An evolutionary conserved function of the JAK-STAT pathway in anti-dengue defense

JAK-STAT Pathway Controls Dengue Virus Infection in A. aegypti.

Our previous studies have shown that the Toll pathway is activated by, and controls, dengue virus infection in the A. aegypti mosquito vector (13). The dengue virus infection responsive nature of the JAK-STAT pathway genes, dome and SOCS, also suggested the implication of this pathway in the mosquito's anti-dengue defense. To investigate this hypothesis, we have used a combination of reverse genetic and transcriptomic analyses. We rendered mosquitoes more resistant or more susceptible to DENV-2 through RNAi-mediated gene silencing of the JAK-STAT pathway negative regulator PIAS or positive regulators Hop and Dome, respectively, as a measure of altered virus loads in the midgut tissue at 3 and 7 days post infectious blood meal (pibm) (Fig. 2). The RNAi-mediated silencing of these negative and positive regulators simulates the transient activation or repression, respectively, of the pathway. PIAS gene silencing (77% silencing efficiency) reduced DENV-2 infection by ≈2 (P = 0.44) and 5-fold (P < 0.05) at 3 and 7 days after pibm, respectively (Fig. 2 A and B). Conversely, both Hop gene silencing (60% silencing efficiency) and Dome gene silencing (69% silencing efficiency) resulted in an increase of virus load at 3 and 7 days pibm, respectively; Hop depletion led to a 2.3-fold (P = 0.07) increase in infection at 3 days pibm (Fig. 2A) and DOME depletion resulted in a 3-fold increase of virus loads (P < 0.05) at 7 days pibm (Fig. 2B) compared with their respective GPF dsRNA-treated control mosquitoes. Although nonsignificant when compared with their respective GPF dsRNA-treated controls, the effect of Hop and PIAS gene silencing on DENV-2 infection becomes statistically significant if compared with each other (P < 0.01) and follow the same trend demonstrated by repression or activation of the JAK-STAT pathway at 7 days pibm. Our data show that the pathway is acting against the dengue virus in the mosquito midgut as early at 3 days after virus ingestion.

Dissecting the JAK-STAT Pathway-Regulated Transcriptome and Its Activation in Response to Dengue Virus Infection.

The implication of the JAK-STAT pathway in limiting DENV-2 infection suggests that this immune pathway regulates transcription of one or several dengue virus restriction factors. The JAK-STAT pathway-regulated transcriptome in mosquitoes and other organisms has not been characterized at a genome-wide level as yet. To identify candidate JAK-STAT pathway-regulated anti-dengue effector genes, we determined the A. aegypti putative JAK-STAT pathway-regulated transcriptome by comparing the transcript abundance between PIAS dsRNA-treated and GFP dsRNA-treated control mosquitoes, at 4 days after dsRNA injection, using high-density oligonucleotide microarrays. Similar approaches have been fruitfully used to determine the Toll and Imd pathway regulated transcriptomes in A. aegypti and A. gambiae mosquitoes (13, 28). The depletion of PIAS resulted in the up-regulation of 63 genes and down-regulation of 66 genes, belonging to diverse functional groups (Fig. 3A and Table S1). Only 14% of the putative JAK-STAT-regulated genes could be assigned putative immune functions based on sequence homology, whereas as much as 22% of the regulated genes were of unknown function. Interestingly, a putative Toll receptor ligand gene, spätzle 6 (spz6, AAEL012164), was up-regulated, and five anti-microbial peptide genes (four Cecropin A-like genes: AAEL000611, AAEL000627, AAEL015515, and AAEL000598 and one Defensin 1-like gene: AAEL003832) were down-regulated. This may indicate a competition of resources between the JAK-STAT pathway and the other immune pathways.

