Modeling cue-exposure therapy for alcohol use disorder in rhesus monkeys: Effects of putative cognitive enhancers

Abstract

Background.

Cue-exposure therapy (CET) is an effective approach for anxiety-related disorders, but its effectiveness for substance use disorders is less clear. One potential means of improving CET outcomes is to include a cognitive-enhancing pharmacotherapy. This study evaluated d-cycloserine (DCS) and RY-023, putative cognitive enhancers targeting glutamate and GABA systems, respectively, in a monkey model of CET for alcohol use disorder.

Methods.

Male rhesus monkeys (n=4) underwent multiple cycles of the CET procedure. During baseline (Phase 1), monkeys self-administered an ethanol solution under a fixed-ratio schedule and limited access conditions such that every 5^th response in a 3-h session resulted in 30-s access to a drinking spout and a change in ethanol-paired cue lights from white to red. Behavior then was extinguished (Phase 2) by omitting the ethanol solution yet retaining the ethanol-paired stimulus lights. Monkeys also received injections of vehicle, DCS (3 mg/kg), a partial agonist at the glycine modulatory site on glutamatergic NMDA receptors, or the α5GABA_A receptor-selective inverse agonist RY-023 (0.03 or 0.3 mg/kg). Once responding declined, monkeys underwent a cue reactivity test (Phase 3), and then returned to self-administration the following day to assess reacquisition (Phase 4).

Results.

Through multiple cycles, self-administration remained stable. Compared to vehicle, DCS facilitated extinction of ethanol seeking (Phase 2) and delayed reacquisition of ethanol self-administration (Phase 4). In contrast, RY-023 facilitated extinction (Phase 2) and reduced cue reactivity (Phase 3).

Conclusions.

Adjunctive pharmacotherapy can improve CET outcomes, but the choice of pharmacotherapy should be dependent on the outcome of interest.

1. Introduction

Over the last 10–20 years, the prevalence of alcohol use, high-risk drinking and clinically-defined alcohol use disorder (AUD) has increased significantly in the U.S. population (Grant et al., 2017). The misuse of alcohol is associated with debilitating social, psychological and physical health problems, which place an extreme financial burden on individuals, families, and society as a whole (Sacks et al., 2015). A hallmark of AUD is a vulnerability to return to active drinking and although both behavioral and pharmacological therapies are available to treat AUD (e.g., Mason, 2017; Swift, 1999), the majority of abstinent individuals resume drinking within a year of detoxification (e.g., Grant et al., 2017; Maisto et al., 2000). These high rates underscore the continuing need to develop new therapeutic approaches for AUD.

A return to alcohol use after a period of abstinence can be precipitated by environmental cues that are associated with drinking behavior, eliciting craving and seeking behavior that ultimately leads to resumption of alcohol consumption (Niaura et al., 1988; Witteman et al., 2015). A treatment strategy focused particularly on substance-associated cues that has been considered for substance use disorders, including AUD, is cue-exposure therapy (CET; for review see Mellentin et al., 2017). CET is a behavioral psychological treatment approach whereby individuals are exposed repeatedly to relevant substance-associated cues in the absence of access to the substance with the goal of weakening conditioned associations that promote craving and a return to active use (Conklin and Tiffany, 2002; Drummond et al., 1990; Marlatt, 1990). Although CET has shown promise for the treatment of anxiety-related disorders (e.g., Byrne et al., 2019), its usefulness for the treatment of substance use disorder and AUD is less clear. In drinkers, for example, some studies have reported positive outcomes with CET (e.g., Monti et al., 1993; Sitharthan et al., 1997), whereas others reported no effect (e.g., Dawe et al., 2002; Loeber et al., 2006), leading to the conclusion that there is no consistently strong evidence for the effectiveness of CET (Conklin and Tiffany, 2002; Mellentin et al., 2017).

