ABSTRACT

GABA_A receptor-channels are pentameric ligand-gated ion channels central to the control of excitability and neuroendocrine secretion. Although adrenal chromaffin cells are known to possess functional GABA_A class receptors, their subunit composition andphysiological role remain to be characterized.  The expression profile of GABA_A receptor subunits was therefore determined both within the rat adrenal and the PC12 cell line.  Both serum-deprived PC12 cells and rat adrenal appear to express a3, a4, g1, g2 and g3 subunits as determined by both real-time and conventional RT-PCR.  No b subunit was positively identified within the adrenal, but mRNAs for b1, b2 and b3 subunits were detected in the PC12 cell line.  Immunocytochemistry reveals the colocalized surface expression of both b and g subunits in the PC12 cell.  Insulin, Nerve Growth Factor (NGF) and Pituitary Adenylate Cyclase Activating Polypeptide (PACAP-38) modulate the expression level of a, b and g subunits in the PC12 cell.  These results suggest that the PC12 cell may serve as an appropriate model for the regulation of GABA_A subunit expression and receptor function in the adrenal.  Furthermore, the differential regulation of the expression of GABA_A receptor subunit isoforms by insulin, NGF and PACAP observed emphasizes the centrality of the GABA receptor system in the regulation of plasticity and excitability within the neuronal and neuroendocrine systems.

Key words: PACAP, NGF, matrix factors, PCR, immunocytochemistry, real-time PCR.

INTRODUCTION

GABA_A receptors belong to a superfamily of ligand-gated ion channels which includes the nicotinic acetylcholine, 5-HT₃, GABA_C and glycine receptors (Ortells and Lunt, 1995).     These receptors share a common predicted topology, with an N-terminus that contributes towards the ligand binding domain, four TransMembrane spanning domains (TM1-4), an intracellular regulatory loop that receives phosphorylation signals and receptor anchoring, and a pore region contributed by residues of the second transmembrane domain (Sigel, 1995).  The considerable functional diversity of GABA_A receptor-channels arises not only from the diversity of subunits which coassemble to form ion channels, a repertoire which includes a_1-6, b_1-4, g_1-4, d, e, p and q (Rudolph et al., 2001, Whiting et al., 1999a; 1999b), but also from the alternative splicing and RNA editing of these subunits (Huntsman et al., 1998; Hosie etal., 2001; Quinlan et al., 2000), and post-translational events including receptor clustering (Phillips and Froehner, 2002) and phosphorylation (Sigel, 1995).

Functional GABA_A receptors are known to be present upon adrenal chromaffin cells (Peters et al., 1989), but at present the subunit expression profile and biology of GABA_A receptors in the adrenalremains unknown.  The binding of GABA and its analogues to GABA_A type receptors usually results in an increase in membrane permeability to chloride ions, typically resulting in membrane hyperpolarization and a reduction in membrane excitability.  However, in the adrenal chromaffin cell, GABA application elevates free [Ca²⁺]_i (Doroshenko, 1989) by evoking membrane depolarization (Busik et al., 1996) to the activation threshold of voltage-gated Ca²⁺ channels (Doroshenko, 1989).  This strongly suggests that Cl^- is accumulated above equilibrium in chromaffin cells.  Thus an increase in GABAergic activity might be expected to augment catecholamine release, and any modulation of GABAergic receptor properties by benzodiazepines or neuroactive steroids would thus be expected to alter adrenal catecholamine output.

In addition to GABA binding sites which are believed to be formed at the interface contributed by a and b subunits within the N-terminus (Amin and Weiss, 1993), a number of key neuromodulatory binding sites are thought to be present within the pentameric receptor-channel complex (Whiting, 1999a; 1999b).  Distinct high and low affinity binding sites for benzodiazepines are present within the pentameric GABA subunit complex (Walters et al., 2000), the high affinity site formed within the N-terminus by residues contributed by both a (Renard et al., 1999) and g (Kucken et al., 2000) subunits.  The modulation of GABA-mediated synaptic transmission thus underlies the pharmacotherapy of various neurological and psychiatric disorders including anxiety and epilepsy.

However at present no in vitro model for GABA_A receptor function exists for the adrenal.  Such a model of the adrenal chromaffin cell would allow the regulation of adrenal output to be studied, provided an equivalent GABA_A subunit expression profile and a functional, regulated catecholaminergic output may be demonstrated.  Tyndale et al. (1994) however reported no a subunit expression within the PC12 cell, the expression of which is believed prerequisite to the formation of functionalGABA_A receptor-channels.  Consistent with this finding, no functional GABA currents were recorded from PC12 cells in vitro (Hales and Tyndale, 1994).  Thus the dedifferentiation of chromaffin cells that occurs during transformation may result in a loss of functional GABA_A receptor expression in the PC12 cell.  Equally, if the plasticity necessary to express functional GABA_A receptors were retained by the PC12 cell, it could be established whether growth factors such as insulin either induce or modulate functional GABA_A receptor expression.  This is of particular interest in cases of hyperinsulinaemia or type II diabetes mellitus where insulin levels are elevated (Colagiuri & Brand Miller, 1997; Galassetti &, Davis, 2000). 

