Abstract
Anxiety disorders are debilitating psychiatric diseases that affect ∼16% of the world's population. Although it has been proposed that the central nucleus of the amygdala (CeA) plays a role in anxiety, the molecular and circuit mechanisms through which CeA neurons modulate anxiety-related behaviors are largely uncharacterized. Soluble epoxide hydrolase (sEH) is a key enzyme in the metabolism of polyunsaturated fatty acids (PUFAs), and has been shown to play a role in psychiatric disorders. Here, we reported that sEH was enriched in neurons in the CeA and regulated anxiety-related behaviors in adult male mice. Deletion of sEH in CeA neurons but not astrocytes induced anxiety-like behaviors. Mechanistic studies indicated that sEH was required for maintaining the the excitability of sEH positive neurons (sEHCeA neurons) in the CeA. Using chemogenetic manipulations, we found that sEHCeA neurons bidirectionally regulated anxiety-related behaviors. Notably, we identified that sEHCeA neurons directly projected to the bed nucleus of the stria terminalis (BNST; sEHCeA–BNST). Optogenetic activation and inhibition of the sEHCeA–BNST pathway produced anxiolytic and anxiogenic effects, respectively. In summary, our studies reveal a set of molecular and circuit mechanisms of sEHCeA neurons underlying anxiety.
SIGNIFICANCE STATEMENT Soluble epoxide hydrolase (sEH), a key enzyme that catalyzes the degradation of EETs, is shown to play a key role in mood disorders. It is well known that sEH is mostly localized in astrocytes in the prefrontal cortex and regulates depressive-like behaviors. Notably, sEH is also expressed in central nucleus of the amygdala (CeA) neurons. While the CeA has been studied for its role in the regulation of anxiety, the molecular and circuit mechanism is quite complex. In the present study, we explored a previously unknown cellular and circuitry mechanism that guides sEHCeA neurons response to anxiety. Our findings reveal a critical role of sEH in the CeA, sEHCeA neurons and CeA-bed nucleus of the stria terminalis (BNST) pathway in regulation of anxiety-related behaviors.
Introduction
Anxiety disorders, which have a global lifetime prevalence of ∼16%, rank as a leading cause of disability and global disease burden and are the most common and debilitating psychiatric disorder (Mnookin, 2016; World Health Organization, 2017; Huang et al., 2019; Purves et al., 2020). However, the mechanism of anxiety disorders remains elusive. Evidence from human, primate and rodent studies has implicated that the central nucleus of the amygdala (CeA) plays a critical for the pathogenesis of anxiety disorders. (Kalin et al., 2004; Tye et al., 2011; Birn et al., 2014; Calhoon and Tye, 2015; Gilpin et al., 2015; Tovote et al., 2015; Ahrens et al., 2018; Dedic et al., 2018; Pomrenze et al., 2019; Griessner et al., 2021). Whereas, the CeA-related molecular mechanism and neural circuits involved in anxiety remain incompletely explored.
The brain is highly enriched with polyunsaturated fatty acids (PUFAs; Bazinet and Layé, 2014), especially arachidonic acid (ARA), which has been implicated in the regulation of mood disorders, such as anxiety disorders and depression (Nery et al., 2008; Kim et al., 2011; Mocking et al., 2013). ARA is metabolized to the biologically active product eicosanoid through cyclooxygenase, lipoxygenase and cytochrome P450. The P450 epoxygenase pathway metabolizes ARA to hydroxyeicosatetraenoic acids (n-HETEs) and epoxyeicosatrienoic acids (EETs; Harris and Hammock, 2013; Morisseau and Hammock, 2013; Atone et al., 2020). Soluble epoxide hydrolase (sEH), a key enzyme that catalyzes the degradation of EETs, is involved in anorexia nervosa and mood disorders (Scott-Van Zeeland et al., 2014; Ren et al., 2016; Lee et al., 2019; Xiong et al., 2019). In the prefrontal cortex, sEH is mostly localized in astrocytes and regulates depressive-like behaviors (Marowsky et al., 2009; Xiong et al., 2019). Intriguingly, sEH is suggested to be expressed in neurons in the CeA region (Marowsky et al., 2009). However, the detailed expression pattern of sEH in the CeA and its functions are still unknown.
Here, we found that deletion of sEH in CeA neurons promoted anxiety-like behaviors. Mechanistic studies showed that inhibition of sEH, deletion of sEH both reduced the excitability of sEH-positive neurons in the CeA (sEHCeA neurons). Furthermore, combining chemogenetic methods, viral tracing and optogenetics, we dissected the functional organization of the sEHCeA neurons which then drove downstream targets in the bed nucleus of the stria terminalis (BNST) to modulate anxiety-related behaviors. Taken together, we uncover a previously unknown cellular and circuitry mechanism underlying CeA-involved modulation of anxiety-related behaviors.
Materials and Methods
Animals
All animal protocols were approved by the Southern Medical University Animal Ethics Committee. All experiments were performed on adult male mice (8–16 weeks old). Mice were housed in groups (three to five per cage) and under controlled temperature (22–25°C) conditions in a 12/12 h light/dark cycle (lights on from 7 A.M. to 7 P.M.) with ad libitum access to food and water. Male C57BL/6J mice (aged 8–12 weeks) were obtained from the Southern Medical University Animal Center (Guangzhou, China).
