Stereological estimations and neurochemical characterization of neurons expressing GABAA and GABAB receptors in the rat pedunculopontine and laterodorsal tegmental nuclei

The pedunculopontine (PPN) and the laterodorsal (LDT) tegmental nuclei are two complex brainstem structures characteristically containing cholinergic cells (Mesulam et al. 1983), which form an anatomical continuum in the mesopontine tegmentum and are considered as a functional unit working in an integrated fashion (Mena-Segovia 2016). Some of their functions, like locomotion and motor adaptive control, action selection and reward, are related to their reciprocal connections with basal ganglia structures (Mena-Segovia et al. 2004; Roseberry et al. 2016; Takakusaki et al. 2016; Gut and Winn 2016). Specifically, both the PPN and LDT receive abundant projections from the output nuclei of the basal ganglia together with minor projections from the striatum and lateral globus pallidus (Cornwall et al. 1990; Edley and Graybiel 1983; Rye et al. 1987; Shink et al. 1997; Steininger et al. 1992), all of which are GABAergic. However, despite the fact that GABA actions on the PPN have long been known from electrophysiological and pharmacological studies (Ikeda et al. 2004; Nandi et al 2002; Saitoh et al 2003; Torterolo et al 2002), specific studies about the anatomical distribution of GABA receptors (GABARs) in the PPN or LDT have not been carried out so far.

In addition, intermingled with their characteristic cholinergic neurons, PPN and LDT also comprise abundant glutamatergic and GABAergic cells (Wang and Morales 2009; Mena-Segovia et al. 2009; Luquin et al. 2018). Specifically, some of these non-cholinergic cells seem to be the main target of GABAergic projections from the basal ganglia output nuclei (Grofova and Zhou 1998; Mongia et al. 2015; Sherman et al. 2015; Roseberry et al. 2016; Caggiano et al. 2018). However, little is known about the relative distribution of nigral or pallidal terminations on each of the three neurochemical subpopulations of the PPN and LDT, or about the anatomical localization of postsynaptic GABARs potentially mediating the direct inhibitory control from these projections. As a first step to better understand the interactions between the downstream projections of basal ganglia output nuclei and the PPN and LDT at the cellular level and the neural circuitry within these nuclei, we aimed to anatomically confirm the presence of GABA receptors within the PPN and LDT, and to quantitatively determine their relative distribution among the three cell phenotypes in the two nuclei.

Electrophysiological and pharmacological studies have demonstrated the presence of functional GABAA and GABAB receptors in PPN and LDT (Saitoh et al. 2003; Ikeda et al. 2004; Ulloor et al. 2004; Pal and Mallick 2004; Datta 2007; Heinmiller et al. 2009; Ye and Garcia-Rill 2009; Kohlmeier and Kristiensen 2010; Fogel et al. 2010; Kohlmeier et al, 2013). GABAA receptors (GABAARs) are ligand-gated Cl ion channels structurally assembled from five individual protein subunits which include α1-6, β1-3, γ1-3, δ, ε, π, and τ (Olsen and Sieghart 2009). The most common combination of subunits in GABAARs, however, is triplet α1/β2/γ2, which is detected in various cell types in the CNS (Fritschy et al. 1992; McKernan and Whiting 1996). GABAB receptors (GABABRs) are heterodimers formed by the co-assembly of two subunits, GABAB1 and GABAB2, that couple to G-proteins through second messenger pathways and act as modulators of GABA transmission (Bowery et al. 2002). Whole-brain studies on the distribution of either GABAAR or GABABR subunits reported the presence of specific subunits in the rat PPN or LDT in relation to cell bodies, neural processes or neuropil (Fritschy and Möhler 1995; Margeta-Mitrovic et al. 1999; Pirker et al. 2000). However, a specific analysis of GABAAR or GABABR distribution in the PPN and LDT has not been carried out thus far.

Finally, although the PPN and LDT share many common input structures and efferent target areas, both afferent and efferent projections are often topographically organized at the regional level (Dautan et al. 2016a; Hallanger et al. 1987; Oakman et al. 1995; Woolf and Butcher 1986; Xiao et al. 2016) and also at the cellular one (Dautan et al. 2016b), supporting their specific participation in separate circuits as well as their partially different roles (Mena-Segovia 2016). Thus, the aims of the present study were: to assess the expression of GABAARs and GABABRs in each cell phenotype of both nuclei, to estimate with unbiased stereological methods the mean number of GABAAR- and GABABR-expressing cells in the three cell subpopulations, and to investigate whether differences in these estimations were present between the PPN and LDT.

