The basic question underlying our work is, whether insights from animal model systems can be used to refine medical EEG diagnostics in early preterm babies and especially to identify typical patterns in the EEG which are ubiquitously related to the status of cortical development.
Preterm birth is worldwide of growing incidence . Survival rates of preterm babies are increasing due to modern health monitoring and specialized neonatal care units . Unfortunately, preterm birth is associated with several risk factors during later life, including cognitive impairments . One major problem is the lack of knowledge regarding a “normal” brain maturation during early development . On the one hand, preterm birth per se is a non-physiologic condition. On the other hand, due to ethical constraints, brain monitoring of preterm babies is usually limited to pathophysiological indications as for example epilepsy .
Therefore, our aim is to evaluate the mouse and the piglet as model systems for preterm EEG patterns which are related to cortical maturation. Knowledge about an age related state of cortical maturation is the only way to distinguish between physiological development and pathophysiologic maldevelopments, because preterm EEG differs profoundly from adult EEG as well as from newborn EEG. EEG in preterm babies is discontinuous, interhemispheric coherence is weak  and sleep/wake stages are not as clearly established as in the EEG of adults or older children [7,8]. A discontinuous EEG is characterized by very low overall network activity and some sparse, spontaneously occurring events, like spindle bursts or delta brushes (latter also known as spontaneous activity transients ). It turned out, that very slow oscillations play an important role during cortical maturation . Consequently we used full band EEG (fbEEG) in order to record even very slow potentials. Such waves with a duration up to several seconds are cut off with standard AC coupled EEG amplifiers . A newborn mouse corresponds approximately to the 17th week of gestation in human regarding cortical maturation/EEG  (http://www.translatingtime.net/). 3 to 4 day old mouse pubs correspond to the 22nd week of gestation in human. Very early preterm babies in human can survive from the 23rd week of gestation on. Two week old mice correspond to a cortical maturation around the 46th week of gestation in human, which is already post term. In comparison to humans, mice are born with a preterm brain . Nevertheless, typical EEG patterns like spindle and gamma bursts can also be observed in the mouse model [13, 14, 15, 16].
The piglet is an emerging model system in the field of developmental neurobiology . Therefore, knowledge is still limited but growing. Obvious advantages are the physiological as well as anatomical similarity to humans and a similar body size. A disadvantage is that piglets are precocious and therefore at least partially more mature at the date of birth in comparison to mouse pubs and human babies.
In contrast to scientific research, the method of choice on the neonatal intensive care unit is the scalp EEG. EEGs derived from the scalp do not contain unit activity but instead exclusively global activity patterns of oscillating neural networks. Nevertheless, recent findings suggest that certain patterns in the preterm scalp EEG are predictive for the health status of former preterm children several years later in life . This finding is an important step on the way to use clinical EEG as diagnostic tool for preventive treatments in the case of preterm birth. Another approach is to analyze correlations between EEG channels and/or frequency bands. Indeed, it has been shown that connectivity analysis is correlated with cortical maturation . One such tool of analysis is phase amplitude coupling, also called nesting coefficient . It deciphers the degree of amplitude modulation of one frequency band as a function of the phase relation to a second frequency band of the EEG, thus a special case of correlation. Delta brushes are representative examples for such nested events in the developing cortex .
By using the mouse and the piglet as model systems for the development of scalp EEG patterns we would like to close a methodological gap to acquire the necessary knowledge regarding early cortical maturation and related patterns in the clinical EEG. We were able to extract patterns of cortical development across both model systems with analytic tools, relying on network activity, especially phase amplitude coupling. In the future this kind of analysis in combination with telemetric full band EEG might be an interesting diagnostic tool for the neonatal intensive care unit.
Materials and Methods
All procedures were approved by the local ethics committee (#23177-07/G10-1-010/G 15-15-011), and followed the European and the German national regulations (European Communities Council Directive, 86/609/ECC; Tierschutzgesetz).
All animal procedures were performed in accordance with the [Medical Center of the Johannes Gutenberg-University Mainz] animal care committee’s regulations.
mice of either sex:
For general anesthesia Urethane (1g/kg) was injected IP. Local anesthesia was achieved by lidocaine gel (Xylocain 2%, Astra Zeneca, UK).
For head fixation, a metal ring was fixed with super glue and dental cement on the skull as described in .
Teflon coated silver wires (advent) with a diameter of 50μm have been used for supra cranial recordings (on top of the Dura). For super cranial recordings (on top of the skull) the same wires have been deinsulated and melted at the tip. The resulting metal droplet was pressed in order to acquire a flattened round electrode tip with a diameter of approximately 1mm. The wires were initially fixed with a conductive 1:1 mixture of Kwik Cast (2-component silicone elastomer, World Precision Instruments, Sarasota, FL, USA) and NaCl containing electrode gel (signa crème, parker laboratories inc., USA) for EEG recordings. In case of age groups without known coordinates, Barrel cortex and forelimb region were pre experimentally determined by intrinsic imaging with Whisker and Forelimb stimulation by a motor driven short touch. After precise placement of electrodes with a micromanipulator, electrodes were additionally fixed with dental cement (Paladur, Heraeus, Germany and Tetric evo flow, ivoclar vivadent GmbH, Germany) and subsequently connected via molex pins to a DC multi channel systems full band EEG (fbEEG) recording setup. Finally, the recording site was confirmed by whisker and forelimb stimulation, respectively. piglets of either sex:
We recorded EEGs from piglets directly in the pigpen. To fix the electrodes on the piglets scalp, piglets were wrapped into a piece of soft tissue to calm them down. We used no anesthesia. Disposable adhesive surface silver/silverchloride electrodes (Spes Medica S.r.l., Genova, Italy) were placed above cerebellum (between the ears) as ground, above the nose as reference and between eye and ear to record from the somatosensory cortex region. In order to assure anatomical dimension of the newborn piglets brain, we prepared brains from stillborn piglets. Whole brains were fixed for two weeks in 5% paraformaldehyde and plastinated afterwards.
