|Home | About | Journals | Submit | Contact Us | Français|
The Val66Met polymorphism in the brain-derived neurotropic factor (BDNF) gene results in alterations in fear extinction behavior in both human populations and mouse models. However, it is not clear whether this polymorphism plays a similar role in extinction of appetitive behaviors. Therefore, we examined operant learning and extinction of both food and cocaine self-administration behavior in an inbred genetic knock-in mouse strain expressing the variant Bdnf. These mice provide a unique opportunity to relate alterations in aversive and appetitive extinction learning as well as provide insight into how human genetic variation can lead to differences in behavior. BDNFMet/Met mice exhibited a severe deficit in operant learning as evidenced by an inability to learn the food self-administration task. Therefore, extinction experiments were performed comparing wildtype (BDNFVal/Val) animals to mice heterozygous for the Met allele (BDNFVal/Met), which did not differ in food or cocaine self-administration behavior. In contrast to the deficit in fear extinction previously demonstrated in these mice, we found that BDNFVal/Met mice exhibited more rapid extinction of cocaine responding compared to wildtype mice. No differences were found between the genotypes in the extinction of food self-administration behavior or the reinstatement of cocaine seeking, indicating the effect is specific to extinction of cocaine responding. These results suggest that the molecular mechanisms underlying aversive and appetitive extinction are distinct from one another and BDNF may play opposing roles in the two phenomena.
Addiction is a characterized by compulsive drug use that persists in the face of negative consequences and desire to stop use (Kalivas & O'Brien, 2008; O'Brien, 2011). Conditioned drug craving elicited by drug-paired cues is among the many factors that contribute to this compulsive use. Addicts report that these conditioned cues are one of the most potent factors contributing to relapse (Heather et al., 1991). Understanding how to extinguish this cue-elicited craving is a major goal in addiction research. Genetically modified mice provide a useful model system to examine molecular mechanisms underlying cocaine extinction behavior and how this might lead to alterations in vulnerability in human populations.
Brain-derived neurotrophic factor (BDNF), a member of the neurotrophin family of polypeptide growth factors, is widely expressed throughout the brain and is regulated in an activity dependent manner (Goodman et al., 1996; Egan et al., 2003). BDNF has been implicated in synaptic plasticity, learning, and addiction. Manipulations of BDNF levels in the brain have been demonstrated to modulate addictive behaviors, such as cocaine-induced locomotor activity (Pierce & Bari, 2001), cocaine conditioned place preference (Horger et al., 1999; Hall et al., 2003), and cocaine self-administration (Lu et al., 2004). Furthermore, upregulation of BDNF expression is observed in various limbic nuclei during abstinence from cocaine self-administration (Sadri-Vakili et al., 2010), and levels of BDNF expression correlate with measures of cue-elicited craving (Grimm et al., 2003).
Along with this evidence from the preclinical literature, there is also a relationship between substance abuse and BDNF in human populations. At a genetic level, single nucleotide polymorphisms (SNPs) conferring vulnerability for polysubstance abuse were indentified flanking the BDNF gene (Uhl et al., 2001). More recently, a common SNP in the BDNF gene that leads to a valine (Val) to methionine (Met) substitution at codon 66 (Val66Met) was linked to substance abuse vulnerability. A higher frequency of the Val allele is found among methamphetamine addicts compared to non-addicted individuals, suggesting the Met allele may confer resistance to addiction (Cheng et al., 2005). However, there is an increase in 66Met allele frequency in smokers and healthy Met carriers consume more alcohol per week than 66Val homozygotes (Lang et al., 2007; Colzato et al., 2011). Parsing the role of this SNP in addictive behavior using an animal model could provide unique insight into the pathophysiology of addiction.
