Ethanol Withdrawal-Induced Dysregulation of Neurosteroid Levels in Plasma, Cortex, and Hippocampus in Genetic Animal Models of High and Low Withdrawal

Abstract

Rationale

Endogenous γ-aminobutyric acid_A receptor (GABA_AR)-active neurosteroids (e.g., allopregnanolone) regulate central nervous system excitability and many physiological functions, so fluctuations are implicated in several neuropsychiatric disorders. Pertinently, evidence supports an inverse relationship between endogenous GABA_AR-active neurosteroid levels and behavioral changes in excitability during ethanol withdrawal (WD).

Objectives

The present studies determined mouse genotype differences in ten neurosteroid levels in plasma, cortex, and hippocampus over the time course of ethanol WD in the WD Seizure–Prone (WSP) and –Resistant (WSR) selected lines and in the DBA/2J (DBA) inbred strain.

Methods

Gas chromatography-mass spectrometry was utilized to simultaneously quantify neurosteroid levels from control-treated male WSP-1, WSR-1, and DBA mice and during 8 h and 48 h of WD.

Results

Combined with our prior work, there was a consistent decrease in plasma allopregnanolone levels at 8 h WD in all three genotypes, an effect that persisted at 48 h WD only in DBA mice. WSR-1 and WSP-1 mice exhibited unexpected divergent changes in cortical neurosteroids at 8 h WD, with the majority of neurosteroids (including allopregnanolone) being significantly decreased in WSR-1 mice, but unaffected or significantly increased in WSP-1 mice. In DBA mice, hippocampal allopregnanolone and tetrahydrodeoxycorticosterone were significantly decreased at 8 h WD. The pattern of significant correlations between allopregnanolone and other GABA_AR-active neurosteroid levels differed between controls and withdrawing mice.

Conclusions

Ethanol WD dysregulated neurosteroid synthesis. Results in WSP-1 mice suggest that diminished GABA_AR function is more important for their high WD phenotype than fluctuations in neurosteroid levels.

Introduction

Endogenous neurosteroids that rapidly enhance γ-aminobutyric acid_A receptor (GABA_AR)-mediated inhibition (e.g., Belelli and Lambert, 2005; Carver and Reddy, 2013; Paul and Purdy, 1992) are formed by the 5α-/5β- and then 3α-reduction of the parent steroids progesterone, deoxycorticosterone (DOC), testosterone, and dehydroepiandrosterone (DHEA; e.g., Finn et al., 2004; Porcu et al., 2009, 2016; Snelling et al., 2014). The progesterone metabolites allopregnanolone (ALLO; 3α,5α-THP or tetrahydroprogesterone) and pregnanolone (3α,5β-THP) and the DOC metabolite tetrahydrodeoxycorticosterone (3α,5α-THDOC) are the three most potent neurosteroids characterized to date, as they enhance GABA_AR-mediated inhibition with nanomolar (nM) potencies, directly activate GABA_ARs with micromolar potencies, and exert effects on other ligand gated ion channels with micromolar potencies (Belelli and Lambert, 2005; Belelli et al., 1990; Carver and Reddy, 2013; Paul and Purdy, 1992; Purdy et al., 1990; Rupprecht and Holsboer, 1999; Veleiro and Burton, 2009). The testosterone metabolite 3α,5α-androstanediol and the DHEA metabolite 3α,5α-androsterone potentiate GABA_ARs but with lower potency than ALLO and 3α,5α-THDOC (Carver and Reddy, 2013; Porcu et al., 2016). Importantly, all steroidogenic enzymes are localized in the nervous system and in peripheral steroidogenic tissues (Do Rego et al., 2009; Mellon and Vaudry, 2001), indicating that neurosteroid levels in the brain reflect a combination of compounds produced there de novo and from metabolism of circulating precursors. Because endogenous levels of ALLO and 3α,5α-THDOC fluctuate in the 10 – 100 nM range that enhance GABA_AR inhibition (Barbaccia et al., 2001; Belelli and Lambert, 2005; Carver and Reddy, 2013; Finn et al., 2004; Paul and Purdy, 1992), these neurosteroids can regulate central nervous system (CNS) excitability. Accumulating evidence suggests that dysregulation in neurosteroid synthesis contributes to symptoms in several neuropsychiatric and neurodegenerative diseases and that targeting neurosteroid biosynthesis may be an effective treatment strategy (Porcu et al., 2016).

Alcohol (ethanol) withdrawal (WD) is one important dimension of genetic risk for dependence (Kendler et al., 2012), and WD-induced seizure activity increases with successive WD in both humans and animals (Duka et al., 2004). Repeated WDs are associated with deficits in cognitive function and emotional processing (e.g., Stephens and Duka, 2008; Sullivan et al., 2000) and are suggested to contribute to continued drinking (Ripley and Stephens, 2011), which underscores the importance of examining mechanisms underlying high withdrawal (Heilig et al., 2010). We and others found that WD is associated with a decrease in GABA_AR inhibition mediated by a variety of factors that includes functional changes in GABA_AR properties, a reduction in the steroidogenic effect of acute ethanol administration, and a decrease in endogenous ALLO levels (see Finn et al., 2004; Kumar et al., 2009). In rodents, monkeys, and humans, WD decreases ALLO levels in plasma and several brain regions (Beattie et al., 2017; Cagetti et al., 2004; Hill et al., 2005; Maldonado-Devincci et al., 2014; Romeo et al., 1996; Snelling et al., 2014; Tanchuck et al., 2009). In small cohorts of male and female alcoholics, the decrease in ALLO and 3α,5α-THDOC levels corresponded to an increase in the subjective ratings of anxiety and depression during days 4 – 5 of WD, versus controls (Hill et al., 2005; Romeo et al., 1996). In male cynomolgus monkeys that had been consuming ethanol for over 12 months, there was a significant decrease in plasma ALLO levels and in ALLO immunoreactivity in the lateral and basolateral amygdala (only amygdala examined) at 48 h following the last drinking session versus controls (Beattie et al., 2017). A significant negative correlation between ALLO immunoreactivity in the lateral and basolateral amygdala and average daily ethanol consumption suggested that WD from high ethanol consumption corresponded to lower ALLO immunoreactivity. WD (48 h) from chronic intermittent ethanol (CIE) intragastric exposure significantly decreased hippocampal ALLO levels, which was associated with a significant increase in anxiety and impairment in hippocampal-dependent memory function in male rats (Cagetti et al., 2004). Collectively, chronic ethanol WD produces a consistent reduction in endogenous ALLO levels that may be associated with increased cellular excitability, high prior ethanol consumption, and increased aversive behavioral effects (e.g., anxiety, depression, convulsive activity, cognitive impairment).

