Anopheles gambiae Blood Feeding Initiates an Anticipatory Defense Response to Plasmodium berghei

Mosquitoes transmit numerous human and animal diseases with devastating consequences worldwide. Malaria is caused by protozoan Plasmodium parasites and transmitted to humans by infected female Anopheles mosquitoes. The mosquito life cycle makes it an ideal disease vector as most adult females must feed on vertebrate blood to acquire nutrients for egg production. However, blood feeding also exposes the mosquito to infection from protozoan parasites, viruses and nematode worms. A further consequence of blood feeding is the dramatic rise in levels of endogenous bacteria in the mosquito midgut [1,2], which puts the mosquito at risk of systemic infection. Therefore, the mosquito immune system must defend against blood-borne infections and control its midgut bacterial populations [3,4,5].

The Anopheles gambiae innate immune system is responsible for eliminating the majority of invading Plasmodium ookinetes during the midgut stages of mosquito infection [6]. Two key immune proteins involved in anti-Plasmodium defense are Leucine-Rich repeat IMmune protein 1 (LRIM1) and APL1C, as shown by striking increases in live parasites when these genes are silenced [7,8,9]. LRIM1 and APL1C are closely related proteins that possess leucine-rich repeat (LRR) domains, which are found in host defense proteins of many phyla, such as vertebrate Toll-like receptors [10]. LRIM1 and APL1C circulate in the hemolymph as a disulfide-linked heterodimeric complex [11,12]. This complex is involved in parasite killing through its interaction with the complement-like effector protein, TEP1. LRIM1/APL1C binds to proteolytically processed, mature TEP1 (known as TEP1_cut), promoting its stabilization, preventing it from reacting with self-tissues and enabling it to opsonize parasites [11,12]. Direct binding of TEP1_cut to the ookinete surface triggers parasite lysis and melanization reactions, resulting in parasite killing and clearance [13]. TEP1, LRIM1 and APL1C are core members of the mosquito complement-like pathway, which plays a broad role in innate immunity including defense against bacteria [14]. These proteins are constitutively present in the hemolymph, bathing the basal labyrinth of the midgut and poised to attack malaria parasites as they emerge through invaded cells. Interestingly, the LRIM1/APL1C complex has also been demonstrated to interact with other TEP family members in vitro including TEP3, TEP4 and TEP9 [15].

Bioinformatic searches discovered a novel mosquito-specific family of proteins related to LRIM1 and APL1C [12,16]. To date, 24 members of this LRIM family have been identified in An. gambiae. Orthologs of most LRIMs and additional homologous proteins were discovered in the genomes of mosquitoes Aedes aegypti and Culex quinquefasciatus. However, no LRIM-related genes were found in other organisms, including Drosophila. LRIM members share a distinct genomic organization and protein domain architecture, which distinguishes them from the larger superfamily of LRR genes. An archetypal LRIM comprises a signal peptide, LRR motifs, a conserved pattern of cysteines and a coiled-coil domain. The members are divided into four subfamilies based on variations to this core structure. Long LRIMs have 10 to 13 LRR motifs, whereas Short LRIMs have 6 or 7. Transmembrane LRIMs possess a C-terminal transmembrane domain and Coil-less LRIMs lack a coiled-coil domain. Interestingly, several LRIMs are encoded within tight genomic clusters with evidence of local gene shuffling and duplication [16]. A cluster of Short LRIMs found in all three mosquitoes consists of LRIM7, LRIM8, LRIM9 and LRIM10. LRIM8 has duplicated in An. gambiae to give LRIM8A and LRIM8B.

