limitation but has not yet been used extensively to analyse neuron-specific gene expression.
The combination of these techniques with the biochemical analysis of protein/DNA interactions and the characterization of promoter sequences suitable for transcription factor binding provides knowledge about a range of promoter elements and transcrip-tion factors that may be involved in the expression of neuronal genes and will be discussed below. Still, the emerging complexity of the protein complexes involved in transcriptional regulation is far from fully understood and results in statements such as
‘surprisingly little is known about how individual genes are turned on or off ’ (Lemon and Tjian 2000). This is illustrated by the unfolding complexity of the interaction of coactivators and corepressors with the basal transcription machinery, and also by the considerable variation of the regulatory architecture of neuron-specific genes with respect to activating and silencing elements that has been appreciated already a decade ago (Mandel and McKinnon 1993; Grant and Wisden 1997).
3.3
Regulation of neuronal gene expression by
REST/NRSF protein silences expression from a reporter construct carrying the RE1/NRSE domain in the promoter region. Conversely, a dominant-negative form of REST/NRSF that obstructs the activity of endogenous REST/NRSF relieves silencing mediated by the native protein (Chong et al. 1995; Eggen and Mandel 1997).
Importantly, similar silencer elements were identified in a large number of neuron-specific genes (Schoenherr et al. 1996) such as those coding for neurotransmitter receptors or transmitter release-associated proteins (Myers et al. 1999; Thiel et al.
1999). In many cases they are able to bind REST/NRSF and repress expression from a promoter/reporter construct. This exciting observation led to the conclusion that REST/NRSF is a silencing transcription factor expressed specifically in non-neuronal cells that prevents the expression of a large battery of neuron-specific genes in non-neuronal cells.
3.3.2 The ββ2 neuronal acetylcholine receptor subunit gene — turning repressors into activators?
Detection of REST/NRSF expression in adult neurons (Palm et al. 1998) strongly suggests that a silencing hypothesis assuming the exclusive action of the NRSE/RE1 in non-neuronal cells may be too simplistic. Also reporter gene expression mediated by the NRSE/RE1 motif in the β2 neuronal acetylcholine receptor (nAChR) subunit promoter casts doubt on the exclusive silencing action of the motif.
The β2 subunit of the nAChR is the most widely expressed subunit in neurons (Hill et al. 1993). A β-galactosidase reporter driven by the β2 nAChR promoter in transgenic mice is expressed in the peripheral nervous system and different regions of the central nervous system such as thalamus and colliculus. Mutation of the RE1/NRSE switches off expression in most of these structures but increases expression in some others such as the colliculus (Bessis et al. 1997). This suggests an intrinsic activating or repressing activity of RE1/NRSE in neurons depending on the neuron population. Similarly, in neuronal cell lines the RE1/NRSE element may act as a silencer or as enhancer in trans-fection experiments. Inserting spacers of different length between the RE1/NRSE and the TATA box of the promoter/reporter construct indicates that RE1/NRSE activity depends on its position in the promoter.
These conclusions remain controversial, however. REST/NRSF functions as repressor on model promoters containing strong promoter/enhancers in addition to RE1/NRSE motifs derived from various neuronal genes including the β2 neuronal acetylcholine receptor subunit (Thiel et al. 1998). Regardless of the position of the RE1/NRSE motifs, no activator function could be attributed to REST/NRSF in these experiments.
Moreover, transcriptional activation mediated by activator domains from different transcription factors is blocked by REST/NRSF independent of the RE1/NRSE position (Lietz et al. 2001). Together with these observations, the existence of different repressor domains in REST/NRSF and their recruitment of histone deacetylases provoke the question how REST/NRSF may mediate an activating function.
3.3.3 The cholinergic gene locus — modulating the silencer function by alternative splicing?
The detection of REST/NRSF splice variants and their dynamic control in neural tissue (Palm et al. 1998) adds to the complexity of transcriptional regulation by this factor.
Analysis of the action of the splice variants on the ChAT promoter demonstrates their possible antagonistic interaction.
