|Home | About | Journals | Submit | Contact Us | Français|
The transcription factor, Sox1 has been implicated in the maintenance of neural progenitor cell status, but accumulating evidence suggests that this is only part of its function. This study examined the role of Sox1 expression in proliferation, lineage commitment, and differentiation by telencephalic neural progenitor cells in vitro and in vivo, and further clarified the pattern of Sox1 expression in postnatal and adult mouse brain. Telencephalic neural progenitor cells isolated from Sox1 null embryos formed neurospheres normally, but were specifically deficient in neuronal differentiation. Conversely, overexpression of Sox1 in the embryonic telencephalon in vivo both expanded the progenitor pool and biased neural progenitor cells towards neuronal lineage commitment. Sox1 mRNA and protein were found to be persistently expressed in the postnatal and adult brain in both differentiated and neurogenic regions. Importantly, in differentiated regions Sox1 co-labeled only with neuronal markers. These observations, coupled with previous studies, suggest that Sox1 expression by early embryonic progenitor cells initially helps to maintain the cells in cell cycle, but that continued expression subsequently promotes neuronal lineage commitment.
Sox (Sry-related-HMG box) genes encode a family of 20 proteins (Schepers et al., 2002) characterized by a high-mobility-group (HMG)-box domain that is homologous to the DNA-binding domain of testis-determining gene Sry (Coriat et al., 1993; Denny et al., 1992; Gubbay et al., 1990; Sinclair et al., 1990; Wright et al., 1993). Sox transcription factors participate in cell fate decisions in multiple tissues including the central nervous system (CNS) during development and postnatal life (Bowles et al., 2000; Schepers et al., 2002; Wegner, 1999). The SoxB1 subfamily (Sox1, Sox2 and Sox3) is evolutionarily conserved and is particularly important for the development of the CNS in different species including Drosophila, Xenopus, chicken and mouse (Bylund et al., 2003; Collignon et al., 1996; Kan et al., 2004; Pevny et al., 1998; Rex et al., 1997; Uchikawa et al., 1999; Uwanogho et al., 1995; Wood and Episkopou, 1999). However, there are conflicting views about the functional roles of SoxB1 members in mammalian CNS development, and further detailed studies are needed.
All three SoxB1 members are co-expressed in the murine and avian neuroepithelium and appear to participate in maintaining neural progenitor cell identity. Sox1 has been used as a marker of embryonic neural stem cells (Aubert et al., 2003; Wood and Episkopou, 1999), and Sox1 expression is reportedly downregulated in progenitor cells as they exit cell cycle and terminally differentiate (Pevny et al., 1998). Sox2, another member of the SoxB1 subfamily was shown to be crucial for maintenance of progenitor cell identity (Graham et al., 2003). SoxB1 factors inhibit neuronal differentiation by avian spinal cord progenitor cells by repressing differentiation events downstream of proneural basic helix-loop-helix proteins (Bylund et al., 2003). These observations have led to the perception that SoxB1 factors all function similarly to maintain the progenitor cell state (Pevny and Rao, 2003). However, Sox1 null, Sox3 null, and Sox2 compound heterozygous mice carrying a Sox2 null and a hypomorphic Sox2 allele all have different neuronal defects, including neuronal degeneration, in specific brain regions, while no glial abnormalities are apparent. Specifically, Sox1 null mice have severe developmental deficits of neurons within the olfactory tubercle and the nucleus accumbens shell (Ekonomou et al., 2005; Malas et al., 2003), whereas mice carrying compound Sox2 hypomorphic alleles exhibit variable phenotypes associated with neural degeneration and abnormal neuronal function (Ferri et al., 2004). In Sox3 null mice, a subset of hypothalamic neurons fails to differentiate (Rizzoti et al., 2004). Furthermore, we have previously shown that Sox1 promotes neuronal lineage commitment by telencephalic progenitor cells in vitro (Kan et al., 2004). These observations led us to hypothesize that, in addition to effects on maintenance of the progenitor cell state, Sox1 may also promote neuronal differentiation in vivo. This study focused on the functional roles of Sox1 in mouse brain development and sought to test the above-mentioned hypothesis.
A retroviral expression vector pCLE-IRES2-eGFP, a kind gift from Dr. Jeffrey Nye, was derived from the pCLE retroviral vector (Gaiano et al., 1999; Gaiano et al., 2000) by replacing IRES-PLAP with an IRES2-eGFP sequence. The coding region of Sox1 was subcloned into this vector. Additional information on oligonucleotide primers and the cloning strategy used to generate this construct is available upon request. Virus was produced by double transfection of GP-293 cells with pCLE-IRES2-eGFP and pVSV-G constructs. Viral supernatant was collected for 3 days and 100-fold concentrated by ultracentrifugation at 25,000g for 1 hour 30 minutes.
