LRRK2 deficiency induced mitochondrial Ca2+ efflux inhibition can be rescued by Na+/Ca2+/Li+ exchanger upregulation

Variants of leucine-rich repeat kinase 2 (lrrk2) are associated with an increased risk in developing Parkinson’s disease (PD). Mitochondrial dysfunction and specifically mitochondrial Ca2+ handling has been linked to the pathogenesis of PD. Here we describe for the second time a mitochondrial Ca2+ efflux deficiency in a model displaying alterations in a PD-associated risk protein. LRRK2 deletion, inhibition and mutations led to an impaired mitochondrial Ca2+ extrusion via Na+/Ca2+/Li+ exchanger (NCLX) which in turn lowered mitochondrial permeability transition pore (PTP) opening threshold and increased cell death. The mitochondrial membrane potential was found not to be the underlying cause for the Ca2+ extrusion deficiency. NCLX activity was rescued by a direct (phosphomimetic NCLX mutant) and indirect (protein kinase A) activation which in turn elevated the PTP opening threshold. Therefore, at least two PD-associated risk protein pathways appear to converge on NCLX controlling mitochondrial Ca2+ extrusion and therefore mitochondrial health. Since mitochondrial Ca2+ overload has been described in many neurological disorders this study warrants further studies into NCLX as a potential therapeutic target.


Introduction
Parkinson's disease is a common, disabling, incurable neurodegenerative condition affecting 1% of the population over the age of 60. The past 10 years have seen a shift in the aetiological understanding of PD, moving from a nearly exclusively environmentally mediated disease towards a complex disorder with important genetic contributions 1 . One of the most extensively studied genes containing these mutations is the leucine-rich repeat kinase 2. In particular, the missense mutation G2019S has been reported to be the underlying cause for PD in 1-2% in the UK/USA rising to 10% in Ashkenazi Jews and 40% in North African Arab Berbers 2,3 . LRRK2 possesses a kinase and a ROC-GTPase domain whose exact physiological function is yet to be fully unravelled but studies provided evidence for a role of LRRK2 in vesicle trafficking, inflammatory responses, autophagy and neurite outgrowth [4][5][6][7] ; with few phosphorylation substrates identified 8,9 . The multitude of pathways regulated by LRRK2 demonstrate the complexity of LRRK2 mutations and their potential effects on cellular signalling pathways. In PD, it has long been known that LRRK2 mutations (in particular G2019S) lead to an increased kinase activity and this upregulation is currently being targeted by LRRK2 kinase inhibitors as a therapeutic agent in PD 10-12 . Mitochondrial dysfunction is strongly implicated in PD as toxic models of PD which employ mitochondrial toxins as well as the majority of proteins associated with familial forms of PD affect mitochondrial function. Recently, we have shown how another PD-associated protein, PINK1, is involved in mitochondrial Ca 2+ homoeostasis suggesting a potential role for the recently described sodium Ca 2+ exchanger (NCLX) in PD pathology [13][14][15] .
While mitochondrial Ca 2+ shuttling couples between Ca 2+ signalling and metabolic activity, its impairment can lead to mitochondrial Ca 2+ overload that can be a trigger for neuronal cell death. We have shown that the mitochondrial Ca 2+ overload phenotype in PINK1 deficient cells can be rescued by a PKA/NCLX-mediated pathway 16 . Furthermore, α-synuclein has been linked to Ca 2+ -induced cell death 17,18 suggesting that inhibition of mitochondrial Ca 2+ efflux may be a common feature in the mechanism of neurodegeneration observed in PD. To test this hypothesis, we used LRRK2 deficient cells and assessed mitochondrial Ca 2+ homoeostasis.
This current study established that mitochondrial Ca 2+ homoeostasis is a common phenotype in PD and not specific to PINK1 deficiency only.

Animals
Wild-type and LRRK2 KO mice were obtained from breeding colonies generated by Jackson Laboratories and Sprague Dawley rat pups (1-3 days postpartum) were acquired from the University College London breeding colony. Wild-type rat of either sex were used for neuronal co-cultures and experimental procedures were performed in full compliance with the United Kingdom Animal (Scientific Procedures) Act of 1986.

