Three hundred and fifty micrometer acute slices from 3 to 4 animals of the same age and genotype were prepared to dissect out and pool the striatum under a Nikon SMZ800 bright field microscope with a SCHOTT KL1500 LCD illuminator (Du et al., 2016). RIPA buffer (Sigma-Aldrich, R0278) with a protease inhibitor cocktail (Roche Diagnostics, 11836153001) was used for protein extraction at a ratio of 20 μL per mg of tissue. The brain tissues were disrupted and homogenized by a sonicator (Ultrasons, Annemasse, France) on ice. The homogenates were centrifuged (1–15k, Sigma) at 13000 rpm for 30 min at 4°C. The supernatants were collected as total cell lysates. Protein concentration was measured with a RC DC^TM protein assay kit (BIO-RAD) according to manufacturer instructions. Samples were divided into 15 μg protein aliquots and kept frozen until used. Samples were supplemented with loading buffer, proteins were separated on SDS-PAGE and revealed by Western blotting using specific antibodies (Table 1), and analyzed as previously described (Chaumont et al., 2013). Briefly, protein bands were detected using an ECL chemiluminescence detection system (Lumiglo/Eurobio) with autoradiography film (Amersham Hyperfilm^TM ECL). Bands were quantified using ImageJ software, and the results obtained for each sample were normalized to the amount of GAPDH measured. The mean value in the WT littermate group, which was processed in parallel, was taken as 100%. Note that, as specified in the corresponding figure legends, each data set is representative of four independent western blot analyses. Moreover, individual analyses were performed with pooled brain structures from at least three mice of the same age and genotype. Therefore, we believe that our data represent the average protein expression of the different groups of mice and avoids individual variability.

Tissue samples were homogenized in Tri-reagent (Euromedex, France) and RNA was isolated using a standard chloroform/ isopropanol protocol (Chomczynski and Sacchi, 1987). RNA was processed and analyzed following an adaptation of published methods (Bustin et al., 2009). cDNA was synthesized from 2 μg of total RNA using RevertAid Premium Reverse Transcriptase (Fermentas) and primed with oligo-dT primers (Fermentas) and random primers (Fermentas). q-PCR was performed using a LightCycler^® 480 Real-Time PCR System (Roche, Meylan, France). q-PCR reactions were made in duplicate for each sample, using transcript-specific primers, cDNA (4 ng) and LightCycler 480 SYBR Green I Master (Roche) at a final volume of 10 μl. The PCR data were exported and analyzed using a software tool (Gene Expression Analysis Software Environment) developed at the NeuroCentre Magendie. For determination of the reference gene, the Genorm method was used (Bustin et al., 2009). We analyzed multiple reference genes for normalization of the q-PCR data: glyceraldehyde-3-phosphate dehydrogenase (Gapdh), actin-beta (Actb), a hydroxymethylbilane synthase (Hbms) ribosomal protein L13a (Rpl13a) succinate dehydrogenase complex, subunit A (Sdha), ubiquitin C (Ubc), tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta (Ywhaz) eukaryotic translation elongation factor 1 alpha 1 (Eef1a1), peptidylprolyl isomerase A (Ppia), tubulin, alpha 4a (Tuba4a), non-POU domain containing, octamer-binding (Nono) glucuronidase, beta (Gusb). Relative expression analysis was corrected for PCR efficiency and normalized against Gapdh and Ppia genes that were the most stably expressed. The relative level of expression was calculated using the comparative (2^-ΔΔCT) method (Livak and Schmittgen, 2001). Primer sequences are reported in Table 2.

Animals were anesthetized with ketamine/xylazine and perfused transcardially with ice-cold modified artificial cerebrospinal fluid (ACSF), equilibrated with 95% O₂–5% CO₂, and containing (in mM): 230 sucrose, 26 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, 0.5 CaCl₂, 10 MgSO₄, and 10 glucose. After removal, brains were submerged in ice-cold modified ACSF and sectioned in 350 μm thick slices with a vibratome (VT1200S; Leica Microsystems, Germany) in the parasagittal plane. Slices were fixed in 4% paraformaldehyde for 30 min and processed into 16 μm-thick sections collected on gelatin-coated slides as previously described (Moragues et al., 2002; Schneider Gasser et al., 2006). Slices from the same hemisphere were cut serially in 16 μm thick sections (60–80 sections for one hemisphere) and placed on gelatin coated slides, one section per slide. All sections were stored at -80°C until used. Labeling were carried out on brains obtained from 3 mice for each age and each genotype.

