Diet-Derived Short Chain Fatty Acids Stimulate Intestinal Epithelial Cells To Induce Mucosal Tolerogenic Dendritic Cells

The immune system is influenced by its immediate surroundings, because nutritional intake and dietary changes have large effects on the intestinal and systemic immune system. Studies showed that diet-derived products can influence the immune system within the lamina propria of the intestine in a beneficial manner (1–4). However, dietary intake can also lead to immune-mediated disorders, such as type 2 diabetes and inflammatory bowel disease (1, 4–7). Therefore, the diet in Western countries may underlie immunological disorders, such as food allergies, asthma, and certain autoimmune diseases (8, 9).

The intestines provide the body with essential nutrients by digestion and absorption of dietary products. Microbiota located within the lumen are fundamental to facilitate this process and to allow the uptake of all of the necessary nutrients from the diet. They are able to process diet-derived products in metabolites that may otherwise not be digestible by the host (10, 11). The highest bacterial load in the gastrointestinal tract is found within the colon. Although the number of bacteria is much lower in the small intestines (SIs), it was shown that many species of bacteria are present in the SIs (12–14). In addition, the distribution of microbiota within the intestines can be influenced by the available nutrients in the diet. Many studies, using high-throughput sequencing of gut microbiota, identified specific bacterial taxa that are beneficial for the host or might be associated with immune disorders (7, 13, 15, 16).

Various anaerobic bacteria produce short chain fatty acids (SCFAs), predominantly acetate, butyrate, and propionate, upon metabolism of dietary fiber. The production of these SCFAs is considered a benefit to the host. Although SCFAs are found at higher concentrations in the colon, they are also present in the cecum and SI (12, 14). SCFAs, in particular butyrate, form the major energy source for intestinal epithelial cells (IECs). In addition, SCFAs are described to promote the barrier function of IECs (10, 17). Recently, multiple studies showed the beneficial effect of SCFAs on intestinal regulatory T cells and during intestinal inflammation (18, 19); however, the exact mechanisms of this regulation are not completely understood.

It is necessary that immunological tolerance toward harmless food Ags and commensal flora is established in the intestines, whereas immunity against pathogens is crucial for host protection. Tolerogenic dendritic cells (DCs) are well-described immune cells that interact with other hematopoietic cells to maintain a balanced mucosal immune system (20, 21). These tolerogenic DCs express CD103 and are able to sample Ags in the intestinal lamina propria and migrate to the draining mesenteric lymph node (MLN) to present Ag and to activate lymphocytes. With the production of retinoic acid (RA), the active metabolite of vitamin A, these DCs are crucial for the differentiation of regulatory T cells and promote IgA class-switching of B cells and induce the expression of gut-homing molecules on lymphocytes (20–22). We (23) and other investigators (24–26) showed that these tolerogenic DCs themselves depend on RA, which is produced by IECs, for their function and phenotype. The critical role of retinol metabolism by small intestinal epithelium was shown by the absence of tolerogenic DCs in CRBPII-deficient mice, because CRBP is predominantly expressed in small IECs and is necessary for the uptake of retinol (26).

In this article, we report that fiber-enriched diets promote vitamin A metabolism in epithelial cells in the SIs, as well as enhanced tolerogenic activity in MLN DCs, increased percentages of intestinal regulatory T cells, and higher levels of IgA production in the lumen of the intestine. Furthermore, we demonstrate that SCFAs and the inhibition of histone deacetylase (HDAC) activity are able to induce this vitamin A metabolism in a small IEC line. Strikingly, the immune system is altered upon dietary adjustments, and the composition of the microbiota and SCFA levels within the SI are also affected. Together, these findings reveal a delicate mechanism that safeguards a tolerant mucosal immune system as a result of dietary intake.

