Host response to influenza infections in human blood: association of influenza severity with host genetics and transcriptomic response

Abstract

Introduction: Influenza virus infections are a major global health problem. Influenza can result in mild/moderate disease or progress to more severe disease, leading to high morbidity and mortality. Severity is thought to be primarily driven by immunopathology, but predicting which individuals are at a higher risk of being hospitalized warrants investigation into host genetics and the molecular signatures of the host response during influenza infections.

Methods: Here, we performed transcriptome and genotype analysis in healthy controls and patients exhibiting mild/moderate or severe influenza (ICU patients). A unique aspect of our study was the genotyping of all participants, which allowed us to assign ethnicities based on genetic variation and assess whether the variation was correlated with expression levels.

Results: We identified 169 differentially expressed genes and related molecular pathways between patients in the ICU and those who were not in the ICU. The transcriptome/genotype association analysis identified 871 genes associated to a genetic variant and 39 genes distinct between African-Americans and Caucasians. We also investigated the effects of age and sex and found only a few discernible gene effects in our cohort.

Discussion: Together, our results highlight select risk factors that may contribute to an increased risk of ICU admission for influenza-infected patients. This should help to develop better diagnostic tools based on molecular signatures, in addition to a better understanding of the biological processes in the host response to influenza.

Keywords: DEGs; QTLs; genotype; human; influenza; transcriptome.

Figure 1

Principal component analysis of transcriptome expression. Principal component analysis plot for gene expression values of infected patients and healthy controls. Abbreviations: HC: healthy controls; inf_ICU infected patients at ICU; inf_non-ICU: infected patients not at ICU.

Figure 2

Comparison of infected patients versus healthy controls. (A) Volcano plot of infected patients versus healthy controls. y-axis: -log₁₀ BH multiple testing adjusted p-values, x-axis: log₂ fold change. DEGs are colored red, and the top 20 up- and down-regulated (by log-fold change) DEGs are labeled. Blue: not significant genes with an adjusted p-value < 0.05. Yellow: not significant genes with an absolute fold change of 1.5 (log₂ = 0.5849625). Grey: NS, not significant. (B) Functional analysis using GO term enrichment for the up-regulated DEGs from the contrast of infected versus healthy controls. (C) Functional analysis using GO term enrichment for the down-regulated DEGs from the contrast of infected versus healthy controls.

Figure 3

Comparison of infected non-ICU patients and ICU patients versus healthy controls, ICU versus non-ICU patients and corresponding VENN diagram. (A) Volcano plot of infected non-ICU patients versus healthy controls. (B) Volcano plot of infected ICU patients versus healthy controls. (C) Volcano plot of infected ICU patients versus infected non-ICU patients. See Figure 2 for more details on volcano plots. (D) Venn diagram illustrating the overlaps between the DEGs from contrasts of ICU and non-ICU patients versus healthy controls and between ICU and non-ICU patients. A total of 2,022 DEGs were identified in all three groups (all genes combined), and 75 DEGs were commonly shared between the three groups (central overlap.

Figure 4

Pathway analysis of infected ICU and non-ICU patients versus healthy controls. (A) Functional analysis using GO term enrichment for the up-regulated DEGs from the contrast of infected non-ICU patients versus healthy controls. (B) Functional pathway analysis using GO term enrichment for the up-regulated DEGs from the contrast of infected ICU patients versus healthy controls. (C) Functional pathway analysis using GO term enrichment for the down-regulated DEGs from the contrast of infected ICU patients versus healthy controls.

Figure 5

Pathway analysis of infected ICU versus non-ICU patients. (A) Functional analysis using GO term enrichment for the up-regulated DEGs (higher in ICU) from the contrast of infected ICU patients versus non-ICU patients. (B) Functional pathway analysis using GO term enrichment for the down-regulated DEGs (higher in non-ICU) from the contrast of infected ICU patients versus non-ICU patients.

Figure 6

MDS plots of genotypes. (A) MDS plot showing all samples, with reference samples colored and participants in gray. (B) MDS plot of participants analyzed in this study (only a single representation for each participant, no reference genomes), colored for ethnicity. Abbreviations: African-American (AfAm), Mexican American Indians (Mex_AmInd), Mexican ancestry (Mexican anc), Indian American (Indian_Am).

Figure 7

Principal component analysis and gene expression by genotype. (A) Principal component analysis plot for gene expression values of infected participants, colored by ethnicity. Abbreviations: African-American (AfAm), Caucasian (Caucs), Mexican American Indians (Mex_AmInd), Mexican ancestry (Mexican anc), Indian American (Indian_Am), no unambiguous assignment (admixed). (B) Volcano plot of DEGs for contrasts of infected Caucasian versus African-American patients. See Figure 2 for more details on volcano plots.

Figure 8

Manhattan plot of cis-eQTL analysis. Manhattan plot illustrating the results of cis-eQTL analysis for all 2,022 DEGs combined (DEGs from all comparisons presented in Figures 2 , 3 ). y-axis: -log10 of p-value, x-axis: genome position per chromosome.

Figure 9

Gene expression levels of genes with a cis-eQTL. Boxplots of gene expression values of six top genes (by FDR) for e-QTL mapping, stratified by genotype. Box center line: median, box limits: upper and lower quartiles, whiskers: 1.5x interquartile range.