Embryonic brains were harvested, pre-fixed with 4% paraformaldehyde, cryoprotected in 15% sucrose/PBS, snap-frozen in dry ice/isopentane slurry, and coronal sections were prepared using a cryostat. Brain sections and cultured cells were fixed with 4% paraformaldehyde in phosphate buffered saline (PBS). For BrdU immunohistochemistry, the sections were pre-treated with 1M HCl 37°C for 30min. Non-specific binding was blocked with 10% normal serum diluted in 1% bovine serum albumin and 0.25% Triton X-100 for one hour in room temperature. The sections were then incubated with primary antibodies diluted with 1% BSA + 0.25% Triton X-100 at 4°C overnight. The primary antibodies were: rabbit anti-Sox1 (1:2000, kind gift from Dr. Yusuke Kamachi and Dr. Hisato Kondoh, Osaka University, Japan), mouse anti-BrdU (1:1000), anti-neuronal nuclei (NeuN) (1:500), GluR2/3 (1:200), MAP2 (1:500) and O4(1:500) are from Chemicon (Temecula, CA, USA), mouse anti-glial fibrillary acidic protein (GFAP) (1:1000) and anti-βIII-tubulin (1:1000) are from Sigma Saint Louis, MO, USA, GAD65 (1:200, BD, Franklin Lakes, NJ, USA ), and rabbit anti-GFP (1:750, Invitrogen, Carlsbad, CA,USA). The sections were then incubated with appropriate secondary antibodies (Cy3 or Cy2 conjugated antibodies (Jackson Lab, Bar Harbor, Maine, USA) diluted with 1% BSA + 0.25% Triton X-100 or Alexa Fluor 488, Alexa Fluor 594, and Alexa 647 (1:1000, Invitrogen, Carlsbad, CA, USA), in the dark at room temperature for 2 hours. Counterstaining was then performed with DAPI (1:5000, Sigma, Saint Louis, MO, USA) or Hoechst for 10 min at room temperature. For thick sections, both first and second antibodies incubation conditions were changed to 24 hours at room temperature. The sections were then mounted with anti-fade kit (Invitrogen, Carlsbad, CA, USA). Regular sections or coverslips were photographed using a Zeiss Axiovert fluorescence microscope (Zeiss, NY, U.S.A.). Thick sections were photographed using Zeiss LSM 510 Laser Scanning Confocal Microscope. Fluorescent images were processed with Adobe Photoshop, cells on embryonic brain sections were counted based on the nuclei (stained with Hoechst) of GFP+ cells. The theoretical division number of individual cluster was calculated (theoretical division number=log2 (total number of GFP+ cells in a cluster), and numbers were compared using Student’s t-test.
Mutant mice with β-galactosidase knocked into the Sox1 locus (Sox1lacz/+) have been described elsewhere (Ekonomou et al., 2005; Nishiguchi et al., 1998). For tissue analyses, the animals were deeply anesthetized (Equithesin, 0.3ml/100g, i.p.) and decapitated. The brains were then rapidly removed and frozen in powdered dry ice and stored at −80°C. Coronal sections, 10μm thick, were cut with a cryostat (Leica, Germany).
LacZ staining was performed as previously described (Ekonomou et al., 2005) with slight modifications. Briefly, 1) The brains of Sox1lacz/+ mice were removed, 1mm thick coronal sections were made with a brain slicer, and the sections were immediately transferred to fixative (4% PFA) and incubated at 4°C overnight in the dark or room temperature for 2 hours. 2) Sections were rinsed three times for 15 minutes each time with rinse buffer(Ekonomou et al., 2005) at room temperature. 3) The sections were then incubated in the dark with staining buffer at room temperature overnight. 4) Sections were then transferred to fixative at 4°C (4% PFA) and incubated in the dark overnight. 5) Sections were then transferred to 70% ethanol for long term storage. 6) Images were captured using a Spot Insight Digital Camera (Diagnostic Instruments, Inc., Sterling Heights MI, USA). For LacZ staining of neurospheres, the fixation condition was changed to 5 min at room temperature, and the staining condition was changed to 5 hours at room temperature.
Sox1 mRNA expression was detected in situ with a 50-mer anti-sense oligonucleotide probe complementary to the Sox1 3′-UTR: 5′-CGA GGC GCT GAC ACC AGA CTG GCC TCT TAG ACT GAA CTT TGG TGT TTT CA -3′ with 3′-digoxigenin-labeling (IDT, IA, USA). Hybridization was performed in 50% formamide, 4x standard saline citrate, 1 x Denhardt’s solution, 10% dextran sulphate, 0.5 mg/ml sheared salmon sperm DNA, 1% sarkosyl (N-lauryl sarcosione), 0.02 M phosphate buffer (PB, pH 7.0). The sections were hybridized overnight at 42°C in a humidified chamber with 0.25 ml per slide of hybridization cocktail. They were subsequently rinsed, washed 4 × 15 min at 55°C in 1 x standard saline citrate (SSC), blocked with 10% normal sheep serum and then incubated with anti-digoxigenin-AP antibody (1:5000, Roche, New Jersey, USA) for 5~12h at room temperature. Hybridization was visualized after incubating with 4-nitro blue tetrazolium chloride (NBT) and 5-bromo-4-chloro-3-indolyl-phosphate (BCIP). The slides were then mounted with glycerin jelly.
