The mouse dorsal peduncular cortex encodes fear memory

SUMMARY

The rodent medial prefrontal cortex (mPFC) is functionally organized across the dorsoventral axis, where dorsal and ventral subregions promote and suppress fear, respectively. As the ventral-most subregion, the dorsal peduncular cortex (DP) is hypothesized to function in fear suppression. However, this role has not been explicitly tested. Here, we demonstrate that the DP paradoxically functions as a fear-encoding brain region and plays a minimal role in fear suppression. By using multimodal analyses, we demonstrate that DP neurons exhibit fear-learning-related plasticity and acquire cue-associated activity across learning and memory retrieval and that DP neurons activated by fear memory acquisition are preferentially reactivated upon fear memory retrieval. Further, optogenetic activation and silencing of DP fear-related neural ensembles drive the promotion and suppression of freezing, respectively. Overall, our results suggest that the DP plays a role in fear memory encoding. Moreover, our findings redefine our understanding of the functional organization of the rodent mPFC.

In brief

Campos-Cardoso et al. employ a series of multimodal analyses to demonstrate that the mouse dorsal peduncular cortex encodes fear but not extinction memory. These findings redefine the principles of the functional organization of the rodent mPFC by providing insight into the role of this ventral-most mPFC subregion.

Graphical abstract

INTRODUCTION

Several studies have pointed toward the rodent medial prefrontal cortex (mPFC) as a locus important for both the promotion and suppression of fear. The mPFC is comprised of four anatomically and functionally distinct subregions including the anterior cingulate 1 cortex (Cg1), the prelimbic cortex (PL), the infralimbic cortex (IL), and the dorsal peduncular cortex (DP). Numerous studies have largely ascribed the dichotomous role of the mPFC to PL and IL subdivisions, which play roles in the promotion and suppression of fear, respectively.^1–5 Similar to the PL, the Cg1 has also been implicated in signaling fear.^6,7 In contrast, the specific role of the DP has largely been overlooked. The roles for the Cg1 and PL in the promotion of fear and the IL in the suppression of fear have led to the proposal that the mPFC exhibits a functional architecture that is organized across the dorsoventral axis. Therefore, due to its anatomical location ventral to the IL, some have hypothesized that the DP also functions to suppress fear.⁸ Despite this, the role that the DP might play in fear suppression or some other aspect of memory has largely not been tested (but see the recent study by Botterill et al.⁹).

In contrast to its proposed role in safety learning, several studies suggest a role for the DP in processing aversive stimuli. Following social defeat stress, the DP, along with the region immediately ventral, the dorsal tenia tecta (DTT), have been implicated in driving physiological sympathetic responses, including thermoregulation and cardiac responses.^10–12 Moreover, recent studies have also implicated the DP/DTT in driving anxiety-like behaviors⁹ and circuits of fight or flight.¹³ Given the central role for autonomic responses in fear, we hypothesized that the DP may be playing a role in fear memory processes rather than in extinction. Indeed, in humans, the ventromedial PFC (vmPFC) has been implicated in both the inhibition and expression of fear.^14–17 Nevertheless, despite this, the role for the DP in fear memory processing is largely unknown. Here, we use a combination of whole-cell brain slice electrophysiology, activity-dependent neural tagging, immunohistochemistry, fiber photometry calcium imaging, and in vivo optogenetics to probe the role of the DP in the regulation of fear memory. We surprisingly found that rather than suppressing fear, neural ensembles in the DP paradoxically function to encode fear memory and drive defensive behaviors. Our results provide a role for the DP in fear memory encoding and fundamentally upend our understanding of the functional organization of the rodent mPFC.

RESULTS

DP is activated in response to fear- but not extinction-related stimuli

To test whether the DP is involved in fear or extinction memory processes, we performed cFos immunohistochemical staining following exposure to three different behavioral paradigms. The first group of mice was subjected to tones only, followed by a tone re-exposure test 24 h later (Figures 1A and S1). The second group consisted of mice that were subjected to fear conditioning, which consisted of paired presentations of conditioned stimuli (CS; auditory tone) and unconditioned stimuli (US; foot shock), followed by a CS-evoked fear memory retrieval test 24 h later (Figures 1B and S1). The third group consisted of mice subjected to paired fear conditioning followed by 2 consecutive days of CS extinction training. Animals from this group were then subjected to an extinction memory retrieval test 24 h after the second day of extinction training (Figures 1C and S1). At 90 min after tone re-exposure or CS-evoked fear or extinction memory retrieval, mice were subjected to transcardial perfusion, and brains were processed for cFos immunohistochemistry. cFos⁺ puncta were then quantified across the Cg1, PL, IL, and DP subregions of the mPFC in mice from each behavioral group (Figure 1D). No differences in the density of cFos⁺ cells were observed across behavior groups in either the Cg1 or PL, consistent with our previous observations in the PL¹⁸ (Figure 1E). In the IL, we observed a significant increase in the density of cFos⁺ cells in the extinction group compared to tones-only or fear-conditioned mice (Figure 1E). This result is consistent with the role for the IL in extinction memory processing.¹⁹ Compared to tones-only mice, those that were subjected to conditioning and fear memory retrieval exhibited a significantly higher density of cFos⁺ cells in the DP (Figure 1E). In contrast, mice subjected to extinction training exhibited a significantly lower density of cFos⁺ cells in the DP compared to conditioned mice. These results suggest that rather than participating in extinction memory processing as was hypothesized previously, the DP may paradoxically be involved in fear-related memory processing.

Figure 1. Fear conditioning drives activation of the DP.

(A–C) Mice were subjected to (A) 6 tones (2 kHz, 20 s, 80 dB) followed by a tone re-exposure test consisting of 4 tones (2 kHz, 20 s, 80 dB) presentations 24 h later; (B) paired presentations of a CS (2 kHz, 20 s, 80 dB) and US (2 s, 0.7 mA) followed by a fear memory retrieval test consisting of 4 CS presentations (2 kHz, 20 s, 80 dB) 24 h later; and (C) paired presentations of a CS (2 kHz, 20 s, 80 dB) and US (2 s, 0.7 mA) and 2 consecutive days of extinction training consisting of 20 CS (2 kHz, 20 s, 80 dB) presentations and an extinction memory retrieval test consisting of 4 CS (2 kHz, 20 s, 80 dB) presentations. 90 min after tone re-exposure or fear or extinction memory retrieval, tissue was collected for cFos immunohistochemical analysis.

(D) Representative histological images of prefrontal brain sections obtained from mice subjected to tones only, fear conditioning, or fear extinction. Scale, 500 μm.

(E) Comparisons of cFos⁺ puncta density between tones-only (n = 5 mice, 10 slices), conditioning (n = 5 mice, 10 slices), and extinction (n = 7 mice, 14 slices) groups in the anterior cingulate cortex 1 (Cg1: F_2,14 = 4.08, p = 0.04, one-way ANOVA), prelimbic cortex (PL: F_2,14 = 2.23, p = 0.14, one-way ANOVA), infralimbic cortex (IL: F_2,14 = 19.34, p = 9.4 × 10⁻⁴, one-way ANOVA), and dorsal peduncular cortex (DP: X² = 9.39 [2], p = 0.009, Kruskal-Wallis ANOVA).

