Striatonigral direct pathway 2-arachidonoylglycerol contributes to ethanol effects on synaptic transmission and behavior

Endocannabinoids (eCB) and cannabinoid receptor 1 (CB1) play important roles in mediating short- and long-term synaptic plasticity in many brain regions involved in learning and memory, as well as the reinforcing effects of misused substances. Ethanol-induced plasticity and neuroadaptations predominantly occur in striatal direct pathway projecting medium spiny neurons (dMSNs). It is hypothesized that alterations in eCB neuromodulation may be involved. Recent work has implicated a role of eCB 2-arachidonoylglycerol (2-AG) in the rewarding effects of ethanol. However, there is insufficient research to answer which cellular subtype is responsible for mediating the 2-AG eCB signal that might be involved in the rewarding properties of ethanol and the mechanisms by which that occurs. To examine the role of dMSN mediated 2-AG signaling in ethanol related synaptic transmission and behaviors, we used conditional knockout mice in which the 2-AG-synthesizing enzyme diacylglycerol lipase α (DGLα) was deleted in dMSNs, DGLαD1-Cre+. Using acute brain slice photometry and a genetically encoded fluorescent eCB sensor, GRABeCB2.0, to assess real-time eCB mediated activity of sensorimotor inputs from primary motor cortices (M1/M2) to the dorsolateral striatum, we showed that DGLαD1-Cre+ mice had blunted evoked eCB-mediated presynaptic eCB signaling compared to littermate controls. Furthermore, ethanol induced eCB inhibition was significantly reduced in DGLαD1-Cre+ deficient mice. Additionally, there was a reduction in the duration of loss of righting reflex (LORR) to a high dose of ethanol in the DGLαD1-Cre+ mice compared to controls. These mice also showed a male-specific decrease in ethanol preference accompanied by an increase in ethanol-induced water consumption in a voluntary drinking paradigm. There were no significant differences observed in sucrose and quinine consumption between the genotypes. These findings reveal a novel role for dMSN mediated 2-AG signaling in modulating ethanol effects on presynaptic function and behavior.


INTRODUCTION
Alcohol use is widespread, with 219.2 million people aged 12 or older reporting use [1].Undoubtingly, alcohol misuse which can potentially result in Alcohol Use Disorder (AUD) is a public health concern.Despite this concern, much remains to be determined about the neurobiological underpinnings contributing to alcohol misuse and the pathological progression of AUD.

MATERIALS AND METHODS
Experiments were approved by the National Institute on Alcohol Abuse and Alcoholism Animal Care and Use Committee and were carried out in accordance with NIH guidelines.Cell-specific dMSN DGLα floxed mice (DGLα flx/flx ) were obtained from Vanderbilt University [41] and bred inhouse.Experiments were conducted during the light phase, unless otherwise stated.Animals were housed on a 12 h light-dark cycle with food and water ad libitum.Both sexes (9-18 wks old) were used for experiments.

Brain slice photometry
Photometric recordings were acquired from eCB2.0 and eCBmut GFPexpressing brain hemisections.Slices were perfused with oxygenated aCSF at 1-2 ml/min at 30-32 °C.A twisted bipolar stimulating electrode (Plastics One) was placed in the DLS outside the fluorescence excitation window (180µm 2 ).Single pulse stimuli (0.6 mA, 2 ms) were generated using a current stimulator (DS3, Digitimer Ltd, Hertfordshire, UK).Short trains of stimuli (5 pulses (Ps), 5 Hz; 10Ps, 10 Hz; 20Ps, 20 Hz; 50Ps, 50 Hz; 100Ps, 100 Hz) were generated by a Master 8 stimulator (A.M.P.I., Jerusalem, Israel).Fluorescence emission was collected using a 40x/0.8NAwater immersion objective on an upright fixed stage Olympus BX41 microscope (Olympus Life Science Solutions, Waltham, MA, USA).Fluorescence from the GRAB eCB2.0 or mut sensor containing a circular permuted GFP moiety was excited using 470 nm light emitted from an LED (Thorlabs, Newton, NY, USA) every 3 min in 1 min and 20 s intervals.A shutter (uniblitz), triggered by pCLAMP software (Molecular Devices, CA, USA 11.03) was used to minimize exposure time and photobleaching.Emitted light was filtered with a FITC filter set (Ex 470/40, FT 495, Em 525/50) and directed to a photomultiplier tube (PMT; Model D-104, Photon Technology International, Edison, NJ, USA).The PMT voltage output (time constant: 5 ms; gain: 1 µA) was filtered at 1 kHz and digitized at 10 kHz using Digidata 1500B (Molecular Devices) and Clampex software (Axon instruments).The data were analyzed using pCLAMP software.Evoked eCB transients were measured as the ratio of the evoked peak fluorescence amplitude of the eCB2.0 or eCBmut transient (ΔF) to the average baseline (5 sweeps, 15 mins) fluorescence value (F) before stimulation, generating a ΔF/F value.The data are expressed as percent change of ΔF/F relative to baseline.

