Dopamine (DA) neurons in the substantia nigra (SN) control several essential functions, including the voluntary movement, learning and motivated behavior. Healthy DA SN neurons show diverse firing patterns in vivo, ranging from slow pacemaker-like activity (1-10 Hz) to transient high frequency bursts (<100 Hz), interspersed with pauses that can last hundreds of milliseconds. Recent in vivo patch experiments have started to reveal the subthreshold mechanisms underlying this physiological diversity, but the impact of challenges like cell loss on the in vivo activity of adult DA SN neurons, and how these may relate to behavioral disturbances, are still largely unknown. We investigated the in vivoelectrophysiological properties of surviving SN DA neurons after partial unilateral 6-OHDA lesions, a single-hit, non-progressive model of neuronal cell loss. We show that mice subjected to this model have an initial motor impairment, measured by asymmetrical rotations in the open field test, which recovered over time. At 3 weeks post-lesion, when open field locomotion was strongly impaired, surviving DA SN neurons showed a compressed in vivo dynamic firing range, characterized by a 10-fold reduction of in vivo burst firing compared to controls. This in vivo phenotype was accompanied by pronounced in vitro pacemaker instability. In contrast, in the chronic post-lesion phase (>2 months), where turning symmetry in open field locomotion had recovered, surviving SN DA neurons displayed the full dynamic range of in vivo firing, including in vivo bursting, similar to controls. The normalized in vivo firing pattern was associated with a 2-fold acceleration of stable in vitro pacemaking, mediated by Kv4.3 potassium channel downregulation. Our findings demonstrate the existence of a homeostatic pacemaker plasticity mechanism in surviving DA SN neurons after pronounced cell loss.

Animals

Male and female C57Bl/6N mice (Charles River Laboratories) were used for the study. The mice were 8 weeks old, group housed and maintained on a 12-hour light-dark cycle. All experiments and procedures involving mice were approved by the German Regierungspräsidium Darmstadt (V54-19c20/15-F40/30). In total, 152 mice were used for this study (see table below).

Stereotactic 6-OHDA infusion

All surgeries were performed under general anesthesia in areflexic state. Prior to the induction of anesthesia, a premedication of 0.2 mg/kg atropine (atropine-sulfate, Braun Melsungen AG, Melsungen) was given as an intraperitoneal (i.p.) injection to stabilize circulation. Anesthesia was induced in a plastic chamber, which was flooded with 5% Isoflurane (Florene®, AbbVie Deutschland GmbH & Co. KG, Ludwigshafen, Germany) in pure oxygen (0.4 l/min). For maintenance of anesthesia, isoflurane was delivered through a breathing mask with a flow rate of 0.35 l/min and its concentration was regulated to 1.5-2.2% using an adjustable vaporizer (Uno, Zevenaar, Netherlands). The depth of anesthesia was controlled by testing the toe pinch reflex and the breathing rate (1-2Hz). Body temperature (36°C) and respiration were constantly monitored. Lidocaine/prilocaine ointment (25 mg/g, Emla® creme, AstraZeneca GmbH, 22876 Wedel) was applied prior to surgery and after suturing of the wound as local anesthetics. Additional analgesia was provided by subcutaneous injection of carprofen (4 mg/kg in NaCl, Rimadyl®, Pfilzer GmbH, Berlin, Germany) after infusion. Eye lubricant (Visidic, Bausch and Lomb, Berlin, Germany) was used to protect eyes from desiccation.

Desipramine hydrochloride (20 mg/kg, Sigma Aldrich) was injected i.p. 20-40 min before intracranial infusions to prevent 6-OHDA uptake by noradrenergic neurons. The desipramine solution was prepared in sterile, isotonic NaCl solution (B. Braun Melsungen AG, Germany) at the day of surgery. The infusion solutions are based on sterile artificial cerebrospinal fluid (ACSF, Harvard Apparatus, Holliston, MA, USA) with 0.02% L-ascorbic acid (used also as a vehicle solution). The 6-OHDA solution (0.2% 6-hydroxydopamine hydrochloride in ACSF with 0.02% L-ascorbic acid) was prepared at the day of infusion, stored on ice, and shielded from light.

