The pPVT^glu- CeA^GABA circuit is involved in regulating hyperalgesia in the chronic NTG mouse model in males

Abstract

Background

Chronic migraine imposes a large disease burden; however, its underlying mechanisms remain poorly understood. The paraventricular thalamus (PVT), traditionally linked to arousal, has recently been shown to regulate pain and negative emotions. However, the role of the PVT and its related neural circuits in migraine remains insufficiently investigated. The neglect of the PVT could be due to its high basal neuronal activity under physiological conditions, complicating the isolation of disease-specific changes.

Methods

A mouse model of chronic migraine was established by repeated intraperitoneal injection of nitroglycerin (NTG) i.p. The mechanical hyperalgesia, anxiety, and depression-like behaviors of the mice were evaluated using the von Frey test, open field test, elevated plus maze test, and tail suspension test. The activation status of the posterior PVT (pPVT) was determined via the FosTRAP strategy, and behavioral tests were performed after its neurons were inhibited chemogenetically. Neural circuits potentially involved in regulation were subsequently identified by combining anterograde tracing with activity mapping in downstream regions. Finally, chemogenetic manipulation was employed to further verify the role of the circuit in NTG-induced chronic migraine.

Results

Mice with chronic migraine exhibited mechanical hyperalgesia and anxiety-like behavior, whereas depression-like behavior was not observed. Compared with the control group, the model group showed increased activation in the pPVT, as indicated by the results of the FosTRAP strategy. Chemogenetic inhibition of pPVT glutamatergic neurons can alleviate mechanical hyperalgesia in model mice without affecting anxiety-like behaviors. Projections from the pPVT innervate multiple regions, including the central amygdala (CeA). The model group showed increased activation in the CeA and its afferent input from the pPVT (pPVT^glu- CeA). Chemogenetic inhibition of CeA GABAergic neurons alleviated both mechanical hyperalgesia and anxiety-like behavior in model mice. However, chemogenetic inhibition of both pPVT^glu- CeA and CeA GABAergic neurons receiving input from the pPVT (pPVT- CeA^GABA) can ameliorate mechanical hyperalgesia in model mice without affecting anxiety-like behaviors.

Conclusions

The pPVT^glu- CeA^GABA circuit is involved in mechanical hyperalgesia behaviors in chronic migraine, and could be a promising candidate for chronic migraine treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s10194-026-02285-3.

Background

Migraine is the second leading cause of global disability worldwide [1]. Chronic migraine imposes a substantial disease burden on both patients and society [2, 3], with a high comorbidity of affective disorders, including anxiety and depression [2, 4]. However, the pathophysiological mechanisms that underlie the development of chronic migraine and associated emotional disorders are still poorly understood.

As a midline thalamic nucleus, the paraventricular nucleus of the thalamus (PVT) was previously believed to be associated with general arousal [5], but it is implicated by a growing body of evidence in the regulation of heterogeneous pain and mood disorders [5, 6]. The PVT regulates inflammation pain, as validated in both complete Freund’s adjuvant(CFA)-induced [7] and visceral pain models [8, 9]. Other studies [10, 11] have shown that the PVT is also involved in regulating neuropathic pain. Different subregions of the PVT (anteroposterior distinction) may be predominantly responsible for distinct functions. One study [6] reported that the anterior PVT (aPVT) has higher expression of arousal-related neuropeptides, whereas the posterior PVT (pPVT) expresses emotion-related neuropeptides.

The involvement of PVT in migraine has not been firmly established. The existing data, primarily from one study, suggest increased activation of the PVT in a levcromakalim-induced chronic migraine model [12]. This relative paucity of relevant research could be due to the high baseline activity in the PVT under physiological conditions. For arousal-related nuclei, such as the PVT, conventional activity-mapping methods including c-Fos [12–14] or phosphorylated ERK (pERK) [15] immunofluorescence staining struggle to differentiate acute, stimulus-evoked neural activation (e.g., in response to pain) from accumulated basal expression. The FosTRAP strategy can effectively address this issue. Tamoxifen-induced temporal control allows this strategy to fluorescently label activated neurons within a narrowly defined time window [16]. This enhances the signal-to-noise ratio for identifying pain-responsive neurons.

In this study, we employed a FosTRAP strategy to determine whether the pPVT regulates mechanical hyperalgesia and anxiety-like behavior in NTG-induced migraine and identified a functional circuit connected to the pPVT that is involved in regulating the aforementioned behavior. We will then chemogenetically manipulate the circuit to confirm its role in chronic migraine.

Methods

Animals

Eight-week-old male C57BL/6J mice purchased from SiPeiFu Biotechnology Co., Ltd (Beijing, China) were used for behavioral validation. Ai9 mice (JAX stock #007909) and FosTRAP2 mice (JAX stock #030323) were kindly gifted by Dr. You Wan’s laboratory at Peking University, Beijing, China. The Ai9 mice, which exhibit robust tdTomato fluorescence following Cre-mediated recombination, were used in this study to label and visualize Cre-expressing cells. We generated the male Fos-tdTomato mice (8–10 weeks old) used in our study by crossing homozygous FosTRAP2 mice with homozygous Ai9 mice. VGAT-Cre mice (Genepax Biotechnology Co., Ltd, Changzhou, China; GAP1041) and Vglut2-Cre mice (Cyagen Biosciences Inc., Suzhou, China; stock #C001240) were used for neural circuit tracing and manipulation. Experimental mice (8–10 weeks old) were subsequently bred in-house by crossing homozygous males with homozygous females. All mice were housed under a 12 h light-dark cycle (200-250 lx) with controlled temperature and humidity, and had ad libitum access to food and water. A one-week acclimatization period was provided for all the newly arrived mice before the experimental procedures.

Animal models

Nitroglycerin (NTG, 5 mg/mL in ethanol, Beijing Yimin Pharmaceutical Co., Ltd., Beijing, China) was used in the modeling. We freshly diluted it in 0.9% saline at a ratio of 1:15. We administered a dose of 10 mg/kg intraperitoneally (i.p.) to the NTG group. The injection volume was 30 ml/kg. Absolute ethanol was diluted in saline at an identical dilution ratio for the VEH (vehicle control) group.

Behavior tests

von Frey test

The calibrated von Frey filaments (Aesthesio, Danmic Global, USA) were used to test periorbital and hind paw mechanical threshold. To avoid potential harm, the von Frey filaments used for testing are limited to a cutoff force of 2 grams.

Prior to the formal periorbital von Frey test, mice underwent a 3-day habituation phase where they were gently placed on the palm of researchers without constraint for 10 min daily. On the formal test day, mice stayed calmly on the palm, and the monofilament was forced perpendicularly on the periorbital skins according to the sequence specified in the up-down method [17]. The filaments remained bent as “C” or “S” for 3s with an inter-stimulus interval of approximately 20s. Head withdrawal, head shaking, facial grooming, or orbital tightening were defined as positive responses.

Prior to the formal hind paw von Frey test, mice underwent a 3-day habituation phase where they were placed on the testing apparatus for 30 min daily. On the formal test day, the mice stayed in an open-top and open-bottom non-transparent acrylic chamber (10 cm L × 7 cm W × 16 cm H) on a mesh, and the monofilament was forced perpendicularly on the center of the plantar surface of the right hind paw when the mice was calmly standing on all four limbs. Each stimulus lasted for 3 seconds, with a 20-second interval separating consecutive stimuli in the order specified in the up-down method [17]. Paw licking and paw withdrawal were defined as a positive response.

The sequence of filament application and the XO-pattern recording adhered to the up-down method [18]. The XO pattern obtained from the test was used to calculate the 50% withdrawal thresholds on a website [17] (https://bioapps.shinyapps.io/von_frey_ app/).

For the validation of the NTG model, vonFrey tests were conducted before each NTG/VEH administration; for chemogenetic manupulation, vonFrey tests were performed before and 2–2.5 hours after clozapine N-oxide (CNO) administration. The values obtained at these time points were referred to as the preCNO and postCNO thresholds, respectively.

Tail suspension test (TST)

The hardware is a non-transparent box (20 cm L × 20 cm W × 50 cm H) with only one opening on its side (for camera recording). The hardware was cleaned with 75% ethanol before each test. Mice were individually suspended with medical adhesive tape at approximately 1 cm from the tip of the tail from a cylindrical rod mounted vertically at the center of the inside top of the box. The test was recorded with a video camera for 6 minutes (1 min for adaption phase and 5 min for test phase). Immobility was defined as passive hanging and passive swaying (absence of active struggling). Total immobility time was calculated to evaluate depression-like behaviors.

Open field test (OFT)

The hardware was an open-top box (50 cm L × 50 cm W × 40 cm H). The light intensity at the floor level of the open field was 15–20 lx. The open field was wiped with 75% ethanol before testing each mouse. The experimenter positioned the mouse gently in the center of the open field arena, facing it away from the experimenter, and allowed it to freely explore for 15 minutes. The behavioral tests were captured and subsequently analyzed by SuperMaze software (Shanghai Xinruan Information Technology Co., Ltd., Shanghai, China). For analysis, the central 25 cm × 25 cm square was defined as the center area, while the remainder was designated as the peripheral area. The criterion for entry into an area was set at a threshold of > 80% of the area of the mouse being detected within the region. Entrance into the center area and time spent in the center area were calculated to evaluate anxiety-like behaviors.

