ABSTRACT
The spatiotemporal identity of neural progenitors and the regional control of neurogenesis are essential for the development of cerebral cortical architecture. Here, we report that mammalian DM domain factors (Dmrt) determine the identity of cerebral cortical progenitors. Among the Dmrt family genes expressed in the developing dorsal telencephalon, Dmrt3 and Dmrta2 show a medialhigh/laterallow expression gradient. Their simultaneous loss confers a ventral identity to dorsal progenitors, resulting in the ectopic expression of Gsx2 and massive production of GABAergic olfactory bulb interneurons in the dorsal telencephalon. Furthermore, double-mutant progenitors in the medial region exhibit upregulated Pax6 and more lateral characteristics. These ventral and lateral shifts in progenitor identity depend on Dmrt gene dosage. We also found that Dmrt factors bind to Gsx2 and Pax6 enhancers to suppress their expression. Our findings thus reveal that the graded expression of Dmrt factors provide positional information for progenitors by differentially repressing downstream genes in the developing cerebral cortex.
INTRODUCTION
During the development of the central nervous system, neural progenitors acquire regionally restricted properties that enforce a unique identity on their neuronal progeny. These properties, which determine progenitor identity, are involved not only in the determination of the fate of their progeny but also in the spatiotemporal patterns of cell cycle progression, division mode, and neurogenic ability (Martynoga et al., 2012; Paridaen and Huttner, 2014).
The telencephalon is the most anterior region of the developing central nervous system. It comprises the dorsal telencephalon (pallium) and ventral telencephalon (subpallium). The dorsal telencephalon gives rise to the cerebral cortex and the hippocampus by the spatial and temporal regulation of secreted inductive signals to provide positional information that defines progenitor identity. The initial step of cerebral cortical development requires reciprocal and/or independent action of extrinsic factors, such as Bmps, Fgfs, Shh and Wnts, which are secreted from signaling centers in the developing telencephalon (Hoch et al., 2009). In addition, many transcription factors have been identified as downstream targets of these extrinsic factors (Hébert and Fishell, 2008). These effectors in turn specify regional cell types and the neurogenic potential of progenitors. Among such factors, the transcription factors Pax6 and Gli3 play an essential role in the acquisition and maintenance of neural progenitor identity in the developing dorsal telencephalon (Grove et al., 1998; Kroll and O'Leary, 2005; Kuschel et al., 2003; Stoykova et al., 2000; Theil et al., 1999; Tole et al., 2000b; Toresson et al., 2000; Yun et al., 2001). In mice lacking Pax6 or Gli3, the dorsal-ventral (D-V) boundary is severely disorganized, resulting in the ectopic expression of several ventral-specific transcription factors, such as Gsx2 (also called Gsh2), Ascl1 (also called Mash1), and Dlxs, which are involved in the specification of all GABAergic interneurons in the ventral telencephalon (Long et al., 2009; Wang et al., 2013). However, the additional knockout of Shh in mice lacking Pax6 or Gli3 can partly rescue the disorganization of the D-V boundary (Aoto et al., 2002; Fuccillo et al., 2006; Rallu et al., 2002; Rash and Grove, 2007), raising the possibility that there are unknown factors that are required for the determination of the identity of cerebral cortex progenitors.
The neurogenic ability of neural progenitors is also dynamically controlled by spatial and temporal information during cortical development. At the early stages of cortical development, neural progenitors are proliferative, undergoing symmetric cell divisions. In contrast, at later stages, progenitors gradually mature to generate cortical neurons. The spatiotemporal control of the transition of progenitors from a less to a more neurogenic state provides a basis for establishing the complex architecture of the cerebral cortex. D-V patterning factors, such as Shh and Pax6, play central roles in regulating the neurogenic ability of progenitors in the developing cerebral cortex. These factors regulate neurogenesis in part by cooperating with the Notch-Delta signaling pathway, which promotes self-renewal of neural progenitors (Dave et al., 2011; Sansom et al., 2009). However, it remains unclear how the neurogenic ability of cortical neural progenitors are spatiotemporally controlled during development.
Here, we report that three mammalian DM domain (Dmrt) factors, Dmrt3, Dmrta1 (also called Dmrt4) and Dmrta2 (also called Dmrt5) determine the identity of cerebral cortical progenitors and regulate the neurogenic ability of these cells in developing mouse embryos. We provide evidence that low levels of Dmrt factors repress Gsx2 expression in the dorsal telencephalon, thereby defining the pallial-subpallial boundary and specifying neural progenitors that produce glutamatergic neurons. Moreover, a graded expression of Dmrt factors suppresses Pax6 transcription and establishes a mediallow/lateralhigh Pax6 expression gradient. In this way, a neurogenic gradient is formed in the developing cerebral cortex.
RESULTS
Dmrt factors maintain the dorsal-ventral patterning of the telencephalon
We first examined the precise expression patterns of Dmrt3 and Dmrta2, which are the predominant Dmrt factors in the developing dorsal telencephalon of rodents (Fig. S1A-C). Whole-mount immunofluorescence was performed on embryonic day (E) 9.5 mouse embryos, in which the dorsal-ventral patterning of the telencephalon had just been established. In these embryos, Dmrt3 and Dmrta2 were detected predominantly in the telencephalon (Fig. S1D). At E12.5, Dmrt3 and Dmrta2 were detected in a restricted pattern in the dorsal telencephalon, with a medialhigh/laterallow gradient (Fig. S1E). Quantitative RT-PCR (qRT-PCR) analysis confirmed that Dmrt3 and Dmrta2 transcripts were predominantly expressed in the medial telencephalon, and no differences in their expression were observed along the anterior-posterior axis (Fig. S1F-H).
We and others have previously reported that Dmrta2 single mutants exhibit a reduction in size of the dorsal telencephalon, with the complete loss of medial regions, such as the cortical hem (Konno et al., 2012; Saulnier et al., 2013). To test whether the Dmrt factors have overlapping roles, we fist examined Dmrt3 and Dmrta2 double mutants. Dmrt3−/−;Dmrta2−/− mutant mice died soon after birth. At E15.5, the telencephalon of Dmrt3−/−;Dmrta2−/− embryos appeared smaller in size compared with control embryos (Fig. 1A, Fig. S1I). Immunostaining for Gad2, an enzyme involved in GABA synthesis, revealed ectopic production of GABAergic neurons in the dorsal telencephalon of Dmrt3−/−;Dmrta2−/− mutant embryos (Fig. 1B, Fig. S1J). Concomitantly, the production of cortical excitatory neurons, labeled with Tbr1 (cortical deep layer neurons) or Pou3f2 (also called Brn-2; cortical upper layer neurons), was markedly decreased in the dorsal telencephalon of mutant embryos (Fig. 1B, Fig. S1J). These results indicate that the Dmrt factors are involved in determining the identity of neurons generated in the dorsal telencephalon.
Disorganization of cerebral cortical development in Dmrt mutant embryos. (A) Lateral views (left) and top views of the forebrain (right) of E15.5 Dmrt3/Dmrta2 double-mutant embryos. (B) Immunofluorescence for Gad2, Tbr1 and Pou3f2 in coronal sections of the telencephalon in Dmrt3/Dmrta2 mutant E15.5 embryos. (C) Lateral views of the head region in Dmrt3/Dmrta2 mutant E12.5 embryos. (D) Immunofluorescence for Pax6 and Gsx2 in coronal sections of the telencephalon of Dmrt3/Dmrta2 mutant E12.5 embryos. (E) Schematic of the transcriptional cascade giving rise to distinct neuronal subtypes in the nascent ventral telencephalon. (F) Immunofluorescence for Sp8 and Isl1 in coronal sections of the telencephalon of Dmrt3/Dmrta2 mutant E12.5 embryos. In D and F, the right panels represent higher magnifications of the boxed regions indicated in the left panels. D, dorsal; GE, ganglionic eminence; L, lateral; LGE, lateral ganglionic eminence; MGE, medial ganglionic eminence; V, ventricle. In all immunofluorescence images, the yellow dotted lines indicate the ventricular surface. Scale bars: 100 µm.
Disorganization of cerebral cortical development in Dmrt mutant embryos. (A) Lateral views (left) and top views of the forebrain (right) of E15.5 Dmrt3/Dmrta2 double-mutant embryos. (B) Immunofluorescence for Gad2, Tbr1 and Pou3f2 in coronal sections of the telencephalon in Dmrt3/Dmrta2 mutant E15.5 embryos. (C) Lateral views of the head region in Dmrt3/Dmrta2 mutant E12.5 embryos. (D) Immunofluorescence for Pax6 and Gsx2 in coronal sections of the telencephalon of Dmrt3/Dmrta2 mutant E12.5 embryos. (E) Schematic of the transcriptional cascade giving rise to distinct neuronal subtypes in the nascent ventral telencephalon. (F) Immunofluorescence for Sp8 and Isl1 in coronal sections of the telencephalon of Dmrt3/Dmrta2 mutant E12.5 embryos. In D and F, the right panels represent higher magnifications of the boxed regions indicated in the left panels. D, dorsal; GE, ganglionic eminence; L, lateral; LGE, lateral ganglionic eminence; MGE, medial ganglionic eminence; V, ventricle. In all immunofluorescence images, the yellow dotted lines indicate the ventricular surface. Scale bars: 100 µm.
