Abstract
Neuropeptide Y (NPY) increases resilience and buffers behavioral stress responses in male rats in part through decreasing the excitability of principal output neurons in the basolateral amygdala (BLA). Intra-BLA administration of NPY acutely increases social interaction (SI) through activation of either Y1 or Y5 receptors, whereas repeated NPY (rpNPY) injections (once daily for 5 d) produce persistent increases in SI through Y5 receptor-mediated neuroplasticity in the BLA. In this series of studies, we characterized the neural circuits from the BLA that underlie these behavioral responses to NPY. Using neuronal tract tracing, NPY Y1 and Y5 receptor immunoreactivity was identified on subpopulations of BLA neurons projecting to the bed nucleus of the stria terminalis (BNST) and the central nucleus of the amygdala (CeA). Inhibition of BLA→BNST, but not BLA→CeA, neurons using projection-restricted, cre-driven designer receptors exclusively activated by designer drug-Gi expression increased SI and prevented stress-induced decreases in SI produced by a 30 min restraint stress. This behavioral profile was similar to that seen after both acute and rpNPY injections into the BLA. Intracellular recordings of BLA→BNST neurons demonstrated NPY-mediated inhibition via suppression of H currents, as seen previously. Repeated intra-BLA injections of NPY, which are associated with the induction of BLA neuroplasticity, decreased the activity of BLA→BNST neurons and decreased their dendritic complexity. These results demonstrate that NPY modulates the activity of BNST-projecting BLA neurons, suggesting that this pathway contributes to the stress-buffering actions of NPY and provides a novel substrate for the proresilient effects of NPY.
Significance Statement
Neuropeptide Y (NPY) is associated with resilience in humans and mitigates stress-related behaviors in animal models. Through the actions of different NPY receptor subtypes, NPY induces either acute (Y1R) or long-term (Y5R) changes in the activity of basolateral amygdala (BLA) principal neurons that result in the development of a persistent stress resistance. Here, we determined that BLA neurons projecting to the bed nucleus of the stria terminalis (BNST) are inhibited by NPY and demonstrate decreased excitability and dendritic hypotrophy to repeated application of NPY. Inhibiting BLA→BNST neurons elevated social interaction (SI) and prevented stress-induced decreases in SI similar to intra-BLA NPY injections. These findings are significant as they identify a neuronal circuitry associated with NPY-mediated stress resistance.
Introduction
The development of resilience is important for mental well-being, and the neural mechanisms underlying the generation of resilience are the subject of intense investigation (Charney et al., 2012; Pfau and Russo, 2015; Kautz et al., 2017; Cathomas et al., 2019). Clinical and preclinical studies both demonstrate that neuropeptide Y (NPY) expression associates with enhanced resilience or performance under stressful conditions (Morgan et al., 2000; Thorsell et al., 2000). In clinical studies, NPY levels in the cerebrospinal fluid (Heilig et al., 2004; Sah et al., 2009, 2014) and in blood (Morgan et al., 2000; Yehuda et al., 2006) correlate with improvements in mood and symptoms associated with post-traumatic stress disorder. Elucidating the role of NPY as an endogenous anxiolytic neuropeptide provides a novel way to target and study neuronal circuitry associated with the generation of resilience.
NPY treatment mitigates symptoms associated with stress-related disorders such as anxiety and depression both in humans and animals (Yehuda et al., 2006; Sayed et al., 2018; Bertocchi et al., 2020; Nahvi et al., 2021). Deletions of either NPY or NPY receptors (Y1 or Y5) produce exaggerated stress responses and increased susceptibility to anxiety-like behavior in mice (Thorsell and Heilig, 2002), while overexpression of NPY in the hippocampus or amygdala reduces anxiety-like behaviors (Thorsell et al., 2000). Microinjection of NPY into discrete brain regions improves performance in several behavioral tests. Most notably, the basolateral amygdala (BLA) is key for the profound behavioral effects of NPY in the conflict test (Heilig et al., 1992), fear learning and extinction (Gutman et al., 2008), and social interaction (SI; Sajdyk et al., 2008), a sensitive readout for the stress-buffering properties of NPY. Intra-BLA administration of NPY prevents decreases in SI induced by acute stress or local administration of corticotropin-releasing factor (CRF; Sajdyk et al., 2004, 2006, 2008).
Injections of NPY into the BLA acutely produce transient increases in SI, while repeated NPY (rpNPY) injections (once daily for 5 d) produce persistent increases in SI via induction of neural plasticity (Sajdyk et al., 2008; Michaelson et al., 2020). The acute behavioral effects of NPY are mediated via either Y1 or Y5 receptor activation, hyperpolarizing BLA pyramidal neurons through inhibition of the depolarizing H current (Ih), carried by the hyperpolarization-activated cyclic nucleotide-gated channel subunit 1 (HCN1; Giesbrecht et al., 2010). Moreover, rpNPY injection produces persistent increases in SI and stress resistance that require activation of the Y5 receptor (Michaelson et al., 2020). In vitro intracellular recordings demonstrate that BLA principal neurons (PNs) from rats treated with intra-BLA rpNPY injections are hyperpolarized up to 4 weeks (W4) after treatment, produced in part by increases in spontaneous IPSC frequencies within 2 weeks of treatment and decreases in resting Ih (Silveira Villarroel et al., 2018). These results identify novel cellular mechanisms associated with the ability of NPY to modulate stress-sensitive behaviors and promote stress resistance.
Here, we identify components of BLA neural circuitry that are targets for endogenous NPY release and the generation of SI and stress-resistant behavior. BLA projections both to the bed nucleus of the stria terminalis (BNST) and central amygdala (CeA) express NPY receptors. The BNST is clearly important for the expression of anxiety-related behaviors (Avery et al., 2016), including SI (Vantrease et al., 2022), and likely mediates the ability of NPY to buffer stress-induced decreases in SI (Sajdyk et al., 2008). While the CeA is involved in fear responses, little is known about its contribution to SI (Rojek-Sito et al., 2023). We examined the contribution of these projections as they relate to the expression of NPY-induced increases in SI and stress resistance using a combination of projection-restricted DREADDs (designer receptors exclusively activated by designer drugs) and electrophysiological methods. Since intra-BLA injections of NPY do not alter SI in females (unpublished data) and chemogenetic inhibition of BLA→BNST neurons does not affect the expression of SI in female rats (Vantrease et al., 2022), the current studies focus on the male rat.
Materials and Methods
Animals
Adult male Sprague Dawley rats (>70 d; 250–300 g) were used in all experiments and were housed in an Association for Assessment and Accreditation of Laboratory Animal Care-accredited animal facility. The care and use of animals was done in accordance with protocols approved by the Rosalind Franklin University of Medicine and Science (RFUMS) and University of Alberta Institutional Animal Care and Use Committees. Upon arrival, animals were housed in groups of two rats per cage and allowed to acclimate to the facility for at least 7 d prior to experimentation. Rats were supplied with food and water ad libitum and maintained on a 12:12 light/dark cycle. The ambient room temperature was maintained at 23°C and light at ∼200–250 lux. All surgical procedures were conducted using sterile conditions.
Identification of NPY receptor subtypes on BLA projections to the BNST and CeA
Iontophoretic delivery of FluoroGold (FG)
FG (Fluorochrome) was stereotaxically injected in the BNST and CeA in animals under isoflurane anesthesia (2–4% isoflurane; Sigma-Aldrich). Once anesthetized, animals were placed in the stereotaxic apparatus with a digital readout (David Kopf). The iontophoretic delivery of FG (4% in 0.1 M cacodylate buffer, pH 7.0), was performed as previously described (Leitermann et al., 2016). In brief, borosilicate glass pipettes (10 μm tip diameter; Harvard Apparatus) were filled with FG solution and connected to an iontophoresis current source (Stoelting) via a bare silver wire. A steady-state retaining current (+7 µA) was maintained while lowering the pipette to the target coordinates. Unilateral injections of FG were delivered into the BNST (AP, −0.12 mm; ML, ±1.7 mm; DV, −6.80 mm from bregma) or in the CeA (AP, −2.5 mm; ML, −4.3 mm; DV, −8.0 mm from bregma; Paxinos and Watson, 2013). FG was delivered using current pulses (+7 μA; 4 s on/4 s off; 30 min), and the pipette was left in place with retaining current (+7 μA) for an additional 10 min prior to withdrawal.
Delivery of retrobeads
Animals were anesthetized and placed in the stereotaxic apparatus as indicated above. Rhodamine retrobeads (46 nl) were pressure injected into the BNST (AP, −0.12 mm; ML, ±1.7 mm; DV, −6.80 mm from bregma) from a cannula, either alone (to determine the presence of NPY Y5R on these neurons) or at the same time as cannula implantation in the BLA to identify BLA→BNST neurons in electrophysiology experiments after repeated intra-BLA NPY delivery.
After surgery, all animals received flunixin (2.0 mg/kg, i.p.; daily for 48 h; Patterson Veterinary Supply) as a postoperation analgesic. At least 7 d was allotted to allow for retrograde transport of both tracers into the BLA after which animals were perfused and brains were collected and processed for immunohistochemistry.
