Report

Mosquito heat seeking is driven by an ancestral cooling receptor

See allHide authors and affiliations

Science  07 Feb 2020:
Vol. 367, Issue 6478, pp. 681-684
DOI: 10.1126/science.aay9847

Heat seeking is cool

Mosquitoes seek hosts using several cues, one of which is body heat. Greppi et al. hypothesized that cooling-activated receptors could be used for locating mammalian hosts if they were rewired downstream for repulsion responses (see the Perspective by Lazzari). A gene family conserved in insects and known to be responsible for sensing changes in temperature in fruit flies was the starting point. Genome-wide analyses and labeled CRISPR-Cas9 mutants allowed visualization of the receptor in neurons of Anopheles gambiae mosquitoes' antennae and assessment of adult female mosquitoes with a disrupted copy of the receptor. This ancestral insect temperature regulatory system has been repurposed for host-finding by malaria mosquitoes.

Science, this issue p. 681; see also p. 628

Abstract

Mosquitoes transmit pathogens that kill >700,000 people annually. These insects use body heat to locate and feed on warm-blooded hosts, but the molecular basis of such behavior is unknown. Here, we identify ionotropic receptor IR21a, a receptor conserved throughout insects, as a key mediator of heat seeking in the malaria vector Anopheles gambiae. Although Ir21a mediates heat avoidance in Drosophila, we find it drives heat seeking and heat-stimulated blood feeding in Anopheles. At a cellular level, Ir21a is essential for the detection of cooling, suggesting that during evolution mosquito heat seeking relied on cooling-mediated repulsion. Our data indicate that the evolution of blood feeding in Anopheles involves repurposing an ancestral thermoreceptor from non–blood-feeding Diptera.

Insect-borne diseases kill over 700,000 people annually, with >400,000 deaths resulting from malaria, a disease caused by protozoan Plasmodium spp. parasites that are transmitted by blood-feeding anopheline mosquitoes (1). Host seeking by mosquitoes and other pathogen-spreading insects relies on the detection of host-associated cues, including carbon dioxide (CO2), odors, and body heat (25). Receptors for CO2 and host odors have been characterized in mosquitoes (69), but receptors that promote heat seeking and heat-induced blood feeding have remained elusive (4, 1012). As vector mosquitoes are descendants of non–blood-feeding ancestors (13), it remains unknown whether the emergence of heat seeking and warming-induced blood feeding in mosquitoes involved the generation of novel thermoreceptors or the repurposing of existing thermoreceptors.

To date, mosquito orthologs of two Drosophila warmth receptors, TRPA1 (14) and GR28b (15), have been tested as candidate heat-seeking receptors in the yellow fever mosquito Aedes aegypti (1012). However, neither is required for heat seeking in Aedes (12). Rather, TRPA1 promotes heat avoidance in both Aedes and Drosophila (12, 14). Although efforts have focused on warmth receptors, insects also possess cooling-activated receptors, which should be equally capable of supporting heat seeking through cooling-mediated repulsion. In Drosophila, cooling detection is mediated by IR21a, IR25a, and IR93a (1618), three members of the ionotropic receptor (IR) family, a group of invertebrate-specific sensory receptors related to ionotropic glutamate receptors (19). IR21a is specifically required for cooling detection in the fly and can confer cooling sensitivity when ectopically expressed (16, 18), while IR25a and IR93a are more broadly acting co-receptors that support cooling detection and other IR-dependent sensory modalities (17, 19, 20). At the behavioral level in Drosophila, IR21a, IR25a, and IR93a help the fly achieve optimal body temperatures by supporting avoidance of excessively cool and warm temperatures (16, 18). Beyond Drosophila, IR21a, Ir25a, and IR93a are each widely conserved from Diptera (flies and mosquitoes) to Isoptera (termites) (19), raising the possibility that their thermosensory functions may also be conserved. Using Anopheles gambiae, a major vector of malaria in sub-Saharan Africa, we first tested whether IR21a is required for detecting cooling in mosquitoes and subsequently whether it can drive heat attraction and heat-stimulated blood feeding.

