Acute ethanol exposure reduces serotonin receptor 1A internalization by increasing ubiquitination and degradation of β-arrestin2

Acute alcohol exposure alters the trafficking and function of many G-protein–coupled receptors (GPCRs) that are associated with aberrant behavioral responses to alcohol. However, the molecular mechanisms underlying alcohol-induced changes in GPCR function remain unclear. β-Arrestin is a key player involved in the regulation of GPCR internalization and thus controls the magnitude and duration of GPCR signaling. Although β-arrestin levels are influenced by various drugs of abuse, the effect of alcohol exposure on β-arrestin expression and β-arrestin–mediated GPCR trafficking is poorly understood. Here, we found that acute ethanol exposure increases β-arrestin2 degradation via its increased ubiquitination in neuroblastoma-2a (N2A) cells and rat prefrontal cortex (PFC). β-Arrestin2 ubiquitination was likely mediated by the E3 ligase MDM2 homolog (MDM2), indicated by an increased coupling between β-arrestin2 and MDM2 in response to acute ethanol exposure in both N2A cells and rat PFC homogenates. Importantly, ethanol-induced β-arrestin2 reduction was reversed by siRNA-mediated MDM2 knockdown or proteasome inhibition in N2A cells, suggesting β-arrestin2 degradation is mediated by MDM2 through the proteasomal pathway. Using serotonin 5-HT1A receptors (5-HT1ARs) as a model receptor system, we found that ethanol dose-dependently inhibits 5-HT1AR internalization and that MDM2 knockdown reverses this effect. Moreover, ethanol both reduced β-arrestin2 levels and delayed agonist-induced β-arrestin2 recruitment to the membrane. We conclude that β-arrestin2 dysregulation by ethanol impairs 5-HT1AR trafficking. Our findings reveal a critical molecular mechanism underlying ethanol-induced alterations in GPCR internalization and implicate β-arrestin as a potential player mediating behavioral responses to acute alcohol exposure.

