Contralateral electroacupuncture modulates the transmission of nociceptive information in the spinal dorsal horn via GABAergic neurons in the rostral ventromedial medulla

Abstract

Recent studies have shown that electroacupuncture (EA) can exert analgesic effects by modulating wide dynamic range (WDR) neurons in the spinal dorsal horn; however, how EA regulates WDR neurons to inhibit pain signals remains unclear. In this study, we identified a key brain region, rostral ventromedial medulla (RVM), involved in the modulation of spinal WDR neurons. Subsequently, we found evidence suggesting that GABAergic neurons in the RVM may mediate the transmission of nociceptive and non-nociceptive signals from the spinal cord, which underlies EA analgesia. The results demonstrated that the activation of RVM GABAergic neurons enhanced the excitability of WDR neurons, thereby facilitating the transmission of peripheral sensory signals within the spinal cord. Contralateral EA at 2 mA effectively suppressed WDR neuron activity and elevated pain thresholds in rats modeled with complete Freund’s adjuvant (CFA). Notably, heightened activity of RVM GABAergic neurons mitigated the inhibitory effects of EA on WDR neurons and reduced EA-induced analgesia. These findings suggest that EA may attenuate WDR neuronal activity by modulating RVM GABAergic neurons, thereby inhibiting nociceptive transmission. This study highlights the potential involvement of RVM GABAergic neurons and identifies the efficacy of high-intensity, contralateral EA stimulation in producing analgesia.

Introduction

Chronic pain is a prevalent condition, with over 30% of patients globally requiring treatment for chronic pain disorders.^1,2 Recurrent pain attacks contribute to peripheral and central sensitization, and initiate a series of autonomic disturbances and emotional disorders. Although opioids, non-steroidal anti-inflammatory drugs (NSAIDs), and anticonvulsants (e.g. gabapentinoids) are commonly used for pain relief, their long-term use in chronic pain management is often limited by risks such as gastrointestinal complications, renal toxicity, tolerance, dependence, and addiction.^3–8 Consequently, there is an urgent need to identify therapies for chronic pain management that are safe, effective, and have minimal adverse effects. Electroacupuncture (EA) analgesia, as a non-pharmacological intervention, offers unique advantages in the treatment of chronic pain due to its distinctive neuromodulatory mechanisms and favorable safety profile. Clinical evidence suggests that acupuncture/EA achieves an effectiveness rate of up to 50% in managing chronic pain.^9,10

In recent years, growing evidence has highlighted that wide dynamic range (WDR) neurons of the spinal dorsal horn were closely associated with the analgesic mechanisms of EA. ¹¹ The WDR neurons are located in the deeper laminae of the spinal dorsal horn; they constitute a class of sensory neurons proficient in receiving and integrating peripheral sensory inputs from Aβ, Aδ, and C fibers. These neurons subsequently convey the integrated signals to the cerebral cortex, thereby facilitating the perception of pain.^12–20 The firing frequency of WDR neurons corresponds to the intensity of sensory input. In models of chronic pain, there is a significant elevation in both spontaneous and evoked firing rates of these neurons.^21–25 Contemporary acupuncture analgesia theory suggests that low-intensity EA stimulation activates thickly myelinated Aβ fibers, which enhance both pre-synaptic and post-synaptic inhibition of WDR neurons by exciting inhibitory interneurons in the spinal dorsal horn, thus producing localized analgesic effects.^11,26–28 Conversely, high-intensity EA, which is capable of activating C fibers, exhibits superior analgesic efficacy through more pronounced suppression of spinal nociceptive transmission, as evidenced by reduced abnormal firing in WDR neurons.^10,28,29 However, the exact neural pathways underlying the effects of high-intensity EA have yet to be fully elucidated.

Clinical observations suggest that spinal cord transection in patients with chronic pain frequently results in the recurrence of pain within a matter of weeks, highlighting the essential involvement of spinal upstream pathways and higher neural centers in the encoding of pain signals. ³⁰ The rostral ventromedial medulla (RVM) is recognized as a pivotal hub for pain modulation. Research has demonstrated that electrical stimulation of the RVM produces frequency-dependent bidirectional regulatory effects. ³¹ Specifically, acupuncture at a frequency of 2 Hz has been shown to significantly inhibit the activity of ON-cells while simultaneously enhancing the firing frequency of OFF-cells within the RVM.^32,33 This bidirectional modulation is posited to contribute to the sustained analgesic efficacy of acupuncture.^34,35

Although research has demonstrated the modulatory effects of EA on the RVM, evidence concerning a regulatory relationship between the RVM and spinal WDR neurons remains sparse. Moreover, the specific neuronal subtypes within the RVM that mediate the effects of EA have yet to be identified. Recent studies suggest that GABAergic neurons may represent a subpopulation of ON-cells, with long-range-projecting ON-cells potentially facilitating spinal pain transmission through the inhibition of enkephalinergic and GABAergic interneurons.^36,37 Additionally, observations of glutamate- and GABA-mediated modulation of WDR neurons in the spinal dorsal horn imply that GABAergic neurons in the RVM may exert regulatory control over dorsal horn WDR neurons. ²⁶

In this study, we investigated the upstream mechanisms through which high-intensity contralateral EA exerts its analgesic effects, with a focus on its suppression of WDR neurons. We specifically elucidated the critical regulatory role of GABAergic neurons in the RVM in this process. Our findings demonstrate that EA inhibits nociceptive transmission in the spinal dorsal horn by modulating the activity of WDR neurons via these RVM GABAergic neurons. Collectively, this work advances our understanding of the neural circuitry underlying EA-induced analgesia and refines the theoretical basis for optimizing chronic pain management strategies.

