Female-specific decreases in alcohol binge-like drinking resulting from GABAA receptor delta-subunit knockdown in the VTA

Binge drinking is short-term drinking that achieves blood alcohol levels of 0.08 g/dl or above. It exhibits well-established sex differences in GABAergic inhibitory neurotransmission, including extrasynaptic δ subunit-containing GABAA receptors (δ-GABAARs) that mediate tonic inhibition, or synaptic γ2-containing GABAARs which underlie fast, synaptic, phasic inhibition have been implicated in sex differences in binge drinking. Ovarian hormones regulate δ-GABAARs, further implicating these receptors in potential sex differences. Here, we explored the contribution of extrasynaptic δ-GABAARs to male and female binge-like drinking in a critical area of mesolimbic circuitry—the ventral tegmental area (VTA). Quantitative PCR revealed higher Gabrd transcript levels and larger tonic currents in the VTA of females compared to males. In contrast, male and female Gabrg2 transcript levels and measures of phasic inhibition were equivalent. Intra-VTA infusion of AAV-Cre-GFP in floxed Gabrd mice downregulated δ-GABAARs and decreased binge-like drinking in females. There was no significant difference in either male or female mice after GABAAR γ2 subunit reduction in the VTA following AAV-Cre-GFP infusion in floxed Gabrg2 mice. Collectively, these findings suggest sex differences and GABAAR subunit specificity in alcohol intake.


Materials and Methods
Animals. Floxed Gabrd mice were generated and previously characterized by our laboratory (Lee & Maguire 2013). Floxed Gabrg2 mice were acquired from The Jackson Laboratory (Stock #016830). All mice used in the current study were adults (8-14 weeks of age) and were maintained in-house at the Tufts University School of Medicine, Division of Laboratory Animal Medicine and were handled according to protocols approved by the Tufts University Institutional Animal Care and Use Committee. Given the already published findings demonstrating estrous cycle changes in GABA A R δ subunit expression in the VTA 24 , all female mice used in the current study were acyclic as previously established by our group and others 26,27 . Stereotaxic surgery. Adult male and female floxed Gabrd and Gabrg2 mice were stereotaxically injected with either AAV2-GFP or AAV2-Cre-GFP (Vector Biolabs) into the VTA using the following coordinates: A/P −3.6 mm, M/L ±0.5 mm, D/V 4.5 mm. This approach results in expression of GFP or Cre recombinase in a subset of neurons in the VTA.
Drinking in the dark. Approximately two weeks after surgery, mice were individually housed on a reverse light/dark cycle (12:12 h, lights off at 0700). One week later, they were subjected to the four-day Drinking in the Dark (DID), binge-like drinking protocol established by Rhodes et al. 28 . Volume differences across each access period were converted to individual g/kg values for each mouse based on mouse weights obtained on the day prior to DID start. BEC measurements. Blood ethanol concentrations (BECs) were measured on a subset of Gabrg2 and Gabrd DID mice. Submandibular blood samples were obtained immediately after the end of the 4 h binge session and plasma was isolated. An AM1 Alcohol Analyzer (Analox Instruments, Lunenburg, MA) was used to obtain all BEC values. Samples were run in duplicate and averaged to obtain a final value.
Sucrose preference and quinine aversion. A subset of mice were also subjected to successive sucrose preference and quinine aversion testing. Briefly, mice were provided access to either 2% (w/v) sucrose or normal drinking water. Bottle weights were obtained every 24 h (1200 each day) for four, consecutive days. The position of the bottles was alternated after each 24 h period to avoid side preference. The volume (mL) of water consumed was also calculated (1g H2O = 1mL H2O ) and consumption differences were used to calculate preference scores according to the following formula: (g sucrose /[g sucrose + g water ]). After three days of water access, mice were subjected to quinine aversion testing using the same procedure, but with 0.3 mM quinine monohydrochloride solution and water.
