Cross-modal sensory compensation increases mosquito attraction to humans

Sensory compensation is a process that allows individuals with a loss of one sense, for instance hearing or vision, to adapt to changes in their sensory abilities. Where this phenomenon has been observed, there is enhanced perception by another sense to compensate for deficiency of the lost sense. Such compensation is important for humans and non-human animals that use multisensory integration for effective navigation and the execution of vital tasks. Among these, female mosquitoes are sensory specialists that rely heavily on integrating multiple human-emitted cues in their quest for a suitable host to obtain a blood meal. Here, we identify a previously undescribed mechanism of sensory compensation in female Aedes aegypti mosquitoes. Mutant mosquitoes lacking the odorant receptor co-receptor Orco show specific enhancement in heat-seeking behavior. This s compensation does not require the antenna, which was previously assumed to be the primary mosquito thermosensitive organ. Instead, we found that the tips of the forelegs are required to detect heat, and that the heightened sensitivity in heat detection is mediated by increased neuronal activity in foreleg sensory neurons, which are distant from the head appendage neurons that express Orco. By comparative gene expression analysis in wildtype and Orco mutant legs, we identify Ir140, a foreleg-enriched member of the Ionotropic Receptor (IR) superfamily of sensory receptors, as strongly upregulated in Orco mutant legs. Emphasizing the important role of IRs in thermosensation, we find that mutant mosquitoes lacking the IR co-receptor, Ir25a, lose all responses to heat, and Ir140 mutants show strong deficits in responding to human skin temperatures. We generated an Ir140, Orco double mutant and show that these animals lose the remarkable sensory compensation seen in Orco mutants. This strongly suggests that upregulation of Ir140 in the foreleg is the mechanism of sensory compensation in Orco mutants. Odorant receptor expression is sparse in legs, suggesting an indirect, long-range mechanism of sensory compensation. Our findings reveal a novel compensatory mechanism in which loss of one sensory modality in female Aedes aegypti mosquitoes results in greater sensitivity in another to maintain the overall effectiveness of their host-seeking behavior, further enhancing their status as the most dangerous predator of humans.


INTRODUCTION
Animals are endowed with diverse sensory modalities to take in information from their environment, encompassing thermal, chemical, auditory, mechanical, visual, and other cues.These senses work together to form a representation in the brain of the sensory richness of the external world to guide appropriate behaviors.Because of the importance of multi-sensory integration, animals have mechanisms to adapt when they experience a loss or impairment in one of their sensory modalities, either through acute injury, chronic disease, or a congenital loss from birth.In this way, loss of vision, hearing, or touch, leads to the development of a heightened acuity in the remaining senses to compensate for their deficit.This adaptive process involves the reorganization and reallocation of neural resources to amplify the function of intact sensory modalities.For example, blind individuals often exhibit superior auditory and tactile perception as they adapt to prioritize these senses to navigate their environment safely and effectively (1,2).Similarly, people with hearing loss may develop a more acute sense of vision or touch to compensate for their auditory impairment (3).Sensory compensation highlights the power of neural plasticity and capacity to adapt to changing sensory inputs, ultimately enabling individuals to maintain high perceptual awareness and functionality despite sensory impairment.Sensory compensation is a phenomenon observed throughout the animal kingdom.However, the full extent of its prevalence, the precise underlying mechanisms, and how it relates to each animal's final behavioral outputs are not fully understood.
Sensory compensation is best understood at a mechanistic level in mammals and it involves anatomical modifications and reorganization at the thalamic and primary sensory cortex levels.These changes expand and refine connections within neighboring primary sensory cortical areas.For example, classical experiments in the ferret visual system revealed that eliminating retinal axon projection in one hemisphere led to alternative terminal space for these axons within the auditory thalamus (4).These changes were subsequently reflected in changes in cortical representation (5,6).Functional reorganization also occurs between primary sensory cortical regions from different sensory modalities.Braille reading by blind individuals activates the primary visual cortex, suggesting an expansion of somatosensory function into brain regions generally dedicated to visual processing (7).Moreover, in congenitally blind subjects, the occipital lobe, typically associated with visual processing, is activated during auditory localization tasks (8).Sensory compensation has also been observed by modulating existing neuronal circuits without requiring major anatomical rewiring.For example, Caenorhabditis elegans nematodes deprived of their sense of touch exhibited enhanced olfactory-mediated behavioral performance (9).This enhancement was attributed to the strengthening of synaptic transmission in the olfactory circuit that was a consequence of reduction in neuropeptide signaling cause by impaired mechanosensory circuits (9).Similarly, studies in visually deprived rats have revealed the prominent strengthening of α-amino-3-hydroxy-5-methyl-4isoxazolepropionic acid (AMPA) receptor-mediated synaptic transmission in pyramidal neurons of layer 2/3 somatosensory barrel cortex (10,11).These changes depended on long-distance serotonin signaling originating from the raphe nuclei (10,11).More recently, a study conducted in Drosophila melanogaster flies showed that loss of olfaction led to an enhanced sensitivity to detect sugar that was mediated through an elevated sugar response by the protocerebrum anterior medial dopaminergic neurons in the mushroom body (12).These anatomical and functional studies highlight the diverse mechanisms that each sensory system and organism employs while also emphasizing the shared necessity of the process of sensory compensation.
For organisms that depend heavily on integrating various sensory cues, this adaptive process is critical for guiding vital behavioral outputs.
Mosquitoes are sensory specialists that detect and integrate diverse host-emitted cues, most notably, human-emitted odors, carbon dioxide (CO2), and body heat (Figure 1A).Unlike many other organisms, female mosquitoes need substantial blood meals to initiate egg production and can feed multiple times during their lifetime.This characteristic makes Aedes aegypti mosquitoes highly effective vectors for transmitting arboviruses such as dengue, yellow fever, Zika, and chikungunya, and makes Anopheles gambiae mosquitoes extremely dangerous vectors of the Plasmodium malaria parasites.Mosquitoes rely on three large multigene families to detect human-emitted cues, gustatory receptors (GRs), odorant receptors (ORs), and ionotropic receptors (IRs).These proteins form multi-subunit ligand-gated ion channel complexes featuring one or more ligand-selective subunit and an obligatory co-receptor subunit.While GRs are generally used for taste cues, members of this gene family form a heteromultimeric receptor that detects carbon dioxide (CO2) (13)(14)(15).In Aedes aegypti, there are 72 GRs, but for this gene family, the identity of the co-receptors is less well understood.Coreceptors for the OR and IR gene family are better understood and are central to their function.Each insect species can possess hundreds of ligand-selective ORs and IRs but only one odorant receptor co-receptor (Orco) and three IR co-receptors (Ir8a, Ir76b, and Ir25a) (16).The co-receptor subunit is required for the assembly and trafficking of functional ion channel complexes to the plasma membrane, and mutations of Orco and individual IR co-receptor subunits in Drosophila melanogaster disrupts the assembly of functional receptors (16,17).In the case of Aedes aegypti, there are 116 ligand-selective ORs and 132 ligand-selective IRs (18).These expansive gene families of chemosensory receptors enable mosquitoes to detect various human-emitted cues.ORs generally respond to esters, alcohols, ketones, and aldehydes, while IRs have very flexible ligand tuning and have been shown to detect acids, amines, and physical stimuli, such as humidity and temperature (19).Due to this co-receptor configuration, the mutation of a single co-receptor gene can significantly impair an insect's ability to detect entire classes of human-emitted sensory cues.
With the advent of genome engineering in this non-model organism (20,21), it has been possible to show that genetic perturbations that disrupt each sensory modality in isolation have modest effects on mosquito attraction to humans.Mosquitoes lacking the CO2 receptor Gr3 continue to exhibit attraction to humans under semi-field conditions (13).Mosquitoes with a loss-of-function mutation in Orco retain attraction to humans while losing a strong preference for humans over non-human animals, as well as becoming insensitive to the volatile effects of the insect repellent DEET (N,N-diethyl-meta-toluamide) (20).Ir8a mutants exhibit severe deficiencies in detecting lactic acid, a major component of human skin odor, but retain partial attraction to humans (22).Ir76b mutants remain attracted to humans despite losing their ability to feed on blood (23).Ir21a and Ir93b mutants experience reduced heat and humidity-seeking behavior but maintain their overall attraction to humans (24,25).The recent discovery of extensive co-expression of ORs and IRs in single sensory neurons may explain this functional redundancy (26,27).In addition, out of many sensory cues emitted by humans, detection of two cues is sufficient to initiate and synergize the drive for human-seeking behavior (13).
In this study, we discovered and investigated cross-modal sensory compensation between olfaction and thermosensation in the context of human-seeking behavior.We discovered that mutating the three IR co-receptors leads to different effects on thermotaxis behavior, ranging from complete loss of heat seeking (Ir25a), to a shift in preference to lower temperatures (Ir76b), or enhancement in sensitivity to heat (Ir8a).Orco mutant mosquitoes, which lack a functional OR pathway, show a remarkable increase in attraction to humans, which was fully attributable to their heightened sensitivity to heat.Contrary to our current understanding of insect thermosensation that points to the antenna as the major heat-sensitive organ, we discovered that mosquitoes employ their forelegs as an essential sensory structure for heat detection.The heightened heat-seeking behavior in Orco mutants can be traced back to enhanced thermosensitivity in these foreleg neurons.Comparative gene expression studies between wildtype and Orco mutant legs showed that both Ir25a and the ligand-specific IR subunit Ir140 were upregulated in Orco mutant legs.Finally, we determined that Ir140 plays a significant role in general heat-seeking behavior and that the enhanced thermosensitivity in Orco mutants is lost in Ir140 mutants.Our results strongly support the notion that mosquitoes possess the capability for cross-modal sensory compensation, and that this compensation is due to long-range communication between head appendages and the foreleg.This novel mechanism allows these dangerous vector insects to maintain their overall effectiveness in human-seeking behavior even when one sensory modality is compromised.