To label dividing cells in the brain in vivo, bromodeoxyuridine (BrdU, Sigma, Saint Louis, MO, USA ) was injected (50mg/kg, i.p.) four times at 2-hour intervals before sacrificing the animals. To double-label dividing cells both in neurospheres and subsequently at different stages of differentiation, we generally followed the protocol outlined previously (Vega and Peterson, 2005). Briefly, secondary neurospheres (three days after the first passenger) were treated with CldU (Sigma, Saint Louis, MO, USA) for 6 hours (final concentration 10μM), and the neurospheres then were washed with DMEM-F12 medium once. 250 washed neurospheres were plated in DMEM-F12 medium with supplements (N2, B27, P/S/G, but without growth factors). Three days after plating on PDL/Laminin coated coverslips, the cells were treated with IdU (Fluka, USA) for 6 hours (final concentration 10μM). The cells were then fixed and double-stained with rat anti-CldU (Accurate Chemical, #OBT-0030, Westbury, NY, USA ) and mouse anti-IdU(BD, #347580, Franklin Lakes, NJ, USA).
Quantitative RT-PCR primers Sox2 (CAGGGAGTTCGCAAAAGTCT, TGGACATTTGATTGCCATGT), and Sox3 (CGTAACTGTCGGGGTTTTGT, AACCTAGGAATCCGGGAAGA) were designed base on 3′UTR of mouse Sox2 and Sox3 sequences, using the OligoPerfect™ Designer (Invitrogen, Carlsbad, CA, USA), and these primers have no significant homology to any other sequences in the public databases. 1μg Total RNA from E12 neurospheres were used as the templates to perform the quantitative RT-PCR, using SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA). Anti-Sox2 (Bani-Yaghoub et al., 2006) and anti-Sox3(R&D, AF2569, Minneapolis, MN, USA) were used to detect the Sox2 and Sox3 protein from E12 neurospheres on Western Blot.
Embryos generated by matings between heterozygous mice (Sox1lacz/+) were dissected from the uterus in DMEM-F12 medium. The day of the virginal plug was counted as 0.5 day. The isolation of neural progenitor cells and culture techniques were described previously (Kan et al., 2004). )Briefly, progenitors from two brain regions, ganglionic eminence or dorsal telencephalic wall, were isolated and dissociated in the same medium with supplements (N2, B27, P/S/G, heparin and bFGF or EGF in case of E16 neurospheres) and the remaining tissues were collected for genotyping (by PCR and LacZ staining). 105 dissociated cells from each embryo were seeded in 10ml of DMEM-F12 medium with supplements in Petri dishes to test the ability to form neurospheres. The total number of neurospheres was counted after 4 days in culture. These primary spheres were then dissociated, and cell numbers were counted again. 105 dissociated cells from primary spheres were seeded in the same medium in Petri dishes to generate secondary spheres. The total number of secondary neurospheres was counted after 4 days in culture in the same manner. To study the lineage potential of neurospheres from different mouse embryos, 250 secondary spheres from each embryo were plated on PDL/Laminin coated coverslips in 24 well plates, in the same medium with supplements (N2, B27, P/S/G, but without growth factors). Cells were fed every 3 days for a period of 7 days or otherwise specified days. The cells were then fixed and stained with lineage markers (GFAP, βIII-tubulin, MAP2, O4 and CNPase).
We generally followed the protocol outlined previously (Gaiano et al., 1999; Turnbull, 1999; Wichterle et al., 2001). Briefly, timed Swiss-Webster mice were anesthetized and a sterile incision was made along the ventral abdomen. A section of the uterus containing 1–2 embryos was pulled through a slit in a rubber membrane into a PBS-filled Petri dish where the embryos were visualized using a high frequency ultrasound probe. The 30° beveled tip (60 μm O.D.) of a pulled glass capillary was inserted into the telencephalic vesicle of each embryo in utero using ultrasound guidance. 1–2 μl of replication-deficient viral stock (~105 infectious particles) was injected through the glass capillary into the telencephalic vesicle of each embryo. The capillary was then removed, the uterus tucked back inside the mother, the abdominal wall sutured, and the skin clipped closed allowing for recovery of mother and natural growth of embryos.
Sox1 is expressed by neural progenitor cells in the CNS, and null mutation of Sox1 leads to abnormal development of specific brain regions (Ekonomou et al., 2005; Malas et al., 2003). To help define mechanisms underlying these abnormalities, we compared neural progenitor cells from embryos of Sox1 null mice (Sox1lacz/lacz) and WT littermates for their ability to form neurospheres. Neurospheres were first generated from the ganglionic eminence of E13, E14 and E16 embryos produced by crossing heterozygous male and female Sox1lacz/+ mice. We found that most cells in both primary and secondary neurospheres generated from Sox1lacz/lacz or Sox1lacz/+ embryos were LacZ+ (Fig. 1A), indicating that the Sox1 gene locus is expressed in the cultured cells. Quantitative study found that the total numbers of primary and secondary neurospheres generated from progenitors of Sox1lacz/lacz, Sox1lacz/+ and WT littermates were comparable (Fig. 1B). The average sizes of the spheres (number of cells per sphere) also did not differ significantly (Fig. 1C). To further address potential regional specific effects on neurosphere formation ability, we repeated the procedure and generated the neurospheres from the dorsal telencephalic wall, where persistent Sox1 expression is also evident at different developing stages ((Ekonomou et al., 2005) and Fig. 5 & 6). We found that the total numbers of primary and secondary neurospheres generated from progenitors of Sox1lacz/lacz and WT littermates were also comparable (Fig. 1D), and that the average sizes of the spheres (number of cells per sphere) did not differ significantly (Fig. 1E). Therefore, reduction or lack of Sox1 expression does not significantly compromise the ability of stem/progenitor cells to form neurospheres, regardless of the progenitor cells’ temporal-spatial origin in the brain. These observations indicate that neural progenitor cells are able to proliferate in the absence of Sox1, although this could reflect compensatory mechanisms such as upregulation of other SoxB1 factors. To test this possibility, we preformed quantitative RT-PCR and Western blotting to examine levels of Sox2 and Sox3 mRNA and protein in WT and Sox1 null neurospheres. There was a small up-regulation of Sox2 mRNA in Sox1 null neurospheres but no change in levels of Sox3 mRNA. (Figure 2A), indicating a possible compensatory mechanism. However, at the protein level there were no detectable changes in levels of Sox2 or Sox3 (Figure 2B). Overall, these data suggest that up-regulation of other SoxB1 factors, especially Sox2, could play some compensatory role, but that such changes are small.