We next asked whether the D-V patterning in the mutant telencephalon was affected at the progenitor level. Dmrt3−/−;Dmrta2−/− E12.5 embryos showed a marked reduction in size of the telencephalon, with no prominent defects in other CNS regions (Fig. 1C) (data not shown). The expression of the transcriptional factor Gsx2 is restricted in ventral telencephalic progenitors (subpallium) in wild-type embryos. However, this factor was observed ectopically in the dorsal telencephalon of double-mutant embryos, suggesting a ventral shift in cortical progenitor fate in the mutant embryos. Consistent with this finding, transcription factors involved in generating cortical excitatory neurons, such as Emx1, Neurog2 and Tbr2 (Eomes), exhibited markedly decreased expression in the mutant dorsal telencephalon (Fig. S2A,B). Interestingly, the expression of Pax6, a transcription factor involved in the development of the dorsal telencephalon, was maintained in the mutant embryos, resulting in the simultaneous expression of Gsx2 and Pax6 in these neural progenitors (Fig. 1D, Fig. S1K).
In wild-type embryos, Gsx2 and Pax6 are co-expressed in a small subset of neural progenitors in the dorsal-most aspect of the ventral telencephalon, which comprises the pallial-subpallial boundary (PSB, or the boundary between the dorsal and ventral telencephalon) and the dorsal lateral ganglionic eminence (dLGE) (Corbin et al., 2003; Toresson et al., 2000; Yun et al., 2001). This region gives rise to olfactory bulb interneurons by activating the expression of the transcription factor Sp8 (Fig. 1E) (Waclaw et al., 2006). This Gsx2-Pax6 double-positive region is adjacent to the ventral LGE (vLGE), which expresses Gsx2 but not Pax6. The vLGE generates striatal neurons that express Isl1 (Ehrman et al., 2013), another downstream transcription factor known to be regulated by Gsx2. To identify subtypes of neurons generated in the mutant dorsal telencephalon that express both Gsx2 and Pax6 in progenitors, we examined the expression of Sp8 and Isl1 in the double-mutant embryos, and found that they ectopically express Sp8 but not Isl1 in the dorsal telencephalon (Fig. 1F, Fig. S1L), suggesting that the simultaneous depletion of Dmrt3 and Dmrta2 expands the PSB/dLGE more dorsally by converting the identity of the cortical progenitors to that of the PSB and/or dLGE. This result was confirmed by the persistent expression of Pax6, a marker of migrating olfactory bulb (OB) interneurons, in the cells that migrate anteriorly (Fig. S3A). Thus, the expression of Dmrt genes appears to be involved in determining the position and size of PSB/dLGE, the region between the dorsal and ventral telencephalon (pallium and subpallium) along the dorsoventral axis.
Gene dosage-dependent suppression of OB neurons by Dmrt factors
As the expression of Dmrta2 and Dmrt3 exhibits a graded pattern along the lateral-dorsal-medial axis, our results regarding the role of Dmrts on the determination of differential progenitor identity suggest that Dmrt gene dosage affects the position and the size of PSB/dLGE. If so, then Dmrt dosage would consequently alter the relative number different neuronal subtypes. We tested this possibility by examining the proportion of OB projection neurons (OB-p) and OB interneurons (OB-i), which are produced from the dorsal and ventral domains adjacent to the PSB, respectively (Ceci et al., 2012; Wichterle et al., 1999), in mice with various dosage combinations of Dmrt genes. These neurons migrate anteriorly to form the olfactory cortex and are distinguishable by several markers. Specifically, subsets of OB-i are characterized by tyrosine hydroxylase expression, whereas OB-p neurons can be identified by calretinin (calbindin 2) expression (Fig. 2A,B) (Waclaw et al., 2006). In Dmrt3+/−;Dmrta2+/− control embryos, the OB-i and OB-p domains formed adequately in the anterior border of the telencephalon at E12.5, whereas Dmrt3−/−;Dmrta2+/− or Dmrta2−/− embryos exhibited a slight expansion of the OB-i domain (Fig. 2C-E). Interestingly, a more significant expansion of the OB-i was seen in Dmrt3+/−;Dmrta2−/− embryos, and the most striking effect was observed in Dmrt3−/−;Dmrta2−/− embryos. This effect occurred in association with no significant expansion of the OB-p domain (calretinin positive) and the gradual decrease of the cerebral cortical domain (calretinin/TH double negative) (Fig. 2F,G). These observations are consistent with Dmrt factor dosage determining the relative proportions of the primordium of the OB and the cerebral cortex (see also Fig. S3A,B).
Gene dosage-dependent suppression of OB neurons by Dmrt factors. (A) Schematic of the production of the OB projection and interneurons from the telencephalic regions near the PSB at E12.5. Section images of Hematoxylin and Eosin staining are from the public database EPMBA (http://www.epmba.org). A, anterior; D, dorsal; P, posterior; V, ventral. (B) Detection of early OB neuron production with calretinin and tyrosine hydroxylase (TH) in a sagittal section of the developing mouse telencephalon at E12.5. (C) Immunofluorescence for calretinin and TH in sagittal sections of the telencephalon from the various combinations of mutants for Dmrt3 and Dmrta2 at E12.5. Dashed line encircles the TH-positive domain. (D-G) Quantification of the distance from the base of the nasal process to the distal end of the TH-positive domain (yellow line; E), the size of the calretinin-positive domain (magenta line; F), and the size of calretinin/TH-double negative (cortical) domain (light blue line; G) normalized to the length of the dorsal ventricular surface (red line). The panel in D displays how the distance and the size of each domain were measured. Numbers below the graphs in E-G correspond to the numbers next to the genotypes in C. The error bars represent ±s.d. [n=4 per group (samples #1 and 3), n=3 per group (samples #2, #4, #5, #6)]. Statistical significance was determined using Student's t-test with Welch's correction (ns, not significant; *P<0.05; **P<0.01; ***P<0.001). Scale bars: 100 µm.
Gene dosage-dependent suppression of OB neurons by Dmrt factors. (A) Schematic of the production of the OB projection and interneurons from the telencephalic regions near the PSB at E12.5. Section images of Hematoxylin and Eosin staining are from the public database EPMBA (http://www.epmba.org). A, anterior; D, dorsal; P, posterior; V, ventral. (B) Detection of early OB neuron production with calretinin and tyrosine hydroxylase (TH) in a sagittal section of the developing mouse telencephalon at E12.5. (C) Immunofluorescence for calretinin and TH in sagittal sections of the telencephalon from the various combinations of mutants for Dmrt3 and Dmrta2 at E12.5. Dashed line encircles the TH-positive domain. (D-G) Quantification of the distance from the base of the nasal process to the distal end of the TH-positive domain (yellow line; E), the size of the calretinin-positive domain (magenta line; F), and the size of calretinin/TH-double negative (cortical) domain (light blue line; G) normalized to the length of the dorsal ventricular surface (red line). The panel in D displays how the distance and the size of each domain were measured. Numbers below the graphs in E-G correspond to the numbers next to the genotypes in C. The error bars represent ±s.d. [n=4 per group (samples #1 and 3), n=3 per group (samples #2, #4, #5, #6)]. Statistical significance was determined using Student's t-test with Welch's correction (ns, not significant; *P<0.05; **P<0.01; ***P<0.001). Scale bars: 100 µm.
We next addressed the role of the other Dmrt factor, Dmrta1, which is expressed nearly uniformly in the anterior dorsal telencephalon and in a mediallow/lateralhigh gradient in the posterior dorsal telencephalon (Fig. S4A-C). Although Dmrta1−/− mutant embryos exhibited no clear abnormalities (Fig. S4D-F), knocking out Dmrta1 in Dmrta2−/− mice resulted in a marked upregulation of Gsx2 expression in the dorsal telencephalon at E12.5, resembling the phenotype observed in Dmrt3−/−;Dmrta2−/− embryos (Fig. S4G). Dmrt3−/−;Dmrta2−/−;Dmrta1−/− embryos also exhibited ectopic expression of Gsx2 similar to those observed in Dmrta1−/−;Dmrta2−/− or Dmrt3−/−;Dmrta2−/− embryos (Fig. S4H). These results suggest that all Dmrt ‘A’ factors work along the same molecular pathway in the developing cerebral cortex.
Cell-autonomous suppression of ventral cell fate by Dmrt factors
We next examined whether and how the loss of Dmrt factor function affects neighboring cells. To this end, we generated chimeric mutant embryos by injecting Dmrt3−/−;Dmrta2−/−;Dmrta1−/− (TKO) embryonic stem (ES) cells into wild-type blastocysts. Whereas Dmrt3−/−;Dmrta2−/− embryos showed a slight increase in the expression of Gsx2 at the most lateral portion of the dorsal telencephalon at E10.5, TKO chimeric mutant embryos showed a strong expression of Gsx2 over the entire region of the dorsal telencephalon (Fig. 3A). Ectopic expression of Sp8 was also observed to a greater extent in TKO chimeric embryos than in double knockout (DKO) embryos (Fig. 3B). Notably, ectopic Gsx2 expression in the dorsal telencephalon of TKO chimeric embryos was observed only in cells lacking Dmrt factors (Fig. 3A), indicating their cell-autonomous function.
