Levodopa affects spike and local field synchronisation in the pedunculopontine nucleus of a rat model of Parkinson’s disease

Loss of dopaminergic neurones in the SNc following the 6-OHDA lesion

Neurone loss in SNc following the 6-OHDA lesion was confirmed by cresyl violet and TH immunohistochemical staining. As shown in Figure 2A–2F, TH-positive neurones were significantly depleted in the SNc ipsilateral to the 6-OHDA lesion. Cresyl violet staining revealed that remnant neurones on the lesioned side displayed altered morphology compared to those on the control side. Histological analysis showed severe 6-OHDA-induced lesion of SNc dopaminergic neurones, confirming successful establishment of the disease model.

Figure 2. (A–D) Exemplar images of cresyl violet-stained coronal sections of the substantia nigra pars compacta of animals with 6-hydroxydopamine (6-OHDA) lesions. (A) control hemisphere, 100×; (B) lesioned hemisphere, 100×; (C) control hemisphere, 400×; (D) lesioned hemisphere, 400×. (E, F) fluorescence immunohistochemical staining of dopaminergic neurones for tyrosine hydroxylase in the substantia nigra pars compacta. (E) Control hemisphere, 200×; (F) lesioned hemisphere, 200×. (G) Schematic representation of electrode positioning in rats of different groups. (H) Image of a coronal section of PPN stained with cresyl violet. The location of the electrode tip is indicated by a red arrowhead. (I) Schematic representation of the electrode path (red line) into the pedunculopontine nucleus (PPN). (J) Walking time along a control ladder in lesioned and levodopa (L-DOPA) rats. (K) Latency in the rotarod test. (L) distance in the rotarod test. (M) Number of apomorphine (APO)-induced rotations after 30 min in lesioned and L-DOPA rats. * p < 0.05, *** p < 0.001, **** p < 0.0001 compared to the control group; # p < 0.05 compared to the lesioned group.

L-DOPA treatment improved locomotion in animals with PD

In this study, only rats with correct electrode positioning were used for analyses. A total of 9 control, 11 lesioned, and 9 L-DOPA rats were examined. For these animals, cresyl violet-stained PPN sections with electrode tip sites are shown in Figure 2G. An image of a coronal slice as well as the electrode path are shown in Figure 2H, 2I.

The walking time along the ladder, latency, distance in the rotarod test, and APO-induced rotations after 30 min were calculated to assess locomotor capacity. In the walking test shown in Figure 2J (F (2,26) = 11.21, p = 0.0003 (one-way ANOVA test)), lesioned rats required more time to walk along the ladder compared to control rats (p = 0.0002). This effect was reversed upon L-DOPA treatment (p = 0.0290). In the rotarod test shown in Figure 2K, 2L (F (2,26) = 40.94, p < 0.0001 for latency; F (2,26) = 40.93, p < 0.0001 for distance), the shorter latency (p < 0.0001) and distance (p < 0.0001) of the lesioned animals were improved upon L-DOPA treatment (p = 0.0256 and 0.0256, respectively). Animals in the L-DOPA group showed fewer APO-induced rotations (Figure 2M) than rats in the lesioned group (t = 2.225, df = 18, p = 0.0391). These results indicate that L-DOPA treatment significantly improved the locomotor behaviour of animals with experimental PD.

Neuronal classification in PPN

From the recordings of nine control rats in a resting state, 57 neurones were collected and classified into two distinct categories (36 Type A and 21 Type B) according to the duration (trough-to-peak) and shape of the waveforms and firing properties. As shown in Figure 3, Type A neurones exhibited a narrower duration with a “V” shape waveform (Figure 3A), higher firing rate, and more irregular firing pattern than Type B neurones (Figure 3B, 3D). From the principal component analysis (PCA) results in 3D view (two neurone types recorded by one electrode channel) in Figure 3C, the two neurone types showed a distinct distribution according to the three main components (F (3,3352) = 6.005, p < 0.0001). Moreover, as shown by the inter-spike intervals (ISI) histograms (Figure 3E, 3F), Type A neurones showed a Poissonian tonic distribution and Type B neurones showed a concentrated distribution, respectively, indicating distinct firing patterns. The results of PPN neurone classification in lesioned and L-DOPA-treated rats in the resting and locomotion state were as described above (data not shown).

