Aedes aegypti mosquitoes detect acidic volatiles found in human odor using the IR8a pathway

Summary

Mosquitoes use olfaction as a primary means of detecting their hosts. Previously, the functional ablation of a family of Aedes aegypti olfactory receptors, the Odorant Receptors (ORs), was not sufficient to reduce host-seeking in the presence of carbon dioxide (CO₂). This suggests the olfactory receptors that remain, such as the Ionotropic Receptors (IRs), could play a significant role in host detection. To test this, we disrupted the Ir8a co-receptor in Ae. aegypti using CRISPR/Cas9. We found that Ir8a mutant female mosquitoes are not attracted to lactic acid, a behaviorally active component of human sweat, and lack odor-evoked responses to acidic volatiles. The loss of Ir8a reduces mosquito attraction to humans and their odor. We show that the CO₂-detection pathway is necessary but not sufficient for IR8a to detect human odor. Our study reveals that the IR8a pathway is crucial for an anthropophilic vector mosquito to effectively seek hosts.

eTOC Blurb

Raji et al. show that Ae. aegypti mosquitoes use the olfactory coreceptor, IR8a, to detect humans and human odor. IR8a responds to acidic volatiles found in human odor including lactic acid, a component of human sweat. This study highlights the crucial role acidic volatiles play when mosquitoes target a host for blood-feeding.

Introduction

Anthropophilic female mosquitoes, such as Aedes aegypti and Anopheles gambiae, have a strong innate drive to find their human hosts and obtain blood meals, which are required for their egg production. As has been highlighted by the recent Zika outbreak, Aedes mosquitoes are efficient vectors for pathogen transmission because of their host-seeking behavior and susceptibility for infection by flaviviruses [1]. Female mosquitoes, like other haematophagous Diptera, integrate an array of sensory information to find their human hosts including carbon dioxide (CO₂), body odor, heat, moisture, and visual cues [2]. How these different cues are sensed to enable mosquito host-seeking has only begun to be understood. Among these attractive cues, human body odor is a complex blend of volatile chemicals that distinguishes us from other vertebrate hosts [3]. Skin microbiota plays a large role in generating the volatile compounds that attract mosquitoes to human sweat [4]. In Ae. aegypti and An. gambiae, human odors that elicit both electrophysiological and behavioral responses have been identified including ammonia, amines, carboxylic acids, lactic acid, ketones, sulfides, and 1-octen-3-ol [5–10].

Insects respond to volatile chemicals in the environment with a complex repertoire of olfactory receptors that are evolutionarily distinct from vertebrate olfactory receptors [11]. Two families of odor-gated ion channels have been identified in insects that respond to a diverse set of molecules, the odorant receptors (ORs) and the ionotropic receptors (IRs) [12–14]. In addition, there are also gustatory receptors (GRs) that are highly sensitive to CO₂ [15–17]. Odor-tuned ORs rely upon the obligate olfactory co-receptor Orco to form an odor-gated ion channel complex [18–20]. In Ae. aegypti, loss of orco results in a loss of electrophysiological responses to some but not all odorants [21]. In the presence of CO₂, attraction to human odor was not significantly different from wild-type controls in female Ae. aegypti orco mutants. However, in the absence of CO₂, orco mutants lose strong attraction to human odor. Ae. aegypti mutants lacking Gr3, a subunit of the heteromeric CO₂ receptor complex, show no electrophysiological or behavioral responses to CO₂ [22]. Behavioral analysis of Gr3mutants showed that CO₂ can gate multiple cues that are sensed by mosquitoes including heat and human odor.

Drosophila IRs are expressed in olfactory sensory neurons (OSNs) that are distinct from OR and GR lineages [23]. There are at least two IR co-receptors, IR8a and IR25a, and a putative third, IR76b [14,24]. These co-receptors form an odor-responsive ion channel with other odor-tuned IRs. For example, some Drosophila odor-tuned IRs require IR8a as a co-receptor, while some other odor-tuned IRs form a functional complex with IR25a and/or IR76b, but not all three co-receptors. Similarly, some OSNs have been shown to express both IR8a and IR25a protein [24]. There are 30 putative odor-tuned IRs expressed in the Ae. aegypti antennae that could potentially form an odor-responsive ion channel with any of the IR co-receptors [25]. The combinatorial pattern of expression observed in IRs may allow flexibility to respond to more diverse olfactory cues with fewer odor-tuned receptors than the Orco pathway [24], which encompasses at least 117 odor-tuned ORs [26].

Insect IRs have been reported to detect amines, aldehydes, ketones, and acids [10,14,23,27,28]. Since many of these compounds are not represented in the OR chemical space, which includes alcohols and esters, the IR odor-ligands are largely complementary and not overlapping with OR odor-ligands [23,29]. In Drosophila, Ir25a and Ir76b are necessary for odor-evoked electrophysiological responses to amines [30], whereas receptor neurons expressing Ir8a are tuned to volatile acids [24]. IR8a forms a functional complex with AgIR75k to elicit odor-evoked inward currents in response to carboxylic acids including heptanoic acid, octanoic acid and nonanoic acid [10]. Lactic acid is enriched in human skin emanations compared to other vertebrates and may be one of the cues that signals to the mosquito that the target host is a human [31]. How lactic acid is sensed in mosquitoes is unknown, but it is likely to be IR-dependent.

Whereas Drosophila Ir25a is involved in many functions, including olfaction, taste, hygrosensation, thermosensation, and attraction to CO₂ [24,30,32–40], Ir8a appears to function exclusively in detecting odors and is not necessary for attractive responses to CO₂ [24,27,28,40]. In Ae. aegypti, Ir8a expression is localized to the antennae and cannot be detected in other chemosensory tissues, whereas Ir25a is broadly expressed in multiple chemosensory tissues and three-fold more abundant in sugar fed female mosquito antennae than Ir8a [25]. Similarly, Ir8a transcript abundance was detected only in the antennae of An. gambiae adults [41,42]. These results make Ir8a a likely candidate receptor for odor detection during mosquito host-seeking behavior. Ir25a and Ir76b have broader expression patterns that are consistent with these receptors being involved in other sensory modalities in addition to olfaction [24,25,30,32].

Since host-seeking is not completely ablated in orco or Gr3 mutants [21,22], we reasoned that the IR olfactory receptors retained in these mutants are crucial for host-seeking. Here, we used the CRISPR/Cas9 system to disrupt Ae. aegyptiIr8a. We tested the relative contribution of Ir8a in human odor detection and its genetic interaction with other olfactory receptor pathways that have been previously implicated in Ae. aegypti host-seeking. We found that Ir8a mutants are not attracted to lactic acid, a behaviorally active component of human sweat, nor were they able to detect acidic components of human odor. When compared to wild-type controls in membrane blood-feeding assays, Ir8a mutants have reduced responses to human odor but not heat and CO₂. Ir8a mutants are also less responsive to humans and human odor than wild-type controls in uniport olfactometer assays. The genetic interactions of Ir8a and orco as well as Ir8a and Gr3 suggest a crucial role for CO₂ to sensitize mosquitoes to human odor and highlights the importance of human acidic volatile detection during mosquito host-seeking.

Results

Targeted mutagenesis of Ae. aegypti Ir8a, an antennal specific ionotropic receptor

Previous analysis of the neurotranscriptome of Ae. aegypti has suggested that Ir8a is expressed in the antennae and no other chemosensory tissues [25]. To confirm this result, we first asked if Ir8a expression can be detected in any other body tissues. Using quantitative RT-PCR analysis in wild-type mosquitoes, we compared the expression of the intact female to the head plus antennae, antennae, body minus the head, and head minus antennae. We show that Ir8a mRNA expression is enriched in the antennae and nearly undetectable in other tissues (Figure S1A). Given that olfaction is key for mosquitoes to host seek [2], and Ir8a acts as an obligate olfactory co-receptor that can complex with odor-tuned IRs that detect acids in Drosophila [24,28], we speculated that mutations in Ae. aegypti Ir8a should disrupt a distinct subset of IRs that are responsive to acidic volatiles in human odor.

