Persistent gamma oscillations in superficial layers of rat auditory neocortex: experiment and model

Transcription

1 J Physiol (2005) pp SYMPOSIUM REPORT Persistent gamma oscillations in superficial layers of rat auditory neocortex: experiment and model Roger D.Traub 1,Andrea Bibbig 1, Fiona E. N. LeBeau 2,Mark O. Cunningham 2 and Miles A. Whittington 2 1 Departments of Physiology and Pharmacology, and of Neurology, SUNY Downstate Medical Center, 450 Clarkson Avenue, Brooklyn, NY 11203, USA 2 School of Biomedical Sciences, Worsley Building, University of Leeds, Leeds LS2 9JT, UK Persistent in vitro gamma oscillations, induced by bath application of carbachol and kainate (amongst other drugs), were discovered by Eberhard Buhl and collaborators in The oscillations are robust, in that they can continue for hours; but the oscillations are also intricate in their mechanisms: they depend upon phasic synaptic excitation and inhibition, upon electrical coupling between interneurones and between pyramidal neurones, and at least in neocortex they depend upon complex intrinsic properties of some of the neurones. (Received 27 August 2004; accepted after revision 8 October 2004; first published online 15 October 2004) Corresponding author R. D. Traub: Departments of Physiology and Pharmacology, and of Neurology, SUNY Downstate Medical Center, 450 Clarkson Ave., Brooklyn, NY 11203, USA. roger.traub@downstate.edu Persistent, pharmacologically induced, gamma (30 70 Hz) oscillations were first described in hippocampal slices, bathed in carbachol, by Fisahn et al. in Eberhard Buhl was a major force in the discovery of persistent gamma, and in all subsequent efforts to untangle the mechanisms. Since 1998, in vitro persistent gamma oscillations have been observed with induction by other agonists, such as kainate (Hormuzdi et al. 2001), and in other brain regions: entorhinal cortex (Cunningham et al. 2003a) and neocortex (Buhl et al. 1998; Cunningham et al. 2004a). Persistent oscillations are of interest for two reasons: functionally, these oscillations may be an accurate experimental model for the study of in vivo ongoing gamma oscillations (Steriade et al. 1995; Penttonen et al. 1998); and mechanistically because, while they are gated by GABA A receptor-mediated IPSPs, the oscillations nevertheless are associated with phasic EPSPs in interneurones (Fisahn et al. 1998), and are dependent upon gap junctions (Traub et al. 2000; Pais et al. 2003). Furthermore, the requisite gap junctions necessary for the oscillations do not appear to be those between interneurones: gamma oscillations continue to exist in connexin36 knockout mice, albeit at reduced power (Hormuzdi et al. 2001). Indirect evidence suggests that the requisite gap junctions produce electrical coupling between the axons of principal This report is dedicated to the memory of Eberhard H. Buhl. It was presented at The Journal of Physiology Symposium in honour of the late Eberhard H. Buhl on Structure/Function Correlates in Neurons and Networks, Leeds, UK, 10 September It was commissioned by the Editorial Board and reflects the views of the authors. neurones (Schmitz et al. 2001): hippocampal, entorhinal cortical (Cunningham et al. 2004b) and neocortical pyramidal neurones (Cunningham et al. 2004a); and probably also entorhinal and cortical spiny stellate neurones (Gutnick et al. 1985). We shall briefly review the evidence for a critical role of axonal electrical coupling in producing the oscillations, and then speculate on how gap junctions might be involved in the function of the oscillations whatever that function might be. We shall also consider the contribution of the intrinsic properties of a special class of cortical neurones fast rhythmic bursting (FRB) neurones (Steriade et al. 1998). Persistent gamma (30 70 Hz) oscillations actually