Acute alcohol exposure produces a wide range of behavioral and neurobiological effects, including feelings of euphoria, loss of coordination, and cognitive impairment, which are thought to be key predictors of vulnerability to alcohol abuse in humans (1)(2)(3). This notion is further supported by a recent longitudinal study showing that individuals who exhibit higher sensitivity to the rewarding effect and lower sensitivity to the sedative effect of an initial acute alcohol exposure were more likely to develop symptoms of alcohol use disorders and escalate their alcohol consumption over time (4). Many aspects of behavioral responses to alcohol are mediated by various G-proteincoupled receptors (GPCRs). 2 For example, ethanol-induced locomotor stimulation is mediated by serotonin 5-HT 1A receptors (5-HT 1A Rs) because treatment with 5-HT 1A R agonist (8-OH-DPAT) reduced locomotor activity in response to an acute ethanol exposure (2.5 g/kg, i.p.) (5). Moreover, the sedative effect of ethanol is regulated by metabotropic glutamate receptor 5 (mGluR5) as genetic deletion of mGluR5 or treatment with the mGluR5 antagonist, 2-methyl-6-(phenylethynyl)pyridine, increases the duration of loss of righting reflex in mice acutely exposed to ethanol (6). Thus, elucidating the molecular mechanisms underlying ethanol-induced neuroadaptations in GPCRs is important for identifying potential new targets that mediate aberrant behavioral responses to alcohol.
␤-Arrestin is an important modulator of GPCR internalization. Upon GPCR activation, ␤-arrestin is recruited to the receptor followed by uncoupling of the receptor from G␣-proteins and then a direct interaction with clathrin to induce receptor internalization. The rate and amplitude of GPCR internalization mediated by ␤-arrestin determine the duration and strength of GPCR signaling. Thus, a change in ␤-arrestin expression would alter receptor internalization, signaling, and receptor-mediated behavioral responses to ethanol. For instance, knockout of ␤-arrestin2 increases [ 35 S]GTP␥S binding to cannabinoid type 1 receptors (CB1Rs) in the cortex, which is presumably due to increased CB1R protein levels on the plasma membrane (7). The level of CB1R is known to be positively associated with ethanol consumption and preference (8,9). The effect of acute or chronic ethanol exposure on ␤arrestin expression is largely unknown with one exception. Chronic ethanol exposure was reported to increase ␤-arrestin mRNA and protein levels in the hippocampus and the striatum of alcohol-preferring rats after voluntary ethanol consumption (6% v/v, 21 days) compared with nonpreferring rats (10). The route, duration, and dose of ethanol exposure likely influence the effect of ethanol on ␤-arrestin expression. Thus, the first aim of this study was to examine the effect of acute ethanol exposure on ␤-arrestin expression using both cultured cells and rodent brains.
When GPCRs are internalized upon stimulation, ␤-arrestin dissociates from the internalized receptors. The dissociated ␤-arrestin either recycles to the membrane to promote additional cycles of internalization or enters proteasomes or lysosomes for degradation (11,12). The endocytic fate of ␤-arrestin, recycling or degradation, is strongly controlled by post-translational modifications such as ubiquitination (13). Ubiquitination involves the sequential action of three enzymes: ubiquitin-activating E1 enzyme, ubiquitin-carrying E2 enzyme, and ubiquitin protein E3 ligase (14). When E3 ligases bind to their substrates, E2 enzymes are recruited to transfer ubiquitin molecules to the substrate (15). In various model systems, ␤-arrestin is capable of coupling to several E3 ligases, including MDM2 and neural precursor cell expressed developmentally down-regulated protein 4 (NEDD4) (16 -18). For example, ␤-arrestin forms a complex with MDM2 in CHW cells, COS-7 cells, and mouse brain lysate, and this ␤-arrestin2-MDM2 interaction results in an increase in both basal and receptor-stimulated ␤-arrestin ubiquitination (16). However, ubiquitination is a reversible modification as linkages between ubiquitin and substrates are frequently hydrolyzed by deubiquitinating enzymes to maintain protein homeostasis and an available pool of free ubiquitin in the cell (19). E3 ligases can modify one or multiple lysine residues of a protein with either a single ubiquitin molecule (monoubiquitination) or ubiquitin polymers (poly-ubiquitination) (19). Emerging evidence suggests that poly-ubiquitination directs substrates to proteasomes for degradation, whereas mono-ubiquitination is primarily involved in protein trafficking (15,20). It has been well-characterized that there is an increase in ␤-arrestin poly-ubiquitination when receptors are stimulated. For instance, activation of the ␤ 2 -adrenergic receptor in HEK293 and COS-7 cells increases ␤-arrestin poly-ubiquitination in an MDM2-dependent manner (11,16). Additionally, treatment with antidepressants (e.g. citalopram) for 3 days or methamphetamine (10 M) for up to 24 h enhances polyubiquitination of ␤-arrestin in C6 rat glioma and PC12 cells, respectively (21,22). These data suggest that ␤-arrestin ubiquitination can be induced by various stimuli. Ubiquitinated ␤-arrestins are predominantly deubiquitinated and become available for subsequent recruitment to activated receptors; however, persistent ubiquitination of ␤-arrestin can lead to degradation (11,21). Disruption of ␤-arrestin ubiquitination homeostasis could dysregulate the post-endocytic fate of ␤-arrestin and subsequent degradation, which will impact GPCR trafficking and activity. Thus, the second aim of this study was to determine the effect of acute ethanol exposure on ␤-arrestin ubiquitination and degradation as a potential mechanism for altered ␤-arrestin expression.
Acute ethanol exposure disrupts the trafficking of GPCRs in rodent brains and cultured neurons. For example, acute ethanol treatment (30 mM, 2 h) to rat hippocampus cultured neurons significantly increases GABA B receptor surface expression compared with the vehicle treatment (23). Moreover, acute ethanol administration (2.5 g/kg, p.o.) increases [ 3 H]DAMGO binding to -opioid receptors in rat PFC (24). However, the molecular mechanisms underlying ethanol-induced alterations in GPCR trafficking remain unclear. Because ␤-arrestin plays an important role in regulating the trafficking of many GPCRs, a change in ␤-arrestin expression and/or recruitment by ethanol exposure would influence receptor trafficking. We used 5-HT 1A Rs as a receptor model system to determine the effect of acute ethanol exposure on ␤-arrestin-mediated GPCR trafficking because the mechanism of 5-HT 1A R trafficking is wellcharacterized (25), and this receptor plays a role in mediating ethanol drinking behavior (26 -28). This study showed for the first time that acute ethanol exposure increases proteasomedependent ␤-arrestin2 degradation via MDM2 in both neuroblastoma 2a (N2A) cells and rat PFC. Furthermore, ethanol exposure abolished agonist-induced 5-HT 1A R trafficking that is likely attributed to dysregulation of ␤-arrestin2 expression and activity.

Validation of ␤-arrestin2 antibody
We verified the specificity of a commercially available ␤-ar-restin2 antibody in cells treated with ␤-arrestin2 siRNA. The protein levels of ␤-arrestin2 were assessed by Western blotting and immunocytochemistry. Knockdown of ␤-arrestin2 via siRNA significantly reduced ␤-arrestin2 protein levels by 63.43 Ϯ 9.40% when compared with the control cells transfected with a corresponding scrambled siRNA (Fig. S1A, unpaired Student's t test, p Ͻ 0.001). Additionally, ␤-arrestin2 levels were also examined in knockdown and control cells by immunocytochemistry. There was a pronounced reduction of ␤-arrestin2 fluorescent intensity in ␤-arrestin2 siRNA-treated cells when compared with the scrambled siRNA treatment (Fig.  S1B, unpaired Student's t test, p Ͻ 0.01).

