Hold your breath – Differential behavioral and sensory acuity of mosquitoes to acetone and carbon dioxide

Host seeking in the yellow fever mosquito, Aedes aegypti, and the African malaria mosquito, Anopheles coluzzii, relies on specific and generic host-derived odorants. Previous analyses indicate that the behavioral response of these species depends differentially on the presence of carbon dioxide (CO2) and other constituents in human breath for activation and attraction. In this study, we use a flight tube assay and electrophysiological analysis to assess the role of acetone, a major component of exhaled human breath, in modulating the behavioral and sensory neuronal response of these mosquito species, in the presence and absence of CO2. When presented alone at ecologically relevant concentrations, acetone increases attraction in Ae. aegypti, but not in An. coluzzii. Moreover, in combination with CO2, human breath-equivalents of acetone ranging between 0.1 and 10 ppm reproduces a behavioral response similar to that observed to human breath in host-seeking Ae. aegypti, but not in An. coluzzii. Acetone does, however, reduce attraction to CO2 in An. coluzzii, when presented at a higher concentration of 10 ppm. We identify the capitate peg A neuron of the maxillary palp of both species as a dual detector of CO2 and acetone. The sensory response to acetone, or binary blends of acetone and CO2, reflects the observed behavioral output in both Ae. aegypti and An. coluzzii. We conclude that host recognition is contextual and dependent on a combination of ecologically relevant odorants at naturally occurring concentrations that are encoded, in this case, by differences in the temporal structure of the neuronal response. This information should be considered when designing synthetic blends for that optimally attract mosquitoes for monitoring and control.


Introduction
The yellow fever mosquito, Aedes aegypti, and the African malaria mosquito, Anopheles coluzzii (formerly Anopheles gambiae molecular form M), exhibit a feeding preference for humans, female mosquitoes were used; these were provided water but deprived of their regular sugar source 24 h prior to experiments.

Flight tube bioassay
To analyze the behavioral response of Ae. aegypti and An. coluzzii to exhaled breath, as well as to acetone and CO 2 , presented alone or in binary blends, a flight tube bioassay was used ( Fig  1A; [5]. The bioassay was illuminated from above with white light at 280 lux for the diurnal Ae. aegypti, while red light (40 lux) was used for the nocturnal An. coluzzii. Experiments were conducted during their period of peak host-seeking activity of each species [27,28]. Charcoal-filtered humidified air (25 ± 2˚C, RH 65 ± 2%) flowed through the flight tube at 30 cm s −1 . To ensure a laminar flow and a homogeneous plume structure, the air passed through a series of stainless-steel mesh screens prior to entering the flight tube. Homogenous discrete stimulus pulses (1 s on and 1 s off) were introduced into the flight section using a stimulus controller (SEC-2/b, Syntech), which pumped stimulus-laden air from a 40 l Tedler gas sampling bag (Adtech Polymer Ltd, Stroud, UK) into a pulse generator, placed behind the mesh screens [24]. Stimuli were prepared in the following way: 1) breath was collected by a volunteer exhaling normally through a mouthpiece into the bag for 4 min prior to the experiment; 2) 30000 ppm CO 2 was prepared by introducing synthetic air (Strandmöllen AB, Ljungby, Sweden) at 1.5 l min -1 and pure CO 2 (Strandmöllen AB) at 0.2 l min -1 into the gas sampling bag, to obtain the concentration approximating the natural composition in exhaled breath [12]; 3) acetone (99.9%, Chromasolv, Sigma-Aldrich, Stockholm, Sweden) was diluted with distilled water and then pipetted into the bag, which was subsequently filled with synthetic air, to obtain concentrations of 0.1 ppm, 1 ppm and 10 ppm at the downwind end of the flight tube; 4) binary mixtures of 30000 ppm CO 2 and the three concentrations of acetone; and 5) for the negative control experiments, the gas sampling bag was filled with synthetic air at an airflow of l.5 l min -1 . The consistency in amplitude and the structure of the discrete pulsed stimuli was visualized using a mini-PID (Aurora Scientific, Aurora, Ontario, Canada) with known concentration of acetone with different flow rates as described by Majeed et al. [24]. In addition, the concentrations of acetone and CO 2 were measured at the down-wind end of the flight tube using the mini-PID (Aurora Scientific, Aurora, Ontario, Canada) and a CO 2 analyzer (LI-820, LICOR Biosciences, Lincoln, NE, USA), respectively. This revealed a concentration of 1200 ± 12 ppm CO 2 , as well as average concentrations of 0.1 ± 0.02 ppm, 1 ppm ± 0.01 and 10 ± 0.04 ppm acetone.
