The Neuronal Basic Helix-Loop-Helix Transcription Factor NSCL-1 Is Dispensable for Normal Neuronal Development

Abstract

The neuronal stem cell leukemia (NSCL) basic helix-loop-helix factors are neural cell-specific transcription factors. We have disrupted the NSCL-1 gene by homologous recombination and replaced the coding region with a beta-galactosidase reporter cassette to study the role of NSCL-1 in neuronal development and to follow the fate of NSCL-1 mutant cells. NSCL-1 mutant mice are viable and fertile on various genetic backgrounds and do not show any obvious signs of neurological malfunction. No differences in the distribution of NSCL-1 mutant or heterozygous neuronal cells were observed in the diencephalon, hippocampus, neocortex, and cerebellum at different stages of development. Likewise, no defects were found in the laminar organization of the cortex, and the distinct neuronal subpopulation appeared normal during development of the neocortex. Analysis of sensory neurons which strongly express NSCL-1 revealed that the spatiotemporal expression of neuronal differentiation factors, such as NeuroD and SCG-10, was not altered in developing distal and proximal cranial ganglia of mutant mice. In the cerebellum expression of NSCL-1 was confined to the proliferative and premigratory zone of the external granular layer and the internal granular layer. Interestingly, unlike cerebella of Math1^−/− or NeuroD2^−/− mice, NSCL-1-deficient mice have no obvious developmental defect, and neurons of the cerebellum appeared fully differentiated. Despite similar expression patterns of NSCL-1 and NSCL-2 in various areas of the diencephalon, including the arcuate nucleus and paraventricular nucleus, NSCL-1^−/− mice are fertile and show no adult onset of obesity like NSCL-2 mutant mice. Double-mutant NSCL-1^−/−-NSCL-2^−/− mice do not show any additional obvious malformations of the central nervous system, although both genes are expressed in a largely overlapping pattern. Our results argue against a simple functional redundancy within the NSCL gene family.

Transcription factors of the basic helix-loop-helix (bHLH) family play important roles in the control of cell proliferation, determination, and differentiation in numerous tissues. A relatively large group of bHLH genes is expressed in the nervous system and has been shown to be responsible for the development of specific subsets of neurons (14, 15, 27). For example, the proneural bHLH genes Ngn-1 and Ngn-2 are required for the development of several neuronal lineages, including proximal and epibranchial placode-derived sensory ganglia, dorsal root ganglia, and the telencephalon (9, 19, 20, 21, 31). Members of the NeuroD subgroup (NeuroD, Nex-1 [Math-2], Math-3, and NeuroD2 [NDRF]) are predominantly expressed at later stages of neuronal development (2, 22, 29, 30). These genes appear to be involved in terminal steps of neuronal differentiation since they are able to induce cell cycle arrest and to activate transcription of neuronal specific genes (8, 16, 25). Interestingly, overexpression of NeuroD in Xenopuslaevis embryos leads to a conversion of presumptive epidermal cells into neurons (17), although its expression in mammalian embryos suggests a role in neuronal differentiation rather than in determination. NeuroD2 (NDRF), like NeuroD, is expressed in differentiating neurons (22) in the developing cerebellum and cortex. NeuroD2^−/− mice die several days after birth and display a reduced cerebellar granular cell layer, possibly due to increased apoptosis of cerebellar neurons (28). Math-1, another bHLH protein, has been shown to be essential for granule cell development during the development of the cerebellum (5, 13).

