Bombesin-like peptide recruits disinhibitory cortical circuits and enhances fear memories

Abstract

Disinhibitory neurons throughout the mammalian cortex are powerful enhancers of circuit excitability and plasticity. The differential expression of neuropeptide receptors in disinhibitory, inhibitory, and excitatory neurons suggests that each circuit motif may be controlled by distinct neuropeptidergic systems. Here, we reveal that a bombesin-like neuropeptide, gastrin-releasing peptide (GRP), recruits disinhibitory cortical microcircuits through selective targeting and activation of vasoactive intestinal peptide (VIP)-expressing cells. Using a genetically encoded GRP sensor, optogenetic anterograde stimulation, and trans-synaptic tracing, we reveal that GRP regulates VIP cells most likely via extrasynaptic diffusion from several local and long-range sources. In vivo photometry and CRISPR-Cas9-mediated knockout of the GRP receptor (GRPR) in auditory cortex indicate that VIP cells are strongly recruited by novel sounds and aversive shocks, and GRP-GRPR signaling enhances auditory fear memories. Our data establish peptidergic recruitment of selective disinhibitory cortical microcircuits as a mechanism to regulate fear memories.

Keywords: CRISPR-Cas9; VIP cells; cortex; disinhibition; fear memory; gastrin-releasing peptide; neuropeptide.

Fig. 1:. Cortex-wide, cell type-specific expression of GRP and its receptor.

A, Schematic of GRP release and binding to GRP receptor (GRPR) in unidentified cell types. B, GRP and GRPR expression were analyzed in mice in the indicated areas. Abbreviations: M1, primary motor; A1, primary auditory; V1, primary visual; S1BF, primary somatosensory – barrel field; ACC, anterior cingulate; AI, anterior insular cortex. C, Locations of all identified Grp⁺ cells in the indicated areas (n=5 slices for S1BF, V1, A1 and 7 for M1). see also Fig. S1. D, Locations of all identified Grpr⁺ cells in indicated areas (n=34 slices, 4900 cells). Quantification of the proportions (mean ± SEM) of cells that are Grpr⁺, Vip⁺ or Gad⁺ across cortical depth (20 bins). E, Representative confocal images of mouse cortex showing coexpression of Grp (top) and Grpr (bottom) with Vip, glutamatergic markers (Slc17a6-8 encoding vGluT1-3) and GABAergic markers (Gad1,2). Scale bars, 20 μm. F, Quantification of coexpression of Grp (top) and Grpr (bottom) with indicated genes. Numbers of analyzed cells per area are indicated above bars (≥15 slices from 4-7 hemispheres per area). G, Schematic of human visual cortex (BA17) in which GRPR expression was analyzed using FISH. H, Left, Locations of all identified GRPR⁺ cells in 5 sections of human visual cortex. Right, Quantification of the proportions (mean ± SEM) of cells that are GRPR⁺, VIP⁺ and GAD1⁺ (n=882 cells; 20 bins). I, Representative confocal image of a GRPR⁺/VIP⁺/GAD1⁺ human cell. See also Fig. S1.

Fig. 2:. Putative local and long-range sources of GRP

A, Schematic of retrograde tracing with CTB to quantify Grp expression in L6 cortico-thalamic neurons. B, Representative epifluorescent image of CTB injection into auditory thalamus (Thal) and retrograde labeling in auditory cortex (ACx) L6. C, Confocal image of retrogradely labeled cells in L6 of ACx and FISH against Grp. Inset: magnification of the highlighted area. D, Retrograde tracing with CTB injected into ACx to quantify Grp expression in corticopetal projection neurons. Epifluorescent image of a representative injection and FISH against Grp. E, Quantification of Grp⁺ and CTB⁺ cells in the indicated areas following injection as in D. Each dot represents data from one mouse. Mean ± SEM across 358-1142 CTB⁺ cells per area. Abbreviations: LA, lateral amygdala; Per, perirhinal cortex; TeA, temporal association area; clACx, contralateral ACx. F, Schematic of transsynaptic tracing from Vip⁺ starter cells using pseudotyped rabies virus SADΔG-EnVA-H2B-EGFP (RabV). G, Confocal image of an exemplary Vip^Cre+ starter cell in ACx identified by RabV-gp1 (RabV), oG and Cre (FISH). H, Confocal images of RabV-gp1⁺ cells in ACx (left) and of an exemplary Grp⁺/RabV-gp1⁺ cell (right). I, Quantification of numbers of RabV-gp1⁺ and Grp⁺ cells after injections into ACx normalized to the numbers of starter cells. Each dot represents data from one mouse. Mean ± SEM. J, Schematic of putative Grp⁺ inputs to ACx VIP cells. See also Fig. S2.

