Activation of Locus Coeruleus to Rostromedial Tegmental Nucleus (RMTg) Noradrenergic Pathway Blunts Binge-Like Ethanol Drinking and Induces Aversive Responses in Mice

Abstract

There is strong evidence that ethanol entails aversive effects that can act as a deterrent to overconsumption. We have found that in doses that support the development of a conditioned taste aversion ethanol increases the activity of tyrosine hydroxylase (TH) positive neurons in the locus coeruleus (LC), a primary source of norepinephrine (NE). Using cre-inducible AAV5-ChR2 viruses in TH-ires-cre mice we found that the LC provides NE projections that innervate the rostromedial tegmental nucleus (RMTg), a brain region that has been implicated in the aversive properties of drugs. Because the neurocircuitry underlying the aversive effects of ethanol is poorly understood, we characterized the role of the LC to RMTg circuit in modulating aversive unconditioned responses and binge-like ethanol intake. Here, both male and female TH-ires-cre mice were cannulated in the RMTg and injected in the LC with rAVV viruses that encode for a Gq-expressing designer receptor exclusively activated by designer drugs (DREADDs) virus, or its control virus, to directly control the activity of NE neurons. A Latin Square paradigm was used to analyze both 20% ethanol and 3% sucrose consumption using the “drinking-in-the-dark” (DID) paradigm. Chemogenetic activation of the LC to RMTg pathway significantly blunted the binge-ethanol drinking, with no effect on the sucrose consumption, increased the emission of mid-frequency vocalizations and induced malaise-like behaviors in mice. The present findings indicate an important involvement of the LC to RMTg pathway in reducing ethanol consumption, and characterize unconditioned aversive reactions induced by activation of this noradrenergic pathway.

1. Introduction

According to the Centers for Disease Control and Prevention (CDC), despite being preventable, binge drinking is the most common and lethal pattern of excessive alcohol consumption in the United States (Esser et al., 2014; Sacks et al., 2015; Stahre et al., 2014); with men being approximately two times more likely to binge drink than women (Kanny et al., 2018). A ‘binge’ is defined by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) as a pattern of drinking that produces blood ethanol concentrations (BECs) greater than 0.08% (NIAAA, 2004). Binge-like ethanol consumption stimulates neuroplastic changes that may reflect the initial stages of a transition to a dependence-like state (Cox et al., 2013; Lowery-Gionta et al., 2012a; Sparrow et al., 2012), thus it is imperative to discover the central mechanisms that modulates this pattern of excessive ethanol drinking and how changes in these mechanisms drive repeated binge episodes. The “drinking-in-the-dark” (DID) paradigm is used for modeling voluntary binge-like ethanol intake in non-dependent rodents (Bell et al., 2011; Holgate et al., 2017; Rhodes et al., 2005; Thiele & Navarro, 2014). In this model, animals consume large amounts of ethanol and present BECs exceeding 80 mg/dl, which enables the investigation for neuronal circuitries and neurochemical systems that modulate binge-like ethanol intake (Sprow & Thiele, 2012).

Although numerous studies have been focused on the reinforcing effects of ethanol and how these effects lead to binge-like ethanol drinking (Sprow & Thiele, 2012), ethanol also comprises aversive properties that can act as a deterrent to overconsumption. In humans, genetics present an important contribution to the sensitivity to the aversive effects of ethanol (Eng et al., 2007; Ransome et al., 2017; Schuckit, 1986a, 1986b), where increased sensitivity to the aversive properties of ethanol may be a factor that reduces the risk for alcohol use disorders (AUDs). In animals, studies comparing different rodent strains have revealed a relationship between aversive reactions and the tendency to ingest alcohol. For example, mice selectively bred to achieve high BECs while binge drinking (the HDID mice) exhibit reduced sensitivity to the aversive properties of ethanol without alterations in sensitivity to ethanol’s reinforcing properties (Barkley-Levenson et al., 2015). Also, high drinking ethanol strains necessitate elevated doses of ethanol to develop a conditioned taste aversion (CTA) to ethanol (Fernandez et al., 2017; Froehlich et al., 1988; Kulkosky et al., 1995; Risinger & Cunningham, 1992). While evidence suggests that sensitivity to the aversive effects of ethanol modulates ethanol intake (Riley, 2011), the neurocircuitry that modulates the aversive responses to ethanol remain poorly understood.

