Consumption of palatable food primes food approach behavior by rapidly increasing synaptic density in the VTA

Significance

Consumption of palatable food or food-related advertising can prime increased food intake, potentially leading to overeating. We show that short-term exposure to palatable foods induces long-lasting synaptic plasticity in mesolimbic dopamine neurons. Furthermore, short-term exposure to sweetened high-fat food can drive food approach behaviors and consumption days after the initial exposure. Suppressing excitatory synaptic transmission in the ventral tegmental area can reverse increased food approach behaviors and consumption. Targeting this circuit with brain-delivered insulin may provide a strategy to suppress food cravings.

Abstract

In an environment with easy access to highly palatable and energy-dense food, food-related cues drive food-seeking regardless of satiety, an effect that can lead to obesity. The ventral tegmental area (VTA) and its mesolimbic projections are critical structures involved in the learning of environmental cues used to predict motivationally relevant outcomes. Priming effects of food-related advertising and consumption of palatable food can drive food intake. However, the mechanism by which this effect occurs, and whether these priming effects last days after consumption, is unknown. Here, we demonstrate that short-term consumption of palatable food can prime future food approach behaviors and food intake. This effect is mediated by the strengthening of excitatory synaptic transmission onto dopamine neurons that is initially offset by a transient increase in endocannabinoid tone, but lasts days after an initial 24-h exposure to sweetened high-fat food (SHF). This enhanced synaptic strength is mediated by a long-lasting increase in excitatory synaptic density onto VTA dopamine neurons. Administration of insulin into the VTA, which suppresses excitatory synaptic transmission onto dopamine neurons, can abolish food approach behaviors and food intake observed days after 24-h access to SHF. These results suggest that even a short-term exposure to palatable foods can drive future feeding behavior by “rewiring” mesolimbic dopamine neurons.

Priming effects of food-related advertising (1) and consumption of palatable food (2, 3) can drive sated consumption of food. Overconsumption of food relative to need is an important factor in the development of obesity (4). The priming effects of palatable food on food intake are likely a considerable factor in the obesity epidemic. However, the mechanism by which food-priming effects occur is unknown. The ventral tegmental area (VTA) and its mesolimbic projections are critical for learning about environmental cues used to predict motivationally relevant outcomes (5). Indeed, food-related cues activate the dopaminergic circuit to reinforce food intake (6, 7). VTA dopamine neurons can increase or decrease their synaptic efficacy to modulate their consequent dopaminergic output (8, 9). The strength of excitatory synaptic input onto VTA dopamine neurons plays a central role in reward-related behavior with potentiation of synapses formed onto VTA neurons, facilitating the transformation of neutral environmental stimuli into salient reward predictive cues (8). Conversely, depression of excitatory synaptic transmission would likely reduce the intrinsic firing rate and excitability of dopamine neurons, as has been demonstrated for other neurons (10), and would likely reduces the salience of reward predicting cues.

The motivation to eat is regulated by a variety of intrinsic and extrinsic factors. Metabolic signals, including neuronal or circulating peptides released in response to internal states such as hunger or satiety, can promote or inhibit food intake, respectively. Indeed, our previous work has demonstrated that insulin induces a long-term depression of excitatory synaptic transmission onto dopamine neurons via an endocannabinoid-mediated mechanism (11). Furthermore, insulin in the VTA suppresses anticipatory activity and conditioned place preference for food rewards (11). Thus, dopamine transmission may provide one mechanism that bridges internal states signaled by peripheral peptides with goal-directed behaviors (12).

Because dopaminergic neurons are sensitive to metabolic signals (13) and food anticipatory responses are accompanied by a significant increase in dopamine concentration in target regions of the VTA (14), we considered whether VTA dopamine neurons are influenced by the priming effects of palatable food consumption. Here, we tested the hypothesis that short-term consumption of palatable food can modulate excitatory inputs to VTA dopamine neurons to drive future food approach behaviors and food intake.

