In 2016,over 650 million individuals were obese (body mass index>30), representing13% of the adult world population (1). The high prevalence of obesity isa major public health concern because obesity is causally associated withnumerous comorbidities (2) and increases mortality (3). Behavioral treatment options forobesity (calorie restriction, exercise) are only moderately effective to inducelong-term weight loss (4,5), whereas bariatric surgicalprocedures are more effective but irreversible and linked with increasedpost-surgical mortality (6). Therefore, the development of neweffective and safe anti-obesity strategies is still required.Because obesity is primarily due tochronic overconsumption of food with excess energy stored as fat, thephysiological control of appetite has been extensively studied, and criticalpathways have been pharmacologically targeted to reduce eating. Notably, theglucagon-like peptide-1, a gut hormone primarily known for its incretin action,inhibits eating and gastric emptying by activating peripheral and centralmechanisms (7-9).
Therefore, GLP-1 analogssuch as exenatide and liraglutide have been developed, and they have beenapproved for the treatment of T2D and, more recently, obesity (10,11). There is increasing evidence that, beyondclassical satiation, many anorexigenic signals suppress reward function (12). This is also the case for exogenously-administeredGLP-1 or GLP-1R agonists in animal models and inhumans (13-18). As GLP-1 is also producedin a discrete cell populations within the brain (19,20), it is difficult to testwhether the behavioral effects observed following exogenous delivery of GLP-1or GLP-1R agonists mimic an afferent gut-brain signal of a CNS-derived signal. However,because GLP-1R agonists penetrate the CNS even when administered peripherally (21), and because directactivation of central GLP-1R reduces food reward (22), current prevailing hypothesis is that the rewardeffect of exogenously administered GLP-1/GLP-1R agonist is primarily mediatedby the activation of central GLP-1R.KS1 Scholars in the field, however, have not yet adequately addressed whether endogenous intestinalGLP-1 can modulate food reward via a gut-brain afferent mechanism. Elucidatingthis question would a/ increase our current knowledge of the endogenous GLP-1system and b/ open new opportunities to modulate food reward via a gut-brainpathway, notably via vagus nerve stimulation.
1.1.1 OwnresearchThe vagus nerve is a direct neural pathway connecting peripheralorgans and the central nervous system. Notably, vagal afferent terminals in thegut sense postprandial signals, including hormones released fromenteroendocrine cells, mechanical distension, or nutrients, and relay thisinformation to neurons of the nucleus tractus solitarius (NTS) in the brainstem(23).
The phenotype and function of vagal afferent neurons(VAN) only begin to be unraveled (24-26).A subpopulation of vagalafferents innervating the gastrointestinal tract expresses GLP-1R and isideally positioned to detect gut-derived GLP-1 (27). To test the role of VAN GLP-1R, wedeveloped a viral vector to knockdown glp1rin rat VAN via shRNA, and showed that a full expression of glp1r is required for the normal control of meal size and post-mealinsulin release (28). Using the same model, we showedthat vagal GLP-1R signaling engages a gut-brain-brown adipose tissue crosstalk thattunes post-meal energy expenditure (29). Together, our findings demonstratedthat VAN GLP-1R signaling mediates physiological effects that are partiallyredundant (satiation (30)) but also partially divergent(brown adipose tissue thermogenesis (31)) to those of central GLP1-Ractivation.Severalpreliminary observations, including ours, are consistent with the idea that GLP-1 reduces food reward viavagal afferentsKS2 . First, in ratmodels of subdiaphragmatic vagal deafferentation (SDA), spontaneous consumptionof saccharin solutions, but not normal chow food is reduced (32) (Fig. 1A).
In the same experiments,SDA but not Sham rats showed avoidance of an Ex4-paired sweet solution,indicating that vagal GLP-1R activation may counterbalanceKS3 the effects of central GLP-1Ractivation on food preference (Fig.1B). Moreover, rats with intact vagalafferents and injected with a subthreshold dose of the GLP-1R antagonist (ie adose not causing an increase in food intake in chow-fed rats), decreases theintake of a newly-introduced high-fat diet, but not chow (unpublishedobservations, Krieger, Fig. 1C). In addition, VAN glp1r kdrats showed decreased consumption of high-fat diet upon introduction ofthe diet (days 0-3, Fig. 1D-E), and reduced time spent in the reward-paired side in a CPP test inpreliminary experiments (unpublished observations, Krieger, Fig. 1D-F).
