Neural Pathway for Gut Feelings: Vagal Interoceptive Feedback From the Gastrointestinal Tract Is a Critical Modulator of Anxiety-like Behavior

Abstract

Background: Anxiety disorders are associated with an altered perception of the body's internal state. Therefore, understanding the neuronal basis of interoception can foster novel anxiety therapies. In rodents, the feeding status bidirectionally modulates anxiety-like behavior but how the sensing of gastrointestinal state affects anxiety remains unclear.

Methods: We combined chemogenetics, neuropharmacology, and behavioral approaches in male and female rats to test whether vagal afferents terminating in the gastrointestinal tract mediate feeding-induced tuning of anxiety. Using saporin-based lesions and transcriptomics, we investigated the chronic impact of this gut-brain circuit on anxiety-like behavior.

Results: Both feeding and selective chemogenetic activation of gut-innervating vagal afferents increased anxiety-like behavior. Conversely, chemogenetic inhibition blocked the increase in anxiety-like behavior induced by feeding. Using a selective saporin-based lesion, we demonstrate that the loss of gut-innervating vagal afferent signaling chronically reduces anxiety-like behavior in male rats but not in female rats. We next identify a vagal circuit that connects the gut to the central nucleus of the amygdala, using anterograde transsynaptic tracing from the nodose ganglia. Lesion of this gut-brain vagal circuit modulated the central amygdala transcriptome in both sexes but selectively affected a network of GABA (gamma-aminobutyric acid)-related genes only in males, suggesting a potentiation of inhibitory control. Blocking GABAergic signaling in the central amygdala re-established normal anxiety levels in male rats.

Conclusions: Vagal sensory signals from the gastrointestinal tract are critical for baseline and feeding-induced tuning of anxiety via the central amygdala in rats. Our results suggest vagal gut-brain signaling as a target to normalize interoception in anxiety disorders.

Keywords: Amygdala; Anxiety; Gut-brains axis; Interoception; Sex differences; Vagus nerve.

Figure 1.

Refeeding increases anxiety-like behavior in male and female rats compared with fasting. (A) Distance (ANOVA, F_1,43 = 12.79, p = .0009) and (B) number of entries (ANOVA, F_1,43 = 11.38, p = .002) in the center of an open field decrease upon refeeding in male and female rats. (C) The total distance moved in an open field is not significantly changed by the feeding status (ANOVA, F_1,43 = 1.90, p = .18). (D) Representative traces of fasted or refed rats during a 30-minute open field test. (E) Distance (ANOVA, F_1,45 = 4.70, p = .035) and (F) time (ANOVA, F_1,45 = 6.50, p = .014) spent in the open arms of an elevated plus maze decrease upon refeeding in male and female rats. (G) The total distance moved in an elevated plus maze is not significantly changed by the feeding status (ANOVA, F_1,45 = 2.89, p = .10). (H) Average heatmaps of fasted or refed rats during a 5-minute elevated plus maze test. (I) Startle amplitude in response to acoustic stimuli is increased in refed compared with fasted rats (ANOVA 90 dB: F_1,45 = 0.26, p = .61; 105 dB: F_1,45 = 7.54, p = .009; 120 dB F_1,45 = 4.61, p = .037). (J) Time to start eating a novel food increases in refed compared with fasted rats (ANOVA, F_1,91 = 34.31, p < .0001). *p < .05; **p < .01; ***p < .001. ANOVA, analysis of variance; NS, not significant.

Figure 2.

Chemogenetic activation of vagal afferents projecting to the gastrointestinal tract increases anxiety-like behavior in male and female rats. (A) Dual viral injection strategy used to selectively activate vagal afferents terminating in the stomach and duodenum. (B) Representative picture of viral vector expression in the nodose ganglia [green: eGFP from retrograde AAV-Cre; red: mCherry from AAV-DIO-hm3D(Gq)]. (C) One-hour food intake decreases after CNO injection compared with Veh (ANOVA, F_1,34 = 14.55, p = 5.48 ×10⁻⁴). (D) Distance (ANOVA, F_1,32 = 4.73, p = .037) but not (E) number of entries (ANOVA, F_1,32 = 1.25, p = .271) in the center of an open field decreases after CNO administration in male and female rats. (F) The total distance moved in an open field is not significantly changed by CNO injection (ANOVA, F_1,32 = 0.67, p = .42). (G) Representative traces of Veh- or CNO-injected rats during a 30-minute open field test. (H) Distance (ANOVA, F_1,36 = 6.57, p = .015) spent in the open arms of an elevated plus maze decreases after CNO injection compared with vehicle. (I) Time (ANOVA, F_1,36 = 2.93, p = .096) spent in the open arms of an elevated plus maze is not significantly decreased by CNO injection. (J) The total distance moved in an elevated plus maze is not significantly changed by CNO injection (ANOVA, F_1,36 = 1.96, p = .17). (K) Average heatmaps of Veh- or CNO-injected rats during a 5-minute elevated plus maze test. (L) Startle amplitude in response to acoustic stimuli is not significantly changed by CNO injection (ANOVA 90 dB: F_1,36 = 2.55, p = .12; 105 dB: F_1,36 = 2.09, p = .16; 120 dB F_1,36 = 0.11, p = .74). (M) Time to start eating a novel food increases in CNO-injected compared with fasted rats (ANOVA, F_1,73 = 15.40, p = .0002). #p < .1; *p < .05; ***p < .001. ANOVA, analysis of variance; CNO, clozapine N-oxide; eGFP, enhanced green fluorescent protein; NS, not significant; Veh, vehicle.

