1
ASSESSMENT OF PHARMACOLOGICAL
INTERVENTIONS TO REDUCE SLEEP-RELATED HIPPOCAMPAL SPIKING IN A MOUSE MODEL OF
ALZHEIMER’S DISEASE
Aysu Naz Atalay
Master of Science thesis Master´s Degree Programme in Biomedicine
University of Eastern Finland Faculty of Health Sciences School of Medicine
02.06.2021
2 The University of Eastern Finland, Faculty of Health Sciences, School of Medicine
Master´s Degree Programme in Biomedicine
Aysu Naz Atalay:Assessments of pharmacological interventions to reduce sleep-related epileptic spiking in a mouse model of Alzheimer’s disease
Master of Science thesis; 30 pages
Supervisors: Nanxiang Jin, Ph.D., Postdoctoral researcher.; Heikki Tanila, MD, Ph.D., Professor of Translational Neuroscience
02.06.2021
Keywords Alzheimer´s disease, epilepsy, seizures, antiepileptic drugs, epileptiform discharge, hippocampal spike
Abstract
Alzheimer’s disease (AD) is an aging-related neurodegenerative disorder the main hallmarks of which are increased cerebral level of β-amyloid (Aβ) peptides, amyloid plaques, and neurofibrillary tangles in the brain, and progressive memory loss. Recent studies suggest that more than 40% of early AD patients show epileptiform spiking that correlates with faster cognitive decline. Furthermore, the spiking activity happens almost exclusively during sleep.
A recently described new type of hippocampal spike in the APP/PS1 transgenic AD mouse model, the so-called giant spike (GS), can imitate epileptiform spiking in AD patients. So far, there are no specific drug treatments for epileptiform spiking in AD. An effective treatment to reduce epileptiform spiking is a promising way to slow down the cognitive impairment of AD patients. Therefore, this study aimed to develop a fast and reliable methodological approach for testing the effect of drug candidates for AD-related epileptiform spiking. The data of vehicle groups show the potential of this new approach to effectively find candidate drugs against sleep- related GSs in an AD mouse model, to accelerates the preclinical procedure, and to offer important clinical insight into drug selection and understanding of GSs.
3
1 Introduction
Alzheimer's disease (AD) is a progressive age-related neurodegenerative disorder that has a global effect. The main hallmarks of the disease include specific memory deficits, cognitive decline, and certain types of behavioral changes in individuals. The key neuropathological alterations in the disease are amyloid plaques induced by an increased level of β-amyloid (Aβ) peptides and neurofibrillary tangles (NFT) caused by hyperphosphorylation of tau protein in the brain (Paula et al, 2009). Previous studies have established that Aβ42 isneurotoxic and can bind to acetylcholine receptors and increase the release of glutamate, it is the main excitatory neurotransmitter in the central nervous system, and can thereby stimulate the neocortical and hippocampal pyramidal neurons. These findings indicate that low levels of Aβ42 are fundamental for maintaining a normal brain function, and overexpression of this peptide will lead to accumulation and finally neurotoxicity in the brain (Garcia-Marin et al, 2009).
Previous research has demonstrated that AD increases the risk for seizures. The prevalence of seizures in AD has been reported to range between 5% to 64.2%; however, the contributing factors, as well as the diversity among patients, have not been well-studied yet (Friedman et al, 2012). Moreover, epileptiform activities and seizures have also been reported with amyloid plaque forming in transgenic mice. In 2007, the Palop group showed that high Aβ levels in hAPPFAD mice can initiate spontaneous nonconvulsive seizure activity in hippocampal regions (Palop et al, 2007). Additionally, another study with APdE9 transgenic mice revealed the high risk of developing epileptic seizures during the formation of amyloid plaques, the seizure activity was both behavioral and electrographic and was detected in 25% of the animals at age 3 months and 55% at 4.5 months (Minkeviciene et al, 2009). Although AD is accepted as a risk factor for seizures, they are not quite common in patients. In a 3.7-year follow-up study, only 1.5% of patients with AD demonstrated seizure activity, which decreases the clinical importance of seizures for AD research (Scarmeas et al, 2009).
Notwithstanding, accumulating evidence proves the relationship between epileptiform electroencephalogram (EEG) activity and AD (Friedman et al, 2012). The epileptiform activity can appear at the early stages of the disease and can easily go underdiagnosed when it is not associated with clinically observed symptoms. The largest study of subclinical epileptiform activity was conducted with 1,674 patients in clinical settings. However, in this study, only 3%
displayed epileptiform activity (Liedorp et al, 2010). A serious weakness behind this result is
4 that the routine EEG recording is not sufficient to detect epileptiform activity, because the Vossel group showed that epileptiform activity might be more extensive in the early than the late stage of AD. They detected epileptiform events in 42.4% of patients with AD in overnight EEG recordings, which could be overlooked in routine daytime EEG (Vossel et al, 2013, 2016).
Subsequently, Cumbo and his group found epileptiform activity in 55% of AD patients with video-EEG study (Cumbo & Ligori, 2010). As well, it has been shown that the paroxysmal epileptiform activities can hasten cognitive decline, fasten the disease progress, and might result in excessive neuronal loss in AD patients (Friedman et al, 2012), and they usually occur in the temporal and frontal lobe regions which result in impaired brain functioning (Vossel et al, 2013), and makes treatment strategies highly substantial.
The presented papers shed new light on investigating epileptiform activity and hippocampal spiking in AD as a new treatment approach. In 2017, Vossel’s group suggested that some of the anti-epileptic drugs (AEDs) might have disease-modifying features if they target amyloid-β and tau proteins, and decrease the hyperexcitability of the neurons with the mouse models of AD, indeed, treatment with AEDs could be an efficient way for improving cognitive symptoms or slowing the disease progression in AD patients (Vossel et al, 2017). In addition, recent research observed increased cognitive performance in AD patients with epileptiform activity after treatment with the well-known AED drug levetiracetam (Bakker et al, 2015).