Apart from LRIM1 and APL1C, the other LRIM members are largely uncharacterized to date. Certain LRIMs have been implicated in innate immunity, including defense against Plasmodium and bacteria [17,18,19,20,21]. It is unclear whether the LRIM family represents an adaptation to the hematophagous lifestyle of mosquitoes. With their versatile LRR domains, we hypothesize that the LRIMs are pathogen recognition proteins, and the family has diversified to recognize different microbes that mosquitoes encounter. This paper aimed to broaden our understanding of the LRIM family in An. gambiae by investigating whether any uncharacterized LRIMs are involved in anti-Plasmodium defense. We discovered LRIM9 is a novel antagonist of Plasmodium berghei infections with a striking expression profile. LRIM9 is highly enriched in adult female mosquitoes. Expression of LRIM9 is dramatically induced by blood feeding and regulated by ecdysone signaling. Our data suggest that LRIM9 functions via a unique immune mechanism independent of the known mosquito complement-like pathway. We hypothesize that LRIM9 is involved in an anticipatory immune response triggered by blood feeding, which defends against blood-borne infections such as Plasmodium. This is an original concept in An. gambiae innate immunity.

This study was carried out in strict accordance with the United Kingdom Animals (Scientific Procedures) Act 1986. The protocols for mosquito blood feeding and for infection of mosquitoes with P. berghei by blood feeding on parasite-infected mice were approved and carried out under the UK Home Office License PLL70/7185 awarded in 2010. The procedures are of mild to moderate severity and the numbers of animals used are minimized by incorporation of the most economical protocols. Opportunities for reduction, refinement and replacement of animal experiments are constantly monitored and new protocols are implemented following approval by the Imperial College Ethical Review Committee.

An. gambiae N'gousso and L3-5 strains were maintained, blood fed, and assayed for infection with P. berghei CON_GFP strain [22] as described previously [15]. Human blood feeding was performed using an artificial membrane feeding system [23]. Single and double knockdown experiments and parasite counts in dissected midguts were performed as described previously [12]. Primers used for synthesis of double-stranded RNA against LRIM9 are as follows with T7 tags in lower case:

LRIM9 RNAi For: taatacgactcactatagggACTGGCAGAAAAG CTTCCAA;

LRIM9 RNAi Rev: taatacgactcactatagggTGGCATTTTCTCG AACACAG.

Other primers for gene silencing (GFP, LRIM1, TEP1 and CTL4) have been reported elsewhere [23].

Total RNA was extracted from 10 whole mosquitoes per sample using the TRIzol Reagent (Invitrogen). For the developmental profile, 10 eggs, larvae (2nd or 4th instar) or pupae were used per sample. Total RNA was DNase treated using Turbo DNA-free kit (Ambion) and cDNA was synthesized from 1 µg RNA using SuperScript II kit with oligo(dT)_12-18 primers (Invitrogen). Quantitative real-time PCR was performed using the ABI Prism 7500 Fast Real-Time PCR System, as previously described [23]. Ribosomal gene S7 was used as the endogenous reference, and gene expression was quantitated relative to a calibrator control sample (e.g. dsGFP-treated mosquitoes). Primers for S7 and LRIM1 have been reported previously [23]. Primers for LRIM9 are as follows:

LRIM9 qRT-PCR F: TTCAGCATGCACTGGAAAAG;

LRIM9 qRT-PCR R: GTCGGTACCATCGGTTGACT.

An. gambiae LRIM9 was cloned into the pIEx-10(Novagen) expression plasmid in-frame with the plasmid signal peptide for secretion, an N-terminal Strep tag and a C-terminal 10× His tag as follows: first a DNA fragment containing LRIM9 was amplified by PCR from genomic DNA using the primers:

LRIM9 F: TGCAATTTTCGATTCAGTGC;

LRIM9 R: AAAGGACCCACATCTCAACG.

The 1,583-bp product was used as a PCR template using primers containing overhangs for ligase-independent cloning:

LRIM9 LIC F: gacgacgacaagatgGAGATTTCCAGCTCCGT GGTG;

LRIM9 LIC R: gaggagaagcccggtttGGCAGACGGTTCGGACGCCAC.

The resulting expression construct encodes a 445-amino acid fragment of LRIM9 removing its endogenous signal sequence and stop (LRIM9^HIS).

