Blockade of Ghrelin Receptor Signaling Enhances Conditioned Passive Avoidance and Context-Associated cFos Activation in Fasted Male Rats

Avoiding danger is critical for survival, and animals are equipped with neural circuits that enable adaptive avoidance responses to stimuli previously associated with potentially dangerous, aversive experiences. However, excessive conditioned avoidance can be maladaptive, contributing to clinical anxiety and other stress-related disorders [1‒3]. Conditioned avoidance responses are based on learning and memory processes that include both conscious and subconscious components. Sensory feedback from body to brain during the learning and expression of emotional responses plays a major role in the subconscious modulation of motivated behavior [4]. For example, short-term negative energy balance generally biases animals toward approach behaviors and away from avoidance behaviors, thereby facilitating exploratory food-seeking [5, 6]. In recent work, we demonstrated a reduction in learned passive avoidance behavior in male rats that were tested following an overnight fast [7]. The present study was designed to investigate potential mechanisms and neural circuits through which interoceptive signals related to overnight food deprivation may act to suppress conditioned passive avoidance behavior.

Interoceptive feedback about the internal state of the body is continuously communicated to the brain through hormonal and neural pathways. The neural pathways feature vagal afferent inputs to the caudal nucleus of the solitary tract (cNTS) that can powerfully modulate motivated behavior [4, 8‒10]. Vagal afferents communicate a diverse array of cardiovascular, respiratory, gastrointestinal, and inflammatory signals to the cNTS in a manner that is further modulated by the hormonal and metabolic state [11‒15]. Within the cNTS, the A2 noradrenergic (NA) cell group receives direct synaptic input from vagal sensory afferents [16, 17], and A2 neurons are activated to express the immediate-early gene product, cFos, in response to vagal sensory stimulation [17‒20]. Central NA signaling pathways are highly implicated in interoceptive state-dependent modulation of feeding and other motivated behaviors, and genetically distinct subpopulations of brainstem NA neurons appear to play different behavioral roles [21]. For example, stimulation of catecholaminergic NTS neurons (likely including A2 neurons) projecting to the mediobasal hypothalamus is sufficient to increase food intake in mice [22], whereas food intake is suppressed after stimulation of A2 neuronal projections to the parabrachial nucleus [23]. Thus, separate populations of NA neurons may exert distinct effects on different types of motivated behavior.

A caudal subset of A2 neurons that coexpress prolactin-releasing peptide (PrRP) become activated to express cFos in rats exposed to a variety of innate and conditioned stressors that suppress food intake and promote avoidance behavior [6, 7, 24, 25]. Interestingly, overnight food deprivation markedly reduces the ability of these innate and conditioned stress stimuli to activate PrRP+ A2 neurons [6, 7], concurrent with reduced cFos activation in hypothalamic and limbic forebrain regions that are innervated by PrRP+ neurons, including the paraventricular nucleus of the hypothalamus (PVN) [19] and the anterior ventrolateral bed nucleus of the stria terminalis (vlBNST) [6, 7].

Food deprivation increases circulating levels of acyl-ghrelin [26‒28], and acyl-ghrelin reduces tonic firing and stimulus-induced responsiveness of vagal sensory neurons [29, 30] and postsynaptic A2 neurons in the cNTS [31]. We recently demonstrated that systemic administration of a ghrelin receptor antagonist in fasted rats partially restores the ability of vagal sensory stimulation to activate PrRP+ A2 neurons and suppress food intake [20]. Further, caloric restriction reduces innate avoidance behavior in wild-type mice but not in mice lacking ghrelin receptors [5], which display more innate anxiety-like behavior than wild-type mice [32]. Consistent with these anxiolytic effects of ghrelin signaling, baseline acyl-ghrelin levels prior to auditory fear conditioning negatively predict conditioned freezing behavior in rats [33]. However, the potential impact of ghrelin signaling on the expression of learned avoidance has not been examined. Based on our previous research demonstrating that rats fasted overnight display less passive avoidance behavior and less fear context-induced neural cFos activation [7], the present study was designed to test the hypothesis that increased ghrelin receptor signaling contributes to the ability of overnight food deprivation to suppress learned passive avoidance behavior, and to suppress conditioned context-induced activation of PrRP+ neurons in the cNTS and their downstream targets in the PVN and anterior vlBNST.

All experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were reviewed and approved by the Florida State University Animal Care and Use Committee (approval number 202000001).

