Conceived and designed the experiments: SS GD. Performed the experiments: SS JLR. Analyzed the data: SS GD. Contributed reagents/materials/analysis tools: GD. Wrote the paper: SS GD.
The authors have declared that no competing interests exist.
The female
Dengue virus (DENV) is transmitted between humans through the bite of infected
With 2.5 billion people now living in areas at risk for epidemic transmission, dengue has become the most important mosquito-borne viral disease affecting humans
Although vertical transmission of the virus has been reported
The mosquito salivary gland plays important roles in DENV transmission. Firstly, infection of the gland itself is an essential part of the transmission cycle. Secondly, the salivary gland produces numerous anti-coagulant, anti-inflammatory and vasodilatory molecules which facilitate probing and bloodmeal acquisition
Mosquitoes are exposed to a variety of microbes in their natural habitats, and possess an innate immune system capable of mounting a potent response against microbial challenge. In addition to RNA interference (RNAi)
To gain a better understanding of how the
To determine the
Of the total number of salivary gland-expressed transcripts, 2255 (13.2%) were significantly enriched in the salivary gland relative to the carcass, 2565 (15.0%) were significantly enriched in the carcass relative to the gland, while 8722 (51.1%) had a similar level of transcript abundance in the two mosquito compartments. The transcripts of 3805 genes (22.3%) were non-detectable or did not meet our signal-to-noise criteria (
(
Since our microarray-based analyses only provide information on the ratio of differential transcript abundance between the compared samples, we also considered the absolute abundance levels of salivary gland transcripts. Based on the fluorescence intensity of their spots on the microarray, we categorized transcripts into high, medium, and low abundance (
Genes that displayed differential transcript abundance between the salivary gland and the carcass represented a range of functional classes (
24 transcripts putatively involved in digestive functions were enriched in the salivary gland. Among these were six alpha-amylases (three of which belonged to the high abundance category - AAEL009524, AAEL000392, AAEL006719) and an alpha-glycosidase, which most likely play roles in the digestion of nectar meals. Several protein digestive enzyme transcripts, including 12 trypsins, an amidase, a serine protease and an endopeptidase, were enriched in the salivary gland. These enzymes in saliva could be ingested along with the bloodmeal and aid its digestion. Alternatively, they could be involved in proteolytic events that occur in the vertebrate host during blood-feeding, such as clot prevention or digestion of extra-cellular matrix components
148 transcripts with putative immunity-related functions (of which 14 belonged to the high abundance category) were salivary gland-enriched and represented a wide variety of immune gene families. These included two MD2-like proteins, six fibrinogen-related proteins, eight antimicrobial peptides (AMPs), 27 serine proteases, and several Toll pathway-related genes (three Spaetzles and two Tolls). Although nothing is known about the potential roles of these genes in salivary gland immunity, their enrichment in this organ suggests that the gland is capable of mounting potent and diverse immune responses against pathogen challenge to ensure sterility of ingested blood and nectar. AMPs may act against bacteria that the mosquito comes into direct contact with during feeding, for example in sugar sources or on vertebrate skin. One of the salivary gland-enriched AMP transcripts (AAEL000598) encodes a cecropin (AAEL000598) that has previously been found to be induced upon DENV infection in the salivary gland and to possess anti-DENV and antibacterial activities
Blood-feeding activates vertebrate host responses that inhibit blood flow, activate antimicrobial defenses, or call the attention of the host to the feeding mosquito. For example, ATP and ADP released from injured cells and activated platelets at the feeding site stimulate platelet aggregation and mast cell degranulation, and adenosine activates inflammatory responses that result in itching and burning sensations, increasing the likelihood of detection of the mosquito
Thrombin is a key enzyme in the vertebrate blood coagulation cascade. Two putative anti-thrombin transcripts (AAEL007420, AAEL006007) were salivary gland-enriched and also belonged to the highly abundant transcript category. AAEL006007 encodes a Kazal-type serine protease inhibitor that has been partially characterized and found to have anticoagulant and thrombin inhibitory activity
Transcripts of the D7 gene family are widely found in mosquito sialotranscriptomes
62 transcripts with putative chemosensory functions were enriched in the salivary gland, while only 10 were enriched in the carcass, suggesting that the salivary gland may have uncharacterized roles in chemosensory signaling. These transcripts encoded, among other proteins, three putative insect pheromone-binding protein serine/threonine kinases, seven odorant-binding proteins (OBPs), eight gustatory receptors (GRs), and numerous odorant receptors (ORs). The vast majority of these transcripts belonged to the low abundance category, with the exceptions of AAEL006408 (encoding a conserved hypothetical protein with a predicted role in odorant binding) and AAEL002587 (encoding OBP11), which belonged to the high abundance category.
