Reactive enteric glial cells participate in paralytic ileus by damaging nitrergic neurons during endotoxemia

Paralytic ileus is common in patients with septic shock, which causes high morbidity and mortality. Enteric neurons and enteric glial cells (EGCs) regulate intestinal motility, but little is known about their interaction in endotoxemia. We aim to investigate whether reactive EGCs have harmful effects on enteric neurons and participate in intestinal motility disorder in mice during endotoxemia. Endotoxemia was induced by lipopolysaccharide (LPS) intraperitoneal injection in mice. And uorocitrate (FC) was administered before LPS injections to inhibit the reactive EGCs. The effects of reactive EGCs on intestinal motility were analyzed by motility assays in vivo and colonic migrating motor complexes (CMMCs) ex vivo. The number of enteric neurons was evaluated by immunouorescent staining HuCD, nNOS and ChAT in vivo. In addition, we stimulated EGCs with IL-1β and TNF-α in vitro and cultured the primary enteric neurons in the conditioned medium, detecting the apoptosis and morphology of neurons through staining TUNEL, cleaved Caspase-3 protein and Anti-β-III tubulin. and while the reactive vitro, the conditioned medium reactive


Introduction
Sepsis is a life-threatening organ dysfunction caused by the host's dysregulated response to infection [1].
The mortality rate of sepsis and septic shock has commonly been quoted as ranging from 25-60% [2].
Intestinal motility disorder, often manifests as paralytic ileus, is a common complication following sepsis and septic shock. It can cause bacteria accumulation in the intestine, and promote bacterial translocation. In turn, the occurrence of bacterial translocation will induce multiple organ dysfunction syndromes (MODS) and lead to high mortality [3]. Thus improving intestinal motility is critical during sepsis treatment [4]. Intestinal motility is regulated by the enteric nervous system (ENS), a complex network of neurons and glial cells, that mainly resides in the submucosal and myenteric plexus [5]. There is evidence that dysfunction, degeneration, or loss of myenteric neurons will cause several intestinal motility disorders [6,7]. Enteric glial cells (EGCs), as the most abundant cells in the ENS, provide nutritional support for enteric neurons, promote the formation and function of synapses, and maintain neuronal survival [8].
EGCs undergo a dramatic transformation called the "reactive glial phenotype" in response to intestinal infection with bacteria [9], viruses and in ammatory mediators [10,11]. However, there is a controversy regarding the function of reactive EGCs. Previous studies have shown EGCs may have dual protective and harmful effects. For instance, the intestinal ischemia-reperfusion injury (IRI) leads to the signi cant upregulation of glial brillary acidic protein (GFAP), the EGC reactivity marker. That presents a bene cial effect on neurons through the increase of glial cell line-derived neurotrophic factor (GDNF) [12,13]. On the contrary, a study from Gulbransen's group provided evidence that glial reactivity as a driver of enteric neurodegeneration caused the death of enteric neurons through purinergic pathways in colitis induced by dinitrobenzene sulfonic acid (DNBS) [14]. As for endotoxemia, it has been con rmed that lipopolysaccharide (LPS) increasing leads to proin ammatory cytokines release and myenteric neurons apoptosis in vitro [15]. In addition, systemic LPS administration also induces an increase in intestinal GFAP and S100b expression [16]. While the association between reactive EGCs and the loss of enteric neurons is still an issue under discussion. That is to say, it is unclear whether reactive EGCs are bene cial or harmful to the enteric neurons in endotoxemia mice.
In this study, we investigated the effect of reactive EGCs on intestinal motility and enteric neuron function in an endotoxemia model in vivo, and observed the effect of reactive EGCs on the growth and survival rate of enteric neurons under an in ammatory condition in vitro. We observed that LPS-induced reactive EGCs are harmful to the enteric neurons during endotoxemia and are associated with colonic transit delay, which might provide a novel strategy for prevention or treatment of LPS-induced intestinal paralysis and other relevant intestinal motility disorders.
Materials And Methods 1. Animals C57BL/6 mice (male, 6~8 weeks old, weight 20~23 g) were obtained from the Experimental Animal Center of Xi'an Jiaotong University (Xi'an, China) and housed in a temperature-controlled and humiditycontrolled room with a 12:12 hour light: dark cycle, and provided with food and water. All procedures were approved by the Institutional Animal Care and Use Committees at Xi'an Jiaotong University (Xi'an, China).
FC treatment and endotoxemia model Animals were randomly divided into 3 groups: Con: vehicle control, normal saline, LPS: lipopolysaccharide, LPS + uorocitrate (FC): LPS injection after intraperitoneal injections of FC. A schematic diagram of the experiments undertaken in this study was illustrated in Supplementary Figure   S1. First, vehicle or FC (Sigma, St. Louis, MO) was injected intraperitoneally twice per day (20 μmol/kg body weight, 9 AM and 6 PM) for 7 d as described by Nasser et al [17]. The details of FC preparation were presented in Supplementary information 1. Then, mice were injected intraperitoneally with vehicle (controls) or with 20 mg/kg LPS (Sigma-Aldrich, E. coli O55:B55, USA) in a 0.2 ml volume on the morning of the eighth day. Animals were observed and evaluated at least twice a day for post-treatment care until sacri ce, including breath, body temperature and activity. Gastrointestinal or colonic motility was measured until 48 hr later. Finally, Animals were euthanized by cervical dislocation.