During surgery, the animals were placed on a heating pad and were fixed in a stereotactic frame (Model 1900, Kopf Instruments, Tujunga, USA) with a stereotactic arm and a connected three-way digital positioning display. The scalp was opened by a longitudinal cut to expose the skull with bregma and lambda on display. With a centering scope (Model 1915, Kopf Instruments, Tujunga, USA), the bregma-lambda distance was measured and examined for suitable anatomy (4.4 ± 0.2 mm distance). Afterwards, the skull was aligned to a reference frame with a stereotaxic alignment indicator (Model 1905, Kopf Instruments, Tujunga, USA) and the manipulator system was referenced to bregma.

Using a stereotaxic drill (Model 1911, Kopf Instruments, Tujunga, USA) with a 500 µm diameter drill bit, a hole above the right striatum was drilled (coordinates: ML: +1.9 mm, AP: +0.5 mm to bregma). ACSF or 6-OHDA solution were loaded to a 10 µl NanoFil syringe (World Precision Instruments Inc., Sarasota, FL, USA) with a 35G blunt needle, which was mounted on a MicroSyringe Pump (UMP3-1, World Precision Instruments) and controlled by a SYSMicro4 Controller (World Precision Instruments). Using the stereotactic arm, the needle was slowly lowered (about 750 µm/min) to a position of -2.2 mm below the brain surface (infusion site coordinates: ML: +1.9 mm, AP: +0.5 mm, DV: -2.2 mm to bregma). Anatomical references are based on Franklin and Paxinos (2008). A volume of 6 µl was infused with a flow rate of 250 nl/min. Once the volume was infused, the needle rested for 5 minutes in that position before it was slowly moved out of the brain. Directly before and after infusion, proper functioning of the syringe system and the needle was checked. Finally, after suture the animal was placed on a heating pad for full recovery. Oats, wet food pallets and water were placed inside the cage to ease consumption.

Behavioral testing

Open field

Spontaneous locomotion (track length, wall distance, time in center and number of rearings) and rotations of all mice were monitored in open field (50 × 50 cm, center 30 × 30 cm; red illumination, 3 lx) for 10 min in 3 baseline sessions and every 4^th or 7^th day post-infusion of ACSF/6-OHDA till the day of in vivo or in vitro experiment (e.g. 21^st or >68^thpost-operative day). The open field was cleaned before and after each mouse with 0,1% acetic acid in distilled water. Using a video tracking system (Viewer II/III, Biobserve) spontaneous behavior was recorded and analyzed both online and offline. Data was extracted from Viewer as Excel-tables and the final analysis was made with custom made Matlab-scripts.

Cylinder test

Forelimb use during explorative activity was explored with cylinder test. The test was performed at corresponding termination time point (20-21^st and 64^th post-infusion day). Mice were placed individually in a glass beaker (9 cm diameter, 19 cm height) at room light and were video recorded with a camera (Logitech HD Webcam C615) for about 5 min. No habituation was allowed before video recording. The glass cylinder was cleaned before and after every mouse with 0,1% acetic acid in distilled water. Only weight-bearing wall contacts made by each and both forelimb on the cylinder wall were scored. Wall exploration was expressed in terms of the percentage of contralateral to the infusion side (in the 6-OHDA-infused mice also impaired forepaw) to all forelimb wall contacts.