Results

To test whether short-term access to sweetened high-fat food (SHF) can prime food-seeking behavior 2 d later, we placed mice in a light/dark conflict box containing SHF in the light compartment (Fig. 1 A and B). Here, mice must enter an aversive space to seek palatable food. Mice with 24-h SHF exposure traveled significantly more in the food zone [t₍₁₄₎ = 2.30, P = 0.0096; Fig. 1C], had decreased latency of the first exit from the dark box [t₍₁₄₎ = 2.28, P = 0.039; Fig. 1D], and had increased entries to the food zone [t₍₁₄₎ = 2.16, P = 0.04; Fig. 1 E and F], but only when food was present [F_(1,28) = 8.86, P = 0.0060; Fig. 1F and Movies S1 and S2]. Time spent, distance traveled, or average speed in the light compartment was not significantly different between mice exposed to 1- or 24-h SHF (SI Appendix, Fig. S1), suggesting that 24-h SHF does not influence anxiety or locomotor activity. Finally, mice with prior access to 24-h SHF consumed more food after the test [F_(1,28) = 6.40, P = 0.017; Fig. 1G]. Consistent with these results, mice showed elevated anticipatory behavior and food consumption 2 d after 24-h SHF (SI Appendix, Fig. S2). Taken together, these data indicate that mice have increased food approach behaviors and consumption 2 d after the 24-h access to SHF.

Fig. 1.

The 24-h SHF exposure primes future food approach behaviors and consumption. (A) Mice were preexposed to a pellet of SHF (1–5 g) to prevent food neophobia, and 3 d later were given 24- or 1-h unlimited SHF in their home cages and then returned to chow (RF). (B) Two days after access to SHF, mice were placed in a light/dark conflict box containing a pellet of SHF in the center of the “food zone” for 10 min. (C–E) Mice with 24-h SHF exposure traveled more in the food zone (C), had decreased latency to the first exit from the dark box compared with mice with 1-h SHF exposure (D), and had increased food zone entries (E). (F) Entries to the food zone were significantly increased 2 d after mice had access to 24-h SHF compared with 1-h SHF only when the food was present. (G) Mice with 24-h access to SHF consumed more food after the test than mice with 1-h SHF. n, number of mice. Bars represent mean ± SEM. *P < 0.05; **P < 0.01.

Because insulin in the VTA can suppress conditioned place preference and anticipatory responses for food (11), we assessed whether insulin signaling in the VTA could inhibit the enhanced food approach behaviors observed 2 d after 24-h access to SHF. Two-way ANOVAs revealed a significant treatment (vehicle vs. insulin) × group (1 vs. 24 h) interaction on distance traveled in the food zone [F_(1,35) = 8.49, P = 0.006; Fig. 2A], latency to the first exit [F_(1,35)= 5.03, P = 0.03; Fig. 2B], food zone entries [F_(1,35) = 4.24, P = 0.04; Fig. 2C], and food consumed after the test [F_(1,35) = 5.72, P = 0.03; Fig. 2D]. These data suggest that intra-VTA insulin abolished the increased food-approach behaviors observed 2 d after 24-h SHF. Insulin in the VTA did not alter basal locomotor activity (SI Appendix, Fig. S3). However, intra-VTA insulin significantly suppressed cocaine-evoked locomotor activity (SI Appendix, Fig. S4), suggesting that intra-VTA insulin suppresses dopaminergic output. Insulin in the VTA suppresses the priming effect of 24-h exposure to SHF on food approach behaviors and food consumption, likely by decreasing dopamine output.

Fig. 2.

Intra-VTA insulin suppresses enhanced food approach behaviors and food consumption days after access to 24-h SHF. Mice were fed 1- or 24-h SHF. Two days later, insulin or vehicle was microinjected in the VTA 15 min before placing animals in the light/dark conflict box. (A) A Sidak’s multiple comparison test revealed a significant increase in distance traveled in the food zone 2 d after 24-h SHF in intra-VTA vehicle-treated mice (P < 0.001), but no significant effect when insulin was delivered intra-VTA (P > 0.05). (B) Latency to exit the dark box was significantly shorter in vehicle-injected mice given 24-h SHF (P < 0.01), but not significantly different in insulin-injected mice given 24-h SHF (P > 0.05). (C) Entries to the food zone were significantly greater in vehicle-injected mice given 24-h SHF (P < 0.01), but not significantly different in insulin-injected mice given 24-h SHF (P > 0.05). (D) SHF consumed after the test was significantly greater in vehicle-injected mice previously given 24-h SHF (P < 0.001), but not significantly different in insulin-injected mice previously given 24-h SHF (P > 0.05). n, number of mice. Bars represent mean ± SEM. ***P < 0.001; **P < 0.01; *P < 0.05.

