Abstract
Gamma frequency (30–80 Hz) oscillations are recordable from human and rodent entorhinal cortex. A number of mechanisms used by neuronal networks to generate such oscillations in the hippocampus have been characterized. However, it is as yet unclear as to whether these mechanisms apply to other anatomically disparate brain regions. Here we show that the medial entorhinal cortex (mEC) in isolation in vitro generates gamma frequency oscillations in response to kainate receptor agonists. Oscillations had the same horizontal and laminar spatiotemporal distribution as seen in vivo and in the isolated whole-brain preparation. Oscillations occurred in the absence of input from the hippocampal formation and did not spread to lateral entorhinal regions. Pharmacological similarities existed between oscillations in the hippocampus and mEC in that the latter were also sensitive to GABAA receptor blockade, barbiturates, AMPA receptor blockade, and reduction in gap junctional conductance. Stellate and pyramidal neuron recordings revealed a large GABAergic input consisting of gamma frequency IPSP trains. Fast spiking interneurons in the superficial mEC generated action potentials at gamma frequencies phase locked to the local field. Stellate cells also demonstrated a subthreshold membrane potential oscillation at theta frequencies that was temporally correlated with a theta-frequency modulation in field gamma power. Disruption in this stellate theta frequency oscillation by the hyperpolarisation activated current (Ih) blocker ZD7288 also disrupted theta modulation of field gamma frequency oscillations. We propose that similar cellular and network mechanisms to those seen in the hippocampus generate and modulate persistent gamma oscillations in the entorhinal cortex.
Introduction
Cortical gamma band (30–80 Hz) oscillations are thought to grant large ensembles of neurons the ability to engage in associative binding, in particular during processing of incoming sensory signals (Gray et al., 1989; Singer and Gray, 1995) (for review, see Verala et al., 2001). Thus, gamma activity has been recorded in sensorimotor (Murthy and Fetz, 1996), auditory (Barth and MacDonald, 1996), and visual cortices (Roelfsema et al., 1997). Gamma frequency oscillations have been reported in the entorhinal cortex (EC) of humans during wakefulness (Hirai et al., 1999; Uchida et al., 2001), as well as in rodents in vivo (Charpak et al., 1995; Chrobak and Buzsáki, 1998) and in an isolated whole-brain preparation (van Der Linden et al., 1999; Dickson et al., 2000b).
The EC occupies a unique position in the temporal lobe, in that it acts as an interface between neocortical regions and the cornu ammonuis, as well as other associated limbic structures such as the subiculum and dentate gyrus (Van Hoesen, 1982; Witter et al., 1989; Burwell and Witter, 2002). Superficial neurons of the EC supply the main neocortical input to the hippocampus via the perforant pathway (Steward and Scoville, 1976; Witter and Groenewegen, 1984; Amaral and Witter, 1989), whereas the deep layer neurons receive output from the hippocampus via CA1 and the subiculum (Harris et al., 2001; Naber et al., 2001). In this respect, the EC occupies a pivotal position in the process of gating information flow along the neo-archicortical axis. Hence, the EC is assumed to act as a supermodal associational cortical region, in which inputs from a number of sensory areas can assemble (Van Hoesen, 1982; Amaral and Witter, 1989; Witter et al., 1989) and engage in a functional interplay of sensory information with the hippocampus (Fernandez et al., 1999; Fell et al., 2001). The importance of the EC is further emphasized by the fact that it is an early and selective site of pathological changes that occur in disorders of memory function such as Alzhemier's disease (Van Hoesen, 1982; Hyman et al., 1984; Van Hoesen et al., 1986; Braak and Braak, 1991) and schizophrenia (Beckmann and Senitz, 2002).