Figure 3. Neuronal classification results in the pedunculopontine nucleus (PPN) (red: Type A; blue: Type B). (A) Averaged waveforms of each neuronal type. (B) Representative signals of each neuronal type. (C) Principal component analysis (PCA) results in 3D view. (D) Spike duration (trough-to-peak) and firing rate of 36 Type A and 21 Type B neurones. (E) Representative inter-spike interval histogram of a Type A neurone. (F) Representative inter-spike interval histogram of a Type B neurone.

Effect of L-DOPA on the association between spiking activities and LFP in the PPN

Coherence value

To evaluate the effect of 6-OHDA lesions and L-DOPA treatment on the correlation between LFP and spiking activities in the PPN, we analysed the coherence value, SFC, and phase-lock. As shown in Figure 4A–4D, alterations in PPN Type A neurones in the resting state were detected in alpha (F (2,117) = 5.011, p = 0.0082) and beta frequency bands (F (2,117) = 5.879, p = 0.0044): 6-OHDA (n = 54) led to significantly increased coherence values between PPN Type A spikes and LFP in alpha (7–12 Hz) (p = 0.0081) and beta (12–30 Hz) (p = 0.0052) frequency bands compared to controls (n = 36). Notably, treatment with L-DOPA (n = 30) failed to reverse these changes (p = 0.9401, 0.9991 compared to in the lesioned group).

Figure 4. Measurement of the coherence value between spikes and local field potential (LFP) in the pedunculopontine nucleus (PPN). (A) Frequency-coherence value plots in resting Type A neurones. (B) Frequency-coherence value plots in resting Type B neurones. (C) Statistical results of coherence value in resting Type A neurones. (D) Statistical results of coherence value in resting Type B neurones. (E) Frequency-coherence value plots of Type A neurones in locomotion state. (F) Frequency-coherence value plots of Type B neurones in locomotion state. (G) Statistical results of coherence value obtained in Type A neurones in locomotion state. (H) Statistical results of coherence value obtained in Type B neurones in locomotion state. The frequency bands with the most noticeable changes are highlighted via blue boxes. *p < 0.05, **p < 0.01 in comparison to the control group.

In the locomotion state (Figure 4E–4H), PPN Type A neurones in the theta band (F (2,68) = 6.245, p = 0.003), and PPN Type B neurones in the alpha band (F (2,47) = 6.347, p = 0.0047) showed altered coherence values between groups. The coherence value of the theta band was higher in the lesioned group (n = 29) than in the control group (n = 22, p = 0.0329). L-DOPA administration had no effect (n = 20, p = 0.5080 compared to the lesioned group). In addition, the coherence value of lesioned Type B neurones in the alpha band (n = 18, p = 0.0043) was higher compared to that of control rats (n = 17), and L-DOPA did not reverse this effect (n = 15, p = 0.8441 compared to lesioned rats).

SFC

Regarding the SFC in the resting state (Figure 5A–5D), Type A neurones in the beta frequency band showed differences between groups (F (2,117) = 11.942, p = 0.0003). Lesioned rats had a higher SFC than control rats (p = 0.0005), and L-DOPA treatment reversed this effect (p = 0.0002 compared with the lesioned group). In the locomotion state (Figure 5E–5G), significant differences were detected by one-way ANOVA in Type A neurones in the theta (F (2,68) = 10.415, p = 0.0009) and alpha bands (F (2,68) = 5.934, p = 0.0041) and Type B neurones in the alpha band (F (2,47) = 6.525, p = 0.0037). 6-OHDA increased the SFC of Type A neurones in the theta (p = 0.0007) and alpha bands (p = 0.0153) and SFC of Type B neurones in the alpha band (p = 0.0104). These changes were reversed upon L-DOPA administration (p = 0.0113, 0.0134, and 0.0081, respectively, compared to in lesioned rats).

