Morpholinos for splice modificatio

Morpholinos for splice modification

Advertisement

Mash1 and Ngn1 control distinct steps of determination and differentiation in the olfactory sensory neuron lineage
Elise Cau, Simona Casarosa, François Guillemot

Summary

bHLH transcription factors are expressed sequentially during the development of neural lineages, suggesting that they operate in genetic cascades. In the olfactory epithelium, the proneural genes Mash1 and neurogenin1 are expressed at distinct steps in the same olfactory sensory neuron lineage. Here, we show by loss-of-function analysis that both genes are required for the generation of olfactory sensory neurons. However, their mutant phenotypes are strikingly different, indicating that they have divergent functions. In Mash1 null mutant mice, olfactory progenitors are not produced and the Notch signalling pathway is not activated, establishing Mash1 as a determination gene for olfactory sensory neurons. In neurogenin1 null mutant mice, olfactory progenitors are generated but they express only a subset of their normal repertoire of regulatory molecules and their differentiation is blocked. Thus neurogenin1 is required for the activation of one of several parallel genetic programs functioning downstream of Mash1 in the differentiation of olfactory sensory neurons. These results illustrate the versatility of neural bHLH genes which adopt either a determination or a differentiation function, depending primarily on the timing of their expression in neural progenitors.

INTRODUCTION

Cell determination, i.e. the irreversible commitment of progenitor cells to a particular fate is an essential early step in the development of cell lineages, and current evidence indicates that common mechanisms operate in different tissues to regulate this step. In particular, proteins of the basic helix-loop-helix (bHLH) class play a central role in the determination of different cell types, including muscle cells and nerve cells (Weintraub, 1993; Jan and Jan, 1994). bHLH proteins are transcription factors that make specific contact with DNA through a basic region and heterodimerize through an HLH domain (Murre et al., 1989). Tissue-specific bHLH factors usually exist as families of related proteins, which are often expressed transiently at different steps in the differentiation of cell lineages. Genetic experiments have shown that such sequential expression of bHLH proteins represent regulatory cascades whereby early expressed proteins activate the expression of later expressed ones. Hence, in vertebrate muscle lineages, the bHLH genes MyoD (Myod1) and Myf5 are expressed in muscle progenitors, the proliferating myoblasts, and they activate the expression of the bHLH gene myogenin in differentiating myotubes (Molkentin and Olson, 1996). A similar process takes place in vertebrate neural lineages, where the bHLH genes Mash1 (Ascl1) and neurogenins (Ngns), related to the Drosophila proneural genes achaete-scute and atonal, respectively, are expressed in dividing progenitors and induce the expression of other bHLH genes, including NeuroD (Neurod1), in differentiating neurons (Ma et al., 1996; Cau et al., 1997; Fode et al., 1998; Ma et al., 1998). Based on this timing of expression, muscle-specific and neural-specific bHLH proteins have been divided into determination factors (MyoD/Myf5; Mash1/Ngns) and differentiation factors (myogenin; NeuroD). In this model, early expressed factors are involved in the selection of progenitor cells that become competent to acquire defined cell fates and commit to differentiation, while late expressed factors are involved in terminal differentiation of post-mitotic cells. The steps of cell determination and differentiation are temporally separated by a period of expansion of the pool of committed progenitors.

Genetic analysis has lent support to the model of distinct determination and differentiation functions for different bHLH proteins in neural lineages (Lee, 1997). In particular, a null mutation in Ngn2 (Atoh4) results in the elimination of cranial sensory ganglia, due to a lack of production of sensory neuron progenitors by epithelial placodes, accompanied by a failure of placodal cells to express the downstream bHLH gene NeuroD and the Notch receptor ligand Delta1 (delta-like1) (Fode et al., 1998). This defect is reminiscent of the loss-of-function phenotypes of Drosophila proneural genes, which are marked by defects in the delamination of neural progenitors from the neurectoderm and in activation of Notch signalling, a pathway involved in the selection of neural progenitors from groups of equivalent cells (Simpson, 1997; Gridley, 1997; Bray, 1998). Thus, ngns have conserved the neural determination functions of their Drosophila counterparts. Moreover, Ngn2 and Mash1 are involved in the commitment to the neuronal fate and the inhibition of the glial fate in multipotent progenitors of the cerebral cortex (Nieto et al., 2001), indicating that selection of neural progenitors is coupled to a step of lineage restriction in the vertebrate central nervous system. In contrast, mutations in the NeuroD gene result in defects in the cerebellum and hippocampus which are due to the abnormal differentiation and death of cerebellar and dentate granule cells, indicating that NeuroD is required for the differentiation and survival of already determined neuronal progenitors (Miyata et al., 1999; Liu et al., 2000).

Sequential expression of numerous bHLH genes has now been reported in a variety of neural lineages (Cau et al., 1997; Fode et al., 1998; Ma et al., 1998; Perron et al., 1999), but the specific role of each bHLH protein in such regulatory cascades has not yet been systematically addressed. The current models on the roles of neural determination and differentiation genes thus rest on a small amount of data collected in different neural lineages. We thus set out to compare the function of bHLH proteins acting sequentially in the same lineage, choosing for this study the olfactory epithelium (OE), a simple sensory epithelium that contains two populations of progenitor cells: a population of neuroepithelial cells that divide apically and a population of secondary progenitor cells that settle on the basal side of the OE where they continue to divide before differentiating into olfactory sensory neurons (OSNs) (Smart, 1971; Caggiano et al., 1994). The simple structure of the OE has helped to define the following temporal sequence of bHLH gene expression, taking place during the differentiation of OSN progenitor cells: Mash1>Ngn1>NeuroD (Cau et al., 1997). The expression of different bHLH genes at different stages in the OSN lineage raises the possibility that each of these genes has a unique function in the lineage.