Before Fixation of the electrode, the skin was cleaned with an abrasive cream (Abralyt HiCl, Easycap GmbH, Herrsching, Germany) in order to remove dead skin cells and to achieve a lower impedance. The data were recorded and send by a telemetry unit (with an AC coupled amplifier), . Only sleep phases were taken into account for analysis.
Comparison of super cranial EEG recordings with dural EEG in mice
Custom made super cranial electrodes were compared with silver wire recordings on top of the skull in P3 animals. Animals were head fixed and anesthetized as described beforehand. We marked the coordinates for Barrel Cortex on the left and right brain hemisphere. 2mm apart from this position a hole was carefully scratched into the skull by the aid of a small piece of razor blade. The supracranial electrode was slowly pushed forward to the marked position, underneath the skull and afterwards fixed with dental cement. The supercranial electrode was placed on top of the skull at the marked position. With this technique we were able to record simultaneously super- and supracranial EEG at the same position.
Comparison EEG under urethane anesthesia vs. awake (local anesthesia) in mice
Newborn mice (P0 or P1) were locally anesthetized as described beforehand. Supercranial electrodes were fixed on top of the skull (forelimb and barrel cortex region of the primary somatosensory cortex). After the recording of spontaneous and evoked EEG, the same animals were anesthetized with urethane. 40 minutes thereafter EEG of spontaneous as well as evoked activity was measured again under urethane anesthesia.
Initial sampling rate for mouse experiments was 10000Hz, for further analyzes, data were down sampled to 1000Hz. We used a DC coupled amplifier (multi channel systems GmbH, Germany). A Bode diagram was kindly provided by multi channel systems, showing that amplification was stable in the frequency range we were interested in (LFP from 0-250Hz). For the piglets we used an AC coupled amplifier and a sampling rate of 500Hz.
We analyzed the Data with matlab (2015a, Simulink) and with brainstorm . EEG raw data were filtered with digital butterworth filters with a custom written matlab script. The filter was designed with the function butter (n=3rd order). We calculated the normalized cutoff frequency (Wn) for EEG bands delta [0-4Hz], theta [4-8Hz], alpha [8-13Hz], beta [13-30Hz], low gamma [30-80Hz] and high gamma [80-120Hz]. Wn is a number between 0 and 1, where 1 corresponds to the Nyquist frequency which is half the sampling rate (here: 500Hz for down sampled EEG data).
The numerator and denominator values (IIR filter), achieved with the function butter, were used with the matlab function filtfilt to filter the EEG data. For the delta EEG band (0-4Hz), a lowpass was used. We extracted all other EEG frequency bands with a bandpass filter design.
Phase amplitude coupling, Phase Locking Value (PLV) as well as coherence were analyzed with brainstorm software .
To obtain Canolty maps , following procedure was computed (neuroimage.usc. edu/brainstorm/Tutorials/Resting): The EEG was filtered at the low frequency of interest, using a narrow band pass filter. The amplitude troughs of the desired low frequency were detected in the signal. A time window is defined around the detected troughs in order to compute a time frequency decomposition using a set of narrow band-pass filters.
Coherence was calculated as time resolved coherence with an estimation window length of 3500ms and 50% sliding window overlap. The maximum frequency resolution was 1Hz and metric significativity p<0.01. Brainstorm Coherence analysis was based on the matlab function “mscohere”. EEGs from all animals of a group were filtered independently with brainstorm . The delta band was filtered with a low pass filter (0-4Hz). All other bands (theta:4-8Hz, alpha:8-13Hz, beta:13-30Hz, gamma:30-80Hz, hgamma:80-120Hz) were filtered with the appropriate band pass filter. Time resolved coherence was calculated for each animal and each band separately. The results for every EEG band and age group were averaged (arithmetic average, n=5 for P0,P1 and n=6 for P13,P14).The Phase Locking Value (PLV) was calculated as averaged connectivity matrix (one file per age group).
We tested data for distribution with the Lilliefors test (matlab 2015a, Simulink). Thereafter, tests were either performed with the non-parametric Wilcoxon ranksum test or the Kruskal Wallis test and a subsequent multiple comparison test in order to achieve exact statistical relations between groups. All tests are implemented in the matlab statistics toolbox (2015, Simulink).
The main question of this study is, in how far the scalp EEG during the first two weeks of life in mice and pigs does show similar maturational patterns, as for example delta brushes or a transition from discontinuous to continuous EEG, as seen in preterm babies.
In order to use mice as model system for human preterm cortical maturation, we developed miniaturized super cranial fbEEG electrodes useful for long term recordings in neonatal mice with a body weight below 1g. Furthermore, we tested the influence of different anesthesia methods (commonly used in animal experiments) on EEG band power in mice. Recording quality of our super cranial electrodes was compared with electrodes underneath the skull. Mice had an age from P0 to P14, which corresponds to cortical development from human post conceptional day 114 (17th week of gestation, very early preterm) towards post conceptional day 318 (46th week of gestation, post term) (translatingtime.net).
Effect of urethane anesthesia on EEG band power
The comparison between urethane anesthetized and awake, spontaneously active newborn mice (group P0, P1, n=6) revealed no statistically significant difference for the power of EEG frequency bands from 1Hz to 120Hz (p-values (wilcoxon ranksum) from delta to high gamma: 0.7/0.6/0.6/0.24/1/0.1) (Figure 1). Power values for urethane anesthetized mice from delta to high gamma: delta: median[4479μV2/ms], mean[4419μV2/ms] +/- 2474μV2/ms SD; theta: median[0.17μV2/ms], mean[6.76μV2/ms] +/-16.19μV7ms SD; alpha: median[0.17μV2/ms], mean[0.49μV2/ms] +/-0.83 μV2/ms SD; beta: median[0.19μV2/ms], mean[0.19μV2/ms] +/-0.05μV2/ms SD; gamma: median[1.83μV2/ms], mean[2.1μV2/ms] +/-0.76 μV2/ms SD; high gamma: median[0.46μV2/ms], mean[0.48μV2/ms] +/-0.06 μV2ms SD.
