Photostimulation of basal forebrain cholinergic neurons promotes conditioned responding when associated with the opportunity to collect rewards

(A) Window of Opportunity Task (WoOT) to study conditioned responding in the absence of discrete cues. Mice were trained prior to any photostimulation, using an operant, variable-interval reinforcement schedule with limited hold (Ferster and Skinner, 1957). Sessions were divided into trials of 3-second Windows of Opportunity, with variable intertrial intervals (ITIs). Rewards were only delivered if mice licked during an unsignaled Window of Opportunity (green; 90% of trial windows). We also included a subset of Unrewarded Windows, on which, even if the mouse licked, reward would not be delivered (purple; 10% of trial windows), similar to intervening ITIs (white). Because windows were not discretely cued and occurred after variable ITIs, mice did not know when they initiated a lick whether it would be rewarded.

(B) Sample WoOT behavior. If a mouse licked during an uncued Windows of Opportunity (green), then reward was delivered. But if a mouse licked during an uncued Unrewarded Window (purple) or during the ITI (white), reward was not delivered. reward delivery. The use of Unrewarded Windows (purple) allowed us to investigate behavior during epochs temporally matched to Windows of Opportunity (green). Training mice in this manner conditioned them to lick the spout with a relatively stable pattern of intermittent lick bouts (Figure S1).

(C) Optogenetic strategy to photostimulate cholinergic basal forebrain neurons, by driving Cre-dependent expression of Channelrhodopsin-2 (ChR2) or a control fluorophore (eYFP) in mice expressing Cre-recombinase under control of the choline acetyltransferase promoter (ChAT::Cre). The photostimulation parameters displayed were used in later behavioral sessions.

(D) Sample histology of fiber placement over cholinergic neurons in the posterior portion of the basal forebrain, the sublenticular substantia innominata/extended amygdala. Blue=DAPI nucleic acid staining, Green=cre-dependent expression of ChR2 fused to eYFP, Red=anti-ChAT immunohistochemical staining. AP coordinate = -0.46. Scale bar = 500 microns. See also Figure S1A-B.

(E) Behavioral training and testing sessions. After early WoOT training without any photostimulation, mice subsequently received testing during a Photostim-Unreinforced or Photostim-Reinforced sessions.

(F) Photostim-Unreinforced Session. In addition to no photostimulation Windows of Opportunity (green, 80%) and Unrewarded Windows (purple, 8%), mice received 2 sec of photostimulation (blue lines) during a subset of Unrewarded Windows (purple, 12%) to study innate responses to photostimulation.

(G) Photostim-Reinforced Session. Conversely to Photostim-Unreinforced sessions, during Photostim-Reinforced sessions, photostimulation (blue lines) was delivered during a subset of Windows of Opportunity (green, 12%), during which, if mice licked, they would receive a reward. Mice still had many more Windows of Opportunity with no photostimulation (68%).

(H & I) The likelihood of licking depended on an interaction of Virus, Photostimulation Window type, and Photostim-Reinforcement Session type (linear mixed effects model, F1,78=5.26, p=0.025). Thin lines represent data from all individual mice, pooled data are represented as mean ± SEM. During the Photostim-Unreinforced Session (H), there was no significant difference between ChR2 (blue) and eYFP (gray) groups, regardless of photostimulation window type. However, during the Photostim-Reinforced Session (I), ChR2 mice licked significantly more during Photostim windows than No Photostim windows (post-hoc tests with Sidak correction for multiple comparisons: ***p<0.0001). ChR2 mice also licked significantly more than eYFP mice during Photostim windows during Photostim-Reinforced sessions (**p=0.001). There were no detectable effects of photostimulation on licking in eYFP mice, and no detectable differences in licking on no photostimulation windows between ChR2 and eYFP mice (all other post-hoc comparisons p>0.10).

(J) We initially photostimulated with a fixed duration, but in a follow-up session photostimulated with varied durations (0-0.5 sec, using different number of pulses at the same frequency/duty cycle). The likelihood of licking depended on an interaction of Virus and the number of photostimulation pulses (linear mixed effects model, F3,84=16.22, p=2.1x10-8). ChR2 mice licked significantly more on windows with 1, 2, or 10 pulses than windows without photostimulation, and more than eYFP mice with any number of pulses (post-hoc tests with Sidak correction for multiple comparisons: *p=0.015, **p=0.0095, ***p<0.001; all other post-hoc comparisons p>0.10).