Acute ethanol exposure reduces the total protein level of ␤-arrestin2 in N2A-5HT 1A R cells and rat PFC
We examined the effect of acute ethanol exposure on ␤-ar-restin1/2 expression in both N2A cells and rat PFC. In N2A cells, ␤-arrestin1/2 levels were determined following the treatment with ethanol (15-75 mM) or vehicle (media) for 18 h without withdrawal (Fig. 1A). Data were normalized to the protein level of ␤-actin, which was not altered by ethanol treatment. A one-way ANOVA revealed a significant dose effect of ethanol on ␤-arrestin2 protein level, F (3,20) ϭ 5.796, p Ͻ 0.01. Bonferroni post hoc analysis showed that 30 and 75 mM ethanol treatment significantly reduced ␤-arrestin2 levels by 44.8 Ϯ 2.0 and 60.3 Ϯ 1.7%, respectively, when compared with the vehicle treatment (Fig. 1A, p Ͻ 0.01). The ␤-arrestin1 protein levels

Ethanol-induced ␤-arrestin2 ubiquitination and degradation
were not detectable in N2A cells with or without ethanol treatment. The reduced ␤-arrestin2 level was not due to toxicity induced by ethanol because the selected ethanol doses did not alter cell viability and membrane permeability measured by trypan blue assay (Fig. S2A) and immunocytochemistry, respectively (Fig. S2, B and C).
A similar effect was observed in the PFC of rats acutely exposed to ethanol vapor for 12 h without withdrawal (Fig. 1B). Ethanol exposure significantly reduced ␤-arrestin2 protein levels in the PFC by 32.4 Ϯ 2.0% compared with the air exposure ( Fig. 1B, unpaired Student's t test, p Ͻ 0.05). However, ␤-arres-tin1 protein levels did not significantly differ between ethanoland air-treated groups.

Acute ethanol exposure increases ubiquitination of ␤-arrestin2 in N2A-5HT 1A R cells and rat PFC
We further examined whether ethanol-induced reduction in ␤-arrestin2 levels was due to an increase in ubiquitination and subsequent degradation. First, immunocytochemistry was performed to measure the co-localization between fluorescently labeled ␤-arrestin2 (green) and ubiquitin (red) in N2A cells following 18 h of ethanol (15-75 mM) or vehicle (media) treatment ( Fig. 2A). Co-localization was calculated as the number of spatially overlapped pixels ( Fig. 2A, yellow) between ␤-arrestin2 (green) and ubiquitin (red) followed by conversion into a co-localization coefficient using FIJI software (National Institutes of Health). A one-way ANOVA indicated a significant effect of ethanol treatment on the co-localization between ␤-arrestin2 and ubiquitin, F(3,116) ϭ 14.87, p Ͻ 0.01. Bonferroni post hoc analysis revealed that 30 and 75 mM ethanol treatment significantly increased this co-localization by 87.6 Ϯ 2.1% (30 mM) and 201.9 Ϯ 7.5% (75 mM) when compared with the vehicle treatment (Fig. 2B, p Ͻ 0.01).
We also performed immunoprecipitation assays as a complementary approach to immunocytochemistry to examine whether there was an increased coupling between ubiquitin and ␤-arrestin2 following ethanol exposure (30 mM; 18 h). First, to verify the specificity of the ␤-arrestin2 antibody used for immunoprecipitation, we immunoprecipitated ␤-arrestin2 in the presence or absence of a custom-made ␤-arrestin2blocking peptide in drug-naïve N2A cells and rat PFC tissue. The blocking peptide completely eliminated signals produced by antibodies used to immunoblot for ubiquitin and ␤-arrestin2 from immunoprecipitated ␤-arrestin2 samples (Fig. 2C), suggesting the specificity of this ␤-arrestin2 antibody for immunoprecipitation. Then, we used this antibody to immunoprecipitate ␤-arrestin2 followed by immunoblotting for ubiquitin, which resulted in distinct mono-ubiquitination and poly-ubiquitination signals (Fig. 2D). We selected the ϳ60-kDa band as mono-ubiquitinated ␤-arrestin2 based on the presumed ϳ8-kDa upward shift in the molecular mass of ␤-arrestin2 caused by the addition of one ubiquitin chain. The poly-ubiquitination signal for ␤-arrestin2 present at a molecular mass of Ͼ80 kDa was consistent with reports from the literature (22,29,30). In N2A cells, we found that ethanol treatment increased both mono-ubiquitination and poly-ubiquitination of ␤-arres-tin2 by 83.2 Ϯ 1.7 and 83.6 Ϯ 2.7%, respectively, compared with the vehicle treatment (Fig. 2, E and F, unpaired Student's t test, p Ͻ 0.01).
Similar to our observation in N2A cells, immunoprecipitation of ␤-arrestin2 in rat PFC homogenates showed that ethanol exposure significantly increased mono-ubiquitination by 46.8 Ϯ 1.5% and poly-ubiquitination of ␤-arrestin2 by 50.9 Ϯ 2.2% compared with the air treatment (  ). B, representative Western blots and quantification of ␤-arrestin 1/2 protein levels in rat PFC. Rats were acutely exposed to ethanol vapor (EtOH) or control (Air) for 12 h followed by immediate sacrifice. Ethanol exposure significantly reduced ␤-arrestin2 protein level in rat PFC (n ϭ 6, *, p Ͻ 0.05 versus air, unpaired Student's t test).

