). Many Sox genes have been shown to have master roles in cell fate specification and differentiation. For example, Sry and Sox9 specify male Sertoli cell fate and differentiation (Sekido and Lovell-Badge, 2009); Sox2 specifies the identity of embryonic stem cells in conjunction with c-Myc, Klf4, and OCT4 (Takahashi and Yamanaka, 2006); Sox1, Sox2, Sox3, and Sox21 play critical roles in regulating NPC identity and neuronal differentiation (Bylund et al., 2003; Graham et al., 2003; Sandberg et al., 2005); Sox8, Sox9, and Sox10 are important regulators of neural crest cell development (Cheung and Briscoe, 2003; Kim et al., 2003); Sox5, Sox6, and Sox9 specify the fate and differentiation of chondrocytes (Lefebvre and Smits, 2005); and Sox18 specifies lymphatic endothelial cells (Francois et al., 2010).
Sox11, Sox4, and Sox12 make up the SoxC subgroup. They share similar but non-identical expression patterns in the developing embryo (Cheung et al., 2000; Dy et al., 2008; Hargrave et al., 1997; Hoser et al., 2008; Schepers et al., 2002). Functional studies have shown that Sox4(−/−) mice die at embryonic (E) 14 from cardiovascular defects (Schilham et al., 1996). Sox11(−/−) mice die shortly after birth with multiple organ defects (Sock et al., 2004). Sox11 is important for the development of sensory neurons, sympathetic neurons, and spinal cord in addition to many non-neural tissues (Lin et al., 2011; Potzner et al., 2010; Thein et al., 2010). Sox12(−/−) mice are viable, fertile, and do not exhibit obvious defects (Hoser et al., 2008). Analyses of compound mutants of SoxC genes have demonstrated that SoxC proteins play redundant roles in the survival of neural and mesenchymal progenitors during early embryonic development (Bhattaram et al., 2010). Sox11 and Sox4 also control the identity and connectivity of cortical spinal neurons (Shim et al., 2012).
In mice, Sox11 is detected in NPCs throughout the developing neuroepithelium as early as E8.5, prior to the onset of neurogenesis (Hargrave et al., 1997; Kuhlbrodt et al., 1998). During later stages of brain development, Sox11 is continuously expressed in NPCs and some differentiated neurons, most notably in the forebrain (Kuhlbrodt et al., 1998). In the adult brain, Sox11 and Sox4 are both specifically expressed in neurogenic niches including the SVZ of the lateral ventricle and SGZ of the hippocampal dentate gyrus (Haslinger et al., 2009; Mu et al., 2012). Within the adult hippocampal neurogenic lineage, Sox11 and Sox4 are co-expressed in proliferating progenitors and immature neurons but downregulated in mature neurons (Mu et al., 2012). Using a retrovirus approach, it was demonstrated that ablation of both Sox11 and Sox4 inhibits in vitro and in vivo neurogenesis from adult NPCs (Mu et al., 2012).
In the current study, we examined the function of Sox11 during embryonic and adult neurogenesis using Sox11 null embryos and Sox11 conditional knockout mice. During embryogenesis, the ablation of Sox11 perturbed NPC proliferation, neuronal migration, and differentiation. In adult brains, the loss of Sox11 specifically in NPCs impaired hippocampal neurogenesis. Using a functional genomic approach, we identified potential Sox11 target genes. Our data established that Sox11 is a critical regulator of both embryonic and adult neurogenesis.
Generation of Sox11 Mutant Alleles
A Sox11 conditional knockout allele was generated using the Cre-loxP-mediated gene targeting system (Fig. 1) as described in the Experimental Procedures section. To generate a null allele, Sox11(f/+) mice were crossed with Prm-Cre mice, which expressed the Cre recombinase in the male germ line (O'Gorman et al., 1997). Sox11(+/−) mice were intercrossed to generate homozygous Sox11(−/−) mice, which died shortly after birth, similar to Sox11lacZ/lacZ knock-in mice in which the Sox11 open reading frame was replaced with a β-galactosidase reporter gene (Sock et al., 2004). This demonstrated that the Sox11 floxed allele can undergo efficient Cre-mediated recombination in vivo. To specifically target adult neural progenitor cells, Sox11(f/f) mice were crossed with Nestin-CreERT2, a tamoxifen-inducible and NPC-specific Cre driver (Chen et al., 2009; Li et al., 2008). As expected, Sox11(f/f);Nestin-CreERT2 conditional knockout mice were viable, healthy, and grossly indistinguishable from Sox11(f/f) littermates.
Sox11 is Required for Brain Development
To study the role of Sox11 in brain development, we analyzed Sox11(−/−) mutant embryos and wild type littermates at multiple developmental stages. The mutant embryos exhibited reduction in brain size from as early as E11.5, and the trend continued to birth (Fig. 2A–B and data not shown). At postnatal day 0 (P0), the thickness of the cerebral cortex was significantly reduced in the Sox11(−/−) mice compared with wild type littermates (Fig. 2C–D). At the mid-sagittal level, the average thickness of the cerebral cortex was reduced by 30%: 2,222.2 ± 211.1 μm in wild type, and 1,666.7 ± 146.7 μm in Sox11(−/−), respectively (P < 0.05). The hippocampus, olfactory bulb, and cerebellum in Sox11(−/−) mice were also significantly smaller compared with wild type littermates (Fig. 2E–N). Therefore, there was a reduction in size throughout the brain of Sox11(−/−) mice. Furthermore, the hippocampus in Sox11(−/−) mice was disorganized: the CA2/3 region split into two bundles, and cells within the developing dentate gyrus (DG) were much more scattered (Fig. 2E–J). In addition, the olfactory bulb in the Sox11(−/−) mice was disorganized and did not have a visible mitral cell layer while the cerebellum in the Sox11(−/−) mice was under-developed (Fig. 2L–N).
