The resurgence of malaria is at least partly attributed to the absence of an effective vaccine, parasite resistance to antimalarial drugs and resistance to insecticides of the anopheline mosquito vectors. Novel strategies are needed to combat the disease on three fronts: protection (vaccines),prophylaxis/treatment (antimalarial drugs) and transmission blocking. The latter entails either killing the mosquitoes (insecticides), preventing mosquito biting (bednets and repellents), blocking parasite development in the vector (transmission blocking vaccines), genetic manipulation or chemical incapacitation of the vector. During the past decade, mosquito research has been energized by several breakthroughs, including the successful transformation of anopheline vectors, analysis of gene function by RNAi,genome-wide expression profiling using DNA microarrays and, most importantly,sequencing of the Anopheles gambiae genome. These breakthroughs helped unravel some of the mechanisms underlying the dynamic interactions between the parasite and the vector and shed light on the mosquito innate immune system as a set of potential targets to block parasite development. In this context, putative pattern recognition receptors of the mosquito that act as positive and negative regulators of parasite development have been identified recently. Characterizing these molecules and others of similar function, and identifying their ligands on the parasite surface, will provide clues on the nature of the interactions that define an efficient parasite–vector system and open up unprecedented opportunities to control the vectorial capacity of anopheline mosquitoes.
Innate immunity in vertebrates plays a fundamental role in pathogen recognition and subsequent activation of the adaptive immune response, which is characterized by a highly diverse, somatically generated repertoire of antigen receptors. The interplay between adaptive and innate immunity orchestrates the defensive responses against a variety of invaders such as viruses, bacteria, fungi or parasites(Medzhitov and Janeway,1997a). Invertebrates, including insects, lack an adaptive immune system but utilize innate immunity for their defence. The successful evolution of insects as the most species-rich group of terrestrial animals testifies to the efficiency and flexibility of innate immunity in dealing with a diverse array of pathogens.
In insects, the peritrophic membrane (PM) of the midgut, the cuticle of the exoskeleton and the lining of the tracheal respiratory system constitute physical barriers against invaders. The PM is a sleeve-like extracellular layer surrounding the food bolus and is composed of chitin, proteins and proteoglycans (Wang and Granados,2001). Ingested ookinetes of the malaria parasite Plasmodium must penetrate the midgut wall of the Anophelesmosquito vector to develop further into oocysts on the haemocoel basal side of the midgut epithelium (Fig. 1). This process is facilitated by the secretion of chitinases that disrupt the PM chitin (Huber et al., 1991; Langer et al., 2000). Midgut invasion was completely blocked when mosquitoes were fed on infected blood containing the chitinase inhibitor allosamidin(Shahabuddin et al., 1993). Similarly, knockout of the Plasmodium falciparum chitinase gene induced a marked reduction in the number of developing oocysts in the midguts of Anopheles freeborni mosquitoes(Tsai et al., 2001). Microorganisms that successfully overcome any of these physical barriers encounter an array of host innate immune responses within the underlying epithelia or systemically in internal immune tissues such as haemocytes(insect blood cells) and the fat body (liver analogue), which are known to release immune effectors into the open circulatory system, the haemolymph (the insect blood). Examples of such responses are phagocytosis, secretion of antimicrobial peptides, nodule formation, agglutination, encapsulation and melanization.
Studies on Drosophila innate immunity have had a major impact on this field, both in vertebrates and invertebrates, leading to key discoveries and fundamental concepts on how organisms efficiently fight pathogens even in the absence of the clonal system of recognition that is central to adaptive immunity (Hoffmann, 2003; Hoffmann and Reichhart, 2002). What made Drosophila tractable for such studies was the powerful genetic and molecular genetic tools available in the fruitfly, including a fully sequenced genome. The mosquito Anopheles gambiae, the major vector of malaria in Africa, also became a suitable model for the study of innate immunity recently, when its genome was sequenced(Holt et al., 2002),accompanied by comparative genomic analysis with Drosophila(Zdobnov et al., 2002) and the establishment of powerful tools for gene discovery(Dimopoulos et al., 2002),functional analysis (Blandin et al.,2002; Levashina et al.,2001) and transgenesis(Grossman et al., 2001). As the most important vector for transmission of the malaria parasite, A. gambiae offers the advantage of assessing immune reactions in a species of major importance to human health. A. gambiae mounts efficient local and systemic immune responses against Plasmodium infection(Dimopoulos et al., 1997, 1998; Richman et al., 1997; Fig. 1). However, in spite of the hostile environment encountered and major losses in parasite numbers inside the vector, some parasite species or strains successfully complete their sexual life cycle, indicating that they have found ways to escape or subvert to some extent the vector immune responses. Other vector–parasite combinations are either poor (Plasmodium gallinaceum/Anopheles stephensi) or incompatible (Plasmodium berghei/Aedes aegypti), suggesting that key molecular and cellular interactions are a prerequisite for a vector–parasite system to become established and subsequently co-evolve(Alavi et al., 2003). In this review, we describe our current knowledge of A. gambiae innate immunity and its impact on Plasmodium development.