Immunohistochemistry and image analysis. GFP immunofluorescence was performed to enhance the GFP expression following stereotaxic injection of AAV-GFP and AAV-Cre-GFP. VTA slices (40 µm) were blocked with 10% normal goat serum (NGS) and incubated with a polyclonal anti-GFP primary antibody (rabbit, 1:1000, Invitrogen, Waltham, MA, USA) overnight at 4 °C and incubated with a polyclonal AlexaFluor 488 secondary antibody (rabbit, 1:200, Invitrogen, Waltham, MA, USA) for 2 h at room temperature. GFP immunostaining was imaged on a Keyence BZ-X700 (Keyence, Itasca, IL, USA) and analyzed using the Fiji distribution of ImageJ (NIH). Images were converted to binary masks prior to analysis and Integrated Density values were used to define signal intensity within a region of interest. Binary masks cause inversion of pixel values, thus greater ID indicates lower GFP signal.
Real-time quantitative PCR (qPCR). Sex-specific, baseline Gabrd and Gabrg2 transcript levels were compared between ethanol naïve wildtype males and females. Similar experiments were conducted on floxed Gabrd and Gabrg2 mice bilaterally infused with either AAV-GFP or AAV-Cre-GFP to assess gene knockdown.
RNA was isolated from the VTA and hippocampus using a QIAGEN RNeasy Mini Kit according to the manufacturer's instructions (Frederick, MD, USA) and the concentration and integrity was analyzed using an Agilent Genomics 2100 Bioanalyzer (Santa Clara, CA, USA). All samples with RNA integrity numbers (RIN) above 8.0 indicated highly intact RNA and were used to generate cDNA. Reverse transcription was performed using SuperScript III First-Strand Synthesis System according to the manufacturer's instructions (ThermoFisher Scientific, Cambridge, MA, USA). SYBR-green-based qPCR was performed using a Stratagene Mx3000P (Agilent Genomics, Santa Clara, CA, USA) with 3 ng of each VTA cDNA template and primers listed below (Table 1) using www.nature.com/scientificreports www.nature.com/scientificreports/ the following thermocycling protocol: 10 min ramp-up cycle to 95 °C, 40 cycles of 30 sec at 95 °C, 1 min at 55 °C, and 30 s at 72 °C, and a final dissociation curve at 95 °C for 1 min, 55 °C for 30 sec, and 95 °C for 30 sec. All reactions were performed in triplicate and a final, averaged C q value used for analysis. A C q value was excluded only if it was greater or less than 1 cycle relative to the remaining C q values. Relative transcript levels were normalized first to the reference gene β-actin and the relative changes in expression levels compared to controls were calculated according to the 2 −ΔΔct method (Livak & Schmittgen 2001). For the comparison between males and females, the transcript levels were normalized to the females since the Gabrd transcript levels were significantly higher in this group. For the knockdown experiments, each group was normalized to the AAV-GFP controls.
Electrophysiology. Adult Gabrd and Gabrg2 mice were bilaterally injected with either AAV-GFP or AAV-Cre-GFP and allowed to recover for three weeks prior to electrophysiological recording. Tonic and phasic GABAergic currents were measured using whole-cell voltage clamp recording techniques as previously described in visually-identified GFP expressing neurons in the VTA (Lee & Maguire 2013). Briefly, mice were anesthetized with isoflurane and decapitated, and the brain was rapidly removed and placed immediately in ice-cold, oxygenated normal artificial cerebrospinal fluid [nACSF; containing (in mM) 126 NaCl, 26 NaHCO 3 , 1.25 NaH 2 PO 4 , 2.5 KCl, 2 CaCl 2 , 2 MgCl 2 , and 10 dextrose (300-310 mOsm)] containing 3 mM kynurenic acid and bubbled with 95% O 2 -5% CO 2 . Coronal slices (350 mm thick) containing the VTA were prepared using a Leica VT1000S vibratome. For voltage-clamp recordings, slices were maintained in oxygenated nACSF containing 3 mM kynurenic acid. For current-clamp recordings, slices were maintained in oxygenated nACSF. After a 1 h minimum recovery period, VTA-containing slices were placed into a recording chamber maintained at 33 °C (in-line heater; Warner Instruments) and perfused at a high flow rate (4 ml/min) throughout the experiment. SR95531 (200 μM) was added to the extracellular solution where indicated.
