the regulation of neuron-specific gene expression in ...

J. Neurogenerics. 1995. Vol. 10. pp. 67-101

0 1995 Harwood Academic Publishers GmbH

Reprints available directly from the publisher Photocopying permitted by license only

Printed in Malaysia

REVIEW

J Neurogenet Downloaded from informahealthcare.com by RWTH Aachen Hochschulbibliothek on 12/09/10 For personal use only.

THE REGULATION OF NEURON-SPECIFIC GENE EXPRESSION IN THE MAMMALIAN NERVOUS SYSTEM RICHARD M. TWYMAN and ELIZABETH A. JONES Department of Biological Sciences, University of Warwick, Coventry, CV4 7AL (Received December 6, 1994; revised January 19, 1995) Keywords:

Tissue-specific/promofer/mammaliangene regularion/transcriprionfacror/neuron

INTRODUCTION The development of a multicellular organism from a fertilised egg involves the progressive and hierarchical restriction of gene expression in time and space. Eventually, each differentiated tissue comes to express a characteristic set of gene products which confer upon that tissue its unique biochemical and physiological properties. In mammals, many genes are expressed only in the nervous system and more specifically in the neuronal cell lineage. Some of these genes are panneuronal, i.e. they are expressed in all (or the great majority of) neurons and are responsible for those characteristics general to all neuronal cell types (e.g. axon growth, electrical excitability, formation of synaptic junctions). Others are subneuronal, i.e. they are expressed in a subset of neuronal cell types and are responsible for the individual characteristics of particular subpopulations of cells (e.g. neurotransmitter phenotype). In this review, we explore the molecular basis of neuronal gene expression in the mammalian nervous system and summarise what is known about the regulation of both pan-neuronal and subneuronal genes. The mammalian nervous system originates from two major embryonic tissues. The central nervous system (CNS) is derived from the neural tube whilst the majority of the peripheral nervous system (PNS) is derived from the neural crest, the remainder originating from placodes of ectodermal origin. Within the adult nervous system, there is an incredible variety of cell types, the origins of which are considered in two excellent reviews concentrating on the CNS and PNS lineages respectively (Anderson, 1989: McKay, 1989). In the last few years, much attention has been directed towards the funcCorrespondence to: Dr. Richard Twyman, Dept. of Biological Sciences, University of Warwick, Coventry CV4 7AL. England.


68

R . M . TWYMAN A N D E. A. JONES

tion and regulation of genes involved in the developing nervous system, specifically to those genes responsible for the determination of cell fates (reviewed by Lemke, 1993). Other groups have aimed to explain the effects of neurotrophic factors, a family of polypeptides whose members exert development-related effects upon neurons (see Barde, 1991); of these, the best characterised is nerve growth factor (NGF) (LeviMontalclini, 1987). Neurotrophic factors are thought to exert their effects by stimulating the expression of early response or immediate early genes (Sheng and Greenberg, 1990), a class of genes encoding transcription factors which directly or indirectly regulate the expression of the terminal or late response genes in the nervous system, including those considered in this review. Strategies for the Analysis of Neuron-Specific Gene Regulation Most genes are regulated at the level of transcription, primarily by interaction between diffusible trans-acting factors and cis-acting DNA elements located near to the transcribed region of the gene (reviewed by Maniatis et al., 1987). Generally, the cis-acting elements consist of a core basal promoter which is absolutely required for minimal gene expression and any number of additional upstream regulatory elements which modulate that expression either constitutively (i.e. in all environments) or confer upon the gene its spatial, temporal and inducible specificity. It is desirable to identify particular functional elements within this upstream region as such elements often represent sites where trans-acting factors bind to the DNA. Genes may be studied as intact units, in which case a suitable assay must be available to detect and measure the gene product. In such cases, the gene is usually studied in a surrogate environment, e.g. a transgenic mouse containing an integrated human gene. It is usually more convenient, however, to study gene regulation indirectly by fusing the putative control elements of the gene under investigation to a reporter gene, a heterologous gene whose product can be assayed easily and quantitatively; common reporter genes used in animal studies include those encoding the enzymes P-galactosidase, chloramphenicol acetyltransferase (CAT) and luciferase. The advantage of using reporter constructs, rather than intact genes, is that assays can easily be carried out in cell lines or in vivo where the endogenous gene is also expressed. A general strategy for the analysis of gene regulation is to identify important cis-acting elements in the flanking sequence of a given gene by deleting, rearranging, substituting or otherwise modifying specific regions. Preliminary results may be obtained by simple stepwise deletions but these may need to be refined and reinforced by observing the effects of internal deletions, linker scanning mutations and finally, specific point mutations. There are two major approaches to the analysis of gene regulation: a cell line approach, involving the transfection of reporter constructs into a variety of permissive and nonpermissive cell lines, and a transgenic approach, involving the analysis of gene expression in vivo. Each has its advantages and disadvantages (see Table I). Gene expression may also be analysed in vitru using protein extracts from cell lines and organs; in vitro transcription assays may succeed where the transfection approach has failed (see e.g. Schwartz et al., 1994). Usually, the in vifro approach is used to identify protein binding sites, e.g. by using gel retardation and footprinting assays.