Recognition of non-self
Immune reactions are initiated when molecules of microbial origin are detected and recognized as `non-self'. This recognition step involves pattern recognition receptors (PRRs) that recognize and bind to so-called pathogen-associated molecular patterns (PAMPs) that are shared by various microorganisms but absent from eukaryotic cells (Medzhitov and Janeway, 1997b, 2002). PAMPs that are known to be potent immune elicitors include lipopolysaccharides (LPS), peptidoglycan(PGN) and β-1,3-glucans. Various PRRs have been identified and isolated from vertebrates (Holmskov et al.,2003; Takeda et al.,2003) and invertebrates(Gobert et al., 2003; Hoffmann, 2003; Wilson et al., 1999; Yu et al., 2002). The best-studied invertebrate PRRs are the peptidoglycan recognition proteins(PGRPs) and the Gram-negative bacteria-binding proteins (GNBPs). PGRPs are soluble or transmembrane proteins containing a domain similar to the bacterial amidase domain, which is involved in recycling bacterial cell wall fragments. PGRPs have been isolated from both invertebrates and vertebrates(Dziarski et al., 2003; Wang et al., 2003). The first was isolated from the haemolymph of the moth Bombyx mori, where it is involved in activating the prophenoloxidase (PPO) cascade(Yoshida et al., 1996). In Drosophila, three PGRPs were identified as bona fide PRRs. The soluble PGRP-SA activates the Toll pathway in response to Gram-positive bacterial infection (Gobert et al.,2003; Michel et al.,2001) in concert with another PRR, GNBP1. By contrast, PGRP-LC(Choe et al., 2002; Gottar et al., 2002; Ramet et al., 2002) and PGRP-LE (Takehana et al.,2002) are involved in activating the immune deficiency (IMD)pathway in response to Gram-negative bacterial infections. Among the seven identified putative Anopheles PGRPs(Christophides et al., 2002),PGRPLC seems to play a central role in defence against bacterial infection (G. K. Christophides, unpublished data). The orthologous genes in both Drosophila (Werner et al.,2003) and Anopheles(Christophides et al., 2002)undergo alternative splicing resulting in at least three distinct isoforms. Interestingly, the Anopheles PGRPLC isoforms are differentially regulated upon immune challenge, suggesting that splicing can be regulated by the immune signal (Christophides et al.,2002). On the other hand, PGRPLB is transcriptionally upregulated following Plasmodium infection of adult mosquitoes(Dimopoulos et al., 2002) and maintains an elevated expression throughout the parasite's entire life cycle(Christophides et al.,2002).
GNBPs were first described in Bombyx mori(Lee et al., 1996) and share significant sequence similarity with the catalytic region of bacterialβ-1,3- and β-1,3,1,4-glucanases. BmGNBP binds strongly to the surface of Gram-negative bacteria and shows pronounced transcriptional upregulation following bacterial challenge(Lee et al., 1996). The Drosophila GNBP1 binds with high affinity to LPS andβ-1,3-glucan (Kim et al.,2000) and, in concert with PGRP-SA, activates the Toll pathway upon infection with Gram-positive bacteria(Gobert et al., 2003). In A. gambiae, six putative GNBPs have been identified(Christophides et al., 2002). Among them, GNBPB1 and GNBPA1 are upregulated following Plasmodium infection, while only GNBPB1 is responsive to bacteria (Christophides et al.,2002; Dimopoulos et al.,2002). Other putative Anopheles PRRs include the thioester-containing proteins (TEPs), leucine-rich immune proteins (LRIMs) and C-type lectins (CTLs). Members of these protein families were recently implicated in the regulation of Plasmodium development in the mosquito vector and are discussed later in this review.
Signal modulation and transduction
Serine protease cascades
Recognition of non-self usually activates a proteolytic cascade of serine proteases that amplify the signal and trigger downstream effector responses,leading to the killing of the invader. Key components of such cascades are clip-domain serine proteases (CLIPs), which have been implicated in several defence mechanisms in insects and crustaceans, such as the activation of signalling pathways leading to the synthesis of antimicrobial peptides (AMPs)(Ligoxygakis et al., 2002a),haemolymph agglutination (Kawabata et al.,1996) and melanization (Kanost et al., 2001), discussed also in a later section. The role of the clip-domain is not yet known but is believed to include regulation of enzymatic activity or protein localization. In Drosophila, some CLIPs are key components of the dorsoventral pattern formation system (e.g. Easter and Snake), but others, such as Persephone, are also involved in the activation of the Toll pathway upon fungal infection(Ligoxygakis et al., 2002a). The horseshoe crab clotting system involves serine protease zymogens that are activated in a cascade manner, leading to the transformation of coagulogen to insoluble coagulum gel and subsequent clotting of the haemolymph (Miyata et al., 1984a,b; Nakamura et al., 1986). The clot is effective in immobilizing pathogens, which are then eliminated by other host effector mechanisms such as secretion of AMPs.
Melanization requires the enzymatic processing of inactive PPO to active phenoloxidases (PO) by activating serine proteases, referred to in the literature as prophenol-activating proteinase (PAP) or prophenoloxidase-activating enzyme (PPAE). In the tobacco hornworm Manduca sexta, several PAPs have been described and shown biochemically to be involved in PPO activation: PAP1 has one clip-domain(Jiang et al., 1998) while PAP2 (Jiang et al., 2003a) and PAP3 (Jiang et al., 2003b)contain two clip-domains. Interestingly, the Manduca PAPs require non-proteolytic serine protease homologues as cofactors for the efficient activation of PPO (Jiang et al., 2003a,b; Yu et al., 2003). Two A. gambiae CLIPs (CLIPB14 and CLIPB15) were found to be responsive to bacterial and Plasmodium infections: CLIPB14 showed persistent upregulation in Plasmodium-infected mosquitoes while CLIPB15 showed only transient upregulation, precisely during midgut invasion(Christophides et al., 2002). The role of these two genes in the activation of PPO has been demonstrated recently (J. Volz, unpublished data).
