Abstract
Our recent study demonstrated the critical role of the mesolimbic dopamine (DA) circuit and its brain-derived neurotropic factor (BDNF) signaling in mediating neuropathic pain. The present study aims to investigate the functional role of GABAergic inputs from the lateral hypothalamus (LH) to the ventral tegmental area (VTA; LHGABA→VTA) in regulating the mesolimbic DA circuit and its BDNF signaling underlying physiological and pathologic pain. We demonstrated that optogenetic manipulation of the LHGABA→VTA projection bidirectionally regulated pain sensation in naive male mice. Optogenetic inhibition of this projection generated an analgesic effect in mice with pathologic pain induced by chronic constrictive injury (CCI) of the sciatic nerve and persistent inflammatory pain by complete Freund's adjuvant (CFA). Trans-synaptic viral tracing revealed a monosynaptic connection between LH GABAergic neurons and VTA GABAergic neurons. Functionally, in vivo calcium/neurotransmitter imaging showed an increased DA neuronal activity, decreased GABAergic neuronal activity in the VTA, and increased dopamine release in the NAc, in response to optogenetic activation of the LHGABA→VTA projection. Furthermore, repeated activation of the LHGABA→VTA projection was sufficient to increase the expression of mesolimbic BDNF protein, an effect seen in mice with neuropathic pain. Inhibition of this circuit induced a decrease in mesolimbic BDNF expression in CCI mice. Interestingly, the pain behaviors induced by activation of the LHGABA→VTA projection could be prevented by pretreatment with intra-NAc administration of ANA-12, a TrkB receptor antagonist. These results demonstrated that LHGABA→VTA projection regulated pain sensation by targeting local GABAergic interneurons to disinhibit the mesolimbic DA circuit and regulating accumbal BDNF release.
SIGNIFICANCE STATEMENT The mesolimbic dopamine (DA) system and its brain-derived neurotropic factor (BDNF) signaling have been implicated in pain regulation, however, underlying mechanisms remain poorly understood. The lateral hypothalamus (LH) sends different afferent fibers into and strongly influences the function of mesolimbic DA system. Here, utilizing cell type- and projection-specific viral tracing, optogenetics, in vivo calcium and neurotransmitter imaging, our current study identified the LHGABA→VTA projection as a novel neural circuit for pain regulation, possibly by targeting the VTA GABA-ergic neurons to disinhibit mesolimbic pathway-specific DA release and BDNF signaling. This study provides a better understanding of the role of the LH and mesolimbic DA system in physiological and pathological pain.
Introduction
Chronic pain has been reframed as a disease instead of simply a symptom and affects the health and quality of life of >30% of the world's population (Cohen et al., 2021). However, in half of the affected patients, pain and its associated symptoms, e.g., depression and cognitive dysfunction, cannot be effectively treated with the available analgesic strategies. Therefore, innovations in pain and analgesic mechanisms are urgently required for pain elimination (Raffaeli and Arnaudo, 2017; Raja et al., 2020).
Since pain is a complicated sensory and emotional experience, emotion-related brain structures have attracted great interest when investigating the underlying mechanisms of pain modulation (Schweinhardt et al., 2009; Mitsi and Zachariou, 2016). The ventral tegmental area (VTA), the center of the midbrain dopamine (DA) reward system, is intensively involved in mediating pain sensation and related negative emotions (Haber and Knutson, 2010; Mitsi and Zachariou, 2016). Clinical and preclinical studies have shown that the VTA is activated by pain experiences or nociception. For example, a brain functional imaging study found that both visceral and somatic pain activated the VTA region in healthy subjects (Dunckley et al., 2005). Animal studies have also shown that VTA dopaminergic (VTADA) neurons are excited by multiple nociceptive stimuli including formalin and chronic nerve injury (Wood, 2006; Woo et al., 2015).
VTADA neurons are found to regulate pain sensation through the nucleus accumbens (NAc), one of the well-studied downstream regions of the VTA (Lammel et al., 2014; S. Yang et al., 2020). Recently, we also reported that the putative VTA DAergic inputs to the NAc (VTADA→NAc) are activated in chronic pain, and inhibition of these circuit-specific VTADA neurons has analgesic effects (H. Zhang et al., 2017a; H.R. Wang et al., 2021). Further molecular and pharmacological studies demonstrated that the analgesic effect is mediated by mesolimbic brain-derived neurotrophic factor (BDNF) signaling (H. Zhang et al., 2017a). However, how the mesolimbic circuit and its BDNF signaling in mediating pain sensation are regulated remains largely unknown.
Presynaptic input is another contributor to neural and molecular maladaptation in addition to local intrinsic mechanisms in the VTA (Z. Zhang et al., 2013; Pignatelli and Bonci, 2015). A previous study demonstrated that nociceptive signals transmitted from upstream regions to VTADA neurons could regulate DA release in the NAc (H. Yang et al., 2021), indicating a potential role of VTA upstream regions in mediating pain and analgesia by targeting mesolimbic DAergic neurons. Whole-brain mapping studies have indicated that the VTA receives robust inputs from multiple brain regions including the lateral hypothalamus (LH), a critical brain region for pain modulation (Wardach et al., 2016; Siemian et al., 2021).
LH has been validated to send GABAergic, glutamatergic, and peptidergic projections to the VTA (Kempadoo et al., 2013; Rossi et al., 2021), with great potential to regulate pain sensation by forming distinct neural circuitry with different VTA neuronal subtypes (Wardach et al., 2016; Siemian et al., 2021). A recent in vivo fast-scan cyclic voltammetry (FSCV) study indicated that activating the LHGABA→VTA projection increases dopamine neurotransmission in the NAc, suggesting a possible disinhibition of the mesolimbic DA pathway by the LHGABA→VTA projection (Nieh et al., 2016). Activating glutamatergic neurons in the LH→VTA pathway excites VTADA neurons and promotes reward behaviors (Kempadoo et al., 2013). Neuropeptides, i.e., orexin and neurotensin, released by LH neurons are reported to regulate VTADA neuronal activity (Borgland et al., 2006; Opland et al., 2013). However, the cellular and molecular mechanisms underlying LH modulation of the mesolimbic DA circuit in pain sensation lack direct anatomic and functional evidence.
The present study applied in-depth cell type-specific and circuit-specific viral tracing and recording, optogenetics, and fiber photometry with classic histologic, pharmacology and behavioral assessments to demonstrate that the LHGABA→VTA circuitry promoted pain sensation by preferentially targeting VTAGABA neurons. Indirect disinhibition of the NAc-projecting VTADA subpopulation further increased mesolimbic DA release and BDNF signaling. Our findings propose the LHGABA→VTAGABA→VTADA→NAc pathway as a novel circuit for pain sensation and provide a better understanding of the role of the mesolimbic DA system in physiological and pathologic pain.
Materials and Methods
Animals
All experiments were reviewed and approved by the Animal Care and Use Committee of Xuzhou Medical University and performed in accordance with the National Institutes of Health Guidelines and Use of Laboratory Animals and the Committee for Research and Ethical Issues of the International Association for the Study of Pain. Mice were group-housed (maximum five mice per cage) under a 12/12 h light/dark cycle (light from 8 A.M. to 8 P.M.), with food and water available ad libitum. The ambient temperature was maintained at 21–22°C with 55% relative humidity. Only C57BL/6J male mice (8–13 weeks old) of normal appearance and weight were used for all studies. vGAT-ires-cre (B6J.129S6(FVB)-Slc32a1tm2(cre)Lowl/MwarJ, stock #028862) mice were purchased from The Jackson Laboratory and bred onto a C57BL/6J genetic background. All behavioral tests were conducted during the light period, and the investigators were blinded to the experimental conditions during testing.
Adenovirus-associated virus (AAV) vectors
All viruses were purchased from Brain VTA Technology Co Ltd.: AAV-Ef1α-DIO-EYFP [2/9, 5.24E + 12 viral genome/ml (vg/ml)]; AAV-hSyn-DIO-hChR2(H134R)-EYFP; (2/9, 5.63E + 12 vg/ml); AAV-Ef1α-DIO-eNpHR3.0-EYFP (2/9, 3.09E + 12 vg/ml); AAV-Ef1α-DIO-EGFP (2/9, 2.54E + 12 vg/ml); AAV-vGAT1-cre (2/R, 2.15E + 12 vg/ml); AAV-Ef1α-DIO-eNpHR3.0-EGFP (2/9, 3.11E + 12 vg/ml); AAV-hSyn-DIO-ChrimsonR-mCherry (2/9, 5.29E + 12 vg/ml); AAV-Ef1α-DIO-mCherry (2/9, 5.14E + 12 vg/ml); AAV-Ef1α-DIO-GCaMp6s (2/9, 5.00E + 12 vg/ml); AAV-vGAT1-Cre (2/9, 3.26E + 12 vg/ml); AAV-TH-Cre (2/9, 6.29E + 12 vg/ml); AAV-Ef1α-DIO-His-EGFP-2a-TVA (2/9, 1.20E + 11 vg/ml); AAV-Ef1α-DIO-RVG (2/9, 1.20E + 11 vg/ml); RV-ENVA-ΔG-DsRed (2.00E + 08 IFU/ml); AAV-Ef1α-DIO-EGFP-T2A-TK (2/9, 2.78E + 12 vg/ml); HSV-ΔTK-LSL-tdTomato (2/9, 2.00E + 09 PFU/ml); AAV-hSyn-DA4.4 (2/9, 2.00E + 12 vg/ml); AAV-DIO-hChR2-mCherry (2/R, 2.19E + 12 vg/ml); AAV-DIO-mGFP-T2A-Synaptophysin-mRuby (2/9, 4.01E + 12 vg/ml); AAV-cFos-tTA-NLS-FLAG (2/9, 2.57E + 12 vg/ml) and AAV-Tre3g-DIO-hChR2-mCherry (2/9, 1.76E + 12 vg/ml).
