Alcohol reduces the activity of somatostatin interneurons in the mouse prefrontal cortex: a neural basis for its disinhibitory effect?

Abstract

The prefrontal cortex (PFC) is involved in executive (“top-down”) control of behavior and its function is especially susceptible to the effects of alcohol, leading to behavioral disinhibition that is associated with alterations in decision making, response inhibition, social anxiety and working memory. The circuitry of the PFC involves a complex interplay between pyramidal neurons (PNs) and several subclasses of inhibitory interneurons (INs), including somatostatin (SST)-expressing INs. Using in vivo calcium imaging, we showed that alcohol dose-dependently altered network activity in layers 2/3 of the prelimbic subregion of the mouse PFC. Low doses of alcohol (1 g/kg, intraperitoneal, i.p.) caused moderate activation of SST INs and weak inhibition of PNs. At moderate to high doses, alcohol (2–3 g/kg) strongly inhibited the activity of SST INs in vivo, and this effect may result in disinhibition, as the activity of a subpopulation of PNs was simultaneously enhanced. In contrast, recordings in brain slices using ex vivo electrophysiology revealed no direct effect of alcohol on the excitability of either SST INs or PNs over a range of concentrations (20 and 50 mM) consistent with the blood alcohol levels reached in the in vivo experiments. This dose-dependent effect of alcohol on SST INs in vivo may reveal a neural basis for the disinhibitory effect of alcohol in the PFC mediated by other neurons within or external to the PFC circuitry.

1. Introduction

Alcohol use disorders have widespread consequences, negatively affecting the health of individuals, and are also the cause of economic and societal costs. Despite increasing efforts to identify genetic influences and neuroadaptations that cause alcohol use disorders (Koob and Le Moal, 2008; Koob and Volkow, 2016, 2010), progress has been limited by a shortage of effective therapies (Heilig et al., 2019), the complexity of the underlying neural circuitry and by a lack of understanding of how alcohol perturbs network activity.

Alcohol is known to have short-term effects on the function of a variety of brain areas, including the ventral tegmental area, ventral and dorsal striatum, amygdala, thalamus, hippocampus and prefrontal cortex (PFC) (Abrahao et al., 2017; Harrison et al., 2017) that may be mediated via the actions of alcohol on specific molecular targets including ion channels and neurotransmitter receptors. These rapid effects on neuronal networks can then induce long-term neuroadaptations (Koob and Le Moal, 2008) in a variety of brain circuits that result in the behavioral changes associated with alcohol withdrawal, craving and dependence (Koob, 2014). The study of the acute effects of alcohol at low and medium doses in specific brain regions, especially in executive areas such as the PFC, may reveal early mechanisms and cellular targets that might initiate the complex behavioral adaptations to chronic alcohol.

The PFC is an extensively interconnected cortical region that mediates “top-down” executive control over goal-directed behaviors (Ballard et al., 2011; Murty et al., 2017). Dysfunction of the PFC is thought to underlie compulsive drug-taking and relapse in substance abusers, including alcoholics (Koob and Volkow, 2010), but relatively little is known about the effects of alcohol on the neural circuitry in the PFC. An early consensus view based on whole-brain imaging studies in human drinkers was that alcohol depresses activity in neuronal circuitry in the PFC (Volkow et al., 1990, 1988) although the complexity of cortical circuitry was not yet fully appreciated at the time. Comprehensive electrophysiological studies have subsequently yielded sparse evidence for the inhibition of principal cells in PFC by alcohol. Although some studies have shown a depression of NMDA-receptor mediated EPSCs (Weitlauf and Woodward, 2008) and enhancement of glycine receptor-mediated tonic inhibition (Badanich et al., 2013; Nimitvilai et al., 2016) in pyramidal neurons (PNs), these observations were all made at higher levels of alcohol in brain slices. Some reports suggest that chronic alcohol exposure can enhance excitability of PNs in the mouse PFC (Pleil et al., 2015; Salling et al., 2018), whereas exposure to acute ethanol reduces excitability in orbitofrontal cortex neurons (Badanich et al., 2013; Cannady et al., 2020; Nimitvilai et al., 2017) and in addition, there is some evidence for depression of “up-states” and persistent firing in PFC PNs (Tu et al., 2007; Woodward and Pava, 2009).

It has recently become obvious that cortical function is dependent, not just on PNs, but also on a large group of GABAergic interneurons (INs) that are diverse in morphology, neurochemistry, connectivity and physiology (Gouwens et al., 2020; Kepecs and Fishell, 2014; Markram et al., 2004; Rudy et al., 2011; The Petilla Interneuron Nomenclature Group et al., 2008). There are three major distinct subclasses of cortical INs: the parvalbumin (PV)-, somatostatin (SST)- and 5HT3 receptor-expressing subtypes (Rudy et al., 2011). Each subset of INs receives inputs from different subcortical regions (Ährlund-Richter et al., 2019; Canetta et al., 2020) and plays a distinct role in controlling activity within the cortex. INs were once thought to simply prevent excessive activity in PNs, but it is now appreciated that INs exert a critical influence on the timing of PN firing (Barthó et al., 2004; Constantinidis and Goldman-Rakic, 2002; González-Burgos et al., 2005) and hence influence local network rhythmicity (Barthó et al., 2004; Chrobak and Buzsáki, 1998).

In cortical layers (L) 2/3, SST INTs provide the main inhibitory input to PNs (Tremblay et al., 2016; Wang et al., 2004) and also contribute to the control of other inhibitory neurons (Pfeffer et al., 2013). Disinhibitory control of SST INs themselves (Ma et al., 2006; Wang et al., 2004; Xu et al., 2006) is involved in fear and anxiety (Cummings and Clem, 2020) and SST INs are also involved in other stress-related behaviors and substance abuse disorders (Robinson and Thiele, 2020). Recent evidence suggests a sex-dependent alteration of excitability in fast-spiking and Martinotti cells (Hughes et al., 2020) of the prelimbic (PL) PFC after chronic intoxication with ethanol, although voluntary binge drinking models have shown increases in the excitability of PV INs, but not SST INs in the same region of all mice (Joffe et al., 2020). Despite the cumulative evidence for dysfunction of PL PFC after chronic alcohol exposure (Pleil et al., 2015; Salling et al., 2018; Trantham-Davidson et al., 2017), the effects of acute alcohol on PL SST INs and PNs in the intact brain are largely unexplored.

In this study, we examined the activity of L2/3 PNs and SST INs in the PL region of mouse PFC before and after acute alcohol exposure. We investigated three ethanol dose ranges of relevance to human intoxication: low (0.5 and 1 g/kg), moderate (2 g/kg) and high (3 g/kg) doses. Using in vivo Ca²⁺ imaging in awake mice, we showed that the effects of alcohol on neuronal activity were dose-dependent: low dose alcohol moderately activated SST INs and weakly inhibited PNs, whereas moderate to high doses of alcohol reduced SST IN activity and enhanced PN activity. In contrast, ex vivo electrophysiology demonstrated no consistent direct effect of alcohol on the resting excitability of either SST INs or PNs. These findings therefore suggest that the inhibitory actions of alcohol on SST INs in vivo may be consequent on the drug acting on neurons external to the PFC, or on other groups of neurons within the PFC circuitry.

2. Methods

2.1. Mice

We used male and female transgenic mice expressing GCaMP6s in L2/3 PNs (Cichon et al., 2020). Thy1-GCaMP6 slow founder lines 1 and 3 were gifts from Dr. Wen-Biao Gan at New York University School of Medicine. C57BL/6J (Stock No: 000664), SST-IRES-Cre (Stock No: 013044) and Ai9 td-tomato reporter (Stock No: 007909) mice were purchased from the Jackson Laboratory. To identify SST INs for patch-clamp recordings, SST-IRES-Cre mice were crossed with the Ai9 td-tomato reporter mice. Mice were group-housed in temperature-controlled rooms on a 12-h light-dark cycle and were randomly assigned to different treatment groups. One to three-month-old mice of both sexes were used for calcium imaging and patch clamp recordings. All animal procedures were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of Columbia University as consistent with National Institutes of Health (NIH) Guidelines for the Care and Use of Laboratory Animals.

2.2. Alcohol application and blood alcohol measurement

For alcohol application, 100% reagent alcohol (Fisherbrand) was diluted in sterile normal saline to 25% v/v. The doses (0.5, 1, 2 or 3 g/kg) of alcohol were administrated via intraperitoneal (i.p.) injection immediately following the acquisition of the calcium images under baseline conditions. Each animal received only one dose of alcohol by i.p. injection. In a separate group of age-matched mice (n = 19), equivalent doses (0.5, 1, 2, and 3 g/kg) of alcohol were injected i.p. and blood was collected from the submandibular vein at multiple time points (10, 30, 60, and 120 minutes). Plasma was separated via centrifugation and blood alcohol concentration was determined using an Alcohol reagent kit (GMRD-113, Analox Instruments) and AM1 alcohol analyzer (Analox Instruments, London, England).

