Muscarinic control of graded persistent activity in lateral amygdala neurons

Muscarinic control of graded persistent activity in lateralamygdala neurons

Alexei V. Egorov,1,2 Klaus Unsicker1 and Oliver von Bohlen und Halbach1

1Interdisciplinary Center for Neurosciences (IZN), Department of Neuroanatomy, University of Heidelberg, Im Neuenheimer Feld307, D-69120 Heidelberg, Germany2Institute of Higher Nervous Activity and Neurophysiology, Russian Academy of Science, Moscow, Russia

Keywords: intrinsic neuronal excitability, mouse, persistent activity, slice, nucleus basalis of Meynert

Abstract

The cholinergic system is crucially involved in several cognitive processes including attention, learning and memory. Muscarinic actionshave profound effects on the intrinsic firing pattern of neurons. In principal neurons of the entorhinal cortex (EC), muscarinic receptorsactivate an intrinsic cation current that causes multiple self-sustained spiking activity, which represents a potential mechanism fortransiently sustaining information about novel items. The amygdala appears to be important for experience-dependent learning byemotional arousal, and cholinergic muscarinic influences are essential for the amygdala-mediated modulation of memory. Here weshow that principal neurons from the lateral nucleus of the amygdala (LA) can generate intrinsic graded persistent activity that is similarto EC layer V cells. This firing behavior is linked to muscarinic activation of a calcium-sensitive non-specific cation current and can bemimicked by stimulation of cholinergic afferents that originate from the nucleus basalis of Meynert (n. M). Moreover, we demonstratethat the projections from the n. M. are essential and sufficient for the control and modulation of graded firing activity in LA neurons. Wefound that activation of these cholinergic afferents (i) is required to maintain and to increase firing rates in a graded manner, and (ii) issufficient for the graded increases of stable discharge rates even without an associated up-regulation of Ca2+. The induction ofpersistent activity was blocked by flufenamic acid or 2-APB and remained intact after Ca2+-store depletion with thapsigargin. Theinternal ability of LA neurons to generate graded persistent activity could be essential for amygdala-mediated memory operations.

Introduction

The medial temporal lobe (MTL) contains a system of anatomicallyrelated structures, including the hippocampus, the entorhinal cortexand the amygdala, that are essential for learning and memory (Squire& Zola-Morgan, 1991; Squire et al., 2004). Persistent firing activity isthought to be a neural substrate for holding memories over short timedelays (Major & Tank, 2004). Two general mechanisms of persistentactivity have been hypothesized: recurrent networks and intrinsicbiophysical cellular properties. Principal neurons from layer V of theentorhinal cortex (EC LV) can generate oscillatory persistent activitythat can maintain multiple levels of stable firing rates in the absence ofsynaptic reverberations (Egorov et al., 2002a). This firing behavior islinked to cholinergic muscarinic receptor activation, and relies onactivity-dependent changes of a Ca2+-sensitive non-specific cationiccurrent. Thus, intrinsic sustained spiking activity that does not dependon previous strengthening of synapses may represent a potentialmechanism for sustaining information about novel items in a short-term memory buffer by the MTL.

The amygdala appears to be important for emotional and experi-ence-dependent learning (Cahill & McGaugh, 1998; Pare, 2003;McGaugh, 2004). Single neurons which show a selective increase infiring rate for novel or familiar stimuli have been recently identified inthe human amygdala (Rutishauser et al., 2006) and sustained spikingactivity has been reported in the primate amygdala by encoding the

positive and negative value of visual stimuli during learning (Patonet al., 2006). Emotionally salient experiences tend to be wellremembered. According to the modulation hypothesis (McGaughet al., 1996, 2002; McGaugh, 2000), the positive effect of emotion onmemory is due to modulatory influences of the basolateral complex ofthe amygdala (BLA) on encoding and consolidation processesoccurring in MTL memory structures. Cholinergic muscarinic influ-ences are essential for the BLA-mediated memory modulation (Poweret al., 2003). Amygdalar cholinergic activation is involved in bothacquisition and consolidation of memory (Power & McGaugh, 2002)and also influences working memory (Beninger et al., 1994; Malletet al., 1995; Barros et al., 2002). The BLA consists of the lateral (LA),basolateral and basomedial nuclei of the amygdala. Here we show thatactivation of muscarinic receptors of LA principal neurons inducesintrinsic oscillatory persistent activity similar to EC LV neurons.Because the major cholinergic input into the BLA is derived fromafferents that originate from the nucleus basalis of Meynert (n. M)(Emson et al., 1979; Carlsen et al., 1985), which is critically involvedin BLA-mediated memory operations (Mallet et al., 1995; Power &McGaugh, 2002), we investigated whether this projection plays a rolein the induction and modulation of persistent activity of neurons,located in the LA.

Materials and methods

Preparation of brain slices

Brain slices were prepared from adult ether-anesthetized (C57 ⁄ BL6)mice using standard procedures (Egorov et al., 2002a). Briefly, animals

Correspondence: Dr A. V. Egorov, 1Interdisciplinary Center for Neurosciences (IZN), asabove.E-mail: [email protected]

Received 1 August 2006, revised 12 September 2006, accepted 27 September 2006

European Journal of Neuroscience, Vol. 24, pp. 3183–3194, 2006 doi:10.1111/j.1460-9568.2006.05200.x

ª The Authors (2006). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing Ltd

were decapitated, the brain was then rapidly removed, and placed in acold (4–6 �C) oxygenated Ringer solution containing (in mm): 124NaCl, 3 KCl, 1.6 CaCl2, 1.8 MgSO4, 26 NaHCO3, 1.25 NaH2PO4 and10 glucose (pH maintained at 7.4 by saturation with 95%O2 ⁄ 5%CO2).Combined horizontal slices 400 lm thick containing the amygdala, thehippocampus, and entorhinal and perirhinal cortices were cut using amanual vibratome slicer (Campden Instruments, Loughborough, UK).Slices were then incubated in an interface holding chamber at roomtemperature for at least 1 h before use. Individual slices weretransferred to the interface recording chamber (Fine Science Tools,Heidelberg, Germany) one by one, superfused with Ringer solution at arate of 1–2 mL ⁄ min, and maintained at 36 ± 1 �C. The LA wasidentified with a dissecting microscope by transillumination.