Power values for awake mice (local anesthesia): delta: median [2146μV2/ms], mean[5224μV2/ms] +/- 7261μV2/ms SD; theta: median[0.31μV2/ms], mean[0.45μV2/ms] +/- 0.39μV2/ms SD; alpha: median[0.28μV2/ms], mean[0.28μV2/ms] +/- 0.15μV2/ms SD; beta: median[0.26μV2/ms], mean[0.28μV2/ms] +/- 0.13μV2/ms SD; gamma: median[1.77μV2/ms], mean[2.18μV2/ms] +/-1.02μV2/ms SD; high gamma: median[0.7μV2/ms], mean[0.66μV2/ms] +/- 0.23 μV2/ms SD.
EEG pattern during development: from discontinuous to continuous EEG
The transition from discontinuous EEG in newborn mice (age groups P0/P1 and P3/P4, urethane anesthesia) to continuous EEG activity in two week old mice can be seen in Figure 2a-c. Due to our recording system (full band EEG) we are able to record very slow, spontaneously occurring delta waves, with superimposed faster oscillations in P1 mice (Figure 2a). Spontaneous spindles can be seen in a P3 mouse with very sparse background activity, typical for discontinuous EEG (Figure 2b). In two week old animals, the transition towards continuous EEG with ongoing activity is completed (Figure 2c). No such transition can be seen in the EEG of piglets (Figure 2 d-f) from the first day of life towards an age of 2 weeks. Recording site was the somatosensory cortex. A gamma burst can be seen in Figure 2e, recorded from a 4 day old piglet. Apart from traces 2a and 2f, which are low pass filtered (30Hz), all the other traces which are seen in Figure 2 are unfiltered raw data.
Comparison of EEG band power recorded with electrodes on top of the skull and electrodes on top of the Dura
No statistically significant difference can be observed in median EEG band power comparing supracranial electrodes (between skull and Dura)(delta: median[2097μV2/ms], mean[3048 μV2/ms] +/- 3029 μV2/ms SD; theta: median[0.68 μV2/ms], mean[2.22 μV2/ms] +/- 3.99 μV2/ms SD; alpha: median[0.37 μV2/ms], mean[0.41 μV2/ms] +/- 0.32 μV2/ms SD; beta: median[0.19 μV2/ms], mean[0.23 μV2/ms] +/-0.15 μV2/ms SD, gamma: median[1.37 μV2/ms], mean[32.38 μV2/ms] +/-75.88 μV2/ms SD) with electrodes on top of the intact skull (supercranial) (delta: median[233.22μV2/ms], mean[2022μV2/ms] +/- 4151μV2/ms SD; theta: median[0.84μV2/ms], mean[1.76μV2/ms] +/- 2.72μV2/ms SD; alpha: median[0.4μV2/ms], mean[0.38μV2/ms] +/- μV2/ms SD; beta: median[0.27μV2/ms], mean[0.26μV2/ms] +/- 0.09μV2/ms SD, gamma: median[1.21μV2/ms], mean[13μV2/ms] +/-28.85μV2/ms SD) in three and four day old mice (n=6, p-values (wilcoxon ranksum) from delta to gamma: 0.4 /0.7/1/0.3/0.9) (Figure 3). Supra cranial electrodes have higher power by trend. In the high gamma range (80-120Hz), a significant difference in median EEG power between super-(median[0.39μV2/ms], mean[0.58μV2/ms] +/- 0.5μV2/ms SD) and supra cranial electrodes (median [0.28μV2/ms], mean [0.31μV2/ms] +/- 0.11 μV2/ms SD) can be observed for the 95% significance level (alpha=0.05, p-value=0.03). In this case, the EEG recorded with supra cranial electrode does show a higher power in comparison to EEGs recorded with electrodes on top of the skull.
Comparative median EEG band power and amplitudes during the first two weeks of live in mice
Median EEG delta band amplitudes are not statistically significant different between P0/P1 (median[36.35μV], mean[52.68μV]+/-44μV standard deviation (SD)), P3/P4 (median[27.17μV], mean[37.94]+/-34μVSD) and 2 week old animals (P13/P14, median[59.12μV], mean[67.15μV] +/-31V SD) (Figure 4). Between the groups P0P1 and P3P4 no statistically significant difference between median amplitudes can be found for any frequency band. P0P1 mice have statistically significant lower median theta band amplitudes (median[0.48μV], mean[1.42μV] +/-2.3μV SD) in comparison to P13 P14 mice (median[9.61μV], mean[11.26μV] +/-3.3pV SD) at a significance level alpha=0.01. 3 and 4 day old mice have lower theta band amplitudes (median[0.79μV], mean[1.05μV] +/-1μV SD) in comparison to two week old animals. The same statistical relations and significance levels are true for the alpha band with the median EEG amplitudes P0P1 (median[0.48μV], mean[0.63μV] +/-0.4μV SD), P3P4 (median[0.6μV], mean[0.56μV]+/-0.2μV SD) and P13P14 (median[7.36μV], mean[7.41μV] +/-2.1μV SD). For the beta band, P0P1 median amplitudes (median[0.49μV], mean[0.48μV] +/-0.1 μV SD) are statistically significant different for alpha=0.05 in comparison to two week old animals (median[5.27μV], mean[4.92μV] +/-1.4μV SD). Median beta amplitudes in P3P4 age group (median[0.51μV], mean[0.48μV] +/-0.1 μV) are statistically significant different for a significance level of alpha=0.01 in comparison to two week old animals. Median gamma amplitude is not statistically significant different between P0P1 mice (median[1.24μV], mean[1.34μV] +/-0.3μV SD) and P13P14 mice (median[4.1μV], mean[3.96μV] +/-1.2μV SD) whereas P3P4 animals have statistically significant lower gamma band amplitudes (alpha=0.001, median[0.96μV], mean[0.65μV] +/-0.2μV SD) in comparison to two week old animals. Group relations and significance levels for the median high gamma band amplitudesare the same as described for the median beta band amplitudes with following median amplitudes: P0P1 (median[0.71μV], mean[0.74μV] +/-0.1μV SD), P3P4 (median[0.64μV], mean[0.65μV] +/-0.1μVμV SD) and P13P14 (median[1.34μV], mean[1.33μV] +/-0.2μV SD).