Ethanol-induced ␤-arrestin2 ubiquitination and degradation
tion, we further assessed the strength of ␤-arrestin2 coupling to MDM2 in response to ethanol exposure. Ethanol exposure increased the association of MDM2 with immunoprecipitated ␤-arrestin2 by 91.1 Ϯ 2.2 and 32.6 Ϯ 1.3% in N2A cells and rat PFC, respectively, when compared with their corresponding control treatments (Fig. 2, G and L, unpaired Student's t test, p Ͻ 0.05). In contrast, ethanol exposure had no effect on the association of immunoprecipitated ␤-arrestin2 with another E3 ligase, NEDD4, in either model system (Fig. 2, H and M). It is also worth noting that the relative intensities of bands between mono-and poly-ubiquitinated ␤-arrestin2 in N2A cells were notably different from those in rat PFC tissues. This is likely attributed to differential expressions of various E2 and E3 ligases involved in ubiquitination of ␤-arrestin2 in each model. Additionally, bands located between mono-(ϳ60 kDa) and poly-ubiquitinated ␤-arrestin2 (Ͼ80 kDa) likely reflect a spectrum of different ubiquitination states (bi-, tri-, tetra-ubiquitination, etc.) on individual lysine residues, as described previously (31). Because the role of these intermediate ubiquitina-tion states is currently unclear, this study focused on mono-and poly-ubiquitination of ␤-arrestin2.

Ethanol-induced ␤-arrestin2 ubiquitination and degradation
P19S was quantified as the spatial overlap between co-localized pixels (yellow) and P19S (blue), which was represented by white overlapped pixels (Fig. 3D). A one-way ANOVA indicated a significant effect of ethanol exposure on ubiquitin, ␤-arrestin2, and P19S co-localization, F(2,12) ϭ 54.48, p Ͻ 0.01. Bonferroni post hoc analysis showed that 30 and 75 mM ethanol treatment significantly increased the co-localization of ␤-arrestin2, ubiquitin, and P19S by 44.5 Ϯ 2.8% (30 mM) and 58.3 Ϯ 2.5% (75 mM), respectively, when compared with the vehicle treatment (Fig. 3F, p Ͻ 0.01). To test whether the proteasome-dependent degradation pathway is the primary mechanism for ethanolinduced ␤-arrestin2 reduction, N2A-5HT 1A R cells were treated with the proteasome inhibitor MG132 (5 M) or vehicle (media) in the presence or absence of ethanol (30 -75 mM; 18 h) followed by measurement of ␤-arrestin2 levels (Fig. 3G) To assess whether acute ethanol exposure also directed ␤-ar-restin2 into lysosomes for degradation, we measured the co-localization between fluorescently labeled ␤-arrestin2 and a lysosome marker, LAMP1, following treatment with ethanol (30 -75 mM, 18 h) or vehicle (media) (Fig. 3I). A one-way ANOVA showed a significant effect of ethanol exposure on the co-localization between ␤-arrestin2 and LAMP1, F(2,102) ϭ 4.87, p Ͻ 0.01. Bonferroni post hoc analysis showed that only the high-dose ethanol treatment (75 mM) produced a small but significant increase in the co-localization of ␤-arrestin2 with LAMP1 by 14.0 Ϯ 2.6% (Fig. 3J, p Ͻ 0.05). Collectively, these data suggest that acute ethanol exposure primarily directs ␤-ar-restin2 to undergo the proteasome-dependent degradation; however, a small pool of ␤-arrestin2 can also be routed to lysosomes under specific conditions such as high doses of ethanol exposure.