To assess the extent of JAK-STAT pathway activation by DENV-2 infection, at the transcriptomic level, we compared the putative JAK-STAT pathway regulated transcriptome to the DENV-2 infection-responsive midgut and carcass transcriptomes at 10 days pibm (13). We also coclustered the putative Toll pathway regulated transcriptome (13) to assess potential overlaps in gene regulation between the two anti-dengue immune pathways. This analysis revealed 18 genes that were transcriptionally induced by both PIAS gene-silenced and DENV-2-infected mosquitoes (Fig. 3B). Interestingly, none of these genes were induced in Cactus-depleted (Toll pathway-activated) mosquitoes, suggesting that the JAK-STAT and the Toll immune pathways modulate DENV-2 infection through different factors and mechanisms. However, the concomitant down-regulation of 36 genes by both pathways corroborates certain relationships that may mainly involve developmental processes (7, 8, 29–31).

Hierarchical clustering of genes that were differentially expressed in at least two of the three experiments (PIAS silencing, cactus silencing, and DENV-2 infection responses in the carcass tissue) generated six distinct expression clusters (Fig. 3C and Table S2). All 18 JAK-STAT pathway induced and DENV-2 infection responsive genes were grouped in the same cluster (Fig. 3C, Cluster VI, and Table S2), further supporting the relationship between the JAK-STAT pathway and anti-dengue immunity. Expression cluster VI comprises six genes of putatively known functions: two t-box transcription factors, tbx6 (AAEL004174) and tbx20 (AAEL014429); a zinc finger protein (ZFP, AAEL009508); a cadherin (AAEL000700); a forkhead box protein (foxn2, AAEL009900); and a host cell factor C1 (HCF1, AAEL008905) that belongs to a gene family that has been linked to host–virus interactions (32–35). The majority (12 of 18) of genes in cluster VI have no known function. Expression cluster V (Fig. 3C, Cluster V, and Table S2) is mostly populated by known immune genes (18 of 19) that are induced by cactus silencing (Toll pathway) but not by PIAS silencing. One Toll pathway-induced Cecropin gene (cecA-like) (AAEL015515) was suppressed by JAK-STAT activation, further suggesting that the JAK-STAT and the Toll pathway are acting independently against DENV-2.

Screening and Characterization of PIAS-Suppressed (Putative JAK-STAT-Induced) Dengue Virus Restriction Factors (DVRFs).

To identify putative anti-dengue factors among the 18 genes in cluster VI that were induced by both PIAS gene silencing (putative JAK-STAT-induced) and DENV-2 infection, we considered their expression patterns along with sequence characteristics that could be indicative of antiviral proteins. We selected the ZFP, foxn2, HCF1, and two additional genes that coded for hypothetical proteins, Q1HR00 [AAEL008492, putative secreted protein, (36)] and AAEL000896. The selection of AAEL000896 and Q1HR00 was purely based on protein features such as a predicted signal peptide (Table S3).

We used an established RNAi-mediated gene silencing approach, in conjunction with DENV-2 infection assays, to test the putative implication of the five selected gene candidates in anti-dengue defense (Fig. 4). This screen confirmed two of these genes, Q1HR00 (33% silencing efficiency) and AAEL000896 (50% silencing efficiency) as putative JAK-STAT pathway-regulated dengue virus restriction factors DVRF1 and DVRF2, respectively. RNAi-mediated depletion of DVRF1 resulted in a 2.5-fold increase of DENV-2 infection (P < 0.05), and silencing of DVRF2 resulted in 2.2-fold increased infection in midgut tissue (P < 0.05) when compared with GFP dsRNA-treated control mosquitoes. The lack of effects on virus infection for the other candidate genes does not exclude their implication in the anti-dengue defense because they could be acting in different tissues and/or at different infection stages.

To validate the JAK-STAT pathway regulation of DVRF1 and DVRF2, we investigated the effect of PIAS gene silencing on their transcript abundance. PIAS depletion resulted in 1.7- and 1.8-fold induction of DVRF1 and DVRF2, respectively. To confirm the relationship between the JAK-STAT pathway and the anti-dengue effectors, we also performed double gene-silencing assays, in which the effect of PIAS depletion on the DVRFs gene expression was reversed through simultaneous Dome depletion (Fig. 5A and Table S4). To confirm the DENV-2 infection-responsive nature of the DVRFs, we compared the relative mRNA abundance of these two genes in the whole body of uninfected and DENV-2-infected adult females at 24 h pibm. Dengue virus infection induced DVRF1 by a 3-fold and DVRF2 by a 3.6-fold (Fig. 5B and Table S4). The activation of the JAK-STAT pathway at this early stage of dengue virus infection was also confirmed by measuring the relative expression of the JAK-STAT pathway reporter gene SOCS with qRT-PCR. At 24 h pibm, SOCS was 3.4-fold up-regulated in response to DENV-2 infection (Fig. 5B and Table S4).