The loxp-flanked Ephx2 mouse line (Ephx2loxp/loxp) was generated as described in our previous report (Xiong et al., 2019). Ephx2-iCreERT2 mice were constructed in Shanghai Model Organisms Center (Shanghai, China). This model was generated by CRISPR/Cas9 inserting 2A-CreERT2 at the termination codon of the ephx2 gene (MGI: 99500) through homologous recombination. The donor vector, constructed by In-Fusion cloning, included a 4.822 kb 5′ homologous arm, 2A-CreERT2 and a 4.909 kb 3′ homologous arm. Cas9 mRNA, gRNA and donor vector were microinjected into the germ cells of C57BL/6J mice to obtain F0 pups. F0 mice were genotyped with the following primers: forward: 5′-CCGAGGCTGAACTGGAGAAGA-3′ and reverse: 5′-CAGCATTTGTGAACAGAAGGGTC-3′. According to the PCR results, the targeted F0 mice were identified. F0 mice were backcrossed with C57BL/6J mice to obtain F1 mice. The F1 pups were screened for the desired mutant allele using prior genomic PCR genotyping. The Ephx2-iCreERT2 mice were genotyped with the following primers: 5′-TCTGCCCCGCTCAGGTCT-3′, 5′-ATGGCACAGGCATACTCAAA-3′, and 5′-TTCCAGGTATGCTCAGAAAACG-3′. The expected PCR products were one band with 342 bp for wild-type, two bands with 342 and 369 bp for heterozygous and one band with 369 bp for homozygous animals. Cre-dependent ROSA26-tdTomato reporter mouse lines (The Jackson Laboratory catalog #007909, RRID: IMSR_JAX:007909) were purchased from The Jackson Laboratory. Ephx2-tdTomato mice were generated by crossing Ephx2-iCreERT2 mice with ROSA26-tdTomato mice. In order to excise the loxp sites by Cre recombination, two-month-old male mice (Ephx2-tdTomato and Ephx2-iCreERT2 mice) were intraperitoneally injected with tamoxifen (TAM; Sigma-Aldrich catalog #T5648) once a day (100 mg kg−1) for 5 d. TAM was dissolved in corn oil (Sigma-Aldrich catalog #C8267) at a final concentration of 10 mg ml−1. For fiber photometry recording and chemogenetic and optogenetic experiments, TAM was injected 7 d after virus injection. For immunofluorescence, experiments were performed one month after TAM injection. Behavioral tests were conducted on three- or four-month-old male mice studied 30 d or 60 d after the first TAM injection.
All mice were handled per day for 3–4 d before the behavioral experiments, and double-blind behavioral tests were performed between 1 and 4 P.M.
Viruses
The viruses AAV2/9-hSyn-DIO-GCaMP6s-WPRE-pA (viral: 2.09 × 1012 vg ml−1), AAV2/9-hSyn-DIO-hM3D(Gq)-mCherry-WPRE-pA (viral: 7.37 × 1012 vg ml−1), AAV2/9-hSyn-DIO-hM4D(Gi)-mCherry-WPRE-pA (viral: 2.63 × 1012 vg ml−1), AAV2/9-Ef1α-DIO-hChR2(H134R)-EYFP-WPRE-pA (viral: 5.48 × 1012 vg ml−1), AAV2/9-Ef1α-DIO-eNpHR3.0-mCherry-WPRE-pA (viral: 3.25 × 1012 vg ml−1) and AAV2/9-GFAP-CRE-GFP (viral: 2.4 × 1012 vg ml−1) were purchased from BrainVTA (Wuhan, China). The virus AAV2/9-hSyn-FLEX-mGFP-2A-synaptophysin-mRuby (viral: 3.8 × 1012 vg ml−1) and AAV2/2Retro-hEf1α-DIO-EYFP-WPRE-pA (viral: 1.26 × 1013 vg ml−1) was provided by Taitool Bioscience.
The shRNA sequences targeting Ephx2 (GenBank accession: NM_007940) were 5′- GCAGCTGATTGGAGAGTAA-3′. A negative control (NC) sequence (TTCTCCGAACGTGTCACGT) was used as the control shRNA. The specificity and efficiency of the shRNAs were validated, and the engineered AAV (AAV-Ephx2-shRNA, 1.0 × 1011 vg ml−1) and AAV2/9-hSyn-Cre-eGFP- WPRE-pA (viral: 3.59 × 1013 vg ml−1) were produced by Obio Technology Corp. Ltd.
Cannula implantation/virus injection/optical fiber implantation
For pharmacological experiments, a unilateral or bilateral cannula (RWD Life Science) was implanted into the CeA [anteroposterior (AP): −1.06; mediolateral (ML): ±2.8; dorsoventral (DV): −4.4; mm relative to bregma] for infusion of 2 μl sEH inhibitor TPPU at a rate of 1 μl min−1 (1 μm, Sigma-Aldrich, catalog #SML0750, CAS: 1222780-33-7). The cannula and screws were held in place with dental cement. Mice were allowed to recover for one week after implantation. All behavioral tests were performed 30 min after the infusion.
Mice were anesthetized with sodium pentobarbital (50 mg kg−1, i.p.) for bilateral or unilateral stereotaxic injection of viruses into the CeA (AP: −1.06; ML: ±2.8; DV: −4.6; mm relative to bregma). The coordinates were measured from the bregma according to the mouse atlas. A volume of 300-nl virus was injected into each location at a rate of 100 nl min−1. After each injection, the needle was left in place for 6 min and then slowly withdrawn.
For fiber photometry, a ceramic ferrule with an optical fiber [200 μm in diameter, numerical aperture (NA) of 0.37 (Inper)] was implanted with the fiber tip on top of the CeA (bregma −1.06 mm, lateral ±2.8 mm, and dura −4.60 mm) after 30 min of AAV injection. TAM was injected (intraperitoneally) once a day (100 mg kg−1) for 5 d on the eighth day of AAV injection. The GCaMP signals were recorded three weeks after optical fiber implantation. For projection-specific optogenetic manipulations, mice were unilaterally (ChR2) or bilaterally (NpHR) injected with 300 nl of virus into the CeA, and then TAM was injected (intraperitoneally) in the next week. Optic fibers (length 5 mm, NA 0.37; Inper) were implanted over the BNST (AP: +0.14 mm, ML: ±0.9 mm, DV: −4.2 mm) five to seven weeks after virus injection. Behavior tests were conducted 10 d later.
A designer receptor exclusively activated by a designer drug (DREADD)
Ephx2-iCreERT2 mice were bilaterally injected with hM3Dq or hM4Di into the CeA following TAM injection one week later. After four weeks of virus expression, mice were injected with clozapine-N-oxide (CNO; 2 mg/kg, i.p.; Sigma-Aldrich, catalog #C0832, CAS: 34233-69-7) and placed into the test room to habituate for 60 min. Stimulation-induced behaviors were scored manually by a human observer inspecting the videos post hoc (EthoVision by Noldus) while blinded to the underlying conditions.