consist of an interplay between gamma and very fast oscillations (> 70 Hz) Power spectra of extracellular fields of hippocampal gamma oscillations show peaks at frequencies of 70 Hz and above (Traub et al. 2001; Pais et al. 2003). That this peak is not simply a harmonic of the gamma peak is shown by persistence of the very fast oscillations (VFOs) after blockade of phasic synaptic transmission (Traub et al. 2001). In addition, use of Fourier spectra with a sliding time window shows that the oscillation structure consists of brief bursts of VFOs separated by gamma periods of tens of milliseconds (Traub et al. 2003c). A surgical cut in the slice preparation can produce a small piece of tissue that generates VFOs alone (see below). These data suggest that persistent gamma results from the interplay of two types of oscillations: VFO, which may depend on non-synaptic mechanisms; and gamma DOI: /jphysiol

2 4 R.D.Traub and others J Physiol oscillation mediated by recurrent synaptic inhibition. That synaptic inhibition must be involved in persistent gamma is indicated by the slowing of the oscillation frequency produced by barbiturate drugs (Fisahn et al. 1998). VFO can be generated through electrical coupling between principal neurones, without chemical synapses That VFO is produced by mechanisms other than conventional chemical synapses was indicated by the data of Draguhn et al. (1998): spontaneous 200 Hz ripples in hippocampal slices persist and even are potentiated in low calcium media in which evoked synaptic potentials are blocked. That electrical coupling is involved was indicated by the block of the oscillations by halothane, octanol and carbenoxolone, and by their potentiation in an alkaline bathing medium (expected to further open gap junctions) (Draguhn et al. 1998). The extracellular potentials during in vitro ripples were fractions of a millivolt in amplitude, too small for electric field interactions to be plausibly considered. The electrical coupling that generates VFO occurs between principal cell axons The coupling potentials (spikelets) occurring during in vitro hippocampal ripples appear to be too fast to be low-pass filtered versions of action potentials, such as would be expected with conventional (somatic or dendritic) gap junctional coupling; in addition, action potentials during the ripples appeared to be antidromic, as they were virtually always preceded by fast prepotentials (spikelets) or had pronounced inflections on the rising phase (Draguhn et al. 1998). It was therefore hypothesized that the electrical coupling actually took place between axons (rather than somata or dendrites), with spikelets or antidromic spikes appearing due to the decremental, partially decremental, or non-decremental antidromic conduction from the axonal coupling site aformofconductionwhichrequires active membrane Figure 1. Intrinsic properties of layer 2/3 pyramidal cells Increasing the density of persistent g Na,ordecreasing the density of the fast voltage- and calcium-gated K + conductance g K(C), can convert a regular spiking (RS) cell into a fast rhythmic bursting (FRB) cell. Depolarization is also required for FRB behaviour. IbTx: iberiotoxin, a blocker g K(C) ; phen: phenytoin, a blocker of persistent g Na ; SNAP: S-nitroso-N-acetylpenicillamine, a NO source that enhances persistent g Na. A, model prediction that decreasing g K(C) brings out fast rhythmic bursting (FRB). B and C, iberiotoxin (50 nm) induces FRB that is resistant to phenytoin (120 µm), but is further enhanced by SNAP (100 µm). D, iberiotoxin broadens the spike, enhances spike afterdepolarization (leading into a second spike) and decreases the initial spike afterhyperpolarization ( ). Phenytoin has little effect on these actions. From Traub et al. (2003a), used with permission of the American Physiological Society.

