THE JOURNAL OF COMPARATIVE NEUROLOGY 326~580-594 (1992)

Differential Expression of GABAA/Benzodiazepine Receptor PI, p2, and P 3 Subunit mRNAs in the Developing Mouse Cerebellum DARK0 ZDILAR, VERA LUNTZ-LEYBMAN, ADRIENNE FROSTHOLM, AND ANDREJ ROTTER Department of Pharmacology and the Neuroscience Program, The Ohio State University, Columbus, Ohio 43210

ABSTRACT Gamma aminobutyric acid (GABA) is the major inhibitory neurotransmitter in the mammalian cerebellum. Cerebellar granule, Purkinje, and deep nuclear neurons are known to receive GABAergic afferents. Since GABA exerts its inhibitory effects via GABA receptors, it is of interest to determine the temporal relationship between the formation of GABAergic synapses and the expression of genes coding for the GABA receptor. In a previous study, we have examined the developmental expression of binding sites for [3Hlmuscimol, which binds with high affinity to the p subunits of the GABAAIbenzodiazepine (GABAA/BZ)receptor. In the present study, [35SlcRNAprobes were used to examine the appearance and distribution of GABA*iBZ pl, p2,and p3 subunit mRNAs in the developing C57BLi6 mouse cerebellum by in situ hybridization. In the adult cerebellum, the distribution of the three subunit mRNAs was clearly different, despite considerable overlap, and their temporal expression differed throughout postnatal development. The p1 hybridization signal appeared within the cerebellar cortex during the second postnatal week as a discrete band at the interface of the molecular and granule cell layers. Grains were distributed diffusely over small densely staining cells surrounding the Purkinje cells; relatively few grains were visible over Purkinje cell bodies themselves. This distribution may reflect an association with Bergmann glia or basket cells. The p2and p3hybridization signals were present considerably earlier than that of the p1mRNA. The p2 signal was present at birth in the moleculariPurkinje cell layer; as development progressed, the signal became increasingly intense over both granule and Purkinje cells. At birth, the p3 subunit mRNA was present in the external germinal and molecular layers, later becoming largely localized within the granule cell layer. Dense pz and p3 cRNA probe labeling was present over the adult granule cell layer. Moderate levels of p2 signal were seen over Purkinje cell bodies; considerably less labeling was observed with the p3 probe. The adult distribution of pz and p3 cRNA probes showed good spatial correspondence with the known GABAAreceptor p subunit markers, [3Hl-muscimoland the mAb 62-3G1 antibody, each being present within the granule cell layer. Our results indicate that the temporal expression of GABA*/BZ receptor p subunit messages within a given cell type may be independently regulated, and that acquisition of the pz and p3 mRNAs occurs before these cells become integrated into mature synaptic Circuits. D 1992 Wiley-Liss, Inc. Key words: gamma-aminobutyric acid receptor, beta subunit genes, in situ hybridization, ontogeny, Purkinje cells, granule cells, deep cerebellar nuclei

Inhibitory neurotransmission in the mammalian cerebellum is primarily dependent on gamma aminobutyric acid (GABA): its effects are mediated to a large extent bv activation of the GABAAibenzodiazepine(GABiA/BZ)receitor' The GABAAiBZreceptor forms a heterooligomeric "Channel composed Of p, and subunits. With the exception of 6, each of the subunits exists in several

o 1992 WILEY-LISS, INC.

isoforms: ( Y ~ - ~ , and ~ 1 - 3(reviewed by Olsen and Tobin, '90; Seeburg, '90; Burt and Kamatchi, '91). Additional Accepted August 21, 1992, Address reprint requests to Dr. Andrej Rotter, Dept. of Pharmacology and the Neuroscience Program, The Ohio State University, 333 West 10th Ave., Columbus,OH43210.

GABAA RECEPTOR @SUBUNIT mRNAs subunit diversity may be generated by alternative splicing of the p4 and y z mRNAs (Bateson et al., '91; Kofuji et al., '91). Several clinically useful drugs, including the benzodiazepines, barbiturates and neurosteroids, have been shown to modulate GABAergic neurotransmission. It has been demonstrated that benzodiazepine binding sites are located on the a subunits (Sigel et al., '83; Fuchs and Sieghart, '89), while the p subunits contain the binding sites for GABA and for the high affinity GABA receptor ligand, L3H]muscimol (Casalotti et al., '86; Fuchs and Sieghart, '89). The yz subunit has been shown to mediate interactions between GABA and the benzodiazepines (Pritchett et al., '89; Shivers et al., '89); however, the role of the 6 subunit, and the subunit identity of barbiturate and steroid binding sites, has not yet been determined. The considerable pharmacological heterogeneity exhibited by the GABA*/BZ receptor is most likely the result of variations in receptor subunit composition and stoichiometry. A frequently encountered subunit combination in the brain is a1, p2,3, and y2 which gives rise to a receptor with Type I benzodiazepine pharmacology (Benke et al., '91). Four variants of the p subunit have been identified so far, one of them (p4) being present in chicken, but not in rodent brain (Bateson et al., '91). In the rat, the three p subunits are products of three separate genes, rather than alternatively spliced forms (Burt and Kamatchi, '91). The mRNAs coding for the three subunits are not equally abundant: in the rodent brain, p1subunit mRNA is least abundant, while pz is most abundant (Burt and Kamatchi, '91). Although the specific roles of each of the p subunits are unclear, they are generally thought to be required for the efficient assembly of the GABAAiBZ receptor. The inclusion of specific p subunits within the receptor complex effects GABA affinity, with the subunit conferring a higher affinity than pz (Malherbe et al., '90). The p subunits bind ["Hlmuscimol with high affinity (Casalotti et al., '86; Deng et al., '86) and, in addition, the pZl3subunits are recognized by monoclonal antibodies bd-17 (Somogyi et al., '89) and mAb 62-3G1 (de Blas et al., '88; Vitorica et al., '88; Ewert et al., '921, which were generated against the purified GABAAIBZ receptor. Since the formation of functional inhibitory synaptic circuitry within the cerebellum is dependent on the acquisition of postsynaptic GABAA/BZ receptor sites, we have examined the developmental expression of several of the genes coding for the receptor complex in the postnatal murine cerebellum. The cerebellum contains high levels of [3Hlmuscimol binding sites (Palacios et al., '80; Frostholm and Rotter, '87), and bd-17 (Somogyi et al., '89) and mAb 62-3G1 (Vitorica et al., '88; Ewert et al., '92) antibody staining. Recent in situ hybridization experiments have revealed the presence of significant amounts of pi-3 subunit mRNAs within the cerebellum (Gambarana et al., '91; Zhang et al., '91). In previous studies, we found that a1 subunit mRNA and [3H]flunitrazepam binding sites within the cerebellum are largely associated with Purkinje cells (Zdilar et al., '91), and that [3H]muscimol binding sites in the developing mouse cerebellum are localized on granule cells (Frostholm and Rotter, '87). In the present study, we have used in situ hybridization of L3%] cRNA probes to determine the temporal relationship between the formation of GABAergic synapses and the expression of genes coding for the pi, p2, and p3 GABAn receptor subunits in the developing cerebellum.