Conventional RT-PCR was thus used to determine which GABA_A subunits are expressed in the pheochromocytoma cell and adrenal gland.  It was determined that growth factors such as NGF, PACAP-38 and insulin induce or modulate the abundance of GABA_A receptor subunits in the PC12 cell, and that the GABA_A a and g subunit expression profile in the PC12 cell line was essentially consistent with that in the adrenal gland.  Moreover, the subunit expression profile is consistent with that believed necessary for the formation of functional receptor-channels, and the presence of colocalized GABA_A receptor subunits upon the surface of the PC12 cell was confirmed by immunocytochemistry.  These findings demonstrate the utility of the PC12 cell as a model for the regulation of gene expression and GABAergic function within the adrenal.

MATERIALS AND METHODS

Cell Culture (for RNA isolation)

PC12 cells (passage 7) were propagated in RPMI 1640 medium supplemented with 1% FCS (Fetal Calf Serum), 10mM HEPES, 1mM sodium pyruvate, 2mM L-glutamine, 10 U/ml Penicillin-Streptomycin (all purchased from Gibco).  For agonist stimulations for the purposes of isolating total cellular RNA, T75 flasks (Nunc) were pre-coated with either 500µl Engelbreth-Holm-Swarm (EHS) cell Matrigel matrix (frozen as 10% stocks in Dulbecco's Modified Eagles Medium (DMEM, Sigma) or 200µl collagen type I (3mg/ml in 1mM acetic acid, Sigma), and left to dry for 2 hours under sterile conditions before washing twice with 15ml DMEM (Gibco).  PC12 cells were then plated at 25-30% confluency in 1% FCS RPMI 1640 medium and left for 24 hours to adhere.  Cells were washed twice with 12ml of serum-free RPMI 1640 at 37°C containing 1% crystalline fatty acid-free BSA (Bovine Serum Albumin, Sigma), 10mM HEPES, 1mM sodium pyruvate, 2mM L-glutamine, 10 U/ml Penicillin-Streptomycin, and serum deprived for 2 hrs prior to the addition of 6ml of 1% BSA RPMI +/- NGF (50ng/ml, Gibco), PACAP-38 (100nM, Sigma) or insulin (100nM, Gibco).  Flasks were incubated for 10 hours at 37°C in a 5% CO₂ incubator prior to RNA isolation.

Cell Culture (for immunocytochemistry)

18mm² glass cover slips (3.24 cm²) were placed in 35mm Petri dishes (Nunc) and coated with 0.01% poly-D-lysine (Sigma) for 2 hours at room temperature, and then washed three times with Dulbecco's PBS (D-PBS, Gibco).  The cover slips were then pre-coated overnight with either 200ml laminin (0.05mg/ml [3mg/cm²]), 100ml EHS (10% in DMEM) or 10ml collagen type I (CI, 3mg/ml in 1mM  acetic acid [9mg/cm²], all Sigma) prior to three DMEM washes.  PC12 cells propagated in RPMI 1640 + 1% FCS were harvested, and the resulting cell suspension was taken centrifuged at 900 rpm for 3 min at 26°C and washed twice in 20ml RPMI + 1% FCS before resuspension in 5ml RPMI + 1% FCS, prior to cell dispersal by trituration.  The cell suspension was diluted to a final density of 3-400,000 cells/ml using a Neubauer Hemocytometer.  The resulting cell suspension (2ml) was seeded onto 35mm dishes containing matrix-coated cover slips by serological pipette and left to attach for 8 hrs.  Cells were then washed twice with 3ml pre-warmed (37°C) RPMI 1640 + 1% BSA, and serum deprived for 2 hrs prior to the addition of 2ml of 1% BSA RPMI ± NGF (50ng/ml, Gibco), PACAP-38 (100nM), insulin (100nM), GDNF (3nM, Gibco), IGF-I (10nM, Gibco) or Substance P (100nM, Sigma).  Dishes were stimulated with factors for 10 hours at 37°C prior to fixation.  

Cell harvesting and RNA preparation

Flasks were washed with 12ml DMEM and then incubated for 10 minutes at 37°C with 6ml DMEM containing 0.5% BSA and 6 units collagenase (Serva).  Cells were harvested by tapping and washing the flasks gently with 12 ml DMEM.  The cells were centrifuged at 1,200 rpm for 3min, before two successive centrifugation washes for 2 min at 1,000 rpm with 5ml Dulbecco’s Phosphate Buffered saline (D-PBS, Gibco).  The supernatent was completely aspirated and the pellet was lysed with 700µl buffer containing guanidinium isothiocyanate.  For the purposes of analysis, separate stimulations and isolations were performed with the same cell passage number on four consecutive days.  Isolation of total cellular RNA, including a 15 minute DNAse I treatment, was carried out using a Macherey-Nagel Nucleospin RNA II kit, according to manufacturer’s specifications, eluting the RNA with 50µl diethylpyrocarbonate (DEPC)-treated RNAse free water prior to storage at –80°C.

