Guanylate cyclase C limits systemic dissemination of a murine enteric pathogen
© Mann et al.; licensee BioMed Central Ltd. 2013
Received: 17 May 2013
Accepted: 21 August 2013
Published: 2 September 2013
Skip to main content
© Mann et al.; licensee BioMed Central Ltd. 2013
Received: 17 May 2013
Accepted: 21 August 2013
Published: 2 September 2013
Guanylate Cyclase C (GC-C) is an apically-oriented transmembrane receptor that is expressed on epithelial cells of the intestine. Activation of GC-C by the endogenous ligands guanylin or uroguanylin elevates intracellular cGMP and is implicated in intestinal ion secretion, cell proliferation, apoptosis, intestinal barrier function, as well as the susceptibility of the intestine to inflammation. Our aim was to determine if GC-C is required for host defense during infection by the murine enteric pathogen Citrobacter rodentium of the family Enterobacteriacea.
GC-C+/+ control mice or those having GC-C genetically ablated (GC-C−/−) were administered C. rodentium by orogastric gavage and analyzed at multiple time points up to post-infection day 20. Commensal bacteria were characterized in uninfected GC-C+/+ and GC-C−/− mice using 16S rRNA PCR analysis.
GC-C−/− mice had an increase in C. rodentium bacterial load in stool relative to GC-C+/+. C. rodentium infection strongly decreased guanylin expression in GC-C+/+ mice and, to an even greater degree, in GC-C−/− animals. Fluorescent tracer studies indicated that mice lacking GC-C, unlike GC-C+/+ animals, had a substantial loss of intestinal barrier function early in the course of infection. Epithelial cell apoptosis was significantly increased in GC-C−/− mice following 10 days of infection and this was associated with increased frequency and numbers of C. rodentium translocation out of the intestine. Infection led to significant liver histopathology in GC-C−/− mice as well as lymphocyte infiltration and elevated cytokine and chemokine expression. Relative to naïve GC-C+/+ mice, the commensal microflora load in uninfected GC-C−/− mice was decreased and bacterial composition was imbalanced and included outgrowth of the Enterobacteriacea family.
This work demonstrates the novel finding that GC-C signaling is an essential component of host defense during murine enteric infection by reducing bacterial load and preventing systemic dissemination of attaching/effacing-lesion forming bacterial pathogens such as C. rodentium.
Citrobacter rodentium (C. rodentium) are gram-negative bacteria of the family Enterobacteriaceae, and are natural enteric pathogens of mice. Related family members, enteropathogenic Escherichia coli (EPEC) and enterohemorrhagic E. coli (EHEC), are human pathogens that are a major cause of infectious diarrhea worldwide. Similar to these bacteria, C. rodentium are non-invasive. Instead, these attaching/effacing lesion-forming pathogens carry a set of virulence genes termed the locus of enterocyte effacement which enable close association with the apical membrane of intestinal cells, causing local destruction of microvilli and transfer of effector gene products via a type III secretion system . In C57BL/6 mice and other strains that are considered resistant to C. rodentium, colonization results in a self-limiting infection which is cleared by both the innate and the adaptive immune systems. The course of infection is characterized by intestinal epithelial cell apoptosis, infiltration of inflammatory cells, mainly macrophage and neutrophils, and crypt hyperplasia, all of which are largely resolved within 3 weeks. In contrast, susceptible strains exhibit severe, often fatal, diarrhea which has been attributed at least in part to decreased expression of transporters important in chloride ion homeostasis  and to differences in the intestinal microbiota [3, 4].
