Abstract
Dynamic reconfiguration of circuit function subserves the flexibility of innate behaviors tuned to physiological states. Internal energy stores adaptively regulate feeding-associated behaviors and integrate opposing hunger and satiety signals at the level of neural circuits. Across vertebrate lineages, the neuropeptides cocaine- and amphetamine-regulated transcript (CART) and neuropeptide Y (NPY) have potent anorexic and orexic functions, respectively, and show energy-state-dependent expression in interoceptive neurons. However, how the antagonistic activities of these peptides modulate circuit plasticity remains unclear. Using behavioral, neuroanatomical, and activity analysis in adult zebrafish of both sexes, along with pharmacological interventions, we show that CART and NPY activities converge on a population of neurons in the dorsomedial telencephalon (Dm). Although CART facilitates glutamatergic neurotransmission at the Dm, NPY dampens the response to glutamate. In energy-rich states, CART enhances NMDA receptor (NMDAR) function by protein kinase A/protein kinase C (PKA/PKC)-mediated phosphorylation of the NR1 subunit of the NMDAR complex. Conversely, starvation triggers NPY-mediated reduction in phosphorylated NR1 via calcineurin activation and inhibition of cAMP production leading to reduced responsiveness to glutamate. Our data identify convergent integration of CART and NPY inputs by the Dm neurons to generate nutritional state-dependent circuit plasticity that is correlated with the behavioral switch induced by the opposing actions of satiety and hunger signals.
SIGNIFICANCE STATEMENT Internal energy needs reconfigure neuronal circuits to adaptively regulate feeding behavior. Energy-state-dependent neuropeptide release can signal energy status to feeding-associated circuits and modulate circuit function. CART and NPY are major anorexic and orexic factors, respectively, but the intracellular signaling pathways used by these peptides to alter circuit function remain uncharacterized. We show that CART and NPY-expressing neurons from energy-state interoceptive areas project to a novel telencephalic region, Dm, in adult zebrafish. CART increases the excitability of Dm neurons, whereas NPY opposes CART activity. Antagonistic signaling by CART and NPY converge onto NMDA-receptor function to modulate glutamatergic neurotransmission. Thus, opposing activities of anorexic CART and orexic NPY reconfigure circuit function to generate flexibility in feeding behavior.
Introduction
Adaptive regulation of feeding in response to internal energy needs is critical for survival. Internal sensory systems, located in the periphery and in the CNS, monitor energy homeostasis and signal satiety or hunger to the feeding-associated neural circuits. Hunger stimulates feeding circuits to generate motivational states that prioritize feeding and related activities over other behaviors. Conversely, satiety signals suppress the feeding drive, facilitating the pursuit of nonfeeding-related behaviors. The flexibility of the feeding behavior—exploiting or abandoning food resources—requires dynamic reconfiguring of the activity states of the underlying circuits, and the deregulation of these processes forms the basis of eating disorders.
Neuropeptides, acting as hormones or local diffusive modulators convey nutritional state information to the nervous system (van den Pol, 2012). Neuropeptide-based neuromodulation of circuit function, via changes in neuron intrinsic properties or synaptic efficacy, may allow the same circuit to produce multiple outputs and behavioral outcomes (Nadim and Bucher, 2014). Neuropeptides typically engage G-protein-coupled receptors, which, via intracellular signaling, may change circuit properties and extend acute signals of internal needs into long-lasting state changes mediated by biochemical hysteresis. For example, in the rodent arcuate nucleus (Arc), orexigenic neuropeptide ghrelin initiates an AMP-activated protein kinase (AMPK)-mediated positive-feedback to potentiate glutamate release for extended periods (Yang et al., 2011). Consequently, several studies have implicated neuropeptides in mediating transitions between distinct behavioral states (Root et al., 2011; Elbaz et al., 2012; Flavell et al., 2013; Betley and Sternson, 2015).
Cocaine- and amphetamine-regulated transcript (CART) and neuropeptide Y (NPY) are abundantly expressed neuropeptides conserved across vertebrate lineages. Both neuropeptides have been implicated in a diverse range of physiological functions, including energy homoeostasis and food intake (Lau and Herzog, 2014; Subhedar et al., 2014; Singh et al., 2021).
CART is a potent anorexigenic agent and considered to be an endogenous satiety factor (Kristensen et al., 1998), especially in the framework of the Arc in the hypothalamus (Singh et al., 2021). Expression of CART increases in response to feeding in the Arc (Kristensen et al., 1998), and recent chemogenetic studies underscore the requirement of CART peptide in Arc neurons to inhibit food intake (Farzi et al., 2018). In contrast, hypothalamic NPY is orexigenic and promotes food intake and energy conservation (Zhang et al., 2019). Starvation increases NPY expression in the Arc (Sainsbury and Zhang, 2010), and NPY signaling from agouti-related peptide (AgRP)–expressing Arc neurons is critical for the sustained maintenance of the feeding drive (Chen et al., 2016).
Within the Arc, the coordinated coregulation of the AgRP/NPY- and the POMC/CART-containing neurons in response to signals indicating metabolic states is well documented (Dietrich and Horvath, 2013; Andermann and Lowell, 2017). The physiologically opposing activities of these two populations and their effect on second-order neurons that drive satiety is a major regulator of the feeding behavior (Garfield et al., 2015; Andermann and Lowell, 2017). However, the intracellular signaling engaged by these neuropeptides in the second-order neurons to reconfigure the circuit properties and drive homoeostatic plasticity remains inadequately defined.
In zebrafish, NPY and CART are expressed in several brain areas, including the periventricular hypothalamus (homologous to mammalian Arc; Forlano and Cone, 2007) and the entopenduncular nucleus (EN; homologous to the internal globus pallidus, an output nucleus of the basal ganglia, in mammals; Mathieu et al., 2002; Mukherjee et al., 2012; Yokobori et al., 2012; Akash et al., 2014; Amo et al., 2014).
Starvation decreases CART mRNA expression in the periventricular hypothalamus and the EN (Nishio et al., 2012; Akash et al., 2014) while increasing NPY expression (Yokobori et al., 2012). Treatment with glucose also increases CART protein expression in the EN (Mukherjee et al., 2012). The zebrafish periventricular hypothalamus and the EN appear to be nutritional state-sensitive interoceptive areas regulating energy homeostasis, in part via CART/NPY. Accordingly, treatment with NPY increases feeding in zebrafish, and its action is mediated by the NPY Y1 receptor (Yokobori et al., 2012).
Despite the prominent involvement of CART and NPY in vertebrate energy homeostasis, it remains unclear how their activities reconfigure the downstream, second-order circuits to meet behavioral demands. In this study, using zebrafish, we show that CART and NPY shape the feeding drive to match the prevailing energy states. The opposing energy state information represented by the two neuropeptides are integrated by a group of forebrain neurons by transforming the antagonistic signaling into dynamic changes in glutamatergic neurotransmission.
Materials and Methods
Animals
Adult, wild-type zebrafish (Danio rerio) of both sexes were housed in stand-alone housing systems and maintained on 14/10 h light/dark cycle. Fish were fed Ziegler feed and live artemia three times a day. For calcium activity imaging experiments, adult Tg(NeuroD:GcAMP6f) in the nacre background were used. All behavioral assays and activity imaging experiments were performed between 10:00 A.M. and 5:00 P.M. No randomization methods were used to allocate animals to different experimental groups. The Institutional Animal Ethics Committee of the Indian Institute of Science Education and Research, Pune approved all the procedures used in this study. This study was not preregistered.
Chemicals and reagents
The following reagents were administered to the zebrafish brain via intracerebroventricular delivery, and the concentrations indicated reflect the amount delivered by the intracerebroventricular route: Anti-CART antibody (1:500; mouse monoclonal antibody raised against rat CART peptide (54–102 amino acids; gift from Drs. Lars Thim and Jes Clausen, Novo Nordisk, Denmark); d-(+)-glucose monohydrate (40 nmol in 1× PBS, glucose; catalog #49161, Sigma-Aldrich), BIBP-3226 (BIBP; 100 pmol in 0.1% DMSO; catalog #B174, Sigma-Aldrich), MK801 (0.015 nmol in 1× PBS; catalog #77086, Sigma-Aldrich), AP5 (0.1 nmol in 1× PBS; catalog #0106, Tocris Bioscience), KT5720 (1.4 pmol in 1× PBS, PKAi; catalog #K3761, Sigma-Aldrich), GF109206X (1.5 pmol in 1× PBS, PKCi; catalog #0741, Tocris Bioscience), FK506 (0.13 nmol in 1× PBS; catalog #F4679, Sigma-Aldrich), and Forskolin (1.2 nmol in 0.1% ethanol; catalog #F6886, Sigma-Aldrich). The following reagents were used for procedures not requiring intracerebroventricular administration: 2-Phenoxyethanol (1:2000 in system water; catalog #P1126, Sigma-Aldrich), D-(+)-glucose monohydrate (catalog #49161, Sigma-Aldrich), sucrose (catalog #S9703, Sigma-Aldrich), paraformaldehyde (catalog #158127, Sigma-Aldrich), poly-l-lysine (catalog #P8920, Sigma-Aldrich), normal goat serum (catalog #9023, Sigma-Aldrich), bovine serum albumin (BSA; catalog #83803, Sisco Research Laboratories), and 1,1′, dioctadecyl-3,3,3′, 3′-tetramethylindo-carbocyanineperchlorate (DiI; catalog #D3911, Invitrogen).
The peptides used in this study are the rat CART peptide (55–102 amino acids; 2 pmol in 1× PBS; catalog #003-62, Phoenix Pharmaceuticals) and full-length human NPY peptide (10 pmol in 1× PBS; catalog #N5017, Sigma-Aldrich). The amino acid sequences of these peptides aligned with their zebrafish orthologues are shown in Extended Data Figures 1-1E (CART) and 10–1 (NPY).