Similar results can be observed for normalized median power (power per ms) (Figure 5). Between the groups P0P1 and P3P4 no statistically significant difference can be found for any frequency band. Median delta band power/ms is in P0P1 (median[2036μV2/ms], mean[3118 μV2/ms]+/-3273μV2/ms SD) as well as P3P4 (median[766μV2/ms], mean[2354 μV2/ms] +/-4027μV2/ms SD) mice significantly different from two week old animals (median[3507μV2/ms], mean[5321 μV2/ms] +/-5383μV2/ms SD) at a significance level of alpha=0.05. For the age group P0P1, median theta (median[0.23μV2/ms], mean[6.82 μV2/ms] +/-16.16μV2/ms SD), alpha (median[0.23μV2/ms], mean[0.54 μV2/ms]+/-0.81μV2/ms SD) and beta power (median[0.23μV2/ms], mean[0.23 μV2/ms]+/-0.08μV2/ms SD) are significantly different from two week old animals at a level of alpha=0.01, respectively, (theta: median[92.3 μV2/ms], mean[135.75 μV2/ms]+/-81.03μV2/ms SD), (alpha: median[54.24 μV2/ms], mean[58.58 μV2/ms]+/-32.5μV2/ms SD), (beta: median[27.83 μV2/ms], mean[25.79 μV2/ms] +/-13.14μV2/ms SD). For the same frequency bands, P3P4 mice show significantly different EEG band power in comparison to two week old animals at a significance level of alpha=0.05 (theta: median[0.63μV2/ms], mean[1.88 μV2/ms]+/-3.21μV2/ms SD), (alpha: median[ 0.36μV2/ms], mean[0.36 μV2/ms] +/-0.24μWms SD), (beta: median[0.26μV2/ms], mean[0.24 μV2/ms] +/-0.09μV2/ms SD). Median gamma band power is not statistically significant different between P0P1(median[1.6μV2/ms], mean[1.85 μV2/ms] +/-0.76μV2/ms SD) animals and P3/P4 mice (median[16.81 μV2/ms], mean[16.97 μV2/ms]+/-9.57μV2/ms SD) but there is a highly significant difference between P3P4 mice (median[0.92 μV2/ms], mean[1 μV2/ms] +/-0.41μV2/ms SD) and two week old animals at alpha=0.001. P0P1(median[0.49μV2/ms], mean[0.55 μV2/ms] +/-0.13μV2/ms SD) is statistically significant different in comparison to two week old animals(median[1.79μV2/ms], mean[1.8 μV2/ms] +/-0.57μV2/ms SD) regarding the median power of the high gamma band for a level of alpha=0.05. For the same band, the difference between median power of P3P4(median[0.41μV2/ms], mean[0.43 μV2/ms] +/-0.13μV2/ms SD) age group and two week old animals is statistically significant for alpha=0.01.
In summary, 2 week old animals show a typical exponential decline of band power from slow to fast oscillations, whereas newborn mice show a two peak distribution with relatively high delta and gamma power. The two peak distribution is already lost in 3 and 4 day old mice.
Comparative median EEG band power during the first two weeks of live in piglets
In contrast to mice, no statistically significant differences can be observed in median spectral EEG band power between the age groups P1 (one day old piglets), P4 (four day old piglets), P13/P14 (two week old) in piglets (Figure 6, Table 1). There is a tendency towards a two peak distribution for all the age groups, with relatively high delta and gamma power in comparison to the other frequency bands, especially in newborn piglets.
Major changes in Phase amplitude coupling during the first two weeks of live in mice and piglets
Amplitude modulation of a fast frequency in relation to the phase of a slower frequency gives distinguishable patterns between age groups (Figure 7) in mice. For P0/P1 and a phase frequency of 1Hz, a peak coupling in the theta, alpha and beta range can be seen (Figure 7, 1a). The 1Hz phase frequency is indicated as white sine wave. Despite a coupling between 4Hz phase frequency and a narrow band in the high gamma range as seen in the P0/P1 age group (Figure 7, 1b) there is no coupling between any of the phase frequencies tested here and the gamma band in young animals, neither in the age group P0/P1 nor in the group P3/P4. In contrast, in the age group P13/P14, coupling effects are much stronger, with a stronger phase relation and dominantly in the gamma band (Figure 7, 3a-c). Additionally, in these 2 week old animals, nearly no coupling in lower bands can be seen. Furthermore, increasing phase frequency is associated with increasing frequency of the amplitude frequency in two week old animals. For example, in Figure 7, 3b with a phase frequency of 4Hz, the amplitude modulation does take place in the gamma range from 30 to 80Hz with enhanced amplitudes in relation to the rising and falling phase of the phase frequency. With a phase frequency of 8Hz (Figure 7, 3c) the amplitude modulation does take place between 70 and 140Hz, which is already mainly the high gamma range. In this case amplitude is enhanced with peak and trough of the phase frequency. A proportional relation between phase frequency and amplitude frequency is not true for both young age groups. In P0P1 the amplitude modulation with 8Hz phase frequency does take place in the delta, theta and alpha bands (Figure 7, 1c). In the primary somatosensory cortex of P3P4 mice, the amplitude modulation does occur mainly in the delta and alpha bands with 4Hz phase frequency (Figure 7, 2b) and more pronounced with 8Hz phase frequency (Figure 7, 2c). To sum up, phase amplitude coupling changes from slower to faster amplitude frequencies during development and the coupling pattern itself gains contrast as well as strength during the first 2 weeks of life in mice.
