Brain Research 896 (2001) 30–35www.elsevier.com/ locate /bres

Research report

Modulation of oscillatory neural activities by cholinergic activation ofinterneurons in the olfactory center of a terrestrial slug

*Satoshi Watanabe , Tsuyoshi Inoue, Masayoshi Murakami, Yasuko Inokuma,Shigenori Kawahara, Yutaka Kirino

Laboratory of Neurobiophysics, Graduate School of Pharmaceutical Sciences, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033,Japan

Abstract

The neurons in the procerebrum (PC) of the terrestrial slug Limax marginatus show regular oscillation of their membrane potential, andthe oscillation has been implicated in olfactory processing. The neural mechanisms for the generation and modulation of the oscillationhave been poorly understood. In the present work, we examined the ionic conductances evoked by acetylcholine (ACh) in the PC neuronsand the effects of ACh application on the population activities of intrinsic and extrinsic neurons. The PC neurons are categorized intobursting neurons, which are putative local inhibitory neurons, and nonbursting neurons, which likely mediate the input and output ofinformation in the PC. Bath application of ACh augmented the local field potential oscillation in the PC. Perforated patch recording fromsingle PC neurons revealed that ACh has direct excitatory effects on bursting neurons, while it suppresses the activity of nonburstingneurons, possibly via augmented inhibitory synaptic input from bursting neurons. The correlation between the membrane potential ofbursting neurons and the frequency of oscillation suggests that bursting neurons are the main determinant of the oscillation frequency.Application of ACh also resulted in a reduction of the oscillation amplitude in the olfactory nerve, suggesting that the frequencymodulation in the oscillatory network could change the activities in the follower neurons. 2001 Elsevier Science B.V. All rightsreserved.

Theme: Sensory systems

Topic: Olfactory senses

Keywords: Slug; Olfaction; Procerebrum; Acetylcholine; Neural oscillation

1. Introduction delivered to the olfactory receptors in a manner that isdependent on the quality and value of the odorant [13,17].

The olfactory centers of a wide variety of animals show However, two major problems about the neural oscillationsynchronized oscillatory activities [1,6,19]. The procereb- have yet to be solved: one is how the oscillation isrum (PC) of the terrestrial slug Limax is comprised of generated and controlled in the network, and the other is

5 what roles the oscillation could have for informationabout 10 neurons, and these neurons also show aprocessing.synchronized oscillation of the membrane potential [7,16].

Recent researches using patch clamp recording fromThe isolated brain preparation of Limax has the advantagesingle PC neurons have reveled the existence of two typesthat it allows direct observation and manipulation of neuralof neurons in the PC: bursting and nonbursting neuronsactivities without disruption of the network and the input /[18]. Morphological [25] and electrophysiological [14,18]output pathways. The examination of the neural activitiesstudies have suggested that the bursting neurons are localin the Limax PC thus far has revealed that the oscillationinhibitory interneurons, whereas the nonbursting neuronsof the PC neurons is highly regular and observed withoutmake connections with extrinsic neurons. Synchronizedany afferent input [7], and is modulated when an odor isbursting in bursting neurons is thought to evoke inhibitorypostsynaptic potentials (IPSPs) in the nonbursting neurons,*Corresponding author. Tel.: 181-3-5841-4804; fax: 181-3-5841-and the IPSPs result in the oscillatory local field potential4805.

E-mail address: satoshi@mayqueen.f.u-tokyo.ac.jp (S. Watanabe). (LFP), which is recorded from the surface of the PC [18].

0006-8993/01/$ – see front matter 2001 Elsevier Science B.V. All rights reserved.PI I : S0006-8993( 00 )03242-X