Fiber photometry recording
GCaMP6s virus injection and optic fiber implantation were performed on Ephx2-tdTomato mice (approximately eight weeks old) on the same day. Fiber optic calcium recording was conducted three weeks after virus injection. A fiber photometry system (Thinker Tech) was used for recording GCaMP signals from genetically identified neurons. To induce fluorescence signals, a laser beam from a laser tube (488 nm) was reflected by a dichroic mirror, focused by a ×10 (NA of 0.3) lens and coupled to an optical commutator. The GCaMP6s fluorescence was bandpass filtered (MF525-39, Thorlabs) and collected by a photomultiplier tube (R3896, Hamamatsu). An amplifier (C7319, Hamamatsu) was used to convert the photomultiplier tube current output to voltage signals, which were further filtered through a low-pass filter (40-Hz cutoff; Brownlee 440). A flashing light-emitting diode triggered by a 1-s square-wave pulse was simultaneously recorded to synchronize the video and GCaMP signals. To minimize photobleaching, the power intensity at the fiber tip was adjusted to 30 μW. For recordings in freely moving mice, mice with optical fibers connected to the fiber photometry system were freely explored in the elevated plus maze (EPM) for 5 min. After the experiments, the optical fiber tip sites were histologically examined in each mouse. Calcium signals for the bulk fluorescence signals were acquired and analyzed with custom-written MATLAB software. The Z-score of a population of neurons was calculated using the formula: Z = (x-y)/SD (where x = FSignal, y = mean of FBaseline, and SD for standard deviation of FBaseline).
Optogenetic manipulations
An optical fiber was initially implanted into the BNST in the brain of an anesthetized mouse that had been immobilized in a stereotaxic apparatus. The implant fiber was secured to the animal's skull with dental cement. Chronically implantable fibers (diameter, 200 μm, Inper) were connected to a laser generator using optic fiber sleeves. The delivery of blue light (470 nm, 10–15 mW, 5-ms pulses, 20 Hz) or yellow light (580 nm, 5–10 mW, constant) was controlled by a stimulator. The same stimulus protocol was applied in the control group. The location of the fiber was examined after all the experiments, and data obtained from mice in which the fibers were outside the desired brain region were discarded.
Brain slice electrophysiology
Mice were anesthetized with isoflurane and perfused with 15- to 20-ml regular artificial CSF (ACSF) containing the following: 126 mm NaCl, 3 mm KCl, 1 mm MgSO4, 2 mm CaCl2, 1.25 mm NaH2PO4, 26 mm NaHCO3, and 10 mm glucose. The brain was then quickly removed and chilled in ice-cold modified ACSF containing the following: 120 mm choline chloride, 2.5 mm KCl, 7 mm MgCl2, 0.5 mm CaCl2, 1.25 mm NaH2PO4, 25 mm NaHCO3, and 10 mm glucose. Coronal brain slices were cut in ice-cold modified ACSF using a VT-1000S vibratome (Leica) and quickly transferred to an incubation chamber containing regular ACSF at 32°C for 30 min and at room temperature (25 ± 1°C) for an additional 1 h before recording. All liquids were saturated with 95% O2/5% CO2 (v/v).
The brain slices were placed in the recording chamber, with continuous perfusion of ACSF at a flow rate of 2 ml min−1. Whole-cell patch-clamp recording from neurons in the CeA region was visualized with infrared optics using an upright microscope equipped with an infrared-sensitive charge-coupled device (CCD) camera (DAGE-MTI, IR-1000E). The pipettes were pulled by a micropipette puller (P-97, Sutter Instrument) with a resistance of 3–5 MΩ. Recordings were made with a MultiClamp 700B amplifier and 1440A digitizer (Molecular Device).
To detect the excitability of CeA neurons, the APs were recorded under the current-clamp mode by injecting a series of gradually increased depolarizing pulses starting from 0 to 80 pA at a step of 20 pA, with the pipette solution including the following: 125 mm Glu-K, 5 mm KCl, 10 mm HEPES, 0.2 mm EGTA, 1 mm MgCl2, 4 mm MgATP, 0.3 mm NaGTP, and 10 mm Na2-phosphocreatine (pH 7.40, 285 mOsm). The resting membrane potential was considered the value without current injection, and the membrane input resistance was calculated in response to a series of hyperpolarizing pulses.
For the drug treatment experiments, slices from Ephx2-tdTomato mice were incubated with drugs (TPPU, 1 μm; 11,12-EET, 2 μm, Cayman Chemical, catalog #50511) or vehicle for 30 min and then subjected to recording.
In all experiments, cells were rejected if the series resistance was larger than 20 MΩ during the recordings or if the series resistance fluctuated by >20% of the initial values. Data were filtered at 1 kHz and sampled at 10 kHz.
Real-time quantitative PCR
Total RNA was obtained from cells or tissues using RNAiso plus (TaKaRa) according to the manufacturer's instructions and was quantified with a Nanodrop 2000 (Thermo Scientific). cDNA was synthesized using the PrimeScript RT Reagent kit (Takara). Quantitative real-time PCR was performed in an ABI 7500 Real-Time PCR System with SYBR premix Ex Taq (TaKaRa). The ΔΔCt method was used to analyze gene expression, and relative mRNA levels were normalized to 18s mRNA levels. The Ephx2 primer sequences were as follows: forward primer, GGACGACGGAGACAAGAGAG; reverse primer, CTGTGTTGTGGACCAGGATG.
Western blot analysis
Mouse brain tissue from the CeA was homogenized in lysis buffer (RIPA) on ice for 30 min and subsequently centrifuged at 12,000 rpm for 20 min at 4°C. Supernatants were transferred to a clean 1.5-ml tube, and samples containing 80 μg of protein were separated using 10% SDS-PAGE gels. Proteins were electrotransferred onto PVDF membranes in ice-cold buffer (25 mm Tris HCl, 192 mm glycine, and 20% methanol) over the course of 2 h. The membranes were blocked with 5% nonfat milk powder dissolved in TBST. The membranes were incubated overnight at 4°C with monoclonal rabbit anti-sEH (1:1000, Shanghai YouKe) and polyclonal mouse anti-GAPDH (1:10,000, Proteintech Group, catalog #60004-1-Ig, RRID: AB_2107436). HRP-conjugated goat anti-mouse and goat anti-rabbit secondary antibodies were purchased from ABclonal (AS014, AS003). Protein abundance was quantified by analyzing the Western blot bands using AlphaEaseFC (WB) software. Quantification of band intensities was normalized to GAPDH and averaged from at least three independent experiments.
