Mice were anesthetized with pentobarbital, transcardially washed, and perfused with PBS (phosphate buffered saline: 136 mM NaCl, 2.7 mM KCl, 10 mM Na₂PO₄; pH 7.4) and 4% paraformaldehyde (PFA) in PBS solution, respectively. The isolated brains were post-fixed afterwards in 4% PFA/PBS overnight at 4°C. After incubation in 30% sucrose/PBS solution for 48 h, brains were frozen on dry ice and stored at −20°C until use. Next, 30-μm thick coronal brain slices were cut on a cryostat Microtome HM560 (Microm, Walldorf, Germany) and kept at 4°C in cryoprotection solution (25% glycerin, 25% ethylene glycol, 50% PBS) until use. Free-floating sections were rinsed from cryoprotection solution with PBS containing 0.2% Triton X-100 (3051.2, Roth, Germany) (PBS-TX) for 10 min, then pre-incubated for 15 min in blocking solution (4% goat serum in PBS) before incubating with the respective primary antibodies overnight at 4°C (mouse anti-TH: F-11; Santa Cruz Biotechnology Inc., CA, United States; 1:200; rabbit anti-NeuN: EPR12763, Abcam, Cambridge, United Kingdom). On the next day, the sections were washed three times for 10 min with PBS-TX and then incubated for 1 h with the appropriate Alexa488- or Alexa546-conjugated goat IgG (goat anti-rabbit Alexa-488 and donkey anti-mouse Alexa-546, Invitrogen, Carlsbad, CA, United States; 1:1000). Brain slices were incubated with DAPI (4’,6-Diamidino-2-Phenylindole, Dihydrochloride, D1306 Invitrogen, CA diluted 1:5000 in PBS) for 5 min to visualize cell nuclei. After a final washing step in PBS, sections were mounted with Mowiol 4–88 mounting medium (Roth, Karlsruhe, Germany) onto slides and dried overnight. Fluorescence labeling was visualized using a Leica DM5500 fluorescence microscope (Leica microsystems, Wetzlar, Germany). Images were captured by using the 20x objective.

Manual quantification of mesDA neurons in the VTA and SNc was performed by counting every sixth section of TH-positive cells between −2.54 mm and − 3.26 mm posterior to bregma (30-μm-slices x 4 analyzed sections x every 6^th section = 720 μm) along the anterior–posterior axis for each animal (EN1-CB1-KO n = 4; EN1-CB1-WT n = 4 and EN1-CRE n = 5; EN1-WT n = 4). Numbers were then extrapolated for the whole region by multiplying by 6, averaged for every genotype, and compared between groups.

The total number of neurons was determined by counting NeuN-positive cells in this 720 μm analyzed region (EN1-CB1-KO n = 3; EN1-CB1-WT n = 3 and EN1-CRE n = 2; EN1-WT n = 3). VTA and SNc were identified according to anatomical landmarks by the Paxinos mouse brain atlas (2008). For stereological cell counts, the LASX software was applied. Statistical analysis was conducted by a two-tailed unpaired Student’s t-test.

Double fluorescence in situ hybridization tests were carried out on coronal slices. To identify CB1/TH double-positive neurons, VTA and SNc were analyzed using fluorescein isothiocyanate (FITC) labeled riboprobes for CB1 and digoxigenin (DIG) labeled riboprobes for TH, and vice versa, when Gad65 (FITC labeled) was used to detect co-expressed CB1 (DIG labeled). VTA and SNc were analyzed in three EN1-CB1-KO mice and three EN1-CB1-WT littermates between −2.54 mm and − 3.26 mm posterior to the bregma. Adult mice were sacrificed by decapitation under deep isoflurane anesthesia. Brains were isolated, snap-frozen on dry ice, and stored at −80°C. Tissue Freezing Medium (Leica Biosystems) was used to mount the brains for sectioning, and 18 μm-thick coronal slices were cut from rostral on a cryostat Leica CM3050 S, dried on a 42°C heating plate, and stored at −20°C until use.