Ethanol exposure reduces and delays recruitment of ␤-arrestin2 to the plasma membrane
We also investigated whether ethanol-induced disruption of 5-HT 1A R internalization resulted from a deficit in ␤-arrestin2 recruitment to the membrane. The effect of acute ethanol exposure (15-30 mM; 18 h) on 8-OH-DPAT-stimulated ␤-ar-restin2 translocation was examined by performing a segmentation analysis of the redistribution of fluorescently-labeled ␤-ar-restin2 as described previously (Fig. 5A) (34). This method is commonly used to measure the translocation of fluorescently labeled proteins to the membrane (36,37). Briefly, the outer boundary of each cell was defined by the localization of fluorescently labeled surface 5-HT 1A Rs, and the inner boundary was marked as a standardized 2-m distance from the outer boundary. Accordingly, the outer segment was calculated as the cellular area between the outer and inner boundaries, and the remaining cellular area within the inner boundary was indexed as the inner segment. A one-way ANOVA showed a significant effect of ethanol treatment on the basal level of ␤-arrestin2 membrane localization, F(2,84) ϭ 8.85, p Ͻ 0.01. Bonferroni post hoc analysis revealed that the baseline ␤-arrestin2 membrane localization was reduced after ethanol treatment (30 mM) compared with the vehicle treatment (Fig. 5C, p Ͻ 0.01). We also examined ␤-arrestin2 membrane recruitment upon 5-HT 1A R stimulation by 8-OH-DPAT (1 M). A two-way ANOVA revealed significant main effects of treatment time, F(3,336) ϭ 37.19, p Ͻ 0.01, and interaction, F(6,336) ϭ 9.67, p Ͻ 0.01. Bonferroni post hoc analysis showed that 8-OH-DPAT stimulation in vehicle-treated cells significantly increased the redistribution of ␤-arrestin2 into the outer segment (55 Ϯ 3.2%) as early as 5 min (Fig. 5D, p Ͻ 0.01). However, 8-OH-DPAT stimulation in ethanol-treated cells (30 mM, 18 h) did not show a significant increase in ␤-arrestin2 redistribution into the outer segment (31.7 Ϯ 4.0%) until 15 min. Taken together, these findings indicate that acute ethanol exposure reduced and delayed 8-OH-DPAT-stimulated ␤-arrestin2 membrane Ethanol-induced ␤-arrestin2 ubiquitination and degradation translocation, which likely explains the abolished agonist-induced internalization of 5-HT 1A R following acute ethanol exposure.

Discussion
In this study, we show for the first time that acute ethanol exposure reduced ␤-arrestin2 expression level via MDM2-mediated ubiquitination and degradation. An important functional consequence of ethanol-induced dysregulation of ␤-ar-restin2 was the abolished 5-HT 1A R internalization stimulated by agonist, which was likely impacted by delayed and reduced ␤-arrestin2 recruitment to the membrane upon 5-HT 1A R stimulation.
Drugs of abuse such as morphine, heroin, and methamphetamine modulate the expression of ␤-arrestin (22,38,39); however, there is very little knowledge about the effect of ethanol exposure on ␤-arrestin expression. In this study, we observed a notable reduction in ␤-arrestin2 levels in N2A-5HT 1A R cells and rat PFC following acute ethanol exposure. ␤-Arrestin degradation is regulated by ubiquitination, which occurs rapidly in response to various stimuli (13,21). Herein, acute ethanol exposure increases ␤-arrestin2 ubiquitination in both N2A-5HT 1A R cells and rat PFC as evidenced by a marked increase in both mono-and poly-ubiquitination of ␤-arrestin2. Previous reports have focused exclusively on poly-ubiquitination of ␤-arrestin2 in response to stimulation of various types of receptors and other stimuli (16,29,30). Poly-ubiquitination is considered an important signal that directs proteins, such as ␤-arrestin, into proteasomes for degradation and can be induced by a diverse array of stimuli (15,40). For instance, chronic antidepressant treatment increases poly-ubiquitination of ␤-arrestin2, which facilitates P26S proteasome-dependent degradation of ␤-arres-tin2 in C6 rat glioma cells (21). Furthermore, poly-ubiquitination and subsequent proteasome-dependent degradation of ␤-arrestin occur in response to acute methamphetamine treatment (1 M, 5 min) in PC-12 cells (22). In line with these observations, we found that ethanol exposure dose-dependently increased sorting of ␤-arrestin2 into proteasomes, and inhibition of proteasome activity prevented ethanol-induced degradation of ␤-arrestin2, suggesting that ethanol degrades ␤-arres-tin2 via the proteasome-dependent pathway. Interestingly, we also observed increased co-localization of ␤-arrestin2 with the lysosomal marker LAMP1 only with high-dose (75 mM) ethanol treatment in N2A-5HT 1A R cells, suggesting that a subpopulation of ␤-arrestin2 can also be subject to lysosome-dependent degradation under specific conditions. ␤-Arrestin is capable of coupling to several E3 ligases, including MDM2, for induction of ␤-arrestin ubiquitination (16). We found that acute ethanol exposure increased the complex for-

