Gastric electrical stimulation ameliorates antral myenteric neurons in streptozotocin-induced diabetic beagle canines

Abstract

BACKGROUND

The loss of enteric neurons may contribute to the altered gastric emptying observed in diabetic gastroparesis. Multiple lines of evidence indicate that gastric electrical stimulation (GES) can effectively alleviate symptoms in patients with refractory gastroparesis; however, its therapeutic mechanism remains unclear.

AIM

To evaluate the impact of GES on enteric neurons in the antral myenteric plexus of diabetic beagle canines.

METHODS

Adult female beagles were used to establish a diabetes mellitus (DM) (n = 6) model via intravenous injection of streptozotocin, with healthy canines as controls (n = 3). All canines were implanted with stimulation devices and received a real GES session (0.33 millisecond, 5 mA, 15 Hz, 0.2 second on, 3 seconds off) and a sham GES session (SGES) (stimulators off) of 4 weeks each in a randomized crossover design, separated by a 3-week washout period. Full-thickness antral biopsies were obtained after each session, and pancreatic biopsies were conducted at the end of the study. Hematoxylin-eosin staining was employed to measure histopathological changes in the pancreatic islets. Immunofluorescence staining of antral full-thickness sections was employed to evaluate the alterations in choline acetyltransferase (ChAT), vasoactive intestinal peptide, neuronal nitric oxide synthase (nNOS), or calretinin immunoreactive (IR) positive neurons in relation to the anti-human neuronal protein HuC/HuD. Based on different stimulations, the results were analyzed in DM + GES, DM + SGES, control + GES, and control + SGES subgroups.

RESULTS

Diabetic canines exhibited a significant reduction in the number and area of pancreatic islets compared with controls. The proportions of ChAT-IR and nNOS-IR neurons in the DM + SGES subgroup were significantly lower than those in the control + SGES subgroup (67.5% ± 1.8% vs 73.3% ± 1.8%, P = 0.003; 27.3% ± 1.4% vs 32.1% ± 0.7%, P = 0.001), with an upregulation of vasoactive intestinal peptide-IR neurons (51.9% ± 0.8% vs 46.8% ± 1.2%, P < 0.001). Following the 4-week GES intervention, the proportions of ChAT-IR and nNOS-IR neurons were significantly higher (73.3% ± 1.7% vs 67.5% ± 1.8%, P = 0.002; 32.1% ± 1.7% vs 27.3% ± 1.4%, P = 0.002) compared to SGES in diabetic canines. A similar increase was also observed in the proportion of nNOS-IR neurons between the GES and SGES subgroups in controls (37.5% ± 1.3% vs 32.1% ± 0.7%, P = 0.036).

CONCLUSION

Imbalanced regulation of excitatory and inhibitory neurons in the enteric nervous system occurred in streptozotocin-induced diabetic beagles, and GES improved the survival of cholinergic and nitrergic neurons in diabetic canines.

Key Words: Diabetes gastroparesis; Gastric electrical stimulation; Choline acetyltransferase; Vasoactive intestinal peptide; Neuronal nitric oxide synthase; Enteric nervous system

Core Tip: Experiments in small animal models have confirmed that damage to the enteric nervous system severely impairs physiological movement in the gastrointestinal system in diabetic gastroparesis. We found that the morphologic changes in gastric excitatory and inhibitory neurons are also present in streptozotocin-induced diabetic beagle canines, despite the absence of obvious manifestations of gastroparesis. High-frequency gastric electrical stimulation ameliorates the enteric nervous system in diabetic canines by upregulating choline acetyltransferase and neuronal nitric oxide synthase-immunoreactive positive neurons, these neuromodulatory effects on enteric neuronal plasticity could restore slow wave rhythms, enhance gastric emptying, and improve gastric accommodation.

INTRODUCTION

Diabetic gastroparesis (DGP) is a complication of diabetes mellitus (DM) characterized by delayed gastric emptying (GE) without mechanical pyloric obstruction. In a community-based study of individuals with diabetes, the 10-year incidence of gastroparesis in type 1 and type 2 diabetes were 5% and 1%, respectively[1]. DGP is manifested with symptoms such as early satiety, postprandial fullness, belching, nausea, vomiting, abdominal pain, and weight loss. These symptoms can result in dysglycemia, frequent hospitalizations, and a poor quality of life[2]. The pathophysiology of DGP is complex and appears to be multifactorial, such as hyperglycemia, vagal dysfunction, loss of neuronal nitric oxide synthase (nNOS) expression in the myenteric plexus (MP), abnormalities in the interstitial cells of Cajal (ICC) network, and oxidative stress[3,4]. Nevertheless, the exact pathophysiology of diabetic gastroenteropathy remains unknown. Dietary modifications and medications are the first-line treatment for gastroparesis, but approximately 30% of patients do not achieve symptom relief with conservative treatment[5].