Specific pathogen–free C57BL/6 mice (Charles River, Maastricht, the Netherlands) were raised on a standard conventional 2016 Teklad Global rodent diet (Harlan Laboratories) or a synthetic diet that was based on AIN-93M with the use of vitamin-free casein (MP Biomedicals, Solon, OH). Mice were kept under standard animal housing conditions. Wild-type (WT) C57BL/6 mice, aged 8–14 wk, were sacrificed for collection of SIs. The mice fed a conventional diet were raised on a standard Teklad 2016 diet that contains 3.3% of crude fiber and is complemented with a standardized mineral and vitamin mix. Mice fed a synthetic diet received AIN-93M, which contains 4000 IU/kg vitamin A. In another experiment, mice were fed an AIN-93M diet containing 20,000 IU/kg of vitamin A. The Animal Experiments Committee of the VU Medical Center approved all experiments described in this study.

All diet studies were performed at Monash University. Mice were raised on a synthetic AIN-93G diet that contains 3.2% fiber. At the age of 8–10 wk, mice were switched to a diet containing no fiber (0%, SF11-028; Special Feeds) or high levels of fiber (35% enriched for guar gum and cellulose, SF11-029) for 2 wk. Sodium butyrate (Sigma-Aldrich) was administered in the drinking water at 100 mM for 3 wk before mice were sacrificed. All mice were backcrossed onto the C57BL/6 background; GPR43^−/− mice were obtained from Deltagen (San Mateo, CA), and GPR109^−/− mice were obtained from Offermans (Heidelberg, Germany). All experimental procedures were carried out according to protocols approved by the relevant Animal Ethics Committees.

The small IEC line mIC_cl2 was used for in vitro stimulation. Cells were grown in a defined medium that was described earlier (27). Stimulations were performed with sodium butyrate, sodium acetate, sodium propionate (all in a concentration of 1 mM after careful titration) and the HDAC inhibitors trichostatin A (TSA; 300 nm), MS275 (2500 nm), M344 (5 μM), and Droxinostat (50 μM) (all from Sigma-Aldrich). A 30-min preincubation with pertussis toxin (100 ng/ml; Sigma-Aldrich) was performed for the inhibition of GPCR signaling. 4-Chloro-α-(1-methylethyl)-N-2-thiazolyl-benzeneacetamide (4-CMTB) (1 μm; Sigma-Aldrich) was used as a GPR43 agonist. Other dietary- or flora-derived components that were tested and titrated (see Fig. 3 for concentrations) included oleic acid, conjugated linoleic acid (Lipid Nutrition, Wormerveer, the Netherlands), 25-hydroxycholesterol, 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), 3,3′-diindolymethane (DIM), and zymosan (all from Sigma-Aldrich).

SIs were dissected, flushed with PBS, and homogenized in TRIzol Reagent (Invitrogen, Breda, the Netherlands). RNA from mIC_cl2 cells was isolated and cDNA was synthesized as described previously (28). Specific primers for aldh1a1 (forward: 5′-CTCCTCTCACGGCTCTTCA-3′, reverse: 5′-AATGTTTACCACGCCAGGAG-3′) and nlpr3 (forward: 5′-CTGCGGACTGTCCCATCAATG-3′, reverse: 5′-GCAGCTCACCAACCACAGTTTCT-3′) and primers for the housekeeping genes Ubiquitin C (forward: 5′-AGCCCAGTGTTACCACCAAG-3′, reverse: 5′-ACCCAAGAACAAGCACAAGG-3′) and Cyclo (forward: 5′-ACCCATCAAACCATTCCTTCTGTA-3′, reverse: 5′-TGAGGAAAATATGGAACCCAAAGA-3′) (Invitrogen) were used. Quantitative PCR (qPCR) analysis was performed as described earlier (28).