Elevated plus maze test (EPM)

The hardware is a 60-cm-high plus maze (Shanghai Xinruan Information Technology Co., Ltd., Shanghai, China; XR-XG201) consisting of two open arms (35 × 5 cm, 15-20 lx), two closed arms (35cmL × 5cmW × 15cmH, 5 lx), and a central platform (5 × 5 cm). The maze was wiped with 75% ethanol before testing each mouse. The mouse was gently placed in the central platform facing the open arms and away from the experimenter and allowed to freely explore for 15 minutes. The test was recorded and analyzed by SuperMaze software (Shanghai Xinruan Information Technology Co., Ltd., Shanghai, China). The criterion for entry into an area was set at a threshold of > 80% of the mouse’s area. Entrance into open arms and time in open arms were calculated to evaluate anxiety-like behaviors.

4-hydroxytamoxifen (4-OHT) preparation and administration

In our Fos-tdTomato mice, 4-OHT was administered to induce nuclear translocation of Cre recombinase, thereby labeling the neurons that expressed cFos for approximately 6 hours [16] surrounding a 4-OHT injection with red fluorescence. Following 3 days of accumulation, the Cre-dependent fluorescent protein becomes readily detectable [16, 19].

To achieve a concentration of 20 mg/mL, 4-OHT (Sigma-Aldrich; H6278) was dissolved in absolute ethanol with shaking at 37 °C and stored at -80 °C. As previously described [20] with minor modifications, the above solution was placed in a 37 °C shaker until thoroughly dissolved prior to use and then diluted 1:10 in corn oil (MedCHemExpress; HY-Y1888). Following the marking of the meniscus levels for corn oil and absolute ethanol on the centrifuge tube wall, the dilution was vortexed until homogeneous. The resulting mixture was incubated at 37 °C until the absolute ethanol had completely evaporated, yielding a solution of 4-OHT in corn oil. The final dose of 4-OHT administered was via intraperitoneal (i.p.) injection at 50 mg/kg. For the comparison of tdTomato-labeled, Fos-positive neurons in the target brain region of Fos-tdTomato mice treated with NTG or VEH, 4-OHT was administered 0.5 h before the last NTG or VEH injection.

Brain tissue collection and brain sectioning

The timing of brain tissue collection varied between experimental paradigms. For the comparison of tdTomato-labeled, Fos-positive neurons in the target brain region of Fos-tdTomato mice treated with NTG or VEH, brain tissue was collected 3 days after 4-OHT administration. For anterograde tracing, retrograde tracing, and brain-wide labeling of target neurons, brain tissue was collected 3 weeks after viral delivery. For efficacy and injection site validation of chemogenetic virus, a one-week washout period (as in the previous study [21]) after behavioral testing was followed by brain tissue collection. Mice were administered CNO, followed by NTG one hour later, and brain tissue was collected two hours after NTG administration. For the comparison of the density of mRuby-labeled terminals from activated pPVT neurons within the CeA between NTG- and VEH-treated FosTRAP2 mice, brain tissue was collected 21 days after 4-OHT administration. For verification of viral expression and ferrule fiber placement in the optogenetic stimulation and fiber photometry experiment, brain tissue was collected after fiber photometry.

Mice were anesthetized by intraperitoneal (i.p.) injection of 1.25% Avertin at a dose of 20 mL/kg. Then the mice were transcardially sequentially perfused with 37 °C phosphate-buffered saline (PBS, PH7.4, 0.1 M, 20 mL), 4 °C PBS (PH7.4, 0.1 M, 20 mL), and 4 °C paraformaldehyde (PFA, 4%, 20 mL). The brain was carefully dissected out of the skull and then postfixed overnight in 4 °C 4% PFA. Brains were cryoprotected by sequential incubation in 15% and 30% (w/w) sucrose solutions until they sank. After the brains were embedded in Tissue-Tek OCT Compound (Sakura Finetek U.S.A. Inc.), coronal brain sections (30 μm) were obtained using a cryostat (Leica Biosystems, Heidelberger, Germany; CM1950). For subsequent experiments, brain sections were collected at fixed intervals along the anterior-posterior axis to cover the entire brain region of interest, aiming to minimize the impact of potential anterior-posterior heterogeneity.

Immunofluorescence staining

The brain sections were incubated in IHC/IF blocking buffer (YangGuangBio, Beijing, China; C220702) for 15 min at room temperature. At 4 °C for overnight (16–18 h), the sections were incubated with primary antibodies diluted in the above blocking buffer. The brain sections were subjected to three 15-minute washes in PBS on an orbital shaker (40 rpm) at room temperature. The brain slides were again subjected to the same washing procedure after being incubated with the species-matched fluorescent secondary antibody for 2 h at room temperature. Afterward, the brain slides were mounted with antifading mounting medium with DAPI (YangGuangBio, Beijing, China;190401) and coverslipped. The antibodies used in this study included: anti-c-Fos (Cell Signaling Technology, #2250, rabbit), anti-NeuN (Abcam, ab177487, rabbit), and anti-GFP (Abcam, ab13970, chicken), Alexa Fluor 647 AffiniPure goat anti-rabbit IgG (YangGuangBio, C081816, 1:400), and Alexa Fluor 488-conjugated goat anti-chicken IgY (Abcam, ab150173). The antibodies were used at a dilution of 1:1000, unless otherwise stated.

Brain slides that did not require immunofluorescence staining were directly mounted, coverslipped, and scanned.

Slide scanning and fluorescence quantification

The slides were scanned using Pannoramic MIDI (3DHISTECH; Budapest, Hungary) with a 20× objective in Extended Depth of Field mode (7 steps; 2 μm step size). Brain regions were identified by referencing the Paxinos and Franklin mouse brain atlas (4th edition). The CellQuant analysis function of QuantCenter software (3DHISTECH) was used to analyze the fluorescent signal indicating nuclear morphology (c-Fos immunoreactivity and the virally expressed H2B-tagged nuclear EGFP). Automated identification of the same targets was performed with a consistent set of parameters (slide color adjustment, nuclear size, nuclear intensity, nuclear contrast, nuclear merge, and nuclear smoothing). Other cellular morphology signals were visually identified. To minimize bias, the experimenter who was blinded to the experimental groups analyzed the brain sections under the same scan and color adjustment settings. The HsitoQuant analysis function of QuantCenter software (3DHISTECH) was used to quantify the fluorescent signal intensity of mRuby-tagged fiber terminals.

Stereotaxic surgery, virus injection, and optical ferrule fiber implantation

Anesthesia and fixation of the mice before the surgery were carried out in accordance with previous literature [22].

Virus injections were delivered via a glass micropipette injection pump (RWD Life Science, R-480). The stereotaxic coordinates and injection volumes are as follows (coordinates are given relative to the bregma): pPVT (30° angle in the coronal plane, AP-1.25, ML1.5, DV-2.59 100 nL, and AP-1.87, ML1.5, DV-2.59 100 nL), CeA (AP-1.20, ML±2.87, DV-4.82, 150 nL, and AP-1.70, ML±2.87, DV-4.62, 50 nL). Constant rate was 40 nL/min. In this study, viral tools were used for neural circuit tracing, neural circuit chemogenetic manipulation, and combined optogenetics and calcium imaging. All viruses used in this study were expressed for three weeks, unless otherwise specified.

For anterograde tracing, rAAV-hSyn-DIO-mGFP-T2A-Synaptophysin-mRuby-WPRE-hGH polyA (AAV2/9, PT-1244, BrainVTA, Wuhan, China), rAAV-hSyn-CRE-WPRE-hGH polyA (AAV2/1, titer≥1.00E+13vg/ml, PT-0136, BrainVTA), and rAAV-hSyn-DIO-H2B-EGFP-T2A-Synaptophysin-mRuby-WPREs (PT-4596, BrainVTA) were used.

For retrograde tracing, rAAV-EF1a-DIO-EGFP (AAV2/11, BC-0015, Brain Case, Shenzhen, China) was used.

For chemogenetic manipulation: rAAV-EF1a-DIO-hM4D(Gi)-EGFP-WPREs (AAV2/9, PT-0987, BrainVTA), rAAV-EF1a-DIO-EGFP-WPRE-hGH-pA (AAV2/9, PT-0987, BrainVTA), rAAV-CaMKIIα-DIO-hM4D(Gi)-P2A-EGFP (AAV2/9, BC-1440, Brain Case), rAAV-CaMKIIα-DIO-EGFP (AAV2/9, BC-0487, Brain Case), rAAV-hSyn-SV40 NLS-Cre (AAV2/11, BC-0159, Brain Case), rAAV-hSyn-ConFon-hM4D(Gi)-EGFP-WPRE-hGH polyA (AAV2/9, PT-1572, Brain Case), rAAV-hSyn-ConFon-EGFP-WPRE-hGH polyA (AAV2/9, PT-9192, BrainVTA), and rAAV-hSyn-SV40 NLS-Flp-P2A-mCherry (AAV2/1, titer≥1.00E+13vg/mL, BC-0999, Brain Case) were used. Mice with intended viral expression in the target region were included for subsequent behavioral analysis.

