WO1999004025A1

WO1999004025A1 - Improvements in or relating to expression of nucleic acid sequences in cerebellar cells

Info

Publication number: WO1999004025A1
Application number: PCT/GB1998/002137
Authority: WO
Inventors: Alison Mary Jones; William Wisden; Sabine Bahn
Original assignee: Medical Research Council
Priority date: 1997-07-19
Filing date: 1998-07-17
Publication date: 1999-01-28
Also published as: AU8450898A; GB9715164D0

Abstract

Disclosed is a nucleic acid fragment which, when operably linked to a downstream nucleic acid sequence of interest, causes selective expression in cerebellar granule cells of the downstream sequence, the nucleic acid fragment comprising an isolated 5' portion of the η-aminobutyric acid type A (GABAA) receptor α6 subunit gene, the isolated portion comprising: the first seven exons, and a 5' part of exon 8 of the α6 subunit gene, the intervening introns, and at least 500 bp upstream of the transcription start site region. Also disclosed are constructs and host cells comprising the fragment, and methods of expressing nucleic acid sequences in cerebellar granule cells.

Description

Title: Improvements in or Relating to Expression of Nucleic Acid Sequences in Cerebellar Cells

Field of the Invention

This invention relates to nucleic acid sequences and constructs, host cells and organisms comprising the same, and to a method of directing expression of a polypeptide of interest to a particular cell type.

Background of the Invention

In the present specification the following abbreviations are employed:

GAB A (7-aminobutyric acid); IRES (internal ribosome entry site); lacZ (/3-galactosidase) ;

NL (nuclear localization signal); X-gal (5-bromo-4chloro-3-indoyl-, /3-D-galactoside).

γ-aminobutyric acid type A (GABA_A) receptors are transmitter-gated chloride channels mediating fast neuronal inhibition. They are built as pentameric subunit assemblies selected from a large gene family (αl-αό, 01-/33, γl-γ3, δ and e) (Refs. 1-6 [all documents cited in this specification are incorporated herein by reference]). The subunit genes are transcribed in complex patterns throughout the brain (7, 8), but nothing is known about how these patterns are generated. Many of the subunit genes are clustered e.g. the c , β , γ2 and cώ genes are on mouse chromosome 11/ human chromosome 5q

(9, 10), which might imply co-ordinated regulation. As an entry point for analysis, we chose to study the subunit gene (α6) which has the simplest expression pattern. When the whole brain is surveyed, the α6 gene is found to be transcribed in just two cell types: cerebellar granule cells and the lineage-related cochlear nucleus granule cells (11-15).

Within this lineage, the α6 locus is active only in postmigratory and differentiated cells, i.e. expression first begins to appear at the beginning of the second postnatal week (5. 14,

16-19). With one exception (20), it has not proved possible to isolate regulatory DNA fragments which direct expression to just one type of neuronal cell. Most brain- specific transgenes drive wide expression (e.g. 21,22). Here, we have identified an α6-based transgene capable of directing high-level expression uniquely to adult cerebellar granule cells. Summary of the Invention

In a first aspect the invention provides a nucleic acid fragment which, when operably linked to a downstream nucleic acid sequence of interest, causes selective expression in cerebellar granule cells of the downstream sequence, the nucleic acid fragment comprising an isolated 5' portion of the γ-aminobutyric acid type A (GABA receptor aβ subumt gene, the isolated portion comprising: the first seven exons, and a 5' part of exon 8 of the a^β subunit gene, the intervening introns, and at least 500bp upstream of the transcription start site region.

The person skilled in the art will appreciate that the sequence of the mtervering introns may not be essential and the intron sequences may be substituted with functionally equivalent sequences of similar size (and which are recognised as introns) without affecting the functioning of the nucleic acid fragment of the invention.

The aβ subunit gene may have the gene sequence of the mouse or rat aβ subunit gene (as exemplified below). Alternatively, the aβ subunit gene being highly conserved (and showing similar cerebellar granule-cell specific expression) across a wide range of mammals (e.g. humans, mouse, rat) and other animals (e.g. goldfish) and birds (e.g. chicken, Refs. 29, 30), the portions from aβ subunit genes from other sources, being functionally equivalent to the portions of the rat/mouse genes identified above, may successfully be employed, if desired. For example, the homologous fragment of the human aβ subunit gene is considered to be the portion stretching from approximately l.όkb upstream (5') of the ATG start codon through to about the middle of exon 8. Functionally equivalent portions of genes from other sources can readily be located by those skilled in the art, for example by conducting sequence comparisons using commercially available computer programs such as Mac Vector (available from Kodak IBI) or "Blast" (available from European Bio informatics Institute, Hinxton Hall, UK).

