Olfactory responses in a gustatory organ of themalaria vector mosquito Anopheles gambiaeHyung-Wook Kwon, Tan Lu, Michael Rutzler, and Laurence J. Zwiebel*

Department of Biological Sciences, Center for Molecular Neuroscience, Institute of Chemical Biology, and Program in Developmental Biology,Vanderbilt University, Nashville, TN 37235

Edited by Obaid Siddiqi, National Center for Biological Sciences, Bangalore, India, and approved July 17, 2006 (received for review February 9, 2006)

The proboscis is an important head appendage in insects that hasprimarily been thought to process gustatory information duringfood intake. Indeed, in Drosophila and other insects in which theyhave been identified, most gustatory receptors are expressed inproboscis neurons. Our previous characterization of the expressionof AgOR7, a highly conserved odorant receptor (OR) of the Afro-tropical malaria vector mosquito Anopheles gambiae in the label-lum at the tip of the proboscis was suggestive of a potentialolfactory function in this mosquito appendage. To test this hy-pothesis, we used electrophysiological recording and neuronaltracing, and carried out a molecular characterization of candidateOR expression in the labellum of A. gambiae. These studies haveuncovered a set of labial olfactory responses to a small spectrumof human-related odorants, such as isovaleric acid, butylamine, andseveral ketones and oxocarboxylic acids. Molecular analyses indi-cated that at least 24 conventional OR genes are expressedthroughout the proboscis. Furthermore, to more fully examineAgOR expression within this tissue, we characterized the AgORprofile within a single labial olfactory sensillum. This study pro-vides compelling data to support the hypothesis that a cryptic setof olfactory neurons that respond to a small set of odorants arepresent in the mouth parts of hematophagous mosquitoes. Thisresult is consistent with an important role for the labellum in theclose-range discrimination of bloodmeal hosts that directly impactsthe ability of A. gambiae to transmit malaria and other diseases.

olfaction � proboscis � insect � olfactory receptor neuron

In the Afrotropical malaria vector mosquito Anopheles gambiaeand in other insects, olfactory signal transduction is initiated by

G protein-coupled receptors (GPCRs) on the dendrites of olfactoryreceptor neurons (ORNs), which have, thus far, been characterizedin several insect species (1–4). In A. gambiae, 79 GPCR geneshypothesized to encode odorant receptors (AgORs) have beenidentified (5). Genes encoding candidate odorant receptors (ORs)are diverse, with one notable exception comprising AgOR7 andother members of a highly conserved nonconventional OR sub-family that is widely expressed throughout insect olfactory organs(2, 5–7). In addition to widespread expression in olfactory organssuch as the antennae and maxillary palps of A. gambiae and Denguevirus vector mosquito Aedes aegypti, Ag�AaOR7 has recently beenlocalized to �25 distinct type-2 (T2) sensilla on the proboscis (8, 9).In contrast to Ag�AaOR7, their Drosophila ortholog, DOr83b (1),is not expressed in the proboscis of the adult fruit fly. This resultsuggests that mosquitoes and perhaps other bloodfeeding insectsmay contain a set of olfactory inputs derived from the proboscis thatis absent in other arthropods.

Recent studies have demonstrated that Drosophila Dor83b isnot directly responsive to odorants but, rather, is a generalcomponent of the olfactory signal transduction machinery (10–13). In A. gambiae, AgOR7 expression in the proboscis may,therefore, support similar olfactory capacity, leading to theprediction that the proboscis would be responsive to odorantstimuli and that other conventional odorant-activated AgORswould also be expressed in the proboscis. To test this hypothesis,we conducted electrophysiological experiments to characterize

the olfactory responses from the proboscis of A. gambiae to adiverse panel of odorants that included several human sweatcompounds (14, 15). Furthermore, we characterized AgOR geneexpression in the proboscis and axonal projections to antennallobes (ALs), a primary olfactory processing center in the insectbrain (16). Taken together, the resulting data strongly supportthe view that the proboscis is an accessory olfactory organ in themalaria vector mosquito.

ResultsCharacterization of Olfactory Responses from the Proboscis of FemaleAn. gambiae. We initially used the electrolabellogram (ELG), atransepithelial electrophysiological recording adapted from thewell established electroantennogram (EAG) (17) technique,from the surface of the labellum of the proboscis to examinewhether this appendage of female adult A. gambiae manifestsperipheral olfactory responses to a diverse panel of odorantstimuli (Table 1, which is published as supporting information onthe PNAS web site). In these analyses, the mosquito labellumdisplayed robust olfactory responses to several compounds (Fig.1), which elicited either fast downward or, occasionally, upwardvoltage changes. Tested chemicals included butylamine, previ-ously isolated from human effluents (18), butanol, and severalshort-chain aliphatic carboxylic acids, such as acetic, butylic,isovaleric, oxobutylic, and oxovaleric acids, each resulting insignificant ELG responses in the labellum compared with solventcontrols (t test, P � 0.01, Fig. 1 C and D). In these studies, severalacidic stimuli elicited upward responses in ELG recordings athigh (10�2) concentrations (Fig. 1 C and D), whereas controlrecordings from the shaft of A. gambiae proboscis (arrowhead inFig. 1 A) corresponding to a putative nonolfactory area of theproboscis showed no significant responses to acetic acid (filledarrowhead, Fig. 1 C and D) or other acidic odorants (data notshown). Importantly, as a control, olfactory responses were notobserved when the Drosophila melanogaster labellum was testedfor sensitivity toward the full spectrum of odorants used in thisstudy (Fig. 1E). This specificity suggests that upward ELGresponses elicited in these assays reflect bona fide processes thatare due to the excitation of olfactory receptors on the mosquitolabellum that are not present in the fly. Furthermore, small butsignificant downward ELG responses were also observed withketones, such as acetylpyridine and acetylthiophene, as wellas the Henkel 100, a complex odorant mixture (Fig. 1D, t test,P � 0.05).