Animal perfusion and tissue preparation
Rats were deeply anesthetized (pentobarbital, 50 mg/kg i.p. or 4% isoflurane; Sigma-Aldrich) and transcardially perfused with warm (37°C) vascular buffer (0.1% procaine and heparin in 0.1 M PBS) followed by cold 4% paraformaldehyde (PFA; prepared in 0.1 M PBS and 0.05% glutaraldehyde). Brains were removed and postfixed in 4% PFA overnight at 4°C. Brains were sectioned on a vibratome (40 μm; Leica Biosystems), and slices from BNST through BLA were stored in cryoprotectant solution (23.8% glycerol, 28.6% ethylene glycol in 0.1 M PBS) at −20°C until processed for immunohistochemistry.
Immunohistochemistry
HCN1-FG-NPY colocalization
PFA-fixed brain sections containing the BLA were collected and processed for HCN1-FG-NPY immunoreactivities as previously described (Giesbrecht et al., 2010; Leitermann et al., 2016; Silveira Villarroel et al., 2018). Immunolabeling for HCN1 was performed using biotinylated tyramide (BT) for signal amplification (Rostkowski et al., 2009; Giesbrecht et al., 2010). Briefly, sections underwent a series of PBS washes and were incubated with 3% H2O2 followed by a blocking solution [5% normal donkey serum in ICC (immunocytochemistry) buffer; PBS–gelatin]. Primary antibody for HCN1 was added (mouse anti-HCN1; 0.5 μg/ml; UC Davis/NIH NeuroMab Facility; RRID, AB_2877279), and the tissue was incubated for 72 h. The HCN1 signal was visualized using a series of incubations in biotinylated donkey anti-mouse secondary antibody (Jackson Laboratory; RRID, AB_2340785), avidin–biotin complex (ABC Vectastain), 0.3% H2O2, and 3% BT in PBS and Alexa Fluor 594 streptavidin (1:250, Invitrogen; RRID, AB_2337250) with buffer rinses between each step. After a final rinse, sections were further incubated in a cocktail of rabbit anti-FG (1:1,000; Fluorochrome; RRID, AB_2314408) and mouse anti-NPY (1:2,000; Peninsula Laboratories; RRID, AB_2314974) for 72 h. Sections were rinsed in ICC buffer and incubated with fluorescent donkey anti-rabbit Alexa Fluor 488 (1:250; Invitrogen; RRID, AB_2535792) and donkey anti-mouse Alexa Fluor 647 (1:250; Invitrogen; RRID, AB_162542) secondary antibodies diluted in blocking buffer. Finally, sections were washed in Tris-buffered saline and mounted on SuperFrost charged glass slides; coverslips were applied using polyvinyl alcohol; 1,4 diazabicyclo [2.2.2]octane.
Y1R-FG-GAD (GABA decarboxylase) colocalization
Brain sections containing the BLA were collected and processed for Y1-FG-GAD (glutamic acid decarboxylase) immunoreactivities as described above with the following exceptions: Y1R immunolabeling [1:800 (Wolak et al., 2003; Rostkowski et al., 2009); Y1R; RRID, AB_231561] was performed with signal amplification using BT, and FG and GAD (mouse anti-GAD; 1:500; Merck Millipore; catalog #06-983; RRID, AB_310324) were subsequently labeled following standard dual-label immunohistochemistry procedures as indicated above.
Y5R-rhodamine retrobeads colocalization
Prior studies from our group have demonstrated the role of Y5 receptors in the long-term effects of repeated intra-BLA NPY treatment (Michaelson et al., 2020). Complementary to our FG experiments, adult male rats were unilaterally injected with retrobeads in the BNST and allowed 7 d for retrobeads to be transported into BLA projection neurons, following a previously published protocol (Vantrease et al., 2022). Y5R immunolabeling (1:500; Wolak et al., 2003) was performed with signal amplification using BT as indicated above.
The specificity of immunohistochemical signals was assessed using omission of one or more primary or secondary antibodies as well as characterization by Western analysis of knock-out and intact tissues (Wolak et al., 2003; Rostkowski et al., 2009; Silveira Villarroel et al., 2018). Immunoreactivity was observed using an Olympus FluoView confocal microscope (Microscopy and Imaging Facility; RFUMS), and colocalization was confirmed when immunoreactive signals were observed in >2 optical sections. For analysis of the regional distribution of BLA neurons which project to either the BNST or CeA, brain sections were matched to corresponding atlas sections (Paxinos and Watson, 2013), and FG-positive cell bodies in the BLA and nearby related amygdala regions were traced onto atlas templates. The placement and extent of the FG injections into the BNST or CeA were characterized, and the corresponding placement of FG-filled cells in the BLA was mapped using similar color notations (Fig. 1). Only brains that included injections within the specific brain regions (BNST; CeA) were included in the analysis. For presentation, images were scanned from the confocal, imported into Adobe Photoshop (v.10) where brightness and contrast were adjusted to appropriately represent the intensity of the signal on the representative section.
Relative distribution of BNST- and CeA-projecting neurons in the BLA. A, Representative placements of FG tracer in rats that received unilateral injections in the BNST. The relative placement and spread of the injection is color-coded in blue and green hues corresponding to each animal. B, Representative placements of FG tracer in rats that received unilateral injections in the CeA. The relative placement and spread of the injection is color-coded in yellow and red hues corresponding to each animal. C–H, Schematic representation of the distribution of BNST- and CeA-projecting neurons in the BLA from the rats indicated in A and B. FG-backfilled neurons were mapped throughout the anterior–posterior axis of the BLA at representative atlas-matched bregma levels (distance posterior to the bregma indicated in bottom-left corners; Paxinos and Watson, 2013).
Chemogenetic inhibition and activation of BLA neurons projecting to BNST or CeA
Stereotaxic microinjections of CAV2-cre and AAV5-DIO-DREADD-Gi or Gq injections
To test the activity of specific BLA projections on SI and stress resistance, combinatorial chemogenetics was used to either inhibit (hM4Gi) or activate (hM3Gq) BLA projections to the BNST or CeA (Sternson and Roth, 2014). Animals underwent sterile stereotaxic surgery using inhalational anesthesia (4% isoflurane) in a Biosafety Level 2 facility. An aliquot (0.5 μl) of each virus was microinjected into the respective brain region over 5 min, and the injector was maintained in situ for another 5 min prior to withdrawal. To target BLA neurons projecting to each specific site, the retrogradely transported canine-associated virus 2 (CAV2-CMV-Cre; 1 × 1012 pp/ml; Plateforme de Vectorologie de Montpellier, Institut de Génétique Moléculaire de Montpellier) was injected bilaterally into either the BNST (15° angle; AP, −0.1 mm; ML, ±2.4 mm; DV, −6.3 mm) or the CeA (AP, −2.3 mm; ML, ±4.3 mm; DV, −8.0 mm) using a Neurosyringe coupled to a digital pump. Viruses containing a construct for either inhibitory [AAV5 hSyn-DIO-hM4D(Gi)-mCherry plasmid; 1 × 1012 GC/ml; Addgene 44362] or excitatory [AAV5 hSyn-DIO-hM3D(Gq)-mCherry plasmid; 1 × 1012 GC/ml; Addgene 44361] DREADDs were injected into the BLA. Starting 7 d after surgery, animals were tested for SI. Upon completion of the studies, the placements of the injections were verified histologically. Only those animals with viral expression in the targeted regions were included in the analyses.
Validation of DREADD expression in the BLA
Brain sections containing the BNST and the CeA/BLA were collected and processed for mCherry [AAV-DREADD and cre (CAV2) immunoreactivity; Vantrease et al., 2022]. Using a modification of the protocol outlined above, sections were incubated in a cocktail of primary antibodies consisting of chicken-mCherry (1:4,000; catalog #CPCA-mcherry; EnCor Biotechnology; RRID, AB_2572308) and rabbit-cre (1:2,000; catalog #908001; BioLegend; RRID, AB_2565079). Sections were then washed in ICC buffer and incubated at room temperature with fluorescent-tagged secondary antibodies targeting the rabbit (1:250; goat anti-rabbit-488; catalog #A11008; Invitrogen) and chicken (1:250; donkey anti chicken-594; catalog #703-585-155; Jackson ImmunoResearch Laboratories) primary antibodies. Following incubation, the sections were washed and subsequently mounted onto gelatin-coated slides for analysis using epifluorescence microscopy (Nikon; 20× magnification). The number of double-labeled cells within the boundaries of the BLA was determined in representative atlas-matched sections of the BLA (−2.3, −2.56, −2.8, −3.3, −3.6, and −3.8 mm from bregma; Leitermann et al., 2012; Rostkowski et al., 2013; Paxinos and Watson, 2013). The total number of labeled cells was reported for each animal.