(F) Schematic cartoon of laminar demarcation across the mPFC.

(G) Comparisons of cFos⁺ puncta density between tones-only (n = 5 mice, 10 slices), conditioning (n = 5 mice, 10 slices), and extinction (n = 7 mice, 14 slices) groups in layer 2/3 of the Cg1 (F_2,14 = 1.49, p = 0.26, one-way ANOVA), PL (X² = 11.5 [2], p = 0.003, Kruskal-Wallis ANOVA), IL (F_2,14 = 17.9, p = 1.4 × 10⁻⁴, one-way ANOVA), and DP (F_2,14 = 19.6, p = 8.8 × 10⁻⁵, one-way ANOVA).

(H) Comparisons of cFos⁺ puncta density between tones-only (n = 5 mice, 10 slices), conditioning (n = 5 mice, 10 slices), and extinction (n = 7 mice, 14 slices) groups in layer 5/6 of the Cg1 (F_2,14 = 2.69, p = 0.1, one-way ANOVA), PL (F_2,14 = 1.81, p = 0.2, one-way ANOVA), IL (X² = 11.8 [2], p = 0.003, Kruskal-Wallis ANOVA), and DP (X² = 10.8 [2], p = 0.005, Kruskal-Wallis ANOVA).

*p < 0.05, **p < 0.01, and ***p < 0.001, Tukey’s or Dunn’s post hoc test. Boxplots represent the median (center line), mean (square), quartiles, and 10%–90% range (whiskers/error bars). Open circles represent data points for individual mice.

See also Figure S1.

Previous studies have highlighted that rather than exhibiting uniform activity, some mPFC subregions exhibit layer-specific neural activity. This pattern of activity could be related to the complex organization of inputs and outputs across mPFC layers. We therefore next subdivided cFos counts across layers in each subregion (Figure 1F). In layer 2/3, we observed that there was a significant increase in the density of cFos⁺ cells in the PL and DP following fear conditioning, as well as an increase in cFos in the IL following extinction training (Figure 1G). In layer 5/6, while we observed a similar increase in cFos in the IL and DP compared to layer 2/3, we did not observe either a fear- or extinction-training-dependent increase in cFos in layer 5/6 of the PL (Figure 1H). These results suggest that, in addition to differentially processing fear- and extinction-related stimuli, mPFC subregions exhibit differences in the laminar distribution of activated neurons.

DP principal neurons exhibit fear-learning-related plasticity

A hallmark of memory encoding is thought to be the expression of learning-related plasticity. We therefore sought to determine whether fear conditioning results in experience-dependent changes in DP neuron synaptic transmission. We performed whole-cell electrophysiological recordings in DP principal neurons (PNs) in acute brain slices prepared from wild-type mice. Prior to recording, mice were subjected to paired or unpaired fear conditioning (Figures 2A and S2). An additional group of mice were left naive and were only home-cage experienced. Acute brain slices were prepared 24 h after behavioral training, and spontaneous excitatory and inhibitory postsynaptic currents (EPSCs and IPSCs, respectively) were recorded from DP PNs. Mice that were subjected to paired fear conditioning exhibited a higher frequency of spontaneous EPSCs onto DP PNs as compared to unpaired and naive control mice (Figures 2B and 2C). No group-dependent differences in spontaneous EPSC amplitude or IPSC frequency or amplitude were observed (Figures 2D and 2E). Similar patterns of plasticity were observed in recordings of miniature EPSCs and IPSCs (Figure S3). Paired pulse EPSC recordings from DP PNs revealed that paired fear conditioning results in a decreased paired pulse ratio compared to naive and unpaired groups (Figure 2F). Together, these results suggest that paired fear conditioning drives the potentiation of excitatory transmission onto DP PNs.

Figure 2. Fear conditioning drives potentiated glutamatergic transmission onto DP PNs.

(A) Mice were either experienced to their home cage (naive) or subjected to unpaired or paired presentations of a CS (2 kHz, 20 s, 80 dB) and US (2 s, 1 mA). Whole-cell recordings were performed 24 h after training.

(B) Representative spontaneous excitatory postsynaptic current (EPSC) recordings in DP PNs from brain slices prepared from naive (8 slices from 4 mice), unpaired (6 slices from 4 mice), and paired (8 slices from 4 mice) mice.

(C) Quantification of the EPSC amplitude and interevent interval for naive, unpaired, and paired groups. Amplitude: F_2,35 = 1.67, p = 0.2, one-way ANOVA; naive = 14 cells; unpaired = 10 cells; paired = 14 cells. Interevent interval: X² = 14.32 (2), p = 7.77 × 10⁻⁴, Kruskal-Wallis ANOVA; naive = 14 cells; unpaired = 10 cells; paired = 14 cells.

(D) Representative spontaneous inhibitory postsynaptic current (IPSC) recordings in DP PNs from brain slices prepared from naive (8 slices from 4 mice), unpaired (6 slices from 4 mice), and paired (8 slices from 4 mice) mice.

(E) Quantification of the IPSC amplitude and interevent interval for naive, unpaired, and paired groups. Amplitude: F_2,32 = 0.084, p = 0.92, one-way ANOVA; naive = 13 cells; unpaired = 10 cells; paired = 12 cells. Interevent interval: X² = 2.2 (2), p = 0.33, Kruskal-Wallis ANOVA; naive = 13 cells; unpaired = 10 cells; paired = 12 cells.

(F) Left, cartoon schematic of recording configuration and representative excitatory paired pulse recordings (100 ms delay) from naive (7 slices from 4 mice), unpaired (8 slices from 3 mice), and paired (7 slices from 3 mice) mice. Right, quantification of paired pulse ratio: F_6,48 = 6.82, p = 2.83 × 10⁻⁵, interaction between training and delay, two-way repeated-measures ANOVA; naive = 10 cells; unpaired = 9 cells; paired = 10 cells.

*p < 0.05, **p < 0.01, and ***p < 0.001, Tukey’s or Dunn’s post hoc test. Boxplots represent the median (center line), mean (square), quartiles, and 10%–90% range (whiskers/error bars). Open circles represent data points for individual cells.

See also Figures S2 and S3.