Although seizures are the most visible and easily diagnosable forms of epileptic events, their frequency is far lower than subclinical epileptiform events, which makes seizures less useful as a biomarker of AD-related epileptic activity than subclinical epileptiform activity; on the contrary, discovering drug candidates that could reduce or eliminate the epileptic activity without causing any possible locomotion abnormality might bring an insight into the neural basis of the AD-related epileptic activities.
Other non-convulsive epileptiform activities in transgenic animal models are mainly known as spike-wave discharges (SWDs) and single high-voltage spikes (interictal spikes/epileptiform discharges). SWDs represent at least three cycles of surface-positive and surface-negative spikes around 8–10 Hz waves (Gureviciene et al, 2019). In general, these events have 200μV amplitude and the duration varies between 20 and 200ms (Azevedo, 2015). Additionally, Wistar-Albino-Glaxo from Rijswijk (WAG/Rij) rats are shown to be a model for simulating SWDs (Russo et al, 2016), as well as the transgenic rat model TgF344-AD demonstrated SWD
5 activity (Stoiljkovic et al, 2018). Even though they are slightly different, they have been recognized as a model of human absence epilepsy (Holmes GL, McKeever M, 1987).
The epileptiform discharges (EDs) are usually categorized as spikes (20-70 ms) and sharp waves (70-200 ms) and acts as a powerful biomarker for the diagnosis of epilepsy (Scolaro et al, 2011). EDs occur as interictal activities which mainly happen during sleep (Kane et al, 2017), and they refer to paroxysmal discharges of <250 ms duration between spontaneous epileptic events (Staley et al, 2011). The appearance of these events can cause interruption of hippocampus activity and may result in memory loss (Capdevila et al, 2008). This concept also applies to sharp waves, slow waves, and spikes observed with EEG recordings. Another finding that supports their value as a biomarker in relevant EEG research is the role of the glutamatergic system for initiating EDs, since the connection between glutamate release and the epileptic activity in the brain is well-known from epilepsy studies, this insight clarifies the importance of EDs in the research (Chapman, 2000).
Transgenic AD models have also displayed epileptic activities in similar characteristics, but with a diverged frequency during EEG recordings. Previous research analyzed APPswe/PS1dE mice with EEG recordings and demonstrated epileptic spiking (Gureviciene et al, 2019; Jin et al, 2020). Further, a study conducted with Tg2576 mice, which is a common mouse model for β-amyloid pathology showed interictal spikes happen exclusively during REM sleep in the Tg2576 mice, but not the wild-type mates, they have also shown that hyperexcitability of neurons was present when mice were immobile. From these findings, interictal spikes are likely to be a birth of hyperexcitability in the earliest stages of the disease (Kam et al, 2016).
Additively, a previous animal EEG study found one type of interictal spike called “giant spikes (GSs)” which is a type of interictal spike with hippocampal local potential, and almost exclusively occurring during REM sleep (Gureviciene et al, 2019). The GSs generated in the hippocampus and able to reach an amplitude of ± 5 mV among the hippocampus channels and shown to be destructive in terms of memory consolidation, as well as, they might be considered as a surrogate marker for AD-related seizure research (Gureviciene et al, 2019), since the hyperactivity of neurons may be initiating epileptic seizures in some severe cases, the initial study demonstrated APP/PS1 mice with EDs have many similarities with animal models with seizures (Jin et al, 2018).
6 Previous work has managed to discover the epileptic events in AD mouse models, however, some fundamental challenges should be considered. As highlighted before, some types of epileptiform activities, including spiking, happen almost exclusively during sleep. The sleep phase has a distinct impact on the frequency of these events, as well as the non-REM (NREM) sleep stage can initiate the seizure activity by increasing epileptiform discharges. During NREM sleep, the interictal spiking activity is higher compared to REM sleep, mostly because of the progression of neuronal synchronizations (Carreño & Fernández, 2016).
The first challenge is the detection of sleep stages in freely moving mice. This is highly important to identify epileptiform spiking, as well as the most used methods, are video analysis, electroencephalogram (EEG), and electromyogram (EMG) recordings, however, each element has its limitations. On one hand, video recording is precise to separate moving from immobility but not sufficient enough to identify sleep, since mice may show immobility without being in the sleeping state. Although EMGs are widely used to detect movements in rodent sleep studies, one disadvantage of them is the reliability since subtle movements not always affecting all the muscles and implanted EMG electrodes might lose sensitivity when it is used over time (Lampert et al, 2015). The conventional ways to assign behavior states are mainly cortical screw electrodes and power ratio from the EEG recordings (Gurevicius et al, 2013). However, in drugged mice power ratio may largely change by secondary effects of drugs (Kam et al, 2016), so considering EEG alone as a reliable way to assign sleep stages may not be sufficient enough.
Besides the benefits and limitations, animal models with the use of subdural or depth electrodes are advantageous for overcoming many challenges. Since in the studies with AD patients, depth electrodes are not an option, and 30-minutes of EEG recording is not sufficient enough to detect sleep stages, the animal models have further benefits compared to the conventional methods and clinical studies in AD (Vossel et al, 2017).
The second drawback is detecting the sleep time of animals during the EEG recordings, since detecting GSs is the core of this study, and REM sleep is crucial. In this case, a serious challenge can be caused by using multiple drugs since they might lead to deprivation of REM sleep, and induce locomotion abnormalities as a secondary effect. For instance, particular drugs might alter mice's behavior, can make them hyperactive, in other words, can prevent sleeping activity.
Likewise, they can alter power ratios, and spindles in EEG recording.
7 Comprehensively, this research aimed to address those challenges presented above, and we have managed to develop a new methodology that provides stable results and untangles the current challenges.
Overall, the specific aims of the study are:
• To obtain at least 30 minutes of sleep during a daytime 3h recording session
• Preventing the possibility of REM sleep deprivation that may be caused by drugs
• Building set of steps to quickly and reliably test drugs against GSs/hippocampal spikes in EEG recordings of AD mice
8
2 Materials and methods
2.1 Animals
Ten 3-4 month-old male APPswe/PS1dE9 (APP/PS1) mice were supplied from a local colony at the University of Eastern Finland, based on breeders from John Hopkins University, Baltimore, MD, USA (Jankowsky et al, 2004). The mice were housed in a controlled environment (temperature 22 ± 1 °C, humidity 50–60%, lights on 07:00–19:00) with food and water available ad libitum. All animal procedures were carried out following the guidelines of the European Community Council Directives 86/609/EEC and approved by the Animal Experiment Board of Finland.