One possible means by which to improve the effectiveness of CET is to include a pharmacological component to the treatment that is aimed towards augmenting the new learning that presumably is occurring during extinction of the conditioned cues. The utility of this “cognitive-enhancing” pharmacological approach for CET has been studied most extensively with D-cycloserine (DCS), a partial agonist at the glycine modulatory site on glutamatergic NMDA receptors (for review see Myers and Carlezon, 2012). Preclinically and clinically, DCS facilitates NMDA receptor-dependent synaptic plasticity and improve learning and memory (e.g., Bailey et al., 2007; Monahan et al., 1989; Onur et al., 2010; Rouaud and Billard, 2003). In the context of preclinical alcohol models, some studies have shown that DCS facilitates extinction of conditioned alcohol seeking in rats, ultimately slowing resumption of alcohol self-administration (Vengeliene et al., 2008, but see Groblewski et al., 2009). Clinically, DCS appears to facilitate extinction-based treatment strategies for a range of anxiety-related psychiatric disorders (cf. Norberg et al., 2008). In contrast, the effectiveness of DCS to facilitate extinction to alcohol-paired cues and reduce cue reactivity in patients is less clear, with some studies reporting facilitation of extinction and reduced cue reactivity (e.g., Kiefer et al., 2015; MacKillop et al., 2015), others reporting no effects (e.g., Kamboj et al., 2011; Watson et al., 2011), and others reporting actual enhancement of cue reactivity (e.g., Hofmann et al., 2012). While a number of important differences between these studies (e.g., treatment v. non-treatment seeking individuals, magnitude of baseline cue reactivity, DCS regimen, etc.) could contribute to the disparate findings, there appears reason to evaluate other cognitive-enhancing compounds in a CET setting.

Inverse agonists with selectivity for the α5 subunit of GABA_A receptors (i.e., α5GABA_A receptors) have been shown to reverse cognitive deficits and/or improve learning and memory across a variety of models/procedures and species (for reviews, see Atack, 2011; Jacob, 2019). These compounds also are effective at attenuating multiple abuse-related effects of alcohol in rodents, nonhuman primates and humans (Chandler et al., 2019; 2021; Cook et al., 2005; June et al., 2001; Nutt et al., 2007; Platt et al., 2005; Rüedi-Bettschen et al., 2013). Together, these observations make evaluation of this class of compounds in a CET model appealing.

The overall goal of the present study was to determine whether addition of a cognitive enhancing adjunctive pharmacotherapy would improve outcomes in a nonhuman primate model of CET designed to reduce the propensity to return to active drinking after abstinence. Specifically, the ability of the α5GABA-A receptor inverse agonist RY-023 to 1) facilitate extinction of responding, 2) reduce cue reactivity, and 3) slow reacquisition of alcohol self-administration was compared to DCS.

2. Methods and Materials

2.1. Animals and Apparatus

Four adult male rhesus monkeys (Macaca mulatta) served as subjects (see Supplemental Table 1 for demographic information). All four had prior experience in oral alcohol self-administration studies (e.g., Berro et al., 2019). All monkeys were provided with sufficient food to maintain stable body weights (Teklad 25% Monkey Diet; Harlan/Teklad, Madison, WI, USA) and had unlimited access to water via a lick-activated spout mounted in the home cage wall. Fresh and dried fruits and seeds were provided daily, and a vitamin supplement was given once a week. Lighting was maintained under a 12 h light:12 h dark cycle, with lights on at 0600 hours. All animal use procedures were approved by the University of Mississippi Medical Center’s Animal Care and Use Committee and were conducted in accordance with the National Research Council’s Guide for Care and Use of Laboratory Animals (8^th edition, 2011).

Monkeys were housed in individual home cages that had been modified to serve additionally as experimental chambers. Alcohol self-administration occurred via an operant drinking panel attached to one side of the cage. The panel accommodated two response levers, two retractable sippers equipped with solenoids to prevent dripping (model #: ENV-652AM; Med Associates, Inc., Georgia, VT, USA), and two sets of triple stimulus lights (red, green, white; model ENV-622M; Med Associates, Inc.) mounted above the sippers. Each sipper was attached to a stainless-steel liquid reservoir, fixed on the outside of the drinking panel, via Tygon^® tubing. For these studies, only one side of the operant drinking panel was active.

