Differential expression of metabotropic glutamate and GABA receptors at neocortical glutamatergic and GABAergic axon terminals

Metabotropic glutamate (Glu) receptors (mGluRs) and GABAB receptors are highly expressed at presynaptic sites. To verify the possibility that the two classes of metabotropic receptors contribute to axon terminals heterogeneity, we studied the localization of mGluR1α, mGluR5, mGluR2/3, mGluR7, and GABAB1 in VGLUT1-, VGLUT2-, and VGAT- positive terminals in the cerebral cortex of adult rats. VGLUT1-positive puncta expressed mGluR1α (∼5%), mGluR5 (∼6%), mGluR2/3 (∼22%), mGluR7 (∼17%), and GABAB1 (∼40%); VGLUT2-positive terminals expressed mGluR1α (∼10%), mGluR5 (∼11%), mGluR2/3 (∼20%), mGluR7 (∼28%), and GABAB1 (∼25%); whereas VGAT-positive puncta expressed mGluR1α (∼27%), mGluR5 (∼24%), mGluR2/3 (∼38%), mGluR7 (∼31%), and GABAB1 (∼19%). Control experiments ruled out the possibility that postsynaptic mGluRs and GABAB1 might have significantly biased our results. We also performed functional assays in synaptosomal preparations, and showed that all agonists modify Glu and GABA levels, which return to baseline upon exposure to antagonists. Overall, these findings indicate that mGluR1α, mGluR5, mGluR2/3, mGluR7, and GABAB1 expression differ significantly between glutamatergic and GABAergic axon terminals, and that the robust expression of heteroreceptors may contribute to the homeostatic regulation of the balance between excitation and inhibition.

The widespread localization of mGluRs and GABA B at presynaptic sites raises the possibility that both classes of metabotropic receptors contribute to axon terminals heterogeneity. To verify this hypothesis, we studied the localization of mGluR1α, mGluR5, mGluR2/3, mGluR7, and GABA B1 in VGLUT1+, VGLUT2+, and VGAT+ 1 terminals in the cerebral cortex of adult rats. We report that the expression of these receptors differs significantly between glutamatergic and GABAergic axon terminals, and that expression of heteroreceptors is greater than expected. We also performed functional assays in synaptosomal preparations, and showed that all agonists modify Glu and GABA levels, which return to baseline upon antagonists administration.

Animals and Tissue Preparation
Adult male Sprague-Dawley rats (190-220 g; Charles River, Milan, Italy) were used. All experiments were carried out in accordance with the European Community Council Directive dated November 24, 1986 (86/609 EEC), and were approved by the local authority veterinary service (CESA, Comitato Etico per la Sperimentazione Animale; Università Politecnica delle Marche). Animals were kept under a dark-light cycle of 12 h and permitted food and water ad libitum.
For immunocytochemical studies, rats were anesthetized with chloral hydrate (300 mg/kg i.p.) and perfused through the ascending aorta with saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB; pH 7.4). Brains were post-fixed for 1 day at 4 • C in the same fixative, cut with a Vibratome into 30 μm thick sections, and processed.

Antibodies
The primary antibodies used are listed in Table 1. Western blots were performed to verify antibodies specificity; nitrocellulose filters were probed with antibodies to VGLUT1, VGLUT2, VGAT, mGluR1α, mGluR5, mGluR2/3, mGluR7, and GABA B1 at the dilutions reported in Table 1. After exposure to the appropriate peroxidase-conjugated antibodies (Vector; Burlingame, CA, USA), immunoreactive bands were visualized 1 The vesicular transporter described by McIntire et al. (1997) and Sagné et al. (1997) does transport both GABA and glycine (Wojcik et al., 2006), and is termed either VGAT (McIntire et al., 1997) or VIAAT (Sagné et al., 1997). Here, in line with its widespread usage and with the more prominent role of GABA in neocortex, we follow the terminology of McIntire et al. (1997).