Riboprobes labeled with DIG and FITC were used. The DNA templates for the CB1 and TH probes were originally synthesized by reverse transcription polymerase chain reaction (RT-PCR) from whole mouse brain cDNA, as previously described (Marsicano and Lutz, 1999; Guimera et al., 2006). A list of the GenBank accession number, primer sequences, and probe length is reported there.

In vitro transcription of all three probes was performed for 3 h at 37°C in 20 μL of the linearized plasmid containing 2 μg of CB1, TH or Gad65 gene inserts. The following restriction enzymes were used for linearization and RNA polymerases: CB1 antisense: BamHI, T3; CB1 sense: EcoRI, T7; TH antisense: SacII, Sp6; TH sense: NotI, T7; Gad65 antisense: Bam HI, T3; Gad65 sense: Eco RI, T7. Pretreatment, hybridization, and imaging of the fluorescent in situ hybridization protocol were carried out as previously described (Zimmermann et al., 2018). The TH riboprobe was labeled with digoxigenin and used in the hybridization reaction over night at a final concentration of 80 ng/mL hybridization mix, while the CB1 riboprobe was labeled with FITC or DIG and applied onto the sections at a final concentration of 800 ng/mL. Gad65 was labeled with FITC and used at a final concentration of 300 ng/mL. Positive signals were detected by incubating hybridized sections with α-Digoxigenin-AP or α-Fluorescein-AP (Roche Diagnostics), followed by TSA Plus Cyanine and TSA Fluorescein amplification solution, respectively (Perkin Elmer/Akoya Biosciences).

Data are represented as the mean ± standard error of the mean (SEM). Graphs and statistics were generated by GraphPad Prism 7 (GraphPad Software Inc., La Jolla, CA, United States) and IBM SPSS Statistics 23 v software (IBM Corporation, Armonk, NY, United States). Unless otherwise described, data were analyzed using a two-tailed unpaired Student’s t-test. Results were considered as significant at p < 0.05.

The present study investigated whether CB1 deficiency in mesDA neurons beginning from E10.5, when precursor mesDA neurons start to be determined in the isthmic organizer (Smits et al., 2006), affects adult behaviors that are controlled by mesocorticolimbic and nigrostriatal pathways.

First, we investigated whether and to what extent the CB1 receptor is expressed in mesDA neurons in adulthood. The double fluorescent in situ hybridization (FISH) method was carried out on coronal midbrain sections of EN1-CB1-KO mice and EN1-CB1-WT control littermates to describe the presence of CB1 and TH mRNA in adult mesDA neurons. DA neurons in the VTA and SNc were identified by TH-expression and using the atlas of Paxinos and Franklin (2008). While the dorsal tegmentum of the midbrain and the hippocampus exhibited higher CB1 fluorescence signals, the ventral tegmentum contains densely packed, mostly low CB1 expressing cells (Figures 1A,C). In EN1-CB1-WT mice, intense TH signal is detected (Figures 1A,B), and CB1 co-expressing with TH was identified in a subset of VTA cells, exhibiting weaker CB1 fluorescence signals (Figures 1B–D, filled arrows). Most co-expressing neurons were identified in the lateral VTA, a DA (TH⁺) cell body rich zone, comprising the parabrachial pigmented nucleus and paranigral nucleus [according to Baik (2020) and Garritsen et al. (2023)]. No co-expression was detected in the SNc (not shown). In mesDA neurons of EN1-CB1-KO mutant mice, CB1 mRNA expression is lost in TH-positive cells (Figures 1E–G). Only CB1 expression in non-TH-positive cells was still detected (Figure 1F, empty arrows). High CB1 mRNA expressing cells within the VTA do not express TH (Figures 1B–D, empty arrows) being most likely GABAergic neurons as the VTA consists besides of DA neurons (65%) mainly of GABAergic (33%) neurons (Nair-Roberts et al., 2008; Morales and Root, 2014), which are known to express CB1 (Hill et al., 2007; Busquets-Garcia et al., 2018). Indeed, by an additional double FISH carried out on parallel sections with a CB1-DIG- and glutamic acid decarboxylase 65 (Gad65)-FITC-labeled riboprobe, we identified strong CB1 signals in GABAergic interneurons of EN1-CB1-WT mice (Figures 1H–J, filled arrows), while weak CB1 signals do not exhibit Gad65 signals (Figures 1H–J, empty arrows). In EN1-CB1-KO mice, most of the CB1 signal was attributable to a Gad65-expressing GABAergic neuron (Figures 1K–M, filled arrows). Moreover, the VTA was analyzed regarding co-expression of TH and Gad65 (Figures 1N–S). Indeed, a few cells are expressing both, TH and Gad65 (see arrows), revealing a cotransmitter phenotype of some VTA neurons.

