α-DB KO mice show increased responding in the PR schedule
To first determine the motivation of WT and α-DB KO mice for an appetitive reward, mice were tested on a PR task. Initial testing on a PR4 schedule found that KO mice showed significantly higher breakpoints than WT animals (U = 33, p = 0.04; Additional file 1: Figure S1A). However, we observed that 67% of WT and 100% of KO animals were still responding at the end of the PR4 session. Therefore, animals were transitioned systematically through PR8, PR12 and PR16 schedules to find a work requirement where the majority of mice stopped responding by the end of the session (Additional file 1: Figure S1B and C). α-DB KO mice showed significantly higher breakpoint than WT mice in PR12 (t(21) = 2.2, p = 0.04; Additional file 1: Figure S1B), however responding remained above 50% for KO mice (Additional file 1: Figure S1C). Persistent responding decreased to 33% and 42% of WT and KO mice, respectively, in PR16 which was deemed to be sufficiently demanding. α-DB KO mice had a higher breakpoint and greater number of completed trials in PR16 compared to WT animals, although this was not statistically significant due to higher spread of responses in this schedule (breakpoint: t(22) = 1.6, p = 0.13; trials: t(22) = 1.6, p = 0.13; Fig. 1a, b). Target touches/sec were significantly higher in KO vs. WT mice (genotype: F(1,43) = 7.1, p = 0.01, post hoc p = 0.003) and target vs. blank touches/sec were also higher in KO mice (touches: F(1,43) = 6.9, p = 0.01, post hoc p = 0.003), while WT animals made the same number of target and blank touches in PR16 (touches: F(1,43) = 6.9, p = 0.01, post hoc p = 0.93; Fig. 1c). A significant genotype x touch effect was also observed (F(1,43) = 4.7, p = 0.04). No differences were observed between animal groups in average latency to collect reward (t(20) = 1.1, p = 0.3; Fig. 1d) or magazine entries/sec (U = 48, p = 0.44; Fig. 1e). Both WT and KO mice made significantly more front beam breaks/sec compared to rear bream breaks/sec (breaks: F(1,22) = 386.1, p < 0.0001, all post hocs p < 0.0001), however no differences in beam breaks were noted between WT and KO animals (genotype: F(1,44) = 1.3, p = 0.27; Fig. 1f).
α-DB KO mice do not show alterations in executive function, but do exhibit some perseverative behaviour
To determine if the increased response rate of the KO animals in the PR schedule was due to perseveration [29], mice were then tested on the PVD task. As shown in Fig. 2, all animals successfully reached criteria on both learning and reversal. No difference was seen between WT and α-DB KO mice on the number of days to criteria in either the acquisition (t(22) = 0.27, p = 0.79) or reversal phase (U = 56, p = 0.79). The rate of learning and reward collection latency during acquisition and reversal was also similar between the animal groups (learning: χ2 (1) = 0.117, p = 0.74; latency: F(1,18) = 0.15, p = 0.75); Fig. 2c–e). Analysis of the number of below- and above-chance errors during the reversal phase showed a significant genotype x error interaction (F(1, 20) = 13.7, p = 0.001) and post hoc analysis showed that KO mice completed significantly fewer above-chance trials (p = 0.04) and showed a non-significant trend (p = 0.06) towards more below-chance trials than WT mice (Fig. 2f).