Gastric electrical stimulation (GES) uses high-frequency, low-energy stimuli to improve symptoms of nausea, vomiting, and hyperglycemia in patients with gastroparesis and has a response rate of 50%-75%[6,7]. A prospective single-center study of 151 patients with refractory gastroparesis demonstrated that GES significantly improved symptoms in 75% of patients with 43% having at least moderate improvement[7]. Clinical improvement was observed in both diabetic and idiopathic patients, with a higher response rate in diabetics than in non-diabetic patients (85% vs 68%)[7]. However, the mechanisms by which GES improves the gastric motility and hyperglycemia, especially in DGP, need to be further elucidated.

Diabetic enteric neuropathy is associated with alterations in the size and number of excitatory and inhibitory enteric neurons, as well as neurodegenerative changes in the enteric nervous system (ENS)[8]. Excitatory enteric neurons, such as cholinergic neurons, primarily express choline acetyltransferase (ChAT) to augment slow waves and increase the amplitude of peristaltic waves. Du et al[9] found enteric neuropathy [characterized by decreased ChAT-immunoreactive (IR) and nNOS-IR positive neurons] in the proximal and distal colon of 8-week and 12-week diabetic rats compared with the control group. It has been reported that ChAT-IR positive neurons in the myenteric layer of the gastric antrum were also reduced in streptozotocin (STZ)-induced diabetic rats for more than 4 weeks[10]. Vasoactive intestinal peptide (VIP) together with nitric oxide (NO) are the primary inhibitory neurotransmitters, mediating smooth muscle relaxation and VIP, inhibiting gastric acid secretion in the stomach, and stimulating the release of intestinal juice and bile in the intestines[11]. Previous studies in the porcine diabetic model have shown that an increased population of VIP-IR positive neurons in the inner submucosal plexus and MP of the duodenum, jejunum and ileum for 6 weeks contributed to gastrointestinal (GI) dysfunctions[12]. The inhibitory nitrergic neurons, which mainly express nNOS, release NO to mediate the relaxation of smooth muscles in the gastric fundus, and are therefore important in gut motility[13]. Several studies have demonstrated that loss of nNOS-IR positive neurons in the MP of the antrum can result in delayed GE in STZ-induced diabetic rats or gastric relaxation in spontaneously diabetic rats[14,15]. Calretinin (CALR), a calcium-binding protein expressed in specific enteric interneurons, may play a role in sensory signal processing within the GI tract[16]. However, the effects of CALR-IR positive neurons in diabetes remain unknown.

According to available data, the therapeutic mechanisms of GES for DGP include improved gastric accommodation, enhanced vagal activity, alterations in the ENS, and activation of central neurons[17,18]. Chen et al[10] revealed that GES for 6 weeks increased the expression of ChAT-IR and nNOS-IR positive neurons located in the myenteric layer of diabetic rats partly through glial cell line-derived neurotrophic factor expression. Furthermore, Wang et al[19] found that synchronized dual-pulse GES [a long pulse (300 milliseconds, 4 mA) and 5 short pulses (0.33 millisecond, 4 mA, 100 Hz)] improved enteric neuronal survival in the antrum tissue, and enhanced GE in vagotomized rats, possibly via cell line-derived neurotrophic factor and the phosphatidylinositol 3-kinase/protein kinase B pathways. Several studies have focused on the physiology of canine ENS because its higher similarity to humans than other small animals[20], but to date, no information is available on the efficacy mechanisms of GES on ENS in diabetic canines. In this study, we aimed to evaluate the effects of GES in STZ-induced diabetic beagle canines in concert with morphological changes in the MP of the antrum.

MATERIALS AND METHODS

Animals

Adult female beagle dogs (fifteen months) were purchased from Beijing Keyu Animal Breeding Co., Ltd (Beijing, China). They weighed 8.3-12.5 kg and were housed individually in cages in a standardized environment at 16-28 °C, with a relative humidity of 40%-70% and a 12-hour/12-hour light-dark cycle. The canines had free access to food and water and were acclimated for 7 days before the modeling experiments. All procedures involving animals complied with the ARRIVE guidelines and the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Animal Welfare and Ethical Committee of Peking Union Medical College Hospital (No. XHDW-2020-047 and No. XHDW-2020-047-2).