Because insulin transiently suppresses excitatory, but not inhibitory, synaptic transmission on to VTA dopamine neurons (11), we next determined whether insulin reduces excitatory synaptic transmission onto dopamine neurons (SI Appendix, Fig. S5) 2 d after a 24-h SHF exposure. We used regular-food (RF) mice as a control for 24-h SHF access, because we previously found that 1-h SHF induces a transient (<3 h) elevation in endocannabinoid tone and a depression of excitatory synapses (11). Further, food-approach behaviors were increased 2 d after 24-h SHF compared with RF (SI Appendix, Fig. S2) or 1-h SHF (Fig. 1). There was a significant interaction of insulin × diet exposure on miniature excitatory postsynaptic current (mEPSC) frequency [F_(1,17) = 5.407, P = 0.027; Fig. 3 A and B], but not on mEPSC amplitude [F_(1,17) = 0.712, P = 0.41; Fig. 3 A and C], of dopamine neurons from RF mice. Notably, there was a significant increase in mEPSC frequency 2 d after 24-h access to SHF (P < 0.05; Fig. 3B), which was inhibited by insulin (P < 0.05; Fig. 3B). Furthermore, the effect of insulin was significantly greater in mice that had 24-h access to SHF compared with RF [t₍₁₇₎ = 2.3, df = 17, P = 0.32; Fig. 3D]. Insulin did not modulate miniature inhibitory postsynaptic currents (mIPSCs) frequency or amplitude onto VTA dopamine neurons (SI Appendix, Fig. S6). Taken together, these results indicate that mEPSC frequency is elevated 2 d after 24-h SHF and that insulin signaling in the VTA can suppress excitatory synaptic transmission onto dopamine neurons and can inhibit increased food approach behaviors days after SHF exposure.

Fig. 3.

Insulin suppresses excitatory synaptic transmission onto VTA dopamine neurons from mice exposed to RF or 24-h SHF. Two days after 24-h SHF, mice were decapitated, and midbrain slices were prepared for whole-cell patch-clamp electrophysiology. mEPSCs were recorded 10 min before and 20 min after insulin (500 nM) application. (A) Example recordings of mEPSCs in the presence (Lower) or absence (Upper) of insulin from dopamine neurons of mice with access to 24-h SHF (Right) or RF (Left). (B) Insulin significantly suppressed mEPSCs in mice with access to 24-h SHF or RF. A Sidak’s multiple comparison test revealed a significant effect of insulin on mEPSC frequency on mice exposed to 24-h SHF (P < 0.001) or RF (P < 0.05). (C) Insulin did not significantly modulate AMPAR mEPSC amplitudes recorded from dopamine neurons of mice exposed to 24-h SHF or RF. (D) The effect size was significantly greater in mice exposed to 24-h SHF compared with RF (P < 0.05). Bars represent mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001.

To determine whether the 24-h SHF-induced increase in excitatory synaptic strength observed 2 d after exposure was long-lasting, we measured excitatory synaptic transmission onto dopamine neurons 2 and 7 d after 24-h SHF exposure. There was a main effect of SHF on mEPSC frequency [F_(1,43) = 9.5, P = 0.004; Fig. 4 A and B], but not on amplitude [F_(1,43) = 1.15, P = 0.29; Fig. 4 A and C], such that mEPSC frequency was significantly increased 2 d (P < 0.05) and 7 d after 24-h SHF (P < 0.05; Fig. 4 A and B). There was no effect of diet on the paired pulse ratio, a measure that correlates with release probability (15), 2 or 7 d after the feeding exposure [F_(1,34) = 0.059, P = 0.81; Fig. 4 C and D], suggesting that there is an increase in release sites without a change in release probability.

Fig. 4.

The 24-h access to SHF induces a long-lasting increase in glutamate release onto VTA dopamine neurons. (A) AMPAR mEPSCs were recorded from VTA dopamine neurons 2 or 7 d after 24-h SHF or RF exposure. A Sidak’s multiple comparison test revealed a significant effect of SHF on mEPSC frequency at 2 (P < 0.05) and 7 (P < 0.05) d after the initial 24-h SHF exposure. (Upper) Example recordings of mEPSC events from dopamine neurons recorded 2 d after RF or SHF. (B) mEPSC amplitude was not significantly different 2 or 7 d after the initial 24-h SHF exposure. (Upper) Example recordings of mEPSC events from dopamine neurons recorded 7 d after RF or SHF exposure. (C) The paired pulse ratio was not significantly different between RF and SHF mice at 2 or 7 d after 24-h SHF exposure. (D) Sample paired pulse ratio recordings from dopamine neurons of RF (Left) or SHF (Right) mice 2 d (Upper) or 7 d (Lower) after access to SHF or RF. Bars or symbols represent mean ± SEM. *P < 0.05.