Behavioral procedures
SI test
All animals were habituated to the open-field arena (SI arena; 1 m long × 1 m wide × 0.3 m high with an open top; 120 lux illumination) at least 24 h prior to the first SI testing. Each animal was initially placed in the center of the arena and allowed 10 min to explore before being returned to its home cage. The first SI (baseline) testing for every animal was performed prior to surgery which followed the established protocol we and others have implemented to study the effects of intra-BLA NPY on SI (Sajdyk et al., 2008; Silveira Villarroel et al., 2018; Michaelson et al., 2020). To test for SI, two partner rats were placed in opposite corners of the arena and allowed to freely explore and interact over 10 min; behavior was recorded on HD video for off-line analysis. The behavioral arena was wiped with 70% ethanol between each testing session. SI experiments were performed between 10:00 and 15:00 h. Partner rats were of the same sex and similar weight, housed under identical conditions and had no previous contact with their paired partner rat. The videos were scored for SI behavior by individuals blinded to the treatment group. Seven days postsurgery, all animals received a vehicle injection (sterile saline; 1 ml/kg, i.p.), and 45 min later they were placed in the SI arena. At W3 and W4 postsurgery, animals received an injection of vehicle or CNO (1 mg/kg, i.p.; clozapine-N-oxide; Tocris Bioscience) 45 min prior to behavioral testing. SI time is reported as % baseline for each individual animal. Behavior was video-recorded and uploaded into CowLog (Hänninen and Pastell, 2009) for analysis. Distinct components of SI behaviors were assessed [following, sitting together, sniffing (anogenital), and grooming (head/body grooming and sniffing); Ferrara et al., 2022], and the time of the beginning and end of each behavior was noted and tabulated. The data obtained with CowLog were analyzed using a custom script in MATLAB (MathWorks) to calculate the total time in seconds for each behavior.
Stress resistance
To test for stress resistance, animals were subjected to a restraint stress prior to SI. Four days after the W4 SI test session, animals received injections of either vehicle or CNO and were subsequently tested for stress resistance (Sajdyk et al., 1999, 2008). Rats were transferred to a novel room, where they were placed into plastic hemicylinder restrainers (8 inch long × 3.25 inch wide) for 30 min. After removal from the restrainers, rats were placed into the SI arena, and their SI behavior was assessed. To test for nonspecific effects of CNO on SI, SI was measured in a separate group of naive male rats that did not receive virus injections.
NPY effects on BLA neuronal activity
Repeated intra-BLA injections of NPY
Induction of stress resistance after intra-BLA microinjections of NPY in rats has been described previously (Sajdyk et al., 2008; Silveira Villarroel et al., 2018; Michaelson et al., 2020). Briefly, male Sprague Dawley rats (6–8 weeks old) were obtained from the University of Alberta colony and allowed to acclimate to the laboratory for 3–4 d or from Harlan and acclimated to the animal facility for 7 d. Animals were anesthetized with ketamine (90 mg/kg, i.p.) and xylazine (10 mg/kg, i.p.) and placed in a stereotaxic apparatus with a digital readout (David Kopf). Body temperature was maintained with a homeothermic monitor (Harvard Apparatus). Bilateral guide cannulae (26 gauge; Plastics One) were implanted stereotaxically to terminate 2.0 mm dorsal to the BLA (AP, −2.8 mm; ML, ±5.0 mm; DV, −6.3 mm). Cannulae were secured to the skull with stainless steel screws (0–80× 1/8; Plastics One) and dental cement. After completion of surgery, all animals received injections of sterile saline (1 ml, i.p., per animal) and meloxicam (1 mg/kg, s.c.) and were placed on a warming pad until fully recovered from the anesthetic.
After 5–7 d recovery, injections either of NPY (10 pmol/100 nl) or vehicle (PBS) were delivered bilaterally into the BLA on 5 consecutive days. Two 33 gauge injectors (Plastics One) extending 2 mm beyond the guide cannula were each connected via polyethylene tubing (PE 50; Plastics One) to a 1 μl syringe (Hamilton) placed in a syringe pump (model PHD 4400, Harvard Apparatus); the solutions were delivered (100 nl per site) over 30 s. The injectors remained in place for an additional 2 min, after which they were withdrawn and their patency was verified. Animals were conscious and gently hand-restrained during this procedure.
SI test after NPY treatment
To verify the efficacy of NPY injections, SI was determined (W2 and W4 postinjections) as previously described (Silveira Villarroel et al., 2018; Michaelson et al., 2020). After recovery from surgery for 72 h, animals were acclimatized as above for baseline SI with mock injections administered 2 min prior to placement in the open-field arena. Twenty-four hours later, animals received a mock injection and afterward were placed in opposite corners of the open-field area with a partner rat as above, and baseline SI behavior was determined. One week after surgery, each animal received a bilateral injection of either PBS vehicle or NPY and 30 min later was allowed to interact with a novel partner rat for 10 min. All SI experiments were recorded on HD video for off-line analysis. SI was performed as above following injection (saline or NPY) on Days 1 and 5 and then on Days 14 (W2), 21 (in some cases), and 28 (W4) after the first injection. The total time the experimental animal spent interacting (i.e., sniffing, grooming, and following) with the partner rat was assessed over the 10 min period. Behavior was independently scored by two different raters blinded to the treatment.
Brain slice preparation
After 14 and 28 d from the first injection, vehicle- or NPY-treated rats (some with retrobead injections) were killed by decapitation without prior anesthesia. Brains were carefully but rapidly removed and submerged in cold (<4°C) artificial cerebrospinal fluid (ACSF) that contained the following (in mM): 118 NaCl, 3 KCl, 1.3 MgSO4, 1.4 NaH2PO4, 5.0 MgCl2·6H2O, 10 glucose, 26 NaHCO3, and 1.5 CaCl2, 300 mOsm, bubbled with carbogen (95% O2, 5% CO2). Kynurenic acid (1 mM) was added to the slicing solution to prevent damage from ionotropic glutamate receptor activation (Giesbrecht et al., 2010). Coronal brain slices (300 µm) containing the BLA were prepared using a vibrating slicer (Slicer HR2, Sigmann Elektronik). Slices were placed into a room temperature (22°C), carbogenated ACSF solution (bath solution) containing the following (in mM): 124 NaCl, 3 KCl, 1.3 MgSO4, 1.4 NaH2PO4, 10 glucose, 26 NaHCO3, and 2.5 CaCl2, 300 mOsm. Bath solution was used for all remaining experiments. Slices were acclimatized to room temperature for a minimum of 30 min prior to being placed into the recording chamber. Slices were held submerged by a platinum and polyester fiber “harp” in a fixed-stage recording chamber (Giesbrecht et al., 2010) and viewed with a movable upright microscope (Axioskop FS2, Carl Zeiss). The slices were continuously perfused with warm (34°C ± 0.5°C), carbogenated ACSF between 2 and 2.5 ml/min for ∼20 min prior to any recordings.
Electrophysiology
Pipettes were pulled from thin-walled borosilicate glass (TW150F, World Precision Instruments) with a two-stage puller (PP-83, Narishige Scientific Instrument Lab). Tip resistance was 5 MΩ when pipettes were filled with an internal solution containing the following (in mM): 5 HEPES, 2 KCl, 136 K+-gluconate, 5 EGTA, 5 Mg·ATP, and 0.35 GTP. The pH was adjusted to 7.27 with KOH, and osmolarity adjusted to between 285 and 290 mOsm using a micro-osmometer (models 3-MO or 3320, Advanced Instruments). In most cases, the pipette solution also contained neurobiotin (0.2%) to enable post hoc analysis of cell morphology. All recordings were made using either an Axoclamp 2A or a Multiclamp 700B amplifier, and data were acquired via a Digidata 1322 or Digidata 1440 interface, and experiments were controlled and analyzed with pClamp version 9 or 10 (all Molecular Devices). All membrane potentials reported were corrected for the calculated 15 mV liquid junction potential of the solutions (Chee et al., 2010).
The anatomical limits of the BLA were identified based on the atlas of Paxinos and Watson (2013). Pyramidal neurons within the BLA were viewed using infrared-differential interference contrast optics and selected based on morphology and the presence of a large primary dendrite. BNST-projecting BLA neurons were preidentified by the presence of retrobeads. Gigaohm seals were initially established on the soma in current clamp. Once a seal was formed, the patch was ruptured with the cell held near the resting potential in voltage-clamp. Once whole-cell recording had been established, neurons were routinely held in voltage clamp near rest at −75 mV (all reported potentials are corrected for the −15 mV liquid junction potential) except when changes in the resting membrane potential and rheobase were studied in current-clamp mode. Only cells exhibiting both stable holding current and access resistance for at least 10 min prior to experimental manipulations were studied further.
Pharmacological studies
Recordings of BLA PN fundamental properties, including rheobase, resting membrane potential, action potential parameters, and spontaneous postsynaptic currents, were made immediately prior to drug application (control), during drug application (lasting ∼3–4 min), and at regular intervals for 30 min after initiating washout (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018; Michaelson et al., 2020). In some cells, the washout was prolonged and recordings were taken until 45 min after washout began. A similar sequence of recordings, including an assessment of access resistance, was acquired for each cell and condition. Changes in the resting membrane potential were assessed by comparison of continuous (30 s) current-clamp recordings made under each experimental condition in the absence of any imposed current.