Animal models

The modeling procedures were conducted in batches with a healthy canine randomly assigned per batch, thereby dividing the cohort into experimental and control groups. During the initial modeling phase, experimental canines received intravenous injections of freshly dissolved STZ (30 mg/kg; Sigma-Aldrich, St. Louis, MO, United States) while controls were administered the same volume of sodium citrate buffer (0.25 mL/kg; Solarbio, Beijing, China). If experimental canines maintained normoglycemia (defined as fasting blood glucose levels < 5.5 mmol/L) within 1-4 months after the initial modeling, a supplemental dose of STZ (30 mg/kg) was administered to reinforce the model. A diabetic model was considered successfully established when the fasting blood glucose concentration measured by a portable blood glucose meter was consistently maintained at 11.1 mmol/L or higher[21]. Following successful model induction, the canines were divided into a DM group with confirmed hyperglycemia (n = 6) and a healthy control group (n = 3). The time from DM development to electrical stimulation device implantation was 0-24 (15.3 ± 10.5) months on average, and the body weights before the electrical stimulation device implantation were 9.5-14.0 kg and 11.9-14.1 kg in the two groups.

Electrical stimulation device implantation

All canines underwent minor abdominal surgery to implant the electrical stimulation device under general anesthesia. A pair of stimulating electrodes was buried in the serous muscular layer of the antrum near the greater curvature 4 cm above the pylorus at a distance of 1 cm. A pulse generator (G101A, Beijing PINS Medical Co., Ltd., Beijing, China) was placed in the capsular bag in the myometrium-fascial space of the left upper abdominal wall which was bluntly separated. Following implantation, there was a 2-week postoperative recovery period (Figure 1).

Figure 1 Experimental protocol and intervention. GES: Gastric electrical stimulation; SGES: Sham gastric electrical stimulation.

Electrode stimulation

This study was a randomized, controlled, crossover design. All beagle canines underwent a stimulation phase (labeled GES) or a sham GES (SGES) phase (labeled SGES) of 4 weeks each in a randomized order. At the end of the first 4-week phase of GES or SGES, there was a 3-week washout to avoid carry-over effects as well as post biopsy operative recovery, followed by crossover for 4 weeks of the alternate stimulation. GES was continuously applied with high-frequency, short-pulse stimulation (0.33 millisecond, 5 mA, 15 Hz, 0.2 second on, 3 seconds off). The procedures for SGES were the same as those for GES except the stimulators were not turned on. The timeline of the experiment is shown in Figure 1.

Gastric antrum and pancreatic tissues biopsies

Full-thickness gastric antrum biopsies of 1 cm × 0.5 cm were obtained at the end of first phase treatment under general anesthesia, and the second antrum biopsies and pancreas tissue biopsies were conducted at the end of the study (Figure 1). Based on different stimulations, the gastric antrum tissues were grouped into the following four subgroups: (1) DM with GES group (DM + GES); (2) DM with SGES group (DM + SGES); (3) Control with GES group (control + GES); and (4) Control with SGES group (control + SGES).

Hematoxylin-eosin staining

Sections from the pancreatic tail, embedded in paraffin, were dewaxed using xylene, rehydrated with a series of graded ethanol, and rinsed with distilled water. The eosin dye solution was applied after rinsing with running water for 10 minutes and allowed to react for 2 minutes to stain the cytoplasm, and then dehydration and mounting were performed for microscopic examination. Pathological changes in the pancreatic islets of diabetic and control canines were detected using hematoxylin-eosin (HE) staining.

Immunofluorescence staining

The full-thickness stomach tissue was slightly stretched and then fixed in 40 g/L formaldehyde, embedded in paraffin blocks, and sliced at a thickness of 4 μm. Consecutive sections were grouped into sets of 10 slices each, resulting in a total of 5 groups. From each group, the 1^st, 2^nd, 3^rd, and 4^th sections were used for four specific neuronal stainings (ChAT, VIP, nNOS and CALR), respectively. Immunofluorescence staining was employed to examine the distribution and proportions of neurons in the MP of diabetic canines (GES, SGES) and control groups (GES, SGES). Paraffin sections were kept in dewaxing agent and deparaffinized each for 3 × 20 minutes and gradually hydrated in a series of graded alcohol. Epitope retrieval was carried out by boiling sections in sodium citrate buffer (0.01M, pH 6.0) for 20 minutes and cooling down to room temperature. Sections were blocked with 10% normal goat serum in phosphate-buffered saline (PBS) containing 0.3% Triton X-100 at room temperature for 2 hours. The tissues were incubated overnight at 4 °C in a mixture of mouse anti-human neuronal protein HuC/HuD (anti-Hu) antiserum and a rabbit primary antibody for neuronal staining. Anti-Hu was utilized as a general neuronal marker for enteric neurons. Primary antibodies targeting ChAT, VIP, nNOS, and CALR were employed to identify specific neurons based on chemical markers: (1) ChAT indicating cholinergic excitatory motor neurons; (2) VIP for inhibitory secretomotor neurons; (3) The nNOS for inhibitory nitrergic neurons[22,23]; and (4) CALR for sensory neurons. The tissues were washed three times for 10 minutes each in PBS and then incubated with secondary antibodies conjugated with fluorescein isothiocyanate (FITC) and tetramethyl rhodamine isothiocyanate for 2 hours in the dark. To ensure proper labeling, the tissues were washed three times for 10 minutes each with PBS and then examined using a fluorescence microscope (Axio imager Z2, Carl Zeiss AG, Oberkochen, Germany). The details of the primary and secondary antibodies, including their dilutions and sources, are provided in Table 1. The outline of the ganglions in the MP was clear and distinguishable with cluster anti-Hu and specific chemical marker IR positive neurons under the fluorescence microscope. At a magnification of 10 × 20, non-overlapping fields-of-view were used across the sections, with 5 fields in the MP from each section and 25 identical fields from each biopsy to ensure that no less than 250 neurons with anti-Hu-IR positive staining were counted. The numbers of neurons and positive neurons for other specific antibodies in each of the 25 fields of the gastric tissue samples were counted by two researchers who had received uniform standard training in a blinded manner. The results were displayed as a percentage of the total neurons that were labeled with anti-Hu.