Our previous work demonstrated that immediately after 1-h access to SHF, mice have increased insulin levels and increased endocannabinoid tone at glutamate synapses onto dopamine neurons (11). Because plasma insulin levels in 24-h SHF mice were significantly increased compared with RF mice (P < 0.05; SI Appendix, Fig. S7), we measured whether excitatory synaptic transmission immediately after the 24-h SHF exposure was modified by endocannabinoids. Application of the cannabinoid 1 receptor (CB1R) antagonist AM251 increased the maximal effect of evoked AMPA receptor (AMPAR) EPSCs only in SHF mice [RF: 95 ± 4% of baseline; SHF: 121 ± 5% of baseline, t₍₅₎ = 3.22, P = 0.022; Fig. 5A]. Further, the paired-pulse ratio was increased in dopamine neurons of 24-h SHF, but not RF, mice, an effect reversed with application of AM251 [one-way ANOVA, F_(2,27) = 5.05, P = 0.014; Fig. 5B]. This effect was transient, because AM251 did not alter mEPSC frequency, amplitude, or paired pulse ratio 2 d after 24-h SHF exposure. (SI Appendix, Fig. S8). Together, these results indicate that immediately after 24-h SHF, endocannabinoid tone is transiently increased resulting in reduced release probability at glutamatergic synapses onto VTA dopamine neurons.

Fig. 5.

After 24-h SHF, increased endocannabinoid tone offsets increased glutamate release. Immediately after 24-h access to SHF, mice were decapitated, and midbrain slices were prepared for whole-cell patch-clamp electrophysiology. (A) Application of AM251 (2 µM) significantly increased AMPAR EPSCs in mice exposed to 24-h SHF (filled circles) compared with RF (open circles). (Upper) example traces of evoked AMPAR EPSCs onto VTA neurons from mice with 24-h access to SHF (filled circle) or RF (open circle). (B) Paired pulses were evoked with a 50-ms interstimulus interval (ISI). The paired pulse ratio was significantly greater in VTA neurons from mice with 24-h SHF compared with neurons from RF mice or VTA neurons treated with AM251 (2 µM) from SHF mice. (Upper) Example traces of AMPA EPSCs from VTA dopamine neurons from mice with 24-h access to RF (Left), SHF (Center), or SHF + AM251 (Right). (C) A Sidak’s multiple comparison test revealed mEPSC frequency was not significantly different between vehicle-treated slices of RF or SHF mice (P > 0.05). However, in AM251, mEPSC frequency was significantly greater in VTA neurons from SHF mice (P < 0.05). (Upper) Example traces of mEPSCs from VTA neurons from RF or SHF mice. (D) There was no significant effect of diet on mEPSC amplitude in the presence or absence of AM251 (P > 0.05). (Upper) Example traces of mEPSCs from VTA neurons treated with AM251 (2 µM) from RF or SHF mice. n/N, number of cells/number of mice. Bars or symbols represent mean ± SEM. *P < 0.05.

In contrast to elevated mEPSC frequency that was observed 2 d after 24-h SHF (Fig. 3A), immediately after food exposure, increased mEPSC frequency was only observed with application of AM251 (Fig. 5C). There was a significant interaction of treatment × diet on mEPSC frequency [F_(1,42) = 4.17, P = 0.047], suggesting that increased glutamate release is offset by increased endocannabinoid tone after 24-h SHF. Furthermore, there was no significant difference in mEPSC amplitude between dopaminergic neurons from RF or SHF mice in the presence or absence of AM251 confirming a presynaptic mechanism [F_(1,42) = 1.634, P = 0.21; Fig. 5D]. We found that 24-h exposure to SHF induces an increase in glutamate release that is offset by an increase in endocannabinoid tone acting at CB1Rs to reduce the probability of release.

