Optimization of a novel screening model for anti-epileptic
drugs using multi-electrode array technology
Thijmen W. Ligthart
studentID: 11644079
Supervisor: Dr. ing. E.A. van Vliet
Date: 03/07/2020
1
Optimization of a novel screening model for anti-epileptic
drugs using multi-electrode array technology
Thijmen W. Ligthart
Center for Neuroscience, Swammerdam Institute for Life Sciences (SILS), University of Amsterdam, Amsterdam, The Netherlands
Abstract
Epilepsy is one of the most common neurological disorders, which is characterized by recurrent seizures and reduced quality of life. About 30% of patients does not respond to available anti-epileptic treatments. It is therefore important to find new treatment strategies in order to help these pharmacoresistant individuals. However, robust high throughput drug screening models, in which the effects of novel drugs can be tested on epileptogenic networks, are lacking. Therefore, we set up and optimized a novel screening model in which epileptogenic activity was evoked in mouse hippocampal-entorhinal cortex slices using 4-aminopyridine (4-AP) and studied the effects of the anti-epileptic drug carbamazepine on network activity using a multi-electrode array (MEA).
We determined the optimal slicing buffer composition, constructed a flow chamber which creates a laminar flow around the submerged slice and thus provides optimal oxygenation, and determined the optimal flow rate. Finally, we validated the model using carbamazepine, which reduced seizure-like events as expected. This shows that the 4-AP model can be used in combination with an MEA to study the effects of anti-epileptic drugs on network excitability, which is promising for future novel drug screening studies.
Keywords: Multi-electrode array (MEA), Epilepsy, 4-Aminopyridine (4-AP), Carbamazepine, Hippocampus, Entorhinal cortex.
2 Introduction
Epilepsy is one of the most common neurological disorders, with roughly 6 million patients in Europe alone (Baulac et al., 2015). The disease is characterised by seizures and loss of consciousness, and it increases mortality rates significantly due to sudden unexpected death, status epilepticus, suicide, injury during seizures, and surgery among other causes (Hitiris et al., 2007). About 30% of patients does not respond well to available treatment strategies (Kwan and Brodie, 2000). Therefore, it is urgently needed to find novel drug targets and develop new therapeutic strategies. Multi-electrode array (MEA) technology could be of value in the search for new therapeutic strategies. Many currently available treatment strategies, such as carbamazepine, work on sodium channels (Willow et al., 1985). Finding new therapeutic targets to increase the response to treatment may be facilitated by a new screening technique.
In vitro models, such as the 4-aminopyridine (4-AP) model of epilepsy are extensively used to study epileptic activity in hippocampal-entorhinal cortex slices (Avoli & Jefferys, 2016). 4-AP is a potassium (K+) channel blocker that enhances neural transmission and therefore increases
network excitability. In contrast to other in vitro models, such as the bicuculline model in which GABAergic transmission is abolished (Hwa et al., 1991; Karlsson et al., 1992), or the application of glutamatergic agonists to boost excitatory neurotransmission, both GABAergic and glutamatergic transmission remain intact in the 4-AP model. This could be an advantage when screening novel compounds and is thought to mimic the epileptogenic brain better. The 4-AP model is characterized by the appearance of spikes and seizure-like events (SLE) which are also seen in the EEG recordings of epilepsy patients (Gonzalez-Sulser et al., 2011). Typically, field potential recordings or cell patching is performed to measure neuronal or network excitability in this model (Raimondo et al., 2017). However, these techniques do not provide spatial information of the entire hippocampal network, since cellular or very local recordings are performed.
MEA systems, which allow for measurement of potentials at many locations in the hippocampal-entorhinal slice model simultaneously, make it possible to determine the response of the whole network to novel drugs. After all, epilepsy is a consequence of network malfunction. Such a system may create the opportunity to easily screen many potentially anti-epileptic compounds while detecting multiple properties at once. MEA systems also make it possible to study network dynamics, such as the origin of activity and propagation from the origin (Liu et al., 2017). Such recordings across the network of interest make it easy to study the multiple types of activity associated with ictal discharges as described by Avoli et al. (1996) (figure 1). Three distinct types of activity can be characterised in the 4-AP model: type 1: continuous inter-ictal like discharges, type 2: GABA mediated potentials, and type 3: seizure-like events (SLEs). These GABA mediated potentials precede the SLEs and appear to be a “trigger” for the SLE to occur (Uva et al., 2015).
3 Figure 1: Source: Avoli et al., 1996. Three distinct types of activity are shown as recorded with field potential recordings. Type
1 spikes are marked with arrows (A), type 2 spikes are marked with * (B), and type 3 SLE’s are marked with a continuous horizontal line (A).
So far, only one study has been published in which MEA recordings are performed using the 4-AP model (Panuccio et al., 2018). Therefore, not much is known about this novel model and the optimal conditions for recording. Recently, an MEA system was purchased by our research group and the main goal was to build the MEA system and replicate earlier findings. However, there are many parameters that can affect the recordings.
For proper functioning of the 4-AP model in the MEA system, it is hypothesized that both adequate oxygenation as well as extracellular K+ accumulation is required (Morris et al., 1996; Avoli et al., 2002; Panuccio et al., 2018). Furthermore, optimal slice quality is necessary to reliably evoke SLEs. In order to achieve this hypothesized optimal state, custom hardware was used in conjunction with commercially available hardware. Changes in conventional slicing practices were also introduced to benefit the quality of the tissue.
The aim of this thesis is to optimize the 4-AP model of epilepsy in an MEA setup, while using the known anti-epileptic drug carbamazepine (CBZ) as an example drug to prove the validity of the system. Based on previous studies that were performed in the 4-AP model using patch clamp or field potential recordings, it is expected that CBZ will reduce SLEs, but not reduce the number of spikes (Watts & Jefferys, 1993; D’antuono et al., 2010).
