Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 May 1.
Published in final edited form as: Pharmacol Biochem Behav. 2020 Mar 13;192:172907. doi: 10.1016/j.pbb.2020.172907

Environmental Enrichment and a Selective Metabotropic Glutamate Receptor2/3 (mGluR2/3) Agonist Suppress Amphetamine Self-administration: Characterizing Baseline Differences

Erik J Garcia 1, Mary E Cain 1
PMCID: PMC7192569  NIHMSID: NIHMS1580365  PMID: 32179027

Abstract

A challenge for developing effective treatments for substance use disorders (SUDs) is understanding how environmental variables alter the efficacy of therapeutics. Environmental enrichment condition (EC) enhances brain development and protects against behaviors associated with drug abuse vulnerability when compared to rats reared in isolation (IC) or standard conditions (SC). EC rearing enhances the expression and function of metabotropic glutamate receptor2/3 (mGlurR2/3) and activation of mGluR2/3 reduces psychostimulant self-administration (SA). However, the ability for mGluR2/3 activation to suppress amphetamine (AMP) SA in differentially reared rats is not determined. Therefore, we tested the hypothesis EC reduces AMP (SA) by augmenting mGluR2/3 function. At postnatal day 21, male Sprague-Dawley rats were assigned to EC, IC, or SC environments for 30 days. Then, they acquired AMP SA and were moved to a progressive ratio (PR) schedule of reinforcement. EC, IC, and SC rats were pretreated with LY379268 (vehicle, 0.3 and 1 mg/kg), a selective mGluR2/3 agonist, before PR behavioral sessions. Linear mixed effects analysis determined EC rats had reduced motivation for AMP SA when compared to IC or SC rats and that LY379268 dose-dependently suppressed AMP SA, but there was no evidence of an interaction. Cumming/Gardner-Altman estimation plots illustrate that the 0.3 mg/kg dose suppressed infusions in EC rats while the 1 mg/kg dose suppressed infusions in SC rats. LY379268 was incapable of suppressing the motivation for AMP SA in IC rats. Controlling for baseline differences in differentially reared rats remains a challenge. Normalizing to a baseline introduced error which is illustrated in the precision of the estimated effect size differences. The data indicate that environmental enrichment enhances the potency of a selective mGluR2/3 agonist and indicates the functional status of the mGluR2/3 is formed during development. Therefore, environmental history must be considered when evaluating pharmacological therapeutics particularly those aimed at the mGluR2/3.

Keywords: Environmental Enrichment and Differential Rearing, Amphetamine self-administration, Progressive Ratio schedule of reinforcement, Metabotropic glutamate receptors, Statistical inference

1. Introduction

Amphetamine (AMP) use remains a significant public health concern because use continues to rise in the United States (NIDA 2014, Robles 2018, Goodnough 2019). Overdose deaths caused by psychostimulants with abuse potential (including amphetamine/dextroamphetamine (AMP) have increased by 22.3% annually from 2008 to 2017 (Kariisa, Scholl et al. 2019). An obstacle to developing effective treatments for substance use disorders (SUDs) is understanding how environmental variables change the efficacy of therapeutics (Kelly and Hannan 2019).

To determine the impact of broad environmental and social variables on SUDs, the differential rearing paradigm was developed (Rosenzweig and Bennett (1969). Enriched condition (EC) rats are group-housed in large cages with novel objects and handled daily during the post-weaning period. In contrast, isolated condition (IC) rats are reared in hanging metal cages in which the bedding, food, and water are changed without handling the rats. The EC and IC rats differ in the amount of handling, physical activity, social interactions, size of their home cage, the complexity of visual and tactile experiences, and the presence of novel objects (Renner and Rosenzweig 1987). The presence of other animals, contact with novel objects, and the amount of handling are all critical elements that create the EC (Renner and Rosenzweig 1987). Social stimulation alone is not sufficient to produce neurobiological changes similar to those observed with EC rearing and housing (Rosenzweig, Bennett et al. 1978, Renner and Rosenzweig 1987). In the standard condition (SC), rats are housed in pairs without novel stimuli in standard cages. The amount of complexity present in the rearing condition is correlated with greater neuronal and behavioral changes and therefore, environments with fewer cohorts, less contact with novel objects, or reduced handling display more transient behavioral and neurobiological outcomes (Renner and Rosenzweig 1987). EC rearing improves spatial learning, enhances extinction learning, and slows the acquisition of cocaine and AMP self-administration (SA). Furthermore, EC rats self-administer less AMP, cocaine, and methylphenidate when compared to rats reared in the IC (Green, Gehrke et al. 2002, Bardo and Dwoskin 2004, Fone and Porkess 2008, Stairs and Bardo 2009, Wooters, Bardo et al. 2011, Bardo, Neisewander et al. 2013). Unequivocally, the EC induces broad neurobiological changes to brain structure and function that protects against psychostimulant SA.

Differential rearing significantly alters metabotropic glutamate receptors (mGluR), glutamate transporters, and glutamate release (Sharp, McNaughton et al. 1985, Green and Greenough 1986, Renner and Rosenzweig 1987, van Praag, Kempermann et al. 2000, Artola, von Frijtag et al. 2006). Together, these mechanisms maintain optimal glutamate levels by balancing glial and synaptic glutamate release and reuptake. Preclinical evidence indicates that repeated exposure to cocaine and AMP derivatives disrupts glutamate homeostasis and increases drug-seeking (Kalivas 2009). Interestingly, despite no differences in baseline glutamate levels in the nucleus accumbens (NAc) between EC and IC rats, a subcutaneous injection of AMP evokes more glutamate efflux in the nucleus accumbens (NAc) of EC rats when compared to IC rats. (Rahman and Bardo 2008). While not fully elucidated, mGluR2/3 signaling likely contributes to the difference in glutamate efflux. MGluR2/3s are located on the presynaptic neuron and are coupled to Gi/o which inhibits 3’,5’-cyclic adenosine monophosphate (cAMP) formation. Thus, stimulating mGluR2/3 inhibits the presynaptic neuron, and represents the putative mechanism that regulates vesicular and nonvesicular glutamate release in the NAc (Manzoni, Michel et al. 1997). Interestingly, mGluR2/3 expression in the medial prefrontal cortex (mPFC) and striatum are similar between EC and IC rats. However, selective antagonists aimed at the mGluR2/3 elevates extracellular glutamate in EC and SC rats, but not in IC rats (Heidbreder, Weiss et al. 2000, Melendez, Gregory et al. 2004, Rahman and Bardo 2008), suggesting that the deficiency in glutamatergic tone in IC rats is mediated by mGluR2/3 transmission. Behaviorally, systemic administration of a selective mGluR2/3 agonist (LY379268) reduces glutamate overflow and suppresses the development of AMP-induced locomotor sensitization (Kim, Austin et al. 2005). However, the same agonist attenuates AMP-induced sensitization in IC and SC rats, but not in EC rats (Arndt, Arnold et al. 2014), indicating that environmental variables influence the potency of stimulant-induced glutamate efflux. Relatedly, systemic LY379268 administration reduces psychostimulant SA and cue-induced reinstatement (Baptista, Martin-Fardon et al. 2004, Adewale, Platt et al. 2006, Hao, Martin-Fardon et al. 2010, Jin, Semenova et al. 2010), giving rise to the hypothesis that mGluR2/3 regulates psychostimulant reinforcement. With these results in mind, we hypothesized that pretreatment with LY379268 will dose-dependently suppress AMP SA motivation in EC, IC and SC rats such that the lower dose of LY379268 will suppress AMP SA motivation in EC but not IC or SC rats.