Our laboratory has been using two different genetic models of ethanol WD severity [WD Seizure-Prone (WSP, severe) versus WD Seizure-Resistant (WSR, negligible) selected lines; DBA/2J (DBA, high) versus C57BL/6J (C57, mild) inbred strains] to test the hypothesis that genetic differences in ethanol WD severity were due in part to alterations in the GABA_AR modulatory effects of neurosteroids such as ALLO. Both WSP and DBA mice exhibit high WD severity, measured by increased handling-induced convulsions (HICs) and anxiety following 72 h of continuous ethanol vapor or CIE vapor (Crabbe et al., 1985; Crabbe, 1998; Finn and Crabbe, 1999; Finn et al., 2004; Gorin et al., 2005; Kosobud and Crabbe, 1986; McCool and Chappell, 2015; Metten and Crabbe, 2005; Metten et al., 2010). Importantly, both WSP and DBA mice exhibited tolerance to ALLO’s anticonvulsant effect during peak (8 h) WD (Beckley et al., 2008; Finn et al., 2000, 2004, 2006), and this tolerance in WSP mice also corresponded to a reduction in functional sensitivity of GABA_ARs to ALLO (Finn et al., 2006) and reduced anticonvulsant effect of intra-CA1 ALLO (Gililland-Kaufman et al., 2008). Thus, a reduction in GABA_AR sensitivity to neurosteroids may contribute to a severe WD phenotype in genetic animal models of high WD.

While the majority of studies have examined relationships between ALLO and measures of ethanol WD, neurosteroid synthesis is a dynamic process, so disruptions in one or more biosynthetic steps could influence levels of multiple steroids with positive or negative modulatory effects at GABA_ARs. Furthermore, since GABA_ARs and other targets have varying degrees of sensitivity to multiple neurosteroids, the state of GABA_ARs will depend on the relative levels of all neurosteroids working in concert to coordinate the physiological control of CNS excitability. Thus, determining the net effect of ethanol WD on GABA_ARs requires a detailed analysis of multiple neurosteroids simultaneously.

We recently modified a gas chromatography-mass spectrometry (GC-MS) method (Porcu et al., 2009) to simultaneously quantify the GABA_AR-active 3α,5α-/3α,5β-reduced metabolites of progesterone, DOC, testosterone and DHEA as well as the precursors pregnenolone and DHEA in plasma from control WSP and WSR mice from the first genetic replicate (WSP-1 and WSR-1) and at 8 h WD from 72 h continuous ethanol vapor exposure (Snelling et al., 2014); there was no line difference in the WD-induced suppression in plasma DHEA, ALLO, 3α,5α/3α,5β-androstanediol, and 3α,5α-androsterone levels or increase in plasma corticosterone (CORT) levels. While the pattern of changes in steroid levels imply that ethanol WD is exerting independent effects at several steps of the steroidogenic pathway in plasma (discussed in Snelling et al., 2014), it is not known if similar results would be observed in discrete brain regions. So, cortex and hippocampus were examined in the present studies, based on evidence that basal neurosteroid levels in plasma do not fully reflect levels in cortex and hippocampus (Caruso et al., 2013) and that divergent changes in ALLO immunoreactivity were observed during WD in mPFC and CA3 (Maldonado-Devincci et al., 2014). Cortical and hippocampal regions also exhibit increased spike and sharp wave activity in the electroencephalogram (e.g., Veatch and Becker, 2002; Veatch and Gonzalez, 1996) and increased immediate early gene expression (Vilpoux et al., 2009) that was greater in DBA versus C57 mice (Chen et al., 2009) during WD, highlighting the importance of these brain regions in WD circuitry. Thus, one purpose of the present experiments was to use GC-MS to simultaneously quantify a panel of 10 neurosteroids in plasma, cortex and hippocampus in control-treated male WSP-1 and WSR-1 mice and over the time course of WD that corresponded to enhanced convulsive activity in WSP mice (i.e., 8 and 48 h; Finn et al., 2004; Gililland-Kaufman et al., 2008), and that also corresponded to the acute phase of WD in mice (i.e., 0 – 48 h in rodents; Heilig et al., 2010). We predicted that the WD-induced changes in brain neurosteroid levels would differ from the pattern detected in plasma and that there would be a line difference in hippocampal and cortical neurosteroid levels during WD. A second purpose was to conduct similar studies in male DBA mice, and to compare the WD-induced changes in neurosteroid levels across two genetic animal models of high ethanol WD. We predicted that there would be similarities in the pattern of results in DBA and WSP-1 mice, consistent with a genetic influence on the proposed inverse relationship between GABA_AR-active neurosteroid levels and ethanol WD severity.

Materials and methods

Animals

Drug naïve male WSP-1 and WSR-1 mice were bred in the Veterinary Medical Unit at the VA Portland Health Care System (VAPORHCS; Portland, OR), were housed 2–4 per cage with free access to food and water upon weaning, and were acclimated to a 12:12 h light/dark cycle (lights on at 0600). At the time of testing, mice were from selected generation 26 (filial generation 132 for 8 h study and filial generations 130–131 for 48 h study) and were 51–77 days old (8 h study) or 47–101 days old (48 h study). Drug naïve male DBA mice were purchased from Jackson Laboratories West (Sacramento, CA) at 8 weeks of age, housed 4 per cage with free access to food and water in the Department of Comparative Medicine at Oregon Health & Science University (OHSU; Portland, OR), and were acclimated to a 12:12 h light/dark cycle (lights on at 0700) for 2 weeks before experimentation. All procedures complied with the National Institute of Health Guidelines for the Care and Use of Laboratory Animals, 8^th ed (National Research Council of the National Academies, 2011) and were approved by the local Institutional Animal Care and Use Committees at VAPORHCS (WSP and WSR studies) and at OHSU (DBA studies).

Chronic ethanol exposure

Mice were exposed to ethanol vapor or air for 72 h by the Dependence Core of the Portland Alcohol Research Center at the VAPORHCS (WSP and WSR studies) or by the Finn laboratory at OHSU (DBA studies), using identical equipment, facilities, and standardized method for inducing ethanol dependence with continuous ethanol vapor exposure (e.g., Finn and Crabbe, 1999; Tanchuck et al., 2009; Terdal and Crabbe, 1994). All blood ethanol concentrations (BECs) were analyzed by headspace gas chromatography (Finn et al., 2007) by the Finn laboratory, and ethanol vapor concentrations were adjusted to achieve BECs of 1.5 – 2.0 mg/mL. Upon removal from the inhalation chambers, separate groups of mice were euthanized at 8 h and at 48 h, but mice were not scored for HIC.

Blood and brain collection

Trunk blood was collected, and plasma was separated and stored at −20°C. Plasma neurosteroid analysis by GC-MS utilized 300 μL samples. The neurosteroids examined included pregnenolone, DHEA, and the 3α,5α/3α,5β-GABA_AR-active metabolites of progesterone, testosterone, DOC and DHEA. Plasma CORT levels were measured in 5 μL samples with a commercially available double antibody radioimmunoassay kit (MP Biomedicals, Solon, OH), as recently described (Ford et al., 2013). Dissected cortical (including prelimbic, cingulate, and overlying motor cortex) and entire hippocampal tissues were collected, rapidly frozen on dry ice, and stored at -20°C. Analysis of samples was singly determined.

