FMRP regulates an ethanol-dependent shift in GABABR function and expression with rapid antidepressant properties

Alcohol promotes lasting neuroadaptive changes that may provide relief from depressive symptoms, often referred to as the self-medication hypothesis. However, the molecular/synaptic pathways that are shared by alcohol and antidepressants are unknown. In the current study, acute exposure to ethanol produced lasting antidepressant and anxiolytic behaviours. To understand the functional basis of these behaviours, we examined a molecular pathway that is activated by rapid antidepressants. Ethanol, like rapid antidepressants, alters γ-aminobutyric acid type B receptor (GABABR) expression and signalling, to increase dendritic calcium. Furthermore, new GABABRs are synthesized in response to ethanol treatment, requiring fragile-X mental retardation protein (FMRP). Ethanol-dependent changes in GABABR expression, dendritic signalling, and antidepressant efficacy are absent in Fmr1-knockout (KO) mice. These findings indicate that FMRP is an important regulator of protein synthesis following alcohol exposure, providing a molecular basis for the antidepressant efficacy of acute ethanol exposure.

T he presence of major depression increases the risk of alcohol use disorders (AUDs) by B2-fold (and vice versa) 1 . The self-medication hypothesis suggests AUDs may develop when the initial antidepressant actions of alcohol are shifted to depressant allostatic states with chronic abuse 2 . The molecular mechanism underlying the initial antidepressant effects of alcohol is unknown.
A major advance in understanding and treating depression is the recognition that NMDA receptor (NMDAR) antagonists act as rapid and effective antidepressant drugs 3 . A single injection of an NMDAR antagonist or 'rapid antidepressant' is effective within 2 h and has sustained antidepressant efficacy for 2 weeks 4 . These long-lasting properties depend on the activity of mammalian target of rapamycin (mTOR) 5,6 , a serine/threonine kinase essential for messenger RNA translation 7 . Recently, we demonstrated that activation of mTOR-dependent protein synthesis by NMDAR antagonists requires a shift in GABA B R signalling from opening potassium channels to facilitating an increase in dendritic calcium 6,8 . Interestingly, both acute ethanol and rapid antidepressants block NMDARs 4,9 . In light of these data, we propose that ethanol has lasting antidepressant efficacy, shares the same downstream molecular signalling events as rapid antidepressants, and requires de novo protein synthesis ( Supplementary Fig. 1).
Studies suggest that antidepressant efficacy requires two phases-an induction phase and a sustained phase 10,11 . Notably, GABA B R-mediated, mTORC1-dependent protein synthesis is required for the long-lasting sustained phase of rapid antidepressants. Our previous work indicates that both new protein synthesis and an increase in protein stability are required for the GABA B R shift in function necessary to increase mTORC1 activity 8 . However, the mechanism that initiates such dynamic changes in protein expression by rapid antidepressants remains unclear.
FMRP is an RNA-binding protein that has been characterized as a repressor of mRNA translation. Some forms of synaptic activity trigger FMRP to release its targets, allowing them to be translated 12,13 . Moreover, degradation and new protein synthesis of FMRP creates a window for the translation of specific mRNAs, facilitating long-lasting changes in synaptic function 14,15 . Complete loss of FMRP results in Fragile-X syndrome (FXS), the single most common genetic cause of autism 16 . Moreover, reduced levels of FMRP, caused by a pre-mutation, lead to a higher incidence of tremors, ataxia, memory loss, and neuronal neuropathy in older men 17 . These findings argue that precise levels of FMRP protein and its target mRNAs are required for normal neuronal function.
Drugs of abuse promote profound changes in gene expression, mRNA translation rates and synaptic protein composition 18,19 . Some studies suggest that drugs and alcohol highjack the molecular mechanisms that underlie synaptic plasticity 20,21 . In agreement with this premise, FMRP has been implicated in cocaine addiction 22 . However, little is known about the mRNA targets and the signalling mechanisms involved. Here we describe a critical role for FMRP in mediating GABA B R synthesis and plasticity following acute ethanol exposure, a mechanism required for antidepressant efficacy.

Antidepressant and anxiolytic effects of ethanol on behaviour.
To determine if acute alcohol has antidepressant properties, as predicted by the self-medication hypothesis, we first assessed the efficacy of alcohol on antidepressant-and anxiolytic-like effects on behaviour. The forced swim test (FST) is a rodent behavioural test predictive of antidepressant activity in humans 23 . Rodents treated with a single injection of NMDAR antagonists or rapid antidepressants swim longer and thus have reduced immobility relative to controls. Notably, these positive effects on behaviour last long after the drug has been metabolized 5,6,8,24 . Therefore, we considered the possibility that ethanol, which blocks NMDARs 9 , could also act like an antidepressant at 24 h, well beyond the intoxication period 25 . To test this, C57BL/6 mice were injected with ethanol (2.5 g kg À 1 , intraperitoneal (i.p.)), a concentration that is achieved during self-administration in mice 26 . Twentyfour hours after injection, the immobility of ethanol-treated mice was reduced by B15% relative to controls (Fig. 1a), similar to our previous observation in mice that had been exposed to the rapid antidepressant Ro- 25-6981 (refs 6,8). These results demonstrate that acute ethanol elicits a lasting antidepressant effect on behaviour similar to that seen with rapid antidepressants 8 .