For combined optogenetic stimulation and fiber photometry, rAAV-CaMKIIa-Chrimson-mCherry-WPRE-hGH polyA (AAV2/9, PT-1414, BrainVTA), rAAV-CaMKIIa-mCherry-WPRE-hGH polyA (AAV2/9, PT-0108, BrainVTA) and rAAV-EF1α-DIO-GCaMP6s (AAV2/9, BC0086, Brain Case) were used. An optic ferrule fiber (5.3 mm long, 200 μm core diameter, 1.25 mm outer diameter, 0.37 numerical aperture; Inper, Zhejiang, China) was implanted at AP-1.31 mm, ML±2.87 mm, DV-4.62 mm, which was 200 μm above the calcium indicator virus injection site (AP-1.31 mm, ML±2.87 mm, DV-4.82 mm). To secure the ferrule fibers, we used dental cement and screws to fix them to the skull.

Intravenous viral delivery

For the brain-wide labeling of target neurons, a total volume of 20 μL of the rAAV-EF1α-DIO-H2B-EGFP (AAV2/PHP.eB, titer≥5.00E+12 vg/mL, BC-0782, Brain Case) was diluted in 1000 μL 0.9% saline and administered to each mouse via slow tail vein injection using an insulin syringe (30 G, 12.7 mm, U100). We allowed 3 weeks for viral expression.

Optogenetic stimulation and fiber photometry

Optogenetic stimulation and fiber photometry were performed with an Optogenetics Tripple Color MultiChannel Fiber Photometry platform (QAXK-FPS-SS-MC-LED-OG, Thinker Tech Nanjing Bioscience Inc., Nanjing, China) composed of a red-light optogenetic stimulator and a fiber photometry system. Optogenetic stimulation and fiber photometry were integrated through a combination of optical and digital signaling pathways. An optogenetic stimulation laser (635 nm, 20 Hz, 5 s) and a fiber photometry recording excitation light (reference channel: 405 nm, signal channel: 470 nm) were delivered through the same ferrule fiber to the target brain region. Furthermore, the two devices were synchronized via a BNC cable. This connection allowed the optogenetic stimulator to send a precise TTL event mark to the photometry system upon each light pulse, enabling subsequent analysis of peri-stimulation calcium dynamics. The output light power was calibrated prior to animal implantation with a power meter. The power at the tip of the optical fiber (with the same ferrule fiber as the implanted) was measured and adjusted to the following target powers: 5 mW at 635 nm, 30 μW at 405 nm, and 45 μW at 470 nm. For each recording site, five trials of optogenetic stimulation were recorded with a 5-minute interstimulus interval. Following fiber photometry, the mice were perfused for brain collection, and the ferrule fiber track in the tissue was used as a landmark. A placement was deemed correct if the tip of the ferrule fiber track was positioned directly above and adjacent to the CeA. Only fiber photometry data from animals with both correct ferrule fiber placement and correct viral expression were included in the analysis. Analysis of the signal was performed with MATLAB codes provided by Thinker Tech. The calcium signals (ΔF/F) were averaged across recording sites (N = 4 sites per group) for analysis. Calcium activity AUC was compared across the groups.

Chemogenetic manipulation

A stock solution of clozapine N-oxide (CNO, BrainVTA, Wuhan, China) was prepared at 20 mg/mL in DMSO. It was diluted in 0.9% saline to 0.3 mg/mL prior to administration. A dose of 3 mg/kg was administered intraperitoneally. The timing and interval of CNO administration relative to those of the behavioral tests are depicted in the figure.

Statistical analysis

Statistical analyses were conducted with GraphPad Prism (Version 9, GraphPad Software, Inc.). We assessed the normality by the Shapiro-Wilk test and QQ plots. An unpaired t-test and the Mann-Whitney U test were used for comparisons between two groups. A repeated measures two-way ANOVA followed by a Sidak’s multiple comparisons test was performed for comparisons of two groups across multiple time points. The same method was used to compare preCNO and postCNO mechanical thresholds in the two groups. A two-way ANOVA followed by a Sidak’s multiple comparisons test was performed for comparisons of two groups across multiple brain regions. We defined a p value of < 0.05 as indicating statistical significance (two-tailed)

Results

Chronic migraine model mice induced by NTG exhibited mechanical hyperalgesia and anxiety-like behaviors

For the investigation of mechanical hyperalgesia and its associated negative affect in NTG-induced chronic migraine, male C57BL/6J mice treated with NTG or vehicle (VEH) for a total of five doses were evaluated for generalized hyperalgesia, anxiety-like and depression-like behaviors (Fig. 1A). As shown in Fig. 1B, the model mice developed sustained paw and periorbital mechanical hyperalgesia. The model mice also exhibited anxiety-like behavior, as evidenced by reduced entrance into the center (OFT, Fig. 1C), time in the open arms, and entrance into the open arms (EPM, Fig. 1D). However, no significant depression-like behaviors were observed, as no difference in immobility time was detected between groups during the TST (Fig. 1E).

Fig. 1.

pPVT is activated in the NTG-induced chronic migraine model A Procedure for model validation. Following random allocation, we administered five intraperitoneal injections of 10 mg/kg NTG to the NTG group and vehicle to the VEH group. The NTG or VEH was administered every other day. Von Frey tests of the hind paw and periorbital area were conducted prior to each injection. OFT, EPM, and TST were conducted on days 9 and 10. The von Frey test and TST were performed on the same cohort of mice, whereas the OFT and EPM were assessed in a separate cohort. B Basal mechanical threshold of the right hind paw and periorbital skin throughout modeling. The group sizes were N = 7 for the VEH group and N = 8 for the NTG group, respectively. (one mouse was excluded from the VEH group because of postintraperitoneal injection bleeding). Differences were analyzed by repeated measures two-way ANOVA. For the paw mechanical threshold, F = 4.299, p = 0.0044 for the time × group interaction, F = 83.52, p < 0.0001 for the group factor, F = 10.99, p = 0.0002 for the time factor. Significant differences between the two groups were revealed by Sidak’s test on day 2 (adjusted p = 0.0066), day 4 (adjusted p = 0.0028), day 6 (adjusted p < 0.0001) and day 8 (adjusted p = 0.0009). For the periorbital mechanical threshold, F = 3.280, p = 0.0186 for the time × group interaction, F = 35.18, p < 0.0001 for the group factor, F = 1.729, p = 0.2064 for the time factor. Sidak’s multiple comparisons test revealed significant differences between the two groups on day 2 (adjusted p = 0.0002), day 4 (adjusted p = 0.0297), day 6 (adjusted p = 0.0111) and day 8 (adjusted p = 0.0093). The asterisks marked in the two graphs above represent the differences between the two groups at the same time point, as calculated by Sidak’s test. C Comparison of time spent in the center (upper graph) and entrance into center (lower graph) in the 15-minute OFT between the NTG group and VEH group. N = 8 mice per group. Differences were analyzed by an unpaired t test. t(14) = 0.3194, p = 0.7541 for time in the center. t(14) = 2.867, p = 0.0124 for entrance into center. D Comparison of time spent in the open arms (upper graph) and entrance into open arms (lower graph) in the 15-minute EPM test between the NTG group and VEH group. N = 8 mice per group. Differences was analyzed by an unpaired t test. t(14) = 2.157, p = 0.0489 for time in the open arms. t(14) = 0.036, p = 0.0089 for entrance into open arms. E Comparison of immobility time between the NTG group and the VEH group in the 5-minute TST. N = 7 mice for the VEH group and n = 8 for the NTG group (one mouse was excluded from the VEH group because of postintraperitoneal injection bleeding). Differences were analyzed by an unpaired t test. t(13) = 0.9758, p = 0.347. F Experimental procedure for assessing neuronal activation in the pPVT following model establishment. On day 8 of modeling, Fos-tdTomato mice received a 4-OHT (50 mg/kg, i.p.) injection 0.5 hours prior to NTG/VEH. The mice were sacrificed for brain tissue collection 3 days later. G Images showing Fos-tdTomato (red) and DAPI (blue) in the pPVT at different anterior-posterior (AP) coordinates (relative to bregma) from the NTG and VEH groups. Scale bar, 100 µm for AP-1.43, 200 µm for AP-1.55 and AP-1.91. H Comparison of Fos-tdTomato number in the pPVT between the two groups. N = 5 mice per group. The values for each mouse were the sum of 6 brain sections at fixed intervals along the anterior-posterior axis from the entire pPVT. Difference was analyzed by an unpaired t test. t(8) = 3.181, p = 0.0130. Data are expressed as mean ± SEM, unless indicated otherwise. Statistical significance is denoted by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001)

pPVT glutamatergic neurons regulate hyperalgesia in mice with chronic migraine

We first compared the activation of the pPVT between the model and control groups, and then observe the consequent behavioral changes after manipulating the activity of pPVT glutamatergic neurons.

To prevent the baseline activity of the pPVT under general physiological conditions from obscuring the differences between groups, we used the FosTRAP strategy to compare pPVT activation between the NTG and VEH groups. Fos-tdTomato mice generated by intercrossing homozygous Ai9 mice with homozygous FosTRAP2 mice were subjected to a chronic migraine model. On day 8 of modeling, the mice received a 4-OHT (50 mg/kg, i.p.) injection 0.5 hours prior to NTG/VEH, and the activated neurons were labeled with red fluorescent protein. Brain tissues were collected after a 3-day period to allow the accumulation of tdTomato to ensure reliable identification (Fig. 1F). The NTG group presented a greater number of 4-OHT-induced tdTomato-positive pPVT neurons, suggesting increased neuronal activation (Fig. 1G-H).