In a particular embodiment the isolated 5' portion of the aβ subunit gene is a sequence from the mouse gene having a 5' end at or adjacent to an Sphl restriction site at nucleotide -1610 upstream from the ATG start codon on the gene, and a 3' end at or adjacent to a unique Aflϊl site in exon 8 (this includes about 200bp, or approx. 50%, of exon 8). In a second aspect the invention provides a nucleic acid construct which directs the specific expression in cerebellar granule cells of a nucleic acid sequence of interest, the construct comprising an isolated 5' portion of the GABA_A receptor aβ subunit gene operably linked to a downstream nucleic acid sequence of interest; wherein the 5' portion comprises the first seven exons and a 5' part of exon 8 of the aβ subunit gene, the intervening introns, and at least 500bp upstream of d e transcription start site region.

In the various aspects of the invention, the term "expression" is intended to refer to transcription, with optional translation, of the nucleic acid sequence of interest. Thus, for example, it may simply be desired to produce an RNA transcript of the nucleic acid sequence of interest (in particular, where the sequence of interest is operably linked in the antisense orientation to the cerebellar granule cell-specific promoter, so as to inhibit expression of one or more polypeptides in the cerebellum). Alternatively, and more preferably, the nucleic acid sequence of interest includes both transcription and translation thereof, and the polypeptide of interest is synthesised in cerebellar granule cells.

In one embodiment the sequence of interest is expressed as an in- frame fusion with the 5' portion of the aβ subunit gene. Alternatively, and preferably, transcription of the nucleic acid sequence encoding the polypeptide of interest commences separately at an internal ribosome entry site (IRES). Accordingly, those skilled in the an will appreciate that "operable linkage" of the 5' portion of the aβ subunit gene and the sequence of interest does not necessarily require expression thereof as a fusion protein.

The 5' portion of the aβ subunit gene may comprise additional base pairs upstream of the aβ transcription start region. Up to 1Kb has been successfully included, and further 5' extension could be made if desired, although this has not so far been found to have any advantageous effect on the working of the invention. The nuleic acid fragment of the inevntion may comprise just a single base from exon 8, the intention being that substantially all of intron 7 is included in the fragment. Alternatively, the a^β subunit portion may be extended in the 3' direction to include for example all of exon 8 and, if desired, some or all of the subsequent intron. Again, however, it is not anticipated that this will have any significant beneficial effect on the operation of the invention. The isolated 5' portion of the aβ subunit, whilst possibly including regions of the aβ gene in addition to those found important by the present inventors, will not comprise the entire a^β gene sequence as this is not expected to confer cerebellar granule cell-specific expression on the sequence encoding the polypeptide of interest.

Sequences other than an IRES may be included between the 5' portion of the aβ subunit gene and the sequence encoding the polypeptide of interest without disrupting the operable linkage therebetween. Typically such intervening sequences may be regulatory sequences (such as nuclear localization signals) or may be sequences to facilitate manipulation (e.g. restriction endonuclease polylinkers).

In the examples below, the invention is illustrated by the use of a reporter gene sequence, encoding a reporter polypeptide (/3-galactosidase) . In practice, almost any nucleic acid sequence could be used in the construct of the invention. Examples of genes of interest which could be employed in the construct include Cre recombinase (38), to produce cerebellar granule cell-specific "knockout" transgenic animals, or A. victoria green fluorescent protein (GFP) for imaging of living cerebellar granule cells (see e.g. Chalfie et al, 1994 Science 263, 802-805; Yang et al, 1996 Gene 173, 19-23).

In particular, the construct would be of usefulness in gene therapy of brain disorders, such as those diseases which are associated with death of cerebellar granule cells, leading to ataxia (e.g. SCAs or "spinocerebellar ataxias"). Potentially therapeutic genes include those encoding: growth factors e.g. brain-derived neuro trophic factor, or neurotrophin-3 (Neveu & Arenas 1996 J. Cell. Biol. 133, 631-646); Ca²⁺ binding proteins e.g. Calbindin D28 (Airaksinen et al, 1997 Proc. Natl. Acad. Sci. USA 94, 1488-1493); protease inhibitors; and membrane channel-forming proteins e.g. Ca²~ channel β subunit gene Cchb4 (Burgess et al, 1997 Cell 88, 385-392).

