electroantennograms
Drosophila olfactory
Dynamic properties of
O R I G I N A L P A P E R
Dynamic properties of Drosophila olfactory electroantennograms
Julia SchuckelÆShannon MeisnerÆ Pa¨ivi H. TorkkeliÆ Andrew S. French
Received: 3 December 2007 / Revised: 5 February 2008 / Accepted: 20 February 2008 / Published online: 5 March 2008 ÓSpringer-Verlag 2008
Abstract Time-dependent properties of chemical signals are probably crucially important to many animals, but little is known about the dynamics of chemoreceptors. Beha-vioral evidence of dynamic sensitivity includes the control of moth flight by pheromone plume structure, and the ability of some blood-sucking insects to detect varying concentrations of carbon dioxide, possibly matched to host breathing rates. Measurement of chemoreceptor dynamics has been limited by the technical challenge of producing controlled, accurate modulation of olfactory and gustatory chemical concentrations over suitably wide ranges of amplitude and frequency. We used a new servo-controlled laminar flow system, combined with photoionization detection of surrogate tracer gas, to characterize electro-antennograms (EAG) of Drosophila antennae during stimulation with fruit odorants or aggregation pheromone in air. Frequency response functions and coherence func-tions measured over a bandwidth of 0–100 Hz were well characterized by first-order low-pass linear filter functions.
Filter time constant varied over almost a tenfold range, and was characteristic for each odorant, indicating that several dynamically different chemotransduction mechanisms are present. Pheromone response was delayed relative to fruit odors. Amplitude of response, and consequently signal-to-noise ratio, also varied consistently with different com-pounds. Accurate dynamic characterization promises to provide important new information about chemotransduc-tion and odorant-stimulated behavior.
Keywords AntennaChemosensoryNoiseOdor Frequency response
Abbreviations
PID Photoionization detector EAG Electroantennogram ppm Parts per million OR Odorant receptor
Introduction
Time-dependent properties of chemoreception are crucially important for many functions, but poorly understood. For example, the flight path of a male moth is determined by the temporal-spatial structure of the female pheromone plume (Justus et al. 2002; Mafra-Neto and Carde´ 1995;
Willis and Baker 1984), and plume temporal structure signals distance to the source (Justus et al. 2002; Murlis et al. 1992). Neural substrates for decoding this informa-tion have been proposed in moth CNS (Vickers et al.
2001), but little is known about the dynamic responses of antennal sensory neurons that initially detect the odorant.
Dynamic plume sensitivity also controls mosquito orien-tation to CO2 from mammalian hosts (Geier et al. 1999), andTriatoma infestans, a hematophagous insect, is selec-tively attracted to CO2pulsations in the human breathing range (Barrozo and Lazzari2006).
Dynamic characterization has helped to identify func-tional components, and quantitative representation of information, in mechanoreceptors and photoreceptors (Juusola and French 1997; Juusola et al. 2003), but this approach has been very limited in chemoreceptors (Justus J. SchuckelS. MeisnerP. H. TorkkeliA. S. French (&)
Department of Physiology and Biophysics,
method is to release short pulses of chemical into a stream of air or other fluid passing over the sensory organ (Barrozo and Kaissling 2002; Bau et al. 2002), but this limits stimulation to relatively low frequencies (typically below about 30 Hz), and concentration at the sensory receptor is difficult to estimate because of diffusion during transit from stimulator to preparation. In contrast, mechanical or light stimulation can be delivered at fre-quencies several orders of magnitude higher, their physical intensity at the sensory receptor can be measured accu-rately, and they can be delivered with a wide bandwidth of frequencies, allowing techniques such as direct spectral analysis (Bendat and Piersol 1980), nonlinear systems analysis (Marmarelis and Marmarelis 1978) and natura-listic stimulation (Karmeier et al.2006).
Wide bandwidth dynamic odorant stimulation was achieved by turbulent flow in a wind tunnel (Justus et al.
