Voltage-Gated Ca2+ Channels in Dopaminergic Substantia Nigra Neurons: Therapeutic Targets for Neuroprotection in Parkinson's Disease?

The loss of dopamine (DA)-producing neurons in the substantia nigra pars compacta (SN) underlies the core motor symptoms of the progressive movement disorder Parkinson's disease (PD). To date, no treatment to prevent or slow SN DA neurodegeneration exists; thus, the identification of the underlying factors contributing to the high vulnerability of these neurons represents the basis for the development of novel therapies. Disrupted Ca2+ homeostasis and mitochondrial dysfunction seem to be key players in the pathophysiology of PD. The autonomous pacemaker activity of SN DA neurons, in combination with low cytosolic Ca2+ buffering, leads to large somatodendritic fluctuations of intracellular Ca2+ levels that are linked to elevated mitochondrial oxidant stress. L-type voltage-gated Ca2+ channels (LTCCs) contribute to these Ca2+ oscillations in dendrites, and LTCC inhibition was beneficial in cellular and in vivo animal models of PD. However, in a recently completed phase 3 clinical trial, the dihydropyridine (DHP) LTCC inhibitor isradipine failed to slow disease progression in early PD patients, questioning the feasibility of DHPs for PD therapy. Novel evidence also suggests that R- and T-type Ca2+ channels (RTCCs and TTCCs, respectively) represent potential PD drug targets. This short review aims to (re)evaluate the therapeutic potential of LTCC, RTCC, and TTCC inhibition in light of novel preclinical and clinical data and the feasibility of available Ca2+ channel blockers to modify PD disease progression. I also summarize their cell-specific roles for SN DA neuron function and describe how their gating properties allow activity (and thus their contribution to stressful Ca2+ oscillations) during pacemaking.


INTRODUCTION
The primary motor symptoms of the neurodegenerative disorder Parkinson's disease (PD) are caused by a progressive loss of dopamine (DA)-producing neurons in the substantia nigra pars compacta (SN) and associated striatal DA depletion (Obeso et al., 2017). Although PD was first described in 1817 (Parkinson, 2002), to date only symptomatic treatments, but still no cure or disease-modifying therapy, exist (Schulz et al., 2016;Obeso et al., 2017). To develop an effective treatment, it is essential to understand the contributing factors and the disease-underlying cellular mechanisms. Next to globally acting factors (e.g., toxins, aging, and genetic mutations), cellautonomous ones have also been proposed and widely studied Surmeier et al., 2017b). Neighboring DA neurons in the ventral tegmental area (VTA) share many of the intrinsic properties but are spared in PD (Dauer and Przedborski, 2003). Both are autonomous pacemakers, but in vulnerable SN DA neurons, large oscillations of intracellular Ca 2+ levels accompany pacemaking (Wilson and Callaway, 2000;Chan et al., 2007;Guzman et al., 2009Guzman et al., , 2018Hage and Khaliq, 2015). In contrast, no or much smaller Ca 2+ transients were detected in VTA neurons (Guzman et al., 2010(Guzman et al., , 2018Benkert et al., 2019) that rely on a different pacemaking mechanism with less contribution of Ca 2+ currents (Khaliq and Bean, 2010;Philippart et al., 2016). Ca 2+ influx is important to modulate neuronal excitability and to activate Ca 2+ -dependent physiological processes, but the rhythmic Ca 2+ load in SN DA neurons also triggers mitochondrial oxidative stress (Guzman et al., 2010(Guzman et al., , 2018Surmeier et al., 2017a). Disrupted Ca 2+ homeostasis and mitochondrial dysfunction are considered key players in PD pathophysiology (Schapira, 2008;Zaichick et al., 2017;Zampese and Surmeier, 2020), and many mutations causing inherited forms of PD affect proteins associated with mitochondrial homeostasis and stress responses (Park et al., 2018). Thus, reducing the activity-related Ca 2+ load and associated mitochondrial stress in SN DA neurons represents a feasible strategy to increase their resistance to degenerative stressors. L-, R-, and T-type voltage-gated Ca 2+ channels (LTCCs, RTCCs, and TTCCs, respectively) contribute to the stressinducing cytosolic Ca 2+ oscillations, and different approaches to decrease their activity produced promising protective effects in preclinical models of PD. However, a phase 3 clinical trial evaluating the efficacy of LTCC inhibition in early PD patients recently failed. This short review aims to give an overview of the (patho)physiological roles of Ca 2+ channel activity in SN DA neurons and to (re)evaluate the therapeutic potential of Ca 2+ channel inhibition (in light of novel preclinical and clinical evidence) and the availability of clinically applicable selective drugs.