3 J Physiol Gamma oscillations in auditory neocortex 5 properties. Schmitz et al. (2001) provided electrophysiological evidence consistent with this idea: spikelets induced by distal axonal stimulation were conducted actively, could follow stimulation 1 : 1 at frequencies > 200 Hz, were attenuated by carbenoxolone and intracellular acidification, and were indeed conducted antidromically. In addition, Schmitz et al. (2001) showed, in four cases, dye coupling between pyramidal neurones, with the coupled structures appearing to be axons by light microscopic criteria. Simulation studies then showed (Traub et al. 1999) that a network of pyramidal neurones, electrically coupled through their axons with sparse connectivity, could indeed generate VFO without chemical synapses. Persistent gamma: principal cell axonal plexus VFO produces high frequency EPSPs in interneurones. Synaptic inhibition shuts off the axonal plexus, and the cycle repeats Such a scheme for the generation of persistent gamma was proposed by Traub et al. (2000) in order to explain persistent gamma induced by carbachol. The experimental issues that were in need of explanation were these: interneurones in carbachol are not especially depolarized but fire at near gamma frequency. In addition, the oscillation collapsed when phasic synaptic excitation was blocked (Fisahn et al. 1998), suggesting that interneurones fired mainly because of phasic EPSPs. In spite of this, pyramidal cell somata fire rarely during the oscillation. By postulating the existence of a pyramidal cell axonal plexus, capable of high-frequency oscillations on its own, but susceptible to perisomatic inhibition on pyramidal cells, we were able to explain the existence of numerous large phasic EPSPs in interneurones, simultaneously with sparse pyramidal cell somatic firing. We also could explain the requirement for gap junctions in the oscillation, as the axonal plexus oscillation falls apart without electrical coupling between the axons. Direct evidence for two predictions of this model has been provided recently (Traub et al. 2003b). First, CA1 stratum oriens can produce a continuous > 100 Hz oscillation, in kainate, after the cell bodies of pyramidal cells have been cut away. (Note that axons and axon collaterals of CA1 pyramidal cells lie in s. oriens.) In addition, EPSPs in interneurones have fast, multiphasic components, with an appearance strikingly similar to the model. Fast rhythmic bursting can arise in superficial cortical pyramidal cells by alterations in one or more intrinsic conductances, combined with depolarization Neocortical persistent gamma is facilitated by fast rhythmic bursting (FRB) neurones; it is therefore necessary to say a bit about the membrane physiology underlying FRB behaviour. FRB firing behaviour ( chattering ) has been described in cat cortical neurones in response to visual stimulation, and to injection of Figure 2. Persistent kainate-induced gamma oscillations in rat auditory cortex in vitro: experiment and model A, gamma oscillations have maximal amplitude in layer 3. B, comparison of firing patterns of different cell types in network model and cortical slice. A generally subthreshold oscillation (about 30 Hz) is present in all cell types. RS cells fire intermittently; FRB cells fire 1 3 or more spikes on about half the waves, FS (fast-spiking) interneurones fire on more than half the waves, and LTS (low threshold spiking) interneurones fire intermittently. Scale bars: 25 mv, 200 ms. From Cunningham et al. (2004a).

4 6 R.D.Traub and others J Physiol strong depolarizing currents (Gray & McCormick, 1996; Steriade et al. 1998). Brumberg et al. (2000) observed FRB firing in ferret cortex in vitro, which was facilitated by persistent g Na. FRB neurones are more difficult to find in rat cortex, constituting just a few per cent of the pyramidal neurones with kainate in the bath (Cunningham et al. 2004a), but in normal medium can be observed when persistent g Na is pharmocologically enhanced. Interestingly, FRB behaviour is facilitated both by persistent g Na and also by reduction of a fast voltage- and calcium-dependent K + conductance (Fig. 1 Traub et al. 2003a). Persistent gamma in cortex is favoured by the presence of a small number of FRB neurones, which inject bursts of action potentials into the axonal plexus For the most part, persistent gamma in slices of neocortex (rat auditory neocortex), at least in superficial cortical layers, has an appearance (Fig. 2) similar to that of persistent gamma in hippocampus (Fisahn et al. 1998; Hormuzdi et al. 2001; Traub et al. 2003c) and entorhinal cortex (Cunningham et al. 2003, 2004b): there is a prominent gamma oscillation in the field. Figure 3. Spikelets during persistent gamma oscillations, and molecular evidence for the presence of pannexin-2 in the cortex A, spikelets (small, action potential-like events) occur in the network model and experimentally. Spikelets can be confirmed to be of axonal origin in the model by examining axonal and somatic potentials. Experimentally, spikelets can be differentiated from EPSPs because they are faster, and because spikelets increase in amplitude and frequency with depolarization of the cell. (Depolarization is expected to decrease the amplitude of EPSPs, but depolarization should favour spikelets, increasing their amplitude and frequency, because spikelets depend on sodium electrogenesis (Schmitz et al. 2001) provided the depolarization is not so large as to convert all axonal spikes into full somatic antidromic spikes.) Scale bars: 10 mv (model), 5 mv (experiment), 100 ms (40 ms in Inset). B, gamma power in the network model increases with the number of FRB cells, with a threshold of less than 5% FRB. C, pannexin-2 mrna staining in cortical layers 2 6, by non-radioactive in situ hybridization. Upper: low magnification, scale bar 0.2 mm; lower: high magnification, scale bar 10 µm. (Pannexin-2 is an hypothesized constituent of putative axonal gap junctions.) From Cunningham et al. (2004a).