Cell-autonomous suppression of the ventral cell fate by Dmrt. (A) Immunofluorescence for Gsx2, Pax6 and Dmrta2 in coronal sections of the E10.5 mouse telencephalon. Sections are from Dmrt3/Dmrta2 double-mutant embryos (top and middle) or Dmrt3/Dmrta2/Dmrta1 triple-mutant embryos (bottom) established by blastocyst injection of Dmrt3/Dmrta1/Dmrta2 TKO ES cells to generate chimeras. (B) Immunofluorescence for Sp8, Dlx2 and Dmrta2 in coronal sections of the E10.5 mouse telencephalon of Dmrt3/Dmrta2 double-mutant embryos (top and middle) or Dmrt3/Dmrta2/Dmrta1 triple-mutant embryos (chimera, bottom). The yellow dotted lines indicate the ventricular surface. The red lines indicate Dmrta2-positive wild-type-derived cells in the TKO chimeric embryo. Scale bars: 100 µm.
Cell-autonomous suppression of the ventral cell fate by Dmrt. (A) Immunofluorescence for Gsx2, Pax6 and Dmrta2 in coronal sections of the E10.5 mouse telencephalon. Sections are from Dmrt3/Dmrta2 double-mutant embryos (top and middle) or Dmrt3/Dmrta2/Dmrta1 triple-mutant embryos (bottom) established by blastocyst injection of Dmrt3/Dmrta1/Dmrta2 TKO ES cells to generate chimeras. (B) Immunofluorescence for Sp8, Dlx2 and Dmrta2 in coronal sections of the E10.5 mouse telencephalon of Dmrt3/Dmrta2 double-mutant embryos (top and middle) or Dmrt3/Dmrta2/Dmrta1 triple-mutant embryos (chimera, bottom). The yellow dotted lines indicate the ventricular surface. The red lines indicate Dmrta2-positive wild-type-derived cells in the TKO chimeric embryo. Scale bars: 100 µm.
Dmrt factors repress the expression of several ventral genes
Given that all Dmrt factors contain an intertwined zinc finger-like DNA-binding module, the DM domain (Fig. S1A,B), we next sought to reveal the transcriptional network that is regulated by Dmrt3 and Dmrta2 (Fig. 4A). Transcriptomic analysis of the Dmrt3−/−;Dmrta2−/− dorsal telencephalon at E12.5 showed significant upregulation of many genes involved in differentiation and patterning of neural progenitors in the ventral telencephalon (Fig. 4B). This result is consistent with the ventral conversion of dorsal (cortical) progenitors and the ectopic production of GABAergic neurons described above. Dmrt3+/−;Dmrta2−/− embryos also showed a similar phenotype but to a lesser extent than Dmrt3−/−;Dmrta2−/− embryos, indicating that Dmrt function is also dosage dependent at the transcriptome level (Fig. 4B). As a first step to identify direct targets of Dmrt factors, we compared the effects of constitutive Dmrt factor knockout (Fig. 4B) with those of acute loss of function by introducing siRNAs via in utero electroporation into the dorsolateral telencephalon (Fig. 4C, Fig. S5A,B). Intriguingly, among the top 30 upregulated genes, only Gsx2 and its protein product were markedly upregulated in the cells electroporated with siRNAs targeting Dmrta2 and Dmrt3 (or Dmrta1) (hereafter termed DKD cells) (Fig. 4D,E, Fig. S6A,B). By contrast, the expression of other genes, such as Dlx1, a transcription factor controlling differentiation of GABAergic neurons (Anderson et al., 1999), Neurog2 and Ascl1 was changed in DKD cells (Dlx1: 3.8-fold increase, Ascl1: 1.5-fold increase, Pax6: 1.3-fold increase, and Neurog2: 1.1-fold decrease, versus control); however, the rate of change was over 18-fold lower than that for Gsx2 (68.8-fold increase versus control) (Fig. 4D,E). This finding raises the possibility that Gsx2 is a downstream target of Dmrt factors. An exogenously expressed fusion protein of the DM domain and the VP16 transcriptional activation domain also induced ectopic expression of Gsx2 and its protein product in the electroporated cells (Fig. S7A-C), suggesting that Dmrt factors function as transcriptional repressors for Gsx2 expression.
Repression of telencephalic ventral genes by Dmrt factors. (A) Schematic of the experimental design for gene expression profiling of Dmrt mutant brains. (B) Transcriptome analysis in the dorsolateral telencephalon of Dmrt double-mutant E12.5 embryos. The heat-maps reflect normalized gene expression. Genes that have been reported as ventral and dorsal specific are indicated in orange and green, respectively. (C) Schematic summarizing the experimental design for histological (D) and gene expression (E) analyses based on siRNA treatment of developing mouse embryos. siRNAs were electroporated with the EGFP-NLS expression vector in the dorsolateral cortex at E11.5. (D) Immunofluorescence for Gsx2 and Pax6 and EGFP fluorescence in coronal sections of the E15.5 mouse telencephalon electroporated with control siRNAs (top) or siRNAs targeting Dmrt3 and Dmrta2 (bottom). The right panels represent higher magnifications of the boxed regions indicated in the left panels. The yellow dotted lines indicate the ventricular surface. Scale bars: 100 µm. (E) Gene expression of Gsx2, Dlx1, Pax6 and Neurog2 normalized to Sox2 in the control (lane 1) or double-knockdown (D3/A2, lane 2) cells, as determined by qPCR. The E13.5 dorsal and ventral telencephalon [dTel (lane 3) and vTel (lane 4), respectively] were used as references. The error bars represent ±s.d. (n=4 per group). Statistical significance was determined using Student's t-test with Welch's correction (*P<0.05; **P<0.01; ***P<0.001). D, dorsal; EP, electroporation; IHC, immunohistochemistry; L, lateral.
Repression of telencephalic ventral genes by Dmrt factors. (A) Schematic of the experimental design for gene expression profiling of Dmrt mutant brains. (B) Transcriptome analysis in the dorsolateral telencephalon of Dmrt double-mutant E12.5 embryos. The heat-maps reflect normalized gene expression. Genes that have been reported as ventral and dorsal specific are indicated in orange and green, respectively. (C) Schematic summarizing the experimental design for histological (D) and gene expression (E) analyses based on siRNA treatment of developing mouse embryos. siRNAs were electroporated with the EGFP-NLS expression vector in the dorsolateral cortex at E11.5. (D) Immunofluorescence for Gsx2 and Pax6 and EGFP fluorescence in coronal sections of the E15.5 mouse telencephalon electroporated with control siRNAs (top) or siRNAs targeting Dmrt3 and Dmrta2 (bottom). The right panels represent higher magnifications of the boxed regions indicated in the left panels. The yellow dotted lines indicate the ventricular surface. Scale bars: 100 µm. (E) Gene expression of Gsx2, Dlx1, Pax6 and Neurog2 normalized to Sox2 in the control (lane 1) or double-knockdown (D3/A2, lane 2) cells, as determined by qPCR. The E13.5 dorsal and ventral telencephalon [dTel (lane 3) and vTel (lane 4), respectively] were used as references. The error bars represent ±s.d. (n=4 per group). Statistical significance was determined using Student's t-test with Welch's correction (*P<0.05; **P<0.01; ***P<0.001). D, dorsal; EP, electroporation; IHC, immunohistochemistry; L, lateral.
Given that sonic hedgehog (Shh) is a central mediator of D-V patterning at the onset of neural development (Dessaud et al., 2008), we tested for a functional interaction between Shh signaling and Dmrt. No significant changes in the expression levels of genes involved in the Shh signaling pathway, such as Gli1 and Ptch1, were observed in DKD cells (Fig. S8A). In addition, neither the presence nor the absence of Dmrta2 affected transcriptional activation by Gli2 or repression by Gli3R (a repressor form of Gli3) in the Shh-reporter assay (Fig. S8B). These results suggest that Dmrt3 and Dmrta2 act in the D-V patterning of the telencephalon independently of Shh signaling.
Dmrt factors inhibit neurogenesis in the medial telencephalon
The disorganization of the D-V boundary of the telencephalon is a striking phenotype in Dmrt DKO mutant embryos. However, Dmrt3 and Dmrta2 expression is actually much higher in the medial aspect of the telencephalon than dorsolaterally, suggesting that Dmrt3 and Dmrta2 may have additional roles in the medial cortex. Indeed, reports have demonstrated that Dmrta2 single-mutant embryos exhibit a loss of the medial cortex, including the cortical hem, with milder abnormalities in D-V patterning (Fig. 2) (Konno et al., 2012; Saulnier et al., 2013). We next asked which step in the development of the medial telencephalon was dependent on Dmrt factors. To this end, we again performed acute loss-of-function experiments with Dmrt gene siRNAs in the medial telencephalon by in utero electroporation (Fig. 5A). Cells electroporated with siRNAs targeting Dmrt3 and Dmrta2 (DKD cells) at E11.5 largely differentiated into neurons and rapidly disappeared from the ventricular zone at E13.5, whereas the majority of cells with control siRNAs remained in the ventricular zone as Sox2-positive progenitors (Fig. 5B,C). In this assay, none of the ventral genes we examined exhibited significant changes in expression (Fig. S9A) (data not shown) in DKD cells, suggesting that Dmrt factors have a unique role in the medial telencephalon compared with their functions in the lateral telencephalon. We then asked which genes were affected by the loss of function of Dmrt3 and Dmrta2. To this end, we performed transcriptomic analysis 24 h after electroporation, prior to the appearance of the neurogenic phenotype. Intriguingly, DKD cells exhibited the upregulation of dozens of genes (Fig. 5D). Among these upregulated genes, we focused on Pax6 given that it was expressed at the highest levels in our analysis and is an established promoter of neurogenesis. In DKD cells, Pax6 expression was markedly elevated (2.1-fold) compared with control cells (Fig. 5E). Several genes encoding transcription factors, such as Neurog2, Neurod1 and Neurod6, were also upregulated in DKD cells (Fig. 5E). These factors function directly or indirectly downstream of Pax6 (Sansom et al., 2009), consistent with the promotion of neurogenesis observed in DKD cells. Notably, both gene expression and histological analyses revealed that the additional knockdown of Pax6 expression in DKD cells (TKD cells) rescued the neurogenic phenotype observed in DKD cells (Fig. 5F) (Fig. S9B-E). Taken together with the graded expression of Dmrt3 and Dmrta2 in the developing telencephalon, our observations suggest that Dmrt factors regulate region-dependent neurogenic properties of neural progenitors by establishing a lateralhigh/mediallow gradient of Pax6 in the developing dorsal telencephalon.