4 Materials & methods
Animals & ethics
All animals used within this research were surplus animals gathered from the animal facility of the Swammerdam Institute for Life Sciences. All experiments were performed in accordance with European guidelines. The animals (male mice) were wild-type animals born from APP/WT – GRIA3/WT mutant parent mice, genotypically verified using PCR analysis. At the time of experimentation, the mice were 45 to 56 days old.
Tissue preparation & solutions
Mice were decapitated and the brain was removed. The cerebellum was separated and discarded. During preparation of the cutting platform, the brain was temporarily stored in ice-cold slicing artificial cerebrospinal fluid (slicing-ACSF, bubbled with 95% O2 / 5% CO2),
which consists of: 120mM cholinechloride, 3.5mM kCl, 0.5mM CaCl2, 6mM MgSO4,
NaH2PO4, 10mM glucose, and 25mM NaHCO3. Horizontal slices were cut at a thickness of
400µm at a 10o angle with a Leica VT 1200S vibroslicer while submerged in ice cold slicing-ACSF bubbled with 95% O2 / 5% CO2. Hippocampal-entorhinal cortex slices were transferred
to slicing-ACSF bubbled with 95% O2 / 5% CO2 at room temperature (RT), and the Schaffer
collaterals were cut using a microblade and dissection microscope to create partially disconnected slices. The Schaffer collaterals were cut to prevent the hippocampal-entorhinal loop from suppressing seizures (Barbarosie & Avoli, 1997).
Subsequently, slices were rinsed in two beakers containing holding-ACSF, consisting of: 115mM NaCl, 2mM KCl, 2mM CaCl2, 1.3mM MgSO4, 1.25mM KH2PO4, 25mM glucose,
26mM NaHCO3, and 1mM ascorbic acid. After rinsing, the slices were transferred to a recovery
chamber containing holding-ACSF bubbled with 95% O2 / 5% CO2. In all incubation chambers,
slices were submerged and resting on a soft nylon mesh (figure 2D). The submerged slices remained in the recovery chamber for at least 60 minutes before transfer to a pre-warming chamber containing holding-ACSF at 32oC bubbled with 95% O2 / 5% CO2. Slices were
pre-warmed for 30 minutes, after which they were moved to a 4-AP incubation chamber, which was also bubbled with 95% oxygen at 32oC, containing recording-ACSF consisting of: 115mM NaCl, 2mM KCl, 2mM CaCl2, 1mM MgSO4, 1.25mM KH2PO4, 25mM glucose, 26mM
NaHCO3, 1mM ascorbic acid, and 100µM of 4-AP. Slices remained in 4-AP incubation for 40
minutes before transfer to the recording setup. The submerged brain slices were held down to the recording electrodes using a custom-made platinum hold-down anchor with thin wires of dental floss spun between the platinum rods (figure 2E). After this short transfer and positioning period of less than a minute, the recording chamber was perfused with recording-ACSF (bubbled with 95% O2 / 5% CO2) with 0.1% dimethyl sulfoxide (DMSO) as a vehicle fluid.
Slices were positioned with the entorhinal cortex positioned towards the fluid inlet side of the recording chamber. The recording-ACSF was warmed by a PH01 heating cannula (multichannel systems, MCS, Reutlingen, Germany) at 37oC before entering the 32oC MEA bath, creating a fluid temperature in the bath of roughly 32-34oC (Panuccio et al., 2018). The
slice activity was recorded at 10kHz sampling rate with a 200Hz low-pass filter for 20 minutes after reaching a stable activity pattern. Subsequently the recording chamber was perfused with recording-ACSF with CBZ in 0.1% DMSO, which was also recorded for 20 minutes after
5 Figure 2: Custom made materials and slices. The 3d printed laminar flow chamber (A), the MEA chip and commercial flow
chamber (glass ring around MEA) (B), the platinum hold-down anchor (C), the slice beds for in the incubation chambers (D), and an overlay of a hippocampal-entorhinal cortex slice with the MEA which can be used to identify the recording sites DG=dentate gyrus, EC=entorhinal cortex (E) are shown. See appendix 2 for technical details of the custom-made flow chamber.
reaching a stable activity pattern. Stability of the activity pattern usually occurred within 10 minutes of any alterations. Figure 3 provides a general overview of the slice treatment and recording periods.
Figure 3: general protocol overview. See appendix for full detailed protocol.
Electrophysiology
Multi electrode arrays consisted of 60 TiN/SiN planar electrodes, including the reference electrode, with 200µm between the centre of each electrode on an MEA mini 1200 system head stage (Multichannel systems, MCS, Reutlingen, Germany). Signals were recorded at 10kHz frequency with 200Hz low-pass filter using multichannel recorder software and analysed using multichannel analyser in combination with custom scripts in RStudio.
6 Optimisation strategies
Multiple strategies were used to optimise the quality of the 4-AP model in an MEA recording environment. Multiple changes were made in comparison to the general protocols for patching and field potential recordings used in the lab.
The first comparison was made between two flow chambers. A commercial one that was delivered with the MEA system (Cylindrical, 20mm diameter, 6mm high) and a custom-made flow chamber (6mm x 7.2mm, fluid levels up to 0.5mm high). The original cylindrical flow chamber was initially used due to its inclusion in the standard MEA system, yet did not provide the desired slice activity (see results). After examining the existing literature on 4-AP in an MEA setup and consulting with the first author (Panuccio et al., 2018), a custom laminar flow chamber was created. Creating laminar flow was done by using a custom 3D printed flow chamber that fits inside the original MEA glass chamber (figure 2A) and a custom hold-down anchor with no flow restrictions (figure 2B). The laminar flow chamber was significantly smaller, with an estimated volume of 21.6 µL in and around the slice. In comparison, the original flow chamber has an estimated maximum volume of 1700 µL, of which roughly 1/3rd or more was filled.