5 J Physiol Gamma oscillations in auditory neocortex 7 Regular spiking glutamatergic cells show an obvious sawtooth-like oscillation of membrane potential, mostly subthreshold, but with occasional action potentials; while pharmacological blockade of either AMPA receptors, or of GABA A receptors, or of gap junctions, all produce significant reductions in gamma power. The most important gap junctions for generating the oscillations appear to be those hypothesized to exist between principal cell axons (Traub et al. 1999; Schmitz et al. 2001), with dendritic gap junctions between interneurones playing more of a modulatory role (Hormuzdi et al. 2001; Traub et al. 2003b). In addition to the pharmacological similarities, principal neurones in both neocortex (Fig. 3A) and in entorhinal cortex (Cunningham et al. 2004b) show, during kainate-induced gamma, spikelets: small sharp potentials resembling miniature action potentials, and presumed to be of antidromic (i.e. axonal) origin. (The entorhinal cortex is especially rich in spikelets, with both stellate cells and pyramidal cells in the superficial layers exhibiting these events.) The occurrence of spikelets is favoured by the hypothesized axonal gap junctions (Schmitz et al. 2001), and it is interesting that there is mrna staining for pannexin-2 (a candidate constituent of principal cell axonal gap junctions) throughout cortical layers 2 6 (Fig. 3C). Neocortex is different from hippocampus and entorhinal cortex, however, in possessing at least some FRB neurones. These neurones have a different firing pattern during gamma oscillations than do regular spiking (RS) neurones: they fire one to several action potentials on about half the gamma waves (Cunningham et al. 2004a). Network modelling predicts that the presence of a small number of FRB neurones greatly facilitates gamma oscillations in cortex (Fig. 3B), by injecting action potentials into the electrically coupled plexus of principal cell axons. A pharmacological experiment is consistent with this prediction: phenytoin in the bath converts FRB behaviour into RS firing behaviour (Cunningham et al. 2004a). Concomitantly, phenytoin eliminates kainate-induced gamma in auditory cortex, but has no effect on kainate-induced gamma in hippocampus, a structure that appears to lack FRB neurones (Cunningham et al. 2004a). The necessity of FRB neurones in cortex, in order to generate persistent gamma, occurs in the model because of the limited gap junctional connectivity and RS cell excitability that we used in our neocortical network model. It is interesting to speculate that FRB firing behaviour may be a neocortical specialization that arose in order to ensure the robustness of persistent gamma. Discussion Persistent gamma oscillations occur in vivo in at least two circumstances. First, during the theta state in rats (associated with theta EEG oscillations in limbic cortices, and occurring during locomotion, REM sleep, and certain types of anaesthesia), gamma oscillations appear superimposed on the theta waves, in hippocampus and entorhinal cortex (Penttonen et al. 1998; Chrobak & Buzsáki, 1998). Second, in cat neocortex, during the slow (< 1 Hz) oscillation of sleep, short runs of gamma occur during the up states (Steriade et al. 1995, 1996). This latter sort of gamma is not truly persistent, occurring rather in brief runs that, however, keep repeating during prolonged episodes of slow wave sleep. As there appears to be metabotropic activation of at least some neurones during slow wave sleep (in relay nuclei and reticular nucleus of the thalamus, at any rate) (Hughes et al. 2002, 2004), it is possible that similar mechanisms are at work during slow wave