After 1 h incubation in blocking solution containing phosphate-buffered saline (PBS), pH 7.4, 4% donkey serum and 0.3% Triton X-100 (Sigma), slides were incubated overnight at room temperature in primary antibodies (Table 1) diluted with the blocking solution. They were then rinsed in PBS and incubated for 1 h in blocking solution containing a cocktail of secondary antibodies conjugated to fluorescent probes (Alexa-488, -568, or -647-conjugated donkey anti-mouse, -rabbit, -guinea pig, or goat antibody; Jackson Immunoresearch). Sections were washed and coverslipped. For quantification of α1 subunit in PV cells and the neuropil, PV was labeled with visible-wavelength emission (Alexa Fluor 488) and α1 with far-red-emitting dye invisible to the naked eye (Alexa Fluor 647). For each animal, 8 striatal serial sections (one of every 10 sections) were taken for labeling and analysis. Observation and acquisition were performed with a Leica DM6000B microscope (Leica Microsystems, Mannheim, Germany) equipped with a Qimaging RETIGA EXI camera. PV cells were identified by eye and imaged. PV labeling was used to draw the outline of the cell and define a region of interest (width 2 μm) including the cell periphery and surrounding external neuropil (Figure 6E). Mean fluorescence intensity of α1 labeling was then automatically measured using a custom-made macro in ImageJ software. The background was centered on the cell containing the nucleus, where labeling of GABA_AR subunits is assumed to be negative. Quantification was performed on 129 to 143 neurons per group from 3 independent experiments on brains obtained from three mice for each age and each genotype.

Double-labeling of α2- and a3-subunits on MSNs, PV-, nNOS-, and ChAT-positive interneurons, were performed as above and imaged with a BX51 Olympus Fluoview 500 confocal microscope using an oil-immersion 60 × objective and 1.4 numerical aperture. DARPP32, PV, nNOS or ChAT labeling was used to identify each neuronal cell type and to draw a 0.2 μm wide band that outlined identified cells and define regions of interest. Quantification of the number of receptor clusters on the cell periphery were performed using a custom-made macro in ImageJ software. During the analysis, an image in a region of interest was subjected to moment preserving thresholding, and clusters above 0.064 μm² were counted. Counts were carried out on brains obtained from three mice for each age and each genotype.

For Western blot, q-PCR, acetylcholinesterase expression and striatum volume analyses, unpaired two-tailed Student’s t-test was applied to compare differences between two groups. Data are expressed as mean ± standard error of the mean (SEM) and were analyzed statistically with Graph-Pad Prism (GraphPad Software Inc.). “n” indicates the number of independent biological replicates used in each group. In image analyses, test of normality (Shapiro test) and analysis of equal variances (Bartlett-Levene test) were performed. When the two groups expressed a normal distribution (Shapiro test) and had the same variance (Bartlett-Levene test), a two-tailed t-test was applied. In other cases, a Mann–Whitney U test was used. The significance level of the tests (p-value) was 5%. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001.

