Early transcriptional responses to human enteric fever challenge

Abstract

Enteric fever, caused by oral infection with typhoidal Salmonella serovars, presents as a non-specific febrile illness preceded by an incubation period of 5 days or more. The enteric fever human challenge model provides a unique opportunity to investigate the innate immune response during this incubation period, and how this response is altered by vaccination with the Vi polysaccharide or conjugate vaccine. We find that on the same day as ingestion of typhoidal Salmonella, there is already evidence of an immune response, with 199 genes upregulated in the peripheral blood transcriptome 12 hours post-challenge (false discovery rate <0.05). Gene sets relating to neutrophils, monocytes, and innate immunity were over-represented (false discovery rate <0.05). Estimating cell proportions from gene expression data suggested a possible increase in activated monocytes 12 hours post-challenge (P = 0.036, paired Wilcoxon signed-rank test). Furthermore, plasma TNF-α rose following exposure (P = 0.011, paired Wilcoxon signed-rank test). There were no significant differences in gene expression (false discovery rate <0.05) in the 12 hours response between those who did and did not subsequently develop clinical or blood culture confirmed enteric fever or between vaccination groups. Together, these results demonstrate early perturbation of the peripheral blood transcriptome after enteric fever challenge and provide initial insight into early mechanisms of protection.

Keywords: Salmonella; enteric fever; human challenge; transcriptomic; typhoid.

Fig 1

Schematic of the studies and samples analyzed in the following paper. Samples from four enteric fever human challenge studies underwent transcriptional profiling, while plasma samples from two studies were processed and stored for multiplex cytokine assays at the timepoints indicated.

Fig 2

(a) Volcano plot [log₂(fold change) versus –log₁₀(unadjusted P value)] of genes 12 hours post-challenge relative to baseline in a meta-analysis carried out using MetaVolcanoR. A log₂(fold change) >0 indicates upregulation. Non-significant genes are colored light yellow, whereas genes with adjusted P value < 0.05 are colored by their sign consistency (e.g., a sign consistency of three corresponds to upregulation in all three data sets). Ninety-five percent confidence intervals in log₂(fold change) for genes with P value < 0.05 are indicated by error bars. (b) Log₂(fold change) and 95% confidence intervals for each of the top five most upregulated and downregulated genes for each individual data set and meta-analysis. (c) Log₂(fold change) of genes identified in a meta-analysis as being differentially expressed 12 hours post-challenge, 12 and 24 hours post-challenge in the microarray typhoid dose-finding/typhoid oral vaccine trial data set. The log₂(fold change) of the same genes in both of the circadian rhythms data sets are shown for comparison. Upregulated genes are colored red and downregulated genes blue. The mean is indicated by a horizontal line.

Fig 3

(a) Normalized enrichment scores of significantly (adjusted value P < 0.05) over- or under-represented blood transcriptional modules at 12 hours post-challenge relative to baseline. Over-represented modules are colored red and under-represented modules blue. Adjusted P values are indicated by the labels. (b) Log₂(fold change) in the number of neutrophils estimated by CIBERSORT 12 hours post-challenge relative to baseline. (c) Log₂(fold change) in the number of monocytes estimated by CIBERSORT 12 hours post-challenge relative to baseline. (d) Log₂(fold change) in the number of activated monocytes estimated by the deconvolution algorithm developed by Ben–Moshe et al. Twelve hours post-challenge relative to baseline. Each point represents one participant, colored by study cohort. The log₂(fold change) in the circadian rhythms data sets are shown for comparison. The median is indicated by a horizontal line. White labels indicate the P values for a paired Wilcoxon-signed rank test comparing the 12 hours timepoint with baseline.

Fig 4

(a) The log₂(fold change) in TNF-α 12 hours after baseline in unchallenged control participants and challenged participants. Each point represents one participant. For each cytokine and group, the median is indicated by a horizontal line. White labels indicate the P values for a paired Wilcoxon-signed rank test comparing concentrations at the 12 hours timepoint with baseline. (b) TNF-α production by granulocytes, monocytes, and lymphocytes from three donors stimulated with S. Typhi at a multiplicity of infection of 10 after 4 hours relative to an unstimulated control. The P value for a linear mixed effects model with blood donor modeled as a batch is indicated. Points are colored by donor.

Fig 5

(a) Volcano plot [log₂(fold difference) versus –log₁₀(unadjusted P value)] of baseline gene expression in those who remained healthy versus those who went on to develop enteric fever in a meta-analysis carried out using MetaVolcanoR. A log₂(fold difference) >0 indicates a gene having higher expression in those who remained healthy. Non-significant genes after adjusting for multiple testing are colored light yellow. (b) Volcano plot [log₂(fold difference) versus –log₁₀(unadjusted P value)] in the 12 hours transcriptional response in those who remained healthy versus those who went on to develop enteric fever [i.e., (Remained healthy 12 hours – Remained healthy baseline) – (Developed enteric fever 12 hours – Developed enteric fever baseline)]. A log₂(fold difference) >0 indicates a greater response in those who remained healthy. Non-significant genes after adjusting for multiple testing are colored light yellow. (c) Differences in cell counts estimated by deconvolution between those who remained healthy and those who developed enteric fever. In the top panels, baseline differences in the estimated amount of cells is shown, with a P value for a logistic regression adjusting for study as a covariate labeled. In the lower panels, differences in the log₂(fold change) at 12 hours post-challenge relative to baseline is shown, with a P value for a Mann-Whitney test labeled. Each point represents one participant, colored by study. The median for each study is indicated by a colored bar, and the difference in median between the two groups in each study is highlighted by a colored line.

Fig 6

(a) Number of differentially expressed genes for each comparison between vaccine groups in the typhoid parenteral vaccine study. Numbers both before (P < 0.05) and after adjustment for multiple testing (FDR < 0.05) are shown. (b) Log₂(fold change) in activated monocytes 12 hours post-challenge relative to baseline, grouped by vaccination status. White labels indicate the P values for a paired Wilcoxon-signed rank test comparing the 12 hours timepoint with baseline. Each point represents one participant, colored by study. The mean log₂(fold change) is indicated by a black bar. (c) Log₂(fold change) in activated monocytes 12 hours post-challenge relative to baseline, grouped by outcome of challenge and vaccination status. The P value for a Mann-Whitney test comparing the log₂(fold changes) between groups is shown. Each point represents one participant, colored by study. The mean log₂(fold change) is indicated by a black bar.