(K) When tested with different durations of photostimulation, the latency of the first lick depended on an interaction of Virus and the number of photostimulation pulses (linear mixed effects model, F3,84=4.53, p=0.005). ChR2 mice licked significantly sooner on windows with 1, 2, or 10 pulses than windows without photostimulation, and sooner than eYFP mice for 2 and 10 pulses (post-hoc tests with Sidak correction for multiple comparisons: **p=0.0047, ***p<0.001; all other post-hoc comparisons p>0.10).

BFChAT:ChR2-induced conditioned responding is muscarinic receptor-dependent

(A) Experimental strategy to test the necessity of cholinergic receptors in conditioned responding. Cholinergic neurons in the basal forebrain were photostimulated after blockade of cholinergic muscarinic or nicotinic receptors using intraperitoneal injection of pharmacologic antagonists.

(B) Modification of Window of Opportunity Task (WoOT) to include additional tone conditioned responses. To provide additional within-subject controls for pharmacologic testing, mice could now receive rewards on either of two separate types of Windows of Opportunity: with tones (top row) or with photostimulation (middle row, both 2 sec duration). Licks during unsignaled, Unrewarded Windows were recorded but had no consequence (bottom row).

(C) Photostim-induced licking was abolished by systemic muscarinic receptor antagonist administration. Linear mixed effects modelling confirmed that licking depended upon an interaction between Virus group, Stimulus type, and Drug session (F3,363=4.61, p=0.0002). Thin lines represent data from individual mice, pooled data are displayed as mean ± SEM. Saline: Both ChR2 and eYFP mice responded more during Tone Windows of Opportunity than during unsignaled, Unrewarded Windows (Unrwd) (***p<0.001, *p<0.05, #p=0.10, Sidak post-hoc multiple comparisons). However, only ChR2 mice responded more during Photostimulation Windows of Opportunity than during unsignaled, Unrewarded Windows, at a similar likelihood as their responses during tone Windows of Opportunity. Scopolamine 0.3mg/kg: ChR2 mice now responded less during photostimulation than during tones. Scopolamine 1mg/kg: ChR2 mice no longer responded more during photostimulation than during Unrewarded Windows, and no longer responded more during photostimulation than eYFP mice, although they continued to respond more during tones than ITIs. Mecamylamine 1mg/kg: Response patterns were similar to Saline sessions. For each session, the likelihood of licking during Unrewarded Windows was similar between ChR2 and eYFP mice (all p>0.8). Additionally, within each group, the likelihood of licking during Unrewarded Windows was similar to Saline sessions for all drug doses (all p>0.8).

Cholinergic basal forebrain neural activity increases during conditioned stimuli and responses, even in the absence of reward delivery

(A) Strategy to record fluorescent activity from basal forebrain cholinergic neurons expressing the calcium sensor GCaMP6s, using interleaved signal (470 nm, blue) and reference (405 nm, violet) wavelengths to elicit fluorescence (525 nm, green).

(B) Task windows. We recorded fluorescent activity from mice during a traditional operant cue detection task. If mice licked after the onset of a tone, a fluid reward was delivered after a 0.5 sec delay. Licks during the silent Unrewarded Windows had no consequence.

(C) Sample GCaMP photometry fluorescence traces from one mouse demonstrating signal increases around the times of tones, licks, and reward deliveries. Increases were apparently present even for licking in the absence of tones and rewards. The blue trace represents data from the signal wavelength (470 nm) and the violet trace represents interleaved data from the reference wavelength (405 nm).

(D) Changes in fluorescence from basal forebrain cholinergic neurons referenced to the time of tone onset. Heat maps represent trial-averaged data from each mouse. Top heat maps are for 470 nm excited fluorescence (Signal 470), bottom heat maps are for 405 nm reference (Ref 405). The bottom panel summary data are represented as mean ± SEM. Mice are sorted in all heat maps (D-F, I-J), in the order of average post-lick activity in panel E.

(E) Changes in fluorescence from basal forebrain cholinergic neurons referenced to the time of the first lick after tone onset. Licking triggered subsequent reward delivery (first lick at 0 sec, reward delivery at dashed line, 0.5 sec).

(F) Changes in fluorescence from basal forebrain cholinergic neurons referenced to the onset of matched lick bouts that were in the absence of tone cues and did not lead to reward delivery.