THE REGULATION OF NEURON-SPECIFIC GENE

69


TABLE 1 Advantages and Disadvantages of the Cell Line and Transgenic Approaches to the Analysis of Gene Regulation. Cell line approach: Advantages: Easy technique with rapid results. Many constructs can be studied during a single experiment. Can study specific inductive responses and rrans-activationlrepression. Disadvantages: No indication of temporal or spatial regulation. Often limited by properties of available cell lines; sometimes difficult to identify cell lines with appropriate properties. Established cell lines may not accurately represent conditions in vivo. Introduction of too much DNA may titrate out available transcription factors. Transfected DNA does not respond to endogenous cis-acting factors such as distant enhancers, chromatin effects, replication timing and methylation. Transgenic approach:

Advantages:

Spatial and temporal gene regulation can be observed, often at single cell resolution. Transgenes respond to true in vivo conditions. Usually only a limited number of copies integrated so that transcription factors will not be titrated out. Disadvantages: Technique more difficult and time-consuming, making it less easy to study many constructs at once. Integrated gene may be subjected to unfavourable position effects.

General Properties of Mammalian Neuron-Specific Genes

As discussed above, neuron-specific genes are not all expressed in the same manner. Some may be pan-neuronal, whilst others may be expressed in a specific subset of neuronal cell types. Still other genes are expressed preferentially in neurons but also in other, nonneuronal tissues. Even within the pan-neuronal genes, the spatial, temporal and inducible qualities of expression may differ. For instance, some genes are expressed preferentially in immature neurons (e.g. SCGIO, GAP-43) or in the peripheral nervous system (e.g. peripherin). Is there any quality which is particular to genes expressed in neurons? The simple answer to this question is no, there is no particular feature which would enable one to identify a neuronal gene given a cursory glance at the sequence of its promoter. However, recent studies of the regulation of a number of mammalian neuron-specific genes have identified several conserved elements which may indicate shared regulatory mechanisms; these are discussed below. Many neuronal genes possess an undermethylated, GC-rich promoter, lacking a canonical TATA box and the CAAT boxes common to regulated class I1 genes; this type of promoter is usually associated with housekeeping genes but has also been found flanking a number of regulated genes. Examples of neuronal genes with this type of promoter include those encoding neuronspecific enolase, the type I1 sodium channel, synapsins I and I1 and the DtAdopamine receptor. Other neuronal genes (e.g. those encoding the neurofilament proteins and GAP-43)possess normal class I1 promoters. Recent studies indicate that neuronal gene expression can be achieved by three principle mechanisms (see Figs. la-c) which may operate alone or in combination. In each case, further elements are often present which act constitutively. Genes expressed in specific neuronal subpopulations (rather than pan-neuronally) are often regulated by modular elements which act to restrict expression driven by a more promiscuous underlying

70

R. M . TWYMAN A N D E. A. JONES

mechanism (Fig. Id). In the following sections, we consider the four mechanisms of neuronal gene regulation using well-characterised examples from mammalian systems. Later, we provide a gene table which summarises what is known about the regulation of a variety of neuronal genes. MECHANISMS OF NEURON-SPECIFIC GENE EXPRESSION J Neurogenet Downloaded from informahealthcare.com by RWTH Aachen Hochschulbibliothek on 12/09/10 For personal use only.

Neuron-Specific Gene Expression Conferred by a Basal Promoter A number of neuronal genes have been identified and characterised by the presence of functional cell-specific regulatory elements very close to the transcriptional start site; this means that most of the 5' flanking region can be deleted without affecting the specificity of gene expression. The smallest functional regulatory element is often less than 300bp in length but this could be attributed to either a specific basal promoter (Fig. la) or to a nonspecific promoter with unusually proximal upstream regulatory elements (Figs. lb and Ic). One gene for which the existence of a specific basal promoter has been proved beyond doubt is the mouse gene for the neuronal intermediate filament protein peripherin. This gene is pan-neuronal but preferentially expressed in the PNS. The 5' flanking region was dissected by Desmaris et al. (1992) who showed that as little as 98bp of sequence upstream of the transcriptional start site was sufficient for cell-type specific expression in transfection assays. Minute analysis of this region by DNase footprinting identified three protected fragments: PERI, PER2 and PER3. Targeted mutation of these regions showed that PER1, which overlapped the TATA box, was required for cell-type specific expression whilst PER2 and PER3 were required for general, constitutive upregulation. By overlapping the TATA box, the neuron-specific element PER1 probably interacts with the basal transcription apparatus and cannot, therefore, be separated from the basal promoter. Its deletion would therefore cause a loss of expression in all cells and this is the essential point of Figure la. Two possible mechanisms of core promoter specificity have been reported: Tamura et al. ( 1990) have shown that different forms of the basal transcription factor TFIID exist in extracts of brain and liver, and that these factors differentially support expression of the myelin basic protein gene. Thus, neuronal specificity could be brought about by the existence or activity of a basal transcription factor only in neuronal cells. Secondly, Wefald et al. (1990) have shown that the context of the TATA box may influence the way in which ubiquitous transcription factors interact with tissue-specific factors located elsewhere in the gene and this is another possible explanation. Neuron-specific core promoters are also thought to control transcription of the rat genes for GAP-43 (neuromodulin, B-50), synapsin I and calmodulin 11 as well as the mouse gene for synapsin 11. However, these regions have not been so carefully dissected and might conceivably contain proximal regulatory elements upstream of a nonspecific promoter. GAP-43 encodes a pan-neuronal axonal growth-associated protein which is expressed preferentially i n immature and regenerating neurons. Dissection of the 5' flanking region (Nedivi et al., 1992) revealed a neuron specific promoter extending 386bp upstream of the transcriptional start site. Interestingly, this core sequence was

t Y

Y

Min

Min

r :

1-"

N S promoln

NSoromolr

'?

n

NSnhnM

I Rornotr Y

Min

L

-I'

5 ;

N

z

Exmpler Q.2

d) Submuronal-Wecific updream elements reatricting a more general underlying neuron-gecific mechanisn

bmp1.r N a l , SCGlO

n

E

I

N

z

r :

b) Romiseuour promoter with neuron gecific ilcncsr

FIGURE I Principle mechanisms of neuron-specific gene expression based upon the results of hyphetical transfection experiments using neuronal and nonneruonal cells (see text for details). Example genes for each mechanism are given. Abbreviations NS = neuron-specific; Min = minimal gene expression.

