Vagal nerve stimulation (VNS) has been shown to limit intestinal inflammation following injury; however, a direct connection between vagal terminals and resident intestinal immune cells has yet to be identified. We have previously shown that enteric glia cell (EGC) expression is increased after injury through a vagal-mediated pathway to help restore gut barrier function. We hypothesize that EGCs modulate immune cell recruitment following injury and relay vagal anti-inflammatory signals to resident immune cells in the gut. EGCs were selectively ablated from an isolated segment of distal bowel with topical application of benzalkonium chloride (BAC) in male mice. Three days following BAC application, mice were subjected to an ischemia-reperfusion injury (I/R) by superior mesenteric artery occlusion for 30 min. VNS was performed in a separate cohort of animals. EGC+ and EGC− segments were compared utilizing histology, flow cytometry, immunohistochemistry, and intestinal permeability. VNS significantly reduced immune cell recruitment after I/R injury in EGC+ segments with cell percentages similar to sham. VNS failed to limit immune cell recruitment in EGC− segments. Histologic evidence of gut injury was diminished with VNS application in EGC+ segments, whereas EGC− segments showed features of more severe injury. Intestinal permeability increased following I/R injury in both EGC+ and EGC− segments. Permeability was significantly lower after VNS application compared with injury alone in EGC+ segments only (95.1 ± 30.0 vs. 217.6 ± 21.7 μg/ml, P < 0.05). Therefore, EGC ablation uncouples the protective effects of VNS, suggesting that vagal-mediated signals are translated to effector cells through EGCs.
NEW & NOTEWORTHY Intestinal inflammation is initiated by local immune cell activation and epithelial barrier breakdown, resulting in the production of proinflammatory mediators with subsequent leukocyte recruitment. Vagal nerve stimulation (VNS) has been shown to limit intestinal inflammation following injury; however, direct connection between vagal terminals and resident intestinal immune cells has yet to be identified. Here, we demonstrate that intact enteric glia cells are required to transmit the gut anti-inflammatory effects of VNS.
- vagal anti-inflammatory
- entric neurons
- enteric glia cells
the gut is increasingly recognized as a major contributor to the development of the systemic inflammatory response (SIRS) and end organ damage following injury (11, 30, 40, 46). Severe trauma/hemorrhagic shock or cutaneous burn results in splanchnic vasoconstriction, which produces an ischemia-reperfusion (I/R) injury (37, 51). I/R injury generates a local inflammatory response in the gut with activation of resident immune cells and release of cytotoxic and chemotactic factors, further promoting immune cell infiltration (4, 12, 14, 20). The release of cytokines, free oxygen radicals, and vasoactive substrates can be damaging to gut epithelial cells (54). A combination of epithelial cell apoptosis and loss of tight junction proteins compromises gut barrier function, which allows for bacterial and antigen translocation and further propagates the inflammatory response (9, 28). If unregulated, the local inflammatory response can lead to the production of proinflammatory mediators that travel through the mesenteric lymph into the systemic circulation (29, 33). Once in the systemic circulation, these mediators can trigger a SIRS response and cause injury to distant organs, such as the development of acute lung injury (23, 56).
To limit inflammation, the body has a reflexive, anti-inflammatory circuit mediated through the vagus nerve (53). In this pathway, inflammatory signals are transmitted to the brain through afferent vagal nerve fibers and respond by triggering anti-inflammatory signals from efferent vagal fibers back to the gut. This vagal-mediated anti-inflammatory pathway can be augmented through the application of direct current or pharmacologic vagus nerve stimulation (VNS) (21, 27). VNS has been shown to prevent gut barrier breakdown (7, 22), limit intestinal injury, reduce mesenteric lymph toxicity (25, 26, 33), and alleviate the development of SIRS and end organ damage following injury.
The mechanism through which the vagus nerve is capable of producing an anti-inflammatory response is incompletely understood. Cholinergic signaling requires close proximity of vagal terminals to target effector cells (10). Vagal branches to the intestine terminate in the myenteric plexus, but efferent projections have not been observed in close proximity to resident immune cells (31). Furthermore, vagal extension beyond the myenteric plexus to the intestinal mucosa has not been identified, raising questions as to the mechanism by which VNS is capable of affecting mucosal integrity and the expression of tight junction proteins (6, 7).