581

MATERIALS AND METHODS Riboprobe preparation PI and p3 riboprobe preparation from rat cDNAs. Rat cDNAs coding for a full length pi subunit mRNA and a 0.90 kb fragment of p3 subunit mRNA were supplied by Dr. Allan Tobin (UCLA, Los Angeles). The p1 cDNA was processed with EcoRV to generate a 0.94 kb fragment spanning the M3-M4 intracellular domain and the 3' end nontranslated region of the full length pi cDNA. This fragment was ligated "in gel" in EcoRV treated pBluescript I1 SK[+l phagemid vector (Stratagene, La Jolla, CA) as described elsewhere (Kalvakolanu and Livingston 111, '91). P I , pz, and p3riboprobepreparation from mouse cDNAs. pi, p2, and (33 mouse-specific probes were generated by polymerase chain reaction (PCR) as follows: Poly(A)+RNA was isolated from mouse cerebellum by means of the Fast Track mRNA isolation kit (Invitrogen, San Diego, CA). First strand cDNA and PCR reactions were conducted according to the Perkin Elmer Cetus GeneAmp RNA PCR protocol (Perkin Elmer Cetus, Norwalk, CT), briefly described as follows: First strand cDNA reaction: 0.5 pg of poly(A)+ RNA was reverse transcribed a t 42°C for 30 minutes (p3, 40 minutes) in a reaction mixture containing 5 mM MgC12, 50 mM KCl, 10 mM Tris (pH 8.3), 1 mM dNTPs each, 1 U/kl of RNase inhibitor, 2.5 Uip1 cloned Moloney murine leukemia virus reverse transcriptase and 2.5 pM random hexamer, in a final volume of 20 pl. The mixture was heat-inactivated at 99°C for 5 minutes (p3,70°C, 10 minutes) and prepared for PCR. PCR reaction: 1-5 pl of the reverse transcription reaction mixture was amplified for 30 cycles, with a denaturation step at 94°C for 1 minute (first cycle for 3 minutes), an annealing step at 55°C (pi and p3) or 60°C (pz)for 1 minute, and an extension step at 72°C for 1 minute (last cycle for 5 minutes) in a total volume of 100 p1 containing 2 mM MgC12, 50 mM KCl, 10 mM Tris (pH 8.31, 200 mM dNTPs each, 2.5 U AmpliTaq DNA polymerase and 0.2 kM each of upstream and downstream primer. Primers, used for amplification of p1 and Pz probe, spanned a region of the M3-M4 fragment, the most variable portion of these particular GABAA receptor subunits. The sequence of the primers, based on the published rat sequence (Ymer et al., '89), with Hind I11 or Bam HI restriction sites and three additional nucleotides, was as follows: 5'-ATAAAGCTTGGAGCGAGCAAACAAGACCA-3'(upstream p1primer); 5-ATAGGATCCTGGCGCTGTCGTATGAGTAC-3' (downstream pi primer); 5'-ATAAAGCTTGAGAAGATGCGCCTGGATGTC-3' (upstream pz primer); 5'-ATAGGATCCGCACGTCTCCTCAGGCGACTT-3' (downstream pz primer); 5'-ATAAAGCTTCTAGCACCGATGGATGTTCAC-3' (upstream p3 primer); 5'-ATAGGATCCTGCTTCTGTCTCCCATGTACC-3' (downstream p3 primer). Amplified fragments of 202b (pi), 26613 (pz)and 157b (p3) were digested with Hind I11 and Bam HI restriction enzymes and ligated "in gel" in Hind 111-Bam HI treated, dephosphorylated pBluescript I1 SK[+] phagemid vector (Stratagene) as described elsewhere (Kalvakolanu and Livingston 111, '91). The recombinant plasmid was purified, alkali denatured, and sequenced with Taq polymerase (Perkin Elmer Cetus), as described by the manufacturer. Zn vitro transcription. Plasmids were linearized with Hind I11 and BamH I (pi, pz, and p3mouse sequences), and Sal I and Pst I (p3 rat sequence) (Boehringer, Indianapolis,

582 IN) for the subsequent production of antisense and sense RNA probes. The linearized templates were transcribed at 37°C for 60 minutes in a reaction mixture containing 40 mM Tris, 8 mM MgC12,50 mM NaCl, 2 mM spermidine, 10 mM dithiothreitol (DTT), 0.5 units RNase-Block 11,500 pM adenosine-5'-triphosphate (ATP), 500 pM cytidine-5'triphosphate (CTP), 500 pM guanosine-5'-triphosphate (GTP) (Stratagene), 500 ng of linearized template, 25 p,M uridine-5'-c~-[~~Slthiotriphosphate ( [35Sl-UTP)(Amersham, S.A. >SO0 Ci/mmol); the reaction was initiated by the addition of 25 units of T3 RNA polymerase (for antisense RNA), or 25 units of T7 RNA polymerase (for sense RNA) (Stratagene), pH 8.0, to a final volume of 10 p1. After digestion of template with 5 units of DNase (Stratagene) for 30 minutes at 3TC, riboprobe was purified over a Nensorb@ 20 column (NEN, Wilmington, DE). The probes were then dried and subjected to limited alkaline hydrolysis (rat and p3) (200 pl volume, containing 40 mM NaHC03/60 mM Na2C03,pH 10.2) for 65 minutes to generate probes with an average length of 100-150 nucleotides. Finally, the probes were precipitated by adding sequentially 6.6 p13 M sodium acetate, pH 6.0, 1.3 p1 glacial acetic acid, 2 p1 (20 pg) tRNA and 500 pl ethyl alcohol (EtOH). The mixture was then placed on dry ice for 30 minutes, centrifuged (12,000 gi15 minutes), washed in ice cold 70% EtOH and dried. The probes were stored at -20°C in 100 mM DTT. Specific activity of the riboprobes was 1-2 x l o 9 dpm/pg. Northern blots. The number and size of the mRNA species recognized by the PI, pz, and p3 probes was determined by Northern blot hybridization. Poly(A)+RNA was isolated according to the method of Badley et al. ('88). Five pg of poly(A)+RNAwas size-fractioned by horizontal electrophoresis by the method of Fourney et al. ('88). RNA was transfered to Duralon-UVa membranes and UV crosslinked according to the manufacturer's protocol (Stratagene). After prehybridization for 2 hours at 62°C in 5 x saline sodium citrate (SSC), 50% formamide, 0.1% (wiv) N-laurylsarcosine, 0.02% sodium dodecyl sulfate (SDS), and 5% (w/v) blocking reagent (Boehringer), the membrane was hybridized overnight at 62°C in the same solution containing 1 x lo6 cpm/mL [32P]labeledcRNA probe. Probes were labeled as above, but with uridine-5'-triphosphate ([32P]UTP)(Amersham, S.A. > 800 Ci/mmol), and were not hydrolyzed. Membranes were washed twice for 5 minutes in 2~ SSC/O.l% SDS at room temperature R.T., twice for 30 x 2 minutes in 0.1 x SSC/O.l% SDS at 65"C, and exposed to Kodak X-OMAT AR film. In situ hybridization. C57BLi6J mice (JacksonLaboratories, Bar Harbor, ME) were sacrificed between postnatal day (PI1 and P35 (day of birth = Pl), and at P80. Animals were decapitated after inhalation anesthesia (Metofane, Pitman-Moore Inc., NJ). Brains were rapidly removed, frozen on dry ice and stored at - 70°C. Coronal or sagittal sections, 20 pm thick, were thaw-mounted onto 3 x coated (300 bloom gelatin and chrome alum) slides and stored at -70°C. Prior to hybridization, sections were fixed for 30 minutes in 4% paraformaldehyde in 1 x PBS (pH 7.4), and washed (2 x 5 minutes) in PBS. Sections were then acetylated (0.25% acetic anhydride diluted in 0.1 M triethanolamine/0.9% NaCl, pH 8.0) for 10 minutes at room temperature (RT),and dehydrated (1minute each) through 70%, SO%, 90%, and 100%EtOH. Hybridization was conducted as follows: Each slide was covered with 100 pl of hybridization buffer (50% formamide, 4 x SSC, 500 pg/ml salmon sperm DNA, 250 pg/ml

