Complex dietary polysaccharide modulates gut immune function and microbiota, and promotes protection from autoimmune diabetes

Summary

The dietary supplement and prebiotic values of β‐glucan‐rich products have been widely recognized and dietary approaches for modulating autoimmunity have been increasingly explored, we assess the impact of oral administration of high‐purity yeast β‐glucan (YBG) on gut immune function, microbiota and type 1 diabetes (T1D) using mouse models. Oral administration of this non‐digestible complex polysaccharide caused a dectin‐1‐dependent immune response involving increased expression of interleukin‐10 (IL‐10), retinaldehyde dehydrogenase (Raldh) and pro‐inflammatory cytokines in the gut mucosa. YBG‐exposed intestinal dendritic cells induced/expanded primarily Foxp3⁺, IL‐10⁺ and IL‐17⁺ T cells, ex vivo. Importantly, prolonged oral administration of low‐dose YBG at pre‐diabetic stage suppressed insulitis and significantly delayed the appearance of T1D in non‐obese diabetic (NOD) mice. Further, prolonged treatment with YBG showed increased Foxp3⁺ T‐cell frequencies, and a significant change in the gut microbiota, particularly an increase in the abundance of Bacteroidetes and a decrease in the Firmicute members. Oral administration of YBG, together with Raldh‐substrate and β‐cell antigen, resulted in better protection of NOD mice from T1D. These observations suggest that YBG not only has a prebiotic property, but also an oral tolerogenic‐adjuvant‐like effect, and these features could be exploited for modulating autoimmunity in T1D.

Introduction

β‐Glucans are non‐digestible complex polysaccharides, commonly found in plants, cereals, bacteria, yeasts and fungi, e.g. mushrooms. They can potentiate the function of immune system and improve the antimicrobial activity of innate immune cells.1, 2 β‐Glucans can activate both innate and adaptive immunity and hence, have been investigated as an effective systemic adjuvant in immunotherapies for cancer and infections.1, 2 Many of the β‐glucan‐containing preparations have been promoted as daily dietary supplements to enhance immune and metabolic functions,3 and are suggested to have prebiotic properties. Our recent report showed that systemic injection of low amounts of highly purified β‐glucan (purity > 98%) from Baker's yeast (Saccharomyces cerevisiae; YBG) modulates autoimmune progression and prevents type 1 diabetes (T1D) in non‐obese diabetic (NOD) mice,4 suggesting that it has previously unknown immune regulatory properties. Structurally, YBG contains β‐1,3‐linked d‐glucose molecules (β‐1,3‐d‐glucan) with β‐1,6‐linked side chains and it is recognized by innate immune cells, primarily through a pathogen recognition receptor called Dectin‐1.5, 6 Clinical studies using YBG preparations have shown beneficial effects,7, 8, 9, 10 but a clinical trial failed to detect changes in either cytokine production or microbicidal activity of leukocytes on oral administration of YBG.11 Overall, as most of these published studies on biological effects of β‐glucans have administered β‐glucans systemically, and/or used partially purified and poorly defined products, the degree and nature of immune modulation induced by purified β‐glucans and the impact of oral consumption of β‐glucans on autoimmunity are largely unknown.

Dietary approaches have increasingly been considered for modulating immune function, and preventing autoimmunity and other clinical conditions. Therefore, in this study, we examined the gut immune and microbiota modulatory properties of orally administered high‐purity YBG (1‐3,1‐6‐β‐glucan), and the impact on organ‐specific autoimmunity in T1D using NOD mice. We observed that oral administration of YBG resulted in the expression of higher levels of immune regulatory and pro‐inflammatory cytokines, and non‐cytokine factors such as vitamin A metabolizing enzyme, retinaldehyde dehydrogenase (Raldh), in the gut mucosa. This mixed innate immune response also contributed to an increase in the regulatory T (Treg) cell frequency. YBG‐recipient mice also showed a significantly higher abundance in the microbial communities of Bacteroidetes and Verrucomicrobia phyla, to which many polysaccharide‐fermenting bacteria belong, and decrease in Firmicute members, compared with untreated controls. Although oral administration of YBG was sufficient to significantly delay the hyperglycemia in NOD mice, co‐administration of YBG in an antigen‐specific manner (β‐cell‐antigen) and retinol further potentiated the diabetes protection effect in these mice. Our observations suggest that a combination of YBG‐induced innate immune response in the gut mucosa and altered microbiota contributes to the suppression of autoimmune progression in NOD mice, and these features can be exploited for preventing T1D in at‐risk individuals by using dietary‐supplement‐based immune modulators.

Materials and methods

Mice

C57BL/6, NOD/LtJ, OT‐II, NOD‐BDC2.5 and NOD‐Scid mice were purchased from the Jackson Laboratory (Bar Harbor, ME). C57BL/6‐Dectin‐1‐deficient (Dectin‐1^−/−) mice12 were provided by Dr. Iwakura, University of Tokyo, Japan. Foxp3‐GFP‐knockin (Foxp3‐GFP) mice were provided by Dr. Vijay Kuchroo (Harvard Medical School, Harvard, MA). Breeding colonies of these strains were established and maintained in the pathogen‐free facility of the University of Illinois at Chicago (UIC) or Medical University of South Carolina (MUSC). NOD‐Foxp3‐GFP mice were generated by backcrossing B6‐Foxp3‐GFP‐knockin mice to NOD background for 12 generations. To generate NOD‐BDC‐Foxp3‐GFP mice, NOD‐BDC2.5 mice were crossed with NOD‐Foxp3‐GFP mice. To detect hyperglycemia, glucose levels in blood collected from the tail vein of wild‐type NOD/Ltj and NOD‐Scid mice were tested using the Ascensia Micro‐fill blood glucose test strips and an Ascensia Contour blood glucose meter (Bayer, Leverkusen, Germany). Mice with glucose levels > 250 mg/dl for two consecutive weeks were considered diabetic. All animal studies were approved by the animal care and use committees at UIC and MUSC.