1

Z

5 ;

c) Minimal promoter with neuron pecific enhancer 7

E x m p l a G q 4 3 . Rriphrin

EffQ of rqumhd ddmionc

I

a) N wron-rpccific b a d promoter



12

R. M. TWYMAN A N D E. A. JONES

flanked by two unusual DNA elements; one, a short sequence of alternating purines and pyrimidines with the potential to form left-handed Z-DNA (Rich et al., 1984) and the other. a long polypurine sequence with the potential to form triple stranded H-DNA by Hoogsteen base pairing (Htun and Dahlberg, 1989). In transfection experiments, deletion of either of these unusual elements resulted in the severe repression of transcription, but if both were present, the adverse effects were ameliorated thereby supporting the notion that gene expression was mediated by countermodulation involving interaction between the elements. Synapsins are membrane-associated phosphoproteins localised in presynaptic termini throughout the nervous system. There are four synapsin proteins, (Ia and Ib, IIa and IIb) which are produced by alternative splicing of the two synapsin genes. Neuron-specific core promoters are thought to be present in the genes for rat synapsin I and mouse synapsin I1 (Saurwald et al., 1990; Chin et al., 1994; Howland et al., 1991) but not in the gene for human synapsin I (G. Thiel et al., 1991). All the promoters appear to be of the housekeeping type, with no canonical TATA box, but there is only one transcriptional start site for each of the genes. The 5' flanking region of rat synapsin I was dissected by Saurwald et al. (1990) who found that 225bp of upstream sequence and 105 bp downstream of the transcriptional start site was sufficient to drive cell type specific expression in transfection experiments. Thiel et al. (1991) showed, however, that the human synapsin I gene was driven by a nonspecific basal promoter, which extended 115bp upstream of the transcriptional start site, and one or more neuron-specific enhancer elements located farther upstream but within 422bp of the start. The 5' region of the rat synapsin I1 gene was analysed by Chin et al. (1994) and was shown to comprise a neuron-specific promoter extending 153bp upstream of the transcriptional start site and a number of neuron-specific positive and negative elements located further upstream. The synapsin genes are therefore illustrative of the fact that gene regulation is usually more complex than shown in Fig. 1, often involving a combination of alternative mechanisms. Promiscuous Transcription Repressed by a Neuron-Specific Negative Modulator

A neuron-specific negative modulator would function in nonneuronal cells and would confer cell type specificity upon a relatively nonspecific promoter. By sequential deletion, it should be possible to remove such an element and allow the expression of a reporter minigene in a wider variety of cell types. These are the essential features of Figure lb. Should such an element function in an orientation and position independent fashion, and be capable of conferring neuronal specificity upon a heterologous promoter, it would then constitute a silencer element (Brand et al., 1985). The first gene for which a neuron-specific negative element was identified was the rat gene for the type I1 sodium channel (NaII) protein. Sodium channels are transmembrane gated ion channels which confer upon neurons and other cells the property of electrical excitability. The type I1 sodium channel is one of four proteins differentially expressed in neurons and belongs to a large protein family, other members of which are expressed in muscles and glia (for a review, see Mandel, 1992). Of the neuronal sodium channel genes, the type I1 gene is the best candidate for the study of neuron-specific gene regulation; this is because it is the only gene expressed strongly and pan-neuronally in the adult



73

nervous system. A reporter gene, driven by 1051bp of NaII 5‘ flanking sequence was generated by Maue et al. (1990). In transfection experiments, this construct was expressed in cells of neuronal origin but neither in excitable nor nonexcitable cells of nonneuronal origin. Stepwise 5’ deletions identified a number of negative regulatory elements, the strongest of which was located between 983 and 1051bp upstream of the transcriptional start site; when this region was deleted, a ninefold upregulation of reporter expression in nonneuronal cells was observed. The functional element (repressor element, R E l ) was identified by footprinting as a 28bp motif with the sequence ATTGGGTTTCAGAACCACGGACAGCACC (Kraner et al., 1992). This element fulfilled all the properties of a silencer and was later renamed the neuronal restrictive silencer element (NRSE) (Mori et al., 1992). Kraner et al. also performed gel retardation assays and identified a protein present only in nonneuronal cells which bound specifically to the NRSE. The regulation of the rat neuron-specific SCGIO gene was studied by transfection (Mori et a]., 1990) and in transgenic mice (Wuenschell et al., 1990). SCG10 is another axonal growth-associated protein which is homologous to the ubiquitously expressed protein strathmin but unrelated to GAP-43 (for a review, see Okazaki et al., 1993). 500bp of the 5’ flanking sequence was shown to confer no cell type-specificity either in cell lines or in vivo, although a number of constitutive enhancer elements were identified. However a construct containing 4kb of upstream sequence was neuron-specific. Further characterisation of this upstream region (Mori et al., 1992) revealed an element which was very similar in sequence and function to the Null silencer, and appeared to bind the same protein, which was termed the neural-restrictive silencer binding factor (NRSBF). The discovery of similar functional cis-acting elements in two neuron-specific but otherwise unrelated genes prompted a search for conserved elements in other neuronal genes (see Table 11). Homologous elements were identified in both the human and rat TABLE 11 Rat SCGlO

