Neurocomputing 26}27 (1999) 375}380

A biophysical simulation of intrinsic and networkproperties of entorhinal cortex

Erik FranseH n!,*, Gene V. Wallenstein!,2, Angel A. Alonso",Clayton T. Dickson", Michael E. Hasselmo!,3

!Department of Psychology, Harvard University, 33 Kirkland St., Cambridge, MA 02138, USA"Montreal Neurological Institute and McGill University, Montreal, QC H3A Canada

Accepted 18 December 1998

Abstract

Biophysical simulations of stellate and non-stellate (pyramidal-like) cells in entorhinal cortexwere used to analyze the currents underlying intrinsic properties such as subthreshold oscilla-tions and plateau depolarization potential oscillations. In accord with experimental data,simulations demonstrated increases in subthreshold oscillation frequency and amplitude whena cell is depolarized from resting potential. Networks of reduced model cells were used to model"eld potentials due to synchronous population activity induced by muscarinic agonists in slicepreparations of entorhinal cortex. The dynamics of these "eld potentials were replicated,including an initial long-duration ictiform event with decreasing frequency, followed by short-duration ictiform events occurring at regular long intervals. ( 1999 Elsevier Science B.V. Allrights reserved.

Keywords: Biophysical simulation; Entorhinal cortex; Subthreshold oscillations; Plateaudepolarizations; Synchronous population activity

1. Introduction

The entorhinal cortex plays an essential role in the function of hippocampalcircuitry, including episodic memory [5,7,12]. Extensive research has been done on

*Corresponding author. Present address. Department of Numerical Analysis and Computing Science,Royal Institute of Technology, S-100 44 Stockholm, Sweden; e-mail: erikf@nada.kth.se.

2Present address. Department of Psychology, University of Utah, Salt Lake City, UT 84112, USA.3Present address. Department of Psychology, Boston University, 64 Cummington St., Boston, MA

02215, USA.

0925-2312/99/$ } see front matter ( 1999 Elsevier Science B.V. All rights reserved.PII: S 0 9 2 5 - 2 3 1 2 ( 9 9 ) 0 0 0 2 7 - 2

the connectivity and intrinsic properties of entorhinal cortex neurons. It has also beenfound that cholinergic innervation of entorhinal cortex may regulate the hippocampaltheta rhythm [1,2]. The cholinergic in#uence on theta rhythm may largely result fromchanges in intrinsic properties of layer II neurons which have been demonstrated invitro, including a large plateau depolarization potential [8,9] and changes in subthres-hold oscillation dynamics [8,9]. Indeed, it has recently been shown that thesecholinergic e!ects contribute to synchronous population oscillations observed inentorhinal cortex slice preparations [4]. Here we present simulations of entorhinalcortex developed using the GENESIS simulation program [3]. The simulations wereperformed on two levels, one on the network level, and one on the cellular level.Understanding how changes in cellular properties can cause changes in networkoscillations in slices and whole animal preparations requires the use of detailedbiophysical network simulations. In addition, these simulations may demonstratehow cholinergic modulation could induce sustained activity for short-term bu!eringof information during memory tasks.

2. Network simulations

Networks of biophysically modeled cells were used to model "eld potentials due tosynchronous population activity induced by muscarinic agonists in slice preparationsof entorhinal cortex. The network simulations contained 100 stellate cells and 25inhibitory interneurons. Application of carbachol to the slice preparation wasmodeled by altering intrinsic neuron properties to re#ect the membrane changesinduced by muscarinic activation. These changes in intrinsic properties caused "eldevents similar to those observed experimentally in the slice preparation, Fig. 1A in [4].The "eld events started with a long-duration ictiform event with decreasing frequencyfollowed by slow periodic occurrence of short-duration ictiform events.