Adult male Sprague Dawley rats (Envigo; N = 32 [250–300 g body weight]) were pair-housed in standard tub cages in a temperature- and light-controlled housing environment (lights on from 04:00 h to 16:00 h). Only male rats were used in the current study based on evidence that food deprivation does not significantly reduce conditioned passive avoidance in female rats [7]. Rats were acclimated to handling for 3 days, with free access to water and rat chow (Purina 5001). Rats underwent passive avoidance training in a novel light/dark shuttle box (Coulbourn Instruments, Allentown, PA, USA), with training performed 4–6 h after lights on. The light/dark box comprised a light (illuminated) chamber with clear plastic walls and a smooth white plastic floor, and a dark (nonilluminated) chamber with black plastic walls and a metal grid floor. Each chamber measured 25 × 25 cm, with 28-cm-high walls and a ceiling. The two chambers were separated by a metal divider wall with a guillotine door that could be opened and closed remotely. For passive avoidance training, rats were initially placed individually into the light chamber of the box and the dividing guillotine door was lifted immediately to allow access to the dark chamber. As expected, due to their innate aversion to the light and preference for the dark, rats very quickly entered the dark chamber (average latency = 22.73 s). Upon entry into the dark chamber, the guillotine door was closed. After a 5-s delay, rats received a single mild electric footshock (0.6 mA; 1 s). Rats remained in the enclosed dark chamber for 30 s following footshock and then were returned to their home cage. Cagemates were similarly trained on the same day.

Two hours before dark onset on the day before testing (i.e., 2 days after training), all rats were transferred to clean cages with chow removed to initiate overnight food deprivation. On the next day, 4–6 h after lights on, rats received either an intraperitoneal (i.p.) injection of 0.15 M NaCl (saline) vehicle (∼0.3 mL) or the same volume of vehicle containing ghrelin receptor antagonist (GRA; [D-Lys3]-GHRP-6; 3.3 mg/kg BW; Sigma-Aldrich G4535). This GRA dose was based on our previous report demonstrating its ability to restore cholecystokinin-induced NTS neuronal cFos activation in food-deprived rats [20]. Thirty minutes later, rats were tested for passive avoidance retention. For this purpose, rats were placed individually into the light chamber of the light/dark box, the guillotine door was opened, and the latency of rats to fully enter the dark chamber was recorded, with a preset maximum latency of 900 s (15 min). During the retention test, the guillotine door remained open and no footshock was administered. Rats were allowed to freely explore both the light and dark chambers during the 900-s test, and total time spent within each chamber was recorded. After testing was complete, rats were returned to their home cages. Chow was not returned until 2 h prior to dark onset (i.e., approximately 6–8 h after testing).

To confirm that GRA treatment does not alter conditioned avoidance behavior in nonfasted rats, a separate cohort of adult male rats (n = 4) was trained in the passive avoidance task as described above. Two days later, rats were tested in an ad libitum-fed state, 30 min after receiving i.p. injection of GRA.

To examine the impact of GRA on general memory function, rats (N = 32) were tested for novel object recognition at least 5 days after completing passive avoidance testing. For this, rats underwent 3 days of acclimation (10 min/day) to a previously novel tub cage that had a thin layer of bedding on the floor, with the tub enclosed in an illuminated sound-attenuating chamber. On the 4th day, rats underwent object training 4–6 h after lights on, during which two similarly sized but geometrically distinct ceramic objects with rounded contours (i.e., a white pyramidal “Eiffel tower” and a blue spherical “rocket ship”) were placed at opposite ends of the tub cage. Rats were allowed to freely explore both objects for 15 min, while their behavior was recorded and later analyzed using behavior analysis software (ANY-maze, Stoelting Co.). During training, there was no significant difference between time spent with either object, defined as the rat being in contact with or actively examining the object.

Two hours before dark onset on the day before testing (i.e., 2 days after object training), all rats were transferred to clean cages with chow removed. On the subsequent day, fasted rats received either an i.p. injection of saline vehicle (∼0.3 mL) or the same volume of vehicle containing GRA (3.3 mg/kg BW). Thirty minutes later, rats were tested for novel object recognition. For this, the previously explored “Eiffel tower” object was replaced with a novel object (i.e., a gray cube-shaped “robot”). Rats were allowed to freely explore the novel and familiar objects for 15 min. The time each rat spent exploring each object was recorded, and the discrimination index was calculated as the difference between time spent with each object divided by the total time spent with either object:

In the discrimination index, a value of 0 indicates no discrimination between the novel and familiar objects. A positive value indicates more time spent with the novel object, whereas a negative value indicates more time spent with the familiar object. One rat from each injection group was excluded from analysis after one of the objects was knocked over during the novel object recognition test.

