Trk neurotrophin receptors are not expressed as widely as the transmitter release-related proteins but can be detected in a variety of neuron subpopulations. In Brn3a mutant mice, Trk expression is affected in a neuron subpopulation and stage-specific manner resulting in the death of certain populations of sensory neurons such as trigeminal neurons whereas dorsal root ganglion neurons are unaffected (Huang et al. 1999b). The analysis of the trkA/NGF receptor promoter demonstrates a region conserved between mammals and birds that is sufficient to direct reporter gene expres-sion in transgenic mice with the appropriate timing (Ma et al. 2000). Mutation of distinct transcription factor binding motifs in this region demonstrates different binding sites to be required for expression in trigeminal, dorsal root and sympathetic ganglia.
These studies support the crucial role of neuronal subpopulation-specific transcrip-tion factors for neuron-specific genes widely expressed in different neuron popula-tions. As the expression patterns of many neuronal genes do not respect the boundaries of expression of individual subpopulation-specific transcription factors, widely expressed neuron-specific genes may be regulated under the regime of different transcription factors in different neuron populations.
3.4.3 Conclusions
So far, the analysis of activating elements in the regulatory regions of neuron-specific genes has demonstrated transcriptional activation at a level specific for neuronal sub-populations and at a level which is not restricted to neurons at all. What has not been found to date is an activating counterpart to REST, namely a transactivating factor generally expressed in neurons but not non-neuronal cells. Such a factor may still appear with ongoing genomic analysis. If this will not be the case, a combination of neuron-restrictive silencing, general, not neuron-restrictive activation, and neuron subpopulation-specific activation is sufficient to drive the appropriate levels of expres-sion for neuron-specific genes in the different neuron populations.
Importantly, neuron subpopulation-specific activation of gene expression is confirmed in vivo in mutant mice for different classes of homeobox-containing transcription factors and for neuron-specific genes as diverse as those coding for transmitter synthesis enzymes or neurotrophin receptors. These transcription factors belong to different families such as paired homeobox or POU proteins. Certain transcription factors or promoter elements may be important only during restricted periods in a neuron’s lifespan. Consequently, the question arises what developmental cascades of sequential transcription factor action are required for neuronal differentiation and specification.
3.5
Developmental acquisition of a neuronal phenotype —
of neuronal genes. The time schedule of this differentiation process and the different stages on this path, characterized by the expression of distinct sets of neuron-specific genes, are beginning to be understood. Different transcription factors analysed in a number of neuron populations are involved and a pictures begins to unfold how the expression of distinct transcription factors at successive differentiation stages propels this developmental process.
3.5.1 Basic helix-loop-helix transcription factors — proneural function and neuronal subtype specification
The genes of the achaete-scute complex in Drosophila melanogaster, in particular the achaete and the scute genes, play a decisive role in the development of the fly peri-pheral nervous system (Ghysen and Dambly-Chaudiere 1988). Loss of function muta-tions obstruct sensory organ development such that cells normally destined to become neurons adopt epidermal fates. Gain of function mutations resulting in misexpression of achaete or scute lead to formation of sensory organs in ‘false’ positions, so called
‘ectopic’ structures. Due to their early function in singling out sensory organ precursor cells that give rise to sensory neurons, achaete and scute were called proneural genes.
Importantly, they are involved in the development of specific types of sensory organs, the external sensory organs. For chordotonal organs or photoreceptors a different proneural gene, atonal, is required (Jarman et al. 1993, 1994). Genes in the achaete scute complex as well as atonal code for transcription factors of the basic helix-loop-helix (bHLH) family. Information specifying the type of sensory organ generated may reside in the bHLH domain of the protein and involve interactions with other cofactors (Chien et al. 1996). Thus, this class of transcription factors plays a crucial role in the very early decisions on the way from a precursor to a neuron and, at the same time, specifies the differentiation path taken.
The demonstration of bHLH transcription factors in vertebrates and their ability to convert non-neuronal to neuronal fate establishes these transcription factors as impor-tant regulators of neuronal development throughout the animal kingdom (Lee 1997).