Ethanol-induced ␤-arrestin2 ubiquitination and degradation
mation between ␤-arrestin2 and MDM2, but not NEDD4, in N2A-5HT 1A R cells and rat PFC, suggesting a unique involvement of MDM2 in regulation of enhanced ␤-arrestin2 ubiquitination and subsequent degradation induced by ethanol exposure. This notion is further substantiated by our observation that knockdown of MDM2 prevented ethanol-induced degradation of ␤-arrestin2 in N2A-5HT 1A R cells. It has been reported that phosphorylation of MDM2 or its substrates can enhance their interaction (41). Exposure to drugs of abuse, such as morphine, has been shown to increase ␤-arrestin2 phosphorylation (42). Therefore, future studies are necessary to determine whether ethanol exposure increases phosphorylation of ␤-arrestin2 and/or MDM2, which could be a potential mechanism for the increased association between these two proteins.
Given the important role of ␤-arrestin in regulating internalization of many GPCRs, the altered ␤-arrestin level by ethanol exposure would likely impact receptor internalization and function. In fact, we found that acute ethanol exposure blocked agonist-induced 5-HT 1A R internalization in a dose-dependent manner. It has been shown that 5-HT 1A R internalization requires ␤-arrestin as overexpression of a dominant-negative ␤-arrestin mutant, V53D, attenuates 5-HT 1A R internalization in HEK293 cells (32). Thus, the reduced ␤-arrestin level by ethanol exposure likely contributes to inhibition of 5-HT 1A R internalization. Although it has been reported that 5-HT 1A R internalization can increase or decrease following chronic ethanol exposure depending on ethanol dose, duration, and length of withdrawal (43,44), this is the first observation of acute ethanol exposure, which leads to inhibition of agonist-stimulated 5-HT 1A R internalization. Because recruitment of ␤-arrestin to receptors is a critical step in receptor internalization, ethanolinduced disruption in ␤-arrestin recruitment could also account for decreased 5-HT 1A R internalization. We found that ethanol treatment reduced and delayed ␤-arrestin2 recruitment to the membrane upon 5-HT 1A R stimulation. Both factors contribute to the abolished 5-HT 1A R internalization induced by agonist stimulation. These data may underlie altered behavioral responses to ethanol mediated by receptors such as 5-HT 1A R and mGluR5.
The diminution of ␤-arrestin2 localization at the membrane is due to a reduction in the total ␤-arrestin2 levels induced by ethanol exposure. The delayed recruitment of ␤-arrestin2 may have resulted from enhanced mono-ubiquitination of ␤-arres-tin2. Mono-ubiquitination disrupts protein trafficking through a number of distinct mechanisms, such as altering interactions between the substrate proteins and trafficking machinery or acting as a sorting signal to direct the movement of the substrate protein between different cellular compartments (45)(46)(47)(48). For example, mono-ubiquitination delays the recruitment of Jen1, a monocarboxylate transporter, to multivesicular bodies (47). Mono-ubiquitination was also shown to increase trafficking of the G␣ s subunit into Rab5-positive endosomes (49).

Ethanol-induced ␤-arrestin2 ubiquitination and degradation
Therefore, identification of the mono-ubiquitinated site and site-directed mutagenesis will help delineate the role of ␤-ar-restin2 mono-ubiquitination in regulation of ␤-arrestin2 trafficking. Additionally, it was reported that the expression level of ␤-arrestin regulates the rate and extent of ␤-arrestin recruitment to activated GPCRs. For instance, ␤2AR-stimulated ␤-ar-restin2 recruitment is significantly greater and displays a biphasic pattern of recruitment (fast recruitment followed by slow recruitment) in HEK293 cells with a high expression of ␤-arres-tin2 compared with cells with low ␤-arrestin2 expression (50). Therefore, reduced ␤-arrestin2 expression level after ethanol exposure may contribute to both delayed and reduced ␤-arres-tin2 membrane recruitment in response to 5-HT 1A R stimulation. Finally, there are other potential factors that may also be involved in ␤-arrestin2 membrane translocation, including changes in levels of GPCRs and other endocytic proteins, such as G-protein-coupled receptor kinases (GRKs) (33,51). However, we did not observe alterations in the total protein levels of HA-5HT 1A R or GRK2 in N2A-5HT 1A Rs cells treated with ethanol (15-75 mM) when compared with the vehicle treatment (Fig. S4). Furthermore, receptor phosphorylation is an important regulator of ␤-arrestin membrane recruitment (52); therefore, future studies should focus on whether ethanol exposure alters phosphorylation of 5-HT 1A R.
In summary, this study provides for the first time evidence that acute ethanol exposure abolishes 5-HT 1A R internalization through enhanced MDM2-dependent ␤-arrestin2 ubiquitination and proteasome-dependent degradation. These findings may be applicable to other GPCRs whose internalization is ␤-arrestindependent. These ␤-arrestin-mediated changes in GPCR activity may be relevant to behavioral desensitization and/or tolerance to ethanol exposure.