Fig. 3:. GRP depolarizes and increases intracellular Ca²⁺ in cortical VIP cells

A, Schematic of whole-cell recording of ACx VIP cell used to examine effects of GRP. B, Two exemplary VIP cells responding with bursts (top) or depolarization (bottom) following 2 min bath application of GRP (300 nM) in the presence of NBQX, CPP, gabazine, CGP. Inset: magnification of first bursts indicated by an asterisk. C, Depolarization of VIP cells upon application of indicated concentrations of GRP. Mean ± SEM. 6-10 cells per group. Comparison 0 vs. 300 nM GRP: t-test for unequal variance: t(12.52)=3.76, p=0.01. Other comparisons n.s. Bath contains NBQX, CPP, gabazine, CGP. D, Representative firing patterns of ACx VIP, SST, PVALB and pyramidal (Pyr) cells upon −200 pA current injection (bottom), at AP threshold (middle), and at maximal firing rate (top). E, Average time course (left) and amplitude (right) of the membrane potential changes in each indicated cell type in ACx following GRP application. Bath contains NBQX, CPP, gabazine, CGP, TTX. Mean ± SEM. n=10 VIP, SST, Pyr cells and 15 PVALB cells. Comparison to VIP cells (Bonferroni-corrected t-test): SST: t(18)=−5.27, p<0.001; PVALB: t(23)=−3.83, p<0.001; Pyr: t(18)=−4.87, p<0.001. F, Design of plasmid for Cre-dependent stoichiometric expression of GCaMP and mBeRFP for imaging of Ca²⁺ entry and detection of infected cells, respectively. G, Injection of AAV DIO-GCaMP-P2A-mBeRFP into ACx of male Vip^Cre mice (left) and epifluorescent GCaMP imaging (right) in acute slices before (top) and after (bottom) GRP bath application, in the presence of NBQX, CPP, gabazine and CGP. H, Heatmap of fluorescence changes (expressed relative to fluorescence following KCl application, ΔF/F_KCl) across all imaged VIP cells in an exemplary acute ACx slice. I, GCaMP fluorescence changes with or without TTX in two exemplary VIP cells (top) or across all recorded cells (bottom) in the presence of NBQX, CPP, gabazine and CGP. Mean ± SEM, Mann-Whitney U test: U=10178, p<0.0001. n=218 (−TTX) and 179 (+TTX) cells in 7 and 6 slices respectively. See also Fig. S3.

Fig. 4:. GRP disinhibits cortex and induces IEG expression

A, Schematic of the disinhibitory circuit that underlies VIP cell function in cortex. B, Whole-cell recordings of IPSCs in a representative SST cell in ACx before (top) and following (bottom) GRP application in the presence of NBQX and CPP. C, Time courses (left, mean ± SEM) and magnitude (right, median and IQR) of IPSC frequency changes in SST, PVALB and Pyr cells (n=10 cells per group) in the presence of NBQX and CPP (TTX where indicated). Mann-Whitney U test: U=21, p=0.03. D, Representative epifluorescent image of FOS immunostaining after injection of 3 μM GRP, as schematized, into the right motor cortex in anaesthetized mice. E, Confocal images of Fos expression in Vip⁺ (arrows) and glutamatergic cells (arrowheads) analyzed using FISH. F-H, Fos and Npas4 expression levels in Vip⁺ and glutamatergic cells across all cortical layers for the right (green/turquoise) and left (black) motor cortices (mean ± SEM). Intensity-coded map of expression levels (% coverage) of all glutamatergic cells in an exemplary slice shown on the left (G, H). n=398 (Vip), 15108 (Slc17a6,7) cells, 3-5 mice for each condition and cell type. See also Fig. S4.

Fig. 5:. ACx VIP cells encode novel sounds and shocks during fear conditioning

A, Fluorescence emission spectrum of GCaMP alone (green) and GCaMP-P2A-mBeRFP (magenta, pink) expressed in HEK 293T cells and measured with or without application of Ca²⁺/ionomycin, indicating that mBeRFP does not interfere with GCaMP6s fluorescence. 2-sample t-test: t(10)=−0.12, p=0.91; n=6 wells each. B, Confocal image of a Pyr cell after injection of AAV DIO-GCaMP-P2A-mBeRFP and AAV Cre into ACx. C, Schematic of experimental setup for analysis of AP-dependent changes of GCaMP and mBeRFP fluorescence in acute brain slices during electrophysiological induction of spiking in VIP or L5 Pyr cells expressing GCaMP-P2A-mBeRFP in ACx of Vip-IRES-Cre and Rbp4-Cre mice. D, AP bursts (each 5 s at 10 Hz) induced in an exemplary RBP4⁺ neuron through a cell-attached electrode (top) with GCaMP and mBeRFP fluorescence (473 nm excitation, middle) and GCaMP fluorescence (405 nm excitation, bottom). Inset shows individual spikes from the last burst. E, Average GCaMP and mBeRFP (473 nm excitation) and GCaMP (405 nm excitation) fluorescence changes (ΔF/F). Data from cell-attached and whole-cell recordings were pooled. Dashed line: baseline fluorescence. Mean ± SEM from n≥3 mice each; n=12 VIP, 15 RBP4 GCAMP/mBeRFP, and 12 RBP4 GCaMP (405 nm) cells. F, Quantification of fluorescence changes shown in E normalized to GCaMP (473 nm) fluorescence change. mBeRFP fluorescence was largely unaffected by neuronal activity. In comparison, GCaMP fluorescence was reduced when excited at 405 nm (−7.4±0.8%). G, Correlation of GCaMP and mBeRFP fluorescence in Vip⁺ cells after injection of AAV DIO-GCaMP-P2A-mBeRFP into ACx of Vip-IRES-Cre mice (linear regression and correlation coefficient in cyan). n=335 cells from 3 injection sites. H, GCaMP fluorescence changes measured in ACx VIP cells around presentation of conditioned (CS⁺, blue) and unconditioned sounds (CS⁻, grey) and shocks (dashed pink lines) early (trial 1-4) and late (trial 12-15) on the conditioning (top) or retrieval (bottom) day. n=11 mice. I, GCaMP fluorescence changes measured around presentation of CS⁺ (blue) and CS⁻ (grey) and shocks (dashed pink lines) before (top) and following (bottom) infusion of GRP (pink) or control solution (NRR, black). n=8 (GRP) and 7 (NRR) mice. Data in A, F-I: Mean ± SEM. See also Fig. S5.

