Chronic pain selectively reduces the motivation to work for remifentanil but not food reward

Chronic pain differentially modulates opioid taking when work effort is low vs high and causes a hypodopaminergic state that likely contributes to drug-seeking behavior.

Abstract

Currently, preclinical research has reported conflicting evidence as to whether chronic pain imparts resilience or vulnerability to opioid drug seeking. Here, we investigated the impact of chronic pain on the intravenous self-administration (IVSA) profile of the short-acting opioid analgesic remifentanil in a mouse model. Using a chronic constriction injury model of chronic neuropathic pain, 7 days after injury, male and female C57Bl/6J mice began remifentanil IVSA. During the acquisition phase, there were no differences in the total number of reinforcers earned but an increase in the number of active nose pokes in pain mice. An increase in the rate of acquisition within sessions was observed in male but not female mice. When work effort increased (fixed ratio 3 and progressive ratio), pain mice unexpectedly showed a reduction in the number of reinforcers earned and their breakpoint. This change in motivational state was specific to the willingness to work for remifentanil, as these changes were not observed with higher effort for a food reward. We hypothesized that chronic pain altered the dopaminergic state of the striatum, which would impact the motivation to work for a reward. We found that pain mice had significantly decreased phasic dopamine release assessed via fast-scan cyclic voltammetry and reduced potassium-evoked extracellular dopamine measured by microdialysis. Future studies will investigate the causal relationship between this hypo-dopaminergic state and decreased behavioral motivation associated with a chronic pain state.

1. Introduction

Chronic pain affects up to 25% of the US population. Opioids remain commonly used analgesics for alleviating acute and chronic pain and are considered highly efficacious based on a number-needed-to-treat analysis,¹⁷ albeit this metric has some inherent problems.²⁴ However, long-term use of opioids causes numerous adverse side effects. Opioids have high abuse liability, and despite great efforts to educate the public and physicians on the dangers of opioid overprescribing, opioid-related overdose deaths are at an all-time high (https://nida.nih.gov/research-topics/trends-statistics/overdose-death-rates). Critically, research studies report conflicting results as to whether chronic pain is a risk factor for developing opioid misuse and/or an opioid use disorder (OUD).

Previous studies in rodents and humans report mixed and often seemingly contradictory findings regarding the impact of chronic pain on the potential development of opioid misuse or abuse.¹² Rodent opioid self-administration studies have not consistently reported an effect of chronic pain on opioid drug-seeking behavior where the pain model, sex, when the opioid is started with respect to pain onset, opioid and dose administered, and species used likely accounts for at least part of this variability.^{3,4,15,19,22,35,39} This highlights the need for additional studies to investigate the impact of chronic pain on opioid drug-taking and drug-seeking behaviors.

Here, we aimed to parse the effects of chronic pain on opioid intake vs drug-seeking behavior using different operant demands in a mouse model of chronic pain. We designed an intravenous self-administration (IVSA) protocol for use in mice, so that future studies can take advantage of transgenic mice. We used remifentanil to assess the effects of chronic pain on various schedules of operant responding. Although self-administration in rodents is not a direct corollary for OUD, various IVSA operant schedules allowed us to investigate the impact of chronic pain on specific domains of drug use, including the acquisition of use, maintenance of use over time, and the motivation to seek the drug under increasing operant demands.¹⁴ We chose remifentanil for to its pharmacokinetics, with rapid onset and ultrashort duration of action,^25,43 allowing animals to learn operant behavior quite quickly and without the need for prior food training. However, remifentanil has been used in individuals with OUD and physical fentanyl dependence.²⁷ These outcomes enabled a more nuanced examination of the neurobiological impacts of chronic pain on various dimensions of opioid drug use in a rodent model.

We, and others, reported that chronic neuropathic pain modifies reward circuitry whereby opioid-evoked dopamine release in the ventral striatum is impaired.^6,16,44,45 In addition, dopamine neuronal excitability is reduced in chronic pain states.^36,47 Further, opioid-evoked release may in part be reduced because of receptor desensitization, where inflammatory pain reduces agonist-induced receptor activation.⁹ As such, we investigated how chronic pain modifies mesolimbic dopamine circuitry using electrically evoked dopamine release determined by fast-scan cyclic voltammetry and potassium-evoked dopamine release measured by microdialysis. In sum, we sought to expand mechanistic insights into the hypodopaminergic state that results from chronic pain and how this state interacts with opioid self-administration.

2. Materials and methods

2.1. Animals

Experiments were initiated when male and female C57BL/6J mice were 12 weeks old (The Jackson Laboratory, Bar Harbor, ME). Upon arrival, mice were group-housed until jugular catheter implantation, at which point they were singly housed. Mice were maintained in a temperature-controlled room on a reverse light–dark cycle, with lights off between 08:00 and 20:00 hours. All animals had free access to food and water except during the 2 hours drug self-administration testing. All behavior testing was performed during the dark phase. All procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, were preapproved by the University of California Los Angeles Institutional Animal Care and Use Committee, and were compliant with ARRIVE guidelines.

Mice were randomly assigned to groups before beginning experiments. Behavioral testing of male and female mice was conducted separately. All male mice underwent testing and were returned to housing, and all equipment was thoroughly cleaned before bringing female mice to the testing room. After a week of vivarium acclimatization, all mice were habituated to human touch by handling the mice for 10 minutes a day every day for a week before beginning experiments. Mice were habituated to the behavior room daily in their home cages for 30 minutes before being handled. Estrous cycle phases were not tracked in female mice because of the sex-specific additional stress.

The total number of mice used in this study was 83 mice. Separate groups of mice were used for food self-administration (N = 28 mice: 7 males and 7 female sham animals and 7 males and 7 females chronic constriction injury [CCI] mice) and remifentanil self-administration (N = 31: 8 males and 7 female sham animals and 8 males and 8 females CCI mice). Additional separate cohorts were used for either microdialysis (N = 12: 3 male and 3 female shams, as well as 3 male and 3 female CCI or voltammetry experiments [N = 12: 3 male and 3 female shams, as well as 3 male and 3 female CCI]).

2.2. Surgical procedures

2.2.1. Chronic constriction injury

Chronic constriction injury of the left sciatic nerve or sham surgeries were conducted in mice as we previously reported.^26,28,44 Mice were anesthetized with isoflurane (5% induction, 2% maintenance), and a 2-mm cuff constructed of PE20 polyethylene tubing (Intramedic) was placed on the main branch of the left sciatic nerve. Sham mice were maintained on isoflurane for a similar duration of time. The surgical site was prepped, but no incision was made. All mice received acetaminophen (∼3 mg, p.o.) 30 minutes before surgery and again 12 hours later. Mice also received 2 mL of lactated Ringer solution just after surgery to maintain hydration. Their eyes were protected from dehydration during surgery by a layer of sterile ophthalmic ointment. Mice were placed on a heated pad during surgery and recovered in a warm clean cage. Environmental enrichment was provided via 2 Nestlet squares for nest building throughout the experiments.

2.2.2. Jugular catheterization

Two or 3 days after CCI or sham surgeries, mice underwent a second surgery to implant catheters in the right jugular vein as we previously described.^20,42 Two or 3 days after CCI or sham surgeries, mice underwent a second surgery to implant a catheter in the right jugular vein under aseptic conditions using isoflurane anesthesia (5% induction and 2% maintenance) according to procedures previously described.^20,42 In brief, 12 mm of 1 Fr silicone tubing (0.2 mm i.d., 0.4 mm o.d., Norfolk Access, Skokie, IL) was inserted into the right vein and connected to a back-mounted 26-gauge stainless-steel guide cannula (315BM-8-5UP, Plastics One, Roanoke, VA). Catheters were flushed daily with 0.02 to 0.04 mL of sterile saline mixed with heparin (30 U/mL) and cefazolin (10 mg/0.1 mL). Two days after surgery, catheter patency was tested with an infusion of 0.02 mL of propofol (10 mg/mL) through the jugular catheter and at the end of experimentation. The surgery success rate was ∼75% successful in obtaining patent cannulated mice. Only mice that remained patent to the end of the progressive ratio task were included in the study.

2.3. Mechanical withdrawal thresholds

Sensory mechanical thresholds were performed as previously described.¹⁸ Sensory mechanical thresholds were performed as previously described.¹⁸ Briefly, mice were placed in individual plexiglass chambers on top of a mesh floor and allowed to acclimate for a minimum of 15 minutes. A 2-g von Frey filament (Stoelting, Wood Dale, IL) was applied to the middle of the plantar surface of the ipsilateral hind paw 10 times. Withdrawals, defined as lifting or moving the paw away from the filament or splaying the toes, were counted as positive responses.

2.4. Operant self-administration apparatus

Experiments were conducted in 12 operant conditioning chambers (Med Associates, St. Albans, VT), as previously described.^20,42 Briefly, each conditioning chamber was fitted with nose pokes, house and cue lights, and a multiaxis lever arm connected to a syringe pump. Experiments were conducted in 12 operant conditioning chambers (Med Associates, Fairfax, VT), as previously described.^20,42 Briefly, each conditioning chamber was housed inside a sound-attenuating box containing a ventilation fan for white noise. Each chamber contained a 5-unit curved nose-poke wall with cue lights (ENV-115C, Med Associates, Fairfax, VT). Back-mounted cannulas for drug delivery were connected to PE20 tubing threaded up through a multiaxis lever arm (PHM-124 MW, Med Associates) and were connected to a syringe pump (PHM-100, Med Associates, Fairfax, VT).