Table 1 Antibodies used for immunofluorescence experiments.

Antigen	Host	Code	Dilution	Source
Anti-human neuronal protein HuC/HuD	Mouse	A-21271	1:200	Invitrogen
Choline acetyltransferase	Rabbit	20747-1-AP	1:200	Proteintech
Vasoactive intestinal peptide	Rabbit	Ab22736	1:1000	Abcam
Neuronal nitric oxide synthase	Rabbit	T55342S	1:200	Abmart
Calretinin	Rabbit	AB5054	1:1000	Chemicon
Mouse IgG	Goat tetramethyl rhodamine isothiocyanate	SA00013-3	1:200	Proteintech
Rabbit IgG	Goat fluorescein isothiocyanate	SA00013-2	1:200	Proteintech

IgG: Immunoglobulin G.

Statistical analysis

Results are shown as mean ± SD; n refers to the number of canines analyzed. Differences between the GES and SGES subgroups were evaluated using the paired-sample t-test in SPSS 25.0 (SPSS Inc., Chicago, IL, United States). For comparisons between the DM and control groups, Levene’s test was first applied to assess the homogeneity of variances. If the variances were equal, an independent samples t-test was used; if variances were unequal, Welch’s corrected t-test was employed. Multiple comparison correction was not applied as this was an exploratory study with only a few planned comparisons based on our protocol. Differences were considered statistically significant at P < 0.05. All graphs were created by commercial software (GraphPad Prism version 5.00 for Windows, GraphPad Software Inc., La Jolla, CA, United States).

RESULTS

Pancreatic islets in diabetic models and controls

Six beagle canines developed DM following initial and subsequent booster modeling. We observed the shape of islets in the pancreatic tail tissues. The HE staining demonstrated that diabetic canines exhibited a significant reduction in the number and area of pancreatic islets with irregular, ill-defined contours and atrophied cells, which confirmed successful establishment of the DM model (Figure 2A). Representative HE staining of pancreatic tail tissues in healthy controls is shown in Figure 2B.

Figure 2 Representative hematoxylin-eosin staining of pancreatic tail tissues in the diabetic models and controls. A: Diabetes mellitus; B: Control. Scale bar = 50 µm. DM: Diabetes mellitus.

Gastric enteric ganglia and neurons

We observed the number of enteric ganglia, the total number of neurons (anti-Hu-IR positive neurons), and any specific markers (i.e., ChAT, VIP, nNOS, and CALR positive neurons) in the relatively consecutive and consistent MP of antral double-stained sections. There were no significant differences in the numbers of enteric ganglia between the GES and the SGES subgroups in diabetic canines or controls (all P > 0.05; Table 2). The total number of enteric neurons was not significantly different in anti-Hu with any specific markers double-staining analyses between the GES and the SGES subgroups, whether in the diabetic canines or the controls (all P > 0.05; Supplementary Table 1).

Table 2 Number of ganglia in the antral myenteric plexus between gastric electrical stimulation and sham gastric electrical stimulation in diabetic canines and controls, mean ± SD.

Ganglia	Diabetes mellitus (n = 6)			Controls (n = 3)
	GES	SGES	P value	GES	SGES	P value
Choline acetyltransferase	16.8 ± 3.0	14.8 ± 2.8	0.292	13.8 ± 3.5	12.3 ± 0.8	0.538
Vasoactive intestinal peptide	16.9 ± 2.7	14.2 ± 3.2	0.130	14.1 ± 3.6	12.5 ± 0.8	0.435
Neuronal nitric oxide synthase	16.1 ± 2.8	14.5 ± 3.1	0.372	14.0 ± 3.9	12.0 ± 0.8	0.539
Calretinin	16.7 ± 2.7	14.8 ± 3.0	0.190	14.4 ± 3.6	11.9 ± 0.5	0.381

GES: Gastric electrical stimulation; SGES: Sham gastric electrical stimulation.

