Abstract
The transcription factor Bcl11b/Ctip2 plays critical roles in the development of several systems and organs, including the immune system, CNS, skin, and teeth. Here, we show that Bcl11b/Ctip2 is highly expressed in the developing vomeronasal system in mice and is required for its proper development. Bcl11b/Ctip2 is expressed in postmitotic vomeronasal sensory neurons (VSNs) in the vomeronasal epithelium (VNE) as well as projection neurons and GABAergic interneurons in the accessory olfactory bulb (AOB). In the absence of Bcl11b, these neurons are born in the correct number, but VSNs selectively die by apoptosis. The critical role of Bcl11b in vomeronasal system development is demonstrated by the abnormal phenotypes of Bcl11b-deficient mice: disorganization of layer formation of the AOB, impaired axonal projections of VSNs, a significant reduction in the expression of vomeronasal receptor genes, and defective mature differentiation of VSNs. VSNs can be classified into two major types of neurons, vomeronasal 1 receptor (V1r)/Gαi2-positive and vomeronasal 2 receptor (V2r)/Gαo-positive VSNs. We found that all Gαi2-positive cells coexpressed Gαo during embryogenesis. This coexpression is also observed in newly differentiated neurons in the adult VNE. Interestingly, loss of Bcl11b function resulted in an increased number of V1r/Gαi2-type VSNs and a decreased number of V2r/Gαo-type VSNs, suggesting that Bcl11b regulates the fate choice between these two VSN types. These results indicate that Bcl11b/Ctip2 is an essential regulator of the differentiation and dichotomy of VSNs.
Introduction
Most terrestrial vertebrates possess a vomeronasal system, which detects pheromones to mediate social and reproductive behaviors (Keverne, 1999; Dulac and Torello, 2003; Halpern and Martínez-Marcos, 2003; Brennan and Zufall, 2006). In the mouse, pheromone signals are detected by the vomeronasal sensory neurons (VSNs), which generally express vomeronasal receptor (VR) genes that encode a putative seven-transmembrane domain protein (Dulac and Axel, 1995; Herrada and Dulac, 1997; Matsunami and Buck, 1997; Ryba and Tirindelli, 1997). There are two classic types of VSN, which are distinguished by the location of the cell body within the vomeronasal epithelium (VNE), the class of VR expressed, the type of G-protein subunit expressed, and their axonal target site in the accessory olfactory bulb (AOB). VSNs located apically in the VNE coexpress vomeronasal 1 receptor (V1r) genes along with Gαi2 and project their axons to the anterior half of the AOB. In contrast, basally located VSNs coexpress vomeronasal 2 receptor (V2r) genes along with Gαo and project their axons to the posterior half of the AOB (Berghard and Buck, 1996; Jia and Halpern, 1996). The cell bodies of V1r/Gαi2-positive and V2r/Gαo-positive VSNs form nonoverlapping layers in the VNE. Although both types of VSNs are generated from Mash1-positive progenitor cells (Murray et al., 2003), the genetic mechanisms that regulate the differentiation and dichotomy of VSNs are not well understood; only a few critical transcription factors are known to regulate the development of VSNs (Murray et al., 2003; Ikeda et al., 2007; Duggan et al., 2008).
The Bcl11b gene (also known as Ctip2 or Rit1) encodes a C2H2 zinc finger transcription factor that is predominantly expressed in the immune system, CNS, and the embryonic olfactory system in mice (Avram et al., 2000; Wakabayashi et al., 2003a; Leid et al., 2004). Several studies of Bcl11b-deficient (Bcl11b−/−) mice have demonstrated that Bcl11b is required for T-cell development, the axonal projections of corticospinal motor neurons, the differentiation of medium spiny neurons of the striatum, skin morphogenesis, and odontogenesis (Wakabayashi et al., 2003b; Arlotta et al., 2005, 2008; Albu et al., 2007; Golonzhka et al., 2009a,b; Ikawa et al., 2010; Li et al., 2010a,b). In this study, we investigated the expression and function of Bcl11b in the developing vomeronasal system in vivo. We found that the expression of Bcl11b changed dynamically during the development of the VNE and AOB from embryogenesis to adulthood. The loss of function of Bcl11b resulted in severe abnormalities in the vomeronasal system, including the disorganization of layer formation in the AOB, impaired axonal projection of VSNs, a severe reduction of VR gene expression, defects in mature differentiation of VSNs, and distinct changes in the expression of several genes in VSNs. In addition, a deficiency of Bcl11b disturbs the balance between the two types of VSNs produced, suggesting that Bcl11b regulates the fate choice between these cell types, indicating that Bcl11b is a critical regulator during functional development of the vomeronasal system.
Materials and Methods
Mutant mice.
Bcl11b−/− and Mash1−/− mice were generated as described previously(Guillemot et al., 1993; Wakabayashi et al., 2003b). Mutant and wild-type mice/embryos of either sex were used. For embryo staging, mid-day of the day of the vaginal plug was designated as embryonic day (E) 0.5. The day of birth was designated postnatal day (P) 0. All mouse studies were approved by the Institutional Animal Experiment Committee of the Tokyo Institute of Technology, and were performed in accordance with institutional and governmental guidelines.
In situ hybridization.
Probes for Mash1, Ngn1, NeuroD, SCG10, GAP43, and OMP were prepared as previously described (Hirota and Mombaerts, 2004). Probes for Gαi2, Gαo, V1rb1, V1rd16, V1re4, V2ra, V2rb, and V2rc were provided by P. Mombaerts, Max-Planck Institute of Biophysics, Frankfort, Germany (Rodriguez et al., 2002; Ishii et al., 2003). Other probes were prepared from the cDNA of Bcl11b (nucleotides 1501-2560, GenBank accession number BC19503), Vil1 (nucleotides 356-2634, NCBI reference sequence NM_009509), Tcfap2e (nucleotides 916-1875, NCBI reference sequence NM_198960), Cart (nucleotides 96-788, NCBI reference sequence NM_001081493), Big2/Contactin4 (nucleotides 1189-2575, NCBI reference sequence NM_173004.3), Mef2b (nucleotides 500-1255, NCBI reference sequence NM_00445484), Panx3 (nucleotides 165-1598, NCBI reference sequence NM_172454), Meis2 (nucleotides 453-1354, NCBI reference sequence NM_001136072.2), and Olig1 (nucleotides 971-1919, NCBI reference sequence NM_016968.4). Single-color and two-color in situ hybridization (ISH) was performed according to a method described previously (Ishii et al., 2003, 2004). For two-color ISH, tyramide signal amplification (TSA)-biotin, TSA-dinitrophenyl, and TSA-plus biotin systems (PerkinElmer) were used. The images were taken on an Olympus BX51 microscope with a DP71 digital CCD camera for bright-field images and a Leica SP or SPE confocal microscope for fluorescent images.