The enteric nervous system (ENS), which includes enteric glial cells (EGCs) and enteric neurons, is increasingly recognized for its role in regulating various physiologic processes in the gut, including the inflammatory response (47). EGCs outnumber enteric neurons, are positioned throughout the intestine with projections extending to villus tips, can receive and transmit a myriad of signals from neighboring cells, and thus are prime candidates to modulate the gut inflammatory response following injury (44). Additionally, EGCs form a rich network surrounding both enteric neurons and efferent vagal fibers. We have previously demonstrated that VNS increases the expression of EGC markers that was associated with improved gut barrier function after injury (6, 24); however, the relative contribution of EGCs to vagal-mediated gut protection is unknown. Here, we hypothesized that EGCs modulate immune cell mobilization following injury and are required to relay vagal anti-inflammatory signals to the gut.
MATERIALS AND METHODS
Male C57BL/6 mice, 8 to 10 wk old (Jackson Laboratories, Sacramento, CA), were housed in our Association for Assessment and Accreditation of Laboratory Animal Care-accredited animal facility and were exposed to a 12-h light-dark cycle. They were provided food (Teklad Rodent Diet 8604; Envigo) and water ad libitum. On the day of the experiment, animals were transferred to the animal surgery facility and placed under general anesthesia using inhaled isoflurane. Animals were given a subcutaneous injection of 1.5 ml of normal saline with buprenorphine for analgesia. The abdomen was shaved with an electrical clipper and cleansed with betadine solution. The mice were secured onto a heating pad to maintain appropriate body temperature during anesthesia. A midline laparotomy was performed to expose the small bowel. A 5-cm segment of ileum was isolated and marked at the proximal and distal ends and at the midpoint with loose sutures. The surrounding bowel was replaced into the abdominal cavity and protected from inadvertent exposure to topical agents. Sterile gauze was soaked in 0.01% benzalkonium chloride (BAC) in phosphate-buffered saline (PBS), 0.05% BAC, or PBS and layered around the isolated segment of bowel and mesentery for 15 min as previously described (57). After gauze removal, the bowel was thoroughly rinsed with sterile saline and returned to the abdominal cavity. The abdomen was closed and the animals were recovered from anesthesia on a heating pad. The animals were given free access to food and water 6 h after BAC application and monitored daily. Animal experiments were approved of by the University of California animal research committee.
Intestinal I/R injury.
Animals underwent repeat general anesthesia and reopening of their abdominal incision 3 days following BAC application. Food and water was withheld from animals 6 h before reoperation. A medial visceral rotation was performed to expose the superior mesenteric artery. Superior mesenteric artery occlusion (SMAO) was performed with an atraumatic vascular clamp (Fine Science Tools, Foster City, CA) for 30 min. Sham animals underwent medical visceral rotation and anesthesia for 30 min without SMAO. In a separate cohort of animals, VNS was performed through a right cervical incision 15 min after ischemia. The vagus nerve was stimulated for 10 min using a square-wave generator at 5 V, with a frequency of 5 Hz. Following the experimental protocol, tissues were harvested for subsequent analysis. Mice were euthanized using overdose of inhaled anesthetic (isoflurane) followed by cervical dislocation.
Enzymatic gut digestion and flow cytometry.
BAC-treated and -untreated segments of bowel were separately collected 4 h after reperfusion. Bowel segments were rinsed of luminal contents, cut into 2- to 4-cm lengths, and incubated in a predigestion solution of 5 mM DTT at 37oC for 20 min to remove mucous. Gut segments were then washed three times with 5 mM EDTA for 15 min to remove epithelial cells. Minced tissue was transferred to the enzyme solution collagenase A/dispase II and incubated at 37oC for an additional 30 min (16). The gut tissue was then passed through a 70-um filter and digestion was quenched with 5% fetal bovine serum solution.