Fragments of 5′flanking region of the ChAT gene from different mammalian species are able to direct expression of a CAT reporter gene to neuronal but not to non-neuronal cell lines (Misawa et al. 1992; Li et al. 1993; Lönnerberg et al. 1995). In non-neuronal cell lines, the RE1/NRSE motif in the proximal part of the 5′flanking region (Lönnerberg et al. 1996) shows repressing activity which is removed by specific deletion of this motif. Expression of the REST/NRSF splice variant REST 4 in a cell line expressing endogenous full-length REST/NRSF is able to transcriptionally activate the cholinergic gene locus (Shimojo et al. 1999). A direct interaction between REST/NRSF and REST4 could be demonstrated by co-immunoprecipitation. The relatively weak binding of REST4 to the cholinergic RE1/NRSE as compared to full-length REST/NRSF (Lee et al. 2000) suggests that the derepressing action of REST4 rests on a direct inter-action with the full-length protein. Using a synapsin I promoter, the derepressing action of REST4 is not confirmed, however (Magin et al. 2002). Thus, the possible antagonistic action of splice variants remains controversial.
Regulation of REST/NRSF expression and splicing via neuronal activity (Palm et al.
1998) and protein kinase-mediated signal transduction (Shimojo et al. 1999) as well as the action of neurotrophin growth factors via the RE1/NRSE motif (Brene et al. 2000) indicate that this motif and the REST/NRSF splice variants may participate in the dynamic control of neuronal gene expression. The low level of expression of splice variants such as REST4 (Palm et al. 1998) leaves open the question for their significance in vivo.
3.3.4 REST/NRSF as master regulator?
The identification of the neuron-restrictive silencer element, RE1/NRSE, and the zinc finger protein REST/NRSF mark major progress in the quest to identify molecular players involved in the regulation of neuron-specific gene expression. The characteri-zation of this element and its activity in regulating the expression of genes such as voltage-gated sodium channels, neuronal acetylcholine receptors, glutamate receptors, synapsins, choline acetyltransferase, and the like (Schoenherr et al. 1996) pointed out a role in the specification of neuronal properties. The quasi-ubiquitous expression in non-neuronal tissues as well as the expression in undifferentiated neuronal progenitors but not differentiated neurons, as it was initially described (Chong et al. 1995;
Schoenherr and Anderson 1995), shaped the hypothesis that REST/NRSF could be a transcription factor crucial for differentiating gene expression between neurons and non-neuronal cells. Thus, REST/NRSF was considered to be ‘a master negative regulator
of neurogenesis’ (Schoenherr and Anderson 1995) where the expression of neuron-specific genes marks a default pathway which is blocked in non-neuronal cells by the presence of REST/NRSF (Chong et al. 1995).
Several lines of evidence indicate that the situation is not that simple. In mice lacking functional REST/NRSF, de novo ectopic expression in non-neuronal tissue was observed for neuronal class III βtubulin but not for a range of other neuron-specific genes such as SCG 10, synapsin I, and neurofilament M all of which contain RE1/NRSE motifs in their promoters (Chen et al. 1998). The possibility that the mutation was hypomorphic rather than completely blocking the function of REST/NRSF seems low as no RE1/NRSE binding activity could be detected that was immunologically related to REST/NRSF. Analysis is compromised, however, by the early embryonic lethality of the mutant mice. Interestingly, derepression was achieved for more neuronal genes in the chick embryo when a dominant-negative REST/NRSF, which blocks the activity of endogenous REST/NRSF, was overexpressed (Chen et al. 1998). Irrespective of the reason for the low degree of derepression of neuronal genes in non-neuronal tissue of the mouse mutant embryos, the results demonstrate that there may be no simple unitary mechanism regulating the expression of neuronal genes, not even of those containing a NRSE/RE1 motif.
