Distribution of �-Aminobutyric AcidReceptors in Cultured Adrenergic andNoradrenergic Bovine Chromaffin Cells

Enrique Castro,1 Marıa Pilar Gonzalez,2 and Marıa Jesus Oset-Gasque2*1Department of Biochemistry, Molecular Biology and Physiology, Faculty of Medicine and Health Sciences,University of Las Palmas de Gran Canaria, Las Palmas, Spain2Department of Biochemistry and Molecular Biology, Faculty of Pharmacy, Complutense University of Madrid,Madrid, Spain

Fluorescence imaging techniques for recording cytosolic[Ca2�]i from single chromaffin cells were used to char-acterize and discriminate between cell subpopulationscontaining �-aminobutyric acid (GABA)A and GABAB re-ceptor subtypes. By combining this methodology withthe immunoidentification of individual chromaffin cellsusing specific antibodies against tyrosine hydroxylase(TH), phenyl-etanolamine-N-methyl transferase (PNMT),and glutamic acid decarboxylase (GAD) linked to differ-ent fluorescent probes, we have been able to ascribesingle-cell calcium responses to identified adrenergicand noradrenergic chromaffin cells. GAD enzyme ispresent in 30% of the chromaffin cell population, locatedprimarily in adrenergic cells; 86% of GAD� cells werealso PNMT�. GAD expression was not correlated withthe presence of GABA receptors. GABA-responsive cellswere found with equal frequency in the GAD� and GAD–

groups. However, the expression of GABA receptors wascorrelated with the adrenergic phenotype. [Ca2�]i re-sponses to GABA were found more frequently in adren-ergic than in noradrenergic cells. GABAA receptors aremore evenly distributed; about 90% of GABA-responsivecells have them. GABAB receptors have a more restricteddistribution (present in 45% of responding cells). Thecoexpression of both GABAA and GABAB subtypes is therule; only a minor subpopulation (about 12%) displaysexclusively GABAB receptors. GABA receptor subtypesare distributed in a similar way when chromaffin cells areseparated according to GAD�/GAD– or PNMT�/PNMT–

classifications, with only minor differences. These dataindicate that the intrinsic GABAergic system in the adre-nal medulla is not designed as a paracrine model inwhich a group of cells specializes in transmitter synthesisand a different group serves as a specific target.© 2002 Wiley-Liss, Inc.

Key words: �-aminobutyric acid; glutamic acid decar-boxylase; immunofluorescence; adrenal gland

�-Aminobutyric acid (GABA), the main inhibitoryneurotransmitter in the CNS, has been shown to be in-

volved in the regulation of catecholamine (CA) secretionfrom bovine adrenomedullary chromaffin cells (for reviewsee Parramon et al., 1995a) in a neuromodulatory way.GABA is a compound endogenous to the adrenal medulla.GABA is synthesised in and released from chromaffin cells(Kataoka et al., 1984; Oset-Gasque et al., 1990b). Inaddition to this source of GABA, a network of GABAer-gic afferent fibers incoming to adrenomedullary tissue hasbeen described for several species (Kataoka et al., 1984;Ahonen et al., 1989; Oomori et al., 1993; Iwasa et al.,1999). GABA receptors in adrenomedullary tissue physi-ologically modulate secretion from adrenal gland stimu-lated in situ (Kataoka et al., 1986; Fujimoto et al., 1987).The modulation of CA secretion in chromaffin cells re-quires a fine control of Ca2� entry through voltage-dependent Ca2� channels. This neuromodulatory func-tion is achieved through GABA binding to both types ofGABA receptors, i.e., GABAA and GABAB, whose pres-ence in chromaffin cell membranes has been demonstrated(Castro et al., 1988, 1989). Ionotropic GABAA receptorsmodulate CA secretion by a mechanism dependent on themembrane potential, in which the intermediate position ofthe equilibrium potential for chloride plays a pivotal role(Kitayama et al., 1990; Gonzalez et al., 1992): Activationof GABAA receptors depolarizes, per se, the chromaffincells, while dampening the depolarizing action elicited byother excitatory transmitters. On the other hand, metabo-tropic GABAB receptors regulate CA secretion by a mech-anism involving cAMP and Ca2� as second messengers(Oset-Gasque et al., 1993; Parramon et al., 1995b).

Contract grant sponsor: Fondo de Investigaciones Sanitarias, Spanish Min-istry of Health; Contract grant number: FIS 95/1537; Contract grantnumber: FIS 98/0587.

*Correspondence to: Maria Jesus Oset-Gasque, Department of Biochem-istry and Molecular Biology, Faculty of Pharmacy, Complutense Universityof Madrid, E-28040 Madrid, Spain. E-mail: mjoset@farm.ucm.es

Received 18 March 2002; Revised 14 August 2002; Accepted 3 September2002

Journal of Neuroscience Research 71:375–382 (2003)

© 2002 Wiley-Liss, Inc.