ggtTTCAGAACCACGGACAGCACCagagt

Rat NaII

catTTCAGCACCACGGAGAGTGCCtctgct

Rat synapsin I

agcTTCAGCACCGCGGACAGTGCCttcgc

Human synapsin I

ggaTTTAGTACCGCGGACAGAGCCttcgc

Human DBH Rat BDNF

Conmenmum

gGTCAG..CGCTGGACAGCTCCtcg TTCAGCACCTTGGACAGAGCCa, I 11111 11111 I I1 g TGATGGTGGAGCCTGTTTAGGC’

...TTCAGNACCACGGACAGNGCC.... T

~~~~~~~~~~

G

G

A

~~

Comparing the sequences of the known neural restrictive silencer elements and homologous elements in other neuronal genes (adapted from Figure 4 in Mori et al., 1992). In the human DBH element. maximum alignment has been achieved by inserting gaps (represented by stops). Base pair complementarity is shown between the two contiguous sequencer present in the BDNF promoter


14

R. M. TWYMAN AND E. A. JONES

synapsin I promoters, but functional analysis of these sequences (Saurwald et al., 1990; G. Thiel et al., 1991) failed to demonstrate neuron-specific silencer activity. L. Li et al. (1993) reported that the putative silencer element located in the human synapsin I promoter was indeed functional and that deletion of the element caused an upregulation of synapsin I-reporter expression in nonneuronal cell lines, however, statistical analysis of the results suggested that the observed upregulation was not significant. Thiel et al. have recently shown that the silencer element in the human synapsin I promoter overlaps a binding site for the positive transcriptional activator Krox-24 (Lemaire et al., 1990), which is encoded by a cellular immediate early gene (Sheng and Greenberg, 1990). They proposed that interaction between Krox-24 and the NRSBF might play a pivotal role in the regulation of this gene and they have shown that the NRSBF can prevent transactivation of synapsin I reporter constructs by Krox-24 in vitro (G. Thiel et al., 1994). However, contrary to earlier reports (Mori et al., 1992; Kraner et al., 1992) the NRSBF was shown not to be restricted to nonneuronal cells. Two NRSE-like sequences, inverted with respect to each other, have also been identified in the brain-specific flanking region of the gene for brain-derived neurotrophic factor (BDNF) (Timmusk et al., 1992). The BDNF gene is most complex, regulated by four individual promoters, two of which are brain-specific. Although the relevance of the NRSE-like sequences has not been defined by functional analysis, their location between the two brain-specific promoters may indicate a regulatory function. A homologous element has also been identified in the 5’ flanking region of the human dopamine P-hydroxylase (DBH) gene which has been shown to possess a relatively promiscuous basal promoter (Ishiguro et al., 1993). If functional, however, the silencer must work in concert with a CAMP responsive neuron-specific enhancer which was shown by Ishiguro et al. to be essential for cell type specific expression. The equivalent region in the rat DBH promoter does not contain a hornologous silencer element (Shaskus et al., 1992). Putative neuron-specific negative rnodulators have been identified in a number of other genes including synaptophysin (Bargou and Leube, 1991) but these have not been further characterised. Minimal Nonspecific Basal TranscriptionActivated by a Neuron-Speci$c Positive Modulator A neuron-specific positive modulator would function in neuronal cells and would upregulate transcription from a constitutive minimal promoter. By sequential deletion, it should be possible to remove such an element and return the expression of a reporter minigene to a minimal level in all cell types. These are the essential features of Figure lc. Should such an element function in a position and orientation independent fashion and be capable of conferring neuronal-specificity upon a heterologous basal promoter, it would then constitute an enhancer (see Maniatis et al., 1987). As positive regulation is a common feature of eukaryotic gene function, it is not surprising that a number of neuron-specific genes should possess binding sites for specific transcriptional activators. In some cases, however, such elements have been difficult to identify. A number of filament proteins are expressed specifically in neuronal cells including the intermediate neuronal filament proteins NF-H (heavy chain), NF-M (mid



75

range chain) and NF-L (light chain). Early reporter-transfection studies, using considerable lengths of 5' and 3' flanking DNA from the various neurofilament genes, identified a number of constitutive positive and negative regulatory elements but failed to isolate elements required for cell type specificity. Thus, reporter constructs were found to be expressed with equal efficiency in neuronal and nonneuronal cell lines alike (Julien et al., 1987; Monteiro and Cleveland, 1989; Nakahira et al., 1990; Pleasure et al., 1990; Schneidman et al., 1992; Zopf et al., 1990). This was also found to be true for the a-internexin gene, which encodes another neuron-specific intermediate filament protein (Ching and Liem, 1991). Evidence from transgenic mouse experiments showed, however, that reporter transgenes driven by relatively short regions of 5' and 3' flanking sequence were properly regulated, indicating that all the required cis-acting elements were present near to the transcriptional start site (Byrne and Ruddle; 1989; Julien et al.. 1988; 1990; Beaudet et al., 1992; Lee et al., 1992; Reeben et al., 1993; Yazdanbakhsh et al.. 1993). This discrepancy was attributed to the artificial nature of the transfection assay (see