A Further Assessment of a Role for Toll-Like-4 Receptors in the Reinforcing and Reinstating Effects of Opioids

Abstract

The Toll-like receptor 4 (TLR4) antagonists, (+)-naloxone and (+)-naltrexone, have been reported to decrease self-administration of opioids in rats and to reduce other preclinical indicators of abuse potential. However, under the self-administration conditions studied, the effects of TLR4 antagonists were not reinforcer selective, questioning the involvement of those receptors and their mediated inflammatory response in the abuse potential of opioids. The objectives of the current study were to further characterize the reinforcer specificity of TLR4 antagonism in opioid self-administration and to explore its effects in a preclinical model of craving/relapse. The TLR4 antagonist (+)-naltrexone decreased responding in rats trained to self-administer the µ-opioid receptor agonist remifentanil, but with a potency that was not significantly different from that observed in another group of subjects in which responding was maintained by food reinforcement. Responding reinstated by heroin injection was decreased by (+)-naltrexone, however, a similar reduction was not reproduced with administration of another TLR4 antagonist, lipopolysaccharide from Rhodobacter sphaeroides, administered into the NAcc shell. Thus, TLR4 antagonists lacked reinforcer selectivity in reducing opioid self-administration and were not uniformly effective in a model of craving/relapse, suggesting limitations on the development of (+)-naltrexone or TLR4 antagonists as treatments for opioid abuse.

INTRODUCTION

The pattern-recognition receptor, Toll-Like Receptor 4 (TLR4), is expressed in central nervous system (CNS) microglia and reacts to foreign entities, triggering release of pro-inflammatory and neuro-excitatory mediators (Li et al. 2016). Recent studies have shown that these glia-based inflammatory responses can be elicited by drugs, and it has been suggested that such reactions mediated by TLR4 may be involved in the abuse of opioids, such as heroin, or stimulants, such as cocaine. Further, TLR4 has been proposed as a potential target for the development of treatments for substance abuse disorders (Bachtell et al. 2015; Bachtell et al. 2017).

The discovery that the (+)-enantiomers of naloxone and naltrexone (Hutchinson et al. 2012; Hutchinson et al. 2008; Lewis et al. 2012) are TLR4 antagonists provided additional pharmacological tools to study the involvement of TLR4 on the abuse-related effects of drugs. These compounds block TLR4 mediated effects of lipopolysaccharide (LPS), with similar potency (Wang et al. 2016), cross the blood-brain barrier, bind to the MD2/TLR4 complex and lack the affinity for µ-receptors possessed by their corresponding (−)-enantiomers (Iijima et al. 1978). The µ-opioid receptor agonists remifentanil and morphine both bound the MD2/TLR4 complex associated with inflammatory response, and remifentanil self-administration was decreased by (+)-naltrexone treatment (Hutchinson et al. 2012). Further, both place conditioning and the elevation in nucleus accumbens dopamine produced by morphine were inhibited by (+)-naloxone (Hutchinson et al. 2012).

The suggestion that TLR4 may serve as a target in the development of treatments for substance-abuse disorders would be further substantiated if there was specificity in the effects of TLR4 antagonists on the abuse of drugs. More specifically, many preclinical assessments of potential treatments for drug abuse have looked for effects on drug self-administration at doses that have little or no effects on comparable responding established by more conventional reinforcers (Mello and Negus 1996). Tanda et al. (2016) compared the effects of both (+)-naloxone and (+)-naltrexone on responding maintained by remifentanil and food reinforcement. Each of the TLR4 blockers decreased drug self-administration at doses that also affected responding maintained by food reinforcement thus indicating a lack of reinforcer selectivity.

Tanda et al. (2016) trained subjects to self-administer cocaine with remifentanil available only in selected sessions. As the history of drugs self-administrated can be an important determinant of the reinforcing effects of substituted drugs (Young et al. 1981), and potentially the effectiveness of pharmacological pretreatments, it was deemed important to extend previous reports to studies of subjects having exclusive exposure to opioids. Moreover, repeated versus limited exposure to opioids might lead to quantitatively or qualitatively different effects on inflammatory pathways in turn potentially affecting related glia-targeting interventions. For example, repeated (five days) but not acute morphine exposure to rats produced a region-specific increase in the levels of glial fibrillary acidic protein, a protein expressed in astrocytes (Beitner-Johnson et al. 1993). Thus, the absence of opioid-specific effects of TLR4 blockers reported by (Tanda et al. 2016) may have resulted from the particular self-administration conditions under which the treatment was evaluated.

