Enterobacter-Activated Mosquito Immune Responses to Plasmodium Involve Activation of SRPN6 in Anopheles stephensi

Successful development of Plasmodium in the mosquito is essential for the transmission of malaria. A major bottleneck in parasite numbers occurs during midgut invasion, partly as a consequence of the complex interactions between the endogenous microbiota and the mosquito immune response. We previously identified SRPN6 as an immune component which restricts Plasmodium berghei development in the mosquito. Here we demonstrate that SRPN6 is differentially activated by bacteria in Anopheles stephensi, but only when bacteria exposure occurs on the lumenal surface of the midgut epithelium. Our data indicate that AsSRPN6 is strongly induced following exposure to Enterobacter cloacae, a common component of the mosquito midgut microbiota. We conclude that AsSRPN6 is a vital component of the E. cloacae-mediated immune response that restricts Plasmodium development in the mosquito An. stephensi.


Introduction
Malaria is among the deadliest infectious diseases, killing in excess of one million people every year, mostly of African children under the age of five. Transmission is entirely dependent on the completion of the life cycle of Plasmodium, the causative agent of malaria, in its mosquito vector. After ingestion of an infectious blood meal, Plasmodium gametocytes differentiate into male and female gametes that fertilize to generate a diploid zygote. After a round of DNA replication, the tetraploid zygote differentiates into a motile ookinete. At approximately 24 h after ingestion the ookinete invades the mosquito midgut and differentiates into sessile oocysts. Within 7 to 14 days (depending on parasite species), thousands of sporozoites are released from each oocyst into the mosquito hemocoel. Sporozoites must successfully invade the salivary glands to ensure transmission when the infected mosquito bites and inoculates sporozoites into a new individual [1,2]. The parasite life cycle in its mosquito host is complex, and dramatic losses in parasite numbers occur at each stage of Plasmodium development [2,3]. Ookinete midgut invasion represents the largest bottleneck in parasite numbers [2,3], as ookinetes must overcome the effects of the mosquito midgut microbiota and the innate immune responses in order to successfully transition into an oocyst [4].
The mosquito midgut microbiota is very dynamic, with dramatic fluctuations based upon life-stage, nutritional status, and age [5]. After a blood meal, mosquito commensal bacteria undergo changes in their population structure to enrich for enteric gram-negative bacteria capable of surviving the harsh, digestive environment of the mosquito midgut [5]. Within this nutrient-rich environment, bacteria reach high numbers at a time that coincides with ookinete invasion (,24 h post-blood meal) [6], and can greatly influence the success of Plasmodium parasite development [6][7][8][9]. In addition, the presence of endogenous bacteria is also thought to prime the mosquito innate immune response to limit parasite survival [9][10][11]. Basal expression levels of anti-microbial genes controlled by mosquito innate immune pathways limit bacterial proliferation and indirectly contribute to cross immune protection against Plasmodium parasites [9][10][11][12]. In the absence of midgut microbiota, mosquito susceptibility to Plasmodium infection is greatly increased [9].
While the involvement of the mosquito's microbiota in the anti-Plasmodium response is beginning to be explored [8][9][10][11][12], the mosquito innate immune system also contributes a major role in parasite attrition [4,13,14]. As Plasmodium ookinetes reach the basal lamina of the midgut, parasites are subjected to components of the mosquito hemolymph that destroy a large proportion of the invading parasites [15][16][17]. Recent evidence suggests that parasite immune recognition is a critical determinant of invasion success, mediated by epithelial nitration of parasites during the process of midgut invasion [18].
Previously, we identified a putative serine protease inhibitor (SRPN6) that modulates rodent malaria parasite development in anopheline mosquitoes [19,20]. In An. stephensi, SRPN6-silencing leads to a significant increase in P. berghei oocyst numbers [19]. However, in a susceptible line of An. gambiae (G3), SRPN6silencing has no effect on the number of developing parasites [19]. Further experiments would suggest that AgSRPN6 mediates parasite recognition and/or lysis and may additionally be involved in the regulation of the melanization response [19,21], yet the precise function of SRPN6 in the mosquito immune response is still unknown.
In both An. stephensi and An. gambiae, SRPN6 is induced in response to Plasmodium ookinete invasion of the midgut [19,22]. However, induction is more pronounced following P. berghei than P. falciparum infection with both mosquito species [19,22]. The reasons for this difference remain unclear. Here we show that AsSRPN6 can also be induced by certain bacteria, but only through contact with the lumenal surface of the midgut epithelium. Our experiments indicate that E. cloacae strongly induces AsSRPN6 expression in the midgut shortly after feeding establishing this gene as an important component of the E. cloacaemediated response that effectively inhibits P. falciparum development in An. stephensi.

