Electrophysiological Studies on the Temporal Organ of the Japanese House Centipede, Thereuonema Hilgendorfi

Some soil arthropods such as the Chilopoda, Diplopoda, Pauropoda, Symphyla and Apterygota possess a pair of sensory organs on the head behind the insertion of the antenna. This organ is referred to as a ‘temporal organ’ (see Haupt, 1979). There have been several ultrastructural accounts of the temporal organs. Based upon these morphological data, the temporal organs are thought to function in several ways: as chemo-or hygroreceptors (Collembola, Karuhize, 1971; Symphyla, Haupt, 1971) or as chemo-, hygro-and/or thermoreceptors (Protura, Haupt, 1972; Pauropoda, Haupt, 1973; Collembola, Altner & Thies, 1976). Behavioural experiments have suggested that the temporal organ of Lithobius (organ of Tömösváry) functions as a hygroreceptor (Tichy, 1972, 1973). Electrophysiological studies which would be important in determining the adequate stimulus of the sensory organ are lacking.

In the present study, the temporal organ of the Japanese house centipede Thereuonema hilgendorfi (Chilopoda) has been studied electrophysiologically. The main aim of the present study is to advance our understanding of the temporal organ by identifying its adequate stimulus and physiological properties.

Adult Japanese house centipedes, Thereuonema hilgendorfi, of either sex were used throughout this study. Animals were collected in the field around Kyushu University, Fukuoka, Japan.

After immobilization by cooling with ice, the animal was fixed to an acrylic platform. Spike activities were recorded extracellularly from a receptor cell by impaling an active electrode about 50 µm deep into the temporal organ from the side of the domed cuticle. The active electrode was a sharpened tungsten wire insulated with lacquer (tip diameter was about 1·0 μm). Glass-coated tungsten electrodes were also used for recording from the perfused temporal organ. Stable, single unit recordings could be made for over 3 h by using these electrodes.

Air of a desired carbon dioxide concentration was prepared as follows (Fig. 1). During a series of carbon dioxide stimuli, we used dry deodorized air in order to minimize the effects of humidity and air-borne chemicals. Air was dried through concentrated sulphuric acid solution and then decarboxylated through potassium hydroxide pellets. After passing through active carbon, the air flow was divided into two parts by a T-tube. One part was used for a carbon dioxide-free conditioning stream. Another part was used for stepwise dilutions of pure carbon dioxide supplied from a tank. The stimulus concentration of carbon dioxide was changed by manipulating six valves, and determined by air flux using five flow meters (Fig. 1). The head of the animal was always exposed to the carbon dioxide-free conditioning stream (4·51 min^-1), and stimulation was performed by adding the second stream (0·5 1 min^-1) containing a known concentration of carbon dioxide to the conditioning stream. A flow-direction changer was operated manually. The outlet of the changer (8 mm in diameter) was set about 5–10 mm from the head, and the flow rate of the air stream was about 150-165 cms⁻¹. This rate had no effect on the spontaneous activity of the receptor cells. Stimulus concentrations of carbon dioxide between 0·05 % and 10 % were prepared by using this apparatus. Stimulus duration was 120s at intervals of 180 s.

Disposable polyethylene syringes were used to observe the phasic responses to lower concentrations (less than 0·05 %) of carbon dioxide and to abrupt changes of carbon dioxide concentration. The syringe was filled with carbon dioxide of a known concentration, which was injected into the stimulus stream through a vinyl tube (TOmm in diameter). The syringe was driven for 2s by a constant speed syringe pump.

The domed cuticle directly covering the temporal organ was partially removed to allow the perfusate to spread over the presumed receptor site. The head of an intact animal was fixed in a chamber (1·0ml in volume), and perfused at a rate of 4·0 ml min^-1. Carbon dioxide solutions of varying concentrations were prepared in the manner of Yoshii, Kashiwayanagi, Kurihara & Kobatake (1980). To examine the effect of pH, we used 25 mmol 1^-1 phosphate buffer (pH 5-8), 25 mmol 1^-1 Tris-HCl buffer (pH 7·2-9·0) and distilled water whose pH was adjusted with 0·1 mol 1^-1 HC1 and 0·1 moll^-1 NaOH. Perfusion experiments were carried out at 25°C.