2.2 Electrode implantation
The electrode implantation was conducted under general isoflurane anesthesia (induction 4.5%, maintenance at 1.8–2.1%). The electrodes were prepared by Dr. Irina Gureviciene, implantations were done by Dr. Gureviciene and Dr. Nanxiang Jin. Two epidural screws (diameter 1.0 mm, length 2.0 mm, Microbiotech/se AB) were attached in the skull over the frontal cortex bilaterally at AP 2.7 mm, ML ± 2.0 mm from bregma. A bundle of three-wire electrodes (Formwar insulated stainless steel wire, diameter 50µm, California Fine Wire Company Co, Grover Beach, CA, USA) was stereotactically implanted into the hippocampus, with different depths (separation 400 µm) at AP -2.1 mm, ML ± 1.3 mm from bregma and DV -1.7 mm from dura mater. Lastly, one wire electrode was inserted into the neck muscles for electromyogram (EMG) recording. The reference screw electrode was located above the cerebellum. The mice acquired carprofen (5 mg/kg, s.c., Rimadyl R , Vericore, Dundee, UK) after the operation and antibiotic powder on the would (bacitrasin 250 IU/g and neomycinsulfate 5 mg/g, Bacibact, Orion, Finland). Animals were put back into standard laboratory cages with water and food ad libitum, after the implantation of the electrodes. Mice recovered for at least 10 days after surgery and before starting the EEG recordings.
9 2.3 Video-EEG recording
2.3.1 Handling and adaptation to the recording environment
Each mouse received 15 min of gentle handling per day until they were comfortable with the researcher. The handling session included approaching, petting, and observing. Besides handling, the animals were gently introduced to the recording environment. They obtained 3 h- sessions in the arena but without the actual recording, while the behavior of the animal was observed and noted by the researcher.
2.3.2 Recording environment
The video-EEG recording was conducted in a cardboard cylinder (diameter 18.5 cm, wall height 18 cm) located on a frosted glass plate. The mice were video recorded with an overhead webcam connected via USB cable to the computer. Mice were connected to an 18-channel headstage preamplifier with a light-weighted recording cable (Plexon Inc., Dallas, TX, USA) (Figure 1A).
The EEG signals were acquired from the mouse brain with the embedded connector located on the head (Mill-Max, NY, USA). The recording wire was attached to an AC amplifier (A-M Systems 3600, Sequim, WA, USA; gain x 1000, analog band-pass 1–1000 Hz) which was eventually connected to the analog-digital converter inserted in the computer (Figure 1B).
Next, the signal was digitized at 2 kHz for each channel (DT2821 series A/D board; Data Translation, Marlboro, MA, USA). The digitized signal was captured by Sciworks 5.0 program (DataWave Technologies, Loveland, CO, USA) and stored in a computer in .ddf format files.
Each recording lasted 3 hours to give enough time for the animal to sleep.
10
Figure 1 Recording set-up (A) The arena environment from birds' eyes view, a cardboard cylinder was placed on frosted glass and a video camera seated above the arena which records the movements of the mice. (B) An illustration of the data flow shows how signals from the mouse’s brain convert into information for the computer.
2.3.3 EEG recording
The long-term 3- h EEG video recordings took place during the light period and animals were free to move. Since this study aimed to detect a specific type of epileptiform spiking in AD that correlates with REM sleep, we tried to maximize the sleep time of the mice. For that purpose, a well-lit and silent environment was ensured. The schedule of the recordings was based on the sleeping time of rodents and decided as 3 hours. To improve the contrast between the mouse and the background in the video recordings, the recording cylinders were placed on a frosted glass with LED lights placed underneath. This condition was vitally important for video recording, to catch small movements of the animal since it gives the sharpest outline. One other important point is that the LED lights were powered by batteries so that the 50Hz noise from the electrical network was prevented.
The behavior of animals was recorded at ~ 20 frames/s with an over-head camera (Live! Cam, Video IM Pro, Creative, Dublin, Ireland) that was synchronized with the electrophysiological signals. The achieved electrophysiological signals and behavioral videos were collected on the computer with Sciworks 5.0 program (DataWave Technologies, Loveland, CO, USA).
11 2.4 Study protocol
The study was conducted as double-blinded for the persons who prepared drugs and who recorded and analyzed video+EEG to avoid possible biases. Every mouse was treated with 4 different drugs. The drug types and their doses were decided according to available literature (Table 1). In this study, only drugs known as having the ability to penetrate the blood-brain barrier (BBB) were used. Baseline recordings happened for assuring recording quality and making animals familiar with the environment. Drugs were given to the animal with an i.p.
injection 30 min before the recording to make sure that it reaches the brain, as well as to assure the drugs at effective concentrations in the brain during the recording time. The washout time between the recordings was at least 7 days for each animal to prevent the previous drug effects.
Drug Dose Function
Donepezil 0.3 – 1.0 mg/kg i.p Clinical AD drug, AChE inhibitor (Rogers et al, 1998)
Istradefylline 0.1 – 0.3 mg/kg i.p. A2A receptor antagonist (Jenner, 2005) Levetiracetam 30 – 100 mg/kg i.p. Anti-epileptic drug (Vossel et al, 2017) Lamotrigine 10 – 30 mg/kg i.p Anti-epileptic drug (Vossel et al, 2017) Table 1 Drug types and doses used in this study with their function
2.5 Data analysis
The first step of data pre-processing was acquiring the data from EEG which is in (.ddf) format, and conversion of the data to (.mat) format with a customized Matlab program (Mathworks, Natick, MA, USA; R2015b). A further step was the conversion of the (.ddf), to the (.avi) format, and for the aim of that SciWork 5.0 program (DataWave Technologies, Loveland, CO, USA) was used. As a final step, Aiseesoft MOD video converter software (version 9.2.28, Aiseesoft Studio, Hong Kong) was used for compressing the video files, and finally, the data were adjusted for the video analysis with the (.avi) profile.