Ionotropic receptors tune mosquito thermotaxis behavior
Our work began with the aim to identify genes involved in mosquito thermal attraction at the specific thermal range of human skin temperature (34 o C-37 o C).Our search used a previously described assay to model heat-seeking behavior by monitoring mosquitoes landing on a warmed Peltier element in a cage supplemented with CO2 (13,28).This system can measure the attraction of mosquitoes to thermal stimuli by heating the Peltier element to a temperature ranging from ambient (26°C) to noxious (60°C) temperatures, and quantifying landing events.(Figure 1B, C, Supplemental Video S1).Using this system, we took a candidate gene approach by measuring thermotaxis behavior of mosquitoes carrying loss-of-function mutations in the three major IR co-receptors, Ir25a, Ir76b, Ir8a.
First, we looked at a broadly expressed IR co-receptor subunit, Ir25a (Figure 1D) (29).Ir25a is required for detecting acids, amines, humidity, cooling, and temperature-synchronized circadian clock oscillation in Drosophila melanogaster (16,(30)(31)(32)(33)(34).Ir25a also mediates amine detection and human odor attraction in Aedes aegypti (26,35).The role of Ir25a in thermotaxis behavior in any mosquito species has not been examined.Using the heat-seeking assay, we found that Ir25a mutants failed to locate the heated Peltier element at all temperatures tested (Figure 1E).Ir25a mutants display normal CO2-evoked flight activity and noxious heat detection, suggesting that the failure to respond to heat is not due to locomotor deficits or a lack of behavioral participation (Supplemental Figure S1).These results indicate that Ir25a is required to detect all physiologically relevant temperatures.We next looked at another broadly expressed IR co-receptor subunit, Ir76b (Figure 1F).Ir76b is required for the detection of pH, amino acids, fatty acids, and salts in Drosophila melanogaster (36)(37)(38)(39) and mediates amine detection and blood-feeding behavior in Anopheles coluzzii (23).The role of Ir76b in Aedes aegypti thermosensation has not been examined.We tested Ir76b mutants using the heatseeking assay and found that these mutants showed normal responses to temperatures above 40°C but were more sensitive at lower temperatures (Figure 1G).This effect was specific to lower temperatures between 28.5°C and 31°C.These results suggest that Ir76b may be required to tune Aedes aegypti temperature preference towards human skin temperature.Finally, we looked at the antenna-enriched IR co-receptor subunit, Ir8a, which is required for lactic acid detection in Drosophila melanogaster and Aedes aegypti (16,22,40) (Figure 1H).Unlike Ir25a and Ir76b mutants, Ir8a mutants displayed enhanced heat-seeking behavior broadly across most temperature ranges tested (Figure 1I).Our analysis of heatseeking behavior in these three IR co-receptor subunit mutants revealed distinct thermotaxis patterns for each mutant, implying that each IR co-receptor subunit contributes uniquely to the precise regulation of heat-seeking behavior in Aedes aegypti.

Orco mutants display enhanced heat-seeking behavior
We next asked whether the Orco co-receptor is required for heat-seeking behavior in Aedes aegypti.In Aedes aegypti, Orco is mainly expressed in the antenna, proboscis, and maxillary palp (Figure 2A) (20).We tested Orco mutants using the heat-seeking assay to determine if ORs, like the IRs, had any signs of tuning the overall heat-seeking behavior in Aedes aegypti.Using the heat-seeking assay, we found that Orco mutants displayed enhanced heat-seeking behavior compared to their genetic controls at 36°C (Figure 2B, Supplemental Video S2).This enhancement was sustained during the entire duration of the heat stimulus period (Figure 2C).Furthermore, Orco mutants showed enhanced heat-seeking behavior, specifically towards human skin temperatures, and displayed normal avoidance behavior at the noxious temperature range (Figure 2B, D).
Next, we investigated which aspect of mosquito heat-seeking behavior was specifically altered in Orco mutants.We first tested whether Orco mutants spent longer on the heated surface by calculating the average dwell time per landing event for each animal at human skin temperature.Our analysis revealed that Orco mutants spent equal time as controls on the Peltier element per landing event (Figure 2E).Another possibility was that Orco mosquitoes engaged in landing events more frequently than their genetic controls.We analyzed the final 60 seconds of each stimulus period and calculated both landing and take-off event frequencies for each animal.We found that Orco mutants display increased landing and take-off frequencies, demonstrating that Orco mutants had persistent and sustained engagement towards the heated Peltier element throughout the heat trials (Figure 2F, G).
Given the dramatic increase in the attraction of Orco mutants to heat, we next asked if the insect repellent DEET deters these mutants from heat-seeking behavior.Unlike wildtype animals, Orco mutants are attracted to human odor in the presence of DEET, however whether volatile DEET interferes with heat-seeking independently is not known (20).We designed a modified version of the heat-seeking assay by placing a source of DEET in front of the Peltier device set at 40°C (Figure 2H).This source of DEET was placed across a spacer to avoid direct contact of DEET contact by the mosquitoes.Contact chemorepellency is mediated by the taste system of the mosquito leg, and does not require Orco function (41).As expected, we found that DEET was sufficient to drive mosquitoes away from heat in wildtype and heterozygous controls.However, Orco mutant mosquitoes retained their enhanced heatseeking behavior even in the presence of DEET (Figure 2I).We replicated this result with another heteroallelic mutant, Orco 2/5 (Figure 2I).These data indicate that enhanced heatseeking is unaffected by Orco-mediated repellency, thus depending on the absence of Orco.Additionally, these data demonstrate the critical role of Orco in DEET detection and avoidance behavior in the presence of multi-modal human-related cues (20).

Orco mutants display enhanced human-seeking behavior
We next investigated sensory cues of different modalities to test the specificity of the enhanced thermotaxis behavior displayed by Orco mutants.Using the CO2 activation assay, we tested the activity and arousal of Orco mutants in response to a brief 20-second pulse of CO2 (Figure 3A, Supplemental Figure S2A).CO2-evoked flight activity and total distance traveled in response to CO2 were indistinguishable from genetic controls (Figure 3B, Supplemental Figure S2B,C).We further tested whether attraction toward human odor was altered in Orco mutants using the nylon-next-to-cage assay (Figure 3C, E).A previous study showed that Orco mutants show normal attraction to humans but are impaired in discriminating human from non-human odors (20).Consistent with these previous observations, Orco mutants show normal attraction towards human-worn nylon (Figure 3E, F).This attraction was specific to host odors because mosquitoes were not attracted to unworn nylon (Figure 3C, D).We further tested whether the enhanced thermosensitivity of Orco mutants enhances their ability to find live human hosts.To test this, we designed an arm-next-to-cage assay with a human arm positioned close to the cage to allow mosquitoes to detect human-emitted odor, CO2, and body temperature without physical contact (Figure 3G).We found that Orco mutants displayed enhanced human armseeking behavior compared to genetic controls (Figure 3H).We further repeated the experiment in the presence of volatile DEET to test whether the enhanced human-seeking behavior is retained in the presence of this repellent (Figure 3I).We found that Orco mutants were still capable of human arm-seeking behavior in the presence of DEET (Figure 3J).These data, along with the heat-seeking data in the presence of DEET (Figure 2I), suggest that enhancement of Orco human-seeking behavior is unaffected by DEET.

