Cocaine self-administration regulates AMPK activity in the NAcsh
To investigate dynamic changes in AMPK activity that are related to addictive behaviors, we employed a cocaine self-administration model in rats (Fig. 1A). The results showed a decrease in p-AMPK (p = 0.000; Fig. 1B and D) 2 h after the last cocaine training session. Interestingly, the lower expression levels persisted for at least 7 days of withdrawal in the NAcsh (p = 0.000). In chronic yoked rats that received the same number and temporal pattern of cocaine injections passively throughout training, p-AMPK levels in the NAcsh decreased only 2 h after the last cocaine injection (p = 0.000; Fig. 1B) and returned to baseline levels after 1 day of withdrawal (p = 0.346; Fig. 1B). No change in total AMPK levels was found in the NAcsh (Fig. 1C and D). These results demonstrated that passive and active cocaine self-administration differentially affected AMPK activity in the NAcsh, implying a possible regulatory role for AMPK in addictive behaviors.
Pharmacological modulation of AMPK activity in the NAcsh regulates the behavioral response to cocaine
The above results demonstrated dynamic changes in AMPK activity in the NAcsh after cocaine exposure. We further investigated the consequences of regulating AMPK activity in the NAc on behavioral effects of cocaine. Rats were trained to self-administer cocaine (0.5 mg/kg/injection) in daily 3-h sessions until cocaine intake stabilized. After stabilization, the rats received infusions of the AMPK activator AICAR (2.5 µg per side), AMPK inhibitor compound C (1.5 µg per side), or vehicle in the NAcsh immediately before four consecutive sessions (Fig. 2A). Neither AICAR nor compound C infusions in the NAcsh significantly altered baseline cocaine self-administration during the treatment regimen (Fig. 2B and C).
The rats were then tested in a between-session dose-response procedure with one of five cocaine doses that were presented each hour in descending dose order. All treated rats exhibited typical inverted U-shaped dose-response curves that were dose-dependent (dose effect: F4,84 = 52.000, p = 0.000; Fig. 2D). A descending limb was observed, whereby higher injection doses prolonged the duration of cocaine reward, resulting in fewer self-injections over time than with lower injection doses. The repeated-measures ANOVA revealed significant effects of group (F2,21 = 13.564, p = 0.000) and a group × dose interaction (F8,84 = 3.814, p = 0.001). The prior intra-NAcsh infusion of AICAR decreased cocaine self-administration at a threshold dose of 0.1 mg/kg/injection at the peak of the dose-response curve compared with vehicle-infused controls (p = 0.012; Fig. 2D). The prior intra-NAcsh infusion of compound C increased cocaine self-administration at a threshold dose of 0.03 mg/kg/injection on the ascending limb of the dose-response curve compared with vehicle-infused controls (p = 0.013; Fig. 2D). A vertical shift in the cocaine dose-response curve characterizes a transition to an addiction-like state. The above results suggested that the inhibition of AMPK activity in the NAcsh increased the sensitivity to cocaine reinforcement and subsequently facilitated the transition to addiction, whereas augmenting AMPK activity had an opposite effect.
After the dose-response test, cocaine self-administration was re-stabilized at 0.5 mg/kg/injection. Subsequently, the motivation for cocaine was assessed using a progressive-ratio schedule of reinforcement to determine the highest ratio of nosepoke responses per cocaine injection that rats would achieve before they voluntarily ceased self-administration (i.e., the breakpoint). Breakpoints were measured at two cocaine injection doses (0.25 and 0.75 mg/kg/injection). The rats generally worked harder for the higher injection dose (dose effect: F1,19 = 46.070, p = 0.000; Fig. 2E), indicating greater motivation to self-administer the higher dose of cocaine. The repeated-measures ANOVA revealed a significant effect of group (F2,19 = 7.412, p = 0.004) and a group × dose interaction (F2,19 = 5.663, p = 0.012). Rats that received intra-NAcsh infusions of compound C achieved higher breakpoints at the 0.75 mg/kg dose of cocaine compared with vehicle-infused controls (p = 0.050), indicating greater motivation for cocaine. Conversely, rats that received intra-NAcsh AICAR infusions achieved lower breakpoints at both injection doses, but no significant difference from vehicle-infused controls was found.