The videos and mouse location were tracked by Ethovision software (version XT, Noldus Information Technology bv, Wageningen, the Netherlands). Ethovision detects the mouse body and gives the position of the body center at every video frame moment, which allows the detection of movements vs. immobility. The table below (Table 2) gives information about the
12 settings that were adjusted depending on the contrast, length, and possible artifacts of the videos.
Arena diameter 19 cm
Detection Method:Gray Scaling, Range:0-65, Subject contour erosion: 1 pixel, Subject contour dilation: 3 pixels, Subject size: 500-8000 pixels
Table 2 Ethovision settings applied for the study
2.5.1 Detection of giant spikes
Signal analyses were conducted by written algorithms in the Matlab program (Mathworks, Natick, MA, USA; R2015b). Finally, an experienced researcher marked spikes with high voltage (> ±10 SD from the filtered baseline) in one of the channels of the hippocampus as a giant spike (GS). GSs should be simultaneously seen in all channels and could be positive or negative. All GSs were detected by Prof. Heikki Tanila.
2.5.2 Detection of sleep periods
Based on the video analysis, mouse coordinates were extracted and then corrected with customized MATLAB scripts. Furthermore, the three behavioral states were specified, which are, movement, waking immobility, and sleep (including REM and non-REM) based on the instant walking speed of the mouse. If the walking speed was above 1.2 cm/s the video frame was determined as movement, in other cases, immobility. If the immobility period was more than 30 s, from the 31st to the last second of immobility was categorized as sleep. The behavioral assignments were done by Dr. Nanxiang Jin.
2.5.3 Statistics
Preprocessed data were further analyzed using SPSS 14.0 (IBM Software, New York, NY, USA), and Excel (Redmond, WA, USA). Every mouse was compared with itself when different drugs were used. The total recording time was counted in seconds, likewise, the sleeping time of the animal was calculated. The GSs were counted by the experienced researcher and normalized by the sleeping time of the animal.
Data are shown as mean ± standard deviation. The level of significance was set to p < 0.05. The data found to be normally distributed by the Kolmogorov-Smirnov test.
13 To show the frequency of GSs through the first, second, and third weeks, the data from animals (Mouse ID: 11,12,16,35,40,41,67,69) were used, only the data from the first injection per week considered. Firstly, sleep seconds were calculated per recording from the EEG recording, then GSs were counted per sleep second. Secondly, the GS data normalized per sleeping hour, and the mean values of three weeks of recording were calculated for each mice. For analyzing the difference between the average sleep time and GS frequency of each animal, one-way ANOVA was applied.
Furthermore, the GS frequency data was used to compare between the days of injections, weeks of recording, and among individual mice to assure the stability of GSs. This time, both injection days from the first, and second weeks were considered and used for evaluating day-to-day stability.
The data from animals (Mouse ID: 11,12,16,35,40,41,69) was used to indicate the stability between days. The average of GS frequency from the first injection was calculated and stated as Day 1, further the average value from the second injection was measured and stated as Day 2. Finally, paired sample t-test was applied for comparing two days.
The data from animals (Mouse ID: 11,12,16,35,40,41,67,69) was used for indicating week-to- week stability. Firstly, the average values from the first and second injections were calculated for each mouse and stated as the First-Second-Third week. ANOVA repeated measures were applied for testing differences within multiple groups.
For the aim of detecting the general difference in GS frequency between the animals, One-way ANOVA was applied. Furthermore, the Bonferroni method was used for correcting paired comparisons within the groups.
14
3 Results
3.1 Pre-training (handling + arena familiarization) ensures sufficient sleep-time in a 3-h recording session
The video-EEG recording for vehicle groups lasted for three weeks. Each week APP/PS1 mice underwent long-term EEG video recordings (3-h sessions) which were conducted during the daytime (Table 3). Some of the animals could not maintain to the second day of injection, since they died in the home cage overnight before being recorded for the next drug. During the second week of recording the values from mouse #67 could not be derived because of noise and data corruption. The data of average sleep time were normalized to 3h total recording time in all involved mice under vehicle sessions and counted as minutes. The mean was calculated as the average value of 3 vehicle recordings in each mouse (Table 4). The highest amount of sleep observed was 41 min (~0.6 h) in mouse 74. All mice overcame the aimed sleeping amount of time, only the mean sleep time of mouse #40 was lower but very close to the 30 min criterion (~0.45 h), which is very unlikely to be a draw-back in further analyses.
Week 1 Week 2 Week 3
Mouse ID Day 1 Day 2 Day 1 Day 2 Day 1 Day 2
11 X X X X X
12 X X X X X
16 X X X X X
35 X X X X X
40 X X X X X
41 X X X X X
67 X X X X X
69 X X X X X X
74 X X X X
Table 3 Recording sessions involving the sleeping time and giant spike frequency data. Mouse IDs are shown on the left. Each recording included two days and two injections.
15
Mean SD
11 0.6 (39 min) 0.08
12 0.6 (38 min) 0.1
16 0.5 (34 min) 0.2
35 0.5 (32 min) 0.1
40 0.4 (26 min) 0.2
41 0.6 (39 min) 0.1
67 0.6 (39 min) 0.09
69 0.6 (39 min) 0.07
74 0.6 (41 min) 0.08
Table 4 Average sleep time data including all recorded mice, values displayed as Mean and SD.
ANOVA Values
Sum of square
df Mean
square
F Sig.
Between groups
0.140 8 0.017 0.731 0.664
Within groups
0.407 17 0.024
Total 0.547 25
Table 5 One-way ANOVA was used for analyzing the values from average sleep time. No evidence was found for significance among animals (p=0.664).
16
Figure 2 Normalized sleep time of individual mice in all 3 vehicle sessions after pre-training. Each recording session was normalized to 3h and sleep time (min) in each recording was adjusted proportionally. The red horizontal line marks the average sleep hour (36 min) in all mice from all recording sessions. Error bars: Mean (± SEM). One-way ANOVA did not show any significant difference between groups (p>0.05).