To establish the respective functions of Ngn1 and Mash1 in the OE, we have analysed mice carrying null mutations in Ngn1 (Ma et al., 1998) and Mash1 (Guillemot et al., 1993). We show that Mash1 is required for the generation of OSN progenitors and for the activation of Notch signalling in neuroepithelial cells of the OE, therefore establishing Mash1 as a characteristic proneural or determination gene in this tissue. In Ngn1 mutant OE, in contrast, most OSN fail to differentiate initially but basal progenitors are generated and activate Notch signalling, indicating that Ngn1 has characteristics of a differentiation gene for most progenitors of the OE. However, Ngn1 can compensate for the loss of the determination function of Mash1 in a population of early born neurons in the olfactory placode. Therefore the same gene has a determination or a differentiation function in different progenitor populations in the same tissue.

MATERIALS AND METHODS

Generation of mutant mice

Mash1 and Ngn1 mutant embryos were obtained from intercrosses of Mash1+/– (Guillemot et al., 1993) and Ngn1+/– (Ma et al., 1998) mice, respectively. Double mutant embryos were obtained by intercrossing Mash1+/–; Ngn1+/– mice. Genotyping of the Mash1 mutant allele and the Ngn1 mutant allele was as described in Cau et al. (Cau et al., 1997) and Ma et al. (Ma et al., 1998), respectively. For staging of the embryos, mid-day on the day of the appearance of the vaginal plug was considered as E0.5.

In situ hybridisation and immunocytochemistry

Whole-mount RNA in situ hybridisation and hybridisation on cryostat sections were performed as described in Cau et al. (Cau et al., 1997). Probes for the following have been described previously: Mash1, Ngn1, Hes1, NeuroD and SCG10 (Cau et al., 2000), Dll1, Lhx2 and Hes5 (Casarosa et al., 1999), Phd1 (Saito et al., 1997) and Ebf1 (Garel et al., 1997). In situ hybridisation followed by immunohistochemistry was performed as described in Cau et al. (Cau et al., 1997). For double in situ hybridisation, embryos were hybridised with a mixture of two probes labelled with digoxigenin-UTP and fluorescein-UTP respectively (Boehringer Mannheim). After development of the first reaction using NBT/BCIP (Boehringer Mannheim), alkaline phosphatase was inactivated by incubation in glycine (0.1 M, pH 2.2). Embryos were then washed in PBS and blocked before incubation in the antibody (anti-digoxigenin 1/2500 or anti-fluorescein 1/1000; Boehringer Mannheim). INT/BCIP diluted in 0.1 M NaCl; 0.1 M Tris pH 9.5; 50 mM MgCl2; 0.1% Tween 20 was used for the development of the second reaction. Histological preparations and BrdU immunostaining were carried out as described previously (Casarosa et al., 1999). Proliferation was studied at E12.5 by injecting pregnant females intraperitoneally with 2 mg of BrdU 30 minutes prior to sacrifice.

RESULTS

Function of Mash1 and Ngn1 in a unidirectional cascade

To begin to compare the functions of Mash1 and Ngn1 in the OE, we first examined the epistatic relationship between the two genes during development of this tissue. At E12.5, when the OE has begun to acquire its mature layered organisation (Cau et al., 1997; Cau et al., 2000), Mash1 and Ngn1 are expressed in distinct subsets of dividing progenitors which are found at different positions at this stage. While Mash1 is expressed in dividing cells in apical, intermediate and basal positions in the OE, Ngn1 expression is restricted to dividing cells in the basal position (Fig. 1A-C) (Cau et al., 1997). To determine whether the two genes are expressed in the same population of basal progenitor cells, we performed a double labelling experiment using Mash1 and Ngn1 RNA probes. We observed both Mash1-positive, Ngn1-negative cells and Mash1-negative, Ngn1-positive cells, as well as a smaller number of double labelled cells (Fig. 1A, arrowheads). Ngn1 expression was severely reduced in the OE of Mash1 mutant mice at E12.5 (Fig. 1B′) (Cau et al., 1997), while it was unaffected in the olfactory placodes of mutants at E10.0, when the first population of OSN progenitors is generated (data not shown) (Cau et al., 1997). In contrast, Mash1 expression was not significantly affected in the OE of E12.5 Ngn1-null mutant mice (Fig. 1C′). Together, these data indicate that Mash1 and Ngn1 are expressed sequentially at two different progenitor stages in the same lineage, with a transient coexpression of the two genes, and that Mash1 acts upstream of Ngn1 in a unidirectional regulatory cascade in most OSN progenitors at E12.5.

Fig. 1.

