Journal of the American Mosquito Control Association, 16(3):234-24O,20ffi Copyright O 2000 by the American Mosquito Control Association, Inc. A WIND TUNNEL BIOASSAY SYSTEM FOR SCREENING MOSQUITO REPELLENTS P J. SrnnprNcroN,r T. P HEALy AND M. J. W. Copr-lNo2 Department of Biological Sciences, Wye College, University of London, Wye, Ashford, Kent TN25 SAH, United Kingdom ABSTRACT A wind tunnel bioassay system to screen mosquito repellents is described. A wind tunnel is utilized to exploit the upwind flight response of host-seeking mosquitoes. Mosquitoes within the wind tunnel are activated with human breath, fly upwind, and land on heated chick skins. This behavioral sequence results in a consistently high percentage of the test population approaching repellent or control stimuli. The bioassay system is calibrated with diethyl methylbenzamide againstAedes aegypti and demonstrates a reproducible dose- response relationship. The persistence of diethyl methyl benzamide after a 1-h period is also recorded. The design of the bioassay system permits simultaneous, independent testing of 3 candidate repellents. The wind tunnel bioassay system is compared to other techniques for evaluating mosquito repellents. KEY WORDS Mosquito, Aedes aegypti, repellent, diethyl methylbenzamide, wind tunnel INTRODUCTION Deet (diethyl methyl benzamide) was 1st report- ed as a mosquito repellent by McCabe et al. (1954) and is still regarded as a highly effective repellent. Deet does, however, have some unpleasant char- acteristics: it is a plasticizer, is often considered to have an unpleasant odor and feel, and has been as- sociated with some toxic side effects (Gryboski et al. 1961, Miller 1982). Davis and Sokolove (1976) demonstrated that deet altered the response of lactic acid (an attractive component of human sweat) re- ceptors on the antennae of Aedes aegypti (L.) mos- quitoes. However, the structure-activity relation- ship of deet and other structurally unrelated but effective repellents has yet to be fully understood (Skinner and Johnson 1980). This comparative lack of knowledge has led investigators to search for new and more effective repellents to take the ap- proach of screening large numbers of potential can- didate compounds. The human biting test is generally considered to be the most realistic method of evaluating the ef- fectiveness of a compound. A human arm is treated with a known dose of the potential repellent and introduced to a cage of mosquitoes. Recordings are made of the number of mosquitoes attempting to bite and are compared to an untreated control arm. The effectiveness of the compound is assessed in terms of the observed reduction in biting between the treated arm and the control arm. The arm test is a comparatively slow method of testing. A lim- ited number of compounds can be tested on an in- dividual's arm, and time must be allowed for the treated arm to return to the control level of attrac- tiveness. Using several individuals can introduce significant variation into the trial. The volunteers have to be prepared to be bitten by mosquitoes, especially during control studies. Finally, any com- t Deceased. 2 To whom correspondence should be addressed. pounds of unknown or suspect toxicity or irritancy have to be tested on animals prior to human testing. To circumvent these problems, Bar-Zeev and Smith (1959) introduced the concept of evaluating repellents by treating artificial membranes that cov- ered small reservoirs of heated blood. The effec- tiveness of the repellent was assessed in terms of the numbers of blood-fed mosquitoes in compari- son to the amount of repellent applied. Rutledge et al. (1976) modified this technique by introducing mosquitoes into a cage that contained several wells of heated blood covered by membranes that were treated with a range of amounts of repellent. The system is considered to give the mosquitoes a free choice, as they are exposed to a range of amounts and can select membranes that have been treated with tolerable levels of repellent. Rutledge et al. (1976, 1985) consider that this design more closely represents the natural situation, in which a mosqui- to is free to select less well treated areas of skin or even an untreated host. Buescher et al. (1982b) turther modified this de- sign by strapping a similar type of cage onto a hu- man forearm. Buescher et al. (1982a) tested a range of repellents against Ae. aegypti, and this test pro- cedure is recognized as a standard test method (Anonymous 1984). This technique has been used by Buescher et al. (1983) to examine the persis- tence of deet over a period of time, and Rutledge et al. (1985) have constructed mathematical models relating repellency to time and dose. While conducting the human arrn-biting test with deet and Ae. aegypti, we observed consider- able variation between replicates. A major conhib- uting factor appeared to be variation in the per- centage of mosquitoes in the test population that actually responded to either the control or test arm. We decided to try to reduce this variation by de- signing a bioassay system that would result in a consistently high percentage of the test population responding to the control and repellent stimuli. A 234
Transcript
Journal of the American Mosquito Control Association, 16(3):234-24O,20ffiCopyright O 2000 by the American Mosquito Control Association, Inc.