One possible mechanism that could account for increased mEPSC frequency and no change in release probability in the absence of endocannabinoid tone is an increase in glutamate release sites on VTA dopamine neurons of SHF mice. To test this hypothesis, we quantified the density of asymmetrical (presumably excitatory) or symmetrical (presumably inhibitory) synapses being formed onto dopaminergic neurons of the VTA immediately after 24-h exposure to RF or SHF by using immunogold electron microscopy. We used immunogold beads of different sizes to identify dopamine transporter (DAT)-positive neurons (25-nm beads, dopaminergic neurons) and postsynaptic density 95 (PSD-95)–positive synapses (15-nm beads, excitatory synapses), allowing for the reliable identification of excitatory inputs onto dopaminergic synapses. There was a significant interaction of diet (RF vs. SHF) × cell type (DAT vs. no DAT) on excitatory synaptic density [F_(1,12) = 27.41, P = 0.0002]. The number of PSD-95–positive synapses formed onto VTA dopamine neurons, but not nondopaminergic neurons, significantly increased immediately after 24-h exposure to SHF (P < 0.001; Fig. 6 A–C). There was no interaction of diet × cell type on inhibitory synapses [F_(1,12) = 1.11, P = 0.312]. However, there was a significant increase in inhibitory synapses onto nondopaminergic neurons compared with dopaminergic neurons, regardless of diet [F_(1,12) = 16.20, P = 0.002]. Furthermore, there was no significant difference in the proportion of GluA1 at the active zone of dopamine neurons in SHF compared with RF mice [t₍₄₎ = 0, P > 0.99; Fig. 6 D–F], suggesting that, although there was increased density of glutamatergic synapses, there was no change in the number of AMPARs per synapse. The 24-h SHF did not change the average PSD length of asymmetrical or symmetrical synapses onto dopaminergic or nondopaminergic neurons (SI Appendix, Fig. S9). We found that 24-h SHF induces a rapid and long-lasting increase in glutamate release sites that is initially offset by elevated endocannabinoid tone.

Fig. 6.

The 24-h SHF induces increased excitatory synapses onto dopamine neurons. Electron micrographs were prepared from the VTA of RF and SHF mice (n = 4 mice per group, 1,250- to 1,400-µm² area per mouse). (A) The 24-h SHF induced a significant increase in excitatory synapses onto VTA dopamine neurons (P < 0.001), but it did not change excitatory inputs to nondopaminergic neurons (P > 0.05). (B) There was no significant effect of diet on symmetrical synapse number. (C) A representative electron micrograph demonstrating excitatory (immunogold-labeled PSD-95; open arrow) and inhibitory (filled arrow) synapses onto a dopaminergic neuron (immunogold-labeled DAT; blue/asterisks). Shaded arrow represents inhibitory synapse onto a nondopaminergic neuron. (Scale bar, 500 nm.) (D) Percent GluA1 at active zone. (E) The 24-h SHF did not alter GluA1 at the membrane (filled bars) and in recycling pools (open bars) of excitatory synapses onto dopamine neurons from RF (Upper) and SHF (Lower) mice (n = 3 mice per group, >100 synapses per condition). (F) Representative electron micrograph of VTA synapse from a RF (Upper) or SHF (Lower) mouse showing immunogold-labeled DAT (asterisks) and GluA1 (arrows). (Scale bar, 70 nm.) Bars or symbols represent mean ± SEM. ****P < 0.001.

We tested whether there was a change in food approach behaviors immediately after 24-h SHF. We found no significant effect of SHF on distance traveled in the food zone, latency of the first exit from the dark box, and entries to the food zone measured immediately after 24-h SHF (SI Appendix, Fig. S10). However, we found a significant reduction in SHF consumed after the test, likely due to sensory-specific satiety (16) (SI Appendix, Fig. S10). Presumably, there was no effect of diet on food-approach behaviors because endogenous insulin from the meal suppressed the elevated excitatory input. Therefore, we tested this hypothesis by administering the insulin receptor antagonist, S961, to the VTA immediately after 24-h SHF and 10 min before the light/dark box access (SI Appendix, Fig. S11). There was significant treatment (S961 vs. vehicle) × diet (1- vs. 24-h SHF) interaction on distance traveled in the food zone [F_(1,21) = 7.18, P = 0.014; Fig. 7A] and food zone entries [F_(1,21) = 4.88, P = 0.038; Fig. 7C]. Furthermore, there was a main effect of S961 on latency to exit the dark box [F_(1,21) = 11.82, P = 0.0025; Fig. 6B]. However, there was no effect of S961 on food intake after the test [F_(1,21) = 1.015, P = 0.32, Fig. 7D]. To test whether endogenous insulin indeed counteracts the plasticity induced by 24-h SHF, we repeated the experiments in streptozocin-treated mice, a procedure that ablates insulin-producing β-cells (ref. 17; SI Appendix, Fig. S12 and SI Methods). mEPSC frequency [t₍₁₆₎ = 2.4, P = 0.028] and amplitude [t₍₁₆₎ = 2.27, P = 0.037] were significantly increased immediately after 24-h SHF (Fig. 7 E and F). Taken together, these data indicate that relief of endogenous insulin action in the VTA immediately after food consumption enables food approach behaviors and increases excitatory synaptic transmission, but does not relieve the sensory-specific satiety.

