The underlying mechanisms leading to anesthetic-induced loss of consciousness remain unclear. There is a common agreement that thalamic and cortical areas are major targets of anesthetic substances and that thalamocortical interactions are crucial for maintenance of consciousness (Mashour, 2014). Interfering with information processing between cortex and thalamus may represent a key mechanism of anesthetic-induced unconsciousness. Alkire and colleagues proposed that losing consciousness either occurs by a loss of cortical integration or by a reduction of information capacity (Alkire et al., 2008). This means that either the processing of relevant information is impaired or disrupted during anesthetic-induced unconsciousness, or that the activity patterns representing the information are reduced, insufficient to maintain conscious states. Electrophysiological studies showed that anesthetics decrease the information content in the signal and disrupt or impair directed information flow, i.e., cause an impairment of cortical feedback connectivity and frontal-parietal communication (Ku et al., 2011; Jordan et al., 2013; Ranft et al., 2016). These EEG analyses are restricted to cortical activity, and can only infer the influence of subcortical structures like the thalamus.

Our experiments focused on anesthetic induced changes in cortical neuronal activity following a thalamic electrical stimulus in acute murine slice preparations. Anesthetic substances of different chemical compositions produce a similar phenotypic endpoint—unconsciousness—despite likely differences in their mode of action (Rudolph and Antkowiak, 2004; Franks, 2006). By means of voltage-sensitive dye imaging (VSDI) we intended to evaluate the difference in drug-induced changes to post-stimulus fluorescence between the volatile anesthetic sevoflurane and the intravenous agent propofol. While propofol mainly acts via the GABA_A receptor, sevoflurane affects a broader range of molecular targets, such as GABA_A, glycine or glutamate receptors (Campagna et al., 2003; Rudolph and Antkowiak, 2004). Our VSDI approach with thalamic stimulation allows real-time tracking of stimulus propagation from thalamus to cortex by a fluorescent response caused by potential changes in the neuron's membrane. Therefore, we stained acute murine slice preparations with preserved functional thalamocortical network using a fluorescent dye and evaluated the cortical response in different cortical layers after a voltage stimulus was applied to thalamic structures in the absence and presence of either sevoflurane or propofol. The focus of our investigations was on a possible difference in the anesthetics' effect on thalamocortical signal propagation, based on the different molecular action profiles of sevoflurane and propofol.

The described experimental protocols were approved by the Ethical Committee on Animal Care and Use of the Government of Bavaria (Munich, Germany).

We evaluated the effect of sevoflurane and propofol on cortical depolarization following an electric stimulus applied to the thalamic VPM. While we observed a sevoflurane-induced reduction in intensity and activated cortical area following the stimulus, we found that propofol did not change these parameters. The analysis of defined cortical pixel bands revealed that propofol only affected the distribution of intensity maxima in the 5th pixel band, i.e., propofol seems to specifically constrict the stimulus propagation in this band. Sevoflurane in contrast acted on a wide variety of parameters. It reduced the maximum intensity in all five pixel bands and delayed the cortical response to thalamic stimulation. Interestingly, the number of pixels that contained a maximum in a frame increased in the 1st and 5th pixel bands (i.e., in the upper and lower cortical layers) with sevoflurane exposure, perhaps reflecting less specificity in the effects of sevoflurane on cortical information processing. When looking at the absolute stimulus propagation times, we found the first intensity maxima following the thalamic stimulus around 20–30 ms after the stimulus in the 4th and 5th pixel bands. Hence, these bands seem to reflect the input into the cortical area in the deep cortical layers. These findings underline that sevoflurane and propofol seem to induce general anesthesia in different ways as might be expected from drugs of different molecular families. Propofol mainly acts via GABA_A receptors, whereas sevoflurane in clinically relevant concentration enhances GABA signaling (Garcia et al., 2010) but additionally affects glycine and glutamate receptors (Rudolph and Antkowiak, 2004; Grasshoff et al., 2006).