Viral injection and optical fiber/cannula implantation
Stereotaxic surgeries were conducted with a stereotaxic apparatus (RWD Life Technology Co, Ltd.). Mice were anesthetized with ketamine and xylazine (100 and 10 mg/kg, i.p.). For microinjection, 200 nl of AAV vectors was injected into the LH (Anteroposterior (AP): −1.35 mm; Mediolateral (ML): ±0.90 mm; Dorsoventral (DV): 5.40 mm), VTA (AP: −3.30 mm; ML: ±0.50 mm; DV: 4.55 mm) or NAc (AP: 1.40 mm; ML: ±0.60 mm; DV: 4.60 mm) with a 33-gauge Hamilton syringe needle (Hamilton Company Inc.) at a rate of 100 nl/min, followed by a 10-min pause to minimize backflow. For projection-specific laser stimulation, optical fibers [diameter: 200 µm, numerical aperture (N.A.): 0.48; NEWDOON] were implanted above the target regions: LH (AP: −1.35 mm; ML: ±0.90 mm; DV: 5.20 mm) and VTA (AP: −3.30 mm; ML: ±1.05 mm; DV: 4.40 mm with a 7° angle). For in vivo fiber-photometry recording, optical fibers with a back ceramic ferrule [diameter: 200 µm, numerical aperture (N.A.): 0.48; NEWDOON] were implanted into the VTA (AP: −3.30 mm; ML: ±1.05 mm; DV: 4.40 mm with a 7° angle) or the NAc (AP: 1.40 mm; ML: ±0.60 mm; DV: 4.60 mm).
For pharmacological experiments, a bilateral cannula (RWD Life Technology Co, Ltd.) was implanted above the NAc (AP: 1.40 mm; ML: ±0.60 mm; DV: 4.50 mm) for infusion of the TrkB receptor antagonist ANA-12 (1 μg in 300 nl, Sigma). A stainless-steel obturator was inserted into each guide cannula to prevent blockage. Only mice with the correct optical fiber/cannula locations and viral expression were used for further analysis. Erythromycin ointment was applied locally to avoid infection. After the operation, mice were placed in a cage with a heating pad to keep warm and returned to their home cage when fully awake.
Monosynaptic tracing
To map the monosynaptic inputs onto the VTAGABA neurons, in vGAT-Cre mice, a 200 nL virus mixture of AAV-Ef1α-DIO-His-EGFP-2a-TVA and AAV-Ef1α-DIO-RVG was injected into the VTA followed by a second injection of 200 nl RV-EnVA-ΔG-DsRed into the same site three weeks later. Mice were housed for 7 d to allow for transsynaptic rabies spread and fluorescent protein expression and then terminated for tissue harvests for histologic assessments. The cells expressing EGFP and DsRed simultaneously in the VTA were considered starter cells (Wickersham et al., 2007). To trace the downstream cells receiving monosynaptic innervation from the LHGABA neurons, 200 nl of AAV-Ef1α-DIO-EGFP-T2A-TK was injected into the LH of vGAT-Cre mice, and 200 nl HSV-ΔTK-LSL-tdTomato was injected into the same site three weeks later. Because of HSV-based tracers being toxic and lethal, mice were closely observed and killed 3 d following the last injection (Lo and Anderson, 2011). Individual mice were excluded when the two injections were inconsistent or the starter cells (the initial HSV or rabies-infected cells) were not limited to the VTA or LH area (Callaway and Luo, 2015).
Sciatic nerve chronic constriction injury model of neuropathic pain
Chronic constrictive injury (CCI) surgery was conducted as described in our previous research (Xia et al., 2020). Mice were anesthetized with ketamine and xylazine (100 and 10 mg/kg, i.p.). After the nerves were exposed at the mid-thigh level, three ligatures (4–0 silk) were placed around the nerve proximal to the trifurcation with 1 mm between each ligature. The ligatures were loosely tied until a short flick of the ipsilateral hindlimb was observed. Animals in the sham group received surgery without nerve ligation, and erythromycin ointment was applied locally following the suture. After the operation, mice were placed in a cage with a heating pad to keep warm and returned to their home cage when fully awake.
Complete Freund's adjuvant model of persistent inflammatory pain
Completed Freund's adjuvant (CFA; F5881, Sigma) was dissolved in sterile saline at a volume ratio of 1:1. To establish the persistent inflammatory pain model, 10 μl of diluted CFA was subcutaneously injected into the left hind plantar with a 20-gauge micro-injector (Schwartz et al., 2014). Mice from the sham group received the same volume of the vehicle instead. Behavioral tests were performed 4 h and 3 d after the injection to evaluate acute and persistent pain.
Paw withdrawal latency (PWL)
PWLs were measured with the IITC Plantar Analgesia Meter (IITC Life Science Inc., CA) in a double-blinded manner as described in our previous studies (S. Zhang et al., 2017b; Liu et al., 2018). A heat-producing radiant light source was used to stimulate the plantar surface of the left hind paw. The time from the light-on to a typical withdrawal or licking of the tested hind paw was recorded as the paw withdrawal latency. The basal PWLs were set to 9–15 s by adjusting the radiant light intensity, and the radiant heat illumination was automatically cut off at 20 s to prevent tissue damage. The PWLs were measured five times/time points/animal, with the last three measurements used for analysis.
50% paw withdrawal threshold (50% PWT)
The 50% PWT measurement was adapted and conducted with the up-down paradigm as previously described (Chaplan et al., 1994). Mice were acclimatized for 1 h in transparent acrylic enclosures (10 × 10 × 6 cm) on a wire mesh platform in a temperature-controlled and quiet room. A sequence of calibrated von Frey filaments (0.02, 0.04, 0.07, 0.16, 0.4, 1.0, 2.0, and 6.0 g) was chosen. The measurement was initiated with the 0.16-g filament. Each filament was applied perpendicularly to the plantar surface of the left hind paw, with sufficient force to bend the filament. Lifting, shaking, or licking of the paw indicated a positive response and prompted the use of the next weaker (i.e., lighter) filament. The absence of a paw withdrawal response prompted the use of the next stronger filament. This paradigm continued until six measurements were taken or until five consecutive positive or four consecutive negative responses occurred. The 50% PWT was calculated as 50% PWT = 10[log(Xf) + κδ], Xf = value (in log units) of the final von Frey filament used; κ = tabular value (Chaplan et al., 1994; see their appendix) for the pattern of positive/negative responses, and δ = mean difference (in log units) between stimuli (here, 0.411).
Open field test (OFT)
The mice were habituated for at least 1 h in the testing environment before the behavioral test was conducted. Mice were gently placed in the center of a white Plexiglas open-field arena (50 × 50 × 40 cm) illuminated by a 30-W white fluorescent light 2 m overhead. After 2 min of adaptation, the total locomotor activity of each animal was automatically recorded with ANY-maze software (Stoelting Co) for 5 min, and the total distance traveled and mean moving velocity were analyzed.
Histology
Mice were anesthetized and subjected to intracardial perfusion with PBS (pH 7.4) and 4% paraformaldehyde. The brains were separated and postfixed at 4°C overnight, followed by a 30% sucrose solution dip for 48 h. Thirty-micrometer slices were prepared by a freezing microtome (CM1950, Leica Microsystems). For immunofluorescence, slices were incubated overnight with the primary rabbit anti-c-Fos (1:1000, 2250, Cell Signaling Technology) antibody. After being washed in PBS for 5 × 3 min, the sections were tagged with secondary anti-rabbit Alexa-594 (1:200, A21207, Thermo Fisher Scientific) antibody for 2 h. After staining, the sections were mounted onto glass slides with mounting medium containing DAPI; 10×, 20×, and 40× images were obtained using a confocal microscope (LSM 880, Carl Zeiss).
Western blotting
VTA and NAc-containing brain tissue were harvested and sonicated subsequently in 200-μl lysis buffer containing phosphatase and protease inhibitors. Following centrifugation (12,000 rpm, 15 min), the supernatants were collected for protein concentration assay by an OD-assay facility (Thermo Scientific). Proteins (40 μg per lane) were electrophoresed in a 10% SDS-PAGE gel and transferred onto PVDF membranes (Millipore). The membranes were rinsed and then blocked in a 3% BSA solution (V900933, Sigma-Aldrich) for 2 h at room temperature, followed by incubation at 4°C overnight with primary rabbit anti-GAPDH (1:2000, 10 494–1-AP, Proteintech) and rabbit anti-BDNF (1:1000, ab108319, Abcam). Then the membranes were washed and incubated with a secondary antibody conjugated to alkaline phosphatase (1:1000, VA006, Vicmed) for 2 h at room temperature. Afterward, the bands were visualized by a coloring solution consisting of BCIP (S381C, Promega) and NBT (S380C, Promega). The grayscale value of BDNF protein was measured and standardized to that of GAPDH.