Two mutant alleles of A. gambiae Ir21a were generated using CRISPR-Cas9 (see methods). Ir21a+7bp contains a 7-base pair (bp) insertion, introducing a frameshift positioned to disrupt IR21a’s translation within the second of IR21a’s three transmembrane domains; this lesion is predicted to generate a nonfunctional receptor (Fig. 1A). In Ir21aEYFP, a disruption cassette containing an enhanced yellow fluorescent protein (EYFP) marker, was inserted into IR21a’s fourth exon, a lesion also predicted to create a nonfunctional receptor (Fig. 1B). Both mutants lacked detectable IR21a protein expression (Fig. 1, C to E, and fig. S1), consistent with their acting as Ir21a null mutations.

Fig. 1 IR21a is expressed in the antennal tip.

(A, upper) Ir21a locus, with Ir21a+7bp mutations shown in blue. (Lower) Virtual translations, with 450 C-terminal amino acids of wild-type IR21a replaced by 24 novel amino acids (blue) and a premature stop in Ir21a+7bp. (B) Targeted integration generating Ir21aEYFP, a combined bright-field and fluorescent illumination image, and molecular genotyping results for Ir21a+ and Ir21aEYFP homozygotes. Lane m, DNA size markers. (C) Mosquito anterior [drawing based on (27)]. Inset, flagellomere 13. (D and E) Immunostaining of flagellomere 13 in wild-type (D) and Ir21a+7bp (E) females. (n = 13 wild type; n = 7 Ir21a+7bp). Asterisks, IR21a-expressing cell bodies; arrows, sensory endings. HRP, horseradish peroxidase; DAPI, 4′,6-diamidino-2-phenylindole. Anti-HRP labels neuronal membrane proteins, and DAPI labels nuclei.

Genome-wide analyses of A. gambiae sensory tissues suggest that Ir21a RNA is specifically expressed in the antenna (21). To visualize IR21a protein expression and localization with cellular resolution, anti-IR21a antisera were generated. The antenna’s most distal segment (flagellomere 13) contains three coeloconic sensilla that house sensitive thermoreceptors (2224) (Fig. 1C). In females, IR21a expression was detected in three sensory neurons in flagellomere 13, one innervating each of the coeloconic sensilla (Fig. 1D). Consistent with a role in thermosensory transduction, IR21a strongly localized to the sensory ending of each of these neurons (Fig. 1D). IR21a immunostaining was absent in Ir21a mutants, confirming antisera specificity (Fig. 1E and fig. S1A). The male antennal tip also contains thermoreceptors (23), and IR21a expression was detected in sensory endings there as well (fig. S1B).

Extracellular recordings were performed from the IR21a-positive coeloconic sensilla at the antennal tip (Fig. 2A). In wild-type mosquitoes, the activity of the Cooling Cell, a thermosensory neuron stimulated by cooling and inhibited by warming, was readily detected (Fig. 2B). On rare occasions of exceptional signal to noise, a smaller-amplitude spike was also detected, corresponding to a Heating Cell activated by warming and inhibited by cooling (fig. S2). Cooling Cell responses were highly thermosensitive: an ~0.5°C drop from ~30°C increased spiking by ~40%, and an ~0.5°C drop from ~37°C increased spiking by ~80% (Fig. 2C). Response adaptation initiated rapidly, followed by a slower decline to baseline (Fig. 2C and fig. S3). Heating inhibited spiking, in a similarly transient manner (Fig. 2C). Importantly, Cooling Cells remained highly active at warm temperatures (e.g., 37°C) and were neither more active nor more thermosensitive at colder temperatures (Fig. 2, C and D). Thus, while often referred to as Cold Cells in the classical literature, cooling and not cold is their activating stimulus. In addition, while often referred to as “phasic-tonic” receptors, their responses to temperature shifts adapted fully, albeit slowly, requiring sustained observation (>20 s) to fully appreciate (fig. S3). Therefore, although they fire robustly at constant temperature, Cooling Cells are phasic thermoreceptors. Their rate of baseline firing was relatively temperature insensitive [with a fold change upon 10°C increase (Q10) of ~1.6, reflecting a slight increase with warmth], enabling the cell to respond to small temperature fluctuations over a wide range of absolute temperatures. [Physiologically similar Cooling and Heating Cells have been described in A. aegypti (22, 24) and Drosophila melanogaster (18).] Taken together, these data indicate that Cooling Cells are phasic thermoreceptors that respond to temperature change rather than absolute temperature and that they are capable of responding to abrupt changes in temperature over the wide range of absolute temperatures relevant for host seeking (Fig. 2, C and D).