Recording procedures

Intracellular recordings were obtained using sharp microelectrodespulled on a Brown-Flaming puller P-87 (Sutter Instruments, Novato,CA, USA) from 1.5-mm borosilicate glass (Sutter Instruments).Electrodes were backfilled with 2 m K+-acetate (tip resistance of 60–90 MW). Potential changes were amplified using an Axoclamp-2Bamplifier (Axon Instruments, Union City, CA, USA), filtered at 3 kHzand digitized at 10 kHz by an analog-to-digital converter 1401MICRO (Cambridge Electronic Design, Cambridge, UK). Signalswere stored on a computer, and visualized using Spike 2 laboratorysoftware (Cambridge Electronic Design). Intracellular potential wasrecorded in bridge mode and the bridge balance was monitoredthroughout the experiment. Resting membrane potential was estimatedby subtraction of the tip potential following withdrawal from the cell.Input resistant was determined by passing of current pulses ()0.2 to)0.4 nA, 200 ms) through the recording electrode and measuring theresultant voltage deflections (measurements toward the end of thecurrent pulse at steady-state level). Durations of positive and negativecurrent injections (range 0.1–1 nA) were controlled using a Master-8VP stimulator (AMPI, Jerusalem, Israel). Membrane potential wasmanually adjusted by intracellular injection of DC current through therecording electrode and held near threshold for firing (approximately)60 mV) when eliciting persistent activity. An extracellular bipolarelectrode (SNEX-200; Rhodes Medical Instruments, Woodland Hills,CA, USA) was used to induce synaptic responses by local stimulationof the external capsule, the substantia innominata (SI) or within theBLA. Trains of pulses (0.1 ms, 5–20 V) were delivered at 10–100 Hzfor 0.1–10 s with an Iso-Flex stimulus isolator (AMPI) triggered by aMaster-8 VP pulse generator. Upper parts of stimulation artifacts wereeliminated in the represented traces.

Chemicals

Carbachol (CCh, 10 lm) and atropine (1 lm, both from Sigma-Aldrich, Taufkirchen, Germany) were bath-applied by continuousperfusion and were both effective even during the first minutes ofapplication. As the muscarinic effects [i.e. CCh-induced depolariza-tion (Washburn & Moises, 1992b) and persistent firing activity] didnot desensitize, the neurons were in many cases directly impaled in thepresence of CCh. Some recordings were performed during fractionalor full blockade of ionotropic glutamate- and GABA-mediatedneurotransmission with a drug mixture consisting of either a cocktailof kynurenic acid (2 mm) and picrotoxin (100 lm) (n ¼ 14), or amixture of dl-2-amino-5-phosphonovaleric acid (APV, 30 lm),6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 10 lm), picrotoxin(100 lm, all from Sigma-Aldrich) and CGP 55845 (10 lm) (n ¼ 9).

CGP 55845, thapsigargin, 2-aminoethoxydiphenylborate (2-APB) (allfrom Tocris Cookson, Bristol, UK) and flufenamic acid (Sigma-Aldrich) were applied from stock solutions made in DMSO. The finalconcentration of DMSO in Ringer solution was £ 0.1%. Controlexperiments revealed no measurable effects of DMSO on cellularproperties or cholinergic modulatory actions (n ¼ 5). Eserine (1 lm;Sigma-Aldrich), sMCPG (500 lm; Tocris Cookson), thapsigargin(2–3 lm), flufenamic acid (10 lm), 2-APB (100 lm) and bariumchloride (1 mm) were applied by continuous perfusion. When Ba2+

was introduced, MgCl2 (2 mm) was substituted for MgSO4 in theRinger solution to avoid precipitation. The Ca2+-free solutioncontained 6 mm Mg2+.

Data analysis

Electrophysiological data were analysed using Spike 2 laboratorysoftware. Because muscarinic-dependent plateau potentials have beenobserved equally in regular-spiking and late-firing neurons of LA(Washburn & Moises, 1992a), data were summarized from bothneuronal populations. Averaged data are given as mean ± SD.Calculation of the statistical significance was performed usingunpaired, two-tailed Student’s t-test. Spectral (Fourier) analysis andperistimulus histograms were made using Spike 2. The amplitude ofthe terminated plateau potential and of the slow after-depolarizationwas estimated together as a function of the membrane potential. Thedifference in firing frequency between two neighbor levels (frequencygrade) was calculated as the difference in the reference frequency rateof about 6.5 Hz and a neighbor level elicited after SI stimulation goingwith a subsequent current-step or after a simple SI stimulation. Thereference frequency was 6.7 ± 0.8 Hz (n ¼ 6) for SI stimulation(average duration 2.6 ± 0.9 s) with a current-step of 2 ± 1.2 s;6.4 ± 1.8 Hz (n ¼ 7) for simple SI stimulation (2.57 ± 1 s) incontrols; and 6.2 ± 0.9 Hz (n ¼ 3) for simple SI stimulation (2 s) inthe presence of thapsigargin.

Results

Persistent activity in LA neurons

We investigated the possibility that oscillatory graded persistentactivity occurs in amygdala by using sharp microelectrode recordingsfrom LA principal neurons in acute horizontal slices. Principalneurons were identified electrophysiologically on the basis of theirfiring characteristics as previously described (Washburn & Moises,1992a). We found that under activation of cholinergic receptors (bathapplication of CCh, 10 lm) LA neurons (49 ⁄ 62, 79%) could respondto a short suprathreshold current-step stimulus (2–4 s current-step-driven spike train of 15–40 Hz) with delayed firing at a constantfrequency for an apparently indefinite period of time (tested up to5 min; Fig. 1A). This persistent plateau potential relied on theactivation of muscarinic receptors as its induction was completelyblocked by atropine (n ¼ 4; Fig. 1A). Analogous to EC LV cells, themuscarinic-dependent plateau potential of LA neurons was not due tolocal circuit reverberation mechanisms, as the plateau activity could beinduced equally well during intact neurotransmission (n ¼ 28) andafter pharmacological blockade of ionotropic glutamate- and GABA-receptors (n ¼ 21; Fig. 1B).In Ringer solution, the resting membrane potential (RMP) and input

resistance of the neurons was )72.4 ± 2 mV (n ¼ 7) and54 ± 12 MW (n ¼ 7), respectively. In the presence of CCh RMPwas )67 ± 3.8 mV (n ¼ 23) during intact neurotransmission, and)64.6 ± 4.7 mV during AMPA-, NMDA- and GABA-mediated

3184 A. V. Egorov et al.

ª The Authors (2006). Journal Compilation ª Federation of European Neuroscience Societies and Blackwell Publishing LtdEuropean Journal of Neuroscience, 24, 3183–3194

neurotransmission block [n ¼ 20, data were summarized from bothcocktails: kynurenic acid and picrotoxin (RMP ¼ )64.6 ± 4.5 mV,n ¼ 8); APV, CNQX, picrotoxin and CGP 55845(RMP ¼ )64.7 ± 5 mV, n ¼ 12)]. The average input resistance ofthe neurons was 62 ± 18 MW (n ¼ 23) in the presence of CCh duringintact neurotransmission and 68 ± 17 MW (n ¼ 14) during blockadeof neurotransmission.

Similar to EC LV cells, the plateau-potential of LA neurons thatsustained persistent firing displayed very pronounced activity- andvoltage-dependence (Fig. 1B). The persistent firing could be elicitedby stimulation from voltage levels of about 10 mVor less negative tospike threshold and was never induced by stimulation from voltagelevels below � )70 mV. Furthermore, activation of the cells from thesame voltage level by weak stimulation (current pulses in the range of300 ms to 2 s; spike train of about 10 Hz) usually triggered only after-depolarization or terminated plateau potential. However, increasingthe duration ⁄ intensity of the trigger pulse always led to an increase inthe duration of the plateau potential, which in most cases (49 ⁄ 55neurons, 89%) could then provide a rise to a stable state of sustainedspiking (Fig. 1B). In general, a 4-s spike train at 15–40 Hz evokedfrom a resting level of � 10 mV or less from spike threshold almostinvariably elicited persistent activity (37 ⁄ 42 tested neurons). How-ever, stimulus durations in the range of 1–2 s could also be effective intriggering persistent firing (27 ⁄ 44 tested neurons). Importantly, evensmall changes in membrane potential (i.e. ± 1 mV) increased ordecreased the magnitude of the plateau potentials induced bydepolarizing current pulses of equivalent strength. Plateau potential

amplitude decreased sharply with membrane hyperpolarization (seeSupplementary material, Fig. S1). Once persistent firing was initiatedwe found that it could be turned off by prolonged membranehyperpolarizations and re-elicited again. The larger the amplitude ofthe hyperpolarization, the shorter the time required to turn offthe persistent active state. Thus, at least 8–12 s of hyperpolarizationsto � )80 mV were required to stop persistent firing fully.