(G) Fluorescence levels from basal forebrain cholinergic neurons at baseline (-2 to -1.5 sec before each referenced event) and post-event time points (0 to 0.5 sec after events, to standardize analyses between events) in the Cued task. Fluorescence levels depended on an interaction of Wavelength, Time point, and Event type (linear mixed effects model, F2,55=3.28, p=0.045). Fluorescence levels increased at the time of events in the 470 nm wavelength signal channel (blue), but not the 405 reference channel (violet) (Tone: t55=2.64, #p=0.064; Lick leading to Reward Delivery (Lick & Rwd): t55=7.57, ***p<0.001; Licks in absence of Cue & Reward Delivery (Lick, No Rwd): t55=5.29, ***p<0.001; Sidak correction for six multiple comparisons). Thin lines represent data from all individual mice, pooled data are represented as mean ± SEM.

(H) Uncued Window of Opportunity Task (WoOT). All mice were also recorded from at an earlier stage of WoOT training, before experience with tones or other discrete cues. If a mouse licked during an uncued Window of Opportunity, a fluid reward was delivered. A 0.5 sec delay was instituted between lick and reward to account for the slow dynamics of GCaMP6s. Licks during Unrewarded Windows were recorded but had no consequence.

(I) Changes in fluorescence from basal forebrain cholinergic neurons referenced to the time of the first lick that triggered reward delivery (first lick at 0 sec, reward delivery at dashed line, 0.5 sec).

(J) Changes in fluorescence from basal forebrain cholinergic neurons referenced to the onset of matched lick bouts that did not lead to reward delivery.

(K) Fluorescence levels from basal forebrain cholinergic neurons at baseline (-2 to -1.5 sec before each referenced event) and post-event time points (0 to 0.5 sec after events, to standardize analyses between events) in WoOT. Fluorescence levels depended on an interaction of Wavelength and Time point (linear mixed effects model, F1,35=13.59, p=0.0008), without a third order interaction by Event type (F1,35=0.02, p=0.882). Fluorescence levels increased at the time of events in the 470 nm wavelength signal channel (blue), but not the 405 reference channel (violet) (Lick leading to Reward Delivery (Lick & Rwd): t35=3.58, **p=0.004; Licks in absence of Reward Delivery (Lick, No Rwd): t35=3.90, **p<0.002; Sidak correction for four multiple comparisons). Thin lines represent data from all individual mice, pooled data are represented as mean ± SEM.

Local ACh levels in the BLA, measured using a genetically-encoded sensor, increase during conditioned stimuli and responses

(A) Strategy to record local ACh levels in the BLA. A genetically-encoded, fluorescent ACh sensor (GACh3.0, (B)) was expressed in BLA neurons, and imaged using interleaved signal (470 nm, blue) and reference (405 nm, violet) wavelengths to elicit fluorescence (525 nm, green).

(B) The fluorescent ACh sensor, GACh3.0, is a fusion protein between a modified M3 muscarinic receptor and cyclically permuted GFP. GACh3.0 undergoes a conformational change and fluoresces to 470nm light after binding ACh. Please note that kinetics for GACh3.0(Jing et al., 2020) are somewhat faster than those for GCaMP6s.(Chen et al., 2013)

(C) Cued task windows. We recorded fluorescent activity from mice during a traditional operant cue detection task. If mice licked after the onset of a tone, a fluid reward was delivered after a 0.1 sec delay. Licks during ITIs had no consequence.

(D) Sample BLA ACh sensor fluorescence traces from one mouse demonstrating apparent increases around the times of tones, licks, and reward deliveries. Increases were apparently present even for licking in the absence of tones and rewards.

(E) Changes in BLA ACh sensor fluorescence referenced to the time of tone onset. Heat maps represent trial-averaged data from each mouse. Top heat maps are for 470 nm excited fluorescence (Signal 470), bottom heat maps are for 405 nm reference (Ref 405). The bottom panel summary data are represented as mean ± SEM. Mice are sorted in all heat maps (E-G, J-K), in the order of average post-lick activity in panel F.

(F) Changes in BLA ACh sensor fluorescence referenced to the time of the first lick after tone onset. Licking triggered reward delivery (first lick at 0 sec, reward delivery at dashed line, 0.5 sec).

(G) Changes in BLA ACh sensor fluorescence referenced to the onset of matched lick bouts that were in the absence of tone cues and did not lead to reward delivery.