Effects of GES on myenteric ChAT neurons

Myenteric ChAT-IR positive neurons were stained in the cytoplasm, they were the most important cellular component of the MP with a proportion of over 65% in the four subgroups. There was no significant difference in the total number of ChAT-IR positive neurons between the GES intervention group and SGES subgroup across both the diabetic group and control group (P = 0.127; P = 0.390, respectively; Figure 3A-N and Supplementary Table 1). Following GES, the proportion of ChAT-IR positive neurons was significantly higher compared to SGES (73.3% ± 1.7% vs 67.5% ± 1.8%, P = 0.002; Figure 3O and Supplementary Table 1) in diabetic canines, but this neuromodulatory effect was absent in normoglycemic canines (74.7% ± 1.4% vs 73.3% ± 1.8%, P = 0.255; Figure 3O and Supplementary Table 1). In the sham-stimulated subgroups, diabetic canines exhibited a significantly lower proportion of ChAT-IR positive neurons compared to controls (67.5% ± 1.8% vs 73.3% ± 1.8%, P = 0.003; Figure 3O and Supplementary Table 1).

Figure 3 Representative immunofluorescence images of myenteric choline acetyltransferase-immunoreactive positive neurons in full-thickness sections of canine gastric antrum (200 × magnification, scale bar = 50 μm). A: Representative images of anti-human neuronal protein HuC/HuD-immunoreactive (anti-Hu-IR) labeled neurons in diabetes mellitus (DM) + gastric electrical stimulation (GES); B: Representative images of choline acetyltransferase-immunoreactive (ChAT-IR) positive neurons in DM + GES; C: Representative merged images of anti-Hu-IR/ChAT-IR in DM + GES; D: Representative images of anti-Hu-IR labeled neurons in DM + sham GES (SGES); E: Representative images of ChAT-IR positive neurons in DM + SGES; F: Representative merged images of anti-Hu-IR/ChAT-IR in DM + SGES; G: Representative images of anti-Hu-IR labeled neurons in control + GES; H: Representative images of ChAT-IR positive neurons in control + GES; I: Representative merged images of anti-Hu-IR/ChAT-IR in control + GES; J: Representative images of anti-Hu-IR labeled neurons in control + SGES; K: Representative images of ChAT-IR positive neurons in control + SGES; L: Representative merged images of anti-Hu-IR/ChAT-IR in control + SGES; M: Number of anti-Hu-IR positive neurons; N: Number of ChAT-IR positive neurons; O: Proportion of ChAT-IR positive neurons relative to total neurons. The proportion of ChAT-IR neurons was higher in the diabetes mellitus + GES compared with the diabetes mellitus + SGES subgroup (^bP < 0.01). Within sham-stimulated subgroups, the proportion of ChAT-IR neurons was less in the diabetic canines compared with the controls (^bP < 0.01). Anti-Hu: Anti-human neuronal protein HuC/HuD; ChAT: Choline acetyltransferase; DM: Diabetes mellitus; GES: Gastric electrical stimulation; SGES: Sham gastric electrical stimulation.

Effects of GES on myenteric VIP-IR neurons

Myenteric VIP-IR positive neurons exhibited cytoplasmic dot-like staining and accounted for over 40% of the neuronal population in the four subgroups. No significant differences were observed in the count of VIP-IR positive neurons between the GES intervention and SGES across both the diabetic and control group (P = 0.223; P = 0.581, respectively; Figure 4A-N and Supplementary Table 1) as well as the proportion of VIP-IR positive neurons between GES and SGES among diabetic models or controls (P = 0.054; P = 0.227, respectively; Figure 4O and Supplementary Table 1). Within the GES subgroups, diabetic canines exhibited a significantly higher proportion of VIP-IR positive neurons compared to controls (50.8% ± 1.3% vs 44.9% ± 0.8%, P < 0.001; Figure 4O and Supplementary Table 1), and this phenomenon was also observed in the SGES subgroups (51.9% ± 0.8% in diabetic canines vs 46.8% ± 1.2% in controls, P < 0.001; Figure 4O and Supplementary Table 1).