Immunofluorescence
Animals were deeply anesthetized with sodium pentobarbital (50 mg kg−1, i.p.) and perfused transcardially with saline followed by 4% PFA in 0.1 m PBS, pH 7.4. Brains were removed, postfixed overnight in 4% PFA at 4°C and transferred to 30% sucrose in 0.1 m PBS, pH 7.4. Coronal sections (40 μm) were cut on a freezing microtome (Leica CM3050 S) and stored in 0.1 m PBS. In some instances, immunoreactivity was increased by incubating the slices in 10 mm Sodium citrate, 0.05% Tween 20, pH 6 for 10 min at 95°C. The sections were incubated in blocking buffer containing1% bovine serum albumin or 10% donkey serum in 0.3% Triton X-100/PBS for 2 h at room temperature and then with primary antibodies in blocking buffer overnight at 4°C. The primary antibodies used were sEH (1:500, Shanghai Youke Biotechnology Co, Ltd.), NeuN (1:300, Millipore, catalog #MAB377, RRID: AB_2298772), S100β (1:200, Abcam, catalog #ab52642, RRID: AB_882426), corticotropin-releasing factor (CRF; 1:1000, Sigma, catalog #C5348, RRID: AB_258893), and protein kinase C δ (PKCδ; 1:500, BD Biosciences, catalog #610398, RRID: AB_397781), and SST(SOM) (1:500, Santa Cruz, catalog #sc-7819, RRID: AB_2302603; Haubensak et al., 2010; Lemos et al., 2012; Onorati et al., 2014). After three washes with PBS, sections were incubated with either Alexa Fluor 488-conjugated or Alexa Fluor 594/568-conjugated secondary antibodies at room temperature for 2 h. After another three washes in PBS, sections were incubated with Vectashield mounting medium containing DAPI (Vector Laboratories Inc.) and immunofluorescence was assessed using a Nikon A1R confocal microscope (Nikon Instruments Inc.).
RNAscope in situ hybridization
The fixed frozen brain slices (10 μm thick) containing CeA were prepared as described above. We performed single-molecule fluorescence in situ hybridization using RNAscope fluorescence detection assays and probes (Advanced Cell Diagnostics: RNAscope Probe-Mm-Ephx2 #558701; RNAscope Probe-Mm-Prkcd-C2 #441791-C2; RNAscope Probe -Mm-Sst-C2 #404631-C2; RNAscope Probe-Mm-Crh-C2 #422791) in eight- to nine-week-old C57 mice according to the manufacturer's protocols. Slides showing poor staining were not analyzed.
Fluorescence microoptical sectioning tomography (fMOST)
AAV2/9-Ef1α-DIO-EYFP- WPRE-pA was injected into the CeA of Ephx2-iCreERT2 mice to trace the CeAsEH neuron circuit with fMOST system (OE-bio, Co, Ltd.). The whole-brain precise imaging has been described previously (Lin et al., 2018). Briefly, the GMA-embedded mouse brains were imaged by the fMOST system. Whole-brain imaging, imaging process and reconstruction were performed by OE-bio according to the manufacturer's protocols.
EPM
The EPM apparatus comprised two open arms (30 × 5 × 0.5 cm), two closed arms (30 × 5 × 15 cm), and a central platform (5 × 5 cm) elevated 50 cm above the underlying surface. Mice were placed in the center facing one of the two open arms and allowed to explore for 5 min (except the optogenetic experiments). For the optogenetic experiments, individual mice were connected to the patch cable and allowed 1–5 min in the center for recovery from handling before the 9-min session was initiated. Each session was divided into three alternating 3-min epochs: laser stimulation off, stimulation on, and stimulation off (OFF-ON-OFF epochs). Anxiety-related behavior was assessed by time traveled within the open arms. The duration time in the open and closed arms was tracked and recorded by a video tracking system (EthoVision by Noldus).
Novelty-suppressed feeding test
After 24 h of food deprivation and water available ad libitum, mice were placed in a brightly lit open arena (50 × 50 × 50 cm) containing clean wood chip bedding. A filter paper (8 × 8 cm) was placed in the center of the arena, and one familiar food pellet was placed in the center of the filter paper. Mice were removed from their home cage, placed in a holding cage for 60 min before testing and then placed in a corner of the testing arena. For optogenetic experiments, the task was repeated twice on different days for each mouse, counterbalanced for no stimulation (OFF) or light stimulation (ON). The task ended when the mice began to eat the food pellet. The latency to begin a feeding episode was recorded with a video camera suspended above the arena and saved for further analysis (EthoVision by Noldus). Immediately after testing, mice were removed from the arena and placed into their home cage to measure food consumption for 5 min.
Open-field test (OFT)
Mice were placed in an open chamber (Accuscan Instruments) with transparent, plastic walls and allowed to explore freely for 5 min (except the optogenetic experiments). Before the start of the optogenetic experiments, mice were allowed to recover from handling for 1–5 min and then placed in the center of the open field. The OFT consisted of a 9-min session in which there were three alternating 3-min epochs (OFF-ON-OFF epochs). Behavioral tests were recorded by a video camera. The total distance traveled across a session was analyzed by using Versmax analyzer software.
Statistical analysis
All experiments and data analyses were conducted blindly. The number of experimental replicates (n), t, F, and P are indicated in the text and refers to the number of experimental subjects independently treated in each experimental condition. Statistical comparisons were performed using Graph-Pad Prism and SPSS 20.0 software with appropriate inferential methods. Data are presented as the mean ± SEM. Two-tailed Student's t tests, one-way ANOVA, and two-way ANOVA were used for statistical analyses, and differences were considered to be significant if p < 0.05. Significance levels are indicated as follows: *p < 0.05, **p < 0.01, and ***p < 0.001.