1.2 Hypothesisand AimsAgainstthis background, this research project aims to test the hypothesis that activationof glp1r-expressing vagal afferentneurons modulates the reward value of food. Therefore, the goals of thisresearch project are:Aim 1: Test whether the pharmacologicalactivation of peripheral GLP-1R modulates food reward, and whether this effectrequires intact subdiaphragmatic vagal afferents.
Aim 2: Investigate whether vagal afferentsexpressing glp1r are activated by theintake of rewarding food and whether chemogenetic manipulation of these neuronsmodulates food reward.Aim 3: Elucidate the central connectome ofglp1r-expressing vagal afferents, inparticular, whether they trigger the activation of central reward circuits.1.3 Researchquestions and methodsDespitepreliminary observations in rats, the mice model will allow here a more precisemanipulation of glp1r-expressingvagal afferents1.3.1 Aim1: Test whether the pharmacological activation of peripheral GLP-1R modulatesfood reward, and whether this effect requires intact subdiaphragmatic vagalafferents.1.
3.1.a Question 1: Does peripheralGLP-1R activation alone modulate food reward?Overview: BecauseGLP-1 and GLP-1R agonists are believed to enter the brain and directly activatecentral GLP-1R, it is difficult to test the specific role of peripheral GLP-1Rusing pharmacological administration of these compounds. In this experiment(Fig. 2A), we will inject C57/BL6 mice with an exendin-4 conjugate that doesnot cross the blood-brain-barrier (BBB) and compare its effect to those ofvehicle and exendin-4 injection in two well-established food reward paradigms: conditionedplace preference (CPP) and progressive ratiooperantKS4 responding conditioning(POR)(Fig. 2C)Methods: Mice: C57/BL6Drugs: Ex4 ip 1ug/kg (enough for OR?KS5 )Ex-4conjugate name? ip dose? evidence? Ref?)KS6 CPP: In theconditioned place preference paradigm, animals are expected to spend more timein an environment previously associated with a food reward. The apparatusconsists of two connected chambers with distinct visual drawing on the wallsand tactile protrusions on the floor. Mice will be previously habituated to thereward (chocolate pellets, Kraft foods) in their home cage during two dailysessions of 15 min, and to the CPP box.
Then, 20 conditioning sessions of 20min each (2 sessions per day) will be conducted, in which mice havealternatively access to only one of the chambers. Ex4, the Ex4-conjugate or avehicle solution will be administered 10 minutes prior to each conditioningsession. On the test day, animals have access to both chambers, without rewardor water.
The CPP is expressed as the percent increase in time spent in thereward-paired side on the test day compared to the habituation session (13). OR: Inthe OR paradigm, motivation to work for sucrose is tested. Food-induced operantconditioning training and testing will be conducted in conditioning chambers withtwo retractable levers with white light bulbs above them, and a food pelletdispenser that can deliver 45 mg of sucrose pellets (Test Diet, Glaxo-SmithKline)to the food tray. Mice will be trained to press a lever for reward using afour-stage procedure as previously-described (13), until the number of pellets earnedby session does not differ by more than 15% for three consecutive session. Micewill receive Ex4, the Ex4-conjugate or a vehicle solution 10 minutes prior tooperant response testing.1.
3.1.b Question 2: Aresubdiaphragmatic vagal afferents required for the effects of peripheral GLP-1Ractivation on food reward?Overview: Based on the hypothesis that activation of glp1r-expressing vagal afferent neuronsmodulates the reward value of food, we hypothesize that the effect of theexendin-4 conjugate requires intact subdiaphragmatic vagal afferents. C57/BL6mice will undergo subdiaphragmaticvagotomy KS7 or Sham surgery. After 2 weeks ofrecovery, mice will receive injections of the exendin-4 conjugate during theacquisition phase of a CPP, or during OR (Fig. 2B).
Methods: Mice, drugs andbehavioral testing: same as 1.3.1.
a.Subdiaphragmaticvagotomy: Afteranesthesia, a longitudinal midline incision is made and the stomach is gentlypushed aside toward the posterior of the animal to expose the esophagus. Undera surgical microscope, the left and right vagus nerves (located anterior andposterior to the esophagus) are separated by blunt dissection, transectedimmediately below the diaphragm, and a length of nerve of approximately 1 cm isexcised. For the rats receiving the sham operations, the surgical procedure isidentical except that the vagus nerves are only subjected to gentle traction.
Behavioral testing will be conducted after a minimum of two weeks aftersurgery.1.3.2 Aim 2: Investigate whether vagal afferents expressing glp1r are activated by the intake of rewarding food andwhether chemogenetic manipulation of these neurons modulate food reward.
