Distinct Olfactory Signaling Mechanisms in the Malaria Vector Mosquito Anopheles gambiae

A combination of gene silencing and behavioral studies in the malaria vector mosquito Anopheles gambiae sheds light on the olfactory basis of DEET repulsion as well as reveals the role of another family of chemosensory receptors that facilitate olfaction in An. gambiae.


Introduction
Chemosensory cues play a central role in directing much of the behavioral repertoire and are a significant determinant in the vectorial capacity of female An. gambiae mosquitoes, which are responsible for the transmission of human malaria [1]. Significant progress has been made in identifying the components of olfactory pathways in An. gambiae [2,3,4,5,6]. Nonetheless, there is a paucity of information regarding the precise molecular mechanisms that mediate olfactory signaling in An. gambiae.
At the center of the peripheral olfactory signal transduction pathway in An. gambiae is a family of odorant receptors (AgORs) that are selectively expressed in olfactory receptor neurons (ORNs). Although originally identified as candidate G-proteincoupled receptors (GPCRs) [7], several studies have disputed the GPCR nature of Anopheline and other insect ORs [8,9,10,11], which likely form ligand-gated heteromeric ion channels that activate ORNs through ionotropic [10] as well as perhaps metabotropic mechanisms [11]. In addition, members of a family of another set of chemosensory receptors related to ionotropic glutamate receptors have recently been described in Drosophila melanogaster [12].
The majority of insect ORNs typically express at least two ORs that are likely to form complexes of undetermined stoichiometry that are composed of one highly conserved non-conventional OR83b-like protein together with a conventional OR that presumably mediates odorant binding specificity [4,8,13]. In An. gambiae, 73 of the 79 AgORs originally identified [7] are expressed in the adult and 13 are expressed in larval stages [14]. The nonconventional anopheline OR83b-like family member, AgOR7, is widely expressed in nearly all olfactory sensilla with the notable exception of grooved-peg sensilla [5], which are activated in vivo by compounds such as ammonia, lactic acid, and other carboxylic acids that are major components of human sweat [15,16] known to evoke physiological and/or behavioral activity in An. gambiae [17,18]. Indeed, recent functional analyses of AgOR odor space reveal a paucity of responses for these groups of odorants, suggesting anopheline sensitivity to amines and other variant odorants may lie outside of AgOR-based signaling [17,18].
In order to improve our understanding of mosquito olfaction, we have continued to utilize the relative simplicity of the An. gambiae larval olfactory system, which consists of only 12 ORNs [14]. In previous studies utilizing behavioral and functional approaches to describe the molecular and cellular basis for olfactory responses to a range of natural and synthetic chemical stimuli, we identified a subset of AgORs expressed in the larval antenna that are tuned to odorants that elicit specific behavioral responses [14]. Building upon those studies, we now use RNAibased gene-silencing approaches to validate in vivo the role of AgORs in larval olfactory signal transduction and specifically identify the molecular receptor that mediates the repellent activity of N, N-diethyl-m-toluamide (DEET). In addition, we have identified and characterized a family of chemosensory receptors that are related to inotropic glutamate receptors (AgIRs) that underlie a novel-signaling pathway that is independent of AgOR activity. We propose that An. gambiae expresses distinct signaling pathways that participate in larval olfaction and are likely to also be active in mediating adult responses to a diverse range of chemosensory stimuli. These studies further our understanding of the molecular basis of olfaction and olfactory-driven behaviors in An. gambiae and lay the foundation for advancing alternatives to mosquito control strategies focused on adult life stages.

Mosquito Rearing
An. gambiae sensu stricto, originated from Suakoko, Liberia, was reared as described [2]. For stock propagation, 4-to 5-d-old female mosquitoes were blood fed for 30-45 min on anesthetized mice, following the guidelines set by Vanderbilt Institutional Animal Care and Use Committee.

Individual Larval Behavioral Assays
Larval assays were conducted between ZT2 and ZT10 during the standard LD12:12 rearing cycle. Here, An. gambiae 2 nd or 3 rd instar larvae were removed from rearing pans, rinsed carefully with distilled water to eliminate any remaining food residue, and kept in segregated containers with distilled water for 30 min.