Olfactory responses from a representative T2 sensillum on thelabellum of A. gambiae (boxed area Fig. 1 A and arbitrarilydenoted as S1 in Fig. 1B) were also examined by using singlesensillum recordings (SSRs). This sensillum is stereotypically

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: AgOR, Anopheles gambiae odorant receptor; AL, antennal lobe; EAG,electroantennogram; ELG, electrolabellogram; OR, odorant receptor; ORN, olfactory re-ceptor neuron; SOG, subesophageal ganglion; SSR, single sensillum recording.

*To whom correspondence should be addressed. E-mail: l.zwiebel@vanderbilt.edu.

© 2006 by The National Academy of Sciences of the USA

13526–13531 � PNAS � September 5, 2006 � vol. 103 � no. 36 www.pnas.org�cgi�doi�10.1073�pnas.0601107103

Dow

nloa

ded

by g

uest

on

Dec

embe

r 24

, 202

0

located on the medial portion of the right labellum between thethird and fourth mechanosensory hairs (Fig. 1B), closely approx-imating the vicinity of the ELG recording site (Fig. 1B, arrow-head) that has been shown to house neurons expressing AgOR7

(8). To test for olfactory activity, individual S1 sensilla wereexamined for physiological responses to a broad range of odorantstimuli where robust responses to several ketone odorants,including acetothiophene, acetylpyridine, acetylthiazole, andacetylphenone, as well as significant responses to butylamine andother acidic compounds were observed (Fig. 2; and see Fig. 5,which is published as supporting information on the PNAS website). Furthermore, an examination of response amplitudes from

Fig. 1. ELG recordings from the epidermis of the proboscis labellum. (A)Scanning electron micrograph of the head and appendages of a female A.gambiae. The dotted box indicates the labellum at the distal tip of theproboscis. The arrowhead depicts the shaft of the proboscis. (B) Schematicdorsal view of the labellum, anterior (a) and posterior (p) axes, as indicated.The black arrowhead indicates the area of the ELG recordings; the arrowrepresents the S1 sensillum located between the third and fourth long hairs,see Fig. 2. (C) ELG traces recorded from the proboscis of female A. gambiae andD. melanogaster. Stimulus onset and duration of 0.5 s are shown above orbelow each trace as a horizontal bar. Upward ELG responses to isovaleric,oxovaleric, oxobutylic, and acetic acids were frequently observed (arrows). Noprominent ELG responses were found in nonchemosensory areas. (D and E)Summaries of ELG responses in female A. gambiae (D) and D. melanogaster(E). Values (in millivolts) depict means � SE. Hatched and solid bars in Dcorrespond to upward and downward deflections, respectively. Significantdifferences of ELG responses from the solvent control (water, mineral oil, ordiethyl ether) are represented with asterisks. *, 0.01 � P � 0.05; **, 0.001 � P �0.01; ***, P � 0.001, n � �8–13. ELG values without an asterisk indicate nosignificant difference from solvent controls (t test, P � 0.05); n � 12. n.s., notsignificant.

Fig. 2. SSRs from S1 sensillum on the labellum. (A) Representative 10-s extra-cellular recording traces from S1 sensillum in the absence of odorant stimulus(spontaneous response rate; mean values � SE are provided in the adjacenthistogram) and in response to two odorants. Bars above each trace indicate 0.5-sstimulus. Each trace is accompanied by a histogram showing the distribution ofindividual action potentials during each 10-s recording. The odorants elicit spiketrains in S1 sensillum, where two neurons are distinguishable by spike amplitude(labeled with ‘‘a’’ and ‘‘b’’ on each trace). (B) Summary of odorant responseprofiles of S1 sensillum in the labellum to 16 odorants. Acetophenone familycompounds elicit strong responses from both a and b neurons, whereas acidicodorants, such as oxovaleric acid, strongly stimulate only the b neuron. Olfactoryresponses were evaluated by subtracting spike numbers during the 1 s beforeodorant stimulation from spike values 1 s after the onset of stimuli. Asterisksindicate statistical significance of olfactory responses to each chemical comparedwith solvent control. *, 0.01 � P � 0.05; **, 0.05 � P � 0.0001. Bars without anasterisk indicate no significant difference from solvent controls (t test, P � 0.05,n � 10). All values depict mean � SE.

