1. Introduction

A rapid transition from initial drug exposure to regular patterns of drug use is considered an important prognosticator of whether an individual will later develop problems with substance abuse and dependence (U.S. Congress, Office of Technology Assessment, 1994). Consequently, one of the goals of substance abuse prevention programs is to discourage the development of regular patterns of drug use in at-risk populations. The acquisition of regular patterns of drug intake after initial drug exposure can be modeled in the laboratory by exposing an animal to noncontingent drug infusions and then permitting the animal to self-administer that drug in free-operant test sessions. Factors that influence the acquisition of drug self-administration in the laboratory are often identical to the factors that influence the likelihood an individual will develop problems with substance abuse and dependence. For instance, social isolation (Kosten et al., 2000), stress (Tidey and Miczek, 1997), and previous drug exposure (Fletcher et al., 2001) increase the rate of acquisition in laboratory animals and are considered risk factors for developing problems with substance use in humans. Moreover, interventions that decrease the rate of acquisition in laboratory animals, such as access to alternative nondrug reinforcers, decrease the probability that human populations will develop problems with substance use (see review by Campbell and Carroll, 2000). For instance, providing laboratory rats with concurrent access to a palatable drinking solution reduces the acquisition of intravenous cocaine self-administration (Carroll and Lac, 1993), whereas providing adolescents with educational programs and social activities reduces the initiation of alcohol drinking (Perry et al., 1996; 2002).

Physical activity is one intervention that may reduce the likelihood that an individual will transition into regular patterns of drug use. Epidemiological studies report that adolescents who engage in regular physical activity are less likely to use drugs and alcohol (Field et al., 2001; Kirkcaldy et al., 2002; Ströhle et al., 2007; Iannotti et al., 2009) and have fewer risk factors that are associated with the development of substance use disorders (Collingwood et al., 1991; 2000) than their peers who do not engage in regular physical activity. Preclinical studies reveal that physical activity reduces drug self-administration in procedures that model different transitional stages in the development of substance use disorders. For instance, access to a running wheel decreases the maintenance of cocaine-reinforced behavior in animals with well-established self-administration histories (Cosgrove et al., 2002; Smith et al., 2008), and reduces drug-primed (Zlebnik et al., 2010) and cue-induced (Lynch et al., 2010) reinstatement of cocaine-seeking behavior in animals after a period of forced abstinence. The ability of wheel running to influence the acquisition of cocaine self-administration (i.e., the establishment of regular patterns of cocaine self-administration) is not known.

The purpose of the present study was to examine the effects of access to a running wheel on the acquisition of cocaine self-administration. To this end, sedentary (no wheel) and exercising (access to wheel) rats were implanted with intravenous catheters and allowed to self-administer cocaine in free-operant test sessions. Each session began with a priming infusion of cocaine, followed by a 2-hr period in which cocaine was available on a fixed ratio (FR1) schedule of reinforcement. Previous studies report that the rate of acquisition is influenced by both the dose of cocaine used to reinforce behavior (Carrol and Lac, 1997) and the level of caloric restriction at the time of testing (Specker et al., 1994; Campbell and Carroll, 2001). In the present study, acquisition testing advanced through three distinct stages designed to systematically increase the probability that self-administration would be acquired. In phase 1 (days 1–5), responding was reinforced with 0.25 mg/kg/infusion cocaine; in phase 2 (days 6–10), responding was reinforced with 0.75 mg/kg/infusion cocaine; and in phase 3 (days 11–15) responding was reinforced with 0.75 mg/kg/infusion cocaine and animals were calorically restricted to 85% to 95% of their free-feeding body weight. We hypothesized that rates of active lever pressing would increase over the 15-day period, and that access to a running wheel would significantly decrease the rate of acquisition.

2. Material and methods

3. Results

Wheel running, measured in revolutions per day, increased significantly during the six weeks prior to catheter implantation [Figure 1, main effect of week: F (5, 90) = 32.038, p < .001]. Wheel running reached a plateau immediately after surgery and no further increases were observed. Although small day-to-day variations in wheel running were apparent over the 15 days of testing, this effect failed to reach statistical significance (p =. 094). Body weights increased significantly during the first six weeks of the study [main effect of week: F (5, 180) = 113.464, p < .001], but no differences were observed between the two groups. Body weights also varied significantly over the 15 days of behavioral testing [main effect of day: F (14, 504) = 123.184, p < .001], but no group differences were observed.

The percentage of rats acquiring cocaine self-administration was greater in sedentary rats than exercising rats on all days of testing after the first day (Figure 2A). Sedentary rats acquired cocaine self-administration significantly more rapidly than exercising rats [Figure 2B, effect of group: t (36) = 2.502, p = .017], reaching the acquisition criterion, on average, three days sooner. The number of active lever presses increased significantly over the 15 days of testing [Figure 2C, main effect of session: F (14, 504) = 16.289, p < .001], and sedentary rats emitted a significantly greater number of active lever presses than exercising rats [main effect of group: F (1, 36) = 6.698, p = .014]. Over the 15 days of testing, the total number of active lever presses was approximately twice as great in sedentary rats than in exercising rats, an effect that was statistically significant [Figure 2D, effect of group: t (36) = 2.588, p = .014]. In contrast, the number of inactive lever presses did not differ between the two groups during the 15 days of testing (Figure 2E). Although some day-to-day variation was observed, this effect failed to reach statistical significance (p = .072). Unlike that seen with active lever presses, the total number of inactive lever presses over the 15 days of testing did not differ between sedentary and exercising rats (Figure 2F).