Figure 3.

Chemogenetic inhibition of vagal afferents projecting to the gastrointestinal tract blocks the anxiogenic effect of refeeding. (A) Dual viral injection strategy used to selectively inhibit vagal afferents terminating in the stomach and duodenum. (B) Representative picture of viral vector expression in the nodose ganglia [green: eGFP from retrograde AAV-Cre; red: mCherry from AAV-DIO-hm4D(Gi)]. (C) One-hour food intake increases after CNO injection compared with Veh (ANOVA, F_1,35 = 9.49, p = .004). (D) Distance in the center of an open field increases after CNO administration in refed rats (ANOVA, F_1,35 = 5.94, p = .020). (E) The number of entries in the center of an open field shows a trend toward increase after CNO administration in refed rats (ANOVA, F_1,35 = 3.19, p = .083). (F) The effects of CNO on the total distance moved in an open field test depend on sex in refed rats (ANOVA sex × CNO interaction, F_1,34 = 5.72, p = .022, Holm-adjusted comparisons, females p = .57, males p = .0012). (G) Representative traces of Veh- or CNO-injected rats (refed) during a 30-minute open field test. (H) CNO increases the distance traveled in the open arms of an elevated plus maze in refed rats (ANOVA, F_1,37 = 10.09, p = .003). (I) CNO increases the time spent in the open arms of an elevated plus maze in refed rats (ANOVA, F_1,37 = 11.97, p = .001). (J) The effects of CNO on the total distance moved in an elevated plus maze depend on sex in refed rats (ANOVA sex × CNO interaction, F_1,37 = 4.88, p = .047, Holm-adjusted comparisons, females p = .71, males p = .012). (K) Average heatmaps of Veh- or CNO-injected rats (refed) during a 5-minute elevated plus maze test. (L) Startle amplitude in response to acoustic stimuli is increased after CNO injection in refed rats (ANOVA 90 dB: F_1,36 = 2.68, p = .11; 105 dB: F_1,36 = 2.82, p = .10; 120 dB F_1,36 = 4.14, p = .049). (M) CNO inhibition decreases the time to start eating a novel food (ANOVA, F_1,72 = 18.76, p = .00047). (N) Distance in the center of an open field is not modulated by CNO administration in fasted rats (ANOVA, F_1,32 = 1.78, p = .19). (O) The number of entries in the center is not modulated by CNO administration in fasted rats (ANOVA, F_1,32 = 1.71, p = .20). (P) The total distance moved in an open field test is not modulated by CNO administration in fasted rats (ANOVA, F_1,32 = 2.12, p = .16). (Q) Representative traces of Veh- or CNO-injected rats (fasted) during a 30-minute open field test. (R) CNO decreases the distance traveled in the open arms of an elevated plus maze in fasted rats (ANOVA, F_1,36 = 6.52, p = .015). (S) CNO decreases the time spent in the open arms of an elevated plus maze in fasted rats (ANOVA, F_1,37 = 5.67, p = .023). (T) The effects of CNO on the total distance moved in an elevated plus maze depend on sex in refed rats (ANOVA sex × CNO interaction, F_1,36 = 5.71, p = .022, Holm-adjusted comparisons, females p = .48, males p = .010). (U) Average heatmaps of Veh- or CNO-injected rats (fasted) during a 5-minute elevated plus maze test. (V) Startle amplitude in response to acoustic stimuli is not modulated by CNO injection in fasted rats (ANOVA 90 dB: F_1,37 = 0.001, p = .97; 105 dB: F_1,37 = 0.40, p = .53; 120 dB F_1,37 = 1.28, p = .27). (W) CNO inhibition does not modulate the time to start eating a novel food in fasted rats (ANOVA, F_1,76 = 0.83, p = .37). #p < .1; *p < .05; **p < .01; ***p < .001. ANOVA, analysis of variance; CNO, clozapine N-oxide; eGFP, enhanced green fluorescent protein; NS, not significant; Veh, vehicle.

Figure 4.