(H) BLA ACh sensor fluorescence levels at baseline (-2 to -1.5 sec before each referenced event) and post-event time points (0 to 0.5 sec after events, to standardize analyses between events) in the Cued task. Fluorescence levels depended on an interaction of Wavelength and Time point (linear mixed effects model, F1,99=20.41, p<0.001), without a third order interaction by Event type (F2,99=1.20, p=0.305). Fluorescence levels increased at the time of events in the 470 nm wavelength signal channel (blue), but not the 405 reference channel (violet) (Tone: t99=2.80, *p=0.036; Lick leading to Reward Delivery (Lick & Rwd): t99=5.49, ***p<0.001; Licks in absence of Cue & Reward Delivery (Lick, No Rwd): t99=2.91, *p<0.027; Sidak correction for six multiple comparisons). Thin lines represent data from all individual mice, pooled data are represented as mean ± SEM.

(I) Uncued Window of Opportunity Task (WoOT). All mice were also recorded from at an earlier stage of WoOT training, before experience with tones or other discrete cues. If a mouse licked during an uncued Window of Opportunity, a fluid reward was delivered. Licks during Unrewarded Windows were recorded but had no consequence.

(J) Changes in BLA ACh sensor fluorescence referenced to the time of the first lick that triggered reward delivery (first lick at 0 sec, reward delivery at dashed line, 0.1 sec).

(K) Changes in BLA ACh sensor fluorescence referenced to the onset of matched lick bouts that did not lead to reward delivery.

(L) BLA ACh sensor fluorescence levels at baseline (-2 to -1.5 sec before each referenced event) and post-event time points (0 to 0.5 sec after events, to standardize analyses between events) in WoOT. Fluorescence levels depended on an interaction of Wavelength and Time point (linear mixed effects model, F1,63=44.21, p<0.001), without a third order interaction by Event type (F1,63=1.25, p=0.267). Fluorescence levels increased at the time of events in the 470 nm wavelength signal channel (blue), but not the 405 reference channel (violet) (Lick leading to Reward Delivery (Lick & Rwd): t63=8.41, ***p<0.001; Licks in absence of Reward Delivery (Lick, No Rwd): t63=5.94, ***p<0.001; Sidak correction for four multiple comparisons). Thin lines represent data from all individual mice, pooled data are represented as mean ± SEM.

Cholinergic signaling in the BLA is sufficient to promote conditioning responding but ACh release is independent of reward contingency

(A) Optogenetic strategy to photostimulate cholinergic (ChAT::Cre) basal forebrain terminals in the BLA selectively.

(B) The difference in the likelihood of licking between Photostim and No Photostim windows differed depending on the window in which photostimulation was delivered. The effect of photostimulation within each session is calculated for each mouse. A linear mixed-effects model to account for repeated measures demonstrated that the effect of photostimulation depended on an interaction between Virus and Session type (F1,15 = 9.624, p = 0.007). Post-hoc tests Sidak correction for multiple comparisons revealed that the effect of photostimulation was greater for ChR2 mice in Photostim-Reinforced than Photostim-Unreinforced sessions (***p=0.0004), and that the effect of photostimulation in Photostim-Reinforced sessions was greater for ChR2 mice than eYFP mice (*p=0.0215). All other comparisons were not significant (p>0.05).

(C) Schematic showing concurrent photostimulation of cholinergic terminals in the BLA while measuring local ACh using a genetically-encoded fluorescent sensor, through the same optic fiber. Mice either expressed ChrimsonR or a control fluorophore (tdTomato) in basal forebrain ChAT neurons.

(D) Sample fluorescent traces from ACh sensor (orange) from a mouse with ChrimsonR, in relationship to reward delivery (red), licks (black), behavioral windows (Reward green/Unrewarded purple), and photostimulation (orange). Photostimulation was either delivered in a Photostim-Unreinforced session (during Unrewarded Windows, left, purple) or Photostim-Reinforced session (during rewarded windows, right, green).

(E) Sample ACh fluorescent traces from ACh sensor (gray) from a mouse with a control fluorophore, displayed similarly to (E), in relationship to rewards, licks, behavioral windows, and photostimulation, delivered either in a Photostim-Unreinforced (left, purple) or Photostim-Reinforced session (right, green).

(F) Heat maps comparing average ACh measurements for each mouse around the time of photostimulation on the Photostim-Unreinforced session. Mice are separated based on whether they expressed ChrimsonR (orange, n=6) or control fluorophore (gray, n=4). Summary data in the bottom panel are represented as mean ± SEM. Mice are sorted in all panels based mean DF/F during laser stimulation.

(G) Heat maps comparing average ACh measurements for each mouse during photostimulation in the Photostim-Reinforced session. Conventions are as in (G), and mice are sorted in the same order as in (G).