To examine the maintenance of cocaine-reinforced responding in the two groups, cocaine self-administration was compared on day 15 using only data from rats that met the acquisition criterion (Figure 3). In the 17 sedentary rats and 13 exercising rats meeting the acquisition criterion, cocaine self-administration did not differ between the two groups. Analysis of individual event records revealed stable response rates with regular post-reinforcement pauses in both sedentary and exercising rats (data not shown).

Pearson product-moment correlations did not reveal any significant relationship between wheel running (measured before surgery, after surgery, and throughout the entire study) and cocaine acquisition (measured as days to criterion and total active lever presses). Similarly, wheel running was not predictive of active lever pressing during any of the three phases of behavioral testing. Correlations between wheel running and active lever responding ranged from −.227 to .204 for all comparisons (data not shown).

4. Discussion

The main finding of this study is that access to a running wheel significantly decreases the rate of acquisition of cocaine self-administration in male rats responding in a free operant procedure. On average, rats with access to a running wheel acquired cocaine self-administration 3 days later than sedentary rats, and emitted fewer active lever presses during the 15 days of behavioral testing. Differences between the two groups were observed as early as the second day of testing, and remained evident until the conclusion of the experiment. The differential rate of acquisition could not be attributed to differences in nonspecific responding between the two groups. Rates of inactive lever pressing did not differ between the two groups, suggesting that the increased rate of acquisition observed in sedentary rats was not due to higher rates of general operant responding. Additionally, the differential rate of acquisition could not be attributed to differences in sensitivity to cocaine. When only rats meeting the acquisition criterion were compared, cocaine self-administration did not differ between the two groups. This latter finding is consistent with our previous work showing that wheel running does not alter sensitivity to cocaine when unit price (i.e., ratio value) and income (i.e., session length) is low (Smith et al., 2011). Only when ratio values are high, such as on a progressive ratio schedule of reinforcement, or when session lengths are extended to 6 or 23 hours do we see significant differences between sedentary and exercising rats (Smith et al., 2008; Smith et al., 2011).

It is important to note that access to a running wheel is a form of environmental enrichment, and the present data may be due to an enrichment-related effect. Environmental enrichment has been shown to influence the positive reinforcing effects of cocaine (Smith et al., 2009), and its protective effects on measures of drug-seeking behavior have been well documented (Bardo et al., 2001; Stairs et al., 2006; Thiel et al., 2010; 2011). Although traditional models of environmental enrichment typically include multiple enriching stimuli (e.g., access to running or climbing devices, access to novel objects or toys, access to same-sex cagemates), the present data suggest that access to a running wheel in the absence of other enriching stimuli is sufficient for decreasing drug-seeking behavior in acquisition procedures.

Although there is not a standard procedure for measuring the acquisition of drug self-administration in laboratory animals, most studies using intravenous self-administration methods begin by priming the animal with noncontingent infusions of the drug, and then allowing the animal to respond for contingent infusions under a schedule of positive reinforcement. The noncontingent infusions may be provided as a single priming infusion at the beginning of each session (as in the current study) or as part of an auto-shaping procedure in which each infusion is paired with the insertion/retraction of a response lever. Both methods are believed to model human patterns of acquisition in which drugs are initially provided freely (perhaps by friends or dealers) and then the monetary or behavioral costs of obtaining the drug increase as the individual shifts to regular patterns of use (for review and discussion, see Carroll and Meisch, 2011). Regardless of the method employed, studies have consistently shown that increasing the dose of the drug and increasing the level of caloric restriction increases the likelihood that drug self-administration will be acquired (Specker et al., 1994; Carrol and Lac, 1997; Campbell and Carroll, 2001).

In the present study, testing advanced through three distinct phases in which the dose and level of caloric restriction were systematically varied. These phases were not designed to measure the impact of dose and caloric restriction on cocaine self-administration; rather, these phases were designed to progressively increase the likelihood that self-administration would be acquired. Consistent with this aim, acquisition increased progressively in both groups over the 15 days of testing. Sedentary rats acquired cocaine self-administration at a faster rate than exercising rats, especially during the first two phases of testing (days 1–10). The majority of exercising rats that acquired did so during the third phase of testing, during which responding was maintained by the higher dose of cocaine and all rats were calorically restricted to 85% – 95% of their free-feeding body weight. It is important to note that body weights did not differ between sedentary and exercising rats at any point during the study. Previous studies report that rats with access to running wheels increase their energy intake to compensate for their increase in energy expenditure (Holloszy, 1993), and we have found that exercising rats consume approximately 35% more food than sedentary rats under free-feed conditions (unpublished observations). In the present study, exercising rats consumed 33% more food than sedentary rats during the 5 days of caloric restriction when food consumption was specifically being monitored. The finding that body weights did not differ between sedentary and exercising rats suggest that both the dose manipulation (expressed as mg/kg of body weight) and the caloric restriction manipulation (expressed as a percentage of free-feeding body weight) was equivalent between the two groups.