To examine cortical development, we performed immunohistochemistry with anti-Tuj1, a neuronal marker, using E12.5 and E17.5 brain sections and performed morphometric analyses. As shown in Figure 3, the thickness of the Tuj1(+) neuronal layer was significantly reduced in Sox11(−/−) embryos as compared with wild type littermates at both E12.5 and E17.5. At E12.5, the average thickness of the Tuj1(+) neuronal layer at the mid-sagittal level was 145 ± 12.8 μm for wild type cortices, and 75 ± 12 μm for Sox11(−/−) cortices, respectively (P < 0.05). At E17.5, the average thickness of the Tuj1(+) neuronal layer at the mid-sagittal level was 1,030 ± 74.5 μm for wild type cortices, and 825 ± 62.5 μm for Sox11(−/−) cortices, respectively (P < 0.05). This indicated that fewer neurons were generated in the absence of Sox11 in vivo. In addition, the thickness of the Tuj1(−) VZ/SVZ, where proliferating NPCs were located, was also significantly reduced in Sox11(−/−) embryos. At E12.5, the average thickness of the Tuj1(−) VZ/SVZ at the mid-sagittal level was 450 ± 42.5 μm for wild type cortices, and 275 ± 34.5 μm for Sox11(−/−) cortices, respectively (P 0.05). At E17.5, the average thickness of the Tuj1(−) VZ/SVZ at the mid-sagittal level was 325 ± 28.5 μm for wild type cortices, and 180 ± 22.5 μm for Sox11(−/−) cortices, respectively (P 0.05). Together, these data indicated that the loss of Sox11 led to reduced cortical neurogenesis in the developing embryo.
Loss of Sox11 Inhibits Proliferation Without Affecting Cell Death
One possible cause for an overall reduction of brain size and reduced neurogenesis is an early defect in the developing neuroepithelium. We confirmed the expression of Sox11 in the developing neuroepithelium by in situ hybridization using a Sox11 antisense RNA probe (Fig. 4A–C). At E10.5, Sox11 was strongly expressed in the telencephalic VZ where NPCs were located (Fig. 4A). At E12.5, Sox11 was upregulated in the marginal zone of the lateral ventricle, where the early differentiated neurons were located (Fig. 4B), while the expression of Sox11 remained strong in the VZ of both the dorsal-lateral neuroepithelium and the dorsal-medial neuroepithelium, which contained NPCs generating projection neurons for the neocortex and the hippocampus, respectively. Sox11 mRNA was not detectable on the Sox11(−/−) sections, confirming the specificity of this Sox11 antisense RNA probe (Fig. 4C).
To examine whether the reduced neurogenesis in the Sox11(−/−) embryos could be caused by proliferation deficits, we performed anti-BrdU immunohistochemistry. BrdU, 2′-bromo-5′-deoxyuridine, is a thymidine analog and can be incorporated into chromosomal DNA by S-phase dividing cells. Timed pregnant female Sox11(+/−) mice, mated with male Sox11(+/−) mice, were injected with 100 μg/g body weight of BrdU, and embryos were collected after 30 min. Cryostat sections were made and subjected to anti-BrdU immunohistochemsitry (Fig. 4D–I). We counted the number of BrdU(+) cells within a defined area that was identical in size for each sample and covered the entire thickness of the cortical wall. We found that at E11.5, the average number of BrdU(+) cells per unit in Sox11(−/−) embryos was reduced by 32% in the dorsal-lateral neuroepithelium, and by 28% in the dorsal-medial neuroepithelium, respectively (Fig. 4J). However, no significant difference in the number of BrdU(+) cells per unit was detected at E12.5 or E13.5 between Sox11(−/−) embryos and control littermates. Therefore, Sox11 was transiently required for NPC proliferation in the developing forebrain. Furthermore, because the overall size of the brain was smaller in Sox11(−/−) embryos from as early as E11.5 to P0, the total numbers of BrdU(+) cells in the entire neuroepithelium of Sox11(−/−) embryos were reduced as compared with age-matched wild type littermates (date not shown).
We also examined apoptosis by immunohistochemistry using an antibody against activated caspase-3. No apparent cell death was observed in the forebrain neuroepithelium in both control and Sox11(−/−) embryos at E11.5, although there was a significant increase of cell death in the Sox11(−/−) trigeminal ganglion (see arrows in Supp. Fig. S1, which is available online). Similar results were obtained using E12.5 and E13.5 brain sections (data not shown). Taken together, these data indicated that the loss of Sox11 inhibited cortical NPC proliferation without significantly affecting cell death.