Signal amplification by the serine protease cascade is under tight regulation by serpins (serine protease inhibitors), which inhibit serine proteases by acting as irreversible suicide substrates that covalently bind to the active centre of the enzyme. The Drosophila serpin Spn43Ac(Necrotic) downregulates the Toll pathway in response to fungal infections by inhibiting cleavage of the Toll receptor ligand, the cytokine-like polypeptide Spaetzle (Levashina et al.,1999). The involvement of serpins in the regulation of melanization has been described in D. melanogaster(De Gregorio et al., 2002; Ligoxygakis et al., 2002b), M. sexta (Zhu et al.,2003) and A. gambiae (K. Michel, unpublished data). A recently characterized Anopheles serpin, SRPN10, encodes four alternatively spliced inhibitory isoforms(Danielli et al., 2003). Interestingly, two of these forms are specifically upregulated in female mosquitoes in response to midgut invasion by P. berghei ookinetes,making SRPN10 an excellent cell-autonomous marker of invasion (A. Danielli, T. G. Loukeris and C. Barillas-Mury, unpublished data).
Immune signalling pathways
Immune signalling pathways transmit the signal originating from PAMP-associated PRRs to the effector genes. In a now classical series of studies, two signal transduction pathways, the Toll and IMD pathways(Fig. 2), have been identified in Drosophila and linked to innate immunity(Hoffmann, 2003). Fungal or Gram-positive bacterial infections activate the Toll pathway by inducing the proteolytic cleavage of Spaetzle, which binds directly to and activates the transmembrane receptor Toll (Weber et al., 2003). Hence, Toll does not appear to be a direct sensor of microbial compounds, unlike the mammalian Toll-like receptors (TLRs) that recognize and bind to several microbial ligands(Takeda et al., 2003). Toll has an intracytoplasmic TIR (Toll, IL-1R) domain, which upon activation recruits three death domain proteins, MyD88, Tube and Pelle, the first two of which are considered as adaptor proteins. The Toll receptor–adaptor complex signals to two cytoplasmic Rel/NFκB transcription factors,Dorsal and Dif, causing their dissociation from Cactus, an ankyrin repeat inhibitory protein, and subsequent translocation to the nucleus where they activate the transcription of AMPs. Whereas Dif has a crucial role in the immune response (Ip et al.,1993; Petersen et al.,1995), Dorsal is believed to be mainly engaged in transcriptional activation of genes involved in the dorsoventral patterning(Morisato and Anderson, 1995). Interestingly, no orthologue of Dif was found in the Anophelesgenome. Gambif1 (now called REL1), the mosquito orthologue of Dorsal, has been previously characterized and shown to translocate to the nucleus following bacterial but not Plasmodium infection(Barillas-Mury et al., 1996). Ten Toll and six Spaetzle-like proteins have been identified in the genome of A. gambiae, but their phylogenetic relationships with the respective Drosophila homologues are unclear(Christophides et al., 2002). However, identification of the mosquito orthologues of MyD88, Tube and Pelle indicates that the Toll pathway in the mosquito is at least partially conserved (Christophides et al.,2002).
The second immune signalling pathway in Drosophila, the IMD pathway, is activated following Gram-negative bacterial infection leading to the cleavage of another Rel/NFκB family protein, Relish, through the recently proposed proteolytic action of the caspase Dredd(Stoven et al., 2003); this causes the release of the Rel-homology domain of Relish from its inhibitory carboxy-terminal ankyrin domain. The transmembrane receptor of the IMD pathway is as yet unclear; however, a number of studies point to a role for PGRP-LC in this process (Choe et al.,2002; Gottar et al.,2002). Intracellular activation of the pathway commences with recruitment of IMD, a death domain protein sharing similarities with the mammalian TNF-α receptor interacting protein, RIP. The sequence of events between IMD activation and the cleavage of Relish is not fully characterized; however, genetic data point to several other factors downstream of IMD, including the protein kinase dTAK1, another death-domain protein,dFADD, and the Drosophila homologues of the mammalian signalosome equivalent comprising IKK-β and IKK-γ. Interestingly, all the aforementioned components of the IMD pathway are conserved in Anopheles (Christophides et al.,2002), and preliminary data indicate that REL2, the Anopheles orthologue of Relish, is indeed involved in anti-bacterial defence (G. K. Christophides, unpublished data). The Ae. aegyptiRelish gene has three alternatively spliced transcripts encoding three different proteins: Relish, IκB-type, which lacks the Rel-homology domain, and the Rel-type in which the amino-terminal transactivation domain and the carboxy-terminal ankyrin repeats are missing(Shin et al., 2002). The involvement of Aedes Relish in the regulation of immune response to bacterial challenge has been shown using transgenic mosquitoes carrying the Rel-type transgene driven by the bloodmeal-inducible vitellogenin promoter(Shin et al., 2003a). In these mosquitoes, the overexpression of Rel-type transgene following a blood meal resulted in severely reduced expression levels of defensin and cecropin genes and a strong susceptibility to Gram-negative bacterial infections. The interference of Rel-type protein with endogenous Relish was suggested to involve competitive binding to the κB motif.