Consistent with previous studies (Smith and Yancey, 2003; Smith and Lyle, 2006), wheel running increased over the first several weeks of the study. Wheel running leveled off immediately after surgery, and remained relatively stable across the 15 days of testing. Previous studies have reported that both contingent (Cosgrove et al., 2002; Smith et al., 2008) and noncontingent (Iijima et al., 1995; Santucci et al., 2008) cocaine suppresses wheel running in laboratory animals. One possibility that may account for the lack of suppression in the present study is the low levels of cocaine intake when averaged across 19 rats over 15 days of testing. Because very low levels of cocaine intake were observed during the first phase of testing (few rats had acquired), and because some rats never acquired cocaine self-administration by the end of the third phase, average levels of cocaine intake were less in the present study than in earlier studies. It should be noted that our electronic counters record only the number of wheel revolutions and do not permit a temporal analysis of wheel running. It is likely that circadian patterns of wheel running were disrupted during the period of behavioral testing, but future studies using a more detailed analysis of running patterns will be needed to confirm this possibility. We previously reported that wheel running prior to surgery was later predictive of cocaine-maintained breakpoints on a progressive ratio schedule of reinforcement (Smith et al., 2008), and we predicted that wheel running would be predictive of acquisition in the present study. Contrary to our prediction, wheel running was not correlated with any measure of cocaine self-administration during any phase of testing.

Physical activity induces many neurobiological changes that could account for its ability to protect against the establishment of regular patterns of drug self-administration. For instance, running increases central dopamine concentrations (Meeusen and De Meirleir, 1995) and regular bouts of running increase the density of dopamine D₂ receptors (Gilliam et al., 1984; MacRae et al., 1987). The latter finding is particularly relevant to the present study because a higher density of D₂ receptors is associated with decreased cocaine self-administration in laboratory animals (Morgan et al., 2002). Wheel running also modulates the intracellular signaling molecule ERK, which is important for drug-induced synaptic plasticity. ERK is particularly important for drug craving (Thomas et al. 2008), and levels of ERK in central reward pathways are correlated with cocaine-seeking behavior (Koya et al., 2009; Lynch et al., 2010; Lu et al., 2006; Edwards et al., 2011). Importantly, access to a running wheel during a period of forced abstinence from cocaine self-administration reduces phosphorylated levels of ERK and decreases cocaine-seeking behavior (Lynch et al., 2010).

On a neuroanatomical level, voluntary (van Praag et al., 1999; Rhodes et al., 2003; Bednarczyk et al., 2009) and forced (Uda et al., 2006; Lou et al., 2008; Kim et al., 2010) running procedures reliably increase neurogenesis in the hippocampal formation. Hippocampal neurogenesis has protective effects in models of drug-seeking behavior, and reductions in hippocampal neurogenesis have been implicated in an increased propensity to self-administer cocaine (Noonan et al., 2010). The relationship between cocaine exposure and hippocampal neurogenesis is reciprocal, in that contingent (Sudai et al., 2011) and noncontingent (Yamaguchi et al., 2004) cocaine administration reliably decrease neurogenesis. Physical activity may thus be producing it protective effects, in part, by enhancing the ability of the hippocampus to mitigate the establishment of regular patterns of drug intake. Wheel running also increases gliogenesis in the prefrontal cortex (Mandyam et al., 2007) and enhances prefrontal-mediated executive functioning (Small et al., 2006; Yanagisawa et al., 2010). A deficit in prefrontal functioning is believed to play a role in the development of substance use disorders (Goldstein and Volkow, 2002), and physical activity may protect against the establishment of regular and compulsive patterns of drug intake via a prefrontal mechanism.

Epidemiological studies consistently reveal that adolescents who engage in regular physical activity are less likely to use drugs and have fewer risk factors associated with the development of substance use disorders (Collingwood et al., 1991; 2000; Field et al., 2001; Kirkcaldy et al., 2002; Ströhle et al., 2007; Iannotti et al., 2009). This inverse relationship between physical activity and substance use may be due to three possibilities: (1) physical activity may produce a causal decrease in drug use, either by serving as an alternative nondrug reinforcer or by promoting neurobiological changes that protect the individual from developing regular patterns of drug intake; (2) drug use may produce a causal decrease in physical activity, either by decreasing aerobic capacity or by reducing discretionary time that would otherwise be spent on nondrug-related activities; or (3) a third factor (e.g., parents, peers, and role models) may have a causal influence on both activities. Although these three possibilities are not mutually exclusive of one another, an accumulating body of evidence from preclinical studies suggests that physical activity plays a causal role in the reduction of drug-seeking behavior. This point is particularly relevant from a translational perspective, because it supports the expanded use of activity-based interventions in drug abuse prevention programs.