ENHANCEMENT OF PARVALBUMIN INTERNEURON-MEDIATED NEUROTRANSMISSION IN THE RETROSPLENIAL CORTEX OF ADOLESCENT MICE FOLLOWING THIRD TRIMESTER-EQUIVALENT ETHANOL EXPOSURE

Prenatal ethanol exposure causes a variety of cognitive deficits that have a persistent impact on quality of life, some of which may be explained by ethanol-induced alterations in interneuron function. Studies from several laboratories, including our own, have demonstrated that a single binge-like ethanol exposure during the equivalent to the third trimester of human pregnancy leads to acute apoptosis and long-term loss of interneurons in the rodent retrosplenial cortex (RSC). The RSC is interconnected with the hippocampus, thalamus, and other neocortical regions and plays distinct roles in visuospatial processing and storage, as well as retrieval of hippocampal-dependent episodic memories. Here we used slice electrophysiology to characterize the acute effects of ethanol on GABAergic neurotransmission in the RSC of neonatal mice, as well as the long-term effects of neonatal ethanol exposure on parvalbumin-interneuron mediated neurotransmission in adolescent mice. Mice were exposed to ethanol using vapor inhalation chambers. In postnatal day (P) 7 mouse pups, ethanol unexpectedly failed to potentiate GABAA receptor-mediated synaptic transmission. Binge-like ethanol exposure of P7 mice expressing channel rhodopsin in parvalbumin-positive interneurons enhanced the peak amplitudes, asynchronous activity and total charge, while decreasing the rise-times of optically-evoked GABAA receptor-mediated inhibitory postsynaptic currents in adolescent animals. These effects could partially explain the learning and memory deficits that have been documented in adolescent and young adult mice exposed to ethanol during the third trimester-equivalent developmental period.


INTRODUCTION
Exposure to ethanol during fetal development causes a spectrum of deficits, including growth retardation, craniofacial anomalies, and CNS alterations. Ethanol affects multiple developmental processes in the fetal brain, leading to long-lasting neurobehavioral alterations that can have a negative impact on the quality of life.
Learning, memory, planning, judgement, attention, fine motor coordination, social interactions, sleep, and emotional control are among the myriad brain functions that can be disrupted by prenatal ethanol exposure. Studies indicate that dysfunction of Bird et al 13 demonstrated that ethanol vapor exposure during P2-9 (peak BEC = 221 mg/dl) reduces IN numbers in the adult mouse hippocampus; this study also found that a single vapor chamber exposure at P7 (peak BEC = 297 mg/dl) increases the number of INs that express activated caspase-3, suggesting that they are programmed to undergo apoptotic neurodegeneration. Ethanol administration to P7 mice (subcutaneous injection; peak BEC near 500 mg/dl) has been shown to reduce the numbers of PV-INs in the frontal cortex at P82 14 , as well as in the hippocampal formation (at P14 and P90-100) and the pyriform cortex (at P100) 15,16 . The same subcutaneous P7 ethanol administration paradigm has been demonstrated to decrease the numbers of PV and calretinin positive INs in the neocortex of adult mice 17 . It can be concluded from these results that INs are particularly sensitive targets of the effects of ethanol exposure during the third trimester-equivalent of human pregnancy.