Individual mosquitoes were kept in glass release chambers (7 × 2.6 cm i.d.), covered with stainless steel mesh on one side and a cotton plug on the other, in the bioassay room for 24 h prior to the experiments [5]. Following the opening of the release chamber, the time to take off (flight activation) and the time to source contact were recorded. The maximum experimental time was 120 s. Thirty individuals of each species were observed for all treatments. To minimize the effect of daily variation in baseline activity and responses to odors, an equal number of test and control individuals were observed each day. For both species, to compare the time to take off and the time to source contact in response to the various treatments (human breath, CO 2 and acetone alone, binary blends and synthetic air), an ANOVA followed by Tukey's HSD post hoc test was used (GraphPad Prism, v. 7; GraphPad software, La Jolla, CA, USA). The total number of mosquitoes making source contact was analyzed with nominal logistic regression, comparing presence and absence of CO 2 and dose of acetone for each species (JMP Pro v. 14, SAS Institute Inc., Cary, NC, 1989-2019).

Electrophysiology
Initial screening of all previously characterized functional types of sensilla on the antennae of Ae. aegypti and An. coluzzii [26,[29][30][31][32] revealed that the sensory neurons sensitive to acetone are found on the maxillary palps. Specifically, the type of sensory neuron responding to acetone is the same as the one responding to CO 2 , the capitate peg A (cpA) neuron [5]. Electrophysiological recordings from this sensillum type were made and analyzed as previously described [5].
A continuous humidified stream of synthetic air (Strandmöllen AB), lacking CO 2 , was passed over the maxillary palp (2 l min −1 ) via a glass tube (7 mm i.d.). Stimuli, consisting of CO 2 , acetone or binary mixture thereof, were introduced into the air stream through a hole (2 mm i.d.) in the glass tube, 11 cm upstream of the maxillary palps. Delivery of CO 2 was regulated by a two-way Teflon solenoid valve (Teddington, Skogås, Sweden), controlled via the digital output of an IDAC-4 (Syntech, Germany). The valve was connected to a separate gas cylinder containing 1200 ppm CO 2 and oxygen (20%), balanced by nitrogen (Strandmöllen AB). Acetone, dissolved in distilled water, at concentrations ranging from 0.1 to 100000 ppm, were loaded (15 μl) onto a piece of filter paper (5 × 20 mm), placed inside a Pasteur pipette. The pipettes were sealed using Parafilm™ after loading and used once within 5 min to limit the variation due to evaporation. For the dose-response analysis, stimuli were presented in increasing concentrations from 0.1 to 100000 ppm. To test binary mixtures of CO 2 and acetone, pipettes loaded with 0.1, 1 and 10 ppm acetone were gently, yet rapidly, filled with 1200 ppm CO 2 , after establishing a stable contact with a capitate peg sensillum. The rationale for selecting 1200 ppm CO 2 for the binary blends was that this concentration corresponds with the natural level associated with proximity to a host [23,25] and has previously been shown to elicit behavioral response in both mosquito species [5,24]. Stimuli were presented in the following order: CO 2 (1200 ppm), 0.1 ppm acetone, 0.1 ppm acetone + CO 2 , 1 ppm acetone, 1 ppm acetone + CO 2 , 10 ppm acetone, 10 ppm acetone + CO 2 . For all experiments, distilled water was used as a control. The entire panel of stimuli was used once per specimen. The time between stimuli was 10 s. Extracellular spike activity of the A neuron was analyzed by counting the number of spikes 0.5 s before stimulus onset and subtracting this from that of the 0.5 s period following stimulus delivery. Results are presented as spikes s -1 . To further analyze the neural response to the binary mixtures, the temporal characteristics were analyzed by plotting frequency histograms (spikes s -1 ) during 100 ms bins over a period of 2s (500 ms before to 1500 ms after the response onset).