Cell culture
Mixed cultures of cortical and midbrain neurons and glial cells were prepared from postnatal pups (Control/ LRRK2 KO mice or rats; day 0-3; UCL breeding colony). Midbrain were removed and placed into ice-cold Ca 2+ /Mg 2+ free PBS (Invitrogen, UK). The tissue was minced and trypsinised (0.25%; 5 min at 37°C), triturated and plated on poly-D-lysine-coated coverslips. The tissue was cultured in Neurobasal A medium (Invitrogen, UK) supplemented with B-27 (Invitrogen, Paisley, UK) and 2 mM GlutaMAX (Invitrogen, UK) 19 . Cultures were maintained at 37°C in a humidified atmosphere of 5% CO 2 and 95% air; media was changed twice weekly and maintained for a minimum of 12 days before experimental use to ensure expression of glutamatergic and other receptors. Only neurons were used for analysis. Neurons were distinguishable from glia: they appeared phase bright, had smooth rounded somata and distinct processes, and lay just above the focal plane of the glial layer. Cells were used at 12-15 days in vitro. Experimental procedures were performed in full compliance with the United Kingdom Animal (Scientific Procedures) Act of 1986.
Fibroblasts were generated from a skin punch biopsy taken under local anaesthetic following local ethical approval and full informed consent 20 . Biopsies were dissected into~1mm pieces and cultured in DMEM, 10% FBS, 1% GlutaMAX until fibroblasts were seen to grow out from the explants. When fibroblasts reached confluency, they were detached from culture dishes using TrypleE (Invitrogen) and transferred to larger culture vessels for further expansion and cryopreservation.

Plasmids and transfection
Preparation of human WT NCLX plasmids and sitedirected mutagenesis was carried out as previously described 16,21 . Plasmids were transfected using Effectene (Qiagen) according to manufacturers protocol. NCLX plasmids were co-transfected with an empty GFP vector (pcDNA3 backbone) to allow for selection of transfected cells for experimentation and analysis.
For TMRM experiments, cells were placed in a HBSS containing 25 nM TMRM for 40 min at room temperature. TMRM was excited using the 560 nm laser line and fluorescence was measured > 580 nm using a Zeiss 710 VIS CLSM equipped with a META detection system and a ×40 oil-immersion objective. Z-stack images were obtained by confocal microscopy and the basal Δψm was measured using Zen software (Zeiss).
Mitochondrial Ca 2+ efflux was assessed in permeabilized cells. To permeabilize, a buffer (0.137 M NaCl, 5 mM KCl, 0.7 mM NaH 2 PO 4 , 25 mM Tris-HCl, pH 7.1) containing low concentration of digitonin (20 μM) was added to the cells. Upon permeabilization, the buffer and digitonin were replaced by fresh buffer containing mitochondrial substrates (5 mM glutamate and 5 mM malate). Cells were allowed to rest for at least 5 min before measurements were taken. 20 μM CaCl 2 was applied to assess mitochondrial Ca 2+ handling. Confocal images were obtained using a Zeiss 710 equipped with a META detection system and a ×40 oil-immersion objective. Mitochondrial Ca 2+ measurements were undertaken using the 543 nm laser line and 560 nm longpass filter. A 488 nm Argon laser line was used to excite Fluo-4 which was measured between 510 and 550 nm. The pinhole set to give an optical slice of~2 μm and illumination intensity was kept to a minimum (at 0.1-0.2% of laser output) to avoid phototoxicity.
PTP opening threshold was determined with increasing concentrations of the electrogenic Ca 2+ ionophore ferutinin in stepwise fashion 22 . Briefly, ferutinin induces a rise in mitochondrial calcium and the concentration of ferutinin required to induce PTP opening (confirmed by the rapid loss of TMRM signal) was determined.

Cell toxicity experiment
Cells were incubated with propidium iodide (PI; 20 μM) and Hoechst 33342 (4.5 µM; Molecular Probes, Eugene, OR). Viable cells exclude the red fluorescent PI whereas Hoechst stains chromatin blue in all cells thus allowing dead cells to be quantified.

Statistical analysis
Statistical analysis and exponential curve fitting were performed using Origin 9 software. Experimental data are shown as means ± SEM. Statistical analysis between samples was performed using a one-way ANOVA with Bonferroni correction. Differences were considered to be significantly different if p < 0.05.