Studies with humans and animals have demonstrated that fetal ethanol exposure causes learning and memory deficits and that these could be, in part, a consequence of alterations in hippocampal function (reviewed in 18,19 ). Comparatively, little is known about the contribution to these deficits of alterations in the function of other cortical regions that interconnect with the hippocampus. One of such regions is the retrosplenial cortex (RSC), which plays a central role in navigation, spatial learning, memory, and contextual fear conditioning 20,21 . Studies with mice have demonstrated that heavy binge-like exposure to ethanol at P7 triggers apoptosis of pyramidal neurons and PV-INs in the RSC, an effect that could, in part, underlie the deficits in contextual fear conditioning and navigation in the Morris water maze caused by this ethanol exposure paradigm [22][23][24][25][26][27] . Consistent with these studies, we recently reported that P7 ethanol vapor administration (peak BEC = 400 mg/dl) triggers apoptotic neurodegeneration of INs in the murine RSC 28 . In the same study, we also found that acute bath application of ethanol to brain slices from P6-8 mice decreased the amplitude of both NMDA and non-NMDA glutamatergic excitatory postsynaptic currents; however, only inhibition of NMDA receptors affected synaptic excitability in RSC neurons. These results support the hypothesis that inhibition of NMDA receptors mediates the apoptogenic effects of third-trimester ethanol exposure 22,[29][30][31] . Here, we tested whether this ethanol exposure paradigm causes either acute or long-lasting alterations in GABAA receptor-mediated neurotransmission in the RSC. Specifically, to determine the acute effects of ethanol on GABAA receptors, we first bath applied ethanol to tissue slices from P6-8 pups and measured the effect of ethanol on evoked GABAA receptor-mediated postsynaptic currents. Next, to determine long-term effects of ethanol, we used optogenetic techniques in acute brain slices to investigate its impact on GABAergic transmission at PV-IN-to-pyramidal neurons synapses in young-adult mice.  (Figure 1a).

Slice electrophysiology
Slice electrophysiology experiments were performed as described previously 28 .
Briefly, P6-P8 pups were anaesthetized with isoflurane (Piramal Critical Care, Bethelhem, PA) and rapidly decapitated. Brains were quickly removed and submerged for 3 min in a protective sucrose cutting solution. Ventral RSC-containing (equivalent to bregma -1.31 to -2.53 in an adult mouse 33 ) coronal slices (300 µm) were prepared using a vibrating slicer (VT 1000S, Leica Microsystems, Bannockburn, IL). Brain slices were then placed in a holding solution for 40 min at 32-34°C, followed by storage in a holding solution at room temperature (~24°C) for 30 min before recordings began. Recordings were acquired at 10 kHz and filtered at 2 kHz. Recordings were collected and analyzed For sPSC recordings, the last 5 minutes of the baseline, 90 mM ethanol application, and washout phases were analyzed using the Mini Analysis Program (Synaptosoft, Decatur, GA). For ePSC recordings, an input/output curve was initially measured and the stimulation intensity was subsequently adjusted so that the ePSC amplitude was 30-40% of the maximum amplitude. PSCs were evoked at 0.033 Hz.
Amplitudes were normalized to the average amplitude of the sPSCs from the entire

Immunohistochemistry
Tissue sections were stained with an anti-PV antibody to verify both the specificity and penetrance of ChR2-H134R/tdTomato expression. Immunohistochemistry (IHC) experiments were performed as described previously 13 .
Brains were extracted and stored in 4% PFA for 48 h, before being cryoprotected for 48 h in 30% sucrose (w/v in PBS). Brains were frozen at -80°C until sectioning.

Optogenetic slice electrophysiology
Electrophysiology experiments were performed using the same methods as described in Experiment 1 with the modifications detailed below. Slices were prepared from male and female B6 PV cre -Ai27D mice aged P40-P60 using the protective cutting/recovery methodology of Ting et al 45 . Mice were deeply anaesthetized with ketamine (250 mg/kg intraperitoneally) and transcardially perfused with 25 mL of a protective N-methyl-D-glucamine (NMDG)-containing aCSF at 4°C composed of (in mM): 92 NMDG, 2.5 KCl, 1.25 NaH2PO4, 30 NaHCO3, 20 HEPES, 25 glucose, 2 thiourea, 3 sodium pyruvate, 5 ascorbic acid, 10 MgSO4, and 0.5 CaCl2 saturated with 95% O2/5% CO2 (pH 7.3-7.4 with HCl; 300-310 mOsm). Brains were rapidly removed and immersed for 1 min in the same NMDG-containing aCSF. Coronal brain slices (300 µM) were prepared using a vibrating slicer (described above under Experiment 1) in NMDG aCSF at 4°C. Once all slices containing the ventral RSC (bregma -1.31 to -2.53 33 ) were prepared, they were transferred to warm NMDG aCSF holding solution at 32-34°C. Over the course of 25 min, the NaCl concentration of the warm holding solution was gradually increased to 52 mM by adding increasing amounts of NMDG-aCSF containing 2M NaCl (250 µl at 0 min, 250 µl at 5 min, 500 µl at 10 min, 1 ml at 15 min, and 2 ml at 20 min). Slices were then allowed to recover in holding solution at room temperature for 1 h (as described for Experiment 1 above).