2.3. Surgical preparation for imaging awake, head-restrained mice

Calcium imaging was carried out in awake, head-restrained mice in a quiet resting state (Yang et al., 2013). Surgery preparation for awake animal imaging includes attaching a head holder and creating a cranial window as described earlier (Yang et al., 2010). Specifically, mice were deeply anesthetized with an i.p. injection of 100 mg/kg ketamine and 15 mg/kg xylazine. The mouse head was shaved and the skull surface was exposed with a midline scalp incision. The periosteum tissue over the skull surface was removed without damaging the temporal and occipital muscles. A head mount was attached to the animal’s skull with glue (Loctite 495) to help restrain the animal’s head and reduce motion-induced artifact during imaging. A small skull region (~1 mm in diameter) located over the PL cortex based on stereotaxic coordinates (2.68 mm anterior and 0.5 mm lateral to the bregma) was removed and a round glass coverslip (approximately the same size as the bone being removed) was glued to the skull. Dental acrylic cement was applied to surrounding area to secure metal bars with precaution not to cover the glass region with cement. Throughout the surgical procedure and recovery, the animal’s body temperature was maintained at ~37 °C.

Upon awakening, mice with head mounts were habituated on three occasions (10 min / session) in a custom-built imaging platform to minimize potential stress effects of head restraining and imaging. In this device, the animal’s head-mount was secured into metal blocks such that the head was fixed and perpendicular to the two-photon objective. The animal’s body was allowed to rest against the bottom of the solid imaging plate. Imaging experiments were started 24 h after window implantation, at which time all animals had recovered from any anesthetic effects.

2.4. In vivo Ca²⁺ imaging and data analysis

The genetically-encoded Ca²⁺ indicator GCaMP6 slow (GCaMP6s) was used for Ca²⁺ imaging of PNs and INs in the PL. We utilized Thy1-GCaMP6s (founder lines 1 and 3) for imaging PNs and SST-Cre lines for IN imaging. For SST-Cre mice, Cre-dependent GCaMP6s was expressed with recombinant adeno-associated virus (AAV) under the synapsin promoter [AAV, serotype 9; > 1×10¹³ (GC/ml) titer; Addgene]. 0.2 μl of AAV in saline were injected (Picospritzer III; 15 p.s.i., 10 ms, 0.5 Hz) over 10–15 min into L2/3 of the PL using a glass microelectrode around the coordinates of AP +2.68 mm, ML 0.5 mm, DV 0.8 mm (Paxinos and Franklin, 2001). The expression of GCaMP6s was robust in Thy1-GCaMP6s (Fig. 1C, Fig. 2A) and SST-IRES-Cre and (Fig. 3A, Fig. 4A), consistent with previous reports (Adler et al., 2019; Cichon et al., 2017; Dana et al., 2014).

Fig. 1. Low doses of alcohol decrease PN activity in the PL in vivo.

Schematic showing in vivo two-photon imaging in the prelimbic (PL) region of PFC through a glass window. Dashed lines indicate the region of PL and the bottom of L2/3 of the cortex. M2, secondary motor cortex. Whole cell recordings were performed in L2/3 of acute PFC slices. (B) Blood alcohol concentrations (BACs) after a bolus injection of alcohol at different doses. At 10 min, the corresponding BACs (in mM) for 0.5, 1, 2 and 3 g/kg were, respectively, 7.9 ± 1.6, 28.5 ± 1.2, 39.4 ± 5.9 and 63.4 ± 4.5. (C) Representative time lapse images and fluorescence traces of PNs expressing GCaMP6s in L2/3 of PL before and after saline injection. Scale bar, 20 μm. (D) Average integrated Ca²⁺ activity of L2/3 PNs before and after saline injection. (Baseline: 18.57 ± 0.50%, 10 min: 18.28 ± 0.37%, P = 0.3274, 30 min: 18.71 ± 0.53%, P = 0.7785, 60 min: 18.23 ± 0.48%, P = 0.7947, 120 min: 17.91 ± 0.39%, P = 0.3725, n = 225 cells from 4 mice). (E) Average integrated Ca²⁺ activity of L2/3 PNs before and after 0.5 g/kg ethanol (EtOH) injection. (Baseline: 18.67 ± 0.54%, 10 min: 17.91 ± 0.44%, P = 0.5373, 30 min: 18.39 ± 0.64%, P = 0.1279, 60 min: 18.72 ± 0.53%, P = 0.6874, 120 min: 18.03 ± 0.45%, P = 0.7231; n = 184 cells from 3 mice). (F) Average integrated Ca²⁺ activity of L2/3 PNs before and after 1 g/kg alcohol injection. (Baseline: 18.38 ± 0.37%, 10 min: 17.78 ± 0.57%, P = 0.0007, 30 min: 17.80 ± 0.60%, P = 0.0007, 60 min: 18.13 ± 0.35%, P = 0.8302, 120 min: 19.31 ± 0.44%, P = 0.0543, n = 212 cells from 4 mice). (D-F) Bottom, the proportions of neurons showing increases, decreases or no change in Ca²⁺ activity after injection. (G) Average integrated Ca²⁺ activity of L2/3 PNs at 30 min after saline, 0.5 g/kg and 1 g/kg alcohol injection. (Saline: 18.71 ± 0.53%, n = 225 cells from 4 mice, 0.5 g/kg: 18.39 ± 0.64%, P = 0.2077, n = 184 from 3 mice, 1 g/kg: 17.80 ± 0.60%, P = 0.0038, n = 212 from 4 mice). (H) Amplitude of L2/3 PN Ca²⁺ transients at 30 min after saline, 0.5 g/kg and 1 g/kg alcohol injection. (Saline: 48.81 ± 1.08%, n = 225 cells from 4 mice, 0.5 g/kg: 46.69 ± 1.02%, P = 0.5662, n = 184 from 3 mice, 1 g/kg: 48.43 ± 1.30%, P = 0.2868, n = 212 from 4 mice). (I) Frequency of L2/3 PN Ca²⁺ transients at 30 min after saline, 0.5 g/kg and 1 g/kg alcohol injection. (Saline: 4.24 ± 0.14, n = 225 cells from 4 mice, 0.5 g/kg: 3.39 ± 0.11, P = 0.0002, n = 184 from 3 mice, 1 g/kg: 3.31 ± 0.14, P < 0.0001, n = 212 from 4 mice). (J) Duration (s) of L2/3 PN Ca²⁺ transients at 30 min after saline, 0.5 g/kg and 1 g/kg alcohol injection. (Saline: 2.11 ± 0.08, n = 225 cells from 4 mice, 0.5 g/kg: 2.03 ± 0.08, P > 0.9999, n = 184 from 3 mice, 1 g/kg: 2.07 ± 0.11, P = 0.8774, n = 212 from 4 mice). Data in (D-J) are presented as mean ± s.e.m. n.s., not signification, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. baseline or saline by Dunn’s post hoc multiple comparison test. (K) Confocal image of a representative biocytin-filled pyramidal neuron in L2/3 section stained with DAPI (blue). Scale bar, 100 μm. (L) Representative recordings in L2/3 PNs before (gray) and in the presence of 20 mM ethanol (red) in response to −200 and +200 pA stimulus. Rheobase traces (in response to a +80 pA current step) are shown in black. Gray arrow indicates voltage sag and black arrows indicate ADP (top arrow) and mAHP (bottom arrown). (M) Comparison of the excitability of PNs by eliciting APs during 500 ms, 20 pA current steps and examining input-output curves during ACSF (open circles), 20 mM ethanol (red solid circles) and washout (gray solid circles). 20 mM ethanol did not alter the number of elicited spikes and did not change after washout compared to ACSF (two-way mixed ANOVA, P = 0.6348, n = 10 cells, except washout: n = 8 cells, from 8 mice). The values plotted in the I-V graph are expressed as mean ± s.e.m.

Fig. 2. Moderate to high doses of alcohol increase PN activity in the PL cortex in vivo but not intrinsic excitability ex vivo.