Dmrt factors promote self-renewing ability in telencephalic progenitors. (A) Schematic summarizing the experimental design for histological (B,C) and gene expression (D) analyses in the medial telencephalon electroporated with siRNAs and the EGFP-NLS expression vector in the dorsolateral cortex at E11.5. (B) Immunofluorescence for Tubb3 and EGFP fluorescence in coronal sections of an E13.5 mouse telencephalon electroporated with control siRNAs (top) or siRNAs targeting Dmrt3 and Dmrta2 (middle), or siRNAs targeting Dmrt3, Dmrta2 and Pax6 (bottom). The yellow dotted lines indicate the ventricular surface. Scale bars: 100 µm. (C) The quantification of Sox2-positive neural progenitors in the electroporated cells at E13.5. The error bars represent ±s.d. (n=4 per group). (D) Transcriptome analysis of cells electroporated with two distinct sets of siRNAs targeting Dmrt (set1: Dmrt3_#2 and Dmrta2_#3, set2: Dmrt3_#1 and Dmrta2_#1) at E12.5. The heat-maps reflect normalized changes in gene expression that are upregulated following the knockdown of Dmrt genes. (E,F) Gene expression of Pax6 (E), Emx1 (F), Neurog2 (F), Neurod1 (F) and Neurod6 (F) normalized to Sox2 in the electroporated cells, as analyzed by qPCR. The error bars represent ±s.d. (n=6 per group in E and F). Statistical significance in C, E and F was determined using Student's t-test with Welch's correction (ns, not significant; **P<0.01; ***P<0.001; ****P<0.0001). D, dorsal; EP, electroporation; IHC, immunohistochemistry; L, lateral.
Dmrt factors promote self-renewing ability in telencephalic progenitors. (A) Schematic summarizing the experimental design for histological (B,C) and gene expression (D) analyses in the medial telencephalon electroporated with siRNAs and the EGFP-NLS expression vector in the dorsolateral cortex at E11.5. (B) Immunofluorescence for Tubb3 and EGFP fluorescence in coronal sections of an E13.5 mouse telencephalon electroporated with control siRNAs (top) or siRNAs targeting Dmrt3 and Dmrta2 (middle), or siRNAs targeting Dmrt3, Dmrta2 and Pax6 (bottom). The yellow dotted lines indicate the ventricular surface. Scale bars: 100 µm. (C) The quantification of Sox2-positive neural progenitors in the electroporated cells at E13.5. The error bars represent ±s.d. (n=4 per group). (D) Transcriptome analysis of cells electroporated with two distinct sets of siRNAs targeting Dmrt (set1: Dmrt3_#2 and Dmrta2_#3, set2: Dmrt3_#1 and Dmrta2_#1) at E12.5. The heat-maps reflect normalized changes in gene expression that are upregulated following the knockdown of Dmrt genes. (E,F) Gene expression of Pax6 (E), Emx1 (F), Neurog2 (F), Neurod1 (F) and Neurod6 (F) normalized to Sox2 in the electroporated cells, as analyzed by qPCR. The error bars represent ±s.d. (n=6 per group in E and F). Statistical significance in C, E and F was determined using Student's t-test with Welch's correction (ns, not significant; **P<0.01; ***P<0.001; ****P<0.0001). D, dorsal; EP, electroporation; IHC, immunohistochemistry; L, lateral.
Binding of Dmrt3 and Dmrta2 to Gsx2 and Pax6 enhancers
We next sought to identify Dmrt3- and Dmrta2-binding regions using whole-genome chromatin immunoprecipitation (ChIP)-sequencing and ChIP-qPCR analyses in tissue from the dorsal telencephalon of E12.5 mouse embryos. In these assays, we found that Dmrt3 and Dmrta2 bound to Gsx2 and Pax6 loci, and two binding regions were identified within ±100 kb from the transcription start site (TSS) of each locus (Fig. 6A-D). To determine the role of the sequences bound by Dmrt3 and Dmrta2, we performed ChIP-sequencing for histone modifications of histone H3 at lysines 4 and 27 using dorsal and ventral telencephalic samples from E12.5 wild-type mouse embryos. We found that Dmrt3- and Dmrta2-binding sites (DmrtBS) exhibited enrichment in lysine 4 monomethylation (H3K4me1) and lysine 27 acetylation (H3K27ac) on histone H3 (Fig. 6A,B). Given that nucleosomes in the vicinity of active enhancers typically contain both H3K4me1 and H3K27ac modifications (Shlyueva et al., 2014), we asked whether the Dmrt-binding sequences had enhancer activity for Gsx2 or Pax6 expression. To answer this, we generated transgenic mice in which EGFP expression was driven by the DmrtBS and Hsp68 (Hspa1) minimal promoter (Fig. 6E,F). We observed EGFP expression in dLGE cells of transgenic mice harboring the DmrtBS located 6 kb downstream of the transcription termination site (TTS) of Gsx2 (Fig. 6G). We also observed a lateralhigh/mediallow EGFP expression gradient in dorsal telencephalic cells in transgenic mice harboring the DmrtBS located 22 kb downstream of the Pax6 TTS site, which overlaps that of a reported Pax6 forebrain enhancer (Fig. 6H) (McBride et al., 2011; Mi et al., 2013b). These results suggest that Dmrt factors may contribute to shaping of the Gsx2 and Pax6 expression domains by binding to their enhancer sequences.
Dmrt binds to the enhancer sequences of Gsx2 and Pax6. (A,B) Distributions of ChIP-seq peaks obtained with antibodies for Dmrt3 (blue), Dmrta2 (blue), H3K4me1 (green), H3K4me3 (magenta), H3K27Ac (red), and H3K27me3 (light blue) at the Gsx2 (A) and Pax6 (B) loci in the E12.5 mouse telencephalon. The bars on the top with letters indicate the position of the primer sets used for ChIP-qPCR in C and D. (C,D) Quantification of Dmrt3/Dmrta2 binding at sequences within 100 kb from the Gsx2 (C) and Pax6 (D) TSSs, as determined by ChIP-qPCR. The error bars represent ±s.d. (n=4 per group). Statistical significance was determined using Student's t-test with Welch's correction (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001). (E,F) The structure of the transgene for visualizing enhancer activity of the Dmrt-binding site at the Gsx2 (E) or the Pax6 (F) locus. (G,H) Left: Lateral views of EGFP fluorescence of E12.5 transgenic mouse embryos generated by injecting the expression cassette indicated in E and F. Right: Immunostaining for Gsx2 (G) or Pax6 (H), with EGFP fluorescence, in coronal sections of the E12.5 telencephalon of transgenic mice E12.5. Scale bars: 100 µm. The yellow dotted lines indicate the outline of the telencephalon (right) and the ventricular surface (left).
Dmrt binds to the enhancer sequences of Gsx2 and Pax6. (A,B) Distributions of ChIP-seq peaks obtained with antibodies for Dmrt3 (blue), Dmrta2 (blue), H3K4me1 (green), H3K4me3 (magenta), H3K27Ac (red), and H3K27me3 (light blue) at the Gsx2 (A) and Pax6 (B) loci in the E12.5 mouse telencephalon. The bars on the top with letters indicate the position of the primer sets used for ChIP-qPCR in C and D. (C,D) Quantification of Dmrt3/Dmrta2 binding at sequences within 100 kb from the Gsx2 (C) and Pax6 (D) TSSs, as determined by ChIP-qPCR. The error bars represent ±s.d. (n=4 per group). Statistical significance was determined using Student's t-test with Welch's correction (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001). (E,F) The structure of the transgene for visualizing enhancer activity of the Dmrt-binding site at the Gsx2 (E) or the Pax6 (F) locus. (G,H) Left: Lateral views of EGFP fluorescence of E12.5 transgenic mouse embryos generated by injecting the expression cassette indicated in E and F. Right: Immunostaining for Gsx2 (G) or Pax6 (H), with EGFP fluorescence, in coronal sections of the E12.5 telencephalon of transgenic mice E12.5. Scale bars: 100 µm. The yellow dotted lines indicate the outline of the telencephalon (right) and the ventricular surface (left).