Intracerebroventricular injections
Intracerebroventricular injections were used to deliver bioactive peptides, chemicals, and pharmacological agents directly to the fish brain while bypassing the blood–brain barrier. Fish were anaesthetized in 2-Phenoxyethanol (1:2000). The anesthetized fish were placed on a cotton bed containing anesthesia solution to submerge the head. Intracerebroventricular delivery protocol was modified after Yokogawa et al., 2007. Briefly, pharmacological agent or vehicle was delivered directly into the ventricular space using an 31 gauge needle (Becton Dickinson Insulin Syringe U-40 31 gauge) attached via a catheter to a 10 μl Hamilton microsyringe. After injection, the fish were returned to their tanks and allowed to recover before proceeding for either behavioral recordings or immunohistochemistry. Unless otherwise indicated, 30 min following injection, the animals were subjected to behavioral task or killed and the tissues processed for immunofluorescence studies.
Immunofluorescence
Fish were subjected to intracerebroventricular injection with drug or vehicle and allowed to recover for 30 min. After recovery, fish were anesthetized and craniotomized to expose the dorsal surface of the brain. Samples were fixed in 4% PFA overnight at 4°C. After fixation for 12–14 h [10 h for phospho NR1 (pNR1) staining], brain was dissected and cryoprotected with 25% sucrose before sectioning. Serial 15–20-µm-thick sections of the entire telencephalon were mounted onto lysine-coated slides and stored at −40°C until further processing. Sections were allowed to dry for 2 h and washed with 0.5% Triton X-100 in 1× PBS (PBST) three times for 10 min each. PBS containing the following (in mm) was used: 137 NaCl, 2.7 KCL, 10 Na2HPO4, and 1.8 KH2PO4, pH 7.2. Sections were then blocked using 5% BSA in PBST for 1 h before the addition of primary antibodies and incubated overnight at 4°C. The next day, the washes were repeated and sections blocked in 5% BSA/5% heat-inactivated goat serum for 1 h. The following primary antibodies were used in this study: rabbit polyclonal anti-phosphorylated ERK (p-ERK: 1:700; catalog #9101, Cell Signaling Technology), mouse monoclonal anti-p-ERK (1:500; catalog #ab50011, Abcam), rabbit polyclonal pNR1 (Ser-897, 1:70; catalog #ABN99, Millipore), mouse monoclonal anti-synaptophysin antibody (1:200; catalog #S5768, Sigma-Aldrich), rabbit polyclonal anti-vglut 2 (1:500; catalog #181773, Abcam), and rabbit polyclonal anti-CART antibody (1:2000; catalog #H-003-62, Phoenix Pharmaceuticals).
Secondary antibodies were added and incubated at room temperature (RT) for 2.5 h in dark. The following secondary antibodies were used: anti-rabbit Alexa Flour 488/568 (1:500; catalog #A-11034/A-11036, Invitrogen) and anti-mouse Alexa Flour 488/568 (1:500; catalog #A-11029/A-11031, Invitrogen). Sections were washed and mounted in media (0.5% n-propyl gallate, 70% glycerol, 1 m Tris, pH 8.0) containing 4′,6′-diamidino-2-phenylindole (DAPI; catalog #D1306, Invitrogen). Sections were observed under an Axioimager Z1 (Zeiss) epifluorescence microscope, and representative images were acquired using a confocal microscope (SP8, Leica, or LSM 780, Zeiss). To ensure reliable comparisons across different groups and maintain stringency in tissue preparation and staining conditions, all the brain sections were processed concurrently under identical conditions.
Feeding behavior
Fish were isolated from home tanks and housed singly in the experimental tank before behavioral analysis. Figure 1A outlines the protocol followed for all feeding behavior experiments. Briefly, for first 2 d of habituation, the experimental tanks were moved to the recording chamber, and the fish were fed floating food pellets with low solubility (15 ± 5 pellets, Taiyo) for 1 h before returning to the housing rack. For habituation to injection and handling stress, the fish were anesthetized and delivered a mock intracerebroventricular injection (saline, 0.9% NaCl) on days 4 and 5. Fish were allowed to recover in a recording chamber for 1 h before returning the tanks to the housing racks. Depending on the experiment, the fish were either starved or given pellet food for 1 h on these days. Fish were observed on all days of the protocol for any anxiety-like/abnormal behavior. Fish exhibiting anxiety-like behaviors were not included for further analysis. On the day of the experiment, fish were anesthetized and intracerebroventricularly injected with appropriate reagents or vehicle (codelivered for experiments with more than one pharmacological agent). Following a recovery period of 15 min, floating pellets of food (∼15 ± 5) were added to the tank, and the behavior of the fish was recorded individually using a video recorder (Sony Handycam DCR SR-47 and DCR SR-20) for 1 h. All recordings were conducted under conditions of uniform lighting and temperature and were performed between 1100 and 1600 h. In all experiments, animals deprived of food for 2.5 d are referred to as “starved,” and those that received food per the regular feeding schedule and used for experiments within 1 h of feeding are referred to as “fed.” The recorded videos were not blinded, and the number of biting attempts made by the fish during the entire 1 h period was counted manually. The data were plotted as the cumulative number of biting attempts in 4 bins of 15 min each.
The evaluation of the difference in dry weight of the food pellet before and after the 1 h feeding assay indicated, as expected, substantially more ingestion by starved animals compared with fed fish (Extended Data Fig. 1-1A,B). During the course of this behavioral protocol, although starvation increased the number of biting attempts and ingestion, no change in body weight was observed (Extended Data Fig. 1-1C). On the other hand, feeding experiments conducted with nonfood mock pellets revealed no difference in biting attempts between fed and starved fish, underscoring the specificity of this assay for food (Extended Data Fig. 1-1D).
Calcium imaging
Starved adult Tg(NeuroD:GCaMP6f) fish, a transgenic line developed by Rupprecht et al. (2016) with robust GCaMP expression in the dorsal telencephalon (Huang et al., 2020), were injected with either vehicle or drugs. After 15 min, the fish were anesthetized in ice-cold HEPES-based Ringer's solution containing the following (in mm): 134 NaCl, 1.2 MgCl2, 2.1 CaCl2, 2.9 KCl, 10 HEPES, and 10 sucrose (for glucose deprivation), or 10 glucose (to mimic satiety), pH 7.2, bubbled with 100% O2, and were craniotomized to dissect the telencephalon. The optic nerves were cut off, and the telencephalon was dissected out by cutting just rostral to habenula. The telencephalon was carefully lifted from the base while removing the ventral connections and immediately transferred to Ringer's solution supplemented with 100% O2.
For ex vivo whole-brain preparations, starved fish were anesthetized in ice-cold HEPES-based Ringer's solution. First, the optic nerves and connections on the ventral side were severed, and the whole brain was dissected out by separating it from spinal cord.
The intact telencephalon or the whole brain was mounted in 4% low-melt agarose (in 50% Ringer's) on on a RC-26 GLP chamber from Warner Instruments (catalog #640236), with the agarose layer over the telencephalon removed to expose the Dm directly to the buffer/glutamate. The imaging chamber was mounted onto an upright confocal microscope (Leica TCS SP8 MP) and continuously perfused with HEPES-based Ringer's solution via a VC 8 perfusion system (catalog #640186, Warner Instruments) directly triggered by the recording software (Leica). Images were acquired at 9.8 Hz at a resolution of 512 × 300 pixels using a 488 nm laser in the resonant, bidirectional scanning mode and a 25× water-immersion objective (NA 0.95). Baseline imaging was performed for 1 min before the start of the glutamate stimulation. Glutamate (75 μm in Ringer's buffer) was perfused from the second to fifth minute and subsequently washed out.
Data analysis
For immunofluorescence, the tissue slices were observed under Axioimager Z1 (Zeiss, 20× objective 0.8 NA) with the imaging conditions kept constant across all treatment groups. The number of p-ERK-positive cells in Dm were scored manually. To avoid overestimation of cell count because of sectioning, the cell numbers were corrected using Abercrombie's (1946) method.
The average pNR1 intensity per cell in Dm was calculated using ImageJ software. The exposure time and other imaging conditions were kept constant for every sample processed. Each neuron in Dm was selected using manual thresholding on the DAPI filter, and the mean intensity of pNR1 per neuron was calculated using ImageJ built-in algorithms. Images were acquired using a 40× objective (NA 1.3) on a SP8 confocal microscope (Leica).
For calcium imaging, measurements of population-level calcium activity in Dm were performed using ImageJ, and the average GCaMP6f fluorescence in the Dm at each time point was calculated. For this, we manually marked a region of interest (ROI) encompassing the Dm using the wand tool in ImageJ on maximum intensity projection of all the frames. This ROI was then applied to the time lapse series, and the baseline fluorescence (F0) was estimated by taking the average of fluorescence values of 620 frames before the start of glutamate treatment. The relative change in fluorescence (ΔF/F0), was calculated using the formula (F − F0)/F0, where F stands for the fluorescence value at a given time. The maximum response amplitude, that is, the maximum ΔF/F0, was determined from the temporal sequence using Microsoft Excel. The total magnitude of the response was calculated by computing the area under the curve (AUC) from the start of glutamate treatment to the end of the washout period in GraphPad Prism version 8 software.
For the feeding assay, videos were analyzed manually at 1.5× speed, and the number of biting attempts were scored. The data were distributed in bins of 15 min, and cumulative biting attempts were plotted using GraphPad Prism version 8 software.