SIs were dissected and opened longitudinally after removal of Peyer’s patches. SIs were treated with EDTA for the removal of epithelial cells and digested to obtain a single-cell suspension, as previously described (21, 28). Subsequently, cell suspensions were immunomagnetically purified for CD45⁺ cells with PE-Cy7–labeled anti-CD45 (clone 30-F11; eBioscience/Immunosource, Halle-Zoersel, Belgium) and the EasySep PE Positive Selection Kit (STEMCELL Technologies, Grenoble, France). Single-cell suspensions were made by cutting MLNs with scissors, followed by digestion at 37°C for 25 min with constant stirring, using 0.5 mg/ml Blendzyme 2 and 0.2 mg/ml DNase I (both from Roche, Penzberg, Germany) in PBS.

Aldehyde dehydrogenase (ALDH) activity in MLN cells was measured using an ALDEFLUOR Kit (STEMCELL Technologies), according to the manufacturer’s protocol. For flow cytometric analysis of ALDEFLUOR-reacted cells, cells were subsequently stained with anti-CD45–PE–Cy7 (clone 30F11), anti-CD11c–PE (clone N418; both from eBioscience), anti-CD103–biotin (clone M290; BD Biosciences, Breda, the Netherlands), MHC class II (clone M5/114, affinity-purified from hybridoma cell culture supernatants), Percp-conjugated (BD Biosciences) streptavidin, and SYTOX Blue (Invitrogen) to discriminate between live and dead cells. In addition, small intestinal cells were stained after CD45 selection with anti-CD3ε–488 (clone 145-2C11), anti-Foxp3–647 (clone FJK-16s), CD4–Percp–Cy5.5 (clone RM4-5; all eBioscience), and LIVE/DEAD Fixable Near-IR (Invitrogen). Staining was performed using a Foxp3 staining set (eBioscience). Cells were analyzed with a Cyan ADP flow cytometer (Beckman Coulter, Mijdrecht, the Netherlands).

Content from the SIs of mice was collected in cold PBS buffer. Debris was removed by cold centrifugation for 20 min at 2000 rpm to harvest the supernatant for analysis of secretory IgA, as described previously (28).

SCFA and lactic acid measurements were performed on the same fecal supernatants collected for IgA analysis, as described above. Acetic, propionic, and butyric acids were quantitatively determined using a Shimadzu- GC2010 (Shimadzu) equipped with a flame ionization detector. The sample (1 μl, split injection of 20×) was injected at 90°C into the column (ZB-FFAB 15 m × 0.53 × 1.0 μm; Phenomenex) using H₂ as carrier gas. After injection of the sample, the oven was heated to 140°C at a rate of 10°C/min, followed by heating to 220°C at 20°C/min, and finally maintained at temperature of 220°C for 1 min. The temperature of the injector was 200°C, and the detector was 250°C. Lactate was determined enzymatically using a d-lactic acid/l-lactic acid test kit (Boehringer Mannheim/R-Biopharm, Darmstadt, Germany).

Small intestinal contents were added to 250 μl of lysis buffer (Agowa, Berlin, Germany), 250 μl zirconium beads (0.1 mm), and 200 μl phenol, followed by cell disruption by bead beating for 2 min. Extracted DNA was amplified and sequenced as described previously (29).

All preprocessing and taxonomic classifications were performed using modules implemented in the mothur software platform (30), as performed previously (31). Unique sequences were aligned using the align.seqs command and the mothur-compatible Bacterial SILVA SEED database. Sequences were classified taxonomically by the RDP-II Naive Bayesian Classifier using a 60% confidence threshold. Community profiles were compared by Weighted Unifrac clustering of OTU abundances (32).

Results are given as the mean ± SD. Statistical analyses were performed using the two-tailed Student t test or for multiple comparison or using one-way or two-way ANOVA for one or two variables, respectively.