(A) Representative time lapse images and fluorescence traces of PNs expressing GCaMP6s in L2/3 of PL before and after 3 g/kg alcohol injection. Scale bar, 20 μm. (B) Average integrated Ca²⁺ activity of L2/3 PNs before and after 2 g/kg alcohol injection. (Baseline: 18.57 ± 0.40%, 10 min: 22.34 ± 0.73%, P < 0.0001, 30 min: 24.16 ± 1.05%, P < 0.0001, 60 min: 23.13 ± 0.85%, P < 0.0001, 120 min: 18.42 ± 0.47%, P = 0.1633; n = 264 cells from 3 mice). (C) Average integrated Ca²⁺ activity of L2/3 PNs before and after 3 g/kg alcohol injection. (Baseline: 18.95 ± 0.48%, 10 min: 20.17 ± 0.49%, P = 0.0563, 30 min: 23.13 ± 0.74%, P < 0.0001, 60 min: 24.33 ± 0.88%, P < 0.0001, 120 min: 20.32 ± 0.47%, P = 0.0008; n = 271 cells from 4 mice). (B-C) Bottom, the proportions of neurons showing increases, decreases or no change in Ca²⁺ activity after injection. (D) Average integrated Ca²⁺ activity of L2/3 PNs at 30 min after saline, 2 and 3 g/kg alcohol injection. (Saline: 18.71 ± 0.53%, n = 225 cells from 4 mice, 2 g/kg: 24.16 ± 1.05%, P < 0.0001, n = 264 from 3 mice, 3 g/kg: 23.13 ± 0.74%, P < 0.0001, n = 271 from 4 mice). (E) Amplitude of L2/3 PN Ca²⁺ transients at 30 min after saline, 2 g/kg and 3 g/kg alcohol injection. (Saline: 48.81 ± 1.08%, n = 225 cells from 4 mice, 2 g/kg: 58.92 ± 1.88%, P < 0.0001, n = 264 from 3 mice, 3 g/kg: 49.03 ± 1.13%, P > 0.9999, n = 271 from 4 mice). (F) Frequency of Ca²⁺ transients at 30 min after saline, 2 g/kg and 3 g/kg alcohol injection. (Saline: 4.24 ± 0.14, n = 225 cells from 4 mice, 2 g/kg: 4.79 ± 0.13, P = 0.0223, n = 264 from 3 mice, 3 g/kg: 6.15 ± 0.27, P < 0.0001, n = 271 from 4 mice). (G) Duration of Ca²⁺ transients at 30 min after saline, 2 and 3 g/kg alcohol injection. (Saline: 2.11 ± 0.08, n = 225 cells from 4 mice, 2 g/kg: 2.60 ± 0.17, P = 0.4864, n = 264 from 3 mice, 3 g/kg: 2.13 ± 0.09, P > 0.9999, n = 271 from 4 mice). Data in (B-G) are presented as mean ± s.e.m. n.s., not signification, *P < 0.05, **P < 0.01, ****P < 0.0001 vs. baseline or saline by Dunn’s post hoc multiple comparison test. (H) Representative traces in response to current injection steps (−200, +80, +200 pA) do not show differences in action potential spiking in the presence of 50 mM ethanol (brown) compared with control (gray). (I) Ethanol (50 mM, brown circles) did not alter cell firing (mean spikes ± s.e.m) of L2/3 PNs compared to ACSF (open circles) (mixed model two-way ANOVA, treatment F_{(2, 18)} treatment = 1.448, P = 0.2703, n = 10 cells, except washout: n = 6 cells, from 9 mice).

Fig. 3. Low doses of alcohol increase SST activity in the PL in vivo.

(A) Representative time lapse images and fluorescence traces of SST INs expressing GCaMP6s in the PL before and after 1 g/kg alcohol injection. Scale bar, 20 μm. (B) Average integrated Ca²⁺ activity of L2/3 SST INs before and after saline injection. (Baseline: 28.23 ± 1.23%, 10 min: 28.01 ± 0.99%, P = 0.3348; 30 min: 26.79 ± 0.87%, P = 0.7135; 60 min: 26.92 ± 0.87%, P = 0.7163; 120 min: 26.97 ± 1.01%, P = 0.9655; n = 156 cells from 7 mice). (C) Average integrated Ca²⁺ activity of L2/3 SST INs before and after 0.5 g/kg alcohol injection. (Baseline: 28.22 ± 2.58%, 10 min: 27.13 ± 1.79%, P = 0.3355; 30 min: 29.98 ± 1.92%, P = 0.0186; 60 min: 32.10 ± 3.13%, P = 0.2658; 120 min: 37.57 ± 3.10%, P = 0.0002; n = 103 cells from 4 mice). (D) Average Ca²⁺ activity of L2/3 SST INs before and after 1 g/kg alcohol injection. (Baseline: 28.44 ± 1.04%, 10 min: 25.74 ± 1.15%, P = 0.0049, 30 min: 29.75 ± 1.31%, P = 0.6992, 60 min: 32.99 ± 1.51%, P = 0.0695, 120 min: 34.89 ± 1.84%, P = 0.0040, n = 107 cells from 5 mice). (B-D) Bottom, the proportions of neurons showing increases, decreases or no change in Ca²⁺ activity after injection. (E) Average integrated Ca²⁺ activity of SST INs at 120 min after saline, 0.5 and 1 g/kg alcohol injection. (Saline: 26.97 ± 1.01%, n = 156 cells from 7 mice, 0.5 g/kg: 37.57 ± 3.10%, P = 0.0066, n = 103 from 4 mice, 1 g/kg: 34.89 ± 1.84%, P < 0.0001, n = 107 from 5 mice). (F) Amplitude of Ca²⁺ transients at 120 min after saline, 0.5 and 1 g/kg alcohol injection. (Saline: 58.90 ± 1.48%, n = 156 cells from 7 mice, 0.5 g/kg: 80.39 ± 5.37%, P = 0.0089, n = 103 from 4 mice, 1 g/kg: 68.10 ± 2.49%, P = 0.0046, n = 107 from 5 mice). (G) Frequency of Ca²⁺ transients at 120 min after saline, 0.5 and 1 g/kg alcohol injection. (Saline: 2.50 ± 0.17, n = 156 cells from 7 mice, 0.5 g/kg: 2.34 ± 0.18, P > 0.9999, n = 103 from 4 mice, 1 g/kg: 4.03 ± 0.19, P < 0.0001, n = 107 from 5 mice). (H) Duration of Ca²⁺ transients at 120 min after saline, 0.5 and 1 g/kg alcohol injection. (Saline: 4.45 ± 0.52, n = 156 cells from 7 mice, 0.5 g/kg: 6.33 ± 1.05, P = 0.1184, n = 103 from 4 mice, 1 g/kg: 3.80 ± 0.48, P > 0.9999, n = 107 from 5 mice). Data in (B-H) are presented as mean ± s.e.m. n.s., not signification, *P < 0.05, **P < 0.01, ****P < 0.0001 vs. baseline or saline by Dunn’s post hoc multiple comparison test. (I) In the left panel, confocal image showing the distribution of Cre-SST/Td-Tomato cells (in red) in the PL cortex. Scale bar, 200 μm. The right panel shows enlarged view of a representative L2/3 SST IN filled with biocytin (green) during patch-clamp recording. In red, neurons of SST-Cre line expressing td-Tomato. CC: corpus callosum. Scale bar, 100 μm. (J) Representative traces of L2/3 SST INs in ACSF (gray) and during bath application of 20 mM ethanol (red) in response to −200 and +200 pA pulses. Rheobase traces induced by +40 pA current injection are shown in black. (K) Input-output curve of the number of spikes (mean ± s.e.m.) of SST interneurons before (open circles), during exposure to 20 mM (red circles) and after washout (gray circles). Ethanol (20 mM) did not affect the excitability of the L2/3 SST interneurons (mixed model two-way ANOVA, F_{(2, 20)} treatment = 0.2483 P = 0.7825, n = 11 cells, except washout: n = 7 cells, from 9 mice). Data in I-V plot are presented as mean ± s.e.m.

Fig. 4. Moderate to high doses of alcohol decrease SST IN activity in the PL in vivo.