β‐cell‐antigen, peptides and other reagents

Immunodominant β‐cell antigen peptides – namely, (i) Insulin B (9–23), (ii) GAD65 (206–220), (iii) GAD65 (524–543), (iv) IA‐2β (755–777), (v) IGRP (123–145); (vi) BDC2.5 TCR reactive peptide (YVRPLWVRME; referred to as BDC‐peptide) and (vii) OVA (323–339) peptides (ISQAVHAAHAEINEAGR; referred to as OT‐II peptide – were custom synthesized (Genscript Inc., Piscataway, NJ) and used in this study. Peptides (i) to (v) were pooled at an equimolar ratio and used as β‐cell‐antigen as described earlier.13, 14, 15 Highly purified YBG (glucan from Baker's yeast, S. cerevisiae), ≥ 98% pure, was purchased from Sigma‐Aldrich (St Louis, MO); the purification method has been described before.16 This agent was tested for specific activity using thioglycollate‐activated macrophages as described previously.4, 17, 18 l‐Tryptophan, retinol, phorbol 12‐myristate 13‐acetate (PMA), ionomycin, Brefeldin A, and purified and conjugated antibodies were purchased from Sigma‐Aldrich, BD Biosciences (Franklin Lakes, NJ), eBioscience (San Diego, CA), Invitrogen (Carlsbad, CA), R&D Systems (Minneapolis, MN) and Biolegend Laboratories (San Diego, CA). Magnetic bead‐based total and CD4⁺ T cells, and CD11c⁺ dendritic cell isolation kits, as well as magnetic‐bead linked secondary antibodies (Miltenyi Biotec, Bergisch Gladbach, Germany; Invitrogen; and StemCell Technologies, Vancouver, BC, Canada) were used for enriching or depleting cell populations. In some experiments, intestinal epithelial cells were depleted by Percoll gradient centrifugation or using A33 antibody (Santa Cruz Biotechnology, Dallas, TX) and anti‐rabbit IgG‐magnetic beads. Enzyme‐linked immunosorbent assay and magnetic bead‐based suspension array reagents/kits were purchased from R&D Systems, BD Biosciences, Invitrogen, Millipore, Affymetrix‐eBioscience and/or Millipore (Burlington, MA).

Treatments

Six to eight‐week‐old C57BL/6 and NOD‐BDC‐Foxp3‐GFP, 10‐week‐old pre‐diabetic (glucose levels < 100 mg/dl) female NOD/Ltj mice were given YBG suspension in saline (250 μg/mouse/day) for different durations by oral gavage. Control mice were given saline. In some experiments, mice were given a higher dose of YBG (2000 μg/mouse/day). In some experiments, mice were also given synthetic peptides (5 μg/mouse/day) or retinol (0·5 μg/mouse/day) by oral gavage. For depleting gut microbiota, C57BL/6 mice were given drinking water containing a broad‐spectrum antibiotic cocktail [ampicillin (1 g/l), vancomycin (0·5 g/l), neomycin (1 g/l) and metronidazole (1 g/l)] for up to 30 days.

Dendritic cells and ex vivo assays

Dendritic cells (DCs) were isolated from the ileum of control and YBG‐fed mice. Intestinal tissues were cleaned thoroughly by flushing with phosphate‐buffered saline‐containing antibiotics, cut into small pieces, incubated in RPMI‐medium containing collagenase (0·5 mg/ml) and DNAse (50 μg/ml) for 1 hr, passed through a 70‐μm filter, and washed using RPMI medium with 10% fetal bovine serum. Percoll gradient was generated by layering 100%, 40% and 25% Percoll in which intestinal cells were suspended, and centrifuged for 15 min at 2000 g. Cells at the interphase of 100% and 40% Percoll layers were collected as immune‐cell‐rich fraction for DC isolation. Intestinal CD11c⁺ DCs were enriched from the immune cell fraction using magnetic bead‐linked anti‐CD11c antibody reagent. In some experiments, cells were stained using fluorochrome‐labeled CD11c and CD103 antibodies, and CD11c⁺ CD103⁺ and CD11c⁺ CD103⁻ populations were enriched using a MoFlo high‐speed cell sorter. In some experiments, transcript levels of various factors in these cells, and cytokines spontaneously secreted by them upon ex vivo culture for 48 hr without additional stimulation, were determined. In antigen presentation assays, purified T cells (1 × 10⁵ cells/well) were incubated with antigenic peptide‐pulsed DCs (2 × 10⁴ DCs/well). In some assays, these cultures were supplemented with retinol and transforming growth factor‐β ₁. For some assays, in vitro stimulated or freshly isolated T cells were re‐stimulated using PMA (50 ng/ml) and ionomycin (500 ng/ml) in the presence of Brefeldin A (1 μg/ml) for 4 hr before staining for intracellular cytokines. In some assays, spleen, mesenteric lymph node (MLN) and pancreatic lymph node (PnLN) cells (2 × 10⁵ cells/well) from treated and control mice were stimulated with anti‐CD3 antibody (1 μg/ml) for 24 hr or β‐cell‐antigen (5 μg/ml) for 48 hr and spent media were tested for cytokines.

Adoptive T‐cell transfer experiment

Pancreatic LN cell suspension of control and YBG‐fed mice were transferred into 6‐ to 8‐week‐old NOD‐Scid mice (intravenously; 1 × 10⁶ cells/mouse); glucose levels were tested every week.