Forelegs mediate thermotaxis behavior
Previous studies in Drosophila melanogaster have identified thermosensitive sensory structures and neurons activated by diverse thermoreceptors and ion channels (Figure 4A) (31,34,(42)(43)(44)(45)(46).In mosquitoes, prior studies identified a set of thermosensitive neurons at the most distal antennal segment that responded to cooling or heating air (Figure 4B, C) (24,25,(47)(48)(49).Since Orco is expressed in the antenna, we hypothesized that loss of Orco may directly affect the function of thermosensitive neurons at the antennal tip.However, none of the previous studies investigated the direct roles and requirements of these thermosensitive neurons in the context of thermotaxis behaviors.To test this, we removed the three most distal antenna segments by cutting off the tip and performed thermotaxis assays (Figure 4D).We first wanted to confirm that antennal tip removal had minimal consequences on the overall behavior of the mosquitoes using an arm-feeding assay where we measured the ability of the antenna tip-cut mosquitoes to blood-feed on a human arm (Figure 4E).Antenna tip-cut mosquitoes engorged on human arms comparable to fully intact animals (Figure 4F).This result confirms that the antenna tip removal had minimal effect on overall human-seeking behavior, consistent with previous reports (50,51).
We then investigated the role of antenna tip neurons in thermotaxis behavior, first using a Glytube assay, an artificial blood-feeding system that provides mosquitoes with warm blood in the presence of CO2 (Figure 4G) (52).Surprisingly, we found that antenna tip-cut animals still engorged on blood at comparable levels to fully intact mock-treated animals (Figure 4H).We then turned to the heat-seeking assay to monitor heat-seeking behavior during each heat stimulus.Consistent with Glytube assay results, antenna-tip-cut mosquitoes displayed normal heat-seeking behavior, indistinguishable from fully intact mock-treated controls at all temperatures tested (Figure 4I).Because our data show that the first three segments of the antennal tip are dispensable for various thermotaxis tasks, we wanted to know whether any portion of the antenna was required to detect thermal information.Since complete antenna removal disables mosquitoes from flying (50,51), we turned to the opto-thermocycler assay (53).This assay tracks thermotaxis behavior, using probing movement as a proxy for heat detection at high temporal and spatial resolution without requiring flying behavior (Figure 4J).The antenna tip-cut animals displayed probing behavior indistinguishable from mock-treated controls, consistent with our other thermotaxis assays.Furthermore, animals with the entire antenna removed also detected and responded by displaying probing behavior to a thermal stimulus indistinguishable from fully intact mock-treated controls (Figure 4K).These experiments suggest that the antenna tip and even the entire antenna is dispensable for heat detection in Aedes aegypti.
We further tested the possibility that other sensory appendages are responsible for mosquito thermotaxis behavior.We focused on the legs because these sensory appendages make direct contact with heated surfaces, and a previous study showed that Drosophila melanogaster leg neurons responded directly to thermal stimuli (31).Ticks also have a unique sensory organ on their distal leg segment called Haller's organ that functions as a heat sensor (54).To determine whether legs contribute to thermotaxis behavior in Aedes aegypti, we removed the three most distal segments from each pair of legs -forelegs, midlegs, and hindlegs (Figure 4L).Cutting the tips of each pair of legs had no significant effect on humanseeking and blood-feeding behavior on a live human arm (Figure 4M).However, animals with foreleg tips cut had a significant reduction in their ability to engorge on blood in the Glytube assay, where the only cues presented to these animals were heat and CO2 (Figure 4N).Furthermore, animals with foreleg tips cut had a significant reduction in heat-seeking behavior to the Peltier element set to 36 o C (Figure 4P) but not at ambient (Figure 4O) or at a noxious temperature (Figure 4Q).These data suggest that mosquito forelegs significantly contribute to Aedes aegypti thermotaxis behavior.

Neuroanatomy of the Aedes aegypti legs
The neuroanatomy of Aedes aegypti legs has not been extensively studied.Insect legs comprise multiple segments -coxa, trochanter, femur, tibia, and tarsus.We focused on the tarsus, explicitly looking at the fifth tarsomere (Figure 5A), which is the most distal tarsal segment and is considered the primary site for chemoreception in the leg (55,56).We characterized the neuronal anatomy of the fifth tarsomere using a panel of chemosensory receptor driver lines, each expressing a fluorescent reporter, dTomato, in its corresponding neural population (Figure 5B-E).Specifically, we used brp-QF2w > QUAS-dTomato to label all neurons (57), Ir25a-QF2 > QUAS-dTomato and Ir76b-QF2 > QUAS-dTomato to label putative thermosensitive and chemosensitive cells (26), and Gr4-QF2 > QUAS-dTomato to label sugarsensing neurons previously shown to induce appetitive behavior once activated (58).Aedes aegypti tarsi are densely covered with scales, and the neuronal cell bodies extended their dendrites into sensilla that decorate the leg.
Using the brp-QF2w > QUAS-dTomato reporter line, we quantified the number of neurons within the fifth tarsomere of each leg and observed a decreasing order of neuron count from forelegs to midlegs and hindlegs (Figure 5F,G).We identified an average of 75 total neurons in the fifth tarsomere of the foreleg, 62 in the fifth tarsomere of the midleg, and 47 in the fifth tarsomere of the hindleg (Figure 5B,G).We then examined reporter expression for Ir25a, Ir76b, and Gr4.Ir25a-expressing neurons were found in roughly two-thirds of the total fifth tarsal neurons, with a decreasing number in order of forelegs, midlegs, and hindlegs (Figure 5C,G).Ir76b-expression was less abundant, in one-third of the neurons of the forelegs and midlegs and, to a lesser extent, in the hindlegs (Figure 5D,G).In contrast, expression of the Gr4 sugar receptor was equivalent in all three legs (Figure 5E,G).
In our images, we observed and quantified dendritic processes from cell bodies into their respective sensilla (Figure 5H).As with the total neural count, the total sensilla count (brppositive) was much higher in the fifth tarsomere of forelegs and midlegs compared to hindlegs (Figure 5B, I).The number of Ir25a-positive sensilla count mirrored that of brp-positive sensilla count, suggesting that Ir25a-positive neurons innervate all or most chemosensory sensilla (Figure 5C, I).Ir76b-positive sensilla represented half of the total sensilla counts in all three legs (Figure 5D, I).The Gr4-positive sensilla count represented half of the total brp-positive sensilla counts in the forelegs and midlegs and had roughly the same number of hindleg sensilla (Figure 5E, I) (26).
We further investigated the spatial distribution of each chemoreceptor-expressing neuron within the fifth tarsomere.We divided the fifth tarsomere into 50 µm regions and counted the neurons in each region (Figure 5J).Ir25a-expressing neurons were broadly distributed across the entire fifth tarsomere while Ir76b-expressing neurons were more concentrated in the proximal tarsal region.Gr4-expressing neurons were primarily located in the distal area of the fifth tarsomere (Figure 5K-M).
We noticed the neurons were organized in two planes on each side of the tarsus (Figure 5N, Supplemental Video S3).To our knowledge, this has not been described in other insects.When the mosquito leg is outstretched, one side of the tarsus faces toward the head and the other side is oriented toward the caudal end of the abdomen.Thus, we named these two neural planes the "cranial" and "caudal" sides of the tarsus and quantified the number of labeled neurons per plane for each reporter line.While the overall number of foreleg neurons of the fifth tarsomere in each plane was similar, Ir25a and Ir76b represented a higher proportion of neurons on the caudal side compared to the cranial side, while Gr4-expressing neurons displayed an even distribution between the two sides (Figure 5O).This cranial-caudal distribution was similar in the midleg (Figure 5P).

Mosquito leg neurons respond to heat
We next asked which neurons in each of the main Aedes aegypti sensory appendages responded to heat.We used the pan-neuronal imaging line brp-QF2w > QUAS-dTomato-T2A-GCaMP6s to measure Ca 2+ response as a proxy for neuronal activity in response to heated air in the four main sensory appendages -antenna, maxillary palp, proboscis, and legs (Figure 6A).We developed a functional imaging setup that allowed us to acquire GCaMP and dTomato fluorescence signals while applying heated air over the samples.Dual channel imaging is critical to measuring heat-evoked Ca 2+ activity due to the intrinsic thermosensitive properties of fluorescent proteins.We report the change in the ratio of the GCaMP signal over the dTomato signal (R = GCaMP6s/dTomato) normalized to baseline (ΔR/R) to correct for any movement artifact as well as temperature-evoked intrinsic changes in fluorescence signals as previously described to study thermosensitivity in C. elegans thermosensory neurons (59, 60) (Figure 6B, Supplemental Figure S3A-C).
We first tested whether foreleg neurons responded to heated air.Consistent with our behavioral data, foreleg neurons throughout all three distal segments responded robustly to heated air (Figure 6C).These heat-evoked foreleg neural responses were reproducible within each trial (Supplemental Figure S3D).This response was independent of any mechanical artifact from the valve switching between two air outlets, because neuronal response returned after the heat source was turned back on (Supplemental Figure S3E, F).We then tested the specificity of this heat-evoked neuronal response in other sensory structures.We could not detect Ca 2+ activity in the antenna or maxillary palp (Figure 6D, E), while the proboscis had weak responses to heated air (Figure 6F).By far, the most robust heat-evoked activity was seen in foreleg neurons (Figure 6G).These functional imaging data further suggest the critical role of foreleg neurons, and the dispensability of the antenna, during thermotaxis behavior.Now that we identified the primary sensory appendage for heat detection, we tested whether the foreleg is the primary site for enhanced thermosensitivity in Orco mutants.To test this, we generated heteroallelic Orco mutant pan-neuronal imaging mosquitoes by crossing two different Orco mutant alleles into the brp-QF2w > QUAS-dTomato-T2A-GCaMP6s imaging background strain (Figure 6H).We modified the imaging setup to apply a slow heat ramp using a single Peltier outlet to capture any heat-evoked activation threshold differences in Orco mutant leg neurons.Using these Orco mutant pan-neuronal GCaMP mosquitoes and the modified imaging setup, we found that both heterozygous and Orco knockout mutant midlegs, although indistinguishable from each other, displayed higher peak response and total activity in response to heated air than wildtype controls (Figure 6I-K).This heterozygous effect was unexpected because Orco typically displays a fully recessive phenotype (20).In contrast, peak amplitude and total activity were enhanced in Orco knockout forelegs compared to both wildtype and heterozygous controls (Figure 6L-N).These data suggest that the enhanced thermosensitivity in the foreleg neurons underlies the increased thermotaxis behavior of Orco mutants.