Odorant stocks were made by dissolving odorant (.99% pure or of the highest grade commercially available) in pre-heated (70uC) 2% NuSieve, GTG low-melting-temperature agarose (Cambrex Bio Science). The assay was performed in a 1061.5 cm Petri dish containing 50 ml of 27uC distilled water. The odorant and larva dropping spots were located at opposite ends along the diameter and marked by a solid circle and a cross, respectively. The odorant/control stock was placed into the dish for 1 min beforehand to equilibrate, and the larva was gently introduced at the marked spot.
Real-time images of larval movements were obtained and downloaded at 1 s intervals for the duration of the 5 min assay using a custom-designed 30 frames/s video camera/computer/ software system (Model NC-70, DAGE-MTI, Michigan City, Apple PowerMac 8500/Scion Image J v1.63, National Institutes of Health, USA). At the conclusion of each assay, all larvae were individually stored at 280uC for molecular analyses, as described below. The images were subsequently sorted and analyzed using Image J (version 1.40g, NIH, USA) with its Mtrack J plug-in (version 1.3.0). The analysis of larval responses was carried out by tracking the motion of individual larva after marking the position of the larva's anterior, which was easily discernable in our system. In this manner, we were able to monitor and calculate the number of larval turns, overall movement, resting time (s), and average velocity (mm/s) to provide a comprehensive characterization of larval behavior patterns. Similarly, a turn threshold was defined such that if the intersection angle between two successive larval tracking vectors exceeded 45u, the larvae were considered to have carried out a turn ( Figure 1). Similarly, movement thresholds were defined so as to recognize false movements and account for the tendency of An. gambiae larvae to stochastically perform body swirls that appear to lack any horizontal locomotion. In our hands, a movement threshold was set by establishing that an individual larva turns 90u relative to an axis set at the body-length midpoint; the distance between the previous and the current position of the larval head can be calculated using the equation: BodyLength ffiffi ffi 2 p . By setting the movement threshold in such a manner, we were able to compensate for false movements that result from the tendency of An. gambiae larvae to stochastically perform body swirls that appear to lack any horizontal locomotion. After measurement of multiple (n.30) stage-2 and stage-3 larvae, we calculated the average larval body length as ,3.25 mm in our CCD system, thereby establishing a threshold for larval movements at ,2.3 mm, such that any shift in larval head position exceeding this value was defined as a single instance of larval movement ( Figure 1). In addition to analyzing tracking data for the number of movements and turns, we also measured the average velocity (mm/s) and resting time (s) over the course of the entire assay. Arithmetic means for each assay/treatment were analyzed for statistical significance using single-factor ANOVA; significant results were followed up with Tukey-Kramer post-tests to distinguish among groups using JMP software (v. 4.0.4, SAS, Cary, NC). In the cases where antennal and maxillary palp ablations of larvae were conducted, all manipulations were carried out by manual dissection at 2 nd instar stages, after which larvae were allowed to recover for 24 h prior to behavioral testing.

AgIR Identification and Expression
Candidate AgIR sequences were identified in both the An. gambiae genome using DmIR amino acid sequences as tBLASTn and BLASTp queries, respectively. Potential exon-intron gene models were predicted based on homology to DmIRs or AgIRs, as well as with the aid of a Hidden Markov Model-based gene

Author Summary
Anopheles gambiae, the principal Afrotropical mosquito vector for human malaria, uses olfaction to respond to chemical cues that are required for feeding, host preference, and mate selection. At the heart of this process is a large family of An. gambiae odorant receptors (AgORs) that respond to these olfactory cues in peripheral sensory neurons. We have now taken advantage of the relative simplicity of the larval olfactory system to develop a novel behavioral paradigm for individual larva that has been used together with gene silencing approaches in vivo to examine directly the role of AgORs in the olfactory system of An. gambiae. In addition to supporting and extending previous studies on the general olfactory response properties of AgORs, our data directly implicate the activity of specific AgORs in mediating the behavioral responses to the commercial insect repellent DEET and reveal the existence of at least two distinct olfactory signaling pathways in An. gambiae. One system depends directly on AgORs, while the other is AgOR-independent and requires the expression and activity of a newly identified family of candidate chemosensory genes, the An. gambiae variant ionotropic receptors (AgIRs). In addition to clarifying the mechanistic basis for olfaction in this system and the basis of DEET repulsion, these advances may ultimately enhance the development of vector control strategies, targeting olfactory pathways in mosquitoes to reduce the catastrophic effects of malaria and other mosquito-borne diseases.

