The nose knows: Nasal temperature tracks facial attractiveness, not social categorization

(234) 29 The regulation of food intake, a sine qua non requirement for survival, thoroughly 30 shapes feeding and energy balance by integrating both homeostatic and hedonic 31 values of food. Unfortunately, the widespread access to palatable food has led to the 32 development of feeding habits that are independent from metabolic needs. Among 33 these, binge eating (BE) is characterized by uncontrolled voracious eating. While 34 reward deficit seems to be a major contributor of BE, the physiological and molecular 35 underpinnings of BE establishment remain elusive. Here, we combined a 36 physiologically relevant BE mouse model with multiscale in vivo approaches to 37 explore the functional connection between the gut-brain axis and the reward and 38 homeostatic brain structures. 39 Our results show that BE elicits compensatory adaptations requiring the gut-to-brain 40 axis which, through the vagus nerve, relies on the permissive actions of peripheral 41 endocannabinoids (eCBs) signaling. Selective inhibition of peripheral CB1 receptors 42 resulted in a vagus-dependent increased hypothalamic activity, modified metabolic 43 efficiency, and dampened activity of mesolimbic dopamine circuit, altogether leading 44 to the suppression of palatable eating. We provide compelling evidence for a yet 45 unappreciated physiological integrative mechanism by which variations of peripheral 46 eCBs control the activity of the vagus nerve, thereby in turn gating the additive 47 responses of both homeostatic and hedonic brain circuits which govern homeostatic 48 and reward-driven feeding. 49 In conclusion, we reveal that vagus-mediated eCBs/CB1R functions represent an 50 interesting and innovative target to modulate energy balance and counteract food-51 reward disorders. 52 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 3

53 54 Feeding is a complex and highly conserved process whose orchestration results from 55 the dynamic integration of interoceptive and exteroceptive signals. The homeostatic 56 and hedonic components of feeding have been attributed to the hypothalamus and 57 the dopamine (DA) reward system, respectively [1]. While the firsts can be broadly 58 defined as key regulators of food intake to ensure optimal energy balance, the 59 seconds mainly relate to the reinforcing properties of sensory stimuli (perception, 60 cues, taste, odors) and reward-associated features of feeding. However, despite the 61 well-accepted recognition that both feeding components are tightly and functionally 62 interconnected [1], they are usually investigated as isolated systems. In addition, the 63 counterpointing central vs peripheral regulations of feeding add a supplemental 64 degree of complexity in the identification of integrative regulatory mechanisms [2]. 65 While energy homeostasis refers to negative feedback mechanisms maintaining 66 body weight at set-points, the combination of both homeostatic and hedonic 67 components of feeding leads to the establishment of feed-forward mechanisms of 68 physiological adaptations. Feed-forward adaptation, also known as allostasis 69 (stability through changes), is critical for energy balance and metabolic efficiency [3] 70 but also contributes to reward-associated events [4]. Indeed, the widespread 71 availability and consumption of palatable diets have profoundly altered the delicate 72 allostatic integration of homeostatic and hedonic signals, leading to the development 73 of metabolic disorders. This is particularly evident in food reward-driven dysfunctions 74 such as binge eating (BE), where uncontrolled feeding perfectly recapitulates the 75 efforts for an organism to adapt its homeostatic processes to the hedonic aspects of 76 feeding. In fact, short- and/or long-term consumptions of energy-rich palatable diets 77 remodel the DA reward system [5] and promote functional adaptations within the 78 hypothalamus [6–8]. Beyond these two core processors of feeding, recent reports 79 have mechanistically demonstrated that the gut-brain vagal axis, besides sensing 80 interoceptive signals and influencing feeding and energy homeostasis [9–11], is also 81 a major modulator of the reward system [12–14]. However, the physiological 82 processes by which the gut-to-brain axis modulates reward feeding remain unclear. 83 Emerging evidence strongly suggest that, besides a plethora of peripheral hormones 84 (i.e. ghrelin, leptin, GLP-1, CCK), peripheral endocannabinoids (eCBs) may be 85 fundamental players in the regulation of feeding and metabolic efficiency [15–17]. 86 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 4 Indeed, eating disorders-associated alterations in peripheral eCBs have been 87 reported in obese and BE patients [18, 19] as well as in diet-induced obese rodents 88 [15, 20]. However, whether and how peripheral eCBs play a permissive role in 89 guiding reward-based feeding behaviors and in buffering the allostatic regulation of 90 energy balance is unexplored. 91 To tackle this question, we took advantage of a physiologically relevant BE-92 like mouse paradigm which, by promoting anticipatory and escalated consummatory 93 food responses, triggers reward-driven behavioral, molecular and allostatic 94 adaptations. Binge eating, which elicited DA-dependent molecular modifications in 95 the reward-related structures [dorsal striatum (DS) and nucleus accumbens (NAc)], 96 revealed an unappreciated integrative gut-to-brain orchestration requiring the 97 modulatory actions of peripheral eCBs. In particular, we show that BE requires an 98 orchestrated dialog between peripheral eCBs and both central hypothalamic and 99 VTA structures through the gut-brain vagal axis, thus modulating both energy 100 balance and reward-like events. 101 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 5

102 See Suppl. Material for the detailed description of materials, methods and 103 techniques, and experimental sizes used in the whole study. 104 105 Animals 106 All experiments were approved by the Animal Care Committee of the Université de 107 Paris (CEB-25–2016) and carried out following the 2010/63/EU directive. 8-12 weeks 108 old male and/or female C57Bl/6J mice (Janvier, Le Genest St Isle, France) were 109 single-housed one week prior to any experimentation in a room maintained at 22 +/-1 110 °C, with light period from 7 AM to 7 PM. Regular chow diet (3 438 kcal/kg, protein 111 19%, fat 5%, carbohydrates 55%, of total kcal, reference #U8959 version 63 Safe, 112 Augy, France) and water were provided ad libitum . Drd2-Cre mice [Tg(Drd2-cre) 113 ER44Gsat/Mmucd, Jackson laboratory] were used for in vivo fiber photometry Ca 2+ 114 imaging in the VTA. Drd2 -eGFP mice [Tg(Drd2-eGFP)S118Gsat/Mmnc] were 115 generated by GENSAT at the Rockefeller University. See Suppl. Material for further 116 details. 117 118 Behaviors 119 Palatable binge eating-like paradigm. Intermittent daily access to a palatable mixture 120 (Intralipid 20% w/v + sucrose 10% w/v) was provided for 1-hour/day during 10-14 121 consecutive days at 10-11 AM in home-cages. During time-locked binge sessions 122 regular chow pellets were not removed. Volume (mL) of consumed palatable mixture 123 was measured at the end of the session. 124 Locomotor activity. Spontaneous and/or induced locomotor activities were measured 125 using an infrared beam-based activity monitoring system (Phenomaster, TSE 126 Systems GmbH, Germany). 127 Tail suspension. To record the activity of GCaMP6f-expressing VTA neurons, mice 128 were suspended above the ground by their tails. Ca 2+ imaging was performed before 129 and during tail suspension. 130 Exploratory drive in a new environment . To record the activity of GCaMP6f-131 expressing VTA neurons in a novelty-induced exploratory drive, mice were put in a 132 new environment (NE) consisting in a novel cage. Ca 2+ imaging was performed 133 before and immediately after changing the environment. See Suppl. Material for 134 further details. 135 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 6 136 Metabolic efficiency analysis 137 Mice were monitored for whole energy expenditure (EE), O 2 consumption, CO 2 138 production, respiratory exchange rate (RER=VCO 2/VO2, V=volume), and locomotor 139 activity using calorimetric cages (Labmaster, TSE Systems GmbH, Bad Homburg, 140 Germany). Ratio of gases was determined through an indirect open circuit 141 calorimeter. This system monitors O 2 and CO 2 at the inlet ports of a tide cage 142 through which a known flow of air is ventilated (0.4 L/min) and regularly compared to 143 a reference empty cage. O 2 consumption and CO 2 production were recorded every 144 15 min during the entire experiment. EE was calculated using the Weir equation for 145 respiratory gas exchange measurements. Food intake was measured with highly 146 sensitive sensors for automated online measurements. Mice had access to food and 147 water ad libitum. Mice were monitored for body weight and composition at the entry 148 and exit of the experiment using an EchoMRI (Whole Body Composition Analyzers, 149 EchoMRI, Houston, USA). Data analysis was performed on Excel XP using extracted 150 raw values of VO2, VCO2 (mL/h), and EE (kcal/h). 151 152 Brown adipose tissue and telemetry body temperature measurements 153 Infrared camera for BAT temperature . Heat production was visualized using a high-154 resolution infrared camera (FLIR E8; FLIR Systems, Portland, OR, USA). To 155 measure brown adipose tissue (BAT) temperature, images of interscapular regions 156 were captured before and after binge sessions. Infrared thermography was analyzed 157 using the FLIR TOOLS. 158 Telemetry body temperature. Anesthetized mice were implanted with a the telemetric 159 transmitter (HD-XG; Data Sciences International) to measure longitudinal fluctuations 160 of core temperature. They were allowed to recover at 35°C and received a daily 161 injection of ketoprofen (Ketofen® 10%) for 3 days. During a 7-days recovery period, 162 mice were monitored and had facilitated access to food. Data were collected using 163 the Ponemah® software (DSI). The detection of transmitted signals was 164 accomplished by a radio receiver (body temperature and locomotor activity) and 165 processed by a microcomputer system. 