(A) Representative time lapse images and fluorescence traces of SST INs expressing GCaMP6s in L2/3 of the PL before and after 2 g/kg alcohol injection. Scale bar, 20 μm. (B) Average integrated Ca²⁺ activity of L2/3 SST INs before and after 2 g/kg alcohol injection. (Baseline: 27.97 ± 1.06%, 10 min: 20.33 ± 1.19%, P < 0.0001, 30 min: 24.73 ± 1.71%, P < 0.0001, 60 min: 28.70 ± 1.44%, P = 0.6202, 120 min: 30.24 ± 1.58%, P = 0.8417; n = 137 cells from 5 mice). (C) Average integrated Ca²⁺ activity of L2/3 SST INs before and after 3 g/kg alcohol injection. (Baseline: 28.96 ± 1.30%, 10 min: 15.33 ± 1.22%, P < 0.0001, 30 min: 17.60 ± 1.11%, P < 0.0001, 60 min: 24.64 ± 2.04%, P < 0.0001, 120 min: 34.48 ± 3.07%, P = 0.7042; n = 132 cells from 4 mice). (B-C) Bottom, the proportions of neurons showing increases, decreases or no change in Ca²⁺ activity after injection. (D) Average integrated Ca²⁺ activity of SST INs at 30 min after saline, 2 and 3 g/kg alcohol injection. (Saline: 26.79 ± 0.87%, n = 156 cells from 7 mice, 2 g/kg: 24.73 ± 1.71%, P = 0.0002, n = 137 from 5 mice, 3 g/kg: 17.60 ± 1.11%, P < 0.0001, n = 132 from 4 mice). (E) Amplitude of L2/3 SST Ca²⁺ transients at 30 min after saline, 2 g/kg and 3 g/kg alcohol injection. (Saline: 61.87 ± 1.35%, n = 156 cells from 7 mice, 2 g/kg: 54.37 ± 2.46%, P < 0.0001, n = 137 from 5 mice, 3 g/kg: 39.20 ± 1.43%, P < 0.0001, n = 132 from 4 mice). (F) Frequency of Ca²⁺ transients at 30 min after saline, 2 and 3 g/kg alcohol injection. (Saline: 2.77 ± 0.17, n = 156 cells from 7 mice, 2 g/kg: 1.72 ± 0.14, P < 0.0001, n = 137 from 5 mice, 3 g/kg: 1.03 ± 0.10, P < 0.0001, n = 132 from 4 mice). (G) Duration of Ca²⁺ transients at 30 min after saline, 2 g/kg and 3 g/kg alcohol injection. (Saline: 3.66 ± 0.33, n = 156 cells from 7 mice, 2 g/kg: 5.64 ± 1.02, P = 0.0003, n = 137 from 5 mice, 3 g/kg: 2.45 ± 0.62, P < 0.0001, n = 132 from 4 mice). Data in (B-G) are presented as mean ± s.e.m. ***P < 0.001, ****P < 0.0001 vs. baseline or saline by Dunn’s post hoc multiple comparison test. (H) Bath application of 50 mM ethanol (brown) had no effect on firing rate in L2/3 SST INs compared to ACSF (gray). Representative traces in response to current injection steps (−200, +60, +120 pA). Rheobase traces are shown in black. (I) High concentration of ethanol (50 mM, brown circles) and washout (gray circles) did not alter cell firing of SST INs compared to ACSF (open circles) (two-way ANOVA, F_{(2, 20)} treatment = 0.098, P = 0.9064, n = 11 cells, except washout: n = 10 cells, from 8 mice). Data in I-V plot are presented as mean ± s.e.m.

Transgenic mice infected with AAV were prepared for head-fixation and imaging 2 weeks following AAV microinjection. In vivo two-photon imaging was performed with a Scientifica Hyperscope two-photon system equipped with a Ti:Sapphire laser (Vision S, Coherent) tuned to 920 nm. The average laser power on the sample was ~20–30 mW for imaging. All experiments were performed using a 25X objective (1.05 N.A.) immersed in an artificial cerebrospinal fluid (ACSF) solution and without digital zoom. All images were acquired at frame rates of ~1.07 Hz (2-μs pixel dwell time). The resolution of somatic Ca²⁺ imaging was 512 × 512 pixels. Image acquisition was performed using ScanImage software and analyzed post hoc using NIH ImageJ software.

During quiet resting, motion-related artifact due to respiration and heartbeat was typically less than 2 μm as detected in our cortical measurements. Vertical movements were infrequent and minimized by habituation, two micro-metal bars attached to the animal’s skull (described above) by dental acrylic, and a custom-built body platform. If the animal struggled in the body platform, imaging time points from those segments were excluded from quantification.

Regions of interest (ROIs) corresponding to visually identifiable somas were selected for quantification as previously described (Cichon et al., 2017). L2/3 cells were distinguished by their somatic distance to the midline. All GCaMP-expressing neurons were included in data analysis. The fluorescence time course of each cell was measured by averaging all pixels within the ROIs covering the somas. A background value was first subtracted from GCaMP6 fluorescence values.

All Ca²⁺ transients were calculated as: ΔF/F₀, where ΔF/F₀ is (F–F₀)/F₀ and F₀ is the baseline value defined as averaged fluorescence over a 12-s period of lowest fluorescence signal over the recording period. The average integrated calcium activity was the average of ΔF/F₀ over 2 min (Cichon and Gan, 2015). The threshold for determining calcium transients was calculated as three times the standard deviation (SD) of baseline fluorescence and then frequency, duration and amplitude and duration of the calcium transients were calculated (Cichon and Gan, 2015). The frequency of calcium transients was calculated as the number of calcium transients per minute. The duration was calculated as the total time of calcium transients with ΔF/F₀ above the threshold. The amplitude was the peak ΔF/F₀ of the calcium transient. Neurons showing increases or decreases in Ca²⁺ activity were defined as cells whose somatic activity deviated more than 3 SD from the mean baseline activity and neurons showing no change in Ca²⁺ activity were defined as those less than 3 SD (Adler et al., 2019). The baseline Ca²⁺ activity was comparable in PNs of females (n = 10, 18.55 ± 0.28%) and males (n = 8, 18.73 ± 0.30%) and in SST INs of females (n = 17, 27.66 ± 0.77%) and males (n = 8, 29.67 ± 0.77%).

2.5. Electrophysiology in acute brain slices

Mice were fully anesthetized with isoflurane and decapitated into ice-cold (4 °C) artificial cerebrospinal fluid (ACSF) composed of (in mM): 124 NaCl, 2.5 KCl, 2 MgSO₄, 1.25 NaH₂PO₄, 2 CaCl₂, 26 NaHCO₃ and 12.5 Glucose (pH 7.3–7.4, 300–305 mOsm). Brains were dissected and sectioned in cold ACSF using a VT1000S vibratome (Leica Biosystems) into coronal slices (250–300 μm thick) containing the PFC. Slices were placed in a holding chamber containing oxygenated (bubbled with 95% O₂-5% CO₂) ACSF at 34 °C for 30 min. Slices were then transferred to room temperature at least another 30 min before recordings were initiated. Brain slices were placed in a submerged chamber continuously perfused with oxygenated ACSF at 32 °C at a flow rate of 1–2 ml/min. Brain slices were visualized under an upright light microscope (BX51WI, Olympus) coupled to a camera (C8484, Hamamatsu). The PL area was identified according to cytoarchitectural criteria (Paxinos and Franklin, 2001) and L2/3 was defined visually under bright-field microscopy as the band between ~120 and 350 μm depth. SST neurons were identified by Ai9-TdTomato fluorescence and PNs were identified by their size and triangular shape under differential interference contrast microscopy.

Pipettes pulled from borosilicate glass (World Precision Instruments, TW150F-4) with a resistance of 3–6 MΩ were filled with a potassium gluconate intracellular solution containing (in mM): 126 Potassium gluconate, 4 KCl, 10 HEPES, 0.3 EGTA, 4 Mg-ATP, 0.3 Na-GTP, 10 Na-phosphocreatine (300–305 mOsm). The pH was adjusted to 7.2–7.3 using KOH. 0.4% biocytin (Sigma B4261) was added to the intracellular solution for later morphological identification of the recorded neurons. Briefly, slices with biocytin-filled neurons were fixed in 4% paraformaldehyde overnight. Slices were washed in phosphate-buffered saline (PBS) and incubated with streptavidin Alexa Fluor 488 (Invitrogen) for 2 h. Slices were then washed with PBS and stained with DAPI before being mounted on microscope slides.