Loss of the righting-reflex (LORR)
To assess the sedative effect of ethanol by measuring the latency and duration of LORR, mice were administered 3.5 g/kg ethanol intraperitoneally (i.p.).LORR latency was defined as the time to lose righting-reflex, which was determined as the time from injection to the inability of the mouse to right itself within 30 s of being placed in a supine position in a V-shaped plexiglass trough.The duration of LORR was defined as the time between losing and regaining the righting-reflex.Mice were determined to have regained their reflex if they were able to right themselves 3 times within 30 s.

Two-bottle choice 16 h brief ethanol intermittent access (bIA)
Mice were singly housed and allowed to habituate to the two-bottle choice cages for 8-12 days for ethanol intake assessment.During this period, both drinking bottles contained water.For the two-bottle choice ethanol bIA (3 times a week, every-other-day), the home cage water bottles were replaced with one bottle containing ethanol and the other water 1 h prior to lights-off.To prevent side preferences, the position of the bottles was rotated.Mice were given access to increasing concentrations of ethanol (5%, 10% & 20% v/v with water) successively on a weekly basis (1wk /concentration) for a total of 3 weeks.Bottles containing ethanol and water were weighed before and after drinking sessions.On non-ethanol drinking days, mice had free access to water.Differences in weights of the bottles were converted to grams of ethanol consumed per kilogram of body weight.The preference for ethanol was determined as the volume of ethanol consumed over the total fluid (water and ethanol) consumed.Data were collected from two separate cohorts of mice.

Ethanol challenge injection
Mice singly housed with prior exposure to ethanol (LORR testing) were given an i.p. injection of 3.5 g/kg ethanol an hour before the dark cycle.For consistency, we used the same dose of ethanol used for LORR testing.The amount of water consumed after 16 hrs of overnight drinking was measured and the grams of water consumed per kilogram of body weight calculated.

Sucrose & quinine consumption
Mice previously exposed to ethanol were given 16 h access to either a bottle containing sucrose or quinine solution for 6 consecutive days each accompanied with a bottle containing water to determine taste preference.Mice were tested on three ascending concentrations of sucrose, 0.5%, 2%, and 10%, as well as three quinine concentrations, 0.03 mM, 0.10 mM, and 1 mM for 2 days each.Sucrose consumption was measured 2 weeks prior to quinine consumption.Solution preference was determined as the ratio of a given solution intake relative to the total fluid intake.Ratios were averaged across the 2 days.

Water restriction
Following quinine consumption, singly housed mice were habituated to a single water bottle cage for one week.The water bottle was removed overnight and returned after 16 h of water deprivation.The water bottle was weighed after 8 h to determine consumption.Louis, MO), and tetrodotoxin (TTX, Biotium, Fremont, CA) were diluted to their final concentration in aCSF for acute slice recordings.For behavioral experiments, alcohol, sucrose (Sigma, St. Louis, MO), and quinine (Alfa Aesar, Tewksbury, MA) were diluted in water.

Statistical analyses
Data were analyzed using unpaired two-tailed t test and a one-, two-, or three-way ANOVA with repeated measures (rmANOVA) in GraphPad Prism.For alcohol intake, data were analyzed by three-way rmANOVA in SPSS.Data are reported as mean ± SEM.
To determine if increased water intake was driven simply by ethanol exposure, a subset of mice was given a challenge dose of ethanol (3.5 g/kg), and water intake was measured for 16 h (Fig. 4j).There was no significant difference in water intake post ethanol injection in both males and females DGLα D1-Cre+ mice compared to DGLα flx/flx mice (Fig. 4k; n = 5-6 males/group; unpaired two-tailed t test: t 9 = 0.4828, p > 0.05; n = 5 females/ group; unpaired two-tailed t test: t 8 = 0.115, p > 0.05).Furthermore, to determine if water restriction would alter water consumption differently in the DGLα D1-Cre+ mice, we water restricted animals overnight and measured water consumption the following day (Fig. 4l).Water intake was similar in both males and females following water restriction (Fig. 4m; n = 6 males/ group; unpaired two-tailed t test: t 10 = 1.019, p > 0.05; n = 4-5 females/group; unpaired two-tailed t test: t 7 = 1.356, p > 0.05).These findings suggest that the sex-specific water intake increase during two-bottle drinking in the DGLα D1-Cre+ mice is context specific, with voluntary ethanol taking but not passive administration (i.p.) of ethanol resulting in increased water consumption.