Figure 4 Representative immunofluorescence images of myenteric vasoactive intestinal peptide-immunoreactive positive neurons in full-thickness sections of canine gastric antrum (200 × magnification, scale bar = 50 μm). A: Representative images of anti-human neuronal protein HuC/HuD-immunoreactive (anti-Hu-IR) labeled neurons in diabetes mellitus (DM) + gastric electrical stimulation (GES); B: Representative images of vasoactive intestinal peptide-immunoreactive (VIP-IR) positive neurons in DM + GES; C: Representative merged images of anti-Hu-IR/VIP-IR in DM + GES; D: Representative images of anti-Hu-IR labeled neurons in DM + sham GES (SGES); E: Representative images of VIP-IR positive neurons in DM + SGES; F: Representative merged images of anti-Hu-IR/VIP-IR in DM + SGES; G: Representative images of anti-Hu-IR labeled neurons in control + GES; H: Representative images of VIP-IR positive neurons in control + GES; I: Representative merged images of anti-Hu-IR/VIP-IR in control + GES; J: Representative images of anti-Hu-IR labeled neurons in control + SGES; K: Representative images of VIP-IR positive neurons in control + SGES; L: Representative merged images of anti-Hu-IR/VIP-IR in control + SGES; M: Number of anti-Hu-IR positive neurons; N: Number of VIP-IR positive neurons; O: Proportion of VIP-IR positive neurons relative to total neurons. The proportion of VIP-IR neurons was significantly higher in the diabetes mellitus canines compared with controls, whether in GES or SGES subgroups (^cP < 0.001). Anti-Hu: Anti-human neuronal protein HuC/HuD; DM: Diabetes mellitus; GES: Gastric electrical stimulation; IR: Immunoreactive; SGES: Sham gastric electrical stimulation; VIP: Vasoactive intestinal peptide.

Effects of GES on the myenteric nNOS-IR neurons

Approximately 30% of myenteric neurons in the canine gastric antrum were nNOS-IR positive neurons with staining confined to the cytoplasm, and the number of these neurons was similar between the GES and SGES subgroups across both the DM models and controls (P = 0.070, P = 0.282, respectively; Figure 5A-N and Supplementary Table 1). GES intervention resulted in a significantly higher proportion of nNOS-IR positive neurons compared to the SGES subgroup (32.1% ± 1.7% vs 27.3% ± 1.4%, P = 0.002; Figure 5O and Supplementary Table 1) in diabetic canines; this neuromodulatory effect was also found in healthy canines (37.5% ± 1.3% vs 32.1% ± 0.7%, P = 0.036; Figure 5O and Supplementary Table 1), and there was no significant difference in this proportion between the two GES subgroups in diabetic and healthy canines. Under the same sham stimulation, the percentage of nNOS-IR positive neurons was markedly lower in the diabetic canines compared with controls (27.3% ± 1.4% vs 32.1% ± 0.7%, P = 0.001; Figure 5O and Supplementary Table 1).

Figure 5 Representative immunofluorescence images of myenteric neuronal nitric oxide synthase-immunoreactive positive neurons in full-thickness sections of canine gastric antrum (200 × magnification, scale bar = 50 μm). A: Representative images of anti-human neuronal protein HuC/HuD-immunoreactive (anti-Hu-IR) labeled neurons in diabetes mellitus (DM) + gastric electrical stimulation (GES); B: Representative images of neuronal nitric oxide synthase-immunoreactive (nNOS-IR) positive neurons in DM + GES; C: Representative merged images of anti-Hu-IR/nNOS-IR in DM + GES; D: Representative images of anti-Hu-IR labeled neurons in DM + sham GES (SGES); E: Representative images of nNOS-IR positive neurons in DM + SGES; F: Representative merged images of anti-Hu-IR/nNOS-IR in DM + SGES; G: Representative images of anti-Hu-IR labeled neurons in control + GES; H: Representative images of nNOS-IR positive neurons in control + GES; I: Representative merged images of anti-Hu-IR/nNOS-IR in control + GES; J: Representative images of anti-Hu-IR labeled neurons in control + SGES; K: Representative images of nNOS-IR positive neurons in control + SGES; L: Representative merged images of anti-Hu-IR/nNOS-IR in control + SGES; M: Number of anti-Hu-IR positive neurons; N: Number of nNOS-IR positive neurons; O: Proportion of nNOS-IR positive neurons relative to total neurons. The proportion of nNOS-IR neurons was increased in the GES subgroups compared with sham GES in both diabetes mellitus (^bP < 0.01) and control canines (^aP < 0.05). Within sham-stimulated groups, the proportion of nNOS-IR neurons was less in the diabetes mellitus compared with the controls (^bP < 0.01). Anti-Hu: Anti-human neuronal protein HuC/HuD; DM: Diabetes mellitus; GES: Gastric electrical stimulation; nNOS: Neuronal nitric oxide synthase; SGES: Sham gastric electrical stimulation.