It has been reported that the pPVT is largely composed of Vglut2+ excitatory glutamatergic neurons and is devoid of GABAergic inhibitory neurons [5, 6]. We targeted these neurons using Vglut2-Cre mice. To precisely quantify the proportion of Vglut2+ neurons within the pPVT, we used an intravenously delivered blood-brain barrier-crossing virus (AAV2/PHP.eB-EF1α-DIO-H2B-EGFP) to label the nuclei of Vglut2+ neurons with a fluorescent marker (Fig. 2A), which more effectively reveals Vglut2+ neurons within the pPVT brain region than in the offspring of Vglut2-Cre mice and Ai9 mice (Fig. S1). We found that Vglut2+ neurons accounted for 85.8% of all neurons in the pPVT (Fig. 2B–C).

Fig. 2.

Inhibition of pPVT glutamatergic neurons alleviates mechanical hyperalgesia in model mice A Procedure for labeling pPVT glutamatergic neurons. AAV2/PHP.eB-EF1α-DIO-H2B-EGFP was administered to Vglut2-Cre mice via tail vein injection. Brain tissues were subjected to immunofluorescence staining for NeuN three weeks later. B Images showing Vglut2-H2B-EGFP (green) and NeuN (red) in the pPVT. Scale bar, 100 µm. C Pie chart showing the proportion of Vglut2+ neurons in the pPVT. The data are from the sum of 5 brain sections across the anteroposterior extent of the pPVT of one mouse. D Procedure for chemogenetic inhibition of pPVT glutamatergic neurons. AAV2/9-EF1a-DIO-hM4D(Gi)-EGFP-WPREs or the corresponding control virus were stereotaxically injected into the pPVT of Vglut2-Cre mice. Three weeks later, the mice were subjected to the chronic migraine model. On day 9, CNO was administered intraperitoneally (i.p.) at a dosage of 3 mg/kg, and preCNO and 2–2.5 h postCNO von Frey test results were measured. On days 10 and 11, OFT and EPM were performed 2 hours after CNO administration. E Validation of viral efficacy and injection site accuracy. Images show the expression of EGFP(green) and cFos (red) in the pPVT in the NTG+EGFP+CNO group and NTG+hM4Di+CNO group. Scale bar, 100 µm. F Quantification of viral efficacy. Comparison of the number of cFos+ EGFP+ neurons between the two groups. N = 3 mice per group. Each data point represents the sum of 5 brain sections through the anteroposterior extent of the pPVT for each individual mouse. The difference was analyzed by unpaired t test. t(4) = 6.669, p = 0.0026 G Comparison of preCNO and postCNO paw mechanical thresholds (upper graph) and periorbital mechanical thresholds (lower graph) between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N =  10 mice per group (only mice with the intended viral expression were included in the analysis). The data were analyzed by repeated measures two-way ANOVA. For the paw mechanical threshold, F = 27.98, p < 0.0001 for the time × group interaction, F = 7.42, p = 0.0139 for the group factor, F = 35.88, p < 0.0001 for the time factor. Sidak’s test revealed a significant difference in the postCNO mechanical thresholds between the NTG+hM4Di+CNO group and the NTG+EGFP+CNO group. (adjusted p < 0.0001), but no significant difference in the preCNO mechanical thresholds (adjusted p = 0.7736) For the periorbital mechanical threshold, F=9.232, p=0.0071 for the time×group interaction, F=8.412, p=0.0095 for the group factor, F=7.827, p=0.0119 for the time factor. Sidak’s test revealed a significant difference in the postCNO mechanical thresholds between the NTG+hM4Di+CNO group and the NTG+EGFP+CNO group. (adjusted p=0.0004), but no significant difference in the preCNO mechanical thresholds (adjusted p=0.9430). The asterisks marked in the two graphs above represent the differences at the same time point (preCNO or postCNO) between the different groups. H Comparison of time spent in center (upper graph) and entrance into center (lower graph) in the OFT between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N=10 mice per group (only mice with the intended viral expression were included in the analysis). For time in the center, the difference was analyzed by the Mann-Whitney test (U=49, p=0.9705). Data are presented as the median±IQR. For the entrance into center, the difference was analyzed by an unpaired t test, t(18)=0.0595, p=0.9532. I Comparison of time spent in the open arms (upper graph) and the entrance into open arms (lower graph) in the EPM test between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N=10 mice (only mice with the intended viral expression were included in the analysis). Differences were analyzed by an unpaired t test. t(18)=0.6415, p=0.5293 for time in the open arms. t(18)=0.7828, p=0.4439 for entrance into the open arms. Data are expressed as mean±SEM, unless indicated otherwise. Statistical significance is denoted by asterisks (*p<0.05, **p<0.01, ***p<0.001)

To manipulate pPVT glutamatergic neurons, we injected a chemogenetic inhibitory virus (AAV2/9-EF1a-DIO-hM4D(Gi)-EGFP-WPREs) into the pPVT of Vglut2-Cre mice. The mice were subjected to the chronic migraine model three weeks later. Next, we tested mechanical hyperalgesia and anxiety-like behavior after the administration of CNO (to activate hM4D (Gi) and suppress neuronal activity, 3 mg/kg, i.p.) on days 9, 10 and 11 (Fig. 2D). To prevent the confounding influence of behavioral testing itself on c-Fos expression, brain tissue was collected following CNO and NTG administration after a one-week washout period. Viral expression locations were examined across all brain sections, and only behavioral data from mice with the intended injection site were included in the final analysis. Furthermore, a subset of mice (randomly selected, N = 3) that had undergone behavioral tests was used for c-Fos immunofluorescence staining to validate the efficacy of the chemogenetic inhibition virus. Compared with the NTG+EGFP+CNO group, the NTG+hM4Di+CNO group showed a significant alleviation in paw and periorbital mechanical hyperalgesia following chemogenetic inhibition of pPVT Vglut2+ neurons (Fig. 2G). Compared with the NTG+EGFP+CNO group, the NTG+hM4Di+CNO group did not significantly differ in anxiety-like behaviors as evidenced by the OFT and EPM results (Fig. 2H–I). These results suggest that pPVT glutamatergic neurons regulate hyperalgesia in mice with chronic migraine, but do not affect anxiety-like behaviors.

Among several downstream targets of pPVT, the CeA is identified as the most promising candidate for regulating chronic migraine

To identify a functional circuit connected to the pPVT that regulates chronic migraine, we first employed anterograde tracing to identify brain regions that receive projections from pPVT glutamatergic neurons. We subsequently selected candidate downstream regions on the basis of the neuronal activation of the model group. We then employed retrograde tracing to confirm the existence of the target circuit.

We used two anterograde tracing strategies. First, we injected a nontranssynaptic anterograde viral tracer (AAV2/9-hSyn-DIO-mGFP-T2A-Synaptophysin-mRuby-WPRE-hGH polyA) into the pPVT of Vglut2-Cre mice (Fig. 3A). Three weeks later, we observed red fluorescent protein-labeled axonal terminals from pPVT Vglut2+ neurons in the nucleus accumbens (NAc), bed nucleus of the stria terminalis (BNST), interstitial nucleus of the posterior limb of the anterior commissure (IPAC), basolateral amygdala (BLA) and central amygdala (CeA) (Fig. 3B). Anterograde transsynaptic virus (AAV2/1-hSyn-CRE-WPRE-hGH polyA) was injected into the pPVT of Ai9 mice (Fig. 3C). We also observed red fluorescent protein-labeled postsynaptic neuronal somata in the aforementioned downstream regions (Fig. 3D).

Fig. 3.

The pPVT primarily projects to the NAc, BNST, IPAC, BLA and CeA. A Schematic of the anterograde nontranssynaptic tracing strategy. AAV2/9-hSyn-DIO-mGFP-T2A-Synaptophysin-mRuby-WPRE-hGH polyA was stereotaxically injected into the pPVT of Vglut2-Cre mice. Brain tissue was collected at 3 weeks post viral injection. B Images showing the injection site and brain regions with axonal terminals from Vglut2+ neurons of the pPVT. The first image from the left shows the expression of mGFP (green, somata morphology) in the pPVT. The second, third, and fourth images show the terminal projections in the NAc, IPAC/BNST, and CeA/BLA, respectively, visualized by mRuby (red, fibrous morphology). Scale bar, 200 µm for the pPVT and CeA/BLA, and 400 µm for the NAc and IPAC/BNST. C Schematic of the anterograde transsynaptic tracing strategy. High-titer (≥1.00E+13 vg/mL) AAV2/1-hSyn-CRE-WPRE-hGH polyA was injected into the pPVT of Ai9 mice. Brain tissue was collected at 3 weeks post viral injection. D Images showing the injection site and projection site. The first image from the left shows the expression of tdTomato (red, somata morphology) in the pPVT. The second, third, and fourth images show the expression of tdTomato (red, somata morphology) in the NAc, IPAC/BNST, and CeA/BLA, which was mediated by anterograde transsynaptic Cre recombinase from the pPVT. Scale bar, 200 µm for pPVT and CeA/BLA, 400 µm for NAc and IPAC/BNST

To assess neuronal activation in the NAc, BNST, IPAC, BLA, and CeA, we used the FosTRAP strategy to compare the NTG and VEH groups (Fig. 4A). In the CeA in the NTG group, we found a higher density of tdTomato+ neurons, suggesting increased neuronal activation, whereas the density of tdTomato+ neurons in the NAc, IPAC, BNST, and BLA were unchanged (Fig. 4B–C). We subsequently compared neuronal activation in three subregions of the CeA—the capsular part (CeC), lateral part (CeL), and medial part (CeM)—between the NTG and VEH groups. The NTG group exhibited an increasing trend in activated neurons in the CeC, CeL, and CeM, although only the increase in the CeC and CeL reached statistical significance (Fig. 4C). These results imply that the CeA is activated in chronic migraine model mice.

