Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Feb 27, 2007; 104(9): 3574–3578.
Published online Feb 20, 2007. doi:  10.1073/pnas.0700079104
PMCID: PMC1805529

The molecular basis of CO2 reception in Drosophila


CO2 elicits a response from many insects, including mosquito vectors of diseases such as malaria and yellow fever, but the molecular basis of CO2 detection is unknown in insects or other higher eukaryotes. Here we show that Gr21a and Gr63a, members of a large family of Drosophila seven-transmembrane-domain chemoreceptor genes, are coexpressed in chemosensory neurons of both the larva and the adult. The two genes confer CO2 response when coexpressed in an in vivo expression system, the “empty neuron system.” The response is highly specific for CO2 and dependent on CO2 concentration. The response shows an equivalent dependence on the dose of Gr21a and Gr63a. None of 39 other chemosensory receptors confers a comparable response to CO2. The identification of these receptors may now allow the identification of agents that block or activate them. Such agents could affect the responses of insect pests to the humans they seek.

Keywords: chemoreceptors, insect, Gr genes

Ever since the classic experiments of Joseph Priestley in the 18th century, the role of CO2 in the natural world has been a subject of great interest. However, despite its ubiquity and central role in the metabolism of living organisms, some of the most fundamental questions about how CO2 interacts with biological systems remain unanswered. One such question is how CO2 is detected in the animal world.

CO2 elicits behavioral responses in many insects that seek human hosts, including tsetse flies (1), which carry African sleeping sickness; Aedes mosquitoes (2), which carry dengue and yellow fever; and Anopheles mosquitoes (3), which transmit hundreds of millions of cases of malaria each year. CO2 also acts as an attractive cue for many insects that seek plants as food sources and oviposition sites (46). In Drosophila, high concentrations of CO2 evoke an avoidance response (7, 8).

CO2-sensitive neurons have been identified in many insect species (9) and in most cases are dedicated to the detection of CO2. In adult Drosophila, odors are detected by olfactory receptor neurons (ORNs) that are housed in sensilla on the antenna and the maxillary palp (10). One class of antennal ORNs, the ab1C class, detects CO2 (11). Axons of these CO2-sensitive neurons project to a single glomerulus in the antennal lobe of the brain, the V glomerulus, which has been shown to be responsive to CO2 (8).

Drosophila contains a family of 60 Odor receptor (Or) genes (1214), and a related family of 60 Gustatory receptor (Gr) genes (14, 15), both of which encode seven-transmembrane-domain proteins. In most ORN classes, a single Or gene defines the odorant response profile (1618). Typically, the Or gene is coexpressed with the noncanonical receptor Or83b, an atypical family member that is required for efficient localization of the canonical Or receptor to the dendrites (19). CO2-sensitive neurons are unique in that they do not express an Or gene (20, 21). Instead, a Gr gene, Gr21a, has been shown to be expressed in this class of neurons (8). Genetic ablation of Gr21a-positive neurons results in defects in the behavioral avoidance response to CO2 in adults (8) as well as in larvae (7). However, there has been no evidence that Gr21a acts in CO2 detection.

Here we show that another Gr gene, Gr63a, is coexpressed with Gr21a in larvae as well as in the adult. Coexpression of Gr21a and Gr63a in an in vivo expression system confers a CO2 response. The response depends on the presence of both Gr genes; neither gene alone confers a CO2 response. The response is highly specific for CO2 and depends on the concentration of CO2. Our results suggest that Gr21a and Gr63a form a heterodimeric receptor for the detection of CO2.


Gr21a and Gr63a Are Coexpressed in the Larva and the Adult.

In a large-scale study of Gr gene expression in the Drosophila larva, we generated a number of Gr promoter-GAL4 lines and found that the Gr21a and Gr63a promoters each drive expression of a GFP reporter in a single neuron in the terminal organ, a larval chemosensory organ (Fig. 1 A and B). When the two Gr promoter-GAL4 drivers were introduced into the same animal, the expression patterns were not additive; rather, a single neuron was again labeled (Fig. 1C). The simplest interpretation of this result is that Gr21a and Gr63a are coexpressed in the same larval chemosensory neuron.

