Response of the mosquito protein interaction network to dengue infection
© Guo et al; licensee BioMed Central Ltd. 2010
Received: 8 October 2009
Accepted: 16 June 2010
Published: 16 June 2010
Two fifths of the world's population is at risk from dengue. The absence of effective drugs and vaccines leaves vector control as the primary intervention tool. Understanding dengue virus (DENV) host interactions is essential for the development of novel control strategies. The availability of genome sequences for both human and mosquito host greatly facilitates genome-wide studies of DENV-host interactions.
We developed the first draft of the mosquito protein interaction network using a computational approach. The weighted network includes 4,214 Aedes aegypti proteins with 10,209 interactions, among which 3,500 proteins are connected into an interconnected scale-free network. We demonstrated the application of this network for the further annotation of mosquito proteins and dissection of pathway crosstalk. Using three datasets based on physical interaction assays, genome-wide RNA interference (RNAi) screens and microarray assays, we identified 714 putative DENV-associated mosquito proteins. An integrated analysis of these proteins in the network highlighted four regions consisting of highly interconnected proteins with closely related functions in each of replication/transcription/translation (RTT), immunity, transport and metabolism. Putative DENV-associated proteins were further selected for validation by RNAi-mediated gene silencing, and dengue viral titer in mosquito midguts was significantly reduced for five out of ten (50.0%) randomly selected genes.
Our results indicate the presence of common host requirements for DENV in mosquitoes and humans. We discuss the significance of our findings for pharmacological intervention and genetic modification of mosquitoes for blocking dengue transmission.
Dengue fever and associated dengue hemorrhagic fever are emerging globally as the most important arboviral disease threatening human populations. Dengue virus (DENV) is transmitted to humans by the mosquitoes Aedes aegypti and Aedes albopictus. When ingested into the mosquito midgut with blood, DENV first interacts with midgut cell membrane receptors and then enters the cells through receptor-mediated endocytosis. Following replication in midguts, DENV disseminates to salivary glands for transmission to human. In the mosquito, DENV is attacked by the mosquito innate immune system, including RNAi and Toll pathway [1, 2]. In addition, the effects of several host genes were reported to determine the mosquito's susceptibility to dengue infection .
DENV belongs to the genus Flavivirus of the Flaviviridae family. Flaviviruses are small, enveloped viruses with a single-stranded positive-sense RNA, which can be translated into a single polyprotein by host cell proteins. The polyprotein is then cleaved into individual proteins by both viral and host proteases . Three structural proteins (capsid, pre-membrane, and envelope) constitute the virus particle. Seven non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, NS5), 5' and 3' untranslated regions (UTR) are involved in RNA replication. The N-terminus of NS3 codes a serine protease essential for virus replication, and the central domain of NS2B is a cofactor for the NS3 serine protease. Virus-host interactions start with binding of the envelope protein E to a cellular receptor. Then, a complex interplay of viral and cellular proteins modulates the process of viral translation, replication, and assembly . Most previous studies on the replication mechanisms of DENV have been conducted in human cells. These studies identified putative cellular receptors and other host proteins that may interact with viral proteins or RNAs . In contrast, relatively little has been done to elucidate how DENV utilizes the cellular factors to enter and replicate in mosquito cells. It remains unclear which virus replication features are responsible for efficient virus passage between these different host systems.
Availability of genome sequences for both humans and the mosquito vector, Ae. aegypti, now provides us novel opportunities to identify the host factors that are required for dengue infection. Recently, microarray and RNAi screens were used for genome-wide studies of interactions between DENV and their hosts. They led to identification of a number of cellular factors that can determine the susceptibility and resistance of both hosts to DENV [2, 7]. These results also emphasize the complexity of interactions between flaviviruses and its hosts. Understanding of such interactions requires systematic and comprehensive studies of the host biology as it relates to viral infections. Analysis of protein interaction network is one way to dissect such complex interactions and has been broadly used to study the systems biology of host-pathogen interactions [8–11].
This systems biology approach requires the construction of signaling and regulatory networks. Physical interactions between two proteins are the basis for protein complexes and signaling networks, while transcription factors regulate gene transcription by binding to their target DNA sequences in the context of intricate transcriptional regulatory networks. Large amounts of high-throughput data are now available in several model organisms for protein-protein interactions [12–14], which have been extensively used in network reconstruction and systems biology studies. However, the high cost of experimental methods prevents large scale interaction network mapping of newly sequenced genomes. Computational approaches provide a viable alternative for protein-protein interaction mapping. Experimental interactomes in model organisms can be transferred to other species based on homology information. This interolog approach has been widely used to build protein interaction networks in humans [15, 16] and other species . Although unique interactions in a new species can not be found by the interolog method, the predicted interactomes may reveal conserved complexes and pathways, and serve as important tools for annotating uncharacterized proteins . The same concept can be applied to predict interactions between proteins in different species.
Uetz et al. predicted interactions between herpesvirus and human proteins based on known interacting orthologs in model organisms, and co-immuniprecipitation confirmed 68% of their predictions . DENV consists of only 10 proteins, which lack orthologs in model organisms. However, viruses in the genus Flavivirus often target the same host proteins, so known flavivirus-host interactions may be used to predict DENV-mosquito protein interactions. Recently, a high throughput RNAi screen in human cells identified 283 host susceptibility factors and 22 host resistant factors for the infection of West Nile virus (WNV; genus Flavivirus, family Flaviviridae). More importantly, targeted silencing of the WNV host factors found that half of the 283 susceptibility factors and all of the 22 resistant factors affected dengue infection in the same manner as they affected WNV infection . Additional independent studies also provide evidence that multiple flaviviruses, including WNV, Japanese encephalitis (JEV) and DENV, target the same host proteins [21, 22]. It was reported that these flaviviruses share a similar genomic organization and replication strategy although they cause a range of distinct clinical diseases in humans .
