An update on mosquito cell expressed dengue virus receptor proteins


  • D. R. Smith

    Corresponding author
    1. Institute of Molecular Biosciences and Center for Emerging and Neglected Infectious Diseases, Mahidol University, Thailand
    Search for more papers by this author

Duncan R. Smith, Institute of Molecular Biosciences, Mahidol University, Salaya Campus, 25/25 Phuttamontol Sai 4, Salaya, Nakorn Pathom, 73170 Thailand. Tel.: + 662 800 3624-8; fax: + 662 441 9906; e-mail:


Dengue is the most important mosquito transmitted viral disease of humans worldwide. Despite intensive study over several decades, many of the fine details of the dengue virus (DENV) replication cycle remain unknown, although generally more is known about the phase of the replication cycle in mammalian cells as compared to the phase in mosquito cells. This results from a combination of less research emphasis on the mosquito stage, as well as fewer tools such as specific antibodies against mosquito proteins and insect informatics databases. The binding of a virus to a host cell is a first and critical stage in the infectious process and the mechanism and identity of cellular proteins involved in this process remains largely unknown. This short review aims to provide an update on our current understanding of the proteins expressed by mosquito cells that mediate DENV binding as a prerequisite to DENV entry and replication.


Infections with the dengue virus (DENV) represent a significant worldwide public health problem, particularly in tropical and subtropical countries (Gubler, 1998; Guzman & Kouri, 2002). Spread to humans by the bite of infected female mosquitoes belonging to the Aedes genus, DENV is believed to cause some 100 million infections each year resulting in approximately half a million cases of hospitalization (Rigau-Perez et al., 1998; Guzman & Kouri, 2002; Kyle & Harris, 2008). In mosquitoes the infection process starts when the vector take a bloodmeal from a viremic animal or human. DENV is maintained in nature in two primary transmission cycles, an urban cycle where humans are the reservoir and amplifying host and transmission is by the periodomestic mosquito species Aedes aegypti and Aedes albopictus, and a sylvatic cycle where nonhuman primates are the reservoir host and transmission is by arboreal, tree hole dwelling Aedes spp. (Weaver & Barrett, 2004). The urban transmission cycle is independent from the sylvatic cycle, and the epidemic/endemic urban DENVs show distinct evolutionary divergence from the sylvatic DENVs (Rico-Hesse, 1990).

Although the presence of DENV during a bloodmeal is a requirement for infection of a mosquito, in and of itself it is insufficient to ensure infection and susceptibility depends upon a number of both genetic and nongenetic factors. Nongenetic factors that have been implicated in vectorial capacity include mosquito size (Alto et al., 2008b), nutritional status and larval competition (Alto et al., 2008a), whereas genetic factors obviously include mosquito species, although the basis for this is far from clear. For example Culex quinquefasciatus is generally considered to be a nonsusceptible species (Huang et al., 1992), but reports of infection of this species with DENV under experimental conditions have been published (Vazeille-Falcoz et al., 1999). Similarly, significant variation exists in the susceptibility of Aedes mosquitoes to DENV infection. The African sylvatic Ae. aegypti formosus is relatively refractory to DENV infection (Black et al., 2002) whereas domestic Ae. aegypti aegypti can show distinct population variation in susceptibility to DENV infection (Bennett et al., 2002).

Overall vectorial competence is believed to be mediated by the presence of a number of genetically determined barriers to viral transmission, which include a midgut infection barrier, a midgut escape barrier and a salivary escape barrier (Black et al., 2002). The first step in infection of a mosquito as a result of ingestion of a viremic bloodmeal is infection of the midgut epithelial cells, and passage of the midgut infection barrier requires the virus to attach, penetrate and replicate in the midgut epithelial cells. Following the successful establishment of infection in the midgut epithelial cells, to pass the midgut escape barrier the infectious virions generated must pass through the basal lamina, disseminate through the haemocele and establish infection in secondary target organs. Passage of the last potential barrier (the salivary escape barrier) involves successful infection of the salivary glands and subsequent escape into the lumen of the salivary gland where it can be transmitted to the next vertebrate host (Black et al., 2002). Studies on the flaviviral vector competence of Aedes mosquitoes have suggested that the midgut infection barrier is the major determinant of transmission (Gubler et al., 1979; Bosio et al., 1998), and genetic loci associated with this barrier have been identified (Black et al., 2002; Bennett et al., 2005; Gorrochotegui-Escalante et al., 2005). Once infected, the vector remains infected for its whole life span, and evidence suggests that DENV can be transmitted vertically to the egg (Khin & Than, 1983; Rosen et al., 1983; Joshi et al., 2002; Gunther et al., 2007).