Fig. 2 Ir21a is required for thermosensing.

(A) Recording electrode insertion site. (B) Representative recordings, with indicated regions displayed on an expanded time scale. Circles, spikes. The weighted average spike rate is the instantaneous spike frequency smoothed using a 1-s triangular window. Dotted lines, spike thresholds. (C) Peri-stimulus time histograms (averages ± SEM) for wild type (n = 8) and Ir21aEYFP and Ir21a+7bp (n = 6) animals tested at 30°C and 25°C. One heating-cooling trial per animal. (D) Cooling response = (average frequency 0.2 to 0.7 s after cooling onset) – (average frequency 5 to 10 s precooling). Heating response = (average frequency 0.5 to 1.5 s after heating onset) – (average frequency 5 to 10 s preheating). Lowercase letters indicate distinct groups (Tukey’s honest significant difference, α = 0.01, except 32°C→37°C, α = 0.05). Shapiro-Wilk test and analysis of variance values are provided in the statistics section of methods.

Cooling Cell thermosensitivity was eliminated in A. gambiae Ir21a mutants. The large-amplitude spike detected in Ir21a mutants was neither activated by cooling nor inhibited by warming (Fig. 2, B to D). Rather, its activity increased slightly upon warming, with a Q10 under 2, which is average for a biological process. Thus, Ir21a is essential for thermosensing by Cooling Cells in the mosquito, demonstrating that A. gambiae IR21a’s molecular function is conserved with its Drosophila ortholog (18).

In female mosquitoes, heat seeking is part of a multimodal host-seeking program activated upon exposure to CO2, with body heat serving as an important cue close to the host (within ~10 to 15 cm) (3, 8, 12). To assess heat seeking, female mosquitoes were provided a 20-s puff of 4% CO2 and exposed to two targets, a control target at ambient temperature (~26°C) and a heated target at ~37°C (Fig. 3, A and B, and fig. S4A). Wild-type mosquitoes exhibited robust heat seeking, with 43 ± 3% of CO2-activated mosquitoes landing on the 37°C target (average ± SEM) (Fig. 3, C and F; movie S1; and fig. S4B). The loss of Ir21a greatly reduced this behavior, with only 15 ± 4% of Ir21aEYFP mutants and 14 ± 4% of Ir21a+7bp mutants landing on the 37°C target (Fig. 3, D to F; movie S2; and fig. S4B). In all cases, the control target was largely ignored, confirming temperature’s importance in the assay (Fig. 3, C to F). While heat seeking was greatly reduced in Ir21a mutants, it was not entirely eliminated (Fig. 3, D to F). This residual activity likely reflects signaling from other as-yet-uncharacterized thermosensors. However, the strong reduction of heat seeking in Ir21a mutants (Fig. 3G) identifies this receptor as a major driver of mosquito attraction to warmth.

Fig. 3 Ir21a mediates heat seeking.

(A) Heat seeking assay. Assay box is 28 cm deep, 40 cm long, and 16 cm tall. (B) The stimulus sequence and formula for the heat-seeking index. (C) Representative images of 26°C (blue) and 37°C (red) targets before and ~120 s after CO2 pulse initiation in wild type. (D and E) Representative images from ~120 s after CO2 pulse initiation in Ir21aEYFP (D) and Ir21a+7bp (E) mosquitoes. (F) Landing on 37°C and 26°C targets as the percentage of mosquitoes taking flight (averages ± SEM). Wild type, n = 15 independent groups; Ir21aEYFP, n = 6; Ir21a+7bp, n = 7. There were 42 to 52 females per group. (G) Heat seeking index (average from 105 to 135s). Letters denote distinct categories (Steel-Dwass test, P < 0.01). Shapiro-Wilk, Kruskal-Wallis, and Steel-Dwass test values are provided in the statistics section of methods.