As in Drosophila, vertebrate bHLH transcription factors confer not only generic but also subtype-specific neuronal properties (Brunet and Ghysen 1999). The diversity of neuronal bHLH proteins in different neuronal subpopulations and the sequence of expression during neuronal differentiation indicate that they are involved in the deter-mination of neuronal precursors and regulate consecutive steps of differentiation. The sensitivity of their expression to lateral inhibition as well as the timing of their express-ing durexpress-ing differentiation indicate that only some of these transcription factors qualify as proneural genes. Others are involved in more advanced events during neurogenesis.
In mouse cranial sensory ganglia a cascade of bHLH gene expression including neurogenins 1 and 2, Math3, NeuroD, and Nscl1 is correlated with different steps of development (Fode et al. 1998; Ma et al. 1998). Since this differentiation cascade and neurogenesis is blocked in neurogenin mutant mice and neurogenins are the first bHLH factors detectable, it is concluded that these bHLH transcription factors
function like Drosophila proneural genes. The later expressed bHLH proteins, such as neuro D, may in part be direct transcriptional targets of neurogenins and function during more downstream events of neurogenesis.
In the spinal cord, the bHLH factor olig2 is involved in motoneuron and at a later stage in oligodendrocyte development (Marquardt and Pfaff 2001). Interaction with neurogenin 2 and homeodomain proteins promotes neuronal differentiation and specification of neuron subtype identity (Mizuguchi et al. 2001; Novitch et al. 2001;
Scardigli et al. 2001). As shown by gene mutation in mouse and overexpression in chick embryos, neurogenins together with mouse atonal homolog 1 (Math1), may specifiy neuron subpopulations by mutual cross-inhibitory actions and by regulation of expression of homeodomain proteins such as LIM transcription factors (Gowan et al. 2001) as will be discussed below.
In sympathetic ganglia, the bHLH gene Mash1 in mice and Cash1 in chick are expressed early during development (Lo et al. 1991; Ernsberger et al. 1995; Groves et al.
1995). Mutation in mice obstructs generation of sympathetic ganglia (Guillemot et al.
1993), which are populated by partially differentiating precursors expressing neurofilament but not the neuronal marker SCG10 (Sommer et al. 1995), indicating different requirements for the expression of distinct neuronal properties. At more advanced stages of sympathetic neuron development, the bHLH transcription factors dHAND and eHAND as well as the homeodomain proteins Phox2a and 2b are expressed in sympathetic ganglia and implicated in noradrenergic differentiation (Howard et al. 2000; Ernsberger et al. 2000). While the role of Phox2 proteins in noradrenergic induction may be direct as discussed below, the precise mechanism of the regulation of noradrenergic target genes by bHLH factors remains to be determined.
In the retina, the possibility of their direct regulation of target gene expression was demonstrated in addition to their importance for neuronal development. In the mouse retina, mutation of both bHLH factors Mash1 and Math3 leads to a complete loss of bipolar cells (Hatakeyama et al. 2001). Math5 is required for retina ganglion cell formation (Brown et al. 2001). Also in the zebrafish, ATH5 is required for retinal ganglion cell genesis (Kay et al. 2001) indicating that the involvement of bHLH factors of this class in eye development is conserved between vertebrate classes. As shown by promoter analysis in the chick embryo, ATH5 may regulate the expression of the β3 nAChR subunit gene which is specifically expressed in retina ganglion cells. The close correlation ofβ3 and ATH5 expression as well as the activation of expression from a β3 promoter by ATH5 suggest that this bHLH factors may directly regulate expression of transmitter receptor subunits as part of its role in retinal ganglion cell development.
As ATH5 is expressed transiently in retinal ganglion cells, the question arises how ganglion cell-specific expression is maintained. Brn3 POU-domain transcription factors, which are also required for retinal development can be induced by chick and mouse ATH5 and may participate in transcriptional regulation in ganglion cells at more advanced stages of development (Liu et al. 2001).