Fig. 1.
Gr21a and Gr63a are coexpressed in larval and adult Drosophila sensory neurons. (A–C) Expression patterns driven by Gr21a-GAL4 (A), Gr63a-GAL4 (B), and Gr21a-GAL4+Gr63a-GAL4 (C) in larvae were visualized by using a UAS-GFP reporter. Pictures show ...

In the adult, previous studies have shown that Gr21a is expressed in one class of ORN in the antenna, ab1C (20, 21). This ORN is housed in the ab1 sensillum with three other neurons designated ab1A, ab1B, and ab1D (Fig. 1D) (11, 22). ORNs of the ab1C class send axons to the V glomerulus in the most ventral portion of the antennal lobe of the brain (Fig. 1E) (8, 23). We tested the hypothesis that Gr21a and Gr63a are also coexpressed in the adult olfactory system. We found that the Gr21a and Gr63a promoters each drive expression of GFP in a subset of antennal ORNs that project to the V glomerulus (Fig. 1 E and F). The simplest interpretation of these results is that Gr21a and Gr63a are coexpressed not only in the larva but also in the ab1C neurons of the adult.

Gr21a and Gr63a Confer a CO2 Response When Coexpressed.

The ab1C neuron has been shown to respond with high sensitivity and specificity to CO2 (11). The V glomerulus has also been implicated in CO2 response (8). However, the role of Gr21a in CO2 response has not been elucidated. Our results suggested that Gr21a and Gr63a might function together in imparting a CO2 response. We tested this possibility by using an in vivo expression system, the “empty neuron” system (16), to ask directly whether these receptors, either singly or in combination, conferred a CO2 response. This system is based on the ab3A ORN, which in the wild-type antenna responds strongly to several volatile compounds but not to CO2 (11). In the empty neuron system the endogenous receptors in this ORN, Or22a and Or22b, have been genetically removed, thereby eliminating the normal odorant responses of the ORN (16). Gr21a and Gr63a were expressed in this mutant via an Or22a promoter and the GAL4-UAS system (Fig. 2A).

Fig. 2.
Gr21a and Gr63a receptors together confer an electrophysiological response to CO2 in the mutant ab3A neuron. (A) Schematic illustrating the empty neuron system used to test electrophysiological responses of receptor combinations in the mutant ab3A neuron. ...

We found that expression of either Gr21a or Gr63a alone did not confer a response to CO2, but that the two when coexpressed produced a physiological response to CO2 in the ab3A neuron (Fig. 2 B and C). Moreover, the magnitude of the response to CO2 depended on the dosage of each gene: CO2 response was stronger when we increased the dosage of either Gr21a or Gr63a by introducing an additional copy of the respective transgene (Fig. 2 B and C). These results show that the CO2 response depends on both Gr21a and Gr63a.

We note that the level of spontaneous activity also increased with increasing dosage of Gr genes. The empty neuron, without transgenes, produces a spontaneous action potential frequency of 0.6 ± 0.3 spikes per second (SEM; n = 20), and adding either a single copy of Gr21a or a single copy of Gr63a alone did not affect this rate. When one copy of Gr21a was expressed together with one copy of Gr63a, the level of spontaneous firing was 2.3 ± 0.4 spikes per second (n = 14). When one copy of Gr21a and two copies of Gr63a were expressed, the frequency increased to 26 ± 5.7 spikes per second (n = 8); when two copies of Gr21a and one copy of Gr63a were expressed, the frequency was 22 ± 5.7 spikes per second (n = 8). These results show that the spontaneous rate depends on the dosage of these receptors, and that Gr21a and Gr63a are functionally equivalent in their effects on this rate.