Statistical analysis
Data were tested for normality using the Shapiro–Wilk test in GraphPad Prism 8. Behavior and immunohistochemical data were subjected to the t test with Welch's correction or the Mann–Whitney test (for single comparison) and one-way ANOVA with Tukey's test or two-way ANOVA with Bonferroni's post hoc analysis for multiple comparisons. All values are expressed as mean ± SEM of the group (see below, Results), and differences were considered significant at p < 0.05. Graphs were plotted using GraphPad Prism version 8 software. Data for feeding drive are represented as columns (with error bars indicating mean ± SEM) of cumulative biting attempts in 15 min bins over a 1 h period. Change in GCaMP6f fluorescence over time is represented as XY plots with the dark line representing mean and the dotted lines indicating SEM. All data shown as violin plots have the thicker dashed line representing median and dotted lines representing first and third quartile.
Data availability
All data generated or analyzed during this study are included in this article. Raw data are available from the corresponding author on reasonable request.
Results
CART neuropeptide mediates satiety during energy-rich metabolic states
To quantitatively evaluate the feeding drive of adult zebrafish, we developed a behavioral assay that scores the number of biting attempts made by the animal on presentation of solid, floating food pellets (see above, Materials and Methods; Fig. 1A). We found that hungry fish made many more biting attempts over a period of 1 h (794 ± 57.46; Fig. 1B; p = 0.0024) compared with fed animals (338.8 ± 70.4; Fig. 1B). Analysis of the number of biting attempts across 15 min bins revealed a cumulated difference over the 1 h period. To rule out contributions from altered locomotion or stress-induced anxiety-like behaviors, we evaluated animals under fed and starved conditions for locomotion kinematics and indicators of anxiety-associated behaviors but found no change in these parameters (Fig. 2).
CART neuropeptide regulates satiety in adult zebrafish. A, Schematic of experimental design and timeline for feeding behavioral experiment. For protocol, pharmacological and bioactive agents, and doses see Materials and Methods, Feeding behavior. B, The feeding drive as indicated by the cumulative number of biting attempts made by fed and starved animals in 15 min bins (N = 6 animals per group, p = 0.002). Validation of the behavioral assay is described above in Materials and Methods and in Extended Data Figure 1-1A–D. C, The cumulative number of biting attempts made by starved fish intracerebroventricularly injected with either vehicle or CART peptide (N = 6 animals per group; ***p < 0.0001). CLUSTAL Omega alignment of the mammalian and zebrafish CART peptides reveal a high degree of amino acid identity (Extended Data Figure 1-1E). D, The cumulative number of biting attempts made by starved fish intracerebroventricularly injected with either vehicle, glucose, CART antibodies, or coinjected with immunoneutralizing CART antibodies (CART Ab) along with glucose (N = 5 for vehicle control and CART Ab–treated animals and N = 6 for glucose and CART Ab plus glucose–treated animals; p < 0.0001 for vehicle plus vehicle compared with glucose plus vehicle and glucose plus vehicle compared with glucose plus CART Ab). Data are represented as cumulative biting attempts in 15 min bins over a 1 h period and compared using two-way ANOVA, with Bonferroni's post hoc analysis for significance in comparison with the glucose-treated starved fish. Error bars indicate ± SEM; ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001).
Figure 1-1
Characterization of the feeding behavior assay and alignment of mammalian and zebrafish CART peptides. A, The cumulative number of biting attempts made by starved and fed fish (N = 6 animals per group; p < 0.0001). B, Change in dry weight of pellet food during 1 h of feeding in starved and fed fish (N = 6 animals per group; p < 0.001). C, Change in body weight of fish that are either starved for 3 d or fed regularly (N = 6 animals per group). ns, Not significant. The data are compared using the unpaired t test with Welch's correction. D, The cumulative number of biting attempts made by starved and fed fish when exposed to mock food pellets (left) and food pellets (right). The data were compared using two-way ANOVA with Bonferroni's post hoc analysis (N = 6 animals per group, ns; *p < 0.01, **p < 0.001). E, CLUSTAL Omega alignment of peptide sequences of rat CART 55–102 (AAA87897.1), zebrafish CART 1 (ADB12484.1), CART 2 (ADB12485.1), CART 3 (ADB12486.1), and CART 4 (ADB12487.1). Dark purple background represents identical residues, light purple background represents similar residues, and white background represents different residues. The percentage identity with rat CART (55–102) is indicated on the right. Download Figure 1-1, TIF file.
Analysis of anxiety-like behaviors and locomotion kinematics. A, Schematic of experimental design and timeline used for the evaluation of anxiety-like behaviors and locomotion. B–E, Average velocity (B), total distance traveled (C), duration in the top zone (D), ratio of the time spent at the top zone and the bottom zone (top:bottom) (E). F, Transition frequency between the top and bottom zones were assessed under different physiological energy states or on intracerebroventricular administration of CART and NPY peptides. The data were compared using unpaired t test with Welch's correction (N = 6 animals in each group; ns, not significant; all p values > 0.5).
To directly assess the role of CART neuropeptide in anorexia, we delivered CART peptide (55–102 aa) via the intracerebroventricular route to the brain of starved fish. Strikingly, exogenous administration of CART directly to the brain was sufficient to induce acute anorexia in starved fish (36.21 ± 14.89; Fig. 1C; p < 0.0001) compared with starved animals injected with vehicle (829.4 ± 58.03; Fig. 1C). However, CART treatment did not induce anxiety-like behaviors or modify locomotion (Fig. 2).
To test whether central application of glucose resulted in anorexia, glucose was directly introduced via the intracerebroventricular route to the brains of starved fish to mimic energy surfeit conditions and preclude peripheral inputs. In contrast to starved animals (852 ± 133.3), glucose-injected fish displayed a dramatic reduction in biting attempts (221.2 ± 33.67; Fig. 1D; p < 0.0001). As CART promotes anorexia (Lau and Herzog, 2014), and CART peptide expression is sensitive to metabolic states in the zebrafish brains (Mukherjee et al., 2012; Akash et al., 2014), we used an immunoneutralizing antibody against the CART peptide to test whether endogenous CART signaling mediates the glucose-induced reduction in feeding. Indeed, glucose-induced anorexia in starved animals was alleviated in the presence of CART immunoneutralizing antibodies coinfused with glucose by the intracerebroventricular route (629.9 ± 79.31; Fig. 1D; p < 0.0001). CART antibodies on their own did not influence the feeding behavior (658.4 ± 67.13; Fig. 1D).
Collectively, these results establish the role of CART neuropeptide signaling in the zebrafish brain in mediating the reduction in food intake behavior during energy-rich metabolic states.
CART signaling alters the activity of telencephalic Dm neurons
The EN in the ventral telencephalon as well as periventricular hypothalamic nuclei, like the nucleus lateralis tuberis (NLT), show energy state–dependent changes in CART mRNA expression (Akash et al., 2014). Administration of glucose also increased the number of CART-peptide-expressing cells in the EN and the NLT (p < 0.0001 and p = 0.0237, respectively; Fig. 3; Mukherjee et al., 2012).
CART expression is upregulated in the EN and the NLT on intracerebroventricular glucose administration. A–F, Representative micrographs of the EN (A, B) and the NLT (D, E) from transverse sections of the zebrafish brain showing CART-immunoreactive cells in starved fish receiving intracerebroventricular injection of either vehicle or glucose. The quantification of the number of CART-positive cells in EN (p < 0.0001, C) and NLT (p = 0.0237, F) in starved fish intracerebroventricularly injected with either vehicle or glucose. The data were compared using unpaired t test with Welch's correction (n = 8/10 EN/NLT from 4/5 animals per group; *p < 0.05; ***p < 0.001).
CART immunofluorescence analysis of transverse sections of the zebrafish telencephalon was used to evaluate the projections from the EN. CART-positive EN somata were found to project to the dorsomedial telencephalon (Dm; Fig. 4A). This is consistent with earlier reports indicating that EN neurons project to the Dm (Turner et al., 2016). We used the lipophilic dye DiI to confirm whether the Dm receives projections from EN and other energy status-responsive interoceptive areas. DiI-positive fibers from the Dm typically extended caudally and ventrally and merged with the lateral forebrain bundle. Labeled cells were found in the ipsilateral EN (Fig. 4B,C). Fibers were also traced caudally into the hypothalamus, and labeled somata were located in the preglomerular nucleus and the NLT (Fig. 4B,D). This connectivity is consistent with previous reports from teleost fish, including zebrafish (Yáñez et al., 2017). Dual immunostaining for CART peptide and the presynaptic marker synaptophysin revealed significant colocalization, indicating the presence of CART in presynaptic terminals at the Dm (Fig. 4F). However, no CART-expressing somata were detected in the Dm. Furthermore, the CART-immunoreactive somata in the EN also expressed vesicular glutamate transporter 1/2, a marker for glutamatergic neurons, indicating a source of excitatory drive to the Dm in addition to the peptidergic signaling via CART (Fig. 4G–I).
Dm receives inputs from the EN and the periventricular hypothalamus. A, Transverse section of telencephalon showing CART-immunoreactive neurons in the EN and CART-containing fiber tracts projecting to Dm region (dotted line). B–D, Analysis of Dm connectivity by DiI labeling (B); site of DiI application in the Dm (dotted circle). Representative photomicrographs (C, B) of the optical sections showing DiI labeling in EN and NLT neurons. E, Schematic showing the connectivity of EN and the periventricular hypothalamic projections to the Dm (shaded region). OB, Olfactory bulb; ac, anterior commissure; Ha, habenula; Hy, hypothalamus; TeO, optic tectum. F, Presynaptic localization of CART in the Dm. Representative micrographs of Dm showing colocalization of synaptophysin (SYP; green) and CART (red) along with nuclei labeled with DAPI (blue). White rectangle marks the region magnified and shown in the top right corner (G–I). Representative micrographs of EN showing colocalization of CART (green) and vglut2 (magenta). Scale bars: A, 25 µm; B, 1000 µm; C–F, 20 µm; G–I, 5 μm.