D. ZDILAR ET AL. tRNA, 1 x Denhardt's, 100 mM DTT, and 10% dextran) containing 2.0 x lo7 dpmiml riboprobe, coverslipped with parafilm and incubated for 20 hours at 50°C. The parafilm was then removed in 2 x SSC; sections were washed in 2 x SSC at R.T. for 10 minutes, followed by incubation in RNase A solution (20 pgiml RNase A in 10 mM Tris-HC1, 500 mM NaCl, 1 mM ethylenediamine tetracetic acid (EDTA), pH 8.0) for 30 minutes at 37°C. RNase treated sections were washed at increasing stringency, as follows: 2 x SSC (10 minutes), 1 x SSC (10 minutes), 0.5 x SSC (10 minutes), and 0.25 x SSC (60 minutes) (PI, pz) or 0.25 x SSC (10 minutes) and 0.125 x SSC (60 minutes) (p3) at 70°C. After final washes in 0.25 x SSC (10 minutes) at R.T. (PI, pz),and 0.125 x SSC (10 minutes) at R.T. (PSI, sections were dehydrated (1minute each) through 70%, SO%, 90% and 100% EtOH. All washing solutions, except RNase A solution, contained 10 mM 2-mercaptoethanol to prevent nonspecific binding of riboprobe. The nonspecific hybridization signal was determined by exposing adjacent sections to sense RNA probes. Ligand Binding. Slide-mounted sections of adult (PSO) mouse brain were prepared as described previously (Frostholm and Rotter, '87). Sections were preincubated in phosphate buffered saline (PBS),pH 7.4, for 30 minutes at 4"C, incubated for 40 minutes in PBS containing 5 nM [3Hlmuscimol (9 Ciimmole, NEN) at 4"C, and washed twice for 30 seconds in ice cold buffer to remove unbound radioactivity. Slides were then dipped into ice cold doubledistilled water to remove residual salts. Control slides for nonspecific binding were labeled in the presence of 200 pM GABA. Autoradiography. Autoradiograms were generated as follows: acid-washed coverslips (25 x 77 mm, No. 0 Corning Glass Works, Corning, NY),previously coated with a uniform layer of photographic emulsion (Ilford K5-D, Polysciences, Inc., Warrington, PA) were apposed to the slidemounted sections and clamped together with binder clips under minimum sodium safelight illumination. The assemblies were placed in lightproof boxes containing desiccant and exposed for 5 days at 4°C. Coverslips containing autoradiographs were developed in Kodak D-19 developer (diluted 1:lwith distilled water) for 4 minutes at 20°C, fixed for 3.5 minutes in Kodak Rapid-Fix, washed in distilled water for 30 minutes, and mounted onto microscope slides with Depex mounting medium (Bioimedical Specialties, Santa Monica, CA). After the coverslips were developed, slides containing the sections were dipped in Ilford K5-D photographic emulsion at 42"C, dried overnight, exposed for 8 days at 4"C, and developed as above. Emulsion covered sections were counterstained with cresyl fast violet. Sections and autoradiograms were photographed (Kodak Panatomic-X film) under brightfield or darkfield illumination with a Nikon SMZ-10 stereomicroscope (low magnification), or a Nikon Optiphot compound microscope (high magnification). Immunocytochemistry. An adult (P60) mouse brain was sectioned in the coronal plane at 40 pm. Sections were preincubated with 0.1 M lysine in 0.1 M phosphate buffer (pH 7.4) and 10% normal horse serum (NHS) for 1 hour prior to immunoreaction. After 2 x 5 minute washes in cold phosphate buffer, sections were incubated for 16 hours at 4°C with the mAb 62-3G1 antibody (1:1,000 dilution in phosphate buffer) containing 0.3% Triton X-100. Sections were then washed (3 x 10 minutes) in cold phosphate buffer. Antibody binding to cerebellar sections was revealed

GABAA RECEPTOR P-SUBUNIT mRNAs

81

583

82

7.4-

5.32.81.91.6-

Fig. 1. Autoradiographs of Northern blots of adult mouse brain poly A+ mRNA hybridized with PI, pz, and p3 mouse cRNA probes. Molecular sizes of standards are expressed in kilobases.

by reaction with a biotin-labeled antibody (horse antimouse IgG), followed by an avidin-biotin horseradish peroxidase complex (ABC “Vectastain” procedure, Vector Labs, Burlingame, CA). Additional sections were placed in a solution of 0.3% hydrogen peroxide (HzOz)in methanol for 30 minutes, in order to destroy endogenous peroxidase activity. Following 2 x 5 minute washes, sections were transferred to a solution of 0.5% diaminobenzidine in 0.05 M Tris/O.S% NaCl for 10 minutes at R.T. Staining was performed in the above solution, containing 0.01% HzOz, for approximately 20 seconds. Finally, sections were rinsed in phosphate buffer, mounted onto glass slides and allowed to dry overnight. Sections were cleared in xylenes, coverslipped and examined by light microscopy. Controls were processed as above with the medium containing the parental myeloma P3X63Ag8.6.5.3. instead of the 62-3G1 antibody.