To compare the sensitivity of the capitate peg A neurons of the two species to acetone and the binary blends, a repeated measures two-way ANOVA followed by Tukey's multiple comparison test was used (GraphPad Prism, v. 8.2.1). The kinetic curve fits were generated using a non-linear regression association/dissociation model (association then dissociation model The behavioral response of host-seeking Aedes aegypti and Anopheles coluzzii to synthetic air, human breath, acetone and CO 2 , as well as binary blends thereof. A. Behavior was assessed in a flight tube assay: (a) charcoal filtered and humidified air flowed through the assay, (b) pressurized air inlet, (c) stimulus inlet into which the stimulus was injected, (d) stainless-steel mesh plume diffusers, (e) glass flight tube, and (f) release chamber. The upper panels demonstrate that the pulsed stimuli (here shown as five cycles of 1 s on and 1 s off) maintain their amplitude and shape throughout the flight tube and at all tested flow rates. The upper right panel shows the consistent and distinct pulsed stimuli at ascending flow rates of known concentration of acetone in the flight tube. Discrete pulsed stimuli were measured in the center (in red) and at the lateral sides (in black) of the release chamber, at halfway and at the source. The upper left panel presents a graphical representation of the distinct pulsed stimuli, which shows the average amplitude of each of the five distinct pulses (N = 10) at different positions and the regression correlation coefficients (R 2 ) that demonstrate the consistency of the stimulus amplitude at the different positions within the flight tube with increasing flow rates. Time to take off (B, C; top graphs) and source contact (B, C; bottom graphs) are differentially affected by human breath, acetone and by the binary blends in the two species. The number of mosquitoes responding is indicated as numbers inset in each bar. Different letters above the bars denote significant differences between treatments within species (ANOVA, p < 0.05). Vertical bars represent the standard error of means ± SE.

Reliance on acetone for the behavioral response to human breath
The behavioral response of Ae. aegypti and An. coluzzii was differentially affected by acetone and the binary blends (Take off: F (8,261) = 9.93, P < 0.0001 and F (8,261) = 4.43, P < 0.0001, respectively; Source contact: F (8,190) = 15.2, P < 0.0001 and F (8,150) = 7.22, P < 0.0001, respectively) (Fig 1). Compared to synthetic air, acetone alone did not affect the time to take off in either species, but significantly reduced the time to source contact in Ae. aegypti (Fig 1B and  1C). When presented together with CO 2 , on the other hand, acetone significantly reduced the time to take off in Ae. aegypti, but not in An. coluzzii (Fig 1B and 1C; top). Combinations of acetone and CO 2 did not influence the time to source contact in Ae. aegypti (Fig 1B; bottom), but the time to source contact increased at the highest dose tested in An. coluzzii (Fig 1C; bottom). For both species, the number of insects that made source contact was higher in the presence of CO 2 but was not affected by the dose of acetone (Ae. aegypti χ 2 = 1.32, p = 0.52, parameter estimates: intercept = 0.036, acetone = 0.63; An. coluzzii χ 2 = 1.51, p = 0.93, parameter estimates: intercept = 0.031, acetone = 0.46).

Detection of acetone
Electrophysiological recordings from the capitate peg sensilla of Ae. aegypti and An. coluzzii (Fig 2A) revealed a dose-dependent response of the overall rate of firing of the cpA neuron to acetone at concentrations exceeding 1000 ppm (Dose: F (6,196) = 62.34, p < 0.0001), with Ae. aegypti being significantly more sensitive than An. coluzzii (Species: F (3,196) = 121.3, p < 0.0001) (Fig 2B-2D). Interestingly, even at low, and ecologically relevant, concentrations (0.1-100 ppm), acetone elicited a higher response than the control (synthetic air) for both Ae. aegypti (p = 0.0097) and An. coluzzii (p = 0.0047) (Fig 2B-2D). Taken together, these results demonstrate that both species are capable of sensing low, ecologically relevant, concentrations of acetone, however assessing the effect of the dose requires higher resolution analysis.

Detection of binary blends of acetone and CO 2
The response of the cpA neuron, when analyzed at high temporal resolution, to either acetone or CO 2 alone, was significantly different from that to the binary blends for both Ae. aegypti (F (1,10) = 70.84, p < 0.0001) and An. coluzzii (F (1,10) = 53.04, p < 0.0001) (Fig 3A-3D). In Ae. aegypti, the cpA response to acetone was significantly higher than that to the control, in both the presence and absence of CO 2 (F (3,30) = 30.04, p < 0.0001) (Fig 3E; Table 1). In contrast, in An. coluzzii, the only significant change in cpA activity was to acetone in the absence of CO 2 (F (3,30) = 6.47, p = 0.0017) (Fig 3F; Table 1).