Histochemical analysis of pancreatic tissues

Pancreata were fixed in 10% formaldehyde, 5‐μm paraffin sections were cut, and stained with hematoxylin & eosin. Stained sections were analyzed using a grading system in which 0 = no evidence of immune cell infiltration, 1 = peri‐islet infiltration (< 5%), 2 = 5–25% islet infiltration, 3 = 25–50% islet infiltration, and 4 = > 50% islet infiltration and completely destroyed islets, as described in our earlier studies.13, 14, 15, 19

16S rRNA gene targeted sequencing and bacterial community profiling

Total DNA was prepared from the fecal pellets for bacterial community profiling as detailed in our previous reports.15, 20 DNA in the samples was amplified by polymerase chain reaction using 16S rRNA gene‐targeted primer sets to assess the bacterial levels. Sequencing of the V3–V4 region of 16S rRNA gene was performed using Illumina MiSeq platform at the MUSC genomic center. The sequencing reads were fed into the Metagenomics application of basespace (Illumina, San Diego, CA) for performing taxonomic classification of 16S rRNA targeted amplicon reads using an Illumina‐curated version of the GreenGenes taxonomic database, which provided raw classification output at multiple taxonomic levels. The sequences were also fed into QIIME, the open reference operational taxonomic units (OTUs) picking pipeline,21 using the QIIME pre‐processing application of basespace. The OTUs were compiled to different taxonomical levels based upon the percentage identity to greengenes reference sequences (i.e. > 97% identity) and the percentage values of sequences within each sample that mapped to respective taxonomical levels were calculated. The OTUs were also normalized and used for metagenome prediction of the Kyoto encyclopedia of genes and genomes orthologs employing PICRUSt as described before.22, 23, 24, 25 The predictions were summarized to multiple levels and functional categories were compared among control and YBG‐fed groups using the Statistical Analysis of Metagenomic Profile Package (STAMP) as described previously.23

Statistical analysis

Mean, SD and statistical significance (P‐value) were calculated, and graphic visualizations including heatmaps were made using excel (Microsoft, Redmond, WA), GraphPad PRISM (San Diego, CA), morpheus (Broad Institute, Cambridge, MA), and STAMP23 applications. One‐ or two‐tailed t‐test was employed, unless specified, for values from in vitro and ex vivo assays where two groups were compared. Log‐rank analysis, which compares the disease‐free duration of the total monitoring period between two groups, was performed to compare T1D (hyperglycemia) incidences of two groups. Fisher's exact test was used for comparing the number of immune cell infiltrated islets in test versus control groups. All the statistical analyses for microbial sequences were performed employing two‐tailed t‐tests and the P‐values were corrected from multiple tests using the Benjamini and Hochberg approach in STAMP. A P‐value of ≤ 0·05 was considered significant.

Results

Oral administration of YBG results in Dectin‐1‐dependent gut immune activation

Previous studies have suggested that partially purified β‐glucans and β‐glucan‐containing agents can activate the gut immune system.26, 27, 28 In this study, wild‐type and Dectin‐1^−/− mice were given high‐purity (> 98%) YBG (high‐dose: 2000 μg/mouse/day or low‐dose: 250 μg/mouse/day) for 3 days, and overall immune phenotype of the small intestine was determined. Figure 1(a) shows that the small intestine of wild‐type mice, but not Dectin‐1^−/− mice, expressed significantly higher levels of cytokine and non‐cytokine factors compared with that of untreated mice. YBG‐induced gut innate immune response in wild‐type mice involved higher expression of mRNA both immune regulatory and pro‐inflammatory factors including interleukin‐10 (IL‐10), Raldh1A2, tumor necrosis factor‐α, IL‐6 and IL‐1β. Similar analysis of distal colon showed relatively higher expression of a few cytokine and chemokine factors in YBG‐treated mice (not statistically significant; not shown). These results suggest the ability of purified YBG to induce a Dectin‐1‐dependent immune response in the gut mucosa. Short‐treatment with low‐dose YBG induced only modestly higher, not statistically significant, levels of some cytokine and non‐cytokine factors in the small intestine (see Supplementary material, Fig. S1).

Figure 1.

Oral administration of yeast β‐glucan (YBG) causes Dectin‐1‐dependent immune activation in the gut mucosa. (a) Eight‐week‐old wild‐type (WT) and Dectin‐1^−/− B6 mice were given YBG suspension (2000 μg/mouse/day) by oral gavage for three consecutive days and killed after 24 hr to harvest the intestine. cDNA prepared from the distal ileum region was subjected to quantitative PCR and the expression levels of cytokines and non‐cytokine factors were compared. Expression levels relative to β‐actin expression were plotted as a heatmap with Morpheus application; four or five mice/group and the assay was performed in triplicate for each sample. Similar analysis of distal colon did not show significant differences (not shown). (b) WT and Dectin‐1^−/− B6 mice were treated and killed as described above, small intestinal CD11c⁺ dendritic cells (DCs) were MACS‐sorted from intestinal epithelial cell‐depleted (using A33 antibody), enriched intestinal immune cell fraction. Cells were cultured in triplicate for 48 hr (2 × 10⁵/well in 96‐well plate) and supernatants were tested for the levels of spontaneously secreted cytokines by suspension bead array. Mean values of untreated WT and Dectin‐1^−/− mouse cells were considered as baseline value of 1 for calculating fold cytokine secretion by cells from YBG‐treated mice; n = 3 pools of cells (two or three mice/pool). Heatmap was generated using graphpad prism. (c) Freshly isolated DCs were pulsed with OT‐II peptide and cultured (2 × 10⁴ cells/well) with OT‐II‐Foxp3‐GFP mouse splenic T cells (1 × 10⁵ cells/well) for 4 days and examined for CD4⁺ cells with GFP (Foxp3) expression. Representative FACS scatter plots (upper panels) and mean ± SD values of three independent experiments in triplicate (lower) are shown. (d) DCs and OT‐II mouse T cells were cultured similarly, re‐stimulated using phorbol 12‐myristate 13‐acetate and ionomycin for 4 hr and stained for intracellular cytokines. Representative FACS scatter plots of three independent experiments are shown. (e) Supernatants collected from the primary cultures similar to that described for (c) were tested by ELISA for cytokine levels. Mean ± SD values of three independent parallel experiments in triplicate are shown. These experiments were repeated at least twice. Statistical significance was calculated by t‐test.

