Low dietary fiber promotes enteric expansion of a Crohn's disease-associated pathobiont independent of obesity.

Obesity is associated with metabolic, immunological, and infectious disease comorbidities, including an increased risk of enteric infection and inflammatory bowel disease such as Crohn's disease (CD). Expansion of intestinal pathobionts such as adherent-invasive Escherichia coli (AIEC) is a common dysbiotic feature of CD, which is amplified by prior use of oral antibiotics. Although high-fat, high-sugar diets are associated with dysbiotic expansion of E. coli, it is unknown if the content of fat or another dietary component in obesogenic diets is sufficient to promote AIEC expansion. Here, we found that administration of an antibiotic combined with feeding mice an obesogenic low fiber, high sucrose, high fat diet (HFD) that is typically used in rodent obesity studies promoted AIEC intestinal expansion. Even a short-term (i.e., 1-day) pulse of HFD feeding before infection was sufficient to promote AIEC expansion, indicating that the magnitude of obesity was not the main driver of AIEC expansion. Controlled diet experiments demonstrated that neither dietary fat nor sugar were the key determinants of AIEC colonization, but that lowering dietary fiber from approximately 13% to 5-6% was sufficient to promote intestinal expansion of AIEC when combined with antibiotics in mice. When combined with antibiotics, lowering fiber promoted AIEC intestinal expansion to a similar extent as widely used HFDs in mice. However, lowering dietary fiber was sufficient to promote AIEC intestinal expansion without affecting body mass. Our results show that low dietary fiber combined with oral antibiotics are environmental factors that promote expansion of Crohn's disease-associated pathobionts in the gut.


AIEC infection 153
AIEC strain NRG857c (serotype O83:H1) was grown overnight in lysogeny broth (LB) medium 154 with shaking at 37°C. Mice were pretreated with streptomycin (2 mg/mouse) by oral gavage one 155 day prior to infection. Mice were infected by oral gavage with 0.1 ml of phosphate buffered 156 saline solution containing ~2×10 9 colony forming units (cfu) of AIEC. The infectious dose was 157 verified by plating of serial dilutions on LB agar supplemented with ampicillin and 158 chloramphenicol. Body mass was measured throughout infection. Fecal pellets were weighed, 159 homogenised (Retsch Mixer Mill) in 1 mL PBS, serially diluted, and plated onto LB agar plates 160 supplemented with ampicillin (100 µg/ml) and chloramphenicol (34 µg/ml). Intestinal tissues 161 were harvested into cold PBS at necropsy and were flushed with PBS to remove luminal 162 contents, homogenized with a sterile metal bead, and plated in the same manner as feces. 163 Plates were incubated overnight at 37°C and colonies were counted to determine cfu per gram 164 of feces or tissue. were analyzed for each sample and scored according to previously defined criteria (50). Briefly, 171 scant, moderate or dense scores were assigned for multiple variables including, but not mitted 172 to i) lumen: necrotic epithelial cells and polynuclear leukocytes, ii) surface epithelium: 173 regeneration, iii) mucosa: crypt abscesses and iv) submucosa: mononuclear and polynuclear 174 leukocytes cell infiltrate. More detailed criteria are previously defined (11). Crypt length 175 measurements and goblet cell quantification were done using ImageJ software on a Nikon

Statistical analysis 179
Data was assessed for normal distribution using the D'Agostino-Pearson normality test. For 180 non-normally distributed data sets, statistical significance was determined by Mann-Whitney U 181 test for comparison of 2 data sets. A Kruskal-Wallis test with Dunn's multiple comparison was 182 used for comparison of more than 2 non-normally distributed data sets. For normally distributed 183 data sets, a one-way or two-way ANOVA with Tukey post-hoc multiple comparison analyses 184 were used. All analyses were performed using Graph Prism 6.0 (GraphPad Software Inc. San 185 Diego, CA). A P-value of 0.05 or less was considered significant.