As another measure of antidepressant effect of ethanol on behaviour, we assessed the grooming behaviour of mice using the splash test after ethanol or saline administration. The splash test measures latency to groom and dedicated grooming time as indicators of self-care and motivational behaviour 27,28 . Lack of self-care is often observed in humans with depressive disorder 29 . We have previously shown that mice receiving a single i.p. injection of the rapid antidepressant Ro-25-6981 spend more time grooming compared with control mice 8 . We hypothesized that ethanol would produce similar effects on grooming behaviour. Indeed, ethanol-treated mice spent more time grooming and displayed shorter latency to initiate grooming relative to controls (Fig. 1b,c).
Ethanol is a well-known anxiolytic substance 30 . However, the anxiolytic effect of a single dose of ethanol 24 h after administration has not been determined. We subjected ethanoland saline-injected mice to the open field test to assess the influence of ethanol on anxiety-like behaviours after 24 h. Mice that spend more time in the centre of the open field are scored as having reduced anxiety-like behaviour relative to mice that remain close to the perimeter 31 . Indeed, mice that received a single dose of ethanol (2.5 g kg À 1 , i.p.) had reduced anxiety-like behaviour, spending B40% more time in the centre relative to controls (Fig. 1d). There was no significant difference in total distance travelled or average speed between the groups (Fig. 1e,f). These data suggest that the anxiolytic effects of ethanol last up to 24 h post injection.
Acute ethanol increases GABA B R2 and surface GABA B Rs. Both ethanol and rapid antidepressants block NMDARs in the hippocampus 4,9 . One of the first events triggered by NMDAR antagonism is increased dendritic GABA B R2 protein expression 8  To determine if acute ethanol exposure in vivo rapidly increases the levels of GABA B R1 and/or GABA B R2, hippocampal synaptoneurosomes were isolated from mice that had been injected with a single dose of ethanol (2.5 g kg À 1 , i.p.) or saline for western blot analysis. The hippocampi were collected within the initiation phase (30 min post injection), a phase where molecular changes facilitate increased downstream mTORC1 activity 11 . Consistent with rapid antidepressants, acute ethanol injection increased the protein expression of GABA B R2 by B37% in the hippocampus with no significant change in GABA B R1 (Fig. 2a-c; uncropped blots, Supplementary Fig. 7a).
To further identify the subcellular localization of ethanolinduced increase in GABA B R2, we examined GABA B R expression in cultured hippocampal neurons. GABA B R1 and R2 were immunostained and quantified in the dendrites. A concentration of 30 mM ethanol was chosen, as it has been shown to reduce NMDAR activity in hippocampal neurons and reflects that achieved in vivo following i.p. injection 9,33 . Acute ethanol exposure (30 mM, 2 h) increased the dendritic levels of GABA B R2 by B47%, but did not affect GABA B R1 levels ( Fig. 2d-g). We did not observe a difference in the diameter of the primary dendrites between vehicle-and ethanol-treated neurons, demonstrating that ethanol does not modify dendritic calibre ( Supplementary Fig. 2). These in vivo and in vitro findings establish a role for ethanol in increasing GABA B R2 protein expression.
Since GABA B R2 is required for the surface expression of the heteromultimeric receptor, we predicted that the ethanol-induced elevation in GABA B R2 levels would increase expression of receptors at the surface. We measured the surface expression of dendritic GABA B Rs using an antibody directed against the extracellular domain of GABA B R1 in unpermeabilized hippocampal neurons. The surface signal was normalized by the total dendritic GABA B R1 levels after permeabilization 6 . As predicted, surface expression of GABA B Rs in ethanol-treated neurons was significantly higher (B66% increase) relative to controls ( Fig. 2h,i). This ethanol effect was consistent with what we previously observed following rapid antidepressant treatment of cultured hippocampal neurons 6 . Collectively, these results suggest that ethanol promotes the surface expression of GABA B Rs, and this is likely achieved by increasing GABA B R2 protein levels.
FMRP regulates GABA B R1 and GABA B R2 expression. Next, we sought to identify the mechanism by which NMDAR antagonism increases GABA B R2 expression. GABA B R2 mRNA is present in the dendrites of hippocampal neurons 34 , suggesting that this mRNA may be locally regulated at the translational level. Thus, we examined RNA-binding factors that may regulate GABA B R2 mRNA expression in dendrites. Notably, both GABA B R1 and GABA B R2 mRNAs are reported targets of FMRP, an RNA-binding protein that stalls translational elongation of its targets 35,36 .