OBPs are small, water soluble, secreted proteins that are highly abundant in the lymph of the sensilla of insect antennae and maxillary palps. They are specialized for ligand binding, and are thought to act as carriers for hydrophobic odorant molecules by facilitating their transport through the aqueous lymph to OR neurons
Since DENV replication in the salivary gland is a prerequisite for virus transmission, we next performed a microarray analysis to compare transcript abundance between DENV-infected and naïve salivary glands at 14 days post-bloodmeal (dpbm) to gain a better understanding of how this organ responds to infection. DENV infection stimulated a significant enrichment of 130 and depletion of 17 salivary gland transcripts (
DENV altered the abundance of 38 transcripts with functions related to metabolic processes, transport and stress response. The majority of these transcripts (33 of 38; 87%) were enriched in the infected salivary gland, perhaps indicating a shift in cellular metabolic state to support virus replication. Six transcripts with predicted cytoskeletal functions were enriched upon infection; this may reflect maintenance in structural integrity of the infected salivary gland, since cytopathology has been reported in this organ following arbovirus infection
Transcripts encoding a transferrin and a fibrinogen-related protein were also up-regulated. Transferrins bind iron with high affinity and play roles in iron metabolism, immunity, and development. They are up-regulated upon parasite or bacterial infection, and may sequester iron from pathogens; alternatively, proteolytic fragments from these proteins have also been suggested to also act as anti-microbial peptides or inducers of the immune response
Transcripts of three leucine-rich repeat (LRR)-containing proteins and one ankyrin repeat-containing protein were induced by DENV infection. The broader LRR-containing protein family includes the mosquito Tolls, and family members are commonly involved in protein-protein interactions and signal transduction pathways
Three cathepsin B transcripts and a putative cystatin transcript were induced upon DENV infection. Cathepsin Bs are lysosomal cysteine proteases known to be involved in the apoptosis of immune cells
The transcripts of three peptides with sequence similarity to secreted salivary peptides from Culicine mosquito species were also up-regulated by DENV infection. The functions of these peptides remain unknown, but some may be involved in the production of allergic reactions to mosquito bites
Finally, DENV also induced two OBP transcripts (OBP10 and OBP22), which had also been found to be enriched in the naïve salivary gland.
To determine whether our observed salivary gland transcriptomic infection responses were specific for this organ, or if they also occurred in other tissues, we went on to characterize the DENV infection-responsive carcass transcriptome at 14 dpbm. DENV infection significantly up-regulated 61 transcripts and down-regulated 74 in the carcass compartment (
Our transcriptomic analyses suggested that at least some of the DENV infection-responsive transcripts may play roles in limiting infection, or reflect a virus-mediated modulation of salivary gland functions that could have implications for mosquito behavior. We were particularly interested in modulators of DENV replication in the salivary gland. Based on our transcriptomic analyses and literature searches, we selected seven candidate genes for functional analysis via RNAi-mediated gene silencing (
Gene ID | Description | Functional Group | Log2-fold | Comment | Tested for role in | |||
DENV SG | DENV Car | Naïve SG | DENV replication | Feeding behavior | ||||
AAEL015136 | Niemann-Pick Type C-2, putative (MD6) | IMM | 1.60 | −1.19 | −1.89 | MD2-like, lipid-binding domain | X | |
AAEL009760 | Niemann-Pick Type C-2, putative (MD21) | IMM | 2.12 | 0.92 | −3.59 | MD2-like, lipid-binding domain | X | |
AAEL014906 | LAP4 protein, putative | IMM | 0.87 | 0.81 | Leucine-rich repeats | X | ||
AAEL007585 | cathepsin B | IMM | 1.49 | Lysosomal cysteine protease; pro-apoptotic, role in TLR signaling | X | |||
AAEL003728 | conserved hypothetical protein; ankyrin repeats | DIV/UNK | 0.87 | 0.92 | Ankyrin repeats present in several immune-related proteins | X | ||
AAEL017380 | hypothetical protein; putative glycine-rich salivary secreted peptide | DIV/UNK | 1.07 | 1.68 | 1.94 | Putative allergen/anti-coagulant | X | |
