A Prelimbic Cortex-Thalamus Circuit Bidirectionally Regulates Innate and Stress-Induced Anxiety-Like Behavior

Abstract

Anxiety-related disorders respond to cognitive behavioral therapies, which involved the medial prefrontal cortex (mPFC). Previous studies have suggested that subregions of the mPFC have different and even opposite roles in regulating innate anxiety. However, the specific causal targets of their descending projections in modulating innate anxiety and stress-induced anxiety have yet to be fully elucidated. Here, we found that among the various downstream pathways of the prelimbic cortex (PL), a subregion of the mPFC, PL-mediodorsal thalamic nucleus (MD) projection, and PL-ventral tegmental area (VTA) projection exhibited antagonistic effects on anxiety-like behavior, while the PL-MD projection but not PL-VTA projection was necessary for the animal to guide anxiety-related behavior. In addition, MD-projecting PL neurons bidirectionally regulated remote but not recent fear memory retrieval. Notably, restraint stress induced high-anxiety state accompanied by strengthening the excitatory inputs onto MD-projecting PL neurons, and inhibiting PL-MD pathway rescued the stress-induced anxiety. Our findings reveal that the activity of PL-MD pathway may be an essential factor to maintain certain level of anxiety, and stress increased the excitability of this pathway, leading to inappropriate emotional expression, and suggests that targeting specific PL circuits may aid the development of therapies for the treatment of stress-related disorders.

Significance Statement

This study provides insight into PL downstream pathways for regulating innate and stress-induced anxiety-like behavior. We reported that PL-mediodorsal thalamic nucleus (MD) projection and PL-ventral tegmental area (VTA) projection exhibited antagonistic effects on anxiety-like behavior, while the PL-MD projection but not PL-VTA projection was necessary for the animal to guide anxiety-related behavior. In addition, this study provides definite evidence that MD-projecting PL neurons bidirectionally regulated remote fear memory retrieval and concordant with a role for the PL-MD in anxiety. Moreover, this study is the first demonstration that restraint stress induced high-anxiety state accompanied by strengthening the excitatory inputs onto MD-projecting PL neurons and inhibiting PL-MD pathway rescued the stress-induced anxiety.

Introduction

Anxiety is a common emotional experience characterized by subjective experiences such as tension and apprehensive thoughts, alongside physiological changes including perspiration and elevations in blood pressure and heart rate. Occasional anxiety is a normal facet of the emotion that aids survival by heightening awareness and enabling swift responses to potential hazards. However, persistent or disproportionate anxiety that disrupts daily functioning can be incapacitating and is considered pathological (Calhoon and Tye, 2015). At present, anxiety disorders are the most prevalent mental disorders worldwide and have become a major burden of social and economic development in countries around the world (GBD 2019 Mental Disorders Collaborators, 2022). However, current medical treatments for anxiety disorders are not satisfying, owing to the inconsistent results, significant side effects, or high recurrence rates (Ginsburg et al., 2014; Bandelow et al., 2015; Loerinc et al., 2015; Craske and Stein, 2016). Cognitive behavior therapy (CBT) is a psychosocial intervention that helps patients replace cognitive distortions with more realistic and effective thoughts and is an effective treatment for individuals with anxiety disorders (Butler et al., 2006; Olatunji et al., 2010; Hofmann et al., 2012; Kaczkurkin and Foa, 2015). Therefore, the identification of top-down pathways originating from cognitive control regions could facilitate a comprehensive comprehension of the cognitive regulation of emotions and enable targeted interventions to be implemented effectively.

The medial prefrontal cortex (mPFC) is a critical brain region in the control of cognitive processes (including attention, decision-making, and working memory) and emotions (Miller, 2000; Gilmartin et al., 2014; D'Esposito and Postle, 2015; Dixon et al., 2017; Murray and Rudebeck, 2018). Previous studies have suggested that subregions of the mPFC have different and even opposite roles in regulating anxiety and fear (Adhikari et al., 2015; Bagot et al., 2015; Liu et al., 2020). However, the specific causal targets of their descending projections in modulating anxiety and fear have yet to be fully elucidated. The prelimbic cortex (PL), a subregion of the mPFC, has been implicated in anxiety and fear regulation (Do-Monte et al., 2015; Dejean et al., 2016; Marek et al., 2018; DeNardo et al., 2019). But the effects of manipulating PL activity in individuals displaying anxiety-like behaviors have yielded inconsistent findings. Several studies reported that inactivation of PL showed anxiolytic effects (Maaswinkel et al., 1996; Stern et al., 2010), while others reported anxiogenic (Jinks and McGregor, 1997; G. Q. Wang et al., 2015) or no effects (Bi et al., 2013). A similar phenomenon occurs in PL's regulation of fear expression, as some research suggested that activation of PL increased fear expression (Vidal-Gonzalez et al., 2006; Sierra-Mercado et al., 2011), while other studies showed that PL lesions did not disrupt fear expression (Sharpe and Killcross, 2015). The discrepancies may arise from different manipulation methods (such as lesions, pharmacological treatments, and optogenetic manipulations) employed in these studies, and these methods may interfere with different neuronal subpopulations that project to different targets. Indeed, using anterograde tracers, a previous study found that PL neurons anatomically project to multiple subcortical brain regions, including the nucleus accumbens (NAc), the mediodorsal thalamic nucleus (MD), the ventral tegmental area (VTA), the dorsal raphe nuclei (DR), and other brain areas (Vertes, 2004). In addition, it's been shown that optogenetic activation of PL-BLA led to anxiety-like behavior (Luo et al., 2023). Another study found that optogenetic inhibition of PL-BLA projection impaired retrieval of fear memory at early time points, while PL-PVT inhibition impaired late retrieval (Do-Monte et al., 2015). Thus, different PL projection neurons may exhibit distinct functions by recruiting different downstream targets, thereby clarifying the previously ambiguous role of this region in anxiety and fear. Furthermore, the adaptive changes in PL projection neurons in response to stress remain unknown, and it is elusive whether precise circuit-specific function of the PL contributes to stress-related psychiatric disorders.

In the present study, we sought to address the behavioral variety in PL-mediated top-down projections in real time. By utilizing optogenetics/chemogenetics, neural circuit mapping, fiber photometry, and electrophysiological recording, we uncovered a discrete PL-MD circuit that plays a crucial role in regulating anxiety and remote fear memory and elucidated a potential role for the PL-MD circuit in the treatment of stress-induced anxiety.

Materials and Methods

All experimental procedures used in this study were conducted in accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals (China) and were approved by the Southern Medical University Animal Ethics Committee. Approximately 6- to 7-week-old C57BL/6J male mice were purchased from Southern Medical University Animal Center (Guangzhou, China). CaMK2α-Cre mice were a gracious gift from Dr. Joe Tsien, Medical College of Georgia, USA. Behavioral assays were conducted with male mice only, and tasks for each paradigm were separated by at least 1 week. Four male mice were housed in a cage under standard housing conditions (a 12 h light/dark cycle) with ad libitum access to food and water. All mice were handled for 3 d before behavioral assays.

Stereotaxic injection

For virus injection, mice were anesthetized with 1% pentobarbital (intraperitoneal injection), mounted onto a mouse stereotaxic frame. The animal's skin was cut, and a small craniotomy was made. Injection coordinates based on the Mouse Brain atlaswere used to mark the location on the skull directly above the target area, and a small hole was drilled. Viruses were delivered through a Hamilton syringe. The total injection volume was 0.2–0.4 μl at 0.1 μl/min. Following injection, the syringe was left in place for 10 additional minutes and then slowly withdrawn. Animals were allowed to recover from anesthesia on a heating pad and then returned to their home cage.