Stool frequency and uid content
According to the previous method[18], each mouse was placed in a clear cage for 1hr. Fecal pellets were collected, counted, and weighed (wet weight). These were dried overnight at 60℃ and weighed again (dry weight). Fluid content (%) = 100% × (wet weight-dry weight)/wet weight.

Colon bead assay
Distal colon transit was assessed using glass beads (2 mm in diameter) as previously described [18], and executed 48 hr after the LPS injection. After being overnight fasted, mice were anesthetized with inhalation of iso urane and a single bead was inserted through the anus and pushed 2 cm towards the oral by a customized needle with a silicon cannula (1.9 mm in diameter) (Cadence Inc., Cat. No. 9921, Staunton, USA). Then, the needle was withdrawn lightly and the bead expulsion latency was obtained. The time required to eject the bead was measured as an estimate of colonic motility.

Whole-gut transit
Whole intestinal transit time was measured following the previous report with a little modi cation [19] and executed 48 hr after the LPS injection. Male mice (8 weeks) have fasted only with access to water for 18 hr before the experiment. 0.2 mL of a solution containing 6% (w/v) Carmine (Sigma-Aldrich, C1022, USA) and 0.5% (w/v) methylcellulose (Sigma-Aldrich, M0262, USA) dissolved in ultrapure water was orally administered to each mouse and left undisturbed in individual cages with food and water ad libitum. During 2 hr after gavage, pellets coloration was checked regularly every 20 min. Time elapsed from gavage until the appearance of the rst red pellet was obtained.

Colonic migrating motor complexes measuredin vitro
According to the previous method [20], the entire colon (5~6 cm, n = 6) was dissected from the mouse and ushed to remove fecal content with Krebs solution. Then the empty colon was mounted in a horizontal organ bath with oxygenated Krebs solution at 35~36°C and maintained intraluminal pressure at 1~2 cmH 2 O. Preparations were equilibrated for 30 min and four 15-minute videos of contractile activity were captured via a video camera (Logitech, Newark, USA) positioned 7~8 cm above the gut. Finally, these videos were converted to spatiotemporal maps via MATLAB (MathWorks, USA). The frequency of colonic migrating motor complexes (CMMCs) as well as the velocity and length of their propagation were analyzed by a researcher blinded to the test groups.

Primary enteric glial cells
The isolation, identi cation, and culture of primary EGCs were performed as previously described [21]. In brief, the colonic tissue was collected from newborn mice (1~2 days old, C57BL/6 mice), and cells were cultured in DMEM-F-12 supplemented with 10% FBS, 1mM glutamine and 100IU/ml penicillin/streptomycin. More details are in Supplementary information 2. They were passaged to new plates after 12~14 d for purity assay and the following experiments. Culture consisted of approximately 90% of enteric glia, as judged by immunolabelling with a chicken antibody speci c for GFAP (1:500, GeneTex, GTX85454, USA) and rabbit antibody for S100b protein (1:100, GeneTex, GTX57757, USA) (Supplementary Figure S2).

Primary enteric neurons
The isolation and culture of enteric neurons are similar to that of primary EGCs. The difference is that the colonic tissue was collected from embryonic day 14~15 (E14~15) rats, and cells were cultured in a neurobasal medium containing 2 mM glutamine, 1 mM (100 IU/ml) penicillin/streptomycin, and 1 mM B-27 supplement. The purity of neurons is also more than 90% after 12~14 d, staining with mouse anti-HuC/D primary antibody (1:100; Invitrogen) (Supplementary Figure S3).

