NAPE-PLD is functionally expressed in the brain and mediates motivational food-responses.
In order to explore the role of the NAEs-synthetizing enzyme NAPE-PLD in food-motivated behaviors, we first addressed the consequence of whole-body NAPE-PLD developmental knock-out in a food-reward seeking behavioral paradigm. Here, we used the Napepldf/f mouse line in which the two LoxP sites span the exon 3 (24, 30, 41), the gene sequence that encodes for the catalytic activity of the enzyme and that efficiently leads to a reduction of NAPE-PLD-derived bioproducts (18, 24, 30, 41). Napepldf/f mice were bred with mice expressing Cre under the pan-promoter phosphoglycerate kinase 1 (Pgk-Cre) which result in whole-body full knock-out (KO, (34)) of NAPE-PLD in subsequent generations. To study the reinforcing and motivational properties of food, we performed an operant conditioning paradigm where animals were trained, under different schedules, to press a lever to obtain a palatable sugar pellet. In both a fixed ratio 1 schedule (FR1, 1 lever press for 1 sugar pellet during 4 daily sessions) or a progressive ratio schedule (PR), which assesses the motivational component of reinforcement behaviors, Napepld+/+ (controls) and NapepldKO mice displayed similar performances (Suppl. Figure 1A, B). These results suggest that either physiological compensations occurred during development, as reported by previous genetic invalidations (27), and/or that, despite the expression of NAPE-PLD in the brain (23), brain NAPE-PLD plays a marginal role in reward-seeking behavior. To disentangle these two hypotheses, we moved to brain-restricted ablation of NAPE-PLD. Ablation of NAPE-PLD in the central nervous system (CNS) was achieved by crossing Napepldf/f mice with mice expressing Cre under the control of the promoter Nestin (NesCre+/− mice) (31, 32). We observed that CNS genetic deletion of NAPE-PLD was associated with an enhanced response to operant behavior. In fact, under FR1 schedule, NapepldΔCNS mice (males and females) collected a higher number of pellets and had a higher number of active lever presses (Fig. 1A, A1). However, this enhanced reward-like phenotype was not related to differences in learning (% of active lever over inactive lever) as both genotypes were characterized by very similar discrimination scores (Fig. 1A2). Once the operant conditioning established, mice were moved to the PR schedule. Again, we noticed that, despite similar learning scores, NapepldΔCNS mice showed enhanced performances (number of rewards and active lever presses) compared to control mice (Fig. 1B, B1, B2). To exclude that such phenotype was driven by the presence of Cre (NesCre+/− (32, 42)) rather than the proper deletion of NAPE-PLD, we performed the same behavioral battery in NesCre−/− (controls) and NesCre+/− mice which both displayed a very similar phenotype on this paradigm (Suppl. Figure 1C, D), thus indicating that genetic deletion of neuronal NAPE-PLD is responsible for the enhanced reward-behavior observed in NapepldΔCNS mice. This result revealed that tissue-specific ablation of NAPE-PLD generates different outcomes than whole-body gene deletion. This finding, which points to an effective role of brain NAPE-PLD in food-reward seeking behaviors, also raises the possibility that the contribution of NAPE-PLD in multiple organs (full KO mice) might lead to physiological adjustments eventually driving opposite consequences on a particular behavioral output with an overall mitigated consequence.
Among the main organs that might contribute to food-dependent reward processes, the gut has emerged as a critical modulator of reinforced behaviors (3, 36, 43, 44). It has been previously shown that mice with a specific and inducible deletion of NAPE-PLD in the intestinal epithelial cells (IEC) (NapepldΔIEC) exhibited a phenotype associated with specific changes in the homeostatic regulation of food intake and altered metabolic adaptations to high-fat diet (18, 45). Therefore, we explored the potential contribution of intestinal NAPE-PLD in reward-seeking behavior. Interestingly, NapepldΔIEC and control mice showed comparable performances in the operant conditioning paradigm (Suppl. Figure 1E, F), indicating that, while intestinal NAPE-PLD is critical for metabolic control (18) and short-term regulation of food intake (45), brain NAPE-PLD might represent a more direct target as acute regulator of food-reward behaviors.