Intestinal DCs from YBG‐fed mice produce higher amounts of cytokines and skew T‐cell response ex vivo

Since intestinal immune cells, DCs in particular, are known to encounter dietary and microbial agents in the gut lumen,29, 30, 31 in vivo response of intestinal DCs (IDCs) to short‐term treatment with high‐dose YBG was examined. As shown in Fig. 1(b), IDCs from wild‐type mice, but not Dectin‐1^−/− mice, that received YBG orally showed higher secretion of pro‐ and anti‐inflammatory cytokine and chemokine factors compared with respective untreated controls. As YBG exposure caused activation of IDCs in vivo, these cells were tested for their ex vivo ability to promote T‐cell activation and differentiation upon antigen presentation. The IDCs from YBG‐fed mice could induce significantly higher frequencies of Foxp3⁻, IL10⁻ and IL17A⁻‐expressing/producing T cells, compared with their counterparts from untreated mice (Fig. 1c–e). On the other hand, compared with IDCs from control mice, these cells from YBG‐fed mice induced a lower interferon‐γ (IFN‐γ) response in T cells. These observations show that YBG exposure activates IDCs and enhances their immune modulatory/regulatory properties.

CD103⁺ IDCs, but not CD103⁻ IDCs, of YBG‐treated mice induce/expand Foxp3⁺ T cells

As CD103⁺ cells are the key IDC population that expresses the tolerogenic enzyme Raldh1A2 and other immune regulatory factors, as well as directly encountering gut luminal content,29, 32, 33 and because oral administration of YBG caused higher expression of mRNA for these factors in the intestine, CD103⁺ and CD103⁻ IDCs of YBG‐fed mice were sorted and tested for the expression levels of mRNA for key cytokine and non‐cytokine factors. As shown in Fig. 2(a), CD103⁺ IDCs mice fed a high‐dose YBG diet expressed higher levels of mRNA for Raldh1A2 and IL‐10 compared with their counterparts from control mice whereas CD103⁻ IDCs from these YBG‐fed mice expressed relatively higher levels of mRNA for IL‐6. Ex vivo antigen presentation assay revealed that although CD103⁺ IDCs from YBG‐fed mice induced significantly higher frequencies of IL‐10‐producing T cells, CD103⁻ IDCs from these mice induced IL‐17‐producing T cells (Fig. 2b). Further, CD103⁺ and CD103⁻ IDCs from YBG‐fed mice induced lower and higher IFN‐γ production, respectively, albeit not significantly statistically, compared with their counterparts from control mice (not shown). Importantly, CD103⁺ IDCs from YBG‐fed mice showed superior ability to induce/expand Foxp3⁺ T cells, in the absence or presence of Raldh‐substrate (retinol) and transforming growth factor‐β ₁, compared with their counterparts from control mice (Fig. 2c). These observations suggest that YBG exposure activates both CD103⁺ and CD103⁻ IDCs and promotes distinct T‐cell responses.

Figure 2.

CD103⁺ intestinal dendritic cells (IDCs) from yeast β‐glucan (YBG) ‐treated mice induce T cells with regulatory phenotype ex vivo. Wild‐type (WT) B6 mice were treated and killed as described for Fig. 1, small intestinal CD11c⁺ CD103⁻ and CD11c⁺ CD103⁺ DCs were FACS‐sorted from intestinal epithelial cell‐depleted, enriched intestinal immune cell fraction. (a) Freshly isolated cells were subjected to quantitative PCR assay. Expression levels relative to β‐actin expression are shown; n = 3 pools of cells (four or five mice/pool). (b) OT‐II peptide‐pulsed DCs and OT‐II T cells were cultured and processed for detection of intracellular cytokines as described for Fig. 1. Representative FACS scatter plots (upper panel) and mean ± SD values of three independent experiments conducted in triplicate (lower panel) are shown. (c) DCs were cultured similarly with OT‐II‐Foxp3‐GFP T cells in the absence or presence of retinol and examined for CD4⁺ T cells with GFP expression. Representative FACS scatter plots (left) and mean ± SD values of three independent experiments in triplicate (right) are shown. These experiments were repeated at least thrice. Statistical significance was calculated by t‐test.

Prolonged oral administration of YBG protects pre‐diabetic NOD mice from T1D

For the short‐term studies described above, which were conducted using B6 mice due to the availability of Dectin‐1 knockout mice in that background, high‐dose YBG (2000 μg/mouse/day) was used to reliably assess the Dectin‐1‐dependent and overall in vivo effects of orally administered YBG. However, as oral administration of YBG resulted in up‐regulation of both immune regulatory and pro‐inflammatory factors for experiments involving long‐term treatment, clinically relevant low‐dose YBG (250 μg/mouse/day) is desired as a precautionary measure to minimize the likelihood of intestinal inflammation upon prolonged treatment. Modulation of gut immune function upon short‐term treatment with high‐dose YBG prompted studies to determine the impact of low‐dose YBG consumption on gut immune phenotype and pancreatic autoimmunity using a spontaneous model of T1D, which was employed in our previous study to determine the impact of systemic injection of YBG.4 As observed in the Supplementary material (Fig. S2a), the gut mucosa of YBG‐treated NOD mice showed cytokine and Raldh1A2 mRNA expressions similar to that of B6 mice (shown in Fig. 1 and the Supplementary material, Fig. S2b). Importantly, pre‐diabetic NOD mice that were given low‐dose YBG for 30 consecutive days showed significantly delayed hyperglycemia (Fig. 3a). This treatment prevented the disease in > 40% of the mice for at least 18 weeks post‐treatment, whereas 100% of the control mice developed diabetes within this period. One set of mice were killed 7 days post‐treatment and their pancreatic tissues were examined for severity of insulitis. The pancreatic islets of YBG‐treated mice showed significantly less severe immune cell infiltration and insulitis compared with untreated mice (Fig. 3b). About 50% of the islets in mice that received β‐glucan showed insulitis grade ≤ 2, but at least 60% of the islets in untreated euglycemic mice showed insulitis severity grade > 2. These results suggest that immune modulation induced by orally administered YBG leads to curbed immune cell infiltration and β‐cell destruction, resulting in delayed onset of hyperglycemia in NOD mice.

Figure 3.