Graded persistent activity in LA neurons

We also tested whether a sustained level of firing frequency in LAneurons could be either increased or decreased in a step-like manner,analogous to EC neurons (Egorov et al., 2002a). As illustrated inFig. 2A (see also Supplementary material, Fig. S2), we found thatrepetitive application of a given activating current-step stimulus couldgive rise to a series of well-defined graded increases of stabledischarge rates up to a ceiling where no further enhancement in firingrate was observed (24 ⁄ 27 neurons tested). The average maximumpersistent firing frequency induced in this manner was 10.9 ± 3.3 Hz(n ¼ 18). These increases consisted of 3–6 levels by repetitivestimulation with a 4-s depolarizing step. However, the number ofgraded levels critically depended on the stimulation parameters. Adecrease in the duration ⁄ intensity of the given pulse always led toincreases in the number of sustained levels of firing frequency up to itsmaximal rate (e.g. one sustained level obtained by using a 4-s pulsecould also be reached by 3 ± 1 consecutive 1-s stimulations at thesame intensity). In fact, an arbitrary number of stable firing rates couldbe elicited in this fashion. In addition, repetitive application ofhyperpolarizing current pulse steps had an opposite effect, i.e. led tograded decreases in firing rate (9 ⁄ 14 neurons tested, Fig. 2B).However, durations of at least 6–8 s were required for a gradeddecrease in persistent firing rate. Identical to EC LV cells, the states ofstable firing frequency in LA neurons were not affected by relativelyshort de- or hyperpolarizing current-step stimuli. Typically, current-step-driven spike trains of insufficient strength to trigger persistentfiring were also unable to elicit graded changes in firing frequency.Similarly, membrane hyperpolarizations of about 20 mV and shorterthan 2 s were always ineffective in causing graded decreases in firingfrequency.

Induction and modulation of persistent activity by synapticstimulation

To examine whether persistent activity could also be induced byactivating the LA cells synaptically, we tested neuronal responses torepetitive synaptic stimulation during intact neurotransmission. Asillustrated in Fig. 3A, plateau potentials that sustained stable firingcould either be induced by local synaptic activation at 40 Hz for 1–2 sor by step depolarizations (5 ⁄ 8 neurons). However, repetitive synapticstimulation in voltage level near spike threshold often induced a strongGABA-mediated hyperpolarization during the train; this hyperpolar-ization prevented spike trains from being initiated and accordinglyprevented the formation of persistent plateau potential (7 ⁄ 15 testedneurons). During a selective pharmacological block of GABA(A)-receptors with picrotoxin (50–100 lm), neurons responded to asynaptic stimulus with epileptiform events (n ¼ 4).To examine the effects of synaptic stimulation on graded changes in

firing frequency, the synaptic trains were applied during plateaupotential elicited by current step depolarization. During intactneurotransmission a repetitive synaptic stimulation was capable ofproducing a stable increase (n ¼ 5), or in some cases even a decrease

Fig. 1. Muscarinic-dependent persistent activity in LA neurons. (A) CCh-induced persistent firing and its block by the muscarinic antagonist atropine(1 lm). The arrow below the right current trace indicates a DC shift.(B) Activity- and voltage-dependence of persistent firing. Responses tocurrent steps of different duration (left). A depolarizing current pulse ofequivalent strength was not able to elicit persistent firing after membranehyperpolarization (right). The recording in B was obtained during blockade ofneurotransmission via CNQX, APV, picrotoxin and CGP 55845. Initial Vm in Aand B: 52 and )58 mV, respectively.

Graded persistent activity in LA neurons 3185


(n ¼ 4) in firing frequency (data not shown). The activity-dependentcharacteristics of the plateau potentials clearly suggest that Ca2+-influxis an important element (see below). Therefore, the variety in firingfrequency changes might be the result of divergent intracellular [Ca2+]regulation during synaptic stimulation. In order to eliminate a possiblecontribution of NMDA-mediated Ca2+-influx as well as a prolongedGABA(B)-mediated hyperpolarization, we tested neurons duringpartial neurotransmission block with APV (30 lm) and CGP 55845(10 lm). During fractional NMDA- and GABA(B)-blockade RMP ofthe neurons was )67 ± 5.8 mV (n ¼ 5) and input resistance was67 ± 14 MW (n ¼ 4) in the presence of CCh. As illustrated in Fig. 3Band C, synaptic train stimulation, applied during plateau potentialunder these conditions, induced a stable graded decrease in firingfrequency in all cases tested (n ¼ 4). Importantly, the number ofgraded levels was decreased when the train duration was increased(tested from 100 ms to 1 s). We also found that the persistent activitycan be completely prevented by prolonged synaptic stimulation

(subthreshold, 8–12 s, 10 Hz) given immediately after, but not before,the depolarizing current step (n ¼ 4; Supplementary Fig. S3).

Induction of persistent activity in LA neurons by stimulation ofcholinergic fibers

An important goal of this study was to induce persistent activitythrough synaptic release of acetylcholine (ACh). The main cholinergicinput to cells in the basolateral complex of the amygdala is derivedfrom afferents that originate from neurons located in the n. M andpartially in the SI (Emson et al., 1979; Carlsen et al., 1985). The effectof electrical stimulation of the cholinergic pathway on BLA pyramidalneurons [i.e. blockade of slow after-hyperpolarization (sAHP),reduction of spike frequency accommodation and prolonged depolar-ization that follows a train of action potentials] was shown byWashburn & Moises (1992b). Indeed, as illustrated in Fig. 4A, wefound that a repetitive stimulation of the SI (10 Hz, 1–10 s, average

Fig. 2. Graded persistent activity in LA neurons. (A) Repetitive stimulation with a 4-s depolarizing step gives rise to five distinct increases of stable discharge rate(CCh, 10 lm). (B) Repetitive application of 6-s hyperpolarizing steps gives rise to discrete decreases of firing rate up to its termination (recording performed in thepresense of CCh, kynurenic acid and picrotoxin). The lower diagrams correspond to the peristimulus histograms (bin width in A and B: 700 and 600 ms,respectively). Initial Vm in A: )52 mV. Final Vm in B: )64 mV.