To test this hypothesis, we generated targeted null mutations in the Ae. aegypti Ir8a gene using CRISPR/Cas9 RNA-guided gene editing [43]. We integrated two distinct donor DNAs to generate two independent alleles in the Ir8a locus at exon 2 and exon 3 (see methods). These two independent mutant alleles, Ir8a^dsRED (exon 2) and Ir8a^attP (exon 3), are predicted to produce truncated IR8a proteins that would eliminate Ir8a function. The sgRNA was designed to guide the Cas9 endonuclease to exon 2 of Ir8a and enable integration of the polyubiquitin promoter and dsRED fluorescent protein through homology dependent repair by way of a plasmid containing the promoter, fluorescent marker and SV40 terminator flanked by homologous DNA sequences surrounding the predicted cut site (Figure 1A). This insertion visually marked Ir8a^dsRED mutants with dsRED fluorescence and could be detected in the Ir8a locus by PCR using a primer that anneals to sequences in the donor construct homologous right arm in Ir8a exon 2 and a second primer that anneals outside the bounds of the donor construct in Ir8a exon 4 (Figure 1B & C, Table S1). Sanger sequencing of this PCR amplicon showed site-directed integration of the donor plasmid into the Ir8a locus as expected (Data S1 & Table S1). To generate the Ir8a^attP allele, an sgRNA, Cas9 mRNA, and a single stranded DNA oligo containing a 50 bp attP PhiC31 recombination site flanked by 75 bp of homologous sequence surrounding the CRISPR binding site were injected into preblastoderm embryos. However, we observed a 17bp deletion in the CRISPR site and 2bp deletion in the attP site which may reduce PhiC31-mediated recombination frequency at this site. Ir8a^attP sequencing data suggests that it was a null mutation, as the predicted mRNA from the Ir8a^attP allele would contain 20 stop codons (Data S2). To reduce the possibility of off-target effects, both alleles were outcrossed to the wild-type Orlando laboratory strain for five generations and then homozygosed.

Figure 1. Mutagenesis of the Ae. aegypti Ir8a locus, and behavioral phenotypic analysis of Ir8a mutant response to lactic acid.

(A) CRISPR sgRNAs designed to target the Cas9 nuclease to exon 2 and exon 3 of Ir8a. Integration location of the dsRED fluorescent marker and ssODN containing attP site into exon 2 and exon 3 of Ir8a respectively. (B) dsRED fluorescence in the Ir8a^dsRED/dsRED homozygous mutant (left) and wild-type (right) adult female mosquitoes. (C) PCR amplification from Ir8a exon 2 locus, note 1162bp amplicon from the wild-type Ir8a locus versus the larger 3483bp amplicon from Ir8a^dsRED/dsRED homozygous mutant female mosquitoes. (D) Uniport olfactometer assay designed to test behavioral response of Ir8a mutants to lactic acid. Mosquitoes that fail to leave the trap during the assay are scored as unactivated, whereas mosquitoes that leave the trap but remain in the main tunnel are scored as activated. Mosquitoes in the trap adjacent to the stimulus chamber are scored as attracted. (E) Percent female mosquitoes attracted to no odor (blank) and CO₂ alone (CO₂) as well as CO₂ and lactic acid in a uniport olfactometer. Lactic acid and CO₂ were co-presented to wild-type, Ir8a^dsRED/dsRED, and Ir8a^attP/dsRED female mosquitoes (right three panels). Trials with different letters are significantly different (p < 0.0001, n=8–10). Analysis was done using One-Way ANOVA with post hoc Tukey’s HSD test. See also Figure S1 & S2, Table S1 and Data S1 & S2.

We next investigated if Ir8a mRNA is disrupted in Ir8a mutants. Using quantitative RT-PCR, Ir8a mRNA in the homozygous Ir8a^attP/attP mutants was nearly undetectable in both sexes (Figure S1B & C). This suggests that the mRNA is degraded by non-sense mediated decay and that the Ir8a^attP/attP mutant is an RNA null. However, the Ir8a^dsRED allele showed over-expression of Ir8a mRNA in male and female mutants (Figure S1B & C). The increase in Ir8a transcript levels observed in the Ir8a^dsRED allele is likely the result of the insertion of the polyubiquitin promoter dsRed fluorescent protein expression cassette into the locus. The predicted Ir8a^dsRED transcript cannot produce a full length IR8a protein (Data S1). No behavioral phenotypic differences were found between the mosquitoes carrying these alleles suggesting that both Ir8a^attP and the Ir8a^dsRED alleles are null (see below).

A concern with the application of CRISPR/Cas9 mutagenesis is off target mutations which may alter the function of other genes [44]. To rule out non-specific behavioral defects in the Ir8a mutants, we assessed their fitness. Using a locomotor activity assay, we found no difference in activity in Ir8a mutants when compared to the wild-type and heterozygous controls in the number of times they moved past an infrared beam (Figure S2A & B) [21]. We further investigated if Ir8a mutants may have feeding defects using the capillary feeder assay (CAFE) (Figure S2C) [45,46]. In this assay, mosquitoes were allowed to feed through a calibrated capillary tube containing 10% sucrose solution. We quantified sucrose consumption by recording the change in volume relative to control vials without mosquitoes. We recorded no difference in feeding in Ir8a mutants when compared to the wild-type and heterozygous controls after 18 hours of ad libitum feeding, (Figure S2D). We also evaluated the ability of the Ir8a mutants to survive fasting. This was tested by sugar-starving mosquitoes with access to water [21]. Our results suggest that our Ir8a mutant males and females can resist starvation as well as wild-type and heterozygous controls. (Figure S2E & F). Taken together, these studies attest to the fitness of Ir8a mutants as well as their suitability for further behavioral analysis.

The IR8a pathway is required for sensing and responding behaviorally to acidic volatiles

Lactic acid has been previously identified to be a behaviorally active component of human sweat that has been used to lure mosquitoes into traps or attract mosquitoes in olfactometer assays, but olfactory receptors for the compound have not been identified [5,10,47,48]. Lactic acid by itself is not attractive to mosquitoes [5]. However, both lactic acid and CO₂ can synergize when presented together to elicit attraction [22,31]. Using a uniport olfactometer with carbon-filtered air flow, we tested the responses of wild-type and Ir8a mutant female mosquitoes to filtered air, CO₂, as well as lactic acid and CO₂ (Figure 1D & E). In order to control for recessive background mutations that may have occurred in either of the two lines, we tested Ir8a^dsRED/dsRED and heteroallelic Ir8a mutants (Ir8a^attP/dsRED). The responses of Ir8a^dsRED/dsRED and Ir8a^attP/dsRED mutants to lactic acid and CO₂ were not significantly different from each other, or the wild-type responses to CO₂ alone, or filtered air (blank) (Figure 1E).

To confirm that Ir8a mutants have lost the ability to detect lactic acid, we examined the electrophysiological responses of female mosquitoes to a panel of odors. Unlike Drosophila, the olfactory receptor expression pattern in mosquitoes has not been comprehensively mapped [49,50]. The locations of IR8a expressing olfactory receptor neurons in the antennae is unknown making single sensillum recordings difficult to perform. To overcome this limitation, we used electroantennogram (EAG) measurements (Figure 2), which record the average signal output from the an entire mosquito antenna for a given odor volatile [51]. The wild-type mosquitoes showed robust odor-evoked responses to all the odor panels tested (Figure 2A). However, Ir8a^dsRED/dsRED mutants are insensitive to the acidic components of human odor represented in the panel, including lactic acid (Figure 2B). We observed a similar lack of olfactory sensitivity in Ir8a^attP/dsRED heteroallelic mutants (Figure 2C). As shown on the representative current traces (Figure 2D), Ir8a^dsRED/dsRED and Ir8a^attP/dsRED olfactory receptor neurons show weak responses to acidic volatiles similar to the solvent response of wild-type controls.