Preparation of RNA from adrenal gland

Intact adrenals were dissected from female Wistar rats after CO₂ asphyxiation.  Adrenals were immediately snap frozen in liquid nitrogen.  Adrenals were placed into 2ml collection tubes and homogenized on ice for 20-30s in 3ml tubes containing 700µl of lysis buffer containing 10µl/ml b-mercaptoethanol + 3µl of RNAse inhibitor (Roche).  Tissue was passed through a Qiagen tissue shredder kit to remove cellular debris and then total RNA was prepared from the cellular lysate with a Macherey-Nagel Nucleospin RNA II kit, including a 15 minute DNAse I treatment. 

Reverse transcription

The RNA concentration was determined through an A₂₆₀ measurement in a spectrophotometer and a volume equivalent to 1µg of RNA was taken for reverse transcription primed either by random hexamers or oligo-dT15 (Roche).  The desired volume of RNA was supplemented to a total volume of 17µl with DEPC-treated water (Gibco), and to this was added a reaction mix containing avian moloney virus (AMV) reversetranscriptase (RT) 2µl, 4µl of 10x AMV reaction buffer, 8µl of 25mM MgCl₂ stock, 4µl 10mM dNTP mix, 4µl primer (oligo-dT15 or random hexamers, 0.02 U/µl), 1µl RNAse inhibitor (all Roche) to a total reaction volume of 40µl.  The reverse transcription was performed in a Biometra T3 Thermocycler with the following temperature cycling program: 10 minutes at 25°C, 2 hours at 42°C, 5 minutes at 99°C to inactivate the AMV RT, with subsequent cooling to –3°C.

TaqMan Real-Time PCR

The cDNA of interest was added to a standardized real-time PCR reaction mixture containing standard components (see Galon et al., 2002).  Real-time PCR was conducted under identical conditions with 3µl cDNA per reaction to which was added (per reaction), 25µl of TaqManä mastermix (Applied Biosystems), 5µl of 10µM forward outer primer stock (final concentration 1µM), 5µl of 10µM reverse outer primer stock (1µM), 2.5µl of 2µM TaqManä 6-FAM-TAMRA inner primer (0.1µM) all targeted against the gene of interest; and 0.25µl of forward, reverse and inner TaqManäVIC primers directed against 18S transcripts to a total volume of 50µl with sterile water.  Reactions were performed in a 96 well reporter plate with TaqManä primer sets directed against different hypervariable non-homologous regions of the target gene (see below) so as to allow 32 reaction conditions to be performed in duplicate and parallel, following oligo-dT15 primed cDNA synthesis, under essentially identical reaction conditions with non-template (cDNA free) controls.  The TaqMan primer sequences used are as given in Table 1. 

Table 1. Real-time PCR primers used to quantify GABA_A subunit mRNA

* Not detected, N (N-terminus), M (region between transmembrane domains), C (C-terminus) 

18S rRNA "TaqManÔ" probes were obtained from Applied Biosystems (Foster City, CA).  The standard reaction cycle consisted of a 50°C step for 2 minutes, a second stage step to 95°C for 10 minutes, followed by 50 cycles consisting of 15 seconds at 95°C and 1 minute at 60°C to determine the cycle threshold (Ct) number for quantitative measurement.  Data were calculated as means ± SE from 4 replicate experiments.  Real-time TaqMan data was analyzed by a two-tailed Students’ paired T-test, assuming unequal variance.  Significance was assumed when P<0.05.

Conventional RT-PCR

Conventional PCR was performed using cDNA prepared by oligo-dT15 primed reverse transcription from rat brain poly A⁺ purified RNA (Stratagene) and total RNA derived from DNAse I treated isolates obtained from serum-starved PC12 cells as described above.  Conventional PCR primers directed against GABA_A subunits were those designed by Tyndale et al. (1994), and primer sequences are as previously published.  Rat brain mRNA or plasmid vectors containing GABA subunits (a1, g2) were used as positive controls for the presence of subunit and the viability of the primers used. 18S RT-PCR controls were incorporated as an internal control for RNA quality and quantity.  The 18S primers used  were "18S FWD" (sense) 5‘-GTA ACC CGT TGA ACC CCA TT-3’ and "18S REV" 5‘-CCA TCC AAT CGG TAG TAG CG-3’  [Accession No. K03432; Transcript +1718 - +1868, predicted amplicon size 150 bp]. PCR reactions were performed with a reaction mixture containing 2µl forward primer (0.4µmole), 2µl reverse primer (0.4µmole), 2µl cDNA, 25µl Taq Master mix (Roche) and 19µl sterile water.  Due to the wide range of Tm values reported for the original primers of Tyndale et al (1994) PCR products were divided into high or (low) Tm ranges for amplification, using the following reaction cycle time course: 2 min 94°C; {5 cycles of 94°C for 30 sec, 50°C (40°C) for 1 min, 72°C for 3 min}, followed by {35 cycles of 94°C for 30 sec, 45°C (35°C) for 1 min and 72°C for 3 min}, with a final extension at 72°C for 10 min prior to cooling to 4°C using a Perkin Elmer 9700 GeneAmp PCR Thermocycler.  PCR products were run on 2-3% agarose gels with Tris Acetate EDTA (TAE) buffer together with corresponding size markers (ladder VI or VIII, Roche) and bands were digitally imaged with a BioRad imaging system under UV illumination.  All RT-PCR amplifications were repeated at least three times for verification.