2005) combined with photoionization measurement of a surrogate tracer gas. This approach allowed both linear and nonlinear systems analysis of moth pheromone receptors at higher frequencies than achieved previously. However, turbulent stimulation does not allow accurate control of the amplitude and frequency ranges of the stimulus. Addi-tionally, odorant concentration versus distance is unpredictable in turbulent flow, raising the possibility that concentration measured even a short distance from the sensory receptor may not accurately estimate the actual stimulus.
Drosophilaantennal olfactory neurons vary in the rate of decay of their responses to step odorant presentations, and this variation is related to the type of odorant receptor involved (Hallem et al. 2004). Defined wide bandwidth olfactory stimulation of Drosophila has recently been achieved using servo-controlled release of odorant and surrogate gas into a small laminar flow wind tunnel (French and Meisner2007). Here, we used this system to record the first full frequency response measurements of Drosophila electroantennograms (EAG) to a series of fruit odors and to aggregation pheromone.
Materials and methods
Stimulation
Olfactory stimulation was performed by a laminar air flow system (Fig.1) whose operational parameters have been described before (French and Meisner2007). Primary air-flow was produced by a small fan (Proten DFC601512M, Cooler Guys, Kirkland, WA) at one end of a 90 mm square plexiglas box. The fan was driven by a 9 V DC power
plexiglas flow tube (22 mm internal diameter, 110 mm long). The fly was positioned at the far end of the flow tube, within 2–3 mm of the exit and 2–3 mm of the tube center line.
Secondary flow from a cylinder of compressed air containing 1,000 ppm propylene tracer gas (BOC, Halifax, NS, Canada) was regulated to 20 kPa initial pressure. It flowed through an odorant cartridge made from the shaft of a 5 ml transfer pipet (Fisher Scientific, Ottawa, ON, Canada), containing a rectangular piece of filter paper
Fig. 1 Drosophilaantennal recording and stimulation system. Lam-inar air flow, created by a 60 925 mm computer fan pushing air through a 5 mm pitch honeycomb, flowed through a circular plexiglas tube 115 mm long and 22 mm internal diameter (flow tube). Odorant chemicals were introduced by air flowing over a rectangular piece of filter paper (45915 mm) in a cartridge made from a 5 ml transfer pipet. The air source contained 1,000 ppm propylene as a tracer gas.
Air with mixed odorant and propylene tracer flowed through the tip of a Pasteur pipet in the center of the flow tube, with the tip variably occluded by the flat surface of a 5 mm silicone elastomer bead, moved by a servo-controlled electromagnetic pusher. The fly was positioned in the center of the circular tube and 3 mm from its mouth. Tracer propylene was detected by a photoionization detector (PID), with the mouth of the aspirating needle (0.76 mm internal diameter) located about 1 mm above the fly
123
in the center of the flow tube, and 10 mm from its origin.
The tip of the pipet was variably occluded by a bead of silicone elastomer (Mastercraft tub and tile silicone sealant, Canadian Tire Corporation, Toronto, ON Canada), driven by a servo-controlled electromechanical stimulator. Odor-ant chemicals and mineral oil were purchased from Sigma (Oakville, ON, Canada). Odorants were mixed with min-eral oil at 20% v/v and pheromone at 50%. Volumes of 50 ppm were loaded into separate cartridges for each mixture.
Stimulus and response measurements
Tracer gas concentration was measured by a miniature photoionization detector (mini-PID, Model 200A, Aurora Scientific Inc, Aurora, ON, Canada), whose needle inlet probe tip was located directly above and within 1 mm of the fly antenna. The PID had a frequency response of 0–330 Hz and a concentration range of 0.05–500 ppm propylene. The typical measurement range used here was 1–20 ppm. The PID needle probe was 57 mm long with an internal diameter of 0.76 mm, giving a volume of 0.0259 ml. PID pump speed was 660 ml/min giving a time delay of 2.35 ms.