2.5. Operant procedure—remifentanil

One week after the CCI surgery, and after recovery from the jugular vein catheterization, all mice were trained to self-administer remifentanil HCl (NIDA drug supply synthesized by RTI log number 13933-138-1, reference number 030332, 50 μg/kg/infusion, prepared in sterile saline) under a fixed ratio 1 (FR1) schedule, where mice received 0.67 μL/g body weight/infusions. Remifentanil was chosen as the opioid of choice for this study because of its fast onset of action and short duration allowing animals to learn operant behavior quite quickly, avoiding the need for prior food training. In each chamber, 2 pokes were lit, 1 active and 1 inactive. Upon a nose poke into the active poke, the light was turned off and the drug became unavailable for 5 seconds (the drug was infused over ∼2 seconds, varying slightly by animal weight to achieve 50 μg/kg dose, followed by a ∼3-second time-out for a total time-out of 5 seconds). Additional nose pokes made during the time-out were recorded. Inactive nose pokes were recorded but resulted in no consequences. Active and inactive nose-poke assignments were counterbalanced on the left and right sides across different mice. During the first 2 sessions, both active and inactive pokes were baited with sweet and condensed milk every 20 minutes. Mice were removed from the boxes when they reached a maximum of 50 infusions or after 2 hours, whichever came first for both FR1 and FR3 protocols. This maximum number of 50 reinforcers was imposed as it represents 1 mL of IV fluid (0.67 µL/g body weight × 30 g mouse × 50 infusions) administration, which is approximately 25% of the blood volume for a mouse.

After the acquisition of nose-poke behavior, defined as >65% accuracy and <20% variability over 3 days (∼5-7 days from the start of training in the operant boxes), mice individually advanced to an FR3 schedule. After stabilization of FR3 responding (defined as >65% accuracy and <20% variability over 3 days), mice were placed on a progressive ratio (PR) schedule,¹¹ in which an increased number of responses were required after each subsequent infusion (ie, 1, 3, 9, 13, 16, 18, 20, 22, 24, 25, 27, 28, 29, 31, 32, 34, 35, 37, 39, 41, 44, 47, 52, etc.). Each progressive ratio session lasted 5 hours. Mice underwent only 1 progressive ratio test day and were then transitioned back to an FR1 schedule for 3 days, followed by 3 days on a reversal schedule, where the active and inactive pokes were switched.

2.6. Operant procedure—food

One week after the CCI surgery, and after recovery from the jugular vein catheterization, all mice were trained to self-administer ensure high protein shake (50% in water) under a FR1 schedule, where mice received 0.67 μL/g body weight/infusions. As with the protocol for remifentanil, the chamber, 2 pokes were lit, 1 active and 1 inactive. Upon a nose poke into the active poke, the light was turned off and the liquid ensure was delivered into the food cup. Additional nose pokes made during the time-out were recorded. Inactive nose pokes were recorded but resulted in no consequences. Active and inactive nose-poke assignments were counterbalanced on the left and right sides across different mice. During the first 2 sessions, both active and inactive pokes were baited with the ensure every 20 minutes. Mice were removed from the boxes when they reached a maximum of 50 infusions or after 2 hours, whichever came first for both FR1 and FR3 protocols. This maximum number of 50 reinforcers was imposed to be consistent with the remifentanil study. Mice underwent FR3, PR, and reversal task learning as with the remifentanil study described above.

2.7. In vivo fast-scan cyclic voltammetry

Two weeks after CCI or sham surgery, anesthetized mice were implanted with carbon fiber microelectrodes as we previously described.^8,33,46 Carbon fiber microelectrodes were constructed by vacuum aspiration of single 7-μm carbon fiber filaments (Goodfellow, Pittsburgh, PA) into 10-cm borosilicate glass capillary tubes (1 mm o.d. and 0.5 mm i.d.) (Catalog #626000, A-M Systems, Sequin, WA). A vertical microelectrode pulling system (Sutter Instruments, Novato, CA) was used to draw the glass to a fine point and to produce a high resistance seal around the carbon fiber, producing two 2-cm carbon-filled capillary tubes. After confirmation of the integrity and viability of the pulled product, the remaining carbon fiber was cut to ∼10 μm past the end of the glass under a microscope. Silver (Ag) wire (30 AWG) was painted with silver paint, backfilled with silver paint, and sealed with epoxy to make an electrical connection with the carbon fiber. Electrodes were calibrated in dopamine standard solutions (0, 0.5, 1, 2, and 5 µM) using a custom-built flow cell (University of California Los Angeles Psychology Mechanics Shop) fed by a syringe pump (Harvard Apparatus, Holliston, MA) regulating phosphate-buffered saline flow at 3 mL/min.

Surgery was performed on the day of fast-scan cyclic voltammetry recording under stable isoflurane anesthesia (5% induction and 2% maintenance, vaporized in oxygen, and delivered through a ventilator). Anesthetized mice were placed on a heated surgical blanket and held in a 3-axis stereotaxic frame. A midline incision was made to visualize bregma and a carbon fiber microelectrode was implanted. A microelectrode was lowered into the nucleus accumbens core (medial lateral [ML] ± 1.7 mm, anterior posterior [AP] +1.7 mm, dorsal ventral [DV] descent −4.5 mm from Bregma), and a bipolar stimulating electrode was implanted in the medial forebrain bundle (ML ± 1.1 mm, AP −2.4 mm, and DV descent −5 mm from Bregma). A Ag/AgCl reference electrode was placed in contact with the contralateral cortical surface of the brain to form a salt bridge.

Background-subtracted fast-scan cyclic voltammetry was performed in anesthetized mice using the High-Definition Cyclic Voltammetry (HDCV) recording and analysis software suite (University of North Carolina at Chapel Hill, Chapel Hill, NC). Reuptake kinetics were analyzed by Demon Voltammetry software (Dr. Jordan Yorgason, Brigham Young University, Provo, UT). A potentiostat (Department of Chemistry, University of Pittsburgh, Pittsburgh, PA), current amplifier (Keithley 428, Keithley Instruments Inc., Cleveland, OH), and HDCV were used to produce applied potentials held at −0.4 V vs Ag/AgCl. The microelectrode potential was then swept linearly to 1.3 V and back to −0.4 V using a triangular waveform at 400 V/second at a frequency of 10 Hz. Current recordings were taken every 10 seconds. Electrical stimulation was at 60 Hz via 2-millisecond pulses at 250 μA using a constant-current biphasic square waveform via the bipolar stimulating electrode (NeuroLog 512, Digitimeter Ltd., Welwyn Garden City, Hertfordshire, United Kingdom) and a stimulus isolator (NeuroLog 800A, Digitimeter Ltd.) with HDCV. Stimulation was applied to the medial forebrain bundle for 2 seconds after the start of recording. Recording and stimulation were performed over five (5) trials per subject.

A small electrical lesion (12 V AC, 3-5 s) was used to confirm the electrode placement, as carbon fiber tracks are otherwise not visible by light microscopy. Because the lesion procedure destroyed the electrode, dopamine concentrations were determined by the average calibration of 10 electrodes not used in surgery. After experimentation, mice were subjected to ice-cold intracardiac 4% paraformaldehyde for tissue fixation. The brains were harvested and sectioned by a cryostat at 40 μm. Correct microelectrode placement, seen by a tissue lesion from the dorsal limit of the brain down to the nucleus accumbens core, was confirmed for each animal.

2.8. In vivo microdialysis

Male and female mice were allowed to recover for a minimum of 3 to 5 days after CCI surgery. A second surgery was then performed on each mouse to implant a Cannulae for microdialysis probes (CMA)/7 guide cannula (#8010773, CMA Microdialysis, Kista, Sweden) for a microdialysis probe aimed at the ventral striatum (vSTR, ML ± 1.7 mm, AP +1.7 mm, DV descent −3.5 mm from Bregma). Mice were prepared for surgery by shaving and sterilizing the surgical area. They were anesthetized with isoflurane (5% induction and 2% maintenance). Body temperature was maintained on a thermal blanket. After implantation, each guide cannula was secured to the skull with C&B Metabond. After each surgery, mice were given daily carprofen injections (5 mg/kg, 1 mg/mL, subcutaneously) for the first 3 days and a combination of an antibiotic (amoxicillin, 0.25 mg/mL) and a second analgesic (ibuprofen, 0.25 mg/mL) in their drinking water for the first 14 days postoperatively. Animals recovered from the second surgery for at least 3 days before undergoing microdialysis.

