|Home | About | Journals | Submit | Contact Us | Français|
Ileo-cecal resection (ICR) is common in Crohn’s disease (CD). Inflammation and fibrosis frequently recur at the site of anastomosis or in the small intestine (SI). No animal models of post-surgical inflammation and fibrosis exist. We developed a model of ICR in IL-10 null and wild-type (WT) mice to test the hypothesis that that ICR promotes post-surgical inflammation and fibrosis in SI or anastomosis of genetically susceptible IL-10 null, but not WT or germ free (GF)-IL-10 null mice.
GF-IL-10 null mice were conventionalized (CONV) and 3 weeks later randomized to ICR, transection (T) or no treatment (NoTx). Age-matched conventionally raised (CONV) WT and GF-IL-10 null mice received ICR, T or NoTx. Animals were killed 28 days later.
Histological scoring, real-time PCR for TNFα and collagen, and immunostaining for CD3+ T cells, assessed inflammation and fibrosis.
After ICR, CONV-IL-10 null, but not CONV-WT mice, developed significant inflammation and fibrosis in SI and inflammation in anastomosis compared to NoTx or T controls. Fibrosis occurred in anastomosis of both CONV-IL-10 null and CONV-WT following ICR. GF-IL-10 null mice developed little or no inflammation or fibrosis in SI or anastomosis after ICR.
ICR in CONV-IL-10 null mice provides a new animal model of post-surgical inflammation and fibrosis in SI and anastomosis. Absence of inflammation and fibrosis in SI of CONV-WT and GF-IL-10 null following ICR indicates that post-surgical small bowel disease occurs only in genetically susceptible IL-10 null mice and is bacteria dependent.
Crohn’s Disease (CD) is an incurable inflammatory bowel disease (IBD) [1, 2]. CD involves chronic inflammation of the intestine, damage and distortion of tissue architecture and loss of digestive, absorptive and motility functions. Disease onset commonly occurs in the ileum and cecum. Inflammation is transmural in CD, and is associated with transmural increases in collagen deposition, thickening of the muscularis layers, and fibrosis, which can lead to stenosis and bowel obstruction [3, 4]. As many as 65% of CD patients require surgical intervention for unmanageable inflammation or fibrostenosing disease and many require repeat surgery for recurrent disease [5, 6]. Ileo-cecal resections (ICR) are common surgical interventions in CD and are associated with high rates of disease recurrence . After ICR or other surgeries, recurrent inflammation or fibrosis typically occurs at the anastomosis and in small intestine (SI) immediately upstream of the anastomosis [5, 6, 7, 8, 9, 10]. Current therapies do not effectively prevent post-surgical recurrence of CD and there are no approved or effective medical therapies for fibrostenosing disease . Major challenges in developing or testing new therapies, are the inherent heterogeneity of CD at surgery, variable definitions of recurrence and duration of follow-up and the fact that endoscopic evaluation of recurrent disease may not fully inform about fibrosis [4, 6, 7, 8, 9, 10]. An animal model would offer a well-controlled experimental system to address mechanisms of post-surgical inflammation and fibrosis in small bowel or anastomosis, and to test therapies.
Available evidence suggests that commensal microflora initiate and perpetuate IBD in genetically susceptible hosts [11, 12, 13]. Antibiotics have proved useful as therapies for specific complications of CD such as localized peritonitis, bacterial overgrowth due to stricture and during drainage therapy for abscesses and perianal disease [14, 15, 16, 17, 18]. More limited evidence implicates microflora in post-surgical CD. In small numbers of patients, antibiotics have reduced postoperative disease recurrence [18, 19] or promoted remission of ileocolonic anastomosis disease . One study reported that ICR altered the profile of ileal microflora and that recurrent disease was associated with a high prevalence of adherent E. coli, enterococci, bacteroides, and fusobacteria . Very little is known about the in vivo roles of microflora in fibrosis associated with CD or post-surgical fibrosis, despite in vitro evidence that bacterial ligands stimulate proliferation and collagen accumulation in intestinal mesenchymal cells .