Electrophysiological recordings were performed with a MultiClamp 700B (Molecular Devices), digitized with a 1440 Digitizer (Molecular Devices) and interfaced with Clampex 10.7 software (Molecular Devices). Data were collected at 10 kHz and low-pass filtered at 2 kHz. Recordings were obtained using the whole-cell patch clamp technique in the current clamp mode. Intrinsic excitability and other electrophysiological properties were performed in the absence of synaptic blockers in ACSF, during bath application of ethanol for 5–10 min and after washout for another 10 minutes. RMP was measured following equilibration with intracellular pipette solution, 2 to 4 min after break-in, with no current applied (I = 0). Input Resistance (R_in) was measured by Ohm’s law from the steady-state voltage responses to 500 ms current injections of 10 pA increments (6 steps starting at −30 pA). The “sag ratio” was measured using a 500 ms, −200 pA current command and it was expressed as a percentage of the ratio between the difference of the steady-state voltage and the peak voltage and the difference between the baseline and the peak voltage (Salling et al., 2018). Medium afterhyperpolarization (mAHP) and afterdepolarization (ADP) were measured following a previous 500 ms current command. mAHP was calculated as the change in membrane potential between the holding potential (−70 mV) and the hyperpolarize peak after a current step that induced between 6–10 action potentials (APs) in PNs and 12–18 APs in SST INs. ADP was measured as the peak of depolarization just after a hyperpolarizing current step of −200 pA. Input-output curves were constructed to assess possible effects of ethanol on the intrinsic excitability of SST INs and PNs. AP firing was induced by a series of 500 ms current injections (0 to 200 pA, 20 pA increments) and rheobase was defined as the minimum amount of current required to elicit an action potential. These protocols were performed from a standardized potential of −70 mV. Series resistance was compensated using the bridge balance, and cells in which R_s > 20 MΩ were omitted from data analysis. SST INs with R_in < 150 MΩ and rheobase > 150 pA (4 of 26 total) were not included in the analysis to avoid contamination of our sample by a few fast-spiking PV INs found to be labeled in SST-Cre lines (Hu et al., 2013; Lee et al., 2010) as well as fast-spiking non-Martinotti cells (Yavorska and Wehr, 2016). A similar number of females and males were used for ex vivo electrophysiological experiments of PNs (12 females, 8 males) and SSTs (10 females, 12 males). Neuronal firing (number of spikes in response to a 200 pA pulse) was also comparable in PNs of males (8.6 ± 2.8 spikes) and females (8.7 ± 2.1 spikes), and in SST INs of males (17.1 ± 3.4 spikes) and females (16.5 ± 1.4 spikes). Off-line analysis of electrophysiological data was performed in Clampfit 10.7 software (Molecular Devices).

2.6. Statistics

All statistical analyses were performed using Prism 8.0 (GraphPad Software, Inc). Summary data were presented as mean ± standard error of the mean (s.e.m.) or mean ± s.d. Calcium imaging data were analyzed with the Friedman test followed by Dunn’s post hoc multiple comparison test. Normalized calcium imaging data were analyzed with two-way ANOVA. For electrophysiological recording data, input-output curves were analyzed using a (mixed-model) two-way repeated-measures ANOVA. RMP, sag ratio, input resistance, mAHP, ADP and rheobase were analyzed using (mixed-model) one-way ANOVA. Significance levels were defined at P < 0.05. Additional detailed statistical information is provided in the figure legends.

3. Results

3.1. Low dose alcohol does not alter the activity of PNs in the PFC.

To study the effects of acute alcohol on neuronal circuitry in the PFC, we performed in vivo two-photon Ca²⁺ imaging in L2/3 of the PL subregion of the PFC in awake, head-restrained mice (Yang et al., 2013), and patch clamp recordings in acute PFC slices (Fig. 1A). For Ca²⁺ imaging, mice were administered alcohol (i.p.) from very low (0.5 g/kg), low (1 g/kg), to moderate (2 g/kg) and high (3 g/kg) doses. As shown in Fig. 1B, the blood alcohol concentrations (BACs) increased rapidly, peaked within 10 min and slowly returned to the baseline 1 to 2 h after i.p. injection of low doses (0.5 or 1 g/kg) of alcohol (corresponding to an interval below and just above legal intoxication levels). The BACs achieved after i.p. injection of 2 or 3 g/kg alcohol were in the range of concentrations from 200–300 mg/dl (corresponding to five to seven standard drinks in less than 2 h) and remained in this range for more than 2 h. The BACs reached at 10 min in vivo were used to calculate the low (20 mM) and medium-high (50 mM) concentrations used for patch clamp experiments in acute PFC slices.

First, we used transgenic mice expressing the genetically encoded calcium indicator GCaMP6s in cortical PNs to measure neuronal activity in the PFC after acute exposure to low doses of alcohol. Through a glass craniotomy window, GCaMP6s-expressing PNs in the PL were visible at the depth of ~400 μm below the pial surface. PN somas in L2/3 were identified based on their distance to the midline (~120–300 μm to midline) (Fig. 1A) and GCaMP6s allows the detection of spontaneous Ca²⁺ transients in PN somas (Fig. 1C).

When mice were in a quiet resting condition, the integrated Ca²⁺ activity averaged over 2 min was 18.63 ± 0.21% (1156 cells from 18 mice) in L2/3 PNs before the injection. We found that injection of saline or a very low dose (0.5 g/kg) of alcohol had no effect on the Ca²⁺ activity in PNs, although the frequency of Ca²⁺ transients was slightly reduced after low dose alcohol (Fig. 1C-E, G-J, Fig. S1A, B). We next performed Ca²⁺ imaging before and after intraperitoneal injection of a higher dose (1 g/kg, i.p.) of alcohol and we found that PN activity was slightly decreased at 10 to 30 min after 1 g/kg alcohol injection (Baseline: 18.38 ± 0.37%, 10 min: 17.78 ± 0.57%, P = 0.0007, 30 min: 17.80 ± 0.60%, P = 0.0007) and returned to the pre-injection baseline level 60 min after alcohol injection (Fig. 1F, Fig. S1C). Concomitantly, the frequency of Ca²⁺ transients (transients/min) at 30 min after injection (Saline: 4.24 ± 0.14, 1 g/kg: 3.31 ± 0.14, P < 0.0001) was significantly lower in alcohol-injected than in saline injected mice (Fig. 1G-J).

To determine how individual PNs in the PFC respond to alcohol, we calculated the proportion of neurons that showed increases, decreases or no change in Ca²⁺ activity, relative to baseline (Fig. 1D-F, bottom panels). We found that administration of 0.5 g/kg alcohol had no obvious effects on the activity of individual PNs in L2/3 of the PL. At 30 min after 0.5 g/kg alcohol injection ~88% PNs showed no change in Ca²⁺ activity (Fig. 1E), and these results were similar to Ca²⁺ activity measurements made in mice within 2 h following saline injection (Fig. 1D). The percentage of PNs showing a decrease in activity after injection of 1 g/kg alcohol was also similar to that seen after saline, suggesting there is little to no effect on PNs at low concentrations of ethanol (Fig. 1F).

To evaluate whether any changes in Ca²⁺ activity seen in vivo were due to changes in the excitability of PNs, we performed ex vivo electrophysiological recordings on PNs in L2/3 of the PL cortex using patch clamp recordings from acute brain slices (Fig. 1A). We evaluated PNs in L2/3 due to the strong inhibitory input from SST INs onto PNs in these layers. PNs were identified by their larger size and triangular cell body which was confirmed by filling with biocytin in a subset of PNs (Fig. 1K). Supra-threshold and sub-threshold excitability features were measured in PNs before, during and after acute exposure to ethanol (Table S1). To study the effects of alcohol on the intrinsic properties and firing of PNs, current injections from 0 to 200 pA at a holding potential of −70 mV were used to elicit action potentials (Fig. 1L, M).

Under current-clamp mode, neither 20 mM ethanol (equivalent to an in vivo low dose between 0.5–1 g/kg at 10 min post-injection) nor washout altered neuronal firing (mixed-model two-way ANOVA, F_{(2, 18)} treatment = 0.4478, P = 0.6460) (Fig. 1L, M). In response to hyperpolarizing current steps, PNs display voltage sag (Fig. 1L) that is partly mediated by the hyperpolarization-activated cation current (I_h) and is altered by chronic alcohol (Salling et al., 2018). While we did not observe changes in sag or other electrophysiological features such as ADP, R_in and RMP, we observed a small decrease in mAHP during and after exposure to 20 mM ethanol (Table S1).

3.2. Moderate to high alcohol (2–3 g/kg) increases PN activity in vivo.

To investigate the dose-dependent effects of alcohol on PN activity, we then injected mice with moderate to high doses of alcohol (2 or 3 g/kg) (Fig. 2A). We found that the average Ca²⁺ activity was increased in L2/3 PNs after 2 g/kg alcohol injection, and this neuronal hyperactivity occurred within 10 min after injection and lasted for at least 60 min, with a peak at 30 min after injection (Baseline: 18.57 ± 0.40% vs 30 min: 24.16 ± 1.05%, P < 0.0001) (Fig. 2B; Fig. S1D). The amplitude and frequency of Ca²⁺ transients at 30 min after alcohol injection were higher than those in saline injected mice (Amplitude: 58.92 ± 1.88% vs 48.81 ± 1.08%, P < 0.0001; Frequency: 4.79 ± 0.13 vs 4.24 ± 0.14, P = 0.0223; 2 g/kg and saline respectively) (Fig. 2D-G). As compared with mice that received a saline injection, a larger fraction (~23%) of PNs increased neuronal activity within 10 to 60 min after 2 g/kg alcohol application.