To examine the impact of GRA on general locomotion and anxiety-like behavior, the same rats (N = 32) were tested in the open field after overnight food deprivation. On the testing day (4–6 h after lights on), fasted rats received either an i.p. injection of saline vehicle (∼0.3 mL) or the same volume of vehicle containing GRA (3.3 mg/kg BW). Thirty minutes later, rats were placed into the center of a novel 95 × 95 cm open arena and then left to freely explore for 10 min. Behavior in the open field was recorded using a Logitech webcam and was analyzed using automated behavior analysis software (ANY-maze, Stoelting Co.). Locomotion was defined as the total distance traveled during the test. Anxiety-like behaviors were interpreted based on time spent in the center zone of the field (∼30 × 30 cm) and latency to first enter the center zone. Data from one cohort of rats (n = 4/injection group) were excluded due to lighting issues in the testing room.

To examine neural cFos activation in response to the conditioned context associated with footshock, a subset of rats used in the behavioral studies described above (N = 24) were retrained in the passive avoidance task in which they were placed into the light side of the light/dark box. When rats entered the dark chamber (average latency = 133.93 s), the guillotine door closed behind them. Shock-retrained rats immediately received a mild electric footshock. Another cohort of rats (N = 12) were placed into the light side of the light/dark box but did not receive the footshock upon entry into the dark chamber. Rats in both the shock-retrained (i.e., shocked) and nonshocked groups were kept in the dark chamber for 30 s before being returned to their home cages. Two days later, 2 h before dark onset, all rats were transferred to clean cages with chow removed to initiate overnight food deprivation. On the next day, 4–6 h after lights on, fasted rats in both groups received either an i.p. injection of 0.15 M NaCl (saline) vehicle (∼0.3 mL) or the same volume of vehicle containing GRA (3.3 mg/kg BW). Thirty minutes later, rats were placed directly into the dark chamber (an aversively conditioned stimulus in the shock-retrained group but not in the nonshocked group) for 10 min with the guillotine door shut; rats were then returned to their homecage. Sixty minutes later, rats were deeply anesthetized with pentobarbital sodium (39 mg/mL i.p., Fatal Plus Solution; Butler Schein) and transcardially perfused with saline (100 mL) followed by 4% paraformaldehyde (500 mL).

Fixed brains were removed from the skull, postfixed overnight at 4˚C, and then cryoprotected in 20% sucrose. Brains were blocked and sectioned coronally (35 μm) using a Leica freezing-stage sliding microscope. Tissue sections were collected in six serial sets and stored at −20°C in cryopreservant solution [34] until immunohistochemical processing. Primary and secondary antisera were diluted in 0.1 M phosphate buffer containing 0.3% Triton X-100 and 1% normal donkey serum. One set of tissue sections from each rat (with each set containing a complete rostrocaudal series of sections spaced by 210 μm) was incubated in a rabbit polyclonal antiserum against cFos (1:10,000; Cell Signaling, 2250; AB_2247211), followed by biotinylated donkey anti-rabbit IgG (1:500; Jackson ImmunoResearch). Sections were then treated with Elite Vectastain ABC reagents (Vector) and reacted with diaminobenzidine intensified with nickel sulfate to produce a blue-black nuclear cFos reaction product. To visualize cFos within hindbrain PrRP neurons using dual immunoperoxidase, one set of cFos-labeled tissue sections from each rat was subsequently incubated in a rabbit polyclonal antiserum against PrRP (1:10,000; Phoenix Pharmaceuticals, H-008-52), followed by biotinylated donkey anti-rabbit IgG (1:500; Jackson ImmunoResearch) and Elite Vectastain ABC reagents (Vector), and finally reacted with plain diaminobenzidine to produce a brown cytoplasmic reaction product.

To visualize cFos within all NA neurons using dual immunofluorescence, a second set of tissue from each rat was incubated in a cocktail of rabbit anti-cFos (1:1,000) and a mouse monoclonal antiserum against dopamine beta-hydroxylase (DbH; 1:5,000; Millipore, MAB308; AB_2245740), followed by a cocktail of Cy3-conjugated donkey anti-rabbit IgG (red) and Alexa Fluor 488-conjugated donkey anti-mouse IgG (green).