Cell surface staining for flow cytometry was performed on fresh cells in PBS. Digested gut cells were stained for the markers MHCII-FITC (cat. no. 553623; BD Biosciences, San Jose, CA) (1), CD45-PE (cat. no. 12-0451-82; eBioscience, San Diego, CA), CD11b-APC-Cy7 (cat. no. 561039; BD Biosciences), and CD11c-APC (cat. no. 561119; BD Biosciences) to distinguish dendritic cells (DCs) and macrophages (M0) as previously described (15). Neutrophils were identified by staining with CD45-FITC (cat. no. 553079; BD Biosciences), Ly6G-PE (cat. no. 551461; BD Biosciences) (13), CD11b-APC-Cy7, and CD11c-APC as described previously (16). All flow cytometry antibodies were used at 1:200 dilution for 30 min and washed before acquisition on a BD Accuri C6 flow cytometer (BD Biosciences). Quantification of enteric neurons and EGCs was performed with flow cytometry as previously described (5, 36). Digested gut cells were fixed and permeabilized with BD Cytofix/Cytoperm according to the manufacturer’s instruction before staining with NeuN-FITC (neuronal nuclei marker; cat. no. ab190195; Abcam, Cambridge, MA) and S100B-PE-Cy7 (intracellular EGC marker; cat. no. AC12-0160-17; Abcore, Ramona, CA). Analysis was performed using BD Accuri C6 software (BD Biosciences).
Histology and immunohistochemistry.
For section mount imaging, BAC-treated and -untreated segments of gut were fixed in neutral buffered formalin for 24 h and then embedded in paraffin before being sectioned (n = 5 animals per group). For histologic evaluation, sections of paraffin were stained with hematoxylin and eosin by the University of California, San Diego Histology Core Services (n = 5 mice per experimental condition). Histological gut injury was scored by three independent researchers blinded to the experimental conditions. Three randomly selected fields from each specimen (n = 3 per experimental condition) were scored for evidence of intestinal injury. Scoring ranged from 0 to 4, as follows: 0 = normal, no evidence of injury; 1 = mild, focal epithelial edema; 2 = moderate, diffuse swelling necrosis of the villi; 3 = severe; diffuse pathology of the villi; and 4 = major; widespread injury with massive inflammatory cell infiltration as previously described (8).
Immunohistochemistry of the myenteric plexus was performed by first dissecting longitudinal muscles and the myenteric plexus from submucosal tissue as described in previous techniques (49) (n = 3 animals per group). The dissected tissue was fixed in 0.1 mol/l PBS containing 4% paraformaldehyde at room temperature for 30 min and fixed on glass slides. The tissue was then washed with PBS before blocking for 30 min with 3% bovine serum albumin (BSA; Sigma) and incubated overnight with primary antibody (1:200). Primary antibodies included goat anti-protein gene product 9.5 (PGP 9.5), a neuronal marker, and anti-glial fibrillary acidic protein (GFAP) PE-Cy7, a glial marker, (Abcore). Gut tissue was then treated with the secondary antibody Alexa Fluor 488 (chicken anti-goat IgG; Life Technologies, Carlsbad, CA). Antibodies were buffered in 1% BSA for 1 h at room temperature after being washed with PBS (pH 7.4) for 5 min. Slow Fade (Invitrogen) was added before placement of coverslips. Whole mount images of the myenteric plexus were obtained using an Olympus FluoView laser scanning confocal microscope with exposure-matched settings (Advanced Software version 1.6; Olympus).
Intestinal permeability to 4-kDa FITC-dextran was measured after injury (n = 6 animals per group). Permeability in BAC+ and BAC− segments was tested in separate cohorts of animals. Before tissue harvest, the BAC-treated segment or the equivalent length segment of untreated bowel was isolated between silk ties and injected with 25 mg of FITC-dextran dissolved in 200 μl of PBS. The bowel was returned to the abdominal cavity and the skin was closed with silk suture. Animals were kept under anesthesia for 30 min, at which point blood was obtained via cardiac puncture. Blood was placed in heparinized tubes and centrifuged at 10,000 g for 10 min to obtain the plasma. FITC-dextran concentration was measured using a SpectraMax M5 (Molecular Devices) fluorescence spectrophotometer; sample fluorescence was compared with a standard curve using known concentrations of FITC-dextran.
Data are presented as means ± SE. Comparison between groups was made with a two-tailed unpaired Student’s t-test. P ≤ 0.05 was considered statistically significant.