Conditioned medium experiments
IL-1β (80 ng/ml, Peprotech, AF-211-11B, USA) and TNFα (60 pg/ml, Peprotech, AF-315-01A, USA) were used to induce EGCs into reactive for 24 hr according to the previous study [21]. Brie y, the enteric glial cell line was culture about 48 hr in DMEM supplemented with 10% FBS, 1mM glutamine, and penicillin/streptomycin until the cell attachment rate was up to 70~80%. Then, we changed the media into the serum-free one supplemented with 80 ng/ml IL-1β and 60 pg/ml TNFα for 24 hr. Next, the medium containing cytokines was removed and replaced with a fresh neurobasal medium for another 24 hr. At this time, the reactive EGCs conditioned medium (ECM) was centrifuged at 1000 rpm for 5 min, ltered through a 0.22 μm syringe lter (0.22 uM lter, Thermo, USA), and conserved at -80°C until use.
The conditioned medium of control (resting) EGCs was collected as above process except supplementary with IL-1β and TNFα.

Detection of apoptosis
To detect apoptosis, TUNEL staining was carried out using the uorescein tagged In Situ Cell Death kit (Promega, G3250, USA) according to the procedures of speci cation. Cells suspension was obtained after isolating enteric neurons from colons of fetal mice, this was counted using a classic hemocytometer.
Primary enteric neurons were plated about 4×10 4 cells/cm 2 in 24-well plates and grown for 12~14 d after maturing in the neurobasal medium. They were treated with the conditioned medium of control (resting) and reactive EGCs respectively. 48 hr after the treatment, neurons were xed in 4% PFA for 20 min at 25 o C and then cleared with PBS. The next steps were according to the procedures of the speci cation.

Neurite formation assay
Primary enteric neurons were plated about 1,000 cells/cm 2 in 24-well plates when cell suspension was obtained. The neurobasal medium was replaced with the conditioned medium of control and reactive EGCs until enteric neurons were grown for 8~9 d. Then, enteric neurons were xed and stained with antibodies against the neuron-speci c beta III Tubulin (TuJ-1, Abcam, ab78078, USA) 72 hr after incubation. For analysis of neurite formation, Sholl analysis was used to assess the complexity of neurites. Pictures of TuJ-1-labeled cells were taken under the confocal microscope (Olympus, FV1000, Japan) with 40× objective and a template of concentric circles distant from 10 to 500 µm (10 µm interval) from the ganglion center was overlaid on the ganglion using Image J software. The number of primary neurites and branching points was also counted. For each group (3 biological replications), 5 ganglia were analyzed.

Immuno uorescent staining
Whole-mount longitudinal muscle/myenteric plexus (LMMP) preparations of colonic myenteric plexus were prepared according to a published method in Tricia H. Smith et al [22]. An entire colon was divided into two parts, and half part of each colon was used as LMMP preparations, and the other part was for Western blot. After washing in phosphate-buffered saline, colon LMMP preparations that separated from 1-1.5-cm colonic fragments were xed in 4% paraformaldehyde (PFA) 1 hr and then permeabilized with 0.3% Triton-X for 40 min before immunostaining. Preparations were blocked with 10% goat serum for 1 hr and exposed to primary antibodies: 6. Confocal Image Acquisition and Analysis Images of the myenteric ganglia were taken under the FluoView 1000 confocal microscope (Olympus, Tokyo, Japan). A computer-controlled motorized stage was used to scan images (20× or 40× objective) without projections of Z-stacks, and the exposure brightness, contrast, and time were maintained for each photomicrograph. At least, 3 images that were captured with a 20× or 40×objective were counted from each animal (n = 6 biological replications). ImageJ (version 2.0) was used to analyze the uorescent intensity and neuronal counts. For quanti cation of enteric neurons, 3 ganglia were randomly selected from each LMMP preparation, 3 preparations for each animal, and 3 animals in each treatment group. In each ganglion, the number of HuC/D-immunoreactive neuronal bodies and nNOS-immunoreactive neurons was manually counted in a blind fashion. In addition, the intensity uorescence of ChAT was detected by ImageJ, along with several adjacent background readings. Then, the CTCF (corrected total cellular uorescence) = [integrated density − (area of selected cell × mean uorescence of background readings)] was calculated [23]. All the studies were performed in a double-blind fashion.