Results
sEH is enriched in neurons in the CeA and is associated with anxiety-related behaviors
To characterize the expression pattern of sEH in the CeA, we employed a specific antibody for sEH (Xiong et al., 2019) and co-stained with NeuN, a marker for neurons, on brain slices from adult C57BL/6J mice. The immunofluorescence results revealed that sEH was present in part of CeA neurons (Fig. 1A). Moreover, we generated the Ephx2-tdTomato mice by crossing Ephx2-iCreERT2 mice (a mouse line expressing Cre recombinase under the control of the promoter ephx2, which encodes sEH protein) with Ai9 mice (Cre-dependent ROSA26-tdTomato reporter mouse line). We found that ∼58% of Ephx2-tdTomato (sEH-tdTomato) cells in the CeA co-stained with NeuN, whereas ∼34% co-stained with S100β, a marker for astrocytes (t(10) = 4.025, p = 0.0024, unpaired t test; Fig. 1B–D). These results indicate that sEH is enriched in neurons in the CeA. To identify the neuron types that sEH is expressed, we performed fluorescence in situ hybridization analysis for mRNA localization. The results showed that sEH colocalized with CRF (17.6 ± 5.5%), Somatostatin (SOM; 25.3 ± 5.1%), or PKCδ (31.6% ± 5.8%), which are the main cell markers of GABAergic neurons in the CeA, respectively (Fig. 1E,F).
To detect whether sEH in the CeA is involved in anxiety, we subjected adult C57BL/6J mice to the EPM and novelty suppressed feeding test (NSF). Mice was separated into high-anxiety and low-anxiety subgroups based on their anxiety levels [Bi et al., 2015; Fig. 1G–I; time in the open arm, t(12) = 10.42, p = 0.000, time in the closed arm, t(12) = 4.874, p = 0.000 (Fig. 1H); t(12) = 6.25, p = 0.000; unpaired t test (Fig. 1I)]. Real-time quantitative fluorescence PCR analysis showed that sEH mRNA expression in the CeA was significantly lower in the high-anxiety subgroup than that in the low-anxiety subgroup (t(12) = 3.052, p = 0.0101, unpaired t test; Fig. 1J).More importantly, there was positive and significant correlations between the sEH mRNA level and the time in open arm (%; r = 0.5665, R2 = 0.3209, p = 0.0347, Pearson r; Fig. 1K). These results suggest that sEH in the CeA is associated with anxiety-related behaviors.
Neuronal sEH in the CeA modulates anxiety-related behaviors
To investigate the role of sEH in anxiety-related behaviors, we infused the sEH inhibitor 1-(1-propionylpiperidin-4-yl)−3-(4-(trifluoromethoxy)-phenyl)urea (TPPU; 1 μm; Wan et al., 2019) unilaterally or bilaterally into the CeA of adult C57BL/6J mice (Fig. 2A,G). TPPU treatment significantly reduced the duration of open-arm exploration in the EPM and increased latency to feed in the NSF, without an alteration in food consumption [Fig. 2B–D,H–J; time in the open arm, t(20) = 2.235, p = 0.037, time in the closed arm, t(20) = 2.103, p = 0.0483 (Fig. 2B); t(25) = 2.263, p = 0.0326; 2D, t(25) = 0.2355, p = 0.8157 (Fig. 2C); time in the open arm, t(22) = 2.355, p = 0.0278, time in the closed arm, t(22) = 3.56, p = 0.0018 (Fig. 2H); t(20) = 2.409, p = 0.0258 (Fig. 2I); t(20) = 0.1515, p = 0.8811 (Fig. 2J); unpaired t test). In addition, TPPU did not influence performance in the OFT (Fig. 2E,F,K,L; t(25) = 0.4462, p = 0.6593 (Fig. 2E); t(25) = 0.9121, p = 0.3704 (Fig. 2F); t(22) = 0.1208, p = 0.9049 (Fig. 2K); t(22) = 0.9123, p = 0.3715; unpaired t test (Fig. 2L)]. These observations indicate that sEH in the CeA is required for regulation of anxiety-related behaviors.
Because of the limitations of the pharmacological study, we next bilaterally injected AAV-Ephx2-shRNA into the CeA of adult C57BL/6J mice (Fig. 3A). Western blot analysis indicate that the sEH level in the CeA was significantly reduced in sEH knock-down (refer to as sEH-KD) mice (t(10) = 2.572, p = 0.0278, unpaired t test; Fig. 3B). The sEH-KD mice showed shorter duration of open-arm exploration in the EPM and a longer latency to feed during NSF [time in the open arm, t(26) = 2.666, p = 0.013, time in the closed arm, t(26) = 2.638, p = 0.0139 (Fig. 3C); t(26) = 2.502, p = 0.019; unpaired t test (Fig. 3D)] compared with control mice. sEH-KD mice showed no difference on food consumption and performance in the OFT [Fig. 3E–G; t(26) = 1.246, p = 0.224 (Fig. 3E); t(27) = 0.6703, p = 0.5084 (Fig. 3F); t(27) = 0.5522, p = 0.5853 (Fig. 3G); unpaired t test].
To further determine whether sEH in neurons is required for anxiety-related behaviors, we selectively deleted neural sEH in the CeA of adult Ephx2loxp/loxp mice by bilateral injection of AAV-hSyn-Cre virus (Fig. 4A). Western blot analysis showed that sEH protein levels were markedly decreased in sEH conditional knockout (cKO) mice (t(8) = 2.46, p = 0.0393, unpaired t test; Fig. 4B). The cKO mice exhibited significantly decreased duration of open-arms exploration in the EPM and increased latency to feed in NSF [time in the open arm, t(20) = 2.625, p = 0.0162, time in the closed arm, t(20) = 1.997, p = 0.0596 (Fig. 4C); t(20) = 2.447, p = 0.0238 (Fig. 4D); unpaired t test], while there were no difference in food consumption and performance in the OFT between the two groups [t(20) = 1.496, p = 0.1504 (Fig. 4E); t(24) = 0.2382, p = 0.8138 (Fig. 4F); t(24) = 0.09,818, p = 0.9226 (Fig. 4G); unpaired t test].
