A top-down insular cortex circuit crucial for non-nociceptive fear learning

Abstract

Understanding how threats drive fear memory formation is crucial to understanding how organisms adapt to environments and treat threat-related disorders such as PTSD. While traditional Pavlovian conditioning studies have provided valuable insights, the exclusive reliance on electric shock as a threat stimulus has limited our understanding of diverse threats. To address this, we developed a conditioning paradigm using a looming visual stimulus as an unconditioned stimulus (US) in mice and identified a distinct neural circuit for visual threat conditioning. Parabrachial CGRP neurons were necessary for both conditioning and memory retrieval. Upstream neurons in the posterior insular cortex (pIC) responded to looming stimuli, and their projections to the parabrachial nucleus (PBN) induced aversive states and drove conditioning. However, this pIC-to-PBN pathway was not required for foot-shock conditioning. These findings reveal how non-nociceptive visual stimuli can drive aversive states and fear memory formation, expanding our understanding of aversive US processing beyond traditional models.

Fig. 1.. Looming visual stimuli drive formation of auditory and contextual fear memory in mice.

(A) Schematic of behavioral tests for exposure of looming visual stimuli (a 10 times repeat of a single looming for 8 s). (B) Changes in movement speed of mice in response to looming visual stimuli. The blue shaded bars indicate a period during which mice displayed freezing behavior. (C) Ethogram showing defensive behavioral responses to the presentation of looming visual stimuli. Mice displayed robust freezing, either immediately or following an escape response (N = 10, 8 mice showed immediate freezing and other 2 mice showed escape-freezing). (D) Schematic of a threat conditioning protocol with looming visual stimuli. (E) Auditory fear memory was formed only in the paired group (paired group, N = 11; Unpaired group, N = 10; Repeated measures (RM) two-way ANOVA, time × group, F_1,19 = 17.2, P = 0.0005). (F) Both paired and unpaired groups show elevated levels of contextual freezing during re-exposure to the conditioning chamber on day3 (RM two-way ANOVA, time, F_1,19 = 36.38, P < 0.0001). (G) Single and double pairings of tone CS and looming US were not effective in driving the formation of auditory fear memory (single-paired group, N = 7; double-paired group, N = 8; five-times paired group, N = 8; RM two-way ANOVA, group, F_2,20 = 4.500, P = 0.0243; right). All data are means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. ns, not significant.

Fig. 2.. CGRP^PBN neurons selectively respond to looming US and acquired tone CS, but not to neutral stimuli.

(A) Schematic of fiber photometry recordings in CGRP^PBN neurons during visual stimuli. (B) Representative image of GCaMP6m expression in the PBN; the dotted rectangle marks the optic fiber location. (C) Freezing levels during flickering and looming stimuli (N = 13 mice; RM one-way ANOVA, Time, F_12,36 = 2.786, P = 0.0087). (D) z-scored calcium activity (means ± SEM) during visual stimuli. (E and F) Average area under the curve (AUC) comparison of calcium activity during visual stimuli [(E); flickering; N = 13 mice; two-tailed paired t test, t₁₂ = 0.6077, P = 0.5547; looming; t₁₂ = 2.221, P = 0.0464] and during 1 min [(F); flickering; t₁₂ = 0.6251, P = 0.5436; looming; t₁₂ = 3.482, P = 0.0045]. (G) AUC comparison between flickering and looming stimuli (t₁₂ = 3.541, P = 0.0041). (H) Schematic of fiber photometry recordings of CGRP^PBN neurons during visual threat conditioning. (I to J) z-scored calcium activity of CGRP^PBN neurons in the unpaired group during the first to fifth exposures to looming US [(J); N = 55 trials; two-tailed Wilcoxon matched-pairs signed rank test, W = −1042, P < 0.0001]. (K and L) z-scored calcium activity of CGRP^PBN neurons in the paired group during the first to fifth exposures to looming US [(L); N = 65 trials; two-tailed Wilcoxon matched-pairs signed-rank test, W = −1261, P < 0.0001]. (M) z-scored calcium activity of CGRP^PBN neurons during the tone retrieval test. (N) Average AUC of calcium activity in the unpaired (left; two-tailed paired t test, t₁₀ = 0.3602, P = 0.7262) and paired groups (right; t₁₂ = 3.714, P = 0.0030) during the tone retrieval test. (O) Comparison of AUC between the two groups during the tone retrieval test (two-tailed unpaired t test, t₂₂ = 3.431, P = 0.0024). All data are means ± SEM. *P < 0.05, **P < 0.01, and ****P < 0.0001.