Upregulation of Ionotropic Receptor genes in the Orco mutant legs
We next investigated the potential molecular mechanisms underlying enhanced neuronal activity in Orco mutant foreleg neurons.We performed bulk RNAseq from pooled whole forelegs and whole midlegs isolated from wildtype and Orco mutant mosquitoes (Figure 7A).Using an FDR cut-off of 0.01, we looked for differentially expressed genes, their expression level, and the degree of differential expression (Figure 7B).We found 614 upregulated and downregulated genes in this Orco mutant foreleg and midleg transcriptome dataset.
We narrowed our search by focusing on known sensory receptor genes.Twenty-two out of 32 pickpocket genes were detected, one upregulated and one downregulated in Orco mutant legs (Figure 7C).For TRP channels, 13 of 16 genes were detected, with only water witch (wtrw) upregulated in Orco mutant legs (Figure 7D).Odorant receptors (ORs) were mainly absent in the legs, with only 7 out of 116 detected, and none were differentially expressed in Orco mutant legs (Figure 7E).Orco transcripts were still detected in our Orco 5/16 samples, but at a lower level compared to wildtype controls with the expected 5-bp and 16-bp deletion in the mapped reads at each mutant allele locus, likely reflecting transcripts undergoing nonsensemediated decay (Supplemental Figure S4A).At the gene-wide comparison level, Orco failed to meet our stringent FDR < 0.01 cut-off but still displayed a strong trend towards significant downregulation in the Orco 5/16 samples with adjusted P-value at 0.028 and FDR < 0.05 (Supplemental Figure S4B).Only 20 of 72 gustatory receptors (GRs) were expressed, with two upregulated and one downregulated in Orco mutant legs (Figure 7G).In contrast, 78 out of IRs were expressed, with eight upregulated, and ten downregulated in Orco mutant legs.These transcriptome data suggest that in the legs, IRs are under stronger regulation due to Orco mutation than other chemosensory receptor classes.The number of normalized read counts is presented as size, differentially expressed genes are represented with bold, colored outlines, and the degree of differential expression is represented as a color gradient.Read counts were normalized using DESeq2 median of ratio method.(C-G) Gene expression comparisons between wildtype and Orco 5/16 mutant mosquito tarsi for Pickpocket genes (C), TRP channels (D), Odorant Receptors (E), Gustatory Receptors (F), and Ionotropic Receptors (G).Genes are listed from left to right from lowest to highest fold-change enrichment in Orco 5/16 tarsi compared to wildtype.(H) Orco 5/16 upregulated Ionotropic receptor gene expression in the indicated tissues from the female leg neurotranscriptome dataset (29).Data labeled with different letters are significantly different (n = 10-14 replicates/leg, FDR < 0.01, pairwise DESeq2 comparison).
To further narrow down the potential genes underlying the enhanced heat-evoked neuronal activity in Orco mutant forelegs, we reanalyzed the previously published Aedes aegypti neurotranscriptome dataset and quantified which Orco-upregulated IR genes were also enriched in the forelegs.From our reanalysis, only Ir25a and Ir140 were enriched in the foreleg (Figure 7H).This enrichment was specific to female samples as the male transcriptome dataset detected both Ir25a and Ir140, but the foreleg enrichment was not as profound as the female samples (Supplemental Figure S4C).The Ir25a expression data are consistent with our Ir25a heat-seeking behavioral data, suggesting that Ir25a is a core co-receptor subunit required for thermotaxis behavior (Figure 1E).

Ir140 mediates heat-seeking behavior and enhanced Orco thermosensation
To test the role of Ir140 in thermosensation, we used an enhanced method of CRISPR-Cas9 genome editing that builds on our original method (21) to generate Ir140 mutant mosquitoes that lacked functional Ir140 ligand-specific IR subunits.We isolated two Ir140 mutant alleles, Ir140 144 and Ir140 17 , and used the heat-seeking assay to characterize thermotaxis behavior of heterozygous animals and heteroallelic Ir140 17/144 null mutants (Figure 8A, Supplemental Figure S5A-C).Ir140 mutants displayed a significant reduction in heat-seeking behavior compared to their genetic controls only when the Peltier was set at 36°C (Figure 8B).These data suggest that Ir140 is a key molecular player in Aedes aegypti thermotaxis behavior at human skin temperatures.
We further tested the possibility that Ir140 upregulation could underlie the enhanced thermosensitivity in Orco mutants.Because Orco and Ir140 are tightly linked on the 3 rd chromosome, it was not feasible to simply cross Ir140 mutant alleles into the Orco mutant background because of the low likelihood of recovering recombination events between the loci.Therefore, we generated Orco, Ir140 double mutants by targeting Ir140 using CRISPR-Cas9 genome editing in the Orco 16 mutant background strain (Figure 8A, Supplemental Figure S5D-F).We isolated two Orco, Ir140 double mutant alleles, Orco 16 , Ir140 14 and Orco 16 , Ir140 49 , and tested the thermotaxis behavior of the heteroallelic Orco 16/16 , Ir140 14/49 double mutants using the heat-seeking assay.While Orco 16/16 , and the heterozygous mutants displayed enhanced heat-seeking behavior, heteroallelic Orco 16/16 , Ir140 14/49 double mutant animals failed to show this enhancement and were indistinguishable from wildtype controls (Figure 8C).These results indicate that Ir140 is one of the key molecular players involved in general thermosensitivity and is required for the enhanced thermosensitivity displayed by Orco mutant mosquitoes.

Cross-modal sensory compensation enhances mosquito attraction to humans
In this study, we establish a novel sensory compensation mechanism in Aedes aegypti mosquitoes.Our findings demonstrate that Orco mutant mosquitoes exhibit heightened thermosensitivity due to increased neuronal activity in their legs.We further show that many ionotropic receptors were differentially expressed in Orco mutants legs.Finally, our study reveals the importance of Ir140, a ligand-specific IR subunit, in normal heat-seeking behavior and its necessity for the heightened thermosensitivity observed in the Orco mutants (Figure 8D).This cross-modal sensory compensation adds to the repertoire of behavioral strategies Aedes aegypti can utilize for their persistent and seemingly unbreakable drive to seek humans.These observations are consistent with prior work showing that disruptions in specific sensory receptors caused reductions in mosquito behavior related to the affected sensory modality but did not eliminate overall attraction to humans (13,24,25,35).Co-expression of multiple chemosensory receptor gene families within a single olfactory sensory neuron may permit mosquitoes to rely on at least one chemoreceptor gene family if others become compromised (26).It is not known if receptor co-expression is seen in sensory neurons in the leg, the primary thermosensitive organ of the mosquito, but the recent development of techniques for single-cell RNA sequencing of mosquito sensory neurons (26) may shed light on this.