To examine whether astrocytic sEH also play a role in regulation of anxiety, we bilaterally injected AAV-gfaABC1D-cre virus into the CeA of Ephx2loxp/loxp mice. We found that deletion of sEH in astrocytes in the CeA showed little effects on latency to feed in the NSF, duration of open-arm exploration in the EPM, or center time in the OFT [Fig. 4H–M; time in the open arm, t(19) = 0.6195, p = 0.5429, time in the closed arm, t(19) = 0.7746, p = 0.4481 (Fig. 4I); t(19) = 0.06,975, p = 0.9451 (Fig. 4J); t(19) = 0.2226, p = 0.8262 (Fig. 4K); t(16) = 1.859, p = 0.0815 (Fig. 4L); t(16) = 0.07747, p = 0.9392 (Fig. 4M); unpaired t test], suggesting a dispensable role of astrocytic sEH in the CeA for regulation of anxiety-related behaviors. Altogether, these observations demonstrate that neuronal but not astrocytic sEH in the CeA is critical for regulation of anxiety-related behaviors.
sEH mediates the excitability of sEH-positive neurons in the CeA
To address how the sEH in CeA neurons modulates anxiety-related behaviors, we measured the excitability of sEHCeA neurons of Ephx2-tdTomato mice by electrophysiological whole-cell recording (Fig. 5A). While TPPU treatment showed little effects on the resting membrane potential (RMP) or input resistance of sEHCeA neurons [t(28) = 0.02999, p = 0.9763 (Fig. 5B); t(28) = 1.467, p = 0.1534 (Fig. 5C); unpaired t test], the firing frequencies of sEHCeA neurons in response to gradually increased current injections were decreased in TPPU group compared with those in vehicle group (Fig. 5D and 5E, 5E, drug factor, F(1,140) = 5.115, p = 0.0253; two-way ANOVA). Further analysis indicated that the firing threshold of the first action potential (AP) was elevated by TPPU treatment (Fig. 5F,G, t(28) = 2.098, p = 0.0451; unpaired t test). On the other hand, TPPU treatment showed no effects on sEH-negative neurons [Fig. 5H–N; t(20) = 0.8129, p = 0.4258 (Fig. 5I); t(20) = 0.8345, p = 0.4139 (Fig. 5J); t(22) = 0.5923, p = 0.5597 (Fig. 5N), unpaired t test; drug factor, F(1,100) = 0.2981, p = 0.5863 (Fig. 5L); two-way ANOVA]. These results suggest inhibition of sEH reduces the excitability of sEHCeA neurons in a cell-autonomous manner.
To further examine the role of sEH in regulation of neuronal excitability, we injected AAV-Ephx2-eGFP-shRNA or control virus into Ephx2-tdTomato mice (Fig. 6A). The firing frequencies were substantially decreased in sEHCeA neurons infected with Ephx2-shRNA, with no change on RMP or input resistance, when compared with those in sEHCeA neurons infected with control-shRNA [Fig. 6B–E; t(25) = 0.4312, p = 0.67 (Fig. 6B); t(25) = 0.7377, p = 0.4676 (Fig. 6C), unpaired t test; virus factor, F(1,125) = 25.84, p = 0.000 (Fig. 6E); two-way ANOVA]. Moreover, The sEHCeA neurons from Ephx2-shRNA mice had a more depolarized firing threshold in comparison with those from control-shRNA mice (Fig. 6F,G, t(24) = 2.126, p = 0.044, unpaired t test). Again, there were no difference in sEH-negative neurons between Ephx2-shRNA and control-shRNA mice [Fig. 6H–N; t(22) = 0.8001, p = 0.4322 (Fig. 6I); t(22) = 1.078, p = 0.2929 (Fig. 6J); t(22) = 0.7389, p = 0.4678 (Fig. 6N), unpaired t test; virus factor, F(1,110) = 0.5096, p = 0.4768 (Fig. 6L); two-way ANOVA]. These data emphasized the indispensable role of sEH in regulation of excitability of sEHCeA neurons.sEH is a key enzyme of the ARA-EET pathway that catalyzes the degradation of EETs. It has been reported that 11,12-EET substantially reduced the excitability of CA1 pyramidal cells (Mule et al., 2017). To test whether 11,12-EET play a role in regulation of the excitability of sEHCeA neurons, we recorded the firings of sEHCeA neurons in Ephx2-tdTomato mice (Fig. 7A). 11,12-EET (2 μm) treatment significantly reduced the firing frequencies with no effects on RMP or input resistance [Fig. 7B–E; t(46) = 0.2598, p = 0.7961 (Fig. 7B); t(46) = 1.435, p = 0.1581 (Fig. 7C), unpaired t test; drug factor, F(1,230) = 9.578, p = 0.0022 (Fig. 7E); two-way ANOVA]. We further analyzed the shape of AP and found that the sEHCeA neurons in the 11,12-EET treatment group had a more depolarized firing threshold compared with neurons in the vehicle treatment group (Fig. 7F,G, t(46) = 3.075, p = 0.0035, unpaired t test). These data suggest that 11,12-EET inhibits the excitability of sEHCeA neurons through elevation of firing threshold.
The activity of CeAsEH neurons is associated with anxiety-related behaviors
To determine whether the activity of sEHCeA neurons is associated with anxiety, we performed cell-type-specific fiber photometry during exploration of innately anxiogenic environments in the EPM. AAV-hSyn-DIO-GCaMP6s virus was injected into the CeA of Ephx2-tdTomato mice and a fiber was implanted above the infected cells (Fig. 8A). This strategy resulted in the expression of GCaMP6s in sEHCeA neurons (Fig. 8B). We compared the GCaMP6 signals of sEHCeA neurons across successive behavioral entrances and exploration bouts of the arms in the EMP. The activity of sEHCeA neurons was significantly increased when mice entered the open arm from closed arm, but not the open arm to open arm, open arm to closed arm or closed-arm to closed-arm (Fig. 8C,D; F(3,12) = 6.466, p = 0.0075; one-way ANOVA). These findings suggest that the activity of CeAsEH neurons may be involved in regulation of anxiety-related behaviors.