Little is known about the role of the JNK and JAK/STAT pathways in antimicrobial defence in insects. The Drosophila JAK/STAT pathway has one STAT component and is involved in many developmental processes(Luo and Dearolf, 2001). A recent study based on microarray analysis on Drosophila cell lines revealed that the IMD pathway branches downstream of dTAK1: one branch controlling the synthesis of antimicrobial peptides through Relish, and the other the synthesis of cytoskeletal proteins through the JAK/STAT(Boutros et al., 2002). Based on these data, a close link between cytoskeletal remodelling and antimicrobial defence was suggested (Boutros et al.,2002). Two members of the STAT family (STAT1 and STAT2) have been identified in the mosquito genome(Christophides et al., 2002). The observed STAT1 (previously called AgSTAT) translocation into the nucleus of mosquito fat body cells following bacterial infection has provided the first evidence for the involvement of insect STATs in immune defence(Barillas-Mury et al.,1999).
Antimicrobial peptides (AMPs)
Immune responses in insects can be divided into humoral and cellular. The synthesis of AMPs is the final step of inducible humoral immune responses. Following pathogen recognition, AMPs are rapidly produced by the fat body and secreted into the haemolymph, where they accumulate in large concentrations(Meister et al., 1997). AMPs are also locally produced by barrier epithelia of several insect species such as B. mori (Brey et al.,1993), A. gambiae(Brey et al., 1993; Dimopoulos et al., 2002; Richman et al., 1997) and D. melanogaster (Ferrandon et al., 1998; Tingvall et al.,2001). To date, several hundred AMPs have been described in immune-challenged insects, where they exhibit wide and complementary spectra of activity against various microorganisms(Bulet et al., 1999). In Drosophila, the AMP-encoding genes are regulated by the finely tuned activity of the IMD and Toll pathways(Bulet et al., 1999; Hoffmann, 2003; Hoffmann and Reichhart, 2002). IMD controls the expression of anti-Gram-negative peptides such as diptericins and drosocins, while Toll induces the expression of the anti-fungal peptide drosomycin by the fat body cells. Interestingly, the local expression of drosomycin by barrier epithelia is Toll-independent, suggesting that local and systemic expressions of AMPs might be regulated by different signalling pathways (Ferrandon et al.,1998). Signals from both Toll and IMD seem to control jointly the induction of cecropins, attacins and defensins in Drosophila(Hoffmann et al., 1996; Meister et al., 1997). In Ae. aegypti, defensin levels were shown to be strongly upregulated following bacterial challenge, reaching a concentration of approximately 45μmol l–1 in the haemolymph at 24 h post-inoculation(Lowenberger et al., 1999b). Another AMP, cecropin A, was isolated from the haemolymph of bacteria-challenged adult Ae. aegypti mosquitoes. The protein was active against a broad spectrum of Gram-negative bacteria but less so against Gram-positive bacteria and fungi(Lowenberger et al., 1999a). The Anopheles genome encodes four defensin genes (DEFs), four cecropins (CECs), one attacin and one gambicin (GAM1)(Christophides et al., 2002). The anti-Gram-positive effect of Anopheles DEF1 (previously called defensin A) has been demonstrated in vitro recently(Vizioli et al., 2001b). GAM1 is active against both Gram-positive and Gram-negative bacteria but is only marginally active against P. berghei ookinetes(Vizioli et al., 2001a).
Phagocytosis is a hallmark of the classical cellular immune responses of insects whereby haemocytes engulf target pathogens but also apoptotic bodies. Three types of haemocytes have been characterized in Drosophila: the plasmatocytes that are responsible for the disposal of microorganisms and apoptotic cells, the lamellocytes that encapsulate large invaders, and the crystal cells that are involved in melanization(Meister and Lagueux, 2003). In the yellow fever mosquito, Ae. aegypti, four different types of haemocytes have been distinguished: the granulocytes, the oenocytoids, the adipohaemocytes and the thrombocytoids(Hillyer and Christensen,2002). Phagocytosis in Drosophila is usually receptor mediated, involving either soluble or membrane-bound PRRs, including PGRP-LC,which is involved in the phagocytosis of Gram-negative bacteria(Ramet et al., 2002), and Croquemort, a membrane-bound receptor mediating phagocytosis of apoptotic corpses (Franc et al., 1996). A soluble thioester-containing protein, TEP1, identified in A. gambiae, acts as a complement-like opsonin promoting the phagocytosis of Gram-negative bacteria in a mosquito haemocyte-like cell line(Levashina et al., 2001). TEP1 activation seems to result from its infection-inducible proteolytic cleavage and exposure of the highly reactive thioester bond, apparently involved in a nucleophilic attack leading to the covalent binding of TEP1 to the surface of microorganisms.
The mechanisms promoting cellular encapsulation in insects are not well understood, and only a few examples of this innate cellular reaction have been reported. Encapsulation is a process by which insect lamellocytes form a multilayered capsule around large invaders such as parasitoids in the haemocoel, resulting in their isolation, immobilization and subsequent killing by asphyxiation, oxidation or melanization(Gotz, 1986). Drosophila larvae encapsulate and then melanize the eggs of a parasitoid wasp in their haemocoel through the concerted action of lamellocytes and crystal cells (Lanot et al., 2001; Sorrentino et al.,2002). Hemese, a recently identified transmembrane receptor, is expressed on the surface of all Drosophila haemocytes and acts as a negative regulator of the encapsulation response(Kurucz et al., 2003). Knockout of Hemese stimulates the proliferation of lamellocytes following parasitoid infection, resulting in an enhanced cellular response(Kurucz et al., 2003). The mechanisms through which lamellocytes are alerted to the presence of a large invader and the signals involved in their proliferation remain unknown.