AAEL013287 | conserved hypothetical protein; putative cystatin | DIV/UNK | 0.81 | Cysteine protease inhibitor, pro-apoptotic, induces autophagy in mammalian cells | X | |||
AAEL005772 | Odorant-binding protein 99c, putative (OBP22) | CSR | 1.38 | 0.86 | Putative roles in olfaction | X | ||
AAEL007603 | Odorant-binding protein 56a, putative (OBP10) | CSR | 0.92 | 1.22 | Putative roles in olfaction | X |
Candidate genes selected for functional testing via RNAi-mediated gene knockdown and their associated log2-fold values in the DENV-infected salivary gland (DENV SG), DENV-infected carcass (DENV Car), and naïve salivary gland (naïve SG). Functional group abbreviations: IMM, immunity; DIV, diverse functions; UNK, unknown functions; CSR, chemosensory reception.
Mosquitoes were orally infected with DENV through a bloodmeal, and candidate genes were silenced in the salivary gland at 7 dpbm by the injection of 2 ug of dsRNA per mosquito
Silencing of the putative cystatin (AAEL013287) and the conserved hypothetical protein with ankyrin repeats (AAEL003728) genes significantly increased salivary gland DENV titers, while silencing of the cathepsin B (AAEL007585) gene resulted in significantly reduced DENV titers (
Candidate genes were silenced in DENV-infected mosquitoes, and salivary gland virus titers at 14 dpbm were determined by plaque assay. A pool of three biological replicates is shown, and p values were determined using the Mann-Whitney U test (*, p<0.05). Gene name abbreviations: AnkP, ankyrin repeat-containing protein; Cyst, cystatin; SSP, salivary secreted peptide; CatB, cathepsin B.
Since injection of dsRNA into the mosquito thorax results in non-compartment-specific silencing, it is possible that the altered virus titers observed in the salivary gland are a consequence of gene silencing in other parts of the mosquito carcass. However, we consider this less likely for several reasons: firstly, DENV infection induced these genes only in the salivary gland and not in the carcass (
Transcripts of OBPs 10 and 22 (AAEL007603 and AAEL005772) displayed an elevated abundance in the salivary gland upon DENV infection, and were also enriched in the naïve gland. This finding was unexpected and intriguing to us, and we hypothesized that these genes could participate in chemosensory signaling during host-seeking or probing. To test this hypothesis, these genes were individually silenced by the injection of 2 ug dsRNA per mosquito, 4 days prior to a behavioral feeding assay. Mosquitoes were offered an anesthetized Swiss Webster mouse, and the following parameters were measured: a) Probing propensity (percentage of mosquitoes that probed within a fixed time period); b) Probing initiation time (time from the introduction of the mouse to the time at which the mosquito initiated probing – a rough measure of host-seeking ability); and c) Probing time (time from the initial insertion of the proboscis in the skin to the initial engorgement of blood
Silencing of the OBP10 or OBP22 genes resulted in a reduced probing propensity, which was statistically significant for OBP22-silenced mosquitoes (
A behavioral feeding assay was performed 4 days post-silencing of OBPs 10 and 22. Mosquitoes were offered an anesthetized mouse and observed individually for 400 seconds. The following parameters were measured and are represented here: (
The observed effect on feeding behavior could also be due to gene silencing in the chemosensory organs (antennae and maxillary palps) instead of or in addition to the salivary gland. To further investigate the molecular basis of this interesting phenotype, we determined OBP gene silencing efficiency in these two body compartments by quantitative RT-PCR. High silencing efficiencies for both OBP genes were consistently obtained in the salivary gland (averages of 87.4% and 86.8% respectively), while efficiencies were lower and more variable in the antennae and palps (average of 22.9% for OBP10; 0%, 73.4%, and 32.5% for three trials of OBP22 (
(
To test the hypothesis that DENV infects the antennae and as such can influence OBP transcript abundance, immunofluorescent staining was first performed on head squashes of orally-infected mosquitoes at 14 dpbm. DENV-infected cells were clearly present in the antennae of infected mosquitoes but not in mock-infected controls (