Simulations were used to explore the cellular basis of some of the properties of theseictiform events. In particular, the long-duration ictiform event starts with an initialhigh-frequency oscillation (about 15 Hz) which gradually decreases to lower frequen-cies (about 3 Hz), as shown in Fig. 1. This replicates the average change in frequencyobserved experimentally during long-duration events, Fig. 1E in [4]. In the model,this change in frequency resulted from increased concentrations of GABA causingactivation of presynaptic GABA

Breceptors on the presynaptic terminals of inhibitory

interneurons. As shown in Fig. 1 (middle), this causes a progressive change in therelative size of GABA

Aand GABA

Bcurrents in the model which causes a progressive

decrease in frequency of oscillations. Weaker GABAA

receptor activation results inslower oscillations. Changes in GABA

Bactivation have been proposed to cause

functionally important changes in dynamics within the hippocampus itself [10,11]. Inthe simulations, the short-duration ictiform events required a mechanism for rapidlyshutting down activity after a brief period of oscillations. This was obtained in themodel with activation of presynaptic metabotropic glutamate receptors, causingdecreases in excitatory glutamatergic transmission and thereby shutting o! therecurrent drive which maintained the network oscillation.

376 E. Franse&n et al. / Neurocomputing 26}27 (1999) 375}380

Fig. 1. Top: simulation of muscarinic in#uences in the network results in an initial long-duration ictiformevent in the simulated "eld potential. Middle: increased activation of presynaptic GABA

Breceptors on

terminals of interneurons results in a progressive change in the relative magnitude of GABAA

and GABAB

receptor activation. Weaker GABAA

receptor activation results in slower oscillations. Bottom: the networkoscillation shifts from high frequencies (around 15 Hz) to lower frequencies (around 3 Hz), in accordancewith data.

3. Stellate cells

Biophysical modeling was performed to study the ionic basis of the subthresholdmembrane potential oscillations of layer II stellate cells (SCs). In previous models aninteraction between an outward recti"er and a persistent sodium channel has com-monly been assumed. Recent data by Dickson and Alonso 1999 (submitted) suggestthat the hyperpolarization-activated inward current I

)is necessary for the generation

of membrane potential oscillations. The primary objective was thus to study theinterplay between I

)and a sustained persistent-type sodium current I

N!P. The I

)was

modeled with equations in the Hodgkin}Huxley formalism. The steady-state activa-tion of I

)was described by a Boltzmann-style equation. The time constant of

activation was described by two bell-shaped functions of membrane potential, thefaster component with a maximum of about 100 ms and the slower componenta maximum of about 600 ms.

E. Franse&n et al. / Neurocomputing 26}27 (1999) 375}380 377

Fig. 2. Left: simulated voltage clamp. Both time course and relative current amplitude correspond closelyto experimental data for the depolarizing as well as hyperpolarizing voltage steps. Right: subthresholdmembrane oscillation. In the "gure, the cell was depolarized from resting potential !63 to !57 mV bya constant current injection.

Using the derived equations for I), voltage clamp simulations were performed and

the result compared with experimental data. The experimental data has indications ofa space clamp component, presumably due to the spatial distribution of I

)and the

extensive dendritic tree of SCs. To account for this, I)

was placed distally and thedendritic attenuation was modeled. Recent in situ labeling of I

)supports the inter-

pretation that I)

is preferentially located distally. The simulated voltage clampcurrents show good agreement with experimental data in time course and relativeamplitude, Fig. 2 (left).

The model of I)described above was combined with a model for the persistent type

Na-current to investigate if these two currents are su$cient for the generation ofmembrane potential oscillations. The model of the sodium current I

N!Pwas derived

from the modeling and experimental data by Alonso and Magistretti (submitted). Thesimulations show that an interplay between I

)and I

N!Pis su$cient to account for the

generation of membrane potential oscillations in layer II SCs, Fig. 2 (right). Theexistence of oscillations in the model depends on the presence of membrane potentialnoise, here caused by random `backgrounda synaptic activations.

4. Pyramidal-like cells

During cholinergic activation, the pyramidal type cells of layer II show plateaudepolarization potentials which slowly oscillate. This oscillatory plateau depolariz-ation is presumed to rely on an interaction between a calcium sensitive non-speci"ccationic current and #uctuations in the internal Ca-concentration. To model thesee!ects, calcium entering the cell through high threshold voltage-gated Ca-channelsinteracts with an internal Ca-store [6]. The release from the store is Ca-sensitive

378 E. Franse&n et al. / Neurocomputing 26}27 (1999) 375}380

(CICR). This enables an oscillatory exchange of calcium between the store and itssurrounding. Calcium is removed from the cell by active pumping. The internaloscillatory calcium a!ects the non-speci"c cationic current, thereby providingthe basis for the membrane oscillations. Furthermore, calcium entering throughthe Ca-channels also a!ects a fast K

C-type current as well as a slow K

AHP-type

current.Using the Ca-model described above, carbachol induced plateau depolarization

potential oscillations [9] were simulated. Carbachol e!ects were modeled as a gain inthe e!ective transfer of calcium from the Ca-channels to the internal Ca-pool.

