Vagal nerve stimulation protects against burn-induced intestinal injurythrough activation of enteric glia cells

Todd W. Costantini,1 Vishal Bansal,1 Michael Krzyzaniak,1 James G. Putnam,1 Carrie Y. Peterson,1

William H. Loomis,1 Paul Wolf,2 Andrew Baird,1 Brian P. Eliceiri,1 and Raul Coimbra1

1Division of Trauma, Surgical Critical Care, and Burns, Department of Surgery, and 2Department of Pathology Universityof California, San Diego School of Medicine, San Diego, California

Submitted 8 April 2010; accepted in final form 5 August 2010

Costantini TW, Bansal V, Krzyzaniak M, Putnam JG, PetersonCY, Loomis WH, Wolf P, Baird A, Eliceiri BP, Coimbra R. Vagalnerve stimulation protects against burn-induced intestinal injurythrough activation of enteric glia cells. Am J Physiol GastrointestLiver Physiol 299: G1308–G1318, 2010. First published August 12,2010; doi:10.1152/ajpgi.00156.2010.—The enteric nervous systemmay have an important role in modulating gastrointestinal barrierresponse to disease through activation of enteric glia cells. In vitrostudies have shown that enteric glia activation improves intestinalepithelial barrier function by altering the expression of tight junctionproteins. We hypothesized that severe injury would increase expres-sion of glial fibrillary acidic protein (GFAP), a marker of enteric glialactivation. We also sought to define the effects of vagal nervestimulation on enteric glia activation and intestinal barrier functionusing a model of systemic injury and local gut mucosal involvement.Mice with 30% total body surface area steam burn were used as modelof severe injury. Vagal nerve stimulation was performed to assess therole of parasympathetic signaling on enteric glia activation. In vivointestinal permeability was measured to assess barrier function. Intes-tine was collected to investigate changes in histology; GFAP expres-sion was assessed by quantitative PCR, by confocal microscopy, andin GFAP-luciferase transgenic mice. Stimulation of the vagus nerveprevented injury-induced intestinal barrier injury. Intestinal GFAPexpression increased at early time points following burn and returnedto baseline by 24 h after injury. Vagal nerve stimulation prior to injuryincreased GFAP expression to a greater degree than burn alone.Gastrointestinal bioluminescence was imaged in GFAP-luciferasetransgenic animals following either severe burn or vagal stimulationand confirmed the increased expression of intestinal GFAP. Injectionof S-nitrosoglutathione, a signaling molecule released by activatedenteric glia cells, following burn exerts protective effects similar tovagal nerve stimulation. Intestinal expression of GFAP increasesfollowing severe burn injury. Stimulation of the vagus nerve increasesenteric glia activation, which is associated with improved intestinalbarrier function. The vagus nerve may mediate the signaling thatoccurs from the central nervous system to the enteric nervous systemfollowing gastrointestinal injury.

glial fibrillary acidic protein; GFAP-luc; tight junction; splenectomy;S-nitrosoglutathione

THE ENTERIC NERVOUS SYSTEM (ENS) regulates and controlsfunction of the gastrointestinal tract, responding to luminalcontents, mechanical stimulation, and inflammation (13). TheENS can function independently through pathways in themyenteric and submucosal plexus or respond to central nervoussystem (CNS) input through sympathetic and parasympatheticpathways. Enteric glia cells are the predominant cell type in theENS and are similar in structure and function to astrocytes of

the CNS (18). Glia cells express glial fibrillary acidic protein(GFAP) upon activation, increasing the number of GFAP-expressing cells following injury (16).

The enteric glia plays an important role in modulating gutinflammation and maintaining intestinal barrier integrity fol-lowing injury. Genetic ablation of enteric glia cells results insevere inflammation and fatal hemorrhagic necrosis of the gut,highlighting the critical role of the enteric glia in maintainingintestinal integrity (5). Enteric glia has been shown to promoteintestinal barrier integrity in vitro by increasing expression ofthe tight junction proteins occludin and zonula occludensprotein-1 (ZO-1) (20). Enteric glia becomes activated in re-sponse to inflammation, with increased GFAP expression fromenteric glia cells after the in vitro addition of either proinflam-matory cytokines or LPS (24).

Multiple medical conditions including inflammatory boweldisease, necrotizing enterocolitis, trauma, and severe burnresult in significant intestinal barrier injury and serious sys-temic sequelae. We have previously studied the intestinalinjury associated with severe burn, finding changes in tightjunction proteins expression that are associated with increasedintestinal permeability at early time points following injury (8,9). The restoration of small intestinal barrier function requiresa complex set of events that are initiated within minutes ofinjury and are characterized by epithelial cell restitution andreassembly of tight junction proteins to close the paracellularspace (3). The mechanisms for intestinal repair followinginjury are multifactorial, with little understanding of the rolethe enteric nervous system plays in this crucial process.

In this series of experiments, we hypothesized that intestinalinjury would result in activation of enteric glia cells in an invivo model of severe burn injury. We also postulated that vagalnerve stimulation would activate enteric glia and improveintestinal barrier integrity following severe injury. Improvedunderstanding of the response of the enteric glia to intestinalinjury may allow for the development of therapies aimed atlimiting intestinal barrier breakdown and the intestinal inflam-matory response to severe injury.

METHODS

Animal model of severe burn injury. Male BALB/c mice (8–12 wk,Jackson Laboratories, Sacramento, CA) were anesthetized with in-haled isoflurane and had their dorsal fur removed. Animals were thenplaced in a template estimating 30% total body surface area (TBSA)burn and subjected to a steam burn for 7 s as previously described (8).Following burn injury, animals received a subcutaneous injection of1.5 ml normal saline with buprenorphine for fluid resuscitation andpain control. At various time points following burn, animals wereagain anesthetized with inhaled isoflurane for tissue procurement.Sham animals underwent removal of their dorsal fur and received a

Address for reprint requests and other correspondence: R. Coimbra, 200 W.Arbor Dr., #8896, San Diego, CA 92103-8896 (e-mail: rcoimbra@ucsd.edu).