To test the hypothesis that FMRP regulates GABA B R mRNA translation, we first verified that (1) GABA B R mRNAs bind to FMRP, and that (2) the absence of FMRP in knockout mice results in aberrant expression of GABA B Rs. Using a specific antibody against FMRP, bound mRNAs were isolated using RNA immunoprecipitation (RIP). GABA B R1 and GABA B R2 binding were assessed by reverse transcription (RT) and quantitative PCR. Indeed, GABA B R1 and GABA B R2 mRNAs were detected in the  immunoprecipitate, along with CaMKIIa, a well-known FMRP mRNA target (Fig. 3a,b; uncropped representative qPCR gels, Supplementary Fig. 7d). The calcium channel accessory subunit Cacna2d2 mRNA is not a reported target for FMRP 35 and was used as a negative control. Cacna2d2 mRNA was not detected in the FMRP RIP (Fig. 3b). In parallel, we used lysates isolated from brains of mice with a genetic deletion of the Fmr1 gene. We did not observe amplification of any of the mRNAs in Fmr1 KO brains, providing additional evidence for specific binding of FMRP to GABA B R1 and GABA B R2 mRNAs (Fig. 3a-c; uncropped blots, Supplementary Fig. 7b).
Next, we determined if FMRP regulates GABA B R1 and GABA B R2 protein levels. Genetic deletion of Fmr1 leads to the constitutive translation of FMRP target mRNAs and the loss of activity-dependent translation 16 . Protein levels of GABA B R1 and GABA B R2 were compared in hippocampal synaptoneurosomes from Fmr1 KO and wild-type (WT) mice ( Fig. 3c-g). GABA B R2 basal protein levels were elevated by B53% in Fmr1 KO hippocampi (Fig. 3g). GABA B R1 protein levels also increased, albeit to a lesser extent than GABA B R2 (Fig. 3f). Collectively, these data suggest that FMRP regulates the expression of GABA B R1 and GABA B R2.  Ethanol and rapid antidepressants reduce dendritic FMRP. As an initial test to determine if FMRP-regulated translation is linked to alcohol exposure, we compared FMRP target mRNAs 35 with mRNAs that are differentially expressed in the hippocampi of alcohol-dependent humans 37 . Remarkably, 225 or B25% of verified FMRP target mRNAs overlap with mRNAs that are altered in alcohol-dependent individuals, suggesting a role for FMRP in aberrant protein levels observed in humans with AUD ( Fig. 4a) 35,37 . We then determined whether exposure to acute ethanol (30 mM, 2 h) or Ro-25-6981 (10 mM, 2 h) alters FMRP expression in the dendrites of hippocampal neurons. Using immunofluorescence, we found that ethanol and Ro-25-6981 reduced FMRP levels by B38% and 45%, respectively (ethanol: Fig. 4b,c; Ro-25-6981: Supplementary Fig. 3). These data suggest that ethanol and Ro-25-6981 alter protein expression in an FMRP-dependent manner.
Ethanol-induced synthesis of GABA B R2 requires FMRP. Due to the ethanol-induced decreases in FMRP, we hypothesized that FMRP is required for ethanol-induced expression of GABA B Rs.
Specifically, if expression of GABA B Rs is constitutive and unregulated in Fmr1 KO mice, then ethanol-induced changes in GABA B R expression should be absent in Fmr1 KO mice. Hippocampal synaptoneurosomes were isolated from WT and Fmr1 KO mice 30 min after i.p. injection of ethanol (2.5 g kg À 1 ). Western blot analysis indicated that both GABA B R1 and GABA B R2 expression remained constant in vehicle-and ethanoltreated Fmr1 KO mice. As observed in Fig. 2, WT hippocampal synaptoneurosomes showed an B23% increase in GABA B R2 but no change in GABA B R1 expression (Fig. 5a-c; uncropped blots, Supplementary Fig. 7c). These data suggest that ethanol-induced changes in GABA B R expression are dependent on FMRP translational regulation.
To determine whether protein synthesis is essential for the FMRP-dependent changes in GABA B R expression, we measured ethanol-induced GABA B Rs in the presence of cycloheximide (CHX), a protein synthesis inhibitor. As demonstrated previously, ethanol did not influence the dendritic expression of GABA B R1; however, co-treatment with cycloheximide increased GABA B R1 expression by B22%. FMRP deletion did not affect the basal, ethanol-, or cycloheximide-induced dendritic protein expression of GABA B R1 (Fig. 5d,e,h). For GABA B R2, we again saw a significant B28% increase in dendritic expression with acute ethanol treatment; however, in the presence of cycloheximide the ethanol-induced increase was abolished. Notably in Fmr1 KO cultures, no change was observed with ethanol or ethanol þ cycloheximide ( Fig. 5f,g,i).