Humidity stimulation was carried out as described by Yokohari & Tateda (1976), but the flow rate of the air stream was modified to 100 cm s^-1. Heat emission from a tungsten lamp was used as a source of temperature stimulation.

An examination was made of the effects of 10 air-borne chemicals, listed in Table 1, upon impulse frequencies of receptor cells. Each chemical was diluted to 1 % (volume/volume) in paraffin oil. A syringe containing a piece of filter paper immersed in each stimulant solution was used for stimulation. The outlet of the syringe was about 2-0 mm in diameter, and positioned 10 mm from the temporal organ. The flow rate was about 6-4 ms^-1. Stimulation began 2 min after filling a syringe with air, for periods of 2s at 2-min intervals. To examine responses to odours of prey animals and house centipedes themselves, pieces of filter paper were used on which faeces of German cockroaches or house centipedes had been absorbed.

The temporal organ of the house centipede is a paired structure located on the head between the base of the antenna and the pseudo-compound eye (Fig. 2). The organ appears externally as a small protuberance with a central opening (Fig. 3) : like a volcano with a deep crater on its summit. The basal diameter of the protuberance is about 40 μm, and the opening of the pit is about 5 μm in diameter. There are several receptor cells together with about 100 supporting cells under the protuberance. The internal morphology of the temporal organ will be reported elsewhere (K. Yamana & Y. Toh, in preparation).

Single unit recordings were obtained from about 100 receptor cells of the temporal organ. They showed spontaneous discharges of 15-40 impulses s ^! (Fig. 4) which were depressed by carbon dioxide application: the time course was phasic-tonic (Fig. 5). The receptor cell in Fig. 5, spontaneously discharging at about 16 impulses s^-1 in carbon dioxide-free air, was completely inhibited for the first 20 s of 1·0% carbon dioxide application. In the subsequent 70s the impulse frequency slowly increased to a steady discharge level (about 5 impulses s^-1) during continued stimulation. Switching the 1-0% carbon dioxide to carbon dioxide-free air resulted in an abrupt increase in impulse frequency (off response: about 35impulsess^-1) beyond the pre-stimulus level, followed by a gradual recovery to the pre-stimulus level (in about 90s for this specimen). It usually took 1-3 min for a receptor cell to recover its full responsiveness after the cessation of stimulation. When a non-insulated electrode was inserted into the temporal organ, several units could be recorded simultaneously. In such conditions, frequencies of all units decreased when carbon dioxide concentration was raised and increased when carbon dioxide concentration decreased. All receptor cells examined in the present study responded to carbon dioxide with a pattern similar to that shown in Fig. 5.

The receptor was stimulated by air of various carbon dioxide concentrations following adaptation to carbon dioxide-free air. The impulse frequency decreased linearly with the logarithm of carbon dioxide concentration from 0·001 % to 0·1 % in the phasic state and from 0·05% to 5·0% in the tonic state (Fig. 6). The magnitude of the off response also depended upon the stimulus concentration (Fig. 7). Stimulation by air containing carbon dioxide after adaptation to carbon dioxide-free air is unlikely in natural environments. Therefore, responses to changes of carbon dioxide concentration were examined after the animal had been fully adapted to given concentrations of carbon dioxide. It was found that slight increases and decreases of carbon dioxide concentration from a given adapting level resulted in large changes in the impulse frequency (Fig. 8). For instance, steady-state impulse discharges (18 impulses s^-1) in 0·1% carbon dioxide were rapidly decreased to 50% (9impulsess^-1) in 0·13% carbon dioxide in Fig. 8. On average, the impulse frequency in 0·13 % carbon dioxide, after adaptation to 0·1 %, decreased to 60 ± 9 % (N=9).