3.5.2 LIM domain transcription factors — specifying neuronal subpopulations
In the vertebrate spinal cord, partitioning of motoneuron and interneuron popula-tions is apparent along the antero-posterior and the dorso-ventral axes. In addition to the bHLH transcription factors discussed above, homeobox-containing transcription factors are essential for the specification of progenitor cell identity (Briscoe and Erickson 2001) and development of the differentiated neurons (Jurata et al. 2000).
The segregation of neuronal fates along the dorso-ventral and medio-lateral axes in the vertebrate spinal cord has been correlated with a combinatorial expression of LIM homeodomain proteins (Tanabe and Jessell 1996), the so-called LIM code (Lumsden 1995). In the chick spinal cord, the LIM factors Islet-1, Islet-2, Lim-1 and Lim-3 define subpopulations of motoneurons that segregate into distinct columns and select different axon pathways (Tsuchida et al. 1994). Also in zebrafish, primary motoneurons are characterized by particular LIM factor combinations (Appel et al.
1995). The lack of motor and certain interneuron populations in mice mutant for the LIM factor islet-1 demonstrates the crucial importance of these transcription factors for specific neuron subpopulations (Pfaff et al. 1996).
The homeodomain protein MNR2 is expressed in motoneuron progenitors and transiently in postmitotic motoneurons (Tanabe et al. 1998). Its expression precedes that of LIM3, Islet1 and Islet2. The induction by MNR2 overexpression of the motoneuron-specific LIM factors as well as markers of more advanced stages in motoneuron differentiation such as choline acetyltransferase demonstrates a moto-neuron differentiation cascade where different transcription factors act sequentially.
Additional transcription factors such as the Nkx and Pax homeodomain proteins are involved in the patterning of the spinal cord (Briscoe et al. 1999, 2000; Sander et al.
2000). Due to mutual cross-inhibitory regulation, they help to establish sharp bound-aries in transcription factor expression domains in the spinal cord (Briscoe and Ericson 2001). They are required for the commitment of progenitors to particular developmental fates and suppress genes involved in the development of neighbouring neuronal subtypes. By this means they help to segregate different neuron populations in the spinal cord. By regulating LIM factor expression, a sequence of homeodomain transcription factors together with bHLH proteins regulates the specification of neuron populations that differ in their location and connectivity. The regulation of the target genes involved in establishing axonal projections remains to be established.
3.5.3 POU domain transcription factors — regulating development and expression of function-specific genes
The family of POU domain transcription factors emerged when structural similarities between a transactivating factor involved in pituitary development and hormone expression in mammals and a protein affecting neuronal development and behaviour in C. elegans were observed (Schonemann et al. 1998). It was soon recognized that
POU domain proteins constitute a large protein family which is expressed in the devel-oping brain (He et al. 1989) and plays crucial roles in pituitary and brain development (McEvilly and Rosenfeld 1999). Their action is analysed in greatest detail for Pit-1/GHF-1 and its importance for pituitary-specific gene expression.
Searching for activators of pituitary-specific growth hormone and prolactin expres-sion led to the identification and cloning of Pit-1 also called GHF-1 (Bodner et al.
1988; Ingraham et al. 1988), a POU domain transcription factor required in somato-troph and lactosomato-troph cells of the anterior pituitary. Pit-1 protein is able to bind cell type-specific cis-acting elements in the prolactin and growth hormone genes and to activate expression from the respective promoters (Nelson et al. 1988; Mangalam et al.
1989). The good correlation between the appearance of Pit-1/GHF-1 protein and of growth hormone expression indicates that this transcription factor is responsible for the differentiation of somatotrophic cells and hormone expression (Dollé et al. 1990).
Direct proof of the Pit-1 function in vivo comes from dwarf mice which show dis-ruptions of the Pit-1 gene and lack detectable expression of growth hormone and prolactin (Li et al. 1990). The additional loss of TSH expression extends the role of Pit-1 to the development of thyrotroph cells and indicates that additional factors confine prolactin and growth hormone expression to their respective cell types (Simmons et al. 1990).
