Dietary fibre directs microbial tryptophan metabolism via metabolic interactions in the gut microbiota

Abstract

Tryptophan is catabolized by gut microorganisms resulting in a wide range of metabolites implicated in both beneficial and adverse host effects. How gut microbial tryptophan metabolism is directed towards indole, associated with chronic kidney disease, or towards protective indolelactic acid (ILA) and indolepropionic acid (IPA) is unclear. Here we used in vitro culturing and animal experiments to assess gut microbial competition for tryptophan and the resulting metabolites in a controlled three-species defined community and in complex undefined human faecal communities. The generation of specific tryptophan-derived metabolites was not predominantly determined by the abundance of tryptophan-metabolizing bacteria, but rather by substrate-dependent regulation of specific metabolic pathways. Indole-producing Escherichia coli and ILA- and IPA-producing Clostridium sporogenes competed for tryptophan within the three-species community in vitro and in vivo. Importantly, fibre-degrading Bacteroides thetaiotaomicron affected this competition by cross-feeding monosaccharides to E. coli. This inhibited indole production through catabolite repression, thus making more tryptophan available to C. sporogenes, resulting in increased ILA and IPA production. The fibre-dependent reduction in indole was confirmed using human faecal cultures and faecal-microbiota-transplanted gnotobiotic mice. Our findings explain why consumption of fermentable fibres suppresses indole production but promotes the generation of other tryptophan metabolites associated with health benefits.

Fig. 1. Tryptophan supplementation increases tryptophan-derived Stickland fermentation products.

a, Schematic representation of the Stickland fermentation pathway. Stickland fermentation of Trp generates either the oxidative pathway product IAA or the reductive pathway products ILA, IAcrA and IPA. b, Concentrations of Trp, ILA and IPA in mGAM (control) and culture supernatants of C. sporogenes grown in mGAM supplemented with final concentrations of 0.02%, 0.05%, 0.1% and 0.2% free tryptophan. Bars and error lines indicate the mean ± s.d. of three to four independent biological replicates. c, Concentrations of Trp, ILA, IAcrA and IPA in mGAM (control) and culture supernatants of P. anaerobius grown in mGAM supplemented with final concentrations of 0.02%, 0.05%, 0.1% and 0.2% free tryptophan. Bars and error lines indicate the mean ± s.d. of three independent biological replicates. d, Fold change in concentrations of tryptophan metabolites in the culture supernatants of six infant faecal communities cultured either in unsupplemented YCFA medium or in YCFA supplemented with 0.05% or 0.1% or 0.2% of free tryptophan. Specific metabolite concentrations are normalized to the basal level of the given metabolite in the growth medium without tryptophan supplementation. Lines and error lines indicate the mean ± s.d. IPA was detected only in two of the six infant faecal communities owing to the lack of producer species P. anaerobius in the four remaining faecal samples (Extended Data Fig. 2d). Absolute values of metabolites in the individual faecal cultures and the microbiota compositions are shown in Extended Data Fig. 2. Source data

Fig. 2. Carbohydrate supplementation inhibits indole production in faecal cultures.

a, Schematic representation of tryptophanase-mediated catabolism of tryptophan to produce indole, pyruvate and ammonia. b–d, Concentrations of indole (b), Trp (c) and ILA (d) in the faecal culture supernatants after cultivation in YCFA medium supplemented with 0.05%, 0.1% or 0.2% glucose, maltose and cellobiose, collectively referred to as GMC. One infant faecal sample (23.11) was selected for cultivation in three replicates as it contained both ILA and IPA producers (for example, P. anaerobius and Bifidobacterium longum) and indole producers (Escherichia). Metabolites were normalized to the final OD₆₀₀ of the culture in the individual culture supernatants. Results are the mean ± s.d. of three independent experiments. Statistical analysis was done using a two-tailed unpaired t-test comparing lowest and highest GMC concentrations, with *P < 0.05; **P < 0.01 (indole, P = 0.0061; Trp, P = 0.0126; ILA, P = 0.0456). Individual replicates and their 16S rRNA profiles are shown in Extended Data Fig. 3. Source data

Fig. 3. Tryptophan and fibre supplementation alters production of tryptophan metabolites in a defined community of gut microorganisms in vitro.

