Highly focused transcriptional response of Anopheles coluzzii to O'nyong nyong arbovirus during the primary midgut infection

Abstract

Background: Anopheles mosquitoes are efficient vectors of human malaria, but it is unknown why they do not transmit viruses as well as Aedes and Culex mosquitoes. The only arbovirus known to be consistently transmitted by Anopheles mosquitoes is O'nyong nyong virus (ONNV, genus Alphavirus, family Togaviridae). The interaction of Anopheles mosquitoes with RNA viruses has been relatively unexamined.

Results: We transcriptionally profiled the African malaria vector, Anopheles coluzzii, infected with ONNV. Mosquitoes were fed on an infectious bloodmeal and were analyzed by Illumina RNAseq at 3 days post-bloodmeal during the primary virus infection of the midgut epithelium, before systemic dissemination. Virus infection triggers transcriptional regulation of just 30 host candidate genes. Most of the regulated candidate genes are novel, without known function. Of the known genes, a significant cluster includes candidates with predicted involvement in carbohydrate metabolism. Two candidate genes encoding leucine-rich repeat immune (LRIM) factors point to possible involvement of immune protein complexes in the mosquito antiviral response. The primary ONNV infection by bloodmeal shares little transcriptional response in common with ONNV infection by intrathoracic injection, nor with midgut infection by the malaria parasites, Plasmodium falciparum or P. berghei. Profiling of A. coluzzii microRNA (miRNA) identified 118 known miRNAs and 182 potential novel miRNA candidates, with just one miRNA regulated by ONNV infection. This miRNA was not regulated by other previously reported treatments, and may be virus specific. Coexpression analysis of miRNA abundance and messenger RNA expression revealed discrete clusters of genes regulated by Imd and JAK/STAT, immune signaling pathways that are protective against ONNV in the primary infection.

Conclusions: ONNV infection of the A. coluzzii midgut triggers a remarkably limited gene regulation program of mostly novel candidate genes, which likely includes host genes deployed for antiviral defense, as well as genes manipulated by the virus to facilitate infection. Functional dissection of the ONNV-response candidate genes is expected to generate novel insight into the mechanisms of virus-vector interaction.

Keywords: Arbovirus; Host–pathogen interactions; Innate immunity; Insect immunity; Insect vectors; Malaria.

Fig. 1

Thirty A. coluzzii genes are significantly regulated during the primary midgut infection with ONNV. Differential gene expression between ONNV-infected blood meal and non-infected bloodmeal was measured by RNAseq in pools of A. coluzzii mosquitoes 3 d after the bloodmeal, a time point when the infection is restricted to the midgut epithelium and not yet disseminated in the body compartment [6]. Histograms indicate fold change of transcript abundance (adjusted p-values, Additional File 1: Table S1). Gene identities are indicated by Vectorbase AGAP identifiers. The six genes with a gene name to the left of the AGAP identifiers have an annotated function, while the majority of ONNV-regulated genes are without known function, and bioinformatic functional prediction data is shown in Additional File 1: Table S1

Fig. 2

Comparison of ONNV regulated gene sets after bloodmeal or injection-induced infections of A. coluzzii. Venn diagram indicates overlap of ONNV-regulated genes in the primary midgut infection by bloodmeal (Blood-Midgut 3d, current study) and in the systemic infection by injection of virus, bypassing the midgut and without bloodmeal, at 1 d (Injected-Systemic 1d), 4 d (Injected-Systemic 4d), and 9 d (Injected-Systemic 9d) post-injection [10]. Number and names of overlapping genes upregulated (red), downregulated (green) and differently regulated (black) between studies are indicated. Three genes were regulated in both bloodmeal and injection-induced infections: LRIM4 and AAAP were upregulated in both conditions while LRIM10 was downregulated in blood-infected but upregulated in injection-infected mosquitoes

Fig. 3

Correlation between miRNA target gene prediction and differential expression of predicted targets in A. coluzzii. a) Venn diagram of Pearson correlation between predicted miRNA target genes and their transcript abundance levels as measured by RNAseq. Silencing of the Imd pathway factor Rel2 significantly regulated five miRNAs (Rel2_5miRNAs), silencing the JAK/STAT factor Stat-A regulated four miRNAs (StatA_4miRNAs), and control treatment with LacZ dsRNA regulated two miRNAs (LacZ_2miRNAs)(Additional file 6: Figure S4). Diagram indicates the numbers of in silico predicted target genes whose expression was significantly anti-correlated with the abundance of the miRNAs. b) PCA analysis of the Pearson correlation matrix for the 11 regulated miRNAs and the 336 differentially expressed predicted target genes represented in the Venn diagram. Points indicate genes that were differentially expressed and also predicted as miRNA targets, color of points indicates the treatment that caused the differential gene expression. Predicted target genes filtered for anti-correlation of expression with the miRNAs cluster strongly according to the conditions that generated the expression differences of the miRNA-mRNA target pairs

Fig. 4

Genes with mRNA abundance significantly anti-correlated with the 11 regulated miRNAs. a) Similar to Fig. 3, but filtering mRNAs only for significant negative correlation of expression with the 11 induced miRNAs (Additional file 6: Figure S4), without filtering by computational target site prediction. Venn diagram of Pearson correlation between miRNAs and the differentially expressed genes as measured by RNAseq. b) PCA analysis of the differentially expressed genes represented in the Venn diagram based on Pearson correlation coefficients with miRNA expression levels. Other description as in Fig. 3