Gap junctions between interneuron dendrites can enhance synchrony of gamma oscillations in distributed networks

Gamma-frequency (30-70 Hz) oscillations in populations of interneurons may be of functional relevance in the brain by virtue of their ability to induce synchronous firing in principal neurons. Such a role would require that neurons, 1 mm or more apart, be able to synchronize their activity, despite the presence of axonal conduction delays and of the limited axonal spread of many interneurons. We showed previously that interneuron doublet firing can help to synchronize gamma oscillations, provided that sufficiently many pyramidal neurons are active; we also suggested that gap junctions, between the axons of principal neurons, could contribute to the long-range synchrony of gamma oscillations induced in the hippocampus by carbachol in vitro. Here we consider interneuron network gamma: that is, gamma oscillations in pharmacologically isolated networks of tonically excited interneurons, with frequency gated by mutual GABA(A) receptor-mediated IPSPs. We provide simulation and electrophysiological evidence that interneuronal gap junctions (presumably dendritic) can enhance the synchrony of such gamma oscillations, in spatially extended interneuron networks. There appears to be a sharp threshold conductance, below which the interneuron dendritic gap junctions do not exert a synchronizing role.

Iontophoresis in vivo demonstrates a key role for GABA(A) and glycinergic inhibition in shaping frequency response areas in the inferior colliculus of guinea pig

The processing of biologically important sounds depends on the analysis of their frequency content by the cochlea and the CNS. GABAergic inhibition in the inferior colliculus shapes frequency response areas in echolocating bats, but a similar role in nonspecialized mammals has been questioned. We used the powerful combination of iontophoresis with detailed analysis of frequency response areas to test the hypothesis that GABAergic and glycinergic inhibition operating in the inferior colliculus of a nonspecialized mammal (guinea pig) shape the frequency responses of neurons in this nucleus. Our analysis reveals two groups of response areas in the inferior colliculus: V-shaped and non-V-shaped. The response as a function of level in neurons with V-shaped response areas can be either monotonic or nonmonotonic. Application of bicuculline or strychnine in these neurons, to block inhibition mediated by GABA(A) or glycinergic receptors, respectively, increases firing rate primarily within the boundaries of the control response area. In contrast, neurons in the non-V-shaped group have response areas that include narrow, closed, tilted, and double-peaked types. In this group, blockade of GABA(A) and glycine receptors increases firing rate but also changes response area shape, with most becoming more V-shaped. We conclude that (1) non-V-shaped response areas can be generated by GABA and glycinergic synapses within the inferior colliculus and do not simply reflect inhibition acting more peripherally in the pathway and (2) frequency-dependent inhibition is an important general feature of the mammalian inferior colliculus and not a specialization unique to echolocating bats.

Impaired electrical signaling disrupts gamma frequency oscillations in connexin 36-deficient mice

Neural processing occurs in parallel in distant cortical areas even for simple perceptual tasks. Associated cognitive binding is believed to occur through the interareal synchronization of rhythmic activity in the gamma (30-80 Hz) range. Such oscillations arise as an emergent property of the neuronal network and require conventional chemical neurotransmission. To test the potential role of gap junction-mediated electrical signaling in this network property, we generated mice lacking connexin 36, the major neuronal connexin. Here we show that the loss of this protein disrupts gamma frequency network oscillations in vitro but leaves high frequency (150 Hz) rhythms, which may involve gap junctions between principal cells (Schmitz et al., 2001), unaffected. Thus, specific connexins differentially deployed throughout cortical networks are likely to regulate different functional aspects of neuronal information processing in the mature brain.

Differential expression of synaptic and nonsynaptic mechanisms underlying stimulus-induced gamma oscillations in vitro