Chronic lesion of gastrointestinal vagal afferents decreases anxiety-like behavior in a sex-dependent manner. (A) Injection strategy used to selectively lesion gastrointestinal vagal afferents. (B) CCK-SAP reduces the nodose ganglia expression of Cckar compared with SAP (ANOVA, F_1,34 = 42.17, p = 1.98 × 10⁻⁷). (C) The anorexigenic effect of intraperitoneal CCK is abolished in CCK-SAP rats (ANOVA, F_1,39 = 119.14, p = 2.0 × 10⁻¹³). (D) The effect of CCK-SAP on the distance in the center of an open field depends on sex (ANOVA, sex × group interaction, F_1,37 = 4.63, p = .038). CCK-SAP increases the distance spent in the center of an open field in males (Holm-adjusted comparisons p = .028) but not in females (p = .36). (E) The effect of CCK-SAP on the number of entries in the center of an open field depends on sex (ANOVA, sex × group interaction, F_1,37 = 4.52, p = .040). CCK-SAP increases the number of entries in the center of an open field in males (Holm-adjusted comparisons p = .010) but not in females (p = .50). (F) The effect of CCK-SAP on total distance moved in an open field depends on sex (ANOVA, sex × group interaction, F_1,37 = 6.99, p = .012). CCK-SAP increases the total distance moved in males (Holm-adjusted comparisons p = .0013) but not in females (p = .94). (G) Representative traces of SAP and CCK-SAP rats during a 30-minute open field test. (H) Distance (ANOVA, F_1,40 = 2.20, p = .15) and (I) time (ANOVA, F_1,40 = 1.03, p = .32) spent in the open arms of an elevated plus maze do not differ between SAP and CCK-SAP rats. (J) The total distance moved in an elevated plus maze is not significantly affected by CCK-SAP (ANOVA, F_1,40 = 1.50, p = .23). (K) Average heatmaps of SAP and CCK-SAP rats during a 5-minute elevated plus maze test. (L) Startle amplitude in response to acoustic stimuli is modulated by CCK-SAP in a sex-dependent manner (ANOVA 90 dB: group effect F_1,38 = 0.17, p = .69; 105 dB: sex × group interaction F_1,39 = 8.21, p = .007; Holm-adjusted comparisons males p = .021, females p = .21; 120 dB: sex × group interaction F_1,39 = 9.12, p = .004; Holm-adjusted comparisons males p = .10, females p = .91). (M) CCK-SAP decreases the time to start eating a novel food compared with SAP (ANOVA, F_1,81 = 10.6, p = .002). #p < .1; *p < .05; **p < .01; ***p < .001. ANOVA, analysis of variance; CCK, cholecystokinin; CeA, central nucleus of the amygdala; HSV, herpes simplex virus; NS, not significant; SAP, saporin.

Figure 5.

GABAergic signaling in the central amygdala is modulated by vagal afferents and underlies the anxiety-reducing effects of CCK-SAP in male rats. (A) Strategy used for anterograde polysynaptic tracing from the nodose ganglia with HSV-129. (B) Representative pictures of HSV-129 labeling in the amygdala 5 days after HSV-129 injection into the nodose ganglia (CeL, CeM, CeC, BLA, BLP). (C) Strategy used for RNA sequencing of CeA micropunches after CCK-SAP or SAP in male rats. (D) PC analysis plot of CeA gene expression profiles in CCK-SAP and control rats. (E) Heatmaps of gene expression (z score) for significant DEGs in the CeA of CCK-SAP compared with control rats (adjusted p < .05). (F) Inference scores of DEGs in the CeA associated with anxiety disorders according to the Comparative Toxicogenomics Database. (G) Significantly enriched biological processes among DEGs in the CeA (adjusted p < .05). (H) DEGs associated with the 3 significantly enriched biological processes in the CeA. (I) Strategy used for the bilateral administration of the GABA_A receptor antagonist bicuculline in CCK-SAP or SAP male rats. (J) Bicuculline injection into the CeA increases startle amplitude in response to a 105-dB acoustic stimulus in CCK-SAP male rats but not in SAP male rats (within-subject ANOVA CCK-SAP × injection interaction, F_1,13 = 4.90, p = .045; Holm-adjusted comparison in SAP rats p = .42, in CCK-SAP rats p = .041). Bicuculline does not significantly modulate startle amplitude in response to 90 dB (within-subject ANOVA; injection effect F_1,13 = 0.050, p = .83; CCK-SAP × injection, F_1,13 = 0.947, p = .35) and 120 dB (within-subject ANOVA; injection effect F_1,13 = 1.050, p = .32; CCK-SAP × injection, F_1,13 = 0.0001, p = .99) stimuli. *p < .05. ANOVA, analysis of variance; Bic, bicuculline; BLA, basolateral amygdaloid nucleus, anterior part; BLP, basolateral amygdaloid nucleus, posterior part; CCK, cholecystokinin; CeA, central nucleus of the amygdala; CeC, central amygdaloid nucleus, capsular part; CeL, central amygdaloid nucleus, lateral division; CeM, central amygdaloid nucleus, medial division; DEGs, differentially expressed genes; GABA, gamma-aminobutyric acid; NS, not significant; PC, principal component; SAP, saporin; Veh, vehicle.