Evidence also indicates the notable conservation of required dengue host factors between insects and humans. Recently, a genome-wide RNA interference screen in Drosophila melanogaster cells identified 116 dengue virus host factors. Among them, 82 have human homologues. Using targeted siRNA, 51.2% of them proved to be dengue host factors in humans . A number of conserved host factors were also identified in several earlier studies. In both mosquitoes and humans, La protein interacts with 5' and 3' ends of dengue viral RNA and two nonstructural proteins NS3 and NS5. Such interactions seem important for assembly of a functional replication complex of dengue virus [23, 24]. Laminin-binding protein and heat shock proteins were also indicated to mediate the dengue infection in both humans and mosquitoes [21, 25–28]. Moreover, a similar internalization route (clathrin-mediated endocytosis) was used by dengue viruses to enter into human and mosquito cells.
In this study, we have developed the mosquito protein interaction network based on large-scale protein interaction datasets in Saccharomyces cerevisiae, Caenorhabditis elegans and Drosophila melanogaster. 714 flavivirus-associated mosquito proteins were inferred from three data sources based on physical interaction assays, genome-wide RNAi screens and microarray results. An integrated analysis of dengue-associated protein interaction sub-networks highlighted four regions consisting of highly interconnected proteins with closely related functions. Finally, we verified that five out of ten (50.0%) randomly selected genes were required for DENV infection in the main mosquito vector, Ae. aegypti, by RNAi-mediated gene silencing.
Results and Discussion
Construction of the first-draft mosquito protein interaction map
Reliable protein interaction datasets in model organisms are the basis of the interolog method. Because large-scale protein interaction mapping generates a high number of false positive interactions, it is essential to assess the quality of experimentally determined protein interactions. To generate the draft mosquito interaction network, we selected different interaction datasets with available confidence scores or rankings. The dataset in S. cerevisiae includes 47,783 interactions identified by co-immunoprecipitation and yeast two-hybrid screens. Each interaction is associated with a confidence score estimated by the logistic regression model . The two-hybrid-based dataset in D. melanogaster includes 20,405 interactions with a confidence score estimated by a generalized linear model . The dataset in C. elegans (4,736 interactions) includes literature-mined data and two-hybrid generated data divided into three confidence classes . For all three datasets, confidence metrics were generated using only statistical and topological predictors, such as the number of interaction partners and a local clustering coefficient. These scores were combined with the InParanoid  score to generate confidence values for predicted interactions in Ae. aegypti.
Using predicted interactions with confidence values more than 0.5, we constructed a weighted protein interaction network for Ae. aegypti (Fig. 1B). The network includes 4,214 mosquito proteins with 10,209 interactions, of which 62%, 25%, 11% and 2% were inferred from yeast, fly, worm, and more than one reference organism, respectively (Fig. 1B). Among all the proteins in our network, 3,500 proteins are connected into a single interconnected network by 9,719 interactions. The number of interacting partners per node in the interconnected network is 5.55, and the degree distribution fits a power law distribution with a degree exponent of 1.79. As expected, the network also has a high clustering coefficient (0.32) and a short path length (6.46).
Prediction of mosquito protein function
Cellular functions are often carried out by a group of physically or functionally linked molecules that work together to complete distinct tasks. The high clustering coefficient observed in our network indicates that isolated functional modules may be identified from the network. The Markov Cluster algorithm (MCL) identifies highly connected modules based on simulation of stochastic flow in graphs. It has been shown to be the most robust algorithm for the extraction of protein complexes from protein-protein interaction networks . Using this algorithm, we parsed the mosquito protein interaction network into 494 densely connected clusters (Additional file 3).
Emerging hypotheses of biological pathways in mosquitoes
Detailed examination of local interaction networks may reveal interconnections of biological pathways and provide further insights on how a biological process is controlled in mosquitoes. Here we focus on the mosquito biological processes that have attracted increasing interest in recent years.
STAT is a key component in the JAK/STAT pathway. In the mosquito protein interaction network, it interacts with eight proteins including AAEL014557, an ortholog of Drosophila caudal (Cad). Cad also interacts with two proteins associated with the Insulin signaling pathway: Pi3K21B [VectorBase: AAEL013596] and AKT1 [VectorBase: AAEL008823] (Fig. 3). This implies a potential link between the JAK/STAT and Insulin pathways. Further examination of the network found Cad interacting with 49 proteins. Thus, as a hub protein in the networks, Cad may have diverse functions and be essential for the survival of Ae. aegypti.
Sex lethal (Sxl) is involved in sex determination in Drosophila. In the network, eight proteins interact with Ae. aegypti Sxl [VectorBase: AAEL011150]. Two of them are hypothetical proteins. Although no evidence suggests these eight proteins are involved in sex determination, six of them are involved in either RNA binding or RNA helicase activity (Fig. 3). Thus, similar to its Drosophila ortholog, Sxl might function as a splicing and translational regulator in mosquito.
Mosquito odorant-binding proteins (Obp) were input into the network to identify their interaction partners. Five odorant-binding proteins and two hypothetical proteins interact with the sterol carrier protein-2 [VectorBase: AAEL002687], a protein which is thought to be involved in intracellular cholesterol transport. OBP56e [VectorBase: AAEL006650] interacts with Hyperkinetic, a gene involved in potassium ion transport, and one hypothetical protein (Fig. 3).
Prediction and analysis of mosquito responses to dengue infection
A comprehensive mapping of mosquito protein interactions targeted by DENV could provide specific hypotheses and a broad perspective on DENV strategies for replication and persistence in mosquitoes. Previous microarray assays revealed broad responses to DENV infection in mosquitoes that entailed a variety of physiological systems . A recent genome wide RNAi screen in human demonstrated a complex dependence of flaviviruses on host cell physiology, requiring a wide variety of molecules and cellular pathways for successful infection . In addition, studies of physical interactions between flaviviruses and their hosts also provided insight on how viruses hijack the host pathways for their entry and replication. With the mosquito protein interaction network constructed, we can now put all these together and perform an integrated analysis to identify the spectrum of mosquito pathways activated by DENV and other flaviviruses.