Loss of Sox11 Affects the Expression of NeuroD and Prox1 in the Developing Hippocampus
It is known that all projection neurons in the hippocampal CA region are born in the dorsal medial VZ, while the interneurons are born in the ventral telencephalon during embryogenesis (Forster et al., 2006; Marin and Rubenstein, 2003; Molyneaux et al., 2007; Ross et al., 2003; Wonders and Anderson, 2006). During early gestation stages, the granule cells in the dentate gyrus are generated in the VZ and these cells populate the developing dentate gyrus by radial migration. By E16.5, a second germinal zone emerges in the hilus and continues to produce dentate granule cells (Altman and Bayer, 1990; Bagri et al., 2002; Nowakowski and Rakic, 1979, 1981; Rakic and Nowakowski, 1981). To understand the effect of Sox11 on hippocampal neurogenesis, we examined the expression of NeuroD, a marker for hippocampal progenitor cells and immature neurons. At E14.5, expression of NeuroD was reduced in both the dorsal-lateral and dorsal-medial neuroepithemium of Sox11(−/−) embryos compared with wild type littermates (Fig. 5A,B). At E16.5, expression of NeuroD was reduced in the emerging dentate gyrus of the Sox11(−/−) embryos compared with wild type littermates (Fig. 5C,D, boxes). We also examined the expression of Prox1, a homeobox transcription factor that was selectively expressed in dentate granule cells and their progenitors (Zhou et al., 2004). We found there were significantly fewer Prox1(+) cells in the developing dentate gyrus of E18.5 Sox11(−/−) embryos than in control littermates (Fig. 5E–G). These data indicated that the loss of Sox11 caused a reduction of hippocampal progenitors and immature neurons during embryogenesis.
Loss of Sox11 Affects NPC Properties In Vitro
To get a better understanding of the properties of NPCs, we used the neurosphere culture system. Telencephalic cells were isolated from E13.5 wild type, Sox11(+/−), and Sox11(−/−) embryos. Progenitor cells were selected and cultured as neurospheres. At passage 3 (P3), single cell suspension was plated and cultured in the differentiation medium that was free of growth factors (EGF, FGF) to allow in vitro differentiation. Six days later, cells were scored for expression of neuronal marker (Tuj1) or astrocytic marker (GFAP), and the total number of cells was scored by counter-staining using DAPI. NPCs from Sox11(−/−) neurospheres exhibited a near two-fold increase in the percentage of Tuj1(+) cells (Fig. 6, A, C, E). The percentages of Tuj1(+) cells were: for wild type, 5.64 ± 0.74%; for Sox11(+/−), 5.06 ± 0.53%; and for Sox11(−/−), 9.72 ± 0.33% (P < 0.001). In contrast, astrocytic differentiation was not significantly affected (Fig. 6, B, D, F). Thus, the loss of Sox11 led to abnormal neuronal differentiation of embryonic NPCs in vitro.
Next, we examined the proliferation and self-renewal capacities of NPCs in vitro. We plated NPCs from P7 neurospheres at a clonal density (2,000 cells/ml) and examined the formation of new neurospheres and their diameters. After 7 days, NPCs from Sox11(−/−) embryos formed fewer neurospheres and the average size of these neurospheres was smaller compared to those derived from wild type or Sox11(+/−) embryos (Fig. 6, G,H). These results indicated that the loss of Sox11 inhibited the self-renewal and proliferative capacities of cortical progenitor cells. Therefore, the endogenous Sox11 protein likely functions by enhancing NPC proliferation and self-renewal, and inhibiting abnormal neuronal differentiation during development.
Sox11 is Required for Adult Hippocampal Neurogenesis
It was previously reported that Sox11 and Sox4 are coexpressed by type-IIb and type-3 intermediate progenitor cells and immature neurons in the adult hippocampal neurogenic lineage (Mu et al., 2012). Using a retrovirus approach, it was demonstrated that the ablation of both Sox11 and Sox4 significantly inhibited adult hippocampal neurogenesis, while the overexpression of Sox11 and Sox4 enhanced hippocampal neurogenesis in vitro (Mu et al., 2012). To examine whether the ablation of Sox11 alone can inhibit adult hippocampal neurogenesis in vivo, S(f/f) mice were crossed with Nestin-CreERT2, a tamoxifen-inducible and NPC-specific Cre transgenic line (Chen et al., 2009; Li et al., 2008), to generate inducible Sox11 conditional knockout mice. As shown in Figure 7A–D, multiple injections of tamoxifen into 6-week-old S(f/f);Nestin-CreERT2 CKO mice abolished the expression of Sox11 in the hippocampal SGZ, but not in the dentate granule cells or CA regions. This demonstrated the specificity and efficiency of this inducible gene ablation system. Immunohistochemistry using antibodies against BrdU and Ki67, markers for proliferating cells, demonstrated that the specific ablation of Sox11 in adult NPCs caused a significant reduction of proliferation in the SGZ (Fig. 7E–H, M). This also led to reduced numbers of DCX(+) and NeuroD(+) immature neurons in the SGZ (Fig. 7I–L, M). Therefore, the ablation of Sox11 specifically in adult NPCs led to reduced hippocampal neurogenesis in vivo.
Sox11 Regulates the Expression of Nmyc, Lis1, and TAK1
Sox11 can serve as a transcriptional activator or repressor. To identify potential Sox11 downstream target genes, we isolated RNAs from the cortex of E17.5 wild type and Sox11(−/−) embryos and performed gene expression analysis using Mouse Genome 430 2.0 arrays (Affymetrix, Santa Clara, CA). The data have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE44308 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE44308). Among 45,101 probe sets that were examined, 1,082 (2.4%) genes were differentially regulated, including 255 genes that exhibited more than a 3-fold change. These genes encode transcription factors, cell cycle regulators, cell adhesion and migration molecules, kinases, axon guidance molecules, and synaptic proteins.