Reinforced behaviors tightly depend on key brain regions that constitute the reward system, notably the midbrain dopamine (DA)-producing ventral tegmental area (VTA) and its dopaminoceptive structures, including the dorsal striatum (DS)/nucleus accumbens (NAc), the prefrontal cortex (PFC) and the hippocampus (Hippo) (46). Therefore, we first investigated whether NAPE-PLD-produced NAEs were altered within these structures in NapepldΔCNS mice. Lipidomic analyses revealed a significant decrease of several NAEs species [AEA, OEA, PEA, N-stearoylethanolamine (SEA), N-linoleoylethanolamine (LEA) and N-docosahexaenoylethanolamine (DEA)] in the midbrain VTA following CNS deletion of NAPE-PLD (Fig. 1C-H). Interestingly, either no major differences (DS/NAc and PFC), specific significant reductions (SEA, LEA and DEA for the hippocampus) or trends of decrease were detected in the levels of NAEs in the other reward-associated brain regions (Fig. 1C-H). Of note, no alterations were detected for the endocannabinoid (eCB) 2-AG (Fig. 1I). Moreover, within the VTA we did not detect alterations in the levels of other fatty acids (linoleic, arachidonic and oleic acids) (Suppl. Figure 2A-C) or N-acylamides (Suppl. Table 2), thus indicating that lack of NAPE-PLD specifically affects a subset of endogenous bioactive lipids.
In addition, lipidomic analyses also revealed that NAEs, but not 2-AG (Fig. 1I), levels were higher in the VTA compared to the DS/NAc, PFC and hippocampus (Fig. 1C-H).
These biochemical results, associated to the enhanced phenotype of NapepldΔCNS mice in the operant reward-driven behavior (Fig. 1A, B), prompted us to investigate the structure- and/or cell type-specific expression of NAPE-PLD in both rodents (rat and mouse) and human brains. First, by taking advantage of single-nucleus RNA transcriptomics (snRNA-seq) in the rat NAc (47) (GSE137763) and VTA (48) (GSE168156), we performed a clustering meta-analyses of transcripts encoding for eCBs- and NAEs-producing enzymes (Napepld, Dagla, Daglb) as well as for eCBs- and NAEs-related transducing effectors (Cnr1, Trpv1, Gpr119, Ppara, Pparg). Compared to Dagla and Daglb, we observed that accumbal Drd1- and Drd2-medium spiny neurons (MSNs) expressed low levels of Napepld (6% of 2819 Drd1-MSNs and 5% of 1993 Drd2-MSNs, respectively) (Fig. 2A, B). On the contrary, we detected a higher expression of Napepld in VTA-neurons (Fig. 2C-E). Since the VTA harbors different neuronal cell types (49), we restricted our meta-analysis to VTA DA-, GABA- and glutamate (Glut)-neurons. In the VTA, we again detected higher levels of Dagla and Daglb but, interestingly, we observed that Napepld was mainly present in VTA DA- and Glut-neurons [12% of 399 DA-neurons (Fig. 2C) and 20% of 698 Glut-neurons (Fig. 2E), respectively], whereas 9% of GABA-neurons (896 cells) were positive for Napepld (Fig. 2D). Interestingly, this pattern of Napepld expression mirrors the higher levels of NAEs observed in the VTA compared to the DS/NAc (Fig. 1C). Moreover, the transcriptomic profiling of eCBs- and NAEs-producing enzymes was in line with the meta-analysis of bulk VTA transcriptomics in the murine (50) (Slc6a3-bacTRAP mice, GSE64526, Fig. 2F) and human midbrains (51) (GSE114918, Fig. 2G).
Altogether these observations underline the potential role for NAPE-PLD in the midbrain VTA as a regulator of food-associated reward processes.