AsSRPN6 Expression is Differentially Induced by Bacteria
Previous experiments have demonstrated that SRPN6 expression is induced by Plasmodium ookinete invasion of the mosquito midgut [19,22] and sporozoite invasion of the salivary glands [20]. To investigate whether bacteria also play a role in AsSRPN6 activation, we measured changes in gene expression after feeding An. stephensi mosquitoes with different species of bacteria ( Figure 1A). These experiments revealed that the intensity of AsSRPN6 expression is differentially regulated by different bacterial species. AsSRPN6 expression was strongly induced by most gram-negative bacteria examined, while gram-positive bacteria produced much weaker expression. These results agree with previous results that SRPN6 was primarily induced by gramnegative bacteria [9]. Interestingly, AsSRPN6 expression was independent of the anti-microbial protein (AMP) gambicin [23], which is strongly expressed in the mosquito midgut and induced by most agents, including LPS and gram-positive bacteria but not by buffer or P. berghei parasites ( Figure 1B). These observations suggest that multiple pathways may contribute to the midgut immune response to produce specific responses toward endogenous gram-negative bacteria and invading pathogens.

AsSPRN6 Expression Requires Signaling via the Lumenal Surface of the Midgut Epithelium
The midgut epithelium consists of a single cell layer. On the lumenal side, epithelial cells display an extensive network of protrusions (microvilli), while the basal side is made of a complex invagination network (basal labyrinth) and a basal lamina exposed to the circulating hemolymph. The experiments outlined above examined AsSRPN6 immune activation following bacteria exposure to the lumenal side of the epithelium via feeding. To investigate the signaling requirements needed for AsSRPN6 activation, we sought to determine if bacterial exposure to the basal midgut surface was also capable of AsSRPN6 induction. Following the injection of bacteria into the mosquito hemocoel, the levels of AsSRPN6 expression were analyzed in midgut and carcass tissues ( Figure 2). AsSRPN6 transcript was undetectable in midgut samples following injection (Figure 2A), and only weakly detected in carcass samples ( Figure 2B). These results suggest that activation of AsSRPN6 midgut expression occurs via specific signaling mediated by the interaction of bacteria with the lumenal midgut surface.
As a control, carcass samples from bacteria-injected mosquitoes were also analyzed for defensin expression, a potent anti-microbial protein induced in the mosquito fat body [24]. While AsSRPN6 and defensin were both weakly upregulated by injury (buffer injection), only defensin (and not AsSRPN6) was strongly upregulated by both E. coli and M. luteus ( Figure 2C). These results suggest that the weak AsSRPN6 expression in the carcass samples may be in response to injury and is independent of the presence of bacteria ( Figure 2B).