In a further aspect, the invention provides a host cell transformed with the construct defined above, or the progeny of such a cell. The term "transformed" as employed herein is intended to refer to any process by which foreign DNA may be introduced into a cell and includes, for example, electroporation, transduction, cell fusion and the like. The host cell may be a prokaryotic cell, into which the construct is introduced for the purposes of replication (in which case, the construct will preferably comprise a prokaryotic origin of replication). More typically the cell is a eukaryotic cell, conveniently a mammalian cell. If it is desired to express the sequence of interest, the construct may be introduced into a cerebellar granule cell, which cells may be cultured in vitro (Gallo et al, 1987 J. Neurosci. 7, 2203-2213) and transfected with adenoviral vectors (Fritz et al, 1997 J. Neurochem. 68, 204-212). Alternatively (e.g. where the sequence of interest may encode a toxic polypeptide) it may be desirable to maintain the construct in a cell or cell line in which it will not be expressed.

In another aspect the invention provides a transgenic animal (conveniently a transgenic mammal, preferably a rat or mouse) comprising the nucleic acid construct defined above. Such transgenic animals will allow for the cerebellar granule cell-specific expression of nucleic acid sequences encoding polypeptides of interest, and may therefore act as valuable research tools.

In a further aspect the invention provides a method of obtaining cerebellar granule cell- specific expression of a nucleic acid sequence of interest, the method comprising: placing the nucleic acid sequence in operable linkage to, and downstream of, a nucleic acid fragment in accordance with the first aspect of the invention defined above, to form a nucleic acid construct; and introducing the construct into a host cell.

The host cell may be a cerebellar granule cell cultured in vitro, in order to obtain expression of the nucleic acid sequence.

Alternatively, the construct is introduced into a cell from which a transgenic animal may be formed. For example, the construct may be introduced by pronuclear micro injection into a mouse or rat embryonic stem cell or the like. Methods of producing transgenic animals are now well known to those skilled in the art. A general reference on the subject is by Hogan et al (1994, Manipulating the mouse embryo: a laboratory manual. Cold Spring Harbor Laboratory Press, New York). The mαόlRESlacZ transgene can be modified to deliver products other than β- galactosidase to cerebellar granule cells. Specifically, lacZ could be removed, and alternative open reading frames placed downstream of the IRES. This could include Cre recombinase or proteins to correct for diseases such as spinocerebellar ataxia degenerations (SCAs) (e.g. Zhuchenko et al, 1997, Nature Genetics 15, 62-69), or other types of degeneration which afflict cerebellar granule cells (e.g. Burgess et al, Cell 88, 385-392). The a^β gene 5' portion could also be incorporated into viral vectors such as adenovirus (Hashimoto et al, 1996 Human Gene Therapy 7, 149-158).

Thus, in an alternative embodiment, the method of the invention may be used in gene therapy, wherein the nucleic acid sequence encodes a therapeutic polypeptide, such as a nerve growth factor or a nucleic acid sequence which correctly codes for a polypeptide so as to replace a missing or "faulty" (i.e. mutated) gene in the patient (e.g. human homologues of the mouse genes described by Bugess et al cited above). The nucleic construct could be introduced into a human patient within a vector, such as a retro virus or adenovirus, or encapsulated or associated with liposomes or as naked DNA. The presence of the 5' portion of the aβ subumt gene serves to ensure that the therapeutic polypeptide is expressed only in cerebellar granule cells, thereby preventing inappropriate expression should the construct be inadvertently introduced into other cells, which inadvertent expression could otherwise lead to undesirable side effects.

Methods of introducing nucleic acid sequences into human subjects for the purposes of gene therapy are now well known, e.g. adenoviral vectors for infecting the airways of cystic fibrosis sufferers. In principle, herpes simplex virus-based or adenovirus-based vectors (e.g. those described by Bett et al, 1994 Proc. Natl. Acad. Sci. USA 91. 8802- 8806 and those described by Le Gal La Salle et al, 1993 Science 259, 988-990) could be used to introduce the construct of the invention into human subjects, and such techniques have already been applied to experiments with rodents.