Gamma oscillations have been described in a number of in vitro slice preparations. In a number of these, metabotropic receptor activation has induced this activity (Whittington et al., 1995; Fisahn et al., 1998) or application of nanomolar concentrations of kainate (Hájos et al., 2000; Hormuzdi et al., 2001). However, attempts to elicit gamma activity in an in vitro preparation of the EC with application of the metabotropic cholinergic agonist carbachol occasionally resulted in the generation of epileptiform activity (Dickson and Alonso, 1997; Gloveli et al., 1999). Here we present data that demonstrates that the activation of kainate receptors can generate persistent gamma activity in vitro in the EC. This activity was associated with sparse activation of principal cell soma but large-scale phasic interneuronal output. Population gamma frequency activity was associated with trains of IPSPs in a similar manner to that seen in models of hippocampal gamma oscillation.
Parts of this paper have been published previously in abstract form (Cunningham et al., 2002a,b).
Materials and Methods
Preparation of slices. Transverse EC hippocampal slices (450 μm) were prepared from adult Sprague Dawley rats, anesthetized with inhaled isoflurance, immediately followed by an intramuscular injection of ketamine (≥100 mg/kg) and xylazine (≥10 mg/kg). Animals were intracardically perfused with ∼50 ml of modified artificial CSF (ACSF), which was composed of the following (in mm): 252 sucrose, 3 KCl, 1.25 NaH2PO4, 24 NaHCO3, 2 MgSO4, 2 CaCl2, and 10 glucose. All salts were obtained from BDH Chemicals (Poole, UK). The brain was removed and submerged in cold (4–5°C) ACSF during dissection, and horizontal slices were cut using a Campden Instruments (Loughborough, UK) vibroslice. Slices were then transferred either to a holding chamber or directly to a recording chamber. Here, they were maintained at 34°C at the interface between a continuous stream (1.2 ml/min) of ACSF (in mm: 126 NaCl, 3 KCl, 1.25 NaH2PO4, 24 NaHCO3, 2 MgSO4, 2 CaCl2, and 10 glucose) and warm, moist carbogen gas (95% O2–5% CO2). Slices were permitted to equilibrate for 45 min before any recordings commenced.
Drugs. All drugs were bath applied at known concentrations: kainic acid (2S,3S,4R)-carboxy-4-(1methylethenyl)-3-pyrrolidineacetic acid), 200–400 nm; NBQX (2,3-dioxo-6-nitro-1,2,3,4-tetrehydrobenzo[f]qui noxaline-7-sulfonamide), 20 μm; carbachol, 20–50 μm; domoic acid, ([2S-[2a,3b,4b(1Z,3E,5R,]]-2-carboxy-4-(-5-carboxy-1-methyl-1,3-hexadienyl)-3-pyrrolidineacetic acid), 100 nm; bicuculline methochloride, 2 μm; and ZD7288 (4-ethylphenylamino-1,2-dimethyl-6-methylaminopyridinium chloride), 10 μm. All were obtained from Tocris Cookson (Bristol, UK). Pentobarbital (20 μm) and carbenoxenlone (100 μm) were obtained from Sigma (Poole, UK).
Recording, data acquisition, and analysis. Extracellular recordings electrodes were pulled from borosilicate glass (Harvard Apparatus, Kent, UK) filled with ACSF and had resistances in the range of 2–5MΩ. For the purpose of intracellular recordings, electrodes were pulled from borosilicate glass filled with KCH3SO4 and had resistances in the range of 70–130 MΩ. Three types of neuron were recorded from in the entorhinal cortex, each identified by the following specific electrophysiological parameters. Stellate cells had pronounced sag observed with hyperpolarizing current step, with action potentials present on rebound depolarization. Longduration depolarizing current steps (500–800 msec) revealed action potentials of stellates cells had a fast afterhyperpolarization (fAHP), which was followed by a depolarizing after potential (DAP), and action potential had half-widths >1 msec. There was also accommodation of action potentials. During tonic depolarization, stellate cells exhibited a subthreshold membrane oscillation. Pyramidal cells displayed little or no sag in response to hyperpolarizing current, and a rebound depolarization was not observed. No subthreshold membrane oscillation was apparent. Depolarizing current revealed that action potentials of pyramidal cells displayed an fAHP but no DAP and had half-widths <1.5 msec. Action potential accommodation was also observed (van Der Linden and Lopes de Silva, 1998). Maximum firing frequencies on current injection were >400 Hz, there was no action potential accommodation, and action potential half-widths were <0.6 msec. Peak frequency and power values were obtained from power spectra generated with Fourier analysis in the Axograph software package (Axon Instruments, Foster City, CA). Power for a given frequency band was determined as the area under the peak in the power spectra between 20 and 80 Hz for gamma frequency oscillations. All values are given as mean ± SE. Power spectra were constructed offline from digitized data (digitization frequency, 10 kHz) using a 60 sec epoch of recorded activity. MatLab (MathWorks, Natick, MA) was used to generate spectrograms. The kinetics of IPSPs were measured using MiniAnalysis (Jaejin Software, Leonia, NJ), and >200 IPSPs were obtained per slice and pooled for additional analysis.