sleep as during persistent pharmacological gamma. What might be the function of gamma oscillations during sleep? Wolf Singer and colleagues (reviewed in Gray, 1994) have proposed that the transient (hundreds of milliseconds to 1 s or so) gamma oscillations, occurring after a sensory stimulus, could serve to encode the binding together of two cortical regions both activated by the stimulus and bound together by synchronization of the respective oscillations. It is difficult, to envision what percepts or stimulus features would be bound during slow wave sleep. We suggest, however, that conditions during persistent gamma are such that synaptic plasticity could be occurring continuously. The reason has to do with spontaneous activity in the axonal plexus, amplified by axonal electrical coupling (Traub et al. 2003b,c) (Fig. 4). Because of such plexus activity, principal neurones will be bombarded, on every gamma cycle, both with excitation arriving at presynaptic glutamatergic terminals, and with antidromic activation. Should the Figure 4. A possible functional significance of persistent gamma oscillations Persistent gamma oscillations allow synapses on pyramidal cells to be near-simultaneously excited both presynaptically and postsynaptically, perhaps setting the stage for long-lasting synaptic plasticity. Key: in a network with synaptic and gap junction (gj) connections, a spike (say initiating at ) can reach a dendritic site in a different cell (blue cross) by 2 pathways: orthodromically (brown arrow), and via the axonal gj followed by antidromic conduction and then back propagation into the dendrite (green arrow). Such convergent activations of the same dendritic site synaptic and with a back propagating spike could lead to synaptic plasticity.

6 8 R.D.Traub and others J Physiol antidromic activation cause somatic firing, then synapses will also see backpropagating dendritic activation, i.e. postsynaptic dendritic excitation that is near-simultaneous with the presynaptic activation. Thus, one would expect synaptic plasticity to be taking place continuously during a persistent gamma oscillation. Whether synapses are altered in a pattern-specific way or not cannot be predicted. Conceivably, the plasticity could be serving a trophic function. References Brumberg JC, Nowak LG & McCormick DA (2000). Ionic mechanisms underlying repetitive high-frequency burst firing in supragranular cortical neurons. JNeurosci20, Buhl EH, Tamás G & Fisahn A (1998). Cholinergic activation and tonic excitation induce persistent gamma oscillations in mouse somatosensory cortex in vitro. JPhysiol 513, Chrobak JJ & Buzsáki G (1998). Gamma oscillations in the entorhinal cortex of the freely behaving rat. JNeurosci18, Cunningham MO, Davies CH, Buhl EH, Kopell N & Whittington MA (2003). Gamma oscillations induced by kainate receptor activation in the entorhinal cortex in vitro. JNeurosci23, Cunningham MO, Halliday DM, Davies CH, Traub RD, Buhl EH & Whittington MA (2004b). Coexistence of gamma and high-frequency oscillations in the medial entorhinal cortex in vitro. JPhysiol 559, Cunningham MO, Whittington MA, Bibbig A, Roopun A, LeBeau FEN, Vogt A, Monyer H, Buhl EH & Traub RD (2004a). A role for fast rhythmic bursting neurons in cortical gamma oscillations in vitro. Proc Natl Acad Sci USA101, Draguhn A, Traub RD, Schmitz D & Jefferys JGR (1998). Electrical coupling underlies high-frequency oscillations in the hippocampus in vitro. Nature 394, Fisahn A, Pike FG, Buhl EH & Paulsen O (1998). Cholinergic induction of network oscillations at 40 Hz in the hippocampus in vitro. Nature 394, Gray CM (1994). Synchronous oscillations in neuronal systems: mechanisms and functions. JComput