Melanization is a prime humoral immune reaction of insects, being involved in wound healing and sequestration of invaders in a dense melanin coat. In contrast to encapsulation, it does not require the direct involvement of haemocytes (Soderhall and Cerenius,1998). Melanization requires the proteolytic activation of the inactive PPO zymogens to the active POs, a step tightly regulated, as described above, by the balanced action of CLIPs and their inhibitors,serpins. POs oxidize phenolic substances such as tyrosine, DOPA and dopamine to melanine and serve several tasks including wound healing, cuticle pigmentation and sclerotization, and melanization of invading pathogens(Soderhall and Cerenius,1998). PPOs are produced by haemocytes and released into the haemolymph (Ashida, 1971; Durrant et al., 1993; Muller et al., 1999), from where they can also be transported to the cuticle through the underlying cuticular epithelium to facilitate defence against microbial invasion or abrasion of the cuticle (Asano and Ashida,2001; Ashida and Brey,1995). There are nine PPO-encoding genes (PPO1–9)in A. gambiae (Christophides et al., 2002) that show overlapping developmental expression profiles(Jiang et al., 1997; Lee et al., 1998; Muller et al., 1999). PPO5 and PPO6 are mainly expressed in adult mosquitoes,whereas PPO1–4 are expressed in pre-adult stages. Some genes such as PPO2, PPO3 and PPO9 are induced following blood feeding (H. M. Muller, unpublished data).
A. gambiae: an emerging model for the study of innate immunity
Anopheles immune responses
A. gambiae is the major African vector for transmission of malaria, a disease killing approximately two million people every year,largely children between the age of one and five, in sub-Saharan Africa(WHO, 1999). The present alarming prevalence of malaria is attributed, at least in part, to the absence of a protective vaccine (Richie and Saul,2002) and the rapid spread of drug-resistant parasites(Olliaro, 2001) and insecticide-resistant mosquitoes. Novel strategies to control malaria are badly needed. This has generated a growing interest in understanding the complex interactions between the mosquito vector and the Plasmodiumparasite (Varmus et al.,2003). Of particular interest are the vector immune responses,which are associated with parasite killing. Initial attempts monitoring the immune responses of A. gambiae to infections with bacteria and P. berghei, a model rodent malaria parasite, were based on the use of a small number of mRNA immune markers, isolated by differential display techniques or homology cloning (Dimopoulos et al., 1997, 1998; Richman et al., 1997). These studies clearly showed that parasite invasion of the midgut (by ookinetes) and salivary gland epithelium (by sporozoites) induces the upregulation of several immune markers locally but also systemically in the abdomen. The abdominal response most probably involves several organs including fat body and/or haemocytes. This has suggested local immune responses in the infected epithelium and an immune-related signalling process propagated to other tissues (Fig. 1). Whether this process involves parasite-derived diffusible products or specific mosquito factors is currently unknown. Interestingly, it has been reported that midgut invasion by P. falciparum, which is accomplished by fewer ookinetes,does not result in upregulation of the same immune markers as does invasion by P. berghei (Tahar et al.,2002). Expressed sequence tag (EST) libraries and cDNA microarrays(Dimopoulos et al., 2000) were utilized later to analyze on a genomic scale the expression profile of Anopheles genes in response to parasite and bacterial infections. A pilot study based on a 4000 EST chip revealed that Gram-negative and Gram-positive bacterial challenges upregulate overlapping sets of genes, many of which belong to the immunity class, and that the response to Plasmodium partially overlaps with this response(Dimopoulos et al., 2002). The documented upregulated genes included, among others, several putative PRRs belonging to the CTL, PGRP, GNBP, TEP and LRIM gene families, as well as specific serpins, serine proteases and immune signalling molecules such as Cactus. Functional characterization of some of these genes has provided valuable clues concerning the direct involvement of specific immunity genes in the regulation of Plasmodium development in the vector, as discussed below.
The A. gambiae genome: a comparative analysis withDrosophila
Anopheles is the first insect vector whose genome has been sequenced (Holt et al., 2002). A comparative genome analysis of Drosophila and Anopheles,which are thought to have diverged approximately 250 million years ago, has revealed that nearly half the genes are 1:1 orthologues(Zdobnov et al., 2002). Analysis of 18 gene families that include some innate immune-related genes revealed a twofold deficit in orthologous pairs relative to the genomes as a whole (Christophides et al.,2002). The orthologue-poor families included putative pattern recognition, signal modulation and effector molecules. Interestingly, immunity gene families, in particular those involved in pattern recognition, signal modulation and effector mechanisms, showed substantial species-specific expansions. Genes belonging to immune signalling pathways are highly conserved, and this is probably attributed to the multiple functions they serve, many of which involve developmental processes(Christophides et al., 2002; Zdobnov et al., 2002). The expansion and diversification of the PRRs may reflect a strong selective pressure, leading to faster evolution in the face of distinct microbial floras prevailing in the ecological habitats of the different species. An extreme example of this expansion/diversification is the FBN-lectin gene family, which underwent two independent large expansions: one in Anopheles,resulting in 52 genes, and the second in Drosophila, resulting in 11 genes; in these species, only two orthologous pairs persist!