It is known that the noradrenergic system is involved in both rewarding and aversive states that drive ethanol consumption (for review see (Koob, 2014)) and presents an important impact on ethanol seeking and use (for review see (Vazey et al., 2018)). We previously found that acute ethanol induced greater c-Fos activation of the locus coeruleus (LC) in selectively bred low ethanol drinking NP and ANA rats than in high ethanol drinking P and AA rats (Thiele et al., 1997). Similarly, we recently found that iHDID1 mice, an inbred strain created from a line that was selectively bred for achieve high BECs using DID procedures (Barkley-Levenson & Crabbe, 2014), exhibited lower ethanol-induced c-Fos expression in the LC relative to the HS/Npt control line following doses of ethanol that supported CTAs (Robinson et al., 2020). These observations are consistent with data showing that iHDID1 mice exhibit blunted ethanol-induced CTAs relative to HS/Npt mice, suggesting reduced sensitivity to the aversive properties of ethanol (Crabbe et al., 2019; Robinson et al., 2020). Taken together, these findings suggest that LC activation is protective against ethanol ingestion, perhaps by modulating the aversive reactions to ethanol. The LC is an important source of norepinephrine (NE) (Dahlstrom & Fuxe, 1964; Felten & Sladek, 1983) and sends projections to numerous brain structures that modulate motivated behaviors, such as the bed nucleus of the stria terminalis (BNST), amygdala and substantia nigra (Moore & Bloom, 1979). Interestingly, we have recently discovered in preliminary data that the LC provides NE projections that innervate the rostromedial tegmental nucleus (RMTg), a region that has been implicated in modulating aversive responses and the aversive properties of drugs ((Fu, Chen, et al., 2016; Fu, Zuo, et al., 2016; Haack et al., 2014). The RMTg inhibits the activity of dopaminergic neurons in the ventral tegmental area (VTA) via GABAergic innervation (Kaufling et al., 2009), and it is involved in modulating ethanol-induced CTA (Glover et al., 2016) as well as ethanol intake (Fu, Zuo, et al., 2016; Sheth et al., 2016). Thus, the noradrenergic LC to RMTg pathway may serve to blunt ethanol intake by modulating the aversive effects of ethanol.

In the present study, we used chemogenetic approaches in tandem with TH-ires-Cre mice to interrogate two main questions: The first was to assess the role of the NE+ LC → RMTg pathway in the modulation of binge-like ethanol intake using DID procedures, and the second objective was to clarify whether the NE+ LC → RMTg pathway support unconditioned aversive behaviors. It is known that the emission of vocalizations by rodents can reflect the emotional and internal states of those animals (Brudzynski, 2013; Niemczura et al., 2020; Williams et al., 2012), and thus vocalizations can serve as valuable tools to assess unconditioned behavioral parameters. In mice, the recently identified “mid-frequency vocalization” was found to be emitted by mice undergoing stress (Grimsley et al., 2016). Additionally, spontaneous behavioral alterations such as “lying-on-belly” induced by lithium chloride in rats (Navarro & Cubero, 2003) and grooming and head shake in mice (Vuralli et al., 2019) can be used as indicators of the induction of aversion or malaise produced by experimental manipulations. Here, we examined vocalization and aversive-like behaviors by mice to analyze whether activaton of the TH+ LC→RMTg pathway would support unconditioned aversive responses. We hypothesize that the TH+ LC→RMTg pathway activation would decrease the emission of ultrasonic vocalizations (USVs) and increase the mid-frequency vocalizations and increase aversive-like behaviors by the animals. Our results confirm that the activation of this noradrenergic pathway significantly decreases binge-like ethanol consumption without altering sucrose intake and also induced unconditioned behaviors thought to reflect the induction of an aversive- or malaise-like state.

2. Methods

2.1. Animals

Male (total n=23) and female (total n=23) TH-ires-Cre transgenic mice (with Cre-recombinase expression under the tyrosine hydroxylase (TH) promoter) (Savitt et al., 2005) on a C57BL/6J background bred in-house were used in all experiments described here. The animals were group housed 2–5 per cage until weaning with a room temperature maintained at 22°C and a 12 hour forward light/dark cycle. After weaning, animals were maintained at 22°C and a 12 hour reverse light/dark cycle starting at 7 am. Ad libitum access to Prolab^® RMH 3000 (Purina labDiet^®; St. Louis, MO) chow and water were available. For the vocalization experiment, due to the importance of socialization during the initial stages of life on the vocal repertoire development (Chabout et al., 2012; Grimsley et al., 2011), mice were group housed from weaning until the surgery day and the siblings confirmed as negatives to the Cre expression (TH-Cre−) were used as control mice while mice with positive Cre expression (TH-Cre+) served as the experimental group. All experiments were performed following National Institute of Health procedures and in accordance with approved guidelines by the Institutional Animal Care and Use Committee at the University of North Carolina at Chapel Hill (IACUC approval number: 18–096).

2.2. Surgery

Mice between 70 and 90 days old received bilateral micro infusions of either a Cre-dependent Gq-coupled Designer Drug Exclusive Activated by Designer Drug (DREADD) vector (pAAV8-hSyn-DIO-hM3D(Gq)-mCherry, Addgene, Watertown, MA) or a Cre-dependent control vector (AAV8-hSyn-DIO-mCherry, Addgene, Watertown, MA) in the LC (AP: −5.40, ML: −/+0.90, DV:−3.80), and had bilaterally angled (20°) cannula guides (26 GA cannula, 6mm cut bellow pedestal, C315GS-5/SP, Plastics One, Roanoke, VA) placed above the RMTg (AP:−3.80, ML:+1.90 – with 20° angle, DV:−4.70). Animals recovered for at least 21 days after surgeries to allow the maximum expression of the viral vectors (Krashes et al., 2011; Sasaki et al., 2011). Mice were also habituated to handling and connection to tubing for 2 consecutive days prior to drinking and behavioral testing. For the tracing study, an AAV5-ChR2-DIO-mCherry (UNC Vector Core, Chapel Hill, NC) virus was injected into the LC and five weeks later mice were perfused and brains were processed for immunofluorescence.