Immunohistochemistry.
Immunohistochemistry (IHC) was performed according to a previously described method (Hirota et al., 2007). The following primary antibodies and dilutions were used: rabbit anti-Bcl11b (1:200) (Wakabayashi et al., 2003b); goat anti-Ki67 (1:1500 or 1:3000, catalog #SC-7846, Santa Cruz Biotechnology); goat anti-OMP (1:5000, catalog #544-10001, Wako); guinea pig anti-GABA (1:1000 or 1:10,000, catalog #AB175, Millipore); rabbit anti-Tbx21 (1:5000, a gift from Y. Yoshihara, RIKEN, Wako, Japan) (Yoshihara et al., 2005); guinea pig anti-Tbx21 (1:1000, a gift from Y. Yoshihara) (Yoshihara et al., 2005); mouse anti-NCAM (1:500, catalog #C-9672, Sigma); mouse anti-synaptophysin (1:1000, catalog #MAB5258, Millipore); rabbit anti-protocadherin 21 (1:1000, a gift from Dr. Yoshihara) (Kaneko-Goto et al., 2008); and rabbit anti-active caspase-3 (1:500 or 1:5000, catalog #C92–605, BD Pharmingen). The following appropriate secondary antibodies were used: the Alexa series (Invitrogen) and Cy3- and Cy5-conjugated goat anti-rabbit IgG (Jackson Immuno Research Laboratories); biotin-conjugated donkey anti-goat IgG (Rockland); biotin-conjugated goat anti-guinea pig IgG and goat anti-rabbit IgG (Vector Laboratories) antibodies. For the immunostaining that used the anti-Ki67 or anti-active caspase-3 antibodies, the sections were treated with the Vector Avidin/Biotin Blocking Kit (Vector Laboratories) and incubated with primary antibodies. Immunostaining was performed using the Vectastain Detection Kit (Vector Laboratories) in combination with the TSA-biotin system (PerkinElmer) to amplify signals. Anti-Ki67 immunoreactivity was detected using the DAB-chromogenic method with streptavidin-HRP, and anti-active caspase-3 was detected with streptoavidin-Alexa488 fluorescence. For immunostaining with the mouse anti-NCAM and anti-synaptophysin antibodies, the sections were pretreated with the Vector M.O.M. Immunodetection Kit (Vector Laboratories) before treatment with the primary antibody. For immunostaining of the nuclear protein, we performed antigen-retrieval pretreatment in 10 mm sodium citrate for 10 min or in HistoVT One (Nacalai Tesque) at 70°C for 15 min using a microwave oven.
Quantitative analyses.
To quantify the number of DAPI-positive cells, VR-expressing cells, Gαi2 and Gαo-expressing cells, proliferating cells, and apoptotic cells, every fifth or tenth coronal section (10 μm thickness), or every third coronal section (20 μm thickness) throughout the vomeronasal organ (VNO) was collected for each staining experiment, and the number of positive cells was counted.
DiI tracing experiment.
DiIC18(3) (DiI) crystal (Invitrogen) was heat melted to coat a glass capillary. After dissection of the heads of Bcl11b−/− and wild-type mice, the nose tip was cut to expose the rostral vomeronasal organ, and the DiI-coated glass capillary was placed into the lumen of the VNO. Samples were incubated in PBS for 2 h at 37°C, in 0.5% PFA/PBS overnight, and in 4% PFA/PBS for a month at room temperature to allow for the diffusion of DiI from the VSNs to the axonal termini. The DiI fluorescence of sagittally transected heads was imaged using an Olympus BX51 fluorescence microscope with a DP71-digital CCD camera.
Affymetrix microarrays.
VNOs were obtained from Bcl11b−/− and wild-type mice at P0 and stored in RNAlater (Ambion). The total RNA of each preparation was extracted using the RNeasy Mini Kit (Qiagen). Biotinylated cRNA was synthesized using 10 ng of total RNA with the Two-Cycle Target Labeling and Control Reagents Kit (Affymetrix), fragmented, and hybridized to the DNA microarrays (Mouse Genome 430 2.0 Array). To ensure reproducibility, microarray analyses were performed with RNA samples from six Bcl11b−/− and five wild-type mice. The microarray data were linearly normalized with the GAPDH (Probe ID: 1418625_s_at) signal of each preparation using GeneChip operating software (Affymetrix). The statistical significance of gene expression differences between Bcl11b−/− and wild-type was analyzed using GeneSpring version 7.3 (Agilent Technologies), and differentially expressed genes with a false discovery rate <0.25 were extracted from the data for further experiments.
Results
Bcl11b is expressed in the developing vomeronasal epithelium
The expression of Bcl11b/Ctip2 has been reported in the embryonic main olfactory epithelium (MOE) (Leid et al., 2004), but its expression in the VNE has not been studied. Because the MOE and the VNE are derived from the same olfactory placode, we assumed that Bcl11b would also be expressed in the VNE. Therefore, we examined the expression of Bcl11b in the developing VNE using ISH. The ISH studies revealed that Bcl11b was expressed in the VNE during the course of fetal development to adulthood, and its expression levels and patterns changed dynamically (Fig. 1A). The expression of Bcl11b was observed in the vomeronasal groove/VNE, and the MOE at the earliest time point examined, E11.5, which is shortly after the olfactory pits invaginate to develop the VNE and the MOE (Cuschieri and Bannister, 1975; Garrosa et al., 1998). The expression of Bcl11b increased gradually in level and in the number of cells during embryogenesis (Fig. 1A). From E16.5 to P0, a strong expression of Bcl11b was detected in the VSN layer but not in the sustentacular cell layer. Expression levels in individual cells were not uniform; rather, high-, low-, and nonexpressing cells were intermingled in the embryonic VNE. After birth, Bcl11b expression changed dynamically: Bcl11b expression gradually decreased and was restricted to the marginal region of the VNE (Fig. 1A, arrows), where neuronal progenitor/precursor cells and immature neurons localize. This result indicates that Bcl11b is expressed mainly in proliferating cells and/or immature neurons. Because proliferating cells and differentiating and differentiated neurons intermingle in the VNE during embryogenesis and for several days postnatally, the differential expression levels of Bcl11b in individual cells in the embryonic VNE likely represent a differentiation-dependent expression of Bcl11b.