For retrograde tracing experiments, 0.3 μl of cholera toxin β subunit conjugated to fluorophores (CTB488, Life Technologies) was injected unilaterally into the MD (AP: −0.93 mm, ML: +0.44 mm, DV: −3.25 mm).

For PL terminal activation or inhibition experiments, mice aged ∼8 weeks were administered unilateral or bilateral injections of 0.3 μl AAV2/8-CaMK2α-ChR2-eGFP or AAV2/9-CaMK2α-hM4Di-mCherry virus in the PL (AP: +1.90 mm, ML: ±0.30 mm, DV: −2.22 mm). For activation experiments, optical fibers (200 μm core diameter, 0.37 NA) were implanted unilaterally to target the MD (AP: −0.93 mm, ML: +0.44 mm, DV: −3.22 mm), VTA (AP: −3.18 mm, ML: +1.0 mm, DV: −4.52 mm, a 7° angle from lateral to medial), NAc (AP: +1.4 mm, ML: +1.0 mm, DV: −4.5 mm), or DR (AP: −5.0 mm, ML: 0.0 mm, DV: −3.10 mm, a 15° angle from caudal to rostral). For inhibition experiments, annular tubes (OD 0.48 mm) or optical fibers were implanted bilaterally above the MD (AP: −0.93 mm, ML: ±1.50 mm, DV: −2.60 mm, a 20° angle from lateral to medial) or VTA (AP: −3.18 mm, ML: ±1.68 mm, DV: −3.9 mm, a 15° angle from lateral to medial). For PL-MD rescue experiments, annular double tubes (RWD, 62023) were used, MD (AP: −0.93 mm, ML: ±0.44 mm, DV: −3.22 mm).

For fiber-optic recording experiments, to specifically target PL neurons projecting to the MD, mice aged ∼8 weeks were administered injections of 0.3 μl retrogradely transporting CAV-Cre into the MD (AP: −0.93 mm, ML: +0.44 mm, DV: −3.25 mm) and of 0.3 μl AAV2/9-DIO-GCaMP6m into the PL (AP: +1.90 mm, ML: +0.3 mm, DV: −2.22 mm). Six weeks after injection, optic fibers were implanted in the mice over the PL (AP: +1.90 mm, ML: +0.3 mm, DV: −2.19 mm).

To specifically target PL neurons projecting to the MD, mice aged ∼8 weeks were administered injections of 0.3 μl retrogradely transporting AAV retro-Cre into the MD (AP: −0.93 mm, ML: +0.44 mm, DV: −3.25 mm) and of 0.3 μl AAV2/9-DIO-ChR2-eGFP into the PL (AP: +1.90 mm, ML: +0.3 mm, DV: −2.22 mm). Six weeks after injection, optic fibers were implanted in the mice over the PL (AP: +1.90 mm, ML: +0.3 mm, DV: −2.19 mm).

Viruses

Purchased from BrainVTA:

rAAV-Ef1α-DIO-eNpHR3.0-EYFP-WPRE-pA, titer, 5.20 × 10¹² v.g./ml, AAV2/9

rAAV-Ef1α-DIO-EYFP-WPRE-pA, titer, 5.41 × 10¹² v.g./ml, AAV2/9

rAAV-CaMK2α-hM4D(Gi)-mCherry-WPRE-pA, titer, 6.02 × 10¹² v.g./ml, AAV2/9

rAAV-CaMK2α-mCherry-WPRE-pA, titer, 5.01 × 10¹² v.g./ml, AAV2/9

rAAV-Ef1α-DIO-GCaMP6m-WPRE-pA, titer, 6.69 × 10¹² v.g./ml, AAV2/9

CAV-Cre-WPRE-pA, titer, 6.02 × 10¹² v.g./ml

rAAV-DIO-ChR2-eGFP, titer, 3.01 × 10¹² v.g./ml, AAV 2/9

AAV retro-Cre, titer, 5.01 × 10¹² v.g./ml

Purchased from Shanghai Sunbio Medical Biotechnology:

pAAV-CaMK2α-hChR2(H134R)-EGFP, titer, 5.73 × 10¹³ v.g./ml, AAV2/8

pAAV-CaMK2α-EGFP, titer, 4.41 × 10¹³ v.g./ml, AAV2/8

Optogenetic manipulations

For activation experiments, 5 ms 473 nm blue solid-state laser (NEWDOON Technology) pulses were administered. The intensity power of activation at the PL cell body was ∼7–8 mW, and the intensity power of activation at the PL terminals was ∼10–12 mW. The stimulation frequency was 20 Hz. The inhibition experiments entailed continuous 593.5 nm yellow light (THINKERTECH), and the stimulation power was 9–10 mW. Bilateral stimulation was achieved by dividing light from a double-sided fiber-optic rotary (Doric Lenses).

Chemogenetic manipulations

The designer drug clozapine N-oxide (CNO, 5 μM, 0.3 μl each side; Sigma) was administered to animals 30 min before a test.

Immunohistochemistry

Mice were anesthetized with 1% pentobarbital (intraperitoneal injection) and transcardially perfused with 0.9% saline, followed by fixation with 4% paraformaldehyde in 0.01 M PBS (PFA). Brains were dissected out and postfixed in 4% PFA overnight and then rinsed with running water for 1 h before being transferred into 30% sucrose in 0.01 M PBS solution. Coronal sections (40 μm thick) containing the targeted brain region were cut on a freezing microtome.

For the retrograde tracing experiments, we first confirmed that the CTB injections were localized to the targeted brain region (MD). Coronal sections containing the PL were cut, and brain slices 2.34–1.54 mm from bregma were used to evaluate the PL. In every fourth slice, all CTB-labeled cells within the PL were selected for analysis; there were three mice per group and five brain slices per mouse.

For the c-Fos staining experiments, coronal sections containing the PVT and MD were cut, and brain slices −0.71 to −1.31 mm from bregma were used to evaluate the PVT and MD. In every fourth slice, slices were washed with 0.01 M PBS and then blocked in PBS containing 10% goat serum and 0.5% Triton X-100 for 2 h at room temperature. Then, slices were incubated with primary antibodies (1:1,000, CST, #2250) overnight at 4°C after washing three times for 5 min each time with PBS. After that, slices were washed three times for 5 min each time with PBS before incubation with secondary antibodies (1:1,000, Thermo Fisher Scientific, catalog #A-11012) for 2 h at room temperature. Sections were placed on coverslips, embedded in Vectashield mounting medium (VectorLabs, catalog #H-1200) and imaged by confocal microscopy (Nikon C2).

Fiber photometry

After AAV2/9-DIO-GCaMP6m virus was expressed for at least 6 weeks, an optical fiber was placed in a ceramic ferrule and inserted toward the PL (AP: +1.90 mm, ML: ±0.25–0.35 mm, DV: −2.20 mm) through a craniotomy procedure. The ceramic ferrules were fixed to the animals’ heads by dental powder. Mice were housed for at least 1 week to recover.

To record fluorescence signals, a 473 nm laser beam (OBIS 488LS; Coherent) was reflected by a dichroic mirror (MD498; Thorlabs), focused by a 10× objective lens (NA = 0.3; Olympus) and then coupled to an optical commutator (Doric Lenses). An optical fiber (200 μm core diameter, 0.37 NA, 5 m long) guided the light between the rotary joint and the implanted optical fiber. The laser power was adjusted at the tip of the optical fiber to a low level (∼30 μW) to minimize bleaching. GCaMP fluorescence was bandpass filtered (MF525-39, Thorlabs) and recorded by a photomultiplier tube (R3896; Hamamatsu). An amplifier (C7319; Hamamatsu) was used to convert the PMT current output to voltage signals, which were further filtered through a low-pass filter (40 Hz cutoff). The analog voltage signals were digitalized at 500 Hz and recorded with custom software developed in house using MATLAB 2017a (THINKERTECH). Mice were placed in the elevated plus maze (EPM) for 10 min. Animals’ positions were monitored by an overhead camera, location information and calcium signals were integrated via the Sync interface, and the results were output by a custom-written MATLAB program.