The current study utilized an inbred genetic knock-in mouse strain that expresses the variant BDNF allele to recapitulate the specific phenotypic properties of the human polymorphism in vivo. Previous work has demonstrated that the BDNF Val66Met genotype is associated with a resistance to the extinction of fear memories (Soliman et al., 2010). While there is considerable overlap between the neurocircuitry underlying the extinction of aversive and appetitive memories (Ressler et al., 2004; Davis et al., 2006; Peters et al., 2009; Myers & Carlezon, 2010b), it is clear that the mechanisms molecular mechanisms are not identical (Chhatwal et al., 2009; Ward et al., 2009; Bernardi & Lattal, 2010). The objective of this study was to assess the acquisition, stability and extinction of an appetitive behavior (cocaine self-administration) in Vall66Met knock-in mice.
Mice bred and maintained on a C57Bl/6 background as described previously (Chen et al., 2006), were single-housed with food and water available ad libitum. Adult wildtype (BDNFVal/Val), littermate BDNFVal/Met, and littermate BDNFMet/Met male and female mice derived from BDNFVal/Met × BDNFVal/Met crossed parents were used for all experiments. All animals (2-4 months of age) were housed in a temperature and humidity controlled animal care facility with a 12 hr light/dark cycle (lights on a 7:00 A.M.). All procedures were approved by the University of Pennsylvania Animal Care and Use Committee.
Cocaine was obtained from the National Institutes of Drug Abuse Drug Supply Program (Bethesda, MD) and dissolved in sterile 0.9% saline.
Prior to catheterization, mice were trained to perform an operant response for sucrose pellets. Although prior food training can lead to cross-sensitization, mice will not self-administer reliably without this training (Thomsen & Caine, 2007). The mice were placed in operant chambers (Med-Associates) and trained to spin a wheel manipulandum to receive a sucrose pellet. A compound cue stimulus consisting of a cue light above the active lever, a 2,900 Hz tone, and house light off was concurrent with each pellet administration, followed by an additional 8 s time-out when responding had no programmed consequences and the house light remained off. Mice were allowed to self-administer a maximum of 50 pellets per 60 min operant session. During the food self-administration phase, mice were food restricted to approximately 90% of their free feeding weight. Mice received 10 training sessions and animals that did not distinguish between the active and inactive wheel (>60% active responding) were excluded from further study.
Prior to surgery, mice were anesthetized with 80 mg/kg ketamine and 12 mg/kg xylazine. An indwelling silastic catheter was placed into the right jugular vein and sutured in place. The catheter was then threaded subcutaneously over the shoulder blade and was routed to a mesh backmount platform (Strategic Applications, Inc) that secured the placement. Catheters were flushed daily with 0.1 ml of an antibiotic (Timentin, 0.93 mg/ml) dissolved in heparinized saline. The catheters were sealed with plastic obturators when not in use.
After surgery, mice were allowed 3-4 d to recover before beginning behavioral testing. Mice were tested for cocaine self-administration behavior in 2 hour sessions (6 d per week) in the same chamber used for sucrose pellet self-administration. During testing, a quarter turn of a wheel (FR1) on the same wheel used for sucrose training now delivered an intravenous cocaine injection (0.5 mg/kg/infusion). Each session began with a single infusion of cocaine accompanied by the presentation of the tone and light cue. A compound cue stimulus consisting of a cue light above the active lever, a 2,900 Hz tone, and house light off was concurrent with each injection, followed by an additional 8 s time-out when responding had no programmed consequences and the house light remained off. Following 10 days of cocaine self-administration, cocaine-seeking behavior was extinguished by replacing the cocaine with 0.9% saline. Daily 2-h extinction sessions continued for 10 days. Following this 10-day extinction phase, animals that had not reached a criterion of less than 25% of their self-administration responding continued extinction for 5 additional days. Animals underwent cue-induced reinstatement after either meeting the extinction criterion or 15 days of extinction, whichever was shorter. During the cue-induced reinstatement session, the light and tone cues were presented non-contingently for 20 seconds every 2 minutes during the first 10 minutes of the session. After this time period, the cues were presented contingent with operant responding, just as was done during the cocaine self-administration phase. During the reinstatement session, animals received saline infusions following responses on the active wheel.