Sequential expression of Mash1 and Ngn1 in progenitors of the olfactory epithelium. (A) In situ hybridisation with RNA probes for Mash1 (in blue) and Ngn1 (in brown) on frontal sections of OE at E12.5 shows that the two genes are expressed in distinct and partially overlapping progenitor populations. Mash1 is expressed in cells located in apical, intermediate and basal positions while Ngn1 expression is restricted to basal progenitors. The two genes are co-expressed in a subset of basal progenitors (arrowheads in the inset). The top right panel shows the apical (a) and basal (b) sides on a schematic representation of a OE section. (B,B′) Hybridisation with a Ngn1 probe on E12.5 wild-type (B) and Mash1 mutant (B′) OE showing that Ngn1 expression in basal progenitors requires Mash1 function. (C,C′) Hybridisation with a Mash1 probe on E12.5 wild-type (C) and Ngn1 mutant (C′) OE showing that in contrast, Mash1 expression in OE progenitors is independent of Ngn1 function. Scale bars, 25 μm.

Both Mash1 and Ngn1 are required for olfactory neurogenesis but only Mash1 is required for the generation of olfactory progenitors

We then performed a loss-of-function analysis of Mash1 and Ngn1 to compare the role of the two genes in OE development, using mice carrying null mutations in these two genes (Guillemot et al., 1993; Ma et al., 1998). Labelling with the pan-neuronal marker SCG10 shows that in Mash1 mutant embryos at E12.5, there is a severe depletion of neurons, particularly in the rostral part of the OE, while a significant fraction of SCG10+ neurons persist in the caudal OE, as previously reported (Guillemot et al., 1993; Cau et al., 1997) (Fig. 2A′,B). The first neurons to differentiate in the olfactory placode between E10.0 and E10.5 are also generated in absence of Mash1 function (Cau et al., 1997) (Fig. 2C′).

Fig. 2.

Mash1 and Ngn1 are both required for the production of olfactory sensory neurons but only Mash1 is required for the generation of basal OE progenitors. (A-C,A′-C′,A′′,C′′) In situ hybridisation with a probe for the pan-neuronal marker SCG10 on sections of wild-type (A,C), Mash1 mutant (A′,B,C′) and Ngn1 mutant (A′′,B′,C′′) OE at E12.5 (A-A′′,B,B′) and on whole-mount olfactory placodes at E10.0 (C-C″). There is a drastic reduction in the number of neurons differentiating in the OE of both Mash1 and Ngn1 mutant embryos at E12.5, apparent in the medial part of the OE (A′,A′′) and more pronounced in the rostral part of the Mash1 mutant OE (B) and in the caudal part of the Ngn1 mutant OE (B′). Mash1 is not required for the generation of neurons at placodal stage (C′), while these early born neurons are missing in a Ngn1 mutant placode (C′′). (D-D′′) Immunocytochemistry for BrdU after a short period of incorporation followed by Haematoxylin staining reveals progenitors cells in S phase of the cell cycle (brown nuclei) and in mitosis (mitotic figures indicated by arrowheads) in wild-type (D), Mash1 mutant (D′) and Ngn1 mutant (D′′) OE at E12.5. Dividing progenitors are present in apical (top) and basal (bottom) positions in wild-type OE (D). Basal progenitors are missing in the Mash1 mutant OE (D′). Because post-mitotic neurons are also absent, apical progenitors are now present in the whole thickness of the OE. In contrast, basal progenitors are present in the Ngn1 mutant OE (D′′). Scale bars, 25 μm (A-A′′,B′,B′′); 50 μm (C-C′′); 10 μm (D-D′′).

We examined mice carrying a null mutation of Ngn1 to ask whether this gene is also required for neurogenesis in the OE, as shown previously for cranial and dorsal root sensory ganglia (Ma et al., 1998; Ma et al., 1999). SCG10-positive OSNs were missing in Ngn1 mutant olfactory placodes at E10.5 and were severely reduced in number in mutant OE at E12.5 (Fig. 2A′′,C′′). The loss of OSNs was more severe in the caudal than in the rostral part of the OE (Fig. 2B′). By E15.5, however, the number of ORNs appeared to be similar in wild-type and Ngn1 mutant OE (data not shown), suggesting that the loss of Ngn1 leads to a severe but transient neurogenesis defect at early stages of OE development. Taken together, these results indicate that both Mash1 and Ngn1 have essential functions in OSN development.

OSNs are generated from progenitor cells that initially divide on the apical side of the OE, then progressively lose their apical contacts and translocate to the basal side where they continue to divide (Smart, 1971; Caggiano et al., 1994). We asked whether the loss of OSNs in Mash1 and Ngn1 mutants was due to a failure to generate OSN progenitors, or to defects in their differentiation. Progenitor cells in S-phase of the cell cycle were labelled in E12.5 embryos by incorporation of BrdU for 30 minutes. In wild-type OE, BrdU-positive nuclei were found in apical as well as in basal position (Fig. 2D). Progenitors in mitosis, marked by the presence of mitotic figures, were also observed both apically and basally (Fig. 2D, arrowheads). Strikingly, BrdU-positive nuclei and mitotic figures were absent from the basal side of Mash1 mutant OE, whereas they where normally present on the apical side (Fig. 2D′). Thus, basal OSN progenitors are missing in the Mash1 mutant OE. Since E12.5 is the earliest stage at which a distinct population of dividing cells is observed on the basal side of the OE (Smart, 1971), this data strongly suggests that the Mash1 mutation prevents the generation of this population of OSN progenitors.