In vivo electrophysiology

Extracellular recording

In vivo extracellular single-unit activities of SN and VTA neurons were recorded in ACSF-infused (vehicle) and 6-OHDA-infused mice, similar procedures were used in other studies from our lab (Farassat et al. 2019; Schiemann et al. 2012; Subramaniam, Althof, et al. 2014). Briefly, mice were anesthetized (isoflurane; induction 4.5-5%, maintenance 1–2% in 0.4 l/min O2) and placed into a stereotactic frame. The craniotomies were performed as described above to target the lateral SN (AP: -3.08 mm, ML: 1.4 mm) and medial SN (AP: -3.08 mm, ML: 0.9 mm). Borosilicate glass electrodes (10–25 MΩ; Harvard Apparatus, Holliston, MA, USA) were made using a horizontal puller (DMZ-Universal Puller, Zeitz, Germany) and filled with 0.5 M NaCl, 10 mM HEPES (pH 7.4) and 1.5% neurobiotin (Vector Laboratories, Burlingame, CA, USA). A micromanipulator (SM-6; Luigs & Neumann, Ratingen, Germany) was used to lower the electrodes to the recording site. The single-unit activity of each neuron was recorded for at least 10 minutes at a sampling rate of 12.5 kHz (for firing pattern analyses), and then for another one minute at a sampling rate of 20 kHz (for the fine analysis of AP waveforms). Signals were amplified 1000x (ELC-03M; NPI electronics, Tamm, Germany), notch- and bandpass-filtered 0.3–5000 Hz (single-pole, 6 dB/octave, DPA-2FS, NPI electronics) and recorded on a computer with an EPC-10 A/D converter (Heka, Lambrecht, Germany). Simultaneously, the signals were displayed on an analog oscilloscope and an audio monitor (HAMEG Instruments CombiScope HM1508; AUDIS-03/12M NPI electronics). Midbrain DA neurons were initially identified by their broad biphasic AP (> 1.2 ms duration) and slow frequency (1–8 Hz) (Anthony A. Grace and Bunney 1984; Ungless and Grace 2012). AP duration was determined as the interval between the start of initial upward component and the minimum of following downward component.

Juxtacellular labeling of single neurons

In order to identify the anatomical location and neurochemical identity of the recorded neurons, they were labeled post-recording with neurobiotin using the juxtacellular in vivo labeling technique (Pinault 1996). Microiontophoretic currents were applied (1–10 nA positive current, 200ms on/off pulse, ELC-03M, NPI Electronics) via the recording electrode in parallel to the monitoring of single-unit activity. Labeling was considered successful, when: the firing pattern of the neuron was modulated during current injection (i.e., increased activity during on-pulse and absence of activity in the off-pulse), the process was stable for at least 20s, and was followed by the recovery of spontaneous activity. This procedure allowed for the exact mapping of the recorded DA neuron within the SN and VTA subnuclei (Franklin and Paxinos 2012) using custom written scripts in Matlab (MathWorks, Natick, MA, USA), combined with neurochemical identification using TH-immunostaining.

In vitro electrophysiology

Slice preparation

Animals were anesthetized by intraperitoneal injection of ketamine (250 mg/kg, Ketaset, Zoetis) and medetomidine-hydrochloride (2.5 mg/kg, Domitor, OrionPharma) prior to intracardial perfusion using ice-cold ACSF consisting of the following (in mM): 50 sucrose, 125 NaCl, 2.5 KCl, 25 NaHCO3, 1.25 NaH2PO4, 2.5 glucose, 6 MgCl2, 0.1 CaCl2 and 2.96 kynurenic acid (Sigma-Aldrich), oxygenated with 95% O2 and 5% CO2. Rostral coronal midbrain slices (bregma: -2.92 mm to -3.16 mm) were sectioned at 250 µm using a vibrating blade microtome (VT1200s, Leica). Slices were incubated for 1 h before recordings in a 37°C bath with oxygenated extracellular solution with extra 1µM AmmTx3, containing the following (in mM): 22.5 sucrose, 125 NaCl, 3.5 KCl, 25 NaHCO3, 1.25 NaH2PO4, 2.5 glucose, 1.2 MgCl2 and 1.2 CaCl2.