Am J Physiol Gastrointest Liver Physiol 299: G1308–G1318, 2010.First published August 12, 2010; doi:10.1152/ajpgi.00156.2010.

0193-1857/10 Copyright © 2010 the American Physiological Society http://www.ajpgi.orgG1308

subcutaneous injection of normal saline with buprenorphine but werenot subjected to burn injury. All animal experiments were approvedby the University of California, San Diego IACUC and were con-ducted in accordance with accepted guidelines for animal studies.

Vagal nerve stimulation. A right cervical neck incision was per-formed and the right cervical vagus nerve exposed. Vagal nervestimulation was performed by using a VariStim III probe (MedtronicXomed, Jacksonville, FL) set at 2 mA at �1 Hz for 10 min prior toinjury. Sham animals underwent right cervical incision and exposureof the vagus nerve but did not receive stimulation.

In one arm, surgical abdominal vagotomy was performed immedi-ately prior to vagal nerve stimulation and subsequent burn injurythrough an upper-midline laparotomy incision. The gastroesophagealjunction was identified and the dorsal and ventral vagus nerve werevisualized on the distal esophagus with an Olympus SZ61 stereomi-

croscope (Leeds Precision Instruments, Minneapolis, MN). Bothbranches of the vagus nerve were isolated and divided sharply.

Splenectomy. Mice were anesthetized with inhaled isoflurane priorto a midline laparotomy incision. The spleen was removed followingappropriate blood vessel ligation. The abdomen was then closed withinterrupted silk suture.

Injection of S-nitrosoglutathione. Animals were given an intraperi-toneal injection of S-nitrosoglutathione (GSNO, 10 mg/kg) immedi-ately following burn injury (20). Tissues were harvested 4 h followinginjury for histological analysis.

Histopathological evaluation. Sections of distal ileum (n � 3animals per group) were cut 7 �m thick and stained with hematoxylinand eosin (Surgipath, Richmond, IL). Images were viewed with anOlympus IX70 light microscope (Melville, NY) and viewed withQ-imaging software (Surrey, BC, Canada). Histological gut injury

Fig. 1. Vagal nerve stimulation attenuates burn-induced intestinal injury. Distal small intestine harvested from animals 4 h following 30% total body surface area(TBSA) burn (n � 3 animals per group). Sections of small intestine were stained with hematoxylin and eosin. High-power images of the villous tips from eachsection are shown below each image. A: section of normal small intestine from sham animal. B: intestine of burned animals displays evidence of histologicalpathology characterized by blunting of villi. C: gut sections harvested from animals that underwent cervical vagal nerve stimulation prior to severe burn.D: histological gut injury noted in animals that underwent abdominal vagotomy prior to stimulation of the vagus nerve. Size bar � 100 �m. E: intestinal injurywas scored on a scale of 0 (normal) to 4 (severe) by a pathologist blinded to the experimental conditions. *P � 0.05 vs. Sham, †P � 0.05 vs. Burn andBurn/Vagotomy/Vagal Stimulation using the Kruskal-Wallis test followed by pairwise Mann-Whitney test. F: vagal nerve stimulation limits intestinal barrierdysfunction following injury (n � 4 animals per group). Vagal nerve stimulation prevents intestinal barrier dysfunction 4 h following 30% TBSA burn. Dividingthe abdominal (Abd.) vagus nerve abrogates the protective effects of vagal nerve stimulation. Data are expressed as mean systemic FITC-dextran concentration �SE. **P � 0.01 vs. Sham, #P � 0.05 vs. Sham, ‡P � 0.05 vs. Burn and Burn/Vagotomy/Vagal Stimulation by analysis of variance.

G1309VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

was scored by a pathologist blinded to the experimental conditions.Three randomly selected fields from each specimen were gradedbased on a scoring system characterizing gut injury on a scale from 0to 4: 0 � normal, no damage; 1 � mild, focal epithelial edema; 2 �moderate; diffuse swelling necrosis of the villi; 3 � severe; diffusepathology of the villi; with evidence of neutrophil infiltration in thesubmucosa; 4 � major; widespread injury with massive neutrophilinfiltration and hemorrhage as previously described (2).

In vivo intestinal permeability assay. A midline laparotomy wasperformed 4 h following burn and the distal small intestine identifiedas previously described (8, 9). A 5-cm segment of small intestine wasisolated between silk ties prior to intraluminal injection of 25 mg4.4-kDa FITC-dextran (Sigma, St. Louis, MO) diluted in 200 �lphosphate-buffered saline (PBS). The bowel was then returned to theabdominal cavity and the abdominal wall was closed. Animals weremaintained under general anesthesia for 30 min at which time bloodwas obtained via cardiac puncture. Plasma fluorescence was read in afluorescence spectrophotometer (SpectraMax M5, Molecular Diag-nostics, Sunnyvale, CA) and compared with a standard curve ofFITC-dextran diluted in mouse plasma (n � 4 animals per group).

Reverse transcription PCR. Distal ileum was preserved in RNAlater solution and stored at �20°C for no longer than 6 mo. RNA wasextracted from tissue and treated with DNAse via the RNAqueous4PCR kit (Ambion). Reverse transcription was performed (n � 4animals per group) with the High Capacity cDNA Reverse Transcrip-tion kit (Applied Biosciences, Pleasanton, CA). Control for �reversetranscription-PCR was performed separately with Platinum PCR Su-permix (Invitrogen). Reverse transcription quantitative PCR reactionswere run with iQ SYBR Green Supermix (Bio-Rad, Hercules, CA) for40 cycles on a Bio-Rad iQ5 real-time PCR detection system. Cycleswere run at 95°C for 3 min, 95°C for 10 s, 60°C for 30 s, 72°C for 30s. Steps 2–4 were repeated for 40 cycles. A melt curve was obtainedto ensure that only a single species was amplified. The GFAP primerforward 5=-GAGGAGGAGATCCAGTTCTTAAGGA-3=, reverse 5=-GCCTCGTATTGAGTGCGAATC-3= was used. Samples were nor-malized against �-actin or rig/S15 and relative expression levels werecalculated by the ��Ct method (25).

