INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Coherent and periodic oscillations of electrical activity, generated by a population of interconnected neurons, have been observed in some mammalian sensory and limbic systems (Gray and Singer 1989; Vanderwolf 1969). The oscillation of membrane potentials in the olfactory system is a ubiquitous phenomenon, widely observed in many species from vertebrates to invertebrates (Freeman 1978; Gelperin and Tank 1990; Laurent and Naraghi 1994). Recently, some physiological functions of the electrical oscillations in information processing and storage have been demonstrated. For instance, the long-term synaptic modification of the Schaffer collateral-pyramidal cell pathway in the CA1 region of the hippocampus is bidirectional and dependent on the phase of the oscillatory activity (Huerta and Lisman 1995). In the present study, we report a new physiological function of the oscillation, phase-dependent filtering of olfactory information, in the procerebrum (PC) of the terrestrial mollusk Limax marginatus. The PC is the olfactory center of Limax and consists of approximately 10⁵ interneurons that exhibit ongoing synchronous oscillatory activity (Gelperin and Tank 1990).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The slugs, L. marginatus, were anesthetized with Mg²⁺ buffer injected into the body cavity, and then the cerebral ganglion with the posterior olfactory nerve attached was isolated and prepared as previously described (Watanabe et al. 1998). The activity of single PC neurons was recorded in the current-clamp mode of the perforated patch-clamp recording configuration (List Electronic, EPC-7 or EPC-8); the composition of the pipette solution was as follows (in mM): 35.0 KCl, 35.0 K-gluconate, 5.0 MgCl₂, 5.0 HEPES (adjusted to pH 7.6 with KOH), and 250 µg/ml nystatin. The bursting (B) and nonbursting (NB) neurons were identified from their electrical activity (Watanabe et al. 1998). On the other hand, synchronous oscillatory activity in the PC was monitored by recording local field potentials (LFPs) using a glass electrode whose tip diameter was approximately 20-50 µm. The olfactory nerve was stimulated electrically with a conventional suction electrode, and the stimulus voltage and duration were 1-5 V and 1 ms, respectively.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In Limax and the related species, olfactory information is received at the olfactory receptors and conveyed to the PC neurons via the olfactory nerve (Fig. 1A; Chase 1986). Electrical stimulation of the olfactory nerve evoked transient excitatory postsynaptic potentials (EPSPs), and subsequently long-lasting inhibitory postsynaptic potentials (IPSPs) in NB neurons of the PC (Fig. 1Ba). In marked contrast, long-lasting EPSPs were evoked in B neurons by the olfactory nerve stimulation (Fig. 1Bb), and the rising phase of EPSPs in B neurons consisted of multiple components (Fig. 1C). The EPSPs evoked in the NB neurons had their onset approximately 25 ms after the stimulation of the olfactory nerve (24.8 ± 2.3 ms; mean ± SE, n = 5) and the latency was constant over repeated stimulation in each preparation (n = 4). Even if the voltage of stimulation of the olfactory nerve was increased, the input latency was constant while the amplitude of the EPSP became larger (Fig. 1Da). On the other hand, the EPSPs in the B neurons exhibited variable input latencies, which became shorter with a stronger stimulus; the shortest input-latency was about 50 ms (Fig. 1Db; n = 3). These results show that the neuronal connection from the olfactory nerve to NB neurons is monosynaptic excitatory but that to B neurons is polysynaptic excitatory. Taking into account that PC neurons are classified into either NB neurons or B neurons (Kleinfeld et al. 1994; Watanabe et al. 1998), the polysynaptic olfactory nerve-B neuron pathway is suggested to be mediated by NB neurons. Additionally, multiple EPSPs observed in B neurons suggest that a large number of presynaptic NB neurons form convergent synapses onto a single B neuron.

View larger version (14K): [in this window] [in a new window]   Fig. 1. A membrane potential in bursting (B) neurons and nonbursting (NB) neurons evoked by electrical stimulation of the posterior olfactory nerve (ON). A: a schematic drawing of the olfactory system of Limax, based on the previous morphological observations. The PC is clearly divided into the somatic region (SR) and neuropil region (NR). The PC neurons consist of B neurons (filled circle) and NB neurons (empty circles) and only NB neurons extend their neurites to the NR. B: representative responses of NB neurons (Ba) and B neurons (Bb). Dashed lines indicate the membrane potentials just before the ON stimulus is applied. Arrowheads indicate the stimulus artifact. The horizontal scale bar is 200 ms, and the vertical bar is 20 mV for NB neuron and 12.5 mV for B neuron. C: expansion of the rising phase of the evoked excitatory postsynaptic potential (EPSPs) in a B neuron. The time range indicated by the horizontal bar at the bottom of Fig. 1Bb is expanded and the action potential was truncated. Arrows indicate the onset of individual EPSP. Scale bars are 25 ms and 5 mV. D: EPSPs in an NB neuron (Da) and a B neuron (Db) evoked by an ON stimulation. In both neurons, the intensity of the ON stimulus was varied from 1.0 to 3.0 V at the phase range of 1.0-1.4. Since the phase-dependency of the transmissions is not observed in this phase range (see Fig. 2C), we can examine the effects of the stimulus intensity. In these experiments, hyperpolarizing DC currents were injected to keep the membrane potential at about 70 mV to prevent discharges. The horizontal scale bar is 50 ms, and the vertical bar is 3 mV for NB neurons and 8 mV for B neurons.