Immunocytochemistry

Dishes were aspirated and washed twice with 2ml of D-PBS prior to a 10 min fixation with 2ml of 4% paraformaldehyde in PBS at room temperature.  Cover slips were then washed 3 times for 5 min with 2ml of Tris Buffered Saline (TBS, 50mM Tris-HCl, 150mM NaCl, pH 7.4).  Immunolabelling was performed following standard methodologies  (Albers et al. 1999).  For experiments involving colabelling with rhodamine-conjugated phalloidin (RP), cells were incubated in the dark for 2 hours on a horizontal plane rocking tray at a frequency of 100 translations per min (TPM) after addition of 0.5ml of RP at a dilution of 1:1000 in TBS + 1% BSA after preparing the RP stock in methanol according to manufacturer's specifications (Molecular Probes).  Primary antibodies (0.5ml) were added at a dilution of 1:1000 in TBS + 1% BSA.  Affinity purified rabbit anti-rat g2 polyclonal antibodies  (N-terminus) were obtained (Alpha Diagnostic International, GAG21-A), and affinity purified goat anti-rat b2 (N-terminus) polyclonal antibodies (Santa Cruz Biotechnology).  Specificity controls were performed omitting primary antibody and by the use of a blocking antigenic peptide.  Incubations were performed at 4°C overnight after an hour on a rocking tray.  All cover slips were then washed three times with TBS and the first fluorescently-labelled secondary antibody was added (0.5ml) at a dilution of 1:500 in TBS + 1% BSA. All secondary antibodies used were either rhodamine or fluorescein conjugated anti-rabbit or anti-goat polyclonals (Abcam, Cambridge, UK).  .Cover slips were placed upon the rocking tray for 1 hour at 100 TPM for 2 hrs wrapped in foil.  Cover slips were mounted with a drop of Anti-fade reagent (Molecular Probes) and sealed with nail varnish before viewing with a BioRad Confocal scanning microscope wherein sensitivites were held constant to gain an impression of relative immunofluorescence intensities.  As the presence of GABA_A receptor subunits and their cellular distribution was an unknown, the cell was counterlabelled with Rhodamine-conjugated phalloidin (Molecular Probes) to outline the cytoskeleton.

Level of 18S polyadenylation is stable

18S rRNA was determined to be polyadenylated by the reverse transcriptase-polymerase chain reaction (RT-PCR) cloning and sequencing of 18S transcript using oligo-dT15 as the reverse primer.  It was then demonstrated that the relative quantity of the polyadenylated form of 18S, relative to the total pool of 18S rRNA, was invariant by measuring and comparing the Ct values obtained with the internal 18S standard following reverse transcriptions performed with identical concentrations of random hexamer and oligo-dT15 from 1µg of total RNA from the same series of agonist stimulations.  From the data given in table 2 it is demonstrated that the mean Ct  and DCt values (Ct_dT15 – Ct_random) is invariant, therefore demonstrating that levels of polyadenylation are stable.

RESULTS

Determination of GABA_A subunit expression profile in the adrenal

A conventional RT-PCR analysis was performed from adrenal medullae together with a visceral adipose control, using oligo-dT15 and established intron-spanning primers  (Tyndale et al., 1994).   Poly A⁺ purified rat brain mRNA was employed as an internal positive control for primer efficacy.  The predicted sizes of amplicons were contrasted with those observed, both for rat brain and adrenal, to determine whether a given a, b or g subunit is expressed (tables 3& 4).  As the reverse transcription was oligo-dT15 primed, and the subsequent PCR was performed with both intron-spanning primers after a DNAseI treatment step during the preparation of total RNA, the possibility of genomic DNA contamination is excluded.

The comparative expression profile of a subunits 1-6 in adrenal and rat brain is presented in Fig.1, and the exonic regions amplified, and the expected and actual amplicon sizes are presented in table3.  Whilst the rat brain expression profile correlates closely with that predicted, with the exception of additional amplicons detected for a3i (350 bp), a4ii (590 bp) and a 1,300 bp product detected for a3ii.  In the adrenal medulla PCR products correlating to a3i and a4i were detected, if the truncated a3i PCR product observed, which was also amplified from rat brain mRNA, does not represent a non-specific amplification product.  All RT-PCR amplifications were repeated at least three times for verification.