RESULTS Northern blots Each of the riboprobes was specifically designed to include regions complementary to the most variable portion of the corresponding mRNA, in order to maximize specificity of hybridization. Northern blot experiments, conducted under high stringency conditions, using mouse cRNA probes, revealed major bands of approximately 11 kb (PI), 8.2 kb (pz) and 6.1 kb (p3) (Fig. 1). Identical patterns were observed with rat cRNA probes. No cross hybridization between the three subunit mRNAs was observed in either case.

Fig. 2. Low magnification photomicrographs of p subunit mRNAs in the adult (P80) mouse cerebellum, hybridized with mouse cRNA probes. A: PI mRNA hybridization signal. B: p2 mRNA hybridization signal. C : p3 mRNA hybridization signal. g, granule cell layer; rn, molecular layer; P, Purkinje cells; w, white matter; dcn, deep cerebellar nuclei; arrowheads, interface between molecular and granule cell layers. Scale bars, 1mm.

cerebellum. At low magnification, the highest level of PI riboprobe labeling (Fig. 2A) was visible as a discrete band at the interface of the molecular and granule cell layers. When viewed at high magnification (Fig. 3A), many grains appeared to be diffusely distributed over small, densely stained cells surrounding Purkinje cells. Relatively few grains were visible over the Purkinje cell bodies themselves. The molecular and granule cell layers displayed much lower hybridizaAdult distribution of PI, p2, tion signals (Figs. 2A, 3A). The deep cerebellar nuclei were and p3 cRNA probes weakly labeled, as was the white matter (Fig. 2A). The All three GABA,/BZ f3 subunit probes generated strong highest level of pz (Fig. 2B) and p3 (Fig. 2C) probe labeling in situ hybridization signals in the adult C57iBL6 mouse was present over the granule cell layers, while the grain

584

Fig. 3. High magnification photomicrographs of subunit probes in the adult (P80) mouse cerebellum. A: mRNA hybridization signal (rat cRNA probe); B: pa mRNA hybridization signal (mouse cRNA probe); C: p3 mRNA hybridization signal (rat cRNA probe). g, granule cell layer; m, molecular layer. Arrows point to Purkinje cells (PI. Scale bars, 20 km.

density in the molecular layer was low. As with the p1 probe, at high magnification, much of the pz (Fig. 3B) and p3 (Fig. 3C) probe labeling in the Purkinje cell layer was observed over smaller cells. In addition, the pz probe generated a significant number of autoradiographic grains over Purkinje cell bodies; far fewer grains were observed over these cells with the p3 probe. The deep cerebellar nuclei displayed moderate to high levels of p2 (Fig. 2B) and

D. ZDILAR ET AL.

Fig. 4. Localization of GMAA receptor markers. A: Brightfield photomicrograph of a cresyl fast violet stained section, showing the laminar structure of the cerebellar cortex. B: Darkfield photomicrograph showing that the highest level of [3Hlmuscimol binding sites are present in the granule cell layer. C: Immunocytochemical staining with mAb 62-3G1 shows that the antibody is largely localized in the granule cell layer. The tissue has been lightly counterstained with cresyl fast violet to show the Purkinje cell layer; Purkinje cell bodies are not stained with the antibody. g, granule cell layer; m, molecular layer; P, Purkinje cell layer; w, white matter. Scale bars, 50 pm.

GABAARECEPTOR p-SUBUNlT mRNAs p3 (Fig. 2C) labeling. Control slides incubated with corresponding sense probes showed low, uniform labeling with no anatomical pattern (data not shown). The adult cerebellar distribution of pz and p3 mRNA probes showed a high degree of correspondence with other f3 subunit markers (Fig. 4). [3H]Muscimol binding sites (Fig. 4B) and monoclonal antibody mAb 62-3G1 binding (Fig. 4C) were localized by ligand autoradiography and immunocytochemistry, respectively. The highest levels of pz and p3 subunit mRNAs were present in the granule cell layer, while the molecular layer was less densely labeled. High levels of [3Hlmuscimoland mAb 62-3G1binding were also detected in the granule cell layer; in contrast to the pz and p3 mRNA hybridization signal, however, neither [3Hlmuscimol nor mAb 62-3G1 binding were observed in the Purkinje cell layer.

Localization of PI probe during postnatal development The postnatal distribution of p1subunit mRNA is shown in Figures 5, 6, and 7. During the first postnatal week, the p1 hybridization signal was low overall. It was absent from the external germinal layer (egl) and uniformly distributed throughout the rest of the cerebellar cortex (Figs. 5B, 6D). This pattern was observed until P7, when increased grain density was observed over the internal granular layer (Figs. 5D, 6J). Between P 9 and P11, a band of elevated labeling became concentrated at the interface of the molecular and granule cell layers (Figs. 5F, 7D) and was maintained into adulthood (Fig. 2A). The deep cerebellar nuclei were moderately labeled at birth (Fig. 5B); grain density increased through postnatal weeks 1and 2 (Fig. 5D,F), but decreased throughout the third postnatal week, reaching low adult levels at approximately P18 (Fig. 5H). The distribution of the p1 subunit probe was also examined in the 140 day old Purkinje cell degeneration mutant (Fig. 8). A distinct band of labeling was observed between the molecular and granule cell layers, despite the absence of Purkinje cells (Fig. 8A,B,C).Although the grain density was slightly lower than that of normal animals, the overall pattern of labeling was similar (compare Figs. 2A and 8A).

Localization of p2 probe during postnatal development The postnatal distribution of the p2 hybridization signal is shown in Figures 6, 7, and 9. The pz subunit mRNA hybridization signal was clearly detectable at birth and was most concentrated in the moleculariPurkinje cell layer, a pattern that persisted throughout the first postnatal week (Figs. 6E,K and 9A,B). As foliation increased and the Purkinje cells formed a monolayer, a thin band of grains became concentrated over the Purkinje cell layer (Fig. 6K). Between P9-11, as granule cell migration reached its peak, the internal granular layer became more densely labeled (Figs. 7E, 9C); the granule cells of lobules IX and X were labeled at a significantly higher level than the remainder of the cerebellar cortex (Fig. 9C). By P14, the granule cell layer in each cerebellar lobule was labeled at the same high density (Fig. 9D); moderate grain density was also seen in the molecular layer. The adult distribution and grain density was reached by P20, when the highest labeling was present in the granule and Purkinje cell layers (Fig. 7K). The external germinal layer was labeled at background levels. Labeling over the deep cerebellar nuclei was present

585 at birth and increased throughout postnatal development (Fig. 9A-D) into adulthood (Fig. 2B).