Figure 2. IR8a olfactory receptor pathway is required for sensing acidic volatiles that are component of human odor.

(A) Electroantennogram (EAG) responses of wild-type Orlando strain, (B) I8a^dsRED/dsRED mutants, and (C) heteroallelic Ir8a^dsRED/attP to volatiles that are components of human odor. (D) Representative EAG traces of the wild-type and Ir8a mutant mosquitoes. This figure is represented by a mean-max box and whisker plot. The ends of the box are the upper and lower quartiles. The median is marked by a horizontal line inside the box. The whiskers are the two lines outside the box that extend to the highest and lowest EAG response. The odors marked in red are acidic volatiles while the odors represented in blue include alcohols, ketones, aldehydes, monoterpenoid and alkene. EAG responses represented in orange dots are significantly different from solvent control. Responses labelled in grey are not significantly different from the solvent control. Statistical analysis was done using a mixed-effects model with Dunnet’s multiple comparison test. Each column was compared with the mean of the solvent control column. EAG responses marked with asterisks indicate that it is significantly different from response to the control solvent (* = p <0.01; **= p <0.001; *** = p <0.0001; **** = p <0.00001).

Our results show that ¡r8a mutants have lost their behavioral and electrophysiological responses to lactic acid, and strongly suggests that IR8a is required to detect lactic acid. This was true of both the Ir8a^dsRED allele and the heteroallelic combination of the Ir8a^dsRED allele and the RNA-null Ir8a^attP allele, suggesting that both alleles retain no Ir8a gene function. In our study, Ir8a mutants are still able to detect other human volatile compounds including some alcohols, aldehydes and ketones suggesting that other olfactory pathways are not impaired in these mutants.

Ae. aegyptiIR8a pathway responds to human odor cues during blood-feeding

Using a membrane blood-feeding assay (Figure 3A), we assessed the responses of Ir8a mutant and control female mosquitoes to heat, CO₂, and human odor cues, by determining the percentage of females that would blood-feed [22]. While CO₂ and human odor can activate and elicit mosquito attraction towards a blood source, the temperature of the meal is crucial for feeding to occur [52]. Experiments carried out with wild-type mosquitoes showed a robust feeding response when all cues were present, but feeding was reduced when human odor, CO₂, or heat was removed (Figure 3B). In the presence of all three cues, we found that both the Ir8a^dsRED/dsRED and the heteroallelic Ir8a^attP/dsRED mutants exhibited reduced blood-feeding when compared to wild-type or heterozygous controls (Figure 3C).

Figure 3. Ae. aegypti IR8a pathway responds to human odor cues during blood-feeding.

(A) Illustration showing the membrane feeding assay to simulate female mosquito blood-feeding behavior. (B) Percent wild-type female mosquitoes responding to varying sensory cues (n = 5; p < 0.0001). (C) Percent female mosquitoes of indicated genotypes blood-feeding in the presences of CO₂, human odor and heat cues (p < 0.0001; n = 7–10). (D) Percent female mosquitoes of indicated genotypes blood-feeding in the presences of CO₂ and heat cues, without human odor (n = 7; p = 0.468). (E) Percent female mosquitoes of indicated genotypes blood-feeding in the presences of human odor and heat cues, without added CO₂ (n = 5 – 6; p = 0.855). (F) Percent female mosquitoes of indicated genotypes blood-feeding in the presences of human odor and CO₂ cues, without heat cue (n = 6; p = 0.944). For the dot plots in B-F, long lines represent the mean and short lines represent standard error. Statistical analysis was done using One-Way ANOVA with post hoc Tukey’s HSD test. Genotypes marked with different letters are significantly different. Human odor was collected from subject 1, see Table S2.

We attempted to determine the individual contributions of human odor, CO₂, and heat cues in the assay to the Ir8a mutant phenotype by eliminating one cue at a time. When only heated blood and CO₂ were used as attractants there was no significant difference among genotypes (Figure 3D). This suggests that Ir8a mutant females can respond similarly to these cues as wild-type females. When the CO₂ source was removed from the assay, the rate of blood-feeding was very low across all the genotypes (Figure 3E). Similarly, when the temperature of the blood was shifted from body temperature (37°C) to ambient temperature (26°C), the absence of the heat cue nearly eliminated blood-feeding across all genotypes (Figure 3F). The level of blood-feeding was very low in the absence of heated blood or CO₂, so it is difficult to discern genotypic differences. By comparing the results from assays that contain heated blood and CO₂ (Figure 3C & D), we hypothesize that the defect found in Ir8a mutants is due to human odor detection.

Ae. aegypti IR8a pathway is required to detect humans and human odor

We subsequently investigated if the electrophysiological and behavioral defects recorded in Ir8a mutants could translate into impaired responses to human hosts. Using a uniport olfactometer with carbon-filtered air flow and added CO₂, we examined the attraction rate of olfactory receptor mutants to 15 different human subjects (Figure 4A, Table S1). Each subject was tested twice with wild-type (+/+), heterozygous (Ir8a^dsRED/+), Ir8a mutant (Ir8a^dsRED/dsRED), heteroallelic Ir8a mutant (Ir8a^attP/dsRED), orco mutant (orco^16/16), heteroallelic orco mutant (orco^5/16) and double mutant (Ir8a^dsRED/dsRED, orco^16/16) female mosquitoes. In these assays, wild-type mosquitoes and orco mutants showed robust attraction. In contrast, attraction to humans was significantly impaired in Ir8a mutants (Figure 4B). Differences in attraction were significant for both genotype and the 15 human subjects tested (two-way ANOVA, p < 0.0001 for genotype and p = 0.0095 for subject). The statistical difference in mosquito attraction to individual human subjects reflects the differences in attractive cues that emanate from each subject. Surprisingly, the additional loss of orco in the double mutants did not significantly enhance the Ir8a human detection defect in this assay (Figure 4B). This suggests the existence of other olfactory receptors that can respond to human odor since the double mutant still retains some attraction.

Figure 4. Ae. aegypti IR8a pathway is required to detect humans and human odor.

(A) Illustration showing mosquito attraction to a human arm in a uniport olfactometer. (B) Dot plot of percent female mosquitoes of indicated genotypes attracted to 15 human subjects in the presence of CO₂. Each subject was tested twice. An analysis was done using two-way ANOVA to compare the mean difference in attraction between subjects (n = 30, p < 0.0001) and genotypes (n = 30, p< 0.0095). Human subjects (1–15) were differentiated on the plots by dot color in this figure (bottom panel). (C) Percent female mosquitoes of indicated genotypes attracted to a human arm in the absence of CO₂ (p < 0.0001; n = 8 – 10) (D) Illustration showing mosquito attraction to a human scented nylon sleeve in a uniport olfactometer. (E) Percent female mosquitoes of indicated genotypes attracted to human odor trapped on nylon sleeve in the presence of CO₂. Data was analyzed using oneway ANOVA (n = 15, p = 0.0001). (F) Percent female mosquitoes of indicated genotypes attracted to human odor trapped on nylon sleeve in the absence of CO₂ (p = 0.0001; n = 15). For the dot plots, long lines represent the mean and short lines represent standard error. All data above was analyzed using One-Way ANOVA. Genotypes marked with different letters are significantly different by post hoc Tukey’s HSD test. Human subject 1 was used in the experiments in panels C, E, & F. See also Figure S3 & S4 as well as Table S2.