Similar to moderate-dose alcohol, a high dose of alcohol (3 g/kg) increased the frequency of Ca²⁺ transients in L2/3 PNs (6.15 ± 0.27 vs 4.24 ± 0.14, P < 0.0001), causing an increase in the neuronal Ca²⁺ response (Baseline: 18.95 ± 0.48%, 30 min: 23.13 ± 0.74%, P < 0.0001, 60 min: 24.33 ± 0.88%, P < 0.0001). This elevation of neuronal activity lasted for at least 120 min (Fig. 2C-G, Fig. S2E). A larger fraction (~32%) of PNs showed consistently increased Ca²⁺ activity within 2 h after 3 g/kg alcohol exposure, as compared with the effects of saline injection. A small fraction of PNs showing a decrease of Ca²⁺ activity (~10%) may suggest a sub-population of PNs that is directly or indirectly inhibited by ethanol.

We then investigated the effect in brain slice recordings of 50 mM ethanol, equivalent to an in vivo moderate-high dose (2–3 g/kg) 10 min after injection. This concentration did not affect the neuronal firing (Fig. 2H for representative recording traces) compared to ACSF baseline (mixed model two-way ANOVA, treatment F_{(2, 18)} treatment = 1.448, P = 0.2703) (Fig. 2I). Similar to 20 mM ethanol, a higher concentration of ethanol (50 mM) also induced a small decrease in mAHP (ACSF: 3.3 ± 1.2, 50 mM ethanol: 2.4 ± 1.2 mV) but did not alter other electrophysiological characteristics of PNs in vitro (Table S2). Taken together, these results show that moderate to high doses of alcohol (2–3 g/kg) increased PN activity in L2/3 of the PFC in vivo, but not ex vivo.

3.3. Low dose alcohol increases SST IN activity in vivo.

To further understand the effects of alcohol on the neuronal circuitry in the PFC, we examined the activity of local inhibitory cells. Recent studies have shown that SST INs play an important role in regulating the dendritic and somatic activity of PNs (Chiu et al., 2013; Cichon and Gan, 2015; Marlin and Carter, 2014; Muñoz et al., 2017). To measure the effect of alcohol on SST cell responses, we used in vivo Ca²⁺ imaging to examine the activity of GCaMP6s-expressing SST cells before and after alcohol injection (Fig. 3A-H) and performed whole-cell recordings from L2/3 SST INs from slices obtained from SST-Cre/Ai9-TdTomato mouse lines (Fig. 3I-K).

In Ca²⁺ imaging experiments, SST-IRES-Cre mice were injected with a Cre-dependent adeno-associated virus (AAV) to induce the expression of GCaMP6s specifically in SST cells. As expected, saline injection had no effect on the average activity of SST INs in L2/3 of the PL (Fig. 3B, Fig. S2A). When analyzing the response of individual cells to saline injection, the activity changes in the vast majority (~85%) of SST INs remained within 3 SD of their pre-injection baseline for 2 h following the saline injection (Fig. 3B, bottom panel).

We first administered 0.5 g/kg alcohol to examine how SST INs responded to a very low dose of alcohol. In contrast to the lack of response of PNs to low dose alcohol, the average activity of SST INs started to increase at 30 min, and was further increased at 120 min after 0.5 g/kg alcohol injection (L2/3: Baseline: 28.22 ± 2.58%, 120 min: 37.57 ± 3.10%, P = 0.0002) (Fig. 3C, Fig. S2B). Comparing the alcohol and saline groups, the amplitude of Ca²⁺ transients was significantly higher at 120 min after alcohol (80.39 ± 5.37% vs 58.90 ± 1.48%, P = 0.0089) (Fig. 3E-H). When analyzing the response of individual cells, ~30% of L2/3 SST INs showed increased activity within 120 min after injection of 1g/kg ethanol compared (Fig. 3C, bottom panel) with 6.4% after the saline injection (Fig. 3B, bottom panel). Following 1 g/kg alcohol injection, the average Ca²⁺ activity in SST INs was constantly increased until 120 min, although we observed an initial marginal decrease at 10 min (Baseline: 28.44 ± 1.04%, 10 min: 25.74 ± 1.15%, P = 0.0049, 120 min: 34.89 ± 1.84%, P = 0.0040) (Fig. 3D, Fig. S2C). Both the amplitude and frequency of Ca²⁺ transients were increased at 120 min after alcohol injection (Amplitude: 68.10 ± 2.49% vs 58.90 ± 1.48%, P = 0.0046; Frequency: 4.03 ± 0.19 vs 2.50 ± 0.17, P < 0.0001; Duration: 3.80 ± 0.48 vs 4.45 ± 0.52, P > 0.9999; 1 g/kg and saline respectively; Fig. 3E-F). Consistently, when comparing low-dose alcohol (1 g/kg) to saline injection, there was an initial decrease in Ca²⁺ activity at 10 min (1 g/kg: 10.28% with 1 g/kg vs 4.49% with saline), which was followed by a delayed increase in a small proportion (~20%) of SST INs at 120 min (Fig. 3D).

For patch clamp experiments, we generated mice expressing tdTomato in SST-IRES-Cre mice to target SST INs in L2/3 of the PFC (Fig. 3I). Whole-cell patch clamp recordings were performed to interrogate the intrinsic excitability of SST INs during and after acute exposure to a low concentration (20 mM) of alcohol. Under current-clamp, SST INs showed excitability properties (Fig. 3J) that were distinct from PNs (Fig. 1L, Fig. 2H), and the higher frequency of evoked firing also corresponds with a higher Ca²⁺ baseline activity in vivo in SST INs (Fig. 3B). We observed that values of RMP, input resistance and other intrinsic properties (summarized in Table S3) are similar to the values from Martinotti cells reported by others (Ma et al., 2006). Acute application of 20 mM ethanol did not alter the number of evoked action potentials (mixed model two-way ANOVA, F_{(2, 20)} treatment = 0.2483 P = 0.7825) (Fig. 3K) and the rheobase was also unaltered (Table S3). Bath application of 20 mM ethanol induced a small reduction in sag ratio that was not reversed by washout (ACSF: 5.4 ± 4.4%, 50 mM ethanol: 4.2 ± 4.6%, washout: 2.5 ± 2.6%) (Table S3).

3.4. Moderate to high alcohol decreases SST IN activity in vivo.

Following moderate (2 g/kg) and high (3 g/kg) dose alcohol exposure, we found that SST IN activity was persistently decreased for 30 min (Fig. 4A). In L2/3 of the PL PFC, 2 g/kg alcohol decreased the activity in SST INs within 10 min after injection and the effect lasted for at least 30 min (Baseline: 27.97 ± 1.06%, 10 min: 20.33 ± 1.19%, P < 0.0001, 30 min: 24.73 ± 1.71%, P < 0.0001) (Fig. 4B, Fig. S2D). At 30 min, both the amplitude and frequency of Ca²⁺ transients were decreased (Amplitude: 54.37 ± 2.46% vs 61.87 ± 1.35%, P < 0.0001; Frequency: 1.72 ± 0.14 vs 2.77 ± 0.17, P < 0.0001; 5.64 ± 1.02 vs 3.66 ± 0.33, P = 0.0003; 2 g/kg and saline respectively; Fig. 4D-F), though the duration was slightly increased (Fig. 4G). Furthermore, 2 g/kg alcohol substantially increased the number of SST cells with decreases in Ca²⁺ levels (~25% at 10 min) (Fig. 4B, bottom panel). Administration of 3 g/kg alcohol further decreased the population average activity of SST INs at short times (Baseline: 28.96 ± 1.30%, 10 min: 15.33 ± 1.22%, P < 0.0001, 30 min: 17.60 ± 1.11%, P < 0.0001) (Fig. 4C, Fig. S2E). The amplitude, frequency, and duration of Ca²⁺ transients were all decreased (Amplitude: 39.20 ± 1.43% vs 61.87 ± 1.35%, P < 0.0001; Frequency: 1.03 ± 0.10 vs 2.77 ± 0.17, P < 0.0001; Duration: 2.45 ± 0.62 vs 3.66 ± 0.33, P < 0.0001; 3 g/kg and saline respectively; Fig. 4D-G). Moreover, the fraction of SST INs that showed decreased Ca²⁺ levels were highest (~35%) at 10 min after 3 g/kg alcohol injection and then decreased gradually over the next 2 h (Fig. 4C, bottom panel).