Several different neurotransmitter and neuromodu-lators have been implicated in the regulation of the neu-rosecretory function of adrenal medulla. GABA adds tothis great variety of compounds. This diversity of agentsmakes it difficult to understand the local regulation of CAsecretion within the adrenal medulla. Part of this complexvariety is probably addressed to control secretion fromadrenergic and noradrenergic chromaffin cell subpopula-tions within the medulla. For instance, histamine andangiotensin II receptors are distributed mainly to adrener-gic chromaffin cells (Choi et al., 1993; Nunez et al., 1995),whereas ionotropic P2X receptors for ATP are preferen-tially expressed in noradrenergic cells (Castro et al., 1995;Mateo et al., 1997). Thus, the question is whether theseGABA receptors are distributed homogeneously in thewhole chromaffin cell populations or whether these re-ceptors have different roles in the regulation of adrenaline(ADR) and noradrenaline (NA)-secreting cells. This kindof heterogeneity is needed to account for the differentialrelease of ADR and NA that occurs physiologically, insome species, in response to natural stimuli, such as hem-orrhage or hypoxia (Feuerstein and Gutman, 1971).

We have used fluorescence imaging techniques forthe recording of cytosolic free calcium from single chro-maffin cells to characterize and discriminate between thecell subpopulations that contain GABAA and GABABreceptor subtypes. By combining this methodology withthe immunolabeling of individual chromaffin cells fortyrosine hydroxylase (TH), N-phenyl-ethanolaminemethyl transferase (PNMT), and glutamic acid decarbox-ylase (GAD; the enzyme responsible for GABA synthesis),we have been able to assign single-cell calcium responsesto identified chromaffin cells, thus clarifying the differentroles of GABA receptors in the function of the adrenergicand noradrenergic chromaffin cells as well as in GABA(GAD�)-producing cells.

MATERIALS AND METHODS

Cell Culture

Bovine chromaffin cells were obtained essentially as de-scribed previously (Castro et al., 1995; Oset-Gasque et al.,1998). Briefly, after digestion by retrograde perfusion with col-lagenase, the medullary tissue was collected and disaggregatedmechanically. Chromaffin cells were purified in a Percoll densitygradient. Routine preparations yielded �95% chromaffin cells,as assessed by neutral red staining. The cells were plated onto16 mm round coverslips coated with poly-L-lysine and placedinside 35 mm Petri dishes. The cells were plated at a density of106 cells/dish. For combined Ca2� imaging/immunolabelingexperiments, Bellco coverslips engraved with a square grid wereused (one coverslip per dish). Chromaffin cells were maintainedunder a 5% CO2/95% air humidified atmosphere at 37°C in a1:1 mixture of DMEM/F-12 medium supplemented with15 mM HEPES, 25 mM NaHCO3, 5% inactivated fetal calfserum (FCS), penicillin, and streptomycin. Cells were typicallyused between days 2 and 6 after plating. Routine culturescontained 65–70% adrenergic cells, as identified by PNMTimmunolabeling (see Results).

Measurement of [Ca2�]i by Video Imaging

The [Ca2�]i was measured with the fluorescent probeFura-2. The coverslips containing the cells were washed in aphysiological saline containing (in mM): 120 NaCl, 5 KCl,25 NaHCO3, 2 CaCl2, 1 MgCl2, and 10 glucose. The solutionwas constantly gassed with 95% O2/5% CO2, for a final pH of7.4. Chromaffin cells were loaded with 2.5 �M Fura-2/AM for30–45 min at 37°C in this medium supplemented with 1%bovine serum albumin (BSA). After washing, the coverslip wasglued to the bottom of a small (�100 �l) perifusion chamberand placed in the stage of a Nikon Diaphot TMD microscope.The cells were continuously perifused (�1.5 ml/min) withgassed physiological saline. The drugs were applied for shortperiods (15–60 sec), dissolved in the perifusion medium, withthe aid of a four-way stopcock.

The fluorescence changes were recorded with a multipleexcitation MagiCal imaging system (Applied Imaging U.K.).The system was essentially the same as we have describedpreviously (Oset-Gasque et al., 1998). Chromaffin cells werealternately excited at 340 nm and 380 nm by means of a steppingfilter wheel and the epifluorescence optics of the microscope.Emitted fluorescence collected with a �20 fluor objective wasdriven to a Photonics Science SIT camera after passing througha 510 nm bandpass filter. Eight frames (�100 msec exposureeach) were averaged to produce an image. Alternating excitationand image capture and processing were controlled by a singleprocessor in the MagiCal system. Image analysis was performedwith the MagiCal software and custom-made programs (devel-oped by E.C.; details available upon request). Essentially, back-ground fluorescence at each wavelength (obtained from a fielddevoid of cells in each coverslip) was subtracted, and fluores-cence images were ratioed on a pixel-by-pixel basis. Ratio datawere stored as eight-bit pseudocolored images. Rmax, Rmin, and� parameters for Grynkiewicz-type calibration into [Ca2�]iwere obtained from the averaged fluorescence of droplets ofFura-2 at zero and saturating Ca2� (Castro et al., 1995) imagedin with the same system. However, simple transformation ofimage data according to the Grynkiewicz equation may intro-duce severe pitfalls (unless performed on a pixel-by-pixel basiswith cell-specific parameters) and was avoided. Raw ratio dataare presented routinely as a more direct result of the fluorescenceexperiments and usually referred as calcium data, because ratioincreases as [Ca2�]i rises. A contour was drawn around each cellin a field and the averaged ratio value of pixels inside eachcontour evaluated at each time point, to obtain ratio vs. timeplots for all cells. For comparative purposes, [Ca2�]i valuesderived from averaged ratio data using averaged parameters arequoted in the text, in addition to raw ratio data.