2.3. Behavioral experiments

2.3.1. Drinking-in-the-dark procedures

The binge-like ethanol drinking was performed using 4-day DID procedures with access to 20% ethanol beginning 3 hours into the dark cycle (for review see (Thiele & Navarro, 2014)). Ethanol was offered with a 10 ml sipper tube (Thiele et al., 2014) for 2 hours. On the test day (day 4 of DID), 30 minutes before the drinking session, mice were given site-directed injection of either Clozapine-N-oxide (CNO, 900 pmol/0.3 μl/side) dissolved in DMSO (1% v/v) and diluted with 0.9% saline or vehicle (same volume of DMSO diluted in 0.9% saline). Microinjections of CNO and vehicle were infused at a rate of 0.1 μl/min using an injector (P1 Technologies, injector code #8IC315IS5SPC C315IS-5/SPC internal 33GA, fit 6mm C315GS-5/SP with 0.5mm of projection, P1 technologies, Roanoke, VA, USA) connected to a Hamilton syringe (Hamilton Company, code 86200, 22s/2.75”/3, Reno, NV, USA) through a tubbing system (Becton Dickinson and Company, code 427411, ID .58mm, OD .965mm, Sparks, MD, USA), and attached to a Harvard Apparatus PHD 2000 infusion pump (Holliston, MS). The injectors remained in the cannula guide for 1 min after infusion to minimize back flow at their removal (for infusion method see (Rinker et al., 2016). Tail blood samples were collected from each animal immediately after ethanol access, and BECs from mice were assessed via an alcohol analyzer (Analox Instruments, Lunenburg, MA). At the end of each test day, animals underwent 3 days with just water ad libitum, with the total of 7 days completing one drinking cycle. The drinking experiment used a Latin Square design such that all mice were treated with CNO or vehicle over 2 DID cycles. Following assessment of ethanol intake mice were also tested to assess 3% (w/v) sucrose intake after CNO or vehicle administration using a Latin Square design and the same DID procedures noted above.

2.3.2. Emission of vocalizations after LC to RMTg pathway activation

2.3.2.1. Experimental design

Male and female mice were used in the vocalization experiment, but just results from males were analyzed in the present study. It is clear in the literature that there are sex differences in stress responses and anxiety levels that can interfere in the emission of vocalizations (Niemczura et al., 2020). Before each habituation and test session, animals were habituated over 30 minutes to the experimental room under red lights illumination. All the manipulations started 3 hours into the dark cycle to be consistent with DID procedures. On the experimental day, each animal was placed into a clean home cage with low-noise bedding (Tek-Fresh, Envigo, Indianapolis, IN, USA) and covered by a stainless wired top and habituated for 5 minutes before site-directed microinjection. In this experiment TH-Cre− mice served as the control group as they do not express active Cre-dependent DREADD, and TH-Cre+ mice served as the experimental group. Thus, all mice were given CNO on test day. All vocalization trials were recorded using a video camera in order to assess whether certain sounds were a product of the mouse interacting with the cage. Tek-Fresh bedding was used as opposed to corn cob bedding during the vocalization recordings to minimize noise. The bedding was not completely removed because evidence suggest that complete elimination of bedding confounds mouse vocalizations (Chabout et al., 2012; Granon et al., 2003). The animals were returned to their original bedding after being recorded. Between each recording trial, each cage was cleaned using 20% (v/v) ethanol cleaning solution and the bedding was changed.

2.3.2.2. Recording parameters

Emission of vocalizations was monitored by an UltraSoundGate Condenser Microphone CM16/CMPA with sensitive range of 2–200 kHz (flat frequency response; ±6 dB; Avisoft Bioacoustics, Berlin, Germany) attached to the roof of the sound attenuating box (Med Associated Inc., St Albans, VT, USA), 20 cm above the floor. It was connected via an UltraSoundGate 116H audio device (Avisoft Bioacoustics, Berlin, Germany) to a computer, where acoustic data were recorded using a sampling rate of 250 kHz, 16-bit format, FFT-length: 512 points, and a recording range of 0–125 kHz by Avisoft RECORDER (version 4.2, Avisoft Bioacoustics, Berlin, Germany).

2.3.2.3. Analysis of vocal data

For offline acoustical analysis, recordings were transferred to Avisoft SASLab Pro (version 5.20, Avisoft Bioacoustics, Berlin, Germany) and a fast Fourier transform was conducted (512 FFT length, 100% frame, Hamming window, and 75% time window overlap), resulting in spectrograms with 488 Hz of frequency and 0.512 ms of time resolution. Automatic detection programs were not suitable for analyzing vocal data in this study due to background noise which is more severe at low frequencies. Two trained researchers blinded to experimental conditions classified vocalizations manually, using audio playback in lowered sampling rate to distinguish the sounds of animal movement from vocalizations. Vocalizations with a fundamental frequency below 8 kHz were classified as low frequency vocalizations (LFV), or low frequency harmonics (LFH) for those with multiple harmonics. Low frequency harmonics can often contain harmonics that proceed well into the ultrasonic range (Grimsley et al., 2013). Vocalizations with a fundamental frequency between 8–20 kHz were classified as mid-frequency vocalizations (MFV) as previously identified and described by Grimsley and colleagues (Grimsley et al., 2016). Mid-frequency vocalizations often have no harmonics, but MFV vocalizations tend to have more harmonics when they are longer in duration (Grimsley et al., 2016). Noisy vocalizations show a noisy-harmonic chaotic structure (Grimsley et al., 2011). Vocalizations above 20 kHz were classified as ultrasonic vocalizations (USVs) (Chabout et al., 2012). For this study, a bout of vocalizations was defined as a series of three or more vocalizations with less than 1.6 seconds of silence between them. The total numbers of the subtypes of vocalizations were calculated over a 30-minute session to visualize the vocal repertoire response to the DREADDs activation.