Numerous animal models of IBD exist [11, 23]. The IL-10 null model has some features in common with CD [11, 23]. Germ free IL-10 null mice do not develop disease. Colonization of GF-IL-10 null mice with microflora from conventionally raised animals reliably initiates disease, in cecum and colon, with a highly reproducible time course and location . Ciprofloxacin and metronidazole improve disease in CONV-IL-10 null mice , and each antibiotic preferentially affects cecal or colonic disease . This is consistent with findings that different bacteria initiate inflammation in either the cecum or colon of IL-10 null mice . One limitation of IL-10 null mice as a model of CD is that these animals do not typically develop small bowel disease.
Recently, we applied ICR to wild-type C57BL6 mice . In the current study we applied ICR to CONV-IL-10 null mice and CONV-WT mice on the 129SvEv background. We tested the hypothesis that ICR may promote post-surgical inflammation or fibrosis in the small intestine or anastomosis of genetically susceptible IL-10 null, but not wild-type mice. We chose the IL-10 null mouse to develop a model of post-surgical inflammation and fibrosis because of its wide acceptance as an animal model with relevance to human IBD and the availability of germ-free IL-10 null mice. We successfully adapted the ICR model to germ-free IL-10 null mice to definitively test whether enteric microflora promote post-surgical inflammation or fibrosis.
Eight week old germ free (GF) 129SvEv IL-10 null mice were obtained from the UNC gnotobiotic rodent facility. Animals were transferred to specific pathogen free housing and colonized with fecal slurries from conventionally raised mice as previously described . Three weeks later, when colonization reproducibly initiates disease in the cecum and to a lesser extent colon , mice were randomized to sham-operation (T), ileo-cecal resection (ICR) or no surgery (NoTx) groups. Aged matched, conventionally raised WT 129SvEv (CONV-WT) mice were obtained from Jackson labs, Maine, and assigned to NoTx, T or ICR groups. All study groups were given liquid diet (Micro-Stablized Rodent Liquid Diet 101/101A, Purina Mills, MO) 2 days before and 7 days after the start of the experiment. Liquid diet minimizes complications that can occur post-operatively due to obstruction at the anastomosis by solid fecal material [26, 27]. Surgeries were performed under sterile conditions with the aid of the operating microscope (7X magnification). Sham operations (T) consisted of transection and re-anastomosis of bowel approximately 12 cm proximal to the ileocecal junction. For ICR, small bowel was divided 12 cm proximal to the ileocecal junction and at the proximal colon 2–3 cm distal to the cecum. The mesentery was ligated and the intervening SI, cecum and proximal colon removed (Figure 1). Intestinal continuity was restored using an end-to-end, single layered anastomosis with interrupted 9-0 monofilament sutures. Mice were rehydrated with 1–2 ml of warm intraperitoneal saline and the abdomen was closed with running suture. Survival of surgery was 90% for IL-10 null and 87% for WT. The few animals that died or required euthanasia did so within 2–4 days after surgery and were not included in the analyses.
GF-IL-10 null mice were obtained from the same UNC colony as animals that were conventionalized before surgery. Surgery was performed as described above but in a customized GF surgical isolator equipped with a port that permits sterile transfer of mice directly from isolator to hood. Liquid diet was sterilized by irradiation. Pilot studies in CONV-mice, verified that sterile diets result in identical weight gain as non-sterile diets. To verify GF status after surgery, fur and fecal swabs were collected at 4 and 7 days after surgery and from co-housed NoTx controls. Fecal samples were collected weekly thereafter. Aerobic and anaerobic cultures, Gram stains, and culture for molds, performed by our gnotobiotic facility [24, 25], verified lack of contamination. Enlarged cecum in GF-IL-10 null mice given T or NoTx, co-housed with ICR animals, provided an additional measure that GF status was retained after surgery. Culture and Gram stains of feces collected at tissue harvest were uniformly negative. Survival of GF-IL-10 null mice after surgery was 87%.