In contrast, application of 50 mM ethanol to acute brain slices (corresponding to blood levels reached after a medium-high dose of ethanol) did not reduce the excitability of SST INs (two-way ANOVA, F_{(2, 20)} treatment = 0.098, P = 0.9064) (Fig. 4H, I). Analysis of other electrophysiological parameters revealed no effects of alcohol (Table S4).

3.5. Increased PN activity in vivo correlates temporally and inversely with the inhibition of SST INs at moderate to high doses of alcohol.

To investigate the temporal relationship between the changes in Ca²⁺ activity in PNs and SST INs in vivo after acute alcohol exposure, we calculated the ratio of change in Ca²⁺ activity over time as normalized to the baseline activity (Fig. 5A-D). We also compared the proportions of PNs and SST INs more responsive to ethanol (changes exceeding 3 s.d. of the mean baseline activity) at each concentration (Fig. 5E-F). We found that the increase in normalized Ca²⁺ activity in SST INs at low doses of alcohol (0.5 and 1 g/kg) is neither reflected in a change in activity of PNs (Fig. 5A-D) nor in an increase in the fraction of PNs inhibited by alcohol within 2 h after injection (Fig. 5E-F). In contrast, moderate and high doses of alcohol (2 and 3 g/kg) induced an inverse time-dependent response in the normalized Ca²⁺ activity of PNs and SST INs (2 g/kg EtOH at 10 min: PN, 1.20 ± 0.04 vs SST, 0.73 ± 0.04, P <0.0001, 3 g/kg EtOH at 30 min: PN, 1.22 ± 0.04 vs SST, 0.61 ± 0.04, P < 0.0001) (Fig. 5C-D) and a simultaneous increase in the fraction of PNs activated and SST INs inhibited by alcohol (Fig. 5G-H). High-dose alcohol induced a rapid inhibition in a subset of SST INs (10 min) that dropped over time, while the peak activation of PNs occurred later (between 30 and 60 min) and a fraction of PNs remained active 2 h after 3 g/kg alcohol injection (Fig. 5G-H). Altogether, these results show that inhibition of SST INs by moderate to high dose alcohol was temporally associated with an increase in the activity of PNs.

Fig. 5. Dose- and time-dependent effects of moderate to high doses of alcohol on SST IN and PN activity in the PL in vivo.

(A-D) Normalized Ca²⁺ activity of L2/3 PNs (open circles) and SST INs (squares) after 0.5 g/kg (A), 1 g/kg (B), 2 g/kg (C) and 3 g/kg (D) alcohol injection. Dotted line indicates normalized Ca²⁺ baseline. (A) 0.5 g/kg EtOH (120 min: PN, 0.97 ± 0.02, SST, 1.3 ± 0.11, P = 0.0142). (B) 1 g/kg EtOH (60 min: PN, 0.99 ± 0.02, SST, 1.16 ± 0.05, P = 0.0398, 120 min: PN, 1.05 ± 0.02, SST, 1.23 ± 0.06, P = 0.0268). (C) 2 g/kg EtOH (10 min: PN, 1.20 ± 0.04, SST, 0.73 ± 0.04, P < 0.0001, 30 min: PN, 1.30 ± 0.06, SST, 0.88 ± 0.06, P < 0.0001, 60 min: PN, 1.25 ± 0.05, SST, 1.03 ± 0.05, P = 0.0390). (D) 3 g/kg EtOH (10 min: PN, 1.03 ± 0.03, SST, 0.53 ± 0.04, P <0.0001, 30 min: PN, 1.22 ± 0.04, SST, 0.61 ± 0.04, P < 0.0001, 60 min: PN, 1.28 ± 0.05, SST, 0.85 ± 0.07, P < 0.0001). Data are presented as means ± s.e.m. *P < 0.05, ****P < 0.0001, PN vs SST by two-way ANOVA and Sidak’s post hoc multiple comparison test. (E-F). Proportions of PNs (circles) and SST INs (squares) that show decreases (in blue) or increases (in red) in Ca²⁺ activity after 0.5 g/kg (E), 1 g/kg (F), 2 g/kg (G) and 3 g/kg (H) alcohol injection. Dotted line indicates mean proportion (6.5%) of neurons showing increases and decreases in Ca²⁺ activity after saline injection. Responsive neurons (%) as the proportion of neurons that show increases or decreases in Ca²⁺ activity 3 s.d. above or below the mean baseline activity.

4. Discussion

The PFC is implicated in both short and long-term effects of alcohol, and is of special interest in terms of the behavioral effects associated with the moderate alcohol levels commonly achieved during social drinking. Both PL and infralimbic cortex are involved in fear and anxiety (Dejean et al., 2016), reward and drug-seeking behavior (Moorman and Aston-Jones, 2015; Shen et al., 2014), including for alcohol (Pfarr et al., 2015). More recently, strong physiological evidence has been presented for an interaction between the PFC and the amygdala, and specific neuronal ensembles have been identified that are engaged during alcohol-seeking, compulsive drinking and following alcohol withdrawal (Avegno et al., 2018; George and Hope, 2017; Siciliano et al., 2019).

Using in vivo calcium imaging and ex vivo electrophysiological recording, we examined the effects of acute alcohol exposure on the PFC circuitry of the mouse. We found that alcohol dose-dependently altered network activity in the PL region in vivo. Low doses of alcohol (0.5 and 1 g/kg i.p.) caused moderate activation of SST INs and weak inhibition of PNs, whereas moderate to high doses (2–3 g/kg) strongly inhibited the activity of SST INs in vivo. This effect may result in disinhibition, as the activity of a subpopulation of PNs was enhanced at the same time. By contrast, recordings in brain slices revealed no direct effects of alcohol on the excitability of either SST INs or PNs. These results suggest that these effects of alcohol on PNs and SST INs in vivo must involve the participation of other INs within the PFC, or occur via neuromodulation of SST INs by extrinsic inputs, resulting in a net disinhibitory effect in the PFC.

Brain imaging studies in humans have not yet led to a consensus regarding the changes in PFC activity during acute alcohol intake. Early brain imaging studies in alcoholics suggested there are changes in blood flow in the human dorsolateral PFC during drinking, that were interpreted by some as resulting from lower brain activity (Volkow et al., 1990, 1988), but more contemporary studies suggest that activity in the PFC may actually be increased by alcohol (Dudek et al., 2015; Rickenbacher et al., 2011; Strang et al., 2015; Tiihonen et al., 1994). The use of new brain imaging techniques such as Ca²⁺ imaging in animal models (Siciliano and Tye, 2019) has the potential to provide insights into the effects of acute ethanol at the network and cellular level.

While the use of in vivo imaging of genetically encoded calcium indicators (GECIs) is now widely adopted in neuroscience, GECIs have only recently been successfully applied to alcohol (Siciliano et al., 2019) and other animal models of addiction (Siciliano and Tye, 2019). Using the GECI GCaMP6s and two-photon Ca²⁺ imaging allowed us to monitor the acute effects of acute ethanol on a wide area of the PL PFC in vivo as well as to track the cellular responses of PNs and SST INs over a period of 2 h in habituated, head-restrained animals. Although it is worthwhile to mention that longer habituation might be needed to reduce stress levels during behavior in head-fixed animals (Juczewski et al., 2020), the confounding effect of stress on calcium dynamics in our study seems low since we observed little change in neuronal activity over 2 h after saline injection in both PNs and SST INs. Conversely, when mice were exposed to acute alcohol we found diverging effects on different types of neurons in the PFC and these effects were dose- and time-dependent: low-dose alcohol preferentially increases SST IN activity but decreases PN activity, whereas high-dose alcohol decreases SST IN activity and increases PN activity.

