The general anesthetic propofol induces ictal-like seizure activity in hippocampal mouse brain slices

The method of euthanasia was approved by the Waikato Ethics Committee at the University
of Waikato, Hamilton, New Zealand.

Preparation of slices

Brain slices were investigated from 30 wild type mice of C57, 129SV and C57/129SV
genetic background. The animals were 1–6 months old, both sexes and kept with unlimited
access to water and food in a 12-h day/night cycle. All age ranges were represented
in all experimental groups.

The mice were anesthetised with carbon dioxide and the brain dissected and transferred
directly to ice-cold carbogenated (95 % O
₂
, 5 % CO
₂
) “normal” artificial cerebrospinal fluid (aCSF) composed of 125 mM NaCl, 2.5 mM KCl,
1 mM MgCl
₂
, 2 mM CaCl
₂
, 1.25 mM NaH
₂
PO
₄
, 26 mM NaHCO
₃
and 10 mM glucose. To maximise the number of slices containing the hippocampus, the
brain was sectioned (400 ?m, Campden Instruments Ltd, Sileby, Leics, UK) coronally
between Bregma 0 to ?5 mm. Thereafter, the slices were allowed at least 1 h of recovery
at room temperature (19–23 °C) in either normal aCSF or magnesium-free (no-Mg) aCSF,
depending on the experiments. No–Mg aCSF was composed of 124 mM NaCl, 5 mM KCl, 2 mM
CaCl
₂
, 1.25 mM NaH2PO4, 26 mM NaHCO3 and 10 mM glucose. The aCSF solutions were used for
no longer than 1 week following storage at 1–4 °C.

Electrical recording of field potential activity

After recovery, one slice at a time was physically divided using a scalpel blade to
isolate the cerebral cortex from the hippocampus; and one hemisphere from the other
(see Fig. 4). This was to ensure that recordings from the cerebral cortex were independent of
the hippocampus and vice versa. Each slice was then transferred to a submersion-style
bath perfused with carbogenated normal or no-Mg aCSF, depending on the experiment
being performed. Perfusion was by gravity feed at a flow rate of 5 ml/min. The flow
rate was checked repeatedly during the recordings to ensure it remained stable.

Fig. 4. Picture of a mouse brain slice with each of the recording regions numbered adjacent
to the respective electrodes. The lines illustrate where the slice was physically sectioned to isolate the cortex from the
hippocampus. 1 the neocortex, 2 the hippocampus and 3 the entorhinal cortex

Extracellular field potential activity was recorded from one to three 25 µm Teflon-coated
tungsten electrodes, positioned in the neocortex, entorhinal cortex and hippocampus.
For the neocortex no particular subregion was targeted and included the posterior
parietal association, retrosplenial, primary somatosensory, primary motor, piriform,
visual and auditory cortical areas. In the hippocampus electrodes were placed in the
CA1, CA3 or dentate gyrus areas.

Recordings were amplified (1000×, A-M Systems, USA or 250× Kerr Tissue Recording System,
Kerr Scientific Instruments, Christchurch, New Zealand), AD-converted (Power 1401,
CED, UK or PowerLab, ADInstruments, Sydney, Australia) and stored on computer (Spike2,
CED, UK or LabChart, ADInstruments, Sydney, Australia) for later offline analysis
using MatLab software. The recordings were low-pass filtered at either 1000 or 100 Hz
(CED and PowerLab, respectively) and high-pass filtered at 1 Hz. All of the experiments
were performed in a Faraday shielded room to minimize electrical noise in the recordings.

Drug preparation and delivery

Propofol was prepared as a 1 % solution of 2,6 diisopropylphenol 97+ % (SAFC supply
solutions, USA) in 20 % Intralipid (Fresenius Kabi AB, Sweden) and etomidate as the
commercially available Hypnomidate solution (Janssen-Cilag, Belgium). The drug concentrations
used were 28, 56 and 84 ?M for propofol and 8, 16 and 24 ?M for etomidate. In each
case the appropriate amount of each drug was added directly to pre-carbogenated aCSF.
The concentrations were calculated to give tissue levels similar to that achieved
clinically for surgical anesthesia, based on the diffusion charactersitics of etomidate
into brain slice tissue and the relative clinical potencies of etomidate and propofol.
Thus, at a slice recording depth of 100–200 ?m, it takes approximately 10–15 min for
the tissue concentration of etomidate to reach half of the drug concentration in the
perfusion bath (Benkwitz et al. 2007]). For our lowest etomidate concentration this equates to a tissue concentration of
4 ?M. In rodents, a brain effect-site etomidate concentration of 13 ?M induces a surgical
level of anesthesia (De Paepe et al. 1999]). Our relatively higher propofol concentrations were based on clinical data showing
that etomidate is approximately five times more potent than propofol (Avramov et al.
1995]).