For voltage-clamp recordings, the intracellular recording solution contained (in mM) 140 CsCl, 1 MgCl 2 , 10 HEPES, 4 NaCl, 0.1 EGTA, 2 Mg-ATP, and 0.3 Na-GTP (pH 7.25, 280-290 mOsm). Electrodes were used with DC resistance of 5-8 MOhm. Following stabilization of the holding current and series resistance and capacitance measurements, spontaneous inhibitory postsynaptic currents (sIPSCs) were recorded in GFP+ cells from both AAV-GFP and AAV-Cre-GFP slices over a five-minute period at a holding potential of −70 mV. The frequency, peak amplitude, and weighted decay (τ w ) of sIPSCs were measured using MiniAnalysis software (version 6.0.3, Synaptosoft Inc., Fort Lee, NJ, USA). Tonic GABAergic currents were measured as previously described 22,23 in visually identified GFP+ cells from both AAV-GFP and AAV-Cre-GFP slices. The mean current was measured during 10 ms epochs collected every 100 ms throughout the experiment. A Gaussian was fit to points over a 60 s period before and after bath application of SR95531 (>200 µM) to determine the mean holding current in nACSF and in the presence of SR95531. The difference in the holding current in the presence or absence of SR95531 was determined to be a measure of tonic GABAergic inhibition. Series resistance and whole cell capacitance were continually monitored and compensated throughout the course of the experiment. Recordings were eliminated from data analysis if series resistance increased by 20%.
Statistical analyses. All statistical analyses were conducted using Prism 7 (GraphPad La Jolla, CA, USA).
Repeated measures, two-way ANOVA was used to analyze DID 2 h drinking data, with AAV (GFP versus Cre) as the between-subjects factor and time as the within-subjects factor. Repeated measures, two-way ANOVAs were used to analyze both the sucrose preference and quinine aversion experiments, with AAV as the between-subjects factor and time as the within-subjects factor. An unpaired, two-tailed Student's t test was used to analyze DID binge, 4 h drinking data (GFP versus Cre). Unpaired, two-tailed Student's t tests were also used to analyze baseline and Cre-excision qPCR data (male versus female, GFP versus Cre, respectively). A Pearson's correlation was used to analyze the relationship between AAV transduction (optical density) and binge drinking as well as intake (g/ kg) and BECs. Data are presented as the average ± s.e.m. A p-value of p < 0.05 was considered statistically significant for all analyses.

Results
Sex differences in Gabrd mRNA expression in the VTA. Total mRNA was isolated from tissue punches from the VTA of males and females (Fig. 1a). The ratio of Gabrd mRNA to β-actin levels were higher in the VTA of females (females 1.23 ± 0.14 fold) compared to males (0.65 ± 0.16 fold) (t(6) = 2.78, p = 0.032, Fig. 1b). Consistent with higher Gabrd transcript levels in females, tonic inhibitory currents were higher in neurons in the VTA of females (27.23 ± 4.96 pA) relative to males (8.55 ± 4.29 pA) (t(16) = 2.85, p = 0.012, Fig. 1c,d). Collectively, these results show a greater δ-mediated, tonic inhibition in the VTA of female mice relative to males.  Table 1. List of primers used for qRT-PCR. The primers used to detect baseline Gabrd and Gabrg2 transcript levels are designated as WT. The primers used to detect the Cre-mediated excision are designated "Cre excision". Gabrd and Gabrg2 transcript levels were normalized to β-actin.