DP, but not IL, neurons exhibit CS-related activity during fear conditioning and memory retrieval

The previous experiments led to the observation that paired fear conditioning leads to strengthening of excitatory inputs onto DP PNs. These results may point toward a role for the DP in cue-related fear memory processing. To test this possibility, we performed fiber photometry calcium imaging in the DP of mice undergoing fear conditioning and memory retrieval. Wild-type mice received unilateral infusions of GCaMP8f in the DP as well as angled imaging fiber implantation immediately dorsolateral to the DP (Figures 3A, 3B, and S4). We continuously recorded calcium-associated transients from DP neurons during a pre-conditioning CS exposure test, during paired fear conditioning, and during CS-evoked fear memory retrieval. During fear conditioning, we observed a trial-dependent increase in cue-related activity consistent with the ability of DP neurons to track and potentially encode information about the cue (Figure 3C). At 24 h after conditioning, mice were placed in a neutral context and subjected to a cue-evoked fear memory retrieval test. Consistent with the idea that the DP encodes cue-related information, we observed strong cue-elicited calcium signals. These signals were significantly larger than those recorded from the same mice in a pre-conditioning tone exposure test (Figure 3D). In contrast, we did not observe significant changes in calcium signals during spontaneous freezing bouts occurring during inter-CS intervals, suggesting that the DP does not generally signal freezing (Figure S5). Moreover, DP recordings from mice that were subjected to unpaired conditioning followed by context retrieval revealed that DP activity is not related to contextual memory (Figure S6). Together, these results suggest that the DP tracks and encodes cue-related information during fear conditioning and memory retrieval.

Figure 3. DP signals cue-related activity during fear memory acquisition and retrieval.

(A) Wild-type (WT) mice (n = 9 mice) received infusions of a vector encoding GCaMP8f along with imaging fiber implantation directed at the DP. CS calcium-associated activity was monitored before, during, and after conditioning, in addition to after extinction. Fear conditioning consisted of a baseline period (300 s) followed by 6 pairings of CS (2 kHz, 20 s, 80 dB) and US (2 s, 0.7 mA). Pre-conditioning tone exposure, fear memory retrieval, and extinction memory retrieval all consisted of a baseline period (100 s) followed by four CS presentations (2 kHz, 20 s, 80 dB).

(B) Representative histological image of GCaMP8f expression and fiber placement. Scale bar, 500 μm.

(C) Left, raw calcium traces averaged across CS trials for each behavior test. Right, heatmaps depicting individual calcium responses to each CS trial.

(D) Comparison of ΔF/F percentages between pre- and post-conditioning CS presentation, as well as CS-associated activity following extinction training. F_2,15 = 17.7, p = 1.13 × 10⁻⁴.

**p < 0.01, Tukey’s post hoc test. Boxplots represent the median (center line), mean (square), quartiles, and 10%–90% range (whiskers/error bars). Open circles represent data points for individual mice. Lines represent data points from the same animal.

See also Figures S4–S7.

We also tested whether DP neurons may participate in extinction memory processing, as was previously proposed for this region.⁸ To test this possibility, 24 h following the fear memory retrieval test, mice were subjected to 2 consecutive days of cue extinction training followed by an extinction memory retrieval test 24 h later. In contrast to the signals observed during fear memory retrieval, we did not observe significant CS-related activity during cue presentation in the extinction retrieval test (Figures 3C and 3D). Instead, the calcium-associated activity returned to baseline-like levels that were not different from those recorded in the pre-conditioning tone-exposure test. Taken together, these results further support the idea that the DP is involved in encoding cued fear but not extinction memory processes.

The results thus far indicate that the DP and IL likely play opposing roles in fear promotion and suppression, respectively. We therefore further tested this idea in vivo by performing fiber photometry recordings in the IL. GCaMP8f was infused unilaterally, and imaging fibers were implanted directed at the IL. Mice were then subjected to the same behavioral testing as in Figure 3 (Figure S7). During fear conditioning and fear memory retrieval, we did not observe significant cue-related activity; these signals were not significantly different from those recorded during a pre-conditioning CS exposure test. In contrast, following extinction, we observed a significant increase in IL CS-related activity (Figures S7E and S7F), consistent with our cFos experiments (Figure 1). These results indicate that the DP and IL exhibit orthogonal activity patterns in vivo and thus further suggest that they are functionally separate.

DP neurons are activated by fear conditioning and reactivated in response to CS-evoked fear memory retrieval

The previous experiment revealed that DP neurons are activated during both fear memory acquisition and memory retrieval. However, it remained unclear whether the same or different DP neurons are activated during learning and retrieval, a pattern of activity that is thought to be a hallmark of memory-encoding neurons. Toward this goal, we employed an activity-dependent neural tagging strategy that we previously used to tag fear-activated neurons.¹⁸ This strategy entails the use of a vector that allows for the activity-dependent expression of a 4-hydroxytamoxifen (4-OHT)-inducible Cre recombinase under the control of the synthetic enhanced synaptic activity-responsive element (ESARE-ERCreER) promoter.²⁰ A vector encoding ESARE-ERCreER was infused in the mPFC along with additional vectors encoding a Cre-dependent enhanced yellow fluorescent protein (DIO-eYFP) as well as an mCherry under the control of the human synapsin promoter (hSyn-mCherry) (Figure 4A). After 4 weeks, mice were subjected to either paired fear conditioning or tones only (Figure S8). Immediately after training, mice received intraperitoneal (i.p.) injections of either 4-OHT or vehicle. After 2 weeks, which is the amount of time required for recombination and protein expression, mice were subjected to a CS-evoked fear memory retrieval test or a tone re-exposure test in a neutral context, after which they were transcardially perfused. Brain tissue was then processed for cFos immunohistochemistry (Figure 4B).

Figure 4. Learning-activated DP neurons are reactivated following cue-evoked memory retrieval and bidirectionally control fear memory expression.

(A) WT mice received infusions of a cocktail of vectors encoding ESARE-ERCreER-PEST, a DIO-eYFP, and hSyn-mCherry into the DP. After 4 weeks, mice were subjected to tones only (2 kHz, 20 s, 80 dB) or paired presentations of CS (2 kHz, 20 s, 80 dB) and US (2 s, 0.7 mA) and immediately injected with vehicle or 4-OHT after training. Following 2 weeks, mice were subjected to a CS-evoked fear memory retrieval test or tone re-exposure in a neutral context. Mice were subjected to transcardial perfusion 90 min later, and tissue was used to stain against cFos.

(B) Representative histological images from brain sections of tone-exposed and conditioned mice injected with vehicle or 4-OHT. Scale bar, 500 μm. Inset and arrowheads indicate tagged neurons staining positive for cFos (red). Scale bar, 75 μm.

(C–F) Comparisons between tones-only vehicle (n = 4 mice, 8 slices), tones-only 4-OHT (n = 5 mice, 10 slices), conditioned vehicle (n = 4 mice, 8 slices), and conditioned 4-OHT (n = 6 mice, 12 slices) for (C) eYFP⁺ (X² = 13.75 [3], p = 0.003, Kruskal-Wallis ANOVA), (D) cFos⁺ (F_3,15 = 0.288, p = 0.83, one-way ANOVA), (E) eYFP⁺ and cFos⁺ (X² = 12.17 [3], p = 0.006, Kruskal-Wallis ANOVA), and (F) eYFP⁺ and cFos⁺/eYFP⁺ (X² = 15.31 [3], p = 0.0016, Kruskal-Wallis ANOVA).