To study the effect of EGCs on immune cell recruitment and vagal nerve signaling, we created a model of EGC ablation using BAC. BAC is an enteric neurotoxic agent that can be applied topically to the bowel and has previously been used in experimental models of Hirschsprung’s disease (39). The effects of BAC treatment were evaluated 3 days after topical exposure to evaluate gross changes in the gut (Fig. 1A). Application of 0.01% BAC resulted in phenotypic changes to the bowel with luminal dilation in the proximal untreated segments and decompression of the treated segments. These findings, reminiscent of adhesive small bowel obstructions with transition point, likely occur due to motility changes in the treated segment. Given that motility is predominantly controlled through coordinated actions of the ENS, dysmotility suggests damage to underlying ENS components.
The effects of BAC on enteric neuron and EGC quantity were assessed with flow cytometry, utilizing the markers NeuN and S100B, respectively. Application of 0.01% BAC resulted in a 78% reduction in EGCs without significantly affecting enteric neuron quantity (Fig. 1, B–D). Application of a more moderate dose of BAC (0.005%) resulted in a 41% reduction in EGCs. Enteric glial ablation following BAC application was confirmed on immunohistochemistry of dissected myenteric plexus with a dose-dependent decrease in GFAP staining, an EGC marker (Fig. 1E). PGP 9.5 staining, an enteric neuron marker, was not significantly decreased by either dose of BAC.
In previous models, complete EGC ablation resulted in a fatal fulminant jejuno-ileitis (3). Given that the aim of our study was to investigate the effects of EGC ablation on immune cell modulation, we sought to create a model in which EGC ablation was significant enough to demonstrate a differential response to injury, if present, but not significant enough to produce injury alone. On histology, application of 0.01% BAC resulted in intestinal injury, evidenced by decreased villus height, villus architectural distortion, and an increase in immune cell infiltration. These findings were not present after application of 0.005% BAC with histology similar to sham conditions (Fig. 1, F and G). As such, application of 0.005% BAC was chosen to create our model of EGC ablation for all further experiments.
To assess the effects of EGC ablation on injury response, we compared EGC- ablated (EGC−) segments of bowel to adjacent EGC intact (EGC+) segments following ischemia/reperfusion injury. In both segments, I/R caused bowel injury on histology with decreased villus height, cellular infiltration, and mucosal sloughing; however, injury was more severe in EGC− segments with near complete loss of villus architecture (see Fig. 3, A and D).
Immune cell modulation, as characterized by the percentage of leukocytes (CD45+) present in digested gut, was disparate in EGC+ vs. EGC− segments following injury. In EGC+ segments, leukocyte quantity increased from 4.4% in sham conditions to 12.3% following I/R injury. Comparatively, leukocyte recruitment to the gut was higher in the EGC− segments following injury (5.2 to 18.3%, P < 0.05 vs. EGC+; Fig. 2, A and B).
To assess the effects of EGC ablation on vagal-mediated gut protection following I/R injury, we again compared EGC+ to EGC− segments in animals treated with VNS following injury. VNS was protective against intestinal injury in EGC+ segments only. This is demonstrated on histology with maintenance of villus height and architecture, similar to histology from sham animals. In EGC− segments, VNS treatment post-I/R injury produced histologic features similar to injury alone with increased cellular infiltration, mucosal sloughing, and short, disorganized villus structure (Fig. 3, A and D).
VNS prevented the leukocyte infiltration previously observed after injury in EGC+ segments only with leukocyte quantity similar to sham animals (6.8 vs. 4.4%, P = 140) (Fig. 3, B and C). Gut leukocyte subpopulations were characterized using a previously described flow cytometry technique (Fig. 4A) (15, 16). The percentage of DCs (CH45+MHCII+CD11c+CD11b−) increased similarly in EGC+ and EGC− segments following injury and were unaffected by VNS (Fig. 4B). Macrophage (CD45+MHCII+CD11c−CD11b+) quantity, conversely, was significantly altered after treatment using VNS with reduction of macrophage recruitment to near sham levels in EGC+ segments only (Fig. 4C). In EGC− segments, macrophage quantity remained elevated in injured animals treated with VNS. There was no difference in neutrophil (CD45+CD11c+Ly6G+) recruitment to the gut following VNS in EGC+ and EGC− segments (Fig. 4D).