The invention will now be further described by way of illustrative example and with reference to the accompanying figures, in which: Figure 1 shows a schematic representation of various reporter constructs used (on the left hand side), together with columns showing the number of transgenic founders obtained and the number of founders showing reporter gene expression in (from left to right) cerebellar granule cells, cochlear nucleus, or ectopic expression. In the reporter gene constructs, A+ is the SV40 polyadenylation signal; Ex denotes exon; NL is the SV40 nuclear localization sequence; B denotes a Bam HI site; S denotes an Sph 1 site; the arrow indicates the transcriptional start site(s). The open box at the 3' end of the mα6IRES-LacZ6 transgene marks the residual 5' fragment of the neomycin (neo) resistance gene (see 25); (* 3 independent lines were derived, all of which expressed the transgene);

Figures 2A-2F are micrographs showing ectopic (2A-2C), granule cell-specific mosaic (2D-2E), and cochlear nucleus (2F) expression of the rα6LacZ7 transgene. In particular: m6LacZ7 expression in: 2A, cerebellar Purkinje cells (granule cells are non-expressing); 2B, medial habenulae, 2C, olfactory bulb mitral cells; 2D, mosaic rα6LacZ0.9 expression in cerebellar granule cells; 2E, mosaic rα6LacZ7 expression in cerebellar granule cells; 2F, rα6LacZ7 in dorsal cochlear nucleus granule cells. Sections are stained for, /3-galactosidase activity (blue) and counterstained with neutral red; Cb gr = cerebellar granule cells; CN = cochlear nucleus; mol = cerebellar molecular layer; Olf, MC = olfactory bulb mitral cells; mHb = medial habenulae; Scale bars: 2A-C, 120 μm; 2D, 170 μm: IE, 60 μm; 2F, 120 μm.

Figures 3A-3G are micrographs demonstrating that Expression of the mα:6IRES-lacZ6 transgene is restricted to cerebellar granule cells. Figures 3A-C are horizontal sections of three representative independent adult founder brains from the highest expressing (3A) to the lowest expressing (3C) mcu6IRES-lacZ6 integration. Figure 3D is a higher power view of a cerebellar section from the founder shown in 3A; 3E shows that no cochlear nucleus expression is observed even though very strong expression is found in the granule cell layer; Figures 3F-G are high power views of the granule cell layer of an expressing founder from different subregions of the cerebellum to illustrate mosaicism. In Figure 3F, all cells are heavily stained. In Figure 3G, some granule cells are not expressing. In both cases, the molecular layer is heavily stained with X-gal product because of /3-galactosidase in the granule cell axons, the parallel fibres. Figures 3A-D are stained only for /3-galactosidase activity; Figures 3E-G are counterstained with neutral red; (Abbreviations: Cb = cerebellum; CN = dorsal cochlear nucleus; CP = caudate-putamen; ctx = neocortex; gr = cerebellar granule cells; H = hippocampus; IC = inferior colliculus; mol = cerebellar molecular layer; PC = Purkinje cells; T = thalamus; WM = white matter tracts). Scale bars: Figures 3A-C, 1 mm; 3D, 0.5 mm; 3E, 170 μm; 3F-G, 30 μm.

Examples

Example 1

Transgene constructions

Plasmids prα6nLacZ7 and prα;6nLacZ12 were built with the vector prα6nLacZ0.9 (23). This vector contains 500 bp of 5' nontranscribed region, the proximal promoter including the transcription start site(s), and the complete 5' untranslated segment (approx. 350 bp). The first methionine of the aβ gene is replaced with the lacZ coding region incorporating an SV40-derived nuclear localization (NL) sequence (24). The pra6nLacZ0.9 plasmid was first modified to give prα6nLacZ0.9ΔNot 7zo, by inserting unique Not / and Xho I sites into the 5' and 3' polylinkers respectively. Two aβ gene restriction fragments, spanning upstream portions of the aβ gene, were isolated form a λ Dash II Sprague-Dawley rat testis genomic library (Stratagene) (23). These 6.5 kb Not 1/BamH I and 12 kb Not I/BamHl fragments were placed into the 5' poly linker of pro:6nLacZ0.9ΔΝotXho to create pr 6nLacZ7 and prα6nLacZ12 respectively (Figure 1). In each case, the 5' Not I site derives from the λ Dash II polylinker, with the 12 kb fragment originating from a phage with more 5' aβ gene fragments. For pronuclear injections, the plasmid backbone of both constructs was removed by Not I and Xho I digestion.