Gene discovery and functional analysis in A. gambiae
The A. gambiae genome provides unprecedented opportunities for mosquito research. For example, whole genome expression analysis using microarrays is now feasible and is expected to provide new insights into potential functions of the mosquito immune system during its interactions with various pathogens, including Plasmodium. Microarray technology is a very powerful tool for gene discovery; as previously mentioned, a microarray-based study allowed the identification of numerous genes induced by both bacterial and Plasmodium challenge(Dimopoulos et al., 2002);several of these are currently under in-depth analysis. In the framework of an international Mosquito Microarray Consortium (MMC) engaging a number of mosquito groups, our laboratory has constructed a 20 000 EST microarray(MMC1), and full-genome microarrays based on unique genomic amplicons have been designed (G. K. Christophides, unpublished data).
Direct and heritable reverse genetics
Gene discovery and generation of hypotheses based on expression patterns must be complemented with other tools that permit direct assessment of gene function. For this purpose, a direct reverse genetics method based on injection of dsRNA into adult mosquitoes has been established(Blandin et al., 2002), in addition to using the dsRNA treatment of A. gambiae haemocyte-like cell lines (Levashina et al.,2001). For example, injection of dsRNA corresponding to the Anopheles defensin gene (DEF1) in the body cavity of adult mosquitoes efficiently and reproducibly silenced the expression of this gene at the mRNA and protein levels (Blandin et al.,2002). Analysis of the knockout (KO) mosquitoes revealed that defensin is necessary to combat Gram-positive but not Gram-negative bacterial infections in vivo; however, the development of P. bergheiwas not affected by the absence of defensin. The injected dsRNA is detectable for at least 12 days post-injection and, hence, is stable enough to mediate a long-lasting effect. To date, the RNAi approach has been utilized to assay efficiently the function of several A. gambiae genes in response to malaria infection, as described below. The RNAi silencing technique was also established in adult D. melanogaster flies and used to efficiently silence components of the Toll pathway(Goto et al., 2003).
The replacement of wild mosquito populations with genetically modified strains refractory to Plasmodium development has been widely postulated as a potential strategy to control malaria transmission. Towards this aim, substantial effort has been invested to genetically transform Anopheles species. The early single event of A. gambiaetransformation (Miller et al.,1987), using the transposable element (TE) P from Drosophila, proved serendipitous and not transposon mediated. Routine genetic manipulation methods awaited utilization of TEs that were truly mobile in mosquitoes and utilization of effective dominant selectable markers. These tools were developed over the next decade with encouragement from the success achieved in the medfly, Ceratitis capitata(Loukeris et al., 1995), and the mosquito Ae. aegypti (Coates et al., 1998; Jasinskiene et al., 1998). The achievement of efficient transformation of the Asian malaria vector, A. stephensi(Catteruccia et al., 2000),was followed by that of A. gambiae(Grossman et al., 2001) and,more recently, A. albimanus, the south American vector(Perera et al., 2002).
With the methods for transgenesis in place, anopheline mosquitoes refractory to Plasmodium development have been generated. Bloodmeal-induced expression of two alternative transgenes in A. stephensi, a short effector peptide(Ito et al., 2002) and a bee venom phospholipase (Moreira et al.,2002), led to dramatic reduction in oocyst numbers and greatly impaired transmission of P. berghei to naïve mice. Preliminary results also suggested that transgenic lines of Ae. aegypti were rendered resistant to the development of P. gallinaceum through the bloodmeal-induced, systemic expression of defensin A(Shin et al., 2003b).
Although these results illustrate the potential of transgenic technology to study mosquito–parasite interactions, further refinements are vital to the development of this field, especially in view of the negative impact of transformation on mosquito fitness(Catteruccia et al., 2003). Clearly, the identification of anopheline promoters that drive transgene expression in a stage- and tissue-specific manner is a major challenge. These tools will be essential not only for the expression of effector molecules targeted against the critical stages of parasite development in the midgut and salivary gland but also for the silencing of mosquito genes that act as positive regulators of parasite development. To date, only a few promoters have been characterized and used to drive gene expression in mosquitoes. These include Aedes vitellogenin (Shin et al., 2003a), Drosophila actin5C(Brown et al., 2003) and Anopheles carboxypeptidase promoter(Ito et al., 2002). The characterization of mosquito promoter sequences could be greatly aided by utilization of the Cre–LoxP system, which was shown recently to be active in Ae. aegypti(Jasinskiene et al., 2003). This system would allow precise functional comparison of alternative promoters if it is used to target integration of transgenes into the same chromosomal location: vagaries in expression caused by position effects of the transgene insertion site would be eliminated, and promoter activity would be easily compared between transgenic lines.
These approaches will be complemented by the identification of inducible promoters and conditional expression systems permitting temporal and tissue-specific control of transgene expression or gene silencing (using dsRNA-producing transgenes). Such tools will make it possible to finely monitor effector mechanisms and gene function and to dissect in detail immune signalling pathways in the vector. Indeed, anopheline strains have been produced recently that conditionally express transgene in the adult midgut and haemocytes (G. Lycett and T. G. Loukeris, personal communication). In these lines, tissue-specific expression is directed by a promoter of a serpin gene,while conditional regulation is achieved by using the tetracycline transactivator (Gossen and Bujard,1992). Transgene expression can be switched on or off by the supply of tetracycline analogues to the mosquito. With this system,potentially any stage- or tissue-specific promoter can be made`inducible'.
Anopheles and Plasmodium: an interplay of immune attack and evasion?