166 167 Tissue preparation and immunofluorescence 168 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 7 Mice were anaesthetized with pentobarbital (500 mg/kg, i.p., Sanofi-Aventis, France) 169 and transcardially perfused with 4°C PFA 4% for 5 minutes. Sections were processed 170 as previously described [21]. Quantification of immunopositive cells was performed 171 using the cell counter plugin of ImageJ software taking a fixed threshold of 172 fluorescence as standard reference. See Suppl. Material for further details and list of 173 antibodies. 174 175 Western blotting and quantitative RT-PCR 176 At the end of the binge session, the mouse head was cut and immediately immersed 177 in liquid nitrogen for 3 seconds. Brains and sampled structures were processed as 178 previously described [21]. Quantifications were performed using ImageJ software. 179 See Suppl. Material for further details, list of antibodies and primers. 180 181 Drug treatments 182 The following compounds were used: insulin (0.5 U/kg, Novo Nordisk), CCK-8S (10 183 ug/kg, Tocris), liraglutide (100 ug/kg, gift from Novo Nordisk), exendin 4 (10 ug/kg, 184 Tocris), leptin (twice/day for 2 days at 0.25 mg/kg, Tocris), AM251 (3 mg/kg, Tocris), 185 AM6545 (10 mg/kg, Tocris), JD-5037 (3 mg/kg, MedChemExpress), SKF81297 (5 186 mg/kg, Tocris), haloperidol (0.25 and 0.5 mg/kg, Tocris), SCH23390 (0.1 mg/kg, 187 Tocris), GBR12909 (10 mg/kg, Tocris), d-amphetamine sulphate (2 mg/kg, Tocris), 188 JZL184 (8 mg/kg, Tocris). See Suppl. Material for further details. 189 190 Subdiaphragmatic vagotomy 191 Prior to surgery and during 3 post-surgery days, animals were provided with ad 192 libitum jelly food (DietGel Boost #72-04-5022, Clear H 2O) to avoid the presence of 193 solid food in the gastrointestinal tract. Animals received Buprécare® (Buprenorphine 194 0.3 mg) and Ketofen® (100 mg) and were anaesthetized with isoflurane (3.5% for 195 induction, 1.5% for maintenance). Body temperature was maintained at 37°C. Briefly, 196 using a binocular microscope, the right and left vagus nerve branches were carefully 197 isolated along the esophagus and sectioned in vagotomized (VGX) animals or left 198 intact in sham animals. Mice recovered for at least 3 weeks before being used for 199 experimental procedures. See Suppl. Material for further details. 200 201 Quantification of eCBs by UHPLC-MS/MS 202 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 8 Lipids were extracted by liquid/liquid extraction in the presence of deuterated 203 standards. They were then purified by SPE to obtain a fraction containing eCB and 204 NAE that were analyzed by UHPLC-MS/MS using a Xevo TQ-S (ESI source; positive 205 mode). For each analyte, the signal (AUC) of the relevant internal standard was used 206 for normalization. See Suppl. Material for further details. 207 208 Viral production 209 pAAV.Syn.Flex.GCaMP6f.WPRE.SV40 (titer ≥ 1×10 13 vg/ml, working dilution 1:5) 210 was a gift from Douglas Kim (Addgene viral prep #100833-AAV9; 211 http://n2t.net/addgene:100833; RRID:Addgene_100833). 212 pAAV5-hSyn-dLigh1.2 (titer ≥ 4×10¹² vg/mL) was a gift from Lin Tian (Addgene viral 213 prep #111068-AAV5; http://n2t.net/addgene:111068; RRID:Addgene_111068). 214 215 Stereotaxic procedure 216 Mice were anaesthetized with isoflurane, administered with Buprécare® 217 (Buprenorphine 0.3 mg) and Ketofen® (Ketoprofen 100 mg), and placed on a 218 stereotactic frame (Model 940, David Kopf Instruments, California). 219 pAAV.Syn.Flex.GCaMP6f.WPRE.SV40 (0.3 μ l) was injected unilaterally into the VTA 220 (L=−0.5; AP=−3.4; V=−4.4, mm) of Drd2-Cre mice at a rate of 0.05 μ l/min. 221 pAAV5-hSyn-dLight1.2 (0.5 μ l) was injected unilaterally into the NAc (L=−0.75; 222 AP=1.18; V=−4.4, mm) of C57BL6/J male and female mice at a rate of 0.05 μ l/min. 223 The injection needle was carefully removed after 5 minutes waiting at the injection 224 site and 2 minutes waiting halfway to the top. Optical fiber for Ca 2+ (VTA) and 225 dopamine (NAc) imaging was implanted 100 μ m above the viral injection site. 226 Animals were tested 4 weeks after viral stereotaxic injections. 227 228 Fiber photometry and data analysis 229 A chronically implantable cannula (Doric Lenses, Québec, Canada) composed of a 230 bare optical fiber (400 μ m core, 0.48 N.A.) and a fiber ferrule was implanted 100 μ m 231 above the location of the viral injection site in the VTA or in the NAc. The fiber was 232 fixed onto the skull using dental cement (Super-Bond C&B, Sun Medical). Real-time 233 fluorescence signals emitted by GCaMP6f or the DA biosensor dLight1.2 [22] were 234 recorded and analyzed as previously described [21]. Fluorescence was collected 235 using a single optical fiber for both delivery of excitation light streams and collection 236 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 9 of emitted fluorescence. The fiber photometry setup used 2 light-emitting LEDs: 405 237 nm LED sinusoidally modulated at 330 Hz and a 465 nm LED sinusoidally modulated 238 at 533 Hz merged in a FMC4 MiniCube (Doric Lenses) that combines the 2 239 wavelengths excitation light streams and separate them from the emission light. The 240 MiniCube was connected to a Fiberoptic rotary joint connected to the cannula. A 241 RZ5P lock-in digital processor controlled by the Synapse software (Tucker-Davis 242 Technologies, TDT, USA), commanded the voltage signal sent to the emitting LEDs 243 via the LED driver (Doric Lenses). Data are presented as z-score of Δ F/F. See 244 Suppl. Material for further details. 245 246 Statistics 247 All data are presented as mean ± SEM. Statistical tests were performed with Prism 7 248 (GraphPad Software, La Jolla, CA, USA). The detailed statistical analyses are listed 249 in the Suppl. Table 1. Normality was assessed by the D’Agostino-Pearson test. 250 Depending on the experimental design, data were analyzed using either Student t-251 test (paired or unpaired) with equal variances, One-way ANOVA or Two-way 252 ANOVA. In all cases, the significance threshold was automatically set at p < 0.05. 253 ANOVA analyses were followed by Bonferroni post hoc test for specific comparisons 254 only when overall ANOVA revealed a significant difference (at least p < 0.05). 255 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 10

256 257 Time-locked access to palatable diet induces adaptation of nutrient partitioning 258 and metabolic efficiency 259 Several paradigms of bingeing are widely used to model eating disorders [23]. 260 However, the majority of these paradigms mainly rely on ( i) prior alterations of basal 261 homeostasis (food or water restriction/deprivation, stress induction), (ii) dietary 262 exposure to either high-sugar or high-fat foods, or (iii) absence of food choice during 263 bingeing periods. We therefore adapted existing protocols to better study reward and 264 homeostatic components of food intake during binge eating (BE). Since dietary 265 mixtures of fat and sugar lead to enhanced food reward properties [24], we designed 266 a highly palatable diet (sugar and fat) to promote intense reward-driven feeding. 267 Time-locked access to this palatable diet was sufficient to drive escalating binge-like 268 consumption without restricting access to chow diet ( Fig. 1A). In that regard, we are 269 confident that this model is preferentially driven by reward values over metabolic 270 demands since animals are neither food nor water restricted. 271 Male mice intermittently exposed to this dietary palatable mixture maximized their 272 intake within a few days ( Fig. 1B ). This palatable food consumption was 273 simultaneously associated with an increased anticipatory locomotor activity ~2 hours 274 before food access and lasted for another ~1-2 hours ( Fig. 1C, C1), with no changes 275 in the ambulatory activity during the dark phase (Fig. 1C ). The same animals were 276 characterized by a significant reduction in spontaneous nocturnal food intake ( Fig. 277 1D, D1). However, the overall caloric intake [standard diet (SD) + palatable food (PF)] 278 remained similar to controls, thus indicating a conserved maintenance in calories 279 consumption despite reward-driven food intake (Fig. 1E ). Importantly, isocaloric 280 feeding was associated with conserved body weight (BW) and body composition 281 (Fig. 1F, Suppl. Fig. 1A, B). A similar pattern of food reward-driven adaptations was 282 also observed in female mice (Suppl. Fig. 1C-E). 283 Next, we investigated the consequences of BE on metabolic efficiency. Indirect 284 calorimetry analysis revealed an increase in the respiratory exchange ratio (RER) 285 before and after intermittent palatable food consumption ( Fig. 1G, G1), whilst a stark 286 reduction was detected in the dark phase (Fig. 1G ), thereby highlighting a metabolic 287 shift of energy substrates use (from carbohydrates to lipids as indicated by RER ~1 288 or RER ~0.7, respectively). This was further confirmed by the modulation of fatty acid 289 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 11 oxidation (FAO, Suppl. Fig. 1F). In addition, we also observed an increased energy 290 expenditure (EE) during the food anticipatory and consummatory phases ( Fig. 1H, 291 H1). Furthermore, infrared thermographic analysis revealed that BE was associated 292 with a transient increase in brown adipose tissue (BAT) energy dissipation ( Fig. 1I), 293 while telemetric recording of core body temperature revealed a BE-specific increase 294 during the anticipatory, consummatory and post-prandial phases ( Fig. 1J, J1, Suppl. 295 Fig. 1G ) and a sharp reduction during the last hours of the dark phase. Overall, 296 changes in core body temperature were fostered around the time of locked palatable 297 food access and overlapped with the increase in locomotor activity (Fig. 1J, K). 298 Access to calories-rich food and time-restricted feeding are invariably associated with 299 changes in circulating signals reflecting metabolic and behavioral adaptations [25]. In 300 line with this, we observed that our BE model was associated with reduced 301 circulating triglycerides and insulin, and increased circulating corticosterone during 302 the anticipatory phase ( Suppl. Fig. 1H-J ) while insulin sensitivity, as assessed by 303 oral glucose tolerance test, remained unchanged ( Suppl. Fig. 1K, L ). These data 304 indicate that homeostatic adaptations occurring during time-locked palatable feeding 305 lead to changes in lipid-substrates utilization and promote adaptive activation of the 306 hypothalamic-pituitary-adrenal (HPA) axis. However, bingeing did not elicit major 307 changes in the expression of key hunger- and satiety-related hypothalamic genes 308 (Npy, Agrp, Pomc, Cart, Hcrt, Suppl. Fig. 1M). 