Effects of GES on myenteric CALR-IR neurons

Myenteric CALR-IR positive neurons exhibited intense cytoplasmic staining with low fluorescence intensity in the nuclear region, clearly delineating nuclear contours. No significant difference was detected in the total number of CALR-IR positive neurons after GES when compared with the SGES subgroup in both the DM group and the control group (P = 0.174, P = 0.326, respectively; Figure 6A-N and Supplementary Table 1). The percentage of CALR-IR positive neurons was similar between the GES and the SGES subgroups in the DM canines (27.3% ± 1.4% vs 31.5% ± 2.0%, P = 0.631; Figure 6O and Supplementary Table 1). This phenomenon was also observed in the controls (32.2% ± 3.3% in the GES subgroup vs 31.1% ± 1.6% in the SGES subgroup, P = 0.508; Figure 6O and Supplementary Table 1).

Figure 6 Representative immunofluorescence images of myenteric calretinin-immunoreactive positive neurons in full-thickness sections of canine gastric antrum (200 × magnification, scale bar = 50 μm). A: Representative images of anti-human neuronal protein HuC/HuD-immunoreactive (anti-Hu-IR) labeled neurons in diabetes mellitus (DM) + gastric electrical stimulation (GES); B: Representative images of calretinin-immunoreactive (CALR-IR) positive neurons in DM + GES; C: Representative merged images of anti-Hu-IR/CALR-IR in DM + GES; D: Representative images of anti-Hu-IR labeled neurons in DM + SGES; E: Representative images of CALR-IR positive neurons in DM + SGES; F: Representative merged images of anti-Hu-IR/CALR-IR in DM + SGES; G: Representative images of anti-Hu-IR labeled neurons in control + GES; H: Representative images of CALR-IR positive neurons in control + GES; I: Representative merged images of anti-Hu-IR/CALR-IR in control + GES; J: Representative images of anti-Hu-IR labeled neurons in control + SGES; K: Representative images of CALR-IR positive neurons in control + SGES; L: Representative merged images of anti-Hu-IR/CALR-IR in control + SGES; M: Number of anti-Hu-IR positive neurons; N: Number of CALR-IR positive neurons; O: Proportion of CALR-IR positive neurons relative to total neurons. Anti-Hu: Anti-human neuronal protein HuC/HuD; CALR: Calretinin; DM: Diabetes mellitus; GES: Gastric electrical stimulation; IR: Immunoreactive; SGES: Sham gastric electrical stimulation.

DISCUSSION

In this study, a diabetic model was established using large beagle canines. We observed a reduced proportion of ChAT and nNOS-IR positive neurons in the antrum of diabetic beagles. Furthermore, it was noted that the percentage of VIP-IR positive neurons was increased compared with the non-diabetic controls. GES partly reversed these pathological alterations and improved the survival of cholinergic and nitrergic neurons in both the DM groups and controls.

Gastroparesis is the most common GI complication of diabetes. The most accepted mechanisms of DGP are defects in the ICC and changes in enteric neurons, especially musculomotor neurons (i.e., ChAT) and inhibitory neurons (i.e., nNOS)[24]. ChAT affects the neurotransmitter acetylcholine, which activates smooth muscle M3 receptors, thereby increasing the frequency and amplitude of GI contractions[25]. VIP and NO are the key inhibitory neurotransmitters in the MP, which facilitate the receptive relaxation of the gastric fundus, modulate antral contractions, and trigger pyloric relaxation[26,27]. CALR is expressed in a subset of myenteric neurons-including cholinergic motor neurons, interneurons, and sensory neurons, and is crucial for detecting mucosal stimuli and coordinating peristaltic reflexes[28].

Previous studies demonstrated a time-dependent reduction in ChAT expression in the colon of STZ-induced diabetic rats, an increase in VIP-IR positive neurons in the duodenal, jejunal, and colonic tissues of the diabetic porcine model, and a decrease in nNOS expression in the antral MP of DGP rats[9,12,14]. However, there are no experimental data on the changes in CALR-IR positive neurons in the ENS of diabetic models. The impairments in gastric motor functions caused by a high-fat diet seem to be linked to: (1) A reduction in myenteric neurons; (2) Decreased nNOS immunoreactivity in the antrum and fundus; (3) A downregulation of antral ChAT expression; and (4) Increased VIP in the myenteric neurons of the antrum[29]. In diabetes, elevated reactive oxygen species cause the loss and degeneration of nNOS-IR positive neurons, collectively impairing NO-mediated inhibitory neurotransmission. This disruption results in antral hypercontractility, compromised coordination of pyloric sphincter relaxation, and ultimately delayed GE, along with gastric stasis and hypertrophic pyloric stenosis[30,31].