Next, we needed to exclude a potential haploid effect of the En1^Cki/+ locus on a C57BL/6 J wild-type background (EN1-CRE), and we used C57BL/6 J littermates (EN1-WT) as controls. Sonnier et al. (2007) and Nordström et al. (2015) showed that heterozygous En1^+/− mice are born with a normal number of mesDA neurons, but display an age-dependent progressive cell loss reaching by 24 weeks of age a 38 and 23% reduction in the SNc and VTA, respectively. In both publications, heterozygous En1 mice with a lacZ insertion in the En1 locus were used (Hanks et al., 1995). In contrast, our study analyzed the heterozygous En1^Cki/+ mouse line with a Cre recombinase sequence insertion, replacing the first 111 amino acids of one En1 allele (Kimmel et al., 2000), and thereby possibly also compromising endogenous En1 gene function in the heterozygous state. For this reason, we analyzed 42-weeks-old mice using immunohistochemistry to determine whether En1^Cki/+ mice showed a loss of mesDA neurons (Figure 2). We compared not only EN1-CB1-KO mice and EN1-CB1-WT littermate controls but also EN1-CRE mice and EN1-WT littermates. By stereological counts of dopaminergic cells positive for TH, we evaluated the number of mesDA neurons in the VTA and SNc (Figure 2A). Moreover, we quantified the total number of neurons on the same sections by using the pan-neuronal nuclear marker NeuN (Figure 2B). Stereological counts revealed no genotype differences in the total number of TH-positive cells neither in the VTA nor in the SNc of both mouse lines (numbers shown in Figures 2E–H; p > 0.05, two-tailed unpaired Student’s t-test). Furthermore, the total number of NeuN-positive neurons was not altered either in the EN1-CB1-KO or EN1-CRE mice as compared to their respective controls (EN1-CB1-KO: 1725 ± 83.82; EN1-CB1-WT: 1861 ± 74.24, p = 0.2694; EN1-CRE: 1898 ± 144.5; EN1-WT: 1686 ± 58.09, p = 0.2049; two-tailed unpaired Student’s t-test). Hence, we conclude that neither the heterozygosity of the En1 gene in the here used En1^Cki/+ mouse line nor the ablation of CB1 causes degeneration of mesDA neurons.

We first evaluated the number of mesDA neurons in the VTA and SNc of 42 weeks old mice of the EN1-CB1-KO and EN1-CRE mouse lines. Stereological counts revealed no reduction of the total number of TH-positive cells neither in the VTA nor in the SNc of both lines. Hence, we conclude that neither the heterozygosity of the En1 gene nor the ablation of CB1 caused degeneration of mesDA neurons in adult mice. This was an important finding in the light that two studies previously showed that replacing the first 111 amino acids of one En1 allele with the lacZ gene leads to the degeneration of mesDA neurons in the SNc and the VTA (Sonnier et al., 2007; Nordström et al., 2015). However, our results agree with Sgadò et al. (2006) previous report, demonstrating that En1 heterozygous mice display an unchanged number of mesDA neurons in the VTA as compared to wild-type controls. Sonnier et al. (2007) argued that different genetic backgrounds might cause these discrepancies, as they used Swiss mice (Janvier, Le Genest-St.-Isle, France), and Sgadò et al. (2006) used C57BL/6 mice as we did in this study. Nordström et al. (2015) analyzed the mouse line of Sonnier et al. (2007) but maintained on an OF1 genetic background, and the results agreed with the findings of Sonnier et al. (2007). Thus, we conclude that the replacement of the first 111 amino acids of one En1 allele does not lead to a degeneration of mesDA neurons in mice per se, but degeneration might depend on the genetic background. Another possible explanation might be that the integrated gene is responsible for the degeneration of the mesDA neuron population. While the lacZ gene insertion in the En1 locus leads to a progressive decline of mesDA neurons (Sonnier et al., 2007; Nordström et al., 2015), the tau-LacZ (Sgadò et al., 2006) (its protein being not in the cell soma but mostly in axons) and Cre genes do not lead to a cell loss of VTA neurons. Yet, the expression of the lacZ-sequence is a widely used reporter tool to assess the expression of genes of interest in mice. However, once LacZ is activated, it is continuously expressed, leading to protein accumulation representing one of the hallmarks of neurodegenerative diseases (Braak and Braak, 1991; Kobayashi and Chen, 2005). LacZ expression in glutamatergic neurons in the cortex, for example, causes substantial deficits in hippocampus-dependent memory, as well as structural changes of, e.g., dendrite morphology and reduction of hippocampal volume. GFP expression in the same cell populations, on the other hand, does not result in deficiencies in cognition or structure (Reichel et al., 2016). Taken together, the findings of Reichel et al. (2016) might explain the effect of DA cell loss in En1^lacZ/+ mice, which is not present in the En1^Cre/+ lines.