Fig. 4.

The CeA and its afferent input from the pPVT are activated in chronic migraine model mice A Procedure for assessing neuronal activation in the NAc, BNST, IPAC, BLA and CeA following model establishment. On day 8 of modeling, the Fos-tdTomato mice received a 4-OHT (50 mg/kg, i.p.) injection 0.5 hours prior to NTG/VEH. Brain tissue was collected 3 days later. B Images showing Fos-tdTomato (red) in the NAc, BNST, IPAC, BLA and CeA from the NTG and VEH groups. The CeA is shown at two different anterior-posterior (AP) coordinates relative to the bregma (AP-1.43 and AP-1.79). Scale bar, 400 µm for the NAc and IPAC/BNST and 200 µm for the CeA/BLA. C The upper graph shows the difference in the number of Fos-tdTomato per mm² in the NAc, BNST, IPAC, BLA and CeA between the NTG and VEH groups. N = 5 mice per group. The data for each mouse is derived by dividing the total count of Fos-tdTomato-positive cells from bilateral target brain region by the corresponding total area of those regions. Brain sections were collected at consistent intervals along the anterior-posterior axis of the target brain region (4 sections for NAc and BNST, 5 sections for IPAC, 6 sections for BLA and CeA). The data were analyzed by two-way ANOVA and Sidak’s test. A significant difference was revealed between the two groups at the CeA (adjusted p = 0.0143), but no significant difference was found at NAc (adjusted p = 0.8935), IPAC (adjusted p = 0.9991), BNST (adjusted p = 0.8868) or BLA (adjusted p = 0.3086). The lower graph shows the difference in the number of Fos-tdTomato per mm² in the three subregions of the CeA (CeC, CeL, and CeM) between the NTG and VEH groups. N = 5 mice per group. The data were analyzed by two-way ANOVA and Sidak’s test. A significant difference was revealed between the two groups at the CeC (adjusted p = 0.0211), and CeL (adjusted p = 0.0274), but no significant difference was found at CeM (adjusted p = 0.3382). D Schematic of the retrograde tracing strategy from CeA. AAV2/11-EF1α-DIO-EGFP was unilaterally injected into the CeA of Vglut2-Cre mice. Brain tissue was collected at 3 weeks post viral injection. E Images showing EGFP (green) expressed retrogradely from brain CeA in the pPVT at AP-0.95, AP-1.23, AP-1.55 and AP-1.79. Scale bar, 200 µm. F Procedure for tracing activated pPVT neurons and their projections to the CeA in the NTG and VEH groups. AAV2/9-hSyn-DIO-H2B-EGFP-T2A-Synaptophysin-mRuby-WPREs were injected into the pPVT of FosTRAP2 mice. Three weeks later, the mice were subjected to five NTG (i.P., 10 mg/kg, every other day) or VEH injections. On day 8 of modeling, the mice received a 4-OHT (50 mg/kg, i.p.) injection 0.5 hours prior to NTG/VEH. Brain tissue was collected 3 weeks later. G Images showing EGFP (green, somata) indicating the somata of activated pPVT neurons (first column from left) and mRuby (red) indicating the fiber terminals of activated pPVT neurons in CeA (the second, third and fourth columns from left) in the NTG and VEH groups. Scale bar, 200 µm. The images in the fifth column corresponds to a magnified view of the area highlighted by the box in the fourth column, scale bar, 50 µm H Quantification of the fluorescent signal intensity of mRuby-tagged fiber terminals in the CeA (left graph) and its subregions (right graph). N = 4 mice per group. The values for each mouse are the sum of the bilateral CeA from 5 brain sections at fixed intervals along the anterior-posterior axis from the entire CeA. For the left graph, the difference was analyzed by an unpaired t test. t(6) = 2.365, p = 0.0559. For the right graph, the difference was analyzed by two-way ANOVA. A significant difference was revealed between the two groups at the CeC (adjusted p = 0.0399) and CeL (adjusted p = 0.0373), but no significant difference was found at CeM (adjusted p = 0.4408).Data are expressed as mean ± SEM, unless indicated otherwise. Statistical significance is denoted by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001)

We employed retrograde tracing to further confirm the existence of the pPVT-CeA circuit. A retrograde viral tracer (AAV2/11-EF1α-DIO-EGFP) was unilaterally injected into the CeA of Vglut2-Ce mice (Fig. 4D). Three weeks later, we observed green fluorescent protein in the ipsilateral pPVT, which labeled Vglut2+ neurons retrogradely labeled from the CeA (Fig. 4E).

Additionally, to determine whether CeA afferent input from the pPVT was activated in the model group, we traced activated pPVT neurons and their projections in the CeA using FosTRAP2 mice. We injected an anterograde tracing virus (AAV2/9-hSyn-DIO-H2B-EGFP-T2A-Synaptophysin-mRuby-WPREs) into the pPVT of FosTRAP2 mice. The mice were subjected to a chronic migraine model and received a 4-OHT injection 0.5 hours prior to NTG/VEH on day 8 of modeling (Fig. 4F). Three weeks later, we observed more EGFP-labeled somata (indicating activated neurons) within the pPVT and more mRuby-labeled fiber terminals within the CeC and CeL in the NTG group (Fig. 4G–H). These findings suggest that the CeA’s afferent input from the pPVT is activated in the model group.

Overall, we identified the CeA and pPVT-CeA circuit as a promising target for regulating chronic migraine. Our subsequent studies will validate the role of this circuit.

CeA GABAergic neurons regulate both hyperalgesia and anxiety-like behaviors

We next further investigated the role of the CeA by manipulating the neural activity of the CeA and observing subsequent behavioral changes. The vast majority (90~95%) of CeA neurons are GABAergic [23, 24]. We injected a chemogenetic inhibitory virus (AAV2/9-EF1a-DIO-hM4D(Gi)-EGFP-WPREs) into the CeA of VGAT-Cre mice (Fig. 5A). Using the same procedure as described previously in Fig. 2D, we first measured the mechanical threshold in the NTG-induced chronic migraine model mice on day 9 (pre- and post-CNO) (Fig. 5D) and then performed OFT and EPM on days 10 and 11, respectively, following CNO administration (Figs. 5E and 5F). For efficacy and injection site validation of chemogenetic virus, a one-week washout period after behavioral testing was followed by brain tissue collection. Viral expression locations were examined across all brain sections, and only behavioral data from mice with the intended injection site were included in the final analysis. Furthermore, a subset of mice (randomly selected, n = 3) that had undergone behavioral tests was used for c-Fos immunofluorescence staining to validate the efficacy of the chemogenetic inhibition virus (Fig. 5B, C). Compared with the NTG+EGFP+CNO group, the NTG+hM4Di+CNO group showed a significant alleviation in paw and periorbital mechanical hyperalgesia following chemogenetic inhibition of CeA GABAergic neurons in the model mice (Fig. 5D). The NTG+hM4Di+CNO group exhibited alleviated anxiety-like behavior compared to the other group, as evidenced by increased time in the center, entrance into the center (OFT, Fig. 5E) and time in the open arms (EPM, Fig. 5F). We also observed a nonsignificant increase in entrance into open arms in NTG+hM4Di+CNO group (Fig. 5F). CeA GABAergic neurons regulate both hyperalgesia and anxiety-like behaviors.

Fig. 5.