Ironically, the robust and sweeping behavioral effects caused by differential rearing make it challenging to elucidate causal relationships between brain and behavior. EC rearing reduces baseline motivation for numerous reinforcers, including AMP, when compared to IC and SC rats (Bardo and Dwoskin 2004, Stairs and Bardo 2009, Arndt, Johns et al. 2015). Therefore, the potency and efficacy of novel therapeutics can be mischaracterized because baseline differences exist between EC, IC and SC rats in the absence of any pharmacological treatment. Determining the relative contribution of rearing environments holds the potential to understand critical developmental windows that can potentiate or suppress promising therapeutics and uncover critical relationships that lead to a trajectory of SUDs. Revealing these neurobiological underpinnings exposes druggable targets to combat the development of SUDs. An additional purpose of this manuscript is to determine the effects of normalizing response variables to ‘control’ for baselines differences when interpreting the effects of LY379268 pretreatment on PR AMP SA in differentially reared rats.

2. Methods and Materials

2.1. Animals

Twenty-eight male Sprague-Dawley rats (Charles River, Portage, MI, USA) arrived at exactly 21 days of age and were randomly assigned to one of three environmental conditions: enriched (EC), isolated (IC), or standard (SC) using methods described previously (Arndt, Wukitsch et al. 2019, Garcia, Arndt et al. 2019). Rats had ad libitum access to food and water throughout the experiment, with the exception of lever press training (Section 2.5.1). The colony room operated on a 12-hr light-dark cycle and was maintained at approximately 22° C, with humidity ranging from 30–45%. All behavioral tests were conducted during the light portion of the cycle. All experimental procedures were in compliance with the Institutional Animal Care and Use Committee at Kansas State University and complied with NIH guidelines (National Research Council, 2011).

2.2. Environmental Conditions

EC rats were housed with 8–12 other cage mates in a large metal cage (60 × 120 × 45 cm) that was lined with pine wood chip bedding and were handled daily. The EC was created by arranging 14 objects (small children’s toys and PVC pipe) daily. To maintain novelty, seven of the objects were changed daily, and all objects were changed twice weekly. IC rats were reared individually and were housed in hanging wire cages (17 × 24 × 20 cm). The IC rats were not handled throughout the rearing process and were not exposed to novel objects or bedding. The SC rats were housed in pairs in standard shoebox cages (20 × 43 × 20 cm) lined with pine wood chip bedding. SC rats were not supplemented with novel objects and were only handled during the weekly cage change. SC rats were included not as a control for differences between EC and IC rats, but rather to provide us with a known laboratory standard. Rats remained in their respective environmental conditions for the entire experimental period.

2.3. Apparatus: Operant Chamber

Lever press training and AMP SA were conducted in standard operant conditioning chambers (ENV-001, Med Associates, St. Albans, VT). Each chamber was enclosed in a sound-attenuating compartment and operated by a computer interface using MedPC software. During lever press training, sucrose was delivered with a liquid dispenser in a recessed food receptacle. Stainless steel operant response levers were located on each side of the food receptacle 7.3 cm above the metal grid floor. A 28-V, 3-cm diameter white cue light was centered above each response lever. AMP was infused with a mechanical syringe pump (PHM-100, Med Associates) (Arndt, Johns et al. 2015, Arndt, Wukitsch et al. 2019, Garcia, Arndt et al. 2019).

2.4. Drugs: Self-Administration (SA) Testing

Selective mGluR2/3 agonist, LY379268 (Abcam Biochemicals, MA, USA) was dissolved in 0.9% saline and the pH was adjusted to 7.4±0.05 (0.3 and 1.0 mg/kg; 1.0 mg/ml). LY379268 was stored at −20 °C until use. On test days, LY379268 was thawed and injected intraperitoneally (i.p.) 30 minutes prior to the start of AMP SA sessions.

D-amphetamine (Sigma Aldrich, MO, USA) was dissolved in 0.9% saline (0.1 mg/kg/infusion) and was self-administered intravenously.

2.5. Behavioral Procedures

2.5.1. Lever Press Training

At 52 days of age, rats were food restricted to 85% of their free-feeding weights. Then, rats were familiarized with the operant chamber, the sound of the liquid dipper, and sucrose presentation. The next day, the rats were trained on a single lever to press for a 0.1 ml delivery of 20% sucrose solution (dissolved in distilled water) on a fixed-ratio 1 (FR-1) schedule of reinforcement. During the next four sessions, both the active and inactive levers were present. Responses (FR-1) on the active lever delivered 0.1 ml of sucrose and responses on the inactive lever had no programmed consequence. Both active and inactive lever presses were recorded. The same active lever (left or right) was maintained for each rat for both the sucrose training and AMP SA phases of the experiment (Arndt, Johns et al. 2015, Arndt, Wukitsch et al. 2019, Garcia, Arndt et al. 2019)

Following lever press acquisition, rats were returned to free-feeding weights for the remainder of the experiment. After returning to free-feeding weights (~4 days), rats were deeply anesthetized with ketamine (80 mg/kg; 1 mg/ml, i.p.) and diazepam (5 mg/kg; 1 mg/ml, i.p.) prior to jugular catheter implantation. Polyurethane catheters (~12 cm. in length, 0.2 mm internal diameter SAI Infusion Technologies) were inserted through a dorsal incision. The catheter tubing was tunneled under the skin and inserted into the rat’s left jugular vein. Catheter tubing from the jugular vein was connected subcutaneously to a 22 gauge back-mounted cannula (Plastics One; Roanoke, VA) secured and sutured to surgical mesh (Biomedical Structures; Warwick, RI). A stainless-steel bolt covered the catheter cannula cap to prevent damage to the back mount. To maintain catheter patency and to protect against infection, catheters were flushed daily with heparinized saline (10–30 IU/ml; 0.1 ml before SA and 0.1 ml after SA sessions) and cefazolin (50 mg/ml; 0.1 ml intravenous (i.v.) infusion). EC rats were held in shoebox cages with one other EC rat and a novel object for one night following surgery to facilitate surgery recovery while maintaining the enriched environment,

2.5.2. Amphetamine Self-Administration (AMP SA)

Rats recovered from surgery over 9–12 days. Rats were then returned to the operant boxes and allowed to self-administer AMP on an FR-1 schedule of reinforcement daily. AMP SA sessions were 60 minutes. Responses on the active lever infused 100 μl of 0.1 mg/kg/infusion of AMP over 5.9 seconds. To maintain a constant infusion volume and time, various concentrations of AMP were prepared and rats self-administered the concentration specific to its daily weight. After each AMP infusion, a 20-sec timeout period was signaled by the illumination of both cue lights. Active lever presses during the time out period were recorded but not reinforced with AMP infusions. Inactive lever presses were recorded but had no programmed consequence. Rats remained on an FR-1 schedule until stable responding was achieved. Stable responding was defined as: (1) 30% or less variability in active lever presses across three sessions; (2) greater than a 2:1 ratio of active:inactive lever presses across the three sessions; (3) at least ten infusions per session. After the rats met the criteria for stable responding, the effects of LY379268 (the same doses used during PR testing described below) on FR-1 responding were evaluated. These results are not reported in the current manuscript.