Following recovery, slices were transferred to the recording chamber and aCSF containing 3 mM kynurenic acid was applied at a rate of 2 ml/min. Recording electrodes were filled with the same Cs-methanesulfonate internal solution used in Experiment 1.
Optically-evoked inhibitory PSCs (oIPSCs) were generated in layer V pyramidal neurons using a 473 nm laser (IKE-473-100-OP) connected to a power supply (IKE PS-300) (IkeCool Corporation, Anaheim, CA) ( Figure 1b). Laser light was delivered through the 40X objective lens using an IS-OGP optogenetics laser positioner (Siskyou, Grants Pass, OR). Laser output power was 20 mW. Cellular capacitance and membrane resistance were measured immediately before the oIPSC recordings began. Three oIPSCs at 20-s intervals were generated at each of the following laser pulse durations: 0.5 ms, 1 ms, 2 ms, 4 ms, and 8 ms. The oIPSC was then blocked with gabazine (25 μM) to confirm that the oIPSC was a GABA mediated current. Data from any cell with an oIPSC that was not blocked with gabazine were discarded. The average oIPSC at each laser pulse duration for every cell was analyzed for peak amplitude, current density, GABAergic total charge (area under the curve, pA X ms), half-width at half-maximal amplitude, and rise time using Clampfit (Molecular Devices). In addition, the number of individual oIPSC peaks for each evoked current was counted for 100 ms after the onset of the laser stimulation, and the results were averaged together within each laser pulse duration for each cell. To measure oIPSC paired pulse ratios (PPRs), we evoked 10 pairs of oIPSCs at 30 s intervals, with 50 ms between paired laser pulses and 1 ms laser pulse durations. The ratio of the amplitude or the total charge of the second peak divided by the first peak (P2/P1) was measured. IHC data from Experiment 2 was analyzed using an unpaired t-test for parametric data or a Mann-Whitney U test for non-parametric data. Effect sizes for these tests are reported as Hedges' g or r, respectively. For optogenetic electrophysiology experiments from Experiment 2, because multiple cells were recorded from several animals from several litters, we used a linear mixed-model (LMM) approach to data analysis in SPSS  Table 2. Before LMM analyses were performed, outliers for each dependent variable were removed using a 1% ROUT test in GraphPad. The fixed factors were sex, P7 vapor chamber exposure condition, and laser pulse duration as a repeated measure. Models were built stepwise for each dependent variable according to the following procedure. First, a LMM was fit that includes the random effect associated with the intercept for each litter, using homogenous residual error variances between treatment groups, and an unstructured covariance structure for the repeated measures residuals. If the LMM failed to achieve convergence using an unstructured covariance structure for repeated measures residuals, a compound symmetry covariance structure was used instead. Then, a LMM was fit without the random effect included. The -2-log restricted maximum likelihood value for each model fit was then used to perform a likelihood ratio chi-square test. If the p-value for this test was < 0.05, this indicated that random effects significantly improved the model fit and were therefore included in subsequent models. The resulting LMM was then fit using heterogenous residual error variances for treatment groups, and another likelihood ratio chi-square test was performed. If the p-value for this test was < 0.05, including heterogenous residual error variances significantly improved the model and subsequent LMMs included heterogenous residual error variances. F-ratios (containing Satterthwaite approximated degrees of freedom) and p-values for Type III F-tests for treatment, sex, laser pulse duration, and interactions between fixed factors from the final LMM are reported in the results section for each dependent variable. Effect sizes for main effects of exposure and sex for LMMs are reported as Hedges' g and not partial eta squared because SPSS currently does not provide a sum of squares output for these tests.