309 Overall, these results point to rapid reward-driven allostatic adaptations during which 310 animals optimize their palatable food consumption and physiologically adapt, at the 311 cost of regular chow food intake, to maintain a stable body weight. 312 313 BE induces dopamine-related modifications in a D1R-dependent manner 314 Dopamine (DA)-neurons and DA-sensitive structures, such as the dorsal striatum 315 (DS) and the nucleus accumbens (NAc), are critical players in reward-based 316 paradigms and in BE disorders [26, 27]. Here, we investigated whether and how 317 bingeing modulated the DA-associated signaling machinery. The activations 318 (phosphorylation) of the ribosomal protein S6 and the extracellular signal-regulated 319 kinases (ERK) were used as functional readouts of DA-dependent molecular activity 320 [28, 29] ( Suppl. Fig. 2A, B ). The food anticipatory phase was associated with an 321 increase in phospho-ERK only in the DS ( Fig. 2A, B, Suppl. Fig. 2C ), mostly 322 reflecting the increased locomotor activity during the anticipatory phase. Palatable 323 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 12 food consumption induced an increase in phospho-ERK and phospho-S6 (at both 324 Ser235/236 and Ser 240/244 sites) in both DS and NAc ( Fig. 2A, B, Suppl. Fig. 2C). 325 Interestingly, acute (single) consumption of palatable diet failed to trigger ERK and 326 S6 activation ( Fig. 2A, B, Suppl. Fig. 2C), revealing that molecular adaptations of 327 DA signaling in the DS/NAc require the full establishment of BE and not only 328 palatable food consumption. Immunofluorescence analysis further confirmed BE-329 induced S6 activation (Suppl. Fig. 2D, E). 330 Next, we wondered whether food-reward anticipatory and/or consummatory 331 phases were followed by adaptive changes in DA signaling. We treated mice with 332 GBR12909 (10 mg/kg), a specific DA transporter (DAT) blocker that leads to synaptic 333 accumulation of DA. Interestingly, we observed a different behavior depending on BE 334 phases (anticipatory vs consummatory). Before palatable food access, GBR similarly 335 increased locomotor activity in both bingeing and control animals ( Fig. 2C, C 1). 336 However, when GBR was administered following palatable food consumption (1h), 337 GBR-induced locomotor response was blunted in bingeing animals ( Fig. 2D, D 1). 338 These results indicate that BE-induced physiological adaptations are characterized 339 by the enabled ability for palatable food to impinge on DA release and action. 340 At the postsynaptic level, DA acts onto medium spiny neurons (MSNs) which express 341 either the dopamine D1R or D2R. In order to discriminate the role of D1R vs D2R 342 signaling in BE, we pretreated animals with the D1R antagonist SCH23390 (0.1 343 mg/kg) or vehicle (Veh) 30 min prior access to palatable diet. SCH23390 dramatically 344 reduced palatable food consumption (Fig. 2E). On the contrary, 30 min pretreatment 345 with the D2R antagonist haloperidol (0.25 and 0.5 mg/kg) did not dampen palatable 346 food consumption (Fig. 2F), even at cataleptic doses [30]. In line with this evidence, 347 activation of striatal D1R leads to downstream phosphorylation of S6 and ERK [28, 348 29]. The adaptive molecular changes also required D1R activation since SCH23990 349 (0.1 mg/kg) largely suppressed BE-associated phosphorylation of S6 in both DS (Fig. 350 2G, G1, H1) and NAc ( Fig. 2G2, H 2, Suppl. Fig. 2F ). Of note, although SCH23390 351 reduced binge-elicited anticipatory locomotor activity, basal locomotor activity in 352 naive animals was not impaired (Fig. 2I, I 1), thereby excluding the confounding 353 effects due to changes in basal locomotor activity. Furthermore, a compensatory 354 rescue in chow intake was observed in SCH23390-pretreated bingeing animals 355 during the dark phase, excluding potential long-lasting effects of the D1R inhibition 356 (Fig. 2J). To further validate the implication of D1R in BE-elicited DA modifications, 357 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 13 we measured the locomotor activity triggered by the D1R agonist SKF81297 (5 358 mg/kg) at the end of the BE session (1h after food access). Interestingly, we 359 observed an earlier (first 30 min) significant increase in locomotor activity in bingeing 360 animals compared to control mice, although no major differences were detected 361 during the cumulative 2-hrs response (Fig. 2K, K1). 362 Overall, our results reveal that the critical phases surrounding palatable food 363 consumption in the context of BE profoundly affect D1R-associated signaling. 364 365 Peripheral endocannabinoids govern binge eating 366 Recent studies have highlighted the role of neuronal and endocrine gut systems in 367 the regulation of food reward-seeking and DA-associated behaviors [31, 32]. We 368 therefore tested whether gut-born metabolic signals had a privileged action onto BE-369 like consumption of palatable diet when compared to other known circulating satiety 370 signals. 371 First, we observed that peripherally injected leptin (repeated 0.25 mg/kg, Suppl. Fig. 372 3A, B ) or insulin (acute, 0.5 U/kg) did not trigger any reduction in palatable food 373 consumption when administered in bingeing animals (Fig. 3A). Then, we investigated 374 whether gut-born satiety signals retained anorectic properties with a similar protocol. 375 GLP-1R agonists, exendin-4 (10 µg/kg) and liraglutide (100 µg/kg), successfully 376 reduced binge-consumption of palatable diet ( Fig. 3A). Moreover, pharmacological 377 activation of GLP-1R, which did not alter spontaneous ambulatory activity, was also 378 associated with a decrease in the anticipatory and consummatory locomotor phases 379 (Suppl. Fig. 3C, D ). Similarly, the cholecystokinin (CCK) analog CCK-8S (10 µg/kg) 380 acutely decreased palatable food intake ( Fig. 3A). Since only the anorectic action of 381 gut-born signals was efficient in counteracting binge-like consumption, we also 382 investigated the effect of endocannabinoids (eCBs) which are important signals in 383 relaying nutrients-induced adaptive responses in the gut-brain axis [33, 34]. We 384 acutely inhibited the CB1R with the global acting antagonist/inverse agonist AM251 385 (3 mg/kg, i.p.) and observed a dramatic reduction of BE-like consumption ( Fig. 3A). 386 Next, we wondered whether bingeing was accompanied by alterations in circulating 387 peripheral eCBs [anandamide (AEA) and 2-arachidonoylglycerol (2-AG)] and eCBs-388 related species [docosahexanoyl ethanolamide (DHEA), oleoylethanolamide (OEA)]. 389 While circulating N-acylethanolamines (AEA, DHEA, OEA) remained unaffected, 390 bingeing induced a significant increase in circulating 2-AG ( Fig. 3B). No differences 391 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 14 in central 2-AG levels were detected in key brain regions such as the hypothalamus, 392 VTA, DS and NAc (Suppl. Fig. 4A-D). 393 Given the rise in peripheral 2-AG, we were eager to explore the role of peripheral 394 CB1R signaling in BE outputs. Thus, we used the bona fide peripherally restricted 395 CB1R neutral antagonist AM6545 (10 mg/kg, i.p.) or inverse agonist JD-5037 (3 396 mg/kg, i.p.), two compounds with poor blood brain barrier permeability [35–37]. 397 Pretreatment (1h before bingeing) with AM6545 or JD-5037 induced a stark reduction 398 of BE consumption when administered acutely (Fig. 3C). Conversely, the increase of 399 2-AG, achieved through the pharmacological inhibition (JZL184, 8 mg/kg) of its 400 catabolic enzyme monoacylglycerol lipase (MAGL) [38], resulted in an increase of 401 palatable food consumption that was fully prevented by AM6545 ( Fig. 3D ). This 402 bidirectional modulatory action of eCBs/CB1R on BE did not show signs of 403 desensitization and remained efficient throughout 4 days of daily pharmacological 404 intervention (Fig. 3E). In the same line, thermogenic and locomotor activity analyses 405 revealed that acute pretreatment with AM6545 strongly dampened both the 406 anticipatory and consummatory phases of BE ( Fig. 3F, Suppl. Fig. 4E ) as well as 407 the activation of S6 and cFos in the DS and NAc of bingeing mice ( Fig. 3G) with no 408 sex-dependent differences ( Suppl. Fig. 4F ). These results indicate that peripheral 409 CB1R signaling is sufficient to control compulsive eating in BE and its reward-like 410 molecular adaptations. 411 412 We next explored how peripheral CB1R signaling modulates metabolic 413 efficiency in the context of BE. Pretreatment with AM6545 significantly increased fatty 414 acid oxidation (FAO) ( Fig. 3H, H 1). Importantly, this AM6545-induced FAO did not 415 depend on reduced calorie intake (Binge session) or basal calorie contents (NoBinge 416 session) (Fig. 3H2) nor on altered energy expenditure (EE) ( Suppl. Fig. 4G). These 417