In contrast to the absence of basal progenitors in Mash1 mutant OE, BrdU-positive nuclei and mitotic figures were found in both apical and basal positions in Ngn1 mutant OE. The distribution of apical and basal BrdU-positive nuclei was not as distinct as in the wild-type OE, probably because of the absence of an intervening layer of OSNs (Fig. 2D″). The persistence of a basal progenitor population in Ngn1 mutant OE was also inferred from the normal expression of Mash1, which marks a subset of basal progenitors (Cau et al., 1997). Therefore, basal OSN progenitors are generated in absence of Ngn1, indicating that the Ngn1 mutant defect is likely taking place at the level of progenitor differentiation. In contrast, the lack of OSNs in Mash1 mutant OE is due to a failure to generate basal progenitors.

Mash1 and not Ngn1, is required to activate Notch signalling in olfactory progenitors

The lack of OSN progenitors in the Mash1 mutant and not in the Ngn1 mutant OE suggested that Mash1 acts in the OSN lineage at the stage of determination of basal progenitors while Ngn1 acts at the level of their differentiation. An important aspect of the determination function of proneural genes in Drosophila and vertebrates is to induce the expression of ligands of the Notch receptor, and thereby to activate the Notch signalling pathway, which is involved in the selection of neural precursors (Kunisch et al., 1994; Ma et al., 1996; Fode et al., 1998; Ma et al., 1998). To further investigate the function of Mash1 and Ngn1 in the OE, we asked whether the loss of Mash1 and Ngn1 functions affected Notch signalling. In wild-type OE at E12.5, basal progenitors express the Notch ligands Dll3 and Ser2 (Jag2) and the Notch effector gene Hes5 (Fig. 3A,C,E), while the ligand Ser1 (Jag1) and the effector Hes1 are mostly expressed in apical cells (Fig. 3B,D). Ser2 is also expressed strongly in cells in intermediate position and more weakly in apical cells (Fig. 3C). These data suggest that Notch signalling is mediated through distinct ligands and effector molecules in basal and apical cells of the OE. In Mash1 mutant OE, expression of Dll3 and Hes5 was completely abolished at E12.5 (Fig. 3A′,E′), likely reflecting the lack of basal progenitors. Expression of Ser1, Ser2 and Hes1 was also strongly reduced in Mash1 mutant OE, despite the persistence of the apical progenitor population (Fig. 3B′-D′), indicating that Mash1 is required to activate Notch signalling in the OE. In contrast, Ngn1 mutant OE displayed a normal apical expression of Hes1 and Ser1 and a normal basal expression of Dll3, Ser2 and Hes5 (Fig. 3A′′-E′′). Therefore, Mash1 and Ngn1 also differ in their regulation of Notch signalling in the OE, in agreement with the idea that Mash1 has a determination function for OSNs whereas Ngn1 is required in a downstream step for their differentiation.

Fig. 3.

Mash1 and not Ngn1 is required for the expression of Notch signalling genes in the OE. Probes for Notch ligands delta-like 3 (A-A′′), Serrate 1 (B-B′′) and Serrate 2 (C-C′′) and for the Notch bHLH effector genes Hes1 (D-D′′) and Hes5 (E-E′′) hybridised to wild-type (A-E), Mash1 mutant (A′-E′) and Ngn1 mutant (A′′-E′′) OE at E12.5. Expression of Ser1 and Hes1 in apical cells and of Dll3, Ser2 and Hes5 in basal cells is abolished in Mash1 mutant OE. In contrast, expression of these genes is unaffected in Ngn1 mutant OE. Note that Hes1 and Ser1 are also expressed in the mesenchyme surrounding the OE. Scale bar, 25 μm.

Mash1 controls the expression of a larger number of transcription factors than Ngn1 in OSN progenitors

The above results indicated that Ngn1 functions downstream of Mash1 in the OSN progenitors, implying that Ngn1 may mediate some of the activities of Mash1 in this lineage. To address this issue, we examined, in wild-type and mutant OE, the expression of a number of regulatory genes that are candidates to control different aspects of OSN differentiation. The bHLH differentiation gene NeuroD is expressed in basal OSN progenitors after Mash1 and Ngn1 and during the transition between proliferation and differentiation (Cau et al., 1997) (Fig. 4A). At E12.5, expression of NeuroD in the OE was largely eliminated in both Mash1 and Ngn1 mutant embryos (Fig. 4A′,A′′), consistent with NeuroD participating to a regulatory cascade in basal OE progenitors in which Mash1 activates Ngn1 which in turn activates NeuroD. The persistence in Ngn1 mutant OE of basal progenitors (Fig. 2D′′) that fail to activate NeuroD expression (Fig. 4A′′) demonstrates that Ngn1 function is required in basal progenitors to activate a neuronal differentiation program.

Fig. 4.

Ngn1 function is required for the expression of a subset of transcriptional regulators expressed in basal progenitors. Expression of NeuroD (A-A′′), Phd1 (B-B′′), Ebf1 (C-C′′) and Lhx2 (D-D′′) analysed by in situ hybridisation on sections of wild-type (A-D), Mash1 mutant (A′-D′) and Ngn1 mutant (A′′-D′′) OE at E12.5. In wild-type OE, NeuroD is expressed in basal progenitors and only transiently in differentiating neurons (A) (Cau et al., 1997), while Phd1, Ebf1 and Lhx2 are expressed both in dividing basal progenitors and in post-mitotic OSNs (B-D; data not shown). Expression of the four genes is abolished in Mash1 mutants OE (because of the loss of both basal progenitors and post-mitotic neurons in these mutants) except in restricted areas that maintain neurogenesis (A′-D′). In contrast, expression of Ebf1 and Lhx2 is maintained in basal progenitors in Ngn1 mutant OE (C″,D″), while expression of NeuroD and Phd1 is missing (A″,B″). Ebf1 is also expressed in the mesenchyme surrounding the OE. Scale bar, 25 μm.