Statistics
Some skewness and kurtosis in data distribution can be tolerated using LMMs as they are robust against violations of assumptions of normality, as normality of residuals does not affect parameter estimates in multilevel models 49 . However, because residuals from many of the LMMs violated assumptions of normality (Shapiro-Wilkes p-value of < 0.05), we also present non-parametric Mann-Whitney U tests for main effects of vapor chamber exposure condition and sex in Supplemental Table 2 for any LMM that did not pass the Shapiro-Wilkes test. Exposure by laser pulse duration interactions with a pvalue of < 0.05 were subsequently analyzed with non-parametric Mann-Whitney U tests examining the effect of vapor chamber exposure condition within each laser pulse duration. P-values reported for these tests are Bonferroni corrected. Detailed statistics for post hoc tests examining exposure effects within each laser pulse duration appear in Supplemental Table 3. All data presented are mean ± standard error of the mean (SEM).  d. In air exposed animals, mean transgene penetrance was 98.67 ± 1.33% (percentage of PV+ cells expressing ChR2-tdTomato), while in ethanol exposed animals transgene penetrance was 98.82 ± 1.18% (Figure 5e; air male n = 3 animals from 3 litters, air female n = 2 animals from 2 litters; ethanol male n = 3 animals from 3 litters, ethanol female n = 2 animals from 2 litters). Transgene penetrance was not different between vapor-chamber exposure conditions (Mann-Whitney U(n1 = n2 = 5) = 12, p > 0.99, r = 0.047). We also measured transgene specificity to see if ChR2-tdTomato was aberrantly expressed in PV-negative cells. In air exposed animals, non-specific transgene expression (% of ChR2-tdTomato positive cells that were not positive for PV) was 3.76 ± 1.58%, while in ethanol exposed animals non-specific transgene expression There was no effect of sex, or two-or three-way interactions between sex exposure, and laser pulse duration on oIPSC current density (p-values > 0.29). Random effect of litter significantly improved the LMM for oIPSC current density and was included in the final model.

RESULTS
The total charge of the oIPSCs was affected by vapor chamber exposure condition. Ethanol exposed animals had a larger oIPSC total charge than air-exposed animals (Figure 6e There was no effect of sex, or interactions between sex and exposure, sex and laser pulse duration, or sex by exposure by laser pulse duration interactions (p-values > 0. 23 Table 3). There was no effect of sex or other interactions between sex, vapor chamber exposure condition, and laser pulse duration for oIPSC rise time (p-values > 0.13). Random effect of litter did not significantly improve the LMM for oIPSC rise time and was not included in the final model.
When analyzing oIPSC data, we noticed an interesting phenomenon. As laser pulse durations increased during the course of an experiment, oIPSC currents contained an increasing number of individual event peaks (Supplemental Figure 4). This effect has been observed previously 52 , and suggests that increases in laser pulse duration promotes asynchronous GABA release following the synchronous release of this transmitter. To analyze this, we counted the number of individual oIPSC events for 100 ms after each laser stimulation and compared experimental groups using a LMM.
Ethanol exposure at P7 increased the number of oIPSC events compared to air-

DISCUSSION
The results of our study advance knowledge of the short-and long-term effects of developmental ethanol exposure on GABAA receptor-mediated transmission in the cerebral cortex. We demonstrate that acute ethanol exposure does not affect GABAA receptor function in the RSC of neonatal mice, suggesting that potentiation of these receptors is not involved in the mechanism responsible for the apoptotic neurodegeneration triggered in this brain region by binge-like ethanol exposure during the brain growth spurt. Moreover, we show that this ethanol exposure paradigm causes long-lasting functional alterations at PV-IN→pyramidal neuron synapses from adolescent mice. These findings identify a novel mechanism that could contribute to the cognitive alterations associated with fetal ethanol exposure.
