Animals. In accordance with the policy of Grenoble Alpes University and the Grenoble Institut of Neurosciences (B3851610008) and with French legislation, experiments were performed in compliance with the European Community Council Directive of 2010 (2010/63/UE) for care of laboratory animals and the ARRIVE guidelines. All procedures were reviewed and validated by the “Comité éthique du GIN n˚004” and was authorized by the Direction Départementale des Services Vétérinaires de l’Isère – Ministère de l’Agriculture et de la Pêche, France. We used three male adult monkeys (Macaca fascicularis - Mauritius) weighed 6-10kg and were 5-10 years. Animals were kept under controlled conditions, 12h light/dark cycles, 23 ± 2°C, and 50 ± 5% humidity. Animals were pair housed, had access ad libitum to food and water and supplemental fresh fruit was given once a day.
Apparatus. Monkeys were implanted with a polysomnographic equipment, a radio-telemeter transmitter (D70-EEE, Data Science International, France) for long-term recording in freely moving animals. The transmitter had three channels biopotential for recording electroencephalogram (EEG), electro-oculogram (EOG), and electromyogram (EMG) signals with a sampling rate of 500 Hz and a gain of 75 and actimetry count. Signals were acquired via two receivers mounted on the home cage and behavioral cage and then forwarded to a data exchange matrix connected to a computer for data storage and off-line analysis.
Surgery. The surgery was performed under aseptic conditions and general anesthesia. Animals were anesthetized with Ketamine (7mg/kg, i.m.) and Xylazine (0.6mg/kg, i.m.) then intubated and switched to isoflurane mixed with oxygen. Animals were spontaneously breathing. Respiration rate, Et-CO2 and O2 saturation were monitored with a Comdek MD-660P monitor. Saline solution (NaCl 0.9%) was infused intravenously all along the surgery for drug access and hydration. Analgesic therapy (Ketoprofen 2mg/kg i.m.) were provided during one-week post-operative period. The radio-telemeter transmitter was implanted within a subcutaneous pocket in their back and the electrode leads were tunneled to the skull. EEG was recorded using two electrodes screwed unilaterally (one frontal and one parietal) into the skull, EOG was acquired from two electrodes affixed unilaterally at the orbital arch bone (one at the top and one at the external side) and EMG was monitored from two leads sutured into the neck musculature. The reference was fixed on the skull at the left occipital level.
MPTP treatment. Monkeys were intoxicated by intramuscular injection of MPTP under light anesthesia (Ketamine 2-4mg/kg). A progressive protocol was used to obtain a presymptomatic phase, i.e. before the motor symptoms, consisting in small doses of MPTP (0.2-0.5mg/kg, in NaCl 0.9%) at two-weeks interval until the parkinsonian symptoms were stable. M1 received 18 injections (7.55mg/kg total), M2 received 8 injections (2.2mg/kg total) and M3 received 18 injections (6.1mg/kg total). All animals had comparable state of stable parkinsonism.
Motor score. The severity of parkinsonism was evaluated before, during and after MPTP intoxication in the home cage, using a rating scale, combining the most recurring items from eight commonly used parkinsonism scales49. This scale includes eight clinical symptoms (general activity, frequency of each arms movements, posture, bradykinesia, tremor, feeding, freezing and vocalization), rated between 0 (normal) and 2-3 (maximal disability), with a total score out of 25. Evaluations were performed by the same observer once a week at 2 pm for 15min. The spontaneous activity was quantified by the implanted radio-telemeter transmitter and expressed in counts/min. Activity values are collected every 10 seconds, according to the changes of the signal perceived by the receivers mounted in the cage. If the animal is stationary the signal strength remains constant and equal to zero and if the animal moves the signal strength changes and is counted as movement.
DaTscan. In addition to this clinical motor evaluation, the first animal (M1) had an individual follow-up of the striatale dopaminergic system, all along the induction of the parkinsonian syndrome, which gives a global idea of the impact of MPTP on the dopamine degeneration. The fixation of the Ioflupane (123I) tracer, a radioligand with a high affinity for presynaptic dopamine transporters, was quantified in the striatum to observe the dopamine reduction. For all the exams, the animal was anesthetized and a dose of 90 MBq was injected intravenously 3h before the acquisition. Then, the images which had the highest radioactivity count, and the one above and below (three cuts) were selected and averaged. Regions of interest were drawn manually for each striatum and an area of occipital cortex served as a reference for the background noise (region of non-specific 123I uptake). The final activity count was determined as the ratio of striatal activity to occipital activity.