indicate that acute manipulation of peripheral eCB tone affects nutrient 418 partitioning and promotes a shift towards whole body lipid-substrate utilization. 419 Importantly, oral administration (p.o.) of AM6545 did not blunt bingeing responses 420 (Fig. 3I) nor increase FAO ( Fig. 3J). These results suggest that, in our behavioral 421 model, CB1R-mediated homeostatic adaptations do not depend on the lumen-422 oriented apical CB1R of endothelial or enteroendocrine intestinal cells [39, 40] but 423 rather on non-lumen-oriented CB1R. 424 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 15 Given these results, we investigated the anatomo-functional structures at the 425 interface between peripheral and central systems. Compared to control and bingeing 426 mice, AM6545-pretreated bingeing mice showed a more pronounced neuronal 427 activation in the caudal nucleus tractus solitarius (cNTS) and the area postrema (AP) 428 (Fig. 3K) as well as in the cNTS-projecting lateral parabrachial nucleus (lPBN) ( Fig. 429 3L), thus pointing the gut-brain vagal axis as potential mediator of our effects. 430 431 The gut-brain vagal axis is required for eCBs-mediated effects 432 Recent reports have indicated that CB1R is densely expressed in vagal afferent 433 neurons [41]. To discriminate between all vagal afferents, we performed a meta-434 analysis on recent single-cell transcriptomic results [9] obtained through a path-435 specific viral strategy of gut segments ( Fig. 4A). This analysis revealed that Cnr1 436 (gene encoding for CB1R), but not Cnr2, is highly enriched in all segments of the gut-437 brain vagal axis ( Fig. 4B, Suppl. Fig. 4H, I ) and that, together with well-known 438 afferent markers (Slc17a6 , Scn10a, Htr3a, Cartpt, Grin1, Phox2b), Cnr1 may be 439 considered as a constitutive marker of vagal sensory neurons. Thus, we took 440 advantage of subdiaphragmatic vagotomy (VGX) to investigate whether the eCBs-441 vagus axis was necessary/sufficient to mediate the modulatory effects of eCBs on 442 BE. 443 First, we tested whether the pro-bingeing effects of the MAGL inhibitor JZL184 ( Fig. 444 3D, E) required an intact vagal transmission. Although vagotomy per se was 445 associated with a decrease in time-locked hedonic feeding (see Fig. 4C, E ) and 446 consequent BE-derived compensatory homeostatic adaptations ( Suppl. Fig. 5 ), 447 repeated inhibition (4 days) of MAGL (JZL184, 8 mg/kg) induced a significant 448 increase of palatable food consumption only in sham animals (Fig. 4C). 449 Second, we tested whether the anti-bingeing effects of AM6545 were also routed by 450 the vagus nerve. In sham mice, acute AM6545 led to a strong increase of cFos-451 positive neurons in the cNTS, AP and cNTS-projecting lPBN, while the signal was 452 abolished in VGX mice (Fig. 4D, D 1, D2). Of note, blockade of peripheral CB1R did 453 not activate the rostral NTS (rNTS) (Suppl. Fig. 4J), thus indicating and further 454 confirming that gut-to-brain vagal inputs are necessary to mediate the action of 455 AM6545. 456 We also observed that the integrity of the vagus nerve was essential to mediate the 457 anorectic action of AM6545 on BE behavior since the peripheral CB1R antagonist did 458 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 16 not trigger anti-bingeing responses in VGX mice compared to sham mice ( Fig. 4E). 459 Furthermore, vagotomy abolished the increase in FAO following AM6545 460 administration observed in sham mice ( Fig. 4F, F 1, G, G 1), indicating that the gut-461 brain vagal communication routes feeding and the metabolic components associated 462 with BE. 463 These vagus-dependent homeostatic adaptations promoted by the peripheral 464 blockade of CB1R prompted us to investigate whether AM6545 was able to alter the 465 activity of brainstem-projecting hypothalamic structures that control feeding. Indeed, 466 AM6545 induced a strong vagus-dependent increase of cFos-positive neurons in the 467 PVN and DMH regions (Fig. 4H, I), indicating that the metabolic adaptations induced 468 by peripheral blockade of CB1R require a vagus-mediated 469 cNTS→ PBN→ hypothalamus circuit whose nodes’ activation control feeding and 470 energy homeostasis [42–44]. 471 472 Peripheral CB1R signaling routed by the vagus nerve controls the activity of 473 VTA dopamine neurons 474 Palatable bingeing also strongly relies on central DA-dependent mechanisms ( Fig. 475 2). Therefore, we explored the functional connection between peripheral eCBs and 476 the gut-to-brain vagal axis in the modulation of the DA system. Since 477 hedonic/motivated spontaneous feeding is blunted in AM6545-treated mice ( Fig. 3, 478 4), a feature limiting further in vivo investigations of DA/reward events, we promoted 479 and gauged the activity of the DA system by using pharmacological tools. Naive mice 480 were acutely pretreated with AM6545 (or JD-5037) or vehicle 1h before being 481 administered with the DAT blocker GBR12909. Blockade of peripheral CB1R 482 drastically reduced GBR-induced locomotor activity ( Fig. 5A, A 1, B, B 1) as well as 483 GBR-triggered cFos in the striatum ( Fig. 5C, C 1). To further study in vivo DA 484 dynamics, we took advantage of the virally expressed DA biosensor dLight1.2 485 coupled to in vivo fiber photometry [22]. Importantly, pretreatment with AM6545 486 blunted GBR-evoked accumbal DA accumulation (Fig. 5D, D 1). Interestingly, unlike 487 the brain penetrant CB1R antagonist/inverse agonist AM251, the peripheral AM6545 488 failed in counteracting amphetamine-induced locomotor activity ( Fig. 5E, F, Suppl. 489 Fig. 6A). This evidence indicates a clear difference in the action of peripheral vs 490 central CB1R and suggests that inhibition of peripheral CB1R may modulate the 491 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 17 intrinsic spontaneous activity of DA-neurons rather than altering evoked DA release 492 events. 493 Altogether, these results reveal that inhibition of peripheral CB1R, besides promoting 494 satiety and FAO (Fig. 3, 4 ), may dampen reward-driven feeding also by 495 concomitantly reducing DA-neurons spontaneous activity and consequent activation 496 of dopaminoceptive structures. 497 To directly address this point, VGX mice were acutely pretreated with AM6545 1h 498 prior GBR12909. Remarkably, ablation of the vagus nerve abolished the blunting 499 effect of AM6545 on GBR-elicited locomotor activity ( Fig. 5G, G 1). Moreover, 500 consistently with what observed for palatable bingeing (Fig. 3I ), this vagus-to-brain 501 effect was further highlighted by the lack of action of AM6545 when orally 502 administered (Fig. 5H, Suppl. Fig. 6B). 503 Finally, to fully establish that peripheral inhibition of CB1R modulates the activity of 504 VTA DA-neurons, we performed cell type-specific in vivo Ca 2+ imaging of DA-505 neurons in presence or absence of AM6545. We took advantage of Drd2-Cre mice to 506 virally express GCaMP6f in VTA DA-neurons ( Fig. 6A) as they co-express the 507 autoreceptor D2R ( Suppl. Fig. 6C ). Indeed, using this mouse line we were able to 508 detect activation and inhibition of VTA DA-neurons following rewarding (high-fat high-509 sugar pellet) or aversive (scruff restraint) events (Suppl. Fig. 6D, E), respectively. To 510 trigger the activity of DA-neurons independently from food- or drugs-associated 511 stimuli, we used two paradigms that modulate DA-neurons: the tail suspension [45] 512 and exposure to a new environment [46] which promotes exploration ( Fig. 6B). 513 Inhibition of peripheral CB1R led to a reduced activation of VTA DA-neurons during 514 both paradigms (Fig. 6C, D). At the behavioral level, while AM6545 did not alter the 515 time of immobility in the tail suspension test (Fig. 6E), it reduced the exploratory drive 516 in a vagus-dependent manner (Fig. 6F, F 1). Of note, the brain permeable AM251 517 reduced exploratory drive in a vagus-independent manner (Fig. 6G). 518 Collectively, these results reveal that peripheral CB1R signaling routed through the 519 vagal axis exerts an integrative control over metabolic/satietogenic (Fig. 3, 4) and DA 520 paths (Fig. 5, 6), both of which are pivotal for the establishment of BE. 521 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 18

522 523 A characteristic feature of feeding behavior is its key ability to dynamically adapt to 524 sensory and environmental stimuli signaling food availability. Such adaptive strategy 525 is even more pronounced when food is palatable and energy-dense. Indeed, the 526 control of feeding strategies requires complex and highly interactive systems that can 527 hardly be unequivocally attributed to single structures, circuits or mediators. 528 In our study, we observed that, first, palatable time-locked feeding mobilizes both 529 homeostatic and hedonic components of feeding through fast, but yet physiological, 530 allostatic adaptations. Second, such allostatic adaptations require a concerted 531 involvement of central DA (hedonic drive) and peripheral eCBs signaling 532 (homeostatic and hedonic drive). Third, the permissive role of peripheral eCBs fully 533 relies on the vagus nerve which, by a polysynaptic circuit, controls the activity of both 534 satietogenic and reward (dopamine) structures. Fourth, our results point to peripheral 535 CB1R neutral antagonists as promising therapeutic tools to counteract eating as well 536 as reward-related disorders. 537 Overall, our study describes for the first time the fundamental role of eCB gut-brain 538 transmission as a core component of binge eating and its behavioral, cellular and 539 molecular adaptations. 540 541 Here, by investigating the pathways involved in hedonic feeding in absence of 542 hunger or energy deprivation, we provide evidence that the hedonic drive to eat, as 543 triggered by our intermittent time-locked model, promotes rapid homeostatic 544 compensations leading to escalating consumption of palatable food and to allostatic 545 adaptations of energy metabolism. As such, caloric demands are fulfilled and 546 classical energy-mediated homeostatic signals (leptin, insulin) do not seem to 547 spontaneously interfere, thus providing us the opportunity to study food intake-related 548 integrative pathways with the abstraction of the homeostatic vs hedonic discrepancy. 549 In line with clinical data [47, 48], we observed that binge-like feeding in lean animals 550 is not necessarily associated with overweight gain. The allostatic adaptations 551 observed, ranging from increased anticipatory feeding phase to pre-feeding 552 increased corticosterone levels and food intake maximization, all represent key 553 hallmarks of the compulsive and emotional states of BE patients [49–51]. The 554 anticipatory feeding phase was associated with decreased levels of plasma TG and 555 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 19 insulin, whereas both anticipatory and consummatory phases were characterized by 556 increased energy expenditure, core temperature and metabolic efficiency, thereby 557 suggesting a metabolic shift of nutrients’ use. This observation perfectly mirrors the 558 allostatic theory, which stands on the fact that an organism anticipates and adapts to 559 environmental changes while accordingly adjusting several physiological parameters 560 to maintain stable physiological states [52, 53]. Allostatic mechanisms have 561 classically been discussed in terms of stress-related regulatory events. However, the 562 hedonic value of a stimulus (food, recreational drugs) can function as a feed-forward 563 allostatic factor [4]. 