Western blot analysis. Intestinal segments were harvested 4 hfollowing burn for Western blot analysis as previously described (9).Immunoblotting was performed by using standard techniques with anantibody against myosin light chain (MLC) kinase (MLCK, Sigma) or�-actin (Cell Signaling, Danvers, MA). Images were obtained usingthe Xenogen IVIS Lumina imaging system and quantified by use ofUN-SCAN-IT gel digitizing software (Silk Scientific, Orem, UT).

Enzyme-linked immunosorbent assay. Tumor necrosis factor-(TNF-) was measured from plasma and extracts of distal ileumharvested at various time points following injury. A commerciallyavailable enzyme-linked immunosorbent assay kit was obtained fromR&D Systems (Minneapolis, MN). TNF- levels were measured inpicograms per milliliter.

Confocal microscopy. Segments of distal ileum (n � 3 animals pergroup) were embedded in optimal cutting temperature media andstored at �70°C. Sections of intestine were cut 10 �m thick with aReichert-Jung Cryocut 1800 (Reichert Microscopes, Depew, NY).Sections were fixed onto glass slides with 3.7% paraformaldehyde(Electron Microscopy Series, Hatfield, PA) for 10 min then washedwith PBS. Cells were permeabilized with 0.01% Triton X-100(Sigma) for 1 min. Sections were washed in PBS then blocked for 1h in 3% bovine serum albumin (BSA, Sigma). The sections were thenincubated overnight in primary antibody against GFAP-Cy3 conjugate(Sigma), phosphorylated MLC, or 4,6-diamidino-2-phenylindole(DAPI) diluted 1:100. Sections were then treated with the secondaryantibody Alexa Fluor 488 (Invitrogen) in 1% BSA for 1 h. ProlongFade (Invitrogen) was added upon placement of coverslips. Imageswere viewed by using the Olympus Fluoview laser scanning confocalmicroscope (Olympus) with exposure-matched settings (AdvancedSoftware v1.6, Olympus) at 60 magnification.

Detection of in vivo bioluminescence using GFAP-luc mice. Trans-genic GFAP-luc mice with FVB/J background were obtained fromCaliper Life Science (Mountain View, CA). Male GFAP-luc micewere subjected to 30% TBSA steam burn or vagal stimulation aspreviously described (n � 4 animals per group). At various timepoints following injury, animals received an intraperitoneal injectionof 150 �l D-luciferin (15 mg/ml, Caliper Life Sciences) 5 min prior toin vivo imaging with the Xenogen IVIS Lumina while under generalanesthesia (Caliper Life Sciences). Images were obtained with a 1-minexposure time using field of view A. Bioluminescence was analyzedby using Living Image Software (Caliper Life Sciences) with colorbars standardized for animals imaged. Luminescent intensity wasquantified for all animals imaged by using region of interest measure-ments of equivalent areas of the abdominal wall. Following in vivoimaging, animals were euthanized and the brain and intestine wereremoved to confirm ex vivo bioluminescence from these tissues.Transgenic mice also underwent PCR analysis of tail tips to confirmpresence of the GFAP-luc transgene. Nontransgenic FVB/J mice wereimaged to show background fluorescence.

Statistical analysis. Data are expressed as means � SE. Thestatistical significance among groups was determined by t-test oranalysis of variance with Bonferroni correction where appropriate.Statistical significance for gut injury scoring was performed by theKruskal-Wallis test for nonparametric data with post hoc Mann-Whitney test performed in pairwise fashion. Statistical analysis wasperformed by use of SPSS software v11.5 (SPSS, Chicago, IL).Statistical significance was defined as P � 0.05.

RESULTS

Vagal nerve stimulation attenuates intestinal injury. Sec-tions of distal ileum from sham animals and animals 4 hfollowing 30% TBSA burn underwent histological examina-tion to confirm the presence of intestinal injury. The distalsmall intestine from burned animals displayed evidence ofhistological pathology characterized by blunting of villi (Fig. 1B).Stimulation of the vagus nerve prior to burn prevented histo-

Fig. 2. Severe burn increases expression of intestinal glial fibrillary acidicprotein (GFAP). Quantitative PCR (n � 3 animals per group) was performedon intestinal extracts obtained at several time points following burn (open bars)or cervical vagal nerve stimulation (solid bars). Activation of enteric glia cellsoccurs at early time points following burn, with an 8-fold increase in gut GFAPmRNA expression at 2 h following injury. Vagal nerve stimulation also causesactivation of enteric glia cells. Intestinal GFAP mRNA expression increased bynearly 12-fold at 2 and 4 h following vagal nerve stimulation. Data areexpressed as GFAP mRNA expression relative to sham � SE. *P � 0.01 vs.Sham, **P � 0.05 vs. Sham, †P � 0.05 vs. 2-h burn using analysis ofvariance.

G1310 VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

logical gut injury (Fig. 1C). Next, we performed an abdominalvagotomy at the gastroesophageal junction prior to stimulationof the vagus nerve to confirm that efferent vagal nerve signal-ing modulates gut barrier integrity following injury. Abdomi-nal vagotomy (Fig. 1D) prevented the protective effects ofvagal nerve stimulation, with histological gut injury similar toanimals following burn alone. Gut injury scoring was signifi-cantly lower in animals that underwent vagal nerve stimulationprior to burn compared with burn alone (Fig. 1E).

Vagal nerve stimulation prevents burn-induced intestinalpermeability. Intestinal barrier function was assessed with anin vivo intestinal permeability assay (Fig. 1F). Intestinal per-meability to small molecular weight FITC-dextran was in-creased 4 h following burn injury. Stimulation of the vagusnerve immediately prior to injury maintained intestinal barrierintegrity, significantly reducing burn-induced intestinal perme-

ability. Animals undergoing abdominal vagotomy prior tovagal nerve stimulation had intestinal permeability equivalentto animals subjected to burn, eliminating the protective effectsof cervical vagal nerve stimulation. Intestinal permeabilityafter vagal nerve stimulation or vagotomy alone was similar tosham.