Vitamin A is synthesized into its active metabolite RA in two oxidative steps; the second step is irreversible and performed by one of the three isoforms of retinaldehyde dehydrogenase (RALDH) enzymes: RALDH1–3. When we checked for the expression of RALDH enzymes in small intestinal scrapings, which mainly consist of epithelial cells, we observed a high mRNA expression of aldh1a1 (Fig. 1A), the gene coding for RALDH1. These results are in line with previously published data demonstrating that IECs specifically have a high expression of the RALDH1 enzyme, which is needed for the production of RA to induce RALDH2 expression in DCs (21, 23, 24, 26, 33). To address whether RALDH1 expression in IECs is influenced by external factors, we measured aldh1a1 expression in the intestines of mice that were raised on a standard conventional diet or a synthetic AIN-93M diet. Although the precise composition of the synthetic diet is known, the standard conventional diet is composed of natural products that may contain unidentified components. However, the conventional diet is supplemented with the same vitamin and mineral supplements comparable to the levels found in the synthetic diet. SIs were collected and divided into three equal parts from the proximal to distal side to determine whether differences were present along the gut axis. Mice raised on a synthetic diet showed significantly reduced aldh1a1 mRNA expression in the proximal part of the SI compared with mice receiving a standard conventional diet (Fig. 1B). Moreover, ALDH activity in CD103⁺ DCs of the MLN was also significantly reduced in mice that received a synthetic diet (Fig. 1C). The percentages of CD103⁺ versus CD103⁻ DCs, as well as the ALDH activity in CD103⁻ MLN DCs, were not different between the diet groups, pointing to ALDH activity in CD103⁺ DCs as a selective effect of the different diets (Fig. 1D, 1E). Moreover, to demonstrate that ALDH activity of CD103⁺ DCs did not depend on possible differences in dietary vitamin A levels, we tested two levels of vitamin A in the synthetic diet. We raised mice on a control level of vitamin A (4000 IU/kg) or a high level of vitamin A (20,000 IU/kg). As shown in Fig. 1F, we did not observe any difference in ALDH activity in CD103⁺ DCs in mice that received a diet with different vitamin A levels, demonstrating that high levels of dietary vitamin A were not responsible for the differences in ALDH activity in CD103⁺ DCs. Because CD103⁺ DCs are crucial for the induction of regulatory T cells via their production of RA, we examined the percentages of Foxp3⁺ T cells in the SIs of these mice. Indeed, mice on a synthetic diet showed a reduced percentage of regulatory T cells within the CD4⁺ T cell population compared with mice fed conventional chow (Fig. 1G). In addition, when we analyzed the levels of IgA within the content of the SIs, significantly lower IgA concentrations were detected in mice fed the synthetic diet (Fig. 1H). Together, these data demonstrate that mice fed the synthetic AIN-93M diet have reduced expression levels of vitamin A–converting enzymes, as well as fewer intestinal regulatory T cells and diminished production of luminal IgA.

We next investigated whether the mucosal immune system could be influenced by transient changes in the diet. Therefore, mice were raised on a synthetic diet until 10 wk of age, switched to a conventional diet for 3 wk, and compared with mice that were kept on synthetic diet. After the dietary change, mRNA expression of aldh1a1 was increased significantly in the proximal part of SIs from mice that were switched to the conventional diet (Fig. 2A). In addition, ALDH activity of CD103⁺ DCs was enhanced significantly in the MLNs of these mice (Fig. 2B). These data demonstrate that switching to a conventional diet enhances vitamin A metabolism in IECs, leading to more tolerogenic DCs within the intestine.

In a reciprocal type of experiment in which we switched adult mice that were raised on a conventional diet to a synthetic diet, we measured decreased aldh1a1 mRNA expression in the proximal part of the SIs compared with mice that were maintained on the conventional diet (Fig. 2C). In addition, ALDH activity of CD103⁺ MLN DCs was reduced in the switched mice compared with mice that continued to receive a conventional diet (Fig. 2D).

In conclusion, we demonstrate that temporary dietary adjustments can modulate the functionality of the mucosal immune system.