Figure 5. Spike-field coherence (SFC) in the pedunculopontine nucleus (PPN) and local field potential (LFP). (A) Frequency-SFC plots in resting Type A neurones. (B) Frequency-SFC plots in resting Type B neurones. (C) Statistical results of SFC obtained with resting Type A neurones. (D) Statistical results of SFC obtained with resting Type B neurones. (E) Frequency-SFC plots in Type A neurones of rats in locomotion state. (F) Frequency-SFC plots in Type B neurones of rats in locomotion state. (G) Statistical results of SFC obtained with Type A neurones of rats in locomotion state. (H) Statistical results of SFC obtained with Type B neurones of rats in locomotion state. The frequency bands with significant changes are highlighted via blue boxes. ***p < 0.001, *p < 0.05, relative to the control group; ###p < 0.001, ##p < 0.01, #p < 0.05, relative to the lesioned group.

Phase-lock

One-way ANOVA of the phase-lock data (Figure 6) revealed significant differences among resting Type B neurones of rat groups in the alpha band (F (2,54) = 7.537, p = 0.0012). Particularly, neurones in lesioned rats (n = 20, p = 0.0021) showed a higher length of mean phase angle (indicating a higher phase relationship between spikes and LFP) compared to those of control rats (n = 21). This difference was partially abolished upon L-DOPA treatment (n = 16, p = 0.0109). Figure 6 shows representative histograms illustrating the distribution of phase angles in the LFP frequency bands with significant alterations. The circles around the Rose histograms represent the phase orientation of each discharge.

Figure 6. Phase-lock between spikes and local field potential (LFP) in the pedunculopontine nucleus (PPN). (A) Statistical results of phase-lock obtained with resting Type A neurones. (B) Statistical results of phase-lock obtained with resting Type B neurones. (C) Rose histograms representing the distribution of spike phase angles with oscillation in noticeably altered bands. The circles around the histograms display the phase orientation for each discharge, and the red lines arising from the centre indicate the length and vector of the mean phase angles. (D) Statistical results of phase-lock obtained with Type A neurones of rats in locomotion state. (E) Statistical results of phase-lock obtained with Type B neurones of rats in locomotion state. **p < 0.01 relative to control rats; #p < 0.05 relative to lesioned rats.

Subjects

All experiments were conducted using male adult Wistar rats (280–320 g, purchased from Vital River Laboratories, Beijing, China), which were housed at 21 ± 1° C and 55% relative humidity under a 12:12-h light-dark cycle and provided with water and food ad libitum for at least 7 days before use. All experiments were approved by the ethics review board of Shandong University of Traditional Chinese Medicine (No. AE-20180905-01). All efforts were made to minimise the pain of animals as per the guidelines of the National Institute of Health for the Use and Care of Laboratory Animals and Animal Experiment Management Regulations approved by Shandong University of Traditional Chinese Medicine. The animals were separated into a control group (rats with saline injection in the medial forebrain bundle, MFB), lesioned group (rats with 6-OHDA injection in the MFB), and L-DOPA group (rats with 6-OHDA injection in the MFB and L-DOPA treatment).

Walking training

A ladder (100 × 20 cm) was employed for walking training as previously described [17]. The rats were fasted for 24 h and kept at one end of the ladder. A food reward was used to train rats to walk to the opposite end automatically. The rats were trained three times per day for 3 days. After training, the rats exhibited stable and consistent responses and could walk without pausing during electrophysiological signal recording. On the day of signal recording (29th day of the experimental timeline), the walking time of each rat was calculated and used as an indicator of motor function.