Prediction of dengue-associated Ae. aegypti proteins
We inferred Ae. aegypti proteins associated with DENV infection using three separate datasets. The first dataset was based on RNAi screens in human and Drosophila cell lines. Among 82 DENV host factors identified in Drosophila with recognizable human homologues, 42 are also human DENV host factors . In addition, 305 human genes have been shown to be associated with WNV, of which 124 also affect DENV infection . Given these virus-associated host genes, we inferred 243 Ae. aegypti orthologs which may be involved in DENV infection (Additional file 5). These included 46, 168 and 29 Ae.aegypti orthologs from DENV host factors in Drosophila, human and both, respectively. Secondly, we predicted 22 mosquito proteins that may directly bind to flavivirus proteins or RNAs based on known host-flavivirus physical interactions collected by literature search (Additional file 6). The third dataset was derived by genome-wide transcriptional profiling of Ae. aegypti response to dengue infection. A large number of genes (n = 456) in mosquito midgut and the remaining carcass are differentially expressed in response to virus infection . Taken together, we have identified 714 putative dengue associated-mosquito proteins.
Putative DENV-associated Ae. aegypt i proteins supported by more than one experimental evidences
Gene name or description
tubulin beta chain
clathrin coat assembly protein
serine/threonine protein kinase
conserved hypothetical protein
Enriched Gene Ontology (GO) biological process terms in three datasets of DENV-associated Ae. aegypti proteins
Establishment of Localization
Response to stimulus
Network analysis of putative dengue-associated mosquito proteins
RTT is the most complicated region in the major DENV-associated sub-network, and is comprised of 133 proteins (Fig. 4, Additional file 7). Except for 31 proteins with unknown or diverse functions, the largest proportion (92 out of 133; 69.2%) of the proteins in this region plays roles in replication, transcription and translation. This can be explained by DENV hijacking cellular proteins to aid in their own replication and transcription and their reliance on host machinery for their translation. As an example, La protein [VectorBase: AAEL003664] was reported to bind to the 3'and 5' end UTR of viral RNA . La protein has also been shown to inhibit viral RNA synthesis in a dose-dependent manner in mosquito Ae. albopictus. As a RNA-binding protein, La is normally located in the nucleus. It has to be redistributed to the cytoplasm of dengue-infected cells to play its role in the viral replication process. It is still unknown how to trigger the export of La protein from the nucleus to the cytoplasm for the binding with the 3'and 5' UTR and regulation of viral RNA synthesis.
Metabolism is another function that was highlighted in one region of the sub-network comprising highly interconnected proteins (Fig. 4). 15 out of 23 proteins (65.2%) in this region function in metabolism. They include genes that express enzymes to catalyze the reactions in the biosynthesis pathway of L-arginine (AAEL003345), aspartate, phenylalanine (AAEL012579), isoprenoid (AAEL003497) and tetrahydrofolate (AAEL002261). Other proteins in this region are involved in metabolic process of galactose (AAEL001607), phosphorus (AAEL003193), malate (AAEL008166, AAEL001073), lyxose (AAEL013821, AAEL010206), mevalonate pathway I (AAEL005201) and arginase pathway (AAEL005289). As a parasite, DENV depends on the metabolic network of the host cell to provide the energy and macromolecule subunits necessary for their replication. Consequently, host infected by the virus can produce dramatic metabolic alterations. Interestingly, three proteins (AAEL012579, AAEL003497 and AAEL005201) that are related to lipid metabolism were found in this group. Further studies are needed to study the impact of DENV on mosquito metabolism, especially lipid metabolism, and the effects of host metabolic perturbation on DENV replication in mosquitoes.
Immunity is the third function that was highlighted in the major DENV-associated sub-network. A region with highly interconnected proteins consists of the Toll pathway genes and heat shock proteins (Fig. 4). It includes REL1, CACT and three heat shock proteins [VectorBase: AAEL011704, AAEL011708 and AAEL014843]. Highlighting of this Toll pathway-related proteins is consistent with our previous report that the Ae. aegypti Toll pathway can control dengue infection . Previous studies also show that the Onyong-nyong virus induces expression of the heat shock protein cognate 70B in An. gambiae, which has an inhibitory effect on the replication of this virus . It will be interesting to explore whether the Toll pathway can regulate the expression of heat shock proteins. There are eight proteins that have never been reported to be associated with dengue infection in this region. Three of them are hypothetical proteins. The others are AAEL008743, AAEL005080, AAEL013896, AAEL002818 and AAEL003684. Among them, AAEL002818 may be related to immune defense. Knock-down of its Drosophila ortholog, U2af50, can induce the expression of immune genes .
Fourteen highly interconnected proteins are included in the region with a main function in transport (Fig. 4). Ten of them are involved in ion transport while additional four encode tubulin alpha or beta chains, which can mediate microtubule-based movement. Previous studies found that Sindbis viruses induce dramatic changes in the expression of mosquito genes involved in ion transport processes .
Validation of the predicted dengue-mosquito interactions in Ae. aegypti
Notably, knockdown of a putative tumor suppressor protein [VectorBase: AAEL012515] led to a 13-fold decrease in DENV infection in the mosquito midguts (Fig. 6). The Drosophila ortholog of AAEL012515, TSG101, is the negative regulator of the JAK/STAT pathway and functions as a ubiquitin-protein ligase . The JAK/STAT pathway has been demonstrated to be required for anti-viral defense in Drosophila. In humans, silencing of PIAS2, the protein inhibitor of activated STAT, can also reduce both DENV and WNV infections . The above results strongly suggest that the JAK/STAT pathway plays an important role in defense against dengue infection in mosquitoes. Our previous studies also show that several important components of this pathway, including DOME [VectorBase: AAEL012471], were up-regulated in response to dengue infection in Ae. aegypti. Thus, the JAK/STAT pathway might play a role in regulation of dengue infection in mosquitoes.