To determine the functional consequence of immune cell modulation following injury, we evaluated intestinal barrier function of EGC+ and EGC− segments by measuring gut permeability to 4-kDa FITC-Dextran. As previously seen, gut permeability was significantly increased following I/R injury and was similar regardless of EGC status (Fig. 5). VNS was protective against gut barrier breakdown following injury in EGC+ segments only with a 3.9-fold increase from sham compared with the 8.9-fold increase seen following injury alone (P < 0.05). In contrast, permeability following VNS in the EGC− segments was unchanged from injury alone (5.6-fold increase vs. 6.5-fold, P = 0.384) demonstrating that EGCs are required for the gut barrier protective effects of VNS.
Gut inflammation after injury is a result of interrelated processes in the gut including immune cell activation and epithelial barrier breakdown. By comparing EGC-ablated segments to adjacent nonablated segments after injury in this series of experiments, we demonstrate that EGCs are of critical importance to both immune cell recruitment and gut barrier integrity after injury.
Innate immune cell activation in the gut can occur through a variety of infectious and noninfectious pathways. Following I/R injury, tissue hypoxia results in the release of damage-associated molecular patterns (DAMPs), such as high-mobility group box 1 (HMGB1) and adenosine triphosphate (ATP) (10, 12, 17). These molecules, typically sequestered in the intracellular space, bind to pattern recognition receptors on leukocytes and result in either cell priming or activation. Primed cells exhibit increased responsiveness to subsequent stimuli, such as bacterial translocation in the case of gut ischemia (12). Activated cells release proinflammatory cytokines to promote leukocyte trafficking to the area of injury and results in release of reactive oxygen species (ROS), which are damaging to gut epithelial cells and lead to epithelial barrier breakdown (14, 32).
EGCs exhibit numerous characteristics that make them ideal candidates for immune cell modulation. EGCs exist as vast networks and can be detected in all layers of the gut wall, including the lamina propria where most resident innate immune cells reside (35, 50). Additionally, EGCs respond to inflammation and injury through proliferation (19, 24), cytokine secretion (43), and upregulation of genes associated with immune response (42). We found that leukocyte infiltration was increased in EGC-ablated segments of the bowel following injury compared with segments with intact EGCs. Furthermore, macrophage infiltration was decreased when VNS was applied after I/R injury but only in segments with an intact EGCs. These findings suggest that EGCs plays an important role in modulating the immune cell response following injury and that vagal anti-inflammatory signals may be transmitted to resident immune cells through EGCs.
Matteoli et al. (31) arrived at a similar conclusion regarding the transmission of vagal-derived anti-inflammatory signals to resident immune in the gut through an enteric nervous system intermediary. In a model of surgery-induced intestinal inflammation, VNS attenuated proinflammatory cytokine release, MPO-cell infiltration, and improved motility in wild-type, T-cell-depleted, and asplenic animals. Labeling of vagal terminals revealed extensive co-localization with the enteric nervous system but not to resident macrophages. Instead, resident macrophages were found to be in close proximity to enteric neurons, suggesting that enteric neurons relayed vagal signaling to the ultimate target of resident immune cells to limit the inflammatory response. While Matteoli et al. hypothesized the role of enteric neurons, we identified the potential target cell type as EGCs as enteric neurons were unaffected by the quantity and duration of BAC application in our model.
As previously stated, epithelial barrier function is a key component dictating the systemic inflammatory response following injury. The intestinal mucosa is exquisitely sensitive to ischemic insult and results in enterocyte damage and death within 30 min (18). The integrity of the gut barrier is reliant on a continuous lining of epithelial cells with high expression of tight junction proteins to prevent paracellular leakage. Damage to enterocytes, as occurs in I/R injury, compromises gut barrier integrity, which exposes the submucosa and local immune cells to bacteria and antigens present in the intestinal lumen. Exposure of pathogen-associated molecular patterns (PAMPs) triggers innate immune cells, particularly macrophages, to release proinflammatory cytokines and ROS (38). The overlapping processes of immune cell activation and epithelial barrier breakdown can lead to an ever increasing inflammatory response, ultimately leading to SIRS and end organ injury.