Rheobase measurements
To determine the rheobase, neurons were held in the current clamp near their resting membrane potential (for each condition), and families of up to eight depolarizing current ramps were swept from 0 pA initially to 100 pA over 800 ms; the peak current was incremented by 100 pA with each successive sweep. The minimum current necessary to elicit the first action potential during a ramp was designated as the cell's rheobase current under the condition studied.
H current (Ih)
To study Ih, a family of hyperpolarizing voltage-clamp steps (−10 to −70 mV; Giesbrecht et al., 2010) was applied from a holding potential of −55 mV. The magnitude of Ih at a given potential step was determined as the difference between the initial positive current peak after the capacitative transient and the final, steady-state current for each step. The amplitude of Ih was plotted against step potential.
Instantaneous inward rectifier (Iir)
This current, caused by the tonic activation of GABAB receptors on BLA PNs (Mackay et al., 2019), was observed as a time-independent inward current seen immediately upon application of a hyperpolarizing voltage step during the H-current protocol. The amplitude of Iir was plotted against step potential. For clarity, mean net Iir is plotted as a reduction in net current with NPY treatment.
Imaging and neuronal reconstruction
Neurons used for morphological analysis were filled with neurobiotin (0.2%; Vector Laboratories) via the pipette during the recording, fixed overnight in 4% buffered formalin, and then retained for post hoc morphological studies at a later time. Free-floating slices were washed in 3 × 10 min each in PBS before blocking and permeabilization for 2 h with 0.3% Triton X-100 in PBS and 4% normal goat serum (Sigma-Aldrich). Slices were incubated with streptavidin conjugated with Alexa Fluor 555 or −546 (1:1,000; Invitrogen) in 0.3% Triton X-100 in PBS with 4% normal goat serum for 3 h at room temperature. Slices were washed 4 × 10 min each in PBS, mounted on SuperFrost Plus slides, and coverslips were applied with ProLong Gold mounting media (Thermo Fisher Scientific). Slides were allowed to air-dry in the dark at room temperature before being imaged via confocal microscopy. Z-stacks were obtained with a laser scanning confocal microscope (TCS SP5; Leica Microsystems; University of Alberta Faculty of Medicine and Dentistry Cell Imaging Centre) with an excitation wavelength of 543 nm using a 20× objective, with 1,024 × 1,024 resolution and 100 Hz for a series of 0.8 µm steps. The position of each neuron within the slice was determined with multiple single-plane images at 10× magnification (512 × 512 resolution and 100 Hz) and subsequently stitched together (Leica LAS AF software). Only healthy-appearing neurons with stable electrophysiological recordings and a cell body clearly within the boundaries of the BLA were included in the analyses. The neurite tracer function in FIJI (NIH) was used for neuronal reconstruction of soma and dendrites, as previously reported (Michaelson et al., 2020). Sholl analysis of the entire dendritic reconstruction was performed using10 μm steps of concentric radii with the FIJI software. Total dendritic length and dendritic branching were calculated using the FIJI and Microsoft Excel software.
Chemicals and drugs
For components of slicing, bath, and pipette solutions, kynurenic acid was obtained from Abcam, Mg2+ ATP, NaH2PO4, CaCl2•2H2O, and K+-gluconate were obtained from Sigma-Aldrich, and Na-GTP was obtained from Roche Diagnostics. All remaining chemicals were obtained from Thermo Fisher Scientific. NPY (rat/human) was obtained from PolyPeptide Laboratories. All peptides were dissolved in HPLC-grade water (Thermo Fisher Scientific or Sigma-Aldrich) and then stored at −20°C as individual single-dose aliquots of at least 100-fold the applied concentration until immediately prior to use. Neurobiotin was obtained from Vector Laboratories.
Statistical analysis
All data are represented as mean ± standard error of the mean (SEM), and the number of animals (n) is indicated in all graphs. GraphPad Prism software (versions 7–10) was used for calculation of statistical values and graphic illustration of all data. All behavioral analyses were performed using two-way ANOVA (treatment and time point variables) with Tukey's or Sidak's multiple-comparison test. Normality of distributions was assessed using the Shapiro–Wilk test, with an alpha of 0.05 as a cutoff. Sample sizes were determined based on our previously published work; significance was set at p < 0.05. Post hoc power analyses on the behavioral findings for the DREADD experiments were performed using GPower (Faul et al., 2007); providing the effect size F, α = 0.05, and the given sample size and degrees of freedom (df), we have a calculated power of 0.92–1.00 for the data presented.
Results
Relative distribution of BNST- and CeA-projecting neurons within the BLA
The distribution of BNST- and CeA-projecting neurons within the BLA was mapped throughout the anterior–posterior axis of the BLA. The injection sites and distribution of cells of origin within the BLA and surrounding local areas were color-coded for each animal (Fig. 1). The location of the injection sites overlapped within the targeted regions of the BNST and CeA in eight representative animals (Fig. 1A,B). No FG-labeled cell bodies were noted in the BLA from those injections that were outside the targeted areas (BNST or CeA; data not shown), indicating accuracy of injection placement and specificity of BLA projections. While FG neurons projecting to either site were found throughout the rostral–caudal extent of the BLA, the heaviest projections to either the BNST (blue/greens) or CeA (yellow/red) were found in the mid-BLA (−2.80 mm caudal to bregma; Fig. 1E–H). The location of the injection site shifted the exact location of the projection neurons in the BLA, but, in general, BNST- or CeA-projecting neurons were clustered together. BNST- and CeA-projecting populations appear to be segregated in the anterior aspect of the BLA (Fig. 1C,D). In these anterior sections, the cells are present in the medial aspect of the BLA, a region where we observe strong cellular responses to NPY (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018; Mackay et al., 2019). In posterior sections, there was an increased overlap of the two populations which were also expressed in the more lateral aspects of the BLA, forming a banded pattern across the lateral/basal amygdala transition (Fig. 1E–H). BNST-projecting neurons were also found in the CeA. Lastly, the most medial CeA injection yielded FG transport within the amygdalostriatal transition area (rat ID FG4; Fig. 1A). This region contains NPY neurons that project to the BLA and are activated during fear testing (McDonald and Zaric, 2015a; Leitermann et al., 2016).
BLA neurons projecting to the BNST or CeA express NPY receptors and HCN1 channels
As we have demonstrated, activation of Y1 or Y5 receptors in the BLA increases SI with Y5R being important for inducing long-term changes in behavior and electrophysiological properties of BLA PNs (Silveira Villarroel et al., 2018; Michaelson et al., 2020). Furthermore, the acute and long-term effects of NPY are associated with inhibition of the H current (Ih) and downregulation of HCN1 channels (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018). Therefore, to identify the neural circuitry underlying these responses, we determined whether BLA projections to either the BNST or CeA expressed markers associated with NPY receptor signaling and stress resistance in the BLA, namely, Y1 and Y5 receptors and HCN1 channels. As demonstrated previously, NPY, Y1R, and HCN1 immunoreactivities were distributed throughout the BLA (Kopp et al., 2002; Wolak et al., 2003; Rostkowski et al., 2009; Silveira Villarroel et al., 2018). Both Y1R and HCN1 immunoreactivity were present on identified BLA→BNST (Fig. 2A3⇓–5) and BLA→CeA projections (Fig. 2B3⇓–5). In addition, BLA neurons projecting to BNST (Fig. 2A, FG-backfilled neurons) and CeA (Fig. 2B) were found in close apposition with NPY fibers (Fig. 2A3⇓–5,B3⇓–5). This anatomical proximity is consistent with NPY release in the vicinity of projection neurons to directly influence the activity of projection targets. Complementary studies using retrobeads to identify BLA→BNST neurons identified that this projection also expressed Y5R immunoreactivity (Fig. 2C1⇓⇓–4). In sum, these data provide anatomical support that either, or both, of these projections could contribute to the acute and persistent effects of NPY on SI. BLA projection neurons are predominantly glutamatergic, and, as expected, GAD immunostaining was not present in BLA→BNST projection neurons (Fig. 2A6). However, a few, scattered GAD-immunopositive neurons were found projecting to the CeA (Fig. 2B6); while this was somewhat unexpected, small populations of “long-range nonpyramidal” BLA (GABAergic) neurons have previously been reported in this brain region (McDonald et al., 2012; McDonald and Zaric, 2015b).
