Duplicates of punch biopsies obtained from distinct neuroanatomical sites (VP, CeA, BLA, and BST) of HAD and LAD rats were immunoblotted with α1- or α2-specific antibodies, with results quantitated by densitometric scanning as previously described (Smith et al., 1998; Liu et al., 2011). Levels of α1 in HAD rats were significantly higher in the VP (Figure 1A) and CeA (Figure 1B) than in LAD rats, but the two lines expressed similar levels of α1 in the BLA and BST (Figures 1C,D). These findings are in direct contrast to those obtained for the α2 receptor subunit, the levels of which were similar for HAD and LAD rats at all the examined anatomical sites (Figures 1E–H). Our previous report in P rats also revealed an elevation in α1 subunits in the VP and CeA (Liu et al., 2011).

Having seen that α1 is overexpressed in the VP and CeA, we wanted to know whether it is associated with vulnerability to engage in binge alcohol drinking. We focused on the VP to determine if the findings observed in P rats could be extrapolated to HAD rats in a volitional model of binge alcohol drinking. It should also be noted that overexpression of α1 in the CeA of P rats failed to regulate binge drinking, in contrast to similar overexpression in the VP (Liu et al., 2011). These findings indicate that overexpression of the α1 subunit in the CeA in and of itself is not sufficient to regulate the neuroadaptional processes associated with binge drinking.

In the present study, we employed the siRNA technology to examine the contribution of the α1 subunit to binge drinking, as it provides a mechanism for inhibiting specific genes at individual loci. The sequence and specificity of the α1 siRNA used in these studies and the construction of the HSV-based vector for its delivery (pHSVsiLA1), have been described previously (Liu et al., 2011). pHSVsiNC, a scrambled siRNA sequence delivered with a similar HSV-based vector, was studied in parallel with pHSVsiLA1 as a specificity control. Both vectors express enhanced green fluorescent protein (EGFP) in order to facilitate their detection following delivery.

As described in the Section “Materials and Methods,” vectors were delivered into the VP by bilateral stereotaxic microinfusion, and transduction was assessed by confocal microscopy at 72 h post delivery by EGFP visualization. The image shown in Figure 2A for 1 of the 13 bilateral injection sites in the VP of a HAD rat given pHSVsiLA1 (Figure 2B) indicates that transduction was effective, with substantial clusters of transduced neurons seen near the injection site. We conclude that most EGFP-positive cells contain GABA_A receptors, because 90% of the synapses in the VP are GAD immunoreactive, and α1 is the most highly expressed GABA_A subunit in the VP (Churchill et al., 1991; Pirker et al., 2000).

To examine whether α1 expression is specifically inhibited by pHSVsiLA1, cohorts of HAD rats were given pHSVsiLA1, pHSVsiNC, or PBS in the VP, with levels of α1 protein determined at 3 days post-infusion via immunoblotting of duplicate sections with α1-specific antibody. Alpha-2-specific antibody was used as a control. pHSVsiLA1 caused a robust decrease in α1 levels relative to those in the PBS-treated animals, a finding that was not observed in HAD rats given pHSVsiNC (Figure 3A). Moreover, the levels of the α2 subunit were not altered (Figure 3B), consistent with the conclusion that α1 reduction is not secondary to compensatory increases in another subunit, notably α2, in the same tissues. Particularly significant from the standpoint of therapy, α1 expression was still significantly inhibited on day 17 after pHSVsiLA1 infusion, albeit at a less profound level than on day 3 post-infusion (Figure 3C), underscoring the specificity and protracted duration of pHSVsiLA1’s inhibitory effect. The long-term effect of pHSVsiLA1, at least in this animal model, is not an artifact of the experimental procedures. Alpha-2 expression was unchanged (Figure 3D), and levels of α1 expression on day 30 after pHSVsiLA1 infusion were similar to those observed both pre-surgery and for animals given PBS or pHSVsiNC (Figure 3E).