Sections through the cNTS (∼15.46–13.15 mm caudal to bregma) were imaged at 20x using a Keyence microscope (BZ-X700). Using separate sets of dual-labeled tissue sections from each rat, the total numbers of DbH+ and PrRP+ cNTS neurons were counted, and the percentage of each chemically identified neural population coexpressing cFos was determined. Individual neurons were considered cFos positive if their nucleus contained cFos immunolabeling (regardless of intensity) surrounded by visible cytoplasmic DbH or PrRP immunolabeling within the same focal plane.

Sections through the anterior vlBNST (∼0.26–0.40 mm caudal to bregma) were imaged at 10x. Images from two rats were excluded from further analysis due to poor tissue quality. Using ImageJ, a circular region of interest (ROI) of 0.10 mm² was centered on the area containing the highest density of DbH+ fibers and terminals corresponding to the fusiform subnucleus of the anterior vlBNST [6, 35, 36]. The number of cFos-positive nuclei within each 0.10 mm² ROI was determined bilaterally in 2–3 sections per brain and averaged across assessed ROIs in each animal.

Three sections through the PVN (∼1.5–1.9 mm caudal to bregma) were imaged at 10x bilaterally per brain. An ROI encompassing the DbH+ fibers and terminals was drawn using ImageJ, and the number of cFos-positive nuclei within each ROI was determined, divided by the area of the ROI, and averaged across assessed ROIs in each animal to produce the number of cFos-positive neurons per mm².

After finding that GRA treatment significantly impacted conditioned avoidance behavior and cFos activation (see Results), we sought to confirm our assumption that plasma acyl-ghrelin levels increase in rats exposed to single and repeated sessions of overnight food deprivation. For this, terminal blood samples were collected from a separate cohort of adult male rats that were either ad libitum fed with no fasting experience (n = 3), fasted once overnight before morning blood collection (n = 3), or fasted overnight on 3 separate occasions (with at least 2 days between each episode) with the 3rd fast occurring before morning blood collection (n = 4). Prior to blood collection, rats were injected with pentobarbital sodium (39 mg/mL i.p., Fatal Plus Solution; Butler Schein), and blood samples were collected within 5 min from the right atrium of the heart using a 3-mL syringe prerinsed with 1% EDTA. Blood samples were transferred to a K2E EDTA vacuette tube (BD Ref 367841), and aprotinin (50 µL/mL) was added to each sample. Samples were centrifuged at 3,500 rpm for 10 min at 4°C to separate the plasma. Hydrochloride was added to the supernatant (10 µL/100 mL) centrifuging again at 3,500 rpm for 5 min at 4°C. Plasma samples were stored at −80°C. Acyl-ghrelin was measured using an ELISA kit (BioVendor, R&D; RA394062400R) according to the manufacturer's instructions.

Data were analyzed using GraphPad Prism. Passive avoidance behaviors (latency to enter and total time spent in the dark chamber) during the retention test were analyzed using a Welch’s t test [37]. Time spent exploring novel versus familiar objects in the novel object recognition task was analyzed with a mixed 2 × 2 ANOVA with i.p. injection as a between-subjects independent variable and Object (novel, familiar) as a within-subjects independent variable. When ANOVA F values indicated significant main effects and/or interactions, post hoc comparisons were made using Šídák's multiple comparisons tests. Discrimination indices for novel object recognition by each i.p. injection group of rats were analyzed using Welch’s t test, and by individual one-subject t tests with a hypothetical value of 0 (i.e., no discrimination). Each open field behavioral measure (distance traveled, time in center, and latency to first enter the center) was analyzed separately with a Welch’s t test to compare i.p. injection groups. Cell count data (cFos activation of PrRP and DbH neurons, cFos activation within the anterior vlBNST and PVN) were analyzed using between-subjects 2 × 2 ANOVA with previous shock-pairing and i.p. treatment (i.e., GRA vs. vehicle) as independent variables. In addition, Pearson’s r correlations were run between the percentage of DbH+/PrRP+ cNTS neurons expressing cFos and cFos counts within vlBNST/PVN in the same rat. Plasma acyl-ghrelin data were analyzed using a one-way ANOVA with food deprivation history as the independent variable. An alpha level of 0.05 (p ≤ 0.05) was used as the criterion for considering group differences to be statistically significant. Estimation statistics were used to report effect sizes for both passive avoidance behavior and cell count data [38, 39].