Chemicals

All extraction and derivatization solvents were chromatography grade (EMD Chemicals, Billerica, MA). Monodeuterated ethanol was obtained from Acros Organics (New Jersey), the derivatizing reagent heptafluorobutyric acid anhydride (HFAA) was obtained from Fluka (Sigma-Aldrich, St. Louis, MO), and the neurosteroids pregnenolone, pregnanolone, 3α,5β-THDOC, and 3α,5β-androstanediol were obtained from Steraloids (Newport, RI). All other native neurosteroid standards were synthesized by the late Dr. Robert Purdy. Similar to Snelling et al. (2014), the internal standard (ISTD) was 17β-estradiol-2,4,16,16,17-d5 (CDN Isotopes, Pointe-Claire, Quebec, Canada). Ultra-pure water (double distilled; ddH₂O) from a MilliQ system (Millipore Corporation, Bedford, MA) was used. Activated powdered charcoal (Darco-G-60) and Dextran (T-70) were obtained from Sigma-Aldrich.

Extraction of Plasma Steroids

Solid phase extraction (SPE; Snelling et al., 2014) was used, with slight modification. Plasma (300 μL) was spiked with 10 ng estradiol-d5 (ISTD), loaded onto a preconditioned Strata-X-33u Polymeric Reversed Phase SPE column (Phenomenex, Torrance, CA), washed with 1 mL ddH₂O and then with 1 mL 30% v/v methanol:ddH₂O, and dried on a Waters vacuum manifold (Waters, Milford, MA). The dried cartridge was washed with 1 mL n-hexane, and dried for 15 min at 12 in Hg (inches of mercury, equal to 0.49 psi) of vacuum. Steroids were eluted from the SPE columns with 2 aliquots of 750 μL of acetonitrile, and the pooled eluate was dried on an Eppendorf Vacufuge for 90 min. Samples were derivatized or stored in a vacuum desiccator for up to 24 h.

Extraction of Brain Steroids

The extraction used wand homogenization followed by SPE, similar to the plasma steroids. Tissue (5 – 15 mg) was added to 1 mL ice cold ddH₂O, spiked with 10 ng of estradiol-d5 (ISTD), and vortexed for 30 s. The spiked sample was wand homogenized with a Fisher Scientific Sonic Dismembrator Model 500 (Thermo Fischer, Waltham, MA) with a 45 s program of 500 ms on, 1000 ms off at 35% power. The sample was loaded on a preconditioned Strata-X-33u SPE column and treated identical to the extraction of plasma samples. After the pooled eluate was dried on an Eppendorf Vacufuge for 90 min, the sample was then derivatized or stored in a vacuum desiccator for up to 24 h.

Derivatization

Extracted samples were solvated in acetonitrile (450 μL), HFAA (50 μL) was added, and samples were treated as described in Snelling et al. (2014). The dried derivatized samples were resolvated in acetonitrile (10 μL) and transferred to a deactivated vial insert for immediate analysis.

Gas chromatography-mass spectrometry (GC-MS)

Samples and calibration curves (2 μL samples) were analyzed on an Agilent 7890A gas chromatograph coupled to an Agilent 5975C mass spectrometer detector (Agilent Technologies, Santa Clara, CA; Snelling et al., 2014). Briefly, eluates were analyzed by the mass spectrometer in negative chemical ionization using methane (research grade 99.999%, Air Liquide, Houston, TX) as the reagent gas at a flow rate of 40%. The mass spectrometer chemical ionization source was set to 170°C and quadrupole mass filter temperature was set at 150°C. Selected ion monitoring was used, and data were collected for a minimum of two ions per analyte over five retention period groups (details in Snelling et al., 2014).

Calibration curves

Calibration curves were formed by spiking 300 μL of ddH₂O water with 0, 50, 100, 500, 1000, or 3000 pg of all ten analytes of interest and 10 ng of ISTD. The spiked reference standards were then extracted, derivatized, and analyzed by GC-MS. The ratio of the responses of the analyte divided by the response of the ISTD versus the concentration of the analyte was plotted. For each analyte, the response was an integration of the abundance counts collected by the MS detector horn for the target ion from baseline to baseline of the pertinent peak. No analytes co-eluted with other analytes or with the ISTD. Calibration curves were analyzed with linear regression, and unknown plasma or brain samples were interpolated against these regression parameters using their individual response ratios.

Method validation

We conducted additional validation experiments to verify accuracy, reliability, and reproducibility of the current GC-MS procedure for analysis of brain tissue. One mL of ddH₂O was homogenized with 5 – 15 mg of brain tissue, then dextran coated charcoal (250 mg charcoal:25 mg dextran) was added to the homogenate and vortexed for 60 s to strip the brain homogenate of endogenous steroids. The homogenate was centrifuged at 3500 rpm (~1900 × g) for 30 min. The resultant supernatant was transferred to an Eppendorf vial and was spiked with both 10 ng of ISTD and one of four validation concentrations (0, 70, 300, and 700 pg) of all ten analytes of interest. All validation samples were analyzed as described above. To allow for measurement of coefficient of variation for intra-assay and inter-assay data, we independently assayed the four validation concentrations across four sets of steroid stripped brain homogenate samples. We also calculated matrix effects of the stripped brain homogenates by comparing a ratio of the response from the matrix (i.e., homogenate) samples versus the response from our calibration curve from the solvent only spiked samples to determine potential interference of proteins and lipids in the analyte quantification.

Data analysis

Target ion peaks were integrated in ChemStation and interpolated using Prism5 (GraphPad Software Inc., San Diego, CA). Mean picograms (pg) per milliliter (mL; plasma) or milligram (mg; brain) of neurosteroid per extracted sample were calculated from the interpolated value and statistically analyzed with analysis of variance (ANOVA) for line and treatment (WSP and WSR studies) or treatment (DBA studies) effects. A similar strategy was used for the analysis of CORT levels. Significant interactions were followed up with post-hoc tests. Because we were predicting line differences, planned comparisons were conducted with or without the presence of a significant interaction. Effect size estimates (Cohen, 1988) were calculated, with d=0.2 considered a small effect, d=0.5 a medium effect, and d=0.8 a large effect. Pearson correlations analyzed relationships between ALLO and the other 9 neurosteroids in cortex and hippocampus from air- versus ethanol-exposed mice at a time corresponding to peak WD (i.e., 8 h) to evaluate whether treatment differences in relationships would provide evidence for dysregulation in neurosteroid synthesis during WD. Significance was set at a p value of ≤ 0.05.

Results

GC-MS brain assay validation

We initially conducted a variety of control studies to validate the accuracy of our GC-MS approach to quantify neurosteroids in dissected brain tissues (Results in Supplemental Material). Briefly, the limits of quantification of our analytes were comparable to our assay in plasma per 2 μL aliquot, coefficients of variation were low, and there was an excellent linear response for all ten analytes of interest. The precision and accuracy of the control studies confirm the high level of confidence in the optimized assay.