1.3.2.a Question 3: Are glp1r-expressingvagal afferents activated during the acquisition and expression of rewardbehaviour? Overview: Geneticallyencoded calcium indicators (GECI) provide an optical proxy foractivity-dependent calcium transients and make it possible to monitor theactivity of a select subset of neurons in vivo (33).
Here, we will express the most recent GECI variant GCaMP6s (34) in glp1r-expressing vagal afferent neurons. The firing patternactivityof these neurons will be measured in freely-behaving animals during theacquisition of CPP and during OR using photometry. Methods: Mice:Glp1r-ires-cre mice (Jax mice #029283) (35) will be injected in both nodoseganglia (where cell bodies of vagal afferents are located) with an adeno-assoviatedassociatedvirus (AAV) expressing GCaMP6 in a Cre-dependent manner(AAV-DIO-GCaMP6s-tdTomato; AddGene #51082). Photometry: An optical photometry fiber (Doric, Quebec, CA) will be implanted abovethe nucleus tractus solitarii (NTS), where vagal afferents terminate. Afterhabituation to the fiber photometry recording setup, fluorescence signals canbe continuously recorded from freelybehaving mice.
First, the activation of glp1r-expressingvagal afferents will be measured in response to exendin-4 and the exendin-4conjugate used in Aim 1 (1.1.3), as well as in response to isocaloric chow and palatblepalatablefood. Second, the activation of glp1r-expressing vagal afferents will be measured in real-timeduring the acquisition phase of CPP and during OR.Drugs and behavioral testing: same as 1.
2.b Question 4: Does chemogenetic inhibitionof glp1r-expressing vagal afferents modulate food reward?Overview: Here, we aim at expressing the inhibitory designer receptorhm4D, a modified version of the Gi-coupled human muscarinicreceptor-4 (36), in vagal afferent neurons expressing glp1r. Therefore, neuronal activity canbe specifically inhibited in hm4D-expressing neurons by the exogenous ligandclozapine-N-oxide (CNO), and their role in the food reward can be assessed. Methods: Mice: Glp1r-ires-cre mice (Jax mice #029283) will beinjected in the nodose ganglia with an adeno-associated virus(AAV-hSyn-DIO-hm4D-mCherry, AddGene #44362), expressing the inhibitory designerreceptor hm4D in a cre-dependentmanner. Hm4D is a modified version of the Gi-coupled muscarinic receptor 4 thatresponds to the highly selective exogenous ligand clozapine-n-oxide (CNO).Therefore, the activity of glp1r-expressingvagal afferents can be temporarily inhibited by injection of CNO. Under theseconditions, we will test the effect of CNO-induced inhibition on the acuteconsumption of chow and high-fat diet.
In addition, we will test the effect ofthis inhibition on the results of CPP and OR. Behavioral testing: same as 1.3.
3.3 Aim3: Elucidate the central connectome of vagal glp1r-expressing neurons1.3.3.
a Question 5: Are glp1r-expressingvagal afferents synaptically connected to central reward circuits Overview: Here, we aim at tracing the central circuitssynaptically connected to vagal afferent neurons expressing glp1r. Known central reward circuitswill be screened for expression of a cre-dependentanterograde transsynaptic viral tracer.Methods: Mice: Glp1r-ires-cremice (Jax mice #029283) will be injected in the nodose ganglia with a cre-dependent herpes simplex virus (HSV)anterograde transsynaptic tracer (HSV129?TK-TT (37)). Mice will be monitored daily forthe development of infection-induced symptoms (hunched back, increased anxiety,weight loss, …).
Histological analysis: Mice will be euthanized by cardiacperfusion, and brain tissue collected for histological analysis bycryo-sectioning. tdTomato expression will be systematicallyvisualized to characterize the brain connectome linked to glp1r-expressing vagal afferents. Specifically, tdTomato expressionwill be characterized in VTA, NAc, Amygdala, and vHPC 1.3.3.
b Question 6: researchquestion / hypothesis: is thereindication of a functional relationship (does the activation of one engage theactivation of the other?)Functionaltesting of the connection by measuring Dopamine and its metabolites by HPLC inthe connected areas in response to food with or without CNO (mice model of Aim1). We have recentlytested a promoter specific AAV that would allow for simultaneous recording ofactivity in, for example, the VTA DA neurons and the NTS vagal terminals (thisis obviously very ambitious but very cool). The downside here is that we areproposing a lot of photometry experiments, which are not yet setup in thegroup. Maybe to just show something I can borrow some of the recentoptogenetics data gathered by a postdoc in the group with this TH promoterspecific AAV, just to show that we have most elements setup already, and theonly missing factor is the recording (rather than stimulating) fiber.