Western blot analysis
Protein samples were extracted from colonic tissue in RIPA buffer containing complete protease and phosphatase inhibitor cocktail. After assessing the protein content of each sample by BCA Protein Assay kit (Thermo Fisher Scienti c, 23227, USA), they were separated on 10% SDS-PAGE gels and transferred to PVDF membranes. After blocking, the blots were incubated with anti-GFAP (1:10000, GeneTex, GTX100850, USA) primary antibody overnight at 4°C. Subsequently, the membrane was incubated with a secondary antibody (anti-rabbit IgG-horseradish peroxidase, 1:2000, GeneTex, GTX213110-01, USA) followed by visualized with enhanced chemiluminescence. The signals on the blots were detected with Tanon 5200 Multi (Minhang District, Shanghai Municipality, China). β-actin was used as an internal standard. Immunoblots were quanti ed by Image J software (version 2.0).

Statistical Analysis
Results are presented as mean ± Standard Error of Mean (SEM). All statistical analysis was conducted using Prism 7.0 (GraphPad, San Diego, USA). The Student's t-test was used to determine statistical differences between each experimental group and the control group data. One-way ANOVA with Sidak's multiple comparisons test was used for the data with group numbers over two. P < 0.05 was considered signi cant and denoted by *.

LPS induced EGCs reactivity and it was inhibited by FC in mice
We assayed GFAP expression of colonic tissue in LPS-injected animals, as it is a typical identi cation marker of enteric glia and astrocytes to indicate their reactivity. As expected, results of Western blot showed that GFAP expression was signi cantly upregulated after the LPS injection compared with the control group ( Fig. 1a and b, control vs. LPS, P < 0.01). FC is a metabolic compound that makes EGCs metabolism static [17]. To testify whether EGCs are no longer reactive in the mice treated with FC, we injected LPS and assessed the protein expression of GFAP 48 hr later. Western blot analysis showed that FC treatment signi cantly reduced the expression of GFAP in colon tissue compared with the LPS group, while it has no signi cant difference with the control one ( Fig. 1a and b, mean FC + LPS vs. LPS, P < 0.001; vs. control, P > 0.05). In addition, we isolated the LMMPs from colon tissue and detected the intensity of GFAP expression in the myenteric nerve plexus by immunostaining. Compared with the control group, the intensity of GFAP increased in the LPS group ( Fig. 1c-f, mean LPS vs. Control, P < 0.05). However, it decreased in the LPS plus FC group relative to the LPS group ( Fig. 1c-f, mean FC + LPS vs. LPS, P < 0.05). Thus, it indicated that EGCs convert into a reactive state after LPS injection in mice, while, FC treatment impeded this process.

Inhibition of reactive EGCs improves intestinal motility and peristaltic re ex in LPS-injected animals
To determine whether the reactive EGCs are involved in the regulation of GI motility, we measured fecal pellet output, propulsive colorectal motility, total gastrointestinal transit time (TGIT), uid content in vivo. As shown in Figure  CMMCs, which are dependent on ENS [24], were investigated in the isolated preparations of colons to determine whether the observed changes in motility are due to an intrinsic ENS defect. Spatiotemporal maps of CMMCs were constructed and analyzed ( Fig. 2e- Fig. 3c, P < 0.05). No statistically signi cant difference was found in ChAT uorescence intensity between the LPS mice and the LPS mice treated with FC ( Fig. 3d, P > 0.05). These observations suggested that the reactive EGCs cause enteric neuron loss, especially nitrergic neurons in LPS-injected animals.

Reactive EGCs induce primary enteric neurons into apoptosis in vitro
It has been shown that reactive EGCs lead to a decrease of enteric neurons in the myenteric plexus in vivo. We studied how reactive EGCs in uence the state of enteric neurons in vitro. Firstly, IL-1β and TNFα were used to stimulate the enteric glial cell line to develop a serum-free culture model for the reactive EGCs in vitro, mimicking the pathological process in LPS-injected animals (Fig. 4a-d). Meanwhile, it was identi ed that IL-1β and TNFα made primary EGCs reactive (Supplementary Figure S4). Next, puri ed enteric neurons were incubated with Neurobasal medium, the conditioned medium of resting or reactive EGCs. Then the cell apoptosis was assessed using TUNEL tests and cleaved Caspase-3 protein staining. The TUNEL assay showed that primary enteric neurons revealed 27.54 ± 5.28% positive apoptotic neurons at 48 hr after cultured in the conditioned medium of reactive EGCs compared with 3.06% ± 0.34 for cultures in the conditioned medium of resting EGCs ( Fig. 5a and b, P < 0.01). Immuno uorescence of cleaved Caspase-3 protein displayed that 6.84 ± 0.35% positive apoptotic neurons cultured in the conditioned medium of reactive EGCs compared with 3.46 ± 1.16% for neurons incubated with the conditioned medium of the resting EGCs ( Fig. 5c and d, P < 0.01).