Fig. 7.

After 24-h SHF, inhibition of insulin signaling enabled food approach behaviors and plasticity. Mice were fed 1- or 24-h SHF. Immediately after, S961 or vehicle was microinjected in the VTA 15 min before placing animals in the light/dark conflict box. (A) A Sidak’s multiple comparison test revealed a significant increase in distance traveled in the food zone immediately after 24-h SHF in intra-VTA S961-treated mice (P < 0.05), but no significant effect with vehicle (P > 0.05). (B) Latency to exit the dark box was significantly shorter in S961-injected mice given 24-h SHF (P < 0.01), but not vehicle (P > 0.05). (C) Entries to the food zone were significantly greater in S961-injected mice given 24-h SHF (P < 0.05), but not vehicle-injected mice given 24-h SHF (P > 0.05). (D) SHF consumed after the test was significantly less in vehicle- or S961-injected mice previously given 24-h SHF (P < 0.001). n, number of mice. (E) Immediately after SHF exposure in streptozotocin (STZ)-treated mice, mEPSC frequency was significantly greater (P < 0.05). (F) mEPSC amplitude was significantly different between RF and SHF mice (P < 0.05). n/N, number of cells/number of mice. (G) Example traces of mEPSCs from VTA neurons of STZ-treated mice exposed to RF or SHF. Bars represent mean ± SEM. ***P < 0.001; **P < 0.01; *P < 0.05.

Discussion

Here, we demonstrated that a short-term exposure to SHF primes food approach and food consumption days after the initial exposure. The 24-h access to SHF rapidly increases synaptic density and excitatory synaptic transmission onto dopamine neurons lasting at least a week. This increase in synaptic density was initially offset by a transient increase in endocannabinoid tone at glutamatergic synapses onto dopamine neurons, an effect likely mediated by insulin action in the VTA. Furthermore, intra-VTA insulin blocked the primed food approach behaviors and food consumption (SI Appendix, Fig. S13).

The 24-h SHF “Primes” Future Food Approach Behaviors and Consumption.

Priming effects, whereby initial exposure to food-related cues or food consumption influences food intake, have been observed in humans and rodents (2, 3, 18, 19). Consistent with this finding, we observed a priming effect of 24-h SHF, but not 1-h SHF, on food approach behaviors and food consumption 2 d after the exposure. Previous work has demonstrated that food anticipatory responses are accompanied by a significant increase in dopamine concentration in target regions of the VTA (14). Consistent with what others have reported (20), mice exposed to SHF had significantly increased rearing. Secondly, we used a light/dark test whereby mice could access SHF placed in an open light compartment. Because of rodents’ innate fear of bright aversive arenas, time spent in this compartment is minimal. Increased entries into the food zone located in the light compartment can be operationalized as “risk-taking” (21) or “compulsive-like” (22) behavior. Mice exposed to 24-h SHF had increased distance traveled and entries into the food zone 2 d after, but not immediately after, the “priming” experience, unless VTA insulin receptors were blocked. This effect could be due to increased interest in SHF or to an increase in risk-taking associated with a potential withdrawal from the food (21). Elevated anxiety has been associated with withdrawal from palatable foods (21, 23). However, we did not observe any differences in the time spent or distance traveled in either light or dark compartments between groups, indicating that mice do not appear to have elevated anxiety measures 2 d after exposure to 24-h SHF. Therefore, withdrawal from 24-h SHF is not likely to be driving these food approach behaviors. Taken together, these findings demonstrate that increased food approach and food consumption are primed by previous short-term exposure to SHF.