Neuronal stem cell leukemia (NSCL-1) and NSCL-2 (18) genes form a subgroup within the bHLH superfamily. Both genes were originally cloned on the basis of their homology to the human hematopoietic stem cell leukemia gene SCL, also known as Tal-1 and Hen-1 (1, 10). NSCl-1 and NSCL-2 show a high sequence identity of 98% on the amino acid level within the bHLH region. The expression of NSCL-2 overlaps with but is distinct from that of NSCL-1. Inactivation of NSCL-2 results in disruption of the hypothalamic-gonadal axis and leads to infertility and adult onset of obesity (11). NSCL-1 is expressed only in subependymal layers of the neuroepithelium and is not detected in mitotic ventricular layers similar to NSCL-2 (3). NSCL-1 transcripts are first detected from mouse embryonic day 8.75 (E8.75) onwards in the proximal and distal epibranchial placodes (24). At E10.5 transcripts of NSCL-1 are present in different areas of the developing central nervous system (CNS) in the brain and the neural tube. Strong expression was found in dorsal root ganglia, cranial ganglia, the olfactory epithelium, and the developing optic cup (3). At E16.5 to E18.5 NSCL-1 transcripts were observed in the developing cerebellum, the subventricular layer of the cerebral cortex, and the olfactory epithelium. After birth NSCL-1 is restricted to the external granular layer (EGL) and internal granular layer (IGL) of the developing cerebellum (7, 32).

We have mutated the NSCL-1 gene by homologous recombination to decipher its role for neuronal cell formation. NSCL-1 homozygous mice are fully viable and do not show any obvious phenotype. NSCL-1 mutant neurons that are marked by β-galactosidase expression were indistinguishable from their heterozygous counterparts. In order to analyze functional redundancy within the NSCL gene family we have also generated NSCL-1–NSCL-2 double mutant mice. Surprisingly, no obvious differences in the architecture of the cerebellum or any other major malformation in the CNS were observed in double mutant mice.

MATERIALS AND METHODS

Generation and analysis of NSCL-1 knockout mice.

To generate the targeting vector a 6.5-kbp ApaI fragment derived from the 5′ region of the NSCL-1 gene comprising the first exon and 260 bp of the second exon was cloned into the ApaI site of the vector pWH9 in front of an internal ribosomal entry site, the coding region of the lacZ gene and the neomycin resistance cassette. A 1.3-kbp PflMI-BamHI fragment from the 3′ region of the NSCL-1 gene was inserted behind the selection cassette to generate a 3′-homology fragment. The recombination vector replaced the entire DNA binding bHLH region and most of the coding region with lacZ and neomycin resistance genes. Therefore, in the mutant locus, activity of the lacZ reporter gene is directed by NSCL-1 transcriptional control elements.

Electroporation and selection of J1 ES cells were done as described previously (6). To identify correctly targeted clones a 3′ external probe (BamHI/NcoI fragment) was used to detect a 25-kbp EcoRV fragment in the wild-type allele and a 15-kbp fragment in the mutant allele by Southern blot hybridization.

Reverse transcription (RT)-PCR analysis.

Total RNA was prepared from P1 mouse brains by column purification (RNeasy; Qiagen). First-strand cDNA was synthesized using 2 μg of total RNA, Expand reverse transcriptase (Roche), and random primers pd(N)6 (Pharmacia). Ten percent of the reverse transcriptase reaction mixture was used in the PCR (40 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 2 min) with a primer at position 45 (primer a forward, 5′GACTCCAGTTCTGGACTAAGTAAG) and two reverse primers at position 578 (primer b reverse, 5′GGACCACTCCTGGATCCCCGGATC) and at position 1104 (primer c reverse, 5′ GCATGGACTCCATTCTGACCACTC). GAPDH transcripts were detected with a forward (5′GTGGCAAAGTGGAGATTGTTGCC) and a reverse (5′GATGATGACCCGTTTGGCTCC) primer.

Histology and immunohistochemistry.