Oral administration of low‐dose yeast β‐glucan (YBG) resulted in delayed hyperglycemia in non‐obese diabetic (NOD) mice. Female NOD mice were given YBG suspension (250 μg/mouse/day) or saline by oral gavage for 30 consecutive days starting at 10 weeks of age. (a) One set of mice (n = 14 for control and n = 16 for YBG groups) were tested for blood glucose levels, weekly for up to 25 weeks post‐treatment initiation, to detect hyperglycemia/diabetes. (b) One set of mice (five mice/group) were killed 1‐week post‐treatment and hematoxylin & eosin‐stained pancreatic sections were examined for insulitis. Representative islets with insulitis grades (left) and overall percentage of islets with different insulitis grades (right) are shown. Statistical significance was calculated by Log‐rank test (for a) and Fisher's exact test (for b).

Prolonged oral administration of YBG results in modulation of T cell phenotype and function in the pancreatic LNs

The YBG‐fed mice showed less severe insulitis and significant delay in hyperglycemia, so T cells from these mice were examined for their phenotypic and functional properties in comparison with their counterparts from control mice. Foxp3⁺ T‐cell frequencies were profoundly higher in the intestinal mucosa and the PnLNs of YBG‐fed mice compared with control mice (Fig. 4a). Importantly, both β‐cell‐antigen (Fig. 4b) and PMA/ionomycin (not shown) stimulations ex vivo, followed by FACS analysis, showed that IL‐10⁺, IL‐17⁺ and IL‐4⁺ T‐cell frequencies were significantly higher and IFN‐γ ⁺ T‐cell frequency is relatively lower in the PnLNs of YBG‐fed mice compared with controls. Further, levels of these cytokines secreted by PnLN cells from YBG‐fed and control mice in response to β‐cell‐antigen challenge ex vivo showed similar trends (Fig. 4c). To determine if YBG treatment impacted the function of immune cells, diabetogenic property of PnLN cells was examined by adoptively transferring these cells from YBG‐treated and control mice into NOD‐Scid mice. Figure 4(d) shows that NOD‐Scid mice that received immune cells from YBG‐treated mice developed hyperglycemia relatively more slowly compared with control cell recipients. Further, pancreatic tissues of NOD‐Scid mice that received cells from YBG‐treated mice showed significantly lower insulitis severity compared with control mouse cell recipients (Fig. 4e). Overall, these observations suggest that prolonged YBG consumption results in immune modulation not only in the gut mucosa, but also in the systemic compartment including the pancreatic microenvironment, thereby causing the suppression of insulitis in NOD mice.

Figure 4.

Yeast β‐glucan (YBG) treatment caused modulation of T‐cell function in the systemic compartment and suppression of autoimmunity. Female non‐obese diabetic (NOD) mice were given low‐dose YBG by oral gavage for 30 consecutive days starting at 10 weeks of age as described for Fig. 5 and the mice were killed 1 week post‐treatment. (a) Mesenteric lymph node (MLN), SiLP and pancreatic lymph node (PnLN) cells were stained for Foxp3 expression and analyzed by FACS. (b) PnLN cells (2 × 10⁵ cells/well) from these mice were stimulated using β‐cell‐antigen for 24 hr and examined for intracellular cytokine levels by FACS. Representative FACS plots (upper panels) and mean ± SD values of four or five mice (lower panel) are shown for (a) and (b). (c) PnLN cells (2 × 10⁵ cells/well) were stimulated using β‐cell‐antigen for 48 hr and the supernatants were examined for cytokines. (d) PnLN cells from at least five treated and five control mice were pooled separately and injected intravenously into NOD‐Scid mice (1 × 10⁶ cells/mouse; nine recipients/group) and monitored for blood glucose levels to detect diabetes. (e) One set of mice (three/group) was killed 3 weeks post‐cell transfer and hematoxylin & eosin‐stained pancreatic sections were examined for insulitis. Overall percentage of islets with different insulitis grades are shown. Statistical significance was calculated by t‐test (for a–c), by log‐rank test (for up to 7 and 10 weeks) that takes the disease‐free duration of the monitoring period into account (for d), and by Fisher's exact test (for e).

Prolonged oral administration of low‐dose YBG results in changes in the gut microbiota

Complex non‐digestible dietary polysaccharides are thought to be degraded/fermented by colonic bacteria and affect the overall gut microbiota composition and immune function,34, 35, 36 which could impact autoimmune outcomes. Therefore, we determined the impact of prolonged oral treatment of NOD mice with YBG on gut microbiota. The mice were given YBG by oral gavage every day for 30 days, fresh fecal pellets from treated and control mice were collected, and subjected to 16S rRNA targeted gene sequencing to determine the profiles of microbial communities. Compilation of OTUs to different taxonomical levels showed that the abundances of major phyla were significantly different in YBG‐fed mice, compared with control mice (Fig. 5a). YBG‐treated mice showed significantly higher abundance of Bacteroidetes and Verrucomicrobia phyla members and profoundly lower abundance of Firmicutes and Proteobacteria in their fecal samples compared with controls. Examination of OTUs at genus level revealed that the abundances of many major microbial communities were significantly different in YBG‐treated and control mice (Fig. 5b). Overall, a major shift in favor of microbial communities belonging to Bacteroidetes and Verrucomicrobia, which include many polysaccharide‐degrading bacterial communities, over Firmicutes occurred in YBG‐treated mouse gut. For OTU‐based prediction of metabolic functions of gut microbes that may be associated with complex dietary polysaccharide degradation and metabolite production, picrust application was employed. Predicted major pathways that are overrepresented in YBG‐treated mice, based on 16S rRNA gene identity, include carbohydrate metabolism, biosynthesis of secondary metabolites, energy metabolism, cofactor and vitamin metabolism, and xenobiotics biodegradation and metabolism (Fig. 5c). Overrepresentation of these pathways in the gut microbiota of YBG‐fed mice suggests that oral consumption of this dietary polysaccharide can cause selective enrichment of carbohydrate‐fermenting/consuming and metabolite‐producing bacteria in the gut.

Figure 5.