A cell line stably expressing LRIM9^HIS was selected using G418 by cotransfecting Sf9 cells with pIEx-10-LRIM9 and pIE1-neo(Novagen) plasmids as described previously [24]. LRIM9^HIS was affinity purified from 2 l of 0.22 µM sterile-filtered conditioned medium using a 5 ml HisTrap FF column on an ÄKTA purifier (GE Healthcare). After binding, the column was washed with buffer A (1× phosphate-buffered saline, PBS + 0.1% triton, 40 mM imidazole, pH 8.0) and the captured protein was eluted with buffer B (1× PBS + 500 mM imidazole, pH 8.0) and then concentrated using an Amicon Ultra centrifugal filter (Millipore). Purified LRIM9^HIS was analyzed by SDS-PAGE followed by Coomassie staining and quantified by Bradford assay. Approximately 215 µg of LRIM9^HIS was used to immunize guinea pigs for antibody production (Eurogentec). The pre-immune and immune sera were evaluated using Western blotting (see below).

Additionally, a rabbit anti-peptide antibody was generated against the C-terminal peptide NH₂-CDYARRLEVASEPSAK-COOH (Eurogentec). A second peptide against the internal peptide NH₂-DSDGTLLDKSTDGTDC-COOH was unsuccessful. These were used for some initial experiments but were replaced by the more sensitive whole-protein antibody, described above.

Hemolymph was collected from groups of individual mosquitoes as previously described [12]. Final sample volume was adjusted to 1 mosquito/µl; 10 mosquitoes per lane were used for SDS-PAGE and Western blot analysis. An exception was the CLIPA8 cleavage experiment, where 1.5 mosquito/µl and 15 mosquitoes per lane were used. Western blotting with SRPN3, TEP1 and APL1C antibodies was performed as previously described [12]. Positive LRIM9 guinea pig serum was used at 1/500 dilution in PBS + 3% milk and 0.05% Tween-20 for 1 h at room temperature. The TEP1 binding assay was performed using transfected Sua4.0 cells, as detailed previously [12]. Cells were transfected with pIEx10-LRIM9^HIS, pIEx10-GFP_secreted^HIS and pIEx10-APL1C^HIS.

Upon eclosion, mosquitoes were given 3 days to mate in a 30-cm³ population cage prior to injection with double-stranded RNA. Three days after injection, females were allowed to feed on an anesthetized mouse. After 48 h, a group of mosquitoes was dissected to examine blood meal digestion and ovary development. After 72 h, the remaining mosquitoes were placed in 50-mm Petri dishes on filter paper saturated with 0.1% saline. Dishes were stored at 27°C in darkness for 24 h to encourage oviposition, after which mosquitoes were removed and eggs were counted by direct observation under a dissecting microscope. Egg dishes were then half-filled with 0.1% saline, dusted with powdered fish food and incubated at 27°C for 3 days to allow hatching. Larvae were counted by direct observation under a dissecting microscope.

Ampicillin-resistant Escherichia coli OP-50 was grown in Luria-Bertani (LB) broth, harvested during logarithmic growth phase, washed with PBS and resuspended in PBS to give an optical density at 600 nM (OD₆₀₀) of 0.4 [25]. Female mosquitoes were injected with 69 nl of bacterial suspension. Bacterial viability was confirmed by counting colonies formed overnight at 37°C after plating cells onto LB agar. The bacterial proliferation assay was performed as previously described [25], except with 24 h between bacterial inoculation and sample collection. For survival assays, the number of dead mosquitoes was monitored daily for 10 days after bacterial injection. Challenge with bioparticles was performed as described previously [24]. CLIPA8 cleavage was assayed in hemolymph samples analyzed under reducing conditions, as described previously [26]. Female mosquitoes were injected with 69 nl of a 14.5 µg/µl suspension of 20-hydroxyecydsone (Sigma) in sterile PBS. This dose was shown to elicit a maximal increase in protein synthesis in the fat body in vitro [27].

For 4 days prior to blood feeding, newly eclosed mosquitoes were given 10% sucrose supplemented with 10 U/ml penicillin, 10 µg/ml streptomycin and 15 µg/ml gentamicin [3]. Efficacy of antibiotic treatment was assayed by plating homogenates of cohorts of 5 mosquitoes on LB agar and counting the colonies formed after incubation at 27°C for 2 days.