siRNA Preparation and Injection
Double-stranded (ds) RNAs against a specific target gene were prepared and purified using bidirectional in vitro transcription of full-length cDNA templates using flanking T7 transcription initiation sites, and siRNAs were prepared via RNAse III digestion using Silencer siRNA Construction reagents and protocols (Applied BioSystems/Ambion, Austin, TX). Healthy, wild-type 2 nd instar An. gambiae larvae were chosen for micro-injection. They were pre-immobilized on 3MM filter paper on top of a 4uC chill platform (BioQuip Inc, Rancho Dominquez, CA). Additional desiccation was achieved using Kimwipes (Kimberly-Clark, Dallas TX) to gently dry individual larva. Twin styrofoam strips were also employed as temperature sinks to reduce distress from cold temperatures. Single barrel borosilicate glass capillary pipettes (World Precision Instruments, Sarasota, FL) were pulled (using a P-97 puller, Sutter Instruments, Novato, CA) and beveled (using a Narishige EG-5 beveller, Tokyo, Japan) to form microinjection needles. For larval microinjection, 27.6 nl of 100 nM siRNA were injected into the dorsal side of the larval thorax using a Nanoliter 2000 system (World Precision Instruments, Sarasota, FL). Postinjection, larvae were allowed to recover in 27uC distilled H 2 0 with 1 ml of larval food (as described in Mosquito Rearing section) for 48 h. Larvae were monitored every 24 h post-injection, and nonviable individuals were discarded.

Real-Time PCR (qRT-PCR)
Subsequent to experimental treatments and behavioral assays, AgOr7, AgOr40 and AgIr76b transcript levels were determined by means of quantitative RT-PCR. Each sample was comprised of 10 (AgOr7) or 30 (AgOr40, AgIr76b) larval heads that were handdissected from batches of control and experimental An. gambiae larvae. RNA extraction and cDNA synthesis were performed using the QIAGEN RNeasy Mini Kit and Roche Transcriptor First Strand cDNA Synthesis Kit, respectively. All primers in the assay were designed to span predicted introns in order to distinguish well between genomic DNA and cDNA templates. An. gambiae ribosomal protein S7 (rps7), which is constitutively expressed at high levels in all tissues, was chosen as control gene to measure the relative levels of mRNA of target genes in vivo. Primer sequences are as follows: rps7, f59-GGCGATCATCATCTACGTGC-39 and r59-GTAGCTGCTGCAAACTTCGG-39 (product size: 458bp cDNA); AgOr7, f59-ATCTTTGGCAATCGGCTCATC-39 and r59-GGCTCCAAGAACCGAAGC-39 (product size: 346 bp cDNA); AgOr40, f59-GACCCTCAAGAACCAGGGCT-39 and r59-AATGATGGTGTAGTACGAGAAGG-39; AgIr76b, f59-AT-CTTCGATCCAGAGTTGCT-39 and r59-CCGGTCACCAT-GACGAAGTA-39. qRT-PCR was carried out using an Applied Biosystems 7300 Real-time PCR system and SYBR green as fluorescent dye. Three experimental repetitions were analyzed for each biological sample and the data processed using System 7300 Sequence Detection Software (version 1.3.1). Primer efficiency was determined using a standard curve for all the primers used. In the amplification of target genes and Rps7, 8 ml and 2 ml cDNA, respectively, from each group were used as templates. In each trial, cDNA levels of target genes were quantified relative to rps7 levels using the method of Pfaffl [20].

Behavioral Responses of Individual Larva
Previous studies utilized a novel paradigm to assay the behavioral responses of large groups of An. gambiae late instar larvae to various natural and synthetic odorants in order to characterize the molecular and cellular elements of the larval olfactory system [14]. While providing fundamental information about the components underlying the olfactory responses of An. gambiae larvae, these end-point studies did not provide the precise tracking information that would allow us to distinguish between attractive or repulsive behavioral patterns. In addition, the need for a large number of larvae precluded its use in other experimental contexts. To provide such information and utility, a CCD camera-based tracking system was utilized to study the behavior of individual An. gambiae larva in response to odorant stimuli. Visual tracking records ( Figure 2) were then analyzed to distinguish parameters associated with directional movement. These included calculating the total number of turns, the overall number of movements, the average velocity, and the resting time for each larval behavioral assay (Figures 2 and 3).