The minimum alveolar concentration (MAC) defines the anesthetic potency of inhalational anesthetics in terms of a motor response upon a painful stimulus (Eger et al., 1965). MAC values vary among species, and in the case of the species mouse, among strains (Sonner et al., 2000). For mice, a MAC value of 3.22 vol.% for sevoflurane (Ichinose et al., 1998) has been published. We applied sevoflurane by passing the carbogen gas through a calibrated vaporizer. Concentrations given reflect dial settings of the vaporizer. The fact that we did not measure the aqueous concentration (c_aq) of sevoflurane might be a limitation of our study. In our experiments sevoflurane reduced the cortical response upon thalamic stimulation with a calculated IC₅₀ of 3.18%. When calculated with published solubility coefficients for sevoflurane in aqueous solution (Franks and Lieb, 1993, 1996), the application of 1 MAC (3.2%) sevoflurane at room temperature would lead to a calculated c_aq of 0.38 mM. A previous study reported a c_aq of 0.23 mM by application of 2.8% sevoflurane. Other work reported a linear solubility of sevoflurane in aCSF with application of 2.0% sevoflurane leading to a c_aq of 0.42 mM (Haseneder et al., 2009; Nishikawa et al., 2011). Thus, we expect the c_aq of sevoflurane after application of 3.2% sevoflurane under our experimental conditions to be between 0.26 and 0.67 mM. Hence, the applied concentrations of sevoflurane lead to aqueous concentrations that are well in a clinical relevant range and therefore the attenuation of intensity and area of cortical depolarization by sevoflurane occurred at clinically relevant concentrations.

In brain slice preparations, propofol slowly equilibrates in brain tissue and the equilibration time depends on the penetration depth. The recordings in our study were performed after an equilibration time of 30 min. This time is in accordance with patch-clamp studies in acute brain slices that report propofol-induced neuropharmacologic effects at comparable aqueous concentrations of propofol (0.3 μM; Ying and Goldstein, 2005a, 0.6 μM; Ying and Goldstein, 2005b, and 5 μM Chen et al., 2005; Ying et al., 2006) after an equilibration time of 5–20 min. Patch-clamp recordings in acute brain slices are usually obtained from superficially located neurons. Analogously, the recorded VSDI signals have been shown to originate from superficial neuronal structures located up to 50–75 μm deep in the thalamocortical brain slice, since the dye Di-4-ANNEPPS seems to penetrate that deep (Hill and Greenfield, 2013). Published data suggests that at a depth of 50 μm, 50% of the final equilibrium concentration is reached 30 min after propofol application (Gredell et al., 2004). Hence experiments performed at distinct propofol concentrations might underestimate the propofol-induced effects in the brain if unequilibrated. A propofol blood concentration of 3.5 μg/ml seems to produce loss of righting reflex in rodents. This corresponds to a free, i.e., not protein bound, concentration of 0.2 μM (Gredell et al., 2004). We applied propofol in concentrations up to 3 μM. Therefore, our experiments were performed at clinically relevant concentrations despite the slow diffusion kinetics of propofol. A failure to investigate the acute and dynamic effects of propofol on thalamocortical slice physiology is a limitation of our chosen technique. Besides the different actions of sevoflurane and propofol on molecular targets, these differences can be observed at a larger scale.

Phenotypically, the loss of consciousness induced by sevoflurane or propofol seems similar. Human EEG studies revealed a loss of cortical feedback connectivity as a key mechanism of sevoflurane or propofol induced unconsciousness (Ku et al., 2011; Jordan et al., 2013; Ranft et al., 2016). But during general anesthesia patients develop different spectral (frontal) EEG patterns for sevoflurane and propofol (Akeju et al., 2014). The sevoflurane spectrogram shows an unspecific activation of low EEG frequencies especially in the theta range, while propofol causes a distinct alpha peak. Further, the EEG burst suppression pattern in rats, a signature for very deep anesthesia, appears different for sevoflurane and propofol (Kenny et al., 2014); reinforcing the notion that cortical influences of these two drugs differ.