VectorBase Gene Identifiers were as follows: LRIM9, AGAP007453; LRIM1, AGAP006348; APL1C, AGAP007033; TEP1, AGAP010815; TEP3, AGAP010816; TEP4, AGAP010812; TEP9, AGAP010830; CTL4, AGAP005335; CLIPA8, AGAP010731; vitellogenin, AGAP004203; lipophorin, AGAP001826; S7, AGAP010592; SRPN3, AGAP006910.

To elucidate whether any uncharacterized An. gambiae LRIMs play a role in anti-Plasmodium defense, the family was screened for a parasite infection phenotype after gene knockdown by RNAi. Parasite numbers were monitored 7 days after susceptible mosquitoes were infected with GFP-expressing P. berghei. The screen identified LRIM9, a Short family member, as a novel antagonist of P. berghei infections. Silencing LRIM9 resulted in a significant 3-fold increase in live oocysts compared to dsGFP-treated controls (fig. 1a; online suppl. table 1; for all online suppl. material, see www.karger.com/doi/10.1159/000365331). However, there was no change in the proportion of infected mosquitoes (prevalence). LRIM9 was efficiently silenced with 84% average reduction in the transcript (online suppl. fig. 1) lasting for at least 7 days (data not shown).

To further investigate the role of LRIM9 in parasite melanization, the gene was silenced in two An. gambiae experimental models that are refractory to P. berghei infection where virtually all of the invading ookinetes are melanized. In L3-5 mosquitoes, a laboratory-selected refractory strain [28], silencing LRIM9 resulted in a significant decrease in the number of melanized parasites but did not produce live oocysts (fig. 1b, online suppl. table 2). In contrast, silencing LRIM9 did not alter the outcome of infection in susceptible mosquitoes following silencing of CTL4(fig. 1c, online suppl. table 3). In this refractory model, silencing CTL4 alone or in combination with LRIM9 led to the same significant decrease in live oocysts and increase in melanized ookinetes. In comparison, silencing CTL4 in combination with TEP1 completely blocked melanization induced by CTL4 silencing and resulted in a large increase in the number of live oocysts, consistent with previous observations [15]. Silencing efficiency of the LRIM9 transcript in L3-5 and CTL4 knockdown mosquitoes (89 and 76%, respectively) was comparable to the single knockdown in susceptible mosquitoes (online suppl. fig. 1).

To initially characterize LRIM9, we examined its developmental expression profile by quantitative real-time PCR (qRT-PCR). RNA was collected from eggs, two larval stages (2nd and 4th instar), pupae, newly eclosed, and 4-day-old sugar-fed mosquitoes. The transcript was most abundant in adult female mosquitoes (fig. 2). Compared to adult males, LRIM9 transcript was greater than 20-fold enriched in adult females. The LRIM9 transcript level was also extremely low in eggs, larval and pupal stages. These data suggest that LRIM9 functions primarily in adult female mosquitoes.

As LRIM9 is female enriched and a P. berghei antagonist, we investigated whether the LRIM9 transcript is induced in response to P. berghei infection. Indeed, qRT-PCR analysis showed that the LRIM9 transcript was robustly upregulated after feeding on infected blood. When maintained at 19°C, the permissive temperature of P. berghei, LRIM9 expression peaked at 48 h after feeding, where the transcript was 2.6-fold higher than sugar-fed controls (fig. 3a). Expression returned to baseline levels by 72 h. Interestingly, the same regulation was observed when mosquitoes fed on uninfected blood revealing that LRIM9 expression is not triggered by parasites, but instead is a consequence of blood feeding.

To gain insights into its function in vivo, an antibody was raised against LRIM9 and used to assay hemolymph (online suppl. fig. 2). The specific band identified at 50 kDa is consistent with the predicted size of LRIM9, suggesting that LRIM9, unlike LRIM1, APL1C and LRIM4, does not form covalent dimers in the hemolymph [12,15].