To examine effects of Sox1 deletion on lineage commitment, secondary neurospheres from ganglionic eminence were plated onto PDL/Laminin coated coverslips for 7 days differentiation before fixation and staining. Progenitor cells from Sox1 null mice retained the potential to generate cells immunoreactive for GFAP, CNPase, or βIII-tubulin (markers of astrocytes, oligodendrocytes, and neurons, respectively). However, the percentage of βIII-tubulin+ cells generated from Sox1lacz/lacz cells was significantly lower compared to WT littermates in all age groups (P<0.01 in E13 & E14 group, P<0.05 in E16 group, Fig. 3A). The decrease in βIII-tubulin+ cells in the absence of Sox1 could reflect multiple potential defects: 1) a switch from the neuronal lineage to other differentiated lineages, or 2) a specific deficiency in neuronal differentiation, or 3) premature or accelerated differentiation, or 4) delayed neuronal differentiation. To distinguish among these possibilities, we counted the GFAP and CNPase expressing cells as well as the undifferentiated cells lacking any of these lineage markers in cultures from E13 embryos after 7 days differentiation on PDL/Laminin coated coverslips. The percentage of GFAP+ cells generated from Sox1lacz/lacz cells was comparable to WT littermates (Fig. 3B) and there were no obvious morphologic differences in the GFAP+ cells. Further, the percentage of CNPase+ cells was about 1% in both the control and Sox1 null groups, respectively (not shown). However, an apparently higher number of cells lacking any of these markers were present in the Sox1lacz/Lacz group commensurate with the reduction in the number of neurons (Fig. 3B). This suggested that the Sox1 null cells may have persisted as undifferentiated, progenitor cells due to diminished neuronal lineage commitment.
To directly address this issue we performed double-labeling studies of the progenitor cells with CldU and IdU. The double-labeling strategy was chosen so that we could quantify the proliferation of these cells both in the neurosphere form and after plating and differentiation. Secondary neurospheres from ganglionic eminence were grown for 3 days and then labeled with CldU for 6 hours. The cells were then washed and immediately dissociated and plated onto PDL/Laminin coated coverslips in differentiation medium without growth factors. After 3 days in differentiation medium, the cells were then pulsed with IdU for 6 hours before fixation and analysis. There was no statistical difference in CldU labeling between WT and Sox1 null cells although there was a trend towards a decrease with the Sox1 null spheres (Supplementary Fig. 1). This is consistent with the observations in Figure 1 that neurosphere size is comparable between groups. As expected, there was much less labeling with IdU after the cells were plated onto the PDL/laminin, but there were significantly more IdU+ cells in the Sox1 null group compared to WT (Supplementary Fig. 1). This is consistent with the observation that an apparently higher number of undifferentiated cells (presumably progenitor cells that remain in cell cycle) and fewer postmitiotic neurons were present in Sox1 null cultures after plating onto the PDL/Laminin. Notably, the βIII-tubulin+ cells that were generated from Sox1lacz/lacz mice tended to have a more immature morphology with shorter process than neurons generated from WT progenitor cells (Fig. 3C). These observations are consistent with the hypothesis of a specific deficiency in neuronal differentiation of Sox1 null progenitors.
To investigate the potential regional and/or temporal effects on neuronal differentiation, E12, E14 and E16 secondary spheres generated from dorsal telencephalic wall were differentiated on PDL/Laminin coated coverslips for different time periods (3, 7 and 14 days). The shorter and longer differentiation periods (3 days and 14 days) were included to test for the possibilities that neuronal differentiation was simply accelerated or delayed. The percentage of neuronally differentiated cells was determined using two separate neuronal markers, βIII-tubulin and MAP2. The percentage of astrocytes was determined by GFAP staining, and the percentage of oligodendrocytes was determined by staining for O4. Findings with E12, E14 and E16 secondary spheres were similar. The data from E14 spheres are illustrated in Fig. 3D. The numbers of βIII-tubulin+ cells and MAP2+ cells generated from Sox1lacz/lacz cells were both significantly lower compared to WT littermates in all tested conditions (P<0.01 in most conditions, Fig. 3D and data not shown), similar to the findings with the ganglionic eminence progenitor cells (Fig. 3A). However, the number of GFAP+ cells generated from Sox1lacz/lacz cells was significantly higher (after 7 days of differentiation) compared to WT littermates (P<0.01%, Fig. 2D), which differs from findings with the ganglionic eminence cells (Fig. 3B). This may reflect regional differences between the progenitors of ganglionic eminence and dorsal telencephalic wall, although this difference diminished after 14 days of differentiation (Fig. 3D). The cells that were positive for O4, or unlabeled by any lineage markers were comparable between the Sox1 null and WT group at this time point (data not shown).