The CO2 Response Is Dose-Dependent and Highly Specific.

We then examined whether the CO2 response was dose-dependent. We tested a wide range of CO2 concentrations (0.1–100%) and found that responses conferred by Gr21a and Gr63a show a steep dose-dependence over this range (Fig. 3). The responses to 0.035% CO2, a level typically found in air, are low, but the responses to 5% CO2, the level in exhaled breath, are near the half-maximal responses observed for each of the three genotypes we examined in the empty neuron system. Although none of these responses is as strong as those measured in the endogenous ab1C neuron, the response strength clearly depends on the copy number of Gr21a and Gr63a transgenes (Fig. 3). The increase conferred by adding a second copy of Gr21a is the same as the increase conferred by adding a second copy of Gr63a. It is technically difficult to generate a mutant ab3A neuron expressing two copies of each transgene, and we do not know whether such a transgenic neuron would show wild-type responses.

Fig. 3.
CO2 response of Gr21a and Gr63a is dose-dependent. Error bars indicate SEM and are too small to be seen in some cases; n = 8–9.

We next examined the specificity of the response conferred by coexpression of Gr21a and Gr63a. We used an odorant panel that included esters, alcohols, ketones, an aldehyde, and an aromatic compound, all of which were used to categorize the antennal basiconic sensilla into functional classes (11), as well as propionic acid and ammonia, which elicit responses in coeloconic sensilla (24). We found that the response elicited by CO2 was much greater than the response to any other compound (Fig. 4A).

Fig. 4.
Gr21a and Gr63a response is narrowly tuned to CO2. (A) Responses of the empty neuron expressing either Gr21a+Gr63a+Gr63a (Left), Gr21a+Gr21a+Gr63a (center), and of the ab1C neuron, which was analyzed in Or83b2 flies (right). 100% CO2 was used to test ...

To determine whether the specificity of the response conferred by coexpression of Gr21a and Gr63a recapitulated that of the ab1C neuron, we reexamined the specificity of ab1C. A complication in analyzing ab1C has been that it shares a sensillum with three other ORNs: the responses of ab1C to many odorants can be difficult to evaluate because the other ORNs respond strongly to them (11), yielding a complex pattern of action potentials. To increase the sensitivity and resolution of our analysis, we therefore measured ab1C responses in flies that were mutant for Or83b (19). In ab1 sensilla of mutants lacking Or83b, only the ab1C neuron retains its responses (19), thus allowing a sensitive examination of its odorant response profile. Consistent with previous studies, the ab1C neuron in these mutants has a very strong response to CO2, and its response, like that conferred by coexpression of Gr21a and Gr63a, is highly specific (Fig. 4A).

Does CO2 elicit a strong response only from cells coexpressing Gr21a and Gr63a? Of the 39 Or receptors that have been tested with CO2 in the empty neuron system (refs. 17 and 25; S. A. Kreher and J.R.C., data not shown), none confers a response approaching that conferred by Gr21 and Gr63a (Fig. 4B). Although some of the Or receptors appear to confer a weak or modest response, much of this response can be attributed to other factors: of the five Or genes that confer the strongest responses, all five were found to give strong responses to diluent controls. When these control values were subtracted, the singular nature of the Gr21a-Gr63a response is even more striking (Fig. 4C). Taken together, our results support the possibility that Gr21a and Gr63a together are the primary receptors for CO2 detection in Drosophila, a possibility that is consistent with behavioral and imaging analysis of the cells that express them in wild type (7, 8).


We have provided evidence that Gr21a and Gr63a, members of a large family of chemoreceptors, together function as a CO2 receptor in Drosophila. Both Gr21a and Gr63a are expressed in a single class of olfactory neurons that is uniquely and exquisitely tuned to CO2. When expressed in an “empty” olfactory neuron, Gr21a and Gr63a confer a narrowly tuned and dose-dependent response to CO2. In addition, the level of the response depends on the dosage of both genes. Other receptors tested in this system confer little, if any, response to CO2.