Rodent SN DA neurons express all voltage-gated Ca 2+ channel isoforms, except Cav1.1 and Cav1.4 (restricted to skeletal muscle and retina, respectively) (Cardozo and Bean, 1995;Chan et al., 2007;Sinnegger-Brauns et al., 2009;Dufour et al., 2014;Brichta et al., 2015;Shin, 2015;Philippart et al., 2016;Ortner et al., 2017;Guzman et al., 2018;Benkert et al., 2019;Verma and Ravindranath, 2019). The recording of individual current components in intact SN DA neurons is complicated, but in somatic nucleated outside-out patches from juvenile rat SN DA neurons, Ca 2+ currents of all expressed isoforms were found, with a large LTCC contribution that was higher compared to VTA (Philippart et al., 2016). On the transcript level, RTCC Cav2.3 channels are most abundant in mouse SN DA neurons and levels increase with age (Benkert et al., 2019), while LTCCs (Cav1.2/Cav1.3) get downregulated in an age-dependent manner (Branch et al., 2014;Ortner et al., 2017;Benkert et al., 2019). Of the LVA TTCCs, Cav3.1, and Cav3.2 predominate in mouse SN DA neurons (Poetschke et al., 2015;Guzman et al., 2018;Benkert et al., 2019) and immunohistochemical stainings suggest a rise of somatodendritic TTCCs during development (Dufour et al., 2014) [also shown for Cav1.3, but antibody specificity was not demonstrated in brain (SN) tissue lacking Cav1.3]. SN DA neurons are constantly active and fire action potentials in a tonic single spike or transient high-frequency burst mode in vivo, resulting in axonal and somatodendritic DA release (Grace and Bunney, 1984a,b;Chiodo, 1988;Paladini and Roeper, 2014). Cav2 N-and P/Q-type channels drive fast presynaptic neurotransmission, but LTCCs, RTCCs, and TTCCs seem to also contribute to DA release from axonal and/or somatodendritic locations in rodent SN DA neurons (Bergquist and Nissbrandt, 2003;Chen et al., 2006;Brimblecombe et al., 2015;Yee et al., 2019). In vitro, even in complete synaptic isolation, SN DA neurons maintain an intrinsically generated regular pacemaker activity (0.5-4 Hz). Inhibition of Na + channels by tetrodotoxin abolished spike generation and revealed slow oscillatory membrane depolarizations ("SOPs, " slow oscillatory potentials; Fujimura and Matsuda, 1989;Yee et al., 2019) that were absent in neighboring VTA neurons and abolished upon LTCC inhibition (Chan et al., 2007). However, LTCCs are not required for pacemaker generation but rather stabilize precision and robustness of pacemaking (Guzman et al., 2009(Guzman et al., , 2010Poetschke et al., 2015;Ortner et al., 2017). Similarly, pharmacological inhibition of TTCCs decreased pacemaker precision (juvenile) and frequency (adult mice) (Wolfart and Roeper, 2002;Poetschke et al., 2015), while Cav2.3 knockout or its partial pharmacological inhibition reduced spike amplitude and afterhyperpolarization (AHP) (Benkert et al., 2019). Ca 2+ influx through voltage-gated Ca 2+ channels can drive depolarization, but its coupling to K + conductances [e.g., Ca 2+ -sensitive small conductance K + (SK) or A-type K + channels] can also trigger the opposite-a functional coupling important for rhythmic activity (Wolfart and Roeper, 2002;Ji and Shepard, 2006;Duda et al., 2016). During in vivo high FIGURE 1 | Contribution to stressful somatodendritic Ca 2+ oscillations, gating properties, and Ca 2+ current responses of different Ca 2+ channel isoforms during SN DA neuronal pacemaking. (A) Schematic of a plasmalemmal Ca 2+ channel complex consisting of the pore-forming α1 subunit and auxiliary β and α2δ subunits [only HVA α1 subunits require auxiliary subunits for proper function (Dolphin, 2016)]. Membrane depolarization triggers channel opening and subsequent Ca 2+ influx. The middle panel shows mouse SN DA neuron pacemaking (top) with the associated intracellular Ca 2+ transients (bottom; blue) recorded at the cell soma (taken from Ortner et al., 2017). Somatodendritic Ca 2+ oscillations have been linked to increased mitochondrial oxidative stress, implicated in the high vulnerability of SN DA neurons in PD (Guzman et al., 2010). (B) Contribution of different Ca 2+ channel isoforms to intracellular Ca 2+ oscillations in distinct cellular compartments (soma and proximal and distal dendrites) (Ortner et al., 2017;Guzman et al., 2018;Benkert et al., 2019). With a higher isradipine concentration [5 µM (Guzman et al., 2018) compared to 1 µM, 10 nM, or chronic in vivo isradipine application resulting in plasma levels of ∼5 nM isradipine in the abovementioned refs], a complete inhibition of distal dendritic Ca 2+ transients was observed (Guzman et al., 2009(Guzman et al., , 2010. Hyphen indicates no contribution found and question mark indicates not determined. (C-E) The voltage-conductance (activation) curve describes at which potential a certain Ca 2+ channel isoform opens (activation threshold) and is in a Ca 2+ conductive state. The voltage dependence of inactivation gives the proportion of inactivated (non-available) channels at a certain membrane potential. The overlap of these two curves defines the window current (see E) that represents the voltage range at which the respective Ca 2+ channel is steadily active and can thus create a constant background influx of Ca 2+ . (C) Cav1.2 or Cav1.3 α1 with β3 and α2δ1 (2 mM Ca 2+ ; modified from Ortner et al., 2017). Alternative splicing of the Cav1.3 C-terminus results in functionally distinct long (Cav1.3 L , black) and short splice variants (Cav1.3s, gray; Singh et al., 2008;Bock et al., 2011;Tan et al., 2011). (D) Cav2.3e α1 with β3 (black) or β2a (gray) and α2δ1 (2 mM Ca 2+ ). Association with membrane-bound β2a shifts the voltage dependence of inactivation of Cav2.3 channels ∼35 mV toward more positive potentials (Olcese et al., 1994;Jones et al., 1998;Yasuda et al., 2004;Miranda-Laferte et al., 2012). (E) Voltage dependence of activation and inactivation of TTCCs with indicated window current (blue area; taken from Weiss and Zamponi, 2019a). (F) Representative Ca 2+ current traces (I Ca , lower panel) through Cav1.3 S in response to a murine SN DA neuron action potential command voltage shown above (2 mM Ca 2+ , tsA201 cells; modified from Ortner et al., 2017). Simulated pacemaking (2.5 Hz) resulted in a decrease of I Ca to ∼20% in steady state (cyan trace, arrow indicates decreased peak I Ca ) through all three investigated Cav1 subtypes (Cav1.2, Cav1.3 S , and Cav1.3 L ). (G) Enlargement of the action potential spike region indicated in (F) (dotted rectangle). During the interspike interval (before the spike threshold indicated by the dotted vertical line), only Cav1.3 variants conducted Ca 2+ while all LTCCs showed Ca 2+ influx in response to the action potential spike. The horizontal current trace shows full LTCC inhibition at the end of the recording (3 µM isradipine).
Experiments in transgenic mice (mito-GFP mouse) that allow to monitor the oxidation state of mitochondria established a link between Ca 2+ oscillations and elevated mitochondrial oxidative stress (Guzman et al., 2010(Guzman et al., , 2018; Figure 1A). Dendritic Ca 2+ transients and associated mitochondrial oxidation increased with age (Guzman et al., 2018) and were almost absent in neighboring resistant VTA DA neurons (Guzman et al., 2010(Guzman et al., , 2018Benkert et al., 2019). In addition, the already high basal oxidant stress level in SN DA neurons was further exacerbated in a genetic PD mouse model (DJ-1 knockout) (Guzman et al., 2010). Interestingly, factors contributing to vulnerability described in SN DA neurons, i.e., slow pacemaking, cytosolic Ca 2+ oscillations, low intracellular Ca 2+ buffering (Foehring et al., 2009), and elevated levels of mitochondrial oxidant stress, are also found in other vulnerable non-DA neurons (Surmeier et al., 2017b;Zampese and Surmeier, 2020).