2.3.3. Open-field testing (OFT)

The same animals used in the emission of vocalization study had their locomotor activity and anxiety-like behavior assessed in the OFT following TH+ LC → RMTg pathway activation. Starting 3 hours into the dark cycle, animals were placed in the experimental room and allowed to habituate under red lights during 30 min before the experiment. Animals received a site-directed injection of CNO (as described in the DID procedure session) and immediately positioned in the center of the testing chamber (42 cm × 42 cm × 30cm; catalog #71-SFAC; Omnitech Electronics, Inc., Columbus, OH). During a 2-h period, animals’ movements were tracked using VersaMax (AccuScan Instruments, Inc., Columbus, OH) and quantified using VersaDat Version 4.00 (AccuScan Instruments, Inc.). Total distance traveled and time spent in center of the test chamber were considered to assess the anxiety-like behavior (P.Simon, 1994).

2.3.4. Behavioral responses to the LC to RMTg pathway activation

Unconditioned behavioral responses to activation of the TH+ LC → RMTg pathway were assessed in a subset of TH-ires-Cre positive mice expressing hM3D(Gq) DREADD (2 males and 2 females from DID experiment, 3 males and 3 females from vocalization experiment). Before the experiment, animals were habituated over 30 minutes in the behavioral room with lights on. Immediately after a site-directed injection of CNO or vehicle (as described above in the DID methods subsection), mice were recorded with a video camera for 1 hour in their home cages, and the behavioral variables were evaluated and scored by two trained observers blinded to experimental conditions. Animals were randomly assigned to CNO or vehicle and the treatment was alternated with a 2-day interval between treatments using a Latin Square design. The percent of time mice spent performing locomotor activity, compulsive head grooming, and laying with their chin on the bedding were assessed, and the number of rearings, head shakes, and body readjusting performed during the session were recorded. Some behavior aspects considered in the present study were based on behavioral responses following administration of the emetic agent, lithium chloride, by rats as described by Navarro and Cubero (Navarro & Cubero, 2003). Body readjustment while stationary was based on a similar circling behavior induced by stress after social isolation described by Jerussi and Hyde (Jerussi & Hyde, 1985). Head shaking behavior was based on the same behavior presented by mice under nicotine withdrawal-induced stress described by Kotagale and colleagues (Kotagale et al., 2015). All the experiments started 3 hours after the lights went off in the animal facility, respecting a 7 to 10-day interval from the last manipulation in the DID or vocalization tests.

2.4. Microscopy

At the end of the experiments, mice were anesthetized with an overdose of kethamine/xylazine (100/10 mg/kg) and transcardially perfused with phosphate-buffered saline (PBS) for 4 minutes followed immediately by 10% formalin. Brains were then collected and postfixed in 10% formalin for 24 h, followed by cryoprotection in 0.2 M Phosphate buffer (PB) at pH 7.4. Brains were sectioned using a vibratome into 40 μm slices and every other section was mounted on charged glass slides (FisherBrand Superfrost, Thermo Fisher Scientific, Waltham, MA, USA). After allowed to air dry under dark conditions, the mounted tissue was coverslip using HardSet VECTASHIELD mounting medium with DAPI (Vector Laboratories Inc., Burlingame, CA, USA). Tissue that was not mounted was placed in cryopreserve solution and stored in a −20 °C freezer. The location of viral vector expression in the LC were evaluated through of the mCherry fluorescent tag and for cannulas placements in the RMTg were both verified using fluorescent microscopy (Leica DM6000 B widefield light microscope, Leica Microsystems, Buffalo Grove, IL, USA), as presented in Fig. 1. Only animals with bilateral mCherry expression in LC and correct bilateral cannula positions were included in statistical analyses.

Fig. 1 –

DREADDs expression, angled cannulas placements and verification of LC projection to the RMTg plus ethanol consumption, BEC levels and sucrose consumption during drinking-in-the-dark experiment. (A) Schematic showing experimental preparation. Top right photomicrograph shows a Cre-inducible AAV5-DIO viral vector expression (pAAV8-hSyn-DIO-hM3D(Gq)-mCherry) in the locus coeruleus (LC) of TH-ires-Cre+ mouse. Bottom left photomicroph shows cannula placement in 20° angle on the rostromedial tegmental nucleus (RMTg). (B) Representative photomicrographs of channelrhodopsin viral vector (AAV5-EF1a-DIO-hChR2-mCherry) injection into the LC of TH-ires-cre mice (left panel) and anterograde labeling in the RMTg from injection in the LC (right panel). Images were acquired using a widefield fluorescence microscope. IPR= Interpeduncular nucleus. (C) Relative to vehicle treatment, RMTg-directed infusion of CNO decreased binge-like ethanol consumption over 2-hours of mice expressing hM3D(Gq) DREADD in the LC. (D) BEC levels reflected the reduced ethanol consumption by hM3D(Gq) mice after RMTg-directed infusion of CNO relative to vehicle treatment. Treatment had no effect on the ethanol consumption (E) and or BEC levels (F) of mice treated with the control virus. Total consumption of 3% sucrose over 2-hours of drinking in mice expressing hM3D(Gq) (G) and control virus (H) in the LC after site-direction infusion of vehicle or CNO. Data are represented as mean ± SEM, n= 12–18 per group. Total volume (ml) average ± SEM of ethanol solution consumed by group per treatment: Control virus/Vehicle= 0.29 ± 0.04; Control virus/CNO= 0.25 ± 0.03; hM3D(Gq)/Vehicle= 0.38 ± 0.09; hM3D(Gq)/CNO= 0.14 ± 0.09. Total volume (ml) average ± SEM of sucrose solution consumed by group per treatment: Control virus/Vehicle= 1.12 ± 0.09; Control virus/CNO= 1.14 ± 0.11; hM3D(Gq)/Vehicle= 0.9 ± 0.11; hM3D(Gq)/CNO= 0.86 ± 0.11. *p<0.05 of CNO relative to vehicle treatment.