Results
Kainate receptor agonists induced gamma activity in the medial EC
When slices of EC were bath perfused with nanomolar concentrations of kainic acid (200–400 nm), rhythmic (50–200 μV) field activity was observed in all slices studied (n = 36) (Fig. 1A). Activity in the superficial and deep layers had a mean frequency of 46.4 ± 1.9 and 45.4 ± 2.0 Hz, respectively, whereas the mean area power was 622.0 ± 116.6 and 215.9 ± 43.6 μV2 sec-1, respectively. In all cases, the power of the oscillation was significantly greater in superficial layers than with that in the deep layers (p < 0.001) (Fig. 1A), although there was no significant difference in mean frequencies (p < 0.1). The activity between superficial and deep layers of the EC was in anti-phase (Fig. 1A, inset). Once initiated, the network activity was stable for many hours (∼4–5 hr). In addition, the more potent kainate receptor agonist domoic acid (domoate) (100 nm) was also able to elicit network activity in both layers of the EC (n = 5) (Fig. 1B). The activity in superficial and deep layers had a mean frequency of 48.3 ± 3.9 and 48.3 ± 3.5 Hz (n = 5) and a mean area power of 828.4 ± 383.8 and 415.0 ± 171.3 μV2 sec-1, respectively. A cross-correlogram of the activity produced by domoate revealed, in agreement with kainate experiments, that a sharp 180° phase reversal was seen when comparing deep with superficial layers (see Fig. 3).
Spatial profile of gamma activity in the EC
Previously, Fisahn (1999) has demonstrated that the bath application of kainate can generate gamma activity in the hippocampus in vitro. When 400 nm kainate was bath applied to the intact slice preparation (containing both entorhinal cortex and hippocampus), gamma activity was observed in stratum pyramidale in both CA3 and CA1 in this present study. The gamma activity had a significantly lower mean frequency [CA3, 32.6 ± 1.8 Hz; CA1, 33.7 ± 1.1 Hz; n = 9; p < 0.05 compared with medial entorhinal cortex (mEC) data above]. The EC and the hippocampus have extensive and robust reciprocal connections (Witter et al., 1986). Therefore, to determine whether the EC was capable of generating gamma activity independent of the hippocampus, we performed a number of microlesions. Control activity in the EC was recorded before subsequent microlesions. The control activity had a mean area power of 537.0 ± 122.5 (superficial) and 310.7 ± 68.6 μV2 sec-1 (deep). Lesion of area CA3 from the rest of the slice produced no significant change in mean area power in the mEC (superficial, 408.6 ± 118.0 μV2 sec-1; deep, 232.6 ± 77.6 μV2 sec-1; p > 0.05 compared with prelesion values; n = 6). Additional lesion of area CA1 from the mEC recording sites had no significant effect on mean area power (superficial, 371.2 ± 107.1; deep, 238.8 ± 83.8 μV2 sec-1; p > 0.05; n = 6; data not shown).