VTA NAPE-PLD scales food-motivated behaviors and dopamine releasing dynamics
To precisely interrogate the structure-specific functions of NAPE-PLD in driving food-motivated behaviors, we knocked-down the Napepld gene in the VTA using a local and virally mediated delivery of Cre in the VTA of Napepldf/f mice (Fig. 3A). Next, we tested the reinforcing and motivational properties of palatable food using a food-dependent operant conditioning paradigm. In line with the results obtained with NapepldΔCNS mice (Fig. 1A, B), we observed that viral deletion of NAPE-PLD in the VTA promoted food-operant conditioning (increased number of rewards and active lever presses) during both FR1 and PR schedules (Fig. 3B, C), with no differences in learning performances as both groups showed similar active/inactive discrimination index (Fig. 3B, C).
This enhanced reward phenotype was also present following a FR1→FR5→PR training schedule (Suppl. Figure 3A-C, food restriction) and even in sated conditions (Suppl. Figure 3D), therefore excluding the potentially confounding effect of hunger onto motivational drive. Importantly, this phenotype was also confirmed in female mice (Suppl. Figure 3E-H), again in both food-restricted and sated conditions. Of note, in both males and females, no significant differences in initial body weight and body weight loss (food restriction) were observed between experimental groups (Suppl. Figure 4A, B).
Aside from the motivational component, the liking and learning components of feeding are an integral part of food-reward processes (46). These components can be assessed through behavioral measurements of the positive valence assigned to palatable food in the conditioned-place preference (CPP, Fig. 3D) and T-maze (Fig. 3E) paradigms, which both rely on the association between reward value and context. In food-restricted conditions, we observed an increased and similar CPP score in both NapepldVTA−GFP and NapepldΔVTA mice (Fig. 3F). However, in sated conditions, only NapepldΔVTA mice showed an HFD-induced increase in CPP score (Fig. 3G), indicating enhanced susceptibility to the reinforcing properties of palatable foods. Using the T-maze paradigm, we next assessed the ability and flexibility of mice to actively learn in discriminating between a rewarded (HFD) and a non-rewarded arm. During the learning phase (first 5 days), we observed that both groups showed a progressive increase in correct responses (%) over training days, with NapepldΔVTA mice performing significantly better than NapepldVTA−GFP control mice (Fig. 3H). Then, mice were tested for their flexibility to relearn the task under a reversal learning schedule (in which the food reinforcer was switched to the previously unreinforced arm of the T-maze). While both groups displayed good performance in learning/flexibility, VTA-specific deletion of NAPE-PLD resulted in a better performance with a more rapid acquisition of the correct entry into the reinforced arm as compared to NapepldVTA−GFP control mice (Fig. 3H).
Next, we investigated the role of VTA NAPE-PLD in driving palatable food preference during a time-locked window (1h of exposure). First, we tested the reinforcing properties of the non-caloric sweetener sucralose (2 mM). As shown in Fig. 3I, NapepldΔVTA mice consumed more sucralose than NapepldVTA−GFP control mice. A very similar pattern of enhanced preference was measured with the natural caloric sugar sucrose (10%, Fig. 3J) and with emulsified lipids (20%, Fig. 3K).
We therefore decided to investigate whether the enhanced reward-like behavior observed in NapepldΔVTA mice was associated to an increased activity of the nucleus accumbens (NAc), a region highly innervated by VTA projections and whose activity is correlated with food-reward processes (46). However, the enhanced neural response within the reward system might result either from the higher tropism/consumption of palatable food of NapepldΔVTA mice or from the increased rewarding value despite a fixed amount of food-reinforcer. In order to dissociate these two possibilities, we exposed our experimental groups (sated conditions) to an equal amount of HFD during a time-locked window (1h during which all mice consumed the HFD pellet) and then we analyzed the induction of cFos, a molecular proxy of neuronal activity, in the NAc (Fig. 3L). Interestingly, we detected more cFos-positive neurons in the NAc of NapepldΔVTA mice (Fig. 3L), thereby indicating an enhanced responsiveness of the VTA→NAc mesolimbic axis to an equal amount of food-reward consumption.