In the present study, we identified CB1 in mesDA neurons in adult brain. Whereas no CB1 was observed in the SNc, neurons in the VTA co-expressing CB1 and TH (a marker for mesDA neurons in the ventral tegmentum) were detected. Consequently, in mesDA neurons of EN1-CB1-KO mutant mice, CB1 mRNA expression is lost. Given that CB1 protein, e.g., detected by ligand binding, or CB1 mRNA levels in mesDA neurons had been widely reported to be absent or their presence has been controversially discussed, it has been assumed that a direct effect of the ECS on DA cells seemed unlikely (Melis et al., 2012; Wenzel and Cheer, 2018). Thus, CB1 might indirectly modulate DA neurons via excitatory and inhibitory mechanisms of surrounding GABAergic neurons and glutamatergic long-range inputs. However, by going through studies, in situ hybridization, binding, and immunohistochemistry experiments from the 1990s and 2000s revealed both expression and localization of CB1 receptor in mesDA neurons. Hence, our findings agree with the study by Matsuda et al. (1993), reporting also low and scattered CB1 mRNA hybridization signals throughout the rat VTA. Furthermore, the missing CB1 expression in the SNc is in line with a previously described study in rat (Julian et al., 2003). Recently, another study showed CB1 expression in the mouse VTA and very low in the SNc using RNAscope (Han et al., 2023). Considering that the VTA dopamine-releasing neurons are heterogeneous in their afferent and efferent connectivity and, in some cases, release GABA or glutamate in addition to dopamine, no studies to date have demonstrated the direct function of CB1 receptor in a specific subset of VTA mesDA neurons. Interestingly, we identified most co-expressing neurons in the lateral VTA, comprising the parabrachial pigmented nucleus and paranigral nucleus of the VTA (pnVTA). Increasing evidence suggests heterogeneity in neuronal subtypes and anatomical localization in the VTA, including transmitter and neuropeptide systems that may modulate dopaminergic outputs (Jhou et al., 2009; Tan et al., 2012; Van Zessen et al., 2012; Morales and Margolis, 2017). However, the exact processes and regulatory pathways are still an unexplored area. Interestingly, neuropeptide nociceptin-expressing neurons were revealed in the paranigral nucleus (pnVTA) that, in turn, project to mesDA neurons in the VTA – a neurocircuit negatively regulating motivation and reward-seeking in mice (Parker et al., 2019). To date, we did not analyze whether CB1 expressing cells also express nociceptin, but we found in the supplementary data list (Parker et al., 2019) prepronociceptin and CB1 mRNA expressed in the pnVTA neurons (see below).