In vitro electrophysiological recordings

mPFC slices were prepared as described previously (Bi et al., 2013). Mice were anesthetized with 1% pentobarbital (intraperitoneal injection), and then their brains were quickly removed and chilled in an ice-cold solution (artificial cerebrospinal fluid, ACSF) containing the following (in mM): 130 potassium gluconate, 20 KCl, 10 HEPES buffer, 2 MgCl.6H₂O, 4 Mg-ATP, 0.3 Na-GTP, and 10 EGTA. The pH was adjusted to 7.25 with 10 M KOH. Coronal PL slices (300 μm) were cut in ice-cold modified ACSF using a Vibroslice (Leica VT 1000S) and transferred to a storage chamber where they were allowed to recover for at least 1.5 h (0.5 h at 34°C followed by 1 h at 25 ± 1°C) in an ACSF solution containing the following (in mM): 126 NaCl, 3 KCl, 1 MgSO₄, 2 CaCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃, and 10 glucose. Slices were placed in the recording chamber, which was superfused (1.5 ml/min) with ACSF at 32–34°C. All solutions were saturated with 95% O₂/5% CO₂ (vol/vol).

For whole-cell recording in MD-projecting neurons, the pipettes were filled with ACSF solution that contained the following (in mM): 133 potassium gluconate, 20 NaCl, 0.2 EGTA, 10 HEPES, 4 Mg-ATP, and 0.3 Na₃-GTP, pH 7.3 (285 mOsm). The pipettes were pulled by a micropipette puller (P-97, Sutter Instrument) with a resistance of 3–5 MΩ. After the whole-cell recording was complete, PL neurons were held at −70 mV under voltage-clamp mode to record spontaneous excitatory postsynaptic currents (sEPSCs), and the sEPSCs were recorded for 2 min. To record spontaneous inhibitory postsynaptic currents (sIPSCs), the neurons were held at 0 mV under voltage-clamp mode, and the pipettes were filled with ACSF solution containing the following (in mM): 110 Cs₂SO₄, 0.5 CaCl₂, 2 MgCl₂, 5 EGTA, 5 HEPES, 5 TEA, and 5 Mg-ATP, pH 7.3 (285 mOsm). All recordings were acquired using a MultiClamp 700B amplifier, and signals were low-pass filtered at 3 kHz and digitized at 10 kHz (Digidata 1440, Molecular Devices).

To test the monosynaptic response of ChR2 terminal activation, ChR2 was activated using a DG4 light source, and the sodium channel blocker tetrodotoxin (TTX, 1 µM) was perfused in combination with the potassium channel blocker 4-aminopyridine (4-AP, 100 µM) to facilitate release from synaptic terminals. The NMDA receptor antagonist AP5 (50 µM) and AMPA receptor antagonist NBQX (20 µM) were applied to block glutamate receptors.

All analyses were performed using Clampfit 10.2 (Axon Instruments/Molecular Devices) and Mini Analysis Program (Synaptosoft).

Single neuron RNA extraction procedures

To obtain transcriptome data from the different projection single neurons, we applied the recently developed Patch-seq protocol but did not perform electrophysiology recording (Cadwell et al., 2016; Scala et al., 2021). We took meticulous steps to enhance RNA yield and maintain proper osmolarity. Glass capillaries (1.2 mm OD, 0.69 mm ID, Sutter Instrument, catalog #BF120-69-15) underwent autoclaving prior to the fabrication of patch-clamp pipettes. Before use, all surfaces potentially in contact with reagents or solutions were either RNase-free or underwent thorough cleaning using RNaseZap (Life Technologies, catalog #AM9780) and DNA-OFF (Takara, catalog #9036). Intracellular solution was prepared by dissolving 4 mM KCl, 10 mM HEPES, 0.2 mM EGTA and 111 mM potassium gluconate in RNase-free water in a 200 ml conical flask. The mixture was thoroughly blended, following which 4 mM Mg-ATP, 0.3 mM Na₃-GTP, and 5 mM sodium phosphocreatine (all sourced from Sigma-Aldrich) were incorporated, along with RNase inhibitor at a 1:1,000 dilution (Takara, catalog #2313A). The pH was set to 7.40 using RNase-free 0.5 M KOH under the guidance of a specialized pH meter. Subsequently, RNase-free water was added to the solution to achieve the desired volume. The osmolarity of the solution was meticulously verified to be ∼300–310 mOsm. The solution was frozen at −20°C and utilized within a maximum timeframe of 2 weeks.

The cell lysis buffer was prepared using the SMART-Seq HT Kit (Takara, catalog #634437), excluding the addition of 3′ SMART-Seq CDS Primer II A, and reducing 5 μl of RNase-free water reserved for the volume of intracellular solution containing the cell samples. The cell lysis buffer was stored in a 200 μl RNase-free tube (Axygen, catalog #HBDY200UI) on ice. The 3′ SMART-Seq CDS Primer II A was added to the cell lysis buffer before lysis started.

Different from the Patch-seq protocol, the labeled neurons were directly aspirated by the pipettes without recording (Y. H. Chen et al., 2021). The pipette resistance did not exceed 1 MΩ due to the lack of recording requirements. While aspirating, if any extracellular contents were observed to enter the pipette, both the pipette and its contents were discarded. The qualified cells were injected into the cell lysis buffer and each containing a maximum of five cells. The samples were plunged into liquid nitrogen and stored at −80°C. Within a week at most, all samples were subsequently converted into cDNA using the SMART-Seq HT Kit. Samples containing less than ∼1 ng total cDNA or with an average size below 1,500 bp were excluded from sequencing. The cDNA library was purified and constructed using the TruePrep DNA Library Prep Kit V2 for Illumina (Vazyme, catalog #TD503). The DNA library had a total length distribution ranging from 300 to 700 bp. The TruePrep Index Kit V2 for Illumina (Vazyme, catalog #TD202) was used to perform dual-end index tags. The resulting cDNA libraries were then frozen and sent for sequencing in three separate batches within a maximum period of 2 weeks.

RNA-seq analysis

We conducted base calling on the raw image data obtained from the RNA-seq experiment using Bcl2fastq version 2.20.0.422. Subsequently, the RNA-seq data were aligned to the Mus_musculus_GRCm39.107_Ensembl. To ensure the quality of the RNA-seq data, FastQC version 0.10.1 (Bittencourt, 2010) and Cutadapt version 1.9.1 (Martin, 2011) were utilized for adapter sequence removal, elimination of low-quality scores (<20) at the 5′ or 3′ ends, filtering out bases containing “N,” and discarding sequences trimmed to a length <75 bp. DEGs were analyzed by DESeq2 v1.38.3 (Anders and Huber, 2010). Genes with |log2(fold change) | >1 and a p_adj <0.05 were defined as DEGs in this study.

Behavioral assays

Open-field test

The open-field apparatus consisted of a rectangular chamber (40 × 40 × 30 cm) that was made of gray polyvinyl chloride. The mice were gently placed in the center and allowed to explore the area for 9 min, with laser light on during in the second 3 min period; the light was off during the other two 3 min periods. The digitized image of the path taken by each mouse was stored, and the total distance and time spent in the center were analyzed using VERSADAT version software.

Elevated plus maze test

The EPM had a central square platform (5 × 5 cm), from which two opposing open arms (30 × 5 × 0.5 cm) and two opposing enclosed arms (30 × 5 × 15 cm) radiated. Mice were placed in a silent experimental room and were allowed to become acclimated to the area for at least 1 h before testing. The time spent in the open arms and open arm entries was recorded over a 9 min test period, with and without optogenetic stimulation (protocol: 3 min of no light, followed by 3 min of light, followed by 3 min of no light). The maze was cleaned with a solution of 75% ethanol in water between the sessions. All data were analyzed post hoc using EthoVision XT 11.5 software.

