Article Text


Original article
Microbiota fermentation-NLRP3 axis shapes the impact of dietary fibres on intestinal inflammation
  1. Vishal Singh1,
  2. Beng San Yeoh2,
  3. Rachel E Walker3,
  4. Xia Xiao3,
  5. Piu Saha1,
  6. Rachel M Golonka1,
  7. Jingwei Cai4,
  8. Alexis Charles Andre Bretin5,
  9. Xi Cheng1,
  10. Qing Liu4,
  11. Michael D Flythe6,
  12. Benoit Chassaing5,7,
  13. Gregory C Shearer3,
  14. Andrew D Patterson4,
  15. Andrew T Gewirtz5,
  16. Matam Vijay-Kumar1,8
  1. 1Department of Physiology and Pharmacology, College of Medicine and Life Sciences, University of Toledo, Toledo, Ohio, USA
  2. 2Nutritional Sciences, Graduate Program in Immunology and Infectious Diseases, Pennsylvania State University, University Park, Pennsylvania, USA
  3. 3Department of Nutritional Sciences, Pennsylvania State University, University Park, Pennsylvania, USA
  4. 4Department of Veterinary and Biomedical Sciences, Pennsylvania State University, University Park, Pennsylvania, USA
  5. 5Center for Inflammation Immunity and Infection, Institute for Biomedical Sciences, Georgia State University, Atlanta, Georgia, USA
  6. 6USDA-Agriculture Research Service, University of Kentucky Campus, Lexington, Kentucky, USA
  7. 7Neuroscience Institute, Institutefor Biomedical Sciences , Georgia State University, Atlanta, Georgia, USA
  8. 8Department of Medical Microbiology and Immunology, University of Toledo, Toledo, Ohio, USA
  1. Correspondence to Dr Matam Vijay-Kumar, Department of Physiology and Pharmacology, College of Medicine and Life Sciences, University of Toledo, Toledo, OH 43614, USA; MatamVijay.Kumar{at}


Objective Diets rich in fermentable fibres provide an array of health benefits; however, many patients with IBD report poor tolerance to fermentable fibre-rich foods. Intervention studies with dietary fibres in murine models of colonic inflammation have yielded conflicting results on whether fibres ameliorate or exacerbate IBD. Herein, we examined how replacing the insoluble fibre, cellulose, with the fermentable fibres, inulin or pectin, impacted murine colitis resulting from immune dysregulation via inhibition of interleukin (IL)-10 signalling and/or innate immune deficiency (Tlr5KO).

Design Mice were fed with diet containing either cellulose, inulin or pectin and subjected to weekly injections of an IL-10 receptor (αIL-10R) neutralising antibody. Colitis development was examined by serological, biochemical, histological and immunological parameters.

Results Inulin potentiated the severity of αIL10R-induced colitis, while pectin ameliorated the disease. Such exacerbation of colitis following inulin feeding was associated with enrichment of butyrate-producing bacteria and elevated levels of caecal butyrate. Blockade of butyrate production by either metronidazole or hops β-acids ameliorated colitis severity in inulin-fed mice, whereas augmenting caecal butyrate via tributyrin increased colitis severity in cellulose containing diet-fed mice. Elevated butyrate levels were associated with increased IL-1β activity, while inhibition of the NOD-like receptor protein 3 by genetic, pharmacologic or dietary means markedly reduced colitis.

Conclusion These results not only support the notion that fermentable fibres have the potential to ameliorate colitis but also caution that, in some contexts, prebiotic fibres can lead to gut dysbiosis and surfeit colonic butyrate that might exacerbate IBD.

  • inflammatory bowel disease
  • chronic ulcerative colitis
View Full Text

Statistics from

Significance of this study

What is already known on this subject?

  • Fermentable fibres, including inulin, are metabolised into short-chain fatty acids (SCFA) in the large intestine. SCFA are generally reported to have an array of beneficial effects including improvement of gut health.

  • However, diets rich in certain types of fibres worsen disease symptoms in patients with IBD. Animal models similarly revealed both positive and negative health effects of dietary fibres.

  • For example, fibre fermentation-derived SCFA improves gut barrier function but worsens both colitis and colorectal cancer in disease-prone mice.

What are the new findings?

  • Dietary pectin, but not inulin, attenuated colonic inflammation in an immune-mediated mouse colitis model.

  • Surfeit butyrate and elevated Proteobacteria are associated with aggravated colitis in inulin-fed mice.

  • Suppressing butyrate production mitigated colonic inflammation, whereas elevating butyrate levels worsened colitis.

  • Targeting NOD-like receptor protein 3 via genetic, pharmacologic or dietary means reduced the severity of colitis in inulin-fed mice.

Significance of this study

How might it impact on clinical practice in the foreseeable future?

  • Dietary fermentable fibres have many beneficial effects on health but, at least in some cases, may exacerbate intestinal inflammation. Hence, patients with IBD often refrain from consumption of healthy foods rich in these fibres.

  • Our results support the notion that fermentable fibres indeed promote gut health and can dampen inflammation; however, consumption of certain fermentable fibres such as inulin can result in increased total gut bacterial load and elevated levels of butyrate, which collectively resulted in exacerbation of immune-mediated colitis.

  • These findings caution against indiscriminately enriching processed foods with readily fermentable purified fibres and, rather, suggests a strategy to balance consumption of various fibres so as to avoid the risk of flaring in patients with IBD.


Societal changes in diet, including increased consumption of foods rich in simple sugars and fat, have been suggested to contribute to the increased prevalence of chronic inflammatory diseases in developed countries along with low-income and middle-income countries. Conversely, reduced consumption of dietary fibres found naturally in fruits and vegetables may have resulted in a general loss of gut health. Dietary fibres can be broadly categorised as insoluble (eg, cellulose) or soluble (eg, inulin), where the latter is readily fermentable by gut bacteria, thus, generating large quantities of short-chain fatty acids (SCFA; eg, acetate, butyrate and propionate). In addition to differences in solubility, these fibres have distinct chemical structures, for example, cellulose (β-(1→4)-linked poly-D-glucose), inulin (β-(2→1)-linked poly-fructosan) and pectin (α-(1→4)-linked poly-D-galacturonate). Such disparities are likely to impact their fermentability, capacity to nourish distinct groups of bacteria and production of distinct levels of SCFA. Indeed, accumulating evidence suggests that not all soluble fibres exert similar physiological effects; moreover, it is also recognised that purified fibres used in processed foods may be functionally different than their naturally occurring counterparts.