Immunocytochemical Identification ofChromaffin Cells

To study the distribution of GAD, the cells were doublyimmunostained with antibodies against GAD (mouse monoclo-nal anti-GAD) and either TH or PNMT (rabbit polyclonalanti-TH and anti-PNMT). Briefly, cells were fixed for 2 min inan ice-cold 1:1 acetone-methanol mixture (v/v). After blockingin phosphate-buffered saline (PBS) with 3% BSA/0.1% Tx-100for 1 hr, the preparation was incubated with primary antibodies

376 Castro et al.

(mouse anti-GAD 1:100, rabbit anti-TH or anti-PNMT 1:500)for 1 hr at room temperature. After washes in 0.1% Tx-100,secondary antibodies were added (FITC-labeled goat anti-mouse IgG and TRITC-labeled goat anti-rabbit IgG at 1:200)and incubated for another 1 hr. Immunocytochemical controlswere prepared as described, omitting the primary antibodies andincubating with 3% BSA instead. Thus, they test nonspecificbinding of labelled second antibodies to sample structures.

To identify the cells responding to GABA, triple labelingswith mouse anti-GAD, rabbit anti-PNMT, and mouse anti-TH(monoclonal) were performed on the same coverslips used for[Ca2�]i measurements (Bellco coverslips with locating grid).Once the [Ca2�]i experiments had been performed, the cellswere fixed 1:1 acetone-methanol for 2 min at 4°C, followed byfreezing and storage at –20°C until processing. For the immu-nofluorescence assay, the coverslips were thawed by placingthem in 50% acetone:50% methanol cooled at –20°C and wereallowed to warm at room temperature. The coverslips were firstprocessed as for double labeling using anti-GAD and anti-PNMT primary antibodies and FITC and TRITC secondaryantibodies. After washing, the preparation was fixed again incold 1:1 acetone:methanol to denaturalize the already boundmouse anti-GAD primary antibody and processed for labellingwith mouse anti-TH (1:50) and AMCA-conjugated anti-mouseIgG (1:100). The coverslips were mounted in 50% glycerol inPBS containing 2.5% DABCO and digital images taken with thesame MagiCal system used for Ca2� experiments.

Statistical Analysis

Fluorimetric data are expressed as mean � SEM ratio in-creases over basal (net increase). Statistical comparisons betweenmeans were performed with a Student’s t-test. Immunocytochem-ical data are expressed as number of cells counted and as propor-tion � SD of labeled cells vs. total cells. SD were calculatedassuming a binomial distribution of counted cells, employed also totest differences between proportions. Comparisons between pro-portions in 2 � 2 tables were made with the 2 test.

Materials

Fura2/AM was from Molecular Probes (Eugene, OR),and collagenase was from Boehringer Mannheim S.A. (Barce-lona, Spain). Dulbecco’s modified Eagle’s medium (DMEM)and FCS were purchased from Gibco BRL (U.K.). Antibioticswere supplied by Flow Laboratories (Irvine, CA). (–)-Baclofen,CGP 35348, and 3-APPA were a gift from Ciba-Geigy (Basel,Switzerland). GABA, muscimol, BSA, and anti-IgG FITC andTRITC secondary antibodies were obtained from Sigma (Ma-drid, Spain). Rabbit polyclonals anti-TH and rabbit anti-PNMTwere from Eugene Technology International. Mouse monoclo-nal anti-GAD IgG and mouse monoclonal anti-TH were fromBoehringer Mannheim, and AMCA-conjugated anti-mouseIgG was from Calbiochem (San Diego, CA). All other reagentswere from Merck (Darmstadt, Germany).

RESULTSGABA Evoked [Ca2�]i Rises in the Majority ofChromaffin Cells

We observed rapid GABA-induced changes in[Ca2�]i in individual chromaffin cells with the aid of