These results led us to hypothesize that VTA NAPE-PLD and its local NAEs bioproducts may contribute to the regulation of DA dynamics within the VTA→NAc mesolimbic axis. To test this hypothesis, we took advantage of in vivo fiber photometry coupled to virally expressed DA biosensors (GRAB-DA2m (35)) to measure DA dynamics in the NAc of NapepldVTA−GFP and NapepldΔVTA mice (Fig. 4A, B). First, we observed that exposing both fasted (Fig. 4C, D) and ad libitum fed mice (Fig. 4E, F and Suppl. Figure 5A) to HFD triggered a higher DA accumulation/release in the NAc of NapepldΔVTA mice compared to control animals.
Second, to further explore whether and how NAPE-PLD may contribute to the regulation of DA-dependent events, we tested in vivo DA dynamics also in two non-food-dependent paradigms: the administration of cocaine (Fig. 4G, H) and the tail suspension (TS, Fig. 4I, J and Suppl. Figure 5B). In both cases, we observed that NapepldΔVTA mice were characterized by an enhanced accumulation/release of DA in the NAc than NapepldVTA−GFP mice. Lastly, we administered the selective DAT blocker GBR12909 and noticed an enhanced locomotor response in NapepldΔVTA mice compared to controls (Suppl. Figure 5C), further confirming an amplified DA release/tone as a consequence of VTA NAPE-PLD knock-down.
Overall, these results indicate that VTA NAPE-PLD tightly contributes in orchestrating the responses of midbrain DA-neurons to both food- and non-food-related reinforcers by promoting and boosting the release of DA at VTA→NAc synapses.
VTA NAPE-PLD contributes to the regulation of food intake and energy homeostasis
Although the regulation of energy homeostasis has been classically ascribed to the hypothalamus and brainstem (1), new evidence indicates that the reward system also strongly contributes in scaling whole-body metabolic functions (38, 52). We therefore explored the metabolic consequences of VTA NAPE-PLD knock-down in the regulation of whole-body metabolic efficiency and peripheral substrates utilization by using longitudinal measurements of indirect calorimetry. As previously observed (Suppl. Figure 4), no major differences were observed in body weight and body composition between the two experimental groups (Fig. 5A). However, NapepldΔVTA mice displayed a spontaneous increase in locomotor activity and in cumulative food intake compared to NapepldVTA−GFP mice during both the light and dark circadian phases (Fig. 5B, C). These phenotypes were associated with an overall enhanced energy expenditure (Fig. 5D) and to a change in peripheral substrates utilization favoring carbohydrates over lipids-based substrates as indicated by the increase in respiratory exchange ratio (RER, 1 = glucose substrate, 0.7 = lipids substrate) during the light phase (Fig. 5E) and the consequent decrease in fatty acid oxidation (FAO) in NapepldΔVTA mice during both the light and dark phases (Fig. 5F). This feature was also associated with enhanced glucose tolerance during an oral glucose tolerance test at the expense of lower insulin release, suggesting enhanced whole-body glucose dynamics and insulin sensitivity (Suppl. Figure 6).
We then decided to investigate how NapepldΔVTA mice adapted during manipulation of nutrients availability. We noticed that during a food deprivation period (overnight fasting) NapepldΔVTA mice were still characterized by increased locomotor activity (Fig. 5G) and energy expenditure (Fig. 5I), but with no differences in RER (Fig. 5J). While the capability to mobilize lipids-based substrates during the fasting-induced lipolysis was similar between the two groups, as indicated by the RER (Fig. 5J), the proportion of lipids used a primary source of fuel was enhanced in the fasting period as indicated by the FAO (Fig. 5K), thus indicating a metabolic shift toward lipids-based substrates utilization. Interestingly, upon refeeding, mice displayed similar food intake (Fig. 5H), locomotor activity (Fig. 5G) or substrates utilization (Fig. 5J, K), while a slight increase in energy expenditure was still detected in NapepldΔVTA mice (Fig. 5I). These results confirmed the hypothesis that the integrity of NAPE-PLD within the VTA was required for the proper metabolic adaptation to changes in nutrients availability.