(H) Mean ACh measurements evoked by photostimulation on the Unreinforced (left) or Reinforced (right) session. Evoked ACh measurements were higher for ChrimsonR mice than control fluorophore mice, but evoked ACh measurements did not depend upon whether photostimulation was provided on Unreinforced or Reinforced sessions (linear mixed effects model: effect of Virus F1,8=20.21, **p=0.002; effect of Session F1,8=0.47, p=0.51; interaction between Virus and Session type F1,8=0.86, p=0.38).

Cholinergic modulation of neural activity in vivo depends upon reward context in the amygdala, but not in the prefrontal cortex

(A) Strategy for photostimulation of cholinergic basal forebrain neurons and terminal region electrophysiology in the dorsomedial prefrontal cortex (dmPFC) and basolateral amygdala (BLA). Six ChR2 mice had electrodes implanted in both dmPFC and BLA. Five ChR2 mice had electrodes implanted only in the BLA, yielding a total of 11 ChR2 mice with electrodes in BLA. Photostimulation parameters were the same as in ChR2 behavioral experiments (Fig 1).

(B) Activity from all recorded neurons in each target area (dmPFC and BLA, total 963 neurons over all sessions from 11 mice), from sessions in which photostimulation was delivered during ITI Unrewarded Windows (Photostim-Unreinforced, purple) or sessions in which photostimulation was delivered during Windows of Opportunity (Photostim-Reinforced, green). Each row represents activity from a single neuron, normalized to baseline (-2 to 0 sec before photostimulation onset), and smoothed with a 50 ms Gaussian. Neurons are sorted according to mean activity during photostimulation (0 to +2 sec). Summary population data in the bottom panels are represented as mean ± SEM. Black marks underneath the population data represent 10 ms steps when the population activity differed between Photostim-Unreinforced vs. Photostim-Reinforced sessions (rank-sum test, p<0.01).

(C) Licking activity from all mice contributing recordings for each target area, from sessions in which photostimulation was delivered during ITI Unrewarded Windows (Photostim-Unreinforced, purple) or sessions in which photostimulation was delivered during Windows of Opportunity (Photostim-Reinforced, green). Each row represents activity from a single mouse. Summary population data in the bottom panels are represented as mean ± SEM. Black marks underneath the population data represent 10 ms steps when the population licking activity differed between Photostim-Unreinforced vs. Photostim-Reinforced sessions (rank-sum test, p<0.01).

(D) Example neural activity from each target area (dmPFC left, BLA right) around photostimulation of basal forebrain cholinergic neurons (0-2 sec), from Photostim-Unreinforced sessions. Top panels are individual trial rasters and black markers indicate the first lick following Photostim onset. Trials are sorted by lick latency. Summary data in the bottom panels are represented as mean ± SEM, smoothed with a 50 ms Gaussian kernel. We observed neurons that were facilitated and suppressed relative to baseline in both regions (signed-rank test of firing rate in the 1 sec before stimulation vs 0.5 sec after, p<0.01).

(E) Example neural activity from each target area (dmPFC left, BLA right) around photostimulation of basal forebrain cholinergic neurons (0-2 sec), from Photostim-Reinforced sessions. We again observed neurons that were facilitated and suppressed relative to baseline in both regions. Conventions are the same as in (E).

(F) Proportions of neurons that were facilitated (solid bars) or suppressed (open bars) in each area during Photostim-Unreinforced sessions (purple) or Photostim-Reinforced sessions (green). Denominator n’s refer to neurons recorded across all mice during each session type. A higher percentage of BLA neurons were suppressed on Photostim-Reinforced sessions than Photostim-Unreinforced sessions (2-sample tests for equality of proportions: Χ2=6.81, df=1, *p=0.036, corrected for 4 multiple comparisons using Holm’s procedure). There was a trend towards a lower percentage of BLA neurons being facilitated on Photostim-Reinforced sessions than Photostim-Unreinforced sessions (Χ2=4.52, df=1, #p=0.10).

(G) Proportions of neurons that were facilitated (solid bars) or suppressed (open bars) in each area during Photostim-Unreinforced sessions (left, purple) or Photostim-Reinforced sessions (right, green). Data is replotted from (E) to facilitate comparisons between areas for each session type. A higher percentage of BLA neurons than dmPFC neurons were suppressed during both Photostim-Unreinforced and Photostim-Reinforced sessions (2-sample tests for equality of proportions: Photostim-Unreinforced: Χ2=13.11, df=1, ***p<0.001; Photostim-Reinforced: Χ2=35.61, df=1, ***p<0.001; all p values corrected for 4 multiple comparisons using Holm’s procedure). A lower percentage of BLA neurons than PFC neurons were facilitated during Photostim-Reinforced sessions (Χ2=9.21, df=1, **p=0.005).