S. Watanabe et al. / Brain Research 896 (2001) 30 –35 31

There is increasing evidence for the existence of various Kohden, MEG-2100). The field potential of the superiorneurotransmitters and receptors in the PC [8,11]. Especial- tentacle nerve (STN) was recorded from the cut end usingly, fast acting transmitters which have effects within the a suction electrode and amplified in the same way as theorder of milliseconds are thought as essential for the LFP of the PC. The signals were recorded on a DATsynaptic interactions involved in the generation of oscillat- recorder (Sony, PC204Ax).ory activities, and could be used to modulate the neural Puff application of ACh agonists on single PC neuronsactivity reversibly. Our previous work [26] focused on the was made using a microinjector (Medical Systems, PLI-actions of glutamate on the PC neurons, and suggested the 100) and a puff pipette with a tip opening of about 1 mm,possible involvement of glutamate as a transmitter in the which contained a 1 or 2 mM solution of the drug. ThePC. However, the action of glutamate on the PC neurons duration of the puff was 50–200 ms. The drugs used werewas mainly inhibitory, and the existence of an excitatory ACh (Wako Pure Chemical), nicotine (Wako Pure Chemi-neurotransmitter has been predicted to explain the repeti- cal) and muscarine (Sigma). Local application of AChtive synchronized bursting in the bursting neurons. In fact, (Fig. 5) was performed by perfusing the interior of a glassthe bursting neurons show strong excitatory input, possibly pipette (tip diameter, approx. 200 mm) placed on thefrom nonbursting neurons, which could underlie the odor surface of the PC with a saline solution containing 250 mMinput pathway [14]. Acetylcholine (ACh) is one such ACh. Under this condition, the perfused solution willputative fast acting transmitter which is also widely slowly spread over the entire PC and its vicinity, but onlydistributed in molluscan nervous systems [24]. Therefore, with a significant delay and dilution before reaching otherwe examined the effects of ACh on single PC neurons and brain regions, including the STN. In other recordings, AChtheir relationship to the network activity. was bath-applied at a lower concentration (50 mM) in

The information processed by the oscillating network order to allow homogeneous application to the entire PC.21should be transmitted to output neural systems. The For the Co saline, CaCl was replaced by CoCl .2 2

molluscan preparations are also suitable for the studies ofinteractions between the central neural systems and theperiphery because of their well-preserved structure and 3. Resultsfunction in vitro. Therefore, we have also focused on therelationship between the PC neural activity and neurons Bath-application of ACh to the cerebral ganglion aug-receiving input form the PC, and examined the effects of mented the frequency of the LFP oscillation of the PCcentral modulation on the peripheral neural activities. (Fig. 1), as has been demonstrated previously [15]. The

augmentation of frequency was accompanied by a reduc-tion of the amplitude of the LFP events. These results

2. Materials and methods indicate that the PC neurons possess ACh receptors thataugment the oscillatory activity. We then examined the

Recordings were made in an isolated cerebral ganglion cellular mechanisms of ACh-induced augmentation ofof the slug Limax marginatus. Dissection and perforated oscillation frequency with perforated patch recording inpatch recording of PC neurons were made essentially as single PC neurons.described elsewhere [26]. The PC neurons are categorized The ACh-induced responses in the PC neurons were firstinto the two major types, the bursting and nonbursting characterized by voltage-clamp analysis. Puff applicationneurons, and both of these neuron types are found through- of ACh to a bursting neuron evoked an inward current atout the PC. Therefore, we made random selection of holding potentials close to the resting membrane potentialneurons and checked the type of the neurons based on their (about 260 mV) (n510) (Fig. 2A). In four neurons, thefiring patterns. The preparation was placed in a chamber current–voltage relationship was analyzed, and the ex-that was perfused with a saline solution containing (in trapolated reversal potential of this current was 265 mVmM): NaCl 70, KCl 2, CaCl 4.9, MgCl 4.7, Hepes 5, (mean6S.E.M.) (Fig. 2B), suggesting that this current is2 2

and glucose 5 (pH 7.6). The chamber was mounted on a mediated by cation channels with a similar permeability1 1microscope (Olympus, BX50WI, equipped with a 340 for Na and K . In contrast, puff application of ACh on a

water immersion objective), and recordings were made voltage-clamped nonbursting neuron evoked an outwardusing a patch clamp amplifier (List Electronic, EPC-7). current at holding potentials close to the resting membraneThe pipette solution for perforated patch recording con- potential (n59) (Fig. 2C). In three neurons, the current–tained: K gluconate 35, KCl 35, MgCl 5, Hepes 5 (pH voltage relationship was analyzed, and the reversal po-2

7.6), and 250 mg/ml nystatin. The liquid junction potential tential of this current was 296610 mV (mean6S.E.M.)(12 mV) was subtracted offline. (Fig. 2D). These reversal potentials in the nonbursting

The LFP of the PC was recorded from the surface of the neurons were close to the calculated reversal potential for1PC using a low impedance (,1 MV) glass electrode filled K (290 mV), suggesting that the current was mainly

1with the saline solution [16]. The LFP signals were carried by K . The ACh-induced response in the bursting21amplified and band-pass filtered at 0.5–30 Hz (Nihon neurons was still present after perfusion with Co saline.