Vector immune responses are believed to account, at least in part, for the major parasite losses during sporogonic development. In extreme cases, all Plasmodium parasites are killed in genetically selected refractory mosquitoes. Anopheles dirus mosquitoes selected for refractoriness completely block the development of P. yoelii ookinetes by melanotic encapsulation (Somboon et al.,1999), and refractory A. gambiae mosquitoes cause the lysis of P. gallinaceum ookinetes in the cytosol of infected midgut cells (Vernick et al., 1995). The best-studied case of refractoriness is melanotic encapsulation of Plasmodium ookinetes (Collins et al., 1986) in an A. gambiae refractory strain (L35). Melanization takes place in the extracellular space, between the midgut epithelial cells and the basal lamina. Genetic mapping of the L35 phenotype revealed one major (Pen1) and two minor (Pen2 and Pen3) quantitative trait loci (QTL) implicated in this response against P. cynomolgi B (Zheng et al., 1997). A recent study revealed that refractoriness of L35 mosquitoes to P. cynomolgi Ceylon, a different but related species,is controlled by at least three QTLs (Pcen2R, 3R and 3L)(Zheng et al., 2003). Interestingly, while Pcen2R and 3L map near Pen3and Pen2, respectively, and may actually be Pen3 and Pen2, Pcen3R represents a novel QTL unrelated to Pen1,suggesting that different genetic loci may be involved in responses to different malaria parasites. Sequencing of 528 kb of DNA from the Pen1 region revealed a remarkable number of sequence polymorphisms that constitute two alternative haplotypes over at least 121 kb(Thomasova et al., 2002). The significance of these haplotypes, and more generally the molecular basis of the complete melanotic phenotype, is not yet fully understood, although L35 mosquitoes show a high level of reactive oxygen species, which is further enhanced by bloodfeeding (Kumar et al.,2003). Similarly, the molecular basis of P. gallinaceumlysis in a refractory strain of A. gambiae(Vernick et al., 1995) is still unknown. However, recent studies showed that the melanotic response of L35 mosquitoes can be reversed by silencing specific A. gambiaeimmunity genes (Blandin et al.,2004; G. K. Christophides, unpublished data).
Melanization of P. falciparum ookinetes is rare in infected field-caught mosquitoes (Niare et al.,2002; Schwartz and Koella,2002). Rather, these mosquitoes are characterized by a high natural frequency of segregating resistance alleles that apparently attenuate the intensity of infection in the vector(Niare et al., 2002). This suggests that P. falciparum induces a strong selective pressure on Anopheles. Therefore, the parasite and its vector most likely represent a co-evolving system in dynamic equilibrium. The fact that the L35 strain of A. gambiae melanizes several species of malaria parasites,including P. berghei, P. gallinaceum, P. cynomolgi B and allopatric but not sympatric strains of P. falciparum(Collins et al., 1986), adds further support to this hypothesis. Several questions arise concerning mosquito factors that may specifically protect sympatric ookinetes and the means by which the parasite may evade or subvert the mosquito immune responses. A recent high-throughput proteomic approach has revealed that P. falciparum sporozoites express several proteins of the var gene family and other surface receptors(Florens et al., 2002) that were initially thought to be restricted to the mature asexual stages of the parasite. The fact that var genes are involved in immune evasion in the vertebrate host makes it tempting to explore whether var or other specific parasite surface proteins can mediate immune evasion in the vector as well as the vertebrate host.
Protozoan pathogens have evolved several mechanisms to evade the immune responses of the vertebrate host, including antigenic variation, shedding of surface proteins, antigenic mimicry, hiding inside cells and modulation of the host immune responses (Zambrano-Villa et al., 2002). Until recently, the molecular mechanisms that control the number of these parasites in their mosquito vectors, thus facilitating their transmission to vertebrates, remained uncharacterized. They could only be proposed by analogy to antibacterial immunity, by in vitro studies or by inference from descriptions of gene expression patterns. However,application of the dsRNA-mediated gene silencing technique in vivohas changed this situation radically. Recently, specific mosquito gene products, which act as antagonists of parasite development, have been identified in living mosquitoes, as well as others that act as agonists protecting the parasite against antagonists. The antagonists identified to date are the thioester-containing protein TEP1 and a leucine-rich protein,LRIM1. TEP1 is an acute-phase haemocyte-specific protein previously shown to bind and opsonize bacteria in a thioester-dependent manner(Levashina et al., 2001). Recently, TEP1 was also shown to be involved in the killing of P. berghei ookinetes as they cross the midgut epithelium of A. gambiae (Blandin et al.,2004). This was supported by several lines of evidence: (1) TEP1 knockout induces a fivefold increase in ookinete survival in a suceptible (G3)strain of A. gambiae and permits the successful development of P. berghei in a refractory (L35) strain of the same species; (2) TEP1 binds to the surface of ookinetes after they cross the midgut epithelium in both strains, but the timing of binding differs between these strains, and (3)TEP1-associated ookinetes display degeneration, evidenced by parasite blebbing, loss of the vital fluorescent marker GFP, nuclear abnormalities and fragmentation, and perturbations in the distribution of the ookinete-specific surface protein P28. Interestingly, 100% of ookinetes are killed in the L35 strain as compared with 80% in the G3 strain. The identification of TEP1 staining on live, morphologically normal ookinetes has suggested that TEP1 first binds to the ookinetes and then leads to their degeneration. Two polymorphic alleles of TEP1 were identified: TEP1r is associated with the L35 strain and TEP1s with the susceptible strain. However, conclusive evidence is still missing as to whether these polymorphic alleles are associated with the faster binding of TEP1 and more efficient killing of ookinetes in the L35 strain.