To characterize Bcl11b-expressing cells, we performed two-color ISH using riboprobes for specific molecular markers at P0 and P14 (Fig. 1B–M). In the MOE, Mash1 (a marker for neuronal progenitors), Ngn1 (neuronal precursors), NeuroD (differentiating/post mitotic neurons), SCG10 (pan-neurons/immature neurons), GAP43 (immature neurons), and OMP (mature neurons) were used to stage the differentiation of olfactory sensory neurons (OSNs) (Cau et al., 1997, 2002). However, the developmental time course of expression of these marker genes has not been fully established in the VSN lineage. Murray et al. (2003) have shown that the expression of Mash1 and Ngn1 can be detected in progenitor/precursor cells in the VNE; have demonstrated that Mash1 was required for Ngn1 expression in the VNE and for the neuronal differentiation of VSNs; and have proposed that a developmental hierarchy of gene expression in VSN lineage is similar to that of OSN. Accordingly, we used the same set of markers that were established for OSNs to characterize Bcl11b-expressing cells in the VNE.
At P0, Mash1-expressing cells did not coexpress Bcl11b (Fig. 1B), and a few Ngn1-expressing cells coexpressed Bcl11b (Fig. 1C). NeuroD was expressed in the developing VNE, and some NeuroD-positive cells were Bcl11b-positive in the VSN layer (Fig. 1D). SCG10 is expressed in immature neurons after terminal differentiation in the VNE and in the MOE (Camoletto et al., 2001). Most SCG10-positive cells highly expressed Bcl11b (Fig. 1E). The same tendency was observed with another immature neuron marker, GAP43 (Fig. 1F). Expression of OMP partially overlapped with Bcl11b expression (Fig. 1G). At P14, the same pattern of coexpression for Bcl11b and differentiation markers was observed as at P0 (Fig. 1H–M). Because at this developmental time point expression of marker genes for neuronal precursors and immature neurons shifted to the marginal regions of the VNE, the expression of Bcl11b also became gradually restricted to these regions (Fig. 1). The distribution of SCG10-expressing cells overlapped well with high-Bcl11b-expressing cells in the marginal region of the VNE (Fig. 1K). In contrast, most of the fully matured OMP-expressing cells were located in the central region of the VNE, which was complementary to the location of high-Bcl11b-expressing cells (Fig. 1M). These results indicated that Bcl11b is first expressed between the late neuronal precursor stage and the immature neuron stage, and is highly expressed in postmitotic immature neurons. However, its expression level is gradually downregulated during mature differentiation. Additionally, double-label fluorescent IHC against the proliferation marker Ki67 (Schlüter et al., 1993) and Bcl11b showed that a small population of Ki67-positive cells is also positive for Bcl11b at P0 (Fig. 1N) and P14 (Fig. 1O). Yet, most Bcl11b-positive cells are Ki67 negative. These results supported the above observations.
Because a few Bcl11b-expressing cells coexpressed Ngn1 and NeuroD, but not Mash1, Bcl11b is probably located downstream of Mash1 in the VSN lineage. To test this possibility, we analyzed the expression of Bcl11b in Mash1−/− VNE at E18.5 and found that no Bcl11b expression was observed (Fig. 2A). Therefore, Bcl11b might function downstream of Mash1 in the VSN lineage. Mouse VSNs are subdivided into two predominant types of neurons, the apical Gαi2 and V1r-positive neurons and the basal Gαo and V2r-positive neurons. To examine the VSN type-dependent expression of Bcl11b, we performed two-color ISH of Bcl11b with Gαi2, V1rd16 and Gαo, V2ra at P0. ISH showed that Bcl11b was expressed in both types of VSNs (Fig. 2B).
Expression of Bcl11b in the accessory olfactory bulb
We next examined Bcl11b expression in the developing AOB, which is the target of the axonal projections of VSNs by ISH, because expression in the embryonic OB was previously reported (Leid et al., 2004). Bcl11b was strongly expressed in the mitral/tufted cell layer (M/TCL) and the granule cell layer (GCL) of the AOB during embryogenesis, and the expression gradually decreased after birth (Fig. 3A). After P14, weak expression of Bcl11b in the AOB was detected in the glomerular layer (GL) in addition to the M/TCL and the GCL (Fig. 3A, arrowheads). Therefore, Bcl11b shows spatially and temporally restricted expression patterns in the AOB and VNE.
To further characterize Bcl11b-expressing cells in the AOB, we performed double-label fluorescent IHC on sagittal sections using antibodies against cell-type-specific proteins. Tbx21 is specifically detected in the projection neurons, the mitral/tufted cells in the main olfactory bulb and the AOB (Faedo et al., 2002; Yoshihara et al., 2005). At P0, we observed anti-Tbx21 immunostaining in the M/TCL only in the anterior portion of the AOB, but not in the posterior portion (Fig. 3B). Tbx21-positive cells overlapped with Bcl11b immunoreactivity in the anterior AOB (Fig. 3B,D). At P14, anti-Tbx21-labeled cells were detected in the M/TCL of both the anterior and posterior portions of AOB and now colabeled with the anti-Bcl11b antibody (Fig. 3G,I), indicating that Bcl11b-positive cells in the M/TCL are mitral/tufted cells. To characterize the Bcl11b-positive cells in the GL and GCL, coimmunostaining with anti-GABA antibody was conducted because there are many GABAergic interneurons in the GL and the GCL of the AOB (Takami et al., 1992). In addition to the M/TCL, Bcl11b protein was detected in the GCL at P0 and in both the GCL and the GL at P14. At P0, some Bcl11b-positive cells in the GCL overlapped with GABA-positive interneurons (Fig. 3C,E). At P14, the immunostaining signal of the anti-GABA antibody was weaker compared with P0. Some Bcl11b-positive cells were also GABA positive in the GCL (Fig. 3H,J) and in the GL (Fig. 3H,K), indicating that at least a subpopulation of Bcl11b-positive cells in the GCL and the GL are GABAergic interneurons.