Burn injury and vagal nerve stimulation each increaseactivation of enteric glia. Enteric glia activation was assessedby measuring intestinal GFAP mRNA expression via quanti-tative PCR (Fig. 2). The enteric glia became activated at earlytime points following burn, with an eightfold increase in GFAPmRNA from the small intestine at 2 h following injury.Intestinal GFAP mRNA expression is elevated to a lesserdegree by 4 and 6 h following injury. Intestinal GFAP expres-sion returns to sham levels by 24 h following injury (data notshown).

D

Fig. 3. Vagal nerve stimulation alters intestinal GFAP expression. A: quantitative PCR was performed on intestinal extracts obtained 4 h following injury. Resultsare expressed as intestinal GFAP mRNA expression relative to sham � SE. Vagal nerve stimulation (Stim) prior to severe burn results in augmented intestinalGFAP mRNA expression over animals subjected to burn alone. Performing a surgical abdominal vagotomy prior to cervical vagal nerve stimulation andsubsequent severe burn abrogates the ability of vagal nerve stimulation to increase intestinal GFAP mRNA expression. B: arterial blood pressure was measuredduring vagal nerve stimulation to confirm that cervical vagal nerve stimulation did not cause hemodynamically significant changes in blood pressure. There wasno significant change in arterial blood pressure during or immediately following vagal nerve stimulation. Intestinal GFAP� enteric glia localization was assessedby confocal microscopy of intestinal segments stained for GFAP (red) and 4,6-diamidino-2-phenylindole (DAPI; blue). C: there is some background GFAPexpression in sham animals. D: GFAP staining is seen extending toward the intestinal villi in sections from animals 4 h following burn. E: vagal nerve stimulationprior to burn also results in increased staining for GFAP compared with sham. F: abdominal vagotomy prior to vagal nerve stimulation and burn results in apattern of staining similar to sham, suggesting the importance of vagal nerve signaling in enteric glia activation. *P � 0.01 vs. Sham, **P � 0.05 vs. Sham,†P � 0.02 vs. Burn, #P � 0.02 vs. Burn and Burn/Vagotomy/Vagal Stimulation, ‡P � 0.001 vs. all groups except Sham by analysis of variance. Size bar �20 �m.

G1311VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

The enteric nervous system is controlled, in part, throughconnections to the CNS via the vagus nerve. To assess theeffects of signaling via the vagus nerve on enteric glia activa-tion, we performed right cervical vagal nerve stimulation.Intestinal GFAP mRNA expression was measured at severaltime points following vagal nerve stimulation (Fig. 2). There

was a 12-fold increase in GFAP mRNA accumulation at both2 and 4 h following stimulation of the vagus nerve. By 6 hfollowing vagal nerve stimulation intestinal GFAP expressionhad decreased to sham levels.

Stimulation of the vagus nerve modulates activation ofenteric glia following burn. Next, we investigated the changesin intestinal GFAP expression in animals that underwent vagalnerve stimulation prior to severe burn injury (Fig. 3A). Entericglia activation was assessed 4 h following injury as there wasa plateau in GFAP expression following vagal nerve stimula-tion at this time point. Stimulation of the vagus nerve prior toburn was associated with an eightfold increase in intestinalGFAP expression 4 h following injury. This was significantlyhigher than GFAP expression seen from animals followingburn alone. The increased activation of enteric glia cells inanimals receiving vagal nerve stimulation immediately prior tosevere burn correlated to decreased intestinal permeability toFITC-dextran (Fig. 1F).

Next, we performed an abdominal vagotomy prior to vagalstimulation and subsequent severe burn injury to confirm the rela-tionship between efferent vagal nerve signaling and the acti-vation of enteric glia. Abdominal vagotomy prevented theaugmented activation of enteric glia seen in burned animalsundergoing vagal nerve stimulation, with no difference com-pared with animals subjected to burn alone. This also corre-lated with intestinal permeability data showing that abdominalvagotomy abrogated the protective effects of vagal nervestimulation prior to injury (Fig. 1F).

We also confirmed that changes in GFAP expression fol-lowing vagal nerve stimulation were not confounded by sig-nificant changes in arterial blood pressure, which itself couldalter enteric glia activation. Arterial blood pressure was mea-sured in animals during cervical vagal nerve stimulation (Fig.3B). This confirmed that changes seen in enteric glia cellactivation in this model were not due to changes in arterialblood pressure.

Changes in localization of GFAP-expressing enteric gliacells were studied by confocal microscopy of intestinal seg-ments obtained 4 h following injury. There is some baselineexpression of GFAP seen in sham animals as expected (Fig.3C). Additionally, we observed increased staining for GFAP,which was seen projecting toward the intestinal villi in sectionsobtained from burned animal (Fig. 3D). Animals undergoingvagal nerve stimulation prior to burn also have increasedGFAP staining seen projecting into the villi of intestinalsegments (Fig. 3E). The pattern of GFAP expression is lesspronounced in animals subjected to abdominal vagotomy priorto cervical vagal nerve stimulation and burn, suggesting the

Fig. 4. In vivo increases in abdominal luminescence in GFAP-luc mice.Transgenic mice expressing luciferase under control of the GFAP promoterwere imaged in a Xenogen IVIS Lumina 2 h following burn (n � 3 animals pergroup). A: representative sham GFAP-luc mouse demonstrating backgroundGFAP expression with resected intestine from confirming luminescence fromthe intestine. B: in vivo luminescence from the gut of GFAP-luc mice iselevated 4 h following stimulation of the cervical vagus nerve. C: increasedluminescence is also seen from the abdomen of GFAP-luc mice 4 h followingburn. D: quantification of luminescence measured from the abdomen of allGFAP-luc animals imaged, *P � 0.05 vs. Sham, †P � 0.05 vs. Vagal NerveStimulation.

G1312 VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

importance of vagal nerve stimulation in gut GFAP expressionfollowing injury (Fig. 3F).