564 565 In line with this notion, analysis of key DA-activated downstream targets in the 566 DS and NAc highlighted specific patterns of molecular activation. Notably, while the 567 anticipatory phase was associated with an increase in ERK and S6 Ser235/236 568 phosphorylations, the consummatory phase was also accompanied by a robust 569 increase in mTOR-mediated S6 Ser240/244 activation. Such signaling events, which did 570 not depend on a single episode of palatable food intake, required the dopamine D1R. 571 Whether this molecular regulation reflects the full establishment of BE or the amount 572 of ingested palatable food remains to be established. Nevertheless, this D1R 573 mechanism is of interest since, contrary to the well-known molecular insights of drugs 574 of abuse also requiring the D1R [54–56], food-related disorders have usually been 575 predominantly associated with altered D2R signaling [57, 58]. Our results reveal that 576 binge eating, characterized by transients and sudden urges of hedonic drive, 577 requires, at least in its early phases, a D1R-mediated transmission. This D1R-578 dependent mechanism is in line with the affinity and time-dependent dynamics of 579 dopamine effects [54] as well as with the molecular action of released DA which, by 580 binding to Gα (olf)-coupled D1R, would trigger the activation of the aforementioned 581 pathways, while activation of the Gi-coupled D2R would lead to their inhibition. 582 However, in clear opposition to psychostimulants, which directly act at central DA 583 synapses, food and food-mediated behaviors impact on DA transmission through a 584 plethora of indirect and often peripherally born long-range acting mediators. Notably, 585 nutrients, as demonstrated by intragastric infusion of fat and sugar [14, 59–61], or 586 gut-born signals [62–64], are sufficient to modulate DA release in reward-related 587 structures. Here, we observed that gut-born signals such as CCK, GLP1 and 588 endocannabinoids (eCBs) are essential in gating bingeing. In particular, we found 589 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 20 that time-locked consumption of palatable food was associated with a rise in 590 peripheral endogenous eCBs, notably 2-AG. Furthermore, inhibition of the 2-AG-591 degrading enzyme MAG lipase resulted in a potentiation of palatable food 592 consumption. Thus, by taking advantage of low brain permeant CB1R 593 antagonists/inverse agonists, we observed that blockade of CB1R was able to fully 594 abolish both anticipatory and consummatory phases of hedonic feeding as well as 595 the potentiated feeding induced by the MAG lipase inhibitor. These effects agree with 596 the literature showing that endogenous peripheral eCBs are highly and dynamically 597 modulated in eating disorders, and act as powerful mediators of the gut-to-brain 598 integration [17]. 599 600 Previous studies have shown that (chronic) administration of AM6545 601 promoted long-term maintenance of weight loss and reduction of dyslipidemia in 602 obesity [35, 37]. Here, we show that a single, as well as repeated (4 days), 603 administration of AM6545 potently inhibits binge eating and its molecular 604 adaptations. The anorectic effects of peripheral blockade of CB1R have been, at 605 least in part, attributed to the property of global CB1R antagonists to promote fatty 606 acid oxidation (FAO). In agreement with these studies, we have observed that acute 607 administration of AM6545 was able to dramatically increase FAO independently of 608 food intake. However, here we also demonstrate that such effects require the vagus 609 nerve. The action of endogenous eCBs as well as of AM6545 on CB1R-expressing 610 vagal afferents [41] may explain our results. In fact, an increase in endogenous eCBs 611 during palatable feeding would slow the vagus nerve activity through the inhibitory 612 Gi-coupled signaling of CB1R, thus delaying cNTS-reaching satiety signals and 613 promoting food intake. On the contrary, peripheral blockade of CB1R, especially 614 when peripheral eCB levels are endogenously high ( i.e. binge eating, bulimia, 615 obesity), would lead to a prompt disinhibition and to the concomitant activation of 616 satietogenic brain pathways (cNTS → PBN→ PVN). Moreover, it is worth to mention 617 that under fasting or lipoprivic conditions the systemic CB1R inverse agonist 618 SR141716A modulated feeding by the vagal and sympathetic systems [65]. Another 619 site of action for peripheral eCBs is represented by CB1R-expressing gut cells [40, 620 66]. Interestingly, oral administration of a peripheral CB1R antagonist resulted in a 621 reduction of alcohol intake via a ghrelin-dependent and vagus-mediated mechanism 622 [66]. However, in our reward-driven feeding model, oral administration of AM6545 623 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 21 failed to modulate metabolic efficiency as well as to prevent bingeing behavior, thus 624 suggesting that lumen-oriented apical CB1R may not be involved in our mechanism. 625 Intriguingly, recent studies have uncovered that sensory neuropod cells in the gut 626 [67] can synaptically signal with the juxtaposed vagal afferents using, among other 627 possible mediators [68], the fast-acting neurotransmitter glutamate [10]. Whether this 628 specialized gut-to-nerve synapse also mobilizes eCBs, as it occurs at most central 629 excitatory synapses, remains to be determined. In addition, it is important to highlight 630 the key role of peripheral CB1R in adipocytes in the regulation of energy balance 631 [69]. However, whether and how adipocytes may, indeed indirectly, influence the 632 vagal axis is yet unclear. 633 Overall, it would not be hazardous to suggest that peripheral eCBs may impact 634 feeding patterns through different integrative mechanisms which, depending on the 635 location of peripheral CB1R, may strongly modulate distinct hedonic and homeostatic 636 functional outputs. These results call for cell-type and tissue-type-specific strategies 637 to selectively delete CB1R and/or eCBs-producing enzymes in distinct compartments 638 of the gastrointestinal tract and in the neuronal gut-brain axis. 639 To anatomically provide an explanatory gut-to-brain circuit able to support the vagus-640 mediated action of AM6545, we found a stark increase of cFos, a marker of neuronal 641 activity, in key brain regions of the satietogenic neuronal pathway. Importantly, we 642 reveal that blockade of peripheral CB1R signaling resulted in a strong vagus-643 dependent activation of the cNTS as well as of its downstream projecting structures, 644 notably the lPBN and the hypothalamic PVN. This segmented activation of the 645 gut→ brainstem→ hypothalamus path is most likely responsible for the AM6545-646 induced effects on bingeing and energy homeostasis since structure-specific 647 activation of these nodes has been shown to reduce food intake and alters energy 648 homeostasis [43, 70–72]. In addition to this satietogenic path and given the strong 649 reward component of our paradigm, we also uncover that AM6545-mediated vagus 650 activation results in a dampened activation of VTA DA-neurons. However, such effect 651 did not depend on the releasing capabilities of DA-neurons since AM6545 failed to 652 alter amphetamine-evoked locomotor activity. In addition, taking advantage of virally 653 mediated GCaMP6f-mediated in vivo Ca2+ imaging of putative VTA DA-neurons, here 654 we demonstrate that peripheral blockade of CB1R clearly reduced both basal and 655 evoked activity of DA-neurons, a feature resembling some neurochemical effects of 656 vagal nerve stimulation [73, 74]. 657 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 22 The VTA is characterized by a highly heterogeneous connectivity [75], and a single 658 and monosynaptic circuit responsible for the inhibition of DA-neurons through the 659 AM6545-activated vagus nerve cannot be selectively sorted out yet. However, 660 several satiety-related structures in the brainstem and hypothalamus are known to 661 project and modulate, directly and/or indirectly, VTA DA-neurons [13, 76–79]. Among 662 these circuits, the PBN → VTA relay is of particular interest since excitatory PBN 663 neurons also largely contact VTA GABA-neurons [78, 80] which in turn may drive the 664 inhibition of VTA DA-neurons and consequent dampening of motivated 665 behaviors. 666 667 Here, we show that DA-dependent adaptations require orchestrated inputs 668 among which peripheral endocannabinoids, through the vagus nerve, allostatically 669 scale the homeostatic and hedonic components of feeding and act as mandatory 670 gatekeepers for adaptive responses of the reward circuit. The gut-brain axis is 671 increasingly incriminated as a key player of the regulation of energy metabolism [81], 672 and we show for the first time that BE is under the control of the vagus-mediated 673 peripheral inputs. Pointing to peripheral eCBs as permissive actors of this eating 674 disorder certainly brings novelty to the clinical investigations aimed at identifying 675 innovative and non-invasive therapeutic strategies. Importantly, this study further 676 points to the gut-brain axis as a privileged target to modulate brain structures that are 677 functionally responsible for processing cognitive and reward events in an integrative 678 manner. 679 In conclusion, while further studies are warranted to fully untangle the key enteric 680 actors responsible for this phenomenon, our study identifies a novel integrative 681 mechanism by which peripheral endocannabinoids through the gut-brain vagal axis 682 gate allostatic feeding by controlling satiety and reward events, thus also paving the 683 way to target peripheral elements for brain disorders. 684 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 23 Acknowledgments 685 We thank Chloé Morel, Rim Hassouna, Anne-Sophie Delbes, Daniela Herrera Moro 686 and Raphaël Denis for technical advice and support. Adrien Paquot 687 (BPBL/UCLouvain) is acknowledged for his help with eCB quantification. We thank 688 Olja Kacanski for administrative support, Isabelle Le Parco, Ludovic Mai ngault, 689 Angélique Dauvin, Aurélie Djemat, Florianne Michel, Magguy Boa and Daniel 690 Quintas for animals’ care and Sabria Allithi for genotyping. We acknowledge the 691 technical platform Functional and Physiological Exploration platform (FPE) of the 692 Université de Paris (BFA-UMR 8251) and the animal core facility Buffon of the 693 Université de Paris/Institut Jacques Monod. This work was supported by the Fyssen 694 Foundation, Nutricia Research Foundation, Allen Foundation Inc., Université de Paris 695 and CNRS. CB and EM we re supported by fello wships from the Fondation pour la 696 Recherche Médicale (FRM). Telemetry experiments were supported by the 697 Continuous Glucose Telemetry Award 2018 (Dr. Denis) and sponsored by Data 698 Sciences International. 699 700 Author Contributions 701 C.B. and G.G. conceived, designed, performed and analyzed most of the 702 experiments. J.C. performed surgeries and behavioral experiments. E.M. helped with 703 molecular studies. E.F. performed vagotomy. C.M. helped with fiber photometry 704 experiments. G.G.M. and R.T. analyzed endocannabinoids levels. S.L. provided 705 scientific guidance and critical feedback. G.G. supervised the whole project, 706 interpreted the data and wrote the manuscript with contribution from all coauthors. 707 708 Competing interests 709 The authors declare no competing interests. 710 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 24