Primary air flow velocity was estimated from the flow tube dimensions and the delay between movements at the pipet tip and the resulting PID signal, after subtracting the PID needle delay, to be 2.37 m/s, giving a volume flow of 900 ml/s. Secondary airflow relative to primary airflow was estimated from the mean concentration of propylene at the PID to be 1:100, giving a secondary air flow rate of 9 ml/s.
Flies (Drosophila melanogaster, Oregon R #2376, Bloomington Drosophila stock center, Bloomington, IN) were maintained in the laboratory. Flies of both sexes were used within two days of hatching. Procedures for recording EAG were similar to those described previously (Alcorta 1991). Animals were held in the cut end of a 100ll plastic pipet tip. A reference glass microelectrode (*1 lm tip diameter) was inserted into one eye, while a larger glass microelectrode (*20lm tip diameter) was pushed against the distal end of one antenna. Both electrodes were filled with Drosophila saline (Hazel et al. 2003). EAG were recorded as electrical current with a List EPC-7 patch clamp amplifier (ALA Scientific Instruments, Westbury, NY).
Current recordings are proportional to voltage recordings but give lower noise levels (Minor and Kaissling2003; Yao et al. 2005). All experiments were performed at room temperature (20±2°C) and humidity*40%.
Experimental control and data processing
All experiments were controlled by custom-written soft-ware via a personal computer and a data acquisition
computer via a 33-bit maximum length binary sequence algorithm (Golomb1967) driving a 12-bit digital-to-analog convertor into the position servo control. The PID voltage and EAG current were digitized via a 16-bit analog-to-digital convertor and sampled at 5 ms intervals. Sampled time domain data were transferred to the frequency domain using the fast Fourier transform (Cooley and Tukey 1965) in segments of 512 sample pairs. Frequency response functions (amplitude and phase) between PID voltage (input) and EAG current (output) were calculated by direct spectral estimation and plotted as Bode plots of phase and log amplitude versus log frequency. Coherence functions (Bendat and Piersol1980) were calculated from the same data. Frequency response functions were fitted by a coherence-weighted minimum square error process to a first-order low-pass filter function (Justus et al. 2005):
GðjfÞ ¼aexpðj2pfDtÞ 1=ð1þj2pfstÞ ð1Þ whereG(jf) is the complex gain of the frequency response, a is a constant amplitude,Dtis a pure time delay,sis the time constant of the linear filter andj2 ¼ ffiffiffiffiffiffiffi
p 1
:Coherence functions, c2(f) were used to estimate the information capacity of the antennal response (Shannon and Weaver 1949):
R¼ Z
log21=1 c2ðfÞ
df: ð2Þ
Experimental protocols
Experiments used natural fruit odorants that stimulate Drosophila antennal chemoreceptors: butyl butyrate, isoamyl acetate, phenylethyl alcohol, and hexyl acetate (Stensmyr et al. 2003), and the Drosophila aggregation pheromone, (Z)-11-octadecenyl acetate (Hedlund et al.
1996). In the fruit odorant experiments each fly, of either sex, was stimulated with all four odorants, in turn, using a random number sequence to create a different order of stimulation for each animal. Total stimulation time for each odor was 100 s. Between different odors, flies received 100 s of air flow without odor. In pheromone experiments, male flies were stimulated for 100 s. The experimental apparatus was flushed with clean air for periods of at least 10 min between experiments. Ambient air at the apparatus was removed by a direct connection to the building exhaust system. Statistical analysis was performed by Prophet 6.0 software (AbTech Corporation, Charlottesville, VA).
Results
Fruit odorant experiments were conducted on a total of 42
trations at the animal used previously in Drosophila experiments (Yao et al.2005), taking into account the ratio of primary to secondary air flows and the filter paper dimensions. Preliminary experiments showed that EAG signals dropped significantly at lower concentrations but tended to saturate with increased concentrations. A typical frequency response function is shown for stimulation by hexyl acetate (Fig.2). The fitted parameters of Eq.1 were used to draw the solid lines through the amplitude and phase data. True amplitude units of the frequency response cannot be given because the ratio of odorant molecules to tracer gas was unknown. They are shown as antennal
outwards from the electrode, or into the antenna, and positive current increased with olfactory stimulation.