The sensitivity of this system was initially tested with two odorant stimuli, each of which evoked a strong dose-dependent response in the An. gambiae larvae group assay paradigm [14]. The first was DEET, which is a widely used commercial insect repellent. The second was yeast paste, a complex odorant source and a normal component of larval food [14]. The behavioral responses of individual An. gambiae larva to three concentrations of DEET and two concentrations of yeast paste were examined along with the appropriate set of parallel no-odorant controls ( Figure 2). For each assay, the four behavioral parameters described above were quantified. In these studies, yeast paste elicited decreases in overall larval turning (inverse klinokinesis; Figure 2) and movement ( Figure 3) as well as concomitant increases in resting time when compared with no-odorant controls. In contrast, DEET elicited nearly the opposite effect: An. gambiae larvae displayed a dose-dependent increase in the turning rate (direct klinokinesis; Figure 2), number of movements, and average velocity (direct orthokinesis; Figure 3), while the average resting time was reduced to threshold levels at dilutions of 10 23 and 10 24 .
To confirm that the odorant-evoked behavioral responses were mediated by the larval olfactory system, a parallel set of assays were carried out after hand dissection of both larval antennae to effectively eliminate the site of olfactory signal transduction. Antennal-ablated larvae appeared to be largely indifferent to high concentrations of both DEET and yeast, as larval responses were indistinguishable from no-odorant and unablated controls ( Figure 4). In larvae in which the antennae were left intact but maxillary palps removed, responses to DEET and yeast paste were similar to those in unablated controls ( Figure 4). Taken together, these data demonstrate that we have developed a robust behavioral paradigm for examining odorant-induced responses from individual An. gambiae larva.

AgORs Silencing Confirms a Direct Role in the DEET Response
To discern the molecular basis for odorant-evoked behavioral responses of An. gambiae larvae, we initially focused on the role of AgOr7, which is the An. gambiae ortholog of the non-conventional Drosophila OR, DmOr83b [5,7], and is highly expressed in the larval antenna [14]. In the absence of effective strategies to generate mutant or transgenic strains of An. gambiae, we used RNA interference (RNAi) to reduce AgOr7 mRNA levels in individual larva, which could then be tested for abnormal behavioral responses. Individual larval behavioral assays followed by quantitative RNA analyses were conducted to assess the effects of AgOr7 siRNA and control siRNA microinjections on olfactory responses and transcript levels. To account for non-specific effects of siRNA delivery, larvae were microinjected with identical amounts of a siRNA designed against a gene (AT5G39360) from the Arabidopsis thaliana genome lacking significant homology to any cDNA in An. gambiae. Furthermore, buffer-alone microinjections were carried out in parallel to assess any potential effects of microinjection on larval behavior.
In order to assess the efficiency of siRNA-mediated knockdown of AgOr7 transcripts, a series of qRT-PCR studies were carried out on experimental and control larvae after behavioral testing. In these assays, cDNA was prepared from larval heads (with olfactory antennae attached) from individual larva collected immediately  following behavioral testing. These data ( Figure S1) confirm that microinjection of siRNAs targeting AgOr7 resulted in dramatic decreases in levels of this transcript.
Although a modest microinjection effect was observed on the average larval velocity, the overall number of turns ( Figure 5) as well as the number of movements, average velocity, and resting time ( Figure 6) in response to 1.6 mg/ml yeast paste stimuli were largely unaffected by microinjection with AgOr7 or control siRNAs. In contrast, a 1610 23 (v/v) dilution of DEET in individuals that received AgOr7 siRNA showed significant (p,0.01) reductions in turns ( Figure 5), movements, and velocity as well as a significant increase in their average resting time relative to buffer-injected and control larvae (Figure 6). Although a modest microinjection effect was again observed in buffer-injected larvae, these results are consistent with the hypothesis that larval responses to DEET are AgOr7-dependent whilst larval responses to yeast paste are AgOr7-independent.