Co-localization Studies
Sections were incubated for 1 h in normal goat serum (NGS; 10% in PB with 0.2% Triton X-100), and then overnight at room temperature in a solution containing a mixture of the primary antibodies. The next day, sections were incubated in 10% NGS (30 min), and then for 90 min in a mixture of the appropriate secondary fluorescent antibodies. For the VGLUT1 and VGLUT2 series, we used fluorescein isothiocyanate-conjugated goat anti-guinea-pig IgG (FI-7000, Vector; Burlingame, CA, USA) for the vesicular transporters (1:100) and tetramethylrhodamine isothiocyanate-conjugated goat anti-rabbit IgG (T-2769, Molecular Probes; Poort Gebouw, The Netherlands) for the metabotropic receptors (1:100); for the VGAT series, we used fluorescein goat anti-mouse IgG (F-2761, Molecular probes) for the vesicular transporter (1:100) and tetramethylrhodamine isothiocyanate-conjugated goat anti-rabbit IgG (T-2769, Molecular Probes; Poort Gebouw, The Netherlands) for the metabotropic receptors (1:100). Sections were then mounted, air-dried, and coverslipped using Vectashield mounting medium (H-1000; Vector). Doublelabeled sections were examined using a Leica (TCS SP2) confocal laser microscope equipped with an argon (488 nm) and a helium/neon (543 nm) laser for excitation of fluorescein isothiocyanate (FITC) and tetramethylrhodamine isothiocyanate (TRITC), respectively. Green and red immunofluorescence were imaged sequentially and emissions separated with 515/30 nm band pass (FITC) and 570 nm long pass (TRITC) filters. Control experiments with single-labeled sections and sections incubated either with two primary and one secondary antibody, or with one primary and two secondary antibodies revealed no appreciable FITC/TRITC bleed-through or antibody cross-reactivity. Images of experimental series were collected from a region of the parietal cortex characterized by a conspicuous layer IV, with intermingled dysgranular regions, densely packed layers II and III, and a relatively cell-free layer Va. This area corresponds to the first somatic sensory cortex (SI), as identified also by Woolsey and LeMessurier (1948), Welker (1971), Zilles et al. (1980), Donoghue and Wise (1982). Images were acquired from randomly selected subfields in layers II-VI (at least 4-6/layer; 2-4 sections/animal; 10 rats). Layer I was not sampled because it hardly contains VGAT+ puncta (Chaudhry et al., 1998;Minelli et al., 2003). Images were acquired using a 63× oil immersion lens (numerical aperture 1.4; pinhole 1.0 and image size 1,024 × 1,024 pixels, yielding a pixel size of 0.06 μm) from a plane in which the resolution of both stains was optimal and always between 1.3 and 1.8 μm from the surface. To improve the signal/noise ratio, 10 frames/image were averaged.
Quantitative analysis was performed in ∼8,000 randomly selected subfields measuring about 25 × 25 μm from the 1,024 × 1,024 pixel images. In order to minimize the fusion of puncta, the contrast of each image was adjusted manually within the maximum range of levels for each color channel. Analysis of contrast adjustment (not shown) showed that gain/contrast changes within the spectrum used did not alter significantly the percentage of puncta. Then, without reducing the image resolution, the images were binarized and processed by watershed filter using ImageJ software (bfd). Next, each channel was examined separately to identify and count with ImageJ immunopositive puncta; the two channels were then merged and the number of co-localizing puncta was counted manually. Puncta were considered double-labeled when overlap was virtually complete or when a given punctum was entirely included in the other. Moreover, we analyzed ∼2,000 randomly selected subfield (25 × 25 μm) from the 1,024 × 1,024 pixel images acquired in molecular layer of cerebellum and ventrobasal nucleus (10-20/section; 2-4section/animal; 2 animals).
In addition, we compared our manual method with a computerized overlap analysis that defines two objects as colocalized if the centre of mass of one falls within the area of the other (Lachmanovich et al., 2003). To this end, we analyzed about half of all double-labeled sections studied here with the overlap method included in JACoP toolbox of ImageJ (Bolte and Cordelieres, 2006), and found that the percentage of co-localization obtained with the two methods were comparable.