Besides the metabolic challenging pacemaker activity, burst firing is also associated with high Ca 2+ influx and intracellular Ca 2+ levels (Hage and Khaliq, 2015;Philippart et al., 2016;Ortner et al., 2017), and membrane hyperpolarizations (e.g., during postburst pauses) allow channels to recover from inactivation, in particular TTCCs. This mechanism underlies the large TTCCmediated Ca 2+ currents and the resulting afterdepolarizations that facilitate rebound spiking (Evans et al., 2017;Tracy et al., 2018). Interestingly, in PD, cells in the ventrolateral part of the SN are particularly prone to cell death and lateral SN DA neurons also show high in vivo bursting (Schiemann et al., 2012;Farassat et al., 2019). In line, knockout of K ATP K + channels reduced in vivo burst firing and conferred protection in two PD mouse models (Liss et al., 2005;Schiemann et al., 2012).

NEUROPROTECTION BY LTCC INHIBITION: PRECLINICAL AND CLINICAL EVIDENCE
Over the last years, LTCCs (and particularly Cav1.3) were considered the main voltage-gated Ca 2+ channel subtype underlying stressful Ca 2+ oscillations and thus a major driver of SN DA neuronal cell death (Surmeier et al., 2017a;Guzman et al., 2018;Liss and Striessnig, 2019). Epidemiological studies found that the intake of brain-permeable DHP LTCC inhibitors (antihypertensives) reduced the risk to develop PD (Gudala et al., 2015;Lang et al., 2015;Mullapudi et al., 2016). DHPs have been extensively studied in preclinical PD models and showed promising protective effects in most (but not all) studies (Table 1; reviewed in Leandrou et al., 2019;Liss and Striessnig, 2019). The diverging outcomes of DHP treatment in toxin-based PD animal models have been recently discussed in great detail (Liss and Striessnig, 2019), but no definite explanation has been found.
Briefly, in eight out of 13 reports, DHPs significantly reduced mitochondrial-targeting toxin-induced SN DA cell death in mice, rats, and primates (Table 1), but the experimental design of all studies varied, complicating an overall conclusion. Differences included the used PD model (6-OHDA and MPTP), animals (species, strain, age, and sex), treatment regimen (DHP, treatment onset, route of administration, and dosing interval), readout (approach and methodology), and plasma concentrations (if even reported) (Liss and Striessnig, 2019). Thus, a standardized approach in future studies would help to better interpret and compare obtained results. In addition, one recent study (not included in Liss and Striessnig, 2019) tested the DHP felodipine in a genetic PD mouse model (expressing the A53T mutant α-synuclein) and found protective effects on SN DA neuron survival and motor deficits and autophagy-induced clearance of disease-associated proteins from brain (Siddiqi et al., 2019). Noteworthy, in contrast to other DHPs like isradipine (Uchida et al., 1997), felodipine accumulated in the brain with ∼2 to 5fold higher brain levels compared to plasma (Siddiqi et al., 2019). Epidemiological and preclinical evidence, and the availability of safe and clinically approved LTCC DHP inhibitors, prompted the phase 3 STEADY-PD III clinical trial with the DHP isradipine as treatment in early PD patients (336 patients, randomized, double blind, and placebo controlled; Biglan et al., 2017), which however did not reach its primary endpoint (Parkinson Study Group STEADY-PD III Investigators, 2020). Several aspects may have contributed.