Inflammatory bowel disease (IBD), encompassing ulcerative colitis (UC) and Crohn’s disease (CD) are chronic inflammatory diseases of the intestine, whose pathology is influenced by host genetics, diet and the gut microbiota. Genome-wide association studies have identified approximately 200 susceptibility loci that are associated with IBD, including genes related to interleukin-10 (IL-10) function.1 2 Indeed, we and others have shown that the administration of IL-10R neutralising monoclonal antibody (αIL-10R mAb) to susceptible strains of mice induced robust chronic colitis with features reminiscent of human IBD.3–6 Such αIL-10R-induced colitis shares the clinical relevance of the IL-10−/− mouse model, but with the tractability to induce colitis in other strains of mice.

The NOD-like receptor (NLR) family member NLRP3 is a cytosolic protein complex responsible for the proteolytic maturation and secretion of the proinflammatory cytokines IL-1β and IL-18. Although NLRP3 has been highly implicated in IBD, studies to date have yielded conflicting evidence on whether its deficiency ameliorates7 or worsens8–10 chemical-induced colitis. In contrast, in states of IL-10 deficiency, NLRP3 activity promotes colitis.11 12 Such inconsistencies observed in animal studies are reminiscent of observations in patients with IBD, where the role of NLRP3 activation is unclear in UC,13 but is thought to be detrimental in CD, including those associated with a null mutation in the IL-10R.11 14 Herein, we employed an experimental model (ie, αIL-10R-induced colitis) that could be used to study the influence of dietary fibres in multiple genetic contexts, including mice that are dysbiotic due to toll-loke receptor 5 (TLR5) deficiency.

To counteract inflammasome activity, the host constitutively secretes endogenous inhibitors, such as secretory IL-1 receptor antagonist (IL-1Ra) and IL-18 binding protein (IL-18BP). Recently, it has been shown that β-hydroxybutyrate (BHB; a ketone body) is a potent inhibitor of NLRP3.15 16 The level of circulating BHB is known to increase during prolonged fasting, thus, raising the prospect that blunting of NLRP3 activity by BHB may potentially explain, in part, how calorie restriction (CR) and fasting ameliorates inflammatory disorders. Intriguingly, BHB is a structural analogue of butyrate. Yet, in contrast to BHB, butyrate is thought to induce the activation of NLRP3, although this notion remains controversial.15–19

A range of studies have shown that pharmacological intervention with butyrate can ameliorate intestinal inflammation20–24; however, in some studies, administration of butyrate either made no difference or even potentiated the colitis.24–27 These equivocal outcomes are not well understood and are possibly influenced by differences in SCFA administration and/or underlying aetiology of intestinal inflammation. Likewise, the extent to which fermentable fibres and SCFA generated therefrom could impact chronic intestinal inflammation remains poorly defined, in part, due to the lack of approaches to prevent SCFA production. However, we have recently demonstrated that the use of hops β-acids to block fermentation provides a means of surmounting this hurdle.28

Hence, the goal of this study was to test the hypothesis that dietary intervention with soluble fibres, by virtue of their ability to serve as a precursor for SCFA, may be more beneficial than non-fermentable cellulose during IBD. In accord with this hypothesis, we report that the dietary soluble fibre, pectin, improved colonic inflammation in the immune hyperactivation-induced colitis model. Whereas, inulin exacerbated colitis likely due to this fermentable fibre being largely metabolised to butyrate and being associated to promote expansion of Proteobacteria. Taken together, our findings collectively demonstrate that fermentable fibres have a unique impact on microbiota fermentation that dictates gut inflammation.


All experimental methods including mice, diets and induction of acute and chronic colitis are described in the online supplementary material file.


Dietary pectin attenuates acute colonic inflammation

To examine the effect of dietary fibres on the early events of intestinal inflammation, mice were fed ‘open-source’ purified diets (table 1) containing 10% (by weight) cellulose (CCD) or an isocaloric diet where three-fourths of the cellulose was replaced with either inulin (ICD) or pectin (PCD). On day 3, mice were administered αIL-10R mAb and euthanised 1 week later. The 1-week neutralisation of IL-10 did not result in a significant loss in body weight or impact gross appearance of the colon (online supplementary figure S1A–C). Nonetheless, this treatment resulted in splenomegaly and elevated levels of the inflammatory markers serum amyloid A (SAA) and lipocalin 2 (Lcn2, both in serum and faeces). The extent of this increase was potentiated in ICD-fed mice, but ameliorated in PCD-fed mice, when compared with CCD-fed mice (online supplementary figure S1D–G).

Table 1

Isocaloric diet composition (Research Diets Inc.)

The inflammasome cytokine, IL-1β, and its endogenous inhibitor, IL-1Ra, are key determinants of colitis in both humans and mice. In healthy subjects, the ratio of IL-1β to IL-1Ra is close to one29; moreover, an increase in this ratio, due to either decreased IL-1Ra or increased IL-1β, heightens IL-1β bioactivity.30 31 In the present study, quantitation of these cytokines revealed a similar pattern in that induction of IL-1β activity, which was largely driven by changes in levels of IL-1β, was augmented by ICD and abrogated by PCD (online supplementary figure S1H–J). Interestingly, PCD-fed mice exhibited elevated levels of colonic IL-1Ra (online supplementary figure S1I). Together, these results indicate that the moderate degree of colonic inflammation induced by IL-10R neutralisation was exacerbated in ICD while ameliorated in PCD-fed group.