We next asked if the loss of attraction was dependent on olfactory cues from human hosts or on other cues emanating from live humans such as body heat, moisture, or visual cues. To test this, we excluded these other cues by trapping human odor on nylon sleeves previously worn on a human subject for 12 hours and exposing the scented sleeves to mosquitoes in our uniport assay (Figure 4D). In order to ascertain how long the scented sleeves could continue to elicit attraction, we tested the responses of wild-type mosquitoes. Even after 7 trials, which take approximately 85 minutes to perform, we found no significant difference in attraction rate (Figure S3). Thus, a given scented nylon sleeve was never used for more than 7 consecutive trials. Nevertheless, we randomly tested all genotypes to control for any potential bias linked to when human odor was presented. Robust attraction was recorded in the wild-type, heterozygous controls, and orco mutants when the human-scented sleeve was simultaneously presented with CO₂. Consistent with the study carried out with human subjects, Ir8a mutants show deficits in detecting nylon sleeves perfumed with human odor in these assays (Figure 4E). The weak attraction in the mutants cannot be explained by a defect in locomotion or overall fitness of the Ir8a mutants (Figure S2A–F). This suggests that an impairment in sensitivity to Ir8a-dependent odor-ligands presented by humans is responsible for the behavioral defect seen in the uniport olfactometer assays with human subjects.

CO₂ differentially modulates the IR8a and Orco pathways

We asked if CO₂ activation can modulate the attraction rate of Ir8a mutants by testing the attraction of the mutants to a human subject without including added CO₂ in the assay. In contrast to uniport assays that included CO₂ (Figure 4B & E), we recorded a host-seeking deficit in orco mutants in the absence of added CO₂ that was not statistically different than Ir8a mutants (Figure 4C & F). This supports an earlier study which reported that CO₂ synergizes with human odor to rescue host-seeking defects in orco mutants [21]. Similar to what was observed with a human host when CO₂ is not included in the assay, both Ir8a and orco single mutants lose strong attraction to nylon sleeves perfumed with human odor compared to wild-type controls (Figure 4C & F). In contrast to orco mutants, Ir8a mutants’ response to human odor was not changed by the presence or absence of CO₂ (Figure S4A). We found no difference in CO₂ levels in the uniport when airflow is co-presented with a human arm or not (Figure S4B). Passing airflow over a human arm in the presence of CO₂ did not increase CO₂ levels versus airflow and CO₂ alone (Figure S4B).

Similar to what was observed in figure 4C, attraction to human scented nylon sleeves is impaired in Ir8a^dsRED, orco¹⁶ double mutants when CO₂ is absent (Figure 4F). As the genetic interaction between Ir8a and orco is dependent on CO₂ sensation, we hypothesize that CO₂ can activate additional olfactory receptor pathways that integrate the response to host odor cues in the absence of the OR pathway. Unlike in orco mutants, the response of Ir8a mutants to human odor is not modulated by CO₂ (Figure 4C & S4A). This suggests that CO₂ sensitization and host odor detection by other olfactory receptor pathways are not sufficient to rescue the Ir8a host-seeking defect. To explore this difference, we tested the genetic interaction between Ir8a and Gr3 mutants.

The loss of Gr3enhances the Ir8a-dependent host-seeking defect

The integration of host cues requires CO₂ and is dependent on Gr3 in Ae. aegypti [22]. We investigated if Ir8a and Gr3 act together to drive host attraction. Using our uniport assay with a human host and added CO₂, we found that host-seeking behavior is also impaired not only in the Ir8a mutants but also in the Gr3 mutants (Figure 5A). The loss of Gr3 function causes a stronger reduction in host-seeking behavior than the loss of Ir8a function (Figure 5A & B). The loss of both Ir8a and Gr3 causes a host-seeking defect similar to the loss of Gr3 alone (Figure 5A & B). This suggests that Gr3 may be necessary for Ir8a function. Unlike orco mutants or wild-type controls, the attraction rate of Ir8a mutants were not modified when CO₂ was excluded from olfactometer assays with human odor (Figure 5C & S4A). Therefore, CO₂ activation is not sufficient to rescue the Ir8a host-seeking phenotype. This suggests that detection of Ir8a-dependent ligands unlike orco-dependent ligands is a non-redundant component of mosquito host detection.

Figure 5. The loss of Gr3 enhances the Ir8a-dependent host-seeking defect.

(A) Percent female mosquitoes of indicated genotypes attracted to a human arm in the presence of CO₂ (p < 0.0001; n = 10 −12). (B) Percent female mosquitoes of indicated genotypes attracted to a human arm in the absence of CO₂ (p < 0.0001; n = 9–11). Analysis was done using one-way ANOVA by comparing mean attraction across all genotypes. Genotypes marked with different letters are significantly different by post hoc Tukey’s HSD test. For the dot plots in A and B, long lines represent the mean and short lines represent standard error. (C) Percentage response of mosquitoes attracted to human in the presence and absence of CO₂ represented on a bar plot showing mean and standard error. Data compared in C is from figures 5A and 5B and analyzed by Two-Way ANOVA, grouped column statistics comparing Ir8a and Gr3 mutants (p < 0.001). Genotypes marked with asterisks are significantly different. Human subject 1 was used in the experiments in this figure. See also Figure S4 and Table S2.

Discussion

Here we present evidence that strongly suggests that Ir8a is a crucial mediator of Ae. aegypti mosquito attraction to humans and human odor. Ir8a is necessary for the electrophysiological response of the mosquito antennae to acidic volatiles found in human odor including lactic acid. We present evidence that the Ae. aegypti IR8a pathway responds to human odor cues during blood-feeding as well as enables mosquito attraction to humans, human odor, and lactic acid. By varying CO₂ in our behavioral assays and testing the genetic interactions between Ir8a and orco as well as Ir8a and Gr3, we have connected the Ir8a pathway to the overall integration of host cues by mosquitoes. Taken together, our evidence supports the conclusion that the Ir8a pathway is a key participant in the multimodal integration of host odor cues by Ae. aegypti mosquitoes, whose absence cannot be compensated for by other olfactory receptors.

Lactic acid and carboxylic acids are a major component of human sweat and may distinguish humans from other vertebrate hosts [31]. Lactic acid acts synergistically with certain volatile compounds to increase mosquito attraction. Our results support the conclusion that the detection of lactic acid and carboxylic acids by mosquitoes is necessary for robust attraction to humans. Further study will be required to determine if Ir8a mediates the decision to target humans over other vertebrate hosts. In addition, the identification of IR8a as a co-receptor required for lactic acid detection allows for the identification of the odor-tuned receptor(s) that are necessary to sense this important kairomone.

Our results have implications not only for IRs, but for how sensory information from the OR and the CO₂-sensing GR pathways are integrated during mosquito host detection. Our experiments reveal that Ir8a and Gr3 are both required for human host detection, but the genetic interaction of the two genes is not additive. It could have been possible that the loss of both genes could eliminate mosquito host detection, but this was not the case. Double mutants lacking Ir8a and Gr3 show a significant reduction in host detection when compared to Ir8a mutants alone (Figure 5A & B). However, the defects in the double mutant are not enhanced when compared to Gr3 mutants with an intact Ir8a gene (Figure 5A & B). In addition, the phenotype of Ir8a mutants is not rescued by the presence of CO₂ (Figure 5C, S4A). This suggests that Gr3 activation is necessary but not sufficient to promote Ir8a-dependent host attraction. Gr3 activation by CO₂ may make acidic volatiles in human odor salient to the mosquito, but Ir8a is still required to detect these odors.

