Dissociable contributions of the amygdala and ventral hippocampus to stress-induced changes in defensive behavior

Abstract

Stress can have profound consequences on mental health. While much is known about the neural circuits supporting associative memories of stressful events, our understanding of the circuits underlying the non-associative impacts of stress, such as heightened stress sensitivity and anxiety-related behavior, is limited. Here, we demonstrate that the ventral hippocampus (vHC) and basolateral amygdala (BLA) support distinct non-associative behavioral changes following stress. Inhibiting stress-induced protein synthesis in the BLA blocked subsequent increases in stress sensitivity but not anxiety-related behaviors. Conversely, inhibiting stress-induced protein synthesis in the vHC blocked subsequent increases in anxiety-related behavior but not stress sensitivity. Inhibiting neuronal activity in the BLA and vHC during the assessment of stress sensitivity or anxiety-related behavior recapitulated these structures' dissociable contributions to defensive behavior. Lastly, blocking the associative memory of a stressor had no impact on stress-induced changes in anxiety-related behavior. These findings highlight that multiple memory systems support the long-lasting effects of stress.

Keywords: Amygdala; CP: Neuroscience; anxiety; fear; hippocampus; non-associative learning; plasticity; sensitization; stress; stress-enhanced fear learning; trauma.

Figure 1.. Acute stress produces multiple lasting changes in defensive behavior

(A) Animals were exposed to an environment in which they received 10 footshocks (trauma [T]) or were placed in the same environment and received no footshocks (no trauma [NT]). A week later, they were tested for associative fear of the trauma environment, anxiety-related behavior in the light-dark test, and their response to a novel stressor in a new environment to assess stress sensitization. (B) Trauma-exposed animals displayed high levels of post-shock freezing during the trauma. (C) Trauma-exposed animals displayed strong associative fear of the trauma environment. (D) Trauma-exposed animals displayed increased anxiety-related behavior in the light-dark test. (E) Trauma-exposed animals did not differ in baseline levels of freezing when initially placed in the environment of the novel stressor (left) but displayed increased fear of the novel stressor environment when returned to this environment the next day, evidence of stress sensitization (right). (F) Correlation between trauma recall and anxiety-related behavior in light-dark test. (G) Correlation between trauma recall and novel stressor response. (H) Correlation between anxiety-related behavior in light-dark test and novel stressor response. For (B)–(E), NT = 25 (13 female) and T = 31 (16 female) mice. For (F)–(H), T = 40 mice. *p < 0.05, **p < 0.01, and ***p < 0.001. Error bars reflect standard error of the mean.

Figure 2.. Stress-induced protein synthesis in the BLA and vHC supports distinct changes in defensive behavior

(A) After trauma (T) or no trauma (NT), animals were administered 3 injections of anisomycin (ani) or vehicle (veh). A week later, they were tested for associative fear of the trauma environment, anxiety-related behavior in the light-dark test, and their response to a novel stressor in a new environment. (B) No differences were observed between trauma-exposed animals during the initial trauma. (C) Anisomycin given systemically immediately after trauma, but not 48 h later, reduced associative fear of the trauma environment. (D) Anisomycin given systemically immediately after trauma, but not 48 h later, reduced anxiety-related behavior in the light-dark test. (E) Anisomycin given systemically immediately after trauma, but not 48 h later, reduced stress sensitization. (F) No differences were observed between trauma-exposed animals during the initial trauma. (G) Anisomycin in the BLA and vHC reduced associative fear of the trauma environment. (H) Anisomycin in the vHC, but not the BLA, reduced anxiety-related behavior in the light-dark test. (I) Anisomycin in the BLA, but not the vHC, reduced stress sensitization. (J) Example placement of cannula injectors in the BLA and vHC for intracranial infusions. For (B)–(E), anisomycin/vehicle was administered systemically either immediately (0h) or 48 h (48h) after trauma. For (F)–(J), anisomycin/vehicle was administered directly into either the BLA or vHC immediately after trauma. For systemic injections (B–E), NT: veh = 17 (5 female), T: veh = 19 (5 female), T: ani (0h) = 20 (5 female), and T: ani (0h) = 10 (5 female) mice. For intracranial infusions (F–J), NT: veh = 23, T: veh = 40, T: ani-BLA = 19, and T: ani-vHC = 20 mice. Half of the vehicle-treated animals had a cannula in the BLA and the other half in the vHC. *p < 0.05, **p < 0.01, and ***p < 0.001. Error bars reflect standard error of the mean. Statistics are presented in the main text.

Figure 3.. Neuronal activity in the BLA and vHC supports distinct stress-induced defensive behaviors

(A) A pan-neuronal virus expressing the inhibitory chemogenetic receptor HM4D, or EGFP, was infused in the BLA or vHC. (B) HM4D+ neurons, as well as neighboring HM4D− neurons, were recorded before and after cno application. Application of cno dramatically reduced action potentials in HM4D+ neurons. (C) Animals underwent trauma and a week later were tested twice for their associative recall of the traumatic event, first in a drug-free baseline test (bl), and second, after receiving an injection of cno or saline (veh). (D) Animals with HM4D in the BLA and vHC did not differ during the initial trauma. (E) Administration of cno reduced freezing in animals with HM4D in either the BLA or vHC. (F) Animals underwent trauma, and a week later, they were tested for associative fear of the trauma environment, anxiety-related behavior in the light-dark test, and their response to a novel stressor in a new environment. The BLA/vHC were inhibited via cno administration prior to the light-dark test, as well as prior to the novel stressor. (G) No group differences were observed during the initial trauma. (H) No group differences were observed during the drug-free trauma recall test. (I) Inhibition of the vHC, but not the BLA, reduced anxiety-related behavior in the light-dark test. (J) Inhibition of the BLA, but not the vHC, reduced freezing in the test of stress sensitization. For electrophysiological recordings in (B), HM4D− = 2, HM4D+: BLA = 6, and HM4D+: vHC = 7 cells. For effects of inhibition on recall in (C)–(E), BLA: veh = 7, BLA: cno = 7, vHC: veh = 7, and vHC: cno = 9 mice. For effects of inhibition on light-dark and novel stressor in (F)–(J), EGFP: BLA/vHC = 25, HM4D: BLA = 24, and HM4D: vHC = 19 mice. *p < 0.05, **p < 0.01, and ***p < 0.001. Error bars reflect standard error of the mean. Statistics are presented in the main text.

Figure 4.. Reciprocal BLA-vHC connections do not support stress-induced changes in anxiety-related behavior or stress sensitivity

(A) Projection-specific targeting of BLA cells projecting to the vHC, or vice versa, was accomplished by infusing a cre-expressing retrograde virus into the projection target structure and cre-dependent HM4D/control virus into the projection origin structure. EGFP-expressing virus was co-infused into the projection target to confirm surgical placement. (B) Animals underwent trauma and a week later were tested for associative fear of the trauma environment, anxiety-related behavior in the light-dark test, and their response to a novel stressor in a new environment. BLA-vHC connections were inhibited via cno administration prior to the light-dark test, as well as prior to the novel stressor. (C) No group differences were observed during the initial trauma. (D) No group differences were observed during the trauma recall test. (E) No group differences were observed during the light-dark test of anxiety-related behavior. (F) No group differences were observed during the novel stressor test for stress sensitization. mCherry = 11, BLA→vHC = 11, and vHC→BLA = 11 mice. *p < 0.05, **p < 0.01, and ***p < 0.001. Error bars reflect standard error of the mean. Statistics are presented in the main text.