Fig. 3. Synaptic induction and modulation of persistent activity in LA neurons. (A) Synaptric trains (40 Hz, 2 s and 4 s) induced sustained firing in a gradedfashion, which is then stopped by hyperpolarization and re-initiated by depolarizing current steps (right). (B and C) Repetitive synaptic stimulation (40 Hz; 100 msin B and 500 ms in C; same neuron), applied during plateau potential induced by current step depolarization (in CCh), elicited a graded decrease in firing frequency.The control plateau potentials are on the right. Note that the number of graded levels in the same neuron depends on stimulus duration. The diagrams below thecurrent traces correspond to the peristimulus histograms (bin width of 500 ms). Arrows correspond to the time-points of repetitive synaptic stimulation. Drugs wereadded to the Ringer solution as indicated above. Initial Vm in A: )54 mV; in B and C: )53 mV.



4.6 ± 2.7 s) shortly before a suprathreshold current-step was applied,is sufficient to elicit a prolonged (for 1–3 min) sustained firing activityin LA neurons (18 ⁄ 25 tested). Recordings were obtained duringblockade of glutamatergic- and GABAergic-neurotransmission withdrug cocktails of CNQX, APV, picrotoxin, CGP 55845 and in thepresence of the acetylcholinesterase inhibitor eserine (1 lm). Eserineitself had no effect on resting membrane potential ()71.4 ± 4 mV,n ¼ 25 in eserine vs. )72.5 ± 2 mV, n ¼ 12 in control, P ¼ 0.37;input resistance of 55.3 ± 12 MW, n ¼ 17 in eserine vs.53.6 ± 12 MW, n ¼ 7 in control, P ¼ 0.75). Prolonged firing activityobserved under this experimental configuration demonstrated activity-

and voltage-dependence similar to that seen in the experimentsperformed in the presence of CCh. Simple repetitive stimulation ofcholinergic afferents at the subthreshold membrane potential levelswithout spikes was insufficient to elicit persistent firing (n ¼ 6;Fig. 4B left). However, SI stimulation near spike threshold, accom-panied by action potentials, triggered a prolonged firing activity inneurons (n ¼ 8; Fig. 4B right). The long-lasting firing activity,induced by using the SI stimulation modes, was blocked completelyby atropine (1 lm, n ¼ 10; tested with SI stimulation in a range from10 to 100 Hz; Fig. 4C), indicating selective involvement of muscariniccholinergic receptors.

Fig. 4. Induction of prolonged firing activity in LA neurons by stimulation of cholinergic fibers. (A) Responses of an LA neuron to depolarizing current-stepbefore and after a short repetitive stimulation of the substantia innominata (SI, 10 Hz, 4 s). (B) Repetitive stimulation of SI (10 Hz, 10 s) at different membranepotential levels. SI stimulation near spike threshold voltage level, accompanied by action potentials, elicited a prolonged firing activity in the neuron. Insert:schematic representation of the position of the electrodes as well as the location of the nucleus basalis of Meynert. (C) The prolonged firing activity, induced by adepolarizing current-step preceded by SI stimulation, was blocked completely by atropine (1 lm, tested with SI stimulation in a range from 10 to 100 Hz), indicatinga selective involvement of muscarinic cholinergic receptors. The arrow below the current trace signals indicates a DC shift. Drugs were added to the Ringer solutionfor A, B and C as indicated above. Initial Vm in A, B and C: )58, )61 and )57 mV, respectively.



Cholinergic control of graded firing activity in LA neuronsAs shown above, a brief stimulation of cholinergic fibers combinedwith a train of action potentials is sufficient to produce an extendedplateau potential in LA neurons. However, we found that this effect isalways self-terminating and was followed by a long-lasting decrease infiring frequency (i.e. a result different from the persistent plateaupotential observed in the case of continuous activation of muscarinicreceptors by bath-applied CCh). In fact, this might be a consequenceof the dependence of the active ACh concentration on the dur-ation ⁄ strength of fiber stimulation as well as on the diffusion of thesubstance in the slice. Note that the extracellular ACh level decreasedslowly after fiber stimulation in the presense of eserine. This suggestedthat muscarinic activation is necessary not only for the induction ofplateau potentials but also for maintaining a sustained level of firingfrequency and for creating its graded changes. Indeed, we found that a

depolarizing current pulse given 40–60 s after induction of the plateaupotential (SI stimulation at 10 Hz for 2–10 s) was not able to increasethe original firing frequency (n ¼ 13; Fig. 5A, indicated by the openarrowhead). Instead, a long-lasting decrease of firing frequency wasobserved. However, a brief SI stimulation (10 Hz, 1–4 s) given shortlybefore the depolarizing current steps were applied was sufficient forthe graded increases in discharge rates (n ¼ 10; Fig. 5A). Moreover,trains stimulating cholinergic afferents (10 Hz, 1–4 s) were sufficientto produce the graded increases of discharge rates without subsequentdepolarizing current-pulses (n ¼ 11) and even without affecting theoriginal firing frequency (n ¼ 7; Fig. 5B). The average maximumfiring frequency induced by simple SI stimulations was 11.7 ± 1.9 Hz(n ¼ 7) and by SI stimulations plus a following current-step was11 ± 1.8 Hz (n ¼ 6; no difference compared with simple SI,P ¼ 0.49). However, average frequency grade (i.e. difference in

Fig. 5. Cholinergic muscarinic control of graded firing activity in LA neurons. (A) Depolarizing current pulse (open arrowhead) given 50 s after induction of theplateau potential (plateau induced by stimulation of SI at 10 Hz for 10 s, see Fig. 4B) is unable to increase original firing frequency. Instead, a slow decrease of firingfrequency was observed (indicated by filled arrowhead). However, a brief SI stimulation (10 Hz, 2 s) shortly before depolarizing current steps were applied wassufficient for the graded increases of discharge rates. (B) A train of SI stimulations (10 Hz, 1 s) induced a graded increase of discharge rates even without affectingthe original firing frequency (four levels as indicated). The lower diagrams correspond to the peristimulus histograms (bin width in A and B: 600 and 480 ms,respectively). Drugs were added to the Ringer solution for A and B as indicated above.



firing frequency between two neighbor levels) elicited by SIstimulation together with a subsequent current-step was significantlyhigher as compared with a simple SI stimulation [2.9 ± 1.1 Hz for SIwith current-step-driven spikes (n ¼ 6) and 1.37 ± 0.5 Hz for simpleSI (n ¼ 7), P < 0.01].Although the long-lasting firing activity was completely blocked by

atropine, we tested neurons in the presence of (S)-a-methyl-4-carboxyphenylglycine (s-MCPG) (500 lm), in order to exclude apossible influence of metabotropic glutamate receptors (Al-Yahyaet al., 2003). Adding s-MCPG to the perfusion solution did not alterthe ability of LA neurons to express a prolonged firing activity (n ¼ 5)or to exhibit a graded increase in the firing rate (n ¼ 3 for simple SIstimulation; n ¼ 4 for SI stimulation plus current step, data notshown).In addition, we investigated the possible mechanisms underlying the

muscarinic control of firing activity in LA neurons.Prevention of Ca2+ influx by removal of extracellular Ca2+