DENV consists of four closely related but antigenically distinct viruses termed dengue virus 1 (DENV-1) to dengue virus 4 (DENV-4), all of which cause a broadly similar disease spectrum in humans. Some evidence has suggested however that the specific DENV genotype may play a role in determining vectorial capacity in Ae. aegypti (Anderson & Rico-Hesse, 2006). Structurally, the mature DENV is an icosahedral, enveloped virus of approximately 50 nm and consists of three proteins, membrane (M), capsid (C) and envelope (E) and a single stranded, positive sense RNA genome of approximately 11 kb. The virus has a significant lipid component, and approximately 17% of the virion by weight is lipid (Russell et al., 1980), which forms a lipid bilayer between the nucleocapsid core and the E/M outer shell (Kuhn et al., 2002). The RNA genome, which has a 5′-methyl cap but no poly (A) tail encodes for the three structural proteins, with the membrane protein being encoded as a precursor form (preM) as well as seven nonstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5) that direct DENV replication.

In mosquito cells entry of the virus to a susceptible cell is believed to occur by receptor mediated endocytosis into clathrin coated pits (Mosso et al., 2008; Acosta et al., 2009) although some evidence has suggested that direct penetration of the virus into the cell may occur under some circumstances (Hase et al., 1989). The primary viral protein mediating the interaction between the mature virion and proteins expressed on the cell surface in both mosquito and mammalian cells is the E protein (Klasse et al., 1998). This 55 to 60 kDa glycoprotein protein contains 494 to 501 amino acids folded into three distinct structural and functional domains called domains I, II and III, with domain III being involved in the interaction of the virus with cellular receptors (Crill & Roehrig, 2001). The nature of the cellular proteins mediating the initial interaction between the virus and the cell remains poorly characterized. In mammalian cells, a number of proteins including DC-SIGN (Tassaneetrithep et al., 2003), the 37/67 kDa high affinity laminin receptor (Thepparit & Smith, 2004), glucose regulated protein 78 or GRP78 (Jindadamrongwech et al., 2004), heat shock proteins 70 and 90 (Reyes-del Valle et al., 2005), a cluster of differentiation 14 or CD-14 associated protein (Chen et al., 1999) and the mannose receptor (Miller et al., 2008) have all been proposed as dengue virus receptor proteins, with the implication that receptor usage is both cell and possibly serotype specific (Cabrera-Hernandez & Smith, 2005). In addition, considerable evidence has suggested that before the interaction of the virion with a high affinity receptor protein, there is a low affinity interaction of the virus with heparan sulphate that serves to concentrate the virus at the cell surface (Chen et al., 1997).

As noted earlier, the process mediating mosquito infection with DENV encompasses a number of different cell types, including midgut epithelial tissues, cells of secondary organs (such as brain, eye and several abdominal ganglia) and salivary gland cells (Linthicum et al., 1996). Each cell type infected will require, as an initial step in the infection process, an interaction with a receptor or receptors as a prerequisite to infection of the cell. Whether the receptors expressed on different cell types are the same or different remains to be established.

The role of heparan sulphate

Although the contribution of heparan sulphate to the internalization of DENV into mammalian cells has been well characterized (Hilgard & Stockert, 2000; Lin et al., 2000; Germi et al., 2002; Thepparit et al., 2004), the contribution of this molecule to internalization of DENV in mosquito cells is controversial. Although it had been proposed that heparan sulphate is not found on insect cells, contradictory evidence has suggested the presence of this sulphated glycosaminoglycan in insect cells and a role for heparan sulphate was observed in Drosophila melanogaster as a fibroblast growth factor ligand and/or ligand-receptor oligomerization protein (Lin et al., 1999). More recently heparan sulphate has been described in Anopheles stephensi midgut and salivary glands, where it is proposed to play a role in mediating infection and transmission of the Plasmodium parasite (Sinnis et al., 2007). However, specific experiments aimed at determining whether heparan sulphate plays a role in DENV internalization have been contradictory. Whereas some studies have shown the infection of mosquito cells with all four DENVs to be sensitive to heparan or heparinase treatment of the host cells (Sakoonwatanyoo et al., 2006), other studies have shown that this molecule does not play an essential role in dengue virus infection of mosquito cells (Hung et al., 2004; Thaisomboonsuk et al., 2005). Although the reason for the discrepancy amongst the studies is not obvious, an answer may lie in the cell culture adaptation of the virus, which enhances the binding and infectivity by a heparan sulphate-dependent mechanism as has been seen with other viruses (Klimstra et al., 1998; Kroschewski et al., 2003).