Collectively, these results demonstrate that CART peptide expression in EN and the periventricular hypothalamus is controlled by metabolic states, and these areas send projections to the Dm (Fig. 4E). This connectivity suggested that Dm neurons could be responsive to energy state–dependent CART activity. To assess the involvement of Dm, we used p-ERK, a marker of recent neuronal activity (Randlett et al., 2015).
Starved fish injected with CART peptide via the intracerebroventricular route had an elevated number of p-ERK-positive neurons in the Dm (533.9 ± 89.04; p = 0.0067) in comparison to starved, vehicle-injected animals (159.5 ± 34.52; Fig. 5A–C). These data suggest a CART-mediated increase in neuronal activity in the Dm. Further, the Dm receives projections from the glucosensitive EN and the periventricular hypothalamus (Figs. 3, 4A–D). Thus, we tested whether feeding was sufficient to induce neuronal activation of the Dm neurons by comparing starved and recently fed fish (where endogenous levels of CART in the EN and periventricular hypothalamus are high; Fig. 3). Remarkably, Dm also shows increased neuronal activity (423.2 ± 26.61; p < 0.0001) in fed fish compared with starved controls (180.0 ± 12.42; Fig. 5D–F). The subpopulation of Dm neurons responding to CART/fed states was limited to an ∼250 µm region rostral to the level of the anterior commissure (Fig. 5G–I).
CART neuropeptide and satiety increases the activity of the Dm neurons. A, B, Representative micrographs of Dm region from the transverse sections of the telencephalon showing p-ERK-immunoreactive (p-ERK-ir) cells in starved fish intracerebroventricularly injected with either vehicle or CART peptide. C, The number of p-ERK-ir cells in Dm in starved fish intracerebroventricularly injected with either vehicle or CART. The data are compared using the unpaired t test with Welch's correction (N = 6 animals per group; **p < 0.01; p = 0.0067). D, E, Representative micrographs of Dm region from the transverse sections of the telencephalon showing p-ERK-ir cells under starved and fed conditions. F, The number of p-ERK-ir cells in Dm under starved and fed states. The data are compared using the unpaired t test with Welch's correction (N = 6 animals per group; **p < 0.001, ***p < 0.0001). Scale bar, 50 µm. G, The top view of zebrafish telencephalon with shaded region (dark red) indicating the entire dorsomedial telencephalon. The area bounded by dotted lines within dorsomedial telencephalon (light red) shows differential activity on CART treatment and under fed/starved conditions. We refer to this subregion as Dm for the purposes of this study. H, Side view of zebrafish brain with dark red area marking entire dorsomedial telencephalon. The region of our interest, within Dm (light red), is marked by dotted lines (a, The rostral limit of the region that shows differential activity; b, The plane of section shown in I; c, The caudal limit at the level of the anterior commissure; the distance from a to c is ∼250 µm). I, A transverse section of telencephalon with the Dm region marked in light red. ac, Anterior commissure; Dp, posterior zone of dorsal telencephalon; Hy, hypothalamus; Ha, habenula; IOP, integrative olfactory pallium; MP, medial pallium; OB, olfactory bulb; PMPa, posteromedial pallial nucleus; Sy, sulcus yipsiloniformis; TeO, optic tectum; Vd, dorsal nucleus of ventral telencephalic area; Vv, ventral nucleus of ventral telencephalic area.
These findings suggest that enhancement of CART function at the Dm, possibly by the energy state–responsive neurons of the EN and the periventricular hypothalamus, results in increased neuronal activity. However, the possibility of additional inputs from other CART-ergic sources cannot be ruled out.
CART modulates NMDAR signaling to regulate the feeding behavior and the activity of Dm neurons
The dorsal telencephalon of teleosts, including the Dm, is rich in glutamatergic neurons (Aoki et al., 2013) and shows significant binding of labeled kainate and l-glutamate (Tong et al., 1992) implicating glutamatergic neurotransmission. Studies in rodents have previously implicated NMDAR-mediated CART signaling in the context of innate fear processing (Rale et al., 2017) and nociceptive transmission (Chiu et al., 2010) in rat. We investigated NMDAR signaling in CART-mediated regulation of feeding behavior using NMDAR antagonists. Intracerebroventricular delivery of AP5 (competitive NMDAR antagonist) alone in starved fish showed no difference in biting attempts (845.3 ± 89.82; Fig. 6A) compared with vehicle controls (812.8 ± 84.38). However, CART-induced anorexia (47.33 ± 11.4; p = 0.0014) was reduced when CART peptide and AP5 were coadministered (488.7 ± 42.03; Fig. 6A; p = 0.0004). Application of the noncompetitive NMDAR antagonist MK801 produced similar results (Fig. 6C; p < 0.0001 for vehicle plus vehicle compared with vehicle plus CART and for vehicle plus CART compared with MK801 plus CART).
NMDAR signaling is necessary for CART-induced anorexia and increased activation of Dm neurons. A, The cumulative number of biting attempts made by starved zebrafish intracerebroventricularly injected with either vehicle, CART peptide, AP5, or coinjected with AP5 and CART peptide. (N = 6 animals per group; p = 0.0014 for vehicle plus vehicle compared with vehicle plus CART and p = 0.0004 for vehicle plus CART compared with AP5 plus CART). Data are represented as cumulative biting attempts in 15 min bins over a 1 h period and compared using two-way ANOVA with Bonferroni's post hoc analysis. Error bars indicate ± SEM; **p < 0.01, ***p < 0.001). B, The number of p-ERK-ir cells in starved fish intracerebroventricularly injected with either vehicle, CART, AP5, or coinjected with AP5 along with CART. The data are compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the CART-treated starved fish (N = 6 animals per group; **p < 0.001; ***p < 0.0001 for vehicle plus vehicle compared with vehicle plus CART and for vehicle plus CART compared with AP5 plus CART). C, The cumulative number of biting attempts made by starved fish intracerebroventricularly injected with either vehicle, CART peptide, MK801, or coinjected with MK801 and CART peptide. Data are represented as cumulative biting attempts in 15 min bins over a 1 h period and compared using two-way ANOVA with Bonferroni's post hoc analysis analysis for significance in comparison with the CART-treated starved fish. Error bars indicate ± SEM; N = 5 animals for MK801 plus CART and N = 6 animals for the other groups; **p < 0.01, ***p < 0.001; p < 0.0001 for vehicle plus vehicle compared with vehicle plus CART and vehicle plus CART compared with MK801 plus CART). D, The number of p-ERK-immunoreactive cells in starved fish intracerebroventricularly injected with either vehicle, CART peptide, MK801, or coinjected with MK801 and CART peptide. The data were compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the CART-treated starved fish (N = 6 animals per group; ***p < 0.0001 for vehicle plus vehicle compared with vehicle plus CART and vehicle plus CART compared with MK801 plus CART).
AP5 was coinjected with CART peptide in starved animals to evaluate the role of NMDAR signaling in modulating the activity of the Dm neurons. Neuronal activity in the Dm, indicated by the number of p-ERK-positive neurons, was robustly attenuated in CART-peptide-treated and AP5-treated fish (165.7 ± 11.87) compared with animals injected with only CART peptide (312 ± 18.2; Fig. 6B). AP5 alone (125.2 ± 6.96) did not influence the number of p-ERK-positive Dm neurons compared with those in vehicle controls (126.7 ± 13.35; Fig. 6B). The application of MK801 produced similar results (Fig. 6D). Together, these data indicate that NMDAR activity is required for processing CART-induced satiety signals in the Dm neurons.
PKC and PKA phosphorylate NR1 to mediate the modulation of NMDAR signaling by CART
PKC and PKA–mediated phosphorylation of the NR1 subunit of the mammalian NMDAR is known to potentiate NMDAR signaling by increasing the channel opening probability and enhancing calcium permeability (Lan et al., 2001; Skeberdis et al., 2006). PKA and PKC have also been implicated in CART-signaling in in vitro experiments (Chiu et al., 2010). We next evaluated PKA and PKC function in CART signaling using specific pharmacological inhibitors of PKA and PKC kinase activity.
KT5720 [selective PKA inhibitor (PKAi)], when coinjected into the ventricle of starved fish along with CART peptide, alleviated (697.49 ± 105.59; p < 0.0001) the strong reduction in the number of biting attempts induced by CART peptide alone (30.33 ± 10.61 Fig. 7A). The inhibitor on its own had no significant effect on the feeding drive (713.92 ± 215.68) compared with starved control animals (782.9 ± 71.70; Fig. 7A).
CART-induced anorexia and activation of Dm neurons require PKA and PKC activity. A, The cumulative number of biting attempts made by starved fish intracerebroventricularly injected with either vehicle, CART, PKAi, or coinjected with PKAi along with CART (N = 6 animals per group). Data are represented as cumulative biting attempts in 15 min bins over a 1 h period and compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the CART-treated starved fish. Error bars indicate ± SEM; ns, not significant (**p < 0.01, ***p < 0.001; p = 0.0006 for vehicle plus vehicle compared with vehicle plus CART and p < 0.0001 for vehicle plus CART compared with PKAi plus CART). B, The number of p-ERK-ir cells in starved fish intracerebroventricularly injected with either vehicle, CART peptide, PKAi, or coinjected with PKAi and CART peptide. Data are compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the CART-treated starved fish (N = 6 animals per group; ***p < 0.001; p = 0.0002 for vehicle plus vehicle compared with vehicle plus CART and p = 0.0006 for vehicle plus CART compared with PKAi plus CART). C, The cumulative number of biting attempts made by starved fish intracerebroventricularly injected with either vehicle, CART peptide, PKCi, or coinjected with PKCi and CART peptide (N = 6 animals per group; p < 0.0001 for vehicle plus vehicle compared with vehicle plus CART and for vehicle plus CART compared with PKCi plus CART). Data are represented as cumulative biting attempts in 15 min bins over a 1 h period and compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the CART-treated starved fish. Error bars indicate ± SEM (***p < 0.001, **p < 0.01). D, The number of p-ERK-ir cells in starved fish intracerebroventricularly injected with either vehicle, CART, PKCi, or coinjected with PKCi along with CART. The data are compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the CART-treated starved fish (N = 6 animals per group; ***p < 0.001; p < 0.0001 for vehicle plus vehicle compared with vehicle plus CART and for vehicle plus CART compared with PKCi plus CART). E, F, Representative micrographs of one lobe of the Dm region from the transverse sections of the telencephalon showing phospho NR1 (Ser-897) immunoreactivity in starved fish intracerebroventricularly injected with either CART peptide or coinjected with PKAi and PKCi along with CART peptide. G, Quantification of mean pNR1 intensity per neuron in the Dm of starved fish intracerebroventricularly injected with either vehicle and CART or coinjected with PKAi and PKCi along with CART peptide. au, Arbitrary unit. The data were compared using the Mann–Whitney U test (12,253 cells in vehicle plus CART group and 12,015 cells in PKAi plus PKCi plus CART group; N = 3 animals per group; ***p < 0.001, p < 0.0001). Scale bar, 50 µm.