An obesity-causing diet in conjunction with antibiotic exposure promotes AIEC 191 expansion and gut pathology in mice 192
To study the participation of obesity or dietary factors present in an obesogenic diet in AIEC 193 dynamics in the host, we established a standard high-fat feeding regimen in conjunction with a 194 previous developed AIEC infection model (8, 50). We first confirmed that diet-induced obesity 195 worsens blood glucose control because elevated blood glucose can degrade gut barrier function 196 and worsen outcomes from enteric infection in mice (55). We found that mice fed an obesogenic 197 high fat diet (HFD) that contained 60% of calories from fat for 16 weeks had higher blood 198 glucose during a glucose tolerance test, higher fasting blood glucose, higher body mass and 199 higher adiposity compared to mice fed a standard chow diet prior to AIEC infection ( Fig. 1A-D). 200 It is noteworthy that this 60% HFD is often used to study obesity in mice and contains higher 201 sucrose and lower amount of fiber compared to the standard chow diet used in our experiments 202 and many animal facilities. Mice were given a single low dose of antibiotic (streptomycin, 2 203 mg/mouse) one day prior to infection with AIEC. HFD-fed mice maintained a higher body mass 204 compared to chow-fed mice throughout AIEC infection (Fig. 1E). HFD-fed obese mice treated 205 with antibiotics had higher levels of AIEC in the feces and throughout the intestinal track at 17 206 days post-infection. Significantly elevated AIEC colony forming units (CFUs) were found in the 207 ileum, cecum, and colon of HFD-fed mice, but there were no detectable AIEC colonies in spleen 208 homogenates of chow-fed or HFD-fed mice (Fig. 1F). We next analyzed the time course of 209 AIEC infection in antibiotic treated mice and found higher fecal burdens of AIEC in the HFD-fed 210 mice every day between 1-30 days post-infection (Fig. 1G). Antibiotics are known potentiators 211 of AIEC infection, but it was not clear if they were required for the exacerbated AIEC outgrowth 212 observed in HFD-fed mice (40). We found that antibiotic pre-treatment was required to increase 213 AIEC intestinal expansion during an obesogenic diet, since mice fed HFD for 16 weeks had 214 similar AIEC fecal burdens to chow fed mice if mice were not given antibiotics (Fig. 1H). These 215 data established a tractable model of HFD-induced obesity that promoted AIEC expansion in 216 the gut in an antibiotic-dependent manner. 217

218
We next examined intestinal pathology by scoring the severity of crypt hyperplasia, immune cell 219 infiltration, epithelial cell loss, and edema in the gut during AIEC infection. Nine days post AIEC 220 infection, the severity of pathology was higher in the ceca of HFD-fed mice compared to chow-221 fed mice ( Fig. 2A). In addition, HFD-fed mice had crypt elongation in the cecum compared to 222 chow-fed mice at 9 days post AIEC infection ( Fig. 2B-C). Seventeen days post AIEC infection, 223 the severity of pathology was higher in the colon of HFD-fed mice compared to chow-fed mice 224 ( Fig. 2D) In addition, crypt elongation was observed in the ileum, cecum, and colon at 17 days 225 post AIEC infection (Fig. 2E-F). 226 227

Diet rather than level of obesity correlates with intestinal AIEC expansion 228
Transgenic mice that are genetically susceptible to AIEC and fed a Western-style diet high in fat 229 and sugar for 12 weeks have increased fecal AIEC burdens 3 days after infection (1). In order to understand the role of obesity following long-term HFD feeding on AIEC colonization in the gut, 231 we performed feeding studies with a second obesogenic diet in which 45% of calories were 232 derived from fat compared to our previous results using a HFD with 60% of calories from fat. As 233 expected, feeding mice for 16 weeks with a 45% HFD resulted in significantly higher body mass 234 compared to the chow-fed control mice during AIEC infection (Fig. 3A). Our data also confirm 235 the well-established result that mice fed a 60% HFD had a greater increase in body mass 236 compared to mice fed a 45% HFD, an effect that was maintained during AIEC infection. Mice 237 fed a 45% HFD also had elevated fecal burdens of AIEC beginning on day 1 after infection and 238 throughout the 30-day observation period (Fig. 3B). It is well established that mice display 239 interindividual variability in the level of obesity caused by either obesogenic diet (63). This 240 heterogeneity in body mass in mice allowed us to compare AIEC fecal burdens of mice with 241 different levels of obesity. Hence, we tested AIEC fecal burdens at specific time points post 242 infection to determine if body mass correlated with fecal AIEC burden during diet-induced 243 obesity. We found no correlation between body mass and AIEC fecal burden in HFD fed mice at 244 days 8, 13, and 17-days post AIEC infection ( Fig. 3C-E). These data suggested that an aspect 245 of diet composition rather than the magnitude of host obesity was sufficient to promote intestinal 246 AIEC expansion. However, it was unclear if fat content or another dietary component of 247 obesogenic diets was contributing to the expansion of AIEC. 248 249

Dietary components ingested during infection promote intestinal AIEC expansion 250 independent of obesity 251
To parse out the effects of overt obesity from diet, we used short-term HFD feeding in mice that 252 were switched from a standard chow diet to a 60% HFD one day prior to AIEC infection and 253 remained on either the HFD (or chow diet) throughout the course of infection. There were no 254 initial differences in body weight at the onset of HFD feeding, however a slightly higher (i.e., 2-3 255 gram increase) body mass was observed in short-term HFD-fed mice compared to chow fed mice during AIEC infection (Fig. 4A). Fecal burdens of AIEC were significantly higher in the 257 HFD-fed group between days 3-9 post infection (Fig. 4B). At day 9, the differences in organ 258 burdens were only observed in the cecum and proximal colon, where HFD mice had higher 259 AIEC burdens (Fig. 4C). These results indicate that the constituents of the HFD may be a major 260 factor responsible for AIEC expansion, independent of overt obesity. 261