The simplest interpretation of our results is that Gr21a and Gr63a form a heterodimer that responds to CO2. There is precedence for heterodimerization of Or proteins (26, 27), but not Gr proteins. Most receptors of the Or family determine the ligand-specificity of the ORN in which they are expressed (1618) but are believed to require Or83b as a coreceptor for efficient transport and/or stabilization at the membrane (19, 26, 27). Based on our results we cannot determine whether both Gr21a and Gr63a act directly in ligand binding and/or signaling, or whether one of them acts as a cofactor in a manner analogous to Or83b. We note that the two genes were functionally equivalent in terms of the effects of their dosage on both CO2 response and spontaneous firing rate.

There is no precedence for seven-transmembrane-domain proteins that act as receptors for CO2 or any other gas to our knowledge. Previously described gas sensors include soluble guanylate cyclases that have been implicated in responses to NO and CO (28, 29), atypical guanylate cyclases that have been implicated in responses to O2 (30), and a heme-binding nuclear receptor that has been implicated in the response to NO and CO in Drosophila (31).

We do not know whether CO2 acts on the Gr receptor via the extracellular lymph that surrounds the dendrites of ORNs. One alternative possibility is that CO2 enters the cell by an independent mechanism and activates the receptor via the cytoplasm. There is abundant genetic and physiological evidence in unicellular organisms that Amt proteins and Rh proteins act as channels for NH3 and CO2, respectively (3234). It is possible that a similar channel facilitates entry of CO2 into the ab1C cell of Drosophila.

We do not know whether CO2 binds directly to the receptor. CO2 is readily hydrated to HCO3 (bicarbonate), which may bind to the receptor; it is also possible that CO2, a very small molecule, binds to a larger soluble factor that activates the receptor. We note finally that CO2 lowers the pH of water by forming a weak solution of carbonic acid, H2CO3. Such pH changes contribute to responses in central respiratory chemosensory cells as well as in acid-sensing taste cells in vertebrates (35), and could also play a role in CO2-sensing cells of Drosophila.

The responses to CO2 conferred by Gr21a and Gr63a in the empty neuron are lower than that of the ab1C neuron. The lower response may be primarily a result of lower gene dosage. However, the ab1C neuron is unique among ORNs in its dendritic morphology (22), which may be specialized to enhance CO2 reception. The ab1C neuron might also contain soluble factors that optimize CO2 sensing and that are not present in the empty neuron.

Gr21a and Gr63a are among a small number of Gr genes that have orthologs in the malaria vector mosquito Anopheles gambiae (36). We have also identified orthologs in the dengue and yellow fever vector mosquito Aedes aegypti, the silk moth Bombyx mori, and the flour beetle Tribolium castaneum. Interestingly, genes closely related to either Gr21a or Gr63a have not been identified in the honey bee Apis mellifera (37), suggesting that bees employ a different receptor for CO2.

Our finding that Gr21a and Gr63a confer a response to CO2 suggests the possibility of screening for compounds that inhibit or activate these proteins. Such compounds could affect the response of insect disease vectors, which are responsible for hundreds of millions of infections each year, to CO2 emanations from the human hosts they seek.

Materials and Methods

Drosophila Stocks and Transgenes.

Wild-type flies were Canton-S. All transgenic constructs were injected into w1118 flies. For Gr63a-GAL4 flies, ~0.8 kb of DNA upstream of the ATG was used to generate the promoter-GAL4 construct. Primers used to amplify the promoter region were 5′-TGGGAGTGCGCCAATTGTGG-3′ and 5′-TCCGGAGAGACTGTGTCCGG-3′. UAS-Gr constructs were created as described (17). Briefly, coding regions of Gr21a and Gr63a were amplified from Canton-S genomic DNA and inserted into the UAS expression vector, pUAST. Chromosomes bearing multiple copies of UAS-Gr21a or UAS-Gr63a transgenes were obtained by standard recombination techniques and confirmed by PCR. Or83b2 and Gr21a promoter-GAL4 lines were gifts from Leslie Vosshall (The Rockefeller University, New York, NY) and Kristin Scott (University of California, Berkeley, CA), respectively.