Kwon et al. PNAS � September 5, 2006 � vol. 103 � no. 36 � 13527

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Dec

embe

r 24

, 202

0

representative odorant-induced spike trains revealed the exis-tence of at least two distinct neurons associated with each S1sensillum. Of these, an identifiable neuron (denoted as a)characterized by spike amplitudes of �55 �V responded toacetophenone family compounds butylamine and oxovaleric,oxobutylic, and acetic acids (Fig. 2B Inset). A distinctive b neuronwith a larger spike amplitude selectively responded to acetophe-none family compounds (Fig. 2B). In contrast, S1 sensillumneurons showed no response to isovaleric acid, which evokesstrong ELG responses (Fig. 1) as well as 1-octen-3-ol, 2-methylphenol, 4-methyl phenol, indole, geranyl acetone, and ammonia,which have been identified as human sweat compounds andshown to elicit significant responses from A. gambiae antennalpreparations (14, 15, 19–21), (Fig. 1D).

Central Projections of Proboscis Neurons to the Brain. Axonal pro-jections to the ALs and subesophageal ganglion (SOG) of theinsect CNS convey information associated with distinct chemo-sensory modalities (22). Indeed, an examination of these diag-nostic deutocerebrum axonal targets provides a highly reliablemethod that has been used to distinguish olfactory from gusta-tory sensory neurons (23, 24). To determine the projectionpatterns of proboscis ORNs, anterograde dye-tracing studiesfrom the mosquito proboscis to the CNS were carried out. Asexpected, these experiments revealed extensive neuronal ar-borization to the SOG (double arrowhead in Fig. 3 C and D),consistent with a major role for the mosquito proboscis ingustatory processing (25). Importantly, axonal projections (ar-rowhead in Fig. 3 C and D) derived from proboscis neurons andtargeting distinct regions of the ventroposterior domain (pos-teromedial region of the AL in a ventral section in Fig. 3D) ofthe ALs were also observed (Fig. 3 C and D, arrow) furthervalidating the presence of a cryptic set of ORNs in this append-age. Although the Drosophila nc82 monoclonal antibody (thekind gift of R. F. Stocker, University of Fribourg, Fribourg,Switzerland) did not adequately outline AL neuropil to definedistinct mosquito glomeruli, we were able to partially distinguishthe AL glomerular structure as a result of background fluores-cence with excitation at 458 nm. In this manner, we were able toidentify a region corresponding to at least two (denoted as L1and L2, respectively, Fig. 3G) posteromedial glomeruli in bothALs that clearly and reproducibly received afferents from pro-boscis ORNs (arrowheads, Fig. 3 E–J). Of these, the L1 glomer-ulus appears to be predominantly labeled.

Expression and Localization of the ORs in the Proboscis and SingleSensillum. Diverse olfactory responses should correlate with theexpression of several AgOR family members. To examine thisrelationship, we dissected whole proboscises, including shaftsfrom male and female heads, and performed a series of non-quantitative RT-PCR analyses (Fig. 6 and Supporting Methods,which are published as supporting information on the PNAS website) using a set of primers specific to each candidate AgOR (5).In a total of eight experiments (four for each gender), 16 AgORswere reproducibly identified in cDNAs prepared from both maleand female proboscises (shaded rows, Table 2, which is publishedas supporting information on the PNAS web site), and anadditional 9 AgORs (shaded rows with asterisk, Table 2), werereproducibly amplified exclusively from female tissue.

In an attempt to further dissect AgOR expression within theproboscis, we used a single sensillum RT-PCR approach toidentify AgORs within an individual S1 sensillum on the label-lum of the proboscis (see Supporting Methods). Antisense RNAamplification (26) was used to generate sufficient material forcDNA synthesis from individual S1 sensilla after SSR analyses,and AgOR7 expression was used as an assessment of ORNcDNA integrity. In this manner, only cDNA samples positive forAgOR7 were subsequently screened for the presence of other

AgORs identified from the aforementioned whole-proboscisRT-PCR studies (Table 2). Of 10 AgOR7-positive S1 sensillumpreparations, one OR in particular (AgOR6) was consistentlyamplified in 6 preparations (arrow in Table 2). In addition,AgOR53 was detected in 3 preparations, AgOR12 and 18 wereeach detected twice, and seven other AgORs were detected onlyonce (Table 2). Taken together, these data strongly suggest thatAgOR6 is expressed in ORNs associated with S1 sensillum onthe proboscis of A. gambiae. Double-labeling studies usingAgOR6 in situ hybridization coupled with AgOR7 immunostain-ing were also carried out to more precisely localize AgOR6transcripts in the labellum of A. gambiae. Here, AgOR6 mRNAwas detected in multiple cells throughout the medial portion ofthe labellum along with AgOR7 protein (Fig. 4). These studiesalso confirm that AgOR6 mRNA is consistently coexpressed