Functional studies using Xenopus oocytes [14] have previously identified AgOR40 as a conventional ligand-specific larval AgOR that responds to DEET stimulation and, by implication, is likely to be responsible for DEET-elicited behavioral responses in An. gambiae larvae. Inasmuch as the molecular basis for DEET mediated behaviors remains controversial, we tested this hypothesis by using siRNA-mediated gene silencing to examine whether knockdown of AgOr40 transcripts would also perturb behavioral responses to DEET and yeast paste. In these studies, injection of siRNAs targeting AgOr40 echoed the effects of AgOr7 siRNAs and showed a significant reduction in turns and other elements of larval behavior in response to DEET stimuli ( Figure 7A) and were unaffected in response to yeast paste ( Figure 7B). As was the case for AgOr7 silencing, qRT-PCR studies were carried out on experimental and control larvae after behavioral testing to assess the levels of AgOr40 transcripts. These data ( Figure S1) confirm that microinjection of siRNAs targeting AgOr40 resulted in dramatic decreases in AgOr40 transcript levels without significantly altering AgOr7 mRNA pools. Taken together, these data directly validate the role of AgOR40 as a DEET-specific conventional AgOR in the larval olfactory system of An. gambiae.

AgIRs Mediate AgOR Independent Olfactory Responses
Based on the AgOr7-independent response of larvae to yeast paste, we next investigated whether AgOr7-dependent andindependent olfactory signaling exists in An. gambiae larvae. In doing so, we considered that AgOr7 independence of the larval yeast response might, in part, reflect that yeast paste is a complex mixture, some components of which may activate AgOr7independent olfactory signaling pathways. In contrast, DEET is a unitary compound that specifically elicits AgOr-dependent behavioral responses in An. gambiae larvae and physiological responses in Xenopus oocyte-based AgOR functional assays [14].
To examine further the possibility that distinct signaling pathways are active in this system, we searched the An. gambiae genome for homologs of variant ionotropic glutamate receptors that have recently been shown to function as novel chemosensory proteins in D. melanogaster (DmIRs) [12]. We have identified a family of 46 An. gambiae variant ionotropic glutamate receptors, which we have named AgamGLUVIRs, and 9 homologs of ionotropic glutamate receptors, named AgamGLURs or AgamNM-DARs, all according to the convention established by the An. gambiae genome consortium (www.Vectorbase.org). For convenience we refer to the AgamGLUVIR genes as AgIrs and their conceptual peptide products as AgIRs. Another group of researchers has independently identified the same family of genes [21] and we have agreed with them on a unified nomenclature in order to avoid confusion in future publications. A listing of the entire gene family, their chromosome positions, and peptide sequences is given in a Table S1.
A phylogenetic reconstruction comparing the amino acid sequences of AgIRs and DmIRs shows deep branching and low bootstrap support for many of the implied relationships, reflecting the considerable sequence diversity between these proteins both within and across species (Figure 8). The most convincing relationships are observed within the iGluRs, suggesting conser- Figure 5. Differential sensitivity of larval responses in An. gambiae to siRNA-mediated knockdown of AgOr7 is odorant dependent. The average number of turns exhibited by uninjected larvae as well as those receiving mock (buffer-alone), non-specific, or siRNA injections in response to yeast paste and DEET were assessed independently over a 5 min time lapse. Larval responses to 1.6 mg/ml yeast paste were unaffected by any siRNA treatments (A) while larvae receiving AgOr7 siRNAs displayed significant reductions in turning rates in response to a 10 23 v/v dilution of DEET (B). Buffer and non-specific siRNA-injected animals displayed a comparable reduction of the number of turns (p,0.05). Results are shown as means 6 SE, n = 10. doi:10.1371/journal.pbio.1000467.g005 vation of function ( Figure 8). Very few strong homologs are observed between AgIRs and DmIRs. Despite their diversity, topology predictions indicate conservation of 4 hydrophobic stretches of amino acids that likely correlate to the transmembrane and pore regions (Figure 9) of known ionotropic glutamate receptors (for review see [22]).
Interestingly, two of the strongest AgIR homologs of DmIRs are found within the iGluR clade (Figure 8). AgIR25a shares 68% amino acid identity (84% similarity) with DmIR25a, and AgIR8a shares 42% identity (63% similarity) with DmIR8a, genes that are broadly expressed in coeloconic sensilla neurons in the third antennal segment of D. melanogaster [12]. These 2 peptides are also much longer, 891aa and 946aa, respectively, than other AgIRs (average length 664aa) and are closer in size to the iGluRs (avg. 974aa, including partial peptides). Moreover, AgIR25 has retained 2 of the 3 amino acids, an arginine and an aspartic acid ( Figure 9A), in positions that are known to be important for glutamate binding [22]. Importantly, some classes of NMDA receptors also lack the 3rd residue [22]. AgIR8a has potential glutamate-binding residues in all three conserved positions, while several other AgIRs, including AgIR76b, retain one or more ( Figure 9B). Most other AgIRs are divergent at those positions (unpublished data, Table S1).