Modulation of AMPK activity in the NAcsh by adenovirus transfer regulates the behavioral response to cocaine
Next, we further confirmed the effects of manipulating AMPK activity on cocaine reinforcement and motivation by expressing CA-AMPK (T172D mutation in the α2 subunit of AMPK) or DN-AMPK (K45R mutation in the α2 subunit of AMPK) subunits of AMPK in the NAcsh. We first examined the efficiency and localization of modulating AMPK activity by adenoviral transfer in the NAcsh (Fig. 3B and C). The rats were given bilateral intra-NAcsh infusions of Ad-GFP, Ad-CA-AMPK, or Ad-DN-AMPK immediately before the onset of cocaine self-administration (Fig. 3A). The repeated-measures ANOVA of self-infusions revealed a significant group × training session interaction (F24,324 = 3.435, p = 0.000). The repeated-measures ANOVA of active nosepokes revealed a significant effect of group (F2,27 = 5.905, p = 0.007) and a group × training session interaction (F24,324 = 3.990, p = 0.000). No significant differences in overall cocaine self-administration were found on the FR1 or FR3 schedule during acquisition training (Fig. 3D and E). On the FR5 schedule, Ad-CA-AMPK-infused rats exhibited lower cocaine infusions compared with Ad-GFP-infused controls (infusions on day 10: p = 0.009; Fig. 3D). Ad-DN-AMPK-infused rats responded much more vigorously for cocaine infusions (nosepokes on day 10: p = 0.002, nosepokes on day 11: p = 0.012, nosepokes on day 12: p = 0.031; Fig. 3E). Thus, prolonged increases or decreases in AMPK activity in the NAcsh affected the acquisition of cocaine self-administration under conditions of higher motivation.
The rats were subsequently tested in a between-session dose-response procedure. All cocaine-treated rats exhibited inverted U-shaped dose-response curves that were dose-dependent (dose effect: F4,108 = 64.883, p = 0.000; Fig. 3F). The repeated-measures ANOVA revealed a significant effect of group (F2,27 = 16.234, p = 0.000) and a group × dose interaction (F8,108 = 6.244, p = 0.000). Ad-GFP-infused control rats exhibited typical inverted U-shaped dose-response curves (Fig. 3F). The dose-response curve for Ad-CA-AMPK-infused rats was flat and shifted downward, and cocaine self-administered rats exhibited a decrease at a threshold dose of 0.1 mg/kg/injection at the peak of the dose-response curve compared with Ad-GFP-infused controls (p = 0.002; Fig. 3F). The dose-response curve in Ad-DN-AMPK-infused rats shifted upward, and cocaine self-administering rats exhibited an increase at a threshold dose of 0.03 mg/kg/injection on the ascending limb (p = 0.003; Fig. 3F) and a threshold dose of 1 mg/kg/injection on the descending limb of the dose-response curve compared with Ad-GFP-infused controls (p = 0.004; Fig. 3F). These data indicated that AMPK activity in the NAcsh regulated cocaine reinforcement under the more demanding FR5 reinforcement schedule and was critical for vertical shifts in the dose-response curve that would reflect the transition to an addictive state.
After the dose-response test, cocaine self-administration was re-stabilized at 0.5 mg/kg/injection. Subsequently, the motivation for cocaine was assessed using a progressive-ratio schedule of reinforcement, and the breakpoint was measured at two cocaine injection doses (0.25 and 0.75 mg/kg/injection). The rats generally worked harder for the higher injection dose (dose effect: F1,27 = 86.920, p = 0.000; Fig. 3G), indicating greater motivation to self-administer the higher dose of cocaine. The repeated-measures ANOVA revealed a significant effect of group (F2,27 = 9.500, p = 0.001) and a group × dose interaction (F2,27 = 8.296, p = 0.002). Rats that received intra-NAcsh Ad-DN-AMPK infusions achieved higher breakpoints at the 0.75 mg/kg cocaine injection dose compared with Ad-GFP-infused controls (p = 0.028; Fig. 3G), indicating greater motivation for cocaine. Conversely, rats that received intra-NAcsh Ad-CA-AMPK infusions achieved lower breakpoints at both injection doses but exhibited no significant difference from Ad-GFP-infused controls.