Localization of P 3 probe during postnatal development The postnatal distribution of the p3 hybridization signal is shown in Figures 6 , 7 , and 10. The p3 mRNA showed the earliest developmental expression of the three subunits: whereas the p1and pz hybridization signals were generally absent from the external germinal layer, a low, but significant, p3 hybridization signal was first observed over the external germinal layer at birth (Fig. 6F). During postnatal week one, the highest signal was visible over the molecular layer (Fig. 6F, 10B). In postnatal week two, as granule cells migrated across the molecular layer into the internal granular layer with greater frequency, the signal within the latter region became more intense, and the external germinal and internal granular layers became labeled at equal density (Figs. 6L, 7F). Between P11 and P14, the hybridization signal over the external germinal layer became gradually reduced as granule cells completed their migration into the granule cell layer. Grain density in the granule cell layer increased rapidly during this period, until adult levels were reached by approximately P20 (Figs. 7L and 10F). The in situ hybridization signal was present over the deep cerebellar nuclei at birth and increased throughout development, reaching the moderate adult levels by approximately P20 (Fig. 10A-F). Autoradiograms of the corresponding sense probes showed only very faint, uniform patterns of grain density at each developmental age (data not shown).

DISCUSSION Ribonucleotide probes, complementary to either rat or mouse cDNA sequences, were used to localize GABAA receptor p1-3 subunit mRNAs in postnatal mouse cerebellum. Several lines of evidence indicate that the cRNA probes used in these experiments were specific for their respective mRNAs: probes based on mouse sequences (pl, p2, and p3) were complementary to the region of mRNA coding for the putative intracellular loop linking the M3 and M4 transmembrane segments, this segment being the most variable region in each subunit. Probes based on the rat sequence (pl and p3), although of greater length, also included portions coding for the intracellular loops. Northern blot studies showed no cross-hybridization between the three cRNA probes. The sizes of the subunit mRNAs were similar to those reported previously for the rat brain (Ymer et al., '89; Zhang et al., '91), with the exception of the p3 probe where a 2.5 kb band observed in the rat was absent in the mouse. Both rat and mouse subunit probes generated identical autoradiographic distributions throughout the cerebellum. In additional experiments (unpublished observations), using [35S]oligonucleotideprobes complementary to the region coding for the intracellular loops, the distribution of the in situ hybridization signal for each of the three p subunits was similar to that of the cRNA probes. The cerebellar distribution of the in situ hybridization signal for each of the riboprobes was studied in adult mice. The p1 signal was highest over a narrow region at the interface of the granule cell and molecular layers, with much lower levels over the granule cell and molecular layers. Purkinje cells were unlabeled. The f32 probe was present at moderate levels in both granule and Purkinje

586

Fig. 5. Appearance of p1 subunit mRNA (rat cRNA probe) during postnatal development ofthe mouse cerebellum. Left column: Emulsiondipped coronal sections through the cerebellum were counterstained with cresyl fast violet to show laminar organization, and photographed at low power under brightfield illumination. Right column: The same

D. ZDILAR ET AL.

section photographed under darkfield illumination showing the in situ hybridization signal. A, B: P1; C, D: P7; E, F: P11; G, H:P18.Large arrows, deep cerebellar nuclei; small arrows, interface of the molecular and granule cell layers. Scale bars: A, C, 500 pm (also applies to B and D); E, G , 1mm (also applies to F and H).

GABAA RECEPTOR P-SUBUNIT mRNAs

587

Figure 6

588 cells, while the p3probe was largely concentrated in granule cells, with only a minor component in Purkinje cells. Low grain density was also present over the basket and stellate cells in the molecular layer. This distribution is in general agreement with observations by Zhang et al. ('91) and Gambarana et al. ('91), who used oligonucleotide probes to subunit mRNAs in rat brain. In those localize the studies, the granule cell layer was moderately to strongly labeled by both pz and p3probes, while Purkinje cells were strongly labeled with the p2 probe only. The availability of radioligands and monoclonal antibodies for the GABAA receptor makes it possible to compare the localization of the receptor binding sites and the receptor protein with that of the receptor subunit mRNAs. The ligand [3H]muscimol binds to the GABAA receptor with high affinity, and photolabeling studies have revealed that the binding site resides on the p subunits of the receptor (Casalotti et al., '86; Deng et al., '86; Fuchs and Sieghardt, '89). A monoclonal antibody, mAb62-3G1, specifically raised against affinity-purified GABAJBZ receptor (Vitorica et al., '88) was also found to recognize the p subunits, and recent experiments suggest that the specific recognition and p3 polypeptides (Ewart et epitope is located on the al., '92). There appears to be little relationship between the adult pattern of subunit mRNA and that of [3H]muscimol and mAb62-3G1 monoclonal antibody labeling in the mouse cerebellum, as only low levels of p1 mRNA were present in the granule cell layer. In the case of the pz and p3 subunits, however, the mRNAs, mAb62-3G1 and [3H]muscimol binding sites showed good spatial correspondence, each being present at high levels in the adult granule cell layer. Both markers are located on granule cell dendrites which are intermingled with p2 and p3 containing granule cell bodies, thus leading to the observed colocalization within the granule cell layer. The adult molecular layer, which consists primarily of Purkinje cell dendrites, granule cell axons, and basket and stellate cells, had only low levels of the pz and p3 probe signal which appeared to be associated with basket and stellate cell bodies. Labeling with the mAb 62-3G1 antibody and L3H]muscimolwas also low in the molecular layer, indicating that neither basket nor stellate cells synthesize significant levels of the pz and p3 subunit polypeptides. Purkinje cell bodies contained the pz and, to a lesser extent, the p3 signal. The low levels of immunostaining and ligand binding over the cell bodies and over the dendritic regions within the molecular layer may again indicate that only very low levels of the p2 and p3 subunit proteins are synthesized in, or transported to, this region. A unique temporal expression was observed during cerebellar development for each of the three subunits, with the p1mRNA appearing considerably later than either the pzor p3 signals. A low p1 hybridization signal was present

Fig. 6. The distribution of @ subunit mRNAs in individual cerebellar cortical folia during development. Autoradiograms were generated on emulsion-dipped,counterstained sections and photographed under darkfield illumination. Sections were photographed under brightfield illumination and are shown above each autoradiogram. Left column: subunit mRNA (rat cRNA probe). A, D: P1; G , J P7. Center column: pz subunit mRNA (mouse cRNA probe). B, E: P1; H, K: P7. Right column: p3 subunit mRNA (rat cRNA probe). C,F: P1; I, L P7. e, external germinal layer; i, internal granular layer; m, molecular layer; P, Purkinje cell layer; w, white matter. Scale bars, 100 km.