Novelty suppressed feeding test

The testing apparatus consisted of a plastic box (50 × 50 × 20 cm), the floor of which was covered with ∼2 cm of wooden bedding and the arena was brightly lit (1,100–1,200 lux). Twenty-four hours prior to behavioral testing, all food was removed from the home cage, although water was still available ad libitum. At the time of testing, a single pellet of food was placed on a white paper platform positioned in the center of the box. Each animal was placed in a corner of the box, and a stopwatch was immediately started. The latency of the mice to begin eating was recorded. Immediately after this test, the mice were then placed into home cage and the amount of food consumed in 5 min was measured.

Real-time place avoidance test

The real-time place avoidance (RTPA) contained a transparent plastic chamber (50 × 53 × 20, L × W × H in cm) and divided into two equal compartments. One of these was assigned as the photostimulated zone (counterbalanced between animals). At the start of the 20 min session, mice were placed in the center of nonstimulated side of the chamber. Blue light stimulation (20 Hz, 5 ms pulse width, ∼5 mW for CaMKIIα-positive neurons) was delivered whenever the mouse entered or stayed in the photostimulated chamber. Behavior was recorded and analyzed by EthoVision XT video tracking software (Noldus Information Technologies).

Fear conditioning

The fear conditioning chamber consisted of a square cage with numerous parallel stainless-steel grid bars in the floor connected to a shock generator (context A; The FreezeFrame System, Coulbourn Instruments). Two minutes after being placed in the chamber, mice were exposed to 30 s tones (2.8 kHz, 80 dB) terminated with a footshock at 60 s intertrial intervals. The protocol used in the training was four tone-shock pairings (0.75 mA, 1 s shocks). The chamber was cleaned with 75% ethanol at the end of each trial. The following days, mice were tested in a different context (context B, floor and walls were changed) with two tones at both Day 1 and Day 28. The protocol was modified to that used in previous studies (Do-Monte et al., 2015). Data are mean ± SEM in blocks of two trials in Figure 4. Laser illumination was delivered during the two tones at each time point.

Figure 4.

MD-projecting PL neurons exhibit task-related activity patterns when mice are performing the EPM task. A, Schematic of fiber photometry recording of MD-PL neuronal activity during the EPM test in C57 mice. B, Images showing GCaMP6m expressed in MD-projecting PL neurons and optical fiber implantation in the PL. Scale bar, 200 µm. C, A representative mouse movement trace and Ca²⁺ transient heatmap of MD-projecting neurons in the EPM in 10 min. D, EPM scores for MD-projecting neurons. N = 6 mice. E, Heatmap of MD-projecting neuronal Ca²⁺ transients in a representative mouse aligned to the start of entry into the open arm compartment. F, Quantification of the change in calcium signals pre and post entry into the open arm compartment, paired t test, t₍₅₎= 6.287, p = 0.002. G, Average Ca²⁺ transients of MD-projecting PL neurons of mice in the EPM, Ca²⁺ transients during a behavioral transition from the open arm to the closed arm compartment. H, The Ca²⁺ transients of MD-projecting neurons were reduced after the transition, t₍₅₎= 4.053, p = 0.0098. N = 6 mice each group. I, Heatmap of MD-projecting neuronal Ca²⁺ transients in a representative mouse aligned to the start of head dipping, n = 18. J, Quantification of the change in calcium signals pre and post head dips, paired t test, t₍₅₎= 2.669, p = 0.044. The data are presented as the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001.

Restraint stress experiment

Restraint stress was induced by restraining the mice in well-ventilated Perspex restraining tubes for 3 consecutive days, 1 h each day. During the assay, the mice were not physically compressed or did they experience pain. After restraints, the experimental mice were returned to their home cages. Control mice were still in their home cage in the same experimental room during the restraining period.

Statistical analysis

Fiber photometry data analysis

We calculated changes in fluorescence using methods similar to those reported in previous studies (Zhong et al., 2017; Li et al., 2018). Photometry data were saved as MATLAB.mat files for further analysis. We segmented the data based on behavioral events within individual trials. We derived the fluorescence change values (ΔF / F) by calculating (F − F_c) / F_c − F₀, where F₀ is the baseline fluorescence signal averaged over the whole session and F_c is the baseline fluorescence signal averaged over a 2 s control time window. ΔF / F values are presented with heatmaps or the SEM. To examine the response of different PL neurons in the EPM, we calculated Ca²⁺ transients during 5 s after head dips or 10 s after arm transition and averaged the value for all trials.

In our study, EPM scores were computed for all mice to estimate the extent to which the PL encoded the arm type in the EPM (Adhikari et al., 2011, 2015). In brief, the PL neurons’ mean GCaMP intensities were calculated for each of the EPM compartments, and the score was computed by the following formula:

$$\mathrm{EPM}\mathrm{Score}=(A\text{–}B)/(A+B),$$

where

$$A=0.25\times (|F_L\text{–}{F}_U|+|F_L\text{–}{F}_D|+|F_R\text{–}{F}_U|+|F_R\text{–}{F}_D|),$$

and

$$B=0.5\times (|F_L\text{–}{F}_R|+|F_U\text{–}{F}_D|),$$

F_L, F_R, F_U, and F_D are the mean GCaMP intensities for the windows in the left, right, up, and down arms, respectively. A is the mean difference in GCaMP intensity between arms of different types, while B is the mean difference for arms of the same type.

Statistics

For all experiments, coronal sections of the targeted brain area were used to confirm virus expression and to verify targeting of the fibers or annular tubes. Only mice with target regions were used. The data are presented as the mean ± SEM. Comparisons between two groups were performed using Student's unpaired two-tailed t test or paired t test. Comparisons among three groups were performed using one-way ANOVA with post hoc tests. Comparisons between two groups at different time points were performed using two-way ANOVA. Post hoc tests were used following ANOVA only if significant main or interaction effects were detected. The criterion for statistical significance was p < 0.05, and differences were calculated using SPSS 24.

Results

PL-MD projection modulates anxiety-like behavior

Previous animal studies have demonstrated that direct activation of the PL did not elicit any discernible impact on anxiety-like behaviors (Adhikari et al., 2015), whereas its inhibition induced anxiogenic effects (G. Q. Wang et al., 2015). To determine the characteristics of PL modulation over anxiety-like behaviors, we initially employed an optogenetic tool to systematically investigate the causal targets of top-down projections (Beracochea and Krazem, 1991; Gelowitz and Kokkinidis, 1999; Ouhaz et al., 2017). To this end, adeno-associated virus (AAV) containing channelrhodopsin-2 fused to enhanced green fluorescent protein (ChR2-eGFP) under the control of the CaMK2α promoter (AAV2/8-ChR2-eGFP) was injected into the PL (Fig. 1A). We found that the downstream main regions targeted by the PL include the NAc, MD, VTA, and DR (Fig. 1A). The functional expression of ChR2 was examined via electrophysiology (Fig. 1B). To activate the PL terminals, fiber optics were implanted above the MD, VTA, NAc, or DR (Fig. 1C,F,I,L). In these mice, we found that PL-MD activation decreased the time spent in the open arms (Fig. 1D,E) in the EPM test. Conversely, activation of PL-VTA projection increased open-arm exploration of mice (Fig. 1G,H). Additionally, photoactivation of PL-NAc projection or PL-DR projection did not elicit any significant alterations in EPM test (Fig. 1J–N).

Figure 1.