Results from previous studies in PNs of the frontal cortex in vitro show a reduction of NMDA-receptor mediated EPSCs at high concentrations of ethanol (Badanich et al., 2013; Weitlauf and Woodward, 2008) and a decrease in the persistent activity of PFC in cultured slices, but no effect on evoked action potentials (Tu et al., 2007). Although chronic alcohol can increase excitability in PNs of mouse PFC (Pleil et al., 2015; Salling et al., 2018), there is sparse evidence for effects of acute alcohol on excitability (Woodward and Pava, 2009), and the ex vivo results presented here did not reveal any changes in evoked firing of PNs during exposure to acute ethanol, except for a small decrease in mAHP, which has also been reported after alcohol chronic exposure (Nimitvilai et al., 2016; Rau et al., 2015). In vivo electrophysiological results have shown a dose-dependent (0.375–3.5 g/kg ethanol) decrease in the firing rate of PFC neurons after acute alcohol (Tu et al., 2007). In a most recent study, decreases in firing rate in the PFC after medium-high doses (1–2 g/kg ethanol) of ethanol were also found in anesthetized rats but only when the rats were in the resting state (Morningstar et al., 2020). It is possible that the use of anesthetics such as urethane (Hara and Harris, 2002) may confound results from in vivo anesthetized rat recordings. In addition, changes in IN sub-populations are difficult to capture in extracellular electrophysiological recordings in vivo after exposure to ethanol (Morningstar et al., 2020). Here, using in vivo calcium imaging, we were able to identify an increase in the average Ca²⁺ activity in L2/3 PNs in vivo following administration of high-dose (2–3 g/kg) alcohol that correlates temporally and inversely with the inhibition of SST INs seen at the same doses. The frequency and amplitude of the calcium transients, which are correlated with spiking activity (Chen et al., 2013), showed similar trends as the changes of average Ca²⁺ activity in PNs and SST INs. Although it has been recently reported that ethanol may decrease firing variability in the PFC (Morningstar et al., 2020), the frequency of calcium transients were widely increased in PNs and greatly reduced in SST INs short after exposure to high dose of ethanol. Altogether, our results suggest that SST INs exert a direct or indirect disinhibitory effect on PNs during acute alcohol exposure in vivo.

Among the three major groups of GABAergic neurons, SST INs represent around 30% of the INs and mainly innervate distal dendrites of PNs (Rudy et al., 2011). SST INs in L2/3 are mainly Martinotti cells (Wang et al., 2004) and many studies within L2/3 have shown that SST cells provide lateral inhibition to PNs (Adesnik et al., 2012; Fino and Yuste, 2011; Kvitsiani et al., 2013; Xu et al., 2013). The sensitivity of these INs to alcohol has been overlooked until recently, when two recent studies have reported divergent changes in the excitability of Martinotti cells ex vivo after chronic ethanol exposure (Hughes et al., 2020; Joffe et al., 2020). We report a dose-specific and rapid effect of acute ethanol in vivo on the activity of SST INs that may therefore exert its effect via a direct effect on the activity of SST INs and their synaptic activity onto PNs (Fino and Yuste, 2011; Urban-Ciecko et al., 2015; Wang et al., 2004). However, we did not observe a strong inhibition of PNs following low doses of ethanol (0.5–1 g/kg), even when SST INs become more active, in terms of average Ca²⁺ activity as well as the frequency of transients. This might be due to a compensatory effect of excitatory inputs between PNs (Kwan and Dan, 2012) or it may reflect the role of SST INs in regulating the synaptic integration of PNs that may serve to prevent hyperexcitability of cortical circuits (Fino and Yuste, 2011). If we consider SST INs as local organizers of inhibitory activity in the PFC, elevated SST INs activity by acute alcohol exposure may only increase tonic attenuation on PNs but, by silencing SST INs, the loss of gain control over synaptic PNs inputs could lead to a delayed overexcitation of these neurons. However, intrinsic excitability and other intrinsic properties of SST INs were unaltered by exposure to low or high concentration of ethanol and this suggests that acute alcohol may act through the modulation of PFC IN populations other than those studied here, such as VIP or PV subtypes.

Some authors have suggested the key importance of PV in regulating the balance of excitation and inhibition in the PFC (Ferguson and Gao, 2018). PV INs have dense local connections with PNs (Packer and Yuste, 2011) and provide a fast and strong inhibition on PNs in the PFC (Kvitsiani et al., 2013) through synapses on the somas and the axon initial segments (Rudy et al., 2011). There is evidence that acute alcohol disrupts up-states in PN recordings, possibly by reducing the activity of fast-spiking INs (Woodward and Pava, 2009) and it has been recently reported that chronic alcohol alters the excitability of PV INs (Hughes et al., 2020; Joffe et al., 2020). Acute ethanol may reduce the activity of PV INs only at moderate or high doses, resulting in a decrease of inhibition to PNs. SST may also exert an indirect effect on PV INs, whereby an increase in membrane excitability of SST INs can indirectly excite local PNs through inhibition of fast-spiking INs (Cottam et al., 2013), as occurs following exposure to morphine (Jiang et al., 2019).

An additional possibility is that alcohol may have a disinhibitory effect on the PFC mediated by VIP INs that mainly target other INs (Lee et al., 2013; Pfeffer et al., 2013; Pi et al., 2013). The inhibition of SST INs by VIP INs located superficially is supported by electrophysiological (Kvitsiani et al., 2013; Pi et al., 2013) and connectivity studies (Ährlund-Richter et al., 2019) in the PFC. L1 VIP can modulate SST INs and they clearly can mediate PN disinhibition by targeting the dendrites of SST INs (Pfeffer et al., 2013).

It is important to consider that INs form specific circuit motifs depending on their layer location (Fishell and Kepecs, 2020; Urban-Ciecko and Barth, 2016). For instance, VIP INs target SST more strongly in L2/3 than in deeper layers (Pfeffer, 2014; Pfeffer et al., 2013) and so it is possible that SST INs in deeper layers may have differential sensitivity to acute ethanol. We could also hypothesize that the subset of SSTs INs (between 15 and 30%) that strongly respond to acute ethanol, activated at low doses and inhibited at high doses, might correspond to distinct subpopulations of SST INs (Gouwens et al., 2020; Ma et al., 2006; Nigro et al., 2018; Yavorska and Wehr, 2016), and the response to ethanol might reflect diverse effects within the microcircuitry of the PFC. To resolve these questions, it will be important to record from these additional IN subtypes in vitro and in vivo to acquire an integrated understanding of alcohol effects on the PFC.

An alternative to the intrinsic circuitry hypothesis is that ethanol achieves its effects in PFC via extrinsic inputs, by activation of neurons in one or more of the many subcortical regions that send afferents to the PFC, for example, the ventral tegmental area (Avegno et al., 2016; Mrejeru et al., 2015), the amygdala (Cummings and Clem, 2020) and the peri-aqueductal gray (Siciliano et al., 2019). It is known that binge-like alcohol exposure during adolescence in rats results in alterations of dopaminergic neurotransmission within the PL of the adult (Trantham-Davidson et al., 2017, 2014). It is known that D₁ receptors are enriched in PNs in L5/6 of the PFC and absent in PV and SST but present in superficial VIP INs (Anastasiades et al., 2019). These observations suggest at least the possibility that a dopaminergic input to the PFC might play a role in the actions of alcohol. Further pharmacological studies in vivo may help enhance our understanding of the effects of alcohol on activity in the complex circuitry of the PFC. The use of free behaving animals, miniaturized fluorescence microscopes and advanced computational approaches (Pnevmatikakis and Giovannucci, 2017; Zhou et al., 2018) can also be considered for future calcium imaging studies. This may reduce the potential effect of stress related to head-fixed imaging and allow the study of those behaviors that correlate with changes of IN activity after exposure to acute alcohol.

Our findings highlight that acute alcohol does have a complex and cell-specific effect on the neural activity of the PFC and not only a global decrease in neural activity as can be interpreted from previous brain imaging studies in humans (Goldstein and Volkow, 2011; Volkow et al., 1990) and in vivo electrophysiological studies in animals (Morningstar et al., 2020; Tu et al., 2007). These acute effects of ethanol on specific neuronal subpopulations may serve to target the projections that underlie the pathophysiological imbalance between the PFC and other brain areas characterized in alcohol addiction (Koob, 2014; Koob and Volkow, 2016).

Conclusions

The use of in vivo Ca²⁺ imaging for monitoring neuronal activity during acute alcohol exposure, together with classical electrophysiological techniques, as described here, enables rational investigation of the actions of alcohol in terms of the activity of specific neuronal subpopulations in the PFC. Our results show a differential effect between low and high doses of alcohol in the PFC circuitry and strongly suggest that SST INs are involved in a disinhibitory effect of alcohol in the PFC in vivo.

Highlights.

• In vivo calcium imaging enables study of alcohol in the prefrontal cortex (PFC).

• High dose alcohol increases the activity of pyramidal neurons (PNs).

• High dose alcohol decreases the activity of somatostatin (SST) interneurons (INs).

• Excitability of SST INs and PNs is unaltered by acute alcohol ex vivo.

• SST INs may mediate a disinhibitory effect of acute alcohol in the PFC.