Olfactory frequency response functions were reliable, and well-fitted by Eq. 1, with high coherence function values (near unity) over a wide bandwidth.
Initial experiments suggested a reduction in sensitivity, or adaptation, during the first few seconds of stimulation.
To quantify this effect, recorded data were separated into consecutive sets of 2,048 input–output data pairs (10.24 s) and frequency response functions calculated for each set.
Plotting the fitted data versus time showed that adaptation was mild, and complete within 50 s of initial stimulation (Fig.3). Therefore, all further analysis was performed using data collected between 50 and 100 s of initial stimulation, including the data shown in Fig. 2. All further data analysis was applied to experiments where complete responses were obtained for the full 100 s to all four odorants or pheromone. Four parameters, a,s,Dt (Eq. 1) andR(Eq. 2) were calculated for each compound.
The relatively constant response observed during 50 s (Fig.3) also indicates that the concentration of the odorants at the animal was approximately constant during the
Fig. 2 Frequency response function between tracer gas concentration (input) and electroantennogram current (output) for hexyl acetate stimulated Drosophila. Amplitude values represent the ratio of electroantennogram current from the patch clamp amplifier to tracer gas concentration. Amplitude and phase data were fitted by Eq.1 (solid lines) with the following parameters:a= 4.44 pA/ppm,s= 7.06 ms,Dt=-1.78 ms. Fitted amplitude and time constant values are indicated by arrows; first-order filter asymptotic slope is also shown.
Note that the phase relationship lags (decreasing phase value) at low frequencies due to the filter function, but the negative delay caused by the PID needle causes it to lead (increasing phase values) at frequencies above*40 Hz. Inset shows 2 s of original recordings of PID output of tracer gas concentration (upper) and the resulting Drosophilaelectroantennogram current (lower) during pseudorandom stimulation
Fig. 3 Electroantennogram responses adapted slightly for about 30 s after the onset of odorant stimulation. Sampled data was divided into five consecutive sets of 2,048 input–output pairs, giving mean time values of 5.12, 15.36, 25.6, 35.84, and 46.08 s. Frequency response functions were estimated for each set of data and fitted with Eq.1, as in Fig. 2. Data (mean± standard error) are shown for normalized amplitude (a/a0, wherea0was the amplitude at 5.12 s and for time constant,s, for the four odorants used.Numbersof experiments are indicated next to the labels
123
experiments. This could be due to the high ratio (*100) of primary to secondary air flow (see ‘‘Materials and meth-ods’’), so that evaporation was limited by the relatively slow secondary air flow. In preliminary experiments, response amplitude remained approximately constant for periods of at least 500 s (data not shown).
To test for gender differences in fruit odorant sensitivity, statistical analysis was applied to eight experiments on female flies and nine experiments on male flies. Mann–
Whitney rank sum tests (2-sided) failed to show any sig-nificant differences between male and female flies for any of the sixteen groups tested (4 odorants by 4 parameters).
Therefore, all further fruit odorant tests used combined data from both sexes. Pheromone experiments were all con-ducted on male flies.
The four fitted parameters (Fig.4) were measured from 28 flies that all received the four different fruit odorants in random order. Parameter values were compared by
Fried-tion. The hypothesis was rejected for all measurements of time constant, s (P = 0.0001) and information capacity, R, (P = 0.0001). For the amplitude parameter, a, the hypothesis was rejected for the group (P= 0.0001), but not between the effects of isoamyl acetate and hexyl acetate (P[0.05). For the delay parameter,Dt, the hypothesis was not rejected (P = 0.0523) although there was a significant difference between the effects of butyl butyrate and hexyl acetate (P\0.05).