Sleep data analysis. Sleep scoring was performed offline on a software (NeuroScore, Data Science international). EEG and EOG was bandpass-filtered in the range of 0.3 to 35Hz and EMG was bandpass-filtered in the range of 10 to 100Hz. Sleep stages were manually determined according to the American Academy of Sleep Medicine criteria and performed in 30s epoch50. Different stages were identified: active wake (A), quiet wake (W), and non-REM sleep: light sleep stage 1 (S1) and stage 2 (S2), deep sleep stage 3 (S3) and REM sleep (R) (Fig. 4). Movement and chewing artefacts were mostly produced during active wake and were correlated with the simultaneous video observation. EEG power spectral analyses were performed for all scored 30s epochs, and for the different stages of wakefulness and sleep, to verify whether the visual scoring is consistent with the expected frequency bands (Fig. 4). All epochs with artefacts were excluded from spectral analysis.
EMG and EEG spectral analysis. Power spectral density analysis was performed on EMG signals specifically during active wakefulness and REM sleep and on EEG signals specifically during active wakefulness and deep sleep S3. This analysis was performed in healthy, presymptomatic, and symptomatic states to observe any changes between these different conditions. Thus, 300 samples (100 samples for each animal) from random 30s epochs for these different wake and sleep stages and in the different conditions were analyzed. For each 30s epoch, an estimate of the power spectral density was computed using the Fast Fourier Transform and Hamming windowing technique and averaged (Matlab). Data are expressed in mean ± 95% confidence interval.
Nighttime sleep quality evaluation. Nighttime sleep data were analyzed from 7 pm to 7 am. Two 12h recording sessions per week were performed until a total of ten recordings were reached at minimum, in each experimental condition. For each 12h period of nighttime, the sleep latency (SL) (min), total sleep time (TST) (min), duration of each stage (%), wake time after sleep onset (WASO) (min) and sleep efficiency (expressed in %, as the ratio of TST to the 12h nocturnal time) were calculated.
Daytime hypersomnia evaluation. Assessment of hypersomnia was done by long-term recording of 12h from 7 am to 7 pm. Two 12h recording sessions per week were performed until a total of ten recordings were reached at minimum, in each experimental condition. For each 12h period of daytime, the first sleep after wake, the TST and the duration of each stage (%) were calculated.
Daytime sleepiness evaluation. Daytime sleepiness was evaluated using a modified multiple sleep latency test (mMSLT), performed 2h after waking up. In quiet room, the lights were turned off 3 times for a duration of 20min, 1h apart (light-OFF sessions at 10 am, 11 am and 12 am). mMSLT was performed twice a week in healthy state until a minimum of ten experimentations were acquired and all along the parkinsonian syndrome induction for a minimum of ten experimentations in presymptomatic and symptomatic state. SL was determined if a 30s epoch of scorable sleep was observed. If no sleep onset was observed, SL was designated to be 20min as used in human and NHP studies51,52. The two wake stages A and W were pooled together and the two light sleep stages S1 and S2 were pooled together. The following parameters were calculated for each 20min period of light-OFF session: SL (min), duration of wake and sleep (min).
REM sleep heart rate variability analysis. 12 epochs of 2min of EMG during REM sleep were randomly pick up including the perception of the cardiac R peak which gives us an electrocardiogram-like. For each epoch, beat-to-beat (RR) variability was analysed in the time domain including the mean RR (meanNN) interval and its standard deviation (StdNN) and the coefficient of variation of RR intervals |[StdNN/MeanNN] x 100| showing the relative extent of the data. Poincaré plot, which portrays the relationship between successive RR intervals (RRn+1 interval plotted against the preceding one RRn) was done. This plot provides summary information about RR interval and density distribution which are considered as general measures of heart rate variability.
Statistical analysis. Standard statistical methods using GraphPad Prism 8 were applied. A Kruskal–Wallis test followed by Dunn’s multiple comparisons test was used for the comparison of sleep parameters at healthy, presymptomatic and symptomatic states. Data are presented as mean ± standard error of the mean (SEM) and the statistical significance was considered at a probability (p) value ≤0.05.