Dietary pectin, but not inulin, ameliorates chronic colitis

To examine how dietary fibres might impact chronic colitis, mice consuming CCD, ICD or PCD were subjected to 4 weekly injections of αIL-10R mAb. Prolonged IL-10 neutralisation resulted in robust chronic colitis in mice fed CCD that was further potentiated in ICD-fed mice as evident by loss in body weight, emptied and shrunken caeca, enlarged spleen and thickened colon (figure 1A–D). Histological analysis revealed that αIL-10R-treated CCD-fed and ICD-fed mice displayed typical features of chronic colitis: colonic wall thickening, distorted crypt structure, crypt elongation, reduced colonic mucin and goblet cells and infiltration of immune cells (macrophages and neutrophils) in the colonic mucosa (figure 1E–F). Intriguingly, the chronic colitis was abrogated in PCD-fed mice as both serological and histological markers were markedly reduced in this group than CCD-fed mice (figure 1E–F). The elevation of inflammatory markers was also potentiated and ameliorated, respectively, in ICD-fed and PCD-fed mice (figure 1G–J). Increased IL-1β activity was observed in ICD-fed mice while being markedly attenuated in PCD-fed mice (figure 1K–M).

Figure 1

Pectin, but not inulin, prevents chronic colitis. Four-week-old male WT mice were initially maintained on lab chow for 1 week and then switched to purified diets: CCD, ICD or PCD. On day 3, mice were given αIL-10R mAb or isotype control (1.0 mg/mouse, four weekly injections; intraperitoneally) and body weight was monitored weekly. One week post-final injection, mice were euthanised and analysed for colitis parameters. (A) Bar graph shows body weight. (B) Image displays the gross colon appearance. (C–D) Bar graphs represent per cent, (C) spleen weight and (D) colon weight. (E) Colons (proximal to distal portion) were opened longitudinally, Swiss rolled, fixed with 10% neutral buffered formalin, embedded in paraffin, sectioned and used for histochemical and immunohistochemical staining: (1) H&E, (2) Alcian blue for goblet cells, (3) F4/80 for macrophages and (4) Ly6G for neutrophils. (F) Colonic histology score, assessed by visualising the entire H&E stained colon sections microscopically for the extent of immune cell infiltration in mucosa and submucosa, epithelial hyperplasia, goblet cell loss, distorted crypt structure, ulcerations and crypt loss. Bar graphs show serum concentrations of (G) SAA and (H) Lcn2; (I) colonic MPO activity and (J) faecal Lcn2. Proximal colon was used for cytokine analysis. Bar graphs show colonic (K) IL-1β, (L) IL-1Ra and (M) IL-1β/IL-1Ra ratio. The data (mean±SEM) is representative of two independent experiments (*p<0.05). CCD, cellulose containing diet; ICD, inulin containing diet; IL, interleukin; Lcn2, lipocalin 2; MPO, myeloperoxidase; PCD, pectin containing diet; SAA, serum amyloid A; WT, wild type.

Fructo-oligosaccharides (FOS, aka oligofructose, an inulin-type fructan) are mainly produced by partial enzymatic hydrolysis of inulin. Together, inulin and FOS are among the most widely used prebiotic fibres in the food industry. Therefore, we next analysed the effects of FOS-containing diet (FCD) on colitis development. All standard colitis parameters analysed in αIL-10R treated, FCD-fed mice were indeed comparable to the αIL-10R given ICD-fed mice (online supplementary figure S2A–I). Taken together, these results suggest that, at least in the immune dysregulation model of colitis, some fibres exacerbate disease, while some are protective.

Dietary pectin mitigates gut inflammation in colitis-prone TLR5-deficient mice

One general category associated with an increased development of IBD is genetic polymorphisms that result in innate immune deficiencies. In accord with this notion, mice lacking the receptor for flagellin, TLR5, are prone to develop colitis.4 32 Hence, we reasoned that the inclusion of αIL-10R treated Tlr5KO mice would allow us to better discern the procolitogenic and anticolitogenic effects of dietary fibres in the context of severe colitis. Indeed, unlike the relatively modest colitis seen in wild-type (WT) mice, a single injection of αIL-10R induced noticeable colitis in CCD-fed Tlr5KO mice as evident by the thickened colon, splenomegaly and elevated systemic and colonic markers of inflammation (online supplementary figure S3A–J). Analogous to results described above in WT mice, the extent of such colitis in Tlr5KO mice was exacerbated by ICD and markedly ameliorated by PCD (online supplementary figure S3A–J).

In the chronic colitis model, which was induced by 3 weekly injections of αIL-10R, CCD-fed, ICD-fed or PCD-fed mice, showed lower body weight than isotype control-treated mice (online supplementary figure S4A). Macroscopic examination revealed emptied and inflamed caeca, the presence of occult blood in the colon, along with splenomegaly and colomegaly, specifically in ICD-fed Tlr5KO mice (online supplementary figure S4B–D). Correspondingly, histological analysis showed loss of crypt structure and goblet cells and enormous infiltration, typically transmural, of inflammatory cells extending through the mucosal layer to underlying submucosa with marked hyperplasia particularly in ICD-fed Tlr5KO mice (online supplementary figure S4E–F). Additionally, markers of both systemic and colonic inflammation were strikingly elevated in CCD-fed and ICD-fed mice (online supplementary figure S4G–J). The colonic IL-1β/IL-1Ra ratio was markedly elevated in CCD-fed and ICD-fed Tlr5KO mice (online supplementary figure S4K–J). Concomitantly, PCD-fed Tlr5KO mice exhibited reduced spleen weight and colon thickening and lower levels of proinflammatory cytokines when compared with CCD-fed mice. Histological analysis of colonic tissue and cytokine analysis further confirmed that colitis was markedly reduced in αIL-10R treated PCD-fed Tlr5KO mice (online supplementary figure S4A–M).