In vivo and whole-brain studies also demonstrated that entorhinal gamma activity was spatially localized across the horizontal plane of the mEC (van Der Linden et al., 1999). To test this in vitro, a reference electrode was orientated at the most medial portion of the mEC, and a second electrode was then moved in 50 μm steps toward the lateral EC. In the superficial layers, the amplitude and area power of gamma activity was consistent across 300 μm of the medial portion of the EC and sharply diminished on reaching the lateral EC (Fig. 2A). In the deep layers, gamma activity peaked in the medial portion and slowly diminished as the electrode was moved laterally (Fig. 2A). No change in peak frequency of kainate-induced oscillation was seen along the longitudinal axis of EC in either deep or superficial layers. Direct comparison of power in the gamma frequency range between mEC and lateral EC showed a significant reduction in oscillations in the lateral part of EC (p < 0.05; n = 6) (Fig. 2B).
In the isolated guinea pig whole-brain preparation (Chrobak and Buzsáki, 1998; van Der Linden et al., 1999; Dickson et al., 2000b), gamma oscillations were shown to have a marked phase reversal through the laminas. We tested this by placing one reference electrode in the subcortical white matter of the angular bundle and a second electrode close and perpendicular to the reference electrode in 100 μm steps toward the pia. Around level of the perisomatic region (i.e., between layers II and III), a distinct phase reversal was observed (Fig. 3A,B). The maximal area power of gamma was also observed around the site of phase reversal, in layer III (1130.5 ± 174.6 μV2 sec-1; n = 22). To ascertain whether this phase shift was simply attributable to a polarity change of the extracellular field or a genuine phase difference between gamma generators in deep and superficial layers, we performed joint field and intracellular IPSP train measurements. Intracellular events in layer III pyramidal neurons and extracellular activity in layer II were found to be in-phase (Fig. 4A), whereas intracellular events in layer III pyramidal neurons and field activity in layer III were in anti-phase (Fig. 4B). Intracellular events in layer II stellate cells and field activity in layer III were also in anti-phase (Fig. 4C). Thus, principal neurons on either side of the layer in which the field phase shift was seen share the same temporal pattern of inhibitory drive despite changes in the phase of the field (see Discussion).
GABAergic and glutamatergic pharmacology of gamma activity in the mEC
Gamma activity in the hippocampus is critically dependent on the rhythmic synchronous output of populations of interneurons (Whittington et al., 1995; Fisahn et al., 1998). To test whether this hypothesis also held for the entorhinal cortex, we first reduced fast GABAergic transmission. After inducing gamma activity with kainate, coapplication of a low concentration of bicuculline (2 μm) resulted in the virtual abolition of the population gamma activity (n = 7) (Fig. 5A). Prolongation of the decay kinetics of GABAA receptor-mediated IPSPs has been shown to have a marked effect on gamma oscillations driven by tonic interneuronal excitation but much less of an effect when phasic excitation is required (Whittington et al., 2000). Pentobarbital (20 μm) (n = 6) (Fig. 5B) produced a small but significant reduction (p < 0.001) in the peak frequency of gamma activity (superficial: control, 45.8 ± 2.6; pentobarbital, 37.4 ± 2.8; washout, 45.4 ± 2.4 Hz; deep: control, 43.7 ± 2.6; pentobarbital, 34.9 ± 2.6; washout, 44.8 ± 2.8 Hz). Fast phasic glutamatergic transmission has been demonstrated to influence persistent gamma activity in hippocampal and neocortical slices (Buhl et al., 1998; Fisahn et al., 1998). Thus, we investigated the possible involvement of this mechanism in mEC by blocking AMPA receptors. Addition of SYM 2206 (20 μm; n = 3), a potent noncompetitive AMPA receptor antagonist (Li et al., 1999; Behr et al., 2001), significantly reduced gamma frequency activity in both the superficial and deep layers (p < 0.05) (Fig. 6A). In superficial layers only, this near abolition of activity in the gamma band was accompanied by a small but significant increase in activity at lower frequencies (4–12 Hz range) (peak frequency: control vs SYM 2206, 44.8 ± 2.5 vs 11.6 ± 1.6 Hz; p < 0.001). In a subsequent set of experiments, the application of the mixed AMPA/kainate receptor antagonist NBQX (20 mm; n = 3), abolished all rhythmic activity in both layers (Fig. 6B).