Of particular interest were genes encoding Nmyc (downregulated in Sox11−/−cortex by 4.5-fold), Lis1 (also known as Pafah1b1, upregulated in Sox11−/− cortex by more than 10-fold), and TAK1 (also known as mitogen activated protein kinase kinase kinase 7, or Map3k7, upregulated in Sox11−/− cortex by 2-fold). Nmyc is an important transcription factor that functions downstream of sonic hedgehog to promote cell proliferation (Kenney et al., 2003; Knoepfler and Kenney, 2006; Sjostrom et al., 2005). The deletion of Nmyc specifically in NPCs significantly reduced proliferation and caused abnormal neuronal differentiation (Knoepfler et al., 2002). The downregulation of Nmyc in the Sox11(−/−) cortex likely contributed to reduced proliferation and abnormal neuronal differentiation. Lis1 is a microtubule-associated protein and plays an important role in cortical neuronal migration and lamination (Assadi et al., 2003; Ayala et al., 2007; Hirotsune et al., 1998; Marin and Rubenstein, 2003). The upregulation of Lis1 in the Sox11(−/−) cortex likely contributed to abnormal neuronal migration. TAK1 and the MAPK pathway play important functions in cell proliferation, differentiation, and axonal growth (Johnson and Lapadat, 2002; Markus et al., 2002). The upregulation of TAK1 likely contributed to reduced proliferation and abnormal neuronal differentiation.
We verified the expression of Nmyc and Lis1 in the cortex by Western blot and real-time RT-PCR (Fig. 8). It was clear that the expression of Nmyc was significantly reduced in the cortex of Sox11(−/−) embryos at both the mRNA and protein levels, while the expression of Lis1 was significantly increased in the cortex of Sox11(−/−) embryos at both the mRNA and protein levels (Fig. 8A–C). We also verified expression of TAK1 by Western blot. Consistent with the upregulation of TAK1 mRNA, the protein level of TAK1 in the Sox11(−/−) cortex was significantly increased, which led to a marked increase of P-Erk, although the protein level of Erk was not affected (Fig. 8D). Promoter bashing and chromatin immunoprecipitation experiments will be necessary in the future to determine whether Nmyc, Lis1, or TAK1 are direct transcriptional targets of Sox11.
Expression of Sox11 is Not Affected in the Absence of Pro-Neural bHLH Proteins
In a chick in ovo electroporation system using the developing spinal cord, Sox11 was shown to function downstream of pro-neural bHLH (basic helix-loop-helix) proteins Ngn1 and/or Ngn2 to promote neuronal differentiation (Bergsland et al., 2006). We examined whether Sox11 was downstream of pro-neural bHLH proteins in the dorsal forebrain of mice by performing Sox11 RNA in situ hybridization using E10.5 wild type, Ngn1(−/−), Ngn2(−/−), and Mash1(−/−) embryos (Fig. 9). Compared with wild type embryo, Sox11 expression in the dorsal forebrain was not affected in any of these mutant embryos. Moreover, there was no change of Sox11 expression in the Ngn1(−/−);Ngn2(−/−) double knockout embryos. Therefore, loss of these pro-neural bHLH proteins did not significantly affect the expression of Sox11 in the developing neuroepithelium in mice. The transcription factors and signaling molecules that control Sox11 expression in the developing neuroepithelium remain to be identified.
Sox11 is detected in the developing neuroepithelium from E8.5 on, and remains to be expressed in adult NPCs (Hargrave et al., 1997; Haslinger et al., 2009; Kuhlbrodt et al., 1998). In the adult hippocampal lineage, Sox11 is expressed in proliferating progenitors and immature neurons but downregulated in mature neurons (Mu et al., 2012). We confirmed the expression of Sox11 in the developing neuroepithelium by RNA in situ hybridization. However, the specific distribution of Sox11 protein in stem cells and/or intermediate precursors during embryonic development remains unclear.
Previous studies have shown that the ablation of both Sox11 and Sox4 in the mouse germline causes a significant increase of apoptosis of neural and mesenchymal progenitors, while the ablation of both Sox11 and Sox4 in adult NPCs inhibits hippocampal neurogenesis, suggesting Sox11 and Sox4 play redundant roles in these processes (Bhattaram et al., 2010; Mu et al., 2012). Our study revealed that the ablation of Sox11 alone affects both embryonic and adult neurogenesis. Sox11(−/−) embryos have smaller and disorganized brains, reminiscent of microcephaly and lissencephaly in humans. In addition, ablation of Sox11 specifically in adult NPCs inhibits hippocampal neurogenesis in vivo.
Sox11 Regulates NPC Proliferation
The initial phases of cortical development require the rapid expansion of NPCs through symmetric division in which one cell generates two identical proliferating daughter cells. At the time of neurogenesis, a subset of NPCs becomes restricted to a neuronal lineage. This restriction likely involves asymmetric division in which one daughter cell is maintained as a multipotent progenitor, while the other is fated to differentiate into a neuron (Ayala et al., 2007; Zhong, 2003). In order to supply the cortex with the appropriate number of neurons, it is critical that cortical progenitor cells proliferate sufficiently before differentiation. Therefore, the maintenance of NPCs in a proliferative state and the timing of their differentiation are tightly controlled in vivo. Using Sox11(−/−) embryos, we demonstrated the in vivo proliferation deficits in the neuroepithelium at E11.5. Our neurosphere culture experiments showed there were decreased proliferation and increased generation of neurons in vitro in the absence of Sox11. However, we have not yet found direct in vivo evidence for premature neuronal differentiation in the absence of Sox11 (data not shown).