CRTC1 mediates the effects of AMPK on cocaine reinforcement and motivation
The above results demonstrated that cocaine reinforcement behaviors resulted in a decrease in AMPK activity, and the suppression of AMPK activity further enhanced self-administration behaviors. Given that the decrease in AMPK activity promoted the dephosphorylation of CRTC1, we investigated whether AMPK regulated cocaine reinforcement through the CRTC1 pathway. We bilaterally injected AAV vectors that expressed CA-CRTC1 or GFP control in the NAcsh (Fig. 4A-C). Four weeks later, the rats were trained to self-administer cocaine (0.5 mg/kg/injection) in daily 3-h sessions until cocaine intake stabilized. After stabilization, the rats received the AMPK activator AICAR (2.5 µg per side) or vehicle infusions in the NAcsh immediately before four consecutive sessions to augment AMPK activity. A two-way repeated-measures ANOVA, with training session as the within-subjects factor and vector (GFP and CA-CRTC1) and AICAR dose (0 and 2.5 µg per side) as between-subjects factors, was performed to analyze cocaine self-administration behavior. The ANOVA of self-infusions (Fig. 4D) revealed effects of a vector × training session interaction (F13,429 = 2.731, p = 0.001), and an AICAR dose × training session interaction (F13,429 = 1.823, p = 0.038). The ANOVA of active nosepokes (Fig. 4E) revealed a significant effect of vector (F1,33 = 4.749, p = 0.037) and a vector × training session interaction (F13,429 = 1.859, p = 0.033). The expression of CA-CRTC1 in the NAcsh increased cocaine self-administration and rescued the decrease in cocaine intake in AICAR-treated rats.
In the between-session dose-response procedure, all cocaine-treated rats exhibited inverted U-shaped dose-response curves that were dose-dependent (dose effect: F4,132 = 75.821, p = 0.000; Fig. 4F). The repeated-measures ANOVA revealed significant effects of vector (F1,33 = 9.097, p = 0.005) and AICAR (F1,33 = 16.309, p = 0.000), a vector × dose interaction (F4,132 = 4.938, p = 0.001), an AICAR × dose interaction (F4,132 = 9.421, p = 0.000), and a vector × AICAR × dose interaction (F4,132 = 2.544, p = 0.043). The expression of CA-CRTC1 increased cocaine self-administration at threshold doses of 0.03 mg/kg/injection (p = 0.002; Fig. 4F) and 0.3 mg/kg/injection (p = 0.023; Fig. 4F) and blunted the decrease in cocaine self-administration at a threshold dose of 0.1 mg/kg/injection (p = 0.005, compared with CA-CRTC1 + AICAR group; Fig. 4F) that was induced by AMPK activation (p = 0.008, compared with GFP + vehicle group; Fig. 4F).
During the progressive-ratio test, the rats generally worked harder for the higher injection dose (dose effect: F1,33 = 210.281, p = 0.000). The repeated-measures ANOVA of breakpoints revealed significant effects of vector (F1,33 = 6.674, p = 0.014) and AICAR (F1,33 = 8.182, p = 0.007), a vector × dose interaction (F1,33 = 13.510, p = 0.001), and an AICAR × dose interaction (F1,33 = 20.601, p = 0.000). The expression of CA-CRTC1 increased breakpoints (p = 0.001; Fig. 4G) and blunted the decrease in breakpoint (p = 0.000, compared with CA-CRTC1 + AICAR group; Fig. 4G) that was caused by AMPK activation at the 0.75 mg/kg cocaine injection dose.