To expand upon previous findings demonstrating that male rats display less conditioned passive avoidance behavior when tested in a food-deprived state [7], the present study sought to examine whether blocking ghrelin signaling in fasted rats would restore passive avoidance behavior. Indeed, fasted rats injected with GRA demonstrated more passive avoidance behavior than fasted rats injected with saline. Compared to saline-treated fasted rats, GRA-treated fasted rats took longer to enter the shock-paired dark chamber (t(26.1) = 1.728; p = 0.0479) and spent less time in the dark chamber (t(18.67) = 2.162; p = 0.0219) during the testing period (Fig. 1). Specifically, fasted GRA-treated rats took an average of 195 s (3.25 min) longer to enter the shock-paired dark chamber compared to saline-injected controls (95% CI: −11.4 s, 419 s) and spent an average of 178 s (2.97 min) less time in the shock-paired dark chamber during the 900-s retention test (95% CI: 47.2 s, 369 s). Thus, pharmacological blockade of ghrelin signaling enhanced conditioned passive avoidance in fasted rats.

To examine whether GRA treatment by itself might alter conditioned passive avoidance, a separate group of avoidance-trained, nonfasted rats (n = 4) were treated with GRA before the test. These rats displayed a long average latency to enter (895.75 s ± 4.25) and minimal time in the shock-paired dark chamber (1 s ± 1), quite similar to the behavior of fasted GRA-treated rats shown in Figure 1. The similarly robust conditioned avoidance behavior displayed by nonfasted and fasted GRA-treated rats also was similar to published data from our laboratory obtained from avoidance-trained, nonfasted male rats with no i.p. treatment prior to testing [7]. Thus, pharmacological blockade of ghrelin receptors by itself has no apparent effect on conditioned passive avoidance in nonfasted rats.

We considered the possibility that the GRA enhanced passive avoidance behavior in fasted rats due to pharmacological enhancement of memory-based performance. To test this possibility, we examined whether GRA improved performance in a novel object recognition task. Pretest treatment with GRA did not alter the behavior of fasted rats in this task (Fig. 2). Rats demonstrated that they successfully learned the task as indicated by a significant main effect of Object on exploration time (F(1, 28) = 15.59; p = 0.0005), but there was no main effect of i.p. treatment on exploration time and no interaction. Specifically, fasted rats spent significantly more time exploring the novel object compared to the familiar object in both the saline- (p = 0.0093) and GRA-injected (p = 0.0361) groups. The discrimination indices for both the saline- (t(14) = 3.575, p = 0.0030) and GRA-injected groups (t(14) = 2.294, p = 0.0378) were significantly higher than 0, evidence that both groups were similarly able to discriminate between the familiar and novel objects. Thus, pharmacological blockade of ghrelin receptors does not alter general memory in fasted rats as assessed by performance in the novel object recognition task.

We also considered the possibility that the effect of GRA to enhance passive avoidance behavior in fasted rats was due to a general effect on locomotor or anxiety-like behavior. To test this, we examined whether GRA treatment of fasted rats altered these behaviors during 10-min exposure to a novel open field. As shown in Figure 3a, fasted rats displayed similar locomotor behavior (i.e., total distance traveled) regardless of saline (mean = 17.9 m) or GRA treatment (mean = 17.3 m). There were also no significant effects of GRA on anxiety-like behaviors as measured by either time spent in the center of the open field (Fig. 3b) or latency to first enter the central zone (Fig. 3c). Thus, pharmacological blockade of ghrelin receptors does not alter general locomotor or anxiety-like behavior in fasted rats as assessed in a 10-min novel open field test.

We previously reported that PrRP+ A2 neurons are activated to express cFos when ad libitum-fed rats are reexposed to the (previously footshock-paired) dark chamber, whereas these neurons do not express cFos in fasted rats undergoing the same reexposure condition [7]. The present study examined whether GRA restores A2 neural cFos activation in fasted rats after dark chamber reexposure. PrRP+ neurons in fasted rats were responsive to the shock-paired dark chamber as indicated by a significant main effect of shock training (F(1, 30 = 14.02; p < 0.001). We also observed a trending main effect of GRA treatment prior to dark chamber reexposure (F(1, 30) = 3.14; p = 0.087) and a trending interaction between shock training and GRA treatment (F(1, 30) = 3.33; p = 0.078; Fig. 4a). Specifically, cFos expression in PrRP+ neurons was elevated in shock-retrained fasted rats treated with GRA compared to rats treated with saline before dark chamber reexposure (p = 0.010). Approximately 12.1% more PrRP+ neurons were activated to express cFos in shock-retrained, fasted rats pretreated with GRA before dark chamber reexposure compared to cFos activation of PrRP+ neurons in similarly shock-retrained, fasted rats treated with saline before dark chamber reexposure (95% CI: 2.39%, 21.9%). Conversely, there was no GRA-related effect on PrRP+ neural cFos activation in fasted rats that were not shocked during training. A significant interaction between shock training and GRA treatment (F(1, 32) = 7.29; p = 0.011) on cFos activation of DbH+ neurons also was present (Fig. 4d), with no main effects of either condition. Specifically, GRA treatment increased DbH+ cFos expression in shock-retrained rats that were reexposed to the dark chamber (p = 0.024), which activated 11.9% more DbH+ neurons in GRA-treated versus saline-treated rats (95% CI: 1.59%, 20.4%). Conversely, GRA had no effect on DbH+ neuronal cFos activation in nonshocked rats.