Tryptophan metabolites produced by the defined community in vitro. E. coli was selected as the major indole producer, B. thetaiotaomicron was selected as the fibre degrader and C. sporogenes was selected because of its ability to generate Stickland fermentation products. a–d, Concentrations of ILA (a), IPA (b), indole (c) and Trp (d) in the supernatants of the defined community cultured in mGAM supplemented with either 0.02% or 0.05% free tryptophan and with or without 0.5% apple pectin. Bars and error lines indicate the mean ± s.d. of three independent biological replicates. Statistical analysis was done using the Brown–Forsythe ANOVA test using an unpaired two-tailed t-test with Welch’s correction. Only two replicates are shown in the group of 0.02% Trp without pectin for ILA, IPA and Trp owing to technical issues during analysis. e, Relative expression from RT-qPCR targeting tnaA mRNA in E. coli in response to tryptophan and pectin supplementation. f, Relative expression from RT-qPCR targeting mRNAs of arabinose-utilizing genes (araA and araF), rhamnose-utilizing genes (rhaA and rhaT) and xylose-utilizing genes (xylA and xylG) in E. coli in response to tryptophan and pectin supplementation. Total RNA was extracted from early stationary phase cultures (∼1 OD), and mRNA levels were measured as described in Methods and reported as relative difference (fold change) to the 0.02% Trp condition. Results are the mean ± s.e.m. of three independent experiments. Unpaired two-tailed t-tests were performed on the expression ratios to determine the statistical significance of the relative expression differences. P values are shown in the figure panels. Source data

Fig. 4. Dietary fibre and tryptophan supplementation modulates production of tryptophan metabolites by a defined community in vivo.

a, Schematic representation of the experimental plan to evaluate the effect of dietary tryptophan and pectin on the production of tryptophan metabolites in vivo. GF mice were placed in four groups (n = 5 per group) and fed a diet containing 2 g kg⁻¹ tryptophan and 50 g kg⁻¹ pectin for 7 days for adaptation. They were then orally gavaged with a mixed culture of E. coli, B. thetaiotaomicron and C. sporogenes in equal amounts (OD₆₀₀) and remained for another 7 days on the same diet for stabilization. The diets were then changed; the mice were fed a diet with either 2 g kg⁻¹ or 16 g kg⁻¹ tryptophan, with or without 50 g kg⁻¹ pectin for two more weeks. Diet compositions are described in Supplementary Table 1. Samples were collected as shown in the scheme. b, The 16S rRNA gene sequencing profiles show the composition of the defined community in the caecum of each mouse in the four groups, overlaid with indole values measured in the individual caeca. c, Absolute concentrations of indole in caeca. d, Indole concentration in the caeca, normalized to the relative abundance of E. coli. e, Absolute concentrations of Trp, ILA, IAcrA and IPA in serum. f, Serum tryptophan metabolites (ILA, IAcrA and IPA) normalized to C. sporogenes relative abundance in caecum. For c, lines and error bars indicate means and standard deviations, respectively; for d–f, lines and error bars indicate medians and interquartile ranges (IQRs), respectively. Statistical analysis was done across groups within each metabolite measured using one-way ANOVA (c) or Kruskal–Wallis tests (d–f), using uncorrected Fisher’s LSD or Dunn’s post hoc tests (two tailed) to compare between individual groups. For c–f, n = 5 mice samples per group were used for statistical analysis. However, for e and f, one value for tryptophan and ILA was excluded as an extreme outlier (Grubbs test, alpha < 0.01). P values are shown in the figure panels. Panel a was created with BioRender.com. Source data

Fig. 5. Dietary fibre supplementation inhibits indole production by complex human gut microbial communities both in vitro and in vivo.