2.2. Drugs

Ethanol (95%; Pharmco Products, Brookfield, CT, USA) was diluted to 2% w/v using tap water. RY-023 was synthesized at the University of Wisconsin-Milwaukee, as described previously in Huang et al. (2000) and was dissolved in 50% propylene glycol/50% sterile water solution. DCS (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in 0.9% saline solution. Vehicle (saline) and experimental compounds were administered intramuscularly (0.1–0.3 ml per injection).

2.3. Behavioral Procedures

2.3.1. Overview.

Each monkey underwent four blocks of testing, with each block consisting of four cycles. Each cycle consisted of four experimental phases: self-administration-baseline, extinction, cue-reactivity, and self-administration-reacquisition (Figure 1). During the extinction phases within a block, the drug conditions of saline alone (i.m., 30 min pretreatment, daily), DCS (3.0 mg/kg, i.m., 30 min pretreatment, weekly + saline on non DCS days), or RY-023 (0.03 or 0.3 mg/kg, i.m., 10 min pretreatment, daily) were administered with the constraint that, across every block, the order of drug treatments was randomized for each monkey (Supplemental Table 1). This experimental design resulted in each monkey experiencing a given drug condition four times. This design was adopted to avoid potential confounds associated with the monkeys learning the task across repeated cycles/blocks. DCS dose selection was based on previously published work in macaque monkeys showing that DCS can reverse learning and memory deficits across different cognitive tasks (Matsuoka and Aigner, 1996; Schneider et al., 2000). Further, DCS administration parameters (e.g., weekly + saline on non-DCS days) were based on evidence that chronic administration of DCS results in rapid tolerance to its behavioral effects; its positive effects on learning are evident after acute dosing (Hofmann, 2014; Lopes et al., 1997; Myers and Carlezon, 2012; Rothbaum, 2008). RY-023 dose selection and administration parameters also were based on previously published work in monkeys and chosen to be those that did not alter the ability of an individual to complete an operant task (Soto et al., 2013), even across repeated daily dosing (unpublished). Note that only saline was evaluated as a control condition despite different test drugs requiring different vehicles (i.e., saline versus 50% propylene glycol (PG)-containing vehicle) and the possibility that the PG vehicle might produce non-specific effects of behavior. However, this decision was informed by published data (Berro et al., 2019; Rüedi-Bettschen et al., 2013, Sawyer et al., 2014) showing that neither saline nor PG-containing vehicles (including up to 80% PG) affect baseline alcohol or sucrose self-administration. Further, they do not affect rate of responding in an IV alcohol discrimination procedure (Platt et al., 2005). Lastly, data (unpublished) from a behavioral observation procedure show that the two vehicles do not differ in their effects on species-typical behavior. A PC equipped with Med-Associates software (Georgia, VT, USA) controlled experimental conditions and recorded data.

Figure 1.

Schematic of study design. “SA” = self-administration.

2.3.2. Self-administration-baseline.

Experimental sessions occurred 5 days per week and lasted 180 min. Monkeys self-administered ethanol (2%, w/v) as described previously (cf. Berro et al. 2019). Briefly, illumination of white lights signaled the start of the session. Every fifth lever press (FR 5) resulted in a switch from white light to red light (i.e., ethanol-paired stimulus) and access to the drinking spout for 30 s. Depression of the spout during extension resulted in ethanol delivery, continuing as long as the sipper was depressed by the monkey during the 30-s period. The actual duration (up to 30 s), as well as intake volume, were controlled by the subject. Self-administration sessions continued for a minimum of five consecutive sessions or until intake was stable, which was defined as no upward or downward trend in intake (mls) over three consecutive days. These parameters typically engender levels of intake that result in the majority of monkeys being classified as “chronic moderate drinkers” (cf. Chandler et al. 2017; Vivian et al. 2001).

2.3.3. Extinction.

Extinction sessions were initiated once baseline ethanol self-administration was stable. All aspects of the experimental session remained the same, except that no solution was available for self-administration. During extinction, subjects were administered vehicle (saline), RY-023 (0.03 or 0.3 mg/kg), or DCS (3 mg/kg) prior to the start of the session. Extinction sessions continued until lever pressing declined to less than 10% of the number of responses maintained under active self-administration sessions. The monkey then was allowed a four-day rest period in the home cage during which no experimental sessions were conducted.