Bidirectional manipulations of the activity of sEHCeA neurons in regulation of anxiety-related behaviors
To define the role of the activity of sEHCeA neurons in anxiety-related behaviors, we bidirectionally injected the AAV-hSyn-DIO-hM3D(Gq)-mCherry virus into the CeA of Ephx2-iCreERT2 mice to specifically manipulate the activity of sEHCeA neurons (Fig. 9A,B). Confocal images showed that hM3Dq was expressed in sEHCeA neurons (Fig. 9C). Electrophysiologically, the CNO (1 μm) significantly induced firing in hM3Dq-expressing sEHCeA neurons (Fig. 9D). We found that intraperitoneal injection of CNO (2 mg kg−1) significantly increased the duration of open-arm exploration in the EPM, reduced the latency to feed without altering the food consumption in the NSF when compared with those in saline group [time in the open arm, t(27) = 2.321, p = 0.0281, time in the closed arm, t(27) = 3.564, p = 0.0014 (Fig. 9E); t(23) = 3233, p = 0.0037 (Fig. 9F); t(23) = 0.3377, p = 0.7387 (Fig. 9G); unpaired t test]. CNO had little effect on the performance in the OFT (t(23) = 0.6932, p = 0.4951; unpaired t test; Fig. 9H). These data suggest that activation of sEHCeA neurons produced an anxiolytic effect. In contrast, when we expressed hM4Di virus (AAV-hSyn-DIO-hM4D(Gi)-mCherry) into the CeA of Ephx2-iCreERT2 mice (Fig. 9I), CNO significantly inhibited the firing of sEHCeA neurons (Fig. 9J). Behaviorally, intraperitoneal injection of CNO dramatically reduced the duration of open-arm exploration in the EPM and increased the latency to feed without altering food consumption in the NSF, whereas it had little effect on the performance in the OFT [time in the open arm, t(12) = 2.731, p = 0.0182, time in the closed arm, t(12) = 1.671, p = 0.1206 (Fig. 9K); t(18) = 2.191, p = 0.0419 (Fig. 9L); t(18) = 1.004, p = 0.3289 (Fig. 9M); t(24) = 1.321, p = 0.1989 (Fig. 9N); unpaired t test]. These results demonstrate the critical role of the activity of sEHCeA neurons in regulation of anxiety-related behaviors.
Outputs of sEHCeA neurons and characterization of BNST-projecting CeA neurons
To investigate the circuitry mechanism underlying theeffects of sEHCeA neurons on anxiety-related behaviors, we traced the axonal projections of sEHCeAneurons by injection of the AAV2/9-Ef1α-DIO-EYFP-WPRE-pA virus into the CeA of Ephx2-iCreERT2 mice with fMOST system (OE-bio, Co, Ltd.).We observed a dense EYFP signal in the BNST, and a moderate signal in the substantia nigra pars lateralis (SNL) regions (Movie 1), implicating the existence of projection pathways from sEHCeA neurons to BNST and SNL. This is further verified by injecting AAV-hSyn-FLEX-mGFP-2A-Synaptophysin-mRuby virus, which is capable of indicating the terminal of projections, into the CeA of Ephx2-iCreERT2 mice (Fig. 10A). As shown in Figure 10B, there was a dense signal of both mCherry and GFP in the BNST and SNL (Fig. 10B; Extended Data Fig. 10-1A). Together, these results demonstrate that sEHCeA neurons directly project to BNST and SNL. There were no obvious fluorescence signals in other brain regions (Extended Data Fig. 10-1B–E). While, it is known that the BNST is also strongly implicated in maladapted anxiety disorder (Kim et al., 2013; Marcinkiewcz et al., 2016). We mainly focused on the circuit of sEHCeA-BNST. To determine which population of sEHCeA neurons project to BNST, we back-labeled sEHCeAneurons that projected to BNST by injecting AAV2-Retro-Ef1α-DIO-EYFP virus into the BNST of Ephx2-creERT2 mice (Fig. 10C), and co-stained with CRF, SOM, and PKCδ antibodies (Fig. 10D–F). As shown in Figure 9G, 47.2% of sEHCeA-BNST neurons co-labeled PKCδ, and 19.7% co-labeled CRF. There was no colocalization between sEHCeA-BNST and SOM (Fig. 10G).
Extended Data Figure 10-1
Outputs of sEH-positive CeA neurons. Related to Figure 10. A, Representative images of mCherry and eGFP expression in the SNL of Ephx2-iCreERT2 mice after CeA injection of AAV-hSyn-FLEx-mGFP-2A-Synaptophysin-mRuby virus. Scale bars: 500 μm (left) and 50 μm (right). B–E, No mCherry or eGFP signal in the ventral hippocampus (vHPC), medial prefrontal cortex (mPFC), VTA, or striatum. Scale bars: 1000 μm (B) and 500 μm (C–E). Download Figure 10-1, TIF file.
Optogenetic manipulation of the sEHCeA-BNST circuit bidirectionally governs anxiety-related behaviors
To test the functional role of sEHCeA-BNST neuronal pathway in anxiety, we bilaterally injected ChR2-eYFP (AAV2/9-Ef1α-DIO-hChR2(H134R)-EYFP-WPRE-pA) or eYFP (AAV2/9-Ef1α-DIO- -EYFP-WPRE-pA) virus into the CeA of Ephx2-iCreERT2 mice. An optic fiber was inserted above the BNST to allow for the delivery of blue (470 nm) light to the axons in the BNST (Fig. 11A). Confocal images showed dense expression of ChR2 protein on sEHCeA neurons and axon terminals in the BNST (Fig. 11B). We performed optogenetic modulation of the sEHCeA-BNST circuit in the EPM, NSF, and OFT. Results showed that, while mice spent similar time in open-arm during OFF epoch, light stimulation of the sEHCeA-BNST circuit significantly increased the duration of open-arm exploration in ChR2 mice than that in eYFP mice (Fig. 11C,D; virus factor, F(1,42) = 4.21, p = 0.0465, Bonferroni's multiple comparisons test, ON, p = 0.0222; two-way ANOVA). These observations suggest that activation of sEHCeA-BNST pathway elicits an anxiolytic effect. Note that, we failed to detect an effect of optogenetic stimulation in the ChR2 mice in the OFT and NSF [virus factor, F(1,48) = 0.3762, p = 0.5425, Bonferroni's multiple comparisons test, ON, p = 0.9999 (Fig. 11E); virus factor, F(1,48) = 1.21, p = 0.2767, Bonferroni's multiple comparisons test, ON, p = 0.9999 (Fig. 11F); interaction, F(2,45) = 0.4544, p = 0.6377, Bonferroni's multiple comparisons test, ON, p = 0.946 (Fig. 11G); two-way ANOVA].