To find out whether accelerated or delayed neuronal differentiation was responsible for the lower neuronal counts at 7 days of differentiation of Sox1 null progenitors derived from the dorsal telencephalic wall, we compared the differentiation pattern of the cells in our culture system at several time points. After 3 days of differentiation on PDL/Laminin coated coverslips, the cells appeared incompletely differentiated with an immature morphology, and the percentages of cells expressing astrocytic or neuronal markers were very low both in Sox1 null and WT cells (Fig. 3D). These findings indicate that Sox1 null progenitors do not have premature or accelerated differentiation. The deficiency in neuronal differentiation by Sox1 null progenitor cells persisted after 14 days in differentiation medium (P<0.01 by βIII-tubulin staining and P<0.05 by MAP2 staining, respectively), indicating that it did not simply reflect a delay in differentiation.
Overall, the studies in vitro found that Sox1 null neural progenitors both from ganglionic eminence and dorsal telencephalic wall showed a consistent deficiency in neuronal differentiation at several developmental stages. This deficiency was unlikely to be due simply to premature/accelerated or delayed neuronal differentiation. However, progenitors from ganglionic eminence and dorsal telencephalic wall do show different patterns of differentiation, and Sox1 null progenitors from the two different regions also behaved somewhat differently, which suggests context dependent roles of Sox1 in neuronal differentiation. To determine whether some subpopulations of neurons were specifically affected, we counted the numbers of GluR2/3 and GAD65 immunopositive cells after 14 days of differentiation on PDL/Laminin coated coverslips. Sox1 null progenitors gave rise to significantly fewer GAD65+ cells compared to WT cells(P<0.01, Fig. 3D), whereas the numbers of GluR2/3+ cells was comparable in both groups (Fig. 3D) This suggests that Sox1 plays different roles in the development of different neuronal subpopulations, and specifically that it is involved in the differentiation of GAD65-expressing neurons, an observation consistent with findings in Sox1 null brains in vivo (Ekonomou et al., 2005; Malas et al., 2003).
Since Sox1 promoted neuronal cell fate determination and differentiation of neural progenitor cells in vitro (Kan et al., 2004, and Fig. 3), we sought to determine whether Sox1 exerts similar effects in vivo. We therefore overexpressed the transcription factor in cells in the developing mouse telencephalon by microinjecting pCLE-Sox1 or control retrovirus (each containing an IRES2-eGFP sequence) into the telencephalic vesicle of E10.75 embryos using ultrasound guidance. The promoter assembly in the pCLE vector is optimized such that expression of Sox1-IRES2-eGFP (or control IRES2-eGFP) is constitutively maintained in both neural progenitor and differentiated cell populations(Gaiano et al., 1999). After 4 days of normal growth and development of the embryos in utero, the brains were removed and coronal sections (10μm) were stained for GFP and NeuN, and were counterstained with Hoechst. NeuN expression was used as a marker of neuronal lineage commitment, and GFP was used to trace the Sox1 or control infected progeny (Supplementary Fig. 2). For control and Sox1 injected embryos, a total of 29 and 35 different microscopic fields, respectively, from similar regions of the dorsolateral telencephalon from different embryos were used for cell counting (Fig. 4A). In control animals, on average 41.3% of the progeny of infected cells (GFP+) also expressed NeuN. Overexpression of Sox1 in stem/progenitor cells significantly increased the average proportion of neurons generated in the dorsolateral wall of telencephalon to 56.4% of the progeny (p<0.00001) (Fig. 4B). Study of the retroviral infected cell clusters from other brain regions and later time points after injection revealed the same Sox1-mediated bias towards neuronal differentiation (data not shown).
Previous reports indicated that SoxB1 factors may play key roles in the maintenance of the progenitor state. However, our studies in vitro (Figs. 1 and and3)3) indicated that Sox1 is not absolutely necessary for neurosphere formation. To further address this question in vivo, we repeated the injections at E10.75 and allowed the embryos to develop in utero for 4 days. We then compared the cluster sizes (number of GFP+ cells per cluster) of the control and pCLE-Sox1 injected embryos by staining thick coronal brain sections (100μm) with GFP antibody, and then counting all GFP+ cells contributing to a clonal cluster from all optical sections of similar regions of the dorsolateral telencephalon. We chose 100μm thick sections in order to maximize the likelihood of obtaining complete clusters, and clusters which were obviously incomplete or without well-defined borders were therefore excluded from counting. Sox1 overexpressing clusters were on average larger than control clusters (Figure 4C), and the average theoretical number of cell divisions for Sox1 overexpressing clusters (6.14) was significantly larger than that of the control clusters (5.10, P<0.05). The apparent one cell cycle difference between control and Sox1 overexpressing cells (Fig. 4D) indicated that constitutive overexpression of Sox1 causes moderate expansion of the progenitor pool in vivo. Overall, the gain-of-function study suggests that Sox1 expression initially helps to maintain progenitor cells in cell cycle and moderately increases their numbers. As these progenitor cells differentiate over the experimental time period, they not only produce a higher number of neurons (in absolute numbers as expected), but they also produce a higher percentage of neurons (as a proportion of total GFP+ cells). This higher percentage of neurons is not expected if we believe Sox1 overexpression to play no role in neuronal differentiation of neural progenitors. To the contrary, our in utero gain of function studies indicate that continued overexpression of Sox1 promotes neuronal lineage commitment as compared to control.