The precise timing of phasic excitatory inputs to interneurons during gamma oscillations in the hippocampus has been proposed to involve gap junctions between principal cell axons (Traub et al., 2000). In the mEC, application of the gap junction blocker carbenoxolone (100 μm) significantly (p < 0.05; n = 6) reduced gamma power in both the superficial (control vs drug, 399.9 ± 97.2 vs 115.1 ± 38.7 μV2 sec-1) and deep (56.9 ± 10.1 vs 14.3 ± 5.1 μV2 sec-1) layers (data not shown).
Characterization of cellular and synaptic behavior during gamma activity in the mEC
Using sharp microelectrodes, we examined the intrinsic and synaptic properties of both principal excitatory neurons and putative, fast spiking inhibitory interneurons. When stellate neurons in layer II and pyramidal neurons in layer III were recorded at depolarized holding potentials (at least -40 mV) (Fig. 4), trains of large-amplitude IPSPs were seen (8.3 ± 0.2 mV; n = 5871 IPSPs from eight neurons). Spectral analysis demonstrated a peak in the gamma range (layer II vs layer III, 42.3 ± 2.3 vs 37.3 ± 1.1 Hz). With no holding current applied, stellate neurons in layer II and pyramidal neurons in layer III fired spontaneous action potentials, with significantly different (p < 0.05) mean frequencies of 0.7 ± 0.1 and 5.0 ± 0.6 Hz (n = 6) (Fig. 7A), respectively. When the membrane potential of neurons was adjusted to close to the reversal potential of IPSPs (at least -70 mV; n = 7), smallamplitude EPSPs (1.4 ± 0.01 mV; n = 5900) were observed. Analysis of these events demonstrated that they were generated over a faster frequency range than concurrent stellate–pyramidal cell output (stellate vs pyramid, 7.6 ± 0.6 vs 11.8 ± 0.7 Hz; p < 0.05) (Fig. 7B,C). Recordings from fast spiking interneurons located in layer II (n = 3) revealed that, during field gamma oscillations, these cells generated action potentials in the gamma range (45.1 ± 2.8 Hz). These interneurons received significantly larger EPSPs at faster frequencies than in the two principal neuron types studied. However, EPSPs were still present in these cells at lower frequencies than their output (EPSP frequency was 29.3 ± 2.8 Hz; compare with action potential frequency above).
The origin of theta frequency modulation of field gamma oscillations
Unlike in vivo and isolated whole-brain superficial mEC recordings, no clear peak at theta frequency was seen in power spectra (Fig. 1). However, the amplitude of field gamma activity in the superficial mEC (layers II and III) was strongly modulated at theta frequencies throughout kainate application (modulation frequency, 3.2 ± 0.2 Hz; n = 12) (Fig. 8A,B). In contrast, as observed in vivo (Chrobak and Buzsáki, 1998), field gamma frequency activity in the deep layers demonstrated no such clear modulation (Fig. 8C). This frequency modulation was significantly different from the pattern of synaptic excitation see in either principal cell type (3.2 vs 7.6 Hz; 3.2 vs 11.8 Hz; p < 0.001) (Fig. 7).