2.5.3. Progressive-Ratio (PR) Testing

After the FR phase, rats were trained to stability on a PR schedule of reinforcement. Due to the low response rate of the EC rats (Arndt, Johns et al. 2015), the PR session was 60 minutes long and terminated regardless of lever press activity (Corrigall, Coen et al. 2001, Ross, Corrigall et al. 2007, Garcia, Le et al. 2014, Neugebauer, Cortright et al. 2014). When response rates are low, this PR design reduces the propensity for EC rats to extinguish within a session and maintains prolonged response rates across days. Under this PR schedule, the number of responses required to earn each successive AMP infusion was determined by Response Ratio (rounded nearest integer) = [5 × e(0.25 × infusion number) − 5] to produce the following sequence of required lever presses: 1, 3, 6, 9, 12, 17, 24, 32, 42, 56, 73, etc. (Richardson and Roberts 1996). On test days, rats were injected 30 minutes prior with LY379268 (Veh, 0.3, or 1.0 mg/kg; i.p.) using a counterbalanced design. Rats had a minimum of two AMP SA sessions without LY379268 to allow for the washout of any drug effect. Following the last SA session, catheter patency was verified by infusing Brevital (10 mg/ml; 0.1–0.15 ml, i.v.). Rats that failed the catheter patency checks were excluded from the reported analyses. After patency checks, 8 EC rats, 10 IC rats, and 10 SC rats had patent catheters and were included in analyses.

2.6. Statistical Analysis

The focus on the current experiment was to determine the extent to which mGluR2/3 activation moderates AMP SA on a PR schedule of reinforcement in EC, IC, and SC rats. Linear mixed effects were used to analyze SA data using the freely available statistical environment R and the R statistical software packages: lme4 (Bates, Mächler et al. 2015), lmerTest (Kuznetsova, Brockhoff et al. 2017), psycho (Makowski 2018), tidyverse (Wickham, 2019, https://CRAN.R-project.org/packages=tidyverse), and dabestr (Ho, Tumkaya et al. 2019). Linear or nonlinear mixed effects models are the recommended analytical strategy for repeated measures designs in psychology and neuroscience (Young, Clark et al. 2009, Boisgontier and Cheval 2016, Young 2016). LY379268 pretreatment (Veh, 0.3, 1 mg/kg, i.p.) and rearing group (EC, IC, SC) were treated as fixed effects. The intercept and individual rats were defined as random effects. Significance was determined using the lmerTest package which utilizes the Satterthwaite’s method to estimate the degrees of freedom (Type III sums of squares for t-test comparisons between fixed effects). The Satterthwaite’s method is built into the lmerTest package to determine the statistical significance of fixed and random effects, which had been a cumbersome deterrent for researchers making the transition to R and mixed effects analysis. Model fits were completed with restricted maximum likelihood estimation to allow for comparisons between models during model fit optimization.

For normalization analysis and comparisons, the dependent variables were transformed with a ‘group’ normalization and an ‘individual’ normalization transformation to determine the effects of different normalization strategies on effect size estimates and observed confidence interval range. ‘Group’ normalization was calculated as a percent change from the vehicle. First, the dependent variable of interest (active, inactive, breakpoint, or infusions) within an environmental group was averaged after pretreatment with vehicle or LY379268. Then the ratio to the group averages was calculated and multiplied by 100%.

The ‘individual’ normalization was calculated as a percent change from each animal’s response after pretreatment with the vehicle. The dependent variables were calculated as a ratio of each animal’s vehicle response and multiplied by 100%. The normalization strategy removes individual baselines and the effects of LY379268 are a percent change from each animal’s individual response.

Cumming and/or Gardner-Altman plots called ‘estimation plots’ moving forward were created with the R package dabestr (Ho, Tumkaya et al. 2019). Estimation graphics offer several advantages including a representation of the full distribution of the data, effect size comparison and the error in the effect estimate. A complete description including the advantages and disadvantages of estimation graphics and resources available to use estimation analysis can be found in these resources (Calin-Jageman and Cumming 2019, Ho, Tumkaya et al. 2019). Briefly, a bootstrap resampling strategy resamples each experimental condition with replacement and calculates a difference score or effect size between the experimental conditions (n = 5,000). The distribution of all the calculated effect sizes can then be used to calculate the 95% CI of the estimated effect size (Ho, Tumkaya et al. 2019). All R code, data, graphs, and summaries are freely available in a Bitbucket repository (erikgarcia22/ec-ic-baseline-pub.git)and available upon request.

3. Results

3.1. AMP SA (0.1 mg/kg/infusion) in EC, IC, and SC rats

Differentially reared rats acquired AMP (0.1 mg/kg/inf) SA until stable. We and others have previously reported that EC, IC, and SC self-administer similar amounts of AMP when a high dose of AMP is used because EC rearing is most protective at low-unit doses of psychomotor stimulants (Arndt, Johns et al. 2015, Arndt, Wukitsch et al. 2019, Garcia, Arndt et al. 2019). Importantly, because the purpose of this paper is to examine the effects of differential rearing on PR schedules of reinforcement on AMP SA and analytical strategies, we focus the remaining results sections on the effects of LY379268 pretreatment on AMP SA in a PR schedule of reinforcement behavioral procedure.

3.2. EC Rearing Reduces the Motivation for AMP SA (0.1 mg/kg/inf)

For all analyses, we employed the lowest ordered random intercept model with restricted maximum likelihood estimation to determine the effects of rearing group and LY379268 pretreatment on the motivation to earn AMP reinforcers. Beyond this, similarly structured linear mixed effect models were used when the data were transformed with ‘group’ and ‘individual’ normalizing transformations to determine the effects of different strategies on the fixed effects, model fit, effect size estimates, and error using the estimation’ plots.

Linear mixed effects analyses determined that the rearing group significantly altered the number of total active lever presses during the AMP SA PR sessions (F(2, 28.19) = 4.44, p = 0.021). When compared to EC rats, rearing in the IC (b = 37.13, p = 0.045) or SC (b = 56.73, p = 0.003) increased active lever presses. Pretreatment with LY379268, a mGluR2/3 agonist, significantly reduced the number of active lever presses for AMP (F(2, 55.35) = 4.35, p = 0.018). The omnibus model test did not reveal evidence for a significant interaction between the rearing group and LY379269 (F(4, 55.33) = 1.54, p = 0.204). Figure 1A shows the estimation plot for active lever presses for each group (EC, IC, and SC) as a function of LY379268 treatment (Veh, 0.3, and 1 mg/kg; i.p.). The top panel shows the full distribution of the dependent variable (active lever presses) for each respective experimental group. The bottom panel is the repeatedly calculated average difference (large black circle) using bootstrap resampling methods and the corresponding distribution of all difference calculations (shaded gray distribution, n = 5,000 resamples). The error bars surrounding the average effect size represent the 95% bootstrapped confidence intervals. Errors bars overlapping the zero-difference line indicate that ‘no effect/difference’ is within this range and therefore, interpreting differences between groups should be refrained. Bootstrapped resampling confirmed that neither dose of LY379268 (0.3 or 1 mg/kg) significantly altered active lever presses in EC or IC rats. However, when compared to the vehicle, the 1 mg/kg dose significantly reduced active lever presses in SC rats (Δ = −31.8, 95%CI [−58; −8.7]).

Figure 1A.

Figure 1A.

Figure 1A.

Open in a new tab

EC rearing reduces total active lever presses for AMP (0.1 mg/kg/inf) SA on a PR schedule independent of mGluR2/3 activation. Estimation plots for (Left) EC (Middle) IC or (Right) SC rats reveal that pretreatment with LY379268 (0.3 mg/kg) does not affect the average number of active lever presses when compared to vehicle. However, pretreatment with 1 mg/kg of LY379268 reduced active lever presses compared to vehicle and effect driven by SC rats despite a non-significant interaction. The estimation plots illustrate two critical aspects of statistical inference: effect size and confidence in the estimate of the effect size. In the top panel of each estimation plot are the observed individual rat data (red, blue, green-filled circles), and the black vertical lines represent the 95% confidence interval for the observed data. In each of the bottom panels, are the randomized bootstrapped resampled distribution of the effect sizes for each dose of LY379268 compared to the vehicle. The shaded distribution illustrates the complete resampling distribution (n = 5000), the large black circle represents the average difference between LY379268 doses and vehicle for each rearing group, the error bars on the large black circle indicate the 95% confidence interval for the calculated difference. The 95% confidence interval of the difference between groups indicates the range of possible values given the observed data. When the average difference and 95% confidence intervals (large black circles with error bars) cross the horizontal line at zero, an effect size equal to zero is possible, and therefore, a reliably measured difference is unlikely. Progressive ratio (PR); self-administration (SA) Enriched condition (EC), Isolated condition (IC), Standard condition (SC). Figure 1B. The observed breakpoint for the progressive ratio (PR) AMP SA assay. Estimation plots for (Left) EC (Middle) IC or (Right) SC rats reveal that EC rats have reduced breakpoints when compared to IC and SC rats. Pretreatment with LY379268 (0.3 or 1 mg/kg) did not reduce the average breakpoint when compared to the vehicle.

