Direct impact of commonly used dietary emulsifiers on human gut microbiota

Abstract

Background: Epidemiologic evidence and animal studies implicate dietary emulsifiers in contributing to the increased prevalence of diseases associated with intestinal inflammation, including inflammatory bowel diseases and metabolic syndrome. Two synthetic emulsifiers in particular, carboxymethylcellulose and polysorbate 80, profoundly impact intestinal microbiota in a manner that promotes gut inflammation and associated disease states. In contrast, the extent to which other food additives with emulsifying properties might impact intestinal microbiota composition and function is not yet known.

Methods: To help fill this knowledge gap, we examined here the extent to which a human microbiota, maintained ex vivo in the MiniBioReactor Array model, was impacted by 20 different commonly used dietary emulsifiers. Microbiota density, composition, gene expression, and pro-inflammatory potential (bioactive lipopolysaccharide and flagellin) were measured daily.

Results: In accordance with previous studies, both carboxymethylcellulose and polysorbate 80 induced a lasting seemingly detrimental impact on microbiota composition and function. While many of the other 18 additives tested had impacts of similar extent, some, such as lecithin, did not significantly impact microbiota in this model. Particularly stark detrimental impacts were observed in response to various carrageenans and gums, which altered microbiota density, composition, and expression of pro-inflammatory molecules.

Conclusions: These results indicate that numerous, but not all, commonly used emulsifiers can directly alter gut microbiota in a manner expected to promote intestinal inflammation. Moreover, these data suggest that clinical trials are needed to reduce the usage of the most detrimental compounds in favor of the use of emulsifying agents with no or low impact on the microbiota. Video abstract.

Keywords: Dietary emulsifier; Gut microbiota; IBD; Intestinal inflammation.

Fig. 1

Impact of dietary emulsifiers on bacterial density. Bacterial density was assessed by qPCR during the treatment (a) and the post-treatment phases (b). Area under curve (AUC) was calculated for both increase and decrease in bacterial density, as detailed in Figure S2 and in the “Methods” section. Data are the means ± S.E.M, with individual data points being represented (N = 3). *P < 0.05 compared to the untreated group, determined by a one-way analysis of variance corrected for multiple comparisons with a Bonferroni post-test

Fig. 2

Impact of dietary emulsifiers on microbiota composition. Microbiota composition was analyzed by 16S rRNA gene sequencing and beta diversity was computed through QIIME2 pipeline using the Jaccard matrix. a Principal coordinate analysis (PCoA) of the Jaccard matrix from all the time points and treatments analyzed from the three independent experiments. Dots are colored by time point. b, c PCoA of the Jaccard matrix from all the time points of the untreated MBRA chambers from the three independent experiments and colored by time point (b) or experiment (c). d–f Principal coordinate analysis (PCoA) of the Jaccard matrix from all the time points and treatments analyzed from each independent experiment. Dots are colored by treatment. N = 3

Fig. 3

Impact of dietary emulsifiers on microbiota composition. Microbiota composition was analyzed by 16S rRNA gene sequencing and beta diversity was computed through QIIME2 pipeline using the Jaccard matrix (a, b) and the unweighted UniFrac distance (c, d) during the treatment (a, c) and the post-treatment phases (b, d). Area under curve (AUC) was calculated, as detailed in Figure S2 and in the “Methods” section. Data are the means ± S.E.M, with individual data points being represented (N = 3). *P < 0.05 compared to the untreated group, determined by a one-way analysis of variance corrected for multiple comparisons with a Bonferroni post-test

Fig. 4

Impact of dietary emulsifiers on microbiota diversity. Microbiota composition was analyzed by 16S rRNA gene sequencing and alpha diversity was computed through QIIME2 pipeline using the evenness index (a, b) and the number of observed OTUs (c, d) during the treatment (a, c) and the post-treatment phases (b, d). Area under curve (AUC) was calculated for both increase and decrease in alpha diversity, as detailed in Figure S2 and in the “Methods” section. Data are the means ± S.E.M, with individual data points being represented (N = 3). *P < 0.05 compared to the untreated group, determined by a one-way analysis of variance corrected for multiple comparisons with a Bonferroni post-test

Fig. 5

Impact of dietary emulsifiers on microbiota composition at various taxonomic levels. a Microbiota composition was analyzed by 16S rRNA gene sequencing and taxonomic analysis were computed through QIIME2 pipeline at the order level (144-h time point). b Microbiota composition was analyzed by 16S rRNA gene sequencing and taxonomic analysis were computed through QIIME2 pipeline at the genus level (144-h time point). Only the 12 more abundant genera are represented, from the more abundant (top left) to the least abundant (bottom right). Data are the means ± S.E.M, with individual data points being represented (N = 3). *P < 0.05 compared to the untreated group, determined by a one-way analysis of variance corrected for multiple comparisons with a Bonferroni post-test

Fig. 6

Impact of dietary emulsifiers on the expression of microbiota-derived pro-inflammatory molecules. Microbiota-derived expression of pro-inflammatory molecules was analyzed using HEK cells expressing TLR4 or TLR5 in order to quantify bioactive levels of lipopolysaccharide (LPS) (a, b) and flagellin (FliC) (c, d), respectively, during the treatment (a, c) and the post-treatment phases (b, d). Area under curve (AUC) was calculated, as detailed in Figure S2 and in the “Methods” section. Data are the means ± S.E.M, with individual data points being represented (N = 3). *P < 0.05 compared to the untreated group, determined by a one-way analysis of variance corrected for multiple comparisons with a Bonferroni post-test

Fig. 7

Impact of dietary emulsifiers on in vitro microbiota metatranscriptomes. Total RNAs were extracted from MBRA suspension collected at the 120-h time point and subjected to RNA sequencing. a HUMAnN2 program was used to profile for the abundance of microbial pathways and gene families, and a transcript table with relative abundance was generated for each sample and compared using principal coordinate analysis of the Bray-Curtis distance. b Bray-Curtis distance separating each sample from control (untreated) microbiota was determined and plotted. Data are the means ± S.E.M, with individual data points being represented (N = 3). *P < 0.05 compared to the untreated group, determined by a one-way analysis of variance corrected for multiple comparisons with a Bonferroni post-test

Fig. 8

Global compositional and functional effects of dietary emulsifiers on the human microbiota. Heatmap visualization of the impact of dietary emulsifiers on bacterial density (presented in Fig. 1), microbiota composition (beta (presented in Fig. 3) and alpha (presented in Fig. 4) diversity), expression of microbiota-derived pro-inflammatory molecules (presented in Fig. 6), and metatranscriptome (presented in Fig. 7) the human microbiota. *P < 0.05 compared to the untreated group, determined by a one-way analysis of variance corrected for multiple comparisons with a Bonferroni post-test. Emulsifiers are listed from the lowest (left side) to the highest (right side) effect on microbiota composition and function