Guanylate cyclase 2C (encoded by Gucy2c, hereafter termed GC-C) is a transmembrane receptor present in the intestine which, in concert with its ligands guanylin (GN, encoded by Guca2a,)  and uroguanylin (UGN, encoded by Guca2b) , is known to regulate both chloride/bicarbonate ion secretion (via the cystic fibrosis transmembrane conductance regulator, CFTR) [7, 8] and sodium absorption (via the Na+/H+ exchanger 3, NHE3, Slc9a3) . The first step in this regulatory cascade is ligand activation of GC-C to generate the second messenger cyclic guanosine monophosphate (cGMP). Some strains of infectious enterotoxigenic E. coli produce the heat-stable enterotoxin, ST, which acts as a GC-C superagonist . Unlike activation of GC-C by the endogenous ligands GN or UGN, robust overproduction of cGMP by ST-bound GC-C deregulates downstream signaling pathways and causes secretory diarrhea. Mice carrying a targeted disruption of Gucy2c have reduced levels of intestinal cGMP, and are resistant to ST [11, 12]. Importantly, further studies of Gucy2c knockout (hereafter called GC-C−/−) mice have provided evidence for actions of GC-C beyond fluid and ion homeostasis in the gut. Signaling via GC-C has been implicated in intestinal epithelial cell cycle regulation, apoptosis, and progression to gastrointestinal cancer [13–15]. More recently, we have shown the presence of a defective intestinal epithelial barrier as well as sensitivity to DNA damage-induced cell death in mice lacking GC-C [16, 17]. In a model of chemical-induced intestinal injury, GC-C−/− mice exhibited a blunted intestinal inflammatory response that was accompanied by reduced cytokine expression . Based on these findings, we utilized the C. rodentium model of enteric bacterial pathogen infection to investigate the hypothesis that GC-C effector pathways are an important aspect of host defense in the intestinal mucosa.
GC-C−/− mice carrying a targeted deletion of Gucy2c, the gene encoding GC-C, have previously been described by us . Heterozygous GC-C+/− mice have been back-crossed to the C57BL/6 J strain (Jackson Laboratory, Bar Harbor, Me) for > 10 generations, and GC-C−/− and GC-C+/+ mouse lines were generated by this process. As in our recent work [16, 18], both GC-C+/+ and GC-C−/− mouse lines were maintained in the same room of our animal facility under identical specific-pathogen free conditions. Mice of both sexes, aged 6–12 weeks were used, and experiments were performed on age- and gender- matched groups. All studies were approved by the Cincinnati Children’s Hospital Medical Center Institutional Animal Care and Use Committee.
C. rodentium bacteria from frozen stocks of strain DBS 100 (gift of Philip M. Sherman, Hospital for Sick Children, Toronto, Canada) were grown on MacConkey’s agar (Becton, Dickinson and Company, Sparks, MD) overnight at 37°C. A single colony was then cultured in Luria-Bertani broth overnight at 37°C. We used optical density measurements at 600 nm to assess the concentration of bacteria, and confirmed colony forming units (CFU) by plating serial dilutions on MacConkey’s agar. Within each study, groups of GC-C+/+ and GC-C−/− mice with free access to food and water were infected by oral gavage with the same freshly prepared mixture of C. rodentium (~1.5 × 109 CFU C. rodentium in 100 μl sterile PBS per mouse). Stool from individual mice was collected and weighed every few days after gavage, beginning at day 4. To determine bacterial burden, stool was homogenized in sterile PBS (0.1 g stool/1 ml PBS) and serial dilutions were plated on MacConkey’s agar. Colonies were counted after overnight incubation at 37°C. The limit of detection was 5 × 102 CFU/g stool. A mouse was excluded from further analysis if the level of C. rodentium was below the limit of detection at 4 days post-infection. PCR analysis of the C. rodentium espB gene was performed on representative colonies to confirm identity.
Groups of mice were sacrificed at 4, 10, and 20 days after infection. Colon, feces, and liver were aseptically removed. A caudal piece (0.5 cm) of the distal colon was frozen in liquid nitrogen for subsequent RNA extraction, while the remaining section was split longitudinally into “swiss rolls” for formalin fixation/paraffin embedding and for frozen OCT embedding. Sections of liver were frozen in liquid nitrogen for RNA analysis or fixed in formalin/paraffin for histology. In addition, liver sections (trimmed to 0.1 g weight) were homogenized in 1 ml of sterile PBS, and plated on MacConkey’s agar. C. rodentium colonies were counted after incubation at 37°C overnight. The limit of detection for liver homogenates was 40 CFU/g and identity was confirmed by PCR analysis performed as above. Tissues from additional naïve (non-infected) age- and gender-matched mice were similarly obtained and processed.