The cell culture data summarized in Figure 4 indicate that approximately 50% of the total α1 protein levels are biotinylated (i.e., expressed on the cell surface). Identical levels of biotinylated α1 protein were seen in cells treated with pHSVsiNC; however, treatment with pHSVsiLA1 caused a significant decrease in levels of biotinylated α1 protein. The data indicate that inhibition of α1 expression by pHSVsiLA1 also causes a significant reduction in surface protein expression.

Because pHSVsiLA1 specifically reduced α1 gene expression, including on the surfaces of cells, we wanted to know whether it interferes with the density of the GABA_A receptor. HAD rats were given pHSVsiLA1 or PBS in the VP and tissues were collected at 3 days after infusion, when levels of α1 expression are significantly decreased. The collected VP tissues were assayed for radioligand binding of [³H]EBOB, a specific radioligand for the non-competitive convulsant blocking site of the GABA receptor, as previously described (Korpi et al., 1995). Binding was significantly reduced in the pHSVsiLA1-treated rats, as revealed by competition of picrotoxin with [³H]EBOB (Figure 5). A saturation isotherm showed significant differences in specific binding between the group given pHSVsiLA1 (n = 10) relative to the PBS group (n = 10). [³H]EBOB binding (B max) was profoundly reduced from 861 ± 16 fmol/mg protein in the control to 554 ± 8 fmol/mg protein in the pSHVsiLA1-treated rats, with no significant change in affinity (K_d; control, 2.67 ± 0.24 nM; pSHVsiLA1, 2.70 ± 0.18 nM; Figures 5A,B). Reduced binding levels evidenced after pHSVsiLA1 treatment (64% of control levels) correlate well with the reductions in α1 protein expression on the cell surface (50% of control levels; Figure 4). A similar reduction was not seen in any of the neuroanatomical sites of the PBS control rats that were not injected with the virus, nor in animals given pHSVsiNC or pHSVsiLA1 at sites that do not express high levels of α1 (data not shown), confirming the specificity of the pHSVsiLA1-mediated inhibition. Collectively, the data summarized in Figures 3–5 indicate that use of vector-delivered siRNA allows for specific inhibition of gene expression in one targeted neuroanatomical site, thereby providing the means to examine its contribution to alcohol drinking.

Having seen that pHSVsiLA1 specifically inhibits α1 expression and reduces the density of the GABA_A receptor in the VP, we wanted to know whether this reduction is associated with inhibition of binge alcohol drinking. Cohorts of HAD rats (n = 5–6 per group) were trained to self-administer alcohol (Bell et al., 2006; Integrative Neuroscience Initiative on Alcoholism (INIA)-West, 2008) and randomly administered, by cohort, pHSVsiLA1, pHSVsiNC, or PBS into the VP. After 3 days, during which the animals recovered from the stress of surgery, they were allowed to engage in daily alcohol drinking for 30 days. The results of these studies are summarized in Figure 6. In the rats given pHSVsiLA1, alcohol drinking was virtually eliminated on days 3–6 after infusion, when α1 expression was robustly inhibited. Drinking was still profoundly reduced by comparison to the PBS-treated animals on days 7–17 after pHSVsiLA1 infusion, when α1 expression was still inhibited. From days 18–26 the pHSVsiLA1-treated animals tended to show a daily elevation in responding. However, their intake levels were significantly lower compared with the control animals. By days 26–30 there was no differentiation in ethanol administration observed between the pHSVsiLA1 and control groups (Figure 6A). However, the precise time at which protein expression returned to pre-surgery levels in the pHSVsiLA1-treated animals is unknown. At 17 days post-infusion, α1 expression was significantly attenuated relative to control-treated animals, albeit markedly above that of animals 72 h post-treatment. At present, we are investigating the protracted time-course of protein reduction vis-à-vis reduction in volitional alcohol intake following administration of the α1 siRNA amplicon into the VP, in order to more precisely determine the significance of the relationship between the two. Nevertheless, we propose that the effect of pHSVsiLA1 is specific for alcohol, because water ingestion was similar between rats treated with pHSVsiLA1 or PBS control throughout the study interval (Figure 6B), and because the effects of pHSVsiLA1 was undistinguishable from those of PBS in HAD rats trained to drink sucrose in an analogous model (Figure 6C). We conclude that inhibition of alcohol intake by pHSVsiLA1 is related to α1 inhibition in the VP. This is not an artifact of vector-induced toxicity or inflammation, because alcohol drinking was not reduced by pHSVsiNC, which was constructed exactly like pHSVsiLA1, but does not inhibit α1 expression or interfere with receptor density (Figure 6D). Toxicity determined based on loss of body weight and general activity levels was not seen for pHSVsiLA1 or pHSVsiNC (Figures 7A,B).