a–c, Concentrations of tryptophan metabolites indole (a) and ILA (c) and tryptophan (b) in the culture supernatants of nine separate human faecal microbial communities grown in mGAM with no fibre supplementation or supplemented with a mixture of fibres or pectin. Lines and error bars indicate medians and IQRs, respectively. Statistical analysis was done using the Friedman test, with Dunn’s post hoc test (two tailed). P values are shown in the figure panels. d, Schematic representation of the experimental plan to evaluate the effect of dietary fibre on production of tryptophan metabolites in vivo. GF mice were placed in two groups (n = 5 or 6 per group) and fed a complex fibre diet (Altromin 1314) for 14 days before FMT of the mice was done with communities originating from two different human adult donors. Subsequently, the mice remained for another 27 days on the same diet for stabilization. The mice were then fed a diet depleted of fermentable fibres (D10012G) for 2 days before feeding them a complex fibre diet for ten more days. Thereafter, the mice were fed a diet with 2 g kg⁻¹ tryptophan without pectin for 2 weeks and then a diet with 2 g kg⁻¹ tryptophan and 50 g kg⁻¹ pectin for two more weeks. Pectin diet compositions are described in Supplementary Table 1. Faecal samples were collected as shown in the scheme. e, The 16S rRNA gene sequencing profiles show the average relative abundance of individual ASVs in faeces across all mice at each sampling point. Only ASVs with relative abundance >5% in at least one sample are shown, and the rest are grouped into ‘others’. f, Absolute faecal indole concentrations showing means and 95% CIs, as well as the data point of each individual mouse. Statistical analysis was done using repeated-measure one-way ANOVA, with Sidak’s multiple-comparison test (two tailed; q values). The q values are shown in the figure panels. Panel d was created with BioRender.com. Source data

Fig. 6. Impacts on tryptophan metabolism mediated by dietary fibre and substrate.

In the gut, multiple bacterial species require tryptophan for their metabolism and produce bioactive molecules important for host health. E. coli catabolizes tryptophan into indole to generate pyruvate, while C. sporogenes regenerates NAD⁺ and produces ILA and IPA through the Stickland fermentation reductive pathway. The fibre degrader B. thetaiotaomicron degrades pectin and thereby releases monosaccharides available to E. coli. The monosaccharides repress expression of the E. coli tnaA gene encoding tryptophanase, thereby making more tryptophan available to Stickland fermenters in the gut environment. Blue arrows show events occurring in the absence of fibre, while green arrows designate events preferentially occurring in the presence of fibre. Thick and thin arrows depict enhanced and reduced flow of tryptophan, respectively. Although E. coli is shown here as a representative species of indole producers, we argue that the catabolite repression of the tnaA gene is widespread and applies to many other indole producers in the gut. Similarly, in a complex microbiota, fibre degradation and tryptophan utilization can also be performed by other bacterial species than B. thetaiotaomicron and C. sporogenes, as shown here, thus contributing to the diverse bacterial metabolite accumulation in the gut. Figure was created with BioRender.com.

Extended Data Fig. 1. The gut bacteria Clostridium sporogenes and Peptostreptococcus anaerobius produce tryptophan metabolites.

(a) Tryptophan metabolites in the medium and culture supernatants of C. sporogenes grown for 48 hrs in mGAM medium. Bars and error lines show mean ± s.d. of four independent biological replicates for C. sporogenes and three for control medium. (b) Tryptophan metabolites in the medium and culture supernatants of P. anaerobius grown for 48 hrs in mGAM medium in triplicates. Bars and error lines show mean ± s.d. of three independent biological replicates. (c) Tryptophan metabolites in serum (five) and in cecum (four) of germ-free (GF) and six gnotobiotic mice monocolonized with C. sporogenes (C. spo). Source data

Extended Data Fig. 2. Effect of glucose and tryptophan supplementation on production of tryptophan metabolites.