Two weeks postsurgery, the day before microdialysis testing, each mouse was briefly anesthetized with isoflurane for insertion of a CMA/7 microdialysis probe into the guide cannula followed by continuous perfusion of artificial cerebral spinal fluid through the probe at 2 μL/min for 30 to 60 minutes followed by a 0.3-μL/min flow rate for an additional 12 to 14 hours to allow recovery from acute tissue damage because of probe insertion. The following day, the dialysate flow rate was increased to 2 μL/min for 60 to 120 minutes before collecting dialysate samples for online analysis. Serial high-K⁺ aCSF infusions of 1 to 5 min durations with 30-minute intertrial intervals were conducted.

Dialysate sample analysis was performed using an integrated high-performance liquid chromatography (HPLC) system (HTEC-500, formerly Eicom. Amuza Inc., San Diego, CA) with an Insight autosampler and 2 EAS-20s autoinjectors as previously described.^13,29,48 Electrochemical detection was performed using a WE-3G graphite working electrode with an applied potential of 450 mV vs a Ag/AgCl reference electrode. All dialysate samples for in vivo experiments were collected at 5-minute intervals at a dialysate flow rate of 2 μL/min. Samples were injected immediately into the HPLC system in an online analysis configuration.

2.8.1. High-performance liquid chromatography

Analysis of dialysate neurotransmitter levels was performed using an integrated HPLC system (HTEC-500, Amuza Inc., San Diego, CA) with an Insight autosampler and 2 EAS-20s online autoinjectors. Chromatographic separation was achieved using an Eicompak PP-ODS II stationary phase (4.6 mm i.d. × 30 mm, 2-μm particle diameter) and a phosphate-buffered mobile phase with 96 mM NaH₂PO₄ (Fluka #17844) and 3.8 mM Na₂HPO₄ (Fluka #71633), pH 5.4 at room temperature. The final mobile phase composition was 1.8 to 2.5% MeOH (EMD #MX0475–1), 50 mg/L EDTA•Na₂ (Fluka #03682), and 500 mg/L sodium decanesulfonate (TCI #I0348) in water purified via a Milli-Q Synthesis A10 system (EMD Millipore). The column temperature was maintained at 20 to 21°C. The volumetric flow rate was 350 to 500 μL/min.

Electrochemical detection was performed using an Eicom WE-3G graphite working electrode with an applied potential of 450 mV vs a Ag/AgCl reference electrode. Dopamine (Sigma-Aldrich #H8502) and serotonin (Sigma-Aldrich #H9523) standards were prepared in ice-cold regular artificial cerebrospinal fluid (aCSF):mobile phase (vol/vol 1:1).⁵⁰ Regular aCSF contained 147 mM NaCl (Fluka #73575), 3.5 mM KCl (Fluka #05257), 1.0 mM CaCl₂ (Sigma-Aldrich #499609), 1.0 mM NaH₂PO₄, 2.5 mM NaHCO₃ (Fluka #88208), and 1.2 mM MgCl₂ (Sigma-Aldrich #449172), pH 7.3 ±0.03 at room temperature. Standard curves, which were verified daily, encompassed physiological dopamine and serotonin concentration ranges (0, 7.8, 15.6, 31.2, 62.5, 125, 250, 500 pM, and 1.0, 1.25, 2.5, 3.1, 6.3 nM using 20 μL sample volumes). The limit of detection was <150 fmol (<7.8 pm), and the practical limit of quantification was <300 fmol (<15.6 pm) using a 20 μL sample volume.⁴⁹ All dialysate samples for in vivo experiments were collected at 5-minute intervals at a dialysate flow rate of 2 μL/min using EAS-20s autoinjectors. Samples were injected immediately into the HPLC system in an online analysis configuration.³⁸

2.8.2. Probe insertion

During 16:00 to 18:00, the day before the testing day, each subject was briefly (1-3 minutes) anesthetized using isoflurane for insertion of a CMA/7 microdialysis probe (#8010771, CMA Microdialysis, Kista, Sweden) into the guide cannula. Immediately after insertion, regular aCSF was continuously perfused through the probe at 2 μL/min for 30 to 60 minutes followed by a 0.3-μL/min flow rate for an additional 12 to 14 hours to allow brain tissue recovery from acute neurotransmitter release because of probe insertion.

2.8.3. High-K+–induced dopamine and serotonin overflow

For all mice, at 06:00 to 07:00 hours the following day, the regular aCSF flow rate through the probes was increased back to 2 μL/min for 60 to 120 minutes before collecting dialysate samples for analysis. Serial high-K⁺ aCSF infusions of 1-, 2-, 3-, 4-, and 5-minute durations with 30-minute interstimulus intervals were conducted. The high-K⁺ aCSF contained 31 mM NaCl, 120 mM KCl, 1.0 mM CaCl₂, 1.0 mM NaH₂PO₄, 2.5 mM NaHCO₃, 1.2 mM MgCl₂, at pH 7.3 ± 0.03 at room temperature.

2.9. Experimental design and statistical analysis

Experiments were designed to ask (1) does chronic pain impact opioid acquisition and modify work effort, (2) what is the impact of chronic pain on mu opioid receptor function, and (3) what is the impact of chronic pain on extracellular dopamine content in the nucleus accumbens. All data sets were tested for normality and post hoc tests were selected based on the results. Self-administration time course data were analyzed by GraphPad Prism v10.1.1 using mixed 2- or 3-way ANOVAs with repeated measures, with surgical group and nose poke as the between-subjects factor and session number as the within-subject factor. All time course data were performed by repeated measures ANOVAs using a Geisser–Greenhouse correction. A survival cure analysis to determine differences between sham and chronic pain groups on the number of animals that achieved maximum number of reinforcers each day. This analysis consisted of a Kaplan–Meier survival analysis. The IVSA intrasession cumulative data were analyzed by repeated measures 2-way ANOVA, as well as by 2-way ANOVA of the area under the curve data. Progressive ratio data were analyzed by 2-tailed unpaired student t-tests. Voltammetry data were analyzed by repeated measures 2-way ANOVA and 2-tailed unpaired student t-tests. Basal dopamine and serotonin dialysate concentration data are expressed as means ± SEMs with scatter plots for the individual data overlaid. Statistical analysis was conducted using a Welch t test. High-K⁺ infusion-induced dopamine and serotonin overflows are as time courses with the areas under the curve used for overflow peak analysis using a 2-way ANOVA.

3. Results

3.1. Chronic pain impacts the acquisition of remifentanil operant behavior

To assess the impact of chronic pain on opioid self-administration, we designed a schedule that allowed us to evaluate numerous domains of the operant behavior, including acquisition of self-administration, performance on low vs high demand operant schedules, motivation to work for reward, and reversal learning (Fig. 1A). We determined mechanical nociceptive responses once a week using a 10-poke Von Frey method and found that CCI male and female mice had significantly higher numbers of mechanical responses 7 days after surgery (post hoc, P = 0.0002 vs sham), which lasted the duration of the study (Fig. 1B). We found a main effect of time (P < 0.0001), and surgery (P < 0.0001), with a significant interaction of pain and time (P < 0.001), with the pain state peaking during weeks 3 and 4 after surgery. The statistics for von Frey data are presented in Table 1.

Figure 1.

Chronic neuropathic pain increased the acquisition of remifentanil intravenous self-administration (IVSA). Male and female mice underwent chronic constriction injury (CCI) to induce neuropathic pain or sham surgery 7 days before operant testing with IVSA of remifentanil. (A) A schematic representation of the behavioral paradigm. (B) The numbers of responses to mechanical von Frey filament (2 g force) application to the ipsilateral hind paw were increased in both male (N = 10) and female (N = 7) mice compared to sham mice (N = 6 female and N = 9 male). (C) All mice acquired remifentanil IVSA, which increased with time. Mice with chronic pain did not show an increase in the total number of reinforcers earned compared to sham animals. However, examining the time to reach maximum number of reinforcers on each day of training or the probability of mice that reached the maximum number of reinforcers on each day did show an effect of chronic pain. (D) Total numbers of active (AP) and inactive (IP) nose pokes are plotted over the acquisition phase of remifentanil IVSA. The total number of active nose pokes was higher for CCI mice compared to sham mice when the sexes were combined. When these data were divided by sex, both male and female CCI mice showed greater numbers of active nose pokes compared to controls. Statistics are presented in Tables 1 and 2. Data are expressed as means ± SEMs for N = 6 to 16 per group. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Table 1.

Statistical analysis using a repeated measures 2-way analysis of variance of mechanical sensitivity and self-administration presented in Figure 1.

