Understanding Human Cerebral Malaria through a Blood Transcriptomic Signature: Evidences for Erythrocyte Alteration, Immune/Inflammatory Dysregulation, and Brain Dysfunction

Abstract

Background: Cerebral malaria (CM), a reversible encephalopathy affecting young children, is a medical emergency requiring rapid clinical assessment and treatment. However, understanding of the genes/proteins and the biological pathways involved in the disease outcome is still limited.

Methods: We have performed a whole transcriptomic analysis of blood samples from Malian children with CM or uncomplicated malaria (UM). Hierarchical clustering and pathway, network, and upstream regulator analyses were performed to explore differentially expressed genes (DEGs). We validated gene expression for 8 genes using real-time quantitative PCR (RT-qPCR). Plasma levels were measured for IP-10/CXCL10 and IL-18.

Results: A blood RNA signature including 538 DEGs (∣FC | ≥2.0, adjusted P value ≤ 0.01) allowed to discriminate between CM and UM. Ingenuity Pathway Analysis (IPA) and Kyoto Encyclopedia of Genes and Genomes (KEGG) revealed novel genes and biological pathways related to immune/inflammatory responses, erythrocyte alteration, and neurodegenerative disorders. Gene expressions of CXCL10, IL12RB2, IL18BP, IL2RA, AXIN2, and NET were significantly lower in CM whereas ARG1 and SLC6A9 were higher in CM compared to UM. Plasma protein levels of IP-10/CXCL10 were significantly lower in CM than in UM while levels of IL-18 were higher. Interestingly, among children with CM, those who died from a complication of malaria tended to have higher concentrations of IP-10/CXCL10 and IFN-γ than those who recovered.

Conclusions: This study identified some new factors and mechanisms that play crucial roles in CM and characterized their respective biological pathways as well as some upstream regulators.

Figure 1

(a) Heatmap for differentially expressed genes between cerebral malaria (CM) and uncomplicated malaria (UM) children. Hierarchical clustering of microarrays was obtained using Pearson's correlation on probes with a fold change greater than two and adjusted P value ≤ 0.01. The red color represents high expression, while the green color represents low expression. The blue and purple bars at the top represent the CM and UM children, respectively. Some interesting candidate genes are indicated, in red for the new players of CM pathophysiology, in black for the genes previously described in malaria. (b) Stacked bar chart displaying the top canonical pathways found to be differentially represented in comparing gene expression in CM and UM children absolute fold change ≥2, adjusted P value ≤ 0.05. The total number of genes in each pathway is displayed above each bar. Pathways are ranked by statistical significance; the orange dot indicate the -log (P value). The percentage of dysregulated genes is indicated for each pathway, downregulated genes are in green, and the upregulated genes are in red. The ratio of the numbers of DEGs to the total number of genes in the pathways ranged from 9% to 100%.

Figure 2

Gene network and upstream analysis by Ingenuity Pathway Analysis (IPA; Qiagen Inc.). Gene network highlighting the candidate genes and their interaction with the genes presented as nodes and relationship between two indicated as a line. Upstream regulator analysis allows the identification of transcriptional regulators and their target genes dysregulated in our dataset. (a) Gene network for connective tissue development and function, tissue morphology, and connective tissue disorders. (b) Gene network for cell death and survival, connective tissue disorders, and hematological disease. (c) The transcriptional regulator TNF was ranked first among the upstream regulators, and 38 genes were predicted to be activated or inhibited by TNF. Four additional upstream regulators SLC4A1, IL-1B, STAT3, and IL-6 have been identified.

Figure 3

Real-time polymerase chain reaction-based validation of messenger RNA levels for 8 significantly dysregulated genes between CM and UM. Samples from 13 children with cerebral malaria (CM) and 12 with uncomplicated malaria (UM) were analyzed. Relative expression levels were calculated from 2^-ΔΔCt values. Values for children with CM are represented by black squares, and values for children with UM are represented by gray circles. The horizontal lines indicate median values. We used the Mann-Whitney U test to compare the results for the CM and UM groups.

Figure 4

(a) Plasma IP-10 and IL-18 concentrations in children with CM (n = 58, black squares) and UM (n = 41, gray circles). The nonparametric Mann-Whitney U test was used to assess differences. (b) Plasma IP-10 and IFN-γ concentrations in children with fatal CM and nonfatal CM and UM. The levels are represented in log. (c) Plasma IP-10 and IFN-γ concentrations for each child with CM or UM. The levels are represented in log. We used Spearman's correlation coefficient.