Analysis for inactive lever presses revealed a main effect of rearing group, (F(2, 28.26) = 5.25, p = 0.011). The difference in inactive lever presses was driven by an increase in SC rats (b = 11.25, p = 0.040) but not IC rats (b = −1.25, p = 0.816) when compared to EC rats. Pretreatment with LY379268 did not affect inactive lever presses (F(2, 55.51) = 0.07, p = 0.931) and there was no group × treatment interaction (F(4, 55.48) = 0.18, p = 0.968) Bootstrapped estimation revealed that LY379238 (1 mg/kg) reduced inactive lever presses in EC rats when compared to vehicle (Δ = −2, 95%CI [−6.86; −0.143]). LY379269 did not affect IC or SC inactive lever presses (data not shown).

Rearing group significantly changed the number of AMP infusions earned (F(2, 28.26) = 6.07, p = 0.006). Compared to EC vehicle treated rats, IC (b = 1.93, p = 0.074) and SC (b = 2.63, p = 0.016) rats earned more AMP infusions. Treatment with LY379268 did not significantly alter the number of AMP infusions (F(2, 55.48) = 2.94, p = 0.061) and there was also no interaction between rearing group and treatment, (F(4, 55.45) = 1.82, p = 0.139), suggesting that activating mGluR2/3 does not suppress AMP intake in IC or SC rats. However, the bootstrap estimation analysis revealed that the lower dose of LY379268 suppressed AMP infusions in EC rats compared to vehicle (Δ = −1.86, 95%CI [−4.57; −0.143]), while a higher dose (1 mg/kg) was required to suppress AMP infusions in SC rats (Δ = −1.2, 95%CI [−2.6; −0.2]). Neither dose of LY379268 was capable of suppressing AMP infusions in IC rats.

In PR schedules of reinforcement, the observed breakpoint represents the threshold point where the response requirement exceeds the level of reinforcement and therefore, animals cease responding and do not earn subsequent reinforcers. Analysis of breakpoints determined that rearing group, (F(2, 28.22) = 4.53, p = 0.020) and treatment (F(2, 55.40) = 3.22, p = 0.048) significantly altered breakpoints, but there was no evidence of a group × treatment interaction, (F(4, 55.38) = 1.29, p = 0.285). Figure 1B shows the estimation plots for the breakpoint data. These plots demonstrate that the 95% confidence intervals for the average difference between LY379268 (0.3 or 1 mg/kg) and vehicle include the zero-point difference, suggesting that LY379268 does not alter the breakpoint for AMP reinforcers in EC, IC, or SC rats.

3.3. Group Average Normalization Removes Group Level Effects

Normalizing data to 100% is a common analytical approach to determine if an intervention caused a deviation from the normalized baseline. In this next series of analyses, we normalized the variables of interest to 100% for each rearing group following pretreatment with the vehicle. This transformation allowed for individual variability around 100% vehicle (see Figure 2A or 2B vehicle conditions). Expectedly, because all rearing groups are normalized to 100%, the linear mixed effects analysis determined that there was no effect of the rearing group, (F(2, 28.17) = 0.08, p = 0.920). Furthermore, there was no effect of treatment (F(2, 55.31) = 1.85, p = 0.167) or interaction of rearing group × treatment, (F(4, 55.31) = 0.95, p = 0.443) on group normalized active lever presses (Figure 2A). Bootstrapped estimation methods determined that when the data are normalized, LY379268 does not significantly alter group average normalized active lever presses in EC or IC rats. However, the 1 mg/kg dose of LY379268 reduced group normalized active lever presses compared to vehicle (Δ = −34.1%, 95%CI [−62.3; −9.01]) in SC rats.

Figure 2A.

Figure 2A.

Open in a new tab

Group normalized active lever presses for AMP (0.1 mg/kg/inf) SA on a PR schedule. All rats were normalized to their respective rearing group’s average vehicle response. Estimation plots for (Left) EC (Middle) IC or (Right) SC rats reveal baseline differences between rearing groups are ‘masked’ by a group normalization strategy. Rearing environment does not affect group normalized active lever presses. Pretreatment with LY379268 does not affect group normalized active lever presses when compared to vehicle (100%) in EC, IC, or SC rats. Figure 2B. Group normalized breakpoints in a PR AMP SA assay. All rats were normalized to their respective rearing group’s average vehicle response. Estimation plots for (Left) EC (Middle) IC or (Right) SC rats reveal the rearing environment does not affect group normalized breakpoints. Pretreatment with LY379268 does not affect group normalized breakpoints when compared to the vehicle in EC, IC, or SC rats.

Similarly, group normalized inactive lever presses were not changed by rearing group (F(2, 28.21) = 0.51, p = 0.608), treatment (F(2, 55.53) = 0.02, p = 0.975), or rearing group × treatment interaction (F(4, 55.49) = 0.37, p = 0.827). Bootstrap estimation confirmed that LY379268 did not affect group normalized inactive lever presses in EC, IC, or SC rats (all 95% CI include 0). Taken together, these results indicate that normalizing the data to each respective groups’ 100% average can mitigate potential differences between rearing groups and suppressed the once significant treatment effect of LY379268 on active and inactive lever presses when using linear mixed effects analysis. While the bootstrapped estimation methods generally matched the untransformed data, the confidence intervals are considerably wider than the untransformed confidence interval ranges suggesting that more error is introduced in the analysis.

The random intercept linear mixed effects analysis with rearing group, treatment and the interaction was used to analyze the group normalized infusions and breakpoint data. Analysis determined that rearing group (F(2, 28.24) = 0.20, p = 0.823), treatment (F(2, 55.45) = 2.25, p = 0.115), and the rearing group × treatment interaction, (F(4, 55.42) = 1.87, p = 0.128) did not significantly alter group normalized AMP infusions. Bootstrap estimation analysis determined that 0.3 mg/kg (Δ = −34.6%, 95%CI [−82.1; −3]), but not 1 mg/kg suppressed group normalized infusions in EC rats (Δ = −13.4%, 95%CI [−42.6; 8]). Alternatively, 1 mg/kg LY379268 suppressed group normalized AMP infusions in SC rats (Δ = −15%, 95%CI [−32.6; −1.3]), while 0.3 mg/kg was without effect (Δ = 3.9%, 95%CI [−7.5; −17.8]). Interestingly, neither dose of LY379268 significantly changed group normalized AMP infusions in IC rats.

The random intercept linear mixed effect analysis determined that there were no differences between group (F(2, 28.21) = 0.002, p = 0.998), treatment (F(2, 55.38) = 1.93, p = 0.155), and no group × treatment interaction (F(4, 55.35) = 1.18, p = 0.328), suggesting that none of the fixed effects significantly altered breakpoints during AMP SA. Bootstrap estimation analysis determined that neither dose of LY379268 affected group normalized breakpoints in EC, IC, or SC rats (all 95% Ci include zero effect). These results reiterate that the group normalization strategy employed expands the confidence intervals and eliminates the main effects of the linear mixed effects analysis. However, the estimation plots method closely mimicked the results from the untransformed breakpoint data (Section 3.2).