Materials and methods

Animals

Male Sprague-Dawley rats (10–12-week-old adult rats, weighing 220–250 g, or P12–P15-day-old neonatal rats) were used for experiments. All procedures described were approved by the Ethics Committee of the Institutional Animal Welfare and Use Committee of the Acupuncture and Moxibustion Institute of China Academy of Chinese Medical Sciences. All experimental rats were housed in an environment maintained at 22 ± 2°C with a humidity of 50% ± 10%, had ad libitum access to water and food, and were subjected to a 12-h light/dark cycle. A total of 94 animals were used in this study, with 40 for behavioral experiments, 51 for spinal cord electrophysiological recordings, and 3 for neuronal labeling or tracing.

Anesthesia rationale

Different anesthetics were used according to the specific experimental procedure to optimize animal welfare and data quality: Isoflurane (2%) was used for brief surgeries (modeling, EA, fiber implantation) due to rapid induction/recovery. Urethane (10%, i.p. 1.5 mL/100 g) was used for prolonged in vivo electrophysiology due to its stable, long-lasting plane of anesthesia with minimal depression of neural responses. Tribromoethanol (1.25%) was used for stereotaxic virus injection surgery as it provides suitable duration and depth for aseptic procedures.

Establishment of a chronic pain model

Under 2% isoflurane anesthesia (0.5 L/min), a muscle inflammatory pain model was established in the gastrocnemius muscle of the left hind limb of rats. A needle was carefully inserted approximately 2 mm above the insertion point of the gastrocnemius muscle, reaching a depth of about 5–7 mm. Subsequently, 200 μL of complete Freund’s adjuvant (CFA) was slowly injected into the muscle. After the injection, the syringe was held in place for 1 min to ensure proper distribution of the CFA before being slowly withdrawn. Behavioral testing or electrophysiological recordings commenced after a post-injection stabilization period of 3 days.

EA treatment protocol

Behavioral experiments assessed the therapeutic effect of repeated EA, interventions commenced on Day 3 after CFA injection and were administered once daily for three consecutive days (i.e. Days 3, 4, and 5). EA treatment was conducted under 2% isoflurane anesthesia (0.5 L/min). A pair of sterile acupuncture needles (0.25 mm × 13 mm, Beijing Zhongyan Taihe Medical Instruments Co., Ltd., China) were inserted to a depth of 5 mm at the right ST36 acupoint, located lateral to the knee joint approximately 5 mm inferior to the fibula head. Contralateral stimulation was selected to specifically examine the supraspinal descending analgesic pathway, thereby minimizing potential confounding effects from segmental mechanisms that might be more pronounced with ipsilateral stimulation. ¹⁰ The needles were connected to an eight-channel programmable stimulator (STG4008, Reutlingen, Germany), which delivered a constant-current output with an intensity of 2 mA, an alternating frequency of 2/15 Hz, and a pulse width of 0.5 ms. The 2/15 Hz alternating frequency paradigm was chosen based on clinical evidence supporting its efficacy in chronic pain conditions involving central sensitization.^38–40 This paradigm is also proposed to synergistically engage multiple endogenous analgesic systems while reducing the risk of neural adaptation compared with single-frequency stimulation.^41–43 In behavioral experiments, each EA session lasted 20 min. For in vivo electrophysiological recordings designed to capture acute neuronal responses, a single 3-min EA session was applied during the recording period, which took place between Days 3 and 5 after CFA injection.

Nociceptive behavioral tests

All rats were subjected to a 60 min adaptation period with each testing apparatus for three consecutive days prior to any experimental manipulation (e.g. CFA injection or virus surgery). All experiments were conducted in a blinded manner, and no subjects or data were excluded for any reason during the experimental process. For mechanical pain threshold testing, rats were placed in individual polyvinyl chloride boxes with a metal mesh floor, and von Frey filaments ranging from 0.4 to 26 g were applied to the surface of the left hind paw. The up-down method was employed to determine the 50% withdrawal threshold of the subjects. ⁴⁴ For the spontaneous pain, bilateral weight-bearing test, a non-invasive pain-sensitive bilateral balance device was utilized, with pressure sensors that reflect the force exerted on both hind limbs when the rat is standing, using differences in weight distribution between the two feet as an indicator of pain. Each rat was measured three times, and the average value was taken as the measured value. ⁴⁵

In vivo electrophysiological recordings

In vivo electrophysiological recordings were conducted under anesthesia with 10% urethane (i.p. 1.5 mL/100 g). Recordings were performed on either naïve rats or on CFA-treated rats between Days 3 and 5 post-injections, as specified in the respective results sections. Physiological homeostasis was maintained and monitored throughout the experimental process. A laminectomy was performed to expose the lumbar enlargement, and recording electrode was vertically inserted into the dorsal horn of the L4-6 spinal segments using a computer-controlled micromanipulator, with a depth of approximately 800–1400 μm from the spinal cord surface, 0.3 mm lateral to the central blood vessel, and the reference electrode was placed in the muscle layer near the recording area. Signals were amplified by a preamplifier and introduced into a multichannel data acquisition system (bandwidth: 250 Hz–5 kHz), and were stored and analyzed using Spike2 software (CED, Cambridge Electronic Design). ⁴⁶ The WDR neurons in the spinal dorsal horn were identified based on their responses to mechanical stimuli of varying intensities. During electrophysiological recording of dorsal horn neurons, a rubber blunt probe (diameter 0.5 cm) was used to apply mechanical pressure stimuli of 60 g and 200 g to the local pain source, serving as non-nociceptive and nociceptive stimuli to evoke WDR firing, respectively, with each stimulus lasting 20 s. ²⁷