To evaluate GABAergic neurotransmission during HD progression, we quantified expression levels of several indicators essential for GABAergic synaptic function using Western blot on protein extracted from the striatum of mice at 2 and 6 months of age. The expression level of the GABA_AR α1 subunit in R6/1 mice at 2 months was not modified statistically (WT: 100.0 ± 9.9; R6/1: 133.8 ± 22.6%) but was significantly increased (WT: 100.0 ± 3.5; R6/1: 159.9 ± 22.1%) at 6 months (Figure 1A). GABA_ARs present in synapses are associated with the γ2 subunit, while extrasynaptic receptors are mainly associated with GABA_ARs containing either α5 or δ subunits (Luscher et al., 2011). Expression of the γ2 subunit (Figure 1B) in R6/1 mice at 2 months (WT: 100.0 ± 1.4; R6/1: 108.2 ± 18.6%) and 6 months (WT: 100.0 ± 2.5; R6/1: 109.5 ± 8.1%) was not modified statistically, while the expression of the δ subunit showed a non-significant then significant decrease (WT: 100.0 ± 2.8; R6/1: 85.6 ± 8.2%, WT: 100.0 ± 2.1; R6/1: 67.4 ± 6.7%) at 2 and 6 months, respectively (Figure 1C). In direct contrast to δ, however, a non-significant change in α5 at 2 months (WT: 100.0 ± 4.9; R6/1: 145.8 ± 32.6%), became significant at 6 months (WT: 100.0 ± 3.7; R6/1: 196.8 ± 32.6%), in the R6/1 striatum (Figure 1D). GABA_ARs are pentameric complexes that contain β1-3 as mandatory subunits. Therefore, the expression of these subunits may reflect the expression level of GABA_ARs as a whole. The β3 subunit is the major subunit expressed in the striatum (Boyes and Bolam, 2007; Hortnagl et al., 2013). However, no change in expression level was detected for β3 at 2 or 6 months in R6/1 mice (Figure 1E). Furthermore, the β2 subunit, which is moderately expressed in the striatum, displayed an increased expression at 2 months that was then non-significant at 6 months (WT: 100.0 ± 2.9; R6/1: 152.1 ± 14.0% and WT: 100.0 ± 5.7; R6/1: 125.8 ± 20.0%, respectively; Figure 1F).

Because GABAergic neurotransmission is also regulated by presynaptic mechanisms, the expression level of proteins involved in GABA synthesis or release was also measured. The level of GAD 65 was not altered in the striatum at 2 and 6 months in R6/1 (Figure 2A) while a significant decrease in GAD 67 levels was measured at 2 months (WT: 100.0 ± 4.4; R6/1: 82.1 ± 2.2%), but not at 6 months (Figure 2B). Consistent with these findings, no significant change was observed at 6 months with an antibody that recognizes both GAD 67 and GAD 65 (Figure 2C). Moreover, no significant changes in the expression of the vesicular transporter Vgat were observed (Figure 2D). Because GABA release by VMAT2-containing vesicles has been reported in the striatum (Tritsch et al., 2012), we measured the expression level of this vesicular monoamine transporter (Figure 2E); no significant changes were seen in the striatum of R6/1 mice. Finally, gephyrin and neuroligin 2 are predominantly associated with inhibitory synapses (Fritschy et al., 2008; Poulopoulos et al., 2009). The expression levels of these two proteins were not changed in the striatum of either 2 or 6 month-old R6/1 mice (Figures 2F,G).

To corroborate our findings in the R6/1 model, we tested the HdhQ111 knock-in mouse line that is genetically distinct from R6/1 and has a different disease progression phenotype (Crook and Housman, 2011; Pouladi et al., 2013). Western blot analysis of membrane preparations extracted from the striatum of 8 month-old HdhQ111 mice revealed a statistically significant increase in GABA_AR α1 subunit expression levels (WT: 100.0 ± 2.9; R6/1: 173.4 ± 16.8%, Figure 3A). In addition, the level of GAD 65 was significantly increased (WT: 100.0 ± 4.1; R6/1: 138.3 ± 12.6%) whereas the expression of GAD 67 was non-significantly reduced WT: 100.0 ± 4.1; R6/1: 98.5 ± 8.4%, Figures 3B,C). These results therefore are in accord with our data from the R6/1 mouse.