The generation and specificity of the rabbit-derived GABA_A α1 and α2 antibodies have been described previously (Pirker et al., 2000). These recognize amino acids 1–9 and 322–357 of the α1 and α2 proteins, respectively. The antibody controls (GAPDH and actin) can be used interchangeably, based on availability. The GAPDH antibody was unavailable during the middle portion of the study period.

The siRNAs LA1 (α1 subunit) and NC (scrambled, non-specific), as well as the construction and functionality of the respective HSV-1-based amplicon vectors, are described in Table S1 and Figures S2–S4 in Supplementary Material (Liu et al., 2011). Construction used the basic plasmid pHSVsi that contains an HSV-1 origin of DNA replication (ori), the HSV-1 DNA packaging/cleavage signal (pac), and the HSV-1 IE 4/5 promoter to express the EGFP reporter gene. A second transcription unit consists of the RNA polymerase III-dependent H1 promoter and start and termination signals for siRNA synthesis (Saydam et al., 2005). The siRNAs were inserted into pHSVsi between the BglII and HindIII sites. Amplicon DNA was packaged into HSV-1 particles and the titers of the vector stocks were determined by counting the number of green cells at 24 h after transduction and expressing them as transduction units (TU)/mL.

Ventral pallidum tissues were harvested by micropunch (140 μm thick). Immunoblotting was conducted as previously described (Wales et al., 2007). Lysis was performed in RIPA buffer (20 mM Tris–HCl (pH 7.4), 0.15 mM NaCl, 1% Nonidet P-40, 0.1% sodium dodecyl sulfate, 0.5% sodium deoxycholate) with 1% phosphatase and protease inhibitor cocktails (Sigma, St. Louis, MO, USA), and total protein was determined by the bicinchoninic assay (Pierce, Rockford, IL, USA). Proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Blots were exposed to antibody overnight at 4°C, followed by horseradish peroxidase-labeled goat anti-rabbit IgG (Cell Signaling, Beverly, MA, USA) for 1 h at room temperature. Detection was performed using ECL kit reagents (Amersham Life Science, Arlington Heights, IL, USA) and quantitation was by densitometric scanning with a Bio-Rad GS-700 imaging densitometer.

Cerebellum, VP, CeA, and BST membrane homogenates were prepared from adult male HAD rats, and binding of [³H]EBOB was determined by filtration assay (Kralic et al., 2002). Briefly, homogenates were incubated with 6.0 nM [³H]EBOB in 50 mM Tris–HCl buffer (pH 7.4), containing 300 mM NaCl, at 23°C for 1 h. Non-specific binding was determined with 50 μM picrotoxin. After incubation, bound and free ligand were separated by rapid filtration over Whatman GF/B glass-fiber filters (presoaked in 0.05% polyethylenimine for 20 min), followed by three 4 mL washes with ice-cold 0.9% NaCl solution. Radioactivity was counted by liquid scintillation spectroscopy (Packard TRI-CARB 2900 TR, Downers Grove, IL, USA).