Differential regulation of Gsx2 and Pax6 by Dmrt factors
Our results above show that Pax6 and Gsx2 expressions differentially respond to the expression levels of Dmrt factors. We next asked how Gsx2 and Pax6 gene expression is regulated by different doses of Dmrt factors. To address this question, Dmrt mutant ES cells were subjected to an in vitro differentiation protocol by which ES cells can be differentiated into the dorsal telencephalic cell fate with no addition of instructive growth factors (the SFEBq method) (Fig. 7A,B, Fig. S10A-C, Fig. S11A,B) (Eiraku et al., 2008). When Dmrt mutant ES cells were treated with smoothened agonist (SAG), which activates canonical Shh signaling, the expression levels of Gsx2 and also Dlx1 were induced in a dose-dependent manner (Fig. 7C). Notably, their expression was increased more efficiently in DKO and TKO cells compared with wild-type cells at the same concentration of SAG (Fig. 7C). By contrast, Pax6 expression, which is negatively regulated by Shh signaling, was less sensitive to SAG treatment, and a much higher concentration was needed to reduce Pax6 expression even in DKO and TKO cells (Fig. 7C). These results indicate that Pax6 expression is robust against the weak Shh signaling that is sufficient to induce Gsx2 expression in progenitors. Furthermore, consistent with the dose-dependent function of Dmrt factors in vivo, the concentration of SAG that is required to activate Gsx2 expression was lower in TKO cells than that in DKO cells (Fig. 7C). When exogenous Dmrta2 was introduced by the Tet-On system (Fig. 7D,E), the expression of both Gsx2 and Pax6 was gradually and linearly decreased by step-wise increases of Dmrta2 expression (Fig. 7F). The decrease in Pax6 expression by Dmrta2 induction was much more attenuated than that of Gsx2 (Fig. 7F), suggesting that Gsx2 expression is more sensitive to Dmrt factors than Pax6. Thus, the differential sensitivity of Gsx2 and Pax6 gene expression to Dmrt factors may explain the differential responses of these genes to manipulations of Dmrt levels.
Differential regulation of Gsx2 and Pax6 by Dmrt. (A) Schematic summarizing the formation of cerebral organoids from ES cells (SFEBq method). Control and mutant ES cells were differentiated into telencephalic (cerebral) progenitors in vitro. (B) Bright-field images of cerebral organoids established via the SFEBq method (day 2 and day 8) treated with or without SAG, a smoothened antagonist, at a concentration from 0 to 3.0 nM. (C) Gene expression of Gsx2, Dlx1 and Pax6 normalized to Sox2 in organoids treated with SAG at the different concentrations (indicated in B) on day 9. The error bars represent ±s.e.m. (n=6 per group). (D) Schematic of the establishment of Dmrt3/Dmrta2 double-knockout ES cells inducibly expressing Dmrta2 under the control of the nestin enhancer and the Tet-ON system (TetON-Dmrta2). (E) mKO2 fluorescence induced with 1 µg/ml Dox from day 8 to day 10 in TetON-Dmrta2 cells differentiated by SFEBq. (F) Gene expression of mKO2, exoDmrta2 (exogenously expressed Dmrta2), Gsx2 and Pax6 normalized to Sox2 in SFEBq-differentiated TetON-Dmrta2 cells treated with Dox at concentrations from 0.3 to 3.0 µg/ml. n=4 per group. The solid lines represent a model fit obtained by non-linear regression analysis.
Differential regulation of Gsx2 and Pax6 by Dmrt. (A) Schematic summarizing the formation of cerebral organoids from ES cells (SFEBq method). Control and mutant ES cells were differentiated into telencephalic (cerebral) progenitors in vitro. (B) Bright-field images of cerebral organoids established via the SFEBq method (day 2 and day 8) treated with or without SAG, a smoothened antagonist, at a concentration from 0 to 3.0 nM. (C) Gene expression of Gsx2, Dlx1 and Pax6 normalized to Sox2 in organoids treated with SAG at the different concentrations (indicated in B) on day 9. The error bars represent ±s.e.m. (n=6 per group). (D) Schematic of the establishment of Dmrt3/Dmrta2 double-knockout ES cells inducibly expressing Dmrta2 under the control of the nestin enhancer and the Tet-ON system (TetON-Dmrta2). (E) mKO2 fluorescence induced with 1 µg/ml Dox from day 8 to day 10 in TetON-Dmrta2 cells differentiated by SFEBq. (F) Gene expression of mKO2, exoDmrta2 (exogenously expressed Dmrta2), Gsx2 and Pax6 normalized to Sox2 in SFEBq-differentiated TetON-Dmrta2 cells treated with Dox at concentrations from 0.3 to 3.0 µg/ml. n=4 per group. The solid lines represent a model fit obtained by non-linear regression analysis.
Increased levels of Dmrt expands the cerebral cortical area
We lastly examined the effect of Dmrt factor overexpression in vivo. As described above, Dmrt factors (1) suppress the conversion of progenitor identity to dLGE fate by repressing Gsx2 expression in the dorsolateral telencephalon and (2) maintain progenitors in a less neurogenic state by repressing Pax6 in the medial telencephalon. We then speculated that the overexpression of Dmrt factors in progenitors would lead to a conversion of their division mode from more neurogenic to less neurogenic, resulting in expansion of the cerebral cortex. To test this hypothesis, we generated transgenic mice in which Flag-tagged Dmrt3 was overexpressed in neural progenitors using the nestin enhancer and the minimal thymidine kinase (Tk) gene promoter (Fig. 8A). The transgenic mice exhibited massive planar expansion of the ventricular surface, in association with marked reduction of the expression of Pax6 with no increase in ventral gene expression. This finding suggests that the division mode of telencephalic progenitors shifted to a more proliferative mode (Fig. 8B). Consistent with this conclusion, the number of cells expressing Tbr2, a transcription factor that positively regulates neuronal differentiation (Arnold et al., 2008; Sessa et al., 2008), was decreased in the transgenic mice compared with wild-type brains (Fig. 8C). We also found a subtle increase in the number of Tbr1-positive cells in mutants compared with wild-type embryos although this was not statistically significant (data not shown). Further study is needed to reveal the involvement of Dmrt factors in cortical layer formation. These findings imply that the graded expression of Dmrt factors along the D-V axis governs the proportional development of each brain region in the telencephalon and confers region-specific neurogenic properties to neural progenitors, consequently defining the pattern of the dorsal telencephalon.
Increased expression of Dmrt expands the cerebral cortical area. (A) Structure of the transgene for overexpressing FLAG-tagged Dmrt3 under the control of the nestin enhancer. (B) Immunofluorescence for Tbr1, Tbr2, Sox2 and Pax6 in coronal sections of wild-type and Dmrt3-FLAG transgenic mice at E15.5. The yellow dotted lines indicate the ventricular surface. Scale bars: 100 µm. CAS, caudal amygdaloid stream. (C) Quantification of Tbr2-positive basal progenitors in a unit area (left) and the relative length of the ventricular surface in the dorsal telencephalic area of transgenic animals (right). The error bars represent ±s.d. (n=4 per group). Statistical significance was determined using Student's t-test with Welch's correction (*P<0.05; **P<0.01). (D) Model illustrating the Dmrt transcriptional cascade and the production of cortical/subcortical neurons. Graded expression of Dmrt factors determines the positional information of cerebral cortical progenitors by differentially suppressing Gsx2 and Pax6. HC-p, hippocampal projection neuron; hypo, hypoexpression; ML, medial-lateral; NC-p, neocortical projection neuron; OE, overexpression; WT, wild type.
Increased expression of Dmrt expands the cerebral cortical area. (A) Structure of the transgene for overexpressing FLAG-tagged Dmrt3 under the control of the nestin enhancer. (B) Immunofluorescence for Tbr1, Tbr2, Sox2 and Pax6 in coronal sections of wild-type and Dmrt3-FLAG transgenic mice at E15.5. The yellow dotted lines indicate the ventricular surface. Scale bars: 100 µm. CAS, caudal amygdaloid stream. (C) Quantification of Tbr2-positive basal progenitors in a unit area (left) and the relative length of the ventricular surface in the dorsal telencephalic area of transgenic animals (right). The error bars represent ±s.d. (n=4 per group). Statistical significance was determined using Student's t-test with Welch's correction (*P<0.05; **P<0.01). (D) Model illustrating the Dmrt transcriptional cascade and the production of cortical/subcortical neurons. Graded expression of Dmrt factors determines the positional information of cerebral cortical progenitors by differentially suppressing Gsx2 and Pax6. HC-p, hippocampal projection neuron; hypo, hypoexpression; ML, medial-lateral; NC-p, neocortical projection neuron; OE, overexpression; WT, wild type.
DISCUSSION
Our findings describe a novel transcriptional network in which mammalian Dmrt factors orchestrate the development of the cerebral cortex. The medialhigh/laterallow gradient of Dmrt factor expression operates in two ways: (1) determination of the D-V boundary by suppressing expression of the ventral gene Gsx2, a transcription factor essential for LGE and caudal ganglionic eminence cell fates; and (2) restriction of neurogenesis in the medial region through negative regulation of Pax6 expression.