(G) WT mice received infusions of vectors encoding ESARE-ERCreER-PEST, DIO-ChR2-eYFP or FLEX-Arch-GFP, and hSyn-mCherry along with implantation of optical fibers directed at the DP. Mice were subjected to paired fear conditioning and immediately injected with either vehicle or 4-OHT. After 2 weeks, mice were subjected to a CS-evoked fear memory retrieval test in a neutral context with and without optogenetic manipulation.

(H) Representative histological images from mice expressing Arch-GFP or ChR2-eYFP and receiving injections of vehicle or 4-OHT following fear conditioning. Scale, 400 μm.

(I) Freezing quantified during CS (2 kHz, 20 s, 80 dB) and light (473 nm, 20 Hz, 5 ms pulses, 20 s epochs) presentation for animals expressing ChR2 in fear-tagged DP neurons. Vehicle: F_3,15 = 22.14, p = 9.18 × 10⁻⁶, one-way repeated-measures ANOVA; n = 6 mice. 4-OHT: F_3,15 = 73.39, p = 3.42 × 10⁻⁹, one-way repeated-measures ANOVA; n = 6 mice.

(J) Freezing quantified during CS (2 kHz, 20 s, 80 dB) and light (532 nm, solid light, 20 s epochs) presentation for animals expressing Arch in fear-tagged DP neurons. Vehicle: F_3,15 = 97.14, p = 4.73 × 10⁻¹°, one-way repeated-measures ANOVA; n = 6 mice. 4-OHT: F_3,15 = 53.29, p = 3.12 × 10⁻⁸, one-way repeated-measures ANOVA; n = 6 mice.

*p < 0.05, **p < 0.01, and ***p < 0.001, Tukey’s post hoc test. Boxplots represent the median (center line), mean (square), quartiles, and 10%−90% range (whiskers/error bars). Open circles represent data points for individual mice. Lines represent data points from the same animal.

See also Figures S8–S13.

We observed that there were significantly more eYFP⁺ neurons in the DP of 4-OHT-injected conditioned mice compared to vehicle-injected conditioned mice or 4-OHT- or vehicle-injected tone-exposed mice (Figure 4C). While cFos levels were comparable across groups (Figure 4D), there was an enrichment in the number of tagged neurons that were positive for cFos in the 4-OHT-injected conditioned group compared to those injected with vehicle or vehicle- and 4-OHT-injected tones-only mice (Figure 4E). Comparisons of the proportion of cFos⁺ tagged neurons revealed that 4-OHT-injected conditioned mice had significantly higher levels (~40%) of fear retrieval-reactivated learning-tagged DP neurons compared to any other group (Figure 4F). These results reveal that ensembles of DP neurons exhibit activity patterns consistent with a role in fear memory encoding.

The results from our cFos (Figure 1) and fiber photometry experiments (Figure 3) suggested that DP neurons are activated in response to fear conditioning and cue-evoked fear memory retrieval but not following extinction. To further test whether the DP may be playing a role in extinction, we performed activity-dependent neural tagging following extinction training (Figure S9). Following vector cocktail infusion in the IL and DP, mice were subjected to paired fear conditioning and 3 consecutive days of cue extinction training. Following the third extinction trial, mice received i.p. injections of either vehicle or 4-OHT. Two weeks later, brains were processed for immunohistochemistry. Consistent with our previous results, we only observed a small fraction of the total number of tagged neurons in the DP, and there was no difference in the number of tagged neurons in the DP in the vehicle compared to the 4-OHT group. Importantly, a vast majority of tagged neurons instead resided in the IL, which has been implicated in extinction. Consistent with its role in extinction, there were significantly more tagged neurons in the IL of 4-OHT-injected mice compared to those injected with vehicle. These results are also consistent with cFos immunohistochemistry results suggesting that cells in the IL, but not the DP, are activated following extinction training (Figures 1G and 1H). Overall, these results provide additional evidence suggesting that the DP is involved in fear- but not extinction-related processes.

We additionally performed neural tagging in the DP following unpaired conditioning, in which mice were subjected to both the CS and US but in an explicitly unpaired manner (Figure S10). Immediately after US delivery, mice received i.p. injections of either vehicle or 4-OHT. Two weeks after training, mice were subjected to a CS-evoked fear memory retrieval test, and cFos immunohistochemistry was performed. Comparisons of the proportion of cFos⁺ eYFP neurons in vehicle- versus 4-OHT-injected mice revealed no significant group differences. These results further support the idea that cued fear learning-tagged DP neurons represent a distinct learning-related population rather than an ensemble activated by stress, contextual elements, or some other stimulus.

Activity of fear-tagged DP neurons bidirectionally controls fear memory expression

Given that learning-activated DP neurons are reactivated following CS-evoked fear memory retrieval, we next sought to test whether the activity of fear-tagged DP neurons can modulate fear memory expression. Toward this goal, we infused vectors encoding ESARE-ERCreER, Cre-dependent eYFP-tagged channelrhodopsin (DIO-ChR2-eYFP) or green fluorescent protein-tagged archaerhodopsin (FLEX-Arch-GFP), and hSyn-mCherry bilaterally into the DP of wild-type mice (Figures 4G and 4H). Following vector infusion, optic fibers were implanted bilaterally and directed at the DP (Figure S11). After 4 weeks, mice were subjected to paired fear conditioning and immediately injected with vehicle or 4-OHT. After 2 weeks, mice were subjected to a fear memory retrieval test in a neutral context, and laser stimulation-dependent alterations in freezing were assessed. In 4-OHT-injected mice expressing ChR2, optogenetic activation (473 nm, 20 Hz, 5 ms pulses, 20 s epochs) of fear-related DP neurons led to a significant increase in freezing over baseline (Figure 4I). Conversely, optogenetic silencing (532 nm, solid light, 20 s epochs) of fear-tagged DP neurons resulted in a reduction in CS-evoked freezing compared to CS presentation in the absence of optogenetic silencing (Figure 4J). An open field test during which optogenetic activation and silencing were performed revealed that there was no significant light-dependent effect on either locomotion or anxiety-like behaviors (Figure S12). Importantly, recordings from brain slices containing fear-tagged DP neurons expressing Arch-GFP or ChR2-eYFP revealed that the in vivo optogenetic manipulation procedures used result in reliable neural silencing or activation, respectively (Figure S13). Overall, these results suggest that the activity of fear-tagged DP neurons bidirectionally controls fear memory expression.