			F (DFn, DFd)	P
Figure 1B von frey (VF)	Combined sexes	Time	F (4, 148) = 22.98	P < 0.0001****
		Surgery condition	F (1, 148) = 71.35	P < 0.0001****
		Time × surgery	F (4, 148) = 8.045	P < 0.0001****
	Males	Time	F (4, 84) = 17.79	P < 0.0001****
		Surgery condition	F (1, 84) = 35.75	P < 0.0001****
		Time × surgery	F (4, 84) = 4.769	P = 0.0016**
	Females	Time	F (4, 55) = 7.918	P < 0.0001****
		Surgery condition	F (1, 55) = 65.62	P < 0.0001****
		Time × surgery	F (4, 55) = 6.916	P = 0.0001****
Figure 1C Both sexes	Reinforcers	Time	F (4, 116) = 20.60	P < 0.0001****
		Pain	F (1, 29) = 4.068	P = 0.0531
		Time × pain	F (4, 116) = 0.6599	P = 0.6211
Males	Reinforcers	Time	F (4, 64) = 7.772	P < 0.0001****
		Pain	F (1, 16) = 2.794	P = 0.1140
		Time × pain	F (4, 64) = 1.338	P = 0.2655
Females	Reinforcers	Time	F (4, 44) = 19.90	P < 0.0001****
		Pain	F (1, 11) = 1.174	P = 0.3018
		Time × pain	F (4, 44) = 0.4512	P = 0.7709
Figure 1C	Time to achieve max reinforcers	Time	F (3.261, 94.57) = 6.866	P = 0.0002***
		Pain	F (1, 30) = 4.444	P = 0.0435*
		Time × pain	F (4, 116) = 0.6698	P = 0.6142
Figure 1c	Survival curve		X2 = 9.015	P = 0.0027**
Figure 1D Both sexes	Nose pokes	Time	F (4, 145) = 2.564	P = 0.0408*
		Nose poke	F (1, 145) = 56.11	P < 0.0001****
		Pain	F (1, 145) = 17.09	P < 0.0001****
		Time × nose poke	F (4, 145) = 3.005	P = 0.0204*
		Time × pain	F (4, 145) = 0.8918	P = 0.4706
		Nose poke × pain	F (1, 145) = 1.184	P = 0.2784
		Time × nose poke × pain	F (4, 145) = 0.2866	P = 0.8863
Figure 1D Males	Nose pokes	Time	F (4, 145) = 2.420	P = 0.0511
		Nose poke	F (1, 80) = 50.88	P < 0.0001****
		Pain	F (1, 145) = 11.79	P = 0.0008***
		Time × nose poke	F (4, 80) = 1.410	P = 0.2383
		Time × pain	F (4, 145) = 1.641	P = 0.1670
		Nose poke × pain	F (1, 80) = 0.2959	P = 0.5880
		Time × nose poke × pain	F (4, 80) = 1.356	P = 0.2566
Figure 1D Females	Nose pokes	Time	F (4, 145) = 2.566	P = 0.0407*
		Nose poke	F (1, 54) = 22.80	P = 0.0002***
		Pain	F (1, 145) = 12.77	P = 0.0005***
		Time × nose poke	F (4, 54) = 3.607	P = 0.0112*
		Time × pain	F (4, 145) = 0.4045	P = 0.8052
		Nose poke × pain	F (1, 54) = 0.7256	P = 0.3981
		Time × nose poke × pain	F (4, 54) = 1.166	P = 0.3361

P values are provided for each of the factors (time, surgical condition, and if there was an interaction between time and surgical condition). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We began operant training with remifentanil on an FR1 schedule 7 days after CCI or sham surgery. Based on a dose–response pilot study (data not shown), we found that mice were able to learn nose-poke behavior at a 50 μg/kg/infusion dose without the need for prior operant training enabling the assessment of the effects of chronic pain on the acquisition of remifentanil responding. We found no difference in the number of reinforcers earned between control and CCI mice (Fig. 1C), or when segregated by sex. Statistical analysis revealed an effect of time (P < 0.0001) but not pain (P = 0.0531) when data from both sexes were combined. However, we did see a significant difference between groups for the time to reach maximum number of reinforcers within each operant day (Fig. 1C) and for a survival curve analysis, where we compared the number of animals that reached the maximum number of reinforcers on each day of acquisition training (Fig. 1C). Statistical analysis (Table 1) of time to reach maximum reinforcers showed an effect of time (P = 0.0002) and pain (P = 0.0435), but there was no interaction (P = 0.6142). Statistical analysis of the survival curve revealed an effect between sham and pain groups (P = 0.0027).

Although there was no significant increase in the number of reinforcers earned, the drug-seeking metric of active nose pokes was significantly higher in the CCI mice compared to sham animals (Fig. 1D, Table 1). The CCI mice also more quickly acquired the nose-poke behavior for remifentanil compared to sham mice (Fig. 1D). Both male and female CCI mice exhibited greater active nose pokes than sham animals, which reflects drug seeking during the time out as there was no difference in number of reinforcers earned. Statistical analysis using a 3-way ANOVA revealed a main effect of time (P = 0.0079), pain condition (P < 0.0001), and nose poke (P < 0.0001). No interactions were identified for the data with the sexes combined. Similar results were seen for both male and female mice when data for each sex were analyzed separately. Here, a significant interaction was revealed between time and nose pokes for female mice (Table 1). Although active nose pokes were greater in the CCI mice, their accuracy or discrimination index for active compared to total numbers of nose pokes was not different between pain and sham groups (data not shown).

We next examined how mice acquired remifentanil responding for each day of the schedule within each 2 hours session on individual days by assessing the cumulative number of reinforcers earned (Fig. 2A). Figure 2B shows that CCI mice have faster acquisition for remifentanil compared to sham mice on days 2 and 3 of acquisition. Statistical analysis revealed an effect of time, pain condition, and their interaction (Table 2). The area under the curve for each animal confirmed that the differences in behavior were driven by male but not female mice (Fig. 2B, Table 2). The magnitude of the response above sham levels is presented in Figure 2C demonstrating that males but not females show higher rates of acquisition on the earlier days of self-administration.

Figure 2.

Chronic neuropathic pain increased the acquisition of remifentanil intravenous self-administration (IVSA) within session. Male and female mice underwent chronic constriction injury (CCI) to induce neuropathic pain or sham surgery 7 days before operant testing with IVSA of remifentanil. (A) A schematic representation of the behavioral paradigm. (B) Intrasession data (2-hour operant self-administration session data for each day of acquisition) for the data presented in Figure 1. Data are presented with the sexes combined (xy plots) and as areas under the curve (AUC) for male and female mice for day 2 days of remifentanil acquisition. Chronic pain animals acquired remifentanil faster than sham animals. The AUC data demonstrate that this effect shows that this effect was driven primarily by differences in male mice where there was a significant increase in total reinforcers on days 2 and 3 of acquisition. (C) We examined the cumulative numbers of reinforcers normalized to sham animals on each day. Only male CCI mice showed increased cumulative numbers of reinforces earned on days 2 and 3 of the acquisition phase. Data represent means ± SEMs for N = 6 to 16 mice per group. *P < 0.05, **P < 0.01. Statistics for data sets are presented in Table 2.

Table 2.

Statistical analysis using a repeated measures 2- or 3-way analysis of variance of remifentanil during acquisition of opioid intravenous self-administration, where data are presented in Figure 2.

Figure 2			F (DFn, DFd)	P
Remifentanil IVSA
Day 1	Cumulative reinforcers	Time	F (1.385, 40.17) = 40.52	P < 0.0001****
		Pain	F (1, 29) = 3.117	P = 0.0880
		Time × pain	F (120, 3480) = 1.838	P < 0.0001****
	AUC	Pain	F (1, 27) = 2.902	P = 0.1000
		Sex	F (1, 27) = 1.711	P = 0.2019
		Interaction	F (1, 27) = 0.1663	P = 0.6866
Day 2	Cumulative reinforcers	Time	F (120, 3480) = 69.37	P < 0.0001****
		Pain	F (1, 29) = 4.377	P = 0.0453*
		Time × pain	F (120, 3480) = 2.447	P < 0.0001****
	AUC	Pain	F (1, 27) = 5.155	P = 0.0314*
		Sex	F (1, 27) = 3.51	P = 0.0718
		Interaction	F (1, 27) = 4.41	P = 0.0451*
Day 3	Cumulative reinforcers	Time	F (2.151, 62.39) = 107.1	P < 0.0001****
		Pain	F (1, 29) = 2.865	P = 0.1013
		Time × pain	F (120, 3480) = 1.383	P = 0.0042**
	AUC	Pain	F (1, 27) = 3.498	P = 0.0723
		Sex	F (1, 27) = 1.006	P = 0.3249
		Interaction	F (1, 27) = 1.417	P = 0.2443
Day 4	Cumulative reinforcers	Time	F (2.340, 67.86) = 81.03	P < 0.0001****
		Pain	F (1, 29) = 1.252	P = 0.2723
		Time × pain	F (120, 3480) = 1.065	P = 0.3005
	AUC	Pain	F (1, 27) = 0.2393	P = 0.6287
		Sex	F (1, 27) = 0.9762	P = 0.3319
		Interaction	F (1, 27) = 0.2393	P = 0.6287
Day 5	Cumulative reinforcers	Time	F (2.185, 63.36) = 119.9	P < 0.0001****
		Pain	F (1, 29) = 2.476	P = 0.1264
		Time × pain	F (120, 3480) = 0.7456	P = 0.9819
	AUC	Pain	F (1, 27) = 0.2233	P = 0.6403
		Sex	F (1, 27) = 0.004630	P = 0.9462
		Interaction	F (1, 27) = 0.2233	P = 0.6403
Figure 1C	Cumulative reinforcers by day and sex	Male day 1	t = 2.000, df = 8	P = 0.0805
		Male day 2	t = 3.015, df = 8	P = 0.0167*
		Male day 3	t = 2.639, df = 8	P = 0.0297*
		Male day 4	t = 0.3926, df = 8	P = 0.7049
		Male day 5	t = 1.168, df = 8	P = 0.2763
		Female day 1	t = 1.155, df = 6	P = 0.2919
		Female day 2	t = 0.4681, df = 6	P = 0.6563
		Female day 3	t = 0.6504, df = 6	P = 0.5395
		Female day 4	t = 1.223, df = 6	P = 0.2673
		Female day 5	t = 1.641, df = 6	P = 0.1519

A 1-sample t test and Wilcoxon test were used for the cumulative reinforcers being different than theoretical value of zero in panel c. P values are provided for each of the factors (time, pain, and appropriate interactions) for active and inactive nose pokes, and number of reinforcers earned. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

AUC, area under the curve; IVSA, intravenous self-administration.

