Targeting gut microbiota and barrier function with prebiotics to alleviate autoimmune manifestations in NOD mice
Adopting a diet containing indigestible fibre compounds such as prebiotics to fuel advantageous bacteria has proven beneficial for alleviating inflammation. The effect of the microbial changes on autoimmunity, however, remains unknown. We studied the effects of prebiotic xylooligosaccharides (XOS) on pancreatic islet and salivary gland inflammation in NOD mice and tested whether these were mediated by the gut microbiota.
Mother and offspring mice were fed an XOS-supplemented diet until diabetes onset or weaning and were compared with a control-fed group. Diabetes incidence was monitored, insulitis and sialadenitis were scored in histological sections from adult mice, and several metabolic and immune variables were analysed in mice before the development of diabetes. Gut barrier function was assessed using an in vivo FITC-dextran permeability test. The importance of XOS-mediated gut microbial changes were evaluated in antibiotic-treated mice fed either XOS or control diet or given a faecal microbiota transplant from test animals.
Diabetes onset was delayed in the XOS-fed mice, which also had fewer cellular infiltrations in their pancreatic islets and salivary glands. Interestingly, insulitis was most reduced in the XOS-fed groups when the mice were also treated with an antibiotic cocktail. There was no difference in sialadenitis between the dietary groups treated with antibiotics; the mice were protected by microbiota depletion regardless of diet. Faecal microbiota transplantation was not able to transfer protection. No major differences in glucose–insulin regulation, glucagon-like peptide-1, or short-chain fatty acid production were related to the XOS diet. The XOS diet did, however, reduce gut permeability markers in the small and large intestine. This was accompanied by a more anti-inflammatory environment locally and systemically, dominated by a shift from M1 to M2 macrophages, a higher abundance of activated regulatory T cells, and lower levels of induction of natural killer T cells and cytotoxic T cells.
Prebiotic XOS have microbiota-dependent effects on salivary gland inflammation and microbiota-independent effects on pancreatic islet pathology that are accompanied by an improved gut barrier that seems able to heighten control of intestinal diabetogenic antigens that have the potential to penetrate the mucosa to activate autoreactive immune responses.
KeywordsGut barrier Gut microbiota Insulitis Mucosal immunology Prebiotics Sialadenitis XOS
Natural killer (T)
Principal coordinates analysis
Real-time quantitative PCR
Short-chain fatty acids
Analysis of composition of microbiomes
Permutational multivariate ANOVA
The gut microbiota has become a field of extensive research, and increasing evidence demonstrates the extent to which the bacteria colonising our gut contribute to the development of immune-mediated and metabolic disorders [1, 2]. Means of effective alleviation of dysbiosis are much needed to help prevent and understand such diseases; xylooligosaccharides (XOS) are interesting in this regard. XOS pass undigested through the gastrointestinal tract until they reach the ascending colon, where the saccharide molecules selectively propagate residing gut bacteria, including bifidobacteria and lactobacilli [3, 4]. Both bifidobacteria and lactobacilli have, in rodent models, been used successfully to induce an anti-inflammatory response and reduce the incidence of diabetes [5, 6].
Antibiotic treatments have also been shown to modify the incidence of type 1 diabetes in NOD mice and bio-breeding diabetes-prone (BB-DP) rats. Depending on age, the spectrum and combination of antibiotics used and the microbes that remain in the gut, antibiotic treatments have been shown to both reduce and accelerate diabetes onset [7, 8, 9, 10, 11]. Thus, it is clear that there are both protective and diabetogenic bacteria regulating autoimmune diabetes. Besides modulating the composition of the gut microbiota, dietary fibre is fermentable in the large intestine and forms an important source of short-chain fatty acids (SCFA) with a nutritional role for the epithelial cells lining the gut barrier . Some polysaccharides, such as pectin and xylan, are associated with higher levels of proinflammatory and stress-related transcripts in the colon and have diabetogenic potential in NOD mice , whereas human milk oligosaccharides, for example, have a tolerogenic effect on the immune system and reduce diabetes incidence in NOD mice . Toivonen et al  explain that differences in fibre digestion result in different effects on the microbiota and variation in outcome for the host. The role of the gut microbiota has not been addressed in these studies, however. XOS also shows immunomodulatory effects that seem related to microbial regulation [3, 14]. Because of the proposed anti-inflammatory effects, we wanted to investigate whether XOS would protect against autoimmune destruction of beta cells and salivary glands and whether the protection was dependent on gut microbial changes.