pm 6IRES-lacZ6: This transgene was isolated from homozygous ΔαόlacZ 129/Sv x C57BL/6 mouse liver genomic DNA (25). The transgene comprises 6 kb of mouse GABA_A receptor 0:6 subunit gene containing 1 kb of sequence upstream of the transcription start sites, exons 1 through to 8 (up to the AflϊL site contained in exon 8), linked to an internal robosome entry site and lacZ reporter/poly adenylation sequences. To obtain the transgene from the mouse, ΔαόlacZ liver genomic DNA was partially Sau 3 A digested, ligated into a λ Fix II Xho Partial Fill-In vector (Stratagene), packaged and amplified. The resulting library was screened with an aβ cDNA probe (12). A phage containing the promoter region through to the exon 8-internal ribosome entry site (IRES)-lacZ and 5' end of the neo gene, was digested with Sph I to give a lacZ-positive 12 kb fragment. This was subcloned into pUC BM20 (Boehringer Marmheim) to give pmc lRES-lacZό (Figure 1). The 5' Sph I site is 1 kb 5' of the transcription start site(s); the 3' end of the transgene contains a herpes simplex virus thymidine kinase promoter linked to the region encoding the first 180 amino acids of the neomycin phosphotransferase protein. For pronuclear injections, the insert was released using Sph I.

All constructs were verified by sequencing across fragment boundaries.

Although the present inventors obtained the moTRES-lacZό construct from transgenic mice, the construct could readily be prepared by those skilled in the art by following the instructions below: screen a genomic λ phage or cosmid-based library (constructed with mouse strain 129/Sv DNA) using an aβ cDNA probe. Isolate the unique 12 kb Sphl restriction fragment spanning lkb upstream of the transcription start sites through to exon

9. The 5' half of this fragment forms the basis of the transgene. Subclone this 12 kb Sph

I fragment into any standard pUC based cloning plasmid, e.g. by first changing the Smal site in the pBluescript SK (Stratagene) poly linker to Sphl. An IRES-lacZ-neo cassette is then introduced into exon 8 within the 12 kb Sphl fragment as follows: Linerize the plasmid with Aflll (this site is unique, and located in exon 8). Fill in the Afllϊ ends with

Klenow DNA polymerase, and ligate on blunt-end Smal-Bam HI adaptors [adaptors made by annealing d(GATCCCCGGG) and d(CCCGGG) oligonucleotides] to give mα6Sphl2ΔBam. The IRES-lacZ-neo cassette is publicly available from Nehls et al, 1996

(Science 272, 886-889) or Jones et al, 1997 (J. Neurosci. 17, 1350-1362); it is a 5 kb

S HI fragment inserted into pBluescript. The purified 5 kb insert is ligated into the linear Sphl2ΔBam backbone. The correct orientation is selected by sequencing ligation products with the mouse α6-specific sense primer 5'-GAATCACCACTGTTTTA-3' (Seq.

ID No. 1), located in exon 8 just upstream of the modified Aflll site. On a plasmid with the correct orientation of IRES insert, partially digest with Sphl, so that two of the three Sphl sites within the plasmid are cleaved, and religate the backbone (Inserting the IRES- lacZ-neo cassette introduces an additional Sphl site - located in the neomycin resistance gene. Correct partial digests cut only the neo Sphl site, and the site in the 3' polylinker. Religating such a partially digested backbone will delete the 6kb aβ gene fragment containing intron 8 and exon 9). Confirm by sequencing: The 5' end of the appropriate fragment will have the following sequence -

5 'GCATGCTAGGCAAGGATTAACTGAGATATCTTCACTGCTTTTGTTTTTTTG-3 ' (Seq. ID No. 2)

The Sphl restriction enzyme site is underlined. The 3' end is defined by the unique Aflll site in exon 8. The transgene mαόlRESlacZ can be released from the plasmid backbone by Sphl digestion.

Example 2

Transgenic mouse production and analysis

Transgenic mice (CBA/cba x C57BL/6) were produced by pro nuclear micro injection (26). Founders were identified by blotting Bam Hi-digested tail-derived genomic DNA, and hybridizing with a lacZ probe. Anaesthetised adult mice were transcardially perfused with 4% paraformaldehyde. Sections from brains and selected organs (liver, kidney, heart) were then incubated with Xgal (5-bromo-4-chloro-3-indoyl-β-D-galactoside) (27). Low-power images were obtained by photographing wet, non-coverslipped sections using a Leica Wild Heerbrugg microscope. Selected sections were counterstained with neutral red (Sigma), coverslipped and photographed with a Leica Orthomat E microscope.