Table I) but deletion analysis in transgenic mice also failed to identify specific functional control elements. In conflicting reports, Yazdanbakhsh et al. (1993) claimed to have identified a neuron-specific positive element in the flanking region of the human NF-L gene, located between 190 and 300bp upstream of the transcriptional start site whilst Beaudet et al. (1992) claimed that 300bp of upstream sequence was insuficient to confer neuronal-specificity upon a reporter transgene but that 4.6kb of downstream information, combined with the minimal promoter, could confer cell type specificity in transfection experiments. To overcome the limitations of the transfection assay, Schwartz et al. (1994) have analysed the regulation of the mouse NF-H gene by in vitro transcription using protein extracts from brain and liver. They found that a minimal promoter extending 65bp upstream of the transcriptional start site was capable of sponsoring relatively strong reporter expression using extracts from both tissues but that a region located between 65 and 115bp upstream of the start site was responsible for brain-specific enhancement of transcription. Within this region, a sequence displaying perfect dyad symmetry was shown to be essential: GGGGAGGAGG N 1 5 CCTCCTCCCC. Interestingly, deletion analysis showed that only the pyrimidine-rich repeat was essential and that the complementary, purine-rich element could be mutated or deleted without effect. Identical or highly conserved elements have been found near to the transcriptional start sites in all three human and murine neurofilament genes suggesting a common overall mechanism of spatial regulation, however, as the genes are differentially regulated during development (Julien et al., 1986) further controls may be superimposed upon this general regime. Interestingly, in the human and murine NF-L genes, the enhancer is found downstream of the transcriptional start site, supporting the conclusions of Beaudet et a]. (1992). The NF enhancer has not been found in the flanking regions of any other genes and is presumed to be unique to this closely related gene family. Cell-specific transcription of the rat tyrosine hydroxylase gene has also been traced to a neuron-specific enhancer. Transfection analysis in PC8b cells located the enhancer just upstream of the transcriptional start site (Yoon et al., 1992). The TH enhancer comprises an AP- I binding motif (TGATTCA) with an overlapping 20bp element displaying imperfect dyad symmetry. This element has an E-box core, which features in other tissue-specific enhancers such as the exocrine specific Pan element and the im-

R . M. TWYMAN A N D E. A. JONES

76


munoglobulin light chain enhancer kE2 (Nelson et al., 1990; Murre et al., 1989). Mutational analysis carried out by Yoon et al. has shown that both the E-box motif and the AP-I binding site are required for enhancer function and that their spatial relationship is also important, indicating a close interaction between them. More recently, Wong et al. (1994) have shown that the AP-I/ E-box dyad motif is both insufficient and nonessential for neuron-specific expression of rat TH reporter constructs in PC 12 cells. Instead, a functional enhancer sequence was located 500bp upstream of the transcriptional start site. Gene Expression in Subsets of Neurons may involve Combinatorial Controls Superimposed upon a more Promiscuous Mechanism As discussed above (see Table I), the transfection approach to the analysis of gene regulation can provide information concerning the mechanism of gene expression in particular cell lines and the response to applied inducing factors but can provide no information with respect to the temporal or spatial regulation of gene expression. Studies of subneuronal gene expression must therefore be carried out either in very carefully chosen cell lines or in vivo, using transgenic mice. Many neuron-specific genes have been analysed in transgenic mice but these studies are usually limited to showing that a particular length of flanking DNA is capable of conferring correct spatial and temporal expression. Relatively few genes have been dissected in a rigorous manner, to identify elements responsible for spatiotemporal control, but those encoding enzymes of the catecholamine synthesis pathway are an exception. The genes involved in neurotransmitter synthesis and function are prime examples of subneuronal-specific genes. The transmitter phenotype of a particular neuron depends upon the combination of enzymes made in the cell (for example, the pathway to catecholamine synthesis is shown below (Figure 2 ) ) and neurons which synthesise particular neurotransmitters are located in characteristic parts of the nervous system. One would expect that the enzymes involved in neurotransmitter synthesis, activity and degradation would be synthesised in specific overlapping spatial domains, and this would be reflected in the promoters of the corresponding genes.

Tyrosine

U

TYROSINE HYDROXYLASE(TH)

L-DOPA

U

AROMATIC AMINO AlCD DECARBOXYUSE (AADC)

Dopamine

U

DOPAMINE p-HYDROXYUSE (DBH)

Noradrenaline

U

PHENYLETHANOLAMINEN-METHYLTRANSFERASE(PNMT)

Adrenaline FIGURE 2 Catecholamine synthesis in the mammalian nervous system. The pathway is shown with the enzymes to the right.

10 the

lrft



77