Fig. 3. Central projection patterns of proboscis neurons to the brain. (A andB) Schematic sagittal (A) and ventral (B) views of the mosquito brain. Eachsubregion indicates neuropil in the mosquito brain. Pr, protocerebrum; OL,optic lobe. (C) Twelve-micrometer Spurr’s plastic section of the mosquito brainin the sagittal plane (corresponding to A) stained with neurobiotin from theproboscis labellum (magenta). Proboscis neurons intensely arborize into theSOG (double arrowheads). A subpopulation of neurons sends axonal projec-tions (arrowhead) to a ventroposterior region (arrow) in the AL (dashedcircle). (D) Confocal micrograph of the ventral view of the whole mosquitobrain stained with neurobiotin from proboscis neurons (magenta). Neuropilstaining with monoclonal Drosophila nc82 antibody (green) did not demar-cate subregions in the mosquito brain such as the ALs. Cell body staining byTOTO3 is shown as yellow. Dashed boxes represent the AL in each side of thebrain. Arborization into the SOG (double arrowheads) and projection (arrow-heads) to the posteromedial region of the ALs (arrows) from proboscis neu-rons are consistent with the sagittal view in C. (E–G) Enlarged confocal imageof the right AL (as in B) of a single 12-�m mosquito brain section stained withneurobiotin from proboscis neurons. Background autofluorescence (detectedby excitation at 458 nm with a 505- to 550-nm band-pass filter) providesdemarcation of each glomerulus of the AL. Bright green represents DNAstaining of cell bodies (E and G). (F) Two glomeruli targeted from proboscisORNs are distinguishable in posteromedial glomeruli of the AL. (G) Onestrongly stained glomerulus denoted as L1 and the other weakly stainedglomerulus denoted as L2. (H–J) Detailed confocal image of the left AL (as inB) of a single 12-�m plastic section. Bright green represents DNA staining ofcell bodies (H and J). A distinct posteromedial glomerulus (denoted as L1)stained with neuronal projections of proboscis ORNs is contralateral to the L1glomerulus in the right AL in G. L2 was not found in this section. (Scale bars,50 �m.)

13528 � www.pnas.org�cgi�doi�10.1073�pnas.0601107103 Kwon et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 24

, 202

0

with AgOR7 protein in a subset of labellum ORNs (Fig. 4C,arrowheads), whereas the remainder of ORNs presumably ex-presses other conventional AgORs. AgOR6 transcripts were alsolocalized to ORNs in the adult antennae (data not shown).

DiscussionOlfactory Responses in the Gustatory Organ of the Mosquito. Thisreport extends previous studies (8) that localized AgOR7 to theantennae, maxillary palpi, and, unexpectedly, the labellum of theproboscis of A. gambiae, which has, until now, been thought tobe an exclusively gustatory sensory organ. Here, ELG analysesrevealed significant labial olfactory responses to odorants suchas several ketone odorants, butylamine, which has been identi-fied in human skin emanations (18), and other short-chaincarboxlyic acids (e.g., acetic, isovaleric, lactic, oxobutylic, andoxovaleric acids). These acidic odorants frequently elicited up-ward ELG responses at high (10�2) and low (10�4) concentra-tions (Fig. 1 C and D). To verify whether these ELG responseswere the result of nonphysiological artifacts as reported in otherinsects (27), an identical recording electrode was used to assessolfactory responses from nonchemosensory proboscis regions,such as the shaft (Fig. 1 A, arrowhead), where no activity wasobserved (Fig. 1 C and D, arrowhead). Furthermore, parallelrecordings were made from the labellum of Drosophila, where noOR expression has been observed (6) and where, once again, weobserved no significant olfactory responses (Fig. 1E). It has beenreported that acetic acid elicits downward responses in EAGrecordings in A. gambiae (15), whereas upward EAG responsesat high doses have been reported in both houseflies (24) and thesable fly, Stomoxys calcitrans (28). However, at lower doses(�0.1-mg application to a filter paper), acetic acid and otheracidic compounds (propionic, butylic, and valeric acids) haveelicited downward responses in the housefly (27). Moreover, inA. gambiae, the ELG responses to acidic compounds showed afast rising phase at 10�2 dilutions compared with identical stimulirecorded from nonchemosensory areas in the proboscis shaft(Fig. 1D, arrows and arrowheads, respectively).

Taken together, these results strongly suggest that upwardELG responses to acidic compounds as well as the downwardresponses (depolarizations) associated with acetylpyridine,acetylthiophene, butylamine, and other compounds in thelabellum are bona fide electrophysiological responses fromORNs. Indeed, all of the acidic stimuli identified in this studyhave been shown to be present in human sweat�skin emana-tions and, in many cases, have been directly linked to behav-ioral responses in A. gambiae (15, 29–32). The same is true forammonia, which works synergistically with other odorants toattract A. gambiae (19).

A more detailed characterization of olfactory function in onerepresentative (S1) labellum sensillum is provided when SSRstudies are used, where vigorous spike trains to ketones andacidic compounds (Figs. 2 and 6) were observed. In keeping with

our ELG data, acetothiophene, acetylpyridine, and butylamineas well as oxocarboxylic and acetic acids elicited robust andcharacteristic olfactory responses from identifiable S1 neurons.However, in contrast to a significant ELG response, isovalericacid failed to evoke significant SSR recordings, whereas strongresponses were observed for acetylthiazole and acetophenone,two odorants that did not evoke significant ELG deflections. Itis not surprising that individual SSR profiles would diverge frommore broadly tuned ELG responses that are subject to severalvariables, including placement of the recording electrode as wellas sensilla and ORN population density. Indeed, equivalentdilutions of one odorant, (�-pinene) in Drosophila similarly doesnot provoke EAG responses (33) but strongly activates at leastone specific (ab7a) antennal ORN in SSR studies (34). Ingeneral, we note that the large-amplitude S1 sensillum ORNappears to be more specialized as its responses were restricted toacetophenone and other ketone stimuli, whereas the small-amplitude S1 ORN displays more generalized responses to bothacidic and ketone odorants as well as robust sensitivity tobutylamine.