In order to get a long enough sleep time for analysis, we pre-trained each APP/PS1 mouse in 3 steps (see Materials and Methods session); each pre-training took 3-4 days. After pre-training, the mean normalized sleep hour of all mice from all vehicles recording sessions reached 0.612 h (=36 min) (red line in Figure 2), which achieved our aim of 30 min.
One-way ANOVA analysis was used to confirm the positive impact of pre-training on the sleeping time of the animal (Mouse ID: 11,12,16,35,40,41,67,69,74). Each mouse counted as one group and calculations were made based on their average value from 3 vehicle sessions (Table 4). There was no statistical difference observed between groups (Table 5). Herewith, the handling + arena familiarization of mice has a positive impact on the sleeping time of the animal (F(8,17)=0.731, p=0.664). Overall, the results are indicating all mice had a similar amount of sleep after the same process of pre-training (Figure 2).
0 5 10 15 20 25 30 35 40 45 50
11 12 16 35 40 41 67 69 74
Sleeping time during recording (min)
Mouse ID
17 3.2 Frequency of giant spikes in sleep
Week 1 Week 2 Week 3 Overall M Overall SD
11 7.6 27.5 39.2 24.8 15.9
12 82.4 36.5 50.7 56.5 23.5
16 33.0 31.9 29.7 31.5 1.6
35 2.7 3.5 12.4 6.2 5.3
40 1.3 3.3 4.2 2.9 1.4
41 3.6 8.5 14.3 8.8 5.3
67 26.4 32.8 186.6 81.9 90.6
69 26.4 21.3 35.4 27.7 7.0
Table 6 Frequency of giant spikes (sp/h) in sleep among the recording weeks including overall mean and standard deviation of individual mice.
The GS number per hour derived only from the first injection/day for the reason of missing data from the second injection in some mice. Mouse #74 was excluded from this analysis due to this reason.
The GS numbers normalized with the sleep time of the animal further displayed as spike per hour (Table 6). The average value from all animals was calculated as approximately 20 spikes per hour. The highest value observed is for mouse #87 during the third week of recording (186 sp/h), and the lowest GS number derived as approximately 1 spike per hour.
18 3.3 Stability of giant spike frequency in sleep during long-term recordings
3.3.1 Comparison between consecutive days
Week 1
M SD
Day 1
22.9 32.0
Day 2
27.8 20.2
Table 7 The mean and standard deviation of GS frequency from 2 different weeks’ vehicle recordings. (A) First week (B) Second week.
Figure 3 Giant spike number (sp) normalized to sleeping time (h) of each animal. (A) The first week of recording (B) Second week of recording. Paired sample t-test (p>0.05) was used for comparing weeks of recording. Error bars: Mean (± SEM).
The values from two days of injections were calculated separately for the first and second weeks of vehicle recordings. The data from giant spike frequency in sleep used for comparing days and weeks of recordings. The mean values varied between 20 to 27 spikes per hour (Table 7).
0 5 10 15 20 25 30 35 40
Day 1 Day 2
Giantspikefrequency(sp/h)
Recording days
Week 1
0 5 10 15 20 25 30
Day 1 Day 2
Giantspikefrequency(sp/h)
Recording days
Week 2
Week 2
M SD
Day 1
20.7 15.1
Day 2
20.4 13.9
A B
A B
19 The daily comparison of giant spikes is illustrated (Figure 3). The paired sample t-test did not indicate any significant differences between recording days (p=0.409), meaning the giant spike frequency showed stability between the days of recordings. The second week of the recording showed the highest stability with almost no difference among the days.
3.3.2 Comparison between different recording weeks
Mean SD
First week 25.4 22.3
Second week 21.3 13.7
Third week 39.9 43.8
Table 8 Giant spike frequency per hour between different vehicle recording weeks, data displayed as mean and standard deviation
Figure 4 The stability of giant spike frequency compared between three weeks of recording. Error bars: Mean (± SEM). Repeated measures ANOVA did not show any significant difference between groups (p>0.05).
0 10 20 30 40 50 60
Week 1 Week 2 Week 3
Giantspikefrequency(sp/h)
Recording weeks
20 The values for giant spike frequency were applied for comparing between 3 groups of recordings (Table 8). The means ranged from 25 giant spikes per hour to 39 spikes. No group significantly different from any other (RM ANOVA, p=0.358). It can simply be seen that the frequency of GSs was stable enough within the weeks of vehicle recording (Figure 4).
3.3.3 Comparison between individual mice
Mean SD
11 20.8 13.6
12 51.6 20.7
16 33.3 2.7
35 8.9 7.8
40 3.7 1.6
41 9.9 5.3
67 74.7 69.0
69 31.3 10.9
Table 9 Giant spike frequency comparison between individual mice, values calculated as mean and SD.
To assess the difference between individual mice, the GS frequency data from each animal (Mouse ID: 11,12,16,35,40,41,69) including all recording days were included. The overall giant spike frequency ranged from 3 to 74 spikes per hour (Table 9).
21
Figure 5 Giant spike frequency (sp/h) between individual mice. One-way ANOVA indicated a significant correlation between the groups. Further, post-hoc comparison using the Bonferroni method showed a significant difference exclusively at mouse 67 (p =0.002).
Occurrence frequency of giant spikes in APP/PS1 mice shown above. A one-way ANOVA was performed to determine the variation between animals. Each mouse counted as one group, the lowest value observed was from mouse #40 (3.70 sp/h).
The results showed significance (F(7,32)=4.209, p=0.002) for the GS frequency value of mouse
#67 (74.72 sp/h), and point to the probability that even though the GS frequency reasonably stable through days, and weeks of comparisons, individual mice can be quite distinct from each other. Overall, data correlate favorably with the importance of gentle handling and adaptation on the sleeping time, and relevantly with the frequency of GSs.
0 20 40 60 80 100 120
11 12 16 35 40 41 67 69
Giantspikefrequency(sp/h)
Mouse ID
*
22
4 Discussion
This study aimed to address four major challenges presented in the Introduction that crucial for researching AD-related epileptiform activity. We have highlighted the importance of handling sleep-related studies, as well as proposed a way to prevent biases caused by the pharmacodynamics of drugs. Furthermore, we used low EMG amplitude as the main sleep criterion and considered video-EEG analysis to double-confirm sleep. This procedure was independent of drug influence on the EEG power ratio or sleep spindle. The currently devised methodology showed stability through days and weeks of recordings.