Our evidence suggests that Gr3 activation gates the responses of both the OR pathway and the IR8a pathway to promote host-seeking, but how this interaction is achieved is not the same. Based on our results, the loss of Ir8a cannot be compensated for by using other olfactory receptors as in orco mutants. Our genetic interaction experiments support the conclusion that Gr3 activation is required for IR8a pathway function, and CO₂-dependent gating of additional Ir8a-independent receptors can compensate for the loss of orco function. The identification of these receptors is imperative for understanding how mosquitoes finds their human hosts.

The functional characterization of an ionotropic chemoreceptor family member in Ae. aegypti we present provides insight into how olfactory receptor pathways can interact to mediate the detection of humans and their odor by mosquitoes. However, the understanding of how these sensory responses are integrated to facilitate host detection is far from complete. Determining if the integration of cues occurs within the antennal lobe or in higher olfactory processing centers in the brain will help reveal the dynamics of the innate neural circuits that enable mosquito host detection. Further study is necessary to uncover how mosquitoes sense their human hosts starting from the peripheral perception of cues, to the integration and processing of the information in the central nervous system, and finally, the motor circuit outputs that drive host-seeking behavior.

STAR * METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Matthew DeGennaro (mdegenna@fiu.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Statement of Research Ethics

All research was conducted in compliance with the NIH guidelines and the Florida International University Environmental Health and Safety guideline. Laboratory practices, facilities and equipment were reviewed and approved by the Florida International University Institutional Biosafety Committee (IBC-18–004) and Institutional Review Board (IRB-16–0388 & IRB-16–0386-CR02). Informed consent was obtained from human subject volunteers before their participation in this study.

Insect rearing

Aedes aegypti mosquitoes were reared and maintained at 25–28°C, 75% relative humidity under a 14:10 light-dark cycle (lights on at 8 am). All mosquitoes used in these experiments were generated from the Orlando laboratory strain. Mosquito eggs were hatched in deionized, deoxygenated water containing dissolved tablets of Tetramin tropical fish food (catalog #16152, Tetra, Melle, Germany), which served as food for the emerged larvae. Adult mosquitoes were given ad libitum access to 10% sucrose solution. About 1 to 2-week-old adult females were fed on defibrinated sheep blood to generate eggs. Before behavioral assays, 5 to 7 day old sugar-fed mosquitoes were sorted and sexed under hypothermic (4°C) conditions and fasted for up to 24 hours on water. All mosquitoes were tested only once in the behavioral assays in this study and then sacrificed.

Subject volunteers

A total of 18 human subject volunteers participated in host seeking experiments (Table S2). However, three subjects were later excluded because they failed either, to complete all the trials or their body odor was not attractive to wild-type mosquitoes. Each subject was tested twice to assess their attractiveness to mosquitoes. The volunteers were diverse in age (19 to 41 years), sex (Male = 7, Female = 8), and race (White = 5, Hispanic = 7, Asian = 1, Black = 2).

METHOD DETAILS

CRISPR/Cas9 nuclease reagents

CRISPR short guided RNAs (sgRNA) were designed according to standard protocols [43]. The sgRNAs sequences for the mutagenesis of exon 2 and exon 3 of ir8a gene were chosen by the presence of protospacer-adjacent motifs (PAMs) with the sequence NGG. Generated sgRNA sequences were checked for potential off-target binding sites using ZiFit (http://zifit.partners.org/ZiFiT/). Double-stranded DNA templates for specific sgRNAs were produced by performing a template-free PCR with two overlapping oligonucleotides. One containing the specific target sequences (ir8a^dsRED exon2: GGGCGGACAAAATGGCGTAT and ir8a^attP exon 3: GGACATCTGTCGACGATAAC), and the universal CRISPR reverse primer (Table S1). The forward primers used were IR8aExon2CRISPRF and IR8aExon3CRISPRF. The reverse primer used was sgRNArev. Both sgRNAs were produced using MEGAscript T7 Transcription Kit (Catalog #AM1334, Life Technologies, Carlsbad, CA, USA). Following incubation, sgRNA transcripts were purified using MEGAclear Transcription Clean-Up Kit (Catalog #AM1908, Life Technologies) and verified on the Agilent bioanalyzer (Agilent, Santa Clara, CA, USA).

The Cas9 polyadenylated mRNA was made by digesting the MLM3613 plasmid [53] with the PmeI restriction enzyme (New England Biolabs, Ipswich, MA, USA), miniprep purifying the linearized plasmid, and in vitro transcribing the mRNA from the DNA fragment using the mMessage mMachine T7 ultra kit (Catalog #AM1345, Thermo Fisher Scientific). The mRNA transcript was purified using MEGAclear Transcription Clean-Up Kit (Catalog #AM1908, Life Technologies) and checked on the Agilent bioanalyzer (Agilent). MLM3613 was a gift from Keith Joung (Addgene plasmid #42251; http://n2t.net/addgene:42251; RRID: Addgene_42251).

Donor DNAs

The single stranded oligo nucleotide (ssODN) used to integrate attP recombination site sequences into exon 3 of ir8a was synthesized by Integrated DNA Technologies (www.idtdna.com) using the 20 nmole ultramer service with standard desalting. The ssODN contains a 50bp attP site flanked by 75bp of homologous sequences to the target site on both sides: GATTCTCGGTTCTGGATGCCTGACGGCAGTCGAATGTTACGATACAATTTGAACGTTTCGATTTGGACATCTGTCCT ACGCCCCCAACTGAGAGAACTCAAAGGTTACCCCAGTTGGGGCACTACAACGGGTCTGCTGCAATACACCTTGGGACGATCGAGAAGG GTGATGTGGTACCGTTCGTGGGTCAGAAGATCAAA (attP sequence in bold).

To generate the donor DNA to integrate into exon 2 of IR8a, the pSL1180:polyUBdsRED [54] was modified to include homologous sequences surrounding the CRISPR target site. The 796bp left arm and 1005bp right arms were amplified from Orlando strain Ae. aegypti genomic DNA using Novagen KOD polymerase (EMD Millipore, Temecula, CA, USA) and these primers: infusionIR8LA_1, infusionIR8LA_2, infusion_IR8RA1, and infusion_IR8RA2 (Table S1).

After the PCR fragments were amplified and purified with Qiaquick PCR purification kit (Catalog #28106, Qiagen, Hilgen, Germany), pSL 1180:polyUBdsRED was digested with both EcoRI and XhoI (New England Biolabs). The 3253bp fragment (pSL 1180 backbone) and 2321bp fragment (polyubiquitin:dsRED:SV40) were both gel purified. The left homologous arm PCR, polyubiquitin:dsRED fragment, the right homologous arm PCR, and psl1180 backbone were simultaneously assembled using a recombination based method, Infusion HD (Catalog #638909, Clontech). The resulting DNA was cut with the MluI restriction enzyme (New England Biolabs) and Sanger sequenced to confirm the proper integration of the fragments. The final plasmid DNA, pGT-Ir8a, was purified from E. coli using the Qiagen Endo-free Maxiprep DNA (Catalog #12362, Qiagen) isolation kit.

Ir8a^dsRED and Ir8a^attP mutant allele generation and detection

To generate stable germline mutations, CRISPR-Cas9 reagents were injected into the posterior end of the pre-blastoderm embryos. This stage allows nuclei that will become somatic and germline cells to be exposed to the CRISPR-Cas9 complex since the dividing nuclei have not yet undergone cytokinesis. Microinjection into Ae. aegypti pre-blastoderm embryos was performed at the Insect Transformation Facility at the University of Maryland. The microinjection mixes were prepared as follows: for Ir8a^dsRED, sgRNA (25 ng/μl), Cas9 mRNA (500 ng/μl), and donor plasmid (500 ng/μl) and, for Ir8a^attP, sgRNA (25 ng/μl), Cas9 mRNA (500 ng/μl), and ssODN (500 ng/μl) were diluted into nuclease-free H₂O. A total of 1,069 and 509 embryos were injected to target exon 2 and exon 3, respectively.