In summary, the four fruit odorants each gave signi-ficantly different dynamic responses, with mean time constant varying from 4.40 (isoamyl acetate) to 35.5 ms (phenylethyl alcohol). Amplitude also varied with odorant from 3.64 (isoamyl acetate) to 1.47 pA/ppm (phenylethyl alcohol) and was matched by variation in information capacity. Time delay was small and negative with no consistent evidence that it varied with different odorants.
Pheromone responses had much lower amplitude than fruit odor responses (Fig.4), which made the experiments more difficult and strongly reduced the estimated infor-mation capacity. Pheromone experiments were conducted separately on 24 male flies. Amplitudes,a, and information capacities, R, were about one third of the lowest fruit odorant parameters (phenylethyl alcohol). Time constant values,s, were within the range of the fruit odorants (mean 21.1 ms) but the delay parameter, Dt, was now positive (mean 4.8 ms) instead of negative.
Discussion
Different fruit odors produced characteristically different dynamic responses
The four odorants gave different time constants, amplitudes and information capacities (indicating different signal-to-noise ratios). What is the basis for these differences?
Frequency response functions were well fitted by a first-order low-pass filter function (Eq.1) that also fitted moth EAG responses to pheromones (Justus et al. 2005). For moths it was argued that the filter occurs between odorant arrival at the antennal surface and opening of ion channels to produce receptor current in the chemosensory neuron.
A similar argument can be made for Drosophila, because some time constant measurements were much longer than expected for cell membranes, and adaptation during action potential encoding tends to cause high-pass behavior (Carlsson and Hansson2002). Time-dependent steps could include odorant diffusion through cuticular pores, binding to extracellular proteins, binding to membrane odorant receptor molecules, or intracellular second messenger Fig. 4 Fitted experimental parameters for experiments on 52 flies.
Four fruit odorants were used in random order to stimulate each of 28 flies. Pheromone was used to stimulate a different set of 24 flies.
Amplitude,a, time constant,s, and delay, Dt, were obtained from frequency response functions (Fig. 2) fitted by Eq.1. Information capacity,R, was obtained from corresponding coherence functions via Eq.2. Data are shown as mean values±standard error
Previous work linked initial firing rates and adaptation rates to olfactory receptor molecules (ORs), rather than neuron type, when receptor molecules were expressed in differentDrosophilaneurons (Hallem et al.2004), but the ORs corresponding to the four odorants used here are not yet well defined. Stensmyr et al. (2003) identified eight major functional types of Drosophila antennal sensilla, each containing 1–4 action potential producing neurons, although morphological characterization was not possible.
Using their nomenclature, the four odorants used here would be expected to excite neurons S3-A (butyl butyrate and isoamyl acetate), S4-A (butyl butyrate and isoamyl acetate), S4-B (isoamyl acetate), S5-B (isoamyl acetate and hexyl acetate) and S8-B (phenylethyl alcohol), suggesting that at least three ORs were stimulated here.
Hallem et al. (2004) mapped 13 Drosophila ORs to different receptor neurons in antennal basiconic sensilla.
Their map indicates that isoamyl acetate would stimulate the OR10a and OR22a receptors in ab1D and ab3A neu-rons, as well as the OR19a receptor in an unknown neuron.
The other three odorants were not tested. None of the four odorants were tested in coeloconic sensilla recordings (Yao et al.2005). There is evidence of sexual differences in the density of some odorant receptors inDrosophila(Dobritsa et al. 2003) but no differences in EAG responses to fruit odors were observed here.
Therefore, the available evidence indicates that several different ORs were stimulated by the four odorants, and that some of the odorants stimulated more than one type of OR. However, there was no evidence for multiple filter time constants in the frequency response functions.
Therefore, if dynamic behavior is determined by ORs, the data suggest that different ORs responding to a single odorant have similar time constants. Linkage of dynamic properties to an OR does not necessarily mean that the OR function itself determines the time constant, which could also be due to associations between ORs and different groups of odorant binding proteins or second messenger pathways. The finding that dynamic behavior follows ORs into different neurons (Hallem et al. 2004) makes it less likely that penetration through the cuticle or diffusion to the receptor neuron is controlling dynamics.