It has been shown previously that stellate cells in the EC have theta frequency subthreshold membrane potential oscillations (Alonso and Llinás, 1989; Alonso and Llinás, 1993; Jones, 1994). In this model, stellate cells demonstrated a subthreshold theta frequency membrane potential oscillation of 3.2 ± 0.1 Hz (n = 11) with a fixed phase relationship with the modulated concurrent field recordings (5.9 ± 1.1 msec; n = 11). To investigate whether this behavior in these cells was responsible for the observed modulation of gamma frequency oscillations, we recorded from stellate cells concurrently with superficial EC field recordings in the presence and absence of 10 μm ZD7288. The subthreshold membrane potential oscillation has been proposed to be, in part, mediated by the action of a hyperpolarization-activated current (Ih), which is blocked by ZD7288. Bath application of ZD7288 had no significant effect on overall field gamma power measured as area (control, 379.8 ± 134.9 μV2 sec-1; ZD7288, 324.3 ± 121.0 μV2 sec-1; p > 0.1; n = 8). However, no rhythmic pattern of modulation of field gamma frequency was seen in the presence of this drug (Fig. 9). Cross-correlograms of subthreshold stellate cell activity and the field oscillations showed a significant reduction in central peak amplitude (control, -0.05 ± 0.00; ZD7288, -0.02 ± 0.01; p < 0.05; n = 5) and a significant reduction in the peak frequency of stellate cell membrane potential oscillations from 3.2 Hz (above) to 2.0 ± 0.2 Hz (p < 0.05), with a considerable broadening of the spectral peak within the theta frequency range.
Discussion
The entorhinal cortex plays a pivotal role in transferring information to and from the neocortex and the limbic regions of the brain (Fernandez et al., 1999; Fell et al., 2001). It has been proposed to act as a major associational area, taking information from many neocortical areas (Van Hoesen, 1982; Amaral and Witter, 1989; Witter et al., 1989). Associating activity from spatially distant brain regions in a single, separate region may be seen as an extension of the binding hypothesis for associating activity in areas coding for separate features of a sensory object. The role of rhythmogenesis (particularly at gamma frequencies) may therefore be seen as a critical tool used by the entorhinal cortex to temporally organize multiple inputs of diffuse spatial origin. Different forms of gamma frequency network activity have different properties relevant to the temporal organization of local and spatially distributed firing patterns (Whittington et al., 2000). Thus, it is important to understand the mechanisms used by the entorhinal cortex to generate gamma frequency oscillations.
Gamma frequency rhythmogenesis in the hippocampal formation is becoming better understood and provides a reference point with which to compare entorhinal rhythms. The present data demonstrated that, at least superficially, the medial entorhinal cortex can generate persistent field gamma rhythms in a manner similar to the hippocampus in the presence of kainate receptor activation. The lateral entorhinal cortex did not respond in such a manner in the slice as in whole-brain studies (van Der Linden et al., 1999; Dickson et al., 2000b). There is a lack of evidence for gross anatomical and cytoarchitectonic differences between the lateral and medial entorhinal cortex (van Der Linden et al., 1999). However, the lateral entorhinal cortex does receive afferent input from different sources to the medial entorhinal cortex (Kosel et al., 1982; Swanson and Köhler 1986), further providing evidence for a functional difference between the two regions.
In vivo and whole-brain studies also demonstrated a marked 180° phase change between laminas II and III in the mEC. This observation was replicated in the slice preparation here but could not be ascribed to two independent, anti-phase, gamma generators. Instead, principal neurons on either side of the phase reversal point received temporally identical patterns of gamma frequency IPSPs. These data suggested that the phase reversal seen in field recordings may represent a source–sink interaction. The main polar neurons in the mEC are the pyramidal neurons found predominantly in layer III. Spatially discrete, perisomatic and proximal apical dendritic synaptic inputs from interneurons onto these cells, as seen for hippocampal pyramids (Papp et al., 2001), would be expected to generate the field phase relationship seen here and in the in vivo—whole-brain situation.
