Differential expression of soluble receptor for advanced glycation end-products (sRAGE) in mice susceptible or resistant to chronic colitis

Aims Identifying the factors that contribute to chronicity in inflamed colitic tissue is not trivial. However, in mouse models of colitis, we can investigate at preclinical timepoints. We sought to validate murine Trichuris muris infection as a model for identification of factors that promote development of chronic colitis. Methods We compared preclinical changes in mice with a resolving immune response to T. muris (resistant) versus mice that fail to expel the worms and develop chronic colitis (susceptible). Findings were then validated in healthy controls and patients with suspected or confirmed IBD. Results The Receptor for Advanced Glycation End Products (Rage) was highly dysregulated between resistant and susceptible mice prior to the onset of any pathological signs. Increased soluble RAGE (sRAGE) in the serum and faeces of resistant mice correlated with reduced colitis scores. Mouse model findings were validated in a preliminary clinical study: faecal sRAGE was differentially expressed in patients with active IBD compared with IBD in remission, patients with IBD excluded or healthy controls. Conclusion Pre-clinical changes in mouse models can identify early pathways in the development of chronic inflammation that human studies cannot. We identified the decoy receptor sRAGE as a potential mechanism for protection against chronic inflammation in colitis in mice and humans. We propose that the RAGE pathway is clinically relevant in the onset of chronic colitis, and that further study of sRAGE in IBD may provide a novel diagnostic and therapeutic target.


Introduction 47
Inflammatory bowel diseases (IBD) are a group of intestinal immune disorders, including 48 Crohn's disease (CD) and ulcerative colitis (UC), that cause chronic inflammation in the gut 49 [1]. The cause of IBD is currently not known, but dysregulation of intestinal immunity, 50 microbial dysbiosis, genetics and environmental factors contribute to disease onset. 51 Unpredictable cycles of remission and relapse require careful monitoring and the long-term 52 damage from inflammation often warrants potent immunomodulatory therapy or surgical 53 intervention [2]. 54 It is impossible to reliably predict onset, relapse or remission of IBD [3] and currently, only 55 animal models provide a means of studying the perturbations in the gut that precede colitis. 56 Infecting susceptible mouse strains with the enteric nematode parasite Trichuris muris 57 closely parallels human Crohn's disease in both the pathological and transcriptional changes 58 Exploiting these early differences in the host immune response to T. muris infection 68 experimentally may provide information on the factors that promote the onset of chronic, 69 rather than resolving, inflammation in the gut [7]. Early factors are impossible to distinguish from the inflammatory milieu present in IBD patients at the point of diagnosis as chronic 71 inflammation is already well established. Identification and validation of early changes 72 during chronic colitis onset in mice could provide a useful pipeline for developing diagnostic 73 and disease-management biomarkers or therapeutic targets in human colitis. 74 In this study, we carried out a T. muris infection study, investigating preclinical 75 transcriptional changes 24 hours post infection (PI). We identified the receptor for advanced 76 glycation end-products (Rage) as highly upregulated in mice susceptible to T. muris 77 infection. We further investigated the presence of RAGE  parasite stocks were carried as described by Wakelin, 1967 [8]. Experimental mice were 95 infected with 200 embryonated eggs in 200 μl of ultra-pure distilled water via oral gavage. 96 Worm burden was assessed at day 21 PI. Caecum and proximal colon were harvested at 97 autopsy to determine parasite clearance of each mouse at the end of each experiment as 98 described by Else et al., 1990 [9]. 99 in remission at the of time testing (10 ulcerative colitis and 9 Crohn's disease) and 6 patients 112 had active IBD (n=5 CD, n=1 UC) at the time of testing. 113

Statistics and analysis 114
Where statistics are quoted, experimental groups were compared using linear regression, 115 Mann-Whitney U test or two-way analysis of variance (ANOVA) test followed by Sidak's post 116 hoc multiple comparisons test, where appropriate. P values <0.05 were considered 117 significant. Data are presented as mean ± SEM unless otherwise stated. Statistical analyses 118 were carried out using GraphPad Prism 7 (GraphPad Software, La Jolla, California, USA; 119 www.graphpad.com). 120

