The effect of quinine in two bottle choice procedures in C57BL6 mice: opioid preference, somatic withdrawal, and pharmacokinetic outcomes. *

Abstract

Previous reports assessing morphine effects in two bottle choice (TBC) paradigms often use taste adulterants such as sweeteners (e.g., saccharin) and/or bitterants (e.g., quinine) to demonstrate morphine preference with C57BL6 mice. The effect of these additional components on the morphine preference of C57BL6 remains poorly understood. Thus, we sought to elucidate the interrelationship of morphine and quinine in the TBC paradigm. As expected, when morphine was included in the opposite bottle from quinine, a preference for the morphine solution was observed. Conversely, when quinine was included in each bottle, or when fentanyl without quinine was used, no preference was observed. All opioid-drinking mice displayed withdrawal signs, and morphine was detectable in plasma and brain. When these results were compared to previous results via conversion to quinine preference scores, quinine was revealed to determine largely the measured morphine preference. Thus, quinine is effective to drive morphine consumption and engender dependence but may confound the ability to measure oral abuse liability of morphine. Together, these results suggest future TBC procedures should consider the effect of quinine upon measured preference for compounds in the opposite bottle, and that excessively high quinine concentrations appear to influence preference more so than the opposite solute when using C57BL6 mice. Alternative conditions to assess oral abuse liability may be necessary to complement and confirm results from TBC experiments utilizing morphine or other opioids.

1. Introduction

Opioid abuse remains a persistent public health concern in the United States, resulting in over 33,000 deaths in 2015 alone (Rudd et al., 2016). From 1999 to 2015, the number of overdose deaths attributable to prescription opioids has quadrupled (Center for Disease Control, 2017), highlighting prescribed analgesics as a substantial contributing factor to the epidemic. Similar to heroin, prescription opioids possess unwanted effects in addition to their therapeutic properties, notably abuse liability (Jimenez et al., 2017). Unlike heroin, the typical route of administration for prescription opioids is not intravenous, but oral. In one study, more than 90% of prescription opioid abusers report having swallowed pills (McCabe et al., 2007), and in another study over half of the oxycodone, hydrocodone, and morphine users reported oral ingestion for nonmedical purposes (Katz et al., 2008). A more recent meta-analysis suggests 72–97% of prescription opioid abusers utilize the oral route of administration (Kirsh et al., 2012), highlighting a need to understand the abuse-related properties of opioid analgesics via oral drug taking.

Studies examining oral self-administration of drugs of abuse often utilize choice procedures in which preference for drug-containing solutions is assessed against non-drug solutions. Of the many variations of this procedure utilizing opioids, two bottle choice (TBC) features morphine, which exerts most of its effects in vivo via the MOR opioid receptor (Matthes et al., 1996), and has often been used to examine strain differences exhibited by C57BL6 versus DBA mice. Generally, C57BL6 mice exhibit a preference for morphine solutions while DBA mice prefer controlled or non-morphine containing solutions (Doyle et al., 2014, 2008). Morphine is often described as having a bitter taste; thus, mouse models of oral opioid self-administration commonly employ sweeteners like sucrose or saccharin in the morphine solution, and bitter-tasting quinine opposite morphine, or both, to eliminate this confound. Indeed, when presented in otherwise unadulterated tap water, both C57BL6 and DBA mice prefer non-morphine solutions (Horowitz, 1977; Horowitz et al., 1977). Addition of sucrose or saccharin to the morphine solution elevates morphine consumption (Belknap et al., 1993; Horowitz, 1977; Horowitz et al., 1977), while the addition of quinine to the opposite bottle produces a shift in preference toward the morphine bottle (Forgie et al., 1988). Some studies utilize both saccharin and quinine to shift morphine preference (Doyle et al., 2014, 2008; Ferraro et al., 2004; Forgie et al., 1988), although the interrelationship among the three solutes is typically not investigated in great detail. A summary of the variety of conditions used to assess morphine and other opioid preference in the two-bottle choice paradigm can be viewed in Supplementary Table 1.¹