AAV virus injection

For experiments involving optogenetic manipulation, AAV viruses were injected into naïve rats approximately 4–5 weeks prior to the planned behavioral or electrophysiological endpoint. This allowed sufficient time for viral expression, subsequent optical fiber implantation, and recovery before CFA modeling and testing. All viruses used in the experiments were purchased from Wuhan Privy Brain Science and Technology (Wuhan Privy Brain Science and Technology Co., Ltd., Wuhan, China). For virus injection, rats were anesthetized with 1.25% tribromoethanol and placed in a stereotaxic apparatus. A small hole was drilled into the skull to facilitate the injection of needles into RVM (0.0 M/L; −11.28 A/P; −10.3 D/V) as per the Paxinos and Watson Atlas. ³⁶ Viral solution (120 nL per rat, 3.64 × 10¹² vg/mL) was injected into the RVM at a rate of 30 nL/min. For the spinal cord, virus injections were made into the intervertebral foramina of the L4–L6 segments, with a total of four injection sites, each receiving 30 nL of viral solution. These injection sites were located 0.3 mm lateral to the spinal midline, with a depth of 0.8–1.2 mm. The Hamilton micro-syringe was removed 15 min after microinjection was completed, and the skin incision was closed with a suture. Animals were placed on a heating pad and monitored until fully awakened. Rats were used for experiments 3 weeks after the injection, ensuring stable transgene expression before initiating any pain modeling or experimental interventions.

Optical fiber implantation and optogenetics manipulation

For rats undergoing optogenetic manipulation, after allowing 3 weeks for sufficient AAV virus expression following injection, optical fibers were implanted into the RVM. An optical fiber stub, encased in a 1.25 mm ceramic ferrule (0.37 NA, 200 μm diameter, Qian Aoxingke), was subsequently implanted into the RVM and secured using dental cement. Rats were then allowed to recover for 1 week following fiber implantation before proceeding with any subsequent procedures (e.g. CFA modeling and behavioral or electrophysiological testing). Thus, the final optogenetic experiments were conducted approximately 4 weeks after the initial virus injection.

In the context of in vivo electrophysiological experiments, variations in the firing frequency of WDR neurons, elicited by mechanical stimuli of differing intensities, were measured during both baseline and light stimulation phases. The laser was similarly activated 5 s before the initiation of the test and sustained throughout the electrophysiological recording period.

During behavioral experiments, the 50% withdrawal threshold and left hind limb weight-bearing capacity of the subjects were evaluated at baseline, during light stimulation, and post-light stimulation, with a minimum interval of 20 min between assessments. The laser was activated 5 s prior to the commencement of each test and remained active throughout the behavioral testing period. In experiments examining the optogenetic activation of GABAergic neurons within the RVM to mitigate the effects of EA, the firing frequency of WDR neurons elicited by mechanical stimuli of varying intensities was assessed both pre- and post-intervention. The laser was initiated 5 s prior to the EA intervention, sustained throughout the duration of the intervention, and subsequently deactivated following its conclusion.

Perfusion and sectioning

Rats were anesthetized with 10% urethane (i.p., 1.5 mL/100 g) and perfused. After tissue collection, the brain was fixed in 4% PFA overnight and then transferred to a gradient sucrose solution ranging from 10% to 30% for sucrose dehydration until the tissue block sank to the bottom of the solution. The tissue samples were then frozen and sliced on a sliding microtome (Leica Biosystems) at −20°C. The sections were placed on slides or stored in cryoprotectant at −80°C.

Statistical analyses

For experiments involving optogenetic manipulation, the locations of virus injections and optical fiber implants were verified histologically using a fluorescence microscope (Leitz DMRB, Leica). Only data from animals with accurately positioned virus injection sites and fiber implants were included in the statistical analysis. For in vivo electrophysiological experiments, only complete data recordings were included in the statistical analysis. Data from recordings that were interrupted before the experimental protocol was completed were excluded from the analysis. Statistical analyses were conducted using GraphPad Prism 8 software. The normality of data distribution was evaluated using the Shapiro-Wilk test. For comparisons between two groups, normally distributed data were analyzed with either paired or unpaired t-tests, while non-normally distributed data were assessed using the Mann-Whitney U test or the Wilcoxon signed-rank test. For comparisons involving multiple groups, one-way or two-way analysis of variance (ANOVA) was performed, followed by Tukey’s or Bonferroni’s post hoc tests, as appropriate. Data are presented as the mean ± standard error of the mean (SEM). Sample sizes for each experiment were determined from prior studies and preliminary power analyses to ensure sufficient statistical power (α = 0.05, power = 0.8). p value less than 0.05 was considered significant.

Results

EA alleviated CFA-induced chronic pain

To evaluate the stability of the CFA-induced chronic pain model, pain thresholds were measured daily for five consecutive days following the model’s establishment (Figure 1(a)). By Day 2 post-CFA, the 50% mechanical withdrawal threshold in the experimental rats had significantly decreased and remained at a relatively low level from Days 3 to 5 (Figure 1(d)). Regarding spontaneous pain induced by CFA, the weight-bearing capacity of the left hind limb was significantly reduced on Day 1. Although still diminished on Day 2, weight-bearing began to recover by Day 3 and returned to baseline levels on Days 4 and 5 (Figure 1(e)). These results indicate that CFA-induced mechanical allodynia is sustained and stable over time, whereas CFA-induced spontaneous pain is transient, with recovery initiating as early as Day 3 post-induction.

Figure 1.