A WIND TUNNEL BIOASSAY SYSTEMFOR SCREENING MOSQUITO REPELLENTS
P J. SrnnprNcroN,r T. P HEALy AND M. J. W. Copr-lNo2
Department of Biological Sciences, Wye College, University of London, Wye,Ashford, Kent TN25 SAH, United Kingdom
ABSTRACT A wind tunnel bioassay system to screen mosquito repellents is described. A wind tunnel isutilized to exploit the upwind flight response of host-seeking mosquitoes. Mosquitoes within the wind tunnelare activated with human breath, fly upwind, and land on heated chick skins. This behavioral sequence resultsin a consistently high percentage of the test population approaching repellent or control stimuli. The bioassaysystem is calibrated with diethyl methylbenzamide againstAedes aegypti and demonstrates a reproducible dose-response relationship. The persistence of diethyl methyl benzamide after a 1-h period is also recorded. Thedesign of the bioassay system permits simultaneous, independent testing of 3 candidate repellents. The windtunnel bioassay system is compared to other techniques for evaluating mosquito repellents.
KEY WORDS Mosquito, Aedes aegypti, repellent, diethyl methylbenzamide, wind tunnel
INTRODUCTION
Deet (diethyl methyl benzamide) was 1st report-ed as a mosquito repellent by McCabe et al. (1954)and is still regarded as a highly effective repellent.Deet does, however, have some unpleasant char-acteristics: it is a plasticizer, is often considered tohave an unpleasant odor and feel, and has been as-sociated with some toxic side effects (Gryboski etal. 1961, Miller 1982). Davis and Sokolove (1976)demonstrated that deet altered the response of lacticacid (an attractive component of human sweat) re-ceptors on the antennae of Aedes aegypti (L.) mos-quitoes. However, the structure-activity relation-ship of deet and other structurally unrelated buteffective repellents has yet to be fully understood(Skinner and Johnson 1980). This comparative lackof knowledge has led investigators to search fornew and more effective repellents to take the ap-proach of screening large numbers of potential can-didate compounds.
The human biting test is generally considered tobe the most realistic method of evaluating the ef-fectiveness of a compound. A human arm is treatedwith a known dose of the potential repellent andintroduced to a cage of mosquitoes. Recordings aremade of the number of mosquitoes attempting tobite and are compared to an untreated control arm.The effectiveness of the compound is assessed interms of the observed reduction in biting betweenthe treated arm and the control arm. The arm testis a comparatively slow method of testing. A lim-ited number of compounds can be tested on an in-dividual's arm, and time must be allowed for thetreated arm to return to the control level of attrac-tiveness. Using several individuals can introducesignificant variation into the trial. The volunteershave to be prepared to be bitten by mosquitoes,especially during control studies. Finally, any com-
t Deceased.2 To whom correspondence should be addressed.
pounds of unknown or suspect toxicity or irritancyhave to be tested on animals prior to human testing.
To circumvent these problems, Bar-Zeev andSmith (1959) introduced the concept of evaluatingrepellents by treating artificial membranes that cov-ered small reservoirs of heated blood. The effec-tiveness of the repellent was assessed in terms ofthe numbers of blood-fed mosquitoes in compari-son to the amount of repellent applied. Rutledge etal. (1976) modified this technique by introducingmosquitoes into a cage that contained several wellsof heated blood covered by membranes that weretreated with a range of amounts of repellent. Thesystem is considered to give the mosquitoes a freechoice, as they are exposed to a range of amountsand can select membranes that have been treatedwith tolerable levels of repellent. Rutledge et al.(1976, 1985) consider that this design more closelyrepresents the natural situation, in which a mosqui-to is free to select less well treated areas of skin oreven an untreated host.