Fig. 3.. The inhibition of CGRP^PBN neurons impairs fear memory formation and retrieval but not innate defensive behavior to looming stimuli.

(A) Schematic of TetTox-mediated inhibition of CGRP^PBN neurons during visual threat conditioning. (B) Representative image showing the expression of dTomato in the PBN, especially in the external lateral subdivision of PBN (PBel). Scale bars, 200 μm. SCP, superior cerebellar peduncle. (C) Average of movement (left; TetTox, N = 10; tdTomato, N = 13; two-tailed unpaired t test, t₂₁ = 1.518, P = 0.1439) and center zone time (right, two-tailed Mann-Whitney test, U = 46, P = 0.2569) were comparable between groups. (D) Freezing levels during the first CS-US pairing on day 1 (RM two-way ANOVA, time × group, F_2,42 = 5.334, P = 0.0086). (E and F) TetTox inhibition of CGRP^PBN neurons impairs auditory [(E); RM two-way ANOVA, time × group, F_1,21 = 7.251, P = 0.0136] and contextual fear memory [(F); RM two-way ANOVA, time × group F_1,21 = 7.798, P = 0.0109]. (G) Schematic of optogenetic inhibition of CGRP^PBN neurons during visual threat conditioning. (H) Representative image of eNpHR3.0-EYFP expression in CGRP^PBN neurons; the dotted rectangle marks the optic fiber tip location. Scale bar, 200 μm. (I) Freezing during the first CS-US pairing (eNpHR3.0, N = 7; EYFP, N = 8; RM two-way ANOVA, time, F_2,26 = 7.167, P = 0.0033). (J) Inhibition of CGRP^PBN neurons during visual US conditioning impaired the formation of auditory fear memory (RM two-way ANOVA, time × group, F_1,13 = 9.965, P = 0.0076). (K) Schematic of optogenetic inhibition of CGRP^PBN neurons during auditory memory retrieval. (L) Inhibition of CGRP^PBN neurons impaired retrieval of auditory fear memory (eNpHR3.0, N = 10; EYFP, N = 12; RM two-way ANOVA, time × group, F_1,20 = 18.16, P = 0.0004). All data are means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 4.. Identification of CGRP^PBN-projecting neurons activated by looming visual stimuli.

(A) Schematic of a monosynaptic retrograde tracing of CGRP^PBN neurons and c-Fos imaging in response to looming stimuli. Bilateral injections of AAV2/1-hsyn-DIO-TVA-HA-N2cG into the PBN of Calca^Cre mice were followed by bilateral injections of SADB19-GFP rabies virus into the same site. (B) Looming visual stimuli induced a robust freezing response in mice, whereas flickering stimuli did not (flickering, N = 5, looming, N = 7; RM two-way ANOVA, time × group, F_1,10 = 9.893, P = 0.0104). (C) Representative confocal microscopic images showing CGRP^PBN-projecting GFP⁺ neurons (green) and c-Fos⁺ neurons (red) in the pIC, CeA, PAG, and SC. A white arrowhead indicates co-localized neurons. Images of starter cells in the PBN are presented in the fig. S4. Scale bars, 100 μm. LPAG, lateral PAG; SCiw, intermediate white layer of the SC; SCig, intermediate gray layer of the SC. (D to G) The ratio of double-positive neurons (GFP⁺ and c-Fos⁺) within GFP⁺ or GFP⁻ cell populations was compared between the groups (flickering versus looming) in each analyzed region: pIC (D) (flickering, N = 5; looming, N = 7; left; two-tailed unpaired t test, t₁₀ = 4.952, P = 0.0006; right; t₁₀ = 1.145, P = 0.2787), CeA (E) (left; t₁₀ = 1.404, P = 0.1906; right; t₁₀ = 0.4580, P = 0.6567), PAG (F) (left; t₁₀ = 0.2072, P = 0.8400; right; t₁₀ = 0.2809, P = 0.7845), and SC (G) (left; t₁₀ = 2.593, P = 0.0268; right; t₁₀ = 0.1301, P = 0.8901). All data are means ± SEM. *P < 0.05, ***P < 0.001, and ****P < 0.0001.

Fig. 5.. Fiber photometry recordings of pIC^→CGRP-PBN neuronal activity in response to visual and foot-shock stimuli.