Protective DHP effects might require an earlier treatment onset, since at the time when first PD symptoms occur, pathologic mechanisms and neurodegeneration already started-which however would require reliable PD biomarkers . In addition, the age-dependent decrease of LTCCs (Branch et al., 2014;Ortner et al., 2017) could limit the therapeutic window of DHPs in PD (especially in elderly patients), although disease state seems to affect LTCC expression [SN neurons from MPTP-treated mice (Verma and Ravindranath, 2019) and post-mortem brains of early-stage PD patients (Hurley et al., 2015) showed robust Cav1.3 levels, despite profound neuron loss]. When considering Cav1.3 as the primary target, the used maximal tolerable isradipine dose [10 mg/day; limited by Cav1.2mediated peripheral side effects (Parkinson Study Group, 2013)] might be too low to sufficiently engage Cav1.3 LTCCs in SN DA neurons due to their low apparent drug sensitivity (Ortner et al., 2017). This especially applies to C-terminally short splice variants (Huang et al., 2013;Ortner et al., 2017) that are associated with higher Ca 2+ influx (Singh et al., 2008;Bock et al., 2011;Tan et al., 2011; Figure 1C) and account for ∼50% of Cav1.3 transcript in mouse SN DA neurons (Ortner et al., 2017;Verma and Ravindranath, 2019) (higher compared to the cortex and striatum; Verma and Ravindranath, 2019). In this context, the use of an immediate-release isradipine formulation in the STEADY-PD III trial was unfavorable, since slow-onset continuousrelease tablets [as used in the phase 2 trial (Parkinson Study Group, 2013), NCT00753636] result in higher average steadystate plasma levels (Liss and Striessnig, 2019). Thus, strategies to increase DHP brain concentrations (Yiu and Knaus, 1996;Ji et al., 2017) or DHPs that accumulate in the brain (e.g., nimodipine and felodipine; Kupsch et al., 1996;Siddiqi et al., 2019) could be an option. Noteworthy, microglia-specific Cav1.2 knockdown augmented MPTP-induced SN DA neurodegeneration and motor deficits in mice, associated with enhanced activation of "neuroinflammatory" M1 microglia (Wang et al., 2019). Thus, Cav1.2 inhibition might even be disadvantageous, which highlights the need for the development of reliable Cav1.3selective inhibitory drugs. So far, only one putatively selective compound has been described (Cp8 in Kang et al., 2012 andcp-PYT in Cooper et al., 2020) but showed diverging results in follow-up studies from other laboratories (Huang et al., 2014;. Nevertheless, micromolar concentrations of cp-PYT could lower dendritic transients to a similar extent as isradipine and did not affect Ca 2+ currents in Cav1.3 knockout mice (Cooper et al., 2020), suggesting Cav1.3-selective inhibition [lack of non-LTCC modulation was also shown in adrenal mouse chromaffin cells ]. In the meantime, mice expressing DHP-insensitive Cav1.2 channels could be used to mimic selective pharmacological Cav1.3 inhibition in vivo (Sinnegger-Brauns et al., 2004). Lastly, another important aspect is a potential compensation by other Ca 2+ channel isoforms during chronic isradipine treatment, as found in Cav1.3deficient mice [upregulation of Cav3.1 TTCCs (Poetschke et al., 2015)]. However, this was not observed with systemic isradipine treatment for 7-10 days in mice [3 µg/g/day; ∼5 nM plasma isradipine (Guzman et al., 2018)], but longer exposure has not been tested.

RTCCs AND TTCCs EMERGE AS NOVEL PD DRUG TARGETS
Global Cav2.3 knockout fully prevented SN DA neuron degeneration in the gold-standard chronic low-dose MPTP/probenecid PD mouse model ( Table 1) and profoundly reduced somatic Ca 2+ oscillations (∼50%, dendrites not tested) (Benkert et al., 2019). Although a direct proof for knockoutinduced lowering of high mitochondrial stress levels is missing, a similar inhibition of LTCC-mediated Ca 2+ transients in proximal and distal dendrites (∼35-60%) was sufficient to lower mitochondrial oxidation (Guzman et al., 2010(Guzman et al., , 2018. The high Cav2.3 levels in vulnerable SN DA neurons (compared to VTA) and their increase with aging further strengthen a possible involvement of RTCCs in PD pathology. Unlike what is observed for Cav1.3 (Poetschke et al., 2015), loss of Cav2.3 did not trigger compensatory upregulation of other Ca 2+ channels in SN DA neurons (Cav1.2, Cav1.3, and Cav3.1 were tested) (Benkert et al., 2019). In light of these promising findings, it is unfortunate that to date (like for Cav1.3) no selective RTCC inhibitors exist (Schneider et al., 2013). Note