Responding after or during extinction (i.e. when reinforcement is withheld) has been proposed as a model of relapse and drug craving (Shaham et al. 2003). Theberge et al. (2013) evaluated in rats the effects of (+)-naltrexone on responding during extinction 13 days following self-administration training (because those response rates in extinction at 13 days were greater than those immediately after self-administration the authors referred to these rates as reflecting “incubation of craving”). Continuous administration of the TLR4 antagonist (+)-naltrexone during that 13-day interim period by osmotic minipumps decreased responding during extinction when the self-administered drug had been heroin but not methamphetamine, nor during extinction following food reinforcement. The authors suggested a critical role of TLR4 in the “incubation” of heroin craving.

Administration of morphine has been shown to induce a rapid expression of inflammatory genes, cytokine and chemokine and of TLR4 in the NAcc (Schwarz and Bilbo 2013; Schwarz et al. 2011, 2013). The involvement of this intra NAcc inflammatory process in opioid craving/relapse has been assessed in extinction after place-conditioning procedures in which drug-prime injections reinstated the preference for a compartment previously paired with morphine administration. In these models, environmental (neonatal handling) or pharmacological interventions (administration of the anti-inflammatory ibudilast) that prevented opioid-induced inflammatory response and TLR4 expression in the NAcc during adolescence also blocked the reinstatement of preference for the opioid-paired compartment later in life (Schwarz and Bilbo 2013; Schwarz et al. 2011). These studies identified the NAcc as a potential crucial area in which inflammatory cascades might mediate some of the relapse-related effects of opioids through mechanisms involving TLR4 and cytokines (i.e. interleukin 10, IL-10). Nevertheless, a more direct evaluation of the effect of blockage of TLR4 localized in the NAcc on behavior within a model of craving/relapse has not yet been reported.

The present study extended the characterizations of the effects of TLR4 antagonists in animal models of opioid abuse (self administration) and craving/relapse (reinstatement). In particular, effects of (+)-naltrexone were evaluated in rats that had daily sessions of remifentanil self administration and compared the effects with those obtained in subjects trained to respond under a similar schedule of food reinforcement. In a second set of experiments, the TLR4 antagonist (+)-naltrexone was tested on reinstated responding during extinction after heroin self administration. Initial studies of effects with (+)-naltrexone were followed by studies of direct NAcc infusion of lipopolysaccharide from the bacterium Rhodobacter sphaeroides (LPS-RS), a TLR4 antagonist (Stevens et al. 2013).

METHODS

Subjects

Male Sprague-Dawley rats (Taconic Farms, Germantown, NY) served as subjects and were acclimated to a temperature- and humidity-controlled vivarium for at least one week, with food (2018 Teklad global 18% protein rodent diets, Envigo, Huntingdon, UK) and water available at all times in their home cages. The housing room was under a 12:12-h light/dark cycle with lights on at 7:00 AM. After acclimation, body weights of subjects in the reinstatement experiments were maintained at approximately 320 g whereas remifentanil self-administering rats and a comparison group of subjects studied with food-reinforcement were maintained at 350 g by adjusting daily food rations. The animal facilities are fully accredited by AAALAC International and all procedures were conducted in accordance with the guidelines of the Institutional Care and Use Committee of the NIDA Intramural Research Program.

Apparatus

Daily experimental sessions were conducted with subjects placed in experimental chambers (25.5 × 32.0 × 26.5 cm, modified ENV-203; Med Associates) that were contained within sound-attenuating cubicles equipped with ventilation fans. White noise was present throughout sessions to mask extraneous sounds. Ambient illumination was provided by a lamp in the top center of the front panel (house light). Centered 2.0 cm above the floor on the front wall of each chamber was a 5.0- x 5.0-cm opening through which food pellets (45 mg, BioServ, Frenchtown, NJ) could be delivered. Mounted 4.0 cm above the grid floor and 5.0 cm on both sides of the midline were response levers with three LEDs positioned in a row 10 cm above each. A downward displacement of either lever with a force of 0.2 N defined a response, and produced an audible feedback click of a relay mounted behind the front wall of the chamber.

Responses on the right lever were capable of producing drug injections as per procedures detailed below. Drug injections were delivered (i.v.) by an injection system which consisted of a syringe pump (Model 55–2222, Harvard Apparatus, Holliston, MA) connected to a surgically implanted i.v. catheter by Tygon tubing threaded through a balance arm to one end of a single-channel fluid swivel (375 Series Single Channel Swivels, Plymouth Meeting, PA). The other end of the swivel was connected by polyethylene tubing (BPTE-50) to the mid-scapular externalization of the subject’s catheter. The tubing was protected by a surrounding metal spring. During drug self-administration sessions, a 10-ml syringe containing a drug solution was mounted within the syringe pump.