Different Patterns of AsSRPN6 Induction by Bacteria and P. berghei
To better understand the mechanism of AsSRPN6 induction by gram-negative bacteria, we compared the time course of AsSRPN6 expression with that of P. berghei, which served as a positive control. Bacterial induction of AsSRPN6 was rapid as mRNA abundance and protein expression reached peak levels at ,6 h after bacteria feeding ( Figure 3A and 3C). In contrast, peak expression after a P. berghei-infected blood meal occurred at ,24 h ( Figure 3A). This difference is likely explained by the timing of the physical interactions between the inducing agent and the midgut epithelium. Following bacteria feeding, the contact between bacteria and the midgut epithelium likely occurs soon after ingestion. However, the ingestion of a blood meal triggers the formation of the peritrophic matrix, a chitinous acellular layer that completely surrounds the blood bolus, physically separating it from the midgut epithelium [25]. Not until Plasmodium ookinetes differentiate, traverse the peritrophic matrix, and invade the midgut epithelium are the presumed signals leading to AsSRPN6 activation initiated.
This interpretation implicating the role of the peritrophic matrix in preventing immune recognition is consistent with the spatial pattern of SRPN6 expression by immunofluorescence. As shown in Figure 3B, bacteria feeding trigger a generalized activation of AsSRPN6 protein accumulation across the majority of the midgut epithelium. This is in contrast to Plasmodium infection, where AsSRPN6 and AgSRPN6 protein seems to accumulate only in invaded cells [19].
In addition, the mechanistic role of AsSRPN6 as a putative serine protease inhibitor may differ in its response to bacteria or Plasmodium parasites. AsSRPN6 is detected in both its native form and in a higher molecular weight complex following E. cloacae feeding ( Figure 3C). These experiments suggest that AsSRPN6 may form a covalent complex with a serine protease (of ,25 kDa), similar to the inhibitory complexes formed by other mosquito serpins [26]. This is in contrast to the degraded or processed forms of AsSRPN6 detected following Plasmodium infection [19], and may imply that AsSRPN6 has alternative functions in the mosquito immune response to different pathogens. Alternatively, these differences could also be explained by digestive enzymes present within the blood meal or by direct interactions with different pathogen-specific proteases. One may speculate that the ,80 kDa complex that we identified following E. cloacae infection is similar to the recombinant AgSRPN6 complexes produced in vitro with hemolymph proteases from the lepidopteran insect, Manduca sexta [21]. However, further experiments must be performed to identify the target proteinases of AsSRPN6 and AgSRPN6 in vivo.

E. cloacae Inhibits P. falciparum Development in An. stephensi
Previous reports have demonstrated that the presence of bacteria within the mosquito midgut greatly influences Plasmodium development [8,9,12], and that field isolates of an Enterobacter species have a profound effect on P. falciparum development [8]. Our results (Figures 1 and 3) show that E. cloacae strongly induces AsSRPN6 expression (and other components of the mosquito immune response) in the mosquito midgut. To determine the effects of our Enterobacter strain on parasite development, we fed An. stephensi mosquitoes on a P. falciparum gametocyte culture mixed with E. cloacae bacteria [8].  The bacteria caused a dramatic decrease in the number of parasites that developed to the oocyst stage, as compared to control mosquitoes ( Figure 4A). In addition, there was a significant decrease in prevalence (percent mosquitoes that were infected) that accompanied this reduction in oocyst numbers ( Figure 4B). To determine whether this decrease is due to interference of Plasmodium development prior to midgut invasion similar to an Enterobacter strain described in Cirimotich et al. [8], we measured the effect of bacteria on the formation of mature ookinetes. As shown in Table 1, the presence of bacteria up to a concentration of 10 6 /ml had no effect on the development of P. falciparum gametocytes to mature ookinetes in the mosquito midgut. These results suggest that E. cloacae interferes with ookinete invasion and traversal of the midgut or alternatively, with the differentiation of ookinetes into oocysts. This is in contrast with the pre-invasion phenotype described for a natural Enterobacter isolate [8], thus highlighting the differences in the inhibitory mechanisms against Plasmodium development even between bacteria of the same genus. However, considering that these experiments were performed in different mosquito species, we cannot rule out that the observed differences are due to differences in how An. stephensi and An. gambiae respond to bacteria rather than differences between Enterobacter species.
AsSRPN6 Contributes to E. cloacae Inhibition of P. falciparum Development in An. stephensi Bacteria induce a broad spectrum of immune responses in the mosquito midgut to promote resistance to malaria parasites [9,27,28]. As shown in Figure 3, E. cloacae strongly induce AsSRPN6 expression in the mosquito midgut. Since SRPN6 activity has previously been implicated in the mosquito anti-Plasmodium immune response [19,20], we wanted to evaluate the contribution of AsSRPN6 toward this bacterial-induced resistance. To investigate the role of AsSRPN6 in the bacteria-induced inhibition of P. falciparum development, three groups of mosquitoes were fed as follows: (i) naïve mosquitoes fed with an infectious P. falciparum gametocyte culture; (ii) dsGFP-injected control mosquitoes fed with the same gametocyte culture mixed with E. cloacae; and (iii) dsSRPN6-injected experimental mosquitoes fed with the same gametocyte culture mixed with E. cloacae. Group (i) served as a reference, while group (ii) was used to confirm the inhibition of P. falciparum by the bacterium (as in Figure 4) and to control for any non-specific effects of dsRNA injection. Group (iii) was to assess the role played by AsSRPN6, if any, in the inhibition by bacteria.
The efficiency of dsRNA-mediated silencing of AsSRPN6 was evaluated by RT-PCR ( Figure 5A). Similar to our experiments in Figure 4, the addition of E. cloacae significantly inhibited parasite development in both the dsGFP and dsSRPN6 ( Figure 5B), Enterobacter-Mediated Refractoriness to Plasmodium PLOS ONE | www.plosone.org although we cannot distinguish between the effects of E. cloacae feeding and dsRNA treatment. However, there was a significant increase in the number of P. falciparum oocysts that develop in the dsSRPN6 mosquitoes when compared to the dsGFP controls ( Figure 5B). Silencing of AsSRPN6 expression also resulted in an increased prevalence of infection ( Figure 5C). This would suggest that the loss of AsSRPN6 function following dsRNA-silencing partially rescues the inhibitory effects of E. cloacae that limit parasite growth. While other immune components are likely triggered by E. cloacae, our data suggest that inhibition of parasite development is partially mediated by AsSRPN6 function.

