Disynaptic specificity of serial information flow for conditioned fear

Abstract

Memory encoding and retrieval rely on specific interactions across multiple brain areas. Although connections between individual brain areas have been extensively studied, the anatomical and functional specificity of neuronal circuit organization underlying information transfer across multiple brain areas remains unclear. Here, we combine transsynaptic viral tracing, optogenetic manipulations, and calcium dynamics recordings to dissect the multisynaptic functional connectivity of the amygdala. We identify a distinct basolateral amygdala (BLA) subpopulation that connects disynaptically to the periaqueductal gray (PAG) via the central amygdala (CeA). This disynaptic pathway serves as a core circuit element necessary for the learning and expression of conditioned fear and exhibits learning-related plasticity. Together, our findings demonstrate the utility of multisynaptic approaches for functional circuit analysis and indicate that disynaptic specificity may be a general feature of neuronal circuit organization.

Fig. 1.. Disynaptic specificity of BLAp→CeA→vlPAG pathway for conditioned fear.

(A) Scheme illustrating viral injection strategy to express Arch in neurons projecting from the BLA to CeA. CAV2-Cre was injected into the CeA, and then, AAV-flex-Arch-GFP was injected into the BLA. Optic fibers were bilaterally implanted to target labeled BLA neurons. hCAR mice were used to facilitate transduction by CAV2 viruses. (B) Histology of injection and fiber implantation sites: (left) CeA and (right) BLA. Scale bars, 250 μm. (C) Scheme illustrating viral injection strategy to express ArchT in neurons projecting from the BLAp to the CeA-vlPAG pathway. CAV2-Cre was injected into the vlPAG and AAV-DIO-TVA-2A-G was injected into the CeA in hCAR mice. Three days before FC, rabies-ArchT-GFP or rabies-GFP was injected into the CeA. Optic fibers were bilaterally implanted to target posterolateral BLA neurons. (D) Histology of the fiber implantation site. Scale bar, 200 μm. (E) Average freezing levels (means ± SEM) during auditory FC (day 1). Optogenetic manipulation was applied to cover CS + US periods. Freezing levels during a 2-min baseline (BL) and during each CS presentation (trials 1 to 5). N = 13, 10, and 7 mice for GFP (GFP control group), Mono (monosynaptic manipulation group), and Di (disynaptic manipulation group), respectively. (F) Average freezing levels (means ± SEM) during eight CS presentations during fear memory retrieval (day 2). *P = 0.0001; not significant (n.s.) P = 0.99; Dunnett’s test.

Fig. 2.. SST⁺ CeA neurons connect BLAp to vlPAG.

(A) Scheme illustrating viral injection strategy to express ArchT in neurons projecting from the BLAp to the CeA(SST⁺)-vlPAG pathway. AAV-DIO-TVA-2A-G was injected into the CeA in SST-Cre mice, and then, rabies-ArchT-GFP or rabies-GFP was injected into the vlPAG. Optic fibers were bilaterally implanted to target BLAp. Experimental animals underwent optogenetic manipulations during either FC (ArchT-FC) or retrieval (Arch-retrieval). GFP controls had the laser on both during FC and retrieval. Optogenetic manipulations were performed to temporally cover CS + US or CS presentations. (B) Average freezing levels (means ± SEM) during auditory FC (day 1). Freezing levels during a 2-min baseline and during each CS presentation (one to five trials). For the ArchT-FC group, optogenetic manipulation was applied to cover CS + US periods during FC but not during CS presentation during retrieval. In contrast, for the ArchT-retrieval group, optogenetic manipulation was applied to cover CS periods during retrieval but not during CS + US presentation during FC. N = 4, 5, and 4 mice for GFP, ArchT-FC, and ArchT-retrieval, respectively. (C) Average freezing levels (means ± SEM) during eight CS presentations in fear memory retrieval (day 2). ArchT-FC versus GFP: *P = 0.012; ArchT-retrieval versus GFP: *P = 0.002; Dunnett’s test. (D) Scheme illustrating viral injection strategy to express ArchT in CeA(SST⁺) neurons. AAV-flex-NpHR was injected into the CeA in SST-Cre mice. Optic fibers were bilaterally implanted to target vlPAG. (E) Average freezing levels (means ± SEM) during auditory FC (day 1) without optogenetic manipulation. N = 5 and 8 for GFP and SST-NpHR, respectively. (F) Average freezing levels (means ± SEM) during eight CS presentations in fear memory retrieval (day 2). Optogenetic manipulations were performed to temporally cover CS presentations. *P = 0.0031; rank sum test.

Fig. 3.. BLAp neurons that disynaptically connect to vlPAG show learning-related CS response plasticity.

(A and B) Scheme illustrating viral injection strategy to express GCaMP6s in neurons projecting from the BLA to CeA. AAV-DIO-TVA-2A-G was injected into the CeA in SST-Cre mice, and then, rabies-GCaMP6s was injected into the CeA. Optic fibers were unilaterally implanted to target labeled posterior BLA neurons. (C and D) Averaged z-scored photometry traces on day 1 (C) or day 2 (D) from monosynaptically labeled mice. Means ± SEM (N = 5 mice). (E) Scheme illustrating viral injection strategy to express GCaMP6s in neurons projecting from the BLAp to the CeA(SST⁺)-vlPAG pathway. AAV-DIO-TVA-2A-G was injected into the CeA in SST-Cre mice, and then, rabies-GCaMP6s was injected into the vlPAG instead of CeA. Optic fibers were bilaterally implanted to target GCaMP6s-expressing BLAp neurons. (F and G) Averaged z-scored photometry traces on day 1 (F) or day 2 (G) from disynaptically labeled mice. Means ± SEM (N = 4 mice). (H) Summary of responses to CSs. CS responses to the first CS on day 1 were subtracted from those to each CS. Thick lines show group averages. Thin lines represent data from individual animals. (I) CS response comparisons within day 1 (left) and across days (right). Thick circles show group averages. Thin circles represent data from individual animals. *P = 0.016 (left); *P = 0.032 (right); rank sum test.