(A) Schematic of fiber photometry recordings of pIC^→CGRP-PBN neuronal responses to looming (or flickering) and foot-shock stimuli. (B) Representative image of GCaMP6s (green) and FlpO (red) expression in the pIC. All GCaMP6s⁺ cells were FlpO⁺. The dotted rectangle marks the optic fiber location. Scale bar, 200 μm. (C) Representative images of starter cells in the PBN. Yellow arrowheads indicate FlpO-mCherry⁺ and HA⁺ starter cells. Scale bars, 100 μm. (D) z-scored calcium activity (means ± SEM) in response to looming and foot-shock stimuli (the blue shaded area: 8-s looming stimuli; the red shaded area: 2-s foot shock). (E) Average during looming and foot-shock (N = 9 mice; left; baseline = −60 to 0 s, looming = 0 to 60 s after the onset of visual stimuli; Two-tailed paired t test, t₈ = 3.291, P = 0.011; right; shock = 0 to 20 s after the onset of shock stimuli; two-tailed Wilcoxon matched pairs signed-rank test, W = 45, P = 0.0039). (F) Time to reach half of the peak z score after the peak (two-tailed Mann-Whitney test, U = 11, P = 0.0072). (G) z-scored calcium activity (means ± SEM) during flickering and foot shock (the gray shaded area: 8-s flickering stimuli; the red shaded area: 2-s foot shock). (H) Average AUC during flickering and foot shock (N = 8 mice; left; baseline = −60 to 0 s, flickering = 0–60 s after the onset of visual stimuli; Two-tailed Wilcoxon matched pairs signed-rank test, W = −6, P = 0.7422; right; shock = 0–20 s after the onset of shock stimuli; two-tailed paired t test, t₇ = 2.791, P = 0.0269). (I) Comparison of AUC between flickering and looming (two-tailed unpaired t test, t₁₅ = 2.798, P = 0.0135). All data are presented as means ± SEM. *P < 0.05 and **P < 0.01.

Fig. 6.. pIC^→PBN neurons are necessary for activation of CGRP^PBN neurons to looming but not foot-shock stimuli.

(A) Schematic of photometry recording of CGRP^PBN neurons with hM4Di inhibition of pIC^→PBN neurons. (B) Representative images showing expression of hM4Di-mCherry in the pIC^→PBN neurons (left) and GCaMP6m in CGRP^PBN neurons (right). The dotted rectangle indicates the location of optic fiber. Scale bars, 200 μm. (C) z-scored calcium activities of CGRP^PBN neurons to visual stimuli (blue shaded area: 8-s looming stimuli). (D and E) Average calcium activity during visual stimuli in the mCherry group (D) (baseline = −5 to 0 s, looming = 0 to 8 s; N = 12 trials; two trials per mouse, six mice; one sample t test, t₁₁ = 2.659, P = 0.0222) and hM4Di group (E) (N = 14 trials; two trials per mouse, seven mice; t₁₃ = 3.266, P = 0.0061). (F and G) Comparison of average changes in calcium activity evoked by looming stimuli between two groups during stimuli (F) (two-tailed unpaired t test, t₂₄ = 4.104, P = 0.0004) and for 1 min (G) (two-tailed unpaired t test, t₂₄ = 4.203, P = 0.0003). (H) z-scored calcium activities of CGRP^PBN neurons to foot-shock stimuli (red shaded area: 2-s foot shock). (I and J) Average calcium activity during foot-shock in the mCherry group (I) (baseline = −5 to 0 s, shock = 0 to 2 s; N = 12 trials; one sample t test, t₁₁ = 4.887, P = 0.0005) and hM4Di group (J) (N = 14 trials; one sample Wilcoxon test, W = 89, P = 0.0031). (K and L) Comparison of average changes in calcium activity evoked by foot-shock stimuli between two groups during stimuli (K) (two-tailed Mann-Whitney test, U = 63, P = 0.2972) and for 1 min (L) (two-tailed Mann-Whitney test, U = 69, P = 0.4624). All data are means ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. 7.. pIC projections to the PBN are required for threat conditioning with looming but not with foot-shock stimuli.