Ectopic expression studies in chick neural tube suggest that Sox11 and Sox4 play a redundant role in promoting neuronal maturation downstream of pro-neural bHLH proteins (Bergsland et al., 2006). In addition, a genome-wide binding study demonstrates enrichment of Sox11 on neuron-specific genes in embryonic stem cell–derived neurons (Bergsland et al., 2011). Our analyses of mouse mutants for Ngn1, Ngn2, and Mash1 demonstrate that Sox11 is an unlikely downstream target of pro-neural bHLH proteins in the developing mouse neuroepithelium. Consistently, analyses of the spinal cord from mouse embryos deficient for both Sox11 and Sox4 indicate that an absolute requirement for Sox11 and Sox4 in neuronal maturation is unlikely (Thein et al., 2010). While a role of Sox11 in neuronal differentiation is possible, our data argue for an additional role of Sox11 in promoting NPC proliferation and self-renewal.
Nmyc plays an essential role in regulating progenitor cell proliferation and cortical development (Knoepfler et al., 2002). The ablation of Nmyc in NPCs leads to reduced proliferation and aberrant neuronal differentiation, similar to the defects observed in Sox11(−/−) mice. We demonstrated that the expression of Nmyc is inhibited in the absence of Sox11. We also demonstrated that TAK1 is upregulated leading to an increased level of P-Erk, which likely contributes to the abnormal neuronal differentiation of Sox11(−/−) cortical progenitors. Therefore, it is possible that Sox11 maintains the cortical progenitor identity in part by stimulating NPC proliferation through upregulation of Nmyc expression and by inhibiting abnormal neuronal differentiation through downregulation of TAK1 expression. In the future, promoter analysis and chromatin immunoprecipitation (ChIP) experiments will be necessary to confirm whether Nmyc and TAK1 are direct transcriptional targets of Sox11 in vivo.
Sox11 Regulates Hippocampal Neuronal Migration
One of the most interesting observations of the Sox11(−/−) embryos was the disorganized hippocampus, most notable in the CA2/3 region. This indicates that Sox11 is essential for hippocampal neuronal migration in vivo. It was recently reported that the cortex-specific ablation of both Sox11 and Sox4 leads to an inversion of cortical layers similar to that of the reeler mutant mice (Caviness and Sidman, 1973; Shim et al., 2012). Therefore, Sox11 and Sox4 play an additive or synergistic role in regulating cortical neuronal migration.
Our data demonstrate that the loss of Sox11 causes abnormal expression of Lis1, which plays an important role in cortical neuronal migration (Ayala et al., 2007; Bix and Clark, 1998; Kholmanskikh et al., 2003, 2006; Marin and Rubenstein, 2003; Sapir et al., 1997). Heterozygous mutation or deletion of Lis1 in humans is associated with type I lissencephaly, the most severe developmental brain disorder resulting from abnormal neuronal migration. While Lis1(−/−) mice are embryonic lethal soon after implantation, Lis1(+/−) mice display cortical, hippocampal, and olfactory bulb disorganization that results from delayed neuronal migration by a cell-autonomous pathway (Hirotsune et al., 1998). Furthermore, a genetic and biochemical interaction between Lis1 and the reelin signaling pathway has been proposed to regulate neuronal migration (Assadi et al., 2003). We demonstrate that Lis1 is upregulated in the Sox11(−/−) cortex, suggesting that the level of Lis1 protein in vivo is critical for proper neuronal migration. In the future, promoter analysis and ChIP experiments will be necessary to confirm whether Lis1 is a direct transcriptional target of Sox11.
Sox11 Regulates Adult Hippocampal Neurogenesis
Adult hippocampal neurogenesis is a unique type of neural plasticity that leads to the generation of new neurons in the hippocampal dentate gyrus throughout life, and has been implicated in memory, depression, neurological disorders, and functional recovery after traumatic brain injury (Abrous et al., 2005; Blaiss et al., 2011; Deng et al., 2010; Imayoshi et al., 2008; Li et al., 2008; Ming and Song, 2005; Sahay and Hen, 2007; Sahay et al., 2011; Zhao et al., 2008). Although significant progress has been made in our understanding of the activity-dependent extrinsic regulation of adult neurogenesis, our knowledge of cell-intrinsic programs including transcriptional regulation in adult neurogenesis is still limited (Ma et al., 2009, 2010). It was previously reported that Sox11 and Sox4 are co-expressed in the adult neurogenic niches, and the ablation of both genes using a retrovirus approach significantly inhibits adult hippocampal neurogenesis (Haslinger et al., 2009; Mu et al., 2012). It was proposed that Sox11 and Sox4 play redundant roles in adult neurogenesis. In this study, we demonstrated that the ablation of Sox11 alone in adult NPCs caused a significant reduction of proliferation in the SGZ and led to reduced numbers of DCX(+) and NeuroD(+) immature neurons in vivo. It remains unclear which precursor populations in the adult hippocampus were specifically affected by the loss of Sox11, and whether the decreased proliferation and generation of neurons was a consequence of the slowly proliferating Sox11−/− stem cells to progress to a highly proliferative precursor stage.