The degree to which quinine and/or saccharin influence the ensuing preference for morphine of C57BL6 mice remains poorly understood. C57BL6 mice are typically described as morphine-preferring and will consume from 50–200 mg/kg/day (Belknap, 1990; Belknap et al., 1993; Doyle et al., 2014, 2008; Ferraro et al., 2004), depending on the concentration of morphine, and the presence or absence of taste adulterants. This level of intake results in somatic signs of opioid withdrawal following treatment with naloxone (Belknap, 1990), suggesting these conditions are sufficient to produce dependence in C57BL6 mice. The degree to which morphine, taste adulterants, or the interaction of these variables influences the ensuing intake remains unclear, however. The purpose of these experiments was to elucidate the role of morphine in the test bottle versus quinine in control bottles, and whether the concentrations utilized were sufficient to produce preference and dependence under a variety of conditions. The potent MOR agonist fentanyl was included to control for potentially confounding taste effects of morphine, and the non-steroidal anti-inflammatory drug (NSAID) ketoprofen was included as a non-psychoactive analgesic control. Following the TBC procedure, precipitated withdrawal was induced with naloxone to assess dependence. Plasma and brain were also collected and assessed for the presence of the test compounds via mass spectrometry to confirm intake. A novel vehicle consisting of 1% DMSO, 0.2% saccharin, and 0.5% Captisol® was utilized to enhance the solubility of the compounds tested, and as such this vehicle was also assessed for its amenability to the TBC procedure.

2. Methods

2.1. Animal Subjects

Male C57BL6J mice originating from Jackson Labs (Bar Harbor, ME) were approximately 12–17 weeks of age at the beginning of testing and were bred in-house. A total of 28 mice were used for these experiments. Each mouse was only used once and maintained on a 6 am to 6 pm light cycle. All procedures were in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals with approval by The Scripps Research Institute Animal Care and Use Committee.

2.2. Chemicals and Drugs

Morphine sulfate pentahydrate was supplied by the National Institute on Drug Abuse (NIDA) Drug Supply Program at RTI International (Durham, NC). Ketoprofen, fentanyl citrate, naloxone hydrochloride, saccharin, and quinine hydrochloride were acquired from Sigma-Aldrich (St. Louis, MO). Captisol (β-cyclodextrin sulfobutyl ethers) was purchased from Cydex (Lawrence, KS).

2.3. Bottle Construction

To measure daily liquid intake and solution preference, bottles were constructed from 50 ml Falcon centrifuge tubes (ThermoFisher; Waltham, PA), 2.5 in straight sipper tubes (ThermoFisher; Waltham, PA), and washers (1/4 in inner diameter). Briefly, the flat bottom of the Falcon tube was removed with a Dremel, and a sipper tube pushed through until 2 inches of the sipper tube protruded from the end. Parafilm was wrapped around the junction of the tube and sipper to minimize leaking, and four washers were placed over the sipper tube to minimize the mice’s abilities to access the Parafilm. Prior to use in experiments, each bottle was filled with 25 ml of water and placed in food hoppers of empty cages to assess baseline leakage. Any bottle which leaked more than 0.3 ml in a 24-hour period was discarded.

2.4. Two Bottle Choice Procedure

At the start of the two-bottle choice (TBC) procedure, mice were single-housed and provided two bottles, one “test” and one “control”. Each bottle contained 15–20 ml of drinking liquid that was replenished daily. At least 50 g of standard rodent chow was provided daily, and consumption of liquid from each bottle and food intake was assessed each day between 10:00 am and 12:00 pm. The initial left/right positions of the test and control bottles were distributed evenly among subjects for each experiment, and the position of the test and control bottles was alternated daily to control for any potential positional preference. Initially, both bottles were filled with deionized water, and the mice were acclimated to this set-up for 4–7 days. After acclimation, a vehicle drinking solution containing 1% DMSO (v/v), 0.2% saccharin (w/v), and 0.5% Captisol (w/v) was provided in both test and control bottles for 4 days. The exact dosing paradigms for the different experimental protocols can be found in each figure. Pilot experiments demonstrated 1% DMSO did not significantly change the total liquid intake, and the 0.2% saccharin concentration was selected because previous studies have demonstrated this is a maximally preferred concentration of saccharin for C57BL6 mice. Whenever the contents of the test or control bottles were changed, bottles were flushed with dH₂O at least 3 times to eliminate the previous contents.