For immunohistochemistry mice were transcardially perfused with phosphate-buffered saline, followed by 4% paraformaldehyde. Brains were postfixed overnight with the same fixative followed by paraffin embedding or vibratome sectioning. Histological examination of brain structures was performed on routine hematoxylin-eosin (H/E)- and Nissl-stained paraffin sections (8 μm thick). Immunohistological analysis was accomplished using serial frontal and sagittal free-floating vibratome sections (50 μm thick) from adult (4- to 6-month-old) NSCL-1^−/−, NSCL-1^+/−, and wild-type littermates. The following antibodies were used: anti-calbindin D28K (diluted 1:4,000; Sigma), anticalretinin (1:2,000; Swant), anti-GFAP (1:500; DAKO), and anti-neurofilament (1:400; Sigma). Bound antibodies were visualized using the Vectastain Elite kit (Vector). Whole-mount in situ hybridizations were done according to the protocol of Wilkinson and Nieto (33). The NeuroD antisense probe was synthesized from a 1.3-kbp cDNA fragment (175 to 1,474 bp) using T7 RNA polymerase. The SCG-10 antisense probe (gift from A. Mansouri) was synthesized from a rat cDNA clone. Id1 transcripts were detected using a probe from plasmid pMH18Δ that contains an RsaI-EcoRI Id1 cDNA fragment (nucleotides 5 to 927) (gift from H. Weintraub). An Id2 probe was kindly provided by A. Mansouri, Göttingen, Germany. The probe for Math-1 was generated by RT-PCR. LacZ staining of embryos and cryosections (20 μm) was performed as described by Zweigerdt et al. (34).

RESULTS

We have generated NSCL-1 mutant mice by replacing the bHLH domain with a promotorless cytoplasmatic lacZ reporter gene and a neomycin resistance cassette (Fig. 1A). Southern blot analysis of genomic DNA from F₂ littermates proved that the NSCL-1 genomic locus had been targeted as expected (Fig. 1B). No RNA transcripts were generated from the mutant locus, while NSCL-1 mRNAs were readily detectable in wild-type mice as shown by RT-PCR using two different sets of primers (Fig. 1C).

FIG. 1.

Targeted disruption of the NSCL-1 genomic locus. (A) Schematic representation of NSCL-1 locus (upper row), the targeting vector (middle row), and the disrupted allele generated by homologous recombination (lower row). Insertion of the lacZ gene-neo resistance cassette (white box) deleted the second exon, including almost the entire bHLH region. Noncoding regions of the NSCL-1 gene are indicated by grey boxes; the bHLH domain is represented by a black box. Abbreviations for restriction sites: A, ApaI; B, BamHI; E, EcoRV; N, NotI; P, PstI; X, XbaI. (B) Southern blot analysis of EcoRV-digested DNA isolated from F₂ progeny. The external probe detects a 25-kb fragment in wild-type mice and a 15-kb fragment in mutant mice. Mice heterozygous for the mutation show both the 25-kb and the 15-kb band. (C) RT-PCR analysis of NSCL-1 RNA isolated from brains of E18.5 wild-type and homozygous mutant mice. Locations of the primers in exons 1 and 2 are indicated by arrows (a to c) in panel A. Lane 1, molecular weight markers; lanes 2 and 4, primer pair a and c; lanes 3 and 5, primer pair a and b. In homozygous mutant mice (lanes 4 and 5), no transcripts were detectable. To ensure equal loadings, RT-PCR with primers specific for GAPDH was performed.

NSCL-1-deficient mice are viable, fertile, and not obese.

In the F₂ generation (>200 offspring tested) wild-type, heterozygous, and homozygous mutant animals were found at the expected Mendelian ratio. Both NSCL-1^−/− males and females were fertile and able to give rise to normal numbers of progeny (data not shown). Despite the similar expression patterns of NSCL-1 mutant and NSCL-2 mice within various areas of the hypothalamus, such as the arcuate nucleus and paraventricular nucleus, adult onset of obesity was not observed in NSCL-1 male and female mutant mice. To determine the body weight, 20 wild-type and homozygous littermates were raised in single cages for 1 year. The body weights of NSCL-1^−/− animals were not higher than those of their wild-type littermates during the monitoring period (data not shown).

Development of cranial ganglia is not affected in NSCL-1-deficient mice.

The epibranchial placodes are transient ectodermal thickenings present between E8.5 and E10.5 of mouse development. NSCL-1-expressing neuronal cells delaminate from placodes, migrate, and form the distal aspect of cranial ganglia. Distal ganglia, which are mostly derived from epibranchial placodes, can be distinguished from proximal ganglia that are mostly derived from neural crest. Nevertheless, a considerable variation exists in the relative contributions of neural crest and placodal cells among distal and proximal ganglia.