In this study, we discovered that the morphological changes in gastric musculomotor and inhibitory neurons are also present in STZ-induced diabetic beagles. These changes included a marked reduction in the proportions of ChAT- and nNOS-IR positive neurons and a significant increase in VIP-IR positive neurons within the antral MP, even though the diabetic canines did not show obvious manifestations of gastroparesis (according to the results of radionuclide GE scintigraphy). Some studies have shown that the quantity of VIP-IR positive neurons and the level of VIP in the antrum of diabetic animals are significantly reduced[32,33]. Our results are contrary to these findings but are consistent with the study by Bulc et al[12]. Our analysis revealed no significant differences in the proportion of CALR-IR positive neurons between the diabetic and non-diabetic groups. These results are speculated based on the comparison of the proportions of specific neurons between diabetic and control canines receiving the same SGES. The lack of baseline gastric antral histopathology is due to welfare and ethical considerations in animals.

GES is a treatment option for managing refractory gastroparesis, showing significant improvements in key symptoms such as nausea and vomiting, as well as in glucose regulation[34]. The therapeutic mechanisms of GES therapy may involve multifactorial neuromodulatory effects, including an increase in vagal efferent autonomic activity, an enhancement of gastric accommodation, and the potential restoration of ICC proliferation[6,18,35]. Can short-term GES therapy similarly ameliorate the damage to enteric neurons in diabetic beagle canines, as observed by Chen et al[10] in diabetic rats?

We found that GES increased the percentage of ChAT-IR positive neurons in DM models compared to SGES. These findings suggest that GES may regulate abnormal gastric motility by promoting smooth muscle contraction, providing non-pharmacological therapeutic evidence for GES treatment in DGP. However, this neuromodulatory effect was not observed in healthy canines. This pathology-dependent neuromodulation implies that GES counteracts hyperglycemia-induced transcriptional silencing (i.e., histone deacetylase activation) rather than broadly enhancing cholinergic plasticity. Our current study also demonstrated that 4-week GES restored the lost proportion of nNOS-IR positive neurons in both diabetic and control canines. A previous study indicated that nNOS dysregulation contributes to gastro-pyloric dysfunction in diabetic models, and insulin treatment restores its expression and function to normal levels[36]. Xing and Chen[17] provided clear evidence that long-pulse GES has an inhibitory impact on gastric tone and accommodation in healthy canines, with partial mediation through vagal and inhibitory nitrergic pathways. We propose that a selective increase in nNOS-IR neurons likely enhances receptive relaxation, gastric accommodation, and antro-pyloro-duodenal coordination. Notably, the proportions of VIP-IR and CALR-IR positive neurons remained unchanged after stimulation in both the diabetic and control groups. We consider that the stability of CALR-IR neuronal subtypes suggests that GES does not broadly alter the sensory neurons. Instead, its effect is highly selective neuromodulation on cholinergic and nitrergic neurons that regulate the abnormality of gastric motility and accommodation, without alterations in fundamental sensory architecture. Of course, these results may also be ascribed to the relatively short duration of GES (i.e., 4 weeks), which might be inadequate to observe long-term neuroplasticity or functional recovery.

This study provides the first evidence that GES significantly increased the proportions of cholinergic and nitrergic neurons within the gastric MP of diabetic beagle canines. As beagles have a higher physiological, anatomical, and metabolic similarity to humans than other small animals, we believe that the results obtained from canine models might be more representative in relation to humans. However, this study has several limitations. First, the limited sample size of animals (DM, n = 6; controls, n = 3) may reduce statistical power and limit the generalizability of the findings. The small sample size was mainly due to ethical constraints in large-animal research and difficulties in blood glucose management with long-term diabetic models. Secondly, the lack of exploration on molecular mechanisms related to neurotrophic signaling pathways, oxidative stress, and inflammatory markers precludes definitive conclusions about the molecular mechanisms responsible for the observed phenotypic changes. Thirdly, diabetic beagles did not exhibit delayed GE before GES, and no significant changes in GE were observed after GES (data not shown). Therefore, the relationship between the observed neuronal alterations and motility improvements needs to be confirmed in the gastroparesis models. Incorporating cellular mechanisms and functional assessments into our study is crucial for determining how GES promotes neuronal survival and whether GES affects gastric functions.

CONCLUSION

In summary, diabetic canines exhibited notable reductions in both ChAT-IR and nNOS-IR positive neuronal populations in the antral MP compared to controls, along with an upregulation of the proportion of VIP-IR positive neurons. Importantly, GES intervention had neuroprotective effects, effectively improving the survival of both cholinergic and nitrergic neurons in diabetic groups. These findings offer a theoretical foundation for the clinical application of GES in the treatment of DGP, especially through its neuromodulatory impacts on enteric neuronal plasticity.

ACKNOWLEDGEMENTS

The authors would like to acknowledge Beijing PINS Medical Co., Ltd., Beijing, China for their providing gastric electrical stimulation device and technical assistance.