Characteristics of untargeted metabolomics of common bacterial pathogens in respiratory diseases

To the Editor: Common respiratory diseases such as chronic obstructive pulmonary disease (COPD) and asthma induce significant health burdens worldwide.[1,2] Airway/lung microbiome and microbial metabolites have been proven to be closely related to multiple respiratory diseases.[3] However, few relevant studies show characteristics and horizontal comparisons of metabolites of dominant bacteria in respiratory diseases. Additionally, most results are based on association analyses of microbiome sequences, especially those of the gut microbiome, and untargeted metabolomics, without evidence that the metabolites are directly produced and secreted by certain bacteria.[4] In this context, our study aimed to analyze and compare the key metabolites and metabolic pathways of common bacteria involved in respiratory diseases. Based on the culture results of bacteria in clinical samples of patients with respiratory diseases and the reports of cohort studies,[5] we analyzed metabolic changes of five common bacteria, namely, Streptococcus pneumoniae (Sp), Staphylococcus aureus (Sa), Moraxella catarrhal (Mc), Pseudomonas aeruginosa (Pa), and Haemophilus influenzae (Hin). Firstly, we cultivated type strains of the five bacteria until the optical density at 600 nm (OD600) reached 1.0, collected their supernatants [Supplementary Materials, https://links.lww.com/CM9/C341], and then conducted untargeted metabolomics analysis of supernatants to investigate the metabolic profile under the same conditions, attempting to minimize the influence of variabilities and unknown factors. The metabolites with variable influence in projection (VIP) >1 and P <0.05 and fold change (FC) ≥2 or FC ≤0.5 were considered differentially abundant metabolites. Then, we compared the similarity and specificity of differentially abundant metabolites among these bacteria to identify possible key metabolites and related pathways, which might provide a shortcut for further exploration of the role of bacterial metabolites in respiratory diseases. Metabolites annotated by each secondary classification under the primary classification of Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were shown in the positive and negative ion modes. In the positive ion mode, most metabolites were annotated to biological pathways such as metabolism. The top 5 secondary classifications included global and overview maps (100 metabolites), amino acid metabolism (47 metabolites), metabolism of cofactors and vitamins (20 metabolites), nucleotide metabolism (17 metabolites), and metabolism of other amino acids (16 metabolites), accounting for 81.97% (200/244) of the metabolites annotated to the metabolism pathway. In the negative ion mode, most metabolites were annotated to biological pathways such as metabolism again. The top 5 secondary classifications included global and overview maps (107 metabolites), amino acid metabolism (38 metabolites), carbohydrate metabolism (29 metabolites), nucleotide metabolism (27 metabolites), and lipid metabolism (22 metabolites), accounting for 76.90% (223/290) of the metabolites annotated to the metabolism pathway. The metabolites annotated by superclass in the human metabolome database (HMDB) in the positive ion mode were as follows: hydrocarbon derivatives (1 metabolite), alkaloids and derivatives (6 metabolites), organic nitrogen compounds (11 metabolites), phenylpropanoids and polyketides (14 metabolites), nucleosides, nucleotides, and analogues (20 metabolites), organic oxygen compounds (20 metabolites), benzenoids (34 metabolites), lipids and lipid-like molecules (58 metabolites), organoheterocyclic compounds (65 metabolites), and organic acids and derivatives (98 metabolites). In the negative ion mode, the metabolites annotated by the superclass in the HMDB were as follows: alkaloids and derivatives (3 metabolites), organic nitrogen compounds (5 metabolites), phenylpropanoids and polyketides (10 metabolites), benzenoids (19 metabolites), organic oxygen compounds (26 metabolites), nucleosides, nucleotides, and analogues (29 metabolites), organoheterocyclic