Vagal stimulation increases bioluminescence from GFAP-luc transgenic mice. Transgenic GFAP-luc mice were used toassess in vivo changes in GFAP expression. These transgenicanimals express luciferase under transcriptional control of theGFAP promoter. After injection of the substrate luciferin, wewere able to image bioluminescence as a marker for GFAPexpression using the Xenogen IVIS Lumina imaging system.GFAP-luc transgenic mice were used to examine in vivochanges in GFAP expression 4 h following vagal nerve stim-ulation. There was increased luminescence from the abdomenof animals imaged 4 h following stimulation of the cervicalvagus nerve (Fig. 4B) compared with sham (Fig. 4A). Excisionof the intestine following in vivo imaging confirmed lumines-cence from the small intestine. Quantification of all GFAP-lucanimals imaged confirms increased luminescence from animals4 h following stimulation of the vagus nerve (Fig. 4D).

Abdominal bioluminescence is increased in GFAP-luc micefollowing burn. There is increased luminescence noted fromthe abdomen of GFAP-luc animals 4 h following burn (Fig.4C) compared with GFAP-luc mice who were not subjected toburn (Fig. 4A). To confirm that the luminescence detected fromthe abdomen of these animals was from the intestine, the gut wasresected en bloc for imaging. Luminescent imaging from theexcised intestine confirms that the abdominal luminescence seenon in vivo imaging originated from the intestine of the GFAP-lucmice. Quantification of abdominal bioluminescence from all miceimaged demonstrates a greater luminescence from the burnedGFAP-luc mice compared with sham (Fig. 4D). Nontransgenicsham and burn mice were imaged to confirm that there isminimal background luminescence at the settings used in thisseries of experiments (data not shown).

Vagal stimulation increases bioluminescence from GFAP-luc transgenic mice. GFAP-luc transgenic mice were used toexamine in vivo changes in GFAP expression 4 h following

Fig. 5. Vagal nerve stimulation prevents burn-induced intestinal barrier injury. A: intestinal TNF- levels measured from distal ileum harvested 4 h followinginjury using ELISA. Vagal nerve stimulation prevents the increase in TNF- seen following burn injury, *P � 0.05 vs. Burn. B: changes in intestinal myosinlight chain kinase (MLCK) expression 4 h following burn by Western blot. There is nearly a 4-fold increase in intestinal MLCK expression following burn.Animals undergoing vagal nerve stimulation exhibited intestinal MLCK expression similar to sham, **P � 0.05 vs. Burn by analysis of variance. Intestinalsegments were harvested 4 h following burn and stained for phosphorylated myosin light chain (MLC, green) and DAPI (blue) by confocal microscopy.C: minimal staining seen in sham animals. D: staining for phosphorylated MLC is increased following burn. E: vagal nerve stimulation prior to burn decreasesstaining for phosphorylated MLC. Bar � 20 �m.

G1313VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

vagal nerve stimulation. There was increased luminescencefrom the abdomen of animals imaged 4 h following stimulationof the cervical vagus nerve (Fig. 4E) compared with sham (Fig.4D). Excision of the intestine following in vivo imagingconfirmed luminescence from the small intestine. Quantifica-tion of all GFAP-luc animals imaged confirms increased lumi-nescence from animals 4 h following stimulation of the vagusnerve (Fig. 4F).

Vagal nerve stimulation attenuates burn-induced intestinalbarrier injury. Next, we evaluated whether vagal nerve stim-ulation would modulate intestinal inflammation and intestinaltight junction protein expression following injury. Vagal nervestimulation prevented the burn-induced increase in intestinalTNF- levels (Fig. 5A), which correlates with results of the invivo intestinal permeability assay (Fig. 1F) MLCK is a tightjunction protein that plays a key role in modulating intestinalbarrier function. Increased MLCK expression increases para-cellular permeability across the intestinal tight junction and isregulated, in part, by increases in TNF- (15). Changes inintestinal MLCK expression were analyzed by Western blot,showing that vagal nerve stimulation prevents the increase inMLCK activation that is seen after burn injury (Fig. 5B). Theability of vagal nerve stimulation to modulate intestinal tightjunction protein expression was further confirmed by imagingchanges in phosphorylated MLC by use of confocal micros-

Fig. 6. Postinjury vagal nerve stimulation protects against gut barrier injury. Invivo intestinal permeability to 4-kDa FITC-dextran was measured 4 h follow-ing in burn injury. Animals underwent either postinjury vagal nerve stimula-tion or S-nitrosoglutathione (GSNO) injection. Stimulation of the vagus nerve4 h following burn injury decreases intestinal permeability compared withsham. GSNO injection also modulates gut barrier function following severeburn, attenuating burn-induced intestinal permeability. *P � 0.03 vs. Burn.

Fig. 7. Injection of GSNO decreases intestinal injury following severe burn. Intestinal injury was assessed by histology. B: sham. B: there is clear evidence ofgut injury in sections obtained from animals subjected to severe burn. C: sections of intestine from burned animals injected with GSNO show minimal evidenceof injury, similar to sham. Black size bar � 100 �m. Changes in phosphorylated MLC (green) were assessed by confocal microscopy. Cell nuclei were stainedwith DAPI (blue). D: there is minimal staining for phosphorylated MLC seen in sham animals. E: staining for phosphorylated MLC is increased following burn.F: injection of GSNO prior to burn decreases staining for phosphorylated MLC. White size bar � 30 �m.

G1314 VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

copy. MLCK activation results in phosphorylation of MLC andsubsequent contraction of the actin-myosin ring in the intesti-nal epithelium, resulting in opening of the tight junction.Severe burn injury resulted in increased staining for phosphor-ylated MLC (Fig. 5D). Animals undergoing vagal nerve stim-ulation prior to burn had a pattern of phosphorylated MLCexpression similar to sham (Fig. 5, C and E), with clearly lessstaining than seen from the gut of burn injured animals.

Postinjury vagal nerve stimulation protects against burn-induced intestinal permeability. The clinical utility of vagalnerve stimulation will be determined by the immunomodula-tory effects of stimulating the vagus nerve after injury. Weperformed a set of experiments with vagal nerve stimulationoccurring at 30 min after burn injury. Postinjury vagal nervestimulation resulted in decreased intestinal permeability fol-lowing severe burn injury (Fig. 6), with no loss in the protec-tive effects measured in animals that underwent vagal nervestimulation prior to injury.