Separate groups of mice were trained to lever press for sucrose as described above, and then received and additional 10 days of 2-hour sucrose self-administration sessions (6 d per week) in the same chamber used for sucrose training. A compound cue stimulus consisting of a cue light above the active lever, a 2,900 Hz tone, and house light off was concurrent with each pellet administration, followed by an additional 8 s time-out when responding had no programmed consequences and the house light remained off. Each session began with a pellet delivery accompanied by the presentation of a tone and light cue. Following 10 days of food self-administration, food-seeking behavior was extinguished by eliminating the food delivery following the operant response. Daily 2-h extinction sessions continued for 10 days.
Total active operant responses over the course of the ten days of food training were analyzed using a repeated measures two-way ANOVA with genotype (Val/Val, Val/Met, Met/Met) and session as the independent variables and active responses as the dependent variable. Bonferonni's post hoc comparisons were conducted when main effects or interactions were present. Similar analyses were performed for the cocaine self-administration, extinction and reinstatement phases of the experiment.
The number of pellets earned and active responses during each sucrose self-administration session was measured. BDNFVal/Val and BDNFVal/Met mice acquired stable responding for food, readily responding for 20 or more pellets after 5 days of training. However, BDNFMet/Met mice failed to acquire the operant response for food despite continued training. While no differences were seen between BDNFVal/Val and BDNFVal/Met mice with either variable, BDNFMet/Met mice performed significantly fewer active responses (Fig. 1; two-way repeated measures ANOVA, effect of genotype: F(2, 441)= 4.41, p=.017; Bonferroni post-hoc BDNFMet/Met v. BDNFVal/Val and BDNFVal/Met p<.05) and received significantly fewer pellets [Fig. 1; two-way repeated measures ANOVA, effect of genotype: F(2, 441)= 11.57, p=.0001; interaction: F(18, 441)= 1.73, p=.032; Bonferroni post-hoc BDNFMet/Met v. BDNFVal/Val and BDNFVal/Met p<.05]. The BDNFMet/Met mice were not included in the cocaine or food extinction experiments because they were unable to acquire the operant response for food during training.
The number of infusions earned, active responses and percent active responding during each cocaine self-administration session in shown in Fig. 2. Both BDNFVal/Val and BDNFVal/Met mice acquired stable cocaine self-administration behavior after 10 sessions. Although BDNFVal/Met mice showed a trend toward slower acquisition on days 1-4, statistical analyses indicated no differences between the groups in the number of infusions earned or active responses performed over the 10 sessions. Additionally, no differences were seen in the percent active responding with both groups stabilizing at ~70% correct (Fig. 2).
The number of active responses as well as the percent of responding compared to the average responding on the last two days of cocaine self-administration was measured. BDNFVal/Met mice exhibited significantly faster extinction of cocaine responding compared to BDNFVal/Val mice, as measured by both a decrease in active responding during extinction [Fig. 3; two-way repeated measures ANOVA, effect of genotype: F(1, 198)= 4.46, p=.04; effect of session: F(9, 198)= 8.81, p=.0001] and a decrease in the percent responding compared to the cocaine self-administration phase [Fig. 3; two-way repeated measures ANOVA, effect of genotype: F(1,198)= 24.56, p=.0001; effect of session: F(9,198)= 11.73, p=.0001]. Both BDNFVal/Val and BDNFVal/Met mice exhibited significant reinstatement of responding compared to the prior extinction day, however no differences were seen between the genotypes [two-way repeated measures ANOVA, effect of test: F(1,17)=31.42, p=.0001].
The number of active responses as well as the percent of responding compared to the food self-administration phase was measured. BDNFVal/Val and BDNFVal/Met mice exhibited similar rates of extinction of food responding as measured by equivalent active responding during extinction and equivalent percent responding compared to the food self-administration phase [Fig. 4; two-way repeated measures ANOVA, effect of session: F(9,99)= 15.04, p=.0006].
