We used Ai9 (RCL-tdT), GAD2-Cre, and C57BL/6J mice (purchased from Jackson Laboratories or Charles River) at 8–10 weeks of age. All mice were maintained at a stable temperature (23°C–25°C) under a 12-h light/dark cycle. Before surgery, mice were housed in cages of six in a colony. The animal protocols were approved by the Animal Care and Use Committee of Anhui University of Chinese Medicine.

We used restraint stress (RS) combined with irregular feeding to establish our GMD model. Mice were periodically restrained in a 50 mL syringe for 1 h once per day to constrain their movements and they were fed irregularly (fed on 1 day, fasted for 1 day) for 21 days. Holes were drilled in the syringes to allow the mice to breathe. Control mice were allowed to freely move, and they had ad libitum access to food.

The EA group received EA stimulation on the 15th day of the experiment (Figure 1B). Mice were anesthetized by inhalation of isoflurane (1.5%–3.0%) via an anesthesia machine. EA was performed with a disperse-dense wave mode for 20 min once per day for 7 days, with electrical current range of 1–2 mA, and a frequency of 2/15 Hz by using a SDZ-IV electronic instrument (Huatuo Brand). EA stimulation was performed at the abdominal RN12 (Zhongwan) and dorsal BL21 (Weishu) acupoints by inserting a 0.18 × 13 mm acupuncture needle. The RN12 acupoint is located on the anterior median line of the upper abdomen, 10 mm below the xiphisternal synchondroses. The BL21 acupoint is located under the spinous process of the 12th thoracic vertebra.

Food intake and gastric emptying studies were performed as in other studies (Sampath et al., 2021). Mice were fasted overnight (water available ad libitum). The next day, each mouse was caged separately and fed with pre-weighed food for 30 min. In optogenetic experiments, each mouse was fed with light on, and was deprived with light off, the food intake recorded also for 30 min. The remaining unconsumed food was weighed to determine how much food each mouse ate. Food intake (g) = pre-weighed amount-remaining amount. Thereafter, each mouse was placed in a separate clean cage without food and water for 90 min. Mice were then anaesthetized with pentobarbital and euthanized by cervical dislocation. The stomach was removed and weighed (whole stomach weight). The gastric contents were removed from the stomach and the stomach was weighed again (empty stomach weight). The rate of gastric emptying (GE) was calculated as follows: GE(%) = [1—(whole stomach weight-empty stomach weight)/weight of food intake] × 100.

Gastric motility studies were performed as described previously (Wang et al., 2020). Mice were fasted overnight (water ad libitum) and anesthetized with pentobarbital (50 mg/kg, intraperitoneally). Following a laparotomy, a miniature strain gauge (1 mm × 1 mm, 120 Ω) was fixed to the circular smooth muscle of the gastric antrum. The laparotomy was then closed with a 5–0 suture with the strain gauge leads exteriorized. Strain gauge signals were amplified, filtered, digitized via powerlab 8/30 signal system, and recorded using labchart software (AD Instruments). The waves of gastric contraction were monitored for at least 30 min.

A custom-made tetrode array was implanted into the DMV (ML: −0.4 mm, AP: −7.64 mm, DV: −4.45 mm) or CeA (ML: −2.77 mm, AP: −1.22 mm, DV: −4.52 mm) in different batches of mice at the beginning of experiments. The coordinates were defined as dorsal-ventral (DV) from brain surface, anterior-posterior (AP) from bregma and mediolateral (ML) from midline. The screw-based microdrive scaffolds for lowering the electrodes were cemented onto the skull. Each tetrode was made of four twisted fine platinum/iridium wires (diameter 12.5 μm, California Fine Wire). The mice were allowed to recover for 3 days, and the electrodes were attached to a 16-channel signal acquisition system to collect the spontaneous discharge. Then the mice received restraint stress and electroacupuncture treatment. On the next day after the completion of EA treatment, the neuronal activities were recorded, and the data filtered at a bandwidth of 300–5,000 Hz were stored using Neurostudio software. Neuroexplorer 4 (Nex Technologies, United States) was used to calculate the firing rates of the sorted units. We sorted more than 24 units from the recorded spikes of different groups. Only those units with signal noise ratio exceeding 2.5 and average amplitude exceeding 50 μV were included for comparison. However, we removed those with much noise or smaller amplitude to minimize the potential artifact effect.

We anaesthetized the mice and restrained them in a stereotaxic frame (RWD). A volume of 200 nL of virus was injected into the CeA using a glass microelectrode, which was connected to an infusion pump (micro 4, WPI, United States).

For anterograde tracing of the CeA^GABA → DVC circuit, we injected the Cre-dependent virus AAV-DIO-mCherry into the unilateral CeA of GAD2-Cre mice. For inhibition of the CeA^GABA neurons or CeA^GABA → DVC circuit, we injected the AAV-DIO-eNpHR3.0-EYFP viruses into the bilateral CeA; However, we injected the AAV-DIO-ChR2-mCherry viruses into the unilateral CeA to activate the CeA^GABA neurons or CeA^GABA → DVC circuit with optogenetic manipulation. The AAV-DIO-mCherry and AAV-DIO-EYFP viruses were used as controls. All viruses were packaged by BrainVTA (Wuhan). 21 days later, the mice were transcardially perfused with 0.9% saline followed by 4% paraformaldehyde. The brain was sectioned for imaging. Images of viral expression were obtained using a confocal microscope (LSM 710; ZEISS, Germany).