EA alleviates CFA-induced chronic pain and inhibits WDR neuronal activity. (a) Experimental timeline of EA treatment for CFA-induced chronic pain. (b) Mechanical pain thresholds on Day 3 post-CFA (***p < 0.001, **p < 0.01, *p < 0.05 vs Ctrl; n = 8; one-way ANOVA, Tukey’s post hoc test). (c) Left hindlimb weight-bearing on Day 3 post-CFA (***p < 0.001, *p < 0.05 vs Ctrl; n.s. p > 0.05; n = 8; one-way ANOVA, Tukey’s post hoc test). (d) Mechanical pain stability in CFA rats and EA effects over three consecutive days (***P < 0.001 vs Ctrl; ###p < 0.001, ##p < 0.01 vs CFA; n = 8; two-way ANOVA, Bonferroni post hoc test). (e) Spontaneous pain stability in CFA rats and EA effects over three consecutive days (***p < 0.001 vs Ctrl; #p < 0.05 vs CFA; n = 8; two-way ANOVA, Bonferroni post hoc test). (f) Experimental timeline for WDR neuronal activity modulation by EA. (g) Representative traces of WDR neuron firing evoked by 60 g non-nociceptive stimulation pre- and post-EA in CFA rats. (h) Frequency changes of 60 g-evoked WDR firing pre- versus post-EA (*p < 0.05; paired t-test; 17 neurons from 6 rats). (i) Paired comparison of 60 g-evoked WDR firing frequency pre- and post-EA. (j) Representative traces of WDR neuron firing evoked by 200 g nociceptive stimulation pre- and post-EA in CFA rats. (k) Frequency changes of 200 g-evoked WDR firing pre- versus post-EA (***p < 0.001; paired t-test; 19 neurons from 6 rats). (l) Paired comparison of 200 g-evoked WDR firing frequency pre- and post-EA.

Following the establishment of the chronic pain model, rats received daily EA treatment for three consecutive days. The stimulation parameters were set at 2 mA, an alternating frequency of 2/15 Hz, a pulse width of 0.5 ms, and a duration of 20 min per session. Prolonged EA treatment significantly alleviated mechanical allodynia from Days 3 to 5 after CFA induction (Figure 1(d)). Given the transient nature of CFA-induced spontaneous pain, the EA-mediated improvement in left hind limb weight-bearing was observed only on Day 3 post-CFA (Figure 1(e)). Therefore, Day 3 after CFA induction was selected as the standardized time point for administering a single EA session in subsequent mechanistic studies. The therapeutic outcomes of this single EA session delivered on Day 3 are shown separately in Figure 1(b) and (c).

EA reduces the nociceptive discharges of WDR neurons in CFA-treated rats

WDR neurons are integral to pain signal processing, and their hyperactivity is a hallmark of pain sensitization. This study supports the inhibitory effect of EA on WDR neurons in a chronic pain model induced by CFA in rats. The findings revealed that in rats with CFA-induced chronic pain, the firing frequency of WDR neurons, when stimulated by a 200 g nociceptive stimulus, was significantly diminished following EA treatment (Figure 1(j)–(l)). Given that WDR neurons function as sensory relay neurons, these results imply that EA may attenuate nociceptive signal transmission at the spinal cord level. Importantly, the firing frequency of WDR neurons in response to a 60 g non-nociceptive stimulus also showed a significant reduction post-EA intervention (Figure 1(g)–(i)). This observation may be attributed to pain sensitization associated with chronic pain, where stimuli that are typically innocuous, such as a 60 g stimulus, acquire nociceptive characteristics under sensitized conditions. In comparison to naïve rats, those treated with CFA demonstrated increased WDR neuronal responsiveness to 60 g tactile stimulation, indicative of mechanical hyperalgesia.^47,48 EA treatment alleviated this sensitized state, thereby restoring WDR neuronal activity to normal levels in response to previously non-noxious stimuli.

Spinal dorsal horn neurons receive the inhibitory projections from RVM

To investigate the upstream mechanisms by which EA inhibits nociceptive discharges of WDR neurons in the spinal dorsal horn, fluorescent gold was injected into the spinal dorsal horn to retrogradely label brain regions projecting to spinal dorsal horn neurons. Fluorescently labeled neuronal cell bodies were identified in the RVM, a critical center for pain modulation (Figure 2(a) and (b)). Previous studies utilizing transgenic mice have demonstrated inhibitory projections from the RVM to spinal dorsal horn interneurons. ³⁷ In this study, we further characterized the specific neuronal subtypes involved in the RVM-spinal dorsal horn pathway in Sprague-Dawley rats. A GABA-specific promoter-driven adeno-associated virus (rAAV-mDlx-mCherry) was injected into the RVM. Concurrently, the retrograde tracer AAVretro-EF1α-EGFP was bilaterally administered to the L4–L6 segments of the spinal dorsal horn (Figure 2(c)). Colocalization of red (GABAergic) and green (projection neuron) fluorescence was observed in the RVM, with GABAergic neurons comprising approximately 75% of the RVM-spinal projection neurons (Figure 2(d)–(f)).

Figure 2.

Spinal dorsal horn neurons receive inhibitory projections from the RVM. (a) Top: Retrograde tracing strategy from the spinal dorsal horn to the RVM. Bottom: Fluorescence image of the viral injection site in the spinal dorsal horn. (b) Retrograde-labeled neuronal cell bodies in the RVM (includes Gigantocellular nucleus ventral part, GiA, and Nucleus raphe magnus, RMg). (c) Top: AAV strategy using mDlx-specific promoters (GABAergic) and retrograde viruses to label from the RVM to the spinal dorsal horn projections. Bottom: Cross-sectional fluorescence of viral injection sites in the spinal dorsal horn. (d) Fluorescence co-localization of GABAergic neurons (red) and the spinal dorsal horn-projecting neurons (green) in the RVM. (e) Enlarged view of boxed region in (d). (f) Quantification of GABAergic/spinal-projecting neuron co-expression in the RVM (n = 3).