To test whether the JAK-STAT pathway is a general anti-dengue defense system in A. aegypti or whether it is specific for the Rockefeller strain, we also evaluated the activation of these genes in the whole body of the recently colonized A. aegypti Costa Rica strain (CRICA) upon infection with Vero cell (mammalian)-derived dengue virus (mDENV-2) at 24 h pibm. We used the mammalian cell-derived virus to better simulate a field condition when mosquitoes feed on human-derived virus and thereby test whether the potentially different glycosylation pattern of the virion surface could influence the immune response. DVRF2 and SOCS were 2- and 1.7-fold induced, respectively, in the CRICA/mDENV-2 combination (Fig. 5C and Table S4), suggesting that the JAK-STAT pathway is activated in different A. aegypti strains when infected with mammalian cell-derived virus. The lack of DVRF1 induction is likely to indicate variations in function and regulation patterns for this gene between different mosquito strains.

At least one STAT-binding site, TTCN_2–4GAA, was identified within the 1-kb upstream genome region of both DVRFs (Fig. 6A and Table S5) (37). DVRF1 encodes a putative 120-aa transmembrane protein, suggesting a putative receptor function. The 210-aa predicted polypeptide encoded by DVRF2 is a putative secreted protein that contains putative antifreezing 1 (AF1) and allergen C (All_C) domains (Fig. 6B and Table S6) and shares 64% and 53% identity with the A. gambiae cuticular proteins 1 (CPF1, AGAP010900) and 2 (CPF2, AGAP010901), respectively.

Phylogenetic and Protein Sequence Predictions.

The BLAST program was used to align the JAK-STAT factor sequences with the peptide sequences retrieved from the Ensembl database according to procedures (50). Full-length or partial predicted sequences were aligned by using the ClustalX program, and cladograms were constructed by Bootstrap neighbor-joining analysis and displayed through MEGA (Molecular Evolutionary Genetic Analysis) software. Protein domains were determined by using Ensembl or SMART database version 5.1, and signal peptide prediction was carried out through SignalP 3.0. Cladrograms are represented in Fig. S1.

Mosquito Rearing and Cell Line Culture.

The A. aegypti Rockefeller/UGAL strain was maintained as previously described (13). A. aegypti mosquitoes of the Costa Rica strain (MRA-726) were obtained from MR4. All experiments were performed with the Rockefeller strain, except when otherwise specified. The A. albopictus C6/36 cell line was maintained as previously described (13).

RNAi Gene-Silencing Assays.

RNAi assays were carried out according to standard methodology (48, 51). Both dsGPF- (control) and gene dsRNA-injected (experimental) groups were kept in the insectary at the conditions mentioned above. Gene silencing efficiency was determined in at least three biological replicates at 3–4 days after dsRNA injection by using qRT-PCR (Table S7). The primers used for dsRNA synthesis and qRT-PCR are presented in SI Text.

Semiquantitative RT-PCR (qRT-PCR) Assays.

qRT-PCR assays were conducted as previously described (50). RNA samples were treated with Turbo DNase (Ambion) before reverse transcription with SuperScript III kit (Invitrogen), performed according to manufacturer's instructions in 20-μL reaction mixtures containing oligo(dT) primers and 1 μg of total RNA. The PCRs were standardized to yield a nonsaturated amount of PCR products, indicating linear amplification. Optimal number of cycles was empirically determined from test PCR runs for each tested gene. The PCR products were separated by gel electrophoresis, and the gel image was documented by using a Fuji image documenter system. Signal intensities from gel bands were quantified by using Image Gauge software (Fuji-film Imaging System). The relative fold induction or repression in gene expression of experimental samples was determined by comparing these values to their respective controls after normalization of transcript level by the A. aegypti ribosomal S7 gene. Primer sequences are listed in SI Text. Numeric data for the semiquantitative assays are presented in Table S7.