Conservation of temperature detection between Drosophila and mosquitoes
Thermosensation has been extensively explored in Drosophila melanogaster flies, leading to the identification of numerous genes and sensory organs in detecting cooling and heating temperatures.For noxious heat detection, painless, which is expressed throughout the nervous system, is essential for the detection of harmful heat at all life stages (61,62).Pyrexia is required for noxious heat-evoked paralysis behavior (63).Innocuous warmth detection involves genes including TrpA1, required for warmth detection by AC neurons in the brain, controlling the slowly developing preference response of flies exposed to a shallow thermal gradient (45,64).The gustatory receptor Gr28b(d) is necessary for rapid noxious heat avoidance (42,43).Detection and avoidance of innocuous cool temperatures relies on brv1, brv2, and brv3 expression in cold receptors located in the arista (44) and Ir21a, along with Ir93a and Ir25a in larval dorsal organ cool cells, and in cooling cells located in the adult arista (65).Iav in the chordotonal organ (66) and Trp and Trpl mutants also display deficits in cool avoidance (64).Moreover, Ir93a and the IR co-receptor Ir25a mediate the detection of warm and cool temperatures (31,32).In addition to ion channels, opsins have been shown to mediate larval temperature preference behavior through rh5 and rh6 signaling through the Gq/PLC pathway coupled to the transduction channel TrpA1 (67,68).
Recent studies have identified genes conserved from Drosophila melanogaster that mediate cooling detection in mosquitoes.In Anopheles gambiae, Ir21a and Ir93b are expressed in sensory neurons at the antennal tip.Mutating these genes significantly reduced neuronal activity in response to cooling air and deficits in heat-seeking behavior at 37°C (24,25).In Aedes aegypti, Ir93a-expressing neurons in the antenna are thermosensitive, but the role of this ligand-selective IR in heat-seeking behavior remains to be tested (25).In contrast to cooling detection, the functional similarity between Drosophila melanogaster and mosquito heat detection is not well conserved.Unlike Drosophila melanogaster TrpA1 mutants, Aedes aegypti TrpA1 mutants displayed normal thermotaxis behavior in the range of human skin temperature between (26-40°C) but are defective in avoidance of temperatures greater than 40°C, suggesting that TrpA1 is a noxious heat sensor in mosquitoes (28).Additionally, Gr19, the Aedes aegypti homologue of the Drosophila melanogaster Gr28b(d) rapid heat avoidance gene, is not required for thermotaxis behavior (28).These findings collectively indicate that while studies in Drosophila melanogaster have contributed significantly to our understanding of thermosensation, mechanisms in the fly do not directly translate to the mosquito.This is understandable given the marked difference between the species in how they utilize thermal information.While mosquitoes are attracted to the high temperature of human skin, Drosophila strongly avoid these high temperatures.Our current work identified Ir140 as the first mosquitospecific gene that mediates thermotaxis behavior in Aedes aegypti.Based on the complete abolishment of heat-seeking behavior in Ir25a mutants, we speculate that additional IR ligandspecific subunits are likely to function as additional heat sensors.

What is the role of the antenna during thermotaxis behavior?
The mosquito antenna has long been regarded as the essential sensory appendage for mosquito thermotaxis behavior.Earlier studies identified a pair of sensory neurons within the small coeloconic sensilla located at the antennal tip that respond to changes in air temperature (24,(47)(48)(49).These studies further revealed that one of these sensory neurons responds to rising air temperatures, while the other is sensitive to decreasing temperatures (47)(48)(49).However, whether the thermosensitive properties of these sensory neurons have any influence over mosquito thermotaxis behavior has never been investigated.An early study demonstrated that removing the antenna in Aedes aegypti had minimal impact on their ability to seek a combination of human odor and heat unless more than ten distal antenna segments were removed (50).Consistent with these findings, a recent study found that removing the distal six antenna segments in Aedes albopictus had little effect on blood-feeding on a human arm (51).In this study, we found no evidence that the antenna is required for heat-seeking behavior or that the antenna responds to heat.This is contrary to a recent study showing that Aedes aegypti antennal Ir93a neurons are thermosensitive (25).Although the basis for differences in observations in these two studies is unknown, our data strongly indicate that the thermosensitivity of the antenna tip neurons is not necessary for mosquito thermotaxis behavior.The exact functions of antennal tip thermosensory neurons, if they exist, remains to be determined.

Aedes aegypti tarsi as heat sensors
Insect legs are one of the primary sites for taste detection.In Drosophila melanogaster, tarsal sensory neurons detect various non-volatile chemosensory cues, including sugars, bitters, sours salts, fatty acids, water pH, and pheromones (38,39,55,69).In Aedes aegypti, tarsal neurons play a crucial role in detecting water and salt for oviposition site assessment and in responding to DEET contact repellency (41,70,71).In this study, we described the first observation of heat-induced activation of tarsal sensory neurons and established a role for the foreleg distal tarsal segments in heat-seeking behavior.On a behavioral level, our findings demonstrate that removing a pair of distal tarsal segments in the forelegs, as opposed to those in the mid-or hindlegs, significantly impairs the mosquito's ability to detect and land on heated surfaces.At the level of neuronal activity, we show that heat activates a substantial subset of tarsal sensory neurons spanning the entire length of the three most distal segments of both forelegs and midlegs.Furthermore, we present evidence of selective enhancement in heatevoked neuronal activity within the forelegs of Orco mutant animals.These findings indicate the critical role foreleg tarsal neurons play in Aedes aegypti thermotaxis behavior.
Removing pairs of tarsal segments from the forelegs significantly reduced heat-seeking and engorgement on the artificial blood feeder, though the behaviors were not completely abolished.We speculate that this residual behavior may be attributed to the redundant thermosensory abilities of the neurons in the midlegs, given that these neurons also exhibited heat-evoked neuronal activity.We could not obtain functional imaging data in the hindleg tarsi due to high background fluorescence in this tissue.Thus, the contribution of hindlegs to heat detection remains to be determined.As established in previous reports, heat dissipates rapidly from its source (25,72).Therefore, mosquitoes must be close to the surface to detect the presence of heat and swiftly assess the surface temperature.The behavioral aspects of mosquito landing events have received limited attention, and future studies using highresolution behavioral tracking approaches should investigate how each sensory appendage is utilized during heat-seeking behavior.

IRs regulate mosquito thermosensitivity toward host skin temperature
Our data reveal that functional IR complexes, formed from a unique combination of IR ligandspecific subunits along with co-receptor subunits, contribute to various aspects of heat-seeking behavior.We found that Ir25a is required for thermotaxis behavior across all temperature ranges.This suggests that a functional IR complex containing Ir25a is crucial for thermosensation.On the other hand, Ir76b mutants exhibited heightened sensitivity to milder heat while retaining wildtype responses to higher temperatures.The heightened sensitivity towards temperatures lower than human skin temperature implies that Ir76b-containing IR ion channel complexes are involved in adjusting the preferred temperature to a host-relevant range.The mechanism by which such adjustment could be made is unknown, but we speculate that Ir76b contributes to a thermosensitive IR complex that is insensitive to lower temperatures, and that its loss releases inhibition on the complex, activating positive thermotaxis.
Ir8a mutants display yet another unique heat-seeking behavior, showing enhanced heatseeking behavior throughout the entire temperature range.This enhancement in Ir8a mutants is similar to the heat-seeking behavior seen in Orco mutants, except that this enhancement extends across a wider temperature range and is not specific to human skin temperature.Ir8a, like Orco, is selectively expressed in sensory neurons in head appendages.It will be interesting to learn if Ir8a mutants also show selective upregulation of IR subunits in the legs, or whether Ir8a mutants utilize another mechanism of sensory compensation.We speculate that Ir8a mutants enhance their heat-seeking behavior to compensate for their loss of lactic acid detection, a major chemical component of human sweat odor.Interestingly, Ir8a mutants also show enhanced water-seeking behavior while searching for an egg-laying site (22).These data suggest that behavioral enhancement extends beyond human-seeking behavior to encompass other crucial tasks necessary for the propagation of their offspring.Future studies are needed to investigate whether the IR signaling pathways tune additional sensory modalities.
What is the signaling mechanism underlying mosquito sensory compensation?Sensory compensation has traditionally been associated with localized changes in synaptic connections and strength (8), reorganization of sensory maps (6), direct neuromodulation at the synapse by opposing sensory modalities (9), and the release of neuromodulators from remote brain structures (10,11).However, we believe these mechanisms are unlikely to play a role in the case of Orco mutants, given the presence of multiple synapses between the sensory appendages in the head and the affected neurons in the leg.Head appendage sensory neurons synapse in the brain while leg sensory neurons synapse in the ventral nerve cord, the equivalent of the vertebrate spinal cord.
We speculate that this form of sensory compensation arises from a non-cell autonomous mechanism independent of Orco expression in the legs.First, Orco-expression is relatively low compared to other classes of chemosensory ion channels.Using normalized read counts from DESeq2 analysis of foreleg and midleg bulk RNA-seq, we found very low levels of the head appendage co-receptors Orco (83±40) and Ir8a (88±7), compared to the IRs that we speculate play a role in heat-seeking in the leg: Ir25a (17,841±500), Ir76b (15,621±1023), and Ir140 (571 ±14).Moreover, we rarely observed Orco-positive cells when using the Orco-QF2 >QUAS-dTomato reporter line (data not shown), further supporting the hypothesis that the contribution of Orco to the compensatory mechanism is indirect.
We propose several alternative mechanisms that could mediate the form of sensory compensation in our study.One potential mechanism involves long-range excitability changes due to neuromodulation.Our RNAseq data reveals enrichment of upregulated neuromodulatory receptor gene expression in the legs of Orco mutants.These data suggest sustained neuromodulation via circulating hemolymph leading to transcriptional changes and upregulation of thermosensors to enhance sensitivity to heat.Because our functional imaging data involves the examination of detached legs in Orco mutants, acute neuronal modulation is unlikely to be happening.It is also possible that modulation occurs through multi-synaptic modulation by descending neurons from the brain to signal the leg deficit of Orco via the ventral nerve cord.One future approach to probe the possible role of neural activity in sensory compensation would be to acutely genetically silence all Orco-expressing neurons in head appendages and ask if sensory compensation emerges in foreleg sensory neurons.
Another possibility is that these changes occur during development because the Orco mutant is a constitutive knockout, leading to a loss of Orco across all life stages of the mosquito.The early loss of Orco expression and function could have a lasting impact on the neuronal identity and properties of leg sensory neurons.It is worth noting that Orco expression has been detected during the early embryonic stage (73), and transcriptomic analysis has hinted at a regulatory role for Orco expression in ion channel signaling in Aedes aegypti embryos (74).With the increased sophistication of genetic tools in the mosquito, it may be possible in the future to knock out the Orco gene acutely in adult mosquitoes to ask whether sensory compensation occurs, and if it does over what time course after the knock-out of the gene in head appendage neurons.