RT-PCR
Total RNA from the tissue blocks of the NAc and mPFC was extracted using an RNA purification kit (B518651, Sangon Biotech). cDNA was generated by reverse transcription of total RNA (500 ng-1 µg) with HiScript II Q RT SuperMix (R223-01, Vazyme), followed by the preparation of 20 µL reaction mixtures in triplicate with ChamQ SYBR Color qPCR Master Mix (Q421-01, Vazyme). Subsequently, qRT–PCR was performed with a Light Cycler 480 II (Roche Diagnostics), followed by the quantification of BDNF standardized to GAPDH. Finally, the results were analyzed by the comparative CT method (2–ΔΔCT). The sequences of the primers (Sangon Biotech) are as follows: bdnf:
TCATACTTCGGTTGCATGAAGG (forward),
AGACCTCTCGAACCTGCCC (reverse).
gapdh:
CAATGTGTCCGTCGTGGATCT (forward),
GTCCTCAGTGTAGCCCAAGATG (reverse).
Optogenetics
To optogenetically manipulate the LHGABA-VTA circuit in vGAT-Cre mice, AAV-DIO-hChR2-EYFP, AAV-DIO-ChrimsonR-mCherry and AAV-DIO-eNpHR3.0-EYFP were bilaterally injected into the LH followed by optical fibers implantation above the VTA or the LH. Optical fibers were connected to a combined laser generator and the stimulator (NEWDOON), which was used to generate 473-nm blue light or 589-nm yellow light. For optogenetic activation and inhibition of LHGABA→VTA terminals, pulses of 473-nm light (10 Hz, 10 ms, 10 mW) and constant 589-nm light (8-s-on/2-s-off, 10 mW) were delivered to the hChR2 and eNPHR3.0 groups, respectively, in this study. Additionally, 589-nm light (10 Hz, 10 ms, 10 mW) was used to activate the cell bodies of LHGABA→VTA neurons that express ChrimsonR in fiber photometry and molecular experiments.
In vivo fiber photometry
To record dopamine release evoked by projection-specific activation, AAV-hSyn-DA4.4 was injected into the NAc of C57BL/6J mice to express a GPCR-activation-based-DA sensor, and AAV-hSyn-DIO-ChR2-mCherry was injected into the LH simultaneously.
After two to three weeks, optical fibers for recording were implanted into the NAc, and optical fibers for stimulation were implanted into the VTA. Then each mouse was allowed to recover for one week before recording. A fiber photometry system (THINKERTECH) was used for recording. After being connected with a recording fiber wire, mice were placed in a hamster cage and allowed to move freely. The starting and ending points of stimulation were marked by the digital tag produced by the optical stimulator. The fluorescence response (ΔF/F) from 10 s preceding the stimulation (F0, baseline) to 20 s after the stimulation (F) was calculated using the formula: ΔF/F = (F – F0)/F0. The ΔF/F values of all optical stimuli were then averaged and plotted with a shaded area indicating the SEM. At the end of the experiment, all animals were perfused. Only data from animals with correct optical fiber implantation sites and virus expression were included in the analysis. The data analysis was processed using MATLAB software (version R2018a, MathWorks).
Activity-dependent expression of channelrhodopsins (ChR2)
In the Tet-off system, gene expression is turned on when doxycycline (Dox) is removed from the extracellular environment. Thus, 1% doxycycline was added to the drinking water of mice 3 d before virus injection to increase the concentration of doxycycline in CSF. Then, 200 nl of a virus mixture of AAV-cFos-tTA-NLS-FLAG and AAV-Tre3g-DIO-hChR2-mCherry was injected into the LH of vGAT-Cre mice. Doxycycline was added continuously until day 21 to inhibit c-Fos-derived expression. After the doxycycline was removed, CFA was injected into the left hind paw to evoke the pain-related activation of LH neurons and the c-Fos-derived expression of tetracycline-controlled transactivator (tTA). As soon as tTA was combined with tetracycline-responsive element (Tre3g), the GABAergic neurons started to express hChR2. Three days after CFA injection, 1% doxycycline was added to the drinking water again to avoid the nonpain-induced expression of hChR2.
In vitro electrophysiology
As we previously reported (H. Zhang et al., 2017a), mice were anesthetized with isoflurane and perfused immediately with ice-cold artificial CSF [aCSF; which contained (in mm): 128 NaCl, 3 KCl, 1.25 NaH2PO4, 10 D-glucose, 24 NaHCO3, 2 CaCl2, and 2 MgCl2 (oxygenated with 95% O2 and 5% CO2, pH 7.4, 295–305 mOsm)]. Acute brain slices containing LH were cut using a microslicing vibratome (DTK-1000, Ted Pella) in ice-cold sucrose aCSF, which was derived by entirely replacing the NaCl in the solution with 254 mm sucrose and saturated by 95% O2 and 5% CO2. Slices were maintained in holding chambers with aCSF for 1 h of recovery at 32°C. For cell-attached recordings, the patch pipettes (3–8 mm) were filled with 130 mm K-methanesulfate, 10 mm KCl, 3 mm ATP–Na+, 0.5 mm GTP–Na+, 0.4 mm EGTA, 10 mm HEPES, and 7.5 mm phosphocreatine (Na2; pH adjusted to 7.45 with KOH, osmolarity adjusted to 276–280 mOsm). The recording was performed in gap-free mode. The signal was acquired via a MultiClamp 700B amplifier, low-pass filtered at 2.8 kHz, digitized at 10 kHz, and analyzed with Clampfit 10.7 software (Molecular Devices).
Light-evoked responses
Optical stimulation was delivered using a laser (NEWDOON) through an optical fiber 200 μm in diameter positioned 0.2 mm away from the surface of the brain slice. To test the functional characteristics of hChR2, fluorescently labeled neurons expressing hChR2 from vGAT-Cre mice were visualized and stimulated with a blue laser (10 Hz, 10 ms, 473 nm, 10 mW). To test eNpHR3.0, sustained photostimulation (594 nm, 8 s, 10 mW) was delivered instead.
Cell counting
Sections (30 mm, from bregma −1.26 to −1.52) were reserved for counting at an interval of one section. Two to four sections were sampled from each mouse (the exact N values are provided in the figure legends). The lateral hypothalamus (LH) was delineated according to a reference atlas (Paxinos et al., 2019). All of the labeled cells within the LH were counted. Manual counting by ZEN software (version 2.6, Carl Zeiss) was used to count the c-Fos-positive, EGFP-positive and double-labeled cells, during which the investigator was blinded to the experimental conditions. The labeled neurons were identified independently for each fluorescent channel, among which those with colocalization of two kinds of fluorescence were considered double-labeled neurons.
Statistics
Data are presented as the mean ± SEM. All analyses were performed with Prism software (version 7.0) without specification. The normality of the data was statistically tested by the D'Agostino–Pearson omnibus normality test. Normally distributed data from multiple groups were compared with one-way or two-way analysis of variance with/without repeated factors (one-way or two-way ANOVA), followed by post-Bonferroni's multiple comparison tests when appropriate. Comparisons between two groups were analyzed with an unpaired Student's t test. Statistical significance was set at *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
Results
LHGABA→VTA projection bidirectionally regulates pain sensation under physiological condition
To determine the functional role of the GABAergic inputs from the LH to the VTA in pain sensation regulation, we bilaterally injected a cocktail of Cre recombinase-inducible adeno-associated viruses (AAVs) expressing optogenetic channelrhodopsin (AAV-DIO-hChR2) and halorhodopsin (AAV-DIO-eNpHR3.0) fused with an enhanced yellow fluorescent protein (EYFP) reporter into the LH of vesicular GABA transporter (vGAT)–Cre mice to label LHGABA neurons (Fig. 1A). Confocal immunofluorescence staining showed the expression of EYFP in LHGABA neurons and their axon terminals in the VTA (Fig. 1B). In vitro electrophysiological recording demonstrated that optogenetic stimulation with 473-nm blue laser or 589-nm yellow laser evoked real-time increase or decrease in neuronal firing activity of LHGABA neurons in acutely-isolated LH-containing brain slices (Fig. 1C). Pain behavioral tests were conducted at 2-h intervals: baseline→yellow laser on→laser off→blue laser on→laser off. In vGAT-Cre mice with hChR2 and eNpHR3.0 expression, bilateral yellow laser stimulation (8-s-on/2-s-off continuous stimulation) in the VTA increased animals' paw withdrawal latencies (PWLs) to noxious thermal stimulation and their 50% mechanical paw withdrawal threshold (50% PWTs) to the Von Frey mechanical stimuli when compared with their EYFP counterparts (Fig. 1D,E). The elevated PWLs and 50% PWTs returned to the baseline levels 2 h after the laser was turned off. Correspondingly, bilateral VTA blue laser stimulation significantly decreased the PWLs and 50% PWTs, which recovered to the baseline levels 2 h poststimulation (Fig. 1D,E). These cell-type and projection-specific optogenetic experiments strongly indicate a bidirectional regulatory property in the LHGABA→VTA circuit during pain sensation under a physiological condition.