completely and reversibly blocked the muscarinic plateau potentialsinduced by CCh (n ¼ 5, reversibility tested for three cells; seeFig. 7A below). This clearly suggested that Ca2+ influx associatedwith spiking is required. The graded changes of firing frequency arealso expected to be regulated by a Ca2+-dependent mechanism(Fransen et al., 2006). However, in the present study, we observedincreases of firing rates after stimulation of cholinergic fibers evenwithout affecting the original firing frequency during the stimulus.Indeed, the activation of muscarinic receptors (type 1) could enhancethe intracellular Ca2+ concentration by IP3-mediated Ca2+ releasefrom internal stores. In order to examine whether graded increases ofdischarge rate could also be produced after depleting intracellularCa2+ stores, we tested neurons in the presence of the sarcoplasmic–endoplasmic reticulum ATPase inhibitor thapsigargin (2–3 lm). In allneurons tested (n ¼ 4), we found that thapsigargin did not preventthe production of long-lasting firing activity via SI stimulation modes(Fig. 6A and B). Moreover, we observed well-defined gradedincreases in firing frequency after SI stimulation even withoutchanges in spike frequency during the stimulation (n ¼ 3; Fig. 6A).The average frequency grade elicited by simple SI stimulation in thepresence of thapsigargin was 1.4 ± 0.2 Hz (n ¼ 3; not different tocontrol, P ¼ 0.9, Fig. 6C).Activation of muscarinic receptors in BLA neurons has been shown

to result in a suppression of different K+-conductances (Womble &Moises, 1992, 1993) as well as in the activation of a non-selectivecation current (Yajeya et al., 1999). Both effects might account for theslow membrane depolarization and therefore the graded increase infiring rate. However, a graded increase in firing frequency caused bysimple stimulation of cholinergic fibers was also observed during bathapplication of Ba2+ (1 mm; n ¼ 3; Supplementary Fig. S4), a cationbelieved to block I‘leak’, IM and G-protein-coupled inwardly rectifyingK+ (GIRK) currents.As the Ca2+-dependent plateau potentials in EC LV neurons were

mediated by a Ca2+-activated non-specific cation current (ICAN), wetested the effects of the ICAN current blocking agent flufenamic acid(10 lm) (Partridge & Valenzuela, 2000) on the ability of LA neuronsto generate plateau potentials. Indeed, as illustrated in Fig. 7B,addition of flufenamic acid consistently prevented the cells’ ability togenerate plateau potentials triggered through Ca2+ up-regulationpreceded by cholinergic fiber stimulation (n ¼ 4). Finally, CCh-induced persistent activity in LA neurons was blocked by 2-APB(100 lm; n ¼ 4; Fig. 7C). 2-APB, as well as flufenamic acid, isknown to be a non-selective antagonist of several transient receptorpotential (TRP)-mediated currents (Ramsey et al., 2006), suggestingthat the channels underlying ICAN may belong to the TRP family.

Discussion

Graded persistent activity is a common intrinsic behaviorof EC and LA neurons

In the present study we show that principal LA neurons, activated bymuscarinic receptors, can generate oscillatory graded persistentactivity. Analogous to EC LV cells, persistent activity of LA neuronsis not due to local circuit reverberation mechanisms, but represents theintrinsic cellular property. It displays similar activity-, voltage- andcalcium-dependent characteristics, is robust to distractors (i.e. briefdepolarizing or hyperpolarizing current injections), and can beblocked by a CAN current blocking agent. Moreover, the cationicconductance underlying persistent firing in LA neurons is possiblymediated by the TRP family of ion channels, as previously reportedfor EC LV cells (Al-Yahya et al., 2003). Taken together, these findingsdemonstrate that graded persistent activity is a common intrinsicbehavior of EC LV and LA neurons. In fact, electrophysiological andmorphological properties of rat BLA and EC LV neurons are quitesimilar (Washburn & Moises, 1992a; Hamam et al., 2000; Egorovet al., 2002b); in these cell types activation of muscarinic receptors islinked to modulation of comparable conductances (i.e. the reduction ofI‘leak’, IM, IsAHP and the activation of the non-selective cationiccurrents (Washburn & Moises, 1992b; Womble & Moises, 1992;Yajeya et al., 1999; Egorov et al., 2003). CCh-induced depolarizationfor about 8–10 mV from the resting membrane potential could be aresult of a muscarinic activation of a Ca2+-independent nonselectivecationic conductance in both LA and EC LV principal neurons(Egorov et al., 2003; Yajeya et al., 1999). However, the finding thatthe CCh-induced depolarization in LA neurons was associated with anincrease in input resistant (Washburn & Moises, 1992b, and our data)suggested that the muscarinic blockade of potassium conductances isalso involved. The persistent firing could be induced by stimulationfrom voltage levels of about 10 mV negative to spike threshold. Infact, in vivo recordings showed definitely more depolarized (i.e. about)63 mV) levels of resting membrane potential as compared within vitro, which might provide the trigger for persistent activity.The ability of neurons to maintain the sustained multiple levels of

firing frequency may represent a quite universal form of neuronalactivity (Major & Tank, 2004) that might be vital for transientlysustaining memory events. Moreover, the ‘single cell’ mechanismunderlying the generation of graded persistent activity may constitutea widespread intrinsic neuronal property.

Cholinergic control of graded firing activity in LA neurons

One important novel finding in the present study is that the cholinergicprojections from the n. M and the SI are essential and sufficient for thecontrol and modulation of graded firing activity in LA neurons.Several sets of data predict that the dense cholinergic projection fromthe n. M to the BLA is involved in working memory (Beninger et al.,1994; Mallet et al., 1995) as well as in memory acquisition andconsolidation processes (Power & McGaugh, 2002), and in thepathophysiology of Alzheimer’s disease (Shinotoh et al., 2003). Herewe show that activation of these cholinergic afferents: (i) elicits aprolonged firing activity in LA neurons, isolated from glutamatergicand GABAergic innervations, via muscarinic pathways; (ii) is requiredto maintain and to increase firing rates in a graded manner; and (iii) issufficient for the graded increases of stable discharge rates evenwithout associated up-regulation of Ca2+. This suggests that the n. Mmakes a significant contribution in regulation of the persistent firing ofLA neurons. Whereas simple SI stimulation could be sufficient ingraded increases of firing rate, making the grade by SI stimulation



together with subsequent current-step-driven spikes was more effect-ive (i.e. higher difference in firing frequency between two neighborlevels). This is not surprising as the activity of the LA neurons ismodulated by cholinergic and glutamatergic components (Smith &

Pare, 1994; Sah & Lopez De Armentia, 2003), and their co-activationmight have a stronger impact on the graded increase of firing activity.By contrast, GABA-mediated inhibition represents an effective way ofproducing graded decreases of firing frequency. Thus, the intrinsic

Fig. 6. Depletion of Ca2+ stores does not prevent graded firing activity. The frequency rates can be increased in a graded fashion after simple SI stimulation(A, eight levels as indicated) or SI stimulation followed by subsequent current steps (B, from 1 to 5 as indicated) in the presence of thapsigargin (3 lm). The lowerdiagrams correspond to the peristimulus histograms (bin width of 600 ms). Note in A that the graded increase in firing frequency after SI stimulations was observedeven without affecting spike frequency during the stimulation. Initial Vm in A and B is )62 mV. Drugs were added to the Ringer solution for A and B as indicatedabove. (C) Plots of firing rates (from � 6 Hz) vs. number of stimulations. Note that simple SI stimulation (1 s SI in A) requires more steps to reach 10 Hz than the1 s SI-stimulation plus action potential (1 s SI + AP in B).



graded persistent activity is under multiple (i.e. muscarinic, glutama-tergic and GABAergic) controls.However, the spontaneous firing rate of BLA projection cells in vivo

is quite low (Pare & Gaudreau, 1996) and emotional arousal seems tobe accompanied by only a modest increase in firing rate (Pare &Collins, 2000). Persistent firings at a low frequency (1–3 Hz) wereobserved also in our in vitro study. However, in most cases, neuronalbehavior was characterized by a significantly higher (up to � 14 Hz)firing frequency. This may suggest that in vivo the internal neuronalability to maintain persistent activity is under dominant control of

synaptic inhibition; alternatively, the firing patterns could be modifiedby external inputs.