Molecular identities of thermosensors in the mosquito legs
In this study, we identified Ir140 as a critical mediator of heat-seeking behavior, as demonstrated by the significant reduction in heat-seeking behavior observed in Ir140 mutants.While there is a significant reduction in Ir140 mutant heat-seeking behavior, the behavior is not entirely abolished.Therefore, we speculate that additional IR subunits may form functional ion channels contributing to heat detection.One of the upregulated ligand-specific IR subunits in Orco mutant legs may be among the remaining genes that play a role in mosquito heatseeking behavior.We speculate that Ir140, in conjunction with Ir25a and potentially other IR subunits, forms an ion channel within the tarsal sensory neurons.This ion channel complex could serve as either a direct thermosensor or function downstream in the signal transduction pathway of the thermosensory neurons.Further investigation into their biophysical properties using a heterologous expression system is needed to determine whether these ion channel complexes are intrinsically thermosensitive.

Concluding remarks
In this work, we have discovered a novel mechanism of sensory compensation in the mosquito involving a transcriptional response in the tips of the foreleg in response to the loss of the major olfactory co-receptor in head appendages.We further show that the leg, not the antenna, is the behaviorally relevant appendage for sensing heat during human host seeking.Taken together this work highlights the extraordinary ability of the female mosquito to retain attraction to humans, whose blood is key to the generation of her offspring.
to recover overnight at 26°C with 70-80% relative humidity and fasted by replacing 10% sucrose with deionized water.A human forearm was placed adjacent to one side of the cage, separated by two 10 mL serological pipet tips, and perpendicularly to the arm.This arm positioning allowed mosquitoes to detect human odor, CO2, and heat while preventing the mosquitoes from directly making skin contact.A monochrome CMOS camera (The Imaging Source, DMK37BUX178) with a 12 mm/F1.8lens (Edmond Optics, #33-303) was positioned to take images of mosquitoes responding to the human arm.Trials ran for 5 minutes, and images were acquired at one frame/second using the IC Capture Image Acquisition software (The Imaging Source).We manually counted the number of mosquitoes resting overlaying the human arm to quantify mosquito responses.For the human arm experiment with DEET, 10% DEET (N,N-Diethly-3-methylbenamide; Sigma D100951, mixed with 100% ethanol, v/v) was applied on the entire length of the forearm surface facing the cage.Applied 10% DEET was allowed to dry for 10 minutes before each trial.After drying, behavior experiments were performed as described above.

Nylon-next-to-cage assay
This assay is a modified version of the arm-next-to-cage assay as above.Instead of a live human arm, a clear 80well microcentrifuge test tube rack was inserted into a worn or unworn nylon sleeve.Human odor collection using nylon sleeves was described previously (35).The nylon sleeves were kept taut using binder clips.This nylon sleeve stimulus was further mounted to a ring stand and placed adjacent to one side of the cage, ensuring that it did not make direct contact with the cage.An eighty-well microcentrifuge test tube rack was used to mimic the surface area of the human arm that mosquitoes encountered during the arm-next-to-cage assay (6.5 cm x 22.5 cm surface).Image acquisition was performed as described in the arm-next-to-cage assay.

CO2 activation assay
Twenty to 40 adult female mosquitoes were sorted under cold anesthesia (4°C) and transferred to a 28 cm 3 cage and allowed to recover overnight at 26°C with 70-80% relative humidity and fasted by replacing 10% sucrose with deionized water.Sorted mosquitoes were transferred into a custom-made Plexiglass box (30 cm 3 ), with carbonfiltered air pumped continuously into the box via a diffusion pad installed on the ceiling of the enclosure.Mosquitoes were allowed to acclimate for 10 minutes.Following acclimation, a 20-second pulse of CO2 was applied, and flying activity was monitored for 10 minutes.A monochrome CMOS camera (The Imaging Source, DMK37BUX178) with a 12 mm/F1.8lens (Edmond Optics, #33-303) was positioned to acquire images at one frame/second using the IC Capture Image Acquisition software (The Imaging Source) of the entire behavior box to capture of mosquitoes flying in response to CO2.

Heat-seeking assay
Experiments were performed as previously described (13,28).Briefly, 40 -50 adult female mosquitoes were sorted under cold anesthesia and allowed to recover overnight at 26°C with 70-80% relative humidity and fasted by replacing 10% sucrose with deionized water.Before each trial, sorted and fasted mosquitoes were transferred into a custom-made Plexiglass box (30 cm 3 ), with carbon-filtered air pumped continuously into the box via a diffusion pad installed on the enclosure's ceiling and each stimulus period lasted 3 minutes on a single Peltier element (6 × 9 cm, Tellurex).The surface was covered with 15 × 17 cm standard white letter-size printer paper (NMP1120, Navigator) and held taut by a magnetic frame.CO2 pulses (20 seconds, to > 1000 parts per million above background levels) were added to the air stream and accompanied all heat stimulus period onsets.Mosquito landings on the Peltier were monitored by fixed cameras (FFMV-03M2M-CS, Point Grey Research), with images acquired at one frame/second.Images were analyzed using custom MATLAB scripts to count mosquito landings within a fixed target region.Mosquito occupancy on the Peltier was quantified during seconds 90-180 of each stimulus period.

Heat-seeking assay with DEET
To test if DEET could disrupt mosquito attraction without host odor, we tested the female mosquito's thermotaxis to a 40˚C Peltier element in the presence or absence of DEET.A standard 28 x 28 x 28 cm rearing cage (Bioquip) was placed inside a clear vinyl bag.The bag and cage were positioned adjacent to a PCR machine so that the Peltier element was approximately 1 cm from the cage screen.A Firefly MV camera was positioned to take stereotyped images of mosquitoes responding to the heat source (Point Grey Research).A Flypad (Flystuff.com)was placed inside the bag, on top of the cage, as a source of carbon-filtered air and CO2.The assay was assembled on a metal peg board to maintain the fixed position of the components: a rearing cage inside a vinyl bag, a camera, and a PCR machine.Ten to thirty minutes prior to the start of the assay, 25 adult female mosquitoes were released into each cage to acclimate.Before starting each trial, 100 µl of 100% ethanol or 10% DEET in ethanol was applied to a 2 cm x 6 cm strip of Whatman filter paper (GE Healthcare).The treated paper was hung from a rectangular plastic frame in front of the Peltier element, so it was not in contact with the cage screen.For each trial, carbon-filtered air was pumped into the assay for 60 minutes to remove the host odor.After the 60-minute pre-stimulus period, 10% CO2 was released using the Flypad while the Peltier element was heated to 40˚C.For each 5-minute trial, images were taken every 10 seconds.After each trial, the mosquitoes were placed under cold anesthesia and counted to verify the number used.To quantify mosquito behavioral responses in this assay, we counted the number of mosquitoes on the cage screen near the Peltier element.Each data point represents a percentage of mosquitoes attracted to the heat source.This was determined by the average number of mosquitoes from the last ten images of each trial divided by the total number of mosquitoes in the assay, multiplied by 100.All scored images were cropped to the same size with Fiji software using a macro [makeRectangle(269, 176, 347, 237); run("Crop"); run("Save"); run("Open Next")].Once cropped, images were visually scored with assistance from the multipoint counting tool in Fiji.

