Modeling undernutrition with enteropathy in mice

Undernutrition is a global health issue leading to 1 out 5 all deaths in children under 5 years. Undernutrition is often associated with environmental enteric dysfunction (EED), a syndrome associated with increased intestinal permeability and gut inflammation. We aimed to develop a novel murine model of undernutrition with these EED features. Post-weaning mice were fed with low-protein diet (LP) alone or combined with a gastrointestinal insult trigger (indomethacin or liposaccharides). Growth, intestinal permeability and inflammation were assessed. LP diet induced stunting and wasting in post-weaning mice but did not impact gut barrier. We therefore combined LP diet with a single administration of indomethacin or liposaccharides (LPS). Indomethacin increased fecal calprotectin production while LPS did not. To amplify indomethacin effects, we investigated its repeated administration in addition to LP diet and mice exhibited stunting and wasting with intestinal hyperpermeability and gut inflammation. The combination of 3-weeks LP diet with repeated oral indomethacin administration induced wasting, stunting and gut barrier dysfunction as observed in undernourished children with EED. As noninvasive methods for investigating gut function in undernourished children are scarce, the present pre-clinical model provides an affordable tool to attempt to elucidate pathophysiological processes involved in EED and to identify novel therapeutic strategies.


Discussion
In the present study, we developed a murine model combining undernutrition with enteropathy resulting in (i) wasting and stunting, (ii) inflammation and (iii) gut hyperpermeability.
We first investigated the impact of undernutrition alone for 3 weeks by limiting the amount of calories by 50% or by feeding mice with a LP diet. Both approaches had a significant impact on growth by reducing body weight and exhibiting shorter tails. These approaches had no effect on gut barrier function or intestinal permeability. Caloric restriction-induced undernutrition models are used in the literature from 15 to 50% of caloric restriction but are not associated with gut barrier dysfunction 8 . Dietary protein restriction is used from 0 to 7% of proteins and are associated with gut barrier dysfunction, only in the case of drastic dietary protein restriction 8 . Similarly, rats fed with LP diet (4% protein) for 20 days had growth retardation without effect on colonic or ileum permeability to macromolecules 10 . Brown et al. study showed no impact of 3 weeks of LP diet (7%) in weaning mice on fecal calprotectin levels 9 while it increased intestinal permeability 9 . By exposing mice to protein malnutrition from 5 or 14 days, fecal calprotectin levels decreased while a longer exposure for 21 days had no influence on these levels. We speculated that it may result from physiological adaptations. First, mice are able to cope with protein malnutrition and they develop survival-promoting strategies with a reduced inflammatory state. Then, there is a progressive exhaustion of the adaptive mechanisms. A similar pattern was observed in malnourished mice with focal cerebra ischemia 21 . While malnutrition for 7 to 14 days induced survival-promoting mechanisms such as a neuroprotection and immunoregulation, longer exposure to malnutrition for 30 days impairs stroke outcome 21 .
As dietary protein deficiency had a higher effect on body weight compared to caloric restriction and a lower impact on animal's welfare 22 , we further chose low protein diet to reproduces undernutrition features. As dietary protein deficiency alone was not sufficient to impact gut barrier, we then investigated the combination of LP diet with a compound triggering a gut barrier insult. We thus explored two approaches: bacterial LPS and indomethacin. Bacterial LPS is the main cell wall component of gram-negative bacteria and increased anti-LPS Immunoglobulins concentrations in children are associated to poor growth outcomes 16,17 . In our experimental design, intraperitoneal injection of LPS in combination with LP diet, did not result in poorer growth than LP diet alone and did not induce EED features. LPS injection at the same dose was able to induce intestinal permeability in normo-nourished mice 23 . We thus speculated that undernutrition may impair the inflammatory response to a LPS challenge. Indeed, we observed that LP diet for 2 weeks decreased mRNA levels for cytokines. In addition, Neyestani et al. showed altered immunity response after 14 days with LP diet (2%) in weaning mice 24 . This mechanism has already been described in central inflammation 25 . Actually, LP diet during pregnancy decreased offspring inflammatory response to acute LPS in the hypothalamus 25 . Indomethacin, a non-steroidal anti-inflammatory drug has been used to induce enteropathy in experimental models 19,20 . Indomethacin challenge at 10 mg/kg in malnourished mice did not worsen body weight loss or growth faltering but increased fecal calprotectin levels compared to mice fed with LP diet. Similar results on intestinal inflammation were found in non-malnourished mice from day 1 to day 4 after a single injection of indomethacin at the same dose 26,27 . No impact of indomethacin at 10 mg/kg in gut permeability was observed in malnourished mice in the present study.
To increase the effect of indomethacin on intestinal barrier, we orally exposed mice to a single gavage of indomethacin before LP diet started and studied the impact 1 week later. No additional effects of oral indomethacin administration compared to LP diet alone were observed on growth parameters or EED features. Concerning intestinal permeability, Jacob et al. observed higher intestinal permeability 1-6 h in rats receiving indomethacin Figure 2. Growth, intestinal inflammation and permeability assessment in C57BL/6 mice fed a low protein in combination with lipopolysaccharide or indomethacin i.p. injection. (a) 3-weeks-old mice were fed with standard or isocaloric low protein diet post weaning during 14 days. At day 11, mice received a single intraperitoneal injection of LPS (1 mg/kg) or indomethacin (10 mg/kg). (b) Body weight was recorded at D1, D4, D6, D8 and every day from D11 until the end of the experimentation (n = 20 per group). Plots represent the mean ± SEM (***P < 0.001 vs. SD, Two Way ANOVA). At D14, (c) weight (n = 20 per group) and (d) tail length (n = 20 per group) were calculated. Bars indicate the mean ± SEM. One-way analysis of variance with post hoc Tukey's test was performed (***P < 0.0001).  www.nature.com/scientificreports/ gavage at 20 mg/kg and returned to normal 4 days post treatment 28 . We thus hypothesised that indomethacininduced increase in intestinal permeability may be transient.
To strengthen indomethacin impact on gut barrier, we then set up repeated gavages of indomethacin. Indeed, Whitfield-Cargile et al. demonstrated that 6 days of repeated indomethacin gavage at 5 mg/kg led to enteropathy in well-nourished mice 20 . Post-weaning mice were therefore fed with LP diet or SD diet for 3 weeks. In our murine model, much of the body weight loss and the lower tail length was driven by the low protein diet as expected 9 however, indomethacin exposure also induced a lower body weight. Body weight loss is observed in indomethacin-treated mice 29 . It may result from multiple mechanisms such as (i) intestinal damage leading to a reduced food intake, (ii) decreased cyclooxygenase-2 expression leading to decreased mucosal protection 19 , (iii) microbiota changes 29 . Xiao et al. have shown that microbiota depletion by antibiotics treatment improved body weight in indomethacin-treated mice 29 . Neither dietary exposure, nor indomethacin treatment alone were able to induce intestinal permeability but the combination of both induced intestinal hyperpermeability. Low protein diet is not sufficient to impact intestinal permeability but can alter gut barrier components such as mRNA levels for Muc2, Tff3 and Ocln. It may create a favourable environment to increase susceptibility to the negative effects of indomethacin. It has been demonstrated that infection by worms is more severe in mice fed with low protein diet 30 . In addition, a recent study demonstrated that MUC2 deficient mice are more susceptible to sepsis 31 .
This discrepancy between dietary impact on intestinal permeability and gut barrier components is already documented in the literature 32 while these terms are often used interchangeably. We have previously shown that dietary supplementation by l-Glutamine was able to restore intestinal permeability without changing occludin mRNA levels in mice with activity-based anorexia 33  In the present study, the combination of 3 weeks LP diet with chronic oral gavage of indomethacin significantly decreased Cldn2 and Ocln mRNA levels. These results are in accordance previous studies showing a reduced ileal and colonic occludin protein expression in protein-deprived rats 10,11 .
To better understand gut barrier response, we also studied Muc2 and Tff3 which contribute to epithelial protection. We observed that LP diet reduced jejunal Muc2 and Tff3 mRNA levels, suggesting a decrease in epithelium protection. Similarly, dietary protein restriction in pregnant dams alters barrier function by reducing Muc2 and Tff3 mRNA levels in colonic mucosa of offspring suggesting direct impact of undernutrition in mucus alteration 34 . By contrast, dietary high protein diet increased ileal Muc2 mRNA expression in rats 35 . The combination of LP diet with oral indomethacin at 2.5 mg/kg increased jejunal Ccl2 mRNA level compared to mice fed with SD diet as previously described by Harusato et al. in C57BL/6 normonourished mice 36 . In contrast, Il1b and Tnfa mRNA remained unchanged. In our study, LP diet did not impact villous height or crypt depth and similar results were observed after 26-days fed with LP diet (2%) or 3-weeks fed with LP diet (7%) 9,37 . In contrast, repeated indomethacin gavages induced villous blunting as previously shown 20,38 .
Throughout our four experiments, we observed that mice body weight decreased by exposing mice to protein malnutrition from 5, 14 or days, while dietary exposure to a shorter protein malnutrition for 5 days is not sufficient to impact tail length. Similarly, a longer protein malnutrition exposure is necessary to induce a vulnerable environment to enable indomethacin impact on gut barrier function. Our present murine model reflects features observed in children with undernutrition and EED. LP diet reflects the poor nutritional environment leading to undernutrition. LP diet was able to induce wasting and linear growth failure but was not sufficient alone to impact gut barrier function. By repeated oral exposure to indomethacin, a gut barrier dysfunction was induced. Higher intestinal permeability was observed as described in human EED 39 and this effect was confirmed by decreased mRNA levels encoding for tight junction proteins. Fecal calprotectin, a proposed biomarker in EED 40 , was also increased in the present model. Human EED is also characterized by inflammatory cell infiltrate 41 and we detected increased jejunal Ccl2 mRNA levels for MCP-1 suggesting this infiltration also occurred in our model.
As higher endotoxins levels have been observed in EED 15 , we initially speculated that LPS would be more relevant to use than indomethacin in order to induce EED. Through our experiments, we finally demonstrated the opposite. Indomethacin use may not reproduce identical mechanisms involved in EED etiology but enabled to display many EED features while LPS did not. Experimental models of enteropathy induced by chemical agents such as indomethacin are used in the analysis of pathological mechanisms of enteropathy 29,42 as well as the development of therapeutic agents 43 . Although NSAID-induced enteropathy models do not have the complexity of human EED, the present model can contribute to the study of the disease such as microbiota changes 29 or to the evaluation of nutritional intervention 44 . In addition, use of indomethacin provides an easy and reproducible model with a more controlled inflammatory response that may compare to subclinical symptoms found in human pathophysiology. The present model recapitulates key features of human EED such as growth failure, intestinal hyperpermeability and inflammation and is comparable to the model developed by Brown et al. using LP diet and bacterial challenge 9 . Brown's model with bacterial challenge reflects more a primary mechanism of the human disease but our model encompasses methodologies that are considered easy to induce, and its simplicity allows it to be used in several experimental protocols. While regional LP diets such as Regional Basic Diet from Northeast Brazil consider geographic differences in EED development, we chose a commercial LP diet as used in Brown's model.
In addition, human EED is a complex syndrome with multiple phenotypes depending on various adverse exposure 45 and geographic differences 46  In conclusion, we developed a murine model of undernutrition with EED features (intestinal inflammation and hyperpermeability) that may be compared to what may be observed in humans (Fig. 7), and particularly in children aged less than 5 years who become severely wasted prior to or after having accumulated significant linear growth retardation. Understanding the pathophysiological mechanisms involved during an episode of undernutrition (wasting and /or stunting) associated to EED is a critical step to develop novel therapeutic strategies. Obvious methodological limitations hamper the investigation of gut function in undernourished children, and particularly the lack of validated non-invasive methods. A stable and reproducible animal model is therefore an interesting and affordable tool to elucidate pathophysiological processes and potentially evaluate innovative therapeutic applications.

Methods
Ethics. Animal care and experimentation complied according to the European directive for the use and care of laboratory animals (2010/63/UE) and received the agreement of the local animal ethics committee (Comité National de Réflexion Ethique sur l'EXpérimentation Animale) and of the ministerial committee for animal experimentation (registration number: APAFIS#6185). Animal welfare was monitored daily by visual inspection. All interventions were done during the light cycle and mice were given paper nesting material as enrichment.
Animals and treatment regimen. The murine model was initiated at post-weaning stage to enable further studies investigating the impact of early-life nutritional interventions to limit or reverse EED development. Post weaning 3-week-old male mice C57BL/6 were ordered for each experiment (Janvier, Le Genest-Saint-Isle, France). They were housed to a cage and acclimatized for 1 week. During this period, they received standard diet ad libitum (A03 21.4% protein, 5.1% fat-SAFE, Augy, France) and had access to tap water. All experiments took place in a climate-controlled facility with 12/12 light/dark cycle and mice were randomized per cage and assigned to a specific regimen. At the end of each protocol, mice were killed by the intraperitoneal administration of a combination of lethal anesthetics (ketamine 40 mg/kg plus xylazine 1 mg/kg). Samples were stored immediately at − 80 °C. Experiment 1-effect of protein or caloric restriction. Mice were fed with either standard (SD, 21.4% protein, 5.1% fat, n = 20-SAFE A003), isocaloric low protein (LP, 5.8% protein, 6% fat, n = 20-SAFE) diet ad libitum or standard diet with 50% caloric restriction (CR) for 21 days (Table 1, Fig. 1a). CR were calculated from CT mice pellet consumption (3.5 g/day) and divided by 2 (1.75 g/day).

Experiment 3-impact of undernutrition and single gavage of indomethacin.
Mice were fed with either standard or isocaloric low protein diet for 5 days. At D1, a single gavage of indomethacin (10 mg/kg, n = 20) diluted in 1% carboxymethylcellulose (Sigma Aldrich) was performed while control groups were gavaged with vehicle (Fig. 3a).

Experiment 4-impact of undernutrition and repeated gavages of indomethacin.
Mice were fed with either standard or isocaloric low protein diet for 21 days. At D14, indomethacin (1 or 2.5 mg/kg, n = 10 for each group) diluted in dimethyl sulfoxide 20 was given while control groups were gavaged with vehicle. Gavages were performed once a day for 7 days (Fig. 4a). . Growth, intestinal inflammation and permeability assessment in C57BL/6 mice fed a low protein or isocaloric standard diet combined with chronic indomethacin gavages. (a) 3-weeks-old mice were fed with standard or low protein diet for 21 days. At day 14, indomethacin (1 or 2.5 mg/kg) gavage was performed once a day for 7 days. (b) Body weight was recorded every day from D14 until the end of the experimentation (n = 10 per group). Plots represent the mean ± SEM (***P < 0.001, Two-way ANOVA). At D21, (c) body weight (n = 9-10 per group) and (d) tail length (n = 9-10 per group) were measured. Bars indicate the mean ± SEM One-way analysis of variance with post hoc Tukey's test was performed (***P < 0.0001). Fecal calprotectin concentration. Mice feces were collected at the end of each experiment, weighted, homogenized in 500 µL PBS + 1% inhibitors (protease and phosphatase inhibitor cocktail, Sigma Aldrich) and centrifuged at 13,000g, 20 min. Supernatant was stored at − 80 °C. Fecal calprotectin measurement was performed on supernatants using a S100A8/S100A calprotectin ELISA kit following manufacturer's instructions (R&D System, Mineapolis, USA). Concentration was determined by assessing the OD 450 nm using a plate reader (Tecan, Männedorf, Suisse), comparing with a standard curve of known concentration of calprotectin. Intestinal permeability and inflammation were used as enteropathic markers. CR and LP for 3 weeks induced stunting and wasting but had no intestinal impact. We therefore decided to combine LP diet to a gastrointestinal insult trigger by liposaccharides (LPS) or indomethacin. LPS did not significantly impact small intestine while indomethacin increased fecal calprotectin production. To accentuate the effects, we investigated the effects of repeated gavages of indomethacin in addition to LP diet and mice exhibited stunting and wasting with intestinal hyperpermeability and gut inflammation. www.nature.com/scientificreports/ Intestinal permeability assessment. Jejunal permeability was assessed by measuring 4 kDa Fluorescein-isothiocyanate (FITC, Sigma Aldrich)-dextran mucosal to serosal flux level in Ussing chambers (Harvard Apparatus, Holliston, United States) as previously described 47 . Intestinal permeability was assessed by serum FITC-dextran flux concentration 9 . Briefly, mice were fasted 6 h before FITC-dextran gavage (60 mg/kg). Three hours post gavage, plasma was collected post-mortem. FITC-dextran fluorescence level was measured by using a 96-well black plate reader (Chameleon V-Hidex, Turku, Finland) read with the excitation of 485 nm and emission of 530 nm. A standard curve was used to convert values to concentration.
RT-q-PCR. First, 1.5 μg total RNA into cDNA by using 200 units of SuperScript II Reverse Transcriptase (ThermoFischer, Whaltham, Massachussets, USA) was used for reverse transcription as previously described 48 . SYBR Green technology on BioRad CFX96 real time PCR system (BioRad Laboratories, Marnes la Coquette, France) was used to perform qPCR in duplicate for each jejunal sample as previously described 48 . Gapdh (glyceraldehyde-3-phosphate dehydrogenase), B2m (Beta-2-Microglobulin) and Rn18s (18S ribosomal RNA) were used as reference genes. Sense and anti-sense primers are described in the supplementary Table 1.
Histology. Jejunum samples from experiment 4 were embedded in paraffin. Sections of 4 mm were cut with a microtome and stained with a solution of hematoxylin-eosin-saffron (HES) as previously described 49 .
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.