The first major challenge in preclinical EEG recording is achieving adequate sleeping time during the recordings. For that purpose, we used a method including gentle handling of the animal, with proper adaptation time to the recording environment. Gentle handling mainly includes movements in researchers’ hands, tickling, petting, and allowing freedom to the animal. Rodents are social animals, and human-animal interactions appear to be critical, as well as lack of handling can cause aggression, stress, anxiety, and depression in the animal.
Regarding how stress alters the statistical data in research, it has been shown that quality of sleep could be impaired by a high level of stress (Henderson et al, 2017). Besides handling, they have received adaptation sessions in the recording arena. First, they were able to move in the arena without the cable, once they showed sleeping behavior, they had the 3-h sessions with connected cable overhead but without actual recording, and meanwhile, the behavior was observed by the researcher. Hence, the recording with drugs only conducted when animals were truly adapted and could sleep in the recording conditions. Overall, this methodological study indicates the positive outcomes of gentle handling and arena familiarization of mice, and the current data shows these steps critical to have sufficient sleep time under the effect of drugs.
Along with this, the present procedure is a clear way to ensure sleeping states on long-term EEG recording studies. The good pre-training can ensure long and undisturbed sleep during 3h EEG recordings, this increases the chance to detect REM sleep and record GSs since they occur predominantly during REM sleep.
23 The second pitfall was to identify sleep stages (NREM and REM), and for that purpose, the method including 3 elements is preferred and long-term video, EEG, and EMG recordings are performed in freely moving mice. The traditional methods for judging sleep have been mainly video and/or EEG recordings, and one way to assess sleeping behavior is by evaluating the EEG waveforms. Activities of brain states can indicate the sleep stages, and the frequencies in the brain range between alpha (α), beta(β), sigma(σ), theta(θ), delta(δ). NREM sleep is defined by high amplitude EEG together with low voltage EMG. Conversely, REM sleep has a high frequency of EEG without EMG activity. The delta activity from EEG recordings can be used for characterizing these states (Robert et al, 1999).
These behavioral states on EEG determined as,
(1) alpha (10–16 Hz) by gamma (30–80 Hz) power ratio (alpha/gamma) as NREM sleep, (2) theta (7–9 Hz) by delta (1–4 Hz) power ratio (theta/delta) as REM sleep, and
(3) the muscular electrical activity amplitude (EMG, 1–100 Hz) as movement (Gurevicius et al, 2013).
One other known way for assuring sleep time is analyzing the sleep spindles, the term spindle refers to progressive rhythmic waves with high amplitude among the central regions of the brain (Kane et al, 2017), and they are known as the part of EEG rhythms that can be seen during the NREM sleep as events range between 8 and 15 Hz (Kim et al, 2012). Spindles occur during sleep onset every 3 to 6 seconds, found in all mammals, and they have an important role in- memory processing (Antony et al, 2018). However, assessing sleep only from EEG signals might be challenging since the spindle activity in the brain may be increased or decreased by secondary effects of various drugs (Jankel WR, 1985). Additionally, drugs can alter the EEG power ratios (such as alpha/gamma) of each behavior state, including REM and NREM sleep, and this can create some misinterpretation of the sleeping data (Kam et al, 2016; Sanchez et al, 2012).
Several studies have been carried out on EMG analysis and stated that EMG decreases wakefulness, NREM, and REM sleep since muscles usually lose their strength during NREM sleep, and this has been a traditional marker in sleep-related research (Silvani et al, 2017).
However, a combination of EEG and EMG analysis has also been preferred by many research
24 since it shows higher accuracy, and REM sleep determined by low voltage of EEG together with occasional muscle activities detected by EMG, most likely caused by frequency of heart or blood pressure (Tagluk et al, 2010). The current study has managed to find a way to neglect EEG patterns altered by a side-effect of drugs since my method aimed to improve the presented challenge. Additionally, from the previous studies conducted by video-analysis, if the immobility period was higher than the 30 seconds, it was assigned as sleeping state, however, their paper might have been more convincing, if they benefited from the EMG recording. The sleeping stages are decided only from the video recording, which makes it less precise since the video recording does not provide a sharp border to distinguish between the immobility and sleeping behavior of mice (Jin et al, 2020).
Consequently, for the aim of assigning sleeping states, we have used a method including three dimensions which were video, EEG, and EMG recordings. Since it demonstrates the tense of the muscles, EMG is crucial for deciding either the animal is immobilized or sleeping. During awake state, EMG waves are highly active, and if the mouse is sleeping, EMG waves are quite flat, that is why it is a vital parameter to have in this study. Concisely, this method relies on the most reliable index for overall sleep, which is, EMG signals from neck muscles along with EEG signals from skull electrodes, and acquisition of the movement from the video recordings.
The third major drawback in the preclinical drug studies on epileptic spikes is applying multiple drugs on a small number of animals since the data may be influenced by drug effects, e.g., carry-over effect. Preventing the drug-drug interactions would provide more accurate results and minimize the secondary effect. The current study suggests that wash-out time between drugs is crucial for ensuring the effects arise by the ingested drug, as well as restraining the drug-drug interactions. The clinical data from human studies showed, the AChE inhibition caused by Donepezil begins in the first 25 minutes and lasts less than 24-h for a 2.0 mg dose (Rogers et al, 1998), the A2A antagonist Istradefylline has a longer half-life up to 70 hours (Jenner, 2005). The anti-epileptic drug Levetiracetam shows its effects 15 minutes after administration, and the plasma half-life is about 6 to 8 hours (Wright et al, 2013). Lastly, another antiepileptic drug Lamotrigine occurs in plasma around 1-5 hours and lasts approximately 29 hours (Ramsay et al, 1991).
25 Based on the related literature, a wash-out time of approximately one week is decided to be sufficient for this study. Beyond, we randomized the order of drugs to minimize the chance that a drug effect was caused by the dosing order of drugs rather than the drug itself.