(a) Concentration of tryptophan metabolites in the mGAM medium and culture supernatants of C. sporogenes grown in mGAM medium supplemented with final concentrations of 0.05, 0.2 or 0.4 and 0.6 % glucose. Bars and error lines show mean ± s.d. of three independent biological replicates. (b) Concentration of tryptophan metabolites in the mGAM medium and culture supernatants of P. anaerobius grown in mGAM medium supplemented with final concentrations of 0.05, 0.2 or 0.4 and 0.6 % glucose. Bars and error lines show mean ± s.d. of three independent biological replicates. (c) Absolute concentrations of tryptophan metabolites in the culture supernatants of six infant faecal microbiotas and culture medium control (media control). The six different infant faecal microbial communities are cultured either in YCFA medium or YCFA supplemented with 0.05 or 0.1 or 0.2 % of free tryptophan. (d) 16S rRNA gene sequencing profile of the six infant faecal microbiota compositions. Source data

Extended Data Fig. 3. Carbohydrates supplementation inhibits indole production and stimulates ILA production in faecal cultures.

(a) Normalized concentrations of tryptophan metabolites in the culture supernatants from faecal cultures in YCFA medium supplemented either with 0.05 or 0.1 or 0.2 % each of glucose (G), maltose (M) and cellobiose (C), collectively referred here as GMC. A single infant faecal sample (23.11) was selected for cultivation in three replicates since it contained both ILA/IPA producers (for example P. anaerobius and B. longum), and indole producers (Escherichia). Metabolites in the individual culture supernatants were normalized against final OD₆₀₀ of the culture. All individual replicates are shown here. (b) 16S rRNA profiles of all replicates. Note that Peptostreptococcus is frequently lost thus preventing IPA accumulation in many samples (for example replicate #1 and #3). 16S rRNA profile of 0.2 % of #2 replicate is missing due to the loss of the sample during processing. Source data

Extended Data Fig. 4. Supplementation with glucose or pectin inhibits indole production by repressing tnaA gene expression.

(a) Indole production measured in the culture supernatants of E. coli, B. theta and C. sporogenes cultured individually in either LB broth or LB broth supplemented with 0.5 % glucose or 1 % glucose. Bars and error lines show mean ± SD of two independent biological replicates. (b) Indole production measured in the culture supernatants of defined communities consisting of E. coli (E), B. theta (B), P. anaerobius (P) and C. spo (C) cultured in either mGAM or mGAM supplemented with 0.5 % pectin. Bars and error lines show mean ± SD of two independent biological replicates. (c) Defined community consists of E. coli, B. theta and C. spo are cultured together in mGAM medium supplemented with either 0.02 or 0.05 % free tryptophan and with or without 0.5 % apple pectin. Total RNA was extracted after 24 hrs fermentation and mRNA levels were measured as described in methods and reported as relative difference (fold change) to the 0.02% Trp condition. Bars and error lines show mean ± SEM of three independent biological replicates. Metabolites data are shown in Fig. 3. Relative expression from RT-qPCR to measure tnaA mRNA levels in E. coli and in B. theta in response to pectin supplemented (0.5%) in the growth medium after 24 hrs fermentation at either low (0.02%) or high tryptophan (0.05%) concentrations is shown here. Unpaired two-tailed t-tests were performed on the expression ratios to determine the statistical significance of the relative expression differences. P values are shown in the figure panels. Source data

Extended Data Fig. 5. Additional data from the defined community experiment in vivo assessing the effect of dietary fibre and tryptophan supplementation on production of tryptophan metabolites.