Candidate receptor proteins characterized by molecular weight

Several studies investigating DENV mosquito cell receptors have utilized the virus overlay protein binding assay (VOPBA). This methodology entails separating membrane proteins from a suitable source on sodium dodecyl sulphate polyacrylamide gels and transferring the proteins to a solid matrix support. The filters are then incubated with virus and the positions of proteins that bind the virus are detected through the use of specific antivirus antibodies or through the use of radiolabelled virus in the original hybridization. Although the methodology is commonly used it has several drawbacks, and in particular the identification of the binding protein can be problematic because of the number of proteins comigrating at the same molecular weight. Although mass spectroscopy can help in the identification of candidate proteins, the large number of candidate proteins from a single analysed band means that full characterization requires additional methodologies, and many of the published studies investigating DENV receptor proteins describe only a specific molecular weight, without further identification of the specific protein. Almost of the studies undertaken investigating the role of DENV receptor molecules have been undertaken using C6/36 cells, a cell line derived from whole Ae. albopictus larvae (Singh, 1967).

In the earliest study, Salas-Benito & del Angel (1997) identified two membrane proteins sized 40 and 45 kDa as DENV-4 binding molecules expressed on the surface of C6/36 cells (see Table 1) using 35S-labelled DENV-4 as the binding moiety. The interaction of DENV E protein with these two proteins was later confirmed by affinity column chromatography using recombinant dengue E protein as the ligand (Reyes-del Valle & del Angel, 2004). These proteins were shown to be glycoproteins, although the carbohydrate moieties did not apparently participate in the interaction. The specificity of the 45 kDa protein as a DENV receptor protein was further supported by the inhibition of infection in the presence of polyclonal antibodies raised against the 45 kDa protein, and by the wide distribution of this protein in Ae. aegypti tissues (Yazi Mendoza et al., 2002). It was further proposed that the absence of this protein in a malaria vector, Anopheles albimanus supported the contention that this protein was a DENV receptor protein (Yazi Mendoza et al., 2002). A subsequent study suggested that the 45 kDa protein may be immunologically related to heat shock protein (Hsp) 90 (Salas-Benito et al., 2007).

Table 1.  Characterized and partly characterized dengue virus (DENV) receptor proteins expressed by insect cells
Cell line/mosquitoCell typeSerotype; strainReceptor characteristicsReference
C6/36Aedes albopictus cell lineDENV-440 and 45 kDa glycoproteinsSalas-Benito & del Angel, 1997
C6/36Ae. albopictus cell lineDENV-280 and 67 kDa proteinsMunoz et al., 1998
C6/36Ae. albopictus cell lineDENV-2Tubulin protein or like-tubulin proteinChee & AbuBakar, 2004
C6/36Ae. albopictus cell lineDENV-3, 4Laminin-binding proteinSakoonwatanyoo et al., 2006
C6/36Ae. albopictus cell lineDENV-2Heat shock related proteinSalas-Benito et al., 2007
C6/36Ae. albopictus cell lineDENV-2ProhibitinKuadkitkan et al., 2010
CCL-125Aedes aegypti cell lineDENV-2ProhibitinKuadkitkan et al., 2010
Ae. aegyptiMidgut, ovary and salivary gland cell, eggs, larvae and pupae cell extractDENV-445 kDa glycoproteinYazi Mendoza et al., 2002
Ae. aegyptiSalivary glandDENV-1, 2, 3 and 437, 54, 58 and 77 kDa proteinsCao-Lormeau, 2009
Aedes polynesiensisSalivary glandDENV-1, 448, 50, 54, 56 and 67 kDa proteinsCao-Lormeau, 2009

Using broadly the same methodology Munoz et al. (1998) identified two proteins of 67 and 80 kDa on C6/36 cells as putative DENV-2 receptor proteins. Polyclonal antibodies raised against C6/36 membrane proteins both inhibited binding of DENV-2 to C6/36 cells, as well as specifically to the 67 kDa protein in a VOPBA analysis. In later work it was suggested that these two proteins act as receptor proteins for all four DENVs (Mercado-Curiel et al., 2006), and these authors were able to purify the two proteins from Ae. aegypti midgut. In a subsequent study, utilizing three different Ae. aegypti strains (DS3, which is susceptible to DENV infection, IBO-11, which is refractory to infection and DMEB, which is an escape barrier strain that only becomes infected in midgut epithelial cells), the 67 kDa protein was reported to be a marker of vector competence for DENV with the 67 kDa protein being primarily responsible for mediating DENV infection of midgut cells (Mercado-Curiel et al., 2008). This work would however seem to be directly contradicted by a recent study that showed that vector competence is not dependent upon mosquito midgut binding affinity (Cox et al., 2011).