Fibres with distinct structures differentially impact the composition of gut microbiota and its metabolites

Considering the differential impacts of inulin and pectin on colitis severity, we next investigated whether changes in microbiota composition may underlie or be associated with this disparity. In accord with the notion that fermentable fibres nourish microbiota, replacing the insoluble fibre, cellulose, with inulin elevated the total faecal bacterial loads (figure 2A). Notably, while inulin enhanced levels of γ-Proteobacteria, which are well linked to inflammation, this class was decreased in abundance both before and following αIL-10R in PCD-fed mice (figure 2B). Acknowledging the notion that γ-Proteobacteria promote and are promoted by inflammation, levels of γ-Proteobacteria were correlated with the inflammatory marker faecal Lcn2 (figure 2C). Bacteria that are known to readily metabolise fibre into SCFA, including Clostridia cluster XIVa, Lachnospiraceae and Ruminococcaceae,33 were preferentially enhanced by ICD (figure 2D–E). Such increases in butyrogenic bacteria following ICD, but not PCD feeding, was correlated with expression of butyryl-CoA:acetate CoA-transferase (BCoAT), a key bacterial enzyme involved in butyrate synthesis (figure 2F). While replacing cellulose with either fermentable fibre resulted in increased levels of caecal SCFA, PCD preferentially enhanced acetate, whereas ICD preferentially increased butyrate (figure 2G–I).

Figure 2

ICD-fed mice display enrichment of γ-Proteobacteria and butyrate producers as well as gut barrier dysfunction and inflammation. Faecal bacterial DNA was isolated from αIL-10R mAb treated (single injection) CCD, ICD or PCD-fed WT mice (n=6) and subjected to quantitative real-time PCR for gut bacterial analysis. Caecal contents were collected for short-chain fatty acid quantification via gas chromatography–mass spectrometry. Bar graphs represent relative level of faecal (A) total bacterial load and (B) γ-Proteobacteria. (C) A correlation was analysed between the faecal Lcn2 and relative abundance of γ-Proteobacteria (Pearson correlation coefficient r=0.85, p (two-tailed)<0.0001)). Relative abundance of (D) Clostridium cluster XIVa, (E) Lachnospiraceae and Ruminococcaceae and (F) butyryl-CoA:acetate CoA-transferase (BCoAT) gene expression. (G–I) Caecal contents were analysed for short-chain fatty acids. Bar graphs display the level of (G) acetate, (H) propionate and (I) butyrate levels. Serum samples obtained from chronic colitis group were analysed for LPS-specific and flagellin-specific immunoglobulin (Ig) G. Bar graphs represent serum immunoreactivity to (J) LPS and (K) flagellin. The data (mean±SEM) is representative of two independent experiments (*p<0.05). CCD, cellulose containing diet; ICD, inulin containing diet; IL, interleukin; Lcn2, lipocalin 2; LPS, lipopolysaccharide; mAb, monoclonal antibody; PCD, pectin containing diet; WT, wild type.

Defects in intestinal barrier are frequently observed during IBD. Accordingly, serum immunoreactivity to bacterial products, LPS and flagellin—markers of colonic barrier permeability—were increased in both αIL-10R-treated (4-weekly injections) CCD-fed and ICD-fed mice (figure 2J–K). Importantly, the systemic immunoreactivity to bacterial products were relatively lower in αIL-10R-treated PCD-fed mice (figure 2J–K), suggesting that their intestinal barrier function was relatively less compromised than the ICD-fed group. Comprehensive analysis of gut microbiota during chronic colitis, via 16S sequencing, revealed that faecal microbial communities in CCD-fed, ICD-fed and PCD-fed mice clustered separately from each other (online supplementary figure S5A). Such clustering was seen across multiple cages per condition, indicating it was a consequence of diet and inflammation rather than a cage clustering artefact. Linear discriminant analysis effect size revealed that relative abundance of Firmicutes was increased in PCD-fed mice when compared with ICD-fed mice, but not CCD-fed mice (online supplementary figure S5B–D). Surprisingly, we observed a marked decrease in the relative abundance of Verrucomicrobia, which encompasses Akkermansia muciniphila, a mucus-degrading, gut health-promoting bacterium, when compared with both CCD-fed and ICD-fed mice (online supplementary figure S5B–D). As expected, Bifidobacterium was enriched in ICD-fed mice when compared with CCD-fed and PCD-fed mice. Interestingly, in contrast to the quantitative PCR (qPCR)-based microbiota analysis performed during early stage of disease, we did not observe expansion of γ-Proteobacteria on ICD feeding. This is likely due to differences in the extent of colonic inflammation between early and late stages of disease and differences in analytical approach, for example, qPCR can pick up absolute differences while sequencing can only detect relative increases in phyla/taxa.

Suppressing intestinal butyrate production alleviates chronic colitis

While many of the beneficial effects of fibre are ascribed to SCFA, recent findings show that SCFA can also have adverse effects depending on host genetics and physiological state. For instance, butyrate was shown to inhibit the proliferation of colonic epithelial stem/progenitor cells and impair mucosal healing.25 Furthermore, on colonic infusion, butyrate aggravated the colitis in dextran sodium sulfate (DSS)-treated mice.25 More importantly, patients with IBD have been reported to develop intestinal side effects to dietary inulin at doses well tolerated by >90% of healthy subjects.34–38 Therefore, to identify whether high butyrate levels in ICD-fed mice contributes to their colonic inflammation, we treated mice with metronidazole, which preferentially depleted butyrate producers and increased Lactobacillus species, without altering the total gut bacterial load (online supplementary figure S6A–E). Loss of butyrate producers in the gut correlated with reduced expression of BCoAT (online supplementary figure S6F). Butyrate levels were almost undetectable in the caecal and faecal contents (online supplementary figure S6G and S6J), whereas the levels of acetate and propionate were moderately altered on metronidazole treatment (online supplementary figure S6H–I and S6K–L), supporting the notion that metronidazole depletes the butyrate producers in the gut.25 Metronidazole treatment substantially reduced colonic inflammation in the ICD-fed group (figure 3A–I), suggesting that elevated butyrate, specifically in the inflamed colon, may have fueled inflammation in the ICD-fed mice.