D. ZDILAR ET AL. throughout the cerebellar medullary region during the first postnatal week. During postnatal week two, the signal increased rapidly over a band of small, densely stained cells at the interface of the molecular and developing internal granular layers. In previous studies in the rat, a weak signal was observed over both granule (Zhang et al., '91; Gambarana et al., '91) and Purkinje cell layers (Zhang et al., '91). However, in neither of the two studies did the labeling form a distinct narrow band as was observed in the mouse, suggesting that the cellular localization of p1mRNA is not identical in the two species. In the mouse, the hybridization signal within the band was diffuse rather than punctate and localized to a narrow region in the general vicinity of the Purkinje cells and the surrounding Bergmann glia and basket cells. It is unclear, however, which of the three cell types contain the p1 subunit. Very few grains were seen over Purkinje cell bodies. Furthermore, a distinct band of elevated grain density was clearly present in the 140-day-oldPurkinje cell degeneration mouse, in which virtually all Purkinje cells degenerate by P45 following an initially normal developmental sequence (Mullen et al., '76). The presence of a band of grains similar to that found in the normal murine cerebellum indicates a cellular localization other than that of the Purkinje cell. In previous studies, GABAA receptors have been shown to be present on cultured glial cells (Kettenmann et al., '871, and more recently, p1 subunit immunoreactivity has been observed in white matter (Gu et al., '92). Bergmann glial cells are present prior to birth, and their cell bodies remain restricted to the Purkinje cell layer throughout postnatal development (de Blas and Cherwinski, '85). It is unclear whether the increase in p1 labeling observed between postnatal days 9-11 is associated with these cells, or with the "very low basket" cells observed in the mouse (LemkeyJohnston and Larramendi, '681, which appear at the interface of the granule and molecular layers at approximately the same time as the hybridization signal. Further immunocytochemical studies are required for a more precise cellular localization of this subunit. The in situ hybridization signals of both pz and p3 mRNAs were detectable at birth; the strongest signal was primarily localized in the superficial layers of the cerebellar cortex and gradually became translocated to deeper regions during postnatal weeks one and two. Both subunit mRNAs appeared to be associated largely with two major cell types, the Purkinje and granule cells, each of which have different developmental origins and migration patterns. At approximately embryonic day 13, Purkinje cells migrate radially from the ventricular germinal zone to the dorsal surface of the cerebellar cortex (Uzman, '60; Miale and Sidman, '61; Yuasa et al., '91). By postnatal day 1, they form a multiply stacked band in the molecular layer, between the external germinal and internal granular layers. The characteristic Purkinje cell monolayer is formed between postnatal days 5-7, as foliation increases. During the first postnatal week, the pattern of labeling generally followed the development of the Purkinje cell layer. Both pz and p3 signals were clearly visible in the developing molecular layer at the beginning of the first postnatal week. Grains became clearly concentrated over Purkinje cell bodies by P7; this was particularly evident with the pz probe. During the second postnatal week, granule cells, which are generated in the external germinal layer almost entirely after birth, migrate through the molecular layer into the internal granular layer where they become integrated into cerebellar cortical

GABAARECEPTOR P-SUBUNIT mRNAs

589

Figure 7

590

D. ZDILAR ET AL.

Fig. 8. The autoradiographic distribution of b1 subunit mRNA (rat cRNA probe) in the Purkinje cell degeneration (pcdipcd) mutant. A Cerebellar lobule of pcdipcd mouse, showing a band of autoradiographic grains at the interface of the molecular and granule cell layers, photographed under darkfield illumination. B: Brightfield autoradio-

graph at higher magnification; note absence of Purkinje cells. C: Darkfield autoradiograph of same area. g, granule cell layer; m, molecular layer; arrowheads indicate band of high grain density. Scale bars: A, 100 km; B, C, 25 km.

circuitry (Fujita et al.,'66; Larramendi, '69; Altman, '72). Increasing autoradiographic grain density was clearly associated with the developing granule cell layer. However,

although the peak of granule cell migration occurs between postnatal days 7-10 (Fujita et al., '661, early migrating granule cells within the molecular layer may be labeled in addition to Purkinje cells stacked within this region, making the precise cellular localization of p subunit mRNAs difficult to resolve. It is noteworthy that grain density in the granule cell layer of the nodulus and uvula (lobules IX and X of the archicerebellum) increased before that of other cortical regions. In a previous study using an ctl subunit cRNA probe, we observed that Purkinje cells within the same cerebellar region had a similar developmental expression (Zdilar et al., '91). A mediolateral and caudal to rostra1 sequence of Purkinje cell development has been described

Fig. 7. The distribution of p subunit mRNAs in individual cerebellar cortical folia during development (continued from Fig. 6). Left column: p1 subunit mRNA (rat cRNA probe). A, D: P11; G, J: P20. Center column: pz subunit mRNA (mouse cRNA probe). B, E: P11; H, K P20. Right column: p3 subunit mRNA (rat cRNA probe). C, F: P11; I, L P20. e, external germinal layer; g, granule cell layer; i, internal granular layer; m, molecular layer; P, Purkinje cell layer; w, white matter; arrowheads, interface between molecular and granule cell layers. Scale bars: A-F, 100 km; G-L, 250 km.

GABAARECEPTOR P-SUBUNIT mRNAs

591

Fig. 9. Darkfield autoradiographs of sagittal sections through the developing mouse cerebellum, showing the distribution of the pzmouse cRNA probe. Autoradiographs were generated on emulsion coated coverslips. A P1; B: P5; C: P9; D: P14. g, granule cell layer; m,

molecular layer; P, Purkinje cell layer; arrow, deep cerebellar nuclei; arrowhead indicates elevated grain density in the granule cell layer of lobule IX of the archicerebellum. Scale bars, 500 pm.

in the rat (Altman and Bayer, '85) and in the mouse (Inouye et al., '80). Unlike the p2 hybridization signal, the p3 signal was present in the external germinal layer throughout its existence. A similar distribution has been observed in the developing rat cerebellum (Gambarana et al., '91). Previous studies on the ontogeny of [3H]muscimol binding sites in the murine cerebellum (Frostholm and Rotter, '87) have demonstrated that granule cells begin to express GABAA receptors coincidentally with their migration across the molecular layer during postnatal week two. This finding implies that acquisition of receptors precedes the formation of synaptic contacts, which are not formed until granule cells reach their final position in the granule cell layer (Shimono et al., '76). The expression of pz and p3 (but not PI) subunit mRNAs follows a pattern which is consistent with this interpretation. Both p2and p3 mRNAs are present in the molecular layer of early postnatal mice. Presumably, either or both subunit messages are available for translation into the [3Hlmuscimol binding polypeptide. The presence of the p3 mRNA hybridization signal in the external germinal layer implies that granule cells are committed to the expression of p3 subunit message even earlier in their development, possibly during proliferation, and prior to their migration. Whether this message is translated into a corresponding subunit protein while the cells are still in the external germinal layer remains to be determined.