Time course of changes in neurosteroids during ethanol WD in WSP-1 and WSR-1 mice

Hippocampal and cortical neurosteroid levels at 8 h WD

Plasma neurosteroid levels were analyzed and reported in a separate study (Snelling et al., 2014). In the present study, BECs upon removal from the vapor chambers at 72 h were 1.5 ± 0.12 mg/mL (WSP-1, n=13) and 1.6 ± 0.08 mg/mL (WSR-1, n=11); these values were not significantly different from those in the earlier study [BEC = 1.6 ± 0.10 mg/mL (WSP-1) & 1.6 ± 0.16 mg/mL (WSR-1)]. Plasma CORT levels in the air-exposed mice were 16.87 ± 8.54 μg/dL (WSP-1; n=13) and 17.46 ± 2.72 μg/dL (WSR-1; n=15). Ethanol WD significantly increased plasma CORT levels in WSP-1 mice (135.36 ± 25.96 μg/dL, n=11) and WSR-1 mice (80.93 ± 10.67 μg/dL, n=10) [main effect of treatment: F(1,45)=51.30, p<0.001 and post-hoc tests: p<0.001]. The significant interaction between line and treatment [F(1,45)=4.69, p<0.05] was due to the significantly greater WD-induced increase in plasma CORT levels in WSP-1 (703%) versus WSR-1 (363%) mice.

Cortical neurosteroid levels are depicted in Figure 1, and the two-way ANOVA results are summarized in Supplemental Table 2. In general, the results in cortex differed from our results in plasma (Snelling et al, 2014), where similar WD-induced reductions in plasma levels of 5 neurosteroids, including ALLO, were observed in WSP-1 and WSR-1 mice. In the present study, ethanol WD significantly decreased 8 cortical neurosteroid levels by 31 – 48% only in WSR-1 mice (Figure 1). In contrast, ethanol WD significantly increased 3α,5α-androstanediol levels (Figure 1e) in both WSP-1 and WSR-1 mice (19 & 34%, respectively), and increased levels of 3α,5α/3α,5β-androsterone by 122% and 68%, respectively (Figure 1g & 1h), only in WSP-1 mice.

Figure 1. Line differences in the effect of peak chronic ethanol WD (i.e., 8 h) on cortical neurosteroid levels in male WSP-1 and WSR-1 mice.

Mice were exposed to 72 h continuous ethanol vapor or air and euthanized at a time corresponding to peak WD in WSP-1 mice. Ethanol WD significantly decreased cortical levels of 8 of the 10 neurosteroids examined only in WSR-1 mice, with a significant increase in 3 cortical neurosteroid levels in WSP-1 mice. Values are shown as pg/mg and represent the mean ± SEM for the number of mice in parentheses. Also shown are the WD-induced percent changes (vs respective air control) that were > 30%, including those that did not reach statistical significance. Note the different range of the y-axes across panels and that y-axes were kept constant for each pair of 3α,5α/3α,5β-neurosteroid metabolites. Slight differences in group size reflect cases where a statistical outlier (>2 standard deviation from the mean) was removed. *p<0.05, **p<0.01, ***p<0.001 vs respective air control (planned comparison post-hoc tests; effect size range = 0.89–1.52); ^#p<0.05 (main effect of treatment; effect size = 0.64).

Correlations revealed line and treatment differences in the relationships between cortical ALLO levels and levels of the other nine neurosteroids (Table 1). In WSP-1 air-exposed mice (Table 1, top), ALLO levels were significantly positively correlated with levels of 5 neurosteroids, but significantly negatively correlated with 3α,5α-androstanediol levels. In contrast, only pregnanolone and pregnenolone levels remained significantly positively correlated with ALLO levels during ethanol WD. A different pattern of results was observed in WSR-1 mice (Table 1, middle), with more significant correlations during WD than in controls (6 versus 4, respectively). There were some similarities with the results in WSP-1 mice; in both the air- and ethanol-exposed WSR-1 mice, ALLO levels were significantly positively correlated with pregnanolone and pregnenolone levels (similar to WSP-1 mice) and significantly negatively correlated with 3α,5α-androstanediol levels (similar to air-exposed WSP-1 mice).

Table 1.

Cortical neurosteroid correlations between levels of allopregnanolone (3α,5α-THP) and 9 neurosteroids in WSP-1, WSR-1, and DBA mice at 8 h

	3α,5α-diol	3α,5β-diol	3α,5α-one	3α,5β-one	DHEA	3α,5β-THP	PREG	3α,5α-THDOC	3α,5β-THDOC
WSP-1; air n=12–13	r = −0.88 p<0.001 n=13	ns	r = 0.56 p<0.05 n=13	r = 0.74 p<0.01 n=13	ns	r = 0.72 p<0.01 n=13	r = 0.79 p<0.01 n=12	r = 0.79 p=0.001 n=13	ns
WSP-1; WD n=13	ns	ns	ns	ns	ns	r = 0.55 p=0.05 n=13	r = 0.74 p<0.01 n=13	ns	ns
WSR-1; air n=14–15	r = −0.63 p=0.01 n=15	r = 0.52 p<0.05 n=15	r = 0.52 p<0.06 n=14	r = 0.51 p=0.06 n=14	ns	r = 0.57 p<0.05 n=15	r = 0.74 p<0.01 n=15	ns	ns
WSR-1; WD n=10–11	r = −0.60 p=0.05 n=11	r = 0.88 p=0.001 n=10	r = 0.77 p<0.01 n=11	ns	ns	r = 0.84 p<0.01 n=10	r = 0.77 p<0.01 n=11	ns	r = 0.67 p<0.05 n=11
DBA; air n=16	ns	ns	ns	ns	ns	ns	ns	ns	ns
DBA; WD n=18	ns	ns	ns	ns	ns	ns	ns	ns	ns

Pearson Correlations were conducted between allopregnanolone and the other 9 neurosteroids on the cortical neurosteroid data in mice at the 8 h time point for air-exposed mice (air) and for ethanol-exposed mice undergoing withdrawal (WD). Depicted are the correlations that were significant (in bold font) or were statistical trends (p≤0.06). ns = not significant

3α,5α/3α,5β-diol = 3α,5α/3α,5β-androstanediol; 3α,5α/3α,5β-one = 3α,5α/3α,5β-androsterone; DHEA = dehydroepiandrosterone; 3α,5β-THP = pregnanolone; PREG = pregnenolone; 3α,5α/3α,5β-THDOC = 3α,5α/3α,5β-tetrahydrodeoxycorticosterone

Hippocampal neurosteroid levels are depicted in Figure 2, and the two-way ANOVA results are summarized in Supplemental Table 3. Although no treatment effects reached statistical significance, there were some similarities in the pattern of the results that were found in the cortex in terms of percent (%) changes in neurosteroid levels during ethanol WD. Of the 8 neurosteroids that were decreased significantly during WD in cortex in WSR-1 mice, 6 of them were decreased by 34 – 49% in hippocampus. In WSP-1 mice, the WD-induced increase in 3α,5α-androstanediol (Figures 1e & 2e) and 3α,5β-androsterone (Figures 1h & 2h) were similar in cortex and hippocampus.

Figure 2. The effect of peak chronic ethanol WD (i.e., 8 h) on hippocampal neurosteroid levels in male WSP-1 and WSR-1 mice.