2.5. Naloxone-Precipitated Withdrawal and Tissue Collection

On the final day of TBC, mice were removed from their home cages, injected with naloxone HCl (1 mg/kg i.p., in 0.9% saline), placed in a Plexiglass tube (16 in tall x 6 in diameter), and monitored for paw tremors, wet dog shakes, jumps, mastication, and diarrhea for 30 min. Global withdrawal scores were calculated using the following equation: 0.8x(# of jumps) + 1x(# of wet dog shakes) + 0.35x(# of paw tremors) + 1.5x(diarrhea present each 5 minute bin) + 1.5x(mastication present each 5 minute bin) (Raehal and Bohn, 2011).

2.6. Quantification of brain and plasma levels

Immediately following withdrawal observation, mice were humanely sacrificed via rapid decapitation under isoflurane anesthesia. Brains were frozen in liquid nitrogen and stored at −80°C. Blood was collected into EDTA-coated CapiJect tubes and centrifuged at 20,000 x g for 10–15 min at 4°C. The plasma fraction was transferred to 1.7 ml microcentrifuge tubes and stored at −80°C.The plasma supernatant was collected and the remaining pellet discarded. Brains and plasma were stored at −80°C until quantification. Compound concentrations in brain and plasma were evaluated using liquid chromatography (Shimadzu)–tandem mass spectrometry (AB Sciex) (LCMS) operated in a positive-ion mode using multiple reaction monitoring methods (Brust et al., 2016; Zhou et al., 2013).

2.6. Data Analysis

All data points represent the means ± S.E.M. of two or four days of testing. Liquid intake was assessed by measuring the difference in daily bottle weight and assuming a density of 1 g/ml. Total liquid intake was calculated by adding together the intake from both the test and control bottles, and preference was calculated as follows: (test bottle intake) / (total liquid intake) *100. Drug intake was calculated as: test bottle intake * test compound concentration (mg/ml) / daily weight (kg). Preference, total liquid intake, and somatic signs of withdrawal were analyzed via one-way or two-way ANOVA where appropriate, followed by Tukey’s post-hoc test. In cases where a separate vehicle group was not included, a statistical assessment of preference was performed by calculating 95% confidence limits for each concentration of the test drug. If the confidence limits did not include 50%, then the concentration was considered preferred.

3. Results

3.1. C57BL6 Mice Exhibit A Preference for The Morphine Solution With 0.2 Mg/Ml Quinine in The Opposite Bottle.

To assess whether the novel vehicle would be amenable to the morphine TBC procedure, mice were acclimated to diH₂O in both bottles for 4 days, acclimated to the vehicle in both bottles for 4 days, and then morphine or ketoprofen was added to the test bottles (Fig 1A). Some mice were maintained on the vehicle for the duration of the experiment. Morphine alone (0.3 mg/ml) induced a significant decrease in preference for the test bottle, while ketoprofen (0.3 mg/ml) did not result in a significant decrease relative to vehicle (Fig 1B, p<0.0001, F (10, 48) = 16.41). When quinine (0.2 mg/ml) was introduced in the control bottle, the mice preferred the test bottles containing ketoprofen (0.3 mg/ml), vehicle or morphine (0.3 mg/ml), with morphine inducing elevated levels of preference relative to both vehicle and ketoprofen (Fig 1B, p<0.0001, F (10,48) = 16.41). The highest concentration of morphine (0.6 mg/ml) resulted in an equal preference for a vehicle, whereas the highest concentration of ketoprofen (0.6 mg/ml) was not preferred relative to the control bottle (Fig 1B, p<0.0001, F (10, 48) = 16.41). An additional phase of morphine (0.6 mg/ml), ketoprofen (0.6 mg/ml), and vehicle resulted in no change in preference for any test solution (Fig 1B, p<0.0001, F (10, 48) = 16.41). Vehicle increased liquid intake relative to diH₂O in all groups, and intake remained at vehicle levels through the remainder of testing (Fig 1C, p<0.001, F (10, 48) = 5.128). Morphine alone resulted in the relatively low intake, and the addition of quinine to the control bottle significantly increased morphine intake (Fig 1D, p<0.0001, F (3, 20) = 32.20). Increasing the morphine concentration resulted in a further increase of morphine intake, while the continuation of the TBC procedure resulted in a small increase of morphine intake (Fig 1D, p<0.0001. F (3, 20) = 32.20). Ketoprofen intake did not differ significantly throughout testing.