Analyses focused on mice subject to ICR or T, and NoTx at 28 days after surgery to study the chronic responses to surgery rather than immediate, acute inflammation or wound healing responses to surgery. A subset of CONV IL-10 null and WT animals were also studied at 7 and 14 days after ICR to assess the time course of post-surgical inflammation. Animals were weighed, anesthetized and killed and entire small bowel and colon were dissected. A one cm segment spanning the anastomosis and two one cm segments of small intestine proximal to the anastomosis were collected for histology or RNA extraction (Figure 1). A segment of colon was collected from CONV-IL-10 null mice after ICR, T or NoTx for histological scoring of colitis. Histology samples were fixed in 10% formalin and paraffin embedded and samples for RNA were frozen.
Coded hematoxylin and eosin sections were scored for inflammation by an observer blinded to treatments. Scoring used a well-validated system developed for IL-10 null mice , which assigns a score of 0 – 4 for inflammation and mucosal damage based on degree and extent of transmural inflammation, goblet cell depletion and architecture distortion. Sections were stained with Sirius red to visualize collagen accumulation and scored for fibrosis using a validated scoring system as detailed in . Briefly, experimental sections were compared with Sirius red stained sections of normal small intestine and scored on a scale of 0–5 for increased extent of collagen deposition throughout different layers of the bowel wall, multiplied by percent of section involved (1=< 25%; 2=<50%, 3=<75% and 4=<100%), or intensity of staining (1=no increase and 2=increased intensity). In prior studies , fibrosis scores showed highly significant correlation with biochemical measures of fibrosis based on collagen mRNA or protein.
TNFα and procollagen α1( I) mRNAs were quantified as independent measures of inflammation and fibrosis. Total RNA was extracted using TRIZOL (Invitrogen, CA). Four micrograms of DNAse treated (TurboDNA, Ambion, TX) RNA was reverse transcribed into cDNA using oligodT primers, reverse transcriptase (Promega, WI) and standard conditions. TNFα and collagen mRNA abundance was assessed using quantitative real-time PCR (qRT-PCR) performed on the Rotorgene (Corbett Instruments, Sydney, Australia) using Syber green reagent (Sigma-Aldrich, MO) and cycling conditions optimized for each primer and probe set. Primers were as follows: TNFα: F: 5′-CTGTCTACTGAACTTCGGGGTGAT-3′; R: 5′-GGTCTGGGCCATAGAACTGATG-3′ and procollagen α1(I): F: 5′-GGTATGCTTGATCTGTATCTGC-3′; R: 5′-AGTCCAGTTCTTCATTGCATT-3′ and Hydroxymethylbilane synthase (HMBS), F: 5′-TGTGTTGCACGATCCTGAAAC-3′ and R: 5′-CTCCTTCCAGGTGCCTCAGAA-3′ was used as a constitutively expressed housekeeping control. All test samples in given comparison groups were run in the same qRT-PCR assay, together with a standard curve comprising serial 10-fold dilutions of known concentrations of product. Cycle thresholds for each test mRNA were normalized to HMBS using Rotorgene software (Corbett, Australlia). Abundance of TNFα and collagen mRNAs were expressed as fold-change from the mean expression levels in wild type mice given no treatment.
5μm sections were de-paraffinized and antigen retrival performed using Tris-Triton buffer pH7.6 before overnight incubation at 4°C with rabbit anti-CD3 (A0452 Dako, Denmark) diluted 1:200 with 5% normal goat serum in Tris-Triton buffer. Samples were incubated for 60 min at room temperature with goat anti-rabbit (111-065-144, Jackson) diluted 1:500 with Tris-Triton buffer. Immunostaining was visualized using avidin-biotin DAB kit (Zymed/Invitrogen cat#00-2014).
All data are expressed as mean ± standard error. ANOVA was used to test for statistical differences across groups. Tukeys test was used for post-hoc pair-wise comparisons and statistical significance was set at p≤0.05.