Inactivation of the bHLH gene Ngn-1 prevents the determination and development of proximal ganglia, including the trigeminal (Vth), superior, jugular, accessory, and vestibulo-cochlear (VIIIth) ganglia, while ganglia originating from the epibranchial placodes, such as the geniculate (VIIth), petrosal (IXth), and nodose (Xth) ganglia, are dependent on the closely related bHLH gene Ngn-2 (9, 19). As shown in Fig. 2A and D, NSCL-1 lacZ-positive cells were present in both the distal (VIIth, IXth, and Xth) and proximal (Vth and VIIIth) ganglia in wild-type and mutant animals. Expression was also observed in the midbrain, the diencephalon, the neural tube, and the dorsal root ganglia (Fig. 2A and D). In agreement with previous reports using an NSCL-1 antisense probe (3), NSCL-1 lacZ-positive cells were found in subependymal layers of the neuroepithelium and not in the mitotically active ventricular zone, with the exception of the developing cerebellum.

FIG. 2.

Normal development of cranial ganglia of E10.5 NSCL-1^−/− embryos. (A and D) Whole-mount NSCL-1 lacZ staining of heterozygous (A) and homozygous (D) mutant embryos. (B, C, E, and F) In situ hybridization of heterozygous (B and C) and homozygous (E and F) mutant embryos with cRNA probes for NeuroD (B and E) and SCG-10 (C and F). Note that no differences in the distribution and strength of NeuroD and SCG-10 signals (indicated by arrows) are discernible, indicating a normal development of proximal and distal ganglia. Abbreviations: t, trigeminal; g, geniculate; v, vestibulo-cochlear; p, petrosal; n, nodose ganglia; ov, otic vesicle.

To analyze at the molecular level whether successive steps of cranial ganglion cell development occurred normally in NSCL-1 mutants, in situ hybridization experiments were performed. No differences were found at E9.5 (not shown) and E10.5 (Fig. 2B to F) in the expression of the bHLH gene encoding NeuroD, which marks an early stage of neuronal cell development, and of SCG-10, which detects differentiated neurons.

Expression pattern of NSCL-1 in developing cerebral cortex and cerebellum.

Next, we took advantage of the lacZ reporter gene and compared the appearances of NSCL-1-expressing, lacZ-positive neurons between heterozygous and homozygous mutant animals in different areas of the brain during development and adult life. Basically, no significant differences were found, with the exception of a more intense staining in homozygous mutant animals, which is due to the higher copy number of the reporter gene. Strong and widespread lacZ staining was detected in distinct laminated structures of the brain, including the cerebral cortex, cerebellum, hippocampus, and olfactory bulb. Between E14.5 and E16.5 NSCL-1 was expressed in the subventricular layer of the developing cortex with no detectable expression in the mitotically active ventricular layer (Fig. 3A to D). At E18.5, in addition to the expression in the subventricular layer, NSCL-1 was expressed in the intermediate zone and marginal zone of the cortex (Fig. 3E and F). Postnatally, NSCL-1 expression decreased and became undetectable in the adult cerebral cortex (data not shown). Additional expression domains were found in the dentate gyrus, in CA1, CA2, and CA3 fields of the developing hippocampus, and in dorsal areas of the diencephalon, including subsets of thalamic nuclei and the anterior lobe of the developing pituitary gland (data not shown).

FIG. 3.

Normal distribution of lacZ-positive cells in NSCL-1 heterozygous and homozygous mutant mice. (A to F) Frontal sections through subventricular layers of the developing cortex of heterozygous (A, C, and E) and homozygous (B, D, and F) mutant embryos at E14.5 to 16.5 and through the intermediate and marginal zones of the cortexes of E18.5 fetuses (E and F). In the cerebellum at E16.5 NSCL-1 expression in heterozygous (G) and homozygous (H) animals is restricted to the EGL. At E18.5 identical distributions of lacZ-positive cells were found in the proliferating and premigratory zones of the EGL in heterozygous (I) and homozygous (J) animals. A weaker expression is present in the developing IGL (I and J). Postnatally (P10), lacZ-positive cells were localized in the EGL and in the newly formed IGL of heterozygous (K) and homozygous (L) animals. Abbreviations: vz, ventricular zone; svz, subventricular zone; iz, intermediate zone; cp, cortical plate; mz, marginal zone; m, molecular layer.