As a first step toward characterizing the potential role of AgIRs in larval olfactory signaling, we carried out RT-PCR using cDNA derived from An. gambiae larval antennae and gene-specific primers to 5 AgIr genes. These studies indicated that multiple members of this class of candidate chemosensory genes are expressed in the larval antenna ( Figure 10) as 4 of the 5 AgIrs could be amplified from larval antennae. Additionally, expression of one member of the ionotropic glutamate receptor family, AgNMDAR2, was observed in larval antennae ( Figure 10). We expect future work to elucidate the expression profiles of all AgIrs in both the larval and adult olfactory tissues of An. gambiae.
In order to examine whether AgORs and AgIRs perform distinct functional roles in the olfactory system of An. gambiae, we carried out behavioral assays using two additional unitary odorants that have been used successfully in previous behavioral and functional studies [14]. The first was 3-methylphenol (3MP), which was shown to activate AgOR-dependent pathways and evoke robust behavioral responses in larvae [14]. In our current studies, larvae manifest dose-dependent reductions in turns and overall movement, as well as threshold-dependent increases in average resting time ( Figure 11). Furthermore, larval responses to 10 24 dilutions of 3MP were significantly altered in larvae injected with AgOr7 siRNA, whereas control or buffer-injected larval responses were statistically equivalent to uninjected control larvae (Figures 12  and 13A). AgOR40 is one of 3 larval AgORs with a demonstrated sensitivity to 3MP [14]. In that light, we also tested the ability of siRNA mediated silencing of AgOr40 expression to alter larval responses to 3MP-in these studies a marginal but not statistically significant effect was observed (unpublished data) that is consistent with the role of multiple AgORs in mediating larval sensitivity to 3MP.
The next set of studies employed butylamine, a unitary odorant which has been shown to activate grooved-peg ORNs in An. gambiae [23] and Culex quinquefasciatus mosquitoes [24]. As was the case for 3MP, uninjected An. gambiae larvae displayed robust dosedependent responses to butylamine (Figure 11). In contrast to the AgOr7-dependent nature of larval responses to 3MP, larval responses to butylamine were indistinguishable among animals treated with AgOr7 and control siRNAs or microinjected with buffer alone (Figures 12 and 13B).
Based on their homology to DmIRs, which have been shown to mediate responses to amines and other odorants in Drosophila [12], we postulated that AgIRs mediate larval responses to butylamine. To test this hypothesis, siRNA-mediated gene knockdowns were used in an attempt to silence larval AgIRs and subsequently examine the responses of larvae to butylamine. Of the AgIrs tested, microinjection of only one-AgIr76b-displayed siRNA-specific effects on larval responses to butylamine. Microinjection of AgIr76b siRNAs reduced AgIr76b mRNA levels Figure 6. Larval behaviors after injection of non-specific small interfering RNA (siRNA). Averaged responses of buffer, non-specific, and AgOr7 siRNA-injected larvae in the presence of 1.6 mg/ml yeast paste and a 10 23 v/v dilution of DEET. Larval movement (A), velocity (B), and resting time (C) behaviors of larvae in response to yeast paste and DEET. Knockdown of AgOr7 mRNA levels has no effect on the ability of larvae to respond to yeast paste yet evokes significant behavioral alterations in larval responses to DEET (p,0.01). Results are shown as means 6 SE, n = 10. doi:10.1371/journal.pbio.1000467.g006 Figure 7. Differential sensitivity of larval responses in An. gambiae to siRNA-mediated knockdown of AgOr40 is odorant dependent. Larval responses exhibited by uninjected larvae as well as those receiving mock (buffer-alone), non-specific, or siRNA injections in response to DEET (A) and yeast paste (B) were assessed independently over a 5 min time lapse. Larval responses to 1.6 mg/ml yeast paste were unaffected by any siRNA treatments while larvae receiving AgOr40 siRNAs displayed significant reductions in turning rates (top panel) in response to a 10 23 v/v dilution of DEET. Buffer and non-specific siRNA-injected animals displayed a comparable reduction of the number of turns (p,0.05). Larval movement, velocity, and resting time behaviors (from top to bottom) of larvae in response to DEET (A) and yeast paste (B) where knockdown of AgOr40 mRNA levels had no effect on the ability of larvae to respond to yeast paste yet evoked significant behavioral alterations in larval responses to DEET (p,0.01). Results are shown as means 6 SE, n = 10. doi:10.1371/journal.pbio.1000467.g007 Figure 8. AgIR/DmIR phylogenetic tree. Neighbor-joining tree based on amino acid alignments of AgIR and DmIR peptides. AgIR names are shown in bold type and DmIR names are shown in plain type. Black dots indicate branch points where bootstrap support is less than 50%. doi:10.1371/journal.pbio.1000467.g008 ( Figure S2) and led to significant alterations in larval responses to butylamine (Figures 12 and 13B). Larval responses to butylamine were unaffected in AgOr7 knockdowns and by microinjection of non-specific siRNAs or buffer-alone controls (Figures 12 and 13B).