Neurosci 1, Gray CM & McCormick DA (1996). Chattering cells: superficial pyramidal neurons contributing to the generation of synchronous oscillations in the visual cortex. Science 274, Gutnick MJ, Lobel-Yaakov R & Rimon G (1985). Incidence of neuronal dye-coupling in neocortical slices depends on the plane of section. Neuroscience 15, Hormuzdi SG, Pais I, LeBeau FEN, Towers SK, Rozov A, Buhl EH, Whittington MA & Monyer H (2001). Impaired electrical signaling disrupts gamma frequency oscillations in connexin 36-deficient mice. Neuron 31, Hughes SW, Cope DW, Blethyn KL & Crunelli V (2002). Cellular mechanisms of the slow (<1 Hz) oscillation in thalamocortical neurons in vitro. Neuron 33, Hughes SW, Lörincz M, Cope DW, Blethyn KL, Kékesi KA, Parri HR, JuhászG&CrunelliV(2004). Synchronized oscillations at α and θ frequencies in the lateral geniculate nucleus. Neuron 42, Pais I, Hormuzdi SG, Monyer H, Traub RD, Wood IC, Buhl EH, Whittington MA & LeBeau FEN (2003). Sharp wave-like activity in the hippocampus in vitro in mice lacking the gap junction protein connexin 36. JNeurophysiol 89, Penttonen M, Kamondi A, Acsády L & Buzsáki G (1998). Gamma frequency oscillation in the hippocampus: intracellular analysis in vivo. Eur J Neurosci 10, Schmitz D, Schuchmann S, Fisahn A, Draguhn A, Buhl EH, Petrasch-Parwez RE, Dermietzel R, Heinemann U & Traub RD (2001). Axo-axonal coupling: a novel mechanism for ultrafast neuronal communication. Neuron 31, Steriade M, Amzica F & Contreras D (1995). Synchronization of fast (30 40 Hz) spontaneous cortical rhythms during brain activation. JNeurosci16, Steriade M, Contreras D, Amzica F & Timofeev I (1996). Synchronization of fast (30 40 Hz) spontaneous oscillations in intrathalamic and thalamocortical networks. JNeurosci 16, Steriade M, Timofeev I, Dürmüller N & Grenier F (1998). Dynamic properties of corticothalamic neurons and local cortical interneurons generating fast rhythmic (30 40 Hz) spike bursts. JNeurophysiol 79, Traub RD, Bibbig A, Fisahn A, LeBeau FEN, Whittington MA &Buhl EH (2000). A model of gamma-frequency network oscillations induced in the rat CA3 region by carbachol in vitro. Eur J Neurosci 12, Traub RD, Buhl EH, Gloveli T & Whittington MA (2003a). Fast rhythmic bursting can be induced in layer 2/3 cortical neurons by enhancing persistent Na + conductance or by blocking BK channels. JNeurophysiol 89, Traub RD, Cunningham MO, Gloveli T, LeBeau FEN, Bibbig A, Buhl EH & Whittington MA (2003c). GABA-enhanced collective behavior in neuronal axons underlies persistent gamma-frequency oscillations. Proc Natl Acad SciUSA100, Traub RD, Pais I, Bibbig A, LeBeau FEN, Buhl EH, Hormuzdi SG, Monyer H& Whittington MA (2003b). Contrasting roles of axonal (pyramidal cell) and dendritic (interneuron) electrical coupling in the generation of gamma oscillations in the hippocampus in vitro. ProcNatlAcadSciUSA100, Traub RD, Schmitz D, Jefferys JGR & Draguhn A (1999). High-frequency population oscillations are predicted to occur in hippocampal pyramidal neuronal networks interconnected by axoaxonal gap junctions. Neuroscience 92, Traub RD, Whittington MA, Buhl EH, Le Beau FEN, Boys S, Cross H & Baldeweg T (2001). A possible role for gap junctions in generation of very fast EEG oscillations preceding the onset of, and perhaps initiating, seizures. Epilepsia 42, Acknowledgements This work was supported by the N.I.H. (NINDS), the Volkswagen Stiftung, the Wellcome Trust, and the Medical Research Council (UK). We thank our close colleagues, including (but not limited to) Dr Diego Contreras, Professor Nancy Kopell, and Professor Hannah Monyer. We thank Dr Robert Walkup of IBM Corp. for assistance with parallel programming.