Our results indicate that Dmrt factors repress Gsx2 by binding to its enhancer, thereby determining the D-V boundary. A remaining question is the nature of the relationship between this Dmrt function and those of known patterning factors, such as Pax6 and Gli3, which have also been implicated in restriction of Gsx2 expression. Key findings in this regard come from previous studies analyzing the genetic interaction between Pax6 or Gli3, and Shh signaling. In the dorsal telencephalon of Pax6 or Gli3 mutant mice, Gsx2 expression was shown to be ectopically increased, resulting in disorganization of the DV boundary (Toresson et al., 2000; Yun et al., 2001), similar to the phenotype seen in Dmrt mutant embryos. Intriguingly, additional knockout of the Shh gene in Pax6 or Gli3 mutants led to a partial rescue of this phenotype, indicating that the ventral restriction of Gsx2 expression by Pax6 and Gli3 requires Shh function (Aoto et al., 2002; Fuccillo et al., 2006; Rallu et al., 2002; Rash and Grove, 2007). These results are consistent with reports that both Pax6 and Gli3 directly repress the expression of the Shh gene itself or that of Shh target genes, such as Ptch1 (Caballero et al., 2014; Vokes et al., 2008). Taking these findings together with our results, it is likely that Dmrt factors repress Gsx2 expression by direct binding to its enhancer independently of Shh function. In contrast, Pax6 and Gli3 likely repress Gsx2 expression indirectly by suppressing Shh signaling and subsequently suppressing another downstream pathway, such as Fgf signaling. Consistent with this model, the additional knockout of the Shh gene in Dmrt3 and Dmrta2 DKO mice did not rescue the ectopic expression of Gsx2, and all neural progenitors located in the telencephalon of the triple-mutant embryos co-expressed both Pax6 and Gsx2 (data not shown). This result suggests that a ventralizing pathway other than Shh signaling positively regulates Gsx2 expression. Remarkably, our in vitro experiments in which ES cells were differentiated to neural progenitors using cortical organoids showed that Gsx2 expression did not change, or very slightly increased, in the absence of Shh agonist treatment, even in DKO and TKO cells. This result suggests that Gsx2 expression in vivo is positively regulated indirectly by Shh or directly by unidentified exogenous factors (Fig. 8D).
Dmrt factors have different roles in medial telencephalic neural progenitors from those in the lateral telencephalon in developing mice. Our experiments using an acute loss-of-function system in the medial telencephalon showed that Dmrt factors maintain a less neurogenic state by reducing the expression of Pax6, which positively regulates neurogenic genes (Fig. 8D) (Sansom et al., 2009). On the other hand, Pax6 in the developing telencephalon negatively regulates cell cycle progression by directly suppressing the expression of cell cycle regulators, such as Cdk6 (Mi et al., 2013a). In this manner, the graded expression pattern of Pax6 across the developing cortex leads to region-specific regulation of cell cycle progression. Reports have shown that changing cell cycle length can alter the mode of cell division of neural progenitors in the developing cerebral cortex (Lange et al., 2009; Pilaz et al., 2009). Therefore, Dmrt factors may reduce the neurogenic potential of progenitors by repressing Pax6 activity and thereby shortening the cell cycle.
Our results showed that Dmrt factors negatively regulate the expression of two transcription factors, Gsx2 and Pax6, in the developing dorsal telencephalon. These factors exhibit differential sensitivity to the dosage of Dmrt genes. In the case of Gsx2, a relatively lower amount of Dmrt factors can repress its expression in the dorsolateral telencephalon, whereas much higher Dmrt expression is required to suppress Pax6 expression. Of particular interest, the differential response of Gsx2 and Pax6 expression to Dmrt factors permits a situation in which only Pax6 but not Gsx2 is expressed. In this way, graded neurogenic potential of neural progenitors across the developing cerebral cortex can be achieved and cortical glutamatergic neurons can be generated. The question arises of why Gsx2 and Pax6 are differentially sensitive to the same levels of Dmrt expression. The results of our in vitro differentiation experiments provide a clue in this regard. Increasing the expression of exogenous Dmrta2 in SFEBq-mediated neural differentiation induced a gradual decrease in the expression levels of both Gsx2 and Pax6. However, the rate of decrease in the expression of Gsx2 was higher than that of Pax6 expression (Fig. 7F), suggesting at least three possibilities. First, Dmrt factors may have differential binding characteristics at Gsx2 and Pax6 enhancers. Second, the strength of the upstream pathways controlling Gsx2 and Pax6 expression might differ for these two genes. Third, differential regulation of histone modifications in these two enhancer regions may lead to the different response to the Dmrt binding. Indeed, suppressive H3K27me3 modification in the Gsx2 enhancer appeared to be abundant compared with that in the Pax6 enhancer in the dorsal telencephalon (Fig. 6A,B). Future studies to dissect the molecular properties of the Dmrt factors will likely reveal the mechanisms that underlie the differential dose dependence of Dmrt factors on the regulation of their target genes.
Recently, Desmaris et al. have reported a similar defect of dorsal-ventral patterning in Dmrt3/Dmrt5 and Emx2/Dmrt5 double mutants (Desmaris et al., 2018). This study is consistent with their previous report that Emx2 is a possible target of Dmrt5 (Saulnier et al., 2013). However, our study demonstrated that acute knockdown of Dmrt3/Dmrta2 resulted in no significant changes in Emx2 or Emx1 (data not shown). We also confirmed that there was no binding of Dmrt3 and Dmrta2 on Emx1 and Emx2 gene loci by ChIP-seq analysis. This observation raises a possibility that the expression change of Emx1 and Emx2 observed in Dmrt mutants is a secondary effect of lacking the cortical hem, a primordium expressing multiple Wnt genes. Given that Emx2 is partly required for the formation of the cortical hem and the expression of Wnt genes (Tole et al., 2000a), we hypothesize that Dmrt factors interact with Wnt signaling to suppress the expression of Gsx2. Further study is needed to understand the precise molecular mechanisms for the interaction among Dmrt factors, Emx1/2, and Wnt signaling.
The degree of evolutionary conservation of Dmrt factor function in cortical development remains to be elucidated. Several reports have shown the specific expression of a few Dmrt ‘A’ family genes in the dorsal telencephalon of other vertebrate embryos, including chicken, Xenopus and zebrafish (Hong et al., 2007). Therefore, it seems likely that the essential role of Dmrt factors reported in the present manuscript is, at least in part, conserved across vertebrates. Taken together with our present findings regarding the crucial role of Dmrt factors in the developing dorsal telencephalon, we hypothesize that acquiring Dmrt expression might have been one of the key steps in the emergence of the cerebral cortex during evolution. This possibility is an intriguing question for future study.
MATERIALS AND METHODS
Animals
Embryonic stages were calculated by defining noon on the day of vaginal plug as E0.5. Dmrt3 and Dmrta2 mutant mice were maintained in a C57BL/6 background, as described previously (Konno et al., 2012). Genotyping of the Dmrt3 allele was performed by PCR using the following primers: Dmrt3-P1: 5′-GGACCTGCGGGTGGAGCCTG-3′; Dmrt3-P2: 5′-GTCTGTCCTAGCTTCCTCACTG-3′; Dmrt3-P3: 5′-GATGAGGCTCTCCAGGCTCTCGTTG-3′. These primers yield bands of 462 and 336 bp for the wild-type and mutant alleles, respectively. Genotyping of Dmrta2 alleles was performed by PCR using the following primers: Dmrta2-P1: 5′-ACTTCGGATCCTAGTGAACCTCTTCGAG-3′; Dmrta2-P2: 5′-TGCCTACGAAGTCTTTGGCTCGGTTTG-3′; Dmrta2-P3: 5′-TGGAGAGCCACAGTTAAGTAGTTGGAGC-3′. These primers yielded bands of 249 and 210 bp for the wild-type and mutant alleles, respectively. The generation of Dmrta1 mutant mice was achieved by inducing a deletion in the vicinity of the initiation codon of Dmrta1 using the CRISPR/Cas9 system, followed by injection of the mutant ES cells into 8-cell-stage mouse embryos (accession number CDB1303K, http://www2.clst.riken.jp/arg/mutant%20mice%20list.html) (see also Fig. S4D). Genotyping of Dmrta1 alleles was performed by PCR using the following primers: Dmrta1-P1: 5′-TCCAGCCTGGCCCTTCTAGGCTC-3′; Dmrta1-P2: 5′-ACCGGAGCTTGCGGTCGATCGCAG-3′. These primers yielded bands of 450 and 350 bp for the wild-type and mutant alleles, respectively.
Dmrta2;Dmrt3-double heterozygotes exhibited normal fertility and CNS development. We therefore used these mice as controls in this study, except as noted.
All of the animal manipulations were performed according to the guidelines for animal experiments by Institutional Animal Care and Use Committee (IACUC) of RIKEN Kobe Branch.
Plasmids
pCAG-EGFP3NLS is an expression plasmid for 3xNLS-tagged EGFP and was described previously (Konno et al., 2008). pNesE-tk-rtTA2SM2 and pTRETi-mKO2_Dmrta2 are plasmids for the conditional expression of Dmrta2 and are based on the second generation of the Tet-On system (Tet-On Advanced, Invitrogen). pNesE-tk-rtTA2SM2 was constructed by inserting the enhancer sequence of the rat nestin gene (NesE) (Lothian and Lendahl, 1997) obtained using rat genomic DNA by PCR and inserted together with the thymidine kinase minimal promoter (tk) into the rtTA2SM2 expression plasmid (pTet-On Advanced, Clontech). pTRETi-mKO2_ Dmrta2 was constructed by inserting mKO2 [a red fluorescent protein (MBL, Nagoya, Japan)], and Dmrta2 amplified using E12.5 mouse forebrain cDNA by PCR into multiple cloning sites 1 and 2, respectively. The pNesE-tk-rtTA2SM2 and pTRETI-mKO2;Dmrta2 expression cassettes were then excised and inserted into pT2AL200R150G, in which the cloning site is flanked by the Tol2 transposon elements (Urasaki et al., 2006). These elements then integrate into the genome in the presence of Tol2 transposase. pCAGGS-T2TP is a plasmid expressing the Tol2 transposase under the control of the CAG promoter (Kawakami and Noda, 2004). pPGK-PuroR is a plasmid expressing the puromycin-resistant gene (PuroR) under the control of the PGK promoter and was obtained from our animal facility (Laboratories of Animal Resource Development and Genetic Engineering, RIKEN CDB). pNesE-tk-Dmrt3-Flag is a plasmid for the transgenic expression of Flag-tagged Dmrt3 driven by the NesE-tk enhancer–promoter cassette. The plasmid was used to generate transgenic mice after removing the plasmid backbone by digesting with restriction enzymes.