Surgery

Subjects were implanted with a chronic indwelling catheter in the right jugular vein that was externalized in the mid-scapular region, as described previously (Hiranita et al. 2011; Weeks 1962). In some experiments bilateral intracranial guide cannulae were also stereotaxically implanted to a depth of 1 mm above the NAcc shell. Surgical procedures were conducted with subjects under ketamine (60.0 mg/kg) and xylazine (20.0 mg/kg, i.p.) anesthesia.

For stereotaxic cannula placements, the skull was exposed, and two small holes were drilled to expose the dura for bilateral implantation of guide cannulas (26 gauge; 14 mm) above the NAcc shell (with nose bar = −3.3 mm, measured from bregma: anterior/posterior, + 1.5 mm; lateral, ± 1 mm; dorsal/ventral, −6.5 mm), as described previously (Tanda et al. 2005; Tanda et al. 2007). Coordinates were based on the rat brain atlas (Paxinos and Watson 1998) and Figure 1 shows confirmed placements. All subjects were allowed to recover from surgery for approximately seven days before self-administration studies were initiated. Venous catheters were subsequently infused daily with 0.2 ml of a sterile saline solution containing heparin (30.0 IU/ml) and penicillin G potassium (250,000 IU/ml) to minimize the likelihood of infection and the formation of clots or fibroids.

Fig. 1:

Coronal sections (Paxinos and Watson, 1998), showing selected placements (filled circles) of the tip of internal cannulas used for the microinjection procedure. Distance from Bregma is indicated on the left side of each coronal section.

Remifentanil self administration and food-maintained responding

Subjects were initially trained with food reinforcement to respond on the right lever and were subsequently trained under a fixed ratio (FR) 5-response schedule of reinforcement in which each fifth response produced a food pellet. Six of the subjects continued with food reinforcement, whereas the remaining six subjects were surgically implanted (as described above) with chronic-indwelling jugular-vein catheters, and after recovery trained to self-administer remifentanil. Animals were randomly assigned to the two groups. Both food and remifentanil sessions were divided into five 20-min components, each proceeded by a 2-min timeout (TO) period in which responding had no scheduled consequences other than feedback clicks. The first component of extinction was followed by successive components in which the available dose of remifentanil was increased in ascending order to capture in each subject both the ascending and the descending limb of the response-rate dose-effect curve. Similarly, for the subjects maintained on the schedule of food reinforcement, the number of food pellets per ratio completed was 1, 2, 3 or 4 (45, 90, 135 or 180 mg) in successive components following the initial extinction component.

Response rates were calculated as responses divided by elapsed time in each component (excluding responses and time during drug injections or food presentations and the following TO). In each subject, test sessions were conducted when the following criteria were met for two consecutive sessions: total number of reinforcers earned did not vary more than 25% and the component in which the maximum response rate during the session was the same or contiguous to that in the previous two sessions. Subjects in which responding was maintained by food reinforcement had 22-h access to chocolate-flavored pellets (Bio-Serv, Flemington, NJ) before each test session (i.e. when a drug pretreatment was administered). Access to chocolate-flavored pellets was not provided before training sessions. This procedure reduced response rates by 2- to 3-fold, rendering response rates in the groups more comparable.

Saline, (+)-naltrexone (32–100 mg/kg), or (−)-naltrexone (0.032 mg/kg) were administered s.c. 15 min before test sessions. The test compound (−)-naltrexone was evaluated in 3 of the animals responding with food reinforcement, whereas all the other doses of pretreatment drugs were tested in 5–6 animals.

Heroin self-administration training and extinction

Subjects were trained to self-administer heroin during 14 daily 3-h sessions. Sessions consisted of an initial 10-s TO period in which all lights were off and responding had no programmed consequences. Subsequently the house light was turned on, and each response (fixed-ratio one-response, or FR 1, schedule) produced a heroin injection accompanied by illumination of the LEDs above the lever, and the house light turning off. The LEDs were turned off and the house light was re-illuminated after the 2.3-s activation of the injection pump, during which responses had no scheduled consequences other than the feedback clicks. Heroin was dissolved in sterile saline and infused in a volume of 43 μl over 2.3 s at a dose of 0.1 (first seven sessions) and 0.056 (last seven sessions) mg/kg/injection.

Following the last session of heroin self-administration, subjects were exposed to daily 3-h extinction sessions. Extinction sessions were identical to heroin self-administration sessions with the exception that no syringe was placed in the injection pump and therefore responding on the active lever did not produce injections but did produce the stimuli previously paired with the injection of heroin. Subjects were returned to their home cages in the animal vivarium after each session. Subjects continued under extinction conditions for at least 14 and no more than 25 sessions until there was an average of less than 40 responses during the last 3 sessions. In a pilot study with 5 subjects, 40 responses approximated an asymptote obtained by fitting a one phase exponential decay model to data from 14 extinction sessions. One animal had health complications not related to drug effects and could not be tested whereas two animals did not meet extinction criteria over the course of 25 sessions and were removed from the study.