Discussion
Midgut invasion is a critical step in the propagation of many pathogens in the mosquito and essential for their transmission. Moreover, during midgut invasion parasite numbers are the lowest of their life cycle in the mosquito, making this stage a prime target for transmission control [1]. In recent years, progress has been made in outlining the mechanisms of the mosquito immune response against Plasmodium, but our understanding of this process remains incomplete. Here we report on experiments that provide new insights into the mechanisms of action of the immune protein AsSRPN6, and the interplay between the mosquito microbiota and the mosquito response to Plasmodium infection. Our findings demonstrate that AsSRPN6 is differentially activated by certain bacteria and that this activation is more efficient when bacteria are exposed to the lumenal surface of the midgut epithelium. In addition, we demonstrate the importance of bacteria in limiting parasite development through the activation of the innate immune response and find that the E. cloacae-mediated inhibition of P. falciparum development in An. stephensi is partially mediated by AsSRPN6 function.
Mosquito midgut epithelial cells serve as the first point of contact between ingested pathogens and their vector. These cells display an active immune response that limits parasite propagation [9,27,29]. Previous observations suggest that SRPN6 is an important component of the mosquito epithelial response to P. berghei [19], and this report shows that in addition to Plasmodium, AsSRPN6 is selectively induced primarily by gram-negative bacteria. Our experiments suggest that the gram-negative E. cloacae induces the highest levels of AsSRPN6 transcript, but the specific bacterial elicitors of this response are unknown. Recent  work suggests that AgSRPN6 may be induced by physical damage to the midgut epithelium as a result of P. berghei ookinete invasion [22], but it is unknown if E. cloacae is able to similarly damage the midgut epithelium. Other gram-negative bacteria have been shown to compromise the integrity of the Drosophila midgut epithelium through the production of pore-forming toxins, raising the possibility that AsSRPN6 induction mediated by E. cloacae may occur via similar mechanisms [30]. We have recently identified that a LITAF-like transcription factor (LL3) is involved in the transcriptional regulation of SRPN6 in response to Plasmodium invasion [22], but further experiments must be performed to determine if LL3 is involved in the regulation of the AsSRPN6 immune response to E. cloacae. It is unclear whether the mechanisms of SRPN6 activation are conserved between bacteria and Plasmodium parasites, yet our experiments demonstrate that AsSRPN6 immune activation by E. cloacae leads to cross immune protection against P. falciparum as previously described with [9,31].
To examine the signals that lead to AsSRPN6 immune activation, we determined that the injection of bacteria into the mosquito hemocoel had no effect on AsSRPN6 expression in the midgut. AsSRPN6 expression in the carcass was minimally activated as a result of wounding (physical damage resulting from injection), likely through expression in hemocytes or fat body. These observations suggest that the basal epithelial cell surface does not possess the appropriate sensing mechanisms (i.e.receptors) or that the basal lamina constitutes a barrier that prevents physical interaction between the bacteria and the basal midgut epithelial cell surface. In addition, it also implies that AsSRPN6 is not a major component of An. stephensi humoral immune response since bacteria introduced into hemocoel activate anti-microbial gene expression in the fat body [27], but not AsSRPN6. This would suggest that the ''natural'' evolutionary role of AsSRPN6 may be to attenuate the growth of endogenous flora and ingested bacteria within the midgut, rather than having a role in the systemic humoral response. Most likely, the required signaling needed for AsSRPN6 immune activation depends either on direct interactions on the lumenal surface or on intracellular signals produced in response of pathogen invasion.
Our data suggest a mechanism of inhibition that is dependent upon mosquito immune activation, mediated in part, by AsSRPN6 function. This activation occurs soon after feeding (,6 h post blood meal peak) much before parasite invasion of the midgut epithelium (,24 h post blood meal). SRPN6 and other immune genes may be activated through the direct interaction of E. cloacae with the midgut epithelium before the onset of ookinete invasion. It has been well documented that the mosquito immune system is capable of being primed by bacterial injection [31], or by the presence of the intracellular symbiont Wolbachia to confer resistance to Plasmodium development in mosquitoes [32,33]. Through priming of the immune response by feeding E. cloacae, midgut epithelial cells are ''loaded'' with immune proteins that also confer anti-Plasmodium properties, resulting in fewer parasites that are able to successfully develop in the mosquito host.
Data presented here raise the possibility that AsSRPN6 may be an integral component of the mosquito innate immune response to bacteria, in addition to its previously defined roles in anti-Plasmodium immunity [19,20]. As a predicted serine protease inhibitor, SRPN6 may interact with proteases that regulate mosquito immune signaling pathways [26,34] or through direct interactions with the pathogen as in other species [35,36]. Future studies aim to determine the molecular mechanisms leading to SRPN6 activation and understanding how expression of SRPN6 leads to inhibition of Plasmodium development by identifying its mode of action.
In summary, our data demonstrate that lumenal exposure to E. cloacae mediates an anti-Plasmodium response in An. stephensi that is partially dependent on AsSRPN6 function. Moreover, our results support applications that rely on the midgut microbiome of anopheline mosquitoes (natural isolates or genetically engineered) to interfere with the transmission of malaria [8,37].