Parallel studies evaluating the activity of Dm neurons revealed a robust suppression of CART-induced neuronal activation in fish coadministered with PKAi and CART peptide (158.1 ± 56.47; Fig. 7B; p = 0.0006) compared with those injected with only CART peptide (554.5 ± 82.75; Fig. 7B).
As observed for PKAi, inhibition of PKC by GF109206X [selective PKC inhibitor (PKCi)] also reversed the suppression of the feeding drive by CART peptide (489.8 ± 29.38 for PKCi plus CART and 29.83 ± 6.76 for CART alone; Fig. 7C; p < 0.0001) and abrogated the activation of Dm neurons by CART (247.7 ± 28.99 for PKCi plus CART and 464.9 ± 19.57 for CART alone; Fig. 7D; p < 0.0001).
Together, these data implicate PKA and PKC activity in CART-mediated activation of Dm neurons and satiety. PKC and PKA are known to phosphorylate mouse NR1 at multiple residues, including serine 897 (Chen and Roche, 2007). Specific antibodies against phosphoserine 897 of the mouse NR1 (pNR1) were used to evaluate the status of NR1 phosphorylation in the Dm neurons. Although the Dm of CART-treated animals displayed strong pNR1 fluorescence (892.1 ± 5), coinjection of PKAi, PKCi, and CART peptide greatly reduced the pNR1 signal (505.9 ± 2.02; p < 0.0001; Fig. 7E–G).
Collectively, these data identify CART signaling mediated phosphorylation of NR1 via PKA and PKC. The heightened neuronal activity of Dm neurons could be attributed to the potentiation of NMDAR function driven by NR1 phosphorylation.
CART treatment enhances NMDA receptor function resulting in increased sensitivity of Dm neurons to excitatory stimuli
Phosphorylation of NR1 subunit of NMDARs by PKA and PKC is known to enhance NMDAR signaling (Lan et al., 2001; Skeberdis et al., 2006). We speculated that a CART-induced increase in phosphorylation of NMDARs would sensitize and enhance the response of Dm neurons to glutamatergic inputs. We tested this hypothesis directly using the Tg(NeuroD:GCaMP6f) zebrafish, which shows strong expression of the genetically encoded calcium sensor GCaMP6f in the Dm (Huang et al., 2020). Ex vivo preparations of the intact telencephalon were dissected from starved animals treated earlier with either saline or CART via the intracerebroventricular route (see above, Materials and Methods; Fig. 8A). Dm neurons of vehicle-treated animals failed to respond to glutamate stimulation. However, CART-peptide-treated animals showed heightened activity of Dm neurons in response to glutamate. Both the maximum amplitude of the response (0.44 ± 0.09 for vehicle control and 3.43 ± 0.6 for CART) and the magnitude (area under the curve) of the response (69.04 ± 15.29 for vehicle control and 522.6 ± 94.66 for CART) was substantially higher in starved animals receiving CART peptide (Fig. 8B–E; p = 0.0041 and p = 0.0046 for maximum amplitude and AUC, respectively). Cotreatment of AP5 along with CART peptide prevented the activation of the Dm neurons by glutamate (maximum amplitude, 0.29 ± 0.11 compared with 1.724 ± 0.44 for CART alone, p = 0.0308; magnitude, 64.76 ± 10.88 compared with 230.7 ± 41.33 for CART alone, p = 0.0139) and underscored the involvement of NMDAR signaling (Fig. 8F–H).
CART treatment enhances NMDAR function resulting in increased sensitivity of Dm neurons to glutamate. A–I, A timeline of the experimental protocol (A) followed for all calcium imaging experiments using ex vivo telencephalon preparations (see above, Materials and Methods). Representative micrographs of neural activity (GCaMP6f fluorescence) in the Dm of vehicle (B) and CART peptide (B')–injected Tg(NeuroD:GCaMP6f) transgenic animals following stimulation with glutamate for the same time. The ventricle is toward the top of the image; C and R indicate caudal and rostral directions, respectively. Scale bar, 50 µm. In all graphs (C, F, I), the traces represent relative change in fluorescence intensity (ΔF/F0) across time. The dark line represents the mean, and the dotted lines denote SEM. The shaded rectangular box indicates the duration of glutamate presentation. The change in fluorescence intensity (ΔF/F0) of Dm neurons (C) in starved fish treated with either CART peptide (blue trace) or vehicle (brown trace). The maximum response (ΔF/F0; p = 0.0041, D) and the extent of the total response (E) AUC (p = 0.0046) for Dm neurons in starved fish treated with either CART peptide or vehicle. The data were compared using the t test with Welch's correction (N = 6 telencephalons per group; **p < 0.01). The change in fluorescence intensity (ΔF/F0) of Dm neurons (F) in starved fish treated with either CART peptide (blue trace) or coinjected with AP5 and CART peptide (brown trace). The maximum response (ΔF/F0; p = 0.0308, G) and the extent of the total response of Dm neurons (p = 0.0139, H) in starved fish treated with either CART peptide or coinjected with AP5 and CART. The data were compared using the t test with Welch's correction (N = 5 for CART and N = 6 for AP5 plus CART telencephalons; *p < 0.05). The change in fluorescence intensity (ΔF/F0) of Dm neurons (I) in starved fish treated with either CART peptide (blue trace) or coinjected with PKAi and PKCi along with CART peptide (brown trace). J, The maximum response (ΔF/F0; p = 0.0033). K, The extent of the total response of Dm neurons (p = 0.0058) in starved fish treated with either CART peptide or coinjected with PKAi, PKCi, along with CART peptide. The data were compared using the t test with Welch's correction (N = 6 telencephalons per group; **p < 0.01).
To determine whether the change in the glutamate sensitivity of Dm neurons in response to CART required PKA and PKC activity, we coadministered selective PKA and PKC inhibitors along with CART. Interestingly, blocking PKA and PKC activity abolished the CART-induced increased sensitivity of Dm neurons to glutamate (Fig. 8I–K). The maximum amplitude of the response was reduced to 0.21 ± 0.06 (PKAi plus PKCi plus CART) from 2.48 ± 0.43 (CART alone; p = 0.0033), and the magnitude of the response dropped from 380.3 ± 75.86 (CART alone) to 33.39 ± 7.71 (PKAi plus PKCi plus CART; p = 0.0058).
Together these data suggest that CART signaling via PKA and PKC sensitizes NMDA receptors in the Dm neurons, possibly via post-translational modification of NR1 subunits, leading to the enhanced activation of these neurons. This CART neuropeptide-mediated response of the Dm neurons may constitute a neural representation of the sated state of the animal.
The activity of Dm neurons is tuned to changes in energy states
To assess whether Dm exhibits heightened activity in energy-rich metabolic states, we used ex vivo whole-brain preparations from starved animals maintained in sucrose (to maintain gluco-deprived conditions). Following initial evaluation of the response of Dm neurons to glutamate stimulation, the preparation was incubated in glucose to mimic energy-rich conditions and was re-evaluated for neural activity in response to glutamate stimulation (see above, Materials and Methods; Fig. 9A).
Activity of Dm neurons increases in response to glucose. A, A timeline of the experimental protocol followed for calcium imaging experiments using ex vivo whole-brain preparations (see above, Materials and Methods). B, The traces represent relative change in fluorescence intensity (ΔF/F0) over time in response to glutamate. The thicker line represents the mean, whereas the associated finer lines indicate ± SEM. The shaded rectangular box indicates the duration of glutamate presentation. Response of Dm neurons in ex vivo whole-brain samples after 30 min incubation with glucose (dark blue trace, postglucose) is substantial compared with preglucose treatment (dark brown trace). Control experiments showed no response of Dm neurons before (light brown, precontrol) and after (light blue, postcontrol) 30 min incubation in glucose-free media. C, The maximum response (ΔF/F0). D, The extent of the total response of Dm neurons preincubation and postincubation with glucose/glucose-free media. The data were analyzed using one-way ANOVA and Tukey's multiple comparisons test (N = 5 brains in each group; ***p < 0.0001 for preglucose compared with postglucose, precontrol compared with postglucose, and postcontrol compared with postglucose in terms of maximum ΔF/F0 and AUC).
Glutamate stimulation of whole-brains of starved animals maintained in gluco-deprived conditions (preglucose) failed to activate Dm neurons (maximum amplitude of response, 0.243 ± 0.1, and magnitude of response, 48.29 ± 14.71; Fig. 9B–D). However, on incubation of the same whole-brain preparation with 10 mm glucose for 30 min (postglucose), the Dm neurons responded strongly to glutamate stimulation (maximum amplitude of response, 2.33 ± 0.29, and magnitude of response, 485 ± 48.05; p < 0.0001 for both parameters; Fig. 9B–D). In control experiments, whole-brain preparations from starved animals were maintained throughout in glucose-free media, including the 30 min period between the two glutamate stimulations. In these control experiments, Dm neurons did not respond to glutamatergic stimulation either before (precontrol) or after the 30 min control incubation (postcontrol; Fig. 9B–D).