3.2. Chronic pain significantly reduces the motivation to work for rewards on higher-demand operant schedules

After the acquisition of a stable number of reinforcers earned on an FR1 schedule, mice advanced to a FR3 schedule that requires more effort to obtain a reward (Fig. 3A). Contrary to our initial hypothesis, the CCI mice significantly decreased their overall operant responding, earning significantly fewer reinforcers compared to sham mice under the FR3 schedule (Fig. 3B; main effect of pain [P < 0.0001] but no effect of time [P = 0.3598]). Similar to the reduced number of reinforcers earned, the CCI mice performed fewer active pokes than sham animals (main effect of pain [P = 0.0226], time [P = 0.0120], nose poke [P < 0.0001], and a nose pose by pain interaction [P = 0.0018]) (Fig. 3C), although the pain animals showed lower accuracy compared to sham animals yet remained >50% accurate, ie, they still responded better than chance (Fig. 3D). Statistical analysis showed a main effect of pain (P = 0.0021), but not of time (P = 0.8885). The CCI mice did not increase their responses on the inactive poke, such that the decrease in accuracy was because of decreases in active poking, and not an increase in inactive poking.

Figure 3.

Chronic neuropathic pain decreased the motivation for remifentanil intravenous self-administration (IVSA) at higher effort levels. After the acquisition of IVSA of remifentanil, chronic constriction injury (CCI) and sham mice were transitioned to a fixed ratio 3 (FR3) schedule where mice were required to actively poke 3 times before a reward was delivered. (A) A schematic representation of the behavioral paradigm. (B) All mice continued to self-administer remifentanil. Mice with chronic pain showed a decrease in the number of reinforcers earned at most time points compared to sham animals. (C) Total numbers of active (AP) and inactive (IP) nose pokes are plotted for remifentanil IVSA. The total number of active nose pokes was lower for CCI compared to sham mice, without loss of accuracy as evidenced by a similar discrimination index (D). (E) Both male (E) and female (F) mice with chronic pain showed a decrease in the number of reinforcers earned at most time points compared to sham animals. Statistical analysis of area under the curve (AUCs) for the 4 days shows a significant difference for male (t = 2.45, P = 0.0402*) and female (t = 2.261, P = 0.0450*) pain compared to sham mice. The total number of active (AP) and inactive (IP) nose pokes are plotted for remifentanil IVSA. The total number of active nose pokes was lower for CCI compared to sham mice, which was not dependent on sex. For male data (E), statistical analysis did not identify an effect of time (F (2.336, 98.10) = 2.829, P = 0.0556) but did reveal an effect of pain condition (F (3, 42) = 3.907, P = 0.0151*). For female data (F), statistical analysis did identify an effect of time (F (3, 80) = 4.113, P = 0.0091**) and an effect of pain condition (F (3, 80) = 5.009, P = 0.0031**). (G) The intrasession data (2-hour operant self-administration session) for each day under the FR3 schedule are presented. Chronic pain animals were slower in acquiring remifentanil self-administration compared to sham animals within sessions. The AUC revealed differences on days 2, 3, and 5. (H) The cumulative numbers of reinforcers normalized to sham animals are presented on each day of the FR3 schedule. Both male and female CCI mice showed decreased cumulative numbers of reinforces earned on most days. Statistics are presented in Table 3. Data are expressed as means ± SEMs for N = 15 to 16 per group. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Both male and female CCI mice demonstrated this reduced number of reinforcers earned (Fig. 3E, Table 3) as well as reduced active nose pokes (Fig. 3F, Table 3). Within each session, the CCI mice earned their rewards more slowly than sham mice (Fig. 3G, Table 3). We also examined the numbers of cumulative reinforcers for each day of the schedule separated by sex relative to the mean number of reinforcers earned for sham animals on each given day. Both sexes of CCI mice exhibited a reduction in the number of reinforcers earned (Fig. 3H, Table 3).

Table 3.

Statistical analysis using repeated measures 2-way analysis of variance or 3-way analysis of variance of remifentanil during acquisition of opioid intravenous self-administration, where data are presented in Figure 3.

Figure 2			F (DFn, DFd)	P
Remifentanil IVSA
Figure 3B	Reinforcers	Time	F (3, 81) = 2.581	P = 0.0592
		Pain	F (1, 27) = 6.568	P = 0.0163*
		Time × pain	F (3, 81) = 0.6613	P = 0.5782
Figure 3C	Nose pokes	Time	F (3, 216) = 3.738	P = 0.0120*
		Pain	F (1, 216) = 5.271	P = 0.0226*
		Nose poke	F (1, 216) = 30.16	P < 0.0001****
		Time × poke	F (3, 216) = 0.4089	P = 0.7468
		Time × pain	F (3, 216) = 0.1054	P = 0.9569
		Poke × pain	F (1, 216) = 10.03	P = 0.0018**
		Time × poke × pain	F (3, 216) = 0.1837	P = 0.9074
Figure 3D	Discrimination index	Time	F (3, 78) = 0.2823	P = 0.8381
		Pain	F (1, 26) = 5.687	P = 0.0247*
		Time × pain	F (3, 78) = 0.1187	P = 0.9489
Figure 3G Day 1	Cumulative reinforcers	Time × pain	F (120, 3240) = 2.796	P < 0.0001****
		Time	F (1.819, 49.10) = 51.26	P < 0.0001
		Pain	F (1, 27) = 3.071	P = 0.0910
	AUC		t = 1.753, df = 27	P = 0.0910
Figure 3G Day 2	Cumulative reinforcers	Time × pain	F (120, 3120) = 3.883	P < 0.0001****
		Time	F (1.244, 32.35) = 51.42	P < 0.0001****
		Pain	F (1, 26) = 5.814	P = 0.0233*
	AUC		t = 2.411, df = 26	P = 0.0233*
Figure 3G Day 3	Cumulative reinforcers	Time × pain	F (120, 3120) = 6.429	P < 0.0001****
		Time	F (1.413, 36.74) = 46.72	P < 0.0001****
		Pain	F (1, 26) = 10.49	P = 0.0033**
	AUC		t = 3.239, df = 26	P = 0.0033**
Figure 3G Day 4	Cumulative reinforcers	Time × pain	F (120, 3120) = 2.247	P < 0.0001****
		Time	F (1.250, 32.50) = 50.43	P < 0.0001****
		Pain	F (1, 26) = 3.472	P = 0.0737
	AUC		t = 1.863, df = 26	P = 0.0737
Figure 3G Day 5	Cumulative reinforcers	Time × pain	F (120, 2880) = 6.219	P < 0.0001****
		Time	F (1.444, 34.67) = 60.29	P < 0.0001****
		Pain	F (1, 24) = 8.333	P = 0.0081**
	AUC		t = 2.934, df = 23	P = 0.0075**
Figure 3H	Cumulative reinforcers by day and sex	Male day 1	t = 9.343, df = 8	P < 0.0001****
		Male day 2	t = 5.239, df = 8	P = 0.0008***
		Male day 3	t = 5.360, df = 8	P = 0.0007***
		Male day 4	t = 3.827, df = 8	P = 0.0050**
		Male day 5	t = 8.418, df = 7	P < 0.0001****
		Female day 1	t = 0.3854, df = 5	P = 0.7158
		Female day 2	t = 2.676, df = 5	P = 0.0440*
		Female day 3	t = 12.27, df = 5	P < 0.0001****
		Female day 4	t = 1.355, df = 5	P = 0.2336
		Female day 5	t = 3.625, df = 5	P = 0.0151*

A 1-sample t test and Wilcoxon test were used for the cumulative reinforcers being different than theoretical value of zero. P values are provided for each of the factors (time, pain, and if there was an interaction) for active and inactive nose pokes, and number of reinforcers earned. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

AUC, area under the curve; IVSA, intravenous self-administration.

To further assess the motivation to work for remifentanil, mice were placed on a 5-hour PR schedule (Fig. 4A) where each subsequent reward required an increase in the number of nose pokes. We found that the CCI mice were less willing to work for remifentanil compared to the sham mice, as evidenced by earning fewer reinforcers (Fig. 4B). The CCI mice had a significantly lower breakpoint (Fig. 4C, Table 4). They also performed fewer active nose pokes (Fig. 4D) with no difference in inactive nose pokes compared to sham mice (Fig. 4D). Female CCI mice primarily drove this reduced work effort, although male mice showed lower but not statistically significant decreases in work for remifentanil (Figs. 4B–D). Accuracy was maintained, as there was no significant difference in the number of pokes or the inactive pokes (Fig. 4E).