Mice were treated as described in the legend to Figure 1. Depicted are the mean ± SEM pg/mg for the number of mice in parentheses. Also shown are the WD-induced percent changes (vs respective air control) that were > 30%. Note the different range of the y-axes across panels and that y-axes were kept constant for each pair of 3α,5α/3α,5β-neurosteroid metabolites. Slight differences in group size reflect cases where a statistical outlier (>2 standard deviation from the mean) was removed.

There were line and treatment differences in the relationships between the levels of ALLO and other neurosteroids in hippocampus. In WSP-1 air-exposed mice, ALLO levels were significantly positively correlated with levels of the three 3α,5α-reduced neurosteroids and with 3α,5β-THDOC levels (Table 2, top). In contrast, during WD, ALLO levels were significantly positively correlated with levels of the three 3α,5β-reduced neurosteroids. In WSR-1 mice (Table 2, middle), ALLO levels were significantly positively correlated with levels of 6 neurosteroids in the air controls, but there were only 2 significant positive correlations during ethanol WD. It is interesting that in both the air- and ethanol-exposed WSR-1 mice, ALLO levels were significantly positively correlated with 3α,5β-THDOC levels (similar to WSP-1 mice) and significantly positively correlated with 3α,5α-androsterone levels (similar to air-exposed WSP-1 mice).

Table 2.

Hippocampal neurosteroid correlations between levels of allopregnanolone (3α,5α-THP) and 9 neurosteroids in WSP-1, WSR-1, and DBA mice at 8 h

	3α,5α-diol	3α,5β-diol	3α,5α-one	3α,5β-one	DHEA	3α,5β-THP	PREG	3α,5α-THDOC	3α,5β-THDOC
WSP-1; air n=10–11	r = 0.94 p<0.001 n=11	ns	r = 0.90 p<0.001 n=11	ns	ns	ns	ns	r = 0.86 p=0.001 n=10	r = 0.84 p=0.001 n=11
WSP-1; WD n=10–12	ns	r = 0.61 p<0.05 n=11	ns	ns	ns	r = 0.74 p<0.01 n=11	ns	ns	r = 0.77 p<0.01 n=12
WSR-1; air n=13–15	ns	r = 0.85 p<0.001 n=14	r = 0.76 p=0.001 n=15	r = 0.67 p<0.01 n=14	r = 0.52 p<0.05 n=15	ns	r = 0.61 p<0.05 n=13	r = 0.52 p<0.06 n=14	r = 0.90 p<0.001 n=14
WSR-1; WD n=8–9	ns	ns	r = 0.72 p<0.05 n=9	ns	ns	ns	ns	ns	r = 0.76 p<0.05 n=8
DBA; air n=12	ns	ns	ns	ns	ns	ns	ns	ns	ns
DBA; WD n=13	ns	ns	r = −0.58 p<0.05 n=13	ns	ns	ns	r = −0.54 p<0.06 n=13	ns	ns

Pearson Correlations were conducted between allopregnanolone and the other 9 neurosteroids on the hippocampal neurosteroid data in mice at the 8 h time point for air-exposed mice (air) and for ethanol-exposed mice undergoing withdrawal (WD). Depicted are the correlations that were significant (in bold font) or were statistical trends (p<0.06). ns = not significant

3α,5α/3α,5β-diol = 3α,5α/3α,5β-androstanediol; 3α,5α/3α,5β-one = 3α,5α/3α,5β-androsterone; DHEA = dehydroepiandrosterone; 3α,5β-THP = pregnanolone; PREG = pregnenolone; 3α,5α/3α,5β-THDOC = 3α,5α/3α,5β-tetrahydrodeoxycorticosterone

Plasma, hippocampal and cortical neurosteroid levels at 48 h WD

Similar to the cohort of mice used for the 8 h WD studies, there was no line difference in BECs upon removal from the vapor chambers at 72 h between WSP-1 (1.79 ± 0.20 mg/mL, n=8) and WSR-1 (1.75 ± 0.15 mg/mL, n=12) mice in the 48 h WD cohort. We were unable to analyze CORT levels in this cohort of mice, because of insufficient plasma remaining after the analysis of neurosteroid levels.

Plasma neurosteroids were considerably lower than in our prior work (Snelling et al., 2014), and there were minimal changes at the 48 hr WD time point versus values in the air controls (not shown). Depending on the neurosteroid, the values ranged from approximately 40 – 500 pg/mL, which are comparable to basal levels reported for C57 and DBA male mice (Porcu et al., 2010). The only WD-induced changes that were observed included a significant decrease in plasma pregnenolone levels by 8% in WSR-1 mice (p<0.05) and a strong trend for an increase in plasma 3α,5α-androsterone levels of 9 & 14% in WSP-1 and WSR-1 mice, respectively [F(1,34)=3.82, p<0.06].

The majority of cortical neurosteroid levels were unaltered at 48 h WD (Figure 3; Supplemental Table 4 for ANOVA results). Only 3α,5β-androstanediol (Figure 3f) was increased significantly in WSR-1 mice, which contrasted with the significant 40% decrease at 8 h WD (Figure 1f).

Figure 3. The effect of 48 h ethanol WD on cortical neurosteroid levels in male WSP-1 and WSR-1 mice.

Mice were exposed to 72 h continuous ethanol vapor or air and euthanized at a time corresponding to termination of WD-induced convulsive activity in WSP-1 mice. Ethanol WD significantly increased 3α,5β-androstanediol levels in WSR-1 mice. Values are shown as pg/mg and represent the mean ± SEM for the number of mice in parentheses. Also shown is the WD-induced percent change (vs respective air control) that was associated with a significant difference. Note the different range of the y-axes across panels and that y-axes were kept constant for each pair of 3α,5α/3α,5β-neurosteroid metabolites. Slight differences in group size reflect cases where a statistical outlier (>2 standard deviation from the mean) was removed. *p<0.05 vs respective air control (planned comparison post-hoc test; effect size = 0.92).

WD at 48 h produced a different pattern of results in hippocampal neurosteroid levels (Figure 4; ANOVAs in Supplemental Table 5) than seen at 8 h. Ethanol WD produced a similar significant decrease in 3α,5β-androstanediol levels (Figure 4f) in the lines, while pregnanolone (Figure 4d) and 3α,5β-androsterone (Figure 4h) levels were increased significantly in both WSP-1 and WSR-1 mice. In WSR-1 mice, there were opposite WD-induced changes in pregnenolone and DHEA levels (Figure 4a & 4b), and WD also produced line differences in the change in 3α,5α-androstanediol levels (Figure 4e).

Figure 4. Ethanol WD (48 h) produces divergent changes in the pattern of hippocampal neurosteroid levels in male WSP-1 and WSR-1 mice.

Mice were treated as described in the legend to Figure 3. While ethanol WD produced a similar significant decrease in 3α,5β-androstanediol levels in the lines, pregnanolone and 3α,5β-androsterone levels were increased significantly in both WSP-1 and WSR-1 mice. In WSR-1 mice, 48 h WD produced opposite significant changes in pregnenolone and DHEA levels, and WD also produced a line difference in the significant change in 3α,5α-androstanediol levels. Depicted are the mean ± SEM pg/mg for the number of mice in parentheses. Also shown are the WD-induced percent changes (vs respective air control) that were associated with significant differences. Note the different range of the y-axes across panels and that y-axes were kept constant for each pair of 3α,5α/3α,5β-neurosteroid metabolites. Slight differences in group size reflect cases where a statistical outlier (>2 standard deviation from the mean) was removed. *p<0.05, **p<0.01, ***p<0.001 vs respective air control (planned comparison post-hoc tests; effect size = 1.23–2.28); ^#p<0.05, ^##p<0.01 (main effect of treatment; effect size = 0.92–1.52).