Postinjury treatment with GSNO protects against burn-induced intestinal barrier injury. Activated enteric glia cellshave been shown to release GSNO, a known modulator ofintestinal barrier function. GSNO was injected following se-vere burn injury to explore a potential mechanism by whichintestinal barrier function is maintained through activation ofenteric glia cells. Injection of GSNO following injury alsoprevented the burn-induced increase in intestinal permeabilityto 4-kDa FITC-dextran (Fig. 6) and also improved gut histol-ogy compared with burned animals (Fig. 7, A–C). Injection ofGSNO also modulated intestinal tight junction protein ex-pression. Animals injected with GSNO had decreased stain-ing for phosphorylated MLC (Fig. 7F) compared with burnalone (Fig. 7E).

The protective effects of vagal nerve stimulation are not dueto modulation of systemic TNF-�. Previous studies have shownthat vagal nerve stimulation has a major effect on circulatingcytokines, specifically causing a dramatic decrease in TNF-production from the spleen (12). We questioned whether theprotective effects of vagal nerve stimulation in this model weresimply due to a decrease in circulating TNF-. To answer thisquestion we performed vagal nerve stimulation on burn injuredanimals immediately following splenectomy, thus eliminatingthe effects of TNF- produced by the spleen following injury.Burn injury caused increased intestinal permeability and his-tological gut injury in splenectomized animals. Vagal nervestimulation maintained its gut-protective effect after burn in-jury in animals that were splenectomized (Figs. 8 and 9, A–E).This suggests that the protective effects of vagal nerve stimu-lation are not due to modulation of splenic production ofcirculating TNF-. This was confirmed by measuring de-creased circulating TNF- levels in both animals that hadundergone splenectomy and splenectomized animals that un-derwent vagal nerve stimulation following injury (Fig. 9F).Elevated intestinal TNF- correlated with increased intestinalpermeability and histological gut injury, suggesting a localrather than systemic inflammatory response (Fig. 9G).

DISCUSSION

The ability of vagal nerve stimulation to modulate theintestinal response to inflammation is just beginning to beunderstood. In this series of experiments, we investigated

changes in small intestine enteric glia activation in a model ofintestinal injury caused by severe burn. The enteric glia may, atleast in part, be responsible for restoring intestinal barrierintegrity following injury. Here we propose that the responseof the enteric glia to intestinal injury may be modulated by theCNS via the vagus nerve. Our findings suggest that stimulatingthe vagus nerve at the time of injury may augment enteric gliacell activation and either prevent intestinal barrier injury orspeed its recovery.

The results of this study demonstrated that severe burninjury increases activation of enteric glia cells as measured byincreased intestinal GFAP mRNA expression, by confocalmicroscopy of intestinal segments stained for GFAP, and byutilizing in vivo imaging of luminescence from the intestine ofGFAP-luc transgenic animals. We also investigated the effectsof vagal nerve stimulation on enteric glia activation. Theseexperiments showed that electrical stimulation of the cervicalvagus nerve alone causes increased expression of intestinalGFAP.

This is the first study, to our knowledge, to utilize GFAP-luctransgenic mice to study changes in the enteric glia. Thesetransgenic mice have been used to image changes to CNSastrocytes; however, abdominal imaging of changes in GFAPexpression following injury has not previously been reported(6, 14). This technique allows for in vivo imaging of changesin GFAP expression and allows for a powerful visual confir-mation of results of classic methods such as quantitative PCR.In the future, this nonlethal, in vivo imaging technique may beideally suited for following changes in GFAP expression in thesame animal over time.

We found that stimulation of the vagus nerve preventedburn-induced intestinal permeability and attenuated histo-

Fig. 8. Vagal nerve stimulation prevents burn-induced intestinal permeabilityin splenectomized animal. In vivo intestinal permeability to 4-kDa FITC-dextran was measured 4 h following injury. Increased intestinal permeabilitywas seen in animals that underwent splenectomy immediately prior to burninjury. The protective effects of vagal nerve stimulation are seen in burnedanimals that underwent splenectomy. Taken together, this suggests the gut-protective effects of vagal nerve stimulation are not due to modulation ofsplenic cytokine production. *P � 0.02 vs. Sham, Splenectomy, and Burn/Splenectomy/Vagal Stim groups.

G1315VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

logical gut injury. To confirm that the effects of vagal nervestimulation were due to an efferent signaling pathway, weperformed surgical abdominal vagotomy prior to vagalnerve stimulation. Animals undergoing abdominal vagot-omy prior to vagal stimulation had intestinal permeabilitysimilar to animals subjected to burn alone, confirming theprotective effects of efferent vagal nerve signaling. Theimproved intestinal barrier function seen in injured animalsthat underwent vagal nerve stimulation was associated witha significant increase in intestinal GFAP expression viaPCR. Limiting intestinal barrier injury may play an impor-tant role in decreasing the systemic inflammatory responseassociated with severe injury. Vagal nerve stimulation haspreviously been shown to decrease histological gut injuryand intestinal inflammation in a model of colitis (10). Vagalnerve stimulation has also been shown to decrease postop-erative ileus in a murine model, further supporting theability of vagal nerve stimulation to modulate the entericnervous system (22). Importantly, we also demonstrated thatvagal nerve stimulation given 30 min after burn injurycontinues to prevent intestinal barrier injury. This is cer-tainly important, since the clinical utility of vagal nerve

stimulation lies in its ability to have protective effects whenadministered after an insult.

In this series of experiments we found that vagal nervestimulation prevented the burn-induced increase in intestinalTNF-. This was associated with modulation of the intestinaltight junction proteins MLCK and phosphorylated MLC. In-creased expression of both MLCK and phosphorylated MLCplay an important role in the loss of tight junction integrity.Therefore, the ability of vagal nerve stimulation to attenuatethe increase in MLCK and phosphorylated MLC after burngives insight into the mechanism by which vagal nerve stim-ulation prevents intestinal barrier loss and subsequent intestinalinflammation.