Mechanism of graded persistent activity in EC LV and LAneurons

As graded persistent activity is a common intrinsic behavior of EC LVand LA neurons, it can be hypothesized that the mechanismsunderlying this phemomenon are identical. Thus, in the case ofcontinuously activating muscarinic receptors by bath-applied CCh, thespike-induced Ca2+ influx triggering a cationic current-mediatedpotential is linked to a basic mechanism for the generation ofpersistent activity at the single cell level. The potential mechanisms ofgraded persistent activity, which allows input integration and main-tenance of graded levels of stable firing rate in EC LV neurons, havebeen recently described in a modeling study by Fransen et al. (2006).This point attractor model proposes that the CAN channels areunder dynamic control of two opposing Ca2+-dependent metabolicprocesses – separated by a neutral zone – which stabilize a high and alow conductive state of the channel. When the number of channels islarge, the macroscopic transitions appear graded. Thus, gradedchanges in firing frequency can be achieved by Ca2+-dependentmetabolic regulation of a total CAN-channel conductance.In our present study, we also observed increases of firing rates after

stimulation of cholinergic fibers without changes in firing frequencyduring the stimuli (i.e. without spike-induced Ca2+ up-regulation).Neither an IP3-mediated Ca2+ release from internal stores nor amuscarinic receptor-mediated potassium conductance inhibition arerequired to produce a graded increase in firing rate. The depletion ofCa2+ stores also does not prevent graded persistent activity in EC LVneurons (Fransen et al., 2006). This may lend support to speculationsthat entry of Ca2+ through voltage-gated channels during an actionpotential is essential for maintaining CAN current and persistentspiking; hovever, a graded increase in firing frequency (e.g. increase inthe maximal conductance of a subset of the CAN channel) is – in thismode – under control of the activation of muscarinic secondmessenger cascades. The possible mechanisms involved in themuscarinic control of graded firing activity need further detailedinvestigation.

Functional implications of graded persistent activityin EC and LA neurons

It is assumed that intrinsic graded persistent activity in the EC neuronscould underlie the sustained spiking activity observed duringperformance of delayed non-match and delayed match to sampletasks in the EC of rats (Young et al., 1997) and monkeys (Suzukiet al., 1997). Moreover, the recent demonstration that lesions of thecholinergic innervation of the EC in rats selectively impaired encodingof novel but not familiar stimuli in a delay non-match-to-sample task(McGaughy et al., 2005) indicates an important role for cholinergicmodulation of the EC in working memory for novel stimuli.Functional magnetic resonance imaging (fMRI) studies in humansalso show persistent activity in parahippocampal structures duringworking memory for novel but not familiar stimuli (Stern et al., 2001;Schon et al., 2004). Furthermore, this sustained encoding-related delayactivity was reduced in subjects receiving the muscarinic cholinergicantagonist scopolamine (Schon et al., 2005). Thus, intrinsic sustainedspiking activity that does not depend on previous strengthening ofsynapses may be a potential mechanism for sustaining informationabout novel items in a short-term memory buffer that was predicted by

Fig. 7. ICAN underlying persistent firing activity in an LA neuron. (A) Com-plete block of persistent activity by removal of extracellular Ca2+. (B) Flu-fenamic acid (10 lm) blocks plateau potential triggered by a depolarizing steppreceded by SI stimulation. The arrow below the right current trace indicates aDC shift. (C) Complete block of persistent firing by 2-APB (100 lm). InitialVm in A, B and C: )60.5, )59 and )58 mV, respectively. Drugs were added tothe Ringer solution as indicated above.



computational modeling studies (Lisman & Idiart, 1995; Fransenet al., 2002). However, our experimental observation for intrinsicgraded persistent activity in the LA points to the possibility that thisphenomenon may be of more general function.

The amygdala appears to be important for emotional and experi-ence-dependent learning (McGaugh, 2004). Single neurons that signalnovelty or familiarity with an increase in firing rate have been recentlyidentified in the human amygdala (Rutishauser et al., 2006). Sustainedspiking activity has been reported in the primate amygdala byencoding the positive and negative value of visual stimuli duringlearning (Paton et al., 2006). A behavioral study from Pelletier et al.(2005) has provided direct evidence that emotional arousal graduallyincreases the spontaneous firing rates of BLA neurons in a sustainedmanner (i.e. for about 2 h after the introduction of the footshock).According to pharmaco-behavioral observations (reviewed by Pare,2003), such long-lasting increases in the firing activity might be aconsequence of the release of noradrenaline and ⁄ or ACh in the BLAafter emotional arousal. Our present in vitro study shows that AChactivates the intrinsic cation current that causes sustained spikingactivity in LA neurons; this might contribute to the lasting increase ofBLA activity following emotional arousal. Interestingly, microdialysisstudies have revealed that ACh levels in the BLA were long-lastingelevated following learning episodes (McIntyre et al., 2003). Thissupports our observation that continuous muscarinic activation isrequired to maintain a sustained level of firing frequency and to createits graded changes. Release of ACh in the amygdala is positivelyrelated to performance during a hippocampus-dependent spatialworking memory task (McIntyre et al., 2003), a finding consistentwith the view that activation of muscarinic cholinergic receptors in theBLA plays an essential role in modulation of memory formation inMTL structures (Power & McGaugh, 2002). The modulatoryinfluences of the BLA on encoding and consolidation processesoccurring in MTL memory structures, including the hippocampusand associated parahippocampal regions (i.e. entorhinal, perirhinaland parahippocampal cortices), might be fundamental for thememorizing of emotionally salient experiences, as evidenced bybehavioral studies and functional neuroimaging (Roesler et al., 2002;Kilpatrick & Cahill, 2003; McGaugh, 2004). Recently, an fMRIstudy has provided direct evidence that special interactions betweenthe amygdala and the EC provide better memory for emotionalevents (Dolcos et al., 2004). Reciprocal functional connectivitybetween the BLA and the EC has been confirmed anatomically andphysiologically (Brothers & Finch, 1985; Finch et al., 1986;Pikkarainen et al., 1999; Pitkanen et al., 2000; von Bohlen undHalbach & Albrecht, 2002). The functional projections from the LAto the EC may carry the modulatory influences of the BLA to thepersistent activity occurring in the EC. Moreover, because principalneurons of both areas are able to generate graded persistent activity,synchronization of oscillatory persistent activity between thesestructures might not be excluded.