We next examined the expression in wild-type and mutant OE of several genes encoding transcriptional regulators. The HLH gene Ebf1 (Wang et al., 1997), the LIM-homeobox gene Lhx2 (Xu et al., 1993) and the paired-homeobox gene Phd1 (Saito et al., 1996) are all expressed in basal progenitors as well as post-mitotic neurons in the OE at E12.5 (Fig. 4B-D). Double labelling experiments with RNA probes and an antibody to BrdU after 30 minutes of BrdU incorporation confirmed that Lhx2 and Phd1 were expressed in dividing basal progenitors (data not shown). In Mash1 mutants, expression of Ebf1, Lhx2 and Phd1 was strongly reduced, as expected from the lack of OSNs and their progenitors (Fig. 4B′-D′). In Ngn1 mutants, expression of Phd1 was very severely affected (Fig. 4B″). In contrast, expression of Lhx2 and Ebf1 appeared only slightly reduced (Fig. 4C″,D″), the reduction being due at least in part to the lack of OSN neurons in Ngn1 mutant OE. Together, these results show that a subset of the regulatory genes normally expressed in basal progenitors fail to be activated in Ngn1 mutant OE (i.e. NeuroD and Phd1), while other regulatory genes appear to be expressed independently of Ngn1 (i.e. Ebf1, Lhx2, Dll3 and Hes5). The OSN differentiation programme, which is activated downstream of Mash1 in OSN progenitors, therefore includes at least two parallel sub-programmes, one which is dependent on Ngn1 function and another which is Ngn1 independent (Fig. 5).

Fig. 5.

A model of the regulatory interactions involved in the determination and differentiation of the olfactory sensory neuron lineage. Mash1 is expressed in both apical and basal progenitors in the OE. Apical progenitors are primary neuroepithelial cells that give rise to secondary progenitors that settle basally. Mash1, which is required for the expression of the Notch signalling genes Serrate 1 and Hes1 in apical progenitors, and for the generation of basal progenitors, has the characteristics of a determination gene in the OE. A number of genes encoding transcriptional regulators are activated downstream of Mash1 in basal progenitors, including Ngn1. Ngn1 function is required for expression of the neuronal differentiation gene NeuroD and the neuronal subtype determinant Phd1 and for differentiation of olfactory sensory neurons. In the absence of Ngn1, however, basal progenitors express the Notch signalling genes Ser2, Dll3 and Hes5 and the transcriptional regulators Ebf1 and Lhx2. Ngn1 is therefore required to initiate one of several parallel programs of OSN differentiation activated downstream of Mash1 in basal progenitors. Note that the pathway depicted in this figure only operates in a subset of OE progenitors (e.g. most progenitors at E12.5; see Discussion).

Ngn1 functions as a determination gene for a subset of OE progenitors

The regulatory cascade involving Mash1 and Ngn1 is essential for the determination and differentiation of the majority of OSN progenitors. Some progenitors are however independent of either Mash1 or Ngn1 function, as shown by the differentiation of different populations of OSNs in Mash1 and Ngn1 single mutant OE (Fig. 2A′,A′′,C′) (Cau et al., 1997). We investigated whether the two genes have redundant functions in these progenitors, by examining OE development in embryos double mutants for Mash1 and Ngn1, using SCG10 to mark post-mitotic neurons and Ebf1 to mark both basal progenitors and neurons. No cells expressing either of these markers were found in the OE of Mash1;Ngn1 double mutant embryos at E12.5, E15.5 and E17.5 (Fig. 6A′′′,B′′′, and data not shown), whereas scattered SCG10-positive and Ebf1-positive cells remained in the OE of single Mash1 and Ngn1 mutant embryos at E12.5, and SCG10-positive neurons were found in similar numbers in the OE of wild-type and Ngn1 mutant embryos at E15.5 (Fig. 6A′,A′′,B′,B′′, and data not shown). Therefore, all OSN lineages in the main OE are dependent on either Mash1 or Ngn1 function. Interestingly, the vomeronasal organ (VNO), which originates from a medial recess of the olfactory placode, still contained SCG10-positive OSNs in Mash1; Ngn1 double mutant embryos, albeit in reduced number (Fig. 6C-C′′′), suggesting that another gene than Mash1 or Ngn1 is involved in the generation of VNO neurons.

Fig. 6.

Ngn1 is redundant with Mash1 for the determination of a subset of olfactory sensory neuron progenitors. Expression of the pan-neuronal marker SCG10 (A-A′′′,C-C′′′) and of the progenitor and neuronal marker Ebf1 (B-B′′′) in the OE (A-A′′′) and the vomeronasal organ (C-C′′′) of wild-type (A-C), Mash1 mutant (A′-C′), Ngn1 mutant (A′′-C′′), and Mash1, Ngn1 double mutant (A′′′-C′′′) embryos at E12.5. Subsets of OSNs differentiate in the absence of Mash1 or Ngn1 function (A′,A′′,B′,B′′), while in the absence of both genes, Ebf1-positive progenitors and SCG10-positive OSNs are completely missing (A′′′,B′′′). In contrast, some OSNs persist in the VNO of Mash1, Ngn1 double mutants (C′′′). Expression of the Notch ligand Dll1 (D-D′′′) and the Notch effector Hes5 (E-E′′′) in wild-type (D,E), Mash1 mutant (D′,E′), Ngn1 mutant (D′′-E′′) and Mash1, Ngn1 double mutants (D′′′-E′′′) in E10.0 olfactory placodes. Dll1 is expressed in isolated cells and Hes5 in cell clusters in wild-type and single mutant placodes, while expression of the two genes is missing in Mash1, Ngn1 double mutant placodes (D′′′,E′′′), indicating that Mash1 and Ngn1 have redundant functions in the determination of placodal progenitors. Scale bars, 25 μm (A-A′′′,B-B′′′,C-C′′′); 50 μm (D-D′′′,E-E′′′).

