6 April 2006. Schizophrenia is associated with disruptions in gamma rhythms, low-frequency oscillations typically between 40 and 80 Hz that can be measured by electroencephalogram (EEG). Gamma (γ) oscillations are primarily generated by GABAergic inhibitory interneurons of the cerebral cortex (see SRF related news story), and the rhythms are intimately related to processing of sensory information (for review, see Engel et al., 2001). A broadly drawn case has been made by David Lewis's group at the University of Pittsburgh and others that γ band abnormalities in prefrontal cortex in schizophrenia can be traced to the parvalbumin-positive subpopulation of interneurons (for review, see Lewis et al., 2005).
But what about in other cortical areas? In the March 8 Journal of Neuroscience, researchers in England report that in two very different putative animal models of schizophrenia—one model associated with a loss of layer II parvalbumin-positive interneurons, and the other with acute reduction of NMDA receptor function—a similar pattern of γ oscillation occurs in the entorhinal cortex (EC). The hippocampus, on the other hand, is spared both the loss of this class of interneurons and disruptions in γ band in these models. The findings suggest that functional deficits specific to the EC interneurons could underlie some of the cognitive problems manifested by schizophrenia patients, and further suggest that both GABAergic and glutamatergic disruption can contribute to the disorder.
Unlike the prefrontal cortex, the entorhinal cortex is not neocortex. However, it is the most "neocortex-like" of the phylogenetically older, archicortex areas, possessing nearly as many identifiable layers as neocortex (six, in traditional schemas, though many researchers have postulated as many nine layers in neocortex), and it does generate γ rhythms. Thus, it makes sense to look for similarities to prefrontal cortex abnormalities in schizophrenia. Furthermore, the EC lies at the crossroads between sensory input and motor output. Axons from the stellate cells of the outer, superficial layers of the region form the perforant pathway, which funnels excitatory stimuli to the hippocampus. Because the latter is involved in learning and memory, the EC is also perfectly poised to facilitate sensorimotor gating, or the process of modulating motor responses to sensory information, such as sights and sounds. In schizophrenia patients, however, this gating system does not function normally (see SRF related news story). Finally, a number of EC abnormalities have been reported in schizophrenia, including morphologic and neurochemical changes specific to the stellate cells of the superficial layers, to interneurons, and to synapses.
Employing slice preparations of EC and hippocampus, Claudia Racca and colleagues at the University of Newcastle, England, together with collaborators at the University of Leeds and GlaxoSmithKline, investigated the two very different models of schizophrenia, both of which feature sensorimotor gating deficiency. The first model, ablation of the gene for the lysophosphatidic acid receptor, causes developmental problems in mice that lead to, among other things, loss of prepulse inhibition, a measure of sensorimotor gating (see Harrison et al., 2003). The role of the neuroactive LPA is still unclear, but it is heavily expressed during certain developmental stages, and then appears confined to myelin in adulthood. The second model, administration of the N-methyl-D-aspartate receptor antagonist ketamine, is well known for the psychosis-like state it induces in humans. Ketamine has been shown to induce sensorimotor gating deficits in both rats and humans (see de Bruin et al., 1999 and Oranje et al., 2002).
Lead author Mark Cunningham and colleagues examined γ rhythms elicited by bathing EC or hippocampal tissue slices in the glutamate agonist kainite, one effect of which is to excite interneurons expressing NMDA receptors. They found that γ oscillation in the superficial layer II of the EC was considerably reduced in LPA1-negative slices compared to wild-type tissue. In contrast, γ oscillations in the deep EC were no different from those recorded from normal slices.
In support of this, the authors found that the number of neurons that test positive for γ-aminobutyric acid (GABA) is dramatically reduced (by about twofold) in the EC of LPA1 receptor-negative mice: GABAergic inhibitory neurons are thought to be the source of the γ oscillations. In fact, the authors found that this reduction was confined to parvalbumin-positive interneurons. This loss was restricted to layer II of the EC, also, confirming the importance of this particular zone in the receptor-deficient model.
Loss of interneuron activity in the EC would most likely lead to greater excitability in the stellate cells, those neurons that are reined in by the interneurons. In fact, the authors tested this hypothesis and found elevated excitatory activity in the superficial layers of the EC. When Cunningham and colleagues measured excitatory postsynaptic potentials from stellate cells in LPA1-deficient tissue slices, they found that the cells fired off action potentials over four times more frequently than did cells in normal slices.
By contrast, neither γ oscillations nor the quantities of interneurons showed any change in the hippocampus, a region that has been implicated in schizophrenia by various lines of evidence.
The authors found strikingly similar effects in the second model. When they examined the effects of ketamine on slices of normal brain tissue, they found that the NMDA antagonist significantly weakened γ oscillations in the EC but not in the hippocampus. γ field power was reduced by over 40 percent, and again, the effect was confined to the superficial EC; deep layers were unaffected. “In summary, two models of psychiatric illness associated with sensorimotor-gating deficit generated a pattern of disruption in γ rhythms specific to the superficial entorhinal cortex,” write the authors.
These findings do not rule out a role for the hippocampus in schizophrenia. Curiously, postmortem studies show that parvalbumin-positive neurons are lost from all subfields of the hippocampus in schizophrenia patients (see Zhang and Reynolds, 2002). In the mouse models used in this study, γ oscillations in the hippocampus may be unaffected for several reasons. For one, LPA1-decficiency does not seem to affect parvalbumin-positive neurons in the hippocampus—their stability may be related to the fact that cholinergic rather than glutamatergic innervation is sufficient to generate γ oscillations in the hippocampus (see Fisahn et al., 1998). In addition, the authors suggest their ketamine regimen may be too short to have any effect on the hippocampus.—Tom Fagan and Hakon Heimer.
Cunningham MO, Hunt J, Middleton S, LeBeau FEN, Gillies MG, Davies CH, Maycox PR, Whittington MA, Racca C. Region-specific reduction in entorhinal γ oscillations and parvalbumin-immunoreactive neurons in animal models of psychiatric illness. Journal of Neuroscience. March 8, 2006;26:2767-2776. Abstract