Fig. 6:. GRPR signaling in the ACx enhances fear memories

A, Design of the plasmid for CRIPSR/Cas9-mediated KO of Grpr (here abbreviated as SaCas9-sgRNA) (Tervo et al., 2016). B, GCaMP fluorescence changes measured in acute slices upon GRP application for VIP cells expressing GCaMP-P2A-mBeRFP and SaCas9-sgRNA targeting either Grpr (KO, Grpr1) or lacZ (ctrl). Mann-Whitney U test: U=25956; p<0.0001; n=602 and 298 cells in 10 (ctrl) and 9 (KO) slices. C: Quantification of bilateral SaCas9-HA expression after injection of SaCas9-sgRNA targeting lacZ (ctrl, grey) or Grpr (KO, turquoise). See Fig. S6 for analysis of expression in the whole brain. Color-code: % of mice with SaCas9-HA expression. D, Auditory fear acquisition, measured as the percentage of time spent freezing during presentation of 15 CS⁺ and CS⁻ on conditioning day (histology shown in C). 2-way ANOVA, main effect of genotypes: CS⁺: p=0.90, F=0.01; CS⁻: p=0.78, F=0.08, no significant interaction of genotype and stimulus number. N=15 mice per group. E, Auditory fear memory retrieval, measured as the percentage of time spent freezing averaged across 15 presentations of CS⁺ and CS⁻ on the retrieval day. 2-way ANOVA: main effect of genotype: p=0.01, F=6.41, no significant interaction of genotype and stimulus (CS⁺ vs CS⁻). F, Time courses of average freezing probability across all CS⁺ and CS⁻ during fear memory retrieval. G, Sound discrimination indices measured during retrieval. T-test for unequal variance: t(20.13)=0.53, p=0.60. H, Auditory fear memory retrieval in CRISPR/Cas9-expressing Grp^−/− KO mice, measured as the percentage of time spent freezing averaged across 15 presentations of CS⁺ and CS⁻ on the retrieval day. 2-way ANOVA: main effect of genotype: p=0.28, F=1.21, no significant interaction of genotype and stimulus (CS⁺ vs CS⁻), n = 14 mice per group. All summary data shown as Mean ± SEM. See also Fig. S6.

Fig. 7:. Impaired fear memory in mice with conditional KO of GRPR in the ACx

A, GCaMP fluorescence changes in VIP cells following GRP application in acute ACx slices from ctrl mice (Vip-IRES-Cre;Grpr^wt/y) or mice lacking GRPR in VIP cells (Vip-IRES-Cre;Grpr^fl/y). Two-sample t-test: t(1178)=26.97, p<0.0001, n=655 cells in 14 slices (ctrl) and 525 cells in 11 slices (KO). B, Injection of AAV hSyn-Cre-mCherry into ACx of Grpr^wt/y or Grpr^fl/y mice to locally KO Grpr. Right: Epifluorescent image of exemplary injection sites. Quantification of expression levels across all mice: Fig. S7. C, Time spent freezing during fear memory retrieval in Grpr^wt/y (ctrl) and Grpr^fl/y (KO) mice injected into ACx with AAV encoding Cre-mCherry. 2-way ANOVA: main effect of genotype: p=0.004, F=8.99, no significant interaction of genotype and stimulus (CS⁺ vs CS⁻). N=14 mice per group. D, Time course of average freezing probability across all CS⁺ and CS⁻ presentations during fear memory retrieval. E, Photometric recordings from LA/BLA projection neurons, retrogradely tagged with AAVretro-Cre injections into the ACx. GCaMP fluorescence changes measured during CS⁺ (blue), CS⁻ (grey) and shocks (dashed pink lines) during early conditioning trials (trial 1-4). n=10 mice. F, Photometric recordings from thalamic SG/MGM projection neurons, retrogradely tagged with AAVretro-Cre injections into the ventral ACx and TeA. GCaMP fluorescence changes measured during early conditioning trials (trial 1-4). n=5 mice. All summary data shown as Mean ± SEM. See also Fig. S7.