Overall, the compounds that elicit strong responses from labialsensilla represent odorants of lower volatility relative to thoseknown to evoke responses from the antennae of A. gambiae (14).Indeed, the vapor pressures of these odorants comprise a narrowrange [lactic acid 0.08 torr (1 torr � 133 Pa); isovaleric acid 0.36torr; acetylpyridine 0.37 torr; acetophenone 0.4 torr] relative toodorants that evoke antennal responses, such as ammonia andindole, with vapor pressures of 6,650 and 10.4 torr, respectively.This finding suggests that A. gambiae, and perhaps other hema-tophagous insects, may use their labella to detect low volatilekairomones, which may be important for orientation behaviorsat close proximity. The importance of close-range olfactory cuesis not without biological precedent. Indeed, part of the penul-timate stage for oviposition site selection occurs at extremelyclose range in many species of lepidopteran insects, where,presumably, chemosensory information is processed (35). Thissuggests that these animals obtain critical chemical informationat close proximity but without direct contact. Several mosquitospecies show enhanced attractive orientation flights towardhuman skin compounds as temperature is increased (36), indi-cating that human kairomones are likely to be evaporated anddetected by the insects. This result is consistent with the hy-pothesis that a small set of sensory neurons expressing AgOR7in the proboscis of A. gambiae, and perhaps other mosquitoes,may be important for determining olfactory profiles in closeproximity to a host, where they provide critical olfactory infor-mation to the mosquito as part of the penultimate steps inalighting, probing, and bloodfeeding behaviors.

Central Projections of Proboscis Neurons to the Antennal Lobes of theBrain. Anterograde dye fillings from proboscis neurons revealedextensive arborization to a distinct set of at least two glomeruli

Fig. 4. Colabeling of AgOR6 and AgOR7 by in situ hybridization and immunostaining on a sagittal section of the labellum. (A) In situ hybridization withantisense AgOR6 probes (red), where AgOR6-positive cells are evident. (B) AgOR7-positive cells detected with antibody immunostaining (green). (C) Mergedimage of A and B; colocalization of AgOR6 and AgOR7 is observed (arrowheads). (D) Merged image from C together with bright-field illumination.AgOR6-positive cells are located in a medial region of the labellum. v, ventral; d, dorsal; a, anterior; p, posterior. (Scale bars, 20 �m.)

Kwon et al. PNAS � September 5, 2006 � vol. 103 � no. 36 � 13529

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Dec

embe

r 24

, 202

0

in A. gambiae ALs (Fig. 3), It is, therefore, possible that, in A.gambiae, these posteromedial glomeruli represent a proboscis-specific projection area. If this region encompasses the entireproboscis projection zone of the A. gambiae AL, where theresponses of 24 proboscis AgORs are directed, then it is rea-sonable to suggest that each of these glomeruli is targeted bymultiple AgORs expressing ORNs, implying that a previouslyuncharacterized mechanism may underlie the encoding of ol-factory information from the labellum, where the majority ofORNs express multiple AgORs that target common AL glo-meruli. Alternatively, individual labellum ORNs that expressdistinct AgORs converge onto a restricted number of commonAL glomeruli. In any case, the presence of labial projectionaxons that target the AL provides compelling support for thepresence of a cryptic set of ORNs on what is certainly apredominantly gustatory appendage.

Expression and Function of ORs in the Labellum. Previous studies inHeliothis virescens using whole-appendage expression surveyshave revealed the presence of candidate OR transcripts in theproboscis (37, 38). This result is reminiscent of physiologicalstudies in another lepidopteran, where the labial pit organ (LPO)and its projections of Manduca sexta have defined accessoryolfactory pathways that are exclusively responsive to CO2 (39),where they play a key role in finding host plants at a distance (40,41). Interestingly, the LPO is not responsive to other volatileodorants, suggesting that this structure may be functionally moreclosely related to the maxillary palps in mosquitoes, which arethe site of CO2 sensitivity in these insects (42). In this report, weprovide a demonstration that fully functional AgORs are ex-pressed in chemosensory sensilla located on the labial portion ofthe proboscis, which is a predominantly gustatory appendage inA. gambiae. Importantly, OR expression in the proboscis has,until now, not been described in any other dipteran insects,consistent with electrophysiological studies presented here,demonstrating a lack of olfactory sensitivity in the D. melano-gaster labellum (Fig. 1E). These findings imply that there may beimportant functional and organizational differences between thechemosensory processes of A. gambiae and D. melanogaster thatmay reflect significantly different life-cycle characteristics, in-cluding feeding habits, oviposition demands, and other elements.In this light, it is especially tempting to focus on the anautog-enous requirement for vertebrate blood that is characteristic ofA. gambiae and other hematophagous mosquitoes as a criticaldistinction between the life cycles of these dipterans.