The Animal Experiments Inspectorate under the Ministry of Environment and Food approved the study according to the principles of EU Directive 2010/63/EU and the Danish Animal Experimentation Act (number 474 15/05/2014). The studies were performed in accordance with the Guide for the Care and Use of Laboratory Animals of the National Research Council.
Dams of the strain NOD/MrkTac from Taconic (Germantown, NY, USA) were housed at the barrier-protected animal facility, Faculty of Health and Medical Sciences, University of Copenhagen, under standard conditions and in open cages without filter lids. Housing and health monitoring was conducted according to the Federation of European Laboratory Animal Science Associations (FELASA) guidelines . Mice were fed, ad libitum, a modified 1430 Altromin natural diet (Lage, Germany) supplemented with either XOS (Shandong Longlive, Qingdao, China) or standard corn starch as control to balance the level of polysaccharides in the diets in amounts equal to 10% of the carbohydrate fraction. The experiments were divided into four study set ups, as outlined in electronic supplementary material (ESM) Methods and ESM Fig. 1.
Insulitis and sialadenitis were scored in a blinded fashion on at least 50 islets per pancreas and in 8 sequential sections of each salivary gland from 13-week-old mice on H&E-stained 5 μm sections with 40 μm intervals; they were graded as previously described .
Gut microbiota composition
Faeces were sampled from 4-week-old untreated and antibiotic-treated female NOD mice on the control or XOS-supplemented diet as well as 7-week-old NOD mouse recipients of faecal microbiota transplants from control and XOS-fed NOD mice. DNA extraction, tag-encoded 16S rRNA gene amplicon MiSeq-based sequencing and sequencing analysis are described in detail in ESM Methods as well as the real-time quantitative PCR (qPCR) of bacterial DNA from faeces and the mesenteric lymph node.
For OGTT, 8-week-old mice were fasted for 6 h prior to oral gavage of glucose, 75 mg/mouse (Amgros, Copenhagen, Denmark); blood glucose was measured at 0, 15, 30, 60, 90 and 120 min post glucose challenge. Blood glucose was measured from the lateral tail vein using a glucometer (Abbott Diabetes Care, Copenhagen, Denmark).
Serum from 8-week-old mice fasted for 6 h was analysed for circulating insulin and leptin levels using the Mouse Metabolic Kit (Meso-Scale Discovery, Rockville, MD, USA) according to the manufacturer’s instructions. Samples were read using the Sector Imager 6000 (Meso-Scale Discovery).
Colon content was sampled from 13-week-old mice and stored on dry ice at −80°C until analysis. SCFA concentration was measured by gas chromatography to assess bacterial fermentation and metabolic activity of the gut microbiota, as previously described .
Glucagon-like peptide-1 analysis
Serum samples were collected from 13-week-old mice to analyse the total level of glucagon-like peptide-1 (GLP-1), capturing both intact GLP-1(7–37) and truncated GLP-1(9–37), using a luminescence oxygen-channelling immunoassay (LOCI) [18, 19]. The specific assay run had a lower limit of quantification of 30 pmol/l and an upper limit of 2000 pmol/l.
An in vivo intestinal permeability assay was made after 7-week-old mice were fasted and water deprived for 4 h and subsequently dosed by oral gavage with 600 mg/kg FITC-dextran (4 kDa; Sigma-Aldrich, St Louis, MO, USA). After 2 h, mice were anaesthetised with fentanyl/fluanisone (Hypnorm, VetaPharma, Leeds, UK) and midazolam (Roche, Hvidovre, Denmark) for retro-orbital blood sampling. The FITC-dextran plasma concentration was analysed in triplicate using a spectrophotometer Spektra MAX Gemini XS (Molecular Devices, Sunnyvale, CA, USA) with excitation λ 485 nm and emission λ 535 nm.
Zonulin concentration in serum collected from 13-week-old mice was measured with Zonulin BioAssay ELISA Kit (mouse) (United States Biological, Salem, MA, USA) according to the manufacturer’s description.
Gene expression by quantitative PCR
Immediately after the 13-week-old mice were killed, a 1 cm fragment of colon and ileum was placed in RNAlater (Ambion, Austin, TX, USA). Homogenisation, RNA isolation with MagMAX-96 RNA Isolation Kit (Ambion) and cDNA synthesis using the High-Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems, Foster City, CA, USA) were performed as described previously . Muc1, Muc2, Ocln and Tjp1 were used to assess gut barrier and Arg1, Tgfβ (also known as Tgfb1), Gzmb, Fasl and Cd8a TaqMan gene expression assays (Applied Biosystem) for macrophage  and regulatory and cytotoxic T cell-related genes were used for qPCR analysis. Data were analysed as described previously . Statistical comparison was done on ΔCt values.