In the hippocampus, two main subtypes of GABA-dependent gamma frequency network activity can be seen experimentally. The first subtype, interneuron network gamma (ING), is seen transiently in response to brief periods of direct excitation of populations of interneurons (Whittington et al., 1995). It requires no phasic excitatory input from principal cells and generates a gamma frequency network output on the basis of the kinetics of interneuron–interneuron IPSPs. As such, it is exquisitely sensitive, in terms of frequency, to barbiturates and requires no gap junctional communication for local generation (Traub et al., 2001). The second type, pyramidal–interneuron network gamma (PING), is seen persistently and does require phasic synaptic excitation of interneurons via AMPA receptors (Fisahn et al., 1998; Traub et al., 2000; Hormuzdi et al., 2001). It is less sensitive to barbiturates, in terms of frequency, and requires gap junctional communication in conditions in which individual principal cell firing rates are much lower than population frequency. The gamma oscillations seen in the mEC correspond more closely to PING than to ING: they require AMPA receptors and gap junctions, and the frequency is altered only a little (but significantly) by barbiturates. However, the spike rates for layer III pyramidal cells was higher than that seen in CA3 and CA1 pyramids. Despite this, the frequency of EPSPs invading fast spiking interneurons in the superficial entorhinal cortex was significantly lower than the spike frequency in these cells. The interneurons generated an output that matched the field gamma frequency, suggesting that some action potentials were generated as a consequence of tonic excitatory drive and not through the phasic drive from population EPSPs, a situation reminiscent of ING models in the hippocampal formation.
Field theta oscillations are a prominent feature of in vivo recordings from the entorhinal cortex. However, these rhythms are thought to arise mainly from inputs from the medial septum/diagonal band (Mitchell et al., 1982; Dickson et al., 1994; Jeffery et al., 1995; Leranth et al., 1999), a structure absent from the slices used in the present work. However, strong-amplitude modulation of field gamma rhythms was seen at theta frequencies, suggesting that the EC can generate theta frequency oscillations intrinsically. The most likely source for this modulation comes from stellate cells in the superficial layers. No theta modulation was seen in the deep layers (Fig. 8), and stellate cells have been proposed to generate theta frequency oscillations both alone, as a consequence of intrinsic conductances, and as a network of neurons coupled by recurrent excitation (Haas and White, 2002).
Intrinsic conductances such as Ih have been shown to be involved in generating theta frequency outputs from theta generating interneurons in the hippocampus (Gillies et al., 2002), and stellate cells express large amounts of this conductance (Alonso and Llinás, 1989; Alonso and Llinás, 1993; White et al., 1995; Dickson et al., 2000a). Blockade of Ih with ZD7266 significantly disrupted both subthreshold stellate cell theta rhythms and the theta frequency modulation of field gamma amplitude. These data suggest that stellate cells may modulate gamma frequency activity in the superficial EC, in addition to providing a population theta frequency output to the hippocampal formation via the perforant path (Holsheimer et al., 1983; Boeijinga and Lopes da Silva, 1988; Heynen and Bilkey, 1994; Bragin et al., 1995; Charpak et al., 1995; Chrobak and Buzsáki, 1998).
In summary, the medial entorhinal cortex in vitro generates persistent gamma oscillations in a manner similar to that seen in the cornu ammonis. Differential involvement of pyramidal neurons and stellate cells (in terms of mean firing frequency) suggests different patterns of input to the hippocampus via the temporoammonic and perforant pathways during gamma oscillations in vivo. The greater frequency of action potential generation in pyramidal cells compared with stellate cells suggests that persistent entorhinal gamma frequency oscillations exert more of an active influence on cornu ammonis subfields directly as opposed to via the dentate gyrus. The similarities in mechanisms used by the hippocampal cornu ammonis and superficial mEC to generate gamma frequency population oscillations may facilitate the temporal organization of information passing along this axis.
Footnotes
This work was supported by the Medical Research Council (United Kingdom) and The National Institutes of Health.
Correspondence should be addressed to M. A. Whittington, School of Biomedical Sciences, The Worsley Building, University of Leeds, Leeds LS2 9NQ, UK. E-mail: m.a.whittington{at}leeds.ac.uk.
Copyright © 2003 Society for Neuroscience 0270-6474/03/239761-09$15.00/0