Studies by both the Munoz (Munoz et al., 1998) and del Angel (Salas-Benito & del Angel, 1997) groups showed no effect of neuraminidase treatment of C6/36 cells on virus binding, showing that sialic acid, a common virus binding element expressed by mammalian cells has no role in the binding or entry of DENV into C6/36 cells.

A comprehensive analysis investigating the binding of all four DENVs to C6/36 cells using the VOPBA methodology showed a possible serotype specific component to binding of DENV to mosquito cell proteins (Sakoonwatanyoo et al., 2006). This study identified a common band of approximately 50 kDa that bound DENV-2, DENV-3 and DENV-4, and a second band of 100 kDa that bound DENV-4, but no prominent band was seen to bind DENV-1. Investigation of the 50 kDa band, based on a similar molecular size to the previously identified 37/67 kDa high affinity laminin receptor protein used by DENV-1 to enter human liver cells (Thepparit & Smith, 2004), suggested that the 50 kDa protein may be a laminin binding homologue that may function in the internalization of DENV-3 and DENV-4 to C6/36 cells. Other studies have suggested that laminin binding proteins may function as receptor proteins mediating the entry of Venezuelan equine encephalitis virus (Ludwig et al., 1996) and Japanese encephalitis virus (Boonsanay & Smith, 2007) to C6/36 cells.

In a recent study Cao-Lormeau (2009) investigated the binding of all four DENVs to salivary gland extracts from two mosquito species, namely Ae. aegypti and Aedes polynesiensis. Results showed that four proteins of 37, 54, 58 and 77 kDa from Ae. aegypti bound all four DENVs, whereas five proteins of 48, 50, 54, 56 and 67 kDa from Ae. polynesiensis bound DENV-1 and DENV-4 (Cao-Lormeau, 2009).

Characterized receptor molecules

The first identified molecule that was purported to have a function in the internalization of DENV to C6/36 cells was identified by Chee & AbuBakar (2004). Through the use of VOPBA, Chee & AbuBakar identified a 48 kDa protein that specifically bound DENV-2, and the authors further showed that the binding was abolished by the presence of free unbound protein. The identity of the protein was determined by undertaking matrix assisted laser desorption/ionization-time of flight (MALDI-ToF) mass spectrometry and liquid chromatography mass spectrometry/mass spectrometry (LC-MS/MS) and analysing the subsequently generated peptide masses. The analysed peptide masses were consistent with either alpha-tubulin or beta-tubulin from a number of species, leading Chee & AbuBakar to propose that the 48 kDa binding protein was tubulin or a tubulin-like protein that was involved in directly binding DENV-2 to C6/36 cells (Chee & AbuBakar, 2004). However, Chee &AbuBakar did not observe binding of DENV-2 to membrane fractions of C6/36 cells, and only observed binding in cytosolic fractions. It seems unlikely therefore that this protein plays a role in the initial direct interaction of DENV with C6/36 cells, but it may play a role in either the later stages of internalization or trafficking of the virus particle.

The study of Chee & AbuBakar (2004) highlights the significant problem in using VOPBA to isolate DENV receptor proteins as VOPBA does not distinguish between binding proteins and receptor proteins. Many proteins expressed on the cell surface, or closely associated with cellular membranes may have the ability to bind DENV, without necessarily playing a role in the internalization of the virus to the cell, and this is seen by the multiple bands frequently seen in VOPBA analysis. As such, the ability to discriminate binding proteins (either specific or nonspecific) and bona fide receptor proteins is a key element in the identification of receptor proteins. To address this issue the most recent published study (Kuadkitkan et al., 2010) utilized mosquitoes from a susceptible (Ae. aegypti) and a nonsusceptible species (C. quinquefasciatus) as well as two different cell lines derived from Aedes mosquitoes, namely the Ae. albopictus derived C6/36 cell line used in the majority of earlier studies as well as the Ae. aegypti derived CCL-125 cell line, which had recently been shown to be susceptible to DENV infection (Wikan et al., 2009) in contrast to the original reports on this cell line, which proposed that the cell line was not susceptible to DENV (Singh, 1967, 1971; Singh & Paul, 1968). In common with other reports, several bands were observed on one-dimensional VOPBA analysis, including bands seen in studies by other groups. However, one band was clearly seen to segregate with susceptibility to infection. This protein was identified through multiple methodologies to be prohibitin, a ubiquitously expressed protein in eukaryotic cells. The interaction between prohibitin and DENV-2 was shown by co-immunoprecipitation, and the role of prohibitin in mediating entry of the virus to mosquito cells confirmed by antibody mediated inhibition of infection and small interfering RNA (siRNA) mediated down-regulation of prohibitin as well as colocalization of the virus and protein on the cell surface. Collectively the data showed an unambiguous role for prohibitin in mediating DENV-2 internalization to mosquito cells (Kuadkitkan et al., 2010). Interestingly, prohibitin was shown to be specific for DENV-2, and did not participate in the internalization of Japanese encephalitis virus to these cells. Prohibitin was additionally shown to be significantly colocalized with DENV-2 E protein inside the cell, suggesting that the prohibitin–DENV-2 interaction may be a multifunctional interaction occurring at several stages during the virus replication cycle (Kuadkitkan et al., 2010).