To test the specificity of the Ir21a mutant behavioral deficit for heat seeking, we examined their ability to perform an activation-dependent behavior less reliant on thermosensation. While body heat is a powerful short-range cue, a multimodal combination of longer-range chemosensory and visual cues mediates initial approach (3), suggesting that such behavior should be largely unaffected by a specific thermosensory deficit. To assess approach behavior, mosquitoes were activated by five human breaths and presented a human hand, positioned on a platform to prevent physical contact with the mosquitoes but otherwise providing host-associated cues (Fig. 4A). Hand approach was strong in the wild type, with 57 ± 2% of mosquitoes landing on the surface beneath the hand (Fig. 4, B and C). Hand-associated cues were critical, as hand withdrawal prompted rapid dispersal (Fig. 4B). Ir21aEYFP mutants remained robustly responsive, exhibiting maximum levels of approach (55 ± 2%) similar to wild-type levels (Fig. 4, B and C). Careful examination of response kinetics revealed that, compared to wild type, their initial accumulation rate decreased by ~25% (Fig. 4, B and D), and their dispersal rate upon hand removal increased by ~40% (Fig. 4, B and E), potentially reflecting subtle contributions of the warmth gradient created by the presence of the hand (fig. S4C) to the avidity of host approach. Similar results were obtained for Ir21a+7bp (fig. S4, D to G). Overall, these data demonstrate that the loss of Ir21a does not broadly disrupt orientation toward sensory cues and argue against the presence of global behavioral deficits in the mutants. These results are consistent with prior work indicating that the disruption of single sensory modalities is insufficient to completely eliminate host approach (2, 3, 8, 9).

Fig. 4 Ir21a promotes warmth-stimulated blood feeding.

(A to E) Host approach. (A) Mosquitoes were activated by five breaths and then presented a hand. (B) Mosquitoes landed (accumulated) on the cage roof below the hand. Values are averages ± SEM for 33 to 75 females per assay. Wild type (wt), n = 19 independent groups; Ir21aEYFP , n = 21. (C) Average maximum accumulation, 180 s to 300s. (D) Accumulation rate, {[(accumulation at 45 s) – (accumulation pre–hand exposure)]/[(maximum accumulation) – (accumulation pre–hand exposure)]}/45 s. (E) Departure rate, 1 – [(accumulation at 340 s)/(accumulation at 300 s)]/40 s. (F to H) Blood feeding. (F) Two meals (one dyed green) were placed on the cage. (G) RT meal was dyed. (H) Warm meal was dyed. Dotted lines link pairs. Total of 33 to 75 females per assay. n = 6 independent groups per genotype, except in (H), where Ir21aEYFP n = 7. P values indicate t test results. The green tracking dye partially reduced relative consumption of meals to which it was added. Cages are 17.5-cm-sided cubes. Shapiro-Wilk test values are provided in the statistics section of methods.

Heat strongly stimulates mosquito blood feeding (8, 9). To assess the effect of warmth on blood feeding, artificial membrane feeders (25) were used to present human blood meals at different temperatures (Fig. 4F). One meal was held at room temperature (RT, ~23°C) and the other warmed to ~31°C, a temperature similar to the surface temperatures (~29°C to 33°C) of human torsos and extremities in a 23°C to 24°C room (26). In each trial, green food coloring was added to one meal so that the consumption of warm versus RT food could be distinguished (fig. S5). Each class of trial was assessed independently (RT meal dyed green in Fig. 4G and fig. S4H; warm meal dyed green in Fig. 4H and fig. S4I), and each yielded similar results. In wild type, elevated temperature robustly promoted feeding, as reflected in the greater percentages of mosquitoes consuming warm versus RT meals (Fig. 4, G and H, and fig. S4, H and I). For both Ir21aEYFP (Fig. 4, G and H) and Ir21+7bp (fig. S4, H and I) mosquitoes, this preference for warm blood was significantly reduced. Thus, similar to heat seeking, warmth-promoted blood feeding was reduced in the absence of Ir21a.

These data identify IR21a as a key mediator of heat-seeking behavior in A. gambiae mosquitoes. Although a cooling-activated receptor driving heat seeking is superficially counterintuitive, repulsion from cooling would yield a similar behavioral outcome as attraction to warming. Furthermore, Cooling Cells are bidirectional and are not only activated by cooling but also inhibited by heating (Fig. 2, D and E); each phase of the response could modulate downstream circuits to control behavior. Ultimately, the detection of temperature change by the Cooling Cells is critical, but is just one step in heat seeking, a response that involves the processing of multiple sensory inputs to generate a coherent response. Identification of a key molecular receptor for heat seeking provides a starting point for a deeper understanding of this complex behavior and its contribution to the multimodal process that culminates in mosquito blood feeding.