We next examined the requirement for protein synthesis and FMRP in ethanol-dependent surface expression of GABA B Rs. Using WT and Fmr1 KO hippocampal neurons, we measured ethanol-induced surface expression of GABA B R1 with or without cycloheximide. Co-assembly of GABA B R1 and R2 is required to express GABA B R heterodimers in the membrane 32  Relative fold-enrichment as determined by real-time qPCR relative to input control (DDCt ¼ 2 À (Ct FMRP RIP À Ct IgG RIP) À (Ct FMRP input À Ct IgG input) ). FMRP binds GABA B R1, GABA B R2, and the positive control CaMKIIa mRNA as detected in the RIP sample by real-time qPCR. Cacna2d2 served as a negative control and was not detected above background. WT: Western blot analysis of hippocampal synaptoneurosomes isolated from C57BL/6 WT and Fmr1 KO mice on a C57BL/6 background indicates the absence of (e) FMRP and increased protein expression of (f) GABA B R1 and (g) GABA B R2. Representative western blots are pseudocoloured to indicate intensity of bands, and the normalized optical density for each band is indicated below blot (Lookup  Fig. 6a-c). These data suggest that GABA B R2 protein synthesis is required for the increased surface expression of the hetermultimeric receptor with ethanol exposure. FMRP is reported to repress new protein synthesis. Considering the effects of the protein synthesis inhibitor, these data suggest that the ethanol-mediated reduction in FMRP results in the increase in dendritic GABA B R2 by de novo protein synthesis. To provide more direct evidence, we performed bioorthogonal noncanonical amino-acid tagging (BONCAT) in conjunction with proximity ligation assay (PLA-Duolink) 38 . BONCAT þ PLA can be used to detect new synthesis of proteins of interest, such as GABA B R1 and GABA B R2. Through click chemistry, noncanonical amino acids that are incorporated during mRNA translation are biotinylated. PLA, on the other hand, generates a fluorescent signal when two antibodies are within 30-40 nm of each other (that is, anti-GABA B R1 or GABA B R2 and anti-biotin). By combining these methods, we determined that ethanol treatment increases new protein synthesis of GABA B R2 by B40%, but does not alter GABA B R1 synthesis, similar to the GABA B R changes induced by rapid antidepressants 8 . In Fmr1 KO dendrites, basal levels of GABA B R2s increased by B48%, while a significant decrease was observed in GABA B R1 levels. In addition, ethanol-induced translation of GABA B R2 was lost in Fmr1 KO dendrites (Fig. 6d-i). These data provide additional evidence that the ethanol-induced increase in GABA B R2 expression is due to new protein synthesis that requires the release of translational repression by FMRP.
Ethanol-induced GABA B R plasticity requires FMRP. We previously demonstrated that rapid antidepressants shift GABA B R signalling from opening potassium channels to increasing dendritic calcium 6 . To determine whether ethanol (30 mM, 2 h) causes the same plasticity in GABA B R signalling, we performed fluorescence calcium imaging in cultured WT and Fmr1 KO hippocampal neurons. A transient rise or fall in calcium in dendritic compartments can be detected using a fluorescent indicator that exhibits changes in fluorescent properties depending on the amount of bound calcium 39 . We used baclofen, a GABA B R agonist, to activate GABA B Rs in the presence or absence of ethanol. After establishing a baseline measurement, baclofen reduced dendritic calcium fluorescence in saline-treated WT neurons by B11%, a characteristic signature of GABA B R signalling increasing outward potassium conductance 8 . However, in ethanol-treated WT neurons, baclofen induced distinct calcium waves and an overall averaged increase in calcium signal of B9% (Fig. 7a-c and Supplementary Fig. 4a,c). These results recapitulate our previous observations with NMDAR antagonists 6 . In addition, these findings in WT mouse neurons are consistent with what we observed in rat cultured hippocampal neurons treated with ethanol or the clinically relevant rapid antidepressant Ro-25-6981 ( Supplementary Fig. 5). Unexpectedly, GABA B R activation in saline-treated Fmr1 KO neurons failed to reduce the calcium signal. Moreover, in ethanoltreated Fmr1 KO neurons, GABA B R activation failed to increase dendritic calcium signal (Fig. 7b,c and Supplementary Fig. 4b,c). These findings suggest that the loss of FMRP in Fmr1 KO dendrites decouples GABA B Rs from potassium channels. These results also suggest that the dynamic, ethanol-induced plasticity in GABA B R signalling, which is observed with rapid antidepressants, requires FMRP 6 .
To further substantiate that FMRP regulates ethanoldependent GABA B R plasticity, we overexpressed FMRP in rat hippocampal neurons. Overexpression of FMRP did not alter the GABA B R activation in saline-treated neurons because baclofen reduced the dendritic calcium signal. However, in ethanol-treated neurons, overexpressing FMRP blocked the ethanol-induced GABA B R plasticity ( Fig. 7d and Supplementary Fig. 4d,e). These results provide additional evidence that the dynamic reduction of FMRP with ethanol exposure is important for the expression of GABA B R plasticity.      administration did not affect the behaviours of Fmr1 KO mice in the splash and open field tests compared with saline-treated mice ( Supplementary Fig. 6a-e). Surprisingly, the basal state of immobility in the FST in Fmr1 KO mice is equivalent to ethanolinjected WT mice (Fig. 8). To explore this paradox, we examined the requirement of GABA B R activation in ethanol-induced decreases in immobility by using CGP-35348 to inhibit postsynaptic GABA B Rs. We previously showed that GABA B R antagonism blocked the antidepressant-like behaviour produced by NMDAR antagonist in the FST 6 . GABA B R inhibition alone did not affect the immobility of saline-injected WT mice in the FST, similar to what we observed previously (Fig. 8)    . Ethanol-induced increase in dendritic calcium is prevented by FMRP overexpression. Vector: Veh þ Bac: À 6 ± 1.6, n ¼ 17 dendrites; ETOH þ Bac: 2.5 ± 2.5, n ¼ 17 dendrites. FMRP overexpression: Veh þ Bac: À 4 ± 2, n ¼ 11 dendrites; ETOH þ Bac: À 4.7 ± 1.4, n ¼ 27 dendrites. Significance determined by two-way analysis of variance, followed by Tukey's multiple comparison. Values represent mean ± s.e.m. from saline-treated Fmr1 KO mice. These findings collectively demonstrate that GABA B Rs and FMRP are necessary to elicit the ethanol-mediated antidepressant response.