Having established that GRA treatment increased the ability of the footshock-conditioned context to activate cFos in DbH+/PrRP+ A2 neurons in fasted rats, we next examined cFos activation within two specific forebrain regions that receive dense input from hindbrain A2 neurons, the vlBNST and the PVN. There was a significant main effect of shock training (F(1, 30) = 57.08; p < 0.001) and a significant interaction between shock training and GRA treatment (F(1, 30) = 5.55; p = 0.025) on dark chamber-induced cFos in the vlBNST (Fig. 5a). Specifically, more vlBNST cFos activation was observed in shock retrained rats that were pretreated with GRA compared to those pretreated with saline (p = 0.031), with no effect of GRA treatment on vlBNST cFos in nonshocked rats. Approximately 12.9 more cFos+ nuclei per vlBNST ROI were present in rats pretreated with GRA compared to saline before reexposure to the shock-paired dark chamber (95% CI: 3.4, 25.3). A significant main effect of shock training on cFos activation within the PVN was also observed (F(1, 32) = 19.54; p < 0.001), with a marginally trending interaction between shock training and GRA treatment (F(1, 32) = 2.296; p = 0.14; Fig. 5d). Specifically, reexposure to the dark chamber in shock retrained rats strongly trended toward inducing more PVN cFos following GRA pretreatment compared to saline (p = 0.059), with no effect of GRA treatment in nonshocked rats. Approximately 74.6 more cFos+ nuclei/mm² were counted within the PVN in shock-retrained rats pretreated with GRA versus saline (95% CI: 8.64, 137).

We next examined potential relationships between cFos activation of A2 neurons and cFos activation in forebrain regions receiving input from A2 neurons (i.e., vlBNST and PVN). Across all shock retrained rats, there were strong positive within-subjects relationships between cFos activation of DbH+ cNTS neurons and cFos activation in the vlBNST (r(20) = 0.5093; p = 0.0155; Fig. 6a) and in the PVN (r(22) = 0.6494; p = 0.0006; Fig. 6b). Similar positive relationships occurred between cFos activation of PrRP+ cNTS neurons and cFos activation in the vlBNST (r(20) = 0.5497; p = 0.0080; Fig. 6c) and PVN (r(22) = 0.5488; p = 0.0055; Fig. 6d). No significant parallel relationships were observed in nonshocked rats. While such correlations do not prove causal relationships, they are consistent with the hypothesis that A2 neurons contribute to recruitment of neurons in downstream limbic and hypothalamic target regions.

To confirm that one or more rounds of overnight food deprivation elevates circulating ghrelin, we measured plasma levels of acyl-ghrelin in adult male rats exposed to a single overnight fast, and in rats exposed to a third round of overnight fasting. Acyl-ghrelin level in ad libitum-fed rats was measured to be 120.1 ± 8.0 pg/mL. Acyl-ghrelin levels rose to 192.3 ± 24.5 pg/mL following a single overnight fast, and to 399.9 ± 105.2 pg/mL following a third overnight fast. These data revealed a trending effect of food deprivation to increase plasma acyl-ghrelin (F(2, 7) = 3.851; p = 0.0745) and confirmed no decrement in the ability of overnight fasting to increase plasma ghrelin in rats exposed to three episodes of overnight fasting.

Results from the present study build on our previous findings that overnight food deprivation reduces conditioned passive avoidance behavior in adult male rats, with concurrent reduction of conditioned context-induced activation of cFos in PrRP+ NA neurons in the NTS and downstream target neurons in the vlBNST [7]. Here we demonstrate a role for ghrelin signaling in these fasting-induced behavioral and neural effects. Specifically, we show that systemic administration of GRA (presumed to counteract fasting-induced increases in ghrelin signaling) prior to the retention test enhances conditioned passive avoidance behavior (shown in Fig. 1) and neuronal cFos activation (Fig. 4, 5) in fasted rats. These novel findings support the view that reduced avoidance behavior and cFos activation in food-deprived rats is at least partly due to elevated ghrelin signaling.