6-OHDA lesion and apomorphine-induced rotation test

To establish a rat model of PD, rats in the lesioned and L-DOPA groups were subjected to unilateral 6-OHDA lesion surgery [37]. The rats were anaesthetised with 4% chloralhydrate (360 mg/kg, intraperitoneally [i.p.]) and mounted in a stereotaxic frame. A burr hole was drilled above the left MFB (2.2 mm posterior and 2.1 mm lateral to the bregma), and 3 μL of 6-OHDA (4 μg/μL dissolved in 0.9% sodium chloride and 0.02% ascorbic acid; Sigma, St. Louis, MO, USA) was vertically injected 8.53 mm ventral to the surface of the skull at a rate of 1 μL/min using a microsyringe. To favour diffusion, we retained the syringe at the injection site for approximately 10 min, and then gradually withdrew it. Control rats underwent sham operation with injection of 3 μL of solvent. Suturing was performed at the incision, and the rat was administered 0.5 mL of penicillin (160 mg/kg, i.p.) each day for 3 days to avoid infection. Fourteen days after surgery (on day 14 of the experimental timeline), rats in lesioned and L-DOPA groups were injected with 0.5 mg/kg apomorphine (APO; Sigma) intramuscularly to induce contralateral rotation. The PD model was considered as successful if the rat performed seven turns per min within 30 min after injection. Only rats with successful lesions were included in the lesioned and L-DOPA groups. On day 29 of the experimental timeline (after treatment with L-DOPA), the above-described test was repeated to evaluate the drug effect.

L-DOPA treatment

To avoid the side effects of drugs that are common in clinical or animal research [40], we chose a low-dose, long-term regimen. Rats in the L-DOPA group were treated daily with 4.5 mg/kg L-DOPA (i.p.; Sigma) combined with 4.5 mg/kg benserazide (i.p.; Sigma) for 14 days. The drugs were dissolved in 0.5 mL of 0.9% sodium chloride with 0.02% ascorbic acid and administered three times daily (1.5 mg/kg L-DOPA and 1.5 mg/kg benserazide each time), i.e., at 9:00, 15:00, and 21:00 h [42]. Rats in the control and lesioned groups were mock-injected with solvent. No L-DOPA-induced dyskinesia was observed throughout the experiment.

Electrode implantation

During days 20–22 of the experimental timeline, a multi-unit recording electrode was implanted in all rats. The animals were anaesthetised with chloralhydrate (360 mg/kg, i.p.) and mounted in a stereotaxic frame. A home-made 16-channel electrode (consisting of 16 nickel-chromium Teflon-insulated microwires) was mounted on a stereotaxic frame into the rat PPN with the following coordinates: 1.9 mm lateral and 7.8 mm posterior to bregma, and 7.3 mm ventral to the skull surface, ipsilaterally to the 6-OHDA lesion surgery. The diameter of each microwire (purchased from California Fine Wire, Grover Beach, CA, USA) was 25 μm, and the tip impedance was 0.3–0.7MΩ at 1004 Hz. The Hong Kong Plexon Company provided guidelines for fabricating the electrodes. The Teflon was removed from one end of the microwire and welded to the socket. The other end was converged in a bundle of electrodes (with a cross-sectional area of 0.15 × 0.15 mm² and an array of 4 × 4) to be placed in the PPN. Considering the shadowing effects of single-unit recordings in adjacent electrodes, if the recordings collected from two adjacent electrodes were consistent in all parameters, one of the channels was removed from the analysis. The brain tissue was covered with melted agarose, and the ground wire of the electrode was bound tightly around the stainless-steel screws placed on the skull. The signal displayed in the process of electrode implantation provided stereotaxic assistant and guidance for the electrode position. Finally, the dental cement was used to fix the electrode to the skull, and the rat was administered 0.5 mL of penicillin (160 mg/kg, i.p.) every day for 3 days to prevent infection.