The antibody was next used to determine the temporal dynamics of the LRIM9 protein following a blood meal. Mosquitoes were given a murine blood meal and maintained at 19°C. Hemolymph was collected 3, 8, 12, 24, and 48 h after feeding and analyzed by Western blot (fig. 3b). Compared to the LRIM1/APL1C complex and SRPN3, which both remained relatively stable across all time points, LRIM9 was massively enriched in hemolymph after blood feeding. The earliest induction was observed at 8 h with a striking peak between 24 and 48 h followed by a rapid decrease by 72 h. However, even at 96 h, the LRIM9 protein was more abundant than it was prior to blood feeding.

Given that LRIM9 protein is so strongly induced by blood feeding, RNAi knockdown was measured in blood-fed mosquitoes (online suppl. fig. 3). Indeed, following knockdown, LRIM9 protein levels were significantly reduced even after blood feeding with only a very faint band detectable. We also found that LRIM9 was upregulated when mosquitoes fed on human blood (online suppl. fig. 4). Protein abundance peaked at 24 h after human blood feeding, which is likely because mosquitoes were kept at 27°C rather than 19°C as in the previous experiment using murine blood.

We next asked whether LRIM9 upregulation occurs after subsequent blood meals or is specific to the first. To test this, mosquitoes were allowed to take two consecutive murine blood meals (separated by 96 h), and hemolymph was collected from cohorts 24 h after their first and second blood meal. Hemolymph from sugar-fed controls, which were the same age as the blood-fed mosquitoes but were not blood fed, was analyzed for comparison. Western blot analysis using the LRIM9 antibody demonstrated that the LRIM9 protein (and transcript, not shown) is highly induced after both the first and second blood meal, with no obvious dampening of the second response (fig. 3c).

We then investigated whether LRIM9 induction is triggered by the increase in midgut bacteria that occurs after blood feeding [1,2]. To test this hypothesis, mosquitoes were fed a spectrum of antibiotics to significantly reduce their midgut flora prior to blood feeding [3]. LRIM9 protein levels were unaffected by antibiotic treatment in both sugar-fed and blood-fed mosquitoes (fig. 4a). These findings suggest that LRIM9 upregulation after blood feeding is independent of the endogenous midgut bacteria. To further test whether LRIM9 is bacterial responsive, we performed qRT-PCR after injection of bacteria directly into the hemocoel of sugar-fed mosquitoes. LRIM9 transcript was unaffected 24, 48 and 72 h after Escherichia coli(gram-negative bacteria) injection, whereas LRIM1 was highly induced at 24 and 48 h (fig. 4b). Furthermore, LRIM9 silencing had no effect on the number of colony-forming units or the survival of blood-fed mosquitoes after E. coli injection (fig. 4c, d). In contrast, silencing LRIM1 significantly increased mosquito mortality and bacterial proliferation. Finally, silencing LRIM9 did not affect the bacterial-induced cleavage activation of CLIPA8 (fig. 4e), which is deemed a functional marker of antibacterial defense [26,29]. Together, these data show that LRIM9 is not regulated by midgut bacteria or infection with exogenous bacteria.

As we had excluded bacterial regulation, we investigated whether the hormonal changes that occur after blood feeding influence LRIM9 regulation. In particular, the steroid hormone ecdysone drives the transcription of many key genes for vitellogenesis. Ecdysone is primarily secreted by the ovaries 10-36 h after blood feeding (peaking at 24 to 36 h), and is hydroxylated into 20-hydroxyecdysone (20E) [30]. As LRIM9 is strongly induced by blood feeding, we directly tested whether it is 20E responsive. Using qRT-PCR, the LRIM9 transcript was highly induced 24 h after intrathoracic injection of 20E (fig. 5a). The LRIM1 transcript was relatively unresponsive to both 20E and blood feeding. Furthermore, the LRIM9 protein was strongly enriched in the hemolymph in response to 20E injection (fig. 5b). In contrast to the effect on transcript levels, protein induction was stronger after blood feeding than after 20E injection. This suggests that in addition to regulating the LRIM9 transcript, blood feeding enhances its translation. A similar effect has been shown previously for blood feeding activation of translation of AaGATAa transcripts in the fat body of Ae. aegypti [31]. Again, the LRIM1/APL1C complex was unresponsive to this treatment. Coomassie staining of samples confirmed that vitellogenin and lipophorin, two nutrient transport proteins known to be regulated by 20E, were induced after 20E injection (fig. 5c).