Our coexpression analysis revealed that the expression of Bcl11b changed dynamically during the development of the vomeronasal system in a temporally and spatially restricted manner. The strong expression of Bcl11b during embryogenesis suggests that Bcl11b might play important roles in the development of the vomeronasal system in mice.
Impaired axonal projection of VSNs to the AOB in Bcl11b-deficient mice
To investigate whether Bcl11b has an irreplaceable role in development of the vomeronasal system, we analyzed Bcl11b−/− mice at P0 and during embryonic stages because Bcl11b−/− mice die within a day after birth. First, we analyzed Bcl11b−/− mice by hematoxylin-eosin staining of coronal sections of the VNE and sagittal sections of the AOB. In the VNE, there were no morphological differences between Bcl11b−/− and wild-type mice throughout the developmental stages analyzed (Fig. 4A,B). Quantification of the size and cell numbers in the VNE were nearly identical between Bcl11b−/− and wild-type mice at P0 (size: 0.102 ± 0.010 mm2/section in Bcl11b−/− and 0.110 ± 0.0060 mm2/section in wild-type mice, p = 0.21; cell numbers: 1879 ± 118 cells/section in Bcl11b−/− and 1955 ± 148 cells/section in wild-type mice, p = 0.53; n = 3), which suggests that the VNE develops grossly normal in Bcl11b−/− mice.
In contrast, differences in morphology between the Bcl11b−/− and wild-type animals gradually became obvious during development of the AOB (Fig. 4C). There was no difference in the morphology of the AOB between Bcl11b−/− and wild-type mice at E15.5, a time point at which the nerve layer is not yet observed. As the AOB is innervated by VSN axons, morphological differences become obvious. In wild-type mice, the AOB was clearly divided into five layers comprising the vomeronasal nerve layer (VNL)/GL, the external plexiform layer (EPL), the M/TCL, the internal plexiform layer/the lateral olfactory tract (IPL/LOT), and the GCL at P0, but the Bcl11b−/− AOB lacked the VNL/GL (Fig. 4C). To confirm this observation, we performed IHC using antibodies against Tbx21 for mitral/tufted cells and against NCAM for VSN axons (Fig. 4D). In wild-type mice, as axons reach the AOB to form the VNL/GL, mitral/tufted cells shifted to deeper layers to form the M/TCL. In contrast, few VSN axons reached to the Bcl11b−/− AOB; therefore, the mitral/tufted cells occupied a broad part of the AOB that corresponds to the VNL/GL, the EPL, and the M/TCL in wild-type mice (Fig. 4D). This impaired axonal projection was confirmed with double immunostaining for a presynaptic molecule of VSNs and a dendritic marker of mitral/tufted cells using anti-synaptophysin and anti-pcdh21 antibodies, respectively. In wild-type mice, these immunostains were overlapped in the boundary between a synaptophysin-positive region and a pcdh21-region. In the Bcl11b−/− AOB, however, less immunostaining for anti-synaptophysin antibody was observed, confirming that fewer axonal termini reached to the VNL/GL in the AOB (Fig. 4E).
Next, to visualize the axonal projections of VSNs to the AOB in their entirety, we performed DiI-tracing experiments by injecting a DiI-coated glass capillary into the lumen of the VNO. Figure 4F shows the sagittally dissected OB/AOB and nasal cavity and the top view of the AOB. In wild-type mice, axons were fasciculated to form several axon bundles that reached both the anterior and the posterior parts of the AOB. In Bcl11b−/− mice, DiI fluorescence was dimmer, and the axon bundles were narrower compared with the wild-type mice. Some axon bundles were directed predominantly to the anterior AOB rather than the posterior AOB. Other axon bundles were misdirected to somewhere in the OB. These results indicated that a Bcl11b deficiency caused a significant reduction in the axonal outgrowth from VSNs and impaired axonal projections to the AOB.
To further investigate whether there were any abnormalities caused by loss of Bcl11b activity in the AOB, we examined the production and survival of mitral/tufted cells, which are axonal target of VSNs. In the M/TCL, Tbx21-positive cells were observed in both Bcl11b−/− and wild-type mice at P0, and there was no obvious difference in the number of Tbx21-positive cells between Bcl11b−/− and wild-type mice (141 ± 27.0 cells/section in Bcl11b−/− and 137 ± 14.4 cells/section in wild-type mice, p = 0.82; n = 3), indicating that the primary postsynaptic neurons of VSNs were produced normally in Bcl11b−/− mice (Fig. 4D). Additionally, we observed few apoptotic cells in the AOB of both Bcl11b−/− and wild-type mice. These results indicated that the loss of Bcl11b function did not cause fatal defects in the production and survival of mitral/tufted cells in the AOB.
Incomplete development of VSNs in Bcl11b-deficient mice
The impaired axonal outgrowth was indicative of the defective development of VSNs in Bcl11b−/− mice. Therefore, we examined differentiation of VSNs in Bcl11b−/− mice at P0 by ISH using RNA probes for neuronal marker genes and expression levels of these genes with microarray analyses. The expression of Mash1 in Bcl11b−/− was almost the same as in wild-type mice (Fig. 5A) (signal intensity of microarray analyses of probe set for Mash1, Probe ID, 1437086_at: 313 ± 94.3 in Bcl11b−/−, n = 6, and 245 ± 40.4 in wild-type, n = 5, p = 0.17), indicating that the generation of progenitor cells proceeds normally. Expression of Ngn1 and NeuroD were slightly but significantly increased in Bcl11b−/− (Fig. 5A) (Ngn1, 1438551_at: 958 ± 354 in Bcl11b−/− and 494 ± 86.5 in wild-type mice, p = 0.020; NeuroD, 1426412_at: 5476 ± 803 in Bcl11b−/− and 3541 ± 775 in wild-type mice, p = 0.0030). There was no significant difference in the pan-neuronal/immature neuron marker gene, SCG10, between Bcl11b−/− and wild-type mice (Fig. 5A) (SCG10, 1423280_at: 6671 ± 475 in Bcl11b−/− and 7025 ± 1276 in wild-type, p = 0.053), indicating that the terminal differentiation into neurons is not blocked. The expression of GAP43 was decreased (Fig. 5A) (GAP43, 1423537_at: 1753 ± 192 in Bcl11b−/− and 2774 ± 556 in wild-type mice, p = 0.0021, n = 5). The expression of OMP was severely reduced in Bcl11b−/− mice (Fig. 5A) (OMP, 1422200_at: 49.9 ± 16.6 in Bcl11b−/− and 189 ± 48.3 in wild-type mice, p = 0.00010), which indicates that the differentiation of VSNs is arrested in postmitotic neurons in Bcl11b−/− mice.