Previous studies have shown that the vast majority of GABAAR in brain contain at least one α subunit variant, along with the β2,3- and γ2-subunits (Pirker et al. 2000;Fritschy and Möhler 1995) and, in general, α1 and γ2 are two of the predominating subunits in the brainstem (Fritschy and Möhler 1995). Thus, for the anatomical detection of GABAAR in the PPN and LDT we used immunoreactivity against α1 and γ2 GABAAR subunits, whereas immunoreactivity against the GABAB R2 subunit was used for the GABABR. Subsequently, the total numbers of cells expressing the GABAAR γ2- and GABAB R2 subunits in both nuclei were estimated using stereological methods and the neurochemical phenotypes of GABAAR γ2- and GABABR R2-expressing cells were investigated. Our results demonstrate the abundant expression of both GABAAR and GABABR in the three cell subpopulations of the PPN and LDT. At the same time, they reveal a heterogeneous distribution of both GABAAR and GABABR among glutamatergic and GABAergic cells that suggests the existence of functionally different subsets within those two subpopulations. Finally, they show marked differences between PPN and LDT regarding GABAAR subunit composition in the cholinergic cell population, as well as in relation to the synaptic versus extrasynaptic location of GABAARs in the non-cholinergic subpopulations. Some of these differences in the mechanisms of inhibitory control of specific cell subsets within the two nuclei may contribute to some of the functional differences observed between them.

Mouse monoclonal antibodies against GABAAR α1 and GABAB R2 and a rabbit polyclonal antibody against GABAAR γ2 were used (Table 1), all of which had been characterized previously. Mouse monoclonal antibodies were developed by UC Davis/NIH NeuroMab Facility in mouse generating hybridomas. The anti-GABAAR α1 mouse monoclonal antibody (clone N95/35 monoclonal IgG2a) recognizes protein amino acids 355–394 (cytoplasmic loop) of the α1 subunit and has been used in immunofluorescence labeling in both rat and mice (Micheva et al. 2010; Eyre et al. 2012). The anti-GABAB R2 mouse monoclonal antibody used here (clone N81/2) recognizes protein amino acids 862–913 of the R2 subunit, and has also been used in immunoblotting and immunofluorescence labeling (i.e., Nassirpour et al. 2010; Broussard et al. 2011; Maity et al. 2012). Finally, the anti-GABAAR γ2 rabbit polyclonal antibody recognizes 30 amino acids of an epitope of the extracellular domain of N-terminus of the rat γ2 subunit. This antibody has been used previously in immunofluorescence and immunohistochemistry studies (Sergeeva et al. 2002; Vassias et al. 2005).

One series of sections per receptor subunit—GABAAR α1, GABAAR γ2, and GABAB R2-, was processed for dual colorimetric labeling to simultaneously visualize the cholinergic cells and immunoreactivity against each GABAR subunit. The histochemical staining for nicotinamide adenine dinucleotide phosphate-diaphorase (NADPH-d) was carried out first to visualize the cholinergic neurons of the PPN and LDT (Vincent et al. 1983). Free-floating sections were incubated in Tris buffer (0.1 M, pH 8) containing 1 mM β-NADPH (Sigma), 1 mM Nitroblue Tetrazolium (Sigma), and 0.3% Triton X-100 (Tx), at room temperature (RT). The histochemical reaction was visually controlled, taking generally ~5 min. After rinsing with PBS and prior to the immunocytochemical procedure the sections were incubated in darkness in a quenching solution containing 0.3% hydrogen peroxide (H₂O₂) in 50% ethanol for 30 min at RT, to remove endogenous peroxidase activity. After rinsing, the sections were preincubated for 1 h in a blocking solution containing either 10% of normal horse serum (NHS) and 1% bovine serum albumin (BSA) for GABAB R2- and GABAAR α1-immunoreactivities (GABAB R2- and GABAAR α1-ir), or 4% BSA for GABAAR γ2-ir, followed by incubation in the primary antibody solution. All solutions were prepared in PBS and incubations carried out at RT unless otherwise specified. In the case of GABAB R2 and GABAAR α1, the mouse antibodies were diluted in a solution containing 1% NHS, 1% BSA, and 0.3% Tx and incubated for 48 h at 4 °C; the rabbit anti-GABAAR γ2 solution was prepared in 1% BSA and 0.3% Tx and incubated overnight. Tables 1 and 2 summarize the information about the primary and secondary antibodies used in the present study. After rinsing the sections were incubated with either biotinylated horse anti-mouse in 1% NHS and 1% BSA for 2 h, or goat anti-rabbit IgG in the same solution without NHS. The blocking solutions prior to primary and secondary antibody incubations were the same in all experiments unless otherwise specified. Next, the sections were incubated for 1 h in the avidin–biotin complex (ABC standard, Vector laboratories, Burlingame, CA, USA) and followed by 2X rinses. After pre-equilibrating in 0.1 M Tris HCl pH 7.6, the sections were finally incubated in a diaminobenzidine tetrahydrochloride (DAB) solution containing 0.022% DAB and 0.003% H₂O₂ in Tris HCl for 15 min and stopped with Tris HCl; then they were mounted on slides using 1:3 PB, dehydrated through graded alcohols, defatted in xylene, and coverslipped with DPX (VWR, Leuven, Belgium).