Results and Discussion

We searched for aβ gene regions which confer cerebellar granule cell-specific expression. Our initial analysis showed that the proximal 500 bp promoter fragment (rα61acz0.9) has neuronal specificity, and can occasionally direct weak variegated expression to cerebellar granule cells in certain lobules (Figure 1; Fig. 2D) (23). More usually, integration events did not generate any expression (23). Here, we have extended this analysis using lacZ reporter transgenes containing more upstream regions of the rodent aβ subunit gene. The rα6nlacZ7 transgene contains 7 kb upstream of the start site(s) (Figure 1). 7/15 independent transgene integrations gave expression (Figure 1), all of which was neuron-specific, but ectopic. Differing genomic integration positions generated, for example, Purkinje cell (Figure 2A), medial habenula (Figure 2B) and mitral cell (Figure 2C) expression. Two founders with rα6nlacZ7 integrations had cerebellar granule cell-specific expression (Figure 2E). As for the rα61acZ0.9 transgene (Figure 2D), cerebellar granule cell expression obtained with rα6nlacZ7 was markedly mosaic; the majority of granule cells did not express the gene (Figure 2E). However, the expressing cell minority had high lacZ activity (Figure 2E). Additionally, the two r 6nlacZ7 founders with cerebellar granule cell expression had strong lacZ expression in the dorsal cochlear nucleus granule cells (Figure 2F). Overall, granule cell specific expression from the rα6nlacZ7 transgene did not significantly improve on that obtained with rα61acZ0.9, except that the frequency of expression was roughly twofold higher.

Expression was even less from a transgene incorporating a 12 kb upstream fragment (Figure and data not shown). With this construct, only weak ectopic expression was detected in dispersed midbrain cells. It seems that the 500 bp proximal promoter fragment is, to a certain extent, primed to give granule cell-specific expression, but that the chromosomal integration position usually dominates. Adding on more 5^' DNA does not significantly improve expression. Even when expressed, all these constructs showed strong position effect variegation (28).

We therefore tested the role of sequences downstream of the promoter in regulating the a6 subunit gene expression. Previously, a lacZ "knock-in" mouse line (ΔαόlacZ) had been generated via homologous recombination at the aβ gene (25). In the ΔαόlacZ line, an IRES-lacZ cassette insertion in exon 8 gave high levels of lacZ activity in cerebellar and cochlear nucleus granule cells (25). This mouse line contains "ready-made" transgenes embedded in the aβ gene locus. In particular, a 12 kb Sph I genomic fragment (mα6IRES-lacZ6) contains 1 kb upstream of the transcription start site(s), 5 kb of exon and intronic regions through to exon 8, followed by an IRES-lacZ poly A sequence (25; and see Table 1). We therefore cloned this fragment out from a ΔαόlacZ mouse liver genomic library, as described above.

Expression from the microinjected mo:6IRES-lacZ6 transgene closely resembles that of the native 6 gene (Figure 1 and Figure 3). Cerebellar granule cell-restricted lacZ expression was seen in every expressing transgenic founder, and the number of expressing transgenes was high (occurring in 6/10 independent integrations - see Figure 1). In adult mice, depending on the genomic integration position, mαόlRES-lacZό transgene expression ranged from very strong (Figure 3A), strong (Figure 3B) to moderate (Figure 3C). No ectopic expression in other parts of the brain (Figure 3 A to C), white matter (Figure 3D) or in heart, liver and kidney (data not shown) was found. In some cases, expression was so strong that lacZ activity could be detected in the molecular layer corresponding to enzyme transported into the granule cell axons - the parallel fibres (Figure 3F & G). In these cases, sections required only a short incubation in X-gal reagent to fully develop the blue reaction product. Some variegation in expression, depending on the exact cerebellar lobule was seen. This ranged from virtually all granule cells expressing (Figure 3F) to a minority not expressing (Figure 3G) the transgene. In the strongest expressing founders, variegation was at a minimum. Three of the founders were bred as heterozygotic lines. LacZ expression was stably inherited and correctly developmentally regulated, with transgene expression only appearing in postmigratory granule cells after the first postnatal week (data not shown).