Discussion
Emerging behavioural and molecular evidence demonstrate that NMDAR antagonists act as rapid antidepressants 5,6,8,24,40 . Because it has long been speculated that individuals with major depressive disorders self-medicate with alcohol, we examined whether ethanol, which blocks NMDARs 9 , acts through the same synaptic pathways as NMDAR antagonists. Until this study, the molecular mechanisms shared by alcohol and antidepressants were unexplored. Here we provide molecular and behavioural evidence that acute alcohol exposure elicits antidepressant-like behaviours that persist up to 24 h after administration (Fig. 1), supporting the hypothesis that ethanol initiates lasting antidepressant activity. We have previously demonstrated that NMDAR inhibition by rapid antidepressants induces two key molecular changes that are responsible for the rapid antidepressant response, namely (1) an increase in GABA B R protein synthesis and (2) a shift in GABA B R function that increases dendritic calcium signalling 6,8 . Our current work shows that these same signature changes are produced by acute ethanol exposure (Figs 5-7).
Surface expression of functional GABA B Rs requires the dimerization of GABA B R1 and R2 subunits. Without GABA B R2, GABA B R1 is retained in the endoplasmic reticulum 32 . Our current studies show that the release of GABA B R2 mRNA translational repression by FMRP is necessary for the ethanolinduced increase in surface GABA B Rs with NMDAR blockade (Figs 4,6 and 8; Supplementary Fig. 3). Reduction of FMRP, as seen in animal models of FXS, is associated with elevated protein synthesis of target mRNAs 16 . While we have demonstrated that FMRP associates with GABA B R1 and R2 mRNAs, the loss of FMRP has a profound effect on GABA B R2 protein expression in dendrites. Constitutive loss of FMRP, as observed in many of its targets, abrogates stimulus-dependent mRNA transport and translation of target mRNAs [41][42][43][44] . We show that the ethanolinduced increase in GABA B R2 protein is also absent in Fmr1 KO mice.
The question of why GABA B R2 is uniquely affected by acute ethanol and its dependence on reduced FMRP levels is intriguing. Both GABA B R1 and GABA B R2 mRNAs were detected in the FMRP RIP (Fig. 3). Interestingly, in vivo FMRP influences GABA B R1 expression to a lesser extent compared with GABA B R2, suggesting that FMRP may act in concert with other repressors such as microRNAs to tightly regulate GABA B R1 expression 45 . Moreover, our immunostaining and protein synthesis assays suggest that the increased protein levels of GABA B R1 are not due to new protein synthesis in the dendrites, even in Fmr1 KO mice (Figs 5 and 6). These results indicate that the overall increase in synaptic GABA B R1 expression in Fmr1 KO mice may be due to an increase in presynaptic GABA B R1 expression, either through protein synthesis or increased protein stability. Notably, FMRP has been localized to axons and presynaptic terminals 46,47 . Further exploration into presynaptic and postsynaptic GABA B R expression and function in AUD and FXS is warranted.
The 'GABA hypoinhibition theory' posits that loss of inhibition is a leading cause in many of the neurological symptoms observed in FXS 48 . While studies showing reduced inhibition in models of FXS have focused on decreased expression of GABA A R subunit mRNA and protein 49 , GABA B R protein expression and dendritic signalling has not been explored. Interestingly, the GABA B R agonist baclofen has shown promise in treating FXS. (R)-baclofen administration in vitro corrects the elevated basal protein synthesis normally seen in Fmr1 KO mice, and rescues synaptic abnormalities such as increased spine density 50 . In addition, baclofen administration reduced symptoms related to FXS in Fmr1 KO mice 50,51 . Recent studies have shown that the excitatory drive to fast spiking inhibitory neurons is reduced in the cortex of Fmr1 KO mice 52 . Our data expand on these findings by suggesting that postsynaptic GABA B R coupling to inwardly rectifying G protein-coupled potassium channels (GIRK) is absent in Fmr1 KO neurons (Fig. 7). Collectively, these results may imply that the therapeutic effects of baclofen in the Fmr1 KO mouse may be due to the activation of presynaptic GABA B receptors that may in turn reduce glutamate release and reduce hyperactive metabatropic glutamate receptor mGluR signalling in the hippocampus 53 .