To examine whether receptor cells of the temporal organ may be stimulated by carbon dioxide which is dissolved in the surrounding receptor lymph, changes of impulse frequency were recorded while perfusing carbon dioxide solution over the receptor site. The receptor cells were first adapted to a perfusate of carbon dioxide-free distilled water before replacing the perfusate with carbon dioxide-containing distilled water. The impulse frequency was unchanged for the first 15 s after this replacement, and then decreased gradually (Fig. 9). The time course of the response to carbon dioxide solution differed from the response to gaseous carbon dioxide, the phasic state being gradual with a smaller peak depression.

As carbon dioxide concentration was increased, impulse frequencies in both phasic and tonic states decreased (Fig. 10A). The dynamic range of the response in the tonic state was as wide as that obtained to gaseous carbon dioxide stimuli.

Dissolving carbon dioxide in an aqueous solution is accompanied by a change of pH. To examine the effect of a lowered pH on the response, the receptor site of the temporal organ was perfused with 25 mmol 1^-1 phosphate buffer solutions of varying pH after adaptation to phosphate buffer of pH 7·5. When the pH was lowered, the impulse frequency in the phasic state decreased, but the depression was smaller than that caused by carbon dioxide solution (Fig. 10B). A carbon dioxide solution in equilibrium with air containing T0% carbon dioxide has a pH of about 5·0, and it was found to depress the impulse frequency to 10% of the pre-stimulus level in the phasic state (open circles in Fig. 10A) whereas the buffer solution at pH 5·0 depressed the impulse frequency to 70% of the pre-stimulus level (filled circles in Fig. 10B). Moreover, in the tonic state (180s after the stimulus onset), impulse frequencies to buffer solutions of pH 5, 6 and 7 were near the pre-stimulus level (Fig. 10B). When the pH was raised with phosphate buffer solution to 8·0, the impulse frequency increased to about 115% of the pre-stimulus level, and was maintained during continued stimulation.

To observe the effects of pH over a wider range, we examined the pH-response relationships with applications of 25 mmol 1^-1 phosphate buffer, 25 mmol 1^-1 Tris-HC1 buffer and distilled water whose pH was adjusted with 0·1 mol 1^-1 NaOH and 0·1 moll^-1 HC1 (Fig. 11). A solution was perfused after adaptation to phosphate buffer of pH 7·5. In all three kinds of solutions, when the pH was raised there was an increase in impulse frequency, whereas when the pH was lowered, impulse frequencies decreased. However, the maximum response amplitude (reduction of impulse frequency), found at pH2 was smaller than the response to 1·0% carbon dioxide either in air or in solution. Stimulation with a solution of pH 2 does not seem to be physiological.

When carbon dioxide is dissolved in solution, CO₂, H₂CO₃, HCO₃^- and CO₃^2- coexist in the solution (Fig. 12A). The degree of hydrolysis of the carbon dioxide is determined by the pH of the solution. Since concentration of H₂CO₃ is always less than 0·4% of that of CO₂(K = 2·6X10^-3), [H₂CO₃] is negligible. Molar fractions of CO₂, HCO₃^- and CO₃^2- depend upon the pH of the solution (Kitano et al. 1976) as shown in Fig. 12A. If the solution becomes acidic, the molar fraction of CO₂ increases (broken line in Fig. 12A). For example, more than 95 % of the total carbon dioxide content in the solution exists in the form of CO₂ molecules at pH 5·0. Because of the properties of carbon dioxide in solution, the following experiment was carried out to determine which molecules or ions stimulate the receptor cell of the temporal organ.

