Gene expression analysis reveals early changes in several molecular pathways in cerebral malaria-susceptible mice versus cerebral malaria-resistant mice

Abstract

Background: Microarray analyses allow the identification and assessment of molecular signatures in whole tissues undergoing pathological processes. To better understand cerebral malaria pathogenesis, we investigated intra-cerebral gene-expression profiles in well-defined genetically cerebral malaria-resistant (CM-R) and CM-susceptible (CM-S) mice, upon infection by Plasmodium berghei ANKA (PbA). We investigated mouse transcriptional responses at early and late stages of infection by use of cDNA microarrays.

Results: Through a rigorous statistical approach with multiple testing corrections, we showed that PbA significantly altered brain gene expression in CM-R (BALB/c), and in CM-S (CBA/J and C57BL/6) mice, and that 327 genes discriminated between early and late infection stages, between mouse strains, and between CM-R and CM-S mice. We further identified 104, 56, 84 genes with significant differential expression between CM-R and CM-S mice on days 2, 5, and 7 respectively. The analysis of their functional annotation indicates that genes involved in metabolic energy pathways, the inflammatory response, and the neuroprotection/neurotoxicity balance play a major role in cerebral malaria pathogenesis. In addition, our data suggest that cerebral malaria and Alzheimer's disease may share some common mechanisms of pathogenesis, as illustrated by the accumulation of beta-amyloid proteins in brains of CM-S mice, but not of CM-R mice.

Conclusion: Our microarray analysis highlighted marked changes in several molecular pathways in CM-S compared to CM-R mice, particularly at early stages of infection. This study revealed some promising areas for exploration that may both provide new insight into the knowledge of CM pathogenesis and the development of novel therapeutic strategies.

Figure 1

Schematic outline of data analysis. HCL: hierarchical clustering. ^a Brain gene expression prior to infection was compared with brain gene expression on days 2, 5 and 7. ^b Brain gene expression in CM-R mice was compared with that in CM-S mice at each time point. As represented by dashed arrows, we considered the whole data set (n = 2012 genes) to carry out multiple testing correction.

Figure 2

Distribution of genes regulated by PbA infection according to the time of infection. The genes whose expression was significantly altered by PbA infection were identified by using pair-wise Welch t tests with a Bonferroni correction.A. The number of genes with significant changes is shown at each time point for each mouse strain. B, C and D. The number of genes significantly up- (positive values) and down-regulated (negative values) is shown at each time point for each mouse strain.

Figure 3

Overlapping genes with significant differential expression at different time points during PbA infection. This Venn diagrams show the number of overlapping genes with significant expression changes on days 2, 5, and 7 post-infection with PbA, for C57BL/6 mice (A and B), CBA/J mice (C and D) and BALB/c mice (E and F). Genes that are either induced or suppressed by infection are distinguished for each strain: the number of genes induced is shown in A, C, and E, while the number of genes suppressed is shown in B, D, and F.

Figure 4

Hierarchical classification of mouse-strain-specific and CM-R/CM-S genes according to the time of infection. A. Hierarchical clustering of the 58 brain tissue samples, using expression levels of 327 significant genes differentially expressed between the mouse strains at early and late stages of infection. This set of genes was extracted from the full data set (n = 2012) by use of a SAM procedure and a false discovery rate of 0%. Each row represents a gene and each column represents a sample. Red and green indicate expression levels above and below the median, respectively. Grey indicates missing data. Dendograms of samples (above matrix) and genes (to the left of matrix) represent overall similarities in gene expression profiles. B. Dendogram of samples representing the results of the same global hierarchical clustering applied to the 58 brain tissue samples. Clustering of technical replicates (CBA/J-5, C57BL6-20 and BALB/c-13) is shown: the samples taken from a CBA/J mouse on day 2, a C57L/6 mouse on day 7, and a BALB/c mouse on day 7 were run on 5, 2, and 2 microarrays, respectively. Samples taken from mice on days 2, 5, 7, 9, and 15 post-infection are coded D2, D5, D7, D9, and D15, respectively.

Figure 5

View of biological functional annotation repartition of the genes grouped in clusters B and C. The KEGG pathways, in which the genes were known to be involved, are shown. Of the 245 genes of the clusters B and C, 80 were annotated. We represented only the KEGG pathways that contained at least three genes.

Figure 6

Genes differentially expressed between CM-R and CM-S mice on days 2, 5, and 7. The number of genes up-regulated in CM-S mice compared to CM-R mice (A), and the number of genes up-regulated in CM-R mice compared to CM-S mice (B) are shown.

Figure 7

Schematic diagram showing the possible effects of the reelin pathway in protection from CM. Reelin (RELN) is an extracellular matrix serine protease expressed in some neurons, such as GABAergic interneurons, which inhibit excitotoxic neurotransmission [45]. RELN that is secreted into the extracellular space acts by paracrine and autocrine mechanisms. RELN interacts with very low-density lipoprotein receptors (VLDLR) and apolipoprotein E type 2 receptors (ApoER2) leading to tyrosine phosphorylation of the adaptor protein Disabled-1 (DAB1) by the SRC family kinases (SRC) [42]. DAB1 activation, in turn, activates PI3K/Akt signalling, which has been implicated in neuronal migration during development and adulthood. In addition, phosphorylated DAB1 interacts with LIS1, a protein encoded by Pafah1b1, which associates with microtubules and modulates neuronal migration [46]. LIS1 may be required for regulating crucial steps of reelin-dependent neuronal positioning. In parallel, phosphorylated DAB1 inhibits glycogen synthase kinase 3β (GSK3β), a kinase known to phosphorylate Tau protein at multiple sites. Therefore, the activation of RELN pathway diminishes the level of hyperphosphorylated Tau protein, which is a biomarker of brain injury. In particular, hyperphosphorylated Tau protein is a component of the neurofibrillary tangles involved in Alzheimer's disease. Reln, Dab1 and Pafah1b1 were shown to be over-expressed in CM-R mice compared to CM-S mice. The activation of RELN signalling may inhibit excitotoxic neurotransmission and Tau phosphorylation, and may activate neurogenesis in CM-R mice. This may lead to diminished brain injury and to increased brain injury repair. Solid arrows represent influences on the activity of proteins or physiological mechanisms. Dashed arrows represent impaired effects on the activity of proteins or physiological mechanisms. Negative signs indicate inhibition, and positive signs indicate activation.