A similar analysis is presented in table 2 for the detection of b and g subunits expressed in rat brain and the adrenal gland (Fig.1c and 1d).  For rat brain, amplicon sizes correlate closely with those predicted, with the exception of shorter amplicons than those expected for b3 and g1.  Additional PCR products were detected in rat brain cDNA with b1 (150 bp) and g1 (700 bp) primers.  No d subunit was detected using the established primer set (Tyndale et al., 1994).  Despite parallel internal PCR positive controls (rat brain cDNA), no b subunit expression was positively identified in the rat adrenal, despite the knowledge that functional GABA_A receptor-channels are expressed within adrenal medullary chromaffin cells (Peters et al., 1989).  Although an amplicon approximating to the size expected for g3 was detected, the highest apparent expression level detected was for a 100 bp product generated using g2 sequence specific primers.  The visceral adipose control showed no PCR products with the exception of an unexpected amplicon for b3 primers (not shown).  In conclusion the adrenal medulla appears to express a3, a4, g2 and g3 subunits, but no b subunits could be detected.

Determination of GABA_A subunit expression profiles in agonist-stimulated PC12 cells

PC12 cells cultured upon the secreted Engelbreth-Holm-Swarm tumor cell matrix (EHS/Matrigel)  were stimulated with the addition of insulin (100nM), Nerve Growth Factor (NGF, 50ng/ml), Pituitary Adenylate Cyclase Activating Polypeptide (PACAP-38, 100nM), Glial-Derived Neurotrophic Factor (GDNF, 3nM) or under growth factor free control conditions for 10 hrs in the presence of either1% Fetal Calf Serum (FCS) or 1% Bovine Serum Albumin (BSA) in attempt to determine whether GABA_A subunit expression could either be induced or modulated.  The data for the PCR profiling of a subunit expression in the PC12 cell are illustrated in Fig.2 and presented in table 3.

RT-PCR analysis indicates that the PC12 cell, as shown for the adrenal, expresses a3 and a4 subunits, and that these may be differentially regulated by growth factors.  In contrast to the adrenal, there is apparent expression of the exon spanning region 6-8 (E6-8) of the a3 subunit (designated ii), along with a number of potential splice variants.  Similarly a product corresponding to a4ii (exons 1-4) is detected in the PC12 cell, which is again absent from the adrenal.  Thus, although the PC12 cell may express the same a subunit isoforms as the adrenal, there may be functional differences generated through alternative splicing.  There is also a notable difference between profiles generated in the presence and absence of 1% serum, providing a possible explanation for the inability of Tyndale et al. (1994) to detect a subunit expression after culturing PC12 cells with 10% serum.  

A similar characterization of the b and g subunit expression profiles in response to a panel of growth factors was performed, and the results are shown in Fig.3 and presented in table 4.  In contrast to the findings for the adrenal, conventional RT-PCR suggests that the PC12 cell appears to express both b2 and b3subunits.  Additionally, there is the possible expression of a truncated (150 bp) splice variant of b1.  Expression of different b subunits appeared to be highly variable, dependent upon the presence of serum and the growth factor present.  In contrast g1 expression appeared robust, with only a trace of g3 expression.  A truncated 150 bp g2 variant was detected, as observed previously in the adrenal.  Although the expression levels of the various GABA_A receptor subunits appear to be modulated by growth factors, both the affirmation of the GABA_A subunit expression profiles, and the quantification of agonist-induced changes in expression levels obligate real-time RT-PCR measurements.

Quantitative measurement of GABA_A subunit expression

Real-time PCR primers were designed to quantify agonist-induced changes in the relative expression levels of individual subunit isoforms, and also to validate the subunit profile in the PC12 cell as determined by conventional RT-PCR.  Serum-deprived PC12 cells were stimulated with 100nM insulin, 50ng/ml NGF, 100nM PACAP-38 or under control conditions for 10 hours in the presence of either collagen type I (CI) or EHS.  Real-time RT-PCR measurements were performed using primers directed against two distinct non-conserved regions of the GABA subunits of interest where possible.  For these measurements a3, a4, b2, g1 and g2 subunits were selected for real-time quantification of mRNA levels following their detection by conventional RT-PCR, with the exception of b1, which was taken as an unknown.

Effect of growth factors upon a subunit expression

Figure 4 shows the differential effects of growth factors upon a3 and a4 subunit mRNA levels.  NGF selectively increases the abundance of a3, but not a4 mRNA.  Insulin differentially modulates subunit expression profiles, marginally increasing a3abundance in the presence of EHS (Fig.4), but not collagen I (CI), and conversely decreasing the abundance of a4 in the presence of CI.  Whilst insulin decreased a4 levels in the presence of EHS, this effect was not significantly different from that of EHS alone.  In contrast, PACAP-38 had no significant long-term action upon a subunit abundance.  Results for real-time PCR measurements are summarized qualitatively in table 5. 