RVM GABAergic neurons modulate nociceptive discharges of WDR neuron

To explore the role of RVM GABAergic neurons in modulating pain signal transmission at the spinal cord level, we utilized optogenetic techniques to manipulate the activities of these neurons in naïve rats, in conjunction with in vivo electrophysiological recordings of downstream WDR neurons (Figures 3(a) and 4(a)). Optogenetic activation of RVM GABAergic neurons significantly increased the discharges of WDR neurons in response to 60 g non-nociceptive stimulation (Figure 3(b), (d), (f)), but had no effect on the response of WDR neurons to nociceptive stimulation (Figure 3(c), (d), (g)). In contrast, optogenetic inhibition of RVM GABAergic neurons in naïve rats led to a reduction in WDR firing induced by 200 g nociceptive stimulation (Figure 4(c), (d), (g)), without affecting responses to non-nociceptive stimuli (Figure 4(b), (d), (f)). These results indicate that under physiological conditions (in naïve rats), RVM GABAergic neurons exert bidirectional regulation over WDR neuronal activity in a stimulus-specific manner. In our experiments on naïve rats, activation of RVM GABAergic neurons augmented WDR firing in response to non-nociceptive stimuli, whereas their inhibition diminished WDR activity evoked by nociceptive stimuli, thereby underscoring their potential modulatory role in spinal nociceptive processing.

Figure 3.

Optogenetic activation of RVM GABAergic neurons enhances non-nociceptive-evoked WDR firing. (a) Experimental timeline: Mechanical stimuli (60 g, non-nociceptive; 200 g, nociceptive) applied to the left gastrocnemius receptive field to evoke WDR neuronal firing. (b) Schematic of WDR firing changes evoked by 60 g stimulation pre- versus post-RVM GABAergic activation. (c) Schematic of WDR firing changes evoked by 200 g stimulation pre- versus post-RVM GABAergic activation. (d) Frequency histogram of WDR firing in ChR2 group (activation) pre- versus post-478 nm laser stimulation (*p < 0.05; paired t-test; 40 neurons from 10 rats). (e) Frequency histogram of WDR firing in mCherry group (control) pre- versus post-laser (n.s. p > 0.05; paired t-test; 24 neurons from 6 rats). (f) From Figure (c) and (d), intergroup comparison of 60 g-evoked firing changes (ChR2 vs mCherry; *p < 0.05; unpaired t-test). (g) From Figure (c) and (d), intergroup comparison of 200 g-evoked firing changes (ChR2 vs mCherry; n.s. p > 0.05; unpaired t-test).

Figure 4.

Optogenetic inhibition of RVM GABAergic neurons reduces nociceptive-evoked WDR firing. (a) Experimental timeline. (b) Schematic of WDR firing changes evoked by 60 g non-nociceptive stimulation pre- versus post-RVM GABAergic inhibition. (c) Schematic of WDR firing changes evoked by 200 g nociceptive stimulation pre- versus post-RVM GABAergic inhibition. (d) Frequency histogram of WDR firing in eNpHR3.0 group (inhibition) pre- versus post-589 nm laser stimulation (***p < 0.001; paired t-test; 33 neurons from 11 rats). (e) Frequency histogram of WDR firing in mCherry group (control) pre- versus post-laser (n.s. p > 0.05; paired t-test; 26 neurons from 6 rats). (f) From Figure (c) and (d), intergroup comparison of 60 g-evoked firing changes (eNpHR3.0 vs mCherry; n.s. p > 0.05; unpaired t-test). (g) From Figure (c) and (d), intergroup comparison of 200 g-evoked firing changes (eNpHR3.0 vs mCherry; *p < 0.05; unpaired t-test).

Given that the firing dynamics of WDR neurons reflect the central processing of peripheral pain signals, this study further explored the involvement of RVM GABAergic neurons in pain modulation and analgesia in the naïve state (Figure 5(a) and (d)). Mechanical pain thresholds were evaluated 3 weeks after viral expression. In naïve rats, activation of RVM GABAergic neurons (15 Hz, 478 nm, 0.8–1.2 mW) significantly decreased the 50% paw withdrawal threshold, indicating pronociceptive effects (Figure 5(b) and (c)). In contrast, optogenetic inhibition (continuous, 589 nm, 0.8–1.2 mW) of RVM GABAergic neurons in naïve animals increased baseline pain thresholds (Figure 5(e) and (f)), demonstrating their antinociceptive potential.

Figure 5.

RVM GABAergic neurons modulate pain thresholds and EA analgesia. (a) Viral injection sites (ChR2/mCherry) and fluorescence expression for optogenetic activation of RVM GABAergic neurons. (b) Optogenetic activation reduces 50% mechanical withdrawal threshold (****p < 0.0001, light on vs off; unpaired t-test, n = 8). (c) Laser-evoked threshold changes from (b) (****p < 0.0001; unpaired t-test). (d) Viral injection sites (eNpHR3.0/mCherry) and fluorescence expression for optogenetic inhibition. (e) Optogenetic inhibition increases 50% mechanical withdrawal threshold (****p < 0.0001, light on vs off; unpaired t-test, n = 8). (f) Laser-evoked threshold changes from (e) (**p < 0.001; unpaired t-test). (g) CFA-induced mechanical allodynia (****p < 0.0001, pre- vs post-CFA; unpaired t-test, n = 15). (h) EA increases withdrawal threshold in CFA rats; RVM GABAergic neurons activation attenuates EA effect (***p < 0.001, n.s. p > 0.05; one-way ANOVA, Tukey’s test, n = 9). (i) CFA reduces left hindlimb weight-bearing (****p < 0.0001, pre- vs post-CFA; unpaired t-test, n = 15). (j) EA restores weight-bearing; RVM GABAergic neurons activation blunts EA effect (**p < 0.01, *p < 0.05, n.s. p > 0.05; one-way ANOVA, Tukey’s test, n = 9). (k) RVM GABAergic neurons inhibition elevates withdrawal threshold in CFA rats (****p < 0.0001, light on vs off; unpaired t-test, n = 8). (l) Laser-evoked threshold changes from (k) (****p < 0.0001; unpaired t-test). (m) RVM GABAergic neurons inhibition improves weight-bearing in CFA rats (***p < 0.001, light on vs off; unpaired t-test, n = 8). (n) Laser-evoked weight-bearing changes from (m) (***p < 0.001; unpaired t-test).