DISCUSSION

Here, by using multimodal analyses, we reveal that the DP is involved in fear memory encoding and expression. We found that auditory fear conditioning, but not extinction training, drives activity of DP neurons. Moreover, fear conditioning potentiates glutamatergic transmission onto DP PNs compared to naive and unpaired littermate controls. By using fiber photometry recordings, we observed that DP neurons acquire CS-related activity during memory acquisition and that this activity persists during subsequent fear memory retrieval but is lost after extinction training. Activity-dependent neural tagging following memory acquisition combined with cFos immunohistochemistry after retrieval indicated that ensembles of DP neurons are activated by learning and reactivated following memory retrieval. Finally, optogenetic activation of learning-tagged DP neurons induces fear memory expression, whereas silencing significantly reduces cue-evoked freezing. Taken together, our results point toward a role for the DP in fear memory processing–an observation that is at odds with its proposed role in extinction.⁸

The rodent vmPFC is most well known for its role in memory extinction. In particular, lesion experiments,^21–23 in vivo electrophysiological recordings and electrical stimulation,^24,25 pharmacological microinfusion experiments,²⁶ and in vivo optogenetic manipulations⁵ have pointed toward a role for the vmPFC in extinction memory processing, with the vast majority of this role attributed to the IL. However, some additional studies have provided conflicting results regarding this role. For example, in vivo pharmacological inhibition of the vmPFC has paradoxically been shown to reduce the expression of conditioned fear.²⁷ Moreover, additional lesion studies found that vmPFC lesions had no effect on extinction memory acquisition or retrieval.^28,29 These conflicting findings are also reflected in the human literature, where studies have implicated the vmPFC in extinction,^1,14,15 while other studies have suggested that vmPFC activity contributes to the pathology of post-traumatic stress disorder.^16,17 While the conflicting results in rodents can potentially be explained by compensatory mechanisms or differences in experimental execution, one additional possibility may be due to differences in the lesion/infusion locus within the vmPFC. The results of our current study that implicate the DP in fear, but not extinction, memory processes may potentially account for the conflicting findings in the field.

While our findings have provided numerous lines of evidence suggesting that the DP plays a role in fear memory encoding and retrieval but not extinction, a recent study by Botterill and colleagues reported that the DP/DTT seems play a role in fear extinction rather than fear encoding and retrieval.⁹ These differences may relate to which specific DP outputs were manipulated, the method of manipulation, and anatomical differences in viral targeting. Moreover, an additional study revealed sexual dimorphic activity profiles in the DP following auditory fear conditioning, where female mice exhibited significantly higher levels of conditioning-induced cFos expression compared to males.³⁰ While most of our experiments are not sufficiently powered to perform rigorous analyses of sex differences, our results do not point toward any gross differences in DP activity in male and female mice. Whether conditioning engages distinct cell types or circuits within the DP of male and female mice remains an important area for future investigation.

During the revision of this manuscript, an additional study was published that points toward a role for DP excitatory projections to the central amygdala in the control of fight or flight.¹³ The authors observed that neurons in layer 2/3 project to the central amygdala, whereas those in layer 5/6 project to the dorsal medial hypothalamus, a circuit implicated in mediating sympathetic responses to psychosocial stress.¹⁰ As we observed increased cFos expression across DP cortical layers (Figure 1), the population we examined is likely comprised of neurons projecting to both of these downstream targets. Indeed, sympathetic responses and circuits of fight and flight are both central in fear memory encoding processes. Future studies examining the specific projection profiles of fear-tagged DP neurons will be needed to determine other targets of this population.

As the role of the DP in fear memory encoding and retrieval as well as in processing other emotional stimuli continues to be investigated, additional questions remain to be answered in future studies. One such question is why might both the PL and the DP function to encode fear memory? Are each of these brain regions encoding distinct aspects of the fear memory? Similar to the DP, the PL is activated both during fear memory encoding and retrieval,^31,32 and its activity drives the expression of defensive behaviors.^25,26,33,34 Despite their similar functions in promoting fear, there are also some notable differences in the fear memory processes in the DP and PL. In contrast to the PL, which is involved in both cued and contextual fear,^21,33,35 DP neurons signal cued, but not contextual, fear memory (Figures S6 and S10). Moreover, PL neurons exhibit tone responses in naive mice,^32,36 whereas the same responses do not exist in the DP, at least at the level of our analyses. Insights into the circuits that underlie these differences may be gained through the systematic investigation of projection pathways that differentially engage the PL versus the DP. Indeed, functional in vivo oscillation measurements across mPFC subregions in anesthetized rats have revealed vast differences in the PL and DP, suggesting that these regions likely have unique functions and/or connectivity patterns.³⁷ Moreover, an anatomical analysis reveals bidirectional connectivity between the PL and DP,³⁸ suggesting that they have the capacity to convey information to each other during memory encoding. Therefore, it will be important for future studies to reveal the fundamental connectivity of long-range inputs to the DP to gain insight into these questions.

Our results suggest that a relatively sparse ensemble of DP neurons functions to encode fear memory. While we have characterized the functional contributions of this ensemble, several questions remain. First, it remains unknown what neuron type(s) comprise this ensemble and how these distinct cell types contribute to memory encoding. Our previous work revealed that somatostatin-expressing interneurons (SST-INs) comprise a relatively large part of the fear-activated ensemble in the PL.¹⁸ Whether these or another subset of interneurons are recruited into the DP ensemble remains unknown. Additionally, future studies should focus on revealing the local microcircuit connectivity and plasticity mechanisms that support the recruitment of the DP ensemble. Indeed, our previous work in the PL revealed that there is extensive rebalancing of microcircuit activity that resulted in SST-IN-mediated inhibition of parvalbumin-expressing IN activity, driving the disinhibition of PNs.³⁹ Whether this circuit organization and plasticity mechanism is conserved in the DP is unknown and should be studied in future work.

Overall, our results provide critical insights into fear memory encoding and expression in the mouse DP. These findings open several areas of investigation and fundamentally redefine our understanding of the functional landscape within the rodent mPFC.

Limitations of the study

Although our study provides several lines of evidence suggesting that the DP encodes fear memory, there are some limitations that are important to consider. First, although we performed electrophysiological analyses that revealed that DP PNs undergo fear-related plasticity, we did not identify and test the causal contribution of potential sources of the plasticity or test whether silencing identified pathways blocks plasticity and fear memory. This will be important for identifying how the DP is engaged during fear conditioning and how it is incorporated into our existing understanding of fear circuitry. A second limitation lies in our fiber photometry experiments. Since fiber photometry only reports on bulk calcium-associated neural activity, it is possible that a small population of extinction-related DP neurons exists but that their activity is masked by a larger or more active population of fear-encoding neurons. Since this population, if it exists, would likely be small, it will be important to utilize experimental approaches that enable single-cell resolution, such as silicon probe or microendoscopic recordings, to examine contributions of this potential population. Importantly, however, these limitations do not impact our interpretation that the DP encodes fear memory.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Kirstie A. Cummings (kac3@uab.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• All data reported in this paper will be shared by the lead contact upon request.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Animals

The use of animals in all experimental procedures was approved by the Institutional Animal Care and Use Committee at the Heersink School of Medicine at the University of Alabama at Birmingham. Both male and female mice at the age of postnatal day 42 (P42) were used for stereotaxic surgery and behavioral experiments were then performed at P60-90. Mice used in experiments were the C57Bl/6J genotype (stock no. 000664). Animals were housed in groups of 2–6 with a 12 h light-dark cycle and ad libitum access to food and water. Equal numbers of males and females were used for all experimental groups.