An optical fiber (200 µm in diameter) was implanted into the CeA or DVC. The implant was fixed to the skull of mouse using dental cement. The delivery of a 30 min yellow light (594 nm, 5–8 mW, light on 10 min/light off 5 min, three cycles) or 30 min pulsed blue light (473 nm, 2–5 mW, 10 ms pulses, 20 Hz, light on 10 min/light off 5 min, three cycles) was controlled by a master-8 pulse stimulator (A.M.P.I., Israel) (Figures 3D–L). After the experiment, we examined the location of the fibers and removed data from situations in which the fibers weren’t in the proper target.

SPSS 25.0 software was used for statistical analysis. An unpaired t-test was performed for two-group comparisons. ANOVA (two-way or one-way) with a post hoc Tukey’s test was used for multiple group comparisons. The significance levels are indicated as **p < 0.01, ##p < 0.01. All data are expressed as the mean ± SEM.

Physiological and psychological stress can cause FGIDs, and acupuncture improves gastrointestinal motility. To confirm this phenomenon, we established a GMD model in the context of RS and irregular feeding and performed EA (Figures 1A, B). The results illustrated that RS induced delayed gastric emptying, decreased food intake and gastric motility, and EA alleviated GMD (Figures 1C–F).

The emotional center (CeA) and gastrointestinal center (DVC) are involved in stress-induced GMD. We recorded the neuronal activity across the CeA and DVC in GMD mice. We demonstrated that RS increased the firing rate of CeA neurons and decreased the firing rate in DVC neurons in free-moving mice, with EA reversing this phenomenon (Figures 2A–F). GABAergic neurons are distributed throughout the CeA, and GABAergic projections play an important role in control of DMV neuronal firing. Therefore, we focused on CeA^GABA neurons. Patch clamp recordings were performed in CeA^GABA neurons in acute brain slices. To visualize GABAergic neurons, Ai9 (RCL-tdT) mice were crossed with GAD2-Cre mice to reproduce transgenic mice with red tdTomato-expressing GABAergic neurons (GAD2-tdT, Figure 2G). We demonstrated an increase in action potentials with the current-elicited in GMD mice and a decrease in action potentials in GMD + EA mice (Figures 2H, I). These results indicate that CeA^GABA neurons and the DVC neurons are involved in the regulation of gastric motility with EA. However, the role of CeA^GABA neurons in modulating the activity of DVC neurons to regulate gastric motility is still unclear.

Given the enhanced CeA^GABA neuronal activity in GMD mice, we subsequently aimed to inhibit CeA^GABA neurons and observe the change in gastric motility in GMD mice. We injected Cre-dependent eNpHR3.0 into the CeA to selectively suppress CeA^GABA neurons. The functionality of the eNpHR3.0 virus was verified by patch clamp (Figures 3A–C). Gastric motility data illustrated that optical inhibition of CeA^GABA neurons resulted in significantly increased food intake, gastric movement, and gastric emptying in GMD mice (Figures 3D–H).

In naïve mice, we injected Cre-dependent ChR2 into the CeA to selectively activate CeA^GABA neurons. The functionality of the ChR2 virus was verified by patch clamp (Figures 3I–K). The results showed that activation of CeA^GABA neurons reduced food intake, delayed gastric emptying and decreased gastric motility (Figures 3L–P). These results establish the functional linkage between CeA^GABA neurons and gastric motility.

Previous evidence suggested a direct link between the CeA and DVC. To confirm the presence of a CeA^GABA → DVC projection, an anterograde transmonosynaptic tracing system was employed. Cre-dependent AAV was injected into the CeA (Figure 4A). 21 days later, we examined mCherry⁺ cell bodies within the CeA (Figure 4B), and numerous additional mCherry⁺ signals were observed in the DVC (Figure 4C). These findings suggest an anatomical connection from CeA^GABA neurons to the DVC.

To test the functional connection between the CeA^GABA → DVC pathway, brain slice recordings were performed. Brief light stimulation of ChR2-containing CeA^GABA terminals in the DVC reliably elicited IPSCs in DVC neurons, which were eliminated by the GABA receptor antagonist picrotoxin (PTX) (Figures 4D–F). These data verify that CeA^GABA neurons send inhibitory afferents to the DVC.

We mapped the CeA^GABA → DVC pathway to examine whether it participated in regulation of RS-induced GMD. Optogenetic manipulations were employed in GMD mice. We found that optogenetic inhibition of the activity of the CeA^GABA → DVC circuit significantly promoted food intake, gastric movement, and gastric emptying in GMD mice (Figures 5A–E). However, in naïve mice, we found that activation of the CeA^GABA → DVC circuit reduced food intake, delayed gastric emptying, and subsequently decreased gastric motility (Figures 5F–J). These results suggest that activation of GABAergic projections from CeA to DVC may mimic the symptoms of RS-induced GMD, and inhibition of these projections may relieve symptoms of GMD.