Figure 3

Suppressing caecal butyrate production mitigated colonic inflammation. Four-week-old male C57BL6 (WT) and T5KO mice were maintained on metronidazole (1 g/L in drinking water) and after 3 days switched to ICD. The mice were challenged with αIL-10R mAb (1 mg/mouse, four weekly injections). Mice were euthanised 1 week after the fourth injection. (A) Body weight. (B) Gross colon appearance. (C) Histochemical staining of colons: (1) H&E and (2) Alcian blue for goblet cells. (D) Colonic histology score. Per cent (E) spleen weight and (F) colon weight. Serum level of (G) SAA and (H) Lcn2. (I) Lcn2 level in faeces. In another set of experiments both WT (male, n=7 in each group) and T5KO (male, n=5 in each group) mice were fed with either vehicle (propylene glycol, food grade) or β-acids from Hops (Humulus lupulus, 20 ppm) in drinking water. (J) Butyrate level in the caecal contents. (K) Body weight. (L) Gross images of colon. (M) H&E and alcian blue-stained colonic sections. (N) Colonic histology score. Per cent (O) spleen weight and (P) colon weight. Serum level of (Q) SAA and (R) Lcn2. (S) Lcn2 level in faeces. The data represented as mean±SEM (*p<0.05). CCD, cellulose containing diet; ICD, inulin containing diet; IL, interleukin; Lcn2, lipocalin 2; mAb, monoclonal antibody; PCD, pectin containing diet ; T5KO, toll-like receptor 5 KO; WT, wild type.

To further investigate whether butyrate might be promoting inflammation in ICD-fed mice, we sought a means to block SCFA generation in the gut without having a major impact on bacterial loads and/or composition. We administered hop β-acids, which has been shown to suppress fermentation and thus SCFA production.28 We found that hop β-acids (20 ppm, supplemented in drinking water) suppressed caecal butyrate (figure 3J) and colonic inflammation in the ICD-fed mice when compared with the vehicle-treated group (figure 3K–S). While β-acids are not reported to directly impact bacterial growth, we reasoned that the suppression of fermentation might favour/suppress some species and thus impact overall community structure. In accord with this notion, 16S analysis revealed that microbial communities in the vehicle-acid and β-acid groups clustered separately (online supplementary figure S7A). Differences in composition were driven by reduction in Actinobacteria and Deferribacteres (50%), Firmicutes (~17%), Proteobacteria (~33%) and disappearance of Cyanobacteria and Tenericutes were observed in the β-acid group when compared with the control group. Significant reduction of major butyrate producers, including Clostridiaceae, Lachnospiraceae and Ruminococcaceae were also observed in the β-acid treated group (online supplementary figure S7B–D). However, we are not able, at present, to discern the relative contribution of such compositional changes versus diminished butyrate production in β-acid-induced reduction in colitis severity.

Next, we investigated the extent to which elevated caecal butyrate contributed to the exacerbation of intestinal inflammation in ICD-fed mice. Mice were fed CCD, CCD plus tributyrin (TBT, 5 g/kg body weight; administered orally on every third day for a total period of 21 days) or ICD to WT mice. Mice were then left untreated or subjected to 3 weekly injections of αIL-10R mAb. On day 21, the mice were euthanised and analysed for various colitis markers. Oral administration of TBT markedly elevated caecal butyrate levels in both control and αIL-10R treated CCD-fed mice (figure 4A), while, as expected, no differences were noted in the levels of caecal acetate and propionate (data not shown). Most importantly, relative to CCD-fed mice, the CCD plus TBT-fed mice exhibited severe intestinal inflammation based on multiple parameters including gross appearance (inflamed caeca and proximal colon, splenomegaly and colomegaly), histopathological scoring, proinflammatory markers (Lcn2 and SAA) and colonic IL-1β activity (figure 4B-M). These results suggest that butyrate itself directly contributes to the severe colitis that results from enriching diets with inulin.

Figure 4

TBT, a butyrate precursor, potentiated colonic inflammation in CCD-fed mice. Four-week-old male WT mice were simultaneously maintained on CCD, CCD plus food grade TBT (5 g/kg body weight; orally on every third day for a total period of 21 days) or ICD. Three days later, mice were challenged with weekly injections of αIL-10R neutralisation antibody or isotype control. Mice were euthanised on day 21 (1 week post third injection) and analysed for caecal short-chain fatty acid and markers of colitis. (A) Caecal butyrate, (B) body weight, (C) gross colon (red circle highlights the inflamed caeca), (D) spleen weight, (E) colon weight, (F) H&E-stained colonic sections, (G) colonic histology score, (H) SAA, (I) serum Lcn2 and (J) faecal Lcn2. Proximal colon was homogenised in radioimmunoprecipitation assay buffer supplemented with protease inhibitors and used for cytokine analysis. Bar graphs show colonic (K) IL-1β, (L) IL-1Ra and (M) IL-1β/IL-1Ra ratio. The data represented as mean±SEM (*p<0.05). CCD, cellulose containing diet; ICD, inulin containing diet; IL, interleukin; Lcn2, lipocalin 2; mAb, monoclonal antibody; PCD, pectin containing diet ; SAA, serum amyloid A; TBT, tributyrin; WT, wild type.