Brain-derived neurotrophic factor (BDNF) has been implicated in many facets of addiction in preclinical models, including cocaine conditioned reward (Horger et al., 1999; Lobo et al., 2010), cocaine craving (Grimm et al., 2003), cocaine reinforcement (Sadri-Vakili et al., 2010), and cocaine reinstatement (Berglind et al., 2007). Variants in the BDNF gene have also been implicated in vulnerability to addiction in human addicts (Krebs et al., 2000; Uhl et al., 2001; Tsai, 2007). Using a mouse model system that contains the human SNP in the BDNF gene that leads to a valine (Val) to methionine (Met) substitution at codon 66 (Val66Met), the present study provides several new insights into phenotypes associated with the variant BDNFMet. First, we provide evidence for learning impairment in that BDNFMet/Met mice failed to acquire food self-administration. This learning impairment was specific to the Met homozygotes, as there was no detectable learning deficit in the BDNFVal/Met mice. Second, when examining acquisition and stability of cocaine self-administration behavior, we found no differences between BDNFVal/Val and BDNFVal/Met mice. However, BDNFVal/Met mice exhibited more rapid extinction of cocaine responding compared to BDNFVal/Val controls. This is in direct contrast to the impaired fear extinction behavior seen in both human and mouse BDNF Met carriers (Soliman et al., 2010). Furthermore, there was no difference between the genotypes in extinction of food seeking. Since all mice were food trained prior to cocaine self-administration, it seems unlikely that the cocaine effects resulted from alterations in extinction of food responding during the initial cocaine self-administration sessions. Additionally, this suggests that the molecular mechanisms underlying appetitive and aversive fear extinction are at least somewhat distinct since BDNF appears to play different role in these two forms of learning.
The growth factor, BDNF, is the most abundant neurotrophin in the mammalian central nervous system and is involved in mediating synaptic plasticity associated with various forms of learning (Patterson et al., 1996; Mu et al., 1999; Mizuno et al., 2000; Barco et al., 2005). BDNF plays a critical role in hippocampus-dependent spatial learning (Tyler et al., 2002) as well as amygdala-dependent Pavlovian fear conditioning (Rattiner et al., 2004; Ou & Gean, 2006); however, its role in operant learning is less clear. While learning an operant task, BDNF mRNA levels increase in the medial prefrontal cortex (Rapanelli et al., 2010). Despite this, infusions of BDNF into this brain region do not alter established food seeking behavior (Berglind et al., 2007). BDNF function in the nucleus accumbens, while clearly involved in addiction, does not seem to be related to operant learning in that the increases in accumbens BDNF seen following cocaine self-administration are not seen following food self-administration (Grimm et al., 2003). Furthermore, infusions into the nucleus accumbens of either BDNF or an antibody to BDNF had no effect on established operant responding for food (Graham et al., 2007). Additionally, knockdown of TrkB, the primary receptor for BDNF, in the nucleus accumbens had no effect on the acquisition of operant behavior for either sucrose or cocaine (Graham et al., 2009). In contrast, infusions of BDNF into the dorsomedial striatum led to a decrease in established sucrose self-administration (Jeanblanc et al., 2009). However, indirectly decreasing BDNF levels within the striatum, through knockdown of the transcription factor MeCP2, or directly manipulating BDNF through infusions of a BDNF antibody, leads to a decrease in cocaine intake during extended access cocaine self-administration (Im et al., 2010). Furthermore, viral overexpression of BDNF in the striatum increases cocaine intake during extended access self-administration, but not during limited access paradigms (Im et al., 2010). This suggests that the role of BDNF within operant behavior in the striatum is quite complex and may depend not only on the reinforcer but also on the training regimen.