Independent verification of the interaction between prohibitin and DENV E protein was provided by the study of Paingankar et al. (2010), who used VOPBA to identify DENV binding proteins also using two Aedes cell lines (the Ae. albopictus C6/36 and the Ae. aegypti A7 cell lines) as well as the midgut brush border membrane fraction of Ae. aegypti mosquitoes. In total these authors identified seven DENV binding proteins (actin, ATP synthase beta subunit, heat shock cognate 70, orisis, tubulin beta chain, vav-1 and prohibitin) and proposed a model for DENV-2 entry and transport. In the model the authors propose that the initial interaction between DENV and the cell occurs through a nonprotein-protein interaction with laminin, lectin, heparan sulphate or a similar molecule and that subsequent cellular changes allow an interaction with tubulin that mediates the internalization of the virus (Paingankar et al., 2010). It is proposed that Hsp70, possibly in complex with an ATP synthase, serves to concentrate the virus particle at the membrane and based largely on a relatively modest reduction of levels of prohibitin during infection, the authors propose that the interaction between prohibitin and DENV-2 is a nonreceptor interaction that serves to activate a signal transduction pathway that results in reducing DENV-2 infection (Paingankar et al., 2010). The interactions between actin, vav-1 and tubulin (in a second round of interactions) are proposed to mediate virus transport, possibly facilitating virus exit (Paingankar et al., 2010). Although the model serves to account for the interactions detected between DENV-2 and the identified proteins, the now well-characterized receptor interaction between DENV-2 and prohibitin (Kuadkitkan et al., 2010) serves to cast some doubts on the validity of the model.


There is little doubt that progress in identifying DENV receptor proteins is proceeding more slowly in mosquitoes and mosquito cells than is the case with mammalian cells. To date, prohibitin remains the only clearly characterized and identified mosquito cell expressed DENV receptor protein (Kuadkitkan et al., 2010). However, the interaction has only been shown for DENV-2 and as such a more comprehensive analysis with different serotypes needs to be undertaken. More critically, the functional evaluations of insect receptor proteins undertaken to date have been undertaken in cell lines that are of an indeterminate tissue origin. The commonly used C6/36 and CCL-125 cell lines are derived from whole hatched larvae of Ae. albopictus and Ae. aegypti, respectively (Singh, 1967) and both the immortalized nature of these cell lines, as well as the unknown tissue origins serve to reduce significantly confidence that the events recapitulated in these systems reflect the process of infection in vivo.

As already noted, vectorial capacity has been linked to three barriers, the midgut infection barrier, the midgut escape barrier and the salivary escape barrier. The first two of these barriers, which are essentially involved with the infection of midgut epithelial cells and secondary organs, could be associated with the interaction of DENV with specific receptor proteins. Currently the evidence linking the midgut infection barrier to the expression of specific receptor proteins is contradictory (Mercado-Curiel et al., 2008; Cox et al., 2011). However, initial studies identifying specific quantitative trait loci associated with the infection barriers (Black et al., 2002; Bennett et al., 2005; Gorrochotegui-Escalante et al., 2005) will hopefully result in a greater understanding of the mechanism of action of the barriers.

To date studies investigating insect expressed DENV receptor proteins have suffered from two major deficiencies – an overreliance on VOPBA coupled with a lack of fine discrimination between DENV binding proteins and DENV receptor proteins. In particular DENV, and especially the DENV E protein, may interact with numerous proteins at different stages of the replication cycle and only a small subset of these interactions may occur at the initial binding and internalization steps. The increased use of whole mosquitoes of differing susceptibility to infection and the application of directed methodologies to modulate or ablate specific gene products in Ae. aegypti or other vector species through the use of transgenic or siRNA methodologies will serve to improve significantly our understanding of the mechanics of the DENV : insect receptor interaction.


The author gratefully acknowledges support from the Office of the Higher Education Commission and Mahidol University under the National Research Universities Initiative, the Thailand Research Fund, the National Science and Technology Development Agency and Mahidol University.