Prolonged oral administration of low‐dose yeast β‐glucan (YBG) results in changes in the composition of gut microbiome in non‐obese diabetic (NOD) mice. Ten‐week‐old pre‐diabetic NOD mice were given saline or YBG suspension (250 μg/mouse/day) for 30 consecutive days by oral gavage as described for Fig. 3, and fresh fecal pellets from individual animals (four mice/group) were collected post‐treatment, DNA samples prepared from these fecal pellets were subjected to 16S rRNA gene (V3V4) ‐targeted sequencing using the Illumina MiSeq platform. The operational taxonomic units (OTUs) that were compiled to different taxonomical levels based upon the percentage identity to reference sequences (i.e. > 97% identity) and the percentage values of sequences within each sample that map to specific phylum and genus were calculated. (a) Relative abundances of 16S rRNA gene sequences in fecal samples of individual mice before and after treatment at phyla level are shown. (b) Mean relative abundances of sequences representing the major microbial communities (> 1% of total sequences) in YBG‐treated group at genus level are shown. (c) Normalized OTU‐biome table was used for predicting gene functional categories using PICRUSt application and selected level 2 categories of Kyoto Encyclopedia of Genes and Genomes pathway are shown. Statistical analyses were performed employing t‐test and the P‐values were corrected from multiple tests using Benjamini and Hochberg approach in STAMP.

YBG‐treatment‐associated protection from T1D is, in part, gut microbe dependent

As YBG treatment of NOD mice resulted in significant changes in the composition of gut microbiota, the influence of microbiota on YBG‐treatment‐associated protection of NOD mice from T1D was examined. Pre‐diabetic mice were given a cocktail of broad‐spectrum antibiotics in drinking water to deplete their gut microbiota (see Supplementary material, Fig. S3) and treated concurrently with YBG. Microbiota‐depleted control mice showed relatively rapid progression of insulitis and higher incidence of hyperglycemia, compared with untreated control mice with intact microbiota (Fig. 6a). On the other hand, microbiota‐depleted mice showed only a modest protection from hyperglycemia upon YBG treatment compared with mice that received YBG alone. Correspondingly, the overall severity of insulitis in microbiota‐depleted mice that received YBG was found to be considerably higher than that observed in YBG‐treated mice with intact microbiota, compared with microbiota‐depleted control mice (Fig. 6b). These observations suggest the possibility that both gut immune modulation induced directly by YBG and indirectly by altered gut microbiota contribute to the protection of NOD mice from T1D upon oral consumption of YBG. These observations prompted examination of Treg cell frequencies in the PnLNs of YBG‐treated NOD mice without and with microbial depletion. Figure 6(c) shows, similar to Fig. 4(a), significantly higher frequencies of Foxp3⁺ T cells in the PnLNs of YBG‐treated mice with intact gut microbota compared with their control counterparts. Importantly, although depletion of gut bacteria alone caused substantial reduction in Foxp3⁺ T‐cell frequencies in both control and YBG‐recipient mice compared with their counterparts with intact microbiota, PnLNs from microbiota‐depleted YBG‐recipient NOD mice showed significantly higher frequencies of Foxp3⁺ T cells compared with microbiota‐depleted controls. These observations show that although gut microbes appear to contribute to the maintenance of overall Treg cell function, oral consumption of YBG has a direct impact on immune activation and T‐cell phenotype.

Figure 6.

Yeast β‐glucan (YBG) treatment‐associated protection of non‐obese diabetic (NOD) mice from type 1 diabetes (T1D) is partially gut microbe dependent. Female NOD mice were given YBG for 30 consecutive days starting at 10 weeks of age as described for Fig. 5, and maintained on regular drinking water or drinking water containing a broad‐spectrum antibiotic cocktail throughout the experimental period. (a) One set of mice (eight mice/group) were tested for blood glucose levels, weekly for up to 12 weeks post‐treatment initiation, to detect hyperglycemia/diabetes. (b) One set of mice (five mice/group) were killed 1‐week post‐treatment and hematoxylin & eosin‐stained pancreatic sections were examined for insulitis as described for Fig. 5. Overall percentage of islets with different insulitis grades are shown. (c) Single‐cell suspension of pancreatic lymph node from mice killed 1 week post‐treatment were stained for CD4⁺ cells with Foxp3 expression, and representative FACS graphs (left panel) and mean ± SD values of five mice tested individually (right panel) are shown. Statistical significance was calculated by Log‐rank test (for a), Fisher's exact test (for b), or t‐test (for c).

Co‐administration of YBG, Raldh substrate and β‐cell‐antigen results in expansion of Foxp3⁺ cells and better protection of NOD mice from T1D

Treatment with YBG appears to have a direct, in addition to microbiota‐dependent, impact on enhancing gut immune regulation. Further, the YBG‐induced gut innate immune response involves increased Raldh1A2 and IL‐10 expression, and this regulatory innate response appears to contribute to the protection of NOD mice from T1D. Therefore, whether the YBG‐induced gut innate immune response can be exploited to promote β‐cell‐antigen‐specific immune regulation and better protection from T1D was investigated. In one set of experiments, NOD‐BDC‐Foxp3‐GFP mice were fed with YBG, BDC‐peptide and/or retinol for a brief period (three consecutive days and rested for 4 days) and examined for the frequencies of GFP⁺ (Foxp3⁺) T cells in the gut mucosa. As shown in Fig. 7, mice that received a combination of YBG, BDC‐peptide and retinol showed a robust increase in Foxp3⁺ T cells in Peyer's patches compared with other groups of mice. This brief treatment with YBG, BDC‐peptide and retinol combination also caused a significant increase in IL‐10 response in gut mucosa compared with mice that received individual agents (not shown).

Figure 7.

Short‐term treatment with yeast β‐glucan (YBG), antigen and retinol resulted in a robust regulatory T (Treg) cell response in vivo. Female non‐obese diabetic (NOD)‐BDC2.5‐Foxp3‐GFP mice were given YBG (250 μg/mouse/day), BDC2.5 peptide (5 μg/mouse/day) and/or retinol (0·5 μg/mouse/day) for three consecutive days, killed on day 7 and examined for GFP ⁺ (Foxp3⁺) CD4⁺ T cells in the Peyer's patch by FACS. Representative FACS graphs (left panel) and mean ± SD values of three mice/group tested individually (right panel) are shown. Statistical significance was calculated by t‐test.