SHF Induces Rapid, Long-Term Rewiring of the VTA.

Because strengthening of glutamatergic inputs to VTA dopamine neurons underlies learning of cues that predict rewards (8), we tested whether this possibility was a mechanism associated with enhanced food-approach behaviors. Several lines of evidence demonstrate that 24-h SHF can induce an increase in glutamate release sites. First, we observed an increase mEPSC frequency 2 d after 24-h access to SHF and immediately after 24-h SHF access when CB1Rs were blocked or in insulin-deficient mice. Secondly, 2 d after 24-h SHF and immediately after 24-h SHF in the presence of the CB1R antagonist, there was no change in paired pulse ratio, a measure that correlates well with release probability (15). Therefore, there was no change in the function of the presynaptic terminals per se, but an increase in release sites. This result was confirmed by quantifying synaptic density. Here, we observed a selective increase in glutamatergic synapses onto dopamine neurons. Interestingly, consistent with no change in mEPSC amplitude, there was no change in GluA1 expression per synapse. These results indicate that, although 24-h SHF increases glutamate release sites, there is no change in the postsynaptic AMPAR complement per synapse. This result does not rule out the possibility of changes in AMPAR subunit composition. However, because we observed no effect on mEPSC amplitudes after 24-h SHF, any possible changes in subunit composition likely had little or no effect on overall AMPAR conductance.

The elevated endocannabinoid tone observed immediately after 24-h SHF could be directly due to the SHF diet or due to elevated insulin levels. Indeed, plasma insulin was increased immediately after 24-h access to SHF. Furthermore, insulin-induced long-term depression in the VTA requires elevated endocannabinoids and is blocked by AM251 (11), suggesting that it is feasible that elevated insulin from 24-h exposure to SHF may act in the VTA to induce an endocannabinoid-mediated suppression of glutamate release that is offsetting the effect of increased release sites. Consistent with this hypothesis, insulin-deficient mice had elevated excitatory synaptic transmission immediately after 24-h SHF. It is unlikely that elevated endocannabinoid tone is simply due to constitutive activity of CB1Rs, as has been observed in CB1R expression systems (24), because AM251 did not alter evoked AMPARs in RF mice.

Although the endocannabinoid-mediated decrease in release probability was transient, the effect of 24-h SHF on glutamate release sites was longer lasting. We observed an increase in mEPSC frequency onto VTA dopamine neurons 7 d after 24-h SHF exposure. Single exposures of addictive drugs can also elicit long-lasting increases in excitatory synaptic efficacy of VTA dopamine neurons (25, 26) by trafficking of calcium-permeable AMPARs to the synapse (27). Here, we demonstrate that a short-term exposure to palatable food induces a long-lasting increase in synaptic strength, but acts through a presynaptic mechanism by rapidly increasing glutamate release sites. Because enhanced excitatory synaptic transmission onto dopamine neurons is thought to transform neutral stimuli to salient information (9), these changes in excitatory synaptic transmission may underlie the increased food-approach behavior observed days after exposure to SHF and potentially prime increased food consumption.

Intra-VTA Insulin Reverses SHF-Induced Priming of Food Approach Behavior and Intake.

Interestingly, we observed that intra-VTA insulin can reverse increased food-approach behaviors and food consumption in mice exposed to 24-h SHF. In our previous work (11), we demonstrated that insulin suppresses excitatory synapses onto dopamine neurons of RF mice. Here, we show that 2 d after SHF or RF exposure, the effect of insulin at suppressing excitatory synapses was greater in SHF mice compared with RF mice. Insulin induces an excitatory synaptic depression in the VTA by endocannabinoid-mediated inhibition of presynaptic glutamatergic inputs to dopamine neurons (11). This effect is selective for excitatory, but not inhibitory, synapses onto dopamine neurons (11). Because 24-h SHF increases the number of glutamate release sites onto dopamine neurons, the effect of insulin would be greater, assuming that the new release sites express CB1Rs. Intra-VTA insulin likely reversed increased food-approach behaviors and food consumption by suppressing excitatory synaptic transmission onto VTA dopamine neurons. Intranasal insulin primarily targets the CNS and has been effective in suppressing food intake and attention to food-related cues (28–30). Thus, future work should determine whether intranasal insulin can decrease overeating due to food priming induced by palatable food consumption or food-related cues.