Head squashes from mosquitoes at 14 dpbm were stained with mouse hyperimmune ascitic fluid to DENV and an AlexaFluor568-conjugated anti-mouse antibody. (
We next compared OBP transcript abundance in the chemosensory organs (antennae, palps and proboscis) of DENV- and mock-infected mosquitoes. While OBP22 transcript abundance was not affected by DENV infection, OBP10 transcripts were enriched by almost 2.5-fold at 14 dpbm (
Insect ORs form heteromeric complexes consisting of a conventional OR and a highly conserved universal co-receptor, termed OR co-receptor (Orco)
The behavioral and gene expression data presented above suggest that DENV infection may heighten the chemosensory abilities of mosquitoes, rendering them more efficient at bloodmeal acquisition. To test this hypothesis, we compared the blood-feeding behavior of DENV- and mock-infected mosquitoes at 14 dpbm. Slightly shorter probing initiation and probing times were observed for DENV-infected mosquitoes compared to mock-infected mosquitoes, but these differences were not statistically significant (
A behavioral feeding assay was performed on DENV- and mock-infected mosquitoes at 14 dpbm. Mosquitoes were offered an anesthetized mouse and observed for 400 seconds. The following parameters were measured and are represented here: (
We have used genome-wide microarray analyses to characterize the naïve and DENV-infected
DENV induced 130 and repressed 17 transcripts in the salivary gland at 14 dpbm, indicating that significant molecular and biochemical changes are induced in this organ by infection. In contrast, DENV infection appeared to have an overall negative effect on transcript abundance in the mosquito carcass, repressing more than half of the 135 differentially represented transcripts in this compartment at the same time point. The carcass response to infection at this late stage of infection was also subdued compared to what we have previously observed at 10 dpbm
As has been previously reported
Despite their high abundance in the antennae and maxillary palps, mosquito OBPs are not exclusively expressed in olfactory tissue. OBPs have previously been detected in the salivary gland, as well as in other body compartments such as the proboscis, thoracic spiracles, midgut, and even
The intimate association of the salivary gland and the proboscis raises the possibility that salivary gland-expressed OBPs and ORs may function in gustatory-related roles in the proboscis during blood- and sugar-feeding. Indeed, OBPs and ORs have been detected in the mosquito and fly proboscis
RNAi-mediated gene silencing assays were performed to identify genes that may modulate DENV replication in the salivary gland as a mosquito defense response. The silencing of a cathepsin B gene significantly reduced salivary gland DENV titers, while silencing of a putative cystatin significantly increased DENV titers in the salivary gland. These genes could potentially play roles in apoptosis: cathepsins are lysosomal cysteine proteases that trigger apoptosis through both caspase-dependent and –independent pathways, probably by leaking or translocating from the lysosome into the cytosol
A possible hypothesis for our observations is that apoptosis of infected cells facilitates the cell-to-cell spread of DENV in the salivary gland. A separate but related possibility is that infected cells that produce dsRNA triggers of RNAi are removed from the population by apoptosis, thereby facilitating infection by the virus. Inhibition of apoptosis (by cathepsin B silencing for example) would preserve these cells, maintaining dsRNA production and impairing DENV replication through RNAi. Since silencing a cysteine protease (cathepsin B) and a cysteine protease inhibitor (the putative cystatin) resulted in opposite effects on virus titers, it is tempting to speculate that these two genes are involved in the same process, but further studies are obviously required to test this hypothesis.
Silencing of a gene encoding a hypothetical protein containing ankyrin repeats resulted in significantly elevated salivary gland DENV titers. Ankyrin repeats mediate protein-protein interactions and are present in immune-related proteins such as the IkB inhibitory domain of Rel2, the NFkB-like transcription factor of the mosquito IMD immune signaling pathway. This protein lacks a predicted signal peptide, and could act intracellularly to regulate immune signaling.