I willalso send Nick Bentley an email, asking whether we can include his name in thisgrant as a collaborating lab, since he has as much photometry experience as onecan have. Significanceparagraph somewhere here would be useful. Why is it important? What clinicalimpact could these studies have? Outlining a novel way to target brain rewardcircuitry maybe? There could besomething to say for a drug to be effective on the reward system without havingto penetrate the BBB. Also there is the curious case of opposite than expectedeffects with the vagal kd. Somaybe one interesting angle here could be that the current treatmentsare less effective than they couldbe since they are stimulating two opposingglp1r populations (or something to that effects).Whatresearch avenues/therapeutic potential are we opening? If there is an effect onfood reward – could ourmanipulations also affect drug-related reward? Thesex paragraph belowwould fit right into significance – nearly nothing is knows about vagal impacton the brain in females, so we have a tremendous opportunity here. Nearlyall preclinical studies done on males.
1.4 +paragraph on sex specificityI would emphasize sex in a separate paragraphmaybe after the 3 aims, since pretty much nothing is known about the vagalglp-1r (or vagus-brain transmission) in females…We can use our recent MolecularPsychiatry paper to highlight how different the function and downstream signalscan be between the sexes.To be donein males + females mice evidence for estrogen effect on vagal afferentprojection in the brainstem, Ciriello and Caverson, Brain Research, 2016 1.5 +overall limitations and potential to pivotLimitations:here mention liberles, 24 fasting vs RonveauxPotentialto pivot. Can do aims 2 and 3 just for the control of FI, not necessarilyreward… Hereyou could also emphasize why all these experiments are feasible.
You do a greatjob illustrating this in the images, but this would be a god place to drivethis point home –you already have a uniqueexpertise targeting the nodose ganglia, and we have all the behavioralparadigms along with opto and chemogenetic methods in place etc. Thisis a good place to convince the reviewer that the combination of you and my labis the BEST chance for these studies to be done well.Here would also be a place to mentionthat photometry will be conducted with the help of the Bentley lab at UPenn,and Matt Hayes will provide the novel B12-Ex4 conjugate. I know it’s only 2 years, and may have to cutsome stuff in the end to not be overly ambitious but at this stage it may beuseful to include more rather than less, and then shape to leave the mostinteresting experiments. 2 Scheduleand milestones3 Reasonsfor the choice of the host institutionMy motivation for choosing Prof. Skibicka’s laboratory stemsfrom the many scientific interests I share with her. Her work focuses on theneuronal mechanisms controlling eating and energy expenditure, in particularhow gut-brain hormones control ingestive behavior.
I have myself conductedrelated studies during my PhD (glp1 / glp1kd / BAT / cart). In addition,throughout my readings and preliminary observations from my PhD, I havedeveloped a solid interest in understanding the mechanisms « Gut-brainhormones regulate a wide range of behaviors: impulsivity, novelty seeking,anxiety and depression-like behavior » underlying reward-learning and motivation, especiallygiven their importance for day-to-day human behavior in general, and forneuropsychiatric disorders in particular. I have included such questions intomy application (ResearchLine 2).
I am also greatly attracted to Prof. Kellendonk’s scientificapproach, which aims at deciphering fundamental neurobiological mechanisms inmouse models, while also generating data directly relevant to human medicine.Furthermore, he uses cutting-edge tools such as optogenetic and chemogeneticcontrol of neural activity combined with in vitro and in vivo electrophysiological methods andstate-of-the art behavioral paradigms. This was well illustrated by his lateststudy 13, in which he employed such techniques to identify a novel andfunctional axonal collateral population within the basal ganglia which hasrobust implications for schizophrenia physiopathology.
The project I havedeveloped builds on these results. I also believe the working conditions are ideal: thenecessary mouse lines/viruses and methods are available. All team members haveprovided extremely positive feedback regarding Prof. Kellendonk’s mentorship.The laboratory also belongs to the prestigious Columbia University MedicalCenter, which offers many opportunities for interactions with world-class neurobiologistsand clinicians. Finally, although I am very experienced with some of therelevant techniques, I will have the exciting opportunity to train oncutting-edge technology once in the Kellendonk laboratory, such asoptogenetics/chemogenetics. I intend to complement this with anelectrophysiology course at Cold Spring Harbor.
Such research tools and theoverall experience at Columbia will represent essential building blocks for mycareer in the long-term, as detailed below.