Silencing of the other four genes, AAEL001005, AAEL013989, AAEL005351 and AAEL013275, also significantly reduced the dengue infection in mosquito midguts (Fig. 6). AAEL001005 is a putative mosquito calreticulin protein, whose ortholog in human has been shown to form a complex with 3'UTR in the minus-strand RNA of DENV . Considering that the regulatory sequences involved in viral replication reside in both ends of the flavivirus genome, it is very likely that calreticulin plays a role in the replication of DENV. AAEL013989 is a putative protein translocation complex beta subunit with protein transporter activity. This gene was induced in Ae. aegypti in response to Sindbis virus infection . Silencing of its Drosophila and human ortholog, Sec61β, significantly reduced the dengue infection in S2 cell line and HuH-7 cells, respectively . AAEL005351 is Leucine-rich transmembrane protein. Similarly, its Drosophila and human orthologs were proven to be required for dengue infection . Importin beta-1, AAEL013275, is also a protein with transmembrane transporter activity. Previous studies found that dengue virus NS5 is able to interact with the nuclear import receptor importin-β although the roles of this interaction is still unclear.
Our RNAi results indicate the notable conservation of required factors between mosquito and human hosts. As described in the above, all the five genes verified here are known to play important roles in interactions of DENV with human host. It appears that common host factors present in human and mosquito, and their interactions with DENV may determine the high efficiency of DENV in entry and replication within these two distantly related animals. In order to maintain the transmission cycle of DENV between mosquito and human, a perfect match in the cellular components and conserved DENV-host interactions may be required between the two host species. Our study makes the first steps towards identification of common pathways that are hijacked by DENV in humans and mosquitoes. A recent study using a genome-wide RNA interference screen in D. melanogaster cells found that 51.2% of the human homolog of dengue associated insect host factors were also required for dengue infection in human . This is consistent with our results that 50.0% mosquito orthologs of dengue human host factors are required for its infection in mosquito.
Our studies also show that the protein interaction network is a powerful tool in the study of vector biology. It can not only provide an additional tool for further annotation of vector genome, but can also serve as a novel platform to integrate datasets from different experiments in order to dissect complicated biological processes in insect vectors. By integrating the results from high-throughput assays such as genome-wide RNAi screens, microarrays and yeast two-hybrid system, a protein interaction network can significantly increase prediction accuracy. Moreover, multiple and concurrent signal-transduction pathways operate and are deregulated during the course of vector-pathogen interactions. Such highly interconnected and temporally and spatially regulated signaling pathways can only be dissected by system-level approaches. Construction of a protein interaction network can lay a foundation to study the system biology of pathogen-vector interactions, as shown in our integrated analysis of DENV-associated mosquito proteins and mosquito protein interaction network.
As a first step, we used protein interactions data from model organisms to infer a mosquito protein interaction network. In order to understand the unique biological processes in mosquitoes, it is necessary to construct the network using experimental data directly from mosquitoes, such as a high-throughput yeast two-hybrid system. Such an independent mosquito network will allow us to compare and contrast the network's response to dengue infection between humans and mosquitoes. This can reveal unique features of dengue-host interactions in these two species, which will help us understand why DENV causes serious disease in human but no deleterious effects in the mosquito vector.
Arboviruses pass through two different hosts in their life cycle, with great benefit taken from vectors for their mobility, diffusion and evolution. From an implementation standpoint, this also provides unique opportunity for identification of weak linkages in their life history to target using antiviral agents. For example, conserved DENV host factors may be used as targets for anti-dengue agents for blocking dengue transmission in mosquitoes or for pharmacological intervention in humans. With the genome available, insect vectors may be used as an animal model to study the basic biology of arboviruses, and develop novel therapeutics or control strategies.
We have developed a mosquito protein interaction network based on large-scale protein interaction datasets in yeast, worm, and fly. The weighted interaction network includes 4,214 mosquito proteins with 10,209 interactions, among which 3,500 proteins are connected into a single interconnected network by 9,719 interactions. This network not only provides a valuable resource for the further annotation of the Ae. aegypti genome, but also makes the first step towards studies of mosquito systems biology. In this regard, we identified a cluster of Toll pathway related protein interactions, and predicted the potential link between the Toll pathway, RNAi pathway and expression of heat shock proteins. We then used datasets from physical interaction assays, RNAi and microarray to infer Ae. aegypti proteins associated with dengue infection. This led to the identification of 714 putative dengue-associated mosquito proteins. An integrated analysis of these proteins in the network highlighted four regions consisting of highly interconnected proteins with closely related functions in each of replication/transcription/translation (RTT), immunity, transport and metabolism. Finally, by knocking down the predicted DENV-associated mosquito proteins with RNAi, we successfully verified that five of ten candidate genes were required for DENV infection. Our results support the presence of common host requirement of DENV in human and mosquito.
Prediction of protein-protein interactions in Ae. aegypti
We retrieved three high-throughput protein interaction datasets in model organisms along with their confidence scores [12, 13, 29]. InParanoid was used to determine homology between Ae. aegypti and three model organisms, S. cerevisiae, D. melanogaster, and C. elegans. Proteins A and B in Ae. aegypti are predicted to interact if their respective orthologs A' and B' are interacting partners in a model organism. We use IP(AA') and IP(BB') to denote InParanoid scores for ortholog pairs AA' and BB', CS(A'B') to indicate confidence values of protein interactions in the model organism. Then, the confidence of predicted interactions in Ae. aegypti, CS(AB) is CS(A'B') × average of IP(AA') and IP(BB'). If orthologs of proteins A and B in more than one model organism show evidence of interaction, we will have more than one confidence score for A and B. In this case, the maximum value is taken as interaction confidence.