METHOD DETAILS

Behavioral training paradigms

Paired auditory fear conditioning was performed by subjecting mice to 6 pairings of a pure auditory tone (CS; 80 dB, 2 kHz, 20 s) with a co-terminating scrambled foot shock (US; 0.7–1.0 mA, 2 s). Unpaired conditioned mice were subjected to 6 CS presentations after which they were returned to their home cage for 15 min and then placed back into the arena during which they received 6 US presentations. Naive mice only experienced the home cage. Tones-exposed mice were placed in the conditioning arena and subjected to 6 CS tones in the absence of the US. For extinction training, 24 h after paired fear conditioning, mice were placed in a neutral context (context B) consisting of distinct visual, olfactory, and tactile cues and subjected to 20 CS tones. An additional extinction session was performed the next day for a total of 2 consecutive days. For either fear or extinction memory retrieval in cFos, electrophysiological, and calcium imaging experiments (Figures 1, 2, and 3), mice were placed in context B and subjected to presentation of 4 CS tones at 24 h after either fear conditioning or the second day of extinction training, respectively. Tone re-exposure was performed in context B by presenting animals 4 CS tones at 24 h after tone exposure. For fear memory retrieval or tones re-exposure following neural tagging (Figure 4), mice were placed in context B and subjected to presentation of 4 CS tones at 2 weeks (due to limitations of vector expression timeline) after fear conditioning or tone exposure, respectively. Littermates were used for each experiment whenever possible and were randomly distributed across behavior groups in an interleaved fashion.

Immunohistochemistry

Immunohistochemistry was performed for cFos and/or GFP. 90 min after the last behavioral test, after deep isoflurane anesthesia was induced, mice were transcardially perfused with phosphate buffered saline (PBS; pH 7.4) followed by 4% paraformaldehyde in PBS (PFA; pH 7.4) before removal of the brains. Brains were then fixed in 4% PFA overnight before sectioning 50 μm coronal slices using a Leica VT1000S vibratome (Leica Biosystems, Deer Park, IL, USA).

Brain sections were blocked with 5% normal goat serum in PBS +0.3% Tween 20 for 1 h at room temperature. Afterward, slices were incubated at 4°C overnight with rabbit anti-cFos primary antibody (1:1000; Millipore ABE457). The following day, slices were incubated for 2 h with a secondary goat anti-rabbit antibody conjugated to Alexa 647 (1:500; Jackson Immunoresearch, 111-605-045) with 5% normal goat serum in PBS +0.3% Tween 20.

For experiments requiring staining against GFP, slices were then incubated overnight with a chicken anti-GFP primary antibody (1:500; Millipore AB16901) with 5% normal donkey serum in PBS +0.3% Tween 20. The next day, slices were incubated with a secondary donkey anti-chicken antibody conjugated to fluorescein isothiocyanate (FITC) (1:500; Jackson Immunoresearch, 703-095-155) for 2 h at room temperature.

Following immunohistochemical staining, slices were mounted with ProLong Gold antifade reagent with DAPI (Invitrogen – ThermoFisher Scientific, Eugene, OR, USA). Imaging was performed using an upright Echo Revolution hybrid microscope (BICO, San Diego, CA, USA) or a Nikon Eclipse Ni-E microscope (Nikon Instruments, Melville, NY, USA). Cell counting for cFos or/and eYFP labeled cells was performed with the Cell Counter plug-in through the ImageJ program (NIH, Bethesda, MD, USA). Quantification was performed for either or both cFos⁺ and eYFP⁺ cells, as well as for overlapping cFos⁺/eYFP⁺ cells by experimenters blind to experimental conditions and groups.

Anatomical demarcations were determined by creating overlays aligned to the Paxinos and Franklin’s The Mouse Brain In Stereotaxic Coordinates atlas⁴⁴ in Photoshop. These overlays were then imported into ImageJ (NIH, Bethesda, MD, USA) for analysis. Overlays were created by experimenters blind to experimental conditions, and cFos expression, tagged neuron distribution, etc. were not used as a factor in determining anatomical borders. In an effort to enhance robustness and reliability of data, several blinded experimenters performed these demarcations across multiple cohorts of mice.

Whole-cell brain slice electrophysiology

Mice were subjected to paired fear conditioning, unpaired fear conditioning, or were naive. 24 h after training, mice were used for electrophysiology recordings.

Mice were deeply anesthetized by way of isoflurane administration prior to decapitation. Brain slices containing the mPFC were sectioned at 300 μm on a Leica VT1200S vibratome in chilled carbogen-bubbled low sodium sucrose cutting solution containing (in mM) 210 sucrose, 11 glucose, 26.2 NaHCO₃, 2.5 KCl, 1 NaH₂PO₄, 4 MgCl₂, 0.5 CaCl₂, and 0.5 ascorbate. Slices were then recovered for 45 min in warmed (32°C) carbogen-bubbled normal artificial cerebrospinal fluid containing (in mM) 119 NaCl, 26.2 NaHCO₃, 11 glucose, 1 NaH₂PO₄, 2 MgCl₂, 2.5 KCl, 2 CaCl₂. Slices were kept at room temperature. Whole-cell electrodes (3–4 MΩ) were fabricated from fire-polished borosilicate capillaries and filled with internal solution containing (in mM) 120 Cs-methanesulfonate, 10 HEPES, 10 Na-phosphocreatine, 1 QX-314, 8 NaCl, 0.5 EGTA, 4 Mg-ATP, 0.4 Na-ATP (pH 7.25; 290–300 mOsmol). Cells were visualized on an upright microscope equipped with differential interference contrast optics. Principal neurons were identified visually (large pyramidal somas) and electrophysiologically (low membrane resistance (<75 MΩ) and high capacitance (>100 pF). Cells were sampled across superficial and deep layers and pooled.

Spontaneous excitatory (EPSC) and inhibitory (IPSC) postsynaptic currents were recorded in standard artificial cerebrospinal fluid (ACSF) by clamping cells at −60 mV or 0 mV, respectively. Miniature EPSCs and IPSCs were recorded in standard ACSF with the addition of 1 μM tetrodotoxin (TTX). Each EPSC and IPSC recording lasted at least 5 min. Paired pulse measurements were performed with a bipolar stimulating electrode placed adjacent to recorded cells. Averages of 7–10 sweeps per cell per delay were used for calculating paired pulse ratio. Paired pulse ratio was calculated by dividing the amplitude of the second current by the amplitude of the first. Data were sampled at 10 kHz and low-pass-filtered at 3 kHz using Multiclamp 700B amplifier (Molecular Devices, version 2.2.2.2). Data were analyzed in Easy Electrophysiology (version 2.6.1) by experimenters blind to experimental groups. Access resistance and leak current were monitored throughout each recording. Recordings that yielded unstable currents (>150 pA leak; >20% change in access resistance) or which did not meet cell type characterization requirements were excluded. Cell exclusion was rare and occurred <2% of the time.