compounds (32 metabolites), organic acids and derivatives (64 metabolites), lipids and lipid-like molecules (68 metabolites). The results of screening of differentially abundant metabolites revealed that between the supernatants of the Sp group and the control medium group, 432 differentially abundant metabolites were identified in the positive ion mode, including 176 upregulated and 256 downregulated metabolites. In the negative ion mode, 249 differentially abundant metabolites were identified, including 137 upregulated and 112 downregulated metabolites. Among the supernatants of the Sa group and the control medium group, 103 differentially abundant metabolites were identified in the positive ion mode, including 38 upregulated and 65 downregulated metabolites. In the negative ion mode, 56 differentially abundant metabolites were identified, including 16 upregulated and 40 downregulated metabolites. Between the supernatants of Mc group and the control medium group, 137 differentially abundant metabolites were identified in the positive ion mode, including 71 upregulated and 66 downregulated metabolites. In the negative ion mode, 76 differentially abundant metabolites were identified, including 33 upregulated and 43 downregulated metabolites. Among the supernatants of the Pa group and the control medium group, 407 differentially abundant metabolites were identified in the positive ion mode, including 178 upregulated and 229 downregulated metabolites. In the negative ion mode, 253 differentially abundant metabolites were identified, including 123 upregulated and 130 downregulated metabolites. Among the supernatants of the Hin group and the control medium group, 133 differentially abundant metabolites were identified in the positive ion mode, including 73 upregulated and 60 downregulated metabolites. In the negative ion mode, 81 differentially abundant metabolites were identified, including 31 upregulated and 50 downregulated metabolites. We then compared overlapping and unique differentially abundant metabolites among different bacteria [Figure 1]. In the positive ion mode, there were no common metabolites among the five kinds of bacteria. However, 114, 18, 33, 76, and 46 metabolites were found to be specific for Sp, Sa, Mc, Pa, and Hin, respectively. The top 5 upregulated metabolites in the supernatant of Sp were thioguanine, bicyclo prostaglandin E2, triacanthine, 4-hydroxyretinoic acid, and xanthurenic acid, whereas the top 5 downregulated metabolites were alanine-leucine (ala-leu), 3′-hydroxy-stanozolol, maltotetrose, 6-hydroxymelatonin, and S-adenosylhomocysteine. The top 5 upregulated metabolites in the supernatant of Sa were dihydroceramide (d16:0/16:0), 8-hydroxy-eicosapentaenoic acid, digalactosyl monoacylglycerol (18:2), tetramethylpyrazine, and norfentanyl, whereas the top 5 downregulated metabolites were ergothioneine, cafestol, trans-2-butene-1,4-dicarboxylic acid, 4-hydroxy estrone 1-N3-adenine, and prostaglandin A3. The top 5 upregulated metabolites in the supernatant of Mc were L-tyrosine methyl ester, urocanic acid, 6-aminonicotinamide, T-2 triol, and haloperidol-d4, whereas the top 5 downregulated metabolites were pantethine, ingenol-3-angelate, UR-144 N-(2-hydroxypentyl) metabolite, all-trans-retinal and gatifloxacin. The top 5 upregulated metabolites in the supernatant of the Pa were sterigmatocystin, astragaloside IV, 6β-naltrexol-d3, norbutorphanol, and milbemycin A4 oxime, whereas the top 5 downregulated metabolites were stearamide, N-acetyl-L-tyrosine, L-asparagine, L-adrenaline, and S-adenosyl-L-methionine. The top 5 upregulated metabolites in the supernatant of Hin belonged to phosphatidyl ethanolamine, including PE (12:0/12:0), PE (8:0/8:0), PE (9:0/9:0), PE (11:0/11:0) and PE (11:0/13:1), whereas the top 5 downregulated metabolites were spermidine, cytosine, cytidine 5′-monophosphate (hydrate), N4-acetylcytidine, and gedunin.Figure 1: Common and specific differentially abundant metabolites among supernatants of different bacteria in the positive (A) and negative (B) ion modes. BHI: Brain-heart infusion medium; Hin: Haemophilus influenzae; Mc: Moraxella catarrhal; Pa: Pseudomonas