Polysynaptic transmission from the olfactory nerve to B neurons was further characterized. Electrical activities of a B neuron and an NB neuron were recorded simultaneously (Fig. 2A). The neural activities of both types of neurons were spontaneously oscillating and phase-locked to the LFP of the PC and to each other (Kleinfeld et al. 1994; see Fig. 2A). When the phase of the periodic activities of the PC is defined as shown at the bottom of Fig. 2A, the input latency of evoked EPSPs in B neurons was longer when the stimulus was applied at an earlier phase, i.e., just after the peak of depolarization (Fig. 2B). In Fig. 2C, input latencies and averaged membrane potentials of NB neurons were both indicated as a function of the phase of the PC oscillation. Recovery of the IPSP in NB neurons and input latencies in B neurons were resolved into fast and slow components. The phase at the boundary between the two components was approximately . The input latencies of B neurons were well correlated with the membrane potentials of NB neurons (Fig. 2D), but not with those of B neurons (data not shown). These results indicate that the response of B neurons to a stimulus depends on the phase of the oscillatory activity when the stimulus was applied, and the phase-dependency of latency of the evoked response in B neurons may arise from the phasic oscillation of the intervening NB neurons. Thus, electrical oscillation of NB neurons operates as a neuronal filter in the polysynaptic pathway from the olfactory nerve to B neurons.

View larger version (29K): [in this window] [in a new window]   Fig. 2. Relationship between the membrane potentials of NB neurons and input latencies of B neurons in response to ON stimulation. A: simultaneous recording from an NB neuron (upper) and an adjacent B neuron (lower). The phase of oscillation was defined based on the interval between adjacent spikes of B neurons, as shown at the bottom. Scale bars are 1 s and 10 mV. B: phase-dependence of input latencies of the EPSPs in B neurons. The arrowhead indicates the stimulus artifact and the arrows indicate the onset of the first EPSP. Input latency is defined as the time between the arrowhead and the arrow. Input latency (68 ms) evoked by the stimulus at phase  was shorter than that (152 ms) evoked at phase 0.5. Scale bars are 150 ms and 10 mV. C: input latencies of the first EPSPs in a B neuron (empty marks) and membrane potentials of averaged NB neuron (solid thick lines) as a function of the phase when the stimulus was applied. Since the phase is determined on the basis of the spike intervals in B neurons, the averaged membrane potential of NB neuron as a function of the phase was estimated from the data of paired B and NB neuron recordings (n = 3), while the latencies of the first EPSPs in the B neuron were estimated from recordings of B neurons alone (n = 5). In Fig. 2, Ca and Cb, representative data of the input latency in a B neuron and the individual data of the normalized input latencies in all preparations (indicated by the 5 different marks) are indicated, respectively. In Fig. 2Cb, normalized input latency in B neurons were calculated from the following equation: normalized IL = (IL  ILmin)/(ILmax  ILmin), where IL is input latency, ILmin or ILmax are minimum or maximum input latency in each preparation, respectively. The input latencies at phase 0.4-1.0 and phase 1.0-1.6 were fitted using the least-squares approximation, which are indicated by the 2 thin straight lines. D: correlation between the input latencies of B neurons and the membrane potential of NB neurons. Figure 2, Da and Db, are derived from Fig. 2, Ca and Cb, respectively. In the inset of Fig. 2Db, the correlation coefficient in each preparation is indicated.

Then, what will be the consequence of the phase-dependent response evoked in B neurons? The B neurons are an essential factor for frequency modulation of the ongoing synchronous oscillation (Gelperin 1994; S. Watanabe, T. Inoue, M. Murakami, Y. Inokuma, S. Kawahara, and Y. Kirino, unpublished observations). Frequency modification is known to be evoked also by odor application to the olfactory receptors (Gelperin and Tank 1990). When a single electrical stimulus was applied to the olfactory nerve, it induced modulation in the oscillation frequency of the PC depending on the phase of the PC oscillation (Fig. 3). The PC oscillation frequency was more strongly modulated when the stimulus was applied at a later phase, while stimulation of the olfactory nerve affected only slightly the oscillation frequency when the stimulus was given at an earlier phase (Fig. 3). These results clearly indicate that a modulation of the frequency of the PC oscillation is also phase-dependent and neuronal spikes in the olfactory nerve have different effects on the PC oscillation frequency, depending on the phase at their arrival. In other words, only selected spikes can modulate the oscillation frequency.