Mice were killed at 13 weeks old and the organs immediately placed on ice. Antibody cocktails were used for staining natural killer T (NKT) cells, cytotoxic T cells, macrophages and regulatory T cells with CD3-FITC, NK/NKT-PE, CD8α-APC, CD11c-FITC, F4.80-PerCP-CY5.5, CD4-PerCP-CY5.5, forkhead box P3 (FOXP3)-PE and CD69-APC (eBioscience, San Diego, CA, USA) in single-cell suspensions prepared from mesenteric and pancreatic lymph nodes and spleens as previously described . Analysis was performed using an Accuri C6 flow cytometer (Accuri Cytometers, Ann Arbor, MI, USA), with the gating strategy as outlined in ESM Fig. 2.
GraphPad Prism version 6.03 (GraphPad Software, San Diego, CA, USA) was used for statistical analysis; p values <0.05 were considered significant and p values <0.1 were considered tendencies. Cumulative diabetes incidence curves were calculated using the Kaplan–Meier estimation, while statistical significance between groups was evaluated by the logrank test. For all other data comparing the two groups of mice on a control or XOS diet throughout life, differences were estimated by an unpaired two-tailed Student’s t test, or Mann–Whitney U test if the data did not assume Gaussian distribution. Welch’s correction was included in the t test if variances were unequal by the F test. Unless stated otherwise in the figure legends, all data comparing the four groups of mice fed the control or XOS diet until weaning with or without antibiotic treatment were compared by two-way ANOVA with Sidak’s corrected multiple comparisons test to compare (factor 1) the two dietary groups within (factor 2) either the untreated or the antibiotic-treated group. For microbiota analysis, significant clusters in principal coordinates analysis (PCoA) were estimated with permutational multivariate ANOVA (PERMANOVA), and the differences in taxa abundance and prevalence between the groups were estimated with the statistical framework analysis of composition of microbiomes (ANCOM).
Dietary XOS regulates autoimmune manifestations in NOD mice
Effects of diets on islet and salivary gland inflammation in antibiotic-treated mice
In addition to the effect XOS had on pancreatic and salivary gland inflammation when fed throughout life, a tendency was also observed for reduced insulitis scores in 13-week-old mice fed an XOS diet only during pregnancy and lactation compared with mice on a continuous control diet (Fig. 1e,f). To test whether microbes were necessary for the observed effects, additional groups of pregnant dams on either the control or XOS diet were given a full antibiotic cocktail in the water to deplete the majority of the gut microbiota until all pups were weaned to a control diet. Antibiotic-treated mice on a control diet developed insulitis fully to the same degree as untreated mice (Fig. 1e,f). Interestingly, the slight reduction in insulitis in mice fed an XOS-supplemented diet only until weaning was even greater in mice that were also given antibiotics (Fig. 1e,f). Male NOD mice developed less insulitis than the female mice on antibiotics, but even less when fed an XOS-supplemented diet (Fig. 1g–h). Microbes did not seem necessary for XOS to reduce insulitis, the effect of which, unexpectedly, was more pronounced when microbes were absent.
When the degree of sialadenitis was evaluated in the same groups of mice, we found that an XOS diet early in life was sufficient to reduce the sialadenitis score later in life (Fig. 1d). Both the control- and the XOS-fed mice treated with antibiotics presented with the same low degree of sialadenitis as the untreated XOS-fed mice (Fig. 1d). Hence, no synergistic protective effect was observed for the XOS-fed antibiotic-treated mice as was evident for the development of insulitis.
Antibiotics deplete the gut microbiota
The antibiotic cocktail had no effect on the weight of the mice or the amount of water they drank before weaning (ESM Fig. 4a,b). The remaining microbiome of the antibiotic-treated mice was analysed to investigate whether resistant species might remain in the gut and act synergistically with the XOS diet to protect the pancreas. Targeting the 16S rRNA gene (V3 region) with qPCR demonstrated that the overall number of 16S copies was reduced more than 4000 fold by the antibiotic cocktail (Fig. 2j). Amplicon sequencing indicated that the few remaining bacteria in dietary groups consisted primarily of Streptophyta spp., which reached 98% (±1%) of the amplicon abundance (Fig. 2f). Streptophyta spp. have not previously been associated with any immune or disease function. It was furthermore shown that solely one genus at very low abundance, Flavobacterium spp., was significantly elevated in XOS-fed (0.1% of the remaining microbiome) compared with control-fed (0%) antibiotic-treated mice (Fig. 2g). This single difference in a low-abundance taxon was the main driver for the separate clustering between the two dietary groups in the unweighted PCoA plot: p < 0.05, R2 = 0.086 (Fig. 2c). Hence, no difference in the weighted PCoA plot could be observed: p = 0.17, R2 = 0.17 (Fig. 2d) and, thus, collectively, the data strongly suggest that XOS had direct microbiota-independent effects on insulitis.
Dietary impact on metabolism and glucose tolerance
Having a different diet, one based on more long-chained carbohydrates instead of monomer sugar molecules, would naturally be expected to manifest as differences in the host metabolism. Prebiotics have been shown to improve glucose homeostasis and insulin sensitivity by SCFA-mediated activation of intestinal gluconeogenesis . We found no differences in weight gain or food intake based on the nitrogen-free extract (NFE) of the diets (ESM Fig. 4c,d), and XOS supplementation had no effect on fasting blood glucose levels, serum leptin and insulin, insulin resistance or GLP-1 (ESM Fig. 5a-f), even though fermentation of XOS slightly increased the production of acetic acid in the colon (ESM Fig. 5g). There was no effect of XOS on n-butyric acid or propionic acid (ESM Fig. 5h,i). From these findings, we concluded that the effects of XOS on the host were not mediated through any major shifts in metabolism. A gluten-free diet has been shown to have favourable effects in NOD mice [22, 23], but the effect of gluten seems, in contrast to the XOS diet, to be mediated by direct regulation of beta cell activity [24, 25]. We therefore suspected that a gluten-free diet in combination with XOS could provide further protection. However, the incidence of diabetes remained low, at around 20%, on a gluten-free diet regardless of XOS supplementation (ESM Fig. 5j).
Prebiotic improvement of the intestinal barrier function and mucosal inflammation
Microbiota-mediated dietary changes in mucosal inflammation
Here, we demonstrate that the prebiotic compound XOS can change the gut microbiota and delay the onset of type 1 diabetes in NOD mice and reduce autoimmune cellular reactions against pancreatic beta cells and salivary glands. This is in line with previous results demonstrating that a variety of intestinal bacteria can regulate the degree of both insulitis and sialadenitis in NOD mice [16, 28, 29]. The high degree of insulitis in antibiotic-treated mice resembles that observed in germ-free NOD mice . The effect is similar to that seen in a previous study using the same broad-spectrum antibiotic cocktail during pregnancy to eradicate the microbiota of the mothers before faecal transplants of a complete microbiota at birth; 100% of the female offspring developed diabetes . However, the unexpected reduction in insulitis in antibiotic-treated mice fed XOS early in life suggested that the alleviating effect is not necessarily a product of microbiota-induced anti-diabetogenic signals. Increasing evidence is establishing direct pathways by which dietary oligosaccharides can act on the immune system: for example, the gut barrier is permeable to milk oligosaccharides during infancy , and it can bind specific sugar receptors on human cells [32, 33] and thereby modulate the immune response systemically. Of course, the shift in microbial composition towards species with known probiotic potential cannot be excluded from also playing a protective role. There was, for example, a clear increase in Parabacteroides spp. which have been associated with anti-inflammatory and diabetes-reducing properties [34, 35]. Nonetheless, faecal microbiota transplantation from XOS-fed mice was not sufficient to transfer a reduction in time to diabetes onset. It is worth considering that in such a scenario, without XOS as an available nutrient source, the microbiota may function differently and, therefore, a microbiota-mediated anti-inflammatory effect of an XOS-supplemented diet should therefore not be ruled out. In contrast to the effect on insulitis, the development of sialadenitis was clearly regulated by gut bacteria, which is comparable with our previous study in germ-free mice . These findings warrant further investigation into the regulatory role of Parabacteroides spp. in the development of sialadenitis rather than insulitis.
According to the ‘perfect storm’ hypothesis, an increased intestinal permeability would allow for inappropriate food and microbial antigens to traverse the gut mucosa, triggering an aberrant immune response to otherwise harmless proteins . Fewer CD8+ NKT cells and cytotoxic CD8+ T cells in the lymphoid tissues and reduced translocation of bacteria to the mesenteric lymph nodes further support the minimal passage of intestinal antigens across the mucosal barrier in XOS-fed mice. Functional studies in BB-DP rats demonstrate that increased permeability of the small intestine, but not the colon, appeared before the development of insulitis . A later study verified that BB-DP rats had a higher urinary excretion ratio of lactulose:mannitol than Wistar rats before the onset of diabetes . However, no difference was observed in survival or islet inflammation for BB-DP rats when butyrate was administered to strengthen intestinal tight junctions, bringing into question the importance of a leaky gut, at least in the small intestine.