Distribution of NPY Y1 and Y5 receptor and HCN1 immunoreactivity within BLA→BNST- and BLA→CeA-projecting neurons. A, Immunoreactive patterns of Y1R and HCN1 expression in BLA→BNST neurons. A1, Photomicrograph of FG placement in the BNST of a male rat and of A2, FG-backfilled neurons within the BLA of the same animal. A3–5, Photomicrographs showing coexpression of A3, FG (yellow) and NPY (cyan), and A4, FG and HCN1 (red) immunoreactivity within the BLA. A5, Merged image. Arrows indicate coexpression of HCN1 immunoreactivity in FG-containing neurons; NPY fibers are found in close apposition to these cells (arrow heads). A6–8, Images showing A6, FG (yellow) and GAD (magenta), and A7, FG and Y1R (green) and immunoreactivity within the BLA. A8, Merged image. FG immunoreactivity is found within neurons expressing Y1R immunoreactivity (arrows), and they are surrounded by GAD-immunoreactive puncta (arrowheads). B, Immunoreactive patterns of Y1R and HCN1 expression in BLA→CeA neurons. B1, Yellow deposition shows placement of a FG injection site within the CeA and of B2, FG-backfilled neurons within the BLA of the same animal. B3–5, Higher-magnification images showing the following: B3, FG (yellow) and NPY (cyan), and B4, FG and HCN1 (red) immunoreactivity within the BLA. B5, Merged image. HCN1 channel immunoreactivity is present in FG-immunoreactive neurons (arrows), and they are in close apposition to NPY fibers (arrow heads). B6–8, High-magnification images showing B6, FG (yellow) and GAD (magenta), and B7, FG and Y1R (green) and immunoreactivity within the BLA of the same animal. B8, Merged image. Neurons expressing Y1R (arrows) and which receive perisomatic GAD-immunoreactive puncta (arrow heads) but do not express GAD within their cell bodies. C, Immunoreactive patterns of Y5R expression in BLA→BNST neurons (red). C1, Retrobead-backfilled neurons in the BLA after injection of tracer in the BNST. Horizontal arrows indicate C2, Y5R immunoreactivity coexpressed in C3, BLA→ BNST (red) neurons; C4, Merged image. Some populations of BLA→BNST neurons do not express Y5R immunoreactivity (vertical arrows). Scale bars: 100 μm for A1, A2, B1, B2, C1; 10 μm for A3–8, B3–8, C2–4. LA, lateral amygdala; ot, optic tract.
Chemogenetic inhibition of BLA→BNST but not BLA→CeA pathway increases SI
Since application of NPY inhibits the activity of BLA neurons (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018) and both BLA→BNST and BLA→CeA projections express NPY receptors and HCN1 immunoreactivity, we sought to determine whether inhibition of either of these pathways would recapitulate the effects of NPY on SI and stress-induced changes in SI. Based on previous studies, the BNST was an optimal target since intra-BLA administration of NPY inhibited stress-induced c-Fos expression in this region (Sajdyk et al., 2008). Using a viral approach combining bilateral injections of the retrograde CAV2-cre virus into the BNST and of AAV5-DIO-DREADD-Gi into the BLA, the BLA→BNST circuit was targeted and inhibited upon administration of CNO (Fig. 3A; Vantrease et al., 2022). This combination of viral injections allowed for the expression of DREADD-Gi selectively restricted to BLA neurons projecting to the BNST (Fig. 3B). SI was assessed 45 min after systemic injection of CNO (1 mg/kg) and at various times after viral injections (Fig. 3C). Injection of CNO significantly increased SI at W3 and W4 after viral injections when compared with their baseline or with vehicle controls at the same time points (Fig. 3E). Injections of saline vehicle did not alter SI either at W1 (a time before appropriate transfection of the virus had occurred) or at W3 and W4 in either group. Subjecting rats to a 30 min restraint stress (4 d after the W4 time point; Fig. 3C) significantly decreased SI in animals receiving vehicle treatment, but these stress-induced decreases in SI were prevented in animals receiving CNO prior to restraint (Fig. 3F). Assessment of the different scored behaviors comprising the composite SI score indicated that in unstressed rats (“no restraint”), CNO increased the time following (Fig. 3G) and sitting together (Fig. 3H) but did not alter grooming/sniffing time (anogenital and head to body; Fig. 3I). Subjecting animals to the restraint stress significantly decreased the amount of time spent grooming/sniffing in the vehicle group, and this decrease was prevented by treatment with CNO (Fig. 3I); restraint stress did not significantly impact either following (Fig. 3G) or time sitting together (Fig. 3H).
Chemogenetic inhibition of the BLA to BNST projection increases SI and rescues stress-induced deficits in SI. A, A schematic representation of BNST and BLA injection sites for CAV2-cre in the BNST and AAV5-DIO-DREADD-Gi in the BLA. CAV2-cre is retrogradely transported to the BLA and enables expression of DREADDs only in BLA neurons expressing CAV2-cre, i.e., BLA→BNST neurons. B, the representative CAV2-cre injection site in the BNST (cre; B1) and expression of AAV5-DIO-DREADD-Gi in the BLA (mCherry; B2). B3–5, Demonstration of coexpression of cre (B3) and mCherry (B4) in the same cells (B5, merged image). C, Experimental timeline. Baseline SI was recorded before the surgeries. One week after surgeries, all animals received vehicle injections, and their SI behavior was assessed 45 min later to familiarize animals with the coupling of injections with the SI protocol. CNO (1 mg/kg, i.p.) or saline was administered 45 min prior to the SI testing. Four days after the W4 data point, animals received vehicle or CNO injection, and 45 min later they were subjected to 30 min of restraint stress. SI testing was performed immediately after the end of the stress session. D, Systemic CNO injection to control rats not expressing DREADD receptors did not affect SI performance in unstressed or stressed animals (veh n = 4; CNO n = 5; main effect of treatment F(1,7) = 0.575; p = 0.4730; main effect of stress F(1,6) = 29.03; p = 0.0017; two-way RM ANOVA; no restraint, p = 0.9926; restraint, p = 0.4658; Sidak's post hoc; veh, p = 0.0428; CNO, p = 0.0075; Sidak's post hoc). E, Administration of CNO increased SI at W3 (p = 0.0315; Sidak's post hoc) and at W4 (p = 0.0057; Sidak's post hoc) post-virus injections when compared with vehicle treatment (veh n = 5; CNO n = 6; main effect of treatment F(1,9) = 5.95; p = 0.0373; main effect of time F(2,18) = 8.235; p = 0.0029; time × group interaction F(2,18) = 7.574; p = 0.0041; two-way RM ANOVA). Vehicle injections did not alter SI performance (p > 0.9992; Sidak's post hoc). F, Subjecting animals to a 30 min restraint reduced SI time, and this decrease was prevented in CNO-treated animals (main effect of treatment F(1,9) = 14.43; p = 0.0042; two-way RM ANOVA; no restraint, p = 0.0135; restraint, p = 0.0021; Sidak's post hoc). G, The time rats spent following was increased after CNO injection (main effect of treatment F(1,9) = 14.29; p = 0.0043; two-way RM ANOVA) in both unstressed (p = 0.0072; uncorrected Fisher's LSD) and stressed (p = 0.0147; uncorrected Fisher's LSD) animals, but subjecting animals to restraint stress did not alter following activity (main effect of stress F(1,9) = 2.789; p = 0.1293; stress × treatment interaction, F(1,9) = 0.0648; p = 0.8048; two-way RM ANOVA). H, CNO treatment significantly increased time sitting together (main effect of treatment F(1,9) = 16.28; p = 0.0003; two-way RM ANOVA) in both unstressed (p = 0.0076; uncorrected Fisher's LSD) and stressed (p = 0.0045; uncorrected Fisher's LSD) animals (main effect of stress F(1,9) = 6.258; p = 0.0309; with no interaction, F(1,9) = 0.0364; p = 0.8530; two-way RM ANOVA). I, Administration of CNO did not alter total grooming/sniffing time in unstressed rats (p = 0.9748; uncorrected Fisher's LSD) but did prevent the stress-induced decrease in this behavior (p = 0.0132; uncorrected Fisher's LSD; main effect of treatment F(1,9) = 2.337; p = 0.1606; main effect of stress, F(1,9) = 4.965; p = 0.0529; stress × treatment interaction, F(1,9) = 9.144; p = 0.0144; two-way RM ANOVA). Data are reported as mean ± SEM; *p < 0.05; **p < 0.01 post hoc test.
Prior studies have challenged the exact mechanism of DREADD engagement by systemic CNO (Gomez et al., 2017) thereby raising concerns about off-target effects of CNO (MacLaren et al., 2016; Manvich et al., 2018). While the dose of CNO used is relatively low, we determined possible CNO effects on SI behavior in animals not expressing DREADDs (Fig. 3D). Systemic CNO delivery (1 mg/kg, i.p.) did not alter SI in these animals when compared with their respective baselines or vehicle control responses (Fig. 3D, “no restraint”). Additionally, CNO treatment prior to stress did not modulate stress-induced declines in SI (Fig. 3D, “restraint”). These findings increase our confidence that our observations during SI are indeed attributable to reduced activation of BLA→BNST projections and that potential off-target effects of CNO do not contribute to this finding.