Figure 1.

C57BL6 mice display preference for morphine in the DMSO/Captisol vehicle.

Note: Previous results with TBC paradigms show a preference for morphine (0.3–0.7 mg/ml) in saccharin-sweetened water, opposite quinine. The novel vehicle developed to enhance solubility was assessed under similar conditions. A. A schematic of the test and control bottle contents throughout the experiment. B. Mice did not exhibit a positional preference for diH2O or vehicle in any group. When morphine (MS, 0.3 mg/ml) was added to the control bottle, a preference for the vehicle bottle was observed. Addition of ketoprofen (K, 0.3 mg/ml) did not shift preference relative to mice maintained on vehicle. Addition of quinine to the opposite control bottle shifted preference to the morphine bottle, while addition of ketoprofen had no effect. Increasing the concentration of morphine (M, 0.6 mg/ml) did not enhance preference beyond vehicle levels, while increasing the ketoprofen (K, 0.6 mg/ml) concentration resulted in a preference for the quinine containing bottle, relative to vehicle. C. Mice from each group consumed 3–5 ml/day of diH2O and displayed a moderate increase in liquid intake when switched to saccharin-sweetened vehicle. Total liquid intake remained consistent throughout testing once switched to vehicle irrespective of bottle contents. D. When morphine was included in the test bottle, mice consumed 14.2±2.3 mg/kg/day, and subsequent addition of quinine (0.2 mg/ml) increased consumption to 26.7±3.7 mg/kg/day. Higher morphine concentrations further increased consumption (50±5.7 mg/kg/day), and a second four-day period resulted in a slight but significantly higher increase in morphine intake (to 64.7±11.5 mg/kg/day). Ketoprofen intake did not significantly differ with respect to test and control bottle contents for the duration of testing. N=7, one- or two-way ANOVA where appropriate, followed by Tukey post hoc test, morphine versus vehicle, *p<0.05, **p<0.01, ****p<0.0001; morphine versus ketoprofen, ^^^^p<0.0001; ketoprofen versus vehicle, $ $ $ $p<0.0001.

3.2. The Presence of Quinine (0.1 Mg/Ml) In Both Test and Control Bottles Eliminates Preference for Morphine

To determine the effect of quinine upon the preference of morphine, quinine was included in the control bottle alone (Fig 2A), or in both the test and control bottles (Fig 2B). A lower concentration of quinine (0.1 mg/ml) was used for these experiments to avoid an overall suppression of liquid intake. No preference was observed when quinine was included in both bottles, while the addition of quinine to the control bottle alone induced a preference for the morphine containing solution morphine (Fig 2C). Addition of 0.1 mg/ml morphine had no effect on preference, while 0.17 mg/ml elicited a significant change in preference (preference score (± 95% confidence limits: 67.4 ± 13.0)) (Fig 2C). Increasing the morphine concentration (0.3 mg/ml and 0.56 mg/ml) returned preference to baseline scores in both groups (Fig 2C). Total liquid intake did not significantly change throughout testing (Fig 2D). To ascertain whether morphine intake was greater than would be expected proportionally to the concentration in the bottle, the log concentration of morphine was plotted against the log of morphine intake calculated as mg/kg/day (Fig 2E). In both cases, the slope of the line did not significantly differ from unity (Table 1).

Figure 2.

Preference for morphine solutions is observed when quinine is included in the control bottle, but not both.

Note: One group of mice had quinine added to both the test and control bottles throughout the testing (both), the other group of mice had quinine only in the control bottle (control). Both groups were exposed to morphine starting on day 8. A. A schematic detailing the contents of test and control bottles for the “control” cohort in which quinine was only present in the control bottle. B. The schematic displays the contents of each bottle for the “both” cohort in which quinine was present in both test and control bottles in addition to their other contents. C. Mice displayed preference for the test bottle at 0.17 mg/ml morphine when only the control bottle contained quinine. No preference was observed for vehicle or any dose of morphine assessed in the cohort receiving quinine in both bottles. D. Total liquid intake remained at 4–5 ml/day throughout testing and did not differ between groups at any dose morphine (p=0.1617). E. The log of the concentration of MS in the test bottle was compared to the log of the resulting intake of morphine (mg/kg/day). The slope did not differ from unity for either the “control” cohort or the “both” cohort (Table 1). *confidence limits did not include 50, n=5.

