Phasic signaling in the bed nucleus of the stria terminalis during fear learning predicts within- and across-session cued fear expression

Abstract

While results from many past studies have implicated the bed nucleus of the stria terminalis (BNST) in mediating the expression of sustained negative affect, recent studies have highlighted a more complex role for BNST that includes aspects of fear learning in addition to defensive responding. As BNST is thought to encode ambiguous or unpredictable threat, it seems plausible that it may be involved in encoding early cued fear learning, especially immediately following a first tone-shock pairing when the conditioned stimulus-unconditioned stimulus (CS-US) contingency is not fully apparent. To investigate this, we conducted in vivo electrophysiological recording studies to examine neural dynamics of BNST units during cued fear acquisition and recall. We identified two functionally distinct subpopulations of BNST neurons that encode the intertrial interval (ITI) and may contribute to within- and across-session fear learning. "Ramping" cell activity during cued fear acquisition parallels the increase in freezing expression as mice learn the CS-US contingency, while "Phasic" cells encode postshock (US_post) periods (30 sec following encounter with footshock) only during early trials. Importantly, the magnitude of Phasic unit responsivity to the first US_post period predicted not only freezing expression in response to the subsequent CS during acquisition, but also CS freezing evoked 24 h later during CS retrieval. These findings suggest for the first time that BNST activity may serve as an instructive signal during cued fear learning.

Figure 1.

Freezing expression during cued fear acquisition, CS retrieval, and CX test. (A) Mice (n = 10) were trained to fear an auditory stimulus by pairing it with footshock in Context A. Freezing expression increased over the five acquisition trials, and was significantly different from baseline (noted by “*”) as well as CS-off periods (noted by “#”) during trials 4 and 5. (B) Individual animals’ movement speeds (normalized to baseline speeds) reflected US-induced movement and a reduction in movement during the CS and ITI over the course of the session. Gray boxes represent CS-on times, and black vertical bars indicate time of US delivery. (C) On day 2, mice received 10 CS presentations in Context B. Relative to baseline, the CS evoked freezing on trials 1–6, and 8. Significant freezing levels were observed during the ITI (relative to baseline) on trials 2, 4, and 7. (D) Individual animals’ movement speed during the CS retrieval session, each normalized to its baseline movement speed. Gray boxes indicate CS-on times. (E) On day 3, mice returned to Context A for 5 min to assess contextual fear expression. Freezing was significantly greater during block 3 versus other time blocks. (F) Individual animals’ raw movement speed over the context exposure. (G) Each animal's freezing to the final CS on acquisition day positively correlated with freezing expression during CS retrieval (top) and CX test (bottom). Data in A, C, and E represent the mean ± S.E.M. (*) P < 0.05, (**) P < 0.01, (***) P < 0.001, (###)P < 0.001.

Figure 2.

BNST phasic encoding predicts within- and across session fear expression. (A) Perievent time histograms (PETHs; 10 sec bins) of representative BNST units recorded during the cued fear acquisition session. Phasic units (n = 34 of 79) exhibited significant firing rate changes following early CS–US pairings that returned to baseline by later trials (top, middle), while Ramping units (n = 19 of 79) exhibited gradual, sustained firing rate changes that emerged some time after the first shock presentation and persisted throughout the session (bottom). (B) Population heat plot of Z scored activity (relative to 3 min baseline) of all units recorded across the acquisition session (10 sec bins, organized from greatest to least Z scored US_post response for each unit classification type). Phasic units exhibited significant changes in spike rate following US_post 1 or 2, while Ramping units exhibited significant rate changes lasting through the end of the session. (C) Histological reconstruction of recorded units, color coded by unit type. Blue = Phasic, red = Ramping, and black = no response. Placements overlaid on sagittal atlas sections, 0.72, 0.84, and 0.96 mm lateral relative to midline. (D) The majority of units recorded during acquisition were responsive to the ITI (top, lilac) as well as the US_post periods (bottom, aqua). Very few units were responsive only to the CS-on period (top, violet). ITI and US_post responsive units were comprised of both Phasic (blue) and Ramping (red) units. Lighter blue and red versus darker blue and red shading indicate the proportion of units that exhibited significantly increased versus decreased firing rates relative to baseline, respectively. (E) Average freezing behavior (gray) overlaid on line graphs depicting the fraction of responsive Phasic (blue) and Ramping (red) units for the CS-on, US_post, and ITI periods across trials. As freezing expression increased across trials, fraction of responsive Phasic units decreased as the fraction of responsive Ramping units increased. (F) For each unit type, average maximum absolute Z score during the US_post period for each trial. Average Phasic response to the US_post decreased across trials, while average Ramping response to the US_post increased across trials. (G) Responsiveness to the first US_post period negatively predicted freezing behavior during the subsequent CS presentation for Phasic, but not Ramping, units. (H) On day 2, very few of the 81 units recorded were responsive to either the CS-only or ITI periods (top). However, acquisition day responsiveness to the first US_post period negatively predicted CS-evoked freezing behavior during early trials in Context B 24 h later. (I) About half of all units recorded on day 3 (n = 74) when mice returned to Context A were responsive to the context. Unlike previous days, responsiveness to the first US_post period on acquisition day did not predict freezing to the context on day 3. Panels A, B, and E: Gray boxes represent CS-on times, and black vertical bars indicate time of US delivery. Data in E and F represent the mean ± S.E.M. (*) P < 0.05.