Changes in gene expression and protein synthesis are essential for normal neuroplasticity, but these crucial processes are dysregulated by drug addiction 18,54 . Several lines of evidence support parallel changes in GABA B R mRNA translation/ signalling as a result of NMDAR blockade that may be critical for alcohol actions. First, NMDAR antagonists mimic some effects of ethanol in humans 55 , suggesting common biochemical/ electrophysiological signalling pathway(s). Second, changes in GABA B R2 brain gene expression correlates with lifetime alcohol consumption, supporting a role for altered GABA B R signalling in AUD 56 . Third, although controversial, the GABA B R agonist baclofen may decrease alcohol consumption in some alcoholics 57 . In summary, our data define a common molecular paradigm for alcohol and rapid antidepressants, and identify a mechanism for the initial antidepressant effects of alcohol. A shift in GABA B R signalling is observed with both rapid antidepressants and acute ethanol treatment, which may provide insight into the molecular basis for the high comorbidity between major depressive disorder and AUD.
Note added in proof: Spencer et al. found fmrp mediated changes ion channels with chronic ethanol exposure 58 .

Methods
Cell culture. Primary hippocampal neurons were prepared as previously described by Niere et al. 43 . Briefly, hippocampi were extracted from postnatal day 1-3 Sprague-Dawley rat pups, WT C57BL/6 mouse pups, or Fmr1-knockout (Fmr1 KO) mouse pups on a C57BL/6 background. The tissue was dissociated and plated in neurobasal A medium supplemented with B27, glutamine, and 1% fetal bovine serum. Cultures were plated at a density of B100,000 cells per 12 mm on glass coverslips that had been coated overnight with 50 mg ml À 1 poly-D-lysine and 25 mg ml À 1 laminin in borate buffer. Cultures were fed after 1 day in vitro (DIV), and media was replaced approximately once a week with either fresh rat culture media (neurobasal A supplemented with B27, glutamine and 3 mM AraC) or fresh mouse culture media (glial-conditioned media with 3 mM AraC) until cultures were used at DIV 14-21.
In vitro pharmacology. Primary hippocampal neurons were treated in ethanol vapour chambers according to a method adapted from Chandler et al. 59 . Ethanol vapour chambers were prepared by placing a reservoir of 31.5 mM ethanol (105% of the desired ethanol concentration, that is, 30 mM) in a plastic container with 24-well culture plates containing neuronal cultures in which 30 mM ethanol was added to the culture media. Chambers were filled with 95% O 2 /5% CO 2 and cultures were incubated for 2 h at 37°C. Cultures treated with vehicle (H 2 O) were incubated in the same manner but in the absence of ethanol. For calcium imaging, ethanol was added directly to HEPES-based artificial cerebral spinal fluid (ACSF in mM: 100 NaCl, 10 HEPES (pH 7.4), 3 KCl, 2 CaCl 2 , 1 MgCl 2 , 10 glucose) that was adjusted to match the osmolarity of cell culture media for live imaging. For GABA B R activation neurons were treated with (R)-baclofen (50 mM, Tocris). For Figs 5 and 6, cultured hippocampal neurons were pre-treated with cycloheximide (50 mM, Tocris) for 10 min before ethanol treatment. For Supplementary Fig. 3, neurons were treated with Ro-25-6981 (10 mM, Tocris) or Veh (H 2 O) for 2 h. All cultures were treated at 14-21 DIV. Following treatment, cultures were immediately fixed or live imaged.
injections of either 200 ml saline or 2.5 g kg À 1 ethanol (in a volume of 200 ml saline) 23 . For Fig. 8, CGP-35348 (100 mg kg À 1 ) was i.p. injected with or without ethanol (in a volume of 200 ml saline). All animals were housed four mice per cage according to genotype. All treatments were administered to one mouse per home cage. At the time of drug treatment, animals were coded by number. During the behavioural tasks, animal performance was video recorded, and then later blindly scored. In certain tasks (for example, open field), the animals were scored by a computer program and blinding was not necessary during that process. For western blots the hippocampi were isolated 30 min post injection and flash-frozen. All experiments were carried out in accordance with the National Institutes of Health's Guide for the Care and Use of Laboratory Animals and approved by the UT-Austin Institutional Animal Care and Use Committee (IACUC).