Early immune response informs resistance to T. muris-induced colitis 122
Following challenge with Trichuris muris, as expected BALB/c mice expelled most or all of 123 the worms by 21 days PI, whereas AKR mice were unable to expel all worms and remained 124 infected with a significantly higher worm burden (P=0.016, Mann-Whitney U test) ( Figure  125 1A). Colitis scoring revealed increased histological changes associated with inflammation in 126 both AKR and BALB/c mice after infection ( Figure 1B). These changes included influx of 127 immune cells, presence of immune cells in the submucosa, crypt hyperplasia and goblet cell 128 loss. In agreement with previous data, the colitis scores in BALB/c mice peaked at 21 days PI 129 and had begun to return to normal by 31 days PI. As expected, colitis scores in AKR mice 130 rose after infection and peaked at 31 days PI where the colitis score was significantly greater 131 than that of the BALB/c mice (P=0.046, ANOVA; Figure 1B). Representative images of 132 haematoxylin and eosin stained proximal colon sections in naïve mice and at 31 days PI are 133 shown in Figure 1C Faecal sRAGE in the BALB/c mice increases relative to colitis scores (R 2 =0.99). However, 209 sRAGE levels in susceptible mice did not change during the course of infection relative to 210 increasing colitis scores from D0 to D21 post-infection (serum R 2 =0.99, faecal R 2 =0.73). 211 We then investigated levels of the RAGE ligand S100A8 (one part of the heterodimeric 212 calprotectin protein, currently used as a clinical biomarker for IBD) in serum and faeces as 213 an indicator of whether sRAGE might be quenching the proinflammatory effects of 214 circulating RAGE ligands by acting as a decoy receptor. Serum and faecal S100A8 did increase slightly during the course of infection in both AKR and BALB/c mice, but no 216 statistical differences were observed between the two strains and there was high variability 217 between mice. At 21 days PI BALB/c mice had greater levels of serum S100A8 than AKR mice 218 (not significant, ANOVA; Supplementary data). Faecal S100A8 remained broadly similar in 219 naïve and T. muris infected mice of both AKR and BALB/c strains. As with serum S100A8, 220 faecal S100A8 was slightly raised in BALB/c mice at 21 days PI compared to AKR mice but 221 this was not significant (ANOVA; (Supplementary data)). In addition to being highly variable, 222 S100A8 correlated poorly with colitis scores (Supplementary data). 223

sRAGE is differentially expressed in IBD 224
As the differences in sRAGE were most apparent and consistent in faeces we focused on 225 analysis of faecal specimens from healthy volunteers and patients with IBD or suspected 226 IBD. sRAGE was not detected in the faecal samples of healthy volunteers. In contrast, s-227 RAGE was detectable in the patient cohort ( Figure 4A). The highest levels of sRAGE were 228 seen in patients with IBD excluded (largely IBS ascribed) and IBD in remission, although 229 remission patients were more variable. Patients with active IBD characterised by severe 230 inflammation had low levels of sRAGE and increased calprotectin. 231 We then transformed the faecal RAGE and calprotectin ELISA data into present (1) or absent 232 (0), where protein is scored as present if ≥ 3*SD above baseline. When patient data was 233 stratified into groups (active IBD, IBD in remission, IBD excluded (IBS) and healthy controls) 234 the ratio of RAGE to calprotectin clearly identified healthy controls (0, 0) and active IBD (0, 235 1) from IBS and IBD in remission ( Figure 4B). In healthy individuals, we saw no sRAGE or 236 calprotectin signal in any subject. In active IBD we had a consistent pattern of Calprotectin 237 present, but no sRAGE. For patients whose disease is resolving we saw a more complicated picture, where we had a subset of patients undergoing routine test that were expressing 239 both sRAGE and Calprotectin. 240

Discussion 241
Animal models are crucial for examining the causative events that lead to diseases such as 242 chronic colitis. Genetic factors influence the likelihood of developing colitis, but it is 243 impossible to continually monitor individuals with potential genetically susceptibility in 244 order to identify the pathways that drive the development of chronic intestinal 245 inflammation. Models that accurately simulate preclinical changes in the gut allow us to 246 interrogate the pathways leading to chronic colitis, prior to the development of the complex 247 inflammatory environment when disease is established. Levison et al. [14]  detected in patient's faecal samples. Akin to the mouse data we saw lower sRAGE in 263 patients with active chronic inflammation. Surprisingly, symptomatic patients with IBD excluded on a basis of normal FCP and/or colonoscopy had higher levels of sRAGE than 265 healthy volunteers. Largely, IBS was clinically ascribed but that was not based on formal 266 diagnostic criteria, and other diagnoses such as bile acid diarrhoea or microscopic colitis 267 were not formally excluded. The sample size was small and more prospective studies are 268 needed to confirm this preliminary observation. Patients whose IBD was reported to be in 269 remission had variable levels of sRAGE. It is tempting to speculate that lower levels of sRAGE 270 are associated with a risk of subsequent flare of inflammation but as we only had single 271 samples from the patients, we cannot assess this, however it would be interesting to track 272 differences in DC migration as early as day 1 PI, we saw no differences between cell surface 299 RAGE expression between AKR and BALB/c mice that might account for the altered 300 dynamics of DC recruitment. Similarly, RAGE has been linked to neutrophil recruitment but 301 we only observed modest neutrophil infiltration in the first 24 hours PI and no difference 302 between colitis-susceptible or colitis-resistant mice. Despite minimal differences in immune 303 cell presence during the early stages of T. muris infection, prolonged activation of RAGE 304 results in activation of inflammatory signalling molecules including NF-κB and MAP kinases 305 [16]. Consequently, an environment where RAGE ligands such as HMGB1, and S100 proteins 306 are continually present results in perpetual NF-κB activation and subsequent chronic 307 inflammatory conditions. 308 We observed striking differences in the levels of faecal and systemic sRAGE between colitis-309 resistant and susceptible mice, with BALB/c mice rapidly producing sRAGE in response to T.