The pyramidal-like cell of layer II shows several other characteristics thatdistinguish it from e.g. hippocampal pyramidal cells. The action potential repolariz-ation, for instance, is dependent on a Ca-dependent K

C-current. When Ca-

channels are blocked, a supra-threshold plateau follows spike initiation. Activationof a slow K

M-type current and Na-inactivation terminates the plateau in the

model.

Acknowledgements

This work was supported by a grant from the Human Frontier Science Program,the Swedish Foundation for International Cooperation in Research and HigherEducation (STINT), and P.E. Lindahls fund.

References

[1] A. Alonso, E. Gracia-Austt, Neuronal sources of theta rhythm in the entorhinal cortex of the rat.II. Phase relations between unit discharges and theta "eld potentials, Exp. Brain Res. 67 (1987)502.

[2] A. Alonso, C. Kohler, A study of the reciprocal connections between the septum and the entorhinalarea using anterograde and retrograde axonal transport methods in the rat brain, J. Comp. Neurol.225 (1984) 327.

[3] J.M. Bower, D. Beeman, The Book of GENESIS: Exploring Realistic Neural Models with theGEneral NEural SImulation System, Springer, New York, 1995.

[4] C.T. Dickson, A. Alonso, Muscarinic induction of synchronous population activity in the entorhinalcortex, J. Neurosci. 17 (1997) 6729.

[5] H. Eichenbaum, T. Otto, N.J. Cohen, Two functional components of the hippocampal memorysystem, Behav. Brain Sci. 17 (1994) 449.

[6] D.D. Friel, [Ca2`]ioscillations in sympathetic neurons: An experimental test of a theoretical model,

Biophys. J. 68 (1995) 1752.[7] M.E. Hasselmo, B.P. Wyble, Simulation of the e!ects of scopolamine on free recall and recognition in

a network model of the hippocampus, Behav. Brain Res. 89 (1997) 1.[8] R. Klink, A. Alonso, Muscarinic modulation of the oscillatory and repetitive "ring properties of

entorhinal cortex layer II neurons, J. Neurophysiol. 77 (1997) 1813.[9] R. Klink, A. Alonso, Ionic mechanisms of muscarinic depolarization in entorhinal cortex layer II

neurons, J. Neurophysiol. 77 (1997) 1829.[10] V.S. Sohal, M.E. Hasselmo, Changes in GABA

Bmodulation during a theta cycle may be analogous to

the fall of temperature during annealing, Neural Comp. 10 (1998) 869.

E. Franse&n et al. / Neurocomputing 26}27 (1999) 375}380 379

[11] G.V. Wallenstein, M.E. Hasselmo, GABAergic modulation of hippocampal population activity:Sequence learning, place "eld development and the phase precession e!ect, J. Neurophysiol. 78 (1997)393.

[12] B.J. Young, T. Otto, G.D. Fox, H. Eichenbaum, Memory representation within the parahippocampalregion, J. Neurosci. 17 (1997) 5183.

Erik Franse2 n is an assistant professor at the Department of Numerical Analysisand Computing Science at the Royal Institute of Technology, Stockholm, Sweden.He was a post doctoral fellow at the Department of Psychology at HarvardUniversity during 1997}1998. He received his Ph.D. in Computer Science from theRoyal Institute of Technology, Stockholm, Sweden in 1996. He received his B.S. inPhysics from Uppsala University, Sweden in 1987. His research interests includemathematical modeling and biophysical simulation of neurons and networks ofneurons. He studies the ionic basis of cellular and network properties and thefunctional role of neuromodulators.

380 E. Franse&n et al. / Neurocomputing 26}27 (1999) 375}380