Although no studies have specifically examined instrumental learning, the BDNF Val66Met polymorphism has been studied clinically in other forms of learning. BDNF Met carriers exhibit deficits in episodic memory (Egan et al., 2003; Dempster et al., 2005), recognition memory (Hariri et al., 2003; Goldberg et al., 2008) and spatial navigation (Banner et al., 2011). Furthermore, these deficits in memory performance are accompanied by a decrease in hippocampal activity in BDNF Met carriers during the performance of memory tasks (Hariri et al., 2003; Hashimoto et al., 2008; Banner et al., 2011). The results of the current study clearly indicate that the deficits in learning seen in BDNF Met carriers may extend beyond tasks dependent upon the hippocampus and demonstrate a need for examining the role of this BDNF gene variant in striataldependent learning.
Recent work demonstrated a role for the BDNF Val66Met polymorphism in aversive extinction learning. Using the same mouse strain utilized in the current studies, the Met allele was associated with both increased anxiety-related behaviors (Chen et al., 2006) and impairment of conditioned fear and conditioned taste aversion extinction (Yu et al., 2009; Soliman et al., 2010). This deficit in extinction learning has also been documented in human BDNF Met carriers (Soliman et al., 2010). In contrast, the current study demonstrated facilitation of drug-related extinction learning in BDNFVal/Met mice compared to wildtype controls. This suggests that the role of BDNF in these two forms of extinction is distinct.
Understanding the role that BDNF may play in these two phenomena may require a better understanding of how the BDNF Val66Met polymorphism affects the levels of BDNF in different brain regions and how BDNF in different brain regions may play diverse roles in appetitive and aversive extinction learning. Knockdown of BDNF within the hippocampus or the amygdala leads to impaired aversive extinction behavior, suggesting that BDNF is necessary in these regions for normal extinction (Chhatwal et al., 2006; Heldt et al., 2007). Furthermore, infusions of BDNF into the infralimbic cortex, following fear conditioning, mimic the effects of extinction training (Peters et al., 2010). These findings are all consistent with the idea that it is a decrease of regulated BDNF that is mediating the impaired aversive extinction behavior seen in the BDNF Met carriers (Soliman et al., 2010).
The effects of BDNF on the extinction of appetitive learning are less clear. Similarly to what is seen with aversive extinction, infusions of BDNF into the prefrontal cortex lead to facilitation of extinction of cocaine seeking (Berglind et al., 2007). However, an infusion of BDNF into the ventral tegmental area blocks extinction of cocaine seeking (Lu et al., 2004). These results suggest that a decrease in BDNF may lead to a decrease in cocaine craving, which is consistent with other preclinical data. During withdrawal from cocaine self-administration, BDNF levels within the ventral tegmental area increase, and this increase has been correlated with levels of cocaine craving (Grimm et al., 2003). Furthermore, knockdown of BDNF within either the VTA or the nucleus accumbens leads a decrease in cocaine-conditioned reward (Graham et al., 2009). Taken together, these data indicate that the decrease in regulated BDNF secretion within the VTA of BDNF Met mice may predominate over any other alterations in BDNF and thus lead to facilitation of the extinction of cocaine seeking behavior.
Although there is some overlap in the brain regions thought to mediate aversive and appetitive extinction (Peters et al., 2009), it is apparent that the exact mechanisms underlying these forms of learning are not identical. Just as BDNF seems to play opposing roles in aversive and appetitive extinction, other signaling systems also lead to discriminate effects in these two phenomena. For example, while the cannabinoid receptor antagonist, rimonabant, blocks fear extinction (Chhatwal et al., 2009); it facilitates the extinction of operant responding for food or cocaine (Ward et al., 2009). Similarly, the alpha-adrenergic receptor antagonist, prazosin, inhibits fear extinction while having no effect on appetitive extinction (Bernardi & Lattal, 2010). However, the administration of a partial NMDA agonist, a drug that arguably has more widespread effects than either rimonabant or prazosin, facilitates both aversive and appetitive extinction (Ressler et al., 2004; Parnas et al., 2005; Botreau et al., 2006; Davis et al., 2006; Shaw et al., 2009; Myers & Carlezon, 2010a; Thanos et al., 2011a; Thanos et al., 2011b). These studies, along with inactivation studies (Morgan et al., 1993; Morgan & LeDoux, 1995; Peters et al., 2008), suggest that there is some overlap between the circuits mediating these two types of extinction learning but the underlying molecular mechanisms differ.