www.nature.com/scientificreports www.nature.com/scientificreports/ No sex differences in VTA Gabrg2 mRNA expression. We next sought to determine whether these effects were specific to δ-mediated tonic inhibition or if they applied to GABAergic inhibition more broadly. First, we determined that there are no sex differences in ratio of Gabrg2 transcript levels compared to β-actin in the VTA between males (0.98 ± 0.07 fold) and females (1.21 ± 0.16 fold) (t(6) = 1.384, p = 0.2156, Fig. 1e) and there was no significant difference in frequency of sIPSCs in VTA neurons between females (5.77 ± 1.05 Hz) and males (frequency: 7.00 ± 0.71 Hz) (t(18) = 1, p = 0.3304, Fig. 1f,g). There is also no difference in the peak amplitude of sIPSCs in VTA neurons between females (59.54 ± 6.49 pA) and males (69.49 ± 6.72 pA) (t(18) = 1.05, p = 0.3075, Fig. 1f,h). Taken together, these results indicate that there are similar levels of Gabrg2 transcript and ɣ2-mediated phasic inhibition in the VTA of males and females.
Knockdown of Gabrd in the VTA. Given the baseline differences in δ-mediated tonic inhibition in the VTA between males and females ( Fig. 1), it is likely that excision of the Gabrd gene would have a greater impact on females. As such, we next sought to determine whether excision of the Gabrd gene would result in differential changes in VTA tonic inhibition between males and females.
Collectively, these results indicate that Cre-mediated excision of the Gabrd gene can effectively decrease both Gabrd mRNA transcript levels and tonic inhibition in the VTA. While this effect was only notable in females, its apparent ineffectiveness in males is likely due to the floor effect of low baseline Gabrd expression.
Knockdown of Gabrd in the VTA results in sex-specific differences in binge-like drinking. Given the putative role for δ subunit-containing GABA A receptors in mediating the effects of low dose alcohol, we next sought to determine whether δ subunit-containing GABA A receptors had a sex-dependent effect on binge www.nature.com/scientificreports www.nature.com/scientificreports/ drinking. To address this question, we implemented the Drinking in the Dark paradigm (DID 28 ). During the first three days, mice were allowed 2 h of access to 20% (v/v) ethanol that was timed to occur three hours after the start of the dark cycle. There was no interaction between Gabrd mRNA knockdown and alcohol drinking for the first three days for either females or males (females: F(2,28) = 1.93, p = 0.16; males: F(2,46) = 0.07, p = 0.93) (Fig. 3a,d). The fourth and final day was the 'binge' day, where animals had access to alcohol for 4 h. Females in the intra-VTA AAV-Cre-GFP group consumed significantly less alcohol during the binge day (0.97 ± 0.55 g/kg) compared to female intra-VTA AAV-GFP controls (3.90 ± 1.14 g/kg) (t(14) = 2.31, p = 0.037, Fig. 3b). Males, however, showed no difference in binge-like drinking between AAV-Cre-GFP (3.67 ± 0.87 g/kg) and AAV-GFP groups (3.88 ± 0.84 g/kg) (t(23) = 0.17, p = 0.868, Fig. 3e).
To validate g/kg intake values against measurable BECs, a subset of the male and female mice subjected to DID had submandibular blood taken immediately after the 4 h 'binge' session. There was a significant correlation across subjects and AAV group between BEC and g/kg intake (r = 0.976, n = 12, p < 0.0001, data not shown), verifying that g/kg values were an accurate reflection of g/kg alcohol consumed.
To determine if these effects were specific for alcohol, sucrose preference and quinine aversion was assessed. There were no significant differences between female AAV-Cre-GFP or AAV-GFP groups (Fig. 3c) or male AAV-Cre-GFP or AAV-GFP groups (Fig. 3f)   , we next examined whether there was a neuroanatomical relationship within the VTA between AAV transduction efficiency and binge-like drinking in females and males. An initial placement analysis revealed that both AAV-Cre-GFP and AAV-GFP transduction were predominantly targeted to the posterior VTA (pVTA) (Fig. 4a), with little viral spread to the anterior VTA (aVTA) (Fig. 4b).