Similar to the role of astrocytes in maintaining the blood-brain barrier, enteric glia play an important role in regulat-ing intestinal permeability (21). The importance of entericglia was noted in studies using targeted ablation of entericglia cells in transgenic mice. Loss of enteric glia cells wasfatal in that animal model owing to hemorrhagic necrosis ofthe small intestine (5). Clinically, loss of enteric glia cellshas been implicated in the pathogenesis of inflammatorybowel disease (7).

Fig. 9. Gut-protective effects of vagal nerve stimulation are independent of changes in circulating TNF-. Distal small intestine harvested from animals 4 hfollowing injury (n � 3 animals per group). Sections of small intestine stained with hematoxylin and eosin. Representative sections are shown of normal smallintestine from sham (A) and following splenectomy (B). Intestines of a representative animal following burn (C) and an animal that underwent splenectomy priorto burn (D) display evidence of histological gut injury. E: a normal gut section harvested from a representative animal that underwent splenectomy prior to vagalnerve stimulation and severe burn. F: serum TNF- was measured 4 h following injury using ELISA. There is a decrease in circulating TNF- in animals thatunderwent splenectomy. *P � 0.001 vs. Burn, †P � 0.01 vs. Burn. G: TNF- measured from intestinal extracts obtained 4 h following burn injury. There isevidence of local intestinal TNF- production that correlates with the intestinal injury seen by intestinal permeability and histology. **P � 0.05 vs. Burn andBurn/Splenectomy.

G1316 VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

Neurons of the ENS communicate with enteric glia cells,allowing for cross talk between cell types (18). Gulbransen andSharkey (11) have recently shown that electric stimulation ofenteric neurons causes activation of enteric glia cells throughrelease of the neurotransmitter adenosine triphosphate. Herewe demonstrated that cervical vagal nerve stimulation causesactivation of enteric glia cells as measured by increased GFAPexpression. This demonstrates a link between the vagus nerveand enteric glia and gives insight into possible mechanisms bywhich the CNS communicates with ENS following gut inflam-mation. We found that stimulation of the vagus nerve prior toburn augmented the activation of enteric glia cells in responseto intestinal injury.

Studies by Savidge (20) et al. have shown that enteric gliaimprove intestinal barrier function by augmenting the expres-sion of tight junction proteins, resulting in improved intestinalbarrier integrity. They demonstrated that glial cells secreteGSNO, which modulates intestinal tight junction integrity.This was confirmed in an animal model of gut inflammation, inwhich injection of exogenous GSNO attenuated gut inflamma-tion, improved intestinal barrier integrity, and increased ex-pression of the tight junction protein ZO-1. Enteric glia cellsare also known to secrete transforming growth factor-�, an-other molecule that improves intestinal barrier integrity (1, 17).Here we have shown that injection of GSNO attenuates burn-induced intestinal barrier injury with results similar to animalsundergoing vagal nerve stimulation. This further suggests thatvagal nerve stimulation may maintain barrier integrity andintestinal tight junction protein expression through the abilityof activated enteric glia cells to produce GSNO.

It has been hypothesized that a subpopulation of enteric gliacells can rapidly upregulate GFAP expression in response toinflammation, whereas other glia cells respond more slowlyand require proliferation and GFAP synthesis (4, 24). The CNSmay signal the ENS via the vagus nerve to increase enteric glialcell GFAP synthesis in response to injury. Since enteric gliacells improve intestinal barrier integrity, this may representanother vagal-mediated anti-inflammatory pathway that is ac-tivated in response to gut injury. It is important to acknowledgethat there may be other mechanisms by which stimulating thevagus nerve may protect against burn-induced gut injury. Inaddition to secreting GSNO, activated enteric glia cells causenumerous changes in gene transcription within the intestinalepithelial cell that may alter barrier function (23). Vagal nervestimulation may also alter intestinal function after injury byother pathways including neuropeptide release.

Vagal nerve stimulation has previously been shown to exertits protective effects through its ability to decrease productionof splenic TNF- following injury via the 7 nicotinic acetyl-choline receptor (19). We sought to determine whether thegut-protective effects of vagal nerve stimulation in this studywere due to modulation of splenic TNF-. We first splenecto-mized animals immediately prior to burn injury to eliminate themain source of immediate TNF production. We found thatthere continued to be increased permeability and histologicalgut injury in burned animals after splenectomy despite de-creased levels of circulating TNF-. This suggests that the gutinjury seen in this model is not due to circulating TNF-. Next,we performed vagal nerve stimulation in burned animals thathad been splenectomized immediately prior to injury. Weagain found that vagal nerve stimulation had gut-protective

effects as measured by permeability and histology. We alsomeasured TNF- from gut tissue and found elevated levels inburned animals that had been splenectomized, indicating alocal but not systemic inflammatory response. These datasuggest that the protective effects of vagal nerve stimulation onthe gut are not due to modulation of splenic production ofcirculating TNF-.

In summary, we demonstrated that stimulation of the vagusnerve is associated with increased activation of enteric gliacells and resulted in attenuation of burn-induced intestinalbarrier injury. The ability of vagus nerve stimulation to activateenteric glia cells may give insight into the signaling that occursfrom the CNS to the ENS following gut injury. These resultsadvance previous studies documenting the anti-inflammatoryeffects of vagal nerve stimulation, now showing the ability tomodulate enteric glia cells within the ENS. Exploiting thebarrier inducing effects of enteric glia activation may preventor limit intestinal barrier breakdown and may be a therapeutictarget in diseases causing intestinal inflammation.

ACKNOWLEDGMENTS

The authors thank Alexandra Borboa, Billie Chun, and Luiz GuilhermeReys, MD for technical assistance with this project.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

REFERENCES

1. Al-Sadi R, Boivin M, Ma T. Mechanism of cytokine modulation ofepithelial tight junction barrier. Front Biosci 14: 2765–2778, 2009.

2. Bansal V, Costantini T, Kroll L, Peterson C, Loomis W, Eliceiri B,Baird A, Wolf P, Coimbra R. Traumatic brain injury and intestinaldysfunction: uncovering the neuro-enteric axis. J Neurotrauma 26: 1353–1359, 2009.

3. Blikslager AT, Moeser AJ, Gookin JL, Jones SL, Odle J. Restorationof barrier function in injured intestinal mucosa. Physiol Rev 87: 545–564,2007.