As An. gambiae female mosquitoes require a blood meal to enable egg development, and the 20E targets vitellogenin and lipophorin play an important role in this process, we hypothesized that, in addition to its role in anti-Plasmodium immunity, LRIM9 may also be involved in mosquito reproduction. To examine this possibility, mated dsGFP- and dsLRIM9-treated female mosquitoes were blood fed and encouraged to lay eggs. Initial microscopic analysis of both treatment groups 72 h after blood feeding revealed no differences in ovarian development or blood meal digestion (data not shown). Furthermore, we found that total eggs laid, mean eggs laid per female, larval hatching and percentage of fertile females were equivalent between dsGFP and dsLRIM9(table 1). Therefore, LRIM9 silencing had no impact on mosquito reproduction or blood meal digestion.

The mosquito complement-like pathway has a well-established role in parasite killing and melanization. To characterize whether LRIM9 functions in this pathway, we first investigated whether silencing it affects the stability or abundance of LRIM1, APL1C and TEP1. Hemolymph collected from blood-fed mosquitoes following LRIM9 silencing was indistinguishable from the control; there was no effect on the abundance of TEP1-F, TEP1_cut, the LRIM1/APL1C complex or the loading control, SRPN3 (fig. 6a). There was also no effect of LRIM1 silencing on LRIM9 abundance or mobility. In contrast, as previously shown, silencing LRIM1 abolishes the LRIM1/APL1C complex and results in loss of TEP1_cut from the hemolymph [11,12]. These results suggest that LRIM9 does not interact with TEP1.

To test this directly, we expressed a recombinant His-tagged LRIM9 in Sua4.0 cells, a mosquito hemocyte-like cell line that naturally secretes endogenous TEP1-F and TEP1_cut[12]. LRIM9^HIS was affinity purified from the conditioned medium, and the bound material was assayed for the presence of TEP1 (fig. 6b). Interestingly, there was no detectable signal for either TEP1-F or TEP1_cut in the bound material, indicating that LRIM9 does not interact with TEP1. Furthermore, because Sua4.0 cells endogenously produce the LRIM1/APL1C/TEP1_cut complex, the lack of TEP1 signal in the captured material also reveals that LRIM9 does not interact with the LRIM1/APL1C complex. As a positive control, TEP1_cut was robustly present in samples purified from the conditioned medium of APL1C^HIS transfected cells. These results indicate that LRIM9 does not directly interact with TEP1. Therefore, LRIM9 acts via an unknown immunity mechanism independent of the known complement system.

Here, we have characterized LRIM9, a novel member of the LRIM family in An. gambiae. LRIM9 was found to be a novel antagonist of P. berghei infection. However, the precise function of LRIM9 in defense against Plasmodium remains unclear and requires further study. The characteristic LRR and coiled-coil domains of LRIM9 suggest involvement in pathogen recognition and interactions with other immune proteins [16,32,33]. Our data suggest that LRIM9 is not directly involved in the known mosquito complement-like pathway. However, LRIM9 might interact with other TEP family members or could function downstream of TEP1. We currently lack a suitable assay to test the latter. The observation that LRIM9 silencing does not recover live oocysts in refractory L3-5 mosquitoes suggests LRIM9 might function downstream of parasite lysis/killing mechanisms to promote melanization. In Drosophila, melanization has been shown to increase the efficiency of other immune reactions [34] and the same has been suggested in An. gambiae [26]. Melanization has been proposed as a clearance mechanism for dead parasites in refractory L3-5 mosquitoes, whereas in CTL4 knockdown mosquitoes parasites are thought to be directly killed by melanization [29]. As silencing LRIM9 has a phenotype in L3-5 mosquitoes but not in the CTL4 knockdown, we believe that LRIM9 is not a component of the melanization cascade per se but is promoting recognition of dead parasites, which then leads to their melanization. It remains to be determined whether LRIM9 is itself a scavenger receptor of dead parasites.