3.4. Individual Normalization Reduces Precision of Fixed Effect Estimates

In this series, the dependent variables were normalized to the vehicle treatment for each individual rat. Thus, each rat served as its own 100% baseline. A similarly structured random intercept linear mixed effects analysis was used to analyze the effects of group, treatment and the interaction on ‘individual normalized’ responses (active and, inactive lever presses, infusions, and breakpoints). When data are normalized with this strategy every rat is positioned at 100% and creates a distribution with no variability (Figures 3A and 3B vehicle groups). Any deviation from this baseline is analyzed as an effect attributed to the independent variable regardless of the independent variable true effect size. Therefore, large variability is expected following treatment with LY379268 because individual differences are greatly influential and confounded with treatment effects estimates.

Figure 3A.

Figure 3A.

Open in a new tab

Individual normalized active lever presses for AMP (0.1 mg/kg/inf) SA on a PR schedule. All rats were normalized to their own vehicle response. Estimation plots for (Left) EC (Middle) IC or (Right) SC rats reveal reduced variability in vehicle treatment and ‘masked’ baseline differences across rearing groups by an individual normalization strategy. Even more, the effect size confidence intervals are exaggerated, indicating a high degree of introduced error. Rearing environment does not affect individual normalized active lever presses. Pretreatment with LY379268 does not affect individual normalized active lever presses when compared to vehicle (100%) in EC, IC, or SC rats. Figure 2B. Individually normalized breakpoints in a PR AMP SA assay. All rats were normalized to their vehicle response. Estimation plots for (Left) EC (Middle) IC or (Right) SC rats reveal the rearing environment does not affect individually normalized breakpoints. Pretreatment with LY379268 does not affect group normalized breakpoints when compared to vehicles in EC, IC, or SC rats.

Linear mixed effects analysis determined that there were no main effects of group (F(2, 28.21) = 0.71, p = 0.502), treatment (F(2, 55.51) = 0.40, p = 0.672), or the group × treatment interaction (F(4, 55.47) = 0.68, p = 0.612) on individual normalized active lever presses. Bootstrapped estimation analysis and plots illustrate that neither dose of LY379268 significantly altered individual normalized active lever presses when compared to the vehicle for EC, IC, or SC rats. All estimates or 95% CI include the value of 0, suggesting that no effect is within the range of the CI and differences between the comparison groups are unlikely (Figure 3A).

Inactive lever presses revealed a similar trend. The analysis determined that there were no main effects of group (F(2, 29.21) = 0.006, p = 0.994), treatment (F(2, 52.91) = 2.44, p = 0.097), or a group × treatment interaction (F(4, 52.86) = 0.47, p = 0.758) on individual normalized inactive lever presses. Bootstrapped estimation analyses determined that neither dose of LY379268 changed individualized inactive lever presses in any of the rearing groups.

Similarly, analysis determined that there were no main effects of group (F(2, 27.92) = 1.06, p = 0.361), treatment (F(2, 55.26) = 0.55, p = 0.579), or the group × treatment interaction (F(4, 55.22) = 1.41, p = 0.242) on individual normalized AMP infusions. Bootstrapped estimation analyses determined that neither dose of LY379268 changed individualized infusions SC rats. However, 0.3 mg/kg of LY379268 decreased individual normalized AMP infusions in in EC (Δ = −42.1%, 95%CI [−77.4; −0.857]) and increased them in IC rats (Δ = 17.9%, 95%CI [0.092; 68.5]). The 1 mg/kg dose of LY379268 did not affect individual normalized AMP infusions in EC, IC, or SC rats.

Analysis of individual normalized breakpoints determined that there were no main effects of group (F(2, 28.13) = 0.66, p = 0.524), treatment (F(2, 55.45) = 0.27, p = 0.761), or the group × treatment interaction (F(4, 55.41) = 0.71, p = 0.587). Bootstrapped estimation analyses determined that neither dose of LY379268 changed individually normalized breakpoints in EC or SC rats. However, 0.3 mg/kg of LY379268 increased individual normalized breakpoints in IC rats (Δ = 40.8%, 95%CI [0.2; 160]) while the 1 mg/kg dose did not affect IC rats (Δ = 31.7%, 95%CI [−44.8; 287]). Because the data are normalized to each rat, the distributions of the data are skewed with a single large outlier. Even with random resampling from these otherwise typical distributions, the outlier significantly impacts the data (Figure 3B IC distribution for 0.3 and 1 mg/kg).

3.4. The Precision of Effect Size Estimates is Reduced with Normalizing Transformations.

Although the Akaike Information Criterion (AIC) or the Bayesian Information Criterion (BIC) are not able to be directly compared across untransformed, and group/individual normalization because the response variables are different, Supplemental Table 1 provides the summary AIC and BIC values. Another way to visualize the impact of normalizing on the precision of estimating the effect is summarized in Figures 4A and 4B. These figures illustrate that the confidence interval ranges extracted from the bootstrap resamples used to create the confidence intervals for all the estimation plots. These figures illustrate that the 95% confidence interval gets considerably wider after the observed data are normalized using a group and individual normalizing transformations. Collectively, these data provide strong evidence that normalizing transformation can introduce error into the analysis. Therefore, to maintain the 95% confidence interval the interval range becomes wider.

Figure 4A.

Figure 4A.

Open in a new tab

The 95% CI range for the average difference between LY379268 (0.3 or 1 mg/kg) and vehicle as a function of rearing group and normalization strategy. 4A. Active Lever presses or 4B. Breakpoints Group and individual normalization strategies broaden the 95% confidence interval; indicative of introduced error. Blue denotes LY379268 0.3 mg/kg vs vehicle and green denotes LY379268 1 mg/kg vs vehicle. Solid blue and green bars are the untransformed data, checkered blue and green bars are the group normalized data, and slashed blue and green bars are the individual normalized data.

4. Discussion

The present experiments and analysis determined that EC rearing is a powerful preclinical manipulation to reduce the motivation for AMP reinforcers. Despite equal AMP SA intake during maintenance between rearing conditions, the motivation to respond for AMP reinforcers is reduced in EC rats. This novel result suggests that EC, IC, and SC rats have a quantifiably different motivation for earning AMP reinforcers. Thus, solely measuring drug intake is not sufficient to determine the sweeping changes caused by differences in early life rearing and housing. Overall, transforming the dependent variables with different normalization strategies altered the interpretation. In all mixed effects analyses, evidence of a significant group by treatment interaction was never observed. However, when the estimation plots were used to compare the doses of LY379268 to the vehicle within each rearing group they revealed a dose-related effect in EC and SC rats. In EC rats, the 0.3 mg/kg but not the 1 mg/kg dose suppressed AMP infusions. Alternatively, in SC rats the 0.3 mg/kg dose was ineffective, while the 1 mg/kg dose suppressed AMP infusions, while in IC rats neither dose of LY379268 suppressed AMP infusions. Finally, commonly employed normalization strategies to control for baseline differences between EC, IC, and SC rats can inflate the range of a bootstrapped 95% confidence interval, indicating more error is introduced into statistical analysis and inference.