We, then, examined the impact of short‐term co‐administration of YBG, β‐cell‐antigen (peptide mix) and retinol on T1D in NOD mice. Ten‐ to twelve‐week‐old pre‐diabetic mice were treated with YBG, β‐cell‐antigen and retinol individually or in combination for 3 days, rested for 4 days, and the treatment was repeated for an additional 3 days (a total of 6 days instead of 30 consecutive days of treatment as described for Fig. 4), and monitored for hyperglycemia or examined for insulitis and immune cell phenotypes. As shown in Fig. 8(a), although this short‐term treatment with YBG alone, or other individual agents, did not produce substantial protection from T1D, mice that received the YBG, β‐cell‐antigen and retinol combination showed a significant delay in hyperglycemia compared with untreated controls as well as YBG‐treated mice. Pancreatic islets of mice that received this combination treatment showed less severe insulitis compared with the control groups (Fig. 8b). Importantly, PnLN cells from mice that received this combination treatment showed relatively higher frequencies of Foxp3⁺ T cells (Fig. 8c). Furthermore, higher frequencies of IL‐10⁺ and IL‐17⁺ and lower frequencies of IFN‐γ ⁺ cells were detected in ex vivo β‐cell‐antigen challenged PnLNs from mice that received the combination therapy (Fig. 8d, and see Supplementary material, Fig. S4). The PnLN cells from these mice also secreted higher amounts of IL‐10 and IL‐17 and lower amounts of IFN‐γ upon ex vivo challenge with β‐cell‐antigen (Fig. 8e). T cells enriched from the pancreatic tissues of mice that received the combination treatment also showed a similar trend in Foxp3⁺, IFN‐γ⁺ and IL‐10⁺ Treg cells (see Supplementary material, Fig. S5). Mice that received short‐term treatment with YBG or retinol alone showed no profound difference in cytokine response. Of note, these pre‐diabetic mice that received oral administration of β‐cell‐antigen alone showed not only severe insulitis and rapid progression to hyperglycemia, but also secreted higher pro‐inflammatory IFN‐γ and lower IL‐17 and IL‐10 compared with untreated controls. Lower Foxp3⁺, IL‐10⁺ and IL‐17⁺ cell frequencies, and higher IFN‐γ ⁺ cell frequencies, were also detected in the PnLNs of mice that received β‐cell‐antigen alone. Overall, these observations suggest that although oral administration of β‐cell‐antigen alone, at close to disease onset, could accelerate the disease progression by enhancing pro‐inflammatory T helper type 1 (Th1) cytokine responses and suppressing Treg cell function, the YBG‐induced gut regulatory immune response can be exploited to enhance/promote oral tolerance against self‐antigens, even at advanced stages of disease progression, to prevent T1D.

Figure 8.

Short‐term treatment with yeast β‐glucan (YBG), β‐cell‐antigen and retinol resulted in robust modulation of the immune response and better protection of non‐obese diabetic (NOD) mice from type 1 diabetes (T1D). Ten‐ to 12‐week‐old female NOD mice (eight mice/group) were given YBG, β‐cell‐antigen peptide mix and/or retinol for 3 consecutive days by oral gavage and the treatments were repeated once with a 4‐day interval. (a) One set of mice was tested for blood glucose levels weekly for up to 12‐weeks post‐treatment initiation and percentage of mice in each group with hyperglycemia are shown (left panel). (b) One set of mice (four mice/group) was killed, 2‐week post‐treatment and hematoxylin & eosin‐stained pancreatic sections were examined for insulitis. Overall percentage of islets with different insulitis grades are shown. (c) Fresh pancreatic lymph node (PnLN) cells from one set of mice (three mice/group) killed 7 days post‐treatment were examined for Foxp3⁺ T‐cell frequencies by FACS. (d) PnLN cells from an additional set of mice were cultured with β‐cell‐antigen peptide mix for 24 hr and examined for intracellular cytokine levels by FACS. Mean ± SD values of three mice/group tested individually are shown for (c) and (d). Representative FACS graphs are shown in the Supplementary material (Fig. S4). (e) PnLN cells of mice from a similar experiment were cultured with β‐cell‐antigen peptide mix for 48 hr and supernatants were tested for cytokine levels by ELISA. Statistical significance was calculated by Log‐rank test (for a), Fisher's exact test (for b) or t‐test (for c–e).

Discussion

A large body of literature suggests that certain complex dietary polysaccharides such as β‐glucans of different microbial and plant origins affect the host immune function.26, 27, 28, 37, 38 Much of this literature consists of in vitro experiments or studies in which these polysaccharide preparations were systemically injected, so the true impact of orally administering β‐glucans on gut immune function is largely unknown. Also, use of poorly defined and crude polysaccharide preparations, β‐glucan of yeast origin in particular, has contributed greatly to the confusion, misconceptions and controversies regarding its biological effects. Hence, the true immunological effects following systemic and oral administration of well‐defined β‐1,3/1,6‐glucans and the impact on autoimmunity are largely unknown. In this regard, our recent report showed that systemic administration of low‐dose purified YBG prevents autoimmune diabetes in NOD mice by inducing an immuno‐adjuvant effect,4 suggesting that β‐1,3/1,6‐glucan interaction with innate immune receptors in the systemic compartment enhances overall immune regulation under autoimmune susceptibility. As orally consumed dietary polysaccharides can influence the gut immune function by directly interacting with cell surface receptors and indirectly through gut microbes,27, 28, 39 in this study, we determined the impact of oral administration of purified YBG as a dietary supplement on gut immune function, fecal microbiota composition and autoimmune diabetes. Here, we show that oral administration of YBG results in enhanced gut immune regulation, altered gut microbiome and protection of NOD mice from T1D. We also show that YBG not only has a prebiotic effect on gut microbes, but also has a direct immune regulatory oral‐adjuvant property, which can be exploited to modulate autoimmunity for preventing T1D.