SciTracks assay
The assay was performed as previously described (13).A multi-insect 3D tracking system was custom-designed and built in collaboration with SciTrackS GmbH (Sci-Trak).It consists of a flight chamber (1.25 m wide x 1m high x .75cm deep) with clear acrylic sides and front.A pump continuously provides carbon-filtered air (Cole-Parmer Quiet Pressure Pump #79610-81).CO2 can be added to this airstream through the use of a computer-controlled solenoid valve, a valve controller (NResearch, Inc. model 360D1X75R) operated by a MATLAB script through an analog/digital interface (Measurement Computing Corporation, model USB-1208FS).Humidity was controlled using a closed-loop humidification system consisting of an ultrasonic humidifier (SPT model SU-1051B) controlled by a temperature and humidity probe attached to a microcontroller.Humidity remained at 45% +/-4% for the duration of each trial.Tracking is accomplished by imaging the chamber with two offset cameras (Basler piA640-210gm; Basler AG, Ahrenburg, Germany with 1/2 4-12mm F/1.2 IR Aspherical objects, Tamron, Commack, NY, USA).Lighting is provided by two banks of IR LEDs (Metaphase model ISO-23-IRN-24-AL2, Metaphase Technologies, Bensalem, PA).3D position information of each visible mosquito was calculated in real-time at 200Hz using custom-built software based on previously published algorithms (76).Tracks were further processed, filtered, and analyzed using MATLAB.CO2 activation: For each trial, 20 female mosquitoes (5-10 days old, fasted overnight with access to water) were added to the flight chamber with a mechanical aspirator.After acclimation and tracking baseline activity for 30 minutes, a 40 second pulse of 100% CO2 was added to the airstream at a flow rate of 1360 mL/minute.This raised the concentration of CO2 to approximately 1%, as measured in the middle of the flight chamber.(CARBOCAP® Hand-Held Carbon Dioxide Meter GM70, Vaisala Inc., Woburn, MA, USA).Tracking continued for another 20 minutes to measure activity in response to this pulse of CO2.Data are presented as the population distance tracked in 10 second bins or as cumulative distance tracked per mosquito in the 6 minutes immediately before or after CO2 addition.

Optothermocycler assay
The Optothermocycler assay was performed as previously described (53).The assay was constructed on top of a PCR thermocycler (Eppendorf Mastercycler) using optomechanical components (Thorlabs).Green light at a wavelength of 530 nm was emitted using six Luxeon Star SP-01-G4 LEDs under the control of an Arduino Uno board.The surface of the PCR block was covered with black tape to minimize glare (Thorlabs T137-2.0).Temperature monitoring was conducted through a type T thermocouple (Harold G Schaevitz Industries LLC CPTC-120-X-N), which was connected to an Arduino board (Arduino A000066) via a thermocouple amplifier (Adafruit MAX31856).The thermocouple sensor was securely positioned on the lower right surface of the PCR block using black tape (Thorlabs T137-2.0).Using a custom Processing script, temperature measurements, and light output were logged at 100-millisecond intervals through the Arduino.The video was synchronized with the light and temperature stimuli by employing an infrared 940 nm LED (Adafruit 387) positioned within the camera's field of view.Mosquitoes were illuminated with an 850 nm infrared LED strip (Waveform Lighting) surrounding the mosquitoes' plate orthogonal to the camera's view.The video was recorded using a Blackfly camera (FLIR BFS-U3-16S2M-CS) outfitted with a 780 nm long pass filter (Vision Light Tech LP780-25.5)at 30 frames/second using Spinview software (Teledyne FLIR).The PCR thermocycler was programmed to apply heat stimuli directly from ambient to a noxious temperature range of 24-60°C.

Arm feeding assay
Female adult mosquitoes were cold-anesthetized, and 10 -20 mosquitoes were sorted into a behavioral container made from a 32-ounce HDPE plastic cup (VWR #89009-668).The behavioral container was prepared by cutting a 10 cm hole in the lid with a razor blade and covering the cup with a 20 cm x 20 cm piece of white 0.8 mm polyester mosquito netting (American Home & Habit Inc. #F03A-PONO-MOSQ-M008-ZS) and securing the mesh to the cup using the modified lid.Sorted mosquitoes were allowed to recover overnight at 26°C with 70-80% relative humidity and fasted by replacing 10% sucrose with a cotton wick saturated with deionized water placed on top of the container across the mesh netting.For each experimental trial, mosquitoes were allowed to acclimate for 10 minutes, and a live human arm was placed on top of the container.Mosquitoes were allowed to feed on the human arm through the mesh netting for 10 minutes and were then anesthetized at 4°C and scored as fed if any level of feeding was observed, as assessed by visual inspection of the abdomen of the animal.

Glytube feeding assay
Female adult mosquitoes were sorted into individual behavioral containers as described with the arm feeding assay.The assay chamber was a modification of previously published methods (58) and used a translucent polypropylene storage box 36 cm L x 31 cm W x 32 cm H with a removable lid.One 1.5 cm hole was made on the chamber wall and was used to introduce silicone tubing for CO2 delivery.The CO2 diffusion pad (8.9 cm x 12.7 cm; Tritech Research) was affixed to the inner center of the lid to allow delivery of purified air and CO2 to condition the chamber atmosphere during the trial.Females were fed sheep blood (Hemostat DSB100) supplemented with ATP (Sigma A6419) at a final concentration of 1 mM using a Glytube membrane feeder (58).All blood meals and Glytubes were preheated for at least 10 minutes in a 45°C water bath, and ATP was added to meals immediately before feeding.At the start of each trial, up to four cups were placed in the assay chamber, and the animals were allowed to acclimate for 10 minutes before 1 Glytube containing 1.5 mL of sheep blood+ATP mixture was placed on top of each cup, and CO2 was turned on for 10 minutes.Mosquitoes were then anesthetized at 4°C and scored as fed if any level of feeding was observed, as assessed by visual inspection of the abdomen of the animal.

RNA extraction and sequencing
Seven to 10-day-old female wildtype and Orco 5/16 mosquitoes were cold-anesthetized and kept on ice for up to 30 minutes or until dissections were complete.Forelegs and midlegs were dissected and collected on ice in a 1.5 mL tube.Five samples from each genotype were collected, and each replicate consisted of legs from 20 animals.After collecting each sample, tubes were immediately snap-frozen using a cold block (Simport S700-14) chilled with ethanol containing dry ice.Dissected tissue was stored at −80°C until tissue collection was complete.RNA was extracted by homogenizing each sample in TRIzol reagent (Invitrogen 15596026) using a handheld homogenizer (DWK Life Sciences 749540-0000).To the homogenized sample, 20% of the total TRIzol regent volume of chloroform:isoamyl alcohol (Fisher Scientific AC327155000) was added for phase separation.The aqueous phase containing total RNA was transferred into a fresh RNAse-free tube and cleaned using the PicoPure Kit (ThermoFisher KIT0204) following the manufacturer's instructions, including the DNase treatment step.Samples were run on a Bioanalyzer RNA Pico Chip (Agilent 5067-1513) to determine RNA quantity and quality.Libraries were prepared using the Illumina TruSeq Stranded mRNA kit (Illumina 20020594).Samples were pooled, and sequencing was performed at The Rockefeller University Genomics Resource Center on an Illumina NovaSeq 6000 sequencer using V1.5 reagents, the SP flow cell, and NovaSeq Control Software V1.7.0 to generate 150 bp paired-end reads.All samples were sequenced on the same flow cell to correct for potential batch effects.Data were demultiplexed and delivered as fastq files for each library.Sequencing reads have been deposited at the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under BioProject PRJNA1020561.

Transcriptome alignment, quantification, and differential gene expression analysis
The sequencing data was aligned and quantified with the Salmon quantification software (77) using the reference transcriptome generated from Aedes aegypti LVP_AGWG (https://vectorbase.org/vectorbase/app/record/dataset/DS_cc8d875d2e) genome and the updated version of the previously published gene annotation file from Jove et al. (58).The most current version of the reference transcriptome is available at https://github.com/VosshallLab/Morita_Vosshall2023.Gene expression levels as raw read counts and abundances were retrieved using tximport (version 1.22.0)(78).Normalization of raw read counts and differential expression analysis was performed using DESeq2 (version 1.34.0)(79).The false discovery rate was calculated using Benjamini & Hochberg multiple testing correction, using the FDR cut-off at 0.01.Foreleg, midleg, and hindleg samples from the neurotranscriptome dataset were reanalyzed following the above steps (29).Leg samples from sugar-fed and ovipositing samples were used for analysis (SRR1167521-533, SRR1166473-484, SRR1167469-477, SRR1167554-562, SRR1167491-492).Pair-wise differential expression analysis using DESeq2 was performed.The false discovery rate was calculated using Benjamini & Hochberg multiple testing correction, using the FDR cut-off at 0.01.
(Warner Instruments, SC-20), and the outlet was placed directly above the sensory appendage to be imaged.The air temperature was adjusted by controlling the Peltier heater/cooler device temperature using an external bipolar temperature controller (Warner Instruments, CL-100).By default, this heated air line is kept closed unless a heated stimulus is applied to the sensory appendages.The experiment started with the ambient air line opened, and the heat air line closed.The ambient air line was closed for each heat stimulus while the heat air line was opened.When the temperature of the sample reached ~33°C, the ambient air line was opened while the heat air line was closed.The opening of each air line was controlled through analog signals provided by the Image acquisition software.For the slow heat ramp protocol used in Figure 6H-N, the temperature stimulus was delivered to the dissected sensory appendage through a single air line with an air source as previously described.This air line was passed directly through the Peltier device, where the temperature can be changed throughout the experiment.At the beginning of each trial, the Peltier device was set at 20°C for ambient air temperature delivery.During the experimental trial, the Peltier temperature was changed to 50°C, so the target temperature that the sensory appendages were experiencing was around 33°C.Once the temperature reached our target temperature, the Peltier temperature was switched back to 20°C.To monitor the temperature of the sensory appendages, a thermocouple microprobe was placed adjacent to the samples (Physitemp, IT-23) and was recorded using a digital thermometer (Physitemp, BAT-12) during the entire experiment.