32 S. Watanabe et al. / Brain Research 896 (2001) 30 –35

Fig. 2. ACh-induced currents in PC neurons. (A) The response of aFig. 1. Effects of bath-applied ACh on the LFP oscillation of the PC. (A) voltage-clamped bursting neuron to ACh (applied as a puff at the arrow;Perfusion of ACh augmented the frequency and reduced the amplitude of duration 100 ms). The holding potential was varied from 292 to 212 mVthe LFP events. ACh was perfused during the period indicated by the bar (20-mV steps). (B) Current–voltage plot of the ACh-induced current inbelow the trace. (B) Summary of the LFP frequencies before and after the the bursting neuron in (A), measured at the time indicated by theapplication of ACh, which shows a significant increase in the LFP horizontal bar below the traces. Dashed line indicates the least-squarefrequency (**P,0.01; n58). Control and ACh indicate the averages of approximation of the plot, indicating an estimated reversal potential ofthe frequencies for the 30-s periods before and after the application of 21 mV. (C) The response of a voltage-clamped nonbursting neuron toACh. ACh (applied at the arrow; duration 200 ms). The holding potential was

varied from 292 to 232 mV (20-mV steps). (D) Current–voltage plot ofthe ACh-induced current in the nonbursting neuron in (C), measured at

In contrast, the ACh-induced response in the nonbursting the time indicated by the horizontal bar below the traces.21neurons was blocked by Co saline (data not shown).

1This suggests that the K current induced by the ACh puffin the nonbursting neurons is via synaptic input from panied by an increased burst frequency (n53) (Fig. 3A,C).bursting neurons, although it is also possible that ACh On the other hand, nonbursting neurons showed an in-

1directly evoked a K current that is dependent on extracel- crease in the frequency of inhibitory synaptic potentials21lular Ca . IPSPs accompanied by a reduction of the firing rate (n53)

Like acetylcholine, puff-applied nicotine evoked an (Fig. 3B,D). The amplitude of the IPSPs in the nonburstinginward current in all the bursting neurons tested (n54) and neurons was also reduced and the bottom levels of thean outward current in one of the three nonbursting neurons IPSPs were slightly depolarized. The reduction in the IPSPtested. In another nonbursting neuron, an inward current amplitude could be attributed to an incomplete recovery ofwas evoked, and in the other nonbursting neuron no IPSPs due to increased IPSP frequencies, as well as toresponse was observed. The variety in the response of reduced transmitter release from bursting neurons due tononbursting neurons to nicotine could be due to the the augmented firing rate.existence of both direct and synaptic effects. In contrast, The bottom level of the membrane potential in burstingmuscarine had no effects on either bursting or nonbursting neurons and the frequency of oscillation (Fig. 4) showed aneurons. Therefore, the ACh-induced responses in the PC good correlation; the correlation coefficients in threeneurons are mediated by nicotinic-like receptors. neurons were 0.93, 0.87 and 0.86. These results suggest

The relationship between the ACh-induced conductances that ACh causes a tonic depolarization of bursting neuronsand the modulation of synchronized oscillatory activity that facilitates burst generation, and hence the augmenta-was further analyzed by recording the membrane potential tion of the frequency of the synchronized oscillation. Inof the PC neurons in the current-clamp mode during contrast, since the changes induced in the nonburstingperfusion with ACh. Bursting neurons showed a sustained neurons were in the opposite direction, their contributiondepolarization of the bottom level of the membrane to the frequency of the oscillation was thought to bepotential during perfusion with ACh, which was accom- indirect.