H&E staining of intestinal and hepatic sections was done using standard techniques as previously described [18, 19]. Images were obtained on an Olympus BX51 microscope equipped with an Olympus DP71 camera and DP Manager software. Measurements of crypt depth were taken on micrographs of all well-oriented crypts from H&E-stained “swiss roll” sections of distal colon using ImageJ version 1.38 (National Institutes of Health, Bethesda, MD). Stained slides (distal colon) were also examined to assess colitis severity in terms of disease scores composed of the degree of inflammation, hyperplasia, and infiltrate composition using a modified scoring system based on previous studies [18, 20, 21]. Briefly, this system assessed inflammation (scored 0–4 where 0 is no inflammation and 4 is severe ulceration and crypt abscess), hyperplasia (0 – 3 where 0 is no hyperplasia and 3 is severe hyperplasia with mucin depletion), and inflammatory composition (0 – 3 where 0 is none and 3 is substantial mononuclear cells with high neutrophil content). C. rodentium was localized in the distal colon by incubating OCT frozen sections with anti-E. coli Polyvalent 8 LPS antibody (1:500, Denka Seiken Co., Tokyo, Japan) as according to protocol and helpful advice from Dr. Bruce Vallance (University of British Colombia, Canada) . Sections were then incubated with anti-swine CF-488A-conjugated IgG secondary antibody (1:2000, Sigma-Aldrich Corp., St. Louis, MO) and counter-stained with DAPI. Apoptosis was quantitated via immunofluorescence with cleaved caspase 3 antibody (CC3, antibody #9661, Cell Signaling Technology, Inc., Danvers, MA) as performed previously on OCT frozen sections [19, 23]. Positive-stained intestinal epithelial cells were tabulated from 8–12 micrograph fields (original magnification 200x) per mouse. Immunofluorescence of E-cadherin (#13-1900, Life Technologies; Carlsbad, CA), claudin 2 (#51-6100, Life Technologies) and claudin 3 (#341700, Life Technologies) was performed on frozen sections in a similar manner. Tubulin, claudin 2, and claudin 3 immunoblotting was performed as previously described on extracts from homogenized colon.
Real time RT-PCR primers
For 5’- CAC TGT GCA GGA TGG AGA C-3’
Rev 5’-CTC GGC GTT GGG TTT CT-3’
For 5’-GCA CAG CAG CTC AAA CAA CTG GAA-3’
Rev 5’-TTC TCA TTT GGA ACC AGC GCA AGC-3’
For 5’-AAA CTG TTC CGA GGA GTC AGT GCT
Rev 5’-GCT GAG CTG ATT GCT GAG TTT GGT-3’
For 5’-TTT CAT CAC GCC CTT GAG CCT AGT-3’
Rev 5’-TTT GGT GAC GTG AGC CTC AGA AGT-3’
For 5’-ACC ACA GTC CAT GCC ATC AC-3’
Rev 5’-TCC ACC ACC CTG TTG CTG TA-3’
Intestinal permeability was assessed by measuring movement of a fluorescent tracer (4 kDa FITC-dextran, Sigma-Aldrich, St. Louis, MO) from the gut lumen to the blood. Uninfected and day 4 C. rodentium infected mice of both GC-C+/+ and GC-C−/− genotypes were orally gavaged with 200ul of 70 mg/ml FITC-dextran dissolved in water. Mice were given FITC-dextran solution four hours prior to blood collection at sacrifice. Serum samples were diluted and read in a 96 well plate at 485 excitation/535 emission along with a FITC-dextran standard curve. The calculated serum concentration of FITC-dextran was interpreted as an indication of intestinal permeability.