Table 1.

The concentration of the MOR agonist in the test bottle was plotted against the measured daily consumption and the slope of the line was assessed for divergence for unity, indicating escalation of intake.

Bottle Contents		[MOR agonist] vs consumption (mg/kg/day)
Test	Control	slope (95% confidence limits)	R2
morphine (0.1–0.56 mg/ml)	quinine (0.1 mg/ml)	0.9175 (0.7472 to 1.088)	0.9051
morphine (0.1–0.56 mg/ml)+ quinine (0.1 mg/ml)	quinine (0.1 mg/ml)	0.9919 (0.8056 to 1.1780)	0.9031
fentanyl (0.003–0.056 mg/ml)	vehicle	0.8876 (0.7822 to 0.9933)	0.9588

In all cases, the concentration of the MOR agonist in the test bottle was strongly correlated with the intake measured. Confidence limits of the slope for both morphine groups included unity, while those of the fentanyl group were below 1.

3.3. Mice Display No Preference for The Fentanyl Solution in The Absence of Quinine

To control for potentially confounding taste effects of morphine, the chemically distinct, potent MOR agonist fentanyl was assessed for preference in the absence of quinine. Following diH₂O and vehicle acclimation, mice were provided with increasing concentrations of fentanyl (3–56 µg/ml) in the test bottle (Fig 3A). An initial preference for the test bottle was present when both bottles contained diH₂O (preference score (95% confidence limits): 62.8 ± 2.9), this preference did not persist when the vehicle was introduced into both bottles (preference score (95% confidence limits): 55.3 ± 10.0) (Fig 3B). Mice exhibited no preference at any concentration of fentanyl tested (Fig 3B, p=0.0609, F (1.612, 4.835) = 5.463), and total liquid intake was not significantly different from that observed when water alone was in the bottles for any condition tested (Fig 3C, p=0.6309, F (1.92, 3.576). The log (mg/kg/day) of fentanyl consumed displayed a linear relationship to the log concentration of fentanyl in the test bottle, although the 95% confidence limits did not include unity (Table 1).

Figure 3.

No preference is observed for fentanyl at any concentration tested, in the absence of quinine.

Note: Fentanyl was included in the test bottle while only vehicle was present in the control bottle. A. The schematic depicts bottle contents throughout the duration of testing. All concentrations of fentanyl are expressed as µg/ml. B. An initial preference for the test bottle was observed when both the test and control bottle contained dH₂O (preference score (95% confidence limits): 62.8 ± 2.9), though this effect was absent in the vehicle phase of testing (preference score (95% confidence limits): 55.3 ± 10.0). No dose of fentanyl resulted in preference. C. Total liquid intake did significantly differ for the duration of testing (p=0.6309). D. The fentanyl consumed (log (mg/kg/day)) was strongly correlated with the log concentration of fentanyl in the test bottle (slope (95% confidence limits): 0.8876 (0.7822 to 0.9933), R²=0.9588).

3.4. C57BL6 Mice Which Consumed Morphine and Fentanyl Display Somatic Signs of Opioid Withdrawal

On the final day of testing for all of the studies presented above, mice were administered naloxone (1 mg/kg, i.p.) and observed for somatic signs of opioid withdrawal (paw tremors, wet dog shakes, jumps, mastication, and diarrhea) for 30 minutes. Morphine- and fentanyl-drinking groups displayed significant increases of paw flutters (Fig 4A, p<0.01, F (5, 29) = 4.688), wet dog shakes (Fig 4B, p<0.05, F (5, 29) = 3.135), jumps (Fig 4C, p<0.01, F (5, 29) = 4.425), diarrhea (Fig 4D, p<0.01, F (5, 29) = 5.396), and mastication (Fig 4E, p<0.0001, F (5, 29) = 11.42) while vehicle- and ketoprofen-drinking groups did not exhibit these withdrawal signs. Calculated global withdrawal scores reveal morphine- and fentanyl-drinking groups displayed significant somatic signs of naloxone-precipitated withdrawal (Fig 4F, p<0.0001, F (5, 29) = 7.932).

Figure 4.