The comparatively low impact of the REST/NRSF mutation on neuronal gene expression in mice is in agreement with the expression of a lacZ reporter in transgenic mice that is driven by regulating regions of the L1 cell adhesion molecule gene (Kallunki et al. 1997) or the β2 nAChR subunit gene (Bessis et al. 1997). In particular the analysis of the β2 nAChR regulatory regions demonstrates that a mutation of the NRSE/RE1 motif does not lead to a general non-neuronal expression of the promoter except for a few oligodendrocytes. Together, the studies show that the expression of REST/NRSF by itself and the presence of the respective binding motif, RE1/NRSE, in the regulatory region of a gene may not suffice to correctly differentiate neuronal from non-neuronal gene expression.
Surprisingly, the mutation of the NRSE/RE1 motif in a β2 nACHR subunit transgene results in massive alterations in the neuronal expression of the transgene. The detection of REST/NRSF expression in neurons of different brain regions (Palm et al. 1998) strongly indicates that this factor may not only silence the expression of neuron-specific genes in non-neuronal cells but regulate, by activation or repression, gene expression in neurons. The different activity of the β2 RE1/NRSE motif in different brain regions suggests interaction with other transcriptional regulators. Alternatively different variants of REST/NRSF may be active in different brain regions.
An additional twist of complication lies in the observation that NRSE/RE1 may be found not only in neuronal genes, but also in non-neuronal genes (Schoenherr et al.
1996). Moreover, their intragenic position may be conserved between species and they may be functional in promoter/reporter constructs. This further suggests that the NRSE/RE1 motif must not necessarily result in repression of gene expression in cells expressing REST/NRSF.
3.3.5 Molecular mechanism of REST/NRSF action
Biochemical analysis begins to illuminate the mechanism by which such gene regulation may be accomplished. The important observation that REST/NRSF binds the corepressor Sin3 and recruits histone deacetylase (HDAC) into a complex (Naruse et al. 1999; Huang et al. 1999a; Roopra et al. 2000) suggests that REST/NRSF-mediated repression involves histone deacetylation. Indeed, REST/NRSF binding to RE1/NRSE is accompanied by a decrease of histone acetylation around the RE1/NRSE motif and inhibition of histone deacetylation leads to expression of neuron-specific genes in non-neuronal cells. Whereas the corepressor mSin3A/B interacts with the N terminus of REST/NRSF, another co-repressor, CoRest, interacts with the C-terminal repressor domain of REST/NRSF (Andres et al. 1999). Significantly, CoRest is tightly associated with HDAC1/2 and the combination of REST/NRSF, CoRest and HDAC2 may repress the induction of type II sodium channel expression by nerve growth factor in PC12 cells (Ballas et al. 2001). These observations point out the recruitment of histone deacatylating activity by REST/NRSF and the role of corepressor proteins. On the other hand, due to the involvement of different corepressors binding to different repressor domains on REST/NRSF, the possibility arises that the successive recruitment of different corepressor complexes may help to explain dynamic versus stable, long-lasting regulation of neuronal gene expression (Griffith et al. 2001).
3.3.6 Conclusions
Currently, we have no complete picture of the functional role of the NRSE/RE1 motif and its binding factor REST/NRSF in neuron-specific gene expression. It may rather serve as a model to illustrate the complexity of protein–protein and protein–DNA interactions involved in gene regulation.
A plausible hypothesis for the role of REST/NRSF could be an early differentiation of neuronal and non-neuronal lineages and an additional, later action in differentiated neurons. Its role in non-neuronal cells may rest on an interaction with other transcrip-tion factors which differs between different genes and their regulatory regions. This may explain the non-neuronal expression of genes which despite containing the NRSE/RE1 motif are not neuron-specific.
While REST/NRSF has been analysed in vertebrates, tramtrack, another zinc finger transcription factor found in Drosophila represses neuroblast-specific genes and controls glial development (Badenhorst et al. 1996; Badenhorst 2001). This further underscores an early role of zinc finger transcription factors in lineage decisions. The comparison of vertebrate and invertebrate REST/NRSF and tramtrack homologues will show whether there is a conserved interaction or succession of such transcription factors during the early division of neuronal and non-neuronal development.
3.4