Adaptation of olfactory response was small for all odorants and did not seem to affect the time constant (Fig.2). This agrees with single unit recordings of both basiconic (Dobritsa et al.2003) and coeloconic (Yao et al.
2005) sensilla. Single unit recordings to step olfactory stimuli showed some adaptation during the initial second, which would probably not have been detected by the fre-quency response measurements if it occurred with random stimulation.
The negative delay in fruit odor responses (Fig. 4) can be explained by the experimental arrangement. Flow through the PID sampling needle should give a measurement delay of about -2.5 ms, so observed values of -1 to -2 ms imply a maximum EAG delay of less than 1 ms for fruit odors and *7 ms for pheromone. The origin of EAG current is not clearly established. Electrical models have been proposed that include several current sources (Kapitskii and Gribakin 1992), and any contribution of action potentials to EAG has been challenged by experi-ments using chemical treatment to block action potentials (Lucas and Renou1992).
Although the basis of the delay in olfactory responses cannot yet be identified, it is interesting that significant delay was seen in both moth (Justus et al. 2005) and Drosophila pheromone responses, but not in Drosophila fruit odor responses. This could be related to molecular size because the pheromone used here was much larger than any of the fruit odors.
It is difficult to interpret the dependence of frequency response amplitude, a, or information capacity, R, on the different odorants. EAG amplitude could be affected by receptor current amplitude, number of active neurons and positions of the active neurons relative to the recording electrode. Information capacity is directly dependent on signal-to-noise ratio in a linear system. There is no sig-nificant evidence for nonlinearity in the Drosophila EAG recordings, andRclearly varied witha(Fig.4), suggesting that all recordings included a similar background noise level.
Why doDrosophilaneed such rapid responses to fruit odors?
The time constant of *4.5 ms for isoamyl acetate transduction was three times faster than moth pheromone transduction, and would allow detection of odorant changes up to*50 Hz. Pheromone responses were much more delayed than those of fruit odors. Are these differ-ences behaviorally important? Why is there such a wide range of time constants for different odorants? Which stages of chemotransduction are rate limiting? The pres-ent study has raised several interesting questions about olfactory dynamics, and must be followed by detailed analysis of single olfactory neuron responses. However, it has already provided a simple, quantitative dynamic model that can be used to test hypotheses about chemo-transduction. Accurately characterizing the dynamic properties of these neurons should provide important new information about the physiology and ethology of chemosensation.
123
Research and the Dalhousie Medical Research Foundation. The experiments complied with the ‘‘Principles of animal care’’, publi-cation No. 86-23, revised 1985 of the National Institute of Health, and were approved by the Dalhousie University Committee on Animal Care.
References
Alcorta E (1991) Characterization of the electroantennogram in Drosophila melanogaster and its use for identifying olfactory capture and transduction mutants. J Neurophysiol 65:702–714 Barrozo RB, Kaissling KE (2002) Repetitive stimulation of olfactory
receptor cells in female silkmoths Bombyx mori L. J Insect Physiol 48:825–834
Barrozo RB, Lazzari CR (2006) Orientation response of haemato-phagous bugs to CO2: the effect of the temporal structure of the stimulus. J Comp Physiol A 192:827–831
Bau J, Justus KA, Carde´ RT (2002) Antennal resolution of pulsed pheromone plumes in three moth species. J Insect Physiol 48:433–442