Ethics Statement
This project was carried out in accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The animal protocol was approved by the Animal Care and Use Committee of the Johns Hopkins University (protocol number M009H58). Anonymous human blood used for parasite cultures and mosquito feeding was obtained under IRB protocol NA 00019050 approved by the Johns Hopkins School of Public Health Ethics Committee. Informed consent is not applicable.

Mosquito Rearing and Plasmodium Infections
An. stephensi mosquitoes were reared under standard insectary conditions of 27uC and 80% relative humidity and maintained on 10% sucrose. P. berghei ANKA 2.34 was maintained and used to infect mosquitoes as previously described [38]. NF54 isolates of P. falciparum gametocyte cultures were obtained from the Johns Hopkins Malaria Research Institute Parasite Core facility. The culture was washed and brought up in normal human serum (Interstate Blood Bank, Memphis, TN) plus human RBCs to 45% hematocrit and 0.3% gametocytemia. Infective blood was placed into water-jacketed glass membrane feeders warmed to 37uC. Mosquitoes were allowed to feed for 20 min, and were maintained thereafter for 8 d in an incubator at 26uC and 80% humidity. Only midguts from fully gravid females were analyzed for Plasmodium infection by staining with 0.1% mercurochrome to count oocyst numbers.

Bacteria Feeding
Bacteria were washed in phosphate buffered saline (PBS) and suspended in latex feeding buffer [39] at a final concentration of 1610 6 colony forming units (cfu)/ml or as indicated. Adult mosquitoes were maintained from the time of hatching in sterile 10% sucrose solution containing 10 units/ml penicillin and streptomycin. Antibiotic solution was replaced by sterile sugar solution two days prior to the infectious blood meal. Mosquitoes were starved overnight and fed on medium or bacteria by use of water-jacketed glass-membrane feeders warmed to 37uC. Bacteria suspended in PBS at 10 6 cfu/ml were injected (,100 nl) into the mosquito hemocoel through the thorax. For mixed gametocyte/ bacteria feeding, E. cloacae was washed in ookinete medium [40] and then mixed with gametocytes to final 0.3% gametocytemia at the desired bacterial concentration.