DISCUSSION
We assessed the effects of targeted deletion of DGLα, the 2-AG biosynthesis enzyme, in striatal dMSNs on presynaptic eCB signaling and ethanol effects on this signaling and behavior.We first used the GRAB eCB2.0 biosensor to demonstrate that real-time eCB mobilization that affects corticostriatal presynaptic terminals is reduced in dMSN DGLα KO mouse striatum in response to single and burst stimuli.These findings support a role for dMSN 2-AG signaling and are consistent with previous findings that such stimulation induces mainly 2-AG mobilization from the entire MSN population that affects corticostriatal synapses [42].Interestingly, the decrease in eCB signaling in the dMSN DGLα KO mouse slices appears to be greater than might be expected if both d-and iMSNs contribute equally to stimulus-induced 2-AG production.However, the contribution of other neuronal eCB sources, as well as residual signaling from other 2-AG biosynthesis pathways or eCBs, such as AEA, remains unclear [49].It is also important to mention that 2-AG levels are reduced in the dMSN DGLα KO mice [41].Both 2-AG and AEA have been implicated in striatal plasticity, particularly LTD [30,[50][51][52][53].However, most of the evidence suggests a role of AEA in LTD [30,50,51].Thus, our findings and those of Liput and coworkers [42], are surprising.However, LTD is generally induced by stimuli with frequencies and durations greater than those used in the present study and by Liput and coworkers [42].The corticostriatal afferents are an important site of eCB signaling in the striatum [3,[54][55][56][57][58], and thus this role for dMSN-derived 2-AG has important implications for striatal function and related behaviors.However, it is worth mentioning that eCBs modulate synaptic functions at GABAergic striatal synapses which may be even more sensitive than corticostriatal synapses [59][60][61][62].
Our findings also demonstrate that ethanol induces transient alterations in 2-AG mobilization at corticostriatal afferents.Furthermore, we showed that the selective deletion of DGLα from dMSNs resulted in dampening of ethanol effects on eCB signaling at these terminals.These findings indicate that acute ethanol exposure dampens retrograde eCB signaling involved in synaptic depression, most likely via inhibition of eCB production or release.Endocannabinoid modulation is thought to contribute to compulsive drug seeking and taking behaviors, including those induced by ethanol [9,63,64].Endocannabinoid subtypes like 2-AG might have differential roles in different brain regions and reward-driven behaviors [3,39].The present data show that acute ethanol exposure reduces striatal 2-AG release/CB1 receptor activation in the dorsolateral striatum at corticostriatal terminals.Previous reports have shown that 2-AG levels increase in the nucleus accumbens during ethanol self-administration in vivo [39].Our data do not contradict previous findings but highlight and further support the hypothesis that eCB subtypes may mediate different signaling profiles in different brain areas.It will be interesting to determine the contribution of ethanol inhibition of 2-AG signaling at specific corticostriatal synapses to ethanolrelated behaviors.Also, more research is needed to determine whether ethanol induces differential eCB signaling in response to acute and repeated drug exposure.
Although the newly developed eCB sensor has different affinities for 2-AG and AEA, the sensor can detect both eCBs [43].Given that the sensor has a lower affinity for 2-AG, it would be difficult to detect very low levels of 2-AG signaling.However, our work confirms previous findings from our group that showed eCB transients evoked by brief stimuli detected at these corticostriatal afferents arising from primary motor cortex are primarily 2-AG mediated [42].Together, our findings further support our hypothesis that 2-AG and AEA may have differential functions in different brain regions at different synapses that may be dependent on sensory input and/or external stimuli (e.g., electrical stimulation) [3].In the next generation of eCB sensors, new variants with improved dynamic range and eCB type specificity are needed to fully understand the contributions of eCB subtypes to synaptic signaling and behavior.
The eCB system is altered by acute and chronic ethanol exposure [22,65].Although the CB1 receptor role in ethanol's rewarding properties has been extensively studied, the roles of 2-AG and AEA in these processes are not fully understood [26,27,40,[66][67][68].Ethanol self-administration increases striatal 2-AG levels, without accompanying changes in AEA levels [39].These findings suggest that 2-AG may play a significant role in the reinforcing effects of ethanol.This is further supported by a reduction in ethanol consumption in global 2-AG deficient mice [40].The present data reveal that ethanol inhibits 2-AG signaling predominantly at terminals innervating dMSNs in the dorsolateral striatum, as shown in Fig. 2. CB1 receptor-2-AG activation has the potential to reduce synaptic activation of dMSNs.This in turn, can reduce striatal GABAergic transmission in output brain regions, such as the substantia nigra pars compacta (SNc) and pars reticulata (SNr).For example, decreased inhibitory transmission in the SNc may result in the disinhibition of dopamine neurons, such that dMSN-mediated 2-AG-CB1 receptor activation can contribute to ethanol consumption by enabling striatal dopamine release [69,70].Alternatively, the loss of ethanol-induced eCB modulation of dMSN synapses may result in the enhancement of dMSN roles in behavior through the disinhibition of dorsolateral striatal synaptic activity.This hypothesis is supported by previous work that showed a strengthening of dMSN function after ethanol exposure [35].