Time course of changes in neurosteroids during ethanol WD in DBA mice

Plasma, hippocampal and cortical neurosteroid levels at 8 h WD

BECs upon removal from the vapor chambers at 72 h were 1.49 ± 0.10 mg/mL (n=18). Plasma CORT levels were measured in a subset of the animals [11.32 ± 2.37 μg/dL for air (n=11); 72.64 ± 4.08 μg/dL for WD (n=14)], representing a significant increase of 542% at the 8 h ethanol WD time point [F(1,23)=145.95, p<0.001].

As depicted in Figure 5, plasma levels of 4 neurosteroids were decreased significantly at 8 h WD in DBA mice. DHEA [F(1,32)=13.51, p=0.001], ALLO [F(1,32)=38.0, p<0.001], and 3α,5β-androstanediol [F(1,32)=36.25, p<0.001] levels were decreased by 20–21% during WD (Figure 5b, 5c & 5f), and 3α,5α-androstanediol [F(1,32)=13.27, p=0.001] levels were decreased by 13% during WD (Figure 5e).

Figure 5. Decrease in plasma neurosteroid levels at 8 h WD in male DBA mice.

Mice were exposed to 72 h continuous ethanol vapor or air and euthanized at a time corresponding to peak WD in DBA mice. Ethanol WD significantly decreased plasma levels of 4 neurosteroids, including allopregnanolone. Values are the mean ± SEM pg/mL for the number of mice in parentheses (Group size is the same for all neurosteroids). Also shown are the WD-induced percent changes (vs air control) that were associated with significant differences. Note the different range of the y-axes across panels and that y-axes were kept constant for each pair of 3α,5α/3α,5β-neurosteroid metabolites. ***p≤0.001 vs air control (effect size = 1.26–2.81).

There were minimal changes in cortical neurosteroid levels at 8 h WD (Figure 6). WD significantly increased 3α,5β-androstanediol levels by 30% [F(1,32)=5.85, p<0.05] (Figure 6f) and tended to decrease 3α,5α-androsterone levels by 24% [F(1,32)=3.79, p=0.06] (Figure 6g). There were no significant correlations between the levels of ALLO and other neurosteroids in the controls or during WD (Table 1, bottom).

Figure 6. Minimal changes in cortical neurosteroid levels at 8 h WD in male DBA mice.

Mice were treated as described in the legend to Figure 5. Ethanol WD significantly increased 3α,5β-androstanediol levels and tended to decrease 3α,5α-androsterone levels. Values are shown as pg/mg and represent the mean ± SEM for the number of mice in parentheses (Group size is the same for all neurosteroids). Also shown are the WD-induced percent changes (vs air control) that were associated with significant differences or statistical trends. Note the different range of the y-axes across panels and that y-axes were kept constant for each pair of 3α,5α/3α,5β-neurosteroid metabolites. ⁺p=0.06, *p<0.05 vs air control (effect size = 0.83 for significant difference and 0.67 for trend).

Basal hippocampal neurosteroid levels were very similar to those in cortex (compare air controls in Figures 6 & 7), and some of the WD-induced changes were similar in hippocampus and cortex. Hippocampal ALLO and 3α,5β-THDOC levels were significantly decreased by 7% [F(1,23)=7.35, p=0.01] and by 12% [F(1,23)=11.47, p<0.01] during WD (Figure 7c & 7j), but a similar 7% reduction in cortical ALLO levels and 15% reduction in cortical 3α,5β-THDOC levels during WD (Figure 6c & 6j) did not reach significance. WD also significantly increased 3α,5β-androstanediol levels by 19% [F(1,23)=6.49, p<0.05] in hippocampus (Figure 7f), similar to what was observed in cortex (Figure 6f). In contrast, the WD-induced significant increase (20%) in hippocampal 3α,5α-androsterone [F(1,23)=8.36, p<0.01] (Figure 7g) was opposite of that observed in cortex (Figure 6g). There were no significant correlations between neurosteroid levels in the air controls, but ALLO levels were significantly negatively correlated with 3α,5α-androsterone levels during WD (Table 2, bottom).

Figure 7. Divergent changes in hippocampal neurosteroid levels at 8 h WD in male DBA mice.

Mice were treated as described in the legend to Figure 5. Ethanol WD significantly decreased ALLO and 3α,5β-THDOC levels and significantly increased 3α,5β-androstanediol and 3α,5α-androsterone levels. Depicted are the mean ± SEM pg/mg for the number of mice in parentheses (Group size is the same for all neurosteroids). Also shown are the WD-induced percent changes (vs air control) that were associated with significant differences. Note the different range of the y-axes across panels and that y-axes were kept constant for each pair of 3α,5α/3α,5β-neurosteroid metabolites. *p<0.05, **p≤0.01 vs air control (effect size = 1.08–1.36).

Plasma, hippocampal and cortical neurosteroid levels at 48 h WD

In this cohort of mice, BECs upon removal from the vapor chambers at 72 h were 1.84 ± 0.06 mg/mL (n=12). Plasma CORT levels were 8.25 ± 2.08 μg/dL for air-exposed (n=10) and 75.21 ± 5.70 μg/dL for 48 h WD (n=12) mice, and they were similar to the levels measured at the 8 h WD time point. WD significantly increased plasma CORT by 812% at the 48 h time point [F(1,20)=8.59, p<0.01].

The quantification of plasma neurosteroids revealed significant divergent changes and a persistent suppression in 3 neurosteroid levels at 48 h WD (Figure 8; Supplemental Table 6 for ANOVA results). There was a sustained significant reduction in plasma DHEA, ALLO, and 3α,5β-androstanediol levels (compare Figure 8b, 8c & 8f with Figure 5b, 5c & 5f). Plasma 3α,5α-THDOC also tended to be decreased at 48 h WD. In contrast, plasma pregnenolone, pregnanolone, 3α,5α-androstanediol and 3α,5α-androsterone levels were significantly increased at 48 h WD.

Figure 8. Sustained reduction in some plasma neurosteroid levels at 48 h WD in male DBA mice.

Mice were exposed to 72 h continuous ethanol vapor or air and euthanized at a time corresponding to the termination of WD-induced convulsive activity. When compared with the results in Figure 5, there was a sustained significant reduction in plasma DHEA, ALLO, and 3α,5β-androstanediol levels. In contrast, plasma pregnenolone, pregnanolone, 3α,5α-androstanediol and 3α,5α-androsterone levels were increased significantly at 48 h WD. Values are shown as pg/mL and represent the mean ± SEM for the number of mice in parentheses (Group size is the same for all neurosteroids). Also shown are the WD-induced percent changes (vs air control) that were associated with significant differences or statistical trends. Note the different range of the y-axes across panels and that y-axes were kept constant for each pair of 3α,5α/3α,5β-neurosteroid metabolites. ⁺p<0.06, *p<0.05, **p≤0.01, ***p≤0.001 vs air control (effect size = 1.16–6.40 for significant differences and 0.87 for trend).