Furthermore, the mechanisms underlying different types of appetitive extinction may also be distinct. The facilitation of drug-related extinction behavior seen in the BDNFVal/Met mice in the current study did not extend to all forms of appetitive learning, as no effect of the BDNF SNP was seen in the extinction of food seeking behavior. This is consistent with the literature demonstrating a clear role for BDNF in cocaine seeking that is distinct from food seeking. For example, increases in BDNF within the ventral tegmental area, nucleus accumbens and amygdala are seen following withdrawal from cocaine seeking but not from sucrose self-administration (Grimm et al., 2003). Furthermore, manipulations of BDNF that alter cocaine seeking do not have any effect on food seeking behavior (Berglind et al., 2007).
In recent years, BDNF gene variants, most commonly the Val66Met SNP examined in the current study, have been implicated in a wide range of psychiatric disorders. BDNF Met carriers are at a higher risk for depression (Schumacher et al., 2005; Hwang et al., 2006; Verhagen et al., 2010), aggressive forms of schizophrenia (Spalletta et al., 2010), severe depressive episodes in bipolar depression (Hosang et al., 2010) and anxiety-related traits (Montag et al., 2010). Furthermore, BDNF Met carriers have a higher susceptibility to stress, which may contribute to the effect of this genotype on depression and anxiety (Shalev et al., 2009; Alexander et al., 2010). In contrast, the current study found that BDNFVal/Met mice exhibit an increased rate of extinction of cocaine responding. Despite this increased rate of extinction, BDNFVal/Met mice exhibited equivalent levels of cue-induced reinstatement. As extinction is believed to be a new form of learning that is dependent upon the ventral prefrontal cortex, whereas cue-induced reinstatement depends on projections from the amygdala to the dorsal prefrontal cortex. Therefore, our data suggest that BDNF Met carriers may have an increase in the ability to form new associations. In support of this hypothesis, human studies of BDNF Met carriers have demonstrated in enhanced behavioral flexibility in memory based switching task as well as enhanced response inhibition (Beste et al., 2010; Gajewski et al., 2011). Although BDNF Met carriers clearly exhibit some learning deficits, our study, as well as these human studies, suggest that this SNP can also confer an advantage in some cognitive domains.
Through the use of a mouse model system, we determined that the variant BDNFMet/Met leads to impaired instrumental learning. This abnormality is not seen in BDNFMet/Val heterozygotes suggesting that there may be a threshold level of activity-dependent BDNF release that is needed for learning in this task and only the BDNFMet/Met mice are below this threshold. Furthermore, we demonstrate that BDNFMet/Val heterozygotes exhibit a facilitation of extinction of cocaine seeking. This is in contrast to previous findings demonstrating impairment in extinction of aversive learning (Chen et al., 2006; Soliman et al., 2010). To our knowledge, this is the first evidence that a functionally relevant polymorphism can lead to opposing behavioral phenotypes in these two types of extinction learning. These findings provide the basis for further studies to determine whether human BDNF Met carriers exhibit a similar disparity in behavioral extinction phenotypes.
We thank Ted Huang, Blake Kimmey and Jennifer Xue for technical assistance. Research supported by NIDA F32 DA026660 (LB), R01DA15214 (RCP), K02 DA18678 (RCP), R01 DA-011649 (JAB) and NINDS R01 NS052819 (FSL). The authors report no biomedical financial interests or potential conflicts of interest.