2.3.4. Cue reactivity.

A one-day cue-reactivity test occurred four days after extinction. Conditions for the test were identical to those for extinction (i.e., re-exposure to response-contingent presentations of the cue light in the absence of ethanol).

2.3.5. Self-administration-reacquisition.

Ethanol self-administration sessions resumed the day after the cue-reactivity test. Reacquisition sessions were identical to baseline self-administration sessions, as described above. A monkey was considered to have “reacquired” self-administration when ethanol intake (mls) reached the average of active self-administration associated with the current cycle (i.e., self-administration-baseline).

2.4. Data Analysis

The last five ethanol self-administration sessions were used to establish the baseline level of ethanol intake (g/kg and mls) and total responses for each cycle. For each self-administration measure, data were averaged by monkey for each block of tests and presented for the group of monkeys by block. For analysis of the extinction training sessions, total responses on the active lever were converted to a percentage of self-administration responses. “Self-administration responses” refers to the average of the active lever responses across the final 5 days of baseline self-administration. For each individual monkey, data were averaged separately for the four treatment conditions. Subsequently, the number of days to reach the extinction criterion for each treatment condition was averaged across monkeys. For analysis of the cue reactivity test, the measure used was the total responses on the active lever expressed as a percentage of self-administration-responses; note that data for the last day of extinction also were included in the analysis. For analysis of reacquisition of self-administration, ethanol intake (mls) was converted to a percentage of ethanol intake (mls) during self-administration-baseline. For each individual monkey, data were averaged separately for the four treatment conditions. Subsequently, the number of days to reach the reacquisition criterion for each treatment condition was averaged across monkeys. For all dependent measures, one-way repeated-measures ANOVAs with Bonferroni t-tests were used to assess significant drug/dose effects compared to vehicle (or extinction, in the case of cue reactivity) in each experimental phase. Additionally, because each treatment condition was evaluated four times (once in each block), separate one-way repeated-measures ANOVAs were used to assess the consistency of the drug effects across time on measures of days to extinction, cue reactivity, and days to reacquire self-administration.

3. Results

3.1. Self-administration baseline

Figure 2 summarizes the average number of responses (panel A), total volume (mls; panel B), and ethanol dose (g/kg; panel C) aggregated across subjects by test block. Although individual variability was evident within each of the measures (e.g., M-Fa tended to emit considerably more lever presses than the other monkeys; M-Lo consistently consumed the lowest dose of ethanol), as a group, self-administration generally was consistent across blocks. No significant differences in lever presses, intake volume (mls) or dose were evident across the course of the study.

Figure 2.

Ethanol self-administration-baseline. A) Number of responses by individual and by group across treatment blocks. B) Ethanol intake in mls by individual and by group across treatment blocks. C) Ethanol intake in g/kg by individual and by group across treatment blocks. Different symbols represent data from different individuals; bars represent group mean (+ SEM).

3.2. Extinction

Figure 3 illustrates responding under extinction conditions and different drug treatments in individual subjects (panels A-D), as well as the average number of days to meet the extinction criterion (i.e., decline in responding to < 10% of active self-administration responding) by treatment (panel E). Although responding under all treatment conditions and in most subjects tended to increase/decrease non-systematically across days, monkeys always reached the extinction criterion within 10 days of the start of this phase. The rapidity with which the group of monkeys met the criterion differed significantly with treatment condition (F(3,9) = 11.533, P = 0.002). Multiple comparison tests revealed that all 3 drug treatment conditions facilitated extinction and reduced the number of days needed to meet extinction criterion when compared to the vehicle treatment (P < 0.05). The effects of the drug pretreatments generally were consistent across the 4 blocks of testing, with no evidence of the development of tolerance (RY-023 [0.03 mg/kg]: F(3,9) = 0.442, P = 0.729; RY-023 [0.3 mg/kg]: F(3,9) = 0.140, P = 0.933). However, there was a strong trend for extinction to be quicker across subsequent blocks with DCS treatment (F(3,9) = 3.841, P = 0.051).

Figure 3.