Bacterial DNA was isolated with the QIAamp DNA Stool Mini Kit (QIAGEN, Valencia, CA) according to manufacturer’s instructions for pathogen DNA, including the optional 95°C incubation step. Stool from individual adult mice was weighed, frozen at −80°C, and homogenized directly in ASL lysis buffer (QIAGEN) prior to DNA extraction. Genomic DNA from cultures of the laboratory Escherichia coli strain, DH5α, was also prepared using this kit. Integrity of DNA was checked by agarose gel electrophoresis, and the concentration determined using the Quant-iT dsDNA Broad-Range Assay Kit (Invitrogen, Molecular Probes, Inc., Eugene, OR). To quantify bacteria, real-time PCR was performed using Brilliant II SYBR Green QPCR mix (Stratagene) and published sets of group- or family-specific 16S ribosomal RNA (rRNA) gene primers as follows: Eubacterium rectales/Clostridiae cocoides, Enterobacteriacea, AtopobiumLactobacillus, Bacteroides-Prevotella-Porphyromonas and Bifidobacteria. For determining total bacterial load, universal 16S rRNA gene primers that recognize all eubacteria were used . A standard curve of serial dilutions of a known amount of DH5α DNA was generated, and used to interpolate the quantity of bacteria from each experimental DNA sample. To assess the relative quantity of each group the ∆Ct method was used with normalization by total bacteria values .
Categorical variables were evaluated by Fisher exact test. Differences in continuous variables were compared using the Mann–Whitney test or 2-tailed Student t test. For all analyses, statistical significance was set at P ≤ 0.05 and was determined using Prism Version 5.03 (GraphPad Software, Inc., San Diego, CA).
We next performed a histological survey of liver tissue from wildtype and GC-C−/− mice. Day 4 p.i. liver histology revealed mild inflammatory infiltrate with no indication of increased severity in GC-C−/− mice as compared to GC-C+/+ animals (data not shown). Consistent with a lack of intestinal containment of C. rodentium, focal areas of necrosis at day 10 p.i., were noted in the liver of GC-C−/− mice upon necropsy whereas none could be seen in the livers of any mice of the GC-C+/+ group (Figure 6B). This damage was apparent histologically as well, and included multiple areas of inflammatory infiltrate visible upon microscopic examination of GC-C−/− liver sections (Figure 6C).
Targeting of GC-C by ST results in severe secretory diarrhea and is a significant cause of morbidity and mortality in children of developing countries. It has long been speculated that this GC-C-conferred susceptibility to some forms of enteric infection is offset by a critical role for this receptor in intestinal homeostasis. We and others have suggested that GC-C is important for epithelial ion transport, cell proliferation, and barrier function and that GC-C modulates intestinal disorders ranging from cystic fibrosis to gastrointestinal cancer to intestinal injury and inflammation via these effects [14, 16–18, 45]. Here, we extend the role of GC-C in the intestine by demonstrating that, although it is exploited by ST-producing ETEC, this pathway is highly protective during enteric infection by attaching/effacing bacterial pathogens which do not produce ST toxin. Specifically, we show that GC-C is required to minimize bacterial load during the early to middle phases of C. rodentium infection. We also demonstrate that GC-C activity is essential for maintaining the integrity of the intestinal epithelial barrier, both by reducing permeability early in infection as well as by suppressing epithelial apoptosis at later time points. The important role of GC-C in controlling pathogen load and intestinal barrier function is likely instrumental in minimizing systemic dissemination of attaching/effacing lesion forming bacteria such as C. rodentium. Furthermore, we demonstrate several additional novel aspects of epithelial GC-C/cGMP signaling. First, in the context of the C57BL/6 genetic background, infection by attaching/effacing bacteria such as C. rodentium does not significantly disrupt intestinal fluid homeostasis in the absence of GC-C. Second, loss of GC-C has no apparent impact on the hyperplasia of epithelial crypts that occurs in response to infection by attaching/effacing bacterial pathogens. Third, Gn expression is profoundly reduced during C. rodentium infection and this occurs prior to epithelial hypertrophy or substantial immune cell infiltration. Fourth, GC-C regulates the number and composition of intestinal commensal microflora in naïve mice.