digital imaging fluorescence microscopy. Figure 1A showscalcium images of a group of chromaffin cells before(basal) and after stimulation with 50 �M GABA, a con-centration close to the EC50 for GABA-evoked stimula-tion of CA secretion in chromaffin cells (Castro et al.,1989). From the total cell population, 72% � 3% re-sponded to 50 �M GABA (209 of 289 cells, in 15 exper-iments). The stimulation with GABA producedmonophasic peaks in Fura-2 F340/F380 ratio signal of dif-ferent amplitudes. Representative time courses of changesin [Ca2�]i in a few cells are shown in Figure 1B, in whichit can be appreciated that GABA induces rapid and tran-sient peaks, of very different magnitudes from cell to cell,returning to basal levels after removal of GABA. Chro-maffin cells are equipped with nicotinic acetylcholine re-ceptors with an EC50 of about 5–8 �M (Livett et al.,1983; Castro et al., 1989). We used short challenges withnicotine 20 �M as positive-response control pulses. Sucha challenge is enough to elicit near-maximal [Ca2�]i re-sponses without imposing a severe Ca2� load. Only cellsthat displayed healthy responses to 20 �M nicotine wereused for further studies. The time course of GABA-induced [Ca2�]i increases was similar after challengingcells with 20 �M nicotine; in both cases, cells displayedrapidly increasing ratio transients that gradually decayedafter washing of the receptor agonist. However, in the caseof nicotine, peak heights were always larger than thoseobtained with GABA. The ratio increases after GABAchallenges represented between 5% and 60% of the ratiorises obtained with 20 �M nicotine in the same cells.Healthy responses to nicotine were recorded from virtu-ally all chromaffin cells. Moreover, cells that did notresponded to GABA did show clear ratio increases inresponse to nicotine, indicating that the lack of response toGABA probably is due to the absence of GABA receptorsin those cells and not to deficient Ca2� pathways.

The resting Fura-2 fluorescence ratio was about0.75 � 0.16 (n 289), which corresponds to a [Ca2�]i ofabout 114 nM. Stimulation of cells with 50 �M GABAresulted in a net increase in the Fura-2 F340/F380 ratiobetween 0.3 and 2.5. The frequency histogram showingthe distribution of GABA increases in Fura-2 ratios (�ratio) in the whole chromaffin cell population is shown inFigure 1C. The average ratio increase elicited by GABAwas 0.955 � 0.024 (�420 nM, n 209 cells).

Expression of GAD Among Chromaffin CellsThe immunoidentification of chromaffin cells was

carried out by labeling with different specific antibodiesagainst TH, the rate-limiting enzyme for CA biosynthesis(which stain the total chromaffin cell population); PNMT,the specific enzyme for adrenaline biosynthesis (marker ofthe adrenergic population of chromaffin cells); and GAD,the enzyme responsible of GABA biosynthesis and amarker of GABAergic cells. Noradrenergic cells werethose stained by anti-TH but not by anti-PNMT antibodies.

GAD was expressed by almost one-third of all chro-maffin cells. In TH/GAD double immunolabelings, 30%� 1% of chromaffin cells were GAD� (288 of 970 cells in

GABA Receptors in Identified Chromaffin Cells 377

total). Figure 2 shows immunofluorescence images ofchromaffin cells identified by triple immunostaining,which allows us to distinguish adrenergic and noradren-ergic cells in addition to GAD� and GAD– populations. Inthe GAD panel in Figure 2, some cells are prominentlylabeled, but others cells labeled with anti-TH antibodiesshowed no or very faint fluorescence, comparable to thatin unlabeled controls. GAD expression was more fre-quently observed in adrenergic cells. About 40% ofPNMT� cells were also GAD�, a higher proportion than inthe whole population, indicating a selective enrichment ofGAD� cells within PNMT� cells. However, only 14% ofnoradrenergic PNMT– chromaffin cells were labeled forGAD. These data are summarized in Table I. As can beseen in Table I, there is an asymmetric distribution of

GAD expression in adrenergic and noradrenergic popula-tions. GAD– cells are more frequently noradrenergic(44%) than the averaged population, although PNMT–

cells are a minor proportion of chromaffin cells (a 36% inthis sample). The inverse is also highly segregated. Withinthe GAD� population, 83% of the cells were adrenergic.Thus, the expression of GAD in chromaffin cells is notrandom but correlates with PNMT expression.

Distribution of Responses to GABA in DifferentChromaffin Cell Populations

Once we had the ability to immunoidentify the typeof chromaffin cell as either adrenergic/noradrenergic orGAD�/GAD–, it was possible to assign single-cell calciumresponses to identified chromaffin cells localized on grid-

Fig. 1. Calcium responses to GABA in chromaffin cells. A: Calciumimages showing a group of chromaffin cells before (basal) and afterstimulation with 50 �M GABA (�GABA). Cells placed in a perifusionchamber on the stage of the microscope were imaged through a �20fluor objective. The cells were stimulated with 50 �M GABA for45 sec and then with 20 �M nicotine for 20 sec. Cells were allowed torest for 5 min between the consecutive stimulations. Pairs of images offields were captured at 1.5 sec intervals, and the F340/F380 ratio for eachcell in the field was determined on a pixel-by-pixel basis as described inMaterials and Methods. The images have been coded in grayscale toshow differences in Fura-2 F340/F380 ratio. White/light tones meanlow ratios, and black/dark tones mean high ratios. The contour of eachcell in the raw fluorescence image has been drawn to indicate celllocations. The stimulated image correspond to 10 sec after GABA

challenge, around the peak of the response. B: Time courses of changesin Fura-2 ratio in individual chromaffin cells in response to stimulationwith GABA. Traces show representative examples of typical F340/F380

ratio responses of individual cells and correspond to different fieldsduring different experiments. The lines show the application period ofGABA or nicotine. C: Frequency histogram for the distribution ofincreases in Fura-2 ratio (� ratio) evoked by 50 �M GABA measuredas indicated in Materials and Methods. The first bar represents thosecells unresponsive to GABA. A log-normal distribution was fitted tothe histogram, representing responding cells (without the first bin) andshown as the superimposed line. The fitted mean � ratio and �SDconfidence limits for this distribution of ratio increases are shown by thepoint and lines above the histogram. Data are representative of nineexperiments with a total of 289 cells recorded.