aeruginosa; Sa: Staphylococcus aureus; Sp: Streptococcus pneumoniae.In the negative ion mode, docosapentaenoic acid and 8,11,14-eicosatrienoic acid were two common metabolites among the five kinds of bacteria. In addition, there were 75, 5, 14, 53, and 37 metabolites specific for Sp, Sa, Mc, Pa, and Hin, respectively. The top 5 upregulated metabolites in the supernatant of Sp were deoxyribose 5-phosphate, d-mannose 6-phosphate, d-glucose 6-phosphate, chaetocin, and piceatannol, whereas the top 5 downregulated metabolites were stachyose, methylmalonate, pantothenic acid, serotonin, and xanthosine. The upregulated metabolites in the supernatant of Sa were N-lauroylsarcosine, thromboxane B1, inosine-5′-monophosphate and N-acetyl-L-histidine, whereas the only one downregulated metabolite was 5′-adenylic acid. The top 5 upregulated metabolites in the supernatant of Mc were guanosine 5′-diphosphate (GDP), 18-hydroxy-eicosapentaenoic acid, nicotinamide N-oxide, 19-nortestosterone and geranylgeranyl pyrophosphate, whereas the top downregulated metabolites were oleoyl ethanolamide, bilirubin, stearic acid and cymarin. The top 5 upregulated metabolites in the supernatant of Pa were 3-hydroxydecanoic acid, 12-hydroxydodecanoic acid, 3-hydroxypicolinic acid, cholic acid, and protoporphyrin IX, whereas the top 5 downregulated metabolites were adenylosuccinic acid, adenine, quinic acid, 3-(3-methoxyphenyl) propionic acid and 7-hydroxy-3,4-dihydrocarbostyril. The top 5 upregulated metabolites in the supernatant of Hin belonged to phosphatidyl ethanolamine class, including PE (12:0/12:0), PE (8:0/8:0), PE (9:0/9:0), PE (11:0/11:0), and PE (11:0/13:1), and top 5 downregulated metabolites were uridine monophosphate (UMP), guanosine-3′,5′-cyclic monophosphate, adenosine-3′,5′-cyclic monophosphate, biopterin and ribulose-5-phosphate. Metabolic pathway enrichment analysis revealed that in the positive ion mode, there were no significantly enriched metabolic pathways of differentially abundant metabolites in supernatants of Sp and Pa. The most significantly enriched pathways were purine metabolism in the supernatant of Sa and arachidonic acid metabolism in the supernatant of Mc. The significantly enriched pathways of the differentially abundant metabolites in the supernatant of Hin were purine metabolism and biosynthesis of unsaturated fatty acids, respectively. In the negative ion mode, there were no significantly enriched metabolic pathways of differentially abundant metabolites in the supernatants of Sp and Pa. The most enriched significant pathway was purine metabolism in the supernatant of Sa. The most enriched pathways in the supernatant of Mc were biosynthesis of unsaturated fatty acids, nicotinate and nicotinamide metabolism, fatty acid biosynthesis, and carbon fixation in photosynthetic organisms, respectively. The significantly enriched pathways of differentially abundant metabolites in the supernatant of Hin were also identified, including pathways such as alanine, aspartate and glutamate metabolism, the citrate cycle, carbon fixation pathways in prokaryotes, β-alanine metabolism, furfural degradation, nicotinate and nicotinamide metabolism, histidine metabolism, amino sugar and nucleotide sugar metabolism, glyoxylate and dicarboxylate metabolism, and the biosynthesis of unsaturated fatty acids. In summary, we conducted untargeted metabolomics on common bacteria detected clinically in respiratory diseases and described the features of their metabolic profiles. We also screened a series of common and specific differentially abundant metabolites of each bacterium. We hope these data will provide the basis for further exploring the role of microbiome metabolites in the pathogenesis, diagnosis, and prevention of respiratory diseases. Acknowledgments We thank Robin James Storer, PhD, from Liwen Bianji (Edanz) (www.liwenbianji.cn) for editing the English text of a draft of this manuscript. Funding This study was supported by grants from the National Natural Science Foundation of China (Nos. 82090013, 82071805, and 82241011) and Beijing Natural Science Foundation (Nos. 7212001, 7242010, and 7242002). Conflicts of interest None.