So far, no cochlear nucleus granule cell lacZ expression has been observed in any mα6IRES-lacZ6 expressing brain (Figure 3E, CN), even though cerebellar granule cell lacZ expression is prominent in the same sections (Figure 3E, gr). This might imply the existence of discrete cochlear nucleus- and cerebellar granule cell-specific regulatory elements. Such a cochlear nucleus-specific element could reside between -7 and - 1 kb of the aβ gene, since the ra6nlacZ7 transgene can direct cochlear nucleus granule cell expression. We did not test the contribution of more 3' elements (the 7 kb intron 8, and fragments 3' to exon 9). Clearly, information that directs gene expression specifically to adult cerebellar granule cells is contained in the 5' half (6 kb) of the aβ subunit gene. Expression from mαόlRES-lacZό transgene is robust; no ectopic expression has so far been seen with this construct, and it gives a high expression frequency i.e. it is apparently relatively insensitive to genomic integration position. Therefore, aβ gene regulation can be uncoupled from the αl-jS2-γ2 subunit gene cluster.

This could be because of powerful granule cell-specific enhancer elements. The proximal 500 bp promoter contains information for neuronal specificity, and seems primed to drive expression in granule cells, but with low efficiency (23). It might interact with a downstream element which enforces transcriptional specificity. Alternatively, an optimum gene architecture could account for the high fidelity of transgene expression in granule cells e.g. the exon- intron structure may be particularly suited to granule cell expression.

Throughout the cerebellum's evolutionary history, aβ subunit gene transcription has remained restricted to granule cells (29-30). Presumably, DNA regulatory elements have also been conserved. Comparative sequencing across the 5' half of the gene might help to identify these putative regulatory regions (31). Eventually, by combining this approach with transgene deletions and point mutations, we will be able to identify the transcription factors which act on this gene. Similar factors may be involved in regulating other GABA_A receptor subunit genes (32), or more generally in defining the final stages of neuronal differentiation (33).

The cerebellum participates in the acquisition and deployment of motor skills (34-35), and in higher cognitive aspects of learning and anticipating patterns (36-37). The mα6IRES-lacZ6 transgene described here can be adapted to deliver other gene products uniquely to cerebellar granule cells. By coupling this with recent developments in transgenic technology (22, 38), the specific contributions of granule cell components to the physiology of cerebellar processes can now be investigated. References

1. Seeburg et al (1990) Cold Spring Harb. Symp. Quant. Biol. 55, 29-40.

2. Tyndale et al (1995) in Handbook of Receptors and Channels: Ligand- and Voltage-Gated Ion Channels, ed. North, R. A. (CRC Press, Inc. , Boca Raton), pp.265-290.