The complete loss of Ebf1 expression in Mash1;Ngn1 double mutants, which contrasts with the persistence of subsets of Ebf1-positive progenitors in Mash1 single mutants (Figs 4C′, 6B′′), suggested that Ngn1 may have a determination function for Mash1-independent progenitors. To directly address this possibility, we examined the expression of the Notch signalling genes Dll1 and Hes5, which can be used as markers of neural determination activity (Fig. 3 and data not shown), focusing on the first population of OSNs that differentiates in the olfactory placode because it is easily identifiable and is independent of Mash1 function (Fig. 2C′) (Cau et al., 1997). In wild-type olfactory placodes at E10.0, Dll1 and Hes5 are expressed in scattered cells and small cell clusters, respectively (Fig. 6D,E). Expression of the two genes was unaffected in both Mash1 mutant placodes, as expected (Fig. 6D′,E′), but also in Ngn1 mutant placodes (Fig. 6D′′,E′′), indicating that a determination activity persists in the placodes in absence of Ngn1. The absence of neurons in Ngn1 mutant placodes (Fig. 2C′′) is thus not due to a defect in the generation of Mash1-independent OSN progenitors, but rather to a block in their differentiation. However, placodal expression of Dll1 and Hes5 was abolished when both Mash1 and Ngn1 are mutated (Fig. 6E′′′,D′′′), indicating that either Mash1 or Ngn1 is sufficient to promote the determination of OSN progenitors and activate Notch signalling in the placode. This data thus indicates that although Ngn1 functions mostly as a differentiation gene in the OE, it is completely redundant with Mash1 for the determination of the early OSNs progenitors produced in the olfactory placode.

DISCUSSION

In this article, we have compared the functions of Mash1 and Ngn1 in the OE. The clearly distinct loss-of-function phenotypes of the two genes indicate that they function sequentially in most OSN progenitors at E12.5 as determination and differentiation genes, respectively, while the two genes have a redundant determination function in the olfactory placode at E10.0.

Determination function of Mash1 in the OE

Analysis of the loss-of-function phenotype of Mash1 shows that Mash1 is required for the generation of a subset of progenitors located on the basal side of the OE and for the expression of a Notch ligand (Ser1) and a Notch effector gene (Hes1) in another subset of progenitors, located on the apical side. The lack of basal progenitors is apparent as early as E12.5, the first stage when these cells can be identified (Smart et al., 1971), indicating that loss of Mash1 results in a failure to generate basal progenitors rather than in defects in proliferation or survival of progenitors subsequent to their production. There is indeed no significant cell death in Mash1 mutant OE until E13.5-E14.5 (Cau et al., 1997). The presence of two clearly separated populations of dividing progenitor cells located on opposite sides of the neuroepithelium is a situation unique to the OE that helps to define the cellular function of Mash1 in this tissue. The cells that divide on the apical side of the OE are neuroepithelial cells, which constitute the only progenitor cell population present at the placodal stage (Smart, 1971), and are thought to produce sustentacular cells, the supporting cells of the OE, at postnatal stages (Graziadei and Monti Graziadei, 1978). The second population of dividing progenitors appears on the basal side of the OE around E12.5, and corresponds to the immediate progenitors of olfactory sensory neurons (Graziadei and Monti Graziadei, 1978; Caggiano et al., 1994). These progenitors are likely to derive from neuroepithelial cells that lose their apical attachment and subsequently divide basally, possibly resulting from the asymmetric division of neuroepithelial cells along an horizontal cleavage plan (Smart, 1971). Mash1 is expressed in dividing cells in apical, intermediate and basal positions (Fig. 1) (Cau et al., 1997) and the Mash1 mutation results in the loss of basal but not apical progenitors (Fig. 2). This data suggests that Mash1 function is primarily required in apical neuroepithelial cells for the generation of basal progenitors, which is clearly reminiscent of the role of Drosophila proneural genes in the delamination of neural precursors from the neurectoderm, although it cannot be ruled out at present that Mash1-positive progenitor cells in apical and basal locations are not lineally related. The parallel between the function of Mash1 and of Drosophila proneural genes holds for the regulation of Notch signalling activity. Notch signalling has been implicated in both vertebrates and insects in the singling out of neural progenitors from groups of equivalent cells in the neuroepithelium (Chitnis et al., 1995; Chitnis and Kintner, 1996; Henrique et al., 1997). The regulation by Mash1 of Ser1 and Hes1 expression in OE apical cells (Fig. 3) thus suggests that, through Notch activity, Mash1 controls the selection of apical cells destined to generate basal progenitors and OSNs. Mash1 has thus the hallmarks of a proneural gene for the OSN lineage.