We then retrieved GO terms for all Ae. aegypti proteins based on the TIGR annotation pipeline . GO is comprised of three subcategories, biological process (BP), molecular function (MF), and cellular component (CC). Semantic similarity values were calculated for each predicted interacting protein pair in terms of BP, MF and CC using the SemSim package in BioConductor http://www.bioconductor.org/packages/2.2/bioc/vignettes/SemSim/inst/doc/SemSim.pdf. Specifically, Resnik's approach is applied to calculate GO term similarity based on the assumption that the more information two terms share, the more similar they are . The shared information of two terms is indicated by the information content of their parents, defined as the frequency of each term, or any of its children, occurring in an annotated dataset. Less frequently occurring terms are said to be more informative. GO term similarity defines semantic similarity of proteins annotated with these terms. In the case of multiple annotations for a protein, maximum term similarity is taken as the protein similarity .
Network analysis of weighted protein interactions
Global measures of network topology provide quantitative insight into biological systems. We used the igraph package in the R computing environment to analyze the degree distribution, clustering coefficient, characteristic path length, and diameter of our network. Then, Markov Clustering algorithm (MCL) was applied to partition our mosquito interaction network into densely connected clusters , which were displayed and analyzed using the GenePro plugin in Cytoscape . GenePro displays protein clusters as individual nodes interconnected through interacting proteins. Each individual cluster may be expanded to allow detailed analysis in terms of component proteins and their interactions in the context of the full network.
Prediction and analysis of dengue-Ae. aegypti mosquito protein interactions
Known physical interactions between flavivirus proteins/RNA and their host proteins were collected by literature search. If an ortholog can be identified in Ae. aegypti for the host protein, then the mosquito protein and corresponding virus protein/RNA are predicted to interact. Both Inparanoid and Ensembl were used for ortholog identification, and only the mosquito protein(s) with the highest Inparanoid score were used for downstream analysis. Similarly, mosquito orthologs were identified for a set of human and Drosophila genes associated with DENV and WNV that were derived from RNA interference screen . Then, a hypergeometric test of GO annotations was used to discover enriched functions in the predicted sets of mosquito proteins using the BinGO plugin  in Cytoscape. These proteins were also mapped to the mosquito protein interaction network and the resulting subnetwork was visualized by Cytoscape. The functional classification of mosquito proteins in each module was performed as described previously . The graph density of each functional group or random protein groups with the same size was calculated by graph.density function in an R package igraph http://igraph.sourceforge.net/doc/R/graph.density.html.
Mosquito rearing and cell culture maintenance
Ae. aegypti mosquitoes of the Rockefeller/UGAL strain were maintained on a sugar solution at 27°C and 85% humidity with a 12-hr light/dark cycle according to standard rearing procedures. The Ae. albopictus cell line C6/36 was grown in minimal essential medium (MEM) with 10% heat inactivated FBS, 1% L-glutamine, and 1% non-essential amino acids at 32°C with 5% CO2.
DsRNA were synthesized from PCR-amplified gene fragments using the T7 Megascript kit (Ambion). The sequences of the primers are listed in Additional file 9. RNAi-based gene-silencing assays were conducted as previously reported: Approximately 69 nl dsRNAs (4 mg/ml) in water were injected into the thorax of cold-anesthetized 4-day-old female mosquitoes using a nano-injector . Three to four days after injection, mosquitoes were fed on a DENV-2-supplemented blood meal.
The New Guinea C strain of DENV-2 was propagated in C6/36 cells according to standard conditions. In brief, 0.5 ml aliquots of virus stock were used to infect 75-cm2 flasks of C6/36 cells at 80% confluency with a multiplicity of infection (MOI) of 3.5 virus particles/cell. Infected cells were incubated for 5-7 days. Cells were harvested with a cell scraper and lysed to release the virus particles by repeated freezing and thawing in dry CO2 and a 37°C water bath. The virus suspension was mixed 1:1 with commercial sheep blood. The blood meal (with 2 × 107 PFU/ml DENV) was maintained at 37°C for 30 min prior to feeding 6- to 7-day-old mosquitoes http://www.jove.com/index/Details.stp?ID=220. We confirmed that the virus titer we used resulted in a 90% infection rate in experimentally infected females by an indirect fluorescent antibody assay performed on head squashes at 14 days after the infectious meal.
Mosquito dissection and Plaque assay
Seven days after blood meal, mosquitoes were briefly washed in 70% ethanol, then rinsed in sterile distilled water. The midguts were dissected in sterile PBS and transferred separately to microcentrifuge tubes containing 150 ul of MEM, then homogenized with a Kontes pellet pestle motor in a sterile environment. Each replicate contain 4-5 midguts. At least three independent biological replicate assays were performed for each gene.
The virus-containing homogenates were serially diluted and inoculated into C6/36 cells in 24-well plates. After incubation for 5 days at 32°C and 5% CO2, the plates were assayed for plaque formation 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.
List of abbreviations
Japanese encephalitis virus
Hepatitis C virus
Biomolecular Interaction Network Database
We are grateful to Dr. George Dimopoulos for providing DENV-2 strain and C6/36 cell line for this study. We are thankful to the Arbovirus Diseases Branch at the CDC for providing us with the anti-dengue antibodies (Mouse Hyperimmune Ascetic Fluid). We also thank Drs. Suzanne Thiem, C. Titus Brown and Edward D. Walker for editorial assistance and comments on the manuscript. This work was supported by the National Institutes of Health [1R01AI080597 to Z. X.]; and Michigan State University [08-IRGP-1452].