Figure 4.

Reduced motivation for remifentanil in chronic pain states is not driven by changes in reward value or cognitive impairment. After the fixed ratio 3 schedule for intravenous self-administration (IVSA) of remifentanil, mice were tested in a progressive ratio task. (A) A schematic representation of the behavioral paradigm. (B) In the 5-hour progressive ratio task, chronic constriction injury (CCI) mice earned fewer reinforcers than sham animals. This was evident in female mice but did not quite reach significance in male CCI mice. (C) The breakpoint for CCI mice was significantly lower in chronic pain compared to sham control mice. This was associated primarily with female rather than male pain animals. (D) In the 5-hour progressive ratio task, chronic pain mice made fewer active nose pokes than sham animals, which was evident in female mice but not in male CCI mice. (E) The increase in work effort did not affect the number of inactive nose pokes. The statistical analysis presented was performed by a student unpaired t test. Data are expressed as means ± SEMs for N = 15 to 16 per group, Table 4. *P < 0.05, **P < 0.01.

Table 4.

Statistical analysis using repeated measures 2-way analysis of variance of remifentanil during acquisition of opioid intravenous self-administration, where data are presented in Figure 4.

Figure 4			F (DFn, DFd)	P
Remifentanil IVSA
Figure 4B Progressive ratio	Reinforcers	Both sexes	t = 3.646, df = 26	P = 0.0012**
		Males	t = 1.905, df = 13	P = 0.0791
		Females	t = 3.850, df = 10	P = 0.0032**
Figure 4C	Break point	Both sexes	t = 3.321, df = 27	P = 0.0027**
		Males	t = 1.864, df = 14	P = 0.0834
		Females	t = 3.769, df = 11	P = 0.0031**
Figure 4D	Nose pokes (active)	Both sexes	t = 2.797, df = 26	P = 0.0096**
		Males	t = 1.743, df = 14	P = 0.1032
		Females	t = 2.747, df = 10	P = 0.0206*
Figure 4E	Nose pokes (inactive)	Both sexes	t = 1.385, df = 26	P = 0.1779
		Males	t = 1.028, df = 14	P = 0.3214
		Females	t = 0.8402, df = 10	P = 0.4204

An unpaired t test was used for area under the curve (AUC) data. P values are provided for each of the factors (time, pain, and appropriate interactions) for active and inactive nose pokes, and number of reinforcers earned. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

IVSA, intravenous self-administration.

After the PR schedule, mice were placed back on the FR1 schedule the day after PR testing, and the performance of the CCI mice immediately rebounded, with no significant differences in the number of rewards earned compared to sham mice (Fig. 5, Table 5), indicating the CCI mice still found remifentanil rewarding. We then exposed mice to a reversal-learning task, where the positions of active and inactive pokes were switched. The sham and CCI mice were able to unlearn and relearn at similar rates, indicating that no operant learning deficits were present in the CCI mice (Figs. 5B–D, Table 5). Intrasession data from days 1 and 3 of the return to FR1 and reversal FR1 also revealed no differences between CCI and sham animals (Fig. 5E, Table 5). Taken together, these findings indicate that the decrease in operant performance by CCI mice at FR3 was likely because of a loss in motivation to work and not a loss of the hedonic value of the drug or an inability to learn the operant demands.

Figure 5.

All animals show return to similar drug taking after returning to fixed ratio 1 (FR1) and in a FR1 reversal. The day after the progressive ratio task, mice were placed back on an FR1 schedule. (A) A schematic representation of the behavioral paradigm. Once on this FR1 schedule, there were no differences in responding for number of reinforcers earned (B), nose pokes (C), or accuracy between active and inactive nose pokes (D) between pain and sham animals. After 3 days, the active and inactive ports were reversed to test for cognitive flexibility. All mice earned similar numbers of reinforcers. (C) Active (AP) and inactive (IP) nose pokes for intravenous self-administration (IVSA) remifentanil were similar between pain and sham animals when transferred back to the FR1 schedule. On the first day of reversal, all mice showed increased inactive nose pokes that decreased at similar rates in sham and pain animals. The intrasession (2 hours operant self-administration session for each day of acquisition) data for the first and last day of (E) FR1 and FR1 reversal are presented. There was no effect on the rate of drug acquisition within sessions for any of the time points. Statistics for these data are presented in Table 5. Data are expressed as means ± SEMs for N = 6 to 11 mice per group.

Table 5.

Statistical analysis using repeated measures 2-way analysis of variance of remifentanil during acquisition of opioid intravenous self-administration, where data are presented in Figure 5.

Figure 5			F (DFn, DFd)	P
Remifentanil IVSA
Figure 5B	Reinforcers	Time	F (2.602, 54.12) = 4.781	P = 0.0071**
		Pain	F (1, 22) = 1.278	P = 0.2705
		Time × pain	F (5, 104) = 0.5308	P = 0.7525
Figure 5C	Nose pokes	Time	F (5.000, 130.0) = 4.848	P = 0.0004***
		Nose poke	F (0.4568, 11.88) = 12.55	P = 0.0120*
		Pain	F (1, 26) = 0.01537	P = 0.9023
		Time × nose poke	F (3.056, 42.17) = 31.15	P < 0.0001****
		Time × pain	F (5, 130) = 0.6918	P = 0.6305
		Nose poke × pain	F (1, 26) = 0.2111	P = 0.6198
		Time × nose poke × pain	F (5, 69) = 0.6418	P = 0.6686
Figure 5C	% accuracy	Time	F (3.402, 68.05) = 38.54	P < 0.0001****
		Pain	F (1, 22) = 0.1057	0.7482
		Time × pain	F (5, 100) = 0.1788	0.9700
Figure 5E (before reversal)	Day 1	Time × pain	F (120, 2040) = 2.275	P < 0.0001****
		Time	F (1.595, 27.12) = 66.34	P < 0.0001****
		Pain	F (1, 17) = 2.187	P = 0.1575
	AUC		t = 1.479, df = 17	P = 0.1575
	Day 3	Time × pain	F (120, 2040) = 0.06291	P > 0.9999
		Time	F (2.161, 36.74) = 116.4	P < 0.0001****
		Pain	F (1, 17) = 0.1357	P = 0.7171
	AUC		t = 0.3684, df = 17	P = 0.7171
Figure 5E (after reversal)	Day 1	Time × pain	F (120, 2040) = 1.371	P = 0.0058**
		Time	F (1.515, 25.76) = 38.84	P < 0.0001****
		Pain	F (1, 17) = 0.4521	P = 0.5104
	AUC		t = 0.6724, df = 17	P = 0.5104
	Day 3	Time × pain	F (120, 2040) = 0.5605	P > 0.9999
		Time	F (1.400, 23.80) = 72.10	P < 0.0001****
		Pain	F (1, 17) = 0.6479	P = 0.4320
	AUC		t = 0.8049, df = 17	P = 0.4320

An unpaired t test was used for area under the curve (AUC) data. P values are provided for each of the factors (time, pain, and appropriate interactions) for active and inactive nose pokes, and number of reinforcers earned. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

AUC, area under the curve; IVSA, intravenous self-administration.

3.3. Chronic pain does not impact the motivation to work for food reward

We conducted the full operant schedule (Fig. 6A) in a new cohort of mice using a food reward (50% Ensure diluted with water) to assess if the changes for remifentanil self-administration generalized to natural rewards. On average, CCI and sham mice similarly acquired FR1 responding for Ensure by day 7 (Fig. 6B, main effect of time, F_{(6, 151)} = 9.849, P < 0.0001), yet CCI mice earned slightly fewer rewards (main effect of pain, F_{(1, 151)} = 5.814; P = 0.0171) where there was also a time by pain interaction (P < 0.0001, Table 6). When the sexes were analyzed separately, the effect of pain on responding for food reward was evident in male (P = 0.0063) but not female (P = 0.4617) mice (Table 6). There was no clear effect of chronic pain on the responding accuracy or number of nose pokes (Table 6). All mice learned and performed above 70% accuracy by day 7 (Fig. 6C, main effect of nose poke, F _(1,366) = 195.3; P < 0.0001). Within a session, the CCI mice earned reinforcers more slowly as they learned the food reward task as evidenced by no difference in cumulate reinforcers earned on day 2, but there was a difference on days 3 and 5 when data from both sexes were combined (Fig. 6D). However, neither sex reached significance (Fig. 6D) compared to control mice; there was a time × pain interaction for both sexes.

Figure 6.