Minimal changes were observed in cortical and hippocampal neurosteroid levels at the 48 h WD time point (not shown), and the range in levels was very similar to those depicted in Figures 6 & 7. The only significant WD-induced difference was in 3α,5α-THDOC levels, which were increased by 40% in the cortex [F(1,20)=4.62, p<0.05] and decreased by 9% in the hippocampus [F(1,20)=13.55, p=0.001].

Discussion

The present studies extended our recent report that simultaneously quantified neurosteroids in plasma of WSP-1 and WSR-1 male mice at 8 h WD to include analysis of neurosteroids in cortical and hippocampal tissue across the time course of WD-induced convulsive activity, and to compare the results to those obtained in DBA mice, another genetic animal model of high WD. Importantly, assay validation studies confirmed the high precision and accuracy of the optimized brain and plasma assays (See Supplemental Material), and significant differences were associated primarily with a large effect size (d>0.8) or a medium effect size (d>0.5) in a few cases. Thus, the present findings demonstrate that we are able to simultaneously identify and quantify 10 neurosteroids in dissected brain tissues and that the power is sufficient to support the significant differences observed.

Chronic ethanol exposure was comparable for all three genotypes in this and the previous studies, and ethanol WD produced a significant increase in plasma CORT across all three genotypes. However, the overall pattern of the neurosteroid results revealed significant genotype differences across the time course of WD. And, WD produced some divergent changes among neurosteroid levels (some increased and some decreased) and altered correlations between neurosteroid levels, indicating a broad and complex dysregulation in neurosteroid biosynthesis during WD. There also were differences in the WD-induced changes in plasma versus brain tissue, which fits with previous findings that basal neurosteroid levels in plasma do not simply reflect levels in cortex and hippocampus (Caruso et al., 2013), and argues for independent regulation of neurosteroid synthesis in periphery and brain during WD.

The significant WD-induced increase in plasma CORT levels across all three genotypes is consistent with results in rodents (e.g., Janis et al., 1998; Finn et al., 2000; Koob, 2008; Tanchuck et al., 2009) and in human alcoholics (e.g., Stalder et al., 2010; Stephens and Wand; and references therein) during the early phase of WD. Earlier work also indicated that sensitivity to the proconvulsant effect of CORT differed in WSP versus WSR and in DBA versus C57 mice. Specifically, administration of CORT to naïve mice to mimic levels associated with WD significantly increased HIC scores and increased acute ethanol WD-induced HICs only in WSP and DBA mice (Roberts et al., 1991, 1992). Additionally, administration of aminoglutethimide (a glucocorticoid synthesis blocker to decrease CORT levels) decreased acute ethanol WD severity in WSP mice (Roberts et al., 1991). Thus, genetic differences in the excitatory effects of glucocorticoids may influence WD severity.

WD-induced changes in plasma neurosteroids in DBA mice at 8 h were similar to those observed in WSP-1 and WSR-1 mice (Snelling et al., 2014), with significant decreases in levels of DHEA, ALLO, and 3α,5α/3α,5β-androstanediol. Notably, 3 of the 8 h WD-induced changes in plasma neurosteroid levels persisted at the 48 h time point only in DBA mice. However, there also were opposite changes in levels of other GABA_AR-active neurosteroids (see Figure 8). The divergent WD-induced changes in the 3α,5α- versus 3α,5β- derivatives of progesterone, testosterone, and DHEA were surprising, as they do not fit with a simple WD-induced effect on 5α- versus 5β-reductase activity on steroid precursors (for example), because an opposite pattern of changes was observed for the progesterone versus the testosterone and DHEA derived neurosteroids (compare Figure 8, panels c & d with panels e – h). Regardless, it is possible that the differing changes in neurosteroid levels at 48 h WD in DBA mice reflect an attempt to maintain a homeostatic balance in overall neurosteroid levels that is not disruptive to GABA_AR-mediated inhibition and that corresponds to the termination of WD-induced changes in convulsive activity. Likewise, lack of significant WD-induced changes in neurosteroid levels at 48 h in WSP-1 and WSR-1 mice also can be interpreted as a return to baseline levels upon the termination of the acute phase of WD in mice (at 24 – 48 h in rodents; Heilig et al., 2010). Collectively, the plasma results in WSP-1 and WSR-1 mice suggest that there was a transient suppression in the DHEA and testosterone pathway during WD and that the reduction in ALLO levels occurred at a point downstream from progesterone (discussed in Snelling et al., 2014). In DBA mice, there was a persistent suppression in plasma DHEA levels, which may reflect an effect of ethanol WD on activity of P450c17, the enzyme responsible for the conversion of pregnenolone to 17α-hydroxypregnenolone and then to DHEA (Mellon and Vaudry, 2001). It is likely that the persistent suppression in plasma ALLO and 3α,5β-androstanediol levels occurred at a point downstream of progesterone and testosterone, respectively, given the opposite changes in pregnanolone and 3α,5α-androstanediol that were observed at the 48 h time point. Notably, the finding that different modes of chronic ethanol exposure or consumption and WD produced a consistent reduction in plasma ALLO levels across rodent genotypes, in monkeys, and in humans (Beattie et al., 2017; Cagetti et al., 2004; Hill et al., 2005; Maldonado-Devincci et al., 2014; Romeo et al., 1996; Tanchuck et al., 2009) suggests that adrenal ALLO biosynthesis is suppressed during the acute phase of WD.

Absolute values of the neurosteroids during WD also should be considered when thinking about the potential functional implication of the consistent decreases in plasma ALLO levels during WD, as a decrease in levels below the physiologically relevant range would no longer potentiate GABA_AR function. This is a distinct possibility for WSP-1 versus WSR-1 mice at 8 h WD (discussion in Snelling et al., 2014). Additionally, earlier studies in human subjects found that individuals possessing a minor C-allele of the SRD5A1 gene, which encodes the enzyme 5α-reductase type 1, expressed both a higher ratio of dihydrotestosterone to testosterone (Ellis et al., 2005) and a decreased risk for alcohol dependence (Milivojevic et al., 2011), suggesting that a heightened level of 5α-reduced neurosteroid production may be protective against the development of dependence. In this context, a decline in enzyme function might be expected to decrease the biosynthesis of 5α-reduced neurosteroids and exacerbate the risk of dependence.