Buescher et al. (1982b) turther modified this de-sign by strapping a similar type of cage onto a hu-man forearm. Buescher et al. (1982a) tested a rangeof repellents against Ae. aegypti, and this test pro-cedure is recognized as a standard test method(Anonymous 1984). This technique has been usedby Buescher et al. (1983) to examine the persis-tence of deet over a period of time, and Rutledgeet al. (1985) have constructed mathematical modelsrelating repellency to time and dose.
While conducting the human arrn-biting testwith deet and Ae. aegypti, we observed consider-able variation between replicates. A major conhib-uting factor appeared to be variation in the per-centage of mosquitoes in the test population thatactually responded to either the control or test arm.We decided to try to reduce this variation by de-signing a bioassay system that would result in aconsistently high percentage of the test populationresponding to the control and repellent stimuli. A
234
236 JounNx- oF THE AMERIcAN MosQUITo CoNTRoL AssocIATIoN VoL. 16, No. 3
Vhw in dir€dlon h'
KEY:ffinry; z: lctvateo dercoat f lterl 3: lilol€culer sievo f lter; /t: stainle$ siaol mosh: 5: l-loat€d glss qlindor;6:
Gl83s so€s;7: Mcqufto oioo*say aram;i, at vd"o o*'";gtwaterltor' In; 10: Weteriow out todEin; I l: wataf
female Ae. aegypti were placed into each bioassaychamber. Prior to testing, the female mosquitoeshad been kept in cages containing 400-50O mos-quitoes of mixed sex with access to sugar solution.To reduce variation in mosquito activity due to cir-cadian rhythms, all bioassays began at 1200 h.
The open ends of the 4 glass cylinders were cov-ered in 2 layers of cleanly plucked, 3-week-oldchick skin. The temperature of the skins was raisedto 34'C (human skin temperature) by passing heat-ed water at 3OO mymin through each cylinder. Ini-tial trials with mature chicken skin often leaked wa-ter where adult feathers had been removed,Howeve! the double layer of chick skin rarelyleaked. To avoid contamination, we did not recyclethe water. Temperature variation during the bioas-say was minimized (<l'C) by the large capacity ofthe water tank. Three of the skins were individuallytreated with 30 pl of ethanol, which was just suf-ficient to wet the surface area of a skin. The ethanolcontained deet (Sigma Chemical Company, Poole,United Kingdom; minimum pttity 97Eo) diluted togive a range of doses (O.32, 0.16,0.08, 0,04, 0.02,0.01, 0.005, or 0.0025 mg/cm'�) when applied to askin. The 4th skin, the control, was untreated.
The 4 bioassay chambers were aligned with the4 cylinders, and air was passed through the bioas-say equipment. To prevent cross contamination oc-curring between the chambers, a glass cross was
inserted at the midpoint of the 4 chambers. Thecross protruded upwind, isolating each cylinder.
To activate the mosquitoes in the flight chambers,an experimenter exhaled twice into the inlet ductof the wind tunnel. Human exhalations were knownto produce pulselike fluctuations in the concentra-tion of carbon dioxide within the airflow in thewind tunnel. Pulses of carbon dioxide had been ob-served to activate Anopheles gambiae Giles s.s.mosquitoes, which took off and flew upwind (Healyand Copland 1995). The exhalations had the sameeffect on Ae. aegypti, and the mosquitoes in theflight chambers flew upwind toward the skins.
After the exhalations, a video recording of themosquitoes that landed on the skins was taken forlO min. The bioassay chambers were then removed,and the skins were left for I h at 34"C within theairflow of the wind tunnel. The chambers were thenrealigned with the cylinders as iri the initial bioas-say, and the recording procedure was repeated. Af-ter the l-h bioassay, the mosquitoes were killed,and the Layflat tubing, mosquito netting, and skinswere discarded. All the stainless steel frames weresoaked overnight in 5Vo decon solution and washedin water. The glass cylinders were soaked overnightin chromic acid. washed in water, soaked in decon,and again washed in water. To prevent contamina-tion of the bioassay equipment with any behavior-ally active compounds derived from human skin,
Fig. 4. Proportion of Ae. aegypti repelled by deet during initial bioassay.
of 12.55,12.00, 11.88, and.13.12 (95Vo CL,11.48-13.29) during the l-h bioassay.