The role of EGCs in the maintenance of gut barrier function first came to light when conditional genetic EGC ablation resulted in a fatal fulminate jejuno-ileitis (3). Since that time, several studies have investigated the underlying mechanisms through which EGCs exert influence over gut mucosal integrity. ECG bodies lie <1 μM from the epithelial border and have terminal plates that appear to contact the basement membrane (58). The close proximity of EGCs to epithelial cells allows for signal transduction through secreted factors. Secretion of the trophic factor S-nitrosoglutathione (GSNO) has been identified in cultured EGC media and increases transepithelial resistance in vitro. Furthermore, exogenous administration of GSNO in EGC-ablated animal models restores barrier function and prevents inflammation (45). Additional EGC-secreted factors such as VIP and transforming growth factor B (TGF-B), have been shown to increase epithelial cell density and surface area (34).
In addition to providing support to epithelial barrier function in physiologic conditions, EGCs also respond to injury and inflammation to limit barrier breakdown and/or hasten recovery (58). EGCs secrete glial-derived neurotrophic factor (GDNF) that has strong antiapoptotic effects of both enteric neurons and epithelial cells. In the setting of inflammation, there is increased uptake of GDNF by epithelial cells, correlated with a reduction in epithelial cell apoptosis (55). Following injury, EGCs are activated with increased expression of GFAP, which is associated with an increase in tight junction proteins (6). VNS augments the activation of EGCs following injury and has been shown to increase tight junction protein expression in several models (7, 22, 59). The mechanism by which vagal signals mediate epithelial barrier funciton is, to date, unknown. Our finding of decreased intestinal permeability following VNS in EGC-intact segments suggests that vagal-mediated barrier protection occurs through EGCs rather than direct innervation.
Initial studies by Borovikova et al. (2) demonstrated that VNS limits proinflammatory cytokine release in models of sepsis by modulating splenic macrophages. The presumed pathway suggested that cholinergic signals from the vagus nerve were relayed to the splenic nerve, a purely sympathetic nerve incapable of ACh release, and ultimately acted through a cholinergic intermediary in the spleen to transmit signals to macrophages (41). In addition to the spleen-dependent pathway that mediates the systemic TNF-a response (52), an alternative pathway was discovered in our laboratory in which VNS prevented gut barrier breakdown and intestinal inflammation in the absence of the spleen (6). This pathway, termed the spleen-independent pathway, presumed a direct connection between vagal terminals in the gut and target cells. Studies supporting the spleen-independent pathway showed decreased immune cell infiltration and cytokine production, prevention of tight junction loss, and maintenance of barrier function with VNS following injury in animals that previously underwent splenectomy (8, 31). Mechanistic studies have been ongoing to determine the pathway by which VNS alters gut barrier function in light of the absence of cholinergic vagal fibers in close proximity to epithelial cells and resident immune cells.
Our data suggest an alternate mechanism for the spleen-independent vagal-mediated anti-inflammatory pathway, namely through activation of EGCs. By ablating EGCs in a segment of bowel, we uncouple the protective effects of VNS on immune cell infiltration and gut barrier integrity. Moreover, EGCs have been shown to be in close proximity to vagal terminals in vivo (48) and the α7nAChR required for vagal signaling has been detected in culture EGCs (8).
Improved understanding of the vagal-mediated pathway helps further our knowledge of this complex signaling cascade but also may be useful in designing treatments to augment the protective effects of VNS following injury. Currently, VNS can be applied through direct current or with pharmacologic vagal agonists (33). By revealing the involvement of EGCs, new therapies may be developed that target EGC activation and provide an alternate treatment strategy aimed at limiting the SIRS response to injury.
The study supported by a grant from the American College of Surgeons C. James Carrico Faculty Research Fellowship for the Study of Trauma and Critical Care (to T. W. Costantini).
No conflicts of interest, financial or otherwise, are declared by the authors.
S.L. performed experiments; S.L., M.K., B.P.E., and T.W.C. analyzed data; S.L., M.K., R.C., B.P.E., and T.W.C. interpreted results of experiments; S.L. and T.W.C. prepared figures; S.L. drafted manuscript; S.L., M.K., R.C., B.P.E., and T.W.C. approved final version of manuscript; R.C., B.P.E., and T.W.C. edited and revised manuscript; T.W.C. conceived and designed research.
- Copyright © 2017 the American Physiological Society