2.5. Statistics

Ethanol consumption was obtained by calculating the volume of solution ingested (ml) corrected by body weight (kg) and ethanol concentration (g) for each subject. For sucrose consumption, the volume of solution ingested (ml) was adjusted per body weight (kg). A Latin square design was used in the drinking experiments, and to compare the binge-like drinking levels of each solution repeated-measures (RM) ANOVAs were used to assess the total volume ingested and BEC levels following vehicle or CNO treatment (within-subjects factor) on the test day (separate analyses for each virus condition). Sex was evaluated as between-subject factors, but as the analyses failed to detect a significant effect of sex on the drinking behavior, data were collapsed across sexes. Similarly, order of vehicle versus CNO injection were evaluated as between-subject factors, but as the analyses failed to a detect a significant effect of order of presentation on the drinking behavior, data were collapsed across order. This way, paired 2-tailed Student t-tests followed by Bonferroni post hoc comparisons, as appropriate, were used to analyze total consumption and BECs. For vocalization, anxiety-like behavior and behavioral response experiments, each behavioral parameter was analyzed using 2-tailed paired-samples Student t-tests. Analyses were calculated using SPSS Software version 26 (IBM Corporation, Armonk, NY, USA), graphs were made using GraphPad Prism Software version 8.0.0 224 (La Jolla, CA).

Results

Activation of the TH+ LC → RMTg pathway blunts binge-like ethanol drinking in male and female mice and has no effect on sucrose consumption

Our first objective in the present study was to investigate the TH+ LC → RMTg pathway activation in modulating the binge-like ethanol drinking. A schematic showing our experimental preparation and viral expression are presented in Figs. 1A and 1B. Results from this experiment are demonstrated in Fig. 1. In the DID experiment, as shown in Fig. 1C, the paired 2-tailed Student t-test showed that the site-directed microinfusion of CNO reduced the volume of ethanol solution consumed by mice expressing the hM3D(Gq) DREADD in the LC [t(15)=4.076, p=0.001] relative to vehicle. The diminished ethanol consumption by the animals at the end of 2 hours of drinking after the noradrenergic pathway activation reflected on their blood ethanol concentrations (BECs) where the CNO treatment reduced their BEC levels [t(15)=3.913, p=0.001] compared to vehicle of mice expressing hDM3(Gq) DREADD (Fig. 1D). No effect of treatment was observed in the ethanol consumption [t(12)=0.046, p=0.964] or in the BEC levels [t(10)=0.046, p=0.958] of mice expressing control virus in the LC, as shown in Figs. 1E and 1F, respectively. Given the role played by the TH+ LC → RMTg activation in decreasing the binge-like ethanol drinking, the next step was to investigate if the effects of activating this pathway were specific to ethanol intake by assessing the consumption of 3% sucrose solution intake in the same DID paradigm. The results from this experiment are presented in Fig. 1. The statistical analyses highlighted that CNO had no effect on the 3% sucrose consumption of animals expressing both hDM3(Gq) [t(14)=0.116, p=0.909] (Fig. 1G) and control virus [t(10)=−0.981, p=0.350] (Fig. 1H) in the LC. These results confirm that the activation of the noradrenergic LC to RMTg circuit reduces ethanol consumption specifically. We also found that general activation of the LC → RMTg pathway using a retrograde Cre virus strategy in C57BL/6J mice blunted binge-like ethanol intake, but also reduced sucrose intake (Suppl Figs. 3 and 4). Suppl. Fig. 2 shows that activation of the hDM3(Gq) DREADD induced by peripheral injection of CNO significantly induced c-Fos expression in the LC, confirming DREADD-induced activation of TH+ neurons in the LC. It is notable that while this approach confirms functional activation of the Gq-DREADD, this approach does not directly demonstrate the functional activity of the TH+ LC → RMTg circuit.