Collectively, our data support a unique role of the ECS in the mesolimbic system. While in most literature, a direct effect of the ECS on DA system in the VTA and SNc is assumed to be absent and research focuses on the GABAergic and glutamatergic afferents, we assume that it will be of great interest to determine whether CB1 acts also directly in mesDA neurons, in particular in the VTA. The ECS in the dopaminergic system could be responsible to withdraw the inhibition of reward seeking behavior conducted by stimulated nociceptin expressing pnVTA neurons, when endocannabinoids from DA neurons cause a retrograde inhibition upon stimulation by pnVTA neurons, which leads to a withdrawal of inhibiting the motivation and thereby a dampening of negatively controlling an exaggerating motivation. In EN1-CB1-KO mice, the suppression of inhibition is missing and the lower reward-seeking behavior is prominent. This could be an explanation for our results in the progressive ratio task. In fact, to corroborate the possibility of this mode of functioning, we found in the TRAP RNAseq data list that in pnVTA cells, besides prepronociceptin, also En1, TH and CB1 is expressed (Gene counts table from GEO, Accession: GSE 108813) (Parker et al., 2019). Thus, En1-Cre in these nociceptin positive neurons could delete CB1 expression, and the retrograde inhibition through CB1 on these motivation inhibiting neurons cannot take place in EN1-CB1 mutants, and for this reason they show less motivation.

Considering our histological data indicating CB1 loss in the ventral midbrain, we were interested whether the CB1 deletion in this area and in precursors of mesDA neurons leads to alterations in the behavior of adult EN1-CB1-KO mice. Therefore, we decided to investigate test paradigms addressing the nigrostriatal and mesolimbic systems. Impairments of the nigrostriatal pathway lead to motor dysfunctions, whereas the mesolimbic system is involved in motivation, reward, and addiction. Although we could not find a decline of mesDA neurons in adult mice, we performed all behavior tests with both mouse lines, EN1-CB1-KO and EN1-CRE, to exclude a behavioral phenotype due to the impairment of one En1 allele.

Concurrent with the histological outcomes, we could not find any differences in motor performance, as mice of both lines performed similarly in the OF and in the accelerating rotarod test. Furthermore, we could exclude a genotype-dependent decline of locomotion in older mice by OF and rotarod tests. Thus, neither the deletion of one En1 allele nor the CB1 ablation impairs locomotion or sensorimotor learning in young and old mice. In the OF test, the age-related increase in anxiety-like behavior, measured as decreased time spent in the center was observed in EN1-CB1-WT at 30 weeks of age as compared to 12 weeks of age. This is congruent with the work by Li et al. (2020), who reported increased anxiety-like behavior in 12 months old C57BL/6 J mice as compared to mice at 2 months of age. The age related increase of anxiety appears not to be present in 30 weeks old EN1-CB1-KO mice, nor is it apparent in the EN1-CRE line. At this point, we are not able to explain this phenotype, nor can we explain why it is not observed in the EN1-WT mice. In addition, there was no difference seen in the LD test regarding risk assessment behavior based on avoidance of light illuminated area. Thus, further investigations would be required to study in detail anxiety-like behavior upon CB1 loss in EN1-positive cells.

In the accelerating rotarod test mice improved in motor coordination, balance and sensorimotor learning without genotype differences and without showing an age effect. However, Sonnier et al. (2007) reported that En1^lacZ/+ mice have a deficit in motor coordination and sensorimotor learning in the rotarod and show an abnormal spontaneous motor activity in the OF test. Instead, the EN1-CB1-KO mice used in our study did not show a phenotype in motor capabilities. The reason why the nigrostriatal pathway related functions are not affected, is not clear but is congruent with our expression studies showing no CB1 mRNA expression in SNc DA neurons.