A similar comparison between lacZ-positive cells of wild-type and mutant animals was performed for the cerebellum between E12.5 and postnatal day 10 (P10). NSCL-1 was first detected at E12.5 in the cerebellar primordium (data not shown), in agreement with previous results using RNA probes (32). Between E14.5 and E16.5 NSCL-1 expression became confined to migrating granule precursor cells of the ventricular zone of the cerebellar epithelium (Fig. 3G and H), which originate from the rhombic lip. At E18.5, NSCL-1 expression persisted in the single-cell layer over the entire surface of the EGL in the developing cerebellum (Fig. 3I and J). At P10 lacZ-positive cells were found in the EGL and the newly formed IGL (Fig. 3K and L).

Neuronal differentiation of the cortex and cerebellum in the absence of NSCL-1.

The tissue architecture of the cortex was investigated using Nissl and H/E staining. The diameters of the different layers of the cortex were measured, and the number of cells within each layer was counted. At least 10 sections were counted per mouse. No obvious differences in the laminar organization of the adult cortex were apparent between heterozygous (Fig. 4A, C, and E) and homozygous (Fig. 4B, D, and F) mice. Immunostaining using antibodies against calbindin-D28K and calretinin revealed the presence of distinct subpopulations of interneurons throughout the cortex in the absence of NSCL-1 (Fig. 4B, D, and F). The expression of calretinin indicated the presence of hippocampal dendrites of dentate gyrus granule cells that were differentiated normally both in wild-type and NSCL-1 mutant brains (data not shown). Likewise, no morphological differences were detected between the cerebella of wild-type and mutant mice using H/E and Nissl stainings (Fig. 5A to H). Mid-sagittal sections demonstrated the same number of folia and no changes in the size of molecular and granule cell layers. The use of antibodies against calbindin and neurofilament revealed the presence of distinct neuronal cell types, such as Purkinje cells and basket cell interneurons, respectively, in normal numbers and in a normal distribution in various areas of the cerebellum of adult NSCL-1^−/− animals (Fig. 5I to L). At least 1,000 cells were counted per animal. Similarly, staining with a GFAP antibody yielded no obvious alterations (data not shown).

FIG. 4.

Normal laminar organization of the cortex of NSCL-1 mutant mice. Frontal sections were taken from NSCL-1^+/− (A, C, and E) and NSCL-1^−/− (B, D, and F) mice. Sections were Nissl stained (A and B) or stained with antibodies against calretinin (C and D) or calbindin (E and F). No differences between NSCL-1⁺/⁻ and NSCL-1^−/− mice are evident. Note that no changes in the thickness of the cortex or the distribution of calretinin- or calbindin-positive cells between heterozygous and homozygous animals are apparent.

FIG. 5.

Normal organization of the cerebellum of adult NSCL-1 mutant mice. Sagittal sections through the cerebellum of wild-type (WT) (A, C, E, G, H, I, and K) and NSCL-1 mutant (B, D, F, H, J, and L) mice are shown. (A to H) No differences in the lobulation of the cerebellum and morphology of the molecular layer (ml) and IGL (gl) are discernible by H/E (A to D) and Nissl (E to H) staining. No significant differences were found in the distribution of Purkinje (p) cells that were identified by staining with an anticalbindin antibody (I and J) and basket cell interneurons (K and L) (stained with antineurofilament).

NSCL-1–NSCL-2 double mutant mice show normal neuronal differentiation in the cerebellum and no major malformation in the CNS.