DISCUSSION
The wind tunnel design resulted in a high per-centage of Ae. aegypri responding to the test stim-uli. When human breath was introduced to the airflow, the mosquitoes took off, flew upwind, and
landed on the upwind netting. Mosquitoes in thecontrol flight chamber rapidly located the skin andprobed its surface. Freeze-frame counts typicallyrecorded around TOVo of the mosquitoes on the con-trol skin at any one time. There was a continualflux of mosquitoes landing, probing, or leaving theskin. It was considered that, apart from rare indi-viduals, all of the mosquitoes would contact theskin during the bioassay period. At the higher doses
Eoo -o.oL 1
ot ooCL9 - ro..:CDo
-3
5
4
3!to= )CLoL 1
troE ooc9 - ro.
ctt -.0
-3
4
-5
Deet log"dose mg/cm2
aegypti repelled by deet after I h.
Y = 2.121+ 0.899X r2= O.6737
ED 50= 0.094 95% CL(0.086-0.104)
Fig. 5. Proportion of Ae
SePTENrssn 2000 WnrD TUNNEL Broessly non Mosqurro RrpelLnNrs
Table l. Effective dose values for deet agunst Aedes aegypti.l
ED5G ED9OAuthor Date Methodology (mg/cm') (mg/cm2)
ED95(mg/cm'�)
Bar-Znev and SmithRutledge et al.Rutledge et al.Buescher et al.Rutledge et al.Buescher et aI.Curtis et al.Buescher et al.Gupta et al.
Cockcroft et al.
Present study
o.260.0310.0244.O420.0034o.02874
0.00160.02870.00180.00260.0s
(0.03-0.0e)'0.35
(0.17-{.45)0.016
(0.015-{.017)
orro
o.0145
0.00570.13010.00860.0137
0.54(0.20-7.88)o.79
(o.s9--2.26)0 .189
(o.1614.22s)
1959 Independent membrane feeder1976 Free-choice membrane feeder1978 Free-choice membrane feeder1982 Free-choice cage on human arm1983 Free-choice membrane feeder1983 Free-choice cage on human arm1987 Free-choice cage on human arm1987 Free-choice membrane feeder1989 Free-choice cage on human arm
1998 Independent biting test on humanarm
Independent membrane feeder
Independent wind tunnel bioassay
t Vafues in parentheses ue 95Vo confidence limits.2ED5O, SOVo effective dose; ED90, 9OVo effective dose; ED95, 9596 effective dose.
of deet, the mosquitoes were observed to approachwithin a few centimeters of the treated skin andthen turn away sharply. Mosquitoes exposed tolower doses landed on the skin but took off after acomparatively short time.
The increase in variation between replicates ofthe lower doses during the l-h trial was unexpect-ed. The linear dose-response relationship impliesthat these low doses should result in almost negli-gible levels of repellency. However, the results in-dicate that occasionally a low dose cala repel 36Voof the mosquitoes.
One possible explanation is that variation in thephysical factors that affect the processes of absorp-tion and evaporation that reduce the dose of deetover the hour period may have a more pronouncedeffect on the lower doses in comparison to the high-er doses.
Comparing previously published values of effec-tive doses ofdeet for Ae. aegypti (Table 1) indicatesthat different techniques produce different values.ED50 values for the free-choice cage, when usedon membranes (approximately 0.03 mg/cm2), arctypically an order of magnitude lower than inde-pendent membrane tests (approximately 0.3 mg/cm'�). It is expected that a free-choice design willproduce lower values than an independent systembecause the mosquitoes can move to the lower dos-es (Rutledge et al. 1976). Curtis et al. (1987) havedemonstrated that the apparent effectiveness of arepellent can be influenced by the choice availableto the mosquitoes. When the free-choice cage isstrapped onto a human arm, the ED50 values (ap-proximately 0.003 mg/cm'�) are an order of magni-tude lower than the free-choice membrane results.This apparent increase in effectiveness is probablyrelated to the untreated human arm being a moreattractive stimulus than a membrane: this results in
more mosquitoes biting the control and the lowdoses.