Figure 2 shows representative photographs of entire bowel of CONV-WT or CONV-IL-10 null mice after T or ICR. Note the evident cecal and colonic inflammation in CONV-IL-10 null mice given T but normal small bowel. Note the evident inflammation in small intestine and at anastomosis of CONV-IL-10 null mice but not WT after ICR. Figure 3 shows histological inflammation scores for small intestine and anastomosis of CONV-WT and CONV-IL-10 null mice given NoTx, T or ICR. In WT mice at 28 days following ICR, inflammation scores in small intestine and anastomosis were low and did not differ significantly from T or NoTx controls. IL-10 null mice showed significantly increased inflammation scores in both the small intestine and anastomosis compared with IL-10 null given T or NoTx. Inflammation scores for small intestine and anastomosis of un-operated IL-10 null mice, or animals given transection, did not differ significantly from WT mice, consistent with previous studies demonstrating lack of small bowel disease in un-operated CONV IL-10 null mice . Thus, development of inflammation in small intestine and anastomosis after ICR occurs only in genetically susceptible IL-10 null mice. Scoring of inflammation in colon confirmed colitis in CONV-IL-10 null mice and revealed that ICR or T did not affect disease severity in colon (colitis scores: NoTx 2.8 ± 0.4; T 2.4 ± 0.5; ICR 2.5 ± 0.4). Photomicrographs of representative H&E stained sections in small intestine of CONV-IL-10 null mice after ICR (Figure 4) illustrate transmural inflammation (Fig. 4A and 4B) and infiltration of small intestine with CD3+ T cells ICR (Fig. 4C). Representative photomicrographs of the anastomosis demonstrate inflammation in IL-10 null and not WT mice after ICR (Figure 4B). Figure 5 shows low and high power histology in small intestine of CONV-IL-10 null mice at 7, 14 and 28 days after ICR compared with WT at 7 days after ICR. Note the transmural inflammation in CONV-IL-10 null mice at 7 and 14 as well as 28 days after ICR with no evidence for inflammation in WT at 7 days (Figure 5) or other times. Subsequent studies focused on the 28 day time point.
In small intestine and anastomosis real-time qRT-PCR revealed significantly increased TNFα mRNA abundance in IL-10 null mice after ICR compared with all other IL-10 null or WT groups (Figure 6). These biochemical data verify histological inflammation scores.
In small intestine of CONV-WT mice, neither ICR nor T significantly increased fibrosis scores compared with NoTx controls (Figure 7). CONV-IL-10 null mice showed major increases in fibrosis scores in small intestine after ICR whereas control transection did not result in fibrosis of small intestine. The anastomosis showed a different pattern. After ICR, significant fibrosis was observed at the anastomosis of both CONV-WT and CONV-IL-10 null mice (Figure 7). In mice given T, mean anastomosis fibrosis scores in both IL-10 null and WT were higher than but not significantly different than scores in untreated small intestine which reflected high inter-animal variability. These results indicate that fibrosis at the anastomosis is likely a consequence of ICR or surgical manipulation and is not influenced by IL-10 deficiency. Figure 8 shows representative Sirius red stained sections to illustrate fibrosis in small intestine of only CONV-IL-10 null mice after ICR, but fibrosis at anastomosis of CONV-WT and CONV-IL-10 null mice after ICR or T.
In small intestine RT-qPCR revealed significantly increased levels of procollagen α1(I) mRNAs in IL-10 null mice after ICR compared with all other groups (Figure 9). These results provide important biochemical confirmation of histological scoring data. At the anastomosis RT-qPCR did not reveal an increase in collagen mRNA in WT or IL-10 null mice after T or ICR compared with NoTx controls (Figure 9). IGF-I and TGFβ mRNA assays revealed no close correlation between expression levels of these fibrogenic mediators and collagen mRNA or fibrosis scores within small intestine or anastomosis (supplemental Figure 1).