In utero electroporation
In utero electroporation was performed at E11.5 as described previously (Fukuchi-Shimogori and Grove, 2001; Saito and Nakatsuji, 2001; Tabata and Nakajima, 2001). Briefly, CD-1 mouse embryos were electroporated with expression plasmids and siRNAs using an electroporator (CUY21, BEX, Tokyo, Japan) at the following concentrations: pCAG-EGFP3NLS, (0.5 µg/µl); siRNAs, 75 µM for Dmrt3, Dmrta1 and Pax6; 150 µM for Dmrta2.
siRNAs
The following target sequences were used to synthesize the siRNAs (Invitrogen stealth siRNA) for knockdown experiments of mouse Dmrt3, Dmrta2 and Dmrta1: Dmrt3_control (mutated Dmrt3_#2): 5′-GCGCGTTCGATAACCGATACACTGA-3′; Dmrt3_#1: 5′-TGAGGTCCCAGTATGTCAGTCCATT-3′; Dmrt3_#2: 5′-GCGCAGCTTGCTAAACCAGATCTGA-3′; Dmrt3_#3: 5′-GCCCTCTAGCGGCCATATCTTTGAA-3′; Dmrta2_control (mutated Dmrta2_#3): 5′-CAAGTACAGGATTGTTATCGTATTT-3′; Dmrta2_#1: 5′-GCCTACGAAGTCTTTGGCTCGGTTT-3′ Dmrta2_#2: 5′-GAAGGACTGCCTGTGCGCCAAGTGT-3′; Dmrta2_#3: 5′-CAAATTGCAGAAGTTTGATCTGTTT-3′; Dmrta1_control (mutated Dmrta1_#3): 5′-GGAAGACTTCAATCTATCACGGGTT-3′; Dmrta1_#1: 5′-GAGTGGGCCAGAGACTACATTGCTA-3′; Dmrta1_#2: 5′-CCACGAGACCCTCTCGGAATTCTTA-3′; Dmrta1_#3: 5′-GGAGGAGATTCAACTCTCTACAGTT-3′; Pax6_control (mutated in Pax6_#2): 5′-CCACGAACAAACCGTCCATTGTGCA-3′; Pax6_#1: 5′-ACCACACCTGTCTCCTCCTTCACAT-3′; Pax6_#2: 5′-CCATGGCAAACAACCTGCCTATGCA-3′; Pax6_#3: 5′-CATCAATAAACAGAGTTCTTCGCAA-3′. The knockdown efficiency of each siRNA was examined in HEK293 cells transfected with the siRNAs/expression plasmids at 24 h by western blotting. The transfections were performed using the Lipofectamine 2000 reagent (Invitrogen), according to the manufacturer's protocol.
Isolation of electroporated cells
Electroporated brain regions were dissected, incubated in 0.05% trypsin/HBSS(−) at 37°C for 10 min, and dissociated by adding 0.75% BSA/PBS(−) and pipetting gently. EGFP-positive transfected cells were collected using a SH800 cell sorter (Sony Corporation, Tokyo, Japan) with the ultra-purity mode. The cells were collected directly into TRIzol LS reagent (Invitrogen) and stored at −80°C until the purification of total RNA.
Establishment of ES cells
Knockout ES cells for Dmrt3 and Dmrta2 were established according to a previously described protocol (Kiyonari et al., 2010). Briefly, blastocysts obtained by crossing Dmrt3+/− and Dmrta2flox/flox mutant mice were cultured in iSTEM mouse ES cell media supplemented with 0.8 µM PD184352 (Stem Cell Sciences), 2 µM SU5402 (Stem Cell Sciences), 3 µM CHIR99021 (Stem Cell Sciences) and 1000 U/ml LIF (ESGRO) in feeder-free conditions. After hatching, the blastocyst inner cell masses were plated on mouse embryonic fibroblast (MEF) feeder cells, cultured for 5-7 days, trypsinized, and re-plated on MEF feeder cells. ES cell-like colonies were picked and expanded 3-4 days after re-plating. After establishing Dmrt3+/+;Dmrta2flox/flox or Dmrt3−/−;Dmrta2flox/flox ES cells, the cells were transfected with a plasmid expressing Cre recombinase driven by polyoma enhancer/herpes simplex virus thymidine kinase (MC1) promoter to delete Dmrta2 exon 2.
To establish the ES cell line that conditionally expresses Dmrta2 in Dmrt3/Dmrta2 knockout ES cells, Dmrt3−/−;Dmrta2−/− ES cells were transfected with pT2-NesEtk-rtTA2SM2, pT2-pTRETI-mKO2_Dmrta2, pCAGGS-T2TP, and pPGK-PuroR using the NEON Transfection System (Invitrogen), according to the manufacturer's protocol. Puromycin (2 µg/ml) was added at 24 h and removed 72 h after transfection to kill non-transfected cells. After recovering for a few days, individual colonies were picked and re-plated into two wells of a 96-well plate by equal splitting, allowing for the colony to expand and for checking transgene expression. The transgene expression was confirmed by the red fluorescence of mKO2 and by western blotting for Dmrta2 and mKO2 in SFEBq-differentiated ES cells treated with 1 µg/ml doxycycline (Dox) for 72 h from culture day 7.
To establish triple-mutant ES cells for Dmrt3, Dmrta2 and Dmrta1 (referred to as Dmrt3−/−;Dmrta2−/−;Dmrta1−/−), the CRISPR/Cas9 system (Addgene) (Cong et al., 2013) was applied using the established ES cells. Briefly, Dmrt3−/−;Dmrta2−/− ES cells were transfected with expression plasmids encoding SpCas9 and sgRNAs to delete the genomic sequence containing the start codon of Dmrta1. The isolation and expansion of ES cell clones were performed according to the procedures described above. The sgRNA target sequences and combinations for knocking out Dmrta1 are described as follows: Set 1, 5′-GACCTAGGCGGGTCCTCAGC-3′ and 5′-GCGGGCTGCTGCGCCCGCTT-3′; Set 2, 5′-GACCTAGGCGGGTCCTCAGC-3′ and 5′-GTCGGTGTCGTCGGGGATTC-3′. Homozygous deletion and the introduction of a frame-shift error in Dmrta1 were confirmed by DNA sequencing.
cDNA synthesis
The appropriate amounts of tissue and cells were dissociated in TRIzol (for tissue samples) (Invitrogen) or TRIzol LS (for sorted cells) (Invitrogen) reagents. Total RNA was purified using a Qiagen RNeasy Mini kit with on-column DNaseI digestion (Qiagen). Approximately 100-500 ng of total RNA was used for the first-strand cDNA synthesis using SuperScript VILO cDNA Synthesis Kit (Invitrogen). All of the procedures were performed according to the manufacturer's protocols.
Real-time quantitative PCR
Real-time quantitative PCR (qPCR) was performed using Thermal Cycler Dice Real Time System TP860 (Takara Bio) with FastStart Universal SYBR Green Master Mix (Roche). For the RT-qPCR analysis, the Ct value was determined using the comparative Ct method (ΔΔCT method), and the expression level of target genes is reported as relative to that of Sox2 or Gapdh. In ChIP-qPCR analysis, the amount of target DNA sequences was determined using the ΔΔCT method and reported as a percentage of that in the input sample in each experiment. The primer sequences used in RT-qPCR and ChIP-qPCR are listed in Table S1.
Microarray analysis
Cy3-labeled cRNA (1.65 µg) was fragmented at 60°C for 30 min in a reaction volume of 55 µl containing 25× fragmentation buffer and 10× blocking agent, following the manufacturer's instructions (Agilent). On completion of the fragmentation reaction, 55 µl of 2× GEx Hybridization Buffer HI-RPM was added to the fragmentation mixture and hybridized to Agilent Whole Mouse Genome Microarrays (G4846A) for 17 h at 65°C in a rotating Agilent hybridization oven. After hybridization, the microarrays were washed for 1 min at room temperature with GE Wash Buffer 1 (Agilent) and 1 min with 37°C GE Wash buffer 2 (Agilent). The slides were scanned immediately after washing on the Agilent DNA Microarray Scanner (G2505C) using the one-color scan setting for 4×44k array slides (scan area 61×21.6 mm, scan resolution 5 µm, dye channel set to green, green PMT set to 100%, NoXDR). The scanned images were analyzed with Feature Extraction Software 10.5.1.1 (Agilent) using default parameters (protocol GE1_105_Jan09 and Grid:026655_D_F_20100123) to obtain background-subtracted and spatially detrended Processed Signal intensities. The normalized signal intensities (log2 values) were calculated by Genespring Agilent GX11.0 software.