Discussion
In the face of a dearth of traditional genetic tools and a robust transgenic capacity, the ability to carry out RNAi-mediated gene silencing on individual An. gambiae larva provides an opportunity to examine the molecular basis for olfactory driven behaviors in this disease vector. Furthermore, the relative simplicity of the larval nervous system provides a considerably more tractable model within a non-model system for understanding similar processes that are presumed to underlie chemosensory responses in adults that directly contribute to anopheline vectorial capacity.
In this study, we have developed a simple behavioral paradigm that can be used to track the olfactory responses of individual An. gambiae larva to a range of chemical stimuli. Overall, these data are consistent with the hypothesis that when larvae are exposed to a repellent compound, such as DEET, they exhibit an increased rate   of turning and a rise in overall movement and velocity. In contrast, an attractant such as yeast paste or 3MP leads to a reduction in the number of movements, turns, and average velocity while the average resting time is increased.
Together with gene-silencing approaches, we have employed a novel behavioral assay to provide compelling in vivo evidence that, for the first time, supports a direct in vivo role of AgORs in olfactory processes in An. gambiae. Furthermore, these studies go further to address the molecular mechanism responsible for DEET mediated repulsion of insects. Previous studies [25] suggesting that DEET's mode of action is to inhibit the activation of a subset of insect ORs that would otherwise be activated by attractants are in contrast to models that suggest DEET acts via direct excitation of OR-expressing ORNs that, in turn, evoke downstream behavioral repulsion. The excito-repellent hypothesis is consistent with our previous study on the larval olfactory system in An. gambiae [14] that showed robust DEET-mediated behavioral responses that correlated with a discrete population of larval ORNs co-expressing AgOR7/AgOR40 as well as specific DEET stimulation of Xenopus oocytes injected with AgOR7/AgOR40 cRNAs. This hypothesis is also supported by other studies that describe DEETmediated activation of a subset of ORNs in Culex mosquitoes [26] and more recent work in Aedes aeqypti suggesting that DEET sensitivity is a genetically determined characteristic affecting the functionality of discrete ORNs [27]. While the reduction in DEET-mediated repellent responses in larvae undergoing RNAi mediated silencing of AgOr7 is consistent with a general requirement for AgOR-based signaling, the similar effects of AgOr40 silencing specifically supports the role of both these molecular targets in mediating DEET repellency. That these behavioral effects were manifest by DEET alone, i.e. in the absence of any other stimuli, further validates our earlier study and supports a direct excito-repellent mechanism for DEET activity.
Lastly, these studies uncover the existence of at least two parallel chemosensory transduction systems in larval-stage An. gambiae that respond to distinct classes of odorant stimuli. One pathway, which is in keeping with the established literature for insect olfactory signal transduction, is based on the obligatory role of the nonconventional anopheline Or83b family member AgOr7, which acts together with other conventional AgORs in the formation of functional receptors. It is likely that AgOR-dependent signaling pathways impact responses to a wide range of odorant cues that play important roles in several aspects of anopheline behavior. These pathways are exemplified by the dramatic alterations in the DEET and 3MP responses of An. gambiae larvae after RNAimediated silencing of AgOr7 transcripts (Figures 5, 12). The other  pathway depends on the function of the AgIr gene family, which likely recognizes different odor classes than the AgOr pathway. Moreover, the similarities between AgIRs8a and 25a and iGluRs suggest that cellular receptors for glutamate in the antenna could act as a neuromodulator of ORN function. This hypothesis is consistent with the inability of AgIr25a siRNAs to alter larval behavioral responses to odors (unpublished data).