View larger version (30K): [in this window] [in a new window]   Fig. 3. Phase-dependent modulation of the LFP oscillation frequency in the PC. A: representative data of frequency modulation of LFP oscillation in the PC by a single stimulus to the ON. Arrowhead indicates the stimulus artifact. The olfactory nerves were stimulated at a later phase in lower traces of the LFP. Scale bar is 2 s. B: summary of the data (mean ± SE, n = 4). The phase was determined on the basis of the onset of local field potential (LFP) events, which approximately coincided with the spike in B neurons. The vertical axis indicates the normalized frequency change, (fpost  fpre)/fpre, where fpre refers to instantaneous LFP frequency, which is the inverse of the interval between the two LFP events, just before the ON stimulus, and fpost is average of instantaneous LFP frequency for 3 cycles just after the ON stimulus.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this paper, we have addressed mainly two points: 1) the synaptic connection from the olfactory nerve to the PC neurons, and 2) the neuronal mechanism of phase-dependent synaptic transmission of sensory information, which results in phase-dependent modulation of the PC oscillatory activity. This neuronal mechanism suggests that the PC oscillation functions as a phase-dependent filter of sensory information. This paper is the first report that described phase-dependent transmission, elucidated its neuronal mechanism, and proposed the physiological function as a neuronal filter of electrical oscillation in the olfactory system, although phase-dependent modification of PC activity has been previously reported (Gelperin and Tank 1990).

Figure 1 shows that olfactory information conveyed from the olfactory nerve first excites NB neurons in the PC, and subsequently B neurons. Taking into account the previously suggested inhibitory connection from B neurons to NB neurons (Kleinfeld et al. 1994; Watanabe et al., unpublished observations), the PC can be viewed as a feedback inhibitory circuit; activity levels of NB neurons evoked by olfactory input depend both on the direct depolarization and on the recurrent hyperpolarization via B neurons. This synaptic circuit is similar to that in the olfactory bulb, the vertebrate olfactory center, where olfactory information is received by mitral/tufted cells, which form an inhibitory feedback loop via the granule cells (Isaacson and Strowbridge 1998; Jahr and Nicoll 1980). In addition to the similarity in the morphology of the neurons (Scott 1986; Watanabe et al. 1998) and the existence of synchronous oscillations (Freeman 1978; Gelperin and Tank 1990), the synaptic connections revealed in the present study indicate a novel type of conserved function between the vertebrates and invertebrates.

As shown in Fig. 2, B and C, the efficacy of synaptic transmission from the olfactory nerve to B neurons was modified phase-dependently, presumably because of the phasic oscillation of the intervening NB neurons. Namely, the fact that a 500-ms difference in stimulus timing (i.e., approximately 0.5 phase difference) leads to a 110-ms difference in EPSP latency (Fig. 2B) is presumably due to the depth of the hyperpolarizing IPSP in the NB neurons and the steep nonlinear depolarizing ramp during recovery from the IPSP. Additionally, more hyperpolarizing membrane potentials of the NB neurons not only would produce a longer interval between the onset of the EPSP and the spike generation, but also might inhibit the EPSP from reaching the threshold in the extreme case. On the other hand, more depolarizing membrane potentials would elicit spike generation more easily and the interval is shorter. Since the excitability in B neurons determines oscillation frequency in the PC (Gelperin 1994; Watanabe et al., unpublished observations), the phase-dependency of evoked responses in B neurons in turn results in the phase-dependent modulation of synchronous oscillatory activity in the PC by the olfactory nerve stimulus (Fig. 3).

Then, how does this phase-dependent filtering affect the physiological operation of olfactory information processing? In Limax and the related species, tentacle ganglion (TG) is located upstream of the PC in information flow (Chase 1986) and some of the neurons in the TG project their axons to the olfactory nerve (Chase and Kamil 1983). In recent studies, synchronous oscillation in the TG and the olfactory nerve, which are noncoherent with spontaneous oscillation in the PC, have been observed and spiking activities in the olfactory nerve were shown to be phase-locked with the oscillation in the TG (Kimura et al., personal communication). Thus, spiking activities in the olfactory nerve are noncoherent with the spontaneous oscillation in the PC, and filtering effect of olfactory information in the PC might be determined by the phase-relationship between the two synchronous oscillatory networks: the TG and the PC. In other words, phase-dependent filtering system in the PC might transmit olfactory information to the further information processing unit in the PC only when the TG and PC oscillations are in a certain phase relationship. Odor-induced dynamic interactions between the primary and secondary olfactory processing regions will be a key mechanism for understanding physiological function of the newly proposed mechanism in olfactory information processing.