Calcium imaging data analysis
Fluorescent intensities emitted by fluorescent proteins are temperature-dependent and have an inverse relationship with temperature (82,83).To account for this biophysical property of fluorescent proteins, we took advantage of animals co-expressing GCaMP6s and dTomato in the same cell by capturing images at both fluorescent channels.dTomato functioned as a reference imaging channel to correct for both movement artifacts and temperature-induced changes in fluorescence intensity.Regions of interest (ROI) were marked using the dTomato signal captured at the beginning of each experiment.ROI information was overlaid with all image frames captured during each experimental trial, and fluorescent intensities were recorded for each time point from both GCaMP and dTomato imaging channels.The ratio of GCaMP and dTomato was used to report the temperaturedependent changes in fluorescent intensities at each time point and was normalized to the first frame to report the data as ΔR/R.Ir140 crRNA design Ir140 single and Orco, Ir140 double mutants were generated using methods described previously (21).sgRNA sequences were designed with the CHOPCHOP sgRNA design tool (https://chopchop.cbu.uib.no/)(84) using the following parameters: In, Aedes aegypti (AaegL5.0);Using, CRISPR/Cas9; For, knock-out, with Ir140 CDS sequence (NM_001358695.1)as the target sequence.For each type of Ir140 mutants, pairs of target sequences were selected for targeted double-stranded break-induced mutagenesis, with each pair flanking roughly 250 base pairs.The following sequences were used for each type of Ir140 mutant generation: Ir140 single mutant: 1) CAGAAGTTATCATAGCACCTGTTTTAGAGCTATGCT 2) TCTCCATCATGCCAACACACGTTTTAGAGCTATGCT Orco 16 , Ir140 double mutant: 1) CCGCCTTCATAATCAATCAGGTTTTAGAGCTATGCT 2) CTCATCGTGATCTACTCTGAGTTTTAGAGCTATGCT Orco 5 , Ir140 double mutant: 1) CCGCCTTCATAATCAATCAGGTTTTAGAGCTATGCT 2) CTCATCGTGATCTACTCTGAGTTTTAGAGCTATGCT CRISPR injection mix 2 µl of 100 µM Alt-R CRISPR-Cas9 tracrRNA (IDT.#1072532) was mixed with 1 µl of each custom-designed Alt-R CRISPR-Cas9 crRNA (100 µM) and 1 µl of Nuclease-Free Duplex Buffer (IDT.#11-01-03-01).This initial mix was then incubated at 95°C on a thermocycler for 5 minutes, followed by 5 minutes at room temperature.After both incubations, 3µl of Alt-R™ S.p. Cas9 V3, glycerol-free, at 1µg/µl (IDT #10007806) and 2 µl of nuclease-free water were added.The solution was incubated at room temperature for 5 minutes and then placed on ice until the time of injection.

Injection procedure
Mosquito embryos were injected in-house using a Sutter Instrument injection system (Product # FG-BRE, FG-MPC385) and a Zeiss SteREO Discovery stereo microscope (#495007-9880-010).Embryo injection was carried out as previously described (85).About 100 eggs were injected per injection session 60 minutes after laying them.

Figure 1 :
Figure 1: Ionotropic receptors tune thermotaxis behavior.(A) Female mosquitoes use redundant human-emitted cues (CO2, odor, and heat) to pursue blood meal.The black ovals with sensory cues displayed here and throughout the manuscript show the sensory stimuli present in the corresponding behavioral experiment.(B) Schematic of heat-seeking assay apparatus (30 cm 3 ).(C) Representative experimental image of wildtype female mosquitoes on and near the Peltier (dotted line) set at 36°C.(D, F, H) Schematic of female body parts that express Ir25a (D), Ir76b (F), and Ir8a (H).(E, G, I) Percent of mosquitoes of indicated genotypes on Peltier during seconds 90-180 of stimuli of indicated temperature (mean ± SEM, n = 5-13 trials per genotype; datapoints marked with * indicate that the knockout mutant differs significantly from all other tested genotypes within each tested temperature at p < 0.05; one-way ANOVA with Tukey's HSD post hoc test).

Figure 2 :
Figure 2: Orco mutant mosquitoes display enhanced heat-seeking behavior.(A) Schematic of female body parts that express Orco.(B) Heat maps showing mean mosquito occupancy for the indicated genotypes on the Peltier (dotted lines) and surrounding area at indicated Peltier temperature during seconds 90-180 of each stimulus period.(C) Mean ± SEM percent of mosquitoes of indicated genotypes on Peltier (top) during the 36°C trial (bottom).A 20 second pulse of CO2 was applied at the beginning of each stimulus period.(D) Percent of mosquitoes of indicated genotypes on Peltier during seconds 90-180 of stimuli of indicated temperature (mean ± SEM, n = 9 trials per genotype; datapoints marked with * indicate that the knockout mutant differs significantly from all other tested genotypes within each tested temperature at p < 0.05; one-way ANOVA with Tukey's HSD post hoc test).(E -G) Mean dwell time (E), landing frequency (F), and takeoff frequency (G) of indicated genotypes on the Peltier surface during the 36°C trial (n = 9 trials/genotype).(H) Schematic representation of the modified heat-seeking assay (28 cm 3 ) in the presence of DEET.(I) Percent of mosquitoes of indicated genotype and treatment groups heat-seeking during a 40°C trial (n = 6-10 trials/genotype).Data are plotted as scatter-box plots (individual data points, median as horizontal line, interquartile range as box) for (E,F,G,I).Data labeled with different letters differ significantly (p < 0.05; one-way ANOVA with Tukey's HSD post hoc test at each tested temperature).

Figure 3 :
Figure 3: Human attraction is enhanced in Orco mutant mosquitoes.(A) Schematic of the CO2 activation assay.(B) Percent mosquitoes flying in the presence of CO2 quantified every minute (mean ± SEM, n = 10 trials per genotype).(C, E) Nylon-next-to-cage assay schematic with unworn (C) and worn (E) nylon.(D, F) Percent mosquitoes next to unworn (D, n=10 trials/genotype) or worn (F, n=5 trials/genotype) nylon, quantified once per minute (mean ± SEM).(G, I) Arm-next-to-cage assay schematic with human arm without (G) or with 10% DEET (I).(H, J) Percent mosquitoes next to the human arm without (H, n=10-13 trials/genotype) or with 10% DEET (J, n= 6 trials/genotype) quantified once per minute (mean ± SEM).In panels D, F, H, and J, datapoints marked with * indicate that knockout mutant differs significantly from all other tested genotypes within each tested temperature at p < 0.05; one-way ANOVA with Tukey's HSD post hoc test).

Figure 5 :
Figure 5: Morphology of tarsal neurons (A) Schematic representation of a mosquito leg and representative tiled confocal image of tarsal segments of brp-wQF2>QUAS-dTomato mosquitoes expressing dTomato in all neurons.For this and all other images in the figure, dorsal is left and distal is top.Individual tarsomeres are numbered from proximal to distal segments.(B-E)

Figure 6 :
Figure 6: Tarsal neurons respond to heat (A) Schematic of female body parts that express GCaMP6s in the pan-neuronal imaging strain brp-QF2w > dTomato-T2A-GCaMP6s. (B) Schematic of ex vivo ratiometric sensory appendage Ca 2+ imaging setup.(C) Widefield images of foreleg distal tarsal segments, prior to and during heat stimulus.Neuronal responses are indicated

Figure 7 :
Figure 7: Aedes aegypti chemosensory receptor expression changes in the Orco mutant legs (A) Workflow of pooled foreleg and midleg bulk RNA-sequencing comparing gene expression between wildtype and Orco 5/16 mutant mosquito tarsi.(B) A legend describing bulk RNA-seq data presented in panels C-G.The

Figure 8 :
Figure 8: Ir140 contributes to heat-seeking behavior and is required for the enhanced Orco mutant heatseeking behavior (A) Schematic of Aedes aegypti Ir140 and Orco loci.Arrows indicate Ir140 mutant alleles in wildtype and Orco mutant backgrounds.Introns are not drawn to scale.(B,C) Percent of mosquitoes of indicated genotype on Peltier