711 1. Berthoud H-R, Münzberg H, Morrison CD. Blaming the Brain for Obesity: 712 Integration of Hedonic and Homeostatic Mechanisms. Gastroenterology. 713 2017;152:1728–1738. 714 2. Lenard NR, Berthoud H-R. Central and peripheral regulation of food intake 715 and physical activity: pathways and genes. Obesity (Silver Spring). 2008;16 Suppl 716 3:S11-22. 717 3. McEwen BS, Wingfield JC. The concept of allostasis in biology and 718 biomedicine. Horm Behav. 2003;43:2–15. 719 4. George O, Le Moal M, Koob GF. Allostasis and addiction: role of the 720 dopamine and corticotropin-releasing factor systems. Physiol Behav. 2012;106:58–721 64. 722 5. Volkow ND, Wang G-J, Baler RD. Reward, dopamine and the control of food 723 intake: implications for obesity. Trends Cogn Sci (Regul Ed). 2011;15:37–46. 724 6. Mazier W, Saucisse N, Simon V, Cannich A, Marsicano G, Massa F, et al. 725 mTORC1 and CB1 receptor signaling regulate excitatory glutamatergic inputs onto 726 the hypothalamic paraventricular nucleus in response to energy availability. Mol 727 Metab. 2019;28:151–159. 728 7. Rossi MA, Basiri ML, McHenry JA, Kosyk O, Otis JM, van den Munkhof HE, et 729 al. Obesity remodels activity and transcriptional state of a lateral hypothalamic brake 730 on feeding. Science. 2019;364:1271–1274. 731 8. Beutler LR, Corpuz TV, Ahn JS, Kosar S, Song W, Chen Y, et al. Obesity 732 causes selective and long-lasting desensitization of AgRP neurons to dietary fat. 733 Elife. 2020;9. 734 9. Bai L, Mesgarzadeh S, Ramesh KS, Huey EL, Liu Y, Gray LA, et al. Genetic 735 Identification of Vagal Sensory Neurons That Control Feeding. Cell. 2019;179:1129-736 1143.e23. 737 10. Kaelberer MM, Buchanan KL, Klein ME, Barth BB, Montoya MM, Shen X, et 738 al. A gut-brain neural circuit for nutrient sensory transduction. Science. 2018;361. 739 11. de Lartigue G. Role of the vagus nerve in the development and treatment of 740 diet-induced obesity. J Physiol. 2016;594:5791–5815. 741 12. Fernandes AB, Alves da Silva J, Almeida J, Cui G, Gerfen CR, Costa RM, et 742 al. Postingestive Modulation of Food Seeking Depends on Vagus-Mediated 743 Dopamine Neuron Activity. Neuron. 2020;106:778-788.e6. 744 13. Han W, Tellez LA, Perkins MH, Perez IO, Qu T, Ferreira J, et al. A Neural 745 Circuit for Gut-Induced Reward. Cell. 2018;175:665-678.e23. 746 14. Hankir MK, Seyfried F, Hintschich CA, Diep T-A, Kleberg K, Kranz M, et al. 747 Gastric Bypass Surgery Recruits a Gut PPAR-α -Striatal D1R Pathway to Reduce Fat 748 Appetite in Obese Rats. Cell Metab. 2017;25:335–344. 749 15. Argueta DA, DiPatrizio NV. Peripheral endocannabinoid signaling controls 750 hyperphagia in western diet-induced obesity. Physiol Behav. 2017;171:32–39. 751 16. DiPatrizio NV, Joslin A, Jung K-M, Piomelli D. Endocannabinoid signaling in 752 the gut mediates preference for dietary unsaturated fats. FASEB J. 2013;27:2513–753 2520. 754 17. Gómez R, Navarro M, Ferrer B, Trigo JM, Bilbao A, Del Arco I, et al. A 755 peripheral mechanism for CB1 cannabinoid receptor-dependent modulation of 756 feeding. J Neurosci. 2002;22:9612–9617. 757 18. Monteleone P, Matias I, Martiadis V, De Petrocellis L, Maj M, Di Marzo V. 758 Blood levels of the endocannabinoid anandamide are increased in anorexia nervosa 759 and in binge-eating disorder, but not in bulimia nervosa. Neuropsychopharmacology. 760 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 25 2005;30:1216–1221. 761 19. Monteleone AM, Piscitelli F, Dalle Grave R, El Ghoch M, Di Marzo V, Maj M, 762 et al. Peripheral Endocannabinoid Responses to Hedonic Eating in Binge-Eating 763 Disorder. Nutrients. 2017;9. 764 20. Kuipers EN, Kantae V, Maarse BCE, van den Berg SM, van Eenige R, Nahon 765 KJ, et al. High Fat Diet Increases Circulating Endocannabinoids Accompanied by 766 Increased Synthesis Enzymes in Adipose Tissue. Front Physiol. 2018;9:1913. 767 21. Berland C, Montalban E, Perrin E, Di Miceli M, Nakamura Y, Martinat M, et al. 768 Circulating Triglycerides Gate Dopam ine-Associated Beha viors through DRD2-769 Expressing Neurons. Cell Metab. 2020;31:773-790.e11. 770 22. Patriarchi T, Cho JR, Merten K, Howe MW, Marley A, Xiong W-H, et al. 771 Ultrafast neuronal imaging of dopamine dynamics with designed genetically encoded 772 sensors. Science. 2018;360. 773 23. Avena NM. The study of food addiction using animal models of binge eating. 774 Appetite. 2010;55:734–737. 775 24. DiFeliceantonio AG, Coppin G, Rigoux L, Edwin Thanarajah S, Dagher A, 776 Tittgemeyer M, et al. Supra-Additive Effects of Combining Fat and Carbohydrate on 777 Food Reward. Cell Metab. 2018;28:33-44.e3. 778 25. Oosterman JE, Koekkoek LL, Foppen E, Unmehopa UA, Eggels L, Verheij J, 779 et al. Synergistic Effect of Feeding Time and Diet on Hepatic Steatosis and Gene 780 Expression in Male Wistar Rats. Obesity (Silver Spring). 2020;28 Suppl 1:S81–S92. 781 26. Wang G-J, Geliebter A, Volkow ND, Telang FW, Logan J, Jayne MC, et al. 782 Enhanced striatal dopamine release during food stimulation in binge eating disorder. 783 Obesity (Silver Spring). 2011;19:1601–1608. 784 27. Spierling S, de Guglielmo G, Kirson D, Kreisler A, Roberto M, George O, et al. 785 Insula to ventral striatal projections mediate compulsive eating produced by 786 intermittent access to palatable food. Neuropsychopharmacology. 2020;45:579–588. 787 28. Biever A, Puighermanal E, Nishi A, David A, Panciatici C, Longueville S, et al. 788 PKA-dependent phosphorylation of ribosomal protein S6 does not correlate with 789 translation efficiency in striatonigral and striatopallidal medium-sized spiny neurons. J 790 Neurosci. 2015;35:4113–4130. 791 29. Gangarossa G, Perroy J, Valjent E. Combinatorial topography and cell-type 792 specific regulation of the ERK pathway by dopaminergic agonists in the mouse 793 striatum. Brain Struct Funct. 2013;218:405–419. 794 30. Radl D, Chiacchiaretta M, Lewis RG, Brami-Cherrier K, Arcuri L, Borrelli E. 795 Differential regulation of striatal motor behavior and related cellular responses by 796 dopamine D2L and D2S isoforms. Proc Natl Acad Sci U S A. 2018;115:198–203. 797 31. Reichelt AC, Westbrook RF, Morris MJ. Integration of reward signalling and 798 appetite regulating peptide systems in the control of food-cue responses. Br J 799 Pharmacol. 2015;172:5225–5238. 800 32. de Araujo IE, Schatzker M, Small DM. Rethinking Food Reward. Annual 801 Review of Psychology. 2020;71:139–164. 802 33. DiPatrizio NV, Piomelli D. Intestinal lipid-derived signals that sense dietary fat. 803 J Clin Invest. 2015;125:891–898. 804 34. Lau BK, Cota D, Cristino L, Borgland SL. Endocannabinoid modulation of 805 homeostatic and non-homeostatic feeding circuits. Neuropharmacology. 806 2017;124:38–51. 807 35. Cluny NL, Vemuri VK, Chambers AP, Limebeer CL, Bedard H, Wood JT, et al. 808 A novel peripherally restricted cannabinoid receptor antagonist, AM6545, reduces 809 food intake and body weight, but does not cause malaise, in rodents. Br J 810 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 26 Pharmacol. 2010;161:629–642. 811 36. Tam J, Cinar R, Liu J, Godlewski G, Wesley D, Jourdan T, et al. Peripheral 812 cannabinoid-1 receptor inverse agonism reduces obesity by reversing leptin 813 resistance. Cell Metab. 2012;16:167–179. 814 37. Tam J, Vemuri VK, Liu J, Bátkai S, Mukhopadhyay B, Godlewski G, et al. 815 Peripheral CB1 cannabinoid receptor blockade improves cardiometabolic risk in 816 mouse models of obesity. J Clin Invest. 2010;120:2953–2966. 817 38. Long JZ, Li W, Booker L, Burston JJ, Kinsey SG, Schlosburg JE, et al. 818 Selective blockade of 2-arachidonoylglycerol hydrolysis produces cannabinoid 819 behavioral effects. Nat Chem Biol. 2009;5:37–44. 820 39. Sykaras AG, Demenis C, Case RM, McLaughlin JT, Smith CP. Duodenal 821 enteroendocrine I-cells contain mRNA transcripts encoding key endocannabinoid and 822 fatty acid receptors. PLoS One. 2012;7:e42373. 823 40. Argueta DA, Perez PA, Makriyannis A, DiPatrizio NV. Cannabinoid CB1 824 Receptors Inhibit Gut-Brain Satiation Signaling in Diet-Induced Obesity. Front 825 Physiol. 2019;10:704. 826 41. Egerod KL, Petersen N, Timshel PN, Rekling JC, Wang Y, Liu Q, et al. 827 Profiling of G protein-coupled receptors in vagal afferents reveals novel gut-to-brain 828 sensing mechanisms. Mol Metab. 2018;12:62–75. 829 42. Grill HJ, Hayes MR. Hindbrain neurons as an essential hub in the 830 neuroanatomically distributed control of energy balance. Cell Metab. 2012;16:296–831 309. 832 43. D’Agostino G, Lyons DJ, Cristiano C, Burke LK, Madara JC, Campbell JN, et 833 al. Appetite controlled by a cholecystokinin nucleus of the solitary tract to 834 hypothalamus neurocircuit. Elife. 2016;5. 835 44. Cheng W, Gonzalez I, Pan W, Tsang AH, Adams J, Ndoka E, et al. Calcitonin 836 Receptor Neurons in the Mouse Nucleus Tractus Solitarius Control Energy Balance 837 via the Non-aversive Suppression of Feeding. Cell Metab. 2020;31:301-312.e5. 