In A. gambiae, the presence of 24 conventional AgORs hasbeen detected from whole-proboscis RT-PCR screens (Table 2).Among the AgORs identified in the S1 sensillum, AgOR6 isobserved in the majority (6 of 10) of these assays, indicating thatit is highly expressed in one or more ORN associated withindividual S1 sensilla and, moreover, may reasonably be ex-pected to be tuned to one or more of the odorants that evoke thestrongest responses in these assays. Analyses of SSR spike trainamplitudes (Fig. 2) suggest that two ORNs are likely to belocated in the S1 sensillum of the proboscis labellum. Further-more, our in situ hybridization and RT-PCR data are consistentwith the view that AgOR6 is expressed in a subset of AgOR7-positive labellum ORNs, and AgOR6 is one of several AgORsthat facilitate olfactory responses in this appendage.

These data argue against our earlier hypothesis that AgOR7may also function in a gustatory role on the labellum of A.gambiae (43) and, instead, provide compelling evidence for thepresence of cryptic ORNs on this mosquito chemosensoryappendage. Although these data do not formally rule out a rolefor AgOR7 in gustation, we favor the hypothesis that AgOR7 isa true homolog to the nonconventional Drosophila Dor83bprotein and that, accordingly, its expression defines the majorityof ORNs in this system. The strong olfactory responses recorded

in this study from the labellum of A. gambiae may conveyinformation that is critical to the later-stage events in blood-feeding, host preference, and other behaviors of this mosquitoand, therefore, may have profound effects on its vectorialcapacity.

MethodsInsect Preparations. A. gambiae sensu stricto (G3 strain) werereared as described (3). Nonbloodfed 3- to 4-day-old femalemosquitoes were used for electrophysiological recordings andneuroanatomical studies. Before the electrophysiological exper-iments described below, adult mosquitoes were cooled at 4°C andrestrained in a pipette tip, holding head and appendages in place.

Odorant Stimulation. Odorants (�98% purity) were obtainedfrom Sigma or Aldrich. Henkel 100, which contains 100 differentvolatile chemicals (44), was supplied by Henkel (Dusseldorf,Germany). Other odorants were chosen on the basis of behav-ioral and electrophysiological responses to A. gambiae, as shownin previous studies (Table 1). A humidified and purified, con-tinuous air stream (4 ml�s) was delivered to mosquitoes througha glass pipette by using a stimulus controller (Syntech, Hilver-sum, The Netherlands). Twenty-five milliliters of diluted odor-ants (10�2 and 10�4, vol�vol) were applied to a filter disk (VWR,West Chester, PA) that was inserted into a Pasteur pipette. Eachodorant was delivered in a 0.5-s, air pulse through the Pasteurpipette to the glass pipette.

ELG. A restrained mosquito in a pipette tip was positioned onmodeling wax (Hygenic, Akron, OH). Antennae, proboscises,and maxillary palpi were carefully attached to Scotch double-stick tape (3M, St. Paul, MN) on a coverglass mounted onmodeling wax. A sharp glass recording electrode with an0.84-mm i.d. with 1–2 M� resistance (World Precision Instru-ments, Sarasota, FL) was prepared by using a horizontal elec-trode puller (Model P-97; Sutter Instruments, Novato, CA) andfilled with 0.1 M KCl. This electrode was placed in contact withthe epithelium of the labellum, together with a similarly pre-pared reference electrode placed on the thorax, as modified fromEAG procedures described in ref. 45. Data were imported to a4-channel IDAC-UAB analyzed with EAG2000 software (Syn-tech, Hilversum, The Netherlands) on a personal computer.

SSR. Intact mosquito proboscises were hand-dissected and placedonto a coverglass by using Scotch double-stick tape (3M).Electrode gel (Spectra 360; Parker Laboratories, Fairfield, NJ)was applied to the amputated part of the proboscis. Glassrecording and reference electrodes (1 mm i.d. with 5–7 M�resistance) were prepared (Model P-97; Sutter Instruments) andfilled with insect saline (46). The reference electrode wasinserted into the electrode gel, and the recording electrode wasinserted into the lumen of a sensillum on the labellum untilaction potentials were achieved (47). The preparation wasviewed at a 1,200� magnification by using a BX-40 microscopewith a 50� LMPlanFl objective lens (Olympus, Melville, NY).Signals were imported via a 4-channel IDAC-USB and analyzedwith AutoSpike software (Syntech) on a personal computer.Offline analysis of each individual 10-s recording was carried outby using AutoSpike software (47) such that small a and large bspike neurons were clearly distinguishable according to theirspike amplitude. Olfactory responses were quantified by sub-tracting the number of action potentials 1 s before odor stimu-lation from the number of spikes 1 s after the onset of odorstimulation from individual preparations.