Activation of PL-MD and PL-VTA projections exert antagonistic effects on anxiety-like behavior. A, Left, Schematic of AAV2/8-CaMK2α-ChR2-eGFP injection into PL. Right, Representative image showing virus expression in the PL (scale bar, 500 µm) and the PL terminals in the NAc, MD, VTA, and DR (scale bar, 200 µm). B, Ex vivo whole-cell patch-clamp recording shows the response of a ChR2 cell to a train of light pulses (20 Hz, 5 ms pulse width, blue bars) for 1 s in current-clamp mode. C, Representative image showing the PL terminals in the MD. Scale bar, 500 µm. D, E, Optogenetic activation of PL-MD projection decreased the time animals spent in the open arms (D) and had no effect on animals’ open-arm entries (E). Two-way repeated-measures ANOVA, main effect of opsin, F_(1,18)= 4.769, p = 0.042, post hoc Mann–Whitney U test, p = 0.048, n = 10 mice for eGFP, 10 mice for ChR2. F, Representative image showing the PL terminals in the VTA. G, H, Optogenetic activation of PL-VTA projection increased the time spent in the open arms (G) and had no effect on the open-arm entries (H). Two-way repeated-measures ANOVA with Mann–Whitney U test, p = 0.046, n = 9 mice for eGFP, 8 mice for ChR2. I, Representative image showing the PL terminals in the NAc. J, K, Optogenetic activation of the PL-NAc projection had no effect on the time mice spent in the open arms (J) or the number of open-arm entries (K). PL-NAc, n = 9 mice for eGFP, 10 mice for ChR2. L, Representative image showing the PL terminals in the DR. M, N, Optogenetic activation of the PL-DR projection had no effect on the time mice spent in the open arms (M) or the number of open-arm entries (N). PL-DR, n = 10 mice for eGFP, 10 mice for ChR2. O, Time spent in the stimulation chamber during optogenetic activation of PL-MD projection in the RTPA test. Unpaired t test, t₍₁₇₎= 2.654, p = 0.0167, n = 10 eGFP mice, n = 9 ChR2 mice. P, Optogenetic activation of PL-MD projection had no effect on the time mice stayed in the center (right) or the distance mice traveled (left). Q, R, Optogenetic activation of PL-MD projection increased the latency of the mice to begin eating food in NSFT (Q) with no effects on food consumption (R). Q, Two-way repeated-measures ANOVA, main effect, F_(1,14)= 22.86, p = 0.0006, Bonferroni's multiple-comparisons test, eGFP versus ChR2, p < 0.0001, n = 8 mice for eGFP, 8 mice for ChR2. S, Time spent in the stimulation chamber during optogenetic activation of PL-VTA projection in RTPA test. Unpaired t test, t₍₁₅₎= 1.164, p = 0.2626, n = 9 eGFP mice, n = 8 ChR2 mice. T, Optogenetic activation of PL-VTA projection had no effect on the time mice stayed in the center (left) or the distance mice traveled (right). U, V, Optogenetic activation of PL-VTA projection did not affect behavior in NSFT. W–Y, Activation of PL-MD projection only in the closed arms did not elicit aversion toward the closed arms (W), and increased the time spent in the open arms (X), but had no effect on the open-arm entries (Y). W, unpaired t test, t₍₂₂₎= 1.954, p = 0.0636; X, unpaired t test, t₍₂₂₎= 2.257, p = 0.0343; Y, unpaired t test, t₍₂₂₎= 0.2030, p = 0.8410; n = 12 eGFP mice, n = 12 ChR2 mice. The data are presented as the mean ± SEM; *p < 0.05, ***p < 0.001, ****p < 0.0001.

The aforementioned findings suggest that either the MD or the VTA may serve as subcortical output streams through which PL modulates anxiety-related behavior. Subsequently, we employed two more behavioral assays, namely, the RTPA assay (Sun et al., 2020) and the novelty suppressed feeding test (NSFT) (Bodnoff et al., 1989; Bi et al., 2013), to further ascertain the potential involvement of PL-MD/VTA projections in modulating the anxiety-like behaviors. As expected, optogenetic activation of the PL-MD projection resulted in a significant reduction in the time spent in the stimulated chamber compared with the control group (Fig. 1O) without alteration in locomotor activity (Fig. 1P), whereas ChR2 and eGFP mice spent a similar amount of time on the side during light illumination of PL-VTA projection (Fig. 1O,P) in the RTPA test. Consistently, the latency to initiate eating was prolonged compared with control mice when PL-MD projection was activated with on effect on total food consumption (Fig. 1Q,R), while no significant effect was observed upon optogenetic activation of PL-VTA projection in NSFT (Fig. 1S–V). To determine whether the decreased open time was caused by light-induced aversion, we performed additional experiments and found that the activation of PL-MD pathway only in the closed arms did not lead to an increase in time spent in closed arms (Fig. 1W–Y), indicating that this manipulation elicited heightened anxiety responses instead of place avoidance in EPM test. Taken together, these results suggest that activation of PL-MD and PL-VTA projections can modulate anxiety-related behaviors in opposite directions.

We subsequently aimed to evaluate the necessity of PL-MD and PL-VTA projections in regulating anxiety-related behaviors. DREADD (designer receptors exclusively activated by designer drugs) method was employed to inactivate PL-MD or PL-VTA projection. These DREADDs are variants of muscarinic receptors (-hM4Di) that have been modified to be selectively activated by the pharmacologically inert compound CNO. When activated by CNO, hM4Di suppresses neuronal firing through signaling by activating G-protein inwardly rectifying potassium channels (Armbruster et al., 2007; Alexander et al., 2009). AAV2 expressing the engineered hM4Di receptor under the control of the CaMK2α promoter (AAV2/9-CaMK2α-hM4Di-mCherry) was injected into the PL of C57 mice (Fig. 2A), and functional verification of hM4Di expression was conducted in brain slices by perfusion of CNO (Fig. 2B,C). Seven weeks after AAV2/9-CaMK2α-hM4Di-mCherry or control virus injection, bilateral cannulas were implanted above the MD (Fig. 2D) or VTA of the mice (Fig. 2K). CNO (5 µm) were administered into the MD or VTA through the cannulas 30 min prior to behavioral assays. DREADD inhibition of PL-MD projection resulted in anxiolytic effects, as evidenced by increased time spent in the open arms and entries in hM4Di-CNO mice in the EPM test (Fig. 2E,F). Moreover, inhibition of PL-MD projection had no effect on the OFT (Fig. 2G,H) or NSFT (Fig. 2I,J). However, inhibition of PL-VTA projection had no effect on the same test (Fig. 2L,M), neither did on the OFT (Fig. 2N,O) or NSFT (Fig. 2P,Q). Together, these observations suggest that the endogenous activity of PL-MD projection is necessary to maintain certain level of anxiety.

Figure 2.

Inhibition of PL-MD projection produces antianxiety-like behavior. A, Representative image of the PL showing AAV2/9-CaMK2α-hM4Di-mCherry injections. Scale bar, 500 µm. B, C, The neuronal activity of CaMK2α-hM4Di–expressing neurons induced by injected current was suppressed in by CNO perfusion. Paired t test, t₍₃₎= 3.826, p = 0.031, n = 4. D, Left, Schematic of bilateral injection of AAV2/9-CaMK2α-hM4Di-mCherry in the PL and fiber-optic placement targeted at MD terminals. Right, Representative image showing the PL terminals in the MD. Scale bar, 200 µm. E, F, Chemogenetic inhibition of PL-MD projection increased the time animals spent in the open arms (E) and open-arm entries (F). E, unpaired t test, t₍₁₄₎= 3.535, p = 0.0033; F, unpaired t test, t₍₁₄₎= 2.173, p = 0.0474, n = 8 mice for mCherry + CNO, 8 mice for hM4Di + CNO. G–J, Chemogenetic inhibition of PL-MD projection had no effect on the time mice stayed in the center (G), the distance mice traveled (H), the latency to food (I), or food consumption (J). K, Left, Schematic of bilateral injection of AAV2/9-CaMK2α-hM4Di-mCherry in the PL and fiber-optic placement targeted at VTA terminals. Right, Representative image showing the PL terminals in the VTA. Scale bar, 200 µm. L, Q, Chemogenetic inhibition of PL-VTA projection had no effect on the time animals spent in the open arms (L), animals’ open-arm entries (M), the time mice stayed in the center (N), the distance mice traveled (O), the latency to food (P), or food consumption (Q). n = 7 mice for mCherry + CNO, 7 mice for hM4Di + CNO. The data are presented as the mean ± SEM; *p < 0.05, **p < 0.01.