Effect of growth factors upon b subunit expression

The same rationale was applied to the regulation of b subunit expression, excepting that b1 expression had not been positively determined by conventional RT-PCR analysis.  Of the two regions targeted sequences, one central and one within the C-terminus (see methods), only the C-terminal region was detected.  Both NGF and PACAP-38 substantially augmented the abundance of the C-terminal region of b1, irrespective of the extracellular matrix employed.  In contrast, of these two growth factors, only PACAP-38 increased the abundance of the b2 subunit, and then only in the presence of CI.  Conversely insulin reduced b1 mRNA levels in the presence of EHS, but not CI, although b2 mRNA levels were diminished irrespective of the identity of the extracellular matrix.

Effect of growth factors upon g subunit expression

As non-homologous regions within the g1 and g2 subunits were smaller and less frequent, only one real-time PCR probe was designed per subunit.  PACAP-38 increased both g1 and g2 subunit mRNA abundance in the presence of CI, but not EHS, an effect which could not be readily attributed to a direct action of EHS, by reason of the augmentation of mRNA levels with EHS alone.  NGF increased g2, but not g1 mRNA levels in PC12 cells cultured upon CI, although it potently inhibited g1, but not g2 expression in cells cultured upon EHS.  This is indicative of a polarizing action of matrix factors upon the balance of g subunit mRNA expression.  Insulin displayed similarly complex actions upon the g subunit expression profile,  increasing g1 and depressing g2 levels in cells cultured upon CI.   In contrast, insulin dramatically augmented the expression of the g2, but not the g1 subuniton EHS.

A determination of the relative abundance of each subunit was attempted by a direct comparison of the averaged mean DCt values derived from each subunit, the DCt values comparing abundance relative to that of the 18S rRNA internal standard.  The values were taken for the agonist-free CI control, and ranked in order of the ascending mean values (i.e. decreasing relative abundance) of the DCt values.  The relative abundance of mRNA species was, in descending order, a3(DCt = 22.8) = a4 (22.8) = b1 (23.3) ≥ b2 (24.3) ≥ g1 (24.9) > g2 (25.7).  Given that subunits co-assemble, with the g subunit being stoichiometrically the least abundant (after the classical pentameric arrangement, a₂b₂g), this appears superfically to be consistent.  As the values presented are for agonist-free controls cultured upon CI, these findings suggest that the ratio of a3:a4; b1:b2 and g1:g2 mRNA levels would be expected to vary under the influence of growth factors.

Immunocytochemistry reveals surface labelling of b and g  subunits

From the observation that PC12 cells express mRNAs for a, b and g  subunits, it does not necessarily follow that the component subunits are translated and coassembled to produce GABA_A receptor-channels that are transported to the cell membrane.  Antibodies raised against the N-termini of b2 and g2 subunits were employed to determine whether these subunits coassemble, and whether they can be detected at the PC12 cell membrane.

Serum-deprived PC12 cells were fixed with 4% paraformaldehyde after 10 hours of stimulation in the presence or absence of growth factor.  Triton X-100 treatment was excluded so as to better emphasize surface labelling.  Figure 5 shows that a widespread immunofluorescence is observed at the PC12 cell membrane after labelling for the b2 subunit, thereby illustrating the classical neurite outgrowth induced by both NGF (Fig.5b) and PACAP-38 (Fig.5d), and the comparatively low levels of neurite outgrowth observed in PC12 cells stimulated with insulin in the presence of CI.  The neurite outgrowth induced by EHS in the absence of NGF is also shown (Fig.5e), together with CI control (Fig.5a).

Although all images were captured at equivalent camera sensitivities and gain, these provide a qualititative estimation of the level of protein expression.  However, b2  fluorescence intensity appeared consistently lower after treatment with PACAP-38 (Fig.5d), especially in PC12 cells cultured upon ECM (not shown).  The pattern of fluorescence labelling seen with antibodies directed against b2 and g2 subunits was similar in intensity and distribution, except that g2 labelling in the presence of ECM was evidently less intense (Fig.6e), especially following stimulation with PACAP-38 (Fig.6d).  However, measurements obtained by quantitative Western blotting and immunofluorescence will be required before the effects of growth factors upon GABA_A receptor subunit expression can be reliably be determined at a translational and post-translational level.

In order to further address antibody specificity and to verify that subunit coassembly had taken place, colabelling of g2 and b2 subunits was performed using affinity-purified rhodamine and fluorescein-conjugated secondary antibodies respectively.  Figure 7 reveals the extent of surface co-labelling of  g2 and b2 subunits following stimulation with either NGF (Fig.7a) or insulin (Fig.7b).  Cells which had been double-labelled were selected where GABA_A receptor expression was demonstrably punctate and asymmetrical, so that any coincident immunolabelling overlapped to a degree that would exclude coincidence.  Therefore if there were non-specific labeling, or an absence of coassembly of b2 (green) and g2 (red) subunits then double labeling (yellow) would not be evident or predominant.  Cells with punctate and polarized receptor clustering were selected to best illustrate this control.  If labeling were universal for both antibodies over the surface of the cell, then no conclusion could be drawn as regards colabelling, as both antibody pairs might theoretically bind to different non-specific antigens with a universal distribution, whereas the chances that two unrelated antigens would have the same punctate and polarized distribution seems unlikely.