Reduced VTA Gabrg2 expression does not alter binge-like drinking. Floxed Gabrg2 male and female mice were bilaterally injected into the VTA with either AAV-Cre-GFP or AAV-GFP and subjected to one cycle of binge-like drinking (DID) as previously described. There was no significant difference in binge-like drinking in AAV-Cre-GFP or AAV-GFP females (3.63 ± 0.77 g/kg, 4.42 ± 0.4 g/kg, respectively; p = 0.41, Fig. 5i). Similarly, there is no difference in binge-like drinking in AAV-Cre-GFP and AAV-GFP males (3.09 ± 0.42 g/kg, 3.85 ± 0.64 g/kg, respectively; p = 0.34, Fig. 5j). There was a significant correlation between BEC measures and g/ kg intake (r = 0.5786, n = 26, p = 0.002; data not shown), validating the use of g/kg as a drinking measure.

Discussion
The results presented here show a striking and novel sex-specific difference in murine binge-like alcohol drinking that hinges on Gabrd expression in the pVTA. Concurrently, we also show higher baseline Gabrd transcript levels as well as δ-mediated tonic inhibition in the VTA of female mice as compared to males. In contrast, sex-specific differences were not observed in transcript levels of Gabrg2 or in phasic GABAergic inhibition. Subsequent Cre-mediated Gabrd excision primarily from the pVTA significantly reduced female binge-like drinking, with www.nature.com/scientificreports www.nature.com/scientificreports/ Cre-mediated Gabrg2 excision having no effect. Taken together, these results indicate that δ-mediated tonic inhibition in the VTA plays an integral role in mediating binge drinking behavior in females.
Recent work has demonstrated estrous cycle-dependent changes in Gabrd transcript levels in the VTA 24 . Specifically, Gabrd levels in non-estrus females were found to be approximately 80% greater than those in estrus. More broadly, this pattern underscores that ovarian hormones and, relatedly, biological sex plays a relevant role in regulating Gabrd expression. The female mice used in these experiments were maintained in same-sex housed conditions under separate ventilation from males, thus avoiding estrus cycle-inducing male cues 26,27 , focusing the study solely on sex-dependent effects on binge drinking. The highest relative Gabrd female transcript level was approximately 50% greater than our lowest relative Gabrd transcript level in acyclic females, which is markedly less variable than that seen between estrous and non-estrous states in freely cycling females 24 , indicating a likely less important role for estrous cycle in these acyclic females. Perhaps most importantly, Cre-mediated excision reduced VTA Gabrd levels in females to nearly negligible levels. Thus, any cycling-dependent effects on subunit expression for Gabrd mice would be irrelevant once Cre-mediated excision had occurred.
While there have been conflicting reports 29 , it has been suggested that low alcohol concentrations influence the activity of δ-containing GABA A receptors in recombinant receptor systems as well as in slices [9][10][11]30,31 . However, it is also likely that the well-documented effects of ethanol on presynaptic GABA release [15][16][17] and/or increases  48 . Gabrg2 transcript levels were significantly decreased in female (b) and male (c) AAV-Cre-GFP VTA relative to AAV-GFP (n = 4-7 mice per group). (d) Representative trace demonstrating decreased phasic inhibition after Cre-mediated excision of Gabrg2. Cre-mediated excision resulted in decreases in both sIPSC frequency and peak amplitude in GFP+ cells from the VTA of female (e,g) and male (f,h) AAV-Cre-GFP relative to AAV-GFP (***p < 0.0001, ^p = 0.053, ***p < 0.0001 and **p = 0.01; n = 9-12 cells, 3 mice per experimental group). VTA Gabrg2 excision had no effect on binge-like drinking in females (i) or males (j).
www.nature.com/scientificreports www.nature.com/scientificreports/ in neurosteroid levels [18][19][20]32 may indirectly mediate the effects of ethanol on these receptors. Regardless, global deletion of the Gabrd gene was found to be associated with reduced 24 hr alcohol consumption and preference in males and females 12 . Yet, whether similar sex-independent reductions in intake would occur when modeling binge-like drinking had not been determined. The present work allowed for site-specific Gabrd deletion within the VTA, demonstrating a sex-dependent role of GABA A R δ subunit-containing receptors in binge-like alcohol consumption. It is also possible that Gabrd excision in other areas of mesolimbic circuitry (e.g., dorsomedial shell of the nucleus accumbens, see 33 ) would be more effective in males and/or could occlude the bias conferred by the δ in the VTA of females.