Previous studies suggest that chronic in vivo exposure to misused substances, including alcohol, results in predominate changes in synaptic transmission and anatomical properties involving dMSNs [35,65].Our findings suggest that most of the ethanol-induced 2-AG inhibition is mediated by dMSNs (Fig. 2).However, this study cannot unequivocally rule out the contributions of iMSN 2-AG signaling to the ethanol induced inhibition observed in Fig. 2. Therefore, more research on the role of iMSN 2-AG signaling in ethanol effects on transmission and behavior is needed to further understand the cell-specific mechanisms involved in the effects of ethanol on the eCB system.
Ethanol induced alteration in striatal 2-AG signaling in the dMSN DGLα KO mice is accompanied by a reduction in the sedative/hypnotic effects of ethanol (Fig. 3).This finding could suggest either a change in sensitivity to ethanol or a change in ethanol metabolism that results in faster reductions in blood ethanol concentration leading to faster regain of righting reflex in the dMSN DGLα KO mice.A metabolic change seems unlikely given the neuronal targeting of the enzyme knockout.Nonetheless, future experiments examining blood ethanol concentration can assess possible changes in metabolism.This finding differs from previous experiments in the global CB1 knockout mouse indicating increased ethanol-induced sedation [32] an effect that likely involves synapses other than the corticostriatal inputs to dMSNs.Genetic background can also influence ethanol-induced behavioral responses, such as sedation [71].
Additionally, the eCB system has been implicated in the development of habitual behaviors, as well as the voluntary taking of ethanol [12,72,73].It was previously unclear whether ethanol can induce differential eCB responses involved in different ethanol-related behaviors.Our findings extend a previous report that indicates involvement of 2-AG signaling in voluntary ethanol consumption [40].Deletion of DGLα from dMSNs decreased ethanol preference in a voluntary intake paradigm in males but not females.This change was not accompanied by a decrease in ethanol intake, but rather an increase in water consumption and total fluid intake in males.Ethanol intake varied depending on the day/session in which the ethanol concentration was administered.However, the day-dependent changes in ethanol consumption accounts for ~5% of the variance in the data.It is possible that loss of 2-AG signaling at cortical inputs to dMSNs increases fluid intake, but this effect is counteracted in the presence of ethanol due to taste or the neural effects of the drug.However, increased water intake was not seen with ethanol exposure via intraperitoneal injection, nor when ethanol was exchanged for sucrose or quinine.Thus, we find no evidence that ethanol intoxication per se or the presence of preferred or non-preferred tastants drives increased water intake in the dMSN DGLα KO mice.Thus, there is something specific about voluntary ethanol ingestion that drives up water intake in these mice, perhaps due to the unique taste of ethanol or presence of mild neural effects of ethanol that induce greater thirst and cannot be mimicked by intraperitoneal drug injection.Although our findings suggest that ethanol exposure and 2-AG signaling may promote and/or facilitate ethanol preference via eCB depression of transmission onto dMSNs, other paradigms of drinking, such as ethanol-primed and cue-induced ethanolseeking behavior should be investigated to further evaluate possible 2-AG eCB effects on voluntary ethanol taking behaviors.
Although our cell-type specific deletion of 2-AG signaling in corticostriatal synapses onto dMSNs in DLS may suggest a role for this signaling in the development of habitual behavior, this is likely not the only mechanism by which ethanol exerts its function.Undoubtedly, 2-AG signaling at other synapses within different brain regions may contribute to alcohol-related behaviors.Nucleus accumbens 2-AG levels are increased in vivo during ethanol selfadministration [39].Therefore, dMSN 2-AG signaling deficits in this area likely also contribute to ethanol-related behaviors.Furthermore, prefrontal cortex has been implicated in the modulation of ethanol seeking and taking behaviors [74][75][76][77].It is likely that 2-AG signaling deficits in the cortex may alter glutamatergic output in subcortical areas, thereby altering the motivational salience of ethanol [78].Drug salience is important for the establishment of drug dependence [79].For example, if cortical drive is diminished due to deficits in 2-AG signaling, this may result in the perceived decrease in ethanol's rewarding effects.
Our work has highlighted that eCB signaling mechanisms at specific synapses and release from specific neurons can have profound signaling and behavioral effects produced by ethanol.Moreover, our findings provide a new potential therapeutic target research avenue of exploration, involving dMSN-mediated 2-AG signaling and direct pathway function for treating AUD.