Additionally, the effect of the flow rate of the recording solution on the spiking frequency in the slice was examined to determine the optimal flow rate. One slice was recorded at multiple flow rates for at least 5 minutes each to determine the average spiking frequency over the entire duration. Higher flow rates were tested first to prevent damage due to possible oxygen shortage at lower flow rates. A flow of 0.3, 0.4, 0.6, and 1 ml/min were tested.
Lastly, a comparison was also made between two different slicing solutions: Cholinechloride slicing ACSF and Sucrose slicing ACSF. (table 1) This comparison was performed in order to determine the effect of different neuroprotective solutions on the slice activity.
Table 1: Contents of cholinechloride-based slicing-ACSF and sucrose-based slicing-ACSF.
Cholinechloride-based slicing-ACSF Sucrose-based slicing-ACSF
Cholinechloride 120 mM Sucrose 208 mM KCl 3.5 mM KCl 2 mM CaCl2 0.5 mM CaCl2 0.5 mM MgSO4 6 mM MgSO4 2 mM NaH2PO4 1.25 mM MgCl2 5 mM Glucose 10 mM KH2PO4 1.25 mM NaHCO3 25 mM Glucose 10 mM NaHCO3 26 mM Pyruvic acid 3 mM Ascorbic acid 1 mM Analysis
Before analysis, all data was filtered at a 200Hz low pass filter. Spike detection was performed using the Multichannel Analyzer software, after which the exported spike data was analysed using a custom script in RStudio. SLE analysis was done after removal of artefacts by counting
7
the number of SLEs within the last 20 minutes of each recording period. Determination of the SLE duration was also performed and the initial type 2 spike was chosen as the starting point of the SLE and the last type 1 spike as the end point of the event. Channels with excessive noise were also removed from the analysis before any spike calculations were performed.
For frequency calculation, the average frequency within each channel was viewed as an individual data point and analysed with paired comparisons.
Paired t-tests were used for the frequency analysis between two groups with exception of the average spiking frequency between the 4-AP vehicle condition and 4-AP with carbamazepine, where the difference scores were not normally distributed. These difference scores were analysed using a non-parametric Wilcoxon test. A repeated measures ANOVA was used for comparison of multiple flow rates, followed by Tukey’s multiple comparisons test. P<0.05 was assumed to indicate a significant difference.
Results
Comparison between two flow chambers
Both slices in this comparison were cut in cholinechloride based slicing-ACSF. First, slice activity was recorded in the commercial flow chamber at fluid levels of 2mm or higher above the slice with circular turbulent flow at a flow rate of 1 mL/min. The total volume of the volume in the chamber was roughly 570µL, therefore the fluid in the chamber is being replaced every 34.2 seconds (=1.8 times/min). When a recording was performed within the commercial flow chamber, the average spiking frequency of the hippocampus was 0.018Hz (n = 58 channels, SEM = 0.003) and only one type of spike was observed. Nearly all spiking activity originated from the Dentate Gyrus area of the slice (figure 4B/C). Next, the slice activity was recorded with a different slice in the custom-made laminar flow chamber, which is much smaller with an estimated volume of 21.6 µL above and around the slice, therefore refreshing the entire volume of fluid roughly every 1.3 seconds (=48 times/min) at a flow rate of 1 mL/min. The fluid height was a total of 0.5mm (0.1mm above the slice), regulated by the dam like structure at the outlet side of the recording chamber (figure 2 & appendix). When the custom flow chamber was used with low fluid levels, the average spiking frequency significantly increased across the entire hippocampal network (figure 4A, p = <0.0001, t = 30.46, df = 58, SEM = 0.018) to an average of 0.566Hz (n = 54 channels, SEM = 0.016). Furthermore, the change to lower fluid levels and laminar flow in the custom-made flow chamber resulted in the presence of type 2 spiking activity, while these types of spikes were absent when the flow was turbulent within the commercial chamber (figure 4C/E). These type 2 spikes are mostly present in the entorhinal cortex but carry over to the Dentate Gyrus of the hippocampus as can be seen in figure 4 D.
8 Figure 4: The effects of the commercial and custom chamber on spiking activity. The figure shows the differences in network
activity when flow and fluid levels are altered. (A) The adaptation from turbulent flow at high fluid levels to low fluid levels with laminar flow by using a custom flow chamber, which increases the spiking frequency in the hippocampal network on average by 0.547Hz (p < 0.0001, ***, t = 30.46, df = 58, SEM = 0.018), is shown in a bar plot. (B, C, D, E) Unfiltered sample traces that are representative of activity across the entire duration of the recording are also shown within this figure, where type 2 spikes are indicated with a *. Both Dentate Gyrus (B, D) and entorhinal cortex (C, E) traces are shown for both flow chamber designs. Laminar flow spiking activity is more frequent and type 2 spikes are present, while being absent in the turbulent flow recording. These type 2 spikes are more present in the entorhinal cortex (E) when compared to Dentate Gyrus (D).
Optimal laminar flow rate
For the second optimization the flow rate of the recording-ACSF was altered within the recording of one single slice in order to determine the optimal flow rate for the highest spiking activity. The slicewas cut in cholinechloride based slicing-ACSF. The slice used was not prone to generating SLEs yet showed consistent spiking activity of both type 1 and type 2 spikes. The effect of the flow rate was determined by recording the average spiking frequency at a flow rate of 1, 0.6, 0.4, and 0.3 mL/min at 58 electrodes. Each flow rate was recorded for an average of 442±73 seconds. High flow rates were recorded first, in order to minimize any potential damage due to lack of oxygen.