EC rearing significantly reduces AMP and cocaine SA at low-unit doses (Green, Gehrke et al. 2002, Bardo and Dwoskin 2004, Arndt, Johns et al. 2015), which gives rise to the criticism that EC rearing has limited efficacy and translational relevance, especially when considering higher doses of psychostimulants or long access drug SA models. In opposition to that criticism, the current results and those presented by Yates et al. suggest that despite relatively equal drug self-administration history and overall AMP intake, the protective effect of enrichment remains intact (Yates, Bardo et al. 2019). The EC effect simply requires a more expansive evaluation of reinforcer cost. When the cost (responses/mg) is low, EC and IC rats will consume similar amounts of the drug. However, when the cost of psychostimulant infusions increases, EC rats are more sensitive to price changes and cease psychostimulant SA (Yates, Bardo et al. 2019). Our current results are in agreement with this aforementioned explanation of AMP SA differences between rearing groups, because as the response requirement increased in the PR schedule of reinforcement, the EC rats consistently ceased AMP before IC or SC rats, suggesting that they were more sensitive to price changes and that EC rats remain more flexible in AMP SA when compared to IC or SC rats. Therefore, the reinforcing value of a high dose of AMP is blunted by EC rearing.

The current results indicate a dynamic role for mGluR2/3 agonists to suppress psychostimulant SA. Long, but not short access cocaine SA disrupts mGluR2/3 function because LY379268 decreases cocaine SA motivation only after long access cocaine SA sessions (Hao, Martin-Fardon et al. 2010). AMP and methamphetamine elevate glutamate release in the prefrontal cortex (PFC) and NAc (Shoblock, Sullivan et al. 2003). Interestingly, LY379268 was equally effective at reducing methamphetamine SA motivation in short and long access SA models, even when animals had escalated methamphetamine intake (Crawford, Roberts et al. 2013), suggesting that escalated drug consumption is not necessary for mGluR2/3-mediated disruptions in glutamate homeostasis. Given that LY379268 suppressed AMP infusions using a short access SA model, this suggests that AMP and methamphetamine deviate from cocaine in how each of the stimulants affects mGluR2/3 function and expression (Crawford, Roberts et al. 2013). Therefore, AMP may upregulate mGluR2/3 function and/or expression because the efficacy of LY379268 requires mGluR2/3: Gi/o coupling (Hao, Martin-Fardon et al. 2010). LY379268 is a highly selective mGluR2/3 agonist with greater than 80-fold selectivity over other mGluRs (Monn, Valli et al. 1999), suggesting that at the doses tested it is unlikely that neurotransmission action at other mGluRs was responsible for the suppression of AMP SA.

Rearing in the IC condition blocked the efficacy of LY379268 to suppress AMP SA in a PR schedule of reinforcement. While the IC animals did not demonstrate the highest level of baseline motivation during PR sessions, none of the doses of LY379268 tested significantly reduced active lever presses, AMP infusions earned, or breakpoint. These results suggest that rearing in the IC condition creates a vulnerable phenotype by altering the functional capability of the mGluR2/3 receptor because LY379268 was unable to block AMP SA in this group. Furthermore, it indicates that the mGluR2/3 does not fully mediate AMP reinforcement under a PR schedule of reinforcement. Whether the mGluR2/3 is significantly altered before or after AMP SA is not determined in the current experiments. However, given the observed potency differences between EC, IC, and SC rats with LY379268 it does suggest that early life environmental variables impact the dynamics of the mGluR2/3. This result is in agreement with other results suggesting that blocking the mGluR2/3 is incapable of augmenting glutamate release in IC rats (Melendez, Gregory et al. 2004) and this dysfunction is at least one putative mechanism by which IC rats are a vulnerable phenotype in preclinical models of psychostimulant use disorders.

A daunting challenge for differential rearing experiments becomes apparent during analysis and interpretation because of differences in baseline behaviors. Analytical methods to address baseline differences in motivated behaviors are well described (Jamieson 1999), yet an accepted analytical approach remains elusive within the context of differential rearing. Psychologists and neuroscientists commonly normalize to 100% to ‘remove’ baseline differences and then compare the experimental condition to the 100% normalized baseline to determine if the intervention altered the dependent variable. However, there are numerous statistical transformations to achieve a ‘100% baseline’. Intuitively, normalizing to 100% can simplify the interpretations. For example, a 25% reduction is readily accessible and is not as cumbersome as interpreting other statistical test parameters such as regression coefficients and eta2 (ANOVA effect sizes). However, the impact broadly applied normalizations can be overlooked. How these oversights strongly influence data structures are best observed in the ‘group’ and ‘individual’ normalized inactive lever press data. The sample size is comparable to a typical sample in behavioral neuroscience research (n = 10). However, when the actual untransformed values are low, even small changes from baseline appear to be large differences when the values are normalized to 100%. Even more worrisome, going from 1 to 8 lever presses creates an inflated 800% increase which significantly skews the effect size estimate in the direction of the outlier (IC rats in Figure 3A/3B). Thus, even a randomized, robust, and nonparametric method to estimate average differences is subject to significant deviations. More importantly, there is an inherent cost to normalizing to percent change which is that the normalization masks the original data making it difficult to reproduce. A 25% reduction could similarly suggest a decrease from 100 to 75 or a reduction from 4 to 3? Despite the percent (%) magnitudes of change being equal, there is more confidence in the former. Without the original data, a researcher never knows typical values or how an intervention changes them which leads to Type I and Type II errors. Therefore, the original individual data points, their distributions and effect size estimate and precision should be included and is paramount for deciphering the contribution that EC rearing has on motivated behavior when compared to IC or SC rats.

Estimation plots offer the ability to quickly see the whole distribution of the original observed data (upper panels of figures) to determine the spread or error in measurement within each experimental group. Second, within the same figure (lower panels), estimation plots show the effect size and the error/precision in estimating the effect size given the observed data. In short, how much and how confident, respectively. Even when the original observed data are not normally distributed, the repeated resampling bootstrap method will come to approximate a normal distribution of calculated effect sizes (experiment - control) based on principles of the Central Limit Theorem. If the measurement error is small in the original data, then the 95% CI will be narrow. When measurement error infiltrates the data, the 95% CI expands because the calculated effect size is large is some resamples and small in other resamples. Therefore, when the 95% CI is estimated, there are many small, medium, and large effects, causing the precision of the estimate and the confidence in that estimate to suffer.

Estimation methods question the continued use of significance testing (Calin-Jageman and Cumming 2019), but at this time, eliminating significance testing goes beyond the scope of the current manuscript which led us to include the linear mixed effect analysis as a robust analytical strategy to supplement the estimation plots (Bates, Mächler et al. 2015). Mixed effects analysis is becoming increasingly more prevalent in psychological and neuroscience research (Young, Clark et al. 2009, Boisgontier and Cheval 2016, Garcia, Jorgensen et al. 2017, Wang, Marshall et al. 2017, Harrison, Donaldson et al. 2018, Yates, Bardo et al. 2019). Recommendations about fitting more complex hierarchical multilevel models deserves its own discussion (Barr, Levy et al. 2013). We chose a simple, low-level random intercept model for all analyses, which can oversimplify and overlook more complex and better fitting models, especially when distributions do not meet the assumptions of a linear trajectory (Babyak 2004, Barr, Levy et al. 2013, Young 2016). With this in mind, we examined all fitted vs predicted quantile-quantile plots and model fit residuals and determined they were normally distributed and there was no distinctive pattern for the residuals, suggesting that the random intercept linear model fit accommodated the untransformed and transformed data without significantly violating model fit assumptions (all plots available in repository). Largely, the interpretation of estimation methods and the mixed model analyses were in accord. The benefit added with the estimation method is removing the dichotomous p-value and supplementing it with “how much and how confident?”(Calin-Jageman and Cumming 2019, Ho, Tumkaya et al. 2019). There will never be a gold-standard for data analysis because observed data structures and designs are ever more complex and disciplines have their own traditions (Boisgontier and Cheval 2016). However, at the crux of any research design is understanding the data. Linear mixed effects models are highly flexible and capable of describing complex relationships, and estimation methods can quickly illustrate simple fixed effect comparisons. Estimation plots have one distinct advantage over any type of significance testing which is estimation plots illustrate to the author and the reader the full distribution of the data, effect sizes, and calculated confidence intervals in a single figure (Calin-Jageman and Cumming 2019). Estimation plots may even provide a proverbial ‘light bulb’ to turn on when researchers normalize data and lose or gain ‘significant’ effects because the size and confidence in that ‘significant’ effect are reviewed by the author and reader.