As already seen in the mouse model, gamma band coupling can only be seen in the age groups P1 (Figure 8.1b) and P13/P14 (Figure 8.3a-c) in piglets but not in 4 day old piglets (Figure 8.2a-c). Furthermore, the coupled gamma band is shifted towards higher frequencies (here called “high gamma”, up to 250Hz) values in two week old animals.
Coherence of brain hemispheres during development
Coherence in the EEG delta to beta band is much stronger in newborn mice (P0/P1, n=5) in comparison to two week old animals (P13/P14, n=6, Figure 9). Two week old mice show mainly coherence between corresponding regions: For the Barrel Cortex in the delta band and for the forelimb region of the primary somatosensory Cortex in the theta band. Similarly, newborn mice (P0/P1) do also have the strongest coherence between left and right Barrel cortex in the delta range and the strongest correlation between left and right forelimb region of the primary somatosensory Cortex in the theta range. In contrast to two week old animals, P0/P1 mice show strong intrahemispheric coherence in the delta range between ipsilateral Barrel Cortex and forelimb region of the primary somatosensory Cortex. In the gamma as well as high gamma band of newborn mice, coherence is only apparent between hemispheres. The opposite is true for two week old animals: only ipsilateral, right hemispheric gamma band coherence.
Phase locking value in the somatosensory Cortex during development
Phase locking values in the somatosensory Cortex are stronger for newborn mice (P0/P1) in comparison to two week old animals, especially in the EEG delta and gamma band (Figure 10). For newborn mice, no clear trend towards ipsi- or contralateral phase locking is visible. In the high gamma range (newborn mice, P0/P1), phase locking values are by far weaker than in the gamma band with strongest phase locking occurring between hemispheres. Phase locking values for two week old animals are gradually declining from EEG delta band to the high gamma band. This is in contrast to newborn mice (P0/P1), where maximal phase locking values occur in the gamma range. Similarly to newborn mice, Phase locking for the lower EEG bands in two week old mice is apparent between adjacent regions of the primary somatosensory Cortex (Barrel Cortex and forelimb region) as well as between cortical hemispheres. For the faster EEG bands (beta to high gamma), phase locking is occurring particularly between hemispheres.
Our main finding is, that the fbEEG of neonatal mice show similarities to preterm human full band EEG patterns as for example less pronounced hemisphere coherence and a discontinuous EEG with spontaneous activity transients . We were able to show fundamental changes in the EEG spectral band power and network connectivity patterns during a phase of maturation in mouse pubs, which corresponds to the critical transition period from very early preterm to full term in human infants. We observed similar results for piglets.
Why an animal model system for preterm birth at all?
In first instance, because the measurement of EEG in preterm infants is limited to cases with clinical indication. Since very early preterm birth is a life threatening status , other physiological parameters, as for example breathing, have priority and make EEG acquisition in the neonatal intensive care unit challenging . Nevertheless, it turned out that former preterm infants have not only a statistically increased risk for cognitive impairments throughout their later lives in comparison to term born adults, they have also corresponding alterations in brain physiology and anatomy  . Hence, there is an urgent need to distinguish “normal” developmental EEG patterns from pathophysiologic predictors of cognitive or other impairments during later life. In a first mathematical approach, such patterns have been identified from EEG measurements .
Influences of analgesic treatment on EEG
A general problem are medication associated EEG alterations . Preterm babies underwent very often some kind of analgesic treatment   whereas laboratory animals most often receive anesthesia. We were able to show for newborn mice, that the spectral band power and amplitude of EEG recordings is not statistically significant altered by using a light urethane anesthesia in comparison to awake animals (fig.1). Furthermore we could show certain typical preterm EEG patterns as for example delta brushes or spindle bursts during urethane anesthesia (Figure 2). Spindle bursts under light urethane anesthesia have also been described in the EEG of rat pubs .
Adaptation to the clinical context: non invasive fbEEG electrodes for newborn mice with a body weight below 1g
We recorded the EEG in mice with non invasive skull electrodes in order to come closer to the recording situation on the neonatal intensive care unit. To our knowledge, this has been done for the first time in newborn mice. We validated our newly designed method by comparing the power spectral density between surface and conservative invasive electrodes on top of the Dura, as they are usually used in the scientific context for animal experiments. Despite the high gamma band, we could not find any statistically significant differences between invasive and the newly designed non-invasive electrodes (Figure 3). This might be due to the thin and soft skull in young mice and an associated very high conductivity, as has been proposed for neonates .
Power spectral density as developmental marker
Very slow oscillations play an important role during early cortical development . We used a full band EEG recording system to measure slow delta potentials. In fact, there is a high proportion of delta band power in the EEG of newborn mice (Figure 5). In comparison to two week old mice, which have exponentially decreasing power as well as amplitude values (Figure 4) from slow to fast oscillations, newborn mice (P0/P1) show a two peak distribution (Figure 5). They have relatively high amount of delta power as well as gamma. This pattern is already resolved in three day old mouse pubs, which have very low power for all bands but delta. One explanation might be the shift from a subplate driven, gap junction coupled syncytium, towards a network which is mainly driven by chemical synapses . It has been shown, that gap junction coupled ensembles are capable of producing oscillations of high frequencies, as for example beta and gamma . Therefore, the EEG gamma band in newborn mice is most likely of different physiologic origin than the gamma band in two week old animals.