The same injection and viral treatment as above were used to target and inhibit the BLA→CeA projection and measure resultant changes in SI both in unstressed and stressed animals (Fig. 4A,C). DREADD-Gi expression was specifically restricted to BLA neurons projecting to the CeA (Fig. 4B). Vehicle injections W1 postsurgery did not alter SI values when compared with their respective baseline values, indicating that surgery did not affect SI (Fig. 4D). At W3 and W4 postsurgery, injection of CNO did not alter SI times when compared with their baseline or vehicle-injected controls (Fig. 4D). On W4, 4 d after the previous CNO injection, all animals received either vehicle or CNO, 45 min prior to a 30 min session of restraint stress. Restraint stress significantly decreased SI in vehicle-treated animals, and this was not affected by CNO treatment nor was it different between the groups (Fig. 4E) suggesting that inhibition of BLA→CeA projections does not alter SI levels. This was also evident by the lack of any effect of CNO on the time spent following (Fig. 4F), sitting together (Fig. 4G), or grooming/sniffing (Fig. 4H) in unstressed rats. Subjecting animals to the restraint stress significantly decreased the amount of time spent grooming/sniffing in the vehicle group, and this was not prevented by CNO treatment (Fig. 4H). The number of DREADD-expressing BLA neurons projecting to either the BNST or CeA was not significantly different between the groups suggesting that differences in the number of transfected cells were not a contributing factor to the observed differences in behavior (BLA→BNST neurons, 554 ± 6 cells; BLA→CeA neurons, 578 ± 22 cells; Student’s t test, t = 1.067; df = 8; p = 0.3173).
Chemogenetic inhibition of the BLA to CeA projection does not affect SI times. A, A schematic representation of CeA and BLA injection sites for bilateral CAV2-cre in the CeA and bilateral AAV5-DIO-DREADD-Gi in the BLA. B, A representative CAV2-cre injection site in the CeA (B1, cre) and AAV5-DIO-DREADD-Gi in the BLA (B2, mCherry). B3–5, Demonstration of coexpression of cre (B3) and mCherry (B4) in the same cells (B5, merged image). C, Experimental timeline. Baseline SI performance of all animals was recorded before the surgeries and time course follows that as described in Figure 4C. D, Systemic CNO did not affect SI times at the times tested (veh n = 5; CNO n = 6; main effect of treatment, F(2,18) = 0.09047; p = 0.7704; main effect of time, F(2,18) = 0.07599; p = 0.9271; time × group interaction, F(2,18) = 0.3155; p = 0.7334; two-way RM ANOVA). Vehicle injections did not affect SI performance (W1; p = 0.9992). E, Restraint stress produced a significant decrease in SI that was not affected by CNO treatment (main effect of treatment, F(1,9) = 0.1542; p = 0.7037; main effect of stress, F(1,9) = 47.82; p < 0.0001; interaction, F(1,9) = 0.7206; p = 0.4180; two-way RM ANOVA; no restraint, p = 0.9190; restraint, p = 0.6193; veh, p = 0.0010; CNO, p = 0.0030; Sidak's post hoc). F, The time rats spent following was not affected by CNO treatment (main effect of treatment, F(1,9) = 0.0318; p = 0.8623; two-way RM ANOVA) or subjecting animals to restraint stress (main effect of stress, F(1,9) = 4.918; p = 0.0538; stress × treatment interaction, F(1,9) = 0.8338; p = 0.3850; two-way RM ANOVA). G, Neither CNO treatment or stress exposure affected the time spent sitting together (main effect of treatment, F(1,9) = 0.1008; p = 0.7581; main effect of stress F(1,9) = 0.0536; p = 0.8221; with no interaction, F(1,9) = 0.1427; p = 0.7143; two-way RM ANOVA). H, Subjecting animals to restraint stress decreased time spent grooming/sniffing in both the vehicle- (p = 0.0095; uncorrected Fisher's LSD) and CNO-treated groups (p = 0.0284; uncorrected Fisher's LSD). Administration of CNO did not alter total grooming/sniffing time (main effect of treatment, F(1,9) = 0.7077; p = 0.4220; main effect of stress, F(1,9) = 17.4800; p = 0.0024; stress × treatment interaction, F(1,9) = 0.4411; p = 0.5232; two-way RM ANOVA). Data presented as mean ± SEM; *p < 0.05; **p < 0.01 post hoc test.
Chemogenetic activation of either BLA→BNST or BLA→CeA pathways decreases SI
To test whether activation of either of these circuits, such as would occur during exposure to a stress, could alter SI, we used the same paradigm as above to activate BLA→BNST or BLA→CeA projections using CAV2-cre into the BNST or CeA and the excitatory, AAV5-DIO-DREADD-Gq, into the BLA. In our previous experiments, vehicle treatment did not affect SI behavior in animals that had DREADD-Gi expression in the BLA→BNST pathway or the BLA→CeA pathway. Therefore, vehicle-treated animals were combined so that equal numbers of BLA→BNST and BLA→CeA were pooled together to reduce the unnecessary use of animals. Similar to previous experiments, baseline SI was recorded prior to the surgeries. Again, SI was not different among the groups receiving saline injections at W1 (Fig. 5A). At W3 and W4 after surgery, administration of CNO significantly decreased SI performance in both BLA→BNST and BLA→CeA groups compared with vehicle-treated animals (Fig. 5A). Restraint stress decreased SI similarly across all groups, and interestingly, no additive effect of stress and CNO treatment on SI (Fig. 5B) or grooming/sniffing (Fig. 5E) was observed. CNO treatment did not alter the time following (Fig. 5C) but decreased the time the animals sat together (Fig. 5D); there were no additive effects of stress with CNO on any of the behaviors scored (Fig. 5C–E).
Chemogenetic activation of the BLA→BNST or the BLA→CeA projection decreases SI times. Animals received bilateral CAV2-cre in the BNST or the CeA and bilateral AAV-DIO-DREADD-Gq in the BLA to selectively decrease the threshold for activation of either the BLA→BNST or the BLA→CeA projections as outlined in Figures 3C and 4C. A, Administration of CNO decreased SI in all groups receiving viral injections (W3 and W4; veh n = 4; CNO–BNST n = 5; CNO–CeA n = 6; main effect of treatment, F(2,12) = 6.732; p = 0.0110; main effect of time, F(1.178,14.14) = 68.54; p < 0.0001; interaction, F(4,24) = 11.37; p < 0.0001; two-way RM ANOVA). At W1, saline did not alter SI in any of the groups tested (CNO–BNST, p = 0.9912; CNO–CeA, p = 0.8067). At W3 and W4, CNO significantly decreased SI in the virus-treated groups (CNO–BNST: W3, p = 0.0004; W4, p = 0.0027; Tukey's post hoc; and CNO–CeA: W3, p = 0.0004; W4, p = 0.0015; Sidak's post hoc). B, Subjecting animals to restraint stress significantly decreased SI in the saline-treated group (gray diagonal hatch bars). Administration of CNO 45 min before a 30 min restraint stress session did not alter the stress-induced decrease in SI, nor was there a further decrease in SI after treatment. All animals had decreased SI from their own baselines but no difference between treatments (main effect of treatment, F(2,12) = 1.575; p = 2,470; main effect of stress, F(1,12) = 8.105; p = 0.0147; interaction, F(2,12) = 7.710; p = 0.0070; two-way RM ANOVA; veh p = 0.0034; Sidak's post hoc; no restraint, CNO–BNST, p = 0.0448; CNO–CeA, p = 0.0069; restraint, CNO–BNST, p = 0.9913; CNO–CeA, p = 0.8085; Tukey's post hoc). C, There were no significant effects of either treatment or stress on the time the rats spent following (main effect of treatment, F(2,12) = 0.0941; p = 0.9109; main effect of stress F(1,12) = 2.754; p = 0.1229; stress × treatment interaction, F(2,12) = 2.1920; p = 0.1543; two-way RM ANOVA). D, There were no significant effects of either treatment or stress on the time the rats spent sitting together (main effect of treatment, F(2,12) = 2.278; p = 0.1450; main effect of stress, F(1,12) = 0.7837; p = 0.3932; interaction, F(2,12) = 1.939; p = 0.1864; two-way RM ANOVA). E, Administration of CNO decreased grooming in unstressed rats receiving viral injections. Subjecting rats to restraint stress decreased grooming/sniffing in the vehicle-treated group (main effect of treatment, F(2,12) = 1.677; p = 0.2278; main effect of stress, F(1,12) = 5.092; p = 0.0435; interaction F(2,12) = 3.218; p = 0.0761; two-way RM ANOVA; CNO–BNST, p = 0.0275; CNO–CeA, p = 0.0144; uncorrected Fisher's LSD). Data presented as mean ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001.
Electrophysiological properties of BLA→BNST neurons and cellular responses to NPY
After allowing 7 d for transport of retrobeads, acute slice preparations of BLA were made as previously described (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018; Mackay et al., 2019; Michaelson et al., 2020). BLA PNs whose somata demonstrated the presence of retrobeads (Fig. 6A,B) were identified as BLA→BNST projection neurons, and their properties were compared with adjacent, unlabeled neurons (nonBNST-projecting neurons). Identified BLA→BNST neurons had properties characteristic of BLA PNs from previous studies, and we observed no significant differences in their capacitance (Fig. 6C), H current (Ih; Fig. 6D), inward rectifier current (IR; Fig. 6E), or membrane potential (Fig. 6F) from those of nonBNST-projecting cells. The holding current at −75 mV (Irest; Fig. 6G) and the rheobase (Fig. 6H) were significantly less in the BNST-projecting neurons, whereas input resistance was not significantly different (Fig. 6I). Based on our immunohistochemical findings (Fig. 2), we tested whether these neurons responded to acute application of NPY. Bath application of 1 μM NPY hyperpolarized BLA→BNST neurons (Fig. 7A) and decreased both Ih (Fig. 7B) and IiR (Fig. 7C). Acute NPY increased Irest (Fig. 7D) and increased the rheobase (Fig. 7E) of BNST-projecting neurons, but had no effect on their input resistance (Fig. 7F).