Since little is known about the overall endogenous expression pattern and function of Sox1 in the postnatal and adult CNS, we used heterozygous mice with the β-galactosidase gene knocked into the Sox1 locus (Ekonomou et al., 2005) (Sox1lacz/+) to study the postnatal pattern of expression of the Sox1 gene. These mice have no detectable brain phenotype and provide a powerful and specific approach since they eliminate the potential immunohistochemistry or in situ hybridization problems of cross reactions with other members of the highly homologous Sox gene family. At postnatal day 0 (P0), LacZ+ cells were detected in most brain regions, but high levels of expression were found only in the region of the VZ/SVZ (Fig. 5A and B), the rostral migratory stream, and the olfactory bulb (Fig. 5A). In contrast, in the P0 spinal cord, most cells were negative for LacZ staining and only a small population of LacZ+ cells was detected in ventral spinal cord and around the central canal (Fig. 5C).
In the adult mouse brain (3 months old), LacZ was expressed in a more restricted pattern although many LacZ+ cells were present in both differentiated and neurogenic regions (Fig. 5E–K). Specifically, LacZ staining was intense in a number of specific ventral brain regions including ventral striatum, nucleus accumbens, ventral caudato-putamen, olfactory tubercle, islands of Calleja, intercalated nuclei of the amygdala, and dorsomedial hypothalamic nuclei. Other specific regions that contained many positive cells included lateral septum, lambdoid septal zone, and some dorsal nuclei of the forebrain such as the lateral habenular nuclei (Fig. 5E–K). Specific staining was also found in the hippocampus (CA1, CA2, and CA3), and sparse positive cells were found within the neocortex (Fig. 5J). In contrast, in the adult spinal cord only a few LacZ+ cells were detected around the central canal (Fig. 5D).
The widespread expression of LacZ in multiple differentiated regions of postnatal and adult brain was surprising in view of previous reports suggesting that its expression is downregulated in neural progenitor cells as they exit cell cycle and terminally differentiate (Pevny et al., 1998). To confirm this finding, we first performed double antibody staining for Sox1 and β-Gal in adult brain sections and found that Sox1 and β-Gal co-localized to the same cells with predominantly nuclear staining for Sox1 staining and cytoplasmic staining for β-Gal (Fig. 6A–C). By contrast, this staining was not present in Sox1 null animals (data not shown), and the antibody recognized a single band of right size on Western blot analysis (Fig. 6D). Furthermore, this band is different in size from the band that recognized by Sox2 or Sox3 antibody (Fig. 2 and data not shown), so it is unlikely that Sox1 antibody is cross-reacting with other SoxB1 factors. These findings reinforced the validity of the LacZ staining and also indicated that the Sox1 antibody we used is specific and sensitive enough to study endogenous Sox1 expression. We then used this antibody to stain or double-stain brain and spinal cord sections of wildtype mice to further characterize the endogenous expression pattern of Sox1. At P0, high levels of Sox1 protein were found in the VZ/SVZ as expected (Fig. 6E) with lower levels of expression in other brain regions. Two and four weeks after birth, Sox1 immunopositive cells were still found in neurogenic region, such as the subgranular cell layer in the hippocampal dentate gyrus (DG) and the subependymal regions of the lateral ventricles, third ventricle, cerebral aqueduct, and fourth ventricle of both the mouse and rat brain (Fig. 6F and data not shown). Sox1 immunopositive cells in the mouse brain were also found in differentiated regions similar to those that contained LacZ+ cells (Figs. 6G). Sox1+ cells were also observed in the rostral migratory stream and the olfactory bulb (data not shown). Similar results were observed in rats (Sprague Dawley) (data not shown). Consistent with the LacZ staining, only a few Sox1+ cells were detected in the adult spinal cord around the central canal (Figs. 6F). In situ hybridization of adult brain sections with a 50-mer anti-sense oligonucleotide probe complementary to the 3′-UTR of Sox1 (which has no significant homology to any other SoxB1 factors) found that Sox1 mRNA was abundantly expressed both in differentiated and neurogenic regions of the brain in a pattern similar to the one observed by immunohistochemistry (Supplementary Fig. 3).