DENV infection induced OBP10 and OBP22 transcripts in the salivary gland, a finding that surprised and intrigued us. OBPs facilitate the olfactory processes of host-seeking and probing, which mosquitoes rely on for bloodmeal acquisition. Since DENV transmission also relies on these same processes, we investigated the possibility that these OBPs influence feeding behavior. Indeed, silencing of these OBP genes reduced the percentage of mosquitoes that probed on mice, and also increased probing initiation and probing times, indicating less efficient feeding behavior. To our knowledge, this is the first observation of potential arbovirus modulation of mosquito feeding behavior through chemosensory-related molecules.
We could not conclusively determine if OBP gene silencing in the salivary glands or in the chemosensory organs was responsible for this impaired feeding behavior. OBP silencing efficiency in the salivary gland was much higher and more consistent than in the antennae and maxillary palps, suggesting that silencing in the salivary glands might be responsible for at least part of the observed feeding impairment. More efficient gene silencing in
Since both probing initiation time (a rough measure of host-seeking behavior, associated with antennal function) and probing time (associated with salivary proteins that inhibit hemostasis and inflammation) were negatively affected by gene silencing, both compartments could potentially be involved. In consideration of this, we provide evidence that DENV successfully infects and replicates in the female
Insect OBPs are quite diverse, and more than 60 members of this family have been found in the
Our data imply that viral induction of OBPs could facilitate mosquito host-seeking and/or probing behavior, and thus at least theoretically increase transmission efficiency. DENV readily infects the mosquito brain, nervous system
Apart from our data, the effect of DENV infection on mosquito feeding behavior has been previously studied, with conflicting results: one study found no effect of infection status on feeding behavior
Our transcriptomic analysis suggests novel and uncharacterized roles for many genes in salivary gland function and response to pathogens. In addition, DENV infection in the salivary gland not only regulates genes that modulate virus replication, but also genes that potentially affect bloodmeal acquisition (and hence DENV transmission) by modifying mosquito host-seeking or probing behavior. Further characterization of these genes will yield a clearer picture of these reciprocal host-pathogen interactions in this poorly-studied organ.
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Animal Care and Use Committee of the Johns Hopkins University (Permit Number: M006H300). Commercial anonymous human blood was used for dengue virus infection assays in mosquitoes, and informed consent was therefore not applicable. The Johns Hopkins School of Public Health Ethics Committee has approved this protocol.
Mosquito infections with DENV were carried out as previously described
DENV-infected and control mosquitoes were dissected at 14 days post-bloodmeal, and salivary glands and carcasses were collected and stored in Buffer RLT (Qiagen) with 1% β-mercaptoethanol. Three independent biological replicates were performed, with approximately 200 salivary glands and 20 carcasses per replicate. Total RNA was extracted from samples using the RNeasy Mini kit (Qiagen).
The Low Input Quick Amp Labeling kit (Agilent Technologies) was used to synthesize Cy-3- or Cy-5-labeled cRNA probes from total RNA (100 ng for salivary gland samples and 200 ng for carcass samples). In addition to three biological replicates, a pseudo-replicate containing an equal amount of Cy-5-labeled probe from each experimental biological replicate was also included. Hybridizations were carried out on an Agilent-based microarray platform using custom-designed whole genome 4×44K
For classification of transcripts by abundance, microarray spot hybridization fluorescence intensities were used as an indicator of transcript abundance. Spot intensity values were averaged across replicate assays for each spot, and then for each gene using the GEPAS software. Transcripts were then categorized into three categories based on mean fluorescence value (at 635 nm): high abundance (fluorescence intensity ≥5000), medium abundance (1000–5000), and low abundance (≤1000). Data are presented in
RNA interference (RNAi)-mediated candidate gene silencing in mosquitoes was performed as previously described
DENV titers in midguts and salivary glands were determined by plaque assay on BHK-21 (clone 15) cells. Individual midguts and salivary glands were homogenized in DMEM with a Bullet Blender (NextAdvance), serially diluted, and then inoculated onto cells seeded to 80% confluency in 24-well plates (100 ul per well). Plates were rocked for 15 min at room temperature, and then incubated for 45 min at 37°C and 5% CO2. Subsequently, 1 ml of DMEM containing 2% FBS and 0.8% methylcellulose was added to each well, and plates were incubated for 5 days at 37°C and 5% CO2. Plates were fixed with a methanol/acetone mixture (1∶1 volume) for >1 h at 4°C, and plaque-forming units were visualized by staining with 1% crystal violet solution for 10 min at room temperature.