The pattern of changes in hippocampal and cortical neurosteroids varied across the time course of the acute WD phase in the 3 genotypes examined. It is interesting that the WSR-1 mice, which were selectively bred for low WD-induced convulsive activity, exhibited the most consistent decrease in cortical neurosteroid levels at 8 h WD; 8 neurosteroids, including ALLO, were significantly decreased. This result is consistent with the decrease in ALLO immunoreactivity in the mPFC of low WD C57 mice at 8 h WD (Maldonado-Devincci et al., 2014). However, the WD-induced decrease in ALLO immunoreactivity in the mPFC persisted at 72 h WD in the Maldonado-Devincci et al. (2014) study, whereas cortical ALLO levels were not significantly altered at 48 h WD in WSR-1 mice. These differences may be due to the differing duration and/or pattern of ethanol vapor exposure between the two studies (72 hrs continuous for WSR-1 mice; 4 bouts of repeated CIE for C57 mice). When comparing the cortical and hippocampal neurosteroid results during WD, similar % decreases in levels of 6 hippocampal neurosteroids to that seen in cortex also were observed at 8 h WD in WSR-1 mice, although the changes did not reach significance. ALLO immunoreactivity was significantly increased at 72 h WD only in the CA3 (Maldonado-Devincci et al., 2014), and we found that hippocampal pregnanolone (the 5β-stereoisomer of ALLO) levels were significantly increased at 48 h WD in WSR-1 mice. Additionally, divergent changes in hippocampal levels of other neurosteroids were observed (overall significant increases in 3 steroids, significant decreases in 3 steroids; Figure 4), although we do not know if the WD-induced changes were limited to one or several hippocampal subregions. Collectively, the relatively consistent decrease in levels of cortical and some hippocampal neurosteroids at 8 h WD in WSR-1 mice may reduce GABA_AR-neuronal inhibition in these brain regions, with a subsequent increase in glutamatergic output of the projection neurons. At 48 hr WD, the opposite changes in levels of GABA_AR-active neurosteroid levels may offset putative changes in GABA_AR inhibition to maintain homeostasis at a time corresponding to the termination of the acute WD phase. Additionally, physiologically relevant ALLO levels exert effects on presynaptic GABA_ARs that are located on GABAergic or glutamatergic nerve terminals to increase GABA (e.g., Park et al., 2011; also reviewed in Herd et al., 2007) or glutamate release (e.g., Iwata et al., 2013), respectively. While it is not known whether all GABA_AR-active neurosteroids exert similar influences on presynaptic GABA and glutamate release, brain regional differences in the neuroanatomical localization of presynaptic GABA_ARs could produce mixed effects on CNS excitability.

Comparison of the hippocampal and cortical neurosteroid results in WSP-1 and DBA mice revealed that ethanol WD produced a different pattern of outcomes in these two genetic animal models of high ethanol WD. For instance, ALLO only was significantly decreased in hippocampus at 8 h WD in DBA mice, and significant changes in the levels of other hippocampal neurosteroids in DBA mice corresponded to an unchanged neurosteroid level in WSP-1 mice. Another interesting finding is that of the small proportion of cortical neurosteroids that were significantly changed, all were significantly increased in both WSP-1 mice and DBA mice (although different neurosteroids were increased in WSP-1 versus DBA mice). Because GABA levels can produce an inverse effect on steroidogenic enzyme activity in rodent brain (Do Rego et al., 2009), it is possible that the increase in neurosteroid levels reflects a decrease in GABA release during ethanol WD. Consistent with this idea, preliminary evidence suggests that 72 h continuous ethanol vapor decreased presynaptic GABA release in dentate gyrus granule cells, measured by a decrease in frequency of miniature inhibitory post-synaptic currents in WSP-1 mice (Mohr, Richardson, Finn & Rossi, unpublished), although we do not know if similar WD-induced changes in GABA release occur in DBA mice. Thus, at least in WSP-1 mice, the increase in neurosteroid levels may reflect a homeostatic counter to decreases in GABA release. Similarly, 8 h ethanol WD also produced tolerance to the anticonvulsant effect of ALLO, and this decreased behavioral sensitivity to ALLO was accompanied by significantly reduced functional sensitivity of GABA_ARs to ALLO in WSP mice (Finn et al., 2006), again compatible with increases in neurosteroid levels being a homeostatic counter to reduced sensitivity during WD, albeit clearly not adequate, given the heightened WD phenotype. Collectively, the results suggest that the decreased sensitivity of GABA_ARs to ALLO may exert a greater functional influence on GABA_AR-mediated inhibition than the detected increases in neurosteroid levels in cortex and hippocampus and that, taken in conjunction with the concomitant decrease in presynaptic GABA release, would increase neuronal excitability in WSP-1 mice.

As mentioned above, hippocampal ALLO levels were decreased significantly in DBA mice at 8 h WD, as were 3α,5β-THDOC levels. As ALLO and THDOC are the most potent positive allosteric neurosteroid modulators of GABA_ARs, the functional implication of these WD-induced changes would be a decrease in GABA_AR inhibition. In contrast, levels of 3α,5β-androstanediol and 3α,5α-androsterone were significantly increased. It is possible that these opposite changes in neurosteroid levels produce a net effect that does not functionally alter GABA_AR-inhibition at 8 h WD. However, other studies have determined that 8 h ethanol WD produced tolerance to the anticonvulsant efficacy of ALLO, suggestive of decreased sensitivity of GABA_ARs to ALLO during WD in DBA mice (Finn et al., 2000). Notably, acute ethanol WD severity in male DBA mice is altered in a bidirectional manner by manipulation of GABA_AR-active neurosteroid levels (Gililland and Finn, 2007; Kaufman et al., 2010); i.e., adrenalectomy and gonadectomy (ADX/GDX) to reduce neurosteroids in the periphery increased acute ethanol WD severity, replacement with progesterone and DOC (GABA_AR-active precursors) offset the ADX/GDX-induced increase in WD severity, and the “rescue” effect of progesterone and DOC was blocked with co-administration of finasteride (a 5α-reductase inhibitor that blocks metabolism of progesterone and DOC to GABA_AR-active neurosteroids ALLO and 3α,5α-THDOC). Thus, the ADX/GDX results (Gililland and Finn, 2007; Kaufman et al., 2010) provide evidence that endogenous neurosteroid levels can influence acute ethanol WD severity. Additional studies are necessary to determine whether different GABA_AR-active neurosteroid mechanisms influence ethanol WD severity in DBA versus WSP-1 mice, which would not be surprising given the multiple other differences between these genotypes. Importantly, genotypic characteristics of DBA mice arose fortuitously during inbreeding, while the genotypic characteristics of WSP and WSR mice were markedly influenced by direction selective pressures.

In conclusion, chronic ethanol exposure and WD dysregulated neurosteroid synthesis, with evidence for independent regulation of neurosteroid synthesis in the periphery and brain regions examined. The decrease in ALLO levels at 8 h WD in WSR-1 mice in plasma, cortex, and hippocampus, in DBA mice in plasma and hippocampus, and in WSP-1 mice in plasma (Snelling et al., 2014; this study) is consistent with data from other studies showing decreases in ALLO levels during the early phase of WD in rodents, monkeys, and humans (Beattie et al., 2017; Cagetti et al., 2004; Hill et al., 2005; Maldonado-Devincci et al., 2014; Romeo et al., 1996; Tanchuck et al., 2009). The results also add to evidence that neurosteroid levels are altered in several neuropsychiatric and neurological diseases and suggest that strategies to enhance aspects of neurosteroid synthesis may have therapeutic potential (Porcu et al., 2016).