We extended our analyses to the fetal VNE. From E12.5 to E15.5 (Fig. 5B,C), when few OMP-positive mature neurons were observed in wild-type embryos, no obvious differences in the expression of Mash1, Ngn1, NeuroD, SCG10, and GAP43 were observed between Bcl11b−/− and wild-type embryos. These results indicate that the production of progenitor/precursor cells and the terminal differentiation of VSNs proceed normally during early fetal development. After E16.5, when a severe reduction in OMP expression became evident in Bcl11b−/− compared with wild-type mice, Ngn1- and NeuroD-expressing cells gradually increased in Bcl11b−/− mice. These results suggest that the slightly increased expression of Ngn1 and NeuroD after E16.5 are not due to a direct effect of the loss of function of Bcl11b. The increased expression is probably due to secondary effects that are caused by the reduction of mature VSNs, such as a reduction of feedback inhibition signals of neurogenesis from mature VSNs (Kawauchi et al., 2009).
Increased numbers of proliferative and apoptotic cells in Bcl11b-deficient mice
The morphological normality of the VNE, the upregulation of Ngn1 and NeuroD expression, and the profound defects in mature differentiation of VSNs in Bcl11b−/− mice at P0 suggested an alteration of the balance of proliferation and apoptosis in the VNE. To test this hypothesis, we quantified proliferative cells and apoptotic cells by immunostaining using anti-Ki67 and anti-active caspase-3 antibodies, respectively (Fig. 6A,B). The number of proliferative cells was increased by 20% (303 ± 11 cells/section in Bcl11b−/− and 253 ± 14 cells/section in wild-type mice, p = 0.0082, n = 3), and the number of apoptotic cells was increased by 820% (59 ± 13 cells/section in Bcl11b−/− and 6.4 ± 0.79 cells/section in wild-type mice, p = 0.0022, n = 3) at P0 (Fig. 6B). The normal morphology and cellular numbers in the VNE of Bcl11b−/− mice (Fig. 4A,B) could be due to an increased number of both apoptotic and proliferative cells (i.e., the net balance of dividing and dying cells might be almost the same as in the wild-type mice).
To define the step in VSN differentiation that apoptosis occurs in Bcl11b−/− mice, we conducted double-immunostaining with anti-Ki67 and anti-active caspase-3 antibodies (Fig. 6C), and we performed ISH using GAP43 or OMP probes combined with the immunostaining of active caspase-3 (Fig. 6D,E). Active caspase-3-positive cells did not coimmunostain with the anti-Ki67 antibody, which indicates that proliferative cells hardly undergo apoptosis in Bcl11b−/− and wild-type mice. Apoptotic cells were predominantly observed in GAP43-positive cells in Bcl11b−/− and wild-type mice (Fig. 6D). The number of GAP43-positive apoptotic cells increased 10-fold in the Bcl11b−/− VNE (29 ± 6.1 cells/section in Bcl11b−/− and 2.6 ± 0.73 cells/section in wild-type mice, p = 0.0018, n = 3). In contrast, no significant difference in the number of OMP-positive apoptotic cells was observed between the Bcl11b−/− and wild-type VNE (1.1 ± 0.22 cells/section in Bcl11b−/− and 1.1 ± 0.35 cells/section in wild-type VNE, p = 0.87, n = 3) (Fig. 6E). Therefore, apoptosis occurs predominantly in the immature neuron stage in Bcl11b−/− mice that results in a severe reduction in the number of mature VSNs.
Identification of genes associated with the abnormal differentiation of VSNs in Bcl11b-deficient mice
Because Bcl11b functions as either a transcription activator or a repressor, depending on the cell type (Senawong et al., 2003; Topark-Ngarm et al., 2006; Marban et al., 2007; Cismasiu et al., 2008; Cherrier et al., 2009), a null mutation should result in a decrease or an increase in the expression of target and downstream genes. To identify genes whose expressions are associated with abnormalities in VSNs in Bcl11b-deficient mice, we directly compared gene expression profiles between the Bcl11b−/− and wild-type VNOs at P0 using DNA microarray analysis and selected the most significantly downregulated or upregulated genes for further analysis. Because there were multiple cell types that were included in the dissected VNOs, we verified the microarray data using ISH and excluded those genes that were expressed in areas other than the VNE. We first found that the expression of both V1rs and V2rs, as indicated by the microarray data, was significantly reduced in Bcl11b−/− mice (Table 1), suggesting that the abnormal differentiation of VSNs might associate with a reduction in the expression of VR genes. Additionally, we identified eight genes that were either specifically downregulated (Vil1, Panx3, Cart, Big2/Contactin4, Mef2b, and Tcfap2e) (Table 1, Fig. 7A) or upregulated (Meis2 and Olig1) (Table 1, Fig. 7B) in the VNE of Bcl11b−/− mice. Among the downregulated genes identified, ISH analyses revealed that Vil1, Panx3, Cart, Big2/Contactin4, and Mef2b were expressed in both the apical and basal VSN layer, but Tcfap2e was expressed only in the basal VSN layer in wild-type mice (Fig. 7A). Because Big2/Contactin4 functions as an axonal guidance molecule in the main olfactory system, a severe reduction in Big2/Contactin4 could be involved in the phenotype of defective axonal projection of VSNs in Bcl11b−/− mice. Of the upregulated genes, Meis2 was expressed in the apical VSN layer in wild-type mice. Its expression was significantly increased and was detected in both the apical and basal portions in the Bcl11b−/− VNE (Fig. 7B). The transcription factor Olig1 was not expressed in wild-type mice. However, it was expressed in Bcl11b−/− mice, which indicates that the expression of Olig1 in the VNE is allowed by the absence of Bcl11b function. Therefore, we identified eight candidate genes, in addition to the V1rs and V2rs, that may be located downstream of Bcl11b. Moreover, we identified Tcfap2e and Meis2 as novel marker genes that are specifically expressed in the basal and apical VSNs, respectively. Interestingly, the expression of the novel basal marker Tcfap2e was downregulated, whereas the expression of the apical marker Meis2 was upregulated in Bcl11b−/− mice, which suggest that the deficiency of Bcl11b results in a decrease in basal VSNs and an increase in apical VSNs. These data demonstrate that the loss of Bcl11b function causes substantial changes in gene expression in the VNE and also provides a molecular foundation for the understanding of the function of Bcl11b in VSN differentiation.