To identify the lineage of the Sox1+ cells, we double-labeled adult brain sections with Sox1 antibody and different cell lineage markers. Most Sox1+ cells in non-neurogenic regions were co-labeled with the neuronal marker, NeuN, which labels most mature neurons (Fig. 7B, upper panel). In the VZ/SVZ, Sox1+ cells were not co-labeled with NeuN (Fig. 7A) but rather were BrdU+. By contrast, in the hippocampus the subgranular and the granular layer cells of the dentate gyrus were co-labeled with Sox1 and NeuN (data not shown). Interestingly, there were cells in the cerebellar Purkinje cell layer that were immunoreactive for Sox1 but negative for NeuN staining (Fig. 7B, lower panel). This is consistent with a recent report that indicates that some Sox1+ cells are Bergmann glia (Sottile et al., 2006), a specific progenitor/stem cell population present in the Purkinje cell layer. More importantly, there was no overlap between Sox1 and the oligodendrocyte marker, CNPase, anywhere in the brain. Sox1 co-localized with glial fibrillary acidic protein (GFAP), which labels progenitor cells in neurogenic regions and astrocytes in differentiated regions, only in a small population of subgranular layer cells in the dentate gyrus (presumably neural progenitors). No co-localization of Sox1 with GFAP was found in differentiated regions (Fig. 7C).
To determine the subtypes of neurons that express Sox1, we performed additional double-labeling with Sox1 and different subtype-specific neuronal markers. GAD65, a marker of inhibitory neurons, co-labeled with Sox1 in ventral brain regions (Fig. 8A). We also found that Sox1+ cells co-labeled extensively with GluR2/3, a marker of excitatory neurons, in multiple different brain regions, including the neocortex (Fig. 8B, upper panel) and the hippocampus (Fig. 8B, lower panel). Sox1 expression was predominantly nuclear whereas GluR2/3 expression was predominately cytoplasmic.
Sox1 and the other SoxB1 family members, Sox2 and Sox3, are expressed by neural progenitor cells in the developing CNS (Bylund et al., 2003; Collignon et al., 1996; Graham et al., 2003) and in the subgranular zone (SGZ) in the adult brain (Aubert et al., 2003; Limke et al., 2003). Sox1 has been considered to be a pan-neural or universal stem cell/progenitor cell marker (Pevny and Rao, 2003), and the prevailing view is that SoxB1 factors (including Sox1) function to maintain the progenitor cell state in the brain. However, evidence is accumulating that SoxB1 factors may also exert other effects in the mammalian brain. We reported that overexpression of Sox1 biases mouse progenitor cells towards neuronal lineage commitment and differentiation in vitro (Kan et al., 2004). Analysis of Sox1 null mice found that Sox1 is essential for migration and terminal differentiation of specific subpopulations of neurons (Ekonomou et al., 2005; Malas et al., 2003). Furthermore, Ferri et al (Ferri et al., 2004) found that compound heterozygotes (Sox2β-geo/ΔENH, which express only about 25% of WT amount of Sox2) show multiple CNS abnormalities including decreased generation of new neurons in adult neurogenic regions and age-related neuronal degeneration. These findings cannot easily be explained by the hypothesis that SoxB1 factors act solely to maintain the progenitor cell state.
To reconcile these conflicting views, we hypothesized that each SoxB1 gene plays unique as well as overlapping roles in the CNS in a context dependent manner, i.e. they are interchangeable in some conditions but not in others. More specifically, we hypothesized that Sox1 serves a unique role in neuronal differentiation in addition to the overlapping role with other SoxB1 family members in the proliferation of neural progenitors. To test this hypothesis, we first performed a quantitative loss-of-function study examining neurosphere formation and lineage commitment by cultured Sox1 null stem/progenitor cells. We found that Sox1 null progenitor cells formed neurospheres normally, but that Sox1 null neurospheres were specifically deficient in neuronal lineage differentiation. We further tested the hypothesis with a gain-of-function study in embryonic telencephalic progenitor cells in vivo, since there is no report of a gain-of-function study in mouse telencephalon. We found that overexpression of Sox1 in embryonic telencephalon appeared both to moderately expand the progenitor pool and to bias neural progenitors towards neuronal commitment. Immunohistochemistry and in situ hybridization studies provided additional supporting evidence for the hypothesis that Sox1 serves roles in neuron specification and function in addition to its role in progenitor cell proliferation. Sox1 continues to be expressed abundantly throughout life by neurons in specific brain regions including the neocortex and hippocampus. By contrast, Sox1 expression is not detectable in either astrocytes or oligodendroglia (Milosevic and Goldman, 2002, and present study). This finding is consistent with the hypothesis that Sox1 plays key roles in neuronal development and function.
Our loss-of-function studies suggest that Sox1 is not absolutely necessary for maintenance of a proliferative progenitor cell state. However, the gain-of-function studies in vivo indicated that constitutive expression of Sox1 is sufficient to modestly expand the progenitor pool at a rate consistent with one extra round of cell cycle. Interestingly, our previous in vitro study also found that constitutive overexpression of Sox1 in neural progenitors appeared to cause one extra cell division (Kan et al., 2004). One potential mechanism for the moderate expansion of the progenitor pool is through shifts in the relative tempo of symmetric to asymmetric cell divisions.