Generation of a Sox11 Conditional Knockout Allele
A Sox11 conditional knockout allele was generated using the Cre-loxP-mediated gene targeting system (Gu et al., 1994). A 12.4-kb Sox11 genomic clone was obtained by screening a mouse sv129 genomic DNA library using a 2.1-kb mouse Sox11 cDNA as a probe. A Sox11 conditional targeting vector was generated using standard recombinant DNA techniques to insert two loxP sites flanking the only exon of the Sox11 gene. The neomycin-resistant cassette (Neo) flanked by FRT sites was used as a positive selection marker and the diphtheria toxin cassette (dta) was used as a negative selection marker. The targeting vector was linearized, electroporated into mouse R1 embryonic stem (ES) cells, and 200 G418-resistant clones were picked. Among them, 17 clones underwent homologous recombination as examined by Southern blots. Three ES clones carrying the targeted allele were injected into E3.5 C57BL/6 mouse blastocysts to produce chimeric mice. Male chimeras were mated with wild type C57BL/6 females to obtain F1 mice carrying the targeted allele as examined by coat color and Southern blots. F1 mice carrying the targeted allele were mated with the hACTB::FLPe transgenic mice (Rodriguez et al., 2000) to remove the neomycin-resistant cassette to obtain Sox11loxP mice, also referred to as Sox11(f/+) mice. Sox11(f/+) mice were intercrossed to generate Sox11(f/f) homozygous mice, which are healthy and fertile. Sox11(f/f) mice were crossed with Nestin-CreERT2, a tamoxifen-inducible and NPC-specific Cre transgenic line (Chen et al., 2009; Li et al., 2008), to generate Sox11 conditional knockout mice. Sox11 conditional knockout mice are viable and fertile and appear indistinguishable from Sox11(f/f) littermates.
Generation of a Sox11 Null Allele
Sox11(f/+) mice were mated with Prm-Cre mice, which expresses the Cre recombinase in the male germline (O'Gorman et al., 1997), to obtain Sox11(f/+);Prm-Cre mice. Male Sox11(f/+);Prm-Cre mice were crossed with wild type female mice and Sox11(+/−);Prm-Cre mice containing a Sox11 null allele were obtained. Sox11(+/−);Prm-GP mice were crossed with wild type mice and Sox11(+/−) progeny that did not carry the Prm-Cre transgene were retained and used to generate Sox11(−/−) embryos. Sox11(−/−) mice die shortly after birth, similar to the Sox11LacZ/LacZ knock-in mice (Sock et al., 2004).
Mouse Husbandry and Genotyping
Mice were housed and maintained according to the IACUC guidelines of the University of New England, the Arizona State University, and the University of Texas Southwestern Medical Center at Dallas. The Prm-Cre mice (stock number 003328) were obtained from the Jackson Laboratories (Bar Harbor, ME). The Nestin-CreERT2 mice were previously described (Chen et al., 2009; Li et al., 2008). Sox11(+/−) mice were intercrossed to generate Sox11(−/−) embryos at appropriate developmental stages. Sox11(f/f) mice were crossed with Nestin-CreERT2 mice to generate Sox11(f/f);Nestin-CreERT2 conditional knockout mice. Genotyping was performed by PCR using DNA from tails or yolk sacs. For the Sox11 floxed allele, two primers were used: FP1, 5′-GTG ATT GCA ACA AAG GCA GA-3′; and RP, 5′-TCT GCC GAT GTC TTT CAG AC-3′. The PCR condition was: 94°C for 2 min; 94°C for 30 sec, 54°C for 45 sec, 72°C for 60 sec; 40 cycles. Sox11(+/+) mice had one band of 110 bp, Sox11(f/+) mice had two bands of 110 bp and 150 bp, and Sox11(f/f) mice had one band of 150 bp. For the Sox11 null allele, three primers were used: FP1, 5′-GTG ATT GCA ACA AAG GCA GA-3′; RP, 5′-TCT GCC GAT GTC TTT CAG AC-3′; and FP2, 5′-TGA ATC CGT GAA CTC CTT T-3′. The PCR condition was the same as that used for the Sox11 floxed allele. Sox11(+/+) mice had one band of 110 bp, Sox11(+/−) mice had two bands of 110 bp and 200 bp, and Sox11(−/ −) mice had one band of 200 bp. The primers and PCR condition for Prm-Cre and Nestin-CreERT2 were previously described (Chen et al., 2009; O'Gorman et al., 1997).
All histological examinations were performed on age-matched littermates. Timed-pregnant female mice were sacrificed at various stages to collect embryos. Gestational age was calculated by assigning the morning of the appearance of the mother's vaginal plug as embryonic day (E) 0.5. Embryos were dissected out, fixed overnight in 4% (w/v) paraformaldehyde (PFA), cryoprotected in 30% sucrose until sunk, and embedded in OCT. Postnatal mice were intra-cardially perfused with phosphate buffered saline (PBS) followed by 4% PFA. Brains were dissected out, postfixed overnight with 4% PFA, cryoprotected in 30% sucrose until sunk, and embedded in OCT. Fourteen-micrometer serial sections were made using a Cryostat. For histology, Nissl staining was performed and anatomically matched sections were used in comparisons. Digital images were captured with a Nikon DS-Fi1 CCD camera under the brightfield using a Nikon Eclipse TE2000-U phase-contrast microscope.