In Experiment #1, we demonstrate that eGABAA-PSCs in layer V neurons of the mouse RSC are not affected by acute ethanol exposure during the third trimesterequivalent of human pregnancy. These results follow up on those of a recent study from our laboratory, in which we showed that acute exposure to ethanol during P6-P8 inhibits the excitability of RSC layer V pyramidal neurons and INs mainly through effects exerted post-synaptically on NMDA receptors 28 . Together, the outcomes of these two studies suggest that apoptosis caused by binge-like exposure to ethanol in the RSC during this critical developmental period is caused by inhibition of neuronal activity mediated by a decrease in the function of NMDA receptors, rather than potentiation of Clcurrent flow through GABAA receptors. To determine if this lack of an effect of ethanol on eGABAA-PSCs was specific to the RSC, we also measured currents in developing CA1 hippocampal pyramidal neurons. We found that acute 90 mM ethanol exposure did increase eGABAA-PSC amplitude but for a relatively short duration. A previous study with hippocampal slices from mice showed that protein kinase C (PKC) delta increases, whereas PKC epsilon decreases, sensitivity of IPSCs to acute ethanol (80 mM) exposure (including duration of ethanol's potentiation) 54 56 and 57 ). This also appears to be the case for developing neurons during the brain growth spurt, as indicated by electrophysiological studies with slices from neonatal rats showing that acute ethanol increases GABA release while having minimal effects on postsynaptic GABAA receptor function in layer II and III pyramidal neurons of the parietal cortex 58 , pyramidal neurons and INs of the CA3 hippocampal region 59,60 , and hypoglossal motoneurons in the brain stem 61 . In contrast to the findings of these studies, we found that ethanol does not affect the amplitude of electrically evoked GABAA-PSCs in RSC pyramidal neurons and INs. As a positive control, we measured the effect of flunitrazepam on eGABAA-PSCs and found that this agent prolongs the duration of these events. This finding indicates that the insensitivity of GABAA receptors to ethanol is not due to a general lack of sensitivity of the receptors to allosteric modulators. In addition, ethanol did not modulate the amplitude, decay, or frequency of spontaneous GABAA receptor-mediated PSCs in these cells. These It has been demonstrated that third-trimester equivalent ethanol exposure results in long-term loss of INs in the rodent brain 13,64 , including PV-INs in the RSC 17,65 . In Experiment #2, we demonstrate that binge-like ethanol exposure during this developmental period causes long-lasting alterations in the function of PV-IN→pyramidal neuron synapses. oIPSC amplitude and total charge were increased in ethanol-exposed animals, suggesting that postsynaptic GABAA receptor function is altered in RSC pyramidal neurons. Future studies will be needed to determine the underlying mechanism of this effect (e.g., increases in receptor expression or changes in phosphorylation state). An unexpected finding from this study was that ethanolexposed animals had an increased number of individual oIPSC events following a laser pulse when compared to air-exposed animals. This observation is similar to electrophysiology experiments in which currents are evoked in the presence of extracellular strontium instead of calcium, which leads to asynchronous presynaptic release of neurotransmitter 66 . In these studies, an increase in the frequency of postsynaptic events following stimulation is indicative an increase in the probability of presynaptic neurotransmitter release 67 . Similar to our study, Pulizzi et al. observed an increasing number of synaptic release events with increasing laser stimulus durations, which they attributed to increases in presynaptic neurotransmitter release probability 52 .