The wind tunnel bioassay is a system where t}tedoses are tested independently, and chick skin isused as an alternative to human skin. The wind tun-nel system also has the effect of activating host-seeking behavior in a high percentage of the testpopulation of mosquitoes. The ED values are ap-proximately equivalent to the free-choice mem-brane bioassay results. Both of these bioassays giveresults that are lower than, but in the same order ofmagnitude as, the independent biting test on a hu-man arm conducted by Cockroft et al. (1998).
The wind tunnel bioassay has been used in com-mercial studies (patent WO 96/08 147) to screen anextensive range of compounds for repellency. Threecompounds are tested simultaneously at tle deetED9O dose, allowing the performance of each com-pound to be compared against deet. The l-h trialhas proven an important feafure, as some com-pounds can elicit comparable repellency to deetduring the initial trial, but very few are comparableafter t h. Compounds or combinations of com-pounds that produce effective levels of repellencyin both trials are further evaluated by a biting teston a human arm.
ACKNOWLEDGMENTS
We wish to thank Quest International Ltd., Ash-ford, Kent, United Kingdom, for financial supportthroughout the development and operation of thebioassay system.
REFERENCES CITED
[Anonymous]. 1984. Standard test methods for laboratorytesting of non-commercial mosquito repellent formula-
Cockcroft A, Cosgrove JB, Wood RJ. 1998. Comparativerepellency of commercial formulations of deet, per-methrin and citronella against the mosquito Aedes ae-gypti, using a collagen membrane technique comparedwith human arm tests. Med Vet Entomol 12:289-294.
Collet D. 1991. Modelling binary data London, UnitedKingdom: Chapman and Hall.
Crawley MJ. 1993. Glim for ecologists Oxford, UnitedKingdom: Blackwell Scientifi c Publications.
Curtis CE Lines JD, Ijumba J, Callaghan A, Hill N, Kar-imzad MA. 1987. The relative efficacy of repellentsagainst mosquito vectors of disease. Med Vet Entornoll : 109 - l 19 .
Davis EE, Sokolove PG. 1976. Lactic acid-sensitive re-
ceptors on the antennae of the mosquito Aedes aegypti.J Comp Physiol 1O5:43-54.
Gryboski J, Weinstein D, Ordway NK. 1961. Toxic en-cephalopathy apparently related to the use of an insectrepellent. N Engl J Med 264:.289-291,.
Gupta RK, Rutledge LC, Letourneau WJ. 1989. An im-proved laboratory test cage for testing repellents on hu-man volunteers. J Am Mosq Control Assoc 5:436-438.
Healy TP, Copland MJW. 1995. Activation of Anophelesgambiae mosquitoes by carbon dioxide and humanbreath. Med Vet Entomol 9:331-336.
McCabe ET, Barthel WE Gertler SI, Hall SA. 1954. Insectrepellents. III. N,N-Diethylamides. J Org Chem 19:493-498.
McCullagh R Nelder JA. 1983. Generalised linear modelsLondon, United Kingdom: Chapman and Hall.
Miller JD. 1982. Anaphylaxis associated with insect re-pellent. N Engl J Med 3O7:1341-1342.
Rutledge LC, Collister DM, Meixsell VE, EisenbergGHG. 1983. Comparative sensitivity of representativemosquitoes (Diptera: Culicidae) to repellents. J MedEntomol 20:506-51O.
Rutledge LC, Moussa MA, Belletti CI. 1976. An in vitroblood-feeding system for quantitative testing of mos-quito repellents. Mosq News 36:283-293.
Rutledge LC, Moussa MA, Lowe CA, Sofield RK. 1978.Comparative sensitivity of mosquito species and strainsto the repellent diethyl toluamide. J Med Entomol 14:536 541.
Rutledge LC, Wirtz RA, Buescher MD, Mehr ZA. 1985.Mathematical models of the effectiveness and persis-tence of mosquito repellents. J Am Mosq Control Assocl:56-62.
Skinner WA, Johnson HL. 1980. The design of insect re-pellents. In: Ariens EJ, ed. Drug design Yolume lONew York: Academic Press. p 277-305.