Morphine- and fentanyl-drinking groups exhibited somatic signs of opioid withdrawal.

Note: On the final day of the TBC procedure, mice were removed from their home cages, administered naloxone (1 mg/kg, i.p., in 0.9% saline), and monitored for somatic signs of opioid withdrawal. Morphine- and fentanyl-drinking mice displayed significant increases in A. paw tremors, B. jumps, C. wet dog shakes, D. diarrhea, and E. mastication, while vehicle- and ketoprofen-drinking groups did not. Global withdrawal scores indicate only groups which received MOR agonists in the test bottle exhibited somatic signs of withdrawal associated with opioid dependence. The quinine concentration (mg/ml) and the presence of quinine in either control (C) or both (B) bottles is indicated in the legend below each panel. *<p0.05, **p<0.01, ****p<0.0001 versus vehicle.

3.5. Morphine Is Detectable in Brain and Plasma of Morphine-Drinking Mice Following Naloxone-Precipitated Withdrawal

Immediately following withdrawal scoring, plasma and whole brain were collected, and the concentration of the drugs in the plasma and brains were quantified via mass spectrometry to confirm drug intake. Morphine was detectable in both plasma and brain in all morphine-drinking groups, while fentanyl was below the limit of detection. Ketoprofen was also detectable in both plasma and brain (Table 2).

Table 2.

Following naloxone-precipitated withdrawal experiments, mice were euthanized, plasma and brain collected, and levels of each test compound quantified by LC/MS analysis.

Bottle Contents		Samples
Test	Control	brain (ng/ml) ± SEM	plasma (ng/ml) ± SEM	N
morphine (0.3–0.6 mg/ml)	quinine (0.2 mg/ml)	10.1 ± 1.1	17.0 ± 3.7	7
ketoprofen (0.3–0.6 mg/ml)	quinine (0.2 mg/ml)	11.2 ± 3.3	1049.6 ± 320.7	7
morphine (0.1–0.56 mg/ml)	quinine (0.1 mg/ml)	33.1 ± 12.8	116.8 ± 61.5	5
morphine (0.1–0.56 mg/ml) + quinine (0.1 mg/ml)	quinine (0.1 mg/ml)	17.0 ± 4.8	66.9 ± 34.9	5
fentanyl (0.003–0.056) mg/ml)	vehicle	< 5	< 5	4

Morphine and ketoprofen were detectable in both plasma and brain, while fentanyl levels were below the limit of detection.

3.6. Preference for Morphine-Containing Solutions Is Dependent Upon the Concentration of Quinine in The Opposite Bottle

To compare the morphine versus quinine preferences measured here to previously published reports, morphine preference scores for studies utilizing similar conditions were re-calculated as quinine preference scores by inverting preference values (i.e., 100 – morphine preference = quinine preference). These values were then plotted against a standard curve for quinine preference versus tap water for C57BL6 mice generated from Bachmanov et al., 1996. For previous studies, no calculated quinine aversion score fell outside of the 95% confidence limits of the standard curve (Fig 5A). For the preference scores generated here (Fig 1A and Fig 2A), only 0.3 mg/ml and 0.56 mg/ml morphine versus 0.1 mg/ml quinine lay outside of the 95% confidence limits of the quinine standard curve (Fig 5B).

Figure 5.

The concentration of quinine in the control bottle dictates the observed morphine preference.

Note: A standard curve of quinine aversion for C57BL6 mice was generated from Bachmanov et al. 1996 to assess the influence of quinine on the measured morphine preference across previously published resulted in those generated herein. A. Quinine aversion scores from recent studies utilizing morphine versus quinine in a two-bottle choice procedure were calculated by inverting morphine preference scores. These recalculated values were plotted against the generated standard curve (with 95% confidence limits). No calculated values differed from the standard curve. B. Preference scores from Figure 1A and2A were similarly inverted and plotted on the quinine standard curve. Only the two highest doses of morphine (0.3 and 0.56 mg/ml) tested at the lower concentration of quinine (0.1 mg/ml) fell outside of the 95% confidence limits of the quinine standard.