Aggravated colitis in ICD-fed mice is NLRP3 dependent

We next tested our presumption that the exacerbated colitis in ICD-fed mice was due to heightened proinflammatory innate-immune signalling rather than an unanticipated direct effect of butyrate on some other aspect of gut function. Indeed, in the absence of adaptor protein, MyD88, which mediates signalling for all TLR, except TLR3 and inflammasome cytokines, the severity of colitis was dramatically reduced in αIL-10R-treated mice fed with ICD (online supplementary figure S8A–I). We further hypothesised that inflammasome, specifically NLRP3, signalling might be critical in this model. A recent report14 demonstrated that NLRP3 activation and excess IL-1β production drives the colitis in the absence of IL-10 in mice and humans. In both WT and Tlr5KO mice, the level of colonic IL-1β was approximately 6-fold and 19-fold, respectively, more in the ICD-fed group when compared with the PCD-fed mice. Hence, we next considered the possibility that heightened colonic inflammation in ICD-fed mice might be mediated by NLRP3. Indeed, the αIL-10R-induced colitis is NLRP3 dependent as colonic inflammation was substantially reduced in the ICD-fed NLRP3-deficient (Nlrp3KO) mice when compared with WT mice (online supplementary figure S9A–L).

Next, we investigated if deletion of NLRP3 might prevent Tlr5KO mice from developing severe colitis on IL-10R neutralisation. Indeed, mice lacking both TLR5 and NLRP3 (T5/Np3-DKO mice) were protected from loss in body weight, colomegaly and splenomegaly ((figure 5A-D). In accord, T5/Np3-DKO mice exhibited reduced colonic inflammation as assessed by inflammatory markers and histopathology (figure 5A–L). Such improvement in disease pathology was not associated with altered levels of caecal butyrate in the T5/Np3-DKO mice (figure 5M), indicating that butyrate may require NLRP3 to fuel the colonic inflammation in ICD-fed mice. NLRP3-deficient mice, whether fed inulin or CCD plus TBT, did not display significant colitis in response to αIL-10R treatment (online supplementary figure S10A–I). Collectively, these results support the notion that this colitis is NLRP3 dependent and is consistent with the possibility that butyrate’s exacerbation of anti-IL-10-induced colitis is dependent on NLRP3.

Figure 5

Deletion of NLRP3 protected ICD-fed Tlr5KO mice from chronic colitis. Male Tlr5KO (T5KO) and Tlr5/Nlrp3-DKO mice (T5/NP3DKO, 4-weeks-old) were fed with ICD and after 3 days, mice were subjected to intraperitoneal injection of αIL-10R mAb (1 mg/mouse, three weekly injections). Mice were euthanised 2 weeks after the third injection. (A) Body weight. (B) Gross colon appearance (red circle highlights the presence of occult blood). Per cent weight of (C) spleen and (D) colon. (E) Image display colonic histochemical and immunohistochemical for: (1) H&E, (2) Alcian blue, (3) macrophages (F4/80) and (4) neutrophil (Ly6G). Bar graph represents colonic (F) histology score, (G) IL-1β, (H) IL-1Ra and (I) IL-1β/IL-1Ra ratio and (J) Lcn2 level in faeces. Serum levels of (K) SAA and (L) Lcn2. (M) Concentration of butyrate in caecal content. The data represent as mean±SEM (*p<0.05). ICD, inulin containing diet; IL, interleukin; Lcn2, lipocalin 2; mAb, monoclonal antibody; NLRP3, Nod-like receptor protein 3; PCD, pectin containing diet ; SAA, serum amyloid A; WT, wild type.

Systemic elevation of endogenous NLRP3 inhibitor, BHB, mitigated chronic colitis

We next sought to determine whether NLRP3 can be targeted to treat the severe colitis in ICD-fed αIL-10R-treated mice. BHB, a major ketone body present in circulation, is reported to inhibit NLRP3.15 Dietary intake of 1,3 butanediol (BD), a monoester of BHB which gets metabolised into BHB in systemic circulation, elevates serum BHB.39 Notably, fasting can increase serum BHB by several folds without any dietary interventions. Therefore, we hypothesised that elevating circulating BHB, either via dietary supplementation or CR, up to a level that can suppress NLRP3 could diminish colonic inflammation in ICD-fed mice. Accordingly, we observed a significant increase in the serum BHB in both BD-fed group and in mice maintained on CR (a combination of regular 5-hour fasting and allowed only to eat 70% of average daily food intake) (figure 6A). Notably, such elevation in BHB, via BD or CR, mitigated the αIL-10R-induced colitis in ICD-fed WT mice when compared with ad libitum fed mice. Such improvement in colitis was reflected at gross, histological (figure 6B–G) and at colonic and serological level (figure 6H–M). Specifically, colonic levels of the inflammasome cytokine, IL-1β, was markedly reduced on both interventions (figure 6H). Similarly, the above approaches increased serum BHB and attenuated colitis in Tlr5KO mice (online supplementary figure S11A–L).

Figure 6

Dietary and CR-induced BHB suppressed colonic inflammation. Male WT mice (4-weeks-old) were divided into three groups: group 1 received ICD, group 2 received ICD along with food grade 1,3 butanediol (20% v/v) and group 3 had daily 5-hour fasting followed by controlled access (~70% of total daily food intake) to the diet. All mice were subjected to intraperitoneal injection of αIL-10R mAb (1 mg/mouse, four weekly injections) and then analysed for standard colitis parameters. (A) BHB. (B) Body weight. (C) Gross colons appearance (red circle highlights the presence of occult blood). (D) Images show the (1) H&E and (2) alcian blue staining and immunostaining for (3) macrophages (F4/80) and (4) neutrophils. (E) Colonic histology score. (F–G) Per cent weight of (F) spleen and (G) colon. Proximal colon was processed in radioimmunoprecipitation assay buffer and supernatant was used for cytokine estimation. (H–J) Colonic level of (H) IL-1β, (I) IL-1Ra and (J) IL-1β/IL-1Ra. Serum level of (K) SAA and (L) Lcn2. (M) Lcn2 level in faeces. The data represented as mean±SEM (*p<0.05). BHB, β-hydroxybutyrate; CR, calorie restriction; ICD, inulin containing diet; IL, interleukin; Lcn2, lipocalin 2; mAb, monoclonal antibody; PCD, pectin containing diet; SAA, serum amyloid A; WT, wild type.