More generally, studies of synchronized oscillatory activity withinthe BLA propose that such activity may facilitate the temporal lobe aswell as neocortical processes involved in consolidating explicit ordeclarative memory (Pare, 2003; Pelletier & Pare, 2004). BLA-mediated facilitation of rhinal (perirhinal–entorhinal) interactions hasbeen recently observed (Raz et al., 2006). To conclude, our presentin vitro findings provide evidence for how muscarinic cholinergicactivation may control the graded firing activity in LA neurons.Moreover, we propose that the internal ability of LA neurons togenerate graded persistent activity could play an essential role in theBLA-mediated modulation of memory processing; however, this pointneeds further investigation.

Supplementary material

The following supplementary material may be found on www.blackwell-synergy.comFig. S1. Plateau potential amplitude in LA neurons decreased withmembrane hyperpolarization.Fig. S2. Graded persistent activity in LA neurons does not depend onsynaptic neurotransmission.Fig. S3. Elimination of persistent activity in LA neurons by prolongedsynaptic stimulation.Fig. S4. Bath application of Ba2+ does not prevent a graded increasein firing frequency evoked by simple SI stimulation.

Acknowledgements

We thank Andreas Draguhn (Heidelberg) for the support of this work; MichaelHasselmo (Boston), Kai Kaila (Helsinki) and William Wisden (Glasgow) forhelpful comments on the manuscript. This work was funded by the DeutscheForschungsgemeinschaft (SFB 636 ⁄ A5).

Abbreviations

ACh, acetylcholine; BLA, basolateral complex of the amygdala; CCh,carbachol; EC, entorhinal cortex; EC LV, layer V of the entorhinal cortex;fMRI, functional magnetic resonance imaging; ICAN, Ca2+-activated non-specific cation current; LA, the lateral nucleus of the amygdala; MCPG, (S)-a-methyl-4-carboxyphenylglycine; MTL, medial temporal lobe; n. M, nucleusbasalis of Meynert; RMP, resting membrane potential; SI, substantia innom-inata; TRP, transient receptor potential.

References

Al-Yahya, E., Hamel, E., Kennedy, T.E., Alonso, A.A. & Egorov, A.V. (2003)Persistent activity in entorhinal cortex neurons induced by muscarinic andmetabotropic glutamate receptor activation and its dependence on TRPchannels. Soc. Neurosci. Abstracts, 377.5.

Barros, D.M., Pereira, P., Medina, J.H. & Izquierdo, I. (2002) Modulation ofworking memory and of long- but not short-term memory by cholinergicmechanisms in the basolateral amygdala. Behav. Pharmacol., 13, 163–167.

Beninger, R.J., Kuhnemann, S., Ingles, J.L., Jhamandas, K. & Boegman, R.J.(1994) Mnemonic deficits in the double Y-maze are related to the effects ofnucleus basalis injections of ibotenic and quisqualic acid on cholineacetyltransferase in the rat amygdala. Brain. Res. Bull., 35, 147–152.

von Bohlen und Halbach, O. & Albrecht, D. (2002) Reciprocal connections ofthe hippocampal area CA1, the lateral nucleus of the amygdala and corticalareas in a combined horizontal slice preparation. Neurosci. Res., 44, 91–100.

Brothers, L.A. & Finch, D.M. (1985) Physiological evidence for an excitatorypathway from entorhinal cortex to amygdala in the rat. Brain. Res., 359,10–20.

Cahill, L. & McGaugh, J.L. (1998) Mechanisms of emotional arousal andlasting declarative memory. Trends Neurosci., 21, 294–299.

Carlsen, J., Zaborszky, L. & Heimer, L. (1985) Cholinergic projections fromthe basal forebrain to the basolateral amygdaloid complex: a combinedretrograde fluorescent and immunohistochemical study. J. Comp. Neurol.,234, 155–167.

Dolcos, F., LaBar, K.S. & Cabeza, R. (2004) Interaction between the amygdalaand the medial temporal lobe memory system predicts better memory foremotional events. Neuron, 42, 855–863.

Egorov, A.V., Angelova, P.R., Heinemann, U. & Muller, W. (2003) Ca2+-independent muscarinic excitation of rat medial entorhinal cortex layer Vneurons. Eur. J. Neurosci., 18, 3343–3351.

Egorov, A.V., Hamam, B.N., Fransen, E., Hasselmo, M.E. & Alonso, A.A.(2002a) Graded persistent activity in entorhinal cortex neurons. Nature, 420,173–178.

Egorov, A.V., Heinemann, U. & Muller, W. (2002b) Differential excitabilityand voltage-dependent Ca2+ signalling in two types of medial entorhinalcortex layer V neurons. Eur. J. Neurosci., 16, 1305–1312.

Emson, P.C., Paxinos, G., Le Gal La Salle, G., Ben-Ari, Y. & Silver, A. (1979)Choline acetyltransferase and acetylcholinesterase containing projections



from the basal forebrain to the amygdaloid complex of the rat. Brain Res.,13, 271–282.

Finch, D.M., Wong, E.E., Derian, E.L., Chen, X.H., Nowlin-Finch, N.L. &Brothers, L.A. (1986) Neurophysiology of limbic system pathways in the rat:projections from the amygdala to the entorhinal cortex. Brain Res., 9, 273–284.

Fransen, E., Alonso, A.A. & Hasselmo, M.E. (2002) Simulations of the role ofthe muscarinic-activated calcium-sensitive nonspecific cation current INCMin entorhinal neuronal activity during delayed matching tasks. J. Neurosci.,22, 1081–1097.

Fransen, E., Tahvildari, B., Egorov, A.V., Hasselmo, M.E. & Alonso, A.A.(2006) Mechanism of graded persistent cellular activity in entorhinal cortexlayer V neurons. Neuron, 49, 735–746.

Hamam, B.N., Kennedy, T.E., Alonso, A. & Amaral, D.G. (2000)Morphological and electrophysiological characteristics of layer V neuronsof the rat medial entorhinal cortex. J. Comp. Neurol., 418, 457–472.

Kilpatrick, L. & Cahill, L. (2003) Amygdala modulation of parahippocampaland frontal regions during emotionally influenced memory storage. Neuro-image, 20, 2091–2099.

Lisman, J.E. & Idiart, M.A. (1995) Storage of 7 + ⁄ - 2 short-term memories inoscillatory subcycles. Science, 267, 1512–1515.

Major, G. & Tank, D. (2004) Persistent neural activity: prevalence andmechanisms. Curr. Opin. Neurobiol., 14, 675–684.

Mallet, P.E., Beninger, R.J., Flesher, S.N., Jhamandas, K. & Boegman, R.J.(1995) Nucleus basalis lesions: implication of basoamygdaloid cholinergicpathways in memory. Brain Res. Bull., 36, 51–56.

McGaugh, J.L. (2000) Memory – a century of consolidation. Science, 287,248–251.

McGaugh, J.L. (2004) The amygdala modulates the consolidation of memoriesof emotionally arousing experiences. Annu. Rev. Neurosci., 27, 1–28.

McGaugh, J.L., Cahill, L. & Roozendaal, B. (1996) Involvement of theamygdala in memory storage: interaction with other brain systems. Proc.Natl Acad. Sci. USA, 93, 13508–13514.