378 Castro et al.

engraved coverslips. Figure 3 shows the response histo-grams from a set of immunoidentified chromaffin cells inwhich Fura-2 ratio was measured during stimulation with50 �M GABA.

Subclassifying chromaffin cells according to the ex-pression of GAD into GAD� or GAD– did not reveal anyheterogeneity. In both groups, about 25% of the cells wereunresponsive to the GABA challenge. This is the samefraction found in the whole population (28%). The mag-nitude of the [Ca2�]i response to GABA was somewhatsmaller in the GAD� sample (0.848 � 0.048, [Ca2�]i �380 nM) than in the GAD– group of cells (0.990 � 0.027,[Ca2�]i � 425 nM). However, the averaged differencewas not statistically significant, and the histograms showthat in both types of cells the responses to GABA span asimilar range (see Fig. 3A). Thus, GABA receptors seem tobe equally distributed between chromaffin cells lackingand expressing the GAD enzyme.

Chromaffin cells can be separated also into adrener-gic and noradrenergic groups, as shown in Figure 3B. Thisclassification revealed a difference in the behavior of thetwo types of cells when challenged with GABA. Adren-ergic cells tended to show a higher frequency of positiveresponses to GABA. More than 80% of the PNMT� cellsdisplayed [Ca2�]i increases when challenged with 50 �MGABA. On the other hand, the noradrenergic populationshowed a very different response histogram. Up to 41% ofPNMT– cells failed to respond to GABA with [Ca2�]iincreases, a significantly lower proportion of responsesthan in adrenergic cells (P � 0.001). In addition, largeincreases in Fura-2 ratio evoked by GABA were observedmore often in adrenergic cells, although the population-averaged mean increase was not statistically different be-tween the two cell populations (0.984 � 0.024 vs.0.952 � 0.066 in adrenergic and noradrenergic cells,respectively).

Fig. 2. Immunofluorescence images showing the identification ofchromaffin cells by specific enzymatic antibodies. Triple immu-nostainnig of chromaffin cells with antibodies anti-TH (totalpopulation), anti-PNMT (adrenergic cells), and anti-GAD(GABAergic cells). The plates are representative of seven differentexperiments. Controls without primary antibodies showed a veryfaint, almost indiscernible fluorescence, in photographs takenusing the same exposure times (comparable to the three cells inthe top left panel, in the GAD plate). Counting data are summa-rized in Table I. Scale bars 50 �m.

GABA Receptors in Identified Chromaffin Cells 379

Pharmacological Identification of GABA ReceptorSubtypes

We had shown previously that GABA may modulatecatecholamine secretion by activation of two separate re-ceptors: ionotropic GABAA and metabotropic GABABreceptors (Castro et al., 1989; Gonzalez et al., 1992; Par-ramon et al., 1995b). By challenging chromaffin cells withspecific GABAA or GABAB agonists and antagonists andmeasuring the [Ca2�]i increase responses in single adrenalchromaffin cells, it is possible to identify cells containingeither one or the other of the two subtypes of GABAreceptors. Figure 4 shows panels representing typical ratioresponses for single chromaffin cells containing eitherGABAA or GABAB receptors alone or a mixed populationof both.

In cells expressing only GABAA receptors, bicucul-line, a specific GABAA antagonist at the concentration

used (20 �M), completely blocked the GABA response,whereas muscimol (a specific GABAA agonist, used at100 �M) evoked vigorous [Ca2�]i transients. In contrast,baclofen (100 �M), a GABAB agonist, and CGP-35348(250 �M), a specific GABAB antagonist, were totallyineffective (Fig. 4A, left). On the other hand, in cellsexpressing only GABAB receptors, CGP-35348, but notbicuculline, completely abolished the GABA response. Inaddition, muscimol was ineffective, whereas baclofenshowed clear positive responses (Fig. 4A, center). Finally,in those cells expressing both GABAA and GABAB recep-tors, either bicuculline or CGP-35348 produced a partialblock of GABA-evoked responses. In the same way, bothmuscimol and baclofen evoked calcium responses, butthese were smaller than those elicited by GABA (Fig. 4A,right).