Electrophysiological properties of BLA→BNST neurons. A, Schematic representation of retrobead injection in the BNST. Backfilled neurons were visualized in freshly cut BLA slices to perform projection-specific electrophysiological recordings. B, Experimental timeline. C, Comparison of capacitance measurements of BNST and nonBNST-projecting BLA neurons (BNST-projecting n = 21; nonBNST-projecting n = 26; t45 = 0.1274; p = 0.8992; unpaired t test). D, Measurement of Ih in BNST- and nonBNST-projecting BLA neurons (main projection effect, F(7,360) = 1.139; p = 0.2866; two-way ANOVA). E, Comparison of IIR in BNST- and nonBNST-projecting BLA neurons (main projection effect, F(1,405) = 5.875; p = 0.0158; two-way ANOVA). Comparison of (F) membrane potential (t45 = 1.116; p = 0.2703; unpaired t test); (G) membrane holding current at Vh = −75 mV (t41 = 3.204; p = 0.0026; unpaired t test); (H) rheobase (t45 = 4.237; p = 0.0001; unpaired t test); and (I) input resistance (t45 = 1.039; p = 0.3046; unpaired t test) in BNST- and nonBNST-projecting BLA neurons. Bars represent mean values ± SEM; *p < 0.05; **p < 0.01; ***p < 0.001.
Acute effects of NPY application on BLA neurons projecting to the BNST. Effect of bath application of 1 μM NPY on (A) resting membrane potential (main effect of treatment, F(2,68) = 15.66; p < 0.0001; one-way ANOVA); (B) the amplitude of Ih (main effect of treatment, F(1,50) = 6.824; p = 0.0118; two-way ANOVA); (C) IIR (main effect of treatment, F(1,50) = 5.542; p = 0.0225; two-way ANOVA); (D) membrane current at the nominal resting potential of −75 mV (main effect of treatment, F(2,68) = 8.275; p = 0.0006; one-way ANOVA; control vs NPY, p = 0.0006; control vs wash, p = 0.7137; NPY vs wash, p = 0.0178; Tukey's post hoc); (E) rheobase (t25 = 5.220; p < 0.0001; paired t test); and (F) input resistance (t21 = 1.401; p = 0.1758; unpaired t test). Data presented as mean ± SEM; *p < 0.05; ***p < 0.001; ****p < 0.0001.
Repeated intra-BLA NPY injections induce neuroplasticity of BLA→BNST neurons
Repeated intra-BLA NPY treatment produces long-lasting changes in the excitability and dendritic structure of BLA neurons, which is associated with the development of stress resistance (Sajdyk et al., 2008; Silveira Villarroel et al., 2018; Michaelson et al., 2020). We wanted to determine whether BLA→BNST neurons also show similar responses to in vivo rpNPY treatment which would support their involvement in the NPY-mediated generation of stress resistance. Intra-BLA injections of NPY or PBS (vehicle) were administered over 5 d to adult rats (Sajdyk et al., 2008; Silveira Villarroel et al., 2018), and retrobeads were simultaneously injected into the BNST to selectively label BLA→BNST neurons (Fig. 8A). At W4 after the injections, acute BLA slices were prepared, and the properties of BLA→BNST neurons were compared with nonBNST-projecting BLA neurons in both NPY- and vehicle-treated animals (Fig. 8B). Intra-BLA injections of NPY increased SI and reduced whole-cell capacitance of BLA→BNST neurons compared with that of nonBNST-projecting neurons, as well as compared with BLA→BNST and nonBNST-projecting neurons from animals that received only PBS (Fig. 8); this suggests that the effect of NPY on capacitance occurs predominantly on BLA→BNST neurons (Fig. 8C). RpNPY treatment had no effect on the maximal Ih amplitude (Fig. 8D) but decreased the IIR in BLA→BNST compared with nonBNST-projecting neurons and BLA neurons from control animals (both BLA→BNST and nonBNST-projecting; Fig. 8E). Resting membrane potential (Fig. 8F) and resting membrane current (when held at −75 mV; Fig. 8G) were similar in all groups of BLA neurons. Interestingly, within the NPY-treated group, rheobase increased in nonBNST-projecting compared with that in BLA→BNST neurons (Fig. 8H). NPY treatment also increased input resistance in BLA→BNST compared with that in nonBNST-projecting BLA neurons (Fig. 8I).
Intra-BLA injections of NPY decrease capacitance of BLA→BNST neurons. A, A schematic representation of retrobead injections in the BNST and bilateral intra-BLA cannulation for in vivo repeated NPY injections. B, Experimental timeline. Surgery was performed and 1 week later, the animals received bilateral daily injections of NPY for 5 d. At W4, brains were harvested for slice recordings. Effect of intra-BLA treatment on (C) capacitance (PBS, BNST-projecting n = 10; nonBNST-projecting n = 36; NPY, BNST-projecting n = 19; nonBNST-projecting n = 30; main treatment effect, F(1,91) = 15.24; p = 0.0002; main projection effect, F(1,91) = 5.269; p = 0.0240; two-way ANOVA); (D) H current (Ih; main treatment effect, F(3,640) = 1.277; p = 0.2814; main voltage effect, F(7,640) = 401.5; p < 0.0001; interaction, F(21,640) = 0.2900; p = 0.9993; two-way ANOVA); and (E) amplitude of IIR (main treatment effect, F(3,711) = 5.489; p = 0.0010; main voltage effect, F(8,711) = 848.2; p < 0.0001; interaction, F(24,711) = 3.427; p < 0.0001; two-way ANOVA; Tukey's post hoc). Both the (F) membrane potential (main treatment effect, F(1,80) = 0.09754; p = 0.7556; main projection effect, F(1,80) = 1.711; p = 0.1946; interaction, F(1,80) = 0.3974; p = 0.5302; two-way ANOVA) and the (G) resting membrane potential (main treatment effect, F(1,80) = 0.09754; p = 0.7556; main projection effect, F(1,80) = 1.711; p = 0.1946; interaction, F(1,80) = 0.3974; p = 0.5302; two-way ANOVA) were not different. H, Rheobase of BLA→BNST from animals treated with repeated NPY or vehicle injections (main treatment effect, F(1,179) = 0.8906; p = 0.3482; main projection effect, F(1,79) = 5.804; p = 0.0183; interaction, F(1,79) = 2.516; p = 0.1167; two-way ANOVA; Sidak's post hoc). I, Input resistance in BNST-projecting neurons from animals that received repeated NPY injections (main treatment effect, F(1,78) = 0.3187; p = 0.5740; main projection effect, F(1,78) = 5.454; p = 0.0221; interaction, F(1,78) = 1.266;p = 0.2640; two-way ANOVA; Sidak's post hoc). Data presented as mean ± SEM; **p < 0.01; ***p < 0.001; ##p < 0.01; ####p < 0.0001.
After recording, the neurobiotin-filled neurons were fixed and stained, and the overall structure and neuronal complexity of BLA neurons were determined (Michaelson et al., 2020). Intra-BLA NPY treatment significantly decreased the total dendritic length of BLA→BNST neurons (Fig. 9A) without significantly altering the number of dendrites (Fig. 9B) or the number of branch points (Fig. 9C). Sholl analysis revealed that the decrease in the dendritic arborization occurred most prominently between 190 and 210 μm from the soma (Fig. 9D). Together, these results suggest that NPY reduces excitability and induces hypotrophy of BLA→BNST neurons, which associates with the prolonged increase we observe with SI (Sajdyk et al., 2008; Silveira Villarroel et al., 2018; Michaelson et al., 2020).
Repeated intra-BLA injections of NPY decrease dendritic length of BLA→BNST-projecting neurons. A, Repeated NPY treatment reduces total dendritic length of BLA→BNST-projecting neurons (BNST–PBS n = 14; BNST–NPY n = 20; t29 = 2.288; p = 0.0296; unpaired t test). B, The number of dendrites was not different between the two groups (t8 = 0.1055; p = 0.9186; unpaired t test). C, The number of branch points was not different between groups (t28 = 1.096; p = 0.2824; unpaired t test). D, Sholl analysis of the number of intersections across the distance from soma revealed a significant decrease at 190–210 μm distance in BNST-projecting neurons from animals that received repeated intra-BLA NPY compared with PBS-treated controls (main treatment effect, F(1,1,479) = 44.06; p < 0.0001; main distance effect, F(50,1,479) = 123.0; p < 0.0001; interaction, F(50,1,479) = 1.525; p = 0.0113; two-way ANOVA; Sidak's post hoc). Bars represent mean values ± SEM; *p < 0.05.