Immunofluorescence. Primary neuronal cultures on glass coverslips were fixed with 4% paraformaldehyde (PFA) at room temperature for 20 min, washed three times with phosphate-buffered saline (PBS) and permeabilized in 0.25% Triton X-100 in PBS for 5 min. For FMRP staining, neurons were fixed and permeabilized in 100% methanol at À 20°C for 10 min. Neurons were washed three times in PBS and then blocked (10% normal goat serum in PBS) for 30 min at room temperature. Primary antibodies were incubated in blocking buffer at 4°C overnight. Neurons were washed three times for 10 min with PBS, and then incubated in secondary antibody in blocking buffer for 1 h at room temperature, and washed three times for 10 min with PBS before mounting in Fluoromount with DAPI (SouthernBiotech, 0100-20) as outlined in Sosanya et al. 60 . Surface staining was performed similarly to Workman et al. 6 . Neurons were first fixed in 4% PFA for 10 min on ice, washed three times in PBS, blocked with 3% normal goat serum, and then incubated in primary surface antibody in 3% blocking buffer overnight at 4°C. Following primary surface antibody incubation, neurons were washed six times for 10 min each in PBS, then permeabilized with 0.25% Triton X-100 in PBS for 5 min, followed by three washes for 10 min each in PBS, and again incubated in primary total antibody in 3% blocking buffer overnight at 4°C. Neurons were washed three times for 10 min each with PBS, and then incubated in secondary antibody in 3% blocking buffer for 1 h at room temperature, and finally washed four times for 10 min with PBS before mounting in Fluoromount with DAPI to slides. The primary antibodies used were: Adeno-associated viral vectors. The FMRP and tdTomato coding sequences were cloned into separate adeno-associated viral (AAV) vectors containing a mouse synapsin promoter, a woodchuck post-transcriptional regulatory element and SV40 poly-adenylation sequence between flanking AAV2 inverted terminal repeats. rAAVs were assembled using a modified helper-free system (Stratagene) as serotype 2/1 (rep/cap genes) viruses, and collected and purified over sequential caesium chloride gradients as previously described 61 . Viral titres were 41 Â 10 9 infectious particles per microlitre. For FMRP and&#8232;tdTomato co-infections, rAAV:mSYN-FMRP and rAAV:mSYN-tdTomato were mixed at a ratio of 4:1. One microlitre of the resulting rAAV mix was used per coverslip of primary cultured neurons. Imaging was performed B1 week after infection.
Live calcium imaging. Dissociated hippocampal cultures were prepared from WT and Fmr1 KO mice as described 43 . Neurons at 14-21 DIV were used for live calcium imaging. Neurons were treated as outlined in in vitro pharmacology above. Before imaging, cells were incubated in ACSF with Oregon Green 488 BAPTA-1 AM (OGB, 200 mM; 30 min; 37°C; ThermoFisher) as described 6 . After OGB incubation, cells were transferred to fresh ACSF (37°C) for imaging (1 frame per 20 s). Baseline calcium signal was imaged (1 min), after which (R)-baclofen (50 mM, Tocris) or vehicle (H 2 O) was added. For ethanol-treated cells, the neurons were incubated with OGB and imaged in ACSF containing ethanol (30 mM). Neurons were imaged for 800 s at room temperature. Quantification of the calcium signal was performed using Metamorph (Molecular Devices) as described 6 . Briefly, dendritic regions of interest (ROI) that were at least 5 mm from the soma were analysed. The mean intensity values for each ROI at each time were averaged as baseline (F 0 ). The ROI intensity values obtained at each time point after the addition of baclofen or vehicle were averaged (F). The equation, DF/F ¼ ((F À F 0 )/F 0 ), was used to measure the change in signal and data were plotted as a percentage of the baseline.
BONCAT-PLA. BONCAT-PLA was performed using Click-it Metabolic Labeling azidohomoalaine (AHA), Biotin-Alkyne and Click-iT Reaction buffer kit (Life Technologies). Proximity ligation assay (PLA) was performed using Duolink kit (Duolink, Sigma) 38 . Briefly, primary hippocampal neuronal cultures were incubated in a methionine-free ACSF media for 30 min. AHA was then added to the media just before neurons were treated with ethanol for 2 h as previously described. Neurons were fixed in 4% PFA for 15 min, washed two times for 5 min with 3% bovine serum albumin (BSA) in PBS, followed by permeabilization with 0.25% Triton X-100 in PBS for 15 min, and washed as before. Neurons were incubated for 30 min at room temperature in Cell Buffer Additive/Click-it Cocktail according to manufacturer directions. Neurons were washed as before and then blocked and incubated with primary antibody as previously described. Next, neurons were incubated in the appropriate PLA probes diluted in blocking buffer and secondary antibody at 37°C for 1 h. Neurons were washed in RT Buffer A two times for 5 min, and incubated in ligation solution at 37°C for 30 min, and washed again in Buffer A. Neurons were incubated in amplification solution at 37°C for 2-3 h, followed by washing in RT Buffer B two times for 10 min and 1% buffer B for 1 min. Last, neurons were mounted to slides in Duolink mounting media for imaging. Primary and secondary antibodies included: GABA B R1 (1/50 dilution; Santa Cruz, sc-14006), GABA B R2 (1/100; Neuromab 75-124), MAP2 (1/2,000 dilution; Abcam, ab5392), biotin/a-rabbit (1/500; Sigma, SAB3700857), Alexa488 (1/500; Life Technologies, A-11039). PLA probes used: Rabbit Minus (1/5; Duolink, 82005), Mouse Plus (1/5; Duolink, 82001).
Microscopy and analysis. Images were acquired with a Leica SP5 confocal microscope under a Â 63 oil immersion lens for fixed tissue or a Â 63 water immersion lens for live imaging. Max projected images were used for immunostaining analysis from 10 mm Z-stacks of 1,024 Â 1,024 pixels obtained using a 400-Hz scan rate 60 . For each experiment, all images were collected using the same settings. Fixed images were analysed using NIH imaging software ImageJ, and live imaging quantification was performed with Metamorph (Molecular Devices, Sunnyvale, CA). Background signal was determined by shifting the ROI adjacent to the dendrite being traced, but void of all processes. Dendritic signal was background subtracted and averaged every 10 mm using a customized R script.