MD-projecting PL neurons show task-related activity patterns in the EPM

To determine whether the PL innervates the MD directly, the retrograde tracer recombinant cholera toxin subunit (CTB) was injected into the MD (Fig. 3A), and we found a distinct laminar distribution of MD-projecting PL neurons 2 weeks later (Fig. 3B). Additionally, light stimulation of PL terminals in the MD were able to evoke robust spiking, which were completely abolished by the application of tetrodotoxin (TTX) and restored by 4AP, indicating that the exciting responses in the MD neurons originated from monosynaptic activation of PL neurons. We also validated the release of glutamate from these terminals by observing that the light-induced excitation in the same cells was abolished upon application of glutamate receptor antagonists NBQX and AP5 (Fig. 3C–E). Taken together, these results suggest that MD neurons receive monosynaptic input from PL neurons.

Figure 3.

Monosynaptic inputs to the MD from the PL. A, Left and middle, Schematic of CTB injection in the MD (CTB488, green). Scale bar, 500 µm. Right, A coronal section of PL neurons labeled with CTB in green. Scale bar, 200 µm. B, Quantification of the medial-lateral (M/L) distribution of PL-MD neurons. C, Voltage-clamp recording of ChR2-expressing PL neuronal terminals in the MD with light stimulation. D, Example traces from one mouse showing stimulation of ChR2-expressing PL neuronal terminals in the MD (acute slice). Recordings used TTX and 4-AP to detect monosynaptic responses. E, Summary of the data with different treatments, n = 4 mice, one-way ANOVA with Tukey’s post hoc analysis, F_(2,9)= 737.8. The data are presented as the mean ± SEM; ****p < 0.0001.

To investigate the real-time dynamics of MD-projecting PL neurons during mice explore in the EPM, a dual-virus strategy that selectively expressed the genetically encoded calcium indicator GCaMP6m in a projection-specific manner (T. W. Chen et al., 2013) was employed. We injected canine adenovirus encoding Cre recombinase (CAV-Cre) into the MD (Lerner et al., 2015; Li et al., 2018); concurrently with Cre-dependent vectors, AAV2/9-DIO-GCaMP6m injected into the PL to selectively express GCaMP6m in PL neurons projecting to the MD (Fig. 4A). To record the GCaMP fluorescence changes, an optical fiber was implanted above the PL (Fig. 4B). Fiber photometry recording revealed that the GCaMP intensity of MD-projecting PL neurons remained high when the mice were in the open arms (Fig. 4C). To confirm the observed sustained response in the open arms were not due to chance, an EPM score was calculated (Adhikari et al., 2011, 2015). We found that the EPM score of MD-projecting PL neurons was 0.38817 (Fig. 4D), a positive score indicating the response represents the “open” structure of the EPM (Adhikari et al., 2011). In addition, we observed a rapid increase in GCaMP intensity of MD-projecting neurons when mice switched from the closed arm to the open arm (Fig. 4E,F), while a decrease when mice switched from the open arm to the closed arm (Fig. 4G,H). Head dips were characterized by brief periods during which mice lowered their heads over the edges of the open arms toward the floor (Walf and Frye, 2007; Adhikari et al., 2011), evaluating the frequency and/or duration of head dips can be considered as a “risk assessment.” Notably, we found a significant increase in GCaMP signals in MD-projecting PL neurons after head dips (Fig. 4I,J). These results suggested that subpopulations of PL neurons project to the MD encode anxiety-associated features. Consistent with this active response of MD-projecting PL cells in anxiogenic settings, and the hypothesized countervailing role for inhibitory anxiolytic inputs to MD, our above results showed PL-MD inhibition decreased avoidance of open arms (Fig. 2E).

PL-MD projection regulates remote fear memory retrieval

High trait anxiety individuals tend to show deficiency in achieving extinction of fear response when exposure to the feared situation without an aversive outcome (Perales et al., 2013; Zika et al., 2023). Therefore, we investigated the role of the PL-MD projection in modulation of fear memory by employing tone-cued fear conditioning and optogenetic manipulations. For optogenetic activation of PL-MD projection, we bilaterally injected AAV2/8-CaMK2α-ChR2-eGFP or control virus into the PL of C57 mice and implanted fiber optics above the MD (Fig. 5A). After cued fear conditioning, optogenetic activation of PL-MD projection on Day 28 but not on Day 1 increased conditioned freezing (Fig. 5B). For optogenetic inhibition of the PL-MD projection, AAV2/9-DIO-eNpHR-eYFP and control virus was injected into the PL of CaMK2α-Cre mice and implanted fiber optics above the MD (Fig. 5C–E). Acute brain slice recording revealed that activation of eNpHR by yellow light inhibited the firing of PL projecting neurons (Fig. 5F). Yellow light illumination of PL-MD projection suppressed conditioned freezing on Day 28 instead of Day 1 (Fig. 5G). To further elucidate the involvement of paraventricular nucleus of the thalamus (PVT), a region next to MD, in remote fear memory retrieval at Day 28, we performed c-Fos staining after retrieval Day 28 and found that both the MD and PVT activities were increased during remote fear retrieval (Fig. 5H,I). Moreover, we used an intersectional strategy to label the fibers of MD-projecting PL neurons by injecting an AAV retro-expressing Cre recombinase (AAV retro-Cre) into the MD concurrently with an AAV-expressing ChR2-eGFP in a Cre-dependent manner (AAV-DIO-ChR2-eGFP) into the PL (Fig. 5J,K). The results showed that MD-projecting PL neurons mainly innervated the MD (Fig. 5K). Additionally, we observed that specific activation of MD-projecting neurons in PL enhanced remote fear conditioning memory in mice (Fig. 5L). These results suggested that MD-projecting PL neurons influence bidirectionally remote fear memory retrieval and concordant with a role for the PL-MD in anxiety.

Figure 5.

PL-MD regulates remote fear retrieval. A, Schematic of unilateral injection of AAV2/8-CaMK2α-ChR2-eGFP in the PL of C57 and representative image of PL terminals in MD. Scale bar, 500 μm. B, Activation of PL-MD projection enhanced fear retrieval on Day 28, t₍₁₇₎ = -2.256, p = 0.038, n = 8 mice for eGFP, 11 mice for ChR2. C, Schematic of bilateral injection of AAV2/9-DIO-eNpHR-eYFP in the PL of CaMK2α-Cre mice with fiber optics above the MD. D, E, Representative image showing eNpHR-eYFP expression in PL and MD. Scale bar, 500 μm. F, Current-clamp traces from an in vitro slice recording of a DIO-eNpHR-expressing neuron illuminated by a 594 nm laser confirmed that DIO-eNpHR viruses were efficient. G, Inhibition of PL-MD projection impaired fear retrieval on Day 28, t₍₂₁₎ = 2.520, p = 0.020, n = 11 mice for eYFP, 12 mice for eNpHR. H, Micrographs showing c-Fos expression in the MD and PVT in the control or conditioned groups following fear retrieval at Day 28 time points. I, Fear retrieval at Day 28 increased the density of c-Fos in both MD and PVT. MD c-Fos density, Unpaired t test, t₍₁₀₎ = 5.773, p = 0.0002. PVT c-Fos density, Unpaired t test, t₍₁₀₎ = 4.268, p = 0.0016. Scale bar, 500 μm. J, Schematic of unilateral injection of AAV-DIO-ChR2-eGFP in the PL and AAV retro-Cre into MD. K, Micrographs showing virus expression in the PL with fiber optics above the PL and axon terminals of MD-projecting PL neurons in the MD. Scale bar, 100 μm. L, Activation of PL-MD projection enhanced fear retrieval at Day 28, t₍₂₂₎ = 3.075, p = 0.0055, n = 12 mice for eGFP, 12 mice for ChR2. The data are presented as the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001.