Extinction. A-D) Responding under extinction conditions expressed as a percentage of self-administration responding by treatment and individual subject (M-Mo, M-Ro, M-Fa, M-Lo; mean ± SEM). E) Number of days until extinction criterion met (i.e., responding declined to < 10% of active self-administration). Bars represent group mean (+ SEM). N = 4 * P<0.05 compared to vehicle (“VEH”).

3.3. Cue reactivity

Figure 4 illustrates average responding during the cue reactivity tests for each treatment condition, as well as for the last day of extinction. Responding is expressed as a percentage of active self-administration responding. By-and-large, re-exposure to the ethanol-paired cue induced modest yet significant levels of ethanol seeking-behavior under vehicle conditions compared to levels of responding on the final day of extinction (F(4,12) = 6.680, P = 0.005; Bonferroni t-tests, P < 0.05). Compared to vehicle, treatment with DCS and RY-023 did reduce responding upon re-exposure to the cue, but only the highest dose of RY-023 produced a significant decrease (F(4,12) = 6.680, P = 0.005; Bonferroni t-tests, P < 0.05). Moreover, the effect observed with 0.3 mg/kg RY-023 was consistent across the testing blocks (F(3,9) = 0.428, P = 0.738).

Figure 4.

Cue reactivity. Responding on the last day of extinction (“EXT”) and on cue reactivity tests expressed as a percentage of baseline self-administration responding. Different symbols represent data from different individuals; bars represent group mean (+ SEM). * P<0.05 compared to extinction (“EXT”); # P<0.05 compared to vehicle (“VEH”).

3.4. Self-administration-reacquisition

Figure 5 illustrates responding during reacquisition under different drug treatments in individual subjects (plotted as a percentage of self-administration responding; panels A-D), as well as the average number of days to reacquire under each treatment condition. Obvious individual differences are evident in the rate at which subjects reacquired self-administration to baseline levels. At one extreme is subject M-Mo that reacquired in under 10 days with no strong differences between treatment conditions. In contrast, the three other subjects took between 30–40 days to reacquire self-administration under at least one (and sometimes more than one) treatment condition. The rapidity with which the group of monkeys met the reacquisition criterion differed significantly with treatment condition (F(3,9) = 6.096, P = 0.015). Multiple comparison tests revealed that only DCS deterred reacquisition of self-administration significantly longer than vehicle treatment (P < 0.05). Further, this deterring effect of DCS was observed consistently across all of the testing blocks (F(3,9) = 1.439, P = 0.295).

Figure 5.

Ethanol self-administration-reacquisition. A-D) Responding during reacquisition expressed as a percentage of baseline self-administration responding by treatment and individual subject (M-Mo, M-Ro, M-Fa, M-Lo; mean ± SEM). E) Number of days until reacquisition criterion met. Bars represent mean (+ SEM). N = 4. * P<0.05 compared to vehicle (“VEH”).

4. Discussion

A return to active drinking after a period of abstinence is known to be precipitated by re-exposure to environmental cues associated with drug taking. CET, a therapy designed to reduce the psychological/physiological impact of environmental cues, is being considered as a treatment strategy for substance use disorders. Its effectiveness in treating AUD patients (as well as patients with other use disorders) is not universal, but there is the potential for improved outcomes with the addition of an adjunctive pharmacotherapeutic drug. In the present study, we developed a CET model in male rhesus monkeys and then evaluated two adjunctive drugs: RY-023 and DCS. RY-023 is an α5GABA_A inverse agonist, and this class of drugs has been shown to both improve performance in cognitive tasks and attenuate many of the abuse-related effects of alcohol. DCS was chosen as a comparator because it is the most studied adjunctive drug in the context of CET, and it has shown promising results in some studies with AUD patients.