Anterograde Dye Filling. Anterograde labeling using fluorescencedyes was performed as described (48), with slight modifications.The labellum was severed, and the remaining proboscis shaft was

13530 � www.pnas.org�cgi�doi�10.1073�pnas.0601107103 Kwon et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 24

, 202

0

immediately immersed in a glass electrode filled with 1%neurobiotin (Vector Laboratories, Burlingame, CA) in PBS.Animals were kept in the humid chamber for 5–7 h, after whichmosquito heads were fixed in 4% paraformaldehyde solution at4°C overnight. Brains were dissected and washed in PBS for2–4 h in the dark, followed by dehydration with a 0–100%ethanol series, followed by incubation in 100% propylene oxide(Sigma, St. Louis, MO) for 5 min before a descending ethanol-to-PBS rehydration procedure. The brains were then incubatedwith a 1:50 dilution of streptavidin–Alexa Fluor 546 conjugate(Molecular Probes, Carlsbad, CA) at 4°C overnight. DNAcounterstaining of cells in the mosquito brain was carried outwith a 1:1,000 dilution of TOTO3 (Molecular Probes) in PBSTfor 20 min. Whole mounts of the brain were washed in PBS for1–2 h and mounted on a glass slide with Vectashield (VectorLaboratories). For plastic sectioning, brain preparations weredehydrated in a 25–100% ethanol series, followed by 100%acetone, and subsequently embedded in Spurr’s epoxy resin (49)before 12-�m sections were prepared on a sliding microtome(HM340E; Microm, Waldorf, Germany). Whole mounts as wellas plastic sectioned preparations of the stained mosquito brainwere observed by using an LSM 510 confocal microscope (Zeiss,Thornwood, NY) under which optical sections were scanned at0.5- to 2-�m intervals to capture fluorescence images fromback-filling experiments.

In Situ Hybridization and Immunolabeling. Paraffin-embeddedpreparations were sectioned at 10- to 12-�m thickness by usinga sliding microtome (HM340E; Microm), subsequently dew-axed with Citri-Solv (Fisher BioSciences, Rockville, MD), andrehydrated in an ethanol series to PBS. In situ hybridizationand probe preparation were carried out as described (6, 50),with digoxigenin-labeled RNA probes comprising �800 bp ofAgOR6 coding sequence. Signals were visualized by alkalinephosphatase (AP) coupled to anti-DIG antibodies (Roche,Indianapolis, IN) at 1:1,000 dilution. AP signals were detectedby using Fast Red tablets (Roche) according to the manufac-turer’s instructions. Anti-AgOR7 immunostaining was carriedout as described (8). Images were captured with confocalmicroscopy as described above.

We thank Drs. J. Carlson and E. Hallem (Yale University, New Haven,CT), Dr. H. W Honegger, Mr. R. J. Pitts, and other colleagues in theZwiebel laboratory for their comments on this manuscript; Drs. N.Strausfeld and W. Gronenberg for advice about graphics; Dr. R. F.Stocker for nc82 antiserum; Drs. B. Appel, C. Carter, T. Fitzwater, A.Goldman, H. C. Park, and H. Yan for help with in situ hybridizations;Drs. G. Pitts and R. Baugh for advice on single sensillum PCR; and Ms.P. Russell and Z. Li for mosquito rearing. This work was supported byNational Institutes of Health Grants A1056402 and DC04692 (to L.J.Z.).

1. Vosshall, L. B., Amrein, H., Morozov, P. S., Rzhetsky, A. & Axel, R. (1999) Cell96, 725–736.

2. Krieger, J., Raming, K., Dewer, Y. M., Bette, S., Conzelmann, S. & Breer, H.(2002) Eur. J. Neurosci. 16, 619–628.

3. Fox, A. N., Pitts, R. J., Robertson, H. M., Carlson, J. R. & Zwiebel, L. J. (2001)Proc. Natl. Acad. Sci. USA 98, 14693–14697.

4. Clyne, P. J., Warr, C. G., Freeman, M. R., Lessing, D., Kim, J. & Carlson, J. R.(1999) Neuron 22, 327–338.

5. Hill, C. A., Fox, A. N., Pitts, R. J., Kent, L. B., Tan, P. L., Chrystal, M. A.,Cravchik, A., Collins, F. H., Robertson, H. M. & Zwiebel, L. J. (2002) Science298, 176–178.

6. Vosshall, L. B., Wong, A. M. & Axel, R. (2000) Cell 102,147–159.7. Krieger, J., Klink, O., Mohl, C., Raming, K. & Breer, H. (2003) J. Comp.

Physiol. A 189, 519–526.8. Pitts, R. J., Fox, A. N. & Zwiebel, L. J. (2004) Proc. Natl. Acad. Sci. USA 101,

5058–5063.9. Melo, A. C., Rutzler, M., Pitts, R. J. & Zwiebel, L. J. (2004) Chem. Senses 29,

403–410.10. Neuhaus, E. M., Gisselmann, G., Zhang, W., Dooley, R., Stortkuhl, K. & Hatt,

H. (2005) Nat. Neurosci. 8, 15–17.11. Nakagawa, T., Sakurai, T., Nishioka, T. & Touhara, K. (2005) Science 307,

1638–1642.12. Larsson, M. C., Domingos, A. I., Jones, W. D., Chiappe, M. E., Amrein, H. &

Vosshall, L. B. (2004) Neuron 43, 703–714.13. Elmore, T., Ignell, R., Carlson, J. R. & Smith, D. P. (2003) J. Neurosci. 23,

9906–9912.14. Meijerink, J., Braks, M. A. & Van Loon, J. J. (2001) J. Insect Physiol. 47,

455–464.15. Cork, A. & Park, K. C. (1996) Med. Vet. Entomol. 10, 269–726.16. Stocker, R. F. (1994) Cell Tissue Res. 275, 3–26.17. Alcorta, E. (1991) J. Neurophysiol. 65, 702–714.18. Ellin, R. I., Farrand, R. L., Oberst, F. W., Crouse, C. L., Billups, N. B., Koon

W. S., Musselman, N. P. & Sidell, F. R. (1974) J. Chromatogr. 100, 137–152.19. Smallegange, R. C., Qiu, Y. T., van Loon, J. J. & Takken, W. (2005) Chem.