Our in vitro differentiation assay provides evidence that neural progenitors from different brain regions show different patterns of differentiation. More importantly, null mutation of Sox1 results in different deficiencies in progenitor cells from different regions. Specifically, ventral progenitors (ganglionic eminence) tended to remain as progenitor cells in the absence of Sox1 expression, whereas dorsal progenitors (dorsal telencephalic wall) tended to switch to a glial fate in the absence of Sox1 expression. These observations are consistent with the idea that Sox1 plays context dependent roles in neuronal differentiation.
Although our observations and those of Ekonomou et al (Ekonomou et al., 2005) both suggest a role for Sox1 in the differentiation and maturation of subpopulations of neurons, there are some differences in the findings. Our study found that the percentage of βIII-tubulin or MAP2 positive cells generated from Sox1lacz/lacz progenitor cells was significantly lower compared to WT littermates in all age groups (E12 through E16). By contrast, Ekonomou et al stained brain sections of E13 embryos with TuJ1, an immature neuron marker (Menezes and Luskin, 1994), and did not find significant differences in general neuronal lineage commitment between Sox1 mutant and WT embryos. This discrepancy may reflect differences in the methodologies used for the analysis of the phenotype since it is very likely that more cell populations, and more compensatory pathways and mechanisms, are involved to mask the phenotype in the whole brain in vivo compared to neurosphere cultures. Importantly, the previous studies by Ekonomou and coworkers found that Sox1 is required for the differentiation and/or migration of specific subpopulations of neurons, including Gad and preproenkephalin-expressing neurons (Ekonomou et al., 2005; Malas et al., 2003), and this specific defect was recapitulated in our culture system (Fig. 3C).
One important finding from our current expression study is that while Sox1 is widely expressed in postnatal and adult rodent brain, it is not widely expressed in the spinal cord. These markedly different patterns of expression, coupled with the fact that no major phenotype has been described in Sox1 null mouse spinal cord, suggest that findings with spinal cord progenitor cells cannot be generalized to telencephalic progenitor cells. Since the studies of Bylund et al (2003) analyzed avian spinal cord progenitor cells, species differences and/or regional differences could explain the conflicting findings regarding Sox1 function (Bylund et al., 2003; Malas et al., 2003).
Our current and previous observations (Kan et al., 2004), coupled with the findings of other investigators (Bylund et al., 2003; Graham et al., 2003; Pevny et al., 1998) thus suggest that Sox1 expression by progenitor cells initially helps to maintain the cells in cell cycle, but that continued expression subsequently promotes a neuronal phenotype. A similar dual role has been described for other factors such as β-catenin which promotes either reentry of progenitor cells into cell cycle or neuronal differentiation depending upon the presence or absence of signaling from other factors such as FGF2 or the age of the progenitor cells (Israsena et al., 2004).
Overall, the current report revealed a previously unidentified neuronal differentiation deficiency of Sox1 null progenitor/stem cells using neurosphere assays. More importantly, this study directly demonstrated, for the first time, the in vivo gain-of-function phenotypes in mouse telencephalon using viral overexpression of Sox1 in embryonic telencephalic neural progenitor cells. Additionally, our expression studies in postnatal and adult mice further clarified the expression pattern of Sox1 and provided additional supporting evidence for the general hypothesis that Sox1 is involved in neuronal differentiation. Our findings, coupled with recent studies of SoxB1 null mutants, suggest that Sox1, and probably other SoxB1 factors (Sox2 and Sox3), play multiple roles in the brain other than solely maintaining progenitor cell identity.
A) The secondary neurospheres from ganglionic eminence after 3 days in culture were pulsed with CldU for 6 hours. Then, the cells were washed and immediately plated on the PDL/Laminin coated coverslips in differentiation medium without growth factors for 3 days. Finally, cells were pulsed with IdU for 6 hours before being fixed and double-stained. The average percentages of CldU+ cells, IdU+ cells and CldU/IdU+ cells ± s.d. are shown. *Differs from WT by ANOVA at P < 0.05.
B) Typical images of CldU(red)/IdU(green) double-staining, counterstained with DAPI (blue).
Note that more IdU+ cells (green nuclei) are present in Sox1−/− culture.
A) NeuN is used as a marker of neuronal differentiation.
B) GFP expression is used to trace the infected progeny.
C) Hoechst staining of the nuclei facilitates cell counting.
D) Merged picture demonstrates neuronal differentiation of a subset of the infected progeny where GFP and NeuN are colocalized over the same cells (arrowheads). Arrows point to the GFP+ cells that were not differentiated into neurons (NeuN−) at the time the embryonic brain was harvested. Ventricular surface is towards the bottom of the panel.
A) Sox1 mRNA is expressed in hippocampus (CA1 is shown here).
B) Sox1 mRNA is expressed in the piriform region.
C) Sox1 mRNA is expressed in the cerebellar purkinje layer. Bar=40 μm
The authors are grateful to Dr. Yusuke Kamachi and Dr. Hisato Kondoh, Osaka University, Japan for their gift of Sox1 antibody, and to Dr. Mahmud Bani-Yaghoub, Institute for Biological Sciences, National Research Council of Canada for their gift of Sox2 antibody. We appreciate the help from other members of Kessler lab. We thank Dr. Enrico Mugnaini and Gabrielia Sekerkova for sharing their knowledge of the cerebellum. This study was supported by NIH grants NS34758 and NS20778. AJ and AB are supported by NS51962 and NS4817, respectively.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.