The developmental distributions of the three probes within the deep cerebellar nuclei were not identical. The PI signal was transiently expressed: although punctate clusters of grains were present over the cells of each of the three subnuclei at birth and through the first two postnatal weeks, they were virtually absent in adult animals. The absence of hybridization signal in the deep cerebellar nuclei of adult mice is in general agreement with observations on the rat by Gambarana et al. ('91) and Zhang et al. ('91). The pz and P3 mRNAs were also present in each of the subnuclei at birth; the signals increased and remained strong into adulthood. The same authors also observed strong pzlabeling in the deep cerebellar nuclei of adult rats. The p3 mRNA signal observed in mice was reported to be absent in the adult rat, however, suggesting possible species differences. One might expect that pz and p3 subunit mRNAs within the deep cerebellar nuclei would be accompanied by an appropriate dendritic distribution of [3H]muscimol binding sites within the same area. However, the developmental distribution of [3Hlmuscimol binding sites within this region appears to be more similar to that of the probe, since moderate [3H]muscimollabeling was present in the deep cerebellar nuclei of young animals and grains were virtually absent in adult mice (Frostholm and Rotter, '87). Studies performed to date have indicated that in order to obtain a functional receptor with properties similar to those

592

D. ZDILAR ET AL.

Fig. 10. Darkfield autoradiographs of coronal sections through the developing mouse cerebellum and brainstem taken at low power, showing the p3 subunit mRNA hybridization signal (rat cRNA probe). Autoradiographs were generated on emulsion-dipped sections, counter-

stained with cresyl fast violet Autoradiographic grains over the granule cells are partial1 obscured by the cresyl fast violet staining. A P1; B: P4; C: P7; D: P9; E: P11; F P20. Arrow, deep cerebellar nuclei. Scale bars, 1 mm.

of the native receptor, the oligomeric complex must include, in addition to the p subunits, an ci and a y subunit (Pritchett et al., '89). The presence of the a and f3 subunits is required for the receptor to possess benzodiazepine and GABA binding sites, respectively (Sigel et al., '83; Casalotti et al., '86). The y subunit confers the ability of benzodiazepines t o allosterically modulate GABA sensitivity (Pritchett et al., '89). The ontogenic appearance of the cil and y2 subunits has been compared with that of the f3 subunits

throughout the rat brain (Gambarana et al., '91). The a1 and yz mRNAs were found t o rise in parallel with the p2 subunit in many brain regions, including the cerebellar granule and Purkinje cells. In the mouse cerebellum, the expression of cil (Zdilar et al., '91; Frostholm et al., '91) and y2 (unpublished data) mRNA shares most of the features with the f3z developmental pattern. As with the p2 signal, the a1 and yz mRNAs are present in the molecular/Purkinje cell layer from birth, and as development proceeds, they

GABAA RECEPTOR @-SUBUNITmRNAs become clearly associated with Purkinje and granule cells. The observations in the rat and mouse cerebellum indicate that the developmental expression of p subunits is accompanied by the appearance of other subunit mRNAs, thus permitting the assembly of a fully functional GABAAiBZ receptor. Indeed, the p subunit antibody, mAb 62-3G1, has been found to coprecipitate the (Y subunit in the developing rat at all postnatal ages. This suggests that the coupling between the cx and p subunits is present at birth and is maintained thereafter (Vitorica et al., '90). In summary, our results lead to several broad conclusions. Firstly, the three p subunit mRNAs are differentially distributed in the mouse cerebellum, suggesting that the expression of a specific GABAA/BZreceptor subunit combination is cell type-dependent. Secondly,the temporal appearance of the p messages in a given cell type is not identical, suggesting that their expression is independently regulated. Thirdly, acquisition of pz and p3 mRNAs by granule cells occurs at a stage before these cells have become integrated into mature synaptic circuits. This suggests that synaptogenesis does not regulate receptor expression, but rather may indicate that GABAA~BZreceptor expression may influence subsequent GABAergic synapse formation.

ACKNOWLEDGMENTS We are very grateful to Dr. Allan Tobin for providing us with the rat p1 and p3 cDNA clones and to Dr. Angel deBlas for providing the mAb 62-3G1 antibody. This work was supported by USPHS grants AA08520, NS18089 to A. Rotter and NS27784 to A. Frostholm. V. Luntz-Leybman was supported by Training Grant NS07291.

LITERATURE CITED Altman, J. (1972) Postnatal development of the cerebellar cortex in the rat. 111. Maturation of the components of the granular layer. J. Comp. Neurol. 145:465-514. Altman, J., and S.A. Bayer (1985) Embryonic development of the rat cerebellum. 111. Regional differences in the time of origin, migration, and settling of Purkinje cells. J. Comp. Neurol. 231:42-65. Badly, J.E., G.A. Bishop, T. St John, and J.A. Frelinger (1988) A simple, rapid method for the purification of poly A+ RNA. Biotechniques 6:114-116. Bateson, A.N., A. Lasham, and M.G. Darlison (1991) y-Aminobutyric acid, receptor heterogeneity is increased by alternative splicing of a novel p subunit gene transcript. J. Neurochem. 56:1437-1440. Benke, D., S. Mertens, A. Trezciak, D. Gillessen, and H. Mohler (1991) GABA, receptors display association of y2-subunit with a1 and p213 subunits. J. Biol. Chem. 26614478-4483. Burt, D.R., and G.L. Kamatchi (1991) GABA, receptor subtypes: From pharmacology to molecular biology. FASEB J. 52916-2923. Casalotti, S.O., F.A. Stephenson, and E.A. Barnard (1986) Separate subunits for agonist and benzodiazepine binding in the y-aminobutyric acid, receptor oligomer. J. Biol. Chem. 261:15013-15016. de Blas, A.L., and H.L. Cherwinski (1985) The development of the Bergmann fiber palisades in the cerebellum of the normal rat and in the weaver mouse. Brain Res. 342234-241. de Blas, A.L., J. Vitorica, and P. Friedrich (1988) Localization of the GABA, receptor in the rat brain with a monoclonal antibody to the 57,000 M, peptide of the GABA, receptoribenzodiazepine receptoriC1- channel complex. J. Neurosci. 8:602-614. Deng, L., R.W. Ransom, and R.W. Olsen (1986) 3H-Muscimol photolabels the gamma aminobutyric acid receptor binding site on a peptide subunit distinct from that labeled with benzodiazepines. Biochem. Biophys. Res. Commun. 138:1308-1314. Ewart, M., A.L. de Blas, H. Mohler, and P.H. Seeburg (1992) A prominent