- Franz AW, Sanchez-Vargas I, Adelman ZN, Blair CD, Beaty BJ, James AA, Olson KE: Engineering RNA interference-based resistance to dengue virus type 2 in genetically modified Aedes aegypti. Proc Natl Acad Sci USA. 2006, 103 (11): 4198-4203. 10.1073/pnas.0600479103.PubMed CentralPubMedView ArticleGoogle Scholar
- Xi Z, Ramirez JL, Dimopoulos G: The Aedes aegypti toll pathway controls dengue virus infection. PLoS Pathog. 2008, 4 (7): e1000098-10.1371/journal.ppat.1000098.PubMed CentralPubMedView ArticleGoogle Scholar
- Gomez-Machorro C, Bennett KE, del Lourdes Munoz M, Black WCt: Quantitative trait loci affecting dengue midgut infection barriers in an advanced intercross line of Aedes aegypti. Insect Mol Biol. 2004, 13 (6): 637-648. 10.1111/j.0962-1075.2004.00522.x.PubMedView ArticleGoogle Scholar
- Chambers TJ, Hahn CS, Galler R, Rice CM: Flavivirus genome organization, expression, and replication. Annu Rev Microbiol. 1990, 44: 649-688. 10.1146/annurev.mi.44.100190.003245.PubMedView ArticleGoogle Scholar
- Lindenbach BD, Rice CM: Molecular biology of flaviviruses. Adv Virus Res. 2003, 59: 23-61. full_text.PubMedView ArticleGoogle Scholar
- Fernandez-Garcia MD, Mazzon M, Jacobs M, Amara A: Pathogenesis of flavivirus infections: using and abusing the host cell. Cell Host Microbe. 2009, 5 (4): 318-328. 10.1016/j.chom.2009.04.001.PubMedView ArticleGoogle Scholar
- Sessions OM, Barrows NJ, Souza-Neto JA, Robinson TJ, Hershey CL, Rodgers MA, Ramirez JL, Dimopoulos G, Yang PL, Pearson JL: Discovery of insect and human dengue virus host factors. Nature. 2009, 458 (7241): 1047-1050. 10.1038/nature07967.PubMed CentralPubMedView ArticleGoogle Scholar
- Calderwood MA, Venkatesan K, Xing L, Chase MR, Vazquez A, Holthaus AM, Ewence AE, Li N, Hirozane-Kishikawa T, Hill DE: Epstein-Barr virus and virus human protein interaction maps. Proc Natl Acad Sci USA. 2007, 104 (18): 7606-7611. 10.1073/pnas.0702332104.PubMed CentralPubMedView ArticleGoogle Scholar
- Forst CV: Host-pathogen systems biology. Drug Discov Today. 2006, 11 (5-6): 220-227. 10.1016/S1359-6446(05)03735-9.PubMedView ArticleGoogle Scholar
- de Chassey B, Navratil V, Tafforeau L, Hiet MS, Aublin-Gex A, Agaugue S, Meiffren G, Pradezynski F, Faria BF, Chantier T: Hepatitis C virus infection protein network. Mol Syst Biol. 2008, 4: 230-10.1038/msb.2008.66.PubMed CentralPubMedView ArticleGoogle Scholar
- Dyer MD, Murali TM, Sobral BW: The landscape of human proteins interacting with viruses and other pathogens. PLoS Pathog. 2008, 4 (2): e32-10.1371/journal.ppat.0040032.PubMed CentralPubMedView ArticleGoogle Scholar
- Giot L, Bader JS, Brouwer C, Chaudhuri A, Kuang B, Li Y, Hao YL, Ooi CE, Godwin B, Vitols E: A protein interaction map of Drosophila melanogaster. Science. 2003, 302 (5651): 1727-1736. 10.1126/science.1090289.PubMedView ArticleGoogle Scholar
- Li S, Armstrong CM, Bertin N, Ge H, Milstein S, Boxem M, Vidalain PO, Han JD, Chesneau A, Hao T: A map of the interactome network of the metazoan C. elegans. Science. 2004, 303 (5657): 540-543. 10.1126/science.1091403.PubMed CentralPubMedView ArticleGoogle Scholar
- Uetz P, Giot L, Cagney G, Mansfield TA, Judson RS, Knight JR, Lockshon D, Narayan V, Srinivasan M, Pochart P: A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae. Nature. 2000, 403 (6770): 623-627. 10.1038/35001009.PubMedView ArticleGoogle Scholar
- Lehner B, Fraser AG: A first-draft human protein-interaction map. Genome Biol. 2004, 5 (9): R63-10.1186/gb-2004-5-9-r63.PubMed CentralPubMedView ArticleGoogle Scholar
- Brown KR, Jurisica I: Online predicted human interaction database. Bioinformatics. 2005, 21 (9): 2076-2082. 10.1093/bioinformatics/bti273.PubMedView ArticleGoogle Scholar
- Geisler-Lee J, O'Toole N, Ammar R, Provart NJ, Millar AH, Geisler M: A predicted interactome for Arabidopsis. Plant Physiol. 2007, 145 (2): 317-329. 10.1104/pp.107.103465.PubMed CentralPubMedView ArticleGoogle Scholar
- Ideker T, Lauffenburger D: Building with a scaffold: emerging strategies for high- to low-level cellular modeling. Trends Biotechnol. 2003, 21 (6): 255-262. 10.1016/S0167-7799(03)00115-X.PubMedView ArticleGoogle Scholar
- Uetz P, Dong YA, Zeretzke C, Atzler C, Baiker A, Berger B, Rajagopala SV, Roupelieva M, Rose D, Fossum E: Herpesviral protein networks and their interaction with the human proteome. Science. 2006, 311 (5758): 239-242. 10.1126/science.1116804.PubMedView ArticleGoogle Scholar
- Krishnan MN, Ng A, Sukumaran B, Gilfoy FD, Uchil PD, Sultana H, Brass AL, Adametz R, Tsui M, Qian F: RNA interference screen for human genes associated with West Nile virus infection. Nature. 2008, 455 (7210): 242-245. 10.1038/nature07207.PubMed CentralPubMedView ArticleGoogle Scholar