Recent functional analyses [17,18] of AgOR-based odor coding against a diverse panel of compounds suggest that, in An. gambiae, olfactory pathways respond to a wide range of odorant stimuli with particular affinity for heterocyclics and aromatics that are associated with human skin emanations [15,16]. These groups of odorants are thought to play essential roles in host-seeking, oviposition, and other behaviors that are critical for anopheline life cycles [28]. Coincidently, this AgOR-based odor space is characterized by sparse responses to the majority of acids, aldehydes, and esters that were tested in addition to being particularly devoid of amine-elicited responses [17,18]. This raised the suggestion that sensitivity to these classes of odorants might lie outside of AgOr-dependent olfactory signaling pathways.
We have identified several AgIrs that are expressed in larval olfactory tissues ( Figure 10) and have used RNAi-mediated gene silencing to demonstrate the role of one of these genes in mediating larval responses to the AgOR-independent odorant butylamine. Critically, while knockdown of AgIr76b specifically altered larval responses to butylamine, there was no effect on responses to two other unitary odorants that were dependent on AgOr7 expression. These data are consistent with the hypothesis that, in contrast to the AgOR-dependent sensitivity to 3MP, DEET, and a broad range of ''general'' odorants [17,18], anopheline responses to other odorants (e.g., butylamine) are mediated through AgIrdependent signaling. There is reason to assume that these parallel pathways persist through to adult An. gambiae where AgIrs are likely to be responsible for olfactory sensitivity to important human kairomones, such as ammonia and lactic acid that are known to activate ORNs in grooved peg sensilla [29] that are devoid of AgOR7 [5]. Indeed, we have observed expression of multiple AgIrs in adult olfactory appendages, supporting the hypothesis that this family of genes is involved in chemosensory signaling in adults (manuscript in preparation).
Current efforts are directed toward expanding our understanding of AgIr-based odor coding in An. gambiae. Improving our understanding of olfactory signal transduction in An. gambiae may lead to new opportunities to target olfactory mediated behaviors at the molecular level. In turn, this may reduce the vectorial capacity of An. gambiae and help reduce the transmission of malaria and other important human diseases. Figure S1 Quantitative analysis demonstrates significant transcript level reduction of AgOr7 and AgOr40 after siRNA treatment. Larval cDNAs for qRT-PCR were generated using equal amounts (2 mg for AgOr7 and 4 mg for AgOr40) of RNA extracted from hand-dissected larval heads from each injection treatment group, and three technical replicates were performed for each experimental group. AgOr7 and AgOr40 mRNA levels were quantified as fold-changes relative to Rps7 using the method of Pfaffl [20]. AgOr7 and AgOr40 levels are shown after normalization to buffer-alone controls in each of three experimental replicates. Histograms showing averaged AgOr7 and AgOr40 levels normalized to buffer-alone injection controls. Standard errors were 60.041 and 60.029 for non-specific and AgOr7 siRNA injections; 60.127 and 60.392 for non-specific and AgOr40 siRNA injections, respectively. Raw data from each qRT-PCR reaction indicating cycle-threshold (CT) and primer efficiency information for each technical replicate. Found at: doi:10.1371/journal.pbio.1000467.s001 (0.71 MB JPG) Figure S2 Quantitative mRNA analysis demonstrates significant transcript level reduction of AgIr76b after siRNA treatment. Larval cDNAs for qRT-PCR were generated using equal amounts (,3.5 mg) of RNA extracted from hand-dissected larval heads from each injection treatment group. Two independent biological replicates were performed, each consisting of three technical replicates for every experimental group. AgIr76b mRNA levels were quantified as fold-changes relative to Rps7 using the method of Pfaffl [20]. AgIr76b levels are shown as averaged values of both biological replicates after normalization to buffer alone controls in each of three technical replicates. Histograms showing averaged AgIr76b levels normalized to buffer alone injection controls. Standard errors were 60.04 and 60.003 for non-specific and AgIr76b siRNA injections, respectively. Raw data from each qRT-PCR reaction indicating cycle-threshold (CT) and primer efficiency information for each biological/technical replicate.