5. Conclusion

Differential rearing changes baseline motivation for AMP reinforcers and AMP reinforcement is moderated in part by mGluR2/3. The ability for LY379268 to dose-dependently suppress AMP SA differently in EC and SC rats while not being effective in IC rats strongly indicates that differential rearing alters the functional status of the mGluR2/3. These differences in LY379268 potency between EC, IC, and SC rats indicate that the expression of the mGluR2/3 or the ability for the receptor to couple to its G protein is enhanced in EC rats and diminished in IC rats. These results provide the premise that early-life development impacts drug abuse vulnerability but more critically, determines the effectiveness of potential therapeutics.

Supplementary Material

1

Highlights.

  • Environmental enrichment reduces the motivation for amphetamine

  • Environmental enrichment increases the potency of a selective mGluR2/3 agonist

  • A mGluR2/3 agonist does not reverse isolation-induced detriments in mGluR function

  • Estimation plots summarize the distribution, effect size, and precision in one figure

  • Normalizing baseline differences to 100% can introduce error and reduce confidence

Acknowledgments

The authors would like to thank David L. Arndt, Ph.D. for contributing to collecting the behavioral data. The experimental work was supported by the National Institute on Drug Abuse (NIDA) R15 grant DA035435. EJG was supported by a Promoting Health Diversity Supplement awarded by the NIDA DA035435-S1. The authors do not have any conflicts of interest that would confound the research design or the interpretation of the data presented.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Adewale AS, Platt DM and Spealman RD (2006). “Pharmacological stimulation of group ii metabotropic glutamate receptors reduces cocaine self-administration and cocaine-induced reinstatement of drug seeking in squirrel monkeys.” J Pharmacol Exp Ther 318(2): 922–931. [DOI] [PubMed] [Google Scholar]
  2. Arndt DL, Arnold JC and Cain ME (2014). “The effects of mGluR2/3 activation on acute and repeated amphetamine-induced locomotor activity in differentially reared male rats.” Exp Clin Psychopharmacol 22(3): 257–265. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Arndt DL, Johns KC, Dietz ZK and Cain ME (2015). “Environmental condition alters amphetamine self-administration: role of the MGluR(5) receptor and schedule of reinforcement.” Psychopharmacology (Berl) 232(20): 3741–3752. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Arndt DL, Wukitsch TJ, Garcia EJ and Cain M (2019). “Histone deacetylase inhibition differentially attenuates cue-induced reinstatement: An interaction of environment and acH3K9 expression in the dorsal striatum.” Behav Neurosci 133(5): 478–488. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Artola A, von Frijtag JC, Fermont PC, Gispen WH, Schrama LH, Kamal A and Spruijt BM (2006). “Long-lasting modulation of the induction of LTD and LTP in rat hippocampal CA1 by behavioural stress and environmental enrichment.” The European Journal of Neuroscience 23(1): 261–272. [DOI] [PubMed] [Google Scholar]
  6. Babyak MA (2004). “What you see may not be what you get: a brief, nontechnical introduction to overfitting in regression-type models.” Psychosomatic medicine 66(3): 411–421. [DOI] [PubMed] [Google Scholar]
  7. Baptista MA, Martin-Fardon R and Weiss F (2004). “Preferential effects of the metabotropic glutamate 2/3 receptor agonist LY379268 on conditioned reinstatement versus primary reinforcement: comparison between cocaine and a potent conventional reinforcer.” J Neurosci 24(20): 4723–4727. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bardo MT and Dwoskin LP (2004). “Biological connection between novelty- and drug-seeking motivational systems.” Nebr Symp Motiv 50: 127–158. [PubMed] [Google Scholar]
  9. Bardo MT, Neisewander JL and Kelly TH (2013). “Individual differences and social influences on the neurobehavioral pharmacology of abused drugs.” Pharmacol Rev 65(1): 255–290. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Barr DJ, Levy R, Scheepers C and Tily HJ (2013). “Random effects structure for confirmatory hypothesis testing: Keep it maximal.” Journal of memory and language 68(3): 255–278. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bates D, Mächler M, Bolker B and Walker S (2015). “Fitting Linear Mixed-Effects Models Using lme4.” Journal of Statistical Software; Vol 1, Issue 1 (2015). [Google Scholar]
  12. Boisgontier MP and Cheval B (2016). “The anova to mixed model transition.” Neurosci Biobehav Rev 68: 1004–1005. [DOI] [PubMed] [Google Scholar]
  13. Calin-Jageman RJ and Cumming G (2019). “Estimation for Better Inference in Neuroscience.” eNeuro 6(4): ENEURO.0205–0219.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Corrigall WA, Coen KM, Zhang J and Adamson KL (2001). “GABA mechanisms in the pedunculopontine tegmental nucleus influence particular aspects of nicotine self-administration selectively in the rat.” Psychopharmacology (Berl) 158(2): 190–197. [DOI] [PubMed] [Google Scholar]
  15. Crawford JT, Roberts DC and Beveridge TJ (2013). “The group II metabotropic glutamate receptor agonist, LY379268, decreases methamphetamine self-administration in rats.” Drug Alcohol Depend 132(3): 414–419. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Fone KC and Porkess MV (2008). “Behavioural and neurochemical effects of post-weaning social isolation in rodents-relevance to developmental neuropsychiatric disorders.” Neurosci Biobehav Rev 32(6): 1087–1102. [DOI] [PubMed] [Google Scholar]
  17. Garcia EJ, Arndt DL and Cain ME (2019). “Dynamic interactions of ceftriaxone and environmental variables suppress amphetamine seeking.” Brain Res 1712: 63–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Garcia EJ, Jorgensen ET, Sprick LS and Cain ME (2017). “Voluntary ethanol consumption changes anticipatory ultrasonic vocalizations but not novelty response.” Behav Brain Res 320: 186–194. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Garcia KL, Le AD and Tyndale RF (2014). “Effect of food training and training dose on nicotine self-administration in rats.” Behav Brain Res 274: 10–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Goodnough A (2019). A New Drug Scourge: Deaths Involving Meth Are Rising Fast. The New York Times. [Google Scholar]
  21. Green EJ and Greenough WT (1986). “Altered synaptic transmission in dentate gyrus of rats reared in complex environments: evidence from hippocampal slices maintained in vitro.” Journal of Neurophysiology 55(4): 739–750. [DOI] [PubMed] [Google Scholar]
  22. Green TA, Gehrke BJ and Bardo MT (2002). “Environmental enrichment decreases intravenous amphetamine self-administration in rats: dose-response functions for fixed- and progressive-ratio schedules.” Psychopharmacology (Berl) 162(4): 373–378. [DOI] [PubMed] [Google Scholar]
  23. Hao Y, Martin-Fardon R and Weiss F (2010). “Behavioral and functional evidence of metabotropic glutamate receptor 2/3 and metabotropic glutamate receptor 5 dysregulation in cocaine-escalated rats: factor in the transition to dependence.” Biol Psychiatry 68(3): 240–248. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Harrison XA, Donaldson L, Correa-Cano M, Evans J, Fisher DN, Goodwin CED, Robinson BS, Hodgson DJ and Inger R (2018). “A brief introduction to mixed effects modelling and multi-model inference in ecology.” PeerJ 6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Heidbreder CA, Weiss IC, Domeney AM, Pryce C, Homberg J, Hedou G, Feldon J, Moran MC and Nelson P (2000). “Behavioral, neurochemical and endocrinological characterization of the early social isolation syndrome.” Neuroscience 100(4): 749–768. [DOI] [PubMed] [Google Scholar]