In parallel studies (Osta et al.,2004), LRIM1 was identified as a new parasite antagonist whose absence induces a dramatic, nearly fourfold, increase in the number of parasites in susceptible mosquitoes. LRIM1 is also predominantly expressed in the carcass (mosquito remnant following midgut isolation) as compared with the midgut and is specifically upregulated in the carcass of infected mosquitoes as compared with non-infected mosquitoes. Interestingly, LRIM1 expression in the midgut is strong and transient: upregulation occurs at 24–28 h post-infection, coinciding with the period of ookinete invasion of the midgut epithelium. Additional experiments revealed that two A. gambiaeC-type lectins, CTL4 and CTLMA2, act as agonists protecting the parasite from mosquito immune responses (Osta et al.,2004). Silencing either lectin gene by RNAi induces massive melanization of ookinetes in the susceptible A. gambiae G3 strain:the CTL4 KO results in melanization of nearly all ookinetes, while partial melanization is observed in CTLMA2 KO mosquitoes. Both lectins are predominantly expressed in the mosquito carcass (possibly in fat body and/or haemocytes) as compared with the midgut and are specifically upregulated in the carcass during ookinete invasion of the midgut epithelium, suggesting some immune signalling between tissues. These results highlight the potential use of CTL genes or proteins as targets to block Plasmodium transmission in the vector.
Genetic epistasis analysis revealed that the melanization response induced in the absence of CTLs requires LRIM1 function: the double KO of LRIM1 with either lectin gene completely abolishes the melanization phenotype and induces a fourfold increase in oocyst numbers, a phenotype similar to that of the single LRIM1 KO. Further research is being conducted to address the detailed mechanisms by which LRIM1 dramatically limits the parasite load in the vector, while CTLs protect the parasite. For now, our understanding can be summarized as in Fig. 3.
It is clear that parasite transmission depends upon complex molecular and cellular interactions acting at different levels of the parasite's life cycle in the vector (Alavi et al.,2003). Deciphering these interactions and identifying the molecules (defence or non-defence) that negatively and positively regulate parasite development in the vector will provide valuable information that can be exploited in designing novel transmission-blocking strategies for vector-borne pathogens. It will obviously be a matter of considerable importance to determine whether TEP1, LRIM1, CTL4 and CTLMA2 are induced by and act on P. falciparum in the same manner as on P. berghei.
Concluding remarks and perspectives
A decade ago, dissection of Anopheles innate immunity and the molecular interactions occurring between the vector and the parasite was a distant dream. Now, there is conclusive evidence that the Anophelesimmune system is a determining factor of vectorial capacity, and our knowledge of the specific molecules that are involved is advancing rapidly. Several breakthroughs underlie this remarkable progress: the development of tools for transformation of both the Plasmodium parasite and its Anopheles vector, the establishment of a direct reverse genetic technique (RNAi) in the mosquito for rapid investigation of gene function, the development of microarrays for gene discovery and genome-wide expression profiling and, most importantly, the sequencing of the genomes of A. gambiae (Holt et al.,2002), the human malaria parasite, P. falciparum(Gardner et al., 2002), and the model rodent malaria parasite, P. yoelii yoelii(Carlton et al., 2002). The genome sequence of the widely used model rodent malaria parasite P. berghei will also be available soon. With these tools in hand, the scientific community can now decipher the key interactions occurring between the parasite and the mosquito during the different stages of parasite development. Of particular interest is the in-depth characterization of the Anopheles immunity proteins that control parasite development in the vector and the immune signalling pathways responsible for regulated production of these proteins. Identification of both parasite antagonists and agonists in the vector is an important conceptual advance that sets the stage for dissecting the molecular mosquito–parasite interactions in detail. It also suggests a novel avenue for potential control of the malaria parasite in the mosquito. Establishment of the infection in Anopheles is essential for disease transmission and undoubtedly depends on multiple interactions of the parasite with other bloodmeal components and with multiple mosquito tissues that it encounters sequentially. Of these interactions, the ones that occur during the transition from ookinete to oocyst, when parasite numbers are at a minimum, are especially important and most probably reflect a fine balance between positive and negative mosquito factors. Such a balance may prove a favourable point for chemical intervention to reduce malaria transmission. Future detailed understanding of these molecular interactions may permit development of antimalarial `smart sprays': chemicals that are delivered like pesticides but are designed to disrupt interactions protective of the parasite or to reinforce others that are antagonistic to the parasite.
List of abbreviations
clip-domain serine proteases
Gram-negative bacteria-binding protein
leucine-rich immune protein
pathogen-associated molecular pattern
peptidoglycan recognition protein
pattern recognition receptor
quantitative trait locus
This paper constitutes a review for inclusion with the 2003 `Experimental Biology of Malaria and its Vectors' special issue [J. Exp. Biol.206(21)].
We thank previous and current members of the Kafatos laboratory, as well as our collaborators in the Mosquito Immunity Consortium who contributed to the work summarized in this review, and especially those who authorized the references to unpublished data. We are grateful to Gareth Lycett for critical reading. M.A.O. is supported by a Marie Curie Intra-European Fellowship. D.V. was initially supported by a European Commission Marie Curie Fellowship. The work on innate immunity in the laboratory was funded by the EMBL, NIH,MacArthour Foundation, SFB and European Commission.