The average spiking frequency was highest at a flow rate of 0.6 mL/min with an average frequency of 0.863Hz (SE = 0.004), while being slightly reduced at a flow rate of 1.0 (p < 0.0001, ***, SE = 0.034, q = 7.097), and 0.4 mL/min (p < 0.0001, ***, SE = 0.025, q = 9.954), with average spiking frequencies of 0.693Hz (SE = 0.032) and 0.662Hz (SE = 0.026) respectively. At 0.3 mL/min the spiking activity was the least frequent (p <0.0001, ***, SE = 0.024, q = 23.21), with an average frequency of 0.476Hz (SE = 0.021). There is no significant difference between the average spiking frequency of 0.4 and 1.0 mL/min flow rates (p = 0.532, SE = 0.022, q = 1.918). Com mer cial cha mbe r Cus tom cha mbe r 0.0 0.2 0.4 0.6 0.8 a v g s p ik in g f re q u e n c y ( H z ) ✱✱✱ -500 0 500 1000 Custom chamber DG [s] [µ V ] 1s -400 -200 0 200 400 Custom chamber EC [s] [µ V ] 1s -100 -50 0 50 100 150 Commercial chamber DG [s] [µ V ] 1s -100 -50 0 50 100 150 Commercial chamber EC [s] [µ V ] 1s
A
B
C
D
E
* *9 Figure 5: Network activity based on the ACSF flow rate. All mean spiking activity for each recorded flow rate is shown.
Significance is indicated with *** (p < 0.001). The highest spiking frequency occurred at a flow rate of 0.6 mL/min, while the lowest amount of activity was recorded at 0.3 mL/min.
Slicing solution
Two different slicing-ACSF solutions were used in the slicing procedure with different results in the activity of the recordings. The first slicing solution was based on cholinechloride as the main ingredient of the ACSF, while the second solution was based on sucrose. The change to sucrose slicing solution appears to have little to no effect on the firing frequency of inter-ictal spikes, yet appears to impact the frequency and reliability with which seizure like events can be induced. One slice was recorded with SLEs that was cut in cholinechloride based slicing-ACSF, and one slice was recorded with SLEs that was cut in sucrose-based slicing-ACSF. Within the cholinechloride based recording, one SLE was recorded over the entire duration of the two hour and ten-minute lasting recording. In contrast, the sucrose-based recording captured five SLEs in a duration of 20 minutes. Thus, it appears that the use of sucrose-based slicing-ACSF creates a more reliable chance to evoke SLEs using 4-AP.
Carbamazepine
In order to determine the validity of the model, carbamazepine was applied to the slice that was cut in sucrose-based slicing-ACSF. This anti-epileptic drug is known to reduce the number of SLEs in the recorded slice, but does not affect spiking frequency (D’Antuono et al., 2010, Colombi et al., 2013). While only one slice was recorded, the effect was apparent. When solely 4-AP + vehicle was applied, five SLEs occurred within a recording period of 20 minutes (fig 6A). When 4-AP was applied in combination with 100µM CBZ however, the number of SLEs was reduced to one at the start of the recording period and the SLEs disappeared entirely in the later periods of the recording (fig 6A). Furthermore, the duration of the one remaining SLE in
0.3 0.4 0.6 1.0 0.0 0.2 0.4 0.6 0.8 1.0
flow rate (mL/min)
a
v
e
ra
g
e
s
p
ik
in
g
f
re
q
u
e
n
c
y
(
H
z
)
✱✱✱ ✱✱✱ ✱✱✱ ✱✱✱ ✱✱✱10 Figure 6: SLE dataset of CBZ validation recording. The figure shows the number of SLEs recorded in the last 20 minutes of
each condition (A), as well as the average duration of these SLEs (B). The median spiking frequency with IQR is also shown in a bar plot (C). Two sample traces of the raw recording are shown. One shows the activity pattern of an SLE in the Dentate gyrus (D), while the other shows the short SLE on the same electrode that was present when CBZ was introduced to the system (E). The duration of the SLE is marked with a continuous line.
the CBZ condition (8 seconds) was highly reduced (fig 6B) when compared to the average duration of the events in the vehicle condition (27±3.391 seconds).
Interictal spiking however did not change significantly across the entire recorded network (figure 6 C). When only 4-AP with vehicle was applied, the median spiking frequency for each channel was 1.408Hz (IQR = 0.491). When compared to 4-AP with 100µM CBZ, where the median frequency was 1.380Hz (IQR = 0.578), the results are not significantly different (p = 0.075, W = -436). 4AP+veh 4AP+CBZ 0 2 4 6 N u m b e r o f S L E /2 0 m in 4AP+veh 4AP+CBZ 0 10 20 30 40 d u ra ti o n o f S L E (s ) 4AP+veh 4AP+CBZ 0.0 0.5 1.0 1.5 2.0 M e d ia n s p ik in g f re q u e n c y ( H z )
A
B
C
D
-100 -50 0 50 100 [s] [µ V ] 1s 4-AP + veh -100 -50 0 50 100 [s] [µ V ] 1s 4-AP + CBZE
11 Discussion
In summary, it appears multiple optimizations are required in order to evoke SLEs using the 4-AP model in an MEA setup. In order to be able to generate SLEs low fluid levels, flow laminarity, optimal flow rate and use of the sucrose based slicing solution are paramount. Furthermore, 4-AP induced SLEs could be inhibited by the anti-epileptic drug CBZ exactly as expected, without altering spiking activity.