S. Watanabe et al. / Brain Research 896 (2001) 30 –35 33

received an input from them. The major source of nervefibers making contacts with the PC is the superior tentaclenerve (STN) [7]. The STN also shows a synchronizedoscillation of the membrane potential [16] and could beregarded as containing fibers receiving input from PCneurons, or as follower neurons of the PC. Before AChapplication to the PC, the STN showed a regular oscilla-tion of its field potential composed of negative deflectionsthat were tightly synchronized with the LFP oscillation ofthe PC (Fig. 5). The amplitude of the negative potentialswas smaller when the recording electrode was further fromthe cerebral ganglion. The negative events in the STN,therefore, indicate electronic spread of outward currentsdriven by an inhibitory input at a distant position from therecording electrode, presumably in the PC. Local applica-tion of 250 mM ACh to the posterior (cell body layer)surface of the PC transiently evoked a large negativedeflection of the field potential in the STN, with anaugmented frequency and reduced amplitude both in thePC and STN, possibly due to an increased inhibitory input(Fig. 5). After cutting the STN between the recording siteand the cerebral ganglion, only very weak effects wereevoked in the STN, indicating that the effects of ACh onthe STN were via the PC and not the direct effects on thefibers in the STN (data not shown). Therefore, the changesin the activity of PC neurons by ACh also affect theactivity of the fibers in the STN, the follower neurons ofthe PC.

4. Discussion

The present work showed that: (1) ACh stimulates thebursting neurons and suppresses the nonbursting neurons;(2) ACh augments the synchronized oscillation; and (3)the augmented oscillatory activity also suppresses theactivity of the neurons receiving an input from the PCneurons.

Fig. 3. Effects of bath-applied ACh on the membrane potential of singleThe correlation between the excitation of burstingPC neurons recorded in the current-clamp mode of the perforated patch

neurons and the augmentation of the synchronized oscillat-recording. (A) In a bursting neuron, perfusion with ACh (indicated by thebar below the trace) depolarized the membrane potential and augmented ory activity of the PC suggests that the augmentation of thethe frequency of bursting. (B) In a nonbursting neuron, perfusion with oscillatory activity by ACh is caused by activation of theACh (indicated by the bar below the trace) increased the frequency of the bursting neurons. This assumption is consistent withIPSPs and suppressed firing. (C) Changes in the bottom level (indicated

previous studies that used nitric oxide [9,12] and glutamateby the arrow in the inset) of the membrane potential in three bursting[10]. The generation of the synchronized oscillation in PCneurons tested. (D) Changes in the peak level of the membrane potential

(indicated by the black arrow in the inset) in three nonbursting neurons neurons driven by bursting neurons is similar to manytested. The periods in which an action potential was generated (gray other neural systems [2,3,5,20,23].arrow in the inset) were excluded. In spite of the ubiquity of oscillatory neural networks,

the mechanisms for the output of information from oscil-The results described above suggest that ACh would latory central networks have been generally unknown,

make a novel pharmacological tool to modulate the except for a few reports in invertebrates [21]. Our presentoscillatory activity of PC neurons with known ionic results suggest a novel role for the neural oscillation that itmechanisms, i.e., the excitation of bursting neurons and could also determine the activity level of the outputinhibition of nonbursting neurons. Therefore, we then neurons. The olfactory nerves in Limax contain fiberstested using this potential tool to see how the changes in which show an oscillation in synchrony with the PCthe activity of PC neurons would affect the neurons that neurons, thus providing a model system for the study of

34 S. Watanabe et al. / Brain Research 896 (2001) 30 –35

Fig. 4. Correlation between the membrane potential of a bursting neuron and the synchronized oscillatory activity (the same neuron as in Fig. 3A). (A)Traces of the membrane potential before (black) and during (gray) the application of ACh, indicating the sustained depolarization of the bottom level of themembrane potential. (B) Plot of the instantaneous oscillation frequency, which is the inverse of the burst interval, against the bottom level of the membranepotential of the bursting neuron. Dashed line indicates the least-square approximation.