These experiments indicate that activity in the Dm is tuned to changes in glucose levels in the brain, and neuronal activity in Dm is correlated to energy states. As CART levels are sensitive to metabolic states (Fig. 3; Mukherjee et al., 2012; Akash et al., 2014), the glucosensitive neurons of the EN and hypothalamus may contribute to elevating endogenous CART signaling at the Dm neurons to mediate this effect. Interestingly, as our ex vivo preparation only included the brain, the glucose-mediated changes in glutamate-induced Dm activity appears to be independent of peripheral mediators of energy homeostasis.
NPY promotes food intake in zebrafish
Beyond sensing energy sufficiency, interoceptive awareness is also expected to involve signals indicating depletion of energy stores. Neuropeptide Y (NPY) is a potent orexigenic neuropeptide (Zhang et al., 2019) that is expressed in the EN and periventricular hypothalamic regions of teleosts, including zebrafish (Mathieu et al., 2002; Mukherjee et al., 2012; Akash et al., 2014; Turner et al., 2016) and regulates feeding behavior (Yokobori et al., 2012). We tested whether NPY signaling regulated zebrafish feeding drive in a manner obverse to that of CART.
A selective NPY receptor Y1R (Y1R) antagonist, BIBP-3226 (Doods et al., 1995; Yokobori et al., 2012; Kaniganti et al., 2021) was administered to starved fish to evaluate endogenous NPY signaling in regulating the feeding drive. Compared with vehicle-injected starved animals (789 ± 69.46), BIBP introduced by the intracerebroventricular route robustly suppressed the biting frequency (144.7 ± 21.57; p < 0.0001; Fig. 10A) and is consistent with earlier reports (Yokobori et al., 2012).
NPY promotes food intake in zebrafish. A, The cumulative number of biting attempts made by starved fish intracerebroventricularly injected with either vehicle or BIBP3226 (N = 6; **p < 0.01, ***p < 0.001). B, The cumulative number of biting attempts made by fed fish intracerebroventricularly injected with either vehicle or NPY peptide (N = 6; **p < 0.01, ***p < 0.001). CLUSTAL Omega alignment of the human and zebrafish NPY peptides reveal a high degree of amino acid identity (Extended Data Figure 10-1). Data are represented as cumulative biting attempts in 15 min bins over a 1 h period and were compared using two-way ANOVA with Bonferroni's post hoc analysis.
Figure 10-1
Alignment of human and zebrafish NPY amino acid sequences. A, CLUSTAL Omega alignment of peptide sequences of human NPY (36 aa; EAW93807.1) and zebrafish NPY (AAI62071.1). Dark purple background represents identical residues, light purple background represents similar residues, and white background represents different residues. The percentage identity with human NPY (55–102) is indicated on the right. Download Figure 10-1, TIF file.
To evaluate whether NPY could increase the feeding drive, we intracerebroventricularly introduced NPY peptide into recently fed fish. As expected, vehicle-injected fed fish made infrequent biting attempts (338.8 ± 70.4), whereas NPY-injected animals showed an increased feeding drive (983.7 ± 75.84; p = 0.0004) that was comparable with that of starved animals (Fig. 10B). However, there was no change in locomotion or anxiety-like behaviors in NPY-treated animals (Fig. 2).
NPY signaling converges on the Dm neurons to decrease neuronal activity
Increased expression of NPY in response to starvation has been reported in several species (Loh et al., 2015), and also observed in the hypothalamic nuclei of zebrafish (Yokobori et al., 2012). Compared with fed zebrafish, the number of NPY-positive neurons increased in the EN and the hypothalamic nucleus recessus lateralis (NRL) of starved animals (Fig. 11A–F).
NPY expression is upregulated in the EN and the NRL in response to starvation. A–F, Representative micrographs of the EN (A, B) and the NRL (D, E) regions from transverse sections of the zebrafish brain showing NPY-immunoreactive cells in starved and fed fish. The quantification of the number of NPY-positive cells in EN (C) and NRL (F) in starved and fed fish. The data were compared using unpaired t test with Welch's correction (n = 6 EN/NRL from 3 animals per group; *p < 0.05, **p < 0.01). G, Transverse section of telencephalon showing NPY-immunoreactive neurons in EN and NPY-containing fiber tracts projecting to Dm region (dotted line). H, Schematic showing connectivity of EN and hypothalamic CART and NPY projections to the Dm (shaded region). OB, Olfactory bulb; ac, anterior commissure; Ha, habenula; Hy, hypothalamus; TeO, optic tectum. Scale bar, 25 µm.
In transverse slices of the telencephalon, NPY-expressing EN somata were found to project to the Dm (Fig. 11G), similar to what was seen for CART-positive EN neurons. EN to Dm projections are consistent with previous reports (Turner et al., 2016). Together with DiI tracing experiments reported earlier (Fig. 4B–D), the connectivity of periventricular hypothalamic nuclei and the EN to Dm (Fig. 11H) raises the possibility of NPY signaling also converging on the Dm and opposing CART activity. We tested this possibility by evaluating the activity of the Dm neurons in response to NPY.
Intracerebroventricular administration of NPY to recently fed animals reduced the number of p-ERK-positive Dm neurons (109.5 ± 4.74) compared with vehicle controls (353.5 ± 49.73; p = 0.0043; Fig. 12A–C). Thus, the difference in the activity of Dm neurons observed between fed and starved animals (Fig. 5D–F) can be partially explained by the opposing orexic and anorexic signaling by NPY and CART neuropeptides, respectively.
NPY reduces the activation of Dm neurons. A, B, Representative micrographs of Dm region from the transverse sections across telencephalon showing p-ERK-ir cells in fed fish intracerebroventricularly injected with either vehicle or NPY. C, The number of p-ERK-ir cells in Dm of fed fish intracerebroventricularly injected with either vehicle or NPY. The data were compared using unpaired t test with Welch's correction (N = 6; **p < 0.01, p = 0.0043). Scale bar, 50 µm. D, E, Representative micrographs of one lobe of Dm region from the transverse sections across telencephalon showing pNR1 immunoreactivity in starved fish intracerebroventricularly injected with either glucose or coinjected with NPY along with glucose. F, Quantification of mean pNR1 intensity per neuron in Dm in starved fish intracerebroventricularly injected with either glucose or coinjected with NPY along with glucose. au, Arbitrary unit. The data were compared using unpaired t test with Welch's correction (12,253 cells in vehicle plus glucose group and 11,901 in NPY plus glucose group; N = 3; ***p < 0.001, p < 0.0001) Scale bar, 50 µm. G, The traces represent the relative change in fluorescence intensity (ΔF/F0) over time in response to 75 μm glutamate. The dark line represents mean ± SEM, with the dotted lines denoting the limits for SE. The colored rectangular box indicates the duration during which glutamate was presented. The response of Dm neurons in starved fish intracerebroventricularly injected with glucose (blue trace) and coinjected with NPY along with glucose (brown trace). H, I, The maximum (ΔF/F0; p = 0.0463, H) as well as AUC of response of Dm neurons (p = 0.0132, I) in starved fish intracerebroventricularly injected with glucose compared with the starved fish coinjected with NPY along with glucose. The data were analyzed using the t test with Welch's correction (N = 6 telencephalons per group; *p < 0.05).
As CART signaling increased the phosphorylation of the NR1 subunit of NMDAR to increase neuronal activity, we tested whether the reduction of Dm activity following NPY treatment was a result of decreased NR1 phosphorylation. Indeed, glucose-stimulated (satiety mimic) pNR1 fluorescence (944.3 ± 2.98) was substantially decreased (478.2 ± 2.03; p < 0.0001) when NPY peptide was coinjected with glucose into the brain of starved fish (Fig. 12D–F).
These data identify NPY signaling, originating from the energy state–sensitive populations of the hypothalamus and EN under energy-poor metabolic states in modulating zebrafish feeding behavior. NPY signaling converges on the Dm neurons, where it acts antagonistically to CART to reduce NR1 phosphorylation and decreases neuronal activity at the Dm.
NPY signaling reduced the glutamate responsiveness of Dm neurons
To directly test whether NPY signaling reduced the response of Dm neurons to glutamatergic stimulation, we evaluated the Dm activity in the Tg(NeuroD:GCaMP6f) zebrafish line. Ex vivo preparations of the intact telencephalon were dissected from starved animals treated earlier with either glucose (satiety mimic) or glucose and NPY via the intracerebroventricular route.
As expected, Dm of glucose-injected animals responded to glutamate stimulation (maximum amplitude of response, 0.39 ± 0.09, and magnitude of response, 75.45 ± 12.93). However, the response to glutamate was substantially damped in the Dm of fish coinjected with glucose and NPY (maximum amplitude of response, 0.13 ± 0.06; p = 0.0463, and magnitude of response, 28.43 ± 4.7; p = 0.0132; Fig. 12G–I).
These data show that NPY signaling alters the responsiveness of Dm neurons to glutamate, and this altered activity state of Dm may be a neuronal representation of the energy deficit conditions. The orexic NPY signaling and the anorexic activities of the CART neuropeptide converge on the Dm neurons. Dm thus appears to be an important integrative center whose activity is correlated with the homoeostatic regulation of feeding.
Orexic signaling by NPY is mediated by the activation of calcineurin and inhibition of adenylyl cyclase
Our data identify PKA and PKC–mediated phosphorylation of NR1 in the Dm neurons in response to CART in inducing satiety. Further, NPY signaling appears to oppose CART function, including the dephosphorylation of NR1 (Fig. 12F). Thus, we investigated NPY-induced signaling that could oppose CART function.