Release Experiments
Synaptosomes (from 32 rats) were incubated at 37 • C for 15 min; aliquots of synaptosomal suspension (150 μg) were layered on microporous filters placed at the bottom of a set of parallel superfusion chambers maintained at 37 • C (Superfusion System; Ugo Basile, Comerio, Italy; (Raiteri et al., 1984). Superfusion was started with standard medium at a rate of 0.5 ml/min and continued for 48 min. In the experiments aimed at measuring basal Glu and GABA release, after 36 min of superfusion to equilibrate the system, fractions were collected according to the following scheme: four 3-min samples (t = 36-39, basal release; t = 39-42, t = 42-45, and t = 45-48, druginduced release). The mGluR1 and mGluR5 agonist 3,5-DHPG (30 μM) was introduced at t = 39, after the first sample was collected. When appropriate, the selective mGluR1 and mGluR5 antagonists LY367385 (1 μM) and MPEP (1 μM), respectively, were introduced at t = 30 and maintained until the end of the experiment. Drug effects were evaluated by comparing the amount of endogenous Glu or GABA (pmol/mg synaptosomal protein) in the fourth sample collected (in which maximum effect of 3,5-DHPG was generally reached) with the content in the first sample (basal efflux). In experiments aimed at measuring the stimulus-evoked Glu and GABA release, after 36 min of superfusion to equilibrate the system, fractions were collected according to the following scheme: two 3-min samples (t = 36-39 min and t = 45-48 min; basal outflow) before and after one 6-min sample (t = 39-45 min; stimulusevoked release). A 90-s period of stimulation was applied at t = 39 min, after the collection of the first sample. Stimulation of synaptosomes was performed by 15 mM KCl, substituting for equimolar concentration of NaCl. The mGluR2 and mGluR3 agonist LY379268 (100 nM), the mGluR7 agonist AMN082 (100 nM) or the GABA B receptor agonist (-)baclofen (10 μM) was introduced at t = 39 min concomitantly to the KCl pulse. When appropriate, the selective mGluR2 and mGluR3 antagonist LY341495 (100 nM), the selective mGluR7 antagonist MMPIP (10 nM) or the selective GABA B antagonist CGP52432 (1 μM) was introduced at t = 30 min and maintained until the end of the experiment. The stimulus-evoked overflow was estimated by subtracting the endogenous neurotransmitter release content of the two 3-min fractions (basal outflow) from the endogenous neurotransmitter amount collected in the 6-min fraction (stimulus-evoked overflow).
Endogenous Glu and GABA content in the samples collected was measured by high performance liquid chromatography following pre-column derivatization with o-phthalaldehyde and gradient separation on a C18 reverse-phase chromatographic column (10 mm × 4.6 mm, 3 μm; at 30 • C; Chrompack, Middleburg, The Netherlands) coupled with fluorometric detection (excitation wavelength 350 nm; emission wavelength 450 nm). Homoserine was used as an internal standard (Bonanno et al., 2005). The following buffers were used: solvent A, 0.1 M sodium acetate (pH 5.8)/methanol 80:20; solvent B, 0.1 M sodium acetate (pH 5.8)/methanol, 20:80; solvent C, sodium acetate (pH 6.0)/methanol, 80:20. The gradient program was as follows: 100% C for 4 min from the initiation of the program; 90% A and 10% B in 1 min; 42% A and 58% B in 14 min; 100% B in 1 min; isocratic flow 2 min; 100% C in 3 min; flow rate 0.9 mL/min. Amino acid release content was expressed as picomoles per milligram of synaptosomal protein.

Statistical Analysis
Statistical significance was evaluated by non-parametric Mann-Whitney U test (for confocal microscopy in cerebellum and ventrobasal complex), and non-parametric one way ANOVA (Kruskal-Wallis with Dunn's post-test for confocal microscopy and release experiments in cerebral cortex).
Expression of mGluRs and GABA B1 in VGLUT1+ cortical axon terminals was studied in 25 sections from eight rats (Figures 2 and 3).
Since mGluRs and GABA B are located both pre-and postsynaptically, we tested whether our data on pre-synaptic localization had been biased by the presence of postsynaptic receptors. GABA B1 is expressed presynaptically in cerebellum but not in the thalamic ventrobasal complex ; we therefore studied GABA B1 expression in VGLUT1, VGLUT2, and VGAT+ puncta in cerebellum and ventrobasal complex (2 rats, 12 sections) to verify whether our confocal method was able to differentiate pre-synaptic from postsynaptic localization. We found that whereas in cerebellum 19 ± 1% of VGLUT1, 9.6 ± 0.3% of VGLUT2, and 9.2 ± 0.3% of VGAT were FIGURE 3 | Expression of mGluR1α, mGluR5, mGluR2/3, mGluR7, and GABA B1 in VGLUT1+, VGLUT2+, and VGAT+ axon terminals in cerebral cortex. Values are mean ± SEM; * * p < 0.01, * * * p < 0.001. GABA B1 +, in ventrobasal complex only 4.9 ± 0.1%, of VGLUT1, 4.4 ± 0.1% of VGLUT2, and 4.2 ± 0.2% of VGAT were GABA B1 + (Figure 4). These findings indicate that most of labeling identified as presynaptic is indeed presynaptic, even though values given in the preceding paragraph may be slightly overestimated (see Discussion).
To further confirm the presynaptic localization of mGluRs and GABA B1 and to gain insights on their functional role, we studied the impact of activation of the two classes of metabotropic receptors on the release of endogenous glutamate and GABA in rat parietal cortex. The spontaneous efflux of GABA and Glu from cerebral cortex synaptosomes in superfusion, in the absence of stimulus amounted to 51.4 ± 4.62 and to 99.6 ± 4.95 pmol/mg of proteins, respectively. The 15 mM FIGURE 4 | Control analysis of confocal microscopy images. VGLUT1/GABA B1 in molecular layer of cerebellum (Cb) and in ventrobasal complex (VB). Puncta were considered double-labeled when overlap was virtually complete or when a given punctum was entirely included in the other puncta (arrows in third and fourth columns). Puncta that did not meet these criteria (arrowheads) were not considered double-labeled; Bars: 2 μm.