Anterior regions of larvae were dissected (Institut für Zoologie, Regensburg, Germany) and mounted in PBS containing 50% glycerol. Adult brains were dissected and prepared as described (38). Samples were immunostained with nc82 monoclonal antibody (a gift from Alois Hofbauer). All tissues were visualized by using a Bio-Rad (Hercules, CA) 1024 laser-scanning confocal microscope.


Extracellular recordings were performed on female antennal sensilla, and the action potentials of ORNs were quantified as described (16). All odor stimuli, with the exception of CO2, were puffed in a 500-ms pulse into a continuous airstream directed at the antenna (16). All odorants in the odor panel were dissolved in paraffin oil except propionic acid and ammonia, which were dissolved in water, and the responses to the diluent were subtracted from the responses to the odorants.

For CO2 responses, the flow rate for a CO2 gas stream was calibrated to match that of the continuous airstream (37.5 ml/sec). A stimulation box was devised such that the airstream directed at the antenna could be diverted and replaced for 500 ms by a steady stream from tanks containing various concentrations of CO2 (0.1%, 0.5%, 1%, 5%, 20%, and 100%, diluted with N2) (Airgas). For all CO2 recordings, the number of spikes in 500 ms of prestimulus spontaneous activity was subtracted from the number in the 500 ms after the onset of CO2 stimulation unless indicated otherwise.

Supporting Text.

The orders of receptors along the abscissas in Fig. 4 B and C are listed in SI Text.

Supplementary Material

Supporting Text:


We thank Wynand van der Goes van Naters for help with electrophysiology and suggestions and Scott Kreher for sharing unpublished data. This work was supported by National Institutes of Health Grants GM63364, DC04729, and DC02174.


olfactory receptor neuron.

Note Added in Proof.

Note Added in Proof.

Similar results were obtained in ref. 39.


The authors declare no conflict of interest.

This article contains supporting information online at www.pnas.org/cgi/content/full/0700079104/DC1.