Because our data indicate that the synthetic diet has a reduced capacity to induce a tolerant mucosal immune system, we wanted to determine which dietary components are responsible for the induction of RALDH1 in small IECs. We tested various dietary components that are differentially expressed in the two distinct diets and that might have an effect on vitamin A metabolism in epithelial cells. Therefore, we set up an in vitro system and made use of a mouse small IEC line, mIC_cl2 (27). First, we tested the effect of dietary conjugated linoleic acid, which was shown to affect vitamin A levels within the liver (34). However, conjugated linoleic acid did not affect aldh1a1 mRNA expression compared with its control, oleic acid (Fig. 3A). In addition, incubation of the epithelial cell line with 25-hydroxycholesterol, which can be found in the diet as a result of oxidation of cholesterol, also did not alter aldh1a1 mRNA levels in the epithelial cell line (Fig. 3B).

Furthermore, we tested aryl hydrocarbon receptor (AhR) ligands, which are formed during digestion of cruciferous vegetables, because their importance in mucosal immune regulation and development is well known (2, 35). Although the synthetic AhR ligand TCDD did not exert any effect (Fig. 3C), the natural AhR ligand DIM was able to increase aldh1a1 mRNA expression after 24 h of treatment (Fig. 3D). However, when we analyzed whether the expression of aldh1a1 was changed in the SI of AhR-knockout mice, we did not observe any difference compared with control mice (Fig. 3E), thereby excluding a central role for this ligand–receptor pair in vivo.

To further examine whether fungal-derived ligands, which are present within the intestinal lumen, affect aldh1a1 expression, we incubated the epithelial cell line with zymosan, a β-glucan present in the cell wall of fungi. However, we did not observe any effects of zymosan on aldh1a1 expression (Fig. 3F). To further focus on microorganism-derived products present within the intestines, we used different SCFAs (sodium butyrate, sodium acetate, and sodium propionate), which are produced upon fermentation of diet-derived fibers, in our culture system. Of the different SCFAs tested, only sodium butyrate led to significantly higher mRNA levels of the vitamin A–metabolizing enzyme aldh1a1 upon 24-h stimulation of the epithelial cell line (Fig. 3G).

In sum, of the various molecules tested, sodium butyrate and natural Ahr ligands were able to induce aldh1a1 expression in small IECs in culture.

Because SCFAs are produced upon digestion of fiber by intestinal microorganisms, we examined whether dietary fiber could specifically induce vitamin A metabolism in small IECs in vivo in a controlled experiment in which only dietary fiber content was varied. Therefore, mice were raised on a synthetic diet containing 3.2% fiber that was followed, for 2 wk, by a synthetic diet containing no fiber (0%) or a diet with a high fiber level (35%). As shown previously, high-fiber diets led to the increased production of different SCFAs by microbiota in the gut (18, 36). Analysis of SIs revealed a significantly higher expression of aldh1a1 mRNA in mice receiving a high-fiber diet compared with mice receiving a no-fiber diet (Fig. 4A), demonstrating the ability of dietary fiber to induce vitamin A metabolism in IECs. Furthermore, to confirm that SCFAs are responsible for increased RALDH1 levels when mice received a high-fiber diet, sodium butyrate was supplied in the drinking water for 3 wk. Subsequent analysis revealed that sodium butyrate increased aldh1a1 mRNA expression levels in the proximal part of the SI (Fig. 4B).