262
If a dietary component of HFD was the factor responsible for AIEC expansion, rather than the 263 comorbid effects of obesity, then feeding mice a short pulse of HFD would be expected to 264 modulate AIEC loads. To test this, we compared AIEC burdens in mice fed a HFD for 16 weeks 265 (i.e. long-term) or for one day (i.e. pulse) before AIEC infection. We also compared these HFD-266 fed mice to mice fed a chow diet for equal periods of time. Consistent with our previous results, 267 mice on the long-term HFD feeding regimen had significantly higher fecal AIEC burdens 268 compared to chow-fed controls (Fig. 5A). Mice exposed to a HFD pulse also had significantly 269 higher AIEC fecal burdens than chow-fed control mice from day 3-9 post-infection (Fig. 5A). 270 Importantly, the body mass of short term HFD-fed mice was significantly lower compared to 271 mice fed a chow-diet in the long-term (Fig 5B). Thus, despite lower body mass, mice fed a HFD 272 pulse still had significantly higher AIEC fecal burdens compared to chow-fed control mice in the 273 long-term cohort. These data indicate that factors such as diet can regulate AIEC infectious 274 burden independent of changes in body mass leading to obesity. Indeed, only the long-term 275 HFD-fed mice had significantly higher adiposity compared to all other groups of mice (Fig. 5C). 276 Overall, these data indicate that diet composition, independent of body mass or adiposity, plays 277 a role in the regulation of AIEC infection. 278

Low dietary fiber promotes intestinal AIEC expansion 280
In addition to higher fat, obesogenic diets also contain higher sucrose and lower fiber content, establish which component of HFD was promoting AIEC expansion, we established two defined 283 diet regimens that were controlled for ingredient composition but where the majority of the fat 284 and sucrose content was substituted with corn starch and maltodextrin. First, we compared the 285 60% HFD to a sucrose and fiber matched diet that has the same (high) sucrose and same (low) 286 fiber content, but lower fat compared to the obesogenic HFD. The goal was to test the impact 287 that fat content has on the AIEC infection, hence we compared the HFD to a low-fat high 288 sucrose, low fiber-matched diet (HSLF) using both the short-term and long-term feeding 289 regimens with this control diet. 290 291 From short-term feeding experiments, we first isolated dietary fat content as a composition 292 variable by comparing mice fed high (HFD) or low fat-containing diets (HSLF) that were 293 equivalent for fiber and sucrose content. Following short-term feeding and AIEC infection, both 294 HFD-fed and HSLF-fed mice had similar levels of AIEC in their feces for the 15-day observation 295 period ( Fig. 6A). HFD-fed mice had higher body mass starting at day 8 post AIEC infection 296 compared to HSLF-fed mice (Fig 6B). The HFD mice also had significantly higher adiposity 297 compared to the HSLF-fed mice (16.6% versus 8.5%, respectively; Fig. 6C). The magnitude of 298 increase in adiposity in short-term HFD feeding compared to a HSLF diet was similar to our 299 results using short-term feeding of HFD compared to a chow diet (Fig. 5C). On day 17 post-300 infection, there were also no differences in tissue associated AIEC burdens in any section of the 301 intestinal tract after short-term HFD feeding compared to a HSLF diet (Fig. 6D). We also 302 performed long-term feeding studies for 16 weeks using HFD or HSLF diets and found no 303 significant differences in AIEC fecal burden at any time during the infection (Fig. 6E), despite 304 increased body mass in HFD-mice throughout infection (Fig. 6F). Similarly, 16 weeks of feeding 305 HFD or HSLF diets did not change tissue burden or dissemination were observed at day 17 post 306 infection in long-term HFD-and HSLF-fed mice (Fig. 6G). These data strongly suggested that We next examined the effect of fiber as a dietary variable in obesogenic diets on AIEC 310 colonization. We used a no-sucrose, low-fat (NSLF) diet that contains the same amount of fiber 311 as a 60% HFD but has low fat and no sucrose. Diets low in fiber have been associated with 312 decreased levels of butyrate as well as loss of mucosal barrier integrity (16, 59). Diets higher in 313 fiber have shown to be protective from DSS-induced colitis and mitigate aspects of metabolic 314 disease (24, 32, 64). However, high fiber diets can potentiate enterohemorrhagic E. coli 315 infections concomitant with lower commensal E. coli abundance (65). To assess the influence of 316 low dietary fiber, we used the low fat, no sucrose, low fiber diet (NSLF) compared to a chow diet 317 in our short-term feeding model. We found elevated fecal burdens of AIEC from day 1-15 post 318 infection in the NSLF-fed mice compared to chow-fed mice ( Fig. 7A) despite no differences in 319 body mass (Fig. 7B). We also observed higher tissue burdens in the distal ileum, cecum, and 320 proximal colon at day 17 post infection (Fig. 7C). These data indicate that ingestion of lower 321 dietary fiber is sufficient to promote expansion of AIEC throughout the gut. 322