Carbon dioxide solutions in equilibrium with air containing 2 % carbon dioxide were pH adjusted with phosphate buffer or NaOH, so as to stimulate the receptor cell with a solution of known composition. As pH was lowered over the range pH 5–10, thus altering the molar fraction of CO₂ in solution, there was a decrease in impulse frequency (Fig. 12B). These results, together with those in Fig. 10A, suggest that the impulse frequency of a receptor cell decreases with increase in concentration of CO₂ molecules around the receptor site.

The temporal organ was found to respond to changes in humidity, by stimulation with moist and dry air (100% and 0% RH, respectively). When the relative humidity was changed from 100 % to 0 %, the impulse frequency decreased (Fig. 13A). Conversely, when the humidity was changed from 0% to 100%, the impulse frequency increased (Fig. 13B). The recordings shown in Fig. 13A,B were obtained from the same cell. This cell showed the largest responses to humidity stimuli of 29 cells examined. On average, relative impulse frequency was increased by a factor of 1 ·29 ± 0·37 (mean ± S.D., N = 29) when the humidity was raised from 0 % to 100 % (measured for the first second after the beginning of stimulation). The impulse frequency during adaptation to 100% RH was also higher than that during adaptation to 0% RH by a factor of 1·16 ± 0·28 (N = 29).

In response to aldehydes and acids, a decrease in impulse frequency was observed, an increase was observed with amines. Alcohols, the odour of house centipedes and the odour of German cockroaches have little or no effects on the impulse frequencies (Table 1). For all these chemicals, the responses of the recorded cells had quite identical spectra.

Spontaneous activity was not altered by changing the temperature (not shown).

The temporal organ of arthropods is thought to have different functions in different animals; namely, hygroreception, thermoreception or olfaction (see Introduction). In the present study, the temporal organ of the house centipede is shown to act primarily as a CO₂ receptor.

Humidity stimuli could result in changes in the impulse frequency of a receptor cell (Fig. 13). Impulse frequency was changed by a factor of 1·29 ±0·37 in the phasic state, and 1·16 ±0·28 in the steady state when the humidity was increased from 0% to 100%. These responses are less than those generally observed in insect hygroreceptors: a factor of 6 between 70% and 0% RH (locust; Waldow, 1970), a factor of 3 between 100% and 0% RH (cockroach; Yokohari & Tateda, 1976) and a factor of 2 between 90% and 10% RH (honey-bee; Yokohari, Tominaga & Tateda, 1982). The locust moisture receptor responds to a rise in relative humidity from 0% to 70% with a rise in impulse frequency from 10 to 175 impulses s^-1 (measured for the first second, Waldow, 1970). Moreover, hygroreceptive sensilla of insects have a hygroscopic cap structure upon which sensory cilia with mechanoreceptive structures impinge (Yokohari, 1981, 1983), whereas the sensory cilia in the temporal organ of the house centipede do not possess such structures (K. Yamana & Y. Toh, in preparation). These physiological and morphological data suggest that the temporal organ of the house centipede may not be a functional hygroreceptor. Changes in the impulse frequency may be secondary effects due to transient changes in the ionic composition of the receptor lymph.

Receptor cells of the temporal organ also respond to some air-borne chemicals: stimulation with aldehydes and fatty acids decreases the activities of receptor cells, whereas stimulation with three amines increases them. This does not indicate that the temporal organ is an olfactory organ, however, for the following reasons. (1) In general, olfactory organs contain many types of receptor cells of different spectra. Outputs from such different cells are thought to be decoded in the central nervous system for olfactory discrimination (Boeckh, 1980). However, as shown in Table 1, the temporal organ appears to contain receptor cells with identical spectra.

(2) Receptor cells of the temporal organ did not respond to two natural olfactory stimuli examined, the odours of the German cockroach and the house centipede.

(3) The house centipede possesses a pair of long, well-developed antennae, which function as an olfactory organ. It is more likely that the observed effects of the airborne stimuli tested here are indirect, influencing the pH and/or carbon dioxide concentration at the receptor site..

Since spontaneous activities were not affected by temperature changes, the temporal organ does not appear to be a thermoreceptor.