838 45. Kolata SM, Nakao K, Jeevakumar V, Farmer-Alroth EL, Fujita Y, Bartley AF, et 839 al. Neuropsychiatric Phenotypes Produced by GABA Reduction in Mouse Cortex and 840 Hippocampus. Neuropsychopharmacology. 2018;43:1445–1456. 841 46. Takeuchi T, Duszkiewicz AJ, Sonneborn A, Spooner PA, Yamasaki M, 842 Watanabe M, et al. Locus coeruleus and dopaminergic consolidation of everyday 843 memory. Nature. 2016;537:357–362. 844 47. Hutson PH, Balodis IM, Potenza MN. Binge-eating disorder: Clinical and 845 therapeutic advances. Pharmacol Ther. 2018;182:15–27. 846 48. Carr MM, Grilo CM. Examining heterogeneity of binge-eating disorder using 847 latent class analysis. J Psychiatr Res. 2020;130:194–200. 848 49. Naish KR, Laliberte M, MacKillop J, Balodis IM. Systematic review of the 849 effects of acute stress in binge eating disorder. Eur J Neurosci. 2019;50:2415–2429. 850 50. Bake T, Murphy M, Morgan DGA, Mercer JG. Large, binge-type meals of high 851 fat diet change feeding behaviour and entrain food anticipatory activity in mice. 852 Appetite. 2014;77:60–71. 853 51. Muñoz-Escobar G, Guerrero-Vargas NN, Escobar C. Random access to 854 palatable food stimulates similar addiction-like responses as a fixed schedule, but 855 only a fixed schedule elicits anticipatory activation. Sci Rep. 2019;9:18223. 856 52. Ramsay DS, Woods SC. Clarifying the roles of homeostasis and allostasis in 857 physiological regulation. Psychol Rev. 2014;121:225–247. 858 53. De Ridder D, Manning P, Leong SL, Ross S, Vanneste S. Allostasis in health 859 and food addiction. Sci Rep. 2016;6:37126. 860 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 27 54. Luo Z, Volkow ND, Heintz N, Pan Y, Du C. Acute cocaine induces fast 861 activation of D1 receptor and progressive deactivation of D2 receptor striatal 862 neurons: in vivo optical microprobe [Ca2+]i imaging. J Neurosci. 2011;31:13180–863 13190. 864 55. Kai N, Nishizawa K, Tsutsui Y, Ueda S, Kobayashi K. Differential roles of 865 dopamine D1 and D2 receptor-containing neurons of the nucleus accumbens shell in 866 behavioral sensitization. J Neurochem. 2015;135:1232–1241. 867 56. Bertran-Gonzalez J, Bosch C, Maroteaux M, Matamales M, Hervé D, Valjent 868 E, et al. Opposing patterns of signaling activation in dopamine D1 and D2 receptor-869 expressing striatal neurons in response to cocaine and haloperidol. J Neurosci. 870 2008;28:5671–5685. 871 57. Kenny PJ, Voren G, Johnson PM. Dopamine D2 receptors and striatopallidal 872 transmission in addiction and obesity. Curr Opin Neurobiol. 2013;23:535–538. 873 58. Caravaggio F, Raitsin S, Gerretsen P, Nakajima S, Wilson A, Graff-Guerrero 874 A. Ventral striatum binding of a dopamine D2/3 receptor agonist but not antagonist 875 predicts normal body mass index. Biol Psychiatry. 2015;77:196–202. 876 59. Han W, Tellez LA, Niu J, Medina S, Ferreira TL, Zhang X, et al. Striatal 877 Dopamine Links Gastrointestinal Rerouting to Altered Sweet Appetite. Cell Metab. 878 2016;23:103–112. 879 60. Tellez LA, Han W, Zhang X, Ferreira TL, Perez IO, Shammah-Lagnado SJ, et 880 al. Separate circuitries encode the hedonic and nutritional values of sugar. Nat 881 Neurosci. 2016;19:465–470. 882 61. Alhadeff AL, Goldstein N, Park O, Klima ML, Vargas A, Betley JN. Natural and 883 Drug Rewards Engage Distinct Pathways that Converge on Coordinated 884 Hypothalamic and Reward Circuits. Neuron. 2019;103:891-908.e6. 885 62. Cone JJ, McCutcheon JE, Roitman MF. Ghrelin acts as an interface between 886 physiological state and phasic dopamine signaling. J Neurosci. 2014;34:4905–4913. 887 63. Fulton S, Pissios P, Manchon RP, Stiles L, Frank L, Pothos EN, et al. Leptin 888 regulation of the mesoaccumbens dopamine pathway. Neuron. 2006;51:811–822. 889 64. Reddy IA, Smith NK, Erreger K, Ghose D, Saunders C, Foster DJ, et al. Bile 890 diversion, a bariatric surgery, and bile acid signaling reduce central cocaine reward. 891 PLoS Biol. 2018;16:e2006682. 892 65. Bellocchio L, Soria-Gómez E, Quarta C, Metna-Laurent M, Cardinal P, Binder 893 E, et al. Activation of the sympathetic nervous system mediates hypophagic and 894 anxiety-like effects of CB ₁ receptor blockade. Proc Natl Acad Sci U S A. 895 2013;110:4786–4791. 896 66. Godlewski G, Cinar R, Coffey NJ, Liu J, Jourdan T, Mukhopadhyay B, et al. 897 Targeting Peripheral CB1 Receptors Reduces Ethanol Intake via a Gut-Brain Axis. 898 Cell Metab. 2019;29:1320-1333.e8. 899 67. Bohórquez DV, Shahid RA, Erdmann A, Kreger AM, Wang Y, Calakos N, et al. 900 Neuroepithelial circuit formed by innervation of sensory enteroendocrine cells. J Clin 901 Invest. 2015;125:782–786. 902 68. Haber AL, Biton M, Rogel N, Herbst RH, Shekhar K, Smillie C, et al. A single-903 cell survey of the small intestinal epithelium. Nature. 2017;551:333–339. 904 69. Ruiz de Azua I, Mancini G, Srivastava RK, Rey AA, Cardinal P, Tedesco L, et 905 al. Adipocyte cannabinoid receptor CB1 regulates energy homeostasis and 906 alternatively activated macrophages. J Clin Invest. 2017;127:4148–4162. 907 70. Roman CW, Derkach VA, Palmiter RD. Genetically and functionally defined 908 NTS to PBN brain circuits mediating anorexia. Nat Commun. 2016;7:11905. 909 71. Carter ME, Soden ME, Zweifel LS, Palmiter RD. Genetic identification of a 910 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 28 neural circuit that suppresses appetite. Nature. 2013;503:111–114. 911 72. Li MM, Madara JC, Steger JS, Krashes MJ, Balthasar N, Campbell JN, et al. 912 The Paraventricular Hypothalamus Regulates Satiety and Prevents Obesity via Two 913 Genetically Distinct Circuits. Neuron. 2019;102:653-667.e6. 914 73. Manta S, El Mansari M, Debonnel G, Blier P. Electrophysiological and 915 neurochemical effects of long-term vagus nerve stimulation on the rat monoaminergic 916 systems. Int J Neuropsychopharmacol. 2013;16:459–470. 917 74. Perez SM, Carreno FR, Frazer A, Lodge DJ. Vagal nerve stimulation reverses 918 aberrant dopamine system function in the methylazoxymethanol acetate rodent 919 model of schizophrenia. J Neurosci. 2014;34:9261–9267. 920 75. Morales M, Margolis EB. Ventral tegmental area: cellular heterogeneity, 921 connectivity and behaviour. Nat Rev Neurosci. 2017;18:73–85. 922 76. Alhadeff AL, Rupprecht LE, Hayes MR. GLP-1 neurons in the nucleus of the 923 solitary tract project directly to the ventral tegmental area and nucleus accumbens to 924 control for food intake. Endocrinology. 2012;153:647–658. 925 77. Nieh EH, Vander Weele CM, Matthews GA, Presbrey KN, Wichmann R, 926 Leppla CA, et al. Inhibitory Input from the Lateral Hypothalamus to the Ventral 927 Tegmental Area Disinhibits Dopamine Neurons and Promotes Behavioral Activation. 928 Neuron. 2016;90:1286–1298. 929 78. Faget L, Osakada F, Duan J, Ressler R, Johnson AB, Proudfoot JA, et al. 930 Afferent Inputs to Neurotransmitter-Defined Cell Types in the Ventral Tegmental 931 Area. Cell Rep. 2016;15:2796–2808. 932 79. Wang X-F, Liu J-J, Xia J, Liu J, Mirabella V, Pang ZP. Endogenous Glucagon-933 like Peptide-1 Suppresses High-Fat Food Intake by Reducing Synaptic Drive onto 934 Mesolimbic Dopamine Neurons. Cell Rep. 2015;12:726–733. 935 80. Beier KT, Steinberg EE, DeLoach KE, Xie S, Miyamichi K, Schwarz L, et al. 936 Circuit Architecture of VTA Dopamine Neurons Revealed by Systematic Input-Output 937 Mapping. Cell. 2015;162:622–634. 938 81. Clemmensen C, Müller TD, Woods SC, Berthoud H-R, Seeley RJ, Tschöp 939 MH. Gut-Brain Cross-Talk in Metabolic Control. Cell. 2017;168:758–774. 940 941 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 29 Figure legends 942 943 Fig. 1: Allostatic adaptations of metabolic efficiency to time-locked access to 944 palatable diet. (A) Experimental design. Control (Ctr) or bingeing animals (Binge) 945 had daily intermittent access to water or a palatable lipids/sucrose mixture for 1 946 hour/day during 10-14 consecutive days. Regular chow pellets were provided ad 947 libitum throughout the entire experiment. ( B) Daily binge consumption (ml) of 948 palatable mixture during a 14-days protocol. Statistics: ***p<0.001 Binge vs Control. 949 (C) 24 hrs locomotor activity in calorimetric chambers (average of 3 consecutive 950 days). Red dotted rectangles indicate the locomotor activity 2-hrs prior and after 951 palatable food access. (C1) Cumulative locomotor activity 2-hrs prior and after 952 palatable food access. Results are expressed as beam breaks (bb). Statistics: 953 *p<0.05 and ***p<0.001 Binge vs Control. (D) Temporal pattern of regular chow food 954 intake (FI, kcal/h) during 24 hrs (average of 3 consecutive days). Statistics: **p<0.01 955 Binge vs Control. (D1) Cumulative chow food intake during the dark period. Statistics: 956 ***p<0.001 Binge vs Control. (E) 24 hrs food intake considering all calories: standard 957 diet (SD) and palatable food (PF). Statistics: ***p<0.001 Binge(SD) vs Control(SD), 958 ###p<0.001 Binge(SD+PF) vs Binge(SD). ( F) Body weight throughout the 14-days 959 experimental procedure. (G) Longitudinal profile of the respiratory energy ratio (RER) 960 from indirect calorimetry (average of 3 consecutive days) and ( G1) averaged RER 961 values 2-hrs prior and after palatable food access. Statistics: **p<0.01 and 962 ***p<0.001 Binge vs Control. (H) Longitudinal profile of energy expenditure (EE) from 963 indirect calorimetry (average of 3 consecutive days) and ( H1) averaged EE values 2-964 hrs prior and after palatable food access. Statistics: *p<0.05 and **p<0.01 Binge vs 965 Control. ( I) Brown adipose tissue (BAT) temperature during bingeing. Statistics: 966 *p<0.05 and **p<0.01 Binge vs Control. ( J) Real-time core temperature recording 967 during 24 hrs and ( J1) averaged values 2-hrs prior and after palatable food access. 968 Statistics: ***p<0.001 Binge vs Control. (K ) Matching locomotor activity from core 969 temperature measurements. Statistics: ***p<0.001 Binge vs Control. For number of 970 mice/group and statistical details see Suppl. Table 1. 971 972 Fig. 2: Binge eating induces dopamine D1R-related molecular modifications . 