the interaction between the oscillatory center and the augmented. In both the PC and the STN, the oscillationperipheral neurons. In the present study, we showed that frequency increased while the power decreased. Assumingapplication of ACh to the PC evokes a negative potential that the positive potentials in the LFP of the PC reflectin the STN when the oscillation frequency in the PC was IPSPs in nonbursting neurons [18] and that the negative

potentials in the STN field potential also suggest IPSPs inthe STN fibers, the membrane potential changes would besimilar for the nonbursting PC neurons and the STN fibers.We have two possible explanations for the similar potentialchanges in these two types of neurons: (1) increased inputfrom the bursting neurons to the STN, i.e., commoninhibitory input to the nonbursting PC neurons and theSTN; and (2) gap junctional coupling between nonburstingneurons and the STN. Recently, the connection from theSTN to the PC neurons has been clarified electrophysio-logically [14] and the fibers in the STN have been shownto make direct connections onto nonbursting neurons butnot onto bursting neurons, although it is still necessary toexamine the connections from the PC neurons to the STNfibers. The analysis of the morphology of PC neurons [25]also revealed that nonbursting neurons, but not burstingneurons, projected to the terminal mass region of the PC,which is the only projection area of the STN fibers in thePC and a putative connection site between the STN fibersand PC neurons. Based on these results, the second schemeis preferred at present. The membrane potential oscillationin the STN may cause phase-dependent modulation ofinformation flow [14], and the augmentation of the oscil-latory activity by ACh may, therefore, reduce the in-formation flow through the STN.

Fig. 5. Effects of ACh application to the PC on the STN activity. (A)ACh is a ubiquitous neurotransmitter in both inverte-Before application of ACh, both the PC (upper trace) and the STN (lower

brates and vertebrates, and evokes a variety of responsestrace) showed synchronized oscillations of their field potentials, whichwere opposite in polarity. Application of ACh to the PC (horizontal bar) [24]. Cholinergic neurons and receptors have been alsoevoked a negative potential shift in the STN field potential, accompanied found in insect olfactory systems. The antennal lobeby a reduction in the amplitude. (B) Changes in the power of the field neurons have been suggested to be cholinergic [22], andpotentials in the PC and the STN induced by the application of ACh to

cultured mushroom body neurons also show an excitatorythe PC. The ordinate indicates the power for the 20-s period just after theresponse to ACh [4]. Although the localization of choliner-ACh application relative to that for the 20-s period just before the

application. gic neurons in the Limax CNS is still unknown, the present

S. Watanabe et al. / Brain Research 896 (2001) 30 –35 35

[10] A. Gelperin, J. Flores, Vital staining from dye-coated microprobesresults predict the existence of cholinergic neurons in theidentifies new olfactory interneurons for optical and electricalLimax cerebral ganglion. Since the bursting neurons mayrecording, J. Neurosci. Methods 72 (1997) 97–108.

receive excitatory input from nonbursting neurons [14] as [11] A. Gelperin, Oscillatory dynamics and information processing inwell as from other bursting neurons, either type of PC olfactory systems, J. Exp. Biol. 202 (1999) 1855–1864.neurons may be cholinergic. [12] A. Gelperin, J. Flores, F. Raccuia-Behling, I.R.C. Cooke, Nitric

oxide and carbon monoxide modulate oscillations of olfactoryThe mechanisms of the action of ACh characterized ininterneurons in a terrestrial mollusk, J. Neurophysiol. 83 (2000)the present study allow selective modulation of the oscil-116–127.latory activity of the PC. Together with glutamate [26] and

[13] R. Gervais, D. Kleinfeld, K.R. Delaney, A. Gelperin, Central andnitric oxide [9,12] as tools, the PC neural circuitry could reflex neuronal responses elicited by odor in a terrestrial mollusk, J.be further analyzed with regard to its input /output flow of Neurophysiol. 76 (1996) 1327–1339.

[14] T. Inoue, S. Watanabe, S. Kawahara, Y. Kirino, Phase-dependentinformation.filtering of sensory information in the oscillatory olfactory center ofa terrestrial mollusk, J. Neurophysiol. 84 (2000) 1112–1115.

[15] A. Iwama, A. Yabunaka, E. Kono, T. Kimura, S. Yoshida, T.Acknowledgements Sekiguchi, Properties of wave propagation in the oscillatory neural

network in Limax marginatus, Zool. Sci. 16 (1999) 407–416.[16] S. Kawahara, S. Toda, Y. Suzuki, S. Watanabe, Y. Kirino, Compara-We thank Dr. H. Ooya for supplying the slugs. This

tive study on neural oscillation in the procerebrum of the terrestrialwork was supported by Grants-in-Aid for Scientific Re-slugs Incilaria bilineata and Limax marginatus, J. Exp. Biol. 200search from the Ministry of Education, Science, Sports and(1997) 1851–1861.