The design of our rhesus monkey model was based on a similar study in squirrel monkeys with cocaine (Nic Dhonnchadha et al., 2010). Similar to the squirrel monkey study, rhesus monkeys underwent repeated cycles of self-administration followed by extinction, a cue reactivity test, and then reacquisition of self-administration. Using this within-subjects approach, all monkeys received conditions randomly in a testing block (vehicle, one dose of DCS, two doses of RY-023). Unlike the squirrel monkey study, but because of the potential for order effects, rhesus monkeys underwent four blocks of testing; and for a given monkey, the order of test conditions varied across blocks. Thus, each test condition was evaluated four times across the course of the study and the outcomes of the four determinations of a condition were averaged for each monkey. It should be noted that this design in which all phases of the study were conducted in the monkey’s home cage, differs from the typical design of preclinical/clinical studies with humans in which the experimental manipulation (extinction therapy with or without adjunctive pharmacotherapy) occurs in an environment (e.g., lab) where alcohol/drug taking does not usually occur. There is substantial evidence to suggest, though, that distal cues (e.g., environments/people associated with drug taking) are just as important to cue reactivity as proximal cues (e.g., exposure to favorite drink, drug-taking paraphernalia; e.g., Ahnallen and Tidey, 2011; Conklin et al., 2008). Further, it has been suggested that in order to reduce craving through extinction procedures, it is important to use a wide range of stimuli that help to promote generalization of responses in different contexts and real-life situations (e.g., Garcia-Rodriguez et al., 2012). If this is the case, then, perhaps the present design represents a more translationally-relevant approach for evaluating the usefulness of adjunctive therapies in the context of CET.

Across the study, ethanol self-administration remained stable for the group of monkeys, regardless of whether the variable of interest was intake (in terms of volume or dose) or number of responses. This stability likely can be attributed to the experience of the subjects in the basic self-administration model and the parameters of the self-administration session (e.g., limited access v. “free” access; cf. Roberts et al., 2002). Under extinction conditions and vehicle pretreatment, responding dropped immediately on day 1 to 15–30% of active self-administration levels, depending on the subject. It then took an additional 5–9 days to meet the extinction criterion. When re-exposed to the ethanol-paired cues in the cue reactivity test, responding increased significantly despite the absence of ethanol as a reinforcer. When given the opportunity to resume ethanol self-administration, it took approximately 7 days for the monkeys to return to “baseline” levels of intake. These results obtained under vehicle pretreatment delineate the control conditions of our CET model, and based on these results, there is obvious room for improvement in multiple outcomes of the model (e.g., time to extinguish, level of cue reactivity, time to reacquire) with adjunctive pharmacotherapy.

As has been shown in previous preclinical and clinical studies with alcohol, as well as with other drugs of abuse (e.g., cocaine: Nic Dhonnchadha et al., 2010; Thanos et al., 2011a; 2011b), DCS influenced the rate of extinction of responding in the presence of ethanol-paired cues, significantly reducing the time required to achieve the extinction criterion. Likewise, RY-023 reliably and dose-dependently facilitated the rate of extinction. Interestingly, in the context of fear conditioning, nonselective GABA_A inverse agonists have been shown to disrupt, rather than enhance, extinction learning (Harris and Westbrook, 1998; Kim and Richardson, 2009). This difference from the present findings suggests that selectivity for α5 subunit-containing GABA_A receptors and/or the nature of the cue (fear-related v. positive/motivational) may underlie the favorable outcome with RY-023. Importantly, the results with both DCS and RY-023 support the notion that cognitive enhancing drugs, regardless of whether they target glutamate or GABA systems, can enhance the rate of extinction to conditioned drug cues.

In cue reactivity tests, although all of the drug conditions decreased responding to the ethanol-paired cues to some degree, only the highest dose of RY-023 produced significant reductions. Given the assumption that during extinction the cognitive enhancing drugs were facilitating new learning that should counter the conditioned reinforcing effects of the ethanol-paired stimuli (Davis et al., 2006), it is somewhat surprising that DCS did not reliably reduce cue reactivity. However, this observation is consistent with other preclinical and clinical studies showing a lack of effect of DCS in drug cue re-exposure scenarios (e.g., cocaine: Nic Dhonnchadha et al., 2010; nicotine: Kamboj et al., 2012, but see Santa Ana et al., 2009), or even a worsening of cue-elicited craving (Price et al., 2009). In preclinical studies where DCS did attenuate responding in a cue reactivity test (e.g., Vengeliene et al., 2008), the effect appeared to be nonspecific (i.e., reduced responding induced by cues associated with ethanol reinforcement and by cues associated with non-ethanol [water] reinforcement). Further, in clinical studies that report DCS as having efficacy to reduce cue reactivity, the study populations have mainly consisted of detoxified, abstinent AUD patients (Kiefer et al., 2015) or treatment-seeking individuals (MacKillop et al., 2015). In clinical studies that report no efficacy of DCS, the study populations consisted of people who drink excessively in social settings (Kamboj et al., 2011) or non-treatment-seeking people with hazardous drinking (Hofmann et al., 2012). These disparate findings suggest that DCS’s capacity to reduce cue reactivity may be largely dependent on the motivation of the subjects to achieve reductions in drinking/abstinence. Given that the monkeys had no external motivation to achieve abstinence and that drug treatment always was initiated off a drinking baseline (rather than abstinence baseline), perhaps it is not surprising that DCS failed to alter cue reactivity in the current study.