Inhibition of CeA GABAergic neurons alleviated mechanical hyperalgesia and anxiety-like behaviors in model mice A Schematic of the chemogenetic inhibition of CeA GABAergic neurons. AAV2/9-EF1a-DIO-hM4D(Gi)-EGFP-WPREs or the corresponding control virus were stereotaxically injected into the CeA of VGAT-Cre mice. The subsequent procedure is the same as in that Fig. 2D. B Validation of viral efficacy. Images show the expression of EGFP(green) and cFos (red) in the CeA from the NTG+EGFP+CNO group and NTG+hM4Di+CNO group. Scale bar, 200 µm. C Quantification of viral efficacy. Comparison of the number of cFos+EGFP+ neurons between the two groups. N = 3 mice per group. Each data point represents the sum of 5 brain sections through the anteroposterior extent of the CeA for each individual mouse. The difference was analyzed by unpaired t test. t(4) = 6.106, p = 0.0036. D Comparison of preCNO and postCNO paw mechanical thresholds (upper graph) and periorbital mechanical thresholds (lower graph) between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 8 mice per group (only mice with the intended viral expression were included in the analysis). The data were analyzed by repeated measures two-way ANOVA. For the paw mechanical threshold, F = 12.25, p = 0.0035 for the time × group interaction, F = 3.563, p = 0.0800 for the group factor, F = 7.685, p = 0.0150 for the time factor. Sidak’s test revealed a significant difference in the postCNO mechanical thresholds between the NTG+hM4Di+CNO group and the NTG+EGFP+CNO group.(adjusted p = 0.0056), but no significant difference in the preCNO mechanical thresholds (adjusted p = 0.9963). For the periorbital mechanical threshold, F = 13.89, p = 0.0023 for the time × group interaction, F = 5.705, p = 0.0316 for the group factor, F = 13.89, p = 0.0023 for the time factor. Sidak’s test revealed a significant difference in the postCNO mechanical thresholds between the NTG+hM4Di+CNO group and the NTG+EGFP+CNO group (adjusted p = 0.0006), but no significant difference in the preCNO mechanical thresholds (adjusted p = 0.9367). The asterisks marked in the two graphs above represent the differences at the same time point (preCNO or postCNO) between the different groups. E Comparison of time spent in the center (upper graph) and entrance into the center (lower graph) in the OFT between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 8 mice per group (only mice with the intended viral expression were included in the analysis). Difference were analyzed by an unpaired t test. t(14) = 2.217, p = 0.0437 for time in the center. t(14) = 2.857, p = 0.0127 for entrance into the center. F Comparison of time spent in the open arms (upper graph) and entrance into the open arms (lower graph) in EPM between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 8 mice per group (only mice with the intended viral expression were included in the analysis). Difference were analyzed by an unpaired t test. t(14) = 2.154, p = 0.0492 for time in the open arms. t(14) = 1.231, p = 0.2386 for entrance into the open arms. Data are expressed as mean ± SEM, unless indicated otherwise. Statistical significance is denoted by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001)

The functional connectivity between the pPVT and CeA is predominantly excitatory

Given that the CeA contains local inhibitory interneurons [25], it is difficult to predict whether activation of the pPVT would increase or decrease overall neuronal activation in the CeA. To investigate the functional connectivity between the pPVT and the CeA, we injected an optogenetic activatory virus (AAV2/9-CaMKIIa-Chrimson-mCherry-WPRE-hGH polyA) or a corresponding control virus (AAV2/9-CaMKIIa-mCherry-WPRE-hGH polyA) into the pPVT and injected a calcium indicator virus (AAV2/9-EF1α-DIO-GCaMP6s) into the CeA of VGAT-Cre mice. An optical fiber in CeA was used to optogenetically activate the fiber terminals of CeA-projecting CaMKIIα+ neurons in the pPVT and to record the calcium activity of CeA GABAergic neuron (Fig. 6A–B). We found that activation of the pPVT increased the overall neuronal activation of CeA GABAergic neurons, as evidenced by increased ΔF/F in response to optogenetic stimulation (Fig. 6C–D). Compared to the mCherry+GCaMP6s group, the AUC of calcium signals in the Chrimson+GCaMP6s group was significantly increased (Fig. 6E). These findings suggest that the functional connectivity between the pPVT and CeA is predominantly excitatory.

Fig. 6.

Activation of CeA-projecting glutamatergic neurons in the pPVT (PVT^glu- CeA) drives activity in CeA GABAergic neurons A Schematic of the recording of the activity of CeA GABAergic neurons using fiber photometry during the optogenetic activation of CeA-projecting glutamatergic neurons in the pPVT (PVT^glu- CeA). AAV2/9-CaMKIIa-Chrimson-mCherry-WPRE-hGH polyA or AAV2/9-CaMKIIa-mCherry-WPRE-hGH polyA was injected into the pPVT. Injection of AAV2/9-EF1α-DIO-GCaMP6s and implantation of a ferrule fiber were performed in the CeA of VGAT-Cre mice. Optogenetic stimulation and fiber photometry were performed 3 weeks later. B Validation of viral efficacy and ferrule fiber implantation of mCherry+GCaMP6s and Chrimson+GCaMP6s groups. The images in the first column show mCherry expression (red, punctate, perinuclear) in the pPVT. Scale bar, 200 µm. The images in the second column show GCaMP6s expression in the CeA (green, somata morphology) and ferrule fiber track (white arrow). Scale bar, 1000 µm the magnified images in the third column show GCaMP6s expression (green, somata morphology, white arrow) and terminal projections from the pPVT visualized by mCherry (red, punctate and fibrous morphology, white arrowhead) in the CeA. Scale bar, 200 µm. C Heatmap of calcium activity of CeA GABAergic neurons following optic stimulation of the PVT^glu- CeA circuit. The data show five trials of optogenetic stimulation recorded with a 5-minute interstimulus interval from one site. The thick red line represents optogenetic stimulation (635 nm, 20 Hz, 5 s). D Average fiber photometry (Δ F/F %) trace of CeA GABAergic neurons of mCherry+GCaMP6s and Chrimson+GCaMP6s groups. The black line and shaded black area represents the mean ±SEM ΔF/F response from the mCherry+gcamp6s group (N = 20 trials from 4 sites). The red line and shaded red area represents the mean ±SEM ΔF/F response from the Chrimson+GCaMP6s group (N = 20 trials from 4 sites). The thick red line represents optogenetic stimulation (635 nm, 20 Hz, 5 s). E Comparison of calcium activity AUC in CeA GABAergic neurons following optic stimulation of the PVT^glu- CeA circuit. N = 4 sites per group (only sites with the intended viral expression were included in the analysis). Difference were analyzed by an unpaired t test. t(6) = 2.545, p = 0.0438. Data are expressed as mean ± SEM. Statistical significance is denoted by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001)

The pPVT^glu-CeA^GABA circuit regulates mechanical hyperalgesia in mice with chronic migraine

We next manipulated the pPVT^glu-CeA^GABA circuit and observed the subsequent behavioral changes. Chemogenetic manipulation was performed in two parts: one targeting the presynaptic neurons (CeA-projecting glutamatergic neurons in the pPVT, pPVT^glu- CeA) and the other targeting the postsynaptic neurons (CeA GABAergic neurons innervated by the pPVT, pPVT- CeA^GABA).

To chemogenetically inhibit pPVT^glu- CeA, we injected AAV2/9-CaMKIIα-DIO-hM4D (Gi)-P2A-EGFP into the pPVT of C57BL/6J mice. This vector was paired with a retrograde virus (AAV2/11-hSyn-SV40 NLS-Cre) injected into the bilateral CeA (Fig. 7A). Using the same procedure as described previously in Fig. 2D, we first measured the mechanical threshold in the NTG-induced chronic migraine model mice on day 9 (pre- and post-CNO) (Fig. 7D) and then performed OFT and EPM on days 10 and 11, respectively, following CNO administration (Figs. 7E and 7F). For efficacy and injection site validation of chemogenetic virus, a one-week washout period after behavioral testing was followed by brain tissue collection. Viral expression locations were examined across all brain sections, and only behavioral data from mice with the intended injection site were included in the final analysis. Furthermore, a subset of mice (randomly selected, n = 3) that had undergone behavioral tests was used for c-Fos immunofluorescence staining to validate the efficacy of the chemogenetic inhibition virus (Figs. 7B and 7C). Compared with the NTG+EGFP+CNO group, the NTG+hM4Di+CNO group showed a significant alleviation in paw and periorbital mechanical hyperalgesia following chemogenetic inhibition of pPVT^CaMKIIα- CeA neurons in the model mice (Fig. 7D). In terms of anxiety-like behaviors, the NTG+hM4Di+CNO group did not exhibit significant changes in anxiety-like behaviors compared with the NTG+EGFP+CNO group (Fig. 7E–F). The presynaptic neurons of the pPVT^glu-CeA^GABA circuit regulate mechanical hyperalgesia in chronic migraine mice.

Fig. 7.