Culture, Japan (Nos. 11771408, 12048209 and 12307053) [17] T. Kimura, S. Toda, T. Sekiguchi, Y. Kirino, Behavioral modulationand by a grant from the Program for Promotion of Basic induced by food odor aversive conditioning, and its influence on the

olfactory responses of an oscillatory brain network in the slug LimaxResearch Activities for Innovative Biosciences, Japan.marginatus, Learn. Mem. 4 (1998) 365–375.

[18] D. Kleinfeld, K.R. Delaney, M.S. Fee, J.A. Flores, D.W. Tank, A.Gelperin, Dynamics of propagating waves in the olfactory network

References of a terrestrial mollusk: an electrical and optical study, J. Neuro-physiol. 72 (1994) 1402–1419.

[1] E.D. Adrian, Olfactory reactions in the brain of the hedgehog, J. [19] G. Laurent, M. Naraghi, Odorant-induced oscillations in the mus-Physiol. 100 (1942) 459–473. hroom bodies of the locust, J. Neurosci. 14 (1994) 2993–3004.

[2] P. Andersen, J. Eccles, Inhibitory phasing of neuronal discharge, [20] K. MacLeod, G. Laurent, Distinct mechanisms for synchronizationNature 196 (1962) 645–647. and temporal patterning of odor-encoding neural assemblies, Sci-

[3] L.S. Benardo, Recruitment of GABAergic inhibition and synchroni- ence 274 (1996) 976–979.zation of inhibitory interneurons in rat neocortex, J. Neurophysiol. ¨[21] K. MacLeod, A. Backer, G. Laurent, Who reads temporal in-77 (1997) 3134–3144. formation contained across synchronized and oscillatory spike

[4] M. Cayre, S.D. Buckingham, S. Yagodin, D.B. Sattelle, Cultured trains?, Nature 395 (1998) 693–698.insect mushroom body neurons express functional receptors for [22] S. Oleskevich, Cholinergic synaptic transmission in insect mus-acetylcholine, GABA, glutamate, octopamine, and dopamine, J. hroom bodies in vitro, J. Neurophysiol. 82 (1999) 1091–1096.Neurophysiol. 81 (1999) 1–14. [23] W. Rall, G.M. Shepherd, T.S. Reese, M.W. Brightman, Dendroden-

[5] S.R. Cobb, E.H. Buhl, K. Halasy, O. Paulsen, P. Somogyi, dritic synaptic pathway for inhibition in the olfactory bulb, Exp.Synchronization of neuronal activity in hippocampus by individual Neurol. 14 (1966) 44–56.GABAergic interneurons, Nature 378 (1995) 75–78. ´[24] K.S.-Rozsa, The pharmacology of molluscan neurons, Prog. Neuro-

[6] W.J. Freeman, C.A. Skarda, Spatial EEG patterns, non-linear biol. 23 (1984) 79–150.dynamics and perception: the neo-Sherringtonian view, Brain Res. [25] S. Watanabe, S. Kawahara, Y. Kirino, Morphological characteriza-Rev. 10 (1985) 147–175. tion of the bursting and nonbursting neurones in the olfactory centre

[7] A. Gelperin, D.W. Tank, Odour-modulated collective network oscil- of the terrestrial slug Limax marginatus, J. Exp. Biol. 201 (1998)lations of olfactory interneurons in a terrestrial mollusc, Nature 345 925–930.

2(1990) 437–440. [26] S. Watanabe, S. Kawahara, Y. Kirino, Glutamate induces Cl and1[8] A. Gelperin, L.D. Rhines, J. Flores, D.W. Tank, Coherent network K currents in the olfactory interneurons of a terrestrial slug, J.

oscillations by olfactory interneurons: modulation by endogenous Comp. Physiol. A 184 (1999) 553–562.amines, J. Neurophysiol. 69 (1993) 1930–1939.

[9] A. Gelperin, Nitric oxide mediates network oscillations of olfactoryinterneurons in a terrestrial mollusc, Nature 369 (1994) 61–63.