that the peptide toxin SNX-482 is selective for Cav2.3 channels only at low concentrations and also inhibits other channels at elevated concentrations ("Cav2.3prevalent") (Newcomb et al., 1998;Bourinet et al., 2001;Schneider et al., 2013;Kimm and Bean, 2014). As mentioned above, association of Cav2.3 channels with membrane-bound β2 splice variants (β2a/β2e) may stabilize sustained Cav2.3 activity-a possible explanation for the large contribution of   Tabata et al., 2018). Preclinical studies with LTCC inhibitors have been recently reviewed (Leandrou et al., 2019;Liss and Striessnig, 2019 Evidence for a role of TTCCs in PD pathology comes from a study employing DA neurons derived from PD patientspecific iPSCs as in vitro PD model (Tabata et al., 2018). Compared to controls, PD patient-specific DA neurons showed pathological signs including reduced neurite length, enhanced oxidative stress, and elevated resting intracellular Ca 2+ levels and apoptosis. Interestingly, levels of all three TTCC isoforms were also increased. Treatment with the mitochondrial toxin rotenone ("PD trigger") aggravated the observed effects on cell morphology and survival, which could be prevented by drugs targeting TTCCs as well as individual knockout of all TTCC isoforms. Two of the TTCC inhibitory compounds have complex mechanism of action besides inhibition of TTCCs, but one drug, ML218, specifically blocks TTCCs (Xiang et al., 2011). Interestingly, while LTCC inhibition (nifedipine and isradipine) had no effect, the DHP benidipine that acts on L-, N-, and T-type Ca 2+ channels could prevent rotenone-induced apoptosis. Although promising, proof of SN DA neuroprotection in in vivo PD models is still missing. TTCC inhibition has been explored for the treatment of parkinsonism, but mainly based on their involvement in abnormal burst discharges within the thalamocortical circuitry and associated motor effects (Tai et al., 2011;Kopecky et al., 2014;Yang et al., 2014;Galvan et al., 2016). Zonisamide, an unselective TTCC inhibitor, ameliorated symptoms in PD patients (Murata, 2010), but due to its broad mechanism of action, it is difficult to assign observations to TTCC inhibition. Selective and clinically suitable TTCC inhibitors exist (analgesic/antiepileptic drugs) (Weiss and Zamponi, 2019a,b) and showed good tolerability and safety in phase 2 clinical trials (Richard et al., 2019(Richard et al., , 2020, and one phase 2 trial plans to evaluate the selective blocker CX-8998 as treatment for PD-associated tremor (ClinicalTrials.gov #NCT03436953).

SUMMARY AND CONCLUSION
There is accumulating evidence that voltage-gated Ca 2+ channels represent an attractive drug target for the therapy of PD. Their contribution to large somatodendritic Ca 2+ oscillations, associated with increased mitochondrial oxidative stress, seems a likely mechanism by which their activity contributes to SN DA neuron degeneration. Pharmacological and genetic strategies to decrease the activity of LTCCs, RTCCs, and TTCCs showed promising neuroprotective effects in preclinical models of PD, but a phase 3 clinical trial found no slowing of disease progression in early PD patients upon treatment with the DHP LTCC inhibitor isradipine. Although DHPs are safe, brain permeable, and clinically available, their lack of selectivity for Cav1.3, the more likely LTCC target for neuroprotection, and adverse Cav1.2mediated effects limit their potential for PD therapy. Selective TTCC blockers exist, show good tolerability and safety, and could be repurposed for the therapy of PD; however, evidence for neuroprotection in in vivo PD models is still missing. In contrast, no selective inhibitors for Cav1.3 and Cav2.3 exist. The convincing preclinical data described in this review and the discovery of activity-enhancing mutations in neurological diseases (Ortner et al., 2020;Schneider et al., 2020;Weiss and Zamponi, 2020) highlight the urgent need for isoform-selective blockers that are suitable for clinical application. Further studies with available TTCC or multiple Ca 2+ channel targeting blockers and, if once available, Cav1.3and Cav2.3-selective inhibitors will help to uncover the full therapeutic potential of Ca 2+ channel inhibition for neuroprotection in PD.

AUTHOR CONTRIBUTIONS
The author confirms being the sole contributor of this work and has approved it for publication.

FUNDING
This work was supported by the Erika-Cremer habilitation fellowship of the University of Innsbruck and the Austrian Science Fund (P27809).