Anopheles gambiae densovirus (AgDNV) has negligible effects on adult survival and transcriptome of its mosquito host

Cell culture, mosquito infection and rearing conditions

Sua5B cells, which are naturally infected with AgDNV (Ren, Hoiczyk & Rasgon, 2008), were passaged weekly in Schneider’s media with 10% FBS. AgDNV was obtained from the infected cell line Sua5B and first-instar mosquito larvae infected by exposure to purified virus as previously described (Ren, Hoiczyk & Rasgon, 2008). Control mosquitoes were mock infected with Schneider’s medium. After infection, mosquitoes were reared in 2L pans according to a standard feeding protocol as described (Ren & Rasgon, 2010).

Life-table analysis

At the pupal stage, cups of emerging pupae were placed in cages and removed 12 h later ensuring that adults were of similar ages. Mosquitoes were allowed access to 10% sucrose but were not blood fed. Mortality was accessed daily at the same time ±1 h. There were 3–4 cages per treatment (approximately 50 mosquitoes per cage), and the entire experiment was replicated three times (total = 740 AgDNV-infected mosquitoes, 860 control mosquitoes). For AgDNV treatments, collected dead mosquitoes were assayed for AgDNV infection by PCR amplification of a ∼300 bp fragment of the AgDNV capsid gene as described (Ren, Hoiczyk & Rasgon, 2008); infection rates per cage ranged from 87% to 100%. The experiment included both males and females, but mosquitoes were not sexed for analysis. Data were analyzed by the Gehan–Breslow–Wilcoxon test using GraphPad Prism 5.

RNA extraction and microarrays

Affymetrix GeneChip microarrays were used to assess the effect of AgDNV infection on An. gambiae gene transcription. First instar larvae were infected or mock infected as described above and reared to adulthood. At 10 days post-emergence when AgDNV titers are at their highest (Ren & Rasgon, 2010) mosquitoes were processed for analysis. For each biological replicate, pools of 20 adult mosquitoes were processed (mosquitoes were not sexed). There were three replicates per treatment. 20 randomly selected mosquitoes per replicate were tested by PCR (Ren, Hoiczyk & Rasgon, 2008) to confirm AgDNV infection; 100% of mosquitoes treated with virus as larvae were positive for infection by PCR compared to 0% of control mosquitoes. Mosquitoes were homogenized and lysed with Lysing Matrix D (MP bio) in 1 ml of Trizol reagent (Invitrogen) by rapid agitation in a FastPrep 120 Instrument (MP Biomedicals, Solon, OH) for 45 s at speed setting 6 and placed on ice for 2 min. Homogenization and ice incubation was repeated twice or until the samples were completely homogenized. After homogenization, RNA was extracted, purified and quantified as previously described (Hughes et al., 2011b). For each array 100 ng total RNA was hybridized to the Affymetrix Plasmodium/Anopheles microarray using the Affymetrix 3′ IVT express kit, according to manufacturer’s recommended protocol. Hybridization cocktails were prepared as recommended for arrays of Standard format using reagents in the Affymetrix Hybridization, Wash, and Stain kit. Hybridization was performed at 45 °C for 16 h at 60 rpm in the Affymetrix rotisserie hybridization oven. The signal amplification protocol for washing and staining of eukaryotic targets was performed in an automated fluidics station (Affymetrix FS450) using Affymetrix protocol FS450_0004. Arrays were transferred to a GCS3000 laser scanner with autoloader and 3G upgrade (Affymetrix) and scanned at an emission wavelength of 570 nm at 2.5 µm resolution. Quality assessment of hybridizations and scans was performed with Expression Console software. Detailed analyses were performed using Partek Genomics Suite version 6.4. GC-RMA algorithm defaults were used for background correction, normalization and summarization of probesets. Analysis of variance (AVOVA) was performed with linear contrasts for each densovirus treatment vs. control. Gene lists were developed based on 1.75 or 2.0 fold change (FC) or greater gene expression using a false discovery rate of P < 0.05. Raw Affymetrix CEL files are available as Supplemental data at http://rasgonlab.files.wordpress.com/2014/08/renetalrawaffymetrixdata.zip (controls: W1.CEL, W2.CEL, W3.CEL; treatment: AgDNV1.CEL, AgDNV2.CEL, AgDNV3.CEL).

qPCR validation of microarray analysis

qPCR analysis was completed as previously described (Hughes et al., 2011b). Briefly, RNA was extracted from pools of ten mosquitoes either infected with densovirus or uninfected using an RNeasy mini kit (Qiagen). Five biological replicates were completed for each treatment. RNA was DNase treated (Ambion) and cDNA synthesized using superscript III (Invitrogen) following manufactures guidelines. qPCR was completed using a Rotor gene Q (Qiagen) using Rotor gene SYBR green PCR kit (Qiagen) according to manufactures guidelines. To validate microarray results we assayed five genes selected both from genes affected and not affected in the microarray analysis (AMMECR1, Rel1, Rel2, cactus and caspar; primers in Table 1). Expression of each target gene was normalizing to expression of the ribosomal protein S7 gene (Table 1). All qPCRs were performed in triplicate. Melt curve analysis was completed on all PCRs. Determinations of relative expression were calculated using qGENE (Joehanes & Nelson, 2008).