As can be seen in the distribution graph in Figure 4B,GABAA receptor subtype was more widespread in chro-

Fig. 3. Distribution of chromaffin cell response to GABA in whole,adrenergic, and noradrenergic chromaffin cell populations. Frequencyhistograms for the distribution of increases in Fura-2 ratio (� ratio)evoked by 50 �M GABA measured as indicated in Materials andMethods. The first bar represents those cells unresponsive to GABA. Alog-normal distribution was fitted to the histogram, representing re-sponding cells (without the first bin) and shown as the superimposedline. A: Cells classified by GAD expression. B: Cells classified byPNMT expression. Data are presented normalized for each subgroupand come from nine representative experiments.

TABLE I. Relative Proportion of GAD� and GAD� Cells WithinAdrenergic (PNMT�) and Noradrenergic (PNMT�) ChromaffinCell Subpopulations*

PNMT (%) PNMT� (%)

GAD 44 � 2 (210/471) 55 � 2 (261/471)GAD� 16 � 2 (34/206) 83 � 2 (172/206)P �0.001 �0.001

*Data collected from six separate experiments and counting 18 independentmicroscopic fields. The values in parentheses are the total accumulatedcounts for all data (677 cells). Statistical analysis: P values at the bottomrepresent testing for equality of individual binomial proportion in eachcolumn. The observed differences are significant at P � 0.01 level accord-ing to the results of a 2 test for random distribution of GAD.

Fig. 4. Distribution of GABA receptors in different phenotypes ofchromaffin cells. A: Effects of GABA and GABA receptor agonists andantagonists on [Ca2�]i recorded from single adrenal chromaffin cells.Each panel represent typical ratio responses of single chromaffin cellscontaining only GABAA or GABAB receptors or a mixed population ofboth GABAA and GABAB receptors simultaneously. GABAA: In cellsexpressing only GABAA receptors, bicuculline completely blocked theGABA response, whereas muscimol evoked vigorous transients. Incontrast, baclofen and CGP-35348 were totally ineffective. GABAB: Incells expressing only GABAB receptors, CGP-35348, but not bicucul-line, completely abolished the GABA response. GABAA � GABAB: Inthose cells expressing both GABAA and GABAB receptors, eitherbicuculline or CGP-35348 produced only a partial block of GABA-evoked responses. Drugs and concentrations used: G, GABA, 50 �M;B, baclofen, 100 �M; M, muscimol, 100 �M; Bic, bicuculline, 20 �M;CGP, CGP-35348, 250 �M. B: Bar diagrams showing the distributionof the different types of GABA receptors in the total population,adrenergic, noradrenergic, GAD�, and GAD– chromaffin cells. Dataare presented normalized for each subgroup and come from ninerepresentative experiments.

380 Castro et al.

maffin cells. In the whole population, over 88% of the cellsthat did respond to GABA with [Ca2�]i increases con-tained receptors of the GABAA subtype. This finding wasalso true for all the subgroups of chromaffin cells that wehave considered. GABAA receptors are present in 80–100% of the GABA-responding cells. Furthermore, a sub-stantial fraction of the cells, about half (45–60%, depend-ing on the group considered) express the GABAA subtypeexclusively and lack the GABAB receptor.

On the other hand, GABAB receptors are much lesswidely distributed in the chromaffin cell population. Inthe whole population, they are found in 47% of GABA-responding cells. Their frequency varies from 40% (nor-adrenergic cells) to 55% (adrenergic cells) within the dif-ferent subpopulations considered. Generally, GABABreceptors are coexpressed with GABAA receptors, as couldbe inferred from the prevalence of the GABAA subtype.The proportion of GABA-responding chromaffin cellsthat lack the GABAA subtype and express only theGABAB receptor subtype is very small (12%).

The occurrence of both GABA receptor subtypes inGAD� and GAD– subpopulations considered separatelywas very similar to that of the whole population, with44–45% of cells displaying GABAB responses and a pre-dominant expression of GABAA receptor subtypes. It isremarkable that GABAA receptors were present in 100%of GAD� cells that responded to GABA.

In considering adrenergic and noradrenergic cellsseparately, some specific enrichment of one subtype isobserved. GABAB was slightly more frequent among nor-adrenergic cells than among adrenergic cells (55% vs. 40%P � 0.01). Conversely, cells containing GABAA receptorswere enriched in the PNMT� adrenergic population(92% vs. 72% P � 0.01). The proportion of cells thatcoexpressed both GABAA and GABAB receptor subtypesremained constant in adrenergic and noradrenergic cellsand similar to the proportion observed in the whole pop-ulation (34%). As a consequence, the proportion of cellsexpressing exclusively the GABAB receptor subtype is 2.5times higher within the noradrenergic subpopulation thanwithin the adrenergic subpopulation.

DISCUSSIONOnly 30% of bovine chromaffin cells express the

GABA-synthesizing enzyme GAD, and over 70% of themexpress GABA receptors. Thus, it was tempting to spec-ulate that GABA may serve an exclusively paracrine rolewithin the adrenal medulla; being released by a group ofcells, it acts on a different target population of cells, ascenario analogous to what we have described for nitricoxide (NO) signalling, in which noradrenergic cells spe-cialize in NO synthesis (NO synthase expression), whereasadrenergic cells are the targets of the NO action (Oset-Gasque et al., 1998).