To gain more insight in the mechanisms by which SCFAs influence epithelial cells, we performed the same fiber diet experiments with GPR43- and GPR109-knockout mice, because SCFAs can mediate their effects via various GPCRs (8, 37). In the absence of GPR109, aldh1a1 expression still increased upon consumption of high-fiber diet compared with mice that received a diet that lacked fiber, indicating that induction of aldh1a1 expression can occur independently of GPR109 (Fig. 4C). In contrast, aldh1a1 expression was not significantly induced in GPR43-knockout mice upon consumption of a high-fiber diet, suggesting a possible involvement for this receptor (Fig. 4D). However, compared with WT mice (Fig. 4A), expression of aldh1a1 within the SIs of mice receiving a no-fiber diet was higher in GPR43^−/− mice. These data suggest that, in the absence of GPR43, aldh1a1 expression is induced independently of fiber consumption and is likely to be regulated via other mechanisms. In addition, when we stimulated the epithelial cell line with sodium butyrate after it was preincubated with pertussis toxin, which inhibits G_i-coupled GPCR signaling, aldh1a1 mRNA expression could still be induced (Fig. 4E). To further explore the role of GPR43, we directly stimulated mIC_cl2 cells with the GPR43 agonist 4-CMTB. First, to confirm that 4-CMTB stimulation worked, we analyzed expression of nlpr3 as a positive control, because we demonstrated recently that SCFA binding to GPR43 leads to NLPR3 inflammasome activation in IECs (38). Indeed, we observed a significant induction of nlpr3 after 4-CMTB stimulation of the epithelial cells (Fig. 4F). However, when we analyzed the expression of aldh1a1, we did not see any differences after 4-CMTB stimulation, whereas sodium butyrate significantly induced aldh1a1 expression (Fig. 4G).

These results demonstrated that dietary fiber was able to induce the expression of RALDH1 in the intestines, which was not dependent on GPR signaling.

In addition to their ability to signal via GPCRs, SCFAs can act as HDAC inhibitors. Because the DNA is wrapped around histones within the nucleus of the cell, acetyl and deacetyl transferases can manipulate the chromatin conformation, leading to more open chromatin and thus active or silenced gene expression, respectively (39). Therefore, HDAC inhibitors can have a direct effect on the expression of genes that are regulated by the specific HDACs. To test whether SCFAs induce aldh1a1 levels via the inhibition of HDAC, we stimulated mIC_cl2 cells with a general HDAC inhibitor, TSA (Fig. 5A). Indeed, aldh1a1 expression was strongly induced upon culture with TSA. Moreover, no additive effect was found when we stimulated the epithelial cell line with sodium butyrate and TSA at the same time (Fig. 5B). These results indicate that SCFAs induce expression of the vitamin A–converting enzyme RALDH1 in IECs by inhibiting HDAC activity. To further define which HDAC could be involved in regulating RALDH enzyme expression levels, we stimulated the epithelial cell line with a more specific HDAC inhibitor, MS275, which inhibits HDAC1 and HDAC3. After stimulation with this specific HDAC inhibitor, we observed a significant induction of aldh1a1 expression in the epithelial cells (Fig. 5A). These data implicate a role for HDAC1 or HDAC3 in controlling vitamin A metabolism in IECs. However, when we stimulated the epithelial cell line with Droxinostat, which specifically inhibits HDAC3, HDAC6, and HDAC8, we observed only a slight increase in aldh1a1 expression. Stimulation of epithelial cells with M344, which specifically inhibits HDAC1 and HDAC6, resulted in a much higher expression of aldh1a1, similar to the effects of TSA or MS275. These data imply that HDAC1 is involved in controlling aldh1a1 expression in IECs. As a control, we measured expression of the actin-binding protein villin, which is highly expressed in epithelial cells. When epithelial cells were stimulated with sodium butyrate or one of the HDAC inhibitors, we did not observe any differences in villin expression (Fig. 5C). These results demonstrate that HDAC inhibition does not generally increase gene expression and that the induction of aldh1a1 expression after HDAC inhibition is specific.

In sum, our results indicate that dietary fiber is able to induce vitamin A metabolism in small IECs via the inhibition of HDAC.