Hoyle et al. have recently investigated the regulation of the human D B H gene in transgenic mice using a nested set of 5’ deletions (Hoyle et al., 1994). Endogenous DBH expression is restricted to noradrenergic neurons (i.e. most sympathetic and parasympathetic neurons) and adrenal chromaffin cells (anaxonal neurons which secrete adrenaline and noradrenaline into the blood). Transgenic mice carrying 5.8kb of the human DBH 5’ flanking region express the reporter-transgene in those cells where endogenous D B H mRNA is detected but also in ectopic tissues such as the dorsal root ganglia and dopaminergic and noncatecholaminergic neurons of the brain. Transient ectopic expression is also observed during development in the spinal cord and facial mesenchyme (Mercer et al., 1991; Kapur et al., 1991; Hoyle et al., 1994). The ectopic expression was thought to be due to the absence of negative modulators located outside the limited regulatory DNA flanking the reporter gene. Deletion of 5’ flanking sequence to 1.1kb upstream of the transcriptional start site allowed further ectopic expression in the hypothalamus, septum and olfactory bulb, however neither this expression, nor the ectopic expression observed for the 5.8kb 5’ sequence, was observed if 1.5kb of flanking DNA was present. Hoyle et al. therefore suggested that the region between 1.1 and 1Skb upstream of the transcriptional start site was responsible for restricting the pattern of expression conferred by the more promiscuous 1.1kb sequence but that further elements, 1e cated between 1.5 and 5.8kb upstream permitted expression in other cells and that these elements were repressed by sequences elsewhere. Finally, deletion of the 5’ flanking sequence to within 600bp of the transcriptional start site abolished reporter gene expression altogether suggesting the presence of an essential positive element between -600 and -1 100bp. It is noteworthy that transfection analysis using the same promoter (Ishiguro et al., 1993) identified a functional NRSE 500bp upstream of the transcriptional start site and an essential CAMPresponsive element (CRE) less that 300bp from the transcriptional start site. The loss of expression in transgenic mice could not be due to the removal of either of these elements and might be caused by unfavourable position effects. Because all cells expressing DBH should also express T H , Hoyle et al. put forward the idea that the ectopic expression from the D B H minigene might correspond to the endogenous expression of T H . This would further suggest that the genes involved in catecholamine synthesis could be regulated according to a common paradigm with extra levels of restriction imposed upon those genes encoding downstream enzymes. Although this was an attractive proposal, comparison of promiscuous D B H reporter expression with endogenous T H mRNA showed that some populations of cells expressed the DBH reporter but not T H , and that the domains of expression of the two genes did not match. Furthermore, Shaskus et al. (1992) has compared the proximal flanking regions of the rat D B H and T H genes and although two conserved regions were found, deletion analysis showed that neither were required for cell type-specific expression and that the conserved region was not present in the human T H promoter. A striking example of subneuronal-specific gene regulation imposed upon an underlying pan-neuronal mechanism comes from study of the Purkinje cell protein 2 (Pcp-2) gene (Vandaele et al., 1991). Pcp-2 is expressed specifically in cerebellar Purkinje cells and retinal bipolar neurons, and transgenic mice carrying 4kb of 5’ flanking sequence express the reporter gene faithfully in the Purkinje cells (Oberdick et al., 1990). Vandaele et al. showed that a transgene driven by only 400bp of the flanking sequence was ex-

78

R. M. TWYMAN AND E. A. JONES

pressed in a multitude of neuronal cell types whilst a transgene driven by 3.5kb was expressed in the same manner as the endogenous gene. This showed that negative elements, located between -400 and -3500bp,were responsible for the restriction of expression to certain subneuronal cell types and that this restriction had been imposed upon a more promiscuous (but still neuron-specific) core region.


Other Conserved Features of Neuron-Specific Genes

The studies outlined above have identified a number of cis-acting motifs which appear to confer neuronal cell type specificity; these include the neuronal restrictive silencer element or NRSE (Mori et al., 1992) and the neurofilament enhancer (Schwartz et al., 1994). A number of further motifs have been found to be conserved between otherwise unrelated neuronal genes and these are described below: a ) A nonfunctional consensus? Maue et al. (1990) first noted a conserved sequence with the core motif CCAGG in the 5’ flanking regions of four neuronal genes: the rat type I1 sodium channel and peripherin genes, the mouse NF-L gene and the Drosophila dopamine decarboxylase gene. The sequence has been found to be required but not sufficient for CNS expression in Drosophila (Scholnick et al., 1986) and was termed element 1, however, in the sodium channel gene, its deletion was shown to have no effect (Maue et al., 1990). Later, the list of neuronal genes canying the consensus sequence in their flanking sequences was expanded by Vandaele et al. (1991). Homologous elements have also been reported upstream of the mouse synapsin I1 gene (Chin et al., 1994). the rat GAP-43 gene (Nedivi et al., 1992) and the mouse OMP gene (Kudrycki et al., 1993); all the known sequences are aligned for comparison in Table 111. In all cases where the sequence