Activating RVM GABAergic neurons reverses EA’s inhibitory effects on WDR neurons

To explore the involvement of RVM GABAergic neurons in the EA-mediated regulation of WDR sensory neurons, we employed optogenetic activation of RVM GABAergic neurons during EA intervention, utilizing an empty virus as a control. This approach aimed to evaluate whether activation of RVM GABAergic neurons mitigates the EA-induced suppression of WDR neuronal activity (Figure 6(a)). In the group receiving the mCherry control virus, the concurrent application of EA and blue laser intervention resulted in a significant reduction in the firing frequency of WDR neurons in response to nociceptive stimulation (Figure 6(b) and (d)) and a moderate decrease in firing elicited by non-nociceptive stimulation (Figure 6(b) and (d)). Conversely, in the ChR2 group, no significant alterations in WDR firing frequencies were observed for either nociceptive or non-nociceptive stimuli before and after the combined EA and laser intervention (Figure 6(c)–(e)). A comparative analysis of WDR neuron firing modulation between the mCherry and ChR2 groups demonstrated that EA-induced suppression of both 60 g-evoked (non-nociceptive) and 200 g-evoked (nociceptive) WDR activity was significantly more pronounced in the mCherry group (Figure 6(f) and (g)). These findings suggest that activation of RVM GABAergic neurons nearly nullifies the inhibitory effects of EA on WDR neurons.

Figure 6.

Optogenetic activation of RVM GABAergic neurons counteracts EA’s suppression of WDR neurons. (a) Experimental timeline. (b) Schematic of WDR neuron firing changes in the mCherry group (control) pre- and post-EA with 478 nm laser intervention. (c) Schematic of WDR neuron firing changes in the ChR2 group (activation) pre- and post-EA with 478 nm laser intervention. (d) Frequency histogram of WDR firing in the mCherry group pre- versus post-EA + laser (*p < 0.05, ***p < 0.001; paired t-test; 16 neurons from 6 rats). (e) Frequency histogram of WDR firing in the ChR2 group pre- versus post-EA + laser (n.s. p > 0.05; paired t-test; 32 neurons from 7 rats). (f) From Figure (d) and (e), intergroup comparison of 60 g non-nociceptive-evoked firing changes (mCherry vs ChR2; *p < 0.05; unpaired t-test). (g) From Figure (d) and (e), intergroup comparison of 200 g nociceptive-evoked firing changes (mCherry vs ChR2; *p < 0.05; unpaired t-test).

Inhibiting RVM GABAergic neurons mimics EA’s suppressive effects on WDR neurons

To further substantiate the hypothesis, optogenetic inhibition of GABAergic neurons in the RVM of CFA rats was conducted to determine whether it replicates the inhibitory effects of EA on WDR neurons (Figure 7(a)). In the eNpHR3.0 group, inhibition of RVM GABAergic neurons significantly diminished nociceptive stimulation-evoked WDR firing compared to baseline levels (Figure 7(c) and (e)). No significant alterations in WDR firing were detected in the mCherry control group (Figure 7(b)–(d)). Comparative analysis of WDR firing modulation between the mCherry (control) and eNpHR3.0 (inhibition) groups demonstrated that nociceptive stimulation-evoked WDR activity was significantly more suppressed in the eNpHR3.0 group (Figure 7(f) and (g)). These findings suggest that transient inhibition of RVM GABAergic neurons partially emulates the EA-induced suppression of nociceptive WDR neurons, thereby supporting the hypothesis that the anti-nociceptive effects of EA at the spinal level are mediated, at least in part, by the inhibition of RVM GABAergic neuronal activity.

Figure 7.

Optogenetic inhibition of RVM GABAergic neurons mimics EA’s suppression of WDR neurons. (a) Experimental timeline. (b) Schematic of WDR neuron firing in the mCherry control group pre- and post-589 nm laser stimulation. (c) Schematic of WDR neuron firing in the eNpHR3.0 inhibition group pre- and post-589 nm laser stimulation. (d) Frequency histogram of WDR firing in the mCherry group pre- versus post-laser (n.s. p > 0.05; paired t-test; 22 neurons from 6 rats). (e) Frequency histogram of WDR firing in the eNpHR3.0 group pre- versus post-laser (n.s. p > 0.05, **p < 0.01; paired t-test; 19 neurons from 6 rats). (f) From Figure (d) and (e), intergroup comparison of 60 g non-nociceptive-evoked firing changes (mCherry vs eNpHR3.0; n.s. p > 0.05; unpaired t-test). (g) From Figure (d) and (e), intergroup comparison of 200 g nociceptive-evoked firing changes (mCherry vs eNpHR3.0; ***p < 0.001; unpaired t-test).