Experimental protocols

Experiments were performed in either normal or no-Mg aCSF. In the case of the former,
no baseline SLE activity was established in the tissue, according to standard practice.
Under these conditions, viable tissue shows a transient burst of high frequency activity
(see Fig. 5) when an electrode is inserted into the tissue (Voss et al. 2013]). All recordings in normal aCSF showed this characteristic response. When using no-Mg
aCSF, SLE activity was already established in the tissue and recordings were made
from locations where robust and stable activity was found. In these cases, at least
10 min of stable SLE activity was recorded prior to delivery of the first anesthetic
concentration. A corresponding “baseline” time-period was allowed for slices in normal
aCSF. Thereafter, either propofol (n = 27 in no-Mg and n = 7 in normal aCSF) or etomidate
(n = 19 in no-Mg and n = 12 in normal aCSF) were delivered at three increasing concentrations
at 30 min intervals.

Fig. 5. Illustration from one slice perfused with normal artificial cerebrospinal fluid (aCSF)
showing a burst of high frequency activity upon electrode insertion. This characteristic
response was used as a gauge of tissue viability in slices perfused with normal aCSF,
which do not ordinarily express seizure-like event activity

The no-Mg experiments were terminated when either the maximum anesthetic concentration
was reached or SLE frequency dropped to 50 % of that established during baseline.
When one of these criteria was met for all recordings in the slice, wash-out with
drug-free aCSF followed. Wash out was continued for 40 min or until the return of
SLE activity could be confirmed.

To control for the possibility of time effects: 3 slices were perfused with no-Mg
aCSF with addition of three consecutive doses of drug-free 20 % intralipid solution
equivalent in amount to the propofol experiments; and 13 slices were perfused with
no-Mg aCSF for at least 3 h without delivery of anesthetic.

All slices were photographed and recording locations identified in consultation with
the Allen Interactive Brain Atlas.

Statistical analysis

For 36 out of a total of 178 recording locations, the protocol was not completed for
one or other of the following reasons:

1. 10 min of stable baseline no-Mg SLE activity was not achieved.

2. Return of stable no-Mg SLE activity was not achieved during drug wash out,

3. A burst of activity with electrode insertion could not be identified in normal
aCSF recordings.

4. Stable no-Mg SLE activity could not be achieved for a minimum of three hours during
control recordings.

5. Electrodes were inadvertently moved during recording.

These recordings were excluded from the study, giving a total of 142 recordings for
statistical analysis.

During the course of the experiments, it became apparent that propofol was inducing
a unique pattern of activity in the majority of hippocampal recordings, characterised
by a transition from interictal-like SLE activity into long ictal-like bursts (see
“Results and discussion” for details). The propofol hippocampal data was therefore
analysed as a separate group by quantifying the proportion of recordings exhibiting
ictal-like bursting versus interictal-like SLE activity using the Fisher’s exact test.
The qualitative characteristics of these recordings were assessed in detail.

The remainder of the data were analysed in MatLab for changes in SLE amplitude, frequency
and length. Background noise and artefacts were first removed by visual inspection.
The data were then quantified and averaged over three time periods covering:

1.
Baseline 10 min before drug delivery.

2.
Last dose 30 min during perfusion of either the maximum drug concentration or that concentration
required to effect a 50 % reduction in SLE frequency. The 30 min analysis period was
offset by 10 min from the start of perfusion of the relevant concentration, since
at a flow rate of 5 ml/min, 10 min is needed for the drug concentration in the perfusion
bath to equilibrate (data not shown).

3.
End period the last 30–40 min of recording following drug washout.

In the control slices, corresponding time frames were selected. For a small proportion
of recordings (10 out of 142) the automated MatLab scripts failed to accurately report
SLE length. In these cases SLE length was manually quantified and averaged across
five sequential SLEs corresponding to equivalent time points in the recordings.

Graphpad (GraphPad Software Inc 2003, v3.06) was used for all statistical analyses.
Data normality was tested using the Kolmogorov–Smirnov test, and parametric or non-parametric
ANOVA and pair-wise tests applied as appropriate. The data are presented as median
(range). A P value 0.05 was considered statistically significant.