LHGABA→VTA circuit bidirectionally regulates pain perception in naive mice. A, Experimental timeline and schematic for optogenetic manipulation of the LHGABA→VTA circuit in the VTA. B, Representative immunofluorescent images showing virus expression in the LH GABAergic neurons and their axon terminals in the VTA. Scale bar, 200 µm. C, In vitro electrophysiology for neuronal responses to laser illumination in acute isolated LH-containing brain slice. D, Acute real-time optogenetic inhibition/activation of the LHGABA→VTA circuit in the VTA increased/decreased PWLs in naive mice. n = 8 mice/group. PWLs: group, F(1,14) = 1.177, p = 0.2963; EYFP versus eNpHR3.0/hChR2-EYFP, yellow laser on p = 0.0011, blue laser on p = 0.0007. E, Acute real-time optogenetic inhibition/activation of the LHGABA→VTA circuit in the VTA increased/decreased 50% PWTs in naive mice. n = 7–8 mice/group. 50% PWTs: group, F(1,13) = 3.081, p = 0.1027; EYFP versus eNpHR3.0/hChR2-EYFP, yellow laser on p = 0.0151, blue laser on p = 0.0004. *p < 0.05, **p < 0.01, ***p < 0.001. Data analyzed by (D, E) two-way ANOVA with Šídák post hoc tests. Error bars indicate SEM LH, lateral hypothalamic area; VTA, ventral tegmental area; ic, internal capsule; BL, baseline.
LHGABA→VTA projection is required for the maintenance of pathologic pain
To explore the response and potential role of LHGABA→VTA projection in the context of pathologic pain, we examined the activity of LHGABA→VTA neurons and their necessity during the development of CCI-induced neuropathic pain. We first labeled LHGABA→VTA neurons by injecting a retrograde AAV expressing Cre driven by the vGAT1 promoter into the ipsilateral VTA and a Cre-inducible AAV expressing EGFP into the ipsilateral LH to specifically label VTA projecting LHGABA neurons. The mice then underwent CCI surgery on the contralateral hindpaw on the side of the injection to model neuropathic pain three weeks after the virus injection. On day 7 after CCI surgery, mice were perfused to assess the expression of c-Fos protein, an indicator of neuronal activity (Fig. 2A). The results demonstrated an increased c-Fos protein expression in EGFP-labeled LHGABA→VTA neurons in the contralateral LH of the CCI-affected paw compared with the sham-treated mice (Fig. 2B,C). The data suggest that CCI-induced neuropathic pain activates LHGABA→VTA neurons. Correspondingly, real-time optogenetic inhibition of LHGABA terminals (Fig. 2D) in the VTA was sufficient to elevate the PWLs and 50% PWTs in mice expressing eNpHR3.0 compared with those EGFP control mice (Fig. 2E,F).
CCI surgery-induced neuropathic pain and CFA-induced inflammatory pain require activation of the LHGABA→VTA circuit. A, Experimental timeline and schematic for retrograde labeling of LHGABA→VTA neurons in C57BL/6J mice. B, Representative immunofluorescent images for c-Fos protein expression in LHGABA→VTA projecting neurons in sham and CCI mice 7 d after CCI surgery. Scale bar, 100 µm. C, Quantitative data of c-Fos protein expression. Left, Total number of c-Fos-positive cells that co-express EGFP in CCI and sham groups, n = 2–3 slices from each of 3 mice/group, Sham, 6.57 ± 1.15; CCI, 25.44 ± 2.46; t(14) = 6.331, p < 0.0001. Right, Percentages of c-Fos-positive neurons out of EGFP-labeled LHGABA→VTA projecting neurons in CCI and sham groups, n = 2–3 slices from each of 3 mice/group, Sham, 11.57 ± 1.45; CCI, 21.00 ± 1.03; t(14) = 5.460, p < 0.0001. D, Experimental timeline and schematic for optogenetic inhibition of LHGABA→VTA circuit in the VTA. E, F, Acute real-time optogenetic inhibition of the LHGABA→VTA circuit in the VTA increased (E) PWLs and (F) 50% PWTs in CCI mice. nSham + EGFP = 6; nCCI + EGFP = 7; nCCI + eNpHR3.0 = 7. PWLs: group, F(2,17) = 109.1, p < 0.0001; CCI + EGFP versus CCI + eNpHR3.0, laser on p = 0.0006. 50% PWTs: group, F(2,17) = 5.986, p = 0.0108; CCI + EGFP versus CCI + eNpHR3.0, laser on p = 0.0222. G, Experimental timeline and schematic for retrograde labeling of LHGABA→VTA neurons in C57BL/6J mice. H, Representative immunofluorescent images for c-Fos protein expression in LHGABA→VTA projecting neurons in sham and CFA mice 3 d after CFA injection. Scale bar, 100 µm. I, Quantitative data of c-Fos protein expression. Left, Total number of c-Fos-positive cells that co-express EGFP in CFA and sham groups, n = 3 slices from each of 3 mice/group. Sham, 25.22 ± 2.98; CFA, 54.56 ± 6.14; t(16) = 4.297, p = 0.0006. Right, Percentages of c-Fos protein-positive neurons out of EGFP-labeled LHGABA→VTA projecting neurons, n = 2–3 slices from 3 mice/group. Sham, 11.57 ± 1.45; CFA, 21.00 ± 1.80; t(14) = 3.915, p = 0.0016. J, Experimental timeline and schematic for optogenetic inhibition of LHGABA→VTA circuit in the VTA. K, L, Four hours following CFA injection: acute real-time optogenetic inhibition of the LHGABA→VTA circuit in the VTA increased (K) PWLs and (L) 50% PWTs in CFA mice. n = 8 mice/group. PWLs: group, F(2,21) = 141.3, p < 0.0001. CFA + EGFP versus CFA + eNpHR3.0 laser on p < 0.0001. 50% PWTs: group, F(2,21) = 21.02, p < 0.0001. CFA + EGFP versus CFA + eNpHR3.0, laser on p = 0.0008. M, N, Three days following CFA injection: acute real-time optogenetic inhibition of the LHGABA→VTA circuit in the VTA increased (M) PWLs and (N) 50% PWTs in CFA mice. n = 8 mice/group. PWLs: group, F(2,21) = 50.43, p < 0.0001. CFA + EGFP versus CFA + eNpHR3.0 laser on p = 0.0002. 50% PWTs: group, F(2,21) = 20.13, p < 0.0001. CFA + EGFP versus CFA + eNpHR3.0, laser on p < 0.0001. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data analyzed by (C, I) unpaired t test, or (E, F, K, L, M, N) two-way ANOVA with Tukey's post hoc tests. Error bars indicate SEM LH, lateral hypothalamic area; ic, internal capsule; 3V, third ventricle; VTA, ventral tegmental area.
To evaluate the necessity of LHGABA→VTA neuronal activation in mediating pathologic pain sensation, we repeated the histologic and optogenetic inhibition experiments in another well-established inflammatory pain model in rodents, i.e., intraplantar injection of CFA. Immunofluorescence staining for c-Fos protein was performed on brain slices from mice that were injected with CFA for 3 d. Consistent with the CCI mice, the CFA mice also displayed a significant increase in c-Fos expression in LHGABA→VTA neurons in the contralateral LH of the CFA-affected paw (Fig. 2G–I). Behavioral tests and optogenetic modulation were performed at two selected time points, 4 h following CFA injection for acute pain and 3 d following injection for persistent pain (Fig. 2J). Compared with the EGFP group, optogenetic inhibition of LHGABA→VTA neurons significantly increased PWLs and 50% PWTs in the acute pain phase (Fig. 2K,L), as well as the persistent pain phase (Fig. 2M,N).These data indicate that both CCI-induced neuropathic pain and CFA-induced inflammatory pain activate the LHGABA→VTA circuit which is required for the maintenance of pathologic pain.
LHGABA→VTA projection distinctly modulates VTA DAergic and GABAergic neuronal activity
Both GABAergic and DAergic neurons in the VTA receive GABAergic inputs from the LH (Beier et al., 2015; Bouarab et al., 2019). However, whether synaptic specificity, i.e., a preference for specific postsynaptic partners, exists between LHGABA and VTA subpopulations remains unknown. We next examined how activation of LH GABAergic projections affects the neuronal activity of VTAGABA and VTADA, respectively, in free-moving mice using in vivo fiber photometry recording. To target VTA GABAergic neurons, Cre inducible AAV-DIO-GCaMp6s was injected into the VTA and AAV-DIO-ChrimsonR into the LH of vGAT-Cre mice (Fig. 3A–C). Multiple trials in different animals consistently showed a significant decrease in fluorescent calcium signaling in VTAGABA neurons when LHGABA neurons were activated (Fig. 3D–F). This result indicates that activation of LHGABA neurons directly inhibits VTAGABA neurons at the population level. In vivo fiber photometry was performed to record the activity of VTADA neurons. In C57BL/6J mice, a cocktail of AAV-TH-Cre and AAV-DIO-GCaMp6s was injected into the VTA to label VTADA neurons, while another viral cocktail of AAV-vGAT1-Cre and AAV-DIO-ChrimsonR was injected into the LH (Fig. 3G–I). Unlike VTAGABA neurons, VTADA neurons displayed a significant increase in intracellular calcium signaling when stimulating LHGABA neurons with a yellow laser (Fig. 3J–L), suggesting that an indirect disinhibition circuit plays a role. These findings led us to hypothesize that GABAergic inputs from the LH preferentially synapse on VTAGABA neurons and further disinhibit VTADA neurons.