The conservation of IR21a’s thermosensory function between Drosophila (18) and Anopheles (Fig. 2), whose last common ancestor lived ~250 million years ago, suggests thermosensing is an ancestral function of IR21a. As this ancestor predates the evolution of blood feeding (13), its IR21a would have regulated other behaviors, such as thermoregulation. Thus, our findings indicate that the evolution of blood feeding in A. gambiae mosquitoes involved repurposing an ancestral thermoreceptor to facilitate host seeking. Alterations in the connectivity or function of downstream circuits would likely have been crucial in this behavioral shift. Given the conservation of IR21a as well as IR25a and IR93a (IR21a’s coreceptors in Drosophila) across insects (19), these IRs may be used in heat seeking not only by other mosquitoes but also across a range of hematophagous insect taxa.

In addition to Ir21a’s role in heat seeking, IR21a expression in the antennae of A. gambiae males suggests it continues to serve additional thermosensory functions. It will be interesting to assess whether IR21a mediates thermal preference in male and possibly female mosquitoes and the extent to which thermal preference and heat-seeking circuits overlap. Not all thermoreceptors appear to have been repurposed, as the TRPA1 warmth receptor has a similar role in flies and mosquitoes, mediating heat avoidance in both (12). At a practical level, exploiting and manipulating the sensory systems of vector insects offer an avenue for disease control strategies.

Supplementary Materials

science.sciencemag.org/content/367/6478/681/suppl/DC1

Materials and Methods

Figs. S1 to S5

References (2832)

Movies S1 and S2

References and Notes

Acknowledgments: We thank A. Hammond, R. Harrell, T. Nolan, S. McIver, A. Crisanti, E. Marois, B. White, and L. Vosshall for reagents and advice, R. Albuquerque for assistance with data analysis, R. Gerber for assistance with mosquito husbandry and behavioral assays, and A. Lee, E. Marder, M. Rosbash, P. Sengupta, and C. Zhu for comments on the manuscript. Funding: This work was supported by grants from the National Institute of Allergy and Infectious Diseases (F31 AI133945 to C.G.; R01 AI122802 and R21 AI140018 to P.A.G. and F.C.), the National Institute of Neurological Disorders and Stroke (2T32NS007292-31) to W.J.L., the National Institute of General Medicine (F32 GM113318) to B.G., the Swiss National Science Foundation (P2FRP3_168480) to L.V.G., a Faculty Research Scholar Award by the Howard Hughes Medical Institute and the Bill & Melinda Gates Foundation (grant OPP1158190) to F.C., and the National Science Foundation (IOS 1557781) to P.A.G. Author contributions: C.G., W.J.L., G.B., L.V.G., A.L.S., F.C., and P.A.G. designed experiments. C.G., W.J.L., E.C.C., A.M.D., and L.V.G. performed husbandry, molecular genetics, and heat-seeking behavior experiments. C.G. performed NHEJ-based gene disruption. W.J.L. performed gene targeting. W.J.L. and E.C.C. performed hand approach and blood-feeding behavior analyses. L.V.G. and C.G. performed immunohistochemistry. G.B. and C.G. performed electrophysiology. A.L.S. performed transgenesis of gRNA expression vector. W.J.L. and P.A.G. performed data analysis. C.G., W.J.L., and P.A.G. wrote the paper with input from all authors. Competing interests: A.L.S. is a coinventor on patent WO2015105928A1 (WIPO PCT pending; inventors K. Esvelt and A. L. Smidler), “RNA-guided gene drives.” The patent involves spreading desirable traits genetically through mosquito populations using Cas9-based gene drives. IR21a could potentially be used as a target for such a gene drive. P.A.G. is a coinventor on patent WO2017196861A1 (WIPO PCT pending; inventors Z. Knecht, P.Garrity, L. Ni) “Methods for modulating insect hygro- and/or thermosensation.” This patent proposes using members of the ionotropic receptor family as targets for strategies to disrupt hygro- and thermo-sensation in insects. Data and materials availability: The datasets generated and analyzed during the current study are available at DRYAD (https://doi.org/10.5061/dryad.pzgmsbcg3).

Stay Connected to Science

Navigate This Article