4. Discussion

This series of experiments investigated the role of morphine and quinine and their ability to affect preference in the TBC procedure, utilizing a novel vehicle. A maximally preferred saccharin concentration (0.2%) was included to enhance intake from both bottles, with morphine present in the test bottle and quinine in control bottle to replicate recent results using a tap water vehicle (Doyle et al., 2014, 2008; Ferraro et al., 2004). Similar to previous results, a preference for the morphine solution was observed, validating the use of the novel vehicle. The inclusion of the NSAID ketoprofen instead of morphine resulted in a starkly different pattern of preference as compared to morphine, in which ketoprofen was preferred at the low concentration, but not at the high concentration.

Quinine concentrations are often reported to be comparably bitter to the selected morphine dose. When 0.1 mg/ml quinine was included in the control bottle, only one concentration of morphine (0.17 mg/ml) produced preference, suggesting a narrow dose range at which morphine would be orally reinforcing. Higher concentrations of morphine resulted in no preference for either solution, suggesting these doses were equivalently reinforcing to 0.1 mg/ml quinine. The inclusion of quinine in both bottles did not result in a change in preference at any concentration of morphine tested. Because no preference for morphine developed when controlling for the taste effects of quinine, the preference observed for morphine when opposite quinine is likely a function of the concentration of quinine, rather than morphine. Additionally, morphine intake was correlated with the concentration of morphine, indicating no escalation of intake.

To eliminate the use of quinine and to avoid potential taste-related effects of a MOR agonist, the highly potent MOR agonist fentanyl was included in the test bottle versus vehicle in the control bottle. A subset of these doses has been used in oral operant paradigms with C57BL6 (Wade et al., 2013, 2008) and three bottle choice procedures with 129-derived mice (Jimenez et al., 2017), and served as reinforcers in some cases. Under our experimental conditions, no preference was observed at any concentration tested. Though fentanyl has not been previously utilized in TBC to our knowledge, the highly potent MOR agonist etonitazene has been assessed for preference-inducing properties (Forgie et al., 1988). Because etonitazene is approximately 1000-fold more potent than morphine (Madia et al., 2009), it could provide separation in potency between preference-related activation of MOR, and taste-related aversive-like effects. Similar to results of fentanyl reported here, no significant preference was observed, although only a single concentration of etonitazene was tested (Forgie et al., 1988). The correlation between fentanyl concentration and intake was below unity, demonstrating higher concentrations may lead to the lower overall intake. However, because the upper 95% confidence limit differed only slightly from unity, it remains possible that intake scales linearly with the concentration present in the bottle, as with the morphine groups.

Dependence was assessed via naloxone-precipitated withdrawal. All morphine and fentanyl treated groups, irrespective of calculated preference, displayed significant somatic signs of withdrawal relative to the vehicle controls. Accordingly, ketoprofen-drinking groups did not display significant withdrawal signs relative to vehicle. These experiments reaffirm the TBC paradigm as an effective means to induce oral dependence and mechanistically demonstrates chronic oral ingestion of opioids.

Ingestion of test compounds was verified with mass spectrometry of plasma and brain tissue collected following the precipitated withdrawal procedure. Morphine was detectable in samples from morphine-drinking groups, though there was notable and statistically significant variability between groups. One possible explanation for the differences measured between morphine-drinking groups lies in the selected conditions for each experiment. For mice with quinine (0.2 mg/ml) in the opposite bottle, they on average consumed more morphine (64.7 ± 11.5 mg/kg/day, Fig 1D) at the final tested dose (0.6 mg/ml) than their counterparts with, either 0.1 mg/ml quinine group (42.6 ± 4.1 and 37.0 ± 2.5, Fig 2E). Somewhat counterintuitively, the amount of morphine detected in plasma and brain was lower in the group with the highest preference (Table 2). This observation may be explained by several considerations regarding the design of this assay. First, mice had constant access to the bottles in their cages, excepting the time required to measure consumption and refill the bottles. Though the amount of total liquid consumed in a 24-hour period may be easily assessed, these experiments lack temporal sensitivity insofar as the patterns of drinking, i.e., frequency and magnitude of ingestion over time. Mice could ingest morphine constantly, or consume it intermittently in some number of bouts; thus, the preference merely reflects an overall distribution of drinking behavior.