Ca 2+ dysregulation in LRRK2 KO primary cultures
First we assessed whether deletion of LRRK2 leads to a dysregulation in Ca 2+ homeostasis. To do so, we used primary neuron/glia co-cultures from the cerebral cortex/midbrain of wild-type (WT) and LRRK2 knockout (KO) mice. Both cytosolic and mitochondrial Ca 2+ was assessed upon adenosine triphosphate (ATP) application. Intracellular Ca 2+ levels are tightly controlled by the ER which is the main Ca 2+ store and mitochondria which "fine-tunes" Ca 2+ transients. Exogenous applied ATP triggers an intracellular Ca 2+ release from the endoplasmic reticulum (ER) via P2Y receptors followed by activation of the phospholipase C and IP 3 pathway in astrocytes 23 . The Ca 2+ release from the ER into the cytosol leads to a simultaneous increase in mitochondrial Ca 2+ . Application of ATP led to increased cytosolic and mitochondrial Ca 2+ in WT astrocytes and Ca 2+ levels recovered back to basal levels typically within 1 min (Fig. 1a). However, it was noted that after a short recovery mitochondrial Ca 2+ levels in KO astrocytes induced a delayed calcium deregulation in form of a secondary Ca 2+ increase. The underlying reason could be induced by opening of store-operated calcium channels or energy deprivation followed by mitochondrial Ca 2+ uptake 24 .
Altered mitochondrial Ca 2+ efflux in LRRK2 deficient cells Next, we investigated the underlying reason for the observed inhibited mitochondrial Ca 2+ efflux. Midbrain WT and KO neurons were permeabilized and mitochondrial Ca 2+ buffering was assessed by application of exogenous CaCl 2 . Ca 2+ was taken up by the mitochondria and the rate of mitochondrial Ca 2+ efflux was significantly lower (0.19 ± 0.04; p < 0.001) in KO mitochondria when compared with WT (0.4 ± 0.03; Fig. 1b, c). The same effect was observed in cortical cells. Figure 1d exemplifies that normalised mitochondrial Ca 2+ mirror non-normalised data. The maximal mitochondrial Ca 2+ uptake was not significantly different in KO neurons when compared with WT neurons indicating that the slower Ca 2+ efflux is not caused by an increased Ca 2+ uptake (Fig. 1e).
In excitable cells, mitochondrial Ca 2+ is predominantly removed by the sodium/Ca 2+ /lithium exchanger (NCLX) and, to a lesser extent, by other efflux pathways. NCLX translocates Na + and Ca 2+ ions across the inner mitochondrial membrane which can be visualised by loading cells with CoroNA (Na 2+ indicator) as well as xRhod1 (mitochondrial Ca 2+ indicator). In permeabilized WT neurons, application of CaCl 2 resulted in an increase in mitochondrial Ca 2+ and decrease in mitochondrial Na + , before Ca 2+ is extruded and Na + is taken up by the mitochondria (Fig. 1f). However, in KO neurons, NCLXmediated mitochondrial Ca 2+ extrusion is impaired suggesting an altered NCLX activity leading to the inhibited mitochondrial Ca 2+ efflux.
Inhibition of kinase activity (loss of function) was employed to confirm the mitochondrial Ca 2+ efflux deficiency is a LRRK2 mediated impairment. Therefore, two LRRK2 kinase inhibitors were employed to assess whether mitochondrial Ca 2+ is a direct consequence of LRRK2 deficiency. To do so, WT rat primary neuron/glia cocultures were exposed to either LRRK2 inhibitor 1 (LRRK-IN-1; 1 µM), CZC-25146 (5 µM) or DMSO vehicle control and mitochondrial Ca 2+ homoeostasis was assessed in neurons. Inhibition of LRRK2 kinase mirrored the significant inhibited mitochondrial Ca 2+ efflux observed in KO mitochondria (Fig. 1g, h; p < 0.001). As a positive control, the NCLX inhibitor CGP 37157 (10 µM) was employed which mirrored the significantly inhibited mitochondrial Ca 2+ efflux similar to that observed in LRRK2 deficient models (Fig. 1g, h; 0.05 ± 0.01; p < 0.001). Figure 1i shows that maximal mitochondrial Ca 2+ uptake was not significantly different when compared with control. Further, non-normalised data mirror the normalised efflux data.
Fibroblast from PD patients bearing LRRK2 mutations were employed to investigate whether this mitochondrial phenotype can also be observed in LRRK2associated PD, mirroring the PINK1 phenotype. Fibroblast from patients bearing either LRRK2 G2019S (kinase domain), R1441G (ROC domain) or Y1699C (COR domain) mutation and unaffected controls were employed to assess whether mitochondrial Ca 2+ buffering was a common phenotype in PD as well as LRRK2 deficiencies. Indeed, the mitochondrial Ca 2+ efflux was significantly inhibited in G2019S (Fig. 1j, k; 0.14 ± 0.015; p < 0.001), R1441G (0.21 ± 0.03; p < 0.001; n = 3  Fig. 1j, k). Maximal mitochondrial Ca 2+ uptake was not significantly different between cell types (Fig. 1l) PD-associated mutations result in lower mitochondrial Ca 2+ handling capacity The observed altered mitochondrial Ca 2+ efflux deficiency is likely to lower the mitochondrial Ca 2+ buffering capacity. To test this, we employed a natural electrogenic Ca 2+ ionophore ferutinin to promote mitochondrial Ca 2+ overload independently of MCU and trigger Ca 2+dependent PTP opening 22,25 . PTP opening can be visualised by a significant drop in ΔΨm (Fig. 2a). Ferutinin was applied and the PTP in LRRK2 mutation bearing fibroblasts opened at significantly lower ferutinin concentrations. At 20 μM 62% of control cells ( ± 5.1%; p < 0.001; Fig. 2b) depolarised, whereas 94% (G2019S), 100% (R1441G) and 99% (Y1699C) were found to be depolarised. These results indicate that these LRRK2 mutations lead to a significantly lower Ca 2+ buffering capacity when compared with control which may lead to an increased bioenergetic imbalance in cells with a highenergy demand such as neurons.