Optogenetic activation of the LHGABA→VTA circuit decreases GABAergic neuronal activity while increases VTA DAergic neuronal activity at a population level. A, Experimental timeline and schematic for optogenetic activation of LHGABA→VTA circuit in the LH and in vivo fiber photometry recording in VTA GABAergic neurons in free-moving vGAT-Cre mice. B, C, Representative immunofluorescent images showing optogenetic virus expression in (B) LH and GCaMp6s expression in (C) VTA GABAergic neurons. Scale bar, 200 µm. D, Representative false heatmap showing calcium signaling change over time with optogenetic activation of LHGABA→VTA neurons in the LH in ChrimsonR and mCherry groups. n = 23 trials from 5 mice. E, Average quantification curve showing calcium-signaling change over time. n = 23 trials from 5 mice in ChrimsonR group; n = 13 trials from 3 mice in mCherry group. F, Area under the curve (AUC) over three periods: baseline, yellow laser on, and laser off. n = 23 trials from 5 mice in ChrimsonR group; n = 13 trials from 3 mice in mCherry group. ChrimsonR: group, F(1.645,37.83) = 41.96, p < 0.0001; prestim versus stim, p < 0.0001. mCherry: group, F(1.098,14.27) = 3.008, p = 0.1020; prestim versus stim, p = 0.4245. G, Experimental timeline and schematic for optogenetic activation of LHGABA→VTA circuit in the LH and in vivo fiber photometry recording in VTA DAergic neurons of free-moving C57BL/6J mice. H, I, Representative immunofluorescent images showing optogenetic virus expression in (H) LH and GCaMp6s expression in (I) VTA DAergic neurons. Scale bar, 200 µm. J, Representative false heatmap showing calcium signaling change over time with optogenetic activation of LHGABA→VTA neurons in the LH in ChrimsonR and mCherry groups. n = 17 trials from 4 mice/group. K, Average quantification curve showing calcium signaling change over time. n = 17 trials from 4 mice in ChrimsonR and mCherry group. L, AUC over three periods: baseline, yellow laser on, and laser off. n = 17 trials from 4 mice in ChrimsonR and mCherry group. ChrimsonR: group, F(1.937,32.93) = 70.03, p < 0.0001; prestim versus stim, p < 0.0001. mCherry: group, F(1.946,31.38) = 0.6809, p = 0.5022; prestim versus stim, p = 0.7535. ****p < 0.0001, n.s, no significance. Data analyzed by (F, L) repeated measures one-way ANOVA with Tukey's post hoc test. Error bars indicate SEM LH, lateral hypothalamic area; VTA, ventral tegmental area; ic, internal capsule; f, fornix; mt, mammillothalamic tract; ml, medial lemniscus; ZI, zona incerta; IP, interpeduncular nucleus; AUC, Area under the curve; BL, baseline.
Monosynaptic connection between LH GABAergic neurons and VTA GABAergic neurons
To validate the disinhibitory hypothesis, in vGAT-Cre mice, we employed modified rabies virus with monosynaptic retrograde tracing properties to strictly identify presynaptic components that target VTAGABA neurons. Coexpression of the avian receptor TVA and rabies glycoprotein G (RVG) in VTAGABA neurons was achieved through VTA injection of AAV-DIO helper viruses (AAV-DIO-EGFP-TVA and AAV-DIO-RVG; Fig. 4A). Rabies virus (RV-EnVA-ΔG-DsRed) was injected into the same site three weeks later. As the rabies virus could only transfect cells expressing the TVA receptor, transfection and labeling were limited to Cre-expressing VTAGABA neurons (EGFP+), which were considered starter cells (Fig. 4B). Cre-inducible AAV-DIO-EGFP virus was injected into the LH; thus, retrograde transsynaptic labeled VTAGABA-projecting LHGABA neurons (DsRed+: EGFP+) could be distinguished and quantified (Fig. 4C). Quantitative data showed that ∼90% of VTAGABA-innervating neurons (EGFP+) in the LH were GABAergic (Fig. 4D).
Preference of monosynaptic connection between the LH GABAergic neurons and the VTA GABAergic neurons. A, Experimental timeline and surgery schematic for retrograde mapping of presynaptic neurons projecting onto the VTA GABAergic neurons. B, Representative immunofluorescent images showing VTA GABAergic neurons labeling with TVA-EGFP and RV-DsRed. The co-labeled ones are called starter cells, scale bar, 100 µm. C, Representative immunofluorescent images showing EGFP-labeling GABAergic neurons and RV-DsRed labeling presynaptic neurons in the LH. Scale bar, 100 µm. D, Quantitative data showing that ∼90% of (VTA GABAergic neuron)-innervating neurons in the LH are GABAergic. n = 32 from 4 mice. E, Experimental surgery schematic for anterograde mapping of postsynaptic neurons of the LH GABAergic neurons. F, Representative immunofluorescent images showing LH GABAergic neurons labeling with EGFP. Scale bar, 100 µm. G, Representative immunofluorescent images showing EGFP-labeling VTA GABAergic neurons and tdTomato-labeling neurons. Scale bar, 100 µm. H, Quantitative data showing that ∼65% of postsynaptic VTA neurons of the LH GABAergic projection are GABAergic. n = 48 from 5 mice. LH, lateral hypothalamic area; VTA, ventral tegmental area; f, fornix.
Then, we checked the postsynaptic neuronal types of the LH GABAergic projections in the VTA. A herpes simplex virus 1 (HSV-1) strain-based anterograde monosynaptic tracing strategy was adopted using Cre-inducible AAVs expressing the TK gene. Briefly, a helper virus (AAV- DIO-EGFP-T2A-TK) was injected into the LH of vGAT-Cre mice to express TK in GABAergic neurons (EGFP+), followed by injection of HSV-ΔTK-LSL-tdTomato that specifically transfected TK-expressing LHGABA starter cells (tdTomato+: EGFP+) and anterogradely transferred down to the postsynaptic neurons (Fig. 4E). Together with the injection of Cre-inducible AAV-DIO-EGFP vector into the VTA, the postsynaptic VTAGABA neurons could be determined by co-expression of EGFP and tdTomato (Fig. 4F). The results showed that ∼65% of the postsynaptic neurons (tdTomato+) in VTA that receiving LH GABAergic innervation were GABAergic (Fig. 4G,H). These data strongly suggest that the LH mainly sends GABAergic projections to VTAGABA neurons, and that LHGABA neurons prefer to innervate GABAergic neurons in the VTA.
Optogenetic activation of the LHGABA→VTA projection induces DA release in the NAc
We recently reported that the mesolimbic DA pathway regulates pain sensation (H. Zhang et al., 2017a). However, whether and how the mesolimbic DA pathway is regulated by LHGABA→VTA projection in free-moving animals remains unknown. To answer this question, the functional connection between the LHGABA→VTA projection and the mesolimbic DA circuit was examined. A Cre recombinase-inducible genetically encoded GPCR-activation-based-DA (GRABDA) sensor that enables simultaneous monitoring of DA release was virally expressed in NAcGABA neurons in vGAT-Cre mice (Fig. 5A,B). To observe DA release alteration on LHGABA→VTA projection activation, we targeted LHGABA→VTA neurons with AAV-DIO-ChrimsonR-mCherry to permit their optogenetic activation (Fig. 5A–C). Optogenetic activation of LHGABA terminals in the VTA reliably induced an increase in the fluorescence signal in the NAc of GRABDA sensor-expressing mice, indicating an increase of DA release on NAcGABA neurons (Fig. 5D–F).