Activation of the TH+ LC → RMTg pathway increases the emission of aversive-related vocalizations in male mice

In acoustic communication, the emission of vocalizations carries information on the emotional states of animals, including rodents (Brudzynski, 2013; Wohr & Schwarting, 2013). Hypothesizing that activation of the LC to RMTg circuit drives aversive reactions that may modulate ethanol drinking, our objective was to assess whether chemogenetic activation of the TH+ LC → RMTg pathway modifies the animals’ emission of vocalizations as a function of an aversive-like state induced by this stimulus. The spectrograms of all vocalization subtypes emitted by TH-ires-Cre transgenic male mice used in the present study are represented in Fig. 2A–E. Before experimental manipulations, the TH-Cre+ and TH-Cre− mice presented similar baselines for anxiety-like behavior and calling rates (results are presented in Suppl. Figs. 5 and 6, respectively). As presented in Fig. 2, the 2-tailed independent-Sample Student t-test showed that the total number of vocalizations emitted within 30 min after RMTg-directed injection of CNO did not vary between the TH-ires-Cre phenotypes [t(8)=−0.399, p=0.700] (Fig. 2F). However, the statistical analyses showed that the percentage of mid-frequency vocalizations (MFVs) emitted by the TH-ires-Cre positives (expressing hM3D(Gq)) was higher when compared to the siblings confirmed as negatives to the Cre expression [t(8)=−2.650, p=0.029]. The percentages of LFVs [t(8)=0.594, p=0.569], LFHs [t(8)=0.394, p=0.704], USVs [t(8)=2.122, p=0.067] and noisy vocalizations [t(8)=0.689, p=0.511] were similar between the groups. These results are presented in Fig 2G. Similarly described by Grimsley and colleagues ((Grimsley et al., 2016)), which described emission of MFVs by CBA/CaJ mice under high stress, our results confirm that the noradrenergic pathway manipulation alters the vocal behavior of the animals, and that it reflects the level of aversion produced by this pathway activation.

Fig. 2 –

Spectograms of vocalizations from male TH-ires-Cre mice and vocal repertoire and anxiety-like behavior of mice expressing hM3D(Gq) in the LC after chemogenetic manipulation of TH+ LC → RMTg pathway. (A) Bouts of ultrasonic vocalizations (USVs). (B) Noisy vocalization. (C) Bouts of mid-frequency vocalizations (MFVs). (D) Low-frequency harmonic vocalizations (LFH). (E) Bouts of LFH and low-frequency vocalizations (LFVs). No alteration in the call rates 30-min (F) after RMTg-directed infusion of CNO. Percentage of vocal repertoire within 30 min (G) after the pathway activation indicated higher emissions of MFVs by TH-Cre+ mice. The time spent (H) and distance traveled in the center (I), and total distance traveled in the OFT (J) during 2 hours after RMTg-directed infusion of CNO were not affected by the pathway activation. Data is presented as mean ± SEM, n=5 per group. *p<0.05 of TH-ires-Cre positive relative to TH-ires-Cre negative mice.

To evaluate whether the higher emission of MFVs observed in male TH-Cre+ mice after the noradrenergic pathway manipulation is related to anxiety, the same TH-Cre+ and TH-Cre− mice animals had their anxiety-like behavior evaluated in the OFT after LC → RMTg pathway activation. The RMTg-site directed injection of CNO had no effect on the time spent [t(8)=−0.278, p=0.791] and distance traveled in the center of the chamber [t(8)=0.392, p=0.706] (Fig. 2H and Fig. 2I, respectively), or in the total distance traveled by the animals in the chamber [t(8)=1.453, p=0.184] (Fig 2J).

Activation of TH+ LC → RMTg pathway increases behaviors associated with unconditioned aversive reactions in male and female mice

To test the hypothesis that the activation of the TH+ LC → RMTg pathway induces aversive-like behaviors, male and female TH-ires-Cre+ mice expressing hM3D(Gq) DREADD in the LC were given RMTg-directed infusion of CNO or vehicle, and immediately video recorded in their homecages. The results from this experiment are presented in Fig. 3. The 2-tailed paired-samples Student t-tests revealed that the RMTg-directed injection of CNO increased the number of body readjusting movements while mice were stationary [t(8)=−2.788, p=0.024] (Fig. 3A), the number of head shakes [t(8)=−6.249, p<0.001] (Fig. 3B), and the time spent with the chin on bedding [t(8)=−3.524, p=0.008] (Fig. 3D) when compared to the vehicle treatment. No difference was found in the total number of rearings [t(8)=0.693, p=0.508] (Fig. 3C), the time spent in overgrooming head [t(8)=−1.710, p=0.126] (Fig. 3E) and locomotor activity [t(8)=−0.082, p=0.936] (Fig. 3F) after treatments.

Fig. 3 –

Behavioral responses of male and female mice expressing hM3D(Gq) in the LC after RMTg-directed infusion of vehicle or CNO. (A) Number of readjustings. (B) Number of head shakings. (C) Number of rearings. (D) Percentage of time with chin on bedding. (E) Percentage of time overgrooming head. (F) Percentage of time in locomotor activity. Data are represented as mean ± SEM, n=4–5 per group *p<0.05 vehicle vs CNO.

Discussion

Here we provide novel evidence of a TH+ LC → RMTg pathway, and that chemogenetic activation of this pathway significantly blunts binge-like intake of a 20% ethanol solution in both male and female mice. Activation of this pathway did not significantly alter consumption of a 3% sucrose solution, suggesting that this noradrenergic pathway selectively modulates binge-like ethanol intake. Because the RMTg has been found to modulate CTA learning (Glover et al., 2016; Sheth et al., 2016), we further characterized the role of this pathway in modulating unconditioned behaviors thought to reflect the induction of an aversive-like state. Consistently, chemogenetic activation of the TH+ LC → RMTg circuit promoted a significant increase of mid-frequency vocalizations in male mice as well as aversion-related behaviors in male and female mice. Because activation of this pathway did not alter anxiety-like behaviors, the aversive responses elicited by activation of this pathway are likely not related to fear-like emotional reactions but rather aversive responses consistent with malaise. Taken together, the present results provide the first direct evidence that activation of a NE+ pathway originating in the LC and innervating the RMTg protects against binge-like ethanol drinking and promotes aversion-related behaviors. We speculate that activation of this pathway while consuming ethanol provides a braking mechanism to protect against excessive consumption by triggering aversive responses.