A dual fluorescence labeling protocol was used to investigate the potential colocalization of GABAAR- or GABABR-immunoreactivity (GABAAR- or GABABR-ir) with cholinergic neurons using an immunohistochemistry against choline acetyltransferase (ChAT). To enhance the visualization of GABAAR α1- and GABAB R2-ir, a water-bath heating was carried out for antigen retrieval, and an amplification system was used (TSA® detection kit). Thus, the free-floating sections were rinsed and transferred to a PB solution preheated and maintained at 80ºC for 30 min by a surrounding water-bath. Once the sections were lukewarm they were incubated in the quenching solution mentioned previously, for 20 min (Luquin et al. 2010). After rinsing, the sections were preincubated in their respective blocking solutions, followed by incubation in the primary antibody solution containing either mouse anti-GABAB R2 or mouse anti-GABAAR α1 and goat α-ChAT, which was added 16 h before the end of the incubation. The sections were then incubated in a solution with Alexa Fluor 488 donkey anti-goat IgG, for 2 h in darkness. After rinsing, sections were incubated with biotinylated horse anti-mouse IgG in 1% NHS in TNB [0.5% blocking reagent in TN buffer (0.1 M Tris–HCl, pH 7.5, 0.15 M NaCl)] for 45 min, then rinsed once in TNT buffer (TN, 0.05% Tween 20) and 2X in TN, they were incubated in a solution containing streptavidin-conjugated horseradish peroxidase (1:500, TSA^® detection kit) in TNB buffer for 30 min, followed by incubation for 10 min with Biotin–tyramide (1:250 in amplification diluent from TSA™). Finally, the sections were incubated with streptavidin-Alexa 633 for 90 min in TNB, and they were mounted on gelatine-coated slides, air dried, defatted in toluene, and covered with DPX. The sections were then examined under a confocal microscope (LSM 880, Zeiss).

For GABAAR γ2-immunofluorescence detection, neither antigen retrieval nor amplification was needed. Instead, the sections were incubated in the corresponding blocking solution, followed by incubation in the primary antibody solution containing rabbit anti-GABAAR γ2 and goat anti-ChAT, and finally incubated in Alexa Fluor 488 donkey anti-goat IgG to label the cholinergic neurons and Alexa Fluor 555 donkey anti-rabbit IgG to visualize GABAAR γ2 neurons.

After inactivation of the endogenous peroxidase activity and incubation in their respective blocking solutions, the sections were incubated in the primary antibody solution containing either mouse α-GABAB R2 or mouse α-GABAAR α1, while goat α-ChAT and rabbit α-Neuronal nuclear protein (NeuN) were added 16 h before the end of the incubation. To enhance the visualization of GABAAR α1 and GABAB R2, the TSA^® Plus fluorescein detection kit was used. Thus, the sections were incubated in a solution with biotinylated horse anti-mouse IgG in 1% NHS in TNB for 45 min and after rinsing in TNT buffer and TN, the sections were incubated in a solution containing streptavidin-conjugated horseradish peroxidase (1:200, TSA^® Plus fluorescein detection kit) in TNB for 30 min and then incubated for 10 min with FITC-tyramide (1:100 in amplification diluent from TSA Plus kit). After rinses with PBS, the sections were incubated in the secondary antibody solution containing Alexa Fluor 546 donkey anti-goat IgG and an Alexa Fluor 647 donkey anti-rabbit IgG, and finally mounted.

For GABAAR γ2-immunofluorescence detection, the sections were incubated first in the respective blocking solution and then in a solution containing mouse anti-NeuN, rabbit anti-GABAAR γ2 and goat anti-ChAT. To visualize NeuN, the sections were incubated with a biotinylated horse anti-mouse IgG antibody with 1% BSA for 2 h and then in a streptavidin-Alexa 633 solution for 90 min; subsequently they were incubated in a solution containing Alexa Fluor 488 donkey anti-goat IgG and Alexa Fluor 555 donkey anti-rabbit IgG to visualize the cholinergic neurons and GABAAR γ2-positive neurons, respectively.