Optogenetic activation of the LHGABA→VTA projection increases NAc DA release. A, Surgery schematic for optogenetic activation of LHGABA→VTA projection in the VTA and in vivo NAc DA release imaging in free-moving vGAT-Cre mice. B, C, Representative immunofluorescent images showing mCherry expression in (B) LH and GRAB expression in (C) NAc neurons. Scale bar, 200 µm. D, Representative false heatmap showing extracellular NAc DA signaling change over time with or without optogenetic activation of LHGABA→VTA neurons in the VTA. n = 36 trials from 6 mice. E, Average quantification curve showing extracellular NAc DA signaling change over time. n = 36 trials from 6 mice in ChrimsonR group; n = 32 trials from 6 mice in mCherry group. F, AUC of NAc DA signaling over three periods: baseline, yellow laser on, and laser off. n = 36 trials from 6 mice in ChrimsonR group; n = 32 trials from 6 mice in mCherry group. ChrimsonR: group, F(1.698,59.41) = 173.2, p < 0.0001; BL versus laser on, p < 0.0001. mCherry: group, F(1.78,56.95) = 1.890, p = 0.1646; BL versus laser on, p = 0.5081. G, Experimental timeline for the c-Fos-driven hChR2 expression in the LH GABAergic neurons activated by the CFA injection. H, Surgery schematic for optogenetic activation of LHGABA→NAc neurons activated by the CFA injection and in vivo NAc DA release imaging in free-moving vGAT-Cre mice. I, Representative immunofluorescent images showing the mCherry expression in the LH GABAergic neurons before and after transgene induction and decay. Scale bar, 200 µm. J, Representative immunofluorescent images showing the (left) VTA axon terminals of mCherry-labeled GABAergic neurons which are activated by CFA injection from the LH and the recording fiber in (right) NAc. Scale bar, 200 µm. K, L, Acute real-time optogenetic activation of the pain-activated LHGABA→VTA neurons in the VTA decreased (K) PWLs and (L) 50% PWTs in mice. n Saline + hChR2 = 13; n CFA + hChR2 = 7. PWLs: group, F(2,36) = 73.85, p < 0.0001; Saline + hChR2 versus CFA + hChR2, laser on p < 0.0001. 50% PWTs: group, F(2,36) = 19.05, p < 0.0001; Saline + hChR2 versus CFA + hChR2, laser on p < 0.0001. M, Representative false heatmap showing extracellular NAc DA signaling change over time with or without optogenetic activation of LHGABA→VTA neurons in the VTA. n = 21 trials from 6 mice. N, Average quantification curve showing extracellular NAc DA signaling change over time. n = 21 trials from 6 mice in CFA + hChR2 group; n = 18 trials from 6 mice in mCherry group. O, AUC of NAc DA signaling over three periods: BL, blue laser on, and laser off. n = 20 trials from 6 mice in CFA + hChR2 group; n = 18 trials from 6 mice in mCherry group. CFA + hChR2: group, F(1.669,33.38) = 27.45, p < 0.0001; prestim versus stim, p < 0.0001. mCherry: group, F(1.540,27.72) = 0.7595, p = 0.4452; prestim versus stim, p = 0.5714. ****p < 0.0001, n.s, no significance. Error bars indicate SEM BL, baseline; LH, lateral hypothalamic area; VTA, ventral tegmental area; f, fornix; ic, internal capsule; ml, medial lemniscus.
To build the direct link between LHGABA→VTA circuit activation and accumbal DA release in pain regulation, we used the Tet-off system to express hChR2 in LHGABA neurons activated by CFA inflammation pain and then detected the evoked accumbal DA release when its terminals were optically activated in the VTA (Fig. 5G). To achieve this, the AAV-cFos-tTA-NLS-FLAG and AAV-Tre3g-DIO-hChR2-mCherry were injected into the LH of vGAT-Cre mice to drive hChR2 expression in the LHGABA when doxycycline (Dox) was absent (Fig. 5H). To avoid nonpain-related hChR2 expression, 1% Dox was added to the drinking water 3 d before virus injection and only removed during the first 3 d after CFA injection. Two weeks after the second administration of 1% Dox, we detected hChR2-mCherry expression in the LH, indicating that LHGABA neurons were activated by CFA injection (Fig. 5I). We activated the terminals of these pain-related LHGABA neurons via the optical fibers in the VTA, and simultaneously recorded DA release in the NAc (Fig. 5J). Consistent with the activation of the LHGABA→NAc circuit described above, optogenetic activation of pain-activated neurons in the LHGABA→VTA circuit also decreased the PWLs and 50% PWTs in naive mice (Fig. 5K,L), and induced significant DA release in the NAc (Fig. 5M–O). These results suggested that the LHGABA→VTA circuit was involved in pain sensation regulation by modulating accumbal DA release.
Together with these cell type-specific and pathway-specific gain-of-function behavioral outcomes and a recent fast-scan cyclic voltammetry recording (Nieh et al., 2016), these real-time neurotransmitter release experiments in free-moving mice built a functional connection between the LHGABA→VTA and VTADA→NAc circuits in mediating pain sensation regulation.
LHGABA→VTA projection disinhibits mesolimbic BDNF signaling and regulates pain sensation
The anatomic and functional studies described above indicate that GABAergic inputs from the LH target VTA GABAergic neurons, disinhibit the mesolimbic DA circuit and regulate pain behaviors. Our recent studies suggested that mesolimbic BDNF signaling is a critical nociceptive effector, and the release of BDNF regulated by VTA→NAc circuitry is sufficient to modulate pain behavior in a neuropathic pain model (H. Zhang et al., 2017a). Thus, we designed optogenetics-assisted molecular and pharmacological experiments to investigate the possible role and mechanisms of LHGABA→VTA projection regulated mesolimbic BDNF signaling in pain sensation.
We first tested whether repeated optogenetic inhibition of the LHGABA→VTA projection can attenuate the increase in mesolimbic BDNF expression induced by neuropathic pain (Fig. 6A). As previously reported, CCI surgery increased BDNF protein expression in the VTA and NAc, and these changes were reversed by repeated optogenetic inhibition of LHGABA→VTA projection (Fig. 6B). Correspondingly, repeated activation of LHGABA→VTA projection significantly increased BDNF protein expression in the VTA and NAc (Fig. 6C,D), a phenomenon observed in mice with neuropathic pain. In addition, the bdnf mRNA level increased in the VTA but not in the NAc (Fig. 6E), suggesting that the increased BDNF in the NAc was most likely released by terminals from other upstream regions. Previous studies indicated that the increase of protein expression in the NAc was because of elevated BDNF synthesis in the VTA and its downstream release in the NAc, which was controlled by the activity of putative NAc-projecting VTA dopamine neurons (Krishnan et al., 2007; H. Zhang et al., 2017a). Thus, we next investigated whether the activity of VTADA neurons contributes to BDNF transcription and release in the mesolimbic system. To inhibit the activation of VTA neurons induced by the activation of LHGABA→VTA circuit, we implanted a bilateral VTA cannula for local infusion of an Ih blocker (DK-AH 269, 1 μg/0.3 μl), which can inhibit VTADA neuronal firing (Fig. 6F). VTA injection of DK-AH 269 1 h before optical stimulation prevented the mechanical and thermal hyperpathia induced by the optogenetic activation of LHGABA→VTA projection (Fig. 6G). The increases in BDNF protein in the VTA and NAc (Fig. 6H) and bdnf mRNA levels in the VTA (Fig. 6I) were also attenuated by VTA DK-AH 269 infusion. These results indicate that the changes in mesolimbic BDNF expression and its accumbal release depend on VTADA neuronal activity.