In mammals, NPY increases the activity of the protein phosphatase calcineurin (Chen et al., 2005; Sajdyk et al., 2008), and calcineurin is known to dephosphorylate NR1 (Choe et al., 2005). We blocked calcineurin activity using FK506 (calcineurin inhibitor) in NPY-peptide-treated fed fish and evaluated their feeding behavior. FK506 alone (222 ± 57.39) did not show any significant effect on the number of biting attempts compared with vehicle controls (291.4 ± 79.85; Fig. 13A). However, the coinjection of FK506 along with NPY (368 ± 50.97) reversed NPY-induced orexia (954.7 ± 67.55; p < 0.0001; Fig. 13A).
NPY inhibits adenylyl cyclase and activates of calcineurin to promote food intake. A, The cumulative number of biting attempts made by fed fish intracerebroventricularly injected with either vehicle, NPY, FK506, or coinjected with FK506 along with NPY (N = 5 animals per group; p < 0.0001 for vehicle plus vehicle compared with vehicle plus NPY and for vehicle plus NPY compared with FK506 plus NPY). Data are represented as cumulative biting attempts in 15 min bins over a 1 h period and compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the NPY-treated fed fish. Error bars indicate ± SEM (***p < 0.001). B, The number of p-ERK-ir cells in fed fish intracerebroventricularly injected with either vehicle, NPY, FK506, or coinjected with FK506 along with NPY. Data were compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the NPY-treated fed fish (N = 5; ***p < 0.001; p = 0.001 for vehicle plus vehicle compared with vehicle plus NPY and p < 0.0001 for vehicle plus NPY compared with FK506 plus NPY). C, The cumulative number of biting attempts made by fed fish intracerebroventricularly injected with either vehicle, NPY, Forskolin, or coinjected with Forskolin along with NPY (N = 5 animals per group). Data are represented as cumulative biting attempts in 15 min bins over a 1 h period and were compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the NPY-treated fed fish. Error bars represent ± SEM (*p < 0.05, ***p < 0.001; p < 0.0001 for vehicle plus vehicle compared with vehicle plus NPY and for vehicle plus NPY compared with Forskolin plus NPY). D, The number of p-ERK-ir cells in fed fish intracerebroventricularly injected with either vehicle, NPY, Forskolin, or coinjected with Forskolin along with NPY. Data are compared using two-way ANOVA with Bonferroni's post hoc analysis for significance in comparison with the NPY-treated fed fish (N = 5; **p < 0.01; p = 0.0043 for vehicle plus vehicle compared with vehicle plus NPY and p = 0.0030 for vehicle plus NPY compared with Forskolin plus NPY). E, Schematic showing CART and NPY projections to Dm from interoceptive regions in zebrafish brain. OB, Olfactory bulb; ac, anterior commissure; Ha, Habenula; Hy, Hypothalamus; TeO, Optic tectum. E', Summary of modulatory changes induced by CART and NPY leading to a corresponding change in neural activity and food intake under fed and starved conditions, respectively. Under energy-enriched conditions (Fed), CART signaling at the Dm is increased. CART activates PKA and PKC to increase the phosphorylation of NR1 subunit of NMDARs and thereby increases NMDAR activity in Dm neurons. CART-mediated increase in Dm neuronal activity is correlated with a decrease in food intake in zebrafish. In contrast, under energy-deprived conditions, NPY signaling at the Dm is increased. NPY acting via Y1R reduces PKA activity and activates calcineurin to decrease phosphorylation of NR1, in turn, decreasing NMDAR activity. NPY-mediated reduction in Dm neuronal activity is correlated with increased food intake in zebrafish.
To test whether blocking calcineurin activity affects NPY-induced reduction in Dm activity, we coinjected FK506 and NPY peptide in fed fish. Treatment with the FK506 alone (299.8 ± 38.11) did not affect the number of p-ERK-positive neurons in the Dm of fed fish (409.6 ± 32.62; Fig. 13B). However, blocking calcineurin activity in NPY-treated fish (502.4 ± 56.7) abrogated NPY-induced (111 ± 9.38; p < 0.0001) reduction in p-ERK-positive neurons in Dm.
We have demonstrated that NPY function in regulating food intake in zebrafish is via the Y1 receptor (Fig. 10A). NPY Y1R is a G-protein-coupled receptor that inhibits adenylyl cyclase activity (Molosh et al., 2013). A fall in cAMP production by adenylyl cyclase is expected to reduce the activation of PKA, and PKA is critical for CART signaling (Fig. 7A,B). Therefore, we activated adenylyl cyclase with Forskolin (an adenylyl cyclase agonist) in fed fish cotreated with NPY. Forskolin alone (236 ± 77.40) did not influence the number of biting attempts compared with vehicle controls (261.8 ± 86.98; Fig. 13C); however, application of Forskolin in NPY coinjected fed fish reversed (194.7 ± 40.7) NPY-induced orexia (929 ± 96.86; p < 0.0001; Fig. 13C).
Forskolin coadministered along with NPY in fed fish, via the intracerebroventricular route, alleviated (304.1 ± 49.90; p = 0.0030) NPY-mediated reduction of p-ERK-positive neurons in the Dm of fed animals (145.1 ± 23.68; p = 0.0043; Fig. 13D). Treatment with forskolin alone (223.8 ± 20.05) was comparable to vehicle-injected fed fish (296.6 ± 56.19; Fig. 13D).
Collectively, these experiments demonstrate that orexic NPY signaling is mediated by the downregulation of cAMP activity and upregulation of calcineurin activity, which decreases the activity of Dm neurons via reduced phosphorylation of NR1. NPY signaling is antagonistic to CART-induced phosphorylation of NR1, and the energy state–dependent expression of CART/NPY results in the bimodal regulation of Dm activity and feeding behavior.
Discussion
The flexibility of innate, motivated behaviors is driven by context-dependent recruitment and outputs of the underlying neural circuits (Sternson, 2013; Flavell et al., 2020). Neuromodulator-based signaling has emerged as a key mechanism facilitating reconfiguration of circuit activity (Lee and Dan, 2012). For instance, state changes between component behaviors in Caenorhabditis elegans foraging is regulated by the differential recruitment of circuits by the opposing activities of serotonin and the PDF neuropeptide (Flavell et al., 2013). Neuropeptide activities may also change the response of specific neurons or circuits. Peptidergic modulation of the worm ASH neuron mediates feeding state-dependent nociceptive responses (Ezcurra et al., 2016). Similarly, NPY signaling regulates the sensitivity of zebrafish olfactory receptor neurons in response to hunger (Kaniganti et al., 2021).
Homoeostasis necessitates altering food intake in response to changing energy stores. Therefore, the neural circuits regulating food intake need to be dynamic and responsive to physiological states. Neuropeptidergic signaling may represent energy state information and facilitate biochemical modulation of synaptic transmission to regulate food-intake-associated behaviors. Indeed, a diverse and complicated network of peptidergic activity, both peripheral and central, influence the plasticity of neural circuits regulating food intake (Crespo et al., 2014). However, the neuropeptide-induced signaling facilitating the integration of opposing energy state signals to generate appropriate behavioral outputs remains poorly understood.
In the mammalian Arc, the POMC/CART-expressing neurons of the Arc release α-MSH onto the MC4R-expressing neurons of the paraventricular hypothalamic nucleus (PVH) to enhance glutamatergic neurotransmission and induce satiety (Fenselau et al., 2017). Conversely, the GABAergic AgRP/NPY neurons, on the other hand, provide an inhibitory tone onto MC4R-expressing PVH neurons to promote food intake (Fenselau et al., 2017). Although opposing inputs from the Arc converge onto the PVH neurons, the intracellular mechanisms recruited by these peptidergic neuromodulators are inadequately characterized.
We demonstrate the convergence of antagonistic intracellular signaling mediated by the anorexic neuropeptide CART and the orexic NPY on a zebrafish telencephalic neuronal population to reconfigure its activity. Energy state–sensitive expression of the neuropeptides imposes opposing biochemical constraints on their target neurons at the dorsal telencephalon via the modulation of NMDAR function and regulates feeding behavior.
We show that not only do the endogenous activities of CART and NPY modulate the feeding drive but the intracerebroventricular delivery of these peptides acutely alters food intake in the expected direction. We limit our analysis to intracerebroventricular delivery of bioactive molecules and pharmacological agents to focus on central activities and avoid potential confounds originating from peripheral inputs. Consistent with earlier work (Subhedar et al., 2011; Mukherjee et al., 2012; Yokobori et al., 2012; Akash et al., 2014), we also find that expression of CART peptide increases, under energy-rich conditions, in energy state interoceptive regions of the periventricular hypothalamus and the EN of zebrafish, whereas NPY expression increases on starvation. Unlike NPY, at least four different CART genes have been identified in zebrafish with the CART 2 gene product sharing the highest amino acid identity (79.2%) with the mammalian CART peptide (Akash et al., 2014). In the EN, only CART 2 expression is sensitive to energy states (Akash et al., 2014), although all the other zebrafish CART gene products show high amino acid identity with the mammalian peptide (Extended Data Fig. 1-1). The immunoneutralizing antibody used in our experiments was raised against the rat CART peptide, and our results suggest that it successfully blocks functional CART peptide signaling in zebrafish. It is likely that all the zebrafish CART peptides, especially the CART 2 gene product, are neutralized by the antibody.
Anatomical analysis indicated that the peptidergic neurons of the periventricular hypothalamus and the EN project to the Dm and suggested that the latter population may be the target for CART or NPY under different metabolic states. The connectivity observed in our tracing studies is in line with observations reported in other teleosts (Kanwal et al., 1988), including zebrafish (Turner et al., 2016; Yáñez et al., 2017). Thus, our subsequent studies focused on the neuronal activity at the Dm in response to CART and NPY.