Increased response of α-DB KO in ERC schedule
To more fully explore the increased motivation of α-DB KO mice, the same animals were then tested in the ERC task. As expected, the number of completed trials decreased significantly in both the WT and KO groups as the work requirement to receive a reward increased across the ERC schedules (schedule: F(1.78, 39.1) = 60.1, p < 0.0001; genotype F(1, 22) = 7.5, p = 0.01, schedule x genotype: F(3, 66) = 4.9, p = 0.004; all post hocs p < 0.05; Fig. 3a, b). This corresponded with a significant increase in food consumption in the testing chamber in both groups (schedule: F(1.42, 31,2) = 36.5, p < 0.0001, all post hocs p < 0.05), and WT and KO animals ate equal amounts of food (genotype: F(1, 22) = 1.02, p = 0.32; Fig. 3c, d). In the ERC64 schedule, α-DB KO mice completed significantly more trials than WT animals (genotype: F(1, 22) = 7.5, p = 0.01, post hoc p = 0.03; schedule x genotype F(3. 66) = 4.9, p = 0.004; Fig. 3b). Moreover, whereas WT mice showed a progressive decrease in total touches between ERC16, ERC32 and ERC64 (schedule: F(1.33, 29.3) = 33.7, p < 0.0001, all post hocs p < 0.01 vs. ERC64), KO mice maintained a constant number of touches between all ERC schedules (p > 0.05 for all post hocs) and made significantly more touches than WT mice in ERC64 (genotype: F(1,22) = 15.2, p = 0.0008, post hoc p = 0.03; schedule x genotype: F(3,66) = 23.8, p < 0.0001; Fig. 3e, f).
To confirm that the decreased engagement with the ERC task was due to effortful choice, mice were then tested on an FR64 schedule without food pellets in the chamber. Both WT and KO mice completed significantly more trials (schedule: F(1.78, 39.1) = 60.1, p < 0.0001, all post hocs p < 0.01; Fig. 3a, b) and made more total touches in the FR64 task compared to ERC64 (schedule: F(1.33, 29.3) = 33.7, p < 0.0001, all post hoc p < 0.05; Fig. 3e, f). In addition, KO mice made more total touches (genotype: F(1, 22) = 15.2, p = 0.0008, post hoc p = 0.0009) and completed more trials than WT mice during FR64 (genotype: F(1,22) = 7.5, p = 0.01, post hoc p = 0.001), suggesting a higher motivation for the milkshake (Fig. 3b, f). No difference in latency to collect reward (genotype: F(1,16) = 3.77, p = 0.07), front beam breaks/sec (genotype: F(1,18) = 0.37, p = 0.55), rear beam breaks/sec (genotype: F(1,22) = 0.38, p = 0.54) or magazine entries/sec (genotype: F(1,21) = 0.77, p = 0.39) was observed between WT and KO in any of the ERC schedules (Additional file 2: Figure S2A-C).
Because α-DB KO mice maintained a consistent number of touches over the ERC schedules despite decreasing frequency of milkshake reward, mice were then progressed through an extinction protocol to determine if they would continue to engage with the task in the absence of appetitive stimuli. A significant decrease in the number of completed trials was observed by day 4 of extinction in both animal groups (time: F(5.02, 110.4) = 58.3, p < 0.0001, all post hocs p < 0.001 from day 4–12 vs. day 1), but no differences were noted between WT and KO mice on any day of the protocol (genotype: F(1,22) = 1.95, p = 0.18; Additional file 2: Figure S2D).
Fully fed α-DB KO mice show increased ad lib intake of milkshake
To determine if the increased motivation for the milkshake reward was due to a generally elevated appetite, food-restricted mice were allowed to freely consume food pellets, water or milkshake over a 60 min period. Both WT and KO mice drank significantly more milkshake than water (food type: F(3,87) = 177.6, p < 0.001, all post hocs p < 0.0001; genotype x food type: F(3,87) = 3.5, p = 0.02; Fig. 4a, b), but there were no differences between animal groups on food, water or milkshake intake (genotype: F(1,87) = 0.05, p = 0.83). To evaluate if food restriction was masking potential genotype differences, animals were then put back on ad lib feeding for 5 days, following which they were allowed access to milkshake for 24 h alongside food and water in their home cage. Daily food and water intake per cage was similar between WT and KO mice (food: F(1,5) = 0.65, p = 0.46; water: F(1,5) = 0.25, p = 0.64; Fig. 4c, d). However, α-DB KO animals drank significantly more milkshake than WT mice (genotype: F(1,65) = 4.5, p = 0.04, post hoc p = 0.0009; genotype x food type: F(2, 65) = 5.1, p = 0.009; Fig. 4e). Comparison of mouse body weight at the start and end of the experiment did not differ between animal groups (genotype: F(1,22) = 13.6, p = 0.22), although both WT and KO mice gained a significant amount of weight over the course of the experiment (time: F(1,22) = 236.9, p < 0.0001, all post hoc p < 0.0001; time x genotype: F(1,22) = 5.7, p = 0.03) and there was a non-significant trend (p = 0.07) for greater weight gain over the course of the experiment in the KO mice (Fig. 4f).