Chemogenetic activation of the TH+ LC → RMTg pathway reduced binge-like ethanol drinking without affecting the consumption of sucrose in male and female mice, suggesting that this circuit modulates ethanol intake specifically. Importantly, the effect of CNO was observed only in mice expressing the hM3D(Gq) DREADD and not the control virus, which argues against possible of off-target effects of CNO, such as effects caused by back-metabolism of CNO to clozapine (Manvich et al., 2018). Our results are in accordance with recent findings from our group, where chemogenetic activation of NE neurons in the LC blunted excessive ethanol consumption in the DID paradigm (Burnham et al., 2021). Additionally, an optogenetic study showed that driving LC-NE neurotransmission into a phasic activation decreased ethanol consumption in rats (Deal et al., 2020). Moreover, down-regulation of NE transporters was found in the LC of rat lines selectively bred for high alcohol preference (P and HAD) in comparison to their related low alcohol preference lines (NP and LAD) (Hwang et al., 2000), suggesting that the protective role played by the LC may be compromised in high ethanol drinking genetic lines. Taken together, given that catecholamine neurons of the LC are composed primarily of NE (Cunningham & Sawchenko, 1988; Robertson et al., 2013), these results suggest that noradrenergic neurons in the LC may play important protective role against excessive ethanol consumption. However, because LC is a molecularly diverse structure (Schwarz & Luo, 2015), that have been shown to co-express glutamate (Fung et al., 1994), it is important to highlight the caveat that chemogenetic activation of TH+ neurons in the LC → RMTg circuit blunted binge-like ethanol intake by activation of co-expressed neurochemicals such as glutamate. Future research employing pharmacological manipulations will help address this caveat. Interestingly, the use of a retrograde cre virus technology in tandem with cre-inducible DREADDs to promote general activation of the LC → RMTg circuit reduced the ethanol intake, but also affected sucrose consumption (see Suppl. Fig. 4). While it is not entirely clear why we observed different results, it is possible that there was more robust activation of the LC → RMTg pathway using the retrograde virus approach, which in turn may have produced non-specific reductions of both ethanol and sucrose. One caveat related to sucrose consumption control studies here is that the volumes of 20% ethanol and 3% sucrose differed, suggesting potential differences between these stimuli over multiple dimensions (e.g., rewarding or caloric properties). While it is very difficult to equate different stimuli at all levels, we used 3% sucrose in the present work to be consistent with control conditions used in our previous published work (Burnham et al., 2021).

It is known that amplification of LC activity induces stressful and anxiogenic responses (Morris et al., 2020) and that the NE system is involved in the negative emotional states that both promote alcohol consumption through withdrawal (den Hartog et al., 2020) and reduce alcohol consumption through aversion (Thiele et al., 2000). In that regard, the emission of innate and learned vocal sounds convey important information about the internal and emotional states of animals (Simola & Granon, 2019), and have been used as a tool to asses several psychiatric disorders (Caruso et al., 2020; Luchetti et al., 2021; Premoli et al., 2021; Wendler et al., 2019), including alcohol use disorder (for review see (Mittal et al., 2018)). For example, in rats, vocal repertoire of alcohol-naïve, high-alcohol (P and HAD-1) and low-alcohol (NP and LAD-1) drinking lines served as a predicting factor for the level of alcohol consumption among them (Mittal et al., 2018). Mice present a rich vocal repertoire, where high-frequency vocalizations emerge prominently in social interactions and courtship (Chen et al., 2021; Grimsley et al., 2011), whereas lower frequency vocalizations are produced in response to stress (Grimsley et al., 2016; Niemczura et al., 2020). Here, we demonstrated that activation of the TH+ LC → RMTg pathway selectively increases the emission of a newly identified mid-frequency vocalization (MFV) (Grimsley et al., 2016) by TH-Cre+ mice expressing hM3D(Gq) DREADD when compared to the TH-Cre− mice, with no change in the total call rate. Interesting, the MFVs subtype was emitted by CBA/CaJ mice undergoing restraint stress and correlated with elevated levels of corticosterone and higher stress relative to mice in isolation (Grimsley et al., 2016). Moreover, here, despite the fact that emissions of LFHs (audible squeak) and noisy vocalization subtypes were similar between both TH-Cre+ and TH-Cre− mice after CNO, their considerable presence in both groups could be caused by the social isolation period that animals underwent after surgeries (Grimsley et al., 2016). LFHs were commonly found in aversive contexts including pain (Hong et al., 2018) and agitation (Grimsley et al., 2013), while noisy vocalizations were intensively emitted during social isolation (Grimsley et al., 2016). In addition, we have found that both TH-Cre genotypes emitted a similar non-harmonic low-frequency vocalization (LFV), a subtype that is not described in the literature. It is important to highlight that vocal repertoire may vary among different genetic backgrounds (Heckman et al., 2016), reflecting by variance in the structural and spectrotemporal characteristics of the vocalizations. Finally, USV was not influenced by activation of the TH+ LC → RMTg pathway, which is not surprising given that USVs are typically associated with positive emotional behaviors. Together, these findings suggest that the noradrenergic LC → RMTg pathway activation induces an aversive-like state that is reflected on the emission of non-USVs stress-related vocalizations.