(A) Schematic of optogenetic inhibition of pIC-PBN circuit during visual threat conditioning. (B) Representative images of eNpHR3.0-EYFP expression in pIC neurons (left) and their axons in the PBN (right). Dotted rectangle indicates optic fiber tip. Scale bars, 200 μm. (C) Freezing levels during first CS-US pairing (EYFP, N = 12; eNpHR3.0, N = 12; RM two-way ANOVA, time, F_2,44 = 125.6, P < 0.0001). (D) Inhibition of pIC-PBN projections during conditioning impaired auditory fear memory formation (RM two-way ANOVA, time × group, F_1,22 = 4.550, P = 0.0443). (E and F) Schematic and representative images of pIC-PBN inhibition during foot-shock conditioning. The dotted rectangle marks the optic fiber tip. Scale bars, 200 μm. (G) Average of movement during the first pairing of tone CS and foot-shock US on day 1 (EYFP, N = 8; eNpHR3.0, N = 7; RM two-way ANOVA, time, F_2,26 = 71.34, P < 0.0001). (H) Inhibition of pIC-PBN projections during foot-shock conditioning did not affect the formation of fear memory (RM two-way ANOVA, time, F_1,13 = 30.91, P < 0.0001). (I) Optogenetic inhibition of pIC-PBN projections during US exposure (2 s before to 2 s after US; total 8 s) did not impair auditory fear memory formation (EYFP, N = 7; eNpHR3.0, N = 8; RM two-way ANOVA, time, F_1,13 = 38.2, P < 0.0001). (J) Optogenetic inhibition of pIC axon terminals in the PBN during ITI impaired auditory fear memory formation (EYFP, N = 7; eNpHR3.0, N = 8; RM two-way ANOVA, time × group, F_1,13 = 12.76, P = 0.0034). All data are means ± SEM. **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 8.. Optogenetic stimulation of pIC^→CGRP-PBN projections induces anxiogenic behaviors.

(A) Schematic of viral injections and behavior tests. (B) Representative images of ChR2 (green) and/or retrogradely labeled FlpO (red) expression in the pIC (left) and PBN (right). The dotted rectangle marks the optic fiber location. (C) Representative heatmaps of time spent in RTPA test chambers during photostimulation. (D to F) ChR2 group avoided the photostimulation-paired chamber. (D) Time spent in the photostimulated chamber (left, EYFP; N = 9; two-tailed paired t test, t₈ = 0.6490, P = 0.5345; right, ChR2; N = 8, t₇ = 6.837, P = 0.0002). (E) Preference ratio for the photostimulated chamber (two-tailed Mann-Whitney test, U = 6, P = 0.0025). (F) Difference score for each chamber (RM two-way ANOVA, time × group, F_2,30 = 10.77, P = 0.0003). (G) Representative movement trajectory during the OFT. (H to L) ChR2 photostimulation increased anxiety-like behaviors. (H) Photostimulation did not induce a freezing (EYFP, N = 9; ChR2, N = 9; RM two-way ANOVA, time × group, F_9,144 = 0.8810, P = 0.5439). Averages of freezing levels (I) (Two-tailed unpaired t test, t₁₆ = 1.379, P = 0.1870), time spent in the center zone (J) (t₁₆ = 2.627, P = 0.0183), number of center zone crossings (K) (t₁₆ = 2.167, P = 0.0456), and moving distance (L) (t₁₆ = 1.131, P = 0.2748). (M) Representative movement trajectory during EPM. (N to R) Time spent in open arms (N) (EYFP, N = 9; ChR2, N = 9; t₁₆ = 1.406, P = 0.0894), time spent in closed arms (O) (t₁₆ = 2.141, P = 0.0480), open arm entries (P) (U = 18.5, P = 0.0495), closed arm entries (Q) (t₁₆ = 1.012, P = 0.3267), and moving distance (R) (t₁₆ = 1.129, P = 0.2757). Two-tailed Mann-Whitney test in (P) and two-tailed unpaired t test in [(N), (O), (Q), and (R)] were used. All data are means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 9.. Optogenetic stimulation of pIC^→CGRP-PBN projections is sufficient to drive formation of fear memory.

(A) Schematic of a threat conditioning paradigm with optogenetic stimulation of pIC^→CGRP-PBN projections as s a US. During conditioning, mice were exposed to a 30-s tone, during which the last 10-s tone presentation was overlapped with optic stimulation. This CS-US pairing was repeated five times at a 120-s interval. (B) Representative images showing the neurons expressing ChR2 (green) and/or retrograde labeled FlpO (Red) in the pIC (left) and ChR2-expressing neurons projections (green) in the PBN (right). Neurons expressing FlpO were also detected in the PBN (right). The dotted rectangle indicates the location of the optic fiber. Scale bars, 100 μm. (C) ChR2-expressing mice exhibited increased freezing levels to the conditioned tone during the fear memory retrieval test (EYFP, N = 9; ChR2, N = 8; RM two-way ANOVA, time × group, F_1,15 = 15.22, P = 0.0014). All data are means ± SEM. **P < 0.01 and ****P < 0.0001.