Gamma frequency oscillations occur in hippocampus in vitro after brief tetani delivered to afferent pathways. Previous reports have characterized these oscillations as either (1) trains of GABA(A) inhibitory synaptic events mediated by depolarization of both pyramidal cells and interneurons at least in part mediated by metabotropic glutamate and acetylcholine receptors, or (2) field potential oscillations occurring in the near absence of an inhibitory synaptic oscillation when cells are driven by depolarizing GABA responses and local synchrony is produced by field effects. The aim of this study was to investigate factors involved in the differential expression of these synaptically and nonsynaptically gated oscillations. Field effects were undetectable in control recordings but manifested when slices were perfused with hypo-osmotic solutions or a reduced level of normal perfusate. These manipulations also reduced the amplitude of the train of inhibitory synaptic events associated with an oscillation and enhanced the depolarizing GABA component underlying the post-tetanic depolarization. The resulting field oscillation was still dependent, at least in part, on inhibitory synaptic transmission, but spatiotemporal aspects of the oscillation were severely disrupted. These changes were also accompanied by an increase in estimated [K(+)](o) compared with control. We suggest that nonsynaptic oscillations occur under conditions also associated with epileptiform activity and constitute a phenomenon that is distinct from synaptically gated oscillations. The latter remain a viable model for in vivo oscillations of cognitive relevance.

A possible role for gap junctions in generation of very fast EEG oscillations preceding the onset of, and perhaps initiating, seizures

PURPOSE

We propose an experimentally and clinically testable hypothesis, concerning the origin of very fast (> approximately 70 Hz) EEG oscillations that sometimes precede the onset of focal seizures. These oscillations are important, as they may play a causal role in the initiation of seizures.

METHODS

Subdural EEG recordings were obtained from children with focal cortical dysplasias and intractable seizures. Intra- and extracellular recordings were performed in rat hippocampal slices, with induction of population activity, as follows: (a) bath-applied tetramethylamine (an intracellular alkalinizing agent, that opens gap junctions); (b) bath-applied carbachol, a cholinergic agonist; and (c) focal pressure ejection of hypertonic K+ solution. Detailed network simulations were performed, the better to understand the cellular mechanisms underlying oscillations. A major feature of the simulations was inclusion of axon-axon gap junctions between principal neurons, as supported by recent experimental data.

RESULTS

Very fast oscillations were found in children before seizure onset, but also superimposed on bursts during the seizure, and on interictal bursts. In slice experiments, very fast oscillations had previously been seen on interictal-like bursts; we now show such oscillations before, between, and after epileptiform bursts. Very fast oscillations were also seen superimposed on gamma (30-70 Hz) oscillations induced by carbachol or hypertonic K+, and in the latter case, very fast oscillations became continuous when chemical synapses were blocked. Simulations replicate these data, when axonal gap junctions are included.

CONCLUSIONS

Electrical coupling between principal neurons, perhaps via axonal gap junctions, could underlie very fast population oscillations, in seizure-prone brain, but possibly also in normal brain. The anticonvulsant potential of gap-junction blockers such as carbenoxolone, now in clinical use for treatment of ulcer disease, should be considered.

A model of gamma-frequency network oscillations induced in the rat CA3 region by carbachol in vitro

Carbachol (> 20 microM) and kainate (100 nM) induce, in the in vitro CA3 region, synchronized neuronal population oscillations at approximately 40 Hz having distinctive features: (i) the oscillations persist for hours; (ii) interneurons in kainate fire at 5-20 Hz and their firing is tightly locked to field potential maxima (recorded in s. radiatum); (iii) in contrast, pyramidal cells, in both carbachol and kainate, fire at frequencies as low as 2 Hz, and their firing is less tightly locked to field potentials; (iv) the oscillations require GABAA receptors, AMPA receptors and gap junctions. Using a network of 3072 pyramidal cells and 384 interneurons (each multicompartmental and containing a segment of unmyelinated axon), we employed computer simulations to examine conditions under which network oscillations might occur with the experimentally determined properties. We found that such network oscillations could be generated, robustly, when gap junctions were located between pyramidal cell axons, as suggested to occur based on studies of spontaneous high-frequency (> 100 Hz) network oscillations in the in vitro hippocampus. In the model, pyramidal cell somatic firing was not essential for the oscillations. Critical components of the model are (i) the plexus of pyramidal cell axons, randomly and sparsely interconnected by gap junctions; (ii) glutamate synapses onto interneurons; (iii) synaptic inhibition between interneurons and onto pyramidal cell axons and somata; (iv) a sufficiently high rate of spontaneous action potentials generated in pyramidal cell axons. This model explains the dependence of network oscillations on GABA(A) and AMPA receptors, as well as on gap junctions. Besides the existence of axon-axon gap junctions, the model predicts that many of the pyramidal cell action potentials, during sustained gamma oscillations, are initiated in axons.