Flow chamber
Lowering fluid levels and creating laminarity in the flow greatly increased the quality of the signal while introducing type 2 spiking activity to the network. The presence of these spikes is important for the generation of SLEs (Uva et al., 2015). There are multiple possibilities as to why this effect is so drastic. Firstly, it is expected that creating laminar flow increases the amount of oxygen available to the slice, as the fluid that is in direct contact with the tissue is more reliably refreshed, allowing for better transfer of oxygen to the slice tissue (Guha, 2008). Lowering the fluid levels may also contribute to the availability of oxygen to the slice in multiple ways. With a decrease in the amount of fluid in the recording chamber, the entire content of the chamber will be renewed at a higher rate with similar pump speeds. Also, the increased surface area to volume ratio of the liquid may increase the oxygen available to the slice due to exchange of oxygen with the air around it. Generally, higher amounts of oxygen available provide more energy to the slice, increasing cell survival and potential for activity generation (Hájos & Mody, 2009; Ivanov & Zilberter, 2011). Finally, the decrease in liquid volume above the slice increases the potential for potassium accumulation around the cells, therefore increasing the excitability of the neurons by depolarising the membrane potential in the presence of the K+ channel blocker 4-AP (Panuccio et al., 2018). It is known that the availability of high amounts of K+ are linked to the presence of type 2 spikes and the onset of
seizure like events (Morris et al., 1996; Avoli et al., 2002; Panuccio et al., 2018). Therefore, the presence of type 2 spikes in the custom flow chamber confirms the expected increase in K+ availability at smaller fluid volumes and lower fluid levels.
Flow rate
Changes in flow rate also have an impact on the occurrence of spikes within the network. There is an optimal flow rate at roughly 0.6 mL/min. Both higher and lower flow rates decrease the activity of the network significantly. A possible explanation for this is the balance between K+ availability and ample oxygen. In general, slice activity is higher when more oxygen is provided due to higher flow rates (Hájos & Mody, 2009; Ivanov & Zilberter, 2011). Previous research has found that spontaneous activity is higher when flow rates increase. However, when using the 4-AP model, both oxygen availability and K+ accumulation are required for desirable
activity patterns (Panuccio et al., 2018). The volume of the recording chamber and slice thickness should also be considered. At lower volumes, lower flow rates can more easily provide oxygen due to higher rates at which the chamber contents are refreshed (Panuccio et al., 2018). Slice thickness also impacts oxygen availability, with thicker slices requiring more oxygen to spike reliably (Ivanov & Zilberter, 2011). Using a slice thickness of 400 µm was found to be a fair trade-off between slice viability and ample intrinsic connections (Dossi et al.,
12
2014; Panuccio et al., 2018). When the flow rate is too low, the slice may not be provided with enough oxygen to be able to have enough energy to fire frequently, let alone have an SLE. However, when the flow rate is raised too high, K+ will not be able to accumulate around the
slice (Panuccio et al., 2018).
Slicing solution
The change from cholinechloride-based slicing solution to sucrose-based slicing solution may affect the quality of the slice to be more beneficial for the generation of SLEs. Interneurons are generally thought to be more vulnerable to deterioration in in vitro models (Ting et al., 2014); however, they are thought to be important to generate SLEs (Uva et al., 2015). This explains why the frequency of type 1 spikes remains the same when the slicing solution is changed. Sucrose-based slicing solutions are beneficial for reducing swelling and damage in the superficial layers and have also been shown to improve success rate of patch clamp recordings (Kuenzi et al., 2000; Walz et al., 2002; Ting et al., 2014). Ting et al. (2014) have also shown that using neuroprotective recovery methods involving NMDG-based solutions increases the cell survivability in older mice, providing a more reliable method for recording activity in brain slices cut from aged mice.
Implications and further research opportunities
With a fully optimized model to generate SLEs in an in vitro slice on the MEA, the screening of new anti-epileptic drugs could potentially provide more detailed spatial information about hippocampal-entorhinal cortex network excitability. It would also be much more cost-effective and less time consuming than conventional methods, such as cell patching, field potential recording or in vivo electrophysiology. Furthermore, because of the relative ease of this model and the long duration of recordings, less animals are required for statistically sound research. While within this research paper the model was only tested with CBZ as an anti-epileptic compound, many more compounds can be tested with a setup like the one described. Not only drugs, but also dietary compounds or mutant mice can be used with slight alterations to the setup. The array also allows for all electrodes to be used as stimulating electrodes, opening up the possibility to look at input/output curves instead of taking seizure activity as a measurement. For future research, adding a wash-out condition after the anti-epileptic compound is applied, would improve validity of the research, as this would rule out the general degradation of the slice. However,due to time constraints this may be difficult depending on the protocol used for the compound of interest. Alternatively, an extra slice can be recorded for the same duration as the experimental slice, without applying the anti-epileptic compound to prove that the slice remains viable for the entire duration of the experimental recording. Furthermore, the pre-incubation within 4-AP recording-ACSF can also be skipped if so desired. Although, it is recommended to use this incubation period, as application of 4-AP on the MEA results in a long delay before the desired activity occurs.
Potential future research could use this model to research any compound of interest that may influence seizure like activity and/or interictal activity. Another interesting potential use of MEA technology for epilepsy could be the use with cultured slice models (Gong et al., 2016;
13
Drion et al., 2019), as these models may also provide insights into the effect of the compound on epileptogenesis. Many of the optimizations, such as flow rate, protective ACSF, and fluid levels can be carried over to such models to screen for compounds that may reduce epileptogenesis. Furthermore, using mutant mice for the dissection of brain slices can provide insight into the processes underlying increased network excitability in epilepsy or epileptogenesis.