Additionally, detection of morphine in plasma and brain depends on the amount consumed and how close to a collection that morphine was imbibed. Because plasma and brain were collected approximately 40 minutes after removal from the cage, ongoing drug metabolism and clearance that occurred would also diminish the total amount of morphine ultimately measured in samples (Kalvass et al., 2007). In the case of fentanyl, the half-life in mice is 4.9 ± 1.3 min (Kalvass et al., 2007), rendering the detection in plasma and brain unlikely after 40 minutes since the last ingestion of fentanyl. Fentanyl is also a substrate of P-glycoprotein, found in the blood-brain barrier (BBB), which enhances the efflux of fentanyl contributing to its rapid clearance (Dagenais et al., 2004; Mercer and Coop, 2011). Conversely, ketoprofen was readily detected in plasma, but comparatively little was measured in brain, likely reflecting its poor brain penetrance.

A brief review of previous results utilizing quinine opposite of morphine revealed that in all cases C57BL6 mice preferred sweetened morphine to sweetened quinine. The concentration of quinine varied from 0.06 mg/ml to 0.4 mg/ml, though the selected quinine concentration did not vary systematically across studies with respect to the selected morphine concentration. Importantly, these morphine preference scores could be recalculated as quinine preference scores simply by inverting the values (i.e., 100 – morphine preference = quinine preference). When these recalculated values were plotted against a standard curve generated from a quinine versus tap water study in C57BL6 mice (Bachmanov et al., 1996), none of the recalculated values deviated from the 95% confidence limits generated from linear regression of the quinine curve (Fig 5A). Plotting our own quinine preference values (measured with morphine in the opposite bottle) revealed that only 0.3 and 0.56 mg/ml morphine opposite 0.1 mg/ml significantly deviated from the quinine standard curve (Fig 5B). Whether or not higher concentrations of morphine would engender a preference for quinine was not determined, but these results strongly implicate the selected quinine concentration as an important determining factor in the measured preference for the morphine-containing bottle.

Though more recent studies have included a sweetener, usually saccharin, in both bottles to enhance palatability, initial TBC procedures utilized sweetened morphine solutions versus tap water. The most widely used saccharin concentration (0.2%) was selected on the basis of being maximally preferred by both C57BL6 and DBA mice (Horowitz et al., 1977). When this concentration is compared to a concentration-response curve for saccharin preference (Lush, 1989), 0.2% is revealed to induce a maximal preference for C57BL6 mice. Though not implicated explicitly in TBC, morphine and saccharin have known pharmacological interactions. Some studies indicate saccharin may potentiate morphine-induced anti-nociception in mice (Abdollahi et al., 2000; Nikfar et al., 1997), while non-contingent morphine also seems to influence saccharin preference in rats (Touzani et al., 1991). The use of a saccharin concentration which is only slightly preferred to tap water may help to reveal the reinforcing properties of oral morphine by potentiating consumption such that a higher preference would be observed for morphine-saccharin as opposed to morphine or saccharin alone. Another possibility would be integration of other paradigms which are known to increase liquid intake such as intermittent access (Skupio et al., 2016), drinking in the dark (Jimenez et al., 2017), or post-prandial conditions (Enga et al., 2016) which may increase the likelihood to unveil the reinforcing properties of oral opioids. Adenosine monophosphate, an apparent antagonist of bitter receptors (Ming et al., 1999), may also be used to increase the palatability of morphine or other bitter alkaloids. Regardless of the chosen method, continued efforts should focus on identifying conditions which produce a preference for opioid-containing solutions with minimal influence of taste adulterants, as these additions potentially confound the interpretation of measured preferences. Because of the influence of quinine on solution choice, care must be taken when interpreting the results of these studies. Many researchers refer to C57BL6 mice as “morphine-preferring”, a moniker which can overemphasize the influence of morphine upon the measured choice behavior. The results of these experiments suggest “quinine-avoiding” may be a more appropriate descriptor, and that a two-bottle choice procedure which employs quinine may be a reliable method to induce oral morphine intake but may not speak the abuse-related properties of morphine. Despite the taste-related limitations of opioid preference studies, identification of an oral abuse liability assay would be of great utility for preclinical assessment of candidate opioid analgesics.

Highlights.

• C57BL6 morphine/quinine preference

• Dependence and pharmacokinetic outcomes

Abbreviations:

(TBC)

Two Bottle Choice

(MOR)

Mu Opioid Receptor

(NSAID)

Non-Steroidal Anti-Inflammatory Drug