323
Finally, we tested if supplementation of a HFD with a specific fiber could alter AIEC colonization 324 by comparing a 60% HFD and 60% HFD containing an additional 150 g/kg of cellulose. We 325 found no change in fecal AIEC burden at nearly all days post AIEC infection during short-term 326 feeding of a HFD compared to a HFD containing extra cellulose, despite mice having lower 327 body mass between 9-24 days post-AIEC infection when fed the HFD containing extra cellulose 328 ( Fig. 8 A-B). In fact, the diet supplemented with cellulose transiently increased fecal AIEC 329 burden 9 days post infection (Fig. 8A). Further, mice had a small increase fecal AIEC burden 3- We examined how pre-existing obesity versus individual diet components altered enteric 338 pathobiont expansion in mice. We first developed a "two-hit" model of environmental factors that 339 altered AIEC infection. We used pre-treatment with a single antibiotic plus ingestion of an 340 obesogenic diet and found that this combination exacerbated AIEC burdens in the intestine and 341 feces after enteric infection. The interaction of two environmental stressors is consistent with 342 recent work showing that antibiotic use combined with ingestion of a high fat Western-style diet 343 worsen signs of early IBD in humans and mice (28). We found that an obesogenic HFD 344 worsened markers of intestinal pathology during AIEC infection. Although we cannot determine 345 directionality between increased AIEC burden and worsened intestinal pathology during HFD 346 feeding, our results are consistent with obesogenic diets promoting both host and microbial 347 indicators of IBD pathology and risk. We then questioned whether pre-existing obesity at the 348 time of enteric infection or ingestion of Western-style obesogenic diet during infection was 349 sufficient to influence AIEC expansion in the intestine. It is well established that long-term HFD 350 feeding (for over 16 weeks) causes obesity in mice and we found that protracted HFD feeding 351 also promoted AIEC colonization and worsened pathology in mice. We also found that short-352 term HFD initiated just before or just after AIEC infection was sufficient to increase AIEC 353 intestinal expansion even in the absence of overt obesity. 354 355 HFD feeding influences many factors beyond obesity including, metabolic inflammation, poor 356 blood glucose control, and insulin resistance (13, 29). HFD feeding also changes the 357 composition of the commensal intestinal microbiota (4, 36). In fact, as early as one day of HFD-358 feeding can transiently alter the composition of gut microbial populations (19, 47, 48). Our results showing that AIEC expansion induced by HFD-feeding only occurs when combined with 360 the use of antibiotics implies a role for the commensal microbiota as a mediator in the 361 interaction between these two environmental factors. Although, we did not define the precise 362 role of the commensal microbiota, others have found that antibiotics plus a HFD combine to 363 impair mitochondrial metabolism in the intestinal epithelium, and promote an inflammatory gut 364 environment that allows expansion of Enterobacterales (28). It is logical for future studies to 365 investigate if a similar mechanism opens an intestinal niche for AIEC colonization, expansion or 366 hinders host control of this pathobiont. study showed that maltodextrin, irrespective of chain length, promoted AIEC growth and biofilm 405 formation (39). This study also found that sucrose did not confer a growth advantage to AIEC, 406 which supports our hypothesis that the low fiber content, not high sucrose in the diet can 407 promote AIEC growth. 408

409
The types of fibre contributing to AIEC expansion remain unclear. We only tested a single type 410 of insoluble fiber and found that increased dietary cellulose only caused a transient increase or decrease in AIEC burden, but cellulose supplementation was not sufficient to alter the 412 magnitude of AIEC burden consistently during obesogenic HFD feeding. Cellulose 413 supplementation did lower body mass in mice, which is consistent with all the other results 414 showing that changes in body mass do not necessarily coincide with changes in AIEC burden. 415 Soluble fibers can be fermented by the microbiota into SCFAs, while insoluble fibers provide 416 bulk for waste disposal (48, 62). HFD supplemented with inulin increases mucosal immunity and 417 increases intestinal microbial diversity compared to HFD supplemented with cellulose (64).