Using p-ERK as a readout of recent neuronal activity and calcium imaging of a genetically encoded calcium sensor, we found that energy-rich states resulting from recent feeding (and consequent increase in endogenous CART neuropeptide) or exogenous application of CART peptide to the brain led to heightened activity of the Dm neurons. The increased activity of the Dm neurons was found to be NMDAR dependent, and blocking NMDAR signaling reduced Dm activity and CART-mediated anorexia.
Studies in rodents have previously reported the involvement of NMDA receptors in the regulation of food intake. Under satiety conditions, NMDA receptors in the hindbrain are required for CCK-induced reduction of food intake (Wright et al., 2011). NMDAR function has also been implicated in leptin (Neyens et al., 2020) and PACAP (pituitary adenylate cyclase-activating polypeptide; Resch et al., 2014) mediated reduction in food intake. Furthermore, NMDAR activity in the lateral hypothalamus regulates food intake in hungry animals (Stanley et al., 2011), whereas it is necessary for the activity of the AgRP-expressing neurons to establish the starvation state (Liu et al., 2012). In addition, CART function in processing innate fear in the central amygdala (Rale et al., 2017) and nociception by sensory neurons (Chiu et al., 2010) also involve NMDAR function and may represent a common modality in CART signaling.
Further investigations identified PKA and PKC activities in mediating CART-induced anorexia since inhibiting the activity of either kinases led to the restoration of food intake and decreased neuronal activation in the Dm. Using phosphospecific NR1 antibodies, we found that CART-induced activation of PKA and PKC results in phosphorylation of NR1 on specific residues.
Phosphorylation of NR1 by PKC and PKA has been known to enhance NMDAR signaling by increasing both the Ca2+ permeability (Skeberdis et al., 2006) and the opening probability of the receptor complex (Lan et al., 2001). Thus CART-induced phosphorylation is likely to increase the responsiveness of Dm neurons to glutamatergic inputs and is consistent with our results showing heightened calcium response of Dm neurons on stimulation by glutamate.
The modulation of Dm activity by endogenous CART release in response to energy-rich conditions was demonstrated using calcium imaging of ex vivo whole-brain preparations where the inputs from the hypothalamus and EN are preserved. Brains isolated from starved animals (low endogenous CART and high endogenous NPY) failed to respond to glutamate when maintained under glucose-deprived conditions. However, incubation with glucose restored glutamate responsiveness and suggests altered inputs (including CART) from the glucose-sensing interoceptive areas. Although this response may involve modulatory inputs to the Dm not limited to CART, it is consistent with our observations of CART function. Importantly, as the experimental preparation precludes peripheral signals of energy state, the modulation of Dm activity appears to be primarily driven by neuronal inputs.
NPY-containing neurons of the EN and periventricular hypothalamus also project to the Dm. Thus, the Dm neurons could function as a common substrate for orexic and anorexic inputs, and their activity could be dynamically tuned to the prevailing energy states. Indeed, both starvation and exogenous application of NPY to fed fish (low levels of endogenous NPY) led to a reduction in Dm activity with a concomitant increase in the feeding drive.
Our investigation reveals that NPY opposes CART-mediated elevation of NR1 phosphorylation and consequent increase in Dm activity and anorexia by reducing cAMP production (thereby reducing PKA activation) and activating the phosphatase calcineurin. Our observations are consistent with NPY-mediated activation of calcineurin (Chen et al., 2005; Sajdyk et al., 2008) and calcineurin mediated dephosphorylation of NR1(Choe et al., 2005). Further, NPY Y1R–mediated reduction in cAMP and PKA has been shown to decrease NMDA-mediated postsynaptic currents (Molosh et al., 2013) in the neurons of the basolateral amygdala. Although no colocalization of Y1R and NMDAR in the zebrafish Dm is reported so far, these receptors are known to colocalize in the mammalian amygdala. (Molosh et al., 2013).
Our data suggest that Dm neurons are modulated by the opposing activities of CART and NPY to achieve two distinct activity states (Fig. 13E,E'). Dm neurons in zebrafish may represent second-order neurons immediately downstream of the interoceptive first-order population in the hypothalamus and EN. Although our experiments are unable to distinguish whether the same individual neurons respond differentially to CART and NPY, the tight correlation of Dm activity with alterations in food intake behavior suggest a pivotal role for Dm neurons in regulating energy homoeostasis (Fig. 13E'). A major limitation in delineating CART function has been the lack of well-characterized receptors and selective pharmacology. This drawback has limited our analysis of Dm neurons. However, collateral evidence from the gustatory systems underscores the role of Dm in food intake. In a range of teleosts, the preglomerular tertiary gustatory nucleus is reciprocally connected to the Dm (Kanwal et al., 1988; Kato et al., 2012). Although the downstream targets of the Dm relevant to the feeding drive are unknown, it appears CART/NPY activities modulate the responsiveness of the Dm to glutamatergic inputs to release or inhibit food intake.
CART neurons of the EN are glutamatergic (Fig. 4G–I) and suggest corelease of CART and glutamate from EN neurons onto the Dm. However, glutamatergic inputs to the Dm may also come from sources other than the CART neurons of the EN. For example, Dm inputs from the dorsolateral region of the telencephalon and the nucleus of the lateral olfactory tract (nLOT) are also glutamatergic (Ng et al., 2012; Porter and Mueller, 2020; Wu et al., 2017). EN and periventricular hypothalamus–derived CART neuropeptide could be released diffusively in volume transmission mode and act on synapses receiving glutamatergic inputs from non-EN sources.
The Dm is considered to be homologous to the mammalian amygdaloid complex (Wullimann and Mueller, 2004; Porter and Mueller, 2020). Several previous reports identify the mammalian amygdalar complex in the regulation of feeding behaviors (Padilla et al., 2016; Kim et al., 2017; Izadi and Radahmadi, 2022). Consistent with our findings, the amygdala receives projections from both the NPY and CART neurons of the hypothalamic Arc (Hwang and Lee, 2018) and employs NPY Y1–mediated signaling to increase food intake (Padilla et al., 2016). Additionally, projections from dopamine receptor-positive neurons in prefrontal cortex are known to activate medial basolateral amygdala (BLA) to increase food intake (Land et al., 2014). The amygdala is also involved in hindbrain satiety circuits, where PKC-δ+ neurons in the central amygdala (CeA) integrate multiple anorexigenic signals to inhibit food intake (Cai et al., 2014). Conversely, Htr2a+ neurons of the CeA have been shown to stimulate food intake (Douglass et al., 2017). In addition, several anorexigenic neuromodulators, when administered to the CeA, have been shown to affect food intake (Fekete et al., 2007; Beckman et al., 2009; Kovács et al., 2012). A previous study has identified appetitive information specific projections from the BLA to the CeA and isolated specific CeA subpopulations that differentially mediate fear processing and feeding-associated behaviors (Kim et al., 2017).
The fish EN is homologous to the globus pallidus, an output nucleus of the basal ganglia, in mammals (Amo et al., 2014; Turner et al., 2016). The mammalian globus pallidus/EN expresses both CART (Abraham et al., 2009) and NPY (Busch-Sørensen et al., 1989) and is known to contain glucosensitive neurons (Karádi et al., 1995; Lénárd et al., 1995). Further, the globus pallidus and the amygdala are reciprocally connected (Saunders et al., 2015; Giovanniello et al., 2020; Dong et al., 2021). Thus, the EN–Dm/globus pallidus–amygdala circuit, although poorly investigated, is found in both teleosts and mammals and may serve a role in energy homoeostasis across vertebrate phyla along with the better characterized Arc–PVH (preoptic area in fish; Soengas et al., 2018) axis. Further investigations will be necessary to evaluate the relative valence of these two circuits and their evolutionary origins.
Although the microcircuitry and the diversity of neuronal subtypes in the zebrafish Dm remain unknown, the feeding-regulation-associated CART/NPY-responsive neurons could be distinct from those processing emotions. Consistent with this, no difference in anxiety-like behaviors was seen in fish under fed and starved states or treated with either of the neuropeptides (Fig. 2). The CART/NPY-responsive neurons are a subpopulation restricted to a specific region of the Dm, starting at the level of the anterior commissure and extending 250 µm rostrally (Fig. 5G–I). Identification of neuronal subtypes within the Dm and development of intersectional genetics will be critical in characterizing the CART/NPY-responsive microcircuits in the future.
This study investigates neural mechanisms mediating the opposing activities of the neuropeptides CART and NPY in regulating the feeding drive in zebrafish and identifies plasticity in a group of neurons in the zebrafish forebrain. CART and NPY inputs converge on Dm neurons to antagonistically modulate NMDAR function, resulting in an energy state–dependent alteration in glutamatergic neurotransmission that is tightly correlated with the behavioral switch in food intake.
Footnotes
This work was supported by Department of Biotechnology, Ministry of Science and Technology (DBT) Grant BT/PR26241/GET/119/244/2017 to A.G. D.S.B. was supported by a fellowship from the Indian Institute of Science Education and Research (IISER), Pune, and A.M. was supported by a fellowship from the Council for Scientific and Industrial Research. The National Facility for Gene Function in Health and Disease at IISER Pune is supported by DBT Grant BT/INF/22/SP17358/2016. We thank Dr. G. Deshpande, Princeton University; Dr. S. Rath and Dr. R. Rajan, IISER Pune, for reading the manuscript and discussions; Dr. V. Thirumalai, National Centre for Biological Sciences, Bengaluru, for the gift of the Tg(NeuroD:GCaMP6f) line; N. Tiwari, IISER Pune for contributing to the development of the feeding assay; the IISER Pune Microscopy Facility, and the National Facility for Gene Function in Health and Disease at IISER Pune for access to equipment and infrastructure.
The authors declare no competing financial interests.
- Correspondence should be addressed to Aurnab Ghose at aurnab{at}iiserpune.ac.in
