Genetic Susceptibility to Enteric Fever in Experimentally Challenged Human Volunteers

Abstract

Infections with Salmonella enterica serovars Typhi and Paratyphi A cause an estimated 14 million cases of enteric fever annually. Here, the controlled nature of challenge studies is exploited to identify genetic variants associated with enteric fever susceptibility. Human challenge participants were genotyped by Illumina OmniExpress-24 BeadChip array (n = 176) and/or transcriptionally profiled by RNA sequencing (n = 174). While the study was underpowered to detect any single nucleotide polymorphisms (SNPs) significant at the whole-genome level, two SNPs within CAPN14 and MIATNB were identified with P < 10-5 for association with development of symptoms or bacteremia following oral S. Typhi or S. Paratyphi A challenge. Imputation of classical human leukocyte antigen (HLA) types from genomic and transcriptomic data identified HLA-B*27:05, previously associated with nontyphoidal Salmonella-induced reactive arthritis, as the HLA type most strongly associated with enteric fever susceptibility (P = 0.011). Gene sets relating to the unfolded protein response/heat shock and endoplasmic reticulum-associated protein degradation were overrepresented in HLA-B*27:05+ participants following challenge. Furthermore, intracellular replication of S. Typhi is higher in C1R cells transfected with HLA-B*27:05 (P = 0.02). These data suggest that activation of the unfolded protein response by HLA-B*27:05 misfolding may create an intracellular environment conducive to S. Typhi replication, increasing susceptibility to enteric fever.

Keywords: HLA antigens; Salmonella Typhi; genomics; single nucleotide polymorphism; transcriptome; typhoid fever; unfolded protein response.

Conflict of interest statement

The authors declare a conflict of interest. A.J.P. is Chair of the UK Department of Health and Social Care's (DHSC) Joint Committee on Vaccination & Immunisation (JCVI) and is a member of the WHO's Strategic Advisory Group of Experts. C.J.B. is currently employed by GlaxoSmithKline.

Figures

FIG 1

Number of participants and samples at each stage of the analysis pipeline.

FIG 2

(a) Population structure (principal components 1, 2, and 3) of the enteric fever cohorts within the context of the 1000 Genomes superpopulations. Each point corresponds to one individual, with individuals from the enteric fever cohorts highlighted in purple (light purple = remained healthy following challenge, dark purple = developed enteric fever following challenge) and those from the 1000 Genomes project in blue (differing shades correspond to different superpopulations). Where enteric fever challenge participants do not cluster with the 1000 Genomes European superpopulation, self-reported ethnicity is indicated. (b) Self-reported demographics (sex and ethnicity) of participants included in the GWAS and HLA analyses, by study. (c) Manhattan plot showing the significance [−log10(unadjusted P value)] of the relationship between each single nucleotide polymorphism (SNP) and development of symptoms or bacteremia following oral S. Typhi or S. Paratyphi A challenge, for each chromosome. The dashed line indicates a suggestive P value of 10−5. The five SNPs with the lowest P values are highlighted, with the overlapping gene as identified by SNPnexus indicated as well as the odds ratio (OR).

FIG 3

(a) Difference in HLA dosage versus mean HLA dosage for four participants with an outlying time point. For each participant and each HLA type (2-digit resolution), the mean dosage of each HLA type was calculated, as well as the amount by which each time point deviated from the mean. Each point represents an HLA type at a certain time point, color coded by time point. Points which were subsequently excluded due to belonging to an outlying time point are circled in red. The thresholds at which points deviate >50% from the mean are indicated. (b) Difference in HLA dosage versus mean HLA dosage for all participants with multiple time points profiled (n = 50) following exclusion of outlying time points. For each participant and each HLA type (2-digit resolution), the mean dosage of each HLA type was calculated, as well as the amount by which each time point deviated from the mean. Each point represents an HLA type in a certain participant at a certain time point, color coded by participant. The thresholds at which points deviate >50% from the mean are indicated. (c) Intraclass correlation coefficients (one way, single measurement) for agreement between HLA type (2-digit resolution) dosages at different time points, as calculated by the R package irrNA. Each point represents the intraclass coefficient for one HLA type, with 95% confidence intervals indicated by error bars. (d) Agreement (weighted Cohen’s kappa) between SNP2HLA and HISAT-genotype for participants HLA typed by both methods (n = 71), as calculated by the R package irr. Each point represents the weighted Cohen’s kappa for one HLA type (2-digit resolution), with 95% confidence intervals indicated by error bars. The strength of agreement for each range of kappa, as assigned in the work of Landis and Koch (14), is indicated.

FIG 4

(a) Relative frequency of each HLA type at a resolution of 2 digits for HLA-A, HLA-B, HLA-C, HLA-DQA1, HLA-DQB1, and HLA-DRB1 in the entire combined cohort, including participants from the typhoid dose-finding study, typhoid oral vaccine trial, typhoid Vi vaccine trial, paratyphoid dose-finding study, and typhoid toxin study. (b) Odds ratios (odds ratio of >1 indicates association with susceptibility and of P values are indicated for each. (c) Odds ratios for the two HLA-B*27 subtypes at a resolution of 4 digits with 95% confidence intervals. P values are indicated for each. (d) Percentage of participants who were diagnosed with enteric fever following challenge, stratified by the presence or absence of one copy of HLA-B*27:05. The proportion of participants diagnosed is indicated for each group.

FIG 5

(a) CFU per milliliter recovered from C1R cells infected with S. Typhi Quailes strain, in the presence or absence of HLA-B*27 expression, 24 h postinfection. Parent and HLA-B*27:05+ cells were seeded in a 96-well plate at a density of 100,000 cells per well and infected with the S. Typhi Quailes strain at an MOI of 0 or 10 in triplicate. After 1 h gentamicin was added to kill extracellular bacteria. At 24 h postinoculation, cells were lysed using 1% Triton X-100, and lysates were serially diluted and plated onto tryptone soya agar. Colonies were counted following overnight incubation at 37°C. A P value for a t test is indicated. Points represent replicates within a single experiment. (b) Volcano plot showing the log2(fold difference) in gene expression between HLA-B*27:05-positive and -negative participants 12 h postchallenge against the −log10(P value). A dashed line indicating where P equals 0.05 is shown, and genes relating to the unfolded protein response and heat shock proteins are highlighted. Genes more highly expressed in participants who were HLA-B*27:05 positive are shown further to the right, and those more highly expressed in HLA-B*27:05-negative participants are shown further to the left. RNA expression was characterized by RNA sequencing. Data were filtered, normalized, and transformed, and differential expression was then assessed using the limma R package, using participant ID, sequencing pool, vaccination status, challenge strain, and dose as blocking variables. (c) Expression of MICA and CALR following normalization and transformation using the edgeR and limma packages, in HLA-B*27:05-positive and -negative participants at baseline and 12 h postchallenge. Nominal P values before adjustment for multiple testing are indicated. (d) Normalized enrichment scores for gene sets relating to the unfolded protein response (UPR), heat shock, and endoplasmic reticulum-associated protein degradation (ERAD) 12 h following enteric fever challenge in participants heterozygous for HLA-B*27:05 relative to those without HLA-B*27:05. Genes were ranked by t statistic, and relevant gene sets were downloaded from the Molecular Signatures Database. A custom gene set based on the genes highlighted in panel b was also included. Gene set enrichment analysis was carried out on the preranked list using GSEA_4.1.0. Bars are color coded and ordered by significance, and nominal P values are indicated.