Power spectral density in newborn piglets differs somewhat from mice regarding differences between age groups. Piglets do not show significant differences between the age groups P1, P4 and P13/P14. Nevertheless, there is a similar two peak distribution with relatively high gamma as well as delta power in comparison to the other frequency bands for all the piglet age groups in contrast to mice, where this kind of spectral EEG power distribution can only be observed for the younger age groups. One reason might be that piglets do not develop as fast as mice so that it is impossible (in the given developmental window of this study) to observe the transition towards an exponential decay of power from slow to fast EEG band oscillations, as seen in two week old mice. We do not know the exact developmental correlation between age groups in mice and piglets. But one day old piglets show EEG patterns similar to K-Komplexes which are occurring in human from the third month of life onwards .
Does this mean, mouse pubs do not show any activity in the theta to beta band?
We used phase amplitude coupling analysis to compare the extend of nested processes between newborn mice, which correspond to human very early preterm and two week old mice corresponding to approximately the 46th week of gestation in human regarding EEG (translatingtime.net) which is already post term in human. The computed modulation index indicates, whether a faster wave changes its amplitude (power) when superimposed on top of a slower wave. With other words, whether the amplitude of the fast oscillation is dependent on the phase of a slower oscillation. This network phenomenon is also known as cross frequency coupling  or nesting coefficient .
To come back to the initial question, we observed most phase amplitude coupling events between delta and the faster EEG bands theta, alpha and beta in newborn mice (Figure 7, 1a-2c). Taking our power analysis into account it seems, as if these intermediate bands do occur mostly during nested events. This does make sense, because preterm EEG has sparse overall activity with spontaneous activity transients  which are most often very slow delta oscillations with nested faster oscillations, also called delta brushes . Surprisingly, we found also coupling patterns in the high gamma range in newborn mice but not in three day old mouse pubs. This does fit to our spectral power analysis, where P0P1 mice have higher power in the gamma as well as fast gamma range by trend in comparison to 3 day old mice. Since the coupling pattern has no similarity with corresponding theta gamma coupling we observed in two week old mice, the fast gamma band in newborn mice might be of different physiological origin, as already discussed for the power analysis. The EEG gamma band in adult mice is associated with GABAergic network activity, especially PV interneurons . Since the physiology of GABA changes drastically during early cortical development in mice   and human  this assumption is very likely. And it might be a precise indicator of development or mal-development. Prior to GABA driven gamma oscillations, gap junction coupled subplate networks  may induce very early gamma rhythms in the cortex of newborn mice, measured here with non invasive EEG electrodes. In two week old mice, delta band is less involved in PAC (phase amplitude coupling) patterns. This is in contrast to all younger mice we recorded (P0/P1 and P3/P4) and in good agreement with data from preterm infants, published recently. It has been shown, that PAC is decreasing for nesting frequencies in the delta range and nested frequencies from 10-32Hz during the first two weeks of life in full term infants . From our data it becomes clear, that two week old animals have higher nested frequencies, the higher the nesting frequency becomes. Also for 1Hz nesting frequency, the main nested frequencies lie well above 30 Hz. In contrast, in newborn mice, there is no proportional relation between nested and nesting frequency. Higher nesting frequencies result also in delta coupling as seen in the Canolty maps (Figure 7: 1c, 2c). This is not the case for two week old animals (Figure 7: 3b,3c).
We think, that these differences in the phase amplitude coupling pattern represent drastic network changes during this critical period of early cortical development. This is not only in good agreement with data from newborn infants where delta associated PAC decreases during maturation  but it is also in line with the occurrence of delta brushes in newborn rodents and preterm human infants  as well as with theta gamma PACs, seen in the EEG of adult rodents  and human . Furthermore, we were able to show similar amplitude coupling patterns for piglets. Gamma coupling is only occurring in 1 day old (Figure 8.1b) and two week old piglets (Figure 8.3a-c) but not in 4 day old piglets (Figure 8.2a-c). Additionally, the coupled gamma band changes to higher frequencies during development (Figure 8.3a-c). Hence, it might be interesting to use phase amplitude coupling as diagnostic tool for brain maturation in the clinical context.
Developmental changes in hemispheric coherence
It is known that hemispheric coherence in preterm infants is probably brainstem driven up to the 35th week of gestation, before callosal hemisphere connections are functional . Up to 28 weeks of gestational age synchronization between brain hemispheres is strong for high amplitude potentials . After a desynchronized phase, synchronous activity is reactivated from a gestational age of 31 weeks onwards . This might be an proximate explanation for the relatively high absolute values of interhemispheric coherence in newborn mice in comparison to two week old animals (Figure 9). Ultimately, this phenomenon might be explained by the syncytial, gap junction coupled network in younger mice in comparison to the fine tuned network, driven by chemical synapses which is already established during later stages of development in two week old mice . Another explanation could be the extended distance between electrodes in older animals, because coherence may decrease with increasing distance . Nevertheless, other authors do not find volume conduction effects for coherence but an asymmetric pattern for coherence values between different anatomical locations and head axis . For newborn mice, coherence is strongest for the delta band, without a clear preference towards ipsilateral or contralateral (between hemispheres) coherence. This is in strong contrast to two week old animals, where coherence is strongest for corresponding areas of the primary somatosensory cortex between hemispheres (Figure 9). Interestingly, for the Barrel Cortex in the delta band and for the forelimb region in the theta band. It has been shown that EEG coherence is occurring in topographically distinct frequency bands . For comparable vigilance states, coherence might therefore be an interesting tool to tag neurodevelopment with topographical precision.