Chronic neuropathic pain reduces the acquisition of food reward but does not affect work effort and motivation. Male and female mice underwent chronic constriction injury (CCI) to induce neuropathic pain or sham surgery 7 days before operant testing with oral Ensure (50% in water). (A) A schematic representation of the behavioral paradigm. (B) All mice acquired food reward that increased with time. Mice with chronic pain showed a reduced number of reinforcers earned compared to sham animals. This effect was associated with male mice, whereas there was no effect of pain on the rate of acquisition in chronic pain female mice. (C) Both sexes had similar numbers of active (AP) and inactive (IP) nose pokes for the first few days until learning to discriminate the ports. There was no effect of pain. (D) The intrasession (2-hour operant self-administration session for days 2, 3, and 5 of acquisition) data are shown. Data are presented as areas under the curve (AUC) with the sexes combined where chronic pain animals acquired fewer cumulative numbers of reinforcers compared to sham animals. When the sexes were analyzed separately, neither female nor male CCI mice showed slower acquisition of food reward on day 5 of the fixed ratio 1 (FR1) schedule. However, there was a pain × time interaction. These data demonstrate that the effect was associated with male rather than female mice. (E) After the acquisition of 50% Ensure responding, the CCI and sham mice were transitioned to a FR3 schedule. All mice continued to self-administer food reward. There were no differences between chronic pain and sham animals in the numbers of reinforcers earned or in the numbers of active nose pokes. (F) After 5 days on the FR3 schedule, mice were transitioned to a 5-hour progressive ratio task. Here, there was also no effect of pain on work effort to receive the food reward. (G) The day after the progressive ratio task, mice were placed back on an FR1 schedule. Once on this schedule, there were no behavioral differences between pain and sham animals. After 3 days, the active and inactive ports were reversed to test for cognitive flexibility. All mice showed similar numbers of reinforcers earned. Active (AP) and inactive (IP) nose pokes for 50% Ensure were also similar between pain and sham animals when transferred to FR1. On the first day of reversal, all mice showed increased inactive nose pokes that decreased at similar rates in sham and pain animals. The number of magazine entries was also not different between pain and sham mice. Statistics are presented in Table 6. Data are expressed as means ± SEMs for N = 6 to 12 per group.

Table 6.

Statistical analysis using a repeated measures 2-way analysis of variance or 3-way analysis of variance of ensure food acquisition of data presented in Figure 6.

Figure 6			F (DFn, DFd)	P
Ensure food
Figure 6B FR1	Reinforcers Both sexes	Time	F (6, 129) = 23.25	P < 0.0001****
		Pain	F (1, 22) = 1.439	P = 0.2431
		Interaction	F (6, 129) = 0.2529	P = 0.9573
	Reinforcers Males	Time	F (3.176, 31.76) = 11.72	P < 0.0001****
		Pain	F (1, 10) = 2.287	P = 0.1614
		Interaction	F (6, 60) = 0.2059	P = 0.9736
	Reinforcers Females	Time	F (2.741, 26.04) = 11.21	P < 0.0001****
		Pain	F (1, 10) = 0.1585	P = 0.6989
		Interaction	F (6, 57) = 1.219	P = 0.3099
Figure 6C FR1	Nose pokes	Time	F (9, 99) = 28.08	P < 0.0001****
		Nose poke	F (1, 11) = 91.29	P < 0.0001****
		Pain	F (1, 11) = 0.2100	P = 0.6557
		Time × nose poke	F (9, 99) = 7.760	P < 0.0001****
		Time × pain	F (9, 99) = 0.8956	P = 0.5323
		Nose poke × pain	F (1, 11) = 2.026	P = 0.1824
		Time × nose poke × pain	F (9, 25) = 0.7876	P = 0.6301
Figure 6D Intersession cumulative reinforcers	AUC	Day 3	t = 2.934, df = 18	P = 0.0089***
		Day 5	t = 2.970, df = 19	P = 0.0079**
	Day 5 Males	Time × pain	F (120, 2520) = 2.111	P < 0.0001****
		Time	F (2.317, 48.65) = 56.78	P < 0.0001****
		Pain	F (1, 21) = 4.227	P = 0.0524
	Day 5 Females	Time × pain	F (120, 2520) = 1.330	P = 0.0109*
		Time	F (2.311, 48.53) = 49.66	P < 0.0001****
		Pain	F (1, 21) = 2.648	P = 0.1186
Figure 6E FR3	Reinforcers Both sexes	Time × pain	F (4, 107) = 0.3334	P = 0.8550
		Time	F (2.196, 46.65) = 1.917	P = 0.1548
		Pain	F (1, 22) = 0.04718	P = 0.8300
	Nose poke Both sexes	Time	F (7, 77) = 0.4868	P = 0.8414
		Nose poke	F (1, 11) = 1137	P < 0.0001****
		Pain	F (1, 11) = 0.9074	P = 0.3613
		Time × nose poke	F (7, 77) = 1.619	P = 0.1428
		Time × pain	F (7, 77) = 0.6145	P = 0.7423
		Nose poke × pain	F (1, 11) = 1.047	P = 0.3282
		Time × nose poke × pain	F (7, 7) = 0.7977	P = 0.6134
Figure 6F Progressive ratio	Break point	Pain	F (1, 20) = 0.1870	P = 0.6701
		Sex	F (1, 20) = 0.7723	P = 0.3899
		Pain × sex	F (1, 20) = 0.2384	P = 0.6307
Figure 6G FR1 reversal	Reinforcers	Time × pain	F (9, 178) = 0.7763	P = 0.6385
		Time	F (1.925, 38.07) = 2.221	P = 0.1241
		Pain	F (1, 22) = 1.412	P = 0.2474
	Nose pokes	Time	F (9, 99) = 52.85	P < 0.0001****
		Nose poke	F (1, 11) = 153.9	P < 0.0001****
		Pain	F (1, 11) = 0.05763	P = 0.8147
		Time × nose poke	F (9, 99) = 67.80	P < 0.0001****
		Time × pain	F (9, 99) = 1.014	P = 0.4341
		Nose poke × pain	F (1, 11) = 0.2603	P = 0.6200
		Time × nose poke × pain	F (9, 59) = 1.344	P = 0.2348
	Magazine Entries	Time × pain	F (10, 174) = 0.6821	P = 0.7402
		Time	F (4.484, 78.03) = 8.614	P < 0.0001****
		Pain	F (1, 22) = 0.3819	P = 0.5430

An unpaired t test was used for area under the curve (AUC) data. Cumulative reinforcers were analyzed by a repeated measures 2-way repeated ANOVA. P values are provided for each of the factors (time, pain, and appropriate interactions) for active and inactive nose pokes, and number of reinforcers earned. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

AUC, area under the curve; FR, fixed ratio.

When advanced to the FR3 schedule, the performance of CCI mice was indistinguishable from control mice for the number of rewards earned across days (Fig. 6E), as well as for the rate of acquisition of Ensure rewards within a session (data not shown). Accuracy for CCI mice was slightly decreased vs control mice (main effect of pain, F_{(1, 141)} = 18.94; P < 0.001; no interaction or main effect of time) but remained high around 80%.

Furthermore, when challenged on a PR schedule, the CCI mice were willing to work for the Ensure reward at rates similar to the sham mice (Fig. 6F). And, when returned to the FR1 schedule, both CCI and sham mice performance remained high. All mice performed well on the reversal learning task, earning similar numbers of Ensure rewards across days (Fig. 6G, Table 6). In summary, the CCI mice were willing to work for the Ensure reward, although they showed minor deficits in operant responding on the FR1 schedule, which is the reverse of what was observed on the same experimental timeline for remifentanil reward.

3.4. Chronic pain alters dopamine dynamics in the striatum

To investigate whether dopamine signaling is altered by chronic pain in the nucleus accumbens, we used fast-scan cyclic voltammetry to assess electrically stimulated dopamine levels in the nucleus accumbens core in anesthetized CCI and sham mice. Figure 7A illustrates the correct targeting of carbon-fiber microelectrodes upon postmortem examination. The CCI mice exhibited significantly reduced electrically stimulated extracellular dopamine levels compared to sham animals (Fig. 7B), which was also evident in the pseudocolor plots of voltammograms with respect to time (Fig. 7C). In a separate group of animals, we used microdialysis in behaving mice to measure basal and potassium-evoked release of dopamine and serotonin in the ventral striatum. Basal dopamine levels were not significantly reduced in the ventral striatum compared to sham animals (Fig. 7D). We used increasing lengths of high-potassium pulses delivered locally through the microdialysis probe to investigate stimulated dopamine and serotonin levels in the ventral striatum. Stimulated extracellular dopamine was increased in a manner dependent on the length of the high-potassium stimulus (Figs. 7E and F). Although potassium-evoked increases in extracellular dopamine were evident in the CCI mice, the magnitude was significantly reduced in pain animals compared to sham controls (Figs. 7E and F). We did not observe difference in ventral striatal basal serotonin levels between CCI and sham mice (Fig. 7G). Like dopamine, pulses of high potassium perfusion increased ventral striatal extracellular serotonin levels (Figs. 7H and I). However, only the longest potassium pulse was associated with a depression in potassium-evoked serotonin levels in the CCI mice.

Figure 7.