All things considered, MEA technology in combination with a well-functioning model for epilepsy provides many new research opportunities. By optimizing the model properly, SLEs can be reliably evoked, and suppressed by anti-epileptic compounds, providing a good model for investigating these compounds more effectively.
Conclusion
Our data show that the 4-AP model can be used in combination with an MEA to study the effects of anti-epileptic drugs on network excitability when properly optimized, which is promising for future novel drug screening studies.
Acknowledgements
I would like to thank E.A. van Vliet for supervising this project and investing time and work into our cooperation. I would also like to extend my gratitude to G. Panuccio for her excellent advice regarding potential optimizations of the model.
14 References
Avoli, M., Barbarosie, M., Lücke, A., Nagao, T., Lopantsev, V., & Köhling, R. (1996). Synchronous GABA-mediated potentials and epileptiform discharges in the rat limbic system in vitro. Journal of Neuroscience, 16(12), 3912-3924.
Avoli, M., D’Antuono, M., Louvel, J., Köhling, R., Biagini, G., Pumain, R., ... & Tancredi, V. (2002). Network and pharmacological mechanisms leading to epileptiform synchronization in the limbic system in vitro. Progress in neurobiology, 68(3), 167-207.
Avoli, M., & Jefferys, J. G. (2016). Models of drug-induced epileptiform synchronization in vitro. Journal of neuroscience methods, 260, 26-32.
Barbarosie, M., & Avoli, M. (1997). CA3-driven hippocampal-entorhinal loop controls rather than sustains in vitro limbic seizures. Journal of Neuroscience, 17(23), 9308-9314.
Baulac, M., De Boer, H., Elger, C., Glynn, M., Kälviäinen, R., Little, A., ... & Ryvlin, P. (2015). Epilepsy priorities in Europe: A report of the ILAE‐IBE epilepsy advocacy Europe task force. Epilepsia, 56(11), 1687-1695.
Colombi, I., Mahajani, S., Frega, M., Gasparini, L., & Chiappalone, M. (2013). Effects of antiepileptic drugs on hippocampal neurons coupled to micro-electrode arrays. Frontiers in
neuroengineering, 6, 10.
D’Antuono, M., Köhling, R., Ricalzone, S., Gotman, J., Biagini, G., & Avoli, M. (2010). Antiepileptic drugs abolish ictal but not interictal epileptiform discharges in vitro. Epilepsia,
51(3), 423-431.
Dossi, E., Blauwblomme, T., Nabbout, R., Huberfeld, G., & Rouach, N. (2014). Multi-electrode array recordings of human epileptic postoperative cortical tissue. JoVE (Journal of
Visualized Experiments), (92), e51870.
Drion, C. M., Kooijman, L., Aronica, E., van Vliet, E. A., Wadman, W. J., Chameau, P., & Gorter, J. A. (2019). Curcumin reduces development of seizurelike events in the
hippocampal‐entorhinal cortex slice culture model for epileptogenesis. Epilepsia, 60(4), 605-614.
Gong, W., Senčar, J., Bakkum, D. J., Jäckel, D., Obien, M. E. J., Radivojevic, M., & Hierlemann, A. R. (2016). Multiple single-unit long-term tracking on organotypic
hippocampal slices using high-density microelectrode arrays. Frontiers in neuroscience, 10, 537.
Gonzalez-Sulser, A., Wang, J., Motamedi, G. K., Avoli, M., Vicini, S., & Dzakpasu, R. (2011). The 4-aminopyridine in vitro epilepsy model analyzed with a perforated multi-electrode array. Neuropharmacology, 60(7-8), 1142-1153.
Guha, A. (2008). Transport and deposition of particles in turbulent and laminar flow. Annu.
Rev. Fluid Mech., 40, 311-341.
Hájos, N., & Mody, I. (2009). Establishing a physiological environment for visualized in vitro brain slice recordings by increasing oxygen supply and modifying aCSF content. Journal of
15
Hitiris, N., Mohanraj, R., Norrie, J., & Brodie, M. J. (2007). Mortality in epilepsy. Epilepsy &
behavior, 10(3), 363-376.
Hwa, G. G. C., Avoli, M., Oliver, A., & Villemure, J. G. (1991). Bicuculline-induced epileptogenesis in the human neocortex maintained in vitro. Experimental brain research, 83(2), 329-339.
Ivanov, A., & Zilberter, Y. (2011). Critical state of energy metabolism in brain slices: the principal role of oxygen delivery and energy substrates in shaping neuronal activity. Frontiers
in neuroenergetics, 3, 9.
Karlsson, G., Kolb, C., Hausdorf, A., Portet, C., Schmutz, M., & Olpe, H. R. (1992). GABAB receptors in various in vitro and in vivo models of epilepsy: a study with the GABAB receptor blocker CGP 35348. Neuroscience, 47(1), 63-68.
Kuenzi, F. M., Fitzjohn, S. M., Morton, R. A., Collingridge, G. L., & Seabrook, G. R. (2000). Reduced long-term potentiation in hippocampal slices prepared using sucrose-based artificial cerebrospinal fluid. Journal of neuroscience methods, 100(1-2), 117-122.
Kwan, P., & Brodie, M. J. (2000). Epilepsy after the first drug fails: substitution or add-on?.
Seizure, 9(7), 464-468.
Liu, B. W., Mao, J. W., Shi, Y. J., Lu, Q. C., Liang, P. J., & Zhang, P. M. (2017). Dynamics in the neural network of an in vitro epilepsy model. International Journal of Data Mining and
Bioinformatics, 18(2), 125-143.