Bendat JS, Piersol AG (1980) Engineering applications of correlation and spectral analysis. Wiley, New York, pp 1–302
Benton R (2007) Sensitivity and specificity inDrosophilapheromone perception. Trends Neurosci 30:512-519
Carlsson MA, Hansson BS (2002) Responses in highly selective sensory neurons to blends of pheromone components in the moth Agrotis segetum. J Insect Physiol 48:443–451
Cooley JW, Tukey JW (1965) An algorithm for the machine calculation of complex Fourier series. Math Comput 19:297–301 Dobritsa AA, van der Goes van Naters W, Warr CG, Steinbrecht RA, Carlson JR (2003) Integrating the molecular and cellular basis of odor coding in theDrosophilaantenna. Neuron 37:827–841 French AS, Meisner S (2007) A new method for wide frequency
range dynamic olfactory stimulation and characterization. Chem Senses 32:681–688
Geier M, Bosch OJ, Boeckh J (1999) Influence of odour plume structure on upwind flight of mosquitoes towards hosts. J Exp Biol 202:1639–1648
Golomb S (1967) Shift register sequences. Holden-Day, San Francisco, pp 1–224
Hallem EA, Ho MG, Carlson JR (2004) The molecular basis of odor coding in theDrosophilaantenna. Cell 117:965–979
Hazel MH, Ianowski JP, Christensen RJ, Maddrell SH, O’Donnell MJ (2003) Amino acids modulate ion transport and fluid secretion by insect Malpighian tubules. J Exp Biol 206:79–91
pheromones of Drosophila immigrans, D. phalerata, and D. subobscura. J Chem Ecol 22:1835–1844
Justus KA, Schofield SW, Murlis J, Carde´ RT (2002) Flight behaviour ofCadra cautellamales in rapidly pulsed pheromone plumes.
Physiol Entomol 27:59–66
Justus KA, Carde´ RT, French AS (2005) Dynamic properties of antennal responses to pheromone in two moth species.
J Neurophysiol 93:2233–2239
Juusola M, French AS (1997) The efficiency of sensory information coding by mechanoreceptor neurons. Neuron 18:959–968 Juusola M, Niven JE, French AS (2003) Shaker K+ channels
contribute early nonlinear amplification to the light response in Drosophilaphotoreceptors. J Neurophysiol 90:2014–2021 Karmeier K, van Hateren JH, Kern R, Egelhaaf M (2006) Encoding of
naturalistic optic flow by a population of blowfly motion-sensitive neurons. J Neurophysiol 96:1602–1614
Kapitskii SV, Gribakin FG (1992) Electroantennogram of the American cockroach: effect of oxygen and an electrical model.
J Comp Physiol A 170:651–663
Lucas P, Renou M (1992) Electrophysiological study of the effects of deltamethrin, bioresmethrin, and DDT on the activity of pheromone receptor neurones in two moth species. Pest Biochem Physiol 43:103–115
Mafra-Neto A, Carde´ RT (1995) Influence of plume structure and pheromone concentration on upwind flight of Cadra cautella males. Physiol Entomol 20:117–133
Marmarelis PZ, Marmarelis VZ (1978) Analysis of physiological systems: the white-noise approach. Plenum Press, New York Minor AV, Kaissling KE (2003) Cell responses to single pheromone
molecules may reflect the activation kinetics of olfactory receptor molecules. J Comp Physiol A 189:221–230
Murlis J, Elkington JS, Carde´ RT (1992) Odor plumes and how insects use them. Annu Rev Entomol 37:505–532
Shannon CE, Weaver W (1949) The mathematical theory of communication. University of Illinois Press, Urbana
Stensmyr MC, Giordano E, Balloi A, Angioy AM, Hansson BS (2003) Novel natural ligands forDrosophilaolfactory receptor neurones. J Exp Biol 206:715–724
Vickers NJ, Christensen TA, Baker TC, Hildebrand JG (2001) Odour-plume dynamics influence the brain’s olfactory code. Nature 410:466–470
Willis MA, Baker TC (1984) Effects of intermittent and continuous pheromone stimulation on the flight behaviour of the oriental fruit moth,Grapholita molesta. Physiol Entomol 9:341–358 Yao CA, Ignell R, Carlson JR (2005) Chemosensory coding by
neurons in the coeloconic sensilla of theDrosophila antenna.