4. Bradley JS Jr, Parr EJ, Sharkey KA. Effects of inflammation on cellproliferation in the myenteric plexus of the guinea-pig ileum. Cell TissueRes 289: 455–461, 1997.

5. Bush TG, Savidge TC, Freeman TC, Cox HJ, Campbell EA, Mucke L,Johnson MH, Sofroniew MV. Fulminant jejuno-ileitis following ablationof enteric glia in adult transgenic mice. Cell 93: 189–201, 1998.

6. Cordeau P, Jr, Lalancette-Hebert M, Weng YC, Kriz J. Live imagingof neuroinflammation reveals sex and estrogen effects on astrocyte re-sponse to ischemic injury. Stroke 39: 935–942, 2008.

7. Cornet A, Savidge TC, Cabarrocas J, Deng WL, Colombel JF, Lass-mann H, Desreumaux P, Liblau RS. Enterocolitis induced by autoim-mune targeting of enteric glial cells: a possible mechanism in Crohn’sdisease? Proc Natl Acad Sci USA 98: 13306–13311, 2001.

8. Costantini TW, Loomis WH, Putnam JG, Drusinsky D, Deree J, ChoiS, Wolf P, Baird A, Eliceiri B, Bansal V, Coimbra R. Burn-induced gutbarrier injury is attenuated by phosphodiesterase inhibition: effects ontight junction structural proteins. Shock 31: 416–422, 2009.

9. Costantini TW, Loomis WH, Putnam JG, Kroll L, Eliceiri BP, BairdA, Bansal V, Coimbra R. Pentoxifylline modulates intestinal tight junc-tion signaling after burn injury: effects on myosin light chain kinase. JTrauma 66: 17–24; discussion 24–15, 2009.

10. Ghia JE, Blennerhassett P, El-Sharkawy RT, Collins SM. The protec-tive effect of the vagus nerve in a murine model of chronic relapsingcolitis. Am J Physiol Gastrointest Liver Physiol 293: G711–G718, 2007.

11. Gulbransen BD, Sharkey KA. Purinergic neuron-to-glia signaling in theenteric nervous system. Gastroenterology 136: 1349–1358, 2009.

12. Huston JM, Ochani M, Rosas-Ballina M, Liao H, Ochani K, PavlovVA, Gallowitsch-Puerta M, Ashok M, Czura CJ, Foxwell B, TraceyKJ, Ulloa L. Splenectomy inactivates the cholinergic antiinflammatorypathway during lethal endotoxemia and polymicrobial sepsis. J Exp Med203: 1623–1628, 2006.

G1317VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org

13. Lomax AE, Linden DR, Mawe GM, Sharkey KA. Effects of gastroin-testinal inflammation on enteroendocrine cells and enteric neural reflexcircuits. Auton Neurosci 126–127: 250–257, 2006.

14. Luo J, Ho P, Steinman L, Wyss-Coray T. Bioluminescence in vivoimaging of autoimmune encephalomyelitis predicts disease. J Neuroin-flammation 5: 6–12, 2008.

15. Ma TY, Boivin MA, Ye D, Pedram A, Said HM. Mechanism of TNF-modulation of Caco-2 intestinal epithelial tight junction barrier: role ofmyosin light-chain kinase protein expression. Am J Physiol GastrointestLiver Physiol 288: G422–G430, 2005.

16. McGraw J, Hiebert GW, Steeves JD. Modulating astrogliosis afterneurotrauma. J Neurosci Res 63: 109–115, 2001.

17. Neunlist M, Aubert P, Bonnaud S, Van Landeghem L, Coron E,Wedel T, Naveilhan P, Ruhl A, Lardeux B, Savidge T, Paris F,Galmiche JP. Enteric glia inhibit intestinal epithelial cell proliferationpartly through a TGF-�1-dependent pathway. Am J Physiol GastrointestLiver Physiol 292: G231–G241, 2007.

18. Neunlist M, Van Landeghem L, Bourreille A, Savidge T. Neuro-glial crosstalkin inflammatory bowel disease. J Intern Med 263: 577–583, 2008.

19. Rosas-Ballina M, Ochani M, Parrish WR, Ochani K, Harris YT,Huston JM, Chavan S, Tracey KJ. Splenic nerve is required for

cholinergic antiinflammatory pathway control of TNF in endotoxemia.Proc Natl Acad Sci USA 105: 11008–11013, 2008.

20. Savidge TC, Newman P, Pothoulakis C, Ruhl A, Neunlist M, Bour-reille A, Hurst R, Sofroniew MV. Enteric glia regulate intestinal barrierfunction and inflammation via release of S-nitrosoglutathione. Gastroen-terology 132: 1344–1358, 2007.

21. Savidge TC, Sofroniew MV, Neunlist M. Starring roles for astroglia inbarrier pathologies of gut and brain. Lab Invest 87: 731–736, 2007.

22. The FO, Boeckxstaens GE, Snoek SA, Cash JL, Bennink R, LarosaGJ, van den Wijngaard RM, Greaves DR, de Jonge WJ. Activation ofthe cholinergic anti-inflammatory pathway ameliorates postoperative ileusin mice. Gastroenterology 133: 1219–1228, 2007.

23. Van Landeghem L, Mahe MM, Teusan R, Leger J, Guisle I, HoulgatteR, Neunlist M. Regulation of intestinal epithelial cells transcriptome byenteric glial cells: impact on intestinal epithelial barrier functions. BMCGenomics 10: 507–527, 2009.

24. Von Boyen GB, Steinkamp M, Reinshagen M, Schafer KH, Adler G,Kirsch J. Proinflammatory cytokines increase glial fibrillary acidic pro-tein expression in enteric glia. Gut 53: 222–228, 2004.

25. Willems E, Leyns L, Vandesompele J. Standardization of real-time PCRgene expression data from independent biological replicates. Anal Bio-chem 379: 127–129, 2008.

G1318 VAGAL STIMULATION PROTECTS THE GUT BARRIER AFTER INJURY

AJP-Gastrointest Liver Physiol • VOL 299 • DECEMBER 2010 • www.ajpgi.org