Candidate genes were silenced in 4-day old female mosquitoes by the injection of 2 ug of dsRNA per mosquito (dsRNA to GFP was used as a control), and behavioral assays were carried out at 4 days post-injection. Mosquitoes were deprived of sucrose solution overnight prior to the assay. Mosquitoes were transferred in pairs to a small cage and allowed to rest for at least 10 min before being offered an anesthetized Swiss Webster mouse. The mosquitoes were observed for a maximum of 400 seconds, and the following parameters were measured for each mosquito: a) Probing propensity (percentage of mosquitoes that probed within the fixed time period of 400 seconds); b) Probing initiation time (time from the introduction of the mouse to the time at which the mosquito begins to probe); and c) Probing time (time from the initial insertion of the mouthparts in the skin to the initial engorgement of blood; if the mosquito makes multiple probing attempts, the subsequent probing times are added to the first until blood is ingested, and the interprobing times are not included
Behavioral assays involving DENV-infected mosquitoes were carried out at 14 dpbm. A maximum of seven mosquitoes were placed in a single cage and deprived of sucrose solution overnight. A Swiss-Webster mouse was euthanized via the administration of an overdose of ketamine, and offered to the mosquitoes. To counteract the drop in body temperature, a heat pack was placed over the mouse for the duration of the assay. Because of the larger number of mosquitoes per cage, video recordings were made during the assay to allow the experimenter to keep track of individual mosquitoes. As above, probing propensity, probing initiation time and probing time were measured.
Quantitative RT-PCR was used to measure relative transcript abundance of odorant-binding proteins in the mosquito chemosensory apparatus following DENV infection, as well as to measure relative DENV loads in these organs. Antennae and maxillary palps were dissected from DENV- and mock-infected mosquitoes at 10 and 14 days post-bloodmeal. Total RNA was extracted from samples using the RNeasy Mini Kit (Qiagen), treated with Turbo DNase (Ambion) and reverse-transcribed with M-MLV Reverse Transcriptase (Promega) and oligo-dT20. For detection and relative quantification of DENV transcripts, total RNA was reverse-transcribed with a DENV-specific reverse primer. Real-time quantification was performed using SYBR Green PCR Master Mix and the StepOne Plus Real-Time PCR system (Applied Biosystems). Three independent biological replicates were analyzed, and technical duplicates were run for each sample. Expression values were normalized against the ribosomal gene S7. p-values were determined with the student's t test. Primers used in this assay are listed in
Heads were dissected from DENV- and mock-infected mosquitoes at 14 days post-bloodmeal, squashed on 3-aminopropyltriethoxysilane (APES)-treated glass slides, and fixed in acetone at 4°C overnight. Slides were incubated with mouse hyper-immune ascitic fluid specific for DENV2 (diluted 1∶1000 in PBS with 0.1% Triton X-100 and 0.2% BSA) at 4°C overnight, and washed three times in PBS. Slides were then incubated with AlexaFluor 568-conjugated goat anti-mouse IgG (Molecular Probes) for 1 hour at room temperature, and washed three times in PBS. Samples were covered with ProLong Gold Antifade with DAPI (Invitrogen), and sealed with a cover-slip and nail varnish. Slides were visualized under a Leica fluorescence microscope.
(TIF)
(TIF)
(XLSX)
(XLSX)
(XLSX)
(XLSX)
(DOCX)
(DOCX)
We are grateful to Janece M. Lovchik, Bhavin Thumar, and Anna P. Durbin for providing technical support and materials (DENV-2 strains and C6/36 cell line). We thank the insectary personnel at the Johns Hopkins Malaria Research Institute for assistance with mosquito rearing, and the Microarray Core Facilities at the Johns Hopkins School of Public Health and School of Medicine for help with array scanning. We also thank Dr. Deborah McClellan for editorial assistance.