Severely reduced expression of V1r and V2r genes in Bcl11b-deficient mice
To confirm the microarray data, we examined VR gene expression in Bcl11b−/− mice. We performed ISH using probes for the V1r family of V1rb1 (Vmn1r50), V1rd16 (Vmn1r180), and V1re4 (Vmn1r232); and for the V2r family of V2ra (Vmn2r8-17, Vmn2r84–89, and Vmn2r121), V2rb (Vmn2r28-52), and V2rc (Vmn2r91-110). Quantification of the number of VR-expressing cells revealed that V1rs- and V2rs-expressing cells were severely decreased in Bcl11b−/− mice at P0 [number of cells expressing V1rb1 (in cells/section) as follows: 0.63 ± 0.23 in Bcl11b−/− vs 2.0 ± 0.13 in wild-type, p = 0.0024; expressing V1rd16: 1.8 ± 0.42 vs 9.6 ± 2.3, p = 0.025; expressing V1re4: 0.47 ± 0.06 vs 1.3 ± 0.03, p = 0.00027; expressing V2ra: 3.4 ± 0.79 vs 31 ± 6.4, p = 0.0019; expressing V2rb: 3.2 ± 0.93 vs 49 ± 6.2, p = 0.00023; expressing V2rc: 1.6 ± 1.0 vs 15 ± 2.1, p = 0.0023] (Fig. 8A,B). The impact of Bcl11b deficiency on the reduction of V2rs-expressing cells (89–94% reduction in Bcl11b−/−) is larger than on the reduction of V1rs-expressing cells (64–81% reduction). The defective expression of VR genes was also observed in Bcl11b−/− mice at E14.5 before the contact of vomeronasal axonal termini with the AOB (Fig. 8A,B) (V1rd16: none in Bcl11b−/− and 0.57 ± 0.41 cells/section in wild-type mice; V2ra: none in Bcl11b−/− and 1.5 ± 0.66 cells/section in wild-type mice), and at E16.5 (Fig. 8A,B) (V1rd16: 1.2 ± 0.57 in Bcl11b−/− and 5.6 ± 0.84 cells/section in wild-type mice, p = 0.0017; V2ra: 0.058 ± 0.056 in Bcl11b−/− and 17 ± 1.6 cells/section in wild-type mice, p = 0.000051), suggesting that the decrease in the number of VR-expressing cells in Bcl11b−/− mice is a cell-autonomous effect of the loss of function of Bcl11b.
VR-expressing cells in Bcl11b-deficient mice do not undergo apoptosis
As described above, we found a profound defect of VR gene expression in Bcl11b−/− mice, in addition to the increased apoptosis during the immature stage and incomplete mature differentiation of VSNs. To examine whether VR-expressing VSNs underwent apoptosis, we performed a combination of ISH and IHC using probes for V1rd16 and V2ra and an anti-active caspase-3 antibody, respectively. In both Bcl11b−/− and wild-type mice, none of the V1rd16-positive and V2ra-positive cells overlapped with anti-active caspase-3 immunostaining, which indicates that neither V1rd16-expressing nor V2ra-expressing cells undergo apoptosis (Fig. 8C) (number of V1rd16+ cells counted: 169 in Bcl11b−/− vs 881 in wild-type mice; number of V2ra+ cells counted: 128 vs 2540, respectively; n = 2). Therefore, increased apoptosis most likely occurs in non-receptor-expressing immature neurons in Bcl11b−/− mice. In addition, two-color ISH using probes for V1rd16 and V2ra vs OMP showed that most of the V1rd16- or V2ra-expressing cells were OMP positive (Fig. 8D), indicating that VRs-expressing VSNs have the potential to differentiate into mature VSNs.
Bcl11b-mediated regulation of the dichotomy of VSNs
During the quantification of the number of VRs-expressing neurons in the VNE of Bcl11b−/− mice, we found an abnormal distribution of these cells in Bcl11b−/− mice compared with wild-type mice. The V1rd16-expressing neurons were distributed broadly in both the apical and basal VSN layer in Bcl11b−/− mice (Fig. 8A), whereas the V2ra-expressing cells were restricted more basally (Fig. 8B). Together with the microarray and ISH results showing the increased and broadening expression of the apical marker, Meis2, and the decrease in the basal marker, Tcfap2e, in Bcl11b−/− mice, it is conceivable that loss of Bcl11b function may differentially affect the differentiation pathway to the two VSN types (i.e., an increase in the apical VSN lineage and a decrease in the basal VSN lineage). To confirm this hypothesis, we investigated the expression of the G-protein α subunits, Gαi2 and Gαo, which were specifically coexpressed in nonoverlapping layers in the adult VNE with V1rs and V2rs, respectively. The expression of Gαi2 was barely detectable at E14.5 in both the Bcl11b−/− and wild-type VNEs (Fig. 9A). At E16.5 and P0, the expression of Gαi2 was observed in both Bcl11b−/− and wild-type mice, but its pattern was different. In wild-type mice, Gαi2 was expressed in the apical VSN layer. In Bcl11b−/− mice, however, Gαi2-expressing neurons increased in number and were broadly distributed in both the apical and basal VSN layers (Fig. 9A). On the other hand, the expression of Gαo was already detectable at E14.5 (Fig. 9B), suggesting that the onset of Gαo expression is earlier than that of Gαi2. At E16.5 and P0, Gαo-expressing neurons were broadly distributed in both the Bcl11b−/− and wild-type VNEs, but the expression level of Gαo was obviously reduced in the Bcl11b−/− VNE (Fig. 9B).