RNA In Situ Hybridization
The digoxigenin (DIG) nonisotopic RNA in situ hybridization was carried out on cryostat sections as previously described (Lei et al., 2005, 2006). An antisense RNA probe for Sox11 was used (Lin et al., 2011).
Immunohistochemistry and Cell Counts
Immunohistochemistry was carried out as previously described (Lei et al., 2005, 2006). Sections were blocked in 3% normal donkey serum in PBS. Primary antibodies included mouse anti-TUJ1 (1:500, Covance, Princeton, NJ), goat anti-NeuroD (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anti-activated caspase 3 (AC3, 1:200, Cell Signaling, Danvers, MA), rabbit anti-Prox1 (1;2,000, Chemicon, Temecula, CA), rabbit anti-Ki67 (1:300, Lab Vision, Fremont, CA), goat anti-DCX (1:200, Santa Cruz Biotechnology), and mouse anti-BrdU conjugated with Alexa488 (Anti-BrdU-Alexa488, 1:500, Molecular Probes, Eugene, OR). Appropriate Cy2-, Cy3-, Cy5-, or biotin-conjugated secondary antibodies (1:200, Jackson ImmunoResearch, Bar Harbor, ME) were used for the detection of primary antibodies. For immunofluorescence, DAPI (1 μg/ml, Molecular Probes) was added together with the secondary antibodies to stain nuclei. Immunostaining was visualized using a Nikon (Melville, NY) Eclipse TE2000-U inverted fluorescence microscope or a Zeiss (Thornwood, NY) 510 confocal microscope. Digital images were captured using a Photometrics CoolSnap CCD camera (Roper Scientific, Martinsried, Germany) and analyzed using the Metamorph software (Universal Imaging, Bedford Hills, NY). For biotin-conjugated secondary antibodies, an ABC kit and DAB reagent were used to detect the signals (Vector Laboratories, Burlingame, CA). Digital images were captured with a Nikon DS-Fi1 CCD camera under the brightfield using the Nikon Eclipse TE2000-U microscope.
To examine proliferation during embryogenesis, embryos were collected from timed pregnant females injected intraperitoneally with 100 μg/g body weight of 2′-bromo-5′-deoxyuridine (BrdU, Sigma, St. Louis, MO) 30 min before sacrifice. Immunohistochemistry for BrdU-positive cells was done using 14-μm sagittal sections through the telencephalic vesicle. Boxes of the same size were drawn to cover the thickness of the dorsal-lateral and dorsal-medial cortical walls, respectively, and the number of BrdU(+) cells within each box was counted. To examine neurogenesis in adult mice, P42 male S(f/f) mice and S(f/f);Nes-CreERT2 mice were injected intraperitoneally with tamoxifen (85 μg/g body weight, twice a day for 5 days). One day after the final tamoxifen injection, mice were injected once with 200 μg/g body weight of BrdU. Two hours later, mice were perfused with PBS followed by 4% PFA. Brains were dissected out, postfixed overnight in 4% PFA, cryoprotected, and embedded in OCT. Immunohistochemistry for BrdU, Ki67, doublecortin, and NeuroD-positive cells was done using 14-μm coronal sections through the hippocampus (bregma: −0.82 mm to −4.24 mm). The granular and sub-granular layer of the dentate gyrus in every fifth section was examined, from which the total number of positive cells in both hemispheres was calculated. All cell counts were performed in a genotype-blind manner.
Neurosphere culture was performed following the instruction of StemCell Technologies (Vancouver, Canada). Dissociated telencephalic cells were prepared from E13.5 mouse cortices by triturating through fire polished glass pipette in L-15 media (Invitrogen, Carlsbad, CA) until single-cell suspension was achieved. Cells were then plated on 6-well plates at a 2×105/ml density in NeuroCult NSC Basal Medium (Mouse, StemCell Technologies) supplied with NeuroCult NSC Proliferation Supplements (Mouse, StemCell Technologies), 1% Pen/Strep, 20 ng/ml epidermal growth factor (EGF, Invitrogen), and 10 ng/ml fibroblast growth factor (hFGF, Invitrogen) (abbreviated as "NSC-P" medium). The same media was added every 3 days during a 6–8-day time course until neurospheres were formed. Passage of neurospheres was done by chemical dissociation using accutase (1×, Chemicon) for 10 min at 37°C. To examine in vitro differentiation, at passage 3 (P3), single cell suspensions dissociated from neurospheres were plated at a 5×104/cm2 density on the poly-D-ornithine and laminin (Invitrogen) coated 8-well chamber slides in NeuroCult NSC Basal Medium (Mouse, StemCell Technologies) plus NeuroCult NSC Differentiation Supplements (Mouse, StemCell Technologies) (abbreviated as "NSC-D" medium). Six days after plating, cells were fixed with 4% PFA for 30 min and subjected to triple-labeling using anti-TUJ1 (neuronal marker), anti-GFAP (astroglial marker), and DAPI (staining all nuclei). For the differentiation assay, 6 wells each were plated with wild type cells or Sox11(+/−) cells and 4 wells were plated with Sox11(−/−) cells. TUJ1(+), GFAP(+), or DAPI(+) cells were counted separately for each well using 12 fields that cover the entire area. On average, over 1,700 DAPI(+) cells were counted for each well. The percentage of DAPI(+) cells that also expressed TUJ1 or GFAP was calculated, respectively. To examine in vitro proliferation and self-renewal, dissociated cells from P7 neurospheres were plated at clonal density (2,000 cells/ml) in the NSC-P medium and cultured as above. Seven days later, the number of neurospheres per well and the size of the neurospheres were measured. Neurospheres with a diameter larger than 50 μm were counted.