Interestingly, using a Pit-1/lacZ transgene in Pit-1-defective dwarf mice demon-strates that Pit-1 expression is regulated by distinct enhancers for initial gene activation and subsequent autoregulation (DiMattia et al. 1997). This observation suggests a developmental sequence where distinct transcription factors initiate Pit-1 expression, which in turn initiates hormone production and maintains its own expression. The detection of the paired-like homeodomain transcription factor Prop-1 before the onset of Pit-1 expression and its requirement for Pit-1 gene activation illustrates such a cascade of tissue-specific regulators (Sornson et al. 1996). A range of additional tran-scription factors from different families are involved in different aspects of pituitary cell development (Dasen and Rosenfeld 1999). These include the LIM homeodomain factors Lhx3 and 4 as well as the Pitx homeodomain factors Pitx1 and 2 during early phases of pituitary development and nuclear hormone receptors involved in pituitary cell type-specific expression of hormone-encoding genes (Dasen and Rosenfeld 2001).
In the nervous system, POU domain proteins of the Brn3 subfamily are required for neuronal development. Targeted mutations in mice show that Brn3b and 3c are essen-tial for differentiation and survival of retinal ganglion cells (Xiang 1998; Gan et al.
1999) and inner ear hair cells (Xiang et al. 1998), respectively. Axonal growth is affected by all Brn3 subfamily members as observed in Brn3a-deficient trigeminal neurons (Eng et al. 2001) or Brn3b and 3c-deficient retinal ganglion cells (Erkman et al. 2000;
Wang et al. 2000, 2002). Whether direct regulation by Brn3 transcription factors of the SNAP-25 gene as discussed above and of GAP-43 expression as suggested by micro-array analysis of cDNA from wild-type and Brn3b mutant retinas (Mu et al. 2001) are involved remains to be proven. A range of other transcription factors is affected by the
Brn3b mutation (Erkman et al. 2000) and may contribute to the altered expression of target genes and axonal outgrowth disruption.
3.5.4 Conclusions
The results from different parts of vertebrate and invertebrate nervous systems provide a first impression of transcription factor cascades involved in neuronal differentiation and specification. A general theme is the early role of bHLH proteins which may excert proneural functions and are involved in the specification of developing neurons.
Cascades of bHLH proteins found in vertebrate neurogenesis indicate the diversifica-tion of their funcdiversifica-tion along different stages during neurogenesis. A crucial role for the differentiation of the diverse neuronal lineages can be attributed to transcription factors of distinct homeobox protein families. There is currently no unifying concept for the relative roles played by the different homeobox factor families during differenti-ation of the diverse neuronal lineages.
A classical example for the combinatorial action of different homeobox transcrip-tion factors is the specificatranscrip-tion of neuron lineages in the nematode C. elegans. The POU domain protein unc-86 and the LIM domain protein mec-3, both homeobox-containing transcription factors, are necessary for development of distinct neurons (Finney et al. 1988; Way and Chalfie 1989). The sequential expression of unc-86 before mec-3 suggest hierarchical actions in the touch neuron differentiation cascade (Duggan et al. 1998). Their action must not be strictly sequential, however. unc-86/
mec-3 heterodimers can form and bind to the mec-3 promoter (Xue et al. 1993;
Rockelein et al. 2000) as well as presumed target gene promoters (Duggan et al. 1998).
This illustrates how sequential activation of transcription factors may propel cell fate determination not only by successive production of distinct transcription factors but also by synergistic action of these transcription factors recruited into hetero-oligomere complexes.
Direct target genes of the respective transcription factors are known for only a limited number of examples such as Pit-1, Phox2, Ath-5, or unc-86. Moreover, the proof that the transcription factors bind to the presumed promoter target sequences in vivo is, in the majority of cases, not existent. Nevertheless, the methods to undertake this analysis are available. By the combination of these approaches with the increasing knowledge of model organism genomes, we can within the foreseeable future expect to have a profound look into the similarities and differences between transcription factor cascades specifying different classes of neurons.
3.6