DISCUSSION

The PC12 cell as a molecular model for neuroendocrine cell function

The adrenal is often referred to as the master stress organ and forms part of the integrating Hypothalmic-Pituitary-Adrenal (HPA) axis and the autonomic nervous system.  There is also an intimate functional interaction between the adrenal cortex and medulla, including a neuroendocrine regulation of the adrenal cortex by the adrenal medullary chromaffin cells (Bornstein et al., 1990, Bornstein et al., 2000).  Adrenal chromaffin cells, and in particular the PC12 pheochromocytoma cell line have been used not only as a tool by which to study adrenomedullary function and catecholamine biosynthesis, but serve as an established model by which to understand secretion and neuronal plasticity.  This is the first comprehensive molecular study of the adrenal GABA_A receptor system.  Our data clearly demonstrates the expression of GABA_A receptor subunits in the adrenal gland and PC12 cell line, and the colocalized surface expression of b and g subunits in the PC12 cell line.

The apparent inability of Tyndale and colleagues (1994) to observe a subunit expression was perhaps explained by their use of 10% serum for PC12 cell culture.  This combined with the absence of differentiating growth and extracellular matrix factors would create hyperproliferative culture conditions and result, as observed, in markedly different expression patterns for GABA_A and tyrosine hydroxylase (Brandt et al.,unpublished observations).  These conditions would strongly promote proliferation, rather than creating the differentiating conditions that might promote GABA_A receptor subunit gene expression and function.

Relevance of altered adrenal GABA subunit expression profiles by growth factors

The observed changes in the relative expression levels of mRNAs for a3 and a4 subunits in response to NGF and insulin may have profound consequences for the functional properties of GABA_A receptors.  For instance, in conjunction with the g2 subunit, the a subunit is a major determinant of the properties of the high affinity BDZ binding site present within the GABA_A receptor (Walters et al., 2000).  It is further known that BDZs such as alprazolam mediate their known ‘anti-panic’ effects through a remodelling of GABAergic transmission in the pons-medulla region.  This remodelling is characterized by an increase in the levels of expression of transcripts for the a3, b1 and g2 subunits which coassemble within this region, and it is suggested that a proportional increase in the number of this GABA_A receptor subtype parallels the observed changes in behaviour (Tanay et al., 2001).  Similarly an altered balance of a3 and a4 subunit expression in response to insulin might have potential clinical consequences in cases of hyperinsulinaemia and diabetes mellitus in which insulin levels are elevated. 

Despite the apparent complexity in the regulation of subunit expression profiles in response to growth factors, some clear patterns emerge.  PACAP-38 increases the apparent expression of both b and g, but not a subunits, in the presence of CI.  This suggests that CI is a permissive substrate for the actions of PACAP-38, whilst the tumor derived EHS matrix was occlusive.  Whilst the actions of PACAP-38 appear to be subunit rather than isoform specific, NGF and insulin are more discriminating in their actions upon subunit isoforms.  NGF selectively increases the abundance of a3, b1 and g2 subunit mRNAs, regardless of the extracellular matrix employed, although NGF does show some dependence upon the composition of the extracellular matrix in the modulation of g1 expression.  In contrast, insulin apparently exerts a polarizing action upon the abundance of subunit isoforms, having diametrically opposite effects depending upon the matrix components present.  Insulin may thus act to change the balance of subunit composition and modulate the properties of GABA_A receptors in this way.  Thus growth and extracellular matrix factors may act, either synergistically or antagonistically, to differentially modulate levels of subunit isoform expression, thereby altering the functional characteristics of GABA_A receptor channels.

Is a functional b subunit necessary for the formation of functional GABA_A  receptors?

The apparent absence of a detected b subunit in the rat adrenal gland raises the possibility that a3 or a4 subunits may functionally coassemble with g subunits, despite the widely held doctrine that the incorporation of b subunits is a prerequisite for the formation of functional GABA_A receptors. 

The functional heterogeneity of GABA_A receptors is achieved via combinatorial assembly from a repertoire of at least 18 subunits, excluding potential splice variants, to generate receptor subtypes with functionally distinct properties (Whiting et al., 1999a; 1999b).  Assuming that two a and two b subunits are required, and that the 'fifth' subunit may be a d, e, q, g or p (or an additional a or b) subunit, once again excluding variants generated through alternative splicing, over 10,000 potential subunit combinations may exist.  However, only certain combinations have thus far been identified in situ or been shown to functionally co-assemble within heterologous expression systems.  Furthermore, the expression pattern of receptor subtypes has been shown to be neuron-specific both in adult rat brain and in vitro.  