Reactive EGCs disrupt neurites formation of primary enteric neurons in vitro
To test if reactive EGCs affect neurites formation of enteric neurons in vitro, primary enteric neurons were incubated with the conditioned medium of resting or reactive EGCs and observed the morphology through TuJ-1 immunostaining (Fig. 6a). The dendritic complexity of neurons cultured in reactive EGCs conditioned medium signi cantly declined compared with those grown in the resting EGCs conditioned medium (Fig. 6b, Fig. 6e, P > 0.05), However, there was no signi cant difference between them (Supplementary Figure S5).

Discussion
In this study, we demonstrated that reactive EGCs caused a signi cant decrease in intestinal motility via increasing the apoptosis of nitrergic enteric neurons in LPS-injected animals. However, this tendency was reversed, when the function of reactive EGCs was inhibited by FC. In addition, reactive EGCs also disrupted the neurites formation of primary enteric neurons in vitro. Together, our results indicate that the reactive EGCs could cause damage to the enteric neuron during endotoxemia, which deteriorates the intestinal motility disorder, suggesting that inhibition of the reactive EGCs could be a therapeutic strategy for the treatment of LPS-induced intestinal paralysis and other intestinal motility disorders.
Early research showed that EGCs play an important role in regulating and protecting enteric neurons, and dysfunction of EGCs will break the intestinal homeostasis, leading to a series of gastrointestinal disorders [25]. Previous studies demonstrate that enteric glia could be activated by bacteria, which will respond to an immune reaction in the gastrointestinal tract and release ATP [14], in ammatory cytokines [26], and prostaglandin, etc [27]. However, whether the reactive EGCs participate in neurodegenerative processes and whether the interaction between EGCs and neurons in uences intestinal motility is yet under discussion. To elucidate this question, we tested the intestinal motility of mice injected LPS and FC plus LPS, as well as the number of colonic LMMP neurons. Compared with the control, there was a signi cant delay of intestinal transit accompanied by a signi cant loss of neurons after the LPS injection. The result revealed that inhibition of EGC reactivity could improve intestinal motility and prevent neurons from losing. Thus, we inferred that the reactive EGCs were associated with the decrease of enteric neurons. Meanwhile, it has been shown that ENS plays a signi cant role in motility regulation[28] and the loss of enteric neurons causes intestinal dysmotility [7]. Therefore, if the reactive EGCs are inhibited, the loss of neurons would be alleviated as well as the intestinal dysmotility.
GFAP, a glial brillary acidic protein, is the typical identi cation marker of enteric glia and astrocytes, whose expression in mature EGCs is modulated by cell differentiation, in ammation, and injury [29]. Increased GFAP has been observed in ulcerative colitis, Crohn's disease [25], and Parkinson's disease [30]. Astrocytes and EGCs can be induced into reactive by cytokines and LPS [31,32]. In line with previous research, there was an upregulation of the GFAP expression in our study [16]. However, GFAP is dynamic and its observed expression varies depending on glial state [33], subtype [34], and genetic targeting strategy [35]. Changes in morphology or GFAP expression are not a de nitive indication of a reactive glial phenotype, since the functional change is more crucial for a different cell phenotype. Additionally, LPS triggers immune responses in endotoxemia, leading to a massive release of cytokines such as IL-1β and TNF-α [4,36]. In our study, we stimulated the enteric glial cell line and primary EGCs with IL-1β and TNF-α in vitro to mimic the in ammatory response induced by LPS in vivo. As expected, the GFAP expression levels increased signi cantly. To make a de nitive indication of a reactive glial phenotype, we collected the conditioned medium of the reactive EGCs and incubated it with the primary enteric neurons, detecting effects of reactive EGCs on neurons survival and neurites' formation.
EGCs were primarily considered as a component to provide structural support for the enteric neural net. Currently, it has been con rmed that EGCs could release nerve growth factor (NGF), neurotrophin 3 (NT-3), and GDNF to maintenance and contribute to the survival of enteric neurons [37,38]. Interestingly, glia transforms into a reactive phenotype in response to intestinal infection with bacteria or viruses, in ammatory mediators [39]. The function of reactive EGCs, though unlike the central nervous system (CNS), is still poorly understood. As shown for the CNS, two different types of reactive astrocytes termed "A1" and "A2" were induced by neuroin ammation and ischemia respectively [40]. In addition, A1 reactive astrocytes become neurotoxic, leading to the neuron's death [41]. As for ENS, the intestinal ischemiareperfusion injury (IRI) leads to the signi cant upregulation of GFAP, which may present a bene cial effect on neurons, because GDNF increased under this condition [12,13].
On the contrary, a study from Gulbransen's group provided evidence that glial reactivity is a driver of enteric neurodegeneration in colitis [14]. Besides, Andromeda's group used human EGCs from GI-surgical specimens and treated them with LPS to study the gene expression and relevant functions. Their results showed that the "reactive human EGC phenotype" leads to alterations of important molecular and functional signaling pathways that could disrupt GI motility [42]. After all, there is still a lack of study about the in uence of reactive EGCs on enteric neurons. In our study, we used a conditioned medium of reactive EGCs to interfere with primary enteric neurons, and the results indicated that the reactive EGCs damaged enteric neurons through increasing neurons' apoptosis and disrupting neurites' formation. On the other hand, it suggested that EGCs might release some speci c substances to be harmful to enteric neurons. Thus, potential substances or targets that reactive EGCs damage enteric neurons could be found in the future study, which could be a potential new therapeutic target to modulate the intestinal dysmotility of patients suffering from sepsis or endotoxemia, or other diseases associated with motility disorders.
Fluoroacetate (FAC) and its metabolite FC were important tools that have been used to de ne the role of astrocytes in central [43,44] and peripheral neural networks previously [45,46]. Also, FC can be selectively taken up by enteric glia in the intestine [47] and cause a reversible and selective disruption of glial function owing to a fall in glia ATP levels through inhibiting the tricarboxylic acid TCA cycle. However, it does not in uence glial and neuronal morphology in both the myenteric and submucosal plexuses of the ileum and colon in mice treated with 20 µmol/kg FC [17]. In our research, EGCs were not induced into reactive by LPS after FC administration, meanwhile, gastrointestinal motility and the number of enteric neurons were not decreased anymore, comparing with the LPS group. Hence, it suggested that the reactive EGCs appear to an adverse effect on intestinal motility in LPS-injected animals.
We found that the amount of total and nitrergic neurons in the myenteric nerve clusters was decreased, except for cholinergic neurons, which is similar to a previous study [6]. The ENS contains many different types of neurons [48]. Cholinergic neurons and nitrergic neurons are mainly involved in intestinal motility regulation [49]. At least 70% of all myenteric neurons are cholinergic, which excite smooth muscle [50]. While most enteric neurons that inhibit smooth muscle contain nitric oxide synthase 1 (NOS1) [50]. This might seem contradictory. On the one hand, it may be relevant to the interaction and coordination among different kinds of enteric neurons, and the broken balance of intestinal neural networks might cause intestinal disorders. On the other hand, this study only focused on two kinds of neurons, changes in other kinds of neurons during endotoxemia are still unclear. All these need further research.