Restraint stress severely impairs synaptic inputs onto MD-projecting PL neurons

Stress can lead to behavioral maladaptation associated with anxiety (Guo et al., 2012). We further investigated whether the activity of MD-projecting PL neurons was altered in response to pathological stress. Retrograde tracers CTB488 were injected into the MD of C57 mice to label MD-projecting PL neurons (Fig. 6A). Two weeks later, mice were either exposed to restraint stress for 3 consecutive days, 1 h each day, or kept in the home cage with no stress (Fig. 6B). Mice exhibited heightened anxiety levels in the EPM test following restraint stress, compared with control mice, as evidenced by reduced time spent in the open arms (Fig. 6C,D). Subsequently, whole-cell recordings were performed on CTB488-positive neurons in slices containing the PL region from stressed mice and control mice. We observed that the sEPSC frequency but not amplitude was increased (Fig. 6E–G), whereas the sIPSC frequency was decreased (Fig. 6H–J) in MD-projecting PL neurons of stress groups. These results indicated that synaptic inputs onto MD-projecting PL neurons were increased after restraint stress, implying a potential contribution to the development of stress-induced pathological anxiety.

Figure 6.

Restraint stress severely impairs synaptic inputs onto MD-projecting PL neurons. A, Schematic of the mouse brain. Cholera toxin subunit B (CTB) was injected into the MD (CTB488, green). B, Experimental procedure. C, D, After stress, mice showed decreased exploration in the open arms in the EPM. Two-tailed unpaired t test, n = 7 control mice, n = 9 stressed mice, t₍₁₄₎= 3.075, p = 0.008. E–J, Representative sEPSC (E) or sIPSC (H) traces of MD-projecting PL neurons from control and stressed mice. Calibration: 1 s, 20 pA. Cumulative distribution of sEPSCs (F) or sIPSCs (I) amplitudes and average amplitudes. Cumulative distribution of sEPSCs (G) or sIPSCs (J) interevent intervals and average frequencies. F, G, Control n = 8 cells from 3 mice, stressed n = 9 cells from 3 mice; G, unpaired t tests, t₍₁₅₎= 2.225, p = 0.042. I, J, Control n = 11 cells from 4 mice, stressed n = 12 cells from 4 mice; J, unpaired t tests t₍₂₁₎= 2.775, p = 0.011. The data are presented as the mean ± SEM; *p < 0.05; **p < 0.01.

Inhibition of PL-MD projection reverses stress-induced anxiety-like behavior

Given that PL-MD projection inhibition can suppress anxiety-like behavior in naive mice (Fig. 2E) and their excitatory inputs were greatly enhanced following restraint stress (Fig. 6G), we sought to investigate whether inhibition of this projection might ameliorate behavioral abnormalities in stressed mice. AAV2/9-CaMK2α-hM4Di-mCherry virus and control virus were injected into the PL of C57 mice, and 7 weeks later, cannulas were implanted above the MD of these mice (Fig. 7A); then, mice were subjected to restraint stress for 3 consecutive days for 1 h each day, or kept in the housing room with no stress; 24 h after the last restraint stress session, stressed mice and control mice were both transferred to a testing room and behaviorally assayed (Fig. 7A). Our findings reveal that chemogenetic inhibition of PL-MD projection not only increased open-arm exploration by naive mice in the EPM but also reversed the open arms avoidance in stressed mice (Fig. 7B,C). The similar results were also obtained from NSFT (Fig. 7D–F).

Figure 7.

Inhibition of PL-MD projection or activation of PL-VTA projection reverses stress-induced anxiety-like behavior. A, Top, Experimental procedure. Bottom, Schematic of bilateral injection of AAV2/9-CaMK2α-hM4Di-mCherry in the PL and cannula placement targeted at MD terminals. Scale bar, 200 µm. B, The restraint stress (RS)-induced decrease of open-arm exploration in the EPM test which was reversed by chemogenetic inhibition of PL-MD projection. B, One-way ANOVA with Tukey's multiple-comparisons test, F_(3,39)= 4.351, p = 0.0097; C-mCherry versus C-hM4Di, p = 0.0065; C-mCherry versus S-mCherry, p = 0.0456; S-mCherry versus S-hM4Di, p = 0.0012. C, Open arm entries, one-way ANOVA with Tukey's multiple-comparisons test, F_(3,39)= 2.298, p = 0.0926; C-mCherry versus C-hM4Di, p = 0.4226; C-mCherry versus S-mCherry, p = 0.0617; S-mCherry versus S-hM4Di, p = 0.4275; n = 10 mice for C-mCherry, 10 mice for C-hM4Di, 10 mice for S-mCherry, 13 mice for S-hM4Di. D, Top, Experimental procedure. Bottom, Schematic of the NSFT. E, F, Chemogenetic inhibition of PL-MD projection recued the latency to food (E), but not food consumption (F). One-way ANOVA with Tukey's multiple-comparisons test, F_(3,32)= 8.353, p = 0.0003; C-mCherry versus C-hM4Di, p = 0.9807; C-mCherry versus S-mCherry, p < 0.0001; S-mCherry versus S-hM4Di, p = 0.0007; n = 9 mice for C-mCherry, 9 mice for C-hM4Di, 9 mice for S-mCherry, 9 mice for S-hM4Di. Control-mCherry (C-mCherry), Control-hM4Di (C-hM4Di), Stress-mCherry (S-mCherry), Stress-hM4Di (S-hM4Di). G, Experimental procedure. Bilateral injection of AAV2/9-CaMK2α-hM4Di-mCherry in the PL and cannula placement targeted at VTA terminals. H, I, The restraint stress induced decrease of open-arm exploration in the EPM test which was reversed by chemogenetic activation of PL-VTA projection. H, One-way ANOVA with Tukey's multiple-comparisons test, F_(2,24) = 19.41, p < 0.0001; C-mCherry versus S-mCherry, p = 0.0003; S-mCherry versus S-hM3Gq, p < 0.0001. I, One-way ANOVA with Tukey's multiple-comparisons test, F_(2,24)= 4.495, p = 0.0220; C-mCherry versus S-mCherry, p = 0.0172; S-mCherry versus S-hM3Gq, p = 0.4606. J, K, Chemogenetic activation of PL-VTA projection recued the latency to food (J), but not food consumption (K). J, One-way ANOVA with Tukey's multiple-comparisons test, F_(2,24)= 5.279, p = 0.0126; C-mCherry versus S-mCherry, p = 0.0207; S-mCherry versus S-hM3Gq, p = 0.0313. The data are presented as the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Our results show that PL-MD projection and PL-VTA projection have antagonistic effects on anxiety-like behaviors and inhibition of PL-MD projection can reverse anxiety-like behaviors induced by stress. We next explored whether activating PL-VTA projection had the same effect and found that stimulating PL-VTA projection attenuated anxiety-like behaviors induced by 3 d of restraint stress (Fig. 7G–L).