Levels of CB1 expression are altered in α-DB KO mice
To determine if the increased motivation in the α-DB KO mice was due to alterations in dysbindin-1 or in the DA, opioid and cannabinoid systems, mRNA levels of Dtnbp1, Scl6a3, Drd1a, Drd2, Oprm1 and Cnr1 were evaluated by RT-qPCR in the PfCtx, Cpu, nAc and/or tegmentum of WT and KO mice. As shown in Fig. 5a, no differences were observed in Dtnbp1 expression between mouse groups in any brain region (genotype: F(1,48) = 1.2, p = 0.27). Scl6a3 expression, which was only reliably detected in the tegmentum, did not differ significantly between mouse groups, although 4 of the 7 KO mice showed a > twofold increase in Scl6a3 mRNA levels (t(11) = 1,56, p = 0.15; Fig. 5b). As Drd1a mRNA expression was below the level of detection in the tegmentum of several animals, analysis of Drd1a was only performed in the PfCtx, Cpu and nAc. No difference was observed between WT and α-DB KO mice in Drd1a mRNA levels (genotype: F(1,32) = 1.1, p = 0.29; Fig. 5c). Similarly, Drd2 expression did not differ between mouse groups in any brain areas (genotype: F(1, 44) = 1.3, p = 0.24; Fig. 5d). Cnr1 mRNA levels were not significantly different between animal groups, although there was a trend towards increased expression in the PfCtx and decreased expression in the tegmentum of KO mice (genotype: F(1, 45) = 0.01, p = 0.9; tissue: F(3,45) = 0.06, post hoc p = 0.07; Fig. 5e). Oprm1 expression was similar between genotypes in the PfCtx and tegmentum, but levels were significantly decreased in the Cpu and nAc of α-DB KO compared to WT animals (genotype: F(1,40) = 8.0, p = 0.007; all post hocs p < 0.05; Fig. 5f).
Based on the RT-qPCR results, brain tissues were processed by Western blotting to determine if protein levels of DAT, mOR1 and CB1 differed between animal groups. Levels of DAT were similar between WT and KO animals in all brain regions examined (genotype: F(1, 44) = 2.1, p = 0.16; Fig. 6a). The anti-mOR1 receptor detected two strong bands at 60 and 90 kDa. Quantification of the 60 kDa band, which most closely matched the predicted molecular weight of mOR1 [30], detected no differences between WT and KO mice (genotype: F(1,47) = 3.7, p = 0.06; Fig. 6b). However, CB1 protein levels were significantly higher in the PfCtx and lower in the nAc of KO mice compared to WT animals (genotype x area: F(3, 46) = 5.38, p = 0.003; all post hocs p < 0.05; Fig. 6c).
No changes CB1 protein levels were observed in astrocytes
Although CB1 is predominantly expressed in neurons, the receptor is also expressed in astrocytes [31, 32]. Because astrocyte dysfunction is reported in α-DB KO mice, brain tissues were processed for triple-labelling immunohistochemistry of CB1, NeuN and GFAP to determine if CB1 expression was specifically altered in the astrocytes of α-DB KO mice. CB1 expression appeared punctate within fibers surrounding NeuN-positive cells throughout the cortex in a distinct pattern to that of GFAP expression (Fig. 7a–h). 3D reconstruction of the confocal images found that NeuN-positive neurons made direct contact with CB1, while little contact was observed between GFAP-positive astrocytes and CB1 in either WT or α-DB KO mice (Fig. 7I-L). This suggests that the observed alterations in CB1 levels were not specific to astrocytes, but likely due to changes in neuronal expression.