LRIM9 has a unique expression profile: it is enriched in adult female mosquitoes and strongly induced by blood feeding. Our results correlate well with previous whole-genome transcriptomics analyses [35,36,37]. Mosquitoes responded similarly to murine and human blood, which indicated that possible causes of upregulation of LRIM9 may include common mammalian blood components, formation of the blood bolus, distension of the gut or signaling occurring after blood feeding.

We have shown that LRIM9 is regulated by 20E, a steroid hormone secreted by the ovaries in response to blood feeding. In agreement with our findings, LRIM9 expression is 3-fold higher 24 h after blood feeding in mosquitoes with ovaries compared to genetically manipulated ovaryless mosquitoes [Magnusson and Crisanti, pers. commun.]. The most parsimonious hypothesis is that LRIM9 is transcriptionally regulated by 20E, and is activated directly by the ecdysone receptor or indirectly via an ecdysone-regulated transcription factor. It would be insightful to determine whether LRIM9 expression is dramatically reduced after 20E injection when the ecdysone receptor is silenced. Alternatively, transcript stability, translational repression or protein turnover could be regulated by 20E [38].

We have determined that neither midgut bacterial flora nor exogenous bacteria influence LRIM9 regulation. However, the nutrient-sensitive target of rapamycin (TOR) pathway, insulin/insulin-like growth factor signaling or microRNAs [39,40,41] could feasibly contribute to LRIM9 expression. In mosquitoes, steroid hormones and TOR signaling work synergistically to control expression of yolk protein precursors [42].

Regulation of immune genes by 20E has been observed previously in insects. Melanization pathway components, such as PPO, are under 20E regulation in An. gambiae [43,44] and other blood-feeding insects [45]. Like LRIM9, PPO2, PPO3 and PPO9 are strongly induced after blood feeding [6,44]. As we demonstrated that LRIM9 can promote melanization, future studies should investigate a possible interaction between LRIM9 and PPO activation. Using the genome-wide expression map available for An. gambiae [46], we found that LRIM9 and vitellogenin are coregulated. Like LRIM9, vitellogenin is regulated by 20E and by multiple blood meals [47,48,49]. As vitellogenin is produced by the mosquito fat body and ecdysone is hydroxylated into 20E by this tissue [30], we hypothesize that LRIM9 is also produced by the fat body. Indeed, previous microarray analyses demonstrate that LRIM9 expression is significantly higher in the fat body compared to the midgut or ovaries [36,37]. Furthermore, LRIM9 was not enriched in the transcriptome of circulating An. gambiae hemocytes [50]. However, this warrants further investigation as blood feeding was recently demonstrated to induce hemocyte proliferation and activation [51]. Interestingly, vitellogenin in the hemolymph interferes with TEP1-mediated killing of Plasmodium parasites by reducing the efficiency of TEP1 binding to parasite surfaces [52]. However, we did not observe a link between LRIM9 and mosquito fecundity or TEP1 levels in the hemolymph. Therefore, LRIM9 is likely to function via a novel immunity mechanism.

We hypothesize that LRIM9 is induced in anticipation of blood-borne infections rather than in response to infection, which is an original concept in An. gambiae immunity. In support of this hypothesis, LRIM9 is induced after both uninfected and infected blood meals. Hematophagous insects, like mosquitoes, are at high risk of infection from blood-borne pathogens, including Plasmodium parasites, filarial nematodes and viruses. Our proposed theory of anticipatory immunity in mosquitoes would be a highly important defense mechanism against such infections. By assuming every blood meal is infectious and inducing immune effectors, like LRIM9, in anticipation of such infections, the mosquito does not need to specifically recognise each pathogen but is prepared for imminent danger. It is unknown whether LRIM9 plays a generalized role or is specific to particular blood-borne pathogens, such as Plasmodium. Importantly, it should be determined whether anticipatory immunity and LRIM9 are involved in defense against the human malaria parasite, P. falciparum.