Reinstatement and microinjection procedures

After extinction, all subjects were adapted to injection procedures by administering a saline injection (1 ml/kg, s.c.) and/or a microinjection of 0.5 µl of saline solution. The next session, subjects underwent reinstatement testing which was in all respects, other than the pre-session injections, identical to the extinction sessions. Subjects were injected s.c. with saline or heroin (0.56 mg/kg) and saline or (+)-naltrexone (15 mg/kg), and after 15 min they were placed in the chamber for 3 h during which responses on either lever were recorded (systemic administration groups). A total of sixteen subjects were used for this study (saline-heroin: n= 6; (+)-naltrexone and heroin: n = 5; (+)-naltrexone and saline: n=5).

Other groups of subjects (microinjection groups) received, instead of (+)-naltrexone, a bilateral injection of LPS-RS (2.5 µg/0.5 µl /side) or saline (0.5 µl) into the NAcc shell, delivered over 1 min. Both microinjectors (33 gauge; 16 mm) were attached by polyethylene tubing to a 10 μl Hamilton syringe with injections being driven by a microinjection pump (Harvard Apparatus, USA). The injectors remained in position for 1 min after the completion of the microinjection to ensure that the entire dose had diffused. A total of seventeen subjects were used for this study (saline-heroin: N= 6; LPS-RS and heroin: N = 6; LPS-RS and saline: N=5).

Each rat was tested only once. The dose of heroin used in the reinstatement experiment (0.56 mg/kg, i.p.) was selected as the one inducing the highest level of responding in a pilot study. In each of the studies (systemic or microninjection), animals were assigned to the different treatments by minimizing the differences between groups in the number of active-lever responses in the last day of extinction.

After the session, the subjects were anesthetized, decapitated, and brains were removed and stored in 10% formalin before sectioning. The brains were sectioned in the coronal plane (50 μm) and verified for cannula placement under a light microscope.

Drugs

Heroin (3, 6-diacetylmorphine hydrochloride) was obtained from the NIDA Drug Supply Program. Ketamine, xylazine and (−)-naltrexone were purchased commercially. The isomer (+)-naltrexone was generously provided by Dr. Kenner Rice of the Chemical Biology Research Branch (NIDA-IRP, NIH). LPS-RS (Rhodobacter sphaeroides LPS) was purchased from InvivoGen (San Diego, CA, USA). Remifentanil HCl was purchased from Hospira (Lake Forest, IL). All drugs were dissolved in 0.9 % a sterile NaCl solution (saline).

Statistical analysis

Rates of responding maintained by remifentanil injection or food presentation during the FR components of the experimental sessions (excluding TO periods) were plotted as a function of dose. For each reinforcer, a repeated-measures ANOVA was performed, with dose of treatment [(+)-naltrexone or (−)-naltrexone] and magnitude of reinforcer as independent factors and response rates as the dependent variable. For each reinforcer magnitude, a Dunnett’s multiple comparisons test was used to compare means after vehicle versus after administration of the treatment. The dose-effect curve of remifentanil was determined twice in the same animals. A repeated-measures ANOVA with determination number and dose of remifentanil as independent factors and response rates as the dependent variable was performed, and followed by Dunnett’s multiple comparisons.

To assess specificity of drug effects on remifentanil self administration response rates after administration of (+)-naltrexone were normalized by the corresponding value after administration of vehicle for each subject and dose of (+)-naltrexone. Normalized data were then analyzed by a repeated-measures ANOVA, with dose of the treatment and reinforcer as independent factors and percent baseline response rats as the dependent variable, followed by Dunnett’s post-hoc tests.

To assess whether heroin self-administration extinguished, the numbers of responses on the active lever during the last three heroin self-administration sessions were compared to corresponding means from the last three extinction sessions using a one-tailed paired t-test. A one-way ANOVA indicated that there were no significant differences between groups prior to treatments in the mean numbers of active and inactive responses during the last sessions of extinction.

Effect of TLR4 antagonism on prime-induce reinstatement was assessed using a two-way repeated-measures ANOVA, with session (last day extinction vs heroin prime) and drug treatment (vehicle, (+)-naltrexone or LPS-RS) as factors, and numbers of responses as dependent variables, followed by Bonferroni post-hoc tests. This was done for both active and inactive responses. The numbers of responses after heroin-prime injection in the systemic versus microinjection group were compared by unpaired t test with Welch’s correction. Data were displayed as the mean ± SEM. The level of significance for ANOVAs and post-hoc tests was set at p<0.05.

RESULTS

Remifentanil self-administration and food-maintained responding

Remifentanil maintained increasing rates of responding as a function of doses from extinction (no drug available) to 1 μg/kg/injection. The dose-effect curve had an inverted-U shape (Fig. 2a, b; filled symbols) with decreases in rates of responding at the highest doses. Prior injection (s.c.) of (−)-naltrexone (0.032 mg/kg) produced, as expected, a rightward (approximately 12-fold) and downward shift in the remifentanil dose-effect curve (Fig. 2a). Statistical analysis indicated significant effects of (−)-naltrexone (F_1,43=9.02, p<0.01), of the unit dose of remifentanil (F_5,43=8.35, p<0.001) and of their interaction (F_5,43=3.27, p<0.05). In comparison with vehicle, administration of 0.032 mg/kg (−)-naltrexone produced a significant reduction of response rates at 1 µg/kg/inf (t₄₃=4.66, p<0.001) and a trend of reduction at 0.32 µg/kg/inf (t₄₃=2.65, p=0.05)

Fig. 2:

Effects of the µ-opioid receptor antagonist (−)-naltrexone (s.c.) and the TLR4 antagonist (+)-naltrexone (b.,d.) or vehicle (veh) on the response-rates (mean ± SEM) of rats self-administering the opioid remifentanil (a., b.) or food (c.,d.). Vertical axes: Response Rate (resp/sec). Horizontal axes: Unit dose of remifentanil (µg/kg; a., b) and number of 20 mg pellets (c.,d.).

Food reinforcement maintained increasing rates of responding as a function of number of pellets from extinction (no pellets available) to four pellets, with an inverted-U shape (Fig. 2c, d; filled symbols). Prior injection (s.c.) of (−)-naltrexone (0.032 mg/kg) did not significantly alter response rates across the range of pellets per ratio completed (dose of (−)-naltrexone: F_4,18=0.37, NS; Fig. 2c).

Administration of (+)-naltrexone dose-dependently shifted the remifentanil dose-effect curve downward and to the right (Fig. 2b). There was a significant effect of (+)-naltrexone (F_3,86=4.49; p < 0.01) and remifentanil (F_5,86=12.5, p<0.001) doses, but no significant interaction (F_15,86=1.47, NS). Post-hoc analysis indicated that in comparison with saline, doses of (+)-naltrexone decreased response rates maintained by remifentanil at doses of 32, 56, and 100 mg/kg at 1.0 µg/kg of remifentanil (t₈₆= 3.29, 3.17 and 4.82, respectively, with all p values < 0.05). At 100 mg/kg of (+)-naltrexone the same comparison also indicated significant decreases in response rates maintained by 3.2 µg/kg/inf (t=3.03, df=86, p<0.05). Statistical analysis (2-way ANOVA) of the two determinations of the remifentanil dose-effect curves after saline pretreatment (Figure 2a and 2b) indicated a significant effect of remifentanil dose (F_{5, 43}=16.33, p<0.001) but no significant effect of the determination (F_{1, 43}=0.84, NS).

Treatment with (+)-naltrexone also decreased response rates maintained by food reinforcement (Fig. 2d), to less of an extent at the lower two doses than at the 100 mg/kg dose. The ANOVA revealed a significant effect of the dose of (+)-naltrexone (F_3,90=7.12, p<0.001), the number of pellets per ratio (F_4,90=22.5, p<0.001) and of their interaction (F_12,90=1.88, p<0.05). Post-hoc tests showed that in comparisons with saline, decreases in response rates were obtained after administration of 32 mg/kg (+)-naltrexone at 4 pellets (t₉₀=2.54, p=0.05) and by 100 mg/kg (+)-naltrexone at 3 (t₉₀=3.78, p<0.001) and 4 pellets (t₉₀=4.45, p<0.001).

Figure 3 compares the dose-related decreases produced by (+)-naltrexone on responding maintained by 1.0 µg/kg/injection of remifentanil or three food pellets (60 mg). For both remifentanil and food, (+)-naltrexone produced a decrease in the baseline response rates. Statistical analysis indicated a significant effect of the dose of (+)-naltrexone (F_3,29=4.33, p<0.05) but not of the type of reinforcer (F_1,10=0.06, NS) nor of an interaction of the two factors (F_3,29=0.54, NS). Post-hoc tests did not indicate any significant differences between the effects of (+)-naltrexone on response rates in the remifentanil vs food groups.

Fig. 3:

Effects of (+)-naltrexone on the % Control Response Rates (mean ± SEM) of remifentanil (1 µg/kg/inf; filled circles) or food (3 × 20 mg pellets; open circles).

Acquisition and extinction of heroin self-administration

At the 0.1 and subsequent 0.056 mg/kg/injection doses of heroin there were progressive increases in the number of active-lever responses across sessions and a consequent increase in number of injections received and total intake of heroin (data not shown). In contrast, the number of inactive responses did not increase across sessions (Fig 4: gray symbols).

Fig. 4:

Extinction of responding (mean ± SEM; active lever: black circles; inactive lever: grey circles) in rats self-administering heroin (3h session). Extinction was identical to training with the exception that lever pressing did not result in drug delivery. The gray area in the bottom panel indicates the number of active lever responses (± SEM) in the last day of acquisition (unit dose of heroin: 0.056 mg/kg/inf). Vertical axes: Lever Presses. Horizontal axes: Session; labels L3 and L2 indicate indicates the third and second to last session whereas L1 the last extinction session in which criteria were met.

During the first extinction session, the number of responses increased on both the active and inactive levers (Fig. 4). Thereafter response rates decreased on the previously active lever to levels below those previously maintained by heroin. A significant reduction in responses on the active lever was present with the mean of the last three extinction sessions compared to the mean of the last three heroin self-administration sessions (t₃₂=3.29, p<0.05). The number of sessions to extinction criteria (at least 14 sessions and active responses < 40) was 16.24 ± 0.53 (mean ± SEM).

Effects of (+)-naltrexone on heroin-induced reinstatement

In the last sessions of extinction neither active- (F_2,15=1.07, ns) nor inactive-lever responses (F_2,15=0.81, NS) after saline administration were significantly different among the treatment groups (Fig. 5 open bars). Heroin (0.56 mg/kg, s.c.) and saline injections (Fig. 5, upper panel) increased responding on the active lever (leftmost solid bar) when compared to that during the immediately preceding last session of extinction (corresponding unfilled bar). However, when the heroin injection was combined with (+)-naltrexone (15 mg/kg, s.c.) there was no increase in active-lever responding (Fig. 5, upper middle two bars). Naltrexone and saline injections did not significantly alter active-lever responding compared to the preceding last session of extinction (Fig. 5, rightmost two bars). The two-way repeated-measures ANOVA of responding in the reinstatement test indicated significant effects of session, drug treatment and an interaction between the two factors (F_1,13=29.0; F_2,13=8.88; F_2,13=14.2; p<0.05). Post-hoc tests revealed that heroin increased responding on the active lever compared to responding during last session of extinction (t₁₃=7.79, p˂0.001), and that this increase was blocked by (+)-naltrexone (t₁₃=1.13, p<0.05). Additionally, (+)-naltrexone and saline did not significantly alter responding compared to the preceding last session of extinction (t₁₃=0.653, NS). Finally, there were no significant drug-induced differences between the three groups in responding on the inactive levers during the reinstatement test versus the last extinction session (Fig. 5, lower panel).

Fig. 5:

Effects of (+)-naltrexone (15 mg/kg, s.c.) or saline on extinction active (top panel) and inactive (bottom panel) lever responding (mean ± SEM) after an injection of heroin (0.56 mg/kg, s.c.) or saline. Open bars indicate sessions in which pretreatment was two saline injections whereas filled bars represent test sessions in the corresponding subjects. Vertical axes: Upper panel. Active Lever Responses; Lower panel: Inactive Lever Responses. Horizontal axes: Pretreatment administered in test sessions.

Effects of LPS-RS injection into NAcc shell on heroin-induced reinstatement

In the last sessions of extinction neither active- (F_2,16=1.47, NS) nor inactive-lever responses (F_2,16=0.58, NS) after saline administration were significantly different among the treatment groups (Fig. 6 open bars). Heroin injection (0.56 mg/kg, s.c.) with saline (0.5 µl) microinjection into the NAcc shell (Fig. 6, upper panel) increased responding on the active lever (leftmost solid bar) when compared to that during the immediately preceding last session of extinction (corresponding unfilled bar). Further, the same effect of heroin was found when it was accompanied by LPS-RS (2.5 µg/0.5 µl /side) microinjection into the NAcc shell. Finally, LPS-RS microinjection with saline (s.c.) did not significantly alter active-lever responding from last session of extinction. A significant main effect of session (F_1,14=31.5, p<0.005) and session X drug treatment interaction that approached significance (F_2,14=10.3, p=0.054) was found in the analysis of active lever responding in the reinstatement test (Fig. 6, upper panel). Post-hoc tests revealed that heroin accompanied by microinjection of saline increased responding on the active lever compared to responding during last session of extinction (t₁₄=4.34, p˂0.01). Further, this increase was also obtained with heroin and LPS-RS microinjection (t₁₄=3.55, p˂0.01). Additionally, LPS-RS with saline (s.c.) injection did not significantly alter responding compared to the preceding last session of extinction (t₁₄=0.47, NS). Finally, no significant effect by ANOVAs or post-hoc analysis was detected for the inactive-lever responding in the three experimental groups (Fig. 6, lower panel).

Fig. 6:

Effects of microinjections of LPS-RS (2.5 µg/0.5 µl /side) or saline into the NAcc shell on extinction active (top panel) and inactive (bottom panel) lever responding (mean ± SEM) after an injection of heroin (0.56 mg/kg, s.c.) or saline. Open bars indicate sessions in which pretreatment was two saline injections whereas filled bars represent test sessions in the corresponding subjects. Vertical axes: Upper panel. Active Lever Responses; Lower panel: Inactive Lever Responses. Horizontal axes: Pretreatment administered in test sessions.

Responding (mean ± SEM) after heroin-prime injection was greater for the microinjection (71.8 ± 16) than for the systemic-injection group (53.2 ± 6.5), but that difference was not statistically significant (t₆=1.08, NS; Fig. 5, Fig. 6)

DISCUSSION

In the last decade, evidence has accumulated for an involvement of inflammatory processes and glia activation in models of opioid abuse (self administration) and craving/relapse (reinstatement) (Bachtell et al. 2015, 2017). In particular, (+)-enantiomers of opioid antagonists have been shown to have activity at TLR4, a receptor involved in inflammatory reaction to foreign entities, and have been shown to reduce self-administration of the opioid remifentanil as well as responding during extinction following a history of heroin self administration in rats (Hutchinson et al. 2012; Theberge et al. 2013). The aim of the present investigation was to further characterize the reinforcer-specificity of TLR4 antagonism in reducing opioid self-administration and to explore its possible effects in a preclinical model of opioid craving/relapse.

The TLR4 antagonist, (+)-naltrexone, reduced self-administration of remifentanil in the current study at a dose approximately 3,000-fold higher than a comparably effective dose of (−)-naltrexone. However, this effect was not specific for remifentanil-maintained responding, as (+)-naltrexone also decreased food-maintained responding with similar potency. The present results parallel recent findings of a lack of specificity between the effects of the TLR4 antagonists (+)-naloxone or (+)-naltrexone on remifentanil- or food-maintained responding in rats trained to self-administer cocaine (Tanda et al. 2016). In contrast to the study by (Tanda et al. 2016) subjects in the current study were tested but also trained with remifentanil, rendering them exposed to opioids on a daily rather than occasional basis. Further, the subjects in the present study were not exposed to stimulant drugs. Thus, the current findings extend the generalizability of previous reports, and suggest that previous negative findings with (+)-naltrexone were not restricted to the history of training with a stimulant drug. The present conclusion is supported also by the recent evidence that neither (+)-naloxone nor (+)-naltrexone attenuated heroin-induced-DA elevation in the NAcc of rats under a wide range of dosing and experimental conditions (Tanda et al. 2016) and that (+)-naloxone did not block the locomotor stimulating effects of morphine and heroin (Eriksen et al. 2016).

While reinforcers may differ in many features, their functional maintenance of behavior (baseline/control response rates) has been shown repeatedly to be a crucial determinant of the effects of many drugs (e.g. Dews 1955; Kelleher and Morse 1968). In the current study, the reinforcer selectivity of (+)-naltrexone was evaluated by comparing its effects on groups of subjects responding under fixed schedule of food presentation or remifentanil infusion. These groups also differed in their baseline response rate with food maintaining higher responding that remifentanil. This between-group difference was partially mitigated by the implementation of the present feeding conditions, but response rates were not identical, which represents a possible caveat to the generality of the present findings. Nevertheless, it is noteworthy that the current procedure detected effects of (−)-naltrexone that were reinforcer selective, thus providing a prima facia evidence supporting its use to evaluate the selectivity of the effects of (+)-naltrexone.

In the present study, systemic administration of (+)-naltrexone attenuated the prime-induced reinstatement of lever responding in subjects with a history of acquisition and subsequent extinction of heroin self-administration. This initial result suggested a possible involvement of microglia and TLR4 in opioid reinstatement, but is inconsistent with studies that failed to find an acute effect of systemic administration of microglia inhibitors in models of opioid reinstatement. In particular, injections (24 h or 30 min before) of minocycline, a drug that inhibits microglial activation, did not affect prime-induced reinstatement of morphine place conditioning (Schwarz and Bilbo 2013). However, that reinstatement of place conditioning was made effective by adolescent morphine exposure, an effect blocked by adolescent co-administration of ibudilast, which among other actions suppresses glial cell activation (Schwarz and Bilbo 2013), suggesting that the roles of microglia in this behavioral procedure are complex.