Rotarod test

To assess the beneficial effects of 6-OHDA lesion and L-DOPA on the locomotor function of rats, we performed a rotarod test using a rotarod treadmill (rod diameter: 64 mm, UGO Basile, Gemonio Varese, Italy) three times on day 29 of the experimental timeline at a constant speed (20 rounds per min) for 30 min. The longest latency and distance obtained with each animal were used for statistical analysis.

Signal recording

The electrophysiological signals were recorded on day 29 of the experimental timeline. Extracellular signals, i.e., spikes (sample rate 20 kHz, band-pass filtering 300–8000 Hz) and LFP (sample rate 1 kHz, band-pass filtering 0.5–200 Hz) were obtained using the 16-channel OmniPlex D Neural Data Acquisition System (Plexon, Dallas, TX, USA) in rats in resting (awake but not paying attention to their environment) and locomotion (ladder-walking) state. Videos were used to monitor animal conditions and recorded synchronously. The basal noise of the wideband (0.5–8000 Hz) signals indicated animal state differences compared to under resting conditions. Spike signals were recorded only when the signal-to-noise ratio was >2. During locomotion, the signals recorded throughout the walking task were extracted for analysis. Signal analysis was performed using Offline Sorter V4 (Plexon), Neuroexplorer V5 (Nex Technologies, Nays, KS, USA), and MATLAB (Mathworks, Natick, MA, USA).

Histological staining

On day 30 of the experimental timeline, the rats were sacrificed with 4% chloralhydrate (i.p.). An electrical microlesion (10 μA, 10 s × 5) was induced via an anodal current through the electrodes. The brains were collected after intracardial perfusion with 0.01 M phosphate-buffered saline (PBS) and paraformaldehyde (4%) containing potassium ferricyanide (1%). After post-fixation in 4% paraformaldehyde, the brains were dehydrated in 30% sucrose. Frozen sections were obtained, and 40-μm coronal sections encompassing the PPN were collected. The sections were stained with cresyl violet to examine the tracks and sites of electrodes, and only animals with correct electrode positioning were used for data analysis.

To determine the level of dopaminergic neurone degeneration, we collected sections encompassing the SNc of animals that underwent 6-OHDA injection surgery, and half of of the sections were stained with cresyl violet. The remaining sections were examined via histochemical tyrosine hydroxylase staining. The sections were washed with 0.01 M PBS, incubated for 6 h with blocking buffer containing Triton X-100 (0.5%) and normal donkey serum (5%), and then incubated with the primary polyclonal rabbit antibody (1:1000; ab6211; Abcam, Cambridge, UK) at 4° C overnight. The sections were incubated with fluorescein isothiocyanate-conjugated donkey anti-rabbit IgG (1:200; sc-2090; Santa Cruz Biotechnology, Dallas, TX, USA) for 2 h, washed with PBS, mounted with antifading mounting medium (S2100; Solarbio, Beijing, China), and the coverslips were fixed with nail polish. The primary and secondary antibodies were diluted with 0.01 M PBS containing Triton X-100 (0.3%) and donkey serum (3%). For the negative control, the same procedure was followed but the primary antibody was omitted. The sections stained with cresyl violet were observed, and the images were captured with an optical microscope (Axio Lab A1; Zeiss, Oberkochen, Germany) and tyrosine hydroxylase-stained sections were captured with a laser confocal microscope (Leica TCS SPE; Wetzlar, Germany).

Data analysis

Neurone classification

The recorded signals were imported into Offline Sorter Software, and cross-channel artefacts and waveforms with short ISIs (less than 2 ms) were eliminated with the built-in function. Using the k-means clustering algorithm in PCA, the waveforms in each channel were sorted. Waveforms with analogous characteristics were gathered as a cluster in the 2D/3D view. Two separate clusters were defined as different neurones after application of the F statistic parametric multivariate ANOVA (analysis of variance, p < 0.05). Manual checking was conducted to guarantee that spurious signals were eliminated. The characteristics of each neurone in the PPN were examined, and the spikes of each neurone were identified based on the waveform shape, duration of peak-to-trough, inter-spike interval histograms, PCA analysis, and firing rate [17].