Upregulation of NCLX rescues the inhibited mitochondrial Ca 2+ efflux
We have previously shown 16  mitochondrial Ca 2+ efflux. Therefore, this study tested whether activation of NCLX is able to overcome the mitochondrial Ca 2+ efflux phenotype observed in LRRK2 deficient models.
Direct upregulation of NCLX is achieved by expression of a constitutively active NCLX. We have previously reported that the phosphomimetic mutant of NCLX (NCLX S258D ) prevents mitochondrial Ca 2+ overload in a PINK1 deficient model 16 . Hence, this study expressed NCLX S258D in LRRK2 KO primary cultures and mitochondrial Ca 2+ was recorded in neurons. Expression of NCLX S258D restored the efflux back to levels similar to those observed in WT neurons (0.23 ± 0.02 and 0.2 ± 0.01, respectively; p < 0.001; Fig. 4a). Furthermore, expression of NCLX S258D in patient fibroblast bearing the G2019S fibroblasts restored mitochondrial Ca 2+ efflux back to unaffected control levels (0.38 ± 0.02 and 0.36 ± 0.01, respectively; p < 0.001; Fig. 4b).

Discussion
Mitochondrial dysfunction has been linked to both familial and sporadic forms of PD. Several PD risk genes, such as PINK-1, Parkin and DJ-1 have been associated with mitochondrial pathogenesis observed in PD with a key role in mitochondrial Ca 2+ homoeostasis [26][27][28] . Mitochondrial Ca 2+ overload has a direct effect on mitochondrial bioenergetics and lowers the threshold for PTP opening leading to premature cell death 29,30 and has previously been demonstrated in loss-of-function, recessively inherited mutations of PINK1 17,31 .
In this present study, we show that modulation of LRRK2 through deletion (KO), inhibition (de-phosphorylation) or mutagenesis of LRRK2 alters the mitochondrial Ca 2+ efflux; therefore linking another PD risk gene to NCLX leading to decreased PTP opening threshold and an increased dopamine-induced neuronal death (Figs. 1, 2 , 5). LRRK2 has previously been associated with mitochondrial membranes but no direct interaction between LRRK2 and NCLX could be confirmed in our previous manuscript (supplementary data) and is therefore unlikely to directly modulate NCLX Ca 2+ efflux 11,16,32 .
We and other have previously described that a lower ΔΨ m can result in reduced NCLX Ca 2+ efflux 16,17 . However, ΔΨ m in all LRRK2 models employed in this current study is either not changed (fibroblasts) or elevated (inhibitor/KO) when compared with control conditions suggesting that the ΔΨ m is not the underlying reason for the mitochondrial Ca 2+ efflux deficiency (Fig. 3). Papkovskaia et al. 33 have previously reported that fibroblasts bearing G2019S bear a lower ΔΨ m . This current study employed three matched control cell lines as well as three mutations affecting different domains of LRRK2 (ROC, COR and kinase domain) and could not find any significant changes in the ΔΨ m .
A global Ca 2+ dysregulation can lead to mitochondrial Ca 2+ dysregulation, For example, Bedford et al. 34 have provided evidence that LRRK2 stimulates voltage-gated Ca 2+ channels, while the PD mutation G2019S leads to an increased channel stimulation when compared with wildtype LRRK2. Furthermore, LRRK2 has been proposed to stimulate plasma membrane sodium/calcium exchanger activity 35 and nicotinic acid adenine dinucleotide phosphate receptors 36 . These reports provide evidence for a global Ca 2+ regulation through LRRK2 where mutations can lead to a Ca 2+ dysregulation affecting organelles and cellular functions. Indeed, Verma et al. 37 has shown that mitochondrial Ca 2+ contributes to dendrite injury in a LRRK2 PD model. The authors have also reported transcriptional upregulation the mitochondrial calcium uniporter and the mitochondrial calcium uptake 1 protein leading to increased mitochondrial Ca 2+ influx. Interestingly, we did not observe an altered mitochondrial Ca 2+ uptake in any of the models employed in this current study indicating other driving factor(s) for the mitochondrial Ca 2+ efflux pathology (Fig. 1e, i, l).
The exact mechanism through which LRRK2 acts upon NCLX Ca 2+ extrusion requires further investigation. LRRK2 has previously been reported to interact with the regulatory subunit of PKA, negatively regulating activity during synaptogenesis 38 . While LRRK2 is predominantly expressed in the cytosol, it has previously been suggested to affect mitochondrial dynamics 39,40 , with few studies providing evidence for a mitochondrial localisation 11,33 .