Morris, M. E., Obrocea, G. V., & Avoli, M. (1996). Extracellular K+ accumulations and synchronous GABA-mediated potentials evoked by 4-aminopyridine in the adult rat hippocampus. Experimental brain research, 109(1), 71-82.
Panuccio, G., Colombi, I., & Chiappalone, M. (2018). Recording and modulation of
epileptiform activity in rodent brain slices coupled to microelectrode arrays. JoVE (Journal of
Visualized Experiments), (135), e57548.
Raimondo, J. V., Heinemann, U., de Curtis, M., Goodkin, H. P., Dulla, C. G., Janigro, D., ... & Bernard, C. (2017). Methodological standards for in vitro models of epilepsy and epileptic seizures. A TASK 1‐WG 4 report of the AES/ILAE Translational Task Force of the ILAE.
Epilepsia, 58, 40-52.
Ting, J. T., Daigle, T. L., Chen, Q., & Feng, G. (2014). Acute brain slice methods for adult and aging animals: application of targeted patch clamp analysis and optogenetics. In
Patch-clamp methods and protocols (pp. 221-242). Humana Press, New York, NY.
Uva, L., Breschi, G. L., Gnatkovsky, V., Taverna, S., & de Curtis, M. (2015). Synchronous inhibitory potentials precede seizure-like events in acute models of focal limbic seizures.
Journal of Neuroscience, 35(7), 3048-3055.
Walz, W., Boulton, A. A., & Baker, G. B. (Eds.). (2002). Patch-clamp analysis: advanced
techniques (No. 35). Springer Science & Business Media, 172-173.
Watts, A. E., & Jefferys, J. G. (1993). Effects of carbamazepine and baclofen on 4-aminopyridine-induced epileptic activity in rat hippocampal slices. British journal of
16
Willow, M., Gonoi, T., & Catterall, W. L. (1985). Voltage clamp analysis of the inhibitory actions of diphenylhydantoin and carbamazepine on voltage-sensitive sodium channels in neuroblastoma cells. Molecular Pharmacology, 27(5), 549-558.
17 Appendix 1. Full step by step protocol of optimal 4-AP recording procedure on an MEA.
1. Prepare stock solutions for later use (duration: 1.5 hours)
Note: start 1 (or more) day(s) before the experiment.
Prepare stock solutions according to the following concentrations:
recording-ACSF salt chemical Molecular weight concentration (mM)
NaCl 58.44 2300
solvent: distilled water KCl 74.56 40
concentrate: 20x CaCl2 147.02 40
storage temperature: 4-6 ºC MgSO
4 246.48 20
maximum storage duration: 7
days KH2PO4 136.09 25
Holding-ACSF salt chemical Molecular weight concentration (mM)
NaCl 58.44 2300
solvent: distilled water KCl 74.56 40
concentrate: 20x CaCl2 147.02 40
storage temperature: 4-6 ºC MgSO
4 246.48 26
maximum storage duration: 7
days KH2PO4 136.09 25
ACSF Glucose chemical Molecular weight concentration (mM)
glucose 180.2 500
solvent: distilled water NaHCO3 84.01 520
concentrate: 20x
storage temperature: 4-6 ºC
maximum storage duration: 7
days
4-AP stock chemical Molecular weight concentration (mM)
4AP 94.115 100
solvent: distilled water
concentrate: 1000x
storage temperature: 4-6 ºC
maximum storage duration: 14
days
CBZ stock chemical Molecular weight concentration (mM)
CBZ 236.269 100
solvent: dimethyl sulfoxide
(DMSO)
concentrate: 1000x
storage temperature: -20 ºC
maximum storage duration: 6
weeks
It is also recommended to make 1M stock solutions of the following chemicals, these can be stored over longer periods of time (up to 1 year):
18
Chemical Molecular weight Concentration (M)
KCl 74.56 1
CaCl2 147.02 1
MgSO4 246.48 1
MgCl2 203.3 1
KH2PO4 136.09 1
2. Prepare the slicing-ACSF (duration: 30 minutes)
Note: start on the same day as the experiment.
1. Prepare a 1.5 mL 1M ascorbic acid stock by dissolving 0.2641g ascorbic acid (mw: 0.1751 g/mol) in 1.5 mL distilled water.
2. Pour 500 mL of distilled water in a 1L flask and stir using a magnetic stirrer. 3. Add the following stocks:
a. 1 mL KCl b. 0.25 mL CaCl2 c. 1 mL MgSO4 d. 2.5 mL MgCl2 e. 0.625 mL KH2PO4 f. 0.5 mL ascorbic acid
4. Add the following amounts of chemicals.
Chemical Molecular weight Weight (g) Sucrose 342.29 35.598 Glucose 180.2 0.901 NaHCO3 84.01 1.092 Pyruvic acid 110 0.1650
5. Cool the slicing-ACSF to 4 ºC in the fridge.
3. Prepare the holding-ACSF (duration: 5-10 minutes)
Note: the holding-ACSF will be used for rinsing, as well as storing and pre-warming. Always prepare on the day of the experiment.
1. Pour 180 mL distilled water in a 400 mL or 500 mL beaker that will be used for storing the slices.
2. Add 10 mL of holding-ACSF salt stock and 10 mL of ACSF glucose stock. 3. Add 200 µL ascorbic acid stock.
4. Stir the fluid until properly mixed.
5. Place the slice holding bed in the beaker and remove any excess fluid. 6. Repeat step 1-5 for the pre-warming chamber.
7. Pour 80 mL distilled water in a beaker that will be used for rinsing. 8. Add 5 mL of holding-ACSF salt stock and 5 mL of ACSF glucose stock. 9. Add 100 µL ascorbic acid stock.