Since fasting increases the motivational drive and responsiveness to food, we wondered whether the expression of NAPE-PLD was required to promote DA releasing dynamics in fasted mice exposed to a chow pellet. In contrast to the acute response to palatable HFD (Fig. 4C-F), consumption of a chow pellet resulted in similar DA releasing dynamics in both NapepldVTA−GFP and NapepldΔVTA mice (Fig. 5L, M). This may suggest that (i) the action of VTA NAPE-PLD in the modulation of adaptive metabolic responses to nutritional manipulations can be dissociated from DA release in the fast-refeeding transition and/or (ii) VTA NAPE-PLD plays an active role in discriminating between palatable (HFD, Fig. 4C-F) and regular (chow, Fig. 5L, M) foods through the control of reward-dependent DA dynamics.
VTA NAPE-PLD does not contribute to exercise-motivated behaviors but still regulates energy homeostasis
In mammals, exercise can function as a rewarding/motivational stimulus (53) and the eCBs system, especially within the VTA, has been identified as a key regulator of exercise-induced reinforced behaviors (54–56). We thus decided to extend our investigations to exercise-motivated behaviors (i) to investigate whether VTA NAPE-PLD was also important in mediating the reinforcing properties of exercise and (ii) to study whether metabolic adaptations observed in NapepldΔVTA mice (Fig. 5) were solely dependent on enhanced locomotor activity.
First, we performed a time-locked access (30 min session/day) to a running wheel. Despite both NapepldVTA−GFP and NapepldΔVTA mice progressively spent more time wheel-running, we surprisingly noticed a reduced performance in NapepldΔVTA mice compared to control animals (Fig. 6A, A1). This led us to investigate whole-body metabolism and metabolic efficiency in calorimetric chambers equipped with running wheels.
Again, we observed that NapepldΔVTA mice were characterized by an enhanced spontaneous locomotor activity (Fig. 6B, B1) and reduced wheel-running activity (Fig. 6C, C1). When combining both forms of activity (spontaneous + wheel running activities), we detected no differences in the light phase and a lower global activity in NapepldΔVTA mice during the dark phase (Fig. 6D). Of interest, the peculiar metabolic signature associated with VTA NAPE-PLD deletion also remained in this exercise-based paradigm and was characterized by enhanced energy expenditure (Fig. 6E, E1), food intake (Fig. 6F) and RER (Fig. 6G), and lower FAO (Fig. 6H) in NapepldΔVTA mice.
Altogether, these results indicate that the role of VTA NAPE-PLD in regulating reward-like processes cannot be generalized to all natural rewards (food vs exercise) and that the metabolic adaptations observed in VTA NAPE-PLD-deleted mice are not solely dependent on locomotor activity.
VTA NAPE-PLD controls metabolic adaptation to an obesogenic environment.
Food-reward drive, together with changes in metabolic outputs in response to food environment, are important contributors to the obesity pandemics. Given the above-mentioned results showing a key role of VTA NAPE-PLD in controlling reward and metabolic processes, we hypothesized that NAPE-PLD may influence the (mal)adaptive responses to an obesogenic environment. Thus, NapepldVTA−GFP and NapepldΔVTA mice were chronically exposed to an obesogenic diet (3 months of HFD) and then metabolically characterized.
First, we noticed no significant differences in the body weight and lean mass composition of both HFD-exposed experimental groups (Fig. 7A). However, fat body mass was significantly lower in NapepldΔVTA mice (Fig. 7A). The analysis of metabolic efficiency revealed that obese NapepldΔVTA mice displayed increased nocturnal locomotor activity (Fig. 7B) and enhanced nocturnal food intake (Fig. 7C, C1). Surprisingly, we detected a higher energy expenditure (Fig. 7D, D1) and FAO (Fig. 7E, E1) in obese NapepldΔVTA mice, whereas the RER resulted unchanged (Fig. 7F). This metabolic blueprint suggests that, depending on diets (chow vs HFD) and metabolic profiles (lean vs obese), VTA NAPE-PLD readily allows the plastic adaptation of nutrient partitioning (Fig. 5E-F vs Fig. 7E-F) in order to maintain a higher energy expenditure.
These results indicate that, within the VTA, the deletion of NAPE-PLD partially protects against diet-induced obesity.