Protein expression analysis is an accurate way to assess gene expression downstream of transcriptional and translational regulation (Tian et al., 2004). However, Western blot is hampered by the limitation of finding suitable antibodies. Therefore, to validate and extend our findings, we performed quantitative PCR on mRNAs extracted from the striatum of 6 month-old WT and R6/1 mice (Figure 4). The significantly increased expression of striatal α1 subunit protein was not associated with a significant increase in mRNA expression, whereas the finding of a significant increase or decrease in the expression of α5 and δ mRNAs, respectively, was well correlated with our Western blot data (see Figures 1C,D). It should be noted that the mRNA of the α4 subunit that is associated with the δ subunit within extrasynaptic GABA_ARs was found to be also down-regulated in R6/1 mice. Interestingly, our q-PCR experiments showed that mRNA expression of α2 is decreased while the α3 subunit is increased in the striatum. Finally, the mRNA expression of the mandatory β subunits (either highly expressed β3 or the low-abundant β2 and β1 subunits) was not significantly modified. We also performed q-PCR on mRNAs encoding GAD 65, GAD 67, Vgat, neuroligin 2 and gephyrin extracted from the striatum (Figure 4). These experiments showed a decreased expression of GAD 67, Vgat and gephyrin mRNAs.

Taken together, our analyses of GABA_AR subunit expression and pre- and post-synaptic protein markers of inhibitory synapses revealed subtle and complex changes in the striatum of R6/1 mice, with major alterations in α1, α2, and α3 subunit expression suggesting a change in the GABA_A receptor subtype. We therefore focussed attention on the expression of these three subunits in the striatum of WT and R6/1 mice.

Our Western blot data described above showed an increase in GABA_AR α1 subunit expression in the striatum of R6/1 mice. Since this subunit plays a major role in the kinetics of inhibitory post-synaptic currents (Luo et al., 2013), we performed multiple immuno-labeling on brain sections to assess its expression at the cellular level (Figure 5). Some staining was detected on MSN cell bodies revealed by DARPP 32 labeling, while major staining was found throughout the striatum, likely reflecting the subunit’s widespread distribution in the neuropil (Figure 5A). Parvalbumin (PV) interneurons were decorated with α1 puncta on their cell body membranes and proximal dendrites (Figure 5B). No co-localization with either calretinin or somatostatin dense-core vesicle labeling was detected (Figure 5C). Cholinergic neurons labeled by choline acetyltransferase (ChAT) antibodies did not display detectable α1 labeling (Figure 5D). In addition to the α1 labeling on PV interneurons, we also found a stronger labeling on NeuN positive-PV negative cells (Figures 5E,F). These neurons that expressed a high level of α1 are sparsely distributed (∼1 cell per section) and are characterized by a large dendritic tree displaying varicosities (Figure 5F). Labeling of 2 and 6 month-old R6/1 mice showed that this strong α1 staining on non-identified neurons was virtually absent in the mutant striatum.

The expression of α1 on PV neurons and in the neuropil was further analyzed in striatal sections from 2 and 6 month-old mice (Figure 6). While fluorescent α1 labeling was clearly evident in confocal images of PV neurons from WT mice at 2 and 6 months (Figures 6A,C), it was less pronounced in R6/1 littermates at both 2 and 6 months (Figures 6B,D). To further assess this apparent overall decrease, the mean fluorescence intensity of α1 labeling was measured on the cell body periphery of PV-positive cells and on the surrounding neuropil (Figure 6E) and normalized to the corresponding expression values for PV positive cells from WT mice (Figures 6F,G). This quantitative analysis showed that a significant decrease in α1 subunit expression (83.8 ± 5.4% compared to WT PV cells) had indeed occurred on the somata of striatal PV positive interneurons in 2 month old R6/1 mice (Figure 6F). In contrast, the expression of neuropilar α1, which in WT was 78.7 ± 3.4% of the corresponding cell body expression, was increased significantly to 109.2 ± 4.8% in R6/1 mice (Figure 6F). Similarly at 6 months, α1 expression on PV neuron somata was decreased to 79.2 ± 5.8%, whereas the subunit’s presence in the neuropil, which in WT was 77.8 ± 4.4% of soma expression, had increased to 151.5 ± 10.2% in R6/1 mice (Figure 6G). Therefore, altogether these data support the conclusion that the increase in striatal α1 subunit presence in the R6/1 mutant revealed by Western blot analyses (Figure 1A) is due to an increased expression of α1 in MSNs, although intriguingly, this cell-specific increase is associated with an early decrease in the subunit’s expression in PV interneurons.