Fig. 1
Fig. 1 GRAB eCB2.0 detects 2-AG signaling deficits in dMSN DGLα KO mice.a Schematic of breeding strategy.b Diagram showing viral eCB2.0 injections into M1 and recording site in the DLS.c eCB2.0 immunoreactivity in M1 somata and DLS inputs.d Top.Representative traces of normalized eCB transient evoked by a single electrical pulse.Bottom.Averaged peak amplitude.e Representative traces of eCB transients evoked by a train of electrical stimulation.f Averaged Peak amplitude of eCB transients as a function of stimulation frequency and number of pulses.g Representative traces and h averaged peak amplitude of eCB transients evoked by train electrical stimulation after preincubation with D034.i Representative traces before and after CB1 receptor antagonist, AM251.j Evoked eCB transients in the presence of AM251.k Representative traces before and after bath application of sodium channel blocker, TTX.l Evoked eCB transients in the presence of TTX.Data shown as the mean ± SEM. *p < 0.05, **p < 0.01.

Fig. 3
Fig. 3 Deletion of striatal dMSN 2-AG signaling reduces duration of loss of righting reflex (LORR).a Experimental schematic and pictures of LORR testing paradigm.b Latency to lose (c) and duration of ethanol induced LORR.Data from individual animals are overlaid as circles, with females in pink and males in blue.DGLα flx/flx in gray and DGLα D1 Cre+ knockout mice in red.All error bars represent SEM.t test, *p < 0.05.

Fig. 4 2 -
Fig. 4 2-AG dMSN signaling deletion results in decreased ethanol preference and increased water consumption in males.Individual data points are overlaid as circles, with DGLα flx/flx in gray and DGLα D1 Cre+ in red.a Schematic of 16 h intermittent ethanol 2-bottle choice drinking paradigm.Ethanol intake in males (b) and females (c) across 16 h of intermittent drinking.Preference for ethanol, consumption of water, and total fluids in males (d-f) and females (g-i) in DGLα flx/flx and DGLα D1 Cre+ mice.Littermate controls, DGLα flx/flx , represented as gray bars and DGLα D1 Cre+ mice represented as red bars.Individual data points for females shown as pink circles and males as blue circles for each genotype.j Schematic of experimental procedure.k Water consumption following a challenge injection of ethanol.l Schematic of water restriction experimental procedure.m Water consumption following 16 h of water restriction.Data presented as mean ± SEM. ***p < 0.001.

Fig. 5
Fig. 5 DGLα D1 Cre+ knockout mice display similar taste preference and aversion for sweet-and bitter-flavored solutions.Summary bar graph depicting DGLα flx/flx mice in gray and DGLα D1 Cre+ mice in red.Individual differences between males (blue circles) and female (pink circles) are shown.a Sucrose consumption, (b) preference, (c) water consumption, and (d) total fluid consumption during two bottle choice testing for increasing concentration of sucrose solutions.e Quinine consumption, (f) preference, (g) water consumption, and (h) total fluid consumption during quinine two bottle choice testing.All data reported as mean ± SEM.