High degree of phase locking in newborn mice
We compared the phase locking values of distinct frequency bands between neonatal mice (P0/P1) and two week old animals (Figure 10). As seen for the PAC, absolute phase locking values are higher for newborn mice. Two week old animals show most phase locking in the delta band, whereas newborn mice have most phase locking in the gamma band. This is a surprising result, since the high frequency component of the EEG is most often not recorded in the clinical context. Our data indicate, that it might be advantageous to extend neonatal clinical EEG recordings to frequencies up to 100Hz. The two peak distribution in the delta and gamma band in newborn mice is hence not only a quantitative phenomenon of our power spectral density calculation (Figure 5) but also of functional relevance regarding cortical connectivity at an early stage of cortical development. Gamma driven phase locking between Hippocampus and prefrontal cortex has been shown to occur during early development in rat pubs .
Taken together, the mouse as well as the piglet might be promising model systems for clinical (non-invasive) preterm brain electrophysiology and development. Simple calculations as for example power spectral density as well as network analysis, like phase locking or phase amplitude coupling show stable differences between age groups, which are corresponding to very early preterm until post term in human. We established a non invasive fbEEG method for newborn mice in order to adapt to clinical methods in a better way. The raw traces of these super cranial EEG traces in developing mice nicely illustrate the transition from discontinuous to continuous EEG (Figure 2a-c). Hence, the mouse might be a powerful model system to understand early EEG patterns in a better way and especially, to distinguish between normal EEG patterns and indicators of future dysplasia in the preterm EEG. The piglet is partially also a good model system, keeping in mind that piglets, as all undulates, are precocious. We were not able to see a discontinuous EEG pattern in newborn piglets (Figure 2d-f) but instead k-komplex like EEG patterns, which are seen in human from an age of several month on . This might either be due to the precocious nature of piglets or it might be due to the fact that piglets are simply much more mature during birth in comparison to mice. The last possibility would indeed mean, that piglets are very much closer to human in comparsion to mice, regarding cortical maturation. Therefore, the combination of both model systems might be ideal to study clinical patterns of cortical maturation in the scientific context.
This work was supported by a grant from the federal ministry of education and science (BMBF funding no 13GW0033B) in Germany, the federal ministry for economic affairs and energy and the European social fund (funding no 03EFJBE108 EXIST).
Howson C.P., Kinney M.V., McDougall L., Lawn J.E. Born Too Soon: Preterm birth matters” Reproductive Health, 2013, 10 (Suppl 1) : S1 Google Scholar
Nosarti C., Nam K.W., Walshe M., Murray R.M., Cuddy M., Rifkin L., Allin M.P., Preterm birth and structural brain alterations in early adulthood. Neuroimage Clinical, 2014, 6: 180-191 Web of SciencePubMedCrossrefGoogle Scholar
Stevenson N.J., Palmu K., Wikstrom S., Hellstrom-Westas L., and Vanhatalo S., Measuring brain activity cycling (BAC) in long term EEG monitoring of preterm babies. Physiological Measurement, 2014, 35: 1493-1508. Web of ScienceCrossrefPubMedGoogle Scholar
Tolonen M., Palva J.M., Andersson S., and Vanhatalo S., Development of the spontaneous activity transients and ongoing cortical activity in human preterm babies. Neuroscience, 2007, 145: 997-1006. CrossrefWeb of SciencePubMedGoogle Scholar
Vanhatalo S., Tallgren P., Andersson S., Sainio K., Voipio J., and Kaila K., DC-EEG discloses prominent, very slow activity patterns during sleep in preterm infants. Clin.Neurophysiol., 2002, 113: 1822-1825 CrossrefPubMedGoogle Scholar
Clancy B., Finlay B.L., Darlington R.B., and Anand K.J. Extrapolating brain development from experimental species to humans, Neurotoxicology, 2007, 28: 931-937. Web of ScienceCrossrefPubMedGoogle Scholar
Yang J.W., Hanganu-Opatz I.L., Sun J.J., and Luhmann H.J. Three patterns of oscillatory activity differentially synchronize developing neocortical networks in vivo. Journal of Neuroscience, 2009, 29: 9011-9025 CrossrefWeb of ScienceGoogle Scholar
Conrad M.S., Sutton B.P., Dilger R.N., Johnson R.W. An In Vivo Three-Dimensional Magnetic Resonance Imaging-Based Averaged Brain Collection of the Neonatal Piglet (Sus scrofa). PLoS ONE, 2014, 9(9):e107650. Web of ScienceCrossrefGoogle Scholar
Iyer K.K., Roberts J.A., Hellstrom-Westas L., Wikström S., Hansen-Pupp I., Ley D., Vanhatalo S., and Breakspear M. Cortical burst dynamics predict clinical outcome early in extremely preterm infants. Brain, 2015, 138: 2206-2218. PubMedCrossrefWeb of ScienceGoogle Scholar
Tokariev A., Videman M., Palva J.M., Vanhatalo S. Functional Brain Connectivity Develops Rapidly Around Term Age and Changes Between Vigilance States in the Human Newborn. Cerebral Cortex, 2015, 26(12): 4540-4550 Web of ScienceGoogle Scholar