Sense and antisense riboprobes of rat GAD67 and Vglut2 were transcribed as described previously (Erlander et al. 1991; Stornetta et al. 2002a,b; Tillakaratne et al. 1992). GAD67 plasmids were generously donated by Drs. A.J. Tobin and N.J.K. Tillakaratne (Department of Biology, University of California, Los Angeles, CA), while the Vglut2 plasmid was kindly gifted by Drs. R.L. Stornetta and P. Guyenet (Department of Pharmacology, University of Virginia, Charlottesville, VA). The riboprobe synthesis was carried out as previously described (Luquin et al. 2018).

A triple fluorescence labeling protocol was used to investigate the potential colocalization of GABAB R2 with either GAD67 or Vglut2 mRNA, carrying out ISH first followed by immunohistochemistry against ChAT. In these sets of experiments, brain tissue from rats perfused with DEPC-treated solutions was used.

For ISH, we first determined the optimal concentrations of GAD67 and Vglut2 sense and antisense digoxigenin riboprobes, which were 110 and 83 ng/mL, respectively. Then, the selected sections were processed for the triple fluorescent protocol. The ISH was carried out first, following a previously described protocol (Luquin et al. 2018). After rinsing with PBS the sections were processed for immunofluorescence, undergoing first preincubations in the respective blocking solutions for 40 min, followed by an overnight incubation in darkness in a solution containing anti-GABAB R2 and anti-ChAT, in which Tx was specifically omitted. Finally, sections were processed according to the dual immunofluorescence labeling protocol.

A similar triple fluorescence labeling protocol was used to investigate the colocalization of GABAAR γ2 with GAD67 and Vglut2 mRNA. The optimal concentrations of GAD67 and Vglut2 sense and antisense biotin riboprobes in these cases were 50 and 194 ng/mL, respectively. Selected sections were then processed for ISH as previously described (Luquin et al. 2018) and the biotin riboprobes incubated with streptavidin-Alexa 633 in TNB in darkness. After this, the sections were processed for immunofluorescence undergoing first a preincubation in the respective blocking solution and followed by an overnight incubation in a solution containing anti-ChAT and anti-GABAAR γ2 antibodies. After that, the sections were incubated in a solution containing 1% BSA, an Alexa Fluor 488 donkey anti-goat IgG and an Alexa Fluor 555 donkey anti-rabbit IgG in PBS, and finally mounted.

To analyze the potential colocalization of markers, the full rostrocaudal extent of the PPN or LDT was studied in series of one out of six sections (9–10 sections per experiment). In every section, the whole extent of both PPN and LDT was scanned and photographed with a confocal microscope (LSM 800; Zeiss) using a 40× oil-immersion lens with differential interference contrast. The most peripheral cholinergic neurons were used as reference to establish the nuclear boundaries. A complete series of optical sections (Z-stack) was acquired from every field and the interval between every two slices was adjusted at 0.48 μm. Every Z-stack was then analyzed slice-by-slice to determine the potential colocalization of markers on the same plane; thus, all single and dually labeled cells were counted using Zen lite 2012 software (https://​www.​zeiss.​com/​microscopy/​int/​products/​microscope-software/​zen-lite.​html).

To investigate potential regional differences in the expression of receptors between the anterior and posterior portions of the PPN and LDT, in each animal we calculated the percentage of GAD67- and VGlut2-positive cells co-expressing each GABAR subunit in the rostral sections of each nucleus, and statistically compared it with the percentage calculated in the caudal ones (Wilcoxon signed-rank test). When a series of sections had an uneven number of them, the extra section was systematically assigned to the anterior portion. The same calculations and analyses were carried out for ChAT-positive cells expressing the GABAAR α1 subunit.

The optical fractionator method (West and Gundersen 1990; Boyce et al. 2010) was used to obtain the total counts of the different cell subpopulations of PPN and LDT because it allows to establish cell numbers independently of volume estimates, eliminating most potential biases due to tissue shrinkage.

The first section for the neuronal counts was randomly selected from the first six containing the PPN, and then one every other 6 (240 µm) was systematically selected throughout the full rostrocaudal extent of PPN and LDT, resulting on average, in 9 sections per animal. Once processed, the sections were analyzed using an Olympus Bx-UCB microscope (Olympus Optical Co, Europe GmbH, Hamburg, Germany) connected to a digital camera (DP71, Olympus) and supplied with a motorized microscope stage ProScan (Prior Scientific Inc, Rockland, MA). The microscope was guided by a computer supplied with NewCast software (NewCast, v.2.16.1.0; Visiopharm, Denmark) which provided a systematic, random, and uniform sampling of optical disectors across tissue sections. These were first examined at 2×; the closed contours of PPN and LDT were outlined at 10X, and cell counts were obtained using a 100× 1.4 NA oil-immersion objective.