Senses 30, 145–152.20. Healy, T. P. & Copland, M. J. (2000) Med. Vet. Entomol. 14, 195–200.21. Dekker, T., Steib, B., Carde, R. T. & Geier, M. (2002) Med. Vet. Entomol. 16,

91–98.22. Hildebrand, J. G. & Shepherd, G. M. (1997) Annu. Rev. Neurosci. 20, 595–631.23. Wang, Z., Singhvi, A., Kong, P. & Scott, K. (2004) Cell 117, 981–991.24. Thorne, N., Chromey, C., Bray, S. & Amrein, H. (2004) Curr. Biol. 14,

1065–1079.25. Shanbhag, S. R., Park, S. K., Pikielny, C. W. & Steinbrecht, R. A. (2001) Cell

Tissue Res. 304, 423–437.

26. Baugh, L. R., Hill, A. A., Brown, E. L. & Hunter, C. P. (2001) Nucleic AcidsRes. 29, E29.

27. Kelling, F. J. (2001) PhD thesis (University of Groningen, Germany), pp.95–104.

28. Warnes, M. L. & Finlayson, L. H. (1986) Physiol. Entomol. 11, 469–473.29. Meijerink, J., Braks, M. A., Braak, A. A., Adam, W., Dekker, T., Posthumus,

M. A., Beek, T. A. & Van Loon, J. J. A. (2000) J. Chem. Ecol. 26, 1367–1382.30. Knols, B. G. J., van Loon, J. J. A., Cork, A., Robinson, R. D., Adam, W.,

Meijerink, J., de Jong, R. & Takken, W. (1997) Bull. Entomol. Res. 87, 151–159.31. Enserink, M. (2002) Science 298, 90–92.32. Engstrom, Y., Kadalayil, L., Sun, S. C., Samakovlis, C., Hultmark, D. & Faye, I.

(1993) J. Mol. Biol. 232, 327–333.33. Park, K. C., Ochieng, S. A., Zhu, J. & Baker T. C. (2002) Chem. Senses 27,

343–352.34. de Bruyne, M., Foster, K. & Carlson, J. R. (2001) Neuron 30, 537–552.35. Mechaber, W. L., Capaldo, C. T. & Hildebrand, J. G. (2002) J. Insect Sci.

www.insectscience.org�2.5.36. Schreck, C. E., Kline, D. L. & Carlson, D. A. (1990) J. Am. Mosquito Control

Assoc. 6, 406–410.37. Krieger, J., Grosse-Wilde, E., Gohl, T., Dewer, Y. M., Raming, K. & Breer, H.

(2004) Proc. Natl. Acad. Sci. USA 101, 11845–11850.38. Krieger, J., Klink, O., Mohl, C., Raming, K. & Breer, H. (2003) J. Comp.

Physiol. A 189, 519–526.39. Kent, K. S., Harrow, I. D., Quartararo, P. & Hildebrand, J. G. (1986) Cell Tissue

Res. 245, 237–245.40. Thom, C., Guerenstein, P. G., Mechaber, W. L. & Hildebrand, J. G. (2004)

J. Chem. Ecol. 30, 1285–1288.41. Guerenstein, P. G., Yepez, E. A., Van Haren, J., Williams, D. G. & Hildebrand,

J. G. (2004) Naturwissenschaften 91, 329–333.42. Grant, A. J., Wigton, B. E., Aghajanian, J. G. & O’Connell, R. J. (1995)

J. Comp. Physiol. A 389–396.43. Pitts, R. J., Fox, A. N. & Zwiebel, L. J. (2004) Proc. Natl. Acad. Sci. USA 101,

5058–5063.44. Wetzel, C. H., Oles, M., Wellerdieck, C., Kuczkowiak, M., Gisselmann, G. &

Hatt, H. (1999) J. Neurosci. 19, 7426–7433.45. Ayer, R. K., Jr. & Carlson, J. (1992) J. Neurobiol. 23, 965–982.46. O’Shea, M. & Adams, M. E. (1981) Science 213, 567–569.47. de Bruyne, M., Clyne, P. J. & Carlson, J. R. (1999) J. Neurosci. 19, 4520–4532.48. Ehmer, B. & Gronenberg, W. (1997) Cell Tissue Res. 290, 153–165.49. Spurr, A. R. (1969) J. Ultrastruct. Res. 26, 31–43.50. Goldman, A. L., Van der Goes van Naters, W., Lessing, D., Warr, C. G. &

Carlson, J. R. (2005) Neuron 45, 661–666.

Kwon et al. PNAS � September 5, 2006 � vol. 103 � no. 36 � 13531

NEU

ROSC

IEN

CE

Dow

nloa

ded

by g

uest

on

Dec

embe

r 24

, 202

0