In vitro patch-clamp recordings

Slices were placed in a heated recording chamber (37°C) that was perfused with oxygenated extracellular solution at 2-4 ml min⁻¹. CNQX (20 μM; Biotrend), gabazine (SR95531, 4 μM; Biotrend) DL-AP5 (10 μM; Cayman Chemical) were added to inhibit excitatory and inhibitory synaptic transmission. For voltage clamp recordings, TTX (500nM; Tocris) was added to the extracellular solution. Neurons were visualized using infrared differential interference contrast videomicroscopy with a digital camera (VX55, Till Photonics) connected to an upright microscope (Axioskop 2, FSplus, Zeiss). Patch pipettes were pulled from borosilicate glass (GC150TF-10; Harvard Apparatus LTD) using a temperature-controlled, horizontal pipette puller (DMZ-Universal Puller, Zeitz). Patch pipettes (4-6 MΩ) were filled with a solution containing the following (in mM): 135 KGlu,  5 KCl, 10 HEPES, 0.1 EGTA, 5 MgCl2, 0.075 CaCl2, 5 NaATP, 1 LiGTP, 0.1% neurobiotin, adjusted to a pH of 7.35 with KOH. Recordings were performed using an EPC-10 patch-clamp amplifier (Heka electronics) with a sampling rate of 20 kHz and a low-pass filter (Bessel, 5 kHz). For voltage clamp recordings only experiments with uncompensated series resistance <10MO were included in this study and series resistance was electronically compensated 75%. Neurons were held at a holding potential of -40mV to minimize HCN activation. To determine Kv4.3 activation kinetics neurons were hyperpolarized to -80mV for 500ms followed by varying voltage steps from -60mV to -20mV in increments of 5mV for 1s. For inactivation kinetics neurons were hyperpolarized from -120mV to -20mV in increments of 10 mV for 1s followed by a fixed voltage step to -20mV for 1s. For analysis of HCN currents neurons were hyperpolarized from -80mV to -120mV in increments of 20mV for 1s. For analysis, action potential thresholds (mV) were determined at dV_m/dt > 10 mV/ms.

Immunohistochemistry

Following in vivo recordings, animals were transcardially perfused, as described previously (Farassat et al. 2019; Schiemann et al. 2012; Subramaniam, Althof, et al. 2014). Fixed brains were sectioned into 60µm (midbrain) or 100µm (forebrain) coronal sections using a vibrating microtome (VT1000S, Leica). In vitro slices were fixed in paraformaldehyde after finishing the experiment. Sections were rinsed in PBS and then incubated (in blocking solution (0.2 M PBS with 10% horse serum, 0.5% Triton X-100, 0.2% BSA). For staining with the polyclonal rabbit alpha-synuclein phospho S129 antibody, an antigen retrieval protocol was applied prior to blocking: sections were incubated for 30 minutes at 80 °C in sodium citrate buffer. Afterwards, the standard histological protocol continued with blocking.

Sections were incubated in carrier solution (room temperature, overnight) with the following primary antibodies: polyclonal guinea pig anti-tyrosine hydroxylase (TH; 1:1000; Synaptic Systems), monoclonal mouse anti-TH (1:1000; Synaptic Systems) or polyclonal rabbit anti-TH (1:1000; Synaptic Systems); mouse anti-Kv4.3 (1:1000, Alomone Labs); polyclonal rabbit alpha-synuclein phospho S129 (1:200, Abcam), monoclonal rabbit anti alpha-synuclein aggregate [MJFR-14-6-4-2] (1:1000, Abcam). In sequence, sections were again washed in PBS and incubated (room temperature, overnight) with the following secondary antibodies: goat anti-guinea pig 488, goat anti-rabbit 488, goat anti-mouse 488, goat anti-mouse 568 , goat anti-rabbit 568 (all 1:750; Thermofisher). Streptavidin AlexaFluor-568 and Streptavidin AlexaFluor-488 (both 1:1000; Invitrogen) were used for identifying neurobiotin-filled cells. Sections were then rinsed in PBS and mounted on slides with fluorescent mounting medium (Vectashield, Vector Laboratories).

DAB immunocytochemistry

For DAB (3,3ʼ-diaminobenzidine) staining procedures, a Vectastain ABC Staining Kit (Vector Laboratories) was used. Coronal sections of midbrain (30 μm) areas were cut and rinsed in PBS (3x10 min). Similar to previous immunolabeling procedures, unspecific antigen binding sites were blocked by incubation of the sections with blocking solution (60 min, room temperature). Subsequently, sections were incubated with primary antibody against TH (rabbit anti TH) overnight, rinsed in PBS (3x10 min), and were incubated with biotinylated secondary antibodies (biotinylated anti-rabbit) for two hours at RT. In parallel, an avidin-biotin complex (ABC) was formed by pre-incubation of avidin (1:1000) with biotinylated HRP (1:1000) in PBS for two hours at room temperature. Sections were rinsed in PBS (3x10 min) prior to incubation with ABC solution (60 min, room temperature). Next, sections were rinsed in PBS (2x10 min) and Tris-buffer (1x10 min). Finally, DAB oxidation was developed by application of 2 % H₂O₂, 2 % NiCl₂ and 4 % DAB in Tris-buffer using a DAB Substrate Kit (Vector Laboratories, Burlingame, USA). NiCl₂ enhances sensitivity and intensity of DAB precipitation product. DAB oxidation was developed for 2 to 5 minutes and was stopped with Tris-buffer once a specific high-contrast signal was detectable. Sections were rinsed in Tris-buffer (3x10 min) and transferred onto gelatin-covered slides, air-dried overnight, and dehydrated in consecutive ascending alcohol concentrations (50 %, 70 %, 90 % and 2x 100 %; 10 min each) followed by dehydration in xylol (2x 100 %; 10 min each). Finally, sections were mounted under glass coverslips with hardening mounting medium (Vectamount, Vector Laboratories, Burlingame, USA).

Unbiased stereology measurements

For quantification of total cell loss, TH-DAB labeled SN DA neurons were counted using unbiased stereology based on optical dissection (Gundersen 1986). In coronal sections (30 μm), the region of interest was selected based on anatomical landmarks including the medial lemniscus, which separates SN and adjacent VTA. Stereological counting provides unbiased data based on random, systematic sampling using an optical fractionator. This method involves counting of neurons with an optical dissector, a three-dimensional probe placed through a reference space (Gundersen 1986). The optical dissector forms a focal plane with a thin depth of field through the selected sections. Objects in focus of this focal plane are located within the reference section and are counted, while objects outside of the focal plane are not counted. On top of the optical dissector, a counting frame is applied. Counting frames ensure that all neurons have equal probabilities of being selected, regardless of shape, size, orientation, and distribution. To avoid counting capped neurons at the border of a section, an additional guard zone was deployed at the upper and lower borders of each section. DA neurons within the counting frame as well as those crossing the green line (acceptance line) were counted, while DA neurons crossing the red line (rejection line) excluded. Moreover, only neurons with a detectable nucleus in focus within the optical dissector were counted. For quantification of total cell loss, StereoInvestigator software (V5, MicroBrightField, Colchester, USA) was used in combination with BX61 microscope (Olympus, Hamburg, Germany). The region of interest was selected and marked using a low magnification objective lens (2x, NA 0.25, Olympus) and 12-30 serial sections of 30 μm thickness were counted, covering the entire caudo-rostral extent of the SN. To count the number of DA neurons in the area of the SN pars compacta, a high magnification oil-immersion objective lens (100x, NA 1.30, Olympus) (counting frame, 50 × 50 μm; sampling grid, 125 × 100 μm) was used. After counting was finished, the total number of neurons was calculated by the software using the formula described by West et al (West 1993).

Optical density measurements

Optical density measurements of TH-DAB labeled striatal sections were performed using ImageJ software (http://rsbweb.nih.gov/ij/). Following TH-DAB labeling and TH-immunohistochemistry, images of five coronal sections (100 μm) covering the rostrocaudal axis of the striatum were captured using laser-scanning microscope (Nikon Eclipse90i, Nikon GmbH). Images were gray scale converted and mean gray values of desired striatal areas were encircled and measured. Unspecific mean gray values were measured in a defined cortical area (100x100 pixels) that displayed no specific TH signal due to the absence of DA innervation and were subtracted. The ventral edge of lateral ventricles served as an anatomical landmark to separate dorsal and ventral areas. For all animals, the measurement from the ipsilateral to the infusion side were divided by the contralateral side to calculate the relative optical density of the striatum.

Immunohistochemical Kv4.3 channel signal quantification

A Nikon Eclipse 90i microscope was used for fluorescent signal detection, excitation wavelength of 488 nm for TH-signal and 568 nm for Kv4.3-channel signal. From each animal, 4 midbrain slices covering the caudal, intermedial and rostral regions, were selected and imaged for overview with 4x magnification. Then 60x magnification was used to acquire data from 4 areas within each SN (4 images on the ipsilateral and 4 images on the contralateral to infusion side). All images were acquired using the same laser and camera settings. Images were exported from Nikon NIS-Elements Advanced Research (Version 4.20.03) software as 8-bit TIFF files for quantification. Data was analyzed using custom made Python 3.0 scripts with matplotlib, numpy, scimage and scipy modules. First, TH immunosignals were converted to a binary image via Otsu-thresholding algorithm to detect TH areas bigger than 400 pixels. Then, the resulting binary image was used as a mask for Kv4.3 channel immunosignal detection. For all ROIs surface areas and mean Kv.4.3 channel signal intensity was measured. By applying erosion and dilatation algorithms on the ROIs, membrane and cytoplasm areas were segregated, allowing isolation of Kv4.3 channel immunosignal intensity for these cell compartments. Background Kv4.3 channel immunosignal was quantified in TH-immunosignal areas below the Otsu-threshold. All data were then grouped according to medio-lateral and ipsi-/contralateral position for both vehicle and 6-OHDA group. Graphs and statistical analysis for this data was performed using Python custom made scripts.

Statistical Analysis

Spike train analyses

Spike time-stamps were extracted by thresholding above noise levels with IgorPro 6.02 (WaveMetrics, Lake Oswego, OR, USA). Firing pattern properties such as mean frequency, coefficient of variation (CV) and bursting measures were analyzed using custom scripts in Matlab. In order to estimate burstiness and intra-burst properties, we used the burst detection methods described in Grace & Bunney (A. A. Grace and Bunney 1984; Anthony A. Grace and Bunney 1984; Ungless and Grace 2012). All non-burst related ISIs (excluding all ISIs that followed the Grace and Bunney criteria, as well as all pre- and post-burst ISIs) were used to calculate the single spike firing frequency and single spike coefficient of variation.

For analysis of general firing patterns, autocorrelation histograms (ACH) were plotted using custom Matlab scripts. We used established criteria for classification of in vivo firing patterns based on visual inspection of autocorrelograms (Farassat et al. 2019; Schiemann et al. 2012; Subramaniam, Althof, et al. 2014): single spike-oscillatory (≥3 equidistant peaks with decreasing amplitudes), single spike-irregular (<3 peaks, increasing from zero approximating a steady state), bursty-irregular (narrow peak with steep increase at short ISIs) and bursty-oscillatory (narrow peak reflecting fast intraburst ISIs followed by clear trough and repetitive broader peaks).

Statistics

Categorical data is represented as stacked bar graphs. To investigate the assumption of normal distribution, we performed the single-sample Kolmogorov-Smirnov test. The Mann-Whitney-Test, one-/two-way ANOVA were performed in non-parametric data to determine statistical significance. Categorical parameters, such as ACH-based firing pattern, were analyzed with the Chi-squared test. Statistical significance level was set to p < 0.05. All data values are presented as means ± SEM. Statistical tests were performed using GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA), Matlab and Python. The scatter plots are represented with median or mean ± SEM. The resulting pvalues were compared with Bonferroni-corrected α-level or Tukey post hoc comparison. A value of p ≤ 0.05 was considered to be statistically significant; p ≤ 0.05 = *p ≤ 0.005 = **p ≤ 0.0005 = ***. Graphs were plotted using GraphPad Prism software (9.0c), Matlab and Python.