Chan D.W., Yamazaki M., Akiyama T., Chu B., Donner E.J., Otsubo H. Rapid oscillatory activity in delta brushes of premature and term neonatal EEG. Brain Dev, 2010, 32(d): 482-486. Web of ScienceCrossrefPubMedGoogle Scholar
Lapray D., Bergeler J., Dupont E., Thews O., and Luhmann H.J., A novel telemetric system for recording brain activity in small animals. In Telemetry: Research, Technology and Applications, D. Barculo and J. Daniels, eds. Nova Science Publishers), pp. 195-203, 2009 Google Scholar
Canolty, R. T. et al. “High gamma power is phase-locked to theta oscillations in human neocortex.” Science 313.5793 (2006): 1626-28. Google Scholar
Meng C., Bäuml J.G., Daamen M., Jaekel J., Neitzel J., Scheef L., Busch B., Baumann N., Boecker H., Zimmer C., Bartmann P., Wolke D., Wohlschläger A.M., Sorg C., Extensive and interrelated subcortical white and gray matter alterations in preterm-born adults. Brain Struct Funct., 2015, 221(4): 2109-21 PubMedWeb of ScienceGoogle Scholar
Kahn D.J., Richardson D.K., Gray J.E., Bednarek F., Rubin L.P., Shah B., Frantz I.D. 3rd, Pursley D.M. Variation among neonatal intensive care units in narcotic administration. Arch Pediatr Adolesc Med, 1998, 152: 844-851. PubMedGoogle Scholar
Steinhorn R., McPherson C., Anderson P.J., Neil J., Doyle L.W., Inder T., Neonatal Morphine Exposure in Very Preterm Infants— CerebralDevelopment and Outcomes. J Pediatr, 2015, 166: 1200-1207 CrossrefPubMedGoogle Scholar
Odabaee M., Tokariev A., Layeghy S., Mesbah M., Colditz P.B., Ramon C., and Vanhatalo S., Neonatal EEG at scalp is focal and implies high skull conductivity in realistic neonatal head models. Neuroimage, 2014, 96, 73-80 PubMedCrossrefWeb of ScienceGoogle Scholar
Dupont E., Hanganu I.L., Kilb W., Hirsch S., and Luhmann H.J. Rapid developmental switch in the mechanisms driving early cortical columnar networks. Nature, 2006, 439: 79-83. CrossrefPubMedGoogle Scholar
LeBeau F.E., Traub R.D., Monyer H., Whittington M.A., and Buhl E.H. The role of electrical signaling via gap junctions in the generation of fast network oscillations. Brain Res.Bull., 2003, 62, 3-13 CrossrefPubMedGoogle Scholar
Kirchberg, S., ENTWICKLUNG VON K-KOMPLEXEN ALS ELEMENT DES SCHLAFSTADIUMS NREM 2 IM KINDESALTER, PhD thesis, Friedrich Schiller University Jena, Germany, 2015 Google Scholar
Tort A.B., Komorowski R., Eichenbaum H., Kopell N. Measuring Phase-Amplitude Coupling Between Neuronal Oscillations of Different Frequencies. J Neurophysiol, 2010, 104: 1195–1210 CrossrefPubMedWeb of ScienceGoogle Scholar
Vanhatalo S., Palva, J.M., Andersson, S., Rivera, C., Voipio, J., and Kaila, K., Slow endogenous activity transients and developmental expression of K+-Cl-cotransporter 2 in the immature human cortex.Eur.J Neurosci., 2005a, 22: 2799-2804 CrossrefGoogle Scholar
Milh M., Kaminska A., Huon C., Lapillonne A., Ben-Ari Y., and Khazipov R., Rapid cortical oscillations and early motor activity in premature human neonate. Cerebral Cortex, 2007, 17: 1582-1594 Web of ScienceCrossrefGoogle Scholar
Kim T., Thankachan S., McKenna J.T., McNally J.M., Yang C., Choi J.H., Chen L.C., Kocsis B., Deisseroth K., Strecker R.E., Basheer R., Brown R.E., and McCarley R.W. Cortically projecting basal forebrain parvalbumin neurons regulate cortical gamma band oscillations. Proceedings of the National Academy of Sciences of the United States of America, 2015, 112: 3535-3540. CrossrefPubMedWeb of ScienceGoogle Scholar
Kirmse, K., Kummer, M., Kovalchuk, Y., Witte, O., W., Garaschuk, O., Holthoff, K., GABA depolarizes immature neurons and inhibits network activity in the neonatal neocortex in vivo, Nature Communications, 2015, 6: 7750 Web of SciencePubMedCrossrefGoogle Scholar
Rivera C., Voipio J., Payne J.A., Ruusuvuori E., Lahtinen H., Lamsa K., Pirvola U., Saarma M., and Kaila K. The K+/Cl- co-transporter KCC2 renders GABA hyperpolarizing during neuronal maturation. Nature, 1999, 397: 251-255 CrossrefGoogle Scholar
Wang J., Li D., Li X., Liu F.Y., Xing G.G., Cai J., Wan Y., Phase–amplitude coupling between theta and gamma oscillations during nociception in rat electroencephalography. Neuroscience Letters, 2011, 499: 84-87 PubMedWeb of ScienceCrossrefGoogle Scholar
Andre M., Lamblin M.D., d’Allest A.M., Curzi-Dascalova L., Moussalli-Salefranque F., Tich S.N.T., Vecchierini-Blineau M.F., Wallois F., Walls-Esquivel E., and Plouin P. Electroencephalography in premature and full-term infants. Developmental features and glossary. Neurophysiologie Clinique-Clinical Neurophysiology, 2010, 40: 59-124. CrossrefWeb of SciencePubMedGoogle Scholar
Thatcher R.W., Krause P.J., Hrybyk M., Cortico-Cortical Associations and EEG Coherence: A Two-Compartmental Model. Electroencephalography and clinical Neurophysiology, 1986, 64: 123-143. PubMedCrossrefGoogle Scholar
Brockmann M.D., Poschel B., Cichon N., and Hanganu-Opatz I.L. Coupled oscillations mediate directed interactions between prefrontal cortex and hippocampus of the neonatal rat. Neuron, 2011, 71: 332-347. Web of SciencePubMedCrossrefGoogle Scholar
About the article
Published Online: 2017-12-29
Citation Information: Translational Neuroscience, Volume 8, Issue 1, Pages 211–224, ISSN (Online) 2081-6936, DOI: https://doi.org/10.1515/tnsci-2017-0029.
© 2017 Nora Vanessa de Camp et al.. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. BY-NC-ND 3.0