Stereotaxic vector infusion and fiber implantation

AAV2/8-ESARE-ERCreER-PEST was received as a gift from Dr. Haruhiko Bito (U. Tokyo) and custom packaged at Boston Children’s Hospital Vector Core. AAV1-DIO-eYFP (Cat. No. 27056), AAV8-hSyn-mCherry (Cat. No. 114472), AAV1-hSyn-GCaMP8f (Cat. No. 162376), AAV1-DIO-hChR2(H134R)-eYFP-WPRE (Cat. No. 20298), and AAV1-FLEX-Arch-GFP (Cat. No. 22222) were purchased from Addgene. For all experiments, vector infusions occurred at P42. For activity-dependent tagging quantification experiments, vectors were infused bilaterally in the DP (200 μL; A/P: 1.55; M/L: +/− 0.25; D/V: −3.7) or in DP and IL (400 μL; A/P: 1.55; M/L: +/− 0.25; D/V: −3.7-3.5). Vectors were mixed at a 2 (ESARE-ERCreER-PEST) to 7 (DIO-eYFP) to 1 (hSyn-mCherry) ratio just prior to stereotaxic infusion. For fiber photometry experiments and in vivo optogenetics, vectors were infused unilaterally (fiber photometry) or bilaterally (in vivo optogenetics) into the DP (A/P: 1.55; M/L: + and/or − 2.3; D/V: −4.1; 32° ⦠) or IL (A/P: 1.55; M/L: + and/or − 1.25; D/V: −2.7; 15° ⦠). For fiber photometry experiments, an imaging fiber (400 μm diameter, 0.48 NA, Doric Lenses) was implanted immediately above the DP (A/P: 1.6; M/L: +/− 1.65; D/V: −3.9-4.0; 32° ⦠) or IL (A/P: 1.55; M/L: + and/or − 1.25; D/V: −2.6; 15° ⦠). For in vivo optogenetic experiments, optic fibers (200 μm diameter, 0.22 NA, Doric Lenses) were implanted directed at DP (A/P: +1.6; M/L: +/− 1.65; D/V: −3.8; 32° ⦠). Vector incubation times were previously determined^18,39 or were determined empirically for this study.

Activity-dependent neural tagging and cFos immunofluorescence

At 3.5 weeks post vector cocktail infusion, mice were handled for three days consecutively prior to behavior. Mice were subjected either to auditory paired fear conditioning, tones only, or unpaired fear conditioning. For mice used to tag extinction-related mPFC neural ensembles, animals were subjected to extinction training for 3 consecutive days starting at 24 h after fear conditioning. Neural tagging was performed through an intraperitoneal (IP) injection of 4-OHT (10 mg/kg) or vehicle solution immediately after the completion of the last day of training, as was done previously.¹⁸ Two weeks after training and activity-dependent tagging, mice were subjected to a fear memory retrieval test in a neutral context (context B). Mice were sacrificed 90 min post-retrieval for cFos and GFP immunohistochemistry as described above.

Fiber photometry recording and analysis

Fiber photometry recordings were performed by using a complete system from Tucker-Davis Technologies, as we have done previously.³⁹ At 4 weeks after vector infusion and imaging fiber implantation (Doric Lenses), mice were habituated to handling and patch cord tethering for 3 consecutive days. Mice were placed in a neutral arena with white floor and rounded wall inserts (context B) and subjected to 4 CS presentations (CS: 80 dB, 2 kHz, 20 s). 24 h later, mice were subjected to either paired or unpaired fear conditioning in an arena with a metal grid floor and square metal walls (context A). Mice were then subjected to a fear memory retrieval test in context B at 24 h post-conditioning. 24 h later, mice were subjected to two consecutive days of cue extinction training in context B. At 24 h post extinction training, mice were subjected to an extinction memory retrieval test in context B. Calcium signals were sampled at 6 kHz and continuously recorded throughout the duration of each behavioral test. Transistor-transistor logic (TTL) signals were generated through MedAssociates software and were used to produce timestamps for the start and stop of each session as well as for CS presentation epochs. Prior to the start of each session, calcium signals were recorded for a duration of 2 min to ensure stable recordings.

Data were processed and analyzed using custom-written code in MATLAB (R2019b version 9.7.0.1190202) from the Tucker-Davis Technologies website as we have done previously.³⁹ Briefly, CS-associated activity changes were calculated by normalizing the calcium signal during the 20 s CS presentation to a 20 s baseline period occurring immediately prior to CS onset. The percentage change in fluorescence (%ΔF/F) was calculated by subtracting the peak baseline signal from the peak signal during CS presentation and dividing by the peak baseline signal (%ΔF/F = (F_peakCS - F_peakBL)/F_peakCS)). Reported values represent within-animal changes. At the conclusion of experiments, mice were subjected to transcardial perfusion with PBS followed by 4% paraformaldehyde and imaging fiber placement and GCaMP8f expression were assessed. Mice with incorrect fiber placements and/or GCaMP8f expression were excluded from analyses.

In vivo optogenetics

At 3.5 weeks after vector infusion and optic fiber implantation (Doric lenses), mice were habituated to handling and tethering to patch cords for 3 consecutive days. Mice were subjected to paired fear conditioning and then received intraperitoneal injections of either 4-OHT (10 mg/kg) or vehicle immediately after training and returned to their home cage. After two weeks, mice were subjected to a CS-evoked fear memory retrieval test in a neutral context (context B). Optogenetic activation (473 nm, 20 Hz, 5 ms pulses, 20 s epochs) or silencing (532 nm, solid light, 20 s epochs) was performed in animals expressing ChR2 or Arch, respectively. For silencing experiments, light was ramped down over the course of 5 sec to avoid propagation of rebound action potentials.

Light intensities delivered to the brain were measured and calculated prior to each experimental day and ranged from 6 to 8 mW. Optogenetic manipulations were performed alone or in combination with the CS in an alternating manner (e.g., Baseline: light-on, light-off; CS: CS + light, CS alone, etc) and the order of light presentation was counterbalanced. Each epoch (baseline, light-on; baseline, light-off; CS, light-on; CS, light-off) was repeated twice per animal and the average was reported. Videos were scored by experimenters blind to experimental conditions and groups. At the end of the experiment, mice were subjected to transcardial perfusion and brains were sectioned for immunohistochemical analysis to confirm correct optic fiber placement and vector expression. Mice with incorrect fiber placement and/or viral expression were removed from analyses.

Open field test

Mice from the in vivo optogenetics experiments (Figure 4) were also subjected to optogenetic stimulation in an open field. The test was 10 min in duration, and consisted of alternating and counterbalanced epochs of 1 min light-on and 1 min light-off periods. The open field measured 40 cm length x 40 cm width x 40 cm height and tracked movement via infrared beam breaks (MedAssociates, Fairfax, VT). Average distance moved and average percent time spent in the margin space was calculated and reported for each mouse for the light-on/light-off epochs.