DENV-2 Infections and Titration.

Dengue virus serotype 2 (DENV-2) from the New Guinea C strain was propagated in mosquito-derived C6/36 cells as previously described (52). Virus titers in the midguts were determined at 7 days after DENV-2 infection according to established protocol (13, 49) (www.jove.com/index/Details.stp?ID=220). At least three biological replicates were performed for each experimental group. The virus-containing homogenates were serially diluted, inoculated into 80% confluent C6/36 cells in 24-well plates and kept for 5–7 days at 32 °C and 5% CO_2. Plaque formation units (PFU) were visualized by peroxidase immunostaining, using mouse hyperimmune ascitic fluid (MHIAF, specific for DENV-2) and a goat anti-mouse HRP conjugate as the primary and secondary antibody, respectively. All procedures involving DENV-2 infections were carried out in a Biological Safety Level 2 laboratory. Numeric PFU data are presented in Table S8. New Guinea C DENV-2 propagated in monkey-derived Vero cells was provided by Bhavin Thumar and Anna P. Durbin (Johns Hopkins Bloomberg School of Public Health), and used to infect A. aegypti Costa Rica strain, as described above.

Microarray Assays.

Transcription assays and analysis were conducted following standard protocols with a full-genome Agilent-based microarray platform (13, 17, 48). Relative mRNA abundance was compared between PIAS dsRNA- and GFP dsRNA-injected controls 4 days after injection. In brief, 2–3 μg of total RNA was used for probe synthesis of cy3- and cy5-labeled dCTP. Hybridizations were conducted with an Agilent Technologies In Situ Hybridization kit at 60 °C according to the manufacturer's instructions. Hybridization intensities were determined with an Axon GenePix 4200AL scanner, and images were analyzed with Gene Pix software. The expression data were processed and analyzed as described previously (13, 17). In brief, the background-subtracted median fluorescent values were normalized according to a LOWESS normalization method, and Cy5/Cy3 ratios from replicate assays were subjected to t tests at a significance level of P < 0.05 by using TIGR MIDAS and MeV software (53). Expression data from all replicate assays were averaged with the GEPAS microarray preprocessing software before logarithm (base 2) transformation. Self–self-hybridizations were used to determine the cutoff value for the significance of gene regulation on these type of microarrays to 0.78 in log₂ scale, which corresponds to 1.71- fold regulation (54). For genes with P < 0.05, the average ratio was used as the final fold change; for genes with P > 0.05, the inconsistent replicates (with distance to the median of replicate ratios >0.8) were removed, and only the values from a gene with at least two replicates in the same direction of regulation were further averaged. Three independent biological replicate assays were performed. Numeric microarray gene expression data are presented in Table S1.

Immune Challenge Assays and Endogenous Bacteria Measurement in RNAi Gene-Silenced Mosquitoes.

The effect of gene silencing on mosquito survival after bacterial challenge and the effect of gene silencing on proliferation of opportunistic bacteria in the mosquito hemolymph were carried out essentially according to established protocols (28, 48, 50) (see SI Text for details). In brief, for survival analysis, mosquitoes were injected with PBS or bacteria 3 days after RNAi gene silencing, and dead mosquitoes were counted every day (Fig. S2A). For bacteria measurement, the midgut or hemolymph of dsRNA-injected mosquitoes were removed and plated in LB agar plates 3 days after dsRNA injection, and the number of colony forming units (CFUs) was determined (Fig. S2 B and C).

Accession Numbers.

The Ensembl gene ID for the genes and proteins cited in the text are: AAEL012164 (spätzle 6); AAEL000611, AAEL000627, AAEL015515, and AAEL000598 (cecropin A-like); AAEL003832 (defensin 1-like); AAEL000700 (cadherin); AAEL009900 (foxn2); AAEL008905 (HCF1); AAEL008492 (DVRF1); AAEL000896 (DVRF2).