In the cerebral cortex, the step of selection of progenitors from the ventricular neuroepithelium, regulated by Mash1 and Ngn2, coincides with the restriction of multipotent cells to the neuronal lineage (Nieto et al., 2001). The main cell types in the OE are the OSNs and the sustentacular cells, which are thought to be generated in the adult by distinct progenitor cells, located basally and apically, respectively (Graziadei and Monti Graziadei, 1978). It is not known whether a common progenitor for the two lineages exists at early stages of OE development, in other words whether apically dividing neuroepithelial cells in the embryonic OE are pluripotent and give rise to both sustentacular cells and basal OSN progenitors, or whether they constitute a mosaic of progenitors restricted to the sustentacular and OSN fates. A retroviral lineage study performed in the lesioned OE has revealed the existence of cells that have the capacity to produce both OSNs and sustentacular cells after lesion (Huard et al., 1998), but there is no evidence that such cells are also active during normal development of the OE. Indeed, another lineage study performed in the unlesioned OE of young rats has led to the opposite conclusion that OSNs and sustentacular cells belong to separate lineages (Caggiano et al., 1994). It is therefore unclear whether Mash1, which controls the generation of basal progenitors, is involved in restricting OE progenitors to the OSN fate. Alternatively, the control by Mash1 of the transition from apical to basal progenitors could correspond to other changes in the properties of OE progenitors, for example in their rate of proliferation or in their response to extrinsic signals (Calof et al., 1998).

Neuronal differentiation function of Ngn1

In contrast to Mash1, which is expressed in OE cells in both apical and basal locations, Ngn1 expression is restricted to basal progenitors after E12.5 (Fig. 1). This suggests that Ngn1 acts in the OSN lineage after the generation of basal progenitors and has thus a function distinct from that of Mash1. In support of the idea that Ngn1 has a later function than Mash1, analysis of gene expression in mutant OE shows that Ngn1 acts downstream of Mash1 in a unidirectional regulatory cascade (but see below for Ngn1 function in the olfactory placode). Comparison of the null mutant phenotypes of Mash1 and Ngn1 further demonstrates that these genes are absolutely required at two clearly distinct stages for further progression of the OSN. Basal progenitors are present, express Mash1, Dll3 and Hes5 and divide in absence of Ngn1, demonstrating that in contrast to Mash1, Ngn1 is not required for the generation of basal progenitors or the activation of Notch signalling. Therefore, Ngn1 does not has the characteristics of a determination gene in most of the OE.

In the absence of Ngn1, however, basal progenitors fail to express several regulatory genes presumably involved in specification and differentiation of OSNs, including the paired-homeobox gene Phd1 and the bHLH gene NeuroD, and basal progenitors fail to withdraw from the cell cycle and differentiate. Thus, genetic analysis has identified Ngn1 as a gene required in basal progenitors to activate an OSN differentiation program, and force dividing progenitors out of the cell exit.

Parallel regulatory pathways in OSN progenitors

The analysis of the Ngn1 mutant phenotype in the OE has revealed the expression in the OE of two groups of regulatory genes distinct in their mode of regulation in basal progenitors. Genes of the first group, which include NeuroD and Phd1, require Ngn1 function for their expression. Genes of the second group, which include the Lim-homeobox gene Lhx2 and the HLH gene Ebf1, are activated in basal progenitors independently of Ngn1 activity. None of these genes are expressed in Mash1 mutant OE, indicating that they belong to distinct regulatory pathways that are activated downstream of Mash1 in basal progenitors. It has been proposed that in neural crest-derived progenitors, Mash1 couples two parallel differentiation programs controlling the expression of neuronal subtype genes (e.g. the homeobox gene Phox2a and genes encoding the neurotransmitter synthesising enzymes TH and BDH) and pan-neuronal genes (e.g. genes encoding peripherin and NF160) (Lo et al., 1998). The idea that neuronal differentiation entails the activation of distinct regulatory programs implementing the different aspects of the neuronal phenotype has been amply supported experimentally (reviewed by Jan and Jan, 1994; Anderson and Jan, 1997; Brunet and Ghysen, 1999; Guillemot, 1999). Our results indicate that Mash1 is likely to similarly couple the various components of the OSN phenotype. Among the regulatory genes expressed in basal progenitors and missing in Mash1 mutant OE, NeuroD and Ebf1 are likely to promote generic neuronal differentiation of OSNs (Anderson, 1999). Both genes are broadly expressed in neurons in the embryonic nervous system and have the capacity to induce ectopic neurons when forcibly expressed in Xenopus embryos (Lee et al., 1995; Wang et al., 1997; Garel et al., 1997; Dubois et al., 1998). Other basal OE progenitor genes, e.g. Phd1 and Lhx2, are, in contrast, likely to be involved in the specification of OSN identity. These genes belong to the paired-homeodomain and Lim-homeodomain families of transcription factors, respectively (Xu et al., 1993; Saito et al., 1996), which have been implicated in specification of various aspects of the neuronal phenotype (Anderson and Jan, 1997; Pfaff and Kintner, 1998). It is interesting to note that the Ngn1-dependent and -independent groups of regulators defined in this study do not segregate into ‘pan-neuronal’ (NeuroD and Ebf1) and ‘neuronal subtype’ (Phd1 and Lhx2) categories, but are on the contrary distributed in both. This suggests that the regulatory programs supporting OSN differentiation are not specialised in the acquisition of either generic or OSN-specific traits, but may instead control the acquisition of different combinations of both types of traits. The logic behind this complex regulation of the OSN phenotype remains to be elucidated.