has been manipulated in mammalian genes, it has proved both insufficient and nonessential to confer neuronal specificity upon a reporter construct. Its existence in the promoters of so many neuronal genes is therefore a mystery, but the observed conservation with the Drosophila element 1 would suggest an important function. b) The SNN motif Another conserved motif was identified by Saurwald et al. (1990); this is termed the SNN motif because it is found in the rat and human Synapsin I promoters, the mouse NF-M promoter (Lewis and Cowan, 1986) and the human NGF receptor gene promoter (Sehgal et al., 1988). Although the function of the sequence has not been tested, such analysis may be difficult because in the synapsin I genes, the downstream moiety of the motif overlaps an essential CAMPresponse element and in both synapsin I and the neurofilament promoters, the upstream moiety forms a consensus Krox-24 binding site which may be required for proper gene regulation.

SNN Consensus CG/CTTC/TGCCC/TCC/TGC - N3-7

- CGC/GGCTGNC

c ) The Identifier (ID) sequence In the early 1980s, Sutcliffe et al. identified two small cellular RNA (scRNA) molecules, called BCl and BC2 respectively, which were expressed specifically in the rat brain and peripheral nervous system by RNA polymerase I11 (Sutcliffe et al., 1982; 1984). A middle repetitive DNA element dispersed throughout the rat genome was found to hybridise to these scRNA molecules and it was later discovered that the sequences were preferentially located near to or within postnatally ex-


19

TABLE 111 Sequence

(-289-2 10)

.........

AQAGGCTC aEATCCTG AGATGCTQ

Rat NSE Rat NSE Rat NSE

(-9951-984) (-8231-771) (-662’-64I )

AQACGCTG

Human NSE

(-7631-782)

Hunran NF-M

(502/543)

Rat Pcriphcrin Rat Pcriphcrin Rat Pcripherin

(-404/- z

h)

m

Rat

Human

NtrII (Type I I sodium channel)

NGF receptor

Neurofilament genes ( N F - H , N F - M , NF-L)

Rat (r) Typical Class II Mouse (m) Human (h) Chicken (ch)

Mouse

GC-rich Atypical TATA AT-rich No TATA Multiple starts Housekeeping tY Pe

RPt

a-liiternexin

Pan-neuronal and nonneuronal descendants of neural crest Pan-neuronal

Pan-neuronal Pref. in adult brain

CNS

Pan-neuronal Axonal growth cones Transient expression in nonneuronal cells of chick limb bud.

Typical Class II with unusual DNA flanking elements. Multiple starts

Rat

CAP-43 (B-50,

neuromodulin)

Expression

Promoter type

Species

Gene

Relevant Studies

Second promoter identified downstream of neuron-specific core promoter which is active during development. Transfection analysis: 5kb 5’seq. insufficient for cell type specificity. Transfection analysis: Deletion analysis shows promiscuous basal promoter restricted by distant, cell type specific silencer element, prototype NRSE. Transfection analysis: 1.2kb of 5’seq. allows expression in fibroblasts. Cell type specific negative regulatory element located I .7kb upstream of transcriptional start site contains E-box motif. Tran\fection analysis: Various NF genes with extensive flanking seq. are expressed with equal efficiency in neuronal and nonneuronalcell lines. Most deletion studies fail to idcntify cell type specific elements although a number of constitutive elements are found. Yazdanbakhsh et al. (1993) claim to have identified cell type specific enhancer in human NF-L gene.

Transfection analysis: Deletion analysis shows 386bp core promoter is cell type specific. Negative countennodulation by unusual DNA elements flanking this region. Function of minimal promoter almost fully conserved in transgenic zebra fish. 8.4kb 5’ seq. drives neuron-specific reporter expression in transient Xenopus DNA microinjection assays.

TABLE IV Genes (Continired)


I990 (niNF-L) Monteiro and Cleveland, 1989 (mNF-L) Pleasure et al., 1990 (hNF-M) Schneidman ct 31.. 1992 (mNF-H. and NF-L) Yazdanbakhsh et al.. 1993 (hNF-l Zopf el al., 1990 (chNF-M)

Neuman et al., 1993b Julien ct al.. 1987 (hNF-L) Nakahira et al.,

Chin and Liem. 1991 Maue et al., 1990 Kraner et al.. 1992 (Mori et 81.. 1992) Sehgal ct al., 1988

Reinhardt et al., I994 Verhaagen et al.. I993 Eggen et al., 1994

Nedevi et al., 1992 Starr et al., 1994

References

-

rn Z rn

0

2

c!

rn

~

v)

7

0

C x1

rn

Z

$1

z

>

L

rn

-rn

x1

r rn

-1

NSE (neuronspecific enolase)

Neurofilament genes (Continued)

Gene

Rat

Species

GC-rich Atypical TATA

Promoter type

Pan-neuronal and neuroendocrine

Expression

Transgenic analysis: 1.8kb 5' seq. drives correctly regulatcd reportcr expression in transgenic mice. Herpes virus vectors used to deliver genes to rat brain; reporter genes expressed under the control of I .8kb NSE 5' seq. Transfection studies: Deletion analysis shows that 93bp minimal promoter sufficient for cell type specific expression.

In v i m analysis: h tifro transcription using brain and liver extracts identifies tissue-specific enhancer in mouse NF-H gene; homologous elements in all other NF genes. A number of transcription factor binding sites are identified. including those for Krox-24.

Various NF genes with limited flanking seq. are correctly regulated in transgenic mice indicating that all relevant ci.\-ncting sequences are present. Deletion analyses fail to identify neuron-specific elements, although Beaudet et al. (1992) identify downstream neuron-specific elements in human NF-L.

Cote el al., 1993

Transgenic analysis: Homologous and hcterologous NF genes overexpressed in transgenic mice cause recognisable neuropathological disorders.