To further characterize Gαi2- and Gαo-expressing cells, we performed two-color ISH for Gαi2 and Gαo (Fig. 9C). In wild-type mice at P0, we found that Gαi2-expressing VSNs coexpressed Gαo, and these Gαi2/Gαo double-positive neurons were distributed in the apical VSN layer, whereas Gαo single-positive VSNs were located in the basal VSN layer. In contrast, this layer organization was not observed in Bcl11b−/− mice. In Bcl11b−/− mice, Gαi2-expressing VSNs also coexpressed Gαo, and the number of Gαi2/Gαo double-positive neurons was significantly increased (685 ± 148 neurons/section in Bcl11b−/− and 402 ± 38.2 neurons/section in wild-type mice, p = 0.033; n = 3), but Gαo single-positive neurons were severely decreased (26.9 ± 6.05 neurons/section in Bcl11b−/− and 656 ± 74.3 neurons/section in wild-type mice, p = 0.0044; n = 3) (Fig. 9C). Most interestingly, two-color ISH revealed that Gαi2-positive cells were always Gαo positive in both Bcl11b−/− and wild-type mice at P0 (Fig. 9C). This coexpression was also observed in the marginal regions of the adult VNE, where proliferating cells and postmitotic immature neurons are located (Fig. 9D). Because there are no Gαi2 single-positive VSNs and there are already V1r-positive VSNs that exist at P0, it is possible to speculate that Gαi2/Gαo double-positive cells are the VSNs that are committed to the V1r-type lineage. To test this possibility, we examined coexpression of V1rd16 with either Gαo or Gαi2 in wild-type mice at P0. Indeed, V1rd16-expressing VSNs coexpressed Gαo (63 Gαo-positive VSNs/63 V1rd16-positive VSNs) as well as Gαi2 (261 Gαi2-positive VSNs/263 V1rd16-positive VSNs) (Fig. 9E, left). In contrast, we found that V2ra-expressing VSNs only coexpressed Gαo (68 Gαo-positive VSNs/68 V2ra-positive VSNs), but not Gαi2 (0 Gαi2-positive VSN/242 V2ra-positive VSNs) (Fig. 9E, right). These results indicate that Gαi2/Gαo double-positive VSNs commit to the V1r-VSN lineage, and V1r/Gαi2 type of VSNs would be differentiated via the Gαi2/Gαo double-positive stage. In Bcl11b−/− mice, the number of cells in the VNE and the expression of the pan-neuronal/immature neuron marker, SCG10, were unaltered compared with wild-type mice (Figs. 4B, 5). These results suggest that the terminal differentiation into neurons proceeds normally, and almost the same number of VSNs is produced in Bcl11b−/− and wild-type mice. In Bcl11b−/− mice, the number of Gαi2-positive (Gαi2/Gαo double-positive) cells increased, whereas the number of Gαo single-positive cells decreased compared with that of wild-type mice, which demonstrates the opposite impact of Bcl11b deficiency on the generation of the two types of VSNs. Therefore, these results indicate that Bcl11b functions in the step that determined the fate of VSN types.
Discussion
Here, we report that Bcl11b is expressed in the vomeronasal system of mice and is required for its proper development. We demonstrate that the loss of Bcl11b results in impaired axonal projections of VSNs, a significant reduction in the expression of vomeronasal receptor genes, and increased apoptosis and defective differentiation of VSNs. Interestingly, the loss of Bcl11b function affects the fate choice between the two types of VSNs, Gαi2- and Gαo-positive neurons, in opposite ways. Together, these data indicate that Bcl11b plays an essential role in the development of the vomeronasal system and in the regulation of the fate choice of VSNs.
Bcl11b is required for the survival and differentiation of vomeronasal sensory neurons
The molecular mechanisms that control the developmental processes of the VSN have not been well characterized. Genetic studies by Murray et al. (2003) have demonstrated that Mash1 is a determining factor for the neuronal lineage of VSN in neurogenesis and suggested that the molecular details of neurogenesis in the VNE and MOE are similar. In the MOE, the Mash1 > Ngn1 > NeuroD genetic pathway is involved in neuronal differentiation (Cau et al., 2002). We showed that these bHLH transcription factors were expressed in the VNE, and the expression patterns were similar to the expression patterns in the MOE. These results suggest that the genetic pathway is conserved during the neuronal differentiation of VSNs because it is comparable to the pathway in the OSN that was previously proposed (Murray et al., 2003; Suzuki et al., 2003). This study adds Bcl11b as one of the important transcription factors that regulate the differentiation of VSNs.
During the process of differentiation of VSNs, Bcl11b is highly expressed in SCG10-positive postmitotic immature neurons and is decreased in mature neurons. This stage-specific expression of Bcl11b during neuronal differentiation is similar to that in the neocortex and the striatum, where Bcl11b is expressed at high levels in postmitotic neurons but not in progenitor cells (Arlotta et al., 2005, 2008) and plays important roles in postmitotic mature differentiation. In the Bcl11b−/− VNE, the correct number of neurons is produced, but most of these neurons fail to differentiate to maturity, which leads to increased apoptosis in GAP43-positive immature neurons. Therefore, the observed abnormalities in VSN differentiation were most likely caused by defects in postmitotic maturation rather than by an earlier defect in precursor cells. These results indicate that Bcl11b is required for the survival and mature differentiation of postmitotic VSNs.
The molecular mechanisms that underlie the increased apoptosis, impaired axonal projections and defective mature differentiation of VSNs in Bcl11b−/− mice remain unclear. Because Bcl11b is expressed in both VSNs and their axonal target, the AOB, the phenotypes observed in VSNs could be due to a cell-autonomous effect of the loss of Bcl11b function, a non-cell-autonomous effect, or both. In non-cell-autonomous mechanisms, the Bcl11b−/− AOB may somehow influence the mature differentiation of VSNs and/or cause apoptosis. To reveal this, it will be interesting to perform rescue experiments by forced expression of Bcl11b specifically in either VSNs or in mitral/tufted cells in the AOB. Nevertheless, we speculate that a non-cell-autonomous effect is unlikely due to the following reasons: (1) there are no obvious defects observed in cellular composition and survival in the AOB, except for the lack of the vomeronasal nerve layer; and (2) in the closely related main olfactory system, genetic or surgical removal of the OB, which is the axonal target of olfactory sensory neurons, does not significantly influence cellular proliferation, the differentiation of the olfactory epithelium, or odorant receptor gene expression (Graziadei et al., 1978; Sullivan et al., 1995; St John et al., 2003).