Embryos were hatched 3 days after injection and reared to the adult stage as previously described [42]. A total of 50 G₀ females were sexed during pupation, allowed to mate freely with wild-type Orlando males, and females were blood-fed to generate G₁ progeny. G₀ females were put into oviposition vials to collect G₁ eggs. Thereafter, G₁ eggs were hatched and pupae were screened for Ir8a^dsRED insertion allele using a fluorescent microscope. Fluorescent G₁ individuals were sexed and reared to adulthood. Fluorescent females were outcrossed with wild-type animals for 5 generations and homozygosed.

The dsRED fluorescent protein gene insertion was detected in the Ir8a locus by PCR. Mosquito genomic DNA was purified using DNeasy Blood & Tissue kits (Catalog #69504, Qiagen). PCR amplification was done using the following primers Ir8a^dsREDForLA3 and Ir8aexon4rev3 (Table S1).

To further confirm site specific integration of the pGT-Ir8a donor DNA into Ir8a exon 2 and sequence the integration site, a 3483bp PCR product was amplified using Novagen KOD polymerase (Catalog #71086–5, EMD Millipore) with the forward primer (Ir8a^dsREDForLA3) homologous to the first 20bp of Ir8a exon 2, which is also contained in the left arm of the donor plasmid, and the reverse primer (Ir8aexon4rev3) in Ir8a exon 4 outside the boundary of the right arm of the donor plasmid. This PCR product was then purified using NucleoSpin Gel and PCR cleanup kit (Catalog #740609, Machery-Nagel Inc., Bethlehem, PA, USA), cloned using the NEB PCR cloning kit (Catalog #E1202S, New England Biolabs) and Sanger sequenced using the SimpleSeq premixed Kit (Eurofins Genomics) with the following primers: Ir8adsRedForLA3, Ir8adsRedForLA1, Ir8aexon4rev3, Ir8a_polyU_For, Ir8adsRED_poly_rev2, Ir8a_afterRA_rev, and, SV40For1 (Table S1). Sequencing showed directed insertion into Ir8a exon 2 with no aberrant insertions or deletions.

Ir8a^attP mutant lines were screened for attP insertion and detected by fluorescent size-based genotyping each generation. Pupae were sexed and allowed to eclose in same-sex cages. Virgin G₂ females were allowed to mate freely with introduced wild-type males 24 hours after eclosion. Mosquitoes were then outcrossed to wild type in this manner for 5 generations. Single-pair crosses of G₅ siblings were set up and mosquitoes with the desired mutations were chosen for homozygosing. Females were blood-fed and placed in individual oviposition vials. After eggs were laid, females were sacrificed, and genomic DNA was extracted using the IBI Genomic DNA extraction kit (Catalog #IB47222, IBI Scientific, Peosta, IA, USA). A fluorescent amplicon was generated using Amplitaq Gold 360 PCR master mix (Catalog #4398881, Applied Biosystems, Foster City, CA, USA) for each female using the following primers: Ir8aExon3for1 and Ir8aExon3rev2 (Table S1). Samples were sent for capillary electrophoresis to the DNA Core Lab at Florida International University and were then genotyped using Peak Scanner software (v1.0, Applied Biosystems) to determine the fragment length. To confirm the single stranded donor DNA insertion in the Ir8a^attP allele, PCR products were amplified from mosquitoes that contained the Ir8a^attP allele and were PCR purified using QIAEX II Gel Extraction Kit (Catalog #20051, Qiagen). After purification, the amplicons were ligated into the pCR2.1 TOPO TA vector (Catalog #K462040, Invitrogen, Carlsbad, CA, USA) and transformed with One Shot TOP10 chemically competent E. coli cells (Catalog #C607003, Invitrogen). Thereafter, colonies were picked and prepped using the QIAprep Spin Miniprep Kit (Catalog #27104, Qiagen). The plasmid DNA was sequenced in the DNA Core facility lab at the Florida International University.

Mosquito tissue dissection and preparation

In order to determine where Ir8a is expressed in adult female mosquitoes, we dissected five different body parts including the whole body, head without antennae, antennae alone, head alone, and headless body from sugar-fed 7–10 days old wild-type Ae. aegypti of the Orlando strain. A total of 10 mosquitoes were dissected for each body part except for the antennae, where a total of 25 antennal pairs were used for the assay. In addition, we also dissected antennal tissues from ir8a^dsRED/dsRED, Ir8a^attP/dsRED, Ir8a^dsRED/+, Ir8a^attP/+, and wild-type Ae. aegypti Orlando strain to determine if Ir8a mRNA expression was altered by the mutations we made. For this assay, a total of 10 mosquito heads were used for each genotype. Using a pair of forceps, samples were dissected under cold anesthesia into RNAlater stabilization solution (Catalog #AM7020, Invitrogen) or into a dry ice/ethanol bath. For each genotype or tissue type. at least three biological replicates were used.

Ir8a mRNA extraction and treatment

Mosquito tissues were suspended in a 1 ml solution containing 4 M guanidine thiocyanate (CAS: 593-84-0), 0.5% Sarkosyl (CAS: 137-16-6), Chloroform (CAS 67-66-3), and 0.1 M of 2-mercapthoethanol (CAS: 60-24-2). Thereafter, tissue samples were manually homogenized using RNase-free disposable pellet pestles (Catalog #12-141-364, Thermo Fisher Scientific). Phenol-chloroform phase separation was performed to separate the supernatant. This phase extraction process was repeated twice for the supernatant. RNA extraction was performed using the RNAid Kit supplied by MPBio (catalog #111007200). Beads were washed twice using RNA Wash concentrate before eluting in 20μl DEPC-treated water. Sample concentration and quality were determined using NanoDrop 2000c (Thermo Fisher Scientific).

cDNA synthesis and qPCR

The RNA extracted was treated with Turbo DNA-free kit (Catalog #AM1907, Invitrogen) to remove any DNA contamination. RNA was diluted to 50 ng/ul for each sample. cDNA library was created with reverse transcriptase using the SuperScript II RT reagent kit (Catalog # 18080–051, Invitrogen) and primed using oligo dT. Each reaction comprises 200ng RNA to make up a final volume of 20μl. After cDNA synthesis, samples were treated with RNase cocktail enzyme mix (Catalog #AM2286, Thermo Fisher Scientific).

cDNA samples were amplified using Amplitaq Gold 360 master mix (Catalog #4398901, Fisher Scientific) and primed by ribosomal protein L32 gene. Genomic contamination and PCR amplification was assessed by running the sample on an agarose gel electrophoresis.

Ae. aegypti ir8a mRNA expression was quantified using RT-qPCR (Catalog #4345241, Thermo Fisher Scientific). Quantitative PCR was performed using TaqMan gene expression assay. The reaction consists of two sequence-specific PCR primers with a custom TaqMan probe (Catalog #4331348, Thermo Fisher Scientific) in 2X universal master mix. The TaqMan assay was performed with three technical replicates for each given biological replicate. The real-time PCR was performed using the TaqMan universal PCR master mix (Catalog #4324018, Thermo Fisher Scientific). The forward primer sequence (ATCAGTCCGATCGCTATGACAAG) and the reverse primer sequence (GGTTGTCAATACCTTTCGGCTTAC) were used as control primers (Table S1). Raw data was analyzed using SDS v1.5.1 software with the detection threshold set at 0.2. Outliers with Ct values greater by 0.5 from the nearest technical replicate were discarded. The ribosomal protein L32 gene was used as endogenous control to normalize between cDNA samples. Relative fold change was calculated as previously described [55].