The fourth major challenge is declaring the individual differences in the GS occurrence among the animals. For this purpose, using a within-subject study design is necessary, the suggested statistical method is testing the same mouse with different drugs and comparing the mouse with itself. In addition, during statistical analysis, GSs were normalized per sleep hour, instead of the recording hour since they are predominantly happening during sleep. Additionally, it has been shown that there are huge differences in the GS occurrence between individual animals.
The occurrence of GSs differs within an individual genotype in the present study (Figure 5), and this substantiates previous knowledge in the literature, it is known that EDs are influenced by the sleep-wake cycle and other factors including drug therapy (Gureviciene et al, 2019), and this is the main reason of why animals should be analyzed individually.
This study will be valuable in solving the difficulty of researching sleep-related EDs.
Furthermore, the presented method is reliable enough to determine and treat sleep-related epileptiform spiking.
In conclusion, the recommendations for optimal study design are:
• Gentle handling and adaptation process of the animals until they feel safe with the researcher and comfortable with being in the recording environment
• Ensuring the sleep-states of the animal from EMG neck muscle signals along with EEG brain signals, and video recording for the movement of mice
• Providing the drugs in a randomized order, with a decided wash-out time, for eliminating possible drug interactions
• Deciding a statistical method suitable for comparing each animal within itself, such as the paired test analyses because of the diversity among individual mice
26
Acknowledgments
This study was carried out in the Department of Neurobiology, A.I.Virtanen Institute for Molecular Science, the University of Eastern Finland during the year 2020-2021. I would thank all people that I worked with during this period. Throughout the process of writing the thesis, I have received a great deal of support and assistance.
I genuinely thank Professor Heikki Tanila, MD for admitting me to his research group and allowing me to learn from intelligent people. I want to express my gratitude for this very interesting project and for always providing me
I’m very grateful to my supervisor Nanxiang Jin, Ph.D. for being always available for help and endless support. The practical training combined with genuine encouragement improved me for my further scientific career. His wise reviews and feedbacks helped me to understand my mistakes and brought my scientific skills to a higher level.
I would warmly thank Irina Gureviciene, Ph.D. for her insightful publications in this field, also for inspiring surgeon skills. I thank Jari Nissinen for solving all technical problems immediately. I thank Tiina Kuningas and Jorma Palvimo for their useful feedbacks and teachings during our studies and thesis process.
I thank my colleague Sofya Ziyatdinova, Ph.D. training me for data analysis, and the good conversations we had in our office.
I would warmly thank the official reviewer of this thesis Xavier Ekolle Ndode-Ekane, Ph.D.
for giving his time and effort.
27
References
Antony JW, Piloto L, Wang M, Pacheco P, Norman KA, Paller KA, Antony JW, Piloto L, Wang M, Pacheco P, et al (2018) Sleep Spindle Refractoriness Segregates Periods of Article Sleep Spindle Refractoriness Segregates Periods of Memory Reactivation. Curr Biol 28:
1736-1743.e4
Azevedo CRACFBFM de (2015) Classification of epileptiform events in EEG signals using neural classifier based on SOM. Int Conf Electr Eng Inf Commun Technol (ICEEICT),: 1–
5
Bakker A, Albert MS, Krauss G, Speck CL & Gallagher M (2015) Response of the medial temporal lobe network in amnestic mild cognitive impairment to therapeutic intervention assessed by fMRI and memory task performance. NeuroImage Clin 7: 688–698
Capdevila OS, Dayyat E, Kheirandish-Gozal L & Gozal D (2008) Prevalence of epileptiform activity in healthy children during sleep. Sleep Med 9: 303–309
Carreño M & Fernández S (2016) Sleep-Related Epilepsy. Curr Treat Options Neurol 18 Chapman AG (2000) Glutamate and Epilepsy. J Nutr 130: 1043S–1045S
Cumbo E & Ligori LD (2010) Levetiracetam, lamotrigine, and phenobarbital in patients with epileptic seizures and Alzheimer’s disease. Epilepsy Behav 17: 461–466
Friedman D, Honig LS & Scarmeas N (2012) Seizures and Epilepsy in Alzheimer’s Disease.
CNS Neurosci Ther 18: 285–294
Garcia-Marin V, Blazquez-Llorca L, Rodriguez JR, Boluda S, Muntane G, Ferrer I & DeFelipe J (2009) Diminished perisomatic GABAergic terminals on cortical neurons adjacent to amyloid plaques. Front Neuroanat 3: 1–18
Gureviciene I, Ishchenko I, Ziyatdinova S, Jin N, Lipponen A, Gurevicius K & Tanila H (2019) Characterization of Epileptic Spiking Associated With Brain Amyloidosis in APP/PS1 Mice. Front Neurol 10: 1–22
Gurevicius K, Lipponen A & Tanila H (2013) Increased cortical and thalamic excitability in freely moving APPswe/PS1dE9 mice modeling epileptic activity associated with
28 Alzheimer's disease. Cereb Cortex 23: 1148–1158
Henderson F, Vialou V, El Mestikawy S & Fabre V (2017) Effects of social defeat stress on sleep in mice. Front Behav Neurosci 11: 1–14
Holmes GL, McKeever M AM (1987) Absence seizures in children: Clinical and electroencephalographic features. Ann Neurol 21: 268–273
Jankel WR NE (1985) Sleep Spindles. J Clin Neurophysiol 2: 1–35
Jankowsky JL, Fadale DJ, Anderson J, Xu GM, Gonzales V, Jenkins NA, Copeland NG, Lee MK, Younkin LH, Wagner SL, et al (2004) Mutant presenilins specifically elevate the levels of the 42 residues β-amyloid peptide in vivo: Evidence for augmentation of a 42- specific γ secretase. Hum Mol Genet 13: 159–170
Jenner P (2005) Istradefylline, a novel adenosine A2A receptor antagonist, for the treatment of Parkinson’s disease. Expert Opin Investig Drugs 14: 729–738
Jin N, Lipponen A, Koivisto H, Gurevicius K & Tanila H (2018) Increased cortical beta power and spike-wave discharges in middle-aged APP/PS1 mice. Neurobiol Aging 71: 127–141 Jin N, Ziyatdinova S, Gureviciene I & Tanila H (2020) Response of spike-wave discharges in aged APP/PS1 Alzheimer model mice to antiepileptic, metabolic and cholinergic drugs.