Conclusions
In conclusion, the present data indicated that reactive enteric glial cells are involved in paralytic ileus through promoting nitrergic enteric neuronal apoptosis and impeding the formation of neurites during endotoxemia. These ndings could provide a novel therapeutic strategy for intestinal motility disorders, including LPS-induced intestinal paralysis. In addition, the speci c mechanism about how the reactive  Effects of FC on EGCs reactivity in septic mice. (a-b) Western blot showed that the protein expression of GFAP in colon tissue increased after the LPS injection whereas it was reversed by FC treatment. (c-f) Immunostaining in the colon myenteric plexus showed that the GFAP uorescence intensity increased in the LPS group relative to the control group, however, it was reduced in the LPS plus FC group compared with the LPS group. One-way ANOVA, n = 6 mice per group, * P < 0.05, ** P < 0.01, ns, no signi cant difference. reduced CMMC frequency and velocity in the colon, while they were signi cantly improved by FC treatment. One-way ANOVA, n = 6 mice per group, * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001, ns, no signi cant difference. showed that more positive apoptotic neurons were observed in the conditioned medium of the reactive EGCs group than in the conditioned medium of the resting EGCs. One-way ANOVA, n = 6 biological replications, * P < 0.05, ** P < 0.01, *** P < 0.001, ns, no signi cant difference. Control, blank control (Neurobasal medium); +ECM(+resting EGCs), the conditioned medium of resting EGCs; +ECM(+reactive EGCs), the conditioned medium of reactive EGCs.