Locomotor activity assay

Mosquito activity was monitored using Locomotor Activity Monitors (LAM 25; Trikinetics Inc., Waltham, MA, USA). Sugar-fed 5 to 7-day old female mosquitoes were singly aspirated into glass tubes (25 mm diameter, 125 mm long) directly from the cage without cold anesthesia. The glass tubes were sealed at both ends with cotton plugs. One of the cotton plugs was soaked in water and the other end was left dry. This was done in order to generate more activity in the tube. We hypothesized that mosquitoes would hygrotax to the wet end. The set up was maintained at 25–28°C and 75% relative humidity under a 14:10 light-dark cycle (lights on at 8 am). The movements of the mosquitoes triggered an infrared beam break. This was set to record at every second and tabulated into 1min bins. Bins with more than 60 beam breaks and trials that exceed 2000 beam breaks per day were excluded from the analysis. Activity count was recorded for 24 hours on the fourth full day of fasting.

CAFE assay

The capillary feeder (CAFE) assay was adapted for mosquitoes with slight modifications from the feeding study carried out on fruit flies [45, 46]. Female mosquitoes aged 5 to 7 days were starved from sugar for 24 hours but had unlimited access to water. After fasting, five mosquitoes were transferred into each polystyrene vial (95mm long, 27mm wide) and sealed with a cotton plug. Thereafter, two calibrated 5 μl capillary tubes (catalog #1904637, Blaubrand Intramark) were introduced into the vial by pushing it through the cotton. The capillary tubes were filled with 10% sucrose to the 5 μl mark. The assay lasted for 18 hours and feeding was estimated by recording the change in the sucrose volume. Control vials without mosquitoes were also set up to correct for water loss due to evaporation.

Fasting resistance assay

This assay was carried out accordingly as previously described [21]. A total of sixty male and sixty female mosquitoes from each of the five genotypes (Ir8a^dsRED/dsRED, Ir8a^attP/attP, Ir8a^dsRED/+, Ir8a^ttP/+ and wild-type) were tested for fasting resistance. Prior to the experiment, mosquitoes were fed on 10% sugar. Thereafter, each mosquito was aspirated into a fly vial (25 mm diameter, 95 mm long) containing 1 cotton ball soaked in 10 ml of distilled water and plugged with a cellulose acetate fly vial plug (Catalog # 49–101, Genesee Scientific, San Diego, CA, USA). In order to control for positional effects, vials were randomized for genotype and sex in the racks. Experiments took place in the behavior room at 25–28°C, 75% relative humidity under a 14:10 light-dark cycle (lights on at 8 am). To quantify fasting resistance, visual examination of the vials was carried out each day to observe for any movement. If no movement was noticed, the vial was tapped twice and visually inspected again. If the mosquito failed to move, it was scored as dead.

Electrophysiological studies

Electroantennogram (EAG) recordings were made using Ag-AgCl electrodes and glass capillaries filled with Ringer solution (8.0 g L⁻¹ NaCl 0.4 g L⁻¹ CaCl₂) connected to silver wire, which closed the electric circuit. Non-blood fed female mosquitoes (Ae. aegypti) were recorded at 3–4 days post-eclosion between 8am and 4pm. Mosquitoes were chilled for one minute before their bodies were secured between sticky tape and wax. The glass capillary connected to the indifferent electrode was placed within the eye of the mosquito, and the glass capillary connected to the recording electrode was connected to the tip of the antennae. The signals were passed through a high impedance amplifier (IDAC-4, Syntech 2004, Hilversum, Netherlands) and analyzed using a customized software package (Syntech EAG-Pro 4.6). Ten microliter aliquots of each chemical compound at a concentration of 1× 10⁻³ mg/mL were added onto a precut filter paper (Whatman No. 1, 20 mm), which was inserted into sterilized Pasteur pipette. The stimuli were delivered via an air stream at a flow rate of 1 L.min⁻¹ with a puff (2 seconds duration) at 30 second intervals. Solvent control (Hexane) was tested at the beginning and end of each repetition. Each treatment contained 5 replicates.

The odor volatiles tested in the EAG experiment include: Geranylacetone (C.A.S. 689-67-8), 6-methyl-5-hepten-2-one (Sulcatone, C.A.S. 110-93-0), Linalool (C.A.S. 78-70-6), Dodecanal (C.A.S. 112-54-9), Lactic acid (C.A.S. 79-33-4) and were obtained from Sigma-Aldrich. Octanoic acid (C.A.S.124-07-2), Heptanoic acid (C.A.S. 111-14-8) and Butyric acid (C.A.S. 107-92-6) were supplied by ICN. Octanal (C.A.S. 124-13-0) and Nonanal (C.A.S. 124-19-6) were obtained from Chemicon:Acros. Limonene (C.A.S. 5989-27-5) and Nonanoic acid (C.A.S. 112-05-0) were supplied by Fluka. 2-ethyl hexanol (C.A.S. 104-76-7) was obtained from Lancaster; while 1-octen-3-ol (C.A.S. 3391-86-4) was supplied by Janssen Chimica.

Membrane-feeding assay

This assay was carried out as previously described with slight modifications [22]. A total of four 14 mm diameter glass-jacketed membrane feeders (#CG-1835–70, Chemglass) were connected in series using silicone tubing to a digital water bath (VWR international). The membrane feeder was prepared by stretching a thin layer of Parafilm M laboratory film (Catalog #PM996, Bemis) over the feeders. Thereafter, a total of 1000μl of defibrinated whole sheep blood without any added ATP (Hemostat Laboratories) was transferred into each feeder. A dimension of 10 cm by 10 cm squares of nylon sleeves (Duane Reade), previously worn for 12 hours by a human subject, were perforated and stretched over the Parafilm and tied around the feeder. For each trial, 4 cups of mosquitoes (n =18–20) were set into their feeding positions in a room with regulated temperature (25°C) and relative humidity (75%). The assay was supplemented with CO₂ diffusion pads (8.0 cm by 11.5 cm; Tritech Research), placed at the edge of the cup, and set to release at a rate of 2500–2700ppm measured by a carbon dioxide monitor (Catalog # CO₂-100, Amprobe, Everett, WA). The blood in the membrane feeder was heated to 37°C to mimic human body temperature. For assays without heated blood, the temperature was maintained at room temperature (26°C). Feeding positions were alternated to control for position effects on the genotypes tested. The assay had a duration of 15 minutes. Blood fed mosquitoes were visually scored and any mosquitoes that appeared not to have blood-fed were squashed between paper towels. If blood was found on the paper after squashing, the animal was scored as blood-fed.

Uniport olfactometer assay

A custom-built uniport olfactometer constructed by the Engineering Department at Florida International University was built to assess female mosquito attraction to human host stimuli. The uniport olfactometer is made of a large plexiglass tube (75cm long and 13cm wide) attached to a small cylindrical trap (13cm long and diameter is 5cm wide) which houses the mosquitoes before the experiment. The other end of the plexiglass tube is a hollow box (length is 25cm, breadth is 20cm and diameter is 13cm) connected to the stimulus chamber. Carbon-filtered, humidified air and CO₂ are able to mix with the odorants in the stimulus chamber to attract mosquitoes that have been released from the trap. The CO₂ release rate in the stimulus chamber was measured by an acrylic flowmeter Model VFA-4-SSV (Dwyer Instruments Inc., IN, USA) set at 3 SCFH. The final concentration for CO₂ in the assay was maintained at 2500–2700ppm by a carbon dioxide monitor (Catalog # CO₂-100, Amprobe). Whereas, air flow rate was set at 21 standard cubic feet per hour by an air flowmeter (King Instruments CA, USA). The sealed design of the uniport, air filtration, and the positive pressure caused by the airflow in the apparatus isolated the assay from potential odors in the surrounding environment.