  26. Ho J, Tumkaya T, Aryal S, Choi H and Claridge-Chang A (2019). “Moving beyond P values: data analysis with estimation graphics.” Nat Methods 16(7): 565–566. [DOI] [PubMed] [Google Scholar]
  27. Jamieson J (1999). “Dealing with baseline differences: two principles and two dilemmas.” International journal of psychophysiology : official journal of the International Organization of Psychophysiology 31(2): 155–161. [DOI] [PubMed] [Google Scholar]
  28. Jin X, Semenova S, Yang L, Ardecky R, Sheffler DJ, Dahl R, Conn PJ, Cosford ND and Markou A (2010). “The mGluR2 positive allosteric modulator BINA decreases cocaine self-administration and cue-induced cocaine-seeking and counteracts cocaine-induced enhancement of brain reward function in rats.” Neuropsychopharmacology 35(10): 2021–2036. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kalivas PW (2009). “The glutamate homeostasis hypothesis of addiction.” Nat Rev Neurosci 10(8): 561–572. [DOI] [PubMed] [Google Scholar]
  30. Kariisa M, Scholl L, Wilson N, Seth P and Hoots B (2019). “Drug Overdose Deaths Involving Cocaine and Psychostimulants with Abuse Potential - United States, 2003–2017.” MWR Morb Mortal Wkly Rep. 68: 388–395. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kelly A and Hannan AJ (2019). “Therapeutic impacts of environmental enrichment: Neurobiological mechanisms informing molecular targets for enviromimetics.” Neuropharmacology 145(Pt A): 1–2. [DOI] [PubMed] [Google Scholar]
  32. Kim JH, Austin JD, Tanabe L, Creekmore E and Vezina P (2005). “Activation of group II mGlu receptors blocks the enhanced drug taking induced by previous exposure to amphetamine.” Eur J Neurosci 21(1): 295–300. [DOI] [PubMed] [Google Scholar]
  33. Kuznetsova A, Brockhoff PB and Christensen RHB (2017). “lmerTest Package: Tests in Linear Mixed Effects Models.” Journal of Statistical Software; Vol 1, Issue 13 (2017). [Google Scholar]
  34. Makowski (2018). “The psycho Package: an Efficient and Publishing-Oriented Workflow for Psychological Science.” Journal of Open Source Softward 3(22). [Google Scholar]
  35. Manzoni O, Michel JM and Bockaert J (1997). “Metabotropic glutamate receptors in the rat nucleus accumbens.” The European journal of neuroscience 9(7): 1514–1523. [DOI] [PubMed] [Google Scholar]
  36. Melendez RI, Gregory ML, Bardo MT and Kalivas PW (2004). “Impoverished rearing environment alters metabotropic glutamate receptor expression and function in the prefrontal cortex.” Neuropsychopharmacology 29(11): 1980–1987. [DOI] [PubMed] [Google Scholar]
  37. Monn JA, Valli MJ, Massey SM, Hansen MM, Kress TJ, Wepsiec JP, Harkness AR, Grutsch L, Wright RA, Johnson BG, Andis SL, Kingston A, Tomlinson R, Lewis R, Griffey KR, Tizzano JP and Schoepp DD (1999). “Synthesis, Pharmacological Characterization, and Molecular Modeling of Heterobicyclic Amino Acids Related to (+)-2-Aminobicyclo[3.1.0]hexane- 2,6-dicarboxylic Acid (LY354740): Identification of Two New Potent, Selective, and Systemically Active Agonists for Group II Metabotropic Glutamate Receptors.” Journal of Medicinal Chemistry 42(6): 1027–1040. [DOI] [PubMed] [Google Scholar]
  38. Neugebauer NM, Cortright JJ, Sampedro GR and Vezina P (2014). “Exposure to nicotine enhances its subsequent self-administration: contribution of nicotine-associated contextual stimuli.” Behav Brain Res 260: 155–161. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. NIDA. (2014). “Trends & Statistics.” Retrieved May 2018, from https://www.drugabuse.gov/related-topics/trends-statistics.
  40. Rahman S and Bardo MT (2008). “Environmental enrichment increases amphetamine-induced glutamate neurotransmission in the nucleus accumbens: a neurochemical study.” Brain Res 1197: 40–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Renner MJ and Rosenzweig MR (1987). Enriched and impoverished environments: Effects on brain and behavior. New York, Springer-Verlag. [Google Scholar]
  42. Richardson NR and Roberts DC (1996). “Progressive ratio schedules in drug self-administration studies in rats: a method to evaluate reinforcing efficacy.” J Neurosci Methods 66(1): 1–11. [DOI] [PubMed] [Google Scholar]
  43. Robles F (2018). Meth, the Forgotten Killer, Is Back. And It’s Everywhere. The New York Times: A1. [Google Scholar]
  44. Rosenzweig MR, Bennett EL, Hebert M and Morimoto H (1978). “Social grouping cannot account for cerebral effects of enriched environments.” Brain Research 153(3): 563–576. [DOI] [PubMed] [Google Scholar]
  45. Ross JT, Corrigall WA, Heidbreder CA and LeSage MG (2007). “Effects of the selective dopamine D3 receptor antagonist SB-277011A on the reinforcing effects of nicotine as measured by a progressive-ratio schedule in rats.” Eur J Pharmacol 559(2–3): 173–179. [DOI] [PubMed] [Google Scholar]
  46. Sharp PE, McNaughton BL and Barnes CA (1985). “Enhancement of hippocampal field potentials in rats exposed to a novel, complex environment.” Brain Research 339(2): 361–365. [DOI] [PubMed] [Google Scholar]
  47. Shoblock JR, Sullivan EB, Maisonneuve IM and Glick SD (2003). “Neurochemical and behavioral differences between d-methamphetamine and d-amphetamine in rats.” Psychopharmacology (Berl) 165(4): 359–369. [DOI] [PubMed] [Google Scholar]
  48. Stairs DJ and Bardo MT (2009). “Neurobehavioral effects of environmental enrichment and drug abuse vulnerability.” Pharmacol Biochem Behav 92(3): 377–382. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. van Praag H, Kempermann G and Gage FH (2000). “Neural consequences of environmental enrichment.” Nature Reviews Neuroscience 1(3): 191–198. [DOI] [PubMed] [Google Scholar]
  50. Wang M, Marshall AT and Kirkpatrick K (2017). “Differential effects of social and novelty enrichment on individual differences in impulsivity and behavioral flexibility.” Behavioural Brain Research 327(Neurosci. Biobehav. Rev. 32 2008): 54–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Wickham et al. , (2019). “Welcome to the tidyverse”. Journal of Open Source Software, 4(43), 1686, 10.21105/joss.01686 [DOI] [Google Scholar]
  52. Wooters TE, Bardo MT, Dwoskin LP, Midde NM, Gomez AM, Mactutus CF, Booze RM and Zhu J (2011). “Effect of environmental enrichment on methylphenidate-induced locomotion and dopamine transporter dynamics.” Behav Brain Res 219(1): 98–107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Yates JR, Bardo MT and Beckmann JS (2019). “Environmental enrichment and drug value: a behavioral economic analysis in male rats.” Addict Biol 24(1): 65–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Young ME (2016). “The problem with categorical thinking by psychologists.” Behav Processes 123: 43–53. [DOI] [PubMed] [Google Scholar]
  55. Young ME, Clark MH, Goffus A and Hoane MR (2009). “Mixed effects modeling of Morris water maze data: Advantages and cautionary notes.” Learning & Motivation 40: 160–177. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1580365-supplement-1.docx (56.9KB, docx)

RESOURCES