XApEn calculated from the LFP of closely spaced intracortical sites showed significant changes with anesthetic concentrations. This finding demonstrates that volatile anesthetics coerce small cortical sub-networks, here represented by rat barrel cortex, into uniform, synchronized activity patterns.
Barrel cortex forms a large part of rodent somatosensory cortex, characterized by a one-to-one correspondence between the sensory organs (follicles at the base of the large facial whiskers) and cytoarchitectonically segregated structures in layer 4 termed 'barrels' [35–37]. This columnar, somatotopic organization results in an orderly bottom-up spread of sensory-evoked activity which, in the initial stage of processing, is spatially restricted to the discrete termination zones of the major thalamic afferents [38–40]. Yet, barrel cortex is also characterized by a large degree of synaptic divergence and interconnectivity, characteristic of neocortex in general . Axons of pyramidal neurons may span several barrels, especially in layers 2/3 [40, 42], and other cortical areas including contralateral barrel cortex form reciprocal, spatially dispersed connections [37, 43]. Probably owing to this high degree of cortical interconnectivity, the permanent, ongoing activity therein, unrelated to sensory input, has a spatiotemporal profile which is largely independent of the barrel architecture [44, 45]. In the present study, we recorded and analyzed this activity. Signal characteristics showed a dramatic and fundamental change during administration of volatile anesthetics.
The decrease of XApEn with all three volatile anesthetics suggests that these agents transform the diversity of synaptic inputs impinging on closely spaced pyramidal cells into a more uniform, synchronous pattern. These patterns may arise from an enhancement of GABAergic currents and a weakening of glutamatergic currents , which bias synaptic communication towards inhibition. Specifically, GABAergic interneurons likely gain in influence on cortical activity patterns by entraining local networks to common rhythms. Both the relative insensitivity of some interneuron classes to GABAergic inhibition  and the finding that inhibitory inputs in neighboring pyramidal neurons are more synchronous than excitatory inputs  are consistent with this idea. Furthermore, input via long-range connections from other cortical areas is impaired under anesthesia [9, 48, 49]. Experimental findings on figure-ground separation in monkey visual cortex could also be interpreted along the lines of a functional disconnection of cortical areas with anesthetics . Therefore, it seems likely that the long-range excitatory synaptic inputs emanating from various cortical areas have an overall desynchronizing influence on local network activity in barrel cortex, and, by extension, that the impairment of this input by volatile anesthetics contributes to more local synchrony and thus a decrease of XApEn as observed.
Given that intracortical connections outnumber subcortical afferents  and that volatile anesthetics alter network activity in isolated cortical networks in vitro [16, 51] we argue that the decline of XApEn was to a substantial part due to intracortical effects of the anesthetics. In addition, decreases in XApEn probably also reflect the anesthetic-induced transformation of activity in subcortical areas projecting to cortex. In particular thalamus, with its intricate reciprocal connections to cortex [36, 37], must be considered . Shown to be sensitive to volatile anesthetics and prone to bursting behavior, it may imprint its activity patterns on cortex [53–55]. Other likely candidates include the basal forebrain, which modulates cortical activity via cholinergic, GABAergic and glutamatergic afferents [56–58] as well as hypothalamic sleep pathways .
We found that the three volatile anesthetics differed in their potency to alter cortical activity patterns as quantified by XApEn. While isoflurane and enflurane were indistinguishable, halothane had significantly weaker effects. This finding fits the profile of this anesthetic, which has previously been found to exert weaker effects than isoflurane on XApEn computed from interhemispheric EEG  and on spontaneous action potential activity in neocortex in vitro and in vivo [16, 51]. In rat visual cortex, Imas et al. found an enhancement of event-related gamma oscillations at intermediate concentrations of halothane , a finding which underlines the particular characteristic of this anesthetic. A potential limitation of our results is the fact that body temperature of the animals was not controlled, possibly leading to hypothermia during anesthesia. To minimize the influence of this phenomenon, the animals were placed into a plastic housing which provided some thermal insulation, and therefore at least it seems unlikely that they experienced severe hypothermia.
It is surprising that isoflurane- and halothane-induced changes of XApEn computed from interhemispheric EEG and LFP match qualitatively so well (Hudetz et al. used a different set of filter frequencies and computational parameters, so that a quantitative comparison of values may not lead to valid results [10, 11]). First, although the basal mechanisms underlying EEG and LFP signals are identical - synaptic population currents in pyramidal cells giving rise to extracellular potential gradients - the populations of cells sampled from are not. EEG electrodes record potentials from a much larger population than intracortical electrodes. Furthermore, EEG signals are dominated by the largest dipole-generating contributors, pyramidal cells of layer 5, which extend their dendrites to layer 1. Intracortical signals, by contrast, are sampled from a restricted spatial volume  and are thus lamina-dependent [25, 26]. Second, cortical sites separated by less than 1-2 mm as in our experiments receive a great deal of common synaptic input [20, 61], and consequently exhibit more synchronous activity than cortical networks in different hemispheres. Thus, the qualitatively similar depression of XApEn values reported by Hudetz et al. and here show that anesthetics unfold synchronizing effects on widely different spatial scales.