Electrophysiological validation of in vivo optogenetic manipulations

To validate the modes of optogenetic manipulation of fear-related DP neurons used in vivo (Figure 4), mice that received infusions of vectors encoding ESARE-ERCreER and either DIO-ChR2-eYFP or FLEX-Arch-GFP were subjected to paired fear conditioning and intraperitoneal injection of 4-OHT (10 mg/kg). After 2 weeks, mice were used to prepare brain slices, as described above. DP neurons expressing ChR2 or Arch were targeted in slice via fluorescence. For DP neurons expressing Arch, after achieving whole-cell configuration, slow current was injected to induce spontaneous action potential firing. Green light (532 nm; solid light; 8 mW; 20 s epochs) was delivered via LED-coupled objectives for a total of 4–5 times per cell. The average firing frequency and magnitude of hyperpolarization was calculated during light-off and light-on epochs. For DP neurons expressing ChR2, after achieving the whole-cell configuration, pulses of blue light (473 nm; 20 Hz, 5 ms pulses, 20 s epochs) were delivered via LED-coupled objectives and action potential firing was recorded.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical testing and graph plotting were performed using OriginPro 2019 version 9.6.0.172. Levine’s and Shapiro-Wilk tests were performed prior to parametric statistical testing to test whether data exhibited homogeneity of variance and normal distribution, respectively. If either or both conditions were not met, data were subjected to non-parametric statistical analyses. All statistical details, including statistical tests used, the exact value of each n, and what n represents can be found in each Figure or Supplementary Figure legend. In all Figures and Supplementary Figures, boxplots represent the median (center line), mean (square), quartiles, and 10–90% range (whiskers/error bars). Open circles represent individual data points, which is defined in each legend (e.g., mice, cells, etc.). Tukey’s (for parametric tests) or Dunn’s (for non-parametric tests) post hoc tests were performed where indicated in each figure legend. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-cFos	Millipore	ABE457; RRID:AB_2631318
Chicken anti-GFP	Millipore	AB16901; RRID:AB_90890
Goat anti-rabbit Alexa 647	Jackson Immunoresearch	111-605-045; RRID:AB_2338075
Donkey anti-chicken FITC	Jackson Immunoresearch	703-095-155; RRID:AB_2340356
Bacterial and virus strains
AAV2/8-ESARE-ERCreER-PEST	Boston Children’s Hospital Vector Core	N/A
AAV1-DIO-eYFP	Fenno et al.40	Addgene 27056
AAV1-DIO-ChR2(H134R)-eYFP	pAAV-EF1a-double floxed-hChR2 (H134R)-EYFP-WPRE-HGHpA was a gift from Karl Deisseroth	Addgene 20298 (Addgene viral prep # 20298-AAV1; http://n2t.net/addgene:20298; RRID:Addgene_20298)
AAV1-FLEX-Arch-GFP	Han et al.41	Addgene 22222
AAV8-hSyn-mCherry	pAAV-hSyn-mCherry was a gift from Karl Deisseroth	Addgene 114472 (Addgene viral prep # 114472-AAV1; http://n2t.net/addgene:114472; RRID:Addgene_114472)
AAV1-hSyn-GCaMP8f	Zhang et al.42	Addgene 162376
Chemicals, peptides, and recombinant proteins
4-hydroxytamoxifen (70% Z-isomer)	Sigma	H6278
Tetrodotoxin citrate	Abcam	Ab120055
Invitrogen ProLong Gold antifade reagent with DAPI	Fisher Scientific	P36931
Experimental models: Organisms/strains
Mouse: C57Bl/6J	Jackson Laboratory	000664
Recombinant DNA
pFBAAV-ESARE-ERCreER-PEST	Gift from Dr. Haruhiko Bito	N/A
Software and algorithms
MedAssociates Video Freeze 3.02.00.00	MedAssociates	https://www.med-associates.com/product/video-fear-conditioning/
Ethovision XT	Noldous	https://www.noldus.com/ethovision
Clampex 11.0.3.03	Molecular Devices	https://www.moleculardevices.com/products/axon-patch-clamp-system/acquisition-and-analysis-software/pclamp-software-suite
NIH ImageJ 1.52a	Schneider et al.43	https://imagej.nih.gov/ij/index.html
VLC media player 2.1.3 Rincewind	VideoLAN	https://www.videolan.org/
Easy Electrophysiology v2.6.1	Easy Electrophysiology	https://www.easyelectrophysiology.com/
MATLAB R2019b 9.7.0.1190202	MathWorks	https://www.mathworks.com/products/matlab.html
Custom-written code for fiber photometry data analysis	Tucker-Davis Technologies	https://www.tdt.com/docs/sdk/offline-data-analysis/offline-data-matlab/
Photoshop 24.7.2	Adobe	https://www.adobe.com/products/photoshop/
Nikon Elements 5.42.03 (Build 1812)	Nikon	https://www.microscope.healthcare.nikon.com/products/software/nis-elements
Echo Revolution 1.0.26.3	BICO	https://discover-echo.com/revolution/
OriginPro 9.6.0.172	OriginLab	https://www.originlab.com/origin
BioRender	BioRender	https://www.biorender.com/
Other
Leica VT1000S vibratome	Leica Biosystems	https://www2.leicabiosystems.com/na/vibratomes
Echo Revolution hybrid microscope	BICO	https://discover-echo.com/revolution/
Nikon Eclipse Ni-E microscope	Nikon	https://www.microscope.healthcare.nikon.com/products/upright-microscopes/eclipse-ni-series
Leica VT1200S vibratome	Leica Biosystems	https://www2.leicabiosystems.com/na/vibratomes
Multiclamp 700B amplifier 2.2.2.2	Molecular Devices	https://www.moleculardevices.com/products/axon-patch-clamp-system/amplifiers/axon-instruments-patch-clamp-amplifiers
Imaging fibers (0.48 NA, 400 μm diameter)	Doric Lenses	https://neuro.doriclenses.com/pages/fiber-photometry
Optic fibers (0.22 NA, 200 μm diameter)	Doric Lenses	https://neuro.doriclenses.com/collections/fiber-optic-cannulas/products/mono-fiber-optic-cannulas
Stereotaxic instruments	Stoelting	https://stoeltingco.com/Neuroscience/Digital-Rat-and-Mouse-Stereotaxic-Instruments
Fiber photometry recording system	Tucker-Davis Technologies	https://www.tdt.com/system/fiber-photometry-system/

Highlights.

• The dorsal peduncular cortex (DP) is recruited following fear conditioning but not extinction

• DP neurons exhibit fear-related plasticity and cue-associated activity

• Fear-tagged DP neurons exert bidirectional control over fear memory expression

• Results point toward a revised understanding of the functional organization of the mPFC