Since several studies showed that the ECS is also involved in the modulation of depression (Patel and Hillard, 2009), we were moreover interested in the impact of the CB1 ablation in the VTA in this regard. To date, there are inconsistencies regarding cannabis use observed in the human population: the use of cannabis may lead to the onset of depression, but on the contrary, depression may also lead to the onset or increase in cannabis use frequency. However, there is preclinical evidence that alteration in the ECS could potentially benefit patients suffering from depression (Feingold and Weinstein, 2021), but cannabis abuse is often associated with depression and bipolar disorder (El-Alfy et al., 2010). Therefore, the role of CB1 still remains elusive and needs to be further investigated. To further address the role of CB1 in the mesolimbic system, we performed the FST, a test originally developed in rats to investigate the effects of anti-depressant drugs, whereby the drug-induced increase in struggling and decrease in floating in the inescapable water bath was considered as the positive therapeutic effect. The FST has been widely used in preclinical research, but covers only a subset of possible pathological changes described in human depression, i.e., despair behavior in an inescapable situation, and, therefore gives a limitation to this test. Despite this limitation, the FST in the analysis of mutant mice is useful and can give us information about genotype differences in despair behavior and coping strategy. The term “depressive-like” behavior is often used, but behaviorally intends to describe “despair” behavior. Interestingly, our data demonstrate that EN1-CB1-KO mice show increased immobility and decreased latency to the first immobility compared to control mice. Hence, we conclude that CB1 deletion in En1-expressing cells leads to depressive-like behavior congruent with the results of, e.g., full CB1-KO mice (Steiner et al., 2008b). In contrast, mice lacking CB1 in principal forebrain neurons and in GABAergic neurons, behave like wild-type controls (Steiner et al., 2008a). However, interestingly mice lacking CB1 in cortical glutamatergic neurons show less despair behavior (Steiner et al., 2008a). Because of the fact that both pro- and anti-depressant effects have been reported, and that both, agonists and antagonists of the cannabinoid receptors, can act similarly to anti-depressants (Gorzalka and Hill, 2011), it is important to understand the underlying mechanisms of these complex interactions.

Another hallmark of depression is altered responsiveness to rewarding stimuli, such as impaired hedonic motivation (Willner, 1997). There is solid research suggesting that the ability to process rewarding stimuli requires ECS activity. CB1 deletion or blockade can attenuate the motivational, appetitive, and reinforcing effects of a diversity of rewarding stimuli, such as sucrose (Arnone et al., 1997; Freedland et al., 2000; Sanchis-Segura et al., 2004). Thus, the deficiency of ECS signaling may cause anhedonia. In contrast to full CB1-KO mice showing reduced responsiveness to sucrose (Sanchis-Segura et al., 2004), our set of experiments observed similar sucrose consumption in the two-bottle choice test of mutant and control mice, suggesting no anhedonic phenotype of EN1-CB1-KO mice. Thus, these processes seem to be independent of CB1 receptor expression in En1-expressing DA cells.

The ECS plays a critical role in central mechanisms controlling appetite and food reward. It is well-known that in humans, for instance, cannabis increases the consumption of palatable food (Abel, 1975). In Progressive Ratio (PR) paradigms, designed to evaluate food-related motivation, mice are offered a highly preferred reward that can only be obtained through the exertion of effort (Hodos, 1961). In our first PR paradigm, mice were food-deprived by reducing the initial body weight to 85% during the experiment to motivate animals participating in the task and thus, learning the nose poking response. As a result, we showed that the learning of operant conditioning maintained by chocolate-flavored pellets was impaired in EN1-CB1-KO and EN1-CRE mice as they both require more time to develop learning than their respective control littermates. Consequently, under food-deprived conditions, the deletion of one En1 allele has an impact on operant learning. Nevertheless, CB1 receptor in En1 expressing cells might also be implicated in conditioning learning, supported by the observation that the difference is even more prevalent in the EN1-CB1-KO mouse line.