The temporal pattern of response to acetone and CO 2 alone, as well as binary blends thereof, was phasic-tonic for both Ae. aegypti and An. coluzzii (Fig 2B and 2C and Fig 3C, A,C In Ae. aegypti, the cpA activity in response to CO 2 and the three binary blends of acetone and CO 2 could be modeled by a single, shared association/dissociation curve (F (12,306) = 1.509, p = 0.1194; dotted line in Fig 3E), revealing a consistently higher kinetic response to increasing concentrations of acetone in combination with CO 2 compared to acetone alone (Fig 3E). In contrast, a model of the activity of the Ae. aegypti cpA neuron to acetone alone revealed that each concentration could be represented by distinct association/dissociation curves (F (8,240) = 4.551, p < 0.0001; solid lines in Fig 3E), demonstrating that the response kinetics of this neuron is influenced by acetone concentration (Fig 3E). The reverse was observed for An. coluzzii, in which the cpA response to acetone alone could be modeled by a single, shared curve (F (2,204) = 1.566, p = 0.1367; dotted line in Fig 3F), while the response to CO 2 and the blends were modeled individually (F (12,272) = 4.198, p < 0.0001; solid lines in Fig 3F). This demonstrates that the response kinetics of this neuron is influenced by acetone concentration only in the presence of CO 2 (Fig 3F).

Discussion
The behavioral response of Ae. aegypti and An. coluzzii to human breath relies on the presence of CO 2 [18, 33, this study] and is differentially modulated by acetone. While acetone enhances the behavioral response of Ae. aegypti, in both the presence and absence of CO 2 , it decreases the behavioral response of An. coluzzii in the presence of CO 2 . This demonstrates that blend perception of generic volatile compounds in human breath plays a role in host attraction for these species. The observed differences in behavioral attraction correlate with the different mode of encoding CO 2 and acetone, as well as the binary blends, by the cpA neurons of the two species. The data provided here, together with our previous studies [4,5], clearly emphasize that generic host volatiles, when presented in ecologically relevant concentrations, may provide reliable cues for host attraction and recognition for mosquitoes.
When presented at ecologically relevant concentrations, acetone, in combination with CO 2 , reproduces a behavioral response similar to that observed to human breath of host-seeking Ae. aegypti, but not of An. coluzzii. This is in line with previous observations demonstrating that activation of host-seeking Ae. aegypti and An. gambiae differentially relies on human breath constituents other than CO 2 [18,33]. Acetone, when presented alone at concentrations of 1--10 ppm, significantly decreased the time to source contact in Ae. aegypti when compared with synthetic air. A similar observation was made by Venkatesh and Sen [20], albeit at concentrations of acetone >100,000× higher than that used in the current study, demonstrating that acetone may substitute for CO 2 , as suggested by Bernier et al. [21]. A plausible explanation for the observed behaviour at these extreme concentrations may be a result of the intense activation of the CO 2 sensitive, cpA, neurons, demonstrated here. When presented in combination with CO 2 , 1 ppm acetone, an equivalent concentration of that found in human breath (0.5-2 ppm), significantly decreased the time to take off in Ae. aegypti compared to that of CO 2 alone, to a level similar to that observed to human breath. Acetone has previously been shown to enhance the behavioral response of Ae. aegypti to other host volatiles, in the presence of CO 2 [21,22], verifying that acetone may act additively or synergistically to enhance activation and attraction. In contrast, acetone, whether in the presence or absence of CO 2 , did not significantly affect time to take off and source contact for An. coluzzii. The only exception to this was an observed increase in time to source contact, when acetone was presented at 10 ppm together with CO 2 . A similar inhibition of the behavioural response to high concentrations of acetone has been reported by Takken et al. [19] and Qiu et al. [17]. We conclude that Ae. aegypti and Dose-dependent response of the cpA neuron in Ae. aegypti and An. coluzzii to increasing concentrations of acetone (B-D). Note the significant increased response, above the control, to low concentrations of acetone (D). At higher concentrations, acetone elicits a significantly higher response in the cpA neuron of Ae. aegypti than of An. coluzzii, indicated by different letter designations (two-way repeated-measures ANOVA, Tukey's multiple comparison test, uppercase, Ae. aegypti; lowercase, An. coluzzii) (D). "ns" indicates no significant difference among the doses for each species control. https://doi.org/10.1371/journal.pone.0226815.g002 Differential acuity to acetone and CO2 An. coluzzii appear differentially constrained in their capacity to respond behaviorally to binary blends of acetone and CO 2 .