LHGABA→VTA circuit disinhibits the mesolimbic BDNF signaling to mediate pain sensation. A, Schematics showing viral surgeries and stimulation pattern for optogenetic inhibition of the LHGABA→VTA circuit to examine mesolimbic BDNF expression. B, Representative blots and quantitative data showing (left) VTA and (right) NAc BDNF protein expression in mice from sham, CCI, and CCI-eNpHR3.0 groups. n = 4 mice/group. VTA: group, F(2,9) = 22.88, p = 0.0003; Sham versus CCI, p = 0.0003; Sham versus eNpHR3.0 + CCI, p = 0.3882; CCI versus eNpHR3.0 + CCI, p = 0.0018. NAc: group, F(2,9) = 5.838, p = 0.0237; Sham versus CCI, p = 0.0369; Sham versus eNpHR3.0 + CCI, p = 0.9983; CCI versus eNpHR3.0 + CCI, p = 0.0403. C, Schematics showing viral surgeries and stimulation pattern for optogenetic activation of the LHGABA→VTA circuit to examine mesolimbic BDNF expression. D, Representative blots and quantitative data showing (left) VTA and (right) NAc BDNF protein expression in mice from sham, CCI, and hChR2 groups. n = 4 mice/group. VTA: group, F(2,9) = 10.56, p = 0.0044; Sham versus CCI p = 0.0096; Sham versus hChR2, p = 0.0070; CCI versus hChR2, p = 0.9749. NAc: group, F(2,9) = 11.14, p = 0.0037; Sham versus CCI, p = 0.0129; Sham versus hChR2, p = 0.0044; CCI versus hChR2, p = 0.7483. E, Quantitative data showing (left) VTA and (right) NAc bdnf mRNA expression in mice from sham, CCI, and hChR2 groups. n = 4 mice/group. VTA: group, F(2,9) = 11.36, p = 0.0034; Sham versus CCI p = 0.0038; Sham versus hChR2, p = 0.00,145; CCI versus hChR2, p = 0.6415. NAc: group, F(2,9) = 0.02,251, p = 0.9778; Sham versus CCI, p = 0.9998; Sham versus hChR2, p = 0.9798; CCI versus hChR2, p = 0.9836. F, Schematics showing viral surgeries and stimulation pattern for optogenetic activation of the LHGABA→VTA circuit to examine mesolimbic BDNF expression, the intra-VTA microinjections of DK-AH 269 or Aetinomycin D (ACD) were 1 h before the optical activation. G, Intra-VTA microinjections of DK-AH 269 reversed the decreased (left) PWLs and (right) 50% PWTs in CCI mice induced by repeated optogenetic activation of the LHGABA→VTA. nmCherry + Saline = 8; nmCherry + DK-AH269 = 8; nhChR2 + Saline = 7; nhChR2 + DK-AH269 = 8. PWLs: group, F(6,54) = 7.769, p < 0.0001; hChR2 + Saline versus hChR2 +DK-AH 269, During-stim p = 0.0002. 50% PWTs: group, F(6,54) = 4.050, p = 0.002; hChR2 + Saline versus hChR2 +DK-AH 269, During-stim p = 0.0018. H, Representative blots and quantitative data showing (left) VTA and (right) NAc BDNF protein expression in mice from mCherry + Saline, mCherry + DK-AH269, hChR2 + Saline, and hChR2 + DK-AH269 group. n = 4 mice/group. VTA: group, F(3,12) = 6.957, p = 0.0058; mCherry + Saline versus hChR2 + Saline, p = 0.0254; hChR2 + Saline versus hChR2 + DK-AH269, p = 0.0096; NAc: group, F(3,12) = 6.42, p = 0.0077; mCherry + Saline versus hChR2 + Saline, p = 0.0192; hChR2 + Saline versus hChR2 + DK-AH269, p = 0.0105. I, Quantitative data showing (left) VTA and (right) NAc bdnf mRNA expression in mice from mCherry + Saline, mCherry + DK-AH269, hChR2 + Saline, and hChR2 + DK-AH269 group. n = 4 mice/group. VTA: group, F(3,12) = 5.786, p = 0.011; mCherry + Saline versus hChR2 + Saline, p = 0.0278; hChR2 + Saline versus hChR2 + DK-AH269, p = 0.0190; NAc: group, F(3,12) = 0.1862, p = 0.9038; hChR2 + Saline versus hChR2 + DK-AH269, p > 0.9999. J, Intra-VTA microinjections of ACD reversed the decreased (left) PWLs and (right) 50% PWTs in CCI mice induced by repeated optogenetic activation of the LHGABA→VTA circuit. nmCherry + Saline = 8; nmCherry + ACD = 8; nhChR2 + Saline = 7; nhChR2 + ACD = 7. PWLs: group, F(9,78) = 7.769, p < 0.0001; hChR2 + Saline versus hChR2 +ACD, During-stim p = 0.0002. 50% PWTs: group, F(9,78) = 2.565, p < 0.0001; hChR2 + Saline versus hChR2 +ACD, During-stim p = 0.0068. K, Representative blots and quantitative data showing (left) VTA and (right) NAc BDNF protein expression in mice from mCherry + Saline, mCherry + ACD, hChR2 + Saline, and hChR2 + ACD group. n = 4 mice/group. VTA: group, F(3,12) = 6.532, p = 0.0072; mCherry + Saline versus hChR2 + Saline, p = 0.0129; hChR2 + Saline versus hChR2 + ACD, p = 0.0127; NAc: group, F(3,12) = 13.87, p = 0.0003; mCherry + Saline versus hChR2 + Saline, p = 0.019; hChR2 + Saline versus hChR2 + ACD p = 0.0009. L Quantitative data showing (left) VTA and (right) NAc bdnf mRNA expression in mice from mCherry + Saline, mCherry + ACD, hChR2 + Saline, and hChR2 + ACD group. n = 4 mice/group. VTA: group, F(3,12) = 5.786, p = 0.011; mCherry + Saline versus hChR2 + Saline, p = 0.0278; hChR2 + Saline versus hChR2 + ACD9, p = 0.0190; NAc: group, F(3,12) = 0.1862, p = 0.9038; hChR2 + Saline versus hChR2 + ACD, p >0.9999. M, Schematic shows strategy for optogenetic activation of LHGABA→VTA circuit with AAV-DIO-ChrimsonR, cannula implantation, and experimental timeline. N, O, Intra-NAc microinjection of ANA 12 (1 μg/0.3 μl/side) reversed the decrease of (N) PWLs and (O) 50% PWTs evoked by LHGABA→VTA circuit activation. nmCherry + Saline = 8, nChrimsonR + Saline = 6, nChrimsonR + ANA-12 = 4. PWLs: group, F(2,16) = 0.5338, p = 0.5965; mCherry + Saline versus ChrimsonR + Saline, laser on p = 0.0002; ChrimsonR + Saline versus ChrimsonR + ANA-12, laser on p < 0.0001. 50% PWTs: group, F(2,16) = 2.534, p = 0.1106; mCherry + Saline versus ChrimsonR + Saline, laser on p = 0.0280; ChrimsonR + Saline versus ChrimsonR + ANA-12, laser on p = 0.0006. P, Q, Optogenetic activation of the LHGABA→VTA circuit alone or combined with local infusion of ANA-12 in the NAc did not affect the (P) velocity and (Q) traveled distance in the open field test. nmCherry + Saline = 8, nChrimsonR + Saline = 6, nChrimsonR + ANA-12 = 4. Velocity: group, F(2,15) = 2.604, p = 0.1070. ChrimsonR + Saline versus ChrimsonR + ANA-12 p = 0.1625. Distance traveled: group, F(2,15) = 2.377, p = 0.1268. ChrimsonR + Saline versus ChrimsonR + ANA-12 p = 0.1339. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n.s, no significance. Data were analyzed by (B, D, E, H, I, K, L, P, Q) one-way ANOVA with Tukey's post hoc test or (G, J, N, O) two-way ANOVA with Tukey's post hoc tests. Error bars indicate SEM LH, lateral hypothalamic area; VTA, ventral tegmental area; NAc, nucleus accumbens; BL, baseline; ACD, actinomycin D.
To further confirm the necessity of mesolimbic BDNF signaling in the pronociceptive effect of activating LHGABA→VTA projection, we injected a general inhibitor of transcription, actinomycin D (ACD; 0.1 μg/0.3 μl) into the VTA via the bilateral cannula 1 h before optogenetic stimulation to block BDNF transcription (Fig. 6F). We found that the injection of ACD into the VTA prevented the decreases in PWLs and 50% PWTs induced by the repeated activation of LHGABA→VTA projection (Fig. 6J). In the ACD-injected mice, BDNF protein levels in the VTA and NAc remained unchanged after repeated activation of LHGABA→VTA projection (Fig. 6K), as did bdnf mRNA levels in the VTA (Fig. 6L). Furthermore, in vGAT-Cre mice, we expressed AAV-DIO-ChrimsonR-mCherry in LHGABA→VTA neurons for optogenetic activation and implanted a bilateral NAc cannula for local infusion of ANA-12, an antagonist of the BNDF receptor tyrosine kinase B (TrkB; Fig. 6M). Consistent with our results above, activation of ChrimsonR-expressing LHGABA→VTA projection with a yellow laser markedly decreased the PWLs and 50% PWTs of naive mice, effects that could be entirely prevented by pretreatment with an intra-NAc microinjection of ANA 12 (1 μg/0.3 μl/side) 1 h before laser treatment (Fig. 6N,O). Moreover, these treatments did not affect the animal's locomotor activity (Fig. 6P,Q).
Together, these data identify the LHGABA→VTAGABA→VTADA→NAc pathway as a novel circuit for pain sensation, in which LH GABAergic inputs disinhibit mesolimbic BDNF signaling by targeting VTAGABA neurons (Fig. 7). The present study provides a better understanding of the role of the mesolimbic DA system in regulating pain sensation.
Cell type-specific functional connection of the LH, VTA, and NAc in pain processing based on the LHGABA→VTA circuit. Our anatomic and functional experiments indicate that LHGABA→VTA projecting neurons disinhibit VTA→NAc projection-specific DA neurons by targeting the VTA GABAergic neurons, and BDNF signaling to mediate pain sensation in mice with physiological and pathologic pain. LH, lateral hypothalamic area; VTA, ventral tegmental area; NAc, nucleus accumbens.
Discussion
LH and VTA have been implicated in pain sensation at neural circuitry and molecular levels (Ezzatpanah et al., 2016; Torruella-Suárez and McElligott, 2020). However, whether and how the cell type-based synaptic connections between the two nuclei regulate pain sensation is not yet clear. Using cell type-specific and projection-specific viral tracing, optogenetics, in vivo calcium and neurotransmitter imaging and traditional neuropharmacological approaches, here we have identified that the LHGABA→VTA projection preferentially targets VTAGABA neurons which disinhibit mesolimbic DA release and BDNF signaling to mediate pain sensation (Fig. 7). We recently reported the functional role of the putative VTA→NAc DAergic projection in mediating pain sensation and identified projection-specific BDNF signaling as a molecular effector in the process (H. Zhang et al., 2017a). The present study provides a novel circuit and mechanism by which the mesolimbic DA system is regulated in neuropathic pain.
The neurobiological functions of a neural structure are affected by both local intrinsic molecular mechanisms and presynaptic inputs from upstream brain regions. In the present study, we identified LHGABA neurons as a source of presynaptic inputs in the VTA and demonstrated the bidirectional regulating capabilities of this circuitry in mechanical and thermal pain sensation in naive mice. Notably, optogenetic inhibition of the presynaptic LH GABAergic terminals in the VTA induced a potent analgesic effect in behavioral tests. Consistent with the behavioral data, cell type-specific labeling and staining with c-Fos verified the elevated activity of VTA-projecting LHGABA neurons in mice with CCI-induced neuropathic pain and CFA-induced inflammatory pain. As seen in the naive mice, optogenetic inhibition of GABAergic projection was analgesic and sufficient to alleviate the pain symptoms in mice with pathologic pain. These data demonstrated an essential role of the LHGABA→VTA projection as a previously unknown pathway for pain regulation. The LH also contains glutamatergic and neuropeptidergic projections to the VTA. Future studies are needed to dissect their role in mediating pain sensation.