19
10. Repeat step 7-9 for the second rising beaker.
4. Preparation of experimental setup (duration: 10-15 minutes)
Note: it can take up to 30 minutes for the bath to reach the desired temperature Holding and pre-warming chamber
1. Turn on the warm bath at 32 ºC and place the pre-warming chamber inside.
2. Make sure the holding and pre-warming chamber are bubbled with 95% O2 / 5% CO2
gas mixture.
3. Remove any trapped bubbles using a Pasteur pipette. 4. Cover the chambers with aluminium foil.
Vibratome and dissection microscope
1. Fill a large petri dish with ice-cold slicing-ACSF and surround it with ice. 2. Pour ice cold slicing-ACSF into the slicing tray and surround it with ice.
3. Fill another large petri dish with slicing-ACSF at room temperature and place it under the dissection microscope.
4. Bubble all solutions with 95% O2 / 5% CO2 gas mixture.
5. Place a blade in the vibratome and mount it at a 10º angle. (stripe on the left is just
visible)
6. Set the vibratome to slice at 400µM intervals, cutting speed 0.08 mm/sec and amplitude 0.85 mm.
5. Preparation of the brain slice (30 minutes)
Note: slicing procedure should be performed as quickly as possible. A duration longer than 30 minutes between decapitation of the animal and finishing the slicing procedure is not
recommended. If older animals are used (>8 weeks old) perfusing the mouse with slicing-ACSF before removing the brain is recommended.
1. Decapitate the mouse and extract the brain within 1 minute (write down the time of decapitation).
2. Remove the cerebellum and place the brain in the petri dish with ice-cold slicing-ACSF while preparing the cutting plane.
3. With the use of a spatula, glue the brain to the cutting platform using second glue (gel) with the dorsal side facing downwards and the occipital pole facing the vibratome blade.
4. Place the cutting platform in the slicing tray and begin the slicing procedure after setting the slicing window on the vibratome.
5. Discard the tissue until the hippocampus is properly visible (roughly 900 – 1000 µm). Collect ventral slices.
6. Separate the hemispheres and remove excess tissue, such that hippocampus & entorhinal cortex remain.
20
7. Using an inverted Pasteur pipette, transfer the slices to the petri dish with slicing solution at room tempetature at the dissection microscope and cut the Schaffer collaterals with a microblade.
8. Rinse the slices twice and transfer to the holding chamber, where they will recover for a minimum of 60 minutes at room temperature.
6. Preparation of the recording-ACSF (duration: 5-10 minutes)
Note: this step requires 20 minutes of warming
1. Pour 180 mL distilled water in a 400 mL or 500 mL beaker that will be used for storing the slices.
2. Add 10 mL of recording-ACSF salt stock and 10 mL of ACSF glucose stock. 3. Add 200 µL ascorbic acid stock.
4. Add 200 µL of 4-AP stock.
5. Stir the fluid until properly mixed.
6. Place the slice bed in the tray and remove any excess fluid.
7. Place the tray in the warm bath at 32 ºC (it may take up to 20 minutes for the solution to reach 30-32 oC).
8. Bubble the solution with 95% O2 / 5% CO2 and cover it with aluminium foil.
9. Pour 45 mL of into two 100 mL flasks.
10. Add 2.5 mL of recording-ACSF salt stock and 2.5mL of ACSF glucose stock. 11. Add 50 µL of ascorbic acid stock to each.
12. Add 50 µL of 4-AP stock to each. 13. Add 50 µL of DMSO to one flask.
14. Add 50 µL of CBZ stock to the other flask.
15. Bubble both with 95% O2 / 5% CO2 gas mixture and cover them with aluminium foil.
7. Pre-warming and 4-AP incubation (duration: 70 minutes)
1. Transfer one slice to the pre-warming chamber using an inverted Pasteur pipette 2. Let the slice rest in the pre-warming chamber for 30 minutes before transferring to the
4-AP incubation chamber. Let it rest in 4-AP incubation for 40 minutes.
8. Prepare the MEA setup (duration: 5-15 minutes)
Note: start at least 20 minutes before recording.
1. Place the MEA chip in the headstage and attach the ground. Be aware it fits only in one way!
2. Turn on the amplifier, pump, and heating controller.
3. Start the Multichannel Recorder software and set the sample rate to 10kHz. 4. Choose a file name for the recording and set up the trigger generator.
21
5. Insert the custom laminar flow chamber into the MEA chip and attach the perfusion and waste cannula.
6. Set the pump at a flow rate of 0.6 mL/min with a perfusion/waste ratio of 150%. 7. Attach the pump to the flask containing 4-AP recording solution.
8. Set the MEA bed temperature to 32 ºC and the PH01 cannula temperature to 37 ºC.
9. Turn on data acquisition in the Multichannel Recorder software and check that there is no excessive noise. (noise is acceptable within a 40 µV range)
9. Slice positioning & recording
Note: slice positioning should not take longer than 1 minute. Slice positioning
1. Briefly turn off the perfusion pump when transferring the slice to the MEA recording chamber using an inverted Pasteur pipette. Transfer as little fluid as possible from the incubation chamber to the MEA.
2. Quickly but gently position the slice with the entorhinal cortex facing the perfusion cannula and place the hold-down anchor on top.
3. Turn the perfusion pump back on.
Recording
1. Turn on Data acquisition and verify that the signal is not noisy. 2. Start the recording.
3. Wait for the slice to fire consistently and record for at least another 20 minutes. 4. Change the solution attached to the pump from 4-AP recording-ACSF to 4-AP + CBZ
recording ACSF and mark a triggered event in the software.
5. Wait for the slice to fire consistently and record for at least another 20 minutes. 6. End the recording and save the recorded file.
22 Appendix 2: Technical details of the custom flow chamber