- Ren J, Ding T, Zhang W, Song J, Ma W: Does Japanese encephalitis virus share the same cellular receptor with other mosquito-borne flaviviruses on the C6/36 mosquito cells?. Virol J. 2007, 4: 83-10.1186/1743-422X-4-83.PubMed CentralPubMedView ArticleGoogle Scholar
- Chu JJ, Leong PW, Ng ML: Characterization of plasma membrane-associated proteins from Aedes albopictus mosquito (C6/36) cells that mediate West Nile virus binding and infection. Virology. 2005, 339 (2): 249-260. 10.1016/j.virol.2005.05.026.PubMedView ArticleGoogle Scholar
- Garcia-Montalvo BM, Medina F, del Angel RM: La protein binds to NS5 and NS3 and to the 5' and 3' ends of Dengue 4 virus RNA. Virus Res. 2004, 102 (2): 141-150. 10.1016/j.virusres.2004.01.024.PubMedView ArticleGoogle Scholar
- Yocupicio-Monroy M, Padmanabhan R, Medina F, del Angel RM: Mosquito La protein binds to the 3' untranslated region of the positive and negative polarity dengue virus RNAs and relocates to the cytoplasm of infected cells. Virology. 2007, 357 (1): 29-40. 10.1016/j.virol.2006.07.042.PubMedView ArticleGoogle Scholar
- Thepparit C, Smith DR: Serotype-specific entry of dengue virus into liver cells: identification of the 37-kilodalton/67-kilodalton high-affinity laminin receptor as a dengue virus serotype 1 receptor. J Virol. 2004, 78 (22): 12647-12656. 10.1128/JVI.78.22.12647-12656.2004.PubMed CentralPubMedView ArticleGoogle Scholar
- Sakoonwatanyoo P, Boonsanay V, Smith DR: Growth and production of the dengue virus in C6/36 cells and identification of a laminin-binding protein as a candidate serotype 3 and 4 receptor protein. Intervirology. 2006, 49 (3): 161-172. 10.1159/000089377.PubMedView ArticleGoogle Scholar
- Reyes-Del Valle J, Chavez-Salinas S, Medina F, Del Angel RM: Heat shock protein 90 and heat shock protein 70 are components of dengue virus receptor complex in human cells. J Virol. 2005, 79 (8): 4557-4567. 10.1128/JVI.79.8.4557-4567.2005.PubMed CentralPubMedView ArticleGoogle Scholar
- Salas-Benito J, Reyes-Del Valle J, Salas-Benito M, Ceballos-Olvera I, Mosso C, del Angel RM: Evidence that the 45-kD glycoprotein, part of a putative dengue virus receptor complex in the mosquito cell line C6/36, is a heat-shock related protein. Am J Trop Med Hyg. 2007, 77 (2): 283-290.PubMedGoogle Scholar
- Bader JS, Chaudhuri A, Rothberg JM, Chant J: Gaining confidence in high-throughput protein interaction networks. Nat Biotechnol. 2004, 22 (1): 78-85. 10.1038/nbt924.PubMedView ArticleGoogle Scholar
- Remm M, Storm CE, Sonnhammer EL: Automatic clustering of orthologs and in-paralogs from pairwise species comparisons. J Mol Biol. 2001, 314 (5): 1041-1052. 10.1006/jmbi.2000.5197.PubMedView ArticleGoogle Scholar
- Lu LJ, Xia Y, Paccanaro A, Yu H, Gerstein M: Assessing the limits of genomic data integration for predicting protein networks. Genome Res. 2005, 15 (7): 945-953. 10.1101/gr.3610305.PubMed CentralPubMedView ArticleGoogle Scholar
- Guo X, Liu R, Shriver CD, Hu H, Liebman MN: Assessing semantic similarity measures for the characterization of human regulatory pathways. Bioinformatics. 2006, 22 (8): 967-973. 10.1093/bioinformatics/btl042.PubMedView ArticleGoogle Scholar
- Brohee S, van Helden J: Evaluation of clustering algorithms for protein-protein interaction networks. BMC Bioinformatics. 2006, 7: 488-10.1186/1471-2105-7-488.PubMed CentralPubMedView ArticleGoogle Scholar
- Shin SW, Kokoza V, Bian G, Cheon HM, Kim YJ, Raikhel AS: REL1, a homologue of Drosophila dorsal, regulates toll antifungal immune pathway in the female mosquito Aedes aegypti. J Biol Chem. 2005, 280 (16): 16499-16507. 10.1074/jbc.M500711200.PubMedView ArticleGoogle Scholar
- Christophides GK, Vlachou D, Kafatos FC: Comparative and functional genomics of the innate immune system in the malaria vector Anopheles gambiae. Immunol Rev. 2004, 198: 127-148. 10.1111/j.0105-2896.2004.0127.x.PubMedView ArticleGoogle Scholar
- Hoffmann JA: The immune response of Drosophila. Nature. 2003, 426 (6962): 33-38. 10.1038/nature02021.PubMedView ArticleGoogle Scholar
- Lemaitre B, Nicolas E, Michaut L, Reichhart JM, Hoffmann JA: The dorsoventral regulatory gene cassette spatzle/Toll/cactus controls the potent antifungal response in Drosophila adults. Cell. 1996, 86 (6): 973-983. 10.1016/S0092-8674(00)80172-5.PubMedView ArticleGoogle Scholar
- Ishizuka A, Siomi MC, Siomi H: A Drosophila fragile X protein interacts with components of RNAi and ribosomal proteins. Genes Dev. 2002, 16 (19): 2497-2508. 10.1101/gad.1022002.PubMed CentralPubMedView ArticleGoogle Scholar
- Waterhouse RM, Kriventseva EV, Meister S, Xi Z, Alvarez KS, Bartholomay LC, Barillas-Mury C, Bian G, Blandin S, Christensen BM: Evolutionary dynamics of immune-related genes and pathways in disease-vector mosquitoes. Science. 2007, 316 (5832): 1738-1743. 10.1126/science.1139862.PubMed CentralPubMedView ArticleGoogle Scholar