Neural projections from one brain region target their postsynaptic neurons in another and function as a neural circuit to conduct behavior. The cell type of postsynaptic neurons is another determinant of neural circuit functions. Our anterograde and retrograde viral labeling strategies revealed that LH GABAergic projections preferred to form synaptic connections with VTAGABA neurons, and the presynaptic inputs that VTAGABA neurons received from the LH were mainly GABAergic. Functionally, optogenetic activation of LH GABAergic inputs in the VTA decreased the activity of GABAergic neurons in our in vivo calcium-imaging test. These findings indicate a potential disinhibiting effect of LHGABA neurons on other VTA cell types via VTA GABA neurons. Indeed, increased calcium activity of VTADA neurons was observed when VTA-projecting LHGABA neurons were optogenetically activated. Given that pain-related hyperactivity of mesolimbic DAergic neurons was reported (H. Zhang et al., 2017a), the present study suggests an LH GABAergic modulation of pain by targeting VTAGABA neurons to disinhibit VTADA neurons. This notion was further validated by the increased NAc DA release on optogenetic activation of LHGABA→VTA projection in our in vivo neurotransmitter imaging experiment in free-moving mice. It must be noted that the VTA receives multiple inputs that directly or indirectly affect the neuronal activity of VTA projecting dopamine neurons, including GABAergic innervations from the LH, which have been implicated in behavioral regulation. The present study provided direct anatomic and functional evidence to support that the LH GABAergic input acts as a neural circuit to bidirectionally regulate animals' responses to thermal and mechanical stimuli under physiological and pathologic states. However, excluding the contributions of synaptic inputs from other brain regions is impossible. Even so, in the present study, mere excitation of axonal terminals from inflammatory pain-activated LHGABA neurons in the VTA was sufficient to induce hyperpathia in naive mice, and DA release in the NAc, providing critical evidence that the LHGABA→VTA circuit is one fundamental driver of pain-related DA release in the NAc.
In general, the role of VTA→NAc circuits in pain relief versus pain maintenance is still controversial (Kuner and Kuner, 2021). Traditional views and multiple studies support that pain decreases the activity of VTADA neurons and that pain relief is signaled as a reward, which leads to activation of the midbrain DA system (Ren et al., 2016; Watanabe et al., 2018; Markovic et al., 2021). In contrast, in vivo fiber photometry recordings revealed that acute pain stimuli rapidly increase the activity of VTADA neurons (Moriya et al., 2018), and other studies including ours observed the enhanced activity of VTA→NAc circuits in neuropathic pain animals (Sagheddu et al., 2015; H. Zhang et al., 2017a). These conflicting results may be explained by variations in the models of neuropathy used, the types and methods of the behavioral readouts, the subnuclei involved and the timing of the postinjury analyses (Kuner and Kuner, 2021). Recent studies highlighting the anatomic and functional heterogeneity of DAergic neurons in the VTA→NAc circuits may bridge the abovementioned divergence. The subregions of the NAc (medial shell, lateral shell, and core) receive DAergic innervation from different VTA areas (Beier et al., 2015), and play different roles in the early and late phases of chronic pain (Makary et al., 2020). In addition, neuropathic pain increases DA release in the medial shell of the NAc but decreases DA release in the NAc core (Ren et al., 2021). In the present study, our recording fiber was located in the medial shell and increased DA release induced by the LHGABA→VTA circuit was observed, which supports the idea that activation of VTA DAergic projection to NAc shell is related to pain experience. Although DA release in the NAc core was not directly examined in the present study, the lower DA release in the NAc core during activation of LHGABA→VTA circuit is feasible since LHGABA neurons also seed direct inhibitory projection to VTADA neurons. Both inhibition and activation of VTADA neurons by upstream afferents were observed in chronic pain animal models (Markovic et al., 2021; L. Zhang et al., 2021), suggesting that presynaptic characteristics are involved in VTADA neuronal heterogeneity. Nevertheless, the disinhibition pathway revealed in our results provides a circuitry explanation for the pain-related hyperactivity of VTADA neurons and the mesolimbic DA system, suggesting the necessity of investigating VTA→NAc circuits in more detail.
Multiple lines of evidence suggest that peripheral and spinal BDNF/TrkB signaling can modulate nociceptive transmission and central sensitization in chronic pain states (Coull et al., 2005; Pezet and McMahon, 2006; Merighi et al., 2008; X. Wang et al., 2009). However, less is known at the supraspinal level, especially in the mesolimbic reward circuitry, of the effect of BDNF/TrkB. In our previous study on chronic neuropathic pain, we discovered that mesolimbic BDNF signaling modulated by VTADA neurons plays a crucial role in the maintenance of pain (H. Zhang et al., 2017a). The present study further expanded this finding by demonstrating a disinhibiting effect of LHGABA→VTA projection on mesolimbic BDNF signaling with cell type-specific and projection-specific optogenetic approaches. As a neuropeptide, compared with the fast neurotransmitter, BDNF employs a slow mode of communication and might require higher frequency and prolonged stimulations to be released (Arrigoni and Saper, 2014). Thus, the acute activation/inhibition of LHGABA→VTA projection may not influence BDNF protein levels in the NAc. Clinical studies indicated that acute noxious stimulation elevated real-time NAc dopamine transmission in healthy volunteers (Scott et al., 2006), and NAc dopamine release was also reported to increase in patients with chronic pain (Ren et al., 2021). Together, we hypothesize that acute pain mediation of the LHGABA→VTA circuit is attributed to DA transmission in the mesolimbic reward circuit, as evidenced by the real-time DA release in the NAc in response to activation of the LHGABA→VTA circuit. Considering the continuous hyperactivity of mesolimbic BDNF signaling in chronic pain (H. Zhang et al., 2017a), optogenetic manipulations were applied to evaluate the effect of LHGABA→VTA activity on NAc BDNF expression. As observed in mice with neuropathic pain, repeated inhibition of LHGABA→VTA projection decreased BDNF protein expression in the VTA and NAc, and repeated activation of LHGABA→VTA projection in naive mice sufficiently increased BDNF protein expression in the two nuclei. However, bdnf mRNA expression increased in the VTA, but not the NAc, suggesting that BDNF in the NAc is released by innervating terminals. Previous studies have demonstrated that BDNF release is controlled by local or upstream neuronal activity (Balkowiec and Katz, 2002; Krishnan et al., 2007; H. Zhang et al., 2017a). The present study indicated that activation of VTADA neurons is required for the increased protein and mRNA level of BDNF in the VTA→NAc circuit induced by activation of LHGABA→VTA projection. In addition, pain-like behaviors induced by activating LHGABA→VTA projection can be completely prevented by pretreatment with the TrkB receptor antagonist ANA-12 in the NAc and blocking BDNF transcription in the VTA attenuated the effect induced by the repeated activation of LHGABA→VTA projection. Together with our transsynaptic tracing, fiber photometry calcium, and neurotransmitter release studies, these results support the hypothesis that LH GABAergic neurons target VTA GABAergic neuron to disinhibit the neuronal activity of VTADA neurons, leading to increased BDNF synthesis in the VTA and thus increased BDNF release in the NAc, and providing extensive evidence that mesolimbic BDNF signaling is a critical nociceptive effector of pathologic pain.
In conclusion, this study identified the LHGABA→VTA projection as a novel neural circuit for pain regulation, possibly by targeting VTAGABA neurons to disinhibit the mesolimbic pathway-specific DA release and BDNF signaling. Thus, the present research reveals a novel neural circuitry and its underlying molecular mechanisms for pain and analgesia in the mesolimbic reward system, which may provide helpful information for developing new strategies to treat chronic pain.
Footnotes
This work was supported by the National Key R&D Program of China, the Sci-Tech Innovation 2030 Major Project 2021ZD0203100 (to J.-L.C); National Natural Science Foundation of China Grants 82130033, 82293641, 31970937, 82271255, 82101315, and 82271263; Jiangsu Province Innovative and Entrepreneurial Team Program; the Key Project of Nature Science Foundation of Jiangsu Education Department Grant 22KJA320006; the Innovation and Entrepreneurship Program of Xuzhou Medical University Grant 2021CXFUZX002; China Postdoctoral Science Foundation Grants 2022M710771 and 2022M722676; and Postgraduate Research & Practice Innovation Program of Jiangsu Province Grants KYCX22-2934 and 20-2455.We thank Shi-Ya Zou, M.D., Bing-Qian Fan, M.D., and Li Yang, M.D. for technical support.
The authors declare no competing financial interests.
- Correspondence should be addressed to Jun-Li Cao at caojl0310{at}aliyun.com or Hongxing Zhang at hongxing.zhang{at}xzhmu.edu.cn
