BioSystems 58 2000 133 – 141
Spiking frequency versus odorant concentration in olfactory receptor neurons
Jean-Pierre Rospars
a,
, Petr La´nsky´
b
, Patricia Duchamp-Viret
c
, Andre´ Duchamp
c
a
Unite´ de Biome´trie, INRA,
78352
Jouy-en-Josas, Versailles, Cedex, France
b
Institute of Physiology, Academy of Sciences, CZ-
142 20
Prague
4
, Czech Republic
c
Neurosciences et Syste`mes sensoriels, Uni6ersite´ Claude Bernard, F-
69622
Villeurbanne Cedex, France
Abstract
The spiking response of receptor neurons to various odorants has been analyzed at different concentrations. The interspike intervals were measured extracellularly before, during and after the stimulation from the olfactory
epithelium of the frog Rana ridibunda. First, a quantitative method was developed to distinguish the spikes in the response from the spontaneous activity. Then, the response intensity, characterized by its median instantaneous
frequency, was determined. Finally, based on statistical analyses, this characteristic was related to the concentration and quality of the odorant stimulus. It was found that the olfactory neuron is characterized by a low modulation in
frequency and a short range of discriminated intensities. The significance of the results is discussed from both a biological and a modelling point of view. © 2000 Elsevier Science Ireland Ltd. All rights reserved.
Keywords
:
Intensity coding; Interspike interval; Olfaction; Transduction; Neuron modeling www.elsevier.comlocatebiosystems
1. Introduction
Intensity coding is a major aspect of sensory processes and the most amenable to quantitative
analysis. It is widely believed that a frequency code is used by receptor cells to encode stimulus
intensity e.g. Shepherd, 1988; Duchamp-Viret and Duchamp, 1997; Hildebrand and Shepherd, 1997.
In order to investigate the nature of the frequency code and its role in quality discrimination, one has
begun to analyze the trains of action potentials recorded from olfactory receptor neurons ORNs
in response to various odorants at different concen- trations. A series of concentration – response curves
were described quantitatively and were compared for different odorants. Four main questions were
addressed. How can the response frequency be measured reliably in a noisy background? What are
the average quantitative characteristics of the re- sponse curves in magnitude and sensitivity? What
is the variability of these characteristics among neurons and odorants? Is it possible to account for
these results in the framework of the ORN model proposed Rospars et al., 1996?
Corresponding author. Present address: Unite de Phy- topharmacie et Mediateurs chimiques, INRA, 78026 Versailles
Cedex, France. Fax: + 33-1-30833359. E-mail address
:
rosparsversailles.inra.fr J.-P. Rospars. 0303-264700 - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved.
PII: S 0 3 0 3 - 2 6 4 7 0 0 0 0 1 1 6 - 7
2. Methods
2
.
1
. Recordings Spiking activity was recorded extracellularly
from the olfactory epithelium of the frog Rana ridibunda and the interspike intervals ISIs were
measured 30 s before and 30 s after the onset of 2-s stimulations Fig. 1. The four odorants tested
were
anisole ANI,
DL
-camphor CAM,
isoamyle acetate ISO and
DL
-limonene LIM. They are representative of previously identified
odorant groups, ANI for aromatic, CAM for camphoraceous and LIM for terpenic, except ISO
which is not representative of any group see Duchamp-Viret and Duchamp, 1997. The odor-
ants were delivered as almost square pulses at precisely controled concentrations along a discrete
scale determined by the stimulating apparatus. Due to the very large range of stimulus intensity
used, a logarithmic measure C of the intensity was used, C = logL, where L, expressed in moll, is
the number of molecules of odorants per unit volume of odorized air delivered to the prepara-
tion. A total of 550 stimulations of 40 ORNs were recorded.
2
.
2
. Response detection A two-step quantitative method was developed
to distinguish the ISIs in the response from the spontaneous spikes. For each spike, the instanta-
neous firing frequency F
i
was determined as the inverse of the interspike interval x
i
. The response was defined as the first uninterrupted series of at
least three ISIs having F
i
higher than a given threshold F
u
and significantly shorter than the spontaneous ISIs as judged by a Mann and Whit-
ney test Conover, 1980. F
u
was chosen 1.5 spike s above the median of the spontaneous firing
frequency of the neuron studied the spontaneous ISIs being distributed asymmetrically between 0
and 3.85 spikes with median 1.15 spikes; Rospars et al., 1994. A total of 350 significant
responses were found. Each response was charac- terized by its median firing frequency F. The
curves F versus C were plotted for a given neuron stimulated by a single odorant at different concen-
trations Fig. 2. The plots with only one or two responses were discarded. One was left for statisti-
cal
analyses with
57 curves
including 335
responses.
Fig. 1. Recordings from a neuron at increasing concentration of the odorant limonene, with 30-s prestimulus spontaneous, 2-s stimulation vertical dotted line indicates onset of stimulation and 30-s poststimulus activities. Concentrations C in log moll are
given on the left.
Fig. 2. Concentration – response curve with logistic black line and arc tangent grey line fittings. Same example as in Fig. 1.
Response properties maximum firing frequency F
M
, threshold C
t
, saturation C
s
, and dynamic range DC are indicated for the logistic curve.
zero frequency; in the case of logistic the same was done for all concentrations up to and includ-
ing C . As in a previous modelling work e.g.
Rospars et al., 1996, these calculated parameters were not used directly but converted to more
meaningful characteristics, the maximum response F
M
horizontal asymptote, the logconcentrations at threshold C
t
and close to saturation C
s
, and the dynamic range DC=C
s
− C
t
giving the ratio in log units of the extreme concentrations see Fig.
2.
3. Results