Our q-PCR analyses also indicated a decreased expression of the α2 subunit in the striatum of 2 and 6-month-old R6/1 mice (see Figure 4). However, subsequent Western blot comparison of striatal tissue protein extracted from WT and R6/1 mice did not indicate that significant subunit changes had occurred (Figure 7A). We therefore performed multiple immuno-labeling on brain sections to assess the expression of the α2 subunit at the cellular level. More precisely, striatal sections from 2 and 6 month-old WT and R6/1 mice were subjected to double-immunostaining in order to label the α2 subunit co-localized with dopamine- and cAMP-regulated neuronal phosphoprotein (DARPP32) that is only expressed in MSNs, or acetylcholinesterase, nitric oxide synthase and Calretinin that are specific to striatal GABAergic and non-GABAergic interneurons (Waldvogel et al., 1999; Ghiglieri et al., 2012; Calabresi et al., 2014). Staining of α2 was detected on MSN projection neurons, recognized by the presence of DARPP32, and GABAergic interneuron cell bodies located by co-staining with either PV or nNOS (Figures 7B–D). Subsequent analysis showed a significant increase of α2 subunit expression in striatal MSN and PV positive interneurons of R6/1 mice at 2 months and a significant expression decrease in these neuron subtypes at 6 months (Figures 7B,C). There was a significant decrease of α2 puncta in nNOS interneurons at 2 months, but no significant change was detected at 6 months (Figure 7D). These data thus suggest that, while the overall expression of α2 as revealed by Western blot analyses remains unchanged in the striatum of R6/1 mice, the subunit’s expression in specific neuronal subtypes is altered.

Our q-PCR analyses also showed an increased expression of the α3 subunit in R6/1 mice. Correspondingly, Western blot analysis showed no significant change in R6/1 mice at 2 months and a significant increase (WT: 100.0 ± 8.4%; R6/1: 335.7 ± 29.5%) at 6 months (Figure 8A). Moreover, using the HdhQ111 knock-in mouse model, Western blot analysis of membrane preparations extracted from the striatum of 8-month-old mice also revealed a significant increase in the GABA_AR α3 subunit expression level (WT: 100.0 ± 3.6; HdhQ111: 128.2 ± 7.9%, Figure 8B). We then performed multiple immuno-labeling of R6/1 striatal sections to assess the expression of α3 subunit at the cellular level. Staining was detected on interneuron cell bodies and proximal dendrites identified as cholinergic by Choline Acetyltransferase (ChAT) co-labeling (Figure 8C). Quantitative analysis (Figure 8C, right) showed that α3 subunit expression in such striatal interneurons is significantly enhanced at 6 months in R6/1 mice, suggesting that the subunit’s altered presence revealed by q-PCR and western blot analyses is due, at least in part, to an increased expression in cholinergic neurons.

Because the expression of GABA_AR subunits is already modified at 2 months in our R6/1 mice, we sought evidence to correlate this finding with actual brain pathology. Using acetylcholinesterase staining (Figures 9A,B), we found no change in acetylcholinesterase levels at 2 months (WT: 100 ± 2.4%; R6/1: 95.3 ± 3.0%; n = 5), but by 6 months, a dramatic reduction was evident (WT: 100.0 ± 3.3%; R6/1: 64.07 ± 1.86%; n = 3). Consistent with previous findings (Suzuki et al., 2001; Massouh et al., 2008), this is in turn indicative of a decline in acetylcholinesterase activity in the older animal and the resultant impairment of cholinergic neurotransmission. Additionally in R6/1 mice, striatal volume was significantly reduced at both 2 months (WT: 10.57 ± 0.28 mm³; R6/1: 8.81 ± 0.41 mm³; n = 5) and 6 months (WT: 10.50 ± 0.220 mm³; R6/1: 7.30 ± 0.355 mm³; n = 7) (Figure 9C). This concurs with magnetic resonance imaging data showing a decrease in R6/1 brain volume at 9 weeks (Rattray et al., 2013), and is also in agreement with recent observations on human HD showing that regional brain atrophy begins several years before the emergence of diagnosable symptoms (Ross et al., 2014).