Six signal segments per neurone in resting or locomotion state were randomly selected according to the random digits method. The neurone mean level was defined as the mean value of these six segments. The length of each segment was 10 s in resting state and corresponded to the time of walking along the ladder in the locomotion state.

Coherence analysis

The selected signals were imported into MATLAB, and analysis was conducted using the open-source data analysis toolbox Chronux 2.0 (http://chronux.org) and CircStat Toolbox [43]. To investigate the effect of the 6-OHDA lesion and L-DOPA treatment on the spike-LFP synchronisation of PPN, we applied three algorithms for coherence, SFC, and phase-lock analyses according to our previous studies [16, 37]. The LFP signals were separated into five bands: delta, 1–4 Hz; theta, 4–7 Hz; alpha, 7–12 Hz; beta, 12–30 Hz; gamma, 30–70 Hz. The spike-LFP synchronisation of each neuronal type in the PPN was analysed in different LFP frequency bands. In Figure 7, the LFP filtering process is shown as a representative of the 10-s signal in the resting state, and the synchronously recorded spike raster indicates single-unit activities. In addition, an LFP spectrogram obtained with the fast Fourier transfer-based log of power spectral density method with multiple tapers (3 time-bandwidth and five tapers) is shown.

Figure 7. Local field potential (LFP) filtering process is illustrated as a representative 10-s signal in resting state with synchronously recorded spike raster and an LFP spectrogram.

With the coherence function in Chronux, a multitaper method with 0.5-s window size, 0.1-s time interval, nine tapers, five time-bandwidth and 1 kHz frequency sample were used in coherence analysis. Variations in the values of coherence between zero to one were calculated, and zero indicated that the spike and LFP had no relation, whereas one indicated a perfect correlation.

SFC analysis

In this study, SFC analysis was performed to evaluate the timing relationship between oscillations and spikes. The SFC was calculated as the ratio between the frequency spectrum of spike-triggered average (STA) and the average spike-triggered power (STP) frequency spectrum of the LFP traces in specific frequency bands: SFC = (fSTA/STP) × 100% [18]. The STA was obtained by extracting the unfiltered LFP signals (centred on each spike ± 500 ms), and the STA spectrum was calculated (3 time-bandwidth, five tapers, window overlap of 50%). For each signal segment, the number of spikes was equalised to 50 and 20 in resting state and locomotion state, respectively.

Analysis of phase-lock

The Rayleigh test was performed to assess whether the spikes were noticeably phase-locked with LFP at specific frequency bands. Only segments with p < 0.05 were used for further analysis. As the spikes and LFP were recorded synchronously, spikes occurring at LFP oscillation peaks were considered as 0° C phase, whereas spikes at troughs were considered as 180° C (π) or -180° C (-π) phase. Subsequently, using the CircStat Toolbox, the mean phase angle and resulting vector length were calculated. The mean phase angle length varied from 0 to 1. When the mean phase angle length approached 1, the neurone showed a greater correlation with the LFP oscillations because the angles were more concentrated around the mean phase angle. The mean phase angle vector and length were plotted in a Rose histogram.

Statistics

SPSS software 23.0 (SPSS, Inc., Chicago, IL, USA) was used for statistical analysis, and the data were represented as the mean ± standard error. The Kolmogorov-Smirnov test was used to evaluate the normality of data distribution. The differences between rat groups (control, lesioned, and L-DOPA) were assessed by one-way ANOVA followed by Tukey’s post hoc comparison. For the APO-induced rotation, a two-tailed unpaired t-test was performed between two groups. The threshold for statistical significance was set at p < 0.05, and the level of significance was defined as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.