Metabolomic analysis reveals an important role of sphingosine 1-phosphate in the development of HFMD due to EV-A71 infection

Fig 1. Metabolomic profiling of HFMD plasma. (A and B) The positive and negative ion mode chromatograms of the three groups of samples. (C) The cluster analysis of metabolic profiles by heatmap among different groups. Red is increasing, blue is decreasing across each row. (D-a) PCA score plots, (D-b) PLS-DA score plots, and (D-c) OPLS-DA score plots based on the data from the HCs (green), mild cases (blue), and severe cases (red). (E) Venn diagram analysis identifies differentiated metabolites across four comparison strategies (HCs vs mild cases; HCs vs severe cases; mild cases vs severe cases; HCs vs mild cases vs severe cases).

Fig 2. Differential metabolic profiles and metabolic pathways across HCs, mild and severe EV-A71 patients. (A) Heatmap visualization of hierarchical clustering based on significantly altered metabolites (VIP > 1 and P < 0.05 and fold changes ≥ 1.5 or ≤0.667 between groups). Metabolic pathway analysis was launched using identified differential metabolites. The color and size of each circle are based on its P value and pathway impact value, respectively. (B) Mild cases vs severe cases; (C) HCs vs mild cases vs severe cases. (D) Metabolic pathway map of the major differential metabolites.

Fig 3. Potential predictive values of differential metabolites for the onset of severe EV-A71 infection. Six elected metabolites that significantly differed between mild, severe, and HCs. (A) Eicosapentaenoyl PAF C-16, (B) α-ketoisovaleric acid, (C) L-phenylalanine, (D) S1P, (E) creatine, and (F) choline. The logistic regression model was constructed for calculating mild group and severe group diagnosis score. (G) The corresponding ROC of the logistic regression model. (H) The holdout cross-validation analysis for 100 times. ROC curves showing the performance of models. AUC indicates the area under the curve, and CI indicates the confidence interval. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Fig 4. Quantification of unique plasma metabolites of severe EV-A71 patients. The quantification of unique plasma metabolites in different groups (mild, severe, and HCs) and subgroups between mild and severe cases. Mild and severe patients were then divided into subgroups according to the number of days after onset: 2, 3, 4, and ≥5 days. The number of mild cases in each subgroup was 16 (2 days after onset), 13 (3 days), 11 (4 days), and 5 (≥5 days). The number of severe cases in each subgroup was 12 (2 days), 7 (3 days), 12 (4 days), and 8 (≥5 days). (A) Eicosapentaenoyl PAF C-16, (B) α-ketoisovaleric acid, (C) L-phenylalanine, (D) S1P, (E) creatine, and (F) choline. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Fig 5. Correlation analysis between S1P and some clinical indicators. The concentration of serum S1P showed positive correlation with (A) neutrophils, (B) total protein, (C) globulin, (D) hs-CRP, (E) WBC, and (F) glucose. The r value represents the correlation coefficient between the two variables.

Fig 6. Regulation of S1P metabolism can improve the survival rate of infected mice. (A) Serum S1P levels of infected and control mice were measured by ELISA at 2 and 5 dpi (N = 6–9 per group). (B) A schematic diagram to explain the animal experiment. Following treatment as indicated in the schematic diagram, (C) mean clinical score and (D) survival rate were monitored daily for 15 days (N = 10–17 per group, vs EV-A71). Clinical scores were defined as follows: 0, healthy; 1, reduced mobility; 2, ruffled hair, hunchbacked, or ataxia; 3, weight loss; 4, limb weakness; and 5, dying or death. *P < 0.05; **P < 0.01; ns, no statistical differences. Created with BioRender.com.

Fig 7. Targeting S1P-metabolism may regulate immune function in the brain. (A) A schematic diagram to indicate the procedures for intervention and collecting samples. (B) The blood routine was detected by an automatic blood cell analyzer at 2 dpi (N = 5 per group). The main detection indicators included a: WBC, b: monocyte (MON), c: lymphocyte (LYM), d: granulocyte (Gran), e: red blood cell (RBC), and f: platelet (PLT). (C) The numbers of CD45 positive cells in brain tissues were measured by flow cytometer at 2 dpi (N = 4 per group). (D) The results of flow cytometry were analyzed. (E) qPCR was conducted to measure the gene expression levels of some typical inflammatory cytokines (a, IL-1β; b, IL-6; c, MCP-1; d, TNFα) in the brains of EV-A71-infected and mock mice with MK571 interventions at 2 dpi (N = 3–4 per group). (F) Pathological changes in the brains of mice at 2 dpi were evaluated by H&E staining. The scales are indicated in these images. *P < 0.05; **P < 0.01; ns, no statistical differences. Created with BioRender.com.