In small intestine of GF-IL-10 null mice, in contrast to small intestine of CONV-IL-10 null mice, ICR did not lead to a significant increase in inflammation scores (Figure 10). While GF IL-10 null mice showed higher mean inflammation scores at the anastomosis following ICR, this was not statistically significant. Inflammation scores in small intestine of GF-IL-10 null mice after ICR were significantly lower than in small intestine of CONV-IL-10 null mice after ICR, and a similar pattern was observed at the anastomosis (Figure 10). Photomicrographs of representative H&E stained sections are shown in Figure 11A. CD3 staining for T-cells showed no evidence of increased T-cell infiltration into small intestine of GF-IL-10 null mice after ICR (Figure 11B). Together these data demonstrate that commensal microflora is required to initiate post-surgical small bowel inflammation in IL-10 null mice.
In GF-IL-10 null mice neither ICR nor T induced significant increases in fibrosis scores in small intestine, and fibrosis scores were significantly higher in small intestine of CONV versus GF-IL-10 null mice after ICR (Figure 12). Fibrosis scores at the anastomosis of a few GF animals receiving either T (3 out of 8 animals) or ICR (1 out of 7 animals) were higher than mean NoTx fibrosis scores. However, overall mean anastomosis fibrosis scores in GF- IL-10 null mice given ICR or T were significantly lower than in CONV IL-10 null mice given ICR (Figure 12) suggesting that bacteria promote fibrotic responses at the site of surgery. Figure 13 shows representative photomicrographs of Sirius red stained sections to illustrate little or no fibrosis in small intestine of GF IL-10 null mice following ICR, compared with NoTx, or T, and mild fibrosis at anastomosis in a subset of GF-IL-10 null mice given T or ICR. However comparisons of Figure 8 and and1313 demonstrate much more dramatic fibrosis in CONV IL-10 null mice after ICR.
The primary goal of this study was to develop an animal model of post-surgical small bowel inflammation and fibrosis. Using three independent measures, histological scoring, TNFα mRNA abundance and T cell infiltration, we provide evidence that ICR results in sustained post-surgical inflammation in the small intestine and anastomosis of CONV-IL-10 null mice but not CONV-WT at 28 days after ICR. Un-operated CONV-IL-10 null mice do not typically develop small bowel disease [24, 25]. While we do not yet know the precise mechanisms that elicit post-surgical inflammation in the small intestine and anastomosis of CONV-IL-10 null mice after ICR, a likely contributing factor is exposure of the anastomosis or small intestine to microflora or luminal contents from the colon due to loss of ileo-cecal valve. A recent review highlighted exposure to luminal agents including bacteria, as causative factors in post-surgical disease in CD patients, particularly those with ileocolonic anastomosis . The fact that GF-IL-10 null mice do not develop inflammation after ICR supports this concept. However, findings that inflammation develops in small intestine and anastomosis of CONV-IL-10 null, but not CONV-WT mice, demonstrate that post-surgical small bowel inflammation occurs only in IL-10 null mice with genetic susceptibility to IBD or ongoing IBD. SI and anastomosis are typical sites of inflammation and fibrosis in CD patients after ileo-colonic resection [5, 6, 7, 8]. While we must be cautious in extrapolating the findings in our animal model to post-surgical disease in CD, ICR in CONV and GF-IL-10 null mice provides a new and unique model to pursue future studies of the mechanisms of post-surgical inflammation including the role of specific commensal microflora. It will be of significant interest for example to assess if specific bacteria linked to recurrent disease in CD patients  induce post-surgical small bowel disease in GF-IL-10 null mice after ICR.
There is no consensus that typically employed post-surgical therapies such as immunosuppressives or antibiotics reduce post-surgical disease recurrence in CD patients [6, 7] and nor is it clear whether newer therapies such as anti-TNFα impact on post-surgical disease . Our animal model of post-surgical inflammation provides a well-controlled experimental model to test the effects of anti- TNFα and other emerging therapies such as probiotics [2, 32, 33]. Modulation of luminal flora by probiotics [12, 34, 35, 36, 37, 38] is a potential new therapy for CD or post-surgical CD. Trials of single probiotics in CD [39, 40] or post-surgical CD  did not improve disease, leading to a consensus that more information is needed before probiotics are used more widely in CD [15, 34, 38, 42]. The animal models developed in our study should prove particularly useful for future preclinical testing of probiotics during post-surgical inflammation.
