Genome-wide expression profiling in malaria infection reveals transcriptional changes associated with lethal and nonlethal outcomes

Abstract

High-density oligonucleotide microarrays are widely used to study gene expression in cells exposed to a variety of pathogens. This study addressed the global genome-wide transcriptional activation of genes in hosts infected in vivo, which result in radically different clinical outcomes. We present an analysis of the gene expression profiles that identified a set of host biomarkers which distinguish between lethal and nonlethal blood stage Plasmodium yoelii malaria infections. Multiple biological replicates sampled during the course of infection were used to establish statistically valid sets of differentially expressed genes. These genes that correlated with the intensity of infection were used to identify pathways of cellular processes related to metabolic perturbations, erythropoiesis, and B-cell immune responses and other innate and cellular immune responses. The transcriptional apparatus that controls gene expression in erythropoiesis was also differentially expressed and regulated the expression of target genes involved in the host's response to malaria anemia. The biological systems approach provides unprecedented opportunities to explore the pathophysiology of host-pathogen interactions in experimental malaria infection and to decipher functionally complex networks of gene and protein interactions.

FIG. 1.

Course of infection and differential gene expression in P. yoelii infection. (A) Kinetics of parasite replication in 17XNL and 17XL infection in mice. Error bars indicate standard deviations. (B) Number of differentially expressed genes (P < 0.005) during specific intervals during infection relative to mean gene expression from six uninfected mice. (C and D) Three-dimensional PCA of average expression from six 17XNL-infected mice at six time intervals (C) and from four 17XL-infected mice at 4 time intervals (D) during infection is shown in relation to gene expression from six uninfected mice (yellow circle). Significant differentially expressed genes were collapsed into three-dimensional vectors to reveal differences in transcript abundance as a function of parasite density.

FIG. 2.

Concordance in gene expression of immune response genes between lethal 17XL and nonlethal 17XNL infection. The fold change in intensity of differentially expressed genes is depicted as a heat map at 24 h postinfection (A) and at 5% parasitemia (B) for P. yoelii (17XL and 17XNL)-infected mice compared to six uninfected mice. Statistically significant gene ontology groups derived by Onto-Express from sets of up-regulated genes at 24 h postinfection and at 5% parasitemia are indicated.

FIG. 3.

Functional annotation of differentially expressed genes in 17XNL infection. (A and C) Pie charts depict the representative number of overrepresented GO terms from genes up-regulated (A) or down-regulated (C) during infection. The size of each slice represents the number of differentially expressed genes assigned to the particular GO functional classification. (B) Gene expression for one representative GO term (heat shock protein activity) is tightly regulated in nonlethal infection and varies directly with intensity of infection. (D) Expression of genes with chemotaxis/chemokine activity is repressed during 17XNL infection and remains suppressed even after resolution of parasitemia. The heat maps indicate the average fold change in gene expression greater than (red) or less than (blue) that in uninfected control spleens at each time point or stage of infection.

FIG. 4.

Differential gene expression in erythropoiesis. (A) Centroid plot of mean expression of 13 genes involved in iron transport and erythrocyte membrane proteins, with heat maps showing fold change in transcript abundance (columns represent individual mice). (B) Relative gene expression for genes representing the enzymes of the heme biosynthetic pathway at 25% parasitemia in nonlethal (NL) and lethal (L) malaria. (C) Cartoon representation of gene expression for regulatory transcription factors critical for erythropoiesis (adapted from reference 20). Red or blue colored boxes and ovals represent significantly induced or repressed gene expression, respectively. Gray objects represent no change in gene expression relative to uninfected control spleens.

FIG. 5.

Expression of glycolytic enzyme genes. The log₂ fold changes of the mean expression of glycolysis genes from 0 to 25% parasitemia for the 17XNL (n = 6) (Fig. 4A) and 17XL (n = 4) (Fig. 4B) infections are shown. The genes are listed, from top to bottom, in the order of the glycolytic enzyme pathway.

FIG. 6.

Expression of B-cell proliferation and plasma cell differentiation genes. (A) Hierarchical clustering of genes significantly differentially expressed between 17XNL and 17XL infection. The red-green matrix represents the normalized expression pattern for each gene across the samples. The intensity of the red or green shaded boxes represents the highest or lowest relative expression, respectively. Overrepresented GO terms identified by dCHIP in coexpressed clusters of genes are indicated. NL, nonlethal; L, lethal. (B) Comparisons in gene expression ratios for B-cell activation and proliferation genes. The color matrix plot shows relative expression across spleens from 17XNL and 17XL infection at 25% parasitemia. (C) Color matrix plot of a subset of immunoglobulin genes from the hierarchical cluster at 25% parasitemia, showing relative overexpression during nonlethal infection.