Inhibition of CeA-projecting glutamatergic neurons in the pPVT (PVT^glu-CeA) and CeA GABAergic neurons innervated by the pPVT (pPVT-CeA^GABA) alleviated mechanical hyperalgesia in model mice A Schematic of chemogenetic inhibition of CeA-projecting glutamatergic neurons in the pPVT. (PVT^glu-CeA). AAV2/9-CaMKIIα-DIO-hM4D (Gi)-P2A-EGFP or the corresponding control virus was stereotaxically injected into the pPVT of C57BL/6J mice. In combination, retrograde AAV2/11-hSyn-SV40 NLS-Cre was bilaterally injected into the CeA. The subsequent procedure is the same as that in Fig. 2D. B Validation of viral efficacy and injection site accuracy. The first column (from left) shows the representative images of EGFP (green, fibrous morphology) indicating the fiber terminals of PVT^CaMKIIα-CeA neurons in the CeA. The second, third and fourth columns (from left) show representative images of EGFP (green) and cFos (red) in PVT^CaMKIIα-CeA neurons within the pPVT in NTG+EGFP+CNO group and NTG+hM4Di+CNO group. Scale bar, 200 µm at the CeA, and 100 µm at the pPVT. C Quantification of viral efficacy. Comparison of the number of cFos+EGFP+ neurons between the two groups. N = 3 mice per group. Each data point represents the sum of 5 brain sections through the anteroposterior extent of the pPVT for each individual mouse. The difference was analyzed by unpaired t test. t(4) = 3.190, p = 0.0332. D Comparison of preCNO and postCNO paw mechanical thresholds (upper graph) and periorbital mechanical thresholds (lower graph) between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 7 mice per group (only mice with the intended viral expression were included in the analysis). The data were analyzed by repeated measures two-way ANOVA. For the paw mechanical threshold, F = 48.12, p < 0.0001 for the time × group interaction, F = 49.79, p < 0.0001 for the group factor, and F = 56.26, p < 0.0001 for the time factor. Sidak’s test revealed a significant difference in the postCNO mechanical thresholds between the NTG+hM4Di+CNO group and NTG+EGFP+CNO group (adjusted p < 0.0001), but no significant difference in the preCNO mechanical thresholds (adjusted p = 0.8679). For the periorbital mechanical threshold, F = 8.511, p = 0.0129 for the time × group interaction, F = 1.125, p = 0.3098 for the group factor, F = 6.894, p = 0.0222 for the time factor. Sidak’s test revealed a significant difference in the postCNO mechanical thresholds between the NTG+hM4Di+CNO group and the NTG+EGFP+CNO group (adjusted p = 0.0256), but no significant difference in the preCNO mechanical thresholds (adjusted p = 0.5080). The asterisks marked in the two graphs above represent the differences at the time point (preCNO or postCNO) between different groups. E comparison of time spent in the center (upper graph) and entrance into the center (lower graph) in the OFT between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 7 mice per group (only mice with the intended viral expression were included in the analysis). Differences were analyzed by an unpaired t test. t(12) = 0.1447, p = 0.8873 for time in center. t(12) = 0.9111 p = 03802 for entrance into center. F Comparison of time spent in the open arms (upper graph) and entrance into the open arms (lower graph) in the EPM between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 7 mice (only mice with the intended viral expression were included in the analysis). The difference in time in open arms was analyzed by Mann-Whitney test (U = 12, p = 0.1282), and the data are presented as the median ± IQR. For entrance into open arms, the difference was analyzed with an unpaired t test. t(12) = 1.922, p = 0.0787. G Schematic of the chemogenetic inhibition of CeA GABAergic neurons innervated by the pPVT (pPVT-CeA^GABA). AAV2/9-hSyn-ConFon-hM4D(Gi)-EGFP-WPRE-hGH polyA or the corresponding control virus was bilaterally injected into the CeA of VGAT-Cre mice. In combination, high-titer (≥1.00E+13 vg/mL) transsynaptic AAV2/1-hSyn-SV40 NLS-Flp-P2A-mCherry was injected into the pPVT. The subsequent procedure is the same as that in Fig. 2D. H Validation of viral efficacy and injection site accuracy. The first column (from left) shows the representative images of mCherry (red) indicating the injection site in the pPVT. The second, third and fourth columns (from left) show the representative images of EGFP (green) and cFos (red) in CeA GABAergic neurons innervated by the pPVT in the NTG+EGFP+CNO group and NTG+hM4Di+CNO group. Scale bar, 200 µm. I Quantification of viral efficacy. Comparison of the number of cFos+EGFP+ neurons between the two groups. N = 3 mice per group. Each data point represents the sum of 5 brain sections through the anteroposterior extent of the CeA for each individual mouse. The difference was analyzed by an unpaired t test. t(4) = 10.02, p = 0.0006. J Comparison of preCNO and postCNO paw mechanical thresholds (upper graph) and periorbital mechanical thresholds (lower graph) between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 8 mice per group (only mice with the intended viral expression were included in the analysis). The data were analyzed by repeated measures two-way ANOVA. For the paw mechanical threshold, F = 14.36, p = 0.0020 for the time × group interaction, F = 4.847, p = 0.0450 for the group factor, F = 8.927, p = 0.0098 for the time factor. Sidak’s test revealed a significant difference in the postCNO mechanical thresholds between the NTG+hM4Di+CNO group and NTG+EGFP+CNO group (adjusted p = 0.0011), but no significant difference in the preCNO mechanical thresholds (adjusted p = 0.6987). For the periorbital mechanical threshold, F = 6.821, p = 0.0205 for the time × group interaction, F = 2.566, p = 0.1315 for the group factor, and F = 6.108, p = 0.0269 for the time factor. Sidak’s test revealed a significant difference in the postCNO mechanical thresholds between the NTG+hM4Di+CNO group and NTG+EGFP+CNO group (adjusted p = 0.0232), whereas no significant difference was detected in the preCNO mechanical thresholds (adjusted p = 0.9969. The asterisks in the two graphs above represent the differences at the same time point (preCNO or postCNO) between different groups. K Comparison of time spent in the center (upper graph) and entrance into the center (lower graph) in the OFT between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 8 mice per group (only mice with the intended viral expression were included in the analysis). Differences were analyzed by an unpaired t test. t(14) = 1.062, p = 0.3064 for time in the center. t(14) = 0.6734, p = 0.5116 for entrance into center. L Comparison of time spent in the open arms (upper graph) and entrance into the open arms (lower graph) in the EPM between the NTG+EGFP+CNO group and the NTG+hM4Di+CNO group. N = 8 mice per group (only mice with the intended viral expression were included in the analysis). The difference in time spent in open arms was analyzed by an unpaired t test. t(14) = 1.2245, p = 0.8256. For entrance into the open arms, the difference was analyzed by Mann-Whitney test (U = 27, p = 0.6211) and the data are presented as the median ± IQR. Data are expressed as mean ± SEM, unless indicated otherwise. Statistical significance is denoted by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001)

To chemogenetically inhibit pPVT-CeA^GABA, we injected a Cre- and Flp-dependent virus (AAV2/9-hSyn-ConFon-hM4D(Gi)-EGFP-WPRE-hGH polyA) into the CeA of VGAT-Cre mice, which was used in combination with an anterograde transsynaptic virus (AAV2/1-hSyn-SV40 NLS-Flp-P2A-mCherry) injected into pPVT (Fig. 7G). Using the same procedure as described previously in Fig. 2D, we first measured the mechanical threshold in the NTG-induced chronic migraine model mice on day 9 (pre- and post-CNO) (Fig. 7J) and then performed OFT and EPM on days 10 and 11, respectively, following CNO administration (Figs. 7K and 7L). For efficacy and injection site validation of chemogenetic virus, a one-week washout period after behavioral testing was followed by brain tissue collection. Viral expression locations were examined across all brain sections, and only behavioral data from mice with the intended injection site were included in the final analysis. Furthermore, a subset of mice (randomly selected, n = 3) that had undergone behavioral tests was used for c-Fos immunofluorescence staining to validate the efficacy of the the chemogenetic inhibition virus (Figs. 7H and 7I). Compared with the NTG+EGFP+CNO group, the NTG+hM4Di+CNO group showed a significant alleviation in paw and periorbital mechanical hyperalgesia following chemogenetic inhibition of pPVT-CeA^GABA neurons in the model mice (Fig. 7J). In terms of anxiety-like behaviors, the NTG+hM4Di+CNO group did not exhibit significant changes in anxiety-like behaviors compared to the other group (Fig. 7K–L). The postsynaptic neurons of the pPVT^glu-CeA^GABA circuit regulate mechanical hyperalgesia in chronic migraine mice.

The above results suggest that the pPVT^glu-CeA^GABA circuit regulates mechanical hyperalgesia in chronic migraine mice, but do not regulate anxiety-like behaviors.

Discussion

We investigated mechanical hyperalgesia and anxiety-like behaviors in an NTG-induced chronic migraine model. The pPVT^glu-CeA^GABA circuit regulates mechanical hyperalgesia rather than anxiety-like behaviors. However, CeA GABAergic neurons as a whole regulate both mechanical hyperalgesia and anxiety-like behaviors.

The NTG-induced chronic migraine mouse model is a widely recognized and reliable experimental model of migraine [26]. Consistent with previous studies [27–29], we validated the presence of hyperalgesia and anxiety-like behaviors, but depression-like behaviors were not significant. We found that the pPVT is involved in mechanical hyperalgesia but not anxiety-like behavior in NTG-induced chronic migraine. The role of the PVT has been investigated in other types of pain models. Researchers observed an analgesic effect in both naïve mice and mice with CFA-induced inflammatory pain following inhibition of the PVT-NAc circuit [7]. In a neuropathic pain model, activation of the ZI^GABA(Zona Incerta)-PVT circuit (which is equivalent to inhibiting the PVT) alleviated pain [30]. Similarly, the inhibition or lesions of pPVT neurons alleviated mechanical hyperalgesia induced in an SNI model [10] and chronic constriction injury (CCI) model [11]. Collectively, in inflammation pain, neuropathic pain, and chronic migraine models, both mechanical hyperalgesia and pain-like behavior show a positive correlation with neuronal activity in the PVT.

In migraine neural circuit research, the FosTRAP strategy, owing to its temporal specificity in inducing Cre expression, is suitable for the initial screening of brain regions that may be involved in disease mechanisms—particularly those that are activated under general physiological conditions. FosTRAP2 mice have a TRAPing window of approximately 6 hours surrounding a 4-OHT injection [16]. Furthermore, when combined with Cre reporter mice (such as Ai9 or Ai14 mice), the fosTRAP can be used to map the activity patterns underlying behaviors in the brain [31, 32]. It can also be integrated with neuron activation markers labeled by immunofluorescence staining to observe neuronal activation responses to two sequentially identical or different stimuli [33] (the first stimulus labeled with the FosTRAP strategy and the second with immunofluorescence staining).