Discussion
The present study showed that mGluRs and GABA B receptors are differentially expressed in glutamatergic and GABAergic axon terminals. Interestingly, mGluR1α, mGluR5, mGluR2/3, and mGluR7 are more robustly expressed at GABAergic than at glutamatergic terminals, whereas GABA B1 is more expressed in glutamatergic terminals. In addition, we showed that in cerebral cortex mGluR1 and mGluR5 potentiate Glu and GABA release, whereas mGluR2/3, mGluR7, and GABA B receptors inhibit it.
both pre-and postsynaptically Lopez-Bendito et al., 2002;Boyes and Bolam, 2003;Kulik et al., 2003;Chen et al., 2004;Lujan et al., 2004;Lacey et al., 2005;Lujan and Shigemoto, 2006). Therefore, we performed several control studies to rule out the possibility that our "presynaptic" population had been contaminated by postsynaptic receptors. Since presynaptic localization of GABA B1 occurs in cerebellum but not in the thalamic ventrobasal (VB) nucleus , we first studied the co-localization of GABA B1 receptor and VGLUT1, VGLUT2, and VGAT in both cerebellum and VB, and found a significant differences between co-localization in cerebellum and VB (where co-localization was very low) in the three types of terminals. It is therefore safe to assume that our confocal method can satisfactorily discriminate presynaptic and postsynaptic expression. This view is further reinforced by the observation that in our experimental conditions GABA B1 agonists elicited modulatory responses.
Next, we performed functional assays in synaptosomal preparations, which provided further evidence that the mGluRs and GABA B studied immunocytochemically are indeed located presynaptically. To this purpose, we monitored Glu and GABA release in cerebral cortex synaptosomes under superfusion conditions (Raiteri et al., 1984). It has been demonstrated that by this way all the released transmitters/modulators are immediately removed by the superfusion medium flow, before they can interact with receptors or other proteins expressed at the presynaptic level. As a consequence, any drug-induced effect on the release of a given neurotransmitter can be attributed exclusively to the direct action of the drug at the nerve terminal storing and releasing that neurotransmitter. Thus, the modulation of Glu or GABA release reported here is in all likelihood due to activation of mGluRs or GABA B1 receptors located at Glu or GABA releasing nerve terminals, respectively. Overall, these sets of data provide evidence that, although we might have slightly (<5%) overestimated the population of mGluRs and GABA B receptors, our immunocytochemical localization of presynaptic receptors was not grossly biased by their postsynaptic localization. The present observation that in adult neocortex mGlu and GABA B1 receptors exhibit differences between glutamatergic and GABAergic terminals expands the notion that these axon terminals greatly differ in their complement of presynaptic proteins (Bragina et al., 2007(Bragina et al., , 2010(Bragina et al., , 2011.
The present results show that not all glutamatergic terminals express mGluRs. Indeed, our data indicate that at least 60% of VGLUT1+ and >10% of VGLUT2+ glutamatergic terminals are devoid of glutamate-mediated metabotropic control. Moreover, these percentages could be significantly increased by the possibility that some terminals express more than one mGluR. As far as VGAT+ terminals are concerned, the present data suggest that all of them could express mGluRs, although the lack of triple-labeling studies prevent a firm conclusion, and that only a small fraction do exhibit GABA-mediated metabotropic control. Thus, although the amount of data available is still limited, it is safe to conclude that not all glutamatergic and GABAergic terminals are controlled through metabotropic receptors by glutamate and GABA, respectively.
Besides this, our results provide some novel data on mGluRs and GABA B receptors localization in cerebral cortex. mGluRs and GABA B receptors have been described in both glutamatergic and GABAergic terminals (Gereau and Swanson, 2008;Raiteri, 2008, for reviews), but the present quantitative observation in neocortex that presynaptic mGluRs are so strongly expressed in GABAergic and that presynaptic GABA B receptors are more expressed in glutamatergic than in GABAergic terminals adds much to previous observations and opens a new perspective to unravel cortical microcircuits. Indeed, although the complex relationships between glutamatergic and GABAergic are only beginning to be unveiled (Lujan et al., 2005;Pinheiro and Mulle, 2008;Heja et al., 2009), making any hypothesis fragile, it is conceivable that the strong heterolocalization of mGluRs and GABA B receptors at cortical axon terminals is part of the complex homeostatic mechanism regulating the balance between excitation and inhibition.