Immunohistochemistry
Immunohistochemistry was performed as described previously (Konno et al., 2012). Briefly, the brains were fixed in 1% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) at 4°C for 2 h. After preparing 12-μm-thick frozen coronal or sagittal sections of fixed brains using a cryostat (Leica), the sections were mounted on APS-coated slide glasses (Matsunami) and stained with the following antibodies: anti-Dmrt3 (rabbit, 1:5000) (Konno et al., 2012), anti-Dmrta2 (rabbit, 1:5000) (Konno et al., 2012), anti-Dmrta2 (rat, 1:2000) (Konno et al., 2012), anti-βIII-tubulin (mouse, 1:5000, clone Tuj1, Covance), anti-Gad2 (rabbit, 1:200, 3988, CST Japan), anti-Tbr1 (rabbit, 1:10,000, a gift from Dr. Robert Hevner, University of Washington), anti-Pou3f2 (Brn-2) (goat, 1:500, C-20, Santa Cruz Biotechnology), anti-Zbtb20 (rabbit, 1:500, HPA016815, Sigma-Aldrich), anti-Math2 (Neurod6) (goat, 1:500, L-15, Santa Cruz Biotechnology), anti-Dlx2 (guinea pig, 1:2000, a gift from Dr. Kazuaki Yoshikawa, Osaka University), anti-Gsh2 (Gsx2) (rabbit, 1:10,000, a gift from Dr. Yoshiki Sasai, RIKEN CDB), anti-Pax6 (rabbit, 1:500, PRB-278P, Covance), anti-Sp8 (goat, 1:1000, C-18, Santa Cruz Biotechnology), anti-Isl1 (mouse, 1:20, Developmental Studies Hybridoma Bank), anti-Er81 (rabbit, 1:20, a gift from Dr Silvia Arber, University of Basel), anti-Sox2 (goat, 1:1000, Y-17, Santa Cruz Biotechnology), anti-Sox2 (rat, 1:1000, clone Btjce, eBioscience), anti-calretinin (goat, 1:200, AF5065, R&D Systems), anti-tyrosine hydroxylase (TH) (rabbit, 1:200, AB152, EMD Millipore), anti-Eomes (Tbr2) (rat, 1:1000, clone Dan11mag, eBioscience), anti-DYKDDDDKtag (Flag) (mouse, 1:1000, clone 1E6, Wako Pure Chemical Industries), anti-Emx1 [1:10,000, clone 1H8B11, a monoclonal antibody generated by immunizing a synthetic peptide (CKQANGEDIDVTSND) in rats, Cell Engineering Corporation, Osaka, Japan], anti-Neurog2 (goat, 1:200, C-16, Santa Cruz Biotechnology), anti-hAsc1 (Ascl1) (mouse, 1:1000, clone 24B72D11.1, BD-Japan), anti-ASH1 (Ascl1) (rabbit 1:1000, SK-T01-003, Cosmo Bio), anti-Bf1 (Foxg1) (goat, 1:200, N-15, Santa Cruz Biotechnology), anti-Bf1 (Foxg1) (rabbit, 1:500, M227, Takara Bio), anti-Lhx2 (goat, 1:500, C-20, Santa Cruz Biotechnology) and anti-Lmx1a (rabbit, 1:2000, AB10533, EMD Millipore). The primary antibodies were visualized with the secondary antibodies conjugated to Alexa 488, Cy3 and Cy5 (donkey, 1:500, 715-545-151, 715-165-151, 715-605-151, 711-545-152, 711-165-152, 711-605-152, 712-545-153, 712-165-153, 712-605-153, 706-545-148, 706-165-148, 706-605-148, 705-545-147, 705-165-147, 705-605-147, Jackson ImmunoResearch). For staining with mouse monoclonal antibodies, the sections were pre-incubated with monovalent Fab fragments (Jackson ImmunoResearch) to reduce the background signals on the mouse tissues. Apoptotic cells were detected by TUNEL (terminal deoxynucleotidyl transferase dUTP nick-end labeling) staining using the In Situ Cell Death Detection kit (Roche Diagnostics). All of the fluorescent images were acquired using Olympus FV1000 confocal microscope (Olympus).
Chromatin immunoprecipitation (ChIP)
The E12.5 wild-type CD-1 forebrain samples were fixed in D-PBS(−) containing 0.5% paraformaldehyde at room temperature for 5 min. After washing with ice-cold D-PBS(−) containing 0.1 M glycine twice, the brains were dissociated in ChIP buffer [10 mM Tris-HCl (pH 8.0), 200 mM KCl, 1 mM CaCl2, 0.5% NP40] containing protease inhibitor cocktail (EDTA free) (25955-11, Nacalai Tesque) with a pestle homogenizer and briefly sonicated (Handy Sonic UR-20P, Tomy Seiko Co.). After sonication, the suspension was treated with 50 units/ml micrococcal nuclease (Worthington Biochemical) for 60 min at 37°C, with mild agitation for every 10 min. The nuclease reaction was stopped by adding ethylenediaminetetraacetic acid (EDTA; final concentration of 10 mM). After briefly sonicating again and centrifuging at 15,000 g for 10 min at 4°C, the supernatant was collected. The supernatants were then incubated at 4°C for 1 h with anti-mouse or anti-rabbit IgG magnetic beads (Dynabeads, Life Technologies) pre-bound with the following antibodies (2 μg antibody/20 μl Dynabeads suspension): rabbit normal IgG (Jackson ImmunoResearch), rabbit anti-Dmrt3 (Konno et al., 2012), rabbit anti-Dmrta2 (Konno et al., 2012), mouse anti-HistoneH3K4me (Ab8895, Abcam), mouse anti-HistoneH3K4me3, mouse anti-HistoneH3K27ac (Ab4729, Abcam), or mouse anti-HistoneH3K27me3 (Kimura et al., 2008). The beads were briefly washed three times with ChIP buffer and three times with ChIP wash buffer [10 mM Tris-HCl (pH 8.0), 500 mM KCl, 1 mM CaCl2, 0.5% NP40]. The beads were treated in elution buffer [50 mM Tris-HCl (pH 8.0), 10 mM EDTA, and 1% sodium dodecyl sulfate] containing 20 µg/ml RNase (Sigma-Aldrich) at 37°C for 10 min and eluted by adding 20 µg/ml Proteinase K (Roche Diagnostics) and incubated at 55°C for 16 h. The DNA was recovered using a DNA gel extraction kit (Promega) and used for ChIP-qPCR analysis and ChIP-seq analysis.
Transgenic mice
Transgenic mice were generated by injecting the linearized expression cassettes derived from pNesEtk-Dmrt3-Flag into fertilized oocytes derived from CD-1 mouse strains. All of the transgenic animals were analyzed in transgenic founder (F0) embryos at E15.5.
RNA-sequencing
The RNA-seq libraries were prepared using TruSeq RNA Sample Preparation v2 kit and sequenced using an Illumina Genome analyzer-II X. The expression levels of genes were estimated using Cufflinks (version 2.2.1, options ‘cuffdiff –u –b’) with reads mapped onto the mouse genome (mm9) using TopHat (version 2.0.12) with default parameters.
ChIP-sequencing
The ChIP library was prepared using NEBNext Ultra DNA Library Prep kit for Illumina (New England Biolabs) and sequenced on the Illumina HiSeq1500 system (Illumina). The sequenced reads of ChIP and input DNA controls were mapped to the mouse genome (mm9) using bowtie2 with default parameters (version 2.2.3) (Langmead and Salzberg, 2012). The multi-reads were discarded by filtering ‘XS:I’ tags, and PCR-duplicate reads were discarded using SAMtools (Li et al., 2009). The mapped reads were counted in non-overlapping 100 base windows, and the counts were normalized as reads per million (RPM). We then calculated the normalized ChIP-Seq signal intensities on each window as RPM difference between ChIP and input DNA control data (i.e. RPMChIP–RPMInput).
Quantifications
All of the statistical analyses were performed with Student's t-test with Welch's correction or one-way ANOVA using Prism 6 or R. The data were expressed as the mean±s.d. or ±s.e.m., as indicated in figure legends. Across all experiments, the data distribution was assumed to be normal, although the normality was not formally tested.
Gene ontology (GO) analysis
GO enrichment analysis was performed using PANTHER Classification System (Mi et al., 2013c) on the Gene Ontology Consortium website (http://geneontology.org).
Acknowledgements
We thank Matsuzaki laboratory members for helpful discussions and the animal resource facility members of RIKEN BDR for producing and maintaining the mutant mice.
Footnotes
Author contributions
Conceptualization: D.K., F.M.; Methodology: D.K.; Software: D.K.; Validation: D.K.; Formal analysis: D.K.; Investigation: D.K., C.K., K.M., Y.O., H.K., S.O.; Resources: D.K., H.K.; Data curation: D.K.; Writing - original draft: D.K.; Writing - review & editing: D.K., F.M.; Visualization: D.K.; Supervision: D.K., F.M.; Project administration: D.K., F.M.; Funding acquisition: D.K.
Funding
This work was supported by Ministry of Education, Culture, Sports, Science and Technology/Japan Society for the Promotion of Science KAKENHI grants (JP22700361, JP24500395, JP25123724, JP15K06727, JP19H05266, JP18H05527).
References
Competing interests
The authors declare no competing or financial interests.