Sci Rep 10: 1–34
Kam K, Duffy ÁM, Moretto J, LaFrancois JJ & Scharfman HE (2016) Interictal spikes during sleep are an early defect in the Tg2576 mouse model of β-amyloid neuropathology. Sci Rep 6: 1–16
Kane N, Acharya J, Benickzy S, Caboclo L, Finnigan S, Kaplan PW, Shibasaki H, Pressler R
& van Putten MJAM (2017) A revised glossary of terms most commonly used by clinical electroencephalographers and updated proposal for the report format of the EEG findings.
Revision 2017. Clin Neurophysiol Pract 2: 170–185
Kim A, Latchoumane C, Lee S, Kim GB, Cheong E, Augustine GJ & Shin HS (2012) Optogenetically induced sleep spindle rhythms to alter sleep architectures in mice. Proc Natl Acad Sci U S A 109: 20673–20678
29 Lampert T, Plano A, Austin J & Platt B (2015) On the identification of sleep stages in mouse
electroencephalography time-series. J Neurosci Methods 246: 52–64
Liedorp Maarten, Cornelis J. Stam PS (2010) Prevalence and Clinical Significance of Epileptiform EEG Discharges in a Large. Dement Geriatr Cogn Disord 29: 432–437 Minkeviciene R, Rheims S, Dobszay MB, Zilberter M, Hartikainen J, Fülöp L, Penke B,
Zilberter Y, Harkany T, Pitkänen A, et al (2009) Amyloid β-induced neuronal hyperexcitability triggers progressive epilepsy. J Neurosci 29: 3453–3462
Palop JJ, Chin J, Roberson ED, Wang J, Thwin MT, Bien-Ly N, Yoo J, Ho KO, Yu GQ, Kreitzer A, et al (2007) Aberrant Excitatory Neuronal Activity and Compensatory Remodeling of Inhibitory Hippocampal Circuits in Mouse Models of Alzheimer’s Disease.
Neuron 55: 697–711
Paula V de JR de, Guimarães FM, Diniz BS & Forlenza OV (2009) Neurobiological pathways to Alzheimer’s disease. Dement Neuropsychol 3: 188–194
Ramsay RE, Pellock JM, Garnett WR, Sanchez RM, Valakas AM, Wargin WA, Lai AA, Hubbell J, Chern WH, Allsup T, et al (1991) Pharmacokinetics and safety of lamotrigine (Lamictal®) in patients with epilepsy. Epilepsy Res 10: 191–200
Robert C, Guilpin C & Limoge A (1999) Automated sleep staging systems in rats. J Neurosci Methods 88: 111–122
Rogers SL, Cooper NM, Sukovaty R, Pederson JE, Lee JN & Friedhoff LT (1998) Pharmacokinetic and pharmacodynamic profile of donepezil HCl following multiple oral doses. Br J Clin Pharmacol Suppl 46: 7–12
Russo E, Citraro R, Constanti A, Leo A, Lüttjohann A, van Luijtelaar G & De Sarro G (2016) Upholding WAG/Rij rats as a model of absence epileptogenesis: Hidden mechanisms and a new theory on seizure development. Neurosci Biobehav Rev 71: 388–408
Sanchez PE, Zhu L, Verret L, Vossel KA, Orr AG, Cirrito JR, Devidze N, Ho K, Yu GQ, Palop JJ, et al (2012) Levetiracetam suppresses neuronal network dysfunction and reverses synaptic and cognitive deficits in an Alzheimer’s disease model. Proc Natl Acad Sci U S A 109: E2895–E2903
30 Scarmeas N, Honig LS, Choi H, Cantero J, Brandt J, Blacker D, Albert M, Amatniek JC, Marder K, Bell K, et al (2009) Seizures in Alzheimer disease: Who, when, and how common?
Arch Neurol 66: 992–997
Scolaro GR, De Azevedo FM & Boos CF (2011) Neural classifier for detection and classification of spikes and sharp waves. BIOSIGNALS 2011 - Proc Int Conf Bio-Inspired Syst Signal Process: 504–509
Silvani A, Ferri R, Lo Martire V, Bastianini S, Berteotti C, Salvadè A, Plazzi G, Zucconi M, Ferini-Strambi L, Bassetti CL, et al (2017) Muscle activity during sleep in human subjects, rats, and mice: Towards translational models of REM sleep without atonia. Sleep 40 Staley KJ, White A & Dudek FE (2011) Interictal spikes: harbingers or causes of epilepsy?
Neurosci Lett 497: 247–250
Stoiljkovic M, Kelley C, Horvath TL & Hajós M (2018) Neurophysiological signals as predictive translational biomarkers for Alzheimer’s disease treatment: Effects of donepezil on neuronal network oscillations in TgF344-AD rats. Alzheimer’s Res Ther 10: 1–11 Tagluk ME, Sezgin N & Akin M (2010) Estimation of sleep stages by an artificial neural
network employing EEG, EMG, and EOG. J Med Syst 34: 717–725
Vossel KA, Beagle AJ, Rabinovici GD, Shu H, Lee SE, Naasan G, Hegde M, Cornes SB, Henry ML, Nelson AB, et al (2013) Seizures and epileptiform activity in the early stages of Alzheimer disease. JAMA Neurol 70: 1158–1166
Vossel KA, Ranasinghe KG, Beagle AJ, Mizuiri D, Honma SM, Dowling AF, Darwish SM, Van Berlo V, Barnes DE, Mantle M, et al (2016) Incidence and impact of subclinical epileptiform activity in Alzheimer’s disease. Ann Neurol 80: 858–870
Vossel KA, Tartaglia MC, Nygaard HB, Zeman AZ & Miller BL (2017) Epileptic activity in Alzheimer’s disease: causes and clinical relevance. Lancet Neurol 16: 311–322
Wright C, Downing J, Mungall D, Khan O, Williams A, Fonkem E, Garrett D, Aceves J &
Kirmani B (2013) Clinical pharmacology and pharmacokinetics of levetiracetam. Front Neurol 4: 1–6