One reason for the absence of any obvious phenotypic defects in NSCL-1 knockout mice is the existence of partially overlapping expression patterns of NSCL-1 and NSCL-2 in most parts of the CNS. To elucidate possible functional redundancy between NSCL-1 and NSCL-2, we generated NSCL-1 and NSCL-2 double homozygous mutants. Although NSCL-1–NSCL-2 double mutants die within the first day after birth, histological analysis of various sections of the CNS failed to reveal morphological differences between wild type and NSCL-1–NSCL-2 double homozygous mutants (data not shown). Likewise, in situ hybridization analysis using the neuronal markers NeuroD and Math-1 disclosed only minor differences in expression pattern between double heterozygous and homozygous animals at E18. 5 (Fig. 6A to H). Similarly, NSCL-1 has been postulated to regulate the expression of Id1 and/or Id2, two inhibitory HLH proteins that are strongly expressed in the cerebellum. In agreement with previous reports, strong (26) expression of Id1 was detected in the IGL of the cerebellum at E18.5, while expression of Id2 was confined to the Purkinje cell layer of the cerebellum. We did not observe any alteration of the expression of Id1 in double homozygous compared to double heterozygous mice at E18.5 (Fig. 6I to N). In contrast, we found a slight reduction of Id2 expression within the Purkinje cell layer of the cerebellum in double-knockout mice (Fig. 6K to P). Although these differences were only minor, they were repeatedly observed in several animals in separate experiments.

FIG. 6.

Normal expression of neuronal HLH genes in cerebella of NSCL-1–NSCL-2 double mutant mice at E18.5. Matched frontal and sagittal sections from NSCL-1–NSCL-2 double heterozygous (A, C, E, G, I, K, M, and O) and NSCL-1–NSCL-2 double homozygous mice (B, D, F, H, J, L, N, and P) were hybridized with riboprobes specific for NeuroD (A, B, E, and F), Math-1 (C, D, G, and H), Id1 (I, J, M, and N), and Id2 (K, L, O, and P). Panels E to H and M to P represent higher magnification of sections shown in panels A to D and I to P, respectively. No major differences in the expression of NeuroD, Math-1, and Id1 were detected between double heterozygous and double homozygous littermates. A minor reduction of Id2 expression is apparent in the Purkinje cell layer of double homozygous mutants (O and P) when compared to double heterozygous control animals. pcl, Purkinje cell layer.

DISCUSSION

Two separate groups of neuronal bHLH transcription factors have been distinguished based on their expression profile and assumed function: (i) determination factors, which regulate early events in neurogenesis, such as Ngn-1/2, and (ii) differentiation factors which are expressed either transiently in migrating and/or differentiating neurons, such as NSCL-1/2 and NeuroM, or enduringly in terminally differentiated neurons, such as NeuroD (15). The concept of determination and differentiation factors governing neuronal cell development reflects similar observations made and conclusions drawn originally for myogenic cells (6, 27).

Our analysis of NSCL-1 homozygous mutant mice revealed that these animals were phenotypically indistinguishable from heterozygous and wild-type mice. Cortical and cerebellar neurons, which represent major expression domains of NSCL-1, develop and differentiate without obvious morphological defects based on histological analysis, antibody and lacZ staining, and whole-mount in situ hybridization. Despite a very similar expression pattern of NSCL-2 in the thalamus and hypothalamus, including the paraventricular nucleus and the arcuate nucleus, NSCL-1-deficient mice do not show infertility or adult onset obesity such as that seen in NSCL-2-deficient mice.

The presence of a lacZ reporter gene in the NSCL-1 locus enabled us to monitor the fate of NSCL-1-expressing homozygous and heterozygous mutant precursors of cranial sensory neurons that delaminate from the placodal ectoderm and migrate to a site of ganglion condensation. Initially, the expression of NSCL-1 in both proximal and distal cranial ganglia had led to the idea that NSCL-1 might function as an early migration and differentiation factor for developing cranial ganglia. Surprisingly, no change in the migration pattern and cellular behavior of NSCL-1 mutant cells was detectable, thus excluding this idea. In both homozygous and heterozygous mutant embryos NSCL-1 lacZ expression became apparent at E8.75 (14 to 16 somites) in cells of the proximal placodes (Ngn1 dependent) that give rise to trigeminal and vestibulo-cochlear ganglia. An identical situation was observed in the distal ganglia (Ngn2 dependent), such as geniculate, petrosal, and nodose ganglia, where NSCL-1 first appeared at E9.0. In addition, the normal expression of NeuroD and SCG-10 that is representative for different stages of neuronal cell differentiation in cranial ganglia suggested that NSCL-1 is dispensable for normal differentiation of sensory neurons.