Several multifunctional effector proteins (such as EspF and Map) common to C. rodentium and other attaching/effacing pathogens, cause host epithelial cell apoptosis as well as the disruption of intercellular tight junctions [37, 46, 47]. Bacterial translocation is associated with the intestinal cell death, inflammation, and barrier dysfunction seen during experimental C. rodentium infection [31, 39, 48]. We have previously reported increased radiation-induced intestinal epithelial cell apoptosis and epithelial barrier dysfunction in response to LPS challenge that resulted in translocation of commensal bacteria in GC-C−/− mice [16, 17]. In the current study, we show both a loss of barrier integrity in the intestine at early stages of infection and a significant elevation in intestinal epithelial apoptosis in GC-C−/− mice that culminates in translocation of C. rodentium out of the intestine. While this correlates with the increased bacterial load in stool from GC-C−/− at day 10, it is notable that we did not find major differences in the localization of C. rodentium colonization within the colon of GC-C−/− mice as compared to GC-C+/+. At day 4, prior to translocation of live bacteria, there was significantly increased intestinal permeability in GC-C−/− mice and this correlated with elevated expression of chemokine and acute phase genes in the liver as compared to GC-C+/+. Collectively, this work indicates that GC-C activity is an essential component of epithelial barrier function and anti-apoptosis and is required for multiple aspects of intestinal barrier function during enteric infection.
There is increasing evidence that complex, reciprocal crosstalk between the host and its commensal microflora has a broad and profound impact on gastrointestinal mucosal immunity and is a critical factor in defining susceptibility to cancer, inflammation, obesity as well as infection by bacterial pathogens . Fecal transplant from a C. rodentium-resistant mouse strain reduced mortality in a susceptible murine genetic background [3, 4]. In addition, treatment with the probiotic bacteria Lactobaccillus promoted survival of C. rodentium-infected neonatal mice , while exposure to a stressor (physical restraint) both altered gut microbial composition and increased colonization by C. rodentium. Many factors shape the influence of commensal microflora on colonization resistance to a pathogen, and may include direct inhibition by bacteriocins (antimicrobials generated by bacteria), bacterial metabolites, or competition for nutrients . Alternatively, the abundance of closely related family members may be indicative of the potential success of a pathogen from the same family with related growth characteristics . Because an imbalance in intestinal microflora often occurs in mice with deregulated ion transport pathways [41, 43], we investigated whether this may contribute to the increased C. rodentium colonization noted in GC-C−/− mice. Here, we have demonstrated that loss of GC-C has a significant impact on the abundance and systemic dissemination of an enteric bacterial pathogen and we further show that naïve GC-C−/− mice have a significant imbalance in intestinal bacterial species relative to their GC-C+/+ counterparts. In addition to a decrease in potentially protective Lactobacillus, the Enterobacteriaceae family colonizes the gut of naïve GC-C−/− mice at higher levels than GC-C+/+, as does the Enterobacteriaceae family member C. rodentium during the early stages of the infection course. It is possible that GC-C activity regulates outgrowth of commensal bacteria as well as C. rodentium by affecting epithelial mucus layer dynamics, cell surface or luminal pH, or the ion/solute-dependent aspects of bacterial metabolism and growth. Additional studies will be necessary to define the mechanism through which GC-C influences the composition and load of commensal bacteria.
External factors, such as variation in diet and animal husbandry conditions at any given research institute, impact the composition of gut microflora and can have a dominant influence on disease model phenotype [53–55]. Therefore, genetically-modified mice used as models of gastrointestinal pathology such as bacterial infection, inflammation, obesity, and tumor susceptibility reflect the interaction between investigator-imposed genetic modifications and institution- or animal colony-dependent microflora. Further characterization and study of the microbiome of the GC-C−/− intestine will be necessary to establish its specific role in susceptibility to bacterial infection as well as its putative broader impact on gastrointestinal pathophysiology.
In summary, we have identified a protective role for GC-C signaling in defense against an attaching/effacing lesion-forming bacterial pathogen. Our studies have shown that GC-C is required to maintain an effective epithelial barrier and to suppress systemic dissemination of C. rodentium. We also demonstrate an imbalance in commensal microflora in naïve GC-C−/− mice relative to GC-C+/+ animals from the same colony and speculate that this is an important component in the increased colonization of enteric pathogens in mice lacking GC-C.
Guanylate cyclase 2C, GUCY2C
Cystic fibrosis transmembrane conductance regulator
Na+/H+ exchanger 3
Chemokine (C-X-C motif) ligand 1
Chemokine (C-X-C motif) ligand 9
Lipopolysaccharide binding protein
This work was supported by the AGA Foundation for Digestive Health and Nutrition, the Crohn’s and Colitis Foundation of America, NIH R01DK047318, and, in part, by PHS Grant DK P30DK078392.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.