The total cell counts of each neurochemical subpopulation were estimated using the following equation:

where ssf is the section sampling fraction, asf the area sampling fraction, hsf the height sampling fraction, and ΣQ⁻ the cells counted in every region. ssf was calculated as T/BA, where T represents the distance between sections (240 µm), and BA is the block advance or thickness set at the microtome (40 µm; asf is calculated as (Dx × Dy)/a, where Dx and Dy represent the step length in the X- and Y-axes (155.57 µm in both cases), and a is the counting frame area (6050.7 µm²); finally hsf is calculated as ^̅t_Q⁻ /h, where ^̅t_Q⁻ is the number-weighted mean section thickness and h the height of the disector. Cell number estimations obtained with the optical fractionator design are not affected by tissue shrinkage in the X- and Y-axes; however, to avoid a potential bias due to differential tissue deformation in the Z-axis we used the number-weighted mean section thickness or ^̅t_Q⁻ (Bermejo et al. 2003; Dorph-Petersen et al. 2001). Finally, as the mean section thickness was 13.6 ± 2 µm, the disector height was set at 9 µm, keeping an upper guard zone of 2 µm and a lower one of variable height (2.5 µm on average). Our counting unit was the equator plane of the cell soma, which is the plane of the cell with most sharp borders and it is normally visible in 1–2 microns thickness at most. A cell was counted if the equator was in focus within the height of the disector, which was automatically signaled by the program, and did not touch the forbidden sides (left and bottom) of the disector frame (Luquin et al. 2018).

The sampling fraction was previously determined in a pilot study so as to ensure that a minimum of 100 cells per case from each neuronal phenotype were counted separately in the PPN and LDT, resulting in a coefficient of error (CE) ≤ 0.1 for each of them. The CE was calculated using Eq. 20 from Gundersen et al. (Gundersen et al. 1999). The number of disectors counted ranged between 4 in the smallest areas, and 46 in the largest ones. The volume (V) of PPN and LDT was calculated following the Cavalieri principle (Gundersen and Jensen 1987) using the formula:

where T represents the distance between sections (240 µm; a, is the area per point (0.024 mm²), and ΣP is the sum of points counted. Once the volume was calculated, the cell densities (Nv) were finally estimated using the formula:

Here we have provided the first anatomical evidence of the presence of GABAAR and GABAB receptors in the different cell phenotypes in the PPN and LDT. A detailed quantitative analysis of GABAAR γ2 and GABAB R2 subunits in the two nuclei was carried out, and the mean total number of cells expressing each subunit was estimated. Furthermore, we also quantified the number of cells expressing each receptor subunit within each neurochemical subpopulation. While all cholinergic cells in both nuclei expressed GABAAR γ2 and GABAB R2, only a certain proportion of the total glutamatergic and GABAergic cells expressed the γ2 subunit, suggesting that two functionally heterogeneous subsets of cells may coexist among cells of the same phenotype. The proportion of GABAAR γ2-expressing cells in the two non-cholinergic cell subpopulations differed between the PPN and LDT. This, together with a marked difference between the two nuclei regarding the presence of GABAAR α1 subunit in the cholinergic cells might contribute to functional differences between the PPN and LDT.

Downstream projections from the basal ganglia output nuclei directly target the three cell phenotypes in the PPN and LDT, and as we show here, they are all endowed with GABA receptors enabling GABAergic transmission. What specific cell subsets of cholinergic and non-cholinergic cells are actually controlled by these projections, and what local and/or reciprocal connections to basal ganglia components do they establish, is a matter of future research that will contribute to understand the role of PPN and LDT within basal ganglia circuitry.

Marked differences were found between the PPN and LDT in relation to GABAAR subunit composition in the cholinergic subpopulation, as well as to the synaptic versus extrasynaptic GABAARs in their non-cholinergic subpopulations. The significance of GABAAR heterogeneity for brain function is not yet understood. What we know is that the diverse dynamic characteristics of different GABAAR subtypes give rise to distinct types of inhibitory control, some of which may coexist in a given nucleus (Ye et al. 2017), while the presence of GABABRs likely provides additional types of inhibitory control. Additional work on GABAAR subunit combinations and their particular kinetics, as well as on their potential interactions with GABABRs will help understand inhibitory neurotransmission in these nuclei and may lead to a better understanding of functional differences between the PPN and LDT.