The steady-state activity of these receptor cells is reduced by an increase in carbon dioxide concentration between 0·05% and 5·0% (Fig. 6). In addition, there is a response to slight changes in carbon dioxide concentration (Fig. 8). Carbon dioxide concentration in air is about 0·03 %. House centipedes are soil animals whose habitat usually contains a carbon dioxide concentration exceeding that of normal air by 10–100 times (Aoki, 1973). Thus it appears that the temporal organ of the house centipede functions as a carbon dioxide receptor. These receptors are unusual in that they show an inhibitory response to carbon dioxide, since those in other arthropods show excitatory responses (Lacher, 1964; Kellogg, 1970; Stange & Diesendorf, 1973; Stange, 1974).

Electrophysiological responses of carbon dioxide receptors have been recorded in honey-bees (Lacher, 1964; Stange & Diesendorf, 1973; Stange, 1974), mosquitoes (Kellogg, 1970) and blowflies (Stange, 1975). Although multiple steps are proposed between carbon dioxide stimulation and spike generation in insect carbon dioxide receptors (Stange, 1974, 1975), Fig. 12 suggests that the amplitude of the response (reduction of impulse frequency) appears to depend primarily upon the concentration of CO₂ molecules at the receptor site. Similar results have been reported for fish chemoreceptors which respond specifically to CO₂ molecules (Yoshii et al. 1980).

The effects of pH upon impulse frequency (Figs 10B, 11) suggest that changes of pH around the possible receptor site may be involved in the carbon dioxide receptor process. This is unlikely, however, because changes in impulse frequency caused by changing pH are smaller than those caused by carbon dioxide either in air or in solution. Moreover, there are few differences in the steady-state activities in perfusates of pH 5, 6 and 7 (Figs 10B, 11). Several possible explanations may be proposed for these observed effects of pH. First, the activity of the receptor cell could be directly influenced by pH; this might or might not be related to carbon dioxide reception. Similar effects of pH upon chemoreceptors have been reported in insect contact chemoreceptors (Shiraishi & Morita, 1969). Second, even in perfusion experiments, a small amount of haemolymph containing HCO₃^- and CO₃^2- still remains around the receptor site, and decreasing pH could result in an increase in CO₂ molecules, which in turn could decrease the impulse frequency of the receptor cells. The latter explanation cannot account for the increasing impulse frequency above pH 8-11 (Fig. 11) or the higher impulse frequency at pH 10 than that at pH 8 (Fig. 12B). Because the concentration of CO₂ is near zero at a pH greater than 8, the observed pH effect may be intrinsic to the receptor cell. Acidic chemicals (fatty acids) and basic chemicals (amines) may somehow affect this intrinsic nature of the receptor cell. Aldehydes are also easily oxidized in solution and become acids. The delayed responses observed in the perfusion experiments (Figs 9, 10, 12B) may be related to a greater diffusion time of CO₂ molecules over the receptor site.

It has been demonstrated that carbon dioxide is the most important stimulus involved in host-finding in mosquitoes (see Gillies, 1980), and that carbon dioxide receptors in mosquitoes can detect an increase in carbon dioxide concentration from 0·04% to 0·05% (Kellogg, 1970). It is also known in several insects that the rhythmic movements of the respiratory system are regulated by the external concentration of carbon dioxide (e.g. Wigglesworth, 1935). Thus, two roles can be proposed for the temporal organ; it may serve when an animal seeks its prey, and/or it may be involved in respiratory regulation. Receptor cells of the temporal organ are sufficiently sensitive to carbon dioxide to play either role. Further studies are necessary to understand the function of the temporal organ.

This work was supported in part by a grant-in-aid for Special Project Research on Molecular Mechanism of Bioelectrical Response 59123002 from the Japanese Ministry of Education. The authors express their deepest thanks to Dr D. R. Stokes, Department of Biology, Emory University, Atlanta, GA, for his critical revision of the manuscript.