The VTA receives inputs of varying strength from ∼50 different brain regions, many of which are glutamatergic (31). Increased glutamate in the VTA is necessary for driving phasic firing associated with reward-seeking behavior (32). Thus, a long-lasting increase in glutamatergic input to dopamine neurons induced by a short-term palatable food exposure may lead to increased food-seeking behavior. Indeed, we demonstrate that 24-h SHF primes food-approach behaviors and food consumption days after the initial exposure. Although this adaptation may have been advantageous to reinforce the salience of food-predictive cues in times of food scarcity, in an environment rich with easily accessible, low-cost, palatable food, this plasticity may be a significant driver of overeating.

Methods

Animals.

All protocols were in accordance with the ethical guidelines established by the Canadian Council for Animal Care and were approved by the University of Calgary Animal Care Committee or the University of British Columbia Animal Care Committee. C57BL/6J male mice were obtained from the University of British Columbia breeding facility (Jackson-derived) or from Jackson Laboratories.

Diets.

Mice were fed 0-h (RF), 1-h, or 24-h SHF in their home cage. Diet compositions are provided in SI Appendix, Table S1, and feeding procedure is described in SI Appendix, SI Methods. Although mice preferred SHF to RF when given simultaneously (P < 0.05; SI Appendix, Fig. S14A), mice consumed similar caloric content from SHF as RF when each food was given alone (P > 0.05; SI Appendix, Fig. S14B), indicating that mice maintained their energy homeostasis during the initial exposure.

Electrophysiology.

All electrophysiological recordings were performed in slice preparations from male C57BL/6J mice ranging from postnatal day 21 (P21) to P30 as described (11). Detailed methods are provided in SI Appendix, SI Methods. Dopamine neurons were identified by the presence of a hyperpolarizing cation current (I_h) (33, 34), a reliable predictor of dopamine neurons medial to the medial terminalis of the optic nucleus in mice (35) or the presence of tyrosine hydroxylase (SI Appendix, Fig. S4).

Immunoelectron Microscopy.

Samples were prepared and processed as described in SI Appendix, SI Methods. Primary antibodies used were rat anti-dopamine transporter (Millipore; MAB369) and rabbit anti-PSD-95 (Frontier Institute, Japan; PSD95-Rb-Af628). Images were collected at 98,000× magnification on a FEI Tecnai G2 Spirit transmission electron microscope and quantified. All images were acquired and analyzed blind to the diet treatment of each mouse.

Intra-VTA Cannulations and Microinjections.

Detailed methods are described in SI Appendix, SI Methods. Mice were implanted with bilateral guide cannulas (26 gauge; Plastics One) into the VTA (anteroposterior, −3.2 mm; mediolateral, ± 0.5 mm; dorsoventral, −4.6 mm). For microinfusions, mice had two habituation sessions, by performing mock perfusions with a microinjector cut above the length of the cannula. On test days, microinfusions were conducted by using 33-gauge microinjectors that protruded 0.2 mm below the base of the guide cannula to a final dorsoventral coordinate of −4.8 mm. Insulin [5 mU in 10% (vol/vol) DMSO and saline, 0.2 µL per hemisphere], vehicle [10% (vol/vol) DMSO in saline, 0.2 µL per hemisphere], S961 (1 µg in saline, 0.2 µL per hemisphere), or vehicle (saline, 0.2 µL per hemisphere) were infused bilaterally into the VTA (0.1 μL/min). Microinjectors were left in place for 3 min after the injection. Mice were then returned to their home cage for 15 min before the behavioral assay. Separate groups of mice received intra-VTA insulin or vehicle.

Light/Dark Box Test with SHF.

Animals were habituated to the test room the day before testing. At the beginning of the 10-min test, animals were placed into the light compartment, facing both the food and the entry to the dark compartment. Any-Maze software (Stoelting) was used to record and track the mice. The time spent in the dark was the difference between the total time and time spent in the light compartment. The apparatus was cleaned between each subject.

Data Analysis.

All values are expressed as mean ± SEM. Statistical significance was assessed by using two-tailed Student's t tests. A two-way ANOVA followed by a Sidak’s post hoc test was used for multiple group comparisons unless otherwise indicated. Prism (Version 5; GraphPad Software) was used to perform statistical analysis. Unless otherwise indicated, data met the assumptions of equal variances.