Our study has shown that mutations in LRRK2 phenocopy pharmacological inhibition and knockout of LRRK2 despite opposing reported effects (gain vs loss of function) 41,42 . One possible explanation may be link between LRRK2/Rab32 and PKA. It was shown that the small GTPase Rab32 interacts with the N-terminal armadillo repeat LRRK2. Rab32 itself has been reported to act as a PKA anchoring protein localising PKA to mitochondria and was shown to modulate mitochondrial associated membranes and mitochondrial fission 43,44 . How mutations or inhibition affect these interactions and the consequences on mitochondrial Ca 2+ are yet to be unravelled. Therefore, the link between LRRK2/Rab32/ PKA and its role in mitochondrial health in relation to PD requires further investigations.
We have previously shown that upregulation of NCLX via a cAMP/PKA-dependent activation through the application of forskolin rescues the mitochondrial Ca 2+ Fig. 5 Protective effects of NCLX upregulation from dopamineinduced neuronal cell death. LRRK2 inhibitor treated neuronal cultures were challenged with 70 μM Dopamine for 12 h and cell death was assessed using PI and Hoechst (N = 3 experiments; each data point represents one field of view with an average of 58 neurons). *p < 0.05; **p < 0.01 efflux in PINK1 deficient models 16 . In this current study expression of the constitutively active phosphomimetic mutant of NCLX (NCLX S258D ) also rescued the mitochondrial Ca 2+ efflux in LRRK2 mutation bearing fibroblast and LRRK2 inhibitor treated cells in a similar manner previously described in a PINK PD model 16,45 (Fig. 3). We and others have previously shown that cAMP-mediated PKA activation using forskolin reverses pathologies in a PINK1 and LRRK2 G2019S PD model 16,46 . This was mirrored in this current study as application of forskolin (indirect NCLX upregulation) restored the Ca 2+ buffering capacity back to control levels in LRRK2 mutation bearing fibroblasts (Fig. 2). Delayed mitochondrial Ca 2+ efflux, as described in this study, can lead to mitochondrial swelling and a lowered PTP opening threshold. PTP opening can be either triggered by high Ca 2+ levels and or high superoxide levels and is characterised by the loss of the mitochondrial membrane potential, swelling of the mitochondrial matrix, rupture of the outer membrane and release of internalised Ca 2+ and pro-apoptotic proteins including cytochrome C. Therefore, opening of the PTP is the first trigger for the induction of cell death pathways (apoptosis and necrosis) which can have particular detrimental effects in cells with a high-energy demand such as dopaminergic neurons which are predominantly affected in PD 45,47 . Here, we have shown that LRRK2 inhibitors make neurons more prone to dopamine-induced cell death (Ca 2+ -dependent). By upregulation of NCLX and therefore enhancing mitochondrial Ca 2+ efflux, we were able to rescue this dopamine-induced neuronal cell death (Fig. 5).
Inhibition of NCLX in midbrain neurons in LRRK2 mutation is specifically important because it enhances the mitochondrial calcium level that increase probability to open permeability transition pore in mitochondria and trigger cell death in response to additional stresses which are typical for PD pathology such as oligomeric αsynuclein and dopamine 48,49 .
This study provides evidence that mitochondrial Ca 2+ efflux deficiency is a common phenotype observed in two PD-associated risk genes (PINK1 and LRRK2). This phenotype can be rescued by an upregulation of NCLX via a direct or indirect cAMP/PKA-dependent approach which may offer a novel therapeutic strategy in PD. The observed phenotype associated with LRRK2 modulation may be of particular importance when employing LRRK2 inhibitors as mitochondrial bioenergetics are likely to be negatively affected. Prolonged mitochondrial Ca 2+ dysregulation has detrimental bioenergetic consequences and is of particular importance to high-energy-demanding cells such as neurons. Since mitochondrial Ca 2+ dysregulation is a common phenotype in many neurodegenerative disease models NCLX and its regulators may serve as a novel therapeutic strategy.