When compared with δ-containing GABA A receptors, γ2-containing GABA A receptors are extensively expressed throughout the brain 34 . Clinically, there is some evidence that alterations to Gabrg2 is related to the propensity to develop alcoholism [35][36][37] and past work has shown γ2-subunit containing receptors respond to high concentrations of alcohol (>30-40 mM 31 ). This indicates that γ2-containing GABA A receptors are less sensitive to alcohol-especially relative to those with a δ subunit. When considering our Gabrg2 mice, it is notable that the level of reduction in Gabrg2 levels was not as significant as those seen in Cre-mediated excision of Gabrd. One concern is that the remaining γ2 subunit-containing GABA A receptors might counterbalance any effect of the knockdown. Our Gabrg2 mice drank an average of 3.5-4.5 g/kg for females (Cre and GFP) and 3.0-3.8 g/kg for males (Cre and GFP), corresponding to approximate BECs of 40 mg/dL immediately after the 'binge' . These values are lower than have been reported for DID 7,28,38 , and approximate to alcohol concentrations of 8-9 mM. This is potentially lower than the threshold for γ2 subunit activation, which may impact our results and introduce a bias towards δ-mediated effects.
It is worth noting that the DID protocol requires single housing of the animals, introducing a social isolation stress using this paradigm, which may impact alcohol consumption 39 . In particular, social isolation during adolescence has been shown to impact alcohol consumption in adulthood 40,41 in a sex-specific manner 42 . In the current study, the animals are group housed until being subjected to the DID protocol as adults, suggesting that the differences in alcohol consumption is not related to previous social isolation stress. In addition, all the animals in the study are identically treated and subjected to the same extent of social isolation, arguing that differences in intake following genetic manipulations or the sex-dependent differences are not related to the social isolation component of the experimental paradigm.
Our data revealed two major conclusions: (1) higher baseline Gabrd levels in females and (2) reductions in binge-like drinking after Gabrd excision from the VTA in females. But what does this mean for overall differences in VTA circuit functioning between males and females? In a baseline state, female mice have a larger amount of δ-mediated tonic inhibition within the VTA, whether localized exclusively to the pVTA or more broadly. As previous work supports greater activation of GABAergic vs. DAergic cells in the VTA in response to the δ-specific compound THIP 43 , we posit that the sex difference in δ-GABA A Rs would maintain higher inhibitory tone on these GABAergic cells, allowing for the greater activation of dopamine neurons in the VTA in females 25 (Fig. 6)   Figure 6. Higher δ expression in the VTA of females has implications for overall mesolimbic excitability and effects on binge drinking. (top row, left to right) Female VTA has higher expression of δ-subunit containing GABA A Rs in the VTA, resulting in lower inhibition on DAergic neurons and higher DA release. Low concentrations of alcohol interact with δ-subunit containing GABA A Rs on this population, further increasing DA release. Cre-mediated excision markedly reduces Gabrd expression and δ-mediated tonic inhibition, reducing overall DA release in females. (bottom row, left to right) Male VTA has decreased baseline δ-subunit containing GABAA receptor in the VTA, resulting in greater inhibition on DAergic projection neurons. While low concentrations of alcohol disinhibit DAergic neurons and increase DA release, the effect is not as significant as in females. Due to the low baseline Gabrd levels, Cre-mediated excision has limited effect on both Gabrd levels, δ-mediated tonic inhibition, and resulting DA release. In all cases, γ2-containing GABAA receptors are present at equal levels between males and females and are not affected by low concentrations of alcohol.