J Neurosci 25:8359–8367
A digital sequence method of dynamic
olfactory characterization
Journal of Neuroscience Methods
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j n e u m e t h
A digital sequence method of dynamic olfactory characterization
Julia Schuckel, Andrew S. French∗
Department of Physiology and Biophysics, Dalhousie University, Halifax, Nova Scotia, Canada B3H 1X5
a r t i c l e i n f o
Article history:
Received 11 January 2008
Received in revised form 23 February 2008 Accepted 25 February 2008
Keywords:
Olfaction Drosophila Antenna Sensory receptor Frequency response Photoionization
a b s t r a c t
Measurements of system dynamics, such as input–output frequency response estimation, have been widely used in neuroscience. Dynamic characterization of sensory systems has been particularly useful because both the amplitude and time-dependent properties of sensory input signals can often be accu-rately controlled. However, chemoreceptors have proved less amenable to these approaches because it is often difficult to accurately modulate or measure chemical concentration at a sensory receptor. New meth-ods of dynamic olfactory stimulation have recently been introduced that combine controlled mechanical release of odorant with detection by photoionization of surrogate tracer gas mixed with the odorant. We have developed a modified version of this approach based on rapid binary switching of odorant flow using pseudo-random binary signals (maximum-length sequences, or M-sequences) generated by a software shift register. This system offers several advantages over previous methods, including higher frequency range stimulation, experimental simplicity and the possibility of computational efficiencies. We show that there is predictable dynamic odorant concentration at the sensory receptor and we explore the stimula-tion parameters as funcstimula-tions of total air flow rate and spatial locastimula-tion. A typical applicastimula-tion of the system is shown by measuring the frequency response function of aDrosophilaelectroantennogram.
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
Systems analysis provides an established general method of measuring the dynamic properties of input–output systems (Bendat and Piersol, 1980). Both linear and nonlinear systems anal-ysis have been used to characterize a wide variety of physiological systems (Marmarelis and Marmarelis, 1978), and have become par-ticularly important in neuroscience (French et al., 1972; French and Marmarelis, 1999; Karniel and Inbar, 1999). In the case of linear sys-tems, a complete dynamic characterization, such as the frequency response function or impulse response function, can be used to predict the response of the system to any other stimulus.
Systems analysis often employs a random, or approximately random, signal that is sometimes called “white noise” or “pseudo-random noise”. White noise stimulation can be used to characterize linear, and many nonlinear, biological systems (Marmarelis and Marmarelis, 1978). Practical generation of pseudorandom noise is commonly achieved by deriving an analog signal from an M-sequence shift register (Golomb, 1981), where the term
“sequence” is derived from “maximum length sequence”. M-sequences are generated by applying a feedback loop to a binary shift register so that repeated shift operations cause the binary number stored in the shift register to cycle through every
possi-ble value, giving the maximum length sequence of binary numbers.
M-sequence shift registers can be implemented from physical com-ponents or as computer software. Although not truly random, the order of binary numbers appearing in an M-sequence is sufficiently complex to create a pseudorandom signal, for example, by driving the inputs in parallel to a digital-to-analog converter.
Alternatively, the binary digits appearing at one end of the M-sequence shift register can be used as the input to an unknown system. Because these values are either one or zero, it is possible to perform very efficient computation on the resulting output signals from the system by techniques such as the Walsh–Hadamard trans-form (French and Butz, 1974). This approach has been utilized for rapid characterization of visual function, where multiple receptors and lateral interactions between receptors produce large numbers of parallel output signals from a single input (Sutter, 2001).
Systems analysis of chemoreceptors has proved more difficult than other sensory modalities because of the nature of the stimulus.
The basic requirements for linear and nonlinear systems analysis of sensory receptors are the abilities to modulate the signal in a con-trolled manner over a wide range of frequencies, and to accurately measure both the input stimulus signal and the output neural sig-nal with appropriate time resolution. Dynamic measurements of some insect olfactory receptors have been achieved using pulsed release of chemicals in wind tunnels (Bau et al., 2005). Another approach used randomly varying pheromone concentration