That RY-023 reduced cue reactivity is consistent with a report that another α5GABA_A receptor-selective inverse agonist (L-655,708) inhibited ethanol seeking induced by re-presentation of ethanol-paired cues in rats trained in a cue-induced reinstatement procedure (Chandler et al., 2019). Unlike the present study, in the cue-induced reinstatement study, reinstatement tests were conducted after extinction sessions during which both ethanol and ethanol-paired cues were omitted; thus, the cues presumably remained salient to the subjects. That L-655,708 effectively blocked reinstatement under these conditions indicates that cue-exposure training in the context of CET (and the current study) is not necessary for α5GABA_A receptor inverse agonists to reduce responding induced by ethanol-paired cues.

When ethanol again was made available for self-administration, only DCS reliably delayed reacquisition. In fact, in 3 of the 4 monkeys, it took a month or more before they returned to baseline levels of self-administration. This effect of DCS is consistent with some preclinical (Vengeliene et al., 2008) and clinical studies of CET for AUD (MacKillop et al., 2015), and preclinical studies with cocaine (Nic Dhonnchadha et al., 2010). This delay in a return to self-administration by DCS has been attributed to its capacity to augment consolidation of extinction learning rather than by simply facilitating the rapidity of extinction per se (Nic Dhonnchadha et al., 2010). This interpretation appears to have merit in light of the lack of effect of RY-023 on reacquisition of ethanol self-administration. In the context of other learning/memory tasks, the α5GABA_A inverse agonist α51A-II improved encoding and recall of memory, but did not enhance memory consolidation (Collinson et al., 2006). Together, these results suggest that drugs that enhance consolidation of extinction learning specifically should be considered for adjunctive pharmacotherapies in the context of CET. A possible candidate might be cannabidiol, a non-psychotomimetic cannabinoid that has been shown to facilitate extinction of fear memories in rats (Bitencourt et al., 2008) and to enhance consolidation of fear extinction in humans (Das et al., 2013). Whether cannabidiol would be effective in the context of CET for substance use disorders remains to be determined.

In summary, the present results demonstrate that both DCS and/or RY-023 can improve outcomes in a CET model, depending on the variable of interest. That neither drug uniformly improved all measures of the model likely reflects the complex interplay of GABA and glutamatergic systems in different aspects of extinction learning (for review see Davis and Myers, 2002). Further, because cue reactivity could be attenuated without a concomitant delay in reacquisition of self-administration, our results suggest that cue reactivity levels do not predict rate of reacquisition. This lack of predictability has been noted in some clinical studies (Rohsenow et al., 1994; Witteman et al., 2015) and has been attributed primarily to characteristics of the study population (e.g., severe AUD). While the monkeys cannot be classified as having severe AUD, they do have a long history of chronic moderate ethanol self-administration (5 days/week) that may have contributed to the disconnect between cue reactivity and reacquisition. Thus, from a clinical perspective, the choice of adjunctive pharmacotherapy in the context of CET should be determined on the basis of the outcome of interest. If the goal is to decrease cue reactivity, then targeting the GABA system might be fruitful; if the goal is to delay the return to drinking, then targeting the glutamatergic system might prove useful.

Highlights.

• Adjunctive pharmacotherapies may improve outcomes of cue-exposure therapy (CET).

• D-cycloserine facilitated extinction and slowed resumption of self-administration.

• RY-023 facilitated extinction and attenuated cue reactivity.

• Glutamatergic and GABAergic compounds differentially enhance specific CET outcomes.