RNA immunoprecipitations. Cortices from 6-week-old C57BL/6 and Fmr1 KO male mice were collected and flash-frozen on dry ice. RIP was performed by modified method of Jain et al. and Keene et al. 63,64 . Tissue was homogenized and lysed with a cordless pestle motor and disposable pellet mixers (VWR) in polysome lysis buffer (10 mM HEPES pH 7.0, 100 mM KCl, 25 mM EDTA, 5 mM MgCl 2 , 1 mM DTT, 0.5% NP-40) in a 1:1 tissue-buffer ratio. RNaseOUT (Thermo) and protease/phosphatase inhibitors (Halt Protease and Phosphatase Cocktail, Pierce Biotechnology) were freshly added to samples. Samples were rotated for 10 min at 4°C to induce swelling and then flash-frozen on dry ice. Samples were thawed by holding between fingers at room temperature to lyse and nuclei were pelleted at 3,000g for 10 min. Lysates obtained above were pre-cleared by adding 50 ml of washed magnetic bead slurry (Protein A Dynabeads, Thermo) and rotating for 30 min at 4°C. To bind the antibody to the beads, 50 ml of magnetic beads slurry was washed and then resuspended in eight volumes of NT-2 buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM MgCl 2 , 1 mM DTT, 0.05% NP-40 with RNaseOUT/protease and phosphatase inhibitors added fresh) þ 5% BSA. Ten micrograms of either FMRP (Abcam, ab17722) or IgG (Santa Cruz Biotechnologies, sc-2027) antibodies was added to the beads and rotated for 10 min at room temperature. Antibody-bound beads were washed four times with ice-cold NT-2 buffer. For the immunoprecipitation, 4.5 mg of protein from pre-cleared lysates was added to an RNase-free microcentrifuge tube containing the antibodybound beads. Input collected at this step for downstream analysis was either 1% of the final pre-cleared lysate volume in the immunoprecipitation reaction (for immunoblotting) or 10% of the final pre-cleared lysate volume (to normalize in qPCR). The antibody-bead-lysate mixture was then diluted at a ratio of 1:5 with NET-2 buffer (20 mM EDTA pH 8.0, and 1 mM DTT in NT-2 buffer; RNaseOUT and protease/phosphatase inhibitors added fresh) and rotated for 1 h at room temperature. Beads were quickly washed six times in ice-cold NT-2 buffer and immediately resuspended in 350 ml TRI Reagent Solution (Ambion) for 10 min at room temperature. Beads were pelleted and the supernatant was removed and resuspended in 350 ml of absolute ethanol. RNA was extracted by applying ethanolresuspended samples to spin column from the Direct-zol RNA MiniPrep Kit (Zymogen) according to the manufacturer's instructions. Eluted RNA (25 ml) was DNase treated using the TURBO DNA-free kit (Thermo).
Forced swim test. Male mice were tested in the forced swim test 24 h post injection as described previously 6,8 . Mice were individually placed into a cylinder containing 3 l of water (25°C) for 6 min. Each session was video recorded and the last 4 min of the sessions were later scored blindly for immobility. Animals were scored for escape-directed behaviours. The water was replaced between animals. Experiments were repeated by three independent experimenters. Data were normalized by experimenter. Power analysis was performed in R Programming 66 to predict sample size for all behavioural tests. This sample size was used as a guideline for the WT animals, however since transgenic animals were used, the exact sample size for each group may not have been possible. Transgenic animal sample size was as close as possible to that which was calculated, due to limitations of litter size.
Open field. Twenty-four hours after animals were injected i.p. with either saline or ethanol, they were studied in the open field test similar to Treit et al. 31 . Mice were individually placed in a 40 cm Â 40 cm Â 35 cm arena with opaque walls. Each test session lasted 30 min under 85 lux illumination. Sessions were video recorded and analysed via ANY-Maze (Stoelting, Wood Dale, IL). Mice were considered to be in the centre of the maze if they entered a 18.5 cm Â 18.5 cm area in the centre of the apparatus. Mice were returned to their home cage at the end of the test session, and the arena was wiped down with 70% ethanol before the start of each run.
Splash test. Two and a half hours after open field testing, animals underwent the splash test 27,28 . Cage mates were moved from their home cage to a holding cage, and each animal was individually tested in its home cage. Two-hundred microlitres of 10% sucrose was applied to the dorsal fur of the mouse. Mice were monitored and video recorded for 5 min and then moved to a different holding cage. Videos were later scored blindly for latency to initiate grooming and for total time spent grooming. Grooming behaviour included licking, grooming with forepaws, and scratching.
Statistical analysis. Power analysis was performed in R Programming 66 to predict sample size. Prism software (GraphPad) was used for all statistical analyses. Statistical comparisons were made using one-way analysis of variance (ANOVA), two-way ANOVA, Student's t-test or w 2 -test with Yates. Outliers were determined using Grubbs' test (a ¼ 0.05). All data are expressed as mean ± s.e.m.