A previous study found acute administration of (+)-naltrexone to be ineffective in reducing “incubation of heroin craving” (i.e. increase in responding during extinction after a period without exposure to the self-administration chamber and drugs) (Theberge et al. 2013). However, with (+)-naltrexone delivered via osmotic minipumps over a thirteen-day period, doses from 7.5 to 30 mg/kg/day decreased the enhancement of responding compared to the vehicle condition. These varied outcomes might arise from differences in the acute vs. chronic dosing parameters. Another possibility is that (+)-naltrexone was effective in an experiment with a high control reinstated response rate, whereas no effect of (+)-naltrexone was obtained in experiments with a lower control reinstated response rates.

The present experiments also failed to find an effect of acute NAcc shell microinjections of LPS-RS in the reinstatement procedure. It is widely recognized that the mesolimbic system, including the NAcc in particular, is implicated in the reinforcing effects of drugs as well as in reinstatement induced by environmental context or priming drug injections (Bossert et al. 2012; Di Chiara et al. 1999; Pontieri et al. 1995). Single or repeated administrations of opioids in rats have been shown to produce a rapid structure-specific and age-dependent activation of microglia and astrocytes and to increase the expression of genes, including those for TLR4 and IL-10, that regulate inflammatory response in the NAcc (Lacagnina et al. 2017; Schwarz et al. 2011, 2013). Noteworthy, intra-VTA microinjections of LPS-RS, at doses as small as 0.1 µg/side, during the acquisition phase of the place conditioning in rats have been reported to reduce time allocation to the morphine-paired compartment (Brown et al. 2018). Thus, it seems unlikely that the absence of effectiveness of LPS-RS in the current reinstatement procedure was the result of inadequate dose and consequent limited blocking of TLR4 in the NAcc shell. Thus, the current experiment together with the studies cited above indicate that acute blockage of TLR4 and reduction of microglia activation does not affect the immediate heroin reinstatement of drug lever responding in adult subjects (Schwarz and Bilbo 2013). However, the involvement of morphine-induced glial activation may be more prevalent under specific developmental conditions (e.g. Schwarz and Bilbo 2013; Schwarz et al. 2011).

Several studies have shown that contingent and not contingent [PW1]administration of opioids (including remifentanil) can produce an activation of TLR4 and lead to inflammatory processes (Lacagnina et al. 2017; Schwarz et al. 2011; Schwarz et al. 2013). Nevertheless, the particular dosing conditions of the current experiments may have been insufficient to trigger such inflammatory response, thus precluding the detection of subsequent therapeutic effects of blockage of TLR4. However, that hypothesis would also imply that the effectiveness of TLR4 antagonists for opioid abuse might depend on particular pharmacological conditions and, therefore that their therapeutic use is limited.

The present findings do not rule out a broader involvement of some inflammatory signaling in the effects of opioids. For example, administration of plasmid DNA encoding for the anti-inflammatory cytokine IL-10 in the NAcc of rats has been shown to decrease self-administration of remifentanil but not responding maintained by food (banana pellets or sucrose pellets), and therefore to have reinforcer-selective effects (Lacagnina et al. 2017). Recently, three clinical trials have investigated the effects of the glia modulator ibudilast in opioid-dependent subjects and have reported promising but not always consistent results (Cooper et al. 2016, 2017; Metz et al. 2017). For example, in a concurrent progressive-ratio schedule in which subjects could choose between oxycodone and money, ibudilast reduced drug breakpoints and shifted downward the oxycodone dose-response curve (Metz et al. 2017). However, it is not clear whether that reduction was a specific allocation of behavior to the alternative reinforcer or of a general non-selective decrease in responding. A reduction in the positive subjective rating of opioids and of discontinuation-precipitated withdrawal by administration of ibudilast has been reported in some clinical trials, but not in others (Cooper et al. 2016, 2017; Metz et al. 2017).

In conclusion, TLR4 antagonists did not selectively affect self-administration of opioids across a broad range of doses or consistently attenuate a preclinical indicator postulated as predictive of craving/relapse. This finding is consistent with recent evidence obtained in behavioral as well as neurochemical procedures (Tanda et al. 2016) that employed the (+)-enantiomers of naltrexone and naloxone. Taken together these studies suggest that TLR4 antagonists have limited utility as a potential treatment for opioid abuse.