Through two anterograde tracing approaches, we found that neurons in the pPVT primarily project to five brain areas: the NAc, BNST, IPAC, CeA, and BLA. These findings are largely consistent with previously published literature [5]. Unlike commonly used anterograde nontranssynaptic viral tracers, the approach we selected could minimize the potential for misidentifying fiber pathways of passage as actual synaptic projection sites [32, 34]. We employed an anterograde viral tracer equipped with a presynaptic membrane-targeting element (synaptophysin) to drive the accumulation of fluorescent protein (mRuby) in presynaptic terminals. Additionally, we used the property of high-titer AAV1 serotype viral vectors to mediate the monosynaptic Cre transfer [35]. When combined with Cre reporter mice (Ai9 mice), this approach enabled the labeling of postsynaptic neuronal somata through fluorescent protein expression (tdTomato). These methods can serve as alternatives for researchers who are precluded from using the HSV-based anterograde monosynaptic tracing strategy owing to biosafety level constraints.

Among the four main brain regions that receive projections from the pPVT, we considered the pPVT-CeA circuit to be most likely involved in NTG-induced chronic migraine. This choice was motivated primarily by our experimental findings. Qualitative observations suggested that the activation of the dorsal part of the pPVT was more pronounced than that of the ventral part in the migraine group. Consistent retrograde tracing revealed that the neurons in the pPVT that innervate the CeA are predominantly distributed in the dorsal part of the pPVT. This anatomical congruence led us to target the pPVT-CeA circuit. Furthermore, the migraine model group exhibited increased activation in the CeA, along with a higher density of fiber terminals originating from activated neurons in the pPVT. This rationale was further supported by the literature. Evidence suggests that parabrachial nucleus^CGRP-CeA^{protein kinase c} contributes to the chronification of migraine [15]. Accordingly, we targeted this circuit for subsequent chemogenetic manipulation studies.

We observed a dissociation between mechanical hyperalgesia and anxiety-like behaviors following manipulation of the pPVT-CeA circuit. In mice with NTG-induced chronic migraine, both mechanical hyperalgesia and anxiety-like behavior were observed. However, the ameliorative effect was observed only in mechanical hyperalgesia when pPVT glutamatergic neurons, CeA projecting pPVT^glu neurons, and CeA^GABA neurons receiving inputs from the pPVT were chemogenetically inhibited. These findings suggest that the neural substrates underlying mechanical hyperalgesia are different from those underlying anxiety in chronic migraine mice. The pPVT^glu- CeA^GABA circuit is involved primarily in the regulation of mechanical hyperalgesia rather than anxiety-like behaviors. Similarly, a previous study [36] revealed that anxiety-like behaviors did not change after chemogenetic activation of the PVT-CeA circuit in the chronic restraint stress model. However, this circuit plays a distinct role in prenatal maternal stress-induced anxiety. A study [8] found that the PVT^glu-CeA^GABA circuit regulated the anxiety-like behaviors in offspring of prenatal maternal stress-exposed mice. Beyond anxiety, this circuit has also been investigated in the context of depression, another core negative emotional state. One study [37] demonstrated that suppression of the PVT ameliorated depression-like behaviors in the chronic unpredictable stress model, and that activation of the PVT-CeA induced depression-related behaviors under physiological conditions. Other research [38] has demonstrated that PVT^vglut2 neurons are involved in depression-like behaviors in Parkinson’s disease model mice. Taken together, it seems that the pPVT^glu- CeA^GABA circuit may contribute to emotional regulation in a model-specific manner.

Notably, the therapeutic effect of the inhibition of CeA^GABA neurons was observed in both mechanical hyperalgesia and anxiety-like behaviors. This result suggests that the neuronal population within the CeA is functionally heterogeneous and that GABAergic neurons in CeA which regulate anxiety-like behavior in migraine are not directly modulated by inputs from the pPVT. In addition to inputs from the pPVT, the CeA also receives projections from multiple other upstream brain regions. The BLA-CeA [39], PVN(paraventricular nucleus of the hypothalamus)-CeA [40], and VTA(ventral tegmental area)-CeA [41] circuits have each been implicated to regulate the anxiety-like behaviors in pain models. The individual or combined actions of these circuits above may be responsible for the anxiety-like behaviors in chronic migraine. Additionally, given the neuronal heterogeneity within CeA, a previous study [42] indicated that protein kinase c + neurons primarily modulate mechanical hyperalgesia but not anxiety, whereas the somatostatin+ subtype regulates anxiety without affecting pain processing. On the basis of these findings and our results, we speculate that the pPVT may regulate pain behaviors in model mice through the pPVT^glu-CeA^{protein kinase c} circuit, whereas somatostatin + neurons in the CeA regulate anxiety-related behaviors through their circuitry with other upstream brain regions.

Methodologically, we also identified an approach (AAV-mediated nucleus-localized fluorescent protein expression across the blood-brain barrier in Vglut2-Cre mice) for labeling Vglut2+ neurons in the PVT, which is superior to the most commonly used method (offspring of Vglut2-Cre mice and Ai9 mice). The difference in neuronal labeling efficacy primarily stems from the subcellular localization of the fluorescent protein. The pPVT is predominantly composed of Vglut2+ neuronal cell bodies and their axonal projections. When the protein is localized in the cytoplasm, the somatic signals overlap extensively with the fibrous axonal signals, severely compromising the resolution of individual target neurons. An approach using nucleus-localized fluorescent protein can effectively address this issue.

This study has several limitations. We conducted the experiments only on male mice. The choice of male mice was primarily intended to control for the potential confounding effects of the estrous cycle. However, sex differences are evident in both the clinical features and the pathophysiology of migraine [43, 44]. The generalization of the conclusions requires caution. Future studies should include female mice and incorporate their estrous cycle [45] into the experimental design. It should be acknowledged that the sample sizes common in behavioral studies, like ours, might be underpowered and thus potentially limit our ability to detect subtle effects. In this context, the lack of a statistically significant effect of modulating the specified pPVT-CeA neural circuit on anxiety-like behavior should consequently be interpreted with caution. While we performed chemogenetic inhibition of the pPVT-CeA circuit and observed consequent behavioral changes in the NTG-induced migraine group, we did not activate the circuit in naïve mice. This limitation prevents us from determining whether this neural circuit regulates pain and anxiety under physiological conditions. However, this study focused on investigating the mechanisms underlying migraine. The physiological state was not the focus of our research. Another limitation of this study is that our comparison of neural activation in the target brain region between the NTG and VEH groups relies on Fos-based histological evidence. We did not employ complementary methods such as fiber photometry or electrophysiology. A potential limitation of this study pertains to the heterogeneity within the pPVT and CeA. Based on distinct molecular markers, neurons in the pPVT and CeA can be classified to several subtypes [46, 47]. Our investigation focused on the general functional connectivity between the pPVT and CeA without delineating the specific subpopulations within this circuit that mediate the observed effects. Future studies are warranted to identify which neuronal subpopulations are critical for the pPVT-CeA circuit function.

Conclusions

The FosTRAP strategy is a valuable tool for screening functionally engaged brain regions in a chronic migraine model. The pPVT^glu- CeA^GABA circuit is involved in mechanical hyperalgesia behaviors in chronic migraine, and could be a potential candidate for chronic migraine treatment.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

4-OHT

4-Hydroxytamoxifen

BLA

Basolateral amygdala

BNST

Bed nucleus of the stria terminalis

CCI

Chronic constriction injury

CeA

Central amygdala

CeC

Central amygdala, capsular part

CeL

Central amygdala, lateral part

CeM

Central amygdala, medial part

CFA

Freund’s adjuvant

CNO

Clozapine N-oxide

EPM

Elevated plus maze test

FosTRAP

Fos targeted recombination in active populations

IPAC

Interstitial nucleus of the posterior limb of the anterior commissure

NAc

Nucleus accumbens

NTG

Nitroglycerin

OFT

Open field test

pERK

Phosphorylated Extracellular Signal-Regulated Kinase

PVN

Paraventricular nucleus of the hypothalamus

PVT

Paraventricular thalamus

TST

Tail suspension test

VTA

Ventral tegmental area

ZI

Zona Incerta

Author contributions

All authors approved the final manuscript. W.N. injected the virus, implanted the ferrule fiber and performed behavior tests. W.N., H.Y., C.Y., D.Z., B.L., and Z.H. helped wrote the original draft and optimize the experimental design. C.C., Z.M., W.T., Y.C., X.L., J.L., and S.Y. revised the manuscript. R.L., W.T., and D.Z. contributed to the study guidance and performed project coordination. S.Y. supervised the study.

Funding

We acknowledge the financial support from Research Project of China (145BHQ090003000X12).

Data availability

Within the article and its supplementary figures, readers can find data, reagents, protocols, and resources described in this study. The corresponding author can also provide them upon request.

Declarations

Ethics approval and consent to participate

Experimental protocols received approval from the Institutional Animal Care and Use Committee of the Chinese PLA General Hospital.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.