Chronic pain is associated with reduced evoked and phasic dopamine extracellular concentrations in the ventral striatum/nucleus accumbens. (A) Histological confirmation of carbon fiber microelectrode placement for fast-scan cyclic voltammetry experiments. (B) The medial forebrain bundle was electrically stimulated in anesthetized mice to produce extracellular dopamine release in the nucleus accumbens core (60 Hz, 24 pulses, 2 seconds after the start of recording). Chronic constriction injury mice showed a significant reduction in the amount of stimulated dopamine release compared to sham animals. (C) Representative 2-dimensional pseudo-color plots for the cyclic voltammograms. Changes in current at carbon-fiber microelectrodes are indicated by color. (D) Basal dopamine levels in the ventral striatum were measured in awake mice by high-performance liquid chromatography using microdialysis. Chronic constriction injury mice did not have lower basal dopamine concentrations compared to sham mice. (E) Reverse dialysis with high-potassium artificial cerebrospinal fluid was used to stimulate extracellular dopamine release. Overflow increased with the duration of the stimulation in both sham and CCI mice. (F) Evoked dopamine overflow was significantly reduced in the CCI mice compared to sham animals at all stimulus durations. (G) Serotonin levels were contemporaneously determined in the same mice in the ventral striatum. There was no effect of pain on basal serotonin levels. (H) Potassium stimulation increased extracellular serotonin concentrations in a duration-dependent manner in sham and CCI mice. (I) Although evoked serotonin overflow was reduced in CCI mice compared to control animals at all stimulus durations except the shortest (ie, 1 minute), differences only reached statistical significance at the 5-minute stimulus duration. Data from mixed-sex groups of mice (roughly equal in numbers) are expressed as means ± SEMs for N = 6 to 7 mice per group. *P < 0.05, ***P < 0.0001 CCI vs sham mice.

4. Discussion

Chronic pain causes significant neurobiological changes, but how these changes impact opioid use and the risk of developing OUD remains unclear. We found that although the acquisition of the number of reinforcers for remifentanil self-administration across days (ie, the number of reinforcers earned each day) was not significantly impacted by the chronic pain state, mice experiencing chronic pain associated with sciatic nerve injury showed enhanced drug-seeking behavior during this phase. This was evident in both the time to reach the maximum number of reinforcers as well as the number of animals that reached a maximum number of reinforcers for each day during the acquisition phase. The intrasession data within each 2 hours session on each day on the FR1 schedule show similar findings that pain animals show enhanced acquisition, which appear to be driven by male and not female pain animals. It was surprising then, when chronic pain animals transitioned to higher demand FR3 and progressive ratio schedules, the chronic CCI mice significantly decreased their willingness to work for opioid rewards.

The lack of differences in the numbers of reinforcers earned or the numbers of nose pokes when mice were transferred back to an FR1 schedule after the progressive ratio schedule suggests that the motivation to work for remifentanil was not a function of a change in reward value over time. Furthermore, in a reversal-learning task, both control and CCI mice were able to adjust had similar responding, suggesting that the reduced responding of chronic CCI mice was not because of a learning deficit. After the same schedules for a nondrug food reward (ie, 50% Ensure), the motivation for highly palatable food reward was not different in chronic CCI mice compared to control mice. These findings imply that an opioid agonist has effects that differ from natural food reward, which may be driven by opioid-induced analgesia.

To our knowledge, the impact of chronic neuropathic pain on increased demand/effort for opioid intravenous self-administration in mice has not been previously investigated. There are 2 conflicting reports worth noting using a model of chronic inflammatory pain in rats. One study examined heroin self-administration where decreased or increased drug taking was dose-dependent including when examining motivation during a progressive ratio task.²² The other study examined the effects of the same inflammatory pain model on fentanyl self-administration. This latter study did not observe differences in responding because of the effects of pain regardless of the duration of access or work effort.⁴ Both of these studies were conducted in rats, so it is unclear if our study is related to differences in species or more likely because of differences in chronic pain models. We acknowledge 1 limitation to the study is that the reduced work may be related to the dose, where only 1 dose, albeit high, was tested in this study. Previous studies showed that low vs high dose had opposite effects on motivation with increased work effort after CFA injection to induce persistent pain in rats previously trained to self-administer heroin.²² An early study determining the impact of arthritis on fentanyl self-administration also reported pain increased fentanyl intake.¹⁰ In this study, Colpaert et al. concluded that the arthritic rats consumed more fentanyl when offered as a choice, for analgesic therapeutic effects and explicitly not for reward. However, it is also notable that when fentanyl consumption was forced (no choice), there was reduced consumption relative to controls, which is consistent with some of the present study's observations. Martin et al.³⁰ also identified that rats with chronic neuropathic pain only self-administered various opioids to reach pain relief without further escalation, suggesting motivation for drug taking was negative reinforcement. Oral fentanyl self-administration in mice has been investigated in chronic pain models, including chronic inflammatory pain and 2 models of neuropathic pain. One study identified that the acquisition of oral fentanyl was reduced in chronic pain animals on a FR1 schedule, although food reward was not modified by pain,^15,45 as we report here. There are many possibilities for differences in outcomes between our study and some of the previous studies including the model, route of administration, species, and opioid dose, highlighting the importance of examining the effects of chronic pain on various aspects of drug-seeking behavior.

One interpretation of our findings is that chronic pain may be associated with a reduced risk of developing an OUD. In our study, mice were less willing to work for remifentanil reward as demand increased. That said, chronic CCI mice acquired remifentanil faster within sessions on an FR1 schedule. We cannot rule out that this increase in acquisition may be driven by negative reinforcement, where pain relief is the major motivator for drug taking. In clinical populations involving children and adults, one of the main reasons for the nonmedical use of opioids was for pain relief^2,7,12,40 where only 12% of these cohorts used opioids to “feel good” or “get high” (https://www.samhsa.gov/data/sites/default/files/reports/rpt39443/2021NSDUHFFRRev010323.pdf). Preclinical studies support the idea that pain relief is the driver for drug taking. For example, in a model of arthritis, rats self-administered opioids to obtain pain relief, rather than for its rewarding effects.^10,29 Nonetheless, motivation to take opioids for pain relief in chronic pain patients has been identified as a risk factor for developing a substance use disorder.²¹

Our data show that chronic pain did not modify food reward, although sham mice acquired food rewards faster than pain animals within sessions on the FR1 schedule. Unlike drug seeking, motivation for obtaining food reward was shown to be reduced during a progressive ratio task when conducted 48 hours after an inflammatory pain stimulus induced by complete Freund adjuvant.^31,41 Interestingly, there was no effect of inflammatory pain on motivation for food reward in a progressive ratio task if conducted at later time points after pain onset (Dr. Nicolas Massaly, University of California Los Angeles, personal communication). These findings are consistent with recent data in a mouse model of neuropathic pain where self-administration of food using a progressive ratio task in the home cage demonstrated that pain animals showed reduced work effort or motivation for food.³² The duration of pain may be a factor influencing study outcomes.

Although we cannot point to a causal relationship, we observed a hypodopaminergic state within the striatum that is likely contributing to a loss of motivation to work for the opioid reward. We used 2 approaches to assess mesolimbic dopamine function. Voltammetry, in vivo, showed that dopamine release was significantly reduced in the nucleus accumbens core of chronic neuropathic CCI mice. We also measured extracellular dopamine overflow via microdialysis, where there was a trend toward reduced basal dopamine in chronic pain animals. Importantly, evoked dopamine levels were reduced in chronic CCI mice, whereas there the effect on evoked serotonin was modest. A limitation of our study is that the dopamine measurements were not powered to detect potential sex differences so we cannot comment on possible differences in how chronic pain may modulate dopamine. The observed hypodopaminergic state in chronic neuropathic pain in our study may contribute to a loss of opioid effectiveness, and thus lowered reinforcer value for remifentanil. This may explain why food self-administration is maintained but there is a reduced effect with the opioid.

Previous research has demonstrated altered dopaminergic neurotransmitter and receptor function in chronic pain states in rodents and humans (reviewed in Taylor et al., 2016). We and others showed that opioid-evoked dopamine release in the nucleus accumbens was significantly attenuated in chronic pain animals independent of the etiology of the pain.^22,34,44 Dopamine and its metabolites in the ventral striatum were significantly reduced in a mouse chronic neuropathic pain.⁴⁵ Dopamine neuronal excitability within the lateral ventral tegmental area (VTA) was reduced in a spared nerve injury model of neuropathic pain.²³ This is consistent with another report on reduced VTA dopamine neuronal excitability and reduced extracellular dopamine in the nucleus accumbens in a model of neuropathic pain.³⁶ However, to what extent the hypodopaminergic state induced by chronic pain impacts drug seeking is not clear. Lesion studies in rats showed that dopamine depletion in the nucleus accumbens does not impact low-ratio responding but significantly decreased responding on higher ratio schedules.¹ This is consistent with our findings where the transition to FR3 and progressive ratio tasks uncovered reduced motivation for drug taking in CCI mice.

In conclusion, we provide evidence that chronic neuropathic pain is associated with a hypodopaminergic state that is correlated with reduced motivation (work effort) for opioid drug taking. It is unclear how this change in motivation may impact abuse liability, as initial drug taking was enhanced when effort was low. Significant decreases in VTA dopamine activity have been reported after periods of protracted abstinence that lead to craving and relapse.³⁷ Importantly, reward deficiency syndrome that shows a hypodopaminergic trait were at higher risk of addictive behavior.⁵ Our future studies will examine how reduced motivation impacts drug craving in periods of forced abstinence and in models of relapse.

Conflict of interest statement

The authors have nothing to disclose.