Similar findings were made in the cerebellum and in the cortex. In the cerebellum NSCL-1 is expressed in the outer and inner portions of EGL (this study and reference 32) corresponding to the proliferative and premigratory zones, whereas the closely related NSCL-2 gene is restricted to the inner layer of cerebellar EGL (12). The partially overlapping expression pattern of NSCL-1 and NSCL-2 within the developing EGL indicated that NSCL-2 might rescue the function of NSCL-1, at least in the inner EGL. We therefore reasoned that animals lacking both NSCL-1 and NSCL-2 would exhibit a more severe phenotype. However, our analysis of NSCL-1–NSCL-2 double-knockout mice demonstrated that the presence of NSCL-2 is not critically needed to compensate for the loss of NSCL-1, since double homozygous NSCL-1–NSCL-2 mice did not show any additional morphological malformations in the CNS.

It should be noted, however, that numerous bHLH proteins are expressed in a complex pattern in the cerebellum. Many bHLH proteins are highly expressed in the outer and inner zones of the EGL. For example, Math-1 is expressed in the outer proliferating EGL of the cerebellum, and Math-1^−/− mice fail to make any granule cells in the cerebellum, indicating that Math-1 has a critical role in the generation of granule precursors (4). Another member of the bHLH protein family, NeuroD, is expressed only in the inner postmitotic zone of the EGL. Postnatally, NeuroD^−/− transgenic mice display a massive reduction in the number of granule cells of the cerebellum and a complete loss of DG granule cells in the hippocampus (23). Other bHLH proteins, such as NeuroD2 (NDRF), NEX-1 (Math-2), and Math-3, also have related developmental profiles and partially overlapping expression domains (28, 29).

In principle, the function of individual bHLH genes can be rescued by other bHLH genes as first demonstrated for myogenic bHLH genes of the MyoD family (see reference 27 for a review). It seems therefore likely that the lack of an obvious phenotype in NSCL-1^−/− mice is due to a functional compensation by other neuronal bHLH proteins. In Nex1^−/− mice, which also show no recognizable defect (30), Schwab and coworkers postulate a functional compensation of the lack of Nex1 by NeuroD and NeuroD2.

At the moment no clear candidate has emerged that might rescue the function of NSCL-1 in all parts of the brain in which NSCL-1 is expressed at a high level. The generation and analysis of double homozygous NSCL-1–NSLC-2 mutant mice clearly ruled out that the NSCL-2 gene alone, which is closely related to NSCL-1, is required to compensate for the loss of NSCL-1. In this context it is interesting that in general the targeted disruption of early determination factors, such as Ngn-1 or -2 or Math-1, gives rise to rather severe phenotypes, such as a complete lack of distinct neuronal sublineages. Mutations of bHLH genes that are expressed late in the determination or differentiation process, however, like Nex1, NSCL-1 or NeuroD2, lead to only mild phenotypes. It might be possible that neuronal cells are able to use bHLH genes, which normally act as differentiation factors in different neuronal sublineages, to promote a comparable differentiation step in another cell types, while early determination events, which are based on a less-complicated network, do not have this choice. Nevertheless, it is surprising that the NSCL genes that form a distinct subgroup within the family of neuronal bHLH genes that are highly expressed and evolutionarily well conserved are dispensable to generate a morphologically normal CNS. Ultimately, the high number of and the complex relations between different neuronal bHLH might require the generation of triple- and quadruple-knockout mice to pinpoint specific and overlapping functions of individual neuronal bHLH genes.