To examine whether inhibition of sEHCeA-BNST pathway could induce anxiety-like behaviors, we expressed NpHR (AAV2/9-Ef1α-DIO-eNpHR3.0-mCherry-WPRE-pA) or mCherry (AAV2/9-Ef1α-DIO-mCherry-WPRE-pA) virus in the CeA of Ephx2-iCreERT2 mice along with a bilaterally implanted optical fiber over the BNST to inhibit the sEHCeA-BNST pathway (Fig. 11H,I). The yellow light (580 nm) was constantly delivered during ON epoch. We found that the latency to feed was significantly increased without changes on food consumption on optogenetic inhibition of sEHCeA-BNST circuit in NpHR mice [Fig. 11J–L; virus factor, F(1,40) = 11.07, p = 0.0019, Bonferroni's multiple comparisons test, ON, p = 0.0115 (Fig. 11K); virus factor, F(1,40) = 0.1215, p = 0.7293, Bonferroni's multiple comparisons test, ON, p = 0.7792 (Fig. 11L); two-way ANOVA]. In contrast, there was little effect of optogenetic inhibition of sEHCeA-BNST circuit in the EPM and OFT [virus factor, F(1,45) = 0.7892, p = 0.3791, Bonferroni's multiple comparisons test, ON, p = 0.6679 (Fig. 11M); virus factor, F(1,57) = 0.1363, p = 0.7134, Bonferroni's multiple comparisons test, ON, p = 0.9999 (Fig. 11N); two-way ANOVA]. Nevertheless, these observations suggest that the sEHCeA-BNST circuit is critical for regulation of anxiety-related behaviors.
Discussion
Our study defines the neural mechanisms by which molecularly defined CeA neurons modulate anxiety-related behaviors. First, inhibition or specific deletion of sEH in CeA neurons promotes anxiety-like behaviors, which may be because of downregulated excitability of sEHCeA neurons. Second, the activity of sEHCeA neurons is corelated with anxiety-related behaviors, and chemogenetic manipulations of the activity of sEHCeA neurons bidirectionally regulate anxiety-related behaviors. Last, sEHCeA neurons mainly project to BNST, and optogenetic manipulations of the sEHCeA-BNST circuit are sufficient to affect anxiety-related behaviors. Together, these data support a model in which the activity of sEHCeA neurons leads to anxiety-related behaviors via both molecular and long-range circuit mechanisms.sEH has been reported to mostly localize in astrocytes throughout the brain. Using Ephx2-tdTomato mice and sEH antibody, we demonstrate that sEH is also expressed in neurons in the CeA (Fig. 1), in line with a previous study (Marowsky et al., 2009). By using multiple approaches, including pharmacology, AAV-shRNA and condition knock-out, we find that sEH in neurons in the CeA is required for anxiety-like behaviors (Figs. 2–4). One recent study showed that sEH−/− mice displayed anxiety-like behaviors without clear molecular and neural circuit mechanisms (Lee et al., 2019). Together, these findings demonstrated a novel role for sEH in the regulation of anxiety and might extend the function of the ARA-EET pathway in the brain.
What are the molecular mechanisms underlying the modulation of anxiety by sEH? sEH is a key enzyme of the ARA-EET pathway that catalyzes the degradation of EETs. In the electrophysiology experiments, we find that the sEH inhibitor TPPU and AAV-Ephx2-shRNA decreased the excitability of sEHCeA neurons (Figs. 5, 6). Notably, 11,12-EET, which could be degraded by sEH, also exhibited suppressive effect on the excitability of sEHCeA neurons (Fig. 7), in consistent with a previous report (Mule et al., 2017). It is plausible to speculate that sEH may regulate the excitability of sEHCeA neurons via 11,12-EET. However, the detailed mechanisms need to be further studied.
The identification of neuronal subpopulations in the CeA that positively regulate anxiety has long remained elusive. In our study, chemogenetic and optogenetic manipulations of the sEHCeA neurons and sEHCeA–BNST circuit bidirectionally regulate anxiety-related behaviors. It is noteworthy that while chemogenetic manipulations exhibited regulative effects in both the EPM and NSF, whereas optogenetic manipulations only showed effect either in the EPM (activation of sEHCeA–BNST circuit) or in the NSF (inhibition of sEHCeA–BNST circuit). These different behavioral outcomes between chemogenetic manipulation and optogenetic manipulation studies might be because of the different methods used and/or the different subtypes of neurons manipulated. Note that, except BNST, sEHCeA neurons also send projections to other regions such as SNL (Extended Data Fig. 10-1). The behavioral outcomes of the chemogenetic manipulations of sEHCeA neurons, which theoretically activate all the projecting pathways, may be incompletely consistent with those of optogenetic manipulations of sEHCeA–BNST pathway specifically.
The CeA has been reported to comprises distinct neuronal populations of inhibitory GABAergic neurons that express CRF, SOM, or PKCδ (Fadok et al., 2017), which partly co-stained with sEH (17.6 ± 5.5%, 31.6 ± 5.8%, and 25.3 ± 5.1%, respectively). We find that sEH partly colocalizes with CRF, SOM, and PKCδ at rate of 17.6 ± 5.5%, 31.6 ± 5.8%, and 25.3 ± 5.1%, respectively. Previous studies have shown that CRF-expressing neurons in the CeA that project to the ventral tegmental area (VTA) or BNST regulate anxiety (Dedic et al., 2018). We describe a new population of sEH-expressing cells in the CeA that innervate the BNST, which is required for regulation of anxiety-related behaviors (Figs. 8–11). Nevertheless, the functions of sEHCeA neurons that project to other brain regions, such as SNL, and the mechanisms underlying the regulation of anxiety-like behaviors warranted further investigation.
Footnotes
This work was supported by the National Natural Science Foundation of China (31771187, 81801293), the Guangzhou Science and Technology Project (201904020039, 202007030013), the National Program for Support of Top-notch Young Professionals, The Key Area Research and Development Program of Guangdong Province (2018B030334001, 2018B030340001), the Natural Science Foundation of Guangdong (2020B1515020006) and the Program for Changjiang Scholars and Innovative Research Team in University (IRT_16R37). We thank Xin-Hong Zhu (Southern Medical University) for providing Ephx2loxp/loxp and Ephx2-iCreERT2 mice. We also thank Wen-Chao Xiong, Ying-Ying Fang, Shu-Ji Li, Ting Guo, and Rong-Qing Chen (Southern Medical University) for their technical support.
The authors declare no competing financial interests.
- Correspondence should be addressed to Xiong Cao at caoxiong{at}smu.edu.cn or Xiang-Dong Sun at xisun{at}gzhmu.edu.cn