EEG recordings mainly reflect cortical activity that can be heavily influenced by subcortical structures. Electrophysiological and optical (VSDI) signals result from changes in membrane potential, i.e., from transmembrane voltage changes (Buzsáki et al., 2012). Contreras et al could show a similarity of electrophysiological (LFP) and optical (VSDI) evoked responses (Contreras et al., 2017). They also showed a similarity between these signals for 1 s episodes of spontaneous activity. And since LFP activity can be considered “micro-EEG” (Buzsáki et al., 2012), i.e., the EEG being a more generalized and more blurred version of an LFP covering a larger part of the cortical network, we are confident that findings from electrophysiological and optical (VSDI) experiment can be compared to a big degree. The use of fMRI demonstrates that anesthetics also affect subcortical structures such as the thalamus (Mhuircheartaigh et al., 2010). The investigation of anesthetic-induced effects on interactions between cortical and thalamic areas is highly relevant, since they are necessary to maintain consciousness (Mashour and Alkire, 2013). Distortion of these interactions seems a general key mechanism of anesthesia induced unconsciousness (John and Prichep, 2005).

Latest results from fMRI suggest a difference in how the anesthetics, sevoflurane or propofol, affect the connectivity between thalamus and cortex. Sevoflurane did not cause a significant change in the connectivity between thalamus and sensory cortices (Ranft et al., 2016). Propofol in contrast evoked an uncoupling of the thalamus and cortical independent component networks (Jordan et al., 2013). Of course, the findings from the MRI and VSDI experiments cannot be directly related, but the similarity of the findings highlights the differences between sevoflurane and propofol, when it comes to thalamocortical effects.

Our results revealed that propofol may have a more specific mode of action to negatively influence network activity between thalamus and cortex than sevoflurane. While sevoflurane affected the cortex in general, i.e., in all five observed pixel bands, we only found a propofol effect in the 5th band. Our findings of a persistent thalamocortical stimulus propagation structure with sevoflurane and a funneling of this propagation in the 5th with propofol seem to be in line with the mentioned observations from functional fMRI. One limitation in the VSDI recordings is a somewhat coarse spatial resolution, the fit calculation that transfers the selection into an arc line, and the pixel size. It is therefore not possible to draw definitive conclusions from the pixel bands to effects that are specific for distinct cortical layers. Previous work (Cruikshank et al., 2010) showed, that axonal arbors from thalamocortical projections are concentrated in two regions that are localized further than 500 μm from the edge of the brain in the same angled slice preparation we used for our experiments. Since thalamocortical neurons primarily send projections to cortical layers IV and VI (Lee and Imaizumi, 2013), our results from the 5th band correspond to the deep cortical layers, most likely layer IV–VI.

Cortical layer IV plays a special role in thalamocortical processing (Berbel et al., 2010; Christianson et al., 2011); it serves as input of feed-forward excitation from thalamic core cells, neurons from VPL and VPM (Jones, 1998). Thalamic information undergoes initial intracortical signal processing in cortical layer IV. From there, information is relayed to supragranular laminae, i.e., cortical laver I and II (Jones, 1998). An anesthetic-induced modulation of neurons in layer IV might therefore result in impaired cortical propagation of signals initiated in the thalamus. Cortical layer VI contains pyramidal neurons that provide excitatory feedback to the thalamus (Alitto and Usrey, 2003). Together with cortical layer IV, it presents an important part of the communication structure between thalamus and cortex.