A common cause of surgery or repeat surgery in CD patients is stricture, caused by fibrostenosing disease [5, 6, 7, 8]. Much less is known about mechanisms of fibrosis or stricture in CD than inflammation, and even less is known about mechanisms of post-surgical fibrosis and stricture. Histological scoring and assays of collagen mRNA abundance demonstrate fibrosis as well as inflammation in small intestine and anastomosis of IL-10 null mice after ICR. Only IL-10 null and not WT mice given ICR developed post-surgical fibrosis in small intestine distant from the site of surgery. However, WT given ICR and a subset of IL-10 null and WT given transection, develop fibrosis at the anastomosis in the absence of detectable inflammation. This suggests that development of post-surgical fibrosis in small bowel during active inflammation may have distinct mechanisms than fibrosis at the anastomosis. Consistent with this concept, post surgical fibrosis of small intestine, but not anastomosis, was associated with increases in procollagen α1(I) mRNA indicative of ongoing collagen synthesis. A recent review suggested that strictures which are inflammatory, fibrotic or both, may respond differently to therapies such as anti-TNFα . Testing of anti-TNFα or other therapies in IL-10 null versus WT mice given ICR or control transection could therefore be very valuable in differentiating effects of existing or new therapies on surgery-induced fibrogenic responses at anastomosis versus fibrosis associated with active inflammation in small intestine. Our findings that GF-IL-10 null mice given ICR exhibit only a low incidence and much less severe fibrosis in small intestine or anastomosis, provide new in vivo evidence to support findings in cultured cells that bacteria or bacterial products can drive fibrosis . This sets the stage for future studies where GF IL-10 null or WT mice given ICR or transection can be used as models to evaluate the effects of specific bacteria on surgery or inflammation associated fibrosis.
Future studies are needed to fully define the time course and the mechanisms of post-surgical inflammation and fibrosis in our IL-10 null ICR model. Initial examination of CONV-IL-10 null mice at 7 and 14 days after ICR demonstrates inflammation within submucosal as well as mucosal layers in genetically susceptible IL-10 null mice but not WT early after ICR, indicating that even early post-surgical inflammation is transmural in nature and occurs only in genetically susceptible IL-10 null mice. Future studies aim to better characterize the cellular nature of inflammatory infiltrate at different times after ICR and assess if fibrosis progresses to stricture. It will also be of interest to study the consequences of ICR or ileal resection in spontaneous ileitis models such as the SAMP1/Yit [43, 44] and TNFΔARE  mice. Accumulating evidence suggests that fibrostenosing disease has a genetic component with recent links to CARD15/NOD2 or fractalkine receptor mutations [46, 47]. Evaluation of fibrosis or inflammation after ICR in mouse models carrying similar mutations could provide a useful experimental system to directly define the roles of these genes in post-surgical inflammation or fibrosis.
We would like to thank Dr. Sandra Kim for assistance and guidance in inflammation scoring, and thank Kirk McNaughton and Carolyn Suitt for histological sectioning and staining.
Support for these studies was provided by a post-doctoral fellowship from the Crohns’ and Colitis Foundation of America (R.J. Rigby), NIH (DK067395 M.A. Helmrath; DK080283 P.K. Lund and M.A. Helmrath) and a pilot feasibility study from the UNC Center for Gastrointestinal Biology and Disease (P30DK34987). This study used core facilities of the National Gnotobiotic Rodent Resource Center (P40RR018603).
The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of all authors, an exclusive license (or non-exclusive for government employees) on a worldwide basis to the BMJ Publishing Group Ltd and its Licensees to permit this article to be published in Gut editions and any other BMJPGL products to exploit all subsidiary rights, as set out in our license http://gut.bmjjournals.com/ifora/licence.pdf.