That some fermentable fibres, notably inulin, can alter the microbiota in a manner that results in reduced food consumption28 led us to consider if the amelioration of inflammasome signalling and colitis development by pectin might have involved reduced food intake. In accord with this possibility, mice fed PCD exhibited a modest reduction in food consumption before and during acute  and chronic colitis. However, there was no significant difference in food consumption between mice fed ICD and PCD at any time points (data not shown). Thus, the modest reduction in food consumption that resulted from pectin was not likely a predominant mechanism underlying reduced activation of the inflammasome and amelioration of colitis.

Lastly, to gain a better mechanistic understanding of how pectin, but not inulin, diminishes the colonic inflammation, we profiled immune cells in the lamina propria at the early stage of colitis. Intriguingly, regulatory T cells (Tregs: CD3+CD4+FoxP3+) were abundant in the PCD, but reduced in the ICD-fed mice, when compared with CCD-fed mice (online supplementary figure S11E). Tregs are known to limit intestinal inflammation by preventing activation of the mucosal immune system.40 Other immune cells assayed, including macrophages, neutrophils, CD4+, CD8+, Th17 and NK cells, were unaltered (online supplementary figure S12A–H). Together, these results support the notion that NLRP3 is a key driver of colonic inflammation in ICD-fed mice.


In recent years, the gut microbiota has emerged as a key non-genetic potentially modifiable factor that substantially influences gut health. Indeed, it is now recognised that alterations in microbial composition contribute to the pathogenesis of several chronic diseases including IBD. That diet can alter microbiota   supports speculation that broad societal changes in diet have contributed to post-mid-20th century increase in incidence of IBD.41 Such changes include increased consumption of processed foods that are often fortified with highly refined fibres. Unlike healthy individuals, many patients with IBD develop clinical complications on consumption of diets rich in certain fibres including fermentable oligosaccharides, disaccharides, monosaccharides and polyols (FODMAP), particularly during flares of active disease, whereas reducing fibre intake may help avoid disease flares.34 37 This starkly contrasts with the generally held view that fibre is health promoting outside the context of IBD. This view is largely based on epidemiological studies that assess consumption of fibre-rich whole foods (fruits and vegetables), rather than refined fibres used to fortify processed foods. Hence, we sought to examine the effects of refined fibres on IBD. We observed that dietary pectin protected against development of αIL-10R-induced colitis while inulin promoted it, likely via, at least in part, promoting NLRP3 activation. These results highlight potential risks and benefits of specific dietary fibres to persons at risk of developing IBD.

While reports of gut microbiota dysbiosis in patients with IBD are numerous, the observed compositional differences are quite varied and sometimes conflicting.42 43 Yet, some features of dysbiosis like the loss of butyrate-producing bacteria during IBD are widely presumed to be detrimental,43 thus, giving rise to the notion that butyrate supplementation may help treat the disease. Such views are substantiated by studies demonstrating the physiological effects of butyrate in serving as an energy source to the gut epithelia44 and in promoting differentiation of Tregs.45–47 Hence, while the unpleasant odour and poor palatability of butyrate and other SCFA have slowed their widespread use and incorporation into oral formulations to treat IBD, their dietary precursors, namely, soluble fibres such as inulin and pectin, are increasingly used to enrich processed foods. The notion that use of such fibres would be broadly health promoting spurred the ongoing interest to exploit them as an attractive means to elevate butyrate levels in the gut during IBD. Our results generally support the concept that some dietary fibres have potential to ameliorate inflammation, but caution that strategies which largely focus on enriching butyrate-producing bacteria may also increase the risk of IBD in susceptible hosts.

Fructo-oligosaccharides in processed foods are typical examples of FODMAP, whose fermentation by gut bacteria release SCFA and gases that could result in GI discomfort such as bloating, cramps, diarrhoea and constipation.48 Knowing that dietary fibres, in some conditions, can exacerbate inflammation fits clinical recommendation that individuals with active IBD should exclude FODMAP from their diet.34 37 38 Likewise, WT mice fed a diet-containing prebiotic, inulin and probiotics had decreased LPS-induced inflammatory responses, whereas IBD-susceptible IL-10−/− mice given the similar diet displayed substantial increases in the colonic proinflammatory genes.49 Other negative effects of inulin on intestinal tumorigenesis in susceptible hosts have gradually been characterised in recent years,50 51 thus lending credence to the notion that inulin may be conditionally, but not universally beneficial. These observations generally support a paradigm where fibres are ‘double-edged swords’ that could promote or deter health depending on the pre-existing conditions in the intestine and the specific fibre consumed. Our results suggest that this paradigm may also apply specifically to butyrate-producing bacteria: beneficial in some contexts but also capable of exacerbating inflammation. We note an intriguing study by the Stappenbeck group, who demonstrated that a surfeit of butyrate in the inflamed gut may do more harm than good by potentially inhibiting the proliferation of colonic epithelial stem/progenitor cells and thus impair restitution.25 Mice receiving prolonged and high doses of SCFA have also been reported to develop T cell-mediated urethritis and hydronephrosis52 in spite of the ability of SCFA to promote immune tolerance.45–47 53 More surprisingly, severe disease and higher mortality was observed in DSS-treated mice that received butyrate-producing bacteria.26

That the exacerbated inflammation was reduced not only by eliminating these bacteria but also by using an inhibitor of SCFA production, like hops β-acids, implicates that surfeit caecal butyrate could be one of the drivers of severe colitis observed in inulin-fed mice. In addition, we observed that the β-acids attenuated the severe intestinal bleeding previously described in inulin-fed DSS-treated mice.54 In accord with the notion that butyrate can exacerbate colitis, we note that orally administered TBT, a food additive that increases butyrate levels in the large intestine, worsened colitis in CCD-fed αIL-10R treated mice. These results collectively imply that surfeit caecal butyrate can boost αIL-10R-induced intestinal inflammation. However, we do not exclude the possibility of other undefined factors that may also influence the development of colitis in ICD-fed mice. For instance, we note that the effects of metronidazole are not only limited to depleting butyrate producers but also broadly depleting other anaerobic bacteria55 and promoting the expansion of Lactobacilli, which could, in part, contribute to diminish colonic inflammation. Analogously, β-acids do not only block butyrate production but also overall fermentation; therefore, they are likely to have secondary impacts on other aspects of bacterial metabolism.