In the immune system, the germline deletion of Bcl11b results in impaired T-cell development around CD4/CD8 double-negative stages 2 and 3 and defective T-cell receptor gene expression due to unsuccessful recombination, which consequently causes apoptosis (Wakabayashi et al., 2003b; Ikawa et al., 2010; Li et al., 2010a,b). As designed for the detection of foreign molecules, the vomeronasal system shares similarities with the immune system. Both systems are responsible for self- vs non-self-recognition by expressing one receptor per cell from a large repertoire of receptor genes in a monoallelic and mutually exclusive manner. It is conceivable that Bcl11b may regulate this common and unique mode of gene expression. In future experiments, it will be interesting to investigate genomic region tethering of the transcription factor, Bcl11b, using chromatin immunoprecipitation (ChIP) on DNA chip analysis and ChIP sequencing analysis in combination with microarray data to understand the molecular mechanism that underlies the Bcl11b-mediated regulation of VSN development.
Bcl11b regulates the fate choice between two types of vomeronasal sensory neurons
VSNs can be classified into two major types of neurons: V1r/Gαi2-positive and V2r/Gαo-positive. The fact that both types of VSNs are produced by progenitors in a Mash1-dependent manner and develop side by side in the VNE has raised the question of when their fates are determined and how this dichotomy is regulated. A recent study that used BrdU pulse-labeling experiments suggested that Gαi2- and Gαo-positive neurons mature independently of each other (de la Rosa-Prieto et al., 2010). Our observations provide the different and novel view that the dichotomy of VSNs is regulated by Bcl11b postmitotically. First, we found that all Gαi2-expressing neurons coexpressed Gαo in the embryonic VNE (Fig. 9C), and that this coexpression was observed in the marginal region of the adult VNE, an area where newly differentiated and postmitotic immature neurons are located. Second, Gαi2/Gαo double-positive VSNs expressed V1r, but not V2r, at P0, indicating commitment to the V1r/Gαi2 VSN lineage. These results suggest that V1r/Gαi2-type VSNs are generated via Gαi2/Gαo double-positive neurons. Third, a small number of Gαo-expressing cells colabeled with the differentiating/postmitotic neuron marker NeuroD, but none of the Gαi2-expressing cells were NeuroD-positive in the VNE (data not shown), indicating that the expression of Gαo is earlier than Gαi2 in the course of the differentiation of VSNs. Fourth, we detected the expression of Gαo at E14.5. However, the expression of Gαi2 was observed later, around E16.5, which indicates that Gαo single-positive cells exist solely in the early development of the VNE. Therefore, it is likely that the early postmitotic neuron first expresses Gαo, and then the Gαi2/Gαo-positive cell is generated for the V1r/Gαi2 VSN lineage. It is conceivable that Gαo-positive neurons may be the default type of VSNs. It is interesting that phylogenetically ancient animals, such as fish and frogs, have larger numbers of V2r genes than V1r genes (Shi and Zhang, 2007), which suggests that they predominantly use and develop V2r/Gαo VSNs.
In addition, we demonstrated that the loss of Bcl11b function affects the number of two types of VSNs in opposite ways (i.e., the number of Gαi2/Gαo double-positive apical VSNs increases, but the number of Gαo single-positive basal VSNs decreases). This phenotype is supported by the additional observation of the expression of Meis2 and Tcfap2e, which are novel identifiers of apical and basal VSNs, respectively. Because the number of cells in the VNE was unaltered in Bcl11b−/− mice, it is likely that the balance of the fate choice between the V1r/Gαi2 and V2r/Gαo VSNs was disturbed by the loss of Bcl11b activity. These results led us to propose a model in which the dichotomy of VSNs is regulated by Bcl11b (Fig. 10). In this model, Bcl11b regulates the fate choice in early postmitotic neurons to determine the type of VSN, either Gαi2-positive or Gαo-positive, by suppressing the Gαi2-positive pathway and/or activating the pathway for the Gαo-positive VSNs. The role of Bcl11b in fate determination has been demonstrated in other systems. In the immune system, Bcl11b is required for early T-cell lineage commitment by the suppression of other pathways for the determination of the fates of natural killer, myeloid, and dendritic cells (Wakabayashi et al., 2003a; Ikawa et al., 2010; Li et al., 2010a,b). In the CNS, Bcl11b is involved in the fate choice of subcortical projection neurons (Chen et al., 2008) and in the pathways of medium spiny neuron specification and differentiation (Arlotta et al., 2008), which suggests a common function of Bcl11b in fate choice during differentiation in different systems by the control of the expression of downstream genes.
Notes
Supplemental material for this article is available at http://www.hirota.bio.titech.ac.jp/publication/JNeuroscience_2011Bcl11b.pdf. Additional images of Bcl11b expression analysis and coexpression analysis of NeuroD with Gαo and Gαi2 are presented. This material has not been peer reviewed.
Footnotes
This work was supported in part by the Ministry of Education, Culture, Sports, Science and Technology of Japan, Grant-in-Aid for Young Scientists (B) to M.O, Grant-in-Aid for Young Scientists (A) to I.M., and Grants-in-Aid for Scientific Research (C) to J.H.; by the National Institute on Deafness and Other Communication Disorders (Grant R03 DC011143) to I.M.; by the Japan Society for the Promotion of Science, Grant-in-Aid for Scientific Research on Innovative Areas to J.H.; and by grants from the Senri Life Science Foundation, the Inamori Foundation, and the Sumitomo Foundation to J.H. We thank Drs. P. Mombaerts and Y. Yoshihara for providing reagents, Dr. S. Fuss for instructions on the DiI experiment and critical review of the manuscript, and the members of the Hirota laboratory for their continuous support.
- Correspondence should be addressed to Junji Hirota at the above address. jhirota{at}bio.titech.ac.jp