WF-W Xu et al.. 1993 (mNF-L) Beaudet et el., 1992 (nHF-L) Lee et a]., 1992 (WF-M) Reeben et al., 1993 (rNF-L) Julien et al.. 1988; 1990 ( W F - L ) Monteiro et al.. 1990 (mNF-L) Byrne and Ruddle, 1989 (mNF-L) Vidal-Sanz et al., 1991 ( W F - L ) Schwartz et al., 1994 (mNF-H) Elder et al., 1992a; b ( W F - H )and lvanov and Brown, 1992 (mNF-L) Pospelov et al., 1994 ( W F - L ) Forss-Petter et al.. 1990 J Andercen et al.. 1992; 1993 Twyman and Jones,unpublished results

References

Relevant Studies


LI

?

P

> z

0

m P

Rat

Oxytocid Vasopressin

Human PNMT (Phenylethanolamine N methyltransferase) SCGlO Rat

Rat

Preprotachykinin (Substance P)

Typical Class I1 Multiple starts

GC-rich Atypical TATA

Three TATA Boxes

Inner nuclear layer of retina. Adrenal gland Pan-neuronal Axonal growth cones

Typical Class 11 PNS

Transfection analysis: Deletion analysis shows promiscuous basal promoter restricted by distant, cell type specific silencer element (NRSE).

Transfection analysis: 865bp 5’ seq. confers neuron-specificity to reporter gene but expression is not restricted to those neuronal populations expressing substance P, nor.is reporter NGF inducible. Neuron-specific binding site identified on border between intron I and exon 1. Transgenic analysis: 2kb of 5’ seq. drives correct expression of reporter gene in transgenic mice.

Mouse

Prripherin

Dorsal root ganglia

Mouse

PEP-I9

Transgenic analysis: 3.5kb of 5’ seq. drives correctly regulated reporter expression in transgenic mice. 400bp of 5’seq. drives expression in a variety of neuronal cell types. Transgenic analysis: 1.35kb 5‘ seq insufficient for expression in transgenic mice but this element does specifically bind proteins from cerebellar extract. Transfection analysis: Deletion analysis shows 98bp promoter sufficient for cell type specific expression. Three footprints in this region; one (PERI) is required but insufficient for specificity. PER2 and 3 are essential constitutive elements.

Typical Class 11 Retinal bipolar neurons. Cerebellar Two start sites Purkinje cells.

Mouse

Pcp-Z/LI (Purkinje cell protein 2)

Transgenic analysis: 5.2kb transgene including oxytocin and vasopressin genes allows tissue-specific expression of oxytocin but not vasopressin in transgenic mice. Transgenic analysis: I .25kb of 5’ Vasopressin sequence insufficient for tissue-specific expression. Construct containing 9kb 5’ sequence and I .5kb 3’ sequence is expressed in a tissue-specific manner Transgenic analysis: 0.3kb 5’ seq. sufficient for tissue-specific reporter expression in transgenic mice. This sequence contains one Olf-1 binding site which is also present in several other olfactory neuron-specific genes.

Relevant Studies

Pan-neuronal

GC-rich No TATA Olfactory neuroepithelium

Hypothalamus

Expression

Mouse

Promoter type

O M P (Olfactory Marker Protein)

Bovine

Species

Gene

TABLE IV Genes (Continued)


Mori et al., I990 Mori et al.. 1992

Baetge et al., 1988

Desmaris et al., 1992 Karpov et al., 1992 Thomson et al., 1992 Mulderry et al., 1993 Quinn, 1992

Kudrycki et al., 1993 Wang et al., 1993 Danciger et al.. 1989 Oberdick et al.. 1990 Vandaele et al.. 1992 Sangammeswand and Morgan, I993

Ang el al.. I993

Young et al., I990

References

Mouse

Rat

Rat

Synapsin II

Synaptophysin

TH (Tyrosine Hydroxylase)

Human

Rat

Synapsin I

Species

Rat

~

SCGIO

Gene

Multiple starts Typical Class 11

tY Pe

Housekeeping type Single start Housekeeping

Housekeeping type Single start

Promoter type

CNS-Mainly olfactory bulb Sympathetic ganglia Adrenal gland

Pan-neuronal and neuroendocrine

Pan-neuronal

Pan-neuronal

Expression

Relevant Studies

Transgenic analysis: 4.3kb 5‘ seq. sufficient for correct tissuespecific expression. Transfection analysis: Deletion analysis reveals functional silencer element (NRSE) although activity not significant. Region from -1 15 to +44 shown to act as basal promoter. Region from -22 to -422 contains cell type specific enhancer. This region contains NRSE and binding site for Krox-24; NRSBF and Krox-24 thought to interact to regulate expression. Transfection analysis: Deletion analysis reveals cell type specific core promoter between -79 and +153. Cell type specific silencers and enhancers also located further upstream. Transfection analysis: I .2kb of 5’ seq insufficient for cell type specific expression. Two upstream regions capable of confeming specificity. Transfection analysis: Deletion analysis shows 212bp promoter sufficient for cell type specific expression. Region between -1 87 and -2 I2 essential. AP- I/E box dyad motif in this region shown to act as essential enhancer in PC8b cells. Same region showed to be both insufficient and nonessential for expression in PC12 cells, instead functional enhancer lies about 500bp upstream of transcriptional start site.

Transgenic analysis: 4kb 5’ seq. drives correctly regulated reporter expression. 300bp 5’seq. drives expression in many nonneuronal tissues. Transfection analysis: Deletion analysis shows 225bp promoter and 105hp leader sufficient for cell type specific expression. CRE element found at -155.

TABLE IV Genes (Continued)


Camhi et al., 1989 Fung et al., 1992 Yoon and Chikaraishi, Wong et al.. 1994 Gandleman et al., I990

Bargou and Leube. 1991

Chin et al., 1994

Hoesche et al., I993 L Li et al., 1993 G Thiel et a1.,1991 G Thiel et al., 1994

1991

Saurwald et al., 1990 Howland et al.,

Wuenschell et al., 1990

References

=: U

?

m

0

Z

>

z

I p

-e