However, the results from our study do not supportthis view. In fact, the expression of GAD is correlatedneither with the presence of GABA receptors nor with therelative abundance of GABA receptor subtypes in chro-

maffin cells. GABA-sensitive cells are equally distributedamong GAD� and GAD– subpopulations. About three offour cells express GABA receptors, without any differencewith regard to GAD expression. The distribution of bothGABAA and GABAB receptors is similar in GAD� andGAD– cells and closely resembles the distribution in thewhole population. Even the finding that GAD� cells lacka separate population expressing only GABAB receptors isof minor significance, given the overwhelming prepon-derance of GABAA receptor expression within GABA-sensitive chromaffin cells. Thus, GABA receptors in ad-renal chromaffin cells seem to be tuned to receive nonlocalGABA signals. In addition to GABA released from otherchromaffin cells, GABA signalling can originate fromGABA-ergic nerve endings entering the adrenal medulla.Such innervation has been clearly demonstrated in severalspecies, including bovine, murine, and canine species(Kataoka et al., 1984, 1986; Ahonen et al., 1989; Iwasa etal., 1999). This network is quite dense, and it is probablythe main source of GABA acting on chromaffin cells. Thedetails of the precise cellular distribution of GAD maydiffer from one species to other. In rat adrenal medulla,GAD is found mainly in ADR-containing cells (Ahonenet al., 1989), as we find in bovine tissue. However, in themouse gland, intrinsic GAD in the adrenal medulla ispresent as a weak immunoreactivity within noradrenergiccells (Iwasa et al., 1999), contrary to what could be con-cluded from our work in bovine tissue.

Although the expression of GAD and the presence ofGABA receptors are not correlated with each other, bothparameters show a partial correlation with the adrenergicphenotype, the expression of PNMT enzyme. In fact,most (86%) GABA-synthesizing cells are adrenergic. It isremarkable that this high degree of localization is achievedby a very modest biased selectivity, together with thehigher abundance of PNMT� cells in bovine adrenalmedulla. Our results suggest that even minor asymmetriesin the distribution of some marker may give rise to pro-nounced differences if several such asymmetries accumu-late in the same direction. The segregated compartmen-talization of GAD is mirrored by a moderate asymmetry inthe distribution of GABA receptors together with PNMTexpression. Noradrenergic cells are GABA unresponsive ata disproportionately higher frequency than adrenergiccells, and the expression of GABA receptors is specificallydepressed in noradrenaline-containing cells. Thus, norad-renergic cells are less likely to be a target for GABA action.The distribution of GABAA and GABAB subtypes isslightly different in adrenergic and noradrenergic cells.The physiological meaning of this asymmetry, if any,seems only marginal. Both receptor subtypes mediate aninhibition of catecholamine release stimulated by cholin-ergic stimulation. Therefore, the net effect would be qual-itatively identical for both subtypes. On the other hand,GABAA receptors may enhance the basal catecholaminesecretion when acting alone (Kitayama et al., 1990;Gonzalez et al., 1992). The preponderance of this subtype

GABA Receptors in Identified Chromaffin Cells 381

(more pronounced in adrenergic cells) might be related toa role of GABA in the control of basal secretion tone ofthe gland, as described for the dog (Kataoka et al., 1986).

The role of GABA in the secretion of cat-echolamines by adrenal medulla may be summarized byacknowledging that there are two pathways of GABAsignaling in the adrenal medulla, neural and paracrine. Theparacrine pathway comes from GAD� cells that are over-whelmingly adrenergic. This pathway is probably second-ary to the neural pathway: GABA released from the affer-ent nerve endings. The target of GABA action is mostlikely the adrenergic cell population (either GAD� orGAD–), mainly because adrenergic cells are more abun-dant but also based on the relative GABA unresponsive-ness of NA cells. It is very suggestive that, in two species,incoming GABA fibers make intimate contacts with ad-renergic cells in the adrenomedullary tissue, sparing thenoradrenergic cell clusters (Iwasa et al., 1999).

ACKNOWLEDGMENTSThe experimental part of this study was performed at

the Centre for Neurosciences of Coimbra (Coimbra, Por-tugal) in the laboratory of Dr. Luıs M. Rosario. We aregrateful to Dr. Rosario for warmly welcoming us andallowing to use the Ca2� imaging facility. We also thankDr. A. Tome and G. Baltazar for their help in obtainingchromaffin cell cultures and S. Figueroa for correcting themanuscript. E.C. was supported by a FEBS Long TermFellowship. M.J.O.-G. was the recipient of a short-termfellowship from UCM (Spain).

REFERENCESAhonen M, Soinila S, Wu J-Y, Happola O. 1989. L-Glutamate decarbox-

ylase immunoreactivity in developing sympathetic tissues of the rat. JAuton Nerv Syst 27:155–164.

Castro E, Oset-Gasque MJ, Canadas S, Gimenez A, Gonzalez MP. 1988.GABAA and GABAB sites in bovine adrenal medulla membranes. J Neu-rosci Res 20:241–245.