We considered the possibility that reduced passive avoidance behavior in fasted rats with elevated ghrelin is due to a general reduction in memory retrieval, such that GRA may increase passive avoidance by improving memory of the shock-associated context. However, GRA treatment did not improve novel object recognition memory in food-deprived rats (shown in Fig. 2). A key distinction between mnemonic processes underlying passive avoidance versus novel object recognition is that the former but not the latter involves contextual memory [40, 41]. Previous research has shown that mice with a global ghrelin receptor knockout and rats with a vagal afferent-specific ghrelin receptor knockdown display impairment in hippocampal-dependent spatial and contextual novel object tasks [42, 43]. However, GRA effects on passive avoidance memory in the present study were in the opposite direction, such that GRA treatment improved avoidance memory in fasted rats. This discrepancy may be related to chronic rather than acute disruption of ghrelin signaling in the knockout/knockdown studies, which could interfere with other stages of memory formation. Alternatively, reduced performance in the “novel object in context” task by rodents with ghrelin receptor knockout/knockdown could reflect avoidance of the novel object. For example, Davis and colleagues [43] reported a negative shift in discrimination index in their vagal afferent ghrelin receptor knockdown model, such that knockdown rats spent less time exploring the novel object than would be expected if their behavior was due only to impaired memory of the familiar object.

If a fasting-induced elevation in ghrelin signaling is normally anxiolytic, then GRA treatment may reverse this effect to enhance general/innate anxiety-like behavior and avoidance displayed by nonfasted rats. Supporting this idea, previous work demonstrated that overnight food deprivation suppresses innate anxiety-like behavior in the elevated plus maze and light-enhanced acoustic startle tests [6], and in the open field test [44]. In the present study, GRA treatment did not alter innate anxiety-like behavior displayed by food-deprived rats in the open field test (shown in Fig. 3b, c), evidence that ghrelin signaling alone cannot fully explain the anxiolytic effect of food deprivation. A previous study reported that vagal afferent-specific ghrelin receptor knockdown does not alter anxiety-like behavior in the elevated zero maze [43], although rats in that experiment were not food deprived, so endogenous ghrelin presumably would have been relatively low in both knockdown and control rats.

Our findings demonstrate that GRA treatment enhanced the ability of reexposure to the shock-associated context to activate cFos expression in PrRP+ NA neurons in shock-retrained, fasted rats (shown in Fig. 4). These results are particularly interesting due to the suggested role of central PrRP signaling in stress responses and avoidance [45]. In nonfasted rodents, PrRP+ neurons in the cNTS are recruited by a variety of homeostatic and psychogenic stressors [6, 7, 10, 24, 25, 46], and stress-related neuroendocrine responses are modulated by PrRP signaling [24, 46‒49]. Further, administration of a PrRP analog has been reported to improve memory in a mouse model of Alzheimer’s disease [50]. These data and our current findings suggest a potential role for PrRP signaling in fear and memory and suggest a unique mechanism through which fasting-induced increases in ghrelin signaling may modulate the expression of conditioned avoidance behavior.

The ability of GRA treatment to enhance shock-paired context-induced cFos expression in PrRP+ A2 neurons is consistent with evidence that ghrelin suppresses the activity of vagal sensory afferents [29, 30] and suppresses activation of postsynaptic A2 neurons in the cNTS [31]. Our current results also are in line with our previous report that GRA enhances the ability of pharmacological vagal afferent stimulation to activate PrRP+ A2 neurons [20]. In that study [20] and in the current study, GRA treatment did not completely restore A2 neural cFos activation to levels observed in ad libitum-fed rats (i.e., see [7]). Thus, elevated ghrelin signaling during food deprivation contributes to but does not fully explain the A2 neural “silencing” effect of negative energy balance. Other contributing factors may include altered mechanosensory signals and/or hormonal signals [6, 51, 52].