McGaugh, J.L., McIntyre, C.K. & Power, A.E. (2002) Amygdala modulation ofmemory consolidation: interaction with other brain systems. Neurobiol.Learn. Mem., 78, 539–552.

McGaughy, J., Koene, R.A., Eichenbaum, H. & Hasselmo, M.E. (2005)Cholinergic deafferentation of the entorhinal cortex in rats impairs encodingof novel but not familiar stimuli in a delayed nonmatch-to-sample task.J. Neurosci., 25, 10273–10281.

McIntyre, C.K., Marriott, L.K. & Gold, P.E. (2003) Cooperation betweenmemory systems: acetylcholine release in the amygdala correlates positivelywith performance on a hippocampus-dependent task. Behav. Neurosci., 117,320–326.

Pare, D. (2003) Role of the basolateral amygdala in memory consolidation.Prog. Neurobiol., 70, 409–420.

Pare, D. & Collins, D.R. (2000) Neuronal correlates of fear in the lateralamygdala: multiple extracellular recordings in conscious cats. J. Neurosci.,20, 2701–2710.

Pare, D. & Gaudreau, H. (1996) Projection cells and interneurons of the lateraland basolateral amygdala: distinct firing patterns and differential relation totheta and delta rhythms in conscious cats. J. Neurosci., 16, 3334–3350.

Partridge, L.D. & Valenzuela, C.F. (2000) Block of hippocampal CAN channelsby flufenamate. Brain. Res., 867, 143–148.

Paton, J.J., Belova, M.A., Morrison, S.E. & Salzman, C.D. (2006) The primateamygdala represents the positive and negative value of visual stimuli duringlearning. Nature, 439, 865–870.

Pelletier, J.G., Likhtik, E., Filali, M. & Pare, D. (2005) Lasting increases inbasolateral amygdala activity after emotional arousal: implications forfacilitated consolidation of emotional memories. Learn. Mem., 12, 96–102.

Pelletier, J.G. & Pare, D. (2004) Role of amygdala oscillations in theconsolidation of emotional memories. Biol. Psychiatry, 55, 559–562.

Pikkarainen, M., Ronkko, S., Savander, V., Insausti, R. & Pitkanen, A. (1999)Projections from the lateral, basal, and accessory basal nuclei of theamygdala to the hippocampal formation in rat. J. Comp. Neurol., 403, 229–260.

Pitkanen, A., Pikkarainen, M., Nurminen, N. & Ylinen, A. (2000) Reciprocalconnections between the amygdala and the hippocampal formation,perirhinal cortex, and postrhinal cortex in rat. Ann. NY Acad. Sci., 911,369–391.

Power, A.E. & McGaugh, J.L. (2002) Phthalic acid amygdalopetal lesion of thenucleus basalis magnocellularis induces reversible memory deficits in rats.Neurobiol. Learn. Mem., 77, 372–388.

Power, A.E., Vazdarjanova, A. & McGaugh, J.L. (2003) Muscarinic cholinergicinfluences in memory consolidation. Neurobiol. Learn. Mem., 80, 178–193.

Ramsey, I.S., Delling, M. & Clapham, D.E. (2006) An introduction to trpchannels. Annu. Rev. Physiol., 68, 619–647.

Raz, R., Pelletier, J.G., Bauer, E., P. & Pare, D. (2006) Emotional enhancementof memory via amygdala-driven facilitation of rhinal interactions. Nat.Neurosci., 9, 1321–1329.

Roesler, R., Roozendaal, B. & McGaugh, J.L. (2002) Basolateral amygdalalesions block the memory-enhancing effect of 8-Br.-cAMP infused into theentorhinal cortex of rats after training. Eur. J. Neurosci., 15, 905–910.

Rutishauser, U., Mamelak, A.N. & Schuman, E.M. (2006) Single-trial learningof novel stimuli by individual neurons of the human hippocampus-amygdalacomplex. Neuron, 49, 805–813.

Sah, P. & Lopez De Armentia, M. (2003) Excitatory synaptic transmission inthe lateral and central amygdala. Ann. NY Acad. Sci., 985, 67–77.

Schon, K., Atri, A., Hasselmo, M.E., Tricarico, M.D., Lopresti, M.L. & Stern,C.E. (2005) Scopolamine reduces persistent activity related to long-termencoding in the parahippocampal gyrus during delayed matching in humans.J. Neurosci., 25, 9112–9123.

Schon, K., Hasselmo, M.E., LoPresti, M.L., Tricarico, M.D. & Stern, C.E.(2004) Persistence of parahippocampal representation in the absence ofstimulus input enhances long-term encoding: a functional magneticresonance imaging study of subsequent memory after a delayed match-to-sample task. J. Neurosci., 24, 11088–11097.

Shinotoh, H., Fukushi, K., Nagatsuka, S., Tanaka, N., Aotsuka, A., Ota, T.,Namba, H., Tanada, S. & Irie, T. (2003) The amygdala and Alzheimer’sdisease: positron emission tomographic study of the cholinergic system. Ann.NY Acad. Sci., 985, 411–419.

Smith, Y. & Pare, D. (1994) Intra-amygdaloid projections of the lateral nucleusin the cat: PHA-L anterograde labeling combined with postembeddingGABA and glutamate immunocytochemistry. J. Comp. Neurol., 342, 232–248.

Squire, L.R., Stark, C.E. & Clark, R.E. (2004) The medial temporal lobe. Annu.Rev. Neurosci., 27, 279–306.

Squire, L.R. & Zola-Morgan, S. (1991) The medial temporal lobe memorysystem. Science, 253, 1380–1386.

Stern, C.E., Sherman, S.J., Kirchhoff, B.A. & Hasselmo, M.E. (2001) Medialtemporal and prefrontal contributions to working memory tasks with noveland familiar stimuli. Hippocampus, 11, 337–346.

Suzuki, W.A., Miller, E.K. & Desimone, R. (1997) Object and place memory inthe macaque entorhinal cortex. J. Neurophysiol., 78, 1062–1081.

Washburn, M.S. & Moises, H.C. (1992a) Electrophysiological and morpholo-gical properties of rat basolateral amygdaloid neurons in vitro. J. Neurosci.,12, 4066–4079.

Washburn, M.S. & Moises, H.C. (1992b) Muscarinic responses of ratbasolateral amygdaloid neurons recorded in vitro. J. Physiol., 449, 121–154.

Womble, M.D. & Moises, H.C. (1992) Muscarinic inhibition of M-current anda potassium leak conductance in neurones of the rat basolateral amygdala.J. Physiol., 457, 93–114.

Womble, M.D. & Moises, H.C. (1993) Muscarinic modulation of conductancesunderlying the afterhyperpolarization in neurons of the rat basolateralamygdala. Brain Res., 621, 87–96.

Yajeya, J., Juan, A.D., Bajo, V.M., Riolobos, A.S., Heredia, M. & Criado, J.M.(1999) Muscarinic activation of a non-selective cationic conductancein pyramidal neurons in rat basolateral amygdala. Neuroscience, 88, 159–167.

Young, B.J., Otto, T., Fox, G.D. & Eichenbaum, H. (1997) Memoryrepresentation within the parahippocampal region. J. Neurosci., 17, 5183–5195.



Documents

Muscarinic control of graded persistent activity in lateral amygdala neurons