Animals
Male Sprague-Dawley rats (Envigo) were housed in pairs and maintained on a 12:12 h light/dark cycle (lights on at 9 p.m.) with food and water ad libitum. Animals were ϳ10 -11 weeks of age (300 -350 g) upon arrival. After 7 days of habituation, animals were exposed to ethanol or air as described previously (53,54). Briefly, animals were placed into air-tight Plexiglas chambers (Triad Plastics) in their home cages and were exposed to either ethanol vapor or room air during the light cycle for 12 h. Ethanol vapor, produced by submerging an air stone in 95% ethanol, was mixed with room air and was pumped into the chambers at a rate of 16 liters/min. Ethanol levels were maintained at ϳ35 mg/liter within the chamber. All animals were euthanized immediately following 12 h of ethanol or air exposure. Blood ethanol concentrations determined from the trunk blood were 242.3 Ϯ 21.3 mg/dl at the time of sacrifice. Blood ethanol concentrations were determined using a standard, commercially available ethyl alcohol enzymatic assay kit (Carolina Liquid Chemistries; catalog no. BL421). All animal care procedures were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the Wake Forest Animal Care and Use Committee.

Generation of N2A-5-HT 1A R stable cell line
Parental N2A cells (ATCC) were grown in Opti-MEM media supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. To generate a cell line stably expressing HA-5HT 1A R, N2A cells were transfected with human 5-HT 1A R cDNA with an HA-tag at the N terminus (HA-5-HT 1A R; Missouri S&T cDNA Resource Center) using Lipofectamine 2000 (Invitrogen) followed by an antibiotic selection with geneticin G418 (400 g/ml). The expression of HA-5-HT 1A Rs was confirmed by the mRNA level and the protein presence using quantitative PCR and immunocytochemistry, respectively.
N2A-5HT 1A R cells were subject to fluorescence-activated cell sorting (FACS) to isolate cells with a high expression of HA-tagged 5HT 1A R. Briefly, cells suspended in ice-cold PBS containing 3% BSA, at ϳ5 ϫ 10 6 viable cells/ml, were incubated with HA-tag DyLight488-conjugated antibody (Columbia Biosciences; D8 -1830) on ice for 1 h. Then cells were washed three times by centrifugation at 400 ϫ g for 5 min. Cells were resuspended in ice-cold PBS, 10% fetal bovine serum and sorted using a FACS Aria cell sorter (BD Biosciences). Live cells were gated based on forward/side scatter, and events in the top 3-4% of 488 nm fluorescence were selected, cultured, and expanded in medium containing geneticin (400 g/ml).

Acute ethanol treatment to N2A-5-HT 1A R cells
N2A-5HT 1A R cells were treated with ethanol (15, 30, or 75 mM) or vehicle (media) for 18 h. Ethanol-treated cell culture plates were placed in a closed but not sealed polystyrene box with an open dish containing the corresponding concentration of ethanol. As shown previously, this procedure results in stable ethanol levels for several days (55,56). The concentration of ethanol in the culture medium was determined using a colorimetric ethanol assay (Bioassay Systems; ECET100), which was performed at the end of the 18-h treatment period. Under these growing conditions, the average reduction in ethanol concentration after 18 h for each dose of ethanol was 3.13 Ϯ 0.85% compared with initial ethanol concentrations at the time of treatment (not significant, p ϭ 0.24 versus initial treatment, paired Student's t test, n ϭ 6).

Ethanol-induced ␤-arrestin2 ubiquitination and degradation
intensity of ␤-arrestin2 immunoreactive bands was normalized to ␤-actin and expressed as relative to the scrambled siRNA control.

MDM2 protein knockdown via siRNA
N2A-5HT 1A R cells were transiently transfected with a mouse MDM2 siRNA (Mm01_00139357) or a corresponding scrambled siRNA obtained from Sigma. Cells were lysed in solubilization buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 1% Triton X-100. Western blotting was performed to determine MDM2 levels using mouse anti-MDM2 antibody (Millipore; OP46) followed by a horse anti-mouse HRP-conjugated secondary antibody (Cell Signaling Technology, catalog no. 7076). Bands were visualized using chemiluminescence detection method (Pierce; catalog no. 32106). Quantification of the bands was performed using ImageJ (National Institutes of Health). MDM2 immunoreactive bands were normalized to ␤-actin (Santa Cruz Biotechnology) and expressed as relative to the scrambled siRNA control.

Confocal microscopy
Fluorescent images were acquired with Zeiss LSM 710 and 880 laser-scanning confocal microscopes. Confocal image planes were acquired with a ϫ63/NA 1.4 PlanApo oil-immersion objective. Alexa 405, Alexa 555, and Alexa 647 signals were excited at 405, 561, and 633 nm with diode-pumped solid-state (DPSS), and HeNe lasers, respectively. Fluorescent channels were acquired sequentially to prevent cross-excitation of laser signal and to minimize bleed-through. All images were acquired with identical settings for excitation intensity, detector sensitivity, and pinhole to allow intensity comparison between samples. Images for analysis were obtained by taking a z axis stack of image planes (1024 ϫ 1024 pixels) with 0.2-m steps encompassing the entire cell structure. Fluorescent intensity was quantitated for each channel using ImageJ software (National Institutes of Health). All images were adjusted with Gaussian blur for presentation purposes only. Alexa 405 emission was pseudocolored as green upon acquisition for optimal presentation appearance. All experiments were performed independently at least four times. For each group, ϳ25-35 cells were selected for analysis from each experiment.