To further elucidate key PL neuron subtypes that regulate different aspects of anxiety, we retrogradely labeled VTA- or MD-projecting PL neurons with CTB and conducted patch sequencing (Fig. 8A). Numerous genes were differentially expressed in VTA- or MD-projecting PL neurons (Fig. 8B), confirming that they are two independent neuronal populations. We further analyzed the differentially expressed genes between these two subpopulations and found that Htr2a and Gabra1, genes encoded hydroxytryptamine receptor 2A (5TH-2A) and gamma-aminobutyric acid type A receptor subunit alpha1 (GABAAR-α1), respectively, were higher in VTA-projecting neurons (Fig. 8B), which are targets for the antidepressant and antianxiety drugs (Soyka and Longo, 2017; Kim et al., 2020; Medina-Puche et al., 2020; Dubovsky and Marshall, 2022). Ctgf and SLC17A6, genes encoded connective tissue growth factor (CTGF) and vesicular glutamate transporter 2 (vGlut2), were higher in MD-projecting neurons (Fig. 8B). It has been reported that CTGF expression was significantly increased in major depressive disorder compared with control subjects (Turner et al., 2018). Activation of vGlut2 neurons in the auditory thalamus induced anxiety-like behaviors (Y. Wang et al., 2023). These results suggest that distinct gene expression in VTA- or MD-projecting PL neurons may contribute to different aspects of anxiety.

Figure 8.

Neuronal type-specific RNA-seq. A, Strategy of neuronal type-specific RNA-seq. B, Heatmap showing PL-VTA and PL-MD neuron-enriched genes from four numbered individual mice.

Discussion

There are distinct subpopulations of PL neurons that project to different downstream targets (Murugan et al., 2017; Anastasiades and Carter, 2021). Here, using a combination of in vivo Ca²⁺ recording, circuit tracing, and chemogenetic and optogenetic manipulation, we defined MD-projecting PL neurons that have a crucial role in modulating anxiety and fear. Our results demonstrated that the PL-MD pathway is sufficient and necessary for maintaining anxiety states and regulating remote but not recent fear memory retrieval. Notably, the PL-MD projection implemented top-down control of the stress-induced high-anxiety state.

The MD is one of the largest thalamic nuclei, and it participates in several corticosubcortical networks, primarily those involving the PFC (Klein et al., 2010; Yuan et al., 2015). Pathological involvement of the MD or its cortical connections contributes to cognitive impairment (Pergola et al., 2018), epilepsy (Bertram, 2014), and schizophrenia (Pergola et al., 2015). A study found that corticothalamic (CT) neurons in PL layers 5 and 6 projected to the MD, and thalamocortical neurons in the MD in turn primarily activated layer 2/3 corticocortical (CC) neurons. Additionally, researchers evaluated local connections from layer 2/3 CC neurons to layer 5 CT neurons working as an intermediary between thalamocortical and corticothalamic networks (Collins et al., 2018). Consistently, our findings also indicate that PL neurons in layers 5 and 6 are strongly driving neurons in the MD. Despite the comprehensive characterization of the connection between MD and mPFC, it remains unknown whether and how this pathway is involved in modulating emotional response. In our study, optogenetic activation of the PL-MD projection induced elevated anxiety states in mice, and chemogenetic inhibition of the PL-MD projection demonstrated an anxiolytic effect. These findings provide novel evidence that the PL-MD pathway is both necessary and sufficient for the regulation of anxiety states in mice. Since the connection between PL and MD is bidirectional, further research is required to investigate the role of MD-PL projection in regulating anxiety. Notably, our findings revealed that synaptic inputs onto MD-projecting PL neurons were rebalanced by increasing excitation and reducing inhibition in stressed mice. Consistent with the involvement of this pathway in pathological anxiety, chemogenetic inhibition of PL-MD projection remitted the stress-induced anxiogenic effect.

It has been reported that the mPFC contains three different subtypes of neurons that fire preferentially when animals are in closed, the center, or open arms in the EPM (Adhikari et al., 2011). Our study identified a subpopulation of PL neurons, the MD-projecting PL neurons, were preferentially active when mice entered into the open arms. Moreover, we found that the activity of MD-projecting PL neurons exhibited an immediate increase when the mice engaged in head dipping. Combined with our optogenetic and chemogenetic manipulation of PL-MD during anxiety-related assays, we suspect that when animals were in the open arms, an anxiety-provoking environment, MD-projecting PL neurons keep firing to induce behavioral avoidance. Similar results show that vCA1 neurons exhibit stable representations of anxiogenic environments that are required for anxiety-like behaviors (Jimenez et al., 2018). As the vCA1 neurons send projections to the PL, how the PL and vCA1 neurons interact to represent anxiety states needs to be further studied.

Our results showed that activation of PL-VTA showed anxiolytic effect in the EPM test but had no effects in the RTPA test and NSFT. The RTPA test was conducted in an environment with low light lux (Jimenez et al., 2018), while the EPM test induced a higher basal anxiety level using intensive light lux (Walf and Frye, 2007; Gehrlach et al., 2019). Therefore, no effects of activating PL-VTA in the RTPA test might be due to the mice's low anxiety levels. The NSFT utilizes the stress-induced (intensive light exposure) impact on eating behaviors and measures the latency to consume food as an indicator of anxiety-like behavior (Francois et al., 2022). Therefore, the observed feeding behavior may potentially impact the outcomes of NSFT. The VTA comprises a heterogeneous population of cells with distinct cytoarchitecture, connectivity, and function, encompassing dopaminergic, glutamatergic, and GABAergic neurons. The dopamine neurons play a critical role in driving the motivation to engage in food-seeking behavior (Wise, 2004; Salamone and Correa, 2012). On the other hand, VTA glutamatergic neurons elicit innate defensive behaviors (Barbano et al., 2020, 2024). Therefore, we speculate that the PL projection neurons may mainly innervate local GABA neurons in VTA. The activation of PL-VTA pathway is likely to inhibit VTA glutamatergic neurons and alleviates innate fear responses under intensive light exposure during NSFT. Simultaneously, the PL-VTA activation also suppresses the motivation of feeding by inhibiting the activity of dopamine neurons, both of these have opposite effect on the latency to consume food, resulting in no significant difference observed in the NSFT. The identification of the specific VTA neuron subtype targeted by PL projections necessitates further investigation.

Previous work demonstrated that pharmacologically silencing PL neurons impairs the expression of learned fear (Corcoran and Quirk, 2007). A recent study showed that reducing PL activity impaired remote memory retrieval, and unbiased whole-brain analyses revealed a set of cortical regions; many of these regions were preferentially activated by PL neurons during remote memory retrieval (DeNardo et al., 2019). Accordingly, several studies have managed to manipulate the PL and its subcortical downstream targets in memory retrieval tasks. The PL-BLA projection is crucial for both recent and remote fear memory, as impairment of memory retrieval 2 weeks after contextual fear conditioning was observed when PL neuronal terminals in the BLA were inhibited (Kitamura et al., 2017), and other study found that silencing PL-BLA projection impaired recent tone-cued fear memory (Do-Monte et al., 2015). Additionally, the PL is capable of drive tone-cued fear memory via its projections to the PVT, as silencing PL-PVT projection impaired remote memory (Do-Monte et al., 2015). Similarly, we found PL-MD was important for remote but not recent cued fear memory. Whether PL-MD and PL-PVT are involved in contextual fear memory needs further investigation. Moreover, the precise circuit-specific function of the PL contributes to recent but not remote memory remains to be further investigated. Nevertheless, all of them suggested a time-dependent recruitment of the PL neural circuits required for fear memory retrieval.

In summary, the MD contributes to anxiety and fear retrieval by top-down supporting activity from subpopulations in the PL. These results emphasize the necessity for more comprehensive subregion investigations in the prefrontal cortex to dissect the neural foundations of PTSD in patients, aiming toward circuit-based therapies in future advancements.

Code Availability

The analysis code will be made available from the corresponding author upon reasonable request.

Data Availability

Data will be made available from the corresponding author upon reasonable request.