SRPN6 Gene Expression Analysis
Northern blot analysis was performed as previously described [41]. Total RNA was isolated from 5-20 engorged female mosquito midguts or carcass (all tissues minus the gut) at selected time points following bacteria feeding or a parasite-infected blood meal with TRIzol reagent (Molecular Research Center Inc.). A mosquito mitochondrial rRNA gene was used as a loading control [41].
For quantitative RT-PCR, RNA samples were treated with DNase and reverse-transcribed with Superscript III (Invitrogen) using random hexamer primers. Real-time quantification was performed using the SYBR Green PCR master mix (Applied Biosystems) and ABI Detection System ABI Prism 7000 (Applied Biosystems). All PCR reactions were performed in triplicate. Specificity of the PCR reactions was assessed by analysis of melting curves for each data point. The ribosomal protein S7 gene was used for normalization of cDNA templates as described previously [19].

Immunofluorescence and Immunoblotting
An. stephensi midgut sheets were prepared from buffer-fed or E. cloacae-fed mosquitoes 6 h after a bacterial meal and stained by immunofluorescence as previously described [42]. Midgut sheets were incubated with affinity-purified SRPN6 antibody (1:1000) raised against the full-length An. gambiae SRPN6 protein [19] and SRPN6 was detected with Alexa FluorH 488-labeled goat antirabbit secondary antibody (1:1000, green, Molecular probes). Cell nuclei were stained with DAPI (Roche Applied Science).
For immunoblotting, 10 midgut sheets were prepared from An. stephensi mosquitoes fed on an E. cloacae bacterial meal were placed in 70 ml of 16Laemmli buffer and boiled for 5 min. The equivalent of 2 midgut sheets were separated by 12% SDS-PAGE and transferred to a PVDF membrane. The membrane was incubated with affinity-purified anti-SRPN6 antibody (1:10,000) and the bound antibody was detected with a horseradish peroxidase-linked anti-rabbit IgG (Pierce, 1:25,000 dilution) by exposing the blots to X-ray films. Membranes were stripped by two 30-minute washes in 100 mM 2-mercaptoethanol, 2% (w/v) SDS, 62.5 mM Tris-HCl, and pH 6.7 at 50uC. The stripped membranes were incubated with actin antibody (1:500; A2066, Sigma) for assessment of the relative amount of protein analyzed in each lane.

Detection of P. falciparum Ookinetes
To determine the number of midgut ookinetes, mosquito midguts were dissected in PBS 20 h post gametocyte feeding. Each gut was suspended in 100 ml of PBS, homogenized gently and 5 ml of the homogenate was spotted on a glass slide and air dried. P. falciparum ookinetes were detected by immunofluorescence as described above using an anti-Pfs25 monoclonal antibody (1:1000) and detected with Rhodamine red TM -X-labeled goat anti-mouse secondary antibody (1:1000, Molecular probes).

RNA Interference
A 785 bp (nucleotide 709 to 1693) region of AgSRPN6 cDNA (AGAP009212-RA) was amplified using T7-promoter-flanked primers (F: 59 T7-AGCCCCAGCTCGAGTGTGGTG-39 and R: 59T7-AAATAACGAGACGCGTCAGAAGTA-39). Double stranded RNA (dsRNA) was synthesized using the MEGAscript RNAi kit (Ambion) and purified according to the manufacturer's instructions. Approximately 300 nl of dsRNA targeting SRPN6 or GFP (control) solution at 3 mg/ml was injected into adult female mosquitoes using a nanojet injector as previously described [43]. Two days later, mosquitoes were fed on a P. falciparum gametocyte culture mixed with E. cloacae. To determine silencing, SRPN6 expression was analyzed by semi-quantitative RT-PCR approximately 6 h after an E. cloacae bacterial meal.