If and how the mucosal inflammation contributes to inducing an autoimmune cascade in the nearby pancreas remain unknown, but the antibiotic-treated mice shed some light on the importance of a leaky gut. The small intestinal barrier in these mice was no longer improved by the XOS diet, in contrast to markers of colonic barrier function. This is in accordance with a study from the 90s, demonstrating that dietary fibre increased crypt and mucin-containing goblet cell numbers in germ-free rats [26, 39] and that colonisation of the gut actually abolished this effect. The discrepancy between the different sections of gut is interesting as the XOS diet was also able to reduce islet inflammation in microbiota-depleted mice, suggesting that the colonic barrier function rather than the small intestinal barrier was involved in the alleviating effect. In support of this, a study demonstrated that glucagon-like peptide-2 reduced small intestinal transepithelial resistance in NOD mice but failed to delay the onset of type 1 diabetes . In another study, infection with Citrobacter rodentium increased large intestinal permeability at weaning, and this was followed by systemic infection of mesenteric and pancreatic lymph nodes, proliferation of diabetogenic CD8+ T cells and accelerated development of insulitis; this was in contrast to infection with a modified strain that did not disrupt the large intestinal barrier . The latter study also showed that the small intestinal permeability was increased at 12 weeks of age, long after islet inflammation began, but not before 12 weeks.
In conclusion, the inconsistency between studies investigating the ability of the gut barrier to regulate the development of diabetes appears to highlight the need to distinguish segments of the gastrointestinal tract based on functional differences. It is important to note that the small intestinal barrier function may play a more profound role in the systemic immune response and autoimmune reactions in the salivary glands, as the XOS diet not only failed to alleviate small intestinal permeability in antibiotic-treated mice, but also had no effect on systemic immunity and sialadenitis in the absence of a fully colonised gut. Functional studies combining methods to alleviate and aggravate gut permeability with environmental factors associated with a modified barrier function, such as the XOS diet, in various sections of the gastrointestinal tract are necessary to address such hypotheses. Nonetheless, considering the previous findings on increased intestinal permeability in individuals with type 1 diabetes [42, 43], eating a prebiotic supplement seems a feasible approach that, because of the effects on mucus production and tight junction proteins, may be able to limit the impact of environmental triggers that might otherwise translocate the compromised gut barrier.
The authors thank S. Høgh Madsen and S. Halkier from Research Bioanalysis at Novo Nordisk for analytical support.
CSL, HOP, ŁK, AV, WK, LFZ and CL obtained research data. DSN and AKH contributed to the conception and interpretation of this work. CHFH designed the study, obtained research data and wrote the manuscript. All authors revised and approved the final version of the manuscript. CHFH is responsible for the integrity of the work as a whole.
This project was funded by the 3G Center (Gut, Grain & Greens), supported by the Innovation Fund Denmark.
Duality of interest
The authors confirm there is no duality of interest associated with this manuscript.
- 8.Brugman S, Klatter FA, Visser JTJ et al (2006) Antibiotic treatment partially protects against type 1 diabetes in the Bio-Breeding diabetes-prone rat. Is the gut flora involved in the development of type 1 diabetes? Diabetologia 49(9):2105–2108. https://doi.org/10.1007/s00125-006-0334-0 CrossRefGoogle Scholar
- 19.Pridal L, Ingwersen SH, Larsen FS, Holst JJ, Adelhorst K, Kirk O (1995) Comparison of sandwich enzyme-linked immunoadsorbent assay and radioimmunoassay for determination of exogenous glucagon-like peptide-1(7-36)amide in plasma. J Pharm Biomed Anal 13(7):841–850. https://doi.org/10.1016/0731-7085(95)01312-9 CrossRefGoogle Scholar
- 35.Kverka M, Zakostelska Z, Klimesova K et al (2011) Oral administration of Parabacteroides distasonis antigens attenuates experimental murine colitis through modulation of immunity and microbiota composition. Clin Exp Immunol 163(2):250–259. https://doi.org/10.1111/j.1365-2249.2010.04286.x CrossRefGoogle Scholar
- 38.Neu J, Reverte CM, Mackey AD et al (2005) Changes in intestinal morphology and permeability in the biobreeding rat before the onset of type 1 diabetes. J Pediatr Gastroenterol Nutr 40(5):589–595. https://doi.org/10.1097/01.MPG.0000159636.19346.C1 CrossRefGoogle Scholar