Life table analysis

We used life table analysis to assess the fitness effects of AgDNV infection on adult An. gambiae (Keele strain) mosquitoes. For replicates 1 and 2, survival trajectories of AgDNV infected and uninfected mosquitoes were not significantly different (Table 2, Figs. 1A and 1B). For replicate 3, mosquitoes in both treatments had significantly elevated mortality compared to replicates 1 and 2, and AgDNV-infected mosquitoes had a slight, but statistically significant reduction in lifespan (P = 0.003) (Table 2, Fig. 1C). We do not know why results from replicate 3 differed from replicates 1 and 2, but the significant reduction (greater than 50%) in lifespan for both treatments in replicate 3 suggests other confounding factors besides AgDNV infection are at play.

Microarray analysis

To evaluate the effect of AgDNV infection on host gene expression, we completed microarray analysis comparing infected versus uninfected adult mosquitoes. AgDNV infection had very limited effect on An. gambiae gene expression with global expression analysis identifying only 4 genes modestly differentially up-regulated (fold-change (FC) ≥ 2, P < 0.05) in response to infection, and no genes significantly down-regulated (Table 3). qPCR of selected genes showed similar results (R² = 0.84, P = 0.03; Fig. 2). These results are in stark contrast to the effect of the human pathogen O’Nyong-nyong virus in An. gambiae, where infection resulted in the regulation of 253 genes (152 up-regulated, 102 down-regulated), including many genes involved in anti-viral and innate immune pathways (Waldock, Olson & Christophides, 2012). However, our results are similar to other studies examining gene expression in response to DNV infection in other insects. Minimal effects on host gene expression were seen in the moth Spodoptera frugiperda with the up-regulation of only 8 genes after injection of DNV into fat body tissue (Barat-Houari et al., 2006). Using subtractive hybridization to identify differentially expressed genes in Bombyx mori in response to virus infection, only 28 genes were found in a moth line resistant to DNV, while infection in susceptible line lead to differential expression of 23 genes (Bao et al., 2008). While infection with other microbes profoundly affects gene expression in An. gambiae (Abrantes et al., 2008), AgDNV (and DNVs in general) seem to neither strongly elicit nor suppress global gene expression patterns. Further studies are required to ascertain whether the virus employs mechanism(s) to avoid modulating the host transcriptome.

Of the 4 genes marginally up-regulated by AgDNV infection, two are putatively associated with stress response (cytochrome P450 (CYP450) and AMME syndrome candidate gene 1 (AMMECR1)). In mosquitoes, CYP450s are detoxification enzymes expressed when the insect is under oxidative stress (Feyereisen, 1999) and are known to be expressed when the mosquito is infected by pathogens (Abrantes et al., 2008). No functional studies have been conducted on AMMERCR1 in insects. The other two identified significantly regulated genes are trypsinogen 2 and plugin. Trypsinogen is the precursor of trypsin, which is involved in hydrolysis of peptide bonds during bloodmeal digestion (Noriega et al., 1996) while plugin is one of the major components of the mating plug (Le et al., 2013).

When we applied a less stringent FC criteria (1.75 FC), 15 genes were differently regulated in response to DNV virus infection (Table 3) with 9 genes up-regulated and 6 down-regulated. Besides functionally unknown genes, the rest of the genes are classified to transport-, metabolism- or binding-related transcripts (Table 3).

Paratransgenesis is an approach that attempts to modulate vector competence of the vector by manipulation of microorganisms within the host (Beard, Cordon-Rosales & Durvasula, 2002; Riehle & Jacobs-Lorena, 2005). Malaria researchers have focused on paratransgenesis as a novel alternative to traditional transgenic strategies (Favia et al., 2007). Our results suggest that AgDNV infection has minimal impact on survival or gene expression of its mosquito host making it a potentially attractive agent for paratransgenesis in An. gambiae. It should be noted that survival is only one component of the complex amalgam of traits that, collectively make up “fitness” and that further studies need to be performed to assess the effect of AgDNV infection on other fitness components (such as development time, fecundity, and mating behavior).