Distinct regulation and function of Ngn1 in different OSN progenitor populations

Mice lacking both Mash1 and Ngn1 present a complete depletion of OSNs at all stages examined, indicating that together, Mash1 and Ngn1 are required for the progression of neurogenesis throughout the OE. Analysis of the olfactory placodes of Mash1;Ngn1 double mutant embryos showed that the two genes have redundant functions in the determination of OSN progenitors at this early stage. In particular, a lack of Dll1 and Hes5 expression revealed that Notch signalling is not activated and that OSN progenitors are likely missing from the olfactory placodes in absence of both Mash1 and Ngn1.

The redundancy of Mash1 and Ngn1 function in the determination of placodal progenitors at E10.0 is in marked contrast to the distinct and sequential roles of the two genes in the OE at E12.5. Even at E12.5, the Mash1/Ngn1 cascade does not operate in all OSN progenitors, since Ngn1 is expressed and required in a subset of Mash1 mutant OE progenitors at this stage (Fig. 1B′; Fig. 6) (Cau et al., 1997). The fact that Mash1-independent progenitors are found at reproducible locations in different Mash1 mutant embryos argues for the existence of a distinct progenitor population in which a neurogenesis program including Ngn1 expression can be activated without Mash1 function (Fig. 2) (Guillemot et al., 1993; Cau et al., 1997). Whether Ngn1 is required for the determination of Mash1-independent progenitors at E12.5 as is the case for progenitors of olfactory placodes, or for their differentiation as is the case for other E12.5 OE progenitors, is difficult to address given the rarity of these cells. In any case, the epistatic relationship between Mash1 and Ngn1 observed in most OSN progenitors at E12.5 is not a general feature of the neurogenesis program in the OE, but is instead restricted both temporally (to the E12.5 OE and not the E10.0 placode) and spatially (to the rostral OE and less so to the caudal OE).

The dual function of Ngn1 in the selection of progenitors (in E10.0 placodes) and their differentiation (in most OSN progenitors at E12.5) has previously been reported for other neural bHLH genes. Mash1, which is required for the generation and fate specification of progenitors in the telencephalon (Casarosa et al., 1999; Nieto et al., 2001), is only required for terminal differentiation of sympathetic neurons (Sommer et al., 1995). In Drosophila, atonal is a proneural gene for photoreceptors, chordotonal sense organs and olfactory sensilla (Jarman et al., 1995; Gupta et al., 1998), while it is involved in neurite arborization, but not in progenitor selection, in the embryonic brain (Hassan et al., 2000). Therefore, bHLH factors with neural determination properties can also regulate later aspects of the differentiation program when their expression is maintained in differentiating progenitors and neurons. The observation that Ngn1, which is a determination factor for sensory neurons in cranial ganglia (Ma et al., 1998), is required for the differentiation of olfactory sensory neurons, is therefore in line with these previous findings, although the situation of a bHLH gene having both determination and differentiation functions in the same tissue, as for Ngn1 in the OE, had not been reported before.

How can these findings be reconciled with the observation that in regulatory cascades underlying cell lineage development, distinct subfamilies of bHLH factors are usually used as either early expressed determination factors or later expressed differentiation genes (see Introduction). Our results and previous studies (Sommer et al., 1995; Hassan et al., 2000) support the idea that determination factors also have the necessary properties to participate to neuronal differentiation programs, and that the specific determination or differentiation function of genes like Ngn1 or Mash1 depends primarily on the timing of their expression and on the context of their activity. In contrast, other neural bHLH genes such as NeuroD and related genes, are consistently expressed in late precursors and post-mitotic neurons and have not been implicated in the process of progenitor selection during normal development, although NeuroD shares with Ngns the property to induce neurons and ectopically activate Notch signalling when overexpressed in Xenopus embryos (Lee et al., 1995; Chitnis and Kintner, 1996; Ma et al., 1996), and NeuroD has been shown to participate to the choice between neuronal and glial fates in the retina (Morrow et al., 1999). The fact that differentiation genes such as NeuroD are normally not involved in cell determination suggests that they may lack some of the necessary properties. Indeed, the myogenic determination factors MyoD and Myf5 are more efficient than the differentiation factor myogenin at remodelling chromatin and activate transcription at previously silent loci (Gerber et al., 1997), an activity which is very likely relevant to their determination function. Possibly as a consequence of these divergent activities, myogenin cannot fully substitute for Myf5 when expressed from the Myf5 locus (Wang and Jaenisch, 1997). It will be interesting to test the prediction that neural determination genes (Mash1 and Ngns) can efficiently substitute for differentiation genes (NeuroD and related genes) but not the reverse, in similar gene swapping experiments.

Acknowledgments

We thank Carol Schuurmans and Steve Wilson for their critical comments on the manuscript, and Valérie Meyer and Véronique Pfister for excellent technical help. We thank David Anderson for the generous gift of Ngn1 mutant mice, and David Anderson, Patrick Charnay, Domingos Henrique, Tetsuichiro Saito, Gerry Weinmaster and Heiner Westphal for the gifts of cDNA clones. This work was supported by grants from the European Commission ‘Quality of Life and Management of Living Resources’ Program, the Human Frontiers Science Program, the Association pour la Recherche sur le Cancer, and the Ministère de l’Enseignement et de la Recherche to F. G. and by institutional funds from INSERM, CNRS and Hôpital Universitaire de Strasbourg.

Footnotes

    • Accepted January 25, 2002.

References

View Abstract