(a) Weekly water intake, (b) weekly food intake and (c) calculated weekly tryptophan intake from individual mice in the different experimental groups (n = 5 biologically independent animals per group) described in Fig. 4a. Bars and error bars indicate mean ± s.d. No significant differences were found between groups in water intake (p_{group_overall} = 0.61) or food intake (p_{group_overall} = 0.39) according to Two-way repeated measures ANOVA (p > 0.05 for all pairwise comparisons between feeding groups at week 2 and 3 after Bonferroni correction). Tryptophan intake differed significantly between groups (p_{group_overall} < 0.0001) and were significantly higher for group 3 and 4 compared to group 1 and 2 in both week 2 and 3 (p < 0.01 for all pairwise comparisons across, but not within, trp feeding groups after Bonferroni correction). (d) Total bacterial load quantified using qPCR, showing median and error bars showing IQR. No significant differences were found using Kruskal-Wallis tests (n = 5 biologically independent animals per group). (e) 16S rRNA gene sequencing profiles show the composition of the defined community in colon of each individual mouse, with indole values in the colon overlaid. Indole values are missing for few mice due to the extremely low samples amounts present in their colon. (f) Absolute indole concentration in the colon. (g) Indole concentration in the colon, normalized to the relative abundance of E. coli in colon. (h, i) Absolute abundance of IPA in (h) cecum and in (i) colon, with grey shaded area indicating background noise. (j) Serum tryptophan metabolites normalized to C. sporogenes relative abundance in colon. (k) Scatter plot and Spearman’s rank correlation of relative abundances of isovaleric acid and isobutyric acid versus C. sporogenes relative abundance in cecum. All graphs in panel f-j show median and IQR and statistical analysis was done across groups within each metabolite measured using One-way ANOVA (panel f) or Kruskal Wallis tests (panel g-j), using uncorrected Fisher’s LSD or Dunn’s posthoc tests to compare between individual groups. P values are shown in the figure panels. For panel f-i, n = 5 biologically independent animals per group was used for statistical analysis except ‘Normal Trp+Pectin’ (n = 3) and ‘High Trp’ (n = 4). For panel j, n = 5 biologically independent animals per group was used for statistical analysis, except one value for ILA was excluded as an extreme outlier (Grubbs test, alpha < 0.01). Source data

Extended Data Fig. 6. Fibre mixture or pectin supplementation inhibits indole production and stimulates ILA production in complex human gut microbial communities in vitro.

(a) Concentrations of tryptophan metabolites in the individual culture supernatants of nine adult human faecal microbiota in mGAM medium supplemented either with pectin or with the mixture of fibres as described in methods. Results from one sample (S019) is missing for Tryptophan and ILA due to an error with the internal standard in the LC-MS for this sample. (b) 16S rRNA gene sequencing profiles show the composition of the microbiota community in each individual cultures. Source data

Extended Data Fig. 7. Effect of fibre mixture or pectin supplementation on total bacterial load, Escherichia and B. thetaiotaomicron as well as concentrations of tryptophan metabolites in vivo.

(a) Total bacterial load in faeces were quantified using qPCR, showing mean and 95% CI, and all samples from individual mice connected with a grey line. (b) Absolute abundance of Escherichia species (indole producers) across all faecal samples, showing mean and 95% CI, and all samples from individual mice connected with a grey line. (c) Absolute abundance of Bacteroides thetaiotaomicron (fibre degrader and indole producer) across all faecal samples showing mean and 95% CI and all samples from individual mice connected with a grey line. (d) Concentrations of tryptophan across all faecal samples, showing mean and 95% CI. (e-f) Concentrations of (e) IPA and (f) ILA in the faeces of each individual mice at each sampling point, with grey shaded area indicating background noise, and dashed line limit of detection (LOD). For panel a-d statistical analysis (n = 11 biologically independent animals) was done using mixed effects analysis with Sidak’s multiple comparisons tests (q-values) to compare between sampling periods over time. q-values are shown in the figure panels. Due to the limited number of samples above limit of detection statistical test were not perform for data in panel e-f. Source data

Extended Data Fig. 8. Qualitative analysis of indolic compounds using Kovac’s assay.

Kovac’s reagent specificity against 0, 10, 20, 50 and 100 µM concentration of Indole, IAA, ILA, IAcrA, IPA and skatole were assessed and showed that it is highly specific for indole.