In addition to the assessment of the emission of vocalizations, our next objective was to evaluate whether TH+ LC → RMTg pathway activation would elicit further unconditioned behavioral responses thought to reflect an aversive-like state. Here, activation of the NE+ LC → RMTg pathway increased the number of readjusting movements while the animals were stationary, or spontaneous circling behavior, when compared to the vehicle treatment. In accordance, similar behaviors were observed in mice under stress induced by social isolation (Jerussi & Hyde, 1985). The number of head shakes were also exacerbated after NE+ LC → RMTg pathway activation. Interesting, head shake movements were also increased in mice under nicotine withdrawal-induced stress (Kotagale et al., 2015). Further, activation of the TH+ LC → RMTg pathway increased the time that mice spent with their “chin on bedding”, a somatic sign of malaise that is similar to the “lying on belly” (LOB) behavior observed in rats after treatment with the emetic agent, lithium chloride (Navarro & Cubero, 2003), a drug used as an aversive unconditioned stimulus in learning experiments (Kislal & Blizard, 2016). Taken together, these results demonstrate that the activation of the noradrenergic LC → RMTg circuit induces somatic behaviors that correlate with stress and malaise without affecting the locomotor activity, and are in accordance with the aversive-like vocal repertoire found in the vocalization experiment.

It is of interest to consider the possible mechanisms by which the TH+ LC → RMTg pathway modulates ethanol consumption and aversion. As noted above, we have previously found that in doses equated for the ability to support a CTA, ethanol, but not LiCl, induced robust c-Fos expression in LC TH+ neurons (Thiele et al., 2000). Interestingly, we also found that doses of ethanol that supported CTA induced significantly weaker c-Fos expression in the LC of rats selective bred for high ethanol drinking relative to their low ethanol drinking counterparts (Thiele et al., 1997), and we recently found similar results in the selectively bred iHDID-1 line of mice relative to their control line (Robinson et al., 2020). Notably, relative to the control line the iHDID-1 line also has been shown to exhibit reduced sensitivity to ethanol-induced CTAs (Crabbe et al., 2019). Here we show that the LC provides robust projections that innervate the RMTg. The RMTg is a midbrain GABAergic nucleus that sends inhibitory projects to the midbrain VTA (Jhou, Geisler, et al., 2009). Importantly, the lateral habenula (LHb), which modulates aversive properties of drugs including ethanol (Haack et al., 2014) provides excitatory input to the RMTg, which in turns drives inhibitory input to the VTA (Hong et al., 2011), and it is this LHb → RMTg → VTA circuit that is thought to modulate responses to aversive stimuli including drugs of abuse (Barrot et al., 2012; Jhou, Fields, et al., 2009). As we show that the LC provides NE input to the RMTg, and since NE is an excitatory neurotransmitter, we propose the LC → RMTg NE pathway, much like the LHb → RMTg pathway, as a novel circuit modulating the aversive effects of ethanol, and which in turn curbs binge-like ethanol drinking. Our current investigations will determine if chemogenetic silencing the TH+ LC → RMTg pathway blunts the aversive effects of ethanol by attenuating ethanol-induced CTAs, and if this manipulation will be associated with increased binge-like ethanol intake. Because we used TH-ires-cre+ mice maintained on a C57BL/6J background, one problem that we could encounter after silencing this pathway is a lack of effect on ethanol intake stemming from a ceiling effect as C57BL/6J mice exhibit high levels of ethanol intake at baseline. One way to prevent this potential confound would be to test the animals during the light cycle where it has been shown that C57BL/6J mice exhibit lower levels of ethanol consumption (Lowery-Gionta et al., 2012b).

In conclusion, here we provide novel evidence for a NE+ LC → RMTg pathway, we show that chemogenetic activation of this pathway significantly blunts binge-like ethanol intake without altering sucrose drinking, locomotor activity, or anxiety-like behavior, and we demonstrate that chemogenetic activation of this pathway induces behaviors that are consistent with an aversive- or malaise-like state. Further, we speculate that ethanol-induced activation of NE+ neurons in the LC → RMTg circuit is part of a “braking” mechanism that protects against increased ethanol consumption by inducing aversive reactions to ethanol. Theoretically, with repeated consumption activation of this protective circuit may weaken, leading to escalating levels of ethanol intake that are characteristic as AUDs progresses. Our future work will determine if silencing this NE+ LC → RMTg pathway promotes increased levels of binge-like ethanol intake, and blunts aversive reactions to ethanol, such as by attenuating ethanol-induced CTAs. The present results suggest that increasing NE tone in RMTg-projecting LC neurons may represent a new approach to treating AUDs, especially early on prior to the development of ethanol dependence.

Highlights.

• Chemogenetic activation of a noradrenergic locus coeruleus to rostromedial tegmental nucleus circuit blunted binge-like ethanol intake.

• Activation of this pathway did not alter sucrose consumption, suggesting that this manipulation was specific to effects on ethanol intake.

• Activation of this pathway increased mid-frequency vocalization in mice, consistent with induction of an aversive-like state.

• Activation of this pathway promoted behaviors also induced by drugs that promote visceral illness, suggesting induction of a malaise-like state.