1. Vale GA. Bull Ent Res. 1980;70:563–570.
2. Dekker T, Geier M, Carde RT. J Exp Biol. 2005;208:2963–2972. [PubMed]
3. Takken W, Knols BG. Annu Rev Entomol. 1999;44:131–157. [PubMed]
4. Stange G. Naturwissenschaften. 1999;86:190–192.
5. Guerenstein PG, Yepez EA, Van Haren J, Williams DG, Hildebrand JG. Naturwissenschaften. 2004;91:329–333. [PubMed]
6. Thom C, Guerenstein PG, Mechaber WL, Hildebrand JG. J Chem Ecol. 2004;30:1285–1288. [PubMed]
7. Faucher C, Forstreuter M, Hilker M, de Bruyne M. J Exp Biol. 2006;209:2739–2748. [PubMed]
8. Suh GS, Wong AM, Hergarden AC, Wang JW, Simon AF, Benzer S, Axel R, Anderson DJ. Nature. 2004;431:854–859. [PubMed]
9. Stange G, Stowe S. Microsc Res Tech. 1999;47:416–427. [PubMed]
10. Hallem EA, Dahanukar A, Carlson JR. Annu Rev Entomol. 2006;51:113–135. [PubMed]
11. De Bruyne M, Foster K, Carlson JR. Neuron. 2001;30:537–552. [PubMed]
12. Clyne PJ, Warr CG, Freeman MR, Lessing D, Kim J, Carlson JR. Neuron. 1999;22:327–338. [PubMed]
13. Vosshall LB, Amrein H, Morozov PS, Rzhetsky A, Axel R. Cell. 1999;96:725–736. [PubMed]
14. Robertson HM, Warr CG, Carlson JR. Proc Natl Acad Sci USA. 2003;100:14537–14542. [PMC free article] [PubMed]
15. Clyne PJ, Warr CG, Carlson JR. Science. 2000;287:1830–1834. [PubMed]
16. Dobritsa AA, van der Goes van Naters W, Warr CG, Steinbrecht RA, Carlson JR. Neuron. 2003;37:827–841. [PubMed]
17. Hallem EA, Ho MG, Carlson JR. Cell. 2004;117:965–979. [PubMed]
18. Goldman AL, Van der Goes van Naters W, Lessing D, Warr CG, Carlson JR. Neuron. 2005;45:661–666. [PubMed]
19. Larsson MC, Domingos AI, Jones WD, Chiappe ME, Amrein H, Vosshall LB. Neuron. 2004;43:703–714. [PubMed]
20. Couto A, Alenius M, Dickson BJ. Curr Biol. 2005;15:1–13.
21. Fishilevich E, Vosshall LB. Curr Biol. 2005;15:1548–1553. [PubMed]
22. Shanbhag S, Muller B, Steinbrecht A. Int J Insect Morphol Embryol. 1999;28:377–397.
23. Scott K, Brady R, Jr, Cravchik A, Morozov P, Rzhetsky A, Zuker C, Axel R. Cell. 2001;104:661–673. [PubMed]
24. Yao CA, Ignell R, Carlson JR. J Neurosci. 2005;25:8359–8367. [PubMed]
25. Kreher SA, Kwon JY, Carlson JR. Neuron. 2005;46:445–456. [PubMed]
26. Neuhaus EM, Gisselmann G, Zhang W, Dooley R, Stortkuhl K, Hatt H. Nat Neurosci. 2005;8:15–17. [PubMed]
27. Benton R, Sachse S, Michnick SW, Vosshall LB. PLoS Biol. 2006;4:e20. [PMC free article] [PubMed]
28. Wingrove JA, O'Farrell PH. Cell. 1999;98:105–114. [PMC free article] [PubMed]
29. Verma A, Hirsch DJ, Glatt CE, Ronnett GV, Snyder SH. Science. 1993;259:381–384. [PubMed]
30. Gray JM, Karow DS, Lu H, Chang AJ, Chang JS, Ellis RE, Marletta MA, Bargmann CI. Nature. 2004;430:317–322. [PubMed]
31. Reinking J, Lam MM, Pardee K, Sampson HM, Liu S, Yang P, Williams S, White W, Lajoie G, Edwards A, Krause HM. Cell. 2005;122:195–207. [PubMed]
32. Soupene E, Lee H, Kustu S. Proc Natl Acad Sci USA. 2002;99:3926–3931. [PMC free article] [PubMed]
33. Soupene E, King N, Feild E, Liu P, Niyogi KK, Huang CH, Kustu S. Proc Natl Acad Sci USA. 2002;99:7769–7773. [PMC free article] [PubMed]
34. Kustu S, Inwood W. Transfus Clin Biol. 2006;13:103–110. [PubMed]
35. Putnam RW, Filosa JA, Ritucci NA. Am J Physiol. 2004;287:C1493–C1526. [PubMed]
36. Hill CA, Fox AN, Pitts RJ, Kent LB, Tan PL, Chrystal MA, Cravchik A, Collins FH, Robertson HM, Zwiebel LJ. Science. 2002;298:176–178. [PubMed]
37. Robertson HM, Wanner KW. Genome Res. 2006;16:1395–1403. [PMC free article] [PubMed]
38. Python F, Stocker RF. J Comp Neurol. 2002;445:374–387. [PubMed]
39. Jones WD, Cayirlioglu P, Kadow IG, Vosshall LB. Nature. 2007;445:86–91. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...