EA suppresses pain signals by a modulating RVM GABAergic neuronal activity

To further validate the hypothesis that EA modulates nociception through RVM GABAergic neurons, we conducted pain sensitivity tests on CFA model rats combined with optogenetic manipulation (Figure 5(g) and (i)). Results demonstrated that EA treatment significantly increased pain thresholds in CFA rats compared to untreated controls, including the 50% mechanical withdrawal threshold and affected limb weight-bearing ratio (Figure 5(h) and (j)). However, concurrent optogenetic activation of RVM GABAergic neurons during EA intervention markedly attenuated these analgesic effects (Figure 5(h) and (j)). Optogenetic inhibition of RVM GABAergic neurons using eNpHR3.0 significantly increased both the 50% withdrawal threshold and weight-bearing ratio (Figure 5(k)–(n)), whereas the mCherry control group showed no significant changes. These findings indicate that RVM GABAergic neurons play a crucial role in mediating EA analgesia, consistent with prior research from our group. ⁴⁹ Furthermore, our results elucidate the mechanistic pathway through which EA suppresses the activity of RVM GABAergic neurons, thereby reducing nociceptive transmission in the spinal dorsal horn.

Discussion

In this study, we investigated the potential important role of RVM GABAergic neurons in regulating spinal WDR neurons, focusing on their involvement in nociceptive information processing. Our findings indicate that increased activity of RVM GABAergic neurons augments non-nociceptive stimulus-evoked firing in WDR neurons. In contrast, inhibition of these neurons attenuates chronic pain-induced aberrant firing in WDR neurons and reduces spinal nociceptive transmission. These results are consistent with a critical role of RVM GABAergic neurons in modulating spinal WDR neuronal activity and pain processing. This led us to explore whether the inhibitory effects of EA on WDR neurons are contingent upon RVM GABAergic modulation. Our data show that although EA significantly suppresses CFA-induced abnormal firing in WDR neurons, optogenetic activation of RVM GABAergic neurons almost completely negates this effect of EA. The striking observation that activating RVM GABAergic neurons counteracts the effect of EA, whereas inhibiting them mimics it, strongly suggests that effective EA likely produces analgesia, at least in part, by suppressing the activity of this specific neuronal population in the RVM.

WDR neurons are integral to the transmission and processing of nociceptive signals. ⁵⁰ Previous research has demonstrated that EA can suppress the pathological hyperactivity of WDR neurons, although its effectiveness is contingent upon the stimulation parameters and anatomical sites involved.^51–53 Research indicates that high-intensity EA generally produces superior analgesic effects compared to low-intensity EA. ⁴⁸ Notably, low-intensity EA is effective in producing analgesic effects only when applied directly to the site of pain, whereas high-intensity EA remains effective even when administered to the contralateral limb.^10,26 The neurobiological rationale for choosing contralateral stimulation is significant. At ipsilateral, low-intensity EA may primarily engage segmental gate-control mechanisms mediated by Aβ fibers. But at the contralateral, 2 mA EA, a higher-intensity, is more likely to activate small-diameter (Aδ/C) fibers, which project bilaterally and are more effective in engaging the descending inhibitory pathways (e.g. periaqueductal gray (PAG)-RVM-spinal cord) that were the focus of this study. This approach helps isolate the supraspinal analgesic component and aligns with clinical observations that high-intensity, contralateral EA can produce potent analgesia via central mechanisms, particularly in conditions involving central sensitization. Current theoretical frameworks propose that low-intensity EA primarily activates Aβ fibers, thereby inhibiting spinal nociceptive transmission through the gate control mechanism. ²⁷ In contrast, high-intensity EA sufficiently stimulates C-fibers to activate descending analgesic systems, resulting in enhanced analgesia.^10,28,29,32 Considering that high-intensity EA simultaneously activates both Aβ and C-fibers, we applied EA to the contralateral side of the injury to specifically target C-fiber-mediated mechanisms. Additionally, our electrophysiological assessments confirmed that a 2 mA single-pulse stimulation effectively activates sensory C-fibers (Figure 8). Based on these verifications, we established standardized EA parameters for subsequent experiments: an intensity of 2 mA, an alternating frequency of 2/15 Hz, a pulse width of 0.5 ms, and stimulation at the contralateral ST36 acupoint.

Figure 8.

2 mA electrical stimulation activates sensory C fibers.

Based on distinct conduction velocities of Aβ, Aδ, and C sensory fibers, WDR neurons exhibit differential firing latencies in response to sensory stimuli. During extracellular recordings, single-pulse 2 mA electrical stimulation was applied to the muscle receptive field. The type of activated sensory fibers was determined by WDR neuronal firing latencies: Aβ fibers (fast-conducting): 0–20 ms latency; Aδ fibers (intermediate-conducting): 20–90 ms latency; C fibers (slow-conducting): 90–300 ms latency. ³²

Our experimental data suggest that the activation of GABAergic neurons within the RVM led to an increase in the firing of WDR neurons in response to non-nociceptive stimulation, while it did not enhance firing in response to nociceptive stimulation. Conversely, the inhibition of RVM GABAergic neurons resulted in a reduction of WDR neuron firing elicited by 200 g nociceptive stimulation, with no effect on firing induced by 60 g non-nociceptive stimulation. These findings imply that, in naïve rats under physiological conditions, RVM GABAergic neurons remain inactive and are only activated during pathological states or acute pain, maintaining a state of chronic activation during persistent pain conditions. This observation is consistent with our previous research findings. ⁴⁹

As 60 g stimulation is non-nociceptive, the optogenetic inhibition of RVM GABAergic neurons did not affect baseline information transmission within the RVM-spinal dorsal horn GABAergic pathway under physiological conditions. Consequently, this inhibition did not significantly alter the firing frequency of WDR neurons in response to 60 g stimulation. Conversely, 200 g nociceptive stimulation activates RVM GABAergic neurons, and the inhibition of these neurons significantly reduced WDR neuronal firing in response to the noxious stimulus. For ChR2-mediated activation, while a 60 g stimulus is considered non-nociceptive, prolonged activation of GABAergic neurons simulates the pathological conditions associated with chronic pain and central sensitization. This results in an increased firing frequency of WDR neurons in response to 60 g stimulation. Further analysis indicates that activation of RVM GABAergic neurons negates the inhibitory effects of EA on WDR neurons, whereas their inhibition replicates the suppressive effects of EA. This supports the hypothesis that the regulation of WDR neurons by RVM GABAergic neurons is a crucial mechanism through which EA attenuates WDR neuronal responses to peripheral nociceptive stimuli.