Another four groups of rats were given bilateral intra-NAcsh infusions of AAV-shCRTC1 or AAV-Scramble 4 weeks before the onset of cocaine self-administration (Fig. 5A). After stabilization, the rats received infusions of the AMPK inhibitor compound C (1.5 µg per side) or vehicle in the NAcsh immediately before four consecutive sessions to attenuate AMPK activity. A two-way repeated-measures ANOVA, with training session as the within-subjects factor and vector (Scramble and shCRTC1) and compound C dose (0 and 1.5 µg per side) as between-subjects factors, was performed to analyze cocaine self-administration behavior. The ANOVA of self-infusions (Fig. 5B) revealed effect of a vector × training session interaction (F13,455 = 1.869, p = 0.032). The knockdown of CRTC1 in the NAcsh decreased cocaine self-administration and blocked the increase in cocaine intake in compound C-treated rats.
In the between-session dose-response procedure, all cocaine-treated rats exhibited inverted U-shaped dose-response curves that were dose-dependent (dose effect: F4,140 = 84.494, p = 0.000; Fig. 5D). The repeated-measures ANOVA revealed significant effects of vector (F1,35 = 4.178, p = 0.049), a vector × dose interaction (F4,140 = 8.291, p = 0.000), and a compound C × dose interaction (F4,140 = 4.333, p = 0.002). The knockdown of CRTC1 decreased cocaine self-administration at a threshold dose of 0.1 mg/kg/injection (p = 0.026; Fig. 5D).
During the progressive-ratio test, the rats generally worked harder for the higher injection dose (dose effect: F1,35 = 86.072, p = 0.000; Fig. 5E). The repeated-measures ANOVA of breakpoint revealed significant effects of vector (F1,35 = 6.315, p = 0.017) and compound C (F1,35 = 6.550, p = 0.015), and a compound C × dose interaction (F1,35 = 14.730, p = 0.000).Breakpoints had no change after knockdown of CRTC1, whereas showed a downward trend (p = 0.091, compared with Scramble + vehicle group; Fig. 5E). And the knockdown of CRTC1 prevented the increase in breakpoints (p = 0.025, compared with shCRTC1 + comp C group; Fig. 5E) that was caused by AMPK inhibition at the 0.75 mg/kg cocaine injection dose.
Modulation of AMPK activity in the NAcsh has no effect on the motivation for a natural reward
We then investigated whether the regulation of cocaine reinforcement and motivation by AMPK generalizes to natural rewards. Rats were trained to self-administer sucrose solutions on an FR5 schedule and received a similar infusion of AICAR or compound C in the NAcsh. Infusions of AICAR or compound C in the NAcsh had no effect on the rate of sucrose self-administration over the next 4 days (supplementary Fig. 1A and B). The pharmacological activation or inhibition of AMPK in the NAcsh did not affect the number of sucrose reinforcers earned or breakpoints for sucrose on a progressive-ratio schedule (supplementary Fig. 1C and D).
AMPK regulates cocaine-induced dendritic spine density of neurons in the NAcsh
Chronic cocaine administration increased the density of dendritic spines on medium spiny neurons in the NAc, resulting in functional changes at excitatory glutamatergic synapses in these neurons [33–35], which, in turn, drove addictive behaviors [36]. Given that AMPK activity is also involved in structural plasticity [37], we investigated whether AMPK activity affects the number of spines of NAcsh medium spiny neurons. A two-way ANOVA, with condition (saline and cocaine) and vector (GFP, CA-AMPK, and DN-AMPK) as between-subjects factors, was performed to analyze spine density. We found significant effects of condition (F1,30 = 21.464, p = 0.000) and vector (F2,30 = 28.930, p = 0.000), and a marginally significant condition × vector interaction (F2,30 = 2.677, p = 0.085; Fig. 6A and B). Repeated cocaine administration increased total spine density on medium spiny neurons in the NAcsh (p = 0.002, compared with saline + Ad-GFP group), an effect that was completely blocked by Ad-CA-AMPK (p = 0.002, compared with cocaine + Ad-GFP group; Fig. 6B). Conversely, the overexpression of Ad-DN-AMPK potentiated the baseline (p = 0.036, compared with saline + Ad-GFP group) and cocaine-induced increase (p = 0.014, compared with cocaine + Ad-GFP group; Fig. 6B) in total spine density. These results indicate that AMPK regulated the behavioral response to cocaine through alterations of spine dynamics.