Given the partial restoration of A2 neuronal cFos activation in shock-retrained, GRA-treated rats after reexposure to the shock-paired context, we examined concurrent cFos activation in DbH terminal-rich regions of the vlBNST and PVN; in rats, these DbH+ inputs arise primarily from caudal medullary A2 (and A1) NA neurons [53, 54]. Indeed, GRA increased activation of cFos within both regions in fasted rats that were reexposed to the shock-paired context (shown in Fig. 5). Importantly, GRA treatment prior to dark chamber reexposure did not influence cFos activation in PrRP+ NA neurons or within the vlBNST and PVN in rats that were not shocked during training (Fig. 4, 5). These data support the view that ghrelin signaling in fasted rats acts to limit aversive context-induced neural activation, without suppressing neural activation in response to a nonaversive context.

Based on significant positive relationships between cFos activation of PrRP+ and DbH+ cNTS neurons and activation of neurons within the vlBNST and PVN (shown in Fig. 6), we hypothesize that recruitment of A2 cNTS neurons in response to the shock-paired context promotes activation of downstream neurons within the vlBNST and PVN. To demonstrate a causal link between neural activation in these brain regions and their involvement in passive avoidance behavior, future research should specifically manipulate neural activity in these brain regions and circuits using pharmacological, optogenetic, chemogenetic, and/or lesioning strategies. For example, we previously used an NA-specific lesioning strategy to demonstrate that A2 NA inputs to the vlBNST and PVN are necessary for stress-induced innate avoidance and behavioral suppression [55‒57]. Interestingly, A2-derived NA signaling in the BNST has also been implicated in avoidance behavior conditioned by aversive opiate withdrawal [54, 58]. The current results provide new evidence, suggesting that A2 NA inputs to the vlBNST and/or PVN also participate in conditioned passive avoidance behavior.

The effect of GRA treatment on behavior and central cFos activation in the present study could reflect interference with either central or peripheral ghrelin signaling mechanisms. Ghrelin receptor mRNA is expressed by vagal sensory neurons in the nodose ganglia [29, 59] and by neurons within a variety of central brain regions [60‒64]. These central regions include the cNTS [60, 61] and the PVN [61], and regions considered to play a role in passive avoidance and fear memory [65], including the hippocampus [61, 62] and amygdala [63, 64]. Since systemic administration of the same GRA used in the present study was shown to block the effects of centrally administered ghrelin [66], central blockade of ghrelin receptors within these brain regions may have contributed to the behavioral and neural effects of GRA in our experiments. Additional work will be needed to differentiate between peripheral/vagally mediated versus centrally mediated ghrelin signaling in conditioned avoidance behavior. Future research also should explore sex differences in passive avoidance behavior and the role of ghrelin signaling (and other interoceptive feedback signals) in that behavior, given our previous report that food deprivation suppresses conditioned passive avoidance in male, but not female, rats [7]. In this regard, recent evidence indicates that female rats have significantly higher plasma levels of acyl-ghrelin compared to male rats under both fed and fasted conditions, and that females may be more sensitive to the anxiolytic effects of fasting-induced ghrelin signaling [28]. We reported that compared to male rats, female rats display markedly less conditioned passive avoidance behavior regardless of metabolic state at testing, consistent with other reports that female rodents show less passive and more active responses to threatening stimuli [7]. Thus, an active avoidance learning and testing approach is likely more suitable to explore potential effects of ghrelin signaling on behavioral (and neural) responses to conditioned aversive stimuli in females.

In summary, our results demonstrate a role for endogenous ghrelin signaling in the ability of negative energy balance to reduce conditioned avoidance behavior in male rats. We further demonstrate that GRA treatment enhances the ability of a footshock-conditioned context to activate PrRP+ A2 NA neurons, and to activate neurons in the downstream targets of these cNTS neurons (i.e., vlBNST and PVN) in fasted male rats. These findings support the view that interoceptive feedback from body to brain modulates learned avoidance behavior, and that a ghrelin-sensitive circuit contributes to this modulatory effect.

This study protocol was reviewed and approved by Florida State University Animal Care and Use Committee (#202000001) and was conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

The authors have no conflicts of interest to declare.

Research reported in this paper was funded by the National Institutes of Health grants F31MH119784 (C.M.E.) and R01MH59911 (L.R.).

Caitlyn M. Edwards and Linda Rinaman designed the experiments. Caitlyn M. Edwards, Inge Estefania Guerrero, Huiyuan Zheng, and Tyla Dolezel performed the experiments and analyzed the resulting data. Caitlyn M. Edwards, Inge Estefania Guerrero, and Linda Rinaman wrote the manuscript.

All data generated during this study are included in this article. Further enquiries can be directed to the corresponding author. A preprint version of this article is available on bioRxiv [67].