This study collectively provides evidence for the regulatory function of RVM GABAergic neurons in modulating WDR neurons and identifies the mechanism by which contralateral high-intensity EA could induce analgesia through this pathway, thereby supporting our proposed mechanistic hypothesis (Figure 9). Under normal physiological conditions, non-nociceptive sensory information from the periphery is conveyed to the spinal dorsal horn via the DRG and synapses with inhibitory interneurons in the spinal dorsal horn, which subsequently inhibit WDR neurons. ³⁷ Consequently, non-nociceptive information does not ascend to the cortex to elicit pain. However, during peripheral nociceptive stimulation, RVM GABAergic neurons become activated. Through descending projections, they exert inhibitory control over spinal dorsal horn interneurons,^36,37 facilitating the transmission of nociceptive signals via WDR neurons to central cortical areas for pain perception. In chronic pathological pain conditions, RVM GABAergic neurons remain persistently activated, ⁴⁹ leading to sustained inhibition of spinal dorsal horn interneurons.^36,37 This leads to impaired regulation of ascending WDR neurons, resulting in their persistent activation by peripheral nociceptive stimuli. As a result, even non-noxious peripheral inputs can easily activate WDR neurons, inducing pain perception – a fundamental mechanism underlying central sensitization. The 2 mA EA protocol mitigates this effect by inhibiting RVM GABAergic neurons, ⁴⁹ which in turn disinhibits spinal inhibitory interneurons.^36,37 This reversal of central sensitization reestablishes normal sensory processing of both nociceptive and non-nociceptive information in the dorsal horn.

Figure 9.

Hypothetical mechanism diagram.

Although this study clarifies the spinal modulatory role of RVM GABAergic neurons, a full understanding of EA analgesia requires integrating this circuit into the broader neural network. The analgesic signal triggered by contralateral, high-intensity EA must first ascend to supraspinal centers. A key ascending pathway is the spinoparabrachial tract, which conveys nociceptive information to the parabrachial nucleus (PBN). The PBN, in turn, projects to limbic structures and, importantly, to PAG. As a well-established upstream regulator, the PAG likely serves as a critical relay that forwards processed EA signals to the RVM. Furthermore, EA is known to promote the release of endogenous opioids and endocannabinoids, which may act on RVM GABAergic neurons via μ-opioid receptors (MOR) or cannabinoid CB1 receptors, respectively, to suppress their activity.^33,54 Upon receiving these integrated inputs, the RVM (including its GABAergic population) functions as a pivotal hub within a wider network of descending pain modulatory systems. This network includes noradrenergic projections from the locus coeruleus (LC) and serotonergic pathways from the dorsal raphe nucleus (DRN), both well-documented contributors to endogenous pain inhibition. ⁵⁵ These monoaminergic pathways converge on the spinal dorsal horn, where they enhance the activity of inhibitory interneurons (GABAergic and glycinergic) to attenuate nociceptive transmission. ⁵⁵ Our findings suggest that EA, by inhibiting RVM GABAergic neurons, may disinhibit these spinal interneurons, thereby synergizing with other descending monoaminergic actions to restore spinal inhibitory tone. The coordinated interaction between brainstem GABAergic regulation and other descending pathways likely underpins the comprehensive analgesic effect of EA. Future studies are needed to precisely map the specific ascending circuits to the RVM and elucidate the temporal and spatial dynamics of these multi-system interactions during EA stimulation.

Despite these advances, several limitations warrant consideration. Firstly, although our optogenetic and behavioral data strongly suggest a role for RVM GABAergic neurons, our interpretation that EA exerts its effect by suppressing their activity is based on correlative evidence from loss- and gain-of-function manipulations. Direct confirmation via in vivo recording of these neurons’ activity during EA intervention, such as through electrophysiology or fiber photometry, is a crucial next step to establish causality. Specifically, studies employing fiber photometry or single-unit recordings in RVM GABAergic neurons during EA of varying intensities are necessary to elucidate the precise dose-response relationship. Secondly, the use of different anesthetics tailored to specific procedures, although standard and justified, may introduce variability in the baseline states of neurons. Nonetheless, internal controls within each experimental paradigm help mitigate this concern regarding the interpreted effects. Thirdly, although our findings indicate that the anti-nociceptive effects of EA at the spinal level are mediated by RVM GABAergic neurons, the specific pathways through which acupuncture signals ascend to the RVM and influence GABAergic neuronal activity remain unresolved. Previous studies suggest that EA analgesia may involve the activation of μ-opioid receptors or cannabinoid receptor 1 (CB1R) on GABAergic neurons, leading to a reduction in GABA release.^33,49,54,56 Nonetheless, whether EA directly suppresses GABAergic transmission via opioid or CB1 receptor signaling necessitates further experimental validation. Additionally, this study did not include a control group subjected to non-acupoint electrical stimulation. Future research that incorporates such a control group would be crucial for distinguishing the specific effects of acupoint stimulation from those of general electrical stimulation.

Overall, our data suggest that effective EA may exert its analgesic effects, at least in part, by suppressing the activity of RVM GABAergic neurons. Our findings support the hypothesis that the modulation of RVM GABAergic neurons contributes to EA-induced analgesia, partly by regulating the activity of spinal WDR neurons. This study elucidates a potential circuit mechanism for contralateral, higher-intensity EA and provides an enhanced framework for understanding the central actions of EA in the context of chronic pain.