Several reports have shown that the biological effect of β‐1,3/1,6‐glucans such as YBG is the result of interaction with specific receptors like Dectin‐1 on immune cells.4, 5, 6 Importantly, it is now recognized that gut‐resident DCs have the ability to interact directly with dietary and microbial antigens of the gut lumen.29, 30, 31 Our experiments using wild‐type and Dectin‐1‐deficient mice show that the overall and IDC‐specific immune response of small intestine to orally administered purified YBG is primarily Dectin‐1‐dependent. Interestingly, as indicated by the ability of IDCs to promote Treg cell induction/expansion as well as to modulate the cytokine response of T cells, the YBG‐induced gut innate immune response is immune regulatory in nature.

Colonic microbial communities can ferment non‐digestible dietary polysaccharides and use them as energy sources for their growth.40, 41, 42 Therefore, these polysaccharides are being considered as prebiotics with the ability to promote the growth of specific beneficial microbial communities. We observed that prolonged oral treatment using low‐dose YBG causes changes in the gut microbial composition by reducing the abundance of Firmicutes and increasing the abundance of the Bacteroidetes and Verrucomicrobia phyla, which include many polysaccharide‐fermenting bacteria. Further, predictive functional profiling of fecal microbiota shows overrepresentation of metabolic pathways linked to carbohydrate metabolism, glycan biosynthesis and metabolism, and biosynthesis of secondary metabolites after YBG treatment compared with the pre‐treatment time‐point.

Previous studies have shown that gut colonization by specific microbial communities causes induction and expansion of Foxp3⁺ Treg cells in the intestinal and systemic compartments.43, 44, 45 Further, gut bacteria play a critical role in maintaining peripheral tolerance and gut immune homeostasis.46, 47 Our results show that depletion of gut bacteria using broad‐spectrum antibiotics diminishes Foxp3⁺ T‐cell frequency, compared with mice with an intact gut microbiota, but less significantly in YBG‐treated mice compared with their control counterparts. Our observations that both control and bacteria‐depleted mice, which received YBG, have higher frequencies of Treg cells in the small intestine compared with their control counterparts with intact microbiota suggest that the YBG‐treatment‐associated increase in Treg cell abundance is, at least in part, the result of its direct interaction with gut mucosa. On the other hand, our results show that YBG‐treatment‐induced enhanced immune regulation in the large intestine is primarily microbiota‐dependent. Nevertheless, future studies using gnotobiotic mice conventionalized using control and YBG‐shaped microbiota will be needed to conclusively determine whether the YBG‐treatment‐shaped gut microbiota contributes to Treg cell induction/expansion and overall immune regulation.

Progression of autoimmune diseases including T1D in at‐risk individuals and rodent models can be influenced by dietary factors and gut microbiota.48, 49, 50, 51 Importantly, clinical trials and pre‐clinical studies have exploited the tolerogenic nature of the gut mucosa for inducing self‐antigen‐specific immune modulation to prevent autoimmunity.52, 53, 54 Shaping the gut microbiota as well as enhancing gut immune regulation by employing dietary factors presents attractive approaches for modulating autoimmune diabetes and enhancing the efficacy of tolerogenic oral vaccines. Our observations that prolonged oral treatment with low‐dose YBG modulates T‐cell function in the systemic compartment, suppresses insulitis progression and delays T1D incidence suggest that the YBG‐treatment‐induced regulatory innate immune response has a direct suppressive effect on autoimmunity against pancreatic β‐cells. We found that the YBG‐treatment‐associated gut innate immune response involves higher expression of Raldh along with various anti‐ and pro‐inflammatory cytokines. Raldh enzymes, which convert retinol to retinoic acid, are considered as the key tolerogenic factors expressed in gut‐resident DCs. Importantly, the ability of retinoic acid to promote the conversion of antigen‐recognized effector T cells to Treg cells has been well recognized.55, 56, 57, 58 Our observation that the oral co‐administration of YBG, peptide antigen and retinol results in a robust increase in Treg cells in the gut mucosa suggests that tolerogenic self‐antigen presentation in the gut mucosa could be enhanced through a YBG‐induced innate immune response. This notion has been substantiated by the observation that a better protection of NOD mice from insulitis and hyperglycemia can be achieved by oral co‐administration of YBG, peptide antigen and retinol compared with administering individual agents.

Overall, our observations show that the Dectin‐1‐dependent innate immune response induced by YBG in the gut mucosa involves both immune regulatory and pro‐inflammatory factors. We show that, in addition to causing changes in the gut microbial composition that appear to have an impact on gut immune regulation, the pro‐inflammatory and immune regulatory responses triggered by YBG in the gut mucosa can promote both Treg cell and Th17 responses, leading to suppressed Th1‐dominated β‐cell‐antigen‐specific autoimmunity of the systemic compartment and protection of NOD mice from T1D. The Th17 response has been shown to independently, or in combination with the Treg cell response, have a protective effect against β‐cell autoimmunity in NOD mice.4, 15, 59 Most importantly, our study also shows that the YBG‐induced regulatory innate immune response of the gut mucosa can be exploited to modulate the T‐cell response to pancreatic β‐cell‐antigen, for achieving better protection from T1D, highlighting the oral tolerogenic adjuvant value of low‐dose YBG treatment.

Disclosures

Authors do not have any conflict(s) of interest to disclose.

Author contribution

RG researched and analyzed data, and edited the manuscript; NP, BJ, MHS, RB and SKM researched data and reviewed the manuscript; SQ researched data and reviewed/edited the manuscript; and CV designed experiments, researched and analyzed data, and wrote/reviewed/edited the manuscript.

Supporting information

Figure S1. Effect of short‐term oral administration of low dose yeast β‐glucan on small intestine.

Figure S2. Effect of prolonged oral administration of low‐dose yeast β‐glucan on intestine of non‐obese diabetic (a) and B6 (b) mice.

Figure S3. Depletion of gut microbiota using antibiotics.

Figure S4. Short‐term treatment with yeast β‐glucan, β‐cell‐antigen and retinol resulted in a robust modulation of immune response.

Figure S5. Short‐term treatment with yeast β‐glucan, β‐cell‐antigen and retinol resulted in a modulation of immune response.