To examine whether pHSVsiLA1 alters the expression of the GABA α1 protein on the cell surface, we used the Pierce ® Cell Surface Protein Isolation Kit (Pierce Biotechnology, Rockford, IL, USA) according to the manufacturer’s instructions.

High-alcohol drinking rats were anesthetized by intraperitoneal injection of nembutal (50 mg/kg) and positioned in a stereotaxic apparatus as described previously (Liu et al., 2011). The selected microinjection sites in the rat VP extend from +2.2 mm anterior to bregma to −0.8 mm posterior to bregma, and from 0.5 to 3.0 mm lateral to the midline (Paxinos and Watson, 1998). Since amplicons do not diffuse over long distances in brain, a single large injection would fail to cover the entire VP and likely also result in a pressure lesion. Thus, we performed a pattern of 13 small injections in each hemisphere spaced across the entire VP (Figure 2B). At each site, 200 nL of either pHSVsiLA1 amplicon (2.5 × 10⁵ TU), pHSVsiNC (2.5 × 10⁵ TU) in PBS diluent or PBS control were injected using a calibrated pulled glass micropipette (∼20 μm tip) connected to a Picospritzer II pneumatic pressure injection apparatus (Science Products GmbH, Hofheim, Germany). Injections lasted 30 s and were separated by 1–2 min pauses for tissue recovery prior to insertion of the pipette at the next site. Acrylic microbeads were mixed with the amplicon and control injections to confirm accuracy of the microinjection placement based on the Rat Brain Atlas (Paxinos and Watson, 1998). All surgical and amplicon administration procedures were approved by the IACUC and Biosafety Committees at the University of Maryland School of Medicine.

To initiate excessive alcohol drinking in HAD rats, we used the drinking-in-the-dark-multiple-scheduled-access (DIDMSA) binge model developed by the Integrative Neuroscience Initiative on Alcoholism (INIA)-West, 2008, McBride and Li, 1998, Bell et al., 2006, which emulates the binge alcohol drinking patterns seen in humans (Naimi et al., 2003). Our experimental design comprised both pre- and post-surgery phases of home-cage binge alcohol drinking. In brief, rats were each presented with two 10 mL glass pipette vials equipped with stainless-steel sipper spouts and rubber stoppers. The pipettes were calibrated to allow for measurement of drinking for both alcohol and water. The 10% (v/v) alcohol solution was prepared using 10% alcohol (USP) and 90% deionized water (v/v). This pre-surgery phase included 21 days of consecutive alcohol drinking wherein HAD rats consumed alcohol for 1.5 h daily. The daily 1.5 h alcohol drinking periods were each divided into three 30 min bouts separated by 1 h intervals wherein rats received food and water in their home cages. During the non-binge periods (i.e., 22.5 h), rats also received food and water ad libitum. To determine if the rats were consuming binge alcohol levels of the 10% (v/v) alcohol solution, BAC levels were evaluated beginning on day 11 of the pre-surgery period, and were quantified every other day up to day 21, as previously reported (Harvey et al., 2002). Based on quantification of BACs, the HAD rats were stratified into two cohorts with BACs of 112 mg%/dL ± 13 (n = 6) or 118 mg%/dL ± 25 (n = 5). The two cohorts then received the PBS control or the pHSVsiLA1 amplicon vector, respectively. Seventy-two hours after administration of the experimental treatments, the post-surgery phase began. During the post-surgery phase, all rats consumed their daily 1.5 h binge alcohol for 15 consecutive days.

Data were analyzed by between-group ANOVAs and mixed repeated-measures ANOVAs for group and days. Significant ANOVAs were followed by the Tukey, Newman–Keuls or t-test post hoc tests. Analyses were performed using the StatMost 5.0 (Dataxiom Software Inc., Los Angeles, CA, USA) and GraphPad Prism 6.0 (GraphPad Software, Inc., La Jolla, CA, USA) programs.