Under the influence of propofol we saw a reduced distribution of depolarization in these deep cortical layers following electric stimulation of the thalamus. Changes in dye fluorescence of each pixel represent the combined neuronal activity of axons, dendrites, neuronal cell bodies and glia. Because the surface area of dendrites is much larger than that of somata or axons, the origin of the recorded signal is mainly dendritic (Yuste et al., 1997). In accordance with other authors (Hill and Greenfield, 2013) we conclude that our recordings reflect the activity in the postsynaptic arborization of the thalamocortical afferents. In the cortex simple sensory stimuli lead to concomitant occurrence of synaptic excitation and inhibition (Isaacson and Scanziani, 2011). Principal and inhibitory neurons are reciprocally connected and local network activity is consequently modulated by inhibitory feedback. As a GABA_A receptor agonist, propofol enhances inhibitory synaptic transmission (Eckle et al., 2015) and consequently shifts the balance of excitation and inhibition toward inhibition. Hence, reduced spatial distribution of pixels presenting maximum intensity over a period of time (110 ms) is highly likely to originate from augmented local recurrent inhibition of the deep cortical structures involved in communication between thalamus and cortex. This inhibition and the funneling effect may lead or contribute to decoupling between thalamus and cortex. However, we cannot draw a definitive conclusion from our experimental data, especially if afferent or efferent cortical connections to thalamus or both are affected.

In contrast, sevoflurane caused a general decrease in depolarization upon thalamic stimulation in all cortical layers. In contrast to propofol, sevoflurane impairs local field potential activity in organotypic slice cultures of isolated cortex (Drexler et al., 2013). These findings agree with our results of sevoflurane affecting our parameters in all cortical layers, whereas propofol only affects one parameter in the 5th pixel band.

In addition to the potentiation of GABAergic inhibition, sevoflurane enhances glycine receptor activity and impairs excitatory synaptic transmission at clinical relevant concentrations (Rudolph and Antkowiak, 2004). Hence, the sevoflurane-induced general attenuation of cortical depolarization that we observed in our study possibly results either from reduced excitatory feed-forward inputs from the thalamus or impaired corticocortical neurotransmission. Contrary to propofol, sevoflurane increased the spatial distribution of intensity maxima in the 5th pixel band. It is possible that the effect of sevoflurane on excitatory pathways may be more important for producing unconsciousness than its effect on enhancing activity of local inhibitory networks. This is in line with findings showing that volatile anesthetics mainly act on glutamine-mediated orthodromic pathways in the hippocampus whereas propofol enhances recurrent inhibition (Asahi et al., 2006).

In agreement with our data, Raz et al. detected a strong effect of another volatile anesthetic, isoflurane, on corticocortical processing in the auditory and visual cortex (Raz et al., 2014). The authors describe that isoflurane preferentially inhibits “top-down” projections. With our approach of thalamic stimulation, we could observe significant effects of sevoflurane on the “bottom-up” thalamocortical pathway. These findings may seem contradictory, but the authors state that the action of isoflurane is pathway specific. In contrast to Raz and co-workers we applied electric stimuli to a thalamic nucleus mainly responsible for relaying sensory information (Miyata, 2007) in our experiments. Hence, the results may not be comparable.

Liu et al. (2013) further found that propofol also affects the thalamocortical system. The ventrobasal complex of the thalamus is a first order relay (specific system) because it transmits information from a subcortical source (lemniscal projections; Andersen et al., 1964; Sherman and Guillery, 1996) directly to the cortex. We placed the stimulation electrode onto the ventrobasal complex for our voltage-sensitive dye imaging experiments. Higher-order pathways seem to be preserved in the thalamocortical slice preparation (Lee and Sherman, 2008). We cannot completely exclude that our electrical pulse applied to the ventrobasal complex also stimulated higher-order nuclei of the thalamus finally resulting in cortical depolarization. This might limit the reliability of our findings with respect to their specificity to the ventrobasal complex and we cannot conclude whether propofol also affects the nonspecific system.

In summary, we have demonstrated that sevoflurane and propofol affect thalamocortical stimulus propagation in a substance-specific manner. Sevoflurane globally decreased cortical depolarization upon thalamic stimulation whereas propofol specifically attenuated signal processing in the deep cortical areas. This provides further evidence that the mechanisms of anesthetic-induced loss of consciousness are drug- and pathway-specific.

SK designed and performed experiments, wrote manuscript; CM performed experiments and provided critical feedback on manuscript; SS, EK, PG, GS, and GR provided critical feedback on the manuscript; RH designed experiments and provided critical feedback on manuscript; MK designed and conducted analysis, wrote manuscript.