- Modis Y, Ogata S, Clements D, Harrison SC: A ligand-binding pocket in the dengue virus envelope glycoprotein. Proc Natl Acad Sci USA. 2003, 100 (12): 6986-6991. 10.1073/pnas.0832193100.PubMed CentralPubMedView ArticleGoogle Scholar
- Chee HY, AbuBakar S: Identification of a 48 kDa tubulin or tubulin-like C6/36 mosquito cells protein that binds dengue virus 2 using mass spectrometry. Biochem Biophys Res Commun. 2004, 320 (1): 11-17. 10.1016/j.bbrc.2004.05.124.PubMedView ArticleGoogle Scholar
- Yocupicio-Monroy RM, Medina F, Reyes-del Valle J, del Angel RM: Cellular proteins from human monocytes bind to dengue 4 virus minus-strand 3' untranslated region RNA. J Virol. 2003, 77 (5): 3067-3076. 10.1128/JVI.77.5.3067-3076.2003.PubMed CentralPubMedView ArticleGoogle Scholar
- Shafee N, AbuBakar S: Effects of Dengue 2 Virus Inoculation on. Toxorhynchites splendens Larvae. Journal of Entomology. 2006, 3 (1): 89-94. 10.3923/je.2006.89.94.View ArticleGoogle Scholar
- Sim C, Hong YS, Tsetsarkin KA, Vanlandingham DL, Higgs S, Collins FH: Anopheles gambiae heat shock protein cognate 70B impedes o'nyong-nyong virus replication. BMC Genomics. 2007, 8: 231-10.1186/1471-2164-8-231.PubMed CentralPubMedView ArticleGoogle Scholar
- Foley E, O'Farrell PH: Functional dissection of an innate immune response by a genome-wide RNAi screen. PLoS Biol. 2004, 2 (8): E203-10.1371/journal.pbio.0020203.PubMed CentralPubMedView ArticleGoogle Scholar
- Sanders HR, Foy BD, Evans AM, Ross LS, Beaty BJ, Olson KE, Gill SS: Sindbis virus induces transport processes and alters expression of innate immunity pathway genes in the midgut of the disease vector, Aedes aegypti. Insect Biochem Mol Biol. 2005, 35 (11): 1293-1307. 10.1016/j.ibmb.2005.07.006.PubMedView ArticleGoogle Scholar
- Bogachek MV, Protopopova EV, Loktev VB, Zaitsev BN, Favre M, Sekatskii SK, Dietler G: Immunochemical and single molecule force spectroscopy studies of specific interaction between the laminin binding protein and the West Nile virus surface glycoprotein E domain II. J Mol Recognit. 2008, 21 (1): 55-62. 10.1002/jmr.866.PubMedView ArticleGoogle Scholar
- Salama RH, Muramatsu H, Zou K, Inui T, Kimura T, Muramatsu T: Midkine binds to 37-kDa laminin binding protein precursor, leading to nuclear transport of the complex. Exp Cell Res. 2001, 270 (1): 13-20. 10.1006/excr.2001.5341.PubMedView ArticleGoogle Scholar
- Afshar N, Black BE, Paschal BM: Retrotranslocation of the chaperone calreticulin from the endoplasmic reticulum lumen to the cytosol. Mol Cell Biol. 2005, 25 (20): 8844-8853. 10.1128/MCB.25.20.8844-8853.2005.PubMed CentralPubMedView ArticleGoogle Scholar
- Petersen JM, Her LS, Varvel V, Lund E, Dahlberg JE: The matrix protein of vesicular stomatitis virus inhibits nucleocytoplasmic transport when it is in the nucleus and associated with nuclear pore complexes. Mol Cell Biol. 2000, 20 (22): 8590-8601. 10.1128/MCB.20.22.8590-8601.2000.PubMed CentralPubMedView ArticleGoogle Scholar
- Muller P, Kuttenkeuler D, Gesellchen V, Zeidler MP, Boutros M: Identification of JAK/STAT signalling components by genome-wide RNA interference. Nature. 2005, 436 (7052): 871-875. 10.1038/nature03869.PubMedView ArticleGoogle Scholar
- Dostert C, Jouanguy E, Irving P, Troxler L, Galiana-Arnoux D, Hetru C, Hoffmann JA, Imler JL: The Jak-STAT signaling pathway is required but not sufficient for the antiviral response of drosophila. Nat Immunol. 2005, 6 (9): 946-953. 10.1038/ni1237.PubMedView ArticleGoogle Scholar
- Nene V, Wortman JR, Lawson D, Haas B, Kodira C, Tu ZJ, Loftus B, Xi Z, Megy K, Grabherr M: Genome sequence of Aedes aegypti, a major arbovirus vector. Science. 2007, 316 (5832): 1718-1723. 10.1126/science.1138878.PubMedView ArticleGoogle Scholar
- Resnik P: Semantic similarity in a taxonomy: an information-based measure and its application to problems of ambiguity in natural language. J Artif Intel Res. 1999, 11: 95-130.Google Scholar
- Enright AJ, Van Dongen S, Ouzounis CA: An efficient algorithm for large-scale detection of protein families. Nucleic Acids Res. 2002, 30 (7): 1575-1584. 10.1093/nar/30.7.1575.PubMed CentralPubMedView ArticleGoogle Scholar
- Vlasblom J, Wu S, Pu S, Superina M, Liu G, Orsi C, Wodak SJ: GenePro: a Cytoscape plug-in for advanced visualization and analysis of interaction networks. Bioinformatics. 2006, 22 (17): 2178-2179. 10.1093/bioinformatics/btl356.PubMedView ArticleGoogle Scholar
- Maere S, Heymans K, Kuiper M: BiNGO: a Cytoscape plugin to assess overrepresentation of gene ontology categories in biological networks. Bioinformatics. 2005, 21 (16): 3448-3449. 10.1093/bioinformatics/bti551.PubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.