While operant responding on an FR schedule is a powerful method to evaluate the ability to learn a response to obtain a reinforcer, it might not reflect the degree of motivation to work to obtain the reinforcer. Thus, we also analyzed the behavior in a PR reinforcement schedule, in which increasing responses are progressively required to obtain the reward. Indeed, EN1-CB1-KO mice showed a significantly reduced motivation compared to littermates, measured by the BP, whereas the motivation of EN1-CRE mice remained unaffected as compared to their control littermates. Hence, we can conclude that motivation decreases specifically by the CB1 ablation in mesDA neurons. To ensure that the decreased motivation is independent of differences in hunger levels triggered by the genotype, we furthermore measured the amount of food intake in food-deprived mice. Mice consumed the same amount of food, regardless of the genotype. However, since ECS signaling is critical for stress and food intake regulation (Matias et al., 2006; Steiner et al., 2008a; Kirkham, 2009), we conducted a second self-administration PR paradigm in which the mice were fed ad libitum. Indeed, the mice were not overfed, and operant training did not significantly modify body weight as it was maintained at least at about 95% of the initial body weight. Moreover, we switched the day/night cycles, and mice could perform in their active, i.e., dark phase. Interestingly, in contrast to the previous experimental conditions under food restriction, our data revealed no significant change in acquisition levels of operant learning in fixed ratio schedules between mutant and wildtype in both mouse lines. Furthermore, EN1-CB1-KO mice displayed a decreased reinforcer achievement compared to littermates during the PR session, comparable to food-restricted mice. In the EN1-CRE mouse line, no differences were obtained, congruent with food-restricted mice of the prior PR schedule. Thus, the decreased motivation seems to be solely attributed to the CB1 deletion. Moreover, we evaluated that food restriction was not necessary to produce cue-induced seeking in this mouse line.

It is interesting to note that the decreased breaking point number in the progressive ratio in EN1-CB1-KO as compared to EN1-CB1-WT resembles the pharmacological effect of the D2/D3 receptor antagonist raclopride, Heath et al. (2015) suggesting overlapping signaling pathways of these neuromodulatory systems in the regulation of reward behavior. Our findings also agree with a growing body of research demonstrating that the ECS is implicated in the regulation of positive reward and motivational aspects of highly palatable food (Rowland et al., 2001; Miller et al., 2004; Domingo-Rodriguez et al., 2020). Previous research suggests that exposure to highly palatable food and motivational attributes may induce the eCB tone in the limbic region (e.g., NAc), resulting in dopamine release and heightened rewarding effects after consumption. While THC has been shown to stimulate the intake of chocolate-containing food without having an impact on regular food intake (Koch and Matthews, 2001), the CB1 antagonist rimonabant reduces the consumption of a highly palatable chocolate-flavored drink and blocks palatable food-induced dopamine release in the NAc (Maccioni et al., 2008). Reports of full CB1-KO mice exhibiting reduced sensitivity to the motivating properties of food are in line with these studies as they show lower levels of responding for sweet food and achieving lower breaking points (Sanchis-Segura et al., 2004). Parker et al. (2019) reported a nociceptin circuit regulating motivation and reward-seeking in mice. As already mentioned, nociceptin is enriched in the pnVTA (pnVTA^Pnoc neurons), which projects to the lateral parts of the VTA. Using fiber photometry experiments, it has been shown that the pnVTA^Pnoc neurons are necessary to limit motivation to obtain rewards, and nociceptin and its receptor NOPR play a critical role in this process. While mice performed a PR test to obtain a sucrose reward, Parker et al. showed that the activity of pnVTA^Pnoc neurons was low when the reward was easy to obtain, whereas it increased with the number of nose pokes and was highest at the BP. Selective ablation, chemogenetic inhibition, and photoinhibition of pnVTA neurons showed that the number of nose pokes increased in the PR task. In contrast, photostimulation and chemogenetic stimulation of pnVTA^Pnoc neurons resulted in a reduction in the number of nose pokes and the rewards obtained during the PR task, which was again blocked by the administration of a NOPR antagonist. Administration of a selective NOPR agonist reduces the number of nose pokes in the PR task in wild-type mice but not in Nopr^−/− mice. Deletion of the receptor in mesDA neurons of the VTA resulted in an increased number of rewards in the PR task. In turn, selective reexpression of Nopr in neurons of Nopr−/− mice and administration of a NOPR agonist led to a lower number of rewards obtained in the PR task (Parker et al., 2019). CB1 might possibly be involved in the regulation of this system (see above).

In conclusion, our data demonstrate that the deletion of CB1 in En1-expressing cells was sufficient to reduce motivation for highly palatable food. Additionally, we showed that CB1 is not required for the extinction of the stimulus–response association in this appetitively motivated learning task, implying that distinct molecular pathways may govern the training, acquisition, and extinction phases. Future studies will be required to corroborate the function of CB1 in mesDA neurons and neural circuits implicated in food-related motivation.