Interactions between gastric microbiota and metabolites in gastric cancer

Abstract

The development and progression of gastric cancer (GC) is greatly influenced by gastric microbiota and their metabolites. Here, we characterized the gastric microbiome and metabolome profiles of 37 GC tumor tissues and matched non-tumor tissues using 16s rRNA gene sequencing and ultrahigh performance liquid chromatography tandem mass spectrometry, respectively. Microbial diversity and richness were higher in GC tumor tissues than in non-tumor tissues. The abundance of Helicobacter was increased in non-tumor tissues, while the abundance of Lactobacillus, Streptococcus, Bacteroides, Prevotella, and 6 additional genera was increased in the tumor tissues. The untargeted metabolome analysis revealed 150 discriminative metabolites, among which the relative abundance of the amino acids, carbohydrates and carbohydrate conjugates, glycerophospholipids, and nucleosides was higher in tumor tissues compared to non-tumor tissues. The targeted metabolome analysis further demonstrated that the combination of 1-methylnicotinamide and N-acetyl-D-glucosamine-6-phosphate could serve as a robust biomarker for distinction between GC tumors and non-tumor tissues. Correlation analysis revealed that Helicobacter and Lactobacillus were negatively and positively correlated with the majority of differential metabolites in the classes of amino acids, carbohydrates, nucleosides, nucleotides, and glycerophospholipids, respectively, suggesting that Helicobacter and Lactobacillus might play a role in degradation and synthesis of the majority of differential metabolites in these classes, respectively. Acinetobacter, Comamonas, Faecalibacterium, Sphingomonas, and Streptococcus were also significantly correlated with many differential amino acids, carbohydrates, nucleosides, nucleotides, and glycerophospholipids. In conclusion, the differences in metabolome profiles between GC tumor and matched non-tumor tissues may be partly due to the collective activities of Helicobacter, Lactobacillus, and other bacteria, which eventually affects GC carcinogenesis and progression.

Fig. 1. Altered gastric microbiota in 37 gastric cancer (GC) tissues compared with matched non-tumor tissues.

A, B The observed OTUs and Shannon indices were used to evaluate the microbial diversity of the paired tumor and non-tumor tissues. The Wilcoxon matched-pairs signed rank test was performed. C Principal coordinate analysis (PCoA) of the weighted UniFrac distance demonstrated that the non-tumor and tumor tissues showed two distinct clusters. D The Venn diagram illustrates the overlapped OTUs between the paired GC tumor tissues and non-tumor tissues. E Differential taxa at the phylum and genus levels identified by linear discriminant analysis (LDA) effect size (LEfSe) analysis (LDA > 3.5, Q < 0.05).

Fig. 2. The metabolome profiles of gastric cancer tissues were different from those of matched non-tumor tissues.

A PLS-DA showed that tumor tissues and non-tumor tissues were separated into two distinct clusters. B The test for PLS-DA model showed that the PLS-DA model for this study was valid. PLS-DA, partial least-squares discriminant analysis. QC, quality control. The QC samples were obtained by mixing the equal amounts of metabolites extracted from all samples, which were used for evaluation of the stability of the instrument.

Fig. 3. The heat map shows the differential metabolites between the paired gastric cancer tissues and non-tumor tissues.

The heat map shows the scaled relative abundance (Lg) of 109 differential metabolites (VIP > 1 and Q value < 0.05 and fold change (FC) ≥ 2 or FC ≤ 0.5). The discriminative metabolites from top to bottom are amino acids, carbohydrates and carbohydrate conjugates, indoles and derivatives, nucleosides, nucleotides, steroids and derivatives, fatty acyls, glycerolipids, and glycerophospholipids. The differential metabolites were classified using the Human Metabolome Database. Q value, corrected P value.

Fig. 4. Identification of metabolite biomarkers for discriminating gastric tumor tissues from non-tumor tissues.

A The top 15 metabolites according to VIP values are displayed. VIP, variable importance in projection. B Among the 15 metabolites, the relative abundance of 8 metabolites was higher, but the relative abundance of the rest of the metabolites was lower in the tumor tissues than the non-tumor tissues. Q value, corrected P value. C ROC analysis for 1-methylnicotinamide, N-acetyl-D-glucosamine-6-phosphate, and the combination of the two metabolites. ROC, receiver operating curve.

Fig. 5. Validation of metabolite biomarkers for distinguishing between gastric tumor and non-tumor tissues.

A The concentrations of 1-methylnicotinamide was compared between tumor tissues and matched non-tumor tissues from 20 patients with gastric cancer (GC). B Comparison of the concentration of N-acetyl-D-glucosamine-6-phosphate between 20 paired GC tumor and non-tumor tissues. C–E ROC analysis for 1-methylnicotinamide (C), N-acetyl-D-glucosamine-6-phosphate (D), and the combination of the two metabolites (E). ROC, receiver operating curve.

Fig. 6. Pathway analysis for metabolites and integrated analysis of microbiota and metabolites.

A The functions of these metabolites and metabolic pathways were studied using the KEGG database, and enriched pathways were displayed by a bubble plot. B The association between 11 discriminative genera and 25 differential metabolites in the main enriched pathways were analyzed using the Spearman’s correlation method. Red, positive correlations; blue, negative correlations. *P value < 0.05; **P value < 0.01.

Fig. 7. The analysis of correlation between differential genera and metabolites in different classes.

A–D The analysis of correlation between 11 differential genera and discriminative metabolites in the classes of fatty acyls (n = 24) (A), differential glycerophospholipids (n = 29) (B), amino acids (n = 21) (C), carbohydrates (n = 12), nucleosides (n = 7), and nucleotides (n = 4) (D). Red, positive correlations; blue, negative correlations. *P value < 0.05; **P value < 0.01.

Fig. 8. The heat map shows the association between metabolites and tumor stage.

The differences in metabolites among non-tumor tissues (n = 37), early-stage (stage I–II, n = 16) and late-stage gastric tumor tissues (stage III, n = 21) were displayed. The heat map shows the scaled relative abundance (Lg) of 109 metabolites.