Based on these studies, the increased number of oIPSC events in ethanol-exposed mice observed here indicates that P7 ethanol exposure causes a long-term increase in presynaptic function. However, we did not observe a change in PPRs cause by P7 ethanol exposure, suggesting that ethanol exposure did not produce a global change in synchronous and asynchronous GABA release. Asynchronous transmitter release is mediated by different presynaptic proteins (e.g., the slow Ca 2+ binding sensors synaptotagmin 7 and Doc2) than synchronous release (e.g., the fast Ca 2+ binding sensor synaptotagmin 1) 53 . It is possible that ethanol exposure selectively increased the expression and/or function of these and other proteins that regulate asynchronous GABA release from PV INs 53 . Another interesting finding from this study concerns the reduction in oIPSC rise time in ethanol exposed animals. Alterations in rise-time can be indicative of changes in post-synaptic GABAA receptor subunit composition 68  Deficits in learning and memory processes are among the most common negative consequences of developmental ethanol exposure 19 . Enhancement of postsynaptic GABAergic neurotransmission in the RSC after third-trimester equivalent ethanol exposure could have several impacts on limbic system network activity that may, in part, underlie the learning and memory deficits that have been documented in mice subjected to binge-like ethanol exposure at P7 24,27 . The RSC has reciprocal connections with several brain areas critical to learning and memory functions, including the hippocampal formation, the entorhinal cortex, and anterior/lateral thalamic nuclei 20 .
Importantly, communication between the hippocampus and neocortex via sharp wave ripples involves a subiculum-RSC pathway 75 . Consequently, the RSC has a distinct role in spatial memory; it processes distal spatial cues and allows animals to orient themselves both spatially and directionally 76,77 . Lesioning or inactivating the RSC in rodents and macaques impairs performance on spatial memory tasks, while humans with damage to the RSC have problems with anterograde memory formation and navigation using spatial cues 20,78 . The contribution of PV-mediated GABAergic neurotransmission to RSC function and its role in spatial memory are poorly understood at present. However, recent work has demonstrated that PV-IN activity in the anterior cingulate cortex, to which the RSC is immediately posterior, is necessary for memory consolidation 79 . Given that the RSC and the anterior cingulate are adjacent and strongly interconnected 80 , it is reasonable to hypothesize that a disruption in PV-IN-mediated signaling in the RSC will likewise impact learning and memory processes. Future work can test this hypothesis by capitalizing on advances in chemogenetic strategies to determine if up-or down-regulation of PV-IN activity impacts RSC physiology and/or performance on spatial memory tasks 78        Average oIPSC current traces from air-exposed animals using 0.5 ms (black trace), 1 ms (red trace), 2 ms (green trace), 4 ms (purple trace), and 8 ms (orange trace) laser pulse durations. b) Average oIPSC current traces from ethanol-exposed animals using 0.5, 1, 2, 4, and 8 ms laser pulse durations. Scale bars = 20 ms, 200 pA. Blue arrows indicate onset of laser pulse. c-h) Collected peak amplitudes (c), current densities (d), GABAergic total charge (e), half widths at half-maximum amplitude (f), rise times (g), and # of oIPSC events (h) for all cells from air exposed (black circles) and ethanol exposed (red squares) animals presented for each laser pulse duration. Data are presented collapsed across sex due to a lack of sex effects from LMM analyses. Dagger ( † ) denotes a p-value of < 0.06, and asterisks (*,**,***,****) denote p-values of p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively. Please see Supplemental Tables 2   and 3   neurons at P40-60. a) Average oIPSC PPR current traces from air-exposed (black trace) and ethanol-exposed (red trace) animals, collapsed across sex. Amplitudes of traces from air-and ethanol-exposed animals are normalized to the amplitude of the first peak to illustrate PPRs. Scale bar = 40 ms. b) Collected amplitude PPRs for all cells from air-exposed (black circles) and ethanol-exposed (red squares) animals. c) Collected total charge PPRs for all cells from air-exposed and ethanol-exposed animals.
Asterisk (*) denotes a p-value of p < 0.05. Please see Supplemental Table 2     Air Ethanol Air-exposed   c Air-exposed Ethanol exposed