For each trial, approximately 20 female Ae. aegypti mosquitoes one-week post-eclosion were sorted under cold anesthesia (4 °C) and placed in a small cylindrical trap. All females used in the assay had access to their mating partners but had not obtained a blood meal. Females were fasted with access to water for up to 24 hours prior to the assay. Pre-assay fasting and behavior experiments took place at 25°C and 70–80% relative humidity. In our experiments, we tested multiple genotypes including wild-type, heterozygous Ir8a mutant (Ir8a^dsRED/+), homozygous Ir8a mutant (Ir8a^dsRED/dsRED), heteroallelic Ir8a mutant (Ir8a^attP/dsRED), homozygous orco mutant (orco^16/16), heteroallelic orco mutant (orco^5/16), homozygous Ir8a and orco double mutants Iir8a^dsRED, orco¹⁶), homozygous Gr3 mutant (Gr3^cfp/cfp), heterozygous Gr3 mutant (Gr₃^cfp/+), and homozygous Ir8a and Gr3 double mutants (Ir8a^dsRED, Gr₃^cfp). Heteroallelic mutants were tested to control for background mutations that may have occurred independently in each line. Double mutants were tested to assess the epistatic interaction between the two genes.

To explore the behavior of these mosquito lines, starved mosquitoes were released from the small cylindrical trap and allowed to respond to the stimuli. After 8 mins, the number of mosquitoes attracted by host cues were counted. The 8 mins time point was determined empirically to produce consistently high responses. A blank trial with no odor stimulus was run to ascertain that the set-up was clean. Attraction greater than 20% indicates the presence of residual odor and the whole set up was cleaned up again with an odorless soap and allowed to air dry. Mosquitoes were scored as attracted if they were able to fly upwind through the tube into the attraction trap within the time frame. Mosquitoes that move out of the cylindrical trap within the assay period were scored as activated.

Human host uniport assay

A total of 15 human subjects were recruited for this assay. Each subject was tested twice to assess their attractiveness to mosquitoes in a uniport olfactometer as described above. All subjects were given informed consent before participating in the experiment and approved by the Florida International University Institutional Review Board. The volunteers were diverse in age (19 to 41 years), gender (Male=7, Female=8), and race (White [5], Hispanic [7], Asian [1], Black [2]). All of the human subjects were asked not to wear scented cosmetics, deodorants, or fragrances on the day of the assay. Any subject that did not follow our request was excluded from the study. Human subjects that adhered to this rule were asked to insert their forearm up to the elbow level into the stimulus chamber. The arm was inserted through a tight-fitted glove affixed to the olfactometer to prevent airflow between the assay and the room. None of the human volunteers were bitten by mosquitoes during the experiment.

Human odor perfumed nylon sleeves uniport assay

The toe sections of women’s knee-high pantyhose (Hanes Brands Inc., NC, USA) were cut off with scissors. A single volunteer subject was used for this assay to control for differential attractiveness. The subject wore the nylon sleeves on the arm and stretched it towards the armpit. This was worn overnight for 12 hours while the subject did not shower or apply any scented products. The nylon sleeves were later retrieved from both arms and immediately tested on the library of mutants already generated, as well as the wild type mosquitoes in the uniport olfactometer. New nylon sleeves perfumed with human odor were used for each day of testing without prior storage. Nylon sleeves that had not been previously worn were used as a negative control.

Lactic acid uniport assay

In order to test the response of wild-type and Ir8a mutant mosquitoes to lactic acid, we used the set up for the uniport olfactometer assay with slight modifications. Approximately 30 mosquitoes were used for each trial. A total of 4ml L-(+)- lactic acid solution (C.A.S. 79-33-4, catalog number 27714; 88%−92%, Sigma-Aldrich) was transferred by a pipette directly into the open lid of a polystyrene Petri dish (60mm diameter, 10mm height). The treatment was centrally placed in the stimulus port. For the control, the Petri dish was left blank. The number of mosquitoes attracted was recorded after 10 min. CO₂ concentration (2500ppm −2700ppm) was monitored by a carbon dioxide meter (Catalog # C0₂-100, Amprobe).

Measuring CO₂ abundance in the uniport olfactometer

Carbon dioxide monitor (Amprobe CO₂- 100, Everett, WA) was used to measure CO₂ release rate in a custom-built uniport olfactometer. In this experiment, air flow rate was maintained at 21 standard cubic feet per hour (SCFH) as measured by an airflow meter (King Instruments CA, USA). The final CO₂ concentration in the assay was measured by placing the CO₂ monitor at the end where the trapped mosquitoes were temporarily held before they were released to fly upwind. For the assay with air alone, carbon-filtered humidified air was released into the stimulus port. For the assay with supplemented CO₂, an acrylic flowmeter Model VFA-4-SSV (Dwyer Instruments Inc., IN, USA) set at 3 SCFH was used to control CO₂ release rate into the stimulus chamber of the uniport. For the assay with a human arm, the arm of the subject was inserted through a tight-fitted glove affixed to the olfactometer to prevent airflow between the assay and the room. The sealed design of the uniport and the positive pressure caused by the airflow in the apparatus isolate the assay from potential odors in the surrounding environment. The stimuli (airflow, supplemented CO₂ and skin-emanated CO₂) traveled from the stimulus port into the large plexiglass tube (75cm long and 13cm wide) and exited via the aperture where the mosquitoes were released. The CO₂ final concentration was measured via the aperture where the mosquitoes were released.

A total of ten trials were carried out to measure the CO₂ release rate from the olfactometer. The device was turned on and allowed to calibrate to ambient CO₂ room concentration. For each measurement, the device was placed in the aperture and measurements were recorded when the digital recording was stabilized. The device recording was stable when the beep sound was consistent. After taking the recording, the device was removed from the aperture and allowed to calibrate to ambient room condition. This process was repeated consistently for the ten trials. Carbon dioxide release rates were measured in three different conditions: humidified air, humidified air and CO₂, and humidified air supplemented with CO₂ and human arm. Data was analyzed using one-way AVOVA. Conditions marked with different letters were significantly different according to post hoc Tukey’s HSD test.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis of our behavioral and quantitative PCR data sets was performed using the GraphPad Prism 7 software package (GraphPad Software, San Diego, CA, USA). Statistical analysis of our electrophysiological recordings (EAGs) were performed using GraphPad Prism 8 software package (GraphPad Software, San Diego, CA, USA). All details for statistical analysis including the statistical tests used, number of trials (n), number of animals, and how significance was determined can be found in the figure legends.

Data S1: Ir8a^dsRED sequencing data. Related to Figure 1.

SnapGene and genbank files showing the Sanger sequencing of the PCR amplicon. The sequence confirmed the site-directed integration of the donor plasmid into the Ir8a locus.

Data S2: Ir8a^attP sequencing data. Related to Figure 1.

SnapGene and genbank files showing the Ir8a mRNA that would result from the attP donor DNA insertion into the Ir8a locus. The genomic DNA around the attP insertion side was verified by Sanger sequencing.

Highlights.

• Ae. aegypti Ir8a mutant mosquitoes cannot sense lactic acid, a human sweat component

• Attraction to humans and human odor is reduced in Ir8a mutant mosquitoes

• Ae. aegyptiIR8a pathway responds to human odor cues during blood-feeding

• The Ir8a mutant host-seeking defect cannot be rescued by other olfactory receptors