Castro E, Oset-Gasque MJ, Gonzalez MP. 1989. GABAA and GABAB

receptors are functionally active in the regulation of catecholamine secre-tion by bovine chromaffin cells. J Neurosci Res 23:290–296.

Castro E, Mateo J, Tome AR, Barbosa RM, Miras-Portugal MT, RosarioLM. 1995. Cell-specific purinergic receptors coupled to Ca2� entry andCa2� release from internal stores in adrenal chromaffin cells. J Biol Chem270:5098–5106.

Choi AY, Cahill AL, Perry BD, Perlman RL. 1993. Histamine evokesgreater increases in phosphatidylinositol metabolism and catecholaminesecretion in epineprine-containing than in norepineprine-containingchromaffin cells. J Neurochem 61:541–549.

Feuerstein G, Gutman Y. 1971. Preferential secretion of adrenaline andnoradrenaline by cat adrenal in vivo in response to different stimuli. Br JPharmacol 43:764–775.

Fujimoto M, Kataoka Y, Guidotti A, Hanbauer I. 1987. Effect of gamma-aminobutyric acid receptor agonists and antagonists on the release ofenkephalin-containing peptides from dog adrenal gland. J Pharmacol ExpTher 243:195–199.

Gonzalez MP, Oset-Gasque MJ, Castro E, Bujeda J, Arce C, Parramon M.1992. Mechanisms through which GABA receptor modulates catechol-amine secretion from bovine chromaffin cells. Neuroscience 47:487–494.

Iwasa K, Oomori Y, Tanaka H. 1999. Colocalization of gamma-aminobutyric acid immunoreactivity and acetylcholinesterase activity innerve fibers of the mouse adrenal gland. J Vet Med Sci 61:631–635.

Kataoka Y, Gutman Y, Guidotti A, Panula P, Wroblewski J, Cosenza-Murphy D, Wu JY, Costa E. 1984. Intrinsic GABAergic system of adrenalchromaffin cells. Proc Natl Acad Sci USA 81:3218–3222.

Kataoka Y, Fujimoto M, Alho H, Guidotti A, Geffard M, Kelly GD,Hanbauer I. 1986. Intrinsic gamma aminobutyric acid receptors modulatethe release of catecholamine from canine adrenal gland in situ. J Pharma-col Exp Ther 239:584–590.

Kitayama S, Morita K, Dohi T, Tsujimoto A. 1990. GABAergic modula-tion of catecholamine release from cultured bovine adrenal chromaffincells. Evidence for the involvement of Cl–-dependent Ca2� entry. Nau-nyn Schmiedebergs Arch Pharmacol 341:419–424.

Livett BG, Boksa P, Dean DM, Mizobe F, Lindenbaum MH. 1983. Use ofisolated chromaffin cells to study basic release mechanisms. J Auton NervSyst 7:59–86.

Mateo J, Castro E, Miras-Portugal MT. 1997. P2 purinoceptors in adre-nomedullary cells. In: Kanno T, Nakazato Y, Kumakura K, editors. Theadrenal chromaffin cell. Sapporo: Hokkaido University Press. p 101–113.

Nunez L, De la Fuente MT, Garcıa AG, Garcıa-Sancho J. 1995. DifferentialCa2� responses of adrenergic and noradrenergic cromaffin cells to varioussecretagogues. Am J Physiol 269:C1540–C1546.

Oomori Y, Iuchi H, Nakaya K, Tanaka H, Ishikawa K, Satoh Y, Ono K.1993. Gamma-aminobutyric acid (GABA) immunoreactivity in themouse adrenal gland. Histochemistry 100:203–213.

Oset-Gasque MJ, Castro E, Gonzalez MP. 1990a. GABAergic mechanismsin bovine adrenal medulla. In: Erdo SL, editor. GABA outside the CNS.Berlin: Springer-Verlag. p 167–182.

Oset-Gasque MJ, Castro E, Gonzalez MP. 1990b. Mechanisms of[3H]gamma-aminobutyric acid release by chromaffin cells in primaryculture. J Neurosci Res 26:181–187.

Oset-Gasque MJ, Parramon M, Gonzalez MP. 1993. GABAB receptorsmodulate catecholamine secretion in chromaffin cells by a mechanisminvolving cyclic AMP formation. Br J Pharmacol 110:1586–1592.

Oset-Gasque MJ, Vicente S, Gonzalez MP, Rosario LM, Castro E. 1998.Segregation of nitric oxide synthase expression and calcium response tonitric oxide in adrenergic and noradrenergic bovine chromaffin cells.Neuroscience 83:271–280.

Parramon M, Gonzalez MP, Oset-Gasque MJ. 1995a. Pharmacologicalmodulation of adrenal medullary GABAA receptor: consistent with itssubunit composition. Br J Pharmacol 116:1875–1881.

Parramon M, Gonzalez MP, Herrero MT, Oset-Gasque MJ. 1995b.GABAB receptors increase intracellular calcium concentrations in chro-maffin cells through two different pathways: their role in catecholaminesecretion. J Neurosci Res 41:65–72.

382 Castro et al.