Lactobacillus rhamnosus GG modifies the metabolome of pathobionts in gnotobiotic mice

Abstract

Background: Lactobacillus rhamnosus GG (LGG) is the most widely used probiotic, but the mechanisms underlying its beneficial effects remain unresolved. Previous studies typically inoculated LGG in hosts with established gut microbiota, limiting the understanding of specific impacts of LGG on host due to numerous interactions among LGG, commensal microbes, and the host. There has been a scarcity of studies that used gnotobiotic animals to elucidate LGG-host interaction, in particular for gaining specific insights about how it modifies the metabolome. To evaluate whether LGG affects the metabolite output of pathobionts, we inoculated with LGG gnotobiotic mice containing Propionibacterium acnes, Turicibacter sanguinis, and Staphylococcus aureus (PTS).

Results: 16S rRNA sequencing of fecal samples by Ion Torrent and MinION platforms showed colonization of germ-free mice by PTS or by PTS plus LGG (LTS). Although the body weights and feeding rates of mice remained similar between PTS and LTS groups, co-associating LGG with PTS led to a pronounced reduction in abundance of P. acnes in the gut. Addition of LGG or its secretome inhibited P. acnes growth in culture. After optimizing procedures for fecal metabolite extraction and metabolomic liquid chromatography-mass spectrometry analysis, unsupervised and supervised multivariate analyses revealed a distinct separation among fecal metabolites of PTS, LTS, and germ-free groups. Variables-important-in-projection scores showed that LGG colonization robustly diminished guanine, ornitihine, and sorbitol while significantly elevating acetylated amino acids, ribitol, indolelactic acid, and histamine. In addition, carnitine, betaine, and glutamate increased while thymidine, quinic acid and biotin were reduced in both PTS and LTS groups. Furthermore, LGG association reduced intestinal mucosal expression levels of inflammatory cytokines, such as IL-1α, IL-1β and TNF-α.

Conclusions: LGG co-association had a negative impact on colonization of P. acnes, and markedly altered the metabolic output and inflammatory response elicited by pathobionts.

Keywords: Competitive exclusion; Fecal metabolites; Germ-free mice; Inflammation; Liquid chromatography, mass spectrometry; Microbiota; Propionibacterium acnes.

Fig. 1

Comparison of different LC-MS normalization procedures. a Mice were fed for 9 weeks, and feces were sampled in weeks 0, 4, 5, 6, 7, 8 and 9. Mice remained germ-free (GF) throughout. Equal volumes of extraction buffer were added to all fecal samples regardless of the fecal weight. After LC-MS analysis, the data either remained non-normalized, or were normalized to fecal weight (Post-LC-MS). b PCA analysis of the non-normalized (to fecal weight) LC-MS metabolite data. c PCA analysis of the LC-MS metabolite data normalized to fecal weight. d Feces of GF mice were collected three times over two weeks. Prior to LC-MS analysis, they were weighed and the volume of the extraction buffer was adjusted to correspond to the weight of the feces (Pre-LC-MS). e PCA analysis of the fecal metabolites in GF mice used in (d). f Density plot of the CV (coefficient of variation) under different normalization conditions: non-normalized, post-LC-MS normalized, and pre-LC-MS normalized. g Heatmap of non-normalized LC-MS analysis of fecal metabolites from (a). h Heatmap of post-LC-MS-normalized data of fecal metabolites from (a). i Heatmap of pre-LC-MS normalized data of fecal metabolites from (d)

Fig. 2

LGG modifies bacterial and fecal metabolite composition of PTS. a Schematic diagram of the experimental design. Mice remained GF for 12 days, and on day 0, bacteria (PTS or LTS) were gavaged then feces were sampled on days 0 (before gavage), 5, 9, 14 and 19. Metabolites from feces were extracted using the pre-LC-MS procedure outlined above, then subjected to 16S metagenomic sequencing and metabolomics. b Principal coordinate analysis (PCoA) indicating beta diversity of bacterial communities in PTS and LTS mice after gavage. Each point refers to fecal samples in specific mice, with GF6 to 10 mice in the PTS cage, and GF11 to 15 in the LTS cage. c Microbiota composition (relative OTU abundance) at the genus level of feces collected 14th day postgavage from PTS and LTS mice. d Abundance of 16S rRNA (d), LGG (e), and P. acnes (f) in the feces of PTS and LTS mice (n = 5 in each group) 5, 9, and 19 days after gavage. g Images of aggregation of P. acnes alone, LGG alone, and P. acnes and LGG in medium after 48 h of incubation. Percentage of autoaggregation of P. acnes and LGG and coaggregation of P. acnes with LGG (h) after 48 h of incubation

Fig. 3

Metabolomic signatures of feces from GF, PTS and LTS mice. Partial least squares discriminant analysis (PLS-DA) scores plots of fecal metabolites from GF and PTS mice (a) from GF and LTS mice (b) and from PTS and LTS mice (c). There are 30 samples for GF, from 3 sampling days and 2 cages, n = 5 per cage. There are 20 samples each for PTS and LTS, from 4 sampling days and one cage. d Enrichment of metabolic pathways and pathway analysis of t-test significant upregulated metabolites using pathway dataset and pathway analysis of LTS compared to PTS. e Enrichment and pathway analysis of t-test significant downregulated metabolites using pathway dataset and pathway analysis of LTS compared to PTS. f Metabolic pathway significance plotted against pathway impact of metabolites from PTS and LTS mice. The larger the diameter of the symbols, the greater the impact. g Heat map of metabolites illustrating changes in the metabolomic profiles of five GF mice after LTS and PTS gavage. Note the remarkable wide-ranging changes when GF mice were gavaged with bacteria, and the more specific changes between PTS and LTS mice. The top 10 metabolites selected on the basis of VIP scores after comparative analysis of feces from PTS and LTS (h) GF and PTS (i) and of GF and LTS (j) mice. The ordinate represents the VIP scores of metabolites regardless of ionization mode, and of up or down regulation (Supplemental Table 3A to C). If both positive and negative ionization modes of a single metabolite had high VIP scores, only the higher VIP score was used

Fig. 4

Patterns of regulation of identified metabolites in GF, PTS and LTS mice. a Box-whisker plots of representative upregulated metabolites in LTS and PTS mice compared to GF. b Box-whisker plots of representative downregulated metabolites in LTS and PTS mice compared to GF. c Box-whisker plots of representative downregulated metabolites in LTS compared to PTS and/or GF mice. d Box-whisker plots of representative upregulated metabolites in LTS compared to PTS and/or GF mice. * P < 0.05; ** P < 0.025; *** P < 0.001; NS = not significant, P > 0.05

Fig. 5

Patterns of regulation of unidentified metabolites in GF, PTS and LTS mice. Box-whisker plots of representative upregulated, unidentified metabolites in LTS and PTS mice compared to GF (a), of downregulated metabolites in LTS and PTS mice compared to GF (b), of downregulated metabolites in LTS compared to PTS and/or GF mice (c), and of upregulated metabolites in LTS compared to PTS and/or GF mice (d). Mass spectra of selected unidentified metabolites (e, f)

Fig. 6

LGG diminishes inflammatory response to pathobiont association. Relative mRNA expression of intestinal immune-related genes in distal intestinal (a) and colonic (b) mucosa in GF mice associated PTS or LTS (mean ± SEM (n = 5). AREG = amphiregulin. * P < 0.05, ** P < 0.01