Discussion
NPY is associated with the development of resilience and buffering of stress responses in animals and humans (Yehuda et al., 2013; Sah et al., 2014; Serova et al., 2014; Sayed et al., 2018). While NPY is posited as a novel treatment for anxiety-related disorders, there is very little known about the neural circuitry associated with the behavioral responses to NPY. Initial work by Sajdyk et al. (2008) demonstrates that rpNPY injections into the BLA induce a form of neuroplasticity that both generates persistent decreases in stress-related behaviors and prevents stress-induced decreases in SI in male rats. We have used this model to characterize cellular and molecular changes in the BLA to better understand the NPY-responsive circuitry related to the generation of stress resistance. Within the BLA, NPY acutely reduces the activity of PNs through inhibition of the nonselective cation current, Ih, resulting in increased SI, whereas rpNPY injections downregulate HCN1 channel expression and induce neuronal hypotrophy, producing a less excitable BLA and a persistent increase in SI that lasts beyond 2 months (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018; Michaelson et al., 2020). The studies presented here extend this work and support a key role of BLA→BNST neurons in transducing these stress-buffering actions of NPY on SI.
Activation of NPY receptors in the BLA acutely increases SI and also prevents stress- and CRF-induced decreases in SI, which, in part, defines NPY's role as a stress buffer (Heilig et al., 1994; Shekhar et al., 2005; Sajdyk et al., 2006, 2008; Karlsson et al., 2008; Giesbrecht et al., 2010). Here we identified that BLA projections to both the BNST and the CeA express NPY receptors and HCN1 channel immunoreactivity, indicating that both neuronal populations are equipped to respond to endogenous NPY release. The association of HCN1 immunoreactivity with NPY receptor expression provides anatomical support for NPY and the hyperpolarization of BLA PNs via inhibition of Ih, carried by HCN1 channels (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018). The presence of Y5 receptors within BLA→BNST neurons suggests that they could also demonstrate neuroplastic changes to rpNPY (Leitermann et al., 2012; Silveira Villarroel et al., 2018; Michaelson et al., 2020). While we did not assess the degree of coexpression of Y1 and Y5 receptor immunoreactivity on these projection neurons, we know from previous studies that ∼50% of Y1R-immunoreactive neurons coexpress Y5R immunoreactivity (Wolak et al., 2003; Longo et al., 2015). Therefore, it is possible that populations of these neurons express both receptor subtypes, which could have implications for pharmacological activation of these cells (Gehlert et al., 2007).
RpNPY injections into the BLA produce persistent increases in, and ameliorate the effects of stress on, SI that persist long after the final injection (Sajdyk et al., 2008). This NPY-induced stress resistance is associated with decreased excitability of BLA PNs, decreased HCN1 channel expression, and a pronounced hypotrophy of BLA neurons (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018; Michaelson et al., 2020). While activation of both Y1 and Y5 receptors acutely increases SI, the long-term effects are mediated via activation of Y5 receptors (Michaelson et al., 2020). The expression of immunoreactivity both for the Y5 receptor and for calcineurin in BLA→BNST neurons promotes the induction of long-term increases in SI (Sajdyk et al., 2008; Leitermann et al., 2012) and suggests that this pathway could exhibit hypotrophy in response to rpNPY (Michaelson et al., 2020). To test the ability of BLA→BNST neurons to respond to NPY, we recorded cellular activity from BLA neurons identified as innervating BNST compared with nonBNST-projecting cells. Both categories (BNST- and nonBNST-projecting) of neurons exhibited properties characteristic of BLA PNs from previous studies (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018; Mackay et al., 2019), with no differences noted in their capacitance, Ih, inward rectifier current, membrane potential, or input resistance. However, BNST-projecting neurons had a lower resting current and rheobase, which, in previous work, did not predict responsiveness to NPY (Giesbrecht et al., 2010). Application of NPY hyperpolarized BLA→BNST neurons and reduced Ih as we observed previously (Giesbrecht et al., 2010; Silveira Villarroel et al., 2018).
BLA→BNST neurons also demonstrated hypotrophy in response to rpNPY treatment in vivo. These neurons showed significant decreases in overall dendritic length accompanied by a concomitant decrease in capacitance, an excellent proxy for reduced dendritic length (Michaelson et al., 2020). No changes in excitability or Ih were observed 4 weeks after in vivo NPY treatment as we noted earlier (Silveira Villarroel et al., 2018); this was a bit surprising, but given the neuronal diversity of BLA neurons, it is plausible that other populations (nonBNST-projecting) may also exhibit this response. These previous studies did not characterize BLA neurons by their projections, so while BLA→BNST neurons demonstrate responses to both acute and repeated treatment with NPY, this population may not exhibit downregulation of Ih. Furthermore, we note that labeling for NPY receptors is not uniform in BLA neurons; some express only Y1 or Y5, while others coexpress both, possibly accounting for differences in neuronal responses (Wolak et al., 2003; Gehlert et al., 2007; Longo et al., 2015).
Inhibition of the BLA results in a number of dysregulated behaviors (Elorette et al., 2020), and perhaps the least-studied of these has been modulation of social behaviors (Twining et al., 2017; Elorette et al., 2020; Rojek-Sito et al., 2023). Administration of NPY reduces the activity of the BLA, and we observe the strongest influence on SI while seeing no significant changes in open-field, elevated plus maze or locomotor behaviors (Sajdyk et al., 2008; Silveira Villarroel et al., 2018). Given the widespread distribution of NPY receptors in the BLA, we hypothesized that the effects of NPY on SI would be determined by activation of NPY receptors on neural outputs specifically involved in regulating social behaviors. We focused on projections to the BNST, since NPY treatment reduces stress-induced increases in c-Fos there, consistent with the observed stress-resilient effects on SI (Sajdyk et al., 2008). Suppressing activity of BLA→BNST neurons using inhibitory DREADDs not only increased SI but also prevented stress-induced decreases in SI, replicating the responses seen with intra-BLA administration of NPY (Sajdyk et al., 2004, 2008; Silveira Villarroel et al., 2018). These findings implicate the BLA→BNST projection in the maintenance of SI and modulation of stress on SI in male rats. Activation of this projection with excitatory DREADDs decreased SI without producing an additive effect of stress on SI. Therefore, this supports BLA→BNST neurons as bidirectionally modifying SI, as seen with other brain circuits (Felix-Ortiz et al., 2016). The importance of the BNST in the expression of SI and social behaviors is well documented (Clauss et al., 2019; Flanigan and Kash, 2022; Jacobs et al., 2023); through these studies, we have expanded this circuitry by demonstrating the significance of BLA inputs to the BNST for modulating basal and stress-induced changes in social behaviors (Vantrease et al., 2022).
Inhibition of BLA→CeA neurons did not alter SI, but activation of this circuit, as would occur with stress, did decrease SI. Since the BLA→CeA projection influences the expression and modulation of fear-related behaviors (Massi et al., 2023), it is not surprising that activation would also inhibit SI, as it would ill serve the organism to seek out SIs in the presence of a threat. Relatively little is known about the role of the CeA in the expression of social behaviors. The lack of an SI response upon inhibition of this projection implies that BLA→CeA neurons do not contribute to the basal regulation of SI and that the quiescence of this projection is permissive for SI. Recent studies demonstrate that the CeA serves as a nexus for the initiation and modulation of social behaviors, receiving inputs from the ventral tegmental area and cingulate cortex (Rojek-Sito et al., 2023). It is conceivable that BLA inputs could relay stress signals which would modify the activity of specific neuronal circuits in the CeA and thereby influence the expression of SI (Fadok et al., 2018). Furthermore, collateralizations of BLA neurons to both the BNST and CeA have been identified (Bienkowski and Rinaman, 2013; Reichard et al., 2017) which could contribute to the regulation of more complex behaviors.
The BLA→BNST pathway appears to be a key component of the effects of NPY on SI and the ability of NPY to buffer stress effects on these neurons and related behaviors (Heilig et al., 1994; Sajdyk et al., 2006, 2008). These results highlight the importance of NPY input to the BLA (Rostkowski et al., 2013; McDonald and Zaric, 2015a; Leitermann et al., 2016). The existence of multiple sources of NPY innervating the BLA suggests the possible differential modulation by NPY of BLA output under control and stress conditions. We anticipate that these studies and future work will help elucidate the neural circuitry associated with the actions of an endogenous anxiolytic neuropeptide.
Footnotes
We acknowledge Nicole Ferrara, PhD, for her insightful and helpful discussions on the assessment of social interaction behavior. This work was supported by National Institutes of Health R01 MH090297 (J.H.U., W.F.C.) and R21 MH114184 (J.H.U., W.F.C.).
The authors declare no competing financial interests.
↵*M.B.’s and A.P.M.T.’s present address: Department of Psychiatry, New York State Psychiatric Institute and Columbia University, New York, New York 10032 and National Institute on Drug Abuse Intramural Research Program, Baltimore, Maryland 21224-2816.
- Correspondence should be addressed to Janice H. Urban at janice.urban{at}rosalindfranklin.edu.