Mouse cortical proteins were extracted and subjected to Western blot as previously described (Romero et al., 2007). Briefly, freshly dissected brain tissues were quickly rinsed with ice-cold 1× PBS twice and then homogenized in RIPA buffer (50 mM Tris-HCl, PH 7.4, 1% NP-40, 0.25% Na-deoxycholate, 150 mM NaCl, 1 mM EDTA, 1 mM Na3VO4, and 1 mM NaF) plus protease inhibitor cocktail Complete Mini (1 tablet/10 ml, Roche, Indianapolis, IN). The solutions were incubated on ice for 30 min and centrifuged at 14,000 rpm for 20 min at 4°C. Supernatants were collected and protein concentrations were determined by the BCA Protein Assay Kit (Pierce, Rockford, IL). Twenty micrograms of proteins were loaded into each well and separated by 8% SDS-PAGE gels. Proteins were then transferred to nitrocellulose membranes for blotting. Primary antibodies included rabbit anti-Nmyc (1:200, Santa Cruz Biotechnology), goat anti-Lis1 (1:100, Santa Cruz Biotechnology), rabbit anti-phospho-p44/42 MAPK (P-Erk, 1:1,000, Cell Signaling), rabbit anti-p44/42 MAPK (Erk, 1:1,000, Cell Signaling), rabbit anti-TAK1 (1:1,000, M-579, Santa Cruz Biotechnology), and mouse anti-acetylated tubulin (1:1,000, Sigma). HRP-conjugated secondary antibodies were used (1:5,000, Santa Cruz Biotechnology). Immunoreactivity was detected with the ChemiGlow reagent (Alpha Innotech, San Leandro, CA), and the densitometry analysis was performed with the KODAK (Rochester, NY) 1D Image Analysis Software.
Total RNA was isolated from E17.5 wild type and Sox11(−/−) cortices using TRIzol (Invitrogen). Microarray analysis was performed using Mouse Genome 430 2.0 arrays (Affymetrix). cRNA probe synthesis, hybridization, scanning, and data collection were performed following the manufacturer's instruction at the Microarray Core Facility of University of Texas Southwestern Medical Center at Dallas. Microarray data was further analyzed using the GeneSpring software (Agilent Technologies, Santa Clara, CA).
Total RNA was isolated from E17.5 wild type (n=2) and Sox11(−/−) (n=4) cortices using TRIzol (Invitrogen). cDNA was made using the SuperscriptIII reverse transcriptase (Invitrogen). Quantitative PCR was done using an ABI Prism 7000 machine and SYBR Green (Applied Biosystems, Foster City, CA) (Lei et al., 2005). Two pairs of Lis1 primers and three pairs of Nmyc primers were used, and one pair of β-actin primers was used as an internal control. The primers were: Lis1-1 forward, 5′-CCC AAG AAG TCC TAA CCG CA-3′, Lis1-1 reverse, 5′-ACA CAA ACC AAC CTG CAG CTT-3'; Lis1–2 forward, 5′-GGT GTG CTC AGA GCA CTA CCC-3′, Lis1–2 reverse, 5′-CCG GGA CGG CTT GCT AAT-3′; Nmyc-1 forward, 5′-TGG AAG TTC GGG ACA CTA AGG AGC-3′, Nmyc-1 reverse, 5′-AGG GGC ATC AAA TGG CAA CC-3′; Nmyc-2 forward, 5′-TGC TGG GTG GCT TGT TTT CC-3′, Nmyc-2 reverse, 5′-CAG GGG CAT CAA ATG GCA AC-3′; Nmyc-3 forward, 5′-AAG GCG GTA ACC ACT TTC ACG-3′, Nmyc-3 reverse, 5′-CTG GAA CAA CAC TTT TGA GGG G-3′; β-actin forward, 5′-ACA GTC CGC CTA GAA GCA CT-3′, and β-actin reverse, 5′-TCC GAT GCC CTG AGG CTC TT-3′. For each sample, the reaction using each pair of primers was done in triplicates. The relative expression levels of Lis1 and Nmyc were normalized against the level of β-actin.
All values were presented as mean ± SEM. Student's t-tests (two-sample assuming unequal variance) were applied to data with two groups of samples. ANOVA analyses were used for comparisons of data with more than two groups. Post hoc group comparisons were performed with Bonferroni test. A value of P < 0.05 was accepted as statistically significant.
This work was supported by a grant to L.L. from the National Institute of General Medical Sciences (P20GM103643). We are grateful to Drs. Ed Bilsky, Mike Burman, Ling Cao, and Ian Meng for critical reading of the manuscript. We also thank technical assistance from Brittany Roy, Taylor Cushman, Michael Anderson, Kristen Nash, Junshi Wang, and Emilea Y. Lee. L.F.P. is an American Cancer Society Research Professor.