Immunocytochemistry
Stimulated 5-HT 1A R internalization following treatment with ethanol (15-75 mM) or vehicle (media) for 18 h in N2A-5HT 1A R cells was measured using an antibody feeding technique, as described previously (34). N2A-5HT 1A R cells were washed with PBS and blocked with 6% normal horse serum in PBS at 4°C for 30 min. Surface 5-HT 1A Rs were then labeled with mouse anti-HA tag (Sigma; H6958) for 45 min at 4°C. Cells were warmed to 37°C and stimulated with 8-OH-DPAT (1 M) for 5, 15, 30, or 60 min. Internalization was terminated by washing cells with cold PBS and placing coverslips on ice. The remaining primary antibody-bound surface 5-HT 1A Rs were incubated with rabbit anti-mouse Alexa 647 to determine the pool of surface 5-HT 1A Rs that were not internalized. Next, cells were fixed with 4% formaldehyde for 15 min and permeabilized with 0.1% Triton X-100 for 10 min. Internalized 5-HT 1A Rs bound with HA-tag primary antibody were labeled with goat anti-mouse Alexa 405 for quantification of internalized 5-HT 1A Rs. Stimulated 5-HT 1A R internalization was calculated as the percent fluorescent intensity of Alexa 405 (internalized 5-HT 1A Rs) of the total (Alexa 647 ϩ Alexa 405) (total surface 5-HT 1A Rs before internalization). Data were presented as relative to the baseline level of 5-HT 1A R internalization in vehicle-treated cells.
We examined the co-localization of ␤-arrestin2, ubiquitin, and the proteasome marker P19S. Briefly, cells were pretreated with ethanol (15-75 mM) or vehicle (media) for 18 h, then washed with cold PBS prior to fixation with 4% formaldehyde for 15 min, and permeabilized with 0.1% Triton X-100 in blocking buffer containing 6% normal horse serum for 10 min. After blocking with 6% normal horse serum in PBS for 20 min, cells were incubated with the following antibodies: rabbit anti-␤-arrestin2 (LSBio; LS-B15546), sheep anti-ubiquitin (LSBio; LS-C76671), and goat anti-P19S (Novus Biologicals; NB100-1483) for 1 h followed by incubation with the secondary antibodies donkey anti-goat Alexa 633, donkey anti-sheep Alexa 555, and goat anti-rabbit Alexa 405, respectively. All fluorescent secondary antibodies are from Invitrogen. Coverslips were mounted using Prolong Gold mounting reagent (Invitrogen, P36930) and imaged via confocal microscopy. Images were processed, and spatial overlap was calculated using FIJI and Photoshop. To directly assess spatial overlap of these proteins, background was removed, and variable intensity information from each image was done by thresholding. This method allows each pixel to provide binary information for the presence or absence of each protein and to provide specific quantitation and display of positional overlap. The co-localization between ␤-arrestin2 (green) and ubiquitin (red) was first determined as yellow overlapped pixels. These co-localized pixels (yellow) were then isolated, and triple co-localization with P19S (blue) was determined as white overlapped pixels. Data were reported as a percentage of co-localization by overlapping pixel counts, y/x ϩ y ϩ z, (ubiquitin/␤-arrestin2 co-localization with respect to P19S).
To evaluate whether ethanol-induced ␤-arrestin2 degradation was proteasome-dependent, cells were incubated with or without MG132 (5 M), a proteasome inhibitor as shown previously (57), during treatment with ethanol (30 -75 mM) or vehicle for 18 h. Western blotting was performed to measure ␤-arrestin2 protein levels.
In an additional set of experiments, nitrocellulose membranes were cut according to molecular weight and Western-blotted with rabbit anti-NEDD4 (Cell Signaling Technology; catalog no. 3607S) and mouse anti-␤-arrestin2 (LSBio; LS-B6008).

Statistical analysis
GraphPad Prism 5 (La Jolla, CA) was used for statistical analysis. Data are presented as mean Ϯ S.E. A two-tailed unpaired t test was used for MDM2 protein levels after siRNA knockdown.
A one-way ANOVA followed by Bonferroni post hoc analysis was used for determining ␤-arrestin protein level and ubiquitination. A two-way ANOVA followed by Bonferroni post hoc analysis for multiple comparisons was used for all other experiments. A value of p Յ 0.05 was considered statistically significant.