3. Costa & Guidotti (1996) Trends Pharmacol. Sci. 17, 192-200.

4. McKernan & Whiting (1996) Trends Neurosci. 19, 139-143.

5. Wisden et al (1996) Neuropharmacology 35, 1139-1160.

6. Davies et al (1997) Nature 385,820-823.

7. Wisden et al (1992) J. Neurosci. 12, 1040-1062.

8. Fritschy (1995) J. Comp. Neurol. 359, 154-194.

9. Garrett et al (1997) Mol. Brain Res. 45, 133-137.

10. Hicks et al (1993) Genomics 20, 285-288.

11. Kato (1990) J. Mol. Biol. 214, 619-624.

12. Luddens et al (1990) Nature 346, 648651.

13. Laurie et al (1992) J. Neurosci. 12, 1063-1076.

14. Varecka et al (1994) J. Comp. Neurol. 339, 341-352.

15. Nusser et al (1996) J. Neurosci. 16, 103-114.

16. Korpi et al (1993) Neuroscience 53, 483-488.

17. Kuhar et al (993) Development 117, 97-104.

18. Laurie et al (1992) J. Neurosci. 12, 4151-4172.

19. Zheng et al (1993) Dev. Brain Res. 75, 91-103.

20. Oberdick et al (1990) Science 248, 223-226.

21. Andra et al (1996) Neurobiol. Aging 17, 183-190.

22. Mayford et al (1996) Science 274, 1678-1683.

23. Jones et al (1996) J. Neurochem. 67, 907-916.

24. Mercer et al (1991) Neuron 7, 703-716.

25. Jones et al (1997) J. Neurosci. 17, 1350-1362.

26. Hogan et al (1994) Manipulating the mouse embryo: a laboratory manual. (Cold Spring Harbor Laboratory Press).

27. Bonnerot & Nicolas (1993) Methods Enzymol. 225, 451-469. 28. Dobie et al (1997) Trends Genetics 13, 127-130.

29. Bahn et al (1996) J. Neurochem. 66, 1810-1818.

30. Hadingham et al (1996) Mol. Pharmacol. 49, 253-259.

31. Aparicio et al (1995) Proc. Natl. Acad.Sci. USA 92, 1684- 1688.

32. Motejlek et al (1994) J. Biol. Chem. 269, 15265-15273.

33. Hatten et al (1997) Curr. Opin. Neurobiol. 7, 40-47.

34. Kim & Thompson (1997) Trends Neurosci. 20, 177-181.

35. Raymond et al (1996) Science 272, 1126-1131.

36. Allen et al (1997) Science 275, 1940-1943.

37. Fiez (1996) Neuron 16, 13-15.

38. Tsien et al (1996) Cell 87, 1317-1326.

Claims

1. A nucleic acid fragment which, when operably linked to a downstream nucleic acid sequence of interest, causes selective expression in cerebellar granule cells of the downstream sequence, the nucleic acid fragment comprising an isolated 5' portion of the γ-aminobutyric acid type A (GABA_A) receptor aβ subunit gene, the isolated portion comprising: the first seven exons, and a 5' part of exon 8 of the aβ subunit gene, the intervening introns, and at least 500bp upstream of the transcription start site region.

2. A fragment according to claim 1, comprising an isolated 5^' portion of a mammalian aβ subunit gene.

3. A fragment according to claim 1 or 2, comprising an isolated 5' portion from the mouse aβ subunit gene.

4. A fragment according to any one of claims 1 , 2 or 3, comprising an isolated portion from the mouse aβ subunit gene, which portion has has a 5' end about nucleotide -1610 upstream from the ATG start codon, and a 3' end at or adjacent to a unique Aflll site in exon 8.

5. A fragment according to any one of the preceding claims, wherein the nucleic acid sequence of interest encodes a polypeptide of interest.

6. A fragment according to any one of claims 1-4, wherein the nucleic acid sequence of interest is operably linked in the antisense orientation to the isolated 5' portion of the o;6 subunit gene.

7. A nucleic acid construct which directs the specific expression in cerebellar granule cells of a nucleic acid sequence of interest, the construct comprising an isolated 5' portion of the GABA_A receptor aβ subunit gene operably linked to a downstream nucleic acid sequence of interest; wherein the isolated 5' portion comprises the first seven exons and a 5' part of exon 8 of the aβ subunit gene, the intervening introns, and at least 500bp upstream of the transcription start site region.

8. A nucleic acid construct according to claim 7, and further in accordance with any one of claims 2-6.

9. A nucleic acid construct according to claim 7 or 8, comprising one or more of the following: a restriction endonuclease polylinker; a nuclear localization signal; an IRES.

10. A host cell transformed with a construct according to any one of claims 7, 8 or 9, or the progeny of such a cell.

11. A mammalian cell or cell line according to claim 10.

12. A transgenic animal grown from a cell transformed with a construct according to any one of claims 7, 8 or 9.

13. A method of obtaining cerebellar granule cell-specific expression of a nucleic acid sequence of interest, the method comprising the steps of:

(a) placing the nucleic acid sequence of interest in operable linkage to, and downstream from, an isolated 5' portion of the γ-aminobutyric acid type A (GABA_A) receptor a^β subunit gene, the isolated portion comprising the first seven exons, and a 5' part of exon 8 of the 0:6 subunit gene, the intervening introns. and at least 500bp upstream of the transcription start site region; so as to form a nucleic construct: and

(b) introducing the construct into a host cell.

14. A method according to claim 13, comprising the use of a construct in accordance with any one of claims 7, 8 or 9.

15. A method of inhibiting the expression of a gene naturally expressed in a cerebellar granule cell, the method comprising the steps of:

(a) introducing into a cerebellar cell or precursor thereof a nucleic acid construct according to claim 7, wherein the nucleic acid sequence of interest comprises a sequence homologous to the gene whose expression is to be inhibited, but linked in the antisense orientation; and (b) causing transcription of the sequence of interest.

16. A method of synthesising a polypeptide of interest in a cerebellar granule cell, the method comprising the steps of:

(a) introducing into a cerebellar cell or precursor thereof a nucleic acid construct according to claim 7, wherein the nucleic acid sequence encodes the polypeptide of interest; and

(b) causing transcription and translation of the nucleic acid sequence encoding the polypeptide of interest.