In contrast to LRIM9, LRIM1 and APL1C constitutively circulate at high levels in the hemolymph poised to attack invaders, which has been described as basal immunity [53]. Unlike LRIM9, they are not specific to blood-borne infections because they are also involved in antibacterial defense and phagocytosis [54]. Basal immunity is able to rapidly defend against pathogens that directly enter the hemolymph, such as bacteria and fungi, whereas anticipatory immunity is a slower response best suited to protect against blood-borne pathogens, which take longer to invade the hemolymph. Anticipatory immunity would potentially be less energetically costly than basal immunity, whilst still providing sufficient protection. Anticipatory immunity has not, to our knowledge, been reported in the innate immune response of another organism. Innate immunity is traditionally considered to be poorly specific and nonanticipatory. However, the adaptive immune response of vertebrates has been previously proposed as ‘anticipatory'. In a mechanism distinct from that in mosquitoes, vertebrates generate diverse repertoires of T and B lymphocyte receptors by rearrangement of gene segments to enable recognition of any potential antigen [55].

By coupling the regulation of LRIM9 with hormonal changes that occur after blood feeding, the mosquito enhances its immune system at the most opportune time. The fascinating link between innate immunity and steroid hormones has been widely observed in vertebrates and invertebrates [56,57,58]. Steroid hormones, like juvenile hormone and 20E, are extremely versatile and able to regulate development, growth, reproduction, ageing and immunity in insects [59,60,61,62]. It was recently reported that 20E transferred by male An. gambiae mosquitoes during mating modulates oogenesis in females [63]. Although hormonal regulation of insect development is well understood [64,65,66], the mechanisms regulating immunity are still being uncovered. In Drosophila, developmental production of 20E regulates the innate immune response via a complex network of transcriptional circuits [67].

20E has been shown to regulate the Drosophila Imd pathway and modulate production of antimicrobial peptides [68]. Furthermore, 20E controls the expression of the pattern recognition receptor, PGRP-LC, which protects the fly against bacterial infection [67]. Therefore, reduced 20E signaling can severely immunocompromise the adult fly. A recent study demonstrated that ecdysone is essential for hemocyte activation in Drosophila larvae and normal immune function [69,70]. Without this activation, larvae were defective in bacterial phagocytosis and unable to survive bacterial infection [69]. Ecdysone has also been implicated in the control of hemocyte phagocytosis in Rhodius prolixus, the Hemipteran vector of Chagas disease [71]. Importantly, hormone signaling has been implicated in the synchronization of different immune responses [69], which suggests LRIM9 could be involved in more than one immune function. Future research will aim to shed light on the roles of LRIM9 and hormonal regulation in the mosquito's anticipatory immune response.

The LRIM9 promoter could be favorable for transgenic strategies where adult female-specific expression is required, such as Release of Insects carrying a Dominant Lethal (RIDL) [72]. If the expression profile of LRIM9 in An. gambiae is conserved in other mosquito species, such a strategy could be used to target numerous mosquito-borne diseases, such as arboviruses and filarial worms, as well as malaria. Indeed, using the available data in VectorBase, the LRIM9 ortholog in Ae. aegypti(AAEL001414) is also highly expressed in females and induced 12 h after blood feeding [73]. Furthermore, in all three sequenced mosquito genomes, LRIM9 resides in a genomic cluster with other Short LRIMs (LRIM7, LRIM8A, LRIM8B and LRIM10 in An. gambiae) [12,16]. Based on extant microarray data, this cluster of Short LRIMs seems to share a similar expression profile (with the exception of LRIM7) in An. gambiae and Ae. aegypti [35,37,73]. Future studies should further investigate these Short LRIMs and their relationship with LRIM9 in these and other disease vector mosquitoes where they are present, such as Cx. quinquefasciatus [16].

This work was funded by the Wellcome Trust (093587/Z/10/Z). L.M.U. was supported by a Biotechnology and Biological Sciences Research Council Doctoral Training Grant. We thank Dr. Lavanya Bhagavatula for generating the stable Sf9 cell line expressing LRIM9^HIS. We also wish to thank Katarzyna Sala and Tibebu Habtewold for mosquito production and parasite maintenance.

Current address of M.P.: Department of Pathobiology, School of Veterinary Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA.