Cav2.3 channels contribute to dopaminergic neuron loss in a model of Parkinson’s disease

Degeneration of dopaminergic neurons in the substantia nigra causes the motor symptoms of Parkinson’s disease. The mechanisms underlying this age-dependent and region-selective neurodegeneration remain unclear. Here we identify Cav2.3 channels as regulators of nigral neuronal viability. Cav2.3 transcripts were more abundant than other voltage-gated Ca2+ channels in mouse nigral neurons and upregulated during aging. Plasmalemmal Cav2.3 protein was higher than in dopaminergic neurons of the ventral tegmental area, which do not degenerate in Parkinson’s disease. Cav2.3 knockout reduced activity-associated nigral somatic Ca2+ signals and Ca2+-dependent after-hyperpolarizations, and afforded full protection from degeneration in vivo in a neurotoxin Parkinson’s mouse model. Cav2.3 deficiency upregulated transcripts for NCS-1, a Ca2+-binding protein implicated in neuroprotection. Conversely, NCS-1 knockout exacerbated nigral neurodegeneration and downregulated Cav2.3. Moreover, NCS-1 levels were reduced in a human iPSC-model of familial Parkinson’s. Thus, Cav2.3 and NCS-1 may constitute potential therapeutic targets for combatting Ca2+-dependent neurodegeneration in Parkinson’s disease.

SEM. Significances are indicated by asterisks: * p < 0.05 ** p < 0.01, *** p < 0.001, **** p < 0.0001. All data and statistics are detailed in Supplementary Table 8A. knockout: n=346) expressed as ratio between raw signal and background on Cav2.3 wildtype and Cav2.3 knockout for both regions. Wildtype and knockout sections were processed in parallel, and SN and VTA neurons analysis was performed in parallel on the same sections (Note that signals in TH-positive neurons from wildtype are significantly higher than in knockout, indicating the specificity of the analyzed signal in Fig. 1c). Error bars: SD. Significances are indicated by asterisks: * p < 0.05 ** p < 0.01, *** p < 0.001, **** p < 0.0001. All data and statistics are detailed in Supplementary Table 5B. lower: action potentials upon injection of +40 pA current from a holding potential of -70 mV).

Supplementary
General workflow of UV-LMD & RT-qPCR analysis. SN dopaminergic neurons of adult mice were retrogradely traced with fluorescent retrobeads in vivo. Midbrains were prepared and cut into 12 µm coronal cryosections, fixed with ethanol and -for juvenile mice -stained with cresyl-violet (CV). 10 individual SN neurons were isolated using UV-LMD. After combined cell lysis and cDNA synthesis, cDNA samples were purified via ethanol precipitation 1 and dissolved in 17 µl water (5Prime). To verify the homogeneity of each sample, 5 µl cDNA (equivalent to ~3 cells) was used for a multiplex nested PCR of a panel of positive and negative marker genes [2][3][4] . Only samples expressing the correct marker gene profile (positive for TH and negative for GAD65/67, CB and GFAP) were further analyzed via TaqMan qPCR. For each TaqMan qPCR, 5 µl of a cDNA sample was used as template (equivalent to ~3 cells). Relative or absolute expression levels were determined by running respective standard curve DNA samples in parallel (see next paragraph, cited methods papers, and cartoon (B) for further details). Workflow modified from 5 , graphics created by Julia Benkert.

Supplementary Figure 9.
General workflow of UV-LMD Workflow of standard curve generation for relative and absolute quantification of cDNA via qPCR. For generation of standard curves, total RNA was isolated from murine midbrain tissue using the RNeasy Mini Kit and reverse-transcribed into cDNA and purified, according to 5,6 . For generation of relative standard curves, cDNA concentration was determined using a QuBit 3.0. cDNA samples for generation of relative standard curves were generated by serial dilutions of this cDNA, containing 12.5 to 12,500 pg cDNA as templates. Respective dilutionranges used for relative qPCR quantifications were determined for each analyzed gene while establishing the respective TaqMan assays. These samples for generation of relative standard curves were run in parallel with UV-LMD samples.
For generation of standard curves allowing absolute quantification of cDNA molecule numbers, a PCR fragment for each individual gene was amplified from mouse midbrain tissue derived cDNA, covering the respective location of the TaqMan assay. These PCR products were separated and purified by gel electrophoresis, band-excision and DNA-extraction, using the QIAquick Gel

Animals and chronic MPTP treatment
Juvenile (~PN13), adult (~PN90) and aged (~PN550) male C57BL/6J, NCS-1 knockout and NCS-1 wildtype mice 8 were bred at Ulm University. Cav2.3 knockout and Cav2.3 wildtype mice 9 (obtained from Toni Schneider) were bred at Ulm University as well as at the University of Cologne. The NCS-1 knockout (obtained from Olaf Pongs) is back-crossed at least 10 times into C57BL/6J 8 , leading to a 99.9% analogy with C57BL/6J 10 and losing the 129/SvJ original background. Cav2.3 knockout is back-crossed only 4 times into C57BL/6J as they do not breed well, likely due to the presence of Cav2.3 channels in sperm 11 , leading to a 75% C57BL/6J and 25% 129/SvJ mixed background. Data derived from littermate NCS-1 wildtype and Cav2.3 wildtype were not significantly different from those of C57BL/6J, wildtype data were pooled for some experiments. To obtain sufficiently large cohorts with animals of the same age for the MPTP experiments, animals from litters bred in parallel under identical housing conditions were pooled. Two independent breeding cohorts each were analyzed with this strategy. The MPTP induced degeneration of about 50% of SN dopaminergic neurons in the wildtype animals of this study is similar to that induced in other wildtype strains or in C57BL/6J mice [12][13][14][15] . To probe for differences in MPTP sensitivities due to different genetic backgrounds, we compared effects on wildtype strains from this study and other studies 12 . We robustly detect similar MPTP sensitivities in different wildtype strains and of different cohorts.
MPTP injections and perfusions were carried out at the University of Göttingen. Numbers of mice were calculated in accordance with biometric sample size estimation to detect a biologically relevant difference between compared groups. These calculations are a legal requirement to obtain approval from our federal ethical review committee (Regierungspräsidium Tübingen), to analyze a sufficiently large cohort of animals, but to prevent use of more animals than necessary (3R principle directive 16  ) and carried out in accordance with the approved guidelines. Mice were injected 10 times with MPTP hydrochloride (20 mg/kg saline, subcutaneously, Sigma) and probenecid (250 mg/kg PBS buffer, intraperitoneally, Thermo Fisher) for 5 weeks every 3.5 days 17 , according to the respective safety protocols and guidelines 18 . Control mice were injected with saline and probenecid only. One week after the last injections, mice were sacrificed and PFAperfused for immunohistochemistry 12 .
The chronic low-dose MPTP/probenecid model was chosen as it best simulates the progressive degeneration found in the human Parkinson's disease, and thus is generally accepted as the gold standard for preclinical testing of neuroprotective strategies for Parkinson's disease in rodents and monkeys [18][19][20][21][22] . MPTP induces selective loss of particularly SN dopaminergic neurons and Parkinson symptoms in animals and humans. Probenecid retards renal excretion and thus reduces biological variations 17 . Advantages of this chronic MPTP neurotoxin model, compared to the 6-OHDA neurotoxin Parkinson's disease-model are that MPTP is less invasive and better mimics the progression and pathology of Parkinson's disease 19,23 . 6-OHDA (in contrast to MPTP) cannot cross the blood-brain-barrier, which necessitates its direct injection into the SN or the striatum, and consequently there is a larger variability of its outcome in dependence e.g. on the exact injection sites and 6-OHDA dose. This makes it more difficult to standardize conditions between laboratories 3,24 . Furthermore, the 6-OHDA model is less chronic (thus less chance to activate neuroprotective responses due to the Parkinson's disease-stressor), as neuronal death occurs over a brief time course (similar as in the acute MPTP models), and Lewy-bodies (as present in Parkinson's disease and the chronic MPTP model), are normally not found [19][20][21][22][23]25 .

Human samples
Informed consent was obtained from all Parkinson's disease-patients and control-subjects involved in this study before cell donation. The Hampstead Research Ethical committee previously approved the consent forms. All available information on the human donors is summarized in Supplementary Table 9A.

Combined Ca 2+ Imaging and electrophysiology
Electrophysiological recordings were combined with Ca 2+ imaging 3,28 . SN dopaminergic neurons were identified according to their sag component / slow Ih-current, broad action potentials, and post hoc by TH-immunohistochemistry, mesolimbic VTA dopaminergic neurons (from C57BL/6J mice) were identified by retrograde tracing and/or electrophysiological fingerprints, and post hoc by TH-immunohistochemistry 3,29,30 . Biocytin-streptavidin labeling was combined with THimmuno-histochemistry 28 . Only electrotypical SN dopaminergic neurons and Nucleus Accumbens core / medial shell VTA dopaminergic neurons 29 were analyzed in this study.
Perforated patch recordings were performed using protocols modified from Horn & Marty 31 and Akaike & Harata 32 . Electrodes with tip resistances between 3 and 5 MOhm were fashioned from borosilicate glass (0.86 mm inner diameter; 1.5 mm outer diameter; GB150-8P; Science Products) with a vertical pipette puller (PP-830; Narishige, London, UK). Patch recordings were performed with ATP and GTP free pipette solution containing (in mM): 128 Kgluconate, 10 KCl, 10 HEPES, 0.1 EGTA, 2 MgCl2 and adjusted to pH 7.3 (with KOH) resulting in an osmolarity of ~300 mOsm. ATP and GTP were omitted from the intracellular solution to prevent uncontrolled permeabilization of the cell membrane 33 . The patch pipette was tip filled with internal solution and back filled with 0.02% tetraethylrhodamine-dextran (D3308, Invitrogen, Eugene, OR, USA) and amphotericincontaining internal solution (~200-250 μg·ml -1 ; G4888; Sigma-Aldrich, Taufkirchen, Germany) to achieve perforated patch recordings. Amphotericin was dissolved in dimethyl sulfoxide (final concentration: 0.1 -0.3%; DMSO; D8418, Sigma-Aldrich) 34 ; and was added to the modified pipette solution shortly before use. The used DMSO concentration had no obvious effect on the investigated neurons. During the recordings access resistance (Ra) was constantly monitored and experiments were started after Ra and the action potential amplitude were stable (~15 -20 min). Recordings with Ras > 50MΩ were not considered for analysis of intrinsic electrophysiological parameters. In the analyzed recordings Ras were comparable, did not change significantly over recording time, and were not significantly different between the distinct experimental groups. The mean Ra of all recordings was 41.3 ± 2.8 MΩ (n = 48). For the analysis of the action-potential waveform between the three different experimental groups (n=34; Supplementary Figure 1 The imaging setup consisted of an Imago/SensiCam CCD camera with a 640x480 chip (Till Photonics) and a Polychromator IV (Till Photonics) coupled via an optical fiber into the upright microscope. Fura-2 was excited at 340 nm, 360 nm or 380 nm (410 nm dichroic mirror; DCLP410, Chroma). Emitted fluorescence was detected through a 440 nm long-pass filter. Data were acquired as 80x60 frames using 8x8 on-chip binning. Images were recorded in arbitrary units (AU) and stored and analyzed as 12-bit grayscale images. Before establishing the perforated-patch clamp recordings fura-2 was loaded into the neurons by electroporation (1 V with 1 ms pulse duration at 65 Hz for 10 -15 s). The loading pipette contained intracellular saline and 3.6 mM fura-2 (pentapotassium salt, F1200, Life Technologies). Loading was monitored at 360 nm excitation to reach comparable loading states (129 ± 9 AU, n = 48).
To measure Ca 2+ dynamics, pairs of frames excited with 340 nm and 380 nm were taken at 25 Hz (pacemaker activity) or 10 Hz (current induced actionpotentials, APs). The mean AU value within a region of interest (ROI) from the soma was determined. The ROI was adjusted for each cell. For background subtraction for the whole-time series an adjacent, second ROI was chosen. Data were analyzed as normalized fura-2 F340/F380 ratios. Mean amplitudes of 20 oscillations were calculated for each analyzed neuron. Ca 2+ imaging experiments were designed to assess if the actionpotential (AP) induced fluctuations of free intracellular Ca 2+ (referred to as Ca 2+ oscillations) are reduced in SN dopaminergic neurons from Cav2.3 knockout compared to Cav2.3 wildtype mice due to SNX-482. To achieve this, the relevant experimental conditions (e.g. indicator loading, AP-frequency) were carefully controlled to ensure equal conditions in all analyzed neurons. Since AP-associated Ca 2+ dynamics are strongly frequency dependent, the AP frequency was adjusted for Ca 2+ imaging in all recorded neurons to a similar value of ~1.5 Hz 3 . The distributions of spontaneous frequencies between the experimental groups were similar for a given cell type and accordingly the mean currents that were necessary to adjust the frequency to ~1.5 Hz were not significantly different (Supplementary Table 6B/D). For pharmacology, 100 nM SNX-482 (Tocris) was bath applied (in ACSF) for at least 15 min before and during recordings. As SNX-482 besides Cav2.3 also inhibits voltage-sensitive A-type Kv4.3 K + channels with high nanomolar affinity 35 , and these channels are prominently expressed in SN dopaminergic neurons and crucial for pacemaker frequency 7,36 , SNX-482 Ca 2+ imaging experiments were performed in the presence of the A-type K + channel blocker 4-aminopyridine (4 mM, Sigma) for full channel inhibition. Since SNX-482 may also inhibit other voltage-gated Ca 2+ channels at higher concentrations 37 , we have used a low concentration of 100 nM that should only inhibit Cav2.3 -but that does not fully inhibit Cav2.3 channels. For pharmacological experiments, neurons were hyperpolarized to ~ -70 mV and depolarizing currents (50 -70 ms, 50 -80 pA) were injected to induce two action potentials.
Recordings from voltage step protocols were fitted with a modified Boltzmann function (formula 1), in GraphPad where I is the peak current amplitude, Gmax is the maximal conductance [µS], V is the respective test potential [mV], Vrev the reversal potential (135 mV calculated via Nernst equation), V0.5 is the half-maximal activation voltage [mV], and k is the slope factor. Each individual cell was fitted, and also mean values for all cells were fitted.
Conductance was fitted by a Boltzmann sigmoidal function (formula 2) in GraphPad Prism.
where Gmax is the maximal conductance [nS], V0.5 is the half-maximal activation voltage [mV], and k is the slope factor.
Area under the curves (AUC) were calculated using the integral function in GraphPad Prism.

RNAScope® in situ hybridization
In situ hybridization experiments were performed on fresh frozen mouse brain tissue using the RNAScope® technology (Advanced Cell Diagnostics, ACD), according to the manufacturer's protocol under RNase-free conditions. Briefly, 12 µm coronal cryosections were prepared 5 , mounted on SuperFrost® Plus glass slides and dried for one hour at -20°C. Directly before starting the RNAScope procedure, sections were fixed with 4% PFA for 15 min at 4°C and dehydrated using an increasing ethanol series (50%, 75%, 100%, 100%), for 5 min each. After treatment with protease IV (ACD, Cat# 322340) for 30 min at room temperature, sections were hybridized with the respective target probes for 2 h at 40°C in a HybEZ II hybridization oven (ACD). Target probe signals were amplified using the RNAScope Fluorescent Multiplex Detection Kit (ACD, Cat# 320851). All amplifier solutions were drop on respective sections, incubated at 40°C in the HybEZ hybridization oven and washed twice with wash buffer (ACD) between each amplification step, for 2 min each. Nuclei were counterstained with DAPI ready-to-use solution (ACD, included in Kit) and slides were coverslipped with HardSet mounting medium (VectaShield, Cat# H-1400) and dried overnight.
Target probes were obtained from the library of validated probes provided by Advanced Cell Diagnostics (ACD). Details of all used target probes (RNAScope assays) used for analysis are provided in Supplementary Table 1A. The here used target probes detected but did not discriminate between all known splice variants of the respective target genes, i.e. voltage-gated Ca 2+ channel α1 subunits. Target genes, visualized with Atto550 fluorophore, were co-stained with Tyrosine hydroxylase (TH), and visualized with AlexaFluor488, as a marker for dopaminergic neurons. The gene peptidyl-prolyl isomerase B (PPIB) was used as positive control.
Fluorescent images of midbrain sections were acquired by a Leica CTR6 LED microscope using a Leica DFC365FX camera as z-stacks, covering the full depth of cells at 63x magnification. Z-stacks were reduced to maximum intensity Z-projections using Fiji (http://imagej.net/Fiji) and images were analyzed by utilizing a custom-designed algorithm (Wolution, Munich, Germany). The algorithm delineates cell shapes according to the TH marker gene signal and quantifies the area of fluorescent staining. According to Advanced Cell Diagnostics (ACD), target probe hybridization results in a small fluorescent dot for each mRNA molecule, allowing quantification of absolute number of mRNA molecules independent from fluorescent signal intensity.

Multiplex-nested PCR and qPCR analysis
A multiplex-nested primer approach was used only for qualitative PCRs. Qualitative multiplexnested PCR and quantitative qPCRs were carried out 2-6 , using a GeneAmp PCR System 9700 thermocycler (Thermo Fisher Scientific) and the following PCR conditions. For multiplex PCR Only samples expressing a correct qualitative marker gene profile, defined by qualitative multiplex-nested PCR, were further analyzed by qPCR. The correct qualitative marker gene profile was defined as positive for TH, and negative for calbindin d28k (CB), glial fibrillary acidic protein (GFAP), and glutamic acid decarboxylase (GAD65/67).
Details of all primers, qPCR assays (TaqMan®), and standard curve parameters used for analysis are provided in Supplementary Table 2A & 3. For qPCR assays, we tested at least three different TaqMan assays (i.e. a probe, flanked by two primers), targeting the same gene of interest, for each gene we analyze. For each individual gene (in Supplementary Table 3) only the assays are given that resulted in the most optimal performance over four orders of magnitude of template molecules (ideal slope: -3.32), the most robust reproducibility, and the highest sensitivity. All qPCR amplicons are very small (between 57-73 bp) (Supplementary Table 3). For absolute quantification of Cav1.2, Cav1.3, Cav2.3, and NCS-1 cDNA molecule numbers (data given in Figure  1b, Figure 4d and Supplementary Table 4A/E), defined amounts of cDNA molecules were utilized to generate absolute real-time PCR standard curves 3,7 , and further illustrated in cartoon (B), next page.
To obtain DNA for generation of absolute standard curves, cDNA fragments covering the respective TaqMan® assay locations (used for qPCR) were amplified by PCR (primer sequences are given in Supplementary Table 2B, REDTaq Ready Mix, Taq Polymerase, Merck, 125 U, 50 µl total volume): 3 min 94°C; 35 cycles: 30 s 94°C, 1 min 63°C, 1 min 72°C; 7 min 72°C) and after gel electrophoresis, PCR products were purified (QIAquick Gel Extraction Kit, Qiagen), and quantified (with a Qubit 3.0 fluorometer). DNA molecule numbers were calculated by determining the molecular weight of the individual dsDNA fragments, and by using Avogadro's number. DNA samples containing 1,000,000 molecules down to 1 molecule (in 10-fold diluted steps) were used to generate absolute cDNA standard curves (pipetted in parallel with the respective SN dopaminergic neuron derived cDNA-samples). These curves were used to calculate absolute cDNA molecule numbers, as further illustrated in cartoon (B), next page.
Absolute quantification of cell-specific cDNA levels is a much more complex, difficult, and timeconsuming approach 3,7 ; however only with this approach, the comparison of expression levels between different genes and different ages is valid. Thus we applied it only when necessary to prove a hypothesis (i.e. Figure 1b and Figure 4d but not Figure 3d and Figure 4c). Note again that all qPCR amplicons -for relative and for absolute quantification -are between 57-73 bp only (Supplementary Table 3).
Relative qPCR quantification data are given as cDNA amount [pg/cell] in respect to midbraintissue cDNA standard curves. Data were calculated according to formula (3)   Absolute qPCR data are given as cDNA molecules per cell, derived from absolute cDNA standard curve values for Y-intercept and slope (see Supplementary Table 3). Data were calculated according to formula (4) 6 (with S, Nocells, and cDNA fraction, as defined above): • Histological analysis and immunostaining Briefly 12 , after perfusion and post-fixation in 4% PFA in PBS (pH 7.4) at 4°C overnight, the brains were stored in 0.05% NaN3 in PBS at 4°C. Before cutting, brains were incubated for 1 h in cutting solution (10% sucrose and 0.05% NaN3 in PBS) at 4°C. 30 µm coronal midbrain sections for stereology and for immunofluorescence and 100 µm coronal striatal sections for striatal fiber densitometric analysis were cut with a vibratome (VT 1000S, Leica). Free-floating slices were washed three times in PBS for 10 min each with shaking (300 rpm, microplate shaker, VWR) and treated for 2 h with a blocking solution (10% normal goat serum, 0.2% BSA and 0.5% Triton X-100 in PBS). After washing the slices with PBS, the primary antibody was applied in solution (1% goat serum, 0.2% BSA and 0.5% Triton X-100 in PBS).
Rabbit anti-TH for immunohistochemistry was incubated overnight at room temperature, the other primary antibodies used for immunofluorescence were incubated at 4°C while shaking to better preserve antigen integrity (300 rpm, Microplate shaker, VWR). Slices were washed three times in 0.2% Triton X-100 in PBS for 10 min and incubated with the respective secondary antibody for 2 h (for immunohistochemistry) or for 3 h (for immunofluorescence) at room temperature while shaking (300 rpm, Microplate shaker, VWR) in the solution described above.
The specificity of the Cav2.3 antibody used for immunofluorescence (Proteintech 27225-1-AP) was tested by Western blots and by immunofluorescence on sections from Cav2.3 wildtype and Cav2.3 knockout mice (Supplementary Figure 6). For these Western blots we used 30 µg of total brain lysates pre-heated at 30°C before loading and the Cav2.3 antibody was diluted 1:4000. As a loading control we used anti beta-actin antibody (1:5000, as-15, abcam).
The specificity of the Cav2.3 antibody used for the Western blots on iPSCs was already previously proved by Western blots in brain microsomal membrane lysates from Cav2.3 wildtype and Cav2.3 knockout mice, and published 9,39,40 . Note that the Cav2.3 antibody used for iPSC derived Western blot data was generated against human Cav2.3 protein.
Immunofluorescent stained sections were mounted on glass slides and the coverslip was fixed with VectaShield Mounting hard-set medium with DAPI (Cat#: H-1500, Vector Laboratories) for confocal imaging.
For automated counting of neurons, the sections were incubated in xylene to remove the coverslips and rehydrated by immersion in ethanol series (100%, 90%, 70%, 50%) and H2O (5 min each). After drying for 5 min, the slides were incubated with Vector hematoxylin QS (2 min) for counterstaning of the nuclei (Vector Hematoxylin QS Kit, Cat#: H-3404, Vector Laboratories), after dehydration and xylene, coverslips fixed by Vectamount.

Optical densitometry, stereological and automated counting
Optical densitometry and stereological analysis were performed 12 . Regions of interest namely dorsal striatum (DS), ventral striatum (VS), SN and VTA were identified according to established anatomical landmarks as by K. Franklin and G. Paxinos mouse brain atlas 27 . Analysis of TH-signal in dorsal and ventral striatum is less quantitative than stereological analysis of the remaining THpositive neuron bodies in SN and VTA, and allows only a semiquantitative -but compared to high resolution stereology very quick -analysis of potential effects.
Relative optical densities of striatal TH-immunoreactivity were determined using the digital imaging software (Fiji). Images were acquired with the LMD Software 8.1.0.6156 (Leica) at the LMD7000 microscope. Striatal TH signal was quantified in 18 DS and 10 VS serial sections, covering the whole caudo-rostral axis (Bregma -0.3 to 1.7 for DS and Bregma 0.6 to 1.7 for VS). Respective signals from cortices were subtracted as background.
Total numbers of TH-labeled neurons in SN and VTA were determined with high-resolution stereology 12 , using an unbiased optical fractionator method (StereoInvestigator software; MBF Bioscience). 40 serial TH-stained sections were analyzed, covering the whole caudo-rostral axis (Bregma -3.8 to -2.7).
Estimated total number of TH-positive neurons (N) was calculated for each animal according to equation (5): with ΣQ -= number of counted neurons, t = mean section thickness, h = counting frame height (i.e. 11 µm), asf = area sampling fraction (i.e. 0.44), and ssf = section sampling fraction (i.e. 1 for SN and 2 for VTA). Sampling grid dimensions were 75 x 75 μm (x,y-axes) and counting frame size were 50 x 50 µm (x,y-axes). Reliability of the estimation was evaluated by the Gundersen coefficient (CE, m = 1) according to equation (6): with s 2 = variance due to noise and VARSRS = variance due to systematic random sampling, according to equation (7) for m = 1. CE values were all ≤0.05 for all analyzed animals. Caudorostral axes were generated by plotting the mean relative optical densities, or the mean absolute counted number of neurons for each analyzed section for each animal. The experimenter was blinded for stereological analysis.
For automated neuron counting, stained sections were digitized with a whole slide scanner (3D-Histech Pannoramic 250 Flash III, Sysmex Deutschland GmbH, Norderstedt, Germany). The digital images were acquired at 0.25 µm/pixel resolution using 9 focal layers with 1 µm intervals between layers. The layers were combined using extended focus mode. The digital slides were uploaded on Aiforia® Cloud platform (Fimmic Oy, Helsinki, Finland). The Aiforia platform was used for supervised training of a convolutional neural network (CNN) algorithms to detect all neurons or only TH-positive neurons similarly as in Penttinen et al. 41 . The training was based on nuclear/cell morphology and TH signal. We used 5079 and 4064 neurons to train the CNNs to count all neurons and TH-positive neurons only, respectively. To validate the trained CNNs, we compared the absolute counts of TH-positive neurons per animal with the respective stereological estimates (see Supplementary Figure 5 and Supplementary Table 8A) in SN regions. SN and VTA regions were manually annotated directly on the digital images in Aiforia Cloud. In case of a flawed section the mean value of counted neurons from the section before and the section after was taken, to avoid methodological bias, as with automated counting all neuron were counted, compared to a fraction only and a calculated estimate by stereology.

Semi-quantitative analysis by confocal immunofluorescence
The quantitative analysis of NCS-1 and Cav2.3 antibody immunofluorescence signals was performed in mouse TH-positive SN and VTA neurons 36,42 . High-resolution images of midbrain sections, immunostained by either NCS-1 or Cav2.3 together with TH and DAPI (4′,6-diamidino-2phenylindole) were acquired by a Zeiss LSM 710 confocal microscope (Carl Zeiss, Germany) using a 63x oil objective. Settings were adjusted by the ZEN software (Carl Zeiss, Germany).
Fluorochromes were excited using an argon laser at 405 nm, 488 nm and 546 nm for DAPI, NCS-1 or Cav2.3 and TH respectively, avoiding signal saturation and keeping confocal settings identical for all analyzed neurons / animals. The LUT (Look Up Table) function provided by the Range Indicator in the Zeiss software was used to ensure that the fluorescence signal for quantification was not saturated. In particular, the Range Indicator allows to visualize saturating pixels, i.e. pixel values above a threshold of 200, in red. We ensured that the acquisition had no saturated pixels by reducing the laser power, detector gain and amplifier offset according to the method information related to the Zeiss LSM710 ZEN software. In this way we ensured to be within the dynamic range of the detector critical for semi-quantitative analysis (0-255 for 8-bit images and 0-65536 for 16-bit images).
For NCS-1 we used the following parameters: master gain = 655 for NCS-1, and 800 for TH, 750 for DAPI; digital Gain = 1; shutter speed = 4; pinhole diameter = 29.9, 0.6 AU; one directional acquisition with 1912 X 1912 frame size; bit depth = 12. For Cav2.3 we used: master gain = 800 for Cav2.3, and 800 for TH, 750 for DAPI; digital Gain = 1; shutter speed = 4; pinhole diameter = 29.9, 0.6 AU; one directional acquisition with 1912 X 1912 frame size; bit depth = 12. All the images were converted to 8-bit (0-255), and only TH-positive neurons showing a visible nucleus were used for the analysis. TH immunoreactivity was used to identify and mark TH-positive cytoplasmic regions of interest (ROIs). The DAPI staining was used to identify and mark the nucleus of TH-positive cells, used as background noise ROI. The NCS-1 signal intensities were measured in the cell body / cytoplasm ROI (note that some minor expression is described for (peri)nuclear regions). Cav2.3 signal intensities were measured in the plasma membrane ROI by using the Fiji software 43 (https://fiji.sc/). We determined average signal intensities / pixel of the antibody fluorescent signal in the plasma membranes or cell bodies of individual TH-positive SN and VTA neurons. These values are independent from the respective areas of the neurons. The individual signal intensities were normalized to the background signal in the respective cell nuclei for all analyzed neurons, and given relative to that of VTA DA neurons (=1). The high background signal for NCS-1 and its overlapping Gaussian curve with the signal can be explained by a possible signal of NCS-1 in the perinuclear space as it has been shown in dopaminergic like PC12 neurons and HeLa cells 44 and also in cardiomyocytes 45 . For each mouse, seven coronal midbrain sections (30 µm after fixation, covering the full caudo-rostral axis) and about 20-25 TH-positive neurons per section (130-150 TH-positive DA neurons per animal) were analyzed for the SN as well as for the VTA respectively (N=4 mice).
Cells were extracted with cold PBS and centrifuged at 500 x g for 5 min. Cell pellets were snap frozen and stored at -80°C until required. Frozen cell pellets were lysed in RIPA buffer on ice (TRIS 1M pH8, iGEPAL, Sodium Deoxycolate 10%, protease inhibitor cocktail (Sigma) 1 tablet in 10 ml). Cell lysates were briefly sonicated (10 sec) then centrifuged 10 min at 4°C at full speed. Supernatant protein concentration was quantified by BCA and diluted into appropriate concentration in Laemmli sample buffer (4X). Extracted samples were boiled for 5 min at 90°C, proteins were separated by SDS-polyacrylamide gel electrophoresis (precast gel 4-15%) with a homemade running buffer (144 g glycine; 30.3 g TRIS Base 10 g SDS for 1 L of 10X buffer) and transferred onto PVDF membrane using Trans-Blot Turbo Transfer System (Bio-Rad) (High molecular weight protocol, midi membrane). Blots were stained with Ponceau S for 1 min.
The western blot was blocked in 5% Milk in PBS-T (0.1%) for 1 h at room temperature before incubation with the primary antibody (rabbit anti-NCS-1, 1:500, Abcam ab129166; and rabbit anti-CaV2.3 common (anti-Nast 197B, self-designed, obtained from Toni Schneider) 49 , 1:500; and mouse anti-beta actin HRP conjugated, 1:10000, Abcam) at 4°C, overnight. It was then washed 2 times for 10 min in PBS-Tween (0.05%) and once for 10 min in TBS-Tween (0.05%) before incubation for one hr at room temperature with secondary antibody (goat anti-rabbit IgG (H+L)-HRP Conjugate, 1:20000, Biorad) diluted in 5% Milk in PBS-T (0.1%). Finally, the blot was washed again as before and imaged on the ChemiDoc-Touch (Bio-Rad) with ECL substrate (Millipore) 1:1. Densitometric analysis of each protein was performed and normalized to actin prior to the calculation of Cav2.3 / NCS-1 ratios. Due to differentiation variability, all 4 control lines (see Supplementary Table 9A) were included in every "differentiation batch/experiment" and normalized to 1 for all respective control data in each differentiation.
Immuno-staining for the established markers neuron-specific class III beta-tubulin (Tuj1, mouse anti-Tuj1 antibody, Biolegend, Cat# 801202, 1:500) and tyrosine hydroxylase (TH, rabbit anti-TH, Sigma, Cat# Ab152, 1:500) has been performed 46 , and electrophysiological properties were assessed in order to attest to differentiation into dopaminergic neurons (Supplementary Figure  7). In our experimental design we chose to use multiple independent lines from different control individuals and Parkinson' disease patients to evaluate the effect of the GBA N370S mutation on dopaminergic neurons across six separate differentiations.

Data analysis
Data analysis and graphical illustrations were performed using FitMaster software (v2x90. Sample sizes for all mouse data were chosen in advised and in agreement with the Institute of Epidemiology and Medical Biometry, Ulm University, and in agreement with EU regulations (Principles of Replacement, Reduction and Refinement (3R's), Directive 2010/63/EU) and as approved by the German "Regierungspräsidium". No particular procedure was required for randomization/allocating wildtype and knockout mice to the respective experimental groups.
Data were tested for outliers via the ROUT function of GraphPad Prism 7. Tests for statistically significant differences were performed in GraphPad Prism 7 using statistical tests as indicated in the text. Parametric statistical tests were only chosen if data were normally distributed. Normal distribution was tested with D`Agostino-Pearson omnibus normality test. Unless otherwise stated, unpaired tests were used. Individual test are given in the Supplementary Tables 1-9. Statistical significances are indicated as p < 0.05 (*), p < 0.01 (**), p < 0.001 (***) and p < 0.0001 (****). In graphs, data are given as mean ± SD or SEM as indicated. In boxplots, horizontal lines indicate the median of the data, + sign indicates the mean of the data; boxes indicate the 25 th and 75 th percentile and whiskers were calculated according to the Tukey method. For experiments with low n-numbers more conservative non-parametric tests were applied, even if data were normally distributed. For comparisons of multiple groups, non-parametric Kruskal Wallis test with Dunn's multiple comparisons test or two-way ANOVA with Sidak's multiple comparison test were performed as indicated.

Data availability statement
Most data presented are included in the article and supplementary information. All datasets generated are available from the corresponding author on request.

B) Details of primers for qualitative PCR for amplifying DNA covering the respective TaqMan assays for generation of absolute standard curves (primers not used for quantitative PCR).
Work-flow of the procedures is summarized in the supplementary methods and Figure B within.       Figure   2 (A, C), Supplementary Figure 1 Figure 3 given as mean, ± SD, ±SEM, median and 95% confidence interval (CI  Figure   5b given as mean, ± SD, ±SEM, median and 95% confidence interval (CI