We do not dispute that some of the health-promoting properties of soluble fibres are driven by SCFA, nor do we question their therapeutic potential to treat inflammatory diseases. Rather, we propose that the factors dictating the equivocal effects of dietary fibres should be carefully studied during an active state of colitis. A seminal study by Medzhitov and colleagues14 asserts that IL-10 is indispensable in suppressing IBD in both humans and mice, at least in part, by preventing the aberrant activation of NLRP3 in colonic macrophages. Intriguingly, there have been several reports suggesting that butyrate could also modulate NLRP3 activation and responses.17–19 While the extent to which butyrate promotes or inhibits NLRP3 remains a subject of controversy, our results provide support for a potentially positive correlation between elevated butyrate levels and heightened inflammasome activation in the inflamed gut. Metabolomics analyses revealed that, among the fibres examined herein, dietary inulin generates higher quantities of butyrate in both healthy and colitic mice. These observations led us to argue that perhaps the butyrate-inflammasome axis may, in part, explain why inulin feeding worsened the severity of colitis in mice with loss of IL-10 function in the study herein, and also in mice with DSS-induced colitis in our previous study.54 Consistent with this notion, the interventions which deplete butyrate producers, inhibit bacterial fermentation or ablate NLRP3 were indeed sufficient to alleviate the colitis in inulin-fed αIL-10R treated mice. Increasing the levels of BHB,15 16 either through supplementation or via CR, also conferred substantial mucoprotection to ICD-fed αIL-10R-treated mice. While the aforementioned interventions (eg, metronidazole, β-acids, BHB) are conceived to suppress NRLP3, we accede that they may also contribute to dampening colitis through other mechanisms not analysed in this study.

It would be an oversimplification to view all fibres as FODMAP and thus potentially risky for patients with IBD. Unlike the fructan-type inulin, the galacturonan-type pectin was more effective in improving the colonic inflammation in mice lacking IL-10 function. While pharmacological interventions with butyrate can ameliorate gut inflammation in both mice and humans, our working hypothesis is that the extent to which it does so is context-dependent. Indeed, some studies found no or adverse impacts of butyrate administration on colonic inflammation.25–27 In the present study, the worsened colitis in inulin-fed mice could be due to elevated butyrate coupled with inflammatory disequilibrium, for example, approximately twofold less production of colonic IL-1Ra when compared with PCD-fed group (19.5±1.7 ng/mg protein (PCD-fed) vs 9.9±1.3 (CCD-fed) or 11.2±0.65 (ICD-fed)). One possible factor contributing to this context is that, herein, inulin was added to ‘open-source’ compositionally defined diet (CDD), which, in and of itself impacts microbiota and consequently gut physiology.28 While such CDD are needed to permit careful controlled dietary studies, they are also thought to be reminiscent of highly processed diets that are consumed by many people in developed countries. Thus, the extent to which interventions that increase levels of butyrate promote or dampen inflammation may depend on background diet, microbiota composition and/or level of inflammation. In other words, we are not claiming that butyrate is universally promoting inflammation, but that in some contexts it can do so and thus interventions to increase butyrate generation should proceed with caution as they may help some individuals but may also harm IBD-prone individuals.

While pectin is not classified as a FODMAP, it remains unclear whether pectin has any, if not reduced, side effects when consumed during IBD. At least in comparison to inulin, we demonstrated that pectin-feeding resulted in less caecal butyrate. Considering that inulin and pectin are structurally distinct and require specific enzymes (eg, inulinases and pectinases) for their fermentation, it seems conceivable that these fibres may differentially cater to diverse groups of bacteria, resulting in disparate SCFA profiles and physiological effects. Additionally, we reasoned that the mucoprotection conferred by pectin during colitis may be, in part but strongly, associated with its ability to upregulate IL-1Ra, which is known to counter-regulate IL-1β bioactivity, in addition to promoting elevation of colonic Tregs. Further studies are needed to explore whether pectin may serve as an alternative source of fibre for patients with IBD who frequently develop complications with FODMAPs or similar types of complex carbohydrate-rich foods, to rescue from prolonged fibre deprivation.

In conclusion, our study not only supports the potential of enriching diets with fibre as a means to ameliorate inflammation but also reveals that the manner in which diet-induced alterations in microbiota composition and consequently their metabolic products impact inflammation is very much context dependent and can vary greatly in states of active inflammation. The distinct effects mediated by dietary fibres, studied herein, also serve to emphasise the fact that not all fibres are created equally, nor will they be fermented uniformly. A deeper understanding on the various aspects of such fibres and how they are tolerated by the host may pave the way forward for development of personalised fibre-based interventions for patients with IBD.


View Abstract


  • Patient consent for publication Not required.

  • Contributors VS conceived the project, designed and performed the mouse experiments. Original observation was made by VS. BSY helped with histochemical staining and writing the manuscript. REW, GCS, JC, QL and ADP analysed SCFA in caecal contents and provided inputs for data interpretation. ACAB, XX, RMG and PS performed colonic gene expression and immune cell profiling. VS, BC and XC generated and analysed the microbiota data. MDF provided hops β-acid with intellectual inputs on bacterial fermentation. ATG assisted with data interpretation and study design. MV-K participated in the study design and directed the study. VS, ATG and MV-K wrote the manuscript.

  • Funding This work was supported by National Institutes of Health grants DK097865 (to MV-K), DK083890 (to ATG) and DK099071 (to ATG). VS is supported by research fellowship award (ID# 418507) and career development award (ID# 597229) from Crohn’s & Colitis Foundation (CCF). PS is supported by research fellowship award from CCF. BC is supported by a career development award fellowship from CCF and by an Innovator Award from the Kenneth Rainin Foundation.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.