Because microorganisms present in the intestines are critical for the production of SCFAs, we analyzed the composition of the microbiota within the SIs of mice that were raised on a conventional or a synthetic diet. To determine microbial diversity in these samples, a barcoded amplicon pyrosequencing method was used (31). Indeed, clear differences could be observed between the two diet groups (Fig. 6A). Sequences affiliated with the butyrate-producing family Lachnospiraceae (phylum Firmicutes, order Clostridiales) were less dominant in the mice fed a synthetic diet (Fig. 6B). Similarly, sequences affiliated with the genus Lactobacillus (phylum Firmicutes, order Lactobacillales) were more dominant in the mice fed a conventional diet. This lactate-producing genus was shown to indirectly stimulate bacterial butyrate production (40, 41). In addition, the Bacteroidetes phylum, of which some members are able to produce SCFAs themselves (40, 42), was reduced in mice fed a synthetic diet (Fig. 6B). Therefore, these bacteria species could be involved in the production of SCFAs in mice receiving a conventional diet. When two isomers of lactic acid and SCFAs were measured in the SI, significantly lower levels were detected in mice fed a synthetic diet versus a conventional diet (Fig. 6C, 6D). d-Lactic acid and propionic acid were even below the detection limit in the content of the SI of mice fed a synthetic diet (Fig. 6C, 6D). Additionally, an increased level of Erysipelotrichia was observed in mice that received a synthetic diet (Fig. 6B). These data are in agreement with earlier studies in which low-fiber diets led to an increase in Firmicutes, mainly Erysipelotrichia (36). Furthermore, the synthetic diet led to a clear increase in the mucus-degrading bacterial genus Akkermansia (phylum Verrucomicrobia).

In conclusion, we showed that the microbiota composition in the SI changed upon dietary adjustments. These microbial alterations affected the production of SCFAs by the microbiota in the SIs, having an effect on vitamin A metabolism within IECs.

Although the importance of tolerogenic CD103⁺ DCs for the maintenance of mucosal homeostasis is well established, the exact mechanism by which the differentiation of these cells is regulated is not clear. It is described that the phenotype of these DCs depends on the production of RA by epithelial cells (23, 24, 26); however, it is not understood how the production of RA by epithelial cells themselves is controlled. Although we saw in previous studies that ALDH activity of CD103⁺ DCs is ablated in mice that are deficient for vitamin A, aldh1a1 mRNA expression in the SIs did not depend on the intake of vitamin A (21). In this study, we demonstrated that additional dietary compounds are crucial for the regulation of vitamin A metabolism in IECs, which leads to increased ALDH activity in CD103⁺ MLN DCs and a concomitant increase in regulatory T cells within the intestines and production of luminal IgA. The responsible dietary compound is fiber, which gives rise to SCFAs upon digestion by the microflora. Moreover, SCFAs, especially butyrate, were able to induce vitamin A metabolism in the small IEC line mIC_cl2. Our data confirmed results from an earlier study in which mice that received a diet supplemented with cycloinulooligosaccharides had increased levels of SCFAs and higher concentrations of IgA measured in the content of the SI (43). Recently, we also demonstrated the role of SCFAs in a model of food allergy and found that SCFAs, specifically acetate and butyrate, added to the drinking water increased tolerogenic DCs and tolerogenic T cells in the MLN (44). Furthermore, it was shown that SCFAs, mainly acetate and butyrate, were enhanced in the cecum and colon of mice that received a high-fiber diet (18, 36). We observed increased vitamin A metabolism in mice that received a high-fiber diet. By focusing on the effects within the SIs, we could determine changes in the microbiota and SCFA levels that impacted the epithelial cells and the mucosal immune system. In line with our data is the report that stromal cells in the lamina propria of the SIs can also produce RA, which is dependent on the presence of microbiota in the gut. Like IECs, these RA-producing stromal cells are in close contact with CD103⁺ DCs. Together, these data suggest that, in addition to IECs, stromal cells might play a role in the induction of tolerogenic DCs (45). However, it remains to be seen whether these stromal cells are imprinted directed by luminal contents or whether they depend on IEC-derived factors. However, the importance of the induction of these tolerogenic DCs was further highlighted by Magnusson et al. (46) recently; they found that patients with ulcerative colitis have reduced Aldh1⁺ cells (macrophages and DCs) in the colon, regardless of their inflammation status.