973 (A, B) Protein quantification of phospho-ERK, S6 S235/236 and S6S240/244 in the DS ( A) 974 and NAc ( B). For immunoblot pictures see Suppl. Fig. 2C. Statistics: *p<0.05, 975 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 30 **p<0.01 and ***p<0.001, comparisons to control mice. ( C, D) Temporal profile of 976 locomotor activity and cumulative locomotor response (C1 and D1) of animals treated 977 with the dopamine transporter blocker GBR during the anticipatory phase ( C, C1) or 978 1-hour after intermittent access to water (Ctr + GBR) or palatable diet (Binge + GBR) 979 (D, D1). Results are expressed as beam breaks (bb). Statistics: **p<0.01 Binge+GBR 980 vs Control+GBR. ( E) Palatable diet intake after vehicle (Veh+Binge) or D1R 981 antagonist SCH23390 (SCH+Binge) pretreatment. Note: SCH23390 was acutely 982 administered 30 min before binge session. Statistics: ***p<0.001 SCH+Binge vs 983 Veh+Binge. ( F) Palatable diet intake after vehicle (Veh+Binge) or D2R antagonist 984 haloperidol 0.25 mg/kg or 0.5 mg/kg (H 0.25+Binge and H 0.5+Binge) treatment. Note: 985 haloperidol was acutely administered 30 min before binge session. ( G) 986 Immunolabeling of phospho-S6 in the DS and NAc (for immunolabeling images of 987 NAc see Suppl. Fig. 2E) and their associated quantifications (G1, G2, H1, H2) in mice 988 pretreated with SCH23390 or vehicle and exposed to time-locked palatable diet. 989 Scale bar: 50 μ m. Statistics: ***p<0.001 Veh+Binge vs Veh+Control, ###p<0.001 990 SCH+Binge vs Veh+Binge. ( I) Temporal profile of locomotor activity and cumulative 991 locomotor response ( I1) of animals receiving SCH (SCH+Binge) or vehicle 992 (Veh+Binge) (red arrow) and access to palatable diet (black arrow). Statistics: 993 **p<0.01 SCH+Binge vs Veh+Binge. ( J) Cumulative regular chow diet intake 994 following acute SCH23390 (SCH+Binge) or vehicle (Veh+Binge). Statistics: **p<0.01 995 SCH+Binge vs Veh+Binge. (K) Temporal profile of locomotor activity and cumulative 996 locomotor response (2 hrs and 30 min, K1) induced by the D1R agonist SKF81297 997 administered 1 hour after access to time-locked water (Ctr+SKF) or palatable diet 998 (Binge+SKF). Statistics: *p<0.05 and **p<0.01 Binge+SKF vs Control+SKF. For 999 number of mice/group and statistical details see Suppl. Table 1. 1000 1001 Fig. 3: Peripheral endocannabinoids (eCBs) govern binge eating. (A) Palatable 1002 bingeing in animals pretreated (1h prior binge session) with vehicle (Veh), leptin, 1003 insulin, GLP1 agonists exendin-4 (Exe4) and liraglutide (Lira), CCK octapeptide 1004 sulfated (CCK-8S) or CB1R inverse agonist AM251. Aside leptin (2 injections/day for 1005 2 consecutive days, Suppl. Fig. 3A, B), drugs were administered acutely. Statistics: 1006 ***p<0.001 Exe4-, Lira-, CCK-8S- & AM251-treated Bingeing mice vs Veh+Binge 1007 mice, ###p<0.001 AM251-treated vs Exe4-, Lira & CCK-8S-treated bingeing mice. (B) 1008 Dosage of peripheral and circulating endocannabinoids: anandamide (AEA), 1009 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 31 diacylglycerol (2-AG), docosahexanoyl ethanolamide (DHEA) and 1010 oleoylethanolamide (OEA) 1 hour before and after palatable bingeing. ( C) Palatable 1011 bingeing in mice (3 males and females/group) pre-treated with a single i.p. injection 1012 of vehicle (Veh), peripheral CB1R antagonist AM6545 (10 mg/kg) and peripheral 1013 CB1R inverse agonist JD-5037 (3 mg/kg). Statistics: ***p<0.001 AM6545 and JD-1014 5037 vs Veh-Binge. (D) Palatable bingeing in mice pre-treated with a single i.p. 1015 injection of vehicle (Veh), peripheral CB1R antagonist AM6545 (10 mg/kg), and/or 1016 monoacylglycerol lipase inhibitor JZL184 (8 mg/kg). Statistics: ***p<0.001 AM6545, 1017 JZL184, AM6545+JZL184 vs Veh-Binge. ( E) Chronic (4 days) administration of 1018 JZL184 and AM6545 on palatable bingeing. Statistics: ***p<0.001 AM6545-Binge vs 1019 Veh-Binge, ###p<0.001 JZL184-Binge vs Veh-Binge. (F ) Effects of acute AM6545 on 1020 core temperature. Statistics: **p<0.01 AM6545-Binge vs Veh-Binge. Note: black and 1021 red arrows indicate administration of AM6545 and palatable food access, 1022 respectively. (G) Immunolabeling and quantifications of phospho-S6 and cFos in the 1023 DS and NAc of control or bingeing animals (males and females) acutely pretreated 1024 with Veh or AM6545. Scale bars: 50 μ m. Statistics: ***p<0.001 and *p<0.005 1025 Veh+Binge or AM6545+Binge vs Ctr, ###p<0.001 AM6545+Binge vs Veh+Binge. (H) 1026 Longitudinal measurement of fatty acid oxidation (FAO) following administration of 1027 AM6545 during a Binge session and a NoBinge session. ( H1) Averaged FAO from 1028 time of injection (11h00) till the end of light phase (19h00). ( H2) Ratio of FAO and 1029 food intake (FI) to discriminate between the effect of AM6545 and calories intake. 1030 Statistics: ***p<0.001 AM6545 vs Veh (in both Binge and NoBinge sessions). ( I) 1031 Palatable bingeing after acute oral gavage of AM6545 (10 mg/kg, p.o.) and (J) 1032 associated fatty acid oxidation. ( K, L) Immunolabeling and quantifications of cFos in 1033 the cNTS/AP regions ( K) and in the lPBN (L ) of control or bingeing animals (males 1034 and females) acutely pretreated with Veh or AM6545. Scale bars: 250 μ m. Statistics: 1035 ***p<0.001, **p<0.01 and *p<0.005 Veh+Binge or AM6545+Binge vs Ctr, ###p<0.001 1036 AM6545+Binge vs Veh+Binge. For number of mice/group and statistical details see 1037 Suppl. Table 1. 1038 1039 Fig. 4: The gut-brain vagal axis is required for eCBs-mediated effects. (A) The 1040 scheme indicates gut-originated afferent paths that were virally targeted to perform 1041 single-cell transcriptomic analysis [9]. ( B) Enrichment of different vagal markers 1042 (SLC17a6, Scn10a, Htr3a, Cartpt, Grin1, Phox2b) and comparison with Cnr1 and 1043 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 32 Cnr2 in sensory vagal neurons labeled from microinjections in the stomach, proximal 1044 and middle intestines. Note: for other gut-brain vagal segments see Suppl. Fig. 3H, 1045 I. (C) Palatable food consumption in sham and vagotomized (VGX) animals pre-1046 treated with Veh (Day 10) or JZL184 (Day 11-14) 2-hrs before bingeing sessions. 1047 Statistics: ***p<0.001 Sham+JZL184 vs Sham+Veh, ###p<0.001 VGX+JZL184 vs 1048 Sham+JZL184. (D) cFos immunolabeling in the area postrema (AP), caudal nucleus 1049 tractus solitarius (cNTS), lateral parabrachial nucleus (lPBN) and medial parabrachial 1050 nucleus (mPBN) in sham and vagotomized animals treated with the peripheral CB1R 1051 antagonist AM6545 (10 mg/kg). Scale bars: 250 μ m. ( D1) Scheme indicates the 1052 central vagus → cNTS→ PBN→ target regions path in sham and VGX mice. (D 2) 1053 Quantification of cFos-positive neurons in the AP, cNTS and lPBN in sham and VGX 1054 mice injected with AM6545. Statistics: ***p<0.001 VGX+AM6545 vs Sham+AM6545. 1055 (E) Palatable bingeing in sham and vagotomized (VGX) animals pre-treated with 1056 AM6545 (A) or vehicle (V), and associated measurements of fatty acid oxidation ( F, 1057 F1 and G, G 1). Statistics: ***p<0.001 Sham+AM6545 vs Sham+Veh. (H, I ) cFos 1058 immunolabeling in the paraventricular nucleus (PVN) and dorsomedial nucleus of the 1059 hypothalamus (DMH) of sham or VGX animals acutely treated with vehicle or 1060 AM5646 and associated counting. Scale bars: 250 μ m. Statistics: ***p<0.001 1061 Sham+AM6545 vs Veh, ###p<0.001 VGX+AM6545 vs Sham+AM6545. For number of 1062 mice/group and statistical details see Suppl. Table 1. 1063 1064 Fig. 5: Peripheral CB1R signaling modulates dopamine dynamics. (A, A1) Effect 1065 of AM6545 or Veh on GBR-induced locomotor activity (beam breaks, bb). Statistics: 1066 **p<0.01 AM6545+GBR vs Veh+GBR. ( B, B 1) Effect of JD-5037 or Veh on GBR-1067 induced locomotor activity. Statistics: **p<0.01, *p<0.05 AM6545+GBR vs Veh+GBR. 1068 (C) Effect of AM6545 on GBR-triggered cFos expression in the striatum. Scale bar: 1069 50 μm. Statistics: ***p<0.001 AM6545+GBR vs Veh+GBR. ( D) Longitudinal profile 1070 and heat maps of GBR-induced accumbal DA accumulation (dLight1.2) in mice pre-1071 treated with Veh or AM6545 1h before GBR12909 (red arrow). ( D1) Quantification of 1072 bulk fluorescence (AUC) in Veh+GBR and AM6545+GBR groups. Statistics: 1073 ***p<0.001 AM6545+GBR vs Veh+GBR. (E) Cumulative locomotor activity response 1074 in mice pretreated with vehicle (Veh+Amph) or AM6545 (AM6545+Amph). For 1075 temporal locomotor activity see Suppl. Fig. 6A. ( F) Cumulative locomotor activity 1076 response in mice pretreated with vehicle (Veh+Amph) or AM251 (AM251+Amph). 1077 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint 33 Statistics: ***p<0.001 AM251+Amph vs Veh+Amph. ( G) GBR-induced locomotor 1078 activity, cumulative locomotor response ( G1) in VGX mice pretreated with vehicle 1079 (VGX/Veh+GBR) or AM6545 (VGX/AM6545+GBR). ( H) Cumulative locomotor 1080 response in mice pretreated with oral gavage (po) of vehicle (Veh (po)+GBR) or 1081 AM6545 (AM6545 (po)+GBR). For temporal locomotor activity see Suppl. Fig. 6B. 1082 1083 Fig. 6: Peripheral CB1R signaling routed by the vagus nerve controls VTA DA-1084 neurons activity. (A) Expression of GCaMP6f in VTA DA-neurons of virally injected 1085 Drd2-Cre mice. Please, note ( i) colocalization with TH and GCaMP6f-positive 1086 neurons and (ii) projecting terminals in the DS and NAc. See also, Suppl. Fig. 6C for 1087 TH expression in the VTA of Drd2 -eGFP mice. (B ) Behavioral paradigms used to 1088 trigger the activity of VTA DA-neurons: tail suspension and exposure to a new 1089 environment (NE). [For validation of in vivo recording of Ca2+ transients in VTA D2R-1090 (DA)-neurons see Suppl. Fig. 6D, E ]. (C, D) Temporal dynamics and corresponding 1091 heat maps of DA-neurons activity during the during the tail suspension test ( C) and 1092 exposure to a new environment ( D). Statistics: *p<0.05, **p<0.01 AM6545 vs Veh. 1093 (E) Immobility time (sec) of sham and VGX mice acutely pretreated with Veh or 1094 AM6545 1 hour before tail suspension (6 min). ( F) Effect of AM6545 or Veh in sham 1095 and VGX mice on novel environment-induced locomotor activity. ( F1) Cumulative 1096 locomotor activity response in sham and VGX mice pretreated with vehicle or 1097 AM6545. Statistics: **p<0.01 Sham/AM6545 vs Sham/Veh. (G) Cumulative locomotor 1098 activity response in sham and VGX mice pretreated with vehicle or AM251. Statistics: 1099 **p<0.01 Sham+AM251 vs Sham+Veh, and ##p<0.01 VGX+AM251 vs VGX+AM251. 1100 For number of mice/group and statistical details see Suppl. Table 1. 1101 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted October 14, 2021. ; https://doi.org/10.1101/2020.11.14.382291doi: bioRxiv preprint