The observed differential behavioral response of Ae. aegypti and An. coluzzii to acetone and CO 2 , and binary blends thereof, is reflected in the sensory response of the cpA neurons to these stimuli. While we cannot rule out the existence of rare types of acetone-sensitive neurons on the antennae or differences in higher order processing of either species, this study suggests that the CO 2 -sensitive neuron in mosquitoes, far from being a labelled line for CO 2 , is capable of encoding breath-related blends. While acetone is a novel ligand for the cpA neuron, other odorants are known to be agonists of the cpA neuron in Ae. aegypti, An. gambiae and in the southern house mosquito, Culex quinquefasciatus, when presented at high concentrations [34][35][36]. Our physiological analysis, however, reveals that the cpA neuron of both Ae. aegypti and An. coluzzii demonstrate a high sensitivity to ecologically relevant concentrations of acetone alone, which in Ae. aegypti correlates with behavior. A correlation is also found between the behavior of both Ae. aegypti and An. coluzzii and the sensory response of their cpA neurons to the binary blends. In Ae. aegypti, stimulation with both acetone and CO 2 significantly increased the response of the cpA neuron, without altering the response kinetics, whereas the opposite was found for An. coluzzii. The observed differences in sensory response of the two species to acetone, either alone or in combination with CO 2 , is intriguing. Agonists of the CO 2 -sensitive neuron in both mosquitoes [34][35][36] and Drosophila melanogaster [35,37] have been shown to be detected by gustatory receptors (Grs) expressed in the cpA neuron. Whether structural or stoichiometric differences of the Grs in the two mosquito species [8,38,39] account for the demonstrated differences in sensory response remain to be analyzed.
Previous studies indicate that behavioral responses to complex host odors in mosquitoes are more robust than to single host volatiles [3,6,17,40,41]. While host discrimination and  Ae. aegypti (A, B) and An. coluzzii (C, D) to increasing concentrations of acetone, CO 2 , and binary blends thereof. E and F show, in high temporal resolution, the differential increase in response of the cpA neuron of Ae. aegypti (n recording = 6, N individual = 6) and An. coluzzii (n recording = 6, N individual = 6) to acetone in the presence or absence of CO 2 (two-way repeatedmeasures ANOVA, Tukey's multiple comparison test; �� p < 0.01; ��� p < 0.001).
https://doi.org/10.1371/journal.pone.0226815.g003 selection by anthropophilic mosquitoes may be regulated by species-specific host volatiles [42,43], recent research suggests that these behaviors also rely on a number of generic host volatiles and their relative proportions [3,4,41,[44][45][46][47][48]. In most of these studies, these host volatiles have been shown to be detected by sensory neurons on the antennae, and readily synergize with CO 2 in eliciting host-related behaviors. In this study, we identify acetone as a host volatile, detected by sensory neurons on the maxillary palps, that differentially modulates both the physiological and behavioral responses in Ae. aegypti and An. coluzzii. Behaviorally, Ae. aegypti is activated by the presence of CO 2 and acetone at ecologically relevant concentrations, whereas the activation of An. coluzzii is independent of the presence of acetone. This is reflected in the firing rates of the cpA sensory neuron of each species. In contrast, source contact is mediated in Ae. aegypti by CO 2 , with or without acetone, while this is not the case in An. coluzzii. Moreover, An. coluzzii tolerates only a limited concentration of acetone, less than 10 ppm, above which the time to source contract increases. This differential response to the presence of acetone correlates with the change in temporal kinetics seen in the cpA neuron in An. coluzzii. This suggests both that other stimuli are involved in attraction to the host in An. coluzzii [18], and that acetone is acting as a host discrimination cue in this species, but not in Ae. aegypti. In Ae. aegypti, acetone, together with CO 2 , is acting as an activator, similar to human breath, and may be the missing component in human breath that was proposed by Khan and Maibach [33] to be responsible for the observed greater human attraction. While the role of CO 2 in activating, sensitizing and attracting mosquitoes to potential hosts is well characterized [3,5,49, and references therein], this study reveals that another major component of exhaled breath, acetone, is able to modulate this behavioral response. When presented in binary blends together with CO 2 , acetone, within the natural concentrations found in the exhaled breath of potential hosts, enhances both the sensory and behavioral response of Ae. aegypti. For An. coluzzii, which is highly anthropophilic, the sensory and behavioral response is decreased at higher concentrations of acetone, indicating that acetone may act as a recognition cue to discriminate among hosts. From a vector control perspective, additional identification of ecologically relevant odorants and their naturally occurring concentrations, and how these factors affect vector-host interaction, can aid in optimizing synthetic blends for monitoring and control of mosquito populations.