For example, single-cell RT-PCR analysis of the GABA_A receptor subunit composition in neurons from the reticular nucleus (nRt) and the ventrobasal (VB) complex thalamic nuclei revealed the presence of functional channels in nRt neurons in the apparent absence of a b subunit, adding further credibility to the hypothesis that a and g subunits may functionally coassemble (Browne et al., 2001).

Immunohistochemical studies using confocal laser microscopy immunofluorescence staining reveal an extraordinary heterogeneity in the distribution of GABA_A receptor subunits.  Fritschy and Mohler (1995) identified seven major subunit combinations within defined neurons, the most prevalent of which was a1b2b3g2, a combination which was often found associated with an additional a2, a3 or d subunit.  The triplets a2b2b3g2 and a3b2b3g2 and a5b2b3g2 were also identified within discrete cell populations.  Importantly, five combinations were identified which apparently lacked b2 and b3 subunits, including one which was devoid of the b2 subunit, namely a1a2g2, a2g2, a3g2, a2a3g 2 and a2a5d (Fritschy and Mohler, 1995).  This provides further evidence suggesting the coassembly of a3 and g2 subunits without b, although it is interesting to note that the g2 subunit expressed within the adrenal and PC12 cell appeared to be a truncated variant.

In a similar vein, Qian and Ripps (1999) reported that r1B subunits of the GABA_C receptor functionally coassembled with g2 subunits to form hetero-oligomeric receptors in oocytes.  These recombinant r1Bg2 receptors displayed properties more akin to those of native GABA_C currents than those attributed to r1B homo-oligomeric receptor-channels.

Modulation of GABA receptor function and expression by insulin

Only recently has the extent to which both insulin and the insulin receptor (IR) modulate CNS function become to be appreciated (Zhao and Alkon, 2001).  The IR is abundantly expressed in several specific brain regions that regulate fundamental behaviours such as food intake, reproduction and cognition, and it is also expressed peripherally upon neuroendocrine cells, including the PC12 cell line (Ohmichi et al., 1993).  However IRs expressed within the periphery exhibit fundamental differences in both their structure and function to those present within the CNS.

In addition to that produced within the endocrine pancreas, insulin is also synthesised locally within the brain, where it has been implicated in associative learning and memory formation (Zhao and Alkon, 2001).  Insulin may modulate neurotransmitter release processes at various presynaptic terminals and the activities of both excitatory and inhibitory postsynaptic receptors, including NMDA and GABA_A type receptors (Zhao and Alkon, 2001).

As energy and glucose homeostasis are under the control of both CNS centres and the HPA axis, the neurotransmitter GABA is of pivotal importance as a regulator of these centres.  One notable hypothesis implicates an alteration of the biochemical structure and composition of GABA receptors in the aetiology of the insulin resistance syndrome (Blasi and Jeanrenaud, 1993).  Similarly, putative alterations in the subunit composition of GABA_A receptors within the adrenal, perhaps similar to those demonstrated here, may significantly alter GABA_A function within the HPA axis.  Further evidence for a role for insulin in the modulation of the sympathetic nervous system comes from recent work demonstrating that insulin depresses expression of dopamine b hydroxylase and PNMT within the PC12 cell (Walters et al., submitted).

An increase in the number of active postsynaptic receptors is thought to be the most efficient way of promoting synaptic efficacy.  Hence the significance of the observation that insulin causes the GABA_A receptor to translocate rapidly from the intracellular compartment to the plasma membrane in transfected cell lines (MacDonald et al, 1997).  This relocation of GABA_A receptors requires the b2 subunit of the GABA_A receptor, and serves to increase both the expression level and number of functional GABA_A receptors incorporated both postsynaptically and upon the dendrites (MacDonald et al, 1997). 

Thus insulin may play a fundamental role in the modulation of synaptic plasticity, and the observations from the PC12 cell further suggest a role for insulin in modulating GABA_A receptor expression and function.  The data further critically suggests that insulin alters the balance of a,b and g subunit isoform expression levels, potentially adding yet another dimension to insulin-induced plasticity within the CNS and neuroendocrine systems.

A central role for PACAP in the modulation of adrenal output

PACAP is a sympathoadrenal neurotransmitter, related to secretin and VIP, which is involved in the regulation of catecholamine production and glucohomeostasis.  There is accumulating evidence to suggest that PACAP may function as a transmitter at the adrenomedullary synapse, as it is present in nerve terminals at all mouse adrenomedullary cholinergic synapses (Ogawa et al., 1984).  Targeted disruption of the PACAP gene chronically enhances insulin-induced hypoglycemia, which is lethal at higher insulin concentrations.  The most likely explanation is that, in the absence of PACAP, there is a long-term impairment of epinephrine secretion as a direct consequence of the diminished levels of tyrosine hydroxylase (TH) activity seen following insulin induced hypoglycemia, resulting in a depletion of adrenomedullary epinephrine stores (Hamelink et al., 2002).  Hence it was proposed that PACAP couples epinephrine biosynthesis to its secretion during metabolic stress, PACAP thus functionin