593 epitope on GABA, receptors is recognized by two different monoclonal antibodies. Brain Res. 569157-62. Fourney, R.M., J. Miyakoshi, R.S. Day 111, and M.C. Patterson (1988) Northern blotting: Efficient RNA staining and transfer. Focus 10:5-7. Frostholm, A,, and A. Rotter (1987) The ontogeny of [~Hlmuscimolbinding sites in the C57BLi6J mouse cerebellum. Dev. Brain Res. 37:157-166. Frostholm, A,, D. Zdilar, A. Chang, and A. Rotter (1991) Stability of GABA,/benzodiazepine receptor a1 subunit mRNA expression in reeler mouse cerebellar Purkinje cells during postnatal development. Dev. Brain Res. 64:121-128. Fuchs, K., and W. Sieghart (1989) Evidence for the existence of several different a and @ subunits of the GABAibenzodiazepine receptor complex from rat brain. Neurosci. Lett. 97:329-333. Fujita, S., M. Shimada, and T. Nakamura (1966) HQhymidine autoradiographic studies on the cell proliferation and differentiation in the external and the internal granular layers of the mouse cerebellum. J. Comp. Neurol. 128:191-208. Gambarana, C., C.E. Beattie, Z.R. Rodriguez, and R.E. Siege1 (1991) Region-specific expression of messenger RNAs encoding GABA, receptor subunits in the developing rat brain. Neuroscience 45423432. Gu, Q., J.L.P. Velazquez, K.J. Angelides, and M.S. Cynader (1992) GABA, receptor immunoreactivity in the white matter. NeuroReport. 3: 169172. Inouye, M., and U. Murakame (1980) Temporal and spatial patterns of Purkinje cell formation in the mouse cerebellum. J. Comp. Neurol. 194:499-503. Kalvakolanu, D.V.R., and W.H. Livingston 111(1991) Rapid and inexpensive protocol for generating > 95% recombinants in subcloning experiments. Bio Techniques 10:176-177. Kettenmann, H., K.H. Backus, and M. Schachner (1987) y-aminobutyric acid opens C1-channels in cultured astrocytes. Brain Res. 404: 1-9. Kofuji, P., J.B. Wang, S.J. Moss, R.L. Huganir, and D.R. Burt (1991) Generation of two forms of the y-aminobutyric acid, receptor y,-subunit in mice by alternative splicing. J. Neurochem. 56:713-715. Larramendi, L.M.H. (1969) Analysis of synaptogenesis in the cerebellum of the mouse. In R. Llinas (ed): Neurobiology of Cerebellar Evolution and Development. Chicago: A.M.A. pp. 803-843. Lemkey-Johnston, N., and L.M.H. Larramendi (1968) Types and distributions of synapses upon basket and stellate cells of the mouse cerebellum: An electron microscopic study. J. Comp. Neurol. I34:73-112. Malherbe, P., A. Draguhn, G. Multhaup, K. Bayreuther, and H. Mohler (1990) GABA,-receptor expressed from rat brain a- and @-subunit cDNAs displays potentiation by benzodiazepine receptor ligands. Mol. Brain Res. 8:199-208. Miale, I.L., and R.L. Sidman (1961) An autoradiographic analysis of histogenesis in the mouse cerebellum. Exp. Neurol. 4:277-296. Mullen, R.J., E.M. Eicher, and R.L. Sidman (1976) Purkinje cell degeneration, a new neurological mutation in the mouse. Proc. Natl. Acad. Sci. USA 73:208-212. Olsen, R.W., and A.J. Tobin (19901 Molecular biology of GABA, receptors. FASEB J. 4:1469-1480. Palacios, J.M., W.S. Young 111, and M.J. Kuhar (1980) Autoradiographic localization of GABA receptors in rat cerebellum. Proc. Natl. Acad. Sci. USA. 771670-674. Pritchett, D.B., H. Sontheimer, B.D. Shivers, S. Ymer, H. Kettenmann, P.R. Schofield, and P.H. Seeburg (1989) Importance of a novel GABA, receptor subunit for benzodiazepine pharmacology. Nature 338~582585. Seeburg, P.H. (1990) Elucidating GABA, receptor heterogeneity: An integrated molecular approach. In G. Biggio and E. Costa (eds): GABA and Benzodiazepine Receptor Subtypes. New York: Raven Press, pp. 15-21. Shimono, T., S. Nosaka, and K. Sasaki (1976) Electrophysiological study of the postnatal development of neuronal mechanisms in the rat cerebellar cortex. Brain Res. 108:279-294. Shivers, B.D., I. Killisch, R. Sprengel, H. Sontheimer, M. Kohler, P.R. Schofield, and P.H. Seeburg (1989) Two novel GABA, subunits exist in distinct neuronal subpopulations. Neuron. 3:327-337. Sigel, E., F.A. Stephenson, C. Mamalaki, and E.A. Barnard (1983) A y-aminobutyric acidibenzodiazepine receptor complex of bovine cerebral cortex. Purification and partial characterization. J. Biol. Chem. 258:69656971. Somogyi, P., H. Takagi, J.G. Richards, and H. Mohler (1989) Subcellular localization of benzodiazepineiGABA, receptors in the cerebellum of rat, cat and monkey using monoclonal antibodies. J. Neurosci. 9:2197-2209.

594 Uzman, L.L. (1960) The histogenesis of the mouse cerebellum as studied by its tritiated thymidine uptake. J. Comp. Neurol. 114:137-160. Vitorica, J., D. Park, G. Chin, and A.L. de Blas (1988) Monoclonal antibodies and conventional antisera to the GABA, receptoribenzodiazepine receptor/Clk channel complex. J. Neurosci. 8515-622. Vitorica, J.,D. Park, G. Chin, andA.L. de Blas (1990) Characterization with antibodies of the y aminobutyric acid,ibenzodiazepine receptor complex during development of the rat brain. J. Neurochem. 54.187-194. Ymer, S., P.R. Schofield, A. Draguhn, P. Werner, M. Kohler, and P.H. Seeburg (1989) GABAA receptor p subunit heterogenity: Functional expression of cloned cDNAs. EMBO J. 8:1665-1670.

D. ZDILAR ET AL. Yuasa, S., K. Kawamura, K. Ono, T. Yamakuni, and Y. Takahashi (1991) Development and migration of Purkinje cells in the mouse cerebellar primordium. Anat. Embryol. 184.195-212. Zdilar, D., A. Rotter, and A. Frostholm (1991) Expression of GABAA/ benzodiazepine receptor a, subunit mRNA and PHI flunitrazepam binding sites during postnatal development of mouse cerebellum. Dev. Brain Res. 61.63-71. Zhang, J-H., M. Sato, and M. Tohyama (1991) Region-specific expression of the mRNAs encoding p subunits (PI, Pn, and p3) of GABA, receptor in the rat brain. J. Comp. Neurol. 303.637-657.