Although our data from GPR43- and GPR109-knockout mice did not demonstrate a central role for the induction of vitamin A metabolism upon high fiber intake, based on these data we cannot fully exclude a role for GPCRs in the induction of vitamin A metabolism within IECs. Various GPCR receptors are able to bind the various SCFAs with different affinity (8, 37). Therefore, the lack of only one receptor might not lead to differences in SCFA signaling because of receptor redundancy. To rule out the role of GPCR receptors, triple-knockout mice for the main GPCRs that bind SCFA (GPR41, GPR43, and GPR109) may need to be analyzed. However, when we inhibited G protein signaling during sodium butyrate stimulation in vitro, induction of aldh1a1 expression still occurred, showing no role for G protein–mediated signaling in controlling vitamin A metabolism in IECs. Furthermore, when we stimulated the IEC line with a specific GPR43 agonist, no induction of aldh1a1 expression was observed.

In addition to binding to GPCRs, SCFAs can act as HDAC inhibitors. We were able to show in vitro that inhibition of HDAC activity resulted in increased expression of aldh1a1, and our data implicated HDAC 1 in the control of aldh1a1 expression levels in IECs. In line with our results, it was demonstrated recently that other IEC lines and organoids induce their aldh activity after SCFA stimulation (47).

Moreover, we demonstrated that differences in diet resulted in alterations in the composition of the microbiota. Whether these changes directly or indirectly affect the mucosal immune system cannot be stated at this point; however, we showed that SCFA-producing taxa were reduced in the SI of mice that received a synthetic diet. These bacterial changes led to lower levels of SCFAs compared with conventionally fed mice.

In addition, it was shown that obese humans and mice have an altered gut microbiota, with decreased levels of Bacteroidia bacteria (16, 48); these results were similar to our mice that received a synthetic diet. Furthermore, we demonstrated that bacteria of the genus Erysipelotrichia (phylum Firmicutes) were enhanced significantly in mice that received a synthetic diet, correlating with mice that received a high-fat diet or mice with colorectal cancer (16, 49). Moreover, previous studies demonstrated that microbiota diversity is diminished when a low-fiber diet is consumed. Additionally, they showed that microbiota in mice fed a low-fiber diet was dominated by the phylum Firmicutes, mainly Erysipelotrichia, whereas the levels of Bacteroidia were reduced, resembling our results in mice fed a synthetic diet and that most likely consumed less fiber than mice fed a conventional diet (36). Furthermore, in mice fed a synthetic diet, the levels of Akkermansia (phylum Verrucomicrobia), which can be found in the inner mucous layer where they degrade mucus and, thereby, stimulate mucus production by goblet cells (50), were significantly lower. Moreover, the intestinal mucous layer is altered in a number of intestinal diseases linked to a deficiency in dietary fiber.

A balanced symbiotic relationship between host and microbiota exists during intestinal homeostasis; however, gut dysbiosis can lead to a variety of diseases (13). The best-known therapeutic approaches to restore a healthy gut microflora are antibiotics, prebiotics, and probiotics, (51, 52). Particular combinations of bacteria might be needed to successfully restore symbiosis (15, 51, 53).

In conclusion, we showed that dietary factors are able to induce aldh1a1 expression in IECs. Subsequently, these epithelial cells are able to induce tolerogenic CD103⁺ DCs through their production of RA. In turn, these DCs produce RA themselves and, thereby, are able to induce regulatory T cells and IgA-producing B cells. Furthermore, SCFAs are able to induce aldh1a1 expression in IECs through inhibition of HDAC. The availability of substrates determines the composition of the microbiota and the production of SCFAs, because these are produced upon digestion of the substrates by the microbiota. These findings reveal a delicate mechanism that controls the mucosal immune system and that can have important implications for the prevention and treatment of immune-mediated chronic diseases.

We thank C. Prins, H. van der Laan, and E. van Gelderop (VU University, Amsterdam, the Netherlands) for animal care.

The authors have no financial conflicts of interest.