Cholinergic afferents suppress basolateral amygdala output through muscarinic receptors and feed-forward inhibition

(A) Schematic of injection strategy to express ChR2 in cholinergic neurons of the basal forebrain (BF) and eYFP in GABAergic neurons of the BLA (BLAGABA), using conditional viral expression in ChAT::Cre x VGAT::Flpo mice (VGAT = vesicular GABAergic transporter), along with CTB-647 as a retrograde marker of neurons projecting to dmPFC.

(B) Confocal image of the BLA showing whole-cell patch-clamp recording arrangement in the BLA with optical stimulation of ChR2-expressing BF terminals. AP coordinate = -1.58.

(C) High magnification images of neurobiotin-filled recorded BLA neurons expressing CTB-647 (BLA-mPFC; upper panels) and eYFP (BLAGABA, lower panels).

(D) Passive membrane properties of BLA-mPFC and BLAGABA neurons. BLA-mPFC neurons had significantly greater capacitance (unpaired t-test: t42=11.90, ***p<0.001, n=20 BLA-mPFC, n=24 BLAGABA, from 9 mice), smaller membrane resistance (unpaired t-test: t42=6.326, ***p<0.001, n=20 BLA-mPFC, n=24 BLAGABA, from 9 mice), and more negative resting membrane potential (unpaired t-test: t29=2.857, **p=0.0078, n=13 BLA-mPFC, n=18 BLAGABA, from 8 mice) than BLAGABA neurons.

(E) Example trace and frequency histogram showing suppression of firing in BLA-mPFC neurons and facilitation of firing of BLAGABA neurons following optical stimulation of cholinergic terminals (470 nm light, 20 Hz; scale bars=20 mV, 5 s).

(F) Membrane potential of BLA-mPFC (upper traces) and BLAGABA neurons (lower traces) in response to 1 s 470 nm light delivered at 5, 10, and 20 Hz in current-clamp.

(G) At each stimulation frequency the amplitude of the fast excitatory postsynaptic potential (EPSP) was greater in BLAGABA neurons (green) compared with BLA-mPFC neurons (magenta; 2-way ANOVA, main effect of cell type: F1,40=95.59, ***p<0.001; n=7, 7, 15 BLA-mPFC neurons at 5, 10, 20 Hz, n=5, 5, 7 BLAGABA neurons at 5, 10, 20 Hz, from 9 mice), while the slower inhibitory postsynaptic current (IPSP) was greater in BLA-mPFC neurons (2-way ANOVA, main effect of cell type: F1,40=47.29, ***p<0.001; n=7, 7, 15 BLA-mPFC neurons at 5, 10, 20 Hz, n=5, 5, 7 BLAGABA neurons at 5, 10, 20 Hz, from 9 mice).

(H) Response of BLA-mPFC (upper traces) and BLAGABA neurons (lower traces) to a single 5 ms pulse of 470 nm light, with application of TTX/4AP to isolate monosynaptic currents.

(I) Following application of TTX/4AP, the EPSP was maintained in BLAGABA neurons (green; unpaired t-test: t15=0.367, p=0.719, n=8 (ACSF) and n=9 (TTX/4AP) BLAGABA cells from 4 mice), while the IPSP was maintained in BLA-mPFC neurons (magenta; unpaired t-test: t16=0.094, p=0.926, n=9 (ACSF) and n=9 (TTX/4AP) BLA-mPFC cells from 3 mice).

(J) Example traces showing inhibition of the IPSP in BLA-mPFC neurons (upper panels) by the muscarinic receptor antagonist scopolamine (10 µM) (dark gray), but not nicotinic antagonists (dihydro-ß-erythroidine 10 µM, methyllycaconitine 0.1 µM, mecamylamine 10 µM) (light gray), and inhibition of the EPSP in BLAGABA neurons (lower panels) by nicotinic receptor antagonists, but not muscarinic.

(K) Proposed circuit model showing BF inhibition of BLA output by ACh acting at nicotinic receptors on BLAGABA neurons and muscarinic receptors on projection neurons. Dashed lines represent local BLAGABA neuron synapses onto BLA projector neurons from prior literature (Lee and Kim, 2019; Woodruff and Sah, 2007).